diff --git a/added_tokens.json b/added_tokens.json new file mode 100644 index 0000000000000000000000000000000000000000..301df2d484b892fb69765533defad74b1169af4d --- /dev/null +++ b/added_tokens.json @@ -0,0 +1,6 @@ +{ + "": 32003, + "": 32001, + "": 32002, + "[PAD]": 32000 +} diff --git a/checkpoint-14000/added_tokens.json b/checkpoint-14000/added_tokens.json new file mode 100644 index 0000000000000000000000000000000000000000..301df2d484b892fb69765533defad74b1169af4d --- /dev/null +++ b/checkpoint-14000/added_tokens.json @@ -0,0 +1,6 @@ +{ + "": 32003, + "": 32001, + "": 32002, + "[PAD]": 32000 +} diff --git a/checkpoint-14000/config.json b/checkpoint-14000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..066a0107385020bf48a41a53e63dd916e2f96670 --- /dev/null +++ b/checkpoint-14000/config.json @@ -0,0 +1,30 @@ +{ + "_name_or_path": "/scratch/wzhu59/LLaVA-Med/weights/llava-med-7b-pretrain", + "architectures": [ + "LlavaLlamaForCausalLM" + ], + "bos_token_id": 0, + "eos_token_id": 1, + "freeze_mm_mlp_adapter": false, + "hidden_act": "silu", + "hidden_size": 4096, + "initializer_range": 0.02, + "intermediate_size": 11008, + "max_sequence_length": 2048, + "mm_hidden_size": 1024, + "mm_use_im_start_end": true, + "mm_vision_select_layer": -2, + "mm_vision_tower": "openai/clip-vit-large-patch14", + "model_type": "llava", + "num_attention_heads": 32, + "num_hidden_layers": 32, + "pad_token_id": -1, + "rms_norm_eps": 1e-06, + "tie_word_embeddings": false, + "torch_dtype": "float32", + "transformers_version": "4.28.0.dev0", + "tune_mm_mlp_adapter": false, + "use_cache": false, + "use_mm_proj": true, + "vocab_size": 32004 +} diff --git a/checkpoint-14000/generation_config.json b/checkpoint-14000/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..128b48946713166edcc3e3c62ae5fe285df776be --- /dev/null +++ b/checkpoint-14000/generation_config.json @@ -0,0 +1,7 @@ +{ + "_from_model_config": true, + "bos_token_id": 0, + "eos_token_id": 1, + "pad_token_id": 0, + "transformers_version": "4.28.0.dev0" +} diff --git a/checkpoint-14000/optimizer.pt b/checkpoint-14000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..c43d9ac056c1223e5ce8590c388d5c50f16099ad --- /dev/null +++ b/checkpoint-14000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5839ededeb22df2229708e51393affa15e1bcb85be0626194559f538df34bb55 +size 13485322046 diff --git a/checkpoint-14000/pytorch_model-00001-of-00003.bin b/checkpoint-14000/pytorch_model-00001-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..dcec9adbee0fd926f6daef32d3caa0f129f61a09 --- /dev/null +++ b/checkpoint-14000/pytorch_model-00001-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd16410cf72340bd90c514293b4e21d9e5500ec2c134220454912942528f3b4e +size 9878059026 diff --git a/checkpoint-14000/pytorch_model-00002-of-00003.bin b/checkpoint-14000/pytorch_model-00002-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..26fe043a51549b044c01733844a2fcc032073727 --- /dev/null +++ b/checkpoint-14000/pytorch_model-00002-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4d5a76fb4eb120da4879b4f9df298116a441f82801cf53a65b71fbe1cec37a30 +size 9894805046 diff --git a/checkpoint-14000/pytorch_model-00003-of-00003.bin b/checkpoint-14000/pytorch_model-00003-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..0c092eaee585f9c141a193bfcf00879cc3c564ba --- /dev/null +++ b/checkpoint-14000/pytorch_model-00003-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed60f71dd1637ef65bc1d0fc0f75268333fc995975761e8c3f7324cd3178ed31 +size 7197853429 diff --git a/checkpoint-14000/pytorch_model.bin.index.json b/checkpoint-14000/pytorch_model.bin.index.json new file mode 100644 index 0000000000000000000000000000000000000000..50d9a764a2282762889d821cfaa49b284963c070 --- /dev/null +++ b/checkpoint-14000/pytorch_model.bin.index.json @@ -0,0 +1,332 @@ +{ + "metadata": { + "total_size": 26970595328 + }, + "weight_map": { + "lm_head.weight": "pytorch_model-00003-of-00003.bin", + "model.embed_tokens.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.12.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.20.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.24.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.30.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.mm_projector.bias": "pytorch_model-00003-of-00003.bin", + "model.mm_projector.weight": "pytorch_model-00003-of-00003.bin", + "model.norm.weight": "pytorch_model-00003-of-00003.bin" + } +} diff --git a/checkpoint-14000/rng_state_0.pth b/checkpoint-14000/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..66e030e30e30a2c53a2c60fa0225c4ca873ad8fa --- /dev/null +++ b/checkpoint-14000/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b056bea9e6a63fa35e94c0ac207b3f3ec6085b3546669b620eab15cf871e1df7 +size 14583 diff --git a/checkpoint-14000/rng_state_1.pth b/checkpoint-14000/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..59b4568ccb9870000bee08b1feb64388759ed096 --- /dev/null +++ b/checkpoint-14000/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83eeface4abf9d73bba7f526bfde1a69bbd9473a70613d68e7daabeda4177537 +size 14583 diff --git a/checkpoint-14000/rng_state_2.pth b/checkpoint-14000/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..15857d6785e9b26d87fb15e2f92e632c4b03469d --- /dev/null +++ b/checkpoint-14000/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aacc99f2cb89bb983d4f5e62efd2be8fd4d3aafb32a8d3abf30054ed93208428 +size 14583 diff --git a/checkpoint-14000/rng_state_3.pth b/checkpoint-14000/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..a8ad4c42870d260104c43886530cc1b9d1ea0409 --- /dev/null +++ b/checkpoint-14000/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:351ce80e1bdb15bae6094c50c40b30709211854c1b94f32c30eaf93fd915d5ad +size 14583 diff --git a/checkpoint-14000/scheduler.pt b/checkpoint-14000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..f40923dff02ee3375a576704e12ffc6c07fbb77f --- /dev/null +++ b/checkpoint-14000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d0e7631abbbbfa318209ecb22d76dea9b039d23429383dac9d943ea3e1ce1cde +size 627 diff --git a/checkpoint-14000/special_tokens_map.json b/checkpoint-14000/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..5ce7dbc9759ecf5596c5122abbdf0b431355efa4 --- /dev/null +++ b/checkpoint-14000/special_tokens_map.json @@ -0,0 +1,6 @@ +{ + "bos_token": "", + "eos_token": "", + "pad_token": "[PAD]", + "unk_token": "" +} diff --git a/checkpoint-14000/tokenizer.model b/checkpoint-14000/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-14000/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-14000/tokenizer_config.json b/checkpoint-14000/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..22bcdfb351739d00df7b9b12f99757c6e7e0b540 --- /dev/null +++ b/checkpoint-14000/tokenizer_config.json @@ -0,0 +1,32 @@ +{ + "bos_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + }, + "clean_up_tokenization_spaces": false, + "eos_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + }, + "model_max_length": 2048, + "pad_token": null, + "padding_side": "right", + "special_tokens_map_file": "/nobackup/haotian/.cache/huggingface/hub/models--lmsys--vicuna-7b-delta-v0/snapshots/6d2e645f62e734821be8c392c08c8b2098b35434/special_tokens_map.json", + "tokenizer_class": "LlamaTokenizer", + "unk_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-14000/trainer_state.json b/checkpoint-14000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..5f43b19d3f1f5748aa3d66232a3d91e18fb8ad5f --- /dev/null +++ b/checkpoint-14000/trainer_state.json @@ -0,0 +1,84016 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 5.8614193008164115, + "global_step": 14000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 3.484320557491289e-08, + "loss": 1.487, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 6.968641114982578e-08, + "loss": 1.5418, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 1.045296167247387e-07, + "loss": 1.2413, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 1.3937282229965157e-07, + "loss": 1.1777, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 1.7421602787456448e-07, + "loss": 0.9729, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 2.090592334494774e-07, + "loss": 0.9887, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 2.439024390243903e-07, + "loss": 0.9151, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 2.7874564459930313e-07, + "loss": 0.9341, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 3.1358885017421604e-07, + "loss": 0.9024, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 3.4843205574912896e-07, + "loss": 0.96, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 3.832752613240418e-07, + "loss": 0.9397, + "step": 11 + }, + { + "epoch": 0.01, + "learning_rate": 4.181184668989548e-07, + "loss": 1.0473, + "step": 12 + }, + { + "epoch": 0.01, + "learning_rate": 4.5296167247386764e-07, + "loss": 0.8521, + "step": 13 + }, + { + "epoch": 0.01, + "learning_rate": 4.878048780487805e-07, + "loss": 0.9829, + "step": 14 + }, + { + "epoch": 0.01, + "learning_rate": 5.226480836236935e-07, + "loss": 1.1014, + "step": 15 + }, + { + "epoch": 0.01, + "learning_rate": 5.574912891986063e-07, + "loss": 0.9186, + "step": 16 + }, + { + "epoch": 0.01, + "learning_rate": 5.923344947735192e-07, + "loss": 0.8867, + "step": 17 + }, + { + "epoch": 0.01, + "learning_rate": 6.271777003484321e-07, + "loss": 0.8071, + "step": 18 + }, + { + "epoch": 0.01, + "learning_rate": 6.62020905923345e-07, + "loss": 0.8481, + "step": 19 + }, + { + "epoch": 0.01, + "learning_rate": 6.968641114982579e-07, + "loss": 0.8675, + "step": 20 + }, + { + "epoch": 0.01, + "learning_rate": 7.317073170731707e-07, + "loss": 0.9635, + "step": 21 + }, + { + "epoch": 0.01, + "learning_rate": 7.665505226480836e-07, + "loss": 0.8868, + "step": 22 + }, + { + "epoch": 0.01, + "learning_rate": 8.013937282229965e-07, + "loss": 0.9029, + "step": 23 + }, + { + "epoch": 0.01, + "learning_rate": 8.362369337979096e-07, + "loss": 0.9118, + "step": 24 + }, + { + "epoch": 0.01, + "learning_rate": 8.710801393728225e-07, + "loss": 0.9768, + "step": 25 + }, + { + "epoch": 0.01, + "learning_rate": 9.059233449477353e-07, + "loss": 0.8292, + "step": 26 + }, + { + "epoch": 0.01, + "learning_rate": 9.407665505226482e-07, + "loss": 0.8865, + "step": 27 + }, + { + "epoch": 0.01, + "learning_rate": 9.75609756097561e-07, + "loss": 0.9918, + "step": 28 + }, + { + "epoch": 0.01, + "learning_rate": 1.010452961672474e-06, + "loss": 0.7146, + "step": 29 + }, + { + "epoch": 0.01, + "learning_rate": 1.045296167247387e-06, + "loss": 0.907, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 1.0801393728222997e-06, + "loss": 0.8015, + "step": 31 + }, + { + "epoch": 0.01, + "learning_rate": 1.1149825783972125e-06, + "loss": 0.7597, + "step": 32 + }, + { + "epoch": 0.01, + "learning_rate": 1.1498257839721255e-06, + "loss": 0.8095, + "step": 33 + }, + { + "epoch": 0.01, + "learning_rate": 1.1846689895470384e-06, + "loss": 0.9036, + "step": 34 + }, + { + "epoch": 0.01, + "learning_rate": 1.2195121951219514e-06, + "loss": 0.8162, + "step": 35 + }, + { + "epoch": 0.02, + "learning_rate": 1.2543554006968642e-06, + "loss": 0.9883, + "step": 36 + }, + { + "epoch": 0.02, + "learning_rate": 1.289198606271777e-06, + "loss": 0.8634, + "step": 37 + }, + { + "epoch": 0.02, + "learning_rate": 1.32404181184669e-06, + "loss": 0.7832, + "step": 38 + }, + { + "epoch": 0.02, + "learning_rate": 1.3588850174216028e-06, + "loss": 0.7378, + "step": 39 + }, + { + "epoch": 0.02, + "learning_rate": 1.3937282229965158e-06, + "loss": 0.8348, + "step": 40 + }, + { + "epoch": 0.02, + "learning_rate": 1.4285714285714286e-06, + "loss": 0.6925, + "step": 41 + }, + { + "epoch": 0.02, + "learning_rate": 1.4634146341463414e-06, + "loss": 0.8969, + "step": 42 + }, + { + "epoch": 0.02, + "learning_rate": 1.4982578397212545e-06, + "loss": 0.7099, + "step": 43 + }, + { + "epoch": 0.02, + "learning_rate": 1.5331010452961673e-06, + "loss": 0.8763, + "step": 44 + }, + { + "epoch": 0.02, + "learning_rate": 1.56794425087108e-06, + "loss": 0.7967, + "step": 45 + }, + { + "epoch": 0.02, + "learning_rate": 1.602787456445993e-06, + "loss": 0.7528, + "step": 46 + }, + { + "epoch": 0.02, + "learning_rate": 1.6376306620209059e-06, + "loss": 0.8367, + "step": 47 + }, + { + "epoch": 0.02, + "learning_rate": 1.6724738675958191e-06, + "loss": 0.8172, + "step": 48 + }, + { + "epoch": 0.02, + "learning_rate": 1.707317073170732e-06, + "loss": 0.7745, + "step": 49 + }, + { + "epoch": 0.02, + "learning_rate": 1.742160278745645e-06, + "loss": 0.8643, + "step": 50 + }, + { + "epoch": 0.02, + "learning_rate": 1.7770034843205577e-06, + "loss": 0.7374, + "step": 51 + }, + { + "epoch": 0.02, + "learning_rate": 1.8118466898954705e-06, + "loss": 0.8095, + "step": 52 + }, + { + "epoch": 0.02, + "learning_rate": 1.8466898954703836e-06, + "loss": 0.7755, + "step": 53 + }, + { + "epoch": 0.02, + "learning_rate": 1.8815331010452964e-06, + "loss": 0.8084, + "step": 54 + }, + { + "epoch": 0.02, + "learning_rate": 1.916376306620209e-06, + "loss": 0.7396, + "step": 55 + }, + { + "epoch": 0.02, + "learning_rate": 1.951219512195122e-06, + "loss": 0.7355, + "step": 56 + }, + { + "epoch": 0.02, + "learning_rate": 1.986062717770035e-06, + "loss": 0.7007, + "step": 57 + }, + { + "epoch": 0.02, + "learning_rate": 2.020905923344948e-06, + "loss": 0.8282, + "step": 58 + }, + { + "epoch": 0.02, + "learning_rate": 2.055749128919861e-06, + "loss": 0.7561, + "step": 59 + }, + { + "epoch": 0.03, + "learning_rate": 2.090592334494774e-06, + "loss": 0.8712, + "step": 60 + }, + { + "epoch": 0.03, + "learning_rate": 2.1254355400696864e-06, + "loss": 0.8559, + "step": 61 + }, + { + "epoch": 0.03, + "learning_rate": 2.1602787456445995e-06, + "loss": 0.7767, + "step": 62 + }, + { + "epoch": 0.03, + "learning_rate": 2.1951219512195125e-06, + "loss": 0.81, + "step": 63 + }, + { + "epoch": 0.03, + "learning_rate": 2.229965156794425e-06, + "loss": 0.805, + "step": 64 + }, + { + "epoch": 0.03, + "learning_rate": 2.264808362369338e-06, + "loss": 0.6648, + "step": 65 + }, + { + "epoch": 0.03, + "learning_rate": 2.299651567944251e-06, + "loss": 0.8027, + "step": 66 + }, + { + "epoch": 0.03, + "learning_rate": 2.334494773519164e-06, + "loss": 0.7623, + "step": 67 + }, + { + "epoch": 0.03, + "learning_rate": 2.3693379790940767e-06, + "loss": 0.7903, + "step": 68 + }, + { + "epoch": 0.03, + "learning_rate": 2.4041811846689897e-06, + "loss": 0.6917, + "step": 69 + }, + { + "epoch": 0.03, + "learning_rate": 2.4390243902439027e-06, + "loss": 0.8177, + "step": 70 + }, + { + "epoch": 0.03, + "learning_rate": 2.4738675958188153e-06, + "loss": 0.7348, + "step": 71 + }, + { + "epoch": 0.03, + "learning_rate": 2.5087108013937284e-06, + "loss": 0.6916, + "step": 72 + }, + { + "epoch": 0.03, + "learning_rate": 2.5435540069686414e-06, + "loss": 0.6663, + "step": 73 + }, + { + "epoch": 0.03, + "learning_rate": 2.578397212543554e-06, + "loss": 0.6881, + "step": 74 + }, + { + "epoch": 0.03, + "learning_rate": 2.6132404181184674e-06, + "loss": 0.6372, + "step": 75 + }, + { + "epoch": 0.03, + "learning_rate": 2.64808362369338e-06, + "loss": 0.7268, + "step": 76 + }, + { + "epoch": 0.03, + "learning_rate": 2.682926829268293e-06, + "loss": 0.7138, + "step": 77 + }, + { + "epoch": 0.03, + "learning_rate": 2.7177700348432056e-06, + "loss": 0.7341, + "step": 78 + }, + { + "epoch": 0.03, + "learning_rate": 2.7526132404181186e-06, + "loss": 0.7041, + "step": 79 + }, + { + "epoch": 0.03, + "learning_rate": 2.7874564459930316e-06, + "loss": 0.7239, + "step": 80 + }, + { + "epoch": 0.03, + "learning_rate": 2.8222996515679447e-06, + "loss": 0.6809, + "step": 81 + }, + { + "epoch": 0.03, + "learning_rate": 2.8571428571428573e-06, + "loss": 0.7256, + "step": 82 + }, + { + "epoch": 0.03, + "learning_rate": 2.8919860627177703e-06, + "loss": 0.7205, + "step": 83 + }, + { + "epoch": 0.04, + "learning_rate": 2.926829268292683e-06, + "loss": 0.6596, + "step": 84 + }, + { + "epoch": 0.04, + "learning_rate": 2.9616724738675963e-06, + "loss": 0.657, + "step": 85 + }, + { + "epoch": 0.04, + "learning_rate": 2.996515679442509e-06, + "loss": 0.6502, + "step": 86 + }, + { + "epoch": 0.04, + "learning_rate": 3.031358885017422e-06, + "loss": 0.7095, + "step": 87 + }, + { + "epoch": 0.04, + "learning_rate": 3.0662020905923345e-06, + "loss": 0.6643, + "step": 88 + }, + { + "epoch": 0.04, + "learning_rate": 3.1010452961672475e-06, + "loss": 0.6621, + "step": 89 + }, + { + "epoch": 0.04, + "learning_rate": 3.13588850174216e-06, + "loss": 0.7324, + "step": 90 + }, + { + "epoch": 0.04, + "learning_rate": 3.1707317073170736e-06, + "loss": 0.73, + "step": 91 + }, + { + "epoch": 0.04, + "learning_rate": 3.205574912891986e-06, + "loss": 0.7109, + "step": 92 + }, + { + "epoch": 0.04, + "learning_rate": 3.240418118466899e-06, + "loss": 0.695, + "step": 93 + }, + { + "epoch": 0.04, + "learning_rate": 3.2752613240418118e-06, + "loss": 0.5556, + "step": 94 + }, + { + "epoch": 0.04, + "learning_rate": 3.310104529616725e-06, + "loss": 0.6237, + "step": 95 + }, + { + "epoch": 0.04, + "learning_rate": 3.3449477351916382e-06, + "loss": 0.6923, + "step": 96 + }, + { + "epoch": 0.04, + "learning_rate": 3.379790940766551e-06, + "loss": 0.7276, + "step": 97 + }, + { + "epoch": 0.04, + "learning_rate": 3.414634146341464e-06, + "loss": 0.6814, + "step": 98 + }, + { + "epoch": 0.04, + "learning_rate": 3.4494773519163764e-06, + "loss": 0.7419, + "step": 99 + }, + { + "epoch": 0.04, + "learning_rate": 3.48432055749129e-06, + "loss": 0.6808, + "step": 100 + }, + { + "epoch": 0.04, + "learning_rate": 3.5191637630662025e-06, + "loss": 0.6811, + "step": 101 + }, + { + "epoch": 0.04, + "learning_rate": 3.5540069686411155e-06, + "loss": 0.6577, + "step": 102 + }, + { + "epoch": 0.04, + "learning_rate": 3.588850174216028e-06, + "loss": 0.6782, + "step": 103 + }, + { + "epoch": 0.04, + "learning_rate": 3.623693379790941e-06, + "loss": 0.7048, + "step": 104 + }, + { + "epoch": 0.04, + "learning_rate": 3.6585365853658537e-06, + "loss": 0.6673, + "step": 105 + }, + { + "epoch": 0.04, + "learning_rate": 3.693379790940767e-06, + "loss": 0.6239, + "step": 106 + }, + { + "epoch": 0.04, + "learning_rate": 3.7282229965156797e-06, + "loss": 0.6954, + "step": 107 + }, + { + "epoch": 0.05, + "learning_rate": 3.7630662020905927e-06, + "loss": 0.6656, + "step": 108 + }, + { + "epoch": 0.05, + "learning_rate": 3.7979094076655053e-06, + "loss": 0.6732, + "step": 109 + }, + { + "epoch": 0.05, + "learning_rate": 3.832752613240418e-06, + "loss": 0.7624, + "step": 110 + }, + { + "epoch": 0.05, + "learning_rate": 3.867595818815331e-06, + "loss": 0.7086, + "step": 111 + }, + { + "epoch": 0.05, + "learning_rate": 3.902439024390244e-06, + "loss": 0.6284, + "step": 112 + }, + { + "epoch": 0.05, + "learning_rate": 3.937282229965157e-06, + "loss": 0.6643, + "step": 113 + }, + { + "epoch": 0.05, + "learning_rate": 3.97212543554007e-06, + "loss": 0.6443, + "step": 114 + }, + { + "epoch": 0.05, + "learning_rate": 4.006968641114983e-06, + "loss": 0.6025, + "step": 115 + }, + { + "epoch": 0.05, + "learning_rate": 4.041811846689896e-06, + "loss": 0.6651, + "step": 116 + }, + { + "epoch": 0.05, + "learning_rate": 4.076655052264808e-06, + "loss": 0.6758, + "step": 117 + }, + { + "epoch": 0.05, + "learning_rate": 4.111498257839722e-06, + "loss": 0.6962, + "step": 118 + }, + { + "epoch": 0.05, + "learning_rate": 4.146341463414634e-06, + "loss": 0.6613, + "step": 119 + }, + { + "epoch": 0.05, + "learning_rate": 4.181184668989548e-06, + "loss": 0.6735, + "step": 120 + }, + { + "epoch": 0.05, + "learning_rate": 4.21602787456446e-06, + "loss": 0.5424, + "step": 121 + }, + { + "epoch": 0.05, + "learning_rate": 4.250871080139373e-06, + "loss": 0.7245, + "step": 122 + }, + { + "epoch": 0.05, + "learning_rate": 4.2857142857142855e-06, + "loss": 0.5873, + "step": 123 + }, + { + "epoch": 0.05, + "learning_rate": 4.320557491289199e-06, + "loss": 0.7426, + "step": 124 + }, + { + "epoch": 0.05, + "learning_rate": 4.3554006968641115e-06, + "loss": 0.7477, + "step": 125 + }, + { + "epoch": 0.05, + "learning_rate": 4.390243902439025e-06, + "loss": 0.6925, + "step": 126 + }, + { + "epoch": 0.05, + "learning_rate": 4.4250871080139375e-06, + "loss": 0.7745, + "step": 127 + }, + { + "epoch": 0.05, + "learning_rate": 4.45993031358885e-06, + "loss": 0.6976, + "step": 128 + }, + { + "epoch": 0.05, + "learning_rate": 4.4947735191637636e-06, + "loss": 0.6296, + "step": 129 + }, + { + "epoch": 0.05, + "learning_rate": 4.529616724738676e-06, + "loss": 0.6011, + "step": 130 + }, + { + "epoch": 0.05, + "learning_rate": 4.56445993031359e-06, + "loss": 0.6661, + "step": 131 + }, + { + "epoch": 0.06, + "learning_rate": 4.599303135888502e-06, + "loss": 0.6491, + "step": 132 + }, + { + "epoch": 0.06, + "learning_rate": 4.634146341463416e-06, + "loss": 0.6139, + "step": 133 + }, + { + "epoch": 0.06, + "learning_rate": 4.668989547038328e-06, + "loss": 0.6212, + "step": 134 + }, + { + "epoch": 0.06, + "learning_rate": 4.703832752613241e-06, + "loss": 0.7824, + "step": 135 + }, + { + "epoch": 0.06, + "learning_rate": 4.738675958188153e-06, + "loss": 0.6765, + "step": 136 + }, + { + "epoch": 0.06, + "learning_rate": 4.773519163763067e-06, + "loss": 0.6975, + "step": 137 + }, + { + "epoch": 0.06, + "learning_rate": 4.8083623693379794e-06, + "loss": 0.7251, + "step": 138 + }, + { + "epoch": 0.06, + "learning_rate": 4.843205574912893e-06, + "loss": 0.7232, + "step": 139 + }, + { + "epoch": 0.06, + "learning_rate": 4.8780487804878055e-06, + "loss": 0.6243, + "step": 140 + }, + { + "epoch": 0.06, + "learning_rate": 4.912891986062718e-06, + "loss": 0.6315, + "step": 141 + }, + { + "epoch": 0.06, + "learning_rate": 4.947735191637631e-06, + "loss": 0.632, + "step": 142 + }, + { + "epoch": 0.06, + "learning_rate": 4.982578397212544e-06, + "loss": 0.7962, + "step": 143 + }, + { + "epoch": 0.06, + "learning_rate": 5.017421602787457e-06, + "loss": 0.5554, + "step": 144 + }, + { + "epoch": 0.06, + "learning_rate": 5.052264808362369e-06, + "loss": 0.6004, + "step": 145 + }, + { + "epoch": 0.06, + "learning_rate": 5.087108013937283e-06, + "loss": 0.5713, + "step": 146 + }, + { + "epoch": 0.06, + "learning_rate": 5.121951219512195e-06, + "loss": 0.6607, + "step": 147 + }, + { + "epoch": 0.06, + "learning_rate": 5.156794425087108e-06, + "loss": 0.627, + "step": 148 + }, + { + "epoch": 0.06, + "learning_rate": 5.1916376306620205e-06, + "loss": 0.6944, + "step": 149 + }, + { + "epoch": 0.06, + "learning_rate": 5.226480836236935e-06, + "loss": 0.6617, + "step": 150 + }, + { + "epoch": 0.06, + "learning_rate": 5.261324041811847e-06, + "loss": 0.6516, + "step": 151 + }, + { + "epoch": 0.06, + "learning_rate": 5.29616724738676e-06, + "loss": 0.6271, + "step": 152 + }, + { + "epoch": 0.06, + "learning_rate": 5.331010452961673e-06, + "loss": 0.5682, + "step": 153 + }, + { + "epoch": 0.06, + "learning_rate": 5.365853658536586e-06, + "loss": 0.6439, + "step": 154 + }, + { + "epoch": 0.06, + "learning_rate": 5.400696864111499e-06, + "loss": 0.7387, + "step": 155 + }, + { + "epoch": 0.07, + "learning_rate": 5.435540069686411e-06, + "loss": 0.5682, + "step": 156 + }, + { + "epoch": 0.07, + "learning_rate": 5.470383275261324e-06, + "loss": 0.5933, + "step": 157 + }, + { + "epoch": 0.07, + "learning_rate": 5.505226480836237e-06, + "loss": 0.6817, + "step": 158 + }, + { + "epoch": 0.07, + "learning_rate": 5.540069686411151e-06, + "loss": 0.6364, + "step": 159 + }, + { + "epoch": 0.07, + "learning_rate": 5.574912891986063e-06, + "loss": 0.5878, + "step": 160 + }, + { + "epoch": 0.07, + "learning_rate": 5.609756097560977e-06, + "loss": 0.6615, + "step": 161 + }, + { + "epoch": 0.07, + "learning_rate": 5.644599303135889e-06, + "loss": 0.6586, + "step": 162 + }, + { + "epoch": 0.07, + "learning_rate": 5.679442508710802e-06, + "loss": 0.6446, + "step": 163 + }, + { + "epoch": 0.07, + "learning_rate": 5.7142857142857145e-06, + "loss": 0.733, + "step": 164 + }, + { + "epoch": 0.07, + "learning_rate": 5.749128919860628e-06, + "loss": 0.6642, + "step": 165 + }, + { + "epoch": 0.07, + "learning_rate": 5.7839721254355405e-06, + "loss": 0.6861, + "step": 166 + }, + { + "epoch": 0.07, + "learning_rate": 5.818815331010453e-06, + "loss": 0.623, + "step": 167 + }, + { + "epoch": 0.07, + "learning_rate": 5.853658536585366e-06, + "loss": 0.5951, + "step": 168 + }, + { + "epoch": 0.07, + "learning_rate": 5.88850174216028e-06, + "loss": 0.6887, + "step": 169 + }, + { + "epoch": 0.07, + "learning_rate": 5.923344947735193e-06, + "loss": 0.5783, + "step": 170 + }, + { + "epoch": 0.07, + "learning_rate": 5.958188153310105e-06, + "loss": 0.6036, + "step": 171 + }, + { + "epoch": 0.07, + "learning_rate": 5.993031358885018e-06, + "loss": 0.6031, + "step": 172 + }, + { + "epoch": 0.07, + "learning_rate": 6.027874564459931e-06, + "loss": 0.659, + "step": 173 + }, + { + "epoch": 0.07, + "learning_rate": 6.062717770034844e-06, + "loss": 0.6425, + "step": 174 + }, + { + "epoch": 0.07, + "learning_rate": 6.0975609756097564e-06, + "loss": 0.5897, + "step": 175 + }, + { + "epoch": 0.07, + "learning_rate": 6.132404181184669e-06, + "loss": 0.6508, + "step": 176 + }, + { + "epoch": 0.07, + "learning_rate": 6.1672473867595825e-06, + "loss": 0.6807, + "step": 177 + }, + { + "epoch": 0.07, + "learning_rate": 6.202090592334495e-06, + "loss": 0.5439, + "step": 178 + }, + { + "epoch": 0.07, + "learning_rate": 6.236933797909408e-06, + "loss": 0.5265, + "step": 179 + }, + { + "epoch": 0.08, + "learning_rate": 6.27177700348432e-06, + "loss": 0.6912, + "step": 180 + }, + { + "epoch": 0.08, + "learning_rate": 6.3066202090592345e-06, + "loss": 0.6348, + "step": 181 + }, + { + "epoch": 0.08, + "learning_rate": 6.341463414634147e-06, + "loss": 0.6374, + "step": 182 + }, + { + "epoch": 0.08, + "learning_rate": 6.37630662020906e-06, + "loss": 0.626, + "step": 183 + }, + { + "epoch": 0.08, + "learning_rate": 6.411149825783972e-06, + "loss": 0.6239, + "step": 184 + }, + { + "epoch": 0.08, + "learning_rate": 6.445993031358886e-06, + "loss": 0.6928, + "step": 185 + }, + { + "epoch": 0.08, + "learning_rate": 6.480836236933798e-06, + "loss": 0.6335, + "step": 186 + }, + { + "epoch": 0.08, + "learning_rate": 6.515679442508711e-06, + "loss": 0.5033, + "step": 187 + }, + { + "epoch": 0.08, + "learning_rate": 6.5505226480836235e-06, + "loss": 0.6619, + "step": 188 + }, + { + "epoch": 0.08, + "learning_rate": 6.585365853658538e-06, + "loss": 0.6189, + "step": 189 + }, + { + "epoch": 0.08, + "learning_rate": 6.62020905923345e-06, + "loss": 0.6277, + "step": 190 + }, + { + "epoch": 0.08, + "learning_rate": 6.655052264808363e-06, + "loss": 0.6073, + "step": 191 + }, + { + "epoch": 0.08, + "learning_rate": 6.6898954703832765e-06, + "loss": 0.6097, + "step": 192 + }, + { + "epoch": 0.08, + "learning_rate": 6.724738675958189e-06, + "loss": 0.5943, + "step": 193 + }, + { + "epoch": 0.08, + "learning_rate": 6.759581881533102e-06, + "loss": 0.5979, + "step": 194 + }, + { + "epoch": 0.08, + "learning_rate": 6.794425087108014e-06, + "loss": 0.598, + "step": 195 + }, + { + "epoch": 0.08, + "learning_rate": 6.829268292682928e-06, + "loss": 0.6136, + "step": 196 + }, + { + "epoch": 0.08, + "learning_rate": 6.86411149825784e-06, + "loss": 0.6153, + "step": 197 + }, + { + "epoch": 0.08, + "learning_rate": 6.898954703832753e-06, + "loss": 0.6817, + "step": 198 + }, + { + "epoch": 0.08, + "learning_rate": 6.9337979094076655e-06, + "loss": 0.5501, + "step": 199 + }, + { + "epoch": 0.08, + "learning_rate": 6.96864111498258e-06, + "loss": 0.7031, + "step": 200 + }, + { + "epoch": 0.08, + "learning_rate": 7.003484320557492e-06, + "loss": 0.5989, + "step": 201 + }, + { + "epoch": 0.08, + "learning_rate": 7.038327526132405e-06, + "loss": 0.6308, + "step": 202 + }, + { + "epoch": 0.08, + "learning_rate": 7.0731707317073175e-06, + "loss": 0.5342, + "step": 203 + }, + { + "epoch": 0.09, + "learning_rate": 7.108013937282231e-06, + "loss": 0.6175, + "step": 204 + }, + { + "epoch": 0.09, + "learning_rate": 7.1428571428571436e-06, + "loss": 0.5792, + "step": 205 + }, + { + "epoch": 0.09, + "learning_rate": 7.177700348432056e-06, + "loss": 0.6591, + "step": 206 + }, + { + "epoch": 0.09, + "learning_rate": 7.212543554006969e-06, + "loss": 0.6793, + "step": 207 + }, + { + "epoch": 0.09, + "learning_rate": 7.247386759581882e-06, + "loss": 0.6891, + "step": 208 + }, + { + "epoch": 0.09, + "learning_rate": 7.282229965156795e-06, + "loss": 0.5279, + "step": 209 + }, + { + "epoch": 0.09, + "learning_rate": 7.317073170731707e-06, + "loss": 0.6229, + "step": 210 + }, + { + "epoch": 0.09, + "learning_rate": 7.35191637630662e-06, + "loss": 0.6168, + "step": 211 + }, + { + "epoch": 0.09, + "learning_rate": 7.386759581881534e-06, + "loss": 0.4843, + "step": 212 + }, + { + "epoch": 0.09, + "learning_rate": 7.421602787456447e-06, + "loss": 0.5995, + "step": 213 + }, + { + "epoch": 0.09, + "learning_rate": 7.4564459930313594e-06, + "loss": 0.5968, + "step": 214 + }, + { + "epoch": 0.09, + "learning_rate": 7.491289198606272e-06, + "loss": 0.5597, + "step": 215 + }, + { + "epoch": 0.09, + "learning_rate": 7.5261324041811855e-06, + "loss": 0.6491, + "step": 216 + }, + { + "epoch": 0.09, + "learning_rate": 7.560975609756098e-06, + "loss": 0.643, + "step": 217 + }, + { + "epoch": 0.09, + "learning_rate": 7.595818815331011e-06, + "loss": 0.6927, + "step": 218 + }, + { + "epoch": 0.09, + "learning_rate": 7.630662020905924e-06, + "loss": 0.5916, + "step": 219 + }, + { + "epoch": 0.09, + "learning_rate": 7.665505226480837e-06, + "loss": 0.5445, + "step": 220 + }, + { + "epoch": 0.09, + "learning_rate": 7.70034843205575e-06, + "loss": 0.6517, + "step": 221 + }, + { + "epoch": 0.09, + "learning_rate": 7.735191637630662e-06, + "loss": 0.6, + "step": 222 + }, + { + "epoch": 0.09, + "learning_rate": 7.770034843205574e-06, + "loss": 0.6441, + "step": 223 + }, + { + "epoch": 0.09, + "learning_rate": 7.804878048780489e-06, + "loss": 0.6266, + "step": 224 + }, + { + "epoch": 0.09, + "learning_rate": 7.839721254355401e-06, + "loss": 0.5312, + "step": 225 + }, + { + "epoch": 0.09, + "learning_rate": 7.874564459930314e-06, + "loss": 0.614, + "step": 226 + }, + { + "epoch": 0.1, + "learning_rate": 7.909407665505228e-06, + "loss": 0.6742, + "step": 227 + }, + { + "epoch": 0.1, + "learning_rate": 7.94425087108014e-06, + "loss": 0.588, + "step": 228 + }, + { + "epoch": 0.1, + "learning_rate": 7.979094076655053e-06, + "loss": 0.5782, + "step": 229 + }, + { + "epoch": 0.1, + "learning_rate": 8.013937282229966e-06, + "loss": 0.6051, + "step": 230 + }, + { + "epoch": 0.1, + "learning_rate": 8.048780487804879e-06, + "loss": 0.671, + "step": 231 + }, + { + "epoch": 0.1, + "learning_rate": 8.083623693379791e-06, + "loss": 0.6003, + "step": 232 + }, + { + "epoch": 0.1, + "learning_rate": 8.118466898954704e-06, + "loss": 0.6023, + "step": 233 + }, + { + "epoch": 0.1, + "learning_rate": 8.153310104529616e-06, + "loss": 0.6305, + "step": 234 + }, + { + "epoch": 0.1, + "learning_rate": 8.18815331010453e-06, + "loss": 0.5449, + "step": 235 + }, + { + "epoch": 0.1, + "learning_rate": 8.222996515679443e-06, + "loss": 0.5957, + "step": 236 + }, + { + "epoch": 0.1, + "learning_rate": 8.257839721254356e-06, + "loss": 0.6335, + "step": 237 + }, + { + "epoch": 0.1, + "learning_rate": 8.292682926829268e-06, + "loss": 0.6214, + "step": 238 + }, + { + "epoch": 0.1, + "learning_rate": 8.327526132404183e-06, + "loss": 0.6096, + "step": 239 + }, + { + "epoch": 0.1, + "learning_rate": 8.362369337979095e-06, + "loss": 0.55, + "step": 240 + }, + { + "epoch": 0.1, + "learning_rate": 8.397212543554008e-06, + "loss": 0.641, + "step": 241 + }, + { + "epoch": 0.1, + "learning_rate": 8.43205574912892e-06, + "loss": 0.6294, + "step": 242 + }, + { + "epoch": 0.1, + "learning_rate": 8.466898954703833e-06, + "loss": 0.6216, + "step": 243 + }, + { + "epoch": 0.1, + "learning_rate": 8.501742160278746e-06, + "loss": 0.5949, + "step": 244 + }, + { + "epoch": 0.1, + "learning_rate": 8.536585365853658e-06, + "loss": 0.568, + "step": 245 + }, + { + "epoch": 0.1, + "learning_rate": 8.571428571428571e-06, + "loss": 0.6762, + "step": 246 + }, + { + "epoch": 0.1, + "learning_rate": 8.606271777003485e-06, + "loss": 0.6068, + "step": 247 + }, + { + "epoch": 0.1, + "learning_rate": 8.641114982578398e-06, + "loss": 0.584, + "step": 248 + }, + { + "epoch": 0.1, + "learning_rate": 8.67595818815331e-06, + "loss": 0.5967, + "step": 249 + }, + { + "epoch": 0.1, + "learning_rate": 8.710801393728223e-06, + "loss": 0.5988, + "step": 250 + }, + { + "epoch": 0.11, + "learning_rate": 8.745644599303137e-06, + "loss": 0.611, + "step": 251 + }, + { + "epoch": 0.11, + "learning_rate": 8.78048780487805e-06, + "loss": 0.6672, + "step": 252 + }, + { + "epoch": 0.11, + "learning_rate": 8.815331010452962e-06, + "loss": 0.5487, + "step": 253 + }, + { + "epoch": 0.11, + "learning_rate": 8.850174216027875e-06, + "loss": 0.5397, + "step": 254 + }, + { + "epoch": 0.11, + "learning_rate": 8.885017421602788e-06, + "loss": 0.7101, + "step": 255 + }, + { + "epoch": 0.11, + "learning_rate": 8.9198606271777e-06, + "loss": 0.6165, + "step": 256 + }, + { + "epoch": 0.11, + "learning_rate": 8.954703832752613e-06, + "loss": 0.5816, + "step": 257 + }, + { + "epoch": 0.11, + "learning_rate": 8.989547038327527e-06, + "loss": 0.6, + "step": 258 + }, + { + "epoch": 0.11, + "learning_rate": 9.02439024390244e-06, + "loss": 0.5065, + "step": 259 + }, + { + "epoch": 0.11, + "learning_rate": 9.059233449477352e-06, + "loss": 0.7077, + "step": 260 + }, + { + "epoch": 0.11, + "learning_rate": 9.094076655052265e-06, + "loss": 0.7149, + "step": 261 + }, + { + "epoch": 0.11, + "learning_rate": 9.12891986062718e-06, + "loss": 0.5832, + "step": 262 + }, + { + "epoch": 0.11, + "learning_rate": 9.163763066202092e-06, + "loss": 0.5587, + "step": 263 + }, + { + "epoch": 0.11, + "learning_rate": 9.198606271777004e-06, + "loss": 0.5781, + "step": 264 + }, + { + "epoch": 0.11, + "learning_rate": 9.233449477351917e-06, + "loss": 0.5852, + "step": 265 + }, + { + "epoch": 0.11, + "learning_rate": 9.268292682926831e-06, + "loss": 0.6214, + "step": 266 + }, + { + "epoch": 0.11, + "learning_rate": 9.303135888501744e-06, + "loss": 0.4923, + "step": 267 + }, + { + "epoch": 0.11, + "learning_rate": 9.337979094076656e-06, + "loss": 0.5257, + "step": 268 + }, + { + "epoch": 0.11, + "learning_rate": 9.372822299651569e-06, + "loss": 0.6273, + "step": 269 + }, + { + "epoch": 0.11, + "learning_rate": 9.407665505226482e-06, + "loss": 0.638, + "step": 270 + }, + { + "epoch": 0.11, + "learning_rate": 9.442508710801394e-06, + "loss": 0.5369, + "step": 271 + }, + { + "epoch": 0.11, + "learning_rate": 9.477351916376307e-06, + "loss": 0.6128, + "step": 272 + }, + { + "epoch": 0.11, + "learning_rate": 9.51219512195122e-06, + "loss": 0.61, + "step": 273 + }, + { + "epoch": 0.11, + "learning_rate": 9.547038327526134e-06, + "loss": 0.6117, + "step": 274 + }, + { + "epoch": 0.12, + "learning_rate": 9.581881533101046e-06, + "loss": 0.5191, + "step": 275 + }, + { + "epoch": 0.12, + "learning_rate": 9.616724738675959e-06, + "loss": 0.5381, + "step": 276 + }, + { + "epoch": 0.12, + "learning_rate": 9.651567944250871e-06, + "loss": 0.6686, + "step": 277 + }, + { + "epoch": 0.12, + "learning_rate": 9.686411149825786e-06, + "loss": 0.6415, + "step": 278 + }, + { + "epoch": 0.12, + "learning_rate": 9.721254355400698e-06, + "loss": 0.5333, + "step": 279 + }, + { + "epoch": 0.12, + "learning_rate": 9.756097560975611e-06, + "loss": 0.6254, + "step": 280 + }, + { + "epoch": 0.12, + "learning_rate": 9.790940766550524e-06, + "loss": 0.5899, + "step": 281 + }, + { + "epoch": 0.12, + "learning_rate": 9.825783972125436e-06, + "loss": 0.5635, + "step": 282 + }, + { + "epoch": 0.12, + "learning_rate": 9.860627177700349e-06, + "loss": 0.5338, + "step": 283 + }, + { + "epoch": 0.12, + "learning_rate": 9.895470383275261e-06, + "loss": 0.6047, + "step": 284 + }, + { + "epoch": 0.12, + "learning_rate": 9.930313588850174e-06, + "loss": 0.6124, + "step": 285 + }, + { + "epoch": 0.12, + "learning_rate": 9.965156794425088e-06, + "loss": 0.5827, + "step": 286 + }, + { + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.52, + "step": 287 + }, + { + "epoch": 0.12, + "learning_rate": 1.0034843205574913e-05, + "loss": 0.6033, + "step": 288 + }, + { + "epoch": 0.12, + "learning_rate": 1.0069686411149826e-05, + "loss": 0.5744, + "step": 289 + }, + { + "epoch": 0.12, + "learning_rate": 1.0104529616724739e-05, + "loss": 0.6823, + "step": 290 + }, + { + "epoch": 0.12, + "learning_rate": 1.0139372822299653e-05, + "loss": 0.6204, + "step": 291 + }, + { + "epoch": 0.12, + "learning_rate": 1.0174216027874565e-05, + "loss": 0.5409, + "step": 292 + }, + { + "epoch": 0.12, + "learning_rate": 1.0209059233449478e-05, + "loss": 0.5721, + "step": 293 + }, + { + "epoch": 0.12, + "learning_rate": 1.024390243902439e-05, + "loss": 0.6182, + "step": 294 + }, + { + "epoch": 0.12, + "learning_rate": 1.0278745644599303e-05, + "loss": 0.579, + "step": 295 + }, + { + "epoch": 0.12, + "learning_rate": 1.0313588850174216e-05, + "loss": 0.6389, + "step": 296 + }, + { + "epoch": 0.12, + "learning_rate": 1.0348432055749128e-05, + "loss": 0.6308, + "step": 297 + }, + { + "epoch": 0.12, + "learning_rate": 1.0383275261324041e-05, + "loss": 0.5235, + "step": 298 + }, + { + "epoch": 0.13, + "learning_rate": 1.0418118466898957e-05, + "loss": 0.5905, + "step": 299 + }, + { + "epoch": 0.13, + "learning_rate": 1.045296167247387e-05, + "loss": 0.6244, + "step": 300 + }, + { + "epoch": 0.13, + "learning_rate": 1.0487804878048782e-05, + "loss": 0.692, + "step": 301 + }, + { + "epoch": 0.13, + "learning_rate": 1.0522648083623695e-05, + "loss": 0.5482, + "step": 302 + }, + { + "epoch": 0.13, + "learning_rate": 1.0557491289198607e-05, + "loss": 0.6385, + "step": 303 + }, + { + "epoch": 0.13, + "learning_rate": 1.059233449477352e-05, + "loss": 0.643, + "step": 304 + }, + { + "epoch": 0.13, + "learning_rate": 1.0627177700348433e-05, + "loss": 0.5917, + "step": 305 + }, + { + "epoch": 0.13, + "learning_rate": 1.0662020905923345e-05, + "loss": 0.6183, + "step": 306 + }, + { + "epoch": 0.13, + "learning_rate": 1.069686411149826e-05, + "loss": 0.557, + "step": 307 + }, + { + "epoch": 0.13, + "learning_rate": 1.0731707317073172e-05, + "loss": 0.6811, + "step": 308 + }, + { + "epoch": 0.13, + "learning_rate": 1.0766550522648085e-05, + "loss": 0.6266, + "step": 309 + }, + { + "epoch": 0.13, + "learning_rate": 1.0801393728222997e-05, + "loss": 0.5146, + "step": 310 + }, + { + "epoch": 0.13, + "learning_rate": 1.083623693379791e-05, + "loss": 0.6395, + "step": 311 + }, + { + "epoch": 0.13, + "learning_rate": 1.0871080139372822e-05, + "loss": 0.6394, + "step": 312 + }, + { + "epoch": 0.13, + "learning_rate": 1.0905923344947735e-05, + "loss": 0.5787, + "step": 313 + }, + { + "epoch": 0.13, + "learning_rate": 1.0940766550522648e-05, + "loss": 0.5919, + "step": 314 + }, + { + "epoch": 0.13, + "learning_rate": 1.0975609756097562e-05, + "loss": 0.5283, + "step": 315 + }, + { + "epoch": 0.13, + "learning_rate": 1.1010452961672475e-05, + "loss": 0.5572, + "step": 316 + }, + { + "epoch": 0.13, + "learning_rate": 1.1045296167247389e-05, + "loss": 0.5774, + "step": 317 + }, + { + "epoch": 0.13, + "learning_rate": 1.1080139372822301e-05, + "loss": 0.5864, + "step": 318 + }, + { + "epoch": 0.13, + "learning_rate": 1.1114982578397214e-05, + "loss": 0.5731, + "step": 319 + }, + { + "epoch": 0.13, + "learning_rate": 1.1149825783972127e-05, + "loss": 0.5627, + "step": 320 + }, + { + "epoch": 0.13, + "learning_rate": 1.118466898954704e-05, + "loss": 0.5659, + "step": 321 + }, + { + "epoch": 0.13, + "learning_rate": 1.1219512195121953e-05, + "loss": 0.6295, + "step": 322 + }, + { + "epoch": 0.14, + "learning_rate": 1.1254355400696866e-05, + "loss": 0.5662, + "step": 323 + }, + { + "epoch": 0.14, + "learning_rate": 1.1289198606271779e-05, + "loss": 0.5659, + "step": 324 + }, + { + "epoch": 0.14, + "learning_rate": 1.1324041811846691e-05, + "loss": 0.74, + "step": 325 + }, + { + "epoch": 0.14, + "learning_rate": 1.1358885017421604e-05, + "loss": 0.604, + "step": 326 + }, + { + "epoch": 0.14, + "learning_rate": 1.1393728222996516e-05, + "loss": 0.5228, + "step": 327 + }, + { + "epoch": 0.14, + "learning_rate": 1.1428571428571429e-05, + "loss": 0.5745, + "step": 328 + }, + { + "epoch": 0.14, + "learning_rate": 1.1463414634146342e-05, + "loss": 0.5464, + "step": 329 + }, + { + "epoch": 0.14, + "learning_rate": 1.1498257839721256e-05, + "loss": 0.5523, + "step": 330 + }, + { + "epoch": 0.14, + "learning_rate": 1.1533101045296168e-05, + "loss": 0.5893, + "step": 331 + }, + { + "epoch": 0.14, + "learning_rate": 1.1567944250871081e-05, + "loss": 0.6265, + "step": 332 + }, + { + "epoch": 0.14, + "learning_rate": 1.1602787456445994e-05, + "loss": 0.6626, + "step": 333 + }, + { + "epoch": 0.14, + "learning_rate": 1.1637630662020906e-05, + "loss": 0.6379, + "step": 334 + }, + { + "epoch": 0.14, + "learning_rate": 1.1672473867595819e-05, + "loss": 0.6768, + "step": 335 + }, + { + "epoch": 0.14, + "learning_rate": 1.1707317073170731e-05, + "loss": 0.5926, + "step": 336 + }, + { + "epoch": 0.14, + "learning_rate": 1.1742160278745644e-05, + "loss": 0.5531, + "step": 337 + }, + { + "epoch": 0.14, + "learning_rate": 1.177700348432056e-05, + "loss": 0.5478, + "step": 338 + }, + { + "epoch": 0.14, + "learning_rate": 1.1811846689895473e-05, + "loss": 0.5894, + "step": 339 + }, + { + "epoch": 0.14, + "learning_rate": 1.1846689895470385e-05, + "loss": 0.5819, + "step": 340 + }, + { + "epoch": 0.14, + "learning_rate": 1.1881533101045298e-05, + "loss": 0.4793, + "step": 341 + }, + { + "epoch": 0.14, + "learning_rate": 1.191637630662021e-05, + "loss": 0.6046, + "step": 342 + }, + { + "epoch": 0.14, + "learning_rate": 1.1951219512195123e-05, + "loss": 0.5306, + "step": 343 + }, + { + "epoch": 0.14, + "learning_rate": 1.1986062717770036e-05, + "loss": 0.6153, + "step": 344 + }, + { + "epoch": 0.14, + "learning_rate": 1.2020905923344948e-05, + "loss": 0.6093, + "step": 345 + }, + { + "epoch": 0.14, + "learning_rate": 1.2055749128919862e-05, + "loss": 0.6022, + "step": 346 + }, + { + "epoch": 0.15, + "learning_rate": 1.2090592334494775e-05, + "loss": 0.515, + "step": 347 + }, + { + "epoch": 0.15, + "learning_rate": 1.2125435540069688e-05, + "loss": 0.6889, + "step": 348 + }, + { + "epoch": 0.15, + "learning_rate": 1.21602787456446e-05, + "loss": 0.576, + "step": 349 + }, + { + "epoch": 0.15, + "learning_rate": 1.2195121951219513e-05, + "loss": 0.5454, + "step": 350 + }, + { + "epoch": 0.15, + "learning_rate": 1.2229965156794425e-05, + "loss": 0.4593, + "step": 351 + }, + { + "epoch": 0.15, + "learning_rate": 1.2264808362369338e-05, + "loss": 0.5302, + "step": 352 + }, + { + "epoch": 0.15, + "learning_rate": 1.2299651567944252e-05, + "loss": 0.5591, + "step": 353 + }, + { + "epoch": 0.15, + "learning_rate": 1.2334494773519165e-05, + "loss": 0.6179, + "step": 354 + }, + { + "epoch": 0.15, + "learning_rate": 1.2369337979094078e-05, + "loss": 0.6015, + "step": 355 + }, + { + "epoch": 0.15, + "learning_rate": 1.240418118466899e-05, + "loss": 0.5615, + "step": 356 + }, + { + "epoch": 0.15, + "learning_rate": 1.2439024390243903e-05, + "loss": 0.6798, + "step": 357 + }, + { + "epoch": 0.15, + "learning_rate": 1.2473867595818815e-05, + "loss": 0.578, + "step": 358 + }, + { + "epoch": 0.15, + "learning_rate": 1.2508710801393728e-05, + "loss": 0.629, + "step": 359 + }, + { + "epoch": 0.15, + "learning_rate": 1.254355400696864e-05, + "loss": 0.6236, + "step": 360 + }, + { + "epoch": 0.15, + "learning_rate": 1.2578397212543556e-05, + "loss": 0.577, + "step": 361 + }, + { + "epoch": 0.15, + "learning_rate": 1.2613240418118469e-05, + "loss": 0.5588, + "step": 362 + }, + { + "epoch": 0.15, + "learning_rate": 1.2648083623693382e-05, + "loss": 0.6655, + "step": 363 + }, + { + "epoch": 0.15, + "learning_rate": 1.2682926829268294e-05, + "loss": 0.6124, + "step": 364 + }, + { + "epoch": 0.15, + "learning_rate": 1.2717770034843207e-05, + "loss": 0.4786, + "step": 365 + }, + { + "epoch": 0.15, + "learning_rate": 1.275261324041812e-05, + "loss": 0.566, + "step": 366 + }, + { + "epoch": 0.15, + "learning_rate": 1.2787456445993032e-05, + "loss": 0.5676, + "step": 367 + }, + { + "epoch": 0.15, + "learning_rate": 1.2822299651567945e-05, + "loss": 0.6284, + "step": 368 + }, + { + "epoch": 0.15, + "learning_rate": 1.2857142857142859e-05, + "loss": 0.5828, + "step": 369 + }, + { + "epoch": 0.15, + "learning_rate": 1.2891986062717772e-05, + "loss": 0.6464, + "step": 370 + }, + { + "epoch": 0.16, + "learning_rate": 1.2926829268292684e-05, + "loss": 0.5833, + "step": 371 + }, + { + "epoch": 0.16, + "learning_rate": 1.2961672473867597e-05, + "loss": 0.578, + "step": 372 + }, + { + "epoch": 0.16, + "learning_rate": 1.299651567944251e-05, + "loss": 0.5904, + "step": 373 + }, + { + "epoch": 0.16, + "learning_rate": 1.3031358885017422e-05, + "loss": 0.5454, + "step": 374 + }, + { + "epoch": 0.16, + "learning_rate": 1.3066202090592334e-05, + "loss": 0.5759, + "step": 375 + }, + { + "epoch": 0.16, + "learning_rate": 1.3101045296167247e-05, + "loss": 0.5288, + "step": 376 + }, + { + "epoch": 0.16, + "learning_rate": 1.3135888501742163e-05, + "loss": 0.6116, + "step": 377 + }, + { + "epoch": 0.16, + "learning_rate": 1.3170731707317076e-05, + "loss": 0.5446, + "step": 378 + }, + { + "epoch": 0.16, + "learning_rate": 1.3205574912891988e-05, + "loss": 0.5642, + "step": 379 + }, + { + "epoch": 0.16, + "learning_rate": 1.32404181184669e-05, + "loss": 0.5638, + "step": 380 + }, + { + "epoch": 0.16, + "learning_rate": 1.3275261324041813e-05, + "loss": 0.779, + "step": 381 + }, + { + "epoch": 0.16, + "learning_rate": 1.3310104529616726e-05, + "loss": 0.5757, + "step": 382 + }, + { + "epoch": 0.16, + "learning_rate": 1.3344947735191639e-05, + "loss": 0.6283, + "step": 383 + }, + { + "epoch": 0.16, + "learning_rate": 1.3379790940766553e-05, + "loss": 0.6289, + "step": 384 + }, + { + "epoch": 0.16, + "learning_rate": 1.3414634146341466e-05, + "loss": 0.5548, + "step": 385 + }, + { + "epoch": 0.16, + "learning_rate": 1.3449477351916378e-05, + "loss": 0.5725, + "step": 386 + }, + { + "epoch": 0.16, + "learning_rate": 1.348432055749129e-05, + "loss": 0.6109, + "step": 387 + }, + { + "epoch": 0.16, + "learning_rate": 1.3519163763066203e-05, + "loss": 0.5975, + "step": 388 + }, + { + "epoch": 0.16, + "learning_rate": 1.3554006968641116e-05, + "loss": 0.5972, + "step": 389 + }, + { + "epoch": 0.16, + "learning_rate": 1.3588850174216028e-05, + "loss": 0.6673, + "step": 390 + }, + { + "epoch": 0.16, + "learning_rate": 1.3623693379790941e-05, + "loss": 0.6687, + "step": 391 + }, + { + "epoch": 0.16, + "learning_rate": 1.3658536585365855e-05, + "loss": 0.504, + "step": 392 + }, + { + "epoch": 0.16, + "learning_rate": 1.3693379790940768e-05, + "loss": 0.6478, + "step": 393 + }, + { + "epoch": 0.16, + "learning_rate": 1.372822299651568e-05, + "loss": 0.6042, + "step": 394 + }, + { + "epoch": 0.17, + "learning_rate": 1.3763066202090593e-05, + "loss": 0.6066, + "step": 395 + }, + { + "epoch": 0.17, + "learning_rate": 1.3797909407665506e-05, + "loss": 0.5724, + "step": 396 + }, + { + "epoch": 0.17, + "learning_rate": 1.3832752613240418e-05, + "loss": 0.6821, + "step": 397 + }, + { + "epoch": 0.17, + "learning_rate": 1.3867595818815331e-05, + "loss": 0.544, + "step": 398 + }, + { + "epoch": 0.17, + "learning_rate": 1.3902439024390244e-05, + "loss": 0.5192, + "step": 399 + }, + { + "epoch": 0.17, + "learning_rate": 1.393728222996516e-05, + "loss": 0.6457, + "step": 400 + }, + { + "epoch": 0.17, + "learning_rate": 1.3972125435540072e-05, + "loss": 0.5996, + "step": 401 + }, + { + "epoch": 0.17, + "learning_rate": 1.4006968641114985e-05, + "loss": 0.5737, + "step": 402 + }, + { + "epoch": 0.17, + "learning_rate": 1.4041811846689897e-05, + "loss": 0.6506, + "step": 403 + }, + { + "epoch": 0.17, + "learning_rate": 1.407665505226481e-05, + "loss": 0.5541, + "step": 404 + }, + { + "epoch": 0.17, + "learning_rate": 1.4111498257839722e-05, + "loss": 0.5502, + "step": 405 + }, + { + "epoch": 0.17, + "learning_rate": 1.4146341463414635e-05, + "loss": 0.5909, + "step": 406 + }, + { + "epoch": 0.17, + "learning_rate": 1.4181184668989548e-05, + "loss": 0.5953, + "step": 407 + }, + { + "epoch": 0.17, + "learning_rate": 1.4216027874564462e-05, + "loss": 0.5376, + "step": 408 + }, + { + "epoch": 0.17, + "learning_rate": 1.4250871080139375e-05, + "loss": 0.5377, + "step": 409 + }, + { + "epoch": 0.17, + "learning_rate": 1.4285714285714287e-05, + "loss": 0.6134, + "step": 410 + }, + { + "epoch": 0.17, + "learning_rate": 1.43205574912892e-05, + "loss": 0.6472, + "step": 411 + }, + { + "epoch": 0.17, + "learning_rate": 1.4355400696864112e-05, + "loss": 0.4815, + "step": 412 + }, + { + "epoch": 0.17, + "learning_rate": 1.4390243902439025e-05, + "loss": 0.4832, + "step": 413 + }, + { + "epoch": 0.17, + "learning_rate": 1.4425087108013938e-05, + "loss": 0.5455, + "step": 414 + }, + { + "epoch": 0.17, + "learning_rate": 1.4459930313588852e-05, + "loss": 0.5445, + "step": 415 + }, + { + "epoch": 0.17, + "learning_rate": 1.4494773519163764e-05, + "loss": 0.5713, + "step": 416 + }, + { + "epoch": 0.17, + "learning_rate": 1.4529616724738677e-05, + "loss": 0.5713, + "step": 417 + }, + { + "epoch": 0.18, + "learning_rate": 1.456445993031359e-05, + "loss": 0.6725, + "step": 418 + }, + { + "epoch": 0.18, + "learning_rate": 1.4599303135888502e-05, + "loss": 0.6199, + "step": 419 + }, + { + "epoch": 0.18, + "learning_rate": 1.4634146341463415e-05, + "loss": 0.5049, + "step": 420 + }, + { + "epoch": 0.18, + "learning_rate": 1.4668989547038327e-05, + "loss": 0.5712, + "step": 421 + }, + { + "epoch": 0.18, + "learning_rate": 1.470383275261324e-05, + "loss": 0.6426, + "step": 422 + }, + { + "epoch": 0.18, + "learning_rate": 1.4738675958188156e-05, + "loss": 0.5879, + "step": 423 + }, + { + "epoch": 0.18, + "learning_rate": 1.4773519163763069e-05, + "loss": 0.5529, + "step": 424 + }, + { + "epoch": 0.18, + "learning_rate": 1.4808362369337981e-05, + "loss": 0.6029, + "step": 425 + }, + { + "epoch": 0.18, + "learning_rate": 1.4843205574912894e-05, + "loss": 0.6285, + "step": 426 + }, + { + "epoch": 0.18, + "learning_rate": 1.4878048780487806e-05, + "loss": 0.5832, + "step": 427 + }, + { + "epoch": 0.18, + "learning_rate": 1.4912891986062719e-05, + "loss": 0.506, + "step": 428 + }, + { + "epoch": 0.18, + "learning_rate": 1.4947735191637631e-05, + "loss": 0.6044, + "step": 429 + }, + { + "epoch": 0.18, + "learning_rate": 1.4982578397212544e-05, + "loss": 0.5128, + "step": 430 + }, + { + "epoch": 0.18, + "learning_rate": 1.5017421602787458e-05, + "loss": 0.5998, + "step": 431 + }, + { + "epoch": 0.18, + "learning_rate": 1.5052264808362371e-05, + "loss": 0.6959, + "step": 432 + }, + { + "epoch": 0.18, + "learning_rate": 1.5087108013937284e-05, + "loss": 0.6206, + "step": 433 + }, + { + "epoch": 0.18, + "learning_rate": 1.5121951219512196e-05, + "loss": 0.6337, + "step": 434 + }, + { + "epoch": 0.18, + "learning_rate": 1.5156794425087109e-05, + "loss": 0.5991, + "step": 435 + }, + { + "epoch": 0.18, + "learning_rate": 1.5191637630662021e-05, + "loss": 0.6201, + "step": 436 + }, + { + "epoch": 0.18, + "learning_rate": 1.5226480836236934e-05, + "loss": 0.4951, + "step": 437 + }, + { + "epoch": 0.18, + "learning_rate": 1.5261324041811848e-05, + "loss": 0.5876, + "step": 438 + }, + { + "epoch": 0.18, + "learning_rate": 1.529616724738676e-05, + "loss": 0.5744, + "step": 439 + }, + { + "epoch": 0.18, + "learning_rate": 1.5331010452961673e-05, + "loss": 0.5939, + "step": 440 + }, + { + "epoch": 0.18, + "learning_rate": 1.5365853658536586e-05, + "loss": 0.6046, + "step": 441 + }, + { + "epoch": 0.19, + "learning_rate": 1.54006968641115e-05, + "loss": 0.656, + "step": 442 + }, + { + "epoch": 0.19, + "learning_rate": 1.543554006968641e-05, + "loss": 0.6, + "step": 443 + }, + { + "epoch": 0.19, + "learning_rate": 1.5470383275261324e-05, + "loss": 0.5162, + "step": 444 + }, + { + "epoch": 0.19, + "learning_rate": 1.5505226480836236e-05, + "loss": 0.5459, + "step": 445 + }, + { + "epoch": 0.19, + "learning_rate": 1.554006968641115e-05, + "loss": 0.486, + "step": 446 + }, + { + "epoch": 0.19, + "learning_rate": 1.5574912891986065e-05, + "loss": 0.6013, + "step": 447 + }, + { + "epoch": 0.19, + "learning_rate": 1.5609756097560978e-05, + "loss": 0.6249, + "step": 448 + }, + { + "epoch": 0.19, + "learning_rate": 1.564459930313589e-05, + "loss": 0.545, + "step": 449 + }, + { + "epoch": 0.19, + "learning_rate": 1.5679442508710803e-05, + "loss": 0.6635, + "step": 450 + }, + { + "epoch": 0.19, + "learning_rate": 1.5714285714285715e-05, + "loss": 0.6335, + "step": 451 + }, + { + "epoch": 0.19, + "learning_rate": 1.5749128919860628e-05, + "loss": 0.5401, + "step": 452 + }, + { + "epoch": 0.19, + "learning_rate": 1.578397212543554e-05, + "loss": 0.5021, + "step": 453 + }, + { + "epoch": 0.19, + "learning_rate": 1.5818815331010456e-05, + "loss": 0.5869, + "step": 454 + }, + { + "epoch": 0.19, + "learning_rate": 1.585365853658537e-05, + "loss": 0.6381, + "step": 455 + }, + { + "epoch": 0.19, + "learning_rate": 1.588850174216028e-05, + "loss": 0.4976, + "step": 456 + }, + { + "epoch": 0.19, + "learning_rate": 1.5923344947735194e-05, + "loss": 0.5761, + "step": 457 + }, + { + "epoch": 0.19, + "learning_rate": 1.5958188153310107e-05, + "loss": 0.6614, + "step": 458 + }, + { + "epoch": 0.19, + "learning_rate": 1.599303135888502e-05, + "loss": 0.6228, + "step": 459 + }, + { + "epoch": 0.19, + "learning_rate": 1.6027874564459932e-05, + "loss": 0.5755, + "step": 460 + }, + { + "epoch": 0.19, + "learning_rate": 1.6062717770034845e-05, + "loss": 0.5943, + "step": 461 + }, + { + "epoch": 0.19, + "learning_rate": 1.6097560975609757e-05, + "loss": 0.5631, + "step": 462 + }, + { + "epoch": 0.19, + "learning_rate": 1.613240418118467e-05, + "loss": 0.6802, + "step": 463 + }, + { + "epoch": 0.19, + "learning_rate": 1.6167247386759582e-05, + "loss": 0.5835, + "step": 464 + }, + { + "epoch": 0.19, + "learning_rate": 1.6202090592334495e-05, + "loss": 0.5725, + "step": 465 + }, + { + "epoch": 0.2, + "learning_rate": 1.6236933797909408e-05, + "loss": 0.5487, + "step": 466 + }, + { + "epoch": 0.2, + "learning_rate": 1.627177700348432e-05, + "loss": 0.5889, + "step": 467 + }, + { + "epoch": 0.2, + "learning_rate": 1.6306620209059233e-05, + "loss": 0.5585, + "step": 468 + }, + { + "epoch": 0.2, + "learning_rate": 1.6341463414634145e-05, + "loss": 0.5731, + "step": 469 + }, + { + "epoch": 0.2, + "learning_rate": 1.637630662020906e-05, + "loss": 0.6129, + "step": 470 + }, + { + "epoch": 0.2, + "learning_rate": 1.6411149825783974e-05, + "loss": 0.586, + "step": 471 + }, + { + "epoch": 0.2, + "learning_rate": 1.6445993031358887e-05, + "loss": 0.6418, + "step": 472 + }, + { + "epoch": 0.2, + "learning_rate": 1.64808362369338e-05, + "loss": 0.6442, + "step": 473 + }, + { + "epoch": 0.2, + "learning_rate": 1.6515679442508712e-05, + "loss": 0.5419, + "step": 474 + }, + { + "epoch": 0.2, + "learning_rate": 1.6550522648083624e-05, + "loss": 0.5282, + "step": 475 + }, + { + "epoch": 0.2, + "learning_rate": 1.6585365853658537e-05, + "loss": 0.5588, + "step": 476 + }, + { + "epoch": 0.2, + "learning_rate": 1.662020905923345e-05, + "loss": 0.5536, + "step": 477 + }, + { + "epoch": 0.2, + "learning_rate": 1.6655052264808366e-05, + "loss": 0.5514, + "step": 478 + }, + { + "epoch": 0.2, + "learning_rate": 1.6689895470383278e-05, + "loss": 0.5554, + "step": 479 + }, + { + "epoch": 0.2, + "learning_rate": 1.672473867595819e-05, + "loss": 0.555, + "step": 480 + }, + { + "epoch": 0.2, + "learning_rate": 1.6759581881533103e-05, + "loss": 0.5606, + "step": 481 + }, + { + "epoch": 0.2, + "learning_rate": 1.6794425087108016e-05, + "loss": 0.5154, + "step": 482 + }, + { + "epoch": 0.2, + "learning_rate": 1.682926829268293e-05, + "loss": 0.5407, + "step": 483 + }, + { + "epoch": 0.2, + "learning_rate": 1.686411149825784e-05, + "loss": 0.6047, + "step": 484 + }, + { + "epoch": 0.2, + "learning_rate": 1.6898954703832754e-05, + "loss": 0.5323, + "step": 485 + }, + { + "epoch": 0.2, + "learning_rate": 1.6933797909407666e-05, + "loss": 0.5395, + "step": 486 + }, + { + "epoch": 0.2, + "learning_rate": 1.696864111498258e-05, + "loss": 0.6181, + "step": 487 + }, + { + "epoch": 0.2, + "learning_rate": 1.700348432055749e-05, + "loss": 0.5764, + "step": 488 + }, + { + "epoch": 0.2, + "learning_rate": 1.7038327526132404e-05, + "loss": 0.6155, + "step": 489 + }, + { + "epoch": 0.21, + "learning_rate": 1.7073170731707317e-05, + "loss": 0.5058, + "step": 490 + }, + { + "epoch": 0.21, + "learning_rate": 1.710801393728223e-05, + "loss": 0.6552, + "step": 491 + }, + { + "epoch": 0.21, + "learning_rate": 1.7142857142857142e-05, + "loss": 0.5362, + "step": 492 + }, + { + "epoch": 0.21, + "learning_rate": 1.7177700348432058e-05, + "loss": 0.6479, + "step": 493 + }, + { + "epoch": 0.21, + "learning_rate": 1.721254355400697e-05, + "loss": 0.4939, + "step": 494 + }, + { + "epoch": 0.21, + "learning_rate": 1.7247386759581883e-05, + "loss": 0.6442, + "step": 495 + }, + { + "epoch": 0.21, + "learning_rate": 1.7282229965156796e-05, + "loss": 0.5753, + "step": 496 + }, + { + "epoch": 0.21, + "learning_rate": 1.7317073170731708e-05, + "loss": 0.701, + "step": 497 + }, + { + "epoch": 0.21, + "learning_rate": 1.735191637630662e-05, + "loss": 0.5235, + "step": 498 + }, + { + "epoch": 0.21, + "learning_rate": 1.7386759581881533e-05, + "loss": 0.4857, + "step": 499 + }, + { + "epoch": 0.21, + "learning_rate": 1.7421602787456446e-05, + "loss": 0.5072, + "step": 500 + }, + { + "epoch": 0.21, + "learning_rate": 1.7456445993031362e-05, + "loss": 0.6019, + "step": 501 + }, + { + "epoch": 0.21, + "learning_rate": 1.7491289198606275e-05, + "loss": 0.5263, + "step": 502 + }, + { + "epoch": 0.21, + "learning_rate": 1.7526132404181187e-05, + "loss": 0.5893, + "step": 503 + }, + { + "epoch": 0.21, + "learning_rate": 1.75609756097561e-05, + "loss": 0.6769, + "step": 504 + }, + { + "epoch": 0.21, + "learning_rate": 1.7595818815331012e-05, + "loss": 0.5069, + "step": 505 + }, + { + "epoch": 0.21, + "learning_rate": 1.7630662020905925e-05, + "loss": 0.6241, + "step": 506 + }, + { + "epoch": 0.21, + "learning_rate": 1.7665505226480838e-05, + "loss": 0.5585, + "step": 507 + }, + { + "epoch": 0.21, + "learning_rate": 1.770034843205575e-05, + "loss": 0.5546, + "step": 508 + }, + { + "epoch": 0.21, + "learning_rate": 1.7735191637630663e-05, + "loss": 0.5824, + "step": 509 + }, + { + "epoch": 0.21, + "learning_rate": 1.7770034843205575e-05, + "loss": 0.5741, + "step": 510 + }, + { + "epoch": 0.21, + "learning_rate": 1.7804878048780488e-05, + "loss": 0.5867, + "step": 511 + }, + { + "epoch": 0.21, + "learning_rate": 1.78397212543554e-05, + "loss": 0.5562, + "step": 512 + }, + { + "epoch": 0.21, + "learning_rate": 1.7874564459930313e-05, + "loss": 0.5487, + "step": 513 + }, + { + "epoch": 0.22, + "learning_rate": 1.7909407665505226e-05, + "loss": 0.6476, + "step": 514 + }, + { + "epoch": 0.22, + "learning_rate": 1.7944250871080138e-05, + "loss": 0.5481, + "step": 515 + }, + { + "epoch": 0.22, + "learning_rate": 1.7979094076655054e-05, + "loss": 0.5381, + "step": 516 + }, + { + "epoch": 0.22, + "learning_rate": 1.8013937282229967e-05, + "loss": 0.5591, + "step": 517 + }, + { + "epoch": 0.22, + "learning_rate": 1.804878048780488e-05, + "loss": 0.591, + "step": 518 + }, + { + "epoch": 0.22, + "learning_rate": 1.8083623693379792e-05, + "loss": 0.465, + "step": 519 + }, + { + "epoch": 0.22, + "learning_rate": 1.8118466898954705e-05, + "loss": 0.6022, + "step": 520 + }, + { + "epoch": 0.22, + "learning_rate": 1.8153310104529617e-05, + "loss": 0.4784, + "step": 521 + }, + { + "epoch": 0.22, + "learning_rate": 1.818815331010453e-05, + "loss": 0.5983, + "step": 522 + }, + { + "epoch": 0.22, + "learning_rate": 1.8222996515679442e-05, + "loss": 0.5676, + "step": 523 + }, + { + "epoch": 0.22, + "learning_rate": 1.825783972125436e-05, + "loss": 0.5733, + "step": 524 + }, + { + "epoch": 0.22, + "learning_rate": 1.829268292682927e-05, + "loss": 0.5837, + "step": 525 + }, + { + "epoch": 0.22, + "learning_rate": 1.8327526132404184e-05, + "loss": 0.5893, + "step": 526 + }, + { + "epoch": 0.22, + "learning_rate": 1.8362369337979096e-05, + "loss": 0.6748, + "step": 527 + }, + { + "epoch": 0.22, + "learning_rate": 1.839721254355401e-05, + "loss": 0.5764, + "step": 528 + }, + { + "epoch": 0.22, + "learning_rate": 1.843205574912892e-05, + "loss": 0.6803, + "step": 529 + }, + { + "epoch": 0.22, + "learning_rate": 1.8466898954703834e-05, + "loss": 0.5717, + "step": 530 + }, + { + "epoch": 0.22, + "learning_rate": 1.8501742160278747e-05, + "loss": 0.5755, + "step": 531 + }, + { + "epoch": 0.22, + "learning_rate": 1.8536585365853663e-05, + "loss": 0.5581, + "step": 532 + }, + { + "epoch": 0.22, + "learning_rate": 1.8571428571428575e-05, + "loss": 0.5884, + "step": 533 + }, + { + "epoch": 0.22, + "learning_rate": 1.8606271777003488e-05, + "loss": 0.5363, + "step": 534 + }, + { + "epoch": 0.22, + "learning_rate": 1.86411149825784e-05, + "loss": 0.5798, + "step": 535 + }, + { + "epoch": 0.22, + "learning_rate": 1.8675958188153313e-05, + "loss": 0.5236, + "step": 536 + }, + { + "epoch": 0.22, + "learning_rate": 1.8710801393728226e-05, + "loss": 0.5025, + "step": 537 + }, + { + "epoch": 0.23, + "learning_rate": 1.8745644599303138e-05, + "loss": 0.6242, + "step": 538 + }, + { + "epoch": 0.23, + "learning_rate": 1.878048780487805e-05, + "loss": 0.6023, + "step": 539 + }, + { + "epoch": 0.23, + "learning_rate": 1.8815331010452963e-05, + "loss": 0.6211, + "step": 540 + }, + { + "epoch": 0.23, + "learning_rate": 1.8850174216027876e-05, + "loss": 0.5077, + "step": 541 + }, + { + "epoch": 0.23, + "learning_rate": 1.888501742160279e-05, + "loss": 0.5018, + "step": 542 + }, + { + "epoch": 0.23, + "learning_rate": 1.89198606271777e-05, + "loss": 0.6053, + "step": 543 + }, + { + "epoch": 0.23, + "learning_rate": 1.8954703832752614e-05, + "loss": 0.6225, + "step": 544 + }, + { + "epoch": 0.23, + "learning_rate": 1.8989547038327526e-05, + "loss": 0.6134, + "step": 545 + }, + { + "epoch": 0.23, + "learning_rate": 1.902439024390244e-05, + "loss": 0.5871, + "step": 546 + }, + { + "epoch": 0.23, + "learning_rate": 1.9059233449477355e-05, + "loss": 0.7059, + "step": 547 + }, + { + "epoch": 0.23, + "learning_rate": 1.9094076655052267e-05, + "loss": 0.5859, + "step": 548 + }, + { + "epoch": 0.23, + "learning_rate": 1.912891986062718e-05, + "loss": 0.5957, + "step": 549 + }, + { + "epoch": 0.23, + "learning_rate": 1.9163763066202093e-05, + "loss": 0.5299, + "step": 550 + }, + { + "epoch": 0.23, + "learning_rate": 1.9198606271777005e-05, + "loss": 0.6158, + "step": 551 + }, + { + "epoch": 0.23, + "learning_rate": 1.9233449477351918e-05, + "loss": 0.5808, + "step": 552 + }, + { + "epoch": 0.23, + "learning_rate": 1.926829268292683e-05, + "loss": 0.573, + "step": 553 + }, + { + "epoch": 0.23, + "learning_rate": 1.9303135888501743e-05, + "loss": 0.6362, + "step": 554 + }, + { + "epoch": 0.23, + "learning_rate": 1.933797909407666e-05, + "loss": 0.5059, + "step": 555 + }, + { + "epoch": 0.23, + "learning_rate": 1.937282229965157e-05, + "loss": 0.5784, + "step": 556 + }, + { + "epoch": 0.23, + "learning_rate": 1.9407665505226484e-05, + "loss": 0.649, + "step": 557 + }, + { + "epoch": 0.23, + "learning_rate": 1.9442508710801397e-05, + "loss": 0.6128, + "step": 558 + }, + { + "epoch": 0.23, + "learning_rate": 1.947735191637631e-05, + "loss": 0.595, + "step": 559 + }, + { + "epoch": 0.23, + "learning_rate": 1.9512195121951222e-05, + "loss": 0.5921, + "step": 560 + }, + { + "epoch": 0.23, + "learning_rate": 1.9547038327526135e-05, + "loss": 0.5053, + "step": 561 + }, + { + "epoch": 0.24, + "learning_rate": 1.9581881533101047e-05, + "loss": 0.4499, + "step": 562 + }, + { + "epoch": 0.24, + "learning_rate": 1.961672473867596e-05, + "loss": 0.5918, + "step": 563 + }, + { + "epoch": 0.24, + "learning_rate": 1.9651567944250872e-05, + "loss": 0.5511, + "step": 564 + }, + { + "epoch": 0.24, + "learning_rate": 1.9686411149825785e-05, + "loss": 0.6467, + "step": 565 + }, + { + "epoch": 0.24, + "learning_rate": 1.9721254355400697e-05, + "loss": 0.5508, + "step": 566 + }, + { + "epoch": 0.24, + "learning_rate": 1.975609756097561e-05, + "loss": 0.5592, + "step": 567 + }, + { + "epoch": 0.24, + "learning_rate": 1.9790940766550523e-05, + "loss": 0.5966, + "step": 568 + }, + { + "epoch": 0.24, + "learning_rate": 1.9825783972125435e-05, + "loss": 0.5011, + "step": 569 + }, + { + "epoch": 0.24, + "learning_rate": 1.9860627177700348e-05, + "loss": 0.593, + "step": 570 + }, + { + "epoch": 0.24, + "learning_rate": 1.9895470383275264e-05, + "loss": 0.5659, + "step": 571 + }, + { + "epoch": 0.24, + "learning_rate": 1.9930313588850176e-05, + "loss": 0.5767, + "step": 572 + }, + { + "epoch": 0.24, + "learning_rate": 1.996515679442509e-05, + "loss": 0.5624, + "step": 573 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.4904, + "step": 574 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999985627944e-05, + "loss": 0.5734, + "step": 575 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999999425117754e-05, + "loss": 0.5985, + "step": 576 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999998706514958e-05, + "loss": 0.5729, + "step": 577 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999997700471074e-05, + "loss": 0.558, + "step": 578 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999640698613e-05, + "loss": 0.5446, + "step": 579 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999994826060166e-05, + "loss": 0.5934, + "step": 580 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999295769322e-05, + "loss": 0.6511, + "step": 581 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999990801885353e-05, + "loss": 0.5262, + "step": 582 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999988358636622e-05, + "loss": 0.5777, + "step": 583 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999985627947104e-05, + "loss": 0.4609, + "step": 584 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999982609816867e-05, + "loss": 0.6021, + "step": 585 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999979304246006e-05, + "loss": 0.6274, + "step": 586 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999975711234616e-05, + "loss": 0.517, + "step": 587 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999971830782795e-05, + "loss": 0.562, + "step": 588 + }, + { + "epoch": 0.25, + "learning_rate": 1.999996766289066e-05, + "loss": 0.5583, + "step": 589 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999963207558326e-05, + "loss": 0.5975, + "step": 590 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999958464785925e-05, + "loss": 0.6094, + "step": 591 + }, + { + "epoch": 0.25, + "learning_rate": 1.999995343457359e-05, + "loss": 0.5579, + "step": 592 + }, + { + "epoch": 0.25, + "learning_rate": 1.999994811692147e-05, + "loss": 0.551, + "step": 593 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999942511829714e-05, + "loss": 0.5576, + "step": 594 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999936619298483e-05, + "loss": 0.5486, + "step": 595 + }, + { + "epoch": 0.25, + "learning_rate": 1.999993043932795e-05, + "loss": 0.588, + "step": 596 + }, + { + "epoch": 0.25, + "learning_rate": 1.999992397191829e-05, + "loss": 0.5231, + "step": 597 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999917217069687e-05, + "loss": 0.5773, + "step": 598 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999910174782337e-05, + "loss": 0.5609, + "step": 599 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999902845056444e-05, + "loss": 0.6032, + "step": 600 + }, + { + "epoch": 0.25, + "learning_rate": 1.999989522789222e-05, + "loss": 0.5846, + "step": 601 + }, + { + "epoch": 0.25, + "learning_rate": 1.999988732328988e-05, + "loss": 0.6105, + "step": 602 + }, + { + "epoch": 0.25, + "learning_rate": 1.999987913124965e-05, + "loss": 0.5158, + "step": 603 + }, + { + "epoch": 0.25, + "learning_rate": 1.999987065177177e-05, + "loss": 0.6727, + "step": 604 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999861884856483e-05, + "loss": 0.5327, + "step": 605 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999852830504038e-05, + "loss": 0.5825, + "step": 606 + }, + { + "epoch": 0.25, + "learning_rate": 1.99998434887147e-05, + "loss": 0.5495, + "step": 607 + }, + { + "epoch": 0.25, + "learning_rate": 1.999983385948873e-05, + "loss": 0.542, + "step": 608 + }, + { + "epoch": 0.25, + "learning_rate": 1.999982394282641e-05, + "loss": 0.6037, + "step": 609 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999813738728028e-05, + "loss": 0.6024, + "step": 610 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999803247193873e-05, + "loss": 0.6103, + "step": 611 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999792468224247e-05, + "loss": 0.6136, + "step": 612 + }, + { + "epoch": 0.26, + "learning_rate": 1.999978140181946e-05, + "loss": 0.5756, + "step": 613 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999770047979828e-05, + "loss": 0.6016, + "step": 614 + }, + { + "epoch": 0.26, + "learning_rate": 1.999975840670568e-05, + "loss": 0.6508, + "step": 615 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999746477997352e-05, + "loss": 0.5523, + "step": 616 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999734261855188e-05, + "loss": 0.6448, + "step": 617 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999721758279532e-05, + "loss": 0.5188, + "step": 618 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999708967270747e-05, + "loss": 0.5533, + "step": 619 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999695888829206e-05, + "loss": 0.4754, + "step": 620 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999682522955277e-05, + "loss": 0.5404, + "step": 621 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999668869649347e-05, + "loss": 0.5205, + "step": 622 + }, + { + "epoch": 0.26, + "learning_rate": 1.999965492891181e-05, + "loss": 0.5183, + "step": 623 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999640700743063e-05, + "loss": 0.5141, + "step": 624 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999626185143518e-05, + "loss": 0.5411, + "step": 625 + }, + { + "epoch": 0.26, + "learning_rate": 1.99996113821136e-05, + "loss": 0.5665, + "step": 626 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999596291653715e-05, + "loss": 0.5936, + "step": 627 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999580913764313e-05, + "loss": 0.5419, + "step": 628 + }, + { + "epoch": 0.26, + "learning_rate": 1.999956524844583e-05, + "loss": 0.5021, + "step": 629 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999549295698722e-05, + "loss": 0.5776, + "step": 630 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999533055523437e-05, + "loss": 0.4954, + "step": 631 + }, + { + "epoch": 0.26, + "learning_rate": 1.999951652792045e-05, + "loss": 0.5628, + "step": 632 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999499712890235e-05, + "loss": 0.6356, + "step": 633 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999482610433275e-05, + "loss": 0.5385, + "step": 634 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999465220550057e-05, + "loss": 0.5615, + "step": 635 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999447543241086e-05, + "loss": 0.5947, + "step": 636 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999429578506867e-05, + "loss": 0.6466, + "step": 637 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999411326347924e-05, + "loss": 0.5652, + "step": 638 + }, + { + "epoch": 0.27, + "learning_rate": 1.999939278676477e-05, + "loss": 0.571, + "step": 639 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999373959757945e-05, + "loss": 0.6267, + "step": 640 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999354845327993e-05, + "loss": 0.5371, + "step": 641 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999335443475452e-05, + "loss": 0.5337, + "step": 642 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999315754200892e-05, + "loss": 0.542, + "step": 643 + }, + { + "epoch": 0.27, + "learning_rate": 1.999929577750487e-05, + "loss": 0.5757, + "step": 644 + }, + { + "epoch": 0.27, + "learning_rate": 1.999927551338797e-05, + "loss": 0.5567, + "step": 645 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999254961850765e-05, + "loss": 0.5668, + "step": 646 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999234122893848e-05, + "loss": 0.5143, + "step": 647 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999212996517823e-05, + "loss": 0.5506, + "step": 648 + }, + { + "epoch": 0.27, + "learning_rate": 1.999919158272329e-05, + "loss": 0.6022, + "step": 649 + }, + { + "epoch": 0.27, + "learning_rate": 1.999916988151087e-05, + "loss": 0.5667, + "step": 650 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999147892881182e-05, + "loss": 0.5127, + "step": 651 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999125616834866e-05, + "loss": 0.6013, + "step": 652 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999103053372554e-05, + "loss": 0.4694, + "step": 653 + }, + { + "epoch": 0.27, + "learning_rate": 1.99990802024949e-05, + "loss": 0.5073, + "step": 654 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999057064202555e-05, + "loss": 0.5867, + "step": 655 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999033638496187e-05, + "loss": 0.5578, + "step": 656 + }, + { + "epoch": 0.28, + "learning_rate": 1.9999009925376475e-05, + "loss": 0.588, + "step": 657 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998985924844093e-05, + "loss": 0.5796, + "step": 658 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998961636899736e-05, + "loss": 0.559, + "step": 659 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998937061544096e-05, + "loss": 0.4719, + "step": 660 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998912198777886e-05, + "loss": 0.5707, + "step": 661 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998887048601812e-05, + "loss": 0.5999, + "step": 662 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998861611016608e-05, + "loss": 0.49, + "step": 663 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998835886022997e-05, + "loss": 0.6035, + "step": 664 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998809873621722e-05, + "loss": 0.5609, + "step": 665 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998783573813535e-05, + "loss": 0.5078, + "step": 666 + }, + { + "epoch": 0.28, + "learning_rate": 1.999875698659918e-05, + "loss": 0.5402, + "step": 667 + }, + { + "epoch": 0.28, + "learning_rate": 1.999873011197943e-05, + "loss": 0.5853, + "step": 668 + }, + { + "epoch": 0.28, + "learning_rate": 1.999870294995506e-05, + "loss": 0.4822, + "step": 669 + }, + { + "epoch": 0.28, + "learning_rate": 1.999867550052684e-05, + "loss": 0.6014, + "step": 670 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998647763695568e-05, + "loss": 0.5914, + "step": 671 + }, + { + "epoch": 0.28, + "learning_rate": 1.999861973946204e-05, + "loss": 0.5698, + "step": 672 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998591427827062e-05, + "loss": 0.5685, + "step": 673 + }, + { + "epoch": 0.28, + "learning_rate": 1.999856282879144e-05, + "loss": 0.6114, + "step": 674 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998533942356008e-05, + "loss": 0.5575, + "step": 675 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998504768521588e-05, + "loss": 0.6381, + "step": 676 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998475307289018e-05, + "loss": 0.5428, + "step": 677 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998445558659152e-05, + "loss": 0.5533, + "step": 678 + }, + { + "epoch": 0.28, + "learning_rate": 1.999841552263284e-05, + "loss": 0.4876, + "step": 679 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998385199210945e-05, + "loss": 0.5705, + "step": 680 + }, + { + "epoch": 0.29, + "learning_rate": 1.999835458839434e-05, + "loss": 0.6025, + "step": 681 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998323690183907e-05, + "loss": 0.6213, + "step": 682 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998292504580528e-05, + "loss": 0.5695, + "step": 683 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998261031585104e-05, + "loss": 0.5282, + "step": 684 + }, + { + "epoch": 0.29, + "learning_rate": 1.999822927119854e-05, + "loss": 0.5748, + "step": 685 + }, + { + "epoch": 0.29, + "learning_rate": 1.999819722342175e-05, + "loss": 0.5202, + "step": 686 + }, + { + "epoch": 0.29, + "learning_rate": 1.999816488825565e-05, + "loss": 0.5667, + "step": 687 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998132265701176e-05, + "loss": 0.5365, + "step": 688 + }, + { + "epoch": 0.29, + "learning_rate": 1.999809935575926e-05, + "loss": 0.6048, + "step": 689 + }, + { + "epoch": 0.29, + "learning_rate": 1.999806615843085e-05, + "loss": 0.6033, + "step": 690 + }, + { + "epoch": 0.29, + "learning_rate": 1.99980326737169e-05, + "loss": 0.5729, + "step": 691 + }, + { + "epoch": 0.29, + "learning_rate": 1.999799890161837e-05, + "loss": 0.492, + "step": 692 + }, + { + "epoch": 0.29, + "learning_rate": 1.999796484213624e-05, + "loss": 0.5711, + "step": 693 + }, + { + "epoch": 0.29, + "learning_rate": 1.999793049527148e-05, + "loss": 0.5011, + "step": 694 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997895861025078e-05, + "loss": 0.5375, + "step": 695 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997860939398035e-05, + "loss": 0.5716, + "step": 696 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997825730391345e-05, + "loss": 0.5735, + "step": 697 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997790234006034e-05, + "loss": 0.529, + "step": 698 + }, + { + "epoch": 0.29, + "learning_rate": 1.999775445024311e-05, + "loss": 0.5457, + "step": 699 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997718379103608e-05, + "loss": 0.5196, + "step": 700 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997682020588563e-05, + "loss": 0.5419, + "step": 701 + }, + { + "epoch": 0.29, + "learning_rate": 1.999764537469902e-05, + "loss": 0.4548, + "step": 702 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997608441436033e-05, + "loss": 0.4813, + "step": 703 + }, + { + "epoch": 0.29, + "learning_rate": 1.999757122080066e-05, + "loss": 0.5566, + "step": 704 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997533712793976e-05, + "loss": 0.5593, + "step": 705 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997495917417054e-05, + "loss": 0.5683, + "step": 706 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997457834670988e-05, + "loss": 0.5919, + "step": 707 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997419464556868e-05, + "loss": 0.6129, + "step": 708 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997380807075796e-05, + "loss": 0.4535, + "step": 709 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997341862228886e-05, + "loss": 0.6517, + "step": 710 + }, + { + "epoch": 0.3, + "learning_rate": 1.999730263001725e-05, + "loss": 0.5925, + "step": 711 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997263110442025e-05, + "loss": 0.4815, + "step": 712 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997223303504345e-05, + "loss": 0.5777, + "step": 713 + }, + { + "epoch": 0.3, + "learning_rate": 1.999718320920535e-05, + "loss": 0.4826, + "step": 714 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997142827546196e-05, + "loss": 0.6005, + "step": 715 + }, + { + "epoch": 0.3, + "learning_rate": 1.999710215852804e-05, + "loss": 0.6185, + "step": 716 + }, + { + "epoch": 0.3, + "learning_rate": 1.999706120215206e-05, + "loss": 0.6308, + "step": 717 + }, + { + "epoch": 0.3, + "learning_rate": 1.999701995841942e-05, + "loss": 0.6157, + "step": 718 + }, + { + "epoch": 0.3, + "learning_rate": 1.999697842733132e-05, + "loss": 0.517, + "step": 719 + }, + { + "epoch": 0.3, + "learning_rate": 1.999693660888894e-05, + "loss": 0.5469, + "step": 720 + }, + { + "epoch": 0.3, + "learning_rate": 1.999689450309349e-05, + "loss": 0.5627, + "step": 721 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996852109946178e-05, + "loss": 0.5886, + "step": 722 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996809429448224e-05, + "loss": 0.5238, + "step": 723 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996766461600855e-05, + "loss": 0.6008, + "step": 724 + }, + { + "epoch": 0.3, + "learning_rate": 1.99967232064053e-05, + "loss": 0.4986, + "step": 725 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996679663862815e-05, + "loss": 0.5396, + "step": 726 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996635833974637e-05, + "loss": 0.497, + "step": 727 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996591716742037e-05, + "loss": 0.5982, + "step": 728 + }, + { + "epoch": 0.31, + "learning_rate": 1.999654731216628e-05, + "loss": 0.5701, + "step": 729 + }, + { + "epoch": 0.31, + "learning_rate": 1.999650262024864e-05, + "loss": 0.5235, + "step": 730 + }, + { + "epoch": 0.31, + "learning_rate": 1.99964576409904e-05, + "loss": 0.5917, + "step": 731 + }, + { + "epoch": 0.31, + "learning_rate": 1.999641237439286e-05, + "loss": 0.5691, + "step": 732 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996366820457313e-05, + "loss": 0.6141, + "step": 733 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996320979185074e-05, + "loss": 0.5376, + "step": 734 + }, + { + "epoch": 0.31, + "learning_rate": 1.999627485057746e-05, + "loss": 0.4333, + "step": 735 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996228434635797e-05, + "loss": 0.5657, + "step": 736 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996181731361415e-05, + "loss": 0.5721, + "step": 737 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996134740755662e-05, + "loss": 0.6029, + "step": 738 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996087462819884e-05, + "loss": 0.5711, + "step": 739 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996039897555443e-05, + "loss": 0.5509, + "step": 740 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995992044963705e-05, + "loss": 0.555, + "step": 741 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995943905046048e-05, + "loss": 0.4628, + "step": 742 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995895477803854e-05, + "loss": 0.5218, + "step": 743 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995846763238514e-05, + "loss": 0.4748, + "step": 744 + }, + { + "epoch": 0.31, + "learning_rate": 1.999579776135143e-05, + "loss": 0.6094, + "step": 745 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995748472144005e-05, + "loss": 0.5397, + "step": 746 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995698895617663e-05, + "loss": 0.6416, + "step": 747 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995649031773828e-05, + "loss": 0.5666, + "step": 748 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995598880613928e-05, + "loss": 0.5718, + "step": 749 + }, + { + "epoch": 0.31, + "learning_rate": 1.999554844213941e-05, + "loss": 0.5107, + "step": 750 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995497716351725e-05, + "loss": 0.5182, + "step": 751 + }, + { + "epoch": 0.31, + "learning_rate": 1.999544670325232e-05, + "loss": 0.645, + "step": 752 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995395402842678e-05, + "loss": 0.6027, + "step": 753 + }, + { + "epoch": 0.32, + "learning_rate": 1.999534381512426e-05, + "loss": 0.506, + "step": 754 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995291940098555e-05, + "loss": 0.4984, + "step": 755 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995239777767053e-05, + "loss": 0.4942, + "step": 756 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995187328131252e-05, + "loss": 0.5343, + "step": 757 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995134591192664e-05, + "loss": 0.4752, + "step": 758 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995081566952795e-05, + "loss": 0.4636, + "step": 759 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995028255413184e-05, + "loss": 0.5338, + "step": 760 + }, + { + "epoch": 0.32, + "learning_rate": 1.999497465657535e-05, + "loss": 0.5684, + "step": 761 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994920770440842e-05, + "loss": 0.4523, + "step": 762 + }, + { + "epoch": 0.32, + "learning_rate": 1.99948665970112e-05, + "loss": 0.6039, + "step": 763 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994812136287994e-05, + "loss": 0.4691, + "step": 764 + }, + { + "epoch": 0.32, + "learning_rate": 1.999475738827278e-05, + "loss": 0.5422, + "step": 765 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994702352967132e-05, + "loss": 0.5557, + "step": 766 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994647030372638e-05, + "loss": 0.545, + "step": 767 + }, + { + "epoch": 0.32, + "learning_rate": 1.999459142049088e-05, + "loss": 0.5399, + "step": 768 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994535523323462e-05, + "loss": 0.5359, + "step": 769 + }, + { + "epoch": 0.32, + "learning_rate": 1.999447933887199e-05, + "loss": 0.5282, + "step": 770 + }, + { + "epoch": 0.32, + "learning_rate": 1.999442286713808e-05, + "loss": 0.5645, + "step": 771 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994366108123352e-05, + "loss": 0.5232, + "step": 772 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994309061829438e-05, + "loss": 0.537, + "step": 773 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994251728257978e-05, + "loss": 0.6248, + "step": 774 + }, + { + "epoch": 0.32, + "learning_rate": 1.999419410741062e-05, + "loss": 0.5334, + "step": 775 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994136199289028e-05, + "loss": 0.5189, + "step": 776 + }, + { + "epoch": 0.33, + "learning_rate": 1.9994078003894856e-05, + "loss": 0.64, + "step": 777 + }, + { + "epoch": 0.33, + "learning_rate": 1.999401952122978e-05, + "loss": 0.6003, + "step": 778 + }, + { + "epoch": 0.33, + "learning_rate": 1.999396075129548e-05, + "loss": 0.5584, + "step": 779 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993901694093645e-05, + "loss": 0.4774, + "step": 780 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993842349625978e-05, + "loss": 0.5653, + "step": 781 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993782717894178e-05, + "loss": 0.5828, + "step": 782 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993722798899963e-05, + "loss": 0.6803, + "step": 783 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993662592645057e-05, + "loss": 0.487, + "step": 784 + }, + { + "epoch": 0.33, + "learning_rate": 1.999360209913119e-05, + "loss": 0.5226, + "step": 785 + }, + { + "epoch": 0.33, + "learning_rate": 1.999354131836009e-05, + "loss": 0.5717, + "step": 786 + }, + { + "epoch": 0.33, + "learning_rate": 1.999348025033352e-05, + "loss": 0.5765, + "step": 787 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993418895053227e-05, + "loss": 0.4651, + "step": 788 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993357252520975e-05, + "loss": 0.5336, + "step": 789 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993295322738537e-05, + "loss": 0.5058, + "step": 790 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993233105707693e-05, + "loss": 0.5516, + "step": 791 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993170601430233e-05, + "loss": 0.57, + "step": 792 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993107809907946e-05, + "loss": 0.481, + "step": 793 + }, + { + "epoch": 0.33, + "learning_rate": 1.999304473114265e-05, + "loss": 0.6331, + "step": 794 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992981365136145e-05, + "loss": 0.5095, + "step": 795 + }, + { + "epoch": 0.33, + "learning_rate": 1.999291771189026e-05, + "loss": 0.5001, + "step": 796 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992853771406825e-05, + "loss": 0.5308, + "step": 797 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992789543687673e-05, + "loss": 0.5813, + "step": 798 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992725028734655e-05, + "loss": 0.6555, + "step": 799 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992660226549624e-05, + "loss": 0.4987, + "step": 800 + }, + { + "epoch": 0.34, + "learning_rate": 1.999259513713444e-05, + "loss": 0.5393, + "step": 801 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992529760490976e-05, + "loss": 0.554, + "step": 802 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992464096621113e-05, + "loss": 0.484, + "step": 803 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992398145526738e-05, + "loss": 0.5911, + "step": 804 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992331907209742e-05, + "loss": 0.5786, + "step": 805 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992265381672036e-05, + "loss": 0.5447, + "step": 806 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992198568915526e-05, + "loss": 0.5934, + "step": 807 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992131468942135e-05, + "loss": 0.5529, + "step": 808 + }, + { + "epoch": 0.34, + "learning_rate": 1.999206408175379e-05, + "loss": 0.539, + "step": 809 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991996407352432e-05, + "loss": 0.5555, + "step": 810 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991928445740007e-05, + "loss": 0.5249, + "step": 811 + }, + { + "epoch": 0.34, + "learning_rate": 1.999186019691846e-05, + "loss": 0.5433, + "step": 812 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991791660889762e-05, + "loss": 0.6182, + "step": 813 + }, + { + "epoch": 0.34, + "learning_rate": 1.999172283765588e-05, + "loss": 0.5141, + "step": 814 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991653727218787e-05, + "loss": 0.5161, + "step": 815 + }, + { + "epoch": 0.34, + "learning_rate": 1.999158432958048e-05, + "loss": 0.5049, + "step": 816 + }, + { + "epoch": 0.34, + "learning_rate": 1.999151464474294e-05, + "loss": 0.4883, + "step": 817 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991444672708183e-05, + "loss": 0.5744, + "step": 818 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991374413478214e-05, + "loss": 0.5944, + "step": 819 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991303867055056e-05, + "loss": 0.5067, + "step": 820 + }, + { + "epoch": 0.34, + "learning_rate": 1.999123303344073e-05, + "loss": 0.5539, + "step": 821 + }, + { + "epoch": 0.34, + "learning_rate": 1.999116191263728e-05, + "loss": 0.5636, + "step": 822 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991090504646748e-05, + "loss": 0.63, + "step": 823 + }, + { + "epoch": 0.34, + "learning_rate": 1.999101880947118e-05, + "loss": 0.5853, + "step": 824 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990946827112648e-05, + "loss": 0.5308, + "step": 825 + }, + { + "epoch": 0.35, + "learning_rate": 1.999087455757321e-05, + "loss": 0.5656, + "step": 826 + }, + { + "epoch": 0.35, + "learning_rate": 1.999080200085495e-05, + "loss": 0.5307, + "step": 827 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990729156959956e-05, + "loss": 0.4947, + "step": 828 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990656025890315e-05, + "loss": 0.5177, + "step": 829 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990582607648133e-05, + "loss": 0.5555, + "step": 830 + }, + { + "epoch": 0.35, + "learning_rate": 1.999050890223552e-05, + "loss": 0.4916, + "step": 831 + }, + { + "epoch": 0.35, + "learning_rate": 1.999043490965459e-05, + "loss": 0.5135, + "step": 832 + }, + { + "epoch": 0.35, + "learning_rate": 1.999036062990748e-05, + "loss": 0.63, + "step": 833 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990286062996316e-05, + "loss": 0.5116, + "step": 834 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990211208923242e-05, + "loss": 0.5488, + "step": 835 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990136067690416e-05, + "loss": 0.6538, + "step": 836 + }, + { + "epoch": 0.35, + "learning_rate": 1.999006063929999e-05, + "loss": 0.5465, + "step": 837 + }, + { + "epoch": 0.35, + "learning_rate": 1.998998492375414e-05, + "loss": 0.5929, + "step": 838 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989908921055036e-05, + "loss": 0.5986, + "step": 839 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989832631204863e-05, + "loss": 0.613, + "step": 840 + }, + { + "epoch": 0.35, + "learning_rate": 1.998975605420582e-05, + "loss": 0.5219, + "step": 841 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989679190060104e-05, + "loss": 0.5513, + "step": 842 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989602038769922e-05, + "loss": 0.4865, + "step": 843 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989524600337493e-05, + "loss": 0.6124, + "step": 844 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989446874765048e-05, + "loss": 0.5157, + "step": 845 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989368862054814e-05, + "loss": 0.5596, + "step": 846 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989290562209036e-05, + "loss": 0.5604, + "step": 847 + }, + { + "epoch": 0.36, + "learning_rate": 1.9989211975229966e-05, + "loss": 0.5181, + "step": 848 + }, + { + "epoch": 0.36, + "learning_rate": 1.998913310111986e-05, + "loss": 0.5237, + "step": 849 + }, + { + "epoch": 0.36, + "learning_rate": 1.998905393988099e-05, + "loss": 0.5698, + "step": 850 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988974491515628e-05, + "loss": 0.5563, + "step": 851 + }, + { + "epoch": 0.36, + "learning_rate": 1.998889475602606e-05, + "loss": 0.5638, + "step": 852 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988814733414573e-05, + "loss": 0.5097, + "step": 853 + }, + { + "epoch": 0.36, + "learning_rate": 1.998873442368347e-05, + "loss": 0.4871, + "step": 854 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988653826835062e-05, + "loss": 0.4984, + "step": 855 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988572942871666e-05, + "loss": 0.5244, + "step": 856 + }, + { + "epoch": 0.36, + "learning_rate": 1.99884917717956e-05, + "loss": 0.6335, + "step": 857 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988410313609203e-05, + "loss": 0.5901, + "step": 858 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988328568314818e-05, + "loss": 0.54, + "step": 859 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988246535914787e-05, + "loss": 0.5668, + "step": 860 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988164216411474e-05, + "loss": 0.5125, + "step": 861 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988081609807243e-05, + "loss": 0.5225, + "step": 862 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987998716104473e-05, + "loss": 0.5736, + "step": 863 + }, + { + "epoch": 0.36, + "learning_rate": 1.998791553530554e-05, + "loss": 0.4624, + "step": 864 + }, + { + "epoch": 0.36, + "learning_rate": 1.998783206741284e-05, + "loss": 0.5752, + "step": 865 + }, + { + "epoch": 0.36, + "learning_rate": 1.998774831242877e-05, + "loss": 0.5861, + "step": 866 + }, + { + "epoch": 0.36, + "learning_rate": 1.998766427035574e-05, + "loss": 0.5887, + "step": 867 + }, + { + "epoch": 0.36, + "learning_rate": 1.998757994119616e-05, + "loss": 0.5148, + "step": 868 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987495324952458e-05, + "loss": 0.5131, + "step": 869 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987410421627063e-05, + "loss": 0.5495, + "step": 870 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987325231222423e-05, + "loss": 0.5244, + "step": 871 + }, + { + "epoch": 0.37, + "learning_rate": 1.998723975374098e-05, + "loss": 0.5098, + "step": 872 + }, + { + "epoch": 0.37, + "learning_rate": 1.998715398918519e-05, + "loss": 0.5323, + "step": 873 + }, + { + "epoch": 0.37, + "learning_rate": 1.9987067937557527e-05, + "loss": 0.6008, + "step": 874 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986981598860454e-05, + "loss": 0.613, + "step": 875 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986894973096458e-05, + "loss": 0.5766, + "step": 876 + }, + { + "epoch": 0.37, + "learning_rate": 1.998680806026803e-05, + "loss": 0.5188, + "step": 877 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986720860377664e-05, + "loss": 0.5137, + "step": 878 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986633373427868e-05, + "loss": 0.6744, + "step": 879 + }, + { + "epoch": 0.37, + "learning_rate": 1.998654559942116e-05, + "loss": 0.5557, + "step": 880 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986457538360062e-05, + "loss": 0.5192, + "step": 881 + }, + { + "epoch": 0.37, + "learning_rate": 1.99863691902471e-05, + "loss": 0.5091, + "step": 882 + }, + { + "epoch": 0.37, + "learning_rate": 1.998628055508482e-05, + "loss": 0.568, + "step": 883 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986191632875768e-05, + "loss": 0.5641, + "step": 884 + }, + { + "epoch": 0.37, + "learning_rate": 1.99861024236225e-05, + "loss": 0.5126, + "step": 885 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986012927327575e-05, + "loss": 0.5267, + "step": 886 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985923143993572e-05, + "loss": 0.4799, + "step": 887 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985833073623068e-05, + "loss": 0.5367, + "step": 888 + }, + { + "epoch": 0.37, + "learning_rate": 1.998574271621866e-05, + "loss": 0.5242, + "step": 889 + }, + { + "epoch": 0.37, + "learning_rate": 1.998565207178293e-05, + "loss": 0.5858, + "step": 890 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985561140318496e-05, + "loss": 0.5344, + "step": 891 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985469921827968e-05, + "loss": 0.5244, + "step": 892 + }, + { + "epoch": 0.37, + "learning_rate": 1.998537841631397e-05, + "loss": 0.5289, + "step": 893 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985286623779126e-05, + "loss": 0.5555, + "step": 894 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985194544226082e-05, + "loss": 0.5863, + "step": 895 + }, + { + "epoch": 0.38, + "learning_rate": 1.998510217765748e-05, + "loss": 0.4871, + "step": 896 + }, + { + "epoch": 0.38, + "learning_rate": 1.998500952407598e-05, + "loss": 0.5624, + "step": 897 + }, + { + "epoch": 0.38, + "learning_rate": 1.998491658348424e-05, + "loss": 0.5368, + "step": 898 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984823355884926e-05, + "loss": 0.4979, + "step": 899 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984729841280734e-05, + "loss": 0.6757, + "step": 900 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984636039674342e-05, + "loss": 0.5738, + "step": 901 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984541951068445e-05, + "loss": 0.5586, + "step": 902 + }, + { + "epoch": 0.38, + "learning_rate": 1.998444757546575e-05, + "loss": 0.5576, + "step": 903 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984352912868973e-05, + "loss": 0.5864, + "step": 904 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984257963280823e-05, + "loss": 0.5936, + "step": 905 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984162726704047e-05, + "loss": 0.6112, + "step": 906 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984067203141368e-05, + "loss": 0.5352, + "step": 907 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983971392595537e-05, + "loss": 0.6031, + "step": 908 + }, + { + "epoch": 0.38, + "learning_rate": 1.998387529506931e-05, + "loss": 0.5561, + "step": 909 + }, + { + "epoch": 0.38, + "learning_rate": 1.998377891056545e-05, + "loss": 0.5093, + "step": 910 + }, + { + "epoch": 0.38, + "learning_rate": 1.998368223908672e-05, + "loss": 0.6654, + "step": 911 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983585280635902e-05, + "loss": 0.4926, + "step": 912 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983488035215788e-05, + "loss": 0.5659, + "step": 913 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983390502829168e-05, + "loss": 0.5259, + "step": 914 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983292683478844e-05, + "loss": 0.4566, + "step": 915 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983194577167636e-05, + "loss": 0.524, + "step": 916 + }, + { + "epoch": 0.38, + "learning_rate": 1.998309618389836e-05, + "loss": 0.4984, + "step": 917 + }, + { + "epoch": 0.38, + "learning_rate": 1.9982997503673837e-05, + "loss": 0.5231, + "step": 918 + }, + { + "epoch": 0.38, + "learning_rate": 1.9982898536496912e-05, + "loss": 0.5507, + "step": 919 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982799282370432e-05, + "loss": 0.5801, + "step": 920 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982699741297236e-05, + "loss": 0.5397, + "step": 921 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982599913280202e-05, + "loss": 0.5434, + "step": 922 + }, + { + "epoch": 0.39, + "learning_rate": 1.998249979832219e-05, + "loss": 0.507, + "step": 923 + }, + { + "epoch": 0.39, + "learning_rate": 1.998239939642608e-05, + "loss": 0.5271, + "step": 924 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982298707594757e-05, + "loss": 0.5165, + "step": 925 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982197731831115e-05, + "loss": 0.6061, + "step": 926 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982096469138056e-05, + "loss": 0.6236, + "step": 927 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981994919518496e-05, + "loss": 0.5399, + "step": 928 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981893082975348e-05, + "loss": 0.5696, + "step": 929 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981790959511538e-05, + "loss": 0.5813, + "step": 930 + }, + { + "epoch": 0.39, + "learning_rate": 1.998168854913001e-05, + "loss": 0.5144, + "step": 931 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981585851833697e-05, + "loss": 0.4815, + "step": 932 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981482867625557e-05, + "loss": 0.5817, + "step": 933 + }, + { + "epoch": 0.39, + "learning_rate": 1.998137959650855e-05, + "loss": 0.5457, + "step": 934 + }, + { + "epoch": 0.39, + "learning_rate": 1.998127603848565e-05, + "loss": 0.4847, + "step": 935 + }, + { + "epoch": 0.39, + "learning_rate": 1.998117219355982e-05, + "loss": 0.4814, + "step": 936 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981068061734055e-05, + "loss": 0.5047, + "step": 937 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980963643011346e-05, + "loss": 0.5589, + "step": 938 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980858937394695e-05, + "loss": 0.5293, + "step": 939 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980753944887108e-05, + "loss": 0.5762, + "step": 940 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980648665491606e-05, + "loss": 0.557, + "step": 941 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980543099211215e-05, + "loss": 0.577, + "step": 942 + }, + { + "epoch": 0.39, + "learning_rate": 1.998043724604897e-05, + "loss": 0.5131, + "step": 943 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980331106007912e-05, + "loss": 0.5274, + "step": 944 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980224679091092e-05, + "loss": 0.5165, + "step": 945 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980117965301574e-05, + "loss": 0.5571, + "step": 946 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980010964642417e-05, + "loss": 0.5509, + "step": 947 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979903677116705e-05, + "loss": 0.4886, + "step": 948 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979796102727513e-05, + "loss": 0.5009, + "step": 949 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979688241477938e-05, + "loss": 0.5245, + "step": 950 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979580093371084e-05, + "loss": 0.5203, + "step": 951 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979471658410055e-05, + "loss": 0.5339, + "step": 952 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979362936597967e-05, + "loss": 0.6407, + "step": 953 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979253927937948e-05, + "loss": 0.5562, + "step": 954 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979144632433132e-05, + "loss": 0.5537, + "step": 955 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979035050086656e-05, + "loss": 0.537, + "step": 956 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978925180901675e-05, + "loss": 0.5218, + "step": 957 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978815024881342e-05, + "loss": 0.619, + "step": 958 + }, + { + "epoch": 0.4, + "learning_rate": 1.997870458202883e-05, + "loss": 0.5215, + "step": 959 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978593852347306e-05, + "loss": 0.5597, + "step": 960 + }, + { + "epoch": 0.4, + "learning_rate": 1.997848283583996e-05, + "loss": 0.4835, + "step": 961 + }, + { + "epoch": 0.4, + "learning_rate": 1.997837153250998e-05, + "loss": 0.5569, + "step": 962 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978259942360562e-05, + "loss": 0.5173, + "step": 963 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978148065394917e-05, + "loss": 0.5005, + "step": 964 + }, + { + "epoch": 0.4, + "learning_rate": 1.997803590161626e-05, + "loss": 0.6021, + "step": 965 + }, + { + "epoch": 0.4, + "learning_rate": 1.9977923451027816e-05, + "loss": 0.5151, + "step": 966 + }, + { + "epoch": 0.4, + "learning_rate": 1.9977810713632815e-05, + "loss": 0.5957, + "step": 967 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977697689434504e-05, + "loss": 0.5009, + "step": 968 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977584378436123e-05, + "loss": 0.5259, + "step": 969 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977470780640934e-05, + "loss": 0.4887, + "step": 970 + }, + { + "epoch": 0.41, + "learning_rate": 1.99773568960522e-05, + "loss": 0.5379, + "step": 971 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977242724673198e-05, + "loss": 0.6189, + "step": 972 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977128266507203e-05, + "loss": 0.5615, + "step": 973 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977013521557516e-05, + "loss": 0.5451, + "step": 974 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976898489827423e-05, + "loss": 0.5693, + "step": 975 + }, + { + "epoch": 0.41, + "learning_rate": 1.997678317132024e-05, + "loss": 0.5098, + "step": 976 + }, + { + "epoch": 0.41, + "learning_rate": 1.997666756603928e-05, + "loss": 0.5261, + "step": 977 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976551673987857e-05, + "loss": 0.525, + "step": 978 + }, + { + "epoch": 0.41, + "learning_rate": 1.997643549516931e-05, + "loss": 0.6046, + "step": 979 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976319029586982e-05, + "loss": 0.572, + "step": 980 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976202277244216e-05, + "loss": 0.5908, + "step": 981 + }, + { + "epoch": 0.41, + "learning_rate": 1.997608523814437e-05, + "loss": 0.5843, + "step": 982 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975967912290805e-05, + "loss": 0.5956, + "step": 983 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975850299686894e-05, + "loss": 0.5536, + "step": 984 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975732400336018e-05, + "loss": 0.5441, + "step": 985 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975614214241562e-05, + "loss": 0.5195, + "step": 986 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975495741406934e-05, + "loss": 0.5481, + "step": 987 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975376981835533e-05, + "loss": 0.5622, + "step": 988 + }, + { + "epoch": 0.41, + "learning_rate": 1.997525793553077e-05, + "loss": 0.4714, + "step": 989 + }, + { + "epoch": 0.41, + "learning_rate": 1.997513860249607e-05, + "loss": 0.5424, + "step": 990 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975018982734863e-05, + "loss": 0.5466, + "step": 991 + }, + { + "epoch": 0.42, + "learning_rate": 1.997489907625059e-05, + "loss": 0.5853, + "step": 992 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974778883046688e-05, + "loss": 0.4975, + "step": 993 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974658403126624e-05, + "loss": 0.4859, + "step": 994 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974537636493853e-05, + "loss": 0.4968, + "step": 995 + }, + { + "epoch": 0.42, + "learning_rate": 1.997441658315185e-05, + "loss": 0.5131, + "step": 996 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974295243104093e-05, + "loss": 0.5391, + "step": 997 + }, + { + "epoch": 0.42, + "learning_rate": 1.997417361635407e-05, + "loss": 0.4989, + "step": 998 + }, + { + "epoch": 0.42, + "learning_rate": 1.997405170290528e-05, + "loss": 0.5016, + "step": 999 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973929502761222e-05, + "loss": 0.5196, + "step": 1000 + }, + { + "epoch": 0.42, + "learning_rate": 1.997380701592541e-05, + "loss": 0.5563, + "step": 1001 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973684242401366e-05, + "loss": 0.5229, + "step": 1002 + }, + { + "epoch": 0.42, + "learning_rate": 1.997356118219262e-05, + "loss": 0.4715, + "step": 1003 + }, + { + "epoch": 0.42, + "learning_rate": 1.997343783530271e-05, + "loss": 0.5847, + "step": 1004 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973314201735176e-05, + "loss": 0.6229, + "step": 1005 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973190281493578e-05, + "loss": 0.4641, + "step": 1006 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973066074581475e-05, + "loss": 0.4784, + "step": 1007 + }, + { + "epoch": 0.42, + "learning_rate": 1.997294158100244e-05, + "loss": 0.4818, + "step": 1008 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972816800760046e-05, + "loss": 0.4652, + "step": 1009 + }, + { + "epoch": 0.42, + "learning_rate": 1.997269173385788e-05, + "loss": 0.5223, + "step": 1010 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972566380299544e-05, + "loss": 0.5002, + "step": 1011 + }, + { + "epoch": 0.42, + "learning_rate": 1.997244074008864e-05, + "loss": 0.487, + "step": 1012 + }, + { + "epoch": 0.42, + "learning_rate": 1.997231481322877e-05, + "loss": 0.5619, + "step": 1013 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972188599723564e-05, + "loss": 0.5361, + "step": 1014 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972062099576642e-05, + "loss": 0.524, + "step": 1015 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971935312791646e-05, + "loss": 0.4828, + "step": 1016 + }, + { + "epoch": 0.43, + "learning_rate": 1.997180823937222e-05, + "loss": 0.5571, + "step": 1017 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971680879322012e-05, + "loss": 0.528, + "step": 1018 + }, + { + "epoch": 0.43, + "learning_rate": 1.997155323264469e-05, + "loss": 0.5309, + "step": 1019 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971425299343915e-05, + "loss": 0.4413, + "step": 1020 + }, + { + "epoch": 0.43, + "learning_rate": 1.997129707942337e-05, + "loss": 0.4623, + "step": 1021 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971168572886738e-05, + "loss": 0.4915, + "step": 1022 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971039779737717e-05, + "loss": 0.5633, + "step": 1023 + }, + { + "epoch": 0.43, + "learning_rate": 1.997091069998e-05, + "loss": 0.577, + "step": 1024 + }, + { + "epoch": 0.43, + "learning_rate": 1.9970781333617303e-05, + "loss": 0.5143, + "step": 1025 + }, + { + "epoch": 0.43, + "learning_rate": 1.997065168065335e-05, + "loss": 0.5277, + "step": 1026 + }, + { + "epoch": 0.43, + "learning_rate": 1.997052174109186e-05, + "loss": 0.5045, + "step": 1027 + }, + { + "epoch": 0.43, + "learning_rate": 1.997039151493657e-05, + "loss": 0.5114, + "step": 1028 + }, + { + "epoch": 0.43, + "learning_rate": 1.997026100219122e-05, + "loss": 0.4649, + "step": 1029 + }, + { + "epoch": 0.43, + "learning_rate": 1.997013020285957e-05, + "loss": 0.4951, + "step": 1030 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969999116945374e-05, + "loss": 0.5396, + "step": 1031 + }, + { + "epoch": 0.43, + "learning_rate": 1.99698677444524e-05, + "loss": 0.5606, + "step": 1032 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969736085384422e-05, + "loss": 0.5713, + "step": 1033 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969604139745227e-05, + "loss": 0.5012, + "step": 1034 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969471907538615e-05, + "loss": 0.6019, + "step": 1035 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969339388768372e-05, + "loss": 0.5543, + "step": 1036 + }, + { + "epoch": 0.43, + "learning_rate": 1.996920658343832e-05, + "loss": 0.5305, + "step": 1037 + }, + { + "epoch": 0.43, + "learning_rate": 1.996907349155227e-05, + "loss": 0.5374, + "step": 1038 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968940113114048e-05, + "loss": 0.5252, + "step": 1039 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968806448127488e-05, + "loss": 0.5686, + "step": 1040 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968672496596434e-05, + "loss": 0.5571, + "step": 1041 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968538258524736e-05, + "loss": 0.5308, + "step": 1042 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968403733916246e-05, + "loss": 0.5229, + "step": 1043 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968268922774842e-05, + "loss": 0.5767, + "step": 1044 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968133825104396e-05, + "loss": 0.3965, + "step": 1045 + }, + { + "epoch": 0.44, + "learning_rate": 1.996799844090878e-05, + "loss": 0.6016, + "step": 1046 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967862770191902e-05, + "loss": 0.5493, + "step": 1047 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967726812957652e-05, + "loss": 0.5211, + "step": 1048 + }, + { + "epoch": 0.44, + "learning_rate": 1.996759056920994e-05, + "loss": 0.6184, + "step": 1049 + }, + { + "epoch": 0.44, + "learning_rate": 1.996745403895268e-05, + "loss": 0.5019, + "step": 1050 + }, + { + "epoch": 0.44, + "learning_rate": 1.99673172221898e-05, + "loss": 0.549, + "step": 1051 + }, + { + "epoch": 0.44, + "learning_rate": 1.996718011892523e-05, + "loss": 0.484, + "step": 1052 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967042729162914e-05, + "loss": 0.5575, + "step": 1053 + }, + { + "epoch": 0.44, + "learning_rate": 1.99669050529068e-05, + "loss": 0.5138, + "step": 1054 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966767090160843e-05, + "loss": 0.4945, + "step": 1055 + }, + { + "epoch": 0.44, + "learning_rate": 1.996662884092901e-05, + "loss": 0.4617, + "step": 1056 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966490305215277e-05, + "loss": 0.5356, + "step": 1057 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966351483023627e-05, + "loss": 0.4732, + "step": 1058 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966212374358044e-05, + "loss": 0.4211, + "step": 1059 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966072979222527e-05, + "loss": 0.5779, + "step": 1060 + }, + { + "epoch": 0.44, + "learning_rate": 1.996593329762109e-05, + "loss": 0.5561, + "step": 1061 + }, + { + "epoch": 0.44, + "learning_rate": 1.9965793329557743e-05, + "loss": 0.5145, + "step": 1062 + }, + { + "epoch": 0.45, + "learning_rate": 1.996565307503651e-05, + "loss": 0.5379, + "step": 1063 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965512534061426e-05, + "loss": 0.446, + "step": 1064 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965371706636522e-05, + "loss": 0.5324, + "step": 1065 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965230592765856e-05, + "loss": 0.4506, + "step": 1066 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965089192453475e-05, + "loss": 0.5063, + "step": 1067 + }, + { + "epoch": 0.45, + "learning_rate": 1.996494750570345e-05, + "loss": 0.5347, + "step": 1068 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964805532519852e-05, + "loss": 0.5617, + "step": 1069 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964663272906763e-05, + "loss": 0.4943, + "step": 1070 + }, + { + "epoch": 0.45, + "learning_rate": 1.996452072686827e-05, + "loss": 0.5119, + "step": 1071 + }, + { + "epoch": 0.45, + "learning_rate": 1.996437789440847e-05, + "loss": 0.5418, + "step": 1072 + }, + { + "epoch": 0.45, + "learning_rate": 1.996423477553147e-05, + "loss": 0.5345, + "step": 1073 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964091370241384e-05, + "loss": 0.4945, + "step": 1074 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963947678542335e-05, + "loss": 0.5395, + "step": 1075 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963803700438448e-05, + "loss": 0.5527, + "step": 1076 + }, + { + "epoch": 0.45, + "learning_rate": 1.996365943593387e-05, + "loss": 0.4826, + "step": 1077 + }, + { + "epoch": 0.45, + "learning_rate": 1.996351488503274e-05, + "loss": 0.5785, + "step": 1078 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963370047739218e-05, + "loss": 0.5382, + "step": 1079 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963224924057463e-05, + "loss": 0.5902, + "step": 1080 + }, + { + "epoch": 0.45, + "learning_rate": 1.996307951399165e-05, + "loss": 0.5723, + "step": 1081 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962933817545957e-05, + "loss": 0.5868, + "step": 1082 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962787834724572e-05, + "loss": 0.6129, + "step": 1083 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962641565531694e-05, + "loss": 0.501, + "step": 1084 + }, + { + "epoch": 0.45, + "learning_rate": 1.996249500997152e-05, + "loss": 0.5659, + "step": 1085 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962348168048272e-05, + "loss": 0.556, + "step": 1086 + }, + { + "epoch": 0.46, + "learning_rate": 1.9962201039766163e-05, + "loss": 0.556, + "step": 1087 + }, + { + "epoch": 0.46, + "learning_rate": 1.996205362512943e-05, + "loss": 0.5758, + "step": 1088 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961905924142303e-05, + "loss": 0.4728, + "step": 1089 + }, + { + "epoch": 0.46, + "learning_rate": 1.996175793680903e-05, + "loss": 0.5484, + "step": 1090 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961609663133865e-05, + "loss": 0.5067, + "step": 1091 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961461103121073e-05, + "loss": 0.5013, + "step": 1092 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961312256774915e-05, + "loss": 0.5285, + "step": 1093 + }, + { + "epoch": 0.46, + "learning_rate": 1.996116312409968e-05, + "loss": 0.5439, + "step": 1094 + }, + { + "epoch": 0.46, + "learning_rate": 1.996101370509965e-05, + "loss": 0.5521, + "step": 1095 + }, + { + "epoch": 0.46, + "learning_rate": 1.996086399977912e-05, + "loss": 0.5894, + "step": 1096 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960714008142394e-05, + "loss": 0.4348, + "step": 1097 + }, + { + "epoch": 0.46, + "learning_rate": 1.996056373019378e-05, + "loss": 0.5884, + "step": 1098 + }, + { + "epoch": 0.46, + "learning_rate": 1.99604131659376e-05, + "loss": 0.4819, + "step": 1099 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960262315378188e-05, + "loss": 0.4674, + "step": 1100 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960111178519873e-05, + "loss": 0.5385, + "step": 1101 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959959755366997e-05, + "loss": 0.5105, + "step": 1102 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959808045923917e-05, + "loss": 0.4483, + "step": 1103 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959656050194994e-05, + "loss": 0.5014, + "step": 1104 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959503768184597e-05, + "loss": 0.5339, + "step": 1105 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959351199897102e-05, + "loss": 0.4732, + "step": 1106 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959198345336893e-05, + "loss": 0.5171, + "step": 1107 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959045204508366e-05, + "loss": 0.5483, + "step": 1108 + }, + { + "epoch": 0.46, + "learning_rate": 1.9958891777415917e-05, + "loss": 0.5923, + "step": 1109 + }, + { + "epoch": 0.46, + "learning_rate": 1.9958738064063968e-05, + "loss": 0.5747, + "step": 1110 + }, + { + "epoch": 0.47, + "learning_rate": 1.995858406445693e-05, + "loss": 0.5271, + "step": 1111 + }, + { + "epoch": 0.47, + "learning_rate": 1.9958429778599225e-05, + "loss": 0.5466, + "step": 1112 + }, + { + "epoch": 0.47, + "learning_rate": 1.99582752064953e-05, + "loss": 0.4861, + "step": 1113 + }, + { + "epoch": 0.47, + "learning_rate": 1.9958120348149585e-05, + "loss": 0.5899, + "step": 1114 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957965203566542e-05, + "loss": 0.4815, + "step": 1115 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957809772750623e-05, + "loss": 0.5185, + "step": 1116 + }, + { + "epoch": 0.47, + "learning_rate": 1.99576540557063e-05, + "loss": 0.4802, + "step": 1117 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957498052438046e-05, + "loss": 0.5159, + "step": 1118 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957341762950346e-05, + "loss": 0.6096, + "step": 1119 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957185187247697e-05, + "loss": 0.5007, + "step": 1120 + }, + { + "epoch": 0.47, + "learning_rate": 1.995702832533459e-05, + "loss": 0.5597, + "step": 1121 + }, + { + "epoch": 0.47, + "learning_rate": 1.995687117721554e-05, + "loss": 0.4951, + "step": 1122 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956713742895068e-05, + "loss": 0.4691, + "step": 1123 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956556022377693e-05, + "loss": 0.5219, + "step": 1124 + }, + { + "epoch": 0.47, + "learning_rate": 1.995639801566795e-05, + "loss": 0.545, + "step": 1125 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956239722770382e-05, + "loss": 0.4985, + "step": 1126 + }, + { + "epoch": 0.47, + "learning_rate": 1.995608114368954e-05, + "loss": 0.5149, + "step": 1127 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955922278429975e-05, + "loss": 0.5308, + "step": 1128 + }, + { + "epoch": 0.47, + "learning_rate": 1.995576312699626e-05, + "loss": 0.5482, + "step": 1129 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955603689392974e-05, + "loss": 0.5766, + "step": 1130 + }, + { + "epoch": 0.47, + "learning_rate": 1.995544396562469e-05, + "loss": 0.4821, + "step": 1131 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955283955696004e-05, + "loss": 0.4794, + "step": 1132 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955123659611516e-05, + "loss": 0.5762, + "step": 1133 + }, + { + "epoch": 0.47, + "learning_rate": 1.995496307737583e-05, + "loss": 0.5338, + "step": 1134 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954802208993565e-05, + "loss": 0.574, + "step": 1135 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954641054469347e-05, + "loss": 0.6228, + "step": 1136 + }, + { + "epoch": 0.48, + "learning_rate": 1.99544796138078e-05, + "loss": 0.5616, + "step": 1137 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954317887013576e-05, + "loss": 0.5199, + "step": 1138 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954155874091313e-05, + "loss": 0.5464, + "step": 1139 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953993575045674e-05, + "loss": 0.5484, + "step": 1140 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953830989881324e-05, + "loss": 0.5235, + "step": 1141 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953668118602937e-05, + "loss": 0.5152, + "step": 1142 + }, + { + "epoch": 0.48, + "learning_rate": 1.995350496121519e-05, + "loss": 0.5461, + "step": 1143 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953341517722773e-05, + "loss": 0.4547, + "step": 1144 + }, + { + "epoch": 0.48, + "learning_rate": 1.995317778813039e-05, + "loss": 0.5252, + "step": 1145 + }, + { + "epoch": 0.48, + "learning_rate": 1.995301377244274e-05, + "loss": 0.5046, + "step": 1146 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952849470664547e-05, + "loss": 0.4885, + "step": 1147 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952684882800525e-05, + "loss": 0.5105, + "step": 1148 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952520008855405e-05, + "loss": 0.55, + "step": 1149 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952354848833932e-05, + "loss": 0.5413, + "step": 1150 + }, + { + "epoch": 0.48, + "learning_rate": 1.995218940274085e-05, + "loss": 0.552, + "step": 1151 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952023670580915e-05, + "loss": 0.6011, + "step": 1152 + }, + { + "epoch": 0.48, + "learning_rate": 1.995185765235889e-05, + "loss": 0.5404, + "step": 1153 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951691348079553e-05, + "loss": 0.4995, + "step": 1154 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951524757747675e-05, + "loss": 0.5654, + "step": 1155 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951357881368047e-05, + "loss": 0.5018, + "step": 1156 + }, + { + "epoch": 0.48, + "learning_rate": 1.995119071894547e-05, + "loss": 0.5189, + "step": 1157 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951023270484748e-05, + "loss": 0.5147, + "step": 1158 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950855535990688e-05, + "loss": 0.577, + "step": 1159 + }, + { + "epoch": 0.49, + "learning_rate": 1.995068751546812e-05, + "loss": 0.4954, + "step": 1160 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950519208921866e-05, + "loss": 0.5634, + "step": 1161 + }, + { + "epoch": 0.49, + "learning_rate": 1.995035061635677e-05, + "loss": 0.4313, + "step": 1162 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950181737777673e-05, + "loss": 0.4947, + "step": 1163 + }, + { + "epoch": 0.49, + "learning_rate": 1.995001257318943e-05, + "loss": 0.467, + "step": 1164 + }, + { + "epoch": 0.49, + "learning_rate": 1.994984312259691e-05, + "loss": 0.4954, + "step": 1165 + }, + { + "epoch": 0.49, + "learning_rate": 1.9949673386004975e-05, + "loss": 0.5324, + "step": 1166 + }, + { + "epoch": 0.49, + "learning_rate": 1.994950336341851e-05, + "loss": 0.4987, + "step": 1167 + }, + { + "epoch": 0.49, + "learning_rate": 1.9949333054842398e-05, + "loss": 0.5435, + "step": 1168 + }, + { + "epoch": 0.49, + "learning_rate": 1.994916246028154e-05, + "loss": 0.4749, + "step": 1169 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948991579740832e-05, + "loss": 0.4948, + "step": 1170 + }, + { + "epoch": 0.49, + "learning_rate": 1.994882041322519e-05, + "loss": 0.5437, + "step": 1171 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948648960739536e-05, + "loss": 0.5135, + "step": 1172 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948477222288792e-05, + "loss": 0.5212, + "step": 1173 + }, + { + "epoch": 0.49, + "learning_rate": 1.99483051978779e-05, + "loss": 0.4861, + "step": 1174 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948132887511805e-05, + "loss": 0.5396, + "step": 1175 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947960291195457e-05, + "loss": 0.5566, + "step": 1176 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947787408933816e-05, + "loss": 0.653, + "step": 1177 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947614240731856e-05, + "loss": 0.5085, + "step": 1178 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947440786594552e-05, + "loss": 0.4604, + "step": 1179 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947267046526887e-05, + "loss": 0.538, + "step": 1180 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947093020533863e-05, + "loss": 0.479, + "step": 1181 + }, + { + "epoch": 0.49, + "learning_rate": 1.9946918708620472e-05, + "loss": 0.5215, + "step": 1182 + }, + { + "epoch": 0.5, + "learning_rate": 1.994674411079173e-05, + "loss": 0.49, + "step": 1183 + }, + { + "epoch": 0.5, + "learning_rate": 1.994656922705266e-05, + "loss": 0.5296, + "step": 1184 + }, + { + "epoch": 0.5, + "learning_rate": 1.9946394057408278e-05, + "loss": 0.4532, + "step": 1185 + }, + { + "epoch": 0.5, + "learning_rate": 1.9946218601863626e-05, + "loss": 0.4952, + "step": 1186 + }, + { + "epoch": 0.5, + "learning_rate": 1.994604286042375e-05, + "loss": 0.5172, + "step": 1187 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945866833093694e-05, + "loss": 0.4908, + "step": 1188 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945690519878523e-05, + "loss": 0.6495, + "step": 1189 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945513920783303e-05, + "loss": 0.5633, + "step": 1190 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945337035813114e-05, + "loss": 0.4542, + "step": 1191 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945159864973035e-05, + "loss": 0.5391, + "step": 1192 + }, + { + "epoch": 0.5, + "learning_rate": 1.994498240826816e-05, + "loss": 0.5666, + "step": 1193 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944804665703593e-05, + "loss": 0.5144, + "step": 1194 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944626637284438e-05, + "loss": 0.4959, + "step": 1195 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944448323015818e-05, + "loss": 0.5488, + "step": 1196 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944269722902854e-05, + "loss": 0.4746, + "step": 1197 + }, + { + "epoch": 0.5, + "learning_rate": 1.994409083695068e-05, + "loss": 0.5414, + "step": 1198 + }, + { + "epoch": 0.5, + "learning_rate": 1.994391166516444e-05, + "loss": 0.5357, + "step": 1199 + }, + { + "epoch": 0.5, + "learning_rate": 1.9943732207549287e-05, + "loss": 0.5277, + "step": 1200 + }, + { + "epoch": 0.5, + "learning_rate": 1.994355246411037e-05, + "loss": 0.5926, + "step": 1201 + }, + { + "epoch": 0.5, + "learning_rate": 1.9943372434852864e-05, + "loss": 0.5177, + "step": 1202 + }, + { + "epoch": 0.5, + "learning_rate": 1.994319211978194e-05, + "loss": 0.5034, + "step": 1203 + }, + { + "epoch": 0.5, + "learning_rate": 1.994301151890278e-05, + "loss": 0.5085, + "step": 1204 + }, + { + "epoch": 0.5, + "learning_rate": 1.994283063222058e-05, + "loss": 0.6082, + "step": 1205 + }, + { + "epoch": 0.5, + "learning_rate": 1.9942649459740537e-05, + "loss": 0.4875, + "step": 1206 + }, + { + "epoch": 0.51, + "learning_rate": 1.9942468001467857e-05, + "loss": 0.5117, + "step": 1207 + }, + { + "epoch": 0.51, + "learning_rate": 1.994228625740776e-05, + "loss": 0.453, + "step": 1208 + }, + { + "epoch": 0.51, + "learning_rate": 1.994210422756546e-05, + "loss": 0.509, + "step": 1209 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941921911946202e-05, + "loss": 0.5297, + "step": 1210 + }, + { + "epoch": 0.51, + "learning_rate": 1.994173931055522e-05, + "loss": 0.5244, + "step": 1211 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941556423397762e-05, + "loss": 0.4848, + "step": 1212 + }, + { + "epoch": 0.51, + "learning_rate": 1.994137325047909e-05, + "loss": 0.5528, + "step": 1213 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941189791804466e-05, + "loss": 0.5117, + "step": 1214 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941006047379158e-05, + "loss": 0.4961, + "step": 1215 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940822017208458e-05, + "loss": 0.5069, + "step": 1216 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940637701297647e-05, + "loss": 0.4534, + "step": 1217 + }, + { + "epoch": 0.51, + "learning_rate": 1.994045309965203e-05, + "loss": 0.5828, + "step": 1218 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940268212276904e-05, + "loss": 0.4312, + "step": 1219 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940083039177594e-05, + "loss": 0.4555, + "step": 1220 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939897580359417e-05, + "loss": 0.5016, + "step": 1221 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939711835827703e-05, + "loss": 0.5193, + "step": 1222 + }, + { + "epoch": 0.51, + "learning_rate": 1.993952580558779e-05, + "loss": 0.4985, + "step": 1223 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939339489645034e-05, + "loss": 0.4723, + "step": 1224 + }, + { + "epoch": 0.51, + "learning_rate": 1.993915288800478e-05, + "loss": 0.4698, + "step": 1225 + }, + { + "epoch": 0.51, + "learning_rate": 1.99389660006724e-05, + "loss": 0.4951, + "step": 1226 + }, + { + "epoch": 0.51, + "learning_rate": 1.9938778827653256e-05, + "loss": 0.4973, + "step": 1227 + }, + { + "epoch": 0.51, + "learning_rate": 1.993859136895274e-05, + "loss": 0.5482, + "step": 1228 + }, + { + "epoch": 0.51, + "learning_rate": 1.993840362457623e-05, + "loss": 0.5661, + "step": 1229 + }, + { + "epoch": 0.51, + "learning_rate": 1.993821559452913e-05, + "loss": 0.5084, + "step": 1230 + }, + { + "epoch": 0.52, + "learning_rate": 1.993802727881684e-05, + "loss": 0.6094, + "step": 1231 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937838677444776e-05, + "loss": 0.4842, + "step": 1232 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937649790418357e-05, + "loss": 0.4732, + "step": 1233 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937460617743013e-05, + "loss": 0.5847, + "step": 1234 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937271159424185e-05, + "loss": 0.5418, + "step": 1235 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937081415467313e-05, + "loss": 0.4498, + "step": 1236 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936891385877854e-05, + "loss": 0.5086, + "step": 1237 + }, + { + "epoch": 0.52, + "learning_rate": 1.993670107066127e-05, + "loss": 0.4876, + "step": 1238 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936510469823028e-05, + "loss": 0.5137, + "step": 1239 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936319583368616e-05, + "loss": 0.4546, + "step": 1240 + }, + { + "epoch": 0.52, + "learning_rate": 1.993612841130351e-05, + "loss": 0.5644, + "step": 1241 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935936953633214e-05, + "loss": 0.5763, + "step": 1242 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935745210363224e-05, + "loss": 0.5405, + "step": 1243 + }, + { + "epoch": 0.52, + "learning_rate": 1.993555318149906e-05, + "loss": 0.4923, + "step": 1244 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935360867046228e-05, + "loss": 0.5439, + "step": 1245 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935168267010268e-05, + "loss": 0.4501, + "step": 1246 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934975381396716e-05, + "loss": 0.5177, + "step": 1247 + }, + { + "epoch": 0.52, + "learning_rate": 1.993478221021111e-05, + "loss": 0.5299, + "step": 1248 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934588753459006e-05, + "loss": 0.5357, + "step": 1249 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934395011145967e-05, + "loss": 0.4541, + "step": 1250 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934200983277553e-05, + "loss": 0.464, + "step": 1251 + }, + { + "epoch": 0.52, + "learning_rate": 1.993400666985935e-05, + "loss": 0.5593, + "step": 1252 + }, + { + "epoch": 0.52, + "learning_rate": 1.993381207089694e-05, + "loss": 0.516, + "step": 1253 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933617186395917e-05, + "loss": 0.6846, + "step": 1254 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933422016361886e-05, + "loss": 0.5269, + "step": 1255 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933226560800452e-05, + "loss": 0.5366, + "step": 1256 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933030819717235e-05, + "loss": 0.4933, + "step": 1257 + }, + { + "epoch": 0.53, + "learning_rate": 1.993283479311786e-05, + "loss": 0.5686, + "step": 1258 + }, + { + "epoch": 0.53, + "learning_rate": 1.993263848100796e-05, + "loss": 0.4922, + "step": 1259 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932441883393186e-05, + "loss": 0.546, + "step": 1260 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932245000279184e-05, + "loss": 0.5403, + "step": 1261 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932047831671615e-05, + "loss": 0.4769, + "step": 1262 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931850377576142e-05, + "loss": 0.5271, + "step": 1263 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931652637998443e-05, + "loss": 0.5356, + "step": 1264 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931454612944203e-05, + "loss": 0.5345, + "step": 1265 + }, + { + "epoch": 0.53, + "learning_rate": 1.993125630241911e-05, + "loss": 0.5233, + "step": 1266 + }, + { + "epoch": 0.53, + "learning_rate": 1.993105770642887e-05, + "loss": 0.4715, + "step": 1267 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930858824979192e-05, + "loss": 0.5269, + "step": 1268 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930659658075784e-05, + "loss": 0.5421, + "step": 1269 + }, + { + "epoch": 0.53, + "learning_rate": 1.993046020572438e-05, + "loss": 0.5315, + "step": 1270 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930260467930706e-05, + "loss": 0.6047, + "step": 1271 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930060444700508e-05, + "loss": 0.4419, + "step": 1272 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929860136039537e-05, + "loss": 0.6234, + "step": 1273 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929659541953548e-05, + "loss": 0.4614, + "step": 1274 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929458662448305e-05, + "loss": 0.5827, + "step": 1275 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929257497529584e-05, + "loss": 0.4641, + "step": 1276 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929056047203166e-05, + "loss": 0.4859, + "step": 1277 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928854311474844e-05, + "loss": 0.4513, + "step": 1278 + }, + { + "epoch": 0.54, + "learning_rate": 1.992865229035042e-05, + "loss": 0.6051, + "step": 1279 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928449983835688e-05, + "loss": 0.5138, + "step": 1280 + }, + { + "epoch": 0.54, + "learning_rate": 1.992824739193648e-05, + "loss": 0.5685, + "step": 1281 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928044514658604e-05, + "loss": 0.5332, + "step": 1282 + }, + { + "epoch": 0.54, + "learning_rate": 1.99278413520079e-05, + "loss": 0.5414, + "step": 1283 + }, + { + "epoch": 0.54, + "learning_rate": 1.992763790399021e-05, + "loss": 0.6006, + "step": 1284 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927434170611373e-05, + "loss": 0.519, + "step": 1285 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927230151877253e-05, + "loss": 0.5043, + "step": 1286 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927025847793714e-05, + "loss": 0.5028, + "step": 1287 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926821258366622e-05, + "loss": 0.5702, + "step": 1288 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926616383601862e-05, + "loss": 0.5905, + "step": 1289 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926411223505323e-05, + "loss": 0.5257, + "step": 1290 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926205778082903e-05, + "loss": 0.4561, + "step": 1291 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926000047340507e-05, + "loss": 0.5467, + "step": 1292 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925794031284046e-05, + "loss": 0.5829, + "step": 1293 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925587729919447e-05, + "loss": 0.5134, + "step": 1294 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925381143252633e-05, + "loss": 0.5486, + "step": 1295 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925174271289548e-05, + "loss": 0.5339, + "step": 1296 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924967114036133e-05, + "loss": 0.5039, + "step": 1297 + }, + { + "epoch": 0.54, + "learning_rate": 1.992475967149835e-05, + "loss": 0.5455, + "step": 1298 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924551943682153e-05, + "loss": 0.5704, + "step": 1299 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924343930593516e-05, + "loss": 0.5534, + "step": 1300 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924135632238423e-05, + "loss": 0.4944, + "step": 1301 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923927048622857e-05, + "loss": 0.5032, + "step": 1302 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923718179752815e-05, + "loss": 0.557, + "step": 1303 + }, + { + "epoch": 0.55, + "learning_rate": 1.99235090256343e-05, + "loss": 0.4871, + "step": 1304 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923299586273323e-05, + "loss": 0.5246, + "step": 1305 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923089861675906e-05, + "loss": 0.5811, + "step": 1306 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922879851848077e-05, + "loss": 0.5093, + "step": 1307 + }, + { + "epoch": 0.55, + "learning_rate": 1.992266955679587e-05, + "loss": 0.5393, + "step": 1308 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922458976525335e-05, + "loss": 0.533, + "step": 1309 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922248111042518e-05, + "loss": 0.4694, + "step": 1310 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922036960353486e-05, + "loss": 0.5957, + "step": 1311 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921825524464306e-05, + "loss": 0.5424, + "step": 1312 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921613803381055e-05, + "loss": 0.5007, + "step": 1313 + }, + { + "epoch": 0.55, + "learning_rate": 1.992140179710982e-05, + "loss": 0.4663, + "step": 1314 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921189505656696e-05, + "loss": 0.5117, + "step": 1315 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920976929027782e-05, + "loss": 0.4508, + "step": 1316 + }, + { + "epoch": 0.55, + "learning_rate": 1.992076406722919e-05, + "loss": 0.5698, + "step": 1317 + }, + { + "epoch": 0.55, + "learning_rate": 1.992055092026704e-05, + "loss": 0.5356, + "step": 1318 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920337488147453e-05, + "loss": 0.5155, + "step": 1319 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920123770876573e-05, + "loss": 0.6008, + "step": 1320 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919909768460535e-05, + "loss": 0.5054, + "step": 1321 + }, + { + "epoch": 0.55, + "learning_rate": 1.991969548090549e-05, + "loss": 0.4738, + "step": 1322 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919480908217606e-05, + "loss": 0.4882, + "step": 1323 + }, + { + "epoch": 0.55, + "learning_rate": 1.991926605040305e-05, + "loss": 0.5006, + "step": 1324 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919050907467986e-05, + "loss": 0.5821, + "step": 1325 + }, + { + "epoch": 0.56, + "learning_rate": 1.9918835479418607e-05, + "loss": 0.5611, + "step": 1326 + }, + { + "epoch": 0.56, + "learning_rate": 1.9918619766261106e-05, + "loss": 0.5482, + "step": 1327 + }, + { + "epoch": 0.56, + "learning_rate": 1.991840376800168e-05, + "loss": 0.6395, + "step": 1328 + }, + { + "epoch": 0.56, + "learning_rate": 1.991818748464654e-05, + "loss": 0.5458, + "step": 1329 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917970916201904e-05, + "loss": 0.5189, + "step": 1330 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917754062673992e-05, + "loss": 0.5461, + "step": 1331 + }, + { + "epoch": 0.56, + "learning_rate": 1.991753692406904e-05, + "loss": 0.4995, + "step": 1332 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917319500393294e-05, + "loss": 0.609, + "step": 1333 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917101791652997e-05, + "loss": 0.5375, + "step": 1334 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916883797854407e-05, + "loss": 0.4678, + "step": 1335 + }, + { + "epoch": 0.56, + "learning_rate": 1.991666551900379e-05, + "loss": 0.5375, + "step": 1336 + }, + { + "epoch": 0.56, + "learning_rate": 1.991644695510743e-05, + "loss": 0.441, + "step": 1337 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916228106171596e-05, + "loss": 0.5054, + "step": 1338 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916008972202586e-05, + "loss": 0.5001, + "step": 1339 + }, + { + "epoch": 0.56, + "learning_rate": 1.99157895532067e-05, + "loss": 0.574, + "step": 1340 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915569849190243e-05, + "loss": 0.5413, + "step": 1341 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915349860159528e-05, + "loss": 0.5348, + "step": 1342 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915129586120878e-05, + "loss": 0.4805, + "step": 1343 + }, + { + "epoch": 0.56, + "learning_rate": 1.991490902708063e-05, + "loss": 0.3818, + "step": 1344 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914688183045116e-05, + "loss": 0.5991, + "step": 1345 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914467054020692e-05, + "loss": 0.5398, + "step": 1346 + }, + { + "epoch": 0.56, + "learning_rate": 1.991424564001371e-05, + "loss": 0.5381, + "step": 1347 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914023941030534e-05, + "loss": 0.5163, + "step": 1348 + }, + { + "epoch": 0.56, + "learning_rate": 1.9913801957077535e-05, + "loss": 0.5226, + "step": 1349 + }, + { + "epoch": 0.57, + "learning_rate": 1.99135796881611e-05, + "loss": 0.5332, + "step": 1350 + }, + { + "epoch": 0.57, + "learning_rate": 1.991335713428761e-05, + "loss": 0.5834, + "step": 1351 + }, + { + "epoch": 0.57, + "learning_rate": 1.991313429546347e-05, + "loss": 0.5745, + "step": 1352 + }, + { + "epoch": 0.57, + "learning_rate": 1.991291117169508e-05, + "loss": 0.5629, + "step": 1353 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912687762988855e-05, + "loss": 0.4963, + "step": 1354 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912464069351217e-05, + "loss": 0.4239, + "step": 1355 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912240090788595e-05, + "loss": 0.4935, + "step": 1356 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912015827307426e-05, + "loss": 0.53, + "step": 1357 + }, + { + "epoch": 0.57, + "learning_rate": 1.991179127891416e-05, + "loss": 0.469, + "step": 1358 + }, + { + "epoch": 0.57, + "learning_rate": 1.9911566445615253e-05, + "loss": 0.537, + "step": 1359 + }, + { + "epoch": 0.57, + "learning_rate": 1.9911341327417156e-05, + "loss": 0.5467, + "step": 1360 + }, + { + "epoch": 0.57, + "learning_rate": 1.991111592432635e-05, + "loss": 0.4584, + "step": 1361 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910890236349316e-05, + "loss": 0.6264, + "step": 1362 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910664263492533e-05, + "loss": 0.5039, + "step": 1363 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910438005762503e-05, + "loss": 0.4549, + "step": 1364 + }, + { + "epoch": 0.57, + "learning_rate": 1.991021146316572e-05, + "loss": 0.5509, + "step": 1365 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909984635708714e-05, + "loss": 0.4958, + "step": 1366 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909757523397985e-05, + "loss": 0.5492, + "step": 1367 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909530126240074e-05, + "loss": 0.5841, + "step": 1368 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909302444241508e-05, + "loss": 0.545, + "step": 1369 + }, + { + "epoch": 0.57, + "learning_rate": 1.990907447740884e-05, + "loss": 0.4326, + "step": 1370 + }, + { + "epoch": 0.57, + "learning_rate": 1.9908846225748623e-05, + "loss": 0.5381, + "step": 1371 + }, + { + "epoch": 0.57, + "learning_rate": 1.990861768926741e-05, + "loss": 0.4573, + "step": 1372 + }, + { + "epoch": 0.57, + "learning_rate": 1.9908388867971772e-05, + "loss": 0.4962, + "step": 1373 + }, + { + "epoch": 0.58, + "learning_rate": 1.9908159761868292e-05, + "loss": 0.4974, + "step": 1374 + }, + { + "epoch": 0.58, + "learning_rate": 1.990793037096355e-05, + "loss": 0.4819, + "step": 1375 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907700695264144e-05, + "loss": 0.532, + "step": 1376 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907470734776674e-05, + "loss": 0.4813, + "step": 1377 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907240489507748e-05, + "loss": 0.483, + "step": 1378 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907009959463983e-05, + "loss": 0.6402, + "step": 1379 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906779144652014e-05, + "loss": 0.5279, + "step": 1380 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906548045078464e-05, + "loss": 0.5868, + "step": 1381 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906316660749985e-05, + "loss": 0.4664, + "step": 1382 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906084991673224e-05, + "loss": 0.571, + "step": 1383 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905853037854834e-05, + "loss": 0.5611, + "step": 1384 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905620799301495e-05, + "loss": 0.4523, + "step": 1385 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905388276019873e-05, + "loss": 0.6086, + "step": 1386 + }, + { + "epoch": 0.58, + "learning_rate": 1.990515546801666e-05, + "loss": 0.483, + "step": 1387 + }, + { + "epoch": 0.58, + "learning_rate": 1.990492237529854e-05, + "loss": 0.6169, + "step": 1388 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904688997872215e-05, + "loss": 0.4782, + "step": 1389 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904455335744395e-05, + "loss": 0.5676, + "step": 1390 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904221388921794e-05, + "loss": 0.6075, + "step": 1391 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903987157411138e-05, + "loss": 0.4489, + "step": 1392 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903752641219163e-05, + "loss": 0.5618, + "step": 1393 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903517840352605e-05, + "loss": 0.5132, + "step": 1394 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903282754818216e-05, + "loss": 0.4228, + "step": 1395 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903047384622748e-05, + "loss": 0.513, + "step": 1396 + }, + { + "epoch": 0.58, + "learning_rate": 1.9902811729772974e-05, + "loss": 0.5782, + "step": 1397 + }, + { + "epoch": 0.59, + "learning_rate": 1.990257579027566e-05, + "loss": 0.5138, + "step": 1398 + }, + { + "epoch": 0.59, + "learning_rate": 1.99023395661376e-05, + "loss": 0.5087, + "step": 1399 + }, + { + "epoch": 0.59, + "learning_rate": 1.9902103057365573e-05, + "loss": 0.6183, + "step": 1400 + }, + { + "epoch": 0.59, + "learning_rate": 1.990186626396638e-05, + "loss": 0.5741, + "step": 1401 + }, + { + "epoch": 0.59, + "learning_rate": 1.9901629185946827e-05, + "loss": 0.4439, + "step": 1402 + }, + { + "epoch": 0.59, + "learning_rate": 1.990139182331373e-05, + "loss": 0.6, + "step": 1403 + }, + { + "epoch": 0.59, + "learning_rate": 1.9901154176073914e-05, + "loss": 0.5193, + "step": 1404 + }, + { + "epoch": 0.59, + "learning_rate": 1.9900916244234203e-05, + "loss": 0.6317, + "step": 1405 + }, + { + "epoch": 0.59, + "learning_rate": 1.9900678027801445e-05, + "loss": 0.5093, + "step": 1406 + }, + { + "epoch": 0.59, + "learning_rate": 1.990043952678248e-05, + "loss": 0.4804, + "step": 1407 + }, + { + "epoch": 0.59, + "learning_rate": 1.990020074118417e-05, + "loss": 0.5628, + "step": 1408 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899961671013367e-05, + "loss": 0.5073, + "step": 1409 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899722316276958e-05, + "loss": 0.5252, + "step": 1410 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899482676981812e-05, + "loss": 0.5157, + "step": 1411 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899242753134824e-05, + "loss": 0.5309, + "step": 1412 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899002544742886e-05, + "loss": 0.4717, + "step": 1413 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898762051812906e-05, + "loss": 0.4811, + "step": 1414 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898521274351793e-05, + "loss": 0.496, + "step": 1415 + }, + { + "epoch": 0.59, + "learning_rate": 1.989828021236647e-05, + "loss": 0.4783, + "step": 1416 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898038865863864e-05, + "loss": 0.4625, + "step": 1417 + }, + { + "epoch": 0.59, + "learning_rate": 1.9897797234850916e-05, + "loss": 0.4816, + "step": 1418 + }, + { + "epoch": 0.59, + "learning_rate": 1.989755531933457e-05, + "loss": 0.5341, + "step": 1419 + }, + { + "epoch": 0.59, + "learning_rate": 1.989731311932178e-05, + "loss": 0.5596, + "step": 1420 + }, + { + "epoch": 0.59, + "learning_rate": 1.9897070634819506e-05, + "loss": 0.4642, + "step": 1421 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896827865834717e-05, + "loss": 0.4795, + "step": 1422 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896584812374397e-05, + "loss": 0.4566, + "step": 1423 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896341474445526e-05, + "loss": 0.5784, + "step": 1424 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896097852055103e-05, + "loss": 0.5538, + "step": 1425 + }, + { + "epoch": 0.6, + "learning_rate": 1.989585394521013e-05, + "loss": 0.5432, + "step": 1426 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895609753917614e-05, + "loss": 0.4524, + "step": 1427 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895365278184577e-05, + "loss": 0.4955, + "step": 1428 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895120518018046e-05, + "loss": 0.443, + "step": 1429 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894875473425057e-05, + "loss": 0.5583, + "step": 1430 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894630144412656e-05, + "loss": 0.5227, + "step": 1431 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894384530987885e-05, + "loss": 0.5223, + "step": 1432 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894138633157816e-05, + "loss": 0.4718, + "step": 1433 + }, + { + "epoch": 0.6, + "learning_rate": 1.989389245092951e-05, + "loss": 0.4776, + "step": 1434 + }, + { + "epoch": 0.6, + "learning_rate": 1.9893645984310047e-05, + "loss": 0.4808, + "step": 1435 + }, + { + "epoch": 0.6, + "learning_rate": 1.989339923330651e-05, + "loss": 0.5571, + "step": 1436 + }, + { + "epoch": 0.6, + "learning_rate": 1.989315219792599e-05, + "loss": 0.4956, + "step": 1437 + }, + { + "epoch": 0.6, + "learning_rate": 1.989290487817559e-05, + "loss": 0.5159, + "step": 1438 + }, + { + "epoch": 0.6, + "learning_rate": 1.9892657274062415e-05, + "loss": 0.5385, + "step": 1439 + }, + { + "epoch": 0.6, + "learning_rate": 1.9892409385593586e-05, + "loss": 0.6072, + "step": 1440 + }, + { + "epoch": 0.6, + "learning_rate": 1.989216121277623e-05, + "loss": 0.4643, + "step": 1441 + }, + { + "epoch": 0.6, + "learning_rate": 1.9891912755617477e-05, + "loss": 0.4857, + "step": 1442 + }, + { + "epoch": 0.6, + "learning_rate": 1.989166401412447e-05, + "loss": 0.4994, + "step": 1443 + }, + { + "epoch": 0.6, + "learning_rate": 1.9891414988304357e-05, + "loss": 0.4851, + "step": 1444 + }, + { + "epoch": 0.6, + "learning_rate": 1.98911656781643e-05, + "loss": 0.5248, + "step": 1445 + }, + { + "epoch": 0.61, + "learning_rate": 1.9890916083711463e-05, + "loss": 0.5135, + "step": 1446 + }, + { + "epoch": 0.61, + "learning_rate": 1.9890666204953018e-05, + "loss": 0.5424, + "step": 1447 + }, + { + "epoch": 0.61, + "learning_rate": 1.989041604189615e-05, + "loss": 0.4628, + "step": 1448 + }, + { + "epoch": 0.61, + "learning_rate": 1.989016559454805e-05, + "loss": 0.5106, + "step": 1449 + }, + { + "epoch": 0.61, + "learning_rate": 1.988991486291592e-05, + "loss": 0.5695, + "step": 1450 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889663847006956e-05, + "loss": 0.5715, + "step": 1451 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889412546828386e-05, + "loss": 0.4649, + "step": 1452 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889160962387427e-05, + "loss": 0.5909, + "step": 1453 + }, + { + "epoch": 0.61, + "learning_rate": 1.988890909369131e-05, + "loss": 0.5654, + "step": 1454 + }, + { + "epoch": 0.61, + "learning_rate": 1.988865694074728e-05, + "loss": 0.5696, + "step": 1455 + }, + { + "epoch": 0.61, + "learning_rate": 1.988840450356258e-05, + "loss": 0.486, + "step": 1456 + }, + { + "epoch": 0.61, + "learning_rate": 1.9888151782144463e-05, + "loss": 0.4263, + "step": 1457 + }, + { + "epoch": 0.61, + "learning_rate": 1.9887898776500203e-05, + "loss": 0.4926, + "step": 1458 + }, + { + "epoch": 0.61, + "learning_rate": 1.9887645486637063e-05, + "loss": 0.5234, + "step": 1459 + }, + { + "epoch": 0.61, + "learning_rate": 1.988739191256233e-05, + "loss": 0.4934, + "step": 1460 + }, + { + "epoch": 0.61, + "learning_rate": 1.988713805428329e-05, + "loss": 0.5068, + "step": 1461 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886883911807242e-05, + "loss": 0.5427, + "step": 1462 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886629485141488e-05, + "loss": 0.524, + "step": 1463 + }, + { + "epoch": 0.61, + "learning_rate": 1.988637477429334e-05, + "loss": 0.4719, + "step": 1464 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886119779270126e-05, + "loss": 0.5084, + "step": 1465 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885864500079172e-05, + "loss": 0.4963, + "step": 1466 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885608936727814e-05, + "loss": 0.5211, + "step": 1467 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885353089223397e-05, + "loss": 0.4711, + "step": 1468 + }, + { + "epoch": 0.62, + "learning_rate": 1.988509695757328e-05, + "loss": 0.5843, + "step": 1469 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884840541784824e-05, + "loss": 0.5093, + "step": 1470 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884583841865395e-05, + "loss": 0.4973, + "step": 1471 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884326857822378e-05, + "loss": 0.5495, + "step": 1472 + }, + { + "epoch": 0.62, + "learning_rate": 1.988406958966316e-05, + "loss": 0.6042, + "step": 1473 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883812037395126e-05, + "loss": 0.4894, + "step": 1474 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883554201025687e-05, + "loss": 0.5177, + "step": 1475 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883296080562255e-05, + "loss": 0.4454, + "step": 1476 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883037676012245e-05, + "loss": 0.534, + "step": 1477 + }, + { + "epoch": 0.62, + "learning_rate": 1.988277898738309e-05, + "loss": 0.4713, + "step": 1478 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882520014682218e-05, + "loss": 0.5647, + "step": 1479 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882260757917083e-05, + "loss": 0.5868, + "step": 1480 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882001217095127e-05, + "loss": 0.5463, + "step": 1481 + }, + { + "epoch": 0.62, + "learning_rate": 1.9881741392223822e-05, + "loss": 0.5546, + "step": 1482 + }, + { + "epoch": 0.62, + "learning_rate": 1.9881481283310624e-05, + "loss": 0.5102, + "step": 1483 + }, + { + "epoch": 0.62, + "learning_rate": 1.988122089036302e-05, + "loss": 0.4513, + "step": 1484 + }, + { + "epoch": 0.62, + "learning_rate": 1.9880960213388483e-05, + "loss": 0.5327, + "step": 1485 + }, + { + "epoch": 0.62, + "learning_rate": 1.988069925239452e-05, + "loss": 0.5329, + "step": 1486 + }, + { + "epoch": 0.62, + "learning_rate": 1.988043800738862e-05, + "loss": 0.522, + "step": 1487 + }, + { + "epoch": 0.62, + "learning_rate": 1.98801764783783e-05, + "loss": 0.5105, + "step": 1488 + }, + { + "epoch": 0.62, + "learning_rate": 1.987991466537107e-05, + "loss": 0.5621, + "step": 1489 + }, + { + "epoch": 0.62, + "learning_rate": 1.9879652568374467e-05, + "loss": 0.5259, + "step": 1490 + }, + { + "epoch": 0.62, + "learning_rate": 1.987939018739601e-05, + "loss": 0.464, + "step": 1491 + }, + { + "epoch": 0.62, + "learning_rate": 1.9879127522443255e-05, + "loss": 0.5808, + "step": 1492 + }, + { + "epoch": 0.63, + "learning_rate": 1.9878864573523744e-05, + "loss": 0.5933, + "step": 1493 + }, + { + "epoch": 0.63, + "learning_rate": 1.987860134064504e-05, + "loss": 0.4712, + "step": 1494 + }, + { + "epoch": 0.63, + "learning_rate": 1.98783378238147e-05, + "loss": 0.5199, + "step": 1495 + }, + { + "epoch": 0.63, + "learning_rate": 1.9878074023040313e-05, + "loss": 0.5006, + "step": 1496 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877809938329447e-05, + "loss": 0.5374, + "step": 1497 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877545569689705e-05, + "loss": 0.4978, + "step": 1498 + }, + { + "epoch": 0.63, + "learning_rate": 1.987728091712868e-05, + "loss": 0.4904, + "step": 1499 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877015980653978e-05, + "loss": 0.5521, + "step": 1500 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876750760273218e-05, + "loss": 0.6263, + "step": 1501 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876485255994022e-05, + "loss": 0.5446, + "step": 1502 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876219467824018e-05, + "loss": 0.4413, + "step": 1503 + }, + { + "epoch": 0.63, + "learning_rate": 1.987595339577085e-05, + "loss": 0.4664, + "step": 1504 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875687039842168e-05, + "loss": 0.5025, + "step": 1505 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875420400045625e-05, + "loss": 0.5362, + "step": 1506 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875153476388887e-05, + "loss": 0.5398, + "step": 1507 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874886268879625e-05, + "loss": 0.4848, + "step": 1508 + }, + { + "epoch": 0.63, + "learning_rate": 1.987461877752552e-05, + "loss": 0.4712, + "step": 1509 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874351002334257e-05, + "loss": 0.5298, + "step": 1510 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874082943313543e-05, + "loss": 0.5308, + "step": 1511 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873814600471073e-05, + "loss": 0.5389, + "step": 1512 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873545973814564e-05, + "loss": 0.4738, + "step": 1513 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873277063351737e-05, + "loss": 0.5454, + "step": 1514 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873007869090326e-05, + "loss": 0.5225, + "step": 1515 + }, + { + "epoch": 0.63, + "learning_rate": 1.9872738391038063e-05, + "loss": 0.4892, + "step": 1516 + }, + { + "epoch": 0.64, + "learning_rate": 1.9872468629202696e-05, + "loss": 0.5003, + "step": 1517 + }, + { + "epoch": 0.64, + "learning_rate": 1.9872198583591976e-05, + "loss": 0.5106, + "step": 1518 + }, + { + "epoch": 0.64, + "learning_rate": 1.987192825421367e-05, + "loss": 0.6494, + "step": 1519 + }, + { + "epoch": 0.64, + "learning_rate": 1.987165764107555e-05, + "loss": 0.5218, + "step": 1520 + }, + { + "epoch": 0.64, + "learning_rate": 1.9871386744185386e-05, + "loss": 0.5341, + "step": 1521 + }, + { + "epoch": 0.64, + "learning_rate": 1.9871115563550974e-05, + "loss": 0.4962, + "step": 1522 + }, + { + "epoch": 0.64, + "learning_rate": 1.98708440991801e-05, + "loss": 0.5314, + "step": 1523 + }, + { + "epoch": 0.64, + "learning_rate": 1.9870572351080573e-05, + "loss": 0.5039, + "step": 1524 + }, + { + "epoch": 0.64, + "learning_rate": 1.9870300319260204e-05, + "loss": 0.4734, + "step": 1525 + }, + { + "epoch": 0.64, + "learning_rate": 1.987002800372681e-05, + "loss": 0.565, + "step": 1526 + }, + { + "epoch": 0.64, + "learning_rate": 1.986975540448822e-05, + "loss": 0.5463, + "step": 1527 + }, + { + "epoch": 0.64, + "learning_rate": 1.986948252155227e-05, + "loss": 0.504, + "step": 1528 + }, + { + "epoch": 0.64, + "learning_rate": 1.9869209354926802e-05, + "loss": 0.4311, + "step": 1529 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868935904619673e-05, + "loss": 0.4794, + "step": 1530 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868662170638732e-05, + "loss": 0.5152, + "step": 1531 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868388152991858e-05, + "loss": 0.5112, + "step": 1532 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868113851686925e-05, + "loss": 0.5007, + "step": 1533 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867839266731813e-05, + "loss": 0.5184, + "step": 1534 + }, + { + "epoch": 0.64, + "learning_rate": 1.986756439813442e-05, + "loss": 0.4357, + "step": 1535 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867289245902645e-05, + "loss": 0.532, + "step": 1536 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867013810044394e-05, + "loss": 0.5021, + "step": 1537 + }, + { + "epoch": 0.64, + "learning_rate": 1.9866738090567586e-05, + "loss": 0.4668, + "step": 1538 + }, + { + "epoch": 0.64, + "learning_rate": 1.9866462087480153e-05, + "loss": 0.4493, + "step": 1539 + }, + { + "epoch": 0.64, + "learning_rate": 1.986618580079002e-05, + "loss": 0.503, + "step": 1540 + }, + { + "epoch": 0.65, + "learning_rate": 1.986590923050513e-05, + "loss": 0.5003, + "step": 1541 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865632376633433e-05, + "loss": 0.5519, + "step": 1542 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865355239182892e-05, + "loss": 0.4793, + "step": 1543 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865077818161464e-05, + "loss": 0.4704, + "step": 1544 + }, + { + "epoch": 0.65, + "learning_rate": 1.9864800113577133e-05, + "loss": 0.5161, + "step": 1545 + }, + { + "epoch": 0.65, + "learning_rate": 1.9864522125437874e-05, + "loss": 0.4979, + "step": 1546 + }, + { + "epoch": 0.65, + "learning_rate": 1.986424385375168e-05, + "loss": 0.5197, + "step": 1547 + }, + { + "epoch": 0.65, + "learning_rate": 1.986396529852655e-05, + "loss": 0.4897, + "step": 1548 + }, + { + "epoch": 0.65, + "learning_rate": 1.986368645977049e-05, + "loss": 0.4824, + "step": 1549 + }, + { + "epoch": 0.65, + "learning_rate": 1.9863407337491516e-05, + "loss": 0.565, + "step": 1550 + }, + { + "epoch": 0.65, + "learning_rate": 1.986312793169765e-05, + "loss": 0.5372, + "step": 1551 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862848242396927e-05, + "loss": 0.485, + "step": 1552 + }, + { + "epoch": 0.65, + "learning_rate": 1.986256826959738e-05, + "loss": 0.5229, + "step": 1553 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862288013307062e-05, + "loss": 0.583, + "step": 1554 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862007473534026e-05, + "loss": 0.5444, + "step": 1555 + }, + { + "epoch": 0.65, + "learning_rate": 1.9861726650286338e-05, + "loss": 0.5535, + "step": 1556 + }, + { + "epoch": 0.65, + "learning_rate": 1.986144554357207e-05, + "loss": 0.4921, + "step": 1557 + }, + { + "epoch": 0.65, + "learning_rate": 1.9861164153399297e-05, + "loss": 0.5669, + "step": 1558 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860882479776113e-05, + "loss": 0.5139, + "step": 1559 + }, + { + "epoch": 0.65, + "learning_rate": 1.986060052271061e-05, + "loss": 0.5647, + "step": 1560 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860318282210896e-05, + "loss": 0.6061, + "step": 1561 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860035758285082e-05, + "loss": 0.5984, + "step": 1562 + }, + { + "epoch": 0.65, + "learning_rate": 1.9859752950941292e-05, + "loss": 0.4112, + "step": 1563 + }, + { + "epoch": 0.65, + "learning_rate": 1.9859469860187654e-05, + "loss": 0.4325, + "step": 1564 + }, + { + "epoch": 0.66, + "learning_rate": 1.9859186486032302e-05, + "loss": 0.5618, + "step": 1565 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858902828483382e-05, + "loss": 0.5109, + "step": 1566 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858618887549053e-05, + "loss": 0.5181, + "step": 1567 + }, + { + "epoch": 0.66, + "learning_rate": 1.985833466323747e-05, + "loss": 0.5925, + "step": 1568 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858050155556807e-05, + "loss": 0.5414, + "step": 1569 + }, + { + "epoch": 0.66, + "learning_rate": 1.9857765364515234e-05, + "loss": 0.5568, + "step": 1570 + }, + { + "epoch": 0.66, + "learning_rate": 1.9857480290120947e-05, + "loss": 0.5295, + "step": 1571 + }, + { + "epoch": 0.66, + "learning_rate": 1.985719493238214e-05, + "loss": 0.5144, + "step": 1572 + }, + { + "epoch": 0.66, + "learning_rate": 1.9856909291307008e-05, + "loss": 0.4398, + "step": 1573 + }, + { + "epoch": 0.66, + "learning_rate": 1.9856623366903765e-05, + "loss": 0.5049, + "step": 1574 + }, + { + "epoch": 0.66, + "learning_rate": 1.985633715918063e-05, + "loss": 0.539, + "step": 1575 + }, + { + "epoch": 0.66, + "learning_rate": 1.985605066814583e-05, + "loss": 0.4789, + "step": 1576 + }, + { + "epoch": 0.66, + "learning_rate": 1.9855763893807603e-05, + "loss": 0.4971, + "step": 1577 + }, + { + "epoch": 0.66, + "learning_rate": 1.9855476836174182e-05, + "loss": 0.4828, + "step": 1578 + }, + { + "epoch": 0.66, + "learning_rate": 1.985518949525383e-05, + "loss": 0.4806, + "step": 1579 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854901871054803e-05, + "loss": 0.5103, + "step": 1580 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854613963585362e-05, + "loss": 0.4901, + "step": 1581 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854325772853792e-05, + "loss": 0.5219, + "step": 1582 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854037298868367e-05, + "loss": 0.5073, + "step": 1583 + }, + { + "epoch": 0.66, + "learning_rate": 1.9853748541637387e-05, + "loss": 0.5223, + "step": 1584 + }, + { + "epoch": 0.66, + "learning_rate": 1.985345950116915e-05, + "loss": 0.525, + "step": 1585 + }, + { + "epoch": 0.66, + "learning_rate": 1.9853170177471962e-05, + "loss": 0.4888, + "step": 1586 + }, + { + "epoch": 0.66, + "learning_rate": 1.9852880570554142e-05, + "loss": 0.5425, + "step": 1587 + }, + { + "epoch": 0.66, + "learning_rate": 1.9852590680424012e-05, + "loss": 0.5606, + "step": 1588 + }, + { + "epoch": 0.67, + "learning_rate": 1.9852300507089907e-05, + "loss": 0.5552, + "step": 1589 + }, + { + "epoch": 0.67, + "learning_rate": 1.9852010050560165e-05, + "loss": 0.4529, + "step": 1590 + }, + { + "epoch": 0.67, + "learning_rate": 1.985171931084314e-05, + "loss": 0.3869, + "step": 1591 + }, + { + "epoch": 0.67, + "learning_rate": 1.9851428287947184e-05, + "loss": 0.4744, + "step": 1592 + }, + { + "epoch": 0.67, + "learning_rate": 1.9851136981880662e-05, + "loss": 0.5176, + "step": 1593 + }, + { + "epoch": 0.67, + "learning_rate": 1.985084539265195e-05, + "loss": 0.4943, + "step": 1594 + }, + { + "epoch": 0.67, + "learning_rate": 1.985055352026943e-05, + "loss": 0.4605, + "step": 1595 + }, + { + "epoch": 0.67, + "learning_rate": 1.985026136474149e-05, + "loss": 0.564, + "step": 1596 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849968926076528e-05, + "loss": 0.5434, + "step": 1597 + }, + { + "epoch": 0.67, + "learning_rate": 1.984967620428295e-05, + "loss": 0.5173, + "step": 1598 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849383199369167e-05, + "loss": 0.5304, + "step": 1599 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849089911343608e-05, + "loss": 0.5106, + "step": 1600 + }, + { + "epoch": 0.67, + "learning_rate": 1.98487963402147e-05, + "loss": 0.4964, + "step": 1601 + }, + { + "epoch": 0.67, + "learning_rate": 1.9848502485990876e-05, + "loss": 0.4417, + "step": 1602 + }, + { + "epoch": 0.67, + "learning_rate": 1.9848208348680594e-05, + "loss": 0.4463, + "step": 1603 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847913928292297e-05, + "loss": 0.5096, + "step": 1604 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847619224834455e-05, + "loss": 0.5129, + "step": 1605 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847324238315536e-05, + "loss": 0.4326, + "step": 1606 + }, + { + "epoch": 0.67, + "learning_rate": 1.984702896874402e-05, + "loss": 0.5151, + "step": 1607 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846733416128393e-05, + "loss": 0.6225, + "step": 1608 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846437580477155e-05, + "loss": 0.5483, + "step": 1609 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846141461798804e-05, + "loss": 0.5227, + "step": 1610 + }, + { + "epoch": 0.67, + "learning_rate": 1.984584506010186e-05, + "loss": 0.4964, + "step": 1611 + }, + { + "epoch": 0.67, + "learning_rate": 1.984554837539483e-05, + "loss": 0.5653, + "step": 1612 + }, + { + "epoch": 0.68, + "learning_rate": 1.984525140768625e-05, + "loss": 0.5379, + "step": 1613 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844954156984656e-05, + "loss": 0.5595, + "step": 1614 + }, + { + "epoch": 0.68, + "learning_rate": 1.984465662329859e-05, + "loss": 0.5941, + "step": 1615 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844358806636605e-05, + "loss": 0.4752, + "step": 1616 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844060707007263e-05, + "loss": 0.5019, + "step": 1617 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843762324419132e-05, + "loss": 0.5074, + "step": 1618 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843463658880786e-05, + "loss": 0.628, + "step": 1619 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843164710400815e-05, + "loss": 0.5083, + "step": 1620 + }, + { + "epoch": 0.68, + "learning_rate": 1.9842865478987808e-05, + "loss": 0.4711, + "step": 1621 + }, + { + "epoch": 0.68, + "learning_rate": 1.9842565964650363e-05, + "loss": 0.5057, + "step": 1622 + }, + { + "epoch": 0.68, + "learning_rate": 1.98422661673971e-05, + "loss": 0.5801, + "step": 1623 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841966087236626e-05, + "loss": 0.4771, + "step": 1624 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841665724177572e-05, + "loss": 0.5187, + "step": 1625 + }, + { + "epoch": 0.68, + "learning_rate": 1.984136507822857e-05, + "loss": 0.4719, + "step": 1626 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841064149398262e-05, + "loss": 0.4859, + "step": 1627 + }, + { + "epoch": 0.68, + "learning_rate": 1.9840762937695296e-05, + "loss": 0.4938, + "step": 1628 + }, + { + "epoch": 0.68, + "learning_rate": 1.9840461443128333e-05, + "loss": 0.5628, + "step": 1629 + }, + { + "epoch": 0.68, + "learning_rate": 1.984015966570604e-05, + "loss": 0.5753, + "step": 1630 + }, + { + "epoch": 0.68, + "learning_rate": 1.9839857605437086e-05, + "loss": 0.4204, + "step": 1631 + }, + { + "epoch": 0.68, + "learning_rate": 1.983955526233016e-05, + "loss": 0.4693, + "step": 1632 + }, + { + "epoch": 0.68, + "learning_rate": 1.9839252636393946e-05, + "loss": 0.5352, + "step": 1633 + }, + { + "epoch": 0.68, + "learning_rate": 1.983894972763715e-05, + "loss": 0.5136, + "step": 1634 + }, + { + "epoch": 0.68, + "learning_rate": 1.9838646536068473e-05, + "loss": 0.5446, + "step": 1635 + }, + { + "epoch": 0.68, + "learning_rate": 1.9838343061696633e-05, + "loss": 0.5841, + "step": 1636 + }, + { + "epoch": 0.69, + "learning_rate": 1.983803930453035e-05, + "loss": 0.5229, + "step": 1637 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837735264578358e-05, + "loss": 0.5148, + "step": 1638 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837430941849395e-05, + "loss": 0.4822, + "step": 1639 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837126336352214e-05, + "loss": 0.5169, + "step": 1640 + }, + { + "epoch": 0.69, + "learning_rate": 1.983682144809556e-05, + "loss": 0.4126, + "step": 1641 + }, + { + "epoch": 0.69, + "learning_rate": 1.9836516277088203e-05, + "loss": 0.4774, + "step": 1642 + }, + { + "epoch": 0.69, + "learning_rate": 1.9836210823338917e-05, + "loss": 0.5266, + "step": 1643 + }, + { + "epoch": 0.69, + "learning_rate": 1.983590508685648e-05, + "loss": 0.4671, + "step": 1644 + }, + { + "epoch": 0.69, + "learning_rate": 1.9835599067649676e-05, + "loss": 0.5408, + "step": 1645 + }, + { + "epoch": 0.69, + "learning_rate": 1.983529276572731e-05, + "loss": 0.5083, + "step": 1646 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834986181098176e-05, + "loss": 0.5176, + "step": 1647 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834679313771095e-05, + "loss": 0.5439, + "step": 1648 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834372163754882e-05, + "loss": 0.5107, + "step": 1649 + }, + { + "epoch": 0.69, + "learning_rate": 1.983406473105837e-05, + "loss": 0.5676, + "step": 1650 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833757015690393e-05, + "loss": 0.4696, + "step": 1651 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833449017659796e-05, + "loss": 0.5984, + "step": 1652 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833140736975433e-05, + "loss": 0.5636, + "step": 1653 + }, + { + "epoch": 0.69, + "learning_rate": 1.9832832173646164e-05, + "loss": 0.5231, + "step": 1654 + }, + { + "epoch": 0.69, + "learning_rate": 1.9832523327680865e-05, + "loss": 0.5034, + "step": 1655 + }, + { + "epoch": 0.69, + "learning_rate": 1.98322141990884e-05, + "loss": 0.5617, + "step": 1656 + }, + { + "epoch": 0.69, + "learning_rate": 1.983190478787767e-05, + "loss": 0.4909, + "step": 1657 + }, + { + "epoch": 0.69, + "learning_rate": 1.983159509405756e-05, + "loss": 0.4736, + "step": 1658 + }, + { + "epoch": 0.69, + "learning_rate": 1.9831285117636972e-05, + "loss": 0.592, + "step": 1659 + }, + { + "epoch": 0.69, + "learning_rate": 1.983097485862482e-05, + "loss": 0.5005, + "step": 1660 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830664317030014e-05, + "loss": 0.5292, + "step": 1661 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830353492861493e-05, + "loss": 0.5129, + "step": 1662 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830042386128178e-05, + "loss": 0.575, + "step": 1663 + }, + { + "epoch": 0.7, + "learning_rate": 1.982973099683902e-05, + "loss": 0.4696, + "step": 1664 + }, + { + "epoch": 0.7, + "learning_rate": 1.9829419325002966e-05, + "loss": 0.4746, + "step": 1665 + }, + { + "epoch": 0.7, + "learning_rate": 1.982910737062898e-05, + "loss": 0.4449, + "step": 1666 + }, + { + "epoch": 0.7, + "learning_rate": 1.982879513372602e-05, + "loss": 0.4704, + "step": 1667 + }, + { + "epoch": 0.7, + "learning_rate": 1.9828482614303068e-05, + "loss": 0.5179, + "step": 1668 + }, + { + "epoch": 0.7, + "learning_rate": 1.9828169812369106e-05, + "loss": 0.4781, + "step": 1669 + }, + { + "epoch": 0.7, + "learning_rate": 1.9827856727933122e-05, + "loss": 0.5388, + "step": 1670 + }, + { + "epoch": 0.7, + "learning_rate": 1.9827543361004118e-05, + "loss": 0.5813, + "step": 1671 + }, + { + "epoch": 0.7, + "learning_rate": 1.98272297115911e-05, + "loss": 0.4643, + "step": 1672 + }, + { + "epoch": 0.7, + "learning_rate": 1.9826915779703087e-05, + "loss": 0.4306, + "step": 1673 + }, + { + "epoch": 0.7, + "learning_rate": 1.98266015653491e-05, + "loss": 0.5216, + "step": 1674 + }, + { + "epoch": 0.7, + "learning_rate": 1.9826287068538167e-05, + "loss": 0.5218, + "step": 1675 + }, + { + "epoch": 0.7, + "learning_rate": 1.9825972289279336e-05, + "loss": 0.519, + "step": 1676 + }, + { + "epoch": 0.7, + "learning_rate": 1.982565722758165e-05, + "loss": 0.4881, + "step": 1677 + }, + { + "epoch": 0.7, + "learning_rate": 1.9825341883454165e-05, + "loss": 0.5389, + "step": 1678 + }, + { + "epoch": 0.7, + "learning_rate": 1.982502625690595e-05, + "loss": 0.5541, + "step": 1679 + }, + { + "epoch": 0.7, + "learning_rate": 1.982471034794607e-05, + "loss": 0.4128, + "step": 1680 + }, + { + "epoch": 0.7, + "learning_rate": 1.982439415658361e-05, + "loss": 0.4948, + "step": 1681 + }, + { + "epoch": 0.7, + "learning_rate": 1.9824077682827658e-05, + "loss": 0.4972, + "step": 1682 + }, + { + "epoch": 0.7, + "learning_rate": 1.982376092668731e-05, + "loss": 0.5727, + "step": 1683 + }, + { + "epoch": 0.71, + "learning_rate": 1.9823443888171676e-05, + "loss": 0.5451, + "step": 1684 + }, + { + "epoch": 0.71, + "learning_rate": 1.982312656728986e-05, + "loss": 0.4172, + "step": 1685 + }, + { + "epoch": 0.71, + "learning_rate": 1.9822808964050987e-05, + "loss": 0.5163, + "step": 1686 + }, + { + "epoch": 0.71, + "learning_rate": 1.982249107846419e-05, + "loss": 0.5484, + "step": 1687 + }, + { + "epoch": 0.71, + "learning_rate": 1.98221729105386e-05, + "loss": 0.5193, + "step": 1688 + }, + { + "epoch": 0.71, + "learning_rate": 1.982185446028337e-05, + "loss": 0.483, + "step": 1689 + }, + { + "epoch": 0.71, + "learning_rate": 1.9821535727707647e-05, + "loss": 0.519, + "step": 1690 + }, + { + "epoch": 0.71, + "learning_rate": 1.9821216712820597e-05, + "loss": 0.5161, + "step": 1691 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820897415631383e-05, + "loss": 0.5653, + "step": 1692 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820577836149192e-05, + "loss": 0.4888, + "step": 1693 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820257974383205e-05, + "loss": 0.4977, + "step": 1694 + }, + { + "epoch": 0.71, + "learning_rate": 1.9819937830342618e-05, + "loss": 0.4693, + "step": 1695 + }, + { + "epoch": 0.71, + "learning_rate": 1.981961740403663e-05, + "loss": 0.493, + "step": 1696 + }, + { + "epoch": 0.71, + "learning_rate": 1.9819296695474454e-05, + "loss": 0.5228, + "step": 1697 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818975704665306e-05, + "loss": 0.4938, + "step": 1698 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818654431618416e-05, + "loss": 0.4252, + "step": 1699 + }, + { + "epoch": 0.71, + "learning_rate": 1.981833287634302e-05, + "loss": 0.4584, + "step": 1700 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818011038848355e-05, + "loss": 0.5541, + "step": 1701 + }, + { + "epoch": 0.71, + "learning_rate": 1.9817688919143676e-05, + "loss": 0.412, + "step": 1702 + }, + { + "epoch": 0.71, + "learning_rate": 1.981736651723824e-05, + "loss": 0.5411, + "step": 1703 + }, + { + "epoch": 0.71, + "learning_rate": 1.9817043833141317e-05, + "loss": 0.4207, + "step": 1704 + }, + { + "epoch": 0.71, + "learning_rate": 1.981672086686218e-05, + "loss": 0.502, + "step": 1705 + }, + { + "epoch": 0.71, + "learning_rate": 1.9816397618410112e-05, + "loss": 0.5266, + "step": 1706 + }, + { + "epoch": 0.71, + "learning_rate": 1.9816074087794406e-05, + "loss": 0.5749, + "step": 1707 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815750275024362e-05, + "loss": 0.5609, + "step": 1708 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815426180109286e-05, + "loss": 0.5089, + "step": 1709 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815101803058492e-05, + "loss": 0.5283, + "step": 1710 + }, + { + "epoch": 0.72, + "learning_rate": 1.981477714388131e-05, + "loss": 0.4316, + "step": 1711 + }, + { + "epoch": 0.72, + "learning_rate": 1.9814452202587066e-05, + "loss": 0.4375, + "step": 1712 + }, + { + "epoch": 0.72, + "learning_rate": 1.9814126979185105e-05, + "loss": 0.5121, + "step": 1713 + }, + { + "epoch": 0.72, + "learning_rate": 1.981380147368477e-05, + "loss": 0.4719, + "step": 1714 + }, + { + "epoch": 0.72, + "learning_rate": 1.981347568609542e-05, + "loss": 0.4747, + "step": 1715 + }, + { + "epoch": 0.72, + "learning_rate": 1.9813149616426424e-05, + "loss": 0.5267, + "step": 1716 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812823264687148e-05, + "loss": 0.4687, + "step": 1717 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812496630886974e-05, + "loss": 0.5243, + "step": 1718 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812169715035295e-05, + "loss": 0.5228, + "step": 1719 + }, + { + "epoch": 0.72, + "learning_rate": 1.98118425171415e-05, + "loss": 0.5324, + "step": 1720 + }, + { + "epoch": 0.72, + "learning_rate": 1.9811515037215002e-05, + "loss": 0.5004, + "step": 1721 + }, + { + "epoch": 0.72, + "learning_rate": 1.981118727526521e-05, + "loss": 0.4853, + "step": 1722 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810859231301545e-05, + "loss": 0.551, + "step": 1723 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810530905333437e-05, + "loss": 0.4566, + "step": 1724 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810202297370328e-05, + "loss": 0.4481, + "step": 1725 + }, + { + "epoch": 0.72, + "learning_rate": 1.9809873407421656e-05, + "loss": 0.5405, + "step": 1726 + }, + { + "epoch": 0.72, + "learning_rate": 1.980954423549688e-05, + "loss": 0.5531, + "step": 1727 + }, + { + "epoch": 0.72, + "learning_rate": 1.9809214781605458e-05, + "loss": 0.5072, + "step": 1728 + }, + { + "epoch": 0.72, + "learning_rate": 1.980888504575686e-05, + "loss": 0.6357, + "step": 1729 + }, + { + "epoch": 0.72, + "learning_rate": 1.980855502796057e-05, + "loss": 0.523, + "step": 1730 + }, + { + "epoch": 0.72, + "learning_rate": 1.9808224728226067e-05, + "loss": 0.4832, + "step": 1731 + }, + { + "epoch": 0.73, + "learning_rate": 1.9807894146562847e-05, + "loss": 0.4624, + "step": 1732 + }, + { + "epoch": 0.73, + "learning_rate": 1.9807563282980415e-05, + "loss": 0.444, + "step": 1733 + }, + { + "epoch": 0.73, + "learning_rate": 1.980723213748828e-05, + "loss": 0.485, + "step": 1734 + }, + { + "epoch": 0.73, + "learning_rate": 1.9806900710095958e-05, + "loss": 0.6331, + "step": 1735 + }, + { + "epoch": 0.73, + "learning_rate": 1.980656900081298e-05, + "loss": 0.551, + "step": 1736 + }, + { + "epoch": 0.73, + "learning_rate": 1.9806237009648876e-05, + "loss": 0.4904, + "step": 1737 + }, + { + "epoch": 0.73, + "learning_rate": 1.980590473661319e-05, + "loss": 0.5215, + "step": 1738 + }, + { + "epoch": 0.73, + "learning_rate": 1.9805572181715478e-05, + "loss": 0.5491, + "step": 1739 + }, + { + "epoch": 0.73, + "learning_rate": 1.980523934496529e-05, + "loss": 0.5062, + "step": 1740 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804906226372202e-05, + "loss": 0.5369, + "step": 1741 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804572825945784e-05, + "loss": 0.4987, + "step": 1742 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804239143695622e-05, + "loss": 0.4956, + "step": 1743 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803905179631303e-05, + "loss": 0.4876, + "step": 1744 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803570933762432e-05, + "loss": 0.5259, + "step": 1745 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803236406098612e-05, + "loss": 0.4896, + "step": 1746 + }, + { + "epoch": 0.73, + "learning_rate": 1.9802901596649462e-05, + "loss": 0.4446, + "step": 1747 + }, + { + "epoch": 0.73, + "learning_rate": 1.9802566505424603e-05, + "loss": 0.5677, + "step": 1748 + }, + { + "epoch": 0.73, + "learning_rate": 1.980223113243367e-05, + "loss": 0.5355, + "step": 1749 + }, + { + "epoch": 0.73, + "learning_rate": 1.98018954776863e-05, + "loss": 0.4811, + "step": 1750 + }, + { + "epoch": 0.73, + "learning_rate": 1.9801559541192147e-05, + "loss": 0.4993, + "step": 1751 + }, + { + "epoch": 0.73, + "learning_rate": 1.9801223322960858e-05, + "loss": 0.5684, + "step": 1752 + }, + { + "epoch": 0.73, + "learning_rate": 1.9800886823002105e-05, + "loss": 0.5324, + "step": 1753 + }, + { + "epoch": 0.73, + "learning_rate": 1.9800550041325553e-05, + "loss": 0.5637, + "step": 1754 + }, + { + "epoch": 0.73, + "learning_rate": 1.980021297794089e-05, + "loss": 0.5459, + "step": 1755 + }, + { + "epoch": 0.74, + "learning_rate": 1.9799875632857802e-05, + "loss": 0.5847, + "step": 1756 + }, + { + "epoch": 0.74, + "learning_rate": 1.9799538006085982e-05, + "loss": 0.5407, + "step": 1757 + }, + { + "epoch": 0.74, + "learning_rate": 1.979920009763514e-05, + "loss": 0.4979, + "step": 1758 + }, + { + "epoch": 0.74, + "learning_rate": 1.979886190751499e-05, + "loss": 0.438, + "step": 1759 + }, + { + "epoch": 0.74, + "learning_rate": 1.9798523435735246e-05, + "loss": 0.5051, + "step": 1760 + }, + { + "epoch": 0.74, + "learning_rate": 1.9798184682305645e-05, + "loss": 0.4852, + "step": 1761 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797845647235915e-05, + "loss": 0.5432, + "step": 1762 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797506330535808e-05, + "loss": 0.4505, + "step": 1763 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797166732215078e-05, + "loss": 0.6174, + "step": 1764 + }, + { + "epoch": 0.74, + "learning_rate": 1.979682685228348e-05, + "loss": 0.4877, + "step": 1765 + }, + { + "epoch": 0.74, + "learning_rate": 1.9796486690750792e-05, + "loss": 0.515, + "step": 1766 + }, + { + "epoch": 0.74, + "learning_rate": 1.9796146247626783e-05, + "loss": 0.4858, + "step": 1767 + }, + { + "epoch": 0.74, + "learning_rate": 1.9795805522921246e-05, + "loss": 0.5304, + "step": 1768 + }, + { + "epoch": 0.74, + "learning_rate": 1.979546451664397e-05, + "loss": 0.5342, + "step": 1769 + }, + { + "epoch": 0.74, + "learning_rate": 1.979512322880476e-05, + "loss": 0.5117, + "step": 1770 + }, + { + "epoch": 0.74, + "learning_rate": 1.9794781659413425e-05, + "loss": 0.5361, + "step": 1771 + }, + { + "epoch": 0.74, + "learning_rate": 1.9794439808479783e-05, + "loss": 0.4229, + "step": 1772 + }, + { + "epoch": 0.74, + "learning_rate": 1.979409767601366e-05, + "loss": 0.413, + "step": 1773 + }, + { + "epoch": 0.74, + "learning_rate": 1.979375526202489e-05, + "loss": 0.464, + "step": 1774 + }, + { + "epoch": 0.74, + "learning_rate": 1.9793412566523318e-05, + "loss": 0.5194, + "step": 1775 + }, + { + "epoch": 0.74, + "learning_rate": 1.9793069589518787e-05, + "loss": 0.5144, + "step": 1776 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792726331021166e-05, + "loss": 0.451, + "step": 1777 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792382791040317e-05, + "loss": 0.5394, + "step": 1778 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792038969586113e-05, + "loss": 0.5391, + "step": 1779 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791694866668434e-05, + "loss": 0.5101, + "step": 1780 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791350482297177e-05, + "loss": 0.4901, + "step": 1781 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791005816482242e-05, + "loss": 0.4868, + "step": 1782 + }, + { + "epoch": 0.75, + "learning_rate": 1.9790660869233532e-05, + "loss": 0.4358, + "step": 1783 + }, + { + "epoch": 0.75, + "learning_rate": 1.9790315640560962e-05, + "loss": 0.5186, + "step": 1784 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789970130474455e-05, + "loss": 0.5941, + "step": 1785 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789624338983948e-05, + "loss": 0.4653, + "step": 1786 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789278266099375e-05, + "loss": 0.4051, + "step": 1787 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788931911830688e-05, + "loss": 0.4873, + "step": 1788 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788585276187836e-05, + "loss": 0.5416, + "step": 1789 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788238359180786e-05, + "loss": 0.6702, + "step": 1790 + }, + { + "epoch": 0.75, + "learning_rate": 1.978789116081951e-05, + "loss": 0.5041, + "step": 1791 + }, + { + "epoch": 0.75, + "learning_rate": 1.9787543681113993e-05, + "loss": 0.4556, + "step": 1792 + }, + { + "epoch": 0.75, + "learning_rate": 1.9787195920074213e-05, + "loss": 0.5527, + "step": 1793 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786847877710177e-05, + "loss": 0.5213, + "step": 1794 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786499554031876e-05, + "loss": 0.5291, + "step": 1795 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786150949049335e-05, + "loss": 0.4473, + "step": 1796 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785802062772566e-05, + "loss": 0.5462, + "step": 1797 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785452895211606e-05, + "loss": 0.5014, + "step": 1798 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785103446376483e-05, + "loss": 0.5301, + "step": 1799 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784753716277242e-05, + "loss": 0.5775, + "step": 1800 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784403704923937e-05, + "loss": 0.4802, + "step": 1801 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784053412326637e-05, + "loss": 0.4719, + "step": 1802 + }, + { + "epoch": 0.75, + "learning_rate": 1.97837028384954e-05, + "loss": 0.5638, + "step": 1803 + }, + { + "epoch": 0.76, + "learning_rate": 1.9783351983440306e-05, + "loss": 0.5138, + "step": 1804 + }, + { + "epoch": 0.76, + "learning_rate": 1.978300084717144e-05, + "loss": 0.4597, + "step": 1805 + }, + { + "epoch": 0.76, + "learning_rate": 1.9782649429698896e-05, + "loss": 0.4601, + "step": 1806 + }, + { + "epoch": 0.76, + "learning_rate": 1.9782297731032778e-05, + "loss": 0.5567, + "step": 1807 + }, + { + "epoch": 0.76, + "learning_rate": 1.978194575118319e-05, + "loss": 0.5288, + "step": 1808 + }, + { + "epoch": 0.76, + "learning_rate": 1.9781593490160255e-05, + "loss": 0.5387, + "step": 1809 + }, + { + "epoch": 0.76, + "learning_rate": 1.9781240947974093e-05, + "loss": 0.4789, + "step": 1810 + }, + { + "epoch": 0.76, + "learning_rate": 1.978088812463484e-05, + "loss": 0.4638, + "step": 1811 + }, + { + "epoch": 0.76, + "learning_rate": 1.9780535020152632e-05, + "loss": 0.5389, + "step": 1812 + }, + { + "epoch": 0.76, + "learning_rate": 1.978018163453763e-05, + "loss": 0.4942, + "step": 1813 + }, + { + "epoch": 0.76, + "learning_rate": 1.9779827967799985e-05, + "loss": 0.4926, + "step": 1814 + }, + { + "epoch": 0.76, + "learning_rate": 1.9779474019949864e-05, + "loss": 0.5134, + "step": 1815 + }, + { + "epoch": 0.76, + "learning_rate": 1.977911979099744e-05, + "loss": 0.5047, + "step": 1816 + }, + { + "epoch": 0.76, + "learning_rate": 1.9778765280952896e-05, + "loss": 0.492, + "step": 1817 + }, + { + "epoch": 0.76, + "learning_rate": 1.977841048982642e-05, + "loss": 0.5855, + "step": 1818 + }, + { + "epoch": 0.76, + "learning_rate": 1.977805541762821e-05, + "loss": 0.487, + "step": 1819 + }, + { + "epoch": 0.76, + "learning_rate": 1.9777700064368475e-05, + "loss": 0.562, + "step": 1820 + }, + { + "epoch": 0.76, + "learning_rate": 1.977734443005743e-05, + "loss": 0.4709, + "step": 1821 + }, + { + "epoch": 0.76, + "learning_rate": 1.9776988514705294e-05, + "loss": 0.5106, + "step": 1822 + }, + { + "epoch": 0.76, + "learning_rate": 1.97766323183223e-05, + "loss": 0.5815, + "step": 1823 + }, + { + "epoch": 0.76, + "learning_rate": 1.9776275840918685e-05, + "loss": 0.4957, + "step": 1824 + }, + { + "epoch": 0.76, + "learning_rate": 1.97759190825047e-05, + "loss": 0.4592, + "step": 1825 + }, + { + "epoch": 0.76, + "learning_rate": 1.977556204309059e-05, + "loss": 0.5652, + "step": 1826 + }, + { + "epoch": 0.76, + "learning_rate": 1.9775204722686628e-05, + "loss": 0.4936, + "step": 1827 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774847121303078e-05, + "loss": 0.5499, + "step": 1828 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774489238950222e-05, + "loss": 0.5595, + "step": 1829 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774131075638347e-05, + "loss": 0.5254, + "step": 1830 + }, + { + "epoch": 0.77, + "learning_rate": 1.977377263137775e-05, + "loss": 0.5087, + "step": 1831 + }, + { + "epoch": 0.77, + "learning_rate": 1.977341390617873e-05, + "loss": 0.4864, + "step": 1832 + }, + { + "epoch": 0.77, + "learning_rate": 1.97730549000516e-05, + "loss": 0.5294, + "step": 1833 + }, + { + "epoch": 0.77, + "learning_rate": 1.9772695613006676e-05, + "loss": 0.4775, + "step": 1834 + }, + { + "epoch": 0.77, + "learning_rate": 1.977233604505429e-05, + "loss": 0.5719, + "step": 1835 + }, + { + "epoch": 0.77, + "learning_rate": 1.977197619620478e-05, + "loss": 0.5053, + "step": 1836 + }, + { + "epoch": 0.77, + "learning_rate": 1.9771616066468484e-05, + "loss": 0.4699, + "step": 1837 + }, + { + "epoch": 0.77, + "learning_rate": 1.9771255655855753e-05, + "loss": 0.5384, + "step": 1838 + }, + { + "epoch": 0.77, + "learning_rate": 1.9770894964376956e-05, + "loss": 0.5049, + "step": 1839 + }, + { + "epoch": 0.77, + "learning_rate": 1.9770533992042447e-05, + "loss": 0.5714, + "step": 1840 + }, + { + "epoch": 0.77, + "learning_rate": 1.977017273886261e-05, + "loss": 0.5266, + "step": 1841 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769811204847825e-05, + "loss": 0.5479, + "step": 1842 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769449390008492e-05, + "loss": 0.5449, + "step": 1843 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769087294355006e-05, + "loss": 0.4606, + "step": 1844 + }, + { + "epoch": 0.77, + "learning_rate": 1.976872491789777e-05, + "loss": 0.5291, + "step": 1845 + }, + { + "epoch": 0.77, + "learning_rate": 1.9768362260647205e-05, + "loss": 0.5553, + "step": 1846 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767999322613737e-05, + "loss": 0.5144, + "step": 1847 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767636103807796e-05, + "loss": 0.5458, + "step": 1848 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767272604239823e-05, + "loss": 0.559, + "step": 1849 + }, + { + "epoch": 0.77, + "learning_rate": 1.9766908823920268e-05, + "loss": 0.4688, + "step": 1850 + }, + { + "epoch": 0.77, + "learning_rate": 1.976654476285958e-05, + "loss": 0.558, + "step": 1851 + }, + { + "epoch": 0.78, + "learning_rate": 1.9766180421068236e-05, + "loss": 0.5534, + "step": 1852 + }, + { + "epoch": 0.78, + "learning_rate": 1.97658157985567e-05, + "loss": 0.469, + "step": 1853 + }, + { + "epoch": 0.78, + "learning_rate": 1.9765450895335454e-05, + "loss": 0.4969, + "step": 1854 + }, + { + "epoch": 0.78, + "learning_rate": 1.9765085711414987e-05, + "loss": 0.5451, + "step": 1855 + }, + { + "epoch": 0.78, + "learning_rate": 1.97647202468058e-05, + "loss": 0.5496, + "step": 1856 + }, + { + "epoch": 0.78, + "learning_rate": 1.9764354501518394e-05, + "loss": 0.4641, + "step": 1857 + }, + { + "epoch": 0.78, + "learning_rate": 1.976398847556328e-05, + "loss": 0.513, + "step": 1858 + }, + { + "epoch": 0.78, + "learning_rate": 1.9763622168950982e-05, + "loss": 0.6003, + "step": 1859 + }, + { + "epoch": 0.78, + "learning_rate": 1.9763255581692027e-05, + "loss": 0.5382, + "step": 1860 + }, + { + "epoch": 0.78, + "learning_rate": 1.976288871379696e-05, + "loss": 0.5244, + "step": 1861 + }, + { + "epoch": 0.78, + "learning_rate": 1.9762521565276314e-05, + "loss": 0.5568, + "step": 1862 + }, + { + "epoch": 0.78, + "learning_rate": 1.976215413614065e-05, + "loss": 0.509, + "step": 1863 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761786426400532e-05, + "loss": 0.5482, + "step": 1864 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761418436066523e-05, + "loss": 0.5047, + "step": 1865 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761050165149208e-05, + "loss": 0.4403, + "step": 1866 + }, + { + "epoch": 0.78, + "learning_rate": 1.976068161365916e-05, + "loss": 0.509, + "step": 1867 + }, + { + "epoch": 0.78, + "learning_rate": 1.9760312781606985e-05, + "loss": 0.5031, + "step": 1868 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759943669003282e-05, + "loss": 0.5258, + "step": 1869 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759574275858653e-05, + "loss": 0.5036, + "step": 1870 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759204602183725e-05, + "loss": 0.4541, + "step": 1871 + }, + { + "epoch": 0.78, + "learning_rate": 1.9758834647989123e-05, + "loss": 0.4063, + "step": 1872 + }, + { + "epoch": 0.78, + "learning_rate": 1.9758464413285477e-05, + "loss": 0.5494, + "step": 1873 + }, + { + "epoch": 0.78, + "learning_rate": 1.975809389808343e-05, + "loss": 0.5205, + "step": 1874 + }, + { + "epoch": 0.79, + "learning_rate": 1.9757723102393634e-05, + "loss": 0.4654, + "step": 1875 + }, + { + "epoch": 0.79, + "learning_rate": 1.9757352026226746e-05, + "loss": 0.4682, + "step": 1876 + }, + { + "epoch": 0.79, + "learning_rate": 1.9756980669593432e-05, + "loss": 0.5617, + "step": 1877 + }, + { + "epoch": 0.79, + "learning_rate": 1.975660903250437e-05, + "loss": 0.4518, + "step": 1878 + }, + { + "epoch": 0.79, + "learning_rate": 1.9756237114970235e-05, + "loss": 0.4759, + "step": 1879 + }, + { + "epoch": 0.79, + "learning_rate": 1.9755864917001724e-05, + "loss": 0.6428, + "step": 1880 + }, + { + "epoch": 0.79, + "learning_rate": 1.9755492438609532e-05, + "loss": 0.5162, + "step": 1881 + }, + { + "epoch": 0.79, + "learning_rate": 1.975511967980437e-05, + "loss": 0.4909, + "step": 1882 + }, + { + "epoch": 0.79, + "learning_rate": 1.9754746640596945e-05, + "loss": 0.4749, + "step": 1883 + }, + { + "epoch": 0.79, + "learning_rate": 1.9754373320997987e-05, + "loss": 0.4545, + "step": 1884 + }, + { + "epoch": 0.79, + "learning_rate": 1.975399972101822e-05, + "loss": 0.5069, + "step": 1885 + }, + { + "epoch": 0.79, + "learning_rate": 1.975362584066839e-05, + "loss": 0.4973, + "step": 1886 + }, + { + "epoch": 0.79, + "learning_rate": 1.975325167995924e-05, + "loss": 0.4698, + "step": 1887 + }, + { + "epoch": 0.79, + "learning_rate": 1.9752877238901525e-05, + "loss": 0.4487, + "step": 1888 + }, + { + "epoch": 0.79, + "learning_rate": 1.975250251750601e-05, + "loss": 0.5072, + "step": 1889 + }, + { + "epoch": 0.79, + "learning_rate": 1.9752127515783458e-05, + "loss": 0.5311, + "step": 1890 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751752233744662e-05, + "loss": 0.4764, + "step": 1891 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751376671400398e-05, + "loss": 0.4509, + "step": 1892 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751000828761465e-05, + "loss": 0.5794, + "step": 1893 + }, + { + "epoch": 0.79, + "learning_rate": 1.9750624705838666e-05, + "loss": 0.4593, + "step": 1894 + }, + { + "epoch": 0.79, + "learning_rate": 1.9750248302642813e-05, + "loss": 0.472, + "step": 1895 + }, + { + "epoch": 0.79, + "learning_rate": 1.9749871619184726e-05, + "loss": 0.4706, + "step": 1896 + }, + { + "epoch": 0.79, + "learning_rate": 1.974949465547523e-05, + "loss": 0.5228, + "step": 1897 + }, + { + "epoch": 0.79, + "learning_rate": 1.974911741152516e-05, + "loss": 0.5407, + "step": 1898 + }, + { + "epoch": 0.8, + "learning_rate": 1.9748739887345362e-05, + "loss": 0.6125, + "step": 1899 + }, + { + "epoch": 0.8, + "learning_rate": 1.974836208294669e-05, + "loss": 0.4526, + "step": 1900 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747983998339996e-05, + "loss": 0.4489, + "step": 1901 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747605633536152e-05, + "loss": 0.4619, + "step": 1902 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747226988546038e-05, + "loss": 0.5118, + "step": 1903 + }, + { + "epoch": 0.8, + "learning_rate": 1.974684806338053e-05, + "loss": 0.5502, + "step": 1904 + }, + { + "epoch": 0.8, + "learning_rate": 1.9746468858050526e-05, + "loss": 0.5278, + "step": 1905 + }, + { + "epoch": 0.8, + "learning_rate": 1.9746089372566924e-05, + "loss": 0.4562, + "step": 1906 + }, + { + "epoch": 0.8, + "learning_rate": 1.974570960694063e-05, + "loss": 0.5074, + "step": 1907 + }, + { + "epoch": 0.8, + "learning_rate": 1.974532956118256e-05, + "loss": 0.4635, + "step": 1908 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744949235303644e-05, + "loss": 0.5443, + "step": 1909 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744568629314807e-05, + "loss": 0.4998, + "step": 1910 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744187743226994e-05, + "loss": 0.4901, + "step": 1911 + }, + { + "epoch": 0.8, + "learning_rate": 1.974380657705115e-05, + "loss": 0.5269, + "step": 1912 + }, + { + "epoch": 0.8, + "learning_rate": 1.974342513079823e-05, + "loss": 0.4695, + "step": 1913 + }, + { + "epoch": 0.8, + "learning_rate": 1.9743043404479203e-05, + "loss": 0.4963, + "step": 1914 + }, + { + "epoch": 0.8, + "learning_rate": 1.9742661398105038e-05, + "loss": 0.5258, + "step": 1915 + }, + { + "epoch": 0.8, + "learning_rate": 1.9742279111686717e-05, + "loss": 0.4798, + "step": 1916 + }, + { + "epoch": 0.8, + "learning_rate": 1.974189654523523e-05, + "loss": 0.5146, + "step": 1917 + }, + { + "epoch": 0.8, + "learning_rate": 1.974151369876157e-05, + "loss": 0.4735, + "step": 1918 + }, + { + "epoch": 0.8, + "learning_rate": 1.9741130572276742e-05, + "loss": 0.5074, + "step": 1919 + }, + { + "epoch": 0.8, + "learning_rate": 1.9740747165791762e-05, + "loss": 0.5401, + "step": 1920 + }, + { + "epoch": 0.8, + "learning_rate": 1.974036347931765e-05, + "loss": 0.5378, + "step": 1921 + }, + { + "epoch": 0.8, + "learning_rate": 1.973997951286543e-05, + "loss": 0.5635, + "step": 1922 + }, + { + "epoch": 0.81, + "learning_rate": 1.9739595266446143e-05, + "loss": 0.6037, + "step": 1923 + }, + { + "epoch": 0.81, + "learning_rate": 1.9739210740070833e-05, + "loss": 0.4684, + "step": 1924 + }, + { + "epoch": 0.81, + "learning_rate": 1.973882593375055e-05, + "loss": 0.48, + "step": 1925 + }, + { + "epoch": 0.81, + "learning_rate": 1.973844084749636e-05, + "loss": 0.5285, + "step": 1926 + }, + { + "epoch": 0.81, + "learning_rate": 1.973805548131933e-05, + "loss": 0.4726, + "step": 1927 + }, + { + "epoch": 0.81, + "learning_rate": 1.9737669835230537e-05, + "loss": 0.5762, + "step": 1928 + }, + { + "epoch": 0.81, + "learning_rate": 1.9737283909241065e-05, + "loss": 0.5675, + "step": 1929 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736897703362006e-05, + "loss": 0.5074, + "step": 1930 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736511217604465e-05, + "loss": 0.4967, + "step": 1931 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736124451979547e-05, + "loss": 0.5085, + "step": 1932 + }, + { + "epoch": 0.81, + "learning_rate": 1.9735737406498375e-05, + "loss": 0.5721, + "step": 1933 + }, + { + "epoch": 0.81, + "learning_rate": 1.973535008117207e-05, + "loss": 0.5391, + "step": 1934 + }, + { + "epoch": 0.81, + "learning_rate": 1.9734962476011763e-05, + "loss": 0.4891, + "step": 1935 + }, + { + "epoch": 0.81, + "learning_rate": 1.97345745910286e-05, + "loss": 0.4809, + "step": 1936 + }, + { + "epoch": 0.81, + "learning_rate": 1.9734186426233728e-05, + "loss": 0.4708, + "step": 1937 + }, + { + "epoch": 0.81, + "learning_rate": 1.9733797981638306e-05, + "loss": 0.4815, + "step": 1938 + }, + { + "epoch": 0.81, + "learning_rate": 1.97334092572535e-05, + "loss": 0.5476, + "step": 1939 + }, + { + "epoch": 0.81, + "learning_rate": 1.9733020253090477e-05, + "loss": 0.5123, + "step": 1940 + }, + { + "epoch": 0.81, + "learning_rate": 1.9732630969160428e-05, + "loss": 0.5376, + "step": 1941 + }, + { + "epoch": 0.81, + "learning_rate": 1.9732241405474536e-05, + "loss": 0.5071, + "step": 1942 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731851562044002e-05, + "loss": 0.4348, + "step": 1943 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731461438880034e-05, + "loss": 0.4937, + "step": 1944 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731071035993837e-05, + "loss": 0.5239, + "step": 1945 + }, + { + "epoch": 0.81, + "learning_rate": 1.9730680353396643e-05, + "loss": 0.497, + "step": 1946 + }, + { + "epoch": 0.82, + "learning_rate": 1.9730289391099675e-05, + "loss": 0.563, + "step": 1947 + }, + { + "epoch": 0.82, + "learning_rate": 1.9729898149114174e-05, + "loss": 0.6107, + "step": 1948 + }, + { + "epoch": 0.82, + "learning_rate": 1.9729506627451383e-05, + "loss": 0.5381, + "step": 1949 + }, + { + "epoch": 0.82, + "learning_rate": 1.972911482612256e-05, + "loss": 0.5316, + "step": 1950 + }, + { + "epoch": 0.82, + "learning_rate": 1.972872274513896e-05, + "loss": 0.5669, + "step": 1951 + }, + { + "epoch": 0.82, + "learning_rate": 1.972833038451186e-05, + "loss": 0.5279, + "step": 1952 + }, + { + "epoch": 0.82, + "learning_rate": 1.972793774425254e-05, + "loss": 0.485, + "step": 1953 + }, + { + "epoch": 0.82, + "learning_rate": 1.972754482437228e-05, + "loss": 0.4867, + "step": 1954 + }, + { + "epoch": 0.82, + "learning_rate": 1.972715162488237e-05, + "loss": 0.4395, + "step": 1955 + }, + { + "epoch": 0.82, + "learning_rate": 1.9726758145794127e-05, + "loss": 0.5581, + "step": 1956 + }, + { + "epoch": 0.82, + "learning_rate": 1.9726364387118847e-05, + "loss": 0.4973, + "step": 1957 + }, + { + "epoch": 0.82, + "learning_rate": 1.9725970348867858e-05, + "loss": 0.5836, + "step": 1958 + }, + { + "epoch": 0.82, + "learning_rate": 1.972557603105248e-05, + "loss": 0.461, + "step": 1959 + }, + { + "epoch": 0.82, + "learning_rate": 1.9725181433684047e-05, + "loss": 0.5794, + "step": 1960 + }, + { + "epoch": 0.82, + "learning_rate": 1.9724786556773906e-05, + "loss": 0.4638, + "step": 1961 + }, + { + "epoch": 0.82, + "learning_rate": 1.9724391400333405e-05, + "loss": 0.4794, + "step": 1962 + }, + { + "epoch": 0.82, + "learning_rate": 1.97239959643739e-05, + "loss": 0.4787, + "step": 1963 + }, + { + "epoch": 0.82, + "learning_rate": 1.9723600248906766e-05, + "loss": 0.4904, + "step": 1964 + }, + { + "epoch": 0.82, + "learning_rate": 1.9723204253943366e-05, + "loss": 0.482, + "step": 1965 + }, + { + "epoch": 0.82, + "learning_rate": 1.972280797949509e-05, + "loss": 0.5006, + "step": 1966 + }, + { + "epoch": 0.82, + "learning_rate": 1.9722411425573323e-05, + "loss": 0.506, + "step": 1967 + }, + { + "epoch": 0.82, + "learning_rate": 1.9722014592189472e-05, + "loss": 0.5763, + "step": 1968 + }, + { + "epoch": 0.82, + "learning_rate": 1.9721617479354936e-05, + "loss": 0.5273, + "step": 1969 + }, + { + "epoch": 0.82, + "learning_rate": 1.972122008708113e-05, + "loss": 0.4863, + "step": 1970 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720822415379485e-05, + "loss": 0.4972, + "step": 1971 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720424464261426e-05, + "loss": 0.5359, + "step": 1972 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720026233738388e-05, + "loss": 0.4415, + "step": 1973 + }, + { + "epoch": 0.83, + "learning_rate": 1.9719627723821818e-05, + "loss": 0.5304, + "step": 1974 + }, + { + "epoch": 0.83, + "learning_rate": 1.971922893452318e-05, + "loss": 0.489, + "step": 1975 + }, + { + "epoch": 0.83, + "learning_rate": 1.971882986585393e-05, + "loss": 0.49, + "step": 1976 + }, + { + "epoch": 0.83, + "learning_rate": 1.9718430517825538e-05, + "loss": 0.5563, + "step": 1977 + }, + { + "epoch": 0.83, + "learning_rate": 1.9718030890449483e-05, + "loss": 0.5101, + "step": 1978 + }, + { + "epoch": 0.83, + "learning_rate": 1.9717630983737256e-05, + "loss": 0.5219, + "step": 1979 + }, + { + "epoch": 0.83, + "learning_rate": 1.971723079770035e-05, + "loss": 0.4326, + "step": 1980 + }, + { + "epoch": 0.83, + "learning_rate": 1.971683033235027e-05, + "loss": 0.4955, + "step": 1981 + }, + { + "epoch": 0.83, + "learning_rate": 1.971642958769852e-05, + "loss": 0.5168, + "step": 1982 + }, + { + "epoch": 0.83, + "learning_rate": 1.9716028563756628e-05, + "loss": 0.608, + "step": 1983 + }, + { + "epoch": 0.83, + "learning_rate": 1.971562726053611e-05, + "loss": 0.4624, + "step": 1984 + }, + { + "epoch": 0.83, + "learning_rate": 1.9715225678048515e-05, + "loss": 0.4922, + "step": 1985 + }, + { + "epoch": 0.83, + "learning_rate": 1.9714823816305376e-05, + "loss": 0.6225, + "step": 1986 + }, + { + "epoch": 0.83, + "learning_rate": 1.9714421675318247e-05, + "loss": 0.4787, + "step": 1987 + }, + { + "epoch": 0.83, + "learning_rate": 1.971401925509869e-05, + "loss": 0.4666, + "step": 1988 + }, + { + "epoch": 0.83, + "learning_rate": 1.9713616555658266e-05, + "loss": 0.5071, + "step": 1989 + }, + { + "epoch": 0.83, + "learning_rate": 1.9713213577008556e-05, + "loss": 0.4458, + "step": 1990 + }, + { + "epoch": 0.83, + "learning_rate": 1.971281031916114e-05, + "loss": 0.4941, + "step": 1991 + }, + { + "epoch": 0.83, + "learning_rate": 1.9712406782127612e-05, + "loss": 0.4653, + "step": 1992 + }, + { + "epoch": 0.83, + "learning_rate": 1.9712002965919567e-05, + "loss": 0.5734, + "step": 1993 + }, + { + "epoch": 0.83, + "learning_rate": 1.971159887054862e-05, + "loss": 0.535, + "step": 1994 + }, + { + "epoch": 0.84, + "learning_rate": 1.9711194496026372e-05, + "loss": 0.5193, + "step": 1995 + }, + { + "epoch": 0.84, + "learning_rate": 1.9710789842364465e-05, + "loss": 0.5112, + "step": 1996 + }, + { + "epoch": 0.84, + "learning_rate": 1.971038490957452e-05, + "loss": 0.4812, + "step": 1997 + }, + { + "epoch": 0.84, + "learning_rate": 1.9709979697668172e-05, + "loss": 0.4981, + "step": 1998 + }, + { + "epoch": 0.84, + "learning_rate": 1.970957420665708e-05, + "loss": 0.4752, + "step": 1999 + }, + { + "epoch": 0.84, + "learning_rate": 1.9709168436552887e-05, + "loss": 0.4864, + "step": 2000 + }, + { + "epoch": 0.84, + "learning_rate": 1.9708762387367272e-05, + "loss": 0.4548, + "step": 2001 + }, + { + "epoch": 0.84, + "learning_rate": 1.970835605911189e-05, + "loss": 0.511, + "step": 2002 + }, + { + "epoch": 0.84, + "learning_rate": 1.970794945179843e-05, + "loss": 0.5372, + "step": 2003 + }, + { + "epoch": 0.84, + "learning_rate": 1.970754256543858e-05, + "loss": 0.4788, + "step": 2004 + }, + { + "epoch": 0.84, + "learning_rate": 1.970713540004403e-05, + "loss": 0.5105, + "step": 2005 + }, + { + "epoch": 0.84, + "learning_rate": 1.970672795562649e-05, + "loss": 0.4853, + "step": 2006 + }, + { + "epoch": 0.84, + "learning_rate": 1.970632023219767e-05, + "loss": 0.5353, + "step": 2007 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705912229769285e-05, + "loss": 0.5399, + "step": 2008 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705503948353064e-05, + "loss": 0.4605, + "step": 2009 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705095387960746e-05, + "loss": 0.4799, + "step": 2010 + }, + { + "epoch": 0.84, + "learning_rate": 1.9704686548604076e-05, + "loss": 0.4942, + "step": 2011 + }, + { + "epoch": 0.84, + "learning_rate": 1.9704277430294798e-05, + "loss": 0.5017, + "step": 2012 + }, + { + "epoch": 0.84, + "learning_rate": 1.970386803304468e-05, + "loss": 0.5192, + "step": 2013 + }, + { + "epoch": 0.84, + "learning_rate": 1.9703458356865484e-05, + "loss": 0.4614, + "step": 2014 + }, + { + "epoch": 0.84, + "learning_rate": 1.9703048401768993e-05, + "loss": 0.4899, + "step": 2015 + }, + { + "epoch": 0.84, + "learning_rate": 1.970263816776698e-05, + "loss": 0.4431, + "step": 2016 + }, + { + "epoch": 0.84, + "learning_rate": 1.970222765487125e-05, + "loss": 0.5404, + "step": 2017 + }, + { + "epoch": 0.84, + "learning_rate": 1.970181686309359e-05, + "loss": 0.4927, + "step": 2018 + }, + { + "epoch": 0.85, + "learning_rate": 1.9701405792445815e-05, + "loss": 0.4271, + "step": 2019 + }, + { + "epoch": 0.85, + "learning_rate": 1.970099444293974e-05, + "loss": 0.4929, + "step": 2020 + }, + { + "epoch": 0.85, + "learning_rate": 1.970058281458719e-05, + "loss": 0.4559, + "step": 2021 + }, + { + "epoch": 0.85, + "learning_rate": 1.9700170907399992e-05, + "loss": 0.5275, + "step": 2022 + }, + { + "epoch": 0.85, + "learning_rate": 1.969975872138999e-05, + "loss": 0.5423, + "step": 2023 + }, + { + "epoch": 0.85, + "learning_rate": 1.9699346256569033e-05, + "loss": 0.4918, + "step": 2024 + }, + { + "epoch": 0.85, + "learning_rate": 1.9698933512948973e-05, + "loss": 0.4484, + "step": 2025 + }, + { + "epoch": 0.85, + "learning_rate": 1.9698520490541675e-05, + "loss": 0.4909, + "step": 2026 + }, + { + "epoch": 0.85, + "learning_rate": 1.969810718935901e-05, + "loss": 0.4184, + "step": 2027 + }, + { + "epoch": 0.85, + "learning_rate": 1.9697693609412864e-05, + "loss": 0.516, + "step": 2028 + }, + { + "epoch": 0.85, + "learning_rate": 1.9697279750715118e-05, + "loss": 0.6038, + "step": 2029 + }, + { + "epoch": 0.85, + "learning_rate": 1.9696865613277673e-05, + "loss": 0.5819, + "step": 2030 + }, + { + "epoch": 0.85, + "learning_rate": 1.969645119711243e-05, + "loss": 0.512, + "step": 2031 + }, + { + "epoch": 0.85, + "learning_rate": 1.96960365022313e-05, + "loss": 0.4182, + "step": 2032 + }, + { + "epoch": 0.85, + "learning_rate": 1.969562152864621e-05, + "loss": 0.5777, + "step": 2033 + }, + { + "epoch": 0.85, + "learning_rate": 1.9695206276369077e-05, + "loss": 0.481, + "step": 2034 + }, + { + "epoch": 0.85, + "learning_rate": 1.9694790745411846e-05, + "loss": 0.4749, + "step": 2035 + }, + { + "epoch": 0.85, + "learning_rate": 1.9694374935786457e-05, + "loss": 0.4691, + "step": 2036 + }, + { + "epoch": 0.85, + "learning_rate": 1.9693958847504863e-05, + "loss": 0.5651, + "step": 2037 + }, + { + "epoch": 0.85, + "learning_rate": 1.9693542480579025e-05, + "loss": 0.4447, + "step": 2038 + }, + { + "epoch": 0.85, + "learning_rate": 1.969312583502091e-05, + "loss": 0.4686, + "step": 2039 + }, + { + "epoch": 0.85, + "learning_rate": 1.9692708910842494e-05, + "loss": 0.5077, + "step": 2040 + }, + { + "epoch": 0.85, + "learning_rate": 1.9692291708055762e-05, + "loss": 0.5205, + "step": 2041 + }, + { + "epoch": 0.85, + "learning_rate": 1.9691874226672704e-05, + "loss": 0.4826, + "step": 2042 + }, + { + "epoch": 0.86, + "learning_rate": 1.9691456466705324e-05, + "loss": 0.5561, + "step": 2043 + }, + { + "epoch": 0.86, + "learning_rate": 1.969103842816563e-05, + "loss": 0.4831, + "step": 2044 + }, + { + "epoch": 0.86, + "learning_rate": 1.969062011106563e-05, + "loss": 0.546, + "step": 2045 + }, + { + "epoch": 0.86, + "learning_rate": 1.969020151541736e-05, + "loss": 0.5363, + "step": 2046 + }, + { + "epoch": 0.86, + "learning_rate": 1.968978264123284e-05, + "loss": 0.4719, + "step": 2047 + }, + { + "epoch": 0.86, + "learning_rate": 1.968936348852412e-05, + "loss": 0.5133, + "step": 2048 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688944057303247e-05, + "loss": 0.465, + "step": 2049 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688524347582272e-05, + "loss": 0.5911, + "step": 2050 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688104359373264e-05, + "loss": 0.486, + "step": 2051 + }, + { + "epoch": 0.86, + "learning_rate": 1.9687684092688292e-05, + "loss": 0.4789, + "step": 2052 + }, + { + "epoch": 0.86, + "learning_rate": 1.968726354753944e-05, + "loss": 0.4695, + "step": 2053 + }, + { + "epoch": 0.86, + "learning_rate": 1.968684272393879e-05, + "loss": 0.6171, + "step": 2054 + }, + { + "epoch": 0.86, + "learning_rate": 1.9686421621898443e-05, + "loss": 0.4325, + "step": 2055 + }, + { + "epoch": 0.86, + "learning_rate": 1.9686000241430503e-05, + "loss": 0.513, + "step": 2056 + }, + { + "epoch": 0.86, + "learning_rate": 1.968557858254708e-05, + "loss": 0.4462, + "step": 2057 + }, + { + "epoch": 0.86, + "learning_rate": 1.9685156645260296e-05, + "loss": 0.5465, + "step": 2058 + }, + { + "epoch": 0.86, + "learning_rate": 1.968473442958228e-05, + "loss": 0.4689, + "step": 2059 + }, + { + "epoch": 0.86, + "learning_rate": 1.9684311935525163e-05, + "loss": 0.5581, + "step": 2060 + }, + { + "epoch": 0.86, + "learning_rate": 1.9683889163101096e-05, + "loss": 0.4729, + "step": 2061 + }, + { + "epoch": 0.86, + "learning_rate": 1.968346611232223e-05, + "loss": 0.5558, + "step": 2062 + }, + { + "epoch": 0.86, + "learning_rate": 1.968304278320072e-05, + "loss": 0.5112, + "step": 2063 + }, + { + "epoch": 0.86, + "learning_rate": 1.968261917574874e-05, + "loss": 0.5189, + "step": 2064 + }, + { + "epoch": 0.86, + "learning_rate": 1.9682195289978466e-05, + "loss": 0.5116, + "step": 2065 + }, + { + "epoch": 0.86, + "learning_rate": 1.9681771125902075e-05, + "loss": 0.524, + "step": 2066 + }, + { + "epoch": 0.87, + "learning_rate": 1.9681346683531766e-05, + "loss": 0.5492, + "step": 2067 + }, + { + "epoch": 0.87, + "learning_rate": 1.968092196287974e-05, + "loss": 0.513, + "step": 2068 + }, + { + "epoch": 0.87, + "learning_rate": 1.96804969639582e-05, + "loss": 0.561, + "step": 2069 + }, + { + "epoch": 0.87, + "learning_rate": 1.9680071686779368e-05, + "loss": 0.5189, + "step": 2070 + }, + { + "epoch": 0.87, + "learning_rate": 1.9679646131355463e-05, + "loss": 0.5128, + "step": 2071 + }, + { + "epoch": 0.87, + "learning_rate": 1.9679220297698722e-05, + "loss": 0.45, + "step": 2072 + }, + { + "epoch": 0.87, + "learning_rate": 1.967879418582138e-05, + "loss": 0.5464, + "step": 2073 + }, + { + "epoch": 0.87, + "learning_rate": 1.967836779573569e-05, + "loss": 0.4759, + "step": 2074 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677941127453907e-05, + "loss": 0.5496, + "step": 2075 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677514180988292e-05, + "loss": 0.5275, + "step": 2076 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677086956351118e-05, + "loss": 0.5698, + "step": 2077 + }, + { + "epoch": 0.87, + "learning_rate": 1.967665945355467e-05, + "loss": 0.5322, + "step": 2078 + }, + { + "epoch": 0.87, + "learning_rate": 1.9676231672611233e-05, + "loss": 0.4737, + "step": 2079 + }, + { + "epoch": 0.87, + "learning_rate": 1.96758036135331e-05, + "loss": 0.51, + "step": 2080 + }, + { + "epoch": 0.87, + "learning_rate": 1.9675375276332583e-05, + "loss": 0.4943, + "step": 2081 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674946661021984e-05, + "loss": 0.4394, + "step": 2082 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674517767613633e-05, + "loss": 0.5582, + "step": 2083 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674088596119853e-05, + "loss": 0.4703, + "step": 2084 + }, + { + "epoch": 0.87, + "learning_rate": 1.967365914655298e-05, + "loss": 0.5181, + "step": 2085 + }, + { + "epoch": 0.87, + "learning_rate": 1.9673229418925358e-05, + "loss": 0.4687, + "step": 2086 + }, + { + "epoch": 0.87, + "learning_rate": 1.9672799413249343e-05, + "loss": 0.5428, + "step": 2087 + }, + { + "epoch": 0.87, + "learning_rate": 1.967236912953729e-05, + "loss": 0.5192, + "step": 2088 + }, + { + "epoch": 0.87, + "learning_rate": 1.9671938567801568e-05, + "loss": 0.5889, + "step": 2089 + }, + { + "epoch": 0.88, + "learning_rate": 1.9671507728054558e-05, + "loss": 0.4703, + "step": 2090 + }, + { + "epoch": 0.88, + "learning_rate": 1.9671076610308638e-05, + "loss": 0.5094, + "step": 2091 + }, + { + "epoch": 0.88, + "learning_rate": 1.9670645214576204e-05, + "loss": 0.5278, + "step": 2092 + }, + { + "epoch": 0.88, + "learning_rate": 1.9670213540869655e-05, + "loss": 0.6058, + "step": 2093 + }, + { + "epoch": 0.88, + "learning_rate": 1.96697815892014e-05, + "loss": 0.5246, + "step": 2094 + }, + { + "epoch": 0.88, + "learning_rate": 1.9669349359583852e-05, + "loss": 0.5534, + "step": 2095 + }, + { + "epoch": 0.88, + "learning_rate": 1.966891685202944e-05, + "loss": 0.4617, + "step": 2096 + }, + { + "epoch": 0.88, + "learning_rate": 1.9668484066550592e-05, + "loss": 0.519, + "step": 2097 + }, + { + "epoch": 0.88, + "learning_rate": 1.9668051003159746e-05, + "loss": 0.548, + "step": 2098 + }, + { + "epoch": 0.88, + "learning_rate": 1.9667617661869357e-05, + "loss": 0.4658, + "step": 2099 + }, + { + "epoch": 0.88, + "learning_rate": 1.9667184042691877e-05, + "loss": 0.4929, + "step": 2100 + }, + { + "epoch": 0.88, + "learning_rate": 1.966675014563977e-05, + "loss": 0.4961, + "step": 2101 + }, + { + "epoch": 0.88, + "learning_rate": 1.9666315970725507e-05, + "loss": 0.493, + "step": 2102 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665881517961568e-05, + "loss": 0.4827, + "step": 2103 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665446787360444e-05, + "loss": 0.4241, + "step": 2104 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665011778934628e-05, + "loss": 0.4034, + "step": 2105 + }, + { + "epoch": 0.88, + "learning_rate": 1.9664576492696628e-05, + "loss": 0.463, + "step": 2106 + }, + { + "epoch": 0.88, + "learning_rate": 1.9664140928658947e-05, + "loss": 0.5554, + "step": 2107 + }, + { + "epoch": 0.88, + "learning_rate": 1.9663705086834114e-05, + "loss": 0.481, + "step": 2108 + }, + { + "epoch": 0.88, + "learning_rate": 1.9663268967234655e-05, + "loss": 0.4523, + "step": 2109 + }, + { + "epoch": 0.88, + "learning_rate": 1.9662832569873106e-05, + "loss": 0.5055, + "step": 2110 + }, + { + "epoch": 0.88, + "learning_rate": 1.9662395894762007e-05, + "loss": 0.506, + "step": 2111 + }, + { + "epoch": 0.88, + "learning_rate": 1.9661958941913913e-05, + "loss": 0.4366, + "step": 2112 + }, + { + "epoch": 0.88, + "learning_rate": 1.966152171134138e-05, + "loss": 0.4667, + "step": 2113 + }, + { + "epoch": 0.89, + "learning_rate": 1.9661084203056983e-05, + "loss": 0.4995, + "step": 2114 + }, + { + "epoch": 0.89, + "learning_rate": 1.966064641707329e-05, + "loss": 0.5362, + "step": 2115 + }, + { + "epoch": 0.89, + "learning_rate": 1.9660208353402888e-05, + "loss": 0.5667, + "step": 2116 + }, + { + "epoch": 0.89, + "learning_rate": 1.9659770012058374e-05, + "loss": 0.5167, + "step": 2117 + }, + { + "epoch": 0.89, + "learning_rate": 1.9659331393052342e-05, + "loss": 0.5972, + "step": 2118 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658892496397398e-05, + "loss": 0.4921, + "step": 2119 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658453322106165e-05, + "loss": 0.5383, + "step": 2120 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658013870191258e-05, + "loss": 0.55, + "step": 2121 + }, + { + "epoch": 0.89, + "learning_rate": 1.965757414066531e-05, + "loss": 0.4841, + "step": 2122 + }, + { + "epoch": 0.89, + "learning_rate": 1.9657134133540966e-05, + "loss": 0.4447, + "step": 2123 + }, + { + "epoch": 0.89, + "learning_rate": 1.965669384883087e-05, + "loss": 0.5766, + "step": 2124 + }, + { + "epoch": 0.89, + "learning_rate": 1.9656253286547683e-05, + "loss": 0.53, + "step": 2125 + }, + { + "epoch": 0.89, + "learning_rate": 1.965581244670406e-05, + "loss": 0.5233, + "step": 2126 + }, + { + "epoch": 0.89, + "learning_rate": 1.9655371329312677e-05, + "loss": 0.4755, + "step": 2127 + }, + { + "epoch": 0.89, + "learning_rate": 1.9654929934386215e-05, + "loss": 0.4842, + "step": 2128 + }, + { + "epoch": 0.89, + "learning_rate": 1.965448826193736e-05, + "loss": 0.4566, + "step": 2129 + }, + { + "epoch": 0.89, + "learning_rate": 1.9654046311978803e-05, + "loss": 0.4718, + "step": 2130 + }, + { + "epoch": 0.89, + "learning_rate": 1.9653604084523255e-05, + "loss": 0.4792, + "step": 2131 + }, + { + "epoch": 0.89, + "learning_rate": 1.9653161579583423e-05, + "loss": 0.5052, + "step": 2132 + }, + { + "epoch": 0.89, + "learning_rate": 1.965271879717203e-05, + "loss": 0.5531, + "step": 2133 + }, + { + "epoch": 0.89, + "learning_rate": 1.9652275737301797e-05, + "loss": 0.5135, + "step": 2134 + }, + { + "epoch": 0.89, + "learning_rate": 1.9651832399985466e-05, + "loss": 0.5318, + "step": 2135 + }, + { + "epoch": 0.89, + "learning_rate": 1.9651388785235775e-05, + "loss": 0.548, + "step": 2136 + }, + { + "epoch": 0.89, + "learning_rate": 1.965094489306548e-05, + "loss": 0.4697, + "step": 2137 + }, + { + "epoch": 0.9, + "learning_rate": 1.9650500723487335e-05, + "loss": 0.5317, + "step": 2138 + }, + { + "epoch": 0.9, + "learning_rate": 1.9650056276514115e-05, + "loss": 0.4823, + "step": 2139 + }, + { + "epoch": 0.9, + "learning_rate": 1.964961155215859e-05, + "loss": 0.5023, + "step": 2140 + }, + { + "epoch": 0.9, + "learning_rate": 1.964916655043354e-05, + "loss": 0.4548, + "step": 2141 + }, + { + "epoch": 0.9, + "learning_rate": 1.9648721271351763e-05, + "loss": 0.48, + "step": 2142 + }, + { + "epoch": 0.9, + "learning_rate": 1.9648275714926055e-05, + "loss": 0.5174, + "step": 2143 + }, + { + "epoch": 0.9, + "learning_rate": 1.9647829881169223e-05, + "loss": 0.5626, + "step": 2144 + }, + { + "epoch": 0.9, + "learning_rate": 1.9647383770094083e-05, + "loss": 0.4561, + "step": 2145 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646937381713454e-05, + "loss": 0.4395, + "step": 2146 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646490716040174e-05, + "loss": 0.4942, + "step": 2147 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646043773087078e-05, + "loss": 0.4702, + "step": 2148 + }, + { + "epoch": 0.9, + "learning_rate": 1.9645596552867013e-05, + "loss": 0.4292, + "step": 2149 + }, + { + "epoch": 0.9, + "learning_rate": 1.9645149055392834e-05, + "loss": 0.5833, + "step": 2150 + }, + { + "epoch": 0.9, + "learning_rate": 1.9644701280677406e-05, + "loss": 0.4801, + "step": 2151 + }, + { + "epoch": 0.9, + "learning_rate": 1.9644253228733596e-05, + "loss": 0.4909, + "step": 2152 + }, + { + "epoch": 0.9, + "learning_rate": 1.9643804899574287e-05, + "loss": 0.5129, + "step": 2153 + }, + { + "epoch": 0.9, + "learning_rate": 1.9643356293212363e-05, + "loss": 0.5068, + "step": 2154 + }, + { + "epoch": 0.9, + "learning_rate": 1.964290740966072e-05, + "loss": 0.4858, + "step": 2155 + }, + { + "epoch": 0.9, + "learning_rate": 1.9642458248932262e-05, + "loss": 0.5096, + "step": 2156 + }, + { + "epoch": 0.9, + "learning_rate": 1.9642008811039893e-05, + "loss": 0.5049, + "step": 2157 + }, + { + "epoch": 0.9, + "learning_rate": 1.9641559095996544e-05, + "loss": 0.4885, + "step": 2158 + }, + { + "epoch": 0.9, + "learning_rate": 1.964110910381513e-05, + "loss": 0.4504, + "step": 2159 + }, + { + "epoch": 0.9, + "learning_rate": 1.964065883450859e-05, + "loss": 0.5344, + "step": 2160 + }, + { + "epoch": 0.9, + "learning_rate": 1.9640208288089872e-05, + "loss": 0.4163, + "step": 2161 + }, + { + "epoch": 0.91, + "learning_rate": 1.9639757464571914e-05, + "loss": 0.4949, + "step": 2162 + }, + { + "epoch": 0.91, + "learning_rate": 1.9639306363967686e-05, + "loss": 0.4949, + "step": 2163 + }, + { + "epoch": 0.91, + "learning_rate": 1.963885498629015e-05, + "loss": 0.5139, + "step": 2164 + }, + { + "epoch": 0.91, + "learning_rate": 1.9638403331552277e-05, + "loss": 0.4826, + "step": 2165 + }, + { + "epoch": 0.91, + "learning_rate": 1.963795139976706e-05, + "loss": 0.5396, + "step": 2166 + }, + { + "epoch": 0.91, + "learning_rate": 1.9637499190947478e-05, + "loss": 0.507, + "step": 2167 + }, + { + "epoch": 0.91, + "learning_rate": 1.9637046705106534e-05, + "loss": 0.5509, + "step": 2168 + }, + { + "epoch": 0.91, + "learning_rate": 1.9636593942257237e-05, + "loss": 0.4484, + "step": 2169 + }, + { + "epoch": 0.91, + "learning_rate": 1.9636140902412598e-05, + "loss": 0.4527, + "step": 2170 + }, + { + "epoch": 0.91, + "learning_rate": 1.9635687585585637e-05, + "loss": 0.524, + "step": 2171 + }, + { + "epoch": 0.91, + "learning_rate": 1.963523399178939e-05, + "loss": 0.4281, + "step": 2172 + }, + { + "epoch": 0.91, + "learning_rate": 1.9634780121036887e-05, + "loss": 0.3896, + "step": 2173 + }, + { + "epoch": 0.91, + "learning_rate": 1.9634325973341183e-05, + "loss": 0.5746, + "step": 2174 + }, + { + "epoch": 0.91, + "learning_rate": 1.963387154871533e-05, + "loss": 0.5507, + "step": 2175 + }, + { + "epoch": 0.91, + "learning_rate": 1.9633416847172384e-05, + "loss": 0.4694, + "step": 2176 + }, + { + "epoch": 0.91, + "learning_rate": 1.9632961868725424e-05, + "loss": 0.4982, + "step": 2177 + }, + { + "epoch": 0.91, + "learning_rate": 1.963250661338752e-05, + "loss": 0.4752, + "step": 2178 + }, + { + "epoch": 0.91, + "learning_rate": 1.963205108117176e-05, + "loss": 0.6507, + "step": 2179 + }, + { + "epoch": 0.91, + "learning_rate": 1.963159527209124e-05, + "loss": 0.5155, + "step": 2180 + }, + { + "epoch": 0.91, + "learning_rate": 1.9631139186159064e-05, + "loss": 0.6029, + "step": 2181 + }, + { + "epoch": 0.91, + "learning_rate": 1.9630682823388337e-05, + "loss": 0.5069, + "step": 2182 + }, + { + "epoch": 0.91, + "learning_rate": 1.9630226183792175e-05, + "loss": 0.5005, + "step": 2183 + }, + { + "epoch": 0.91, + "learning_rate": 1.962976926738371e-05, + "loss": 0.5772, + "step": 2184 + }, + { + "epoch": 0.91, + "learning_rate": 1.9629312074176073e-05, + "loss": 0.4924, + "step": 2185 + }, + { + "epoch": 0.92, + "learning_rate": 1.9628854604182403e-05, + "loss": 0.5, + "step": 2186 + }, + { + "epoch": 0.92, + "learning_rate": 1.9628396857415856e-05, + "loss": 0.4859, + "step": 2187 + }, + { + "epoch": 0.92, + "learning_rate": 1.962793883388958e-05, + "loss": 0.4494, + "step": 2188 + }, + { + "epoch": 0.92, + "learning_rate": 1.962748053361675e-05, + "loss": 0.4656, + "step": 2189 + }, + { + "epoch": 0.92, + "learning_rate": 1.9627021956610534e-05, + "loss": 0.475, + "step": 2190 + }, + { + "epoch": 0.92, + "learning_rate": 1.9626563102884117e-05, + "loss": 0.4323, + "step": 2191 + }, + { + "epoch": 0.92, + "learning_rate": 1.9626103972450683e-05, + "loss": 0.578, + "step": 2192 + }, + { + "epoch": 0.92, + "learning_rate": 1.962564456532343e-05, + "loss": 0.4718, + "step": 2193 + }, + { + "epoch": 0.92, + "learning_rate": 1.962518488151557e-05, + "loss": 0.4104, + "step": 2194 + }, + { + "epoch": 0.92, + "learning_rate": 1.962472492104031e-05, + "loss": 0.548, + "step": 2195 + }, + { + "epoch": 0.92, + "learning_rate": 1.9624264683910877e-05, + "loss": 0.555, + "step": 2196 + }, + { + "epoch": 0.92, + "learning_rate": 1.9623804170140494e-05, + "loss": 0.4559, + "step": 2197 + }, + { + "epoch": 0.92, + "learning_rate": 1.9623343379742397e-05, + "loss": 0.5425, + "step": 2198 + }, + { + "epoch": 0.92, + "learning_rate": 1.9622882312729835e-05, + "loss": 0.5496, + "step": 2199 + }, + { + "epoch": 0.92, + "learning_rate": 1.9622420969116062e-05, + "loss": 0.5709, + "step": 2200 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621959348914335e-05, + "loss": 0.5725, + "step": 2201 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621497452137927e-05, + "loss": 0.4704, + "step": 2202 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621035278800115e-05, + "loss": 0.5526, + "step": 2203 + }, + { + "epoch": 0.92, + "learning_rate": 1.9620572828914177e-05, + "loss": 0.5194, + "step": 2204 + }, + { + "epoch": 0.92, + "learning_rate": 1.9620110102493412e-05, + "loss": 0.551, + "step": 2205 + }, + { + "epoch": 0.92, + "learning_rate": 1.9619647099551118e-05, + "loss": 0.53, + "step": 2206 + }, + { + "epoch": 0.92, + "learning_rate": 1.9619183820100603e-05, + "loss": 0.4799, + "step": 2207 + }, + { + "epoch": 0.92, + "learning_rate": 1.9618720264155187e-05, + "loss": 0.4713, + "step": 2208 + }, + { + "epoch": 0.92, + "learning_rate": 1.961825643172819e-05, + "loss": 0.4345, + "step": 2209 + }, + { + "epoch": 0.93, + "learning_rate": 1.9617792322832952e-05, + "loss": 0.5059, + "step": 2210 + }, + { + "epoch": 0.93, + "learning_rate": 1.9617327937482805e-05, + "loss": 0.5279, + "step": 2211 + }, + { + "epoch": 0.93, + "learning_rate": 1.9616863275691102e-05, + "loss": 0.5578, + "step": 2212 + }, + { + "epoch": 0.93, + "learning_rate": 1.9616398337471196e-05, + "loss": 0.5448, + "step": 2213 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615933122836454e-05, + "loss": 0.5377, + "step": 2214 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615467631800248e-05, + "loss": 0.4938, + "step": 2215 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615001864375957e-05, + "loss": 0.4626, + "step": 2216 + }, + { + "epoch": 0.93, + "learning_rate": 1.9614535820576973e-05, + "loss": 0.5208, + "step": 2217 + }, + { + "epoch": 0.93, + "learning_rate": 1.9614069500416685e-05, + "loss": 0.494, + "step": 2218 + }, + { + "epoch": 0.93, + "learning_rate": 1.9613602903908503e-05, + "loss": 0.5347, + "step": 2219 + }, + { + "epoch": 0.93, + "learning_rate": 1.961313603106583e-05, + "loss": 0.6241, + "step": 2220 + }, + { + "epoch": 0.93, + "learning_rate": 1.96126688819021e-05, + "loss": 0.5039, + "step": 2221 + }, + { + "epoch": 0.93, + "learning_rate": 1.9612201456430732e-05, + "loss": 0.4839, + "step": 2222 + }, + { + "epoch": 0.93, + "learning_rate": 1.9611733754665157e-05, + "loss": 0.5787, + "step": 2223 + }, + { + "epoch": 0.93, + "learning_rate": 1.961126577661883e-05, + "loss": 0.5715, + "step": 2224 + }, + { + "epoch": 0.93, + "learning_rate": 1.9610797522305195e-05, + "loss": 0.4819, + "step": 2225 + }, + { + "epoch": 0.93, + "learning_rate": 1.9610328991737716e-05, + "loss": 0.4741, + "step": 2226 + }, + { + "epoch": 0.93, + "learning_rate": 1.9609860184929858e-05, + "loss": 0.4819, + "step": 2227 + }, + { + "epoch": 0.93, + "learning_rate": 1.9609391101895093e-05, + "loss": 0.5646, + "step": 2228 + }, + { + "epoch": 0.93, + "learning_rate": 1.9608921742646913e-05, + "loss": 0.4428, + "step": 2229 + }, + { + "epoch": 0.93, + "learning_rate": 1.9608452107198798e-05, + "loss": 0.4098, + "step": 2230 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607982195564258e-05, + "loss": 0.4311, + "step": 2231 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607512007756793e-05, + "loss": 0.503, + "step": 2232 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607041543789927e-05, + "loss": 0.5069, + "step": 2233 + }, + { + "epoch": 0.94, + "learning_rate": 1.960657080367717e-05, + "loss": 0.4942, + "step": 2234 + }, + { + "epoch": 0.94, + "learning_rate": 1.9606099787432066e-05, + "loss": 0.5272, + "step": 2235 + }, + { + "epoch": 0.94, + "learning_rate": 1.9605628495068143e-05, + "loss": 0.5224, + "step": 2236 + }, + { + "epoch": 0.94, + "learning_rate": 1.9605156926598953e-05, + "loss": 0.6102, + "step": 2237 + }, + { + "epoch": 0.94, + "learning_rate": 1.9604685082038053e-05, + "loss": 0.5718, + "step": 2238 + }, + { + "epoch": 0.94, + "learning_rate": 1.9604212961399004e-05, + "loss": 0.4199, + "step": 2239 + }, + { + "epoch": 0.94, + "learning_rate": 1.960374056469537e-05, + "loss": 0.5303, + "step": 2240 + }, + { + "epoch": 0.94, + "learning_rate": 1.9603267891940742e-05, + "loss": 0.4223, + "step": 2241 + }, + { + "epoch": 0.94, + "learning_rate": 1.9602794943148698e-05, + "loss": 0.6121, + "step": 2242 + }, + { + "epoch": 0.94, + "learning_rate": 1.9602321718332838e-05, + "loss": 0.5694, + "step": 2243 + }, + { + "epoch": 0.94, + "learning_rate": 1.9601848217506755e-05, + "loss": 0.5002, + "step": 2244 + }, + { + "epoch": 0.94, + "learning_rate": 1.960137444068407e-05, + "loss": 0.5088, + "step": 2245 + }, + { + "epoch": 0.94, + "learning_rate": 1.9600900387878397e-05, + "loss": 0.555, + "step": 2246 + }, + { + "epoch": 0.94, + "learning_rate": 1.9600426059103358e-05, + "loss": 0.5601, + "step": 2247 + }, + { + "epoch": 0.94, + "learning_rate": 1.9599951454372594e-05, + "loss": 0.4603, + "step": 2248 + }, + { + "epoch": 0.94, + "learning_rate": 1.9599476573699747e-05, + "loss": 0.5543, + "step": 2249 + }, + { + "epoch": 0.94, + "learning_rate": 1.959900141709846e-05, + "loss": 0.5286, + "step": 2250 + }, + { + "epoch": 0.94, + "learning_rate": 1.9598525984582398e-05, + "loss": 0.494, + "step": 2251 + }, + { + "epoch": 0.94, + "learning_rate": 1.959805027616522e-05, + "loss": 0.4904, + "step": 2252 + }, + { + "epoch": 0.94, + "learning_rate": 1.9597574291860607e-05, + "loss": 0.5076, + "step": 2253 + }, + { + "epoch": 0.94, + "learning_rate": 1.9597098031682238e-05, + "loss": 0.5447, + "step": 2254 + }, + { + "epoch": 0.94, + "learning_rate": 1.9596621495643805e-05, + "loss": 0.4445, + "step": 2255 + }, + { + "epoch": 0.94, + "learning_rate": 1.9596144683759e-05, + "loss": 0.5893, + "step": 2256 + }, + { + "epoch": 0.94, + "learning_rate": 1.959566759604153e-05, + "loss": 0.479, + "step": 2257 + }, + { + "epoch": 0.95, + "learning_rate": 1.959519023250511e-05, + "loss": 0.5299, + "step": 2258 + }, + { + "epoch": 0.95, + "learning_rate": 1.959471259316346e-05, + "loss": 0.4844, + "step": 2259 + }, + { + "epoch": 0.95, + "learning_rate": 1.9594234678030317e-05, + "loss": 0.4575, + "step": 2260 + }, + { + "epoch": 0.95, + "learning_rate": 1.959375648711941e-05, + "loss": 0.538, + "step": 2261 + }, + { + "epoch": 0.95, + "learning_rate": 1.9593278020444483e-05, + "loss": 0.5077, + "step": 2262 + }, + { + "epoch": 0.95, + "learning_rate": 1.9592799278019294e-05, + "loss": 0.5273, + "step": 2263 + }, + { + "epoch": 0.95, + "learning_rate": 1.9592320259857603e-05, + "loss": 0.4944, + "step": 2264 + }, + { + "epoch": 0.95, + "learning_rate": 1.9591840965973175e-05, + "loss": 0.525, + "step": 2265 + }, + { + "epoch": 0.95, + "learning_rate": 1.9591361396379795e-05, + "loss": 0.5158, + "step": 2266 + }, + { + "epoch": 0.95, + "learning_rate": 1.959088155109124e-05, + "loss": 0.5111, + "step": 2267 + }, + { + "epoch": 0.95, + "learning_rate": 1.9590401430121305e-05, + "loss": 0.4892, + "step": 2268 + }, + { + "epoch": 0.95, + "learning_rate": 1.9589921033483793e-05, + "loss": 0.5192, + "step": 2269 + }, + { + "epoch": 0.95, + "learning_rate": 1.958944036119251e-05, + "loss": 0.4518, + "step": 2270 + }, + { + "epoch": 0.95, + "learning_rate": 1.9588959413261272e-05, + "loss": 0.4856, + "step": 2271 + }, + { + "epoch": 0.95, + "learning_rate": 1.9588478189703908e-05, + "loss": 0.4939, + "step": 2272 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587996690534244e-05, + "loss": 0.5192, + "step": 2273 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587514915766124e-05, + "loss": 0.5328, + "step": 2274 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587032865413394e-05, + "loss": 0.4875, + "step": 2275 + }, + { + "epoch": 0.95, + "learning_rate": 1.9586550539489914e-05, + "loss": 0.4893, + "step": 2276 + }, + { + "epoch": 0.95, + "learning_rate": 1.9586067938009546e-05, + "loss": 0.4676, + "step": 2277 + }, + { + "epoch": 0.95, + "learning_rate": 1.9585585060986162e-05, + "loss": 0.4876, + "step": 2278 + }, + { + "epoch": 0.95, + "learning_rate": 1.958510190843364e-05, + "loss": 0.4743, + "step": 2279 + }, + { + "epoch": 0.95, + "learning_rate": 1.958461848036587e-05, + "loss": 0.5614, + "step": 2280 + }, + { + "epoch": 0.95, + "learning_rate": 1.958413477679675e-05, + "loss": 0.4286, + "step": 2281 + }, + { + "epoch": 0.96, + "learning_rate": 1.9583650797740173e-05, + "loss": 0.4811, + "step": 2282 + }, + { + "epoch": 0.96, + "learning_rate": 1.9583166543210065e-05, + "loss": 0.5306, + "step": 2283 + }, + { + "epoch": 0.96, + "learning_rate": 1.9582682013220337e-05, + "loss": 0.4897, + "step": 2284 + }, + { + "epoch": 0.96, + "learning_rate": 1.958219720778492e-05, + "loss": 0.4841, + "step": 2285 + }, + { + "epoch": 0.96, + "learning_rate": 1.9581712126917742e-05, + "loss": 0.5035, + "step": 2286 + }, + { + "epoch": 0.96, + "learning_rate": 1.9581226770632753e-05, + "loss": 0.4354, + "step": 2287 + }, + { + "epoch": 0.96, + "learning_rate": 1.9580741138943906e-05, + "loss": 0.461, + "step": 2288 + }, + { + "epoch": 0.96, + "learning_rate": 1.9580255231865156e-05, + "loss": 0.4559, + "step": 2289 + }, + { + "epoch": 0.96, + "learning_rate": 1.9579769049410468e-05, + "loss": 0.5464, + "step": 2290 + }, + { + "epoch": 0.96, + "learning_rate": 1.957928259159382e-05, + "loss": 0.4297, + "step": 2291 + }, + { + "epoch": 0.96, + "learning_rate": 1.95787958584292e-05, + "loss": 0.5206, + "step": 2292 + }, + { + "epoch": 0.96, + "learning_rate": 1.9578308849930586e-05, + "loss": 0.5657, + "step": 2293 + }, + { + "epoch": 0.96, + "learning_rate": 1.957782156611199e-05, + "loss": 0.4817, + "step": 2294 + }, + { + "epoch": 0.96, + "learning_rate": 1.9577334006987405e-05, + "loss": 0.5052, + "step": 2295 + }, + { + "epoch": 0.96, + "learning_rate": 1.9576846172570855e-05, + "loss": 0.5689, + "step": 2296 + }, + { + "epoch": 0.96, + "learning_rate": 1.9576358062876363e-05, + "loss": 0.5839, + "step": 2297 + }, + { + "epoch": 0.96, + "learning_rate": 1.9575869677917956e-05, + "loss": 0.4988, + "step": 2298 + }, + { + "epoch": 0.96, + "learning_rate": 1.957538101770967e-05, + "loss": 0.4993, + "step": 2299 + }, + { + "epoch": 0.96, + "learning_rate": 1.9574892082265557e-05, + "loss": 0.4793, + "step": 2300 + }, + { + "epoch": 0.96, + "learning_rate": 1.9574402871599666e-05, + "loss": 0.5144, + "step": 2301 + }, + { + "epoch": 0.96, + "learning_rate": 1.957391338572606e-05, + "loss": 0.5921, + "step": 2302 + }, + { + "epoch": 0.96, + "learning_rate": 1.957342362465881e-05, + "loss": 0.5145, + "step": 2303 + }, + { + "epoch": 0.96, + "learning_rate": 1.9572933588411997e-05, + "loss": 0.4551, + "step": 2304 + }, + { + "epoch": 0.97, + "learning_rate": 1.9572443276999696e-05, + "loss": 0.5018, + "step": 2305 + }, + { + "epoch": 0.97, + "learning_rate": 1.957195269043601e-05, + "loss": 0.5091, + "step": 2306 + }, + { + "epoch": 0.97, + "learning_rate": 1.957146182873504e-05, + "loss": 0.5403, + "step": 2307 + }, + { + "epoch": 0.97, + "learning_rate": 1.957097069191089e-05, + "loss": 0.5032, + "step": 2308 + }, + { + "epoch": 0.97, + "learning_rate": 1.9570479279977682e-05, + "loss": 0.5441, + "step": 2309 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569987592949543e-05, + "loss": 0.5361, + "step": 2310 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569495630840597e-05, + "loss": 0.4916, + "step": 2311 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569003393664996e-05, + "loss": 0.4256, + "step": 2312 + }, + { + "epoch": 0.97, + "learning_rate": 1.9568510881436883e-05, + "loss": 0.484, + "step": 2313 + }, + { + "epoch": 0.97, + "learning_rate": 1.9568018094170415e-05, + "loss": 0.4544, + "step": 2314 + }, + { + "epoch": 0.97, + "learning_rate": 1.9567525031879756e-05, + "loss": 0.4739, + "step": 2315 + }, + { + "epoch": 0.97, + "learning_rate": 1.956703169457908e-05, + "loss": 0.4568, + "step": 2316 + }, + { + "epoch": 0.97, + "learning_rate": 1.956653808228257e-05, + "loss": 0.5526, + "step": 2317 + }, + { + "epoch": 0.97, + "learning_rate": 1.956604419500441e-05, + "loss": 0.5471, + "step": 2318 + }, + { + "epoch": 0.97, + "learning_rate": 1.95655500327588e-05, + "loss": 0.4968, + "step": 2319 + }, + { + "epoch": 0.97, + "learning_rate": 1.956505559555994e-05, + "loss": 0.5308, + "step": 2320 + }, + { + "epoch": 0.97, + "learning_rate": 1.9564560883422045e-05, + "loss": 0.6437, + "step": 2321 + }, + { + "epoch": 0.97, + "learning_rate": 1.9564065896359337e-05, + "loss": 0.4779, + "step": 2322 + }, + { + "epoch": 0.97, + "learning_rate": 1.9563570634386042e-05, + "loss": 0.4739, + "step": 2323 + }, + { + "epoch": 0.97, + "learning_rate": 1.9563075097516397e-05, + "loss": 0.4708, + "step": 2324 + }, + { + "epoch": 0.97, + "learning_rate": 1.9562579285764638e-05, + "loss": 0.4772, + "step": 2325 + }, + { + "epoch": 0.97, + "learning_rate": 1.956208319914503e-05, + "loss": 0.5105, + "step": 2326 + }, + { + "epoch": 0.97, + "learning_rate": 1.956158683767182e-05, + "loss": 0.4218, + "step": 2327 + }, + { + "epoch": 0.97, + "learning_rate": 1.9561090201359285e-05, + "loss": 0.5393, + "step": 2328 + }, + { + "epoch": 0.98, + "learning_rate": 1.9560593290221695e-05, + "loss": 0.439, + "step": 2329 + }, + { + "epoch": 0.98, + "learning_rate": 1.9560096104273335e-05, + "loss": 0.4925, + "step": 2330 + }, + { + "epoch": 0.98, + "learning_rate": 1.95595986435285e-05, + "loss": 0.5582, + "step": 2331 + }, + { + "epoch": 0.98, + "learning_rate": 1.9559100908001478e-05, + "loss": 0.4094, + "step": 2332 + }, + { + "epoch": 0.98, + "learning_rate": 1.9558602897706586e-05, + "loss": 0.5477, + "step": 2333 + }, + { + "epoch": 0.98, + "learning_rate": 1.9558104612658137e-05, + "loss": 0.5561, + "step": 2334 + }, + { + "epoch": 0.98, + "learning_rate": 1.9557606052870452e-05, + "loss": 0.4986, + "step": 2335 + }, + { + "epoch": 0.98, + "learning_rate": 1.9557107218357862e-05, + "loss": 0.4391, + "step": 2336 + }, + { + "epoch": 0.98, + "learning_rate": 1.9556608109134707e-05, + "loss": 0.432, + "step": 2337 + }, + { + "epoch": 0.98, + "learning_rate": 1.955610872521533e-05, + "loss": 0.4751, + "step": 2338 + }, + { + "epoch": 0.98, + "learning_rate": 1.955560906661409e-05, + "loss": 0.5742, + "step": 2339 + }, + { + "epoch": 0.98, + "learning_rate": 1.9555109133345347e-05, + "loss": 0.5143, + "step": 2340 + }, + { + "epoch": 0.98, + "learning_rate": 1.9554608925423467e-05, + "loss": 0.4496, + "step": 2341 + }, + { + "epoch": 0.98, + "learning_rate": 1.9554108442862836e-05, + "loss": 0.4794, + "step": 2342 + }, + { + "epoch": 0.98, + "learning_rate": 1.9553607685677834e-05, + "loss": 0.4721, + "step": 2343 + }, + { + "epoch": 0.98, + "learning_rate": 1.955310665388286e-05, + "loss": 0.5335, + "step": 2344 + }, + { + "epoch": 0.98, + "learning_rate": 1.955260534749231e-05, + "loss": 0.4888, + "step": 2345 + }, + { + "epoch": 0.98, + "learning_rate": 1.9552103766520596e-05, + "loss": 0.452, + "step": 2346 + }, + { + "epoch": 0.98, + "learning_rate": 1.9551601910982136e-05, + "loss": 0.5632, + "step": 2347 + }, + { + "epoch": 0.98, + "learning_rate": 1.9551099780891353e-05, + "loss": 0.4978, + "step": 2348 + }, + { + "epoch": 0.98, + "learning_rate": 1.9550597376262685e-05, + "loss": 0.5115, + "step": 2349 + }, + { + "epoch": 0.98, + "learning_rate": 1.955009469711057e-05, + "loss": 0.5622, + "step": 2350 + }, + { + "epoch": 0.98, + "learning_rate": 1.954959174344946e-05, + "loss": 0.5278, + "step": 2351 + }, + { + "epoch": 0.98, + "learning_rate": 1.9549088515293804e-05, + "loss": 0.4838, + "step": 2352 + }, + { + "epoch": 0.99, + "learning_rate": 1.9548585012658074e-05, + "loss": 0.5064, + "step": 2353 + }, + { + "epoch": 0.99, + "learning_rate": 1.954808123555674e-05, + "loss": 0.5339, + "step": 2354 + }, + { + "epoch": 0.99, + "learning_rate": 1.954757718400429e-05, + "loss": 0.4907, + "step": 2355 + }, + { + "epoch": 0.99, + "learning_rate": 1.95470728580152e-05, + "loss": 0.485, + "step": 2356 + }, + { + "epoch": 0.99, + "learning_rate": 1.9546568257603976e-05, + "loss": 0.4973, + "step": 2357 + }, + { + "epoch": 0.99, + "learning_rate": 1.9546063382785117e-05, + "loss": 0.5277, + "step": 2358 + }, + { + "epoch": 0.99, + "learning_rate": 1.9545558233573136e-05, + "loss": 0.4577, + "step": 2359 + }, + { + "epoch": 0.99, + "learning_rate": 1.9545052809982556e-05, + "loss": 0.4347, + "step": 2360 + }, + { + "epoch": 0.99, + "learning_rate": 1.9544547112027904e-05, + "loss": 0.4502, + "step": 2361 + }, + { + "epoch": 0.99, + "learning_rate": 1.9544041139723712e-05, + "loss": 0.446, + "step": 2362 + }, + { + "epoch": 0.99, + "learning_rate": 1.954353489308453e-05, + "loss": 0.618, + "step": 2363 + }, + { + "epoch": 0.99, + "learning_rate": 1.9543028372124903e-05, + "loss": 0.4332, + "step": 2364 + }, + { + "epoch": 0.99, + "learning_rate": 1.9542521576859397e-05, + "loss": 0.4302, + "step": 2365 + }, + { + "epoch": 0.99, + "learning_rate": 1.9542014507302576e-05, + "loss": 0.5073, + "step": 2366 + }, + { + "epoch": 0.99, + "learning_rate": 1.9541507163469012e-05, + "loss": 0.4632, + "step": 2367 + }, + { + "epoch": 0.99, + "learning_rate": 1.9540999545373295e-05, + "loss": 0.5273, + "step": 2368 + }, + { + "epoch": 0.99, + "learning_rate": 1.954049165303001e-05, + "loss": 0.4725, + "step": 2369 + }, + { + "epoch": 0.99, + "learning_rate": 1.9539983486453762e-05, + "loss": 0.4652, + "step": 2370 + }, + { + "epoch": 0.99, + "learning_rate": 1.953947504565915e-05, + "loss": 0.4843, + "step": 2371 + }, + { + "epoch": 0.99, + "learning_rate": 1.9538966330660796e-05, + "loss": 0.511, + "step": 2372 + }, + { + "epoch": 0.99, + "learning_rate": 1.9538457341473318e-05, + "loss": 0.5827, + "step": 2373 + }, + { + "epoch": 0.99, + "learning_rate": 1.9537948078111348e-05, + "loss": 0.4844, + "step": 2374 + }, + { + "epoch": 0.99, + "learning_rate": 1.9537438540589526e-05, + "loss": 0.5425, + "step": 2375 + }, + { + "epoch": 0.99, + "learning_rate": 1.9536928728922496e-05, + "loss": 0.4983, + "step": 2376 + }, + { + "epoch": 1.0, + "learning_rate": 1.953641864312491e-05, + "loss": 0.5085, + "step": 2377 + }, + { + "epoch": 1.0, + "learning_rate": 1.9535908283211435e-05, + "loss": 0.5282, + "step": 2378 + }, + { + "epoch": 1.0, + "learning_rate": 1.9535397649196738e-05, + "loss": 0.4554, + "step": 2379 + }, + { + "epoch": 1.0, + "learning_rate": 1.95348867410955e-05, + "loss": 0.4774, + "step": 2380 + }, + { + "epoch": 1.0, + "learning_rate": 1.95343755589224e-05, + "loss": 0.6091, + "step": 2381 + }, + { + "epoch": 1.0, + "learning_rate": 1.9533864102692134e-05, + "loss": 0.6011, + "step": 2382 + }, + { + "epoch": 1.0, + "learning_rate": 1.9533352372419408e-05, + "loss": 0.4555, + "step": 2383 + }, + { + "epoch": 1.0, + "learning_rate": 1.953284036811893e-05, + "loss": 0.435, + "step": 2384 + }, + { + "epoch": 1.0, + "learning_rate": 1.9532328089805407e-05, + "loss": 0.5451, + "step": 2385 + }, + { + "epoch": 1.0, + "learning_rate": 1.953181553749358e-05, + "loss": 0.4929, + "step": 2386 + }, + { + "epoch": 1.0, + "learning_rate": 1.953130271119817e-05, + "loss": 0.5192, + "step": 2387 + }, + { + "epoch": 1.0, + "learning_rate": 1.953078961093392e-05, + "loss": 0.5065, + "step": 2388 + }, + { + "epoch": 1.0, + "learning_rate": 1.953027623671558e-05, + "loss": 0.4521, + "step": 2389 + }, + { + "epoch": 1.0, + "learning_rate": 1.952976258855791e-05, + "loss": 0.3904, + "step": 2390 + }, + { + "epoch": 1.0, + "learning_rate": 1.9529248666475668e-05, + "loss": 0.3474, + "step": 2391 + }, + { + "epoch": 1.0, + "learning_rate": 1.9528734470483632e-05, + "loss": 0.3506, + "step": 2392 + }, + { + "epoch": 1.0, + "learning_rate": 1.9528220000596577e-05, + "loss": 0.399, + "step": 2393 + }, + { + "epoch": 1.0, + "learning_rate": 1.9527705256829293e-05, + "loss": 0.3954, + "step": 2394 + }, + { + "epoch": 1.0, + "learning_rate": 1.952719023919658e-05, + "loss": 0.3772, + "step": 2395 + }, + { + "epoch": 1.0, + "learning_rate": 1.9526674947713234e-05, + "loss": 0.4016, + "step": 2396 + }, + { + "epoch": 1.0, + "learning_rate": 1.9526159382394068e-05, + "loss": 0.3883, + "step": 2397 + }, + { + "epoch": 1.0, + "learning_rate": 1.952564354325391e-05, + "loss": 0.3803, + "step": 2398 + }, + { + "epoch": 1.0, + "learning_rate": 1.952512743030758e-05, + "loss": 0.3478, + "step": 2399 + }, + { + "epoch": 1.0, + "learning_rate": 1.952461104356991e-05, + "loss": 0.4181, + "step": 2400 + }, + { + "epoch": 1.01, + "learning_rate": 1.952409438305575e-05, + "loss": 0.3764, + "step": 2401 + }, + { + "epoch": 1.01, + "learning_rate": 1.952357744877995e-05, + "loss": 0.412, + "step": 2402 + }, + { + "epoch": 1.01, + "learning_rate": 1.952306024075737e-05, + "loss": 0.3514, + "step": 2403 + }, + { + "epoch": 1.01, + "learning_rate": 1.9522542759002866e-05, + "loss": 0.3896, + "step": 2404 + }, + { + "epoch": 1.01, + "learning_rate": 1.9522025003531327e-05, + "loss": 0.3816, + "step": 2405 + }, + { + "epoch": 1.01, + "learning_rate": 1.9521506974357628e-05, + "loss": 0.4014, + "step": 2406 + }, + { + "epoch": 1.01, + "learning_rate": 1.952098867149666e-05, + "loss": 0.3944, + "step": 2407 + }, + { + "epoch": 1.01, + "learning_rate": 1.952047009496332e-05, + "loss": 0.376, + "step": 2408 + }, + { + "epoch": 1.01, + "learning_rate": 1.9519951244772514e-05, + "loss": 0.363, + "step": 2409 + }, + { + "epoch": 1.01, + "learning_rate": 1.951943212093916e-05, + "loss": 0.4244, + "step": 2410 + }, + { + "epoch": 1.01, + "learning_rate": 1.9518912723478176e-05, + "loss": 0.4064, + "step": 2411 + }, + { + "epoch": 1.01, + "learning_rate": 1.9518393052404495e-05, + "loss": 0.3723, + "step": 2412 + }, + { + "epoch": 1.01, + "learning_rate": 1.951787310773305e-05, + "loss": 0.3509, + "step": 2413 + }, + { + "epoch": 1.01, + "learning_rate": 1.9517352889478787e-05, + "loss": 0.3573, + "step": 2414 + }, + { + "epoch": 1.01, + "learning_rate": 1.9516832397656663e-05, + "loss": 0.3915, + "step": 2415 + }, + { + "epoch": 1.01, + "learning_rate": 1.9516311632281636e-05, + "loss": 0.3676, + "step": 2416 + }, + { + "epoch": 1.01, + "learning_rate": 1.9515790593368677e-05, + "loss": 0.4132, + "step": 2417 + }, + { + "epoch": 1.01, + "learning_rate": 1.951526928093276e-05, + "loss": 0.4165, + "step": 2418 + }, + { + "epoch": 1.01, + "learning_rate": 1.9514747694988873e-05, + "loss": 0.3968, + "step": 2419 + }, + { + "epoch": 1.01, + "learning_rate": 1.9514225835552004e-05, + "loss": 0.3525, + "step": 2420 + }, + { + "epoch": 1.01, + "learning_rate": 1.9513703702637158e-05, + "loss": 0.3924, + "step": 2421 + }, + { + "epoch": 1.01, + "learning_rate": 1.951318129625934e-05, + "loss": 0.3927, + "step": 2422 + }, + { + "epoch": 1.01, + "learning_rate": 1.951265861643357e-05, + "loss": 0.4062, + "step": 2423 + }, + { + "epoch": 1.01, + "learning_rate": 1.951213566317487e-05, + "loss": 0.3723, + "step": 2424 + }, + { + "epoch": 1.02, + "learning_rate": 1.9511612436498267e-05, + "loss": 0.3841, + "step": 2425 + }, + { + "epoch": 1.02, + "learning_rate": 1.9511088936418807e-05, + "loss": 0.3915, + "step": 2426 + }, + { + "epoch": 1.02, + "learning_rate": 1.9510565162951538e-05, + "loss": 0.4367, + "step": 2427 + }, + { + "epoch": 1.02, + "learning_rate": 1.951004111611151e-05, + "loss": 0.3407, + "step": 2428 + }, + { + "epoch": 1.02, + "learning_rate": 1.950951679591379e-05, + "loss": 0.3957, + "step": 2429 + }, + { + "epoch": 1.02, + "learning_rate": 1.9508992202373448e-05, + "loss": 0.3546, + "step": 2430 + }, + { + "epoch": 1.02, + "learning_rate": 1.9508467335505564e-05, + "loss": 0.3823, + "step": 2431 + }, + { + "epoch": 1.02, + "learning_rate": 1.9507942195325223e-05, + "loss": 0.3666, + "step": 2432 + }, + { + "epoch": 1.02, + "learning_rate": 1.9507416781847524e-05, + "loss": 0.4182, + "step": 2433 + }, + { + "epoch": 1.02, + "learning_rate": 1.9506891095087563e-05, + "loss": 0.4023, + "step": 2434 + }, + { + "epoch": 1.02, + "learning_rate": 1.950636513506045e-05, + "loss": 0.4253, + "step": 2435 + }, + { + "epoch": 1.02, + "learning_rate": 1.9505838901781314e-05, + "loss": 0.3536, + "step": 2436 + }, + { + "epoch": 1.02, + "learning_rate": 1.9505312395265273e-05, + "loss": 0.3591, + "step": 2437 + }, + { + "epoch": 1.02, + "learning_rate": 1.950478561552746e-05, + "loss": 0.3788, + "step": 2438 + }, + { + "epoch": 1.02, + "learning_rate": 1.9504258562583018e-05, + "loss": 0.3627, + "step": 2439 + }, + { + "epoch": 1.02, + "learning_rate": 1.95037312364471e-05, + "loss": 0.3955, + "step": 2440 + }, + { + "epoch": 1.02, + "learning_rate": 1.950320363713486e-05, + "loss": 0.3654, + "step": 2441 + }, + { + "epoch": 1.02, + "learning_rate": 1.9502675764661464e-05, + "loss": 0.3787, + "step": 2442 + }, + { + "epoch": 1.02, + "learning_rate": 1.9502147619042086e-05, + "loss": 0.3312, + "step": 2443 + }, + { + "epoch": 1.02, + "learning_rate": 1.950161920029191e-05, + "loss": 0.3557, + "step": 2444 + }, + { + "epoch": 1.02, + "learning_rate": 1.9501090508426117e-05, + "loss": 0.3739, + "step": 2445 + }, + { + "epoch": 1.02, + "learning_rate": 1.950056154345991e-05, + "loss": 0.3747, + "step": 2446 + }, + { + "epoch": 1.02, + "learning_rate": 1.9500032305408494e-05, + "loss": 0.3263, + "step": 2447 + }, + { + "epoch": 1.02, + "learning_rate": 1.9499502794287078e-05, + "loss": 0.3685, + "step": 2448 + }, + { + "epoch": 1.03, + "learning_rate": 1.9498973010110884e-05, + "loss": 0.3698, + "step": 2449 + }, + { + "epoch": 1.03, + "learning_rate": 1.949844295289514e-05, + "loss": 0.3479, + "step": 2450 + }, + { + "epoch": 1.03, + "learning_rate": 1.949791262265508e-05, + "loss": 0.3987, + "step": 2451 + }, + { + "epoch": 1.03, + "learning_rate": 1.949738201940595e-05, + "loss": 0.3598, + "step": 2452 + }, + { + "epoch": 1.03, + "learning_rate": 1.9496851143163004e-05, + "loss": 0.3291, + "step": 2453 + }, + { + "epoch": 1.03, + "learning_rate": 1.9496319993941498e-05, + "loss": 0.4026, + "step": 2454 + }, + { + "epoch": 1.03, + "learning_rate": 1.9495788571756704e-05, + "loss": 0.3992, + "step": 2455 + }, + { + "epoch": 1.03, + "learning_rate": 1.949525687662389e-05, + "loss": 0.3673, + "step": 2456 + }, + { + "epoch": 1.03, + "learning_rate": 1.9494724908558345e-05, + "loss": 0.391, + "step": 2457 + }, + { + "epoch": 1.03, + "learning_rate": 1.9494192667575355e-05, + "loss": 0.3577, + "step": 2458 + }, + { + "epoch": 1.03, + "learning_rate": 1.9493660153690225e-05, + "loss": 0.3866, + "step": 2459 + }, + { + "epoch": 1.03, + "learning_rate": 1.949312736691826e-05, + "loss": 0.3643, + "step": 2460 + }, + { + "epoch": 1.03, + "learning_rate": 1.9492594307274772e-05, + "loss": 0.3734, + "step": 2461 + }, + { + "epoch": 1.03, + "learning_rate": 1.949206097477508e-05, + "loss": 0.4079, + "step": 2462 + }, + { + "epoch": 1.03, + "learning_rate": 1.949152736943452e-05, + "loss": 0.3489, + "step": 2463 + }, + { + "epoch": 1.03, + "learning_rate": 1.949099349126843e-05, + "loss": 0.3724, + "step": 2464 + }, + { + "epoch": 1.03, + "learning_rate": 1.9490459340292153e-05, + "loss": 0.3742, + "step": 2465 + }, + { + "epoch": 1.03, + "learning_rate": 1.9489924916521047e-05, + "loss": 0.3711, + "step": 2466 + }, + { + "epoch": 1.03, + "learning_rate": 1.9489390219970468e-05, + "loss": 0.3826, + "step": 2467 + }, + { + "epoch": 1.03, + "learning_rate": 1.9488855250655785e-05, + "loss": 0.3584, + "step": 2468 + }, + { + "epoch": 1.03, + "learning_rate": 1.948832000859238e-05, + "loss": 0.3569, + "step": 2469 + }, + { + "epoch": 1.03, + "learning_rate": 1.9487784493795637e-05, + "loss": 0.3772, + "step": 2470 + }, + { + "epoch": 1.03, + "learning_rate": 1.948724870628095e-05, + "loss": 0.3452, + "step": 2471 + }, + { + "epoch": 1.03, + "learning_rate": 1.9486712646063715e-05, + "loss": 0.4041, + "step": 2472 + }, + { + "epoch": 1.04, + "learning_rate": 1.9486176313159344e-05, + "loss": 0.3882, + "step": 2473 + }, + { + "epoch": 1.04, + "learning_rate": 1.9485639707583255e-05, + "loss": 0.4075, + "step": 2474 + }, + { + "epoch": 1.04, + "learning_rate": 1.948510282935087e-05, + "loss": 0.355, + "step": 2475 + }, + { + "epoch": 1.04, + "learning_rate": 1.9484565678477616e-05, + "loss": 0.3827, + "step": 2476 + }, + { + "epoch": 1.04, + "learning_rate": 1.9484028254978942e-05, + "loss": 0.3761, + "step": 2477 + }, + { + "epoch": 1.04, + "learning_rate": 1.948349055887029e-05, + "loss": 0.4096, + "step": 2478 + }, + { + "epoch": 1.04, + "learning_rate": 1.9482952590167118e-05, + "loss": 0.3582, + "step": 2479 + }, + { + "epoch": 1.04, + "learning_rate": 1.9482414348884892e-05, + "loss": 0.3936, + "step": 2480 + }, + { + "epoch": 1.04, + "learning_rate": 1.9481875835039075e-05, + "loss": 0.3534, + "step": 2481 + }, + { + "epoch": 1.04, + "learning_rate": 1.9481337048645154e-05, + "loss": 0.3723, + "step": 2482 + }, + { + "epoch": 1.04, + "learning_rate": 1.9480797989718612e-05, + "loss": 0.4351, + "step": 2483 + }, + { + "epoch": 1.04, + "learning_rate": 1.9480258658274947e-05, + "loss": 0.3678, + "step": 2484 + }, + { + "epoch": 1.04, + "learning_rate": 1.9479719054329657e-05, + "loss": 0.3977, + "step": 2485 + }, + { + "epoch": 1.04, + "learning_rate": 1.9479179177898255e-05, + "loss": 0.3488, + "step": 2486 + }, + { + "epoch": 1.04, + "learning_rate": 1.9478639028996262e-05, + "loss": 0.3655, + "step": 2487 + }, + { + "epoch": 1.04, + "learning_rate": 1.9478098607639197e-05, + "loss": 0.4027, + "step": 2488 + }, + { + "epoch": 1.04, + "learning_rate": 1.94775579138426e-05, + "loss": 0.3937, + "step": 2489 + }, + { + "epoch": 1.04, + "learning_rate": 1.9477016947622013e-05, + "loss": 0.354, + "step": 2490 + }, + { + "epoch": 1.04, + "learning_rate": 1.947647570899298e-05, + "loss": 0.4085, + "step": 2491 + }, + { + "epoch": 1.04, + "learning_rate": 1.947593419797107e-05, + "loss": 0.3629, + "step": 2492 + }, + { + "epoch": 1.04, + "learning_rate": 1.947539241457183e-05, + "loss": 0.3365, + "step": 2493 + }, + { + "epoch": 1.04, + "learning_rate": 1.947485035881085e-05, + "loss": 0.3338, + "step": 2494 + }, + { + "epoch": 1.04, + "learning_rate": 1.94743080307037e-05, + "loss": 0.4101, + "step": 2495 + }, + { + "epoch": 1.05, + "learning_rate": 1.947376543026598e-05, + "loss": 0.3708, + "step": 2496 + }, + { + "epoch": 1.05, + "learning_rate": 1.9473222557513273e-05, + "loss": 0.3908, + "step": 2497 + }, + { + "epoch": 1.05, + "learning_rate": 1.947267941246119e-05, + "loss": 0.3852, + "step": 2498 + }, + { + "epoch": 1.05, + "learning_rate": 1.9472135995125344e-05, + "loss": 0.3899, + "step": 2499 + }, + { + "epoch": 1.05, + "learning_rate": 1.9471592305521355e-05, + "loss": 0.4517, + "step": 2500 + }, + { + "epoch": 1.05, + "learning_rate": 1.9471048343664853e-05, + "loss": 0.3975, + "step": 2501 + }, + { + "epoch": 1.05, + "learning_rate": 1.9470504109571467e-05, + "loss": 0.4205, + "step": 2502 + }, + { + "epoch": 1.05, + "learning_rate": 1.9469959603256844e-05, + "loss": 0.3591, + "step": 2503 + }, + { + "epoch": 1.05, + "learning_rate": 1.946941482473664e-05, + "loss": 0.3632, + "step": 2504 + }, + { + "epoch": 1.05, + "learning_rate": 1.9468869774026508e-05, + "loss": 0.4138, + "step": 2505 + }, + { + "epoch": 1.05, + "learning_rate": 1.9468324451142115e-05, + "loss": 0.3705, + "step": 2506 + }, + { + "epoch": 1.05, + "learning_rate": 1.9467778856099134e-05, + "loss": 0.3899, + "step": 2507 + }, + { + "epoch": 1.05, + "learning_rate": 1.946723298891326e-05, + "loss": 0.342, + "step": 2508 + }, + { + "epoch": 1.05, + "learning_rate": 1.9466686849600168e-05, + "loss": 0.4147, + "step": 2509 + }, + { + "epoch": 1.05, + "learning_rate": 1.9466140438175567e-05, + "loss": 0.391, + "step": 2510 + }, + { + "epoch": 1.05, + "learning_rate": 1.9465593754655158e-05, + "loss": 0.3563, + "step": 2511 + }, + { + "epoch": 1.05, + "learning_rate": 1.9465046799054657e-05, + "loss": 0.4002, + "step": 2512 + }, + { + "epoch": 1.05, + "learning_rate": 1.946449957138978e-05, + "loss": 0.4012, + "step": 2513 + }, + { + "epoch": 1.05, + "learning_rate": 1.9463952071676268e-05, + "loss": 0.3771, + "step": 2514 + }, + { + "epoch": 1.05, + "learning_rate": 1.9463404299929852e-05, + "loss": 0.3488, + "step": 2515 + }, + { + "epoch": 1.05, + "learning_rate": 1.9462856256166273e-05, + "loss": 0.379, + "step": 2516 + }, + { + "epoch": 1.05, + "learning_rate": 1.9462307940401288e-05, + "loss": 0.4167, + "step": 2517 + }, + { + "epoch": 1.05, + "learning_rate": 1.9461759352650663e-05, + "loss": 0.3738, + "step": 2518 + }, + { + "epoch": 1.05, + "learning_rate": 1.9461210492930156e-05, + "loss": 0.3786, + "step": 2519 + }, + { + "epoch": 1.06, + "learning_rate": 1.9460661361255552e-05, + "loss": 0.422, + "step": 2520 + }, + { + "epoch": 1.06, + "learning_rate": 1.946011195764263e-05, + "loss": 0.3372, + "step": 2521 + }, + { + "epoch": 1.06, + "learning_rate": 1.9459562282107188e-05, + "loss": 0.3704, + "step": 2522 + }, + { + "epoch": 1.06, + "learning_rate": 1.945901233466502e-05, + "loss": 0.3586, + "step": 2523 + }, + { + "epoch": 1.06, + "learning_rate": 1.9458462115331938e-05, + "loss": 0.3876, + "step": 2524 + }, + { + "epoch": 1.06, + "learning_rate": 1.9457911624123754e-05, + "loss": 0.3911, + "step": 2525 + }, + { + "epoch": 1.06, + "learning_rate": 1.9457360861056294e-05, + "loss": 0.3998, + "step": 2526 + }, + { + "epoch": 1.06, + "learning_rate": 1.9456809826145387e-05, + "loss": 0.3516, + "step": 2527 + }, + { + "epoch": 1.06, + "learning_rate": 1.9456258519406875e-05, + "loss": 0.3591, + "step": 2528 + }, + { + "epoch": 1.06, + "learning_rate": 1.9455706940856602e-05, + "loss": 0.4332, + "step": 2529 + }, + { + "epoch": 1.06, + "learning_rate": 1.9455155090510427e-05, + "loss": 0.4036, + "step": 2530 + }, + { + "epoch": 1.06, + "learning_rate": 1.9454602968384204e-05, + "loss": 0.3703, + "step": 2531 + }, + { + "epoch": 1.06, + "learning_rate": 1.9454050574493815e-05, + "loss": 0.3515, + "step": 2532 + }, + { + "epoch": 1.06, + "learning_rate": 1.9453497908855125e-05, + "loss": 0.4121, + "step": 2533 + }, + { + "epoch": 1.06, + "learning_rate": 1.945294497148403e-05, + "loss": 0.3881, + "step": 2534 + }, + { + "epoch": 1.06, + "learning_rate": 1.945239176239642e-05, + "loss": 0.3798, + "step": 2535 + }, + { + "epoch": 1.06, + "learning_rate": 1.94518382816082e-05, + "loss": 0.3696, + "step": 2536 + }, + { + "epoch": 1.06, + "learning_rate": 1.945128452913527e-05, + "loss": 0.3494, + "step": 2537 + }, + { + "epoch": 1.06, + "learning_rate": 1.9450730504993554e-05, + "loss": 0.4292, + "step": 2538 + }, + { + "epoch": 1.06, + "learning_rate": 1.9450176209198977e-05, + "loss": 0.3643, + "step": 2539 + }, + { + "epoch": 1.06, + "learning_rate": 1.9449621641767474e-05, + "loss": 0.4396, + "step": 2540 + }, + { + "epoch": 1.06, + "learning_rate": 1.944906680271498e-05, + "loss": 0.3905, + "step": 2541 + }, + { + "epoch": 1.06, + "learning_rate": 1.9448511692057446e-05, + "loss": 0.377, + "step": 2542 + }, + { + "epoch": 1.06, + "learning_rate": 1.9447956309810825e-05, + "loss": 0.3605, + "step": 2543 + }, + { + "epoch": 1.07, + "learning_rate": 1.9447400655991086e-05, + "loss": 0.4316, + "step": 2544 + }, + { + "epoch": 1.07, + "learning_rate": 1.94468447306142e-05, + "loss": 0.3723, + "step": 2545 + }, + { + "epoch": 1.07, + "learning_rate": 1.9446288533696145e-05, + "loss": 0.4058, + "step": 2546 + }, + { + "epoch": 1.07, + "learning_rate": 1.9445732065252904e-05, + "loss": 0.379, + "step": 2547 + }, + { + "epoch": 1.07, + "learning_rate": 1.9445175325300482e-05, + "loss": 0.4104, + "step": 2548 + }, + { + "epoch": 1.07, + "learning_rate": 1.9444618313854878e-05, + "loss": 0.4294, + "step": 2549 + }, + { + "epoch": 1.07, + "learning_rate": 1.9444061030932094e-05, + "loss": 0.3712, + "step": 2550 + }, + { + "epoch": 1.07, + "learning_rate": 1.9443503476548162e-05, + "loss": 0.3953, + "step": 2551 + }, + { + "epoch": 1.07, + "learning_rate": 1.9442945650719102e-05, + "loss": 0.3694, + "step": 2552 + }, + { + "epoch": 1.07, + "learning_rate": 1.944238755346095e-05, + "loss": 0.4309, + "step": 2553 + }, + { + "epoch": 1.07, + "learning_rate": 1.9441829184789745e-05, + "loss": 0.3836, + "step": 2554 + }, + { + "epoch": 1.07, + "learning_rate": 1.9441270544721536e-05, + "loss": 0.4283, + "step": 2555 + }, + { + "epoch": 1.07, + "learning_rate": 1.9440711633272387e-05, + "loss": 0.3818, + "step": 2556 + }, + { + "epoch": 1.07, + "learning_rate": 1.944015245045836e-05, + "loss": 0.3676, + "step": 2557 + }, + { + "epoch": 1.07, + "learning_rate": 1.9439592996295525e-05, + "loss": 0.3511, + "step": 2558 + }, + { + "epoch": 1.07, + "learning_rate": 1.943903327079997e-05, + "loss": 0.3633, + "step": 2559 + }, + { + "epoch": 1.07, + "learning_rate": 1.9438473273987775e-05, + "loss": 0.3664, + "step": 2560 + }, + { + "epoch": 1.07, + "learning_rate": 1.9437913005875046e-05, + "loss": 0.4262, + "step": 2561 + }, + { + "epoch": 1.07, + "learning_rate": 1.943735246647788e-05, + "loss": 0.4054, + "step": 2562 + }, + { + "epoch": 1.07, + "learning_rate": 1.9436791655812392e-05, + "loss": 0.4121, + "step": 2563 + }, + { + "epoch": 1.07, + "learning_rate": 1.94362305738947e-05, + "loss": 0.3511, + "step": 2564 + }, + { + "epoch": 1.07, + "learning_rate": 1.9435669220740933e-05, + "loss": 0.3689, + "step": 2565 + }, + { + "epoch": 1.07, + "learning_rate": 1.943510759636723e-05, + "loss": 0.3972, + "step": 2566 + }, + { + "epoch": 1.07, + "learning_rate": 1.9434545700789732e-05, + "loss": 0.404, + "step": 2567 + }, + { + "epoch": 1.08, + "learning_rate": 1.943398353402459e-05, + "loss": 0.3609, + "step": 2568 + }, + { + "epoch": 1.08, + "learning_rate": 1.9433421096087963e-05, + "loss": 0.3381, + "step": 2569 + }, + { + "epoch": 1.08, + "learning_rate": 1.9432858386996018e-05, + "loss": 0.3895, + "step": 2570 + }, + { + "epoch": 1.08, + "learning_rate": 1.9432295406764927e-05, + "loss": 0.3767, + "step": 2571 + }, + { + "epoch": 1.08, + "learning_rate": 1.9431732155410878e-05, + "loss": 0.3764, + "step": 2572 + }, + { + "epoch": 1.08, + "learning_rate": 1.9431168632950053e-05, + "loss": 0.3678, + "step": 2573 + }, + { + "epoch": 1.08, + "learning_rate": 1.9430604839398657e-05, + "loss": 0.3845, + "step": 2574 + }, + { + "epoch": 1.08, + "learning_rate": 1.9430040774772893e-05, + "loss": 0.42, + "step": 2575 + }, + { + "epoch": 1.08, + "learning_rate": 1.9429476439088976e-05, + "loss": 0.385, + "step": 2576 + }, + { + "epoch": 1.08, + "learning_rate": 1.9428911832363126e-05, + "loss": 0.4227, + "step": 2577 + }, + { + "epoch": 1.08, + "learning_rate": 1.9428346954611574e-05, + "loss": 0.3842, + "step": 2578 + }, + { + "epoch": 1.08, + "learning_rate": 1.9427781805850553e-05, + "loss": 0.3735, + "step": 2579 + }, + { + "epoch": 1.08, + "learning_rate": 1.9427216386096313e-05, + "loss": 0.3866, + "step": 2580 + }, + { + "epoch": 1.08, + "learning_rate": 1.94266506953651e-05, + "loss": 0.3834, + "step": 2581 + }, + { + "epoch": 1.08, + "learning_rate": 1.942608473367318e-05, + "loss": 0.4074, + "step": 2582 + }, + { + "epoch": 1.08, + "learning_rate": 1.942551850103682e-05, + "loss": 0.4092, + "step": 2583 + }, + { + "epoch": 1.08, + "learning_rate": 1.942495199747229e-05, + "loss": 0.3959, + "step": 2584 + }, + { + "epoch": 1.08, + "learning_rate": 1.9424385222995883e-05, + "loss": 0.3953, + "step": 2585 + }, + { + "epoch": 1.08, + "learning_rate": 1.9423818177623884e-05, + "loss": 0.3939, + "step": 2586 + }, + { + "epoch": 1.08, + "learning_rate": 1.9423250861372592e-05, + "loss": 0.4049, + "step": 2587 + }, + { + "epoch": 1.08, + "learning_rate": 1.9422683274258318e-05, + "loss": 0.3754, + "step": 2588 + }, + { + "epoch": 1.08, + "learning_rate": 1.9422115416297377e-05, + "loss": 0.4024, + "step": 2589 + }, + { + "epoch": 1.08, + "learning_rate": 1.9421547287506085e-05, + "loss": 0.358, + "step": 2590 + }, + { + "epoch": 1.08, + "learning_rate": 1.942097888790078e-05, + "loss": 0.3966, + "step": 2591 + }, + { + "epoch": 1.09, + "learning_rate": 1.9420410217497795e-05, + "loss": 0.3881, + "step": 2592 + }, + { + "epoch": 1.09, + "learning_rate": 1.941984127631348e-05, + "loss": 0.3901, + "step": 2593 + }, + { + "epoch": 1.09, + "learning_rate": 1.9419272064364186e-05, + "loss": 0.4057, + "step": 2594 + }, + { + "epoch": 1.09, + "learning_rate": 1.9418702581666276e-05, + "loss": 0.351, + "step": 2595 + }, + { + "epoch": 1.09, + "learning_rate": 1.9418132828236114e-05, + "loss": 0.3222, + "step": 2596 + }, + { + "epoch": 1.09, + "learning_rate": 1.9417562804090086e-05, + "loss": 0.3725, + "step": 2597 + }, + { + "epoch": 1.09, + "learning_rate": 1.941699250924457e-05, + "loss": 0.4544, + "step": 2598 + }, + { + "epoch": 1.09, + "learning_rate": 1.941642194371596e-05, + "loss": 0.408, + "step": 2599 + }, + { + "epoch": 1.09, + "learning_rate": 1.941585110752066e-05, + "loss": 0.3905, + "step": 2600 + }, + { + "epoch": 1.09, + "learning_rate": 1.9415280000675074e-05, + "loss": 0.39, + "step": 2601 + }, + { + "epoch": 1.09, + "learning_rate": 1.9414708623195618e-05, + "loss": 0.363, + "step": 2602 + }, + { + "epoch": 1.09, + "learning_rate": 1.9414136975098718e-05, + "loss": 0.3951, + "step": 2603 + }, + { + "epoch": 1.09, + "learning_rate": 1.9413565056400804e-05, + "loss": 0.3983, + "step": 2604 + }, + { + "epoch": 1.09, + "learning_rate": 1.9412992867118318e-05, + "loss": 0.3811, + "step": 2605 + }, + { + "epoch": 1.09, + "learning_rate": 1.9412420407267704e-05, + "loss": 0.3948, + "step": 2606 + }, + { + "epoch": 1.09, + "learning_rate": 1.9411847676865415e-05, + "loss": 0.3897, + "step": 2607 + }, + { + "epoch": 1.09, + "learning_rate": 1.9411274675927922e-05, + "loss": 0.3905, + "step": 2608 + }, + { + "epoch": 1.09, + "learning_rate": 1.9410701404471686e-05, + "loss": 0.4154, + "step": 2609 + }, + { + "epoch": 1.09, + "learning_rate": 1.941012786251319e-05, + "loss": 0.3532, + "step": 2610 + }, + { + "epoch": 1.09, + "learning_rate": 1.9409554050068916e-05, + "loss": 0.3639, + "step": 2611 + }, + { + "epoch": 1.09, + "learning_rate": 1.9408979967155364e-05, + "loss": 0.3617, + "step": 2612 + }, + { + "epoch": 1.09, + "learning_rate": 1.940840561378903e-05, + "loss": 0.3866, + "step": 2613 + }, + { + "epoch": 1.09, + "learning_rate": 1.940783098998643e-05, + "loss": 0.3641, + "step": 2614 + }, + { + "epoch": 1.09, + "learning_rate": 1.9407256095764074e-05, + "loss": 0.3998, + "step": 2615 + }, + { + "epoch": 1.1, + "learning_rate": 1.9406680931138488e-05, + "loss": 0.3803, + "step": 2616 + }, + { + "epoch": 1.1, + "learning_rate": 1.940610549612621e-05, + "loss": 0.3922, + "step": 2617 + }, + { + "epoch": 1.1, + "learning_rate": 1.940552979074377e-05, + "loss": 0.372, + "step": 2618 + }, + { + "epoch": 1.1, + "learning_rate": 1.940495381500773e-05, + "loss": 0.3683, + "step": 2619 + }, + { + "epoch": 1.1, + "learning_rate": 1.9404377568934633e-05, + "loss": 0.4032, + "step": 2620 + }, + { + "epoch": 1.1, + "learning_rate": 1.940380105254105e-05, + "loss": 0.3778, + "step": 2621 + }, + { + "epoch": 1.1, + "learning_rate": 1.9403224265843547e-05, + "loss": 0.3555, + "step": 2622 + }, + { + "epoch": 1.1, + "learning_rate": 1.9402647208858712e-05, + "loss": 0.3768, + "step": 2623 + }, + { + "epoch": 1.1, + "learning_rate": 1.9402069881603123e-05, + "loss": 0.3493, + "step": 2624 + }, + { + "epoch": 1.1, + "learning_rate": 1.9401492284093382e-05, + "loss": 0.4207, + "step": 2625 + }, + { + "epoch": 1.1, + "learning_rate": 1.9400914416346083e-05, + "loss": 0.3887, + "step": 2626 + }, + { + "epoch": 1.1, + "learning_rate": 1.9400336278377845e-05, + "loss": 0.3926, + "step": 2627 + }, + { + "epoch": 1.1, + "learning_rate": 1.939975787020528e-05, + "loss": 0.3902, + "step": 2628 + }, + { + "epoch": 1.1, + "learning_rate": 1.9399179191845022e-05, + "loss": 0.3577, + "step": 2629 + }, + { + "epoch": 1.1, + "learning_rate": 1.9398600243313695e-05, + "loss": 0.3764, + "step": 2630 + }, + { + "epoch": 1.1, + "learning_rate": 1.939802102462794e-05, + "loss": 0.3581, + "step": 2631 + }, + { + "epoch": 1.1, + "learning_rate": 1.9397441535804417e-05, + "loss": 0.3631, + "step": 2632 + }, + { + "epoch": 1.1, + "learning_rate": 1.939686177685977e-05, + "loss": 0.3528, + "step": 2633 + }, + { + "epoch": 1.1, + "learning_rate": 1.939628174781068e-05, + "loss": 0.3927, + "step": 2634 + }, + { + "epoch": 1.1, + "learning_rate": 1.9395701448673797e-05, + "loss": 0.3593, + "step": 2635 + }, + { + "epoch": 1.1, + "learning_rate": 1.939512087946582e-05, + "loss": 0.3985, + "step": 2636 + }, + { + "epoch": 1.1, + "learning_rate": 1.9394540040203428e-05, + "loss": 0.3532, + "step": 2637 + }, + { + "epoch": 1.1, + "learning_rate": 1.9393958930903323e-05, + "loss": 0.3842, + "step": 2638 + }, + { + "epoch": 1.1, + "learning_rate": 1.93933775515822e-05, + "loss": 0.3824, + "step": 2639 + }, + { + "epoch": 1.11, + "learning_rate": 1.939279590225678e-05, + "loss": 0.3778, + "step": 2640 + }, + { + "epoch": 1.11, + "learning_rate": 1.9392213982943774e-05, + "loss": 0.3495, + "step": 2641 + }, + { + "epoch": 1.11, + "learning_rate": 1.9391631793659908e-05, + "loss": 0.3768, + "step": 2642 + }, + { + "epoch": 1.11, + "learning_rate": 1.9391049334421925e-05, + "loss": 0.3629, + "step": 2643 + }, + { + "epoch": 1.11, + "learning_rate": 1.939046660524656e-05, + "loss": 0.375, + "step": 2644 + }, + { + "epoch": 1.11, + "learning_rate": 1.938988360615057e-05, + "loss": 0.3699, + "step": 2645 + }, + { + "epoch": 1.11, + "learning_rate": 1.93893003371507e-05, + "loss": 0.3716, + "step": 2646 + }, + { + "epoch": 1.11, + "learning_rate": 1.9388716798263728e-05, + "loss": 0.4273, + "step": 2647 + }, + { + "epoch": 1.11, + "learning_rate": 1.9388132989506422e-05, + "loss": 0.3377, + "step": 2648 + }, + { + "epoch": 1.11, + "learning_rate": 1.9387548910895567e-05, + "loss": 0.3908, + "step": 2649 + }, + { + "epoch": 1.11, + "learning_rate": 1.9386964562447947e-05, + "loss": 0.3764, + "step": 2650 + }, + { + "epoch": 1.11, + "learning_rate": 1.938637994418036e-05, + "loss": 0.4319, + "step": 2651 + }, + { + "epoch": 1.11, + "learning_rate": 1.9385795056109614e-05, + "loss": 0.3947, + "step": 2652 + }, + { + "epoch": 1.11, + "learning_rate": 1.938520989825251e-05, + "loss": 0.3839, + "step": 2653 + }, + { + "epoch": 1.11, + "learning_rate": 1.9384624470625884e-05, + "loss": 0.4071, + "step": 2654 + }, + { + "epoch": 1.11, + "learning_rate": 1.9384038773246555e-05, + "loss": 0.3845, + "step": 2655 + }, + { + "epoch": 1.11, + "learning_rate": 1.9383452806131356e-05, + "loss": 0.4514, + "step": 2656 + }, + { + "epoch": 1.11, + "learning_rate": 1.9382866569297133e-05, + "loss": 0.34, + "step": 2657 + }, + { + "epoch": 1.11, + "learning_rate": 1.9382280062760738e-05, + "loss": 0.3823, + "step": 2658 + }, + { + "epoch": 1.11, + "learning_rate": 1.9381693286539024e-05, + "loss": 0.381, + "step": 2659 + }, + { + "epoch": 1.11, + "learning_rate": 1.9381106240648867e-05, + "loss": 0.3852, + "step": 2660 + }, + { + "epoch": 1.11, + "learning_rate": 1.9380518925107133e-05, + "loss": 0.4091, + "step": 2661 + }, + { + "epoch": 1.11, + "learning_rate": 1.9379931339930705e-05, + "loss": 0.3806, + "step": 2662 + }, + { + "epoch": 1.11, + "learning_rate": 1.9379343485136478e-05, + "loss": 0.3209, + "step": 2663 + }, + { + "epoch": 1.12, + "learning_rate": 1.9378755360741342e-05, + "loss": 0.3436, + "step": 2664 + }, + { + "epoch": 1.12, + "learning_rate": 1.9378166966762207e-05, + "loss": 0.392, + "step": 2665 + }, + { + "epoch": 1.12, + "learning_rate": 1.9377578303215984e-05, + "loss": 0.3796, + "step": 2666 + }, + { + "epoch": 1.12, + "learning_rate": 1.9376989370119595e-05, + "loss": 0.3759, + "step": 2667 + }, + { + "epoch": 1.12, + "learning_rate": 1.937640016748997e-05, + "loss": 0.4186, + "step": 2668 + }, + { + "epoch": 1.12, + "learning_rate": 1.937581069534404e-05, + "loss": 0.3675, + "step": 2669 + }, + { + "epoch": 1.12, + "learning_rate": 1.937522095369875e-05, + "loss": 0.3504, + "step": 2670 + }, + { + "epoch": 1.12, + "learning_rate": 1.9374630942571055e-05, + "loss": 0.3958, + "step": 2671 + }, + { + "epoch": 1.12, + "learning_rate": 1.9374040661977913e-05, + "loss": 0.3544, + "step": 2672 + }, + { + "epoch": 1.12, + "learning_rate": 1.937345011193629e-05, + "loss": 0.376, + "step": 2673 + }, + { + "epoch": 1.12, + "learning_rate": 1.9372859292463163e-05, + "loss": 0.3849, + "step": 2674 + }, + { + "epoch": 1.12, + "learning_rate": 1.9372268203575512e-05, + "loss": 0.3558, + "step": 2675 + }, + { + "epoch": 1.12, + "learning_rate": 1.9371676845290327e-05, + "loss": 0.3714, + "step": 2676 + }, + { + "epoch": 1.12, + "learning_rate": 1.9371085217624608e-05, + "loss": 0.4111, + "step": 2677 + }, + { + "epoch": 1.12, + "learning_rate": 1.937049332059536e-05, + "loss": 0.3657, + "step": 2678 + }, + { + "epoch": 1.12, + "learning_rate": 1.9369901154219595e-05, + "loss": 0.3718, + "step": 2679 + }, + { + "epoch": 1.12, + "learning_rate": 1.9369308718514342e-05, + "loss": 0.3688, + "step": 2680 + }, + { + "epoch": 1.12, + "learning_rate": 1.9368716013496616e-05, + "loss": 0.4108, + "step": 2681 + }, + { + "epoch": 1.12, + "learning_rate": 1.9368123039183468e-05, + "loss": 0.3761, + "step": 2682 + }, + { + "epoch": 1.12, + "learning_rate": 1.9367529795591935e-05, + "loss": 0.3954, + "step": 2683 + }, + { + "epoch": 1.12, + "learning_rate": 1.9366936282739067e-05, + "loss": 0.3762, + "step": 2684 + }, + { + "epoch": 1.12, + "learning_rate": 1.9366342500641933e-05, + "loss": 0.3663, + "step": 2685 + }, + { + "epoch": 1.12, + "learning_rate": 1.9365748449317593e-05, + "loss": 0.3908, + "step": 2686 + }, + { + "epoch": 1.12, + "learning_rate": 1.9365154128783126e-05, + "loss": 0.3524, + "step": 2687 + }, + { + "epoch": 1.13, + "learning_rate": 1.9364559539055612e-05, + "loss": 0.364, + "step": 2688 + }, + { + "epoch": 1.13, + "learning_rate": 1.9363964680152147e-05, + "loss": 0.3783, + "step": 2689 + }, + { + "epoch": 1.13, + "learning_rate": 1.936336955208982e-05, + "loss": 0.3973, + "step": 2690 + }, + { + "epoch": 1.13, + "learning_rate": 1.936277415488575e-05, + "loss": 0.3912, + "step": 2691 + }, + { + "epoch": 1.13, + "learning_rate": 1.9362178488557043e-05, + "loss": 0.4189, + "step": 2692 + }, + { + "epoch": 1.13, + "learning_rate": 1.9361582553120824e-05, + "loss": 0.3805, + "step": 2693 + }, + { + "epoch": 1.13, + "learning_rate": 1.9360986348594226e-05, + "loss": 0.3507, + "step": 2694 + }, + { + "epoch": 1.13, + "learning_rate": 1.9360389874994374e-05, + "loss": 0.4058, + "step": 2695 + }, + { + "epoch": 1.13, + "learning_rate": 1.9359793132338427e-05, + "loss": 0.3807, + "step": 2696 + }, + { + "epoch": 1.13, + "learning_rate": 1.935919612064353e-05, + "loss": 0.4055, + "step": 2697 + }, + { + "epoch": 1.13, + "learning_rate": 1.9358598839926848e-05, + "loss": 0.3797, + "step": 2698 + }, + { + "epoch": 1.13, + "learning_rate": 1.9358001290205542e-05, + "loss": 0.3917, + "step": 2699 + }, + { + "epoch": 1.13, + "learning_rate": 1.9357403471496797e-05, + "loss": 0.3894, + "step": 2700 + }, + { + "epoch": 1.13, + "learning_rate": 1.9356805383817792e-05, + "loss": 0.3666, + "step": 2701 + }, + { + "epoch": 1.13, + "learning_rate": 1.935620702718572e-05, + "loss": 0.4063, + "step": 2702 + }, + { + "epoch": 1.13, + "learning_rate": 1.9355608401617782e-05, + "loss": 0.3768, + "step": 2703 + }, + { + "epoch": 1.13, + "learning_rate": 1.935500950713118e-05, + "loss": 0.3813, + "step": 2704 + }, + { + "epoch": 1.13, + "learning_rate": 1.9354410343743127e-05, + "loss": 0.3715, + "step": 2705 + }, + { + "epoch": 1.13, + "learning_rate": 1.9353810911470855e-05, + "loss": 0.3931, + "step": 2706 + }, + { + "epoch": 1.13, + "learning_rate": 1.935321121033159e-05, + "loss": 0.41, + "step": 2707 + }, + { + "epoch": 1.13, + "learning_rate": 1.9352611240342565e-05, + "loss": 0.4062, + "step": 2708 + }, + { + "epoch": 1.13, + "learning_rate": 1.935201100152103e-05, + "loss": 0.3632, + "step": 2709 + }, + { + "epoch": 1.13, + "learning_rate": 1.9351410493884234e-05, + "loss": 0.3832, + "step": 2710 + }, + { + "epoch": 1.14, + "learning_rate": 1.9350809717449448e-05, + "loss": 0.3753, + "step": 2711 + }, + { + "epoch": 1.14, + "learning_rate": 1.935020867223393e-05, + "loss": 0.346, + "step": 2712 + }, + { + "epoch": 1.14, + "learning_rate": 1.9349607358254965e-05, + "loss": 0.3275, + "step": 2713 + }, + { + "epoch": 1.14, + "learning_rate": 1.934900577552983e-05, + "loss": 0.4153, + "step": 2714 + }, + { + "epoch": 1.14, + "learning_rate": 1.934840392407582e-05, + "loss": 0.3351, + "step": 2715 + }, + { + "epoch": 1.14, + "learning_rate": 1.9347801803910236e-05, + "loss": 0.4013, + "step": 2716 + }, + { + "epoch": 1.14, + "learning_rate": 1.9347199415050383e-05, + "loss": 0.37, + "step": 2717 + }, + { + "epoch": 1.14, + "learning_rate": 1.934659675751358e-05, + "loss": 0.3452, + "step": 2718 + }, + { + "epoch": 1.14, + "learning_rate": 1.9345993831317147e-05, + "loss": 0.307, + "step": 2719 + }, + { + "epoch": 1.14, + "learning_rate": 1.9345390636478414e-05, + "loss": 0.3732, + "step": 2720 + }, + { + "epoch": 1.14, + "learning_rate": 1.934478717301472e-05, + "loss": 0.4037, + "step": 2721 + }, + { + "epoch": 1.14, + "learning_rate": 1.9344183440943408e-05, + "loss": 0.4228, + "step": 2722 + }, + { + "epoch": 1.14, + "learning_rate": 1.934357944028184e-05, + "loss": 0.3822, + "step": 2723 + }, + { + "epoch": 1.14, + "learning_rate": 1.934297517104737e-05, + "loss": 0.3925, + "step": 2724 + }, + { + "epoch": 1.14, + "learning_rate": 1.9342370633257367e-05, + "loss": 0.3673, + "step": 2725 + }, + { + "epoch": 1.14, + "learning_rate": 1.9341765826929216e-05, + "loss": 0.4204, + "step": 2726 + }, + { + "epoch": 1.14, + "learning_rate": 1.9341160752080294e-05, + "loss": 0.429, + "step": 2727 + }, + { + "epoch": 1.14, + "learning_rate": 1.9340555408727993e-05, + "loss": 0.3993, + "step": 2728 + }, + { + "epoch": 1.14, + "learning_rate": 1.9339949796889717e-05, + "loss": 0.4509, + "step": 2729 + }, + { + "epoch": 1.14, + "learning_rate": 1.933934391658287e-05, + "loss": 0.3768, + "step": 2730 + }, + { + "epoch": 1.14, + "learning_rate": 1.933873776782487e-05, + "loss": 0.3504, + "step": 2731 + }, + { + "epoch": 1.14, + "learning_rate": 1.9338131350633143e-05, + "loss": 0.4084, + "step": 2732 + }, + { + "epoch": 1.14, + "learning_rate": 1.9337524665025114e-05, + "loss": 0.3781, + "step": 2733 + }, + { + "epoch": 1.14, + "learning_rate": 1.9336917711018227e-05, + "loss": 0.4215, + "step": 2734 + }, + { + "epoch": 1.15, + "learning_rate": 1.9336310488629926e-05, + "loss": 0.4066, + "step": 2735 + }, + { + "epoch": 1.15, + "learning_rate": 1.9335702997877664e-05, + "loss": 0.4147, + "step": 2736 + }, + { + "epoch": 1.15, + "learning_rate": 1.9335095238778903e-05, + "loss": 0.3655, + "step": 2737 + }, + { + "epoch": 1.15, + "learning_rate": 1.9334487211351113e-05, + "loss": 0.3868, + "step": 2738 + }, + { + "epoch": 1.15, + "learning_rate": 1.933387891561177e-05, + "loss": 0.3515, + "step": 2739 + }, + { + "epoch": 1.15, + "learning_rate": 1.9333270351578367e-05, + "loss": 0.3973, + "step": 2740 + }, + { + "epoch": 1.15, + "learning_rate": 1.9332661519268384e-05, + "loss": 0.4615, + "step": 2741 + }, + { + "epoch": 1.15, + "learning_rate": 1.933205241869933e-05, + "loss": 0.4009, + "step": 2742 + }, + { + "epoch": 1.15, + "learning_rate": 1.9331443049888706e-05, + "loss": 0.3969, + "step": 2743 + }, + { + "epoch": 1.15, + "learning_rate": 1.9330833412854035e-05, + "loss": 0.3819, + "step": 2744 + }, + { + "epoch": 1.15, + "learning_rate": 1.933022350761284e-05, + "loss": 0.3491, + "step": 2745 + }, + { + "epoch": 1.15, + "learning_rate": 1.9329613334182644e-05, + "loss": 0.3933, + "step": 2746 + }, + { + "epoch": 1.15, + "learning_rate": 1.9329002892580995e-05, + "loss": 0.3607, + "step": 2747 + }, + { + "epoch": 1.15, + "learning_rate": 1.932839218282544e-05, + "loss": 0.3915, + "step": 2748 + }, + { + "epoch": 1.15, + "learning_rate": 1.9327781204933522e-05, + "loss": 0.355, + "step": 2749 + }, + { + "epoch": 1.15, + "learning_rate": 1.9327169958922813e-05, + "loss": 0.4151, + "step": 2750 + }, + { + "epoch": 1.15, + "learning_rate": 1.9326558444810883e-05, + "loss": 0.3827, + "step": 2751 + }, + { + "epoch": 1.15, + "learning_rate": 1.9325946662615304e-05, + "loss": 0.4245, + "step": 2752 + }, + { + "epoch": 1.15, + "learning_rate": 1.932533461235367e-05, + "loss": 0.3551, + "step": 2753 + }, + { + "epoch": 1.15, + "learning_rate": 1.932472229404356e-05, + "loss": 0.3895, + "step": 2754 + }, + { + "epoch": 1.15, + "learning_rate": 1.9324109707702585e-05, + "loss": 0.3821, + "step": 2755 + }, + { + "epoch": 1.15, + "learning_rate": 1.932349685334835e-05, + "loss": 0.3928, + "step": 2756 + }, + { + "epoch": 1.15, + "learning_rate": 1.9322883730998474e-05, + "loss": 0.3673, + "step": 2757 + }, + { + "epoch": 1.15, + "learning_rate": 1.9322270340670574e-05, + "loss": 0.3496, + "step": 2758 + }, + { + "epoch": 1.16, + "learning_rate": 1.9321656682382287e-05, + "loss": 0.3826, + "step": 2759 + }, + { + "epoch": 1.16, + "learning_rate": 1.9321042756151254e-05, + "loss": 0.3732, + "step": 2760 + }, + { + "epoch": 1.16, + "learning_rate": 1.9320428561995114e-05, + "loss": 0.3547, + "step": 2761 + }, + { + "epoch": 1.16, + "learning_rate": 1.931981409993153e-05, + "loss": 0.337, + "step": 2762 + }, + { + "epoch": 1.16, + "learning_rate": 1.9319199369978157e-05, + "loss": 0.3734, + "step": 2763 + }, + { + "epoch": 1.16, + "learning_rate": 1.9318584372152668e-05, + "loss": 0.3379, + "step": 2764 + }, + { + "epoch": 1.16, + "learning_rate": 1.9317969106472742e-05, + "loss": 0.3891, + "step": 2765 + }, + { + "epoch": 1.16, + "learning_rate": 1.9317353572956058e-05, + "loss": 0.4175, + "step": 2766 + }, + { + "epoch": 1.16, + "learning_rate": 1.931673777162032e-05, + "loss": 0.3745, + "step": 2767 + }, + { + "epoch": 1.16, + "learning_rate": 1.9316121702483218e-05, + "loss": 0.421, + "step": 2768 + }, + { + "epoch": 1.16, + "learning_rate": 1.9315505365562464e-05, + "loss": 0.3868, + "step": 2769 + }, + { + "epoch": 1.16, + "learning_rate": 1.931488876087578e-05, + "loss": 0.3765, + "step": 2770 + }, + { + "epoch": 1.16, + "learning_rate": 1.9314271888440878e-05, + "loss": 0.3419, + "step": 2771 + }, + { + "epoch": 1.16, + "learning_rate": 1.9313654748275497e-05, + "loss": 0.4157, + "step": 2772 + }, + { + "epoch": 1.16, + "learning_rate": 1.931303734039738e-05, + "loss": 0.4043, + "step": 2773 + }, + { + "epoch": 1.16, + "learning_rate": 1.9312419664824268e-05, + "loss": 0.4398, + "step": 2774 + }, + { + "epoch": 1.16, + "learning_rate": 1.9311801721573913e-05, + "loss": 0.39, + "step": 2775 + }, + { + "epoch": 1.16, + "learning_rate": 1.931118351066408e-05, + "loss": 0.3564, + "step": 2776 + }, + { + "epoch": 1.16, + "learning_rate": 1.9310565032112542e-05, + "loss": 0.3823, + "step": 2777 + }, + { + "epoch": 1.16, + "learning_rate": 1.9309946285937075e-05, + "loss": 0.3924, + "step": 2778 + }, + { + "epoch": 1.16, + "learning_rate": 1.9309327272155462e-05, + "loss": 0.3895, + "step": 2779 + }, + { + "epoch": 1.16, + "learning_rate": 1.9308707990785497e-05, + "loss": 0.3785, + "step": 2780 + }, + { + "epoch": 1.16, + "learning_rate": 1.9308088441844982e-05, + "loss": 0.3857, + "step": 2781 + }, + { + "epoch": 1.16, + "learning_rate": 1.9307468625351723e-05, + "loss": 0.4335, + "step": 2782 + }, + { + "epoch": 1.17, + "learning_rate": 1.930684854132354e-05, + "loss": 0.3893, + "step": 2783 + }, + { + "epoch": 1.17, + "learning_rate": 1.9306228189778255e-05, + "loss": 0.4093, + "step": 2784 + }, + { + "epoch": 1.17, + "learning_rate": 1.9305607570733696e-05, + "loss": 0.4062, + "step": 2785 + }, + { + "epoch": 1.17, + "learning_rate": 1.930498668420771e-05, + "loss": 0.3321, + "step": 2786 + }, + { + "epoch": 1.17, + "learning_rate": 1.9304365530218134e-05, + "loss": 0.4056, + "step": 2787 + }, + { + "epoch": 1.17, + "learning_rate": 1.930374410878283e-05, + "loss": 0.4217, + "step": 2788 + }, + { + "epoch": 1.17, + "learning_rate": 1.9303122419919655e-05, + "loss": 0.3904, + "step": 2789 + }, + { + "epoch": 1.17, + "learning_rate": 1.9302500463646483e-05, + "loss": 0.3738, + "step": 2790 + }, + { + "epoch": 1.17, + "learning_rate": 1.930187823998119e-05, + "loss": 0.3978, + "step": 2791 + }, + { + "epoch": 1.17, + "learning_rate": 1.930125574894166e-05, + "loss": 0.3531, + "step": 2792 + }, + { + "epoch": 1.17, + "learning_rate": 1.930063299054579e-05, + "loss": 0.3876, + "step": 2793 + }, + { + "epoch": 1.17, + "learning_rate": 1.9300009964811475e-05, + "loss": 0.4144, + "step": 2794 + }, + { + "epoch": 1.17, + "learning_rate": 1.9299386671756628e-05, + "loss": 0.4041, + "step": 2795 + }, + { + "epoch": 1.17, + "learning_rate": 1.9298763111399162e-05, + "loss": 0.4336, + "step": 2796 + }, + { + "epoch": 1.17, + "learning_rate": 1.9298139283757004e-05, + "loss": 0.3942, + "step": 2797 + }, + { + "epoch": 1.17, + "learning_rate": 1.9297515188848083e-05, + "loss": 0.3601, + "step": 2798 + }, + { + "epoch": 1.17, + "learning_rate": 1.9296890826690337e-05, + "loss": 0.3929, + "step": 2799 + }, + { + "epoch": 1.17, + "learning_rate": 1.9296266197301716e-05, + "loss": 0.4062, + "step": 2800 + }, + { + "epoch": 1.17, + "learning_rate": 1.9295641300700173e-05, + "loss": 0.354, + "step": 2801 + }, + { + "epoch": 1.17, + "learning_rate": 1.9295016136903666e-05, + "loss": 0.4031, + "step": 2802 + }, + { + "epoch": 1.17, + "learning_rate": 1.929439070593017e-05, + "loss": 0.4086, + "step": 2803 + }, + { + "epoch": 1.17, + "learning_rate": 1.9293765007797662e-05, + "loss": 0.4178, + "step": 2804 + }, + { + "epoch": 1.17, + "learning_rate": 1.9293139042524124e-05, + "loss": 0.3496, + "step": 2805 + }, + { + "epoch": 1.17, + "learning_rate": 1.9292512810127554e-05, + "loss": 0.397, + "step": 2806 + }, + { + "epoch": 1.18, + "learning_rate": 1.9291886310625948e-05, + "loss": 0.3759, + "step": 2807 + }, + { + "epoch": 1.18, + "learning_rate": 1.9291259544037313e-05, + "loss": 0.3958, + "step": 2808 + }, + { + "epoch": 1.18, + "learning_rate": 1.9290632510379667e-05, + "loss": 0.404, + "step": 2809 + }, + { + "epoch": 1.18, + "learning_rate": 1.9290005209671036e-05, + "loss": 0.394, + "step": 2810 + }, + { + "epoch": 1.18, + "learning_rate": 1.928937764192945e-05, + "loss": 0.4051, + "step": 2811 + }, + { + "epoch": 1.18, + "learning_rate": 1.9288749807172947e-05, + "loss": 0.4203, + "step": 2812 + }, + { + "epoch": 1.18, + "learning_rate": 1.9288121705419572e-05, + "loss": 0.3902, + "step": 2813 + }, + { + "epoch": 1.18, + "learning_rate": 1.928749333668738e-05, + "loss": 0.3657, + "step": 2814 + }, + { + "epoch": 1.18, + "learning_rate": 1.9286864700994433e-05, + "loss": 0.3203, + "step": 2815 + }, + { + "epoch": 1.18, + "learning_rate": 1.9286235798358803e-05, + "loss": 0.3667, + "step": 2816 + }, + { + "epoch": 1.18, + "learning_rate": 1.9285606628798564e-05, + "loss": 0.4075, + "step": 2817 + }, + { + "epoch": 1.18, + "learning_rate": 1.9284977192331807e-05, + "loss": 0.4056, + "step": 2818 + }, + { + "epoch": 1.18, + "learning_rate": 1.9284347488976616e-05, + "loss": 0.4313, + "step": 2819 + }, + { + "epoch": 1.18, + "learning_rate": 1.9283717518751095e-05, + "loss": 0.3834, + "step": 2820 + }, + { + "epoch": 1.18, + "learning_rate": 1.9283087281673352e-05, + "loss": 0.3633, + "step": 2821 + }, + { + "epoch": 1.18, + "learning_rate": 1.9282456777761505e-05, + "loss": 0.3872, + "step": 2822 + }, + { + "epoch": 1.18, + "learning_rate": 1.9281826007033674e-05, + "loss": 0.4124, + "step": 2823 + }, + { + "epoch": 1.18, + "learning_rate": 1.928119496950799e-05, + "loss": 0.4013, + "step": 2824 + }, + { + "epoch": 1.18, + "learning_rate": 1.9280563665202595e-05, + "loss": 0.4354, + "step": 2825 + }, + { + "epoch": 1.18, + "learning_rate": 1.927993209413563e-05, + "loss": 0.3694, + "step": 2826 + }, + { + "epoch": 1.18, + "learning_rate": 1.927930025632525e-05, + "loss": 0.3888, + "step": 2827 + }, + { + "epoch": 1.18, + "learning_rate": 1.9278668151789627e-05, + "loss": 0.4104, + "step": 2828 + }, + { + "epoch": 1.18, + "learning_rate": 1.9278035780546912e-05, + "loss": 0.4169, + "step": 2829 + }, + { + "epoch": 1.18, + "learning_rate": 1.9277403142615297e-05, + "loss": 0.385, + "step": 2830 + }, + { + "epoch": 1.19, + "learning_rate": 1.927677023801296e-05, + "loss": 0.4508, + "step": 2831 + }, + { + "epoch": 1.19, + "learning_rate": 1.9276137066758097e-05, + "loss": 0.4015, + "step": 2832 + }, + { + "epoch": 1.19, + "learning_rate": 1.9275503628868902e-05, + "loss": 0.4012, + "step": 2833 + }, + { + "epoch": 1.19, + "learning_rate": 1.9274869924363585e-05, + "loss": 0.3566, + "step": 2834 + }, + { + "epoch": 1.19, + "learning_rate": 1.9274235953260367e-05, + "loss": 0.3656, + "step": 2835 + }, + { + "epoch": 1.19, + "learning_rate": 1.9273601715577463e-05, + "loss": 0.3389, + "step": 2836 + }, + { + "epoch": 1.19, + "learning_rate": 1.9272967211333107e-05, + "loss": 0.3845, + "step": 2837 + }, + { + "epoch": 1.19, + "learning_rate": 1.927233244054554e-05, + "loss": 0.372, + "step": 2838 + }, + { + "epoch": 1.19, + "learning_rate": 1.9271697403233e-05, + "loss": 0.4459, + "step": 2839 + }, + { + "epoch": 1.19, + "learning_rate": 1.9271062099413752e-05, + "loss": 0.3977, + "step": 2840 + }, + { + "epoch": 1.19, + "learning_rate": 1.9270426529106048e-05, + "loss": 0.3697, + "step": 2841 + }, + { + "epoch": 1.19, + "learning_rate": 1.926979069232816e-05, + "loss": 0.3596, + "step": 2842 + }, + { + "epoch": 1.19, + "learning_rate": 1.9269154589098363e-05, + "loss": 0.388, + "step": 2843 + }, + { + "epoch": 1.19, + "learning_rate": 1.9268518219434943e-05, + "loss": 0.4069, + "step": 2844 + }, + { + "epoch": 1.19, + "learning_rate": 1.9267881583356192e-05, + "loss": 0.367, + "step": 2845 + }, + { + "epoch": 1.19, + "learning_rate": 1.926724468088041e-05, + "loss": 0.3711, + "step": 2846 + }, + { + "epoch": 1.19, + "learning_rate": 1.9266607512025903e-05, + "loss": 0.3994, + "step": 2847 + }, + { + "epoch": 1.19, + "learning_rate": 1.9265970076810987e-05, + "loss": 0.4422, + "step": 2848 + }, + { + "epoch": 1.19, + "learning_rate": 1.9265332375253984e-05, + "loss": 0.3868, + "step": 2849 + }, + { + "epoch": 1.19, + "learning_rate": 1.926469440737322e-05, + "loss": 0.4381, + "step": 2850 + }, + { + "epoch": 1.19, + "learning_rate": 1.9264056173187037e-05, + "loss": 0.4296, + "step": 2851 + }, + { + "epoch": 1.19, + "learning_rate": 1.9263417672713786e-05, + "loss": 0.3709, + "step": 2852 + }, + { + "epoch": 1.19, + "learning_rate": 1.9262778905971808e-05, + "loss": 0.4049, + "step": 2853 + }, + { + "epoch": 1.19, + "learning_rate": 1.9262139872979474e-05, + "loss": 0.3886, + "step": 2854 + }, + { + "epoch": 1.2, + "learning_rate": 1.9261500573755146e-05, + "loss": 0.3822, + "step": 2855 + }, + { + "epoch": 1.2, + "learning_rate": 1.9260861008317203e-05, + "loss": 0.3676, + "step": 2856 + }, + { + "epoch": 1.2, + "learning_rate": 1.9260221176684025e-05, + "loss": 0.386, + "step": 2857 + }, + { + "epoch": 1.2, + "learning_rate": 1.925958107887401e-05, + "loss": 0.3864, + "step": 2858 + }, + { + "epoch": 1.2, + "learning_rate": 1.9258940714905554e-05, + "loss": 0.3853, + "step": 2859 + }, + { + "epoch": 1.2, + "learning_rate": 1.9258300084797063e-05, + "loss": 0.4195, + "step": 2860 + }, + { + "epoch": 1.2, + "learning_rate": 1.925765918856695e-05, + "loss": 0.3893, + "step": 2861 + }, + { + "epoch": 1.2, + "learning_rate": 1.9257018026233642e-05, + "loss": 0.4311, + "step": 2862 + }, + { + "epoch": 1.2, + "learning_rate": 1.9256376597815565e-05, + "loss": 0.3936, + "step": 2863 + }, + { + "epoch": 1.2, + "learning_rate": 1.9255734903331157e-05, + "loss": 0.4135, + "step": 2864 + }, + { + "epoch": 1.2, + "learning_rate": 1.9255092942798858e-05, + "loss": 0.3841, + "step": 2865 + }, + { + "epoch": 1.2, + "learning_rate": 1.9254450716237128e-05, + "loss": 0.394, + "step": 2866 + }, + { + "epoch": 1.2, + "learning_rate": 1.9253808223664427e-05, + "loss": 0.4184, + "step": 2867 + }, + { + "epoch": 1.2, + "learning_rate": 1.9253165465099217e-05, + "loss": 0.3678, + "step": 2868 + }, + { + "epoch": 1.2, + "learning_rate": 1.925252244055998e-05, + "loss": 0.4486, + "step": 2869 + }, + { + "epoch": 1.2, + "learning_rate": 1.9251879150065192e-05, + "loss": 0.3573, + "step": 2870 + }, + { + "epoch": 1.2, + "learning_rate": 1.925123559363335e-05, + "loss": 0.3508, + "step": 2871 + }, + { + "epoch": 1.2, + "learning_rate": 1.925059177128295e-05, + "loss": 0.3864, + "step": 2872 + }, + { + "epoch": 1.2, + "learning_rate": 1.9249947683032497e-05, + "loss": 0.3628, + "step": 2873 + }, + { + "epoch": 1.2, + "learning_rate": 1.9249303328900505e-05, + "loss": 0.3605, + "step": 2874 + }, + { + "epoch": 1.2, + "learning_rate": 1.92486587089055e-05, + "loss": 0.391, + "step": 2875 + }, + { + "epoch": 1.2, + "learning_rate": 1.9248013823066006e-05, + "loss": 0.4173, + "step": 2876 + }, + { + "epoch": 1.2, + "learning_rate": 1.924736867140056e-05, + "loss": 0.413, + "step": 2877 + }, + { + "epoch": 1.2, + "learning_rate": 1.9246723253927714e-05, + "loss": 0.4276, + "step": 2878 + }, + { + "epoch": 1.21, + "learning_rate": 1.9246077570666008e-05, + "loss": 0.4009, + "step": 2879 + }, + { + "epoch": 1.21, + "learning_rate": 1.9245431621634006e-05, + "loss": 0.3918, + "step": 2880 + }, + { + "epoch": 1.21, + "learning_rate": 1.924478540685028e-05, + "loss": 0.3812, + "step": 2881 + }, + { + "epoch": 1.21, + "learning_rate": 1.9244138926333396e-05, + "loss": 0.4158, + "step": 2882 + }, + { + "epoch": 1.21, + "learning_rate": 1.9243492180101947e-05, + "loss": 0.4269, + "step": 2883 + }, + { + "epoch": 1.21, + "learning_rate": 1.9242845168174514e-05, + "loss": 0.3786, + "step": 2884 + }, + { + "epoch": 1.21, + "learning_rate": 1.92421978905697e-05, + "loss": 0.3487, + "step": 2885 + }, + { + "epoch": 1.21, + "learning_rate": 1.924155034730611e-05, + "loss": 0.3736, + "step": 2886 + }, + { + "epoch": 1.21, + "learning_rate": 1.9240902538402354e-05, + "loss": 0.3477, + "step": 2887 + }, + { + "epoch": 1.21, + "learning_rate": 1.9240254463877055e-05, + "loss": 0.3401, + "step": 2888 + }, + { + "epoch": 1.21, + "learning_rate": 1.9239606123748845e-05, + "loss": 0.3752, + "step": 2889 + }, + { + "epoch": 1.21, + "learning_rate": 1.923895751803635e-05, + "loss": 0.37, + "step": 2890 + }, + { + "epoch": 1.21, + "learning_rate": 1.9238308646758224e-05, + "loss": 0.3883, + "step": 2891 + }, + { + "epoch": 1.21, + "learning_rate": 1.9237659509933114e-05, + "loss": 0.3907, + "step": 2892 + }, + { + "epoch": 1.21, + "learning_rate": 1.9237010107579676e-05, + "loss": 0.3955, + "step": 2893 + }, + { + "epoch": 1.21, + "learning_rate": 1.923636043971658e-05, + "loss": 0.3549, + "step": 2894 + }, + { + "epoch": 1.21, + "learning_rate": 1.92357105063625e-05, + "loss": 0.3955, + "step": 2895 + }, + { + "epoch": 1.21, + "learning_rate": 1.9235060307536117e-05, + "loss": 0.3917, + "step": 2896 + }, + { + "epoch": 1.21, + "learning_rate": 1.923440984325612e-05, + "loss": 0.3484, + "step": 2897 + }, + { + "epoch": 1.21, + "learning_rate": 1.923375911354121e-05, + "loss": 0.4145, + "step": 2898 + }, + { + "epoch": 1.21, + "learning_rate": 1.9233108118410082e-05, + "loss": 0.3835, + "step": 2899 + }, + { + "epoch": 1.21, + "learning_rate": 1.923245685788146e-05, + "loss": 0.4056, + "step": 2900 + }, + { + "epoch": 1.21, + "learning_rate": 1.9231805331974053e-05, + "loss": 0.3901, + "step": 2901 + }, + { + "epoch": 1.21, + "learning_rate": 1.9231153540706597e-05, + "loss": 0.4369, + "step": 2902 + }, + { + "epoch": 1.22, + "learning_rate": 1.9230501484097826e-05, + "loss": 0.379, + "step": 2903 + }, + { + "epoch": 1.22, + "learning_rate": 1.9229849162166478e-05, + "loss": 0.3816, + "step": 2904 + }, + { + "epoch": 1.22, + "learning_rate": 1.922919657493131e-05, + "loss": 0.4186, + "step": 2905 + }, + { + "epoch": 1.22, + "learning_rate": 1.922854372241107e-05, + "loss": 0.4063, + "step": 2906 + }, + { + "epoch": 1.22, + "learning_rate": 1.9227890604624534e-05, + "loss": 0.4478, + "step": 2907 + }, + { + "epoch": 1.22, + "learning_rate": 1.9227237221590472e-05, + "loss": 0.4357, + "step": 2908 + }, + { + "epoch": 1.22, + "learning_rate": 1.922658357332766e-05, + "loss": 0.3672, + "step": 2909 + }, + { + "epoch": 1.22, + "learning_rate": 1.9225929659854893e-05, + "loss": 0.3548, + "step": 2910 + }, + { + "epoch": 1.22, + "learning_rate": 1.922527548119097e-05, + "loss": 0.3264, + "step": 2911 + }, + { + "epoch": 1.22, + "learning_rate": 1.922462103735468e-05, + "loss": 0.3426, + "step": 2912 + }, + { + "epoch": 1.22, + "learning_rate": 1.9223966328364853e-05, + "loss": 0.3965, + "step": 2913 + }, + { + "epoch": 1.22, + "learning_rate": 1.922331135424029e-05, + "loss": 0.4179, + "step": 2914 + }, + { + "epoch": 1.22, + "learning_rate": 1.9222656114999835e-05, + "loss": 0.372, + "step": 2915 + }, + { + "epoch": 1.22, + "learning_rate": 1.922200061066231e-05, + "loss": 0.4026, + "step": 2916 + }, + { + "epoch": 1.22, + "learning_rate": 1.9221344841246562e-05, + "loss": 0.356, + "step": 2917 + }, + { + "epoch": 1.22, + "learning_rate": 1.9220688806771443e-05, + "loss": 0.3802, + "step": 2918 + }, + { + "epoch": 1.22, + "learning_rate": 1.92200325072558e-05, + "loss": 0.4339, + "step": 2919 + }, + { + "epoch": 1.22, + "learning_rate": 1.9219375942718508e-05, + "loss": 0.3936, + "step": 2920 + }, + { + "epoch": 1.22, + "learning_rate": 1.9218719113178433e-05, + "loss": 0.3785, + "step": 2921 + }, + { + "epoch": 1.22, + "learning_rate": 1.9218062018654458e-05, + "loss": 0.3854, + "step": 2922 + }, + { + "epoch": 1.22, + "learning_rate": 1.9217404659165476e-05, + "loss": 0.3722, + "step": 2923 + }, + { + "epoch": 1.22, + "learning_rate": 1.921674703473037e-05, + "loss": 0.3839, + "step": 2924 + }, + { + "epoch": 1.22, + "learning_rate": 1.9216089145368052e-05, + "loss": 0.4487, + "step": 2925 + }, + { + "epoch": 1.23, + "learning_rate": 1.921543099109743e-05, + "loss": 0.3922, + "step": 2926 + }, + { + "epoch": 1.23, + "learning_rate": 1.921477257193742e-05, + "loss": 0.393, + "step": 2927 + }, + { + "epoch": 1.23, + "learning_rate": 1.9214113887906947e-05, + "loss": 0.3938, + "step": 2928 + }, + { + "epoch": 1.23, + "learning_rate": 1.9213454939024953e-05, + "loss": 0.3843, + "step": 2929 + }, + { + "epoch": 1.23, + "learning_rate": 1.9212795725310372e-05, + "loss": 0.3286, + "step": 2930 + }, + { + "epoch": 1.23, + "learning_rate": 1.921213624678215e-05, + "loss": 0.3765, + "step": 2931 + }, + { + "epoch": 1.23, + "learning_rate": 1.921147650345925e-05, + "loss": 0.3984, + "step": 2932 + }, + { + "epoch": 1.23, + "learning_rate": 1.9210816495360625e-05, + "loss": 0.4522, + "step": 2933 + }, + { + "epoch": 1.23, + "learning_rate": 1.921015622250526e-05, + "loss": 0.3989, + "step": 2934 + }, + { + "epoch": 1.23, + "learning_rate": 1.920949568491212e-05, + "loss": 0.4184, + "step": 2935 + }, + { + "epoch": 1.23, + "learning_rate": 1.9208834882600207e-05, + "loss": 0.3787, + "step": 2936 + }, + { + "epoch": 1.23, + "learning_rate": 1.9208173815588506e-05, + "loss": 0.4151, + "step": 2937 + }, + { + "epoch": 1.23, + "learning_rate": 1.9207512483896016e-05, + "loss": 0.4006, + "step": 2938 + }, + { + "epoch": 1.23, + "learning_rate": 1.920685088754175e-05, + "loss": 0.3855, + "step": 2939 + }, + { + "epoch": 1.23, + "learning_rate": 1.9206189026544728e-05, + "loss": 0.417, + "step": 2940 + }, + { + "epoch": 1.23, + "learning_rate": 1.920552690092397e-05, + "loss": 0.3611, + "step": 2941 + }, + { + "epoch": 1.23, + "learning_rate": 1.9204864510698513e-05, + "loss": 0.3916, + "step": 2942 + }, + { + "epoch": 1.23, + "learning_rate": 1.9204201855887394e-05, + "loss": 0.3599, + "step": 2943 + }, + { + "epoch": 1.23, + "learning_rate": 1.920353893650966e-05, + "loss": 0.3919, + "step": 2944 + }, + { + "epoch": 1.23, + "learning_rate": 1.9202875752584363e-05, + "loss": 0.3776, + "step": 2945 + }, + { + "epoch": 1.23, + "learning_rate": 1.920221230413057e-05, + "loss": 0.41, + "step": 2946 + }, + { + "epoch": 1.23, + "learning_rate": 1.920154859116735e-05, + "loss": 0.4023, + "step": 2947 + }, + { + "epoch": 1.23, + "learning_rate": 1.9200884613713782e-05, + "loss": 0.3815, + "step": 2948 + }, + { + "epoch": 1.23, + "learning_rate": 1.9200220371788952e-05, + "loss": 0.3629, + "step": 2949 + }, + { + "epoch": 1.24, + "learning_rate": 1.9199555865411948e-05, + "loss": 0.4138, + "step": 2950 + }, + { + "epoch": 1.24, + "learning_rate": 1.9198891094601876e-05, + "loss": 0.3576, + "step": 2951 + }, + { + "epoch": 1.24, + "learning_rate": 1.9198226059377846e-05, + "loss": 0.3736, + "step": 2952 + }, + { + "epoch": 1.24, + "learning_rate": 1.9197560759758963e-05, + "loss": 0.4181, + "step": 2953 + }, + { + "epoch": 1.24, + "learning_rate": 1.9196895195764363e-05, + "loss": 0.3947, + "step": 2954 + }, + { + "epoch": 1.24, + "learning_rate": 1.9196229367413172e-05, + "loss": 0.3668, + "step": 2955 + }, + { + "epoch": 1.24, + "learning_rate": 1.9195563274724523e-05, + "loss": 0.3783, + "step": 2956 + }, + { + "epoch": 1.24, + "learning_rate": 1.9194896917717573e-05, + "loss": 0.3779, + "step": 2957 + }, + { + "epoch": 1.24, + "learning_rate": 1.919423029641147e-05, + "loss": 0.3688, + "step": 2958 + }, + { + "epoch": 1.24, + "learning_rate": 1.9193563410825376e-05, + "loss": 0.3632, + "step": 2959 + }, + { + "epoch": 1.24, + "learning_rate": 1.9192896260978457e-05, + "loss": 0.4211, + "step": 2960 + }, + { + "epoch": 1.24, + "learning_rate": 1.9192228846889896e-05, + "loss": 0.4101, + "step": 2961 + }, + { + "epoch": 1.24, + "learning_rate": 1.9191561168578873e-05, + "loss": 0.3919, + "step": 2962 + }, + { + "epoch": 1.24, + "learning_rate": 1.919089322606458e-05, + "loss": 0.3991, + "step": 2963 + }, + { + "epoch": 1.24, + "learning_rate": 1.919022501936622e-05, + "loss": 0.3567, + "step": 2964 + }, + { + "epoch": 1.24, + "learning_rate": 1.918955654850299e-05, + "loss": 0.3857, + "step": 2965 + }, + { + "epoch": 1.24, + "learning_rate": 1.9188887813494118e-05, + "loss": 0.4223, + "step": 2966 + }, + { + "epoch": 1.24, + "learning_rate": 1.9188218814358817e-05, + "loss": 0.4015, + "step": 2967 + }, + { + "epoch": 1.24, + "learning_rate": 1.9187549551116322e-05, + "loss": 0.4114, + "step": 2968 + }, + { + "epoch": 1.24, + "learning_rate": 1.9186880023785864e-05, + "loss": 0.4167, + "step": 2969 + }, + { + "epoch": 1.24, + "learning_rate": 1.9186210232386694e-05, + "loss": 0.3801, + "step": 2970 + }, + { + "epoch": 1.24, + "learning_rate": 1.9185540176938065e-05, + "loss": 0.3913, + "step": 2971 + }, + { + "epoch": 1.24, + "learning_rate": 1.9184869857459233e-05, + "loss": 0.3867, + "step": 2972 + }, + { + "epoch": 1.24, + "learning_rate": 1.9184199273969465e-05, + "loss": 0.3883, + "step": 2973 + }, + { + "epoch": 1.25, + "learning_rate": 1.918352842648804e-05, + "loss": 0.4711, + "step": 2974 + }, + { + "epoch": 1.25, + "learning_rate": 1.918285731503424e-05, + "loss": 0.4078, + "step": 2975 + }, + { + "epoch": 1.25, + "learning_rate": 1.9182185939627356e-05, + "loss": 0.3993, + "step": 2976 + }, + { + "epoch": 1.25, + "learning_rate": 1.918151430028669e-05, + "loss": 0.3231, + "step": 2977 + }, + { + "epoch": 1.25, + "learning_rate": 1.9180842397031535e-05, + "loss": 0.4097, + "step": 2978 + }, + { + "epoch": 1.25, + "learning_rate": 1.9180170229881216e-05, + "loss": 0.3751, + "step": 2979 + }, + { + "epoch": 1.25, + "learning_rate": 1.9179497798855048e-05, + "loss": 0.3605, + "step": 2980 + }, + { + "epoch": 1.25, + "learning_rate": 1.9178825103972364e-05, + "loss": 0.3395, + "step": 2981 + }, + { + "epoch": 1.25, + "learning_rate": 1.9178152145252497e-05, + "loss": 0.3948, + "step": 2982 + }, + { + "epoch": 1.25, + "learning_rate": 1.917747892271479e-05, + "loss": 0.3671, + "step": 2983 + }, + { + "epoch": 1.25, + "learning_rate": 1.9176805436378597e-05, + "loss": 0.3953, + "step": 2984 + }, + { + "epoch": 1.25, + "learning_rate": 1.9176131686263274e-05, + "loss": 0.4075, + "step": 2985 + }, + { + "epoch": 1.25, + "learning_rate": 1.917545767238819e-05, + "loss": 0.3921, + "step": 2986 + }, + { + "epoch": 1.25, + "learning_rate": 1.9174783394772714e-05, + "loss": 0.4225, + "step": 2987 + }, + { + "epoch": 1.25, + "learning_rate": 1.9174108853436234e-05, + "loss": 0.3768, + "step": 2988 + }, + { + "epoch": 1.25, + "learning_rate": 1.9173434048398133e-05, + "loss": 0.3695, + "step": 2989 + }, + { + "epoch": 1.25, + "learning_rate": 1.917275897967781e-05, + "loss": 0.4141, + "step": 2990 + }, + { + "epoch": 1.25, + "learning_rate": 1.9172083647294676e-05, + "loss": 0.4165, + "step": 2991 + }, + { + "epoch": 1.25, + "learning_rate": 1.9171408051268132e-05, + "loss": 0.4475, + "step": 2992 + }, + { + "epoch": 1.25, + "learning_rate": 1.9170732191617603e-05, + "loss": 0.3717, + "step": 2993 + }, + { + "epoch": 1.25, + "learning_rate": 1.9170056068362516e-05, + "loss": 0.4643, + "step": 2994 + }, + { + "epoch": 1.25, + "learning_rate": 1.9169379681522305e-05, + "loss": 0.3985, + "step": 2995 + }, + { + "epoch": 1.25, + "learning_rate": 1.9168703031116412e-05, + "loss": 0.3602, + "step": 2996 + }, + { + "epoch": 1.25, + "learning_rate": 1.9168026117164285e-05, + "loss": 0.3634, + "step": 2997 + }, + { + "epoch": 1.26, + "learning_rate": 1.916734893968538e-05, + "loss": 0.4244, + "step": 2998 + }, + { + "epoch": 1.26, + "learning_rate": 1.916667149869917e-05, + "loss": 0.3957, + "step": 2999 + }, + { + "epoch": 1.26, + "learning_rate": 1.916599379422512e-05, + "loss": 0.3868, + "step": 3000 + }, + { + "epoch": 1.26, + "learning_rate": 1.9165315826282712e-05, + "loss": 0.3846, + "step": 3001 + }, + { + "epoch": 1.26, + "learning_rate": 1.916463759489143e-05, + "loss": 0.4195, + "step": 3002 + }, + { + "epoch": 1.26, + "learning_rate": 1.916395910007078e-05, + "loss": 0.4048, + "step": 3003 + }, + { + "epoch": 1.26, + "learning_rate": 1.9163280341840253e-05, + "loss": 0.3728, + "step": 3004 + }, + { + "epoch": 1.26, + "learning_rate": 1.9162601320219364e-05, + "loss": 0.419, + "step": 3005 + }, + { + "epoch": 1.26, + "learning_rate": 1.9161922035227633e-05, + "loss": 0.3842, + "step": 3006 + }, + { + "epoch": 1.26, + "learning_rate": 1.9161242486884584e-05, + "loss": 0.4437, + "step": 3007 + }, + { + "epoch": 1.26, + "learning_rate": 1.9160562675209746e-05, + "loss": 0.4448, + "step": 3008 + }, + { + "epoch": 1.26, + "learning_rate": 1.9159882600222664e-05, + "loss": 0.4838, + "step": 3009 + }, + { + "epoch": 1.26, + "learning_rate": 1.9159202261942886e-05, + "loss": 0.3661, + "step": 3010 + }, + { + "epoch": 1.26, + "learning_rate": 1.9158521660389966e-05, + "loss": 0.3588, + "step": 3011 + }, + { + "epoch": 1.26, + "learning_rate": 1.9157840795583466e-05, + "loss": 0.452, + "step": 3012 + }, + { + "epoch": 1.26, + "learning_rate": 1.9157159667542964e-05, + "loss": 0.3669, + "step": 3013 + }, + { + "epoch": 1.26, + "learning_rate": 1.9156478276288027e-05, + "loss": 0.4242, + "step": 3014 + }, + { + "epoch": 1.26, + "learning_rate": 1.9155796621838253e-05, + "loss": 0.3988, + "step": 3015 + }, + { + "epoch": 1.26, + "learning_rate": 1.915511470421323e-05, + "loss": 0.3667, + "step": 3016 + }, + { + "epoch": 1.26, + "learning_rate": 1.9154432523432553e-05, + "loss": 0.4029, + "step": 3017 + }, + { + "epoch": 1.26, + "learning_rate": 1.915375007951584e-05, + "loss": 0.3726, + "step": 3018 + }, + { + "epoch": 1.26, + "learning_rate": 1.9153067372482702e-05, + "loss": 0.4151, + "step": 3019 + }, + { + "epoch": 1.26, + "learning_rate": 1.915238440235277e-05, + "loss": 0.3597, + "step": 3020 + }, + { + "epoch": 1.26, + "learning_rate": 1.9151701169145664e-05, + "loss": 0.3604, + "step": 3021 + }, + { + "epoch": 1.27, + "learning_rate": 1.9151017672881032e-05, + "loss": 0.4356, + "step": 3022 + }, + { + "epoch": 1.27, + "learning_rate": 1.9150333913578518e-05, + "loss": 0.4045, + "step": 3023 + }, + { + "epoch": 1.27, + "learning_rate": 1.9149649891257772e-05, + "loss": 0.379, + "step": 3024 + }, + { + "epoch": 1.27, + "learning_rate": 1.9148965605938463e-05, + "loss": 0.3753, + "step": 3025 + }, + { + "epoch": 1.27, + "learning_rate": 1.9148281057640252e-05, + "loss": 0.4016, + "step": 3026 + }, + { + "epoch": 1.27, + "learning_rate": 1.914759624638282e-05, + "loss": 0.3454, + "step": 3027 + }, + { + "epoch": 1.27, + "learning_rate": 1.914691117218586e-05, + "loss": 0.3758, + "step": 3028 + }, + { + "epoch": 1.27, + "learning_rate": 1.914622583506905e-05, + "loss": 0.449, + "step": 3029 + }, + { + "epoch": 1.27, + "learning_rate": 1.914554023505209e-05, + "loss": 0.3803, + "step": 3030 + }, + { + "epoch": 1.27, + "learning_rate": 1.9144854372154697e-05, + "loss": 0.3536, + "step": 3031 + }, + { + "epoch": 1.27, + "learning_rate": 1.9144168246396576e-05, + "loss": 0.4062, + "step": 3032 + }, + { + "epoch": 1.27, + "learning_rate": 1.914348185779746e-05, + "loss": 0.3956, + "step": 3033 + }, + { + "epoch": 1.27, + "learning_rate": 1.9142795206377066e-05, + "loss": 0.4077, + "step": 3034 + }, + { + "epoch": 1.27, + "learning_rate": 1.914210829215514e-05, + "loss": 0.4146, + "step": 3035 + }, + { + "epoch": 1.27, + "learning_rate": 1.914142111515142e-05, + "loss": 0.4158, + "step": 3036 + }, + { + "epoch": 1.27, + "learning_rate": 1.9140733675385666e-05, + "loss": 0.4014, + "step": 3037 + }, + { + "epoch": 1.27, + "learning_rate": 1.914004597287763e-05, + "loss": 0.3813, + "step": 3038 + }, + { + "epoch": 1.27, + "learning_rate": 1.9139358007647085e-05, + "loss": 0.3681, + "step": 3039 + }, + { + "epoch": 1.27, + "learning_rate": 1.9138669779713807e-05, + "loss": 0.409, + "step": 3040 + }, + { + "epoch": 1.27, + "learning_rate": 1.9137981289097573e-05, + "loss": 0.4381, + "step": 3041 + }, + { + "epoch": 1.27, + "learning_rate": 1.9137292535818174e-05, + "loss": 0.3638, + "step": 3042 + }, + { + "epoch": 1.27, + "learning_rate": 1.913660351989541e-05, + "loss": 0.4291, + "step": 3043 + }, + { + "epoch": 1.27, + "learning_rate": 1.9135914241349088e-05, + "loss": 0.4002, + "step": 3044 + }, + { + "epoch": 1.27, + "learning_rate": 1.9135224700199015e-05, + "loss": 0.3697, + "step": 3045 + }, + { + "epoch": 1.28, + "learning_rate": 1.913453489646502e-05, + "loss": 0.4188, + "step": 3046 + }, + { + "epoch": 1.28, + "learning_rate": 1.913384483016692e-05, + "loss": 0.4102, + "step": 3047 + }, + { + "epoch": 1.28, + "learning_rate": 1.9133154501324557e-05, + "loss": 0.3559, + "step": 3048 + }, + { + "epoch": 1.28, + "learning_rate": 1.9132463909957774e-05, + "loss": 0.4174, + "step": 3049 + }, + { + "epoch": 1.28, + "learning_rate": 1.9131773056086417e-05, + "loss": 0.4116, + "step": 3050 + }, + { + "epoch": 1.28, + "learning_rate": 1.9131081939730348e-05, + "loss": 0.4361, + "step": 3051 + }, + { + "epoch": 1.28, + "learning_rate": 1.913039056090943e-05, + "loss": 0.3952, + "step": 3052 + }, + { + "epoch": 1.28, + "learning_rate": 1.912969891964354e-05, + "loss": 0.3878, + "step": 3053 + }, + { + "epoch": 1.28, + "learning_rate": 1.9129007015952556e-05, + "loss": 0.36, + "step": 3054 + }, + { + "epoch": 1.28, + "learning_rate": 1.9128314849856365e-05, + "loss": 0.3476, + "step": 3055 + }, + { + "epoch": 1.28, + "learning_rate": 1.9127622421374866e-05, + "loss": 0.3948, + "step": 3056 + }, + { + "epoch": 1.28, + "learning_rate": 1.9126929730527957e-05, + "loss": 0.3804, + "step": 3057 + }, + { + "epoch": 1.28, + "learning_rate": 1.9126236777335554e-05, + "loss": 0.4198, + "step": 3058 + }, + { + "epoch": 1.28, + "learning_rate": 1.9125543561817573e-05, + "loss": 0.3692, + "step": 3059 + }, + { + "epoch": 1.28, + "learning_rate": 1.9124850083993942e-05, + "loss": 0.3862, + "step": 3060 + }, + { + "epoch": 1.28, + "learning_rate": 1.912415634388459e-05, + "loss": 0.4109, + "step": 3061 + }, + { + "epoch": 1.28, + "learning_rate": 1.912346234150946e-05, + "loss": 0.3838, + "step": 3062 + }, + { + "epoch": 1.28, + "learning_rate": 1.9122768076888506e-05, + "loss": 0.3874, + "step": 3063 + }, + { + "epoch": 1.28, + "learning_rate": 1.9122073550041675e-05, + "loss": 0.4108, + "step": 3064 + }, + { + "epoch": 1.28, + "learning_rate": 1.9121378760988934e-05, + "loss": 0.3832, + "step": 3065 + }, + { + "epoch": 1.28, + "learning_rate": 1.9120683709750257e-05, + "loss": 0.3902, + "step": 3066 + }, + { + "epoch": 1.28, + "learning_rate": 1.9119988396345615e-05, + "loss": 0.3786, + "step": 3067 + }, + { + "epoch": 1.28, + "learning_rate": 1.9119292820795007e-05, + "loss": 0.4007, + "step": 3068 + }, + { + "epoch": 1.28, + "learning_rate": 1.9118596983118415e-05, + "loss": 0.3548, + "step": 3069 + }, + { + "epoch": 1.29, + "learning_rate": 1.9117900883335846e-05, + "loss": 0.3979, + "step": 3070 + }, + { + "epoch": 1.29, + "learning_rate": 1.9117204521467308e-05, + "loss": 0.3693, + "step": 3071 + }, + { + "epoch": 1.29, + "learning_rate": 1.9116507897532813e-05, + "loss": 0.3875, + "step": 3072 + }, + { + "epoch": 1.29, + "learning_rate": 1.911581101155239e-05, + "loss": 0.3935, + "step": 3073 + }, + { + "epoch": 1.29, + "learning_rate": 1.911511386354607e-05, + "loss": 0.3525, + "step": 3074 + }, + { + "epoch": 1.29, + "learning_rate": 1.911441645353389e-05, + "loss": 0.3911, + "step": 3075 + }, + { + "epoch": 1.29, + "learning_rate": 1.9113718781535896e-05, + "loss": 0.3531, + "step": 3076 + }, + { + "epoch": 1.29, + "learning_rate": 1.9113020847572144e-05, + "loss": 0.4023, + "step": 3077 + }, + { + "epoch": 1.29, + "learning_rate": 1.911232265166269e-05, + "loss": 0.3949, + "step": 3078 + }, + { + "epoch": 1.29, + "learning_rate": 1.9111624193827614e-05, + "loss": 0.3762, + "step": 3079 + }, + { + "epoch": 1.29, + "learning_rate": 1.9110925474086982e-05, + "loss": 0.3759, + "step": 3080 + }, + { + "epoch": 1.29, + "learning_rate": 1.9110226492460886e-05, + "loss": 0.4094, + "step": 3081 + }, + { + "epoch": 1.29, + "learning_rate": 1.9109527248969412e-05, + "loss": 0.4151, + "step": 3082 + }, + { + "epoch": 1.29, + "learning_rate": 1.910882774363266e-05, + "loss": 0.3605, + "step": 3083 + }, + { + "epoch": 1.29, + "learning_rate": 1.9108127976470737e-05, + "loss": 0.3758, + "step": 3084 + }, + { + "epoch": 1.29, + "learning_rate": 1.9107427947503757e-05, + "loss": 0.3833, + "step": 3085 + }, + { + "epoch": 1.29, + "learning_rate": 1.9106727656751846e-05, + "loss": 0.3986, + "step": 3086 + }, + { + "epoch": 1.29, + "learning_rate": 1.9106027104235126e-05, + "loss": 0.3984, + "step": 3087 + }, + { + "epoch": 1.29, + "learning_rate": 1.910532628997374e-05, + "loss": 0.4009, + "step": 3088 + }, + { + "epoch": 1.29, + "learning_rate": 1.910462521398783e-05, + "loss": 0.3672, + "step": 3089 + }, + { + "epoch": 1.29, + "learning_rate": 1.9103923876297544e-05, + "loss": 0.3765, + "step": 3090 + }, + { + "epoch": 1.29, + "learning_rate": 1.9103222276923047e-05, + "loss": 0.3638, + "step": 3091 + }, + { + "epoch": 1.29, + "learning_rate": 1.9102520415884504e-05, + "loss": 0.3996, + "step": 3092 + }, + { + "epoch": 1.29, + "learning_rate": 1.910181829320209e-05, + "loss": 0.3812, + "step": 3093 + }, + { + "epoch": 1.3, + "learning_rate": 1.9101115908895985e-05, + "loss": 0.3676, + "step": 3094 + }, + { + "epoch": 1.3, + "learning_rate": 1.9100413262986383e-05, + "loss": 0.3927, + "step": 3095 + }, + { + "epoch": 1.3, + "learning_rate": 1.909971035549347e-05, + "loss": 0.4282, + "step": 3096 + }, + { + "epoch": 1.3, + "learning_rate": 1.9099007186437465e-05, + "loss": 0.4489, + "step": 3097 + }, + { + "epoch": 1.3, + "learning_rate": 1.9098303755838566e-05, + "loss": 0.3486, + "step": 3098 + }, + { + "epoch": 1.3, + "learning_rate": 1.9097600063717005e-05, + "loss": 0.3687, + "step": 3099 + }, + { + "epoch": 1.3, + "learning_rate": 1.9096896110092998e-05, + "loss": 0.4529, + "step": 3100 + }, + { + "epoch": 1.3, + "learning_rate": 1.909619189498679e-05, + "loss": 0.3496, + "step": 3101 + }, + { + "epoch": 1.3, + "learning_rate": 1.9095487418418614e-05, + "loss": 0.4179, + "step": 3102 + }, + { + "epoch": 1.3, + "learning_rate": 1.9094782680408722e-05, + "loss": 0.4081, + "step": 3103 + }, + { + "epoch": 1.3, + "learning_rate": 1.9094077680977377e-05, + "loss": 0.3881, + "step": 3104 + }, + { + "epoch": 1.3, + "learning_rate": 1.9093372420144835e-05, + "loss": 0.4002, + "step": 3105 + }, + { + "epoch": 1.3, + "learning_rate": 1.9092666897931375e-05, + "loss": 0.3316, + "step": 3106 + }, + { + "epoch": 1.3, + "learning_rate": 1.9091961114357274e-05, + "loss": 0.4296, + "step": 3107 + }, + { + "epoch": 1.3, + "learning_rate": 1.909125506944282e-05, + "loss": 0.3749, + "step": 3108 + }, + { + "epoch": 1.3, + "learning_rate": 1.90905487632083e-05, + "loss": 0.427, + "step": 3109 + }, + { + "epoch": 1.3, + "learning_rate": 1.908984219567403e-05, + "loss": 0.386, + "step": 3110 + }, + { + "epoch": 1.3, + "learning_rate": 1.908913536686031e-05, + "loss": 0.4013, + "step": 3111 + }, + { + "epoch": 1.3, + "learning_rate": 1.9088428276787455e-05, + "loss": 0.3712, + "step": 3112 + }, + { + "epoch": 1.3, + "learning_rate": 1.90877209254758e-05, + "loss": 0.4584, + "step": 3113 + }, + { + "epoch": 1.3, + "learning_rate": 1.9087013312945668e-05, + "loss": 0.41, + "step": 3114 + }, + { + "epoch": 1.3, + "learning_rate": 1.9086305439217402e-05, + "loss": 0.3809, + "step": 3115 + }, + { + "epoch": 1.3, + "learning_rate": 1.908559730431135e-05, + "loss": 0.3908, + "step": 3116 + }, + { + "epoch": 1.31, + "learning_rate": 1.9084888908247864e-05, + "loss": 0.4066, + "step": 3117 + }, + { + "epoch": 1.31, + "learning_rate": 1.908418025104731e-05, + "loss": 0.4241, + "step": 3118 + }, + { + "epoch": 1.31, + "learning_rate": 1.9083471332730058e-05, + "loss": 0.3628, + "step": 3119 + }, + { + "epoch": 1.31, + "learning_rate": 1.908276215331648e-05, + "loss": 0.469, + "step": 3120 + }, + { + "epoch": 1.31, + "learning_rate": 1.9082052712826967e-05, + "loss": 0.3736, + "step": 3121 + }, + { + "epoch": 1.31, + "learning_rate": 1.9081343011281907e-05, + "loss": 0.3628, + "step": 3122 + }, + { + "epoch": 1.31, + "learning_rate": 1.9080633048701696e-05, + "loss": 0.436, + "step": 3123 + }, + { + "epoch": 1.31, + "learning_rate": 1.907992282510675e-05, + "loss": 0.4212, + "step": 3124 + }, + { + "epoch": 1.31, + "learning_rate": 1.907921234051748e-05, + "loss": 0.3986, + "step": 3125 + }, + { + "epoch": 1.31, + "learning_rate": 1.9078501594954305e-05, + "loss": 0.4126, + "step": 3126 + }, + { + "epoch": 1.31, + "learning_rate": 1.9077790588437663e-05, + "loss": 0.3649, + "step": 3127 + }, + { + "epoch": 1.31, + "learning_rate": 1.907707932098798e-05, + "loss": 0.3977, + "step": 3128 + }, + { + "epoch": 1.31, + "learning_rate": 1.9076367792625713e-05, + "loss": 0.368, + "step": 3129 + }, + { + "epoch": 1.31, + "learning_rate": 1.9075656003371304e-05, + "loss": 0.3954, + "step": 3130 + }, + { + "epoch": 1.31, + "learning_rate": 1.907494395324522e-05, + "loss": 0.3712, + "step": 3131 + }, + { + "epoch": 1.31, + "learning_rate": 1.907423164226792e-05, + "loss": 0.3879, + "step": 3132 + }, + { + "epoch": 1.31, + "learning_rate": 1.907351907045989e-05, + "loss": 0.375, + "step": 3133 + }, + { + "epoch": 1.31, + "learning_rate": 1.90728062378416e-05, + "loss": 0.4481, + "step": 3134 + }, + { + "epoch": 1.31, + "learning_rate": 1.907209314443355e-05, + "loss": 0.4373, + "step": 3135 + }, + { + "epoch": 1.31, + "learning_rate": 1.907137979025623e-05, + "loss": 0.4099, + "step": 3136 + }, + { + "epoch": 1.31, + "learning_rate": 1.907066617533015e-05, + "loss": 0.39, + "step": 3137 + }, + { + "epoch": 1.31, + "learning_rate": 1.906995229967582e-05, + "loss": 0.3978, + "step": 3138 + }, + { + "epoch": 1.31, + "learning_rate": 1.9069238163313757e-05, + "loss": 0.4477, + "step": 3139 + }, + { + "epoch": 1.31, + "learning_rate": 1.906852376626449e-05, + "loss": 0.3745, + "step": 3140 + }, + { + "epoch": 1.32, + "learning_rate": 1.906780910854856e-05, + "loss": 0.3677, + "step": 3141 + }, + { + "epoch": 1.32, + "learning_rate": 1.90670941901865e-05, + "loss": 0.359, + "step": 3142 + }, + { + "epoch": 1.32, + "learning_rate": 1.9066379011198863e-05, + "loss": 0.3517, + "step": 3143 + }, + { + "epoch": 1.32, + "learning_rate": 1.906566357160621e-05, + "loss": 0.3351, + "step": 3144 + }, + { + "epoch": 1.32, + "learning_rate": 1.9064947871429102e-05, + "loss": 0.3869, + "step": 3145 + }, + { + "epoch": 1.32, + "learning_rate": 1.9064231910688108e-05, + "loss": 0.4992, + "step": 3146 + }, + { + "epoch": 1.32, + "learning_rate": 1.9063515689403815e-05, + "loss": 0.3779, + "step": 3147 + }, + { + "epoch": 1.32, + "learning_rate": 1.90627992075968e-05, + "loss": 0.3634, + "step": 3148 + }, + { + "epoch": 1.32, + "learning_rate": 1.9062082465287672e-05, + "loss": 0.38, + "step": 3149 + }, + { + "epoch": 1.32, + "learning_rate": 1.9061365462497022e-05, + "loss": 0.3647, + "step": 3150 + }, + { + "epoch": 1.32, + "learning_rate": 1.9060648199245463e-05, + "loss": 0.4137, + "step": 3151 + }, + { + "epoch": 1.32, + "learning_rate": 1.905993067555361e-05, + "loss": 0.392, + "step": 3152 + }, + { + "epoch": 1.32, + "learning_rate": 1.905921289144209e-05, + "loss": 0.3798, + "step": 3153 + }, + { + "epoch": 1.32, + "learning_rate": 1.9058494846931538e-05, + "loss": 0.3981, + "step": 3154 + }, + { + "epoch": 1.32, + "learning_rate": 1.9057776542042587e-05, + "loss": 0.4565, + "step": 3155 + }, + { + "epoch": 1.32, + "learning_rate": 1.9057057976795884e-05, + "loss": 0.3855, + "step": 3156 + }, + { + "epoch": 1.32, + "learning_rate": 1.905633915121209e-05, + "loss": 0.3821, + "step": 3157 + }, + { + "epoch": 1.32, + "learning_rate": 1.905562006531186e-05, + "loss": 0.3874, + "step": 3158 + }, + { + "epoch": 1.32, + "learning_rate": 1.9054900719115874e-05, + "loss": 0.3713, + "step": 3159 + }, + { + "epoch": 1.32, + "learning_rate": 1.9054181112644797e-05, + "loss": 0.3942, + "step": 3160 + }, + { + "epoch": 1.32, + "learning_rate": 1.905346124591932e-05, + "loss": 0.4181, + "step": 3161 + }, + { + "epoch": 1.32, + "learning_rate": 1.9052741118960133e-05, + "loss": 0.405, + "step": 3162 + }, + { + "epoch": 1.32, + "learning_rate": 1.9052020731787936e-05, + "loss": 0.3995, + "step": 3163 + }, + { + "epoch": 1.32, + "learning_rate": 1.9051300084423433e-05, + "loss": 0.4052, + "step": 3164 + }, + { + "epoch": 1.33, + "learning_rate": 1.9050579176887343e-05, + "loss": 0.4243, + "step": 3165 + }, + { + "epoch": 1.33, + "learning_rate": 1.9049858009200388e-05, + "loss": 0.373, + "step": 3166 + }, + { + "epoch": 1.33, + "learning_rate": 1.9049136581383295e-05, + "loss": 0.4055, + "step": 3167 + }, + { + "epoch": 1.33, + "learning_rate": 1.90484148934568e-05, + "loss": 0.3674, + "step": 3168 + }, + { + "epoch": 1.33, + "learning_rate": 1.9047692945441647e-05, + "loss": 0.4224, + "step": 3169 + }, + { + "epoch": 1.33, + "learning_rate": 1.904697073735859e-05, + "loss": 0.4768, + "step": 3170 + }, + { + "epoch": 1.33, + "learning_rate": 1.9046248269228388e-05, + "loss": 0.3627, + "step": 3171 + }, + { + "epoch": 1.33, + "learning_rate": 1.9045525541071807e-05, + "loss": 0.3667, + "step": 3172 + }, + { + "epoch": 1.33, + "learning_rate": 1.904480255290962e-05, + "loss": 0.3859, + "step": 3173 + }, + { + "epoch": 1.33, + "learning_rate": 1.904407930476261e-05, + "loss": 0.3877, + "step": 3174 + }, + { + "epoch": 1.33, + "learning_rate": 1.9043355796651565e-05, + "loss": 0.4244, + "step": 3175 + }, + { + "epoch": 1.33, + "learning_rate": 1.9042632028597286e-05, + "loss": 0.4116, + "step": 3176 + }, + { + "epoch": 1.33, + "learning_rate": 1.904190800062057e-05, + "loss": 0.4461, + "step": 3177 + }, + { + "epoch": 1.33, + "learning_rate": 1.9041183712742233e-05, + "loss": 0.3593, + "step": 3178 + }, + { + "epoch": 1.33, + "learning_rate": 1.9040459164983095e-05, + "loss": 0.3896, + "step": 3179 + }, + { + "epoch": 1.33, + "learning_rate": 1.903973435736398e-05, + "loss": 0.4172, + "step": 3180 + }, + { + "epoch": 1.33, + "learning_rate": 1.903900928990572e-05, + "loss": 0.3866, + "step": 3181 + }, + { + "epoch": 1.33, + "learning_rate": 1.9038283962629164e-05, + "loss": 0.3818, + "step": 3182 + }, + { + "epoch": 1.33, + "learning_rate": 1.903755837555515e-05, + "loss": 0.3322, + "step": 3183 + }, + { + "epoch": 1.33, + "learning_rate": 1.9036832528704547e-05, + "loss": 0.4535, + "step": 3184 + }, + { + "epoch": 1.33, + "learning_rate": 1.9036106422098207e-05, + "loss": 0.4, + "step": 3185 + }, + { + "epoch": 1.33, + "learning_rate": 1.9035380055757007e-05, + "loss": 0.3732, + "step": 3186 + }, + { + "epoch": 1.33, + "learning_rate": 1.9034653429701824e-05, + "loss": 0.3552, + "step": 3187 + }, + { + "epoch": 1.33, + "learning_rate": 1.9033926543953546e-05, + "loss": 0.4101, + "step": 3188 + }, + { + "epoch": 1.34, + "learning_rate": 1.903319939853307e-05, + "loss": 0.409, + "step": 3189 + }, + { + "epoch": 1.34, + "learning_rate": 1.903247199346129e-05, + "loss": 0.3997, + "step": 3190 + }, + { + "epoch": 1.34, + "learning_rate": 1.9031744328759118e-05, + "loss": 0.4128, + "step": 3191 + }, + { + "epoch": 1.34, + "learning_rate": 1.903101640444747e-05, + "loss": 0.4079, + "step": 3192 + }, + { + "epoch": 1.34, + "learning_rate": 1.903028822054727e-05, + "loss": 0.3743, + "step": 3193 + }, + { + "epoch": 1.34, + "learning_rate": 1.902955977707945e-05, + "loss": 0.353, + "step": 3194 + }, + { + "epoch": 1.34, + "learning_rate": 1.9028831074064945e-05, + "loss": 0.3681, + "step": 3195 + }, + { + "epoch": 1.34, + "learning_rate": 1.9028102111524702e-05, + "loss": 0.4156, + "step": 3196 + }, + { + "epoch": 1.34, + "learning_rate": 1.9027372889479676e-05, + "loss": 0.4007, + "step": 3197 + }, + { + "epoch": 1.34, + "learning_rate": 1.9026643407950826e-05, + "loss": 0.3695, + "step": 3198 + }, + { + "epoch": 1.34, + "learning_rate": 1.9025913666959122e-05, + "loss": 0.3651, + "step": 3199 + }, + { + "epoch": 1.34, + "learning_rate": 1.9025183666525543e-05, + "loss": 0.3975, + "step": 3200 + }, + { + "epoch": 1.34, + "learning_rate": 1.9024453406671064e-05, + "loss": 0.4004, + "step": 3201 + }, + { + "epoch": 1.34, + "learning_rate": 1.902372288741668e-05, + "loss": 0.3727, + "step": 3202 + }, + { + "epoch": 1.34, + "learning_rate": 1.902299210878339e-05, + "loss": 0.4255, + "step": 3203 + }, + { + "epoch": 1.34, + "learning_rate": 1.9022261070792197e-05, + "loss": 0.3589, + "step": 3204 + }, + { + "epoch": 1.34, + "learning_rate": 1.902152977346412e-05, + "loss": 0.3909, + "step": 3205 + }, + { + "epoch": 1.34, + "learning_rate": 1.9020798216820174e-05, + "loss": 0.3787, + "step": 3206 + }, + { + "epoch": 1.34, + "learning_rate": 1.9020066400881386e-05, + "loss": 0.3855, + "step": 3207 + }, + { + "epoch": 1.34, + "learning_rate": 1.9019334325668795e-05, + "loss": 0.3937, + "step": 3208 + }, + { + "epoch": 1.34, + "learning_rate": 1.901860199120344e-05, + "loss": 0.448, + "step": 3209 + }, + { + "epoch": 1.34, + "learning_rate": 1.9017869397506374e-05, + "loss": 0.4216, + "step": 3210 + }, + { + "epoch": 1.34, + "learning_rate": 1.9017136544598657e-05, + "loss": 0.3929, + "step": 3211 + }, + { + "epoch": 1.34, + "learning_rate": 1.901640343250135e-05, + "loss": 0.4304, + "step": 3212 + }, + { + "epoch": 1.35, + "learning_rate": 1.9015670061235528e-05, + "loss": 0.3553, + "step": 3213 + }, + { + "epoch": 1.35, + "learning_rate": 1.901493643082227e-05, + "loss": 0.3978, + "step": 3214 + }, + { + "epoch": 1.35, + "learning_rate": 1.9014202541282663e-05, + "loss": 0.425, + "step": 3215 + }, + { + "epoch": 1.35, + "learning_rate": 1.901346839263781e-05, + "loss": 0.4167, + "step": 3216 + }, + { + "epoch": 1.35, + "learning_rate": 1.9012733984908798e-05, + "loss": 0.3343, + "step": 3217 + }, + { + "epoch": 1.35, + "learning_rate": 1.901199931811675e-05, + "loss": 0.3673, + "step": 3218 + }, + { + "epoch": 1.35, + "learning_rate": 1.9011264392282774e-05, + "loss": 0.4347, + "step": 3219 + }, + { + "epoch": 1.35, + "learning_rate": 1.9010529207428004e-05, + "loss": 0.337, + "step": 3220 + }, + { + "epoch": 1.35, + "learning_rate": 1.9009793763573564e-05, + "loss": 0.4128, + "step": 3221 + }, + { + "epoch": 1.35, + "learning_rate": 1.9009058060740603e-05, + "loss": 0.3688, + "step": 3222 + }, + { + "epoch": 1.35, + "learning_rate": 1.900832209895026e-05, + "loss": 0.4163, + "step": 3223 + }, + { + "epoch": 1.35, + "learning_rate": 1.900758587822369e-05, + "loss": 0.417, + "step": 3224 + }, + { + "epoch": 1.35, + "learning_rate": 1.900684939858206e-05, + "loss": 0.3658, + "step": 3225 + }, + { + "epoch": 1.35, + "learning_rate": 1.9006112660046535e-05, + "loss": 0.3795, + "step": 3226 + }, + { + "epoch": 1.35, + "learning_rate": 1.9005375662638294e-05, + "loss": 0.3625, + "step": 3227 + }, + { + "epoch": 1.35, + "learning_rate": 1.900463840637852e-05, + "loss": 0.3675, + "step": 3228 + }, + { + "epoch": 1.35, + "learning_rate": 1.9003900891288406e-05, + "loss": 0.3909, + "step": 3229 + }, + { + "epoch": 1.35, + "learning_rate": 1.900316311738915e-05, + "loss": 0.4061, + "step": 3230 + }, + { + "epoch": 1.35, + "learning_rate": 1.9002425084701958e-05, + "loss": 0.3623, + "step": 3231 + }, + { + "epoch": 1.35, + "learning_rate": 1.900168679324805e-05, + "loss": 0.4009, + "step": 3232 + }, + { + "epoch": 1.35, + "learning_rate": 1.900094824304864e-05, + "loss": 0.373, + "step": 3233 + }, + { + "epoch": 1.35, + "learning_rate": 1.900020943412496e-05, + "loss": 0.3667, + "step": 3234 + }, + { + "epoch": 1.35, + "learning_rate": 1.899947036649825e-05, + "loss": 0.3848, + "step": 3235 + }, + { + "epoch": 1.35, + "learning_rate": 1.8998731040189748e-05, + "loss": 0.4338, + "step": 3236 + }, + { + "epoch": 1.36, + "learning_rate": 1.8997991455220702e-05, + "loss": 0.4409, + "step": 3237 + }, + { + "epoch": 1.36, + "learning_rate": 1.8997251611612385e-05, + "loss": 0.4362, + "step": 3238 + }, + { + "epoch": 1.36, + "learning_rate": 1.899651150938605e-05, + "loss": 0.3413, + "step": 3239 + }, + { + "epoch": 1.36, + "learning_rate": 1.8995771148562978e-05, + "loss": 0.4013, + "step": 3240 + }, + { + "epoch": 1.36, + "learning_rate": 1.8995030529164446e-05, + "loss": 0.3302, + "step": 3241 + }, + { + "epoch": 1.36, + "learning_rate": 1.8994289651211742e-05, + "loss": 0.3261, + "step": 3242 + }, + { + "epoch": 1.36, + "learning_rate": 1.8993548514726162e-05, + "loss": 0.4185, + "step": 3243 + }, + { + "epoch": 1.36, + "learning_rate": 1.8992807119729013e-05, + "loss": 0.4119, + "step": 3244 + }, + { + "epoch": 1.36, + "learning_rate": 1.8992065466241603e-05, + "loss": 0.3402, + "step": 3245 + }, + { + "epoch": 1.36, + "learning_rate": 1.899132355428525e-05, + "loss": 0.3759, + "step": 3246 + }, + { + "epoch": 1.36, + "learning_rate": 1.899058138388128e-05, + "loss": 0.4098, + "step": 3247 + }, + { + "epoch": 1.36, + "learning_rate": 1.8989838955051027e-05, + "loss": 0.4127, + "step": 3248 + }, + { + "epoch": 1.36, + "learning_rate": 1.898909626781583e-05, + "loss": 0.3761, + "step": 3249 + }, + { + "epoch": 1.36, + "learning_rate": 1.898835332219704e-05, + "loss": 0.3541, + "step": 3250 + }, + { + "epoch": 1.36, + "learning_rate": 1.8987610118216006e-05, + "loss": 0.4133, + "step": 3251 + }, + { + "epoch": 1.36, + "learning_rate": 1.8986866655894098e-05, + "loss": 0.4228, + "step": 3252 + }, + { + "epoch": 1.36, + "learning_rate": 1.8986122935252684e-05, + "loss": 0.4267, + "step": 3253 + }, + { + "epoch": 1.36, + "learning_rate": 1.8985378956313133e-05, + "loss": 0.3861, + "step": 3254 + }, + { + "epoch": 1.36, + "learning_rate": 1.8984634719096844e-05, + "loss": 0.3509, + "step": 3255 + }, + { + "epoch": 1.36, + "learning_rate": 1.8983890223625205e-05, + "loss": 0.3746, + "step": 3256 + }, + { + "epoch": 1.36, + "learning_rate": 1.8983145469919606e-05, + "loss": 0.4175, + "step": 3257 + }, + { + "epoch": 1.36, + "learning_rate": 1.8982400458001465e-05, + "loss": 0.4045, + "step": 3258 + }, + { + "epoch": 1.36, + "learning_rate": 1.8981655187892194e-05, + "loss": 0.3661, + "step": 3259 + }, + { + "epoch": 1.36, + "learning_rate": 1.8980909659613217e-05, + "loss": 0.4329, + "step": 3260 + }, + { + "epoch": 1.37, + "learning_rate": 1.898016387318596e-05, + "loss": 0.3702, + "step": 3261 + }, + { + "epoch": 1.37, + "learning_rate": 1.8979417828631858e-05, + "loss": 0.4432, + "step": 3262 + }, + { + "epoch": 1.37, + "learning_rate": 1.8978671525972365e-05, + "loss": 0.3797, + "step": 3263 + }, + { + "epoch": 1.37, + "learning_rate": 1.8977924965228923e-05, + "loss": 0.3628, + "step": 3264 + }, + { + "epoch": 1.37, + "learning_rate": 1.8977178146422992e-05, + "loss": 0.3518, + "step": 3265 + }, + { + "epoch": 1.37, + "learning_rate": 1.8976431069576047e-05, + "loss": 0.3573, + "step": 3266 + }, + { + "epoch": 1.37, + "learning_rate": 1.8975683734709554e-05, + "loss": 0.3757, + "step": 3267 + }, + { + "epoch": 1.37, + "learning_rate": 1.8974936141844997e-05, + "loss": 0.37, + "step": 3268 + }, + { + "epoch": 1.37, + "learning_rate": 1.8974188291003864e-05, + "loss": 0.3833, + "step": 3269 + }, + { + "epoch": 1.37, + "learning_rate": 1.897344018220765e-05, + "loss": 0.3984, + "step": 3270 + }, + { + "epoch": 1.37, + "learning_rate": 1.8972691815477864e-05, + "loss": 0.4031, + "step": 3271 + }, + { + "epoch": 1.37, + "learning_rate": 1.8971943190836013e-05, + "loss": 0.3859, + "step": 3272 + }, + { + "epoch": 1.37, + "learning_rate": 1.8971194308303617e-05, + "loss": 0.3791, + "step": 3273 + }, + { + "epoch": 1.37, + "learning_rate": 1.89704451679022e-05, + "loss": 0.4298, + "step": 3274 + }, + { + "epoch": 1.37, + "learning_rate": 1.89696957696533e-05, + "loss": 0.3776, + "step": 3275 + }, + { + "epoch": 1.37, + "learning_rate": 1.8968946113578455e-05, + "loss": 0.3676, + "step": 3276 + }, + { + "epoch": 1.37, + "learning_rate": 1.896819619969921e-05, + "loss": 0.393, + "step": 3277 + }, + { + "epoch": 1.37, + "learning_rate": 1.8967446028037126e-05, + "loss": 0.372, + "step": 3278 + }, + { + "epoch": 1.37, + "learning_rate": 1.8966695598613758e-05, + "loss": 0.3877, + "step": 3279 + }, + { + "epoch": 1.37, + "learning_rate": 1.8965944911450685e-05, + "loss": 0.3731, + "step": 3280 + }, + { + "epoch": 1.37, + "learning_rate": 1.8965193966569483e-05, + "loss": 0.417, + "step": 3281 + }, + { + "epoch": 1.37, + "learning_rate": 1.8964442763991737e-05, + "loss": 0.4348, + "step": 3282 + }, + { + "epoch": 1.37, + "learning_rate": 1.8963691303739037e-05, + "loss": 0.4124, + "step": 3283 + }, + { + "epoch": 1.37, + "learning_rate": 1.8962939585832985e-05, + "loss": 0.3836, + "step": 3284 + }, + { + "epoch": 1.38, + "learning_rate": 1.8962187610295188e-05, + "loss": 0.4002, + "step": 3285 + }, + { + "epoch": 1.38, + "learning_rate": 1.896143537714726e-05, + "loss": 0.3544, + "step": 3286 + }, + { + "epoch": 1.38, + "learning_rate": 1.8960682886410825e-05, + "loss": 0.3982, + "step": 3287 + }, + { + "epoch": 1.38, + "learning_rate": 1.8959930138107512e-05, + "loss": 0.3865, + "step": 3288 + }, + { + "epoch": 1.38, + "learning_rate": 1.895917713225896e-05, + "loss": 0.3951, + "step": 3289 + }, + { + "epoch": 1.38, + "learning_rate": 1.895842386888681e-05, + "loss": 0.3978, + "step": 3290 + }, + { + "epoch": 1.38, + "learning_rate": 1.8957670348012717e-05, + "loss": 0.3807, + "step": 3291 + }, + { + "epoch": 1.38, + "learning_rate": 1.8956916569658336e-05, + "loss": 0.396, + "step": 3292 + }, + { + "epoch": 1.38, + "learning_rate": 1.895616253384534e-05, + "loss": 0.3715, + "step": 3293 + }, + { + "epoch": 1.38, + "learning_rate": 1.8955408240595396e-05, + "loss": 0.4017, + "step": 3294 + }, + { + "epoch": 1.38, + "learning_rate": 1.895465368993019e-05, + "loss": 0.4133, + "step": 3295 + }, + { + "epoch": 1.38, + "learning_rate": 1.8953898881871412e-05, + "loss": 0.3663, + "step": 3296 + }, + { + "epoch": 1.38, + "learning_rate": 1.895314381644075e-05, + "loss": 0.4083, + "step": 3297 + }, + { + "epoch": 1.38, + "learning_rate": 1.8952388493659918e-05, + "loss": 0.3797, + "step": 3298 + }, + { + "epoch": 1.38, + "learning_rate": 1.8951632913550625e-05, + "loss": 0.3964, + "step": 3299 + }, + { + "epoch": 1.38, + "learning_rate": 1.8950877076134583e-05, + "loss": 0.4279, + "step": 3300 + }, + { + "epoch": 1.38, + "learning_rate": 1.8950120981433527e-05, + "loss": 0.3264, + "step": 3301 + }, + { + "epoch": 1.38, + "learning_rate": 1.8949364629469184e-05, + "loss": 0.3627, + "step": 3302 + }, + { + "epoch": 1.38, + "learning_rate": 1.8948608020263293e-05, + "loss": 0.3707, + "step": 3303 + }, + { + "epoch": 1.38, + "learning_rate": 1.894785115383761e-05, + "loss": 0.3831, + "step": 3304 + }, + { + "epoch": 1.38, + "learning_rate": 1.8947094030213883e-05, + "loss": 0.3846, + "step": 3305 + }, + { + "epoch": 1.38, + "learning_rate": 1.8946336649413883e-05, + "loss": 0.3935, + "step": 3306 + }, + { + "epoch": 1.38, + "learning_rate": 1.8945579011459366e-05, + "loss": 0.4725, + "step": 3307 + }, + { + "epoch": 1.38, + "learning_rate": 1.8944821116372125e-05, + "loss": 0.401, + "step": 3308 + }, + { + "epoch": 1.39, + "learning_rate": 1.8944062964173934e-05, + "loss": 0.446, + "step": 3309 + }, + { + "epoch": 1.39, + "learning_rate": 1.8943304554886594e-05, + "loss": 0.3866, + "step": 3310 + }, + { + "epoch": 1.39, + "learning_rate": 1.8942545888531897e-05, + "loss": 0.4263, + "step": 3311 + }, + { + "epoch": 1.39, + "learning_rate": 1.8941786965131658e-05, + "loss": 0.4343, + "step": 3312 + }, + { + "epoch": 1.39, + "learning_rate": 1.8941027784707685e-05, + "loss": 0.3569, + "step": 3313 + }, + { + "epoch": 1.39, + "learning_rate": 1.8940268347281802e-05, + "loss": 0.3838, + "step": 3314 + }, + { + "epoch": 1.39, + "learning_rate": 1.8939508652875837e-05, + "loss": 0.3983, + "step": 3315 + }, + { + "epoch": 1.39, + "learning_rate": 1.8938748701511633e-05, + "loss": 0.3926, + "step": 3316 + }, + { + "epoch": 1.39, + "learning_rate": 1.8937988493211027e-05, + "loss": 0.3967, + "step": 3317 + }, + { + "epoch": 1.39, + "learning_rate": 1.8937228027995874e-05, + "loss": 0.3469, + "step": 3318 + }, + { + "epoch": 1.39, + "learning_rate": 1.8936467305888033e-05, + "loss": 0.3752, + "step": 3319 + }, + { + "epoch": 1.39, + "learning_rate": 1.893570632690937e-05, + "loss": 0.3537, + "step": 3320 + }, + { + "epoch": 1.39, + "learning_rate": 1.8934945091081755e-05, + "loss": 0.3916, + "step": 3321 + }, + { + "epoch": 1.39, + "learning_rate": 1.8934183598427072e-05, + "loss": 0.3867, + "step": 3322 + }, + { + "epoch": 1.39, + "learning_rate": 1.8933421848967213e-05, + "loss": 0.3915, + "step": 3323 + }, + { + "epoch": 1.39, + "learning_rate": 1.8932659842724067e-05, + "loss": 0.3388, + "step": 3324 + }, + { + "epoch": 1.39, + "learning_rate": 1.8931897579719543e-05, + "loss": 0.4215, + "step": 3325 + }, + { + "epoch": 1.39, + "learning_rate": 1.8931135059975548e-05, + "loss": 0.519, + "step": 3326 + }, + { + "epoch": 1.39, + "learning_rate": 1.8930372283514004e-05, + "loss": 0.3756, + "step": 3327 + }, + { + "epoch": 1.39, + "learning_rate": 1.8929609250356827e-05, + "loss": 0.354, + "step": 3328 + }, + { + "epoch": 1.39, + "learning_rate": 1.8928845960525957e-05, + "loss": 0.3802, + "step": 3329 + }, + { + "epoch": 1.39, + "learning_rate": 1.8928082414043342e-05, + "loss": 0.3898, + "step": 3330 + }, + { + "epoch": 1.39, + "learning_rate": 1.8927318610930912e-05, + "loss": 0.414, + "step": 3331 + }, + { + "epoch": 1.4, + "learning_rate": 1.892655455121063e-05, + "loss": 0.4356, + "step": 3332 + }, + { + "epoch": 1.4, + "learning_rate": 1.8925790234904463e-05, + "loss": 0.3701, + "step": 3333 + }, + { + "epoch": 1.4, + "learning_rate": 1.8925025662034376e-05, + "loss": 0.4053, + "step": 3334 + }, + { + "epoch": 1.4, + "learning_rate": 1.8924260832622347e-05, + "loss": 0.3682, + "step": 3335 + }, + { + "epoch": 1.4, + "learning_rate": 1.8923495746690354e-05, + "loss": 0.3782, + "step": 3336 + }, + { + "epoch": 1.4, + "learning_rate": 1.89227304042604e-05, + "loss": 0.4035, + "step": 3337 + }, + { + "epoch": 1.4, + "learning_rate": 1.8921964805354475e-05, + "loss": 0.3436, + "step": 3338 + }, + { + "epoch": 1.4, + "learning_rate": 1.892119894999459e-05, + "loss": 0.3769, + "step": 3339 + }, + { + "epoch": 1.4, + "learning_rate": 1.892043283820276e-05, + "loss": 0.4044, + "step": 3340 + }, + { + "epoch": 1.4, + "learning_rate": 1.8919666470001003e-05, + "loss": 0.4397, + "step": 3341 + }, + { + "epoch": 1.4, + "learning_rate": 1.8918899845411343e-05, + "loss": 0.3762, + "step": 3342 + }, + { + "epoch": 1.4, + "learning_rate": 1.891813296445583e-05, + "loss": 0.3667, + "step": 3343 + }, + { + "epoch": 1.4, + "learning_rate": 1.8917365827156493e-05, + "loss": 0.4077, + "step": 3344 + }, + { + "epoch": 1.4, + "learning_rate": 1.891659843353539e-05, + "loss": 0.4158, + "step": 3345 + }, + { + "epoch": 1.4, + "learning_rate": 1.8915830783614576e-05, + "loss": 0.3733, + "step": 3346 + }, + { + "epoch": 1.4, + "learning_rate": 1.891506287741612e-05, + "loss": 0.3935, + "step": 3347 + }, + { + "epoch": 1.4, + "learning_rate": 1.8914294714962093e-05, + "loss": 0.3812, + "step": 3348 + }, + { + "epoch": 1.4, + "learning_rate": 1.8913526296274573e-05, + "loss": 0.3571, + "step": 3349 + }, + { + "epoch": 1.4, + "learning_rate": 1.8912757621375654e-05, + "loss": 0.3853, + "step": 3350 + }, + { + "epoch": 1.4, + "learning_rate": 1.8911988690287422e-05, + "loss": 0.412, + "step": 3351 + }, + { + "epoch": 1.4, + "learning_rate": 1.891121950303198e-05, + "loss": 0.4038, + "step": 3352 + }, + { + "epoch": 1.4, + "learning_rate": 1.891045005963145e-05, + "loss": 0.3953, + "step": 3353 + }, + { + "epoch": 1.4, + "learning_rate": 1.890968036010793e-05, + "loss": 0.4022, + "step": 3354 + }, + { + "epoch": 1.4, + "learning_rate": 1.8908910404483564e-05, + "loss": 0.3732, + "step": 3355 + }, + { + "epoch": 1.41, + "learning_rate": 1.8908140192780466e-05, + "loss": 0.3713, + "step": 3356 + }, + { + "epoch": 1.41, + "learning_rate": 1.890736972502079e-05, + "loss": 0.4261, + "step": 3357 + }, + { + "epoch": 1.41, + "learning_rate": 1.890659900122667e-05, + "loss": 0.3903, + "step": 3358 + }, + { + "epoch": 1.41, + "learning_rate": 1.890582802142027e-05, + "loss": 0.4046, + "step": 3359 + }, + { + "epoch": 1.41, + "learning_rate": 1.8905056785623742e-05, + "loss": 0.3788, + "step": 3360 + }, + { + "epoch": 1.41, + "learning_rate": 1.890428529385926e-05, + "loss": 0.3524, + "step": 3361 + }, + { + "epoch": 1.41, + "learning_rate": 1.8903513546149e-05, + "loss": 0.3807, + "step": 3362 + }, + { + "epoch": 1.41, + "learning_rate": 1.8902741542515143e-05, + "loss": 0.33, + "step": 3363 + }, + { + "epoch": 1.41, + "learning_rate": 1.8901969282979882e-05, + "loss": 0.3963, + "step": 3364 + }, + { + "epoch": 1.41, + "learning_rate": 1.890119676756541e-05, + "loss": 0.3798, + "step": 3365 + }, + { + "epoch": 1.41, + "learning_rate": 1.8900423996293938e-05, + "loss": 0.4029, + "step": 3366 + }, + { + "epoch": 1.41, + "learning_rate": 1.889965096918768e-05, + "loss": 0.3793, + "step": 3367 + }, + { + "epoch": 1.41, + "learning_rate": 1.889887768626885e-05, + "loss": 0.3876, + "step": 3368 + }, + { + "epoch": 1.41, + "learning_rate": 1.8898104147559674e-05, + "loss": 0.3914, + "step": 3369 + }, + { + "epoch": 1.41, + "learning_rate": 1.8897330353082392e-05, + "loss": 0.3742, + "step": 3370 + }, + { + "epoch": 1.41, + "learning_rate": 1.8896556302859243e-05, + "loss": 0.4183, + "step": 3371 + }, + { + "epoch": 1.41, + "learning_rate": 1.889578199691248e-05, + "loss": 0.4014, + "step": 3372 + }, + { + "epoch": 1.41, + "learning_rate": 1.889500743526436e-05, + "loss": 0.3522, + "step": 3373 + }, + { + "epoch": 1.41, + "learning_rate": 1.889423261793714e-05, + "loss": 0.4073, + "step": 3374 + }, + { + "epoch": 1.41, + "learning_rate": 1.88934575449531e-05, + "loss": 0.3872, + "step": 3375 + }, + { + "epoch": 1.41, + "learning_rate": 1.889268221633451e-05, + "loss": 0.3361, + "step": 3376 + }, + { + "epoch": 1.41, + "learning_rate": 1.8891906632103664e-05, + "loss": 0.4251, + "step": 3377 + }, + { + "epoch": 1.41, + "learning_rate": 1.889113079228285e-05, + "loss": 0.3913, + "step": 3378 + }, + { + "epoch": 1.41, + "learning_rate": 1.8890354696894374e-05, + "loss": 0.3846, + "step": 3379 + }, + { + "epoch": 1.42, + "learning_rate": 1.888957834596054e-05, + "loss": 0.3444, + "step": 3380 + }, + { + "epoch": 1.42, + "learning_rate": 1.8888801739503667e-05, + "loss": 0.4703, + "step": 3381 + }, + { + "epoch": 1.42, + "learning_rate": 1.8888024877546074e-05, + "loss": 0.3833, + "step": 3382 + }, + { + "epoch": 1.42, + "learning_rate": 1.8887247760110094e-05, + "loss": 0.3652, + "step": 3383 + }, + { + "epoch": 1.42, + "learning_rate": 1.888647038721806e-05, + "loss": 0.4107, + "step": 3384 + }, + { + "epoch": 1.42, + "learning_rate": 1.8885692758892324e-05, + "loss": 0.3597, + "step": 3385 + }, + { + "epoch": 1.42, + "learning_rate": 1.8884914875155235e-05, + "loss": 0.3862, + "step": 3386 + }, + { + "epoch": 1.42, + "learning_rate": 1.8884136736029154e-05, + "loss": 0.4022, + "step": 3387 + }, + { + "epoch": 1.42, + "learning_rate": 1.8883358341536442e-05, + "loss": 0.4315, + "step": 3388 + }, + { + "epoch": 1.42, + "learning_rate": 1.8882579691699475e-05, + "loss": 0.4305, + "step": 3389 + }, + { + "epoch": 1.42, + "learning_rate": 1.8881800786540642e-05, + "loss": 0.3692, + "step": 3390 + }, + { + "epoch": 1.42, + "learning_rate": 1.8881021626082328e-05, + "loss": 0.3415, + "step": 3391 + }, + { + "epoch": 1.42, + "learning_rate": 1.8880242210346925e-05, + "loss": 0.4347, + "step": 3392 + }, + { + "epoch": 1.42, + "learning_rate": 1.8879462539356842e-05, + "loss": 0.3729, + "step": 3393 + }, + { + "epoch": 1.42, + "learning_rate": 1.8878682613134486e-05, + "loss": 0.3826, + "step": 3394 + }, + { + "epoch": 1.42, + "learning_rate": 1.8877902431702274e-05, + "loss": 0.4082, + "step": 3395 + }, + { + "epoch": 1.42, + "learning_rate": 1.8877121995082638e-05, + "loss": 0.3841, + "step": 3396 + }, + { + "epoch": 1.42, + "learning_rate": 1.8876341303298012e-05, + "loss": 0.3887, + "step": 3397 + }, + { + "epoch": 1.42, + "learning_rate": 1.8875560356370827e-05, + "loss": 0.3646, + "step": 3398 + }, + { + "epoch": 1.42, + "learning_rate": 1.8874779154323538e-05, + "loss": 0.3887, + "step": 3399 + }, + { + "epoch": 1.42, + "learning_rate": 1.8873997697178593e-05, + "loss": 0.3638, + "step": 3400 + }, + { + "epoch": 1.42, + "learning_rate": 1.8873215984958465e-05, + "loss": 0.4295, + "step": 3401 + }, + { + "epoch": 1.42, + "learning_rate": 1.8872434017685613e-05, + "loss": 0.3792, + "step": 3402 + }, + { + "epoch": 1.42, + "learning_rate": 1.8871651795382518e-05, + "loss": 0.3996, + "step": 3403 + }, + { + "epoch": 1.43, + "learning_rate": 1.8870869318071667e-05, + "loss": 0.4269, + "step": 3404 + }, + { + "epoch": 1.43, + "learning_rate": 1.8870086585775548e-05, + "loss": 0.3915, + "step": 3405 + }, + { + "epoch": 1.43, + "learning_rate": 1.8869303598516663e-05, + "loss": 0.3593, + "step": 3406 + }, + { + "epoch": 1.43, + "learning_rate": 1.8868520356317517e-05, + "loss": 0.3277, + "step": 3407 + }, + { + "epoch": 1.43, + "learning_rate": 1.886773685920062e-05, + "loss": 0.3957, + "step": 3408 + }, + { + "epoch": 1.43, + "learning_rate": 1.8866953107188498e-05, + "loss": 0.4089, + "step": 3409 + }, + { + "epoch": 1.43, + "learning_rate": 1.8866169100303677e-05, + "loss": 0.368, + "step": 3410 + }, + { + "epoch": 1.43, + "learning_rate": 1.8865384838568693e-05, + "loss": 0.3971, + "step": 3411 + }, + { + "epoch": 1.43, + "learning_rate": 1.886460032200609e-05, + "loss": 0.4052, + "step": 3412 + }, + { + "epoch": 1.43, + "learning_rate": 1.8863815550638414e-05, + "loss": 0.3769, + "step": 3413 + }, + { + "epoch": 1.43, + "learning_rate": 1.886303052448823e-05, + "loss": 0.3886, + "step": 3414 + }, + { + "epoch": 1.43, + "learning_rate": 1.8862245243578097e-05, + "loss": 0.4253, + "step": 3415 + }, + { + "epoch": 1.43, + "learning_rate": 1.8861459707930585e-05, + "loss": 0.3974, + "step": 3416 + }, + { + "epoch": 1.43, + "learning_rate": 1.886067391756828e-05, + "loss": 0.4251, + "step": 3417 + }, + { + "epoch": 1.43, + "learning_rate": 1.885988787251377e-05, + "loss": 0.3961, + "step": 3418 + }, + { + "epoch": 1.43, + "learning_rate": 1.885910157278964e-05, + "loss": 0.3985, + "step": 3419 + }, + { + "epoch": 1.43, + "learning_rate": 1.8858315018418497e-05, + "loss": 0.4277, + "step": 3420 + }, + { + "epoch": 1.43, + "learning_rate": 1.885752820942295e-05, + "loss": 0.4149, + "step": 3421 + }, + { + "epoch": 1.43, + "learning_rate": 1.8856741145825616e-05, + "loss": 0.3769, + "step": 3422 + }, + { + "epoch": 1.43, + "learning_rate": 1.8855953827649115e-05, + "loss": 0.3979, + "step": 3423 + }, + { + "epoch": 1.43, + "learning_rate": 1.8855166254916082e-05, + "loss": 0.398, + "step": 3424 + }, + { + "epoch": 1.43, + "learning_rate": 1.885437842764915e-05, + "loss": 0.3734, + "step": 3425 + }, + { + "epoch": 1.43, + "learning_rate": 1.885359034587097e-05, + "loss": 0.3775, + "step": 3426 + }, + { + "epoch": 1.43, + "learning_rate": 1.885280200960419e-05, + "loss": 0.3443, + "step": 3427 + }, + { + "epoch": 1.44, + "learning_rate": 1.8852013418871474e-05, + "loss": 0.4162, + "step": 3428 + }, + { + "epoch": 1.44, + "learning_rate": 1.885122457369549e-05, + "loss": 0.3529, + "step": 3429 + }, + { + "epoch": 1.44, + "learning_rate": 1.8850435474098903e-05, + "loss": 0.4343, + "step": 3430 + }, + { + "epoch": 1.44, + "learning_rate": 1.884964612010441e-05, + "loss": 0.3688, + "step": 3431 + }, + { + "epoch": 1.44, + "learning_rate": 1.884885651173469e-05, + "loss": 0.4096, + "step": 3432 + }, + { + "epoch": 1.44, + "learning_rate": 1.8848066649012437e-05, + "loss": 0.3725, + "step": 3433 + }, + { + "epoch": 1.44, + "learning_rate": 1.8847276531960363e-05, + "loss": 0.3621, + "step": 3434 + }, + { + "epoch": 1.44, + "learning_rate": 1.8846486160601175e-05, + "loss": 0.3706, + "step": 3435 + }, + { + "epoch": 1.44, + "learning_rate": 1.8845695534957594e-05, + "loss": 0.4113, + "step": 3436 + }, + { + "epoch": 1.44, + "learning_rate": 1.8844904655052344e-05, + "loss": 0.3906, + "step": 3437 + }, + { + "epoch": 1.44, + "learning_rate": 1.884411352090816e-05, + "loss": 0.3552, + "step": 3438 + }, + { + "epoch": 1.44, + "learning_rate": 1.8843322132547778e-05, + "loss": 0.4262, + "step": 3439 + }, + { + "epoch": 1.44, + "learning_rate": 1.884253048999395e-05, + "loss": 0.4011, + "step": 3440 + }, + { + "epoch": 1.44, + "learning_rate": 1.8841738593269432e-05, + "loss": 0.3708, + "step": 3441 + }, + { + "epoch": 1.44, + "learning_rate": 1.884094644239698e-05, + "loss": 0.3951, + "step": 3442 + }, + { + "epoch": 1.44, + "learning_rate": 1.8840154037399368e-05, + "loss": 0.3575, + "step": 3443 + }, + { + "epoch": 1.44, + "learning_rate": 1.8839361378299376e-05, + "loss": 0.4257, + "step": 3444 + }, + { + "epoch": 1.44, + "learning_rate": 1.8838568465119786e-05, + "loss": 0.3638, + "step": 3445 + }, + { + "epoch": 1.44, + "learning_rate": 1.8837775297883384e-05, + "loss": 0.3661, + "step": 3446 + }, + { + "epoch": 1.44, + "learning_rate": 1.883698187661298e-05, + "loss": 0.4143, + "step": 3447 + }, + { + "epoch": 1.44, + "learning_rate": 1.8836188201331368e-05, + "loss": 0.3825, + "step": 3448 + }, + { + "epoch": 1.44, + "learning_rate": 1.8835394272061368e-05, + "loss": 0.3355, + "step": 3449 + }, + { + "epoch": 1.44, + "learning_rate": 1.88346000888258e-05, + "loss": 0.4143, + "step": 3450 + }, + { + "epoch": 1.44, + "learning_rate": 1.8833805651647498e-05, + "loss": 0.428, + "step": 3451 + }, + { + "epoch": 1.45, + "learning_rate": 1.8833010960549286e-05, + "loss": 0.3992, + "step": 3452 + }, + { + "epoch": 1.45, + "learning_rate": 1.883221601555401e-05, + "loss": 0.3523, + "step": 3453 + }, + { + "epoch": 1.45, + "learning_rate": 1.8831420816684528e-05, + "loss": 0.3957, + "step": 3454 + }, + { + "epoch": 1.45, + "learning_rate": 1.8830625363963685e-05, + "loss": 0.366, + "step": 3455 + }, + { + "epoch": 1.45, + "learning_rate": 1.882982965741436e-05, + "loss": 0.3754, + "step": 3456 + }, + { + "epoch": 1.45, + "learning_rate": 1.882903369705941e-05, + "loss": 0.4156, + "step": 3457 + }, + { + "epoch": 1.45, + "learning_rate": 1.8828237482921725e-05, + "loss": 0.4268, + "step": 3458 + }, + { + "epoch": 1.45, + "learning_rate": 1.8827441015024187e-05, + "loss": 0.3932, + "step": 3459 + }, + { + "epoch": 1.45, + "learning_rate": 1.882664429338969e-05, + "loss": 0.4338, + "step": 3460 + }, + { + "epoch": 1.45, + "learning_rate": 1.8825847318041136e-05, + "loss": 0.4078, + "step": 3461 + }, + { + "epoch": 1.45, + "learning_rate": 1.882505008900143e-05, + "loss": 0.4001, + "step": 3462 + }, + { + "epoch": 1.45, + "learning_rate": 1.8824252606293492e-05, + "loss": 0.3735, + "step": 3463 + }, + { + "epoch": 1.45, + "learning_rate": 1.8823454869940243e-05, + "loss": 0.3941, + "step": 3464 + }, + { + "epoch": 1.45, + "learning_rate": 1.8822656879964617e-05, + "loss": 0.3981, + "step": 3465 + }, + { + "epoch": 1.45, + "learning_rate": 1.8821858636389547e-05, + "loss": 0.4284, + "step": 3466 + }, + { + "epoch": 1.45, + "learning_rate": 1.8821060139237975e-05, + "loss": 0.3802, + "step": 3467 + }, + { + "epoch": 1.45, + "learning_rate": 1.8820261388532858e-05, + "loss": 0.3674, + "step": 3468 + }, + { + "epoch": 1.45, + "learning_rate": 1.8819462384297155e-05, + "loss": 0.3818, + "step": 3469 + }, + { + "epoch": 1.45, + "learning_rate": 1.8818663126553833e-05, + "loss": 0.4197, + "step": 3470 + }, + { + "epoch": 1.45, + "learning_rate": 1.8817863615325865e-05, + "loss": 0.425, + "step": 3471 + }, + { + "epoch": 1.45, + "learning_rate": 1.881706385063623e-05, + "loss": 0.3757, + "step": 3472 + }, + { + "epoch": 1.45, + "learning_rate": 1.881626383250792e-05, + "loss": 0.3716, + "step": 3473 + }, + { + "epoch": 1.45, + "learning_rate": 1.8815463560963932e-05, + "loss": 0.392, + "step": 3474 + }, + { + "epoch": 1.45, + "learning_rate": 1.8814663036027262e-05, + "loss": 0.3917, + "step": 3475 + }, + { + "epoch": 1.46, + "learning_rate": 1.881386225772093e-05, + "loss": 0.356, + "step": 3476 + }, + { + "epoch": 1.46, + "learning_rate": 1.8813061226067945e-05, + "loss": 0.3455, + "step": 3477 + }, + { + "epoch": 1.46, + "learning_rate": 1.881225994109134e-05, + "loss": 0.4187, + "step": 3478 + }, + { + "epoch": 1.46, + "learning_rate": 1.881145840281414e-05, + "loss": 0.3953, + "step": 3479 + }, + { + "epoch": 1.46, + "learning_rate": 1.8810656611259387e-05, + "loss": 0.3728, + "step": 3480 + }, + { + "epoch": 1.46, + "learning_rate": 1.8809854566450128e-05, + "loss": 0.3603, + "step": 3481 + }, + { + "epoch": 1.46, + "learning_rate": 1.8809052268409414e-05, + "loss": 0.3356, + "step": 3482 + }, + { + "epoch": 1.46, + "learning_rate": 1.8808249717160318e-05, + "loss": 0.4093, + "step": 3483 + }, + { + "epoch": 1.46, + "learning_rate": 1.8807446912725893e-05, + "loss": 0.4172, + "step": 3484 + }, + { + "epoch": 1.46, + "learning_rate": 1.8806643855129225e-05, + "loss": 0.3647, + "step": 3485 + }, + { + "epoch": 1.46, + "learning_rate": 1.8805840544393393e-05, + "loss": 0.4281, + "step": 3486 + }, + { + "epoch": 1.46, + "learning_rate": 1.880503698054149e-05, + "loss": 0.3747, + "step": 3487 + }, + { + "epoch": 1.46, + "learning_rate": 1.8804233163596613e-05, + "loss": 0.3777, + "step": 3488 + }, + { + "epoch": 1.46, + "learning_rate": 1.8803429093581863e-05, + "loss": 0.3581, + "step": 3489 + }, + { + "epoch": 1.46, + "learning_rate": 1.880262477052036e-05, + "loss": 0.3936, + "step": 3490 + }, + { + "epoch": 1.46, + "learning_rate": 1.8801820194435215e-05, + "loss": 0.3912, + "step": 3491 + }, + { + "epoch": 1.46, + "learning_rate": 1.8801015365349562e-05, + "loss": 0.3921, + "step": 3492 + }, + { + "epoch": 1.46, + "learning_rate": 1.880021028328653e-05, + "loss": 0.358, + "step": 3493 + }, + { + "epoch": 1.46, + "learning_rate": 1.8799404948269267e-05, + "loss": 0.3555, + "step": 3494 + }, + { + "epoch": 1.46, + "learning_rate": 1.879859936032091e-05, + "loss": 0.3767, + "step": 3495 + }, + { + "epoch": 1.46, + "learning_rate": 1.8797793519464628e-05, + "loss": 0.3615, + "step": 3496 + }, + { + "epoch": 1.46, + "learning_rate": 1.8796987425723578e-05, + "loss": 0.3714, + "step": 3497 + }, + { + "epoch": 1.46, + "learning_rate": 1.879618107912093e-05, + "loss": 0.3775, + "step": 3498 + }, + { + "epoch": 1.46, + "learning_rate": 1.8795374479679862e-05, + "loss": 0.4144, + "step": 3499 + }, + { + "epoch": 1.47, + "learning_rate": 1.879456762742356e-05, + "loss": 0.3544, + "step": 3500 + }, + { + "epoch": 1.47, + "learning_rate": 1.8793760522375214e-05, + "loss": 0.3959, + "step": 3501 + }, + { + "epoch": 1.47, + "learning_rate": 1.8792953164558026e-05, + "loss": 0.4117, + "step": 3502 + }, + { + "epoch": 1.47, + "learning_rate": 1.8792145553995203e-05, + "loss": 0.3799, + "step": 3503 + }, + { + "epoch": 1.47, + "learning_rate": 1.879133769070996e-05, + "loss": 0.4163, + "step": 3504 + }, + { + "epoch": 1.47, + "learning_rate": 1.8790529574725514e-05, + "loss": 0.4168, + "step": 3505 + }, + { + "epoch": 1.47, + "learning_rate": 1.8789721206065096e-05, + "loss": 0.3888, + "step": 3506 + }, + { + "epoch": 1.47, + "learning_rate": 1.8788912584751938e-05, + "loss": 0.3918, + "step": 3507 + }, + { + "epoch": 1.47, + "learning_rate": 1.8788103710809292e-05, + "loss": 0.3519, + "step": 3508 + }, + { + "epoch": 1.47, + "learning_rate": 1.8787294584260402e-05, + "loss": 0.377, + "step": 3509 + }, + { + "epoch": 1.47, + "learning_rate": 1.8786485205128526e-05, + "loss": 0.3672, + "step": 3510 + }, + { + "epoch": 1.47, + "learning_rate": 1.878567557343693e-05, + "loss": 0.4139, + "step": 3511 + }, + { + "epoch": 1.47, + "learning_rate": 1.8784865689208888e-05, + "loss": 0.3682, + "step": 3512 + }, + { + "epoch": 1.47, + "learning_rate": 1.8784055552467674e-05, + "loss": 0.4095, + "step": 3513 + }, + { + "epoch": 1.47, + "learning_rate": 1.878324516323658e-05, + "loss": 0.4295, + "step": 3514 + }, + { + "epoch": 1.47, + "learning_rate": 1.87824345215389e-05, + "loss": 0.4504, + "step": 3515 + }, + { + "epoch": 1.47, + "learning_rate": 1.878162362739793e-05, + "loss": 0.3981, + "step": 3516 + }, + { + "epoch": 1.47, + "learning_rate": 1.878081248083698e-05, + "loss": 0.4034, + "step": 3517 + }, + { + "epoch": 1.47, + "learning_rate": 1.878000108187937e-05, + "loss": 0.3617, + "step": 3518 + }, + { + "epoch": 1.47, + "learning_rate": 1.877918943054842e-05, + "loss": 0.3635, + "step": 3519 + }, + { + "epoch": 1.47, + "learning_rate": 1.877837752686746e-05, + "loss": 0.4073, + "step": 3520 + }, + { + "epoch": 1.47, + "learning_rate": 1.8777565370859826e-05, + "loss": 0.3933, + "step": 3521 + }, + { + "epoch": 1.47, + "learning_rate": 1.8776752962548866e-05, + "loss": 0.3756, + "step": 3522 + }, + { + "epoch": 1.47, + "learning_rate": 1.8775940301957932e-05, + "loss": 0.3968, + "step": 3523 + }, + { + "epoch": 1.48, + "learning_rate": 1.877512738911038e-05, + "loss": 0.3968, + "step": 3524 + }, + { + "epoch": 1.48, + "learning_rate": 1.8774314224029578e-05, + "loss": 0.4166, + "step": 3525 + }, + { + "epoch": 1.48, + "learning_rate": 1.87735008067389e-05, + "loss": 0.3903, + "step": 3526 + }, + { + "epoch": 1.48, + "learning_rate": 1.8772687137261728e-05, + "loss": 0.3604, + "step": 3527 + }, + { + "epoch": 1.48, + "learning_rate": 1.8771873215621447e-05, + "loss": 0.3919, + "step": 3528 + }, + { + "epoch": 1.48, + "learning_rate": 1.8771059041841454e-05, + "loss": 0.4071, + "step": 3529 + }, + { + "epoch": 1.48, + "learning_rate": 1.8770244615945153e-05, + "loss": 0.4072, + "step": 3530 + }, + { + "epoch": 1.48, + "learning_rate": 1.8769429937955954e-05, + "loss": 0.3793, + "step": 3531 + }, + { + "epoch": 1.48, + "learning_rate": 1.8768615007897274e-05, + "loss": 0.3788, + "step": 3532 + }, + { + "epoch": 1.48, + "learning_rate": 1.8767799825792533e-05, + "loss": 0.3738, + "step": 3533 + }, + { + "epoch": 1.48, + "learning_rate": 1.8766984391665173e-05, + "loss": 0.3879, + "step": 3534 + }, + { + "epoch": 1.48, + "learning_rate": 1.876616870553862e-05, + "loss": 0.3786, + "step": 3535 + }, + { + "epoch": 1.48, + "learning_rate": 1.876535276743633e-05, + "loss": 0.4138, + "step": 3536 + }, + { + "epoch": 1.48, + "learning_rate": 1.8764536577381748e-05, + "loss": 0.4008, + "step": 3537 + }, + { + "epoch": 1.48, + "learning_rate": 1.8763720135398348e-05, + "loss": 0.3867, + "step": 3538 + }, + { + "epoch": 1.48, + "learning_rate": 1.8762903441509582e-05, + "loss": 0.3631, + "step": 3539 + }, + { + "epoch": 1.48, + "learning_rate": 1.8762086495738937e-05, + "loss": 0.35, + "step": 3540 + }, + { + "epoch": 1.48, + "learning_rate": 1.876126929810989e-05, + "loss": 0.3891, + "step": 3541 + }, + { + "epoch": 1.48, + "learning_rate": 1.8760451848645932e-05, + "loss": 0.3692, + "step": 3542 + }, + { + "epoch": 1.48, + "learning_rate": 1.8759634147370555e-05, + "loss": 0.4024, + "step": 3543 + }, + { + "epoch": 1.48, + "learning_rate": 1.8758816194307272e-05, + "loss": 0.3913, + "step": 3544 + }, + { + "epoch": 1.48, + "learning_rate": 1.875799798947959e-05, + "loss": 0.3868, + "step": 3545 + }, + { + "epoch": 1.48, + "learning_rate": 1.8757179532911026e-05, + "loss": 0.3665, + "step": 3546 + }, + { + "epoch": 1.49, + "learning_rate": 1.8756360824625107e-05, + "loss": 0.4118, + "step": 3547 + }, + { + "epoch": 1.49, + "learning_rate": 1.875554186464537e-05, + "loss": 0.3749, + "step": 3548 + }, + { + "epoch": 1.49, + "learning_rate": 1.8754722652995346e-05, + "loss": 0.4166, + "step": 3549 + }, + { + "epoch": 1.49, + "learning_rate": 1.8753903189698594e-05, + "loss": 0.3962, + "step": 3550 + }, + { + "epoch": 1.49, + "learning_rate": 1.875308347477866e-05, + "loss": 0.4516, + "step": 3551 + }, + { + "epoch": 1.49, + "learning_rate": 1.8752263508259107e-05, + "loss": 0.4046, + "step": 3552 + }, + { + "epoch": 1.49, + "learning_rate": 1.875144329016351e-05, + "loss": 0.4097, + "step": 3553 + }, + { + "epoch": 1.49, + "learning_rate": 1.875062282051544e-05, + "loss": 0.4241, + "step": 3554 + }, + { + "epoch": 1.49, + "learning_rate": 1.874980209933848e-05, + "loss": 0.38, + "step": 3555 + }, + { + "epoch": 1.49, + "learning_rate": 1.8748981126656224e-05, + "loss": 0.3898, + "step": 3556 + }, + { + "epoch": 1.49, + "learning_rate": 1.8748159902492273e-05, + "loss": 0.4338, + "step": 3557 + }, + { + "epoch": 1.49, + "learning_rate": 1.8747338426870222e-05, + "loss": 0.4228, + "step": 3558 + }, + { + "epoch": 1.49, + "learning_rate": 1.8746516699813693e-05, + "loss": 0.35, + "step": 3559 + }, + { + "epoch": 1.49, + "learning_rate": 1.8745694721346305e-05, + "loss": 0.3797, + "step": 3560 + }, + { + "epoch": 1.49, + "learning_rate": 1.874487249149168e-05, + "loss": 0.4204, + "step": 3561 + }, + { + "epoch": 1.49, + "learning_rate": 1.8744050010273458e-05, + "loss": 0.4093, + "step": 3562 + }, + { + "epoch": 1.49, + "learning_rate": 1.874322727771528e-05, + "loss": 0.4614, + "step": 3563 + }, + { + "epoch": 1.49, + "learning_rate": 1.8742404293840784e-05, + "loss": 0.3874, + "step": 3564 + }, + { + "epoch": 1.49, + "learning_rate": 1.8741581058673642e-05, + "loss": 0.3658, + "step": 3565 + }, + { + "epoch": 1.49, + "learning_rate": 1.874075757223751e-05, + "loss": 0.4201, + "step": 3566 + }, + { + "epoch": 1.49, + "learning_rate": 1.8739933834556054e-05, + "loss": 0.3875, + "step": 3567 + }, + { + "epoch": 1.49, + "learning_rate": 1.8739109845652957e-05, + "loss": 0.4209, + "step": 3568 + }, + { + "epoch": 1.49, + "learning_rate": 1.87382856055519e-05, + "loss": 0.347, + "step": 3569 + }, + { + "epoch": 1.49, + "learning_rate": 1.873746111427658e-05, + "loss": 0.4023, + "step": 3570 + }, + { + "epoch": 1.5, + "learning_rate": 1.8736636371850694e-05, + "loss": 0.3956, + "step": 3571 + }, + { + "epoch": 1.5, + "learning_rate": 1.8735811378297946e-05, + "loss": 0.406, + "step": 3572 + }, + { + "epoch": 1.5, + "learning_rate": 1.8734986133642058e-05, + "loss": 0.3692, + "step": 3573 + }, + { + "epoch": 1.5, + "learning_rate": 1.8734160637906738e-05, + "loss": 0.3746, + "step": 3574 + }, + { + "epoch": 1.5, + "learning_rate": 1.8733334891115725e-05, + "loss": 0.3654, + "step": 3575 + }, + { + "epoch": 1.5, + "learning_rate": 1.873250889329275e-05, + "loss": 0.3501, + "step": 3576 + }, + { + "epoch": 1.5, + "learning_rate": 1.8731682644461555e-05, + "loss": 0.4146, + "step": 3577 + }, + { + "epoch": 1.5, + "learning_rate": 1.873085614464589e-05, + "loss": 0.3821, + "step": 3578 + }, + { + "epoch": 1.5, + "learning_rate": 1.8730029393869512e-05, + "loss": 0.3517, + "step": 3579 + }, + { + "epoch": 1.5, + "learning_rate": 1.8729202392156186e-05, + "loss": 0.4147, + "step": 3580 + }, + { + "epoch": 1.5, + "learning_rate": 1.8728375139529688e-05, + "loss": 0.4, + "step": 3581 + }, + { + "epoch": 1.5, + "learning_rate": 1.872754763601379e-05, + "loss": 0.3598, + "step": 3582 + }, + { + "epoch": 1.5, + "learning_rate": 1.8726719881632278e-05, + "loss": 0.405, + "step": 3583 + }, + { + "epoch": 1.5, + "learning_rate": 1.8725891876408946e-05, + "loss": 0.4087, + "step": 3584 + }, + { + "epoch": 1.5, + "learning_rate": 1.8725063620367598e-05, + "loss": 0.4424, + "step": 3585 + }, + { + "epoch": 1.5, + "learning_rate": 1.872423511353204e-05, + "loss": 0.3516, + "step": 3586 + }, + { + "epoch": 1.5, + "learning_rate": 1.8723406355926078e-05, + "loss": 0.4001, + "step": 3587 + }, + { + "epoch": 1.5, + "learning_rate": 1.872257734757355e-05, + "loss": 0.3657, + "step": 3588 + }, + { + "epoch": 1.5, + "learning_rate": 1.8721748088498272e-05, + "loss": 0.3784, + "step": 3589 + }, + { + "epoch": 1.5, + "learning_rate": 1.8720918578724085e-05, + "loss": 0.4014, + "step": 3590 + }, + { + "epoch": 1.5, + "learning_rate": 1.8720088818274835e-05, + "loss": 0.4215, + "step": 3591 + }, + { + "epoch": 1.5, + "learning_rate": 1.8719258807174366e-05, + "loss": 0.3759, + "step": 3592 + }, + { + "epoch": 1.5, + "learning_rate": 1.8718428545446545e-05, + "loss": 0.3367, + "step": 3593 + }, + { + "epoch": 1.5, + "learning_rate": 1.8717598033115224e-05, + "loss": 0.3609, + "step": 3594 + }, + { + "epoch": 1.51, + "learning_rate": 1.871676727020429e-05, + "loss": 0.3814, + "step": 3595 + }, + { + "epoch": 1.51, + "learning_rate": 1.871593625673761e-05, + "loss": 0.4125, + "step": 3596 + }, + { + "epoch": 1.51, + "learning_rate": 1.8715104992739085e-05, + "loss": 0.3841, + "step": 3597 + }, + { + "epoch": 1.51, + "learning_rate": 1.8714273478232597e-05, + "loss": 0.3855, + "step": 3598 + }, + { + "epoch": 1.51, + "learning_rate": 1.871344171324205e-05, + "loss": 0.4007, + "step": 3599 + }, + { + "epoch": 1.51, + "learning_rate": 1.871260969779135e-05, + "loss": 0.3679, + "step": 3600 + }, + { + "epoch": 1.51, + "learning_rate": 1.871177743190442e-05, + "loss": 0.3722, + "step": 3601 + }, + { + "epoch": 1.51, + "learning_rate": 1.871094491560518e-05, + "loss": 0.3753, + "step": 3602 + }, + { + "epoch": 1.51, + "learning_rate": 1.8710112148917555e-05, + "loss": 0.4449, + "step": 3603 + }, + { + "epoch": 1.51, + "learning_rate": 1.8709279131865486e-05, + "loss": 0.4068, + "step": 3604 + }, + { + "epoch": 1.51, + "learning_rate": 1.8708445864472917e-05, + "loss": 0.3895, + "step": 3605 + }, + { + "epoch": 1.51, + "learning_rate": 1.87076123467638e-05, + "loss": 0.3729, + "step": 3606 + }, + { + "epoch": 1.51, + "learning_rate": 1.8706778578762094e-05, + "loss": 0.3501, + "step": 3607 + }, + { + "epoch": 1.51, + "learning_rate": 1.870594456049176e-05, + "loss": 0.337, + "step": 3608 + }, + { + "epoch": 1.51, + "learning_rate": 1.870511029197678e-05, + "loss": 0.3914, + "step": 3609 + }, + { + "epoch": 1.51, + "learning_rate": 1.8704275773241126e-05, + "loss": 0.3952, + "step": 3610 + }, + { + "epoch": 1.51, + "learning_rate": 1.8703441004308792e-05, + "loss": 0.3786, + "step": 3611 + }, + { + "epoch": 1.51, + "learning_rate": 1.8702605985203767e-05, + "loss": 0.38, + "step": 3612 + }, + { + "epoch": 1.51, + "learning_rate": 1.8701770715950058e-05, + "loss": 0.4337, + "step": 3613 + }, + { + "epoch": 1.51, + "learning_rate": 1.8700935196571666e-05, + "loss": 0.3772, + "step": 3614 + }, + { + "epoch": 1.51, + "learning_rate": 1.8700099427092618e-05, + "loss": 0.3557, + "step": 3615 + }, + { + "epoch": 1.51, + "learning_rate": 1.869926340753693e-05, + "loss": 0.363, + "step": 3616 + }, + { + "epoch": 1.51, + "learning_rate": 1.8698427137928637e-05, + "loss": 0.4203, + "step": 3617 + }, + { + "epoch": 1.51, + "learning_rate": 1.8697590618291772e-05, + "loss": 0.4204, + "step": 3618 + }, + { + "epoch": 1.52, + "learning_rate": 1.8696753848650386e-05, + "loss": 0.3818, + "step": 3619 + }, + { + "epoch": 1.52, + "learning_rate": 1.8695916829028525e-05, + "loss": 0.4434, + "step": 3620 + }, + { + "epoch": 1.52, + "learning_rate": 1.8695079559450255e-05, + "loss": 0.4064, + "step": 3621 + }, + { + "epoch": 1.52, + "learning_rate": 1.8694242039939636e-05, + "loss": 0.3619, + "step": 3622 + }, + { + "epoch": 1.52, + "learning_rate": 1.8693404270520745e-05, + "loss": 0.3922, + "step": 3623 + }, + { + "epoch": 1.52, + "learning_rate": 1.8692566251217665e-05, + "loss": 0.3681, + "step": 3624 + }, + { + "epoch": 1.52, + "learning_rate": 1.8691727982054478e-05, + "loss": 0.3653, + "step": 3625 + }, + { + "epoch": 1.52, + "learning_rate": 1.8690889463055285e-05, + "loss": 0.3871, + "step": 3626 + }, + { + "epoch": 1.52, + "learning_rate": 1.8690050694244188e-05, + "loss": 0.3941, + "step": 3627 + }, + { + "epoch": 1.52, + "learning_rate": 1.868921167564529e-05, + "loss": 0.3549, + "step": 3628 + }, + { + "epoch": 1.52, + "learning_rate": 1.8688372407282717e-05, + "loss": 0.3583, + "step": 3629 + }, + { + "epoch": 1.52, + "learning_rate": 1.868753288918059e-05, + "loss": 0.3913, + "step": 3630 + }, + { + "epoch": 1.52, + "learning_rate": 1.868669312136304e-05, + "loss": 0.3775, + "step": 3631 + }, + { + "epoch": 1.52, + "learning_rate": 1.86858531038542e-05, + "loss": 0.3765, + "step": 3632 + }, + { + "epoch": 1.52, + "learning_rate": 1.8685012836678225e-05, + "loss": 0.3596, + "step": 3633 + }, + { + "epoch": 1.52, + "learning_rate": 1.8684172319859258e-05, + "loss": 0.3955, + "step": 3634 + }, + { + "epoch": 1.52, + "learning_rate": 1.8683331553421467e-05, + "loss": 0.468, + "step": 3635 + }, + { + "epoch": 1.52, + "learning_rate": 1.8682490537389017e-05, + "loss": 0.4153, + "step": 3636 + }, + { + "epoch": 1.52, + "learning_rate": 1.868164927178608e-05, + "loss": 0.3604, + "step": 3637 + }, + { + "epoch": 1.52, + "learning_rate": 1.8680807756636837e-05, + "loss": 0.3752, + "step": 3638 + }, + { + "epoch": 1.52, + "learning_rate": 1.8679965991965484e-05, + "loss": 0.3992, + "step": 3639 + }, + { + "epoch": 1.52, + "learning_rate": 1.8679123977796205e-05, + "loss": 0.3463, + "step": 3640 + }, + { + "epoch": 1.52, + "learning_rate": 1.867828171415321e-05, + "loss": 0.412, + "step": 3641 + }, + { + "epoch": 1.52, + "learning_rate": 1.867743920106071e-05, + "loss": 0.3994, + "step": 3642 + }, + { + "epoch": 1.53, + "learning_rate": 1.867659643854292e-05, + "loss": 0.4042, + "step": 3643 + }, + { + "epoch": 1.53, + "learning_rate": 1.867575342662406e-05, + "loss": 0.4215, + "step": 3644 + }, + { + "epoch": 1.53, + "learning_rate": 1.8674910165328374e-05, + "loss": 0.423, + "step": 3645 + }, + { + "epoch": 1.53, + "learning_rate": 1.8674066654680088e-05, + "loss": 0.3863, + "step": 3646 + }, + { + "epoch": 1.53, + "learning_rate": 1.8673222894703456e-05, + "loss": 0.419, + "step": 3647 + }, + { + "epoch": 1.53, + "learning_rate": 1.8672378885422728e-05, + "loss": 0.4253, + "step": 3648 + }, + { + "epoch": 1.53, + "learning_rate": 1.867153462686216e-05, + "loss": 0.3741, + "step": 3649 + }, + { + "epoch": 1.53, + "learning_rate": 1.8670690119046032e-05, + "loss": 0.3937, + "step": 3650 + }, + { + "epoch": 1.53, + "learning_rate": 1.8669845361998607e-05, + "loss": 0.3863, + "step": 3651 + }, + { + "epoch": 1.53, + "learning_rate": 1.8669000355744172e-05, + "loss": 0.4042, + "step": 3652 + }, + { + "epoch": 1.53, + "learning_rate": 1.8668155100307013e-05, + "loss": 0.4318, + "step": 3653 + }, + { + "epoch": 1.53, + "learning_rate": 1.8667309595711426e-05, + "loss": 0.4049, + "step": 3654 + }, + { + "epoch": 1.53, + "learning_rate": 1.8666463841981717e-05, + "loss": 0.4273, + "step": 3655 + }, + { + "epoch": 1.53, + "learning_rate": 1.8665617839142198e-05, + "loss": 0.3994, + "step": 3656 + }, + { + "epoch": 1.53, + "learning_rate": 1.866477158721718e-05, + "loss": 0.3565, + "step": 3657 + }, + { + "epoch": 1.53, + "learning_rate": 1.8663925086230998e-05, + "loss": 0.3925, + "step": 3658 + }, + { + "epoch": 1.53, + "learning_rate": 1.866307833620797e-05, + "loss": 0.446, + "step": 3659 + }, + { + "epoch": 1.53, + "learning_rate": 1.8662231337172446e-05, + "loss": 0.388, + "step": 3660 + }, + { + "epoch": 1.53, + "learning_rate": 1.866138408914877e-05, + "loss": 0.3558, + "step": 3661 + }, + { + "epoch": 1.53, + "learning_rate": 1.8660536592161293e-05, + "loss": 0.3339, + "step": 3662 + }, + { + "epoch": 1.53, + "learning_rate": 1.8659688846234376e-05, + "loss": 0.357, + "step": 3663 + }, + { + "epoch": 1.53, + "learning_rate": 1.8658840851392386e-05, + "loss": 0.4077, + "step": 3664 + }, + { + "epoch": 1.53, + "learning_rate": 1.8657992607659706e-05, + "loss": 0.3655, + "step": 3665 + }, + { + "epoch": 1.53, + "learning_rate": 1.8657144115060707e-05, + "loss": 0.3917, + "step": 3666 + }, + { + "epoch": 1.54, + "learning_rate": 1.8656295373619785e-05, + "loss": 0.4059, + "step": 3667 + }, + { + "epoch": 1.54, + "learning_rate": 1.8655446383361332e-05, + "loss": 0.4232, + "step": 3668 + }, + { + "epoch": 1.54, + "learning_rate": 1.8654597144309753e-05, + "loss": 0.3869, + "step": 3669 + }, + { + "epoch": 1.54, + "learning_rate": 1.865374765648946e-05, + "loss": 0.3615, + "step": 3670 + }, + { + "epoch": 1.54, + "learning_rate": 1.865289791992487e-05, + "loss": 0.368, + "step": 3671 + }, + { + "epoch": 1.54, + "learning_rate": 1.8652047934640405e-05, + "loss": 0.3497, + "step": 3672 + }, + { + "epoch": 1.54, + "learning_rate": 1.86511977006605e-05, + "loss": 0.4531, + "step": 3673 + }, + { + "epoch": 1.54, + "learning_rate": 1.8650347218009596e-05, + "loss": 0.3943, + "step": 3674 + }, + { + "epoch": 1.54, + "learning_rate": 1.864949648671214e-05, + "loss": 0.3597, + "step": 3675 + }, + { + "epoch": 1.54, + "learning_rate": 1.8648645506792578e-05, + "loss": 0.396, + "step": 3676 + }, + { + "epoch": 1.54, + "learning_rate": 1.864779427827538e-05, + "loss": 0.3928, + "step": 3677 + }, + { + "epoch": 1.54, + "learning_rate": 1.8646942801185006e-05, + "loss": 0.3414, + "step": 3678 + }, + { + "epoch": 1.54, + "learning_rate": 1.8646091075545936e-05, + "loss": 0.3588, + "step": 3679 + }, + { + "epoch": 1.54, + "learning_rate": 1.864523910138265e-05, + "loss": 0.4464, + "step": 3680 + }, + { + "epoch": 1.54, + "learning_rate": 1.864438687871964e-05, + "loss": 0.3666, + "step": 3681 + }, + { + "epoch": 1.54, + "learning_rate": 1.8643534407581397e-05, + "loss": 0.4229, + "step": 3682 + }, + { + "epoch": 1.54, + "learning_rate": 1.8642681687992427e-05, + "loss": 0.3988, + "step": 3683 + }, + { + "epoch": 1.54, + "learning_rate": 1.8641828719977244e-05, + "loss": 0.3685, + "step": 3684 + }, + { + "epoch": 1.54, + "learning_rate": 1.864097550356036e-05, + "loss": 0.3398, + "step": 3685 + }, + { + "epoch": 1.54, + "learning_rate": 1.8640122038766307e-05, + "loss": 0.4044, + "step": 3686 + }, + { + "epoch": 1.54, + "learning_rate": 1.863926832561961e-05, + "loss": 0.3999, + "step": 3687 + }, + { + "epoch": 1.54, + "learning_rate": 1.8638414364144813e-05, + "loss": 0.3707, + "step": 3688 + }, + { + "epoch": 1.54, + "learning_rate": 1.863756015436646e-05, + "loss": 0.3947, + "step": 3689 + }, + { + "epoch": 1.54, + "learning_rate": 1.8636705696309104e-05, + "loss": 0.4347, + "step": 3690 + }, + { + "epoch": 1.55, + "learning_rate": 1.8635850989997308e-05, + "loss": 0.3899, + "step": 3691 + }, + { + "epoch": 1.55, + "learning_rate": 1.863499603545564e-05, + "loss": 0.362, + "step": 3692 + }, + { + "epoch": 1.55, + "learning_rate": 1.8634140832708673e-05, + "loss": 0.4191, + "step": 3693 + }, + { + "epoch": 1.55, + "learning_rate": 1.8633285381780987e-05, + "loss": 0.4036, + "step": 3694 + }, + { + "epoch": 1.55, + "learning_rate": 1.8632429682697175e-05, + "loss": 0.4099, + "step": 3695 + }, + { + "epoch": 1.55, + "learning_rate": 1.8631573735481834e-05, + "loss": 0.43, + "step": 3696 + }, + { + "epoch": 1.55, + "learning_rate": 1.8630717540159562e-05, + "loss": 0.3987, + "step": 3697 + }, + { + "epoch": 1.55, + "learning_rate": 1.8629861096754977e-05, + "loss": 0.3406, + "step": 3698 + }, + { + "epoch": 1.55, + "learning_rate": 1.862900440529269e-05, + "loss": 0.3903, + "step": 3699 + }, + { + "epoch": 1.55, + "learning_rate": 1.862814746579733e-05, + "loss": 0.3315, + "step": 3700 + }, + { + "epoch": 1.55, + "learning_rate": 1.8627290278293525e-05, + "loss": 0.3851, + "step": 3701 + }, + { + "epoch": 1.55, + "learning_rate": 1.862643284280592e-05, + "loss": 0.3995, + "step": 3702 + }, + { + "epoch": 1.55, + "learning_rate": 1.8625575159359157e-05, + "loss": 0.393, + "step": 3703 + }, + { + "epoch": 1.55, + "learning_rate": 1.862471722797789e-05, + "loss": 0.405, + "step": 3704 + }, + { + "epoch": 1.55, + "learning_rate": 1.8623859048686777e-05, + "loss": 0.3704, + "step": 3705 + }, + { + "epoch": 1.55, + "learning_rate": 1.862300062151049e-05, + "loss": 0.3837, + "step": 3706 + }, + { + "epoch": 1.55, + "learning_rate": 1.8622141946473705e-05, + "loss": 0.3839, + "step": 3707 + }, + { + "epoch": 1.55, + "learning_rate": 1.8621283023601097e-05, + "loss": 0.4162, + "step": 3708 + }, + { + "epoch": 1.55, + "learning_rate": 1.862042385291736e-05, + "loss": 0.4516, + "step": 3709 + }, + { + "epoch": 1.55, + "learning_rate": 1.8619564434447192e-05, + "loss": 0.3776, + "step": 3710 + }, + { + "epoch": 1.55, + "learning_rate": 1.861870476821529e-05, + "loss": 0.4294, + "step": 3711 + }, + { + "epoch": 1.55, + "learning_rate": 1.8617844854246368e-05, + "loss": 0.3658, + "step": 3712 + }, + { + "epoch": 1.55, + "learning_rate": 1.8616984692565144e-05, + "loss": 0.3766, + "step": 3713 + }, + { + "epoch": 1.55, + "learning_rate": 1.861612428319634e-05, + "loss": 0.3958, + "step": 3714 + }, + { + "epoch": 1.56, + "learning_rate": 1.861526362616469e-05, + "loss": 0.4002, + "step": 3715 + }, + { + "epoch": 1.56, + "learning_rate": 1.8614402721494934e-05, + "loss": 0.3409, + "step": 3716 + }, + { + "epoch": 1.56, + "learning_rate": 1.8613541569211815e-05, + "loss": 0.3693, + "step": 3717 + }, + { + "epoch": 1.56, + "learning_rate": 1.8612680169340085e-05, + "loss": 0.3764, + "step": 3718 + }, + { + "epoch": 1.56, + "learning_rate": 1.861181852190451e-05, + "loss": 0.3525, + "step": 3719 + }, + { + "epoch": 1.56, + "learning_rate": 1.8610956626929852e-05, + "loss": 0.3591, + "step": 3720 + }, + { + "epoch": 1.56, + "learning_rate": 1.8610094484440884e-05, + "loss": 0.3761, + "step": 3721 + }, + { + "epoch": 1.56, + "learning_rate": 1.8609232094462394e-05, + "loss": 0.3549, + "step": 3722 + }, + { + "epoch": 1.56, + "learning_rate": 1.8608369457019166e-05, + "loss": 0.3992, + "step": 3723 + }, + { + "epoch": 1.56, + "learning_rate": 1.8607506572135994e-05, + "loss": 0.3731, + "step": 3724 + }, + { + "epoch": 1.56, + "learning_rate": 1.860664343983769e-05, + "loss": 0.3797, + "step": 3725 + }, + { + "epoch": 1.56, + "learning_rate": 1.8605780060149056e-05, + "loss": 0.3526, + "step": 3726 + }, + { + "epoch": 1.56, + "learning_rate": 1.8604916433094908e-05, + "loss": 0.3517, + "step": 3727 + }, + { + "epoch": 1.56, + "learning_rate": 1.860405255870007e-05, + "loss": 0.3643, + "step": 3728 + }, + { + "epoch": 1.56, + "learning_rate": 1.8603188436989382e-05, + "loss": 0.3997, + "step": 3729 + }, + { + "epoch": 1.56, + "learning_rate": 1.8602324067987678e-05, + "loss": 0.3371, + "step": 3730 + }, + { + "epoch": 1.56, + "learning_rate": 1.86014594517198e-05, + "loss": 0.4331, + "step": 3731 + }, + { + "epoch": 1.56, + "learning_rate": 1.8600594588210602e-05, + "loss": 0.4084, + "step": 3732 + }, + { + "epoch": 1.56, + "learning_rate": 1.8599729477484944e-05, + "loss": 0.3512, + "step": 3733 + }, + { + "epoch": 1.56, + "learning_rate": 1.8598864119567693e-05, + "loss": 0.4022, + "step": 3734 + }, + { + "epoch": 1.56, + "learning_rate": 1.8597998514483724e-05, + "loss": 0.3846, + "step": 3735 + }, + { + "epoch": 1.56, + "learning_rate": 1.859713266225792e-05, + "loss": 0.3828, + "step": 3736 + }, + { + "epoch": 1.56, + "learning_rate": 1.8596266562915163e-05, + "loss": 0.4152, + "step": 3737 + }, + { + "epoch": 1.56, + "learning_rate": 1.8595400216480356e-05, + "loss": 0.347, + "step": 3738 + }, + { + "epoch": 1.57, + "learning_rate": 1.8594533622978394e-05, + "loss": 0.4058, + "step": 3739 + }, + { + "epoch": 1.57, + "learning_rate": 1.859366678243419e-05, + "loss": 0.4159, + "step": 3740 + }, + { + "epoch": 1.57, + "learning_rate": 1.859279969487266e-05, + "loss": 0.3883, + "step": 3741 + }, + { + "epoch": 1.57, + "learning_rate": 1.859193236031873e-05, + "loss": 0.3622, + "step": 3742 + }, + { + "epoch": 1.57, + "learning_rate": 1.859106477879733e-05, + "loss": 0.3714, + "step": 3743 + }, + { + "epoch": 1.57, + "learning_rate": 1.859019695033339e-05, + "loss": 0.3634, + "step": 3744 + }, + { + "epoch": 1.57, + "learning_rate": 1.8589328874951866e-05, + "loss": 0.442, + "step": 3745 + }, + { + "epoch": 1.57, + "learning_rate": 1.8588460552677703e-05, + "loss": 0.426, + "step": 3746 + }, + { + "epoch": 1.57, + "learning_rate": 1.8587591983535868e-05, + "loss": 0.3693, + "step": 3747 + }, + { + "epoch": 1.57, + "learning_rate": 1.8586723167551315e-05, + "loss": 0.4352, + "step": 3748 + }, + { + "epoch": 1.57, + "learning_rate": 1.858585410474903e-05, + "loss": 0.3763, + "step": 3749 + }, + { + "epoch": 1.57, + "learning_rate": 1.8584984795153984e-05, + "loss": 0.4033, + "step": 3750 + }, + { + "epoch": 1.57, + "learning_rate": 1.858411523879117e-05, + "loss": 0.4202, + "step": 3751 + }, + { + "epoch": 1.57, + "learning_rate": 1.858324543568558e-05, + "loss": 0.4004, + "step": 3752 + }, + { + "epoch": 1.57, + "learning_rate": 1.8582375385862215e-05, + "loss": 0.4286, + "step": 3753 + }, + { + "epoch": 1.57, + "learning_rate": 1.858150508934609e-05, + "loss": 0.3823, + "step": 3754 + }, + { + "epoch": 1.57, + "learning_rate": 1.8580634546162212e-05, + "loss": 0.3807, + "step": 3755 + }, + { + "epoch": 1.57, + "learning_rate": 1.857976375633561e-05, + "loss": 0.3462, + "step": 3756 + }, + { + "epoch": 1.57, + "learning_rate": 1.8578892719891314e-05, + "loss": 0.379, + "step": 3757 + }, + { + "epoch": 1.57, + "learning_rate": 1.8578021436854357e-05, + "loss": 0.3501, + "step": 3758 + }, + { + "epoch": 1.57, + "learning_rate": 1.8577149907249786e-05, + "loss": 0.384, + "step": 3759 + }, + { + "epoch": 1.57, + "learning_rate": 1.8576278131102655e-05, + "loss": 0.3863, + "step": 3760 + }, + { + "epoch": 1.57, + "learning_rate": 1.8575406108438017e-05, + "loss": 0.3424, + "step": 3761 + }, + { + "epoch": 1.58, + "learning_rate": 1.857453383928094e-05, + "loss": 0.395, + "step": 3762 + }, + { + "epoch": 1.58, + "learning_rate": 1.8573661323656495e-05, + "loss": 0.3597, + "step": 3763 + }, + { + "epoch": 1.58, + "learning_rate": 1.8572788561589766e-05, + "loss": 0.391, + "step": 3764 + }, + { + "epoch": 1.58, + "learning_rate": 1.857191555310584e-05, + "loss": 0.3818, + "step": 3765 + }, + { + "epoch": 1.58, + "learning_rate": 1.85710422982298e-05, + "loss": 0.4242, + "step": 3766 + }, + { + "epoch": 1.58, + "learning_rate": 1.8570168796986764e-05, + "loss": 0.42, + "step": 3767 + }, + { + "epoch": 1.58, + "learning_rate": 1.8569295049401826e-05, + "loss": 0.3615, + "step": 3768 + }, + { + "epoch": 1.58, + "learning_rate": 1.8568421055500107e-05, + "loss": 0.3709, + "step": 3769 + }, + { + "epoch": 1.58, + "learning_rate": 1.8567546815306726e-05, + "loss": 0.3988, + "step": 3770 + }, + { + "epoch": 1.58, + "learning_rate": 1.8566672328846816e-05, + "loss": 0.3838, + "step": 3771 + }, + { + "epoch": 1.58, + "learning_rate": 1.8565797596145515e-05, + "loss": 0.3679, + "step": 3772 + }, + { + "epoch": 1.58, + "learning_rate": 1.8564922617227958e-05, + "loss": 0.3765, + "step": 3773 + }, + { + "epoch": 1.58, + "learning_rate": 1.8564047392119306e-05, + "loss": 0.4416, + "step": 3774 + }, + { + "epoch": 1.58, + "learning_rate": 1.8563171920844706e-05, + "loss": 0.3616, + "step": 3775 + }, + { + "epoch": 1.58, + "learning_rate": 1.856229620342933e-05, + "loss": 0.3814, + "step": 3776 + }, + { + "epoch": 1.58, + "learning_rate": 1.8561420239898348e-05, + "loss": 0.3511, + "step": 3777 + }, + { + "epoch": 1.58, + "learning_rate": 1.856054403027694e-05, + "loss": 0.3562, + "step": 3778 + }, + { + "epoch": 1.58, + "learning_rate": 1.855966757459029e-05, + "loss": 0.3642, + "step": 3779 + }, + { + "epoch": 1.58, + "learning_rate": 1.855879087286359e-05, + "loss": 0.4004, + "step": 3780 + }, + { + "epoch": 1.58, + "learning_rate": 1.8557913925122042e-05, + "loss": 0.3681, + "step": 3781 + }, + { + "epoch": 1.58, + "learning_rate": 1.8557036731390853e-05, + "loss": 0.3826, + "step": 3782 + }, + { + "epoch": 1.58, + "learning_rate": 1.8556159291695235e-05, + "loss": 0.3754, + "step": 3783 + }, + { + "epoch": 1.58, + "learning_rate": 1.855528160606041e-05, + "loss": 0.3771, + "step": 3784 + }, + { + "epoch": 1.58, + "learning_rate": 1.855440367451161e-05, + "loss": 0.4514, + "step": 3785 + }, + { + "epoch": 1.59, + "learning_rate": 1.8553525497074065e-05, + "loss": 0.5203, + "step": 3786 + }, + { + "epoch": 1.59, + "learning_rate": 1.855264707377302e-05, + "loss": 0.4171, + "step": 3787 + }, + { + "epoch": 1.59, + "learning_rate": 1.8551768404633724e-05, + "loss": 0.3869, + "step": 3788 + }, + { + "epoch": 1.59, + "learning_rate": 1.8550889489681435e-05, + "loss": 0.4159, + "step": 3789 + }, + { + "epoch": 1.59, + "learning_rate": 1.8550010328941417e-05, + "loss": 0.3714, + "step": 3790 + }, + { + "epoch": 1.59, + "learning_rate": 1.8549130922438938e-05, + "loss": 0.375, + "step": 3791 + }, + { + "epoch": 1.59, + "learning_rate": 1.8548251270199274e-05, + "loss": 0.4059, + "step": 3792 + }, + { + "epoch": 1.59, + "learning_rate": 1.854737137224772e-05, + "loss": 0.4218, + "step": 3793 + }, + { + "epoch": 1.59, + "learning_rate": 1.8546491228609552e-05, + "loss": 0.3925, + "step": 3794 + }, + { + "epoch": 1.59, + "learning_rate": 1.8545610839310085e-05, + "loss": 0.3834, + "step": 3795 + }, + { + "epoch": 1.59, + "learning_rate": 1.854473020437461e-05, + "loss": 0.3778, + "step": 3796 + }, + { + "epoch": 1.59, + "learning_rate": 1.8543849323828452e-05, + "loss": 0.427, + "step": 3797 + }, + { + "epoch": 1.59, + "learning_rate": 1.854296819769693e-05, + "loss": 0.3654, + "step": 3798 + }, + { + "epoch": 1.59, + "learning_rate": 1.8542086826005366e-05, + "loss": 0.3874, + "step": 3799 + }, + { + "epoch": 1.59, + "learning_rate": 1.8541205208779095e-05, + "loss": 0.432, + "step": 3800 + }, + { + "epoch": 1.59, + "learning_rate": 1.8540323346043458e-05, + "loss": 0.3539, + "step": 3801 + }, + { + "epoch": 1.59, + "learning_rate": 1.853944123782381e-05, + "loss": 0.421, + "step": 3802 + }, + { + "epoch": 1.59, + "learning_rate": 1.8538558884145497e-05, + "loss": 0.3837, + "step": 3803 + }, + { + "epoch": 1.59, + "learning_rate": 1.8537676285033886e-05, + "loss": 0.3663, + "step": 3804 + }, + { + "epoch": 1.59, + "learning_rate": 1.853679344051435e-05, + "loss": 0.3943, + "step": 3805 + }, + { + "epoch": 1.59, + "learning_rate": 1.8535910350612257e-05, + "loss": 0.3913, + "step": 3806 + }, + { + "epoch": 1.59, + "learning_rate": 1.8535027015353e-05, + "loss": 0.3979, + "step": 3807 + }, + { + "epoch": 1.59, + "learning_rate": 1.8534143434761965e-05, + "loss": 0.4269, + "step": 3808 + }, + { + "epoch": 1.59, + "learning_rate": 1.8533259608864547e-05, + "loss": 0.3765, + "step": 3809 + }, + { + "epoch": 1.6, + "learning_rate": 1.8532375537686155e-05, + "loss": 0.3696, + "step": 3810 + }, + { + "epoch": 1.6, + "learning_rate": 1.85314912212522e-05, + "loss": 0.4306, + "step": 3811 + }, + { + "epoch": 1.6, + "learning_rate": 1.85306066595881e-05, + "loss": 0.3987, + "step": 3812 + }, + { + "epoch": 1.6, + "learning_rate": 1.8529721852719283e-05, + "loss": 0.4022, + "step": 3813 + }, + { + "epoch": 1.6, + "learning_rate": 1.852883680067118e-05, + "loss": 0.4001, + "step": 3814 + }, + { + "epoch": 1.6, + "learning_rate": 1.852795150346923e-05, + "loss": 0.3886, + "step": 3815 + }, + { + "epoch": 1.6, + "learning_rate": 1.8527065961138885e-05, + "loss": 0.339, + "step": 3816 + }, + { + "epoch": 1.6, + "learning_rate": 1.852618017370559e-05, + "loss": 0.3473, + "step": 3817 + }, + { + "epoch": 1.6, + "learning_rate": 1.8525294141194816e-05, + "loss": 0.3534, + "step": 3818 + }, + { + "epoch": 1.6, + "learning_rate": 1.852440786363203e-05, + "loss": 0.3533, + "step": 3819 + }, + { + "epoch": 1.6, + "learning_rate": 1.85235213410427e-05, + "loss": 0.4324, + "step": 3820 + }, + { + "epoch": 1.6, + "learning_rate": 1.8522634573452314e-05, + "loss": 0.3952, + "step": 3821 + }, + { + "epoch": 1.6, + "learning_rate": 1.8521747560886357e-05, + "loss": 0.4197, + "step": 3822 + }, + { + "epoch": 1.6, + "learning_rate": 1.8520860303370333e-05, + "loss": 0.3743, + "step": 3823 + }, + { + "epoch": 1.6, + "learning_rate": 1.851997280092974e-05, + "loss": 0.3709, + "step": 3824 + }, + { + "epoch": 1.6, + "learning_rate": 1.8519085053590083e-05, + "loss": 0.4111, + "step": 3825 + }, + { + "epoch": 1.6, + "learning_rate": 1.8518197061376895e-05, + "loss": 0.4298, + "step": 3826 + }, + { + "epoch": 1.6, + "learning_rate": 1.8517308824315688e-05, + "loss": 0.3054, + "step": 3827 + }, + { + "epoch": 1.6, + "learning_rate": 1.8516420342431994e-05, + "loss": 0.3889, + "step": 3828 + }, + { + "epoch": 1.6, + "learning_rate": 1.8515531615751357e-05, + "loss": 0.387, + "step": 3829 + }, + { + "epoch": 1.6, + "learning_rate": 1.851464264429932e-05, + "loss": 0.3876, + "step": 3830 + }, + { + "epoch": 1.6, + "learning_rate": 1.8513753428101436e-05, + "loss": 0.3743, + "step": 3831 + }, + { + "epoch": 1.6, + "learning_rate": 1.8512863967183265e-05, + "loss": 0.4047, + "step": 3832 + }, + { + "epoch": 1.6, + "learning_rate": 1.8511974261570375e-05, + "loss": 0.3511, + "step": 3833 + }, + { + "epoch": 1.61, + "learning_rate": 1.8511084311288338e-05, + "loss": 0.3873, + "step": 3834 + }, + { + "epoch": 1.61, + "learning_rate": 1.851019411636273e-05, + "loss": 0.3827, + "step": 3835 + }, + { + "epoch": 1.61, + "learning_rate": 1.850930367681915e-05, + "loss": 0.4201, + "step": 3836 + }, + { + "epoch": 1.61, + "learning_rate": 1.8508412992683183e-05, + "loss": 0.3714, + "step": 3837 + }, + { + "epoch": 1.61, + "learning_rate": 1.8507522063980436e-05, + "loss": 0.4159, + "step": 3838 + }, + { + "epoch": 1.61, + "learning_rate": 1.8506630890736515e-05, + "loss": 0.4006, + "step": 3839 + }, + { + "epoch": 1.61, + "learning_rate": 1.8505739472977043e-05, + "loss": 0.3933, + "step": 3840 + }, + { + "epoch": 1.61, + "learning_rate": 1.8504847810727632e-05, + "loss": 0.3986, + "step": 3841 + }, + { + "epoch": 1.61, + "learning_rate": 1.8503955904013922e-05, + "loss": 0.3899, + "step": 3842 + }, + { + "epoch": 1.61, + "learning_rate": 1.8503063752861543e-05, + "loss": 0.4075, + "step": 3843 + }, + { + "epoch": 1.61, + "learning_rate": 1.8502171357296144e-05, + "loss": 0.3603, + "step": 3844 + }, + { + "epoch": 1.61, + "learning_rate": 1.8501278717343372e-05, + "loss": 0.4044, + "step": 3845 + }, + { + "epoch": 1.61, + "learning_rate": 1.850038583302889e-05, + "loss": 0.375, + "step": 3846 + }, + { + "epoch": 1.61, + "learning_rate": 1.8499492704378355e-05, + "loss": 0.3622, + "step": 3847 + }, + { + "epoch": 1.61, + "learning_rate": 1.849859933141745e-05, + "loss": 0.4129, + "step": 3848 + }, + { + "epoch": 1.61, + "learning_rate": 1.8497705714171846e-05, + "loss": 0.4007, + "step": 3849 + }, + { + "epoch": 1.61, + "learning_rate": 1.849681185266723e-05, + "loss": 0.3879, + "step": 3850 + }, + { + "epoch": 1.61, + "learning_rate": 1.8495917746929303e-05, + "loss": 0.4252, + "step": 3851 + }, + { + "epoch": 1.61, + "learning_rate": 1.8495023396983754e-05, + "loss": 0.3669, + "step": 3852 + }, + { + "epoch": 1.61, + "learning_rate": 1.84941288028563e-05, + "loss": 0.3351, + "step": 3853 + }, + { + "epoch": 1.61, + "learning_rate": 1.8493233964572648e-05, + "loss": 0.3886, + "step": 3854 + }, + { + "epoch": 1.61, + "learning_rate": 1.8492338882158526e-05, + "loss": 0.415, + "step": 3855 + }, + { + "epoch": 1.61, + "learning_rate": 1.8491443555639656e-05, + "loss": 0.4072, + "step": 3856 + }, + { + "epoch": 1.61, + "learning_rate": 1.8490547985041774e-05, + "loss": 0.349, + "step": 3857 + }, + { + "epoch": 1.62, + "learning_rate": 1.8489652170390627e-05, + "loss": 0.3835, + "step": 3858 + }, + { + "epoch": 1.62, + "learning_rate": 1.848875611171196e-05, + "loss": 0.4041, + "step": 3859 + }, + { + "epoch": 1.62, + "learning_rate": 1.8487859809031537e-05, + "loss": 0.4144, + "step": 3860 + }, + { + "epoch": 1.62, + "learning_rate": 1.8486963262375113e-05, + "loss": 0.4173, + "step": 3861 + }, + { + "epoch": 1.62, + "learning_rate": 1.8486066471768457e-05, + "loss": 0.329, + "step": 3862 + }, + { + "epoch": 1.62, + "learning_rate": 1.848516943723736e-05, + "loss": 0.4009, + "step": 3863 + }, + { + "epoch": 1.62, + "learning_rate": 1.8484272158807587e-05, + "loss": 0.3595, + "step": 3864 + }, + { + "epoch": 1.62, + "learning_rate": 1.8483374636504943e-05, + "loss": 0.4056, + "step": 3865 + }, + { + "epoch": 1.62, + "learning_rate": 1.8482476870355226e-05, + "loss": 0.3897, + "step": 3866 + }, + { + "epoch": 1.62, + "learning_rate": 1.848157886038424e-05, + "loss": 0.3094, + "step": 3867 + }, + { + "epoch": 1.62, + "learning_rate": 1.848068060661779e-05, + "loss": 0.3604, + "step": 3868 + }, + { + "epoch": 1.62, + "learning_rate": 1.8479782109081702e-05, + "loss": 0.3688, + "step": 3869 + }, + { + "epoch": 1.62, + "learning_rate": 1.8478883367801804e-05, + "loss": 0.3693, + "step": 3870 + }, + { + "epoch": 1.62, + "learning_rate": 1.8477984382803928e-05, + "loss": 0.4919, + "step": 3871 + }, + { + "epoch": 1.62, + "learning_rate": 1.8477085154113913e-05, + "loss": 0.3725, + "step": 3872 + }, + { + "epoch": 1.62, + "learning_rate": 1.847618568175761e-05, + "loss": 0.3605, + "step": 3873 + }, + { + "epoch": 1.62, + "learning_rate": 1.847528596576087e-05, + "loss": 0.4308, + "step": 3874 + }, + { + "epoch": 1.62, + "learning_rate": 1.8474386006149556e-05, + "loss": 0.403, + "step": 3875 + }, + { + "epoch": 1.62, + "learning_rate": 1.847348580294954e-05, + "loss": 0.4061, + "step": 3876 + }, + { + "epoch": 1.62, + "learning_rate": 1.8472585356186688e-05, + "loss": 0.4084, + "step": 3877 + }, + { + "epoch": 1.62, + "learning_rate": 1.847168466588689e-05, + "loss": 0.4078, + "step": 3878 + }, + { + "epoch": 1.62, + "learning_rate": 1.8470783732076034e-05, + "loss": 0.3814, + "step": 3879 + }, + { + "epoch": 1.62, + "learning_rate": 1.8469882554780016e-05, + "loss": 0.4037, + "step": 3880 + }, + { + "epoch": 1.62, + "learning_rate": 1.846898113402474e-05, + "loss": 0.3961, + "step": 3881 + }, + { + "epoch": 1.63, + "learning_rate": 1.8468079469836117e-05, + "loss": 0.3874, + "step": 3882 + }, + { + "epoch": 1.63, + "learning_rate": 1.8467177562240062e-05, + "loss": 0.3401, + "step": 3883 + }, + { + "epoch": 1.63, + "learning_rate": 1.8466275411262504e-05, + "loss": 0.3788, + "step": 3884 + }, + { + "epoch": 1.63, + "learning_rate": 1.8465373016929377e-05, + "loss": 0.3697, + "step": 3885 + }, + { + "epoch": 1.63, + "learning_rate": 1.8464470379266606e-05, + "loss": 0.3385, + "step": 3886 + }, + { + "epoch": 1.63, + "learning_rate": 1.846356749830015e-05, + "loss": 0.4286, + "step": 3887 + }, + { + "epoch": 1.63, + "learning_rate": 1.8462664374055953e-05, + "loss": 0.3876, + "step": 3888 + }, + { + "epoch": 1.63, + "learning_rate": 1.8461761006559982e-05, + "loss": 0.3762, + "step": 3889 + }, + { + "epoch": 1.63, + "learning_rate": 1.84608573958382e-05, + "loss": 0.3909, + "step": 3890 + }, + { + "epoch": 1.63, + "learning_rate": 1.8459953541916576e-05, + "loss": 0.3997, + "step": 3891 + }, + { + "epoch": 1.63, + "learning_rate": 1.84590494448211e-05, + "loss": 0.3785, + "step": 3892 + }, + { + "epoch": 1.63, + "learning_rate": 1.8458145104577752e-05, + "loss": 0.4019, + "step": 3893 + }, + { + "epoch": 1.63, + "learning_rate": 1.8457240521212525e-05, + "loss": 0.3651, + "step": 3894 + }, + { + "epoch": 1.63, + "learning_rate": 1.845633569475143e-05, + "loss": 0.3841, + "step": 3895 + }, + { + "epoch": 1.63, + "learning_rate": 1.8455430625220465e-05, + "loss": 0.4117, + "step": 3896 + }, + { + "epoch": 1.63, + "learning_rate": 1.8454525312645654e-05, + "loss": 0.4056, + "step": 3897 + }, + { + "epoch": 1.63, + "learning_rate": 1.845361975705301e-05, + "loss": 0.379, + "step": 3898 + }, + { + "epoch": 1.63, + "learning_rate": 1.845271395846857e-05, + "loss": 0.4022, + "step": 3899 + }, + { + "epoch": 1.63, + "learning_rate": 1.8451807916918367e-05, + "loss": 0.4109, + "step": 3900 + }, + { + "epoch": 1.63, + "learning_rate": 1.845090163242845e-05, + "loss": 0.3745, + "step": 3901 + }, + { + "epoch": 1.63, + "learning_rate": 1.844999510502486e-05, + "loss": 0.4296, + "step": 3902 + }, + { + "epoch": 1.63, + "learning_rate": 1.844908833473366e-05, + "loss": 0.4323, + "step": 3903 + }, + { + "epoch": 1.63, + "learning_rate": 1.8448181321580916e-05, + "loss": 0.392, + "step": 3904 + }, + { + "epoch": 1.63, + "learning_rate": 1.8447274065592694e-05, + "loss": 0.4409, + "step": 3905 + }, + { + "epoch": 1.64, + "learning_rate": 1.8446366566795075e-05, + "loss": 0.4219, + "step": 3906 + }, + { + "epoch": 1.64, + "learning_rate": 1.8445458825214148e-05, + "loss": 0.4145, + "step": 3907 + }, + { + "epoch": 1.64, + "learning_rate": 1.8444550840875996e-05, + "loss": 0.3887, + "step": 3908 + }, + { + "epoch": 1.64, + "learning_rate": 1.8443642613806727e-05, + "loss": 0.3598, + "step": 3909 + }, + { + "epoch": 1.64, + "learning_rate": 1.8442734144032442e-05, + "loss": 0.3373, + "step": 3910 + }, + { + "epoch": 1.64, + "learning_rate": 1.844182543157926e-05, + "loss": 0.4054, + "step": 3911 + }, + { + "epoch": 1.64, + "learning_rate": 1.844091647647329e-05, + "loss": 0.3754, + "step": 3912 + }, + { + "epoch": 1.64, + "learning_rate": 1.8440007278740673e-05, + "loss": 0.4236, + "step": 3913 + }, + { + "epoch": 1.64, + "learning_rate": 1.8439097838407532e-05, + "loss": 0.3872, + "step": 3914 + }, + { + "epoch": 1.64, + "learning_rate": 1.8438188155500016e-05, + "loss": 0.3718, + "step": 3915 + }, + { + "epoch": 1.64, + "learning_rate": 1.8437278230044266e-05, + "loss": 0.4069, + "step": 3916 + }, + { + "epoch": 1.64, + "learning_rate": 1.8436368062066442e-05, + "loss": 0.4603, + "step": 3917 + }, + { + "epoch": 1.64, + "learning_rate": 1.8435457651592704e-05, + "loss": 0.382, + "step": 3918 + }, + { + "epoch": 1.64, + "learning_rate": 1.843454699864922e-05, + "loss": 0.3786, + "step": 3919 + }, + { + "epoch": 1.64, + "learning_rate": 1.843363610326217e-05, + "loss": 0.3877, + "step": 3920 + }, + { + "epoch": 1.64, + "learning_rate": 1.843272496545773e-05, + "loss": 0.3959, + "step": 3921 + }, + { + "epoch": 1.64, + "learning_rate": 1.8431813585262098e-05, + "loss": 0.3778, + "step": 3922 + }, + { + "epoch": 1.64, + "learning_rate": 1.8430901962701465e-05, + "loss": 0.4074, + "step": 3923 + }, + { + "epoch": 1.64, + "learning_rate": 1.8429990097802037e-05, + "loss": 0.403, + "step": 3924 + }, + { + "epoch": 1.64, + "learning_rate": 1.8429077990590026e-05, + "loss": 0.4485, + "step": 3925 + }, + { + "epoch": 1.64, + "learning_rate": 1.8428165641091643e-05, + "loss": 0.3935, + "step": 3926 + }, + { + "epoch": 1.64, + "learning_rate": 1.8427253049333124e-05, + "loss": 0.3865, + "step": 3927 + }, + { + "epoch": 1.64, + "learning_rate": 1.8426340215340688e-05, + "loss": 0.3605, + "step": 3928 + }, + { + "epoch": 1.64, + "learning_rate": 1.8425427139140586e-05, + "loss": 0.38, + "step": 3929 + }, + { + "epoch": 1.65, + "learning_rate": 1.8424513820759054e-05, + "loss": 0.4266, + "step": 3930 + }, + { + "epoch": 1.65, + "learning_rate": 1.8423600260222346e-05, + "loss": 0.4254, + "step": 3931 + }, + { + "epoch": 1.65, + "learning_rate": 1.8422686457556726e-05, + "loss": 0.398, + "step": 3932 + }, + { + "epoch": 1.65, + "learning_rate": 1.842177241278846e-05, + "loss": 0.3574, + "step": 3933 + }, + { + "epoch": 1.65, + "learning_rate": 1.8420858125943818e-05, + "loss": 0.3633, + "step": 3934 + }, + { + "epoch": 1.65, + "learning_rate": 1.841994359704908e-05, + "loss": 0.4486, + "step": 3935 + }, + { + "epoch": 1.65, + "learning_rate": 1.8419028826130537e-05, + "loss": 0.3329, + "step": 3936 + }, + { + "epoch": 1.65, + "learning_rate": 1.8418113813214477e-05, + "loss": 0.5135, + "step": 3937 + }, + { + "epoch": 1.65, + "learning_rate": 1.841719855832721e-05, + "loss": 0.4119, + "step": 3938 + }, + { + "epoch": 1.65, + "learning_rate": 1.8416283061495038e-05, + "loss": 0.3705, + "step": 3939 + }, + { + "epoch": 1.65, + "learning_rate": 1.841536732274428e-05, + "loss": 0.3832, + "step": 3940 + }, + { + "epoch": 1.65, + "learning_rate": 1.8414451342101253e-05, + "loss": 0.3343, + "step": 3941 + }, + { + "epoch": 1.65, + "learning_rate": 1.8413535119592292e-05, + "loss": 0.364, + "step": 3942 + }, + { + "epoch": 1.65, + "learning_rate": 1.841261865524373e-05, + "loss": 0.3662, + "step": 3943 + }, + { + "epoch": 1.65, + "learning_rate": 1.8411701949081906e-05, + "loss": 0.4828, + "step": 3944 + }, + { + "epoch": 1.65, + "learning_rate": 1.8410785001133177e-05, + "loss": 0.3881, + "step": 3945 + }, + { + "epoch": 1.65, + "learning_rate": 1.8409867811423897e-05, + "loss": 0.4021, + "step": 3946 + }, + { + "epoch": 1.65, + "learning_rate": 1.840895037998043e-05, + "loss": 0.4208, + "step": 3947 + }, + { + "epoch": 1.65, + "learning_rate": 1.840803270682914e-05, + "loss": 0.3583, + "step": 3948 + }, + { + "epoch": 1.65, + "learning_rate": 1.8407114791996416e-05, + "loss": 0.3859, + "step": 3949 + }, + { + "epoch": 1.65, + "learning_rate": 1.840619663550864e-05, + "loss": 0.373, + "step": 3950 + }, + { + "epoch": 1.65, + "learning_rate": 1.84052782373922e-05, + "loss": 0.3767, + "step": 3951 + }, + { + "epoch": 1.65, + "learning_rate": 1.8404359597673492e-05, + "loss": 0.3835, + "step": 3952 + }, + { + "epoch": 1.66, + "learning_rate": 1.840344071637893e-05, + "loss": 0.3904, + "step": 3953 + }, + { + "epoch": 1.66, + "learning_rate": 1.840252159353492e-05, + "loss": 0.3753, + "step": 3954 + }, + { + "epoch": 1.66, + "learning_rate": 1.840160222916788e-05, + "loss": 0.3811, + "step": 3955 + }, + { + "epoch": 1.66, + "learning_rate": 1.8400682623304243e-05, + "loss": 0.362, + "step": 3956 + }, + { + "epoch": 1.66, + "learning_rate": 1.839976277597044e-05, + "loss": 0.4012, + "step": 3957 + }, + { + "epoch": 1.66, + "learning_rate": 1.8398842687192906e-05, + "loss": 0.408, + "step": 3958 + }, + { + "epoch": 1.66, + "learning_rate": 1.839792235699809e-05, + "loss": 0.3913, + "step": 3959 + }, + { + "epoch": 1.66, + "learning_rate": 1.8397001785412455e-05, + "loss": 0.3546, + "step": 3960 + }, + { + "epoch": 1.66, + "learning_rate": 1.839608097246245e-05, + "loss": 0.3846, + "step": 3961 + }, + { + "epoch": 1.66, + "learning_rate": 1.839515991817455e-05, + "loss": 0.3486, + "step": 3962 + }, + { + "epoch": 1.66, + "learning_rate": 1.8394238622575228e-05, + "loss": 0.4618, + "step": 3963 + }, + { + "epoch": 1.66, + "learning_rate": 1.8393317085690964e-05, + "loss": 0.3483, + "step": 3964 + }, + { + "epoch": 1.66, + "learning_rate": 1.839239530754825e-05, + "loss": 0.4279, + "step": 3965 + }, + { + "epoch": 1.66, + "learning_rate": 1.839147328817358e-05, + "loss": 0.4616, + "step": 3966 + }, + { + "epoch": 1.66, + "learning_rate": 1.8390551027593456e-05, + "loss": 0.4158, + "step": 3967 + }, + { + "epoch": 1.66, + "learning_rate": 1.8389628525834385e-05, + "loss": 0.3667, + "step": 3968 + }, + { + "epoch": 1.66, + "learning_rate": 1.8388705782922893e-05, + "loss": 0.4405, + "step": 3969 + }, + { + "epoch": 1.66, + "learning_rate": 1.8387782798885495e-05, + "loss": 0.4028, + "step": 3970 + }, + { + "epoch": 1.66, + "learning_rate": 1.8386859573748724e-05, + "loss": 0.3949, + "step": 3971 + }, + { + "epoch": 1.66, + "learning_rate": 1.8385936107539115e-05, + "loss": 0.3959, + "step": 3972 + }, + { + "epoch": 1.66, + "learning_rate": 1.8385012400283218e-05, + "loss": 0.4073, + "step": 3973 + }, + { + "epoch": 1.66, + "learning_rate": 1.838408845200758e-05, + "loss": 0.3725, + "step": 3974 + }, + { + "epoch": 1.66, + "learning_rate": 1.8383164262738756e-05, + "loss": 0.3761, + "step": 3975 + }, + { + "epoch": 1.66, + "learning_rate": 1.838223983250331e-05, + "loss": 0.3953, + "step": 3976 + }, + { + "epoch": 1.67, + "learning_rate": 1.8381315161327827e-05, + "loss": 0.3925, + "step": 3977 + }, + { + "epoch": 1.67, + "learning_rate": 1.8380390249238876e-05, + "loss": 0.3745, + "step": 3978 + }, + { + "epoch": 1.67, + "learning_rate": 1.837946509626304e-05, + "loss": 0.3934, + "step": 3979 + }, + { + "epoch": 1.67, + "learning_rate": 1.8378539702426917e-05, + "loss": 0.348, + "step": 3980 + }, + { + "epoch": 1.67, + "learning_rate": 1.8377614067757106e-05, + "loss": 0.3714, + "step": 3981 + }, + { + "epoch": 1.67, + "learning_rate": 1.8376688192280213e-05, + "loss": 0.384, + "step": 3982 + }, + { + "epoch": 1.67, + "learning_rate": 1.837576207602285e-05, + "loss": 0.3941, + "step": 3983 + }, + { + "epoch": 1.67, + "learning_rate": 1.837483571901164e-05, + "loss": 0.3607, + "step": 3984 + }, + { + "epoch": 1.67, + "learning_rate": 1.837390912127321e-05, + "loss": 0.3494, + "step": 3985 + }, + { + "epoch": 1.67, + "learning_rate": 1.837298228283419e-05, + "loss": 0.3867, + "step": 3986 + }, + { + "epoch": 1.67, + "learning_rate": 1.837205520372123e-05, + "loss": 0.369, + "step": 3987 + }, + { + "epoch": 1.67, + "learning_rate": 1.8371127883960963e-05, + "loss": 0.3949, + "step": 3988 + }, + { + "epoch": 1.67, + "learning_rate": 1.837020032358006e-05, + "loss": 0.3878, + "step": 3989 + }, + { + "epoch": 1.67, + "learning_rate": 1.8369272522605177e-05, + "loss": 0.4191, + "step": 3990 + }, + { + "epoch": 1.67, + "learning_rate": 1.836834448106298e-05, + "loss": 0.3945, + "step": 3991 + }, + { + "epoch": 1.67, + "learning_rate": 1.836741619898015e-05, + "loss": 0.4298, + "step": 3992 + }, + { + "epoch": 1.67, + "learning_rate": 1.836648767638336e-05, + "loss": 0.3937, + "step": 3993 + }, + { + "epoch": 1.67, + "learning_rate": 1.8365558913299307e-05, + "loss": 0.4084, + "step": 3994 + }, + { + "epoch": 1.67, + "learning_rate": 1.836462990975469e-05, + "loss": 0.3711, + "step": 3995 + }, + { + "epoch": 1.67, + "learning_rate": 1.836370066577621e-05, + "loss": 0.3859, + "step": 3996 + }, + { + "epoch": 1.67, + "learning_rate": 1.836277118139057e-05, + "loss": 0.3787, + "step": 3997 + }, + { + "epoch": 1.67, + "learning_rate": 1.83618414566245e-05, + "loss": 0.387, + "step": 3998 + }, + { + "epoch": 1.67, + "learning_rate": 1.8360911491504714e-05, + "loss": 0.3551, + "step": 3999 + }, + { + "epoch": 1.67, + "learning_rate": 1.835998128605795e-05, + "loss": 0.409, + "step": 4000 + }, + { + "epoch": 1.68, + "learning_rate": 1.8359050840310936e-05, + "loss": 0.3925, + "step": 4001 + }, + { + "epoch": 1.68, + "learning_rate": 1.8358120154290427e-05, + "loss": 0.4395, + "step": 4002 + }, + { + "epoch": 1.68, + "learning_rate": 1.8357189228023174e-05, + "loss": 0.425, + "step": 4003 + }, + { + "epoch": 1.68, + "learning_rate": 1.835625806153593e-05, + "loss": 0.3778, + "step": 4004 + }, + { + "epoch": 1.68, + "learning_rate": 1.835532665485546e-05, + "loss": 0.3671, + "step": 4005 + }, + { + "epoch": 1.68, + "learning_rate": 1.8354395008008545e-05, + "loss": 0.4004, + "step": 4006 + }, + { + "epoch": 1.68, + "learning_rate": 1.835346312102196e-05, + "loss": 0.3547, + "step": 4007 + }, + { + "epoch": 1.68, + "learning_rate": 1.8352530993922483e-05, + "loss": 0.4045, + "step": 4008 + }, + { + "epoch": 1.68, + "learning_rate": 1.835159862673692e-05, + "loss": 0.3707, + "step": 4009 + }, + { + "epoch": 1.68, + "learning_rate": 1.8350666019492065e-05, + "loss": 0.3821, + "step": 4010 + }, + { + "epoch": 1.68, + "learning_rate": 1.8349733172214722e-05, + "loss": 0.4071, + "step": 4011 + }, + { + "epoch": 1.68, + "learning_rate": 1.8348800084931712e-05, + "loss": 0.4002, + "step": 4012 + }, + { + "epoch": 1.68, + "learning_rate": 1.834786675766985e-05, + "loss": 0.3747, + "step": 4013 + }, + { + "epoch": 1.68, + "learning_rate": 1.834693319045597e-05, + "loss": 0.4455, + "step": 4014 + }, + { + "epoch": 1.68, + "learning_rate": 1.8345999383316897e-05, + "loss": 0.3962, + "step": 4015 + }, + { + "epoch": 1.68, + "learning_rate": 1.8345065336279482e-05, + "loss": 0.3184, + "step": 4016 + }, + { + "epoch": 1.68, + "learning_rate": 1.8344131049370566e-05, + "loss": 0.3856, + "step": 4017 + }, + { + "epoch": 1.68, + "learning_rate": 1.834319652261701e-05, + "loss": 0.4466, + "step": 4018 + }, + { + "epoch": 1.68, + "learning_rate": 1.8342261756045673e-05, + "loss": 0.4317, + "step": 4019 + }, + { + "epoch": 1.68, + "learning_rate": 1.834132674968342e-05, + "loss": 0.3718, + "step": 4020 + }, + { + "epoch": 1.68, + "learning_rate": 1.834039150355714e-05, + "loss": 0.3773, + "step": 4021 + }, + { + "epoch": 1.68, + "learning_rate": 1.83394560176937e-05, + "loss": 0.3929, + "step": 4022 + }, + { + "epoch": 1.68, + "learning_rate": 1.8338520292120002e-05, + "loss": 0.4016, + "step": 4023 + }, + { + "epoch": 1.68, + "learning_rate": 1.8337584326862935e-05, + "loss": 0.4165, + "step": 4024 + }, + { + "epoch": 1.69, + "learning_rate": 1.8336648121949405e-05, + "loss": 0.3676, + "step": 4025 + }, + { + "epoch": 1.69, + "learning_rate": 1.8335711677406324e-05, + "loss": 0.3762, + "step": 4026 + }, + { + "epoch": 1.69, + "learning_rate": 1.833477499326061e-05, + "loss": 0.4556, + "step": 4027 + }, + { + "epoch": 1.69, + "learning_rate": 1.833383806953918e-05, + "loss": 0.3767, + "step": 4028 + }, + { + "epoch": 1.69, + "learning_rate": 1.8332900906268972e-05, + "loss": 0.3779, + "step": 4029 + }, + { + "epoch": 1.69, + "learning_rate": 1.8331963503476925e-05, + "loss": 0.3795, + "step": 4030 + }, + { + "epoch": 1.69, + "learning_rate": 1.8331025861189974e-05, + "loss": 0.4227, + "step": 4031 + }, + { + "epoch": 1.69, + "learning_rate": 1.8330087979435083e-05, + "loss": 0.3875, + "step": 4032 + }, + { + "epoch": 1.69, + "learning_rate": 1.8329149858239204e-05, + "loss": 0.4157, + "step": 4033 + }, + { + "epoch": 1.69, + "learning_rate": 1.8328211497629304e-05, + "loss": 0.4011, + "step": 4034 + }, + { + "epoch": 1.69, + "learning_rate": 1.832727289763235e-05, + "loss": 0.3993, + "step": 4035 + }, + { + "epoch": 1.69, + "learning_rate": 1.832633405827533e-05, + "loss": 0.3993, + "step": 4036 + }, + { + "epoch": 1.69, + "learning_rate": 1.8325394979585227e-05, + "loss": 0.4001, + "step": 4037 + }, + { + "epoch": 1.69, + "learning_rate": 1.8324455661589033e-05, + "loss": 0.3964, + "step": 4038 + }, + { + "epoch": 1.69, + "learning_rate": 1.8323516104313746e-05, + "loss": 0.424, + "step": 4039 + }, + { + "epoch": 1.69, + "learning_rate": 1.8322576307786376e-05, + "loss": 0.4006, + "step": 4040 + }, + { + "epoch": 1.69, + "learning_rate": 1.8321636272033934e-05, + "loss": 0.3782, + "step": 4041 + }, + { + "epoch": 1.69, + "learning_rate": 1.8320695997083443e-05, + "loss": 0.3963, + "step": 4042 + }, + { + "epoch": 1.69, + "learning_rate": 1.831975548296193e-05, + "loss": 0.3684, + "step": 4043 + }, + { + "epoch": 1.69, + "learning_rate": 1.8318814729696427e-05, + "loss": 0.4097, + "step": 4044 + }, + { + "epoch": 1.69, + "learning_rate": 1.8317873737313977e-05, + "loss": 0.3876, + "step": 4045 + }, + { + "epoch": 1.69, + "learning_rate": 1.831693250584163e-05, + "loss": 0.4022, + "step": 4046 + }, + { + "epoch": 1.69, + "learning_rate": 1.8315991035306435e-05, + "loss": 0.3946, + "step": 4047 + }, + { + "epoch": 1.69, + "learning_rate": 1.831504932573546e-05, + "loss": 0.4278, + "step": 4048 + }, + { + "epoch": 1.7, + "learning_rate": 1.831410737715577e-05, + "loss": 0.391, + "step": 4049 + }, + { + "epoch": 1.7, + "learning_rate": 1.8313165189594438e-05, + "loss": 0.3522, + "step": 4050 + }, + { + "epoch": 1.7, + "learning_rate": 1.8312222763078552e-05, + "loss": 0.3493, + "step": 4051 + }, + { + "epoch": 1.7, + "learning_rate": 1.83112800976352e-05, + "loss": 0.3808, + "step": 4052 + }, + { + "epoch": 1.7, + "learning_rate": 1.8310337193291473e-05, + "loss": 0.376, + "step": 4053 + }, + { + "epoch": 1.7, + "learning_rate": 1.8309394050074482e-05, + "loss": 0.3485, + "step": 4054 + }, + { + "epoch": 1.7, + "learning_rate": 1.8308450668011334e-05, + "loss": 0.3439, + "step": 4055 + }, + { + "epoch": 1.7, + "learning_rate": 1.830750704712914e-05, + "loss": 0.4055, + "step": 4056 + }, + { + "epoch": 1.7, + "learning_rate": 1.830656318745503e-05, + "loss": 0.4463, + "step": 4057 + }, + { + "epoch": 1.7, + "learning_rate": 1.830561908901613e-05, + "loss": 0.3713, + "step": 4058 + }, + { + "epoch": 1.7, + "learning_rate": 1.8304674751839583e-05, + "loss": 0.3569, + "step": 4059 + }, + { + "epoch": 1.7, + "learning_rate": 1.830373017595253e-05, + "loss": 0.3846, + "step": 4060 + }, + { + "epoch": 1.7, + "learning_rate": 1.8302785361382117e-05, + "loss": 0.4214, + "step": 4061 + }, + { + "epoch": 1.7, + "learning_rate": 1.8301840308155507e-05, + "loss": 0.3752, + "step": 4062 + }, + { + "epoch": 1.7, + "learning_rate": 1.8300895016299868e-05, + "loss": 0.4175, + "step": 4063 + }, + { + "epoch": 1.7, + "learning_rate": 1.8299949485842366e-05, + "loss": 0.4112, + "step": 4064 + }, + { + "epoch": 1.7, + "learning_rate": 1.829900371681018e-05, + "loss": 0.436, + "step": 4065 + }, + { + "epoch": 1.7, + "learning_rate": 1.8298057709230498e-05, + "loss": 0.4509, + "step": 4066 + }, + { + "epoch": 1.7, + "learning_rate": 1.8297111463130512e-05, + "loss": 0.4404, + "step": 4067 + }, + { + "epoch": 1.7, + "learning_rate": 1.8296164978537422e-05, + "loss": 0.3855, + "step": 4068 + }, + { + "epoch": 1.7, + "learning_rate": 1.8295218255478425e-05, + "loss": 0.3585, + "step": 4069 + }, + { + "epoch": 1.7, + "learning_rate": 1.8294271293980744e-05, + "loss": 0.4391, + "step": 4070 + }, + { + "epoch": 1.7, + "learning_rate": 1.8293324094071593e-05, + "loss": 0.34, + "step": 4071 + }, + { + "epoch": 1.7, + "learning_rate": 1.8292376655778202e-05, + "loss": 0.3956, + "step": 4072 + }, + { + "epoch": 1.71, + "learning_rate": 1.8291428979127802e-05, + "loss": 0.3659, + "step": 4073 + }, + { + "epoch": 1.71, + "learning_rate": 1.8290481064147637e-05, + "loss": 0.3756, + "step": 4074 + }, + { + "epoch": 1.71, + "learning_rate": 1.8289532910864943e-05, + "loss": 0.357, + "step": 4075 + }, + { + "epoch": 1.71, + "learning_rate": 1.8288584519306985e-05, + "loss": 0.3498, + "step": 4076 + }, + { + "epoch": 1.71, + "learning_rate": 1.8287635889501024e-05, + "loss": 0.3925, + "step": 4077 + }, + { + "epoch": 1.71, + "learning_rate": 1.828668702147432e-05, + "loss": 0.3581, + "step": 4078 + }, + { + "epoch": 1.71, + "learning_rate": 1.828573791525415e-05, + "loss": 0.372, + "step": 4079 + }, + { + "epoch": 1.71, + "learning_rate": 1.82847885708678e-05, + "loss": 0.388, + "step": 4080 + }, + { + "epoch": 1.71, + "learning_rate": 1.8283838988342553e-05, + "loss": 0.3954, + "step": 4081 + }, + { + "epoch": 1.71, + "learning_rate": 1.8282889167705704e-05, + "loss": 0.3438, + "step": 4082 + }, + { + "epoch": 1.71, + "learning_rate": 1.8281939108984557e-05, + "loss": 0.3651, + "step": 4083 + }, + { + "epoch": 1.71, + "learning_rate": 1.828098881220642e-05, + "loss": 0.3769, + "step": 4084 + }, + { + "epoch": 1.71, + "learning_rate": 1.8280038277398606e-05, + "loss": 0.39, + "step": 4085 + }, + { + "epoch": 1.71, + "learning_rate": 1.827908750458844e-05, + "loss": 0.3994, + "step": 4086 + }, + { + "epoch": 1.71, + "learning_rate": 1.8278136493803253e-05, + "loss": 0.3713, + "step": 4087 + }, + { + "epoch": 1.71, + "learning_rate": 1.8277185245070373e-05, + "loss": 0.4595, + "step": 4088 + }, + { + "epoch": 1.71, + "learning_rate": 1.8276233758417153e-05, + "loss": 0.4077, + "step": 4089 + }, + { + "epoch": 1.71, + "learning_rate": 1.827528203387094e-05, + "loss": 0.4482, + "step": 4090 + }, + { + "epoch": 1.71, + "learning_rate": 1.8274330071459082e-05, + "loss": 0.3704, + "step": 4091 + }, + { + "epoch": 1.71, + "learning_rate": 1.8273377871208952e-05, + "loss": 0.392, + "step": 4092 + }, + { + "epoch": 1.71, + "learning_rate": 1.8272425433147916e-05, + "loss": 0.3883, + "step": 4093 + }, + { + "epoch": 1.71, + "learning_rate": 1.8271472757303352e-05, + "loss": 0.3909, + "step": 4094 + }, + { + "epoch": 1.71, + "learning_rate": 1.827051984370264e-05, + "loss": 0.4504, + "step": 4095 + }, + { + "epoch": 1.71, + "learning_rate": 1.826956669237318e-05, + "loss": 0.3842, + "step": 4096 + }, + { + "epoch": 1.72, + "learning_rate": 1.8268613303342362e-05, + "loss": 0.3769, + "step": 4097 + }, + { + "epoch": 1.72, + "learning_rate": 1.826765967663759e-05, + "loss": 0.3996, + "step": 4098 + }, + { + "epoch": 1.72, + "learning_rate": 1.8266705812286276e-05, + "loss": 0.3751, + "step": 4099 + }, + { + "epoch": 1.72, + "learning_rate": 1.8265751710315843e-05, + "loss": 0.3925, + "step": 4100 + }, + { + "epoch": 1.72, + "learning_rate": 1.826479737075371e-05, + "loss": 0.3749, + "step": 4101 + }, + { + "epoch": 1.72, + "learning_rate": 1.8263842793627314e-05, + "loss": 0.3599, + "step": 4102 + }, + { + "epoch": 1.72, + "learning_rate": 1.8262887978964087e-05, + "loss": 0.4009, + "step": 4103 + }, + { + "epoch": 1.72, + "learning_rate": 1.8261932926791477e-05, + "loss": 0.3794, + "step": 4104 + }, + { + "epoch": 1.72, + "learning_rate": 1.8260977637136937e-05, + "loss": 0.3634, + "step": 4105 + }, + { + "epoch": 1.72, + "learning_rate": 1.8260022110027925e-05, + "loss": 0.3815, + "step": 4106 + }, + { + "epoch": 1.72, + "learning_rate": 1.8259066345491908e-05, + "loss": 0.447, + "step": 4107 + }, + { + "epoch": 1.72, + "learning_rate": 1.8258110343556358e-05, + "loss": 0.4141, + "step": 4108 + }, + { + "epoch": 1.72, + "learning_rate": 1.8257154104248752e-05, + "loss": 0.407, + "step": 4109 + }, + { + "epoch": 1.72, + "learning_rate": 1.8256197627596582e-05, + "loss": 0.395, + "step": 4110 + }, + { + "epoch": 1.72, + "learning_rate": 1.8255240913627335e-05, + "loss": 0.4496, + "step": 4111 + }, + { + "epoch": 1.72, + "learning_rate": 1.8254283962368517e-05, + "loss": 0.366, + "step": 4112 + }, + { + "epoch": 1.72, + "learning_rate": 1.8253326773847628e-05, + "loss": 0.3714, + "step": 4113 + }, + { + "epoch": 1.72, + "learning_rate": 1.8252369348092184e-05, + "loss": 0.3777, + "step": 4114 + }, + { + "epoch": 1.72, + "learning_rate": 1.825141168512971e-05, + "loss": 0.4085, + "step": 4115 + }, + { + "epoch": 1.72, + "learning_rate": 1.825045378498773e-05, + "loss": 0.4169, + "step": 4116 + }, + { + "epoch": 1.72, + "learning_rate": 1.824949564769377e-05, + "loss": 0.4316, + "step": 4117 + }, + { + "epoch": 1.72, + "learning_rate": 1.8248537273275385e-05, + "loss": 0.3995, + "step": 4118 + }, + { + "epoch": 1.72, + "learning_rate": 1.8247578661760114e-05, + "loss": 0.4057, + "step": 4119 + }, + { + "epoch": 1.72, + "learning_rate": 1.824661981317551e-05, + "loss": 0.389, + "step": 4120 + }, + { + "epoch": 1.73, + "learning_rate": 1.8245660727549142e-05, + "loss": 0.4068, + "step": 4121 + }, + { + "epoch": 1.73, + "learning_rate": 1.824470140490857e-05, + "loss": 0.4061, + "step": 4122 + }, + { + "epoch": 1.73, + "learning_rate": 1.8243741845281373e-05, + "loss": 0.4385, + "step": 4123 + }, + { + "epoch": 1.73, + "learning_rate": 1.824278204869513e-05, + "loss": 0.4081, + "step": 4124 + }, + { + "epoch": 1.73, + "learning_rate": 1.8241822015177433e-05, + "loss": 0.4129, + "step": 4125 + }, + { + "epoch": 1.73, + "learning_rate": 1.824086174475588e-05, + "loss": 0.3608, + "step": 4126 + }, + { + "epoch": 1.73, + "learning_rate": 1.8239901237458064e-05, + "loss": 0.4721, + "step": 4127 + }, + { + "epoch": 1.73, + "learning_rate": 1.8238940493311604e-05, + "loss": 0.3642, + "step": 4128 + }, + { + "epoch": 1.73, + "learning_rate": 1.8237979512344107e-05, + "loss": 0.3647, + "step": 4129 + }, + { + "epoch": 1.73, + "learning_rate": 1.8237018294583196e-05, + "loss": 0.4653, + "step": 4130 + }, + { + "epoch": 1.73, + "learning_rate": 1.8236056840056512e-05, + "loss": 0.4066, + "step": 4131 + }, + { + "epoch": 1.73, + "learning_rate": 1.8235095148791675e-05, + "loss": 0.3798, + "step": 4132 + }, + { + "epoch": 1.73, + "learning_rate": 1.823413322081634e-05, + "loss": 0.388, + "step": 4133 + }, + { + "epoch": 1.73, + "learning_rate": 1.823317105615815e-05, + "loss": 0.4043, + "step": 4134 + }, + { + "epoch": 1.73, + "learning_rate": 1.8232208654844766e-05, + "loss": 0.3316, + "step": 4135 + }, + { + "epoch": 1.73, + "learning_rate": 1.823124601690385e-05, + "loss": 0.3562, + "step": 4136 + }, + { + "epoch": 1.73, + "learning_rate": 1.8230283142363068e-05, + "loss": 0.4149, + "step": 4137 + }, + { + "epoch": 1.73, + "learning_rate": 1.8229320031250098e-05, + "loss": 0.3378, + "step": 4138 + }, + { + "epoch": 1.73, + "learning_rate": 1.822835668359263e-05, + "loss": 0.4156, + "step": 4139 + }, + { + "epoch": 1.73, + "learning_rate": 1.8227393099418353e-05, + "loss": 0.3673, + "step": 4140 + }, + { + "epoch": 1.73, + "learning_rate": 1.8226429278754963e-05, + "loss": 0.4593, + "step": 4141 + }, + { + "epoch": 1.73, + "learning_rate": 1.822546522163016e-05, + "loss": 0.3879, + "step": 4142 + }, + { + "epoch": 1.73, + "learning_rate": 1.822450092807166e-05, + "loss": 0.3846, + "step": 4143 + }, + { + "epoch": 1.73, + "learning_rate": 1.8223536398107177e-05, + "loss": 0.3752, + "step": 4144 + }, + { + "epoch": 1.74, + "learning_rate": 1.822257163176444e-05, + "loss": 0.4094, + "step": 4145 + }, + { + "epoch": 1.74, + "learning_rate": 1.8221606629071175e-05, + "loss": 0.3852, + "step": 4146 + }, + { + "epoch": 1.74, + "learning_rate": 1.8220641390055127e-05, + "loss": 0.3741, + "step": 4147 + }, + { + "epoch": 1.74, + "learning_rate": 1.8219675914744035e-05, + "loss": 0.4307, + "step": 4148 + }, + { + "epoch": 1.74, + "learning_rate": 1.8218710203165653e-05, + "loss": 0.3892, + "step": 4149 + }, + { + "epoch": 1.74, + "learning_rate": 1.821774425534774e-05, + "loss": 0.3891, + "step": 4150 + }, + { + "epoch": 1.74, + "learning_rate": 1.821677807131806e-05, + "loss": 0.389, + "step": 4151 + }, + { + "epoch": 1.74, + "learning_rate": 1.8215811651104383e-05, + "loss": 0.3828, + "step": 4152 + }, + { + "epoch": 1.74, + "learning_rate": 1.8214844994734498e-05, + "loss": 0.3996, + "step": 4153 + }, + { + "epoch": 1.74, + "learning_rate": 1.8213878102236177e-05, + "loss": 0.4122, + "step": 4154 + }, + { + "epoch": 1.74, + "learning_rate": 1.821291097363722e-05, + "loss": 0.3949, + "step": 4155 + }, + { + "epoch": 1.74, + "learning_rate": 1.821194360896543e-05, + "loss": 0.3584, + "step": 4156 + }, + { + "epoch": 1.74, + "learning_rate": 1.8210976008248603e-05, + "loss": 0.3853, + "step": 4157 + }, + { + "epoch": 1.74, + "learning_rate": 1.821000817151456e-05, + "loss": 0.4165, + "step": 4158 + }, + { + "epoch": 1.74, + "learning_rate": 1.820904009879111e-05, + "loss": 0.4278, + "step": 4159 + }, + { + "epoch": 1.74, + "learning_rate": 1.8208071790106096e-05, + "loss": 0.3944, + "step": 4160 + }, + { + "epoch": 1.74, + "learning_rate": 1.8207103245487335e-05, + "loss": 0.3955, + "step": 4161 + }, + { + "epoch": 1.74, + "learning_rate": 1.8206134464962678e-05, + "loss": 0.3999, + "step": 4162 + }, + { + "epoch": 1.74, + "learning_rate": 1.820516544855997e-05, + "loss": 0.4308, + "step": 4163 + }, + { + "epoch": 1.74, + "learning_rate": 1.8204196196307058e-05, + "loss": 0.3431, + "step": 4164 + }, + { + "epoch": 1.74, + "learning_rate": 1.8203226708231806e-05, + "loss": 0.3637, + "step": 4165 + }, + { + "epoch": 1.74, + "learning_rate": 1.8202256984362087e-05, + "loss": 0.3876, + "step": 4166 + }, + { + "epoch": 1.74, + "learning_rate": 1.8201287024725765e-05, + "loss": 0.3851, + "step": 4167 + }, + { + "epoch": 1.75, + "learning_rate": 1.8200316829350728e-05, + "loss": 0.3599, + "step": 4168 + }, + { + "epoch": 1.75, + "learning_rate": 1.8199346398264856e-05, + "loss": 0.4277, + "step": 4169 + }, + { + "epoch": 1.75, + "learning_rate": 1.8198375731496054e-05, + "loss": 0.3442, + "step": 4170 + }, + { + "epoch": 1.75, + "learning_rate": 1.8197404829072214e-05, + "loss": 0.4514, + "step": 4171 + }, + { + "epoch": 1.75, + "learning_rate": 1.8196433691021243e-05, + "loss": 0.4127, + "step": 4172 + }, + { + "epoch": 1.75, + "learning_rate": 1.8195462317371063e-05, + "loss": 0.4018, + "step": 4173 + }, + { + "epoch": 1.75, + "learning_rate": 1.819449070814959e-05, + "loss": 0.4155, + "step": 4174 + }, + { + "epoch": 1.75, + "learning_rate": 1.8193518863384753e-05, + "loss": 0.3667, + "step": 4175 + }, + { + "epoch": 1.75, + "learning_rate": 1.819254678310449e-05, + "loss": 0.3539, + "step": 4176 + }, + { + "epoch": 1.75, + "learning_rate": 1.8191574467336737e-05, + "loss": 0.4054, + "step": 4177 + }, + { + "epoch": 1.75, + "learning_rate": 1.8190601916109444e-05, + "loss": 0.4086, + "step": 4178 + }, + { + "epoch": 1.75, + "learning_rate": 1.818962912945057e-05, + "loss": 0.3798, + "step": 4179 + }, + { + "epoch": 1.75, + "learning_rate": 1.818865610738807e-05, + "loss": 0.4052, + "step": 4180 + }, + { + "epoch": 1.75, + "learning_rate": 1.818768284994992e-05, + "loss": 0.3307, + "step": 4181 + }, + { + "epoch": 1.75, + "learning_rate": 1.818670935716409e-05, + "loss": 0.3998, + "step": 4182 + }, + { + "epoch": 1.75, + "learning_rate": 1.8185735629058566e-05, + "loss": 0.3693, + "step": 4183 + }, + { + "epoch": 1.75, + "learning_rate": 1.8184761665661335e-05, + "loss": 0.3871, + "step": 4184 + }, + { + "epoch": 1.75, + "learning_rate": 1.8183787467000392e-05, + "loss": 0.3721, + "step": 4185 + }, + { + "epoch": 1.75, + "learning_rate": 1.8182813033103742e-05, + "loss": 0.3355, + "step": 4186 + }, + { + "epoch": 1.75, + "learning_rate": 1.8181838363999392e-05, + "loss": 0.4374, + "step": 4187 + }, + { + "epoch": 1.75, + "learning_rate": 1.8180863459715358e-05, + "loss": 0.3902, + "step": 4188 + }, + { + "epoch": 1.75, + "learning_rate": 1.8179888320279663e-05, + "loss": 0.4099, + "step": 4189 + }, + { + "epoch": 1.75, + "learning_rate": 1.817891294572034e-05, + "loss": 0.3684, + "step": 4190 + }, + { + "epoch": 1.75, + "learning_rate": 1.817793733606542e-05, + "loss": 0.4386, + "step": 4191 + }, + { + "epoch": 1.76, + "learning_rate": 1.8176961491342947e-05, + "loss": 0.3864, + "step": 4192 + }, + { + "epoch": 1.76, + "learning_rate": 1.8175985411580975e-05, + "loss": 0.4155, + "step": 4193 + }, + { + "epoch": 1.76, + "learning_rate": 1.8175009096807557e-05, + "loss": 0.4204, + "step": 4194 + }, + { + "epoch": 1.76, + "learning_rate": 1.8174032547050754e-05, + "loss": 0.382, + "step": 4195 + }, + { + "epoch": 1.76, + "learning_rate": 1.8173055762338642e-05, + "loss": 0.4276, + "step": 4196 + }, + { + "epoch": 1.76, + "learning_rate": 1.8172078742699293e-05, + "loss": 0.3971, + "step": 4197 + }, + { + "epoch": 1.76, + "learning_rate": 1.817110148816079e-05, + "loss": 0.3628, + "step": 4198 + }, + { + "epoch": 1.76, + "learning_rate": 1.817012399875123e-05, + "loss": 0.4342, + "step": 4199 + }, + { + "epoch": 1.76, + "learning_rate": 1.8169146274498707e-05, + "loss": 0.3692, + "step": 4200 + }, + { + "epoch": 1.76, + "learning_rate": 1.8168168315431318e-05, + "loss": 0.3981, + "step": 4201 + }, + { + "epoch": 1.76, + "learning_rate": 1.8167190121577184e-05, + "loss": 0.4388, + "step": 4202 + }, + { + "epoch": 1.76, + "learning_rate": 1.8166211692964413e-05, + "loss": 0.3119, + "step": 4203 + }, + { + "epoch": 1.76, + "learning_rate": 1.8165233029621136e-05, + "loss": 0.3513, + "step": 4204 + }, + { + "epoch": 1.76, + "learning_rate": 1.816425413157548e-05, + "loss": 0.4071, + "step": 4205 + }, + { + "epoch": 1.76, + "learning_rate": 1.8163274998855585e-05, + "loss": 0.4342, + "step": 4206 + }, + { + "epoch": 1.76, + "learning_rate": 1.8162295631489595e-05, + "loss": 0.418, + "step": 4207 + }, + { + "epoch": 1.76, + "learning_rate": 1.8161316029505656e-05, + "loss": 0.3763, + "step": 4208 + }, + { + "epoch": 1.76, + "learning_rate": 1.8160336192931935e-05, + "loss": 0.4215, + "step": 4209 + }, + { + "epoch": 1.76, + "learning_rate": 1.8159356121796586e-05, + "loss": 0.3893, + "step": 4210 + }, + { + "epoch": 1.76, + "learning_rate": 1.8158375816127792e-05, + "loss": 0.4154, + "step": 4211 + }, + { + "epoch": 1.76, + "learning_rate": 1.8157395275953722e-05, + "loss": 0.375, + "step": 4212 + }, + { + "epoch": 1.76, + "learning_rate": 1.8156414501302565e-05, + "loss": 0.3681, + "step": 4213 + }, + { + "epoch": 1.76, + "learning_rate": 1.8155433492202507e-05, + "loss": 0.3854, + "step": 4214 + }, + { + "epoch": 1.76, + "learning_rate": 1.8154452248681754e-05, + "loss": 0.3947, + "step": 4215 + }, + { + "epoch": 1.77, + "learning_rate": 1.8153470770768506e-05, + "loss": 0.3914, + "step": 4216 + }, + { + "epoch": 1.77, + "learning_rate": 1.8152489058490975e-05, + "loss": 0.3556, + "step": 4217 + }, + { + "epoch": 1.77, + "learning_rate": 1.8151507111877382e-05, + "loss": 0.3644, + "step": 4218 + }, + { + "epoch": 1.77, + "learning_rate": 1.815052493095595e-05, + "loss": 0.3726, + "step": 4219 + }, + { + "epoch": 1.77, + "learning_rate": 1.814954251575491e-05, + "loss": 0.4142, + "step": 4220 + }, + { + "epoch": 1.77, + "learning_rate": 1.8148559866302505e-05, + "loss": 0.3856, + "step": 4221 + }, + { + "epoch": 1.77, + "learning_rate": 1.8147576982626974e-05, + "loss": 0.3581, + "step": 4222 + }, + { + "epoch": 1.77, + "learning_rate": 1.8146593864756577e-05, + "loss": 0.4268, + "step": 4223 + }, + { + "epoch": 1.77, + "learning_rate": 1.8145610512719565e-05, + "loss": 0.3792, + "step": 4224 + }, + { + "epoch": 1.77, + "learning_rate": 1.8144626926544208e-05, + "loss": 0.3814, + "step": 4225 + }, + { + "epoch": 1.77, + "learning_rate": 1.814364310625878e-05, + "loss": 0.3749, + "step": 4226 + }, + { + "epoch": 1.77, + "learning_rate": 1.8142659051891552e-05, + "loss": 0.3667, + "step": 4227 + }, + { + "epoch": 1.77, + "learning_rate": 1.814167476347082e-05, + "loss": 0.3915, + "step": 4228 + }, + { + "epoch": 1.77, + "learning_rate": 1.8140690241024872e-05, + "loss": 0.4039, + "step": 4229 + }, + { + "epoch": 1.77, + "learning_rate": 1.8139705484582005e-05, + "loss": 0.3712, + "step": 4230 + }, + { + "epoch": 1.77, + "learning_rate": 1.813872049417053e-05, + "loss": 0.3634, + "step": 4231 + }, + { + "epoch": 1.77, + "learning_rate": 1.813773526981875e-05, + "loss": 0.4128, + "step": 4232 + }, + { + "epoch": 1.77, + "learning_rate": 1.8136749811554995e-05, + "loss": 0.3902, + "step": 4233 + }, + { + "epoch": 1.77, + "learning_rate": 1.8135764119407585e-05, + "loss": 0.3534, + "step": 4234 + }, + { + "epoch": 1.77, + "learning_rate": 1.8134778193404857e-05, + "loss": 0.3815, + "step": 4235 + }, + { + "epoch": 1.77, + "learning_rate": 1.813379203357515e-05, + "loss": 0.3744, + "step": 4236 + }, + { + "epoch": 1.77, + "learning_rate": 1.8132805639946806e-05, + "loss": 0.3964, + "step": 4237 + }, + { + "epoch": 1.77, + "learning_rate": 1.8131819012548183e-05, + "loss": 0.3992, + "step": 4238 + }, + { + "epoch": 1.77, + "learning_rate": 1.8130832151407634e-05, + "loss": 0.3965, + "step": 4239 + }, + { + "epoch": 1.78, + "learning_rate": 1.8129845056553534e-05, + "loss": 0.4237, + "step": 4240 + }, + { + "epoch": 1.78, + "learning_rate": 1.812885772801425e-05, + "loss": 0.3852, + "step": 4241 + }, + { + "epoch": 1.78, + "learning_rate": 1.8127870165818166e-05, + "loss": 0.4264, + "step": 4242 + }, + { + "epoch": 1.78, + "learning_rate": 1.812688236999367e-05, + "loss": 0.4305, + "step": 4243 + }, + { + "epoch": 1.78, + "learning_rate": 1.8125894340569145e-05, + "loss": 0.3854, + "step": 4244 + }, + { + "epoch": 1.78, + "learning_rate": 1.8124906077573e-05, + "loss": 0.3823, + "step": 4245 + }, + { + "epoch": 1.78, + "learning_rate": 1.8123917581033642e-05, + "loss": 0.3893, + "step": 4246 + }, + { + "epoch": 1.78, + "learning_rate": 1.8122928850979483e-05, + "loss": 0.4558, + "step": 4247 + }, + { + "epoch": 1.78, + "learning_rate": 1.8121939887438936e-05, + "loss": 0.3923, + "step": 4248 + }, + { + "epoch": 1.78, + "learning_rate": 1.812095069044044e-05, + "loss": 0.37, + "step": 4249 + }, + { + "epoch": 1.78, + "learning_rate": 1.8119961260012422e-05, + "loss": 0.423, + "step": 4250 + }, + { + "epoch": 1.78, + "learning_rate": 1.811897159618332e-05, + "loss": 0.3789, + "step": 4251 + }, + { + "epoch": 1.78, + "learning_rate": 1.811798169898159e-05, + "loss": 0.3585, + "step": 4252 + }, + { + "epoch": 1.78, + "learning_rate": 1.8116991568435674e-05, + "loss": 0.3785, + "step": 4253 + }, + { + "epoch": 1.78, + "learning_rate": 1.8116001204574042e-05, + "loss": 0.4106, + "step": 4254 + }, + { + "epoch": 1.78, + "learning_rate": 1.8115010607425154e-05, + "loss": 0.4085, + "step": 4255 + }, + { + "epoch": 1.78, + "learning_rate": 1.811401977701749e-05, + "loss": 0.45, + "step": 4256 + }, + { + "epoch": 1.78, + "learning_rate": 1.8113028713379524e-05, + "loss": 0.4488, + "step": 4257 + }, + { + "epoch": 1.78, + "learning_rate": 1.811203741653975e-05, + "loss": 0.409, + "step": 4258 + }, + { + "epoch": 1.78, + "learning_rate": 1.811104588652666e-05, + "loss": 0.3769, + "step": 4259 + }, + { + "epoch": 1.78, + "learning_rate": 1.8110054123368755e-05, + "loss": 0.3857, + "step": 4260 + }, + { + "epoch": 1.78, + "learning_rate": 1.8109062127094538e-05, + "loss": 0.4303, + "step": 4261 + }, + { + "epoch": 1.78, + "learning_rate": 1.8108069897732527e-05, + "loss": 0.4154, + "step": 4262 + }, + { + "epoch": 1.78, + "learning_rate": 1.810707743531124e-05, + "loss": 0.3739, + "step": 4263 + }, + { + "epoch": 1.79, + "learning_rate": 1.8106084739859206e-05, + "loss": 0.3749, + "step": 4264 + }, + { + "epoch": 1.79, + "learning_rate": 1.8105091811404962e-05, + "loss": 0.4192, + "step": 4265 + }, + { + "epoch": 1.79, + "learning_rate": 1.8104098649977044e-05, + "loss": 0.4326, + "step": 4266 + }, + { + "epoch": 1.79, + "learning_rate": 1.8103105255604003e-05, + "loss": 0.4022, + "step": 4267 + }, + { + "epoch": 1.79, + "learning_rate": 1.8102111628314394e-05, + "loss": 0.4064, + "step": 4268 + }, + { + "epoch": 1.79, + "learning_rate": 1.810111776813677e-05, + "loss": 0.3597, + "step": 4269 + }, + { + "epoch": 1.79, + "learning_rate": 1.810012367509971e-05, + "loss": 0.397, + "step": 4270 + }, + { + "epoch": 1.79, + "learning_rate": 1.8099129349231783e-05, + "loss": 0.4079, + "step": 4271 + }, + { + "epoch": 1.79, + "learning_rate": 1.8098134790561567e-05, + "loss": 0.3914, + "step": 4272 + }, + { + "epoch": 1.79, + "learning_rate": 1.8097139999117655e-05, + "loss": 0.3927, + "step": 4273 + }, + { + "epoch": 1.79, + "learning_rate": 1.8096144974928635e-05, + "loss": 0.4111, + "step": 4274 + }, + { + "epoch": 1.79, + "learning_rate": 1.8095149718023114e-05, + "loss": 0.4026, + "step": 4275 + }, + { + "epoch": 1.79, + "learning_rate": 1.80941542284297e-05, + "loss": 0.3648, + "step": 4276 + }, + { + "epoch": 1.79, + "learning_rate": 1.8093158506177003e-05, + "loss": 0.3546, + "step": 4277 + }, + { + "epoch": 1.79, + "learning_rate": 1.8092162551293646e-05, + "loss": 0.3978, + "step": 4278 + }, + { + "epoch": 1.79, + "learning_rate": 1.8091166363808258e-05, + "loss": 0.3781, + "step": 4279 + }, + { + "epoch": 1.79, + "learning_rate": 1.8090169943749477e-05, + "loss": 0.3416, + "step": 4280 + }, + { + "epoch": 1.79, + "learning_rate": 1.8089173291145935e-05, + "loss": 0.3681, + "step": 4281 + }, + { + "epoch": 1.79, + "learning_rate": 1.808817640602629e-05, + "loss": 0.402, + "step": 4282 + }, + { + "epoch": 1.79, + "learning_rate": 1.8087179288419182e-05, + "loss": 0.4024, + "step": 4283 + }, + { + "epoch": 1.79, + "learning_rate": 1.808618193835329e-05, + "loss": 0.3581, + "step": 4284 + }, + { + "epoch": 1.79, + "learning_rate": 1.808518435585727e-05, + "loss": 0.4094, + "step": 4285 + }, + { + "epoch": 1.79, + "learning_rate": 1.80841865409598e-05, + "loss": 0.3766, + "step": 4286 + }, + { + "epoch": 1.79, + "learning_rate": 1.808318849368957e-05, + "loss": 0.4021, + "step": 4287 + }, + { + "epoch": 1.8, + "learning_rate": 1.8082190214075252e-05, + "loss": 0.3986, + "step": 4288 + }, + { + "epoch": 1.8, + "learning_rate": 1.8081191702145547e-05, + "loss": 0.397, + "step": 4289 + }, + { + "epoch": 1.8, + "learning_rate": 1.8080192957929162e-05, + "loss": 0.3872, + "step": 4290 + }, + { + "epoch": 1.8, + "learning_rate": 1.8079193981454798e-05, + "loss": 0.3809, + "step": 4291 + }, + { + "epoch": 1.8, + "learning_rate": 1.8078194772751173e-05, + "loss": 0.4337, + "step": 4292 + }, + { + "epoch": 1.8, + "learning_rate": 1.8077195331847007e-05, + "loss": 0.4027, + "step": 4293 + }, + { + "epoch": 1.8, + "learning_rate": 1.8076195658771033e-05, + "loss": 0.3748, + "step": 4294 + }, + { + "epoch": 1.8, + "learning_rate": 1.8075195753551974e-05, + "loss": 0.3974, + "step": 4295 + }, + { + "epoch": 1.8, + "learning_rate": 1.8074195616218583e-05, + "loss": 0.3615, + "step": 4296 + }, + { + "epoch": 1.8, + "learning_rate": 1.8073195246799605e-05, + "loss": 0.375, + "step": 4297 + }, + { + "epoch": 1.8, + "learning_rate": 1.8072194645323794e-05, + "loss": 0.449, + "step": 4298 + }, + { + "epoch": 1.8, + "learning_rate": 1.8071193811819912e-05, + "loss": 0.3788, + "step": 4299 + }, + { + "epoch": 1.8, + "learning_rate": 1.807019274631672e-05, + "loss": 0.3921, + "step": 4300 + }, + { + "epoch": 1.8, + "learning_rate": 1.8069191448843005e-05, + "loss": 0.3985, + "step": 4301 + }, + { + "epoch": 1.8, + "learning_rate": 1.806818991942754e-05, + "loss": 0.3922, + "step": 4302 + }, + { + "epoch": 1.8, + "learning_rate": 1.8067188158099112e-05, + "loss": 0.367, + "step": 4303 + }, + { + "epoch": 1.8, + "learning_rate": 1.8066186164886525e-05, + "loss": 0.4072, + "step": 4304 + }, + { + "epoch": 1.8, + "learning_rate": 1.806518393981857e-05, + "loss": 0.3879, + "step": 4305 + }, + { + "epoch": 1.8, + "learning_rate": 1.8064181482924065e-05, + "loss": 0.377, + "step": 4306 + }, + { + "epoch": 1.8, + "learning_rate": 1.8063178794231814e-05, + "loss": 0.4025, + "step": 4307 + }, + { + "epoch": 1.8, + "learning_rate": 1.8062175873770643e-05, + "loss": 0.3922, + "step": 4308 + }, + { + "epoch": 1.8, + "learning_rate": 1.8061172721569383e-05, + "loss": 0.4284, + "step": 4309 + }, + { + "epoch": 1.8, + "learning_rate": 1.8060169337656866e-05, + "loss": 0.4754, + "step": 4310 + }, + { + "epoch": 1.8, + "learning_rate": 1.805916572206194e-05, + "loss": 0.3792, + "step": 4311 + }, + { + "epoch": 1.81, + "learning_rate": 1.805816187481344e-05, + "loss": 0.374, + "step": 4312 + }, + { + "epoch": 1.81, + "learning_rate": 1.8057157795940228e-05, + "loss": 0.4356, + "step": 4313 + }, + { + "epoch": 1.81, + "learning_rate": 1.8056153485471167e-05, + "loss": 0.3891, + "step": 4314 + }, + { + "epoch": 1.81, + "learning_rate": 1.805514894343512e-05, + "loss": 0.3978, + "step": 4315 + }, + { + "epoch": 1.81, + "learning_rate": 1.805414416986097e-05, + "loss": 0.411, + "step": 4316 + }, + { + "epoch": 1.81, + "learning_rate": 1.8053139164777587e-05, + "loss": 0.3439, + "step": 4317 + }, + { + "epoch": 1.81, + "learning_rate": 1.805213392821387e-05, + "loss": 0.42, + "step": 4318 + }, + { + "epoch": 1.81, + "learning_rate": 1.8051128460198703e-05, + "loss": 0.3687, + "step": 4319 + }, + { + "epoch": 1.81, + "learning_rate": 1.8050122760760994e-05, + "loss": 0.3926, + "step": 4320 + }, + { + "epoch": 1.81, + "learning_rate": 1.8049116829929655e-05, + "loss": 0.3792, + "step": 4321 + }, + { + "epoch": 1.81, + "learning_rate": 1.804811066773359e-05, + "loss": 0.3436, + "step": 4322 + }, + { + "epoch": 1.81, + "learning_rate": 1.804710427420173e-05, + "loss": 0.386, + "step": 4323 + }, + { + "epoch": 1.81, + "learning_rate": 1.8046097649362994e-05, + "loss": 0.3816, + "step": 4324 + }, + { + "epoch": 1.81, + "learning_rate": 1.804509079324632e-05, + "loss": 0.3676, + "step": 4325 + }, + { + "epoch": 1.81, + "learning_rate": 1.804408370588066e-05, + "loss": 0.3695, + "step": 4326 + }, + { + "epoch": 1.81, + "learning_rate": 1.8043076387294944e-05, + "loss": 0.4537, + "step": 4327 + }, + { + "epoch": 1.81, + "learning_rate": 1.8042068837518135e-05, + "loss": 0.3724, + "step": 4328 + }, + { + "epoch": 1.81, + "learning_rate": 1.8041061056579194e-05, + "loss": 0.3988, + "step": 4329 + }, + { + "epoch": 1.81, + "learning_rate": 1.8040053044507087e-05, + "loss": 0.3737, + "step": 4330 + }, + { + "epoch": 1.81, + "learning_rate": 1.8039044801330795e-05, + "loss": 0.4418, + "step": 4331 + }, + { + "epoch": 1.81, + "learning_rate": 1.803803632707929e-05, + "loss": 0.3556, + "step": 4332 + }, + { + "epoch": 1.81, + "learning_rate": 1.8037027621781568e-05, + "loss": 0.3917, + "step": 4333 + }, + { + "epoch": 1.81, + "learning_rate": 1.8036018685466614e-05, + "loss": 0.4307, + "step": 4334 + }, + { + "epoch": 1.81, + "learning_rate": 1.8035009518163434e-05, + "loss": 0.4482, + "step": 4335 + }, + { + "epoch": 1.82, + "learning_rate": 1.8034000119901038e-05, + "loss": 0.4407, + "step": 4336 + }, + { + "epoch": 1.82, + "learning_rate": 1.8032990490708434e-05, + "loss": 0.3863, + "step": 4337 + }, + { + "epoch": 1.82, + "learning_rate": 1.8031980630614652e-05, + "loss": 0.3908, + "step": 4338 + }, + { + "epoch": 1.82, + "learning_rate": 1.803097053964871e-05, + "loss": 0.3528, + "step": 4339 + }, + { + "epoch": 1.82, + "learning_rate": 1.8029960217839648e-05, + "loss": 0.3768, + "step": 4340 + }, + { + "epoch": 1.82, + "learning_rate": 1.8028949665216505e-05, + "loss": 0.4015, + "step": 4341 + }, + { + "epoch": 1.82, + "learning_rate": 1.802793888180833e-05, + "loss": 0.4411, + "step": 4342 + }, + { + "epoch": 1.82, + "learning_rate": 1.8026927867644175e-05, + "loss": 0.4535, + "step": 4343 + }, + { + "epoch": 1.82, + "learning_rate": 1.80259166227531e-05, + "loss": 0.3384, + "step": 4344 + }, + { + "epoch": 1.82, + "learning_rate": 1.8024905147164176e-05, + "loss": 0.3852, + "step": 4345 + }, + { + "epoch": 1.82, + "learning_rate": 1.8023893440906476e-05, + "loss": 0.3559, + "step": 4346 + }, + { + "epoch": 1.82, + "learning_rate": 1.8022881504009074e-05, + "loss": 0.4396, + "step": 4347 + }, + { + "epoch": 1.82, + "learning_rate": 1.8021869336501066e-05, + "loss": 0.3677, + "step": 4348 + }, + { + "epoch": 1.82, + "learning_rate": 1.802085693841154e-05, + "loss": 0.3632, + "step": 4349 + }, + { + "epoch": 1.82, + "learning_rate": 1.8019844309769602e-05, + "loss": 0.4229, + "step": 4350 + }, + { + "epoch": 1.82, + "learning_rate": 1.8018831450604356e-05, + "loss": 0.4812, + "step": 4351 + }, + { + "epoch": 1.82, + "learning_rate": 1.801781836094491e-05, + "loss": 0.387, + "step": 4352 + }, + { + "epoch": 1.82, + "learning_rate": 1.80168050408204e-05, + "loss": 0.3763, + "step": 4353 + }, + { + "epoch": 1.82, + "learning_rate": 1.8015791490259932e-05, + "loss": 0.3933, + "step": 4354 + }, + { + "epoch": 1.82, + "learning_rate": 1.801477770929266e-05, + "loss": 0.4103, + "step": 4355 + }, + { + "epoch": 1.82, + "learning_rate": 1.801376369794771e-05, + "loss": 0.4138, + "step": 4356 + }, + { + "epoch": 1.82, + "learning_rate": 1.8012749456254238e-05, + "loss": 0.4028, + "step": 4357 + }, + { + "epoch": 1.82, + "learning_rate": 1.801173498424139e-05, + "loss": 0.3912, + "step": 4358 + }, + { + "epoch": 1.82, + "learning_rate": 1.8010720281938327e-05, + "loss": 0.429, + "step": 4359 + }, + { + "epoch": 1.83, + "learning_rate": 1.800970534937422e-05, + "loss": 0.3638, + "step": 4360 + }, + { + "epoch": 1.83, + "learning_rate": 1.8008690186578243e-05, + "loss": 0.4019, + "step": 4361 + }, + { + "epoch": 1.83, + "learning_rate": 1.800767479357957e-05, + "loss": 0.3711, + "step": 4362 + }, + { + "epoch": 1.83, + "learning_rate": 1.8006659170407396e-05, + "loss": 0.3859, + "step": 4363 + }, + { + "epoch": 1.83, + "learning_rate": 1.80056433170909e-05, + "loss": 0.441, + "step": 4364 + }, + { + "epoch": 1.83, + "learning_rate": 1.80046272336593e-05, + "loss": 0.375, + "step": 4365 + }, + { + "epoch": 1.83, + "learning_rate": 1.800361092014179e-05, + "loss": 0.4191, + "step": 4366 + }, + { + "epoch": 1.83, + "learning_rate": 1.8002594376567585e-05, + "loss": 0.3702, + "step": 4367 + }, + { + "epoch": 1.83, + "learning_rate": 1.8001577602965907e-05, + "loss": 0.4165, + "step": 4368 + }, + { + "epoch": 1.83, + "learning_rate": 1.800056059936598e-05, + "loss": 0.4246, + "step": 4369 + }, + { + "epoch": 1.83, + "learning_rate": 1.7999543365797043e-05, + "loss": 0.4047, + "step": 4370 + }, + { + "epoch": 1.83, + "learning_rate": 1.7998525902288327e-05, + "loss": 0.3736, + "step": 4371 + }, + { + "epoch": 1.83, + "learning_rate": 1.7997508208869082e-05, + "loss": 0.4108, + "step": 4372 + }, + { + "epoch": 1.83, + "learning_rate": 1.799649028556856e-05, + "loss": 0.3926, + "step": 4373 + }, + { + "epoch": 1.83, + "learning_rate": 1.799547213241602e-05, + "loss": 0.403, + "step": 4374 + }, + { + "epoch": 1.83, + "learning_rate": 1.799445374944073e-05, + "loss": 0.3477, + "step": 4375 + }, + { + "epoch": 1.83, + "learning_rate": 1.799343513667196e-05, + "loss": 0.4046, + "step": 4376 + }, + { + "epoch": 1.83, + "learning_rate": 1.7992416294138992e-05, + "loss": 0.3881, + "step": 4377 + }, + { + "epoch": 1.83, + "learning_rate": 1.799139722187111e-05, + "loss": 0.4091, + "step": 4378 + }, + { + "epoch": 1.83, + "learning_rate": 1.7990377919897605e-05, + "loss": 0.399, + "step": 4379 + }, + { + "epoch": 1.83, + "learning_rate": 1.7989358388247776e-05, + "loss": 0.4123, + "step": 4380 + }, + { + "epoch": 1.83, + "learning_rate": 1.798833862695093e-05, + "loss": 0.4001, + "step": 4381 + }, + { + "epoch": 1.83, + "learning_rate": 1.7987318636036382e-05, + "loss": 0.3674, + "step": 4382 + }, + { + "epoch": 1.84, + "learning_rate": 1.7986298415533444e-05, + "loss": 0.3603, + "step": 4383 + }, + { + "epoch": 1.84, + "learning_rate": 1.798527796547145e-05, + "loss": 0.4306, + "step": 4384 + }, + { + "epoch": 1.84, + "learning_rate": 1.7984257285879724e-05, + "loss": 0.3837, + "step": 4385 + }, + { + "epoch": 1.84, + "learning_rate": 1.7983236376787607e-05, + "loss": 0.4104, + "step": 4386 + }, + { + "epoch": 1.84, + "learning_rate": 1.7982215238224445e-05, + "loss": 0.368, + "step": 4387 + }, + { + "epoch": 1.84, + "learning_rate": 1.7981193870219587e-05, + "loss": 0.3786, + "step": 4388 + }, + { + "epoch": 1.84, + "learning_rate": 1.7980172272802398e-05, + "loss": 0.4624, + "step": 4389 + }, + { + "epoch": 1.84, + "learning_rate": 1.7979150446002233e-05, + "loss": 0.4058, + "step": 4390 + }, + { + "epoch": 1.84, + "learning_rate": 1.7978128389848476e-05, + "loss": 0.3636, + "step": 4391 + }, + { + "epoch": 1.84, + "learning_rate": 1.7977106104370493e-05, + "loss": 0.3949, + "step": 4392 + }, + { + "epoch": 1.84, + "learning_rate": 1.7976083589597672e-05, + "loss": 0.3636, + "step": 4393 + }, + { + "epoch": 1.84, + "learning_rate": 1.7975060845559412e-05, + "loss": 0.3717, + "step": 4394 + }, + { + "epoch": 1.84, + "learning_rate": 1.7974037872285102e-05, + "loss": 0.4207, + "step": 4395 + }, + { + "epoch": 1.84, + "learning_rate": 1.797301466980415e-05, + "loss": 0.3758, + "step": 4396 + }, + { + "epoch": 1.84, + "learning_rate": 1.7971991238145965e-05, + "loss": 0.3979, + "step": 4397 + }, + { + "epoch": 1.84, + "learning_rate": 1.7970967577339965e-05, + "loss": 0.4325, + "step": 4398 + }, + { + "epoch": 1.84, + "learning_rate": 1.7969943687415575e-05, + "loss": 0.4123, + "step": 4399 + }, + { + "epoch": 1.84, + "learning_rate": 1.7968919568402234e-05, + "loss": 0.3432, + "step": 4400 + }, + { + "epoch": 1.84, + "learning_rate": 1.7967895220329364e-05, + "loss": 0.3518, + "step": 4401 + }, + { + "epoch": 1.84, + "learning_rate": 1.796687064322642e-05, + "loss": 0.4224, + "step": 4402 + }, + { + "epoch": 1.84, + "learning_rate": 1.7965845837122844e-05, + "loss": 0.3905, + "step": 4403 + }, + { + "epoch": 1.84, + "learning_rate": 1.79648208020481e-05, + "loss": 0.3709, + "step": 4404 + }, + { + "epoch": 1.84, + "learning_rate": 1.796379553803165e-05, + "loss": 0.4077, + "step": 4405 + }, + { + "epoch": 1.84, + "learning_rate": 1.7962770045102968e-05, + "loss": 0.3806, + "step": 4406 + }, + { + "epoch": 1.85, + "learning_rate": 1.7961744323291525e-05, + "loss": 0.3653, + "step": 4407 + }, + { + "epoch": 1.85, + "learning_rate": 1.7960718372626807e-05, + "loss": 0.435, + "step": 4408 + }, + { + "epoch": 1.85, + "learning_rate": 1.7959692193138303e-05, + "loss": 0.3866, + "step": 4409 + }, + { + "epoch": 1.85, + "learning_rate": 1.795866578485551e-05, + "loss": 0.4393, + "step": 4410 + }, + { + "epoch": 1.85, + "learning_rate": 1.795763914780793e-05, + "loss": 0.4025, + "step": 4411 + }, + { + "epoch": 1.85, + "learning_rate": 1.7956612282025078e-05, + "loss": 0.3961, + "step": 4412 + }, + { + "epoch": 1.85, + "learning_rate": 1.7955585187536463e-05, + "loss": 0.4066, + "step": 4413 + }, + { + "epoch": 1.85, + "learning_rate": 1.7954557864371615e-05, + "loss": 0.3567, + "step": 4414 + }, + { + "epoch": 1.85, + "learning_rate": 1.7953530312560056e-05, + "loss": 0.4295, + "step": 4415 + }, + { + "epoch": 1.85, + "learning_rate": 1.7952502532131326e-05, + "loss": 0.3641, + "step": 4416 + }, + { + "epoch": 1.85, + "learning_rate": 1.795147452311497e-05, + "loss": 0.4159, + "step": 4417 + }, + { + "epoch": 1.85, + "learning_rate": 1.7950446285540532e-05, + "loss": 0.3475, + "step": 4418 + }, + { + "epoch": 1.85, + "learning_rate": 1.7949417819437577e-05, + "loss": 0.3599, + "step": 4419 + }, + { + "epoch": 1.85, + "learning_rate": 1.7948389124835654e-05, + "loss": 0.3755, + "step": 4420 + }, + { + "epoch": 1.85, + "learning_rate": 1.794736020176434e-05, + "loss": 0.4222, + "step": 4421 + }, + { + "epoch": 1.85, + "learning_rate": 1.794633105025321e-05, + "loss": 0.3815, + "step": 4422 + }, + { + "epoch": 1.85, + "learning_rate": 1.7945301670331847e-05, + "loss": 0.4236, + "step": 4423 + }, + { + "epoch": 1.85, + "learning_rate": 1.794427206202984e-05, + "loss": 0.3942, + "step": 4424 + }, + { + "epoch": 1.85, + "learning_rate": 1.7943242225376778e-05, + "loss": 0.3716, + "step": 4425 + }, + { + "epoch": 1.85, + "learning_rate": 1.794221216040227e-05, + "loss": 0.3625, + "step": 4426 + }, + { + "epoch": 1.85, + "learning_rate": 1.7941181867135917e-05, + "loss": 0.4135, + "step": 4427 + }, + { + "epoch": 1.85, + "learning_rate": 1.7940151345607343e-05, + "loss": 0.3409, + "step": 4428 + }, + { + "epoch": 1.85, + "learning_rate": 1.793912059584616e-05, + "loss": 0.3957, + "step": 4429 + }, + { + "epoch": 1.85, + "learning_rate": 1.7938089617882004e-05, + "loss": 0.3973, + "step": 4430 + }, + { + "epoch": 1.86, + "learning_rate": 1.7937058411744504e-05, + "loss": 0.4159, + "step": 4431 + }, + { + "epoch": 1.86, + "learning_rate": 1.7936026977463306e-05, + "loss": 0.3604, + "step": 4432 + }, + { + "epoch": 1.86, + "learning_rate": 1.793499531506805e-05, + "loss": 0.3864, + "step": 4433 + }, + { + "epoch": 1.86, + "learning_rate": 1.79339634245884e-05, + "loss": 0.4238, + "step": 4434 + }, + { + "epoch": 1.86, + "learning_rate": 1.7932931306054007e-05, + "loss": 0.3839, + "step": 4435 + }, + { + "epoch": 1.86, + "learning_rate": 1.7931898959494546e-05, + "loss": 0.3724, + "step": 4436 + }, + { + "epoch": 1.86, + "learning_rate": 1.793086638493969e-05, + "loss": 0.383, + "step": 4437 + }, + { + "epoch": 1.86, + "learning_rate": 1.7929833582419115e-05, + "loss": 0.4443, + "step": 4438 + }, + { + "epoch": 1.86, + "learning_rate": 1.792880055196251e-05, + "loss": 0.4123, + "step": 4439 + }, + { + "epoch": 1.86, + "learning_rate": 1.792776729359957e-05, + "loss": 0.3807, + "step": 4440 + }, + { + "epoch": 1.86, + "learning_rate": 1.7926733807359993e-05, + "loss": 0.4245, + "step": 4441 + }, + { + "epoch": 1.86, + "learning_rate": 1.792570009327349e-05, + "loss": 0.3401, + "step": 4442 + }, + { + "epoch": 1.86, + "learning_rate": 1.792466615136977e-05, + "loss": 0.4053, + "step": 4443 + }, + { + "epoch": 1.86, + "learning_rate": 1.7923631981678552e-05, + "loss": 0.4292, + "step": 4444 + }, + { + "epoch": 1.86, + "learning_rate": 1.7922597584229566e-05, + "loss": 0.3729, + "step": 4445 + }, + { + "epoch": 1.86, + "learning_rate": 1.792156295905254e-05, + "loss": 0.4087, + "step": 4446 + }, + { + "epoch": 1.86, + "learning_rate": 1.792052810617722e-05, + "loss": 0.3385, + "step": 4447 + }, + { + "epoch": 1.86, + "learning_rate": 1.7919493025633342e-05, + "loss": 0.4045, + "step": 4448 + }, + { + "epoch": 1.86, + "learning_rate": 1.791845771745067e-05, + "loss": 0.3663, + "step": 4449 + }, + { + "epoch": 1.86, + "learning_rate": 1.7917422181658954e-05, + "loss": 0.4042, + "step": 4450 + }, + { + "epoch": 1.86, + "learning_rate": 1.7916386418287968e-05, + "loss": 0.3809, + "step": 4451 + }, + { + "epoch": 1.86, + "learning_rate": 1.7915350427367476e-05, + "loss": 0.4189, + "step": 4452 + }, + { + "epoch": 1.86, + "learning_rate": 1.791431420892726e-05, + "loss": 0.3534, + "step": 4453 + }, + { + "epoch": 1.86, + "learning_rate": 1.7913277762997105e-05, + "loss": 0.3893, + "step": 4454 + }, + { + "epoch": 1.87, + "learning_rate": 1.7912241089606802e-05, + "loss": 0.3716, + "step": 4455 + }, + { + "epoch": 1.87, + "learning_rate": 1.7911204188786154e-05, + "loss": 0.4082, + "step": 4456 + }, + { + "epoch": 1.87, + "learning_rate": 1.791016706056496e-05, + "loss": 0.3839, + "step": 4457 + }, + { + "epoch": 1.87, + "learning_rate": 1.790912970497303e-05, + "loss": 0.4122, + "step": 4458 + }, + { + "epoch": 1.87, + "learning_rate": 1.790809212204019e-05, + "loss": 0.3856, + "step": 4459 + }, + { + "epoch": 1.87, + "learning_rate": 1.7907054311796254e-05, + "loss": 0.4127, + "step": 4460 + }, + { + "epoch": 1.87, + "learning_rate": 1.7906016274271062e-05, + "loss": 0.3855, + "step": 4461 + }, + { + "epoch": 1.87, + "learning_rate": 1.790497800949445e-05, + "loss": 0.4138, + "step": 4462 + }, + { + "epoch": 1.87, + "learning_rate": 1.7903939517496253e-05, + "loss": 0.4138, + "step": 4463 + }, + { + "epoch": 1.87, + "learning_rate": 1.7902900798306334e-05, + "loss": 0.3514, + "step": 4464 + }, + { + "epoch": 1.87, + "learning_rate": 1.7901861851954543e-05, + "loss": 0.3967, + "step": 4465 + }, + { + "epoch": 1.87, + "learning_rate": 1.7900822678470745e-05, + "loss": 0.4038, + "step": 4466 + }, + { + "epoch": 1.87, + "learning_rate": 1.789978327788481e-05, + "loss": 0.3741, + "step": 4467 + }, + { + "epoch": 1.87, + "learning_rate": 1.7898743650226617e-05, + "loss": 0.4083, + "step": 4468 + }, + { + "epoch": 1.87, + "learning_rate": 1.7897703795526044e-05, + "loss": 0.3672, + "step": 4469 + }, + { + "epoch": 1.87, + "learning_rate": 1.7896663713812984e-05, + "loss": 0.4927, + "step": 4470 + }, + { + "epoch": 1.87, + "learning_rate": 1.7895623405117334e-05, + "loss": 0.4271, + "step": 4471 + }, + { + "epoch": 1.87, + "learning_rate": 1.7894582869468994e-05, + "loss": 0.3796, + "step": 4472 + }, + { + "epoch": 1.87, + "learning_rate": 1.7893542106897878e-05, + "loss": 0.4172, + "step": 4473 + }, + { + "epoch": 1.87, + "learning_rate": 1.7892501117433896e-05, + "loss": 0.3696, + "step": 4474 + }, + { + "epoch": 1.87, + "learning_rate": 1.7891459901106976e-05, + "loss": 0.3844, + "step": 4475 + }, + { + "epoch": 1.87, + "learning_rate": 1.7890418457947038e-05, + "loss": 0.3763, + "step": 4476 + }, + { + "epoch": 1.87, + "learning_rate": 1.788937678798403e-05, + "loss": 0.3572, + "step": 4477 + }, + { + "epoch": 1.87, + "learning_rate": 1.7888334891247883e-05, + "loss": 0.3744, + "step": 4478 + }, + { + "epoch": 1.88, + "learning_rate": 1.788729276776855e-05, + "loss": 0.3865, + "step": 4479 + }, + { + "epoch": 1.88, + "learning_rate": 1.7886250417575986e-05, + "loss": 0.4056, + "step": 4480 + }, + { + "epoch": 1.88, + "learning_rate": 1.7885207840700155e-05, + "loss": 0.4122, + "step": 4481 + }, + { + "epoch": 1.88, + "learning_rate": 1.7884165037171016e-05, + "loss": 0.3764, + "step": 4482 + }, + { + "epoch": 1.88, + "learning_rate": 1.7883122007018554e-05, + "loss": 0.3989, + "step": 4483 + }, + { + "epoch": 1.88, + "learning_rate": 1.788207875027274e-05, + "loss": 0.4024, + "step": 4484 + }, + { + "epoch": 1.88, + "learning_rate": 1.7881035266963575e-05, + "loss": 0.3932, + "step": 4485 + }, + { + "epoch": 1.88, + "learning_rate": 1.787999155712104e-05, + "loss": 0.3933, + "step": 4486 + }, + { + "epoch": 1.88, + "learning_rate": 1.787894762077514e-05, + "loss": 0.3786, + "step": 4487 + }, + { + "epoch": 1.88, + "learning_rate": 1.787790345795588e-05, + "loss": 0.4072, + "step": 4488 + }, + { + "epoch": 1.88, + "learning_rate": 1.787685906869328e-05, + "loss": 0.4105, + "step": 4489 + }, + { + "epoch": 1.88, + "learning_rate": 1.7875814453017355e-05, + "loss": 0.3618, + "step": 4490 + }, + { + "epoch": 1.88, + "learning_rate": 1.787476961095813e-05, + "loss": 0.4214, + "step": 4491 + }, + { + "epoch": 1.88, + "learning_rate": 1.7873724542545647e-05, + "loss": 0.3606, + "step": 4492 + }, + { + "epoch": 1.88, + "learning_rate": 1.7872679247809933e-05, + "loss": 0.397, + "step": 4493 + }, + { + "epoch": 1.88, + "learning_rate": 1.7871633726781042e-05, + "loss": 0.4022, + "step": 4494 + }, + { + "epoch": 1.88, + "learning_rate": 1.787058797948902e-05, + "loss": 0.3629, + "step": 4495 + }, + { + "epoch": 1.88, + "learning_rate": 1.786954200596394e-05, + "loss": 0.427, + "step": 4496 + }, + { + "epoch": 1.88, + "learning_rate": 1.786849580623585e-05, + "loss": 0.3878, + "step": 4497 + }, + { + "epoch": 1.88, + "learning_rate": 1.7867449380334834e-05, + "loss": 0.4738, + "step": 4498 + }, + { + "epoch": 1.88, + "learning_rate": 1.7866402728290967e-05, + "loss": 0.3804, + "step": 4499 + }, + { + "epoch": 1.88, + "learning_rate": 1.786535585013433e-05, + "loss": 0.4011, + "step": 4500 + }, + { + "epoch": 1.88, + "learning_rate": 1.7864308745895024e-05, + "loss": 0.4079, + "step": 4501 + }, + { + "epoch": 1.88, + "learning_rate": 1.7863261415603142e-05, + "loss": 0.3485, + "step": 4502 + }, + { + "epoch": 1.89, + "learning_rate": 1.7862213859288788e-05, + "loss": 0.4219, + "step": 4503 + }, + { + "epoch": 1.89, + "learning_rate": 1.786116607698207e-05, + "loss": 0.3978, + "step": 4504 + }, + { + "epoch": 1.89, + "learning_rate": 1.786011806871311e-05, + "loss": 0.3497, + "step": 4505 + }, + { + "epoch": 1.89, + "learning_rate": 1.7859069834512033e-05, + "loss": 0.354, + "step": 4506 + }, + { + "epoch": 1.89, + "learning_rate": 1.7858021374408968e-05, + "loss": 0.3959, + "step": 4507 + }, + { + "epoch": 1.89, + "learning_rate": 1.785697268843405e-05, + "loss": 0.375, + "step": 4508 + }, + { + "epoch": 1.89, + "learning_rate": 1.7855923776617424e-05, + "loss": 0.3681, + "step": 4509 + }, + { + "epoch": 1.89, + "learning_rate": 1.785487463898924e-05, + "loss": 0.4029, + "step": 4510 + }, + { + "epoch": 1.89, + "learning_rate": 1.7853825275579656e-05, + "loss": 0.4052, + "step": 4511 + }, + { + "epoch": 1.89, + "learning_rate": 1.7852775686418832e-05, + "loss": 0.4151, + "step": 4512 + }, + { + "epoch": 1.89, + "learning_rate": 1.7851725871536945e-05, + "loss": 0.4123, + "step": 4513 + }, + { + "epoch": 1.89, + "learning_rate": 1.785067583096416e-05, + "loss": 0.3862, + "step": 4514 + }, + { + "epoch": 1.89, + "learning_rate": 1.7849625564730667e-05, + "loss": 0.3823, + "step": 4515 + }, + { + "epoch": 1.89, + "learning_rate": 1.784857507286665e-05, + "loss": 0.3856, + "step": 4516 + }, + { + "epoch": 1.89, + "learning_rate": 1.784752435540231e-05, + "loss": 0.3882, + "step": 4517 + }, + { + "epoch": 1.89, + "learning_rate": 1.7846473412367845e-05, + "loss": 0.4147, + "step": 4518 + }, + { + "epoch": 1.89, + "learning_rate": 1.7845422243793464e-05, + "loss": 0.4482, + "step": 4519 + }, + { + "epoch": 1.89, + "learning_rate": 1.784437084970938e-05, + "loss": 0.3736, + "step": 4520 + }, + { + "epoch": 1.89, + "learning_rate": 1.7843319230145822e-05, + "loss": 0.3869, + "step": 4521 + }, + { + "epoch": 1.89, + "learning_rate": 1.7842267385133013e-05, + "loss": 0.4484, + "step": 4522 + }, + { + "epoch": 1.89, + "learning_rate": 1.7841215314701182e-05, + "loss": 0.4245, + "step": 4523 + }, + { + "epoch": 1.89, + "learning_rate": 1.784016301888058e-05, + "loss": 0.3619, + "step": 4524 + }, + { + "epoch": 1.89, + "learning_rate": 1.7839110497701446e-05, + "loss": 0.4076, + "step": 4525 + }, + { + "epoch": 1.89, + "learning_rate": 1.783805775119404e-05, + "loss": 0.4029, + "step": 4526 + }, + { + "epoch": 1.9, + "learning_rate": 1.7837004779388615e-05, + "loss": 0.4297, + "step": 4527 + }, + { + "epoch": 1.9, + "learning_rate": 1.7835951582315445e-05, + "loss": 0.3741, + "step": 4528 + }, + { + "epoch": 1.9, + "learning_rate": 1.78348981600048e-05, + "loss": 0.4124, + "step": 4529 + }, + { + "epoch": 1.9, + "learning_rate": 1.783384451248696e-05, + "loss": 0.3873, + "step": 4530 + }, + { + "epoch": 1.9, + "learning_rate": 1.783279063979221e-05, + "loss": 0.3949, + "step": 4531 + }, + { + "epoch": 1.9, + "learning_rate": 1.7831736541950845e-05, + "loss": 0.399, + "step": 4532 + }, + { + "epoch": 1.9, + "learning_rate": 1.783068221899316e-05, + "loss": 0.4076, + "step": 4533 + }, + { + "epoch": 1.9, + "learning_rate": 1.7829627670949468e-05, + "loss": 0.4031, + "step": 4534 + }, + { + "epoch": 1.9, + "learning_rate": 1.7828572897850076e-05, + "loss": 0.3656, + "step": 4535 + }, + { + "epoch": 1.9, + "learning_rate": 1.78275178997253e-05, + "loss": 0.3846, + "step": 4536 + }, + { + "epoch": 1.9, + "learning_rate": 1.782646267660547e-05, + "loss": 0.409, + "step": 4537 + }, + { + "epoch": 1.9, + "learning_rate": 1.7825407228520917e-05, + "loss": 0.3568, + "step": 4538 + }, + { + "epoch": 1.9, + "learning_rate": 1.7824351555501972e-05, + "loss": 0.4116, + "step": 4539 + }, + { + "epoch": 1.9, + "learning_rate": 1.782329565757899e-05, + "loss": 0.4143, + "step": 4540 + }, + { + "epoch": 1.9, + "learning_rate": 1.7822239534782312e-05, + "loss": 0.3885, + "step": 4541 + }, + { + "epoch": 1.9, + "learning_rate": 1.7821183187142302e-05, + "loss": 0.4098, + "step": 4542 + }, + { + "epoch": 1.9, + "learning_rate": 1.7820126614689326e-05, + "loss": 0.4534, + "step": 4543 + }, + { + "epoch": 1.9, + "learning_rate": 1.7819069817453746e-05, + "loss": 0.3467, + "step": 4544 + }, + { + "epoch": 1.9, + "learning_rate": 1.781801279546594e-05, + "loss": 0.408, + "step": 4545 + }, + { + "epoch": 1.9, + "learning_rate": 1.7816955548756298e-05, + "loss": 0.3673, + "step": 4546 + }, + { + "epoch": 1.9, + "learning_rate": 1.7815898077355204e-05, + "loss": 0.4112, + "step": 4547 + }, + { + "epoch": 1.9, + "learning_rate": 1.7814840381293054e-05, + "loss": 0.3596, + "step": 4548 + }, + { + "epoch": 1.9, + "learning_rate": 1.7813782460600257e-05, + "loss": 0.3993, + "step": 4549 + }, + { + "epoch": 1.9, + "learning_rate": 1.781272431530721e-05, + "loss": 0.3849, + "step": 4550 + }, + { + "epoch": 1.91, + "learning_rate": 1.7811665945444337e-05, + "loss": 0.3657, + "step": 4551 + }, + { + "epoch": 1.91, + "learning_rate": 1.7810607351042062e-05, + "loss": 0.3464, + "step": 4552 + }, + { + "epoch": 1.91, + "learning_rate": 1.7809548532130808e-05, + "loss": 0.3813, + "step": 4553 + }, + { + "epoch": 1.91, + "learning_rate": 1.7808489488741013e-05, + "loss": 0.4107, + "step": 4554 + }, + { + "epoch": 1.91, + "learning_rate": 1.7807430220903113e-05, + "loss": 0.319, + "step": 4555 + }, + { + "epoch": 1.91, + "learning_rate": 1.780637072864756e-05, + "loss": 0.3817, + "step": 4556 + }, + { + "epoch": 1.91, + "learning_rate": 1.7805311012004812e-05, + "loss": 0.36, + "step": 4557 + }, + { + "epoch": 1.91, + "learning_rate": 1.7804251071005323e-05, + "loss": 0.4005, + "step": 4558 + }, + { + "epoch": 1.91, + "learning_rate": 1.7803190905679562e-05, + "loss": 0.3722, + "step": 4559 + }, + { + "epoch": 1.91, + "learning_rate": 1.7802130516058003e-05, + "loss": 0.4129, + "step": 4560 + }, + { + "epoch": 1.91, + "learning_rate": 1.7801069902171123e-05, + "loss": 0.3759, + "step": 4561 + }, + { + "epoch": 1.91, + "learning_rate": 1.7800009064049416e-05, + "loss": 0.414, + "step": 4562 + }, + { + "epoch": 1.91, + "learning_rate": 1.7798948001723368e-05, + "loss": 0.4868, + "step": 4563 + }, + { + "epoch": 1.91, + "learning_rate": 1.779788671522348e-05, + "loss": 0.3889, + "step": 4564 + }, + { + "epoch": 1.91, + "learning_rate": 1.779682520458026e-05, + "loss": 0.3491, + "step": 4565 + }, + { + "epoch": 1.91, + "learning_rate": 1.779576346982421e-05, + "loss": 0.4432, + "step": 4566 + }, + { + "epoch": 1.91, + "learning_rate": 1.7794701510985867e-05, + "loss": 0.3424, + "step": 4567 + }, + { + "epoch": 1.91, + "learning_rate": 1.779363932809574e-05, + "loss": 0.3825, + "step": 4568 + }, + { + "epoch": 1.91, + "learning_rate": 1.7792576921184374e-05, + "loss": 0.3551, + "step": 4569 + }, + { + "epoch": 1.91, + "learning_rate": 1.7791514290282293e-05, + "loss": 0.3705, + "step": 4570 + }, + { + "epoch": 1.91, + "learning_rate": 1.7790451435420046e-05, + "loss": 0.3369, + "step": 4571 + }, + { + "epoch": 1.91, + "learning_rate": 1.778938835662819e-05, + "loss": 0.402, + "step": 4572 + }, + { + "epoch": 1.91, + "learning_rate": 1.7788325053937278e-05, + "loss": 0.3878, + "step": 4573 + }, + { + "epoch": 1.92, + "learning_rate": 1.778726152737787e-05, + "loss": 0.4031, + "step": 4574 + }, + { + "epoch": 1.92, + "learning_rate": 1.7786197776980542e-05, + "loss": 0.384, + "step": 4575 + }, + { + "epoch": 1.92, + "learning_rate": 1.778513380277587e-05, + "loss": 0.3849, + "step": 4576 + }, + { + "epoch": 1.92, + "learning_rate": 1.7784069604794433e-05, + "loss": 0.3563, + "step": 4577 + }, + { + "epoch": 1.92, + "learning_rate": 1.778300518306682e-05, + "loss": 0.4222, + "step": 4578 + }, + { + "epoch": 1.92, + "learning_rate": 1.7781940537623636e-05, + "loss": 0.3769, + "step": 4579 + }, + { + "epoch": 1.92, + "learning_rate": 1.778087566849547e-05, + "loss": 0.3815, + "step": 4580 + }, + { + "epoch": 1.92, + "learning_rate": 1.7779810575712942e-05, + "loss": 0.3749, + "step": 4581 + }, + { + "epoch": 1.92, + "learning_rate": 1.777874525930666e-05, + "loss": 0.3943, + "step": 4582 + }, + { + "epoch": 1.92, + "learning_rate": 1.7777679719307253e-05, + "loss": 0.3639, + "step": 4583 + }, + { + "epoch": 1.92, + "learning_rate": 1.7776613955745337e-05, + "loss": 0.3701, + "step": 4584 + }, + { + "epoch": 1.92, + "learning_rate": 1.7775547968651562e-05, + "loss": 0.3833, + "step": 4585 + }, + { + "epoch": 1.92, + "learning_rate": 1.7774481758056553e-05, + "loss": 0.3607, + "step": 4586 + }, + { + "epoch": 1.92, + "learning_rate": 1.777341532399097e-05, + "loss": 0.3518, + "step": 4587 + }, + { + "epoch": 1.92, + "learning_rate": 1.777234866648546e-05, + "loss": 0.4178, + "step": 4588 + }, + { + "epoch": 1.92, + "learning_rate": 1.7771281785570683e-05, + "loss": 0.4101, + "step": 4589 + }, + { + "epoch": 1.92, + "learning_rate": 1.777021468127731e-05, + "loss": 0.3854, + "step": 4590 + }, + { + "epoch": 1.92, + "learning_rate": 1.776914735363601e-05, + "loss": 0.3975, + "step": 4591 + }, + { + "epoch": 1.92, + "learning_rate": 1.776807980267746e-05, + "loss": 0.409, + "step": 4592 + }, + { + "epoch": 1.92, + "learning_rate": 1.7767012028432352e-05, + "loss": 0.4173, + "step": 4593 + }, + { + "epoch": 1.92, + "learning_rate": 1.776594403093138e-05, + "loss": 0.3405, + "step": 4594 + }, + { + "epoch": 1.92, + "learning_rate": 1.7764875810205234e-05, + "loss": 0.3914, + "step": 4595 + }, + { + "epoch": 1.92, + "learning_rate": 1.776380736628462e-05, + "loss": 0.3719, + "step": 4596 + }, + { + "epoch": 1.92, + "learning_rate": 1.7762738699200256e-05, + "loss": 0.4169, + "step": 4597 + }, + { + "epoch": 1.93, + "learning_rate": 1.7761669808982857e-05, + "loss": 0.4051, + "step": 4598 + }, + { + "epoch": 1.93, + "learning_rate": 1.776060069566315e-05, + "loss": 0.3898, + "step": 4599 + }, + { + "epoch": 1.93, + "learning_rate": 1.7759531359271857e-05, + "loss": 0.3679, + "step": 4600 + }, + { + "epoch": 1.93, + "learning_rate": 1.7758461799839723e-05, + "loss": 0.3402, + "step": 4601 + }, + { + "epoch": 1.93, + "learning_rate": 1.7757392017397493e-05, + "loss": 0.391, + "step": 4602 + }, + { + "epoch": 1.93, + "learning_rate": 1.7756322011975906e-05, + "loss": 0.3498, + "step": 4603 + }, + { + "epoch": 1.93, + "learning_rate": 1.7755251783605733e-05, + "loss": 0.3862, + "step": 4604 + }, + { + "epoch": 1.93, + "learning_rate": 1.7754181332317725e-05, + "loss": 0.4389, + "step": 4605 + }, + { + "epoch": 1.93, + "learning_rate": 1.7753110658142657e-05, + "loss": 0.3772, + "step": 4606 + }, + { + "epoch": 1.93, + "learning_rate": 1.77520397611113e-05, + "loss": 0.4095, + "step": 4607 + }, + { + "epoch": 1.93, + "learning_rate": 1.7750968641254444e-05, + "loss": 0.4164, + "step": 4608 + }, + { + "epoch": 1.93, + "learning_rate": 1.7749897298602867e-05, + "loss": 0.4049, + "step": 4609 + }, + { + "epoch": 1.93, + "learning_rate": 1.7748825733187372e-05, + "loss": 0.3771, + "step": 4610 + }, + { + "epoch": 1.93, + "learning_rate": 1.774775394503876e-05, + "loss": 0.3696, + "step": 4611 + }, + { + "epoch": 1.93, + "learning_rate": 1.7746681934187832e-05, + "loss": 0.3992, + "step": 4612 + }, + { + "epoch": 1.93, + "learning_rate": 1.7745609700665406e-05, + "loss": 0.3612, + "step": 4613 + }, + { + "epoch": 1.93, + "learning_rate": 1.774453724450231e-05, + "loss": 0.3978, + "step": 4614 + }, + { + "epoch": 1.93, + "learning_rate": 1.7743464565729354e-05, + "loss": 0.4378, + "step": 4615 + }, + { + "epoch": 1.93, + "learning_rate": 1.7742391664377384e-05, + "loss": 0.3375, + "step": 4616 + }, + { + "epoch": 1.93, + "learning_rate": 1.7741318540477234e-05, + "loss": 0.4127, + "step": 4617 + }, + { + "epoch": 1.93, + "learning_rate": 1.7740245194059754e-05, + "loss": 0.3915, + "step": 4618 + }, + { + "epoch": 1.93, + "learning_rate": 1.77391716251558e-05, + "loss": 0.3586, + "step": 4619 + }, + { + "epoch": 1.93, + "learning_rate": 1.7738097833796218e-05, + "loss": 0.3378, + "step": 4620 + }, + { + "epoch": 1.93, + "learning_rate": 1.7737023820011884e-05, + "loss": 0.4145, + "step": 4621 + }, + { + "epoch": 1.94, + "learning_rate": 1.7735949583833662e-05, + "loss": 0.4131, + "step": 4622 + }, + { + "epoch": 1.94, + "learning_rate": 1.7734875125292436e-05, + "loss": 0.3654, + "step": 4623 + }, + { + "epoch": 1.94, + "learning_rate": 1.7733800444419088e-05, + "loss": 0.3907, + "step": 4624 + }, + { + "epoch": 1.94, + "learning_rate": 1.773272554124451e-05, + "loss": 0.4214, + "step": 4625 + }, + { + "epoch": 1.94, + "learning_rate": 1.77316504157996e-05, + "loss": 0.3629, + "step": 4626 + }, + { + "epoch": 1.94, + "learning_rate": 1.773057506811526e-05, + "loss": 0.3772, + "step": 4627 + }, + { + "epoch": 1.94, + "learning_rate": 1.7729499498222392e-05, + "loss": 0.4109, + "step": 4628 + }, + { + "epoch": 1.94, + "learning_rate": 1.7728423706151928e-05, + "loss": 0.36, + "step": 4629 + }, + { + "epoch": 1.94, + "learning_rate": 1.772734769193478e-05, + "loss": 0.436, + "step": 4630 + }, + { + "epoch": 1.94, + "learning_rate": 1.772627145560188e-05, + "loss": 0.3648, + "step": 4631 + }, + { + "epoch": 1.94, + "learning_rate": 1.772519499718416e-05, + "loss": 0.4065, + "step": 4632 + }, + { + "epoch": 1.94, + "learning_rate": 1.772411831671257e-05, + "loss": 0.387, + "step": 4633 + }, + { + "epoch": 1.94, + "learning_rate": 1.7723041414218053e-05, + "loss": 0.3432, + "step": 4634 + }, + { + "epoch": 1.94, + "learning_rate": 1.7721964289731562e-05, + "loss": 0.4029, + "step": 4635 + }, + { + "epoch": 1.94, + "learning_rate": 1.772088694328406e-05, + "loss": 0.3698, + "step": 4636 + }, + { + "epoch": 1.94, + "learning_rate": 1.771980937490651e-05, + "loss": 0.4266, + "step": 4637 + }, + { + "epoch": 1.94, + "learning_rate": 1.7718731584629897e-05, + "loss": 0.3723, + "step": 4638 + }, + { + "epoch": 1.94, + "learning_rate": 1.771765357248519e-05, + "loss": 0.3856, + "step": 4639 + }, + { + "epoch": 1.94, + "learning_rate": 1.7716575338503382e-05, + "loss": 0.3812, + "step": 4640 + }, + { + "epoch": 1.94, + "learning_rate": 1.7715496882715463e-05, + "loss": 0.353, + "step": 4641 + }, + { + "epoch": 1.94, + "learning_rate": 1.7714418205152434e-05, + "loss": 0.4068, + "step": 4642 + }, + { + "epoch": 1.94, + "learning_rate": 1.7713339305845293e-05, + "loss": 0.4075, + "step": 4643 + }, + { + "epoch": 1.94, + "learning_rate": 1.7712260184825063e-05, + "loss": 0.3755, + "step": 4644 + }, + { + "epoch": 1.94, + "learning_rate": 1.771118084212276e-05, + "loss": 0.4403, + "step": 4645 + }, + { + "epoch": 1.95, + "learning_rate": 1.7710101277769403e-05, + "loss": 0.4173, + "step": 4646 + }, + { + "epoch": 1.95, + "learning_rate": 1.7709021491796027e-05, + "loss": 0.3637, + "step": 4647 + }, + { + "epoch": 1.95, + "learning_rate": 1.770794148423367e-05, + "loss": 0.4326, + "step": 4648 + }, + { + "epoch": 1.95, + "learning_rate": 1.7706861255113376e-05, + "loss": 0.3747, + "step": 4649 + }, + { + "epoch": 1.95, + "learning_rate": 1.7705780804466192e-05, + "loss": 0.3416, + "step": 4650 + }, + { + "epoch": 1.95, + "learning_rate": 1.7704700132323175e-05, + "loss": 0.3899, + "step": 4651 + }, + { + "epoch": 1.95, + "learning_rate": 1.7703619238715392e-05, + "loss": 0.4293, + "step": 4652 + }, + { + "epoch": 1.95, + "learning_rate": 1.770253812367391e-05, + "loss": 0.4505, + "step": 4653 + }, + { + "epoch": 1.95, + "learning_rate": 1.7701456787229805e-05, + "loss": 0.3799, + "step": 4654 + }, + { + "epoch": 1.95, + "learning_rate": 1.770037522941416e-05, + "loss": 0.3909, + "step": 4655 + }, + { + "epoch": 1.95, + "learning_rate": 1.7699293450258062e-05, + "loss": 0.4465, + "step": 4656 + }, + { + "epoch": 1.95, + "learning_rate": 1.7698211449792604e-05, + "loss": 0.3893, + "step": 4657 + }, + { + "epoch": 1.95, + "learning_rate": 1.769712922804889e-05, + "loss": 0.3736, + "step": 4658 + }, + { + "epoch": 1.95, + "learning_rate": 1.7696046785058024e-05, + "loss": 0.3389, + "step": 4659 + }, + { + "epoch": 1.95, + "learning_rate": 1.7694964120851128e-05, + "loss": 0.4159, + "step": 4660 + }, + { + "epoch": 1.95, + "learning_rate": 1.7693881235459315e-05, + "loss": 0.391, + "step": 4661 + }, + { + "epoch": 1.95, + "learning_rate": 1.7692798128913714e-05, + "loss": 0.3976, + "step": 4662 + }, + { + "epoch": 1.95, + "learning_rate": 1.7691714801245453e-05, + "loss": 0.371, + "step": 4663 + }, + { + "epoch": 1.95, + "learning_rate": 1.7690631252485676e-05, + "loss": 0.3931, + "step": 4664 + }, + { + "epoch": 1.95, + "learning_rate": 1.7689547482665532e-05, + "loss": 0.3735, + "step": 4665 + }, + { + "epoch": 1.95, + "learning_rate": 1.768846349181617e-05, + "loss": 0.4045, + "step": 4666 + }, + { + "epoch": 1.95, + "learning_rate": 1.7687379279968745e-05, + "loss": 0.4001, + "step": 4667 + }, + { + "epoch": 1.95, + "learning_rate": 1.7686294847154423e-05, + "loss": 0.3443, + "step": 4668 + }, + { + "epoch": 1.95, + "learning_rate": 1.7685210193404377e-05, + "loss": 0.4312, + "step": 4669 + }, + { + "epoch": 1.96, + "learning_rate": 1.7684125318749785e-05, + "loss": 0.3976, + "step": 4670 + }, + { + "epoch": 1.96, + "learning_rate": 1.7683040223221828e-05, + "loss": 0.4214, + "step": 4671 + }, + { + "epoch": 1.96, + "learning_rate": 1.76819549068517e-05, + "loss": 0.3772, + "step": 4672 + }, + { + "epoch": 1.96, + "learning_rate": 1.768086936967059e-05, + "loss": 0.3983, + "step": 4673 + }, + { + "epoch": 1.96, + "learning_rate": 1.767978361170971e-05, + "loss": 0.4451, + "step": 4674 + }, + { + "epoch": 1.96, + "learning_rate": 1.7678697633000267e-05, + "loss": 0.3974, + "step": 4675 + }, + { + "epoch": 1.96, + "learning_rate": 1.767761143357347e-05, + "loss": 0.3598, + "step": 4676 + }, + { + "epoch": 1.96, + "learning_rate": 1.7676525013460547e-05, + "loss": 0.3909, + "step": 4677 + }, + { + "epoch": 1.96, + "learning_rate": 1.7675438372692726e-05, + "loss": 0.4153, + "step": 4678 + }, + { + "epoch": 1.96, + "learning_rate": 1.7674351511301238e-05, + "loss": 0.3957, + "step": 4679 + }, + { + "epoch": 1.96, + "learning_rate": 1.767326442931733e-05, + "loss": 0.3571, + "step": 4680 + }, + { + "epoch": 1.96, + "learning_rate": 1.7672177126772237e-05, + "loss": 0.3678, + "step": 4681 + }, + { + "epoch": 1.96, + "learning_rate": 1.7671089603697228e-05, + "loss": 0.4258, + "step": 4682 + }, + { + "epoch": 1.96, + "learning_rate": 1.767000186012355e-05, + "loss": 0.4408, + "step": 4683 + }, + { + "epoch": 1.96, + "learning_rate": 1.766891389608248e-05, + "loss": 0.409, + "step": 4684 + }, + { + "epoch": 1.96, + "learning_rate": 1.7667825711605283e-05, + "loss": 0.4514, + "step": 4685 + }, + { + "epoch": 1.96, + "learning_rate": 1.766673730672324e-05, + "loss": 0.3947, + "step": 4686 + }, + { + "epoch": 1.96, + "learning_rate": 1.766564868146764e-05, + "loss": 0.349, + "step": 4687 + }, + { + "epoch": 1.96, + "learning_rate": 1.7664559835869763e-05, + "loss": 0.4218, + "step": 4688 + }, + { + "epoch": 1.96, + "learning_rate": 1.766347076996092e-05, + "loss": 0.4156, + "step": 4689 + }, + { + "epoch": 1.96, + "learning_rate": 1.7662381483772412e-05, + "loss": 0.3596, + "step": 4690 + }, + { + "epoch": 1.96, + "learning_rate": 1.7661291977335544e-05, + "loss": 0.4368, + "step": 4691 + }, + { + "epoch": 1.96, + "learning_rate": 1.766020225068164e-05, + "loss": 0.3567, + "step": 4692 + }, + { + "epoch": 1.96, + "learning_rate": 1.7659112303842015e-05, + "loss": 0.3661, + "step": 4693 + }, + { + "epoch": 1.97, + "learning_rate": 1.7658022136848008e-05, + "loss": 0.3629, + "step": 4694 + }, + { + "epoch": 1.97, + "learning_rate": 1.7656931749730947e-05, + "loss": 0.3877, + "step": 4695 + }, + { + "epoch": 1.97, + "learning_rate": 1.765584114252218e-05, + "loss": 0.3517, + "step": 4696 + }, + { + "epoch": 1.97, + "learning_rate": 1.7654750315253055e-05, + "loss": 0.3632, + "step": 4697 + }, + { + "epoch": 1.97, + "learning_rate": 1.765365926795492e-05, + "loss": 0.3851, + "step": 4698 + }, + { + "epoch": 1.97, + "learning_rate": 1.7652568000659144e-05, + "loss": 0.4156, + "step": 4699 + }, + { + "epoch": 1.97, + "learning_rate": 1.765147651339709e-05, + "loss": 0.3623, + "step": 4700 + }, + { + "epoch": 1.97, + "learning_rate": 1.7650384806200134e-05, + "loss": 0.4243, + "step": 4701 + }, + { + "epoch": 1.97, + "learning_rate": 1.7649292879099655e-05, + "loss": 0.3677, + "step": 4702 + }, + { + "epoch": 1.97, + "learning_rate": 1.764820073212704e-05, + "loss": 0.4255, + "step": 4703 + }, + { + "epoch": 1.97, + "learning_rate": 1.7647108365313683e-05, + "loss": 0.4099, + "step": 4704 + }, + { + "epoch": 1.97, + "learning_rate": 1.7646015778690983e-05, + "loss": 0.4019, + "step": 4705 + }, + { + "epoch": 1.97, + "learning_rate": 1.7644922972290344e-05, + "loss": 0.4247, + "step": 4706 + }, + { + "epoch": 1.97, + "learning_rate": 1.7643829946143176e-05, + "loss": 0.4158, + "step": 4707 + }, + { + "epoch": 1.97, + "learning_rate": 1.76427367002809e-05, + "loss": 0.354, + "step": 4708 + }, + { + "epoch": 1.97, + "learning_rate": 1.764164323473494e-05, + "loss": 0.4356, + "step": 4709 + }, + { + "epoch": 1.97, + "learning_rate": 1.7640549549536727e-05, + "loss": 0.38, + "step": 4710 + }, + { + "epoch": 1.97, + "learning_rate": 1.7639455644717694e-05, + "loss": 0.3709, + "step": 4711 + }, + { + "epoch": 1.97, + "learning_rate": 1.7638361520309294e-05, + "loss": 0.4102, + "step": 4712 + }, + { + "epoch": 1.97, + "learning_rate": 1.7637267176342964e-05, + "loss": 0.3607, + "step": 4713 + }, + { + "epoch": 1.97, + "learning_rate": 1.763617261285017e-05, + "loss": 0.417, + "step": 4714 + }, + { + "epoch": 1.97, + "learning_rate": 1.763507782986237e-05, + "loss": 0.3558, + "step": 4715 + }, + { + "epoch": 1.97, + "learning_rate": 1.763398282741103e-05, + "loss": 0.381, + "step": 4716 + }, + { + "epoch": 1.97, + "learning_rate": 1.763288760552763e-05, + "loss": 0.3835, + "step": 4717 + }, + { + "epoch": 1.98, + "learning_rate": 1.763179216424365e-05, + "loss": 0.4098, + "step": 4718 + }, + { + "epoch": 1.98, + "learning_rate": 1.7630696503590577e-05, + "loss": 0.3972, + "step": 4719 + }, + { + "epoch": 1.98, + "learning_rate": 1.7629600623599904e-05, + "loss": 0.4298, + "step": 4720 + }, + { + "epoch": 1.98, + "learning_rate": 1.7628504524303128e-05, + "loss": 0.382, + "step": 4721 + }, + { + "epoch": 1.98, + "learning_rate": 1.7627408205731762e-05, + "loss": 0.43, + "step": 4722 + }, + { + "epoch": 1.98, + "learning_rate": 1.7626311667917315e-05, + "loss": 0.3753, + "step": 4723 + }, + { + "epoch": 1.98, + "learning_rate": 1.7625214910891307e-05, + "loss": 0.4242, + "step": 4724 + }, + { + "epoch": 1.98, + "learning_rate": 1.7624117934685264e-05, + "loss": 0.4101, + "step": 4725 + }, + { + "epoch": 1.98, + "learning_rate": 1.7623020739330714e-05, + "loss": 0.3907, + "step": 4726 + }, + { + "epoch": 1.98, + "learning_rate": 1.7621923324859196e-05, + "loss": 0.3705, + "step": 4727 + }, + { + "epoch": 1.98, + "learning_rate": 1.762082569130226e-05, + "loss": 0.3592, + "step": 4728 + }, + { + "epoch": 1.98, + "learning_rate": 1.761972783869145e-05, + "loss": 0.401, + "step": 4729 + }, + { + "epoch": 1.98, + "learning_rate": 1.7618629767058326e-05, + "loss": 0.4288, + "step": 4730 + }, + { + "epoch": 1.98, + "learning_rate": 1.7617531476434447e-05, + "loss": 0.3446, + "step": 4731 + }, + { + "epoch": 1.98, + "learning_rate": 1.7616432966851386e-05, + "loss": 0.3834, + "step": 4732 + }, + { + "epoch": 1.98, + "learning_rate": 1.761533423834072e-05, + "loss": 0.4065, + "step": 4733 + }, + { + "epoch": 1.98, + "learning_rate": 1.761423529093403e-05, + "loss": 0.4022, + "step": 4734 + }, + { + "epoch": 1.98, + "learning_rate": 1.76131361246629e-05, + "loss": 0.3807, + "step": 4735 + }, + { + "epoch": 1.98, + "learning_rate": 1.7612036739558932e-05, + "loss": 0.3898, + "step": 4736 + }, + { + "epoch": 1.98, + "learning_rate": 1.761093713565372e-05, + "loss": 0.3873, + "step": 4737 + }, + { + "epoch": 1.98, + "learning_rate": 1.7609837312978874e-05, + "loss": 0.4046, + "step": 4738 + }, + { + "epoch": 1.98, + "learning_rate": 1.7608737271566004e-05, + "loss": 0.4069, + "step": 4739 + }, + { + "epoch": 1.98, + "learning_rate": 1.7607637011446737e-05, + "loss": 0.3928, + "step": 4740 + }, + { + "epoch": 1.98, + "learning_rate": 1.7606536532652695e-05, + "loss": 0.3691, + "step": 4741 + }, + { + "epoch": 1.99, + "learning_rate": 1.760543583521551e-05, + "loss": 0.4312, + "step": 4742 + }, + { + "epoch": 1.99, + "learning_rate": 1.7604334919166822e-05, + "loss": 0.3853, + "step": 4743 + }, + { + "epoch": 1.99, + "learning_rate": 1.7603233784538275e-05, + "loss": 0.4136, + "step": 4744 + }, + { + "epoch": 1.99, + "learning_rate": 1.7602132431361516e-05, + "loss": 0.4116, + "step": 4745 + }, + { + "epoch": 1.99, + "learning_rate": 1.7601030859668207e-05, + "loss": 0.3834, + "step": 4746 + }, + { + "epoch": 1.99, + "learning_rate": 1.7599929069490014e-05, + "loss": 0.4323, + "step": 4747 + }, + { + "epoch": 1.99, + "learning_rate": 1.7598827060858605e-05, + "loss": 0.4095, + "step": 4748 + }, + { + "epoch": 1.99, + "learning_rate": 1.759772483380565e-05, + "loss": 0.4095, + "step": 4749 + }, + { + "epoch": 1.99, + "learning_rate": 1.759662238836284e-05, + "loss": 0.4272, + "step": 4750 + }, + { + "epoch": 1.99, + "learning_rate": 1.759551972456186e-05, + "loss": 0.405, + "step": 4751 + }, + { + "epoch": 1.99, + "learning_rate": 1.7594416842434405e-05, + "loss": 0.3806, + "step": 4752 + }, + { + "epoch": 1.99, + "learning_rate": 1.759331374201218e-05, + "loss": 0.3747, + "step": 4753 + }, + { + "epoch": 1.99, + "learning_rate": 1.759221042332689e-05, + "loss": 0.3804, + "step": 4754 + }, + { + "epoch": 1.99, + "learning_rate": 1.759110688641024e-05, + "loss": 0.4165, + "step": 4755 + }, + { + "epoch": 1.99, + "learning_rate": 1.7590003131293967e-05, + "loss": 0.3855, + "step": 4756 + }, + { + "epoch": 1.99, + "learning_rate": 1.7588899158009786e-05, + "loss": 0.373, + "step": 4757 + }, + { + "epoch": 1.99, + "learning_rate": 1.7587794966589437e-05, + "loss": 0.3976, + "step": 4758 + }, + { + "epoch": 1.99, + "learning_rate": 1.758669055706465e-05, + "loss": 0.3288, + "step": 4759 + }, + { + "epoch": 1.99, + "learning_rate": 1.758558592946718e-05, + "loss": 0.3985, + "step": 4760 + }, + { + "epoch": 1.99, + "learning_rate": 1.7584481083828766e-05, + "loss": 0.4002, + "step": 4761 + }, + { + "epoch": 1.99, + "learning_rate": 1.7583376020181183e-05, + "loss": 0.3729, + "step": 4762 + }, + { + "epoch": 1.99, + "learning_rate": 1.758227073855618e-05, + "loss": 0.4247, + "step": 4763 + }, + { + "epoch": 1.99, + "learning_rate": 1.7581165238985533e-05, + "loss": 0.4128, + "step": 4764 + }, + { + "epoch": 1.99, + "learning_rate": 1.7580059521501017e-05, + "loss": 0.3545, + "step": 4765 + }, + { + "epoch": 2.0, + "learning_rate": 1.7578953586134423e-05, + "loss": 0.3787, + "step": 4766 + }, + { + "epoch": 2.0, + "learning_rate": 1.7577847432917528e-05, + "loss": 0.3957, + "step": 4767 + }, + { + "epoch": 2.0, + "learning_rate": 1.7576741061882136e-05, + "loss": 0.3975, + "step": 4768 + }, + { + "epoch": 2.0, + "learning_rate": 1.7575634473060045e-05, + "loss": 0.3689, + "step": 4769 + }, + { + "epoch": 2.0, + "learning_rate": 1.757452766648306e-05, + "loss": 0.4133, + "step": 4770 + }, + { + "epoch": 2.0, + "learning_rate": 1.7573420642183004e-05, + "loss": 0.3487, + "step": 4771 + }, + { + "epoch": 2.0, + "learning_rate": 1.757231340019169e-05, + "loss": 0.3746, + "step": 4772 + }, + { + "epoch": 2.0, + "learning_rate": 1.7571205940540945e-05, + "loss": 0.4144, + "step": 4773 + }, + { + "epoch": 2.0, + "learning_rate": 1.7570098263262608e-05, + "loss": 0.3845, + "step": 4774 + }, + { + "epoch": 2.0, + "learning_rate": 1.756899036838851e-05, + "loss": 0.4326, + "step": 4775 + }, + { + "epoch": 2.0, + "learning_rate": 1.7567882255950503e-05, + "loss": 0.3347, + "step": 4776 + }, + { + "epoch": 2.0, + "learning_rate": 1.7566773925980435e-05, + "loss": 0.3743, + "step": 4777 + }, + { + "epoch": 2.0, + "learning_rate": 1.7565665378510165e-05, + "loss": 0.2799, + "step": 4778 + }, + { + "epoch": 2.0, + "learning_rate": 1.756455661357156e-05, + "loss": 0.315, + "step": 4779 + }, + { + "epoch": 2.0, + "learning_rate": 1.7563447631196484e-05, + "loss": 0.2846, + "step": 4780 + }, + { + "epoch": 2.0, + "learning_rate": 1.7562338431416818e-05, + "loss": 0.262, + "step": 4781 + }, + { + "epoch": 2.0, + "learning_rate": 1.7561229014264448e-05, + "loss": 0.2669, + "step": 4782 + }, + { + "epoch": 2.0, + "learning_rate": 1.7560119379771257e-05, + "loss": 0.2785, + "step": 4783 + }, + { + "epoch": 2.0, + "learning_rate": 1.7559009527969145e-05, + "loss": 0.27, + "step": 4784 + }, + { + "epoch": 2.0, + "learning_rate": 1.755789945889001e-05, + "loss": 0.2426, + "step": 4785 + }, + { + "epoch": 2.0, + "learning_rate": 1.7556789172565762e-05, + "loss": 0.266, + "step": 4786 + }, + { + "epoch": 2.0, + "learning_rate": 1.7555678669028315e-05, + "loss": 0.242, + "step": 4787 + }, + { + "epoch": 2.0, + "learning_rate": 1.755456794830959e-05, + "loss": 0.246, + "step": 4788 + }, + { + "epoch": 2.01, + "learning_rate": 1.7553457010441514e-05, + "loss": 0.2788, + "step": 4789 + }, + { + "epoch": 2.01, + "learning_rate": 1.7552345855456017e-05, + "loss": 0.2595, + "step": 4790 + }, + { + "epoch": 2.01, + "learning_rate": 1.755123448338504e-05, + "loss": 0.2886, + "step": 4791 + }, + { + "epoch": 2.01, + "learning_rate": 1.755012289426053e-05, + "loss": 0.2787, + "step": 4792 + }, + { + "epoch": 2.01, + "learning_rate": 1.754901108811444e-05, + "loss": 0.2442, + "step": 4793 + }, + { + "epoch": 2.01, + "learning_rate": 1.754789906497872e-05, + "loss": 0.2656, + "step": 4794 + }, + { + "epoch": 2.01, + "learning_rate": 1.754678682488534e-05, + "loss": 0.2701, + "step": 4795 + }, + { + "epoch": 2.01, + "learning_rate": 1.7545674367866273e-05, + "loss": 0.2597, + "step": 4796 + }, + { + "epoch": 2.01, + "learning_rate": 1.754456169395349e-05, + "loss": 0.2762, + "step": 4797 + }, + { + "epoch": 2.01, + "learning_rate": 1.7543448803178975e-05, + "loss": 0.2363, + "step": 4798 + }, + { + "epoch": 2.01, + "learning_rate": 1.754233569557472e-05, + "loss": 0.2452, + "step": 4799 + }, + { + "epoch": 2.01, + "learning_rate": 1.7541222371172715e-05, + "loss": 0.2096, + "step": 4800 + }, + { + "epoch": 2.01, + "learning_rate": 1.754010883000497e-05, + "loss": 0.264, + "step": 4801 + }, + { + "epoch": 2.01, + "learning_rate": 1.7538995072103485e-05, + "loss": 0.2442, + "step": 4802 + }, + { + "epoch": 2.01, + "learning_rate": 1.7537881097500277e-05, + "loss": 0.2588, + "step": 4803 + }, + { + "epoch": 2.01, + "learning_rate": 1.7536766906227365e-05, + "loss": 0.2637, + "step": 4804 + }, + { + "epoch": 2.01, + "learning_rate": 1.753565249831678e-05, + "loss": 0.255, + "step": 4805 + }, + { + "epoch": 2.01, + "learning_rate": 1.7534537873800544e-05, + "loss": 0.2544, + "step": 4806 + }, + { + "epoch": 2.01, + "learning_rate": 1.7533423032710705e-05, + "loss": 0.2545, + "step": 4807 + }, + { + "epoch": 2.01, + "learning_rate": 1.753230797507931e-05, + "loss": 0.2724, + "step": 4808 + }, + { + "epoch": 2.01, + "learning_rate": 1.7531192700938406e-05, + "loss": 0.2539, + "step": 4809 + }, + { + "epoch": 2.01, + "learning_rate": 1.7530077210320047e-05, + "loss": 0.2681, + "step": 4810 + }, + { + "epoch": 2.01, + "learning_rate": 1.7528961503256302e-05, + "loss": 0.2636, + "step": 4811 + }, + { + "epoch": 2.01, + "learning_rate": 1.752784557977924e-05, + "loss": 0.2572, + "step": 4812 + }, + { + "epoch": 2.02, + "learning_rate": 1.752672943992094e-05, + "loss": 0.27, + "step": 4813 + }, + { + "epoch": 2.02, + "learning_rate": 1.752561308371348e-05, + "loss": 0.2464, + "step": 4814 + }, + { + "epoch": 2.02, + "learning_rate": 1.7524496511188948e-05, + "loss": 0.2508, + "step": 4815 + }, + { + "epoch": 2.02, + "learning_rate": 1.752337972237944e-05, + "loss": 0.2722, + "step": 4816 + }, + { + "epoch": 2.02, + "learning_rate": 1.752226271731706e-05, + "loss": 0.2406, + "step": 4817 + }, + { + "epoch": 2.02, + "learning_rate": 1.7521145496033912e-05, + "loss": 0.258, + "step": 4818 + }, + { + "epoch": 2.02, + "learning_rate": 1.752002805856211e-05, + "loss": 0.2664, + "step": 4819 + }, + { + "epoch": 2.02, + "learning_rate": 1.7518910404933777e-05, + "loss": 0.2426, + "step": 4820 + }, + { + "epoch": 2.02, + "learning_rate": 1.751779253518104e-05, + "loss": 0.2263, + "step": 4821 + }, + { + "epoch": 2.02, + "learning_rate": 1.751667444933602e-05, + "loss": 0.2425, + "step": 4822 + }, + { + "epoch": 2.02, + "learning_rate": 1.7515556147430867e-05, + "loss": 0.2282, + "step": 4823 + }, + { + "epoch": 2.02, + "learning_rate": 1.751443762949772e-05, + "loss": 0.2368, + "step": 4824 + }, + { + "epoch": 2.02, + "learning_rate": 1.7513318895568734e-05, + "loss": 0.2329, + "step": 4825 + }, + { + "epoch": 2.02, + "learning_rate": 1.7512199945676065e-05, + "loss": 0.2456, + "step": 4826 + }, + { + "epoch": 2.02, + "learning_rate": 1.751108077985187e-05, + "loss": 0.2523, + "step": 4827 + }, + { + "epoch": 2.02, + "learning_rate": 1.7509961398128325e-05, + "loss": 0.2661, + "step": 4828 + }, + { + "epoch": 2.02, + "learning_rate": 1.7508841800537608e-05, + "loss": 0.2412, + "step": 4829 + }, + { + "epoch": 2.02, + "learning_rate": 1.750772198711189e-05, + "loss": 0.242, + "step": 4830 + }, + { + "epoch": 2.02, + "learning_rate": 1.750660195788337e-05, + "loss": 0.2648, + "step": 4831 + }, + { + "epoch": 2.02, + "learning_rate": 1.7505481712884235e-05, + "loss": 0.2752, + "step": 4832 + }, + { + "epoch": 2.02, + "learning_rate": 1.750436125214669e-05, + "loss": 0.2557, + "step": 4833 + }, + { + "epoch": 2.02, + "learning_rate": 1.750324057570294e-05, + "loss": 0.2343, + "step": 4834 + }, + { + "epoch": 2.02, + "learning_rate": 1.75021196835852e-05, + "loss": 0.2432, + "step": 4835 + }, + { + "epoch": 2.02, + "learning_rate": 1.750099857582568e-05, + "loss": 0.2425, + "step": 4836 + }, + { + "epoch": 2.03, + "learning_rate": 1.7499877252456624e-05, + "loss": 0.234, + "step": 4837 + }, + { + "epoch": 2.03, + "learning_rate": 1.7498755713510244e-05, + "loss": 0.2568, + "step": 4838 + }, + { + "epoch": 2.03, + "learning_rate": 1.7497633959018784e-05, + "loss": 0.2783, + "step": 4839 + }, + { + "epoch": 2.03, + "learning_rate": 1.749651198901449e-05, + "loss": 0.2693, + "step": 4840 + }, + { + "epoch": 2.03, + "learning_rate": 1.7495389803529615e-05, + "loss": 0.2427, + "step": 4841 + }, + { + "epoch": 2.03, + "learning_rate": 1.749426740259641e-05, + "loss": 0.2645, + "step": 4842 + }, + { + "epoch": 2.03, + "learning_rate": 1.7493144786247136e-05, + "loss": 0.2245, + "step": 4843 + }, + { + "epoch": 2.03, + "learning_rate": 1.7492021954514068e-05, + "loss": 0.2344, + "step": 4844 + }, + { + "epoch": 2.03, + "learning_rate": 1.7490898907429473e-05, + "loss": 0.2408, + "step": 4845 + }, + { + "epoch": 2.03, + "learning_rate": 1.748977564502564e-05, + "loss": 0.2739, + "step": 4846 + }, + { + "epoch": 2.03, + "learning_rate": 1.748865216733485e-05, + "loss": 0.2663, + "step": 4847 + }, + { + "epoch": 2.03, + "learning_rate": 1.7487528474389403e-05, + "loss": 0.2342, + "step": 4848 + }, + { + "epoch": 2.03, + "learning_rate": 1.748640456622159e-05, + "loss": 0.2576, + "step": 4849 + }, + { + "epoch": 2.03, + "learning_rate": 1.7485280442863718e-05, + "loss": 0.2432, + "step": 4850 + }, + { + "epoch": 2.03, + "learning_rate": 1.7484156104348107e-05, + "loss": 0.2371, + "step": 4851 + }, + { + "epoch": 2.03, + "learning_rate": 1.748303155070707e-05, + "loss": 0.2528, + "step": 4852 + }, + { + "epoch": 2.03, + "learning_rate": 1.7481906781972932e-05, + "loss": 0.2573, + "step": 4853 + }, + { + "epoch": 2.03, + "learning_rate": 1.748078179817802e-05, + "loss": 0.2821, + "step": 4854 + }, + { + "epoch": 2.03, + "learning_rate": 1.7479656599354677e-05, + "loss": 0.2348, + "step": 4855 + }, + { + "epoch": 2.03, + "learning_rate": 1.7478531185535235e-05, + "loss": 0.2659, + "step": 4856 + }, + { + "epoch": 2.03, + "learning_rate": 1.747740555675206e-05, + "loss": 0.2252, + "step": 4857 + }, + { + "epoch": 2.03, + "learning_rate": 1.747627971303749e-05, + "loss": 0.2632, + "step": 4858 + }, + { + "epoch": 2.03, + "learning_rate": 1.74751536544239e-05, + "loss": 0.2315, + "step": 4859 + }, + { + "epoch": 2.03, + "learning_rate": 1.747402738094365e-05, + "loss": 0.2415, + "step": 4860 + }, + { + "epoch": 2.04, + "learning_rate": 1.7472900892629114e-05, + "loss": 0.234, + "step": 4861 + }, + { + "epoch": 2.04, + "learning_rate": 1.747177418951267e-05, + "loss": 0.2695, + "step": 4862 + }, + { + "epoch": 2.04, + "learning_rate": 1.7470647271626712e-05, + "loss": 0.2861, + "step": 4863 + }, + { + "epoch": 2.04, + "learning_rate": 1.7469520139003627e-05, + "loss": 0.2673, + "step": 4864 + }, + { + "epoch": 2.04, + "learning_rate": 1.7468392791675812e-05, + "loss": 0.254, + "step": 4865 + }, + { + "epoch": 2.04, + "learning_rate": 1.7467265229675673e-05, + "loss": 0.2536, + "step": 4866 + }, + { + "epoch": 2.04, + "learning_rate": 1.7466137453035626e-05, + "loss": 0.2572, + "step": 4867 + }, + { + "epoch": 2.04, + "learning_rate": 1.746500946178808e-05, + "loss": 0.2533, + "step": 4868 + }, + { + "epoch": 2.04, + "learning_rate": 1.746388125596546e-05, + "loss": 0.2277, + "step": 4869 + }, + { + "epoch": 2.04, + "learning_rate": 1.7462752835600198e-05, + "loss": 0.226, + "step": 4870 + }, + { + "epoch": 2.04, + "learning_rate": 1.746162420072473e-05, + "loss": 0.2779, + "step": 4871 + }, + { + "epoch": 2.04, + "learning_rate": 1.7460495351371492e-05, + "loss": 0.2378, + "step": 4872 + }, + { + "epoch": 2.04, + "learning_rate": 1.745936628757294e-05, + "loss": 0.2827, + "step": 4873 + }, + { + "epoch": 2.04, + "learning_rate": 1.745823700936152e-05, + "loss": 0.2935, + "step": 4874 + }, + { + "epoch": 2.04, + "learning_rate": 1.7457107516769697e-05, + "loss": 0.2524, + "step": 4875 + }, + { + "epoch": 2.04, + "learning_rate": 1.7455977809829933e-05, + "loss": 0.2445, + "step": 4876 + }, + { + "epoch": 2.04, + "learning_rate": 1.7454847888574705e-05, + "loss": 0.2442, + "step": 4877 + }, + { + "epoch": 2.04, + "learning_rate": 1.745371775303649e-05, + "loss": 0.255, + "step": 4878 + }, + { + "epoch": 2.04, + "learning_rate": 1.7452587403247772e-05, + "loss": 0.2344, + "step": 4879 + }, + { + "epoch": 2.04, + "learning_rate": 1.7451456839241044e-05, + "loss": 0.2695, + "step": 4880 + }, + { + "epoch": 2.04, + "learning_rate": 1.7450326061048798e-05, + "loss": 0.2985, + "step": 4881 + }, + { + "epoch": 2.04, + "learning_rate": 1.7449195068703542e-05, + "loss": 0.2883, + "step": 4882 + }, + { + "epoch": 2.04, + "learning_rate": 1.7448063862237784e-05, + "loss": 0.2591, + "step": 4883 + }, + { + "epoch": 2.04, + "learning_rate": 1.7446932441684044e-05, + "loss": 0.2699, + "step": 4884 + }, + { + "epoch": 2.05, + "learning_rate": 1.7445800807074834e-05, + "loss": 0.2448, + "step": 4885 + }, + { + "epoch": 2.05, + "learning_rate": 1.744466895844269e-05, + "loss": 0.2717, + "step": 4886 + }, + { + "epoch": 2.05, + "learning_rate": 1.7443536895820143e-05, + "loss": 0.2549, + "step": 4887 + }, + { + "epoch": 2.05, + "learning_rate": 1.744240461923973e-05, + "loss": 0.261, + "step": 4888 + }, + { + "epoch": 2.05, + "learning_rate": 1.7441272128734004e-05, + "loss": 0.2292, + "step": 4889 + }, + { + "epoch": 2.05, + "learning_rate": 1.7440139424335513e-05, + "loss": 0.2494, + "step": 4890 + }, + { + "epoch": 2.05, + "learning_rate": 1.7439006506076816e-05, + "loss": 0.2344, + "step": 4891 + }, + { + "epoch": 2.05, + "learning_rate": 1.7437873373990478e-05, + "loss": 0.2476, + "step": 4892 + }, + { + "epoch": 2.05, + "learning_rate": 1.743674002810907e-05, + "loss": 0.2553, + "step": 4893 + }, + { + "epoch": 2.05, + "learning_rate": 1.743560646846517e-05, + "loss": 0.2397, + "step": 4894 + }, + { + "epoch": 2.05, + "learning_rate": 1.7434472695091358e-05, + "loss": 0.2433, + "step": 4895 + }, + { + "epoch": 2.05, + "learning_rate": 1.743333870802023e-05, + "loss": 0.2665, + "step": 4896 + }, + { + "epoch": 2.05, + "learning_rate": 1.7432204507284373e-05, + "loss": 0.2498, + "step": 4897 + }, + { + "epoch": 2.05, + "learning_rate": 1.7431070092916395e-05, + "loss": 0.2239, + "step": 4898 + }, + { + "epoch": 2.05, + "learning_rate": 1.74299354649489e-05, + "loss": 0.2373, + "step": 4899 + }, + { + "epoch": 2.05, + "learning_rate": 1.7428800623414502e-05, + "loss": 0.246, + "step": 4900 + }, + { + "epoch": 2.05, + "learning_rate": 1.7427665568345826e-05, + "loss": 0.2617, + "step": 4901 + }, + { + "epoch": 2.05, + "learning_rate": 1.7426530299775492e-05, + "loss": 0.247, + "step": 4902 + }, + { + "epoch": 2.05, + "learning_rate": 1.7425394817736135e-05, + "loss": 0.2407, + "step": 4903 + }, + { + "epoch": 2.05, + "learning_rate": 1.7424259122260396e-05, + "loss": 0.2377, + "step": 4904 + }, + { + "epoch": 2.05, + "learning_rate": 1.7423123213380914e-05, + "loss": 0.2585, + "step": 4905 + }, + { + "epoch": 2.05, + "learning_rate": 1.742198709113034e-05, + "loss": 0.2342, + "step": 4906 + }, + { + "epoch": 2.05, + "learning_rate": 1.7420850755541335e-05, + "loss": 0.2528, + "step": 4907 + }, + { + "epoch": 2.05, + "learning_rate": 1.7419714206646563e-05, + "loss": 0.2572, + "step": 4908 + }, + { + "epoch": 2.06, + "learning_rate": 1.741857744447869e-05, + "loss": 0.2499, + "step": 4909 + }, + { + "epoch": 2.06, + "learning_rate": 1.7417440469070386e-05, + "loss": 0.2417, + "step": 4910 + }, + { + "epoch": 2.06, + "learning_rate": 1.7416303280454343e-05, + "loss": 0.2383, + "step": 4911 + }, + { + "epoch": 2.06, + "learning_rate": 1.741516587866324e-05, + "loss": 0.2331, + "step": 4912 + }, + { + "epoch": 2.06, + "learning_rate": 1.7414028263729776e-05, + "loss": 0.2658, + "step": 4913 + }, + { + "epoch": 2.06, + "learning_rate": 1.741289043568665e-05, + "loss": 0.2369, + "step": 4914 + }, + { + "epoch": 2.06, + "learning_rate": 1.7411752394566565e-05, + "loss": 0.279, + "step": 4915 + }, + { + "epoch": 2.06, + "learning_rate": 1.7410614140402234e-05, + "loss": 0.2687, + "step": 4916 + }, + { + "epoch": 2.06, + "learning_rate": 1.7409475673226376e-05, + "loss": 0.2613, + "step": 4917 + }, + { + "epoch": 2.06, + "learning_rate": 1.7408336993071716e-05, + "loss": 0.2657, + "step": 4918 + }, + { + "epoch": 2.06, + "learning_rate": 1.7407198099970982e-05, + "loss": 0.2621, + "step": 4919 + }, + { + "epoch": 2.06, + "learning_rate": 1.7406058993956908e-05, + "loss": 0.2789, + "step": 4920 + }, + { + "epoch": 2.06, + "learning_rate": 1.7404919675062245e-05, + "loss": 0.2448, + "step": 4921 + }, + { + "epoch": 2.06, + "learning_rate": 1.7403780143319736e-05, + "loss": 0.2405, + "step": 4922 + }, + { + "epoch": 2.06, + "learning_rate": 1.7402640398762137e-05, + "loss": 0.2727, + "step": 4923 + }, + { + "epoch": 2.06, + "learning_rate": 1.740150044142221e-05, + "loss": 0.2417, + "step": 4924 + }, + { + "epoch": 2.06, + "learning_rate": 1.7400360271332718e-05, + "loss": 0.2379, + "step": 4925 + }, + { + "epoch": 2.06, + "learning_rate": 1.7399219888526438e-05, + "loss": 0.2408, + "step": 4926 + }, + { + "epoch": 2.06, + "learning_rate": 1.739807929303615e-05, + "loss": 0.242, + "step": 4927 + }, + { + "epoch": 2.06, + "learning_rate": 1.7396938484894636e-05, + "loss": 0.251, + "step": 4928 + }, + { + "epoch": 2.06, + "learning_rate": 1.739579746413469e-05, + "loss": 0.2409, + "step": 4929 + }, + { + "epoch": 2.06, + "learning_rate": 1.7394656230789106e-05, + "loss": 0.2426, + "step": 4930 + }, + { + "epoch": 2.06, + "learning_rate": 1.7393514784890694e-05, + "loss": 0.2722, + "step": 4931 + }, + { + "epoch": 2.06, + "learning_rate": 1.7392373126472257e-05, + "loss": 0.2441, + "step": 4932 + }, + { + "epoch": 2.07, + "learning_rate": 1.739123125556662e-05, + "loss": 0.2293, + "step": 4933 + }, + { + "epoch": 2.07, + "learning_rate": 1.7390089172206594e-05, + "loss": 0.2598, + "step": 4934 + }, + { + "epoch": 2.07, + "learning_rate": 1.7388946876425016e-05, + "loss": 0.2543, + "step": 4935 + }, + { + "epoch": 2.07, + "learning_rate": 1.7387804368254713e-05, + "loss": 0.2322, + "step": 4936 + }, + { + "epoch": 2.07, + "learning_rate": 1.738666164772853e-05, + "loss": 0.2598, + "step": 4937 + }, + { + "epoch": 2.07, + "learning_rate": 1.7385518714879315e-05, + "loss": 0.2482, + "step": 4938 + }, + { + "epoch": 2.07, + "learning_rate": 1.7384375569739918e-05, + "loss": 0.2783, + "step": 4939 + }, + { + "epoch": 2.07, + "learning_rate": 1.73832322123432e-05, + "loss": 0.2674, + "step": 4940 + }, + { + "epoch": 2.07, + "learning_rate": 1.738208864272202e-05, + "loss": 0.2536, + "step": 4941 + }, + { + "epoch": 2.07, + "learning_rate": 1.7380944860909253e-05, + "loss": 0.2471, + "step": 4942 + }, + { + "epoch": 2.07, + "learning_rate": 1.7379800866937773e-05, + "loss": 0.2384, + "step": 4943 + }, + { + "epoch": 2.07, + "learning_rate": 1.7378656660840475e-05, + "loss": 0.238, + "step": 4944 + }, + { + "epoch": 2.07, + "learning_rate": 1.7377512242650234e-05, + "loss": 0.2664, + "step": 4945 + }, + { + "epoch": 2.07, + "learning_rate": 1.7376367612399947e-05, + "loss": 0.2523, + "step": 4946 + }, + { + "epoch": 2.07, + "learning_rate": 1.737522277012252e-05, + "loss": 0.2584, + "step": 4947 + }, + { + "epoch": 2.07, + "learning_rate": 1.7374077715850865e-05, + "loss": 0.2505, + "step": 4948 + }, + { + "epoch": 2.07, + "learning_rate": 1.7372932449617884e-05, + "loss": 0.2087, + "step": 4949 + }, + { + "epoch": 2.07, + "learning_rate": 1.7371786971456504e-05, + "loss": 0.2631, + "step": 4950 + }, + { + "epoch": 2.07, + "learning_rate": 1.7370641281399647e-05, + "loss": 0.2492, + "step": 4951 + }, + { + "epoch": 2.07, + "learning_rate": 1.736949537948025e-05, + "loss": 0.2934, + "step": 4952 + }, + { + "epoch": 2.07, + "learning_rate": 1.7368349265731246e-05, + "loss": 0.2197, + "step": 4953 + }, + { + "epoch": 2.07, + "learning_rate": 1.7367202940185583e-05, + "loss": 0.2569, + "step": 4954 + }, + { + "epoch": 2.07, + "learning_rate": 1.7366056402876204e-05, + "loss": 0.2391, + "step": 4955 + }, + { + "epoch": 2.07, + "learning_rate": 1.7364909653836074e-05, + "loss": 0.2742, + "step": 4956 + }, + { + "epoch": 2.08, + "learning_rate": 1.7363762693098154e-05, + "loss": 0.2206, + "step": 4957 + }, + { + "epoch": 2.08, + "learning_rate": 1.7362615520695405e-05, + "loss": 0.2756, + "step": 4958 + }, + { + "epoch": 2.08, + "learning_rate": 1.736146813666081e-05, + "loss": 0.2251, + "step": 4959 + }, + { + "epoch": 2.08, + "learning_rate": 1.7360320541027342e-05, + "loss": 0.2713, + "step": 4960 + }, + { + "epoch": 2.08, + "learning_rate": 1.7359172733827995e-05, + "loss": 0.2465, + "step": 4961 + }, + { + "epoch": 2.08, + "learning_rate": 1.735802471509576e-05, + "loss": 0.2497, + "step": 4962 + }, + { + "epoch": 2.08, + "learning_rate": 1.735687648486363e-05, + "loss": 0.2337, + "step": 4963 + }, + { + "epoch": 2.08, + "learning_rate": 1.7355728043164613e-05, + "loss": 0.2509, + "step": 4964 + }, + { + "epoch": 2.08, + "learning_rate": 1.7354579390031724e-05, + "loss": 0.2376, + "step": 4965 + }, + { + "epoch": 2.08, + "learning_rate": 1.7353430525497974e-05, + "loss": 0.2525, + "step": 4966 + }, + { + "epoch": 2.08, + "learning_rate": 1.7352281449596392e-05, + "loss": 0.2847, + "step": 4967 + }, + { + "epoch": 2.08, + "learning_rate": 1.7351132162360002e-05, + "loss": 0.255, + "step": 4968 + }, + { + "epoch": 2.08, + "learning_rate": 1.734998266382184e-05, + "loss": 0.253, + "step": 4969 + }, + { + "epoch": 2.08, + "learning_rate": 1.734883295401495e-05, + "loss": 0.2528, + "step": 4970 + }, + { + "epoch": 2.08, + "learning_rate": 1.7347683032972377e-05, + "loss": 0.2893, + "step": 4971 + }, + { + "epoch": 2.08, + "learning_rate": 1.7346532900727177e-05, + "loss": 0.2915, + "step": 4972 + }, + { + "epoch": 2.08, + "learning_rate": 1.7345382557312408e-05, + "loss": 0.2729, + "step": 4973 + }, + { + "epoch": 2.08, + "learning_rate": 1.7344232002761138e-05, + "loss": 0.2677, + "step": 4974 + }, + { + "epoch": 2.08, + "learning_rate": 1.734308123710643e-05, + "loss": 0.2816, + "step": 4975 + }, + { + "epoch": 2.08, + "learning_rate": 1.734193026038137e-05, + "loss": 0.2447, + "step": 4976 + }, + { + "epoch": 2.08, + "learning_rate": 1.7340779072619043e-05, + "loss": 0.2602, + "step": 4977 + }, + { + "epoch": 2.08, + "learning_rate": 1.7339627673852534e-05, + "loss": 0.2614, + "step": 4978 + }, + { + "epoch": 2.08, + "learning_rate": 1.733847606411494e-05, + "loss": 0.2689, + "step": 4979 + }, + { + "epoch": 2.08, + "learning_rate": 1.7337324243439363e-05, + "loss": 0.2257, + "step": 4980 + }, + { + "epoch": 2.09, + "learning_rate": 1.7336172211858914e-05, + "loss": 0.2506, + "step": 4981 + }, + { + "epoch": 2.09, + "learning_rate": 1.7335019969406702e-05, + "loss": 0.2852, + "step": 4982 + }, + { + "epoch": 2.09, + "learning_rate": 1.7333867516115853e-05, + "loss": 0.226, + "step": 4983 + }, + { + "epoch": 2.09, + "learning_rate": 1.7332714852019487e-05, + "loss": 0.2242, + "step": 4984 + }, + { + "epoch": 2.09, + "learning_rate": 1.733156197715074e-05, + "loss": 0.2465, + "step": 4985 + }, + { + "epoch": 2.09, + "learning_rate": 1.733040889154275e-05, + "loss": 0.2993, + "step": 4986 + }, + { + "epoch": 2.09, + "learning_rate": 1.7329255595228664e-05, + "loss": 0.2579, + "step": 4987 + }, + { + "epoch": 2.09, + "learning_rate": 1.732810208824163e-05, + "loss": 0.2166, + "step": 4988 + }, + { + "epoch": 2.09, + "learning_rate": 1.73269483706148e-05, + "loss": 0.2538, + "step": 4989 + }, + { + "epoch": 2.09, + "learning_rate": 1.7325794442381344e-05, + "loss": 0.2515, + "step": 4990 + }, + { + "epoch": 2.09, + "learning_rate": 1.7324640303574426e-05, + "loss": 0.2579, + "step": 4991 + }, + { + "epoch": 2.09, + "learning_rate": 1.7323485954227223e-05, + "loss": 0.2537, + "step": 4992 + }, + { + "epoch": 2.09, + "learning_rate": 1.7322331394372916e-05, + "loss": 0.2616, + "step": 4993 + }, + { + "epoch": 2.09, + "learning_rate": 1.732117662404469e-05, + "loss": 0.2548, + "step": 4994 + }, + { + "epoch": 2.09, + "learning_rate": 1.7320021643275734e-05, + "loss": 0.2309, + "step": 4995 + }, + { + "epoch": 2.09, + "learning_rate": 1.731886645209926e-05, + "loss": 0.2589, + "step": 4996 + }, + { + "epoch": 2.09, + "learning_rate": 1.731771105054846e-05, + "loss": 0.2616, + "step": 4997 + }, + { + "epoch": 2.09, + "learning_rate": 1.7316555438656547e-05, + "loss": 0.261, + "step": 4998 + }, + { + "epoch": 2.09, + "learning_rate": 1.7315399616456746e-05, + "loss": 0.2548, + "step": 4999 + }, + { + "epoch": 2.09, + "learning_rate": 1.731424358398227e-05, + "loss": 0.2641, + "step": 5000 + }, + { + "epoch": 2.09, + "learning_rate": 1.7313087341266355e-05, + "loss": 0.232, + "step": 5001 + }, + { + "epoch": 2.09, + "learning_rate": 1.7311930888342234e-05, + "loss": 0.2712, + "step": 5002 + }, + { + "epoch": 2.09, + "learning_rate": 1.7310774225243146e-05, + "loss": 0.2451, + "step": 5003 + }, + { + "epoch": 2.1, + "learning_rate": 1.7309617352002343e-05, + "loss": 0.2562, + "step": 5004 + }, + { + "epoch": 2.1, + "learning_rate": 1.7308460268653077e-05, + "loss": 0.2448, + "step": 5005 + }, + { + "epoch": 2.1, + "learning_rate": 1.73073029752286e-05, + "loss": 0.2452, + "step": 5006 + }, + { + "epoch": 2.1, + "learning_rate": 1.730614547176219e-05, + "loss": 0.2327, + "step": 5007 + }, + { + "epoch": 2.1, + "learning_rate": 1.730498775828711e-05, + "loss": 0.2429, + "step": 5008 + }, + { + "epoch": 2.1, + "learning_rate": 1.7303829834836636e-05, + "loss": 0.2622, + "step": 5009 + }, + { + "epoch": 2.1, + "learning_rate": 1.7302671701444057e-05, + "loss": 0.2703, + "step": 5010 + }, + { + "epoch": 2.1, + "learning_rate": 1.730151335814266e-05, + "loss": 0.2536, + "step": 5011 + }, + { + "epoch": 2.1, + "learning_rate": 1.730035480496574e-05, + "loss": 0.2419, + "step": 5012 + }, + { + "epoch": 2.1, + "learning_rate": 1.72991960419466e-05, + "loss": 0.2594, + "step": 5013 + }, + { + "epoch": 2.1, + "learning_rate": 1.7298037069118546e-05, + "loss": 0.2761, + "step": 5014 + }, + { + "epoch": 2.1, + "learning_rate": 1.7296877886514892e-05, + "loss": 0.2803, + "step": 5015 + }, + { + "epoch": 2.1, + "learning_rate": 1.7295718494168963e-05, + "loss": 0.2587, + "step": 5016 + }, + { + "epoch": 2.1, + "learning_rate": 1.7294558892114077e-05, + "loss": 0.2206, + "step": 5017 + }, + { + "epoch": 2.1, + "learning_rate": 1.729339908038357e-05, + "loss": 0.2495, + "step": 5018 + }, + { + "epoch": 2.1, + "learning_rate": 1.7292239059010777e-05, + "loss": 0.2289, + "step": 5019 + }, + { + "epoch": 2.1, + "learning_rate": 1.7291078828029042e-05, + "loss": 0.2807, + "step": 5020 + }, + { + "epoch": 2.1, + "learning_rate": 1.728991838747172e-05, + "loss": 0.242, + "step": 5021 + }, + { + "epoch": 2.1, + "learning_rate": 1.7288757737372157e-05, + "loss": 0.2491, + "step": 5022 + }, + { + "epoch": 2.1, + "learning_rate": 1.7287596877763725e-05, + "loss": 0.2257, + "step": 5023 + }, + { + "epoch": 2.1, + "learning_rate": 1.7286435808679787e-05, + "loss": 0.2492, + "step": 5024 + }, + { + "epoch": 2.1, + "learning_rate": 1.7285274530153718e-05, + "loss": 0.2578, + "step": 5025 + }, + { + "epoch": 2.1, + "learning_rate": 1.7284113042218895e-05, + "loss": 0.26, + "step": 5026 + }, + { + "epoch": 2.1, + "learning_rate": 1.728295134490871e-05, + "loss": 0.2653, + "step": 5027 + }, + { + "epoch": 2.11, + "learning_rate": 1.728178943825655e-05, + "loss": 0.279, + "step": 5028 + }, + { + "epoch": 2.11, + "learning_rate": 1.7280627322295812e-05, + "loss": 0.2164, + "step": 5029 + }, + { + "epoch": 2.11, + "learning_rate": 1.7279464997059905e-05, + "loss": 0.2582, + "step": 5030 + }, + { + "epoch": 2.11, + "learning_rate": 1.727830246258224e-05, + "loss": 0.2465, + "step": 5031 + }, + { + "epoch": 2.11, + "learning_rate": 1.727713971889622e-05, + "loss": 0.2576, + "step": 5032 + }, + { + "epoch": 2.11, + "learning_rate": 1.7275976766035282e-05, + "loss": 0.2423, + "step": 5033 + }, + { + "epoch": 2.11, + "learning_rate": 1.7274813604032847e-05, + "loss": 0.2154, + "step": 5034 + }, + { + "epoch": 2.11, + "learning_rate": 1.727365023292235e-05, + "loss": 0.2747, + "step": 5035 + }, + { + "epoch": 2.11, + "learning_rate": 1.7272486652737236e-05, + "loss": 0.2407, + "step": 5036 + }, + { + "epoch": 2.11, + "learning_rate": 1.7271322863510942e-05, + "loss": 0.2562, + "step": 5037 + }, + { + "epoch": 2.11, + "learning_rate": 1.7270158865276927e-05, + "loss": 0.2562, + "step": 5038 + }, + { + "epoch": 2.11, + "learning_rate": 1.7268994658068645e-05, + "loss": 0.2782, + "step": 5039 + }, + { + "epoch": 2.11, + "learning_rate": 1.7267830241919562e-05, + "loss": 0.2556, + "step": 5040 + }, + { + "epoch": 2.11, + "learning_rate": 1.726666561686315e-05, + "loss": 0.2463, + "step": 5041 + }, + { + "epoch": 2.11, + "learning_rate": 1.7265500782932885e-05, + "loss": 0.2729, + "step": 5042 + }, + { + "epoch": 2.11, + "learning_rate": 1.7264335740162244e-05, + "loss": 0.223, + "step": 5043 + }, + { + "epoch": 2.11, + "learning_rate": 1.7263170488584717e-05, + "loss": 0.2527, + "step": 5044 + }, + { + "epoch": 2.11, + "learning_rate": 1.7262005028233802e-05, + "loss": 0.2513, + "step": 5045 + }, + { + "epoch": 2.11, + "learning_rate": 1.7260839359142995e-05, + "loss": 0.2431, + "step": 5046 + }, + { + "epoch": 2.11, + "learning_rate": 1.7259673481345808e-05, + "loss": 0.2269, + "step": 5047 + }, + { + "epoch": 2.11, + "learning_rate": 1.7258507394875742e-05, + "loss": 0.249, + "step": 5048 + }, + { + "epoch": 2.11, + "learning_rate": 1.725734109976633e-05, + "loss": 0.2325, + "step": 5049 + }, + { + "epoch": 2.11, + "learning_rate": 1.725617459605108e-05, + "loss": 0.2591, + "step": 5050 + }, + { + "epoch": 2.11, + "learning_rate": 1.7255007883763537e-05, + "loss": 0.2405, + "step": 5051 + }, + { + "epoch": 2.12, + "learning_rate": 1.7253840962937226e-05, + "loss": 0.2675, + "step": 5052 + }, + { + "epoch": 2.12, + "learning_rate": 1.7252673833605697e-05, + "loss": 0.2355, + "step": 5053 + }, + { + "epoch": 2.12, + "learning_rate": 1.725150649580249e-05, + "loss": 0.2465, + "step": 5054 + }, + { + "epoch": 2.12, + "learning_rate": 1.7250338949561167e-05, + "loss": 0.2481, + "step": 5055 + }, + { + "epoch": 2.12, + "learning_rate": 1.7249171194915283e-05, + "loss": 0.2396, + "step": 5056 + }, + { + "epoch": 2.12, + "learning_rate": 1.724800323189841e-05, + "loss": 0.2557, + "step": 5057 + }, + { + "epoch": 2.12, + "learning_rate": 1.724683506054411e-05, + "loss": 0.2235, + "step": 5058 + }, + { + "epoch": 2.12, + "learning_rate": 1.7245666680885975e-05, + "loss": 0.2663, + "step": 5059 + }, + { + "epoch": 2.12, + "learning_rate": 1.724449809295757e-05, + "loss": 0.2585, + "step": 5060 + }, + { + "epoch": 2.12, + "learning_rate": 1.7243329296792502e-05, + "loss": 0.2593, + "step": 5061 + }, + { + "epoch": 2.12, + "learning_rate": 1.7242160292424362e-05, + "loss": 0.2311, + "step": 5062 + }, + { + "epoch": 2.12, + "learning_rate": 1.724099107988675e-05, + "loss": 0.2416, + "step": 5063 + }, + { + "epoch": 2.12, + "learning_rate": 1.7239821659213272e-05, + "loss": 0.2705, + "step": 5064 + }, + { + "epoch": 2.12, + "learning_rate": 1.723865203043755e-05, + "loss": 0.2453, + "step": 5065 + }, + { + "epoch": 2.12, + "learning_rate": 1.7237482193593195e-05, + "loss": 0.2731, + "step": 5066 + }, + { + "epoch": 2.12, + "learning_rate": 1.7236312148713838e-05, + "loss": 0.2653, + "step": 5067 + }, + { + "epoch": 2.12, + "learning_rate": 1.723514189583311e-05, + "loss": 0.2623, + "step": 5068 + }, + { + "epoch": 2.12, + "learning_rate": 1.7233971434984648e-05, + "loss": 0.2766, + "step": 5069 + }, + { + "epoch": 2.12, + "learning_rate": 1.72328007662021e-05, + "loss": 0.2531, + "step": 5070 + }, + { + "epoch": 2.12, + "learning_rate": 1.7231629889519106e-05, + "loss": 0.256, + "step": 5071 + }, + { + "epoch": 2.12, + "learning_rate": 1.723045880496933e-05, + "loss": 0.2576, + "step": 5072 + }, + { + "epoch": 2.12, + "learning_rate": 1.7229287512586432e-05, + "loss": 0.2503, + "step": 5073 + }, + { + "epoch": 2.12, + "learning_rate": 1.7228116012404085e-05, + "loss": 0.2751, + "step": 5074 + }, + { + "epoch": 2.12, + "learning_rate": 1.722694430445595e-05, + "loss": 0.258, + "step": 5075 + }, + { + "epoch": 2.13, + "learning_rate": 1.7225772388775716e-05, + "loss": 0.2496, + "step": 5076 + }, + { + "epoch": 2.13, + "learning_rate": 1.722460026539707e-05, + "loss": 0.2667, + "step": 5077 + }, + { + "epoch": 2.13, + "learning_rate": 1.7223427934353695e-05, + "loss": 0.2329, + "step": 5078 + }, + { + "epoch": 2.13, + "learning_rate": 1.7222255395679298e-05, + "loss": 0.2524, + "step": 5079 + }, + { + "epoch": 2.13, + "learning_rate": 1.7221082649407578e-05, + "loss": 0.25, + "step": 5080 + }, + { + "epoch": 2.13, + "learning_rate": 1.7219909695572244e-05, + "loss": 0.2948, + "step": 5081 + }, + { + "epoch": 2.13, + "learning_rate": 1.7218736534207014e-05, + "loss": 0.2579, + "step": 5082 + }, + { + "epoch": 2.13, + "learning_rate": 1.7217563165345607e-05, + "loss": 0.2617, + "step": 5083 + }, + { + "epoch": 2.13, + "learning_rate": 1.721638958902175e-05, + "loss": 0.2608, + "step": 5084 + }, + { + "epoch": 2.13, + "learning_rate": 1.721521580526918e-05, + "loss": 0.2515, + "step": 5085 + }, + { + "epoch": 2.13, + "learning_rate": 1.7214041814121635e-05, + "loss": 0.2472, + "step": 5086 + }, + { + "epoch": 2.13, + "learning_rate": 1.721286761561286e-05, + "loss": 0.2589, + "step": 5087 + }, + { + "epoch": 2.13, + "learning_rate": 1.7211693209776604e-05, + "loss": 0.2718, + "step": 5088 + }, + { + "epoch": 2.13, + "learning_rate": 1.7210518596646627e-05, + "loss": 0.2498, + "step": 5089 + }, + { + "epoch": 2.13, + "learning_rate": 1.720934377625669e-05, + "loss": 0.2506, + "step": 5090 + }, + { + "epoch": 2.13, + "learning_rate": 1.7208168748640565e-05, + "loss": 0.2388, + "step": 5091 + }, + { + "epoch": 2.13, + "learning_rate": 1.7206993513832026e-05, + "loss": 0.2686, + "step": 5092 + }, + { + "epoch": 2.13, + "learning_rate": 1.7205818071864853e-05, + "loss": 0.2474, + "step": 5093 + }, + { + "epoch": 2.13, + "learning_rate": 1.7204642422772836e-05, + "loss": 0.2179, + "step": 5094 + }, + { + "epoch": 2.13, + "learning_rate": 1.7203466566589763e-05, + "loss": 0.257, + "step": 5095 + }, + { + "epoch": 2.13, + "learning_rate": 1.7202290503349436e-05, + "loss": 0.2763, + "step": 5096 + }, + { + "epoch": 2.13, + "learning_rate": 1.7201114233085664e-05, + "loss": 0.2821, + "step": 5097 + }, + { + "epoch": 2.13, + "learning_rate": 1.719993775583225e-05, + "loss": 0.2528, + "step": 5098 + }, + { + "epoch": 2.13, + "learning_rate": 1.7198761071623014e-05, + "loss": 0.2878, + "step": 5099 + }, + { + "epoch": 2.14, + "learning_rate": 1.7197584180491776e-05, + "loss": 0.2428, + "step": 5100 + }, + { + "epoch": 2.14, + "learning_rate": 1.7196407082472375e-05, + "loss": 0.2827, + "step": 5101 + }, + { + "epoch": 2.14, + "learning_rate": 1.7195229777598635e-05, + "loss": 0.2712, + "step": 5102 + }, + { + "epoch": 2.14, + "learning_rate": 1.71940522659044e-05, + "loss": 0.271, + "step": 5103 + }, + { + "epoch": 2.14, + "learning_rate": 1.7192874547423514e-05, + "loss": 0.223, + "step": 5104 + }, + { + "epoch": 2.14, + "learning_rate": 1.7191696622189836e-05, + "loss": 0.2587, + "step": 5105 + }, + { + "epoch": 2.14, + "learning_rate": 1.719051849023722e-05, + "loss": 0.2812, + "step": 5106 + }, + { + "epoch": 2.14, + "learning_rate": 1.718934015159953e-05, + "loss": 0.258, + "step": 5107 + }, + { + "epoch": 2.14, + "learning_rate": 1.7188161606310635e-05, + "loss": 0.2294, + "step": 5108 + }, + { + "epoch": 2.14, + "learning_rate": 1.7186982854404415e-05, + "loss": 0.256, + "step": 5109 + }, + { + "epoch": 2.14, + "learning_rate": 1.718580389591475e-05, + "loss": 0.252, + "step": 5110 + }, + { + "epoch": 2.14, + "learning_rate": 1.7184624730875528e-05, + "loss": 0.2629, + "step": 5111 + }, + { + "epoch": 2.14, + "learning_rate": 1.7183445359320645e-05, + "loss": 0.2399, + "step": 5112 + }, + { + "epoch": 2.14, + "learning_rate": 1.7182265781284e-05, + "loss": 0.257, + "step": 5113 + }, + { + "epoch": 2.14, + "learning_rate": 1.71810859967995e-05, + "loss": 0.2499, + "step": 5114 + }, + { + "epoch": 2.14, + "learning_rate": 1.717990600590105e-05, + "loss": 0.24, + "step": 5115 + }, + { + "epoch": 2.14, + "learning_rate": 1.7178725808622575e-05, + "loss": 0.2659, + "step": 5116 + }, + { + "epoch": 2.14, + "learning_rate": 1.7177545404997998e-05, + "loss": 0.2335, + "step": 5117 + }, + { + "epoch": 2.14, + "learning_rate": 1.717636479506125e-05, + "loss": 0.2443, + "step": 5118 + }, + { + "epoch": 2.14, + "learning_rate": 1.717518397884626e-05, + "loss": 0.2775, + "step": 5119 + }, + { + "epoch": 2.14, + "learning_rate": 1.717400295638698e-05, + "loss": 0.235, + "step": 5120 + }, + { + "epoch": 2.14, + "learning_rate": 1.7172821727717345e-05, + "loss": 0.2412, + "step": 5121 + }, + { + "epoch": 2.14, + "learning_rate": 1.7171640292871314e-05, + "loss": 0.2423, + "step": 5122 + }, + { + "epoch": 2.14, + "learning_rate": 1.7170458651882847e-05, + "loss": 0.2764, + "step": 5123 + }, + { + "epoch": 2.15, + "learning_rate": 1.716927680478591e-05, + "loss": 0.2641, + "step": 5124 + }, + { + "epoch": 2.15, + "learning_rate": 1.716809475161448e-05, + "loss": 0.2793, + "step": 5125 + }, + { + "epoch": 2.15, + "learning_rate": 1.716691249240252e-05, + "loss": 0.2427, + "step": 5126 + }, + { + "epoch": 2.15, + "learning_rate": 1.716573002718402e-05, + "loss": 0.2391, + "step": 5127 + }, + { + "epoch": 2.15, + "learning_rate": 1.7164547355992972e-05, + "loss": 0.2451, + "step": 5128 + }, + { + "epoch": 2.15, + "learning_rate": 1.7163364478863365e-05, + "loss": 0.2525, + "step": 5129 + }, + { + "epoch": 2.15, + "learning_rate": 1.7162181395829204e-05, + "loss": 0.2202, + "step": 5130 + }, + { + "epoch": 2.15, + "learning_rate": 1.7160998106924494e-05, + "loss": 0.2548, + "step": 5131 + }, + { + "epoch": 2.15, + "learning_rate": 1.715981461218325e-05, + "loss": 0.2345, + "step": 5132 + }, + { + "epoch": 2.15, + "learning_rate": 1.7158630911639488e-05, + "loss": 0.2593, + "step": 5133 + }, + { + "epoch": 2.15, + "learning_rate": 1.715744700532723e-05, + "loss": 0.2576, + "step": 5134 + }, + { + "epoch": 2.15, + "learning_rate": 1.7156262893280512e-05, + "loss": 0.2574, + "step": 5135 + }, + { + "epoch": 2.15, + "learning_rate": 1.7155078575533366e-05, + "loss": 0.2137, + "step": 5136 + }, + { + "epoch": 2.15, + "learning_rate": 1.7153894052119837e-05, + "loss": 0.2537, + "step": 5137 + }, + { + "epoch": 2.15, + "learning_rate": 1.7152709323073972e-05, + "loss": 0.2346, + "step": 5138 + }, + { + "epoch": 2.15, + "learning_rate": 1.7151524388429823e-05, + "loss": 0.2037, + "step": 5139 + }, + { + "epoch": 2.15, + "learning_rate": 1.7150339248221454e-05, + "loss": 0.2163, + "step": 5140 + }, + { + "epoch": 2.15, + "learning_rate": 1.7149153902482926e-05, + "loss": 0.2761, + "step": 5141 + }, + { + "epoch": 2.15, + "learning_rate": 1.7147968351248315e-05, + "loss": 0.2492, + "step": 5142 + }, + { + "epoch": 2.15, + "learning_rate": 1.7146782594551698e-05, + "loss": 0.2327, + "step": 5143 + }, + { + "epoch": 2.15, + "learning_rate": 1.714559663242715e-05, + "loss": 0.2507, + "step": 5144 + }, + { + "epoch": 2.15, + "learning_rate": 1.7144410464908774e-05, + "loss": 0.2678, + "step": 5145 + }, + { + "epoch": 2.15, + "learning_rate": 1.714322409203066e-05, + "loss": 0.2523, + "step": 5146 + }, + { + "epoch": 2.15, + "learning_rate": 1.7142037513826904e-05, + "loss": 0.2563, + "step": 5147 + }, + { + "epoch": 2.16, + "learning_rate": 1.714085073033162e-05, + "loss": 0.2264, + "step": 5148 + }, + { + "epoch": 2.16, + "learning_rate": 1.7139663741578917e-05, + "loss": 0.2479, + "step": 5149 + }, + { + "epoch": 2.16, + "learning_rate": 1.7138476547602917e-05, + "loss": 0.2299, + "step": 5150 + }, + { + "epoch": 2.16, + "learning_rate": 1.7137289148437743e-05, + "loss": 0.2625, + "step": 5151 + }, + { + "epoch": 2.16, + "learning_rate": 1.7136101544117526e-05, + "loss": 0.2654, + "step": 5152 + }, + { + "epoch": 2.16, + "learning_rate": 1.7134913734676398e-05, + "loss": 0.2451, + "step": 5153 + }, + { + "epoch": 2.16, + "learning_rate": 1.7133725720148513e-05, + "loss": 0.2287, + "step": 5154 + }, + { + "epoch": 2.16, + "learning_rate": 1.713253750056801e-05, + "loss": 0.2551, + "step": 5155 + }, + { + "epoch": 2.16, + "learning_rate": 1.7131349075969044e-05, + "loss": 0.2604, + "step": 5156 + }, + { + "epoch": 2.16, + "learning_rate": 1.713016044638578e-05, + "loss": 0.2559, + "step": 5157 + }, + { + "epoch": 2.16, + "learning_rate": 1.7128971611852378e-05, + "loss": 0.2613, + "step": 5158 + }, + { + "epoch": 2.16, + "learning_rate": 1.7127782572403015e-05, + "loss": 0.2773, + "step": 5159 + }, + { + "epoch": 2.16, + "learning_rate": 1.7126593328071867e-05, + "loss": 0.2502, + "step": 5160 + }, + { + "epoch": 2.16, + "learning_rate": 1.7125403878893118e-05, + "loss": 0.2821, + "step": 5161 + }, + { + "epoch": 2.16, + "learning_rate": 1.7124214224900955e-05, + "loss": 0.2593, + "step": 5162 + }, + { + "epoch": 2.16, + "learning_rate": 1.712302436612958e-05, + "loss": 0.2377, + "step": 5163 + }, + { + "epoch": 2.16, + "learning_rate": 1.712183430261319e-05, + "loss": 0.2523, + "step": 5164 + }, + { + "epoch": 2.16, + "learning_rate": 1.712064403438599e-05, + "loss": 0.2733, + "step": 5165 + }, + { + "epoch": 2.16, + "learning_rate": 1.71194535614822e-05, + "loss": 0.2376, + "step": 5166 + }, + { + "epoch": 2.16, + "learning_rate": 1.711826288393603e-05, + "loss": 0.2457, + "step": 5167 + }, + { + "epoch": 2.16, + "learning_rate": 1.7117072001781713e-05, + "loss": 0.2349, + "step": 5168 + }, + { + "epoch": 2.16, + "learning_rate": 1.7115880915053476e-05, + "loss": 0.2544, + "step": 5169 + }, + { + "epoch": 2.16, + "learning_rate": 1.7114689623785556e-05, + "loss": 0.2306, + "step": 5170 + }, + { + "epoch": 2.16, + "learning_rate": 1.7113498128012197e-05, + "loss": 0.2666, + "step": 5171 + }, + { + "epoch": 2.17, + "learning_rate": 1.711230642776765e-05, + "loss": 0.2648, + "step": 5172 + }, + { + "epoch": 2.17, + "learning_rate": 1.711111452308616e-05, + "loss": 0.2542, + "step": 5173 + }, + { + "epoch": 2.17, + "learning_rate": 1.7109922414001997e-05, + "loss": 0.2421, + "step": 5174 + }, + { + "epoch": 2.17, + "learning_rate": 1.710873010054942e-05, + "loss": 0.2632, + "step": 5175 + }, + { + "epoch": 2.17, + "learning_rate": 1.710753758276271e-05, + "loss": 0.2471, + "step": 5176 + }, + { + "epoch": 2.17, + "learning_rate": 1.7106344860676137e-05, + "loss": 0.2528, + "step": 5177 + }, + { + "epoch": 2.17, + "learning_rate": 1.7105151934323988e-05, + "loss": 0.2461, + "step": 5178 + }, + { + "epoch": 2.17, + "learning_rate": 1.7103958803740547e-05, + "loss": 0.2155, + "step": 5179 + }, + { + "epoch": 2.17, + "learning_rate": 1.7102765468960116e-05, + "loss": 0.2406, + "step": 5180 + }, + { + "epoch": 2.17, + "learning_rate": 1.7101571930017e-05, + "loss": 0.2621, + "step": 5181 + }, + { + "epoch": 2.17, + "learning_rate": 1.7100378186945494e-05, + "loss": 0.2702, + "step": 5182 + }, + { + "epoch": 2.17, + "learning_rate": 1.709918423977992e-05, + "loss": 0.256, + "step": 5183 + }, + { + "epoch": 2.17, + "learning_rate": 1.7097990088554595e-05, + "loss": 0.2888, + "step": 5184 + }, + { + "epoch": 2.17, + "learning_rate": 1.7096795733303844e-05, + "loss": 0.2608, + "step": 5185 + }, + { + "epoch": 2.17, + "learning_rate": 1.7095601174061997e-05, + "loss": 0.2615, + "step": 5186 + }, + { + "epoch": 2.17, + "learning_rate": 1.709440641086339e-05, + "loss": 0.2309, + "step": 5187 + }, + { + "epoch": 2.17, + "learning_rate": 1.7093211443742368e-05, + "loss": 0.2574, + "step": 5188 + }, + { + "epoch": 2.17, + "learning_rate": 1.7092016272733274e-05, + "loss": 0.2316, + "step": 5189 + }, + { + "epoch": 2.17, + "learning_rate": 1.709082089787047e-05, + "loss": 0.2482, + "step": 5190 + }, + { + "epoch": 2.17, + "learning_rate": 1.7089625319188308e-05, + "loss": 0.264, + "step": 5191 + }, + { + "epoch": 2.17, + "learning_rate": 1.708842953672116e-05, + "loss": 0.2641, + "step": 5192 + }, + { + "epoch": 2.17, + "learning_rate": 1.7087233550503394e-05, + "loss": 0.2737, + "step": 5193 + }, + { + "epoch": 2.17, + "learning_rate": 1.7086037360569386e-05, + "loss": 0.2495, + "step": 5194 + }, + { + "epoch": 2.18, + "learning_rate": 1.7084840966953526e-05, + "loss": 0.2728, + "step": 5195 + }, + { + "epoch": 2.18, + "learning_rate": 1.7083644369690197e-05, + "loss": 0.2599, + "step": 5196 + }, + { + "epoch": 2.18, + "learning_rate": 1.7082447568813798e-05, + "loss": 0.2787, + "step": 5197 + }, + { + "epoch": 2.18, + "learning_rate": 1.708125056435873e-05, + "loss": 0.2641, + "step": 5198 + }, + { + "epoch": 2.18, + "learning_rate": 1.7080053356359396e-05, + "loss": 0.2461, + "step": 5199 + }, + { + "epoch": 2.18, + "learning_rate": 1.707885594485021e-05, + "loss": 0.2538, + "step": 5200 + }, + { + "epoch": 2.18, + "learning_rate": 1.7077658329865593e-05, + "loss": 0.2623, + "step": 5201 + }, + { + "epoch": 2.18, + "learning_rate": 1.7076460511439966e-05, + "loss": 0.2551, + "step": 5202 + }, + { + "epoch": 2.18, + "learning_rate": 1.7075262489607766e-05, + "loss": 0.2842, + "step": 5203 + }, + { + "epoch": 2.18, + "learning_rate": 1.707406426440342e-05, + "loss": 0.2662, + "step": 5204 + }, + { + "epoch": 2.18, + "learning_rate": 1.7072865835861377e-05, + "loss": 0.2783, + "step": 5205 + }, + { + "epoch": 2.18, + "learning_rate": 1.707166720401608e-05, + "loss": 0.2565, + "step": 5206 + }, + { + "epoch": 2.18, + "learning_rate": 1.7070468368901982e-05, + "loss": 0.2397, + "step": 5207 + }, + { + "epoch": 2.18, + "learning_rate": 1.706926933055355e-05, + "loss": 0.2461, + "step": 5208 + }, + { + "epoch": 2.18, + "learning_rate": 1.706807008900524e-05, + "loss": 0.2689, + "step": 5209 + }, + { + "epoch": 2.18, + "learning_rate": 1.7066870644291527e-05, + "loss": 0.2655, + "step": 5210 + }, + { + "epoch": 2.18, + "learning_rate": 1.7065670996446892e-05, + "loss": 0.2626, + "step": 5211 + }, + { + "epoch": 2.18, + "learning_rate": 1.7064471145505813e-05, + "loss": 0.2378, + "step": 5212 + }, + { + "epoch": 2.18, + "learning_rate": 1.706327109150278e-05, + "loss": 0.2428, + "step": 5213 + }, + { + "epoch": 2.18, + "learning_rate": 1.7062070834472287e-05, + "loss": 0.2442, + "step": 5214 + }, + { + "epoch": 2.18, + "learning_rate": 1.7060870374448833e-05, + "loss": 0.2648, + "step": 5215 + }, + { + "epoch": 2.18, + "learning_rate": 1.705966971146693e-05, + "loss": 0.2259, + "step": 5216 + }, + { + "epoch": 2.18, + "learning_rate": 1.7058468845561084e-05, + "loss": 0.2766, + "step": 5217 + }, + { + "epoch": 2.18, + "learning_rate": 1.7057267776765814e-05, + "loss": 0.2512, + "step": 5218 + }, + { + "epoch": 2.19, + "learning_rate": 1.705606650511564e-05, + "loss": 0.2322, + "step": 5219 + }, + { + "epoch": 2.19, + "learning_rate": 1.7054865030645102e-05, + "loss": 0.2896, + "step": 5220 + }, + { + "epoch": 2.19, + "learning_rate": 1.7053663353388727e-05, + "loss": 0.2412, + "step": 5221 + }, + { + "epoch": 2.19, + "learning_rate": 1.7052461473381056e-05, + "loss": 0.2456, + "step": 5222 + }, + { + "epoch": 2.19, + "learning_rate": 1.7051259390656642e-05, + "loss": 0.2572, + "step": 5223 + }, + { + "epoch": 2.19, + "learning_rate": 1.7050057105250034e-05, + "loss": 0.2393, + "step": 5224 + }, + { + "epoch": 2.19, + "learning_rate": 1.7048854617195788e-05, + "loss": 0.2851, + "step": 5225 + }, + { + "epoch": 2.19, + "learning_rate": 1.7047651926528472e-05, + "loss": 0.2603, + "step": 5226 + }, + { + "epoch": 2.19, + "learning_rate": 1.7046449033282656e-05, + "loss": 0.2448, + "step": 5227 + }, + { + "epoch": 2.19, + "learning_rate": 1.7045245937492912e-05, + "loss": 0.2544, + "step": 5228 + }, + { + "epoch": 2.19, + "learning_rate": 1.704404263919383e-05, + "loss": 0.2433, + "step": 5229 + }, + { + "epoch": 2.19, + "learning_rate": 1.7042839138419992e-05, + "loss": 0.2589, + "step": 5230 + }, + { + "epoch": 2.19, + "learning_rate": 1.704163543520599e-05, + "loss": 0.2382, + "step": 5231 + }, + { + "epoch": 2.19, + "learning_rate": 1.7040431529586427e-05, + "loss": 0.2312, + "step": 5232 + }, + { + "epoch": 2.19, + "learning_rate": 1.703922742159591e-05, + "loss": 0.2521, + "step": 5233 + }, + { + "epoch": 2.19, + "learning_rate": 1.7038023111269045e-05, + "loss": 0.258, + "step": 5234 + }, + { + "epoch": 2.19, + "learning_rate": 1.7036818598640456e-05, + "loss": 0.2897, + "step": 5235 + }, + { + "epoch": 2.19, + "learning_rate": 1.7035613883744753e-05, + "loss": 0.2337, + "step": 5236 + }, + { + "epoch": 2.19, + "learning_rate": 1.7034408966616577e-05, + "loss": 0.2502, + "step": 5237 + }, + { + "epoch": 2.19, + "learning_rate": 1.7033203847290553e-05, + "loss": 0.2469, + "step": 5238 + }, + { + "epoch": 2.19, + "learning_rate": 1.703199852580133e-05, + "loss": 0.2517, + "step": 5239 + }, + { + "epoch": 2.19, + "learning_rate": 1.7030793002183546e-05, + "loss": 0.2594, + "step": 5240 + }, + { + "epoch": 2.19, + "learning_rate": 1.702958727647186e-05, + "loss": 0.2306, + "step": 5241 + }, + { + "epoch": 2.19, + "learning_rate": 1.7028381348700926e-05, + "loss": 0.2449, + "step": 5242 + }, + { + "epoch": 2.2, + "learning_rate": 1.70271752189054e-05, + "loss": 0.2469, + "step": 5243 + }, + { + "epoch": 2.2, + "learning_rate": 1.7025968887119962e-05, + "loss": 0.3028, + "step": 5244 + }, + { + "epoch": 2.2, + "learning_rate": 1.702476235337929e-05, + "loss": 0.2763, + "step": 5245 + }, + { + "epoch": 2.2, + "learning_rate": 1.702355561771805e-05, + "loss": 0.2196, + "step": 5246 + }, + { + "epoch": 2.2, + "learning_rate": 1.7022348680170936e-05, + "loss": 0.2704, + "step": 5247 + }, + { + "epoch": 2.2, + "learning_rate": 1.7021141540772645e-05, + "loss": 0.2302, + "step": 5248 + }, + { + "epoch": 2.2, + "learning_rate": 1.7019934199557868e-05, + "loss": 0.2421, + "step": 5249 + }, + { + "epoch": 2.2, + "learning_rate": 1.7018726656561314e-05, + "loss": 0.254, + "step": 5250 + }, + { + "epoch": 2.2, + "learning_rate": 1.7017518911817686e-05, + "loss": 0.251, + "step": 5251 + }, + { + "epoch": 2.2, + "learning_rate": 1.7016310965361708e-05, + "loss": 0.2674, + "step": 5252 + }, + { + "epoch": 2.2, + "learning_rate": 1.70151028172281e-05, + "loss": 0.268, + "step": 5253 + }, + { + "epoch": 2.2, + "learning_rate": 1.701389446745158e-05, + "loss": 0.2271, + "step": 5254 + }, + { + "epoch": 2.2, + "learning_rate": 1.7012685916066894e-05, + "loss": 0.2591, + "step": 5255 + }, + { + "epoch": 2.2, + "learning_rate": 1.7011477163108766e-05, + "loss": 0.2526, + "step": 5256 + }, + { + "epoch": 2.2, + "learning_rate": 1.7010268208611958e-05, + "loss": 0.271, + "step": 5257 + }, + { + "epoch": 2.2, + "learning_rate": 1.7009059052611206e-05, + "loss": 0.238, + "step": 5258 + }, + { + "epoch": 2.2, + "learning_rate": 1.7007849695141275e-05, + "loss": 0.2589, + "step": 5259 + }, + { + "epoch": 2.2, + "learning_rate": 1.700664013623692e-05, + "loss": 0.2659, + "step": 5260 + }, + { + "epoch": 2.2, + "learning_rate": 1.700543037593291e-05, + "loss": 0.2963, + "step": 5261 + }, + { + "epoch": 2.2, + "learning_rate": 1.7004220414264025e-05, + "loss": 0.2727, + "step": 5262 + }, + { + "epoch": 2.2, + "learning_rate": 1.7003010251265036e-05, + "loss": 0.2739, + "step": 5263 + }, + { + "epoch": 2.2, + "learning_rate": 1.7001799886970735e-05, + "loss": 0.2541, + "step": 5264 + }, + { + "epoch": 2.2, + "learning_rate": 1.7000589321415906e-05, + "loss": 0.2445, + "step": 5265 + }, + { + "epoch": 2.2, + "learning_rate": 1.699937855463535e-05, + "loss": 0.2615, + "step": 5266 + }, + { + "epoch": 2.21, + "learning_rate": 1.6998167586663868e-05, + "loss": 0.2552, + "step": 5267 + }, + { + "epoch": 2.21, + "learning_rate": 1.699695641753627e-05, + "loss": 0.2752, + "step": 5268 + }, + { + "epoch": 2.21, + "learning_rate": 1.6995745047287373e-05, + "loss": 0.2623, + "step": 5269 + }, + { + "epoch": 2.21, + "learning_rate": 1.6994533475951985e-05, + "loss": 0.2388, + "step": 5270 + }, + { + "epoch": 2.21, + "learning_rate": 1.6993321703564946e-05, + "loss": 0.2854, + "step": 5271 + }, + { + "epoch": 2.21, + "learning_rate": 1.6992109730161077e-05, + "loss": 0.264, + "step": 5272 + }, + { + "epoch": 2.21, + "learning_rate": 1.699089755577522e-05, + "loss": 0.2667, + "step": 5273 + }, + { + "epoch": 2.21, + "learning_rate": 1.6989685180442212e-05, + "loss": 0.2484, + "step": 5274 + }, + { + "epoch": 2.21, + "learning_rate": 1.6988472604196907e-05, + "loss": 0.2643, + "step": 5275 + }, + { + "epoch": 2.21, + "learning_rate": 1.698725982707416e-05, + "loss": 0.2616, + "step": 5276 + }, + { + "epoch": 2.21, + "learning_rate": 1.6986046849108834e-05, + "loss": 0.2566, + "step": 5277 + }, + { + "epoch": 2.21, + "learning_rate": 1.6984833670335783e-05, + "loss": 0.26, + "step": 5278 + }, + { + "epoch": 2.21, + "learning_rate": 1.698362029078989e-05, + "loss": 0.2434, + "step": 5279 + }, + { + "epoch": 2.21, + "learning_rate": 1.698240671050603e-05, + "loss": 0.2577, + "step": 5280 + }, + { + "epoch": 2.21, + "learning_rate": 1.6981192929519087e-05, + "loss": 0.2326, + "step": 5281 + }, + { + "epoch": 2.21, + "learning_rate": 1.6979978947863946e-05, + "loss": 0.2519, + "step": 5282 + }, + { + "epoch": 2.21, + "learning_rate": 1.69787647655755e-05, + "loss": 0.2774, + "step": 5283 + }, + { + "epoch": 2.21, + "learning_rate": 1.697755038268866e-05, + "loss": 0.2594, + "step": 5284 + }, + { + "epoch": 2.21, + "learning_rate": 1.6976335799238318e-05, + "loss": 0.2421, + "step": 5285 + }, + { + "epoch": 2.21, + "learning_rate": 1.69751210152594e-05, + "loss": 0.2378, + "step": 5286 + }, + { + "epoch": 2.21, + "learning_rate": 1.697390603078682e-05, + "loss": 0.2452, + "step": 5287 + }, + { + "epoch": 2.21, + "learning_rate": 1.6972690845855494e-05, + "loss": 0.2753, + "step": 5288 + }, + { + "epoch": 2.21, + "learning_rate": 1.697147546050036e-05, + "loss": 0.2547, + "step": 5289 + }, + { + "epoch": 2.21, + "learning_rate": 1.6970259874756353e-05, + "loss": 0.2338, + "step": 5290 + }, + { + "epoch": 2.22, + "learning_rate": 1.696904408865841e-05, + "loss": 0.2653, + "step": 5291 + }, + { + "epoch": 2.22, + "learning_rate": 1.6967828102241475e-05, + "loss": 0.2689, + "step": 5292 + }, + { + "epoch": 2.22, + "learning_rate": 1.6966611915540506e-05, + "loss": 0.2678, + "step": 5293 + }, + { + "epoch": 2.22, + "learning_rate": 1.696539552859046e-05, + "loss": 0.2517, + "step": 5294 + }, + { + "epoch": 2.22, + "learning_rate": 1.69641789414263e-05, + "loss": 0.2531, + "step": 5295 + }, + { + "epoch": 2.22, + "learning_rate": 1.6962962154083e-05, + "loss": 0.263, + "step": 5296 + }, + { + "epoch": 2.22, + "learning_rate": 1.6961745166595528e-05, + "loss": 0.26, + "step": 5297 + }, + { + "epoch": 2.22, + "learning_rate": 1.696052797899887e-05, + "loss": 0.2513, + "step": 5298 + }, + { + "epoch": 2.22, + "learning_rate": 1.695931059132801e-05, + "loss": 0.2297, + "step": 5299 + }, + { + "epoch": 2.22, + "learning_rate": 1.6958093003617942e-05, + "loss": 0.2827, + "step": 5300 + }, + { + "epoch": 2.22, + "learning_rate": 1.695687521590367e-05, + "loss": 0.2749, + "step": 5301 + }, + { + "epoch": 2.22, + "learning_rate": 1.695565722822019e-05, + "loss": 0.2556, + "step": 5302 + }, + { + "epoch": 2.22, + "learning_rate": 1.6954439040602516e-05, + "loss": 0.2822, + "step": 5303 + }, + { + "epoch": 2.22, + "learning_rate": 1.6953220653085662e-05, + "loss": 0.2487, + "step": 5304 + }, + { + "epoch": 2.22, + "learning_rate": 1.6952002065704652e-05, + "loss": 0.2533, + "step": 5305 + }, + { + "epoch": 2.22, + "learning_rate": 1.6950783278494513e-05, + "loss": 0.241, + "step": 5306 + }, + { + "epoch": 2.22, + "learning_rate": 1.6949564291490274e-05, + "loss": 0.2709, + "step": 5307 + }, + { + "epoch": 2.22, + "learning_rate": 1.694834510472698e-05, + "loss": 0.2417, + "step": 5308 + }, + { + "epoch": 2.22, + "learning_rate": 1.6947125718239664e-05, + "loss": 0.2468, + "step": 5309 + }, + { + "epoch": 2.22, + "learning_rate": 1.694590613206339e-05, + "loss": 0.2712, + "step": 5310 + }, + { + "epoch": 2.22, + "learning_rate": 1.6944686346233207e-05, + "loss": 0.2649, + "step": 5311 + }, + { + "epoch": 2.22, + "learning_rate": 1.694346636078418e-05, + "loss": 0.25, + "step": 5312 + }, + { + "epoch": 2.22, + "learning_rate": 1.6942246175751372e-05, + "loss": 0.2659, + "step": 5313 + }, + { + "epoch": 2.22, + "learning_rate": 1.6941025791169855e-05, + "loss": 0.2657, + "step": 5314 + }, + { + "epoch": 2.23, + "learning_rate": 1.6939805207074714e-05, + "loss": 0.253, + "step": 5315 + }, + { + "epoch": 2.23, + "learning_rate": 1.693858442350103e-05, + "loss": 0.2573, + "step": 5316 + }, + { + "epoch": 2.23, + "learning_rate": 1.6937363440483892e-05, + "loss": 0.2307, + "step": 5317 + }, + { + "epoch": 2.23, + "learning_rate": 1.6936142258058403e-05, + "loss": 0.2418, + "step": 5318 + }, + { + "epoch": 2.23, + "learning_rate": 1.6934920876259657e-05, + "loss": 0.2507, + "step": 5319 + }, + { + "epoch": 2.23, + "learning_rate": 1.6933699295122766e-05, + "loss": 0.2765, + "step": 5320 + }, + { + "epoch": 2.23, + "learning_rate": 1.6932477514682838e-05, + "loss": 0.2427, + "step": 5321 + }, + { + "epoch": 2.23, + "learning_rate": 1.6931255534974997e-05, + "loss": 0.2684, + "step": 5322 + }, + { + "epoch": 2.23, + "learning_rate": 1.6930033356034368e-05, + "loss": 0.2533, + "step": 5323 + }, + { + "epoch": 2.23, + "learning_rate": 1.692881097789608e-05, + "loss": 0.2579, + "step": 5324 + }, + { + "epoch": 2.23, + "learning_rate": 1.6927588400595265e-05, + "loss": 0.2453, + "step": 5325 + }, + { + "epoch": 2.23, + "learning_rate": 1.6926365624167075e-05, + "loss": 0.2447, + "step": 5326 + }, + { + "epoch": 2.23, + "learning_rate": 1.6925142648646647e-05, + "loss": 0.2478, + "step": 5327 + }, + { + "epoch": 2.23, + "learning_rate": 1.692391947406914e-05, + "loss": 0.2556, + "step": 5328 + }, + { + "epoch": 2.23, + "learning_rate": 1.692269610046971e-05, + "loss": 0.2543, + "step": 5329 + }, + { + "epoch": 2.23, + "learning_rate": 1.6921472527883527e-05, + "loss": 0.2752, + "step": 5330 + }, + { + "epoch": 2.23, + "learning_rate": 1.6920248756345757e-05, + "loss": 0.2648, + "step": 5331 + }, + { + "epoch": 2.23, + "learning_rate": 1.6919024785891576e-05, + "loss": 0.2655, + "step": 5332 + }, + { + "epoch": 2.23, + "learning_rate": 1.691780061655617e-05, + "loss": 0.2572, + "step": 5333 + }, + { + "epoch": 2.23, + "learning_rate": 1.691657624837472e-05, + "loss": 0.2709, + "step": 5334 + }, + { + "epoch": 2.23, + "learning_rate": 1.6915351681382427e-05, + "loss": 0.2539, + "step": 5335 + }, + { + "epoch": 2.23, + "learning_rate": 1.6914126915614483e-05, + "loss": 0.2223, + "step": 5336 + }, + { + "epoch": 2.23, + "learning_rate": 1.69129019511061e-05, + "loss": 0.263, + "step": 5337 + }, + { + "epoch": 2.23, + "learning_rate": 1.6911676787892484e-05, + "loss": 0.2504, + "step": 5338 + }, + { + "epoch": 2.24, + "learning_rate": 1.691045142600885e-05, + "loss": 0.2343, + "step": 5339 + }, + { + "epoch": 2.24, + "learning_rate": 1.6909225865490424e-05, + "loss": 0.2212, + "step": 5340 + }, + { + "epoch": 2.24, + "learning_rate": 1.6908000106372432e-05, + "loss": 0.2872, + "step": 5341 + }, + { + "epoch": 2.24, + "learning_rate": 1.6906774148690106e-05, + "loss": 0.2537, + "step": 5342 + }, + { + "epoch": 2.24, + "learning_rate": 1.6905547992478687e-05, + "loss": 0.2897, + "step": 5343 + }, + { + "epoch": 2.24, + "learning_rate": 1.6904321637773414e-05, + "loss": 0.2736, + "step": 5344 + }, + { + "epoch": 2.24, + "learning_rate": 1.690309508460955e-05, + "loss": 0.2745, + "step": 5345 + }, + { + "epoch": 2.24, + "learning_rate": 1.6901868333022337e-05, + "loss": 0.2602, + "step": 5346 + }, + { + "epoch": 2.24, + "learning_rate": 1.6900641383047046e-05, + "loss": 0.2841, + "step": 5347 + }, + { + "epoch": 2.24, + "learning_rate": 1.6899414234718943e-05, + "loss": 0.2467, + "step": 5348 + }, + { + "epoch": 2.24, + "learning_rate": 1.68981868880733e-05, + "loss": 0.2693, + "step": 5349 + }, + { + "epoch": 2.24, + "learning_rate": 1.6896959343145396e-05, + "loss": 0.267, + "step": 5350 + }, + { + "epoch": 2.24, + "learning_rate": 1.6895731599970516e-05, + "loss": 0.2706, + "step": 5351 + }, + { + "epoch": 2.24, + "learning_rate": 1.689450365858395e-05, + "loss": 0.2446, + "step": 5352 + }, + { + "epoch": 2.24, + "learning_rate": 1.689327551902099e-05, + "loss": 0.2822, + "step": 5353 + }, + { + "epoch": 2.24, + "learning_rate": 1.6892047181316952e-05, + "loss": 0.2618, + "step": 5354 + }, + { + "epoch": 2.24, + "learning_rate": 1.6890818645507126e-05, + "loss": 0.2618, + "step": 5355 + }, + { + "epoch": 2.24, + "learning_rate": 1.6889589911626834e-05, + "loss": 0.2688, + "step": 5356 + }, + { + "epoch": 2.24, + "learning_rate": 1.6888360979711395e-05, + "loss": 0.2756, + "step": 5357 + }, + { + "epoch": 2.24, + "learning_rate": 1.6887131849796132e-05, + "loss": 0.2537, + "step": 5358 + }, + { + "epoch": 2.24, + "learning_rate": 1.6885902521916375e-05, + "loss": 0.2688, + "step": 5359 + }, + { + "epoch": 2.24, + "learning_rate": 1.6884672996107464e-05, + "loss": 0.2811, + "step": 5360 + }, + { + "epoch": 2.24, + "learning_rate": 1.6883443272404735e-05, + "loss": 0.2519, + "step": 5361 + }, + { + "epoch": 2.24, + "learning_rate": 1.688221335084353e-05, + "loss": 0.2722, + "step": 5362 + }, + { + "epoch": 2.25, + "learning_rate": 1.6880983231459217e-05, + "loss": 0.2626, + "step": 5363 + }, + { + "epoch": 2.25, + "learning_rate": 1.687975291428715e-05, + "loss": 0.2585, + "step": 5364 + }, + { + "epoch": 2.25, + "learning_rate": 1.6878522399362683e-05, + "loss": 0.2621, + "step": 5365 + }, + { + "epoch": 2.25, + "learning_rate": 1.68772916867212e-05, + "loss": 0.218, + "step": 5366 + }, + { + "epoch": 2.25, + "learning_rate": 1.6876060776398063e-05, + "loss": 0.2472, + "step": 5367 + }, + { + "epoch": 2.25, + "learning_rate": 1.6874829668428667e-05, + "loss": 0.2508, + "step": 5368 + }, + { + "epoch": 2.25, + "learning_rate": 1.687359836284839e-05, + "loss": 0.2585, + "step": 5369 + }, + { + "epoch": 2.25, + "learning_rate": 1.687236685969263e-05, + "loss": 0.2585, + "step": 5370 + }, + { + "epoch": 2.25, + "learning_rate": 1.6871135158996782e-05, + "loss": 0.2454, + "step": 5371 + }, + { + "epoch": 2.25, + "learning_rate": 1.686990326079625e-05, + "loss": 0.2705, + "step": 5372 + }, + { + "epoch": 2.25, + "learning_rate": 1.6868671165126445e-05, + "loss": 0.299, + "step": 5373 + }, + { + "epoch": 2.25, + "learning_rate": 1.6867438872022782e-05, + "loss": 0.2597, + "step": 5374 + }, + { + "epoch": 2.25, + "learning_rate": 1.6866206381520683e-05, + "loss": 0.2684, + "step": 5375 + }, + { + "epoch": 2.25, + "learning_rate": 1.6864973693655577e-05, + "loss": 0.243, + "step": 5376 + }, + { + "epoch": 2.25, + "learning_rate": 1.686374080846289e-05, + "loss": 0.2693, + "step": 5377 + }, + { + "epoch": 2.25, + "learning_rate": 1.6862507725978067e-05, + "loss": 0.2694, + "step": 5378 + }, + { + "epoch": 2.25, + "learning_rate": 1.686127444623655e-05, + "loss": 0.2304, + "step": 5379 + }, + { + "epoch": 2.25, + "learning_rate": 1.686004096927378e-05, + "loss": 0.2439, + "step": 5380 + }, + { + "epoch": 2.25, + "learning_rate": 1.685880729512523e-05, + "loss": 0.2477, + "step": 5381 + }, + { + "epoch": 2.25, + "learning_rate": 1.6857573423826344e-05, + "loss": 0.264, + "step": 5382 + }, + { + "epoch": 2.25, + "learning_rate": 1.6856339355412602e-05, + "loss": 0.233, + "step": 5383 + }, + { + "epoch": 2.25, + "learning_rate": 1.6855105089919467e-05, + "loss": 0.2768, + "step": 5384 + }, + { + "epoch": 2.25, + "learning_rate": 1.6853870627382416e-05, + "loss": 0.2539, + "step": 5385 + }, + { + "epoch": 2.25, + "learning_rate": 1.6852635967836936e-05, + "loss": 0.2581, + "step": 5386 + }, + { + "epoch": 2.26, + "learning_rate": 1.685140111131852e-05, + "loss": 0.2704, + "step": 5387 + }, + { + "epoch": 2.26, + "learning_rate": 1.6850166057862654e-05, + "loss": 0.2789, + "step": 5388 + }, + { + "epoch": 2.26, + "learning_rate": 1.6848930807504846e-05, + "loss": 0.2843, + "step": 5389 + }, + { + "epoch": 2.26, + "learning_rate": 1.68476953602806e-05, + "loss": 0.2948, + "step": 5390 + }, + { + "epoch": 2.26, + "learning_rate": 1.684645971622543e-05, + "loss": 0.255, + "step": 5391 + }, + { + "epoch": 2.26, + "learning_rate": 1.6845223875374845e-05, + "loss": 0.2635, + "step": 5392 + }, + { + "epoch": 2.26, + "learning_rate": 1.6843987837764375e-05, + "loss": 0.2963, + "step": 5393 + }, + { + "epoch": 2.26, + "learning_rate": 1.684275160342955e-05, + "loss": 0.2592, + "step": 5394 + }, + { + "epoch": 2.26, + "learning_rate": 1.68415151724059e-05, + "loss": 0.2682, + "step": 5395 + }, + { + "epoch": 2.26, + "learning_rate": 1.684027854472897e-05, + "loss": 0.2681, + "step": 5396 + }, + { + "epoch": 2.26, + "learning_rate": 1.6839041720434303e-05, + "loss": 0.2474, + "step": 5397 + }, + { + "epoch": 2.26, + "learning_rate": 1.683780469955745e-05, + "loss": 0.2892, + "step": 5398 + }, + { + "epoch": 2.26, + "learning_rate": 1.683656748213397e-05, + "loss": 0.2526, + "step": 5399 + }, + { + "epoch": 2.26, + "learning_rate": 1.6835330068199424e-05, + "loss": 0.2477, + "step": 5400 + }, + { + "epoch": 2.26, + "learning_rate": 1.683409245778938e-05, + "loss": 0.2851, + "step": 5401 + }, + { + "epoch": 2.26, + "learning_rate": 1.683285465093941e-05, + "loss": 0.2454, + "step": 5402 + }, + { + "epoch": 2.26, + "learning_rate": 1.6831616647685102e-05, + "loss": 0.2666, + "step": 5403 + }, + { + "epoch": 2.26, + "learning_rate": 1.683037844806203e-05, + "loss": 0.2602, + "step": 5404 + }, + { + "epoch": 2.26, + "learning_rate": 1.6829140052105795e-05, + "loss": 0.2423, + "step": 5405 + }, + { + "epoch": 2.26, + "learning_rate": 1.6827901459851986e-05, + "loss": 0.2727, + "step": 5406 + }, + { + "epoch": 2.26, + "learning_rate": 1.6826662671336212e-05, + "loss": 0.2452, + "step": 5407 + }, + { + "epoch": 2.26, + "learning_rate": 1.6825423686594075e-05, + "loss": 0.2498, + "step": 5408 + }, + { + "epoch": 2.26, + "learning_rate": 1.682418450566119e-05, + "loss": 0.2509, + "step": 5409 + }, + { + "epoch": 2.27, + "learning_rate": 1.6822945128573182e-05, + "loss": 0.2815, + "step": 5410 + }, + { + "epoch": 2.27, + "learning_rate": 1.6821705555365665e-05, + "loss": 0.2747, + "step": 5411 + }, + { + "epoch": 2.27, + "learning_rate": 1.6820465786074277e-05, + "loss": 0.2632, + "step": 5412 + }, + { + "epoch": 2.27, + "learning_rate": 1.6819225820734656e-05, + "loss": 0.2268, + "step": 5413 + }, + { + "epoch": 2.27, + "learning_rate": 1.6817985659382437e-05, + "loss": 0.2475, + "step": 5414 + }, + { + "epoch": 2.27, + "learning_rate": 1.681674530205327e-05, + "loss": 0.2406, + "step": 5415 + }, + { + "epoch": 2.27, + "learning_rate": 1.681550474878281e-05, + "loss": 0.2551, + "step": 5416 + }, + { + "epoch": 2.27, + "learning_rate": 1.6814263999606713e-05, + "loss": 0.2745, + "step": 5417 + }, + { + "epoch": 2.27, + "learning_rate": 1.681302305456064e-05, + "loss": 0.2644, + "step": 5418 + }, + { + "epoch": 2.27, + "learning_rate": 1.6811781913680273e-05, + "loss": 0.2814, + "step": 5419 + }, + { + "epoch": 2.27, + "learning_rate": 1.6810540577001277e-05, + "loss": 0.2529, + "step": 5420 + }, + { + "epoch": 2.27, + "learning_rate": 1.6809299044559333e-05, + "loss": 0.2791, + "step": 5421 + }, + { + "epoch": 2.27, + "learning_rate": 1.6808057316390128e-05, + "loss": 0.255, + "step": 5422 + }, + { + "epoch": 2.27, + "learning_rate": 1.6806815392529363e-05, + "loss": 0.26, + "step": 5423 + }, + { + "epoch": 2.27, + "learning_rate": 1.680557327301273e-05, + "loss": 0.256, + "step": 5424 + }, + { + "epoch": 2.27, + "learning_rate": 1.6804330957875926e-05, + "loss": 0.2486, + "step": 5425 + }, + { + "epoch": 2.27, + "learning_rate": 1.6803088447154672e-05, + "loss": 0.2407, + "step": 5426 + }, + { + "epoch": 2.27, + "learning_rate": 1.6801845740884678e-05, + "loss": 0.2346, + "step": 5427 + }, + { + "epoch": 2.27, + "learning_rate": 1.680060283910166e-05, + "loss": 0.2524, + "step": 5428 + }, + { + "epoch": 2.27, + "learning_rate": 1.6799359741841354e-05, + "loss": 0.2708, + "step": 5429 + }, + { + "epoch": 2.27, + "learning_rate": 1.6798116449139483e-05, + "loss": 0.2632, + "step": 5430 + }, + { + "epoch": 2.27, + "learning_rate": 1.6796872961031783e-05, + "loss": 0.2237, + "step": 5431 + }, + { + "epoch": 2.27, + "learning_rate": 1.6795629277554003e-05, + "loss": 0.2382, + "step": 5432 + }, + { + "epoch": 2.27, + "learning_rate": 1.6794385398741894e-05, + "loss": 0.2691, + "step": 5433 + }, + { + "epoch": 2.28, + "learning_rate": 1.67931413246312e-05, + "loss": 0.2636, + "step": 5434 + }, + { + "epoch": 2.28, + "learning_rate": 1.679189705525769e-05, + "loss": 0.2594, + "step": 5435 + }, + { + "epoch": 2.28, + "learning_rate": 1.6790652590657125e-05, + "loss": 0.2543, + "step": 5436 + }, + { + "epoch": 2.28, + "learning_rate": 1.678940793086528e-05, + "loss": 0.2591, + "step": 5437 + }, + { + "epoch": 2.28, + "learning_rate": 1.678816307591793e-05, + "loss": 0.2541, + "step": 5438 + }, + { + "epoch": 2.28, + "learning_rate": 1.6786918025850852e-05, + "loss": 0.2863, + "step": 5439 + }, + { + "epoch": 2.28, + "learning_rate": 1.6785672780699836e-05, + "loss": 0.2611, + "step": 5440 + }, + { + "epoch": 2.28, + "learning_rate": 1.6784427340500684e-05, + "loss": 0.286, + "step": 5441 + }, + { + "epoch": 2.28, + "learning_rate": 1.6783181705289184e-05, + "loss": 0.244, + "step": 5442 + }, + { + "epoch": 2.28, + "learning_rate": 1.6781935875101148e-05, + "loss": 0.2727, + "step": 5443 + }, + { + "epoch": 2.28, + "learning_rate": 1.678068984997238e-05, + "loss": 0.2597, + "step": 5444 + }, + { + "epoch": 2.28, + "learning_rate": 1.67794436299387e-05, + "loss": 0.2972, + "step": 5445 + }, + { + "epoch": 2.28, + "learning_rate": 1.677819721503593e-05, + "loss": 0.2636, + "step": 5446 + }, + { + "epoch": 2.28, + "learning_rate": 1.6776950605299897e-05, + "loss": 0.2327, + "step": 5447 + }, + { + "epoch": 2.28, + "learning_rate": 1.677570380076643e-05, + "loss": 0.275, + "step": 5448 + }, + { + "epoch": 2.28, + "learning_rate": 1.677445680147137e-05, + "loss": 0.2507, + "step": 5449 + }, + { + "epoch": 2.28, + "learning_rate": 1.6773209607450564e-05, + "loss": 0.2583, + "step": 5450 + }, + { + "epoch": 2.28, + "learning_rate": 1.6771962218739856e-05, + "loss": 0.2828, + "step": 5451 + }, + { + "epoch": 2.28, + "learning_rate": 1.6770714635375104e-05, + "loss": 0.2481, + "step": 5452 + }, + { + "epoch": 2.28, + "learning_rate": 1.6769466857392167e-05, + "loss": 0.2325, + "step": 5453 + }, + { + "epoch": 2.28, + "learning_rate": 1.6768218884826916e-05, + "loss": 0.2542, + "step": 5454 + }, + { + "epoch": 2.28, + "learning_rate": 1.6766970717715216e-05, + "loss": 0.2508, + "step": 5455 + }, + { + "epoch": 2.28, + "learning_rate": 1.6765722356092947e-05, + "loss": 0.256, + "step": 5456 + }, + { + "epoch": 2.28, + "learning_rate": 1.6764473799995993e-05, + "loss": 0.261, + "step": 5457 + }, + { + "epoch": 2.29, + "learning_rate": 1.676322504946024e-05, + "loss": 0.2655, + "step": 5458 + }, + { + "epoch": 2.29, + "learning_rate": 1.676197610452159e-05, + "loss": 0.2763, + "step": 5459 + }, + { + "epoch": 2.29, + "learning_rate": 1.6760726965215936e-05, + "loss": 0.2443, + "step": 5460 + }, + { + "epoch": 2.29, + "learning_rate": 1.6759477631579182e-05, + "loss": 0.2783, + "step": 5461 + }, + { + "epoch": 2.29, + "learning_rate": 1.6758228103647246e-05, + "loss": 0.2794, + "step": 5462 + }, + { + "epoch": 2.29, + "learning_rate": 1.6756978381456035e-05, + "loss": 0.2522, + "step": 5463 + }, + { + "epoch": 2.29, + "learning_rate": 1.675572846504148e-05, + "loss": 0.2734, + "step": 5464 + }, + { + "epoch": 2.29, + "learning_rate": 1.6754478354439504e-05, + "loss": 0.2389, + "step": 5465 + }, + { + "epoch": 2.29, + "learning_rate": 1.6753228049686042e-05, + "loss": 0.2517, + "step": 5466 + }, + { + "epoch": 2.29, + "learning_rate": 1.675197755081703e-05, + "loss": 0.2726, + "step": 5467 + }, + { + "epoch": 2.29, + "learning_rate": 1.675072685786842e-05, + "loss": 0.2493, + "step": 5468 + }, + { + "epoch": 2.29, + "learning_rate": 1.674947597087615e-05, + "loss": 0.2413, + "step": 5469 + }, + { + "epoch": 2.29, + "learning_rate": 1.6748224889876188e-05, + "loss": 0.2812, + "step": 5470 + }, + { + "epoch": 2.29, + "learning_rate": 1.674697361490449e-05, + "loss": 0.2636, + "step": 5471 + }, + { + "epoch": 2.29, + "learning_rate": 1.674572214599702e-05, + "loss": 0.2142, + "step": 5472 + }, + { + "epoch": 2.29, + "learning_rate": 1.6744470483189755e-05, + "loss": 0.2527, + "step": 5473 + }, + { + "epoch": 2.29, + "learning_rate": 1.674321862651867e-05, + "loss": 0.2542, + "step": 5474 + }, + { + "epoch": 2.29, + "learning_rate": 1.674196657601975e-05, + "loss": 0.2612, + "step": 5475 + }, + { + "epoch": 2.29, + "learning_rate": 1.6740714331728984e-05, + "loss": 0.2351, + "step": 5476 + }, + { + "epoch": 2.29, + "learning_rate": 1.673946189368236e-05, + "loss": 0.2701, + "step": 5477 + }, + { + "epoch": 2.29, + "learning_rate": 1.673820926191589e-05, + "loss": 0.2584, + "step": 5478 + }, + { + "epoch": 2.29, + "learning_rate": 1.6736956436465573e-05, + "loss": 0.2623, + "step": 5479 + }, + { + "epoch": 2.29, + "learning_rate": 1.6735703417367422e-05, + "loss": 0.2528, + "step": 5480 + }, + { + "epoch": 2.29, + "learning_rate": 1.673445020465745e-05, + "loss": 0.2376, + "step": 5481 + }, + { + "epoch": 2.3, + "learning_rate": 1.673319679837169e-05, + "loss": 0.2837, + "step": 5482 + }, + { + "epoch": 2.3, + "learning_rate": 1.673194319854616e-05, + "loss": 0.261, + "step": 5483 + }, + { + "epoch": 2.3, + "learning_rate": 1.6730689405216895e-05, + "loss": 0.2461, + "step": 5484 + }, + { + "epoch": 2.3, + "learning_rate": 1.6729435418419936e-05, + "loss": 0.2953, + "step": 5485 + }, + { + "epoch": 2.3, + "learning_rate": 1.6728181238191328e-05, + "loss": 0.2853, + "step": 5486 + }, + { + "epoch": 2.3, + "learning_rate": 1.672692686456712e-05, + "loss": 0.2504, + "step": 5487 + }, + { + "epoch": 2.3, + "learning_rate": 1.672567229758337e-05, + "loss": 0.2589, + "step": 5488 + }, + { + "epoch": 2.3, + "learning_rate": 1.6724417537276138e-05, + "loss": 0.2655, + "step": 5489 + }, + { + "epoch": 2.3, + "learning_rate": 1.6723162583681492e-05, + "loss": 0.249, + "step": 5490 + }, + { + "epoch": 2.3, + "learning_rate": 1.67219074368355e-05, + "loss": 0.2526, + "step": 5491 + }, + { + "epoch": 2.3, + "learning_rate": 1.6720652096774245e-05, + "loss": 0.2313, + "step": 5492 + }, + { + "epoch": 2.3, + "learning_rate": 1.6719396563533812e-05, + "loss": 0.2452, + "step": 5493 + }, + { + "epoch": 2.3, + "learning_rate": 1.6718140837150286e-05, + "loss": 0.2643, + "step": 5494 + }, + { + "epoch": 2.3, + "learning_rate": 1.6716884917659763e-05, + "loss": 0.2852, + "step": 5495 + }, + { + "epoch": 2.3, + "learning_rate": 1.671562880509834e-05, + "loss": 0.2577, + "step": 5496 + }, + { + "epoch": 2.3, + "learning_rate": 1.6714372499502133e-05, + "loss": 0.2387, + "step": 5497 + }, + { + "epoch": 2.3, + "learning_rate": 1.671311600090724e-05, + "loss": 0.2763, + "step": 5498 + }, + { + "epoch": 2.3, + "learning_rate": 1.6711859309349788e-05, + "loss": 0.2555, + "step": 5499 + }, + { + "epoch": 2.3, + "learning_rate": 1.67106024248659e-05, + "loss": 0.2557, + "step": 5500 + }, + { + "epoch": 2.3, + "learning_rate": 1.6709345347491695e-05, + "loss": 0.2474, + "step": 5501 + }, + { + "epoch": 2.3, + "learning_rate": 1.670808807726331e-05, + "loss": 0.261, + "step": 5502 + }, + { + "epoch": 2.3, + "learning_rate": 1.670683061421689e-05, + "loss": 0.247, + "step": 5503 + }, + { + "epoch": 2.3, + "learning_rate": 1.6705572958388576e-05, + "loss": 0.26, + "step": 5504 + }, + { + "epoch": 2.3, + "learning_rate": 1.6704315109814514e-05, + "loss": 0.2832, + "step": 5505 + }, + { + "epoch": 2.31, + "learning_rate": 1.670305706853087e-05, + "loss": 0.2386, + "step": 5506 + }, + { + "epoch": 2.31, + "learning_rate": 1.670179883457379e-05, + "loss": 0.286, + "step": 5507 + }, + { + "epoch": 2.31, + "learning_rate": 1.6700540407979457e-05, + "loss": 0.2268, + "step": 5508 + }, + { + "epoch": 2.31, + "learning_rate": 1.669928178878403e-05, + "loss": 0.2397, + "step": 5509 + }, + { + "epoch": 2.31, + "learning_rate": 1.6698022977023697e-05, + "loss": 0.25, + "step": 5510 + }, + { + "epoch": 2.31, + "learning_rate": 1.6696763972734637e-05, + "loss": 0.2595, + "step": 5511 + }, + { + "epoch": 2.31, + "learning_rate": 1.6695504775953036e-05, + "loss": 0.244, + "step": 5512 + }, + { + "epoch": 2.31, + "learning_rate": 1.6694245386715094e-05, + "loss": 0.2475, + "step": 5513 + }, + { + "epoch": 2.31, + "learning_rate": 1.669298580505701e-05, + "loss": 0.253, + "step": 5514 + }, + { + "epoch": 2.31, + "learning_rate": 1.6691726031014985e-05, + "loss": 0.2752, + "step": 5515 + }, + { + "epoch": 2.31, + "learning_rate": 1.669046606462524e-05, + "loss": 0.2726, + "step": 5516 + }, + { + "epoch": 2.31, + "learning_rate": 1.6689205905923977e-05, + "loss": 0.238, + "step": 5517 + }, + { + "epoch": 2.31, + "learning_rate": 1.6687945554947428e-05, + "loss": 0.2627, + "step": 5518 + }, + { + "epoch": 2.31, + "learning_rate": 1.668668501173182e-05, + "loss": 0.2485, + "step": 5519 + }, + { + "epoch": 2.31, + "learning_rate": 1.6685424276313386e-05, + "loss": 0.2596, + "step": 5520 + }, + { + "epoch": 2.31, + "learning_rate": 1.6684163348728364e-05, + "loss": 0.2893, + "step": 5521 + }, + { + "epoch": 2.31, + "learning_rate": 1.6682902229012997e-05, + "loss": 0.2622, + "step": 5522 + }, + { + "epoch": 2.31, + "learning_rate": 1.6681640917203534e-05, + "loss": 0.2604, + "step": 5523 + }, + { + "epoch": 2.31, + "learning_rate": 1.668037941333623e-05, + "loss": 0.2601, + "step": 5524 + }, + { + "epoch": 2.31, + "learning_rate": 1.6679117717447356e-05, + "loss": 0.2353, + "step": 5525 + }, + { + "epoch": 2.31, + "learning_rate": 1.6677855829573164e-05, + "loss": 0.242, + "step": 5526 + }, + { + "epoch": 2.31, + "learning_rate": 1.6676593749749936e-05, + "loss": 0.2716, + "step": 5527 + }, + { + "epoch": 2.31, + "learning_rate": 1.667533147801394e-05, + "loss": 0.258, + "step": 5528 + }, + { + "epoch": 2.31, + "learning_rate": 1.667406901440147e-05, + "loss": 0.273, + "step": 5529 + }, + { + "epoch": 2.32, + "learning_rate": 1.6672806358948803e-05, + "loss": 0.2658, + "step": 5530 + }, + { + "epoch": 2.32, + "learning_rate": 1.667154351169224e-05, + "loss": 0.2419, + "step": 5531 + }, + { + "epoch": 2.32, + "learning_rate": 1.6670280472668082e-05, + "loss": 0.2512, + "step": 5532 + }, + { + "epoch": 2.32, + "learning_rate": 1.666901724191263e-05, + "loss": 0.2694, + "step": 5533 + }, + { + "epoch": 2.32, + "learning_rate": 1.6667753819462196e-05, + "loss": 0.2649, + "step": 5534 + }, + { + "epoch": 2.32, + "learning_rate": 1.6666490205353092e-05, + "loss": 0.2606, + "step": 5535 + }, + { + "epoch": 2.32, + "learning_rate": 1.6665226399621642e-05, + "loss": 0.2565, + "step": 5536 + }, + { + "epoch": 2.32, + "learning_rate": 1.6663962402304178e-05, + "loss": 0.239, + "step": 5537 + }, + { + "epoch": 2.32, + "learning_rate": 1.666269821343703e-05, + "loss": 0.2428, + "step": 5538 + }, + { + "epoch": 2.32, + "learning_rate": 1.6661433833056527e-05, + "loss": 0.2325, + "step": 5539 + }, + { + "epoch": 2.32, + "learning_rate": 1.6660169261199027e-05, + "loss": 0.2501, + "step": 5540 + }, + { + "epoch": 2.32, + "learning_rate": 1.665890449790087e-05, + "loss": 0.2478, + "step": 5541 + }, + { + "epoch": 2.32, + "learning_rate": 1.665763954319841e-05, + "loss": 0.23, + "step": 5542 + }, + { + "epoch": 2.32, + "learning_rate": 1.6656374397128007e-05, + "loss": 0.2581, + "step": 5543 + }, + { + "epoch": 2.32, + "learning_rate": 1.665510905972603e-05, + "loss": 0.2491, + "step": 5544 + }, + { + "epoch": 2.32, + "learning_rate": 1.6653843531028852e-05, + "loss": 0.255, + "step": 5545 + }, + { + "epoch": 2.32, + "learning_rate": 1.6652577811072842e-05, + "loss": 0.2237, + "step": 5546 + }, + { + "epoch": 2.32, + "learning_rate": 1.665131189989439e-05, + "loss": 0.2275, + "step": 5547 + }, + { + "epoch": 2.32, + "learning_rate": 1.665004579752988e-05, + "loss": 0.2532, + "step": 5548 + }, + { + "epoch": 2.32, + "learning_rate": 1.66487795040157e-05, + "loss": 0.2732, + "step": 5549 + }, + { + "epoch": 2.32, + "learning_rate": 1.664751301938826e-05, + "loss": 0.2887, + "step": 5550 + }, + { + "epoch": 2.32, + "learning_rate": 1.6646246343683953e-05, + "loss": 0.2389, + "step": 5551 + }, + { + "epoch": 2.32, + "learning_rate": 1.6644979476939194e-05, + "loss": 0.2584, + "step": 5552 + }, + { + "epoch": 2.32, + "learning_rate": 1.6643712419190396e-05, + "loss": 0.2517, + "step": 5553 + }, + { + "epoch": 2.33, + "learning_rate": 1.664244517047398e-05, + "loss": 0.2563, + "step": 5554 + }, + { + "epoch": 2.33, + "learning_rate": 1.664117773082637e-05, + "loss": 0.2428, + "step": 5555 + }, + { + "epoch": 2.33, + "learning_rate": 1.6639910100284e-05, + "loss": 0.2671, + "step": 5556 + }, + { + "epoch": 2.33, + "learning_rate": 1.663864227888331e-05, + "loss": 0.2196, + "step": 5557 + }, + { + "epoch": 2.33, + "learning_rate": 1.6637374266660734e-05, + "loss": 0.2565, + "step": 5558 + }, + { + "epoch": 2.33, + "learning_rate": 1.6636106063652725e-05, + "loss": 0.2578, + "step": 5559 + }, + { + "epoch": 2.33, + "learning_rate": 1.663483766989574e-05, + "loss": 0.2725, + "step": 5560 + }, + { + "epoch": 2.33, + "learning_rate": 1.663356908542623e-05, + "loss": 0.2808, + "step": 5561 + }, + { + "epoch": 2.33, + "learning_rate": 1.6632300310280664e-05, + "loss": 0.2417, + "step": 5562 + }, + { + "epoch": 2.33, + "learning_rate": 1.663103134449551e-05, + "loss": 0.2653, + "step": 5563 + }, + { + "epoch": 2.33, + "learning_rate": 1.6629762188107248e-05, + "loss": 0.2614, + "step": 5564 + }, + { + "epoch": 2.33, + "learning_rate": 1.6628492841152357e-05, + "loss": 0.2483, + "step": 5565 + }, + { + "epoch": 2.33, + "learning_rate": 1.6627223303667314e-05, + "loss": 0.2439, + "step": 5566 + }, + { + "epoch": 2.33, + "learning_rate": 1.6625953575688624e-05, + "loss": 0.2314, + "step": 5567 + }, + { + "epoch": 2.33, + "learning_rate": 1.6624683657252773e-05, + "loss": 0.216, + "step": 5568 + }, + { + "epoch": 2.33, + "learning_rate": 1.6623413548396272e-05, + "loss": 0.253, + "step": 5569 + }, + { + "epoch": 2.33, + "learning_rate": 1.6622143249155625e-05, + "loss": 0.2446, + "step": 5570 + }, + { + "epoch": 2.33, + "learning_rate": 1.6620872759567345e-05, + "loss": 0.256, + "step": 5571 + }, + { + "epoch": 2.33, + "learning_rate": 1.6619602079667956e-05, + "loss": 0.2411, + "step": 5572 + }, + { + "epoch": 2.33, + "learning_rate": 1.661833120949398e-05, + "loss": 0.2755, + "step": 5573 + }, + { + "epoch": 2.33, + "learning_rate": 1.6617060149081944e-05, + "loss": 0.2411, + "step": 5574 + }, + { + "epoch": 2.33, + "learning_rate": 1.6615788898468384e-05, + "loss": 0.2439, + "step": 5575 + }, + { + "epoch": 2.33, + "learning_rate": 1.661451745768985e-05, + "loss": 0.2606, + "step": 5576 + }, + { + "epoch": 2.33, + "learning_rate": 1.6613245826782875e-05, + "loss": 0.288, + "step": 5577 + }, + { + "epoch": 2.34, + "learning_rate": 1.6611974005784017e-05, + "loss": 0.2454, + "step": 5578 + }, + { + "epoch": 2.34, + "learning_rate": 1.6610701994729838e-05, + "loss": 0.2253, + "step": 5579 + }, + { + "epoch": 2.34, + "learning_rate": 1.660942979365689e-05, + "loss": 0.2392, + "step": 5580 + }, + { + "epoch": 2.34, + "learning_rate": 1.6608157402601753e-05, + "loss": 0.2397, + "step": 5581 + }, + { + "epoch": 2.34, + "learning_rate": 1.6606884821600995e-05, + "loss": 0.2303, + "step": 5582 + }, + { + "epoch": 2.34, + "learning_rate": 1.6605612050691197e-05, + "loss": 0.2431, + "step": 5583 + }, + { + "epoch": 2.34, + "learning_rate": 1.6604339089908935e-05, + "loss": 0.2579, + "step": 5584 + }, + { + "epoch": 2.34, + "learning_rate": 1.6603065939290815e-05, + "loss": 0.2297, + "step": 5585 + }, + { + "epoch": 2.34, + "learning_rate": 1.660179259887342e-05, + "loss": 0.2695, + "step": 5586 + }, + { + "epoch": 2.34, + "learning_rate": 1.6600519068693356e-05, + "loss": 0.2433, + "step": 5587 + }, + { + "epoch": 2.34, + "learning_rate": 1.659924534878723e-05, + "loss": 0.2382, + "step": 5588 + }, + { + "epoch": 2.34, + "learning_rate": 1.659797143919165e-05, + "loss": 0.2639, + "step": 5589 + }, + { + "epoch": 2.34, + "learning_rate": 1.6596697339943234e-05, + "loss": 0.2494, + "step": 5590 + }, + { + "epoch": 2.34, + "learning_rate": 1.659542305107861e-05, + "loss": 0.283, + "step": 5591 + }, + { + "epoch": 2.34, + "learning_rate": 1.6594148572634404e-05, + "loss": 0.2544, + "step": 5592 + }, + { + "epoch": 2.34, + "learning_rate": 1.6592873904647245e-05, + "loss": 0.2498, + "step": 5593 + }, + { + "epoch": 2.34, + "learning_rate": 1.6591599047153778e-05, + "loss": 0.2804, + "step": 5594 + }, + { + "epoch": 2.34, + "learning_rate": 1.6590324000190645e-05, + "loss": 0.2287, + "step": 5595 + }, + { + "epoch": 2.34, + "learning_rate": 1.65890487637945e-05, + "loss": 0.2674, + "step": 5596 + }, + { + "epoch": 2.34, + "learning_rate": 1.6587773338001993e-05, + "loss": 0.2598, + "step": 5597 + }, + { + "epoch": 2.34, + "learning_rate": 1.658649772284979e-05, + "loss": 0.2656, + "step": 5598 + }, + { + "epoch": 2.34, + "learning_rate": 1.658522191837455e-05, + "loss": 0.2451, + "step": 5599 + }, + { + "epoch": 2.34, + "learning_rate": 1.658394592461295e-05, + "loss": 0.2575, + "step": 5600 + }, + { + "epoch": 2.34, + "learning_rate": 1.658266974160167e-05, + "loss": 0.259, + "step": 5601 + }, + { + "epoch": 2.35, + "learning_rate": 1.6581393369377385e-05, + "loss": 0.2611, + "step": 5602 + }, + { + "epoch": 2.35, + "learning_rate": 1.6580116807976795e-05, + "loss": 0.2447, + "step": 5603 + }, + { + "epoch": 2.35, + "learning_rate": 1.657884005743658e-05, + "loss": 0.2716, + "step": 5604 + }, + { + "epoch": 2.35, + "learning_rate": 1.6577563117793447e-05, + "loss": 0.2492, + "step": 5605 + }, + { + "epoch": 2.35, + "learning_rate": 1.65762859890841e-05, + "loss": 0.2713, + "step": 5606 + }, + { + "epoch": 2.35, + "learning_rate": 1.6575008671345248e-05, + "loss": 0.2381, + "step": 5607 + }, + { + "epoch": 2.35, + "learning_rate": 1.6573731164613604e-05, + "loss": 0.2521, + "step": 5608 + }, + { + "epoch": 2.35, + "learning_rate": 1.6572453468925895e-05, + "loss": 0.2688, + "step": 5609 + }, + { + "epoch": 2.35, + "learning_rate": 1.6571175584318845e-05, + "loss": 0.2438, + "step": 5610 + }, + { + "epoch": 2.35, + "learning_rate": 1.656989751082918e-05, + "loss": 0.2655, + "step": 5611 + }, + { + "epoch": 2.35, + "learning_rate": 1.656861924849364e-05, + "loss": 0.2333, + "step": 5612 + }, + { + "epoch": 2.35, + "learning_rate": 1.656734079734897e-05, + "loss": 0.277, + "step": 5613 + }, + { + "epoch": 2.35, + "learning_rate": 1.6566062157431915e-05, + "loss": 0.2572, + "step": 5614 + }, + { + "epoch": 2.35, + "learning_rate": 1.656478332877923e-05, + "loss": 0.2439, + "step": 5615 + }, + { + "epoch": 2.35, + "learning_rate": 1.6563504311427678e-05, + "loss": 0.2518, + "step": 5616 + }, + { + "epoch": 2.35, + "learning_rate": 1.6562225105414016e-05, + "loss": 0.2696, + "step": 5617 + }, + { + "epoch": 2.35, + "learning_rate": 1.656094571077502e-05, + "loss": 0.2628, + "step": 5618 + }, + { + "epoch": 2.35, + "learning_rate": 1.6559666127547455e-05, + "loss": 0.2394, + "step": 5619 + }, + { + "epoch": 2.35, + "learning_rate": 1.655838635576811e-05, + "loss": 0.2871, + "step": 5620 + }, + { + "epoch": 2.35, + "learning_rate": 1.6557106395473775e-05, + "loss": 0.2729, + "step": 5621 + }, + { + "epoch": 2.35, + "learning_rate": 1.655582624670123e-05, + "loss": 0.2643, + "step": 5622 + }, + { + "epoch": 2.35, + "learning_rate": 1.6554545909487277e-05, + "loss": 0.2522, + "step": 5623 + }, + { + "epoch": 2.35, + "learning_rate": 1.655326538386872e-05, + "loss": 0.2616, + "step": 5624 + }, + { + "epoch": 2.36, + "learning_rate": 1.6551984669882362e-05, + "loss": 0.2816, + "step": 5625 + }, + { + "epoch": 2.36, + "learning_rate": 1.655070376756502e-05, + "loss": 0.2558, + "step": 5626 + }, + { + "epoch": 2.36, + "learning_rate": 1.6549422676953513e-05, + "loss": 0.2902, + "step": 5627 + }, + { + "epoch": 2.36, + "learning_rate": 1.6548141398084662e-05, + "loss": 0.2461, + "step": 5628 + }, + { + "epoch": 2.36, + "learning_rate": 1.6546859930995295e-05, + "loss": 0.2601, + "step": 5629 + }, + { + "epoch": 2.36, + "learning_rate": 1.654557827572225e-05, + "loss": 0.2647, + "step": 5630 + }, + { + "epoch": 2.36, + "learning_rate": 1.6544296432302373e-05, + "loss": 0.2559, + "step": 5631 + }, + { + "epoch": 2.36, + "learning_rate": 1.6543014400772496e-05, + "loss": 0.2638, + "step": 5632 + }, + { + "epoch": 2.36, + "learning_rate": 1.6541732181169474e-05, + "loss": 0.2409, + "step": 5633 + }, + { + "epoch": 2.36, + "learning_rate": 1.6540449773530168e-05, + "loss": 0.2524, + "step": 5634 + }, + { + "epoch": 2.36, + "learning_rate": 1.6539167177891436e-05, + "loss": 0.2701, + "step": 5635 + }, + { + "epoch": 2.36, + "learning_rate": 1.653788439429015e-05, + "loss": 0.2518, + "step": 5636 + }, + { + "epoch": 2.36, + "learning_rate": 1.6536601422763173e-05, + "loss": 0.2835, + "step": 5637 + }, + { + "epoch": 2.36, + "learning_rate": 1.6535318263347395e-05, + "loss": 0.2598, + "step": 5638 + }, + { + "epoch": 2.36, + "learning_rate": 1.6534034916079686e-05, + "loss": 0.2588, + "step": 5639 + }, + { + "epoch": 2.36, + "learning_rate": 1.653275138099695e-05, + "loss": 0.222, + "step": 5640 + }, + { + "epoch": 2.36, + "learning_rate": 1.6531467658136064e-05, + "loss": 0.2774, + "step": 5641 + }, + { + "epoch": 2.36, + "learning_rate": 1.6530183747533942e-05, + "loss": 0.2674, + "step": 5642 + }, + { + "epoch": 2.36, + "learning_rate": 1.6528899649227482e-05, + "loss": 0.26, + "step": 5643 + }, + { + "epoch": 2.36, + "learning_rate": 1.652761536325359e-05, + "loss": 0.2622, + "step": 5644 + }, + { + "epoch": 2.36, + "learning_rate": 1.6526330889649194e-05, + "loss": 0.2339, + "step": 5645 + }, + { + "epoch": 2.36, + "learning_rate": 1.6525046228451204e-05, + "loss": 0.279, + "step": 5646 + }, + { + "epoch": 2.36, + "learning_rate": 1.652376137969655e-05, + "loss": 0.2597, + "step": 5647 + }, + { + "epoch": 2.36, + "learning_rate": 1.6522476343422164e-05, + "loss": 0.2777, + "step": 5648 + }, + { + "epoch": 2.37, + "learning_rate": 1.652119111966499e-05, + "loss": 0.2719, + "step": 5649 + }, + { + "epoch": 2.37, + "learning_rate": 1.6519905708461953e-05, + "loss": 0.2532, + "step": 5650 + }, + { + "epoch": 2.37, + "learning_rate": 1.651862010985002e-05, + "loss": 0.2652, + "step": 5651 + }, + { + "epoch": 2.37, + "learning_rate": 1.6517334323866136e-05, + "loss": 0.2313, + "step": 5652 + }, + { + "epoch": 2.37, + "learning_rate": 1.651604835054726e-05, + "loss": 0.253, + "step": 5653 + }, + { + "epoch": 2.37, + "learning_rate": 1.6514762189930353e-05, + "loss": 0.2641, + "step": 5654 + }, + { + "epoch": 2.37, + "learning_rate": 1.651347584205239e-05, + "loss": 0.2449, + "step": 5655 + }, + { + "epoch": 2.37, + "learning_rate": 1.6512189306950345e-05, + "loss": 0.2649, + "step": 5656 + }, + { + "epoch": 2.37, + "learning_rate": 1.6510902584661195e-05, + "loss": 0.2837, + "step": 5657 + }, + { + "epoch": 2.37, + "learning_rate": 1.650961567522193e-05, + "loss": 0.2499, + "step": 5658 + }, + { + "epoch": 2.37, + "learning_rate": 1.6508328578669537e-05, + "loss": 0.2475, + "step": 5659 + }, + { + "epoch": 2.37, + "learning_rate": 1.6507041295041015e-05, + "loss": 0.246, + "step": 5660 + }, + { + "epoch": 2.37, + "learning_rate": 1.6505753824373364e-05, + "loss": 0.2631, + "step": 5661 + }, + { + "epoch": 2.37, + "learning_rate": 1.6504466166703593e-05, + "loss": 0.2594, + "step": 5662 + }, + { + "epoch": 2.37, + "learning_rate": 1.6503178322068718e-05, + "loss": 0.2271, + "step": 5663 + }, + { + "epoch": 2.37, + "learning_rate": 1.650189029050575e-05, + "loss": 0.2303, + "step": 5664 + }, + { + "epoch": 2.37, + "learning_rate": 1.6500602072051714e-05, + "loss": 0.2522, + "step": 5665 + }, + { + "epoch": 2.37, + "learning_rate": 1.6499313666743643e-05, + "loss": 0.2741, + "step": 5666 + }, + { + "epoch": 2.37, + "learning_rate": 1.6498025074618563e-05, + "loss": 0.2291, + "step": 5667 + }, + { + "epoch": 2.37, + "learning_rate": 1.6496736295713526e-05, + "loss": 0.2841, + "step": 5668 + }, + { + "epoch": 2.37, + "learning_rate": 1.649544733006556e-05, + "loss": 0.246, + "step": 5669 + }, + { + "epoch": 2.37, + "learning_rate": 1.649415817771173e-05, + "loss": 0.2773, + "step": 5670 + }, + { + "epoch": 2.37, + "learning_rate": 1.649286883868909e-05, + "loss": 0.2308, + "step": 5671 + }, + { + "epoch": 2.37, + "learning_rate": 1.649157931303469e-05, + "loss": 0.244, + "step": 5672 + }, + { + "epoch": 2.38, + "learning_rate": 1.6490289600785603e-05, + "loss": 0.2649, + "step": 5673 + }, + { + "epoch": 2.38, + "learning_rate": 1.6488999701978905e-05, + "loss": 0.2718, + "step": 5674 + }, + { + "epoch": 2.38, + "learning_rate": 1.6487709616651667e-05, + "loss": 0.2615, + "step": 5675 + }, + { + "epoch": 2.38, + "learning_rate": 1.6486419344840974e-05, + "loss": 0.2741, + "step": 5676 + }, + { + "epoch": 2.38, + "learning_rate": 1.6485128886583913e-05, + "loss": 0.2812, + "step": 5677 + }, + { + "epoch": 2.38, + "learning_rate": 1.6483838241917577e-05, + "loss": 0.2596, + "step": 5678 + }, + { + "epoch": 2.38, + "learning_rate": 1.6482547410879063e-05, + "loss": 0.2359, + "step": 5679 + }, + { + "epoch": 2.38, + "learning_rate": 1.6481256393505477e-05, + "loss": 0.2907, + "step": 5680 + }, + { + "epoch": 2.38, + "learning_rate": 1.6479965189833923e-05, + "loss": 0.2376, + "step": 5681 + }, + { + "epoch": 2.38, + "learning_rate": 1.6478673799901525e-05, + "loss": 0.2418, + "step": 5682 + }, + { + "epoch": 2.38, + "learning_rate": 1.6477382223745396e-05, + "loss": 0.2486, + "step": 5683 + }, + { + "epoch": 2.38, + "learning_rate": 1.6476090461402664e-05, + "loss": 0.2549, + "step": 5684 + }, + { + "epoch": 2.38, + "learning_rate": 1.647479851291046e-05, + "loss": 0.2417, + "step": 5685 + }, + { + "epoch": 2.38, + "learning_rate": 1.6473506378305914e-05, + "loss": 0.2823, + "step": 5686 + }, + { + "epoch": 2.38, + "learning_rate": 1.6472214057626175e-05, + "loss": 0.2542, + "step": 5687 + }, + { + "epoch": 2.38, + "learning_rate": 1.6470921550908385e-05, + "loss": 0.2549, + "step": 5688 + }, + { + "epoch": 2.38, + "learning_rate": 1.6469628858189698e-05, + "loss": 0.2689, + "step": 5689 + }, + { + "epoch": 2.38, + "learning_rate": 1.646833597950727e-05, + "loss": 0.2706, + "step": 5690 + }, + { + "epoch": 2.38, + "learning_rate": 1.6467042914898262e-05, + "loss": 0.2514, + "step": 5691 + }, + { + "epoch": 2.38, + "learning_rate": 1.6465749664399848e-05, + "loss": 0.2491, + "step": 5692 + }, + { + "epoch": 2.38, + "learning_rate": 1.6464456228049195e-05, + "loss": 0.2418, + "step": 5693 + }, + { + "epoch": 2.38, + "learning_rate": 1.6463162605883485e-05, + "loss": 0.2389, + "step": 5694 + }, + { + "epoch": 2.38, + "learning_rate": 1.64618687979399e-05, + "loss": 0.2294, + "step": 5695 + }, + { + "epoch": 2.38, + "learning_rate": 1.6460574804255634e-05, + "loss": 0.2445, + "step": 5696 + }, + { + "epoch": 2.39, + "learning_rate": 1.6459280624867876e-05, + "loss": 0.2614, + "step": 5697 + }, + { + "epoch": 2.39, + "learning_rate": 1.6457986259813827e-05, + "loss": 0.2873, + "step": 5698 + }, + { + "epoch": 2.39, + "learning_rate": 1.6456691709130694e-05, + "loss": 0.251, + "step": 5699 + }, + { + "epoch": 2.39, + "learning_rate": 1.6455396972855688e-05, + "loss": 0.2761, + "step": 5700 + }, + { + "epoch": 2.39, + "learning_rate": 1.6454102051026025e-05, + "loss": 0.2521, + "step": 5701 + }, + { + "epoch": 2.39, + "learning_rate": 1.6452806943678928e-05, + "loss": 0.2566, + "step": 5702 + }, + { + "epoch": 2.39, + "learning_rate": 1.645151165085162e-05, + "loss": 0.2531, + "step": 5703 + }, + { + "epoch": 2.39, + "learning_rate": 1.645021617258133e-05, + "loss": 0.2512, + "step": 5704 + }, + { + "epoch": 2.39, + "learning_rate": 1.6448920508905305e-05, + "loss": 0.239, + "step": 5705 + }, + { + "epoch": 2.39, + "learning_rate": 1.644762465986078e-05, + "loss": 0.2325, + "step": 5706 + }, + { + "epoch": 2.39, + "learning_rate": 1.6446328625485007e-05, + "loss": 0.2521, + "step": 5707 + }, + { + "epoch": 2.39, + "learning_rate": 1.644503240581524e-05, + "loss": 0.2613, + "step": 5708 + }, + { + "epoch": 2.39, + "learning_rate": 1.6443736000888732e-05, + "loss": 0.2535, + "step": 5709 + }, + { + "epoch": 2.39, + "learning_rate": 1.6442439410742753e-05, + "loss": 0.2512, + "step": 5710 + }, + { + "epoch": 2.39, + "learning_rate": 1.6441142635414568e-05, + "loss": 0.2478, + "step": 5711 + }, + { + "epoch": 2.39, + "learning_rate": 1.6439845674941458e-05, + "loss": 0.2688, + "step": 5712 + }, + { + "epoch": 2.39, + "learning_rate": 1.6438548529360698e-05, + "loss": 0.2423, + "step": 5713 + }, + { + "epoch": 2.39, + "learning_rate": 1.6437251198709567e-05, + "loss": 0.2904, + "step": 5714 + }, + { + "epoch": 2.39, + "learning_rate": 1.643595368302537e-05, + "loss": 0.2544, + "step": 5715 + }, + { + "epoch": 2.39, + "learning_rate": 1.643465598234539e-05, + "loss": 0.2706, + "step": 5716 + }, + { + "epoch": 2.39, + "learning_rate": 1.6433358096706937e-05, + "loss": 0.2799, + "step": 5717 + }, + { + "epoch": 2.39, + "learning_rate": 1.6432060026147313e-05, + "loss": 0.2739, + "step": 5718 + }, + { + "epoch": 2.39, + "learning_rate": 1.6430761770703836e-05, + "loss": 0.2693, + "step": 5719 + }, + { + "epoch": 2.39, + "learning_rate": 1.6429463330413812e-05, + "loss": 0.2574, + "step": 5720 + }, + { + "epoch": 2.4, + "learning_rate": 1.6428164705314572e-05, + "loss": 0.2597, + "step": 5721 + }, + { + "epoch": 2.4, + "learning_rate": 1.6426865895443446e-05, + "loss": 0.2549, + "step": 5722 + }, + { + "epoch": 2.4, + "learning_rate": 1.642556690083776e-05, + "loss": 0.2729, + "step": 5723 + }, + { + "epoch": 2.4, + "learning_rate": 1.6424267721534856e-05, + "loss": 0.2518, + "step": 5724 + }, + { + "epoch": 2.4, + "learning_rate": 1.6422968357572076e-05, + "loss": 0.2618, + "step": 5725 + }, + { + "epoch": 2.4, + "learning_rate": 1.642166880898677e-05, + "loss": 0.2654, + "step": 5726 + }, + { + "epoch": 2.4, + "learning_rate": 1.642036907581629e-05, + "loss": 0.2463, + "step": 5727 + }, + { + "epoch": 2.4, + "learning_rate": 1.6419069158098004e-05, + "loss": 0.2646, + "step": 5728 + }, + { + "epoch": 2.4, + "learning_rate": 1.6417769055869266e-05, + "loss": 0.2662, + "step": 5729 + }, + { + "epoch": 2.4, + "learning_rate": 1.641646876916745e-05, + "loss": 0.2697, + "step": 5730 + }, + { + "epoch": 2.4, + "learning_rate": 1.6415168298029943e-05, + "loss": 0.241, + "step": 5731 + }, + { + "epoch": 2.4, + "learning_rate": 1.6413867642494106e-05, + "loss": 0.2836, + "step": 5732 + }, + { + "epoch": 2.4, + "learning_rate": 1.641256680259734e-05, + "loss": 0.2531, + "step": 5733 + }, + { + "epoch": 2.4, + "learning_rate": 1.641126577837703e-05, + "loss": 0.2368, + "step": 5734 + }, + { + "epoch": 2.4, + "learning_rate": 1.640996456987058e-05, + "loss": 0.2644, + "step": 5735 + }, + { + "epoch": 2.4, + "learning_rate": 1.6408663177115382e-05, + "loss": 0.2916, + "step": 5736 + }, + { + "epoch": 2.4, + "learning_rate": 1.6407361600148845e-05, + "loss": 0.2774, + "step": 5737 + }, + { + "epoch": 2.4, + "learning_rate": 1.6406059839008387e-05, + "loss": 0.2308, + "step": 5738 + }, + { + "epoch": 2.4, + "learning_rate": 1.6404757893731424e-05, + "loss": 0.2639, + "step": 5739 + }, + { + "epoch": 2.4, + "learning_rate": 1.640345576435538e-05, + "loss": 0.2969, + "step": 5740 + }, + { + "epoch": 2.4, + "learning_rate": 1.640215345091768e-05, + "loss": 0.2689, + "step": 5741 + }, + { + "epoch": 2.4, + "learning_rate": 1.640085095345576e-05, + "loss": 0.2603, + "step": 5742 + }, + { + "epoch": 2.4, + "learning_rate": 1.6399548272007063e-05, + "loss": 0.2703, + "step": 5743 + }, + { + "epoch": 2.4, + "learning_rate": 1.6398245406609027e-05, + "loss": 0.2651, + "step": 5744 + }, + { + "epoch": 2.41, + "learning_rate": 1.6396942357299104e-05, + "loss": 0.2588, + "step": 5745 + }, + { + "epoch": 2.41, + "learning_rate": 1.639563912411475e-05, + "loss": 0.2697, + "step": 5746 + }, + { + "epoch": 2.41, + "learning_rate": 1.6394335707093423e-05, + "loss": 0.2693, + "step": 5747 + }, + { + "epoch": 2.41, + "learning_rate": 1.6393032106272594e-05, + "loss": 0.2228, + "step": 5748 + }, + { + "epoch": 2.41, + "learning_rate": 1.639172832168973e-05, + "loss": 0.2704, + "step": 5749 + }, + { + "epoch": 2.41, + "learning_rate": 1.639042435338231e-05, + "loss": 0.2572, + "step": 5750 + }, + { + "epoch": 2.41, + "learning_rate": 1.6389120201387806e-05, + "loss": 0.2515, + "step": 5751 + }, + { + "epoch": 2.41, + "learning_rate": 1.6387815865743712e-05, + "loss": 0.2549, + "step": 5752 + }, + { + "epoch": 2.41, + "learning_rate": 1.638651134648752e-05, + "loss": 0.244, + "step": 5753 + }, + { + "epoch": 2.41, + "learning_rate": 1.638520664365673e-05, + "loss": 0.272, + "step": 5754 + }, + { + "epoch": 2.41, + "learning_rate": 1.638390175728884e-05, + "loss": 0.2539, + "step": 5755 + }, + { + "epoch": 2.41, + "learning_rate": 1.638259668742136e-05, + "loss": 0.251, + "step": 5756 + }, + { + "epoch": 2.41, + "learning_rate": 1.63812914340918e-05, + "loss": 0.2558, + "step": 5757 + }, + { + "epoch": 2.41, + "learning_rate": 1.6379985997337678e-05, + "loss": 0.2609, + "step": 5758 + }, + { + "epoch": 2.41, + "learning_rate": 1.6378680377196526e-05, + "loss": 0.2463, + "step": 5759 + }, + { + "epoch": 2.41, + "learning_rate": 1.6377374573705863e-05, + "loss": 0.2639, + "step": 5760 + }, + { + "epoch": 2.41, + "learning_rate": 1.6376068586903227e-05, + "loss": 0.272, + "step": 5761 + }, + { + "epoch": 2.41, + "learning_rate": 1.6374762416826157e-05, + "loss": 0.2646, + "step": 5762 + }, + { + "epoch": 2.41, + "learning_rate": 1.63734560635122e-05, + "loss": 0.251, + "step": 5763 + }, + { + "epoch": 2.41, + "learning_rate": 1.6372149526998905e-05, + "loss": 0.253, + "step": 5764 + }, + { + "epoch": 2.41, + "learning_rate": 1.6370842807323823e-05, + "loss": 0.243, + "step": 5765 + }, + { + "epoch": 2.41, + "learning_rate": 1.636953590452452e-05, + "loss": 0.2471, + "step": 5766 + }, + { + "epoch": 2.41, + "learning_rate": 1.636822881863856e-05, + "loss": 0.2734, + "step": 5767 + }, + { + "epoch": 2.41, + "learning_rate": 1.6366921549703513e-05, + "loss": 0.2282, + "step": 5768 + }, + { + "epoch": 2.42, + "learning_rate": 1.6365614097756956e-05, + "loss": 0.257, + "step": 5769 + }, + { + "epoch": 2.42, + "learning_rate": 1.636430646283647e-05, + "loss": 0.2875, + "step": 5770 + }, + { + "epoch": 2.42, + "learning_rate": 1.6362998644979645e-05, + "loss": 0.2659, + "step": 5771 + }, + { + "epoch": 2.42, + "learning_rate": 1.6361690644224066e-05, + "loss": 0.2742, + "step": 5772 + }, + { + "epoch": 2.42, + "learning_rate": 1.636038246060734e-05, + "loss": 0.2699, + "step": 5773 + }, + { + "epoch": 2.42, + "learning_rate": 1.635907409416706e-05, + "loss": 0.2658, + "step": 5774 + }, + { + "epoch": 2.42, + "learning_rate": 1.6357765544940837e-05, + "loss": 0.2769, + "step": 5775 + }, + { + "epoch": 2.42, + "learning_rate": 1.635645681296629e-05, + "loss": 0.2656, + "step": 5776 + }, + { + "epoch": 2.42, + "learning_rate": 1.6355147898281026e-05, + "loss": 0.2624, + "step": 5777 + }, + { + "epoch": 2.42, + "learning_rate": 1.6353838800922682e-05, + "loss": 0.2353, + "step": 5778 + }, + { + "epoch": 2.42, + "learning_rate": 1.6352529520928877e-05, + "loss": 0.2584, + "step": 5779 + }, + { + "epoch": 2.42, + "learning_rate": 1.6351220058337248e-05, + "loss": 0.2717, + "step": 5780 + }, + { + "epoch": 2.42, + "learning_rate": 1.6349910413185438e-05, + "loss": 0.2829, + "step": 5781 + }, + { + "epoch": 2.42, + "learning_rate": 1.6348600585511082e-05, + "loss": 0.2519, + "step": 5782 + }, + { + "epoch": 2.42, + "learning_rate": 1.634729057535184e-05, + "loss": 0.2674, + "step": 5783 + }, + { + "epoch": 2.42, + "learning_rate": 1.634598038274536e-05, + "loss": 0.2715, + "step": 5784 + }, + { + "epoch": 2.42, + "learning_rate": 1.6344670007729305e-05, + "loss": 0.2771, + "step": 5785 + }, + { + "epoch": 2.42, + "learning_rate": 1.6343359450341344e-05, + "loss": 0.2167, + "step": 5786 + }, + { + "epoch": 2.42, + "learning_rate": 1.6342048710619143e-05, + "loss": 0.2417, + "step": 5787 + }, + { + "epoch": 2.42, + "learning_rate": 1.6340737788600378e-05, + "loss": 0.2649, + "step": 5788 + }, + { + "epoch": 2.42, + "learning_rate": 1.6339426684322735e-05, + "loss": 0.2508, + "step": 5789 + }, + { + "epoch": 2.42, + "learning_rate": 1.6338115397823897e-05, + "loss": 0.2406, + "step": 5790 + }, + { + "epoch": 2.42, + "learning_rate": 1.6336803929141557e-05, + "loss": 0.2676, + "step": 5791 + }, + { + "epoch": 2.42, + "learning_rate": 1.6335492278313405e-05, + "loss": 0.2332, + "step": 5792 + }, + { + "epoch": 2.43, + "learning_rate": 1.6334180445377156e-05, + "loss": 0.251, + "step": 5793 + }, + { + "epoch": 2.43, + "learning_rate": 1.6332868430370508e-05, + "loss": 0.2471, + "step": 5794 + }, + { + "epoch": 2.43, + "learning_rate": 1.6331556233331178e-05, + "loss": 0.2658, + "step": 5795 + }, + { + "epoch": 2.43, + "learning_rate": 1.633024385429688e-05, + "loss": 0.2872, + "step": 5796 + }, + { + "epoch": 2.43, + "learning_rate": 1.6328931293305345e-05, + "loss": 0.2564, + "step": 5797 + }, + { + "epoch": 2.43, + "learning_rate": 1.632761855039429e-05, + "loss": 0.2626, + "step": 5798 + }, + { + "epoch": 2.43, + "learning_rate": 1.632630562560146e-05, + "loss": 0.2719, + "step": 5799 + }, + { + "epoch": 2.43, + "learning_rate": 1.6324992518964586e-05, + "loss": 0.248, + "step": 5800 + }, + { + "epoch": 2.43, + "learning_rate": 1.6323679230521413e-05, + "loss": 0.2723, + "step": 5801 + }, + { + "epoch": 2.43, + "learning_rate": 1.6322365760309698e-05, + "loss": 0.2574, + "step": 5802 + }, + { + "epoch": 2.43, + "learning_rate": 1.6321052108367187e-05, + "loss": 0.2379, + "step": 5803 + }, + { + "epoch": 2.43, + "learning_rate": 1.631973827473164e-05, + "loss": 0.2748, + "step": 5804 + }, + { + "epoch": 2.43, + "learning_rate": 1.6318424259440827e-05, + "loss": 0.2571, + "step": 5805 + }, + { + "epoch": 2.43, + "learning_rate": 1.631711006253251e-05, + "loss": 0.266, + "step": 5806 + }, + { + "epoch": 2.43, + "learning_rate": 1.6315795684044476e-05, + "loss": 0.2561, + "step": 5807 + }, + { + "epoch": 2.43, + "learning_rate": 1.63144811240145e-05, + "loss": 0.2786, + "step": 5808 + }, + { + "epoch": 2.43, + "learning_rate": 1.6313166382480362e-05, + "loss": 0.2648, + "step": 5809 + }, + { + "epoch": 2.43, + "learning_rate": 1.631185145947986e-05, + "loss": 0.2931, + "step": 5810 + }, + { + "epoch": 2.43, + "learning_rate": 1.631053635505079e-05, + "loss": 0.2583, + "step": 5811 + }, + { + "epoch": 2.43, + "learning_rate": 1.6309221069230954e-05, + "loss": 0.2413, + "step": 5812 + }, + { + "epoch": 2.43, + "learning_rate": 1.6307905602058158e-05, + "loss": 0.2587, + "step": 5813 + }, + { + "epoch": 2.43, + "learning_rate": 1.6306589953570207e-05, + "loss": 0.2623, + "step": 5814 + }, + { + "epoch": 2.43, + "learning_rate": 1.6305274123804934e-05, + "loss": 0.2557, + "step": 5815 + }, + { + "epoch": 2.44, + "learning_rate": 1.6303958112800144e-05, + "loss": 0.2457, + "step": 5816 + }, + { + "epoch": 2.44, + "learning_rate": 1.630264192059367e-05, + "loss": 0.2665, + "step": 5817 + }, + { + "epoch": 2.44, + "learning_rate": 1.6301325547223353e-05, + "loss": 0.2545, + "step": 5818 + }, + { + "epoch": 2.44, + "learning_rate": 1.630000899272702e-05, + "loss": 0.2427, + "step": 5819 + }, + { + "epoch": 2.44, + "learning_rate": 1.6298692257142525e-05, + "loss": 0.2687, + "step": 5820 + }, + { + "epoch": 2.44, + "learning_rate": 1.6297375340507705e-05, + "loss": 0.2529, + "step": 5821 + }, + { + "epoch": 2.44, + "learning_rate": 1.629605824286042e-05, + "loss": 0.2595, + "step": 5822 + }, + { + "epoch": 2.44, + "learning_rate": 1.629474096423853e-05, + "loss": 0.2597, + "step": 5823 + }, + { + "epoch": 2.44, + "learning_rate": 1.6293423504679898e-05, + "loss": 0.2469, + "step": 5824 + }, + { + "epoch": 2.44, + "learning_rate": 1.6292105864222386e-05, + "loss": 0.2507, + "step": 5825 + }, + { + "epoch": 2.44, + "learning_rate": 1.629078804290388e-05, + "loss": 0.2638, + "step": 5826 + }, + { + "epoch": 2.44, + "learning_rate": 1.628947004076225e-05, + "loss": 0.2452, + "step": 5827 + }, + { + "epoch": 2.44, + "learning_rate": 1.628815185783539e-05, + "loss": 0.2736, + "step": 5828 + }, + { + "epoch": 2.44, + "learning_rate": 1.628683349416118e-05, + "loss": 0.2595, + "step": 5829 + }, + { + "epoch": 2.44, + "learning_rate": 1.6285514949777522e-05, + "loss": 0.2602, + "step": 5830 + }, + { + "epoch": 2.44, + "learning_rate": 1.628419622472231e-05, + "loss": 0.2559, + "step": 5831 + }, + { + "epoch": 2.44, + "learning_rate": 1.6282877319033462e-05, + "loss": 0.2772, + "step": 5832 + }, + { + "epoch": 2.44, + "learning_rate": 1.6281558232748874e-05, + "loss": 0.2583, + "step": 5833 + }, + { + "epoch": 2.44, + "learning_rate": 1.6280238965906474e-05, + "loss": 0.2928, + "step": 5834 + }, + { + "epoch": 2.44, + "learning_rate": 1.6278919518544175e-05, + "loss": 0.2403, + "step": 5835 + }, + { + "epoch": 2.44, + "learning_rate": 1.627759989069991e-05, + "loss": 0.2579, + "step": 5836 + }, + { + "epoch": 2.44, + "learning_rate": 1.6276280082411604e-05, + "loss": 0.2968, + "step": 5837 + }, + { + "epoch": 2.44, + "learning_rate": 1.6274960093717192e-05, + "loss": 0.2501, + "step": 5838 + }, + { + "epoch": 2.44, + "learning_rate": 1.6273639924654626e-05, + "loss": 0.2598, + "step": 5839 + }, + { + "epoch": 2.45, + "learning_rate": 1.627231957526185e-05, + "loss": 0.2334, + "step": 5840 + }, + { + "epoch": 2.45, + "learning_rate": 1.6270999045576808e-05, + "loss": 0.2504, + "step": 5841 + }, + { + "epoch": 2.45, + "learning_rate": 1.6269678335637464e-05, + "loss": 0.2664, + "step": 5842 + }, + { + "epoch": 2.45, + "learning_rate": 1.626835744548178e-05, + "loss": 0.2308, + "step": 5843 + }, + { + "epoch": 2.45, + "learning_rate": 1.6267036375147728e-05, + "loss": 0.2584, + "step": 5844 + }, + { + "epoch": 2.45, + "learning_rate": 1.6265715124673272e-05, + "loss": 0.2493, + "step": 5845 + }, + { + "epoch": 2.45, + "learning_rate": 1.6264393694096396e-05, + "loss": 0.2382, + "step": 5846 + }, + { + "epoch": 2.45, + "learning_rate": 1.6263072083455084e-05, + "loss": 0.2394, + "step": 5847 + }, + { + "epoch": 2.45, + "learning_rate": 1.626175029278732e-05, + "loss": 0.2658, + "step": 5848 + }, + { + "epoch": 2.45, + "learning_rate": 1.62604283221311e-05, + "loss": 0.2416, + "step": 5849 + }, + { + "epoch": 2.45, + "learning_rate": 1.6259106171524424e-05, + "loss": 0.2642, + "step": 5850 + }, + { + "epoch": 2.45, + "learning_rate": 1.6257783841005297e-05, + "loss": 0.2426, + "step": 5851 + }, + { + "epoch": 2.45, + "learning_rate": 1.6256461330611727e-05, + "loss": 0.2554, + "step": 5852 + }, + { + "epoch": 2.45, + "learning_rate": 1.6255138640381723e-05, + "loss": 0.2516, + "step": 5853 + }, + { + "epoch": 2.45, + "learning_rate": 1.6253815770353316e-05, + "loss": 0.2651, + "step": 5854 + }, + { + "epoch": 2.45, + "learning_rate": 1.625249272056452e-05, + "loss": 0.2797, + "step": 5855 + }, + { + "epoch": 2.45, + "learning_rate": 1.625116949105337e-05, + "loss": 0.2711, + "step": 5856 + }, + { + "epoch": 2.45, + "learning_rate": 1.62498460818579e-05, + "loss": 0.2555, + "step": 5857 + }, + { + "epoch": 2.45, + "learning_rate": 1.6248522493016153e-05, + "loss": 0.2479, + "step": 5858 + }, + { + "epoch": 2.45, + "learning_rate": 1.624719872456617e-05, + "loss": 0.2662, + "step": 5859 + }, + { + "epoch": 2.45, + "learning_rate": 1.6245874776546005e-05, + "loss": 0.2627, + "step": 5860 + }, + { + "epoch": 2.45, + "learning_rate": 1.624455064899371e-05, + "loss": 0.247, + "step": 5861 + }, + { + "epoch": 2.45, + "learning_rate": 1.6243226341947348e-05, + "loss": 0.266, + "step": 5862 + }, + { + "epoch": 2.45, + "learning_rate": 1.624190185544498e-05, + "loss": 0.25, + "step": 5863 + }, + { + "epoch": 2.46, + "learning_rate": 1.6240577189524688e-05, + "loss": 0.2389, + "step": 5864 + }, + { + "epoch": 2.46, + "learning_rate": 1.6239252344224545e-05, + "loss": 0.2448, + "step": 5865 + }, + { + "epoch": 2.46, + "learning_rate": 1.6237927319582624e-05, + "loss": 0.2697, + "step": 5866 + }, + { + "epoch": 2.46, + "learning_rate": 1.623660211563702e-05, + "loss": 0.2607, + "step": 5867 + }, + { + "epoch": 2.46, + "learning_rate": 1.6235276732425823e-05, + "loss": 0.2711, + "step": 5868 + }, + { + "epoch": 2.46, + "learning_rate": 1.623395116998713e-05, + "loss": 0.2854, + "step": 5869 + }, + { + "epoch": 2.46, + "learning_rate": 1.6232625428359042e-05, + "loss": 0.2689, + "step": 5870 + }, + { + "epoch": 2.46, + "learning_rate": 1.6231299507579668e-05, + "loss": 0.2591, + "step": 5871 + }, + { + "epoch": 2.46, + "learning_rate": 1.6229973407687117e-05, + "loss": 0.2603, + "step": 5872 + }, + { + "epoch": 2.46, + "learning_rate": 1.622864712871951e-05, + "loss": 0.2572, + "step": 5873 + }, + { + "epoch": 2.46, + "learning_rate": 1.6227320670714966e-05, + "loss": 0.2879, + "step": 5874 + }, + { + "epoch": 2.46, + "learning_rate": 1.6225994033711617e-05, + "loss": 0.2058, + "step": 5875 + }, + { + "epoch": 2.46, + "learning_rate": 1.6224667217747594e-05, + "loss": 0.26, + "step": 5876 + }, + { + "epoch": 2.46, + "learning_rate": 1.6223340222861037e-05, + "loss": 0.2241, + "step": 5877 + }, + { + "epoch": 2.46, + "learning_rate": 1.6222013049090086e-05, + "loss": 0.2499, + "step": 5878 + }, + { + "epoch": 2.46, + "learning_rate": 1.622068569647289e-05, + "loss": 0.2711, + "step": 5879 + }, + { + "epoch": 2.46, + "learning_rate": 1.6219358165047606e-05, + "loss": 0.2468, + "step": 5880 + }, + { + "epoch": 2.46, + "learning_rate": 1.6218030454852395e-05, + "loss": 0.2711, + "step": 5881 + }, + { + "epoch": 2.46, + "learning_rate": 1.621670256592541e-05, + "loss": 0.2464, + "step": 5882 + }, + { + "epoch": 2.46, + "learning_rate": 1.6215374498304827e-05, + "loss": 0.2739, + "step": 5883 + }, + { + "epoch": 2.46, + "learning_rate": 1.6214046252028823e-05, + "loss": 0.2529, + "step": 5884 + }, + { + "epoch": 2.46, + "learning_rate": 1.621271782713557e-05, + "loss": 0.2508, + "step": 5885 + }, + { + "epoch": 2.46, + "learning_rate": 1.621138922366326e-05, + "loss": 0.245, + "step": 5886 + }, + { + "epoch": 2.46, + "learning_rate": 1.6210060441650073e-05, + "loss": 0.2693, + "step": 5887 + }, + { + "epoch": 2.47, + "learning_rate": 1.6208731481134214e-05, + "loss": 0.2377, + "step": 5888 + }, + { + "epoch": 2.47, + "learning_rate": 1.6207402342153876e-05, + "loss": 0.2442, + "step": 5889 + }, + { + "epoch": 2.47, + "learning_rate": 1.6206073024747264e-05, + "loss": 0.2673, + "step": 5890 + }, + { + "epoch": 2.47, + "learning_rate": 1.6204743528952592e-05, + "loss": 0.2758, + "step": 5891 + }, + { + "epoch": 2.47, + "learning_rate": 1.6203413854808076e-05, + "loss": 0.2612, + "step": 5892 + }, + { + "epoch": 2.47, + "learning_rate": 1.620208400235193e-05, + "loss": 0.2591, + "step": 5893 + }, + { + "epoch": 2.47, + "learning_rate": 1.6200753971622387e-05, + "loss": 0.2512, + "step": 5894 + }, + { + "epoch": 2.47, + "learning_rate": 1.619942376265767e-05, + "loss": 0.2587, + "step": 5895 + }, + { + "epoch": 2.47, + "learning_rate": 1.619809337549602e-05, + "loss": 0.2153, + "step": 5896 + }, + { + "epoch": 2.47, + "learning_rate": 1.6196762810175674e-05, + "loss": 0.2638, + "step": 5897 + }, + { + "epoch": 2.47, + "learning_rate": 1.6195432066734878e-05, + "loss": 0.2553, + "step": 5898 + }, + { + "epoch": 2.47, + "learning_rate": 1.619410114521189e-05, + "loss": 0.2529, + "step": 5899 + }, + { + "epoch": 2.47, + "learning_rate": 1.619277004564496e-05, + "loss": 0.2724, + "step": 5900 + }, + { + "epoch": 2.47, + "learning_rate": 1.619143876807235e-05, + "loss": 0.2458, + "step": 5901 + }, + { + "epoch": 2.47, + "learning_rate": 1.6190107312532327e-05, + "loss": 0.2857, + "step": 5902 + }, + { + "epoch": 2.47, + "learning_rate": 1.618877567906316e-05, + "loss": 0.2431, + "step": 5903 + }, + { + "epoch": 2.47, + "learning_rate": 1.618744386770313e-05, + "loss": 0.2515, + "step": 5904 + }, + { + "epoch": 2.47, + "learning_rate": 1.6186111878490515e-05, + "loss": 0.2522, + "step": 5905 + }, + { + "epoch": 2.47, + "learning_rate": 1.6184779711463607e-05, + "loss": 0.2592, + "step": 5906 + }, + { + "epoch": 2.47, + "learning_rate": 1.618344736666069e-05, + "loss": 0.2663, + "step": 5907 + }, + { + "epoch": 2.47, + "learning_rate": 1.618211484412007e-05, + "loss": 0.2633, + "step": 5908 + }, + { + "epoch": 2.47, + "learning_rate": 1.6180782143880042e-05, + "loss": 0.2422, + "step": 5909 + }, + { + "epoch": 2.47, + "learning_rate": 1.6179449265978913e-05, + "loss": 0.2889, + "step": 5910 + }, + { + "epoch": 2.47, + "learning_rate": 1.6178116210455006e-05, + "loss": 0.3035, + "step": 5911 + }, + { + "epoch": 2.48, + "learning_rate": 1.6176782977346626e-05, + "loss": 0.2714, + "step": 5912 + }, + { + "epoch": 2.48, + "learning_rate": 1.61754495666921e-05, + "loss": 0.254, + "step": 5913 + }, + { + "epoch": 2.48, + "learning_rate": 1.617411597852976e-05, + "loss": 0.2454, + "step": 5914 + }, + { + "epoch": 2.48, + "learning_rate": 1.617278221289793e-05, + "loss": 0.2544, + "step": 5915 + }, + { + "epoch": 2.48, + "learning_rate": 1.6171448269834956e-05, + "loss": 0.2736, + "step": 5916 + }, + { + "epoch": 2.48, + "learning_rate": 1.6170114149379178e-05, + "loss": 0.2428, + "step": 5917 + }, + { + "epoch": 2.48, + "learning_rate": 1.6168779851568947e-05, + "loss": 0.2423, + "step": 5918 + }, + { + "epoch": 2.48, + "learning_rate": 1.6167445376442604e-05, + "loss": 0.2222, + "step": 5919 + }, + { + "epoch": 2.48, + "learning_rate": 1.6166110724038524e-05, + "loss": 0.2606, + "step": 5920 + }, + { + "epoch": 2.48, + "learning_rate": 1.616477589439506e-05, + "loss": 0.2681, + "step": 5921 + }, + { + "epoch": 2.48, + "learning_rate": 1.6163440887550587e-05, + "loss": 0.2595, + "step": 5922 + }, + { + "epoch": 2.48, + "learning_rate": 1.616210570354347e-05, + "loss": 0.2684, + "step": 5923 + }, + { + "epoch": 2.48, + "learning_rate": 1.6160770342412097e-05, + "loss": 0.2561, + "step": 5924 + }, + { + "epoch": 2.48, + "learning_rate": 1.6159434804194845e-05, + "loss": 0.2489, + "step": 5925 + }, + { + "epoch": 2.48, + "learning_rate": 1.615809908893011e-05, + "loss": 0.2594, + "step": 5926 + }, + { + "epoch": 2.48, + "learning_rate": 1.6156763196656276e-05, + "loss": 0.2583, + "step": 5927 + }, + { + "epoch": 2.48, + "learning_rate": 1.6155427127411748e-05, + "loss": 0.2552, + "step": 5928 + }, + { + "epoch": 2.48, + "learning_rate": 1.615409088123493e-05, + "loss": 0.2666, + "step": 5929 + }, + { + "epoch": 2.48, + "learning_rate": 1.615275445816423e-05, + "loss": 0.2609, + "step": 5930 + }, + { + "epoch": 2.48, + "learning_rate": 1.6151417858238064e-05, + "loss": 0.2475, + "step": 5931 + }, + { + "epoch": 2.48, + "learning_rate": 1.615008108149485e-05, + "loss": 0.2644, + "step": 5932 + }, + { + "epoch": 2.48, + "learning_rate": 1.6148744127973014e-05, + "loss": 0.2486, + "step": 5933 + }, + { + "epoch": 2.48, + "learning_rate": 1.614740699771098e-05, + "loss": 0.2463, + "step": 5934 + }, + { + "epoch": 2.48, + "learning_rate": 1.614606969074719e-05, + "loss": 0.2454, + "step": 5935 + }, + { + "epoch": 2.49, + "learning_rate": 1.6144732207120083e-05, + "loss": 0.2456, + "step": 5936 + }, + { + "epoch": 2.49, + "learning_rate": 1.61433945468681e-05, + "loss": 0.2615, + "step": 5937 + }, + { + "epoch": 2.49, + "learning_rate": 1.614205671002969e-05, + "loss": 0.2378, + "step": 5938 + }, + { + "epoch": 2.49, + "learning_rate": 1.6140718696643313e-05, + "loss": 0.2868, + "step": 5939 + }, + { + "epoch": 2.49, + "learning_rate": 1.6139380506747424e-05, + "loss": 0.2496, + "step": 5940 + }, + { + "epoch": 2.49, + "learning_rate": 1.6138042140380493e-05, + "loss": 0.2343, + "step": 5941 + }, + { + "epoch": 2.49, + "learning_rate": 1.6136703597580984e-05, + "loss": 0.2396, + "step": 5942 + }, + { + "epoch": 2.49, + "learning_rate": 1.613536487838738e-05, + "loss": 0.2665, + "step": 5943 + }, + { + "epoch": 2.49, + "learning_rate": 1.6134025982838153e-05, + "loss": 0.2772, + "step": 5944 + }, + { + "epoch": 2.49, + "learning_rate": 1.6132686910971792e-05, + "loss": 0.2544, + "step": 5945 + }, + { + "epoch": 2.49, + "learning_rate": 1.6131347662826793e-05, + "loss": 0.2489, + "step": 5946 + }, + { + "epoch": 2.49, + "learning_rate": 1.613000823844164e-05, + "loss": 0.2798, + "step": 5947 + }, + { + "epoch": 2.49, + "learning_rate": 1.6128668637854847e-05, + "loss": 0.2367, + "step": 5948 + }, + { + "epoch": 2.49, + "learning_rate": 1.6127328861104906e-05, + "loss": 0.2688, + "step": 5949 + }, + { + "epoch": 2.49, + "learning_rate": 1.6125988908230337e-05, + "loss": 0.2514, + "step": 5950 + }, + { + "epoch": 2.49, + "learning_rate": 1.6124648779269654e-05, + "loss": 0.2594, + "step": 5951 + }, + { + "epoch": 2.49, + "learning_rate": 1.6123308474261374e-05, + "loss": 0.2542, + "step": 5952 + }, + { + "epoch": 2.49, + "learning_rate": 1.6121967993244026e-05, + "loss": 0.2548, + "step": 5953 + }, + { + "epoch": 2.49, + "learning_rate": 1.6120627336256142e-05, + "loss": 0.2928, + "step": 5954 + }, + { + "epoch": 2.49, + "learning_rate": 1.6119286503336256e-05, + "loss": 0.2584, + "step": 5955 + }, + { + "epoch": 2.49, + "learning_rate": 1.611794549452291e-05, + "loss": 0.2732, + "step": 5956 + }, + { + "epoch": 2.49, + "learning_rate": 1.611660430985465e-05, + "loss": 0.2218, + "step": 5957 + }, + { + "epoch": 2.49, + "learning_rate": 1.6115262949370025e-05, + "loss": 0.2893, + "step": 5958 + }, + { + "epoch": 2.49, + "learning_rate": 1.6113921413107597e-05, + "loss": 0.2684, + "step": 5959 + }, + { + "epoch": 2.5, + "learning_rate": 1.6112579701105923e-05, + "loss": 0.2812, + "step": 5960 + }, + { + "epoch": 2.5, + "learning_rate": 1.6111237813403568e-05, + "loss": 0.2759, + "step": 5961 + }, + { + "epoch": 2.5, + "learning_rate": 1.61098957500391e-05, + "loss": 0.285, + "step": 5962 + }, + { + "epoch": 2.5, + "learning_rate": 1.6108553511051106e-05, + "loss": 0.2418, + "step": 5963 + }, + { + "epoch": 2.5, + "learning_rate": 1.610721109647816e-05, + "loss": 0.2605, + "step": 5964 + }, + { + "epoch": 2.5, + "learning_rate": 1.610586850635885e-05, + "loss": 0.263, + "step": 5965 + }, + { + "epoch": 2.5, + "learning_rate": 1.6104525740731772e-05, + "loss": 0.28, + "step": 5966 + }, + { + "epoch": 2.5, + "learning_rate": 1.6103182799635514e-05, + "loss": 0.246, + "step": 5967 + }, + { + "epoch": 2.5, + "learning_rate": 1.6101839683108686e-05, + "loss": 0.2725, + "step": 5968 + }, + { + "epoch": 2.5, + "learning_rate": 1.610049639118989e-05, + "loss": 0.2355, + "step": 5969 + }, + { + "epoch": 2.5, + "learning_rate": 1.6099152923917737e-05, + "loss": 0.2961, + "step": 5970 + }, + { + "epoch": 2.5, + "learning_rate": 1.6097809281330848e-05, + "loss": 0.2723, + "step": 5971 + }, + { + "epoch": 2.5, + "learning_rate": 1.609646546346784e-05, + "loss": 0.2695, + "step": 5972 + }, + { + "epoch": 2.5, + "learning_rate": 1.6095121470367345e-05, + "loss": 0.287, + "step": 5973 + }, + { + "epoch": 2.5, + "learning_rate": 1.609377730206799e-05, + "loss": 0.2501, + "step": 5974 + }, + { + "epoch": 2.5, + "learning_rate": 1.6092432958608413e-05, + "loss": 0.2542, + "step": 5975 + }, + { + "epoch": 2.5, + "learning_rate": 1.609108844002726e-05, + "loss": 0.259, + "step": 5976 + }, + { + "epoch": 2.5, + "learning_rate": 1.608974374636317e-05, + "loss": 0.2713, + "step": 5977 + }, + { + "epoch": 2.5, + "learning_rate": 1.6088398877654806e-05, + "loss": 0.2765, + "step": 5978 + }, + { + "epoch": 2.5, + "learning_rate": 1.6087053833940815e-05, + "loss": 0.2254, + "step": 5979 + }, + { + "epoch": 2.5, + "learning_rate": 1.6085708615259863e-05, + "loss": 0.2434, + "step": 5980 + }, + { + "epoch": 2.5, + "learning_rate": 1.608436322165062e-05, + "loss": 0.2638, + "step": 5981 + }, + { + "epoch": 2.5, + "learning_rate": 1.6083017653151752e-05, + "loss": 0.2749, + "step": 5982 + }, + { + "epoch": 2.5, + "learning_rate": 1.6081671909801944e-05, + "loss": 0.2938, + "step": 5983 + }, + { + "epoch": 2.51, + "learning_rate": 1.6080325991639866e-05, + "loss": 0.2419, + "step": 5984 + }, + { + "epoch": 2.51, + "learning_rate": 1.6078979898704223e-05, + "loss": 0.2736, + "step": 5985 + }, + { + "epoch": 2.51, + "learning_rate": 1.607763363103369e-05, + "loss": 0.2498, + "step": 5986 + }, + { + "epoch": 2.51, + "learning_rate": 1.6076287188666972e-05, + "loss": 0.2468, + "step": 5987 + }, + { + "epoch": 2.51, + "learning_rate": 1.6074940571642773e-05, + "loss": 0.2543, + "step": 5988 + }, + { + "epoch": 2.51, + "learning_rate": 1.6073593779999795e-05, + "loss": 0.2549, + "step": 5989 + }, + { + "epoch": 2.51, + "learning_rate": 1.6072246813776758e-05, + "loss": 0.2918, + "step": 5990 + }, + { + "epoch": 2.51, + "learning_rate": 1.6070899673012372e-05, + "loss": 0.2755, + "step": 5991 + }, + { + "epoch": 2.51, + "learning_rate": 1.6069552357745363e-05, + "loss": 0.2557, + "step": 5992 + }, + { + "epoch": 2.51, + "learning_rate": 1.6068204868014455e-05, + "loss": 0.2732, + "step": 5993 + }, + { + "epoch": 2.51, + "learning_rate": 1.6066857203858387e-05, + "loss": 0.2549, + "step": 5994 + }, + { + "epoch": 2.51, + "learning_rate": 1.6065509365315893e-05, + "loss": 0.2497, + "step": 5995 + }, + { + "epoch": 2.51, + "learning_rate": 1.6064161352425708e-05, + "loss": 0.2364, + "step": 5996 + }, + { + "epoch": 2.51, + "learning_rate": 1.6062813165226594e-05, + "loss": 0.2453, + "step": 5997 + }, + { + "epoch": 2.51, + "learning_rate": 1.606146480375729e-05, + "loss": 0.2279, + "step": 5998 + }, + { + "epoch": 2.51, + "learning_rate": 1.606011626805656e-05, + "loss": 0.2577, + "step": 5999 + }, + { + "epoch": 2.51, + "learning_rate": 1.6058767558163168e-05, + "loss": 0.2731, + "step": 6000 + }, + { + "epoch": 2.51, + "learning_rate": 1.6057418674115876e-05, + "loss": 0.2782, + "step": 6001 + }, + { + "epoch": 2.51, + "learning_rate": 1.605606961595346e-05, + "loss": 0.2451, + "step": 6002 + }, + { + "epoch": 2.51, + "learning_rate": 1.60547203837147e-05, + "loss": 0.2513, + "step": 6003 + }, + { + "epoch": 2.51, + "learning_rate": 1.605337097743837e-05, + "loss": 0.2531, + "step": 6004 + }, + { + "epoch": 2.51, + "learning_rate": 1.605202139716327e-05, + "loss": 0.2754, + "step": 6005 + }, + { + "epoch": 2.51, + "learning_rate": 1.605067164292818e-05, + "loss": 0.2522, + "step": 6006 + }, + { + "epoch": 2.51, + "learning_rate": 1.6049321714771907e-05, + "loss": 0.2427, + "step": 6007 + }, + { + "epoch": 2.52, + "learning_rate": 1.6047971612733246e-05, + "loss": 0.2295, + "step": 6008 + }, + { + "epoch": 2.52, + "learning_rate": 1.6046621336851007e-05, + "loss": 0.272, + "step": 6009 + }, + { + "epoch": 2.52, + "learning_rate": 1.6045270887164005e-05, + "loss": 0.2655, + "step": 6010 + }, + { + "epoch": 2.52, + "learning_rate": 1.6043920263711056e-05, + "loss": 0.2788, + "step": 6011 + }, + { + "epoch": 2.52, + "learning_rate": 1.6042569466530985e-05, + "loss": 0.2805, + "step": 6012 + }, + { + "epoch": 2.52, + "learning_rate": 1.604121849566261e-05, + "loss": 0.2553, + "step": 6013 + }, + { + "epoch": 2.52, + "learning_rate": 1.6039867351144778e-05, + "loss": 0.2407, + "step": 6014 + }, + { + "epoch": 2.52, + "learning_rate": 1.6038516033016314e-05, + "loss": 0.2728, + "step": 6015 + }, + { + "epoch": 2.52, + "learning_rate": 1.6037164541316067e-05, + "loss": 0.2442, + "step": 6016 + }, + { + "epoch": 2.52, + "learning_rate": 1.603581287608288e-05, + "loss": 0.2559, + "step": 6017 + }, + { + "epoch": 2.52, + "learning_rate": 1.603446103735561e-05, + "loss": 0.2662, + "step": 6018 + }, + { + "epoch": 2.52, + "learning_rate": 1.6033109025173113e-05, + "loss": 0.2541, + "step": 6019 + }, + { + "epoch": 2.52, + "learning_rate": 1.6031756839574247e-05, + "loss": 0.2909, + "step": 6020 + }, + { + "epoch": 2.52, + "learning_rate": 1.6030404480597888e-05, + "loss": 0.2366, + "step": 6021 + }, + { + "epoch": 2.52, + "learning_rate": 1.60290519482829e-05, + "loss": 0.2578, + "step": 6022 + }, + { + "epoch": 2.52, + "learning_rate": 1.6027699242668166e-05, + "loss": 0.2686, + "step": 6023 + }, + { + "epoch": 2.52, + "learning_rate": 1.6026346363792565e-05, + "loss": 0.2634, + "step": 6024 + }, + { + "epoch": 2.52, + "learning_rate": 1.6024993311694986e-05, + "loss": 0.267, + "step": 6025 + }, + { + "epoch": 2.52, + "learning_rate": 1.6023640086414318e-05, + "loss": 0.2677, + "step": 6026 + }, + { + "epoch": 2.52, + "learning_rate": 1.6022286687989466e-05, + "loss": 0.2588, + "step": 6027 + }, + { + "epoch": 2.52, + "learning_rate": 1.6020933116459324e-05, + "loss": 0.254, + "step": 6028 + }, + { + "epoch": 2.52, + "learning_rate": 1.6019579371862803e-05, + "loss": 0.2767, + "step": 6029 + }, + { + "epoch": 2.52, + "learning_rate": 1.6018225454238816e-05, + "loss": 0.248, + "step": 6030 + }, + { + "epoch": 2.53, + "learning_rate": 1.6016871363626277e-05, + "loss": 0.2448, + "step": 6031 + }, + { + "epoch": 2.53, + "learning_rate": 1.6015517100064114e-05, + "loss": 0.2472, + "step": 6032 + }, + { + "epoch": 2.53, + "learning_rate": 1.601416266359124e-05, + "loss": 0.2665, + "step": 6033 + }, + { + "epoch": 2.53, + "learning_rate": 1.6012808054246608e-05, + "loss": 0.2622, + "step": 6034 + }, + { + "epoch": 2.53, + "learning_rate": 1.601145327206914e-05, + "loss": 0.2483, + "step": 6035 + }, + { + "epoch": 2.53, + "learning_rate": 1.601009831709778e-05, + "loss": 0.2636, + "step": 6036 + }, + { + "epoch": 2.53, + "learning_rate": 1.600874318937148e-05, + "loss": 0.2706, + "step": 6037 + }, + { + "epoch": 2.53, + "learning_rate": 1.600738788892919e-05, + "loss": 0.2803, + "step": 6038 + }, + { + "epoch": 2.53, + "learning_rate": 1.6006032415809866e-05, + "loss": 0.2737, + "step": 6039 + }, + { + "epoch": 2.53, + "learning_rate": 1.6004676770052466e-05, + "loss": 0.2622, + "step": 6040 + }, + { + "epoch": 2.53, + "learning_rate": 1.600332095169596e-05, + "loss": 0.2615, + "step": 6041 + }, + { + "epoch": 2.53, + "learning_rate": 1.6001964960779326e-05, + "loss": 0.2653, + "step": 6042 + }, + { + "epoch": 2.53, + "learning_rate": 1.600060879734153e-05, + "loss": 0.2388, + "step": 6043 + }, + { + "epoch": 2.53, + "learning_rate": 1.5999252461421564e-05, + "loss": 0.2292, + "step": 6044 + }, + { + "epoch": 2.53, + "learning_rate": 1.5997895953058407e-05, + "loss": 0.2427, + "step": 6045 + }, + { + "epoch": 2.53, + "learning_rate": 1.5996539272291052e-05, + "loss": 0.2385, + "step": 6046 + }, + { + "epoch": 2.53, + "learning_rate": 1.59951824191585e-05, + "loss": 0.2729, + "step": 6047 + }, + { + "epoch": 2.53, + "learning_rate": 1.5993825393699746e-05, + "loss": 0.2514, + "step": 6048 + }, + { + "epoch": 2.53, + "learning_rate": 1.59924681959538e-05, + "loss": 0.2577, + "step": 6049 + }, + { + "epoch": 2.53, + "learning_rate": 1.5991110825959676e-05, + "loss": 0.245, + "step": 6050 + }, + { + "epoch": 2.53, + "learning_rate": 1.598975328375639e-05, + "loss": 0.2652, + "step": 6051 + }, + { + "epoch": 2.53, + "learning_rate": 1.5988395569382954e-05, + "loss": 0.2477, + "step": 6052 + }, + { + "epoch": 2.53, + "learning_rate": 1.5987037682878405e-05, + "loss": 0.2296, + "step": 6053 + }, + { + "epoch": 2.53, + "learning_rate": 1.5985679624281772e-05, + "loss": 0.244, + "step": 6054 + }, + { + "epoch": 2.54, + "learning_rate": 1.5984321393632087e-05, + "loss": 0.2329, + "step": 6055 + }, + { + "epoch": 2.54, + "learning_rate": 1.5982962990968397e-05, + "loss": 0.2332, + "step": 6056 + }, + { + "epoch": 2.54, + "learning_rate": 1.5981604416329742e-05, + "loss": 0.243, + "step": 6057 + }, + { + "epoch": 2.54, + "learning_rate": 1.598024566975518e-05, + "loss": 0.2464, + "step": 6058 + }, + { + "epoch": 2.54, + "learning_rate": 1.597888675128376e-05, + "loss": 0.2263, + "step": 6059 + }, + { + "epoch": 2.54, + "learning_rate": 1.5977527660954545e-05, + "loss": 0.2715, + "step": 6060 + }, + { + "epoch": 2.54, + "learning_rate": 1.5976168398806602e-05, + "loss": 0.2625, + "step": 6061 + }, + { + "epoch": 2.54, + "learning_rate": 1.5974808964879005e-05, + "loss": 0.2729, + "step": 6062 + }, + { + "epoch": 2.54, + "learning_rate": 1.5973449359210823e-05, + "loss": 0.2417, + "step": 6063 + }, + { + "epoch": 2.54, + "learning_rate": 1.597208958184114e-05, + "loss": 0.2355, + "step": 6064 + }, + { + "epoch": 2.54, + "learning_rate": 1.5970729632809044e-05, + "loss": 0.2282, + "step": 6065 + }, + { + "epoch": 2.54, + "learning_rate": 1.596936951215362e-05, + "loss": 0.2426, + "step": 6066 + }, + { + "epoch": 2.54, + "learning_rate": 1.5968009219913966e-05, + "loss": 0.2719, + "step": 6067 + }, + { + "epoch": 2.54, + "learning_rate": 1.5966648756129187e-05, + "loss": 0.2581, + "step": 6068 + }, + { + "epoch": 2.54, + "learning_rate": 1.5965288120838383e-05, + "loss": 0.2915, + "step": 6069 + }, + { + "epoch": 2.54, + "learning_rate": 1.596392731408066e-05, + "loss": 0.2598, + "step": 6070 + }, + { + "epoch": 2.54, + "learning_rate": 1.5962566335895146e-05, + "loss": 0.2513, + "step": 6071 + }, + { + "epoch": 2.54, + "learning_rate": 1.5961205186320952e-05, + "loss": 0.2531, + "step": 6072 + }, + { + "epoch": 2.54, + "learning_rate": 1.59598438653972e-05, + "loss": 0.2473, + "step": 6073 + }, + { + "epoch": 2.54, + "learning_rate": 1.595848237316303e-05, + "loss": 0.2286, + "step": 6074 + }, + { + "epoch": 2.54, + "learning_rate": 1.595712070965757e-05, + "loss": 0.2766, + "step": 6075 + }, + { + "epoch": 2.54, + "learning_rate": 1.595575887491996e-05, + "loss": 0.2845, + "step": 6076 + }, + { + "epoch": 2.54, + "learning_rate": 1.595439686898935e-05, + "loss": 0.2496, + "step": 6077 + }, + { + "epoch": 2.54, + "learning_rate": 1.595303469190488e-05, + "loss": 0.2742, + "step": 6078 + }, + { + "epoch": 2.55, + "learning_rate": 1.5951672343705714e-05, + "loss": 0.2465, + "step": 6079 + }, + { + "epoch": 2.55, + "learning_rate": 1.5950309824431008e-05, + "loss": 0.2647, + "step": 6080 + }, + { + "epoch": 2.55, + "learning_rate": 1.5948947134119926e-05, + "loss": 0.2563, + "step": 6081 + }, + { + "epoch": 2.55, + "learning_rate": 1.594758427281164e-05, + "loss": 0.2532, + "step": 6082 + }, + { + "epoch": 2.55, + "learning_rate": 1.594622124054532e-05, + "loss": 0.2513, + "step": 6083 + }, + { + "epoch": 2.55, + "learning_rate": 1.5944858037360145e-05, + "loss": 0.2514, + "step": 6084 + }, + { + "epoch": 2.55, + "learning_rate": 1.5943494663295303e-05, + "loss": 0.2461, + "step": 6085 + }, + { + "epoch": 2.55, + "learning_rate": 1.5942131118389984e-05, + "loss": 0.2538, + "step": 6086 + }, + { + "epoch": 2.55, + "learning_rate": 1.594076740268337e-05, + "loss": 0.2439, + "step": 6087 + }, + { + "epoch": 2.55, + "learning_rate": 1.593940351621468e-05, + "loss": 0.2267, + "step": 6088 + }, + { + "epoch": 2.55, + "learning_rate": 1.5938039459023098e-05, + "loss": 0.2541, + "step": 6089 + }, + { + "epoch": 2.55, + "learning_rate": 1.5936675231147845e-05, + "loss": 0.2191, + "step": 6090 + }, + { + "epoch": 2.55, + "learning_rate": 1.593531083262813e-05, + "loss": 0.268, + "step": 6091 + }, + { + "epoch": 2.55, + "learning_rate": 1.593394626350317e-05, + "loss": 0.2505, + "step": 6092 + }, + { + "epoch": 2.55, + "learning_rate": 1.5932581523812193e-05, + "loss": 0.2636, + "step": 6093 + }, + { + "epoch": 2.55, + "learning_rate": 1.5931216613594426e-05, + "loss": 0.2326, + "step": 6094 + }, + { + "epoch": 2.55, + "learning_rate": 1.5929851532889095e-05, + "loss": 0.2709, + "step": 6095 + }, + { + "epoch": 2.55, + "learning_rate": 1.592848628173545e-05, + "loss": 0.2567, + "step": 6096 + }, + { + "epoch": 2.55, + "learning_rate": 1.5927120860172722e-05, + "loss": 0.2475, + "step": 6097 + }, + { + "epoch": 2.55, + "learning_rate": 1.592575526824017e-05, + "loss": 0.2789, + "step": 6098 + }, + { + "epoch": 2.55, + "learning_rate": 1.5924389505977038e-05, + "loss": 0.2341, + "step": 6099 + }, + { + "epoch": 2.55, + "learning_rate": 1.5923023573422585e-05, + "loss": 0.2605, + "step": 6100 + }, + { + "epoch": 2.55, + "learning_rate": 1.592165747061608e-05, + "loss": 0.2788, + "step": 6101 + }, + { + "epoch": 2.55, + "learning_rate": 1.5920291197596784e-05, + "loss": 0.2772, + "step": 6102 + }, + { + "epoch": 2.56, + "learning_rate": 1.591892475440397e-05, + "loss": 0.232, + "step": 6103 + }, + { + "epoch": 2.56, + "learning_rate": 1.5917558141076915e-05, + "loss": 0.2478, + "step": 6104 + }, + { + "epoch": 2.56, + "learning_rate": 1.5916191357654906e-05, + "loss": 0.2352, + "step": 6105 + }, + { + "epoch": 2.56, + "learning_rate": 1.5914824404177222e-05, + "loss": 0.2803, + "step": 6106 + }, + { + "epoch": 2.56, + "learning_rate": 1.5913457280683163e-05, + "loss": 0.2376, + "step": 6107 + }, + { + "epoch": 2.56, + "learning_rate": 1.591208998721202e-05, + "loss": 0.264, + "step": 6108 + }, + { + "epoch": 2.56, + "learning_rate": 1.5910722523803098e-05, + "loss": 0.2767, + "step": 6109 + }, + { + "epoch": 2.56, + "learning_rate": 1.5909354890495703e-05, + "loss": 0.2921, + "step": 6110 + }, + { + "epoch": 2.56, + "learning_rate": 1.5907987087329143e-05, + "loss": 0.2388, + "step": 6111 + }, + { + "epoch": 2.56, + "learning_rate": 1.590661911434274e-05, + "loss": 0.2408, + "step": 6112 + }, + { + "epoch": 2.56, + "learning_rate": 1.5905250971575808e-05, + "loss": 0.2688, + "step": 6113 + }, + { + "epoch": 2.56, + "learning_rate": 1.590388265906768e-05, + "loss": 0.2435, + "step": 6114 + }, + { + "epoch": 2.56, + "learning_rate": 1.590251417685768e-05, + "loss": 0.2595, + "step": 6115 + }, + { + "epoch": 2.56, + "learning_rate": 1.5901145524985155e-05, + "loss": 0.26, + "step": 6116 + }, + { + "epoch": 2.56, + "learning_rate": 1.5899776703489432e-05, + "loss": 0.304, + "step": 6117 + }, + { + "epoch": 2.56, + "learning_rate": 1.5898407712409863e-05, + "loss": 0.2597, + "step": 6118 + }, + { + "epoch": 2.56, + "learning_rate": 1.58970385517858e-05, + "loss": 0.2655, + "step": 6119 + }, + { + "epoch": 2.56, + "learning_rate": 1.58956692216566e-05, + "loss": 0.2699, + "step": 6120 + }, + { + "epoch": 2.56, + "learning_rate": 1.5894299722061617e-05, + "loss": 0.273, + "step": 6121 + }, + { + "epoch": 2.56, + "learning_rate": 1.5892930053040218e-05, + "loss": 0.2488, + "step": 6122 + }, + { + "epoch": 2.56, + "learning_rate": 1.5891560214631775e-05, + "loss": 0.2928, + "step": 6123 + }, + { + "epoch": 2.56, + "learning_rate": 1.5890190206875663e-05, + "loss": 0.2671, + "step": 6124 + }, + { + "epoch": 2.56, + "learning_rate": 1.588882002981126e-05, + "loss": 0.2488, + "step": 6125 + }, + { + "epoch": 2.56, + "learning_rate": 1.588744968347795e-05, + "loss": 0.2903, + "step": 6126 + }, + { + "epoch": 2.57, + "learning_rate": 1.5886079167915123e-05, + "loss": 0.2526, + "step": 6127 + }, + { + "epoch": 2.57, + "learning_rate": 1.5884708483162175e-05, + "loss": 0.2429, + "step": 6128 + }, + { + "epoch": 2.57, + "learning_rate": 1.5883337629258502e-05, + "loss": 0.2909, + "step": 6129 + }, + { + "epoch": 2.57, + "learning_rate": 1.5881966606243512e-05, + "loss": 0.2554, + "step": 6130 + }, + { + "epoch": 2.57, + "learning_rate": 1.588059541415661e-05, + "loss": 0.2652, + "step": 6131 + }, + { + "epoch": 2.57, + "learning_rate": 1.5879224053037213e-05, + "loss": 0.2657, + "step": 6132 + }, + { + "epoch": 2.57, + "learning_rate": 1.5877852522924733e-05, + "loss": 0.2627, + "step": 6133 + }, + { + "epoch": 2.57, + "learning_rate": 1.58764808238586e-05, + "loss": 0.2544, + "step": 6134 + }, + { + "epoch": 2.57, + "learning_rate": 1.587510895587824e-05, + "loss": 0.2426, + "step": 6135 + }, + { + "epoch": 2.57, + "learning_rate": 1.587373691902309e-05, + "loss": 0.2401, + "step": 6136 + }, + { + "epoch": 2.57, + "learning_rate": 1.5872364713332583e-05, + "loss": 0.2544, + "step": 6137 + }, + { + "epoch": 2.57, + "learning_rate": 1.587099233884616e-05, + "loss": 0.2522, + "step": 6138 + }, + { + "epoch": 2.57, + "learning_rate": 1.586961979560328e-05, + "loss": 0.2804, + "step": 6139 + }, + { + "epoch": 2.57, + "learning_rate": 1.586824708364338e-05, + "loss": 0.2559, + "step": 6140 + }, + { + "epoch": 2.57, + "learning_rate": 1.5866874203005928e-05, + "loss": 0.2641, + "step": 6141 + }, + { + "epoch": 2.57, + "learning_rate": 1.5865501153730383e-05, + "loss": 0.2238, + "step": 6142 + }, + { + "epoch": 2.57, + "learning_rate": 1.586412793585621e-05, + "loss": 0.2681, + "step": 6143 + }, + { + "epoch": 2.57, + "learning_rate": 1.5862754549422884e-05, + "loss": 0.272, + "step": 6144 + }, + { + "epoch": 2.57, + "learning_rate": 1.5861380994469883e-05, + "loss": 0.2693, + "step": 6145 + }, + { + "epoch": 2.57, + "learning_rate": 1.5860007271036682e-05, + "loss": 0.2674, + "step": 6146 + }, + { + "epoch": 2.57, + "learning_rate": 1.585863337916278e-05, + "loss": 0.2741, + "step": 6147 + }, + { + "epoch": 2.57, + "learning_rate": 1.5857259318887653e-05, + "loss": 0.2629, + "step": 6148 + }, + { + "epoch": 2.57, + "learning_rate": 1.585588509025081e-05, + "loss": 0.202, + "step": 6149 + }, + { + "epoch": 2.57, + "learning_rate": 1.5854510693291744e-05, + "loss": 0.26, + "step": 6150 + }, + { + "epoch": 2.58, + "learning_rate": 1.5853136128049963e-05, + "loss": 0.2583, + "step": 6151 + }, + { + "epoch": 2.58, + "learning_rate": 1.585176139456498e-05, + "loss": 0.243, + "step": 6152 + }, + { + "epoch": 2.58, + "learning_rate": 1.5850386492876307e-05, + "loss": 0.2652, + "step": 6153 + }, + { + "epoch": 2.58, + "learning_rate": 1.5849011423023463e-05, + "loss": 0.2457, + "step": 6154 + }, + { + "epoch": 2.58, + "learning_rate": 1.584763618504598e-05, + "loss": 0.2511, + "step": 6155 + }, + { + "epoch": 2.58, + "learning_rate": 1.5846260778983382e-05, + "loss": 0.2497, + "step": 6156 + }, + { + "epoch": 2.58, + "learning_rate": 1.5844885204875207e-05, + "loss": 0.2725, + "step": 6157 + }, + { + "epoch": 2.58, + "learning_rate": 1.584350946276099e-05, + "loss": 0.292, + "step": 6158 + }, + { + "epoch": 2.58, + "learning_rate": 1.5842133552680286e-05, + "loss": 0.2717, + "step": 6159 + }, + { + "epoch": 2.58, + "learning_rate": 1.5840757474672634e-05, + "loss": 0.2678, + "step": 6160 + }, + { + "epoch": 2.58, + "learning_rate": 1.5839381228777586e-05, + "loss": 0.2544, + "step": 6161 + }, + { + "epoch": 2.58, + "learning_rate": 1.583800481503471e-05, + "loss": 0.2599, + "step": 6162 + }, + { + "epoch": 2.58, + "learning_rate": 1.5836628233483568e-05, + "loss": 0.2436, + "step": 6163 + }, + { + "epoch": 2.58, + "learning_rate": 1.5835251484163728e-05, + "loss": 0.2356, + "step": 6164 + }, + { + "epoch": 2.58, + "learning_rate": 1.583387456711476e-05, + "loss": 0.2818, + "step": 6165 + }, + { + "epoch": 2.58, + "learning_rate": 1.5832497482376244e-05, + "loss": 0.2922, + "step": 6166 + }, + { + "epoch": 2.58, + "learning_rate": 1.583112022998776e-05, + "loss": 0.2658, + "step": 6167 + }, + { + "epoch": 2.58, + "learning_rate": 1.5829742809988907e-05, + "loss": 0.2644, + "step": 6168 + }, + { + "epoch": 2.58, + "learning_rate": 1.5828365222419268e-05, + "loss": 0.2485, + "step": 6169 + }, + { + "epoch": 2.58, + "learning_rate": 1.582698746731844e-05, + "loss": 0.2583, + "step": 6170 + }, + { + "epoch": 2.58, + "learning_rate": 1.582560954472603e-05, + "loss": 0.2407, + "step": 6171 + }, + { + "epoch": 2.58, + "learning_rate": 1.582423145468164e-05, + "loss": 0.2561, + "step": 6172 + }, + { + "epoch": 2.58, + "learning_rate": 1.5822853197224888e-05, + "loss": 0.2346, + "step": 6173 + }, + { + "epoch": 2.58, + "learning_rate": 1.5821474772395385e-05, + "loss": 0.2575, + "step": 6174 + }, + { + "epoch": 2.59, + "learning_rate": 1.5820096180232762e-05, + "loss": 0.2469, + "step": 6175 + }, + { + "epoch": 2.59, + "learning_rate": 1.5818717420776635e-05, + "loss": 0.2751, + "step": 6176 + }, + { + "epoch": 2.59, + "learning_rate": 1.5817338494066636e-05, + "loss": 0.251, + "step": 6177 + }, + { + "epoch": 2.59, + "learning_rate": 1.581595940014241e-05, + "loss": 0.2643, + "step": 6178 + }, + { + "epoch": 2.59, + "learning_rate": 1.5814580139043587e-05, + "loss": 0.2536, + "step": 6179 + }, + { + "epoch": 2.59, + "learning_rate": 1.581320071080982e-05, + "loss": 0.2741, + "step": 6180 + }, + { + "epoch": 2.59, + "learning_rate": 1.581182111548076e-05, + "loss": 0.2593, + "step": 6181 + }, + { + "epoch": 2.59, + "learning_rate": 1.5810441353096054e-05, + "loss": 0.2871, + "step": 6182 + }, + { + "epoch": 2.59, + "learning_rate": 1.5809061423695368e-05, + "loss": 0.2528, + "step": 6183 + }, + { + "epoch": 2.59, + "learning_rate": 1.5807681327318372e-05, + "loss": 0.2764, + "step": 6184 + }, + { + "epoch": 2.59, + "learning_rate": 1.5806301064004724e-05, + "loss": 0.2689, + "step": 6185 + }, + { + "epoch": 2.59, + "learning_rate": 1.5804920633794107e-05, + "loss": 0.2624, + "step": 6186 + }, + { + "epoch": 2.59, + "learning_rate": 1.5803540036726198e-05, + "loss": 0.2553, + "step": 6187 + }, + { + "epoch": 2.59, + "learning_rate": 1.5802159272840677e-05, + "loss": 0.24, + "step": 6188 + }, + { + "epoch": 2.59, + "learning_rate": 1.580077834217724e-05, + "loss": 0.2369, + "step": 6189 + }, + { + "epoch": 2.59, + "learning_rate": 1.5799397244775576e-05, + "loss": 0.2515, + "step": 6190 + }, + { + "epoch": 2.59, + "learning_rate": 1.5798015980675384e-05, + "loss": 0.2577, + "step": 6191 + }, + { + "epoch": 2.59, + "learning_rate": 1.579663454991637e-05, + "loss": 0.2721, + "step": 6192 + }, + { + "epoch": 2.59, + "learning_rate": 1.579525295253824e-05, + "loss": 0.2448, + "step": 6193 + }, + { + "epoch": 2.59, + "learning_rate": 1.5793871188580704e-05, + "loss": 0.2729, + "step": 6194 + }, + { + "epoch": 2.59, + "learning_rate": 1.579248925808348e-05, + "loss": 0.2618, + "step": 6195 + }, + { + "epoch": 2.59, + "learning_rate": 1.5791107161086294e-05, + "loss": 0.2722, + "step": 6196 + }, + { + "epoch": 2.59, + "learning_rate": 1.5789724897628873e-05, + "loss": 0.261, + "step": 6197 + }, + { + "epoch": 2.59, + "learning_rate": 1.578834246775095e-05, + "loss": 0.2592, + "step": 6198 + }, + { + "epoch": 2.6, + "learning_rate": 1.5786959871492254e-05, + "loss": 0.2786, + "step": 6199 + }, + { + "epoch": 2.6, + "learning_rate": 1.5785577108892537e-05, + "loss": 0.2406, + "step": 6200 + }, + { + "epoch": 2.6, + "learning_rate": 1.5784194179991534e-05, + "loss": 0.2347, + "step": 6201 + }, + { + "epoch": 2.6, + "learning_rate": 1.5782811084829005e-05, + "loss": 0.2531, + "step": 6202 + }, + { + "epoch": 2.6, + "learning_rate": 1.5781427823444704e-05, + "loss": 0.2519, + "step": 6203 + }, + { + "epoch": 2.6, + "learning_rate": 1.578004439587839e-05, + "loss": 0.2603, + "step": 6204 + }, + { + "epoch": 2.6, + "learning_rate": 1.5778660802169826e-05, + "loss": 0.2648, + "step": 6205 + }, + { + "epoch": 2.6, + "learning_rate": 1.577727704235879e-05, + "loss": 0.2316, + "step": 6206 + }, + { + "epoch": 2.6, + "learning_rate": 1.5775893116485052e-05, + "loss": 0.2395, + "step": 6207 + }, + { + "epoch": 2.6, + "learning_rate": 1.5774509024588388e-05, + "loss": 0.2766, + "step": 6208 + }, + { + "epoch": 2.6, + "learning_rate": 1.5773124766708588e-05, + "loss": 0.2556, + "step": 6209 + }, + { + "epoch": 2.6, + "learning_rate": 1.577174034288544e-05, + "loss": 0.2498, + "step": 6210 + }, + { + "epoch": 2.6, + "learning_rate": 1.5770355753158736e-05, + "loss": 0.2674, + "step": 6211 + }, + { + "epoch": 2.6, + "learning_rate": 1.5768970997568282e-05, + "loss": 0.2553, + "step": 6212 + }, + { + "epoch": 2.6, + "learning_rate": 1.576758607615387e-05, + "loss": 0.2363, + "step": 6213 + }, + { + "epoch": 2.6, + "learning_rate": 1.5766200988955315e-05, + "loss": 0.2762, + "step": 6214 + }, + { + "epoch": 2.6, + "learning_rate": 1.576481573601243e-05, + "loss": 0.2845, + "step": 6215 + }, + { + "epoch": 2.6, + "learning_rate": 1.5763430317365032e-05, + "loss": 0.2653, + "step": 6216 + }, + { + "epoch": 2.6, + "learning_rate": 1.5762044733052944e-05, + "loss": 0.2545, + "step": 6217 + }, + { + "epoch": 2.6, + "learning_rate": 1.576065898311599e-05, + "loss": 0.2495, + "step": 6218 + }, + { + "epoch": 2.6, + "learning_rate": 1.575927306759401e-05, + "loss": 0.2674, + "step": 6219 + }, + { + "epoch": 2.6, + "learning_rate": 1.575788698652684e-05, + "loss": 0.2691, + "step": 6220 + }, + { + "epoch": 2.6, + "learning_rate": 1.5756500739954312e-05, + "loss": 0.2548, + "step": 6221 + }, + { + "epoch": 2.6, + "learning_rate": 1.5755114327916282e-05, + "loss": 0.2469, + "step": 6222 + }, + { + "epoch": 2.61, + "learning_rate": 1.5753727750452596e-05, + "loss": 0.2675, + "step": 6223 + }, + { + "epoch": 2.61, + "learning_rate": 1.575234100760311e-05, + "loss": 0.2719, + "step": 6224 + }, + { + "epoch": 2.61, + "learning_rate": 1.5750954099407688e-05, + "loss": 0.2764, + "step": 6225 + }, + { + "epoch": 2.61, + "learning_rate": 1.5749567025906193e-05, + "loss": 0.2575, + "step": 6226 + }, + { + "epoch": 2.61, + "learning_rate": 1.5748179787138494e-05, + "loss": 0.2601, + "step": 6227 + }, + { + "epoch": 2.61, + "learning_rate": 1.5746792383144472e-05, + "loss": 0.2749, + "step": 6228 + }, + { + "epoch": 2.61, + "learning_rate": 1.5745404813964e-05, + "loss": 0.2885, + "step": 6229 + }, + { + "epoch": 2.61, + "learning_rate": 1.5744017079636967e-05, + "loss": 0.2574, + "step": 6230 + }, + { + "epoch": 2.61, + "learning_rate": 1.5742629180203258e-05, + "loss": 0.2448, + "step": 6231 + }, + { + "epoch": 2.61, + "learning_rate": 1.574124111570277e-05, + "loss": 0.2523, + "step": 6232 + }, + { + "epoch": 2.61, + "learning_rate": 1.57398528861754e-05, + "loss": 0.277, + "step": 6233 + }, + { + "epoch": 2.61, + "learning_rate": 1.5738464491661054e-05, + "loss": 0.2612, + "step": 6234 + }, + { + "epoch": 2.61, + "learning_rate": 1.5737075932199637e-05, + "loss": 0.2514, + "step": 6235 + }, + { + "epoch": 2.61, + "learning_rate": 1.5735687207831064e-05, + "loss": 0.2762, + "step": 6236 + }, + { + "epoch": 2.61, + "learning_rate": 1.5734298318595252e-05, + "loss": 0.266, + "step": 6237 + }, + { + "epoch": 2.61, + "learning_rate": 1.573290926453212e-05, + "loss": 0.2705, + "step": 6238 + }, + { + "epoch": 2.61, + "learning_rate": 1.5731520045681606e-05, + "loss": 0.2766, + "step": 6239 + }, + { + "epoch": 2.61, + "learning_rate": 1.5730130662083628e-05, + "loss": 0.2422, + "step": 6240 + }, + { + "epoch": 2.61, + "learning_rate": 1.572874111377813e-05, + "loss": 0.2674, + "step": 6241 + }, + { + "epoch": 2.61, + "learning_rate": 1.5727351400805054e-05, + "loss": 0.268, + "step": 6242 + }, + { + "epoch": 2.61, + "learning_rate": 1.5725961523204345e-05, + "loss": 0.2462, + "step": 6243 + }, + { + "epoch": 2.61, + "learning_rate": 1.5724571481015953e-05, + "loss": 0.2585, + "step": 6244 + }, + { + "epoch": 2.61, + "learning_rate": 1.572318127427983e-05, + "loss": 0.308, + "step": 6245 + }, + { + "epoch": 2.62, + "learning_rate": 1.5721790903035942e-05, + "loss": 0.2525, + "step": 6246 + }, + { + "epoch": 2.62, + "learning_rate": 1.572040036732425e-05, + "loss": 0.2731, + "step": 6247 + }, + { + "epoch": 2.62, + "learning_rate": 1.571900966718473e-05, + "loss": 0.246, + "step": 6248 + }, + { + "epoch": 2.62, + "learning_rate": 1.571761880265735e-05, + "loss": 0.2731, + "step": 6249 + }, + { + "epoch": 2.62, + "learning_rate": 1.5716227773782088e-05, + "loss": 0.2903, + "step": 6250 + }, + { + "epoch": 2.62, + "learning_rate": 1.5714836580598936e-05, + "loss": 0.2622, + "step": 6251 + }, + { + "epoch": 2.62, + "learning_rate": 1.5713445223147876e-05, + "loss": 0.2476, + "step": 6252 + }, + { + "epoch": 2.62, + "learning_rate": 1.5712053701468902e-05, + "loss": 0.259, + "step": 6253 + }, + { + "epoch": 2.62, + "learning_rate": 1.5710662015602016e-05, + "loss": 0.2912, + "step": 6254 + }, + { + "epoch": 2.62, + "learning_rate": 1.5709270165587216e-05, + "loss": 0.2395, + "step": 6255 + }, + { + "epoch": 2.62, + "learning_rate": 1.570787815146451e-05, + "loss": 0.2619, + "step": 6256 + }, + { + "epoch": 2.62, + "learning_rate": 1.5706485973273916e-05, + "loss": 0.2478, + "step": 6257 + }, + { + "epoch": 2.62, + "learning_rate": 1.570509363105544e-05, + "loss": 0.2812, + "step": 6258 + }, + { + "epoch": 2.62, + "learning_rate": 1.5703701124849115e-05, + "loss": 0.2282, + "step": 6259 + }, + { + "epoch": 2.62, + "learning_rate": 1.5702308454694963e-05, + "loss": 0.2505, + "step": 6260 + }, + { + "epoch": 2.62, + "learning_rate": 1.570091562063301e-05, + "loss": 0.2522, + "step": 6261 + }, + { + "epoch": 2.62, + "learning_rate": 1.5699522622703303e-05, + "loss": 0.2635, + "step": 6262 + }, + { + "epoch": 2.62, + "learning_rate": 1.569812946094587e-05, + "loss": 0.2555, + "step": 6263 + }, + { + "epoch": 2.62, + "learning_rate": 1.5696736135400765e-05, + "loss": 0.2671, + "step": 6264 + }, + { + "epoch": 2.62, + "learning_rate": 1.569534264610803e-05, + "loss": 0.2352, + "step": 6265 + }, + { + "epoch": 2.62, + "learning_rate": 1.5693948993107733e-05, + "loss": 0.2657, + "step": 6266 + }, + { + "epoch": 2.62, + "learning_rate": 1.569255517643992e-05, + "loss": 0.2495, + "step": 6267 + }, + { + "epoch": 2.62, + "learning_rate": 1.569116119614466e-05, + "loss": 0.2657, + "step": 6268 + }, + { + "epoch": 2.62, + "learning_rate": 1.5689767052262028e-05, + "loss": 0.2685, + "step": 6269 + }, + { + "epoch": 2.63, + "learning_rate": 1.568837274483208e-05, + "loss": 0.2572, + "step": 6270 + }, + { + "epoch": 2.63, + "learning_rate": 1.5686978273894914e-05, + "loss": 0.2648, + "step": 6271 + }, + { + "epoch": 2.63, + "learning_rate": 1.5685583639490598e-05, + "loss": 0.2516, + "step": 6272 + }, + { + "epoch": 2.63, + "learning_rate": 1.568418884165923e-05, + "loss": 0.2292, + "step": 6273 + }, + { + "epoch": 2.63, + "learning_rate": 1.5682793880440895e-05, + "loss": 0.2718, + "step": 6274 + }, + { + "epoch": 2.63, + "learning_rate": 1.5681398755875695e-05, + "loss": 0.2613, + "step": 6275 + }, + { + "epoch": 2.63, + "learning_rate": 1.5680003468003725e-05, + "loss": 0.2443, + "step": 6276 + }, + { + "epoch": 2.63, + "learning_rate": 1.56786080168651e-05, + "loss": 0.2538, + "step": 6277 + }, + { + "epoch": 2.63, + "learning_rate": 1.567721240249993e-05, + "loss": 0.2497, + "step": 6278 + }, + { + "epoch": 2.63, + "learning_rate": 1.5675816624948322e-05, + "loss": 0.2681, + "step": 6279 + }, + { + "epoch": 2.63, + "learning_rate": 1.56744206842504e-05, + "loss": 0.2525, + "step": 6280 + }, + { + "epoch": 2.63, + "learning_rate": 1.5673024580446294e-05, + "loss": 0.2496, + "step": 6281 + }, + { + "epoch": 2.63, + "learning_rate": 1.567162831357613e-05, + "loss": 0.2768, + "step": 6282 + }, + { + "epoch": 2.63, + "learning_rate": 1.5670231883680044e-05, + "loss": 0.2721, + "step": 6283 + }, + { + "epoch": 2.63, + "learning_rate": 1.566883529079817e-05, + "loss": 0.274, + "step": 6284 + }, + { + "epoch": 2.63, + "learning_rate": 1.5667438534970664e-05, + "loss": 0.283, + "step": 6285 + }, + { + "epoch": 2.63, + "learning_rate": 1.566604161623766e-05, + "loss": 0.284, + "step": 6286 + }, + { + "epoch": 2.63, + "learning_rate": 1.566464453463932e-05, + "loss": 0.2795, + "step": 6287 + }, + { + "epoch": 2.63, + "learning_rate": 1.5663247290215797e-05, + "loss": 0.2495, + "step": 6288 + }, + { + "epoch": 2.63, + "learning_rate": 1.566184988300726e-05, + "loss": 0.261, + "step": 6289 + }, + { + "epoch": 2.63, + "learning_rate": 1.5660452313053873e-05, + "loss": 0.2663, + "step": 6290 + }, + { + "epoch": 2.63, + "learning_rate": 1.56590545803958e-05, + "loss": 0.2632, + "step": 6291 + }, + { + "epoch": 2.63, + "learning_rate": 1.565765668507323e-05, + "loss": 0.2589, + "step": 6292 + }, + { + "epoch": 2.63, + "learning_rate": 1.565625862712634e-05, + "loss": 0.26, + "step": 6293 + }, + { + "epoch": 2.64, + "learning_rate": 1.5654860406595312e-05, + "loss": 0.2677, + "step": 6294 + }, + { + "epoch": 2.64, + "learning_rate": 1.5653462023520344e-05, + "loss": 0.2423, + "step": 6295 + }, + { + "epoch": 2.64, + "learning_rate": 1.5652063477941625e-05, + "loss": 0.2646, + "step": 6296 + }, + { + "epoch": 2.64, + "learning_rate": 1.5650664769899356e-05, + "loss": 0.2359, + "step": 6297 + }, + { + "epoch": 2.64, + "learning_rate": 1.564926589943374e-05, + "loss": 0.2378, + "step": 6298 + }, + { + "epoch": 2.64, + "learning_rate": 1.5647866866584992e-05, + "loss": 0.2465, + "step": 6299 + }, + { + "epoch": 2.64, + "learning_rate": 1.5646467671393323e-05, + "loss": 0.2394, + "step": 6300 + }, + { + "epoch": 2.64, + "learning_rate": 1.564506831389895e-05, + "loss": 0.2737, + "step": 6301 + }, + { + "epoch": 2.64, + "learning_rate": 1.5643668794142102e-05, + "loss": 0.2833, + "step": 6302 + }, + { + "epoch": 2.64, + "learning_rate": 1.5642269112162998e-05, + "loss": 0.2895, + "step": 6303 + }, + { + "epoch": 2.64, + "learning_rate": 1.5640869268001875e-05, + "loss": 0.2499, + "step": 6304 + }, + { + "epoch": 2.64, + "learning_rate": 1.563946926169897e-05, + "loss": 0.2881, + "step": 6305 + }, + { + "epoch": 2.64, + "learning_rate": 1.5638069093294527e-05, + "loss": 0.271, + "step": 6306 + }, + { + "epoch": 2.64, + "learning_rate": 1.5636668762828793e-05, + "loss": 0.2463, + "step": 6307 + }, + { + "epoch": 2.64, + "learning_rate": 1.5635268270342015e-05, + "loss": 0.2836, + "step": 6308 + }, + { + "epoch": 2.64, + "learning_rate": 1.5633867615874453e-05, + "loss": 0.2646, + "step": 6309 + }, + { + "epoch": 2.64, + "learning_rate": 1.5632466799466362e-05, + "loss": 0.2348, + "step": 6310 + }, + { + "epoch": 2.64, + "learning_rate": 1.5631065821158016e-05, + "loss": 0.2564, + "step": 6311 + }, + { + "epoch": 2.64, + "learning_rate": 1.5629664680989677e-05, + "loss": 0.2469, + "step": 6312 + }, + { + "epoch": 2.64, + "learning_rate": 1.562826337900162e-05, + "loss": 0.2531, + "step": 6313 + }, + { + "epoch": 2.64, + "learning_rate": 1.562686191523413e-05, + "loss": 0.2948, + "step": 6314 + }, + { + "epoch": 2.64, + "learning_rate": 1.562546028972749e-05, + "loss": 0.2857, + "step": 6315 + }, + { + "epoch": 2.64, + "learning_rate": 1.562405850252198e-05, + "loss": 0.2549, + "step": 6316 + }, + { + "epoch": 2.64, + "learning_rate": 1.5622656553657904e-05, + "loss": 0.2772, + "step": 6317 + }, + { + "epoch": 2.65, + "learning_rate": 1.5621254443175553e-05, + "loss": 0.28, + "step": 6318 + }, + { + "epoch": 2.65, + "learning_rate": 1.561985217111523e-05, + "loss": 0.2754, + "step": 6319 + }, + { + "epoch": 2.65, + "learning_rate": 1.5618449737517242e-05, + "loss": 0.2899, + "step": 6320 + }, + { + "epoch": 2.65, + "learning_rate": 1.5617047142421904e-05, + "loss": 0.2448, + "step": 6321 + }, + { + "epoch": 2.65, + "learning_rate": 1.5615644385869532e-05, + "loss": 0.2475, + "step": 6322 + }, + { + "epoch": 2.65, + "learning_rate": 1.5614241467900444e-05, + "loss": 0.276, + "step": 6323 + }, + { + "epoch": 2.65, + "learning_rate": 1.5612838388554966e-05, + "loss": 0.2666, + "step": 6324 + }, + { + "epoch": 2.65, + "learning_rate": 1.561143514787343e-05, + "loss": 0.2741, + "step": 6325 + }, + { + "epoch": 2.65, + "learning_rate": 1.561003174589617e-05, + "loss": 0.2765, + "step": 6326 + }, + { + "epoch": 2.65, + "learning_rate": 1.5608628182663527e-05, + "loss": 0.2711, + "step": 6327 + }, + { + "epoch": 2.65, + "learning_rate": 1.5607224458215843e-05, + "loss": 0.2404, + "step": 6328 + }, + { + "epoch": 2.65, + "learning_rate": 1.560582057259347e-05, + "loss": 0.2638, + "step": 6329 + }, + { + "epoch": 2.65, + "learning_rate": 1.5604416525836756e-05, + "loss": 0.2678, + "step": 6330 + }, + { + "epoch": 2.65, + "learning_rate": 1.560301231798606e-05, + "loss": 0.2781, + "step": 6331 + }, + { + "epoch": 2.65, + "learning_rate": 1.560160794908175e-05, + "loss": 0.2909, + "step": 6332 + }, + { + "epoch": 2.65, + "learning_rate": 1.560020341916419e-05, + "loss": 0.2679, + "step": 6333 + }, + { + "epoch": 2.65, + "learning_rate": 1.5598798728273753e-05, + "loss": 0.251, + "step": 6334 + }, + { + "epoch": 2.65, + "learning_rate": 1.5597393876450813e-05, + "loss": 0.2647, + "step": 6335 + }, + { + "epoch": 2.65, + "learning_rate": 1.5595988863735758e-05, + "loss": 0.2343, + "step": 6336 + }, + { + "epoch": 2.65, + "learning_rate": 1.5594583690168966e-05, + "loss": 0.2831, + "step": 6337 + }, + { + "epoch": 2.65, + "learning_rate": 1.559317835579083e-05, + "loss": 0.2257, + "step": 6338 + }, + { + "epoch": 2.65, + "learning_rate": 1.5591772860641743e-05, + "loss": 0.2514, + "step": 6339 + }, + { + "epoch": 2.65, + "learning_rate": 1.5590367204762113e-05, + "loss": 0.2614, + "step": 6340 + }, + { + "epoch": 2.65, + "learning_rate": 1.5588961388192333e-05, + "loss": 0.2447, + "step": 6341 + }, + { + "epoch": 2.66, + "learning_rate": 1.558755541097282e-05, + "loss": 0.2514, + "step": 6342 + }, + { + "epoch": 2.66, + "learning_rate": 1.5586149273143982e-05, + "loss": 0.2586, + "step": 6343 + }, + { + "epoch": 2.66, + "learning_rate": 1.5584742974746242e-05, + "loss": 0.2454, + "step": 6344 + }, + { + "epoch": 2.66, + "learning_rate": 1.5583336515820024e-05, + "loss": 0.2618, + "step": 6345 + }, + { + "epoch": 2.66, + "learning_rate": 1.558192989640575e-05, + "loss": 0.2594, + "step": 6346 + }, + { + "epoch": 2.66, + "learning_rate": 1.5580523116543853e-05, + "loss": 0.2391, + "step": 6347 + }, + { + "epoch": 2.66, + "learning_rate": 1.557911617627477e-05, + "loss": 0.2645, + "step": 6348 + }, + { + "epoch": 2.66, + "learning_rate": 1.557770907563895e-05, + "loss": 0.2572, + "step": 6349 + }, + { + "epoch": 2.66, + "learning_rate": 1.5576301814676825e-05, + "loss": 0.2587, + "step": 6350 + }, + { + "epoch": 2.66, + "learning_rate": 1.5574894393428856e-05, + "loss": 0.2847, + "step": 6351 + }, + { + "epoch": 2.66, + "learning_rate": 1.5573486811935494e-05, + "loss": 0.2696, + "step": 6352 + }, + { + "epoch": 2.66, + "learning_rate": 1.55720790702372e-05, + "loss": 0.2752, + "step": 6353 + }, + { + "epoch": 2.66, + "learning_rate": 1.557067116837444e-05, + "loss": 0.233, + "step": 6354 + }, + { + "epoch": 2.66, + "learning_rate": 1.5569263106387676e-05, + "loss": 0.2646, + "step": 6355 + }, + { + "epoch": 2.66, + "learning_rate": 1.5567854884317387e-05, + "loss": 0.273, + "step": 6356 + }, + { + "epoch": 2.66, + "learning_rate": 1.5566446502204052e-05, + "loss": 0.2578, + "step": 6357 + }, + { + "epoch": 2.66, + "learning_rate": 1.556503796008815e-05, + "loss": 0.2444, + "step": 6358 + }, + { + "epoch": 2.66, + "learning_rate": 1.5563629258010173e-05, + "loss": 0.2624, + "step": 6359 + }, + { + "epoch": 2.66, + "learning_rate": 1.556222039601061e-05, + "loss": 0.2674, + "step": 6360 + }, + { + "epoch": 2.66, + "learning_rate": 1.5560811374129955e-05, + "loss": 0.2442, + "step": 6361 + }, + { + "epoch": 2.66, + "learning_rate": 1.5559402192408712e-05, + "loss": 0.283, + "step": 6362 + }, + { + "epoch": 2.66, + "learning_rate": 1.5557992850887388e-05, + "loss": 0.2245, + "step": 6363 + }, + { + "epoch": 2.66, + "learning_rate": 1.5556583349606492e-05, + "loss": 0.2622, + "step": 6364 + }, + { + "epoch": 2.66, + "learning_rate": 1.5555173688606537e-05, + "loss": 0.2813, + "step": 6365 + }, + { + "epoch": 2.67, + "learning_rate": 1.5553763867928043e-05, + "loss": 0.2728, + "step": 6366 + }, + { + "epoch": 2.67, + "learning_rate": 1.5552353887611536e-05, + "loss": 0.2805, + "step": 6367 + }, + { + "epoch": 2.67, + "learning_rate": 1.5550943747697546e-05, + "loss": 0.2557, + "step": 6368 + }, + { + "epoch": 2.67, + "learning_rate": 1.5549533448226598e-05, + "loss": 0.2733, + "step": 6369 + }, + { + "epoch": 2.67, + "learning_rate": 1.554812298923924e-05, + "loss": 0.2685, + "step": 6370 + }, + { + "epoch": 2.67, + "learning_rate": 1.5546712370776005e-05, + "loss": 0.2678, + "step": 6371 + }, + { + "epoch": 2.67, + "learning_rate": 1.554530159287745e-05, + "loss": 0.2202, + "step": 6372 + }, + { + "epoch": 2.67, + "learning_rate": 1.5543890655584122e-05, + "loss": 0.2725, + "step": 6373 + }, + { + "epoch": 2.67, + "learning_rate": 1.5542479558936572e-05, + "loss": 0.2627, + "step": 6374 + }, + { + "epoch": 2.67, + "learning_rate": 1.554106830297537e-05, + "loss": 0.2835, + "step": 6375 + }, + { + "epoch": 2.67, + "learning_rate": 1.5539656887741076e-05, + "loss": 0.2782, + "step": 6376 + }, + { + "epoch": 2.67, + "learning_rate": 1.5538245313274258e-05, + "loss": 0.278, + "step": 6377 + }, + { + "epoch": 2.67, + "learning_rate": 1.5536833579615495e-05, + "loss": 0.2575, + "step": 6378 + }, + { + "epoch": 2.67, + "learning_rate": 1.5535421686805367e-05, + "loss": 0.242, + "step": 6379 + }, + { + "epoch": 2.67, + "learning_rate": 1.553400963488445e-05, + "loss": 0.2558, + "step": 6380 + }, + { + "epoch": 2.67, + "learning_rate": 1.553259742389334e-05, + "loss": 0.2637, + "step": 6381 + }, + { + "epoch": 2.67, + "learning_rate": 1.553118505387262e-05, + "loss": 0.2492, + "step": 6382 + }, + { + "epoch": 2.67, + "learning_rate": 1.55297725248629e-05, + "loss": 0.2669, + "step": 6383 + }, + { + "epoch": 2.67, + "learning_rate": 1.5528359836904774e-05, + "loss": 0.2425, + "step": 6384 + }, + { + "epoch": 2.67, + "learning_rate": 1.552694699003885e-05, + "loss": 0.2539, + "step": 6385 + }, + { + "epoch": 2.67, + "learning_rate": 1.5525533984305742e-05, + "loss": 0.2474, + "step": 6386 + }, + { + "epoch": 2.67, + "learning_rate": 1.552412081974606e-05, + "loss": 0.254, + "step": 6387 + }, + { + "epoch": 2.67, + "learning_rate": 1.5522707496400425e-05, + "loss": 0.2662, + "step": 6388 + }, + { + "epoch": 2.67, + "learning_rate": 1.552129401430947e-05, + "loss": 0.2675, + "step": 6389 + }, + { + "epoch": 2.68, + "learning_rate": 1.5519880373513814e-05, + "loss": 0.2582, + "step": 6390 + }, + { + "epoch": 2.68, + "learning_rate": 1.5518466574054097e-05, + "loss": 0.2468, + "step": 6391 + }, + { + "epoch": 2.68, + "learning_rate": 1.5517052615970955e-05, + "loss": 0.2305, + "step": 6392 + }, + { + "epoch": 2.68, + "learning_rate": 1.551563849930503e-05, + "loss": 0.3024, + "step": 6393 + }, + { + "epoch": 2.68, + "learning_rate": 1.551422422409697e-05, + "loss": 0.2718, + "step": 6394 + }, + { + "epoch": 2.68, + "learning_rate": 1.551280979038743e-05, + "loss": 0.2705, + "step": 6395 + }, + { + "epoch": 2.68, + "learning_rate": 1.551139519821706e-05, + "loss": 0.2716, + "step": 6396 + }, + { + "epoch": 2.68, + "learning_rate": 1.5509980447626532e-05, + "loss": 0.2879, + "step": 6397 + }, + { + "epoch": 2.68, + "learning_rate": 1.5508565538656505e-05, + "loss": 0.2648, + "step": 6398 + }, + { + "epoch": 2.68, + "learning_rate": 1.5507150471347644e-05, + "loss": 0.2476, + "step": 6399 + }, + { + "epoch": 2.68, + "learning_rate": 1.5505735245740635e-05, + "loss": 0.2499, + "step": 6400 + }, + { + "epoch": 2.68, + "learning_rate": 1.5504319861876148e-05, + "loss": 0.2537, + "step": 6401 + }, + { + "epoch": 2.68, + "learning_rate": 1.5502904319794875e-05, + "loss": 0.2677, + "step": 6402 + }, + { + "epoch": 2.68, + "learning_rate": 1.55014886195375e-05, + "loss": 0.2583, + "step": 6403 + }, + { + "epoch": 2.68, + "learning_rate": 1.5500072761144712e-05, + "loss": 0.2513, + "step": 6404 + }, + { + "epoch": 2.68, + "learning_rate": 1.5498656744657222e-05, + "loss": 0.2343, + "step": 6405 + }, + { + "epoch": 2.68, + "learning_rate": 1.5497240570115716e-05, + "loss": 0.2468, + "step": 6406 + }, + { + "epoch": 2.68, + "learning_rate": 1.5495824237560907e-05, + "loss": 0.2654, + "step": 6407 + }, + { + "epoch": 2.68, + "learning_rate": 1.5494407747033513e-05, + "loss": 0.2677, + "step": 6408 + }, + { + "epoch": 2.68, + "learning_rate": 1.5492991098574243e-05, + "loss": 0.2488, + "step": 6409 + }, + { + "epoch": 2.68, + "learning_rate": 1.5491574292223814e-05, + "loss": 0.2519, + "step": 6410 + }, + { + "epoch": 2.68, + "learning_rate": 1.5490157328022957e-05, + "loss": 0.2988, + "step": 6411 + }, + { + "epoch": 2.68, + "learning_rate": 1.54887402060124e-05, + "loss": 0.281, + "step": 6412 + }, + { + "epoch": 2.68, + "learning_rate": 1.5487322926232876e-05, + "loss": 0.262, + "step": 6413 + }, + { + "epoch": 2.69, + "learning_rate": 1.5485905488725126e-05, + "loss": 0.2738, + "step": 6414 + }, + { + "epoch": 2.69, + "learning_rate": 1.548448789352989e-05, + "loss": 0.2776, + "step": 6415 + }, + { + "epoch": 2.69, + "learning_rate": 1.5483070140687916e-05, + "loss": 0.2465, + "step": 6416 + }, + { + "epoch": 2.69, + "learning_rate": 1.5481652230239954e-05, + "loss": 0.2877, + "step": 6417 + }, + { + "epoch": 2.69, + "learning_rate": 1.5480234162226763e-05, + "loss": 0.2721, + "step": 6418 + }, + { + "epoch": 2.69, + "learning_rate": 1.5478815936689107e-05, + "loss": 0.2494, + "step": 6419 + }, + { + "epoch": 2.69, + "learning_rate": 1.5477397553667745e-05, + "loss": 0.2774, + "step": 6420 + }, + { + "epoch": 2.69, + "learning_rate": 1.5475979013203456e-05, + "loss": 0.2249, + "step": 6421 + }, + { + "epoch": 2.69, + "learning_rate": 1.5474560315337007e-05, + "loss": 0.2835, + "step": 6422 + }, + { + "epoch": 2.69, + "learning_rate": 1.547314146010918e-05, + "loss": 0.2793, + "step": 6423 + }, + { + "epoch": 2.69, + "learning_rate": 1.5471722447560757e-05, + "loss": 0.2306, + "step": 6424 + }, + { + "epoch": 2.69, + "learning_rate": 1.547030327773253e-05, + "loss": 0.2505, + "step": 6425 + }, + { + "epoch": 2.69, + "learning_rate": 1.546888395066529e-05, + "loss": 0.2627, + "step": 6426 + }, + { + "epoch": 2.69, + "learning_rate": 1.546746446639983e-05, + "loss": 0.2706, + "step": 6427 + }, + { + "epoch": 2.69, + "learning_rate": 1.546604482497696e-05, + "loss": 0.2982, + "step": 6428 + }, + { + "epoch": 2.69, + "learning_rate": 1.546462502643748e-05, + "loss": 0.2484, + "step": 6429 + }, + { + "epoch": 2.69, + "learning_rate": 1.5463205070822202e-05, + "loss": 0.2251, + "step": 6430 + }, + { + "epoch": 2.69, + "learning_rate": 1.5461784958171945e-05, + "loss": 0.2504, + "step": 6431 + }, + { + "epoch": 2.69, + "learning_rate": 1.5460364688527523e-05, + "loss": 0.2677, + "step": 6432 + }, + { + "epoch": 2.69, + "learning_rate": 1.5458944261929763e-05, + "loss": 0.243, + "step": 6433 + }, + { + "epoch": 2.69, + "learning_rate": 1.5457523678419498e-05, + "loss": 0.2672, + "step": 6434 + }, + { + "epoch": 2.69, + "learning_rate": 1.5456102938037555e-05, + "loss": 0.2635, + "step": 6435 + }, + { + "epoch": 2.69, + "learning_rate": 1.5454682040824776e-05, + "loss": 0.2272, + "step": 6436 + }, + { + "epoch": 2.69, + "learning_rate": 1.5453260986822e-05, + "loss": 0.2772, + "step": 6437 + }, + { + "epoch": 2.7, + "learning_rate": 1.545183977607008e-05, + "loss": 0.2579, + "step": 6438 + }, + { + "epoch": 2.7, + "learning_rate": 1.545041840860986e-05, + "loss": 0.2872, + "step": 6439 + }, + { + "epoch": 2.7, + "learning_rate": 1.5448996884482198e-05, + "loss": 0.2657, + "step": 6440 + }, + { + "epoch": 2.7, + "learning_rate": 1.5447575203727964e-05, + "loss": 0.2656, + "step": 6441 + }, + { + "epoch": 2.7, + "learning_rate": 1.5446153366388007e-05, + "loss": 0.2635, + "step": 6442 + }, + { + "epoch": 2.7, + "learning_rate": 1.5444731372503206e-05, + "loss": 0.235, + "step": 6443 + }, + { + "epoch": 2.7, + "learning_rate": 1.5443309222114434e-05, + "loss": 0.2488, + "step": 6444 + }, + { + "epoch": 2.7, + "learning_rate": 1.544188691526257e-05, + "loss": 0.2692, + "step": 6445 + }, + { + "epoch": 2.7, + "learning_rate": 1.5440464451988494e-05, + "loss": 0.2598, + "step": 6446 + }, + { + "epoch": 2.7, + "learning_rate": 1.5439041832333095e-05, + "loss": 0.2663, + "step": 6447 + }, + { + "epoch": 2.7, + "learning_rate": 1.5437619056337267e-05, + "loss": 0.2528, + "step": 6448 + }, + { + "epoch": 2.7, + "learning_rate": 1.5436196124041902e-05, + "loss": 0.2391, + "step": 6449 + }, + { + "epoch": 2.7, + "learning_rate": 1.5434773035487905e-05, + "loss": 0.2803, + "step": 6450 + }, + { + "epoch": 2.7, + "learning_rate": 1.543334979071618e-05, + "loss": 0.2625, + "step": 6451 + }, + { + "epoch": 2.7, + "learning_rate": 1.5431926389767637e-05, + "loss": 0.2697, + "step": 6452 + }, + { + "epoch": 2.7, + "learning_rate": 1.543050283268319e-05, + "loss": 0.2431, + "step": 6453 + }, + { + "epoch": 2.7, + "learning_rate": 1.5429079119503752e-05, + "loss": 0.2628, + "step": 6454 + }, + { + "epoch": 2.7, + "learning_rate": 1.542765525027026e-05, + "loss": 0.266, + "step": 6455 + }, + { + "epoch": 2.7, + "learning_rate": 1.542623122502363e-05, + "loss": 0.2715, + "step": 6456 + }, + { + "epoch": 2.7, + "learning_rate": 1.54248070438048e-05, + "loss": 0.2704, + "step": 6457 + }, + { + "epoch": 2.7, + "learning_rate": 1.5423382706654705e-05, + "loss": 0.2164, + "step": 6458 + }, + { + "epoch": 2.7, + "learning_rate": 1.542195821361429e-05, + "loss": 0.2669, + "step": 6459 + }, + { + "epoch": 2.7, + "learning_rate": 1.5420533564724495e-05, + "loss": 0.2869, + "step": 6460 + }, + { + "epoch": 2.71, + "learning_rate": 1.5419108760026273e-05, + "loss": 0.2427, + "step": 6461 + }, + { + "epoch": 2.71, + "learning_rate": 1.5417683799560576e-05, + "loss": 0.248, + "step": 6462 + }, + { + "epoch": 2.71, + "learning_rate": 1.541625868336837e-05, + "loss": 0.2846, + "step": 6463 + }, + { + "epoch": 2.71, + "learning_rate": 1.5414833411490612e-05, + "loss": 0.2726, + "step": 6464 + }, + { + "epoch": 2.71, + "learning_rate": 1.5413407983968274e-05, + "loss": 0.2504, + "step": 6465 + }, + { + "epoch": 2.71, + "learning_rate": 1.5411982400842328e-05, + "loss": 0.239, + "step": 6466 + }, + { + "epoch": 2.71, + "learning_rate": 1.5410556662153747e-05, + "loss": 0.2547, + "step": 6467 + }, + { + "epoch": 2.71, + "learning_rate": 1.5409130767943516e-05, + "loss": 0.284, + "step": 6468 + }, + { + "epoch": 2.71, + "learning_rate": 1.5407704718252625e-05, + "loss": 0.2341, + "step": 6469 + }, + { + "epoch": 2.71, + "learning_rate": 1.5406278513122058e-05, + "loss": 0.2427, + "step": 6470 + }, + { + "epoch": 2.71, + "learning_rate": 1.5404852152592814e-05, + "loss": 0.2431, + "step": 6471 + }, + { + "epoch": 2.71, + "learning_rate": 1.540342563670589e-05, + "loss": 0.2642, + "step": 6472 + }, + { + "epoch": 2.71, + "learning_rate": 1.5401998965502293e-05, + "loss": 0.2666, + "step": 6473 + }, + { + "epoch": 2.71, + "learning_rate": 1.5400572139023026e-05, + "loss": 0.2834, + "step": 6474 + }, + { + "epoch": 2.71, + "learning_rate": 1.5399145157309105e-05, + "loss": 0.281, + "step": 6475 + }, + { + "epoch": 2.71, + "learning_rate": 1.539771802040155e-05, + "loss": 0.2498, + "step": 6476 + }, + { + "epoch": 2.71, + "learning_rate": 1.539629072834138e-05, + "loss": 0.2607, + "step": 6477 + }, + { + "epoch": 2.71, + "learning_rate": 1.539486328116962e-05, + "loss": 0.2705, + "step": 6478 + }, + { + "epoch": 2.71, + "learning_rate": 1.5393435678927305e-05, + "loss": 0.2686, + "step": 6479 + }, + { + "epoch": 2.71, + "learning_rate": 1.5392007921655466e-05, + "loss": 0.2472, + "step": 6480 + }, + { + "epoch": 2.71, + "learning_rate": 1.5390580009395147e-05, + "loss": 0.2202, + "step": 6481 + }, + { + "epoch": 2.71, + "learning_rate": 1.5389151942187383e-05, + "loss": 0.2606, + "step": 6482 + }, + { + "epoch": 2.71, + "learning_rate": 1.538772372007323e-05, + "loss": 0.2618, + "step": 6483 + }, + { + "epoch": 2.71, + "learning_rate": 1.5386295343093743e-05, + "loss": 0.2498, + "step": 6484 + }, + { + "epoch": 2.72, + "learning_rate": 1.5384866811289976e-05, + "loss": 0.263, + "step": 6485 + }, + { + "epoch": 2.72, + "learning_rate": 1.538343812470299e-05, + "loss": 0.2487, + "step": 6486 + }, + { + "epoch": 2.72, + "learning_rate": 1.538200928337385e-05, + "loss": 0.273, + "step": 6487 + }, + { + "epoch": 2.72, + "learning_rate": 1.538058028734363e-05, + "loss": 0.2388, + "step": 6488 + }, + { + "epoch": 2.72, + "learning_rate": 1.5379151136653407e-05, + "loss": 0.2524, + "step": 6489 + }, + { + "epoch": 2.72, + "learning_rate": 1.5377721831344258e-05, + "loss": 0.244, + "step": 6490 + }, + { + "epoch": 2.72, + "learning_rate": 1.537629237145726e-05, + "loss": 0.2628, + "step": 6491 + }, + { + "epoch": 2.72, + "learning_rate": 1.5374862757033518e-05, + "loss": 0.2961, + "step": 6492 + }, + { + "epoch": 2.72, + "learning_rate": 1.5373432988114113e-05, + "loss": 0.2753, + "step": 6493 + }, + { + "epoch": 2.72, + "learning_rate": 1.537200306474014e-05, + "loss": 0.2629, + "step": 6494 + }, + { + "epoch": 2.72, + "learning_rate": 1.5370572986952712e-05, + "loss": 0.2631, + "step": 6495 + }, + { + "epoch": 2.72, + "learning_rate": 1.5369142754792926e-05, + "loss": 0.2519, + "step": 6496 + }, + { + "epoch": 2.72, + "learning_rate": 1.5367712368301897e-05, + "loss": 0.2696, + "step": 6497 + }, + { + "epoch": 2.72, + "learning_rate": 1.536628182752074e-05, + "loss": 0.2714, + "step": 6498 + }, + { + "epoch": 2.72, + "learning_rate": 1.536485113249057e-05, + "loss": 0.2673, + "step": 6499 + }, + { + "epoch": 2.72, + "learning_rate": 1.5363420283252518e-05, + "loss": 0.2689, + "step": 6500 + }, + { + "epoch": 2.72, + "learning_rate": 1.536198927984771e-05, + "loss": 0.2581, + "step": 6501 + }, + { + "epoch": 2.72, + "learning_rate": 1.5360558122317275e-05, + "loss": 0.2747, + "step": 6502 + }, + { + "epoch": 2.72, + "learning_rate": 1.5359126810702357e-05, + "loss": 0.2599, + "step": 6503 + }, + { + "epoch": 2.72, + "learning_rate": 1.5357695345044096e-05, + "loss": 0.2317, + "step": 6504 + }, + { + "epoch": 2.72, + "learning_rate": 1.5356263725383633e-05, + "loss": 0.264, + "step": 6505 + }, + { + "epoch": 2.72, + "learning_rate": 1.5354831951762122e-05, + "loss": 0.2814, + "step": 6506 + }, + { + "epoch": 2.72, + "learning_rate": 1.535340002422072e-05, + "loss": 0.2611, + "step": 6507 + }, + { + "epoch": 2.72, + "learning_rate": 1.5351967942800584e-05, + "loss": 0.2676, + "step": 6508 + }, + { + "epoch": 2.73, + "learning_rate": 1.535053570754288e-05, + "loss": 0.2563, + "step": 6509 + }, + { + "epoch": 2.73, + "learning_rate": 1.5349103318488772e-05, + "loss": 0.2836, + "step": 6510 + }, + { + "epoch": 2.73, + "learning_rate": 1.534767077567944e-05, + "loss": 0.257, + "step": 6511 + }, + { + "epoch": 2.73, + "learning_rate": 1.5346238079156055e-05, + "loss": 0.2543, + "step": 6512 + }, + { + "epoch": 2.73, + "learning_rate": 1.53448052289598e-05, + "loss": 0.2448, + "step": 6513 + }, + { + "epoch": 2.73, + "learning_rate": 1.5343372225131863e-05, + "loss": 0.2426, + "step": 6514 + }, + { + "epoch": 2.73, + "learning_rate": 1.5341939067713433e-05, + "loss": 0.2579, + "step": 6515 + }, + { + "epoch": 2.73, + "learning_rate": 1.5340505756745703e-05, + "loss": 0.2629, + "step": 6516 + }, + { + "epoch": 2.73, + "learning_rate": 1.5339072292269874e-05, + "loss": 0.2614, + "step": 6517 + }, + { + "epoch": 2.73, + "learning_rate": 1.5337638674327152e-05, + "loss": 0.2702, + "step": 6518 + }, + { + "epoch": 2.73, + "learning_rate": 1.5336204902958743e-05, + "loss": 0.249, + "step": 6519 + }, + { + "epoch": 2.73, + "learning_rate": 1.5334770978205857e-05, + "loss": 0.2522, + "step": 6520 + }, + { + "epoch": 2.73, + "learning_rate": 1.5333336900109715e-05, + "loss": 0.2293, + "step": 6521 + }, + { + "epoch": 2.73, + "learning_rate": 1.5331902668711536e-05, + "loss": 0.2922, + "step": 6522 + }, + { + "epoch": 2.73, + "learning_rate": 1.5330468284052544e-05, + "loss": 0.2786, + "step": 6523 + }, + { + "epoch": 2.73, + "learning_rate": 1.5329033746173975e-05, + "loss": 0.2407, + "step": 6524 + }, + { + "epoch": 2.73, + "learning_rate": 1.5327599055117057e-05, + "loss": 0.2681, + "step": 6525 + }, + { + "epoch": 2.73, + "learning_rate": 1.5326164210923032e-05, + "loss": 0.2567, + "step": 6526 + }, + { + "epoch": 2.73, + "learning_rate": 1.5324729213633146e-05, + "loss": 0.268, + "step": 6527 + }, + { + "epoch": 2.73, + "learning_rate": 1.532329406328864e-05, + "loss": 0.2753, + "step": 6528 + }, + { + "epoch": 2.73, + "learning_rate": 1.5321858759930776e-05, + "loss": 0.256, + "step": 6529 + }, + { + "epoch": 2.73, + "learning_rate": 1.53204233036008e-05, + "loss": 0.2613, + "step": 6530 + }, + { + "epoch": 2.73, + "learning_rate": 1.531898769433998e-05, + "loss": 0.2602, + "step": 6531 + }, + { + "epoch": 2.73, + "learning_rate": 1.5317551932189574e-05, + "loss": 0.2707, + "step": 6532 + }, + { + "epoch": 2.74, + "learning_rate": 1.531611601719086e-05, + "loss": 0.2616, + "step": 6533 + }, + { + "epoch": 2.74, + "learning_rate": 1.5314679949385105e-05, + "loss": 0.2707, + "step": 6534 + }, + { + "epoch": 2.74, + "learning_rate": 1.5313243728813598e-05, + "loss": 0.2467, + "step": 6535 + }, + { + "epoch": 2.74, + "learning_rate": 1.5311807355517608e-05, + "loss": 0.257, + "step": 6536 + }, + { + "epoch": 2.74, + "learning_rate": 1.531037082953843e-05, + "loss": 0.2538, + "step": 6537 + }, + { + "epoch": 2.74, + "learning_rate": 1.530893415091736e-05, + "loss": 0.2585, + "step": 6538 + }, + { + "epoch": 2.74, + "learning_rate": 1.5307497319695683e-05, + "loss": 0.2764, + "step": 6539 + }, + { + "epoch": 2.74, + "learning_rate": 1.5306060335914706e-05, + "loss": 0.2576, + "step": 6540 + }, + { + "epoch": 2.74, + "learning_rate": 1.5304623199615737e-05, + "loss": 0.2602, + "step": 6541 + }, + { + "epoch": 2.74, + "learning_rate": 1.530318591084008e-05, + "loss": 0.2584, + "step": 6542 + }, + { + "epoch": 2.74, + "learning_rate": 1.530174846962905e-05, + "loss": 0.2904, + "step": 6543 + }, + { + "epoch": 2.74, + "learning_rate": 1.530031087602396e-05, + "loss": 0.2618, + "step": 6544 + }, + { + "epoch": 2.74, + "learning_rate": 1.5298873130066144e-05, + "loss": 0.272, + "step": 6545 + }, + { + "epoch": 2.74, + "learning_rate": 1.529743523179692e-05, + "loss": 0.2769, + "step": 6546 + }, + { + "epoch": 2.74, + "learning_rate": 1.529599718125762e-05, + "loss": 0.228, + "step": 6547 + }, + { + "epoch": 2.74, + "learning_rate": 1.5294558978489583e-05, + "loss": 0.2458, + "step": 6548 + }, + { + "epoch": 2.74, + "learning_rate": 1.5293120623534145e-05, + "loss": 0.2771, + "step": 6549 + }, + { + "epoch": 2.74, + "learning_rate": 1.529168211643265e-05, + "loss": 0.244, + "step": 6550 + }, + { + "epoch": 2.74, + "learning_rate": 1.529024345722645e-05, + "loss": 0.2994, + "step": 6551 + }, + { + "epoch": 2.74, + "learning_rate": 1.5288804645956896e-05, + "loss": 0.2486, + "step": 6552 + }, + { + "epoch": 2.74, + "learning_rate": 1.5287365682665346e-05, + "loss": 0.2665, + "step": 6553 + }, + { + "epoch": 2.74, + "learning_rate": 1.5285926567393158e-05, + "loss": 0.2559, + "step": 6554 + }, + { + "epoch": 2.74, + "learning_rate": 1.5284487300181707e-05, + "loss": 0.2747, + "step": 6555 + }, + { + "epoch": 2.74, + "learning_rate": 1.5283047881072357e-05, + "loss": 0.2544, + "step": 6556 + }, + { + "epoch": 2.75, + "learning_rate": 1.5281608310106478e-05, + "loss": 0.2588, + "step": 6557 + }, + { + "epoch": 2.75, + "learning_rate": 1.5280168587325462e-05, + "loss": 0.2612, + "step": 6558 + }, + { + "epoch": 2.75, + "learning_rate": 1.5278728712770683e-05, + "loss": 0.2553, + "step": 6559 + }, + { + "epoch": 2.75, + "learning_rate": 1.5277288686483528e-05, + "loss": 0.2601, + "step": 6560 + }, + { + "epoch": 2.75, + "learning_rate": 1.5275848508505396e-05, + "loss": 0.2463, + "step": 6561 + }, + { + "epoch": 2.75, + "learning_rate": 1.527440817887768e-05, + "loss": 0.2555, + "step": 6562 + }, + { + "epoch": 2.75, + "learning_rate": 1.5272967697641784e-05, + "loss": 0.2678, + "step": 6563 + }, + { + "epoch": 2.75, + "learning_rate": 1.5271527064839107e-05, + "loss": 0.265, + "step": 6564 + }, + { + "epoch": 2.75, + "learning_rate": 1.5270086280511064e-05, + "loss": 0.2693, + "step": 6565 + }, + { + "epoch": 2.75, + "learning_rate": 1.5268645344699066e-05, + "loss": 0.2654, + "step": 6566 + }, + { + "epoch": 2.75, + "learning_rate": 1.5267204257444534e-05, + "loss": 0.2538, + "step": 6567 + }, + { + "epoch": 2.75, + "learning_rate": 1.526576301878889e-05, + "loss": 0.2554, + "step": 6568 + }, + { + "epoch": 2.75, + "learning_rate": 1.526432162877356e-05, + "loss": 0.2549, + "step": 6569 + }, + { + "epoch": 2.75, + "learning_rate": 1.5262880087439975e-05, + "loss": 0.2483, + "step": 6570 + }, + { + "epoch": 2.75, + "learning_rate": 1.526143839482957e-05, + "loss": 0.2568, + "step": 6571 + }, + { + "epoch": 2.75, + "learning_rate": 1.5259996550983797e-05, + "loss": 0.259, + "step": 6572 + }, + { + "epoch": 2.75, + "learning_rate": 1.5258554555944085e-05, + "loss": 0.2885, + "step": 6573 + }, + { + "epoch": 2.75, + "learning_rate": 1.5257112409751886e-05, + "loss": 0.2732, + "step": 6574 + }, + { + "epoch": 2.75, + "learning_rate": 1.5255670112448657e-05, + "loss": 0.2344, + "step": 6575 + }, + { + "epoch": 2.75, + "learning_rate": 1.5254227664075858e-05, + "loss": 0.2616, + "step": 6576 + }, + { + "epoch": 2.75, + "learning_rate": 1.5252785064674945e-05, + "loss": 0.27, + "step": 6577 + }, + { + "epoch": 2.75, + "learning_rate": 1.5251342314287385e-05, + "loss": 0.2484, + "step": 6578 + }, + { + "epoch": 2.75, + "learning_rate": 1.5249899412954652e-05, + "loss": 0.2413, + "step": 6579 + }, + { + "epoch": 2.75, + "learning_rate": 1.5248456360718218e-05, + "loss": 0.2837, + "step": 6580 + }, + { + "epoch": 2.76, + "learning_rate": 1.524701315761956e-05, + "loss": 0.2716, + "step": 6581 + }, + { + "epoch": 2.76, + "learning_rate": 1.5245569803700174e-05, + "loss": 0.2518, + "step": 6582 + }, + { + "epoch": 2.76, + "learning_rate": 1.5244126299001534e-05, + "loss": 0.249, + "step": 6583 + }, + { + "epoch": 2.76, + "learning_rate": 1.5242682643565135e-05, + "loss": 0.2523, + "step": 6584 + }, + { + "epoch": 2.76, + "learning_rate": 1.5241238837432477e-05, + "loss": 0.2322, + "step": 6585 + }, + { + "epoch": 2.76, + "learning_rate": 1.523979488064506e-05, + "loss": 0.2686, + "step": 6586 + }, + { + "epoch": 2.76, + "learning_rate": 1.523835077324439e-05, + "loss": 0.2603, + "step": 6587 + }, + { + "epoch": 2.76, + "learning_rate": 1.5236906515271974e-05, + "loss": 0.2674, + "step": 6588 + }, + { + "epoch": 2.76, + "learning_rate": 1.5235462106769327e-05, + "loss": 0.2468, + "step": 6589 + }, + { + "epoch": 2.76, + "learning_rate": 1.5234017547777969e-05, + "loss": 0.2629, + "step": 6590 + }, + { + "epoch": 2.76, + "learning_rate": 1.5232572838339418e-05, + "loss": 0.2581, + "step": 6591 + }, + { + "epoch": 2.76, + "learning_rate": 1.5231127978495208e-05, + "loss": 0.2561, + "step": 6592 + }, + { + "epoch": 2.76, + "learning_rate": 1.5229682968286862e-05, + "loss": 0.2629, + "step": 6593 + }, + { + "epoch": 2.76, + "learning_rate": 1.5228237807755925e-05, + "loss": 0.254, + "step": 6594 + }, + { + "epoch": 2.76, + "learning_rate": 1.522679249694393e-05, + "loss": 0.2554, + "step": 6595 + }, + { + "epoch": 2.76, + "learning_rate": 1.5225347035892423e-05, + "loss": 0.2422, + "step": 6596 + }, + { + "epoch": 2.76, + "learning_rate": 1.522390142464295e-05, + "loss": 0.2471, + "step": 6597 + }, + { + "epoch": 2.76, + "learning_rate": 1.5222455663237068e-05, + "loss": 0.2608, + "step": 6598 + }, + { + "epoch": 2.76, + "learning_rate": 1.5221009751716333e-05, + "loss": 0.2835, + "step": 6599 + }, + { + "epoch": 2.76, + "learning_rate": 1.5219563690122306e-05, + "loss": 0.2757, + "step": 6600 + }, + { + "epoch": 2.76, + "learning_rate": 1.5218117478496548e-05, + "loss": 0.2538, + "step": 6601 + }, + { + "epoch": 2.76, + "learning_rate": 1.5216671116880637e-05, + "loss": 0.274, + "step": 6602 + }, + { + "epoch": 2.76, + "learning_rate": 1.5215224605316143e-05, + "loss": 0.2488, + "step": 6603 + }, + { + "epoch": 2.76, + "learning_rate": 1.5213777943844648e-05, + "loss": 0.2378, + "step": 6604 + }, + { + "epoch": 2.77, + "learning_rate": 1.521233113250773e-05, + "loss": 0.2754, + "step": 6605 + }, + { + "epoch": 2.77, + "learning_rate": 1.5210884171346983e-05, + "loss": 0.263, + "step": 6606 + }, + { + "epoch": 2.77, + "learning_rate": 1.5209437060403988e-05, + "loss": 0.2472, + "step": 6607 + }, + { + "epoch": 2.77, + "learning_rate": 1.5207989799720352e-05, + "loss": 0.2798, + "step": 6608 + }, + { + "epoch": 2.77, + "learning_rate": 1.520654238933767e-05, + "loss": 0.2764, + "step": 6609 + }, + { + "epoch": 2.77, + "learning_rate": 1.5205094829297546e-05, + "loss": 0.2551, + "step": 6610 + }, + { + "epoch": 2.77, + "learning_rate": 1.5203647119641591e-05, + "loss": 0.2434, + "step": 6611 + }, + { + "epoch": 2.77, + "learning_rate": 1.5202199260411417e-05, + "loss": 0.2527, + "step": 6612 + }, + { + "epoch": 2.77, + "learning_rate": 1.5200751251648641e-05, + "loss": 0.2886, + "step": 6613 + }, + { + "epoch": 2.77, + "learning_rate": 1.5199303093394883e-05, + "loss": 0.2617, + "step": 6614 + }, + { + "epoch": 2.77, + "learning_rate": 1.5197854785691775e-05, + "loss": 0.2527, + "step": 6615 + }, + { + "epoch": 2.77, + "learning_rate": 1.5196406328580943e-05, + "loss": 0.273, + "step": 6616 + }, + { + "epoch": 2.77, + "learning_rate": 1.5194957722104024e-05, + "loss": 0.2436, + "step": 6617 + }, + { + "epoch": 2.77, + "learning_rate": 1.5193508966302655e-05, + "loss": 0.2875, + "step": 6618 + }, + { + "epoch": 2.77, + "learning_rate": 1.5192060061218477e-05, + "loss": 0.2511, + "step": 6619 + }, + { + "epoch": 2.77, + "learning_rate": 1.519061100689314e-05, + "loss": 0.2614, + "step": 6620 + }, + { + "epoch": 2.77, + "learning_rate": 1.5189161803368297e-05, + "loss": 0.2674, + "step": 6621 + }, + { + "epoch": 2.77, + "learning_rate": 1.51877124506856e-05, + "loss": 0.2546, + "step": 6622 + }, + { + "epoch": 2.77, + "learning_rate": 1.5186262948886716e-05, + "loss": 0.2376, + "step": 6623 + }, + { + "epoch": 2.77, + "learning_rate": 1.5184813298013304e-05, + "loss": 0.243, + "step": 6624 + }, + { + "epoch": 2.77, + "learning_rate": 1.5183363498107033e-05, + "loss": 0.2854, + "step": 6625 + }, + { + "epoch": 2.77, + "learning_rate": 1.5181913549209582e-05, + "loss": 0.2663, + "step": 6626 + }, + { + "epoch": 2.77, + "learning_rate": 1.5180463451362621e-05, + "loss": 0.2811, + "step": 6627 + }, + { + "epoch": 2.77, + "learning_rate": 1.5179013204607838e-05, + "loss": 0.2553, + "step": 6628 + }, + { + "epoch": 2.78, + "learning_rate": 1.5177562808986914e-05, + "loss": 0.2549, + "step": 6629 + }, + { + "epoch": 2.78, + "learning_rate": 1.5176112264541542e-05, + "loss": 0.249, + "step": 6630 + }, + { + "epoch": 2.78, + "learning_rate": 1.5174661571313417e-05, + "loss": 0.2608, + "step": 6631 + }, + { + "epoch": 2.78, + "learning_rate": 1.5173210729344234e-05, + "loss": 0.2462, + "step": 6632 + }, + { + "epoch": 2.78, + "learning_rate": 1.5171759738675702e-05, + "loss": 0.2513, + "step": 6633 + }, + { + "epoch": 2.78, + "learning_rate": 1.5170308599349524e-05, + "loss": 0.2768, + "step": 6634 + }, + { + "epoch": 2.78, + "learning_rate": 1.5168857311407411e-05, + "loss": 0.2487, + "step": 6635 + }, + { + "epoch": 2.78, + "learning_rate": 1.516740587489109e-05, + "loss": 0.2677, + "step": 6636 + }, + { + "epoch": 2.78, + "learning_rate": 1.5165954289842267e-05, + "loss": 0.2875, + "step": 6637 + }, + { + "epoch": 2.78, + "learning_rate": 1.516450255630267e-05, + "loss": 0.2793, + "step": 6638 + }, + { + "epoch": 2.78, + "learning_rate": 1.5163050674314033e-05, + "loss": 0.2797, + "step": 6639 + }, + { + "epoch": 2.78, + "learning_rate": 1.5161598643918086e-05, + "loss": 0.2556, + "step": 6640 + }, + { + "epoch": 2.78, + "learning_rate": 1.5160146465156564e-05, + "loss": 0.2648, + "step": 6641 + }, + { + "epoch": 2.78, + "learning_rate": 1.5158694138071213e-05, + "loss": 0.2395, + "step": 6642 + }, + { + "epoch": 2.78, + "learning_rate": 1.5157241662703777e-05, + "loss": 0.2652, + "step": 6643 + }, + { + "epoch": 2.78, + "learning_rate": 1.5155789039096005e-05, + "loss": 0.2498, + "step": 6644 + }, + { + "epoch": 2.78, + "learning_rate": 1.515433626728965e-05, + "loss": 0.2564, + "step": 6645 + }, + { + "epoch": 2.78, + "learning_rate": 1.5152883347326475e-05, + "loss": 0.2747, + "step": 6646 + }, + { + "epoch": 2.78, + "learning_rate": 1.5151430279248242e-05, + "loss": 0.2571, + "step": 6647 + }, + { + "epoch": 2.78, + "learning_rate": 1.5149977063096715e-05, + "loss": 0.2651, + "step": 6648 + }, + { + "epoch": 2.78, + "learning_rate": 1.5148523698913668e-05, + "loss": 0.2875, + "step": 6649 + }, + { + "epoch": 2.78, + "learning_rate": 1.5147070186740877e-05, + "loss": 0.2477, + "step": 6650 + }, + { + "epoch": 2.78, + "learning_rate": 1.514561652662012e-05, + "loss": 0.2882, + "step": 6651 + }, + { + "epoch": 2.79, + "learning_rate": 1.5144162718593183e-05, + "loss": 0.2645, + "step": 6652 + }, + { + "epoch": 2.79, + "learning_rate": 1.5142708762701852e-05, + "loss": 0.279, + "step": 6653 + }, + { + "epoch": 2.79, + "learning_rate": 1.5141254658987922e-05, + "loss": 0.2428, + "step": 6654 + }, + { + "epoch": 2.79, + "learning_rate": 1.5139800407493189e-05, + "loss": 0.2613, + "step": 6655 + }, + { + "epoch": 2.79, + "learning_rate": 1.5138346008259452e-05, + "loss": 0.2361, + "step": 6656 + }, + { + "epoch": 2.79, + "learning_rate": 1.5136891461328525e-05, + "loss": 0.2499, + "step": 6657 + }, + { + "epoch": 2.79, + "learning_rate": 1.5135436766742207e-05, + "loss": 0.2513, + "step": 6658 + }, + { + "epoch": 2.79, + "learning_rate": 1.5133981924542314e-05, + "loss": 0.2727, + "step": 6659 + }, + { + "epoch": 2.79, + "learning_rate": 1.513252693477067e-05, + "loss": 0.2188, + "step": 6660 + }, + { + "epoch": 2.79, + "learning_rate": 1.5131071797469093e-05, + "loss": 0.2486, + "step": 6661 + }, + { + "epoch": 2.79, + "learning_rate": 1.512961651267941e-05, + "loss": 0.2672, + "step": 6662 + }, + { + "epoch": 2.79, + "learning_rate": 1.5128161080443451e-05, + "loss": 0.2455, + "step": 6663 + }, + { + "epoch": 2.79, + "learning_rate": 1.5126705500803055e-05, + "loss": 0.2667, + "step": 6664 + }, + { + "epoch": 2.79, + "learning_rate": 1.5125249773800057e-05, + "loss": 0.2744, + "step": 6665 + }, + { + "epoch": 2.79, + "learning_rate": 1.5123793899476303e-05, + "loss": 0.2415, + "step": 6666 + }, + { + "epoch": 2.79, + "learning_rate": 1.5122337877873641e-05, + "loss": 0.2695, + "step": 6667 + }, + { + "epoch": 2.79, + "learning_rate": 1.5120881709033919e-05, + "loss": 0.2449, + "step": 6668 + }, + { + "epoch": 2.79, + "learning_rate": 1.5119425392999e-05, + "loss": 0.2488, + "step": 6669 + }, + { + "epoch": 2.79, + "learning_rate": 1.511796892981074e-05, + "loss": 0.2597, + "step": 6670 + }, + { + "epoch": 2.79, + "learning_rate": 1.5116512319511003e-05, + "loss": 0.253, + "step": 6671 + }, + { + "epoch": 2.79, + "learning_rate": 1.5115055562141661e-05, + "loss": 0.2759, + "step": 6672 + }, + { + "epoch": 2.79, + "learning_rate": 1.5113598657744585e-05, + "loss": 0.3121, + "step": 6673 + }, + { + "epoch": 2.79, + "learning_rate": 1.5112141606361657e-05, + "loss": 0.2661, + "step": 6674 + }, + { + "epoch": 2.79, + "learning_rate": 1.511068440803475e-05, + "loss": 0.2716, + "step": 6675 + }, + { + "epoch": 2.8, + "learning_rate": 1.5109227062805758e-05, + "loss": 0.2753, + "step": 6676 + }, + { + "epoch": 2.8, + "learning_rate": 1.5107769570716566e-05, + "loss": 0.2742, + "step": 6677 + }, + { + "epoch": 2.8, + "learning_rate": 1.510631193180907e-05, + "loss": 0.2392, + "step": 6678 + }, + { + "epoch": 2.8, + "learning_rate": 1.510485414612517e-05, + "loss": 0.2766, + "step": 6679 + }, + { + "epoch": 2.8, + "learning_rate": 1.510339621370677e-05, + "loss": 0.2514, + "step": 6680 + }, + { + "epoch": 2.8, + "learning_rate": 1.510193813459577e-05, + "loss": 0.2646, + "step": 6681 + }, + { + "epoch": 2.8, + "learning_rate": 1.5100479908834087e-05, + "loss": 0.2824, + "step": 6682 + }, + { + "epoch": 2.8, + "learning_rate": 1.5099021536463636e-05, + "loss": 0.2641, + "step": 6683 + }, + { + "epoch": 2.8, + "learning_rate": 1.5097563017526336e-05, + "loss": 0.2753, + "step": 6684 + }, + { + "epoch": 2.8, + "learning_rate": 1.5096104352064112e-05, + "loss": 0.2636, + "step": 6685 + }, + { + "epoch": 2.8, + "learning_rate": 1.5094645540118888e-05, + "loss": 0.2799, + "step": 6686 + }, + { + "epoch": 2.8, + "learning_rate": 1.5093186581732599e-05, + "loss": 0.2483, + "step": 6687 + }, + { + "epoch": 2.8, + "learning_rate": 1.5091727476947182e-05, + "loss": 0.2439, + "step": 6688 + }, + { + "epoch": 2.8, + "learning_rate": 1.5090268225804577e-05, + "loss": 0.25, + "step": 6689 + }, + { + "epoch": 2.8, + "learning_rate": 1.5088808828346727e-05, + "loss": 0.2747, + "step": 6690 + }, + { + "epoch": 2.8, + "learning_rate": 1.5087349284615584e-05, + "loss": 0.255, + "step": 6691 + }, + { + "epoch": 2.8, + "learning_rate": 1.50858895946531e-05, + "loss": 0.2375, + "step": 6692 + }, + { + "epoch": 2.8, + "learning_rate": 1.5084429758501235e-05, + "loss": 0.2505, + "step": 6693 + }, + { + "epoch": 2.8, + "learning_rate": 1.5082969776201948e-05, + "loss": 0.2722, + "step": 6694 + }, + { + "epoch": 2.8, + "learning_rate": 1.5081509647797204e-05, + "loss": 0.2526, + "step": 6695 + }, + { + "epoch": 2.8, + "learning_rate": 1.5080049373328976e-05, + "loss": 0.263, + "step": 6696 + }, + { + "epoch": 2.8, + "learning_rate": 1.5078588952839235e-05, + "loss": 0.2623, + "step": 6697 + }, + { + "epoch": 2.8, + "learning_rate": 1.507712838636996e-05, + "loss": 0.245, + "step": 6698 + }, + { + "epoch": 2.8, + "learning_rate": 1.5075667673963139e-05, + "loss": 0.2681, + "step": 6699 + }, + { + "epoch": 2.81, + "learning_rate": 1.507420681566075e-05, + "loss": 0.2744, + "step": 6700 + }, + { + "epoch": 2.81, + "learning_rate": 1.5072745811504793e-05, + "loss": 0.2359, + "step": 6701 + }, + { + "epoch": 2.81, + "learning_rate": 1.5071284661537257e-05, + "loss": 0.2707, + "step": 6702 + }, + { + "epoch": 2.81, + "learning_rate": 1.5069823365800146e-05, + "loss": 0.2543, + "step": 6703 + }, + { + "epoch": 2.81, + "learning_rate": 1.5068361924335457e-05, + "loss": 0.2385, + "step": 6704 + }, + { + "epoch": 2.81, + "learning_rate": 1.5066900337185203e-05, + "loss": 0.2427, + "step": 6705 + }, + { + "epoch": 2.81, + "learning_rate": 1.5065438604391398e-05, + "loss": 0.2722, + "step": 6706 + }, + { + "epoch": 2.81, + "learning_rate": 1.5063976725996052e-05, + "loss": 0.2621, + "step": 6707 + }, + { + "epoch": 2.81, + "learning_rate": 1.5062514702041193e-05, + "loss": 0.253, + "step": 6708 + }, + { + "epoch": 2.81, + "learning_rate": 1.5061052532568838e-05, + "loss": 0.2611, + "step": 6709 + }, + { + "epoch": 2.81, + "learning_rate": 1.5059590217621018e-05, + "loss": 0.2644, + "step": 6710 + }, + { + "epoch": 2.81, + "learning_rate": 1.5058127757239772e-05, + "loss": 0.2422, + "step": 6711 + }, + { + "epoch": 2.81, + "learning_rate": 1.5056665151467126e-05, + "loss": 0.2587, + "step": 6712 + }, + { + "epoch": 2.81, + "learning_rate": 1.505520240034513e-05, + "loss": 0.2658, + "step": 6713 + }, + { + "epoch": 2.81, + "learning_rate": 1.505373950391583e-05, + "loss": 0.248, + "step": 6714 + }, + { + "epoch": 2.81, + "learning_rate": 1.5052276462221267e-05, + "loss": 0.3038, + "step": 6715 + }, + { + "epoch": 2.81, + "learning_rate": 1.5050813275303501e-05, + "loss": 0.2474, + "step": 6716 + }, + { + "epoch": 2.81, + "learning_rate": 1.5049349943204595e-05, + "loss": 0.2528, + "step": 6717 + }, + { + "epoch": 2.81, + "learning_rate": 1.50478864659666e-05, + "loss": 0.2764, + "step": 6718 + }, + { + "epoch": 2.81, + "learning_rate": 1.5046422843631594e-05, + "loss": 0.2536, + "step": 6719 + }, + { + "epoch": 2.81, + "learning_rate": 1.5044959076241634e-05, + "loss": 0.246, + "step": 6720 + }, + { + "epoch": 2.81, + "learning_rate": 1.5043495163838809e-05, + "loss": 0.2581, + "step": 6721 + }, + { + "epoch": 2.81, + "learning_rate": 1.5042031106465186e-05, + "loss": 0.2556, + "step": 6722 + }, + { + "epoch": 2.81, + "learning_rate": 1.5040566904162855e-05, + "loss": 0.2449, + "step": 6723 + }, + { + "epoch": 2.82, + "learning_rate": 1.5039102556973899e-05, + "loss": 0.2626, + "step": 6724 + }, + { + "epoch": 2.82, + "learning_rate": 1.5037638064940416e-05, + "loss": 0.2653, + "step": 6725 + }, + { + "epoch": 2.82, + "learning_rate": 1.5036173428104493e-05, + "loss": 0.256, + "step": 6726 + }, + { + "epoch": 2.82, + "learning_rate": 1.5034708646508236e-05, + "loss": 0.2708, + "step": 6727 + }, + { + "epoch": 2.82, + "learning_rate": 1.5033243720193746e-05, + "loss": 0.2576, + "step": 6728 + }, + { + "epoch": 2.82, + "learning_rate": 1.5031778649203133e-05, + "loss": 0.2685, + "step": 6729 + }, + { + "epoch": 2.82, + "learning_rate": 1.5030313433578508e-05, + "loss": 0.2787, + "step": 6730 + }, + { + "epoch": 2.82, + "learning_rate": 1.5028848073361987e-05, + "loss": 0.2648, + "step": 6731 + }, + { + "epoch": 2.82, + "learning_rate": 1.5027382568595688e-05, + "loss": 0.2672, + "step": 6732 + }, + { + "epoch": 2.82, + "learning_rate": 1.5025916919321741e-05, + "loss": 0.2528, + "step": 6733 + }, + { + "epoch": 2.82, + "learning_rate": 1.5024451125582271e-05, + "loss": 0.2618, + "step": 6734 + }, + { + "epoch": 2.82, + "learning_rate": 1.5022985187419411e-05, + "loss": 0.2609, + "step": 6735 + }, + { + "epoch": 2.82, + "learning_rate": 1.50215191048753e-05, + "loss": 0.254, + "step": 6736 + }, + { + "epoch": 2.82, + "learning_rate": 1.5020052877992077e-05, + "loss": 0.2502, + "step": 6737 + }, + { + "epoch": 2.82, + "learning_rate": 1.5018586506811893e-05, + "loss": 0.24, + "step": 6738 + }, + { + "epoch": 2.82, + "learning_rate": 1.501711999137689e-05, + "loss": 0.2637, + "step": 6739 + }, + { + "epoch": 2.82, + "learning_rate": 1.5015653331729225e-05, + "loss": 0.2965, + "step": 6740 + }, + { + "epoch": 2.82, + "learning_rate": 1.5014186527911056e-05, + "loss": 0.2677, + "step": 6741 + }, + { + "epoch": 2.82, + "learning_rate": 1.5012719579964548e-05, + "loss": 0.2725, + "step": 6742 + }, + { + "epoch": 2.82, + "learning_rate": 1.501125248793186e-05, + "loss": 0.2443, + "step": 6743 + }, + { + "epoch": 2.82, + "learning_rate": 1.5009785251855164e-05, + "loss": 0.2601, + "step": 6744 + }, + { + "epoch": 2.82, + "learning_rate": 1.5008317871776643e-05, + "loss": 0.241, + "step": 6745 + }, + { + "epoch": 2.82, + "learning_rate": 1.5006850347738463e-05, + "loss": 0.2925, + "step": 6746 + }, + { + "epoch": 2.82, + "learning_rate": 1.5005382679782817e-05, + "loss": 0.26, + "step": 6747 + }, + { + "epoch": 2.83, + "learning_rate": 1.5003914867951888e-05, + "loss": 0.2512, + "step": 6748 + }, + { + "epoch": 2.83, + "learning_rate": 1.5002446912287862e-05, + "loss": 0.2741, + "step": 6749 + }, + { + "epoch": 2.83, + "learning_rate": 1.5000978812832942e-05, + "loss": 0.2485, + "step": 6750 + }, + { + "epoch": 2.83, + "learning_rate": 1.4999510569629323e-05, + "loss": 0.2567, + "step": 6751 + }, + { + "epoch": 2.83, + "learning_rate": 1.4998042182719207e-05, + "loss": 0.2542, + "step": 6752 + }, + { + "epoch": 2.83, + "learning_rate": 1.4996573652144807e-05, + "loss": 0.2555, + "step": 6753 + }, + { + "epoch": 2.83, + "learning_rate": 1.4995104977948331e-05, + "loss": 0.2351, + "step": 6754 + }, + { + "epoch": 2.83, + "learning_rate": 1.4993636160171994e-05, + "loss": 0.2558, + "step": 6755 + }, + { + "epoch": 2.83, + "learning_rate": 1.4992167198858016e-05, + "loss": 0.2562, + "step": 6756 + }, + { + "epoch": 2.83, + "learning_rate": 1.4990698094048624e-05, + "loss": 0.2462, + "step": 6757 + }, + { + "epoch": 2.83, + "learning_rate": 1.498922884578604e-05, + "loss": 0.2662, + "step": 6758 + }, + { + "epoch": 2.83, + "learning_rate": 1.4987759454112505e-05, + "loss": 0.2444, + "step": 6759 + }, + { + "epoch": 2.83, + "learning_rate": 1.4986289919070247e-05, + "loss": 0.2565, + "step": 6760 + }, + { + "epoch": 2.83, + "learning_rate": 1.4984820240701513e-05, + "loss": 0.2805, + "step": 6761 + }, + { + "epoch": 2.83, + "learning_rate": 1.4983350419048544e-05, + "loss": 0.254, + "step": 6762 + }, + { + "epoch": 2.83, + "learning_rate": 1.4981880454153586e-05, + "loss": 0.2438, + "step": 6763 + }, + { + "epoch": 2.83, + "learning_rate": 1.49804103460589e-05, + "loss": 0.2595, + "step": 6764 + }, + { + "epoch": 2.83, + "learning_rate": 1.4978940094806733e-05, + "loss": 0.254, + "step": 6765 + }, + { + "epoch": 2.83, + "learning_rate": 1.4977469700439358e-05, + "loss": 0.2385, + "step": 6766 + }, + { + "epoch": 2.83, + "learning_rate": 1.4975999162999027e-05, + "loss": 0.2545, + "step": 6767 + }, + { + "epoch": 2.83, + "learning_rate": 1.4974528482528018e-05, + "loss": 0.2445, + "step": 6768 + }, + { + "epoch": 2.83, + "learning_rate": 1.4973057659068601e-05, + "loss": 0.2655, + "step": 6769 + }, + { + "epoch": 2.83, + "learning_rate": 1.4971586692663055e-05, + "loss": 0.2913, + "step": 6770 + }, + { + "epoch": 2.83, + "learning_rate": 1.4970115583353661e-05, + "loss": 0.2701, + "step": 6771 + }, + { + "epoch": 2.84, + "learning_rate": 1.496864433118271e-05, + "loss": 0.263, + "step": 6772 + }, + { + "epoch": 2.84, + "learning_rate": 1.496717293619248e-05, + "loss": 0.2663, + "step": 6773 + }, + { + "epoch": 2.84, + "learning_rate": 1.4965701398425275e-05, + "loss": 0.2575, + "step": 6774 + }, + { + "epoch": 2.84, + "learning_rate": 1.496422971792339e-05, + "loss": 0.2518, + "step": 6775 + }, + { + "epoch": 2.84, + "learning_rate": 1.4962757894729128e-05, + "loss": 0.2548, + "step": 6776 + }, + { + "epoch": 2.84, + "learning_rate": 1.496128592888479e-05, + "loss": 0.2347, + "step": 6777 + }, + { + "epoch": 2.84, + "learning_rate": 1.4959813820432693e-05, + "loss": 0.2599, + "step": 6778 + }, + { + "epoch": 2.84, + "learning_rate": 1.4958341569415149e-05, + "loss": 0.2783, + "step": 6779 + }, + { + "epoch": 2.84, + "learning_rate": 1.4956869175874475e-05, + "loss": 0.2505, + "step": 6780 + }, + { + "epoch": 2.84, + "learning_rate": 1.4955396639852996e-05, + "loss": 0.2879, + "step": 6781 + }, + { + "epoch": 2.84, + "learning_rate": 1.4953923961393043e-05, + "loss": 0.251, + "step": 6782 + }, + { + "epoch": 2.84, + "learning_rate": 1.4952451140536934e-05, + "loss": 0.2481, + "step": 6783 + }, + { + "epoch": 2.84, + "learning_rate": 1.4950978177327016e-05, + "loss": 0.2704, + "step": 6784 + }, + { + "epoch": 2.84, + "learning_rate": 1.4949505071805624e-05, + "loss": 0.2769, + "step": 6785 + }, + { + "epoch": 2.84, + "learning_rate": 1.4948031824015098e-05, + "loss": 0.2836, + "step": 6786 + }, + { + "epoch": 2.84, + "learning_rate": 1.4946558433997792e-05, + "loss": 0.2574, + "step": 6787 + }, + { + "epoch": 2.84, + "learning_rate": 1.4945084901796048e-05, + "loss": 0.2546, + "step": 6788 + }, + { + "epoch": 2.84, + "learning_rate": 1.494361122745223e-05, + "loss": 0.2575, + "step": 6789 + }, + { + "epoch": 2.84, + "learning_rate": 1.4942137411008696e-05, + "loss": 0.2564, + "step": 6790 + }, + { + "epoch": 2.84, + "learning_rate": 1.4940663452507802e-05, + "loss": 0.2227, + "step": 6791 + }, + { + "epoch": 2.84, + "learning_rate": 1.4939189351991928e-05, + "loss": 0.2171, + "step": 6792 + }, + { + "epoch": 2.84, + "learning_rate": 1.4937715109503434e-05, + "loss": 0.2628, + "step": 6793 + }, + { + "epoch": 2.84, + "learning_rate": 1.4936240725084701e-05, + "loss": 0.256, + "step": 6794 + }, + { + "epoch": 2.84, + "learning_rate": 1.4934766198778111e-05, + "loss": 0.2534, + "step": 6795 + }, + { + "epoch": 2.85, + "learning_rate": 1.4933291530626047e-05, + "loss": 0.279, + "step": 6796 + }, + { + "epoch": 2.85, + "learning_rate": 1.493181672067089e-05, + "loss": 0.2554, + "step": 6797 + }, + { + "epoch": 2.85, + "learning_rate": 1.4930341768955043e-05, + "loss": 0.233, + "step": 6798 + }, + { + "epoch": 2.85, + "learning_rate": 1.4928866675520897e-05, + "loss": 0.2738, + "step": 6799 + }, + { + "epoch": 2.85, + "learning_rate": 1.492739144041085e-05, + "loss": 0.2688, + "step": 6800 + }, + { + "epoch": 2.85, + "learning_rate": 1.4925916063667309e-05, + "loss": 0.2855, + "step": 6801 + }, + { + "epoch": 2.85, + "learning_rate": 1.4924440545332681e-05, + "loss": 0.2513, + "step": 6802 + }, + { + "epoch": 2.85, + "learning_rate": 1.4922964885449382e-05, + "loss": 0.2448, + "step": 6803 + }, + { + "epoch": 2.85, + "learning_rate": 1.4921489084059823e-05, + "loss": 0.2518, + "step": 6804 + }, + { + "epoch": 2.85, + "learning_rate": 1.4920013141206428e-05, + "loss": 0.2839, + "step": 6805 + }, + { + "epoch": 2.85, + "learning_rate": 1.4918537056931623e-05, + "loss": 0.2599, + "step": 6806 + }, + { + "epoch": 2.85, + "learning_rate": 1.4917060831277835e-05, + "loss": 0.2528, + "step": 6807 + }, + { + "epoch": 2.85, + "learning_rate": 1.4915584464287494e-05, + "loss": 0.2628, + "step": 6808 + }, + { + "epoch": 2.85, + "learning_rate": 1.4914107956003044e-05, + "loss": 0.2673, + "step": 6809 + }, + { + "epoch": 2.85, + "learning_rate": 1.4912631306466918e-05, + "loss": 0.243, + "step": 6810 + }, + { + "epoch": 2.85, + "learning_rate": 1.4911154515721564e-05, + "loss": 0.2547, + "step": 6811 + }, + { + "epoch": 2.85, + "learning_rate": 1.4909677583809435e-05, + "loss": 0.2695, + "step": 6812 + }, + { + "epoch": 2.85, + "learning_rate": 1.4908200510772979e-05, + "loss": 0.2457, + "step": 6813 + }, + { + "epoch": 2.85, + "learning_rate": 1.4906723296654652e-05, + "loss": 0.2445, + "step": 6814 + }, + { + "epoch": 2.85, + "learning_rate": 1.490524594149692e-05, + "loss": 0.2652, + "step": 6815 + }, + { + "epoch": 2.85, + "learning_rate": 1.4903768445342248e-05, + "loss": 0.2757, + "step": 6816 + }, + { + "epoch": 2.85, + "learning_rate": 1.4902290808233103e-05, + "loss": 0.2322, + "step": 6817 + }, + { + "epoch": 2.85, + "learning_rate": 1.4900813030211956e-05, + "loss": 0.2429, + "step": 6818 + }, + { + "epoch": 2.85, + "learning_rate": 1.4899335111321289e-05, + "loss": 0.2251, + "step": 6819 + }, + { + "epoch": 2.86, + "learning_rate": 1.489785705160358e-05, + "loss": 0.2629, + "step": 6820 + }, + { + "epoch": 2.86, + "learning_rate": 1.4896378851101321e-05, + "loss": 0.2627, + "step": 6821 + }, + { + "epoch": 2.86, + "learning_rate": 1.4894900509856995e-05, + "loss": 0.2647, + "step": 6822 + }, + { + "epoch": 2.86, + "learning_rate": 1.4893422027913099e-05, + "loss": 0.2637, + "step": 6823 + }, + { + "epoch": 2.86, + "learning_rate": 1.4891943405312126e-05, + "loss": 0.25, + "step": 6824 + }, + { + "epoch": 2.86, + "learning_rate": 1.489046464209658e-05, + "loss": 0.2694, + "step": 6825 + }, + { + "epoch": 2.86, + "learning_rate": 1.488898573830897e-05, + "loss": 0.2741, + "step": 6826 + }, + { + "epoch": 2.86, + "learning_rate": 1.4887506693991803e-05, + "loss": 0.269, + "step": 6827 + }, + { + "epoch": 2.86, + "learning_rate": 1.4886027509187593e-05, + "loss": 0.2546, + "step": 6828 + }, + { + "epoch": 2.86, + "learning_rate": 1.4884548183938857e-05, + "loss": 0.2611, + "step": 6829 + }, + { + "epoch": 2.86, + "learning_rate": 1.4883068718288121e-05, + "loss": 0.2603, + "step": 6830 + }, + { + "epoch": 2.86, + "learning_rate": 1.4881589112277903e-05, + "loss": 0.2771, + "step": 6831 + }, + { + "epoch": 2.86, + "learning_rate": 1.4880109365950744e-05, + "loss": 0.2664, + "step": 6832 + }, + { + "epoch": 2.86, + "learning_rate": 1.4878629479349167e-05, + "loss": 0.2743, + "step": 6833 + }, + { + "epoch": 2.86, + "learning_rate": 1.4877149452515718e-05, + "loss": 0.2375, + "step": 6834 + }, + { + "epoch": 2.86, + "learning_rate": 1.4875669285492933e-05, + "loss": 0.2422, + "step": 6835 + }, + { + "epoch": 2.86, + "learning_rate": 1.487418897832336e-05, + "loss": 0.2433, + "step": 6836 + }, + { + "epoch": 2.86, + "learning_rate": 1.4872708531049554e-05, + "loss": 0.2557, + "step": 6837 + }, + { + "epoch": 2.86, + "learning_rate": 1.4871227943714063e-05, + "loss": 0.2563, + "step": 6838 + }, + { + "epoch": 2.86, + "learning_rate": 1.4869747216359446e-05, + "loss": 0.2676, + "step": 6839 + }, + { + "epoch": 2.86, + "learning_rate": 1.486826634902827e-05, + "loss": 0.2543, + "step": 6840 + }, + { + "epoch": 2.86, + "learning_rate": 1.4866785341763097e-05, + "loss": 0.2639, + "step": 6841 + }, + { + "epoch": 2.86, + "learning_rate": 1.4865304194606496e-05, + "loss": 0.2607, + "step": 6842 + }, + { + "epoch": 2.86, + "learning_rate": 1.4863822907601046e-05, + "loss": 0.2511, + "step": 6843 + }, + { + "epoch": 2.87, + "learning_rate": 1.4862341480789322e-05, + "loss": 0.2529, + "step": 6844 + }, + { + "epoch": 2.87, + "learning_rate": 1.4860859914213905e-05, + "loss": 0.2363, + "step": 6845 + }, + { + "epoch": 2.87, + "learning_rate": 1.4859378207917385e-05, + "loss": 0.2423, + "step": 6846 + }, + { + "epoch": 2.87, + "learning_rate": 1.485789636194235e-05, + "loss": 0.2616, + "step": 6847 + }, + { + "epoch": 2.87, + "learning_rate": 1.4856414376331394e-05, + "loss": 0.264, + "step": 6848 + }, + { + "epoch": 2.87, + "learning_rate": 1.4854932251127117e-05, + "loss": 0.2839, + "step": 6849 + }, + { + "epoch": 2.87, + "learning_rate": 1.4853449986372122e-05, + "loss": 0.253, + "step": 6850 + }, + { + "epoch": 2.87, + "learning_rate": 1.4851967582109013e-05, + "loss": 0.288, + "step": 6851 + }, + { + "epoch": 2.87, + "learning_rate": 1.48504850383804e-05, + "loss": 0.2654, + "step": 6852 + }, + { + "epoch": 2.87, + "learning_rate": 1.4849002355228901e-05, + "loss": 0.2839, + "step": 6853 + }, + { + "epoch": 2.87, + "learning_rate": 1.484751953269713e-05, + "loss": 0.2706, + "step": 6854 + }, + { + "epoch": 2.87, + "learning_rate": 1.4846036570827714e-05, + "loss": 0.2648, + "step": 6855 + }, + { + "epoch": 2.87, + "learning_rate": 1.4844553469663274e-05, + "loss": 0.2387, + "step": 6856 + }, + { + "epoch": 2.87, + "learning_rate": 1.4843070229246448e-05, + "loss": 0.2518, + "step": 6857 + }, + { + "epoch": 2.87, + "learning_rate": 1.4841586849619862e-05, + "loss": 0.2505, + "step": 6858 + }, + { + "epoch": 2.87, + "learning_rate": 1.4840103330826158e-05, + "loss": 0.2374, + "step": 6859 + }, + { + "epoch": 2.87, + "learning_rate": 1.4838619672907984e-05, + "loss": 0.2631, + "step": 6860 + }, + { + "epoch": 2.87, + "learning_rate": 1.4837135875907978e-05, + "loss": 0.2686, + "step": 6861 + }, + { + "epoch": 2.87, + "learning_rate": 1.4835651939868792e-05, + "loss": 0.2546, + "step": 6862 + }, + { + "epoch": 2.87, + "learning_rate": 1.4834167864833085e-05, + "loss": 0.2985, + "step": 6863 + }, + { + "epoch": 2.87, + "learning_rate": 1.483268365084351e-05, + "loss": 0.2512, + "step": 6864 + }, + { + "epoch": 2.87, + "learning_rate": 1.4831199297942731e-05, + "loss": 0.2642, + "step": 6865 + }, + { + "epoch": 2.87, + "learning_rate": 1.482971480617342e-05, + "loss": 0.2639, + "step": 6866 + }, + { + "epoch": 2.88, + "learning_rate": 1.482823017557824e-05, + "loss": 0.2675, + "step": 6867 + }, + { + "epoch": 2.88, + "learning_rate": 1.4826745406199865e-05, + "loss": 0.2623, + "step": 6868 + }, + { + "epoch": 2.88, + "learning_rate": 1.482526049808098e-05, + "loss": 0.2545, + "step": 6869 + }, + { + "epoch": 2.88, + "learning_rate": 1.4823775451264262e-05, + "loss": 0.2215, + "step": 6870 + }, + { + "epoch": 2.88, + "learning_rate": 1.4822290265792403e-05, + "loss": 0.229, + "step": 6871 + }, + { + "epoch": 2.88, + "learning_rate": 1.4820804941708086e-05, + "loss": 0.2672, + "step": 6872 + }, + { + "epoch": 2.88, + "learning_rate": 1.4819319479054009e-05, + "loss": 0.2455, + "step": 6873 + }, + { + "epoch": 2.88, + "learning_rate": 1.481783387787287e-05, + "loss": 0.2786, + "step": 6874 + }, + { + "epoch": 2.88, + "learning_rate": 1.4816348138207373e-05, + "loss": 0.2453, + "step": 6875 + }, + { + "epoch": 2.88, + "learning_rate": 1.4814862260100222e-05, + "loss": 0.2477, + "step": 6876 + }, + { + "epoch": 2.88, + "learning_rate": 1.4813376243594128e-05, + "loss": 0.275, + "step": 6877 + }, + { + "epoch": 2.88, + "learning_rate": 1.4811890088731808e-05, + "loss": 0.2415, + "step": 6878 + }, + { + "epoch": 2.88, + "learning_rate": 1.4810403795555975e-05, + "loss": 0.2582, + "step": 6879 + }, + { + "epoch": 2.88, + "learning_rate": 1.4808917364109352e-05, + "loss": 0.2525, + "step": 6880 + }, + { + "epoch": 2.88, + "learning_rate": 1.480743079443467e-05, + "loss": 0.2591, + "step": 6881 + }, + { + "epoch": 2.88, + "learning_rate": 1.4805944086574653e-05, + "loss": 0.2261, + "step": 6882 + }, + { + "epoch": 2.88, + "learning_rate": 1.4804457240572037e-05, + "loss": 0.2787, + "step": 6883 + }, + { + "epoch": 2.88, + "learning_rate": 1.4802970256469564e-05, + "loss": 0.2833, + "step": 6884 + }, + { + "epoch": 2.88, + "learning_rate": 1.4801483134309971e-05, + "loss": 0.2346, + "step": 6885 + }, + { + "epoch": 2.88, + "learning_rate": 1.4799995874136006e-05, + "loss": 0.2583, + "step": 6886 + }, + { + "epoch": 2.88, + "learning_rate": 1.4798508475990421e-05, + "loss": 0.2704, + "step": 6887 + }, + { + "epoch": 2.88, + "learning_rate": 1.4797020939915967e-05, + "loss": 0.2709, + "step": 6888 + }, + { + "epoch": 2.88, + "learning_rate": 1.4795533265955404e-05, + "loss": 0.267, + "step": 6889 + }, + { + "epoch": 2.88, + "learning_rate": 1.4794045454151491e-05, + "loss": 0.2466, + "step": 6890 + }, + { + "epoch": 2.89, + "learning_rate": 1.4792557504546997e-05, + "loss": 0.2482, + "step": 6891 + }, + { + "epoch": 2.89, + "learning_rate": 1.4791069417184689e-05, + "loss": 0.2537, + "step": 6892 + }, + { + "epoch": 2.89, + "learning_rate": 1.4789581192107343e-05, + "loss": 0.2675, + "step": 6893 + }, + { + "epoch": 2.89, + "learning_rate": 1.4788092829357736e-05, + "loss": 0.2571, + "step": 6894 + }, + { + "epoch": 2.89, + "learning_rate": 1.4786604328978648e-05, + "loss": 0.267, + "step": 6895 + }, + { + "epoch": 2.89, + "learning_rate": 1.4785115691012866e-05, + "loss": 0.2748, + "step": 6896 + }, + { + "epoch": 2.89, + "learning_rate": 1.4783626915503182e-05, + "loss": 0.2911, + "step": 6897 + }, + { + "epoch": 2.89, + "learning_rate": 1.4782138002492385e-05, + "loss": 0.2753, + "step": 6898 + }, + { + "epoch": 2.89, + "learning_rate": 1.4780648952023276e-05, + "loss": 0.245, + "step": 6899 + }, + { + "epoch": 2.89, + "learning_rate": 1.4779159764138656e-05, + "loss": 0.2457, + "step": 6900 + }, + { + "epoch": 2.89, + "learning_rate": 1.4777670438881328e-05, + "loss": 0.2733, + "step": 6901 + }, + { + "epoch": 2.89, + "learning_rate": 1.4776180976294105e-05, + "loss": 0.2523, + "step": 6902 + }, + { + "epoch": 2.89, + "learning_rate": 1.4774691376419795e-05, + "loss": 0.2498, + "step": 6903 + }, + { + "epoch": 2.89, + "learning_rate": 1.4773201639301223e-05, + "loss": 0.2705, + "step": 6904 + }, + { + "epoch": 2.89, + "learning_rate": 1.47717117649812e-05, + "loss": 0.2358, + "step": 6905 + }, + { + "epoch": 2.89, + "learning_rate": 1.4770221753502558e-05, + "loss": 0.2867, + "step": 6906 + }, + { + "epoch": 2.89, + "learning_rate": 1.4768731604908125e-05, + "loss": 0.2619, + "step": 6907 + }, + { + "epoch": 2.89, + "learning_rate": 1.4767241319240735e-05, + "loss": 0.2755, + "step": 6908 + }, + { + "epoch": 2.89, + "learning_rate": 1.4765750896543222e-05, + "loss": 0.2694, + "step": 6909 + }, + { + "epoch": 2.89, + "learning_rate": 1.476426033685843e-05, + "loss": 0.294, + "step": 6910 + }, + { + "epoch": 2.89, + "learning_rate": 1.4762769640229202e-05, + "loss": 0.235, + "step": 6911 + }, + { + "epoch": 2.89, + "learning_rate": 1.4761278806698389e-05, + "loss": 0.2529, + "step": 6912 + }, + { + "epoch": 2.89, + "learning_rate": 1.475978783630884e-05, + "loss": 0.2626, + "step": 6913 + }, + { + "epoch": 2.89, + "learning_rate": 1.4758296729103413e-05, + "loss": 0.2572, + "step": 6914 + }, + { + "epoch": 2.9, + "learning_rate": 1.475680548512497e-05, + "loss": 0.2899, + "step": 6915 + }, + { + "epoch": 2.9, + "learning_rate": 1.4755314104416371e-05, + "loss": 0.2459, + "step": 6916 + }, + { + "epoch": 2.9, + "learning_rate": 1.4753822587020491e-05, + "loss": 0.255, + "step": 6917 + }, + { + "epoch": 2.9, + "learning_rate": 1.47523309329802e-05, + "loss": 0.2422, + "step": 6918 + }, + { + "epoch": 2.9, + "learning_rate": 1.4750839142338372e-05, + "loss": 0.2726, + "step": 6919 + }, + { + "epoch": 2.9, + "learning_rate": 1.4749347215137888e-05, + "loss": 0.2683, + "step": 6920 + }, + { + "epoch": 2.9, + "learning_rate": 1.4747855151421636e-05, + "loss": 0.2625, + "step": 6921 + }, + { + "epoch": 2.9, + "learning_rate": 1.4746362951232499e-05, + "loss": 0.253, + "step": 6922 + }, + { + "epoch": 2.9, + "learning_rate": 1.4744870614613369e-05, + "loss": 0.2732, + "step": 6923 + }, + { + "epoch": 2.9, + "learning_rate": 1.4743378141607146e-05, + "loss": 0.2718, + "step": 6924 + }, + { + "epoch": 2.9, + "learning_rate": 1.4741885532256726e-05, + "loss": 0.2532, + "step": 6925 + }, + { + "epoch": 2.9, + "learning_rate": 1.4740392786605014e-05, + "loss": 0.2521, + "step": 6926 + }, + { + "epoch": 2.9, + "learning_rate": 1.473889990469492e-05, + "loss": 0.2389, + "step": 6927 + }, + { + "epoch": 2.9, + "learning_rate": 1.4737406886569352e-05, + "loss": 0.2739, + "step": 6928 + }, + { + "epoch": 2.9, + "learning_rate": 1.4735913732271226e-05, + "loss": 0.2551, + "step": 6929 + }, + { + "epoch": 2.9, + "learning_rate": 1.4734420441843462e-05, + "loss": 0.2653, + "step": 6930 + }, + { + "epoch": 2.9, + "learning_rate": 1.4732927015328989e-05, + "loss": 0.2595, + "step": 6931 + }, + { + "epoch": 2.9, + "learning_rate": 1.4731433452770723e-05, + "loss": 0.2651, + "step": 6932 + }, + { + "epoch": 2.9, + "learning_rate": 1.4729939754211602e-05, + "loss": 0.2988, + "step": 6933 + }, + { + "epoch": 2.9, + "learning_rate": 1.4728445919694563e-05, + "loss": 0.2706, + "step": 6934 + }, + { + "epoch": 2.9, + "learning_rate": 1.472695194926254e-05, + "loss": 0.2431, + "step": 6935 + }, + { + "epoch": 2.9, + "learning_rate": 1.4725457842958477e-05, + "loss": 0.2552, + "step": 6936 + }, + { + "epoch": 2.9, + "learning_rate": 1.4723963600825322e-05, + "loss": 0.2553, + "step": 6937 + }, + { + "epoch": 2.9, + "learning_rate": 1.472246922290603e-05, + "loss": 0.2611, + "step": 6938 + }, + { + "epoch": 2.91, + "learning_rate": 1.4720974709243542e-05, + "loss": 0.2267, + "step": 6939 + }, + { + "epoch": 2.91, + "learning_rate": 1.471948005988083e-05, + "loss": 0.2719, + "step": 6940 + }, + { + "epoch": 2.91, + "learning_rate": 1.471798527486085e-05, + "loss": 0.2622, + "step": 6941 + }, + { + "epoch": 2.91, + "learning_rate": 1.4716490354226573e-05, + "loss": 0.2586, + "step": 6942 + }, + { + "epoch": 2.91, + "learning_rate": 1.4714995298020963e-05, + "loss": 0.2677, + "step": 6943 + }, + { + "epoch": 2.91, + "learning_rate": 1.4713500106287e-05, + "loss": 0.2544, + "step": 6944 + }, + { + "epoch": 2.91, + "learning_rate": 1.4712004779067659e-05, + "loss": 0.2792, + "step": 6945 + }, + { + "epoch": 2.91, + "learning_rate": 1.471050931640592e-05, + "loss": 0.243, + "step": 6946 + }, + { + "epoch": 2.91, + "learning_rate": 1.4709013718344773e-05, + "loss": 0.2542, + "step": 6947 + }, + { + "epoch": 2.91, + "learning_rate": 1.4707517984927201e-05, + "loss": 0.2478, + "step": 6948 + }, + { + "epoch": 2.91, + "learning_rate": 1.4706022116196208e-05, + "loss": 0.2705, + "step": 6949 + }, + { + "epoch": 2.91, + "learning_rate": 1.4704526112194781e-05, + "loss": 0.2619, + "step": 6950 + }, + { + "epoch": 2.91, + "learning_rate": 1.4703029972965927e-05, + "loss": 0.2677, + "step": 6951 + }, + { + "epoch": 2.91, + "learning_rate": 1.470153369855265e-05, + "loss": 0.2332, + "step": 6952 + }, + { + "epoch": 2.91, + "learning_rate": 1.470003728899796e-05, + "loss": 0.2818, + "step": 6953 + }, + { + "epoch": 2.91, + "learning_rate": 1.4698540744344864e-05, + "loss": 0.2309, + "step": 6954 + }, + { + "epoch": 2.91, + "learning_rate": 1.469704406463639e-05, + "loss": 0.2381, + "step": 6955 + }, + { + "epoch": 2.91, + "learning_rate": 1.469554724991555e-05, + "loss": 0.2561, + "step": 6956 + }, + { + "epoch": 2.91, + "learning_rate": 1.469405030022537e-05, + "loss": 0.2776, + "step": 6957 + }, + { + "epoch": 2.91, + "learning_rate": 1.4692553215608878e-05, + "loss": 0.2784, + "step": 6958 + }, + { + "epoch": 2.91, + "learning_rate": 1.4691055996109112e-05, + "loss": 0.265, + "step": 6959 + }, + { + "epoch": 2.91, + "learning_rate": 1.4689558641769101e-05, + "loss": 0.2791, + "step": 6960 + }, + { + "epoch": 2.91, + "learning_rate": 1.4688061152631889e-05, + "loss": 0.2317, + "step": 6961 + }, + { + "epoch": 2.91, + "learning_rate": 1.468656352874052e-05, + "loss": 0.2672, + "step": 6962 + }, + { + "epoch": 2.92, + "learning_rate": 1.4685065770138043e-05, + "loss": 0.2632, + "step": 6963 + }, + { + "epoch": 2.92, + "learning_rate": 1.4683567876867503e-05, + "loss": 0.272, + "step": 6964 + }, + { + "epoch": 2.92, + "learning_rate": 1.4682069848971966e-05, + "loss": 0.2292, + "step": 6965 + }, + { + "epoch": 2.92, + "learning_rate": 1.4680571686494483e-05, + "loss": 0.2913, + "step": 6966 + }, + { + "epoch": 2.92, + "learning_rate": 1.467907338947812e-05, + "loss": 0.2553, + "step": 6967 + }, + { + "epoch": 2.92, + "learning_rate": 1.4677574957965946e-05, + "loss": 0.2323, + "step": 6968 + }, + { + "epoch": 2.92, + "learning_rate": 1.4676076392001032e-05, + "loss": 0.2707, + "step": 6969 + }, + { + "epoch": 2.92, + "learning_rate": 1.4674577691626448e-05, + "loss": 0.2586, + "step": 6970 + }, + { + "epoch": 2.92, + "learning_rate": 1.4673078856885277e-05, + "loss": 0.2988, + "step": 6971 + }, + { + "epoch": 2.92, + "learning_rate": 1.4671579887820602e-05, + "loss": 0.2444, + "step": 6972 + }, + { + "epoch": 2.92, + "learning_rate": 1.467008078447551e-05, + "loss": 0.2504, + "step": 6973 + }, + { + "epoch": 2.92, + "learning_rate": 1.4668581546893086e-05, + "loss": 0.2456, + "step": 6974 + }, + { + "epoch": 2.92, + "learning_rate": 1.4667082175116431e-05, + "loss": 0.2501, + "step": 6975 + }, + { + "epoch": 2.92, + "learning_rate": 1.466558266918864e-05, + "loss": 0.2867, + "step": 6976 + }, + { + "epoch": 2.92, + "learning_rate": 1.4664083029152815e-05, + "loss": 0.2356, + "step": 6977 + }, + { + "epoch": 2.92, + "learning_rate": 1.4662583255052064e-05, + "loss": 0.2852, + "step": 6978 + }, + { + "epoch": 2.92, + "learning_rate": 1.4661083346929492e-05, + "loss": 0.2831, + "step": 6979 + }, + { + "epoch": 2.92, + "learning_rate": 1.465958330482822e-05, + "loss": 0.2642, + "step": 6980 + }, + { + "epoch": 2.92, + "learning_rate": 1.4658083128791356e-05, + "loss": 0.251, + "step": 6981 + }, + { + "epoch": 2.92, + "learning_rate": 1.4656582818862028e-05, + "loss": 0.3004, + "step": 6982 + }, + { + "epoch": 2.92, + "learning_rate": 1.465508237508336e-05, + "loss": 0.2536, + "step": 6983 + }, + { + "epoch": 2.92, + "learning_rate": 1.4653581797498477e-05, + "loss": 0.2684, + "step": 6984 + }, + { + "epoch": 2.92, + "learning_rate": 1.4652081086150516e-05, + "loss": 0.2621, + "step": 6985 + }, + { + "epoch": 2.92, + "learning_rate": 1.4650580241082613e-05, + "loss": 0.2872, + "step": 6986 + }, + { + "epoch": 2.93, + "learning_rate": 1.4649079262337909e-05, + "loss": 0.2488, + "step": 6987 + }, + { + "epoch": 2.93, + "learning_rate": 1.4647578149959545e-05, + "loss": 0.2557, + "step": 6988 + }, + { + "epoch": 2.93, + "learning_rate": 1.4646076903990671e-05, + "loss": 0.2869, + "step": 6989 + }, + { + "epoch": 2.93, + "learning_rate": 1.4644575524474441e-05, + "loss": 0.2733, + "step": 6990 + }, + { + "epoch": 2.93, + "learning_rate": 1.4643074011454007e-05, + "loss": 0.2676, + "step": 6991 + }, + { + "epoch": 2.93, + "learning_rate": 1.4641572364972532e-05, + "loss": 0.2571, + "step": 6992 + }, + { + "epoch": 2.93, + "learning_rate": 1.464007058507318e-05, + "loss": 0.2744, + "step": 6993 + }, + { + "epoch": 2.93, + "learning_rate": 1.4638568671799113e-05, + "loss": 0.2485, + "step": 6994 + }, + { + "epoch": 2.93, + "learning_rate": 1.4637066625193508e-05, + "loss": 0.2437, + "step": 6995 + }, + { + "epoch": 2.93, + "learning_rate": 1.463556444529954e-05, + "loss": 0.2428, + "step": 6996 + }, + { + "epoch": 2.93, + "learning_rate": 1.4634062132160381e-05, + "loss": 0.2526, + "step": 6997 + }, + { + "epoch": 2.93, + "learning_rate": 1.4632559685819219e-05, + "loss": 0.2696, + "step": 6998 + }, + { + "epoch": 2.93, + "learning_rate": 1.4631057106319242e-05, + "loss": 0.2625, + "step": 6999 + }, + { + "epoch": 2.93, + "learning_rate": 1.4629554393703635e-05, + "loss": 0.259, + "step": 7000 + }, + { + "epoch": 2.93, + "learning_rate": 1.4628051548015596e-05, + "loss": 0.2856, + "step": 7001 + }, + { + "epoch": 2.93, + "learning_rate": 1.4626548569298324e-05, + "loss": 0.2357, + "step": 7002 + }, + { + "epoch": 2.93, + "learning_rate": 1.462504545759502e-05, + "loss": 0.2855, + "step": 7003 + }, + { + "epoch": 2.93, + "learning_rate": 1.4623542212948887e-05, + "loss": 0.2517, + "step": 7004 + }, + { + "epoch": 2.93, + "learning_rate": 1.4622038835403135e-05, + "loss": 0.2457, + "step": 7005 + }, + { + "epoch": 2.93, + "learning_rate": 1.4620535325000978e-05, + "loss": 0.2696, + "step": 7006 + }, + { + "epoch": 2.93, + "learning_rate": 1.4619031681785633e-05, + "loss": 0.2551, + "step": 7007 + }, + { + "epoch": 2.93, + "learning_rate": 1.4617527905800321e-05, + "loss": 0.2464, + "step": 7008 + }, + { + "epoch": 2.93, + "learning_rate": 1.461602399708827e-05, + "loss": 0.2891, + "step": 7009 + }, + { + "epoch": 2.93, + "learning_rate": 1.4614519955692701e-05, + "loss": 0.2391, + "step": 7010 + }, + { + "epoch": 2.94, + "learning_rate": 1.4613015781656854e-05, + "loss": 0.2525, + "step": 7011 + }, + { + "epoch": 2.94, + "learning_rate": 1.4611511475023958e-05, + "loss": 0.2371, + "step": 7012 + }, + { + "epoch": 2.94, + "learning_rate": 1.461000703583726e-05, + "loss": 0.2491, + "step": 7013 + }, + { + "epoch": 2.94, + "learning_rate": 1.4608502464140002e-05, + "loss": 0.2961, + "step": 7014 + }, + { + "epoch": 2.94, + "learning_rate": 1.4606997759975427e-05, + "loss": 0.2514, + "step": 7015 + }, + { + "epoch": 2.94, + "learning_rate": 1.4605492923386795e-05, + "loss": 0.2443, + "step": 7016 + }, + { + "epoch": 2.94, + "learning_rate": 1.4603987954417349e-05, + "loss": 0.2637, + "step": 7017 + }, + { + "epoch": 2.94, + "learning_rate": 1.4602482853110358e-05, + "loss": 0.2675, + "step": 7018 + }, + { + "epoch": 2.94, + "learning_rate": 1.460097761950908e-05, + "loss": 0.2693, + "step": 7019 + }, + { + "epoch": 2.94, + "learning_rate": 1.4599472253656786e-05, + "loss": 0.2458, + "step": 7020 + }, + { + "epoch": 2.94, + "learning_rate": 1.459796675559674e-05, + "loss": 0.2845, + "step": 7021 + }, + { + "epoch": 2.94, + "learning_rate": 1.4596461125372221e-05, + "loss": 0.2674, + "step": 7022 + }, + { + "epoch": 2.94, + "learning_rate": 1.4594955363026509e-05, + "loss": 0.2461, + "step": 7023 + }, + { + "epoch": 2.94, + "learning_rate": 1.4593449468602878e-05, + "loss": 0.2608, + "step": 7024 + }, + { + "epoch": 2.94, + "learning_rate": 1.459194344214462e-05, + "loss": 0.25, + "step": 7025 + }, + { + "epoch": 2.94, + "learning_rate": 1.4590437283695022e-05, + "loss": 0.2608, + "step": 7026 + }, + { + "epoch": 2.94, + "learning_rate": 1.4588930993297381e-05, + "loss": 0.2597, + "step": 7027 + }, + { + "epoch": 2.94, + "learning_rate": 1.4587424570994989e-05, + "loss": 0.2485, + "step": 7028 + }, + { + "epoch": 2.94, + "learning_rate": 1.4585918016831144e-05, + "loss": 0.3006, + "step": 7029 + }, + { + "epoch": 2.94, + "learning_rate": 1.4584411330849162e-05, + "loss": 0.2837, + "step": 7030 + }, + { + "epoch": 2.94, + "learning_rate": 1.458290451309234e-05, + "loss": 0.243, + "step": 7031 + }, + { + "epoch": 2.94, + "learning_rate": 1.4581397563603994e-05, + "loss": 0.2504, + "step": 7032 + }, + { + "epoch": 2.94, + "learning_rate": 1.4579890482427446e-05, + "loss": 0.27, + "step": 7033 + }, + { + "epoch": 2.94, + "learning_rate": 1.4578383269606004e-05, + "loss": 0.228, + "step": 7034 + }, + { + "epoch": 2.95, + "learning_rate": 1.4576875925182998e-05, + "loss": 0.2386, + "step": 7035 + }, + { + "epoch": 2.95, + "learning_rate": 1.4575368449201759e-05, + "loss": 0.2873, + "step": 7036 + }, + { + "epoch": 2.95, + "learning_rate": 1.4573860841705613e-05, + "loss": 0.2577, + "step": 7037 + }, + { + "epoch": 2.95, + "learning_rate": 1.4572353102737894e-05, + "loss": 0.2532, + "step": 7038 + }, + { + "epoch": 2.95, + "learning_rate": 1.4570845232341943e-05, + "loss": 0.249, + "step": 7039 + }, + { + "epoch": 2.95, + "learning_rate": 1.4569337230561105e-05, + "loss": 0.2475, + "step": 7040 + }, + { + "epoch": 2.95, + "learning_rate": 1.4567829097438719e-05, + "loss": 0.2634, + "step": 7041 + }, + { + "epoch": 2.95, + "learning_rate": 1.456632083301814e-05, + "loss": 0.2866, + "step": 7042 + }, + { + "epoch": 2.95, + "learning_rate": 1.4564812437342724e-05, + "loss": 0.2312, + "step": 7043 + }, + { + "epoch": 2.95, + "learning_rate": 1.4563303910455821e-05, + "loss": 0.2826, + "step": 7044 + }, + { + "epoch": 2.95, + "learning_rate": 1.4561795252400797e-05, + "loss": 0.2687, + "step": 7045 + }, + { + "epoch": 2.95, + "learning_rate": 1.4560286463221018e-05, + "loss": 0.2409, + "step": 7046 + }, + { + "epoch": 2.95, + "learning_rate": 1.455877754295985e-05, + "loss": 0.2738, + "step": 7047 + }, + { + "epoch": 2.95, + "learning_rate": 1.455726849166067e-05, + "loss": 0.2755, + "step": 7048 + }, + { + "epoch": 2.95, + "learning_rate": 1.4555759309366848e-05, + "loss": 0.2729, + "step": 7049 + }, + { + "epoch": 2.95, + "learning_rate": 1.455424999612177e-05, + "loss": 0.2863, + "step": 7050 + }, + { + "epoch": 2.95, + "learning_rate": 1.4552740551968817e-05, + "loss": 0.2536, + "step": 7051 + }, + { + "epoch": 2.95, + "learning_rate": 1.4551230976951373e-05, + "loss": 0.2727, + "step": 7052 + }, + { + "epoch": 2.95, + "learning_rate": 1.4549721271112835e-05, + "loss": 0.2709, + "step": 7053 + }, + { + "epoch": 2.95, + "learning_rate": 1.4548211434496598e-05, + "loss": 0.2484, + "step": 7054 + }, + { + "epoch": 2.95, + "learning_rate": 1.4546701467146058e-05, + "loss": 0.2855, + "step": 7055 + }, + { + "epoch": 2.95, + "learning_rate": 1.4545191369104618e-05, + "loss": 0.265, + "step": 7056 + }, + { + "epoch": 2.95, + "learning_rate": 1.4543681140415688e-05, + "loss": 0.2499, + "step": 7057 + }, + { + "epoch": 2.95, + "learning_rate": 1.4542170781122673e-05, + "loss": 0.26, + "step": 7058 + }, + { + "epoch": 2.96, + "learning_rate": 1.4540660291268993e-05, + "loss": 0.2607, + "step": 7059 + }, + { + "epoch": 2.96, + "learning_rate": 1.453914967089806e-05, + "loss": 0.2486, + "step": 7060 + }, + { + "epoch": 2.96, + "learning_rate": 1.4537638920053298e-05, + "loss": 0.3044, + "step": 7061 + }, + { + "epoch": 2.96, + "learning_rate": 1.4536128038778132e-05, + "loss": 0.2689, + "step": 7062 + }, + { + "epoch": 2.96, + "learning_rate": 1.453461702711599e-05, + "loss": 0.2797, + "step": 7063 + }, + { + "epoch": 2.96, + "learning_rate": 1.4533105885110307e-05, + "loss": 0.2443, + "step": 7064 + }, + { + "epoch": 2.96, + "learning_rate": 1.4531594612804516e-05, + "loss": 0.2563, + "step": 7065 + }, + { + "epoch": 2.96, + "learning_rate": 1.4530083210242062e-05, + "loss": 0.2787, + "step": 7066 + }, + { + "epoch": 2.96, + "learning_rate": 1.4528571677466383e-05, + "loss": 0.2638, + "step": 7067 + }, + { + "epoch": 2.96, + "learning_rate": 1.4527060014520932e-05, + "loss": 0.2708, + "step": 7068 + }, + { + "epoch": 2.96, + "learning_rate": 1.4525548221449156e-05, + "loss": 0.235, + "step": 7069 + }, + { + "epoch": 2.96, + "learning_rate": 1.4524036298294515e-05, + "loss": 0.2447, + "step": 7070 + }, + { + "epoch": 2.96, + "learning_rate": 1.4522524245100465e-05, + "loss": 0.2411, + "step": 7071 + }, + { + "epoch": 2.96, + "learning_rate": 1.4521012061910466e-05, + "loss": 0.26, + "step": 7072 + }, + { + "epoch": 2.96, + "learning_rate": 1.4519499748767989e-05, + "loss": 0.2866, + "step": 7073 + }, + { + "epoch": 2.96, + "learning_rate": 1.4517987305716502e-05, + "loss": 0.2783, + "step": 7074 + }, + { + "epoch": 2.96, + "learning_rate": 1.4516474732799476e-05, + "loss": 0.2697, + "step": 7075 + }, + { + "epoch": 2.96, + "learning_rate": 1.4514962030060395e-05, + "loss": 0.2484, + "step": 7076 + }, + { + "epoch": 2.96, + "learning_rate": 1.451344919754274e-05, + "loss": 0.2751, + "step": 7077 + }, + { + "epoch": 2.96, + "learning_rate": 1.4511936235289987e-05, + "loss": 0.2818, + "step": 7078 + }, + { + "epoch": 2.96, + "learning_rate": 1.451042314334563e-05, + "loss": 0.2503, + "step": 7079 + }, + { + "epoch": 2.96, + "learning_rate": 1.450890992175317e-05, + "loss": 0.2401, + "step": 7080 + }, + { + "epoch": 2.96, + "learning_rate": 1.450739657055609e-05, + "loss": 0.2721, + "step": 7081 + }, + { + "epoch": 2.97, + "learning_rate": 1.4505883089797896e-05, + "loss": 0.2605, + "step": 7082 + }, + { + "epoch": 2.97, + "learning_rate": 1.4504369479522092e-05, + "loss": 0.2714, + "step": 7083 + }, + { + "epoch": 2.97, + "learning_rate": 1.4502855739772184e-05, + "loss": 0.2754, + "step": 7084 + }, + { + "epoch": 2.97, + "learning_rate": 1.4501341870591683e-05, + "loss": 0.2411, + "step": 7085 + }, + { + "epoch": 2.97, + "learning_rate": 1.4499827872024104e-05, + "loss": 0.2623, + "step": 7086 + }, + { + "epoch": 2.97, + "learning_rate": 1.4498313744112966e-05, + "loss": 0.2632, + "step": 7087 + }, + { + "epoch": 2.97, + "learning_rate": 1.4496799486901791e-05, + "loss": 0.2663, + "step": 7088 + }, + { + "epoch": 2.97, + "learning_rate": 1.4495285100434107e-05, + "loss": 0.2793, + "step": 7089 + }, + { + "epoch": 2.97, + "learning_rate": 1.449377058475344e-05, + "loss": 0.2695, + "step": 7090 + }, + { + "epoch": 2.97, + "learning_rate": 1.4492255939903326e-05, + "loss": 0.2576, + "step": 7091 + }, + { + "epoch": 2.97, + "learning_rate": 1.44907411659273e-05, + "loss": 0.2428, + "step": 7092 + }, + { + "epoch": 2.97, + "learning_rate": 1.4489226262868906e-05, + "loss": 0.2305, + "step": 7093 + }, + { + "epoch": 2.97, + "learning_rate": 1.4487711230771686e-05, + "loss": 0.2953, + "step": 7094 + }, + { + "epoch": 2.97, + "learning_rate": 1.4486196069679191e-05, + "loss": 0.2662, + "step": 7095 + }, + { + "epoch": 2.97, + "learning_rate": 1.4484680779634968e-05, + "loss": 0.2287, + "step": 7096 + }, + { + "epoch": 2.97, + "learning_rate": 1.4483165360682577e-05, + "loss": 0.2736, + "step": 7097 + }, + { + "epoch": 2.97, + "learning_rate": 1.4481649812865574e-05, + "loss": 0.2746, + "step": 7098 + }, + { + "epoch": 2.97, + "learning_rate": 1.4480134136227526e-05, + "loss": 0.2411, + "step": 7099 + }, + { + "epoch": 2.97, + "learning_rate": 1.4478618330811997e-05, + "loss": 0.2587, + "step": 7100 + }, + { + "epoch": 2.97, + "learning_rate": 1.447710239666256e-05, + "loss": 0.2382, + "step": 7101 + }, + { + "epoch": 2.97, + "learning_rate": 1.4475586333822787e-05, + "loss": 0.2577, + "step": 7102 + }, + { + "epoch": 2.97, + "learning_rate": 1.4474070142336254e-05, + "loss": 0.2714, + "step": 7103 + }, + { + "epoch": 2.97, + "learning_rate": 1.4472553822246548e-05, + "loss": 0.2527, + "step": 7104 + }, + { + "epoch": 2.97, + "learning_rate": 1.4471037373597251e-05, + "loss": 0.292, + "step": 7105 + }, + { + "epoch": 2.98, + "learning_rate": 1.446952079643195e-05, + "loss": 0.2405, + "step": 7106 + }, + { + "epoch": 2.98, + "learning_rate": 1.4468004090794241e-05, + "loss": 0.2724, + "step": 7107 + }, + { + "epoch": 2.98, + "learning_rate": 1.4466487256727721e-05, + "loss": 0.2636, + "step": 7108 + }, + { + "epoch": 2.98, + "learning_rate": 1.4464970294275986e-05, + "loss": 0.2888, + "step": 7109 + }, + { + "epoch": 2.98, + "learning_rate": 1.4463453203482643e-05, + "loss": 0.2672, + "step": 7110 + }, + { + "epoch": 2.98, + "learning_rate": 1.44619359843913e-05, + "loss": 0.2623, + "step": 7111 + }, + { + "epoch": 2.98, + "learning_rate": 1.4460418637045564e-05, + "loss": 0.2394, + "step": 7112 + }, + { + "epoch": 2.98, + "learning_rate": 1.4458901161489051e-05, + "loss": 0.2543, + "step": 7113 + }, + { + "epoch": 2.98, + "learning_rate": 1.4457383557765385e-05, + "loss": 0.2451, + "step": 7114 + }, + { + "epoch": 2.98, + "learning_rate": 1.4455865825918182e-05, + "loss": 0.2542, + "step": 7115 + }, + { + "epoch": 2.98, + "learning_rate": 1.445434796599107e-05, + "loss": 0.2598, + "step": 7116 + }, + { + "epoch": 2.98, + "learning_rate": 1.4452829978027678e-05, + "loss": 0.2718, + "step": 7117 + }, + { + "epoch": 2.98, + "learning_rate": 1.4451311862071641e-05, + "loss": 0.2405, + "step": 7118 + }, + { + "epoch": 2.98, + "learning_rate": 1.4449793618166594e-05, + "loss": 0.2497, + "step": 7119 + }, + { + "epoch": 2.98, + "learning_rate": 1.4448275246356175e-05, + "loss": 0.2675, + "step": 7120 + }, + { + "epoch": 2.98, + "learning_rate": 1.4446756746684036e-05, + "loss": 0.2779, + "step": 7121 + }, + { + "epoch": 2.98, + "learning_rate": 1.4445238119193818e-05, + "loss": 0.2645, + "step": 7122 + }, + { + "epoch": 2.98, + "learning_rate": 1.4443719363929173e-05, + "loss": 0.2663, + "step": 7123 + }, + { + "epoch": 2.98, + "learning_rate": 1.4442200480933761e-05, + "loss": 0.2581, + "step": 7124 + }, + { + "epoch": 2.98, + "learning_rate": 1.4440681470251239e-05, + "loss": 0.2753, + "step": 7125 + }, + { + "epoch": 2.98, + "learning_rate": 1.4439162331925268e-05, + "loss": 0.2406, + "step": 7126 + }, + { + "epoch": 2.98, + "learning_rate": 1.4437643065999516e-05, + "loss": 0.2575, + "step": 7127 + }, + { + "epoch": 2.98, + "learning_rate": 1.443612367251765e-05, + "loss": 0.2494, + "step": 7128 + }, + { + "epoch": 2.98, + "learning_rate": 1.4434604151523345e-05, + "loss": 0.2552, + "step": 7129 + }, + { + "epoch": 2.99, + "learning_rate": 1.443308450306028e-05, + "loss": 0.2535, + "step": 7130 + }, + { + "epoch": 2.99, + "learning_rate": 1.4431564727172135e-05, + "loss": 0.2476, + "step": 7131 + }, + { + "epoch": 2.99, + "learning_rate": 1.4430044823902594e-05, + "loss": 0.2381, + "step": 7132 + }, + { + "epoch": 2.99, + "learning_rate": 1.4428524793295344e-05, + "loss": 0.2544, + "step": 7133 + }, + { + "epoch": 2.99, + "learning_rate": 1.442700463539408e-05, + "loss": 0.269, + "step": 7134 + }, + { + "epoch": 2.99, + "learning_rate": 1.4425484350242498e-05, + "loss": 0.24, + "step": 7135 + }, + { + "epoch": 2.99, + "learning_rate": 1.4423963937884293e-05, + "loss": 0.2364, + "step": 7136 + }, + { + "epoch": 2.99, + "learning_rate": 1.442244339836317e-05, + "loss": 0.2322, + "step": 7137 + }, + { + "epoch": 2.99, + "learning_rate": 1.442092273172284e-05, + "loss": 0.2413, + "step": 7138 + }, + { + "epoch": 2.99, + "learning_rate": 1.4419401938007007e-05, + "loss": 0.2707, + "step": 7139 + }, + { + "epoch": 2.99, + "learning_rate": 1.4417881017259385e-05, + "loss": 0.3087, + "step": 7140 + }, + { + "epoch": 2.99, + "learning_rate": 1.4416359969523694e-05, + "loss": 0.265, + "step": 7141 + }, + { + "epoch": 2.99, + "learning_rate": 1.4414838794843658e-05, + "loss": 0.2298, + "step": 7142 + }, + { + "epoch": 2.99, + "learning_rate": 1.4413317493262996e-05, + "loss": 0.2708, + "step": 7143 + }, + { + "epoch": 2.99, + "learning_rate": 1.4411796064825436e-05, + "loss": 0.2349, + "step": 7144 + }, + { + "epoch": 2.99, + "learning_rate": 1.4410274509574718e-05, + "loss": 0.2678, + "step": 7145 + }, + { + "epoch": 2.99, + "learning_rate": 1.4408752827554571e-05, + "loss": 0.2931, + "step": 7146 + }, + { + "epoch": 2.99, + "learning_rate": 1.4407231018808734e-05, + "loss": 0.2673, + "step": 7147 + }, + { + "epoch": 2.99, + "learning_rate": 1.4405709083380957e-05, + "loss": 0.2571, + "step": 7148 + }, + { + "epoch": 2.99, + "learning_rate": 1.4404187021314982e-05, + "loss": 0.2492, + "step": 7149 + }, + { + "epoch": 2.99, + "learning_rate": 1.4402664832654555e-05, + "loss": 0.2863, + "step": 7150 + }, + { + "epoch": 2.99, + "learning_rate": 1.4401142517443434e-05, + "loss": 0.2501, + "step": 7151 + }, + { + "epoch": 2.99, + "learning_rate": 1.439962007572538e-05, + "loss": 0.2647, + "step": 7152 + }, + { + "epoch": 2.99, + "learning_rate": 1.4398097507544152e-05, + "loss": 0.2374, + "step": 7153 + }, + { + "epoch": 3.0, + "learning_rate": 1.439657481294351e-05, + "loss": 0.2462, + "step": 7154 + }, + { + "epoch": 3.0, + "learning_rate": 1.4395051991967228e-05, + "loss": 0.2512, + "step": 7155 + }, + { + "epoch": 3.0, + "learning_rate": 1.4393529044659077e-05, + "loss": 0.2685, + "step": 7156 + }, + { + "epoch": 3.0, + "learning_rate": 1.439200597106283e-05, + "loss": 0.2732, + "step": 7157 + }, + { + "epoch": 3.0, + "learning_rate": 1.4390482771222273e-05, + "loss": 0.2834, + "step": 7158 + }, + { + "epoch": 3.0, + "learning_rate": 1.4388959445181181e-05, + "loss": 0.2718, + "step": 7159 + }, + { + "epoch": 3.0, + "learning_rate": 1.4387435992983347e-05, + "loss": 0.2553, + "step": 7160 + }, + { + "epoch": 3.0, + "learning_rate": 1.4385912414672559e-05, + "loss": 0.23, + "step": 7161 + }, + { + "epoch": 3.0, + "learning_rate": 1.438438871029261e-05, + "loss": 0.2289, + "step": 7162 + }, + { + "epoch": 3.0, + "learning_rate": 1.4382864879887298e-05, + "loss": 0.2558, + "step": 7163 + }, + { + "epoch": 3.0, + "learning_rate": 1.4381340923500422e-05, + "loss": 0.2601, + "step": 7164 + }, + { + "epoch": 3.0, + "learning_rate": 1.437981684117579e-05, + "loss": 0.2692, + "step": 7165 + }, + { + "epoch": 3.0, + "learning_rate": 1.4378292632957214e-05, + "loss": 0.2586, + "step": 7166 + }, + { + "epoch": 3.0, + "learning_rate": 1.4376768298888498e-05, + "loss": 0.1887, + "step": 7167 + }, + { + "epoch": 3.0, + "learning_rate": 1.4375243839013458e-05, + "loss": 0.2059, + "step": 7168 + }, + { + "epoch": 3.0, + "learning_rate": 1.4373719253375921e-05, + "loss": 0.231, + "step": 7169 + }, + { + "epoch": 3.0, + "learning_rate": 1.4372194542019705e-05, + "loss": 0.1802, + "step": 7170 + }, + { + "epoch": 3.0, + "learning_rate": 1.4370669704988634e-05, + "loss": 0.1511, + "step": 7171 + }, + { + "epoch": 3.0, + "learning_rate": 1.4369144742326544e-05, + "loss": 0.1502, + "step": 7172 + }, + { + "epoch": 3.0, + "learning_rate": 1.4367619654077264e-05, + "loss": 0.1817, + "step": 7173 + }, + { + "epoch": 3.0, + "learning_rate": 1.4366094440284632e-05, + "loss": 0.1603, + "step": 7174 + }, + { + "epoch": 3.0, + "learning_rate": 1.4364569100992489e-05, + "loss": 0.2057, + "step": 7175 + }, + { + "epoch": 3.0, + "learning_rate": 1.4363043636244682e-05, + "loss": 0.1826, + "step": 7176 + }, + { + "epoch": 3.0, + "learning_rate": 1.4361518046085057e-05, + "loss": 0.1632, + "step": 7177 + }, + { + "epoch": 3.01, + "learning_rate": 1.4359992330557467e-05, + "loss": 0.1868, + "step": 7178 + }, + { + "epoch": 3.01, + "learning_rate": 1.4358466489705766e-05, + "loss": 0.1837, + "step": 7179 + }, + { + "epoch": 3.01, + "learning_rate": 1.4356940523573812e-05, + "loss": 0.2101, + "step": 7180 + }, + { + "epoch": 3.01, + "learning_rate": 1.4355414432205468e-05, + "loss": 0.1869, + "step": 7181 + }, + { + "epoch": 3.01, + "learning_rate": 1.4353888215644604e-05, + "loss": 0.2015, + "step": 7182 + }, + { + "epoch": 3.01, + "learning_rate": 1.4352361873935088e-05, + "loss": 0.1817, + "step": 7183 + }, + { + "epoch": 3.01, + "learning_rate": 1.435083540712079e-05, + "loss": 0.2035, + "step": 7184 + }, + { + "epoch": 3.01, + "learning_rate": 1.4349308815245586e-05, + "loss": 0.1981, + "step": 7185 + }, + { + "epoch": 3.01, + "learning_rate": 1.4347782098353364e-05, + "loss": 0.1769, + "step": 7186 + }, + { + "epoch": 3.01, + "learning_rate": 1.4346255256488002e-05, + "loss": 0.1934, + "step": 7187 + }, + { + "epoch": 3.01, + "learning_rate": 1.4344728289693391e-05, + "loss": 0.1702, + "step": 7188 + }, + { + "epoch": 3.01, + "learning_rate": 1.434320119801342e-05, + "loss": 0.177, + "step": 7189 + }, + { + "epoch": 3.01, + "learning_rate": 1.4341673981491982e-05, + "loss": 0.1806, + "step": 7190 + }, + { + "epoch": 3.01, + "learning_rate": 1.4340146640172981e-05, + "loss": 0.2014, + "step": 7191 + }, + { + "epoch": 3.01, + "learning_rate": 1.4338619174100315e-05, + "loss": 0.1654, + "step": 7192 + }, + { + "epoch": 3.01, + "learning_rate": 1.433709158331789e-05, + "loss": 0.1999, + "step": 7193 + }, + { + "epoch": 3.01, + "learning_rate": 1.4335563867869618e-05, + "loss": 0.1922, + "step": 7194 + }, + { + "epoch": 3.01, + "learning_rate": 1.433403602779941e-05, + "loss": 0.1618, + "step": 7195 + }, + { + "epoch": 3.01, + "learning_rate": 1.4332508063151184e-05, + "loss": 0.1569, + "step": 7196 + }, + { + "epoch": 3.01, + "learning_rate": 1.4330979973968856e-05, + "loss": 0.2257, + "step": 7197 + }, + { + "epoch": 3.01, + "learning_rate": 1.4329451760296352e-05, + "loss": 0.1733, + "step": 7198 + }, + { + "epoch": 3.01, + "learning_rate": 1.43279234221776e-05, + "loss": 0.2231, + "step": 7199 + }, + { + "epoch": 3.01, + "learning_rate": 1.4326394959656527e-05, + "loss": 0.1967, + "step": 7200 + }, + { + "epoch": 3.01, + "learning_rate": 1.4324866372777072e-05, + "loss": 0.2143, + "step": 7201 + }, + { + "epoch": 3.02, + "learning_rate": 1.4323337661583168e-05, + "loss": 0.1895, + "step": 7202 + }, + { + "epoch": 3.02, + "learning_rate": 1.4321808826118764e-05, + "loss": 0.2105, + "step": 7203 + }, + { + "epoch": 3.02, + "learning_rate": 1.4320279866427798e-05, + "loss": 0.1458, + "step": 7204 + }, + { + "epoch": 3.02, + "learning_rate": 1.431875078255422e-05, + "loss": 0.1854, + "step": 7205 + }, + { + "epoch": 3.02, + "learning_rate": 1.4317221574541985e-05, + "loss": 0.1838, + "step": 7206 + }, + { + "epoch": 3.02, + "learning_rate": 1.4315692242435048e-05, + "loss": 0.1533, + "step": 7207 + }, + { + "epoch": 3.02, + "learning_rate": 1.4314162786277366e-05, + "loss": 0.1994, + "step": 7208 + }, + { + "epoch": 3.02, + "learning_rate": 1.43126332061129e-05, + "loss": 0.2194, + "step": 7209 + }, + { + "epoch": 3.02, + "learning_rate": 1.4311103501985625e-05, + "loss": 0.169, + "step": 7210 + }, + { + "epoch": 3.02, + "learning_rate": 1.4309573673939503e-05, + "loss": 0.2104, + "step": 7211 + }, + { + "epoch": 3.02, + "learning_rate": 1.4308043722018508e-05, + "loss": 0.2042, + "step": 7212 + }, + { + "epoch": 3.02, + "learning_rate": 1.4306513646266621e-05, + "loss": 0.1725, + "step": 7213 + }, + { + "epoch": 3.02, + "learning_rate": 1.430498344672782e-05, + "loss": 0.1701, + "step": 7214 + }, + { + "epoch": 3.02, + "learning_rate": 1.430345312344609e-05, + "loss": 0.1718, + "step": 7215 + }, + { + "epoch": 3.02, + "learning_rate": 1.430192267646542e-05, + "loss": 0.1763, + "step": 7216 + }, + { + "epoch": 3.02, + "learning_rate": 1.4300392105829802e-05, + "loss": 0.17, + "step": 7217 + }, + { + "epoch": 3.02, + "learning_rate": 1.4298861411583224e-05, + "loss": 0.1941, + "step": 7218 + }, + { + "epoch": 3.02, + "learning_rate": 1.4297330593769691e-05, + "loss": 0.2126, + "step": 7219 + }, + { + "epoch": 3.02, + "learning_rate": 1.4295799652433206e-05, + "loss": 0.1735, + "step": 7220 + }, + { + "epoch": 3.02, + "learning_rate": 1.429426858761777e-05, + "loss": 0.1709, + "step": 7221 + }, + { + "epoch": 3.02, + "learning_rate": 1.4292737399367394e-05, + "loss": 0.1972, + "step": 7222 + }, + { + "epoch": 3.02, + "learning_rate": 1.429120608772609e-05, + "loss": 0.19, + "step": 7223 + }, + { + "epoch": 3.02, + "learning_rate": 1.4289674652737877e-05, + "loss": 0.1853, + "step": 7224 + }, + { + "epoch": 3.02, + "learning_rate": 1.4288143094446771e-05, + "loss": 0.1501, + "step": 7225 + }, + { + "epoch": 3.03, + "learning_rate": 1.42866114128968e-05, + "loss": 0.1467, + "step": 7226 + }, + { + "epoch": 3.03, + "learning_rate": 1.4285079608131984e-05, + "loss": 0.148, + "step": 7227 + }, + { + "epoch": 3.03, + "learning_rate": 1.4283547680196359e-05, + "loss": 0.2309, + "step": 7228 + }, + { + "epoch": 3.03, + "learning_rate": 1.4282015629133958e-05, + "loss": 0.2142, + "step": 7229 + }, + { + "epoch": 3.03, + "learning_rate": 1.4280483454988816e-05, + "loss": 0.2085, + "step": 7230 + }, + { + "epoch": 3.03, + "learning_rate": 1.4278951157804977e-05, + "loss": 0.1577, + "step": 7231 + }, + { + "epoch": 3.03, + "learning_rate": 1.4277418737626483e-05, + "loss": 0.1826, + "step": 7232 + }, + { + "epoch": 3.03, + "learning_rate": 1.4275886194497384e-05, + "loss": 0.2294, + "step": 7233 + }, + { + "epoch": 3.03, + "learning_rate": 1.4274353528461728e-05, + "loss": 0.1659, + "step": 7234 + }, + { + "epoch": 3.03, + "learning_rate": 1.4272820739563575e-05, + "loss": 0.226, + "step": 7235 + }, + { + "epoch": 3.03, + "learning_rate": 1.427128782784698e-05, + "loss": 0.1884, + "step": 7236 + }, + { + "epoch": 3.03, + "learning_rate": 1.4269754793356008e-05, + "loss": 0.1751, + "step": 7237 + }, + { + "epoch": 3.03, + "learning_rate": 1.4268221636134722e-05, + "loss": 0.2024, + "step": 7238 + }, + { + "epoch": 3.03, + "learning_rate": 1.4266688356227194e-05, + "loss": 0.1908, + "step": 7239 + }, + { + "epoch": 3.03, + "learning_rate": 1.4265154953677498e-05, + "loss": 0.1581, + "step": 7240 + }, + { + "epoch": 3.03, + "learning_rate": 1.4263621428529703e-05, + "loss": 0.1458, + "step": 7241 + }, + { + "epoch": 3.03, + "learning_rate": 1.4262087780827896e-05, + "loss": 0.18, + "step": 7242 + }, + { + "epoch": 3.03, + "learning_rate": 1.4260554010616156e-05, + "loss": 0.1969, + "step": 7243 + }, + { + "epoch": 3.03, + "learning_rate": 1.4259020117938574e-05, + "loss": 0.1862, + "step": 7244 + }, + { + "epoch": 3.03, + "learning_rate": 1.4257486102839236e-05, + "loss": 0.2063, + "step": 7245 + }, + { + "epoch": 3.03, + "learning_rate": 1.4255951965362238e-05, + "loss": 0.1843, + "step": 7246 + }, + { + "epoch": 3.03, + "learning_rate": 1.4254417705551679e-05, + "loss": 0.1448, + "step": 7247 + }, + { + "epoch": 3.03, + "learning_rate": 1.4252883323451658e-05, + "loss": 0.2114, + "step": 7248 + }, + { + "epoch": 3.03, + "learning_rate": 1.425134881910628e-05, + "loss": 0.2203, + "step": 7249 + }, + { + "epoch": 3.04, + "learning_rate": 1.4249814192559653e-05, + "loss": 0.1884, + "step": 7250 + }, + { + "epoch": 3.04, + "learning_rate": 1.4248279443855891e-05, + "loss": 0.1889, + "step": 7251 + }, + { + "epoch": 3.04, + "learning_rate": 1.42467445730391e-05, + "loss": 0.1804, + "step": 7252 + }, + { + "epoch": 3.04, + "learning_rate": 1.4245209580153408e-05, + "loss": 0.174, + "step": 7253 + }, + { + "epoch": 3.04, + "learning_rate": 1.4243674465242934e-05, + "loss": 0.1691, + "step": 7254 + }, + { + "epoch": 3.04, + "learning_rate": 1.42421392283518e-05, + "loss": 0.1507, + "step": 7255 + }, + { + "epoch": 3.04, + "learning_rate": 1.424060386952414e-05, + "loss": 0.1765, + "step": 7256 + }, + { + "epoch": 3.04, + "learning_rate": 1.4239068388804089e-05, + "loss": 0.1813, + "step": 7257 + }, + { + "epoch": 3.04, + "learning_rate": 1.4237532786235775e-05, + "loss": 0.1951, + "step": 7258 + }, + { + "epoch": 3.04, + "learning_rate": 1.423599706186334e-05, + "loss": 0.2148, + "step": 7259 + }, + { + "epoch": 3.04, + "learning_rate": 1.4234461215730932e-05, + "loss": 0.1641, + "step": 7260 + }, + { + "epoch": 3.04, + "learning_rate": 1.4232925247882692e-05, + "loss": 0.1737, + "step": 7261 + }, + { + "epoch": 3.04, + "learning_rate": 1.4231389158362771e-05, + "loss": 0.1844, + "step": 7262 + }, + { + "epoch": 3.04, + "learning_rate": 1.4229852947215326e-05, + "loss": 0.179, + "step": 7263 + }, + { + "epoch": 3.04, + "learning_rate": 1.4228316614484511e-05, + "loss": 0.1891, + "step": 7264 + }, + { + "epoch": 3.04, + "learning_rate": 1.4226780160214486e-05, + "loss": 0.166, + "step": 7265 + }, + { + "epoch": 3.04, + "learning_rate": 1.4225243584449415e-05, + "loss": 0.1638, + "step": 7266 + }, + { + "epoch": 3.04, + "learning_rate": 1.422370688723347e-05, + "loss": 0.2241, + "step": 7267 + }, + { + "epoch": 3.04, + "learning_rate": 1.4222170068610816e-05, + "loss": 0.2143, + "step": 7268 + }, + { + "epoch": 3.04, + "learning_rate": 1.4220633128625632e-05, + "loss": 0.2018, + "step": 7269 + }, + { + "epoch": 3.04, + "learning_rate": 1.4219096067322094e-05, + "loss": 0.1457, + "step": 7270 + }, + { + "epoch": 3.04, + "learning_rate": 1.421755888474438e-05, + "loss": 0.1974, + "step": 7271 + }, + { + "epoch": 3.04, + "learning_rate": 1.4216021580936681e-05, + "loss": 0.1919, + "step": 7272 + }, + { + "epoch": 3.05, + "learning_rate": 1.421448415594318e-05, + "loss": 0.2016, + "step": 7273 + }, + { + "epoch": 3.05, + "learning_rate": 1.4212946609808079e-05, + "loss": 0.1676, + "step": 7274 + }, + { + "epoch": 3.05, + "learning_rate": 1.421140894257556e-05, + "loss": 0.2124, + "step": 7275 + }, + { + "epoch": 3.05, + "learning_rate": 1.420987115428983e-05, + "loss": 0.2214, + "step": 7276 + }, + { + "epoch": 3.05, + "learning_rate": 1.4208333244995087e-05, + "loss": 0.1987, + "step": 7277 + }, + { + "epoch": 3.05, + "learning_rate": 1.4206795214735543e-05, + "loss": 0.1791, + "step": 7278 + }, + { + "epoch": 3.05, + "learning_rate": 1.42052570635554e-05, + "loss": 0.2003, + "step": 7279 + }, + { + "epoch": 3.05, + "learning_rate": 1.4203718791498877e-05, + "loss": 0.1513, + "step": 7280 + }, + { + "epoch": 3.05, + "learning_rate": 1.4202180398610188e-05, + "loss": 0.1887, + "step": 7281 + }, + { + "epoch": 3.05, + "learning_rate": 1.420064188493355e-05, + "loss": 0.2105, + "step": 7282 + }, + { + "epoch": 3.05, + "learning_rate": 1.4199103250513192e-05, + "loss": 0.1794, + "step": 7283 + }, + { + "epoch": 3.05, + "learning_rate": 1.4197564495393334e-05, + "loss": 0.1558, + "step": 7284 + }, + { + "epoch": 3.05, + "learning_rate": 1.4196025619618212e-05, + "loss": 0.1765, + "step": 7285 + }, + { + "epoch": 3.05, + "learning_rate": 1.4194486623232056e-05, + "loss": 0.1902, + "step": 7286 + }, + { + "epoch": 3.05, + "learning_rate": 1.4192947506279102e-05, + "loss": 0.1742, + "step": 7287 + }, + { + "epoch": 3.05, + "learning_rate": 1.4191408268803596e-05, + "loss": 0.1754, + "step": 7288 + }, + { + "epoch": 3.05, + "learning_rate": 1.4189868910849779e-05, + "loss": 0.1759, + "step": 7289 + }, + { + "epoch": 3.05, + "learning_rate": 1.4188329432461894e-05, + "loss": 0.1869, + "step": 7290 + }, + { + "epoch": 3.05, + "learning_rate": 1.41867898336842e-05, + "loss": 0.1341, + "step": 7291 + }, + { + "epoch": 3.05, + "learning_rate": 1.4185250114560945e-05, + "loss": 0.1637, + "step": 7292 + }, + { + "epoch": 3.05, + "learning_rate": 1.4183710275136391e-05, + "loss": 0.1879, + "step": 7293 + }, + { + "epoch": 3.05, + "learning_rate": 1.4182170315454798e-05, + "loss": 0.1744, + "step": 7294 + }, + { + "epoch": 3.05, + "learning_rate": 1.4180630235560428e-05, + "loss": 0.1949, + "step": 7295 + }, + { + "epoch": 3.05, + "learning_rate": 1.4179090035497554e-05, + "loss": 0.1792, + "step": 7296 + }, + { + "epoch": 3.06, + "learning_rate": 1.4177549715310445e-05, + "loss": 0.169, + "step": 7297 + }, + { + "epoch": 3.06, + "learning_rate": 1.4176009275043376e-05, + "loss": 0.1775, + "step": 7298 + }, + { + "epoch": 3.06, + "learning_rate": 1.4174468714740628e-05, + "loss": 0.1806, + "step": 7299 + }, + { + "epoch": 3.06, + "learning_rate": 1.4172928034446478e-05, + "loss": 0.217, + "step": 7300 + }, + { + "epoch": 3.06, + "learning_rate": 1.4171387234205219e-05, + "loss": 0.1439, + "step": 7301 + }, + { + "epoch": 3.06, + "learning_rate": 1.4169846314061132e-05, + "loss": 0.1562, + "step": 7302 + }, + { + "epoch": 3.06, + "learning_rate": 1.4168305274058511e-05, + "loss": 0.1626, + "step": 7303 + }, + { + "epoch": 3.06, + "learning_rate": 1.4166764114241657e-05, + "loss": 0.1897, + "step": 7304 + }, + { + "epoch": 3.06, + "learning_rate": 1.4165222834654867e-05, + "loss": 0.2121, + "step": 7305 + }, + { + "epoch": 3.06, + "learning_rate": 1.416368143534244e-05, + "loss": 0.1895, + "step": 7306 + }, + { + "epoch": 3.06, + "learning_rate": 1.4162139916348685e-05, + "loss": 0.2021, + "step": 7307 + }, + { + "epoch": 3.06, + "learning_rate": 1.4160598277717916e-05, + "loss": 0.1783, + "step": 7308 + }, + { + "epoch": 3.06, + "learning_rate": 1.4159056519494439e-05, + "loss": 0.2108, + "step": 7309 + }, + { + "epoch": 3.06, + "learning_rate": 1.4157514641722569e-05, + "loss": 0.2187, + "step": 7310 + }, + { + "epoch": 3.06, + "learning_rate": 1.4155972644446635e-05, + "loss": 0.1467, + "step": 7311 + }, + { + "epoch": 3.06, + "learning_rate": 1.4154430527710954e-05, + "loss": 0.1728, + "step": 7312 + }, + { + "epoch": 3.06, + "learning_rate": 1.4152888291559852e-05, + "loss": 0.1972, + "step": 7313 + }, + { + "epoch": 3.06, + "learning_rate": 1.4151345936037664e-05, + "loss": 0.2267, + "step": 7314 + }, + { + "epoch": 3.06, + "learning_rate": 1.414980346118872e-05, + "loss": 0.1653, + "step": 7315 + }, + { + "epoch": 3.06, + "learning_rate": 1.4148260867057357e-05, + "loss": 0.1404, + "step": 7316 + }, + { + "epoch": 3.06, + "learning_rate": 1.4146718153687917e-05, + "loss": 0.1793, + "step": 7317 + }, + { + "epoch": 3.06, + "learning_rate": 1.4145175321124744e-05, + "loss": 0.1372, + "step": 7318 + }, + { + "epoch": 3.06, + "learning_rate": 1.4143632369412185e-05, + "loss": 0.1705, + "step": 7319 + }, + { + "epoch": 3.06, + "learning_rate": 1.414208929859459e-05, + "loss": 0.2271, + "step": 7320 + }, + { + "epoch": 3.07, + "learning_rate": 1.4140546108716313e-05, + "loss": 0.1695, + "step": 7321 + }, + { + "epoch": 3.07, + "learning_rate": 1.4139002799821714e-05, + "loss": 0.1739, + "step": 7322 + }, + { + "epoch": 3.07, + "learning_rate": 1.4137459371955151e-05, + "loss": 0.2014, + "step": 7323 + }, + { + "epoch": 3.07, + "learning_rate": 1.413591582516099e-05, + "loss": 0.1908, + "step": 7324 + }, + { + "epoch": 3.07, + "learning_rate": 1.41343721594836e-05, + "loss": 0.1706, + "step": 7325 + }, + { + "epoch": 3.07, + "learning_rate": 1.4132828374967348e-05, + "loss": 0.1861, + "step": 7326 + }, + { + "epoch": 3.07, + "learning_rate": 1.4131284471656615e-05, + "loss": 0.1634, + "step": 7327 + }, + { + "epoch": 3.07, + "learning_rate": 1.4129740449595777e-05, + "loss": 0.1741, + "step": 7328 + }, + { + "epoch": 3.07, + "learning_rate": 1.4128196308829209e-05, + "loss": 0.1747, + "step": 7329 + }, + { + "epoch": 3.07, + "learning_rate": 1.4126652049401306e-05, + "loss": 0.1682, + "step": 7330 + }, + { + "epoch": 3.07, + "learning_rate": 1.4125107671356451e-05, + "loss": 0.19, + "step": 7331 + }, + { + "epoch": 3.07, + "learning_rate": 1.4123563174739036e-05, + "loss": 0.1916, + "step": 7332 + }, + { + "epoch": 3.07, + "learning_rate": 1.4122018559593457e-05, + "loss": 0.1645, + "step": 7333 + }, + { + "epoch": 3.07, + "learning_rate": 1.4120473825964112e-05, + "loss": 0.1863, + "step": 7334 + }, + { + "epoch": 3.07, + "learning_rate": 1.4118928973895406e-05, + "loss": 0.1731, + "step": 7335 + }, + { + "epoch": 3.07, + "learning_rate": 1.4117384003431738e-05, + "loss": 0.1718, + "step": 7336 + }, + { + "epoch": 3.07, + "learning_rate": 1.4115838914617518e-05, + "loss": 0.2038, + "step": 7337 + }, + { + "epoch": 3.07, + "learning_rate": 1.4114293707497168e-05, + "loss": 0.1613, + "step": 7338 + }, + { + "epoch": 3.07, + "learning_rate": 1.4112748382115091e-05, + "loss": 0.2105, + "step": 7339 + }, + { + "epoch": 3.07, + "learning_rate": 1.4111202938515711e-05, + "loss": 0.1805, + "step": 7340 + }, + { + "epoch": 3.07, + "learning_rate": 1.4109657376743455e-05, + "loss": 0.157, + "step": 7341 + }, + { + "epoch": 3.07, + "learning_rate": 1.4108111696842743e-05, + "loss": 0.2109, + "step": 7342 + }, + { + "epoch": 3.07, + "learning_rate": 1.4106565898858007e-05, + "loss": 0.1733, + "step": 7343 + }, + { + "epoch": 3.07, + "learning_rate": 1.4105019982833674e-05, + "loss": 0.1645, + "step": 7344 + }, + { + "epoch": 3.08, + "learning_rate": 1.4103473948814189e-05, + "loss": 0.1983, + "step": 7345 + }, + { + "epoch": 3.08, + "learning_rate": 1.4101927796843984e-05, + "loss": 0.1839, + "step": 7346 + }, + { + "epoch": 3.08, + "learning_rate": 1.4100381526967502e-05, + "loss": 0.2312, + "step": 7347 + }, + { + "epoch": 3.08, + "learning_rate": 1.4098835139229195e-05, + "loss": 0.1856, + "step": 7348 + }, + { + "epoch": 3.08, + "learning_rate": 1.409728863367351e-05, + "loss": 0.1912, + "step": 7349 + }, + { + "epoch": 3.08, + "learning_rate": 1.4095742010344898e-05, + "loss": 0.1981, + "step": 7350 + }, + { + "epoch": 3.08, + "learning_rate": 1.4094195269287814e-05, + "loss": 0.197, + "step": 7351 + }, + { + "epoch": 3.08, + "learning_rate": 1.4092648410546723e-05, + "loss": 0.1855, + "step": 7352 + }, + { + "epoch": 3.08, + "learning_rate": 1.4091101434166086e-05, + "loss": 0.1751, + "step": 7353 + }, + { + "epoch": 3.08, + "learning_rate": 1.4089554340190365e-05, + "loss": 0.1941, + "step": 7354 + }, + { + "epoch": 3.08, + "learning_rate": 1.4088007128664036e-05, + "loss": 0.1755, + "step": 7355 + }, + { + "epoch": 3.08, + "learning_rate": 1.4086459799631572e-05, + "loss": 0.1973, + "step": 7356 + }, + { + "epoch": 3.08, + "learning_rate": 1.4084912353137444e-05, + "loss": 0.1854, + "step": 7357 + }, + { + "epoch": 3.08, + "learning_rate": 1.4083364789226135e-05, + "loss": 0.2054, + "step": 7358 + }, + { + "epoch": 3.08, + "learning_rate": 1.4081817107942131e-05, + "loss": 0.1948, + "step": 7359 + }, + { + "epoch": 3.08, + "learning_rate": 1.4080269309329912e-05, + "loss": 0.157, + "step": 7360 + }, + { + "epoch": 3.08, + "learning_rate": 1.4078721393433977e-05, + "loss": 0.2108, + "step": 7361 + }, + { + "epoch": 3.08, + "learning_rate": 1.4077173360298812e-05, + "loss": 0.1796, + "step": 7362 + }, + { + "epoch": 3.08, + "learning_rate": 1.407562520996892e-05, + "loss": 0.2147, + "step": 7363 + }, + { + "epoch": 3.08, + "learning_rate": 1.4074076942488794e-05, + "loss": 0.2133, + "step": 7364 + }, + { + "epoch": 3.08, + "learning_rate": 1.4072528557902944e-05, + "loss": 0.1498, + "step": 7365 + }, + { + "epoch": 3.08, + "learning_rate": 1.4070980056255876e-05, + "loss": 0.2077, + "step": 7366 + }, + { + "epoch": 3.08, + "learning_rate": 1.4069431437592095e-05, + "loss": 0.2182, + "step": 7367 + }, + { + "epoch": 3.08, + "learning_rate": 1.4067882701956121e-05, + "loss": 0.2054, + "step": 7368 + }, + { + "epoch": 3.09, + "learning_rate": 1.406633384939247e-05, + "loss": 0.1672, + "step": 7369 + }, + { + "epoch": 3.09, + "learning_rate": 1.4064784879945658e-05, + "loss": 0.1746, + "step": 7370 + }, + { + "epoch": 3.09, + "learning_rate": 1.4063235793660212e-05, + "loss": 0.1901, + "step": 7371 + }, + { + "epoch": 3.09, + "learning_rate": 1.406168659058066e-05, + "loss": 0.1435, + "step": 7372 + }, + { + "epoch": 3.09, + "learning_rate": 1.4060137270751531e-05, + "loss": 0.2062, + "step": 7373 + }, + { + "epoch": 3.09, + "learning_rate": 1.4058587834217356e-05, + "loss": 0.1671, + "step": 7374 + }, + { + "epoch": 3.09, + "learning_rate": 1.405703828102268e-05, + "loss": 0.1781, + "step": 7375 + }, + { + "epoch": 3.09, + "learning_rate": 1.4055488611212038e-05, + "loss": 0.1978, + "step": 7376 + }, + { + "epoch": 3.09, + "learning_rate": 1.405393882482997e-05, + "loss": 0.1653, + "step": 7377 + }, + { + "epoch": 3.09, + "learning_rate": 1.4052388921921033e-05, + "loss": 0.1671, + "step": 7378 + }, + { + "epoch": 3.09, + "learning_rate": 1.405083890252977e-05, + "loss": 0.1926, + "step": 7379 + }, + { + "epoch": 3.09, + "learning_rate": 1.4049288766700739e-05, + "loss": 0.1676, + "step": 7380 + }, + { + "epoch": 3.09, + "learning_rate": 1.4047738514478493e-05, + "loss": 0.1606, + "step": 7381 + }, + { + "epoch": 3.09, + "learning_rate": 1.40461881459076e-05, + "loss": 0.1998, + "step": 7382 + }, + { + "epoch": 3.09, + "learning_rate": 1.4044637661032614e-05, + "loss": 0.2041, + "step": 7383 + }, + { + "epoch": 3.09, + "learning_rate": 1.404308705989811e-05, + "loss": 0.1909, + "step": 7384 + }, + { + "epoch": 3.09, + "learning_rate": 1.4041536342548656e-05, + "loss": 0.1531, + "step": 7385 + }, + { + "epoch": 3.09, + "learning_rate": 1.4039985509028825e-05, + "loss": 0.1778, + "step": 7386 + }, + { + "epoch": 3.09, + "learning_rate": 1.4038434559383198e-05, + "loss": 0.1942, + "step": 7387 + }, + { + "epoch": 3.09, + "learning_rate": 1.4036883493656353e-05, + "loss": 0.2008, + "step": 7388 + }, + { + "epoch": 3.09, + "learning_rate": 1.4035332311892873e-05, + "loss": 0.1685, + "step": 7389 + }, + { + "epoch": 3.09, + "learning_rate": 1.4033781014137348e-05, + "loss": 0.1667, + "step": 7390 + }, + { + "epoch": 3.09, + "learning_rate": 1.4032229600434363e-05, + "loss": 0.1642, + "step": 7391 + }, + { + "epoch": 3.09, + "learning_rate": 1.403067807082852e-05, + "loss": 0.169, + "step": 7392 + }, + { + "epoch": 3.1, + "learning_rate": 1.4029126425364414e-05, + "loss": 0.183, + "step": 7393 + }, + { + "epoch": 3.1, + "learning_rate": 1.4027574664086641e-05, + "loss": 0.1868, + "step": 7394 + }, + { + "epoch": 3.1, + "learning_rate": 1.402602278703981e-05, + "loss": 0.2027, + "step": 7395 + }, + { + "epoch": 3.1, + "learning_rate": 1.4024470794268528e-05, + "loss": 0.2004, + "step": 7396 + }, + { + "epoch": 3.1, + "learning_rate": 1.4022918685817399e-05, + "loss": 0.1987, + "step": 7397 + }, + { + "epoch": 3.1, + "learning_rate": 1.4021366461731047e-05, + "loss": 0.2168, + "step": 7398 + }, + { + "epoch": 3.1, + "learning_rate": 1.4019814122054082e-05, + "loss": 0.1792, + "step": 7399 + }, + { + "epoch": 3.1, + "learning_rate": 1.4018261666831132e-05, + "loss": 0.1826, + "step": 7400 + }, + { + "epoch": 3.1, + "learning_rate": 1.4016709096106813e-05, + "loss": 0.1979, + "step": 7401 + }, + { + "epoch": 3.1, + "learning_rate": 1.4015156409925754e-05, + "loss": 0.1863, + "step": 7402 + }, + { + "epoch": 3.1, + "learning_rate": 1.4013603608332589e-05, + "loss": 0.1953, + "step": 7403 + }, + { + "epoch": 3.1, + "learning_rate": 1.4012050691371948e-05, + "loss": 0.1579, + "step": 7404 + }, + { + "epoch": 3.1, + "learning_rate": 1.4010497659088468e-05, + "loss": 0.2044, + "step": 7405 + }, + { + "epoch": 3.1, + "learning_rate": 1.4008944511526798e-05, + "loss": 0.1969, + "step": 7406 + }, + { + "epoch": 3.1, + "learning_rate": 1.400739124873157e-05, + "loss": 0.1812, + "step": 7407 + }, + { + "epoch": 3.1, + "learning_rate": 1.4005837870747439e-05, + "loss": 0.2105, + "step": 7408 + }, + { + "epoch": 3.1, + "learning_rate": 1.4004284377619053e-05, + "loss": 0.2011, + "step": 7409 + }, + { + "epoch": 3.1, + "learning_rate": 1.4002730769391069e-05, + "loss": 0.1904, + "step": 7410 + }, + { + "epoch": 3.1, + "learning_rate": 1.4001177046108136e-05, + "loss": 0.1919, + "step": 7411 + }, + { + "epoch": 3.1, + "learning_rate": 1.3999623207814921e-05, + "loss": 0.1803, + "step": 7412 + }, + { + "epoch": 3.1, + "learning_rate": 1.3998069254556088e-05, + "loss": 0.1908, + "step": 7413 + }, + { + "epoch": 3.1, + "learning_rate": 1.3996515186376298e-05, + "loss": 0.22, + "step": 7414 + }, + { + "epoch": 3.1, + "learning_rate": 1.399496100332023e-05, + "loss": 0.174, + "step": 7415 + }, + { + "epoch": 3.1, + "learning_rate": 1.399340670543255e-05, + "loss": 0.2192, + "step": 7416 + }, + { + "epoch": 3.11, + "learning_rate": 1.3991852292757937e-05, + "loss": 0.1978, + "step": 7417 + }, + { + "epoch": 3.11, + "learning_rate": 1.3990297765341074e-05, + "loss": 0.1474, + "step": 7418 + }, + { + "epoch": 3.11, + "learning_rate": 1.3988743123226643e-05, + "loss": 0.1784, + "step": 7419 + }, + { + "epoch": 3.11, + "learning_rate": 1.398718836645933e-05, + "loss": 0.177, + "step": 7420 + }, + { + "epoch": 3.11, + "learning_rate": 1.3985633495083828e-05, + "loss": 0.2083, + "step": 7421 + }, + { + "epoch": 3.11, + "learning_rate": 1.3984078509144824e-05, + "loss": 0.2083, + "step": 7422 + }, + { + "epoch": 3.11, + "learning_rate": 1.398252340868702e-05, + "loss": 0.1653, + "step": 7423 + }, + { + "epoch": 3.11, + "learning_rate": 1.3980968193755115e-05, + "loss": 0.1788, + "step": 7424 + }, + { + "epoch": 3.11, + "learning_rate": 1.397941286439381e-05, + "loss": 0.17, + "step": 7425 + }, + { + "epoch": 3.11, + "learning_rate": 1.3977857420647816e-05, + "loss": 0.1635, + "step": 7426 + }, + { + "epoch": 3.11, + "learning_rate": 1.3976301862561841e-05, + "loss": 0.1704, + "step": 7427 + }, + { + "epoch": 3.11, + "learning_rate": 1.3974746190180595e-05, + "loss": 0.1724, + "step": 7428 + }, + { + "epoch": 3.11, + "learning_rate": 1.3973190403548798e-05, + "loss": 0.1812, + "step": 7429 + }, + { + "epoch": 3.11, + "learning_rate": 1.3971634502711169e-05, + "loss": 0.1525, + "step": 7430 + }, + { + "epoch": 3.11, + "learning_rate": 1.397007848771243e-05, + "loss": 0.1913, + "step": 7431 + }, + { + "epoch": 3.11, + "learning_rate": 1.3968522358597307e-05, + "loss": 0.1774, + "step": 7432 + }, + { + "epoch": 3.11, + "learning_rate": 1.3966966115410533e-05, + "loss": 0.1482, + "step": 7433 + }, + { + "epoch": 3.11, + "learning_rate": 1.3965409758196838e-05, + "loss": 0.2149, + "step": 7434 + }, + { + "epoch": 3.11, + "learning_rate": 1.3963853287000956e-05, + "loss": 0.1504, + "step": 7435 + }, + { + "epoch": 3.11, + "learning_rate": 1.396229670186763e-05, + "loss": 0.2362, + "step": 7436 + }, + { + "epoch": 3.11, + "learning_rate": 1.3960740002841602e-05, + "loss": 0.1958, + "step": 7437 + }, + { + "epoch": 3.11, + "learning_rate": 1.3959183189967615e-05, + "loss": 0.1642, + "step": 7438 + }, + { + "epoch": 3.11, + "learning_rate": 1.395762626329042e-05, + "loss": 0.1807, + "step": 7439 + }, + { + "epoch": 3.11, + "learning_rate": 1.3956069222854774e-05, + "loss": 0.1854, + "step": 7440 + }, + { + "epoch": 3.12, + "learning_rate": 1.3954512068705425e-05, + "loss": 0.2109, + "step": 7441 + }, + { + "epoch": 3.12, + "learning_rate": 1.3952954800887137e-05, + "loss": 0.1881, + "step": 7442 + }, + { + "epoch": 3.12, + "learning_rate": 1.3951397419444675e-05, + "loss": 0.1886, + "step": 7443 + }, + { + "epoch": 3.12, + "learning_rate": 1.3949839924422798e-05, + "loss": 0.1795, + "step": 7444 + }, + { + "epoch": 3.12, + "learning_rate": 1.3948282315866274e-05, + "loss": 0.2422, + "step": 7445 + }, + { + "epoch": 3.12, + "learning_rate": 1.394672459381988e-05, + "loss": 0.2021, + "step": 7446 + }, + { + "epoch": 3.12, + "learning_rate": 1.3945166758328394e-05, + "loss": 0.1772, + "step": 7447 + }, + { + "epoch": 3.12, + "learning_rate": 1.3943608809436587e-05, + "loss": 0.2052, + "step": 7448 + }, + { + "epoch": 3.12, + "learning_rate": 1.3942050747189242e-05, + "loss": 0.1767, + "step": 7449 + }, + { + "epoch": 3.12, + "learning_rate": 1.394049257163115e-05, + "loss": 0.1822, + "step": 7450 + }, + { + "epoch": 3.12, + "learning_rate": 1.3938934282807095e-05, + "loss": 0.1777, + "step": 7451 + }, + { + "epoch": 3.12, + "learning_rate": 1.3937375880761868e-05, + "loss": 0.1736, + "step": 7452 + }, + { + "epoch": 3.12, + "learning_rate": 1.3935817365540268e-05, + "loss": 0.1833, + "step": 7453 + }, + { + "epoch": 3.12, + "learning_rate": 1.3934258737187087e-05, + "loss": 0.2147, + "step": 7454 + }, + { + "epoch": 3.12, + "learning_rate": 1.3932699995747134e-05, + "loss": 0.1847, + "step": 7455 + }, + { + "epoch": 3.12, + "learning_rate": 1.3931141141265206e-05, + "loss": 0.1931, + "step": 7456 + }, + { + "epoch": 3.12, + "learning_rate": 1.3929582173786116e-05, + "loss": 0.15, + "step": 7457 + }, + { + "epoch": 3.12, + "learning_rate": 1.3928023093354671e-05, + "loss": 0.2478, + "step": 7458 + }, + { + "epoch": 3.12, + "learning_rate": 1.392646390001569e-05, + "loss": 0.158, + "step": 7459 + }, + { + "epoch": 3.12, + "learning_rate": 1.3924904593813986e-05, + "loss": 0.1903, + "step": 7460 + }, + { + "epoch": 3.12, + "learning_rate": 1.3923345174794386e-05, + "loss": 0.2119, + "step": 7461 + }, + { + "epoch": 3.12, + "learning_rate": 1.3921785643001708e-05, + "loss": 0.165, + "step": 7462 + }, + { + "epoch": 3.12, + "learning_rate": 1.3920225998480782e-05, + "loss": 0.1985, + "step": 7463 + }, + { + "epoch": 3.12, + "learning_rate": 1.3918666241276442e-05, + "loss": 0.1734, + "step": 7464 + }, + { + "epoch": 3.13, + "learning_rate": 1.3917106371433513e-05, + "loss": 0.202, + "step": 7465 + }, + { + "epoch": 3.13, + "learning_rate": 1.3915546388996838e-05, + "loss": 0.1719, + "step": 7466 + }, + { + "epoch": 3.13, + "learning_rate": 1.3913986294011261e-05, + "loss": 0.1945, + "step": 7467 + }, + { + "epoch": 3.13, + "learning_rate": 1.391242608652162e-05, + "loss": 0.1596, + "step": 7468 + }, + { + "epoch": 3.13, + "learning_rate": 1.391086576657276e-05, + "loss": 0.2132, + "step": 7469 + }, + { + "epoch": 3.13, + "learning_rate": 1.3909305334209536e-05, + "loss": 0.1928, + "step": 7470 + }, + { + "epoch": 3.13, + "learning_rate": 1.3907744789476799e-05, + "loss": 0.1837, + "step": 7471 + }, + { + "epoch": 3.13, + "learning_rate": 1.3906184132419405e-05, + "loss": 0.1827, + "step": 7472 + }, + { + "epoch": 3.13, + "learning_rate": 1.3904623363082213e-05, + "loss": 0.192, + "step": 7473 + }, + { + "epoch": 3.13, + "learning_rate": 1.390306248151009e-05, + "loss": 0.1669, + "step": 7474 + }, + { + "epoch": 3.13, + "learning_rate": 1.3901501487747897e-05, + "loss": 0.163, + "step": 7475 + }, + { + "epoch": 3.13, + "learning_rate": 1.389994038184051e-05, + "loss": 0.1953, + "step": 7476 + }, + { + "epoch": 3.13, + "learning_rate": 1.3898379163832794e-05, + "loss": 0.1866, + "step": 7477 + }, + { + "epoch": 3.13, + "learning_rate": 1.3896817833769631e-05, + "loss": 0.1751, + "step": 7478 + }, + { + "epoch": 3.13, + "learning_rate": 1.3895256391695896e-05, + "loss": 0.1909, + "step": 7479 + }, + { + "epoch": 3.13, + "learning_rate": 1.3893694837656471e-05, + "loss": 0.2215, + "step": 7480 + }, + { + "epoch": 3.13, + "learning_rate": 1.3892133171696246e-05, + "loss": 0.1652, + "step": 7481 + }, + { + "epoch": 3.13, + "learning_rate": 1.3890571393860101e-05, + "loss": 0.1398, + "step": 7482 + }, + { + "epoch": 3.13, + "learning_rate": 1.3889009504192938e-05, + "loss": 0.1765, + "step": 7483 + }, + { + "epoch": 3.13, + "learning_rate": 1.3887447502739649e-05, + "loss": 0.1829, + "step": 7484 + }, + { + "epoch": 3.13, + "learning_rate": 1.388588538954513e-05, + "loss": 0.2005, + "step": 7485 + }, + { + "epoch": 3.13, + "learning_rate": 1.3884323164654284e-05, + "loss": 0.1762, + "step": 7486 + }, + { + "epoch": 3.13, + "learning_rate": 1.3882760828112014e-05, + "loss": 0.2078, + "step": 7487 + }, + { + "epoch": 3.14, + "learning_rate": 1.388119837996323e-05, + "loss": 0.1795, + "step": 7488 + }, + { + "epoch": 3.14, + "learning_rate": 1.3879635820252846e-05, + "loss": 0.1307, + "step": 7489 + }, + { + "epoch": 3.14, + "learning_rate": 1.3878073149025768e-05, + "loss": 0.176, + "step": 7490 + }, + { + "epoch": 3.14, + "learning_rate": 1.3876510366326922e-05, + "loss": 0.207, + "step": 7491 + }, + { + "epoch": 3.14, + "learning_rate": 1.3874947472201226e-05, + "loss": 0.1517, + "step": 7492 + }, + { + "epoch": 3.14, + "learning_rate": 1.38733844666936e-05, + "loss": 0.1875, + "step": 7493 + }, + { + "epoch": 3.14, + "learning_rate": 1.3871821349848977e-05, + "loss": 0.1754, + "step": 7494 + }, + { + "epoch": 3.14, + "learning_rate": 1.3870258121712285e-05, + "loss": 0.1777, + "step": 7495 + }, + { + "epoch": 3.14, + "learning_rate": 1.3868694782328456e-05, + "loss": 0.2025, + "step": 7496 + }, + { + "epoch": 3.14, + "learning_rate": 1.3867131331742429e-05, + "loss": 0.1829, + "step": 7497 + }, + { + "epoch": 3.14, + "learning_rate": 1.3865567769999146e-05, + "loss": 0.176, + "step": 7498 + }, + { + "epoch": 3.14, + "learning_rate": 1.3864004097143546e-05, + "loss": 0.1477, + "step": 7499 + }, + { + "epoch": 3.14, + "learning_rate": 1.3862440313220575e-05, + "loss": 0.1657, + "step": 7500 + }, + { + "epoch": 3.14, + "learning_rate": 1.3860876418275189e-05, + "loss": 0.1488, + "step": 7501 + }, + { + "epoch": 3.14, + "learning_rate": 1.3859312412352335e-05, + "loss": 0.193, + "step": 7502 + }, + { + "epoch": 3.14, + "learning_rate": 1.3857748295496972e-05, + "loss": 0.1975, + "step": 7503 + }, + { + "epoch": 3.14, + "learning_rate": 1.3856184067754053e-05, + "loss": 0.1794, + "step": 7504 + }, + { + "epoch": 3.14, + "learning_rate": 1.3854619729168548e-05, + "loss": 0.161, + "step": 7505 + }, + { + "epoch": 3.14, + "learning_rate": 1.3853055279785416e-05, + "loss": 0.1941, + "step": 7506 + }, + { + "epoch": 3.14, + "learning_rate": 1.3851490719649631e-05, + "loss": 0.2179, + "step": 7507 + }, + { + "epoch": 3.14, + "learning_rate": 1.3849926048806166e-05, + "loss": 0.1785, + "step": 7508 + }, + { + "epoch": 3.14, + "learning_rate": 1.384836126729999e-05, + "loss": 0.2004, + "step": 7509 + }, + { + "epoch": 3.14, + "learning_rate": 1.3846796375176083e-05, + "loss": 0.2078, + "step": 7510 + }, + { + "epoch": 3.14, + "learning_rate": 1.3845231372479432e-05, + "loss": 0.1929, + "step": 7511 + }, + { + "epoch": 3.15, + "learning_rate": 1.3843666259255018e-05, + "loss": 0.1956, + "step": 7512 + }, + { + "epoch": 3.15, + "learning_rate": 1.3842101035547827e-05, + "loss": 0.2089, + "step": 7513 + }, + { + "epoch": 3.15, + "learning_rate": 1.3840535701402847e-05, + "loss": 0.184, + "step": 7514 + }, + { + "epoch": 3.15, + "learning_rate": 1.3838970256865078e-05, + "loss": 0.1952, + "step": 7515 + }, + { + "epoch": 3.15, + "learning_rate": 1.3837404701979517e-05, + "loss": 0.1959, + "step": 7516 + }, + { + "epoch": 3.15, + "learning_rate": 1.3835839036791163e-05, + "loss": 0.2267, + "step": 7517 + }, + { + "epoch": 3.15, + "learning_rate": 1.383427326134502e-05, + "loss": 0.2016, + "step": 7518 + }, + { + "epoch": 3.15, + "learning_rate": 1.3832707375686095e-05, + "loss": 0.1755, + "step": 7519 + }, + { + "epoch": 3.15, + "learning_rate": 1.3831141379859394e-05, + "loss": 0.2223, + "step": 7520 + }, + { + "epoch": 3.15, + "learning_rate": 1.3829575273909939e-05, + "loss": 0.1427, + "step": 7521 + }, + { + "epoch": 3.15, + "learning_rate": 1.3828009057882737e-05, + "loss": 0.1875, + "step": 7522 + }, + { + "epoch": 3.15, + "learning_rate": 1.3826442731822814e-05, + "loss": 0.1887, + "step": 7523 + }, + { + "epoch": 3.15, + "learning_rate": 1.3824876295775187e-05, + "loss": 0.1875, + "step": 7524 + }, + { + "epoch": 3.15, + "learning_rate": 1.3823309749784888e-05, + "loss": 0.1904, + "step": 7525 + }, + { + "epoch": 3.15, + "learning_rate": 1.3821743093896942e-05, + "loss": 0.1796, + "step": 7526 + }, + { + "epoch": 3.15, + "learning_rate": 1.3820176328156381e-05, + "loss": 0.1889, + "step": 7527 + }, + { + "epoch": 3.15, + "learning_rate": 1.3818609452608245e-05, + "loss": 0.1661, + "step": 7528 + }, + { + "epoch": 3.15, + "learning_rate": 1.3817042467297565e-05, + "loss": 0.173, + "step": 7529 + }, + { + "epoch": 3.15, + "learning_rate": 1.3815475372269386e-05, + "loss": 0.1865, + "step": 7530 + }, + { + "epoch": 3.15, + "learning_rate": 1.3813908167568757e-05, + "loss": 0.1942, + "step": 7531 + }, + { + "epoch": 3.15, + "learning_rate": 1.381234085324072e-05, + "loss": 0.2108, + "step": 7532 + }, + { + "epoch": 3.15, + "learning_rate": 1.381077342933033e-05, + "loss": 0.1843, + "step": 7533 + }, + { + "epoch": 3.15, + "learning_rate": 1.3809205895882635e-05, + "loss": 0.2048, + "step": 7534 + }, + { + "epoch": 3.15, + "learning_rate": 1.3807638252942704e-05, + "loss": 0.1397, + "step": 7535 + }, + { + "epoch": 3.16, + "learning_rate": 1.3806070500555583e-05, + "loss": 0.2272, + "step": 7536 + }, + { + "epoch": 3.16, + "learning_rate": 1.3804502638766346e-05, + "loss": 0.1862, + "step": 7537 + }, + { + "epoch": 3.16, + "learning_rate": 1.3802934667620055e-05, + "loss": 0.1913, + "step": 7538 + }, + { + "epoch": 3.16, + "learning_rate": 1.3801366587161786e-05, + "loss": 0.1688, + "step": 7539 + }, + { + "epoch": 3.16, + "learning_rate": 1.3799798397436603e-05, + "loss": 0.2065, + "step": 7540 + }, + { + "epoch": 3.16, + "learning_rate": 1.379823009848959e-05, + "loss": 0.1557, + "step": 7541 + }, + { + "epoch": 3.16, + "learning_rate": 1.3796661690365823e-05, + "loss": 0.1426, + "step": 7542 + }, + { + "epoch": 3.16, + "learning_rate": 1.379509317311038e-05, + "loss": 0.2024, + "step": 7543 + }, + { + "epoch": 3.16, + "learning_rate": 1.3793524546768358e-05, + "loss": 0.2058, + "step": 7544 + }, + { + "epoch": 3.16, + "learning_rate": 1.3791955811384837e-05, + "loss": 0.1913, + "step": 7545 + }, + { + "epoch": 3.16, + "learning_rate": 1.3790386967004911e-05, + "loss": 0.1702, + "step": 7546 + }, + { + "epoch": 3.16, + "learning_rate": 1.3788818013673675e-05, + "loss": 0.1924, + "step": 7547 + }, + { + "epoch": 3.16, + "learning_rate": 1.3787248951436225e-05, + "loss": 0.1676, + "step": 7548 + }, + { + "epoch": 3.16, + "learning_rate": 1.3785679780337668e-05, + "loss": 0.1956, + "step": 7549 + }, + { + "epoch": 3.16, + "learning_rate": 1.3784110500423104e-05, + "loss": 0.2423, + "step": 7550 + }, + { + "epoch": 3.16, + "learning_rate": 1.3782541111737642e-05, + "loss": 0.2112, + "step": 7551 + }, + { + "epoch": 3.16, + "learning_rate": 1.3780971614326393e-05, + "loss": 0.1864, + "step": 7552 + }, + { + "epoch": 3.16, + "learning_rate": 1.3779402008234467e-05, + "loss": 0.2137, + "step": 7553 + }, + { + "epoch": 3.16, + "learning_rate": 1.3777832293506985e-05, + "loss": 0.211, + "step": 7554 + }, + { + "epoch": 3.16, + "learning_rate": 1.377626247018907e-05, + "loss": 0.1677, + "step": 7555 + }, + { + "epoch": 3.16, + "learning_rate": 1.3774692538325835e-05, + "loss": 0.2024, + "step": 7556 + }, + { + "epoch": 3.16, + "learning_rate": 1.3773122497962418e-05, + "loss": 0.1875, + "step": 7557 + }, + { + "epoch": 3.16, + "learning_rate": 1.3771552349143942e-05, + "loss": 0.2106, + "step": 7558 + }, + { + "epoch": 3.16, + "learning_rate": 1.3769982091915539e-05, + "loss": 0.1844, + "step": 7559 + }, + { + "epoch": 3.17, + "learning_rate": 1.3768411726322347e-05, + "loss": 0.241, + "step": 7560 + }, + { + "epoch": 3.17, + "learning_rate": 1.3766841252409502e-05, + "loss": 0.1707, + "step": 7561 + }, + { + "epoch": 3.17, + "learning_rate": 1.3765270670222153e-05, + "loss": 0.1793, + "step": 7562 + }, + { + "epoch": 3.17, + "learning_rate": 1.3763699979805434e-05, + "loss": 0.1644, + "step": 7563 + }, + { + "epoch": 3.17, + "learning_rate": 1.3762129181204502e-05, + "loss": 0.2117, + "step": 7564 + }, + { + "epoch": 3.17, + "learning_rate": 1.3760558274464506e-05, + "loss": 0.1701, + "step": 7565 + }, + { + "epoch": 3.17, + "learning_rate": 1.3758987259630597e-05, + "loss": 0.185, + "step": 7566 + }, + { + "epoch": 3.17, + "learning_rate": 1.3757416136747935e-05, + "loss": 0.1732, + "step": 7567 + }, + { + "epoch": 3.17, + "learning_rate": 1.375584490586168e-05, + "loss": 0.218, + "step": 7568 + }, + { + "epoch": 3.17, + "learning_rate": 1.3754273567017e-05, + "loss": 0.1982, + "step": 7569 + }, + { + "epoch": 3.17, + "learning_rate": 1.3752702120259055e-05, + "loss": 0.2061, + "step": 7570 + }, + { + "epoch": 3.17, + "learning_rate": 1.3751130565633015e-05, + "loss": 0.1819, + "step": 7571 + }, + { + "epoch": 3.17, + "learning_rate": 1.3749558903184059e-05, + "loss": 0.1826, + "step": 7572 + }, + { + "epoch": 3.17, + "learning_rate": 1.3747987132957361e-05, + "loss": 0.1503, + "step": 7573 + }, + { + "epoch": 3.17, + "learning_rate": 1.3746415254998095e-05, + "loss": 0.1801, + "step": 7574 + }, + { + "epoch": 3.17, + "learning_rate": 1.3744843269351446e-05, + "loss": 0.1277, + "step": 7575 + }, + { + "epoch": 3.17, + "learning_rate": 1.3743271176062605e-05, + "loss": 0.1731, + "step": 7576 + }, + { + "epoch": 3.17, + "learning_rate": 1.3741698975176754e-05, + "loss": 0.2044, + "step": 7577 + }, + { + "epoch": 3.17, + "learning_rate": 1.3740126666739086e-05, + "loss": 0.1722, + "step": 7578 + }, + { + "epoch": 3.17, + "learning_rate": 1.3738554250794795e-05, + "loss": 0.1917, + "step": 7579 + }, + { + "epoch": 3.17, + "learning_rate": 1.3736981727389083e-05, + "loss": 0.2002, + "step": 7580 + }, + { + "epoch": 3.17, + "learning_rate": 1.3735409096567145e-05, + "loss": 0.1809, + "step": 7581 + }, + { + "epoch": 3.17, + "learning_rate": 1.3733836358374186e-05, + "loss": 0.1463, + "step": 7582 + }, + { + "epoch": 3.17, + "learning_rate": 1.3732263512855416e-05, + "loss": 0.1965, + "step": 7583 + }, + { + "epoch": 3.18, + "learning_rate": 1.3730690560056043e-05, + "loss": 0.1524, + "step": 7584 + }, + { + "epoch": 3.18, + "learning_rate": 1.3729117500021279e-05, + "loss": 0.1706, + "step": 7585 + }, + { + "epoch": 3.18, + "learning_rate": 1.3727544332796344e-05, + "loss": 0.1952, + "step": 7586 + }, + { + "epoch": 3.18, + "learning_rate": 1.3725971058426453e-05, + "loss": 0.1621, + "step": 7587 + }, + { + "epoch": 3.18, + "learning_rate": 1.3724397676956831e-05, + "loss": 0.2063, + "step": 7588 + }, + { + "epoch": 3.18, + "learning_rate": 1.3722824188432702e-05, + "loss": 0.2065, + "step": 7589 + }, + { + "epoch": 3.18, + "learning_rate": 1.3721250592899295e-05, + "loss": 0.2085, + "step": 7590 + }, + { + "epoch": 3.18, + "learning_rate": 1.3719676890401845e-05, + "loss": 0.1861, + "step": 7591 + }, + { + "epoch": 3.18, + "learning_rate": 1.371810308098558e-05, + "loss": 0.1507, + "step": 7592 + }, + { + "epoch": 3.18, + "learning_rate": 1.3716529164695744e-05, + "loss": 0.1772, + "step": 7593 + }, + { + "epoch": 3.18, + "learning_rate": 1.3714955141577573e-05, + "loss": 0.2075, + "step": 7594 + }, + { + "epoch": 3.18, + "learning_rate": 1.3713381011676315e-05, + "loss": 0.2258, + "step": 7595 + }, + { + "epoch": 3.18, + "learning_rate": 1.3711806775037215e-05, + "loss": 0.2036, + "step": 7596 + }, + { + "epoch": 3.18, + "learning_rate": 1.3710232431705521e-05, + "loss": 0.1845, + "step": 7597 + }, + { + "epoch": 3.18, + "learning_rate": 1.3708657981726487e-05, + "loss": 0.1761, + "step": 7598 + }, + { + "epoch": 3.18, + "learning_rate": 1.3707083425145375e-05, + "loss": 0.1685, + "step": 7599 + }, + { + "epoch": 3.18, + "learning_rate": 1.3705508762007436e-05, + "loss": 0.1957, + "step": 7600 + }, + { + "epoch": 3.18, + "learning_rate": 1.3703933992357936e-05, + "loss": 0.1601, + "step": 7601 + }, + { + "epoch": 3.18, + "learning_rate": 1.370235911624214e-05, + "loss": 0.1924, + "step": 7602 + }, + { + "epoch": 3.18, + "learning_rate": 1.370078413370532e-05, + "loss": 0.1813, + "step": 7603 + }, + { + "epoch": 3.18, + "learning_rate": 1.3699209044792743e-05, + "loss": 0.2021, + "step": 7604 + }, + { + "epoch": 3.18, + "learning_rate": 1.3697633849549683e-05, + "loss": 0.1347, + "step": 7605 + }, + { + "epoch": 3.18, + "learning_rate": 1.369605854802142e-05, + "loss": 0.1988, + "step": 7606 + }, + { + "epoch": 3.18, + "learning_rate": 1.3694483140253234e-05, + "loss": 0.1635, + "step": 7607 + }, + { + "epoch": 3.19, + "learning_rate": 1.3692907626290406e-05, + "loss": 0.1745, + "step": 7608 + }, + { + "epoch": 3.19, + "learning_rate": 1.3691332006178226e-05, + "loss": 0.1576, + "step": 7609 + }, + { + "epoch": 3.19, + "learning_rate": 1.3689756279961986e-05, + "loss": 0.1684, + "step": 7610 + }, + { + "epoch": 3.19, + "learning_rate": 1.3688180447686975e-05, + "loss": 0.1723, + "step": 7611 + }, + { + "epoch": 3.19, + "learning_rate": 1.3686604509398489e-05, + "loss": 0.1553, + "step": 7612 + }, + { + "epoch": 3.19, + "learning_rate": 1.3685028465141828e-05, + "loss": 0.1632, + "step": 7613 + }, + { + "epoch": 3.19, + "learning_rate": 1.3683452314962295e-05, + "loss": 0.1662, + "step": 7614 + }, + { + "epoch": 3.19, + "learning_rate": 1.3681876058905193e-05, + "loss": 0.2028, + "step": 7615 + }, + { + "epoch": 3.19, + "learning_rate": 1.368029969701583e-05, + "loss": 0.2133, + "step": 7616 + }, + { + "epoch": 3.19, + "learning_rate": 1.3678723229339518e-05, + "loss": 0.1861, + "step": 7617 + }, + { + "epoch": 3.19, + "learning_rate": 1.3677146655921573e-05, + "loss": 0.2252, + "step": 7618 + }, + { + "epoch": 3.19, + "learning_rate": 1.3675569976807307e-05, + "loss": 0.1522, + "step": 7619 + }, + { + "epoch": 3.19, + "learning_rate": 1.3673993192042048e-05, + "loss": 0.2132, + "step": 7620 + }, + { + "epoch": 3.19, + "learning_rate": 1.3672416301671111e-05, + "loss": 0.1779, + "step": 7621 + }, + { + "epoch": 3.19, + "learning_rate": 1.3670839305739828e-05, + "loss": 0.1832, + "step": 7622 + }, + { + "epoch": 3.19, + "learning_rate": 1.366926220429353e-05, + "loss": 0.1864, + "step": 7623 + }, + { + "epoch": 3.19, + "learning_rate": 1.3667684997377542e-05, + "loss": 0.2032, + "step": 7624 + }, + { + "epoch": 3.19, + "learning_rate": 1.3666107685037204e-05, + "loss": 0.1911, + "step": 7625 + }, + { + "epoch": 3.19, + "learning_rate": 1.3664530267317855e-05, + "loss": 0.1678, + "step": 7626 + }, + { + "epoch": 3.19, + "learning_rate": 1.3662952744264835e-05, + "loss": 0.1906, + "step": 7627 + }, + { + "epoch": 3.19, + "learning_rate": 1.3661375115923487e-05, + "loss": 0.2059, + "step": 7628 + }, + { + "epoch": 3.19, + "learning_rate": 1.3659797382339162e-05, + "loss": 0.1447, + "step": 7629 + }, + { + "epoch": 3.19, + "learning_rate": 1.3658219543557208e-05, + "loss": 0.1612, + "step": 7630 + }, + { + "epoch": 3.19, + "learning_rate": 1.365664159962298e-05, + "loss": 0.1815, + "step": 7631 + }, + { + "epoch": 3.2, + "learning_rate": 1.3655063550581834e-05, + "loss": 0.1874, + "step": 7632 + }, + { + "epoch": 3.2, + "learning_rate": 1.3653485396479129e-05, + "loss": 0.1905, + "step": 7633 + }, + { + "epoch": 3.2, + "learning_rate": 1.365190713736023e-05, + "loss": 0.1837, + "step": 7634 + }, + { + "epoch": 3.2, + "learning_rate": 1.3650328773270498e-05, + "loss": 0.2112, + "step": 7635 + }, + { + "epoch": 3.2, + "learning_rate": 1.3648750304255307e-05, + "loss": 0.1932, + "step": 7636 + }, + { + "epoch": 3.2, + "learning_rate": 1.3647171730360026e-05, + "loss": 0.134, + "step": 7637 + }, + { + "epoch": 3.2, + "learning_rate": 1.3645593051630032e-05, + "loss": 0.18, + "step": 7638 + }, + { + "epoch": 3.2, + "learning_rate": 1.3644014268110697e-05, + "loss": 0.1485, + "step": 7639 + }, + { + "epoch": 3.2, + "learning_rate": 1.364243537984741e-05, + "loss": 0.1891, + "step": 7640 + }, + { + "epoch": 3.2, + "learning_rate": 1.3640856386885546e-05, + "loss": 0.202, + "step": 7641 + }, + { + "epoch": 3.2, + "learning_rate": 1.3639277289270497e-05, + "loss": 0.1828, + "step": 7642 + }, + { + "epoch": 3.2, + "learning_rate": 1.3637698087047651e-05, + "loss": 0.149, + "step": 7643 + }, + { + "epoch": 3.2, + "learning_rate": 1.3636118780262405e-05, + "loss": 0.1346, + "step": 7644 + }, + { + "epoch": 3.2, + "learning_rate": 1.3634539368960147e-05, + "loss": 0.1852, + "step": 7645 + }, + { + "epoch": 3.2, + "learning_rate": 1.363295985318628e-05, + "loss": 0.1923, + "step": 7646 + }, + { + "epoch": 3.2, + "learning_rate": 1.363138023298621e-05, + "loss": 0.182, + "step": 7647 + }, + { + "epoch": 3.2, + "learning_rate": 1.3629800508405338e-05, + "loss": 0.1883, + "step": 7648 + }, + { + "epoch": 3.2, + "learning_rate": 1.3628220679489069e-05, + "loss": 0.2061, + "step": 7649 + }, + { + "epoch": 3.2, + "learning_rate": 1.3626640746282815e-05, + "loss": 0.2, + "step": 7650 + }, + { + "epoch": 3.2, + "learning_rate": 1.3625060708831992e-05, + "loss": 0.1932, + "step": 7651 + }, + { + "epoch": 3.2, + "learning_rate": 1.3623480567182018e-05, + "loss": 0.154, + "step": 7652 + }, + { + "epoch": 3.2, + "learning_rate": 1.3621900321378308e-05, + "loss": 0.1829, + "step": 7653 + }, + { + "epoch": 3.2, + "learning_rate": 1.3620319971466288e-05, + "loss": 0.2042, + "step": 7654 + }, + { + "epoch": 3.2, + "learning_rate": 1.3618739517491381e-05, + "loss": 0.2031, + "step": 7655 + }, + { + "epoch": 3.21, + "learning_rate": 1.361715895949902e-05, + "loss": 0.1759, + "step": 7656 + }, + { + "epoch": 3.21, + "learning_rate": 1.3615578297534635e-05, + "loss": 0.2407, + "step": 7657 + }, + { + "epoch": 3.21, + "learning_rate": 1.3613997531643657e-05, + "loss": 0.1647, + "step": 7658 + }, + { + "epoch": 3.21, + "learning_rate": 1.3612416661871532e-05, + "loss": 0.2102, + "step": 7659 + }, + { + "epoch": 3.21, + "learning_rate": 1.361083568826369e-05, + "loss": 0.1727, + "step": 7660 + }, + { + "epoch": 3.21, + "learning_rate": 1.3609254610865587e-05, + "loss": 0.2066, + "step": 7661 + }, + { + "epoch": 3.21, + "learning_rate": 1.3607673429722658e-05, + "loss": 0.1782, + "step": 7662 + }, + { + "epoch": 3.21, + "learning_rate": 1.3606092144880359e-05, + "loss": 0.1969, + "step": 7663 + }, + { + "epoch": 3.21, + "learning_rate": 1.3604510756384141e-05, + "loss": 0.1575, + "step": 7664 + }, + { + "epoch": 3.21, + "learning_rate": 1.360292926427946e-05, + "loss": 0.1964, + "step": 7665 + }, + { + "epoch": 3.21, + "learning_rate": 1.3601347668611772e-05, + "loss": 0.1888, + "step": 7666 + }, + { + "epoch": 3.21, + "learning_rate": 1.3599765969426546e-05, + "loss": 0.1649, + "step": 7667 + }, + { + "epoch": 3.21, + "learning_rate": 1.359818416676924e-05, + "loss": 0.1653, + "step": 7668 + }, + { + "epoch": 3.21, + "learning_rate": 1.3596602260685321e-05, + "loss": 0.1853, + "step": 7669 + }, + { + "epoch": 3.21, + "learning_rate": 1.3595020251220263e-05, + "loss": 0.1915, + "step": 7670 + }, + { + "epoch": 3.21, + "learning_rate": 1.3593438138419539e-05, + "loss": 0.1966, + "step": 7671 + }, + { + "epoch": 3.21, + "learning_rate": 1.3591855922328624e-05, + "loss": 0.2002, + "step": 7672 + }, + { + "epoch": 3.21, + "learning_rate": 1.3590273602992996e-05, + "loss": 0.1996, + "step": 7673 + }, + { + "epoch": 3.21, + "learning_rate": 1.3588691180458143e-05, + "loss": 0.1883, + "step": 7674 + }, + { + "epoch": 3.21, + "learning_rate": 1.3587108654769542e-05, + "loss": 0.1751, + "step": 7675 + }, + { + "epoch": 3.21, + "learning_rate": 1.3585526025972688e-05, + "loss": 0.1636, + "step": 7676 + }, + { + "epoch": 3.21, + "learning_rate": 1.3583943294113071e-05, + "loss": 0.2001, + "step": 7677 + }, + { + "epoch": 3.21, + "learning_rate": 1.3582360459236183e-05, + "loss": 0.1908, + "step": 7678 + }, + { + "epoch": 3.21, + "learning_rate": 1.3580777521387523e-05, + "loss": 0.1748, + "step": 7679 + }, + { + "epoch": 3.22, + "learning_rate": 1.357919448061259e-05, + "loss": 0.1909, + "step": 7680 + }, + { + "epoch": 3.22, + "learning_rate": 1.3577611336956888e-05, + "loss": 0.1817, + "step": 7681 + }, + { + "epoch": 3.22, + "learning_rate": 1.3576028090465924e-05, + "loss": 0.1583, + "step": 7682 + }, + { + "epoch": 3.22, + "learning_rate": 1.3574444741185207e-05, + "loss": 0.1666, + "step": 7683 + }, + { + "epoch": 3.22, + "learning_rate": 1.3572861289160245e-05, + "loss": 0.2047, + "step": 7684 + }, + { + "epoch": 3.22, + "learning_rate": 1.3571277734436556e-05, + "loss": 0.1697, + "step": 7685 + }, + { + "epoch": 3.22, + "learning_rate": 1.356969407705966e-05, + "loss": 0.1998, + "step": 7686 + }, + { + "epoch": 3.22, + "learning_rate": 1.3568110317075071e-05, + "loss": 0.1905, + "step": 7687 + }, + { + "epoch": 3.22, + "learning_rate": 1.356652645452832e-05, + "loss": 0.1819, + "step": 7688 + }, + { + "epoch": 3.22, + "learning_rate": 1.3564942489464932e-05, + "loss": 0.2108, + "step": 7689 + }, + { + "epoch": 3.22, + "learning_rate": 1.3563358421930433e-05, + "loss": 0.2004, + "step": 7690 + }, + { + "epoch": 3.22, + "learning_rate": 1.356177425197036e-05, + "loss": 0.178, + "step": 7691 + }, + { + "epoch": 3.22, + "learning_rate": 1.3560189979630247e-05, + "loss": 0.2014, + "step": 7692 + }, + { + "epoch": 3.22, + "learning_rate": 1.3558605604955633e-05, + "loss": 0.1777, + "step": 7693 + }, + { + "epoch": 3.22, + "learning_rate": 1.3557021127992055e-05, + "loss": 0.172, + "step": 7694 + }, + { + "epoch": 3.22, + "learning_rate": 1.3555436548785066e-05, + "loss": 0.1939, + "step": 7695 + }, + { + "epoch": 3.22, + "learning_rate": 1.3553851867380205e-05, + "loss": 0.1617, + "step": 7696 + }, + { + "epoch": 3.22, + "learning_rate": 1.3552267083823028e-05, + "loss": 0.1244, + "step": 7697 + }, + { + "epoch": 3.22, + "learning_rate": 1.3550682198159084e-05, + "loss": 0.1537, + "step": 7698 + }, + { + "epoch": 3.22, + "learning_rate": 1.3549097210433932e-05, + "loss": 0.1861, + "step": 7699 + }, + { + "epoch": 3.22, + "learning_rate": 1.354751212069313e-05, + "loss": 0.2075, + "step": 7700 + }, + { + "epoch": 3.22, + "learning_rate": 1.3545926928982241e-05, + "loss": 0.1893, + "step": 7701 + }, + { + "epoch": 3.22, + "learning_rate": 1.3544341635346827e-05, + "loss": 0.215, + "step": 7702 + }, + { + "epoch": 3.23, + "learning_rate": 1.3542756239832458e-05, + "loss": 0.1982, + "step": 7703 + }, + { + "epoch": 3.23, + "learning_rate": 1.3541170742484705e-05, + "loss": 0.2129, + "step": 7704 + }, + { + "epoch": 3.23, + "learning_rate": 1.3539585143349146e-05, + "loss": 0.1981, + "step": 7705 + }, + { + "epoch": 3.23, + "learning_rate": 1.3537999442471349e-05, + "loss": 0.1974, + "step": 7706 + }, + { + "epoch": 3.23, + "learning_rate": 1.3536413639896895e-05, + "loss": 0.176, + "step": 7707 + }, + { + "epoch": 3.23, + "learning_rate": 1.3534827735671373e-05, + "loss": 0.2061, + "step": 7708 + }, + { + "epoch": 3.23, + "learning_rate": 1.3533241729840362e-05, + "loss": 0.1642, + "step": 7709 + }, + { + "epoch": 3.23, + "learning_rate": 1.3531655622449454e-05, + "loss": 0.1684, + "step": 7710 + }, + { + "epoch": 3.23, + "learning_rate": 1.3530069413544241e-05, + "loss": 0.1948, + "step": 7711 + }, + { + "epoch": 3.23, + "learning_rate": 1.3528483103170312e-05, + "loss": 0.1634, + "step": 7712 + }, + { + "epoch": 3.23, + "learning_rate": 1.3526896691373269e-05, + "loss": 0.1526, + "step": 7713 + }, + { + "epoch": 3.23, + "learning_rate": 1.3525310178198707e-05, + "loss": 0.2019, + "step": 7714 + }, + { + "epoch": 3.23, + "learning_rate": 1.3523723563692237e-05, + "loss": 0.2137, + "step": 7715 + }, + { + "epoch": 3.23, + "learning_rate": 1.352213684789946e-05, + "loss": 0.1549, + "step": 7716 + }, + { + "epoch": 3.23, + "learning_rate": 1.3520550030865983e-05, + "loss": 0.1934, + "step": 7717 + }, + { + "epoch": 3.23, + "learning_rate": 1.3518963112637419e-05, + "loss": 0.1872, + "step": 7718 + }, + { + "epoch": 3.23, + "learning_rate": 1.3517376093259383e-05, + "loss": 0.2245, + "step": 7719 + }, + { + "epoch": 3.23, + "learning_rate": 1.3515788972777491e-05, + "loss": 0.1961, + "step": 7720 + }, + { + "epoch": 3.23, + "learning_rate": 1.3514201751237366e-05, + "loss": 0.1684, + "step": 7721 + }, + { + "epoch": 3.23, + "learning_rate": 1.3512614428684633e-05, + "loss": 0.1901, + "step": 7722 + }, + { + "epoch": 3.23, + "learning_rate": 1.3511027005164911e-05, + "loss": 0.1897, + "step": 7723 + }, + { + "epoch": 3.23, + "learning_rate": 1.3509439480723835e-05, + "loss": 0.1878, + "step": 7724 + }, + { + "epoch": 3.23, + "learning_rate": 1.3507851855407038e-05, + "loss": 0.1834, + "step": 7725 + }, + { + "epoch": 3.23, + "learning_rate": 1.350626412926015e-05, + "loss": 0.1637, + "step": 7726 + }, + { + "epoch": 3.24, + "learning_rate": 1.3504676302328813e-05, + "loss": 0.1433, + "step": 7727 + }, + { + "epoch": 3.24, + "learning_rate": 1.3503088374658661e-05, + "loss": 0.2066, + "step": 7728 + }, + { + "epoch": 3.24, + "learning_rate": 1.3501500346295347e-05, + "loss": 0.1597, + "step": 7729 + }, + { + "epoch": 3.24, + "learning_rate": 1.349991221728451e-05, + "loss": 0.201, + "step": 7730 + }, + { + "epoch": 3.24, + "learning_rate": 1.3498323987671801e-05, + "loss": 0.1925, + "step": 7731 + }, + { + "epoch": 3.24, + "learning_rate": 1.3496735657502877e-05, + "loss": 0.1942, + "step": 7732 + }, + { + "epoch": 3.24, + "learning_rate": 1.3495147226823386e-05, + "loss": 0.2095, + "step": 7733 + }, + { + "epoch": 3.24, + "learning_rate": 1.3493558695678992e-05, + "loss": 0.1603, + "step": 7734 + }, + { + "epoch": 3.24, + "learning_rate": 1.3491970064115351e-05, + "loss": 0.17, + "step": 7735 + }, + { + "epoch": 3.24, + "learning_rate": 1.3490381332178131e-05, + "loss": 0.1829, + "step": 7736 + }, + { + "epoch": 3.24, + "learning_rate": 1.3488792499912998e-05, + "loss": 0.2311, + "step": 7737 + }, + { + "epoch": 3.24, + "learning_rate": 1.348720356736562e-05, + "loss": 0.2073, + "step": 7738 + }, + { + "epoch": 3.24, + "learning_rate": 1.348561453458167e-05, + "loss": 0.1692, + "step": 7739 + }, + { + "epoch": 3.24, + "learning_rate": 1.3484025401606821e-05, + "loss": 0.1917, + "step": 7740 + }, + { + "epoch": 3.24, + "learning_rate": 1.3482436168486754e-05, + "loss": 0.1508, + "step": 7741 + }, + { + "epoch": 3.24, + "learning_rate": 1.3480846835267152e-05, + "loss": 0.1405, + "step": 7742 + }, + { + "epoch": 3.24, + "learning_rate": 1.3479257401993696e-05, + "loss": 0.2095, + "step": 7743 + }, + { + "epoch": 3.24, + "learning_rate": 1.347766786871207e-05, + "loss": 0.2075, + "step": 7744 + }, + { + "epoch": 3.24, + "learning_rate": 1.347607823546797e-05, + "loss": 0.1521, + "step": 7745 + }, + { + "epoch": 3.24, + "learning_rate": 1.3474488502307085e-05, + "loss": 0.1693, + "step": 7746 + }, + { + "epoch": 3.24, + "learning_rate": 1.3472898669275112e-05, + "loss": 0.1621, + "step": 7747 + }, + { + "epoch": 3.24, + "learning_rate": 1.347130873641775e-05, + "loss": 0.2106, + "step": 7748 + }, + { + "epoch": 3.24, + "learning_rate": 1.3469718703780694e-05, + "loss": 0.2237, + "step": 7749 + }, + { + "epoch": 3.24, + "learning_rate": 1.3468128571409658e-05, + "loss": 0.1928, + "step": 7750 + }, + { + "epoch": 3.25, + "learning_rate": 1.346653833935034e-05, + "loss": 0.2011, + "step": 7751 + }, + { + "epoch": 3.25, + "learning_rate": 1.3464948007648457e-05, + "loss": 0.1587, + "step": 7752 + }, + { + "epoch": 3.25, + "learning_rate": 1.3463357576349716e-05, + "loss": 0.1522, + "step": 7753 + }, + { + "epoch": 3.25, + "learning_rate": 1.3461767045499834e-05, + "loss": 0.2028, + "step": 7754 + }, + { + "epoch": 3.25, + "learning_rate": 1.346017641514453e-05, + "loss": 0.1807, + "step": 7755 + }, + { + "epoch": 3.25, + "learning_rate": 1.3458585685329529e-05, + "loss": 0.1705, + "step": 7756 + }, + { + "epoch": 3.25, + "learning_rate": 1.3456994856100547e-05, + "loss": 0.2111, + "step": 7757 + }, + { + "epoch": 3.25, + "learning_rate": 1.3455403927503317e-05, + "loss": 0.1834, + "step": 7758 + }, + { + "epoch": 3.25, + "learning_rate": 1.3453812899583572e-05, + "loss": 0.1738, + "step": 7759 + }, + { + "epoch": 3.25, + "learning_rate": 1.3452221772387036e-05, + "loss": 0.1727, + "step": 7760 + }, + { + "epoch": 3.25, + "learning_rate": 1.3450630545959448e-05, + "loss": 0.1906, + "step": 7761 + }, + { + "epoch": 3.25, + "learning_rate": 1.3449039220346547e-05, + "loss": 0.1901, + "step": 7762 + }, + { + "epoch": 3.25, + "learning_rate": 1.3447447795594078e-05, + "loss": 0.1671, + "step": 7763 + }, + { + "epoch": 3.25, + "learning_rate": 1.3445856271747777e-05, + "loss": 0.1833, + "step": 7764 + }, + { + "epoch": 3.25, + "learning_rate": 1.3444264648853395e-05, + "loss": 0.2022, + "step": 7765 + }, + { + "epoch": 3.25, + "learning_rate": 1.3442672926956685e-05, + "loss": 0.1815, + "step": 7766 + }, + { + "epoch": 3.25, + "learning_rate": 1.3441081106103395e-05, + "loss": 0.164, + "step": 7767 + }, + { + "epoch": 3.25, + "learning_rate": 1.3439489186339283e-05, + "loss": 0.208, + "step": 7768 + }, + { + "epoch": 3.25, + "learning_rate": 1.3437897167710105e-05, + "loss": 0.2039, + "step": 7769 + }, + { + "epoch": 3.25, + "learning_rate": 1.3436305050261625e-05, + "loss": 0.1856, + "step": 7770 + }, + { + "epoch": 3.25, + "learning_rate": 1.3434712834039606e-05, + "loss": 0.1896, + "step": 7771 + }, + { + "epoch": 3.25, + "learning_rate": 1.3433120519089813e-05, + "loss": 0.1631, + "step": 7772 + }, + { + "epoch": 3.25, + "learning_rate": 1.3431528105458021e-05, + "loss": 0.2288, + "step": 7773 + }, + { + "epoch": 3.25, + "learning_rate": 1.3429935593189996e-05, + "loss": 0.1845, + "step": 7774 + }, + { + "epoch": 3.26, + "learning_rate": 1.3428342982331515e-05, + "loss": 0.2092, + "step": 7775 + }, + { + "epoch": 3.26, + "learning_rate": 1.3426750272928359e-05, + "loss": 0.1466, + "step": 7776 + }, + { + "epoch": 3.26, + "learning_rate": 1.3425157465026305e-05, + "loss": 0.2149, + "step": 7777 + }, + { + "epoch": 3.26, + "learning_rate": 1.342356455867114e-05, + "loss": 0.2079, + "step": 7778 + }, + { + "epoch": 3.26, + "learning_rate": 1.342197155390865e-05, + "loss": 0.1664, + "step": 7779 + }, + { + "epoch": 3.26, + "learning_rate": 1.3420378450784623e-05, + "loss": 0.1936, + "step": 7780 + }, + { + "epoch": 3.26, + "learning_rate": 1.3418785249344855e-05, + "loss": 0.1982, + "step": 7781 + }, + { + "epoch": 3.26, + "learning_rate": 1.3417191949635137e-05, + "loss": 0.1644, + "step": 7782 + }, + { + "epoch": 3.26, + "learning_rate": 1.341559855170127e-05, + "loss": 0.1735, + "step": 7783 + }, + { + "epoch": 3.26, + "learning_rate": 1.3414005055589057e-05, + "loss": 0.1726, + "step": 7784 + }, + { + "epoch": 3.26, + "learning_rate": 1.3412411461344293e-05, + "loss": 0.1926, + "step": 7785 + }, + { + "epoch": 3.26, + "learning_rate": 1.3410817769012792e-05, + "loss": 0.1661, + "step": 7786 + }, + { + "epoch": 3.26, + "learning_rate": 1.340922397864036e-05, + "loss": 0.2143, + "step": 7787 + }, + { + "epoch": 3.26, + "learning_rate": 1.3407630090272808e-05, + "loss": 0.1829, + "step": 7788 + }, + { + "epoch": 3.26, + "learning_rate": 1.3406036103955957e-05, + "loss": 0.1304, + "step": 7789 + }, + { + "epoch": 3.26, + "learning_rate": 1.3404442019735617e-05, + "loss": 0.1993, + "step": 7790 + }, + { + "epoch": 3.26, + "learning_rate": 1.3402847837657614e-05, + "loss": 0.1636, + "step": 7791 + }, + { + "epoch": 3.26, + "learning_rate": 1.3401253557767768e-05, + "loss": 0.1769, + "step": 7792 + }, + { + "epoch": 3.26, + "learning_rate": 1.339965918011191e-05, + "loss": 0.1884, + "step": 7793 + }, + { + "epoch": 3.26, + "learning_rate": 1.3398064704735862e-05, + "loss": 0.1943, + "step": 7794 + }, + { + "epoch": 3.26, + "learning_rate": 1.3396470131685462e-05, + "loss": 0.2005, + "step": 7795 + }, + { + "epoch": 3.26, + "learning_rate": 1.3394875461006543e-05, + "loss": 0.2422, + "step": 7796 + }, + { + "epoch": 3.26, + "learning_rate": 1.3393280692744938e-05, + "loss": 0.1647, + "step": 7797 + }, + { + "epoch": 3.26, + "learning_rate": 1.3391685826946492e-05, + "loss": 0.2119, + "step": 7798 + }, + { + "epoch": 3.27, + "learning_rate": 1.3390090863657048e-05, + "loss": 0.1554, + "step": 7799 + }, + { + "epoch": 3.27, + "learning_rate": 1.338849580292245e-05, + "loss": 0.1796, + "step": 7800 + }, + { + "epoch": 3.27, + "learning_rate": 1.3386900644788546e-05, + "loss": 0.1523, + "step": 7801 + }, + { + "epoch": 3.27, + "learning_rate": 1.3385305389301187e-05, + "loss": 0.1896, + "step": 7802 + }, + { + "epoch": 3.27, + "learning_rate": 1.3383710036506232e-05, + "loss": 0.2046, + "step": 7803 + }, + { + "epoch": 3.27, + "learning_rate": 1.3382114586449533e-05, + "loss": 0.1643, + "step": 7804 + }, + { + "epoch": 3.27, + "learning_rate": 1.3380519039176951e-05, + "loss": 0.1594, + "step": 7805 + }, + { + "epoch": 3.27, + "learning_rate": 1.337892339473435e-05, + "loss": 0.1781, + "step": 7806 + }, + { + "epoch": 3.27, + "learning_rate": 1.3377327653167596e-05, + "loss": 0.1904, + "step": 7807 + }, + { + "epoch": 3.27, + "learning_rate": 1.3375731814522552e-05, + "loss": 0.1833, + "step": 7808 + }, + { + "epoch": 3.27, + "learning_rate": 1.3374135878845094e-05, + "loss": 0.1631, + "step": 7809 + }, + { + "epoch": 3.27, + "learning_rate": 1.3372539846181096e-05, + "loss": 0.2096, + "step": 7810 + }, + { + "epoch": 3.27, + "learning_rate": 1.337094371657643e-05, + "loss": 0.1732, + "step": 7811 + }, + { + "epoch": 3.27, + "learning_rate": 1.3369347490076981e-05, + "loss": 0.2059, + "step": 7812 + }, + { + "epoch": 3.27, + "learning_rate": 1.3367751166728626e-05, + "loss": 0.1556, + "step": 7813 + }, + { + "epoch": 3.27, + "learning_rate": 1.3366154746577254e-05, + "loss": 0.152, + "step": 7814 + }, + { + "epoch": 3.27, + "learning_rate": 1.3364558229668748e-05, + "loss": 0.1588, + "step": 7815 + }, + { + "epoch": 3.27, + "learning_rate": 1.3362961616049006e-05, + "loss": 0.1508, + "step": 7816 + }, + { + "epoch": 3.27, + "learning_rate": 1.3361364905763912e-05, + "loss": 0.2059, + "step": 7817 + }, + { + "epoch": 3.27, + "learning_rate": 1.335976809885937e-05, + "loss": 0.1803, + "step": 7818 + }, + { + "epoch": 3.27, + "learning_rate": 1.3358171195381272e-05, + "loss": 0.2117, + "step": 7819 + }, + { + "epoch": 3.27, + "learning_rate": 1.3356574195375526e-05, + "loss": 0.118, + "step": 7820 + }, + { + "epoch": 3.27, + "learning_rate": 1.335497709888803e-05, + "loss": 0.1968, + "step": 7821 + }, + { + "epoch": 3.27, + "learning_rate": 1.3353379905964698e-05, + "loss": 0.1957, + "step": 7822 + }, + { + "epoch": 3.28, + "learning_rate": 1.3351782616651436e-05, + "loss": 0.1985, + "step": 7823 + }, + { + "epoch": 3.28, + "learning_rate": 1.3350185230994157e-05, + "loss": 0.227, + "step": 7824 + }, + { + "epoch": 3.28, + "learning_rate": 1.3348587749038773e-05, + "loss": 0.2088, + "step": 7825 + }, + { + "epoch": 3.28, + "learning_rate": 1.334699017083121e-05, + "loss": 0.1898, + "step": 7826 + }, + { + "epoch": 3.28, + "learning_rate": 1.3345392496417381e-05, + "loss": 0.1721, + "step": 7827 + }, + { + "epoch": 3.28, + "learning_rate": 1.3343794725843215e-05, + "loss": 0.1599, + "step": 7828 + }, + { + "epoch": 3.28, + "learning_rate": 1.3342196859154639e-05, + "loss": 0.2064, + "step": 7829 + }, + { + "epoch": 3.28, + "learning_rate": 1.3340598896397579e-05, + "loss": 0.1717, + "step": 7830 + }, + { + "epoch": 3.28, + "learning_rate": 1.3339000837617967e-05, + "loss": 0.161, + "step": 7831 + }, + { + "epoch": 3.28, + "learning_rate": 1.3337402682861737e-05, + "loss": 0.1659, + "step": 7832 + }, + { + "epoch": 3.28, + "learning_rate": 1.333580443217483e-05, + "loss": 0.1689, + "step": 7833 + }, + { + "epoch": 3.28, + "learning_rate": 1.3334206085603186e-05, + "loss": 0.2025, + "step": 7834 + }, + { + "epoch": 3.28, + "learning_rate": 1.3332607643192745e-05, + "loss": 0.1741, + "step": 7835 + }, + { + "epoch": 3.28, + "learning_rate": 1.3331009104989456e-05, + "loss": 0.1395, + "step": 7836 + }, + { + "epoch": 3.28, + "learning_rate": 1.3329410471039266e-05, + "loss": 0.2303, + "step": 7837 + }, + { + "epoch": 3.28, + "learning_rate": 1.3327811741388126e-05, + "loss": 0.1997, + "step": 7838 + }, + { + "epoch": 3.28, + "learning_rate": 1.3326212916081991e-05, + "loss": 0.1643, + "step": 7839 + }, + { + "epoch": 3.28, + "learning_rate": 1.3324613995166818e-05, + "loss": 0.1636, + "step": 7840 + }, + { + "epoch": 3.28, + "learning_rate": 1.3323014978688567e-05, + "loss": 0.1608, + "step": 7841 + }, + { + "epoch": 3.28, + "learning_rate": 1.3321415866693195e-05, + "loss": 0.2052, + "step": 7842 + }, + { + "epoch": 3.28, + "learning_rate": 1.3319816659226672e-05, + "loss": 0.1693, + "step": 7843 + }, + { + "epoch": 3.28, + "learning_rate": 1.3318217356334967e-05, + "loss": 0.1523, + "step": 7844 + }, + { + "epoch": 3.28, + "learning_rate": 1.3316617958064049e-05, + "loss": 0.1823, + "step": 7845 + }, + { + "epoch": 3.28, + "learning_rate": 1.331501846445989e-05, + "loss": 0.2051, + "step": 7846 + }, + { + "epoch": 3.29, + "learning_rate": 1.3313418875568469e-05, + "loss": 0.197, + "step": 7847 + }, + { + "epoch": 3.29, + "learning_rate": 1.331181919143576e-05, + "loss": 0.1993, + "step": 7848 + }, + { + "epoch": 3.29, + "learning_rate": 1.3310219412107745e-05, + "loss": 0.1948, + "step": 7849 + }, + { + "epoch": 3.29, + "learning_rate": 1.3308619537630416e-05, + "loss": 0.1773, + "step": 7850 + }, + { + "epoch": 3.29, + "learning_rate": 1.3307019568049751e-05, + "loss": 0.1666, + "step": 7851 + }, + { + "epoch": 3.29, + "learning_rate": 1.3305419503411745e-05, + "loss": 0.1913, + "step": 7852 + }, + { + "epoch": 3.29, + "learning_rate": 1.3303819343762387e-05, + "loss": 0.2112, + "step": 7853 + }, + { + "epoch": 3.29, + "learning_rate": 1.3302219089147675e-05, + "loss": 0.1746, + "step": 7854 + }, + { + "epoch": 3.29, + "learning_rate": 1.3300618739613605e-05, + "loss": 0.1547, + "step": 7855 + }, + { + "epoch": 3.29, + "learning_rate": 1.3299018295206176e-05, + "loss": 0.1833, + "step": 7856 + }, + { + "epoch": 3.29, + "learning_rate": 1.3297417755971399e-05, + "loss": 0.1845, + "step": 7857 + }, + { + "epoch": 3.29, + "learning_rate": 1.329581712195527e-05, + "loss": 0.1692, + "step": 7858 + }, + { + "epoch": 3.29, + "learning_rate": 1.3294216393203803e-05, + "loss": 0.1781, + "step": 7859 + }, + { + "epoch": 3.29, + "learning_rate": 1.329261556976301e-05, + "loss": 0.1617, + "step": 7860 + }, + { + "epoch": 3.29, + "learning_rate": 1.3291014651678905e-05, + "loss": 0.215, + "step": 7861 + }, + { + "epoch": 3.29, + "learning_rate": 1.3289413638997502e-05, + "loss": 0.1892, + "step": 7862 + }, + { + "epoch": 3.29, + "learning_rate": 1.3287812531764826e-05, + "loss": 0.1876, + "step": 7863 + }, + { + "epoch": 3.29, + "learning_rate": 1.3286211330026894e-05, + "loss": 0.1248, + "step": 7864 + }, + { + "epoch": 3.29, + "learning_rate": 1.3284610033829734e-05, + "loss": 0.2129, + "step": 7865 + }, + { + "epoch": 3.29, + "learning_rate": 1.3283008643219373e-05, + "loss": 0.2075, + "step": 7866 + }, + { + "epoch": 3.29, + "learning_rate": 1.3281407158241842e-05, + "loss": 0.1871, + "step": 7867 + }, + { + "epoch": 3.29, + "learning_rate": 1.3279805578943174e-05, + "loss": 0.1847, + "step": 7868 + }, + { + "epoch": 3.29, + "learning_rate": 1.3278203905369407e-05, + "loss": 0.1944, + "step": 7869 + }, + { + "epoch": 3.29, + "learning_rate": 1.3276602137566575e-05, + "loss": 0.2002, + "step": 7870 + }, + { + "epoch": 3.3, + "learning_rate": 1.3275000275580723e-05, + "loss": 0.1795, + "step": 7871 + }, + { + "epoch": 3.3, + "learning_rate": 1.3273398319457895e-05, + "loss": 0.1858, + "step": 7872 + }, + { + "epoch": 3.3, + "learning_rate": 1.3271796269244135e-05, + "loss": 0.1892, + "step": 7873 + }, + { + "epoch": 3.3, + "learning_rate": 1.3270194124985499e-05, + "loss": 0.1832, + "step": 7874 + }, + { + "epoch": 3.3, + "learning_rate": 1.326859188672803e-05, + "loss": 0.1631, + "step": 7875 + }, + { + "epoch": 3.3, + "learning_rate": 1.3266989554517791e-05, + "loss": 0.2076, + "step": 7876 + }, + { + "epoch": 3.3, + "learning_rate": 1.3265387128400833e-05, + "loss": 0.2149, + "step": 7877 + }, + { + "epoch": 3.3, + "learning_rate": 1.3263784608423225e-05, + "loss": 0.186, + "step": 7878 + }, + { + "epoch": 3.3, + "learning_rate": 1.326218199463102e-05, + "loss": 0.1097, + "step": 7879 + }, + { + "epoch": 3.3, + "learning_rate": 1.3260579287070288e-05, + "loss": 0.1991, + "step": 7880 + }, + { + "epoch": 3.3, + "learning_rate": 1.3258976485787101e-05, + "loss": 0.1648, + "step": 7881 + }, + { + "epoch": 3.3, + "learning_rate": 1.3257373590827525e-05, + "loss": 0.1858, + "step": 7882 + }, + { + "epoch": 3.3, + "learning_rate": 1.3255770602237634e-05, + "loss": 0.2108, + "step": 7883 + }, + { + "epoch": 3.3, + "learning_rate": 1.325416752006351e-05, + "loss": 0.1963, + "step": 7884 + }, + { + "epoch": 3.3, + "learning_rate": 1.3252564344351227e-05, + "loss": 0.2124, + "step": 7885 + }, + { + "epoch": 3.3, + "learning_rate": 1.3250961075146868e-05, + "loss": 0.1754, + "step": 7886 + }, + { + "epoch": 3.3, + "learning_rate": 1.3249357712496518e-05, + "loss": 0.1954, + "step": 7887 + }, + { + "epoch": 3.3, + "learning_rate": 1.3247754256446263e-05, + "loss": 0.1774, + "step": 7888 + }, + { + "epoch": 3.3, + "learning_rate": 1.3246150707042196e-05, + "loss": 0.1806, + "step": 7889 + }, + { + "epoch": 3.3, + "learning_rate": 1.3244547064330404e-05, + "loss": 0.1706, + "step": 7890 + }, + { + "epoch": 3.3, + "learning_rate": 1.3242943328356988e-05, + "loss": 0.1864, + "step": 7891 + }, + { + "epoch": 3.3, + "learning_rate": 1.3241339499168042e-05, + "loss": 0.1964, + "step": 7892 + }, + { + "epoch": 3.3, + "learning_rate": 1.323973557680967e-05, + "loss": 0.1794, + "step": 7893 + }, + { + "epoch": 3.31, + "learning_rate": 1.3238131561327976e-05, + "loss": 0.162, + "step": 7894 + }, + { + "epoch": 3.31, + "learning_rate": 1.3236527452769061e-05, + "loss": 0.1988, + "step": 7895 + }, + { + "epoch": 3.31, + "learning_rate": 1.3234923251179036e-05, + "loss": 0.1611, + "step": 7896 + }, + { + "epoch": 3.31, + "learning_rate": 1.3233318956604013e-05, + "loss": 0.1859, + "step": 7897 + }, + { + "epoch": 3.31, + "learning_rate": 1.3231714569090108e-05, + "loss": 0.1789, + "step": 7898 + }, + { + "epoch": 3.31, + "learning_rate": 1.3230110088683433e-05, + "loss": 0.1955, + "step": 7899 + }, + { + "epoch": 3.31, + "learning_rate": 1.322850551543011e-05, + "loss": 0.2029, + "step": 7900 + }, + { + "epoch": 3.31, + "learning_rate": 1.3226900849376263e-05, + "loss": 0.1774, + "step": 7901 + }, + { + "epoch": 3.31, + "learning_rate": 1.3225296090568011e-05, + "loss": 0.204, + "step": 7902 + }, + { + "epoch": 3.31, + "learning_rate": 1.3223691239051489e-05, + "loss": 0.1998, + "step": 7903 + }, + { + "epoch": 3.31, + "learning_rate": 1.3222086294872818e-05, + "loss": 0.1742, + "step": 7904 + }, + { + "epoch": 3.31, + "learning_rate": 1.3220481258078142e-05, + "loss": 0.2086, + "step": 7905 + }, + { + "epoch": 3.31, + "learning_rate": 1.3218876128713584e-05, + "loss": 0.2133, + "step": 7906 + }, + { + "epoch": 3.31, + "learning_rate": 1.321727090682529e-05, + "loss": 0.2085, + "step": 7907 + }, + { + "epoch": 3.31, + "learning_rate": 1.3215665592459401e-05, + "loss": 0.1646, + "step": 7908 + }, + { + "epoch": 3.31, + "learning_rate": 1.3214060185662058e-05, + "loss": 0.1632, + "step": 7909 + }, + { + "epoch": 3.31, + "learning_rate": 1.3212454686479406e-05, + "loss": 0.1936, + "step": 7910 + }, + { + "epoch": 3.31, + "learning_rate": 1.3210849094957594e-05, + "loss": 0.1792, + "step": 7911 + }, + { + "epoch": 3.31, + "learning_rate": 1.3209243411142776e-05, + "loss": 0.1455, + "step": 7912 + }, + { + "epoch": 3.31, + "learning_rate": 1.32076376350811e-05, + "loss": 0.1829, + "step": 7913 + }, + { + "epoch": 3.31, + "learning_rate": 1.320603176681873e-05, + "loss": 0.2218, + "step": 7914 + }, + { + "epoch": 3.31, + "learning_rate": 1.3204425806401822e-05, + "loss": 0.1586, + "step": 7915 + }, + { + "epoch": 3.31, + "learning_rate": 1.3202819753876535e-05, + "loss": 0.204, + "step": 7916 + }, + { + "epoch": 3.31, + "learning_rate": 1.3201213609289036e-05, + "loss": 0.168, + "step": 7917 + }, + { + "epoch": 3.32, + "learning_rate": 1.3199607372685497e-05, + "loss": 0.1784, + "step": 7918 + }, + { + "epoch": 3.32, + "learning_rate": 1.319800104411208e-05, + "loss": 0.2257, + "step": 7919 + }, + { + "epoch": 3.32, + "learning_rate": 1.3196394623614963e-05, + "loss": 0.1622, + "step": 7920 + }, + { + "epoch": 3.32, + "learning_rate": 1.3194788111240317e-05, + "loss": 0.1386, + "step": 7921 + }, + { + "epoch": 3.32, + "learning_rate": 1.3193181507034323e-05, + "loss": 0.1977, + "step": 7922 + }, + { + "epoch": 3.32, + "learning_rate": 1.3191574811043159e-05, + "loss": 0.2258, + "step": 7923 + }, + { + "epoch": 3.32, + "learning_rate": 1.318996802331301e-05, + "loss": 0.2303, + "step": 7924 + }, + { + "epoch": 3.32, + "learning_rate": 1.318836114389006e-05, + "loss": 0.1982, + "step": 7925 + }, + { + "epoch": 3.32, + "learning_rate": 1.31867541728205e-05, + "loss": 0.1936, + "step": 7926 + }, + { + "epoch": 3.32, + "learning_rate": 1.3185147110150516e-05, + "loss": 0.1944, + "step": 7927 + }, + { + "epoch": 3.32, + "learning_rate": 1.3183539955926307e-05, + "loss": 0.191, + "step": 7928 + }, + { + "epoch": 3.32, + "learning_rate": 1.3181932710194067e-05, + "loss": 0.2047, + "step": 7929 + }, + { + "epoch": 3.32, + "learning_rate": 1.3180325372999994e-05, + "loss": 0.167, + "step": 7930 + }, + { + "epoch": 3.32, + "learning_rate": 1.3178717944390292e-05, + "loss": 0.2035, + "step": 7931 + }, + { + "epoch": 3.32, + "learning_rate": 1.3177110424411164e-05, + "loss": 0.2037, + "step": 7932 + }, + { + "epoch": 3.32, + "learning_rate": 1.3175502813108813e-05, + "loss": 0.2023, + "step": 7933 + }, + { + "epoch": 3.32, + "learning_rate": 1.3173895110529454e-05, + "loss": 0.1541, + "step": 7934 + }, + { + "epoch": 3.32, + "learning_rate": 1.3172287316719297e-05, + "loss": 0.1985, + "step": 7935 + }, + { + "epoch": 3.32, + "learning_rate": 1.3170679431724553e-05, + "loss": 0.1705, + "step": 7936 + }, + { + "epoch": 3.32, + "learning_rate": 1.3169071455591443e-05, + "loss": 0.2345, + "step": 7937 + }, + { + "epoch": 3.32, + "learning_rate": 1.3167463388366186e-05, + "loss": 0.2211, + "step": 7938 + }, + { + "epoch": 3.32, + "learning_rate": 1.3165855230095008e-05, + "loss": 0.2096, + "step": 7939 + }, + { + "epoch": 3.32, + "learning_rate": 1.3164246980824126e-05, + "loss": 0.1247, + "step": 7940 + }, + { + "epoch": 3.32, + "learning_rate": 1.3162638640599777e-05, + "loss": 0.1589, + "step": 7941 + }, + { + "epoch": 3.33, + "learning_rate": 1.3161030209468187e-05, + "loss": 0.1761, + "step": 7942 + }, + { + "epoch": 3.33, + "learning_rate": 1.3159421687475586e-05, + "loss": 0.1564, + "step": 7943 + }, + { + "epoch": 3.33, + "learning_rate": 1.3157813074668213e-05, + "loss": 0.1976, + "step": 7944 + }, + { + "epoch": 3.33, + "learning_rate": 1.3156204371092302e-05, + "loss": 0.1818, + "step": 7945 + }, + { + "epoch": 3.33, + "learning_rate": 1.3154595576794103e-05, + "loss": 0.2022, + "step": 7946 + }, + { + "epoch": 3.33, + "learning_rate": 1.315298669181985e-05, + "loss": 0.1456, + "step": 7947 + }, + { + "epoch": 3.33, + "learning_rate": 1.3151377716215791e-05, + "loss": 0.1878, + "step": 7948 + }, + { + "epoch": 3.33, + "learning_rate": 1.3149768650028181e-05, + "loss": 0.1428, + "step": 7949 + }, + { + "epoch": 3.33, + "learning_rate": 1.3148159493303262e-05, + "loss": 0.1644, + "step": 7950 + }, + { + "epoch": 3.33, + "learning_rate": 1.3146550246087293e-05, + "loss": 0.1733, + "step": 7951 + }, + { + "epoch": 3.33, + "learning_rate": 1.3144940908426532e-05, + "loss": 0.1824, + "step": 7952 + }, + { + "epoch": 3.33, + "learning_rate": 1.314333148036723e-05, + "loss": 0.1695, + "step": 7953 + }, + { + "epoch": 3.33, + "learning_rate": 1.3141721961955658e-05, + "loss": 0.1797, + "step": 7954 + }, + { + "epoch": 3.33, + "learning_rate": 1.3140112353238075e-05, + "loss": 0.1447, + "step": 7955 + }, + { + "epoch": 3.33, + "learning_rate": 1.3138502654260749e-05, + "loss": 0.1697, + "step": 7956 + }, + { + "epoch": 3.33, + "learning_rate": 1.313689286506995e-05, + "loss": 0.1789, + "step": 7957 + }, + { + "epoch": 3.33, + "learning_rate": 1.3135282985711948e-05, + "loss": 0.1959, + "step": 7958 + }, + { + "epoch": 3.33, + "learning_rate": 1.313367301623302e-05, + "loss": 0.1866, + "step": 7959 + }, + { + "epoch": 3.33, + "learning_rate": 1.3132062956679441e-05, + "loss": 0.1782, + "step": 7960 + }, + { + "epoch": 3.33, + "learning_rate": 1.3130452807097491e-05, + "loss": 0.195, + "step": 7961 + }, + { + "epoch": 3.33, + "learning_rate": 1.3128842567533456e-05, + "loss": 0.2005, + "step": 7962 + }, + { + "epoch": 3.33, + "learning_rate": 1.3127232238033613e-05, + "loss": 0.1895, + "step": 7963 + }, + { + "epoch": 3.33, + "learning_rate": 1.3125621818644258e-05, + "loss": 0.2153, + "step": 7964 + }, + { + "epoch": 3.33, + "learning_rate": 1.3124011309411678e-05, + "loss": 0.2004, + "step": 7965 + }, + { + "epoch": 3.34, + "learning_rate": 1.3122400710382164e-05, + "loss": 0.211, + "step": 7966 + }, + { + "epoch": 3.34, + "learning_rate": 1.3120790021602012e-05, + "loss": 0.1833, + "step": 7967 + }, + { + "epoch": 3.34, + "learning_rate": 1.311917924311752e-05, + "loss": 0.2227, + "step": 7968 + }, + { + "epoch": 3.34, + "learning_rate": 1.311756837497499e-05, + "loss": 0.2323, + "step": 7969 + }, + { + "epoch": 3.34, + "learning_rate": 1.3115957417220722e-05, + "loss": 0.1494, + "step": 7970 + }, + { + "epoch": 3.34, + "learning_rate": 1.3114346369901021e-05, + "loss": 0.1479, + "step": 7971 + }, + { + "epoch": 3.34, + "learning_rate": 1.3112735233062201e-05, + "loss": 0.1661, + "step": 7972 + }, + { + "epoch": 3.34, + "learning_rate": 1.311112400675057e-05, + "loss": 0.1722, + "step": 7973 + }, + { + "epoch": 3.34, + "learning_rate": 1.3109512691012434e-05, + "loss": 0.1949, + "step": 7974 + }, + { + "epoch": 3.34, + "learning_rate": 1.3107901285894118e-05, + "loss": 0.2101, + "step": 7975 + }, + { + "epoch": 3.34, + "learning_rate": 1.3106289791441941e-05, + "loss": 0.1923, + "step": 7976 + }, + { + "epoch": 3.34, + "learning_rate": 1.3104678207702218e-05, + "loss": 0.1802, + "step": 7977 + }, + { + "epoch": 3.34, + "learning_rate": 1.3103066534721273e-05, + "loss": 0.2054, + "step": 7978 + }, + { + "epoch": 3.34, + "learning_rate": 1.3101454772545435e-05, + "loss": 0.2031, + "step": 7979 + }, + { + "epoch": 3.34, + "learning_rate": 1.3099842921221034e-05, + "loss": 0.2368, + "step": 7980 + }, + { + "epoch": 3.34, + "learning_rate": 1.3098230980794394e-05, + "loss": 0.1559, + "step": 7981 + }, + { + "epoch": 3.34, + "learning_rate": 1.309661895131186e-05, + "loss": 0.1767, + "step": 7982 + }, + { + "epoch": 3.34, + "learning_rate": 1.3095006832819759e-05, + "loss": 0.1372, + "step": 7983 + }, + { + "epoch": 3.34, + "learning_rate": 1.3093394625364433e-05, + "loss": 0.1748, + "step": 7984 + }, + { + "epoch": 3.34, + "learning_rate": 1.3091782328992225e-05, + "loss": 0.1735, + "step": 7985 + }, + { + "epoch": 3.34, + "learning_rate": 1.3090169943749475e-05, + "loss": 0.1624, + "step": 7986 + }, + { + "epoch": 3.34, + "learning_rate": 1.3088557469682535e-05, + "loss": 0.1809, + "step": 7987 + }, + { + "epoch": 3.34, + "learning_rate": 1.3086944906837751e-05, + "loss": 0.2157, + "step": 7988 + }, + { + "epoch": 3.34, + "learning_rate": 1.3085332255261472e-05, + "loss": 0.1759, + "step": 7989 + }, + { + "epoch": 3.35, + "learning_rate": 1.3083719515000057e-05, + "loss": 0.1399, + "step": 7990 + }, + { + "epoch": 3.35, + "learning_rate": 1.308210668609986e-05, + "loss": 0.1942, + "step": 7991 + }, + { + "epoch": 3.35, + "learning_rate": 1.3080493768607243e-05, + "loss": 0.1867, + "step": 7992 + }, + { + "epoch": 3.35, + "learning_rate": 1.3078880762568566e-05, + "loss": 0.2199, + "step": 7993 + }, + { + "epoch": 3.35, + "learning_rate": 1.3077267668030194e-05, + "loss": 0.2082, + "step": 7994 + }, + { + "epoch": 3.35, + "learning_rate": 1.3075654485038491e-05, + "loss": 0.166, + "step": 7995 + }, + { + "epoch": 3.35, + "learning_rate": 1.3074041213639832e-05, + "loss": 0.1959, + "step": 7996 + }, + { + "epoch": 3.35, + "learning_rate": 1.3072427853880583e-05, + "loss": 0.1902, + "step": 7997 + }, + { + "epoch": 3.35, + "learning_rate": 1.3070814405807124e-05, + "loss": 0.1851, + "step": 7998 + }, + { + "epoch": 3.35, + "learning_rate": 1.3069200869465832e-05, + "loss": 0.1786, + "step": 7999 + }, + { + "epoch": 3.35, + "learning_rate": 1.3067587244903081e-05, + "loss": 0.1853, + "step": 8000 + }, + { + "epoch": 3.35, + "learning_rate": 1.3065973532165259e-05, + "loss": 0.1811, + "step": 8001 + }, + { + "epoch": 3.35, + "learning_rate": 1.3064359731298746e-05, + "loss": 0.2069, + "step": 8002 + }, + { + "epoch": 3.35, + "learning_rate": 1.3062745842349935e-05, + "loss": 0.2145, + "step": 8003 + }, + { + "epoch": 3.35, + "learning_rate": 1.3061131865365211e-05, + "loss": 0.2082, + "step": 8004 + }, + { + "epoch": 3.35, + "learning_rate": 1.3059517800390967e-05, + "loss": 0.1653, + "step": 8005 + }, + { + "epoch": 3.35, + "learning_rate": 1.30579036474736e-05, + "loss": 0.1863, + "step": 8006 + }, + { + "epoch": 3.35, + "learning_rate": 1.3056289406659506e-05, + "loss": 0.172, + "step": 8007 + }, + { + "epoch": 3.35, + "learning_rate": 1.3054675077995082e-05, + "loss": 0.1559, + "step": 8008 + }, + { + "epoch": 3.35, + "learning_rate": 1.3053060661526738e-05, + "loss": 0.202, + "step": 8009 + }, + { + "epoch": 3.35, + "learning_rate": 1.3051446157300875e-05, + "loss": 0.1904, + "step": 8010 + }, + { + "epoch": 3.35, + "learning_rate": 1.3049831565363897e-05, + "loss": 0.2119, + "step": 8011 + }, + { + "epoch": 3.35, + "learning_rate": 1.3048216885762218e-05, + "loss": 0.2065, + "step": 8012 + }, + { + "epoch": 3.35, + "learning_rate": 1.3046602118542251e-05, + "loss": 0.2191, + "step": 8013 + }, + { + "epoch": 3.36, + "learning_rate": 1.3044987263750409e-05, + "loss": 0.1768, + "step": 8014 + }, + { + "epoch": 3.36, + "learning_rate": 1.3043372321433108e-05, + "loss": 0.1888, + "step": 8015 + }, + { + "epoch": 3.36, + "learning_rate": 1.304175729163677e-05, + "loss": 0.1931, + "step": 8016 + }, + { + "epoch": 3.36, + "learning_rate": 1.304014217440782e-05, + "loss": 0.1912, + "step": 8017 + }, + { + "epoch": 3.36, + "learning_rate": 1.303852696979268e-05, + "loss": 0.1944, + "step": 8018 + }, + { + "epoch": 3.36, + "learning_rate": 1.3036911677837778e-05, + "loss": 0.1651, + "step": 8019 + }, + { + "epoch": 3.36, + "learning_rate": 1.3035296298589549e-05, + "loss": 0.2085, + "step": 8020 + }, + { + "epoch": 3.36, + "learning_rate": 1.3033680832094416e-05, + "loss": 0.2003, + "step": 8021 + }, + { + "epoch": 3.36, + "learning_rate": 1.3032065278398823e-05, + "loss": 0.1749, + "step": 8022 + }, + { + "epoch": 3.36, + "learning_rate": 1.3030449637549203e-05, + "loss": 0.1729, + "step": 8023 + }, + { + "epoch": 3.36, + "learning_rate": 1.3028833909591999e-05, + "loss": 0.2078, + "step": 8024 + }, + { + "epoch": 3.36, + "learning_rate": 1.3027218094573648e-05, + "loss": 0.1727, + "step": 8025 + }, + { + "epoch": 3.36, + "learning_rate": 1.30256021925406e-05, + "loss": 0.1386, + "step": 8026 + }, + { + "epoch": 3.36, + "learning_rate": 1.3023986203539304e-05, + "loss": 0.1952, + "step": 8027 + }, + { + "epoch": 3.36, + "learning_rate": 1.3022370127616209e-05, + "loss": 0.1911, + "step": 8028 + }, + { + "epoch": 3.36, + "learning_rate": 1.3020753964817763e-05, + "loss": 0.1608, + "step": 8029 + }, + { + "epoch": 3.36, + "learning_rate": 1.301913771519043e-05, + "loss": 0.225, + "step": 8030 + }, + { + "epoch": 3.36, + "learning_rate": 1.3017521378780655e-05, + "loss": 0.2167, + "step": 8031 + }, + { + "epoch": 3.36, + "learning_rate": 1.301590495563491e-05, + "loss": 0.1651, + "step": 8032 + }, + { + "epoch": 3.36, + "learning_rate": 1.3014288445799656e-05, + "loss": 0.1953, + "step": 8033 + }, + { + "epoch": 3.36, + "learning_rate": 1.3012671849321351e-05, + "loss": 0.2044, + "step": 8034 + }, + { + "epoch": 3.36, + "learning_rate": 1.3011055166246469e-05, + "loss": 0.2039, + "step": 8035 + }, + { + "epoch": 3.36, + "learning_rate": 1.3009438396621479e-05, + "loss": 0.1804, + "step": 8036 + }, + { + "epoch": 3.36, + "learning_rate": 1.3007821540492854e-05, + "loss": 0.1906, + "step": 8037 + }, + { + "epoch": 3.37, + "learning_rate": 1.3006204597907064e-05, + "loss": 0.2217, + "step": 8038 + }, + { + "epoch": 3.37, + "learning_rate": 1.3004587568910592e-05, + "loss": 0.2237, + "step": 8039 + }, + { + "epoch": 3.37, + "learning_rate": 1.3002970453549918e-05, + "loss": 0.1782, + "step": 8040 + }, + { + "epoch": 3.37, + "learning_rate": 1.300135325187152e-05, + "loss": 0.211, + "step": 8041 + }, + { + "epoch": 3.37, + "learning_rate": 1.2999735963921887e-05, + "loss": 0.184, + "step": 8042 + }, + { + "epoch": 3.37, + "learning_rate": 1.2998118589747509e-05, + "loss": 0.1948, + "step": 8043 + }, + { + "epoch": 3.37, + "learning_rate": 1.2996501129394872e-05, + "loss": 0.1965, + "step": 8044 + }, + { + "epoch": 3.37, + "learning_rate": 1.2994883582910465e-05, + "loss": 0.1551, + "step": 8045 + }, + { + "epoch": 3.37, + "learning_rate": 1.299326595034079e-05, + "loss": 0.1572, + "step": 8046 + }, + { + "epoch": 3.37, + "learning_rate": 1.2991648231732342e-05, + "loss": 0.1887, + "step": 8047 + }, + { + "epoch": 3.37, + "learning_rate": 1.2990030427131618e-05, + "loss": 0.1751, + "step": 8048 + }, + { + "epoch": 3.37, + "learning_rate": 1.2988412536585125e-05, + "loss": 0.2004, + "step": 8049 + }, + { + "epoch": 3.37, + "learning_rate": 1.2986794560139362e-05, + "loss": 0.2151, + "step": 8050 + }, + { + "epoch": 3.37, + "learning_rate": 1.2985176497840843e-05, + "loss": 0.2198, + "step": 8051 + }, + { + "epoch": 3.37, + "learning_rate": 1.2983558349736074e-05, + "loss": 0.1641, + "step": 8052 + }, + { + "epoch": 3.37, + "learning_rate": 1.2981940115871566e-05, + "loss": 0.176, + "step": 8053 + }, + { + "epoch": 3.37, + "learning_rate": 1.2980321796293838e-05, + "loss": 0.1961, + "step": 8054 + }, + { + "epoch": 3.37, + "learning_rate": 1.2978703391049402e-05, + "loss": 0.1903, + "step": 8055 + }, + { + "epoch": 3.37, + "learning_rate": 1.2977084900184784e-05, + "loss": 0.1761, + "step": 8056 + }, + { + "epoch": 3.37, + "learning_rate": 1.2975466323746499e-05, + "loss": 0.1296, + "step": 8057 + }, + { + "epoch": 3.37, + "learning_rate": 1.2973847661781078e-05, + "loss": 0.1581, + "step": 8058 + }, + { + "epoch": 3.37, + "learning_rate": 1.2972228914335044e-05, + "loss": 0.1424, + "step": 8059 + }, + { + "epoch": 3.37, + "learning_rate": 1.2970610081454925e-05, + "loss": 0.1769, + "step": 8060 + }, + { + "epoch": 3.37, + "learning_rate": 1.2968991163187256e-05, + "loss": 0.187, + "step": 8061 + }, + { + "epoch": 3.38, + "learning_rate": 1.2967372159578574e-05, + "loss": 0.1795, + "step": 8062 + }, + { + "epoch": 3.38, + "learning_rate": 1.2965753070675406e-05, + "loss": 0.1714, + "step": 8063 + }, + { + "epoch": 3.38, + "learning_rate": 1.2964133896524305e-05, + "loss": 0.1511, + "step": 8064 + }, + { + "epoch": 3.38, + "learning_rate": 1.2962514637171802e-05, + "loss": 0.1504, + "step": 8065 + }, + { + "epoch": 3.38, + "learning_rate": 1.2960895292664443e-05, + "loss": 0.1826, + "step": 8066 + }, + { + "epoch": 3.38, + "learning_rate": 1.295927586304878e-05, + "loss": 0.1782, + "step": 8067 + }, + { + "epoch": 3.38, + "learning_rate": 1.2957656348371357e-05, + "loss": 0.2208, + "step": 8068 + }, + { + "epoch": 3.38, + "learning_rate": 1.2956036748678727e-05, + "loss": 0.1828, + "step": 8069 + }, + { + "epoch": 3.38, + "learning_rate": 1.2954417064017441e-05, + "loss": 0.1838, + "step": 8070 + }, + { + "epoch": 3.38, + "learning_rate": 1.295279729443406e-05, + "loss": 0.1932, + "step": 8071 + }, + { + "epoch": 3.38, + "learning_rate": 1.2951177439975141e-05, + "loss": 0.2389, + "step": 8072 + }, + { + "epoch": 3.38, + "learning_rate": 1.2949557500687244e-05, + "loss": 0.1934, + "step": 8073 + }, + { + "epoch": 3.38, + "learning_rate": 1.2947937476616938e-05, + "loss": 0.1962, + "step": 8074 + }, + { + "epoch": 3.38, + "learning_rate": 1.2946317367810784e-05, + "loss": 0.2111, + "step": 8075 + }, + { + "epoch": 3.38, + "learning_rate": 1.2944697174315349e-05, + "loss": 0.1932, + "step": 8076 + }, + { + "epoch": 3.38, + "learning_rate": 1.2943076896177208e-05, + "loss": 0.1867, + "step": 8077 + }, + { + "epoch": 3.38, + "learning_rate": 1.2941456533442935e-05, + "loss": 0.1605, + "step": 8078 + }, + { + "epoch": 3.38, + "learning_rate": 1.2939836086159102e-05, + "loss": 0.2104, + "step": 8079 + }, + { + "epoch": 3.38, + "learning_rate": 1.2938215554372292e-05, + "loss": 0.1872, + "step": 8080 + }, + { + "epoch": 3.38, + "learning_rate": 1.2936594938129082e-05, + "loss": 0.1978, + "step": 8081 + }, + { + "epoch": 3.38, + "learning_rate": 1.2934974237476054e-05, + "loss": 0.2306, + "step": 8082 + }, + { + "epoch": 3.38, + "learning_rate": 1.2933353452459796e-05, + "loss": 0.1974, + "step": 8083 + }, + { + "epoch": 3.38, + "learning_rate": 1.29317325831269e-05, + "loss": 0.1785, + "step": 8084 + }, + { + "epoch": 3.38, + "learning_rate": 1.293011162952395e-05, + "loss": 0.1864, + "step": 8085 + }, + { + "epoch": 3.39, + "learning_rate": 1.292849059169754e-05, + "loss": 0.2104, + "step": 8086 + }, + { + "epoch": 3.39, + "learning_rate": 1.2926869469694267e-05, + "loss": 0.2014, + "step": 8087 + }, + { + "epoch": 3.39, + "learning_rate": 1.2925248263560733e-05, + "loss": 0.2069, + "step": 8088 + }, + { + "epoch": 3.39, + "learning_rate": 1.2923626973343527e-05, + "loss": 0.1916, + "step": 8089 + }, + { + "epoch": 3.39, + "learning_rate": 1.2922005599089263e-05, + "loss": 0.1521, + "step": 8090 + }, + { + "epoch": 3.39, + "learning_rate": 1.2920384140844539e-05, + "loss": 0.2251, + "step": 8091 + }, + { + "epoch": 3.39, + "learning_rate": 1.2918762598655965e-05, + "loss": 0.2019, + "step": 8092 + }, + { + "epoch": 3.39, + "learning_rate": 1.291714097257015e-05, + "loss": 0.1826, + "step": 8093 + }, + { + "epoch": 3.39, + "learning_rate": 1.2915519262633704e-05, + "loss": 0.2038, + "step": 8094 + }, + { + "epoch": 3.39, + "learning_rate": 1.2913897468893249e-05, + "loss": 0.187, + "step": 8095 + }, + { + "epoch": 3.39, + "learning_rate": 1.2912275591395393e-05, + "loss": 0.2268, + "step": 8096 + }, + { + "epoch": 3.39, + "learning_rate": 1.291065363018676e-05, + "loss": 0.2099, + "step": 8097 + }, + { + "epoch": 3.39, + "learning_rate": 1.290903158531397e-05, + "loss": 0.1623, + "step": 8098 + }, + { + "epoch": 3.39, + "learning_rate": 1.2907409456823651e-05, + "loss": 0.2001, + "step": 8099 + }, + { + "epoch": 3.39, + "learning_rate": 1.2905787244762427e-05, + "loss": 0.1717, + "step": 8100 + }, + { + "epoch": 3.39, + "learning_rate": 1.2904164949176927e-05, + "loss": 0.1823, + "step": 8101 + }, + { + "epoch": 3.39, + "learning_rate": 1.2902542570113786e-05, + "loss": 0.2004, + "step": 8102 + }, + { + "epoch": 3.39, + "learning_rate": 1.2900920107619627e-05, + "loss": 0.1961, + "step": 8103 + }, + { + "epoch": 3.39, + "learning_rate": 1.2899297561741097e-05, + "loss": 0.2076, + "step": 8104 + }, + { + "epoch": 3.39, + "learning_rate": 1.2897674932524832e-05, + "loss": 0.1968, + "step": 8105 + }, + { + "epoch": 3.39, + "learning_rate": 1.2896052220017474e-05, + "loss": 0.1579, + "step": 8106 + }, + { + "epoch": 3.39, + "learning_rate": 1.2894429424265661e-05, + "loss": 0.1819, + "step": 8107 + }, + { + "epoch": 3.39, + "learning_rate": 1.2892806545316048e-05, + "loss": 0.191, + "step": 8108 + }, + { + "epoch": 3.4, + "learning_rate": 1.2891183583215272e-05, + "loss": 0.1599, + "step": 8109 + }, + { + "epoch": 3.4, + "learning_rate": 1.2889560538009993e-05, + "loss": 0.164, + "step": 8110 + }, + { + "epoch": 3.4, + "learning_rate": 1.288793740974686e-05, + "loss": 0.1802, + "step": 8111 + }, + { + "epoch": 3.4, + "learning_rate": 1.288631419847253e-05, + "loss": 0.174, + "step": 8112 + }, + { + "epoch": 3.4, + "learning_rate": 1.2884690904233658e-05, + "loss": 0.2084, + "step": 8113 + }, + { + "epoch": 3.4, + "learning_rate": 1.2883067527076904e-05, + "loss": 0.1475, + "step": 8114 + }, + { + "epoch": 3.4, + "learning_rate": 1.2881444067048936e-05, + "loss": 0.1961, + "step": 8115 + }, + { + "epoch": 3.4, + "learning_rate": 1.2879820524196414e-05, + "loss": 0.1655, + "step": 8116 + }, + { + "epoch": 3.4, + "learning_rate": 1.2878196898566004e-05, + "loss": 0.1942, + "step": 8117 + }, + { + "epoch": 3.4, + "learning_rate": 1.2876573190204381e-05, + "loss": 0.1498, + "step": 8118 + }, + { + "epoch": 3.4, + "learning_rate": 1.2874949399158213e-05, + "loss": 0.2078, + "step": 8119 + }, + { + "epoch": 3.4, + "learning_rate": 1.2873325525474176e-05, + "loss": 0.2005, + "step": 8120 + }, + { + "epoch": 3.4, + "learning_rate": 1.2871701569198946e-05, + "loss": 0.1994, + "step": 8121 + }, + { + "epoch": 3.4, + "learning_rate": 1.2870077530379205e-05, + "loss": 0.1859, + "step": 8122 + }, + { + "epoch": 3.4, + "learning_rate": 1.286845340906163e-05, + "loss": 0.1879, + "step": 8123 + }, + { + "epoch": 3.4, + "learning_rate": 1.286682920529291e-05, + "loss": 0.1767, + "step": 8124 + }, + { + "epoch": 3.4, + "learning_rate": 1.2865204919119726e-05, + "loss": 0.203, + "step": 8125 + }, + { + "epoch": 3.4, + "learning_rate": 1.286358055058877e-05, + "loss": 0.2012, + "step": 8126 + }, + { + "epoch": 3.4, + "learning_rate": 1.2861956099746731e-05, + "loss": 0.1452, + "step": 8127 + }, + { + "epoch": 3.4, + "learning_rate": 1.2860331566640303e-05, + "loss": 0.1434, + "step": 8128 + }, + { + "epoch": 3.4, + "learning_rate": 1.2858706951316184e-05, + "loss": 0.1894, + "step": 8129 + }, + { + "epoch": 3.4, + "learning_rate": 1.285708225382107e-05, + "loss": 0.2196, + "step": 8130 + }, + { + "epoch": 3.4, + "learning_rate": 1.2855457474201663e-05, + "loss": 0.1776, + "step": 8131 + }, + { + "epoch": 3.4, + "learning_rate": 1.2853832612504662e-05, + "loss": 0.1807, + "step": 8132 + }, + { + "epoch": 3.41, + "learning_rate": 1.2852207668776777e-05, + "loss": 0.1557, + "step": 8133 + }, + { + "epoch": 3.41, + "learning_rate": 1.2850582643064713e-05, + "loss": 0.1438, + "step": 8134 + }, + { + "epoch": 3.41, + "learning_rate": 1.2848957535415183e-05, + "loss": 0.179, + "step": 8135 + }, + { + "epoch": 3.41, + "learning_rate": 1.2847332345874895e-05, + "loss": 0.1775, + "step": 8136 + }, + { + "epoch": 3.41, + "learning_rate": 1.2845707074490563e-05, + "loss": 0.1993, + "step": 8137 + }, + { + "epoch": 3.41, + "learning_rate": 1.284408172130891e-05, + "loss": 0.173, + "step": 8138 + }, + { + "epoch": 3.41, + "learning_rate": 1.284245628637665e-05, + "loss": 0.1968, + "step": 8139 + }, + { + "epoch": 3.41, + "learning_rate": 1.2840830769740506e-05, + "loss": 0.1977, + "step": 8140 + }, + { + "epoch": 3.41, + "learning_rate": 1.2839205171447203e-05, + "loss": 0.1816, + "step": 8141 + }, + { + "epoch": 3.41, + "learning_rate": 1.2837579491543468e-05, + "loss": 0.186, + "step": 8142 + }, + { + "epoch": 3.41, + "learning_rate": 1.2835953730076026e-05, + "loss": 0.2, + "step": 8143 + }, + { + "epoch": 3.41, + "learning_rate": 1.2834327887091611e-05, + "loss": 0.2119, + "step": 8144 + }, + { + "epoch": 3.41, + "learning_rate": 1.2832701962636957e-05, + "loss": 0.1799, + "step": 8145 + }, + { + "epoch": 3.41, + "learning_rate": 1.2831075956758801e-05, + "loss": 0.1658, + "step": 8146 + }, + { + "epoch": 3.41, + "learning_rate": 1.2829449869503876e-05, + "loss": 0.1929, + "step": 8147 + }, + { + "epoch": 3.41, + "learning_rate": 1.2827823700918924e-05, + "loss": 0.203, + "step": 8148 + }, + { + "epoch": 3.41, + "learning_rate": 1.2826197451050692e-05, + "loss": 0.1814, + "step": 8149 + }, + { + "epoch": 3.41, + "learning_rate": 1.2824571119945921e-05, + "loss": 0.1706, + "step": 8150 + }, + { + "epoch": 3.41, + "learning_rate": 1.282294470765136e-05, + "loss": 0.1477, + "step": 8151 + }, + { + "epoch": 3.41, + "learning_rate": 1.2821318214213758e-05, + "loss": 0.2028, + "step": 8152 + }, + { + "epoch": 3.41, + "learning_rate": 1.2819691639679867e-05, + "loss": 0.1899, + "step": 8153 + }, + { + "epoch": 3.41, + "learning_rate": 1.2818064984096443e-05, + "loss": 0.2138, + "step": 8154 + }, + { + "epoch": 3.41, + "learning_rate": 1.2816438247510243e-05, + "loss": 0.1841, + "step": 8155 + }, + { + "epoch": 3.41, + "learning_rate": 1.2814811429968022e-05, + "loss": 0.1999, + "step": 8156 + }, + { + "epoch": 3.42, + "learning_rate": 1.2813184531516545e-05, + "loss": 0.1649, + "step": 8157 + }, + { + "epoch": 3.42, + "learning_rate": 1.2811557552202576e-05, + "loss": 0.185, + "step": 8158 + }, + { + "epoch": 3.42, + "learning_rate": 1.2809930492072882e-05, + "loss": 0.1613, + "step": 8159 + }, + { + "epoch": 3.42, + "learning_rate": 1.2808303351174224e-05, + "loss": 0.189, + "step": 8160 + }, + { + "epoch": 3.42, + "learning_rate": 1.2806676129553381e-05, + "loss": 0.2033, + "step": 8161 + }, + { + "epoch": 3.42, + "learning_rate": 1.2805048827257123e-05, + "loss": 0.2193, + "step": 8162 + }, + { + "epoch": 3.42, + "learning_rate": 1.280342144433223e-05, + "loss": 0.2024, + "step": 8163 + }, + { + "epoch": 3.42, + "learning_rate": 1.280179398082547e-05, + "loss": 0.1882, + "step": 8164 + }, + { + "epoch": 3.42, + "learning_rate": 1.2800166436783627e-05, + "loss": 0.1955, + "step": 8165 + }, + { + "epoch": 3.42, + "learning_rate": 1.279853881225349e-05, + "loss": 0.1965, + "step": 8166 + }, + { + "epoch": 3.42, + "learning_rate": 1.2796911107281833e-05, + "loss": 0.1778, + "step": 8167 + }, + { + "epoch": 3.42, + "learning_rate": 1.279528332191545e-05, + "loss": 0.2018, + "step": 8168 + }, + { + "epoch": 3.42, + "learning_rate": 1.279365545620113e-05, + "loss": 0.1709, + "step": 8169 + }, + { + "epoch": 3.42, + "learning_rate": 1.2792027510185665e-05, + "loss": 0.186, + "step": 8170 + }, + { + "epoch": 3.42, + "learning_rate": 1.2790399483915843e-05, + "loss": 0.1931, + "step": 8171 + }, + { + "epoch": 3.42, + "learning_rate": 1.2788771377438465e-05, + "loss": 0.17, + "step": 8172 + }, + { + "epoch": 3.42, + "learning_rate": 1.278714319080033e-05, + "loss": 0.1962, + "step": 8173 + }, + { + "epoch": 3.42, + "learning_rate": 1.2785514924048235e-05, + "loss": 0.1863, + "step": 8174 + }, + { + "epoch": 3.42, + "learning_rate": 1.2783886577228986e-05, + "loss": 0.1965, + "step": 8175 + }, + { + "epoch": 3.42, + "learning_rate": 1.2782258150389387e-05, + "loss": 0.1853, + "step": 8176 + }, + { + "epoch": 3.42, + "learning_rate": 1.2780629643576249e-05, + "loss": 0.2044, + "step": 8177 + }, + { + "epoch": 3.42, + "learning_rate": 1.2779001056836373e-05, + "loss": 0.1349, + "step": 8178 + }, + { + "epoch": 3.42, + "learning_rate": 1.2777372390216585e-05, + "loss": 0.1896, + "step": 8179 + }, + { + "epoch": 3.42, + "learning_rate": 1.2775743643763691e-05, + "loss": 0.1824, + "step": 8180 + }, + { + "epoch": 3.43, + "learning_rate": 1.2774114817524505e-05, + "loss": 0.2179, + "step": 8181 + }, + { + "epoch": 3.43, + "learning_rate": 1.2772485911545854e-05, + "loss": 0.1746, + "step": 8182 + }, + { + "epoch": 3.43, + "learning_rate": 1.2770856925874555e-05, + "loss": 0.1649, + "step": 8183 + }, + { + "epoch": 3.43, + "learning_rate": 1.2769227860557431e-05, + "loss": 0.1964, + "step": 8184 + }, + { + "epoch": 3.43, + "learning_rate": 1.2767598715641309e-05, + "loss": 0.1795, + "step": 8185 + }, + { + "epoch": 3.43, + "learning_rate": 1.2765969491173023e-05, + "loss": 0.1827, + "step": 8186 + }, + { + "epoch": 3.43, + "learning_rate": 1.2764340187199393e-05, + "loss": 0.1434, + "step": 8187 + }, + { + "epoch": 3.43, + "learning_rate": 1.2762710803767258e-05, + "loss": 0.2034, + "step": 8188 + }, + { + "epoch": 3.43, + "learning_rate": 1.2761081340923455e-05, + "loss": 0.1645, + "step": 8189 + }, + { + "epoch": 3.43, + "learning_rate": 1.2759451798714816e-05, + "loss": 0.1894, + "step": 8190 + }, + { + "epoch": 3.43, + "learning_rate": 1.2757822177188187e-05, + "loss": 0.1838, + "step": 8191 + }, + { + "epoch": 3.43, + "learning_rate": 1.2756192476390406e-05, + "loss": 0.2094, + "step": 8192 + }, + { + "epoch": 3.43, + "learning_rate": 1.2754562696368318e-05, + "loss": 0.2115, + "step": 8193 + }, + { + "epoch": 3.43, + "learning_rate": 1.2752932837168768e-05, + "loss": 0.1835, + "step": 8194 + }, + { + "epoch": 3.43, + "learning_rate": 1.2751302898838607e-05, + "loss": 0.1596, + "step": 8195 + }, + { + "epoch": 3.43, + "learning_rate": 1.2749672881424687e-05, + "loss": 0.1958, + "step": 8196 + }, + { + "epoch": 3.43, + "learning_rate": 1.2748042784973857e-05, + "loss": 0.1959, + "step": 8197 + }, + { + "epoch": 3.43, + "learning_rate": 1.274641260953298e-05, + "loss": 0.2248, + "step": 8198 + }, + { + "epoch": 3.43, + "learning_rate": 1.2744782355148905e-05, + "loss": 0.2259, + "step": 8199 + }, + { + "epoch": 3.43, + "learning_rate": 1.27431520218685e-05, + "loss": 0.2132, + "step": 8200 + }, + { + "epoch": 3.43, + "learning_rate": 1.2741521609738623e-05, + "loss": 0.195, + "step": 8201 + }, + { + "epoch": 3.43, + "learning_rate": 1.273989111880614e-05, + "loss": 0.1968, + "step": 8202 + }, + { + "epoch": 3.43, + "learning_rate": 1.273826054911792e-05, + "loss": 0.2022, + "step": 8203 + }, + { + "epoch": 3.43, + "learning_rate": 1.2736629900720832e-05, + "loss": 0.2075, + "step": 8204 + }, + { + "epoch": 3.44, + "learning_rate": 1.2734999173661742e-05, + "loss": 0.173, + "step": 8205 + }, + { + "epoch": 3.44, + "learning_rate": 1.2733368367987528e-05, + "loss": 0.1718, + "step": 8206 + }, + { + "epoch": 3.44, + "learning_rate": 1.2731737483745068e-05, + "loss": 0.1859, + "step": 8207 + }, + { + "epoch": 3.44, + "learning_rate": 1.2730106520981236e-05, + "loss": 0.207, + "step": 8208 + }, + { + "epoch": 3.44, + "learning_rate": 1.2728475479742914e-05, + "loss": 0.1998, + "step": 8209 + }, + { + "epoch": 3.44, + "learning_rate": 1.272684436007699e-05, + "loss": 0.1945, + "step": 8210 + }, + { + "epoch": 3.44, + "learning_rate": 1.272521316203034e-05, + "loss": 0.1999, + "step": 8211 + }, + { + "epoch": 3.44, + "learning_rate": 1.2723581885649858e-05, + "loss": 0.1899, + "step": 8212 + }, + { + "epoch": 3.44, + "learning_rate": 1.2721950530982431e-05, + "loss": 0.1733, + "step": 8213 + }, + { + "epoch": 3.44, + "learning_rate": 1.2720319098074954e-05, + "loss": 0.2122, + "step": 8214 + }, + { + "epoch": 3.44, + "learning_rate": 1.2718687586974317e-05, + "loss": 0.1633, + "step": 8215 + }, + { + "epoch": 3.44, + "learning_rate": 1.2717055997727416e-05, + "loss": 0.1858, + "step": 8216 + }, + { + "epoch": 3.44, + "learning_rate": 1.2715424330381155e-05, + "loss": 0.1657, + "step": 8217 + }, + { + "epoch": 3.44, + "learning_rate": 1.2713792584982426e-05, + "loss": 0.2026, + "step": 8218 + }, + { + "epoch": 3.44, + "learning_rate": 1.2712160761578138e-05, + "loss": 0.1809, + "step": 8219 + }, + { + "epoch": 3.44, + "learning_rate": 1.27105288602152e-05, + "loss": 0.1778, + "step": 8220 + }, + { + "epoch": 3.44, + "learning_rate": 1.2708896880940513e-05, + "loss": 0.2055, + "step": 8221 + }, + { + "epoch": 3.44, + "learning_rate": 1.270726482380099e-05, + "loss": 0.211, + "step": 8222 + }, + { + "epoch": 3.44, + "learning_rate": 1.270563268884354e-05, + "loss": 0.1524, + "step": 8223 + }, + { + "epoch": 3.44, + "learning_rate": 1.2704000476115079e-05, + "loss": 0.1888, + "step": 8224 + }, + { + "epoch": 3.44, + "learning_rate": 1.2702368185662529e-05, + "loss": 0.1963, + "step": 8225 + }, + { + "epoch": 3.44, + "learning_rate": 1.2700735817532796e-05, + "loss": 0.1832, + "step": 8226 + }, + { + "epoch": 3.44, + "learning_rate": 1.2699103371772815e-05, + "loss": 0.1948, + "step": 8227 + }, + { + "epoch": 3.44, + "learning_rate": 1.26974708484295e-05, + "loss": 0.2054, + "step": 8228 + }, + { + "epoch": 3.45, + "learning_rate": 1.2695838247549777e-05, + "loss": 0.1572, + "step": 8229 + }, + { + "epoch": 3.45, + "learning_rate": 1.2694205569180577e-05, + "loss": 0.1622, + "step": 8230 + }, + { + "epoch": 3.45, + "learning_rate": 1.269257281336883e-05, + "loss": 0.1943, + "step": 8231 + }, + { + "epoch": 3.45, + "learning_rate": 1.2690939980161465e-05, + "loss": 0.1662, + "step": 8232 + }, + { + "epoch": 3.45, + "learning_rate": 1.2689307069605418e-05, + "loss": 0.1833, + "step": 8233 + }, + { + "epoch": 3.45, + "learning_rate": 1.268767408174763e-05, + "loss": 0.2034, + "step": 8234 + }, + { + "epoch": 3.45, + "learning_rate": 1.2686041016635031e-05, + "loss": 0.1898, + "step": 8235 + }, + { + "epoch": 3.45, + "learning_rate": 1.2684407874314568e-05, + "loss": 0.1917, + "step": 8236 + }, + { + "epoch": 3.45, + "learning_rate": 1.2682774654833188e-05, + "loss": 0.1471, + "step": 8237 + }, + { + "epoch": 3.45, + "learning_rate": 1.2681141358237823e-05, + "loss": 0.1666, + "step": 8238 + }, + { + "epoch": 3.45, + "learning_rate": 1.2679507984575431e-05, + "loss": 0.162, + "step": 8239 + }, + { + "epoch": 3.45, + "learning_rate": 1.2677874533892963e-05, + "loss": 0.18, + "step": 8240 + }, + { + "epoch": 3.45, + "learning_rate": 1.2676241006237367e-05, + "loss": 0.1862, + "step": 8241 + }, + { + "epoch": 3.45, + "learning_rate": 1.2674607401655595e-05, + "loss": 0.1574, + "step": 8242 + }, + { + "epoch": 3.45, + "learning_rate": 1.2672973720194608e-05, + "loss": 0.1919, + "step": 8243 + }, + { + "epoch": 3.45, + "learning_rate": 1.2671339961901366e-05, + "loss": 0.2196, + "step": 8244 + }, + { + "epoch": 3.45, + "learning_rate": 1.2669706126822823e-05, + "loss": 0.1901, + "step": 8245 + }, + { + "epoch": 3.45, + "learning_rate": 1.2668072215005946e-05, + "loss": 0.1999, + "step": 8246 + }, + { + "epoch": 3.45, + "learning_rate": 1.2666438226497706e-05, + "loss": 0.2016, + "step": 8247 + }, + { + "epoch": 3.45, + "learning_rate": 1.2664804161345064e-05, + "loss": 0.2066, + "step": 8248 + }, + { + "epoch": 3.45, + "learning_rate": 1.2663170019594988e-05, + "loss": 0.1709, + "step": 8249 + }, + { + "epoch": 3.45, + "learning_rate": 1.2661535801294455e-05, + "loss": 0.1545, + "step": 8250 + }, + { + "epoch": 3.45, + "learning_rate": 1.2659901506490437e-05, + "loss": 0.1295, + "step": 8251 + }, + { + "epoch": 3.45, + "learning_rate": 1.2658267135229911e-05, + "loss": 0.1584, + "step": 8252 + }, + { + "epoch": 3.46, + "learning_rate": 1.2656632687559853e-05, + "loss": 0.1862, + "step": 8253 + }, + { + "epoch": 3.46, + "learning_rate": 1.265499816352725e-05, + "loss": 0.1749, + "step": 8254 + }, + { + "epoch": 3.46, + "learning_rate": 1.2653363563179078e-05, + "loss": 0.1752, + "step": 8255 + }, + { + "epoch": 3.46, + "learning_rate": 1.2651728886562324e-05, + "loss": 0.178, + "step": 8256 + }, + { + "epoch": 3.46, + "learning_rate": 1.2650094133723978e-05, + "loss": 0.1874, + "step": 8257 + }, + { + "epoch": 3.46, + "learning_rate": 1.2648459304711026e-05, + "loss": 0.1988, + "step": 8258 + }, + { + "epoch": 3.46, + "learning_rate": 1.2646824399570466e-05, + "loss": 0.1609, + "step": 8259 + }, + { + "epoch": 3.46, + "learning_rate": 1.2645189418349282e-05, + "loss": 0.2107, + "step": 8260 + }, + { + "epoch": 3.46, + "learning_rate": 1.2643554361094478e-05, + "loss": 0.2183, + "step": 8261 + }, + { + "epoch": 3.46, + "learning_rate": 1.264191922785305e-05, + "loss": 0.201, + "step": 8262 + }, + { + "epoch": 3.46, + "learning_rate": 1.2640284018671995e-05, + "loss": 0.1965, + "step": 8263 + }, + { + "epoch": 3.46, + "learning_rate": 1.2638648733598325e-05, + "loss": 0.1527, + "step": 8264 + }, + { + "epoch": 3.46, + "learning_rate": 1.2637013372679032e-05, + "loss": 0.188, + "step": 8265 + }, + { + "epoch": 3.46, + "learning_rate": 1.2635377935961134e-05, + "loss": 0.1662, + "step": 8266 + }, + { + "epoch": 3.46, + "learning_rate": 1.2633742423491636e-05, + "loss": 0.1792, + "step": 8267 + }, + { + "epoch": 3.46, + "learning_rate": 1.2632106835317545e-05, + "loss": 0.1997, + "step": 8268 + }, + { + "epoch": 3.46, + "learning_rate": 1.263047117148588e-05, + "loss": 0.1857, + "step": 8269 + }, + { + "epoch": 3.46, + "learning_rate": 1.2628835432043655e-05, + "loss": 0.2021, + "step": 8270 + }, + { + "epoch": 3.46, + "learning_rate": 1.2627199617037895e-05, + "loss": 0.1827, + "step": 8271 + }, + { + "epoch": 3.46, + "learning_rate": 1.2625563726515609e-05, + "loss": 0.2168, + "step": 8272 + }, + { + "epoch": 3.46, + "learning_rate": 1.2623927760523822e-05, + "loss": 0.2037, + "step": 8273 + }, + { + "epoch": 3.46, + "learning_rate": 1.2622291719109559e-05, + "loss": 0.1773, + "step": 8274 + }, + { + "epoch": 3.46, + "learning_rate": 1.2620655602319851e-05, + "loss": 0.1975, + "step": 8275 + }, + { + "epoch": 3.46, + "learning_rate": 1.261901941020172e-05, + "loss": 0.1958, + "step": 8276 + }, + { + "epoch": 3.47, + "learning_rate": 1.2617383142802202e-05, + "loss": 0.2037, + "step": 8277 + }, + { + "epoch": 3.47, + "learning_rate": 1.261574680016833e-05, + "loss": 0.166, + "step": 8278 + }, + { + "epoch": 3.47, + "learning_rate": 1.2614110382347133e-05, + "loss": 0.2285, + "step": 8279 + }, + { + "epoch": 3.47, + "learning_rate": 1.2612473889385655e-05, + "loss": 0.1713, + "step": 8280 + }, + { + "epoch": 3.47, + "learning_rate": 1.2610837321330935e-05, + "loss": 0.1785, + "step": 8281 + }, + { + "epoch": 3.47, + "learning_rate": 1.2609200678230014e-05, + "loss": 0.1969, + "step": 8282 + }, + { + "epoch": 3.47, + "learning_rate": 1.260756396012993e-05, + "loss": 0.2023, + "step": 8283 + }, + { + "epoch": 3.47, + "learning_rate": 1.2605927167077736e-05, + "loss": 0.2079, + "step": 8284 + }, + { + "epoch": 3.47, + "learning_rate": 1.2604290299120478e-05, + "loss": 0.1478, + "step": 8285 + }, + { + "epoch": 3.47, + "learning_rate": 1.2602653356305207e-05, + "loss": 0.1633, + "step": 8286 + }, + { + "epoch": 3.47, + "learning_rate": 1.2601016338678976e-05, + "loss": 0.2344, + "step": 8287 + }, + { + "epoch": 3.47, + "learning_rate": 1.2599379246288837e-05, + "loss": 0.179, + "step": 8288 + }, + { + "epoch": 3.47, + "learning_rate": 1.2597742079181847e-05, + "loss": 0.2131, + "step": 8289 + }, + { + "epoch": 3.47, + "learning_rate": 1.2596104837405065e-05, + "loss": 0.1779, + "step": 8290 + }, + { + "epoch": 3.47, + "learning_rate": 1.2594467521005557e-05, + "loss": 0.203, + "step": 8291 + }, + { + "epoch": 3.47, + "learning_rate": 1.259283013003038e-05, + "loss": 0.1678, + "step": 8292 + }, + { + "epoch": 3.47, + "learning_rate": 1.2591192664526606e-05, + "loss": 0.1857, + "step": 8293 + }, + { + "epoch": 3.47, + "learning_rate": 1.2589555124541293e-05, + "loss": 0.1401, + "step": 8294 + }, + { + "epoch": 3.47, + "learning_rate": 1.2587917510121518e-05, + "loss": 0.1979, + "step": 8295 + }, + { + "epoch": 3.47, + "learning_rate": 1.2586279821314351e-05, + "loss": 0.1971, + "step": 8296 + }, + { + "epoch": 3.47, + "learning_rate": 1.2584642058166863e-05, + "loss": 0.1511, + "step": 8297 + }, + { + "epoch": 3.47, + "learning_rate": 1.2583004220726136e-05, + "loss": 0.1771, + "step": 8298 + }, + { + "epoch": 3.47, + "learning_rate": 1.2581366309039242e-05, + "loss": 0.1956, + "step": 8299 + }, + { + "epoch": 3.47, + "learning_rate": 1.2579728323153267e-05, + "loss": 0.1676, + "step": 8300 + }, + { + "epoch": 3.48, + "learning_rate": 1.257809026311529e-05, + "loss": 0.1678, + "step": 8301 + }, + { + "epoch": 3.48, + "learning_rate": 1.2576452128972394e-05, + "loss": 0.2065, + "step": 8302 + }, + { + "epoch": 3.48, + "learning_rate": 1.2574813920771667e-05, + "loss": 0.1917, + "step": 8303 + }, + { + "epoch": 3.48, + "learning_rate": 1.25731756385602e-05, + "loss": 0.1817, + "step": 8304 + }, + { + "epoch": 3.48, + "learning_rate": 1.2571537282385087e-05, + "loss": 0.2061, + "step": 8305 + }, + { + "epoch": 3.48, + "learning_rate": 1.2569898852293412e-05, + "loss": 0.1832, + "step": 8306 + }, + { + "epoch": 3.48, + "learning_rate": 1.2568260348332276e-05, + "loss": 0.1655, + "step": 8307 + }, + { + "epoch": 3.48, + "learning_rate": 1.2566621770548777e-05, + "loss": 0.1663, + "step": 8308 + }, + { + "epoch": 3.48, + "learning_rate": 1.256498311899001e-05, + "loss": 0.1934, + "step": 8309 + }, + { + "epoch": 3.48, + "learning_rate": 1.2563344393703081e-05, + "loss": 0.1964, + "step": 8310 + }, + { + "epoch": 3.48, + "learning_rate": 1.256170559473509e-05, + "loss": 0.1858, + "step": 8311 + }, + { + "epoch": 3.48, + "learning_rate": 1.2560066722133146e-05, + "loss": 0.1925, + "step": 8312 + }, + { + "epoch": 3.48, + "learning_rate": 1.2558427775944357e-05, + "loss": 0.2114, + "step": 8313 + }, + { + "epoch": 3.48, + "learning_rate": 1.255678875621583e-05, + "loss": 0.1507, + "step": 8314 + }, + { + "epoch": 3.48, + "learning_rate": 1.2555149662994684e-05, + "loss": 0.1966, + "step": 8315 + }, + { + "epoch": 3.48, + "learning_rate": 1.2553510496328025e-05, + "loss": 0.1924, + "step": 8316 + }, + { + "epoch": 3.48, + "learning_rate": 1.255187125626297e-05, + "loss": 0.1676, + "step": 8317 + }, + { + "epoch": 3.48, + "learning_rate": 1.2550231942846643e-05, + "loss": 0.1972, + "step": 8318 + }, + { + "epoch": 3.48, + "learning_rate": 1.2548592556126163e-05, + "loss": 0.1806, + "step": 8319 + }, + { + "epoch": 3.48, + "learning_rate": 1.2546953096148648e-05, + "loss": 0.18, + "step": 8320 + }, + { + "epoch": 3.48, + "learning_rate": 1.2545313562961228e-05, + "loss": 0.201, + "step": 8321 + }, + { + "epoch": 3.48, + "learning_rate": 1.2543673956611029e-05, + "loss": 0.1876, + "step": 8322 + }, + { + "epoch": 3.48, + "learning_rate": 1.2542034277145178e-05, + "loss": 0.1419, + "step": 8323 + }, + { + "epoch": 3.49, + "learning_rate": 1.2540394524610808e-05, + "loss": 0.1911, + "step": 8324 + }, + { + "epoch": 3.49, + "learning_rate": 1.2538754699055057e-05, + "loss": 0.1678, + "step": 8325 + }, + { + "epoch": 3.49, + "learning_rate": 1.2537114800525047e-05, + "loss": 0.2095, + "step": 8326 + }, + { + "epoch": 3.49, + "learning_rate": 1.2535474829067928e-05, + "loss": 0.2091, + "step": 8327 + }, + { + "epoch": 3.49, + "learning_rate": 1.2533834784730835e-05, + "loss": 0.191, + "step": 8328 + }, + { + "epoch": 3.49, + "learning_rate": 1.2532194667560908e-05, + "loss": 0.2376, + "step": 8329 + }, + { + "epoch": 3.49, + "learning_rate": 1.253055447760529e-05, + "loss": 0.1688, + "step": 8330 + }, + { + "epoch": 3.49, + "learning_rate": 1.2528914214911132e-05, + "loss": 0.2203, + "step": 8331 + }, + { + "epoch": 3.49, + "learning_rate": 1.2527273879525579e-05, + "loss": 0.1921, + "step": 8332 + }, + { + "epoch": 3.49, + "learning_rate": 1.2525633471495782e-05, + "loss": 0.1735, + "step": 8333 + }, + { + "epoch": 3.49, + "learning_rate": 1.252399299086889e-05, + "loss": 0.1715, + "step": 8334 + }, + { + "epoch": 3.49, + "learning_rate": 1.252235243769206e-05, + "loss": 0.2148, + "step": 8335 + }, + { + "epoch": 3.49, + "learning_rate": 1.2520711812012447e-05, + "loss": 0.1942, + "step": 8336 + }, + { + "epoch": 3.49, + "learning_rate": 1.2519071113877211e-05, + "loss": 0.2036, + "step": 8337 + }, + { + "epoch": 3.49, + "learning_rate": 1.2517430343333508e-05, + "loss": 0.1769, + "step": 8338 + }, + { + "epoch": 3.49, + "learning_rate": 1.251578950042851e-05, + "loss": 0.1718, + "step": 8339 + }, + { + "epoch": 3.49, + "learning_rate": 1.2514148585209372e-05, + "loss": 0.1998, + "step": 8340 + }, + { + "epoch": 3.49, + "learning_rate": 1.251250759772326e-05, + "loss": 0.1851, + "step": 8341 + }, + { + "epoch": 3.49, + "learning_rate": 1.2510866538017352e-05, + "loss": 0.2033, + "step": 8342 + }, + { + "epoch": 3.49, + "learning_rate": 1.2509225406138809e-05, + "loss": 0.1818, + "step": 8343 + }, + { + "epoch": 3.49, + "learning_rate": 1.2507584202134812e-05, + "loss": 0.2065, + "step": 8344 + }, + { + "epoch": 3.49, + "learning_rate": 1.250594292605253e-05, + "loss": 0.1722, + "step": 8345 + }, + { + "epoch": 3.49, + "learning_rate": 1.2504301577939146e-05, + "loss": 0.2298, + "step": 8346 + }, + { + "epoch": 3.49, + "learning_rate": 1.250266015784183e-05, + "loss": 0.2453, + "step": 8347 + }, + { + "epoch": 3.5, + "learning_rate": 1.2501018665807772e-05, + "loss": 0.1601, + "step": 8348 + }, + { + "epoch": 3.5, + "learning_rate": 1.2499377101884153e-05, + "loss": 0.1856, + "step": 8349 + }, + { + "epoch": 3.5, + "learning_rate": 1.2497735466118156e-05, + "loss": 0.224, + "step": 8350 + }, + { + "epoch": 3.5, + "learning_rate": 1.249609375855697e-05, + "loss": 0.2147, + "step": 8351 + }, + { + "epoch": 3.5, + "learning_rate": 1.2494451979247782e-05, + "loss": 0.2016, + "step": 8352 + }, + { + "epoch": 3.5, + "learning_rate": 1.2492810128237787e-05, + "loss": 0.1871, + "step": 8353 + }, + { + "epoch": 3.5, + "learning_rate": 1.2491168205574175e-05, + "loss": 0.2007, + "step": 8354 + }, + { + "epoch": 3.5, + "learning_rate": 1.2489526211304143e-05, + "loss": 0.1742, + "step": 8355 + }, + { + "epoch": 3.5, + "learning_rate": 1.2487884145474894e-05, + "loss": 0.1818, + "step": 8356 + }, + { + "epoch": 3.5, + "learning_rate": 1.2486242008133618e-05, + "loss": 0.1848, + "step": 8357 + }, + { + "epoch": 3.5, + "learning_rate": 1.2484599799327524e-05, + "loss": 0.1861, + "step": 8358 + }, + { + "epoch": 3.5, + "learning_rate": 1.2482957519103813e-05, + "loss": 0.1962, + "step": 8359 + }, + { + "epoch": 3.5, + "learning_rate": 1.2481315167509691e-05, + "loss": 0.163, + "step": 8360 + }, + { + "epoch": 3.5, + "learning_rate": 1.2479672744592368e-05, + "loss": 0.1704, + "step": 8361 + }, + { + "epoch": 3.5, + "learning_rate": 1.2478030250399052e-05, + "loss": 0.1996, + "step": 8362 + }, + { + "epoch": 3.5, + "learning_rate": 1.2476387684976956e-05, + "loss": 0.1665, + "step": 8363 + }, + { + "epoch": 3.5, + "learning_rate": 1.2474745048373292e-05, + "loss": 0.1742, + "step": 8364 + }, + { + "epoch": 3.5, + "learning_rate": 1.2473102340635275e-05, + "loss": 0.1708, + "step": 8365 + }, + { + "epoch": 3.5, + "learning_rate": 1.247145956181013e-05, + "loss": 0.2358, + "step": 8366 + }, + { + "epoch": 3.5, + "learning_rate": 1.246981671194507e-05, + "loss": 0.1767, + "step": 8367 + }, + { + "epoch": 3.5, + "learning_rate": 1.2468173791087324e-05, + "loss": 0.1535, + "step": 8368 + }, + { + "epoch": 3.5, + "learning_rate": 1.2466530799284111e-05, + "loss": 0.1793, + "step": 8369 + }, + { + "epoch": 3.5, + "learning_rate": 1.2464887736582659e-05, + "loss": 0.1774, + "step": 8370 + }, + { + "epoch": 3.5, + "learning_rate": 1.2463244603030194e-05, + "loss": 0.1839, + "step": 8371 + }, + { + "epoch": 3.51, + "learning_rate": 1.2461601398673953e-05, + "loss": 0.2015, + "step": 8372 + }, + { + "epoch": 3.51, + "learning_rate": 1.2459958123561163e-05, + "loss": 0.1921, + "step": 8373 + }, + { + "epoch": 3.51, + "learning_rate": 1.2458314777739056e-05, + "loss": 0.1948, + "step": 8374 + }, + { + "epoch": 3.51, + "learning_rate": 1.2456671361254876e-05, + "loss": 0.2294, + "step": 8375 + }, + { + "epoch": 3.51, + "learning_rate": 1.2455027874155856e-05, + "loss": 0.193, + "step": 8376 + }, + { + "epoch": 3.51, + "learning_rate": 1.2453384316489239e-05, + "loss": 0.1784, + "step": 8377 + }, + { + "epoch": 3.51, + "learning_rate": 1.2451740688302265e-05, + "loss": 0.1754, + "step": 8378 + }, + { + "epoch": 3.51, + "learning_rate": 1.2450096989642184e-05, + "loss": 0.1777, + "step": 8379 + }, + { + "epoch": 3.51, + "learning_rate": 1.2448453220556235e-05, + "loss": 0.1634, + "step": 8380 + }, + { + "epoch": 3.51, + "learning_rate": 1.2446809381091674e-05, + "loss": 0.1739, + "step": 8381 + }, + { + "epoch": 3.51, + "learning_rate": 1.2445165471295745e-05, + "loss": 0.1825, + "step": 8382 + }, + { + "epoch": 3.51, + "learning_rate": 1.244352149121571e-05, + "loss": 0.1394, + "step": 8383 + }, + { + "epoch": 3.51, + "learning_rate": 1.2441877440898817e-05, + "loss": 0.1721, + "step": 8384 + }, + { + "epoch": 3.51, + "learning_rate": 1.2440233320392322e-05, + "loss": 0.1992, + "step": 8385 + }, + { + "epoch": 3.51, + "learning_rate": 1.2438589129743484e-05, + "loss": 0.1969, + "step": 8386 + }, + { + "epoch": 3.51, + "learning_rate": 1.243694486899957e-05, + "loss": 0.214, + "step": 8387 + }, + { + "epoch": 3.51, + "learning_rate": 1.2435300538207832e-05, + "loss": 0.1699, + "step": 8388 + }, + { + "epoch": 3.51, + "learning_rate": 1.2433656137415545e-05, + "loss": 0.1888, + "step": 8389 + }, + { + "epoch": 3.51, + "learning_rate": 1.2432011666669972e-05, + "loss": 0.1743, + "step": 8390 + }, + { + "epoch": 3.51, + "learning_rate": 1.243036712601838e-05, + "loss": 0.1822, + "step": 8391 + }, + { + "epoch": 3.51, + "learning_rate": 1.2428722515508043e-05, + "loss": 0.1835, + "step": 8392 + }, + { + "epoch": 3.51, + "learning_rate": 1.2427077835186235e-05, + "loss": 0.2189, + "step": 8393 + }, + { + "epoch": 3.51, + "learning_rate": 1.2425433085100224e-05, + "loss": 0.1872, + "step": 8394 + }, + { + "epoch": 3.51, + "learning_rate": 1.2423788265297294e-05, + "loss": 0.2121, + "step": 8395 + }, + { + "epoch": 3.52, + "learning_rate": 1.242214337582472e-05, + "loss": 0.1972, + "step": 8396 + }, + { + "epoch": 3.52, + "learning_rate": 1.2420498416729785e-05, + "loss": 0.2004, + "step": 8397 + }, + { + "epoch": 3.52, + "learning_rate": 1.2418853388059771e-05, + "loss": 0.1933, + "step": 8398 + }, + { + "epoch": 3.52, + "learning_rate": 1.241720828986196e-05, + "loss": 0.18, + "step": 8399 + }, + { + "epoch": 3.52, + "learning_rate": 1.2415563122183644e-05, + "loss": 0.2336, + "step": 8400 + }, + { + "epoch": 3.52, + "learning_rate": 1.2413917885072109e-05, + "loss": 0.1966, + "step": 8401 + }, + { + "epoch": 3.52, + "learning_rate": 1.2412272578574647e-05, + "loss": 0.2009, + "step": 8402 + }, + { + "epoch": 3.52, + "learning_rate": 1.241062720273855e-05, + "loss": 0.1812, + "step": 8403 + }, + { + "epoch": 3.52, + "learning_rate": 1.2408981757611112e-05, + "loss": 0.2079, + "step": 8404 + }, + { + "epoch": 3.52, + "learning_rate": 1.2407336243239632e-05, + "loss": 0.1831, + "step": 8405 + }, + { + "epoch": 3.52, + "learning_rate": 1.240569065967141e-05, + "loss": 0.2116, + "step": 8406 + }, + { + "epoch": 3.52, + "learning_rate": 1.2404045006953744e-05, + "loss": 0.1673, + "step": 8407 + }, + { + "epoch": 3.52, + "learning_rate": 1.2402399285133935e-05, + "loss": 0.2214, + "step": 8408 + }, + { + "epoch": 3.52, + "learning_rate": 1.2400753494259293e-05, + "loss": 0.1845, + "step": 8409 + }, + { + "epoch": 3.52, + "learning_rate": 1.2399107634377122e-05, + "loss": 0.1506, + "step": 8410 + }, + { + "epoch": 3.52, + "learning_rate": 1.2397461705534727e-05, + "loss": 0.1963, + "step": 8411 + }, + { + "epoch": 3.52, + "learning_rate": 1.2395815707779426e-05, + "loss": 0.2047, + "step": 8412 + }, + { + "epoch": 3.52, + "learning_rate": 1.239416964115853e-05, + "loss": 0.2205, + "step": 8413 + }, + { + "epoch": 3.52, + "learning_rate": 1.2392523505719349e-05, + "loss": 0.1948, + "step": 8414 + }, + { + "epoch": 3.52, + "learning_rate": 1.2390877301509205e-05, + "loss": 0.2205, + "step": 8415 + }, + { + "epoch": 3.52, + "learning_rate": 1.2389231028575412e-05, + "loss": 0.1596, + "step": 8416 + }, + { + "epoch": 3.52, + "learning_rate": 1.2387584686965297e-05, + "loss": 0.2314, + "step": 8417 + }, + { + "epoch": 3.52, + "learning_rate": 1.2385938276726179e-05, + "loss": 0.1764, + "step": 8418 + }, + { + "epoch": 3.52, + "learning_rate": 1.2384291797905382e-05, + "loss": 0.1856, + "step": 8419 + }, + { + "epoch": 3.53, + "learning_rate": 1.2382645250550232e-05, + "loss": 0.2062, + "step": 8420 + }, + { + "epoch": 3.53, + "learning_rate": 1.2380998634708057e-05, + "loss": 0.2007, + "step": 8421 + }, + { + "epoch": 3.53, + "learning_rate": 1.2379351950426188e-05, + "loss": 0.2188, + "step": 8422 + }, + { + "epoch": 3.53, + "learning_rate": 1.2377705197751961e-05, + "loss": 0.212, + "step": 8423 + }, + { + "epoch": 3.53, + "learning_rate": 1.237605837673271e-05, + "loss": 0.2021, + "step": 8424 + }, + { + "epoch": 3.53, + "learning_rate": 1.2374411487415766e-05, + "loss": 0.1813, + "step": 8425 + }, + { + "epoch": 3.53, + "learning_rate": 1.2372764529848471e-05, + "loss": 0.1811, + "step": 8426 + }, + { + "epoch": 3.53, + "learning_rate": 1.237111750407817e-05, + "loss": 0.1925, + "step": 8427 + }, + { + "epoch": 3.53, + "learning_rate": 1.2369470410152195e-05, + "loss": 0.1922, + "step": 8428 + }, + { + "epoch": 3.53, + "learning_rate": 1.2367823248117899e-05, + "loss": 0.2002, + "step": 8429 + }, + { + "epoch": 3.53, + "learning_rate": 1.2366176018022621e-05, + "loss": 0.2103, + "step": 8430 + }, + { + "epoch": 3.53, + "learning_rate": 1.2364528719913717e-05, + "loss": 0.1977, + "step": 8431 + }, + { + "epoch": 3.53, + "learning_rate": 1.2362881353838532e-05, + "loss": 0.1977, + "step": 8432 + }, + { + "epoch": 3.53, + "learning_rate": 1.236123391984442e-05, + "loss": 0.1646, + "step": 8433 + }, + { + "epoch": 3.53, + "learning_rate": 1.2359586417978733e-05, + "loss": 0.1623, + "step": 8434 + }, + { + "epoch": 3.53, + "learning_rate": 1.2357938848288829e-05, + "loss": 0.2135, + "step": 8435 + }, + { + "epoch": 3.53, + "learning_rate": 1.2356291210822063e-05, + "loss": 0.2054, + "step": 8436 + }, + { + "epoch": 3.53, + "learning_rate": 1.23546435056258e-05, + "loss": 0.1851, + "step": 8437 + }, + { + "epoch": 3.53, + "learning_rate": 1.2352995732747398e-05, + "loss": 0.2118, + "step": 8438 + }, + { + "epoch": 3.53, + "learning_rate": 1.235134789223422e-05, + "loss": 0.2034, + "step": 8439 + }, + { + "epoch": 3.53, + "learning_rate": 1.2349699984133636e-05, + "loss": 0.1801, + "step": 8440 + }, + { + "epoch": 3.53, + "learning_rate": 1.234805200849301e-05, + "loss": 0.1968, + "step": 8441 + }, + { + "epoch": 3.53, + "learning_rate": 1.2346403965359712e-05, + "loss": 0.1928, + "step": 8442 + }, + { + "epoch": 3.53, + "learning_rate": 1.2344755854781114e-05, + "loss": 0.1793, + "step": 8443 + }, + { + "epoch": 3.54, + "learning_rate": 1.2343107676804593e-05, + "loss": 0.1609, + "step": 8444 + }, + { + "epoch": 3.54, + "learning_rate": 1.2341459431477516e-05, + "loss": 0.1669, + "step": 8445 + }, + { + "epoch": 3.54, + "learning_rate": 1.2339811118847269e-05, + "loss": 0.1921, + "step": 8446 + }, + { + "epoch": 3.54, + "learning_rate": 1.2338162738961227e-05, + "loss": 0.1854, + "step": 8447 + }, + { + "epoch": 3.54, + "learning_rate": 1.233651429186677e-05, + "loss": 0.1917, + "step": 8448 + }, + { + "epoch": 3.54, + "learning_rate": 1.2334865777611284e-05, + "loss": 0.1943, + "step": 8449 + }, + { + "epoch": 3.54, + "learning_rate": 1.2333217196242152e-05, + "loss": 0.1806, + "step": 8450 + }, + { + "epoch": 3.54, + "learning_rate": 1.2331568547806768e-05, + "loss": 0.1831, + "step": 8451 + }, + { + "epoch": 3.54, + "learning_rate": 1.232991983235251e-05, + "loss": 0.2109, + "step": 8452 + }, + { + "epoch": 3.54, + "learning_rate": 1.2328271049926774e-05, + "loss": 0.213, + "step": 8453 + }, + { + "epoch": 3.54, + "learning_rate": 1.2326622200576956e-05, + "loss": 0.1713, + "step": 8454 + }, + { + "epoch": 3.54, + "learning_rate": 1.2324973284350444e-05, + "loss": 0.2101, + "step": 8455 + }, + { + "epoch": 3.54, + "learning_rate": 1.2323324301294637e-05, + "loss": 0.1893, + "step": 8456 + }, + { + "epoch": 3.54, + "learning_rate": 1.2321675251456937e-05, + "loss": 0.1919, + "step": 8457 + }, + { + "epoch": 3.54, + "learning_rate": 1.2320026134884744e-05, + "loss": 0.1883, + "step": 8458 + }, + { + "epoch": 3.54, + "learning_rate": 1.2318376951625455e-05, + "loss": 0.1351, + "step": 8459 + }, + { + "epoch": 3.54, + "learning_rate": 1.2316727701726479e-05, + "loss": 0.2197, + "step": 8460 + }, + { + "epoch": 3.54, + "learning_rate": 1.2315078385235223e-05, + "loss": 0.1724, + "step": 8461 + }, + { + "epoch": 3.54, + "learning_rate": 1.2313429002199088e-05, + "loss": 0.1657, + "step": 8462 + }, + { + "epoch": 3.54, + "learning_rate": 1.2311779552665495e-05, + "loss": 0.1867, + "step": 8463 + }, + { + "epoch": 3.54, + "learning_rate": 1.2310130036681847e-05, + "loss": 0.1593, + "step": 8464 + }, + { + "epoch": 3.54, + "learning_rate": 1.2308480454295563e-05, + "loss": 0.1857, + "step": 8465 + }, + { + "epoch": 3.54, + "learning_rate": 1.2306830805554055e-05, + "loss": 0.1832, + "step": 8466 + }, + { + "epoch": 3.54, + "learning_rate": 1.2305181090504744e-05, + "loss": 0.1751, + "step": 8467 + }, + { + "epoch": 3.55, + "learning_rate": 1.2303531309195047e-05, + "loss": 0.178, + "step": 8468 + }, + { + "epoch": 3.55, + "learning_rate": 1.2301881461672387e-05, + "loss": 0.1923, + "step": 8469 + }, + { + "epoch": 3.55, + "learning_rate": 1.2300231547984187e-05, + "loss": 0.1603, + "step": 8470 + }, + { + "epoch": 3.55, + "learning_rate": 1.2298581568177874e-05, + "loss": 0.2233, + "step": 8471 + }, + { + "epoch": 3.55, + "learning_rate": 1.229693152230087e-05, + "loss": 0.1936, + "step": 8472 + }, + { + "epoch": 3.55, + "learning_rate": 1.2295281410400611e-05, + "loss": 0.198, + "step": 8473 + }, + { + "epoch": 3.55, + "learning_rate": 1.2293631232524524e-05, + "loss": 0.1754, + "step": 8474 + }, + { + "epoch": 3.55, + "learning_rate": 1.2291980988720042e-05, + "loss": 0.156, + "step": 8475 + }, + { + "epoch": 3.55, + "learning_rate": 1.22903306790346e-05, + "loss": 0.192, + "step": 8476 + }, + { + "epoch": 3.55, + "learning_rate": 1.2288680303515634e-05, + "loss": 0.1959, + "step": 8477 + }, + { + "epoch": 3.55, + "learning_rate": 1.2287029862210587e-05, + "loss": 0.2003, + "step": 8478 + }, + { + "epoch": 3.55, + "learning_rate": 1.2285379355166893e-05, + "loss": 0.1684, + "step": 8479 + }, + { + "epoch": 3.55, + "learning_rate": 1.2283728782431996e-05, + "loss": 0.2383, + "step": 8480 + }, + { + "epoch": 3.55, + "learning_rate": 1.2282078144053347e-05, + "loss": 0.212, + "step": 8481 + }, + { + "epoch": 3.55, + "learning_rate": 1.2280427440078383e-05, + "loss": 0.1748, + "step": 8482 + }, + { + "epoch": 3.55, + "learning_rate": 1.2278776670554555e-05, + "loss": 0.1972, + "step": 8483 + }, + { + "epoch": 3.55, + "learning_rate": 1.2277125835529318e-05, + "loss": 0.1675, + "step": 8484 + }, + { + "epoch": 3.55, + "learning_rate": 1.2275474935050116e-05, + "loss": 0.184, + "step": 8485 + }, + { + "epoch": 3.55, + "learning_rate": 1.227382396916441e-05, + "loss": 0.1557, + "step": 8486 + }, + { + "epoch": 3.55, + "learning_rate": 1.2272172937919646e-05, + "loss": 0.2133, + "step": 8487 + }, + { + "epoch": 3.55, + "learning_rate": 1.2270521841363293e-05, + "loss": 0.2386, + "step": 8488 + }, + { + "epoch": 3.55, + "learning_rate": 1.2268870679542799e-05, + "loss": 0.1974, + "step": 8489 + }, + { + "epoch": 3.55, + "learning_rate": 1.2267219452505635e-05, + "loss": 0.1825, + "step": 8490 + }, + { + "epoch": 3.55, + "learning_rate": 1.226556816029926e-05, + "loss": 0.2028, + "step": 8491 + }, + { + "epoch": 3.56, + "learning_rate": 1.2263916802971134e-05, + "loss": 0.1763, + "step": 8492 + }, + { + "epoch": 3.56, + "learning_rate": 1.2262265380568729e-05, + "loss": 0.1288, + "step": 8493 + }, + { + "epoch": 3.56, + "learning_rate": 1.2260613893139517e-05, + "loss": 0.1676, + "step": 8494 + }, + { + "epoch": 3.56, + "learning_rate": 1.2258962340730965e-05, + "loss": 0.173, + "step": 8495 + }, + { + "epoch": 3.56, + "learning_rate": 1.2257310723390541e-05, + "loss": 0.1756, + "step": 8496 + }, + { + "epoch": 3.56, + "learning_rate": 1.2255659041165726e-05, + "loss": 0.1826, + "step": 8497 + }, + { + "epoch": 3.56, + "learning_rate": 1.2254007294103994e-05, + "loss": 0.2019, + "step": 8498 + }, + { + "epoch": 3.56, + "learning_rate": 1.2252355482252823e-05, + "loss": 0.2117, + "step": 8499 + }, + { + "epoch": 3.56, + "learning_rate": 1.2250703605659688e-05, + "loss": 0.1707, + "step": 8500 + }, + { + "epoch": 3.56, + "learning_rate": 1.2249051664372078e-05, + "loss": 0.1901, + "step": 8501 + }, + { + "epoch": 3.56, + "learning_rate": 1.2247399658437475e-05, + "loss": 0.2087, + "step": 8502 + }, + { + "epoch": 3.56, + "learning_rate": 1.224574758790336e-05, + "loss": 0.2222, + "step": 8503 + }, + { + "epoch": 3.56, + "learning_rate": 1.2244095452817227e-05, + "loss": 0.208, + "step": 8504 + }, + { + "epoch": 3.56, + "learning_rate": 1.2242443253226558e-05, + "loss": 0.2085, + "step": 8505 + }, + { + "epoch": 3.56, + "learning_rate": 1.224079098917885e-05, + "loss": 0.2235, + "step": 8506 + }, + { + "epoch": 3.56, + "learning_rate": 1.2239138660721593e-05, + "loss": 0.1747, + "step": 8507 + }, + { + "epoch": 3.56, + "learning_rate": 1.2237486267902284e-05, + "loss": 0.1916, + "step": 8508 + }, + { + "epoch": 3.56, + "learning_rate": 1.2235833810768417e-05, + "loss": 0.1648, + "step": 8509 + }, + { + "epoch": 3.56, + "learning_rate": 1.2234181289367489e-05, + "loss": 0.2035, + "step": 8510 + }, + { + "epoch": 3.56, + "learning_rate": 1.2232528703747002e-05, + "loss": 0.2268, + "step": 8511 + }, + { + "epoch": 3.56, + "learning_rate": 1.2230876053954462e-05, + "loss": 0.1868, + "step": 8512 + }, + { + "epoch": 3.56, + "learning_rate": 1.2229223340037364e-05, + "loss": 0.1953, + "step": 8513 + }, + { + "epoch": 3.56, + "learning_rate": 1.2227570562043223e-05, + "loss": 0.1992, + "step": 8514 + }, + { + "epoch": 3.56, + "learning_rate": 1.2225917720019544e-05, + "loss": 0.1824, + "step": 8515 + }, + { + "epoch": 3.57, + "learning_rate": 1.2224264814013833e-05, + "loss": 0.1768, + "step": 8516 + }, + { + "epoch": 3.57, + "learning_rate": 1.2222611844073604e-05, + "loss": 0.185, + "step": 8517 + }, + { + "epoch": 3.57, + "learning_rate": 1.2220958810246371e-05, + "loss": 0.1673, + "step": 8518 + }, + { + "epoch": 3.57, + "learning_rate": 1.2219305712579647e-05, + "loss": 0.1732, + "step": 8519 + }, + { + "epoch": 3.57, + "learning_rate": 1.221765255112095e-05, + "loss": 0.2109, + "step": 8520 + }, + { + "epoch": 3.57, + "learning_rate": 1.2215999325917797e-05, + "loss": 0.2072, + "step": 8521 + }, + { + "epoch": 3.57, + "learning_rate": 1.2214346037017715e-05, + "loss": 0.2434, + "step": 8522 + }, + { + "epoch": 3.57, + "learning_rate": 1.2212692684468215e-05, + "loss": 0.1588, + "step": 8523 + }, + { + "epoch": 3.57, + "learning_rate": 1.221103926831683e-05, + "loss": 0.2165, + "step": 8524 + }, + { + "epoch": 3.57, + "learning_rate": 1.2209385788611084e-05, + "loss": 0.2039, + "step": 8525 + }, + { + "epoch": 3.57, + "learning_rate": 1.2207732245398502e-05, + "loss": 0.1687, + "step": 8526 + }, + { + "epoch": 3.57, + "learning_rate": 1.2206078638726615e-05, + "loss": 0.1956, + "step": 8527 + }, + { + "epoch": 3.57, + "learning_rate": 1.2204424968642957e-05, + "loss": 0.2377, + "step": 8528 + }, + { + "epoch": 3.57, + "learning_rate": 1.220277123519506e-05, + "loss": 0.2237, + "step": 8529 + }, + { + "epoch": 3.57, + "learning_rate": 1.2201117438430456e-05, + "loss": 0.1893, + "step": 8530 + }, + { + "epoch": 3.57, + "learning_rate": 1.2199463578396688e-05, + "loss": 0.1917, + "step": 8531 + }, + { + "epoch": 3.57, + "learning_rate": 1.219780965514129e-05, + "loss": 0.2059, + "step": 8532 + }, + { + "epoch": 3.57, + "learning_rate": 1.2196155668711804e-05, + "loss": 0.1924, + "step": 8533 + }, + { + "epoch": 3.57, + "learning_rate": 1.219450161915577e-05, + "loss": 0.1509, + "step": 8534 + }, + { + "epoch": 3.57, + "learning_rate": 1.2192847506520735e-05, + "loss": 0.1846, + "step": 8535 + }, + { + "epoch": 3.57, + "learning_rate": 1.2191193330854246e-05, + "loss": 0.1921, + "step": 8536 + }, + { + "epoch": 3.57, + "learning_rate": 1.2189539092203849e-05, + "loss": 0.1988, + "step": 8537 + }, + { + "epoch": 3.57, + "learning_rate": 1.218788479061709e-05, + "loss": 0.1842, + "step": 8538 + }, + { + "epoch": 3.58, + "learning_rate": 1.2186230426141528e-05, + "loss": 0.1653, + "step": 8539 + }, + { + "epoch": 3.58, + "learning_rate": 1.2184575998824713e-05, + "loss": 0.1869, + "step": 8540 + }, + { + "epoch": 3.58, + "learning_rate": 1.2182921508714196e-05, + "loss": 0.1924, + "step": 8541 + }, + { + "epoch": 3.58, + "learning_rate": 1.218126695585754e-05, + "loss": 0.1826, + "step": 8542 + }, + { + "epoch": 3.58, + "learning_rate": 1.2179612340302302e-05, + "loss": 0.1999, + "step": 8543 + }, + { + "epoch": 3.58, + "learning_rate": 1.2177957662096038e-05, + "loss": 0.217, + "step": 8544 + }, + { + "epoch": 3.58, + "learning_rate": 1.2176302921286316e-05, + "loss": 0.2095, + "step": 8545 + }, + { + "epoch": 3.58, + "learning_rate": 1.2174648117920698e-05, + "loss": 0.202, + "step": 8546 + }, + { + "epoch": 3.58, + "learning_rate": 1.2172993252046747e-05, + "loss": 0.1597, + "step": 8547 + }, + { + "epoch": 3.58, + "learning_rate": 1.2171338323712034e-05, + "loss": 0.1915, + "step": 8548 + }, + { + "epoch": 3.58, + "learning_rate": 1.216968333296413e-05, + "loss": 0.1843, + "step": 8549 + }, + { + "epoch": 3.58, + "learning_rate": 1.2168028279850603e-05, + "loss": 0.2172, + "step": 8550 + }, + { + "epoch": 3.58, + "learning_rate": 1.2166373164419026e-05, + "loss": 0.1981, + "step": 8551 + }, + { + "epoch": 3.58, + "learning_rate": 1.2164717986716977e-05, + "loss": 0.1738, + "step": 8552 + }, + { + "epoch": 3.58, + "learning_rate": 1.2163062746792027e-05, + "loss": 0.1915, + "step": 8553 + }, + { + "epoch": 3.58, + "learning_rate": 1.2161407444691761e-05, + "loss": 0.1887, + "step": 8554 + }, + { + "epoch": 3.58, + "learning_rate": 1.2159752080463753e-05, + "loss": 0.1902, + "step": 8555 + }, + { + "epoch": 3.58, + "learning_rate": 1.2158096654155592e-05, + "loss": 0.1652, + "step": 8556 + }, + { + "epoch": 3.58, + "learning_rate": 1.2156441165814855e-05, + "loss": 0.187, + "step": 8557 + }, + { + "epoch": 3.58, + "learning_rate": 1.2154785615489129e-05, + "loss": 0.1937, + "step": 8558 + }, + { + "epoch": 3.58, + "learning_rate": 1.2153130003226006e-05, + "loss": 0.1814, + "step": 8559 + }, + { + "epoch": 3.58, + "learning_rate": 1.215147432907307e-05, + "loss": 0.1693, + "step": 8560 + }, + { + "epoch": 3.58, + "learning_rate": 1.2149818593077912e-05, + "loss": 0.1944, + "step": 8561 + }, + { + "epoch": 3.58, + "learning_rate": 1.2148162795288127e-05, + "loss": 0.1962, + "step": 8562 + }, + { + "epoch": 3.59, + "learning_rate": 1.214650693575131e-05, + "loss": 0.1754, + "step": 8563 + }, + { + "epoch": 3.59, + "learning_rate": 1.2144851014515055e-05, + "loss": 0.1715, + "step": 8564 + }, + { + "epoch": 3.59, + "learning_rate": 1.2143195031626962e-05, + "loss": 0.1796, + "step": 8565 + }, + { + "epoch": 3.59, + "learning_rate": 1.214153898713463e-05, + "loss": 0.1753, + "step": 8566 + }, + { + "epoch": 3.59, + "learning_rate": 1.2139882881085658e-05, + "loss": 0.1809, + "step": 8567 + }, + { + "epoch": 3.59, + "learning_rate": 1.2138226713527652e-05, + "loss": 0.1781, + "step": 8568 + }, + { + "epoch": 3.59, + "learning_rate": 1.2136570484508217e-05, + "loss": 0.2059, + "step": 8569 + }, + { + "epoch": 3.59, + "learning_rate": 1.213491419407496e-05, + "loss": 0.1986, + "step": 8570 + }, + { + "epoch": 3.59, + "learning_rate": 1.2133257842275487e-05, + "loss": 0.193, + "step": 8571 + }, + { + "epoch": 3.59, + "learning_rate": 1.2131601429157413e-05, + "loss": 0.2063, + "step": 8572 + }, + { + "epoch": 3.59, + "learning_rate": 1.2129944954768346e-05, + "loss": 0.1816, + "step": 8573 + }, + { + "epoch": 3.59, + "learning_rate": 1.2128288419155903e-05, + "loss": 0.2111, + "step": 8574 + }, + { + "epoch": 3.59, + "learning_rate": 1.2126631822367693e-05, + "loss": 0.1456, + "step": 8575 + }, + { + "epoch": 3.59, + "learning_rate": 1.2124975164451345e-05, + "loss": 0.1482, + "step": 8576 + }, + { + "epoch": 3.59, + "learning_rate": 1.212331844545447e-05, + "loss": 0.2008, + "step": 8577 + }, + { + "epoch": 3.59, + "learning_rate": 1.2121661665424688e-05, + "loss": 0.176, + "step": 8578 + }, + { + "epoch": 3.59, + "learning_rate": 1.2120004824409626e-05, + "loss": 0.1779, + "step": 8579 + }, + { + "epoch": 3.59, + "learning_rate": 1.2118347922456905e-05, + "loss": 0.162, + "step": 8580 + }, + { + "epoch": 3.59, + "learning_rate": 1.2116690959614154e-05, + "loss": 0.1954, + "step": 8581 + }, + { + "epoch": 3.59, + "learning_rate": 1.2115033935929e-05, + "loss": 0.183, + "step": 8582 + }, + { + "epoch": 3.59, + "learning_rate": 1.211337685144907e-05, + "loss": 0.1416, + "step": 8583 + }, + { + "epoch": 3.59, + "learning_rate": 1.2111719706222e-05, + "loss": 0.2038, + "step": 8584 + }, + { + "epoch": 3.59, + "learning_rate": 1.2110062500295418e-05, + "loss": 0.1634, + "step": 8585 + }, + { + "epoch": 3.59, + "learning_rate": 1.2108405233716967e-05, + "loss": 0.1703, + "step": 8586 + }, + { + "epoch": 3.6, + "learning_rate": 1.2106747906534275e-05, + "loss": 0.193, + "step": 8587 + }, + { + "epoch": 3.6, + "learning_rate": 1.2105090518794986e-05, + "loss": 0.1664, + "step": 8588 + }, + { + "epoch": 3.6, + "learning_rate": 1.2103433070546737e-05, + "loss": 0.1917, + "step": 8589 + }, + { + "epoch": 3.6, + "learning_rate": 1.2101775561837172e-05, + "loss": 0.1935, + "step": 8590 + }, + { + "epoch": 3.6, + "learning_rate": 1.2100117992713931e-05, + "loss": 0.1685, + "step": 8591 + }, + { + "epoch": 3.6, + "learning_rate": 1.2098460363224662e-05, + "loss": 0.1723, + "step": 8592 + }, + { + "epoch": 3.6, + "learning_rate": 1.2096802673417014e-05, + "loss": 0.1534, + "step": 8593 + }, + { + "epoch": 3.6, + "learning_rate": 1.2095144923338633e-05, + "loss": 0.1846, + "step": 8594 + }, + { + "epoch": 3.6, + "learning_rate": 1.209348711303717e-05, + "loss": 0.217, + "step": 8595 + }, + { + "epoch": 3.6, + "learning_rate": 1.2091829242560278e-05, + "loss": 0.2127, + "step": 8596 + }, + { + "epoch": 3.6, + "learning_rate": 1.209017131195561e-05, + "loss": 0.182, + "step": 8597 + }, + { + "epoch": 3.6, + "learning_rate": 1.2088513321270823e-05, + "loss": 0.2203, + "step": 8598 + }, + { + "epoch": 3.6, + "learning_rate": 1.2086855270553575e-05, + "loss": 0.2172, + "step": 8599 + }, + { + "epoch": 3.6, + "learning_rate": 1.2085197159851525e-05, + "loss": 0.2137, + "step": 8600 + }, + { + "epoch": 3.6, + "learning_rate": 1.208353898921233e-05, + "loss": 0.1924, + "step": 8601 + }, + { + "epoch": 3.6, + "learning_rate": 1.2081880758683656e-05, + "loss": 0.22, + "step": 8602 + }, + { + "epoch": 3.6, + "learning_rate": 1.208022246831317e-05, + "loss": 0.2001, + "step": 8603 + }, + { + "epoch": 3.6, + "learning_rate": 1.2078564118148531e-05, + "loss": 0.2161, + "step": 8604 + }, + { + "epoch": 3.6, + "learning_rate": 1.2076905708237412e-05, + "loss": 0.19, + "step": 8605 + }, + { + "epoch": 3.6, + "learning_rate": 1.2075247238627482e-05, + "loss": 0.1804, + "step": 8606 + }, + { + "epoch": 3.6, + "learning_rate": 1.2073588709366413e-05, + "loss": 0.2071, + "step": 8607 + }, + { + "epoch": 3.6, + "learning_rate": 1.2071930120501874e-05, + "loss": 0.2102, + "step": 8608 + }, + { + "epoch": 3.6, + "learning_rate": 1.2070271472081542e-05, + "loss": 0.1457, + "step": 8609 + }, + { + "epoch": 3.6, + "learning_rate": 1.2068612764153099e-05, + "loss": 0.2246, + "step": 8610 + }, + { + "epoch": 3.61, + "learning_rate": 1.2066953996764215e-05, + "loss": 0.1485, + "step": 8611 + }, + { + "epoch": 3.61, + "learning_rate": 1.2065295169962573e-05, + "loss": 0.1568, + "step": 8612 + }, + { + "epoch": 3.61, + "learning_rate": 1.206363628379585e-05, + "loss": 0.1886, + "step": 8613 + }, + { + "epoch": 3.61, + "learning_rate": 1.2061977338311737e-05, + "loss": 0.1763, + "step": 8614 + }, + { + "epoch": 3.61, + "learning_rate": 1.2060318333557915e-05, + "loss": 0.166, + "step": 8615 + }, + { + "epoch": 3.61, + "learning_rate": 1.2058659269582067e-05, + "loss": 0.1715, + "step": 8616 + }, + { + "epoch": 3.61, + "learning_rate": 1.205700014643189e-05, + "loss": 0.2064, + "step": 8617 + }, + { + "epoch": 3.61, + "learning_rate": 1.2055340964155067e-05, + "loss": 0.2055, + "step": 8618 + }, + { + "epoch": 3.61, + "learning_rate": 1.2053681722799291e-05, + "loss": 0.2265, + "step": 8619 + }, + { + "epoch": 3.61, + "learning_rate": 1.2052022422412259e-05, + "loss": 0.2, + "step": 8620 + }, + { + "epoch": 3.61, + "learning_rate": 1.2050363063041662e-05, + "loss": 0.1541, + "step": 8621 + }, + { + "epoch": 3.61, + "learning_rate": 1.2048703644735198e-05, + "loss": 0.1835, + "step": 8622 + }, + { + "epoch": 3.61, + "learning_rate": 1.2047044167540564e-05, + "loss": 0.1863, + "step": 8623 + }, + { + "epoch": 3.61, + "learning_rate": 1.2045384631505466e-05, + "loss": 0.1417, + "step": 8624 + }, + { + "epoch": 3.61, + "learning_rate": 1.2043725036677599e-05, + "loss": 0.1641, + "step": 8625 + }, + { + "epoch": 3.61, + "learning_rate": 1.204206538310467e-05, + "loss": 0.1823, + "step": 8626 + }, + { + "epoch": 3.61, + "learning_rate": 1.2040405670834385e-05, + "loss": 0.2046, + "step": 8627 + }, + { + "epoch": 3.61, + "learning_rate": 1.2038745899914446e-05, + "loss": 0.1768, + "step": 8628 + }, + { + "epoch": 3.61, + "learning_rate": 1.2037086070392567e-05, + "loss": 0.1889, + "step": 8629 + }, + { + "epoch": 3.61, + "learning_rate": 1.2035426182316459e-05, + "loss": 0.2021, + "step": 8630 + }, + { + "epoch": 3.61, + "learning_rate": 1.2033766235733828e-05, + "loss": 0.2051, + "step": 8631 + }, + { + "epoch": 3.61, + "learning_rate": 1.2032106230692394e-05, + "loss": 0.2008, + "step": 8632 + }, + { + "epoch": 3.61, + "learning_rate": 1.2030446167239869e-05, + "loss": 0.171, + "step": 8633 + }, + { + "epoch": 3.61, + "learning_rate": 1.202878604542397e-05, + "loss": 0.2133, + "step": 8634 + }, + { + "epoch": 3.62, + "learning_rate": 1.2027125865292417e-05, + "loss": 0.1647, + "step": 8635 + }, + { + "epoch": 3.62, + "learning_rate": 1.202546562689293e-05, + "loss": 0.1962, + "step": 8636 + }, + { + "epoch": 3.62, + "learning_rate": 1.202380533027323e-05, + "loss": 0.2258, + "step": 8637 + }, + { + "epoch": 3.62, + "learning_rate": 1.2022144975481043e-05, + "loss": 0.1678, + "step": 8638 + }, + { + "epoch": 3.62, + "learning_rate": 1.202048456256409e-05, + "loss": 0.1997, + "step": 8639 + }, + { + "epoch": 3.62, + "learning_rate": 1.2018824091570103e-05, + "loss": 0.1733, + "step": 8640 + }, + { + "epoch": 3.62, + "learning_rate": 1.201716356254681e-05, + "loss": 0.1818, + "step": 8641 + }, + { + "epoch": 3.62, + "learning_rate": 1.201550297554194e-05, + "loss": 0.1849, + "step": 8642 + }, + { + "epoch": 3.62, + "learning_rate": 1.2013842330603226e-05, + "loss": 0.1728, + "step": 8643 + }, + { + "epoch": 3.62, + "learning_rate": 1.20121816277784e-05, + "loss": 0.1764, + "step": 8644 + }, + { + "epoch": 3.62, + "learning_rate": 1.20105208671152e-05, + "loss": 0.2053, + "step": 8645 + }, + { + "epoch": 3.62, + "learning_rate": 1.2008860048661362e-05, + "loss": 0.1708, + "step": 8646 + }, + { + "epoch": 3.62, + "learning_rate": 1.200719917246462e-05, + "loss": 0.1796, + "step": 8647 + }, + { + "epoch": 3.62, + "learning_rate": 1.2005538238572724e-05, + "loss": 0.1694, + "step": 8648 + }, + { + "epoch": 3.62, + "learning_rate": 1.2003877247033411e-05, + "loss": 0.1919, + "step": 8649 + }, + { + "epoch": 3.62, + "learning_rate": 1.200221619789442e-05, + "loss": 0.242, + "step": 8650 + }, + { + "epoch": 3.62, + "learning_rate": 1.2000555091203507e-05, + "loss": 0.1495, + "step": 8651 + }, + { + "epoch": 3.62, + "learning_rate": 1.1998893927008411e-05, + "loss": 0.1781, + "step": 8652 + }, + { + "epoch": 3.62, + "learning_rate": 1.199723270535688e-05, + "loss": 0.1978, + "step": 8653 + }, + { + "epoch": 3.62, + "learning_rate": 1.1995571426296671e-05, + "loss": 0.2193, + "step": 8654 + }, + { + "epoch": 3.62, + "learning_rate": 1.1993910089875531e-05, + "loss": 0.1623, + "step": 8655 + }, + { + "epoch": 3.62, + "learning_rate": 1.1992248696141215e-05, + "loss": 0.2132, + "step": 8656 + }, + { + "epoch": 3.62, + "learning_rate": 1.1990587245141477e-05, + "loss": 0.1865, + "step": 8657 + }, + { + "epoch": 3.62, + "learning_rate": 1.1988925736924078e-05, + "loss": 0.1778, + "step": 8658 + }, + { + "epoch": 3.63, + "learning_rate": 1.198726417153677e-05, + "loss": 0.1642, + "step": 8659 + }, + { + "epoch": 3.63, + "learning_rate": 1.1985602549027318e-05, + "loss": 0.1997, + "step": 8660 + }, + { + "epoch": 3.63, + "learning_rate": 1.1983940869443484e-05, + "loss": 0.2021, + "step": 8661 + }, + { + "epoch": 3.63, + "learning_rate": 1.1982279132833028e-05, + "loss": 0.1802, + "step": 8662 + }, + { + "epoch": 3.63, + "learning_rate": 1.1980617339243719e-05, + "loss": 0.215, + "step": 8663 + }, + { + "epoch": 3.63, + "learning_rate": 1.1978955488723323e-05, + "loss": 0.1522, + "step": 8664 + }, + { + "epoch": 3.63, + "learning_rate": 1.1977293581319605e-05, + "loss": 0.1754, + "step": 8665 + }, + { + "epoch": 3.63, + "learning_rate": 1.1975631617080339e-05, + "loss": 0.1967, + "step": 8666 + }, + { + "epoch": 3.63, + "learning_rate": 1.1973969596053295e-05, + "loss": 0.1964, + "step": 8667 + }, + { + "epoch": 3.63, + "learning_rate": 1.1972307518286248e-05, + "loss": 0.163, + "step": 8668 + }, + { + "epoch": 3.63, + "learning_rate": 1.1970645383826971e-05, + "loss": 0.2226, + "step": 8669 + }, + { + "epoch": 3.63, + "learning_rate": 1.1968983192723238e-05, + "loss": 0.1526, + "step": 8670 + }, + { + "epoch": 3.63, + "learning_rate": 1.1967320945022836e-05, + "loss": 0.1703, + "step": 8671 + }, + { + "epoch": 3.63, + "learning_rate": 1.1965658640773535e-05, + "loss": 0.1807, + "step": 8672 + }, + { + "epoch": 3.63, + "learning_rate": 1.196399628002312e-05, + "loss": 0.1969, + "step": 8673 + }, + { + "epoch": 3.63, + "learning_rate": 1.1962333862819379e-05, + "loss": 0.1891, + "step": 8674 + }, + { + "epoch": 3.63, + "learning_rate": 1.1960671389210087e-05, + "loss": 0.1824, + "step": 8675 + }, + { + "epoch": 3.63, + "learning_rate": 1.195900885924304e-05, + "loss": 0.1831, + "step": 8676 + }, + { + "epoch": 3.63, + "learning_rate": 1.1957346272966019e-05, + "loss": 0.1629, + "step": 8677 + }, + { + "epoch": 3.63, + "learning_rate": 1.195568363042682e-05, + "loss": 0.187, + "step": 8678 + }, + { + "epoch": 3.63, + "learning_rate": 1.1954020931673229e-05, + "loss": 0.1698, + "step": 8679 + }, + { + "epoch": 3.63, + "learning_rate": 1.1952358176753038e-05, + "loss": 0.1691, + "step": 8680 + }, + { + "epoch": 3.63, + "learning_rate": 1.1950695365714046e-05, + "loss": 0.1688, + "step": 8681 + }, + { + "epoch": 3.63, + "learning_rate": 1.1949032498604047e-05, + "loss": 0.232, + "step": 8682 + }, + { + "epoch": 3.64, + "learning_rate": 1.1947369575470837e-05, + "loss": 0.1872, + "step": 8683 + }, + { + "epoch": 3.64, + "learning_rate": 1.1945706596362216e-05, + "loss": 0.1597, + "step": 8684 + }, + { + "epoch": 3.64, + "learning_rate": 1.1944043561325989e-05, + "loss": 0.2131, + "step": 8685 + }, + { + "epoch": 3.64, + "learning_rate": 1.194238047040995e-05, + "loss": 0.1646, + "step": 8686 + }, + { + "epoch": 3.64, + "learning_rate": 1.194071732366191e-05, + "loss": 0.1981, + "step": 8687 + }, + { + "epoch": 3.64, + "learning_rate": 1.1939054121129677e-05, + "loss": 0.1612, + "step": 8688 + }, + { + "epoch": 3.64, + "learning_rate": 1.1937390862861047e-05, + "loss": 0.1734, + "step": 8689 + }, + { + "epoch": 3.64, + "learning_rate": 1.1935727548903842e-05, + "loss": 0.1866, + "step": 8690 + }, + { + "epoch": 3.64, + "learning_rate": 1.1934064179305861e-05, + "loss": 0.1663, + "step": 8691 + }, + { + "epoch": 3.64, + "learning_rate": 1.1932400754114925e-05, + "loss": 0.1951, + "step": 8692 + }, + { + "epoch": 3.64, + "learning_rate": 1.1930737273378842e-05, + "loss": 0.1756, + "step": 8693 + }, + { + "epoch": 3.64, + "learning_rate": 1.1929073737145432e-05, + "loss": 0.1713, + "step": 8694 + }, + { + "epoch": 3.64, + "learning_rate": 1.1927410145462507e-05, + "loss": 0.2078, + "step": 8695 + }, + { + "epoch": 3.64, + "learning_rate": 1.1925746498377888e-05, + "loss": 0.1786, + "step": 8696 + }, + { + "epoch": 3.64, + "learning_rate": 1.1924082795939394e-05, + "loss": 0.1771, + "step": 8697 + }, + { + "epoch": 3.64, + "learning_rate": 1.192241903819485e-05, + "loss": 0.1918, + "step": 8698 + }, + { + "epoch": 3.64, + "learning_rate": 1.1920755225192074e-05, + "loss": 0.167, + "step": 8699 + }, + { + "epoch": 3.64, + "learning_rate": 1.1919091356978894e-05, + "loss": 0.1738, + "step": 8700 + }, + { + "epoch": 3.64, + "learning_rate": 1.1917427433603138e-05, + "loss": 0.203, + "step": 8701 + }, + { + "epoch": 3.64, + "learning_rate": 1.1915763455112633e-05, + "loss": 0.2271, + "step": 8702 + }, + { + "epoch": 3.64, + "learning_rate": 1.1914099421555203e-05, + "loss": 0.1783, + "step": 8703 + }, + { + "epoch": 3.64, + "learning_rate": 1.1912435332978684e-05, + "loss": 0.1941, + "step": 8704 + }, + { + "epoch": 3.64, + "learning_rate": 1.1910771189430913e-05, + "loss": 0.2174, + "step": 8705 + }, + { + "epoch": 3.64, + "learning_rate": 1.1909106990959716e-05, + "loss": 0.1742, + "step": 8706 + }, + { + "epoch": 3.65, + "learning_rate": 1.1907442737612932e-05, + "loss": 0.1669, + "step": 8707 + }, + { + "epoch": 3.65, + "learning_rate": 1.1905778429438401e-05, + "loss": 0.2056, + "step": 8708 + }, + { + "epoch": 3.65, + "learning_rate": 1.190411406648396e-05, + "loss": 0.2232, + "step": 8709 + }, + { + "epoch": 3.65, + "learning_rate": 1.190244964879745e-05, + "loss": 0.1932, + "step": 8710 + }, + { + "epoch": 3.65, + "learning_rate": 1.190078517642671e-05, + "loss": 0.1979, + "step": 8711 + }, + { + "epoch": 3.65, + "learning_rate": 1.189912064941959e-05, + "loss": 0.1841, + "step": 8712 + }, + { + "epoch": 3.65, + "learning_rate": 1.1897456067823934e-05, + "loss": 0.1931, + "step": 8713 + }, + { + "epoch": 3.65, + "learning_rate": 1.1895791431687585e-05, + "loss": 0.1672, + "step": 8714 + }, + { + "epoch": 3.65, + "learning_rate": 1.1894126741058393e-05, + "loss": 0.1963, + "step": 8715 + }, + { + "epoch": 3.65, + "learning_rate": 1.189246199598421e-05, + "loss": 0.2083, + "step": 8716 + }, + { + "epoch": 3.65, + "learning_rate": 1.1890797196512884e-05, + "loss": 0.218, + "step": 8717 + }, + { + "epoch": 3.65, + "learning_rate": 1.1889132342692272e-05, + "loss": 0.1758, + "step": 8718 + }, + { + "epoch": 3.65, + "learning_rate": 1.1887467434570227e-05, + "loss": 0.1998, + "step": 8719 + }, + { + "epoch": 3.65, + "learning_rate": 1.1885802472194606e-05, + "loss": 0.1669, + "step": 8720 + }, + { + "epoch": 3.65, + "learning_rate": 1.1884137455613266e-05, + "loss": 0.1861, + "step": 8721 + }, + { + "epoch": 3.65, + "learning_rate": 1.1882472384874067e-05, + "loss": 0.2202, + "step": 8722 + }, + { + "epoch": 3.65, + "learning_rate": 1.1880807260024867e-05, + "loss": 0.1837, + "step": 8723 + }, + { + "epoch": 3.65, + "learning_rate": 1.1879142081113535e-05, + "loss": 0.1654, + "step": 8724 + }, + { + "epoch": 3.65, + "learning_rate": 1.187747684818793e-05, + "loss": 0.1711, + "step": 8725 + }, + { + "epoch": 3.65, + "learning_rate": 1.1875811561295921e-05, + "loss": 0.1968, + "step": 8726 + }, + { + "epoch": 3.65, + "learning_rate": 1.187414622048537e-05, + "loss": 0.1816, + "step": 8727 + }, + { + "epoch": 3.65, + "learning_rate": 1.1872480825804149e-05, + "loss": 0.1843, + "step": 8728 + }, + { + "epoch": 3.65, + "learning_rate": 1.1870815377300131e-05, + "loss": 0.171, + "step": 8729 + }, + { + "epoch": 3.66, + "learning_rate": 1.1869149875021184e-05, + "loss": 0.1602, + "step": 8730 + }, + { + "epoch": 3.66, + "learning_rate": 1.186748431901518e-05, + "loss": 0.1338, + "step": 8731 + }, + { + "epoch": 3.66, + "learning_rate": 1.1865818709330001e-05, + "loss": 0.211, + "step": 8732 + }, + { + "epoch": 3.66, + "learning_rate": 1.1864153046013515e-05, + "loss": 0.2019, + "step": 8733 + }, + { + "epoch": 3.66, + "learning_rate": 1.1862487329113606e-05, + "loss": 0.2191, + "step": 8734 + }, + { + "epoch": 3.66, + "learning_rate": 1.1860821558678152e-05, + "loss": 0.1587, + "step": 8735 + }, + { + "epoch": 3.66, + "learning_rate": 1.1859155734755034e-05, + "loss": 0.22, + "step": 8736 + }, + { + "epoch": 3.66, + "learning_rate": 1.1857489857392133e-05, + "loss": 0.1619, + "step": 8737 + }, + { + "epoch": 3.66, + "learning_rate": 1.1855823926637333e-05, + "loss": 0.1612, + "step": 8738 + }, + { + "epoch": 3.66, + "learning_rate": 1.1854157942538522e-05, + "loss": 0.1793, + "step": 8739 + }, + { + "epoch": 3.66, + "learning_rate": 1.1852491905143586e-05, + "loss": 0.1781, + "step": 8740 + }, + { + "epoch": 3.66, + "learning_rate": 1.1850825814500413e-05, + "loss": 0.1956, + "step": 8741 + }, + { + "epoch": 3.66, + "learning_rate": 1.1849159670656897e-05, + "loss": 0.1916, + "step": 8742 + }, + { + "epoch": 3.66, + "learning_rate": 1.1847493473660925e-05, + "loss": 0.1949, + "step": 8743 + }, + { + "epoch": 3.66, + "learning_rate": 1.1845827223560394e-05, + "loss": 0.1847, + "step": 8744 + }, + { + "epoch": 3.66, + "learning_rate": 1.1844160920403197e-05, + "loss": 0.1929, + "step": 8745 + }, + { + "epoch": 3.66, + "learning_rate": 1.1842494564237231e-05, + "loss": 0.131, + "step": 8746 + }, + { + "epoch": 3.66, + "learning_rate": 1.1840828155110395e-05, + "loss": 0.1856, + "step": 8747 + }, + { + "epoch": 3.66, + "learning_rate": 1.1839161693070584e-05, + "loss": 0.1538, + "step": 8748 + }, + { + "epoch": 3.66, + "learning_rate": 1.1837495178165706e-05, + "loss": 0.1963, + "step": 8749 + }, + { + "epoch": 3.66, + "learning_rate": 1.1835828610443656e-05, + "loss": 0.1969, + "step": 8750 + }, + { + "epoch": 3.66, + "learning_rate": 1.1834161989952345e-05, + "loss": 0.1978, + "step": 8751 + }, + { + "epoch": 3.66, + "learning_rate": 1.1832495316739672e-05, + "loss": 0.2095, + "step": 8752 + }, + { + "epoch": 3.66, + "learning_rate": 1.1830828590853548e-05, + "loss": 0.1731, + "step": 8753 + }, + { + "epoch": 3.67, + "learning_rate": 1.1829161812341883e-05, + "loss": 0.1978, + "step": 8754 + }, + { + "epoch": 3.67, + "learning_rate": 1.182749498125258e-05, + "loss": 0.1767, + "step": 8755 + }, + { + "epoch": 3.67, + "learning_rate": 1.1825828097633562e-05, + "loss": 0.1752, + "step": 8756 + }, + { + "epoch": 3.67, + "learning_rate": 1.1824161161532731e-05, + "loss": 0.1937, + "step": 8757 + }, + { + "epoch": 3.67, + "learning_rate": 1.1822494172998008e-05, + "loss": 0.1824, + "step": 8758 + }, + { + "epoch": 3.67, + "learning_rate": 1.1820827132077306e-05, + "loss": 0.1865, + "step": 8759 + }, + { + "epoch": 3.67, + "learning_rate": 1.1819160038818547e-05, + "loss": 0.1787, + "step": 8760 + }, + { + "epoch": 3.67, + "learning_rate": 1.1817492893269645e-05, + "loss": 0.1637, + "step": 8761 + }, + { + "epoch": 3.67, + "learning_rate": 1.181582569547852e-05, + "loss": 0.2163, + "step": 8762 + }, + { + "epoch": 3.67, + "learning_rate": 1.18141584454931e-05, + "loss": 0.1649, + "step": 8763 + }, + { + "epoch": 3.67, + "learning_rate": 1.1812491143361305e-05, + "loss": 0.2028, + "step": 8764 + }, + { + "epoch": 3.67, + "learning_rate": 1.1810823789131059e-05, + "loss": 0.1657, + "step": 8765 + }, + { + "epoch": 3.67, + "learning_rate": 1.1809156382850291e-05, + "loss": 0.1668, + "step": 8766 + }, + { + "epoch": 3.67, + "learning_rate": 1.1807488924566929e-05, + "loss": 0.1856, + "step": 8767 + }, + { + "epoch": 3.67, + "learning_rate": 1.1805821414328897e-05, + "loss": 0.1946, + "step": 8768 + }, + { + "epoch": 3.67, + "learning_rate": 1.1804153852184137e-05, + "loss": 0.2073, + "step": 8769 + }, + { + "epoch": 3.67, + "learning_rate": 1.1802486238180575e-05, + "loss": 0.1951, + "step": 8770 + }, + { + "epoch": 3.67, + "learning_rate": 1.1800818572366144e-05, + "loss": 0.1153, + "step": 8771 + }, + { + "epoch": 3.67, + "learning_rate": 1.179915085478878e-05, + "loss": 0.1781, + "step": 8772 + }, + { + "epoch": 3.67, + "learning_rate": 1.1797483085496424e-05, + "loss": 0.2328, + "step": 8773 + }, + { + "epoch": 3.67, + "learning_rate": 1.1795815264537008e-05, + "loss": 0.2347, + "step": 8774 + }, + { + "epoch": 3.67, + "learning_rate": 1.1794147391958477e-05, + "loss": 0.2036, + "step": 8775 + }, + { + "epoch": 3.67, + "learning_rate": 1.1792479467808774e-05, + "loss": 0.205, + "step": 8776 + }, + { + "epoch": 3.67, + "learning_rate": 1.1790811492135837e-05, + "loss": 0.1933, + "step": 8777 + }, + { + "epoch": 3.68, + "learning_rate": 1.1789143464987611e-05, + "loss": 0.1803, + "step": 8778 + }, + { + "epoch": 3.68, + "learning_rate": 1.1787475386412048e-05, + "loss": 0.2011, + "step": 8779 + }, + { + "epoch": 3.68, + "learning_rate": 1.178580725645709e-05, + "loss": 0.1484, + "step": 8780 + }, + { + "epoch": 3.68, + "learning_rate": 1.1784139075170686e-05, + "loss": 0.1619, + "step": 8781 + }, + { + "epoch": 3.68, + "learning_rate": 1.1782470842600785e-05, + "loss": 0.1636, + "step": 8782 + }, + { + "epoch": 3.68, + "learning_rate": 1.1780802558795346e-05, + "loss": 0.166, + "step": 8783 + }, + { + "epoch": 3.68, + "learning_rate": 1.1779134223802315e-05, + "loss": 0.1686, + "step": 8784 + }, + { + "epoch": 3.68, + "learning_rate": 1.177746583766965e-05, + "loss": 0.224, + "step": 8785 + }, + { + "epoch": 3.68, + "learning_rate": 1.1775797400445307e-05, + "loss": 0.1735, + "step": 8786 + }, + { + "epoch": 3.68, + "learning_rate": 1.1774128912177245e-05, + "loss": 0.1761, + "step": 8787 + }, + { + "epoch": 3.68, + "learning_rate": 1.177246037291342e-05, + "loss": 0.2169, + "step": 8788 + }, + { + "epoch": 3.68, + "learning_rate": 1.1770791782701794e-05, + "loss": 0.1702, + "step": 8789 + }, + { + "epoch": 3.68, + "learning_rate": 1.176912314159033e-05, + "loss": 0.1961, + "step": 8790 + }, + { + "epoch": 3.68, + "learning_rate": 1.1767454449626992e-05, + "loss": 0.2112, + "step": 8791 + }, + { + "epoch": 3.68, + "learning_rate": 1.1765785706859744e-05, + "loss": 0.2216, + "step": 8792 + }, + { + "epoch": 3.68, + "learning_rate": 1.1764116913336552e-05, + "loss": 0.2142, + "step": 8793 + }, + { + "epoch": 3.68, + "learning_rate": 1.1762448069105387e-05, + "loss": 0.2356, + "step": 8794 + }, + { + "epoch": 3.68, + "learning_rate": 1.1760779174214215e-05, + "loss": 0.1855, + "step": 8795 + }, + { + "epoch": 3.68, + "learning_rate": 1.1759110228711006e-05, + "loss": 0.2006, + "step": 8796 + }, + { + "epoch": 3.68, + "learning_rate": 1.1757441232643737e-05, + "loss": 0.2181, + "step": 8797 + }, + { + "epoch": 3.68, + "learning_rate": 1.1755772186060379e-05, + "loss": 0.1845, + "step": 8798 + }, + { + "epoch": 3.68, + "learning_rate": 1.1754103089008905e-05, + "loss": 0.1632, + "step": 8799 + }, + { + "epoch": 3.68, + "learning_rate": 1.1752433941537297e-05, + "loss": 0.1701, + "step": 8800 + }, + { + "epoch": 3.68, + "learning_rate": 1.175076474369353e-05, + "loss": 0.1762, + "step": 8801 + }, + { + "epoch": 3.69, + "learning_rate": 1.1749095495525584e-05, + "loss": 0.2207, + "step": 8802 + }, + { + "epoch": 3.69, + "learning_rate": 1.1747426197081443e-05, + "loss": 0.2055, + "step": 8803 + }, + { + "epoch": 3.69, + "learning_rate": 1.1745756848409083e-05, + "loss": 0.1968, + "step": 8804 + }, + { + "epoch": 3.69, + "learning_rate": 1.1744087449556492e-05, + "loss": 0.164, + "step": 8805 + }, + { + "epoch": 3.69, + "learning_rate": 1.1742418000571653e-05, + "loss": 0.2076, + "step": 8806 + }, + { + "epoch": 3.69, + "learning_rate": 1.1740748501502559e-05, + "loss": 0.1933, + "step": 8807 + }, + { + "epoch": 3.69, + "learning_rate": 1.1739078952397189e-05, + "loss": 0.1662, + "step": 8808 + }, + { + "epoch": 3.69, + "learning_rate": 1.1737409353303541e-05, + "loss": 0.1819, + "step": 8809 + }, + { + "epoch": 3.69, + "learning_rate": 1.1735739704269605e-05, + "loss": 0.1808, + "step": 8810 + }, + { + "epoch": 3.69, + "learning_rate": 1.1734070005343368e-05, + "loss": 0.184, + "step": 8811 + }, + { + "epoch": 3.69, + "learning_rate": 1.1732400256572827e-05, + "loss": 0.1854, + "step": 8812 + }, + { + "epoch": 3.69, + "learning_rate": 1.1730730458005981e-05, + "loss": 0.1776, + "step": 8813 + }, + { + "epoch": 3.69, + "learning_rate": 1.1729060609690824e-05, + "loss": 0.188, + "step": 8814 + }, + { + "epoch": 3.69, + "learning_rate": 1.172739071167535e-05, + "loss": 0.1515, + "step": 8815 + }, + { + "epoch": 3.69, + "learning_rate": 1.1725720764007564e-05, + "loss": 0.2115, + "step": 8816 + }, + { + "epoch": 3.69, + "learning_rate": 1.1724050766735469e-05, + "loss": 0.1856, + "step": 8817 + }, + { + "epoch": 3.69, + "learning_rate": 1.1722380719907063e-05, + "loss": 0.1764, + "step": 8818 + }, + { + "epoch": 3.69, + "learning_rate": 1.172071062357035e-05, + "loss": 0.1929, + "step": 8819 + }, + { + "epoch": 3.69, + "learning_rate": 1.171904047777334e-05, + "loss": 0.1946, + "step": 8820 + }, + { + "epoch": 3.69, + "learning_rate": 1.1717370282564034e-05, + "loss": 0.1833, + "step": 8821 + }, + { + "epoch": 3.69, + "learning_rate": 1.1715700037990443e-05, + "loss": 0.1939, + "step": 8822 + }, + { + "epoch": 3.69, + "learning_rate": 1.171402974410058e-05, + "loss": 0.167, + "step": 8823 + }, + { + "epoch": 3.69, + "learning_rate": 1.171235940094245e-05, + "loss": 0.223, + "step": 8824 + }, + { + "epoch": 3.69, + "learning_rate": 1.1710689008564068e-05, + "loss": 0.1941, + "step": 8825 + }, + { + "epoch": 3.7, + "learning_rate": 1.1709018567013455e-05, + "loss": 0.1697, + "step": 8826 + }, + { + "epoch": 3.7, + "learning_rate": 1.1707348076338616e-05, + "loss": 0.1743, + "step": 8827 + }, + { + "epoch": 3.7, + "learning_rate": 1.170567753658757e-05, + "loss": 0.2077, + "step": 8828 + }, + { + "epoch": 3.7, + "learning_rate": 1.1704006947808338e-05, + "loss": 0.1682, + "step": 8829 + }, + { + "epoch": 3.7, + "learning_rate": 1.1702336310048936e-05, + "loss": 0.1975, + "step": 8830 + }, + { + "epoch": 3.7, + "learning_rate": 1.170066562335739e-05, + "loss": 0.1564, + "step": 8831 + }, + { + "epoch": 3.7, + "learning_rate": 1.1698994887781718e-05, + "loss": 0.1795, + "step": 8832 + }, + { + "epoch": 3.7, + "learning_rate": 1.1697324103369947e-05, + "loss": 0.1962, + "step": 8833 + }, + { + "epoch": 3.7, + "learning_rate": 1.1695653270170102e-05, + "loss": 0.2126, + "step": 8834 + }, + { + "epoch": 3.7, + "learning_rate": 1.1693982388230207e-05, + "loss": 0.2115, + "step": 8835 + }, + { + "epoch": 3.7, + "learning_rate": 1.1692311457598291e-05, + "loss": 0.2035, + "step": 8836 + }, + { + "epoch": 3.7, + "learning_rate": 1.1690640478322386e-05, + "loss": 0.1825, + "step": 8837 + }, + { + "epoch": 3.7, + "learning_rate": 1.168896945045052e-05, + "loss": 0.1644, + "step": 8838 + }, + { + "epoch": 3.7, + "learning_rate": 1.1687298374030724e-05, + "loss": 0.1886, + "step": 8839 + }, + { + "epoch": 3.7, + "learning_rate": 1.1685627249111034e-05, + "loss": 0.1961, + "step": 8840 + }, + { + "epoch": 3.7, + "learning_rate": 1.1683956075739487e-05, + "loss": 0.2219, + "step": 8841 + }, + { + "epoch": 3.7, + "learning_rate": 1.1682284853964114e-05, + "loss": 0.1937, + "step": 8842 + }, + { + "epoch": 3.7, + "learning_rate": 1.1680613583832956e-05, + "loss": 0.1764, + "step": 8843 + }, + { + "epoch": 3.7, + "learning_rate": 1.1678942265394055e-05, + "loss": 0.1845, + "step": 8844 + }, + { + "epoch": 3.7, + "learning_rate": 1.1677270898695445e-05, + "loss": 0.1797, + "step": 8845 + }, + { + "epoch": 3.7, + "learning_rate": 1.1675599483785171e-05, + "loss": 0.1832, + "step": 8846 + }, + { + "epoch": 3.7, + "learning_rate": 1.167392802071128e-05, + "loss": 0.2014, + "step": 8847 + }, + { + "epoch": 3.7, + "learning_rate": 1.1672256509521814e-05, + "loss": 0.2086, + "step": 8848 + }, + { + "epoch": 3.7, + "learning_rate": 1.1670584950264816e-05, + "loss": 0.1592, + "step": 8849 + }, + { + "epoch": 3.71, + "learning_rate": 1.1668913342988335e-05, + "loss": 0.2036, + "step": 8850 + }, + { + "epoch": 3.71, + "learning_rate": 1.1667241687740422e-05, + "loss": 0.1597, + "step": 8851 + }, + { + "epoch": 3.71, + "learning_rate": 1.1665569984569127e-05, + "loss": 0.2258, + "step": 8852 + }, + { + "epoch": 3.71, + "learning_rate": 1.1663898233522499e-05, + "loss": 0.1592, + "step": 8853 + }, + { + "epoch": 3.71, + "learning_rate": 1.1662226434648595e-05, + "loss": 0.1749, + "step": 8854 + }, + { + "epoch": 3.71, + "learning_rate": 1.1660554587995465e-05, + "loss": 0.1716, + "step": 8855 + }, + { + "epoch": 3.71, + "learning_rate": 1.1658882693611169e-05, + "loss": 0.1531, + "step": 8856 + }, + { + "epoch": 3.71, + "learning_rate": 1.165721075154376e-05, + "loss": 0.1848, + "step": 8857 + }, + { + "epoch": 3.71, + "learning_rate": 1.16555387618413e-05, + "loss": 0.1777, + "step": 8858 + }, + { + "epoch": 3.71, + "learning_rate": 1.1653866724551848e-05, + "loss": 0.2096, + "step": 8859 + }, + { + "epoch": 3.71, + "learning_rate": 1.1652194639723465e-05, + "loss": 0.1777, + "step": 8860 + }, + { + "epoch": 3.71, + "learning_rate": 1.1650522507404211e-05, + "loss": 0.2106, + "step": 8861 + }, + { + "epoch": 3.71, + "learning_rate": 1.1648850327642154e-05, + "loss": 0.1917, + "step": 8862 + }, + { + "epoch": 3.71, + "learning_rate": 1.1647178100485354e-05, + "loss": 0.1864, + "step": 8863 + }, + { + "epoch": 3.71, + "learning_rate": 1.1645505825981884e-05, + "loss": 0.1869, + "step": 8864 + }, + { + "epoch": 3.71, + "learning_rate": 1.1643833504179811e-05, + "loss": 0.1971, + "step": 8865 + }, + { + "epoch": 3.71, + "learning_rate": 1.1642161135127198e-05, + "loss": 0.1678, + "step": 8866 + }, + { + "epoch": 3.71, + "learning_rate": 1.1640488718872124e-05, + "loss": 0.1875, + "step": 8867 + }, + { + "epoch": 3.71, + "learning_rate": 1.1638816255462657e-05, + "loss": 0.1905, + "step": 8868 + }, + { + "epoch": 3.71, + "learning_rate": 1.1637143744946871e-05, + "loss": 0.2219, + "step": 8869 + }, + { + "epoch": 3.71, + "learning_rate": 1.163547118737284e-05, + "loss": 0.1915, + "step": 8870 + }, + { + "epoch": 3.71, + "learning_rate": 1.1633798582788642e-05, + "loss": 0.1979, + "step": 8871 + }, + { + "epoch": 3.71, + "learning_rate": 1.1632125931242356e-05, + "loss": 0.2147, + "step": 8872 + }, + { + "epoch": 3.71, + "learning_rate": 1.1630453232782053e-05, + "loss": 0.2346, + "step": 8873 + }, + { + "epoch": 3.72, + "learning_rate": 1.162878048745582e-05, + "loss": 0.1713, + "step": 8874 + }, + { + "epoch": 3.72, + "learning_rate": 1.1627107695311738e-05, + "loss": 0.2076, + "step": 8875 + }, + { + "epoch": 3.72, + "learning_rate": 1.1625434856397891e-05, + "loss": 0.1693, + "step": 8876 + }, + { + "epoch": 3.72, + "learning_rate": 1.1623761970762359e-05, + "loss": 0.1629, + "step": 8877 + }, + { + "epoch": 3.72, + "learning_rate": 1.1622089038453232e-05, + "loss": 0.1809, + "step": 8878 + }, + { + "epoch": 3.72, + "learning_rate": 1.1620416059518593e-05, + "loss": 0.1831, + "step": 8879 + }, + { + "epoch": 3.72, + "learning_rate": 1.1618743034006532e-05, + "loss": 0.1921, + "step": 8880 + }, + { + "epoch": 3.72, + "learning_rate": 1.1617069961965141e-05, + "loss": 0.1877, + "step": 8881 + }, + { + "epoch": 3.72, + "learning_rate": 1.161539684344251e-05, + "loss": 0.2086, + "step": 8882 + }, + { + "epoch": 3.72, + "learning_rate": 1.1613723678486728e-05, + "loss": 0.1888, + "step": 8883 + }, + { + "epoch": 3.72, + "learning_rate": 1.161205046714589e-05, + "loss": 0.2187, + "step": 8884 + }, + { + "epoch": 3.72, + "learning_rate": 1.1610377209468095e-05, + "loss": 0.1716, + "step": 8885 + }, + { + "epoch": 3.72, + "learning_rate": 1.1608703905501432e-05, + "loss": 0.2116, + "step": 8886 + }, + { + "epoch": 3.72, + "learning_rate": 1.1607030555294005e-05, + "loss": 0.2013, + "step": 8887 + }, + { + "epoch": 3.72, + "learning_rate": 1.1605357158893913e-05, + "loss": 0.1809, + "step": 8888 + }, + { + "epoch": 3.72, + "learning_rate": 1.160368371634925e-05, + "loss": 0.196, + "step": 8889 + }, + { + "epoch": 3.72, + "learning_rate": 1.1602010227708122e-05, + "loss": 0.2002, + "step": 8890 + }, + { + "epoch": 3.72, + "learning_rate": 1.1600336693018635e-05, + "loss": 0.1671, + "step": 8891 + }, + { + "epoch": 3.72, + "learning_rate": 1.1598663112328888e-05, + "loss": 0.1837, + "step": 8892 + }, + { + "epoch": 3.72, + "learning_rate": 1.1596989485686987e-05, + "loss": 0.202, + "step": 8893 + }, + { + "epoch": 3.72, + "learning_rate": 1.1595315813141041e-05, + "loss": 0.1697, + "step": 8894 + }, + { + "epoch": 3.72, + "learning_rate": 1.1593642094739161e-05, + "loss": 0.1923, + "step": 8895 + }, + { + "epoch": 3.72, + "learning_rate": 1.159196833052945e-05, + "loss": 0.166, + "step": 8896 + }, + { + "epoch": 3.72, + "learning_rate": 1.1590294520560022e-05, + "loss": 0.2166, + "step": 8897 + }, + { + "epoch": 3.73, + "learning_rate": 1.1588620664878989e-05, + "loss": 0.1934, + "step": 8898 + }, + { + "epoch": 3.73, + "learning_rate": 1.1586946763534465e-05, + "loss": 0.1996, + "step": 8899 + }, + { + "epoch": 3.73, + "learning_rate": 1.1585272816574565e-05, + "loss": 0.1903, + "step": 8900 + }, + { + "epoch": 3.73, + "learning_rate": 1.1583598824047404e-05, + "loss": 0.1923, + "step": 8901 + }, + { + "epoch": 3.73, + "learning_rate": 1.1581924786001102e-05, + "loss": 0.1939, + "step": 8902 + }, + { + "epoch": 3.73, + "learning_rate": 1.1580250702483773e-05, + "loss": 0.1808, + "step": 8903 + }, + { + "epoch": 3.73, + "learning_rate": 1.1578576573543541e-05, + "loss": 0.1719, + "step": 8904 + }, + { + "epoch": 3.73, + "learning_rate": 1.1576902399228527e-05, + "loss": 0.1946, + "step": 8905 + }, + { + "epoch": 3.73, + "learning_rate": 1.1575228179586855e-05, + "loss": 0.1857, + "step": 8906 + }, + { + "epoch": 3.73, + "learning_rate": 1.1573553914666643e-05, + "loss": 0.1814, + "step": 8907 + }, + { + "epoch": 3.73, + "learning_rate": 1.157187960451602e-05, + "loss": 0.1515, + "step": 8908 + }, + { + "epoch": 3.73, + "learning_rate": 1.1570205249183115e-05, + "loss": 0.1651, + "step": 8909 + }, + { + "epoch": 3.73, + "learning_rate": 1.156853084871605e-05, + "loss": 0.1432, + "step": 8910 + }, + { + "epoch": 3.73, + "learning_rate": 1.156685640316296e-05, + "loss": 0.1824, + "step": 8911 + }, + { + "epoch": 3.73, + "learning_rate": 1.1565181912571975e-05, + "loss": 0.1531, + "step": 8912 + }, + { + "epoch": 3.73, + "learning_rate": 1.1563507376991222e-05, + "loss": 0.2062, + "step": 8913 + }, + { + "epoch": 3.73, + "learning_rate": 1.1561832796468837e-05, + "loss": 0.209, + "step": 8914 + }, + { + "epoch": 3.73, + "learning_rate": 1.1560158171052956e-05, + "loss": 0.1206, + "step": 8915 + }, + { + "epoch": 3.73, + "learning_rate": 1.1558483500791713e-05, + "loss": 0.1754, + "step": 8916 + }, + { + "epoch": 3.73, + "learning_rate": 1.1556808785733242e-05, + "loss": 0.2223, + "step": 8917 + }, + { + "epoch": 3.73, + "learning_rate": 1.1555134025925685e-05, + "loss": 0.1631, + "step": 8918 + }, + { + "epoch": 3.73, + "learning_rate": 1.1553459221417182e-05, + "loss": 0.1875, + "step": 8919 + }, + { + "epoch": 3.73, + "learning_rate": 1.155178437225587e-05, + "loss": 0.2032, + "step": 8920 + }, + { + "epoch": 3.73, + "learning_rate": 1.1550109478489896e-05, + "loss": 0.1785, + "step": 8921 + }, + { + "epoch": 3.74, + "learning_rate": 1.1548434540167399e-05, + "loss": 0.1759, + "step": 8922 + }, + { + "epoch": 3.74, + "learning_rate": 1.1546759557336524e-05, + "loss": 0.2091, + "step": 8923 + }, + { + "epoch": 3.74, + "learning_rate": 1.1545084530045419e-05, + "loss": 0.211, + "step": 8924 + }, + { + "epoch": 3.74, + "learning_rate": 1.154340945834223e-05, + "loss": 0.2067, + "step": 8925 + }, + { + "epoch": 3.74, + "learning_rate": 1.1541734342275106e-05, + "loss": 0.2118, + "step": 8926 + }, + { + "epoch": 3.74, + "learning_rate": 1.1540059181892197e-05, + "loss": 0.1759, + "step": 8927 + }, + { + "epoch": 3.74, + "learning_rate": 1.1538383977241654e-05, + "loss": 0.179, + "step": 8928 + }, + { + "epoch": 3.74, + "learning_rate": 1.1536708728371628e-05, + "loss": 0.1841, + "step": 8929 + }, + { + "epoch": 3.74, + "learning_rate": 1.1535033435330272e-05, + "loss": 0.1893, + "step": 8930 + }, + { + "epoch": 3.74, + "learning_rate": 1.1533358098165744e-05, + "loss": 0.2098, + "step": 8931 + }, + { + "epoch": 3.74, + "learning_rate": 1.1531682716926199e-05, + "loss": 0.1792, + "step": 8932 + }, + { + "epoch": 3.74, + "learning_rate": 1.153000729165979e-05, + "loss": 0.1708, + "step": 8933 + }, + { + "epoch": 3.74, + "learning_rate": 1.1528331822414681e-05, + "loss": 0.1958, + "step": 8934 + }, + { + "epoch": 3.74, + "learning_rate": 1.1526656309239028e-05, + "loss": 0.1662, + "step": 8935 + }, + { + "epoch": 3.74, + "learning_rate": 1.1524980752180996e-05, + "loss": 0.208, + "step": 8936 + }, + { + "epoch": 3.74, + "learning_rate": 1.1523305151288747e-05, + "loss": 0.173, + "step": 8937 + }, + { + "epoch": 3.74, + "learning_rate": 1.1521629506610439e-05, + "loss": 0.216, + "step": 8938 + }, + { + "epoch": 3.74, + "learning_rate": 1.1519953818194246e-05, + "loss": 0.1837, + "step": 8939 + }, + { + "epoch": 3.74, + "learning_rate": 1.1518278086088328e-05, + "loss": 0.1763, + "step": 8940 + }, + { + "epoch": 3.74, + "learning_rate": 1.1516602310340852e-05, + "loss": 0.1977, + "step": 8941 + }, + { + "epoch": 3.74, + "learning_rate": 1.151492649099999e-05, + "loss": 0.2517, + "step": 8942 + }, + { + "epoch": 3.74, + "learning_rate": 1.151325062811391e-05, + "loss": 0.234, + "step": 8943 + }, + { + "epoch": 3.74, + "learning_rate": 1.1511574721730781e-05, + "loss": 0.1756, + "step": 8944 + }, + { + "epoch": 3.75, + "learning_rate": 1.150989877189878e-05, + "loss": 0.1934, + "step": 8945 + }, + { + "epoch": 3.75, + "learning_rate": 1.1508222778666078e-05, + "loss": 0.1277, + "step": 8946 + }, + { + "epoch": 3.75, + "learning_rate": 1.1506546742080854e-05, + "loss": 0.1748, + "step": 8947 + }, + { + "epoch": 3.75, + "learning_rate": 1.1504870662191275e-05, + "loss": 0.169, + "step": 8948 + }, + { + "epoch": 3.75, + "learning_rate": 1.1503194539045527e-05, + "loss": 0.1794, + "step": 8949 + }, + { + "epoch": 3.75, + "learning_rate": 1.1501518372691789e-05, + "loss": 0.134, + "step": 8950 + }, + { + "epoch": 3.75, + "learning_rate": 1.1499842163178234e-05, + "loss": 0.1981, + "step": 8951 + }, + { + "epoch": 3.75, + "learning_rate": 1.1498165910553048e-05, + "loss": 0.1996, + "step": 8952 + }, + { + "epoch": 3.75, + "learning_rate": 1.1496489614864412e-05, + "loss": 0.1395, + "step": 8953 + }, + { + "epoch": 3.75, + "learning_rate": 1.1494813276160511e-05, + "loss": 0.1912, + "step": 8954 + }, + { + "epoch": 3.75, + "learning_rate": 1.1493136894489527e-05, + "loss": 0.1953, + "step": 8955 + }, + { + "epoch": 3.75, + "learning_rate": 1.149146046989965e-05, + "loss": 0.1645, + "step": 8956 + }, + { + "epoch": 3.75, + "learning_rate": 1.1489784002439064e-05, + "loss": 0.2151, + "step": 8957 + }, + { + "epoch": 3.75, + "learning_rate": 1.1488107492155958e-05, + "loss": 0.171, + "step": 8958 + }, + { + "epoch": 3.75, + "learning_rate": 1.1486430939098525e-05, + "loss": 0.1875, + "step": 8959 + }, + { + "epoch": 3.75, + "learning_rate": 1.1484754343314952e-05, + "loss": 0.2094, + "step": 8960 + }, + { + "epoch": 3.75, + "learning_rate": 1.1483077704853431e-05, + "loss": 0.1987, + "step": 8961 + }, + { + "epoch": 3.75, + "learning_rate": 1.1481401023762162e-05, + "loss": 0.2137, + "step": 8962 + }, + { + "epoch": 3.75, + "learning_rate": 1.1479724300089335e-05, + "loss": 0.1704, + "step": 8963 + }, + { + "epoch": 3.75, + "learning_rate": 1.1478047533883143e-05, + "loss": 0.1562, + "step": 8964 + }, + { + "epoch": 3.75, + "learning_rate": 1.1476370725191788e-05, + "loss": 0.1986, + "step": 8965 + }, + { + "epoch": 3.75, + "learning_rate": 1.1474693874063467e-05, + "loss": 0.1907, + "step": 8966 + }, + { + "epoch": 3.75, + "learning_rate": 1.1473016980546377e-05, + "loss": 0.1945, + "step": 8967 + }, + { + "epoch": 3.75, + "learning_rate": 1.147134004468872e-05, + "loss": 0.2249, + "step": 8968 + }, + { + "epoch": 3.76, + "learning_rate": 1.1469663066538704e-05, + "loss": 0.1739, + "step": 8969 + }, + { + "epoch": 3.76, + "learning_rate": 1.1467986046144523e-05, + "loss": 0.1978, + "step": 8970 + }, + { + "epoch": 3.76, + "learning_rate": 1.1466308983554386e-05, + "loss": 0.1379, + "step": 8971 + }, + { + "epoch": 3.76, + "learning_rate": 1.1464631878816502e-05, + "loss": 0.1583, + "step": 8972 + }, + { + "epoch": 3.76, + "learning_rate": 1.1462954731979074e-05, + "loss": 0.1796, + "step": 8973 + }, + { + "epoch": 3.76, + "learning_rate": 1.1461277543090308e-05, + "loss": 0.1897, + "step": 8974 + }, + { + "epoch": 3.76, + "learning_rate": 1.1459600312198416e-05, + "loss": 0.2038, + "step": 8975 + }, + { + "epoch": 3.76, + "learning_rate": 1.1457923039351606e-05, + "loss": 0.1697, + "step": 8976 + }, + { + "epoch": 3.76, + "learning_rate": 1.1456245724598096e-05, + "loss": 0.1876, + "step": 8977 + }, + { + "epoch": 3.76, + "learning_rate": 1.1454568367986094e-05, + "loss": 0.19, + "step": 8978 + }, + { + "epoch": 3.76, + "learning_rate": 1.1452890969563813e-05, + "loss": 0.2207, + "step": 8979 + }, + { + "epoch": 3.76, + "learning_rate": 1.1451213529379472e-05, + "loss": 0.1888, + "step": 8980 + }, + { + "epoch": 3.76, + "learning_rate": 1.1449536047481287e-05, + "loss": 0.1905, + "step": 8981 + }, + { + "epoch": 3.76, + "learning_rate": 1.1447858523917472e-05, + "loss": 0.2155, + "step": 8982 + }, + { + "epoch": 3.76, + "learning_rate": 1.144618095873625e-05, + "loss": 0.2026, + "step": 8983 + }, + { + "epoch": 3.76, + "learning_rate": 1.144450335198584e-05, + "loss": 0.1414, + "step": 8984 + }, + { + "epoch": 3.76, + "learning_rate": 1.1442825703714464e-05, + "loss": 0.1805, + "step": 8985 + }, + { + "epoch": 3.76, + "learning_rate": 1.144114801397034e-05, + "loss": 0.1571, + "step": 8986 + }, + { + "epoch": 3.76, + "learning_rate": 1.14394702828017e-05, + "loss": 0.2133, + "step": 8987 + }, + { + "epoch": 3.76, + "learning_rate": 1.143779251025676e-05, + "loss": 0.1764, + "step": 8988 + }, + { + "epoch": 3.76, + "learning_rate": 1.1436114696383749e-05, + "loss": 0.2045, + "step": 8989 + }, + { + "epoch": 3.76, + "learning_rate": 1.14344368412309e-05, + "loss": 0.1676, + "step": 8990 + }, + { + "epoch": 3.76, + "learning_rate": 1.1432758944846433e-05, + "loss": 0.1914, + "step": 8991 + }, + { + "epoch": 3.76, + "learning_rate": 1.1431081007278584e-05, + "loss": 0.1964, + "step": 8992 + }, + { + "epoch": 3.77, + "learning_rate": 1.142940302857558e-05, + "loss": 0.1722, + "step": 8993 + }, + { + "epoch": 3.77, + "learning_rate": 1.1427725008785655e-05, + "loss": 0.173, + "step": 8994 + }, + { + "epoch": 3.77, + "learning_rate": 1.1426046947957041e-05, + "loss": 0.2282, + "step": 8995 + }, + { + "epoch": 3.77, + "learning_rate": 1.1424368846137975e-05, + "loss": 0.1957, + "step": 8996 + }, + { + "epoch": 3.77, + "learning_rate": 1.142269070337669e-05, + "loss": 0.1834, + "step": 8997 + }, + { + "epoch": 3.77, + "learning_rate": 1.1421012519721421e-05, + "loss": 0.1803, + "step": 8998 + }, + { + "epoch": 3.77, + "learning_rate": 1.141933429522041e-05, + "loss": 0.2323, + "step": 8999 + }, + { + "epoch": 3.77, + "learning_rate": 1.1417656029921895e-05, + "loss": 0.2027, + "step": 9000 + }, + { + "epoch": 3.77, + "learning_rate": 1.1415977723874112e-05, + "loss": 0.1768, + "step": 9001 + }, + { + "epoch": 3.77, + "learning_rate": 1.1414299377125308e-05, + "loss": 0.1486, + "step": 9002 + }, + { + "epoch": 3.77, + "learning_rate": 1.1412620989723724e-05, + "loss": 0.1703, + "step": 9003 + }, + { + "epoch": 3.77, + "learning_rate": 1.1410942561717601e-05, + "loss": 0.1961, + "step": 9004 + }, + { + "epoch": 3.77, + "learning_rate": 1.1409264093155189e-05, + "loss": 0.2018, + "step": 9005 + }, + { + "epoch": 3.77, + "learning_rate": 1.1407585584084726e-05, + "loss": 0.1878, + "step": 9006 + }, + { + "epoch": 3.77, + "learning_rate": 1.1405907034554473e-05, + "loss": 0.2096, + "step": 9007 + }, + { + "epoch": 3.77, + "learning_rate": 1.1404228444612664e-05, + "loss": 0.1901, + "step": 9008 + }, + { + "epoch": 3.77, + "learning_rate": 1.1402549814307555e-05, + "loss": 0.1547, + "step": 9009 + }, + { + "epoch": 3.77, + "learning_rate": 1.1400871143687398e-05, + "loss": 0.193, + "step": 9010 + }, + { + "epoch": 3.77, + "learning_rate": 1.1399192432800442e-05, + "loss": 0.1671, + "step": 9011 + }, + { + "epoch": 3.77, + "learning_rate": 1.1397513681694938e-05, + "loss": 0.1926, + "step": 9012 + }, + { + "epoch": 3.77, + "learning_rate": 1.1395834890419147e-05, + "loss": 0.1855, + "step": 9013 + }, + { + "epoch": 3.77, + "learning_rate": 1.139415605902132e-05, + "loss": 0.1667, + "step": 9014 + }, + { + "epoch": 3.77, + "learning_rate": 1.1392477187549714e-05, + "loss": 0.2235, + "step": 9015 + }, + { + "epoch": 3.77, + "learning_rate": 1.1390798276052587e-05, + "loss": 0.206, + "step": 9016 + }, + { + "epoch": 3.78, + "learning_rate": 1.1389119324578197e-05, + "loss": 0.1925, + "step": 9017 + }, + { + "epoch": 3.78, + "learning_rate": 1.1387440333174809e-05, + "loss": 0.1357, + "step": 9018 + }, + { + "epoch": 3.78, + "learning_rate": 1.1385761301890674e-05, + "loss": 0.1902, + "step": 9019 + }, + { + "epoch": 3.78, + "learning_rate": 1.1384082230774063e-05, + "loss": 0.2036, + "step": 9020 + }, + { + "epoch": 3.78, + "learning_rate": 1.1382403119873236e-05, + "loss": 0.1801, + "step": 9021 + }, + { + "epoch": 3.78, + "learning_rate": 1.1380723969236458e-05, + "loss": 0.204, + "step": 9022 + }, + { + "epoch": 3.78, + "learning_rate": 1.1379044778911994e-05, + "loss": 0.2092, + "step": 9023 + }, + { + "epoch": 3.78, + "learning_rate": 1.1377365548948114e-05, + "loss": 0.164, + "step": 9024 + }, + { + "epoch": 3.78, + "learning_rate": 1.137568627939308e-05, + "loss": 0.1681, + "step": 9025 + }, + { + "epoch": 3.78, + "learning_rate": 1.1374006970295167e-05, + "loss": 0.1726, + "step": 9026 + }, + { + "epoch": 3.78, + "learning_rate": 1.137232762170264e-05, + "loss": 0.1698, + "step": 9027 + }, + { + "epoch": 3.78, + "learning_rate": 1.1370648233663776e-05, + "loss": 0.1895, + "step": 9028 + }, + { + "epoch": 3.78, + "learning_rate": 1.1368968806226844e-05, + "loss": 0.195, + "step": 9029 + }, + { + "epoch": 3.78, + "learning_rate": 1.1367289339440123e-05, + "loss": 0.2167, + "step": 9030 + }, + { + "epoch": 3.78, + "learning_rate": 1.1365609833351877e-05, + "loss": 0.232, + "step": 9031 + }, + { + "epoch": 3.78, + "learning_rate": 1.136393028801039e-05, + "loss": 0.1908, + "step": 9032 + }, + { + "epoch": 3.78, + "learning_rate": 1.1362250703463937e-05, + "loss": 0.1637, + "step": 9033 + }, + { + "epoch": 3.78, + "learning_rate": 1.1360571079760798e-05, + "loss": 0.1488, + "step": 9034 + }, + { + "epoch": 3.78, + "learning_rate": 1.1358891416949249e-05, + "loss": 0.2147, + "step": 9035 + }, + { + "epoch": 3.78, + "learning_rate": 1.1357211715077571e-05, + "loss": 0.2032, + "step": 9036 + }, + { + "epoch": 3.78, + "learning_rate": 1.1355531974194051e-05, + "loss": 0.1986, + "step": 9037 + }, + { + "epoch": 3.78, + "learning_rate": 1.1353852194346963e-05, + "loss": 0.1634, + "step": 9038 + }, + { + "epoch": 3.78, + "learning_rate": 1.1352172375584596e-05, + "loss": 0.2022, + "step": 9039 + }, + { + "epoch": 3.78, + "learning_rate": 1.1350492517955234e-05, + "loss": 0.1578, + "step": 9040 + }, + { + "epoch": 3.79, + "learning_rate": 1.1348812621507167e-05, + "loss": 0.1488, + "step": 9041 + }, + { + "epoch": 3.79, + "learning_rate": 1.1347132686288673e-05, + "loss": 0.1921, + "step": 9042 + }, + { + "epoch": 3.79, + "learning_rate": 1.1345452712348048e-05, + "loss": 0.1615, + "step": 9043 + }, + { + "epoch": 3.79, + "learning_rate": 1.1343772699733579e-05, + "loss": 0.2077, + "step": 9044 + }, + { + "epoch": 3.79, + "learning_rate": 1.1342092648493555e-05, + "loss": 0.1969, + "step": 9045 + }, + { + "epoch": 3.79, + "learning_rate": 1.1340412558676266e-05, + "loss": 0.2035, + "step": 9046 + }, + { + "epoch": 3.79, + "learning_rate": 1.1338732430330011e-05, + "loss": 0.1633, + "step": 9047 + }, + { + "epoch": 3.79, + "learning_rate": 1.1337052263503079e-05, + "loss": 0.1723, + "step": 9048 + }, + { + "epoch": 3.79, + "learning_rate": 1.1335372058243768e-05, + "loss": 0.2138, + "step": 9049 + }, + { + "epoch": 3.79, + "learning_rate": 1.1333691814600369e-05, + "loss": 0.1746, + "step": 9050 + }, + { + "epoch": 3.79, + "learning_rate": 1.1332011532621186e-05, + "loss": 0.2055, + "step": 9051 + }, + { + "epoch": 3.79, + "learning_rate": 1.1330331212354513e-05, + "loss": 0.1866, + "step": 9052 + }, + { + "epoch": 3.79, + "learning_rate": 1.1328650853848646e-05, + "loss": 0.1995, + "step": 9053 + }, + { + "epoch": 3.79, + "learning_rate": 1.132697045715189e-05, + "loss": 0.1582, + "step": 9054 + }, + { + "epoch": 3.79, + "learning_rate": 1.132529002231255e-05, + "loss": 0.2029, + "step": 9055 + }, + { + "epoch": 3.79, + "learning_rate": 1.132360954937892e-05, + "loss": 0.1756, + "step": 9056 + }, + { + "epoch": 3.79, + "learning_rate": 1.1321929038399309e-05, + "loss": 0.2151, + "step": 9057 + }, + { + "epoch": 3.79, + "learning_rate": 1.1320248489422024e-05, + "loss": 0.211, + "step": 9058 + }, + { + "epoch": 3.79, + "learning_rate": 1.1318567902495364e-05, + "loss": 0.2204, + "step": 9059 + }, + { + "epoch": 3.79, + "learning_rate": 1.131688727766764e-05, + "loss": 0.1669, + "step": 9060 + }, + { + "epoch": 3.79, + "learning_rate": 1.1315206614987161e-05, + "loss": 0.1861, + "step": 9061 + }, + { + "epoch": 3.79, + "learning_rate": 1.1313525914502235e-05, + "loss": 0.1728, + "step": 9062 + }, + { + "epoch": 3.79, + "learning_rate": 1.1311845176261172e-05, + "loss": 0.1886, + "step": 9063 + }, + { + "epoch": 3.79, + "learning_rate": 1.1310164400312284e-05, + "loss": 0.1917, + "step": 9064 + }, + { + "epoch": 3.8, + "learning_rate": 1.1308483586703885e-05, + "loss": 0.1719, + "step": 9065 + }, + { + "epoch": 3.8, + "learning_rate": 1.1306802735484282e-05, + "loss": 0.1885, + "step": 9066 + }, + { + "epoch": 3.8, + "learning_rate": 1.1305121846701797e-05, + "loss": 0.2081, + "step": 9067 + }, + { + "epoch": 3.8, + "learning_rate": 1.130344092040474e-05, + "loss": 0.2003, + "step": 9068 + }, + { + "epoch": 3.8, + "learning_rate": 1.1301759956641432e-05, + "loss": 0.199, + "step": 9069 + }, + { + "epoch": 3.8, + "learning_rate": 1.1300078955460187e-05, + "loss": 0.2198, + "step": 9070 + }, + { + "epoch": 3.8, + "learning_rate": 1.1298397916909332e-05, + "loss": 0.187, + "step": 9071 + }, + { + "epoch": 3.8, + "learning_rate": 1.1296716841037176e-05, + "loss": 0.2268, + "step": 9072 + }, + { + "epoch": 3.8, + "learning_rate": 1.1295035727892046e-05, + "loss": 0.2108, + "step": 9073 + }, + { + "epoch": 3.8, + "learning_rate": 1.1293354577522264e-05, + "loss": 0.2043, + "step": 9074 + }, + { + "epoch": 3.8, + "learning_rate": 1.1291673389976153e-05, + "loss": 0.1764, + "step": 9075 + }, + { + "epoch": 3.8, + "learning_rate": 1.1289992165302036e-05, + "loss": 0.2057, + "step": 9076 + }, + { + "epoch": 3.8, + "learning_rate": 1.1288310903548236e-05, + "loss": 0.1947, + "step": 9077 + }, + { + "epoch": 3.8, + "learning_rate": 1.1286629604763087e-05, + "loss": 0.2054, + "step": 9078 + }, + { + "epoch": 3.8, + "learning_rate": 1.1284948268994908e-05, + "loss": 0.2173, + "step": 9079 + }, + { + "epoch": 3.8, + "learning_rate": 1.1283266896292033e-05, + "loss": 0.1699, + "step": 9080 + }, + { + "epoch": 3.8, + "learning_rate": 1.1281585486702791e-05, + "loss": 0.1738, + "step": 9081 + }, + { + "epoch": 3.8, + "learning_rate": 1.1279904040275508e-05, + "loss": 0.1923, + "step": 9082 + }, + { + "epoch": 3.8, + "learning_rate": 1.1278222557058521e-05, + "loss": 0.191, + "step": 9083 + }, + { + "epoch": 3.8, + "learning_rate": 1.127654103710016e-05, + "loss": 0.2003, + "step": 9084 + }, + { + "epoch": 3.8, + "learning_rate": 1.127485948044876e-05, + "loss": 0.1608, + "step": 9085 + }, + { + "epoch": 3.8, + "learning_rate": 1.127317788715266e-05, + "loss": 0.1259, + "step": 9086 + }, + { + "epoch": 3.8, + "learning_rate": 1.1271496257260186e-05, + "loss": 0.1972, + "step": 9087 + }, + { + "epoch": 3.8, + "learning_rate": 1.126981459081968e-05, + "loss": 0.1557, + "step": 9088 + }, + { + "epoch": 3.81, + "learning_rate": 1.1268132887879485e-05, + "loss": 0.2127, + "step": 9089 + }, + { + "epoch": 3.81, + "learning_rate": 1.1266451148487934e-05, + "loss": 0.1941, + "step": 9090 + }, + { + "epoch": 3.81, + "learning_rate": 1.1264769372693365e-05, + "loss": 0.1913, + "step": 9091 + }, + { + "epoch": 3.81, + "learning_rate": 1.1263087560544128e-05, + "loss": 0.1689, + "step": 9092 + }, + { + "epoch": 3.81, + "learning_rate": 1.1261405712088556e-05, + "loss": 0.1506, + "step": 9093 + }, + { + "epoch": 3.81, + "learning_rate": 1.1259723827374996e-05, + "loss": 0.2145, + "step": 9094 + }, + { + "epoch": 3.81, + "learning_rate": 1.1258041906451795e-05, + "loss": 0.1616, + "step": 9095 + }, + { + "epoch": 3.81, + "learning_rate": 1.1256359949367293e-05, + "loss": 0.2278, + "step": 9096 + }, + { + "epoch": 3.81, + "learning_rate": 1.1254677956169839e-05, + "loss": 0.1865, + "step": 9097 + }, + { + "epoch": 3.81, + "learning_rate": 1.1252995926907783e-05, + "loss": 0.1616, + "step": 9098 + }, + { + "epoch": 3.81, + "learning_rate": 1.1251313861629474e-05, + "loss": 0.1893, + "step": 9099 + }, + { + "epoch": 3.81, + "learning_rate": 1.1249631760383253e-05, + "loss": 0.2181, + "step": 9100 + }, + { + "epoch": 3.81, + "learning_rate": 1.1247949623217476e-05, + "loss": 0.1697, + "step": 9101 + }, + { + "epoch": 3.81, + "learning_rate": 1.1246267450180497e-05, + "loss": 0.1728, + "step": 9102 + }, + { + "epoch": 3.81, + "learning_rate": 1.1244585241320664e-05, + "loss": 0.2064, + "step": 9103 + }, + { + "epoch": 3.81, + "learning_rate": 1.1242902996686333e-05, + "loss": 0.186, + "step": 9104 + }, + { + "epoch": 3.81, + "learning_rate": 1.1241220716325863e-05, + "loss": 0.1907, + "step": 9105 + }, + { + "epoch": 3.81, + "learning_rate": 1.12395384002876e-05, + "loss": 0.1635, + "step": 9106 + }, + { + "epoch": 3.81, + "learning_rate": 1.1237856048619904e-05, + "loss": 0.1679, + "step": 9107 + }, + { + "epoch": 3.81, + "learning_rate": 1.1236173661371139e-05, + "loss": 0.1374, + "step": 9108 + }, + { + "epoch": 3.81, + "learning_rate": 1.1234491238589658e-05, + "loss": 0.1645, + "step": 9109 + }, + { + "epoch": 3.81, + "learning_rate": 1.1232808780323821e-05, + "loss": 0.2045, + "step": 9110 + }, + { + "epoch": 3.81, + "learning_rate": 1.1231126286621988e-05, + "loss": 0.1994, + "step": 9111 + }, + { + "epoch": 3.81, + "learning_rate": 1.1229443757532526e-05, + "loss": 0.1896, + "step": 9112 + }, + { + "epoch": 3.82, + "learning_rate": 1.122776119310379e-05, + "loss": 0.1959, + "step": 9113 + }, + { + "epoch": 3.82, + "learning_rate": 1.122607859338415e-05, + "loss": 0.2131, + "step": 9114 + }, + { + "epoch": 3.82, + "learning_rate": 1.122439595842197e-05, + "loss": 0.1619, + "step": 9115 + }, + { + "epoch": 3.82, + "learning_rate": 1.1222713288265613e-05, + "loss": 0.1915, + "step": 9116 + }, + { + "epoch": 3.82, + "learning_rate": 1.1221030582963447e-05, + "loss": 0.1749, + "step": 9117 + }, + { + "epoch": 3.82, + "learning_rate": 1.1219347842563838e-05, + "loss": 0.1934, + "step": 9118 + }, + { + "epoch": 3.82, + "learning_rate": 1.1217665067115164e-05, + "loss": 0.1642, + "step": 9119 + }, + { + "epoch": 3.82, + "learning_rate": 1.1215982256665786e-05, + "loss": 0.2118, + "step": 9120 + }, + { + "epoch": 3.82, + "learning_rate": 1.1214299411264075e-05, + "loss": 0.1747, + "step": 9121 + }, + { + "epoch": 3.82, + "learning_rate": 1.1212616530958407e-05, + "loss": 0.1927, + "step": 9122 + }, + { + "epoch": 3.82, + "learning_rate": 1.121093361579715e-05, + "loss": 0.1762, + "step": 9123 + }, + { + "epoch": 3.82, + "learning_rate": 1.1209250665828684e-05, + "loss": 0.1817, + "step": 9124 + }, + { + "epoch": 3.82, + "learning_rate": 1.1207567681101377e-05, + "loss": 0.2285, + "step": 9125 + }, + { + "epoch": 3.82, + "learning_rate": 1.1205884661663614e-05, + "loss": 0.1896, + "step": 9126 + }, + { + "epoch": 3.82, + "learning_rate": 1.1204201607563761e-05, + "loss": 0.2009, + "step": 9127 + }, + { + "epoch": 3.82, + "learning_rate": 1.1202518518850202e-05, + "loss": 0.1918, + "step": 9128 + }, + { + "epoch": 3.82, + "learning_rate": 1.1200835395571319e-05, + "loss": 0.1798, + "step": 9129 + }, + { + "epoch": 3.82, + "learning_rate": 1.1199152237775485e-05, + "loss": 0.1773, + "step": 9130 + }, + { + "epoch": 3.82, + "learning_rate": 1.1197469045511085e-05, + "loss": 0.1865, + "step": 9131 + }, + { + "epoch": 3.82, + "learning_rate": 1.11957858188265e-05, + "loss": 0.1732, + "step": 9132 + }, + { + "epoch": 3.82, + "learning_rate": 1.1194102557770117e-05, + "loss": 0.2204, + "step": 9133 + }, + { + "epoch": 3.82, + "learning_rate": 1.119241926239031e-05, + "loss": 0.1778, + "step": 9134 + }, + { + "epoch": 3.82, + "learning_rate": 1.119073593273547e-05, + "loss": 0.1671, + "step": 9135 + }, + { + "epoch": 3.82, + "learning_rate": 1.1189052568853985e-05, + "loss": 0.2029, + "step": 9136 + }, + { + "epoch": 3.83, + "learning_rate": 1.1187369170794235e-05, + "loss": 0.2035, + "step": 9137 + }, + { + "epoch": 3.83, + "learning_rate": 1.1185685738604615e-05, + "loss": 0.1841, + "step": 9138 + }, + { + "epoch": 3.83, + "learning_rate": 1.1184002272333509e-05, + "loss": 0.1578, + "step": 9139 + }, + { + "epoch": 3.83, + "learning_rate": 1.1182318772029309e-05, + "loss": 0.1937, + "step": 9140 + }, + { + "epoch": 3.83, + "learning_rate": 1.1180635237740405e-05, + "loss": 0.1827, + "step": 9141 + }, + { + "epoch": 3.83, + "learning_rate": 1.117895166951519e-05, + "loss": 0.2154, + "step": 9142 + }, + { + "epoch": 3.83, + "learning_rate": 1.1177268067402057e-05, + "loss": 0.1988, + "step": 9143 + }, + { + "epoch": 3.83, + "learning_rate": 1.1175584431449393e-05, + "loss": 0.1743, + "step": 9144 + }, + { + "epoch": 3.83, + "learning_rate": 1.11739007617056e-05, + "loss": 0.2047, + "step": 9145 + }, + { + "epoch": 3.83, + "learning_rate": 1.1172217058219073e-05, + "loss": 0.1919, + "step": 9146 + }, + { + "epoch": 3.83, + "learning_rate": 1.1170533321038204e-05, + "loss": 0.186, + "step": 9147 + }, + { + "epoch": 3.83, + "learning_rate": 1.1168849550211394e-05, + "loss": 0.1803, + "step": 9148 + }, + { + "epoch": 3.83, + "learning_rate": 1.1167165745787041e-05, + "loss": 0.2222, + "step": 9149 + }, + { + "epoch": 3.83, + "learning_rate": 1.1165481907813544e-05, + "loss": 0.1488, + "step": 9150 + }, + { + "epoch": 3.83, + "learning_rate": 1.1163798036339303e-05, + "loss": 0.1842, + "step": 9151 + }, + { + "epoch": 3.83, + "learning_rate": 1.1162114131412723e-05, + "loss": 0.1747, + "step": 9152 + }, + { + "epoch": 3.83, + "learning_rate": 1.1160430193082203e-05, + "loss": 0.1926, + "step": 9153 + }, + { + "epoch": 3.83, + "learning_rate": 1.1158746221396148e-05, + "loss": 0.1933, + "step": 9154 + }, + { + "epoch": 3.83, + "learning_rate": 1.1157062216402959e-05, + "loss": 0.1534, + "step": 9155 + }, + { + "epoch": 3.83, + "learning_rate": 1.1155378178151044e-05, + "loss": 0.1914, + "step": 9156 + }, + { + "epoch": 3.83, + "learning_rate": 1.1153694106688809e-05, + "loss": 0.1767, + "step": 9157 + }, + { + "epoch": 3.83, + "learning_rate": 1.115201000206466e-05, + "loss": 0.1747, + "step": 9158 + }, + { + "epoch": 3.83, + "learning_rate": 1.1150325864327003e-05, + "loss": 0.193, + "step": 9159 + }, + { + "epoch": 3.84, + "learning_rate": 1.1148641693524254e-05, + "loss": 0.1653, + "step": 9160 + }, + { + "epoch": 3.84, + "learning_rate": 1.1146957489704817e-05, + "loss": 0.1859, + "step": 9161 + }, + { + "epoch": 3.84, + "learning_rate": 1.1145273252917104e-05, + "loss": 0.1975, + "step": 9162 + }, + { + "epoch": 3.84, + "learning_rate": 1.114358898320953e-05, + "loss": 0.1936, + "step": 9163 + }, + { + "epoch": 3.84, + "learning_rate": 1.1141904680630504e-05, + "loss": 0.2159, + "step": 9164 + }, + { + "epoch": 3.84, + "learning_rate": 1.114022034522844e-05, + "loss": 0.1748, + "step": 9165 + }, + { + "epoch": 3.84, + "learning_rate": 1.1138535977051757e-05, + "loss": 0.2104, + "step": 9166 + }, + { + "epoch": 3.84, + "learning_rate": 1.1136851576148867e-05, + "loss": 0.1722, + "step": 9167 + }, + { + "epoch": 3.84, + "learning_rate": 1.1135167142568186e-05, + "loss": 0.2082, + "step": 9168 + }, + { + "epoch": 3.84, + "learning_rate": 1.1133482676358131e-05, + "loss": 0.1616, + "step": 9169 + }, + { + "epoch": 3.84, + "learning_rate": 1.1131798177567129e-05, + "loss": 0.1815, + "step": 9170 + }, + { + "epoch": 3.84, + "learning_rate": 1.1130113646243585e-05, + "loss": 0.2332, + "step": 9171 + }, + { + "epoch": 3.84, + "learning_rate": 1.112842908243593e-05, + "loss": 0.1734, + "step": 9172 + }, + { + "epoch": 3.84, + "learning_rate": 1.1126744486192585e-05, + "loss": 0.1712, + "step": 9173 + }, + { + "epoch": 3.84, + "learning_rate": 1.1125059857561966e-05, + "loss": 0.2049, + "step": 9174 + }, + { + "epoch": 3.84, + "learning_rate": 1.11233751965925e-05, + "loss": 0.2336, + "step": 9175 + }, + { + "epoch": 3.84, + "learning_rate": 1.1121690503332613e-05, + "loss": 0.1608, + "step": 9176 + }, + { + "epoch": 3.84, + "learning_rate": 1.1120005777830729e-05, + "loss": 0.2013, + "step": 9177 + }, + { + "epoch": 3.84, + "learning_rate": 1.111832102013527e-05, + "loss": 0.1843, + "step": 9178 + }, + { + "epoch": 3.84, + "learning_rate": 1.1116636230294663e-05, + "loss": 0.1676, + "step": 9179 + }, + { + "epoch": 3.84, + "learning_rate": 1.1114951408357344e-05, + "loss": 0.1899, + "step": 9180 + }, + { + "epoch": 3.84, + "learning_rate": 1.1113266554371734e-05, + "loss": 0.2414, + "step": 9181 + }, + { + "epoch": 3.84, + "learning_rate": 1.1111581668386262e-05, + "loss": 0.1606, + "step": 9182 + }, + { + "epoch": 3.84, + "learning_rate": 1.1109896750449367e-05, + "loss": 0.1816, + "step": 9183 + }, + { + "epoch": 3.85, + "learning_rate": 1.110821180060947e-05, + "loss": 0.207, + "step": 9184 + }, + { + "epoch": 3.85, + "learning_rate": 1.1106526818915008e-05, + "loss": 0.1648, + "step": 9185 + }, + { + "epoch": 3.85, + "learning_rate": 1.1104841805414418e-05, + "loss": 0.2001, + "step": 9186 + }, + { + "epoch": 3.85, + "learning_rate": 1.1103156760156129e-05, + "loss": 0.154, + "step": 9187 + }, + { + "epoch": 3.85, + "learning_rate": 1.110147168318858e-05, + "loss": 0.183, + "step": 9188 + }, + { + "epoch": 3.85, + "learning_rate": 1.10997865745602e-05, + "loss": 0.2053, + "step": 9189 + }, + { + "epoch": 3.85, + "learning_rate": 1.1098101434319438e-05, + "loss": 0.2275, + "step": 9190 + }, + { + "epoch": 3.85, + "learning_rate": 1.1096416262514717e-05, + "loss": 0.197, + "step": 9191 + }, + { + "epoch": 3.85, + "learning_rate": 1.1094731059194488e-05, + "loss": 0.2249, + "step": 9192 + }, + { + "epoch": 3.85, + "learning_rate": 1.1093045824407186e-05, + "loss": 0.1649, + "step": 9193 + }, + { + "epoch": 3.85, + "learning_rate": 1.1091360558201249e-05, + "loss": 0.205, + "step": 9194 + }, + { + "epoch": 3.85, + "learning_rate": 1.1089675260625124e-05, + "loss": 0.1797, + "step": 9195 + }, + { + "epoch": 3.85, + "learning_rate": 1.1087989931727248e-05, + "loss": 0.1717, + "step": 9196 + }, + { + "epoch": 3.85, + "learning_rate": 1.1086304571556071e-05, + "loss": 0.1843, + "step": 9197 + }, + { + "epoch": 3.85, + "learning_rate": 1.1084619180160028e-05, + "loss": 0.2124, + "step": 9198 + }, + { + "epoch": 3.85, + "learning_rate": 1.1082933757587571e-05, + "loss": 0.1779, + "step": 9199 + }, + { + "epoch": 3.85, + "learning_rate": 1.1081248303887147e-05, + "loss": 0.2332, + "step": 9200 + }, + { + "epoch": 3.85, + "learning_rate": 1.1079562819107198e-05, + "loss": 0.2471, + "step": 9201 + }, + { + "epoch": 3.85, + "learning_rate": 1.1077877303296174e-05, + "loss": 0.1762, + "step": 9202 + }, + { + "epoch": 3.85, + "learning_rate": 1.1076191756502521e-05, + "loss": 0.2314, + "step": 9203 + }, + { + "epoch": 3.85, + "learning_rate": 1.1074506178774696e-05, + "loss": 0.173, + "step": 9204 + }, + { + "epoch": 3.85, + "learning_rate": 1.107282057016114e-05, + "loss": 0.2055, + "step": 9205 + }, + { + "epoch": 3.85, + "learning_rate": 1.1071134930710308e-05, + "loss": 0.1714, + "step": 9206 + }, + { + "epoch": 3.85, + "learning_rate": 1.1069449260470658e-05, + "loss": 0.1473, + "step": 9207 + }, + { + "epoch": 3.86, + "learning_rate": 1.1067763559490632e-05, + "loss": 0.149, + "step": 9208 + }, + { + "epoch": 3.86, + "learning_rate": 1.1066077827818694e-05, + "loss": 0.2025, + "step": 9209 + }, + { + "epoch": 3.86, + "learning_rate": 1.1064392065503294e-05, + "loss": 0.171, + "step": 9210 + }, + { + "epoch": 3.86, + "learning_rate": 1.106270627259289e-05, + "loss": 0.1824, + "step": 9211 + }, + { + "epoch": 3.86, + "learning_rate": 1.1061020449135935e-05, + "loss": 0.186, + "step": 9212 + }, + { + "epoch": 3.86, + "learning_rate": 1.1059334595180888e-05, + "loss": 0.1983, + "step": 9213 + }, + { + "epoch": 3.86, + "learning_rate": 1.1057648710776212e-05, + "loss": 0.1745, + "step": 9214 + }, + { + "epoch": 3.86, + "learning_rate": 1.1055962795970359e-05, + "loss": 0.2321, + "step": 9215 + }, + { + "epoch": 3.86, + "learning_rate": 1.1054276850811791e-05, + "loss": 0.2003, + "step": 9216 + }, + { + "epoch": 3.86, + "learning_rate": 1.1052590875348975e-05, + "loss": 0.2135, + "step": 9217 + }, + { + "epoch": 3.86, + "learning_rate": 1.1050904869630365e-05, + "loss": 0.2082, + "step": 9218 + }, + { + "epoch": 3.86, + "learning_rate": 1.1049218833704427e-05, + "loss": 0.2232, + "step": 9219 + }, + { + "epoch": 3.86, + "learning_rate": 1.1047532767619629e-05, + "loss": 0.2073, + "step": 9220 + }, + { + "epoch": 3.86, + "learning_rate": 1.1045846671424424e-05, + "loss": 0.1981, + "step": 9221 + }, + { + "epoch": 3.86, + "learning_rate": 1.1044160545167294e-05, + "loss": 0.1434, + "step": 9222 + }, + { + "epoch": 3.86, + "learning_rate": 1.1042474388896687e-05, + "loss": 0.1945, + "step": 9223 + }, + { + "epoch": 3.86, + "learning_rate": 1.1040788202661084e-05, + "loss": 0.1881, + "step": 9224 + }, + { + "epoch": 3.86, + "learning_rate": 1.1039101986508945e-05, + "loss": 0.1577, + "step": 9225 + }, + { + "epoch": 3.86, + "learning_rate": 1.1037415740488742e-05, + "loss": 0.1527, + "step": 9226 + }, + { + "epoch": 3.86, + "learning_rate": 1.1035729464648945e-05, + "loss": 0.2051, + "step": 9227 + }, + { + "epoch": 3.86, + "learning_rate": 1.1034043159038021e-05, + "loss": 0.1761, + "step": 9228 + }, + { + "epoch": 3.86, + "learning_rate": 1.1032356823704446e-05, + "loss": 0.2139, + "step": 9229 + }, + { + "epoch": 3.86, + "learning_rate": 1.1030670458696688e-05, + "loss": 0.2011, + "step": 9230 + }, + { + "epoch": 3.86, + "learning_rate": 1.1028984064063226e-05, + "loss": 0.1833, + "step": 9231 + }, + { + "epoch": 3.87, + "learning_rate": 1.1027297639852528e-05, + "loss": 0.1715, + "step": 9232 + }, + { + "epoch": 3.87, + "learning_rate": 1.1025611186113073e-05, + "loss": 0.1462, + "step": 9233 + }, + { + "epoch": 3.87, + "learning_rate": 1.1023924702893334e-05, + "loss": 0.1673, + "step": 9234 + }, + { + "epoch": 3.87, + "learning_rate": 1.1022238190241787e-05, + "loss": 0.1945, + "step": 9235 + }, + { + "epoch": 3.87, + "learning_rate": 1.102055164820691e-05, + "loss": 0.1562, + "step": 9236 + }, + { + "epoch": 3.87, + "learning_rate": 1.1018865076837182e-05, + "loss": 0.1894, + "step": 9237 + }, + { + "epoch": 3.87, + "learning_rate": 1.101717847618108e-05, + "loss": 0.1953, + "step": 9238 + }, + { + "epoch": 3.87, + "learning_rate": 1.1015491846287088e-05, + "loss": 0.2008, + "step": 9239 + }, + { + "epoch": 3.87, + "learning_rate": 1.1013805187203682e-05, + "loss": 0.2021, + "step": 9240 + }, + { + "epoch": 3.87, + "learning_rate": 1.1012118498979347e-05, + "loss": 0.199, + "step": 9241 + }, + { + "epoch": 3.87, + "learning_rate": 1.1010431781662563e-05, + "loss": 0.1611, + "step": 9242 + }, + { + "epoch": 3.87, + "learning_rate": 1.1008745035301814e-05, + "loss": 0.208, + "step": 9243 + }, + { + "epoch": 3.87, + "learning_rate": 1.1007058259945584e-05, + "loss": 0.1939, + "step": 9244 + }, + { + "epoch": 3.87, + "learning_rate": 1.100537145564236e-05, + "loss": 0.1942, + "step": 9245 + }, + { + "epoch": 3.87, + "learning_rate": 1.1003684622440622e-05, + "loss": 0.1912, + "step": 9246 + }, + { + "epoch": 3.87, + "learning_rate": 1.1001997760388862e-05, + "loss": 0.2018, + "step": 9247 + }, + { + "epoch": 3.87, + "learning_rate": 1.100031086953557e-05, + "loss": 0.1898, + "step": 9248 + }, + { + "epoch": 3.87, + "learning_rate": 1.0998623949929223e-05, + "loss": 0.1977, + "step": 9249 + }, + { + "epoch": 3.87, + "learning_rate": 1.0996937001618317e-05, + "loss": 0.1582, + "step": 9250 + }, + { + "epoch": 3.87, + "learning_rate": 1.0995250024651345e-05, + "loss": 0.1945, + "step": 9251 + }, + { + "epoch": 3.87, + "learning_rate": 1.0993563019076794e-05, + "loss": 0.1894, + "step": 9252 + }, + { + "epoch": 3.87, + "learning_rate": 1.0991875984943152e-05, + "loss": 0.1623, + "step": 9253 + }, + { + "epoch": 3.87, + "learning_rate": 1.0990188922298918e-05, + "loss": 0.1477, + "step": 9254 + }, + { + "epoch": 3.87, + "learning_rate": 1.0988501831192582e-05, + "loss": 0.1554, + "step": 9255 + }, + { + "epoch": 3.88, + "learning_rate": 1.098681471167264e-05, + "loss": 0.2056, + "step": 9256 + }, + { + "epoch": 3.88, + "learning_rate": 1.098512756378758e-05, + "loss": 0.2109, + "step": 9257 + }, + { + "epoch": 3.88, + "learning_rate": 1.0983440387585909e-05, + "loss": 0.15, + "step": 9258 + }, + { + "epoch": 3.88, + "learning_rate": 1.0981753183116111e-05, + "loss": 0.2154, + "step": 9259 + }, + { + "epoch": 3.88, + "learning_rate": 1.0980065950426687e-05, + "loss": 0.2, + "step": 9260 + }, + { + "epoch": 3.88, + "learning_rate": 1.0978378689566144e-05, + "loss": 0.217, + "step": 9261 + }, + { + "epoch": 3.88, + "learning_rate": 1.097669140058297e-05, + "loss": 0.2133, + "step": 9262 + }, + { + "epoch": 3.88, + "learning_rate": 1.0975004083525667e-05, + "loss": 0.198, + "step": 9263 + }, + { + "epoch": 3.88, + "learning_rate": 1.0973316738442738e-05, + "loss": 0.2014, + "step": 9264 + }, + { + "epoch": 3.88, + "learning_rate": 1.0971629365382686e-05, + "loss": 0.1903, + "step": 9265 + }, + { + "epoch": 3.88, + "learning_rate": 1.0969941964394006e-05, + "loss": 0.1916, + "step": 9266 + }, + { + "epoch": 3.88, + "learning_rate": 1.0968254535525204e-05, + "loss": 0.2014, + "step": 9267 + }, + { + "epoch": 3.88, + "learning_rate": 1.0966567078824792e-05, + "loss": 0.1985, + "step": 9268 + }, + { + "epoch": 3.88, + "learning_rate": 1.0964879594341262e-05, + "loss": 0.1646, + "step": 9269 + }, + { + "epoch": 3.88, + "learning_rate": 1.0963192082123124e-05, + "loss": 0.1842, + "step": 9270 + }, + { + "epoch": 3.88, + "learning_rate": 1.0961504542218884e-05, + "loss": 0.2025, + "step": 9271 + }, + { + "epoch": 3.88, + "learning_rate": 1.0959816974677052e-05, + "loss": 0.1725, + "step": 9272 + }, + { + "epoch": 3.88, + "learning_rate": 1.095812937954613e-05, + "loss": 0.2005, + "step": 9273 + }, + { + "epoch": 3.88, + "learning_rate": 1.0956441756874628e-05, + "loss": 0.1933, + "step": 9274 + }, + { + "epoch": 3.88, + "learning_rate": 1.095475410671106e-05, + "loss": 0.1843, + "step": 9275 + }, + { + "epoch": 3.88, + "learning_rate": 1.0953066429103929e-05, + "loss": 0.2184, + "step": 9276 + }, + { + "epoch": 3.88, + "learning_rate": 1.0951378724101751e-05, + "loss": 0.2005, + "step": 9277 + }, + { + "epoch": 3.88, + "learning_rate": 1.0949690991753036e-05, + "loss": 0.1757, + "step": 9278 + }, + { + "epoch": 3.88, + "learning_rate": 1.0948003232106298e-05, + "loss": 0.1942, + "step": 9279 + }, + { + "epoch": 3.89, + "learning_rate": 1.0946315445210048e-05, + "loss": 0.1583, + "step": 9280 + }, + { + "epoch": 3.89, + "learning_rate": 1.0944627631112796e-05, + "loss": 0.1767, + "step": 9281 + }, + { + "epoch": 3.89, + "learning_rate": 1.0942939789863065e-05, + "loss": 0.2008, + "step": 9282 + }, + { + "epoch": 3.89, + "learning_rate": 1.0941251921509367e-05, + "loss": 0.1965, + "step": 9283 + }, + { + "epoch": 3.89, + "learning_rate": 1.0939564026100213e-05, + "loss": 0.2036, + "step": 9284 + }, + { + "epoch": 3.89, + "learning_rate": 1.093787610368413e-05, + "loss": 0.1682, + "step": 9285 + }, + { + "epoch": 3.89, + "learning_rate": 1.0936188154309626e-05, + "loss": 0.1771, + "step": 9286 + }, + { + "epoch": 3.89, + "learning_rate": 1.0934500178025228e-05, + "loss": 0.1683, + "step": 9287 + }, + { + "epoch": 3.89, + "learning_rate": 1.093281217487945e-05, + "loss": 0.1783, + "step": 9288 + }, + { + "epoch": 3.89, + "learning_rate": 1.0931124144920813e-05, + "loss": 0.2245, + "step": 9289 + }, + { + "epoch": 3.89, + "learning_rate": 1.0929436088197842e-05, + "loss": 0.2076, + "step": 9290 + }, + { + "epoch": 3.89, + "learning_rate": 1.0927748004759053e-05, + "loss": 0.1446, + "step": 9291 + }, + { + "epoch": 3.89, + "learning_rate": 1.0926059894652972e-05, + "loss": 0.1343, + "step": 9292 + }, + { + "epoch": 3.89, + "learning_rate": 1.0924371757928121e-05, + "loss": 0.1693, + "step": 9293 + }, + { + "epoch": 3.89, + "learning_rate": 1.092268359463302e-05, + "loss": 0.2063, + "step": 9294 + }, + { + "epoch": 3.89, + "learning_rate": 1.0920995404816204e-05, + "loss": 0.1792, + "step": 9295 + }, + { + "epoch": 3.89, + "learning_rate": 1.091930718852619e-05, + "loss": 0.1761, + "step": 9296 + }, + { + "epoch": 3.89, + "learning_rate": 1.0917618945811503e-05, + "loss": 0.1832, + "step": 9297 + }, + { + "epoch": 3.89, + "learning_rate": 1.0915930676720678e-05, + "loss": 0.1779, + "step": 9298 + }, + { + "epoch": 3.89, + "learning_rate": 1.0914242381302238e-05, + "loss": 0.1716, + "step": 9299 + }, + { + "epoch": 3.89, + "learning_rate": 1.091255405960471e-05, + "loss": 0.173, + "step": 9300 + }, + { + "epoch": 3.89, + "learning_rate": 1.0910865711676628e-05, + "loss": 0.1813, + "step": 9301 + }, + { + "epoch": 3.89, + "learning_rate": 1.0909177337566518e-05, + "loss": 0.1746, + "step": 9302 + }, + { + "epoch": 3.89, + "learning_rate": 1.0907488937322915e-05, + "loss": 0.1858, + "step": 9303 + }, + { + "epoch": 3.9, + "learning_rate": 1.0905800510994341e-05, + "loss": 0.2029, + "step": 9304 + }, + { + "epoch": 3.9, + "learning_rate": 1.0904112058629341e-05, + "loss": 0.1727, + "step": 9305 + }, + { + "epoch": 3.9, + "learning_rate": 1.090242358027644e-05, + "loss": 0.1983, + "step": 9306 + }, + { + "epoch": 3.9, + "learning_rate": 1.0900735075984174e-05, + "loss": 0.1724, + "step": 9307 + }, + { + "epoch": 3.9, + "learning_rate": 1.0899046545801078e-05, + "loss": 0.1676, + "step": 9308 + }, + { + "epoch": 3.9, + "learning_rate": 1.0897357989775687e-05, + "loss": 0.1845, + "step": 9309 + }, + { + "epoch": 3.9, + "learning_rate": 1.0895669407956538e-05, + "loss": 0.1733, + "step": 9310 + }, + { + "epoch": 3.9, + "learning_rate": 1.0893980800392164e-05, + "loss": 0.209, + "step": 9311 + }, + { + "epoch": 3.9, + "learning_rate": 1.0892292167131107e-05, + "loss": 0.2143, + "step": 9312 + }, + { + "epoch": 3.9, + "learning_rate": 1.0890603508221905e-05, + "loss": 0.1837, + "step": 9313 + }, + { + "epoch": 3.9, + "learning_rate": 1.0888914823713092e-05, + "loss": 0.2079, + "step": 9314 + }, + { + "epoch": 3.9, + "learning_rate": 1.0887226113653211e-05, + "loss": 0.1714, + "step": 9315 + }, + { + "epoch": 3.9, + "learning_rate": 1.0885537378090807e-05, + "loss": 0.1785, + "step": 9316 + }, + { + "epoch": 3.9, + "learning_rate": 1.0883848617074412e-05, + "loss": 0.1692, + "step": 9317 + }, + { + "epoch": 3.9, + "learning_rate": 1.0882159830652574e-05, + "loss": 0.2079, + "step": 9318 + }, + { + "epoch": 3.9, + "learning_rate": 1.0880471018873836e-05, + "loss": 0.2076, + "step": 9319 + }, + { + "epoch": 3.9, + "learning_rate": 1.087878218178674e-05, + "loss": 0.2011, + "step": 9320 + }, + { + "epoch": 3.9, + "learning_rate": 1.0877093319439828e-05, + "loss": 0.1759, + "step": 9321 + }, + { + "epoch": 3.9, + "learning_rate": 1.0875404431881651e-05, + "loss": 0.2012, + "step": 9322 + }, + { + "epoch": 3.9, + "learning_rate": 1.0873715519160746e-05, + "loss": 0.1937, + "step": 9323 + }, + { + "epoch": 3.9, + "learning_rate": 1.0872026581325667e-05, + "loss": 0.1947, + "step": 9324 + }, + { + "epoch": 3.9, + "learning_rate": 1.0870337618424956e-05, + "loss": 0.1981, + "step": 9325 + }, + { + "epoch": 3.9, + "learning_rate": 1.0868648630507164e-05, + "loss": 0.2003, + "step": 9326 + }, + { + "epoch": 3.9, + "learning_rate": 1.0866959617620835e-05, + "loss": 0.2131, + "step": 9327 + }, + { + "epoch": 3.91, + "learning_rate": 1.0865270579814522e-05, + "loss": 0.1698, + "step": 9328 + }, + { + "epoch": 3.91, + "learning_rate": 1.0863581517136776e-05, + "loss": 0.1846, + "step": 9329 + }, + { + "epoch": 3.91, + "learning_rate": 1.0861892429636143e-05, + "loss": 0.1581, + "step": 9330 + }, + { + "epoch": 3.91, + "learning_rate": 1.0860203317361178e-05, + "loss": 0.1566, + "step": 9331 + }, + { + "epoch": 3.91, + "learning_rate": 1.0858514180360432e-05, + "loss": 0.2144, + "step": 9332 + }, + { + "epoch": 3.91, + "learning_rate": 1.085682501868246e-05, + "loss": 0.1654, + "step": 9333 + }, + { + "epoch": 3.91, + "learning_rate": 1.0855135832375809e-05, + "loss": 0.2098, + "step": 9334 + }, + { + "epoch": 3.91, + "learning_rate": 1.0853446621489042e-05, + "loss": 0.2019, + "step": 9335 + }, + { + "epoch": 3.91, + "learning_rate": 1.0851757386070706e-05, + "loss": 0.1809, + "step": 9336 + }, + { + "epoch": 3.91, + "learning_rate": 1.085006812616936e-05, + "loss": 0.1719, + "step": 9337 + }, + { + "epoch": 3.91, + "learning_rate": 1.084837884183356e-05, + "loss": 0.2099, + "step": 9338 + }, + { + "epoch": 3.91, + "learning_rate": 1.0846689533111865e-05, + "loss": 0.1807, + "step": 9339 + }, + { + "epoch": 3.91, + "learning_rate": 1.0845000200052828e-05, + "loss": 0.198, + "step": 9340 + }, + { + "epoch": 3.91, + "learning_rate": 1.0843310842705011e-05, + "loss": 0.1822, + "step": 9341 + }, + { + "epoch": 3.91, + "learning_rate": 1.0841621461116971e-05, + "loss": 0.2146, + "step": 9342 + }, + { + "epoch": 3.91, + "learning_rate": 1.0839932055337273e-05, + "loss": 0.1954, + "step": 9343 + }, + { + "epoch": 3.91, + "learning_rate": 1.083824262541447e-05, + "loss": 0.1589, + "step": 9344 + }, + { + "epoch": 3.91, + "learning_rate": 1.0836553171397126e-05, + "loss": 0.2323, + "step": 9345 + }, + { + "epoch": 3.91, + "learning_rate": 1.0834863693333805e-05, + "loss": 0.1982, + "step": 9346 + }, + { + "epoch": 3.91, + "learning_rate": 1.083317419127307e-05, + "loss": 0.1637, + "step": 9347 + }, + { + "epoch": 3.91, + "learning_rate": 1.0831484665263479e-05, + "loss": 0.1847, + "step": 9348 + }, + { + "epoch": 3.91, + "learning_rate": 1.0829795115353601e-05, + "loss": 0.1476, + "step": 9349 + }, + { + "epoch": 3.91, + "learning_rate": 1.0828105541591998e-05, + "loss": 0.1757, + "step": 9350 + }, + { + "epoch": 3.92, + "learning_rate": 1.0826415944027235e-05, + "loss": 0.214, + "step": 9351 + }, + { + "epoch": 3.92, + "learning_rate": 1.0824726322707878e-05, + "loss": 0.217, + "step": 9352 + }, + { + "epoch": 3.92, + "learning_rate": 1.0823036677682499e-05, + "loss": 0.1895, + "step": 9353 + }, + { + "epoch": 3.92, + "learning_rate": 1.0821347008999658e-05, + "loss": 0.1615, + "step": 9354 + }, + { + "epoch": 3.92, + "learning_rate": 1.0819657316707926e-05, + "loss": 0.1782, + "step": 9355 + }, + { + "epoch": 3.92, + "learning_rate": 1.0817967600855874e-05, + "loss": 0.1674, + "step": 9356 + }, + { + "epoch": 3.92, + "learning_rate": 1.0816277861492068e-05, + "loss": 0.1784, + "step": 9357 + }, + { + "epoch": 3.92, + "learning_rate": 1.081458809866508e-05, + "loss": 0.1907, + "step": 9358 + }, + { + "epoch": 3.92, + "learning_rate": 1.0812898312423478e-05, + "loss": 0.1914, + "step": 9359 + }, + { + "epoch": 3.92, + "learning_rate": 1.0811208502815838e-05, + "loss": 0.21, + "step": 9360 + }, + { + "epoch": 3.92, + "learning_rate": 1.0809518669890726e-05, + "loss": 0.1671, + "step": 9361 + }, + { + "epoch": 3.92, + "learning_rate": 1.0807828813696719e-05, + "loss": 0.1829, + "step": 9362 + }, + { + "epoch": 3.92, + "learning_rate": 1.0806138934282393e-05, + "loss": 0.2003, + "step": 9363 + }, + { + "epoch": 3.92, + "learning_rate": 1.0804449031696315e-05, + "loss": 0.1604, + "step": 9364 + }, + { + "epoch": 3.92, + "learning_rate": 1.0802759105987065e-05, + "loss": 0.1791, + "step": 9365 + }, + { + "epoch": 3.92, + "learning_rate": 1.0801069157203217e-05, + "loss": 0.1715, + "step": 9366 + }, + { + "epoch": 3.92, + "learning_rate": 1.0799379185393347e-05, + "loss": 0.1648, + "step": 9367 + }, + { + "epoch": 3.92, + "learning_rate": 1.0797689190606028e-05, + "loss": 0.1898, + "step": 9368 + }, + { + "epoch": 3.92, + "learning_rate": 1.0795999172889847e-05, + "loss": 0.1862, + "step": 9369 + }, + { + "epoch": 3.92, + "learning_rate": 1.0794309132293374e-05, + "loss": 0.1888, + "step": 9370 + }, + { + "epoch": 3.92, + "learning_rate": 1.0792619068865189e-05, + "loss": 0.1692, + "step": 9371 + }, + { + "epoch": 3.92, + "learning_rate": 1.0790928982653873e-05, + "loss": 0.2446, + "step": 9372 + }, + { + "epoch": 3.92, + "learning_rate": 1.0789238873708004e-05, + "loss": 0.2284, + "step": 9373 + }, + { + "epoch": 3.92, + "learning_rate": 1.0787548742076164e-05, + "loss": 0.1938, + "step": 9374 + }, + { + "epoch": 3.93, + "learning_rate": 1.0785858587806935e-05, + "loss": 0.1542, + "step": 9375 + }, + { + "epoch": 3.93, + "learning_rate": 1.0784168410948897e-05, + "loss": 0.164, + "step": 9376 + }, + { + "epoch": 3.93, + "learning_rate": 1.0782478211550636e-05, + "loss": 0.1857, + "step": 9377 + }, + { + "epoch": 3.93, + "learning_rate": 1.078078798966073e-05, + "loss": 0.189, + "step": 9378 + }, + { + "epoch": 3.93, + "learning_rate": 1.0779097745327768e-05, + "loss": 0.2207, + "step": 9379 + }, + { + "epoch": 3.93, + "learning_rate": 1.0777407478600334e-05, + "loss": 0.1534, + "step": 9380 + }, + { + "epoch": 3.93, + "learning_rate": 1.0775717189527012e-05, + "loss": 0.2344, + "step": 9381 + }, + { + "epoch": 3.93, + "learning_rate": 1.0774026878156387e-05, + "loss": 0.218, + "step": 9382 + }, + { + "epoch": 3.93, + "learning_rate": 1.0772336544537045e-05, + "loss": 0.2319, + "step": 9383 + }, + { + "epoch": 3.93, + "learning_rate": 1.0770646188717578e-05, + "loss": 0.177, + "step": 9384 + }, + { + "epoch": 3.93, + "learning_rate": 1.0768955810746564e-05, + "loss": 0.1836, + "step": 9385 + }, + { + "epoch": 3.93, + "learning_rate": 1.07672654106726e-05, + "loss": 0.1955, + "step": 9386 + }, + { + "epoch": 3.93, + "learning_rate": 1.0765574988544276e-05, + "loss": 0.1846, + "step": 9387 + }, + { + "epoch": 3.93, + "learning_rate": 1.0763884544410174e-05, + "loss": 0.1924, + "step": 9388 + }, + { + "epoch": 3.93, + "learning_rate": 1.0762194078318891e-05, + "loss": 0.1258, + "step": 9389 + }, + { + "epoch": 3.93, + "learning_rate": 1.0760503590319016e-05, + "loss": 0.1639, + "step": 9390 + }, + { + "epoch": 3.93, + "learning_rate": 1.0758813080459138e-05, + "loss": 0.1734, + "step": 9391 + }, + { + "epoch": 3.93, + "learning_rate": 1.0757122548787855e-05, + "loss": 0.1761, + "step": 9392 + }, + { + "epoch": 3.93, + "learning_rate": 1.0755431995353753e-05, + "loss": 0.199, + "step": 9393 + }, + { + "epoch": 3.93, + "learning_rate": 1.0753741420205431e-05, + "loss": 0.1722, + "step": 9394 + }, + { + "epoch": 3.93, + "learning_rate": 1.0752050823391478e-05, + "loss": 0.1927, + "step": 9395 + }, + { + "epoch": 3.93, + "learning_rate": 1.0750360204960493e-05, + "loss": 0.1614, + "step": 9396 + }, + { + "epoch": 3.93, + "learning_rate": 1.074866956496107e-05, + "loss": 0.2037, + "step": 9397 + }, + { + "epoch": 3.93, + "learning_rate": 1.0746978903441803e-05, + "loss": 0.1731, + "step": 9398 + }, + { + "epoch": 3.94, + "learning_rate": 1.0745288220451291e-05, + "loss": 0.1679, + "step": 9399 + }, + { + "epoch": 3.94, + "learning_rate": 1.0743597516038132e-05, + "loss": 0.1661, + "step": 9400 + }, + { + "epoch": 3.94, + "learning_rate": 1.074190679025092e-05, + "loss": 0.2065, + "step": 9401 + }, + { + "epoch": 3.94, + "learning_rate": 1.0740216043138256e-05, + "loss": 0.1841, + "step": 9402 + }, + { + "epoch": 3.94, + "learning_rate": 1.073852527474874e-05, + "loss": 0.1781, + "step": 9403 + }, + { + "epoch": 3.94, + "learning_rate": 1.0736834485130972e-05, + "loss": 0.1686, + "step": 9404 + }, + { + "epoch": 3.94, + "learning_rate": 1.0735143674333549e-05, + "loss": 0.188, + "step": 9405 + }, + { + "epoch": 3.94, + "learning_rate": 1.073345284240507e-05, + "loss": 0.1495, + "step": 9406 + }, + { + "epoch": 3.94, + "learning_rate": 1.0731761989394145e-05, + "loss": 0.2022, + "step": 9407 + }, + { + "epoch": 3.94, + "learning_rate": 1.0730071115349367e-05, + "loss": 0.2069, + "step": 9408 + }, + { + "epoch": 3.94, + "learning_rate": 1.0728380220319345e-05, + "loss": 0.2076, + "step": 9409 + }, + { + "epoch": 3.94, + "learning_rate": 1.072668930435268e-05, + "loss": 0.2082, + "step": 9410 + }, + { + "epoch": 3.94, + "learning_rate": 1.0724998367497974e-05, + "loss": 0.2061, + "step": 9411 + }, + { + "epoch": 3.94, + "learning_rate": 1.0723307409803834e-05, + "loss": 0.2191, + "step": 9412 + }, + { + "epoch": 3.94, + "learning_rate": 1.0721616431318865e-05, + "loss": 0.1913, + "step": 9413 + }, + { + "epoch": 3.94, + "learning_rate": 1.0719925432091671e-05, + "loss": 0.2349, + "step": 9414 + }, + { + "epoch": 3.94, + "learning_rate": 1.0718234412170862e-05, + "loss": 0.1887, + "step": 9415 + }, + { + "epoch": 3.94, + "learning_rate": 1.0716543371605039e-05, + "loss": 0.1745, + "step": 9416 + }, + { + "epoch": 3.94, + "learning_rate": 1.0714852310442817e-05, + "loss": 0.1596, + "step": 9417 + }, + { + "epoch": 3.94, + "learning_rate": 1.0713161228732795e-05, + "loss": 0.2257, + "step": 9418 + }, + { + "epoch": 3.94, + "learning_rate": 1.0711470126523587e-05, + "loss": 0.2298, + "step": 9419 + }, + { + "epoch": 3.94, + "learning_rate": 1.0709779003863801e-05, + "loss": 0.2116, + "step": 9420 + }, + { + "epoch": 3.94, + "learning_rate": 1.070808786080205e-05, + "loss": 0.1933, + "step": 9421 + }, + { + "epoch": 3.94, + "learning_rate": 1.070639669738694e-05, + "loss": 0.1895, + "step": 9422 + }, + { + "epoch": 3.95, + "learning_rate": 1.0704705513667084e-05, + "loss": 0.2109, + "step": 9423 + }, + { + "epoch": 3.95, + "learning_rate": 1.0703014309691096e-05, + "loss": 0.1643, + "step": 9424 + }, + { + "epoch": 3.95, + "learning_rate": 1.0701323085507583e-05, + "loss": 0.1672, + "step": 9425 + }, + { + "epoch": 3.95, + "learning_rate": 1.0699631841165162e-05, + "loss": 0.1769, + "step": 9426 + }, + { + "epoch": 3.95, + "learning_rate": 1.069794057671244e-05, + "loss": 0.2088, + "step": 9427 + }, + { + "epoch": 3.95, + "learning_rate": 1.069624929219804e-05, + "loss": 0.1518, + "step": 9428 + }, + { + "epoch": 3.95, + "learning_rate": 1.069455798767057e-05, + "loss": 0.1869, + "step": 9429 + }, + { + "epoch": 3.95, + "learning_rate": 1.0692866663178646e-05, + "loss": 0.1764, + "step": 9430 + }, + { + "epoch": 3.95, + "learning_rate": 1.0691175318770887e-05, + "loss": 0.1703, + "step": 9431 + }, + { + "epoch": 3.95, + "learning_rate": 1.0689483954495906e-05, + "loss": 0.1435, + "step": 9432 + }, + { + "epoch": 3.95, + "learning_rate": 1.068779257040232e-05, + "loss": 0.1748, + "step": 9433 + }, + { + "epoch": 3.95, + "learning_rate": 1.068610116653875e-05, + "loss": 0.1518, + "step": 9434 + }, + { + "epoch": 3.95, + "learning_rate": 1.0684409742953805e-05, + "loss": 0.1742, + "step": 9435 + }, + { + "epoch": 3.95, + "learning_rate": 1.0682718299696113e-05, + "loss": 0.1625, + "step": 9436 + }, + { + "epoch": 3.95, + "learning_rate": 1.068102683681429e-05, + "loss": 0.1579, + "step": 9437 + }, + { + "epoch": 3.95, + "learning_rate": 1.0679335354356954e-05, + "loss": 0.2154, + "step": 9438 + }, + { + "epoch": 3.95, + "learning_rate": 1.0677643852372726e-05, + "loss": 0.1967, + "step": 9439 + }, + { + "epoch": 3.95, + "learning_rate": 1.0675952330910224e-05, + "loss": 0.1679, + "step": 9440 + }, + { + "epoch": 3.95, + "learning_rate": 1.0674260790018076e-05, + "loss": 0.1793, + "step": 9441 + }, + { + "epoch": 3.95, + "learning_rate": 1.0672569229744899e-05, + "loss": 0.2155, + "step": 9442 + }, + { + "epoch": 3.95, + "learning_rate": 1.0670877650139316e-05, + "loss": 0.1693, + "step": 9443 + }, + { + "epoch": 3.95, + "learning_rate": 1.0669186051249948e-05, + "loss": 0.1585, + "step": 9444 + }, + { + "epoch": 3.95, + "learning_rate": 1.0667494433125423e-05, + "loss": 0.1796, + "step": 9445 + }, + { + "epoch": 3.95, + "learning_rate": 1.0665802795814362e-05, + "loss": 0.1399, + "step": 9446 + }, + { + "epoch": 3.96, + "learning_rate": 1.0664111139365392e-05, + "loss": 0.1708, + "step": 9447 + }, + { + "epoch": 3.96, + "learning_rate": 1.0662419463827136e-05, + "loss": 0.1761, + "step": 9448 + }, + { + "epoch": 3.96, + "learning_rate": 1.0660727769248221e-05, + "loss": 0.1746, + "step": 9449 + }, + { + "epoch": 3.96, + "learning_rate": 1.0659036055677272e-05, + "loss": 0.1718, + "step": 9450 + }, + { + "epoch": 3.96, + "learning_rate": 1.0657344323162916e-05, + "loss": 0.2022, + "step": 9451 + }, + { + "epoch": 3.96, + "learning_rate": 1.0655652571753781e-05, + "loss": 0.1998, + "step": 9452 + }, + { + "epoch": 3.96, + "learning_rate": 1.0653960801498493e-05, + "loss": 0.1449, + "step": 9453 + }, + { + "epoch": 3.96, + "learning_rate": 1.0652269012445684e-05, + "loss": 0.2073, + "step": 9454 + }, + { + "epoch": 3.96, + "learning_rate": 1.0650577204643981e-05, + "loss": 0.2105, + "step": 9455 + }, + { + "epoch": 3.96, + "learning_rate": 1.0648885378142013e-05, + "loss": 0.159, + "step": 9456 + }, + { + "epoch": 3.96, + "learning_rate": 1.0647193532988409e-05, + "loss": 0.2047, + "step": 9457 + }, + { + "epoch": 3.96, + "learning_rate": 1.0645501669231803e-05, + "loss": 0.1614, + "step": 9458 + }, + { + "epoch": 3.96, + "learning_rate": 1.0643809786920824e-05, + "loss": 0.2312, + "step": 9459 + }, + { + "epoch": 3.96, + "learning_rate": 1.0642117886104106e-05, + "loss": 0.195, + "step": 9460 + }, + { + "epoch": 3.96, + "learning_rate": 1.0640425966830276e-05, + "loss": 0.1706, + "step": 9461 + }, + { + "epoch": 3.96, + "learning_rate": 1.0638734029147974e-05, + "loss": 0.1486, + "step": 9462 + }, + { + "epoch": 3.96, + "learning_rate": 1.0637042073105826e-05, + "loss": 0.1422, + "step": 9463 + }, + { + "epoch": 3.96, + "learning_rate": 1.0635350098752468e-05, + "loss": 0.1756, + "step": 9464 + }, + { + "epoch": 3.96, + "learning_rate": 1.0633658106136538e-05, + "loss": 0.2265, + "step": 9465 + }, + { + "epoch": 3.96, + "learning_rate": 1.0631966095306668e-05, + "loss": 0.2164, + "step": 9466 + }, + { + "epoch": 3.96, + "learning_rate": 1.063027406631149e-05, + "loss": 0.204, + "step": 9467 + }, + { + "epoch": 3.96, + "learning_rate": 1.0628582019199648e-05, + "loss": 0.2023, + "step": 9468 + }, + { + "epoch": 3.96, + "learning_rate": 1.0626889954019768e-05, + "loss": 0.1844, + "step": 9469 + }, + { + "epoch": 3.96, + "learning_rate": 1.0625197870820493e-05, + "loss": 0.1986, + "step": 9470 + }, + { + "epoch": 3.97, + "learning_rate": 1.0623505769650466e-05, + "loss": 0.1838, + "step": 9471 + }, + { + "epoch": 3.97, + "learning_rate": 1.0621813650558317e-05, + "loss": 0.1745, + "step": 9472 + }, + { + "epoch": 3.97, + "learning_rate": 1.0620121513592682e-05, + "loss": 0.1978, + "step": 9473 + }, + { + "epoch": 3.97, + "learning_rate": 1.0618429358802209e-05, + "loss": 0.2103, + "step": 9474 + }, + { + "epoch": 3.97, + "learning_rate": 1.061673718623553e-05, + "loss": 0.1763, + "step": 9475 + }, + { + "epoch": 3.97, + "learning_rate": 1.061504499594129e-05, + "loss": 0.1444, + "step": 9476 + }, + { + "epoch": 3.97, + "learning_rate": 1.0613352787968125e-05, + "loss": 0.1943, + "step": 9477 + }, + { + "epoch": 3.97, + "learning_rate": 1.061166056236468e-05, + "loss": 0.1833, + "step": 9478 + }, + { + "epoch": 3.97, + "learning_rate": 1.0609968319179595e-05, + "loss": 0.1546, + "step": 9479 + }, + { + "epoch": 3.97, + "learning_rate": 1.060827605846151e-05, + "loss": 0.2261, + "step": 9480 + }, + { + "epoch": 3.97, + "learning_rate": 1.0606583780259073e-05, + "loss": 0.1785, + "step": 9481 + }, + { + "epoch": 3.97, + "learning_rate": 1.060489148462092e-05, + "loss": 0.1689, + "step": 9482 + }, + { + "epoch": 3.97, + "learning_rate": 1.0603199171595701e-05, + "loss": 0.1889, + "step": 9483 + }, + { + "epoch": 3.97, + "learning_rate": 1.0601506841232057e-05, + "loss": 0.1917, + "step": 9484 + }, + { + "epoch": 3.97, + "learning_rate": 1.059981449357863e-05, + "loss": 0.2113, + "step": 9485 + }, + { + "epoch": 3.97, + "learning_rate": 1.0598122128684069e-05, + "loss": 0.1792, + "step": 9486 + }, + { + "epoch": 3.97, + "learning_rate": 1.0596429746597017e-05, + "loss": 0.2004, + "step": 9487 + }, + { + "epoch": 3.97, + "learning_rate": 1.0594737347366125e-05, + "loss": 0.166, + "step": 9488 + }, + { + "epoch": 3.97, + "learning_rate": 1.059304493104003e-05, + "loss": 0.1494, + "step": 9489 + }, + { + "epoch": 3.97, + "learning_rate": 1.0591352497667388e-05, + "loss": 0.179, + "step": 9490 + }, + { + "epoch": 3.97, + "learning_rate": 1.0589660047296844e-05, + "loss": 0.1768, + "step": 9491 + }, + { + "epoch": 3.97, + "learning_rate": 1.0587967579977043e-05, + "loss": 0.2044, + "step": 9492 + }, + { + "epoch": 3.97, + "learning_rate": 1.0586275095756638e-05, + "loss": 0.1723, + "step": 9493 + }, + { + "epoch": 3.97, + "learning_rate": 1.0584582594684274e-05, + "loss": 0.1643, + "step": 9494 + }, + { + "epoch": 3.98, + "learning_rate": 1.0582890076808603e-05, + "loss": 0.1814, + "step": 9495 + }, + { + "epoch": 3.98, + "learning_rate": 1.0581197542178274e-05, + "loss": 0.1636, + "step": 9496 + }, + { + "epoch": 3.98, + "learning_rate": 1.0579504990841935e-05, + "loss": 0.2118, + "step": 9497 + }, + { + "epoch": 3.98, + "learning_rate": 1.0577812422848241e-05, + "loss": 0.1787, + "step": 9498 + }, + { + "epoch": 3.98, + "learning_rate": 1.0576119838245843e-05, + "loss": 0.211, + "step": 9499 + }, + { + "epoch": 3.98, + "learning_rate": 1.057442723708339e-05, + "loss": 0.1991, + "step": 9500 + }, + { + "epoch": 3.98, + "learning_rate": 1.0572734619409535e-05, + "loss": 0.1995, + "step": 9501 + }, + { + "epoch": 3.98, + "learning_rate": 1.0571041985272934e-05, + "loss": 0.1729, + "step": 9502 + }, + { + "epoch": 3.98, + "learning_rate": 1.0569349334722236e-05, + "loss": 0.1999, + "step": 9503 + }, + { + "epoch": 3.98, + "learning_rate": 1.0567656667806097e-05, + "loss": 0.192, + "step": 9504 + }, + { + "epoch": 3.98, + "learning_rate": 1.0565963984573174e-05, + "loss": 0.1887, + "step": 9505 + }, + { + "epoch": 3.98, + "learning_rate": 1.0564271285072117e-05, + "loss": 0.1744, + "step": 9506 + }, + { + "epoch": 3.98, + "learning_rate": 1.0562578569351579e-05, + "loss": 0.1877, + "step": 9507 + }, + { + "epoch": 3.98, + "learning_rate": 1.056088583746022e-05, + "loss": 0.1796, + "step": 9508 + }, + { + "epoch": 3.98, + "learning_rate": 1.0559193089446698e-05, + "loss": 0.2141, + "step": 9509 + }, + { + "epoch": 3.98, + "learning_rate": 1.0557500325359665e-05, + "loss": 0.1828, + "step": 9510 + }, + { + "epoch": 3.98, + "learning_rate": 1.055580754524778e-05, + "loss": 0.1784, + "step": 9511 + }, + { + "epoch": 3.98, + "learning_rate": 1.05541147491597e-05, + "loss": 0.1886, + "step": 9512 + }, + { + "epoch": 3.98, + "learning_rate": 1.0552421937144083e-05, + "loss": 0.1875, + "step": 9513 + }, + { + "epoch": 3.98, + "learning_rate": 1.055072910924959e-05, + "loss": 0.1703, + "step": 9514 + }, + { + "epoch": 3.98, + "learning_rate": 1.0549036265524877e-05, + "loss": 0.1968, + "step": 9515 + }, + { + "epoch": 3.98, + "learning_rate": 1.0547343406018602e-05, + "loss": 0.1663, + "step": 9516 + }, + { + "epoch": 3.98, + "learning_rate": 1.0545650530779428e-05, + "loss": 0.1886, + "step": 9517 + }, + { + "epoch": 3.98, + "learning_rate": 1.0543957639856014e-05, + "loss": 0.1585, + "step": 9518 + }, + { + "epoch": 3.99, + "learning_rate": 1.0542264733297021e-05, + "loss": 0.1914, + "step": 9519 + }, + { + "epoch": 3.99, + "learning_rate": 1.0540571811151106e-05, + "loss": 0.163, + "step": 9520 + }, + { + "epoch": 3.99, + "learning_rate": 1.0538878873466936e-05, + "loss": 0.2158, + "step": 9521 + }, + { + "epoch": 3.99, + "learning_rate": 1.0537185920293171e-05, + "loss": 0.1514, + "step": 9522 + }, + { + "epoch": 3.99, + "learning_rate": 1.0535492951678471e-05, + "loss": 0.1868, + "step": 9523 + }, + { + "epoch": 3.99, + "learning_rate": 1.0533799967671503e-05, + "loss": 0.1927, + "step": 9524 + }, + { + "epoch": 3.99, + "learning_rate": 1.053210696832093e-05, + "loss": 0.1469, + "step": 9525 + }, + { + "epoch": 3.99, + "learning_rate": 1.0530413953675414e-05, + "loss": 0.1823, + "step": 9526 + }, + { + "epoch": 3.99, + "learning_rate": 1.0528720923783617e-05, + "loss": 0.2002, + "step": 9527 + }, + { + "epoch": 3.99, + "learning_rate": 1.0527027878694212e-05, + "loss": 0.2211, + "step": 9528 + }, + { + "epoch": 3.99, + "learning_rate": 1.0525334818455855e-05, + "loss": 0.1998, + "step": 9529 + }, + { + "epoch": 3.99, + "learning_rate": 1.0523641743117215e-05, + "loss": 0.1809, + "step": 9530 + }, + { + "epoch": 3.99, + "learning_rate": 1.0521948652726955e-05, + "loss": 0.1798, + "step": 9531 + }, + { + "epoch": 3.99, + "learning_rate": 1.0520255547333744e-05, + "loss": 0.1796, + "step": 9532 + }, + { + "epoch": 3.99, + "learning_rate": 1.0518562426986253e-05, + "loss": 0.202, + "step": 9533 + }, + { + "epoch": 3.99, + "learning_rate": 1.0516869291733142e-05, + "loss": 0.1884, + "step": 9534 + }, + { + "epoch": 3.99, + "learning_rate": 1.0515176141623085e-05, + "loss": 0.1454, + "step": 9535 + }, + { + "epoch": 3.99, + "learning_rate": 1.0513482976704746e-05, + "loss": 0.1957, + "step": 9536 + }, + { + "epoch": 3.99, + "learning_rate": 1.0511789797026792e-05, + "loss": 0.2049, + "step": 9537 + }, + { + "epoch": 3.99, + "learning_rate": 1.0510096602637896e-05, + "loss": 0.179, + "step": 9538 + }, + { + "epoch": 3.99, + "learning_rate": 1.0508403393586728e-05, + "loss": 0.1709, + "step": 9539 + }, + { + "epoch": 3.99, + "learning_rate": 1.0506710169921954e-05, + "loss": 0.1893, + "step": 9540 + }, + { + "epoch": 3.99, + "learning_rate": 1.0505016931692244e-05, + "loss": 0.1522, + "step": 9541 + }, + { + "epoch": 3.99, + "learning_rate": 1.0503323678946272e-05, + "loss": 0.1592, + "step": 9542 + }, + { + "epoch": 4.0, + "learning_rate": 1.0501630411732709e-05, + "loss": 0.1703, + "step": 9543 + }, + { + "epoch": 4.0, + "learning_rate": 1.0499937130100222e-05, + "loss": 0.1883, + "step": 9544 + }, + { + "epoch": 4.0, + "learning_rate": 1.0498243834097485e-05, + "loss": 0.1816, + "step": 9545 + }, + { + "epoch": 4.0, + "learning_rate": 1.0496550523773177e-05, + "loss": 0.1701, + "step": 9546 + }, + { + "epoch": 4.0, + "learning_rate": 1.0494857199175961e-05, + "loss": 0.174, + "step": 9547 + }, + { + "epoch": 4.0, + "learning_rate": 1.0493163860354512e-05, + "loss": 0.1885, + "step": 9548 + }, + { + "epoch": 4.0, + "learning_rate": 1.0491470507357508e-05, + "loss": 0.2248, + "step": 9549 + }, + { + "epoch": 4.0, + "learning_rate": 1.0489777140233619e-05, + "loss": 0.1783, + "step": 9550 + }, + { + "epoch": 4.0, + "learning_rate": 1.0488083759031524e-05, + "loss": 0.1625, + "step": 9551 + }, + { + "epoch": 4.0, + "learning_rate": 1.0486390363799891e-05, + "loss": 0.2177, + "step": 9552 + }, + { + "epoch": 4.0, + "learning_rate": 1.04846969545874e-05, + "loss": 0.1887, + "step": 9553 + }, + { + "epoch": 4.0, + "learning_rate": 1.0483003531442726e-05, + "loss": 0.1784, + "step": 9554 + }, + { + "epoch": 4.0, + "learning_rate": 1.048131009441454e-05, + "loss": 0.1487, + "step": 9555 + }, + { + "epoch": 4.0, + "learning_rate": 1.047961664355153e-05, + "loss": 0.1338, + "step": 9556 + }, + { + "epoch": 4.0, + "learning_rate": 1.0477923178902359e-05, + "loss": 0.1734, + "step": 9557 + }, + { + "epoch": 4.0, + "learning_rate": 1.047622970051571e-05, + "loss": 0.133, + "step": 9558 + }, + { + "epoch": 4.0, + "learning_rate": 1.0474536208440263e-05, + "loss": 0.1673, + "step": 9559 + }, + { + "epoch": 4.0, + "learning_rate": 1.0472842702724695e-05, + "loss": 0.1659, + "step": 9560 + }, + { + "epoch": 4.0, + "learning_rate": 1.047114918341768e-05, + "loss": 0.1201, + "step": 9561 + }, + { + "epoch": 4.0, + "learning_rate": 1.0469455650567903e-05, + "loss": 0.1705, + "step": 9562 + }, + { + "epoch": 4.0, + "learning_rate": 1.0467762104224039e-05, + "loss": 0.164, + "step": 9563 + }, + { + "epoch": 4.0, + "learning_rate": 1.0466068544434766e-05, + "loss": 0.1443, + "step": 9564 + }, + { + "epoch": 4.0, + "learning_rate": 1.046437497124877e-05, + "loss": 0.1925, + "step": 9565 + }, + { + "epoch": 4.01, + "learning_rate": 1.0462681384714723e-05, + "loss": 0.1494, + "step": 9566 + }, + { + "epoch": 4.01, + "learning_rate": 1.0460987784881313e-05, + "loss": 0.1724, + "step": 9567 + }, + { + "epoch": 4.01, + "learning_rate": 1.0459294171797218e-05, + "loss": 0.1702, + "step": 9568 + }, + { + "epoch": 4.01, + "learning_rate": 1.045760054551112e-05, + "loss": 0.1659, + "step": 9569 + }, + { + "epoch": 4.01, + "learning_rate": 1.04559069060717e-05, + "loss": 0.1584, + "step": 9570 + }, + { + "epoch": 4.01, + "learning_rate": 1.045421325352764e-05, + "loss": 0.1523, + "step": 9571 + }, + { + "epoch": 4.01, + "learning_rate": 1.0452519587927623e-05, + "loss": 0.1254, + "step": 9572 + }, + { + "epoch": 4.01, + "learning_rate": 1.0450825909320335e-05, + "loss": 0.136, + "step": 9573 + }, + { + "epoch": 4.01, + "learning_rate": 1.0449132217754455e-05, + "loss": 0.1503, + "step": 9574 + }, + { + "epoch": 4.01, + "learning_rate": 1.0447438513278664e-05, + "loss": 0.1935, + "step": 9575 + }, + { + "epoch": 4.01, + "learning_rate": 1.0445744795941652e-05, + "loss": 0.146, + "step": 9576 + }, + { + "epoch": 4.01, + "learning_rate": 1.0444051065792105e-05, + "loss": 0.192, + "step": 9577 + }, + { + "epoch": 4.01, + "learning_rate": 1.04423573228787e-05, + "loss": 0.0897, + "step": 9578 + }, + { + "epoch": 4.01, + "learning_rate": 1.0440663567250127e-05, + "loss": 0.1551, + "step": 9579 + }, + { + "epoch": 4.01, + "learning_rate": 1.0438969798955072e-05, + "loss": 0.1595, + "step": 9580 + }, + { + "epoch": 4.01, + "learning_rate": 1.0437276018042219e-05, + "loss": 0.1983, + "step": 9581 + }, + { + "epoch": 4.01, + "learning_rate": 1.0435582224560255e-05, + "loss": 0.1687, + "step": 9582 + }, + { + "epoch": 4.01, + "learning_rate": 1.0433888418557867e-05, + "loss": 0.1532, + "step": 9583 + }, + { + "epoch": 4.01, + "learning_rate": 1.043219460008374e-05, + "loss": 0.1476, + "step": 9584 + }, + { + "epoch": 4.01, + "learning_rate": 1.0430500769186565e-05, + "loss": 0.1043, + "step": 9585 + }, + { + "epoch": 4.01, + "learning_rate": 1.0428806925915026e-05, + "loss": 0.1767, + "step": 9586 + }, + { + "epoch": 4.01, + "learning_rate": 1.0427113070317813e-05, + "loss": 0.1923, + "step": 9587 + }, + { + "epoch": 4.01, + "learning_rate": 1.0425419202443614e-05, + "loss": 0.1675, + "step": 9588 + }, + { + "epoch": 4.01, + "learning_rate": 1.0423725322341115e-05, + "loss": 0.1778, + "step": 9589 + }, + { + "epoch": 4.02, + "learning_rate": 1.0422031430059011e-05, + "loss": 0.1971, + "step": 9590 + }, + { + "epoch": 4.02, + "learning_rate": 1.0420337525645984e-05, + "loss": 0.1652, + "step": 9591 + }, + { + "epoch": 4.02, + "learning_rate": 1.0418643609150732e-05, + "loss": 0.1324, + "step": 9592 + }, + { + "epoch": 4.02, + "learning_rate": 1.041694968062194e-05, + "loss": 0.1142, + "step": 9593 + }, + { + "epoch": 4.02, + "learning_rate": 1.0415255740108299e-05, + "loss": 0.1709, + "step": 9594 + }, + { + "epoch": 4.02, + "learning_rate": 1.0413561787658497e-05, + "loss": 0.1278, + "step": 9595 + }, + { + "epoch": 4.02, + "learning_rate": 1.0411867823321233e-05, + "loss": 0.1172, + "step": 9596 + }, + { + "epoch": 4.02, + "learning_rate": 1.0410173847145195e-05, + "loss": 0.1617, + "step": 9597 + }, + { + "epoch": 4.02, + "learning_rate": 1.040847985917907e-05, + "loss": 0.1151, + "step": 9598 + }, + { + "epoch": 4.02, + "learning_rate": 1.0406785859471554e-05, + "loss": 0.1156, + "step": 9599 + }, + { + "epoch": 4.02, + "learning_rate": 1.0405091848071342e-05, + "loss": 0.1262, + "step": 9600 + }, + { + "epoch": 4.02, + "learning_rate": 1.0403397825027122e-05, + "loss": 0.157, + "step": 9601 + }, + { + "epoch": 4.02, + "learning_rate": 1.0401703790387589e-05, + "loss": 0.175, + "step": 9602 + }, + { + "epoch": 4.02, + "learning_rate": 1.0400009744201438e-05, + "loss": 0.1851, + "step": 9603 + }, + { + "epoch": 4.02, + "learning_rate": 1.0398315686517362e-05, + "loss": 0.1766, + "step": 9604 + }, + { + "epoch": 4.02, + "learning_rate": 1.0396621617384055e-05, + "loss": 0.1459, + "step": 9605 + }, + { + "epoch": 4.02, + "learning_rate": 1.039492753685021e-05, + "loss": 0.1324, + "step": 9606 + }, + { + "epoch": 4.02, + "learning_rate": 1.0393233444964528e-05, + "loss": 0.1526, + "step": 9607 + }, + { + "epoch": 4.02, + "learning_rate": 1.0391539341775696e-05, + "loss": 0.1222, + "step": 9608 + }, + { + "epoch": 4.02, + "learning_rate": 1.0389845227332413e-05, + "loss": 0.1661, + "step": 9609 + }, + { + "epoch": 4.02, + "learning_rate": 1.0388151101683375e-05, + "loss": 0.1224, + "step": 9610 + }, + { + "epoch": 4.02, + "learning_rate": 1.0386456964877281e-05, + "loss": 0.154, + "step": 9611 + }, + { + "epoch": 4.02, + "learning_rate": 1.038476281696282e-05, + "loss": 0.1473, + "step": 9612 + }, + { + "epoch": 4.02, + "learning_rate": 1.0383068657988695e-05, + "loss": 0.0998, + "step": 9613 + }, + { + "epoch": 4.03, + "learning_rate": 1.0381374488003603e-05, + "loss": 0.1953, + "step": 9614 + }, + { + "epoch": 4.03, + "learning_rate": 1.0379680307056238e-05, + "loss": 0.1348, + "step": 9615 + }, + { + "epoch": 4.03, + "learning_rate": 1.03779861151953e-05, + "loss": 0.1649, + "step": 9616 + }, + { + "epoch": 4.03, + "learning_rate": 1.0376291912469488e-05, + "loss": 0.1258, + "step": 9617 + }, + { + "epoch": 4.03, + "learning_rate": 1.0374597698927496e-05, + "loss": 0.1482, + "step": 9618 + }, + { + "epoch": 4.03, + "learning_rate": 1.0372903474618027e-05, + "loss": 0.1671, + "step": 9619 + }, + { + "epoch": 4.03, + "learning_rate": 1.0371209239589779e-05, + "loss": 0.1486, + "step": 9620 + }, + { + "epoch": 4.03, + "learning_rate": 1.0369514993891451e-05, + "loss": 0.1308, + "step": 9621 + }, + { + "epoch": 4.03, + "learning_rate": 1.036782073757174e-05, + "loss": 0.1495, + "step": 9622 + }, + { + "epoch": 4.03, + "learning_rate": 1.036612647067935e-05, + "loss": 0.0969, + "step": 9623 + }, + { + "epoch": 4.03, + "learning_rate": 1.036443219326298e-05, + "loss": 0.184, + "step": 9624 + }, + { + "epoch": 4.03, + "learning_rate": 1.036273790537133e-05, + "loss": 0.127, + "step": 9625 + }, + { + "epoch": 4.03, + "learning_rate": 1.0361043607053101e-05, + "loss": 0.1336, + "step": 9626 + }, + { + "epoch": 4.03, + "learning_rate": 1.0359349298356993e-05, + "loss": 0.1505, + "step": 9627 + }, + { + "epoch": 4.03, + "learning_rate": 1.0357654979331708e-05, + "loss": 0.1709, + "step": 9628 + }, + { + "epoch": 4.03, + "learning_rate": 1.0355960650025946e-05, + "loss": 0.1573, + "step": 9629 + }, + { + "epoch": 4.03, + "learning_rate": 1.0354266310488416e-05, + "loss": 0.2003, + "step": 9630 + }, + { + "epoch": 4.03, + "learning_rate": 1.0352571960767814e-05, + "loss": 0.1725, + "step": 9631 + }, + { + "epoch": 4.03, + "learning_rate": 1.0350877600912844e-05, + "loss": 0.1442, + "step": 9632 + }, + { + "epoch": 4.03, + "learning_rate": 1.0349183230972204e-05, + "loss": 0.1851, + "step": 9633 + }, + { + "epoch": 4.03, + "learning_rate": 1.0347488850994608e-05, + "loss": 0.1321, + "step": 9634 + }, + { + "epoch": 4.03, + "learning_rate": 1.034579446102875e-05, + "loss": 0.1416, + "step": 9635 + }, + { + "epoch": 4.03, + "learning_rate": 1.0344100061123337e-05, + "loss": 0.1843, + "step": 9636 + }, + { + "epoch": 4.03, + "learning_rate": 1.0342405651327072e-05, + "loss": 0.1499, + "step": 9637 + }, + { + "epoch": 4.04, + "learning_rate": 1.0340711231688662e-05, + "loss": 0.1437, + "step": 9638 + }, + { + "epoch": 4.04, + "learning_rate": 1.033901680225681e-05, + "loss": 0.1823, + "step": 9639 + }, + { + "epoch": 4.04, + "learning_rate": 1.0337322363080219e-05, + "loss": 0.1039, + "step": 9640 + }, + { + "epoch": 4.04, + "learning_rate": 1.0335627914207602e-05, + "loss": 0.1494, + "step": 9641 + }, + { + "epoch": 4.04, + "learning_rate": 1.0333933455687655e-05, + "loss": 0.166, + "step": 9642 + }, + { + "epoch": 4.04, + "learning_rate": 1.0332238987569086e-05, + "loss": 0.1793, + "step": 9643 + }, + { + "epoch": 4.04, + "learning_rate": 1.0330544509900603e-05, + "loss": 0.1315, + "step": 9644 + }, + { + "epoch": 4.04, + "learning_rate": 1.032885002273091e-05, + "loss": 0.1267, + "step": 9645 + }, + { + "epoch": 4.04, + "learning_rate": 1.0327155526108716e-05, + "loss": 0.1621, + "step": 9646 + }, + { + "epoch": 4.04, + "learning_rate": 1.0325461020082726e-05, + "loss": 0.1321, + "step": 9647 + }, + { + "epoch": 4.04, + "learning_rate": 1.0323766504701649e-05, + "loss": 0.1195, + "step": 9648 + }, + { + "epoch": 4.04, + "learning_rate": 1.032207198001419e-05, + "loss": 0.1324, + "step": 9649 + }, + { + "epoch": 4.04, + "learning_rate": 1.0320377446069059e-05, + "loss": 0.1025, + "step": 9650 + }, + { + "epoch": 4.04, + "learning_rate": 1.0318682902914962e-05, + "loss": 0.1751, + "step": 9651 + }, + { + "epoch": 4.04, + "learning_rate": 1.0316988350600608e-05, + "loss": 0.1372, + "step": 9652 + }, + { + "epoch": 4.04, + "learning_rate": 1.0315293789174705e-05, + "loss": 0.2112, + "step": 9653 + }, + { + "epoch": 4.04, + "learning_rate": 1.0313599218685961e-05, + "loss": 0.1442, + "step": 9654 + }, + { + "epoch": 4.04, + "learning_rate": 1.0311904639183088e-05, + "loss": 0.0953, + "step": 9655 + }, + { + "epoch": 4.04, + "learning_rate": 1.031021005071479e-05, + "loss": 0.1931, + "step": 9656 + }, + { + "epoch": 4.04, + "learning_rate": 1.030851545332978e-05, + "loss": 0.1386, + "step": 9657 + }, + { + "epoch": 4.04, + "learning_rate": 1.0306820847076768e-05, + "loss": 0.1469, + "step": 9658 + }, + { + "epoch": 4.04, + "learning_rate": 1.0305126232004464e-05, + "loss": 0.1834, + "step": 9659 + }, + { + "epoch": 4.04, + "learning_rate": 1.0303431608161574e-05, + "loss": 0.1344, + "step": 9660 + }, + { + "epoch": 4.04, + "learning_rate": 1.0301736975596816e-05, + "loss": 0.1186, + "step": 9661 + }, + { + "epoch": 4.05, + "learning_rate": 1.0300042334358893e-05, + "loss": 0.1086, + "step": 9662 + }, + { + "epoch": 4.05, + "learning_rate": 1.029834768449652e-05, + "loss": 0.1411, + "step": 9663 + }, + { + "epoch": 4.05, + "learning_rate": 1.029665302605841e-05, + "loss": 0.1346, + "step": 9664 + }, + { + "epoch": 4.05, + "learning_rate": 1.029495835909327e-05, + "loss": 0.1431, + "step": 9665 + }, + { + "epoch": 4.05, + "learning_rate": 1.0293263683649813e-05, + "loss": 0.1703, + "step": 9666 + }, + { + "epoch": 4.05, + "learning_rate": 1.0291568999776751e-05, + "loss": 0.164, + "step": 9667 + }, + { + "epoch": 4.05, + "learning_rate": 1.0289874307522801e-05, + "loss": 0.131, + "step": 9668 + }, + { + "epoch": 4.05, + "learning_rate": 1.0288179606936666e-05, + "loss": 0.1263, + "step": 9669 + }, + { + "epoch": 4.05, + "learning_rate": 1.0286484898067067e-05, + "loss": 0.1508, + "step": 9670 + }, + { + "epoch": 4.05, + "learning_rate": 1.0284790180962712e-05, + "loss": 0.1621, + "step": 9671 + }, + { + "epoch": 4.05, + "learning_rate": 1.0283095455672319e-05, + "loss": 0.0984, + "step": 9672 + }, + { + "epoch": 4.05, + "learning_rate": 1.0281400722244594e-05, + "loss": 0.1535, + "step": 9673 + }, + { + "epoch": 4.05, + "learning_rate": 1.0279705980728258e-05, + "loss": 0.1426, + "step": 9674 + }, + { + "epoch": 4.05, + "learning_rate": 1.0278011231172024e-05, + "loss": 0.1577, + "step": 9675 + }, + { + "epoch": 4.05, + "learning_rate": 1.02763164736246e-05, + "loss": 0.1846, + "step": 9676 + }, + { + "epoch": 4.05, + "learning_rate": 1.0274621708134705e-05, + "loss": 0.1882, + "step": 9677 + }, + { + "epoch": 4.05, + "learning_rate": 1.0272926934751053e-05, + "loss": 0.0994, + "step": 9678 + }, + { + "epoch": 4.05, + "learning_rate": 1.0271232153522359e-05, + "loss": 0.193, + "step": 9679 + }, + { + "epoch": 4.05, + "learning_rate": 1.0269537364497337e-05, + "loss": 0.1684, + "step": 9680 + }, + { + "epoch": 4.05, + "learning_rate": 1.0267842567724702e-05, + "loss": 0.1234, + "step": 9681 + }, + { + "epoch": 4.05, + "learning_rate": 1.0266147763253171e-05, + "loss": 0.1486, + "step": 9682 + }, + { + "epoch": 4.05, + "learning_rate": 1.0264452951131459e-05, + "loss": 0.1551, + "step": 9683 + }, + { + "epoch": 4.05, + "learning_rate": 1.026275813140828e-05, + "loss": 0.1232, + "step": 9684 + }, + { + "epoch": 4.05, + "learning_rate": 1.0261063304132353e-05, + "loss": 0.1866, + "step": 9685 + }, + { + "epoch": 4.06, + "learning_rate": 1.025936846935239e-05, + "loss": 0.1786, + "step": 9686 + }, + { + "epoch": 4.06, + "learning_rate": 1.0257673627117115e-05, + "loss": 0.1304, + "step": 9687 + }, + { + "epoch": 4.06, + "learning_rate": 1.0255978777475238e-05, + "loss": 0.1339, + "step": 9688 + }, + { + "epoch": 4.06, + "learning_rate": 1.025428392047548e-05, + "loss": 0.1277, + "step": 9689 + }, + { + "epoch": 4.06, + "learning_rate": 1.0252589056166554e-05, + "loss": 0.1696, + "step": 9690 + }, + { + "epoch": 4.06, + "learning_rate": 1.025089418459718e-05, + "loss": 0.1633, + "step": 9691 + }, + { + "epoch": 4.06, + "learning_rate": 1.0249199305816078e-05, + "loss": 0.179, + "step": 9692 + }, + { + "epoch": 4.06, + "learning_rate": 1.024750441987196e-05, + "loss": 0.1495, + "step": 9693 + }, + { + "epoch": 4.06, + "learning_rate": 1.0245809526813547e-05, + "loss": 0.1376, + "step": 9694 + }, + { + "epoch": 4.06, + "learning_rate": 1.0244114626689558e-05, + "loss": 0.1755, + "step": 9695 + }, + { + "epoch": 4.06, + "learning_rate": 1.0242419719548711e-05, + "loss": 0.0943, + "step": 9696 + }, + { + "epoch": 4.06, + "learning_rate": 1.0240724805439724e-05, + "loss": 0.1556, + "step": 9697 + }, + { + "epoch": 4.06, + "learning_rate": 1.0239029884411318e-05, + "loss": 0.1504, + "step": 9698 + }, + { + "epoch": 4.06, + "learning_rate": 1.0237334956512209e-05, + "loss": 0.1953, + "step": 9699 + }, + { + "epoch": 4.06, + "learning_rate": 1.0235640021791115e-05, + "loss": 0.1541, + "step": 9700 + }, + { + "epoch": 4.06, + "learning_rate": 1.0233945080296758e-05, + "loss": 0.1627, + "step": 9701 + }, + { + "epoch": 4.06, + "learning_rate": 1.0232250132077861e-05, + "loss": 0.1356, + "step": 9702 + }, + { + "epoch": 4.06, + "learning_rate": 1.0230555177183136e-05, + "loss": 0.1696, + "step": 9703 + }, + { + "epoch": 4.06, + "learning_rate": 1.0228860215661309e-05, + "loss": 0.1229, + "step": 9704 + }, + { + "epoch": 4.06, + "learning_rate": 1.02271652475611e-05, + "loss": 0.1677, + "step": 9705 + }, + { + "epoch": 4.06, + "learning_rate": 1.0225470272931224e-05, + "loss": 0.0818, + "step": 9706 + }, + { + "epoch": 4.06, + "learning_rate": 1.0223775291820406e-05, + "loss": 0.163, + "step": 9707 + }, + { + "epoch": 4.06, + "learning_rate": 1.0222080304277366e-05, + "loss": 0.1689, + "step": 9708 + }, + { + "epoch": 4.06, + "learning_rate": 1.0220385310350829e-05, + "loss": 0.1454, + "step": 9709 + }, + { + "epoch": 4.07, + "learning_rate": 1.0218690310089508e-05, + "loss": 0.1118, + "step": 9710 + }, + { + "epoch": 4.07, + "learning_rate": 1.0216995303542126e-05, + "loss": 0.1471, + "step": 9711 + }, + { + "epoch": 4.07, + "learning_rate": 1.021530029075741e-05, + "loss": 0.1493, + "step": 9712 + }, + { + "epoch": 4.07, + "learning_rate": 1.0213605271784076e-05, + "loss": 0.156, + "step": 9713 + }, + { + "epoch": 4.07, + "learning_rate": 1.0211910246670848e-05, + "loss": 0.1622, + "step": 9714 + }, + { + "epoch": 4.07, + "learning_rate": 1.0210215215466447e-05, + "loss": 0.1617, + "step": 9715 + }, + { + "epoch": 4.07, + "learning_rate": 1.0208520178219602e-05, + "loss": 0.1603, + "step": 9716 + }, + { + "epoch": 4.07, + "learning_rate": 1.0206825134979024e-05, + "loss": 0.1312, + "step": 9717 + }, + { + "epoch": 4.07, + "learning_rate": 1.020513008579344e-05, + "loss": 0.1757, + "step": 9718 + }, + { + "epoch": 4.07, + "learning_rate": 1.0203435030711577e-05, + "loss": 0.1677, + "step": 9719 + }, + { + "epoch": 4.07, + "learning_rate": 1.0201739969782154e-05, + "loss": 0.1854, + "step": 9720 + }, + { + "epoch": 4.07, + "learning_rate": 1.0200044903053896e-05, + "loss": 0.1221, + "step": 9721 + }, + { + "epoch": 4.07, + "learning_rate": 1.0198349830575522e-05, + "loss": 0.1636, + "step": 9722 + }, + { + "epoch": 4.07, + "learning_rate": 1.0196654752395761e-05, + "loss": 0.1262, + "step": 9723 + }, + { + "epoch": 4.07, + "learning_rate": 1.0194959668563332e-05, + "loss": 0.1673, + "step": 9724 + }, + { + "epoch": 4.07, + "learning_rate": 1.0193264579126957e-05, + "loss": 0.1389, + "step": 9725 + }, + { + "epoch": 4.07, + "learning_rate": 1.0191569484135368e-05, + "loss": 0.1372, + "step": 9726 + }, + { + "epoch": 4.07, + "learning_rate": 1.0189874383637284e-05, + "loss": 0.149, + "step": 9727 + }, + { + "epoch": 4.07, + "learning_rate": 1.0188179277681426e-05, + "loss": 0.1391, + "step": 9728 + }, + { + "epoch": 4.07, + "learning_rate": 1.0186484166316526e-05, + "loss": 0.1838, + "step": 9729 + }, + { + "epoch": 4.07, + "learning_rate": 1.01847890495913e-05, + "loss": 0.1652, + "step": 9730 + }, + { + "epoch": 4.07, + "learning_rate": 1.0183093927554478e-05, + "loss": 0.1445, + "step": 9731 + }, + { + "epoch": 4.07, + "learning_rate": 1.0181398800254785e-05, + "loss": 0.1057, + "step": 9732 + }, + { + "epoch": 4.07, + "learning_rate": 1.0179703667740946e-05, + "loss": 0.1414, + "step": 9733 + }, + { + "epoch": 4.08, + "learning_rate": 1.017800853006168e-05, + "loss": 0.1351, + "step": 9734 + }, + { + "epoch": 4.08, + "learning_rate": 1.0176313387265717e-05, + "loss": 0.1709, + "step": 9735 + }, + { + "epoch": 4.08, + "learning_rate": 1.0174618239401786e-05, + "loss": 0.1435, + "step": 9736 + }, + { + "epoch": 4.08, + "learning_rate": 1.0172923086518607e-05, + "loss": 0.1688, + "step": 9737 + }, + { + "epoch": 4.08, + "learning_rate": 1.0171227928664905e-05, + "loss": 0.1496, + "step": 9738 + }, + { + "epoch": 4.08, + "learning_rate": 1.016953276588941e-05, + "loss": 0.1329, + "step": 9739 + }, + { + "epoch": 4.08, + "learning_rate": 1.0167837598240847e-05, + "loss": 0.1073, + "step": 9740 + }, + { + "epoch": 4.08, + "learning_rate": 1.016614242576794e-05, + "loss": 0.106, + "step": 9741 + }, + { + "epoch": 4.08, + "learning_rate": 1.0164447248519415e-05, + "loss": 0.1252, + "step": 9742 + }, + { + "epoch": 4.08, + "learning_rate": 1.0162752066544007e-05, + "loss": 0.1417, + "step": 9743 + }, + { + "epoch": 4.08, + "learning_rate": 1.0161056879890427e-05, + "loss": 0.1441, + "step": 9744 + }, + { + "epoch": 4.08, + "learning_rate": 1.0159361688607411e-05, + "loss": 0.1247, + "step": 9745 + }, + { + "epoch": 4.08, + "learning_rate": 1.0157666492743687e-05, + "loss": 0.2127, + "step": 9746 + }, + { + "epoch": 4.08, + "learning_rate": 1.0155971292347978e-05, + "loss": 0.1592, + "step": 9747 + }, + { + "epoch": 4.08, + "learning_rate": 1.0154276087469015e-05, + "loss": 0.0748, + "step": 9748 + }, + { + "epoch": 4.08, + "learning_rate": 1.015258087815552e-05, + "loss": 0.1466, + "step": 9749 + }, + { + "epoch": 4.08, + "learning_rate": 1.0150885664456224e-05, + "loss": 0.1454, + "step": 9750 + }, + { + "epoch": 4.08, + "learning_rate": 1.0149190446419854e-05, + "loss": 0.1237, + "step": 9751 + }, + { + "epoch": 4.08, + "learning_rate": 1.0147495224095136e-05, + "loss": 0.1437, + "step": 9752 + }, + { + "epoch": 4.08, + "learning_rate": 1.0145799997530801e-05, + "loss": 0.1919, + "step": 9753 + }, + { + "epoch": 4.08, + "learning_rate": 1.0144104766775574e-05, + "loss": 0.1484, + "step": 9754 + }, + { + "epoch": 4.08, + "learning_rate": 1.014240953187818e-05, + "loss": 0.0954, + "step": 9755 + }, + { + "epoch": 4.08, + "learning_rate": 1.0140714292887355e-05, + "loss": 0.1143, + "step": 9756 + }, + { + "epoch": 4.08, + "learning_rate": 1.0139019049851818e-05, + "loss": 0.1436, + "step": 9757 + }, + { + "epoch": 4.09, + "learning_rate": 1.0137323802820304e-05, + "loss": 0.1065, + "step": 9758 + }, + { + "epoch": 4.09, + "learning_rate": 1.0135628551841538e-05, + "loss": 0.1562, + "step": 9759 + }, + { + "epoch": 4.09, + "learning_rate": 1.0133933296964254e-05, + "loss": 0.1523, + "step": 9760 + }, + { + "epoch": 4.09, + "learning_rate": 1.0132238038237173e-05, + "loss": 0.1694, + "step": 9761 + }, + { + "epoch": 4.09, + "learning_rate": 1.0130542775709025e-05, + "loss": 0.1982, + "step": 9762 + }, + { + "epoch": 4.09, + "learning_rate": 1.0128847509428543e-05, + "loss": 0.1591, + "step": 9763 + }, + { + "epoch": 4.09, + "learning_rate": 1.0127152239444453e-05, + "loss": 0.1483, + "step": 9764 + }, + { + "epoch": 4.09, + "learning_rate": 1.0125456965805485e-05, + "loss": 0.1438, + "step": 9765 + }, + { + "epoch": 4.09, + "learning_rate": 1.012376168856037e-05, + "loss": 0.1265, + "step": 9766 + }, + { + "epoch": 4.09, + "learning_rate": 1.0122066407757833e-05, + "loss": 0.1673, + "step": 9767 + }, + { + "epoch": 4.09, + "learning_rate": 1.0120371123446603e-05, + "loss": 0.1653, + "step": 9768 + }, + { + "epoch": 4.09, + "learning_rate": 1.0118675835675416e-05, + "loss": 0.1294, + "step": 9769 + }, + { + "epoch": 4.09, + "learning_rate": 1.0116980544492996e-05, + "loss": 0.1593, + "step": 9770 + }, + { + "epoch": 4.09, + "learning_rate": 1.0115285249948073e-05, + "loss": 0.1596, + "step": 9771 + }, + { + "epoch": 4.09, + "learning_rate": 1.0113589952089377e-05, + "loss": 0.1486, + "step": 9772 + }, + { + "epoch": 4.09, + "learning_rate": 1.0111894650965644e-05, + "loss": 0.2042, + "step": 9773 + }, + { + "epoch": 4.09, + "learning_rate": 1.0110199346625593e-05, + "loss": 0.1378, + "step": 9774 + }, + { + "epoch": 4.09, + "learning_rate": 1.010850403911796e-05, + "loss": 0.1603, + "step": 9775 + }, + { + "epoch": 4.09, + "learning_rate": 1.0106808728491477e-05, + "loss": 0.1564, + "step": 9776 + }, + { + "epoch": 4.09, + "learning_rate": 1.0105113414794872e-05, + "loss": 0.1183, + "step": 9777 + }, + { + "epoch": 4.09, + "learning_rate": 1.0103418098076874e-05, + "loss": 0.1432, + "step": 9778 + }, + { + "epoch": 4.09, + "learning_rate": 1.0101722778386214e-05, + "loss": 0.1461, + "step": 9779 + }, + { + "epoch": 4.09, + "learning_rate": 1.0100027455771624e-05, + "loss": 0.1482, + "step": 9780 + }, + { + "epoch": 4.1, + "learning_rate": 1.0098332130281831e-05, + "loss": 0.156, + "step": 9781 + }, + { + "epoch": 4.1, + "learning_rate": 1.0096636801965571e-05, + "loss": 0.1696, + "step": 9782 + }, + { + "epoch": 4.1, + "learning_rate": 1.0094941470871571e-05, + "loss": 0.1552, + "step": 9783 + }, + { + "epoch": 4.1, + "learning_rate": 1.0093246137048563e-05, + "loss": 0.1347, + "step": 9784 + }, + { + "epoch": 4.1, + "learning_rate": 1.0091550800545279e-05, + "loss": 0.1917, + "step": 9785 + }, + { + "epoch": 4.1, + "learning_rate": 1.0089855461410446e-05, + "loss": 0.135, + "step": 9786 + }, + { + "epoch": 4.1, + "learning_rate": 1.0088160119692798e-05, + "loss": 0.1596, + "step": 9787 + }, + { + "epoch": 4.1, + "learning_rate": 1.0086464775441064e-05, + "loss": 0.126, + "step": 9788 + }, + { + "epoch": 4.1, + "learning_rate": 1.008476942870398e-05, + "loss": 0.1321, + "step": 9789 + }, + { + "epoch": 4.1, + "learning_rate": 1.0083074079530272e-05, + "loss": 0.1722, + "step": 9790 + }, + { + "epoch": 4.1, + "learning_rate": 1.0081378727968676e-05, + "loss": 0.1506, + "step": 9791 + }, + { + "epoch": 4.1, + "learning_rate": 1.007968337406792e-05, + "loss": 0.1526, + "step": 9792 + }, + { + "epoch": 4.1, + "learning_rate": 1.0077988017876732e-05, + "loss": 0.147, + "step": 9793 + }, + { + "epoch": 4.1, + "learning_rate": 1.0076292659443852e-05, + "loss": 0.1657, + "step": 9794 + }, + { + "epoch": 4.1, + "learning_rate": 1.0074597298818005e-05, + "loss": 0.1669, + "step": 9795 + }, + { + "epoch": 4.1, + "learning_rate": 1.0072901936047923e-05, + "loss": 0.1401, + "step": 9796 + }, + { + "epoch": 4.1, + "learning_rate": 1.0071206571182344e-05, + "loss": 0.1403, + "step": 9797 + }, + { + "epoch": 4.1, + "learning_rate": 1.0069511204269992e-05, + "loss": 0.1263, + "step": 9798 + }, + { + "epoch": 4.1, + "learning_rate": 1.0067815835359605e-05, + "loss": 0.1474, + "step": 9799 + }, + { + "epoch": 4.1, + "learning_rate": 1.0066120464499912e-05, + "loss": 0.1316, + "step": 9800 + }, + { + "epoch": 4.1, + "learning_rate": 1.0064425091739643e-05, + "loss": 0.156, + "step": 9801 + }, + { + "epoch": 4.1, + "learning_rate": 1.0062729717127531e-05, + "loss": 0.1534, + "step": 9802 + }, + { + "epoch": 4.1, + "learning_rate": 1.006103434071231e-05, + "loss": 0.1773, + "step": 9803 + }, + { + "epoch": 4.1, + "learning_rate": 1.0059338962542713e-05, + "loss": 0.1738, + "step": 9804 + }, + { + "epoch": 4.11, + "learning_rate": 1.0057643582667467e-05, + "loss": 0.1704, + "step": 9805 + }, + { + "epoch": 4.11, + "learning_rate": 1.0055948201135308e-05, + "loss": 0.1433, + "step": 9806 + }, + { + "epoch": 4.11, + "learning_rate": 1.0054252817994969e-05, + "loss": 0.1176, + "step": 9807 + }, + { + "epoch": 4.11, + "learning_rate": 1.0052557433295181e-05, + "loss": 0.1655, + "step": 9808 + }, + { + "epoch": 4.11, + "learning_rate": 1.0050862047084673e-05, + "loss": 0.1327, + "step": 9809 + }, + { + "epoch": 4.11, + "learning_rate": 1.0049166659412185e-05, + "loss": 0.1708, + "step": 9810 + }, + { + "epoch": 4.11, + "learning_rate": 1.0047471270326446e-05, + "loss": 0.1619, + "step": 9811 + }, + { + "epoch": 4.11, + "learning_rate": 1.0045775879876183e-05, + "loss": 0.1612, + "step": 9812 + }, + { + "epoch": 4.11, + "learning_rate": 1.0044080488110133e-05, + "loss": 0.1654, + "step": 9813 + }, + { + "epoch": 4.11, + "learning_rate": 1.0042385095077032e-05, + "loss": 0.1458, + "step": 9814 + }, + { + "epoch": 4.11, + "learning_rate": 1.0040689700825606e-05, + "loss": 0.1526, + "step": 9815 + }, + { + "epoch": 4.11, + "learning_rate": 1.003899430540459e-05, + "loss": 0.18, + "step": 9816 + }, + { + "epoch": 4.11, + "learning_rate": 1.0037298908862721e-05, + "loss": 0.1752, + "step": 9817 + }, + { + "epoch": 4.11, + "learning_rate": 1.0035603511248723e-05, + "loss": 0.1416, + "step": 9818 + }, + { + "epoch": 4.11, + "learning_rate": 1.0033908112611337e-05, + "loss": 0.141, + "step": 9819 + }, + { + "epoch": 4.11, + "learning_rate": 1.0032212712999292e-05, + "loss": 0.1333, + "step": 9820 + }, + { + "epoch": 4.11, + "learning_rate": 1.0030517312461323e-05, + "loss": 0.1397, + "step": 9821 + }, + { + "epoch": 4.11, + "learning_rate": 1.0028821911046158e-05, + "loss": 0.1191, + "step": 9822 + }, + { + "epoch": 4.11, + "learning_rate": 1.0027126508802534e-05, + "loss": 0.1196, + "step": 9823 + }, + { + "epoch": 4.11, + "learning_rate": 1.0025431105779184e-05, + "loss": 0.1096, + "step": 9824 + }, + { + "epoch": 4.11, + "learning_rate": 1.0023735702024838e-05, + "loss": 0.1517, + "step": 9825 + }, + { + "epoch": 4.11, + "learning_rate": 1.0022040297588229e-05, + "loss": 0.147, + "step": 9826 + }, + { + "epoch": 4.11, + "learning_rate": 1.0020344892518093e-05, + "loss": 0.1235, + "step": 9827 + }, + { + "epoch": 4.11, + "learning_rate": 1.001864948686316e-05, + "loss": 0.1635, + "step": 9828 + }, + { + "epoch": 4.12, + "learning_rate": 1.0016954080672163e-05, + "loss": 0.1218, + "step": 9829 + }, + { + "epoch": 4.12, + "learning_rate": 1.0015258673993837e-05, + "loss": 0.1867, + "step": 9830 + }, + { + "epoch": 4.12, + "learning_rate": 1.0013563266876916e-05, + "loss": 0.1583, + "step": 9831 + }, + { + "epoch": 4.12, + "learning_rate": 1.001186785937013e-05, + "loss": 0.1475, + "step": 9832 + }, + { + "epoch": 4.12, + "learning_rate": 1.0010172451522211e-05, + "loss": 0.1283, + "step": 9833 + }, + { + "epoch": 4.12, + "learning_rate": 1.0008477043381896e-05, + "loss": 0.1581, + "step": 9834 + }, + { + "epoch": 4.12, + "learning_rate": 1.0006781634997917e-05, + "loss": 0.1534, + "step": 9835 + }, + { + "epoch": 4.12, + "learning_rate": 1.0005086226419003e-05, + "loss": 0.127, + "step": 9836 + }, + { + "epoch": 4.12, + "learning_rate": 1.000339081769389e-05, + "loss": 0.1332, + "step": 9837 + }, + { + "epoch": 4.12, + "learning_rate": 1.0001695408871312e-05, + "loss": 0.142, + "step": 9838 + }, + { + "epoch": 4.12, + "learning_rate": 1e-05, + "loss": 0.1187, + "step": 9839 + }, + { + "epoch": 4.12, + "learning_rate": 9.99830459112869e-06, + "loss": 0.1665, + "step": 9840 + }, + { + "epoch": 4.12, + "learning_rate": 9.996609182306113e-06, + "loss": 0.1124, + "step": 9841 + }, + { + "epoch": 4.12, + "learning_rate": 9.994913773581e-06, + "loss": 0.1864, + "step": 9842 + }, + { + "epoch": 4.12, + "learning_rate": 9.993218365002086e-06, + "loss": 0.153, + "step": 9843 + }, + { + "epoch": 4.12, + "learning_rate": 9.991522956618107e-06, + "loss": 0.1748, + "step": 9844 + }, + { + "epoch": 4.12, + "learning_rate": 9.98982754847779e-06, + "loss": 0.1491, + "step": 9845 + }, + { + "epoch": 4.12, + "learning_rate": 9.988132140629871e-06, + "loss": 0.1691, + "step": 9846 + }, + { + "epoch": 4.12, + "learning_rate": 9.986436733123086e-06, + "loss": 0.1863, + "step": 9847 + }, + { + "epoch": 4.12, + "learning_rate": 9.984741326006163e-06, + "loss": 0.1493, + "step": 9848 + }, + { + "epoch": 4.12, + "learning_rate": 9.983045919327837e-06, + "loss": 0.1843, + "step": 9849 + }, + { + "epoch": 4.12, + "learning_rate": 9.981350513136841e-06, + "loss": 0.1709, + "step": 9850 + }, + { + "epoch": 4.12, + "learning_rate": 9.979655107481912e-06, + "loss": 0.1936, + "step": 9851 + }, + { + "epoch": 4.12, + "learning_rate": 9.977959702411776e-06, + "loss": 0.1036, + "step": 9852 + }, + { + "epoch": 4.13, + "learning_rate": 9.976264297975167e-06, + "loss": 0.1825, + "step": 9853 + }, + { + "epoch": 4.13, + "learning_rate": 9.974568894220821e-06, + "loss": 0.1521, + "step": 9854 + }, + { + "epoch": 4.13, + "learning_rate": 9.97287349119747e-06, + "loss": 0.1488, + "step": 9855 + }, + { + "epoch": 4.13, + "learning_rate": 9.971178088953845e-06, + "loss": 0.1464, + "step": 9856 + }, + { + "epoch": 4.13, + "learning_rate": 9.96948268753868e-06, + "loss": 0.1594, + "step": 9857 + }, + { + "epoch": 4.13, + "learning_rate": 9.967787287000711e-06, + "loss": 0.1306, + "step": 9858 + }, + { + "epoch": 4.13, + "learning_rate": 9.966091887388665e-06, + "loss": 0.1487, + "step": 9859 + }, + { + "epoch": 4.13, + "learning_rate": 9.964396488751278e-06, + "loss": 0.1637, + "step": 9860 + }, + { + "epoch": 4.13, + "learning_rate": 9.962701091137282e-06, + "loss": 0.1266, + "step": 9861 + }, + { + "epoch": 4.13, + "learning_rate": 9.961005694595413e-06, + "loss": 0.1867, + "step": 9862 + }, + { + "epoch": 4.13, + "learning_rate": 9.959310299174397e-06, + "loss": 0.1562, + "step": 9863 + }, + { + "epoch": 4.13, + "learning_rate": 9.957614904922971e-06, + "loss": 0.1554, + "step": 9864 + }, + { + "epoch": 4.13, + "learning_rate": 9.95591951188987e-06, + "loss": 0.1215, + "step": 9865 + }, + { + "epoch": 4.13, + "learning_rate": 9.954224120123822e-06, + "loss": 0.1551, + "step": 9866 + }, + { + "epoch": 4.13, + "learning_rate": 9.95252872967356e-06, + "loss": 0.1345, + "step": 9867 + }, + { + "epoch": 4.13, + "learning_rate": 9.950833340587818e-06, + "loss": 0.2001, + "step": 9868 + }, + { + "epoch": 4.13, + "learning_rate": 9.949137952915327e-06, + "loss": 0.1969, + "step": 9869 + }, + { + "epoch": 4.13, + "learning_rate": 9.94744256670482e-06, + "loss": 0.1802, + "step": 9870 + }, + { + "epoch": 4.13, + "learning_rate": 9.945747182005033e-06, + "loss": 0.1551, + "step": 9871 + }, + { + "epoch": 4.13, + "learning_rate": 9.944051798864692e-06, + "loss": 0.1237, + "step": 9872 + }, + { + "epoch": 4.13, + "learning_rate": 9.942356417332538e-06, + "loss": 0.1763, + "step": 9873 + }, + { + "epoch": 4.13, + "learning_rate": 9.940661037457294e-06, + "loss": 0.1686, + "step": 9874 + }, + { + "epoch": 4.13, + "learning_rate": 9.938965659287693e-06, + "loss": 0.1286, + "step": 9875 + }, + { + "epoch": 4.13, + "learning_rate": 9.937270282872474e-06, + "loss": 0.1735, + "step": 9876 + }, + { + "epoch": 4.14, + "learning_rate": 9.935574908260362e-06, + "loss": 0.128, + "step": 9877 + }, + { + "epoch": 4.14, + "learning_rate": 9.933879535500093e-06, + "loss": 0.1503, + "step": 9878 + }, + { + "epoch": 4.14, + "learning_rate": 9.932184164640399e-06, + "loss": 0.1645, + "step": 9879 + }, + { + "epoch": 4.14, + "learning_rate": 9.93048879573001e-06, + "loss": 0.1779, + "step": 9880 + }, + { + "epoch": 4.14, + "learning_rate": 9.928793428817658e-06, + "loss": 0.1424, + "step": 9881 + }, + { + "epoch": 4.14, + "learning_rate": 9.927098063952078e-06, + "loss": 0.1156, + "step": 9882 + }, + { + "epoch": 4.14, + "learning_rate": 9.925402701181999e-06, + "loss": 0.1213, + "step": 9883 + }, + { + "epoch": 4.14, + "learning_rate": 9.923707340556151e-06, + "loss": 0.1282, + "step": 9884 + }, + { + "epoch": 4.14, + "learning_rate": 9.92201198212327e-06, + "loss": 0.1577, + "step": 9885 + }, + { + "epoch": 4.14, + "learning_rate": 9.920316625932085e-06, + "loss": 0.1339, + "step": 9886 + }, + { + "epoch": 4.14, + "learning_rate": 9.918621272031327e-06, + "loss": 0.1632, + "step": 9887 + }, + { + "epoch": 4.14, + "learning_rate": 9.91692592046973e-06, + "loss": 0.1693, + "step": 9888 + }, + { + "epoch": 4.14, + "learning_rate": 9.915230571296021e-06, + "loss": 0.1508, + "step": 9889 + }, + { + "epoch": 4.14, + "learning_rate": 9.913535224558936e-06, + "loss": 0.1149, + "step": 9890 + }, + { + "epoch": 4.14, + "learning_rate": 9.911839880307202e-06, + "loss": 0.1464, + "step": 9891 + }, + { + "epoch": 4.14, + "learning_rate": 9.910144538589556e-06, + "loss": 0.1719, + "step": 9892 + }, + { + "epoch": 4.14, + "learning_rate": 9.908449199454723e-06, + "loss": 0.1044, + "step": 9893 + }, + { + "epoch": 4.14, + "learning_rate": 9.906753862951437e-06, + "loss": 0.1697, + "step": 9894 + }, + { + "epoch": 4.14, + "learning_rate": 9.90505852912843e-06, + "loss": 0.173, + "step": 9895 + }, + { + "epoch": 4.14, + "learning_rate": 9.903363198034434e-06, + "loss": 0.158, + "step": 9896 + }, + { + "epoch": 4.14, + "learning_rate": 9.901667869718172e-06, + "loss": 0.1206, + "step": 9897 + }, + { + "epoch": 4.14, + "learning_rate": 9.89997254422838e-06, + "loss": 0.1652, + "step": 9898 + }, + { + "epoch": 4.14, + "learning_rate": 9.89827722161379e-06, + "loss": 0.1206, + "step": 9899 + }, + { + "epoch": 4.14, + "learning_rate": 9.896581901923131e-06, + "loss": 0.1813, + "step": 9900 + }, + { + "epoch": 4.15, + "learning_rate": 9.894886585205132e-06, + "loss": 0.1526, + "step": 9901 + }, + { + "epoch": 4.15, + "learning_rate": 9.893191271508526e-06, + "loss": 0.1389, + "step": 9902 + }, + { + "epoch": 4.15, + "learning_rate": 9.891495960882042e-06, + "loss": 0.1889, + "step": 9903 + }, + { + "epoch": 4.15, + "learning_rate": 9.88980065337441e-06, + "loss": 0.1551, + "step": 9904 + }, + { + "epoch": 4.15, + "learning_rate": 9.88810534903436e-06, + "loss": 0.1207, + "step": 9905 + }, + { + "epoch": 4.15, + "learning_rate": 9.886410047910624e-06, + "loss": 0.1304, + "step": 9906 + }, + { + "epoch": 4.15, + "learning_rate": 9.884714750051929e-06, + "loss": 0.1537, + "step": 9907 + }, + { + "epoch": 4.15, + "learning_rate": 9.883019455507008e-06, + "loss": 0.1206, + "step": 9908 + }, + { + "epoch": 4.15, + "learning_rate": 9.881324164324587e-06, + "loss": 0.14, + "step": 9909 + }, + { + "epoch": 4.15, + "learning_rate": 9.879628876553398e-06, + "loss": 0.1357, + "step": 9910 + }, + { + "epoch": 4.15, + "learning_rate": 9.87793359224217e-06, + "loss": 0.1855, + "step": 9911 + }, + { + "epoch": 4.15, + "learning_rate": 9.876238311439635e-06, + "loss": 0.1721, + "step": 9912 + }, + { + "epoch": 4.15, + "learning_rate": 9.874543034194516e-06, + "loss": 0.172, + "step": 9913 + }, + { + "epoch": 4.15, + "learning_rate": 9.872847760555547e-06, + "loss": 0.1699, + "step": 9914 + }, + { + "epoch": 4.15, + "learning_rate": 9.871152490571459e-06, + "loss": 0.1347, + "step": 9915 + }, + { + "epoch": 4.15, + "learning_rate": 9.869457224290975e-06, + "loss": 0.1038, + "step": 9916 + }, + { + "epoch": 4.15, + "learning_rate": 9.867761961762829e-06, + "loss": 0.1226, + "step": 9917 + }, + { + "epoch": 4.15, + "learning_rate": 9.86606670303575e-06, + "loss": 0.1654, + "step": 9918 + }, + { + "epoch": 4.15, + "learning_rate": 9.864371448158463e-06, + "loss": 0.1471, + "step": 9919 + }, + { + "epoch": 4.15, + "learning_rate": 9.8626761971797e-06, + "loss": 0.1598, + "step": 9920 + }, + { + "epoch": 4.15, + "learning_rate": 9.860980950148185e-06, + "loss": 0.185, + "step": 9921 + }, + { + "epoch": 4.15, + "learning_rate": 9.859285707112648e-06, + "loss": 0.1481, + "step": 9922 + }, + { + "epoch": 4.15, + "learning_rate": 9.857590468121823e-06, + "loss": 0.1243, + "step": 9923 + }, + { + "epoch": 4.15, + "learning_rate": 9.855895233224431e-06, + "loss": 0.1544, + "step": 9924 + }, + { + "epoch": 4.16, + "learning_rate": 9.854200002469202e-06, + "loss": 0.094, + "step": 9925 + }, + { + "epoch": 4.16, + "learning_rate": 9.852504775904867e-06, + "loss": 0.1216, + "step": 9926 + }, + { + "epoch": 4.16, + "learning_rate": 9.850809553580148e-06, + "loss": 0.1627, + "step": 9927 + }, + { + "epoch": 4.16, + "learning_rate": 9.849114335543778e-06, + "loss": 0.1341, + "step": 9928 + }, + { + "epoch": 4.16, + "learning_rate": 9.847419121844484e-06, + "loss": 0.1723, + "step": 9929 + }, + { + "epoch": 4.16, + "learning_rate": 9.845723912530989e-06, + "loss": 0.1964, + "step": 9930 + }, + { + "epoch": 4.16, + "learning_rate": 9.844028707652023e-06, + "loss": 0.1807, + "step": 9931 + }, + { + "epoch": 4.16, + "learning_rate": 9.842333507256315e-06, + "loss": 0.19, + "step": 9932 + }, + { + "epoch": 4.16, + "learning_rate": 9.84063831139259e-06, + "loss": 0.1478, + "step": 9933 + }, + { + "epoch": 4.16, + "learning_rate": 9.838943120109576e-06, + "loss": 0.1816, + "step": 9934 + }, + { + "epoch": 4.16, + "learning_rate": 9.837247933455998e-06, + "loss": 0.1445, + "step": 9935 + }, + { + "epoch": 4.16, + "learning_rate": 9.835552751480587e-06, + "loss": 0.1875, + "step": 9936 + }, + { + "epoch": 4.16, + "learning_rate": 9.833857574232061e-06, + "loss": 0.1343, + "step": 9937 + }, + { + "epoch": 4.16, + "learning_rate": 9.832162401759155e-06, + "loss": 0.1191, + "step": 9938 + }, + { + "epoch": 4.16, + "learning_rate": 9.83046723411059e-06, + "loss": 0.1396, + "step": 9939 + }, + { + "epoch": 4.16, + "learning_rate": 9.828772071335095e-06, + "loss": 0.148, + "step": 9940 + }, + { + "epoch": 4.16, + "learning_rate": 9.8270769134814e-06, + "loss": 0.1541, + "step": 9941 + }, + { + "epoch": 4.16, + "learning_rate": 9.82538176059822e-06, + "loss": 0.1429, + "step": 9942 + }, + { + "epoch": 4.16, + "learning_rate": 9.823686612734285e-06, + "loss": 0.1397, + "step": 9943 + }, + { + "epoch": 4.16, + "learning_rate": 9.821991469938325e-06, + "loss": 0.1751, + "step": 9944 + }, + { + "epoch": 4.16, + "learning_rate": 9.82029633225906e-06, + "loss": 0.1865, + "step": 9945 + }, + { + "epoch": 4.16, + "learning_rate": 9.818601199745218e-06, + "loss": 0.101, + "step": 9946 + }, + { + "epoch": 4.16, + "learning_rate": 9.816906072445526e-06, + "loss": 0.1478, + "step": 9947 + }, + { + "epoch": 4.16, + "learning_rate": 9.815210950408703e-06, + "loss": 0.1287, + "step": 9948 + }, + { + "epoch": 4.17, + "learning_rate": 9.813515833683477e-06, + "loss": 0.168, + "step": 9949 + }, + { + "epoch": 4.17, + "learning_rate": 9.811820722318577e-06, + "loss": 0.1253, + "step": 9950 + }, + { + "epoch": 4.17, + "learning_rate": 9.81012561636272e-06, + "loss": 0.1594, + "step": 9951 + }, + { + "epoch": 4.17, + "learning_rate": 9.808430515864633e-06, + "loss": 0.1339, + "step": 9952 + }, + { + "epoch": 4.17, + "learning_rate": 9.806735420873044e-06, + "loss": 0.1805, + "step": 9953 + }, + { + "epoch": 4.17, + "learning_rate": 9.805040331436672e-06, + "loss": 0.1188, + "step": 9954 + }, + { + "epoch": 4.17, + "learning_rate": 9.803345247604242e-06, + "loss": 0.1549, + "step": 9955 + }, + { + "epoch": 4.17, + "learning_rate": 9.80165016942448e-06, + "loss": 0.1802, + "step": 9956 + }, + { + "epoch": 4.17, + "learning_rate": 9.799955096946107e-06, + "loss": 0.1507, + "step": 9957 + }, + { + "epoch": 4.17, + "learning_rate": 9.798260030217846e-06, + "loss": 0.112, + "step": 9958 + }, + { + "epoch": 4.17, + "learning_rate": 9.796564969288425e-06, + "loss": 0.1529, + "step": 9959 + }, + { + "epoch": 4.17, + "learning_rate": 9.79486991420656e-06, + "loss": 0.1983, + "step": 9960 + }, + { + "epoch": 4.17, + "learning_rate": 9.793174865020978e-06, + "loss": 0.124, + "step": 9961 + }, + { + "epoch": 4.17, + "learning_rate": 9.7914798217804e-06, + "loss": 0.1237, + "step": 9962 + }, + { + "epoch": 4.17, + "learning_rate": 9.789784784533553e-06, + "loss": 0.1132, + "step": 9963 + }, + { + "epoch": 4.17, + "learning_rate": 9.788089753329157e-06, + "loss": 0.0795, + "step": 9964 + }, + { + "epoch": 4.17, + "learning_rate": 9.78639472821593e-06, + "loss": 0.1575, + "step": 9965 + }, + { + "epoch": 4.17, + "learning_rate": 9.784699709242595e-06, + "loss": 0.1362, + "step": 9966 + }, + { + "epoch": 4.17, + "learning_rate": 9.78300469645788e-06, + "loss": 0.1231, + "step": 9967 + }, + { + "epoch": 4.17, + "learning_rate": 9.7813096899105e-06, + "loss": 0.1493, + "step": 9968 + }, + { + "epoch": 4.17, + "learning_rate": 9.779614689649176e-06, + "loss": 0.1198, + "step": 9969 + }, + { + "epoch": 4.17, + "learning_rate": 9.777919695722638e-06, + "loss": 0.1976, + "step": 9970 + }, + { + "epoch": 4.17, + "learning_rate": 9.776224708179597e-06, + "loss": 0.1473, + "step": 9971 + }, + { + "epoch": 4.18, + "learning_rate": 9.774529727068778e-06, + "loss": 0.1737, + "step": 9972 + }, + { + "epoch": 4.18, + "learning_rate": 9.772834752438903e-06, + "loss": 0.1593, + "step": 9973 + }, + { + "epoch": 4.18, + "learning_rate": 9.771139784338693e-06, + "loss": 0.1489, + "step": 9974 + }, + { + "epoch": 4.18, + "learning_rate": 9.769444822816865e-06, + "loss": 0.174, + "step": 9975 + }, + { + "epoch": 4.18, + "learning_rate": 9.76774986792214e-06, + "loss": 0.1863, + "step": 9976 + }, + { + "epoch": 4.18, + "learning_rate": 9.766054919703243e-06, + "loss": 0.1683, + "step": 9977 + }, + { + "epoch": 4.18, + "learning_rate": 9.764359978208886e-06, + "loss": 0.1814, + "step": 9978 + }, + { + "epoch": 4.18, + "learning_rate": 9.762665043487793e-06, + "loss": 0.1657, + "step": 9979 + }, + { + "epoch": 4.18, + "learning_rate": 9.760970115588686e-06, + "loss": 0.1244, + "step": 9980 + }, + { + "epoch": 4.18, + "learning_rate": 9.759275194560278e-06, + "loss": 0.1577, + "step": 9981 + }, + { + "epoch": 4.18, + "learning_rate": 9.757580280451289e-06, + "loss": 0.1613, + "step": 9982 + }, + { + "epoch": 4.18, + "learning_rate": 9.755885373310444e-06, + "loss": 0.1368, + "step": 9983 + }, + { + "epoch": 4.18, + "learning_rate": 9.754190473186455e-06, + "loss": 0.1572, + "step": 9984 + }, + { + "epoch": 4.18, + "learning_rate": 9.752495580128042e-06, + "loss": 0.1111, + "step": 9985 + }, + { + "epoch": 4.18, + "learning_rate": 9.750800694183926e-06, + "loss": 0.1778, + "step": 9986 + }, + { + "epoch": 4.18, + "learning_rate": 9.749105815402823e-06, + "loss": 0.1719, + "step": 9987 + }, + { + "epoch": 4.18, + "learning_rate": 9.747410943833451e-06, + "loss": 0.1589, + "step": 9988 + }, + { + "epoch": 4.18, + "learning_rate": 9.745716079524526e-06, + "loss": 0.1702, + "step": 9989 + }, + { + "epoch": 4.18, + "learning_rate": 9.744021222524764e-06, + "loss": 0.1895, + "step": 9990 + }, + { + "epoch": 4.18, + "learning_rate": 9.74232637288289e-06, + "loss": 0.143, + "step": 9991 + }, + { + "epoch": 4.18, + "learning_rate": 9.740631530647611e-06, + "loss": 0.0912, + "step": 9992 + }, + { + "epoch": 4.18, + "learning_rate": 9.73893669586765e-06, + "loss": 0.1575, + "step": 9993 + }, + { + "epoch": 4.18, + "learning_rate": 9.737241868591724e-06, + "loss": 0.1241, + "step": 9994 + }, + { + "epoch": 4.18, + "learning_rate": 9.735547048868545e-06, + "loss": 0.1368, + "step": 9995 + }, + { + "epoch": 4.19, + "learning_rate": 9.733852236746832e-06, + "loss": 0.1415, + "step": 9996 + }, + { + "epoch": 4.19, + "learning_rate": 9.732157432275301e-06, + "loss": 0.1183, + "step": 9997 + }, + { + "epoch": 4.19, + "learning_rate": 9.730462635502666e-06, + "loss": 0.1642, + "step": 9998 + }, + { + "epoch": 4.19, + "learning_rate": 9.728767846477643e-06, + "loss": 0.153, + "step": 9999 + }, + { + "epoch": 4.19, + "learning_rate": 9.72707306524895e-06, + "loss": 0.1753, + "step": 10000 + }, + { + "epoch": 4.19, + "learning_rate": 9.725378291865297e-06, + "loss": 0.1548, + "step": 10001 + }, + { + "epoch": 4.19, + "learning_rate": 9.723683526375403e-06, + "loss": 0.1527, + "step": 10002 + }, + { + "epoch": 4.19, + "learning_rate": 9.721988768827977e-06, + "loss": 0.1509, + "step": 10003 + }, + { + "epoch": 4.19, + "learning_rate": 9.720294019271744e-06, + "loss": 0.1601, + "step": 10004 + }, + { + "epoch": 4.19, + "learning_rate": 9.718599277755406e-06, + "loss": 0.1228, + "step": 10005 + }, + { + "epoch": 4.19, + "learning_rate": 9.716904544327683e-06, + "loss": 0.1562, + "step": 10006 + }, + { + "epoch": 4.19, + "learning_rate": 9.71520981903729e-06, + "loss": 0.1742, + "step": 10007 + }, + { + "epoch": 4.19, + "learning_rate": 9.713515101932935e-06, + "loss": 0.1533, + "step": 10008 + }, + { + "epoch": 4.19, + "learning_rate": 9.711820393063337e-06, + "loss": 0.1423, + "step": 10009 + }, + { + "epoch": 4.19, + "learning_rate": 9.710125692477204e-06, + "loss": 0.1037, + "step": 10010 + }, + { + "epoch": 4.19, + "learning_rate": 9.708431000223252e-06, + "loss": 0.18, + "step": 10011 + }, + { + "epoch": 4.19, + "learning_rate": 9.706736316350192e-06, + "loss": 0.1598, + "step": 10012 + }, + { + "epoch": 4.19, + "learning_rate": 9.705041640906736e-06, + "loss": 0.1305, + "step": 10013 + }, + { + "epoch": 4.19, + "learning_rate": 9.703346973941595e-06, + "loss": 0.1635, + "step": 10014 + }, + { + "epoch": 4.19, + "learning_rate": 9.701652315503484e-06, + "loss": 0.1586, + "step": 10015 + }, + { + "epoch": 4.19, + "learning_rate": 9.69995766564111e-06, + "loss": 0.1495, + "step": 10016 + }, + { + "epoch": 4.19, + "learning_rate": 9.698263024403187e-06, + "loss": 0.1283, + "step": 10017 + }, + { + "epoch": 4.19, + "learning_rate": 9.696568391838428e-06, + "loss": 0.1545, + "step": 10018 + }, + { + "epoch": 4.19, + "learning_rate": 9.69487376799554e-06, + "loss": 0.1423, + "step": 10019 + }, + { + "epoch": 4.2, + "learning_rate": 9.693179152923233e-06, + "loss": 0.1542, + "step": 10020 + }, + { + "epoch": 4.2, + "learning_rate": 9.691484546670222e-06, + "loss": 0.1238, + "step": 10021 + }, + { + "epoch": 4.2, + "learning_rate": 9.689789949285213e-06, + "loss": 0.1384, + "step": 10022 + }, + { + "epoch": 4.2, + "learning_rate": 9.688095360816916e-06, + "loss": 0.1603, + "step": 10023 + }, + { + "epoch": 4.2, + "learning_rate": 9.686400781314042e-06, + "loss": 0.1407, + "step": 10024 + }, + { + "epoch": 4.2, + "learning_rate": 9.684706210825298e-06, + "loss": 0.156, + "step": 10025 + }, + { + "epoch": 4.2, + "learning_rate": 9.683011649399393e-06, + "loss": 0.1402, + "step": 10026 + }, + { + "epoch": 4.2, + "learning_rate": 9.681317097085041e-06, + "loss": 0.1731, + "step": 10027 + }, + { + "epoch": 4.2, + "learning_rate": 9.679622553930943e-06, + "loss": 0.1694, + "step": 10028 + }, + { + "epoch": 4.2, + "learning_rate": 9.677928019985811e-06, + "loss": 0.179, + "step": 10029 + }, + { + "epoch": 4.2, + "learning_rate": 9.676233495298353e-06, + "loss": 0.1417, + "step": 10030 + }, + { + "epoch": 4.2, + "learning_rate": 9.674538979917274e-06, + "loss": 0.1471, + "step": 10031 + }, + { + "epoch": 4.2, + "learning_rate": 9.67284447389129e-06, + "loss": 0.127, + "step": 10032 + }, + { + "epoch": 4.2, + "learning_rate": 9.671149977269095e-06, + "loss": 0.1081, + "step": 10033 + }, + { + "epoch": 4.2, + "learning_rate": 9.669455490099402e-06, + "loss": 0.1672, + "step": 10034 + }, + { + "epoch": 4.2, + "learning_rate": 9.667761012430919e-06, + "loss": 0.2019, + "step": 10035 + }, + { + "epoch": 4.2, + "learning_rate": 9.66606654431235e-06, + "loss": 0.144, + "step": 10036 + }, + { + "epoch": 4.2, + "learning_rate": 9.664372085792401e-06, + "loss": 0.1272, + "step": 10037 + }, + { + "epoch": 4.2, + "learning_rate": 9.662677636919783e-06, + "loss": 0.1226, + "step": 10038 + }, + { + "epoch": 4.2, + "learning_rate": 9.660983197743192e-06, + "loss": 0.1758, + "step": 10039 + }, + { + "epoch": 4.2, + "learning_rate": 9.65928876831134e-06, + "loss": 0.1523, + "step": 10040 + }, + { + "epoch": 4.2, + "learning_rate": 9.657594348672931e-06, + "loss": 0.1713, + "step": 10041 + }, + { + "epoch": 4.2, + "learning_rate": 9.655899938876668e-06, + "loss": 0.1666, + "step": 10042 + }, + { + "epoch": 4.2, + "learning_rate": 9.654205538971254e-06, + "loss": 0.1977, + "step": 10043 + }, + { + "epoch": 4.21, + "learning_rate": 9.652511149005396e-06, + "loss": 0.1409, + "step": 10044 + }, + { + "epoch": 4.21, + "learning_rate": 9.650816769027797e-06, + "loss": 0.1365, + "step": 10045 + }, + { + "epoch": 4.21, + "learning_rate": 9.649122399087162e-06, + "loss": 0.1512, + "step": 10046 + }, + { + "epoch": 4.21, + "learning_rate": 9.647428039232188e-06, + "loss": 0.1452, + "step": 10047 + }, + { + "epoch": 4.21, + "learning_rate": 9.645733689511585e-06, + "loss": 0.148, + "step": 10048 + }, + { + "epoch": 4.21, + "learning_rate": 9.644039349974054e-06, + "loss": 0.1768, + "step": 10049 + }, + { + "epoch": 4.21, + "learning_rate": 9.642345020668294e-06, + "loss": 0.1461, + "step": 10050 + }, + { + "epoch": 4.21, + "learning_rate": 9.64065070164301e-06, + "loss": 0.1729, + "step": 10051 + }, + { + "epoch": 4.21, + "learning_rate": 9.6389563929469e-06, + "loss": 0.1695, + "step": 10052 + }, + { + "epoch": 4.21, + "learning_rate": 9.63726209462867e-06, + "loss": 0.1668, + "step": 10053 + }, + { + "epoch": 4.21, + "learning_rate": 9.635567806737021e-06, + "loss": 0.1861, + "step": 10054 + }, + { + "epoch": 4.21, + "learning_rate": 9.633873529320652e-06, + "loss": 0.1445, + "step": 10055 + }, + { + "epoch": 4.21, + "learning_rate": 9.632179262428264e-06, + "loss": 0.1459, + "step": 10056 + }, + { + "epoch": 4.21, + "learning_rate": 9.630485006108554e-06, + "loss": 0.2078, + "step": 10057 + }, + { + "epoch": 4.21, + "learning_rate": 9.628790760410224e-06, + "loss": 0.2002, + "step": 10058 + }, + { + "epoch": 4.21, + "learning_rate": 9.627096525381978e-06, + "loss": 0.1857, + "step": 10059 + }, + { + "epoch": 4.21, + "learning_rate": 9.625402301072508e-06, + "loss": 0.1554, + "step": 10060 + }, + { + "epoch": 4.21, + "learning_rate": 9.623708087530515e-06, + "loss": 0.1844, + "step": 10061 + }, + { + "epoch": 4.21, + "learning_rate": 9.622013884804704e-06, + "loss": 0.1169, + "step": 10062 + }, + { + "epoch": 4.21, + "learning_rate": 9.620319692943765e-06, + "loss": 0.1747, + "step": 10063 + }, + { + "epoch": 4.21, + "learning_rate": 9.618625511996398e-06, + "loss": 0.1651, + "step": 10064 + }, + { + "epoch": 4.21, + "learning_rate": 9.616931342011307e-06, + "loss": 0.1641, + "step": 10065 + }, + { + "epoch": 4.21, + "learning_rate": 9.615237183037181e-06, + "loss": 0.1532, + "step": 10066 + }, + { + "epoch": 4.21, + "learning_rate": 9.613543035122722e-06, + "loss": 0.1131, + "step": 10067 + }, + { + "epoch": 4.22, + "learning_rate": 9.611848898316626e-06, + "loss": 0.1774, + "step": 10068 + }, + { + "epoch": 4.22, + "learning_rate": 9.610154772667589e-06, + "loss": 0.1478, + "step": 10069 + }, + { + "epoch": 4.22, + "learning_rate": 9.608460658224306e-06, + "loss": 0.1161, + "step": 10070 + }, + { + "epoch": 4.22, + "learning_rate": 9.606766555035475e-06, + "loss": 0.1271, + "step": 10071 + }, + { + "epoch": 4.22, + "learning_rate": 9.60507246314979e-06, + "loss": 0.1846, + "step": 10072 + }, + { + "epoch": 4.22, + "learning_rate": 9.603378382615946e-06, + "loss": 0.0966, + "step": 10073 + }, + { + "epoch": 4.22, + "learning_rate": 9.601684313482638e-06, + "loss": 0.1844, + "step": 10074 + }, + { + "epoch": 4.22, + "learning_rate": 9.599990255798564e-06, + "loss": 0.1447, + "step": 10075 + }, + { + "epoch": 4.22, + "learning_rate": 9.598296209612413e-06, + "loss": 0.1451, + "step": 10076 + }, + { + "epoch": 4.22, + "learning_rate": 9.596602174972883e-06, + "loss": 0.1175, + "step": 10077 + }, + { + "epoch": 4.22, + "learning_rate": 9.594908151928663e-06, + "loss": 0.1086, + "step": 10078 + }, + { + "epoch": 4.22, + "learning_rate": 9.593214140528451e-06, + "loss": 0.1838, + "step": 10079 + }, + { + "epoch": 4.22, + "learning_rate": 9.591520140820936e-06, + "loss": 0.1618, + "step": 10080 + }, + { + "epoch": 4.22, + "learning_rate": 9.58982615285481e-06, + "loss": 0.1676, + "step": 10081 + }, + { + "epoch": 4.22, + "learning_rate": 9.588132176678772e-06, + "loss": 0.1582, + "step": 10082 + }, + { + "epoch": 4.22, + "learning_rate": 9.586438212341505e-06, + "loss": 0.1614, + "step": 10083 + }, + { + "epoch": 4.22, + "learning_rate": 9.584744259891706e-06, + "loss": 0.1732, + "step": 10084 + }, + { + "epoch": 4.22, + "learning_rate": 9.583050319378063e-06, + "loss": 0.1284, + "step": 10085 + }, + { + "epoch": 4.22, + "learning_rate": 9.581356390849272e-06, + "loss": 0.1317, + "step": 10086 + }, + { + "epoch": 4.22, + "learning_rate": 9.579662474354017e-06, + "loss": 0.1301, + "step": 10087 + }, + { + "epoch": 4.22, + "learning_rate": 9.577968569940992e-06, + "loss": 0.1623, + "step": 10088 + }, + { + "epoch": 4.22, + "learning_rate": 9.576274677658888e-06, + "loss": 0.1329, + "step": 10089 + }, + { + "epoch": 4.22, + "learning_rate": 9.574580797556389e-06, + "loss": 0.162, + "step": 10090 + }, + { + "epoch": 4.22, + "learning_rate": 9.57288692968219e-06, + "loss": 0.1418, + "step": 10091 + }, + { + "epoch": 4.23, + "learning_rate": 9.571193074084976e-06, + "loss": 0.1598, + "step": 10092 + }, + { + "epoch": 4.23, + "learning_rate": 9.569499230813437e-06, + "loss": 0.14, + "step": 10093 + }, + { + "epoch": 4.23, + "learning_rate": 9.56780539991626e-06, + "loss": 0.1781, + "step": 10094 + }, + { + "epoch": 4.23, + "learning_rate": 9.566111581442136e-06, + "loss": 0.0888, + "step": 10095 + }, + { + "epoch": 4.23, + "learning_rate": 9.564417775439747e-06, + "loss": 0.1739, + "step": 10096 + }, + { + "epoch": 4.23, + "learning_rate": 9.562723981957781e-06, + "loss": 0.1608, + "step": 10097 + }, + { + "epoch": 4.23, + "learning_rate": 9.56103020104493e-06, + "loss": 0.1476, + "step": 10098 + }, + { + "epoch": 4.23, + "learning_rate": 9.559336432749873e-06, + "loss": 0.1451, + "step": 10099 + }, + { + "epoch": 4.23, + "learning_rate": 9.557642677121306e-06, + "loss": 0.1617, + "step": 10100 + }, + { + "epoch": 4.23, + "learning_rate": 9.5559489342079e-06, + "loss": 0.2092, + "step": 10101 + }, + { + "epoch": 4.23, + "learning_rate": 9.55425520405835e-06, + "loss": 0.1794, + "step": 10102 + }, + { + "epoch": 4.23, + "learning_rate": 9.55256148672134e-06, + "loss": 0.1469, + "step": 10103 + }, + { + "epoch": 4.23, + "learning_rate": 9.55086778224555e-06, + "loss": 0.1448, + "step": 10104 + }, + { + "epoch": 4.23, + "learning_rate": 9.54917409067967e-06, + "loss": 0.1527, + "step": 10105 + }, + { + "epoch": 4.23, + "learning_rate": 9.547480412072379e-06, + "loss": 0.1258, + "step": 10106 + }, + { + "epoch": 4.23, + "learning_rate": 9.545786746472363e-06, + "loss": 0.1697, + "step": 10107 + }, + { + "epoch": 4.23, + "learning_rate": 9.544093093928302e-06, + "loss": 0.1573, + "step": 10108 + }, + { + "epoch": 4.23, + "learning_rate": 9.542399454488883e-06, + "loss": 0.161, + "step": 10109 + }, + { + "epoch": 4.23, + "learning_rate": 9.540705828202783e-06, + "loss": 0.1356, + "step": 10110 + }, + { + "epoch": 4.23, + "learning_rate": 9.539012215118688e-06, + "loss": 0.1737, + "step": 10111 + }, + { + "epoch": 4.23, + "learning_rate": 9.537318615285279e-06, + "loss": 0.1357, + "step": 10112 + }, + { + "epoch": 4.23, + "learning_rate": 9.535625028751235e-06, + "loss": 0.1329, + "step": 10113 + }, + { + "epoch": 4.23, + "learning_rate": 9.533931455565236e-06, + "loss": 0.1777, + "step": 10114 + }, + { + "epoch": 4.23, + "learning_rate": 9.532237895775963e-06, + "loss": 0.1467, + "step": 10115 + }, + { + "epoch": 4.24, + "learning_rate": 9.5305443494321e-06, + "loss": 0.1216, + "step": 10116 + }, + { + "epoch": 4.24, + "learning_rate": 9.52885081658232e-06, + "loss": 0.1411, + "step": 10117 + }, + { + "epoch": 4.24, + "learning_rate": 9.527157297275306e-06, + "loss": 0.1181, + "step": 10118 + }, + { + "epoch": 4.24, + "learning_rate": 9.525463791559737e-06, + "loss": 0.1709, + "step": 10119 + }, + { + "epoch": 4.24, + "learning_rate": 9.52377029948429e-06, + "loss": 0.1502, + "step": 10120 + }, + { + "epoch": 4.24, + "learning_rate": 9.522076821097643e-06, + "loss": 0.1715, + "step": 10121 + }, + { + "epoch": 4.24, + "learning_rate": 9.520383356448476e-06, + "loss": 0.1614, + "step": 10122 + }, + { + "epoch": 4.24, + "learning_rate": 9.518689905585463e-06, + "loss": 0.1279, + "step": 10123 + }, + { + "epoch": 4.24, + "learning_rate": 9.51699646855728e-06, + "loss": 0.1557, + "step": 10124 + }, + { + "epoch": 4.24, + "learning_rate": 9.515303045412604e-06, + "loss": 0.1346, + "step": 10125 + }, + { + "epoch": 4.24, + "learning_rate": 9.513609636200112e-06, + "loss": 0.1711, + "step": 10126 + }, + { + "epoch": 4.24, + "learning_rate": 9.511916240968481e-06, + "loss": 0.1474, + "step": 10127 + }, + { + "epoch": 4.24, + "learning_rate": 9.510222859766383e-06, + "loss": 0.1593, + "step": 10128 + }, + { + "epoch": 4.24, + "learning_rate": 9.508529492642493e-06, + "loss": 0.1478, + "step": 10129 + }, + { + "epoch": 4.24, + "learning_rate": 9.506836139645491e-06, + "loss": 0.175, + "step": 10130 + }, + { + "epoch": 4.24, + "learning_rate": 9.505142800824044e-06, + "loss": 0.1369, + "step": 10131 + }, + { + "epoch": 4.24, + "learning_rate": 9.503449476226826e-06, + "loss": 0.1794, + "step": 10132 + }, + { + "epoch": 4.24, + "learning_rate": 9.501756165902517e-06, + "loss": 0.1383, + "step": 10133 + }, + { + "epoch": 4.24, + "learning_rate": 9.50006286989978e-06, + "loss": 0.1675, + "step": 10134 + }, + { + "epoch": 4.24, + "learning_rate": 9.498369588267295e-06, + "loss": 0.1657, + "step": 10135 + }, + { + "epoch": 4.24, + "learning_rate": 9.49667632105373e-06, + "loss": 0.1763, + "step": 10136 + }, + { + "epoch": 4.24, + "learning_rate": 9.494983068307759e-06, + "loss": 0.1615, + "step": 10137 + }, + { + "epoch": 4.24, + "learning_rate": 9.493289830078049e-06, + "loss": 0.1786, + "step": 10138 + }, + { + "epoch": 4.24, + "learning_rate": 9.491596606413275e-06, + "loss": 0.1827, + "step": 10139 + }, + { + "epoch": 4.25, + "learning_rate": 9.489903397362104e-06, + "loss": 0.146, + "step": 10140 + }, + { + "epoch": 4.25, + "learning_rate": 9.488210202973208e-06, + "loss": 0.1449, + "step": 10141 + }, + { + "epoch": 4.25, + "learning_rate": 9.486517023295257e-06, + "loss": 0.1465, + "step": 10142 + }, + { + "epoch": 4.25, + "learning_rate": 9.484823858376916e-06, + "loss": 0.1521, + "step": 10143 + }, + { + "epoch": 4.25, + "learning_rate": 9.483130708266856e-06, + "loss": 0.1594, + "step": 10144 + }, + { + "epoch": 4.25, + "learning_rate": 9.48143757301375e-06, + "loss": 0.1279, + "step": 10145 + }, + { + "epoch": 4.25, + "learning_rate": 9.479744452666257e-06, + "loss": 0.2213, + "step": 10146 + }, + { + "epoch": 4.25, + "learning_rate": 9.47805134727305e-06, + "loss": 0.133, + "step": 10147 + }, + { + "epoch": 4.25, + "learning_rate": 9.476358256882792e-06, + "loss": 0.1305, + "step": 10148 + }, + { + "epoch": 4.25, + "learning_rate": 9.47466518154415e-06, + "loss": 0.1897, + "step": 10149 + }, + { + "epoch": 4.25, + "learning_rate": 9.472972121305795e-06, + "loss": 0.1382, + "step": 10150 + }, + { + "epoch": 4.25, + "learning_rate": 9.471279076216385e-06, + "loss": 0.1311, + "step": 10151 + }, + { + "epoch": 4.25, + "learning_rate": 9.46958604632459e-06, + "loss": 0.1497, + "step": 10152 + }, + { + "epoch": 4.25, + "learning_rate": 9.467893031679073e-06, + "loss": 0.1205, + "step": 10153 + }, + { + "epoch": 4.25, + "learning_rate": 9.466200032328499e-06, + "loss": 0.162, + "step": 10154 + }, + { + "epoch": 4.25, + "learning_rate": 9.46450704832153e-06, + "loss": 0.1396, + "step": 10155 + }, + { + "epoch": 4.25, + "learning_rate": 9.462814079706832e-06, + "loss": 0.1599, + "step": 10156 + }, + { + "epoch": 4.25, + "learning_rate": 9.461121126533069e-06, + "loss": 0.1311, + "step": 10157 + }, + { + "epoch": 4.25, + "learning_rate": 9.459428188848897e-06, + "loss": 0.1662, + "step": 10158 + }, + { + "epoch": 4.25, + "learning_rate": 9.457735266702984e-06, + "loss": 0.1174, + "step": 10159 + }, + { + "epoch": 4.25, + "learning_rate": 9.45604236014399e-06, + "loss": 0.1673, + "step": 10160 + }, + { + "epoch": 4.25, + "learning_rate": 9.454349469220573e-06, + "loss": 0.1501, + "step": 10161 + }, + { + "epoch": 4.25, + "learning_rate": 9.452656593981398e-06, + "loss": 0.1671, + "step": 10162 + }, + { + "epoch": 4.25, + "learning_rate": 9.450963734475125e-06, + "loss": 0.1354, + "step": 10163 + }, + { + "epoch": 4.26, + "learning_rate": 9.44927089075041e-06, + "loss": 0.2105, + "step": 10164 + }, + { + "epoch": 4.26, + "learning_rate": 9.447578062855915e-06, + "loss": 0.1904, + "step": 10165 + }, + { + "epoch": 4.26, + "learning_rate": 9.445885250840301e-06, + "loss": 0.1371, + "step": 10166 + }, + { + "epoch": 4.26, + "learning_rate": 9.444192454752221e-06, + "loss": 0.1447, + "step": 10167 + }, + { + "epoch": 4.26, + "learning_rate": 9.44249967464034e-06, + "loss": 0.1443, + "step": 10168 + }, + { + "epoch": 4.26, + "learning_rate": 9.440806910553308e-06, + "loss": 0.1216, + "step": 10169 + }, + { + "epoch": 4.26, + "learning_rate": 9.439114162539784e-06, + "loss": 0.138, + "step": 10170 + }, + { + "epoch": 4.26, + "learning_rate": 9.437421430648426e-06, + "loss": 0.1481, + "step": 10171 + }, + { + "epoch": 4.26, + "learning_rate": 9.43572871492789e-06, + "loss": 0.152, + "step": 10172 + }, + { + "epoch": 4.26, + "learning_rate": 9.434036015426831e-06, + "loss": 0.1472, + "step": 10173 + }, + { + "epoch": 4.26, + "learning_rate": 9.432343332193907e-06, + "loss": 0.1496, + "step": 10174 + }, + { + "epoch": 4.26, + "learning_rate": 9.430650665277767e-06, + "loss": 0.1479, + "step": 10175 + }, + { + "epoch": 4.26, + "learning_rate": 9.428958014727068e-06, + "loss": 0.1076, + "step": 10176 + }, + { + "epoch": 4.26, + "learning_rate": 9.427265380590466e-06, + "loss": 0.1564, + "step": 10177 + }, + { + "epoch": 4.26, + "learning_rate": 9.425572762916613e-06, + "loss": 0.1509, + "step": 10178 + }, + { + "epoch": 4.26, + "learning_rate": 9.423880161754158e-06, + "loss": 0.1453, + "step": 10179 + }, + { + "epoch": 4.26, + "learning_rate": 9.42218757715176e-06, + "loss": 0.1317, + "step": 10180 + }, + { + "epoch": 4.26, + "learning_rate": 9.420495009158066e-06, + "loss": 0.1405, + "step": 10181 + }, + { + "epoch": 4.26, + "learning_rate": 9.418802457821729e-06, + "loss": 0.1627, + "step": 10182 + }, + { + "epoch": 4.26, + "learning_rate": 9.4171099231914e-06, + "loss": 0.1302, + "step": 10183 + }, + { + "epoch": 4.26, + "learning_rate": 9.415417405315727e-06, + "loss": 0.1721, + "step": 10184 + }, + { + "epoch": 4.26, + "learning_rate": 9.413724904243363e-06, + "loss": 0.164, + "step": 10185 + }, + { + "epoch": 4.26, + "learning_rate": 9.412032420022956e-06, + "loss": 0.1351, + "step": 10186 + }, + { + "epoch": 4.27, + "learning_rate": 9.410339952703157e-06, + "loss": 0.116, + "step": 10187 + }, + { + "epoch": 4.27, + "learning_rate": 9.408647502332612e-06, + "loss": 0.1532, + "step": 10188 + }, + { + "epoch": 4.27, + "learning_rate": 9.40695506895997e-06, + "loss": 0.1497, + "step": 10189 + }, + { + "epoch": 4.27, + "learning_rate": 9.405262652633879e-06, + "loss": 0.1906, + "step": 10190 + }, + { + "epoch": 4.27, + "learning_rate": 9.403570253402987e-06, + "loss": 0.1771, + "step": 10191 + }, + { + "epoch": 4.27, + "learning_rate": 9.401877871315936e-06, + "loss": 0.1526, + "step": 10192 + }, + { + "epoch": 4.27, + "learning_rate": 9.400185506421373e-06, + "loss": 0.1485, + "step": 10193 + }, + { + "epoch": 4.27, + "learning_rate": 9.39849315876795e-06, + "loss": 0.0871, + "step": 10194 + }, + { + "epoch": 4.27, + "learning_rate": 9.396800828404304e-06, + "loss": 0.1601, + "step": 10195 + }, + { + "epoch": 4.27, + "learning_rate": 9.395108515379082e-06, + "loss": 0.1222, + "step": 10196 + }, + { + "epoch": 4.27, + "learning_rate": 9.393416219740929e-06, + "loss": 0.1923, + "step": 10197 + }, + { + "epoch": 4.27, + "learning_rate": 9.391723941538492e-06, + "loss": 0.1477, + "step": 10198 + }, + { + "epoch": 4.27, + "learning_rate": 9.390031680820409e-06, + "loss": 0.1737, + "step": 10199 + }, + { + "epoch": 4.27, + "learning_rate": 9.388339437635321e-06, + "loss": 0.1635, + "step": 10200 + }, + { + "epoch": 4.27, + "learning_rate": 9.386647212031877e-06, + "loss": 0.1653, + "step": 10201 + }, + { + "epoch": 4.27, + "learning_rate": 9.384955004058714e-06, + "loss": 0.1863, + "step": 10202 + }, + { + "epoch": 4.27, + "learning_rate": 9.383262813764471e-06, + "loss": 0.1611, + "step": 10203 + }, + { + "epoch": 4.27, + "learning_rate": 9.381570641197795e-06, + "loss": 0.1518, + "step": 10204 + }, + { + "epoch": 4.27, + "learning_rate": 9.37987848640732e-06, + "loss": 0.183, + "step": 10205 + }, + { + "epoch": 4.27, + "learning_rate": 9.378186349441685e-06, + "loss": 0.1258, + "step": 10206 + }, + { + "epoch": 4.27, + "learning_rate": 9.376494230349537e-06, + "loss": 0.1622, + "step": 10207 + }, + { + "epoch": 4.27, + "learning_rate": 9.374802129179505e-06, + "loss": 0.1771, + "step": 10208 + }, + { + "epoch": 4.27, + "learning_rate": 9.373110045980232e-06, + "loss": 0.1735, + "step": 10209 + }, + { + "epoch": 4.27, + "learning_rate": 9.371417980800357e-06, + "loss": 0.1723, + "step": 10210 + }, + { + "epoch": 4.28, + "learning_rate": 9.369725933688512e-06, + "loss": 0.1516, + "step": 10211 + }, + { + "epoch": 4.28, + "learning_rate": 9.368033904693334e-06, + "loss": 0.1744, + "step": 10212 + }, + { + "epoch": 4.28, + "learning_rate": 9.366341893863467e-06, + "loss": 0.1628, + "step": 10213 + }, + { + "epoch": 4.28, + "learning_rate": 9.364649901247534e-06, + "loss": 0.1779, + "step": 10214 + }, + { + "epoch": 4.28, + "learning_rate": 9.362957926894179e-06, + "loss": 0.1468, + "step": 10215 + }, + { + "epoch": 4.28, + "learning_rate": 9.361265970852031e-06, + "loss": 0.1671, + "step": 10216 + }, + { + "epoch": 4.28, + "learning_rate": 9.359574033169725e-06, + "loss": 0.1538, + "step": 10217 + }, + { + "epoch": 4.28, + "learning_rate": 9.357882113895899e-06, + "loss": 0.1673, + "step": 10218 + }, + { + "epoch": 4.28, + "learning_rate": 9.356190213079178e-06, + "loss": 0.1574, + "step": 10219 + }, + { + "epoch": 4.28, + "learning_rate": 9.354498330768198e-06, + "loss": 0.15, + "step": 10220 + }, + { + "epoch": 4.28, + "learning_rate": 9.352806467011595e-06, + "loss": 0.133, + "step": 10221 + }, + { + "epoch": 4.28, + "learning_rate": 9.35111462185799e-06, + "loss": 0.158, + "step": 10222 + }, + { + "epoch": 4.28, + "learning_rate": 9.349422795356022e-06, + "loss": 0.1445, + "step": 10223 + }, + { + "epoch": 4.28, + "learning_rate": 9.34773098755432e-06, + "loss": 0.2036, + "step": 10224 + }, + { + "epoch": 4.28, + "learning_rate": 9.34603919850151e-06, + "loss": 0.1615, + "step": 10225 + }, + { + "epoch": 4.28, + "learning_rate": 9.344347428246222e-06, + "loss": 0.1625, + "step": 10226 + }, + { + "epoch": 4.28, + "learning_rate": 9.342655676837085e-06, + "loss": 0.1402, + "step": 10227 + }, + { + "epoch": 4.28, + "learning_rate": 9.340963944322731e-06, + "loss": 0.185, + "step": 10228 + }, + { + "epoch": 4.28, + "learning_rate": 9.33927223075178e-06, + "loss": 0.1517, + "step": 10229 + }, + { + "epoch": 4.28, + "learning_rate": 9.337580536172864e-06, + "loss": 0.1605, + "step": 10230 + }, + { + "epoch": 4.28, + "learning_rate": 9.33588886063461e-06, + "loss": 0.1153, + "step": 10231 + }, + { + "epoch": 4.28, + "learning_rate": 9.334197204185638e-06, + "loss": 0.1657, + "step": 10232 + }, + { + "epoch": 4.28, + "learning_rate": 9.332505566874577e-06, + "loss": 0.1653, + "step": 10233 + }, + { + "epoch": 4.28, + "learning_rate": 9.330813948750053e-06, + "loss": 0.1501, + "step": 10234 + }, + { + "epoch": 4.29, + "learning_rate": 9.329122349860686e-06, + "loss": 0.1349, + "step": 10235 + }, + { + "epoch": 4.29, + "learning_rate": 9.327430770255106e-06, + "loss": 0.1312, + "step": 10236 + }, + { + "epoch": 4.29, + "learning_rate": 9.325739209981929e-06, + "loss": 0.155, + "step": 10237 + }, + { + "epoch": 4.29, + "learning_rate": 9.324047669089778e-06, + "loss": 0.1361, + "step": 10238 + }, + { + "epoch": 4.29, + "learning_rate": 9.32235614762728e-06, + "loss": 0.1375, + "step": 10239 + }, + { + "epoch": 4.29, + "learning_rate": 9.320664645643051e-06, + "loss": 0.1078, + "step": 10240 + }, + { + "epoch": 4.29, + "learning_rate": 9.318973163185713e-06, + "loss": 0.1587, + "step": 10241 + }, + { + "epoch": 4.29, + "learning_rate": 9.31728170030389e-06, + "loss": 0.172, + "step": 10242 + }, + { + "epoch": 4.29, + "learning_rate": 9.315590257046197e-06, + "loss": 0.1524, + "step": 10243 + }, + { + "epoch": 4.29, + "learning_rate": 9.313898833461255e-06, + "loss": 0.1059, + "step": 10244 + }, + { + "epoch": 4.29, + "learning_rate": 9.312207429597684e-06, + "loss": 0.1627, + "step": 10245 + }, + { + "epoch": 4.29, + "learning_rate": 9.310516045504096e-06, + "loss": 0.1688, + "step": 10246 + }, + { + "epoch": 4.29, + "learning_rate": 9.308824681229115e-06, + "loss": 0.1488, + "step": 10247 + }, + { + "epoch": 4.29, + "learning_rate": 9.307133336821357e-06, + "loss": 0.1414, + "step": 10248 + }, + { + "epoch": 4.29, + "learning_rate": 9.305442012329433e-06, + "loss": 0.1393, + "step": 10249 + }, + { + "epoch": 4.29, + "learning_rate": 9.303750707801963e-06, + "loss": 0.1709, + "step": 10250 + }, + { + "epoch": 4.29, + "learning_rate": 9.302059423287563e-06, + "loss": 0.1623, + "step": 10251 + }, + { + "epoch": 4.29, + "learning_rate": 9.300368158834843e-06, + "loss": 0.1603, + "step": 10252 + }, + { + "epoch": 4.29, + "learning_rate": 9.298676914492419e-06, + "loss": 0.174, + "step": 10253 + }, + { + "epoch": 4.29, + "learning_rate": 9.29698569030891e-06, + "loss": 0.1618, + "step": 10254 + }, + { + "epoch": 4.29, + "learning_rate": 9.295294486332916e-06, + "loss": 0.1749, + "step": 10255 + }, + { + "epoch": 4.29, + "learning_rate": 9.29360330261306e-06, + "loss": 0.175, + "step": 10256 + }, + { + "epoch": 4.29, + "learning_rate": 9.29191213919795e-06, + "loss": 0.1675, + "step": 10257 + }, + { + "epoch": 4.29, + "learning_rate": 9.290220996136199e-06, + "loss": 0.1704, + "step": 10258 + }, + { + "epoch": 4.3, + "learning_rate": 9.288529873476418e-06, + "loss": 0.1415, + "step": 10259 + }, + { + "epoch": 4.3, + "learning_rate": 9.28683877126721e-06, + "loss": 0.1458, + "step": 10260 + }, + { + "epoch": 4.3, + "learning_rate": 9.285147689557188e-06, + "loss": 0.1277, + "step": 10261 + }, + { + "epoch": 4.3, + "learning_rate": 9.283456628394966e-06, + "loss": 0.1414, + "step": 10262 + }, + { + "epoch": 4.3, + "learning_rate": 9.281765587829143e-06, + "loss": 0.1546, + "step": 10263 + }, + { + "epoch": 4.3, + "learning_rate": 9.28007456790833e-06, + "loss": 0.1535, + "step": 10264 + }, + { + "epoch": 4.3, + "learning_rate": 9.278383568681139e-06, + "loss": 0.161, + "step": 10265 + }, + { + "epoch": 4.3, + "learning_rate": 9.27669259019617e-06, + "loss": 0.1447, + "step": 10266 + }, + { + "epoch": 4.3, + "learning_rate": 9.275001632502029e-06, + "loss": 0.1506, + "step": 10267 + }, + { + "epoch": 4.3, + "learning_rate": 9.273310695647323e-06, + "loss": 0.1517, + "step": 10268 + }, + { + "epoch": 4.3, + "learning_rate": 9.271619779680658e-06, + "loss": 0.1645, + "step": 10269 + }, + { + "epoch": 4.3, + "learning_rate": 9.269928884650635e-06, + "loss": 0.1579, + "step": 10270 + }, + { + "epoch": 4.3, + "learning_rate": 9.268238010605858e-06, + "loss": 0.1617, + "step": 10271 + }, + { + "epoch": 4.3, + "learning_rate": 9.266547157594932e-06, + "loss": 0.1459, + "step": 10272 + }, + { + "epoch": 4.3, + "learning_rate": 9.264856325666455e-06, + "loss": 0.1122, + "step": 10273 + }, + { + "epoch": 4.3, + "learning_rate": 9.26316551486903e-06, + "loss": 0.1507, + "step": 10274 + }, + { + "epoch": 4.3, + "learning_rate": 9.261474725251261e-06, + "loss": 0.1768, + "step": 10275 + }, + { + "epoch": 4.3, + "learning_rate": 9.259783956861744e-06, + "loss": 0.1558, + "step": 10276 + }, + { + "epoch": 4.3, + "learning_rate": 9.258093209749081e-06, + "loss": 0.1558, + "step": 10277 + }, + { + "epoch": 4.3, + "learning_rate": 9.25640248396187e-06, + "loss": 0.1407, + "step": 10278 + }, + { + "epoch": 4.3, + "learning_rate": 9.254711779548709e-06, + "loss": 0.1367, + "step": 10279 + }, + { + "epoch": 4.3, + "learning_rate": 9.253021096558197e-06, + "loss": 0.1778, + "step": 10280 + }, + { + "epoch": 4.3, + "learning_rate": 9.251330435038935e-06, + "loss": 0.1635, + "step": 10281 + }, + { + "epoch": 4.3, + "learning_rate": 9.24963979503951e-06, + "loss": 0.111, + "step": 10282 + }, + { + "epoch": 4.31, + "learning_rate": 9.247949176608527e-06, + "loss": 0.1337, + "step": 10283 + }, + { + "epoch": 4.31, + "learning_rate": 9.246258579794574e-06, + "loss": 0.1602, + "step": 10284 + }, + { + "epoch": 4.31, + "learning_rate": 9.244568004646249e-06, + "loss": 0.1061, + "step": 10285 + }, + { + "epoch": 4.31, + "learning_rate": 9.24287745121215e-06, + "loss": 0.1671, + "step": 10286 + }, + { + "epoch": 4.31, + "learning_rate": 9.241186919540864e-06, + "loss": 0.202, + "step": 10287 + }, + { + "epoch": 4.31, + "learning_rate": 9.239496409680986e-06, + "loss": 0.1655, + "step": 10288 + }, + { + "epoch": 4.31, + "learning_rate": 9.237805921681112e-06, + "loss": 0.1347, + "step": 10289 + }, + { + "epoch": 4.31, + "learning_rate": 9.236115455589827e-06, + "loss": 0.1724, + "step": 10290 + }, + { + "epoch": 4.31, + "learning_rate": 9.234425011455727e-06, + "loss": 0.1192, + "step": 10291 + }, + { + "epoch": 4.31, + "learning_rate": 9.232734589327402e-06, + "loss": 0.1302, + "step": 10292 + }, + { + "epoch": 4.31, + "learning_rate": 9.231044189253438e-06, + "loss": 0.144, + "step": 10293 + }, + { + "epoch": 4.31, + "learning_rate": 9.229353811282427e-06, + "loss": 0.1895, + "step": 10294 + }, + { + "epoch": 4.31, + "learning_rate": 9.227663455462957e-06, + "loss": 0.1672, + "step": 10295 + }, + { + "epoch": 4.31, + "learning_rate": 9.225973121843615e-06, + "loss": 0.1519, + "step": 10296 + }, + { + "epoch": 4.31, + "learning_rate": 9.22428281047299e-06, + "loss": 0.1131, + "step": 10297 + }, + { + "epoch": 4.31, + "learning_rate": 9.222592521399666e-06, + "loss": 0.1864, + "step": 10298 + }, + { + "epoch": 4.31, + "learning_rate": 9.220902254672232e-06, + "loss": 0.1448, + "step": 10299 + }, + { + "epoch": 4.31, + "learning_rate": 9.21921201033927e-06, + "loss": 0.1724, + "step": 10300 + }, + { + "epoch": 4.31, + "learning_rate": 9.217521788449365e-06, + "loss": 0.1048, + "step": 10301 + }, + { + "epoch": 4.31, + "learning_rate": 9.215831589051103e-06, + "loss": 0.141, + "step": 10302 + }, + { + "epoch": 4.31, + "learning_rate": 9.214141412193065e-06, + "loss": 0.1361, + "step": 10303 + }, + { + "epoch": 4.31, + "learning_rate": 9.212451257923841e-06, + "loss": 0.1285, + "step": 10304 + }, + { + "epoch": 4.31, + "learning_rate": 9.210761126292e-06, + "loss": 0.1157, + "step": 10305 + }, + { + "epoch": 4.31, + "learning_rate": 9.209071017346134e-06, + "loss": 0.1308, + "step": 10306 + }, + { + "epoch": 4.32, + "learning_rate": 9.207380931134816e-06, + "loss": 0.1388, + "step": 10307 + }, + { + "epoch": 4.32, + "learning_rate": 9.205690867706631e-06, + "loss": 0.1371, + "step": 10308 + }, + { + "epoch": 4.32, + "learning_rate": 9.204000827110157e-06, + "loss": 0.1566, + "step": 10309 + }, + { + "epoch": 4.32, + "learning_rate": 9.202310809393973e-06, + "loss": 0.1379, + "step": 10310 + }, + { + "epoch": 4.32, + "learning_rate": 9.200620814606659e-06, + "loss": 0.1757, + "step": 10311 + }, + { + "epoch": 4.32, + "learning_rate": 9.198930842796786e-06, + "loss": 0.1404, + "step": 10312 + }, + { + "epoch": 4.32, + "learning_rate": 9.197240894012939e-06, + "loss": 0.1312, + "step": 10313 + }, + { + "epoch": 4.32, + "learning_rate": 9.195550968303688e-06, + "loss": 0.1814, + "step": 10314 + }, + { + "epoch": 4.32, + "learning_rate": 9.19386106571761e-06, + "loss": 0.1651, + "step": 10315 + }, + { + "epoch": 4.32, + "learning_rate": 9.192171186303283e-06, + "loss": 0.1172, + "step": 10316 + }, + { + "epoch": 4.32, + "learning_rate": 9.190481330109276e-06, + "loss": 0.1769, + "step": 10317 + }, + { + "epoch": 4.32, + "learning_rate": 9.188791497184165e-06, + "loss": 0.1593, + "step": 10318 + }, + { + "epoch": 4.32, + "learning_rate": 9.187101687576525e-06, + "loss": 0.1895, + "step": 10319 + }, + { + "epoch": 4.32, + "learning_rate": 9.185411901334924e-06, + "loss": 0.1344, + "step": 10320 + }, + { + "epoch": 4.32, + "learning_rate": 9.183722138507934e-06, + "loss": 0.1776, + "step": 10321 + }, + { + "epoch": 4.32, + "learning_rate": 9.18203239914413e-06, + "loss": 0.2088, + "step": 10322 + }, + { + "epoch": 4.32, + "learning_rate": 9.180342683292074e-06, + "loss": 0.1396, + "step": 10323 + }, + { + "epoch": 4.32, + "learning_rate": 9.178652991000342e-06, + "loss": 0.1484, + "step": 10324 + }, + { + "epoch": 4.32, + "learning_rate": 9.176963322317503e-06, + "loss": 0.1141, + "step": 10325 + }, + { + "epoch": 4.32, + "learning_rate": 9.175273677292122e-06, + "loss": 0.1434, + "step": 10326 + }, + { + "epoch": 4.32, + "learning_rate": 9.17358405597277e-06, + "loss": 0.1797, + "step": 10327 + }, + { + "epoch": 4.32, + "learning_rate": 9.171894458408007e-06, + "loss": 0.1074, + "step": 10328 + }, + { + "epoch": 4.32, + "learning_rate": 9.170204884646402e-06, + "loss": 0.1534, + "step": 10329 + }, + { + "epoch": 4.32, + "learning_rate": 9.168515334736526e-06, + "loss": 0.1112, + "step": 10330 + }, + { + "epoch": 4.33, + "learning_rate": 9.166825808726934e-06, + "loss": 0.1436, + "step": 10331 + }, + { + "epoch": 4.33, + "learning_rate": 9.165136306666196e-06, + "loss": 0.1286, + "step": 10332 + }, + { + "epoch": 4.33, + "learning_rate": 9.163446828602876e-06, + "loss": 0.1368, + "step": 10333 + }, + { + "epoch": 4.33, + "learning_rate": 9.161757374585533e-06, + "loss": 0.1154, + "step": 10334 + }, + { + "epoch": 4.33, + "learning_rate": 9.16006794466273e-06, + "loss": 0.1594, + "step": 10335 + }, + { + "epoch": 4.33, + "learning_rate": 9.158378538883032e-06, + "loss": 0.1529, + "step": 10336 + }, + { + "epoch": 4.33, + "learning_rate": 9.156689157294992e-06, + "loss": 0.1316, + "step": 10337 + }, + { + "epoch": 4.33, + "learning_rate": 9.154999799947174e-06, + "loss": 0.1473, + "step": 10338 + }, + { + "epoch": 4.33, + "learning_rate": 9.153310466888138e-06, + "loss": 0.1211, + "step": 10339 + }, + { + "epoch": 4.33, + "learning_rate": 9.151621158166443e-06, + "loss": 0.1527, + "step": 10340 + }, + { + "epoch": 4.33, + "learning_rate": 9.149931873830642e-06, + "loss": 0.1493, + "step": 10341 + }, + { + "epoch": 4.33, + "learning_rate": 9.148242613929295e-06, + "loss": 0.1741, + "step": 10342 + }, + { + "epoch": 4.33, + "learning_rate": 9.146553378510961e-06, + "loss": 0.1669, + "step": 10343 + }, + { + "epoch": 4.33, + "learning_rate": 9.144864167624191e-06, + "loss": 0.1436, + "step": 10344 + }, + { + "epoch": 4.33, + "learning_rate": 9.143174981317542e-06, + "loss": 0.1587, + "step": 10345 + }, + { + "epoch": 4.33, + "learning_rate": 9.141485819639569e-06, + "loss": 0.2177, + "step": 10346 + }, + { + "epoch": 4.33, + "learning_rate": 9.139796682638824e-06, + "loss": 0.1276, + "step": 10347 + }, + { + "epoch": 4.33, + "learning_rate": 9.138107570363857e-06, + "loss": 0.1801, + "step": 10348 + }, + { + "epoch": 4.33, + "learning_rate": 9.13641848286323e-06, + "loss": 0.1598, + "step": 10349 + }, + { + "epoch": 4.33, + "learning_rate": 9.134729420185481e-06, + "loss": 0.1672, + "step": 10350 + }, + { + "epoch": 4.33, + "learning_rate": 9.13304038237917e-06, + "loss": 0.1669, + "step": 10351 + }, + { + "epoch": 4.33, + "learning_rate": 9.13135136949284e-06, + "loss": 0.1917, + "step": 10352 + }, + { + "epoch": 4.33, + "learning_rate": 9.129662381575047e-06, + "loss": 0.1498, + "step": 10353 + }, + { + "epoch": 4.33, + "learning_rate": 9.127973418674338e-06, + "loss": 0.1808, + "step": 10354 + }, + { + "epoch": 4.34, + "learning_rate": 9.126284480839256e-06, + "loss": 0.1474, + "step": 10355 + }, + { + "epoch": 4.34, + "learning_rate": 9.124595568118352e-06, + "loss": 0.1349, + "step": 10356 + }, + { + "epoch": 4.34, + "learning_rate": 9.122906680560173e-06, + "loss": 0.1752, + "step": 10357 + }, + { + "epoch": 4.34, + "learning_rate": 9.121217818213263e-06, + "loss": 0.126, + "step": 10358 + }, + { + "epoch": 4.34, + "learning_rate": 9.119528981126165e-06, + "loss": 0.1602, + "step": 10359 + }, + { + "epoch": 4.34, + "learning_rate": 9.117840169347428e-06, + "loss": 0.1327, + "step": 10360 + }, + { + "epoch": 4.34, + "learning_rate": 9.11615138292559e-06, + "loss": 0.1333, + "step": 10361 + }, + { + "epoch": 4.34, + "learning_rate": 9.114462621909196e-06, + "loss": 0.1549, + "step": 10362 + }, + { + "epoch": 4.34, + "learning_rate": 9.11277388634679e-06, + "loss": 0.2197, + "step": 10363 + }, + { + "epoch": 4.34, + "learning_rate": 9.111085176286912e-06, + "loss": 0.1698, + "step": 10364 + }, + { + "epoch": 4.34, + "learning_rate": 9.109396491778099e-06, + "loss": 0.1588, + "step": 10365 + }, + { + "epoch": 4.34, + "learning_rate": 9.107707832868896e-06, + "loss": 0.1549, + "step": 10366 + }, + { + "epoch": 4.34, + "learning_rate": 9.106019199607838e-06, + "loss": 0.1317, + "step": 10367 + }, + { + "epoch": 4.34, + "learning_rate": 9.104330592043464e-06, + "loss": 0.1306, + "step": 10368 + }, + { + "epoch": 4.34, + "learning_rate": 9.102642010224313e-06, + "loss": 0.1434, + "step": 10369 + }, + { + "epoch": 4.34, + "learning_rate": 9.100953454198924e-06, + "loss": 0.1234, + "step": 10370 + }, + { + "epoch": 4.34, + "learning_rate": 9.099264924015826e-06, + "loss": 0.1644, + "step": 10371 + }, + { + "epoch": 4.34, + "learning_rate": 9.097576419723564e-06, + "loss": 0.1702, + "step": 10372 + }, + { + "epoch": 4.34, + "learning_rate": 9.095887941370662e-06, + "loss": 0.1733, + "step": 10373 + }, + { + "epoch": 4.34, + "learning_rate": 9.094199489005664e-06, + "loss": 0.0816, + "step": 10374 + }, + { + "epoch": 4.34, + "learning_rate": 9.092511062677092e-06, + "loss": 0.1213, + "step": 10375 + }, + { + "epoch": 4.34, + "learning_rate": 9.090822662433485e-06, + "loss": 0.1588, + "step": 10376 + }, + { + "epoch": 4.34, + "learning_rate": 9.089134288323377e-06, + "loss": 0.1595, + "step": 10377 + }, + { + "epoch": 4.34, + "learning_rate": 9.087445940395292e-06, + "loss": 0.1875, + "step": 10378 + }, + { + "epoch": 4.35, + "learning_rate": 9.085757618697765e-06, + "loss": 0.1474, + "step": 10379 + }, + { + "epoch": 4.35, + "learning_rate": 9.084069323279325e-06, + "loss": 0.1392, + "step": 10380 + }, + { + "epoch": 4.35, + "learning_rate": 9.0823810541885e-06, + "loss": 0.1543, + "step": 10381 + }, + { + "epoch": 4.35, + "learning_rate": 9.080692811473815e-06, + "loss": 0.1155, + "step": 10382 + }, + { + "epoch": 4.35, + "learning_rate": 9.079004595183799e-06, + "loss": 0.1661, + "step": 10383 + }, + { + "epoch": 4.35, + "learning_rate": 9.07731640536698e-06, + "loss": 0.1738, + "step": 10384 + }, + { + "epoch": 4.35, + "learning_rate": 9.075628242071884e-06, + "loss": 0.1154, + "step": 10385 + }, + { + "epoch": 4.35, + "learning_rate": 9.073940105347032e-06, + "loss": 0.1177, + "step": 10386 + }, + { + "epoch": 4.35, + "learning_rate": 9.07225199524095e-06, + "loss": 0.1713, + "step": 10387 + }, + { + "epoch": 4.35, + "learning_rate": 9.07056391180216e-06, + "loss": 0.1511, + "step": 10388 + }, + { + "epoch": 4.35, + "learning_rate": 9.068875855079186e-06, + "loss": 0.1077, + "step": 10389 + }, + { + "epoch": 4.35, + "learning_rate": 9.067187825120552e-06, + "loss": 0.1106, + "step": 10390 + }, + { + "epoch": 4.35, + "learning_rate": 9.065499821974774e-06, + "loss": 0.1994, + "step": 10391 + }, + { + "epoch": 4.35, + "learning_rate": 9.063811845690374e-06, + "loss": 0.1749, + "step": 10392 + }, + { + "epoch": 4.35, + "learning_rate": 9.062123896315872e-06, + "loss": 0.1925, + "step": 10393 + }, + { + "epoch": 4.35, + "learning_rate": 9.06043597389979e-06, + "loss": 0.1688, + "step": 10394 + }, + { + "epoch": 4.35, + "learning_rate": 9.05874807849064e-06, + "loss": 0.1216, + "step": 10395 + }, + { + "epoch": 4.35, + "learning_rate": 9.05706021013694e-06, + "loss": 0.1676, + "step": 10396 + }, + { + "epoch": 4.35, + "learning_rate": 9.055372368887207e-06, + "loss": 0.1076, + "step": 10397 + }, + { + "epoch": 4.35, + "learning_rate": 9.053684554789959e-06, + "loss": 0.1369, + "step": 10398 + }, + { + "epoch": 4.35, + "learning_rate": 9.051996767893707e-06, + "loss": 0.1862, + "step": 10399 + }, + { + "epoch": 4.35, + "learning_rate": 9.050309008246965e-06, + "loss": 0.1327, + "step": 10400 + }, + { + "epoch": 4.35, + "learning_rate": 9.048621275898252e-06, + "loss": 0.1802, + "step": 10401 + }, + { + "epoch": 4.36, + "learning_rate": 9.046933570896073e-06, + "loss": 0.1715, + "step": 10402 + }, + { + "epoch": 4.36, + "learning_rate": 9.045245893288943e-06, + "loss": 0.1338, + "step": 10403 + }, + { + "epoch": 4.36, + "learning_rate": 9.043558243125374e-06, + "loss": 0.145, + "step": 10404 + }, + { + "epoch": 4.36, + "learning_rate": 9.041870620453874e-06, + "loss": 0.1623, + "step": 10405 + }, + { + "epoch": 4.36, + "learning_rate": 9.040183025322952e-06, + "loss": 0.199, + "step": 10406 + }, + { + "epoch": 4.36, + "learning_rate": 9.03849545778112e-06, + "loss": 0.149, + "step": 10407 + }, + { + "epoch": 4.36, + "learning_rate": 9.03680791787688e-06, + "loss": 0.1117, + "step": 10408 + }, + { + "epoch": 4.36, + "learning_rate": 9.035120405658741e-06, + "loss": 0.1546, + "step": 10409 + }, + { + "epoch": 4.36, + "learning_rate": 9.03343292117521e-06, + "loss": 0.1345, + "step": 10410 + }, + { + "epoch": 4.36, + "learning_rate": 9.031745464474794e-06, + "loss": 0.131, + "step": 10411 + }, + { + "epoch": 4.36, + "learning_rate": 9.030058035605996e-06, + "loss": 0.1347, + "step": 10412 + }, + { + "epoch": 4.36, + "learning_rate": 9.028370634617316e-06, + "loss": 0.1506, + "step": 10413 + }, + { + "epoch": 4.36, + "learning_rate": 9.026683261557263e-06, + "loss": 0.1459, + "step": 10414 + }, + { + "epoch": 4.36, + "learning_rate": 9.024995916474333e-06, + "loss": 0.1284, + "step": 10415 + }, + { + "epoch": 4.36, + "learning_rate": 9.023308599417032e-06, + "loss": 0.1399, + "step": 10416 + }, + { + "epoch": 4.36, + "learning_rate": 9.021621310433861e-06, + "loss": 0.164, + "step": 10417 + }, + { + "epoch": 4.36, + "learning_rate": 9.019934049573316e-06, + "loss": 0.1458, + "step": 10418 + }, + { + "epoch": 4.36, + "learning_rate": 9.018246816883894e-06, + "loss": 0.1446, + "step": 10419 + }, + { + "epoch": 4.36, + "learning_rate": 9.016559612414098e-06, + "loss": 0.151, + "step": 10420 + }, + { + "epoch": 4.36, + "learning_rate": 9.014872436212423e-06, + "loss": 0.1738, + "step": 10421 + }, + { + "epoch": 4.36, + "learning_rate": 9.013185288327365e-06, + "loss": 0.1546, + "step": 10422 + }, + { + "epoch": 4.36, + "learning_rate": 9.01149816880742e-06, + "loss": 0.182, + "step": 10423 + }, + { + "epoch": 4.36, + "learning_rate": 9.009811077701083e-06, + "loss": 0.1314, + "step": 10424 + }, + { + "epoch": 4.36, + "learning_rate": 9.008124015056851e-06, + "loss": 0.1467, + "step": 10425 + }, + { + "epoch": 4.37, + "learning_rate": 9.00643698092321e-06, + "loss": 0.1292, + "step": 10426 + }, + { + "epoch": 4.37, + "learning_rate": 9.004749975348657e-06, + "loss": 0.0717, + "step": 10427 + }, + { + "epoch": 4.37, + "learning_rate": 9.003062998381685e-06, + "loss": 0.1891, + "step": 10428 + }, + { + "epoch": 4.37, + "learning_rate": 9.00137605007078e-06, + "loss": 0.1756, + "step": 10429 + }, + { + "epoch": 4.37, + "learning_rate": 8.999689130464436e-06, + "loss": 0.1271, + "step": 10430 + }, + { + "epoch": 4.37, + "learning_rate": 8.99800223961114e-06, + "loss": 0.156, + "step": 10431 + }, + { + "epoch": 4.37, + "learning_rate": 8.99631537755938e-06, + "loss": 0.1119, + "step": 10432 + }, + { + "epoch": 4.37, + "learning_rate": 8.994628544357641e-06, + "loss": 0.1482, + "step": 10433 + }, + { + "epoch": 4.37, + "learning_rate": 8.992941740054418e-06, + "loss": 0.1023, + "step": 10434 + }, + { + "epoch": 4.37, + "learning_rate": 8.991254964698188e-06, + "loss": 0.1573, + "step": 10435 + }, + { + "epoch": 4.37, + "learning_rate": 8.989568218337438e-06, + "loss": 0.1465, + "step": 10436 + }, + { + "epoch": 4.37, + "learning_rate": 8.987881501020656e-06, + "loss": 0.155, + "step": 10437 + }, + { + "epoch": 4.37, + "learning_rate": 8.98619481279632e-06, + "loss": 0.1424, + "step": 10438 + }, + { + "epoch": 4.37, + "learning_rate": 8.984508153712912e-06, + "loss": 0.0995, + "step": 10439 + }, + { + "epoch": 4.37, + "learning_rate": 8.982821523818923e-06, + "loss": 0.1584, + "step": 10440 + }, + { + "epoch": 4.37, + "learning_rate": 8.981134923162823e-06, + "loss": 0.1327, + "step": 10441 + }, + { + "epoch": 4.37, + "learning_rate": 8.979448351793095e-06, + "loss": 0.1538, + "step": 10442 + }, + { + "epoch": 4.37, + "learning_rate": 8.977761809758218e-06, + "loss": 0.1266, + "step": 10443 + }, + { + "epoch": 4.37, + "learning_rate": 8.97607529710667e-06, + "loss": 0.1584, + "step": 10444 + }, + { + "epoch": 4.37, + "learning_rate": 8.974388813886932e-06, + "loss": 0.1403, + "step": 10445 + }, + { + "epoch": 4.37, + "learning_rate": 8.972702360147474e-06, + "loss": 0.1293, + "step": 10446 + }, + { + "epoch": 4.37, + "learning_rate": 8.971015935936777e-06, + "loss": 0.142, + "step": 10447 + }, + { + "epoch": 4.37, + "learning_rate": 8.969329541303315e-06, + "loss": 0.1632, + "step": 10448 + }, + { + "epoch": 4.37, + "learning_rate": 8.967643176295557e-06, + "loss": 0.1862, + "step": 10449 + }, + { + "epoch": 4.38, + "learning_rate": 8.96595684096198e-06, + "loss": 0.1808, + "step": 10450 + }, + { + "epoch": 4.38, + "learning_rate": 8.964270535351058e-06, + "loss": 0.1826, + "step": 10451 + }, + { + "epoch": 4.38, + "learning_rate": 8.962584259511262e-06, + "loss": 0.1768, + "step": 10452 + }, + { + "epoch": 4.38, + "learning_rate": 8.960898013491058e-06, + "loss": 0.1202, + "step": 10453 + }, + { + "epoch": 4.38, + "learning_rate": 8.959211797338919e-06, + "loss": 0.1497, + "step": 10454 + }, + { + "epoch": 4.38, + "learning_rate": 8.957525611103316e-06, + "loss": 0.1282, + "step": 10455 + }, + { + "epoch": 4.38, + "learning_rate": 8.955839454832711e-06, + "loss": 0.148, + "step": 10456 + }, + { + "epoch": 4.38, + "learning_rate": 8.954153328575574e-06, + "loss": 0.1241, + "step": 10457 + }, + { + "epoch": 4.38, + "learning_rate": 8.952467232380375e-06, + "loss": 0.1631, + "step": 10458 + }, + { + "epoch": 4.38, + "learning_rate": 8.950781166295573e-06, + "loss": 0.1293, + "step": 10459 + }, + { + "epoch": 4.38, + "learning_rate": 8.949095130369636e-06, + "loss": 0.1547, + "step": 10460 + }, + { + "epoch": 4.38, + "learning_rate": 8.947409124651027e-06, + "loss": 0.1538, + "step": 10461 + }, + { + "epoch": 4.38, + "learning_rate": 8.94572314918821e-06, + "loss": 0.1362, + "step": 10462 + }, + { + "epoch": 4.38, + "learning_rate": 8.944037204029646e-06, + "loss": 0.1879, + "step": 10463 + }, + { + "epoch": 4.38, + "learning_rate": 8.942351289223795e-06, + "loss": 0.1804, + "step": 10464 + }, + { + "epoch": 4.38, + "learning_rate": 8.940665404819114e-06, + "loss": 0.1195, + "step": 10465 + }, + { + "epoch": 4.38, + "learning_rate": 8.93897955086407e-06, + "loss": 0.1211, + "step": 10466 + }, + { + "epoch": 4.38, + "learning_rate": 8.937293727407115e-06, + "loss": 0.1555, + "step": 10467 + }, + { + "epoch": 4.38, + "learning_rate": 8.935607934496708e-06, + "loss": 0.1263, + "step": 10468 + }, + { + "epoch": 4.38, + "learning_rate": 8.93392217218131e-06, + "loss": 0.184, + "step": 10469 + }, + { + "epoch": 4.38, + "learning_rate": 8.93223644050937e-06, + "loss": 0.1298, + "step": 10470 + }, + { + "epoch": 4.38, + "learning_rate": 8.930550739529345e-06, + "loss": 0.13, + "step": 10471 + }, + { + "epoch": 4.38, + "learning_rate": 8.928865069289693e-06, + "loss": 0.134, + "step": 10472 + }, + { + "epoch": 4.38, + "learning_rate": 8.927179429838864e-06, + "loss": 0.145, + "step": 10473 + }, + { + "epoch": 4.39, + "learning_rate": 8.925493821225309e-06, + "loss": 0.1614, + "step": 10474 + }, + { + "epoch": 4.39, + "learning_rate": 8.923808243497482e-06, + "loss": 0.1361, + "step": 10475 + }, + { + "epoch": 4.39, + "learning_rate": 8.92212269670383e-06, + "loss": 0.1655, + "step": 10476 + }, + { + "epoch": 4.39, + "learning_rate": 8.920437180892803e-06, + "loss": 0.1206, + "step": 10477 + }, + { + "epoch": 4.39, + "learning_rate": 8.918751696112856e-06, + "loss": 0.1356, + "step": 10478 + }, + { + "epoch": 4.39, + "learning_rate": 8.917066242412429e-06, + "loss": 0.1348, + "step": 10479 + }, + { + "epoch": 4.39, + "learning_rate": 8.915380819839972e-06, + "loss": 0.1483, + "step": 10480 + }, + { + "epoch": 4.39, + "learning_rate": 8.91369542844393e-06, + "loss": 0.1389, + "step": 10481 + }, + { + "epoch": 4.39, + "learning_rate": 8.912010068272752e-06, + "loss": 0.1384, + "step": 10482 + }, + { + "epoch": 4.39, + "learning_rate": 8.910324739374877e-06, + "loss": 0.1575, + "step": 10483 + }, + { + "epoch": 4.39, + "learning_rate": 8.908639441798751e-06, + "loss": 0.1206, + "step": 10484 + }, + { + "epoch": 4.39, + "learning_rate": 8.906954175592819e-06, + "loss": 0.1597, + "step": 10485 + }, + { + "epoch": 4.39, + "learning_rate": 8.905268940805519e-06, + "loss": 0.1651, + "step": 10486 + }, + { + "epoch": 4.39, + "learning_rate": 8.903583737485286e-06, + "loss": 0.1341, + "step": 10487 + }, + { + "epoch": 4.39, + "learning_rate": 8.901898565680567e-06, + "loss": 0.1445, + "step": 10488 + }, + { + "epoch": 4.39, + "learning_rate": 8.900213425439803e-06, + "loss": 0.1532, + "step": 10489 + }, + { + "epoch": 4.39, + "learning_rate": 8.898528316811425e-06, + "loss": 0.1891, + "step": 10490 + }, + { + "epoch": 4.39, + "learning_rate": 8.896843239843873e-06, + "loss": 0.1499, + "step": 10491 + }, + { + "epoch": 4.39, + "learning_rate": 8.895158194585583e-06, + "loss": 0.1366, + "step": 10492 + }, + { + "epoch": 4.39, + "learning_rate": 8.893473181084993e-06, + "loss": 0.1487, + "step": 10493 + }, + { + "epoch": 4.39, + "learning_rate": 8.891788199390533e-06, + "loss": 0.1514, + "step": 10494 + }, + { + "epoch": 4.39, + "learning_rate": 8.890103249550638e-06, + "loss": 0.1496, + "step": 10495 + }, + { + "epoch": 4.39, + "learning_rate": 8.88841833161374e-06, + "loss": 0.1665, + "step": 10496 + }, + { + "epoch": 4.39, + "learning_rate": 8.88673344562827e-06, + "loss": 0.1539, + "step": 10497 + }, + { + "epoch": 4.4, + "learning_rate": 8.88504859164266e-06, + "loss": 0.1255, + "step": 10498 + }, + { + "epoch": 4.4, + "learning_rate": 8.883363769705338e-06, + "loss": 0.1813, + "step": 10499 + }, + { + "epoch": 4.4, + "learning_rate": 8.881678979864734e-06, + "loss": 0.1768, + "step": 10500 + }, + { + "epoch": 4.4, + "learning_rate": 8.879994222169275e-06, + "loss": 0.1486, + "step": 10501 + }, + { + "epoch": 4.4, + "learning_rate": 8.87830949666739e-06, + "loss": 0.1229, + "step": 10502 + }, + { + "epoch": 4.4, + "learning_rate": 8.876624803407501e-06, + "loss": 0.1755, + "step": 10503 + }, + { + "epoch": 4.4, + "learning_rate": 8.874940142438036e-06, + "loss": 0.1787, + "step": 10504 + }, + { + "epoch": 4.4, + "learning_rate": 8.873255513807418e-06, + "loss": 0.1251, + "step": 10505 + }, + { + "epoch": 4.4, + "learning_rate": 8.87157091756407e-06, + "loss": 0.1068, + "step": 10506 + }, + { + "epoch": 4.4, + "learning_rate": 8.869886353756415e-06, + "loss": 0.1293, + "step": 10507 + }, + { + "epoch": 4.4, + "learning_rate": 8.868201822432878e-06, + "loss": 0.1115, + "step": 10508 + }, + { + "epoch": 4.4, + "learning_rate": 8.86651732364187e-06, + "loss": 0.1715, + "step": 10509 + }, + { + "epoch": 4.4, + "learning_rate": 8.86483285743182e-06, + "loss": 0.1266, + "step": 10510 + }, + { + "epoch": 4.4, + "learning_rate": 8.863148423851138e-06, + "loss": 0.1839, + "step": 10511 + }, + { + "epoch": 4.4, + "learning_rate": 8.861464022948247e-06, + "loss": 0.1993, + "step": 10512 + }, + { + "epoch": 4.4, + "learning_rate": 8.859779654771563e-06, + "loss": 0.163, + "step": 10513 + }, + { + "epoch": 4.4, + "learning_rate": 8.858095319369499e-06, + "loss": 0.1587, + "step": 10514 + }, + { + "epoch": 4.4, + "learning_rate": 8.856411016790472e-06, + "loss": 0.1455, + "step": 10515 + }, + { + "epoch": 4.4, + "learning_rate": 8.8547267470829e-06, + "loss": 0.1502, + "step": 10516 + }, + { + "epoch": 4.4, + "learning_rate": 8.853042510295187e-06, + "loss": 0.1293, + "step": 10517 + }, + { + "epoch": 4.4, + "learning_rate": 8.851358306475748e-06, + "loss": 0.1647, + "step": 10518 + }, + { + "epoch": 4.4, + "learning_rate": 8.849674135672999e-06, + "loss": 0.1932, + "step": 10519 + }, + { + "epoch": 4.4, + "learning_rate": 8.847989997935345e-06, + "loss": 0.1486, + "step": 10520 + }, + { + "epoch": 4.4, + "learning_rate": 8.846305893311194e-06, + "loss": 0.1692, + "step": 10521 + }, + { + "epoch": 4.41, + "learning_rate": 8.844621821848958e-06, + "loss": 0.1121, + "step": 10522 + }, + { + "epoch": 4.41, + "learning_rate": 8.842937783597045e-06, + "loss": 0.1358, + "step": 10523 + }, + { + "epoch": 4.41, + "learning_rate": 8.841253778603855e-06, + "loss": 0.1654, + "step": 10524 + }, + { + "epoch": 4.41, + "learning_rate": 8.839569806917797e-06, + "loss": 0.1473, + "step": 10525 + }, + { + "epoch": 4.41, + "learning_rate": 8.837885868587279e-06, + "loss": 0.1257, + "step": 10526 + }, + { + "epoch": 4.41, + "learning_rate": 8.836201963660695e-06, + "loss": 0.1375, + "step": 10527 + }, + { + "epoch": 4.41, + "learning_rate": 8.834518092186456e-06, + "loss": 0.1417, + "step": 10528 + }, + { + "epoch": 4.41, + "learning_rate": 8.83283425421296e-06, + "loss": 0.139, + "step": 10529 + }, + { + "epoch": 4.41, + "learning_rate": 8.831150449788613e-06, + "loss": 0.1568, + "step": 10530 + }, + { + "epoch": 4.41, + "learning_rate": 8.829466678961803e-06, + "loss": 0.189, + "step": 10531 + }, + { + "epoch": 4.41, + "learning_rate": 8.827782941780934e-06, + "loss": 0.1277, + "step": 10532 + }, + { + "epoch": 4.41, + "learning_rate": 8.826099238294404e-06, + "loss": 0.1296, + "step": 10533 + }, + { + "epoch": 4.41, + "learning_rate": 8.824415568550612e-06, + "loss": 0.1962, + "step": 10534 + }, + { + "epoch": 4.41, + "learning_rate": 8.82273193259795e-06, + "loss": 0.1299, + "step": 10535 + }, + { + "epoch": 4.41, + "learning_rate": 8.821048330484814e-06, + "loss": 0.0992, + "step": 10536 + }, + { + "epoch": 4.41, + "learning_rate": 8.819364762259598e-06, + "loss": 0.1378, + "step": 10537 + }, + { + "epoch": 4.41, + "learning_rate": 8.817681227970693e-06, + "loss": 0.1181, + "step": 10538 + }, + { + "epoch": 4.41, + "learning_rate": 8.815997727666493e-06, + "loss": 0.1476, + "step": 10539 + }, + { + "epoch": 4.41, + "learning_rate": 8.814314261395389e-06, + "loss": 0.1612, + "step": 10540 + }, + { + "epoch": 4.41, + "learning_rate": 8.812630829205766e-06, + "loss": 0.1735, + "step": 10541 + }, + { + "epoch": 4.41, + "learning_rate": 8.810947431146019e-06, + "loss": 0.1622, + "step": 10542 + }, + { + "epoch": 4.41, + "learning_rate": 8.809264067264533e-06, + "loss": 0.1423, + "step": 10543 + }, + { + "epoch": 4.41, + "learning_rate": 8.807580737609693e-06, + "loss": 0.151, + "step": 10544 + }, + { + "epoch": 4.41, + "learning_rate": 8.805897442229888e-06, + "loss": 0.1505, + "step": 10545 + }, + { + "epoch": 4.42, + "learning_rate": 8.804214181173501e-06, + "loss": 0.1616, + "step": 10546 + }, + { + "epoch": 4.42, + "learning_rate": 8.802530954488915e-06, + "loss": 0.1944, + "step": 10547 + }, + { + "epoch": 4.42, + "learning_rate": 8.800847762224514e-06, + "loss": 0.1288, + "step": 10548 + }, + { + "epoch": 4.42, + "learning_rate": 8.799164604428683e-06, + "loss": 0.123, + "step": 10549 + }, + { + "epoch": 4.42, + "learning_rate": 8.797481481149798e-06, + "loss": 0.1643, + "step": 10550 + }, + { + "epoch": 4.42, + "learning_rate": 8.795798392436239e-06, + "loss": 0.1086, + "step": 10551 + }, + { + "epoch": 4.42, + "learning_rate": 8.794115338336388e-06, + "loss": 0.1179, + "step": 10552 + }, + { + "epoch": 4.42, + "learning_rate": 8.792432318898624e-06, + "loss": 0.1645, + "step": 10553 + }, + { + "epoch": 4.42, + "learning_rate": 8.790749334171323e-06, + "loss": 0.1417, + "step": 10554 + }, + { + "epoch": 4.42, + "learning_rate": 8.789066384202853e-06, + "loss": 0.1391, + "step": 10555 + }, + { + "epoch": 4.42, + "learning_rate": 8.787383469041598e-06, + "loss": 0.1525, + "step": 10556 + }, + { + "epoch": 4.42, + "learning_rate": 8.78570058873593e-06, + "loss": 0.1494, + "step": 10557 + }, + { + "epoch": 4.42, + "learning_rate": 8.78401774333422e-06, + "loss": 0.1371, + "step": 10558 + }, + { + "epoch": 4.42, + "learning_rate": 8.78233493288484e-06, + "loss": 0.1782, + "step": 10559 + }, + { + "epoch": 4.42, + "learning_rate": 8.780652157436163e-06, + "loss": 0.1544, + "step": 10560 + }, + { + "epoch": 4.42, + "learning_rate": 8.778969417036557e-06, + "loss": 0.1573, + "step": 10561 + }, + { + "epoch": 4.42, + "learning_rate": 8.77728671173439e-06, + "loss": 0.1482, + "step": 10562 + }, + { + "epoch": 4.42, + "learning_rate": 8.775604041578034e-06, + "loss": 0.1629, + "step": 10563 + }, + { + "epoch": 4.42, + "learning_rate": 8.773921406615854e-06, + "loss": 0.1126, + "step": 10564 + }, + { + "epoch": 4.42, + "learning_rate": 8.772238806896213e-06, + "loss": 0.1316, + "step": 10565 + }, + { + "epoch": 4.42, + "learning_rate": 8.770556242467479e-06, + "loss": 0.1752, + "step": 10566 + }, + { + "epoch": 4.42, + "learning_rate": 8.768873713378015e-06, + "loss": 0.1287, + "step": 10567 + }, + { + "epoch": 4.42, + "learning_rate": 8.767191219676184e-06, + "loss": 0.1922, + "step": 10568 + }, + { + "epoch": 4.42, + "learning_rate": 8.765508761410345e-06, + "loss": 0.1686, + "step": 10569 + }, + { + "epoch": 4.43, + "learning_rate": 8.763826338628865e-06, + "loss": 0.1586, + "step": 10570 + }, + { + "epoch": 4.43, + "learning_rate": 8.762143951380096e-06, + "loss": 0.168, + "step": 10571 + }, + { + "epoch": 4.43, + "learning_rate": 8.760461599712402e-06, + "loss": 0.1332, + "step": 10572 + }, + { + "epoch": 4.43, + "learning_rate": 8.758779283674142e-06, + "loss": 0.1216, + "step": 10573 + }, + { + "epoch": 4.43, + "learning_rate": 8.757097003313665e-06, + "loss": 0.1546, + "step": 10574 + }, + { + "epoch": 4.43, + "learning_rate": 8.755414758679336e-06, + "loss": 0.1556, + "step": 10575 + }, + { + "epoch": 4.43, + "learning_rate": 8.753732549819508e-06, + "loss": 0.1918, + "step": 10576 + }, + { + "epoch": 4.43, + "learning_rate": 8.752050376782525e-06, + "loss": 0.141, + "step": 10577 + }, + { + "epoch": 4.43, + "learning_rate": 8.750368239616752e-06, + "loss": 0.1161, + "step": 10578 + }, + { + "epoch": 4.43, + "learning_rate": 8.748686138370533e-06, + "loss": 0.1649, + "step": 10579 + }, + { + "epoch": 4.43, + "learning_rate": 8.747004073092218e-06, + "loss": 0.1614, + "step": 10580 + }, + { + "epoch": 4.43, + "learning_rate": 8.745322043830163e-06, + "loss": 0.1292, + "step": 10581 + }, + { + "epoch": 4.43, + "learning_rate": 8.743640050632709e-06, + "loss": 0.1331, + "step": 10582 + }, + { + "epoch": 4.43, + "learning_rate": 8.741958093548208e-06, + "loss": 0.166, + "step": 10583 + }, + { + "epoch": 4.43, + "learning_rate": 8.740276172625007e-06, + "loss": 0.1588, + "step": 10584 + }, + { + "epoch": 4.43, + "learning_rate": 8.738594287911447e-06, + "loss": 0.1624, + "step": 10585 + }, + { + "epoch": 4.43, + "learning_rate": 8.736912439455876e-06, + "loss": 0.1911, + "step": 10586 + }, + { + "epoch": 4.43, + "learning_rate": 8.735230627306637e-06, + "loss": 0.1964, + "step": 10587 + }, + { + "epoch": 4.43, + "learning_rate": 8.733548851512071e-06, + "loss": 0.1723, + "step": 10588 + }, + { + "epoch": 4.43, + "learning_rate": 8.731867112120516e-06, + "loss": 0.1644, + "step": 10589 + }, + { + "epoch": 4.43, + "learning_rate": 8.730185409180321e-06, + "loss": 0.1857, + "step": 10590 + }, + { + "epoch": 4.43, + "learning_rate": 8.728503742739816e-06, + "loss": 0.1473, + "step": 10591 + }, + { + "epoch": 4.43, + "learning_rate": 8.726822112847343e-06, + "loss": 0.1511, + "step": 10592 + }, + { + "epoch": 4.44, + "learning_rate": 8.725140519551238e-06, + "loss": 0.1551, + "step": 10593 + }, + { + "epoch": 4.44, + "learning_rate": 8.72345896289984e-06, + "loss": 0.1422, + "step": 10594 + }, + { + "epoch": 4.44, + "learning_rate": 8.721777442941479e-06, + "loss": 0.145, + "step": 10595 + }, + { + "epoch": 4.44, + "learning_rate": 8.720095959724492e-06, + "loss": 0.1595, + "step": 10596 + }, + { + "epoch": 4.44, + "learning_rate": 8.718414513297214e-06, + "loss": 0.1792, + "step": 10597 + }, + { + "epoch": 4.44, + "learning_rate": 8.716733103707972e-06, + "loss": 0.1171, + "step": 10598 + }, + { + "epoch": 4.44, + "learning_rate": 8.715051731005097e-06, + "loss": 0.1347, + "step": 10599 + }, + { + "epoch": 4.44, + "learning_rate": 8.713370395236918e-06, + "loss": 0.1383, + "step": 10600 + }, + { + "epoch": 4.44, + "learning_rate": 8.711689096451769e-06, + "loss": 0.1394, + "step": 10601 + }, + { + "epoch": 4.44, + "learning_rate": 8.71000783469797e-06, + "loss": 0.1451, + "step": 10602 + }, + { + "epoch": 4.44, + "learning_rate": 8.708326610023852e-06, + "loss": 0.1374, + "step": 10603 + }, + { + "epoch": 4.44, + "learning_rate": 8.706645422477739e-06, + "loss": 0.1479, + "step": 10604 + }, + { + "epoch": 4.44, + "learning_rate": 8.704964272107957e-06, + "loss": 0.1753, + "step": 10605 + }, + { + "epoch": 4.44, + "learning_rate": 8.703283158962828e-06, + "loss": 0.1383, + "step": 10606 + }, + { + "epoch": 4.44, + "learning_rate": 8.701602083090673e-06, + "loss": 0.1204, + "step": 10607 + }, + { + "epoch": 4.44, + "learning_rate": 8.699921044539815e-06, + "loss": 0.1408, + "step": 10608 + }, + { + "epoch": 4.44, + "learning_rate": 8.69824004335857e-06, + "loss": 0.1552, + "step": 10609 + }, + { + "epoch": 4.44, + "learning_rate": 8.696559079595261e-06, + "loss": 0.1682, + "step": 10610 + }, + { + "epoch": 4.44, + "learning_rate": 8.694878153298207e-06, + "loss": 0.1135, + "step": 10611 + }, + { + "epoch": 4.44, + "learning_rate": 8.69319726451572e-06, + "loss": 0.1273, + "step": 10612 + }, + { + "epoch": 4.44, + "learning_rate": 8.691516413296118e-06, + "loss": 0.0887, + "step": 10613 + }, + { + "epoch": 4.44, + "learning_rate": 8.689835599687717e-06, + "loss": 0.136, + "step": 10614 + }, + { + "epoch": 4.44, + "learning_rate": 8.68815482373883e-06, + "loss": 0.1311, + "step": 10615 + }, + { + "epoch": 4.44, + "learning_rate": 8.686474085497765e-06, + "loss": 0.1339, + "step": 10616 + }, + { + "epoch": 4.45, + "learning_rate": 8.68479338501284e-06, + "loss": 0.1429, + "step": 10617 + }, + { + "epoch": 4.45, + "learning_rate": 8.683112722332361e-06, + "loss": 0.1564, + "step": 10618 + }, + { + "epoch": 4.45, + "learning_rate": 8.681432097504636e-06, + "loss": 0.1621, + "step": 10619 + }, + { + "epoch": 4.45, + "learning_rate": 8.67975151057798e-06, + "loss": 0.1507, + "step": 10620 + }, + { + "epoch": 4.45, + "learning_rate": 8.678070961600693e-06, + "loss": 0.1407, + "step": 10621 + }, + { + "epoch": 4.45, + "learning_rate": 8.676390450621084e-06, + "loss": 0.1741, + "step": 10622 + }, + { + "epoch": 4.45, + "learning_rate": 8.674709977687456e-06, + "loss": 0.1286, + "step": 10623 + }, + { + "epoch": 4.45, + "learning_rate": 8.67302954284811e-06, + "loss": 0.1865, + "step": 10624 + }, + { + "epoch": 4.45, + "learning_rate": 8.67134914615136e-06, + "loss": 0.1679, + "step": 10625 + }, + { + "epoch": 4.45, + "learning_rate": 8.669668787645494e-06, + "loss": 0.1714, + "step": 10626 + }, + { + "epoch": 4.45, + "learning_rate": 8.667988467378819e-06, + "loss": 0.1482, + "step": 10627 + }, + { + "epoch": 4.45, + "learning_rate": 8.666308185399634e-06, + "loss": 0.2011, + "step": 10628 + }, + { + "epoch": 4.45, + "learning_rate": 8.664627941756236e-06, + "loss": 0.1634, + "step": 10629 + }, + { + "epoch": 4.45, + "learning_rate": 8.662947736496923e-06, + "loss": 0.1275, + "step": 10630 + }, + { + "epoch": 4.45, + "learning_rate": 8.661267569669994e-06, + "loss": 0.1586, + "step": 10631 + }, + { + "epoch": 4.45, + "learning_rate": 8.659587441323736e-06, + "loss": 0.16, + "step": 10632 + }, + { + "epoch": 4.45, + "learning_rate": 8.65790735150645e-06, + "loss": 0.1889, + "step": 10633 + }, + { + "epoch": 4.45, + "learning_rate": 8.656227300266425e-06, + "loss": 0.2088, + "step": 10634 + }, + { + "epoch": 4.45, + "learning_rate": 8.654547287651955e-06, + "loss": 0.1768, + "step": 10635 + }, + { + "epoch": 4.45, + "learning_rate": 8.652867313711328e-06, + "loss": 0.141, + "step": 10636 + }, + { + "epoch": 4.45, + "learning_rate": 8.651187378492836e-06, + "loss": 0.1424, + "step": 10637 + }, + { + "epoch": 4.45, + "learning_rate": 8.649507482044766e-06, + "loss": 0.1509, + "step": 10638 + }, + { + "epoch": 4.45, + "learning_rate": 8.647827624415403e-06, + "loss": 0.176, + "step": 10639 + }, + { + "epoch": 4.45, + "learning_rate": 8.646147805653037e-06, + "loss": 0.1154, + "step": 10640 + }, + { + "epoch": 4.46, + "learning_rate": 8.644468025805952e-06, + "loss": 0.1167, + "step": 10641 + }, + { + "epoch": 4.46, + "learning_rate": 8.642788284922427e-06, + "loss": 0.149, + "step": 10642 + }, + { + "epoch": 4.46, + "learning_rate": 8.641108583050751e-06, + "loss": 0.19, + "step": 10643 + }, + { + "epoch": 4.46, + "learning_rate": 8.639428920239205e-06, + "loss": 0.1419, + "step": 10644 + }, + { + "epoch": 4.46, + "learning_rate": 8.637749296536065e-06, + "loss": 0.2231, + "step": 10645 + }, + { + "epoch": 4.46, + "learning_rate": 8.636069711989614e-06, + "loss": 0.1218, + "step": 10646 + }, + { + "epoch": 4.46, + "learning_rate": 8.634390166648126e-06, + "loss": 0.1739, + "step": 10647 + }, + { + "epoch": 4.46, + "learning_rate": 8.632710660559883e-06, + "loss": 0.1368, + "step": 10648 + }, + { + "epoch": 4.46, + "learning_rate": 8.631031193773158e-06, + "loss": 0.1488, + "step": 10649 + }, + { + "epoch": 4.46, + "learning_rate": 8.629351766336225e-06, + "loss": 0.1991, + "step": 10650 + }, + { + "epoch": 4.46, + "learning_rate": 8.627672378297361e-06, + "loss": 0.154, + "step": 10651 + }, + { + "epoch": 4.46, + "learning_rate": 8.625993029704838e-06, + "loss": 0.1503, + "step": 10652 + }, + { + "epoch": 4.46, + "learning_rate": 8.624313720606922e-06, + "loss": 0.1851, + "step": 10653 + }, + { + "epoch": 4.46, + "learning_rate": 8.62263445105189e-06, + "loss": 0.1845, + "step": 10654 + }, + { + "epoch": 4.46, + "learning_rate": 8.62095522108801e-06, + "loss": 0.1492, + "step": 10655 + }, + { + "epoch": 4.46, + "learning_rate": 8.619276030763545e-06, + "loss": 0.218, + "step": 10656 + }, + { + "epoch": 4.46, + "learning_rate": 8.617596880126766e-06, + "loss": 0.1528, + "step": 10657 + }, + { + "epoch": 4.46, + "learning_rate": 8.61591776922594e-06, + "loss": 0.1055, + "step": 10658 + }, + { + "epoch": 4.46, + "learning_rate": 8.614238698109329e-06, + "loss": 0.1446, + "step": 10659 + }, + { + "epoch": 4.46, + "learning_rate": 8.612559666825196e-06, + "loss": 0.1775, + "step": 10660 + }, + { + "epoch": 4.46, + "learning_rate": 8.610880675421805e-06, + "loss": 0.1524, + "step": 10661 + }, + { + "epoch": 4.46, + "learning_rate": 8.609201723947415e-06, + "loss": 0.1336, + "step": 10662 + }, + { + "epoch": 4.46, + "learning_rate": 8.607522812450288e-06, + "loss": 0.1588, + "step": 10663 + }, + { + "epoch": 4.46, + "learning_rate": 8.60584394097868e-06, + "loss": 0.1851, + "step": 10664 + }, + { + "epoch": 4.47, + "learning_rate": 8.604165109580855e-06, + "loss": 0.1002, + "step": 10665 + }, + { + "epoch": 4.47, + "learning_rate": 8.602486318305067e-06, + "loss": 0.172, + "step": 10666 + }, + { + "epoch": 4.47, + "learning_rate": 8.600807567199565e-06, + "loss": 0.1638, + "step": 10667 + }, + { + "epoch": 4.47, + "learning_rate": 8.599128856312607e-06, + "loss": 0.1558, + "step": 10668 + }, + { + "epoch": 4.47, + "learning_rate": 8.597450185692451e-06, + "loss": 0.1578, + "step": 10669 + }, + { + "epoch": 4.47, + "learning_rate": 8.59577155538734e-06, + "loss": 0.1578, + "step": 10670 + }, + { + "epoch": 4.47, + "learning_rate": 8.594092965445534e-06, + "loss": 0.1351, + "step": 10671 + }, + { + "epoch": 4.47, + "learning_rate": 8.592414415915275e-06, + "loss": 0.1154, + "step": 10672 + }, + { + "epoch": 4.47, + "learning_rate": 8.590735906844816e-06, + "loss": 0.1396, + "step": 10673 + }, + { + "epoch": 4.47, + "learning_rate": 8.5890574382824e-06, + "loss": 0.1288, + "step": 10674 + }, + { + "epoch": 4.47, + "learning_rate": 8.58737901027628e-06, + "loss": 0.154, + "step": 10675 + }, + { + "epoch": 4.47, + "learning_rate": 8.585700622874695e-06, + "loss": 0.1194, + "step": 10676 + }, + { + "epoch": 4.47, + "learning_rate": 8.584022276125891e-06, + "loss": 0.1551, + "step": 10677 + }, + { + "epoch": 4.47, + "learning_rate": 8.582343970078108e-06, + "loss": 0.1489, + "step": 10678 + }, + { + "epoch": 4.47, + "learning_rate": 8.580665704779593e-06, + "loss": 0.1997, + "step": 10679 + }, + { + "epoch": 4.47, + "learning_rate": 8.57898748027858e-06, + "loss": 0.1953, + "step": 10680 + }, + { + "epoch": 4.47, + "learning_rate": 8.577309296623311e-06, + "loss": 0.1389, + "step": 10681 + }, + { + "epoch": 4.47, + "learning_rate": 8.575631153862027e-06, + "loss": 0.1535, + "step": 10682 + }, + { + "epoch": 4.47, + "learning_rate": 8.573953052042959e-06, + "loss": 0.147, + "step": 10683 + }, + { + "epoch": 4.47, + "learning_rate": 8.572274991214345e-06, + "loss": 0.1321, + "step": 10684 + }, + { + "epoch": 4.47, + "learning_rate": 8.570596971424421e-06, + "loss": 0.1436, + "step": 10685 + }, + { + "epoch": 4.47, + "learning_rate": 8.568918992721418e-06, + "loss": 0.121, + "step": 10686 + }, + { + "epoch": 4.47, + "learning_rate": 8.567241055153567e-06, + "loss": 0.137, + "step": 10687 + }, + { + "epoch": 4.47, + "learning_rate": 8.565563158769103e-06, + "loss": 0.1576, + "step": 10688 + }, + { + "epoch": 4.48, + "learning_rate": 8.563885303616253e-06, + "loss": 0.1183, + "step": 10689 + }, + { + "epoch": 4.48, + "learning_rate": 8.562207489743245e-06, + "loss": 0.1324, + "step": 10690 + }, + { + "epoch": 4.48, + "learning_rate": 8.560529717198305e-06, + "loss": 0.1612, + "step": 10691 + }, + { + "epoch": 4.48, + "learning_rate": 8.558851986029663e-06, + "loss": 0.1455, + "step": 10692 + }, + { + "epoch": 4.48, + "learning_rate": 8.557174296285542e-06, + "loss": 0.1424, + "step": 10693 + }, + { + "epoch": 4.48, + "learning_rate": 8.555496648014163e-06, + "loss": 0.1679, + "step": 10694 + }, + { + "epoch": 4.48, + "learning_rate": 8.553819041263752e-06, + "loss": 0.15, + "step": 10695 + }, + { + "epoch": 4.48, + "learning_rate": 8.552141476082531e-06, + "loss": 0.1507, + "step": 10696 + }, + { + "epoch": 4.48, + "learning_rate": 8.550463952518716e-06, + "loss": 0.1299, + "step": 10697 + }, + { + "epoch": 4.48, + "learning_rate": 8.54878647062053e-06, + "loss": 0.1896, + "step": 10698 + }, + { + "epoch": 4.48, + "learning_rate": 8.547109030436189e-06, + "loss": 0.1801, + "step": 10699 + }, + { + "epoch": 4.48, + "learning_rate": 8.545431632013909e-06, + "loss": 0.1252, + "step": 10700 + }, + { + "epoch": 4.48, + "learning_rate": 8.543754275401906e-06, + "loss": 0.1409, + "step": 10701 + }, + { + "epoch": 4.48, + "learning_rate": 8.542076960648395e-06, + "loss": 0.1873, + "step": 10702 + }, + { + "epoch": 4.48, + "learning_rate": 8.540399687801588e-06, + "loss": 0.1732, + "step": 10703 + }, + { + "epoch": 4.48, + "learning_rate": 8.538722456909694e-06, + "loss": 0.1139, + "step": 10704 + }, + { + "epoch": 4.48, + "learning_rate": 8.53704526802093e-06, + "loss": 0.157, + "step": 10705 + }, + { + "epoch": 4.48, + "learning_rate": 8.5353681211835e-06, + "loss": 0.1486, + "step": 10706 + }, + { + "epoch": 4.48, + "learning_rate": 8.533691016445614e-06, + "loss": 0.1798, + "step": 10707 + }, + { + "epoch": 4.48, + "learning_rate": 8.532013953855477e-06, + "loss": 0.1528, + "step": 10708 + }, + { + "epoch": 4.48, + "learning_rate": 8.530336933461298e-06, + "loss": 0.1536, + "step": 10709 + }, + { + "epoch": 4.48, + "learning_rate": 8.52865995531128e-06, + "loss": 0.1856, + "step": 10710 + }, + { + "epoch": 4.48, + "learning_rate": 8.526983019453624e-06, + "loss": 0.1538, + "step": 10711 + }, + { + "epoch": 4.48, + "learning_rate": 8.52530612593654e-06, + "loss": 0.1494, + "step": 10712 + }, + { + "epoch": 4.49, + "learning_rate": 8.523629274808219e-06, + "loss": 0.1811, + "step": 10713 + }, + { + "epoch": 4.49, + "learning_rate": 8.521952466116862e-06, + "loss": 0.2089, + "step": 10714 + }, + { + "epoch": 4.49, + "learning_rate": 8.52027569991067e-06, + "loss": 0.1166, + "step": 10715 + }, + { + "epoch": 4.49, + "learning_rate": 8.51859897623784e-06, + "loss": 0.1852, + "step": 10716 + }, + { + "epoch": 4.49, + "learning_rate": 8.51692229514657e-06, + "loss": 0.1098, + "step": 10717 + }, + { + "epoch": 4.49, + "learning_rate": 8.515245656685051e-06, + "loss": 0.1062, + "step": 10718 + }, + { + "epoch": 4.49, + "learning_rate": 8.513569060901477e-06, + "loss": 0.1651, + "step": 10719 + }, + { + "epoch": 4.49, + "learning_rate": 8.511892507844045e-06, + "loss": 0.1643, + "step": 10720 + }, + { + "epoch": 4.49, + "learning_rate": 8.510215997560939e-06, + "loss": 0.1554, + "step": 10721 + }, + { + "epoch": 4.49, + "learning_rate": 8.508539530100352e-06, + "loss": 0.1887, + "step": 10722 + }, + { + "epoch": 4.49, + "learning_rate": 8.506863105510476e-06, + "loss": 0.1091, + "step": 10723 + }, + { + "epoch": 4.49, + "learning_rate": 8.505186723839492e-06, + "loss": 0.1308, + "step": 10724 + }, + { + "epoch": 4.49, + "learning_rate": 8.50351038513559e-06, + "loss": 0.1375, + "step": 10725 + }, + { + "epoch": 4.49, + "learning_rate": 8.501834089446956e-06, + "loss": 0.1762, + "step": 10726 + }, + { + "epoch": 4.49, + "learning_rate": 8.500157836821768e-06, + "loss": 0.1493, + "step": 10727 + }, + { + "epoch": 4.49, + "learning_rate": 8.498481627308215e-06, + "loss": 0.1416, + "step": 10728 + }, + { + "epoch": 4.49, + "learning_rate": 8.496805460954474e-06, + "loss": 0.1904, + "step": 10729 + }, + { + "epoch": 4.49, + "learning_rate": 8.495129337808725e-06, + "loss": 0.1275, + "step": 10730 + }, + { + "epoch": 4.49, + "learning_rate": 8.493453257919148e-06, + "loss": 0.1303, + "step": 10731 + }, + { + "epoch": 4.49, + "learning_rate": 8.491777221333923e-06, + "loss": 0.1428, + "step": 10732 + }, + { + "epoch": 4.49, + "learning_rate": 8.490101228101222e-06, + "loss": 0.1747, + "step": 10733 + }, + { + "epoch": 4.49, + "learning_rate": 8.488425278269224e-06, + "loss": 0.1431, + "step": 10734 + }, + { + "epoch": 4.49, + "learning_rate": 8.486749371886096e-06, + "loss": 0.1246, + "step": 10735 + }, + { + "epoch": 4.49, + "learning_rate": 8.485073509000015e-06, + "loss": 0.136, + "step": 10736 + }, + { + "epoch": 4.5, + "learning_rate": 8.483397689659153e-06, + "loss": 0.1523, + "step": 10737 + }, + { + "epoch": 4.5, + "learning_rate": 8.481721913911677e-06, + "loss": 0.1385, + "step": 10738 + }, + { + "epoch": 4.5, + "learning_rate": 8.480046181805757e-06, + "loss": 0.1603, + "step": 10739 + }, + { + "epoch": 4.5, + "learning_rate": 8.478370493389563e-06, + "loss": 0.1634, + "step": 10740 + }, + { + "epoch": 4.5, + "learning_rate": 8.476694848711258e-06, + "loss": 0.1531, + "step": 10741 + }, + { + "epoch": 4.5, + "learning_rate": 8.475019247819006e-06, + "loss": 0.1607, + "step": 10742 + }, + { + "epoch": 4.5, + "learning_rate": 8.473343690760976e-06, + "loss": 0.1643, + "step": 10743 + }, + { + "epoch": 4.5, + "learning_rate": 8.471668177585324e-06, + "loss": 0.1783, + "step": 10744 + }, + { + "epoch": 4.5, + "learning_rate": 8.469992708340213e-06, + "loss": 0.1064, + "step": 10745 + }, + { + "epoch": 4.5, + "learning_rate": 8.468317283073805e-06, + "loss": 0.1407, + "step": 10746 + }, + { + "epoch": 4.5, + "learning_rate": 8.46664190183426e-06, + "loss": 0.1761, + "step": 10747 + }, + { + "epoch": 4.5, + "learning_rate": 8.46496656466973e-06, + "loss": 0.143, + "step": 10748 + }, + { + "epoch": 4.5, + "learning_rate": 8.463291271628375e-06, + "loss": 0.1405, + "step": 10749 + }, + { + "epoch": 4.5, + "learning_rate": 8.46161602275835e-06, + "loss": 0.1192, + "step": 10750 + }, + { + "epoch": 4.5, + "learning_rate": 8.459940818107805e-06, + "loss": 0.1546, + "step": 10751 + }, + { + "epoch": 4.5, + "learning_rate": 8.458265657724894e-06, + "loss": 0.1785, + "step": 10752 + }, + { + "epoch": 4.5, + "learning_rate": 8.456590541657772e-06, + "loss": 0.1608, + "step": 10753 + }, + { + "epoch": 4.5, + "learning_rate": 8.454915469954583e-06, + "loss": 0.1609, + "step": 10754 + }, + { + "epoch": 4.5, + "learning_rate": 8.453240442663477e-06, + "loss": 0.1414, + "step": 10755 + }, + { + "epoch": 4.5, + "learning_rate": 8.451565459832604e-06, + "loss": 0.1398, + "step": 10756 + }, + { + "epoch": 4.5, + "learning_rate": 8.44989052151011e-06, + "loss": 0.1418, + "step": 10757 + }, + { + "epoch": 4.5, + "learning_rate": 8.448215627744134e-06, + "loss": 0.1435, + "step": 10758 + }, + { + "epoch": 4.5, + "learning_rate": 8.446540778582823e-06, + "loss": 0.1296, + "step": 10759 + }, + { + "epoch": 4.5, + "learning_rate": 8.444865974074319e-06, + "loss": 0.2264, + "step": 10760 + }, + { + "epoch": 4.51, + "learning_rate": 8.443191214266763e-06, + "loss": 0.1684, + "step": 10761 + }, + { + "epoch": 4.51, + "learning_rate": 8.441516499208292e-06, + "loss": 0.1324, + "step": 10762 + }, + { + "epoch": 4.51, + "learning_rate": 8.439841828947048e-06, + "loss": 0.1965, + "step": 10763 + }, + { + "epoch": 4.51, + "learning_rate": 8.438167203531166e-06, + "loss": 0.1786, + "step": 10764 + }, + { + "epoch": 4.51, + "learning_rate": 8.436492623008782e-06, + "loss": 0.1435, + "step": 10765 + }, + { + "epoch": 4.51, + "learning_rate": 8.434818087428029e-06, + "loss": 0.1287, + "step": 10766 + }, + { + "epoch": 4.51, + "learning_rate": 8.433143596837043e-06, + "loss": 0.1772, + "step": 10767 + }, + { + "epoch": 4.51, + "learning_rate": 8.431469151283953e-06, + "loss": 0.1788, + "step": 10768 + }, + { + "epoch": 4.51, + "learning_rate": 8.429794750816888e-06, + "loss": 0.1539, + "step": 10769 + }, + { + "epoch": 4.51, + "learning_rate": 8.428120395483982e-06, + "loss": 0.14, + "step": 10770 + }, + { + "epoch": 4.51, + "learning_rate": 8.42644608533336e-06, + "loss": 0.1312, + "step": 10771 + }, + { + "epoch": 4.51, + "learning_rate": 8.42477182041315e-06, + "loss": 0.1656, + "step": 10772 + }, + { + "epoch": 4.51, + "learning_rate": 8.423097600771476e-06, + "loss": 0.1824, + "step": 10773 + }, + { + "epoch": 4.51, + "learning_rate": 8.42142342645646e-06, + "loss": 0.1685, + "step": 10774 + }, + { + "epoch": 4.51, + "learning_rate": 8.419749297516227e-06, + "loss": 0.1578, + "step": 10775 + }, + { + "epoch": 4.51, + "learning_rate": 8.4180752139989e-06, + "loss": 0.1387, + "step": 10776 + }, + { + "epoch": 4.51, + "learning_rate": 8.416401175952598e-06, + "loss": 0.1427, + "step": 10777 + }, + { + "epoch": 4.51, + "learning_rate": 8.414727183425437e-06, + "loss": 0.1592, + "step": 10778 + }, + { + "epoch": 4.51, + "learning_rate": 8.413053236465535e-06, + "loss": 0.1332, + "step": 10779 + }, + { + "epoch": 4.51, + "learning_rate": 8.411379335121014e-06, + "loss": 0.1034, + "step": 10780 + }, + { + "epoch": 4.51, + "learning_rate": 8.409705479439985e-06, + "loss": 0.185, + "step": 10781 + }, + { + "epoch": 4.51, + "learning_rate": 8.408031669470555e-06, + "loss": 0.1347, + "step": 10782 + }, + { + "epoch": 4.51, + "learning_rate": 8.406357905260845e-06, + "loss": 0.1821, + "step": 10783 + }, + { + "epoch": 4.51, + "learning_rate": 8.404684186858962e-06, + "loss": 0.1364, + "step": 10784 + }, + { + "epoch": 4.52, + "learning_rate": 8.403010514313016e-06, + "loss": 0.1522, + "step": 10785 + }, + { + "epoch": 4.52, + "learning_rate": 8.401336887671117e-06, + "loss": 0.1601, + "step": 10786 + }, + { + "epoch": 4.52, + "learning_rate": 8.399663306981368e-06, + "loss": 0.1565, + "step": 10787 + }, + { + "epoch": 4.52, + "learning_rate": 8.39798977229188e-06, + "loss": 0.1307, + "step": 10788 + }, + { + "epoch": 4.52, + "learning_rate": 8.396316283650753e-06, + "loss": 0.1278, + "step": 10789 + }, + { + "epoch": 4.52, + "learning_rate": 8.39464284110609e-06, + "loss": 0.1808, + "step": 10790 + }, + { + "epoch": 4.52, + "learning_rate": 8.392969444705998e-06, + "loss": 0.1575, + "step": 10791 + }, + { + "epoch": 4.52, + "learning_rate": 8.39129609449857e-06, + "loss": 0.163, + "step": 10792 + }, + { + "epoch": 4.52, + "learning_rate": 8.389622790531908e-06, + "loss": 0.1786, + "step": 10793 + }, + { + "epoch": 4.52, + "learning_rate": 8.387949532854113e-06, + "loss": 0.1419, + "step": 10794 + }, + { + "epoch": 4.52, + "learning_rate": 8.386276321513275e-06, + "loss": 0.1721, + "step": 10795 + }, + { + "epoch": 4.52, + "learning_rate": 8.384603156557492e-06, + "loss": 0.145, + "step": 10796 + }, + { + "epoch": 4.52, + "learning_rate": 8.38293003803486e-06, + "loss": 0.1321, + "step": 10797 + }, + { + "epoch": 4.52, + "learning_rate": 8.381256965993468e-06, + "loss": 0.1727, + "step": 10798 + }, + { + "epoch": 4.52, + "learning_rate": 8.379583940481407e-06, + "loss": 0.1064, + "step": 10799 + }, + { + "epoch": 4.52, + "learning_rate": 8.377910961546771e-06, + "loss": 0.1787, + "step": 10800 + }, + { + "epoch": 4.52, + "learning_rate": 8.376238029237643e-06, + "loss": 0.1235, + "step": 10801 + }, + { + "epoch": 4.52, + "learning_rate": 8.374565143602114e-06, + "loss": 0.1486, + "step": 10802 + }, + { + "epoch": 4.52, + "learning_rate": 8.372892304688265e-06, + "loss": 0.1193, + "step": 10803 + }, + { + "epoch": 4.52, + "learning_rate": 8.371219512544182e-06, + "loss": 0.1624, + "step": 10804 + }, + { + "epoch": 4.52, + "learning_rate": 8.369546767217952e-06, + "loss": 0.1455, + "step": 10805 + }, + { + "epoch": 4.52, + "learning_rate": 8.367874068757651e-06, + "loss": 0.1264, + "step": 10806 + }, + { + "epoch": 4.52, + "learning_rate": 8.366201417211361e-06, + "loss": 0.171, + "step": 10807 + }, + { + "epoch": 4.53, + "learning_rate": 8.364528812627164e-06, + "loss": 0.1191, + "step": 10808 + }, + { + "epoch": 4.53, + "learning_rate": 8.362856255053132e-06, + "loss": 0.1946, + "step": 10809 + }, + { + "epoch": 4.53, + "learning_rate": 8.361183744537345e-06, + "loss": 0.1596, + "step": 10810 + }, + { + "epoch": 4.53, + "learning_rate": 8.359511281127879e-06, + "loss": 0.2023, + "step": 10811 + }, + { + "epoch": 4.53, + "learning_rate": 8.357838864872803e-06, + "loss": 0.1452, + "step": 10812 + }, + { + "epoch": 4.53, + "learning_rate": 8.356166495820192e-06, + "loss": 0.1487, + "step": 10813 + }, + { + "epoch": 4.53, + "learning_rate": 8.354494174018118e-06, + "loss": 0.1538, + "step": 10814 + }, + { + "epoch": 4.53, + "learning_rate": 8.352821899514648e-06, + "loss": 0.1443, + "step": 10815 + }, + { + "epoch": 4.53, + "learning_rate": 8.35114967235785e-06, + "loss": 0.1427, + "step": 10816 + }, + { + "epoch": 4.53, + "learning_rate": 8.34947749259579e-06, + "loss": 0.1366, + "step": 10817 + }, + { + "epoch": 4.53, + "learning_rate": 8.347805360276538e-06, + "loss": 0.157, + "step": 10818 + }, + { + "epoch": 4.53, + "learning_rate": 8.346133275448153e-06, + "loss": 0.1599, + "step": 10819 + }, + { + "epoch": 4.53, + "learning_rate": 8.3444612381587e-06, + "loss": 0.1406, + "step": 10820 + }, + { + "epoch": 4.53, + "learning_rate": 8.342789248456242e-06, + "loss": 0.1796, + "step": 10821 + }, + { + "epoch": 4.53, + "learning_rate": 8.341117306388833e-06, + "loss": 0.15, + "step": 10822 + }, + { + "epoch": 4.53, + "learning_rate": 8.339445412004533e-06, + "loss": 0.1105, + "step": 10823 + }, + { + "epoch": 4.53, + "learning_rate": 8.337773565351407e-06, + "loss": 0.153, + "step": 10824 + }, + { + "epoch": 4.53, + "learning_rate": 8.336101766477506e-06, + "loss": 0.1288, + "step": 10825 + }, + { + "epoch": 4.53, + "learning_rate": 8.334430015430878e-06, + "loss": 0.1934, + "step": 10826 + }, + { + "epoch": 4.53, + "learning_rate": 8.332758312259583e-06, + "loss": 0.1263, + "step": 10827 + }, + { + "epoch": 4.53, + "learning_rate": 8.331086657011669e-06, + "loss": 0.1337, + "step": 10828 + }, + { + "epoch": 4.53, + "learning_rate": 8.32941504973519e-06, + "loss": 0.1528, + "step": 10829 + }, + { + "epoch": 4.53, + "learning_rate": 8.327743490478192e-06, + "loss": 0.1801, + "step": 10830 + }, + { + "epoch": 4.53, + "learning_rate": 8.326071979288723e-06, + "loss": 0.1432, + "step": 10831 + }, + { + "epoch": 4.54, + "learning_rate": 8.32440051621483e-06, + "loss": 0.1296, + "step": 10832 + }, + { + "epoch": 4.54, + "learning_rate": 8.322729101304558e-06, + "loss": 0.1795, + "step": 10833 + }, + { + "epoch": 4.54, + "learning_rate": 8.321057734605949e-06, + "loss": 0.1599, + "step": 10834 + }, + { + "epoch": 4.54, + "learning_rate": 8.319386416167047e-06, + "loss": 0.1965, + "step": 10835 + }, + { + "epoch": 4.54, + "learning_rate": 8.31771514603589e-06, + "loss": 0.1628, + "step": 10836 + }, + { + "epoch": 4.54, + "learning_rate": 8.316043924260514e-06, + "loss": 0.1537, + "step": 10837 + }, + { + "epoch": 4.54, + "learning_rate": 8.314372750888968e-06, + "loss": 0.1814, + "step": 10838 + }, + { + "epoch": 4.54, + "learning_rate": 8.312701625969278e-06, + "loss": 0.1303, + "step": 10839 + }, + { + "epoch": 4.54, + "learning_rate": 8.311030549549483e-06, + "loss": 0.1532, + "step": 10840 + }, + { + "epoch": 4.54, + "learning_rate": 8.309359521677617e-06, + "loss": 0.1477, + "step": 10841 + }, + { + "epoch": 4.54, + "learning_rate": 8.307688542401709e-06, + "loss": 0.1307, + "step": 10842 + }, + { + "epoch": 4.54, + "learning_rate": 8.306017611769795e-06, + "loss": 0.122, + "step": 10843 + }, + { + "epoch": 4.54, + "learning_rate": 8.3043467298299e-06, + "loss": 0.1751, + "step": 10844 + }, + { + "epoch": 4.54, + "learning_rate": 8.302675896630054e-06, + "loss": 0.1655, + "step": 10845 + }, + { + "epoch": 4.54, + "learning_rate": 8.301005112218282e-06, + "loss": 0.1444, + "step": 10846 + }, + { + "epoch": 4.54, + "learning_rate": 8.299334376642609e-06, + "loss": 0.1179, + "step": 10847 + }, + { + "epoch": 4.54, + "learning_rate": 8.297663689951067e-06, + "loss": 0.1252, + "step": 10848 + }, + { + "epoch": 4.54, + "learning_rate": 8.295993052191669e-06, + "loss": 0.1143, + "step": 10849 + }, + { + "epoch": 4.54, + "learning_rate": 8.294322463412435e-06, + "loss": 0.132, + "step": 10850 + }, + { + "epoch": 4.54, + "learning_rate": 8.292651923661391e-06, + "loss": 0.1182, + "step": 10851 + }, + { + "epoch": 4.54, + "learning_rate": 8.290981432986551e-06, + "loss": 0.1625, + "step": 10852 + }, + { + "epoch": 4.54, + "learning_rate": 8.289310991435933e-06, + "loss": 0.1747, + "step": 10853 + }, + { + "epoch": 4.54, + "learning_rate": 8.287640599057551e-06, + "loss": 0.162, + "step": 10854 + }, + { + "epoch": 4.54, + "learning_rate": 8.285970255899425e-06, + "loss": 0.1723, + "step": 10855 + }, + { + "epoch": 4.55, + "learning_rate": 8.284299962009559e-06, + "loss": 0.1552, + "step": 10856 + }, + { + "epoch": 4.55, + "learning_rate": 8.28262971743597e-06, + "loss": 0.1625, + "step": 10857 + }, + { + "epoch": 4.55, + "learning_rate": 8.280959522226663e-06, + "loss": 0.1036, + "step": 10858 + }, + { + "epoch": 4.55, + "learning_rate": 8.279289376429653e-06, + "loss": 0.1547, + "step": 10859 + }, + { + "epoch": 4.55, + "learning_rate": 8.27761928009294e-06, + "loss": 0.1245, + "step": 10860 + }, + { + "epoch": 4.55, + "learning_rate": 8.275949233264534e-06, + "loss": 0.1447, + "step": 10861 + }, + { + "epoch": 4.55, + "learning_rate": 8.274279235992438e-06, + "loss": 0.1469, + "step": 10862 + }, + { + "epoch": 4.55, + "learning_rate": 8.272609288324653e-06, + "loss": 0.1636, + "step": 10863 + }, + { + "epoch": 4.55, + "learning_rate": 8.270939390309181e-06, + "loss": 0.2064, + "step": 10864 + }, + { + "epoch": 4.55, + "learning_rate": 8.269269541994022e-06, + "loss": 0.1293, + "step": 10865 + }, + { + "epoch": 4.55, + "learning_rate": 8.267599743427173e-06, + "loss": 0.126, + "step": 10866 + }, + { + "epoch": 4.55, + "learning_rate": 8.265929994656634e-06, + "loss": 0.1713, + "step": 10867 + }, + { + "epoch": 4.55, + "learning_rate": 8.2642602957304e-06, + "loss": 0.1381, + "step": 10868 + }, + { + "epoch": 4.55, + "learning_rate": 8.26259064669646e-06, + "loss": 0.0982, + "step": 10869 + }, + { + "epoch": 4.55, + "learning_rate": 8.260921047602814e-06, + "loss": 0.1539, + "step": 10870 + }, + { + "epoch": 4.55, + "learning_rate": 8.259251498497446e-06, + "loss": 0.1549, + "step": 10871 + }, + { + "epoch": 4.55, + "learning_rate": 8.25758199942835e-06, + "loss": 0.1301, + "step": 10872 + }, + { + "epoch": 4.55, + "learning_rate": 8.255912550443515e-06, + "loss": 0.133, + "step": 10873 + }, + { + "epoch": 4.55, + "learning_rate": 8.254243151590922e-06, + "loss": 0.195, + "step": 10874 + }, + { + "epoch": 4.55, + "learning_rate": 8.252573802918562e-06, + "loss": 0.1498, + "step": 10875 + }, + { + "epoch": 4.55, + "learning_rate": 8.25090450447442e-06, + "loss": 0.173, + "step": 10876 + }, + { + "epoch": 4.55, + "learning_rate": 8.249235256306472e-06, + "loss": 0.1879, + "step": 10877 + }, + { + "epoch": 4.55, + "learning_rate": 8.247566058462704e-06, + "loss": 0.1443, + "step": 10878 + }, + { + "epoch": 4.55, + "learning_rate": 8.245896910991096e-06, + "loss": 0.1435, + "step": 10879 + }, + { + "epoch": 4.56, + "learning_rate": 8.244227813939624e-06, + "loss": 0.1807, + "step": 10880 + }, + { + "epoch": 4.56, + "learning_rate": 8.242558767356266e-06, + "loss": 0.1714, + "step": 10881 + }, + { + "epoch": 4.56, + "learning_rate": 8.240889771288997e-06, + "loss": 0.1307, + "step": 10882 + }, + { + "epoch": 4.56, + "learning_rate": 8.239220825785789e-06, + "loss": 0.1448, + "step": 10883 + }, + { + "epoch": 4.56, + "learning_rate": 8.237551930894616e-06, + "loss": 0.1597, + "step": 10884 + }, + { + "epoch": 4.56, + "learning_rate": 8.23588308666345e-06, + "loss": 0.1435, + "step": 10885 + }, + { + "epoch": 4.56, + "learning_rate": 8.234214293140258e-06, + "loss": 0.1298, + "step": 10886 + }, + { + "epoch": 4.56, + "learning_rate": 8.23254555037301e-06, + "loss": 0.1396, + "step": 10887 + }, + { + "epoch": 4.56, + "learning_rate": 8.230876858409669e-06, + "loss": 0.1867, + "step": 10888 + }, + { + "epoch": 4.56, + "learning_rate": 8.229208217298208e-06, + "loss": 0.149, + "step": 10889 + }, + { + "epoch": 4.56, + "learning_rate": 8.227539627086582e-06, + "loss": 0.1266, + "step": 10890 + }, + { + "epoch": 4.56, + "learning_rate": 8.225871087822757e-06, + "loss": 0.1307, + "step": 10891 + }, + { + "epoch": 4.56, + "learning_rate": 8.224202599554695e-06, + "loss": 0.1753, + "step": 10892 + }, + { + "epoch": 4.56, + "learning_rate": 8.222534162330355e-06, + "loss": 0.1343, + "step": 10893 + }, + { + "epoch": 4.56, + "learning_rate": 8.22086577619769e-06, + "loss": 0.145, + "step": 10894 + }, + { + "epoch": 4.56, + "learning_rate": 8.219197441204659e-06, + "loss": 0.1279, + "step": 10895 + }, + { + "epoch": 4.56, + "learning_rate": 8.21752915739922e-06, + "loss": 0.1412, + "step": 10896 + }, + { + "epoch": 4.56, + "learning_rate": 8.21586092482932e-06, + "loss": 0.1737, + "step": 10897 + }, + { + "epoch": 4.56, + "learning_rate": 8.214192743542914e-06, + "loss": 0.1382, + "step": 10898 + }, + { + "epoch": 4.56, + "learning_rate": 8.212524613587955e-06, + "loss": 0.14, + "step": 10899 + }, + { + "epoch": 4.56, + "learning_rate": 8.210856535012392e-06, + "loss": 0.1652, + "step": 10900 + }, + { + "epoch": 4.56, + "learning_rate": 8.209188507864167e-06, + "loss": 0.1024, + "step": 10901 + }, + { + "epoch": 4.56, + "learning_rate": 8.20752053219123e-06, + "loss": 0.2096, + "step": 10902 + }, + { + "epoch": 4.56, + "learning_rate": 8.205852608041526e-06, + "loss": 0.1255, + "step": 10903 + }, + { + "epoch": 4.57, + "learning_rate": 8.204184735462994e-06, + "loss": 0.1135, + "step": 10904 + }, + { + "epoch": 4.57, + "learning_rate": 8.20251691450358e-06, + "loss": 0.1712, + "step": 10905 + }, + { + "epoch": 4.57, + "learning_rate": 8.200849145211224e-06, + "loss": 0.1354, + "step": 10906 + }, + { + "epoch": 4.57, + "learning_rate": 8.19918142763386e-06, + "loss": 0.14, + "step": 10907 + }, + { + "epoch": 4.57, + "learning_rate": 8.197513761819428e-06, + "loss": 0.1466, + "step": 10908 + }, + { + "epoch": 4.57, + "learning_rate": 8.195846147815865e-06, + "loss": 0.1407, + "step": 10909 + }, + { + "epoch": 4.57, + "learning_rate": 8.194178585671102e-06, + "loss": 0.1501, + "step": 10910 + }, + { + "epoch": 4.57, + "learning_rate": 8.192511075433073e-06, + "loss": 0.1264, + "step": 10911 + }, + { + "epoch": 4.57, + "learning_rate": 8.190843617149712e-06, + "loss": 0.181, + "step": 10912 + }, + { + "epoch": 4.57, + "learning_rate": 8.189176210868942e-06, + "loss": 0.1677, + "step": 10913 + }, + { + "epoch": 4.57, + "learning_rate": 8.187508856638697e-06, + "loss": 0.174, + "step": 10914 + }, + { + "epoch": 4.57, + "learning_rate": 8.185841554506901e-06, + "loss": 0.1589, + "step": 10915 + }, + { + "epoch": 4.57, + "learning_rate": 8.184174304521483e-06, + "loss": 0.1424, + "step": 10916 + }, + { + "epoch": 4.57, + "learning_rate": 8.18250710673036e-06, + "loss": 0.1473, + "step": 10917 + }, + { + "epoch": 4.57, + "learning_rate": 8.180839961181458e-06, + "loss": 0.1487, + "step": 10918 + }, + { + "epoch": 4.57, + "learning_rate": 8.179172867922695e-06, + "loss": 0.1064, + "step": 10919 + }, + { + "epoch": 4.57, + "learning_rate": 8.177505827001997e-06, + "loss": 0.1629, + "step": 10920 + }, + { + "epoch": 4.57, + "learning_rate": 8.175838838467272e-06, + "loss": 0.168, + "step": 10921 + }, + { + "epoch": 4.57, + "learning_rate": 8.174171902366441e-06, + "loss": 0.1765, + "step": 10922 + }, + { + "epoch": 4.57, + "learning_rate": 8.172505018747421e-06, + "loss": 0.1488, + "step": 10923 + }, + { + "epoch": 4.57, + "learning_rate": 8.17083818765812e-06, + "loss": 0.1875, + "step": 10924 + }, + { + "epoch": 4.57, + "learning_rate": 8.169171409146453e-06, + "loss": 0.1606, + "step": 10925 + }, + { + "epoch": 4.57, + "learning_rate": 8.167504683260332e-06, + "loss": 0.1725, + "step": 10926 + }, + { + "epoch": 4.57, + "learning_rate": 8.16583801004766e-06, + "loss": 0.129, + "step": 10927 + }, + { + "epoch": 4.58, + "learning_rate": 8.164171389556346e-06, + "loss": 0.1683, + "step": 10928 + }, + { + "epoch": 4.58, + "learning_rate": 8.162504821834296e-06, + "loss": 0.1483, + "step": 10929 + }, + { + "epoch": 4.58, + "learning_rate": 8.160838306929418e-06, + "loss": 0.1661, + "step": 10930 + }, + { + "epoch": 4.58, + "learning_rate": 8.159171844889609e-06, + "loss": 0.149, + "step": 10931 + }, + { + "epoch": 4.58, + "learning_rate": 8.157505435762769e-06, + "loss": 0.161, + "step": 10932 + }, + { + "epoch": 4.58, + "learning_rate": 8.155839079596805e-06, + "loss": 0.1306, + "step": 10933 + }, + { + "epoch": 4.58, + "learning_rate": 8.154172776439608e-06, + "loss": 0.1119, + "step": 10934 + }, + { + "epoch": 4.58, + "learning_rate": 8.152506526339075e-06, + "loss": 0.1548, + "step": 10935 + }, + { + "epoch": 4.58, + "learning_rate": 8.150840329343104e-06, + "loss": 0.1768, + "step": 10936 + }, + { + "epoch": 4.58, + "learning_rate": 8.149174185499587e-06, + "loss": 0.1677, + "step": 10937 + }, + { + "epoch": 4.58, + "learning_rate": 8.14750809485642e-06, + "loss": 0.127, + "step": 10938 + }, + { + "epoch": 4.58, + "learning_rate": 8.145842057461483e-06, + "loss": 0.1572, + "step": 10939 + }, + { + "epoch": 4.58, + "learning_rate": 8.14417607336267e-06, + "loss": 0.1449, + "step": 10940 + }, + { + "epoch": 4.58, + "learning_rate": 8.142510142607874e-06, + "loss": 0.1592, + "step": 10941 + }, + { + "epoch": 4.58, + "learning_rate": 8.140844265244971e-06, + "loss": 0.1251, + "step": 10942 + }, + { + "epoch": 4.58, + "learning_rate": 8.139178441321851e-06, + "loss": 0.1163, + "step": 10943 + }, + { + "epoch": 4.58, + "learning_rate": 8.137512670886397e-06, + "loss": 0.1183, + "step": 10944 + }, + { + "epoch": 4.58, + "learning_rate": 8.135846953986486e-06, + "loss": 0.1977, + "step": 10945 + }, + { + "epoch": 4.58, + "learning_rate": 8.134181290670002e-06, + "loss": 0.1227, + "step": 10946 + }, + { + "epoch": 4.58, + "learning_rate": 8.132515680984821e-06, + "loss": 0.1084, + "step": 10947 + }, + { + "epoch": 4.58, + "learning_rate": 8.130850124978819e-06, + "loss": 0.163, + "step": 10948 + }, + { + "epoch": 4.58, + "learning_rate": 8.12918462269987e-06, + "loss": 0.1608, + "step": 10949 + }, + { + "epoch": 4.58, + "learning_rate": 8.127519174195853e-06, + "loss": 0.1399, + "step": 10950 + }, + { + "epoch": 4.58, + "learning_rate": 8.125853779514632e-06, + "loss": 0.1631, + "step": 10951 + }, + { + "epoch": 4.59, + "learning_rate": 8.124188438704082e-06, + "loss": 0.1099, + "step": 10952 + }, + { + "epoch": 4.59, + "learning_rate": 8.122523151812072e-06, + "loss": 0.1351, + "step": 10953 + }, + { + "epoch": 4.59, + "learning_rate": 8.120857918886467e-06, + "loss": 0.1401, + "step": 10954 + }, + { + "epoch": 4.59, + "learning_rate": 8.119192739975133e-06, + "loss": 0.1397, + "step": 10955 + }, + { + "epoch": 4.59, + "learning_rate": 8.117527615125938e-06, + "loss": 0.1302, + "step": 10956 + }, + { + "epoch": 4.59, + "learning_rate": 8.115862544386736e-06, + "loss": 0.1344, + "step": 10957 + }, + { + "epoch": 4.59, + "learning_rate": 8.114197527805396e-06, + "loss": 0.1636, + "step": 10958 + }, + { + "epoch": 4.59, + "learning_rate": 8.112532565429774e-06, + "loss": 0.1649, + "step": 10959 + }, + { + "epoch": 4.59, + "learning_rate": 8.11086765730773e-06, + "loss": 0.1666, + "step": 10960 + }, + { + "epoch": 4.59, + "learning_rate": 8.109202803487121e-06, + "loss": 0.1521, + "step": 10961 + }, + { + "epoch": 4.59, + "learning_rate": 8.107538004015796e-06, + "loss": 0.1471, + "step": 10962 + }, + { + "epoch": 4.59, + "learning_rate": 8.105873258941612e-06, + "loss": 0.1451, + "step": 10963 + }, + { + "epoch": 4.59, + "learning_rate": 8.10420856831242e-06, + "loss": 0.1674, + "step": 10964 + }, + { + "epoch": 4.59, + "learning_rate": 8.10254393217607e-06, + "loss": 0.1801, + "step": 10965 + }, + { + "epoch": 4.59, + "learning_rate": 8.100879350580411e-06, + "loss": 0.1484, + "step": 10966 + }, + { + "epoch": 4.59, + "learning_rate": 8.099214823573293e-06, + "loss": 0.1657, + "step": 10967 + }, + { + "epoch": 4.59, + "learning_rate": 8.097550351202554e-06, + "loss": 0.179, + "step": 10968 + }, + { + "epoch": 4.59, + "learning_rate": 8.095885933516043e-06, + "loss": 0.1169, + "step": 10969 + }, + { + "epoch": 4.59, + "learning_rate": 8.094221570561602e-06, + "loss": 0.1453, + "step": 10970 + }, + { + "epoch": 4.59, + "learning_rate": 8.09255726238707e-06, + "loss": 0.1403, + "step": 10971 + }, + { + "epoch": 4.59, + "learning_rate": 8.090893009040288e-06, + "loss": 0.1554, + "step": 10972 + }, + { + "epoch": 4.59, + "learning_rate": 8.08922881056909e-06, + "loss": 0.1699, + "step": 10973 + }, + { + "epoch": 4.59, + "learning_rate": 8.087564667021317e-06, + "loss": 0.1783, + "step": 10974 + }, + { + "epoch": 4.59, + "learning_rate": 8.0859005784448e-06, + "loss": 0.1717, + "step": 10975 + }, + { + "epoch": 4.6, + "learning_rate": 8.084236544887372e-06, + "loss": 0.1571, + "step": 10976 + }, + { + "epoch": 4.6, + "learning_rate": 8.082572566396864e-06, + "loss": 0.1587, + "step": 10977 + }, + { + "epoch": 4.6, + "learning_rate": 8.080908643021107e-06, + "loss": 0.1206, + "step": 10978 + }, + { + "epoch": 4.6, + "learning_rate": 8.079244774807926e-06, + "loss": 0.1667, + "step": 10979 + }, + { + "epoch": 4.6, + "learning_rate": 8.077580961805153e-06, + "loss": 0.1552, + "step": 10980 + }, + { + "epoch": 4.6, + "learning_rate": 8.075917204060606e-06, + "loss": 0.1245, + "step": 10981 + }, + { + "epoch": 4.6, + "learning_rate": 8.074253501622113e-06, + "loss": 0.1779, + "step": 10982 + }, + { + "epoch": 4.6, + "learning_rate": 8.072589854537498e-06, + "loss": 0.1806, + "step": 10983 + }, + { + "epoch": 4.6, + "learning_rate": 8.070926262854573e-06, + "loss": 0.1765, + "step": 10984 + }, + { + "epoch": 4.6, + "learning_rate": 8.069262726621163e-06, + "loss": 0.1469, + "step": 10985 + }, + { + "epoch": 4.6, + "learning_rate": 8.06759924588508e-06, + "loss": 0.1461, + "step": 10986 + }, + { + "epoch": 4.6, + "learning_rate": 8.065935820694142e-06, + "loss": 0.1394, + "step": 10987 + }, + { + "epoch": 4.6, + "learning_rate": 8.064272451096165e-06, + "loss": 0.1073, + "step": 10988 + }, + { + "epoch": 4.6, + "learning_rate": 8.062609137138955e-06, + "loss": 0.126, + "step": 10989 + }, + { + "epoch": 4.6, + "learning_rate": 8.060945878870328e-06, + "loss": 0.1782, + "step": 10990 + }, + { + "epoch": 4.6, + "learning_rate": 8.059282676338092e-06, + "loss": 0.1596, + "step": 10991 + }, + { + "epoch": 4.6, + "learning_rate": 8.057619529590051e-06, + "loss": 0.138, + "step": 10992 + }, + { + "epoch": 4.6, + "learning_rate": 8.055956438674015e-06, + "loss": 0.1796, + "step": 10993 + }, + { + "epoch": 4.6, + "learning_rate": 8.054293403637787e-06, + "loss": 0.153, + "step": 10994 + }, + { + "epoch": 4.6, + "learning_rate": 8.052630424529166e-06, + "loss": 0.1554, + "step": 10995 + }, + { + "epoch": 4.6, + "learning_rate": 8.050967501395955e-06, + "loss": 0.1505, + "step": 10996 + }, + { + "epoch": 4.6, + "learning_rate": 8.049304634285957e-06, + "loss": 0.1279, + "step": 10997 + }, + { + "epoch": 4.6, + "learning_rate": 8.047641823246964e-06, + "loss": 0.1643, + "step": 10998 + }, + { + "epoch": 4.6, + "learning_rate": 8.045979068326773e-06, + "loss": 0.1534, + "step": 10999 + }, + { + "epoch": 4.61, + "learning_rate": 8.044316369573181e-06, + "loss": 0.1411, + "step": 11000 + }, + { + "epoch": 4.61, + "learning_rate": 8.042653727033981e-06, + "loss": 0.1839, + "step": 11001 + }, + { + "epoch": 4.61, + "learning_rate": 8.04099114075696e-06, + "loss": 0.158, + "step": 11002 + }, + { + "epoch": 4.61, + "learning_rate": 8.039328610789911e-06, + "loss": 0.1694, + "step": 11003 + }, + { + "epoch": 4.61, + "learning_rate": 8.037666137180624e-06, + "loss": 0.1096, + "step": 11004 + }, + { + "epoch": 4.61, + "learning_rate": 8.03600371997688e-06, + "loss": 0.1871, + "step": 11005 + }, + { + "epoch": 4.61, + "learning_rate": 8.03434135922647e-06, + "loss": 0.12, + "step": 11006 + }, + { + "epoch": 4.61, + "learning_rate": 8.032679054977171e-06, + "loss": 0.1471, + "step": 11007 + }, + { + "epoch": 4.61, + "learning_rate": 8.031016807276765e-06, + "loss": 0.1441, + "step": 11008 + }, + { + "epoch": 4.61, + "learning_rate": 8.029354616173036e-06, + "loss": 0.1719, + "step": 11009 + }, + { + "epoch": 4.61, + "learning_rate": 8.027692481713757e-06, + "loss": 0.1434, + "step": 11010 + }, + { + "epoch": 4.61, + "learning_rate": 8.026030403946707e-06, + "loss": 0.1538, + "step": 11011 + }, + { + "epoch": 4.61, + "learning_rate": 8.024368382919665e-06, + "loss": 0.1316, + "step": 11012 + }, + { + "epoch": 4.61, + "learning_rate": 8.022706418680398e-06, + "loss": 0.1592, + "step": 11013 + }, + { + "epoch": 4.61, + "learning_rate": 8.021044511276682e-06, + "loss": 0.1634, + "step": 11014 + }, + { + "epoch": 4.61, + "learning_rate": 8.019382660756284e-06, + "loss": 0.1346, + "step": 11015 + }, + { + "epoch": 4.61, + "learning_rate": 8.017720867166976e-06, + "loss": 0.142, + "step": 11016 + }, + { + "epoch": 4.61, + "learning_rate": 8.01605913055652e-06, + "loss": 0.1682, + "step": 11017 + }, + { + "epoch": 4.61, + "learning_rate": 8.014397450972685e-06, + "loss": 0.156, + "step": 11018 + }, + { + "epoch": 4.61, + "learning_rate": 8.012735828463233e-06, + "loss": 0.1653, + "step": 11019 + }, + { + "epoch": 4.61, + "learning_rate": 8.011074263075925e-06, + "loss": 0.1269, + "step": 11020 + }, + { + "epoch": 4.61, + "learning_rate": 8.009412754858525e-06, + "loss": 0.1328, + "step": 11021 + }, + { + "epoch": 4.61, + "learning_rate": 8.007751303858788e-06, + "loss": 0.1473, + "step": 11022 + }, + { + "epoch": 4.62, + "learning_rate": 8.00608991012447e-06, + "loss": 0.108, + "step": 11023 + }, + { + "epoch": 4.62, + "learning_rate": 8.00442857370333e-06, + "loss": 0.1977, + "step": 11024 + }, + { + "epoch": 4.62, + "learning_rate": 8.00276729464312e-06, + "loss": 0.1484, + "step": 11025 + }, + { + "epoch": 4.62, + "learning_rate": 8.00110607299159e-06, + "loss": 0.1721, + "step": 11026 + }, + { + "epoch": 4.62, + "learning_rate": 7.999444908796493e-06, + "loss": 0.1437, + "step": 11027 + }, + { + "epoch": 4.62, + "learning_rate": 7.99778380210558e-06, + "loss": 0.132, + "step": 11028 + }, + { + "epoch": 4.62, + "learning_rate": 7.996122752966596e-06, + "loss": 0.148, + "step": 11029 + }, + { + "epoch": 4.62, + "learning_rate": 7.99446176142728e-06, + "loss": 0.1464, + "step": 11030 + }, + { + "epoch": 4.62, + "learning_rate": 7.992800827535381e-06, + "loss": 0.1883, + "step": 11031 + }, + { + "epoch": 4.62, + "learning_rate": 7.991139951338645e-06, + "loss": 0.1545, + "step": 11032 + }, + { + "epoch": 4.62, + "learning_rate": 7.989479132884803e-06, + "loss": 0.1339, + "step": 11033 + }, + { + "epoch": 4.62, + "learning_rate": 7.987818372221603e-06, + "loss": 0.1387, + "step": 11034 + }, + { + "epoch": 4.62, + "learning_rate": 7.98615766939678e-06, + "loss": 0.1584, + "step": 11035 + }, + { + "epoch": 4.62, + "learning_rate": 7.984497024458062e-06, + "loss": 0.128, + "step": 11036 + }, + { + "epoch": 4.62, + "learning_rate": 7.982836437453192e-06, + "loss": 0.1636, + "step": 11037 + }, + { + "epoch": 4.62, + "learning_rate": 7.9811759084299e-06, + "loss": 0.1627, + "step": 11038 + }, + { + "epoch": 4.62, + "learning_rate": 7.979515437435913e-06, + "loss": 0.1573, + "step": 11039 + }, + { + "epoch": 4.62, + "learning_rate": 7.977855024518962e-06, + "loss": 0.1787, + "step": 11040 + }, + { + "epoch": 4.62, + "learning_rate": 7.976194669726773e-06, + "loss": 0.1583, + "step": 11041 + }, + { + "epoch": 4.62, + "learning_rate": 7.974534373107074e-06, + "loss": 0.1613, + "step": 11042 + }, + { + "epoch": 4.62, + "learning_rate": 7.972874134707585e-06, + "loss": 0.1165, + "step": 11043 + }, + { + "epoch": 4.62, + "learning_rate": 7.971213954576031e-06, + "loss": 0.1546, + "step": 11044 + }, + { + "epoch": 4.62, + "learning_rate": 7.969553832760134e-06, + "loss": 0.1246, + "step": 11045 + }, + { + "epoch": 4.62, + "learning_rate": 7.967893769307608e-06, + "loss": 0.1284, + "step": 11046 + }, + { + "epoch": 4.63, + "learning_rate": 7.966233764266172e-06, + "loss": 0.1343, + "step": 11047 + }, + { + "epoch": 4.63, + "learning_rate": 7.964573817683545e-06, + "loss": 0.1521, + "step": 11048 + }, + { + "epoch": 4.63, + "learning_rate": 7.962913929607433e-06, + "loss": 0.1585, + "step": 11049 + }, + { + "epoch": 4.63, + "learning_rate": 7.961254100085554e-06, + "loss": 0.132, + "step": 11050 + }, + { + "epoch": 4.63, + "learning_rate": 7.95959432916562e-06, + "loss": 0.1382, + "step": 11051 + }, + { + "epoch": 4.63, + "learning_rate": 7.957934616895332e-06, + "loss": 0.1273, + "step": 11052 + }, + { + "epoch": 4.63, + "learning_rate": 7.956274963322406e-06, + "loss": 0.1409, + "step": 11053 + }, + { + "epoch": 4.63, + "learning_rate": 7.954615368494538e-06, + "loss": 0.1615, + "step": 11054 + }, + { + "epoch": 4.63, + "learning_rate": 7.952955832459437e-06, + "loss": 0.1566, + "step": 11055 + }, + { + "epoch": 4.63, + "learning_rate": 7.951296355264805e-06, + "loss": 0.1366, + "step": 11056 + }, + { + "epoch": 4.63, + "learning_rate": 7.949636936958342e-06, + "loss": 0.1468, + "step": 11057 + }, + { + "epoch": 4.63, + "learning_rate": 7.947977577587745e-06, + "loss": 0.1938, + "step": 11058 + }, + { + "epoch": 4.63, + "learning_rate": 7.946318277200712e-06, + "loss": 0.1244, + "step": 11059 + }, + { + "epoch": 4.63, + "learning_rate": 7.944659035844937e-06, + "loss": 0.1601, + "step": 11060 + }, + { + "epoch": 4.63, + "learning_rate": 7.942999853568112e-06, + "loss": 0.1841, + "step": 11061 + }, + { + "epoch": 4.63, + "learning_rate": 7.941340730417934e-06, + "loss": 0.1518, + "step": 11062 + }, + { + "epoch": 4.63, + "learning_rate": 7.93968166644209e-06, + "loss": 0.1456, + "step": 11063 + }, + { + "epoch": 4.63, + "learning_rate": 7.938022661688265e-06, + "loss": 0.137, + "step": 11064 + }, + { + "epoch": 4.63, + "learning_rate": 7.936363716204153e-06, + "loss": 0.1501, + "step": 11065 + }, + { + "epoch": 4.63, + "learning_rate": 7.934704830037432e-06, + "loss": 0.1474, + "step": 11066 + }, + { + "epoch": 4.63, + "learning_rate": 7.933046003235789e-06, + "loss": 0.1521, + "step": 11067 + }, + { + "epoch": 4.63, + "learning_rate": 7.931387235846905e-06, + "loss": 0.1458, + "step": 11068 + }, + { + "epoch": 4.63, + "learning_rate": 7.929728527918456e-06, + "loss": 0.1276, + "step": 11069 + }, + { + "epoch": 4.63, + "learning_rate": 7.928069879498126e-06, + "loss": 0.1618, + "step": 11070 + }, + { + "epoch": 4.64, + "learning_rate": 7.926411290633588e-06, + "loss": 0.1723, + "step": 11071 + }, + { + "epoch": 4.64, + "learning_rate": 7.92475276137252e-06, + "loss": 0.151, + "step": 11072 + }, + { + "epoch": 4.64, + "learning_rate": 7.923094291762588e-06, + "loss": 0.1688, + "step": 11073 + }, + { + "epoch": 4.64, + "learning_rate": 7.921435881851474e-06, + "loss": 0.1447, + "step": 11074 + }, + { + "epoch": 4.64, + "learning_rate": 7.919777531686837e-06, + "loss": 0.1317, + "step": 11075 + }, + { + "epoch": 4.64, + "learning_rate": 7.91811924131635e-06, + "loss": 0.1612, + "step": 11076 + }, + { + "epoch": 4.64, + "learning_rate": 7.916461010787676e-06, + "loss": 0.1556, + "step": 11077 + }, + { + "epoch": 4.64, + "learning_rate": 7.914802840148482e-06, + "loss": 0.1761, + "step": 11078 + }, + { + "epoch": 4.64, + "learning_rate": 7.91314472944643e-06, + "loss": 0.1232, + "step": 11079 + }, + { + "epoch": 4.64, + "learning_rate": 7.91148667872918e-06, + "loss": 0.1865, + "step": 11080 + }, + { + "epoch": 4.64, + "learning_rate": 7.909828688044392e-06, + "loss": 0.1404, + "step": 11081 + }, + { + "epoch": 4.64, + "learning_rate": 7.908170757439724e-06, + "loss": 0.148, + "step": 11082 + }, + { + "epoch": 4.64, + "learning_rate": 7.906512886962833e-06, + "loss": 0.1539, + "step": 11083 + }, + { + "epoch": 4.64, + "learning_rate": 7.90485507666137e-06, + "loss": 0.1607, + "step": 11084 + }, + { + "epoch": 4.64, + "learning_rate": 7.903197326582988e-06, + "loss": 0.1351, + "step": 11085 + }, + { + "epoch": 4.64, + "learning_rate": 7.90153963677534e-06, + "loss": 0.1095, + "step": 11086 + }, + { + "epoch": 4.64, + "learning_rate": 7.899882007286074e-06, + "loss": 0.1682, + "step": 11087 + }, + { + "epoch": 4.64, + "learning_rate": 7.898224438162831e-06, + "loss": 0.1204, + "step": 11088 + }, + { + "epoch": 4.64, + "learning_rate": 7.896566929453267e-06, + "loss": 0.166, + "step": 11089 + }, + { + "epoch": 4.64, + "learning_rate": 7.894909481205017e-06, + "loss": 0.118, + "step": 11090 + }, + { + "epoch": 4.64, + "learning_rate": 7.893252093465725e-06, + "loss": 0.133, + "step": 11091 + }, + { + "epoch": 4.64, + "learning_rate": 7.891594766283034e-06, + "loss": 0.1369, + "step": 11092 + }, + { + "epoch": 4.64, + "learning_rate": 7.88993749970458e-06, + "loss": 0.1728, + "step": 11093 + }, + { + "epoch": 4.64, + "learning_rate": 7.888280293778001e-06, + "loss": 0.1538, + "step": 11094 + }, + { + "epoch": 4.65, + "learning_rate": 7.886623148550932e-06, + "loss": 0.1335, + "step": 11095 + }, + { + "epoch": 4.65, + "learning_rate": 7.884966064071002e-06, + "loss": 0.1191, + "step": 11096 + }, + { + "epoch": 4.65, + "learning_rate": 7.88330904038585e-06, + "loss": 0.1459, + "step": 11097 + }, + { + "epoch": 4.65, + "learning_rate": 7.881652077543098e-06, + "loss": 0.1629, + "step": 11098 + }, + { + "epoch": 4.65, + "learning_rate": 7.879995175590378e-06, + "loss": 0.1501, + "step": 11099 + }, + { + "epoch": 4.65, + "learning_rate": 7.878338334575317e-06, + "loss": 0.189, + "step": 11100 + }, + { + "epoch": 4.65, + "learning_rate": 7.876681554545535e-06, + "loss": 0.1861, + "step": 11101 + }, + { + "epoch": 4.65, + "learning_rate": 7.875024835548658e-06, + "loss": 0.1758, + "step": 11102 + }, + { + "epoch": 4.65, + "learning_rate": 7.873368177632308e-06, + "loss": 0.1761, + "step": 11103 + }, + { + "epoch": 4.65, + "learning_rate": 7.871711580844102e-06, + "loss": 0.159, + "step": 11104 + }, + { + "epoch": 4.65, + "learning_rate": 7.870055045231656e-06, + "loss": 0.1334, + "step": 11105 + }, + { + "epoch": 4.65, + "learning_rate": 7.86839857084259e-06, + "loss": 0.1364, + "step": 11106 + }, + { + "epoch": 4.65, + "learning_rate": 7.866742157724515e-06, + "loss": 0.1134, + "step": 11107 + }, + { + "epoch": 4.65, + "learning_rate": 7.865085805925042e-06, + "loss": 0.1967, + "step": 11108 + }, + { + "epoch": 4.65, + "learning_rate": 7.863429515491786e-06, + "loss": 0.1544, + "step": 11109 + }, + { + "epoch": 4.65, + "learning_rate": 7.861773286472351e-06, + "loss": 0.1231, + "step": 11110 + }, + { + "epoch": 4.65, + "learning_rate": 7.860117118914345e-06, + "loss": 0.1251, + "step": 11111 + }, + { + "epoch": 4.65, + "learning_rate": 7.858461012865373e-06, + "loss": 0.1512, + "step": 11112 + }, + { + "epoch": 4.65, + "learning_rate": 7.856804968373041e-06, + "loss": 0.2215, + "step": 11113 + }, + { + "epoch": 4.65, + "learning_rate": 7.855148985484946e-06, + "loss": 0.135, + "step": 11114 + }, + { + "epoch": 4.65, + "learning_rate": 7.85349306424869e-06, + "loss": 0.1436, + "step": 11115 + }, + { + "epoch": 4.65, + "learning_rate": 7.851837204711874e-06, + "loss": 0.2166, + "step": 11116 + }, + { + "epoch": 4.65, + "learning_rate": 7.85018140692209e-06, + "loss": 0.1807, + "step": 11117 + }, + { + "epoch": 4.65, + "learning_rate": 7.848525670926932e-06, + "loss": 0.1591, + "step": 11118 + }, + { + "epoch": 4.66, + "learning_rate": 7.846869996774e-06, + "loss": 0.1464, + "step": 11119 + }, + { + "epoch": 4.66, + "learning_rate": 7.845214384510875e-06, + "loss": 0.1911, + "step": 11120 + }, + { + "epoch": 4.66, + "learning_rate": 7.84355883418515e-06, + "loss": 0.1695, + "step": 11121 + }, + { + "epoch": 4.66, + "learning_rate": 7.841903345844413e-06, + "loss": 0.1471, + "step": 11122 + }, + { + "epoch": 4.66, + "learning_rate": 7.840247919536249e-06, + "loss": 0.1358, + "step": 11123 + }, + { + "epoch": 4.66, + "learning_rate": 7.838592555308244e-06, + "loss": 0.1547, + "step": 11124 + }, + { + "epoch": 4.66, + "learning_rate": 7.836937253207976e-06, + "loss": 0.1534, + "step": 11125 + }, + { + "epoch": 4.66, + "learning_rate": 7.835282013283027e-06, + "loss": 0.1508, + "step": 11126 + }, + { + "epoch": 4.66, + "learning_rate": 7.833626835580977e-06, + "loss": 0.1305, + "step": 11127 + }, + { + "epoch": 4.66, + "learning_rate": 7.8319717201494e-06, + "loss": 0.1499, + "step": 11128 + }, + { + "epoch": 4.66, + "learning_rate": 7.830316667035872e-06, + "loss": 0.1538, + "step": 11129 + }, + { + "epoch": 4.66, + "learning_rate": 7.828661676287968e-06, + "loss": 0.1356, + "step": 11130 + }, + { + "epoch": 4.66, + "learning_rate": 7.827006747953255e-06, + "loss": 0.1237, + "step": 11131 + }, + { + "epoch": 4.66, + "learning_rate": 7.825351882079305e-06, + "loss": 0.1834, + "step": 11132 + }, + { + "epoch": 4.66, + "learning_rate": 7.823697078713689e-06, + "loss": 0.1415, + "step": 11133 + }, + { + "epoch": 4.66, + "learning_rate": 7.822042337903964e-06, + "loss": 0.1404, + "step": 11134 + }, + { + "epoch": 4.66, + "learning_rate": 7.820387659697701e-06, + "loss": 0.1441, + "step": 11135 + }, + { + "epoch": 4.66, + "learning_rate": 7.818733044142462e-06, + "loss": 0.1424, + "step": 11136 + }, + { + "epoch": 4.66, + "learning_rate": 7.817078491285806e-06, + "loss": 0.1405, + "step": 11137 + }, + { + "epoch": 4.66, + "learning_rate": 7.815424001175289e-06, + "loss": 0.1463, + "step": 11138 + }, + { + "epoch": 4.66, + "learning_rate": 7.813769573858474e-06, + "loss": 0.1954, + "step": 11139 + }, + { + "epoch": 4.66, + "learning_rate": 7.81211520938291e-06, + "loss": 0.1358, + "step": 11140 + }, + { + "epoch": 4.66, + "learning_rate": 7.810460907796153e-06, + "loss": 0.1541, + "step": 11141 + }, + { + "epoch": 4.66, + "learning_rate": 7.80880666914576e-06, + "loss": 0.1566, + "step": 11142 + }, + { + "epoch": 4.67, + "learning_rate": 7.807152493479268e-06, + "loss": 0.1288, + "step": 11143 + }, + { + "epoch": 4.67, + "learning_rate": 7.805498380844236e-06, + "loss": 0.1411, + "step": 11144 + }, + { + "epoch": 4.67, + "learning_rate": 7.803844331288201e-06, + "loss": 0.1424, + "step": 11145 + }, + { + "epoch": 4.67, + "learning_rate": 7.802190344858715e-06, + "loss": 0.1793, + "step": 11146 + }, + { + "epoch": 4.67, + "learning_rate": 7.800536421603317e-06, + "loss": 0.1316, + "step": 11147 + }, + { + "epoch": 4.67, + "learning_rate": 7.798882561569546e-06, + "loss": 0.1403, + "step": 11148 + }, + { + "epoch": 4.67, + "learning_rate": 7.797228764804942e-06, + "loss": 0.1246, + "step": 11149 + }, + { + "epoch": 4.67, + "learning_rate": 7.795575031357047e-06, + "loss": 0.1353, + "step": 11150 + }, + { + "epoch": 4.67, + "learning_rate": 7.793921361273388e-06, + "loss": 0.183, + "step": 11151 + }, + { + "epoch": 4.67, + "learning_rate": 7.792267754601501e-06, + "loss": 0.1039, + "step": 11152 + }, + { + "epoch": 4.67, + "learning_rate": 7.79061421138892e-06, + "loss": 0.1485, + "step": 11153 + }, + { + "epoch": 4.67, + "learning_rate": 7.788960731683174e-06, + "loss": 0.1523, + "step": 11154 + }, + { + "epoch": 4.67, + "learning_rate": 7.787307315531787e-06, + "loss": 0.1523, + "step": 11155 + }, + { + "epoch": 4.67, + "learning_rate": 7.785653962982288e-06, + "loss": 0.1329, + "step": 11156 + }, + { + "epoch": 4.67, + "learning_rate": 7.784000674082205e-06, + "loss": 0.1695, + "step": 11157 + }, + { + "epoch": 4.67, + "learning_rate": 7.782347448879051e-06, + "loss": 0.1421, + "step": 11158 + }, + { + "epoch": 4.67, + "learning_rate": 7.780694287420355e-06, + "loss": 0.1418, + "step": 11159 + }, + { + "epoch": 4.67, + "learning_rate": 7.77904118975363e-06, + "loss": 0.1341, + "step": 11160 + }, + { + "epoch": 4.67, + "learning_rate": 7.777388155926397e-06, + "loss": 0.16, + "step": 11161 + }, + { + "epoch": 4.67, + "learning_rate": 7.775735185986168e-06, + "loss": 0.1736, + "step": 11162 + }, + { + "epoch": 4.67, + "learning_rate": 7.774082279980459e-06, + "loss": 0.1503, + "step": 11163 + }, + { + "epoch": 4.67, + "learning_rate": 7.772429437956777e-06, + "loss": 0.1389, + "step": 11164 + }, + { + "epoch": 4.67, + "learning_rate": 7.77077665996264e-06, + "loss": 0.1582, + "step": 11165 + }, + { + "epoch": 4.67, + "learning_rate": 7.769123946045545e-06, + "loss": 0.1867, + "step": 11166 + }, + { + "epoch": 4.68, + "learning_rate": 7.767471296253003e-06, + "loss": 0.1555, + "step": 11167 + }, + { + "epoch": 4.68, + "learning_rate": 7.765818710632516e-06, + "loss": 0.1818, + "step": 11168 + }, + { + "epoch": 4.68, + "learning_rate": 7.764166189231588e-06, + "loss": 0.1169, + "step": 11169 + }, + { + "epoch": 4.68, + "learning_rate": 7.762513732097721e-06, + "loss": 0.1792, + "step": 11170 + }, + { + "epoch": 4.68, + "learning_rate": 7.76086133927841e-06, + "loss": 0.1554, + "step": 11171 + }, + { + "epoch": 4.68, + "learning_rate": 7.759209010821152e-06, + "loss": 0.1676, + "step": 11172 + }, + { + "epoch": 4.68, + "learning_rate": 7.757556746773443e-06, + "loss": 0.1391, + "step": 11173 + }, + { + "epoch": 4.68, + "learning_rate": 7.755904547182778e-06, + "loss": 0.1281, + "step": 11174 + }, + { + "epoch": 4.68, + "learning_rate": 7.754252412096642e-06, + "loss": 0.1502, + "step": 11175 + }, + { + "epoch": 4.68, + "learning_rate": 7.752600341562527e-06, + "loss": 0.152, + "step": 11176 + }, + { + "epoch": 4.68, + "learning_rate": 7.750948335627925e-06, + "loss": 0.1148, + "step": 11177 + }, + { + "epoch": 4.68, + "learning_rate": 7.749296394340313e-06, + "loss": 0.1783, + "step": 11178 + }, + { + "epoch": 4.68, + "learning_rate": 7.747644517747182e-06, + "loss": 0.1686, + "step": 11179 + }, + { + "epoch": 4.68, + "learning_rate": 7.74599270589601e-06, + "loss": 0.1007, + "step": 11180 + }, + { + "epoch": 4.68, + "learning_rate": 7.744340958834275e-06, + "loss": 0.1236, + "step": 11181 + }, + { + "epoch": 4.68, + "learning_rate": 7.742689276609459e-06, + "loss": 0.1305, + "step": 11182 + }, + { + "epoch": 4.68, + "learning_rate": 7.741037659269037e-06, + "loss": 0.1455, + "step": 11183 + }, + { + "epoch": 4.68, + "learning_rate": 7.739386106860485e-06, + "loss": 0.1433, + "step": 11184 + }, + { + "epoch": 4.68, + "learning_rate": 7.73773461943127e-06, + "loss": 0.1277, + "step": 11185 + }, + { + "epoch": 4.68, + "learning_rate": 7.736083197028866e-06, + "loss": 0.2145, + "step": 11186 + }, + { + "epoch": 4.68, + "learning_rate": 7.734431839700747e-06, + "loss": 0.1557, + "step": 11187 + }, + { + "epoch": 4.68, + "learning_rate": 7.732780547494372e-06, + "loss": 0.1628, + "step": 11188 + }, + { + "epoch": 4.68, + "learning_rate": 7.731129320457205e-06, + "loss": 0.1719, + "step": 11189 + }, + { + "epoch": 4.68, + "learning_rate": 7.729478158636712e-06, + "loss": 0.1289, + "step": 11190 + }, + { + "epoch": 4.69, + "learning_rate": 7.727827062080359e-06, + "loss": 0.1182, + "step": 11191 + }, + { + "epoch": 4.69, + "learning_rate": 7.726176030835597e-06, + "loss": 0.1314, + "step": 11192 + }, + { + "epoch": 4.69, + "learning_rate": 7.724525064949887e-06, + "loss": 0.1119, + "step": 11193 + }, + { + "epoch": 4.69, + "learning_rate": 7.722874164470685e-06, + "loss": 0.1527, + "step": 11194 + }, + { + "epoch": 4.69, + "learning_rate": 7.721223329445448e-06, + "loss": 0.1525, + "step": 11195 + }, + { + "epoch": 4.69, + "learning_rate": 7.719572559921619e-06, + "loss": 0.177, + "step": 11196 + }, + { + "epoch": 4.69, + "learning_rate": 7.717921855946656e-06, + "loss": 0.1636, + "step": 11197 + }, + { + "epoch": 4.69, + "learning_rate": 7.716271217568005e-06, + "loss": 0.1738, + "step": 11198 + }, + { + "epoch": 4.69, + "learning_rate": 7.71462064483311e-06, + "loss": 0.1189, + "step": 11199 + }, + { + "epoch": 4.69, + "learning_rate": 7.712970137789416e-06, + "loss": 0.1319, + "step": 11200 + }, + { + "epoch": 4.69, + "learning_rate": 7.711319696484368e-06, + "loss": 0.1306, + "step": 11201 + }, + { + "epoch": 4.69, + "learning_rate": 7.709669320965402e-06, + "loss": 0.1446, + "step": 11202 + }, + { + "epoch": 4.69, + "learning_rate": 7.70801901127996e-06, + "loss": 0.1587, + "step": 11203 + }, + { + "epoch": 4.69, + "learning_rate": 7.70636876747548e-06, + "loss": 0.1619, + "step": 11204 + }, + { + "epoch": 4.69, + "learning_rate": 7.70471858959939e-06, + "loss": 0.1482, + "step": 11205 + }, + { + "epoch": 4.69, + "learning_rate": 7.70306847769913e-06, + "loss": 0.1366, + "step": 11206 + }, + { + "epoch": 4.69, + "learning_rate": 7.70141843182213e-06, + "loss": 0.1257, + "step": 11207 + }, + { + "epoch": 4.69, + "learning_rate": 7.699768452015815e-06, + "loss": 0.151, + "step": 11208 + }, + { + "epoch": 4.69, + "learning_rate": 7.698118538327613e-06, + "loss": 0.1567, + "step": 11209 + }, + { + "epoch": 4.69, + "learning_rate": 7.696468690804956e-06, + "loss": 0.1596, + "step": 11210 + }, + { + "epoch": 4.69, + "learning_rate": 7.69481890949526e-06, + "loss": 0.1584, + "step": 11211 + }, + { + "epoch": 4.69, + "learning_rate": 7.69316919444595e-06, + "loss": 0.1776, + "step": 11212 + }, + { + "epoch": 4.69, + "learning_rate": 7.691519545704442e-06, + "loss": 0.1568, + "step": 11213 + }, + { + "epoch": 4.69, + "learning_rate": 7.689869963318154e-06, + "loss": 0.1598, + "step": 11214 + }, + { + "epoch": 4.7, + "learning_rate": 7.688220447334508e-06, + "loss": 0.1739, + "step": 11215 + }, + { + "epoch": 4.7, + "learning_rate": 7.686570997800914e-06, + "loss": 0.1726, + "step": 11216 + }, + { + "epoch": 4.7, + "learning_rate": 7.68492161476478e-06, + "loss": 0.1606, + "step": 11217 + }, + { + "epoch": 4.7, + "learning_rate": 7.683272298273525e-06, + "loss": 0.1754, + "step": 11218 + }, + { + "epoch": 4.7, + "learning_rate": 7.681623048374549e-06, + "loss": 0.1424, + "step": 11219 + }, + { + "epoch": 4.7, + "learning_rate": 7.67997386511526e-06, + "loss": 0.2035, + "step": 11220 + }, + { + "epoch": 4.7, + "learning_rate": 7.678324748543066e-06, + "loss": 0.2215, + "step": 11221 + }, + { + "epoch": 4.7, + "learning_rate": 7.676675698705364e-06, + "loss": 0.1417, + "step": 11222 + }, + { + "epoch": 4.7, + "learning_rate": 7.67502671564956e-06, + "loss": 0.1331, + "step": 11223 + }, + { + "epoch": 4.7, + "learning_rate": 7.673377799423047e-06, + "loss": 0.1634, + "step": 11224 + }, + { + "epoch": 4.7, + "learning_rate": 7.671728950073228e-06, + "loss": 0.1274, + "step": 11225 + }, + { + "epoch": 4.7, + "learning_rate": 7.670080167647492e-06, + "loss": 0.1679, + "step": 11226 + }, + { + "epoch": 4.7, + "learning_rate": 7.668431452193234e-06, + "loss": 0.1328, + "step": 11227 + }, + { + "epoch": 4.7, + "learning_rate": 7.666782803757848e-06, + "loss": 0.1429, + "step": 11228 + }, + { + "epoch": 4.7, + "learning_rate": 7.665134222388717e-06, + "loss": 0.1673, + "step": 11229 + }, + { + "epoch": 4.7, + "learning_rate": 7.66348570813323e-06, + "loss": 0.1549, + "step": 11230 + }, + { + "epoch": 4.7, + "learning_rate": 7.661837261038776e-06, + "loss": 0.1568, + "step": 11231 + }, + { + "epoch": 4.7, + "learning_rate": 7.660188881152732e-06, + "loss": 0.1671, + "step": 11232 + }, + { + "epoch": 4.7, + "learning_rate": 7.658540568522487e-06, + "loss": 0.1755, + "step": 11233 + }, + { + "epoch": 4.7, + "learning_rate": 7.656892323195414e-06, + "loss": 0.1713, + "step": 11234 + }, + { + "epoch": 4.7, + "learning_rate": 7.655244145218887e-06, + "loss": 0.1667, + "step": 11235 + }, + { + "epoch": 4.7, + "learning_rate": 7.653596034640291e-06, + "loss": 0.166, + "step": 11236 + }, + { + "epoch": 4.7, + "learning_rate": 7.651947991506994e-06, + "loss": 0.1592, + "step": 11237 + }, + { + "epoch": 4.71, + "learning_rate": 7.650300015866367e-06, + "loss": 0.1462, + "step": 11238 + }, + { + "epoch": 4.71, + "learning_rate": 7.648652107765783e-06, + "loss": 0.1501, + "step": 11239 + }, + { + "epoch": 4.71, + "learning_rate": 7.647004267252606e-06, + "loss": 0.1555, + "step": 11240 + }, + { + "epoch": 4.71, + "learning_rate": 7.645356494374203e-06, + "loss": 0.1119, + "step": 11241 + }, + { + "epoch": 4.71, + "learning_rate": 7.643708789177939e-06, + "loss": 0.1433, + "step": 11242 + }, + { + "epoch": 4.71, + "learning_rate": 7.642061151711176e-06, + "loss": 0.1535, + "step": 11243 + }, + { + "epoch": 4.71, + "learning_rate": 7.64041358202127e-06, + "loss": 0.1613, + "step": 11244 + }, + { + "epoch": 4.71, + "learning_rate": 7.638766080155584e-06, + "loss": 0.1381, + "step": 11245 + }, + { + "epoch": 4.71, + "learning_rate": 7.63711864616147e-06, + "loss": 0.1383, + "step": 11246 + }, + { + "epoch": 4.71, + "learning_rate": 7.635471280086285e-06, + "loss": 0.1832, + "step": 11247 + }, + { + "epoch": 4.71, + "learning_rate": 7.63382398197738e-06, + "loss": 0.1316, + "step": 11248 + }, + { + "epoch": 4.71, + "learning_rate": 7.632176751882103e-06, + "loss": 0.1394, + "step": 11249 + }, + { + "epoch": 4.71, + "learning_rate": 7.630529589847807e-06, + "loss": 0.1282, + "step": 11250 + }, + { + "epoch": 4.71, + "learning_rate": 7.628882495921834e-06, + "loss": 0.1677, + "step": 11251 + }, + { + "epoch": 4.71, + "learning_rate": 7.627235470151528e-06, + "loss": 0.2038, + "step": 11252 + }, + { + "epoch": 4.71, + "learning_rate": 7.625588512584234e-06, + "loss": 0.1632, + "step": 11253 + }, + { + "epoch": 4.71, + "learning_rate": 7.6239416232672904e-06, + "loss": 0.1553, + "step": 11254 + }, + { + "epoch": 4.71, + "learning_rate": 7.622294802248041e-06, + "loss": 0.1566, + "step": 11255 + }, + { + "epoch": 4.71, + "learning_rate": 7.620648049573815e-06, + "loss": 0.1448, + "step": 11256 + }, + { + "epoch": 4.71, + "learning_rate": 7.619001365291948e-06, + "loss": 0.141, + "step": 11257 + }, + { + "epoch": 4.71, + "learning_rate": 7.617354749449774e-06, + "loss": 0.1419, + "step": 11258 + }, + { + "epoch": 4.71, + "learning_rate": 7.615708202094625e-06, + "loss": 0.1363, + "step": 11259 + }, + { + "epoch": 4.71, + "learning_rate": 7.614061723273825e-06, + "loss": 0.1554, + "step": 11260 + }, + { + "epoch": 4.71, + "learning_rate": 7.612415313034706e-06, + "loss": 0.1467, + "step": 11261 + }, + { + "epoch": 4.72, + "learning_rate": 7.61076897142459e-06, + "loss": 0.1473, + "step": 11262 + }, + { + "epoch": 4.72, + "learning_rate": 7.6091226984907985e-06, + "loss": 0.1502, + "step": 11263 + }, + { + "epoch": 4.72, + "learning_rate": 7.607476494280653e-06, + "loss": 0.1061, + "step": 11264 + }, + { + "epoch": 4.72, + "learning_rate": 7.6058303588414725e-06, + "loss": 0.2, + "step": 11265 + }, + { + "epoch": 4.72, + "learning_rate": 7.604184292220577e-06, + "loss": 0.1315, + "step": 11266 + }, + { + "epoch": 4.72, + "learning_rate": 7.602538294465273e-06, + "loss": 0.1253, + "step": 11267 + }, + { + "epoch": 4.72, + "learning_rate": 7.6008923656228814e-06, + "loss": 0.1226, + "step": 11268 + }, + { + "epoch": 4.72, + "learning_rate": 7.599246505740711e-06, + "loss": 0.1397, + "step": 11269 + }, + { + "epoch": 4.72, + "learning_rate": 7.597600714866067e-06, + "loss": 0.1627, + "step": 11270 + }, + { + "epoch": 4.72, + "learning_rate": 7.595954993046259e-06, + "loss": 0.1419, + "step": 11271 + }, + { + "epoch": 4.72, + "learning_rate": 7.594309340328594e-06, + "loss": 0.1498, + "step": 11272 + }, + { + "epoch": 4.72, + "learning_rate": 7.592663756760369e-06, + "loss": 0.1064, + "step": 11273 + }, + { + "epoch": 4.72, + "learning_rate": 7.591018242388888e-06, + "loss": 0.1151, + "step": 11274 + }, + { + "epoch": 4.72, + "learning_rate": 7.589372797261453e-06, + "loss": 0.1374, + "step": 11275 + }, + { + "epoch": 4.72, + "learning_rate": 7.587727421425355e-06, + "loss": 0.1407, + "step": 11276 + }, + { + "epoch": 4.72, + "learning_rate": 7.586082114927892e-06, + "loss": 0.1609, + "step": 11277 + }, + { + "epoch": 4.72, + "learning_rate": 7.58443687781636e-06, + "loss": 0.1706, + "step": 11278 + }, + { + "epoch": 4.72, + "learning_rate": 7.582791710138043e-06, + "loss": 0.1377, + "step": 11279 + }, + { + "epoch": 4.72, + "learning_rate": 7.581146611940235e-06, + "loss": 0.1213, + "step": 11280 + }, + { + "epoch": 4.72, + "learning_rate": 7.57950158327022e-06, + "loss": 0.1416, + "step": 11281 + }, + { + "epoch": 4.72, + "learning_rate": 7.577856624175283e-06, + "loss": 0.1573, + "step": 11282 + }, + { + "epoch": 4.72, + "learning_rate": 7.576211734702711e-06, + "loss": 0.155, + "step": 11283 + }, + { + "epoch": 4.72, + "learning_rate": 7.574566914899779e-06, + "loss": 0.0967, + "step": 11284 + }, + { + "epoch": 4.72, + "learning_rate": 7.5729221648137686e-06, + "loss": 0.1409, + "step": 11285 + }, + { + "epoch": 4.73, + "learning_rate": 7.571277484491959e-06, + "loss": 0.1336, + "step": 11286 + }, + { + "epoch": 4.73, + "learning_rate": 7.569632873981622e-06, + "loss": 0.1512, + "step": 11287 + }, + { + "epoch": 4.73, + "learning_rate": 7.56798833333003e-06, + "loss": 0.1534, + "step": 11288 + }, + { + "epoch": 4.73, + "learning_rate": 7.566343862584458e-06, + "loss": 0.1724, + "step": 11289 + }, + { + "epoch": 4.73, + "learning_rate": 7.56469946179217e-06, + "loss": 0.129, + "step": 11290 + }, + { + "epoch": 4.73, + "learning_rate": 7.563055131000434e-06, + "loss": 0.1654, + "step": 11291 + }, + { + "epoch": 4.73, + "learning_rate": 7.561410870256518e-06, + "loss": 0.1414, + "step": 11292 + }, + { + "epoch": 4.73, + "learning_rate": 7.559766679607682e-06, + "loss": 0.1662, + "step": 11293 + }, + { + "epoch": 4.73, + "learning_rate": 7.5581225591011864e-06, + "loss": 0.1258, + "step": 11294 + }, + { + "epoch": 4.73, + "learning_rate": 7.556478508784291e-06, + "loss": 0.1049, + "step": 11295 + }, + { + "epoch": 4.73, + "learning_rate": 7.554834528704254e-06, + "loss": 0.1295, + "step": 11296 + }, + { + "epoch": 4.73, + "learning_rate": 7.553190618908327e-06, + "loss": 0.0941, + "step": 11297 + }, + { + "epoch": 4.73, + "learning_rate": 7.551546779443765e-06, + "loss": 0.0995, + "step": 11298 + }, + { + "epoch": 4.73, + "learning_rate": 7.54990301035782e-06, + "loss": 0.1157, + "step": 11299 + }, + { + "epoch": 4.73, + "learning_rate": 7.5482593116977355e-06, + "loss": 0.1149, + "step": 11300 + }, + { + "epoch": 4.73, + "learning_rate": 7.546615683510767e-06, + "loss": 0.1697, + "step": 11301 + }, + { + "epoch": 4.73, + "learning_rate": 7.544972125844149e-06, + "loss": 0.1343, + "step": 11302 + }, + { + "epoch": 4.73, + "learning_rate": 7.543328638745131e-06, + "loss": 0.099, + "step": 11303 + }, + { + "epoch": 4.73, + "learning_rate": 7.541685222260948e-06, + "loss": 0.1647, + "step": 11304 + }, + { + "epoch": 4.73, + "learning_rate": 7.540041876438842e-06, + "loss": 0.1305, + "step": 11305 + }, + { + "epoch": 4.73, + "learning_rate": 7.538398601326051e-06, + "loss": 0.1737, + "step": 11306 + }, + { + "epoch": 4.73, + "learning_rate": 7.5367553969698095e-06, + "loss": 0.1746, + "step": 11307 + }, + { + "epoch": 4.73, + "learning_rate": 7.535112263417346e-06, + "loss": 0.1647, + "step": 11308 + }, + { + "epoch": 4.73, + "learning_rate": 7.533469200715892e-06, + "loss": 0.1599, + "step": 11309 + }, + { + "epoch": 4.74, + "learning_rate": 7.53182620891268e-06, + "loss": 0.1466, + "step": 11310 + }, + { + "epoch": 4.74, + "learning_rate": 7.530183288054931e-06, + "loss": 0.1535, + "step": 11311 + }, + { + "epoch": 4.74, + "learning_rate": 7.528540438189872e-06, + "loss": 0.1489, + "step": 11312 + }, + { + "epoch": 4.74, + "learning_rate": 7.526897659364727e-06, + "loss": 0.1699, + "step": 11313 + }, + { + "epoch": 4.74, + "learning_rate": 7.525254951626712e-06, + "loss": 0.106, + "step": 11314 + }, + { + "epoch": 4.74, + "learning_rate": 7.523612315023048e-06, + "loss": 0.1492, + "step": 11315 + }, + { + "epoch": 4.74, + "learning_rate": 7.5219697496009525e-06, + "loss": 0.1654, + "step": 11316 + }, + { + "epoch": 4.74, + "learning_rate": 7.520327255407635e-06, + "loss": 0.1983, + "step": 11317 + }, + { + "epoch": 4.74, + "learning_rate": 7.518684832490311e-06, + "loss": 0.167, + "step": 11318 + }, + { + "epoch": 4.74, + "learning_rate": 7.51704248089619e-06, + "loss": 0.1525, + "step": 11319 + }, + { + "epoch": 4.74, + "learning_rate": 7.515400200672478e-06, + "loss": 0.1334, + "step": 11320 + }, + { + "epoch": 4.74, + "learning_rate": 7.513757991866383e-06, + "loss": 0.1492, + "step": 11321 + }, + { + "epoch": 4.74, + "learning_rate": 7.5121158545251085e-06, + "loss": 0.1698, + "step": 11322 + }, + { + "epoch": 4.74, + "learning_rate": 7.51047378869586e-06, + "loss": 0.1589, + "step": 11323 + }, + { + "epoch": 4.74, + "learning_rate": 7.5088317944258305e-06, + "loss": 0.1427, + "step": 11324 + }, + { + "epoch": 4.74, + "learning_rate": 7.50718987176222e-06, + "loss": 0.1504, + "step": 11325 + }, + { + "epoch": 4.74, + "learning_rate": 7.5055480207522225e-06, + "loss": 0.1366, + "step": 11326 + }, + { + "epoch": 4.74, + "learning_rate": 7.503906241443037e-06, + "loss": 0.1122, + "step": 11327 + }, + { + "epoch": 4.74, + "learning_rate": 7.502264533881849e-06, + "loss": 0.1707, + "step": 11328 + }, + { + "epoch": 4.74, + "learning_rate": 7.50062289811585e-06, + "loss": 0.1653, + "step": 11329 + }, + { + "epoch": 4.74, + "learning_rate": 7.49898133419223e-06, + "loss": 0.1242, + "step": 11330 + }, + { + "epoch": 4.74, + "learning_rate": 7.497339842158171e-06, + "loss": 0.1779, + "step": 11331 + }, + { + "epoch": 4.74, + "learning_rate": 7.495698422060857e-06, + "loss": 0.1762, + "step": 11332 + }, + { + "epoch": 4.74, + "learning_rate": 7.494057073947473e-06, + "loss": 0.1771, + "step": 11333 + }, + { + "epoch": 4.75, + "learning_rate": 7.492415797865191e-06, + "loss": 0.1462, + "step": 11334 + }, + { + "epoch": 4.75, + "learning_rate": 7.490774593861192e-06, + "loss": 0.1265, + "step": 11335 + }, + { + "epoch": 4.75, + "learning_rate": 7.489133461982651e-06, + "loss": 0.1422, + "step": 11336 + }, + { + "epoch": 4.75, + "learning_rate": 7.487492402276742e-06, + "loss": 0.1407, + "step": 11337 + }, + { + "epoch": 4.75, + "learning_rate": 7.485851414790632e-06, + "loss": 0.1442, + "step": 11338 + }, + { + "epoch": 4.75, + "learning_rate": 7.4842104995714935e-06, + "loss": 0.2042, + "step": 11339 + }, + { + "epoch": 4.75, + "learning_rate": 7.482569656666493e-06, + "loss": 0.1364, + "step": 11340 + }, + { + "epoch": 4.75, + "learning_rate": 7.480928886122792e-06, + "loss": 0.1347, + "step": 11341 + }, + { + "epoch": 4.75, + "learning_rate": 7.479288187987554e-06, + "loss": 0.1625, + "step": 11342 + }, + { + "epoch": 4.75, + "learning_rate": 7.477647562307943e-06, + "loss": 0.147, + "step": 11343 + }, + { + "epoch": 4.75, + "learning_rate": 7.476007009131112e-06, + "loss": 0.1096, + "step": 11344 + }, + { + "epoch": 4.75, + "learning_rate": 7.47436652850422e-06, + "loss": 0.1525, + "step": 11345 + }, + { + "epoch": 4.75, + "learning_rate": 7.472726120474425e-06, + "loss": 0.1605, + "step": 11346 + }, + { + "epoch": 4.75, + "learning_rate": 7.471085785088871e-06, + "loss": 0.1622, + "step": 11347 + }, + { + "epoch": 4.75, + "learning_rate": 7.469445522394714e-06, + "loss": 0.16, + "step": 11348 + }, + { + "epoch": 4.75, + "learning_rate": 7.467805332439098e-06, + "loss": 0.1244, + "step": 11349 + }, + { + "epoch": 4.75, + "learning_rate": 7.46616521526917e-06, + "loss": 0.1675, + "step": 11350 + }, + { + "epoch": 4.75, + "learning_rate": 7.4645251709320775e-06, + "loss": 0.1349, + "step": 11351 + }, + { + "epoch": 4.75, + "learning_rate": 7.462885199474956e-06, + "loss": 0.164, + "step": 11352 + }, + { + "epoch": 4.75, + "learning_rate": 7.4612453009449485e-06, + "loss": 0.1653, + "step": 11353 + }, + { + "epoch": 4.75, + "learning_rate": 7.459605475389193e-06, + "loss": 0.1166, + "step": 11354 + }, + { + "epoch": 4.75, + "learning_rate": 7.457965722854823e-06, + "loss": 0.1068, + "step": 11355 + }, + { + "epoch": 4.75, + "learning_rate": 7.4563260433889726e-06, + "loss": 0.1797, + "step": 11356 + }, + { + "epoch": 4.75, + "learning_rate": 7.454686437038775e-06, + "loss": 0.1814, + "step": 11357 + }, + { + "epoch": 4.76, + "learning_rate": 7.4530469038513545e-06, + "loss": 0.1696, + "step": 11358 + }, + { + "epoch": 4.76, + "learning_rate": 7.45140744387384e-06, + "loss": 0.1331, + "step": 11359 + }, + { + "epoch": 4.76, + "learning_rate": 7.44976805715336e-06, + "loss": 0.1058, + "step": 11360 + }, + { + "epoch": 4.76, + "learning_rate": 7.448128743737032e-06, + "loss": 0.1342, + "step": 11361 + }, + { + "epoch": 4.76, + "learning_rate": 7.446489503671979e-06, + "loss": 0.1599, + "step": 11362 + }, + { + "epoch": 4.76, + "learning_rate": 7.4448503370053204e-06, + "loss": 0.1291, + "step": 11363 + }, + { + "epoch": 4.76, + "learning_rate": 7.44321124378417e-06, + "loss": 0.1425, + "step": 11364 + }, + { + "epoch": 4.76, + "learning_rate": 7.441572224055644e-06, + "loss": 0.186, + "step": 11365 + }, + { + "epoch": 4.76, + "learning_rate": 7.4399332778668535e-06, + "loss": 0.1956, + "step": 11366 + }, + { + "epoch": 4.76, + "learning_rate": 7.438294405264912e-06, + "loss": 0.2012, + "step": 11367 + }, + { + "epoch": 4.76, + "learning_rate": 7.436655606296921e-06, + "loss": 0.1178, + "step": 11368 + }, + { + "epoch": 4.76, + "learning_rate": 7.4350168810099955e-06, + "loss": 0.1288, + "step": 11369 + }, + { + "epoch": 4.76, + "learning_rate": 7.433378229451229e-06, + "loss": 0.1542, + "step": 11370 + }, + { + "epoch": 4.76, + "learning_rate": 7.431739651667729e-06, + "loss": 0.1506, + "step": 11371 + }, + { + "epoch": 4.76, + "learning_rate": 7.430101147706592e-06, + "loss": 0.2029, + "step": 11372 + }, + { + "epoch": 4.76, + "learning_rate": 7.428462717614917e-06, + "loss": 0.1316, + "step": 11373 + }, + { + "epoch": 4.76, + "learning_rate": 7.426824361439803e-06, + "loss": 0.1284, + "step": 11374 + }, + { + "epoch": 4.76, + "learning_rate": 7.425186079228335e-06, + "loss": 0.1285, + "step": 11375 + }, + { + "epoch": 4.76, + "learning_rate": 7.42354787102761e-06, + "loss": 0.1689, + "step": 11376 + }, + { + "epoch": 4.76, + "learning_rate": 7.421909736884714e-06, + "loss": 0.1667, + "step": 11377 + }, + { + "epoch": 4.76, + "learning_rate": 7.420271676846737e-06, + "loss": 0.1947, + "step": 11378 + }, + { + "epoch": 4.76, + "learning_rate": 7.4186336909607594e-06, + "loss": 0.1408, + "step": 11379 + }, + { + "epoch": 4.76, + "learning_rate": 7.416995779273866e-06, + "loss": 0.1398, + "step": 11380 + }, + { + "epoch": 4.76, + "learning_rate": 7.415357941833139e-06, + "loss": 0.1497, + "step": 11381 + }, + { + "epoch": 4.77, + "learning_rate": 7.413720178685653e-06, + "loss": 0.1695, + "step": 11382 + }, + { + "epoch": 4.77, + "learning_rate": 7.412082489878484e-06, + "loss": 0.1638, + "step": 11383 + }, + { + "epoch": 4.77, + "learning_rate": 7.41044487545871e-06, + "loss": 0.1852, + "step": 11384 + }, + { + "epoch": 4.77, + "learning_rate": 7.408807335473398e-06, + "loss": 0.1481, + "step": 11385 + }, + { + "epoch": 4.77, + "learning_rate": 7.40716986996962e-06, + "loss": 0.1608, + "step": 11386 + }, + { + "epoch": 4.77, + "learning_rate": 7.4055324789944446e-06, + "loss": 0.1231, + "step": 11387 + }, + { + "epoch": 4.77, + "learning_rate": 7.403895162594934e-06, + "loss": 0.1494, + "step": 11388 + }, + { + "epoch": 4.77, + "learning_rate": 7.402257920818154e-06, + "loss": 0.1702, + "step": 11389 + }, + { + "epoch": 4.77, + "learning_rate": 7.400620753711167e-06, + "loss": 0.1696, + "step": 11390 + }, + { + "epoch": 4.77, + "learning_rate": 7.398983661321029e-06, + "loss": 0.1542, + "step": 11391 + }, + { + "epoch": 4.77, + "learning_rate": 7.3973466436947985e-06, + "loss": 0.1124, + "step": 11392 + }, + { + "epoch": 4.77, + "learning_rate": 7.395709700879525e-06, + "loss": 0.1556, + "step": 11393 + }, + { + "epoch": 4.77, + "learning_rate": 7.394072832922268e-06, + "loss": 0.1859, + "step": 11394 + }, + { + "epoch": 4.77, + "learning_rate": 7.3924360398700745e-06, + "loss": 0.1514, + "step": 11395 + }, + { + "epoch": 4.77, + "learning_rate": 7.390799321769993e-06, + "loss": 0.1577, + "step": 11396 + }, + { + "epoch": 4.77, + "learning_rate": 7.389162678669069e-06, + "loss": 0.1659, + "step": 11397 + }, + { + "epoch": 4.77, + "learning_rate": 7.387526110614347e-06, + "loss": 0.1558, + "step": 11398 + }, + { + "epoch": 4.77, + "learning_rate": 7.385889617652868e-06, + "loss": 0.1791, + "step": 11399 + }, + { + "epoch": 4.77, + "learning_rate": 7.384253199831674e-06, + "loss": 0.1584, + "step": 11400 + }, + { + "epoch": 4.77, + "learning_rate": 7.382616857197801e-06, + "loss": 0.1717, + "step": 11401 + }, + { + "epoch": 4.77, + "learning_rate": 7.380980589798281e-06, + "loss": 0.1607, + "step": 11402 + }, + { + "epoch": 4.77, + "learning_rate": 7.379344397680151e-06, + "loss": 0.1706, + "step": 11403 + }, + { + "epoch": 4.77, + "learning_rate": 7.3777082808904435e-06, + "loss": 0.1739, + "step": 11404 + }, + { + "epoch": 4.77, + "learning_rate": 7.376072239476182e-06, + "loss": 0.1403, + "step": 11405 + }, + { + "epoch": 4.78, + "learning_rate": 7.374436273484395e-06, + "loss": 0.1492, + "step": 11406 + }, + { + "epoch": 4.78, + "learning_rate": 7.372800382962107e-06, + "loss": 0.1787, + "step": 11407 + }, + { + "epoch": 4.78, + "learning_rate": 7.371164567956344e-06, + "loss": 0.155, + "step": 11408 + }, + { + "epoch": 4.78, + "learning_rate": 7.36952882851412e-06, + "loss": 0.1437, + "step": 11409 + }, + { + "epoch": 4.78, + "learning_rate": 7.367893164682456e-06, + "loss": 0.1398, + "step": 11410 + }, + { + "epoch": 4.78, + "learning_rate": 7.366257576508369e-06, + "loss": 0.1301, + "step": 11411 + }, + { + "epoch": 4.78, + "learning_rate": 7.364622064038868e-06, + "loss": 0.1722, + "step": 11412 + }, + { + "epoch": 4.78, + "learning_rate": 7.362986627320967e-06, + "loss": 0.1482, + "step": 11413 + }, + { + "epoch": 4.78, + "learning_rate": 7.361351266401682e-06, + "loss": 0.1344, + "step": 11414 + }, + { + "epoch": 4.78, + "learning_rate": 7.359715981328009e-06, + "loss": 0.1519, + "step": 11415 + }, + { + "epoch": 4.78, + "learning_rate": 7.358080772146956e-06, + "loss": 0.1482, + "step": 11416 + }, + { + "epoch": 4.78, + "learning_rate": 7.356445638905526e-06, + "loss": 0.1776, + "step": 11417 + }, + { + "epoch": 4.78, + "learning_rate": 7.354810581650721e-06, + "loss": 0.1388, + "step": 11418 + }, + { + "epoch": 4.78, + "learning_rate": 7.35317560042954e-06, + "loss": 0.1726, + "step": 11419 + }, + { + "epoch": 4.78, + "learning_rate": 7.351540695288977e-06, + "loss": 0.1506, + "step": 11420 + }, + { + "epoch": 4.78, + "learning_rate": 7.349905866276025e-06, + "loss": 0.1139, + "step": 11421 + }, + { + "epoch": 4.78, + "learning_rate": 7.348271113437679e-06, + "loss": 0.1037, + "step": 11422 + }, + { + "epoch": 4.78, + "learning_rate": 7.346636436820926e-06, + "loss": 0.1677, + "step": 11423 + }, + { + "epoch": 4.78, + "learning_rate": 7.345001836472754e-06, + "loss": 0.1564, + "step": 11424 + }, + { + "epoch": 4.78, + "learning_rate": 7.3433673124401495e-06, + "loss": 0.1574, + "step": 11425 + }, + { + "epoch": 4.78, + "learning_rate": 7.341732864770093e-06, + "loss": 0.1255, + "step": 11426 + }, + { + "epoch": 4.78, + "learning_rate": 7.340098493509565e-06, + "loss": 0.1382, + "step": 11427 + }, + { + "epoch": 4.78, + "learning_rate": 7.338464198705548e-06, + "loss": 0.1807, + "step": 11428 + }, + { + "epoch": 4.79, + "learning_rate": 7.336829980405014e-06, + "loss": 0.1531, + "step": 11429 + }, + { + "epoch": 4.79, + "learning_rate": 7.335195838654939e-06, + "loss": 0.1694, + "step": 11430 + }, + { + "epoch": 4.79, + "learning_rate": 7.333561773502297e-06, + "loss": 0.1758, + "step": 11431 + }, + { + "epoch": 4.79, + "learning_rate": 7.3319277849940525e-06, + "loss": 0.1775, + "step": 11432 + }, + { + "epoch": 4.79, + "learning_rate": 7.330293873177179e-06, + "loss": 0.1232, + "step": 11433 + }, + { + "epoch": 4.79, + "learning_rate": 7.328660038098636e-06, + "loss": 0.153, + "step": 11434 + }, + { + "epoch": 4.79, + "learning_rate": 7.327026279805394e-06, + "loss": 0.1797, + "step": 11435 + }, + { + "epoch": 4.79, + "learning_rate": 7.325392598344406e-06, + "loss": 0.1842, + "step": 11436 + }, + { + "epoch": 4.79, + "learning_rate": 7.3237589937626395e-06, + "loss": 0.1748, + "step": 11437 + }, + { + "epoch": 4.79, + "learning_rate": 7.322125466107042e-06, + "loss": 0.1729, + "step": 11438 + }, + { + "epoch": 4.79, + "learning_rate": 7.320492015424572e-06, + "loss": 0.1727, + "step": 11439 + }, + { + "epoch": 4.79, + "learning_rate": 7.318858641762181e-06, + "loss": 0.1761, + "step": 11440 + }, + { + "epoch": 4.79, + "learning_rate": 7.317225345166819e-06, + "loss": 0.1366, + "step": 11441 + }, + { + "epoch": 4.79, + "learning_rate": 7.315592125685434e-06, + "loss": 0.1397, + "step": 11442 + }, + { + "epoch": 4.79, + "learning_rate": 7.31395898336497e-06, + "loss": 0.107, + "step": 11443 + }, + { + "epoch": 4.79, + "learning_rate": 7.312325918252373e-06, + "loss": 0.1468, + "step": 11444 + }, + { + "epoch": 4.79, + "learning_rate": 7.310692930394583e-06, + "loss": 0.1194, + "step": 11445 + }, + { + "epoch": 4.79, + "learning_rate": 7.309060019838536e-06, + "loss": 0.1522, + "step": 11446 + }, + { + "epoch": 4.79, + "learning_rate": 7.307427186631173e-06, + "loss": 0.0958, + "step": 11447 + }, + { + "epoch": 4.79, + "learning_rate": 7.305794430819424e-06, + "loss": 0.1547, + "step": 11448 + }, + { + "epoch": 4.79, + "learning_rate": 7.304161752450227e-06, + "loss": 0.1457, + "step": 11449 + }, + { + "epoch": 4.79, + "learning_rate": 7.302529151570506e-06, + "loss": 0.1221, + "step": 11450 + }, + { + "epoch": 4.79, + "learning_rate": 7.3008966282271885e-06, + "loss": 0.1433, + "step": 11451 + }, + { + "epoch": 4.79, + "learning_rate": 7.299264182467206e-06, + "loss": 0.1421, + "step": 11452 + }, + { + "epoch": 4.8, + "learning_rate": 7.2976318143374755e-06, + "loss": 0.174, + "step": 11453 + }, + { + "epoch": 4.8, + "learning_rate": 7.295999523884921e-06, + "loss": 0.1891, + "step": 11454 + }, + { + "epoch": 4.8, + "learning_rate": 7.294367311156463e-06, + "loss": 0.1325, + "step": 11455 + }, + { + "epoch": 4.8, + "learning_rate": 7.292735176199013e-06, + "loss": 0.1817, + "step": 11456 + }, + { + "epoch": 4.8, + "learning_rate": 7.291103119059488e-06, + "loss": 0.1543, + "step": 11457 + }, + { + "epoch": 4.8, + "learning_rate": 7.289471139784803e-06, + "loss": 0.1808, + "step": 11458 + }, + { + "epoch": 4.8, + "learning_rate": 7.287839238421863e-06, + "loss": 0.1374, + "step": 11459 + }, + { + "epoch": 4.8, + "learning_rate": 7.286207415017579e-06, + "loss": 0.1189, + "step": 11460 + }, + { + "epoch": 4.8, + "learning_rate": 7.284575669618852e-06, + "loss": 0.1296, + "step": 11461 + }, + { + "epoch": 4.8, + "learning_rate": 7.282944002272587e-06, + "loss": 0.1559, + "step": 11462 + }, + { + "epoch": 4.8, + "learning_rate": 7.28131241302569e-06, + "loss": 0.1616, + "step": 11463 + }, + { + "epoch": 4.8, + "learning_rate": 7.279680901925051e-06, + "loss": 0.1827, + "step": 11464 + }, + { + "epoch": 4.8, + "learning_rate": 7.278049469017569e-06, + "loss": 0.1433, + "step": 11465 + }, + { + "epoch": 4.8, + "learning_rate": 7.276418114350145e-06, + "loss": 0.133, + "step": 11466 + }, + { + "epoch": 4.8, + "learning_rate": 7.274786837969663e-06, + "loss": 0.169, + "step": 11467 + }, + { + "epoch": 4.8, + "learning_rate": 7.273155639923013e-06, + "loss": 0.1406, + "step": 11468 + }, + { + "epoch": 4.8, + "learning_rate": 7.271524520257087e-06, + "loss": 0.1203, + "step": 11469 + }, + { + "epoch": 4.8, + "learning_rate": 7.269893479018766e-06, + "loss": 0.1757, + "step": 11470 + }, + { + "epoch": 4.8, + "learning_rate": 7.268262516254934e-06, + "loss": 0.1237, + "step": 11471 + }, + { + "epoch": 4.8, + "learning_rate": 7.2666316320124755e-06, + "loss": 0.1722, + "step": 11472 + }, + { + "epoch": 4.8, + "learning_rate": 7.265000826338261e-06, + "loss": 0.1364, + "step": 11473 + }, + { + "epoch": 4.8, + "learning_rate": 7.263370099279173e-06, + "loss": 0.1386, + "step": 11474 + }, + { + "epoch": 4.8, + "learning_rate": 7.2617394508820815e-06, + "loss": 0.1343, + "step": 11475 + }, + { + "epoch": 4.8, + "learning_rate": 7.26010888119386e-06, + "loss": 0.1714, + "step": 11476 + }, + { + "epoch": 4.81, + "learning_rate": 7.258478390261377e-06, + "loss": 0.156, + "step": 11477 + }, + { + "epoch": 4.81, + "learning_rate": 7.2568479781315e-06, + "loss": 0.1641, + "step": 11478 + }, + { + "epoch": 4.81, + "learning_rate": 7.255217644851096e-06, + "loss": 0.1244, + "step": 11479 + }, + { + "epoch": 4.81, + "learning_rate": 7.253587390467023e-06, + "loss": 0.1458, + "step": 11480 + }, + { + "epoch": 4.81, + "learning_rate": 7.251957215026142e-06, + "loss": 0.1585, + "step": 11481 + }, + { + "epoch": 4.81, + "learning_rate": 7.250327118575318e-06, + "loss": 0.1757, + "step": 11482 + }, + { + "epoch": 4.81, + "learning_rate": 7.2486971011614e-06, + "loss": 0.1324, + "step": 11483 + }, + { + "epoch": 4.81, + "learning_rate": 7.247067162831237e-06, + "loss": 0.1536, + "step": 11484 + }, + { + "epoch": 4.81, + "learning_rate": 7.245437303631687e-06, + "loss": 0.2021, + "step": 11485 + }, + { + "epoch": 4.81, + "learning_rate": 7.243807523609601e-06, + "loss": 0.1557, + "step": 11486 + }, + { + "epoch": 4.81, + "learning_rate": 7.242177822811818e-06, + "loss": 0.1699, + "step": 11487 + }, + { + "epoch": 4.81, + "learning_rate": 7.240548201285186e-06, + "loss": 0.1632, + "step": 11488 + }, + { + "epoch": 4.81, + "learning_rate": 7.238918659076548e-06, + "loss": 0.1633, + "step": 11489 + }, + { + "epoch": 4.81, + "learning_rate": 7.237289196232744e-06, + "loss": 0.1728, + "step": 11490 + }, + { + "epoch": 4.81, + "learning_rate": 7.235659812800611e-06, + "loss": 0.1672, + "step": 11491 + }, + { + "epoch": 4.81, + "learning_rate": 7.234030508826981e-06, + "loss": 0.1507, + "step": 11492 + }, + { + "epoch": 4.81, + "learning_rate": 7.232401284358692e-06, + "loss": 0.174, + "step": 11493 + }, + { + "epoch": 4.81, + "learning_rate": 7.230772139442572e-06, + "loss": 0.0923, + "step": 11494 + }, + { + "epoch": 4.81, + "learning_rate": 7.229143074125447e-06, + "loss": 0.1453, + "step": 11495 + }, + { + "epoch": 4.81, + "learning_rate": 7.2275140884541495e-06, + "loss": 0.1484, + "step": 11496 + }, + { + "epoch": 4.81, + "learning_rate": 7.225885182475496e-06, + "loss": 0.138, + "step": 11497 + }, + { + "epoch": 4.81, + "learning_rate": 7.224256356236312e-06, + "loss": 0.1439, + "step": 11498 + }, + { + "epoch": 4.81, + "learning_rate": 7.222627609783418e-06, + "loss": 0.1385, + "step": 11499 + }, + { + "epoch": 4.81, + "learning_rate": 7.2209989431636265e-06, + "loss": 0.1192, + "step": 11500 + }, + { + "epoch": 4.82, + "learning_rate": 7.219370356423754e-06, + "loss": 0.1667, + "step": 11501 + }, + { + "epoch": 4.82, + "learning_rate": 7.217741849610616e-06, + "loss": 0.1536, + "step": 11502 + }, + { + "epoch": 4.82, + "learning_rate": 7.216113422771016e-06, + "loss": 0.142, + "step": 11503 + }, + { + "epoch": 4.82, + "learning_rate": 7.214485075951771e-06, + "loss": 0.1855, + "step": 11504 + }, + { + "epoch": 4.82, + "learning_rate": 7.212856809199676e-06, + "loss": 0.1364, + "step": 11505 + }, + { + "epoch": 4.82, + "learning_rate": 7.211228622561538e-06, + "loss": 0.1593, + "step": 11506 + }, + { + "epoch": 4.82, + "learning_rate": 7.2096005160841605e-06, + "loss": 0.153, + "step": 11507 + }, + { + "epoch": 4.82, + "learning_rate": 7.207972489814339e-06, + "loss": 0.1313, + "step": 11508 + }, + { + "epoch": 4.82, + "learning_rate": 7.206344543798871e-06, + "loss": 0.1861, + "step": 11509 + }, + { + "epoch": 4.82, + "learning_rate": 7.204716678084551e-06, + "loss": 0.17, + "step": 11510 + }, + { + "epoch": 4.82, + "learning_rate": 7.203088892718168e-06, + "loss": 0.1709, + "step": 11511 + }, + { + "epoch": 4.82, + "learning_rate": 7.201461187746513e-06, + "loss": 0.0981, + "step": 11512 + }, + { + "epoch": 4.82, + "learning_rate": 7.199833563216375e-06, + "loss": 0.1039, + "step": 11513 + }, + { + "epoch": 4.82, + "learning_rate": 7.198206019174534e-06, + "loss": 0.176, + "step": 11514 + }, + { + "epoch": 4.82, + "learning_rate": 7.196578555667775e-06, + "loss": 0.1032, + "step": 11515 + }, + { + "epoch": 4.82, + "learning_rate": 7.194951172742879e-06, + "loss": 0.1818, + "step": 11516 + }, + { + "epoch": 4.82, + "learning_rate": 7.19332387044662e-06, + "loss": 0.1415, + "step": 11517 + }, + { + "epoch": 4.82, + "learning_rate": 7.191696648825778e-06, + "loss": 0.1941, + "step": 11518 + }, + { + "epoch": 4.82, + "learning_rate": 7.190069507927122e-06, + "loss": 0.1256, + "step": 11519 + }, + { + "epoch": 4.82, + "learning_rate": 7.1884424477974265e-06, + "loss": 0.1336, + "step": 11520 + }, + { + "epoch": 4.82, + "learning_rate": 7.186815468483457e-06, + "loss": 0.1809, + "step": 11521 + }, + { + "epoch": 4.82, + "learning_rate": 7.185188570031979e-06, + "loss": 0.1438, + "step": 11522 + }, + { + "epoch": 4.82, + "learning_rate": 7.183561752489761e-06, + "loss": 0.1717, + "step": 11523 + }, + { + "epoch": 4.82, + "learning_rate": 7.181935015903558e-06, + "loss": 0.1621, + "step": 11524 + }, + { + "epoch": 4.83, + "learning_rate": 7.180308360320134e-06, + "loss": 0.1906, + "step": 11525 + }, + { + "epoch": 4.83, + "learning_rate": 7.178681785786245e-06, + "loss": 0.1621, + "step": 11526 + }, + { + "epoch": 4.83, + "learning_rate": 7.177055292348646e-06, + "loss": 0.1375, + "step": 11527 + }, + { + "epoch": 4.83, + "learning_rate": 7.1754288800540836e-06, + "loss": 0.1286, + "step": 11528 + }, + { + "epoch": 4.83, + "learning_rate": 7.173802548949312e-06, + "loss": 0.1323, + "step": 11529 + }, + { + "epoch": 4.83, + "learning_rate": 7.172176299081078e-06, + "loss": 0.1918, + "step": 11530 + }, + { + "epoch": 4.83, + "learning_rate": 7.17055013049613e-06, + "loss": 0.1785, + "step": 11531 + }, + { + "epoch": 4.83, + "learning_rate": 7.168924043241205e-06, + "loss": 0.1526, + "step": 11532 + }, + { + "epoch": 4.83, + "learning_rate": 7.1672980373630455e-06, + "loss": 0.1177, + "step": 11533 + }, + { + "epoch": 4.83, + "learning_rate": 7.165672112908393e-06, + "loss": 0.0849, + "step": 11534 + }, + { + "epoch": 4.83, + "learning_rate": 7.164046269923978e-06, + "loss": 0.2089, + "step": 11535 + }, + { + "epoch": 4.83, + "learning_rate": 7.162420508456536e-06, + "loss": 0.1855, + "step": 11536 + }, + { + "epoch": 4.83, + "learning_rate": 7.160794828552802e-06, + "loss": 0.1502, + "step": 11537 + }, + { + "epoch": 4.83, + "learning_rate": 7.159169230259497e-06, + "loss": 0.1467, + "step": 11538 + }, + { + "epoch": 4.83, + "learning_rate": 7.157543713623353e-06, + "loss": 0.1252, + "step": 11539 + }, + { + "epoch": 4.83, + "learning_rate": 7.155918278691094e-06, + "loss": 0.1651, + "step": 11540 + }, + { + "epoch": 4.83, + "learning_rate": 7.1542929255094386e-06, + "loss": 0.1559, + "step": 11541 + }, + { + "epoch": 4.83, + "learning_rate": 7.1526676541251075e-06, + "loss": 0.1774, + "step": 11542 + }, + { + "epoch": 4.83, + "learning_rate": 7.151042464584821e-06, + "loss": 0.1794, + "step": 11543 + }, + { + "epoch": 4.83, + "learning_rate": 7.149417356935289e-06, + "loss": 0.1324, + "step": 11544 + }, + { + "epoch": 4.83, + "learning_rate": 7.1477923312232235e-06, + "loss": 0.1458, + "step": 11545 + }, + { + "epoch": 4.83, + "learning_rate": 7.14616738749534e-06, + "loss": 0.1399, + "step": 11546 + }, + { + "epoch": 4.83, + "learning_rate": 7.1445425257983395e-06, + "loss": 0.15, + "step": 11547 + }, + { + "epoch": 4.83, + "learning_rate": 7.14291774617893e-06, + "loss": 0.1499, + "step": 11548 + }, + { + "epoch": 4.84, + "learning_rate": 7.141293048683816e-06, + "loss": 0.1482, + "step": 11549 + }, + { + "epoch": 4.84, + "learning_rate": 7.1396684333597e-06, + "loss": 0.1563, + "step": 11550 + }, + { + "epoch": 4.84, + "learning_rate": 7.138043900253275e-06, + "loss": 0.1151, + "step": 11551 + }, + { + "epoch": 4.84, + "learning_rate": 7.136419449411236e-06, + "loss": 0.1529, + "step": 11552 + }, + { + "epoch": 4.84, + "learning_rate": 7.134795080880278e-06, + "loss": 0.1641, + "step": 11553 + }, + { + "epoch": 4.84, + "learning_rate": 7.133170794707096e-06, + "loss": 0.1526, + "step": 11554 + }, + { + "epoch": 4.84, + "learning_rate": 7.131546590938373e-06, + "loss": 0.1107, + "step": 11555 + }, + { + "epoch": 4.84, + "learning_rate": 7.129922469620798e-06, + "loss": 0.1722, + "step": 11556 + }, + { + "epoch": 4.84, + "learning_rate": 7.128298430801057e-06, + "loss": 0.1138, + "step": 11557 + }, + { + "epoch": 4.84, + "learning_rate": 7.1266744745258264e-06, + "loss": 0.1476, + "step": 11558 + }, + { + "epoch": 4.84, + "learning_rate": 7.12505060084179e-06, + "loss": 0.1776, + "step": 11559 + }, + { + "epoch": 4.84, + "learning_rate": 7.123426809795622e-06, + "loss": 0.1628, + "step": 11560 + }, + { + "epoch": 4.84, + "learning_rate": 7.121803101433998e-06, + "loss": 0.1727, + "step": 11561 + }, + { + "epoch": 4.84, + "learning_rate": 7.1201794758035905e-06, + "loss": 0.1536, + "step": 11562 + }, + { + "epoch": 4.84, + "learning_rate": 7.118555932951067e-06, + "loss": 0.1326, + "step": 11563 + }, + { + "epoch": 4.84, + "learning_rate": 7.116932472923098e-06, + "loss": 0.1376, + "step": 11564 + }, + { + "epoch": 4.84, + "learning_rate": 7.115309095766345e-06, + "loss": 0.1419, + "step": 11565 + }, + { + "epoch": 4.84, + "learning_rate": 7.113685801527473e-06, + "loss": 0.111, + "step": 11566 + }, + { + "epoch": 4.84, + "learning_rate": 7.112062590253143e-06, + "loss": 0.1201, + "step": 11567 + }, + { + "epoch": 4.84, + "learning_rate": 7.110439461990008e-06, + "loss": 0.1694, + "step": 11568 + }, + { + "epoch": 4.84, + "learning_rate": 7.108816416784728e-06, + "loss": 0.1499, + "step": 11569 + }, + { + "epoch": 4.84, + "learning_rate": 7.107193454683956e-06, + "loss": 0.1669, + "step": 11570 + }, + { + "epoch": 4.84, + "learning_rate": 7.105570575734339e-06, + "loss": 0.1315, + "step": 11571 + }, + { + "epoch": 4.84, + "learning_rate": 7.103947779982532e-06, + "loss": 0.1531, + "step": 11572 + }, + { + "epoch": 4.85, + "learning_rate": 7.102325067475171e-06, + "loss": 0.13, + "step": 11573 + }, + { + "epoch": 4.85, + "learning_rate": 7.100702438258905e-06, + "loss": 0.1609, + "step": 11574 + }, + { + "epoch": 4.85, + "learning_rate": 7.099079892380376e-06, + "loss": 0.1277, + "step": 11575 + }, + { + "epoch": 4.85, + "learning_rate": 7.097457429886221e-06, + "loss": 0.1312, + "step": 11576 + }, + { + "epoch": 4.85, + "learning_rate": 7.095835050823074e-06, + "loss": 0.1627, + "step": 11577 + }, + { + "epoch": 4.85, + "learning_rate": 7.094212755237576e-06, + "loss": 0.1633, + "step": 11578 + }, + { + "epoch": 4.85, + "learning_rate": 7.092590543176351e-06, + "loss": 0.156, + "step": 11579 + }, + { + "epoch": 4.85, + "learning_rate": 7.09096841468603e-06, + "loss": 0.1543, + "step": 11580 + }, + { + "epoch": 4.85, + "learning_rate": 7.089346369813244e-06, + "loss": 0.1305, + "step": 11581 + }, + { + "epoch": 4.85, + "learning_rate": 7.087724408604611e-06, + "loss": 0.15, + "step": 11582 + }, + { + "epoch": 4.85, + "learning_rate": 7.086102531106755e-06, + "loss": 0.1845, + "step": 11583 + }, + { + "epoch": 4.85, + "learning_rate": 7.084480737366297e-06, + "loss": 0.1297, + "step": 11584 + }, + { + "epoch": 4.85, + "learning_rate": 7.082859027429853e-06, + "loss": 0.1236, + "step": 11585 + }, + { + "epoch": 4.85, + "learning_rate": 7.0812374013440365e-06, + "loss": 0.1535, + "step": 11586 + }, + { + "epoch": 4.85, + "learning_rate": 7.079615859155464e-06, + "loss": 0.1349, + "step": 11587 + }, + { + "epoch": 4.85, + "learning_rate": 7.077994400910739e-06, + "loss": 0.1956, + "step": 11588 + }, + { + "epoch": 4.85, + "learning_rate": 7.0763730266564725e-06, + "loss": 0.1305, + "step": 11589 + }, + { + "epoch": 4.85, + "learning_rate": 7.0747517364392694e-06, + "loss": 0.1581, + "step": 11590 + }, + { + "epoch": 4.85, + "learning_rate": 7.073130530305733e-06, + "loss": 0.149, + "step": 11591 + }, + { + "epoch": 4.85, + "learning_rate": 7.0715094083024604e-06, + "loss": 0.1399, + "step": 11592 + }, + { + "epoch": 4.85, + "learning_rate": 7.069888370476051e-06, + "loss": 0.174, + "step": 11593 + }, + { + "epoch": 4.85, + "learning_rate": 7.068267416873102e-06, + "loss": 0.1423, + "step": 11594 + }, + { + "epoch": 4.85, + "learning_rate": 7.066646547540208e-06, + "loss": 0.148, + "step": 11595 + }, + { + "epoch": 4.85, + "learning_rate": 7.065025762523952e-06, + "loss": 0.174, + "step": 11596 + }, + { + "epoch": 4.86, + "learning_rate": 7.063405061870925e-06, + "loss": 0.1655, + "step": 11597 + }, + { + "epoch": 4.86, + "learning_rate": 7.061784445627715e-06, + "loss": 0.1518, + "step": 11598 + }, + { + "epoch": 4.86, + "learning_rate": 7.060163913840903e-06, + "loss": 0.1504, + "step": 11599 + }, + { + "epoch": 4.86, + "learning_rate": 7.058543466557069e-06, + "loss": 0.1705, + "step": 11600 + }, + { + "epoch": 4.86, + "learning_rate": 7.056923103822795e-06, + "loss": 0.1463, + "step": 11601 + }, + { + "epoch": 4.86, + "learning_rate": 7.0553028256846555e-06, + "loss": 0.1493, + "step": 11602 + }, + { + "epoch": 4.86, + "learning_rate": 7.053682632189221e-06, + "loss": 0.1341, + "step": 11603 + }, + { + "epoch": 4.86, + "learning_rate": 7.052062523383065e-06, + "loss": 0.1391, + "step": 11604 + }, + { + "epoch": 4.86, + "learning_rate": 7.050442499312757e-06, + "loss": 0.1184, + "step": 11605 + }, + { + "epoch": 4.86, + "learning_rate": 7.048822560024862e-06, + "loss": 0.1176, + "step": 11606 + }, + { + "epoch": 4.86, + "learning_rate": 7.047202705565942e-06, + "loss": 0.165, + "step": 11607 + }, + { + "epoch": 4.86, + "learning_rate": 7.045582935982562e-06, + "loss": 0.1549, + "step": 11608 + }, + { + "epoch": 4.86, + "learning_rate": 7.0439632513212775e-06, + "loss": 0.153, + "step": 11609 + }, + { + "epoch": 4.86, + "learning_rate": 7.042343651628646e-06, + "loss": 0.1586, + "step": 11610 + }, + { + "epoch": 4.86, + "learning_rate": 7.040724136951224e-06, + "loss": 0.1513, + "step": 11611 + }, + { + "epoch": 4.86, + "learning_rate": 7.0391047073355575e-06, + "loss": 0.1322, + "step": 11612 + }, + { + "epoch": 4.86, + "learning_rate": 7.0374853628282e-06, + "loss": 0.1278, + "step": 11613 + }, + { + "epoch": 4.86, + "learning_rate": 7.035866103475698e-06, + "loss": 0.1393, + "step": 11614 + }, + { + "epoch": 4.86, + "learning_rate": 7.034246929324593e-06, + "loss": 0.1838, + "step": 11615 + }, + { + "epoch": 4.86, + "learning_rate": 7.032627840421429e-06, + "loss": 0.1764, + "step": 11616 + }, + { + "epoch": 4.86, + "learning_rate": 7.031008836812743e-06, + "loss": 0.1596, + "step": 11617 + }, + { + "epoch": 4.86, + "learning_rate": 7.029389918545078e-06, + "loss": 0.1471, + "step": 11618 + }, + { + "epoch": 4.86, + "learning_rate": 7.0277710856649614e-06, + "loss": 0.1407, + "step": 11619 + }, + { + "epoch": 4.86, + "learning_rate": 7.026152338218925e-06, + "loss": 0.1156, + "step": 11620 + }, + { + "epoch": 4.87, + "learning_rate": 7.024533676253502e-06, + "loss": 0.1556, + "step": 11621 + }, + { + "epoch": 4.87, + "learning_rate": 7.022915099815219e-06, + "loss": 0.1736, + "step": 11622 + }, + { + "epoch": 4.87, + "learning_rate": 7.021296608950599e-06, + "loss": 0.1521, + "step": 11623 + }, + { + "epoch": 4.87, + "learning_rate": 7.019678203706164e-06, + "loss": 0.1597, + "step": 11624 + }, + { + "epoch": 4.87, + "learning_rate": 7.018059884128437e-06, + "loss": 0.1699, + "step": 11625 + }, + { + "epoch": 4.87, + "learning_rate": 7.01644165026393e-06, + "loss": 0.1904, + "step": 11626 + }, + { + "epoch": 4.87, + "learning_rate": 7.014823502159159e-06, + "loss": 0.1579, + "step": 11627 + }, + { + "epoch": 4.87, + "learning_rate": 7.0132054398606405e-06, + "loss": 0.1672, + "step": 11628 + }, + { + "epoch": 4.87, + "learning_rate": 7.011587463414879e-06, + "loss": 0.1292, + "step": 11629 + }, + { + "epoch": 4.87, + "learning_rate": 7.009969572868385e-06, + "loss": 0.1356, + "step": 11630 + }, + { + "epoch": 4.87, + "learning_rate": 7.008351768267661e-06, + "loss": 0.1077, + "step": 11631 + }, + { + "epoch": 4.87, + "learning_rate": 7.006734049659213e-06, + "loss": 0.1405, + "step": 11632 + }, + { + "epoch": 4.87, + "learning_rate": 7.005116417089537e-06, + "loss": 0.1182, + "step": 11633 + }, + { + "epoch": 4.87, + "learning_rate": 7.0034988706051325e-06, + "loss": 0.169, + "step": 11634 + }, + { + "epoch": 4.87, + "learning_rate": 7.001881410252494e-06, + "loss": 0.1308, + "step": 11635 + }, + { + "epoch": 4.87, + "learning_rate": 7.000264036078113e-06, + "loss": 0.1397, + "step": 11636 + }, + { + "epoch": 4.87, + "learning_rate": 6.9986467481284805e-06, + "loss": 0.1342, + "step": 11637 + }, + { + "epoch": 4.87, + "learning_rate": 6.997029546450085e-06, + "loss": 0.112, + "step": 11638 + }, + { + "epoch": 4.87, + "learning_rate": 6.99541243108941e-06, + "loss": 0.1643, + "step": 11639 + }, + { + "epoch": 4.87, + "learning_rate": 6.9937954020929414e-06, + "loss": 0.1515, + "step": 11640 + }, + { + "epoch": 4.87, + "learning_rate": 6.992178459507153e-06, + "loss": 0.1187, + "step": 11641 + }, + { + "epoch": 4.87, + "learning_rate": 6.990561603378524e-06, + "loss": 0.144, + "step": 11642 + }, + { + "epoch": 4.87, + "learning_rate": 6.9889448337535345e-06, + "loss": 0.1462, + "step": 11643 + }, + { + "epoch": 4.88, + "learning_rate": 6.987328150678652e-06, + "loss": 0.1524, + "step": 11644 + }, + { + "epoch": 4.88, + "learning_rate": 6.985711554200347e-06, + "loss": 0.1162, + "step": 11645 + }, + { + "epoch": 4.88, + "learning_rate": 6.984095044365092e-06, + "loss": 0.154, + "step": 11646 + }, + { + "epoch": 4.88, + "learning_rate": 6.982478621219346e-06, + "loss": 0.1642, + "step": 11647 + }, + { + "epoch": 4.88, + "learning_rate": 6.980862284809575e-06, + "loss": 0.1614, + "step": 11648 + }, + { + "epoch": 4.88, + "learning_rate": 6.97924603518224e-06, + "loss": 0.1626, + "step": 11649 + }, + { + "epoch": 4.88, + "learning_rate": 6.9776298723837956e-06, + "loss": 0.1456, + "step": 11650 + }, + { + "epoch": 4.88, + "learning_rate": 6.976013796460698e-06, + "loss": 0.1292, + "step": 11651 + }, + { + "epoch": 4.88, + "learning_rate": 6.974397807459402e-06, + "loss": 0.1422, + "step": 11652 + }, + { + "epoch": 4.88, + "learning_rate": 6.972781905426354e-06, + "loss": 0.133, + "step": 11653 + }, + { + "epoch": 4.88, + "learning_rate": 6.9711660904080055e-06, + "loss": 0.154, + "step": 11654 + }, + { + "epoch": 4.88, + "learning_rate": 6.9695503624508005e-06, + "loss": 0.1427, + "step": 11655 + }, + { + "epoch": 4.88, + "learning_rate": 6.96793472160118e-06, + "loss": 0.1513, + "step": 11656 + }, + { + "epoch": 4.88, + "learning_rate": 6.966319167905584e-06, + "loss": 0.1249, + "step": 11657 + }, + { + "epoch": 4.88, + "learning_rate": 6.964703701410455e-06, + "loss": 0.1561, + "step": 11658 + }, + { + "epoch": 4.88, + "learning_rate": 6.963088322162221e-06, + "loss": 0.139, + "step": 11659 + }, + { + "epoch": 4.88, + "learning_rate": 6.96147303020732e-06, + "loss": 0.194, + "step": 11660 + }, + { + "epoch": 4.88, + "learning_rate": 6.95985782559218e-06, + "loss": 0.1512, + "step": 11661 + }, + { + "epoch": 4.88, + "learning_rate": 6.958242708363231e-06, + "loss": 0.2009, + "step": 11662 + }, + { + "epoch": 4.88, + "learning_rate": 6.956627678566898e-06, + "loss": 0.1388, + "step": 11663 + }, + { + "epoch": 4.88, + "learning_rate": 6.955012736249598e-06, + "loss": 0.1439, + "step": 11664 + }, + { + "epoch": 4.88, + "learning_rate": 6.953397881457754e-06, + "loss": 0.1572, + "step": 11665 + }, + { + "epoch": 4.88, + "learning_rate": 6.951783114237786e-06, + "loss": 0.1202, + "step": 11666 + }, + { + "epoch": 4.88, + "learning_rate": 6.950168434636106e-06, + "loss": 0.1251, + "step": 11667 + }, + { + "epoch": 4.89, + "learning_rate": 6.94855384269913e-06, + "loss": 0.1462, + "step": 11668 + }, + { + "epoch": 4.89, + "learning_rate": 6.946939338473267e-06, + "loss": 0.1682, + "step": 11669 + }, + { + "epoch": 4.89, + "learning_rate": 6.945324922004919e-06, + "loss": 0.1335, + "step": 11670 + }, + { + "epoch": 4.89, + "learning_rate": 6.943710593340498e-06, + "loss": 0.1708, + "step": 11671 + }, + { + "epoch": 4.89, + "learning_rate": 6.9420963525264025e-06, + "loss": 0.1784, + "step": 11672 + }, + { + "epoch": 4.89, + "learning_rate": 6.940482199609036e-06, + "loss": 0.1363, + "step": 11673 + }, + { + "epoch": 4.89, + "learning_rate": 6.938868134634793e-06, + "loss": 0.1453, + "step": 11674 + }, + { + "epoch": 4.89, + "learning_rate": 6.937254157650068e-06, + "loss": 0.1673, + "step": 11675 + }, + { + "epoch": 4.89, + "learning_rate": 6.935640268701256e-06, + "loss": 0.1203, + "step": 11676 + }, + { + "epoch": 4.89, + "learning_rate": 6.934026467834745e-06, + "loss": 0.1493, + "step": 11677 + }, + { + "epoch": 4.89, + "learning_rate": 6.93241275509692e-06, + "loss": 0.1621, + "step": 11678 + }, + { + "epoch": 4.89, + "learning_rate": 6.930799130534172e-06, + "loss": 0.1438, + "step": 11679 + }, + { + "epoch": 4.89, + "learning_rate": 6.929185594192877e-06, + "loss": 0.1704, + "step": 11680 + }, + { + "epoch": 4.89, + "learning_rate": 6.927572146119416e-06, + "loss": 0.1914, + "step": 11681 + }, + { + "epoch": 4.89, + "learning_rate": 6.925958786360172e-06, + "loss": 0.1893, + "step": 11682 + }, + { + "epoch": 4.89, + "learning_rate": 6.92434551496151e-06, + "loss": 0.115, + "step": 11683 + }, + { + "epoch": 4.89, + "learning_rate": 6.922732331969808e-06, + "loss": 0.1342, + "step": 11684 + }, + { + "epoch": 4.89, + "learning_rate": 6.921119237431436e-06, + "loss": 0.1299, + "step": 11685 + }, + { + "epoch": 4.89, + "learning_rate": 6.9195062313927605e-06, + "loss": 0.1314, + "step": 11686 + }, + { + "epoch": 4.89, + "learning_rate": 6.917893313900144e-06, + "loss": 0.1386, + "step": 11687 + }, + { + "epoch": 4.89, + "learning_rate": 6.916280484999946e-06, + "loss": 0.152, + "step": 11688 + }, + { + "epoch": 4.89, + "learning_rate": 6.914667744738531e-06, + "loss": 0.1339, + "step": 11689 + }, + { + "epoch": 4.89, + "learning_rate": 6.9130550931622555e-06, + "loss": 0.1567, + "step": 11690 + }, + { + "epoch": 4.89, + "learning_rate": 6.911442530317469e-06, + "loss": 0.1512, + "step": 11691 + }, + { + "epoch": 4.9, + "learning_rate": 6.909830056250527e-06, + "loss": 0.1504, + "step": 11692 + }, + { + "epoch": 4.9, + "learning_rate": 6.90821767100778e-06, + "loss": 0.192, + "step": 11693 + }, + { + "epoch": 4.9, + "learning_rate": 6.90660537463557e-06, + "loss": 0.1369, + "step": 11694 + }, + { + "epoch": 4.9, + "learning_rate": 6.904993167180243e-06, + "loss": 0.1093, + "step": 11695 + }, + { + "epoch": 4.9, + "learning_rate": 6.9033810486881445e-06, + "loss": 0.159, + "step": 11696 + }, + { + "epoch": 4.9, + "learning_rate": 6.901769019205606e-06, + "loss": 0.1447, + "step": 11697 + }, + { + "epoch": 4.9, + "learning_rate": 6.900157078778969e-06, + "loss": 0.1699, + "step": 11698 + }, + { + "epoch": 4.9, + "learning_rate": 6.898545227454567e-06, + "loss": 0.1377, + "step": 11699 + }, + { + "epoch": 4.9, + "learning_rate": 6.89693346527873e-06, + "loss": 0.1256, + "step": 11700 + }, + { + "epoch": 4.9, + "learning_rate": 6.895321792297785e-06, + "loss": 0.1856, + "step": 11701 + }, + { + "epoch": 4.9, + "learning_rate": 6.893710208558061e-06, + "loss": 0.1304, + "step": 11702 + }, + { + "epoch": 4.9, + "learning_rate": 6.892098714105882e-06, + "loss": 0.1064, + "step": 11703 + }, + { + "epoch": 4.9, + "learning_rate": 6.890487308987567e-06, + "loss": 0.1524, + "step": 11704 + }, + { + "epoch": 4.9, + "learning_rate": 6.888875993249433e-06, + "loss": 0.1516, + "step": 11705 + }, + { + "epoch": 4.9, + "learning_rate": 6.887264766937801e-06, + "loss": 0.1015, + "step": 11706 + }, + { + "epoch": 4.9, + "learning_rate": 6.885653630098978e-06, + "loss": 0.16, + "step": 11707 + }, + { + "epoch": 4.9, + "learning_rate": 6.884042582779285e-06, + "loss": 0.1747, + "step": 11708 + }, + { + "epoch": 4.9, + "learning_rate": 6.882431625025016e-06, + "loss": 0.1781, + "step": 11709 + }, + { + "epoch": 4.9, + "learning_rate": 6.880820756882486e-06, + "loss": 0.1813, + "step": 11710 + }, + { + "epoch": 4.9, + "learning_rate": 6.879209978397993e-06, + "loss": 0.1389, + "step": 11711 + }, + { + "epoch": 4.9, + "learning_rate": 6.87759928961784e-06, + "loss": 0.1352, + "step": 11712 + }, + { + "epoch": 4.9, + "learning_rate": 6.875988690588325e-06, + "loss": 0.2103, + "step": 11713 + }, + { + "epoch": 4.9, + "learning_rate": 6.8743781813557455e-06, + "loss": 0.1271, + "step": 11714 + }, + { + "epoch": 4.9, + "learning_rate": 6.872767761966388e-06, + "loss": 0.1428, + "step": 11715 + }, + { + "epoch": 4.91, + "learning_rate": 6.8711574324665476e-06, + "loss": 0.1631, + "step": 11716 + }, + { + "epoch": 4.91, + "learning_rate": 6.869547192902511e-06, + "loss": 0.1268, + "step": 11717 + }, + { + "epoch": 4.91, + "learning_rate": 6.867937043320562e-06, + "loss": 0.1482, + "step": 11718 + }, + { + "epoch": 4.91, + "learning_rate": 6.866326983766982e-06, + "loss": 0.1294, + "step": 11719 + }, + { + "epoch": 4.91, + "learning_rate": 6.8647170142880545e-06, + "loss": 0.1384, + "step": 11720 + }, + { + "epoch": 4.91, + "learning_rate": 6.863107134930052e-06, + "loss": 0.13, + "step": 11721 + }, + { + "epoch": 4.91, + "learning_rate": 6.861497345739252e-06, + "loss": 0.111, + "step": 11722 + }, + { + "epoch": 4.91, + "learning_rate": 6.859887646761927e-06, + "loss": 0.152, + "step": 11723 + }, + { + "epoch": 4.91, + "learning_rate": 6.858278038044344e-06, + "loss": 0.1708, + "step": 11724 + }, + { + "epoch": 4.91, + "learning_rate": 6.85666851963277e-06, + "loss": 0.1705, + "step": 11725 + }, + { + "epoch": 4.91, + "learning_rate": 6.855059091573472e-06, + "loss": 0.1319, + "step": 11726 + }, + { + "epoch": 4.91, + "learning_rate": 6.853449753912708e-06, + "loss": 0.1653, + "step": 11727 + }, + { + "epoch": 4.91, + "learning_rate": 6.851840506696738e-06, + "loss": 0.1324, + "step": 11728 + }, + { + "epoch": 4.91, + "learning_rate": 6.8502313499718206e-06, + "loss": 0.1431, + "step": 11729 + }, + { + "epoch": 4.91, + "learning_rate": 6.848622283784209e-06, + "loss": 0.1088, + "step": 11730 + }, + { + "epoch": 4.91, + "learning_rate": 6.847013308180155e-06, + "loss": 0.1375, + "step": 11731 + }, + { + "epoch": 4.91, + "learning_rate": 6.845404423205903e-06, + "loss": 0.1367, + "step": 11732 + }, + { + "epoch": 4.91, + "learning_rate": 6.843795628907699e-06, + "loss": 0.1468, + "step": 11733 + }, + { + "epoch": 4.91, + "learning_rate": 6.842186925331793e-06, + "loss": 0.1308, + "step": 11734 + }, + { + "epoch": 4.91, + "learning_rate": 6.8405783125244196e-06, + "loss": 0.1238, + "step": 11735 + }, + { + "epoch": 4.91, + "learning_rate": 6.838969790531817e-06, + "loss": 0.1603, + "step": 11736 + }, + { + "epoch": 4.91, + "learning_rate": 6.837361359400227e-06, + "loss": 0.1417, + "step": 11737 + }, + { + "epoch": 4.91, + "learning_rate": 6.835753019175875e-06, + "loss": 0.1458, + "step": 11738 + }, + { + "epoch": 4.91, + "learning_rate": 6.834144769904996e-06, + "loss": 0.16, + "step": 11739 + }, + { + "epoch": 4.92, + "learning_rate": 6.832536611633816e-06, + "loss": 0.1393, + "step": 11740 + }, + { + "epoch": 4.92, + "learning_rate": 6.830928544408559e-06, + "loss": 0.1548, + "step": 11741 + }, + { + "epoch": 4.92, + "learning_rate": 6.82932056827545e-06, + "loss": 0.1411, + "step": 11742 + }, + { + "epoch": 4.92, + "learning_rate": 6.827712683280707e-06, + "loss": 0.1808, + "step": 11743 + }, + { + "epoch": 4.92, + "learning_rate": 6.8261048894705505e-06, + "loss": 0.181, + "step": 11744 + }, + { + "epoch": 4.92, + "learning_rate": 6.824497186891189e-06, + "loss": 0.1412, + "step": 11745 + }, + { + "epoch": 4.92, + "learning_rate": 6.822889575588838e-06, + "loss": 0.1402, + "step": 11746 + }, + { + "epoch": 4.92, + "learning_rate": 6.8212820556097105e-06, + "loss": 0.1246, + "step": 11747 + }, + { + "epoch": 4.92, + "learning_rate": 6.819674627000006e-06, + "loss": 0.1237, + "step": 11748 + }, + { + "epoch": 4.92, + "learning_rate": 6.818067289805934e-06, + "loss": 0.1781, + "step": 11749 + }, + { + "epoch": 4.92, + "learning_rate": 6.816460044073694e-06, + "loss": 0.144, + "step": 11750 + }, + { + "epoch": 4.92, + "learning_rate": 6.8148528898494845e-06, + "loss": 0.1312, + "step": 11751 + }, + { + "epoch": 4.92, + "learning_rate": 6.813245827179502e-06, + "loss": 0.1339, + "step": 11752 + }, + { + "epoch": 4.92, + "learning_rate": 6.811638856109941e-06, + "loss": 0.1753, + "step": 11753 + }, + { + "epoch": 4.92, + "learning_rate": 6.810031976686994e-06, + "loss": 0.1361, + "step": 11754 + }, + { + "epoch": 4.92, + "learning_rate": 6.808425188956847e-06, + "loss": 0.1599, + "step": 11755 + }, + { + "epoch": 4.92, + "learning_rate": 6.8068184929656815e-06, + "loss": 0.1234, + "step": 11756 + }, + { + "epoch": 4.92, + "learning_rate": 6.8052118887596865e-06, + "loss": 0.1611, + "step": 11757 + }, + { + "epoch": 4.92, + "learning_rate": 6.803605376385042e-06, + "loss": 0.1567, + "step": 11758 + }, + { + "epoch": 4.92, + "learning_rate": 6.801998955887923e-06, + "loss": 0.1952, + "step": 11759 + }, + { + "epoch": 4.92, + "learning_rate": 6.800392627314506e-06, + "loss": 0.1309, + "step": 11760 + }, + { + "epoch": 4.92, + "learning_rate": 6.7987863907109656e-06, + "loss": 0.1267, + "step": 11761 + }, + { + "epoch": 4.92, + "learning_rate": 6.797180246123467e-06, + "loss": 0.2057, + "step": 11762 + }, + { + "epoch": 4.92, + "learning_rate": 6.795574193598182e-06, + "loss": 0.182, + "step": 11763 + }, + { + "epoch": 4.93, + "learning_rate": 6.793968233181274e-06, + "loss": 0.1429, + "step": 11764 + }, + { + "epoch": 4.93, + "learning_rate": 6.792362364918902e-06, + "loss": 0.1341, + "step": 11765 + }, + { + "epoch": 4.93, + "learning_rate": 6.790756588857228e-06, + "loss": 0.1332, + "step": 11766 + }, + { + "epoch": 4.93, + "learning_rate": 6.7891509050424095e-06, + "loss": 0.146, + "step": 11767 + }, + { + "epoch": 4.93, + "learning_rate": 6.787545313520597e-06, + "loss": 0.1507, + "step": 11768 + }, + { + "epoch": 4.93, + "learning_rate": 6.785939814337944e-06, + "loss": 0.1471, + "step": 11769 + }, + { + "epoch": 4.93, + "learning_rate": 6.7843344075406005e-06, + "loss": 0.1545, + "step": 11770 + }, + { + "epoch": 4.93, + "learning_rate": 6.782729093174709e-06, + "loss": 0.1753, + "step": 11771 + }, + { + "epoch": 4.93, + "learning_rate": 6.781123871286416e-06, + "loss": 0.1701, + "step": 11772 + }, + { + "epoch": 4.93, + "learning_rate": 6.77951874192186e-06, + "loss": 0.1585, + "step": 11773 + }, + { + "epoch": 4.93, + "learning_rate": 6.777913705127181e-06, + "loss": 0.1698, + "step": 11774 + }, + { + "epoch": 4.93, + "learning_rate": 6.776308760948514e-06, + "loss": 0.1194, + "step": 11775 + }, + { + "epoch": 4.93, + "learning_rate": 6.774703909431992e-06, + "loss": 0.1379, + "step": 11776 + }, + { + "epoch": 4.93, + "learning_rate": 6.773099150623742e-06, + "loss": 0.1164, + "step": 11777 + }, + { + "epoch": 4.93, + "learning_rate": 6.771494484569895e-06, + "loss": 0.1853, + "step": 11778 + }, + { + "epoch": 4.93, + "learning_rate": 6.7698899113165715e-06, + "loss": 0.1434, + "step": 11779 + }, + { + "epoch": 4.93, + "learning_rate": 6.768285430909896e-06, + "loss": 0.1482, + "step": 11780 + }, + { + "epoch": 4.93, + "learning_rate": 6.76668104339599e-06, + "loss": 0.1466, + "step": 11781 + }, + { + "epoch": 4.93, + "learning_rate": 6.765076748820968e-06, + "loss": 0.1943, + "step": 11782 + }, + { + "epoch": 4.93, + "learning_rate": 6.763472547230943e-06, + "loss": 0.1793, + "step": 11783 + }, + { + "epoch": 4.93, + "learning_rate": 6.761868438672028e-06, + "loss": 0.1735, + "step": 11784 + }, + { + "epoch": 4.93, + "learning_rate": 6.760264423190332e-06, + "loss": 0.1779, + "step": 11785 + }, + { + "epoch": 4.93, + "learning_rate": 6.75866050083196e-06, + "loss": 0.1505, + "step": 11786 + }, + { + "epoch": 4.93, + "learning_rate": 6.757056671643014e-06, + "loss": 0.1513, + "step": 11787 + }, + { + "epoch": 4.94, + "learning_rate": 6.7554529356696e-06, + "loss": 0.1411, + "step": 11788 + }, + { + "epoch": 4.94, + "learning_rate": 6.753849292957808e-06, + "loss": 0.1627, + "step": 11789 + }, + { + "epoch": 4.94, + "learning_rate": 6.752245743553738e-06, + "loss": 0.1364, + "step": 11790 + }, + { + "epoch": 4.94, + "learning_rate": 6.750642287503486e-06, + "loss": 0.1319, + "step": 11791 + }, + { + "epoch": 4.94, + "learning_rate": 6.749038924853134e-06, + "loss": 0.1318, + "step": 11792 + }, + { + "epoch": 4.94, + "learning_rate": 6.747435655648774e-06, + "loss": 0.1795, + "step": 11793 + }, + { + "epoch": 4.94, + "learning_rate": 6.745832479936492e-06, + "loss": 0.1398, + "step": 11794 + }, + { + "epoch": 4.94, + "learning_rate": 6.744229397762365e-06, + "loss": 0.1839, + "step": 11795 + }, + { + "epoch": 4.94, + "learning_rate": 6.742626409172475e-06, + "loss": 0.1448, + "step": 11796 + }, + { + "epoch": 4.94, + "learning_rate": 6.741023514212901e-06, + "loss": 0.1557, + "step": 11797 + }, + { + "epoch": 4.94, + "learning_rate": 6.739420712929712e-06, + "loss": 0.139, + "step": 11798 + }, + { + "epoch": 4.94, + "learning_rate": 6.7378180053689854e-06, + "loss": 0.1092, + "step": 11799 + }, + { + "epoch": 4.94, + "learning_rate": 6.736215391576782e-06, + "loss": 0.1896, + "step": 11800 + }, + { + "epoch": 4.94, + "learning_rate": 6.734612871599169e-06, + "loss": 0.1533, + "step": 11801 + }, + { + "epoch": 4.94, + "learning_rate": 6.733010445482214e-06, + "loss": 0.1236, + "step": 11802 + }, + { + "epoch": 4.94, + "learning_rate": 6.7314081132719735e-06, + "loss": 0.1565, + "step": 11803 + }, + { + "epoch": 4.94, + "learning_rate": 6.729805875014506e-06, + "loss": 0.1567, + "step": 11804 + }, + { + "epoch": 4.94, + "learning_rate": 6.728203730755868e-06, + "loss": 0.1762, + "step": 11805 + }, + { + "epoch": 4.94, + "learning_rate": 6.726601680542109e-06, + "loss": 0.1186, + "step": 11806 + }, + { + "epoch": 4.94, + "learning_rate": 6.724999724419279e-06, + "loss": 0.1542, + "step": 11807 + }, + { + "epoch": 4.94, + "learning_rate": 6.723397862433429e-06, + "loss": 0.1656, + "step": 11808 + }, + { + "epoch": 4.94, + "learning_rate": 6.721796094630597e-06, + "loss": 0.1587, + "step": 11809 + }, + { + "epoch": 4.94, + "learning_rate": 6.720194421056828e-06, + "loss": 0.1193, + "step": 11810 + }, + { + "epoch": 4.94, + "learning_rate": 6.7185928417581615e-06, + "loss": 0.1541, + "step": 11811 + }, + { + "epoch": 4.95, + "learning_rate": 6.71699135678063e-06, + "loss": 0.1437, + "step": 11812 + }, + { + "epoch": 4.95, + "learning_rate": 6.715389966170269e-06, + "loss": 0.1602, + "step": 11813 + }, + { + "epoch": 4.95, + "learning_rate": 6.713788669973108e-06, + "loss": 0.1463, + "step": 11814 + }, + { + "epoch": 4.95, + "learning_rate": 6.7121874682351784e-06, + "loss": 0.1272, + "step": 11815 + }, + { + "epoch": 4.95, + "learning_rate": 6.710586361002499e-06, + "loss": 0.1211, + "step": 11816 + }, + { + "epoch": 4.95, + "learning_rate": 6.708985348321098e-06, + "loss": 0.1686, + "step": 11817 + }, + { + "epoch": 4.95, + "learning_rate": 6.707384430236991e-06, + "loss": 0.1725, + "step": 11818 + }, + { + "epoch": 4.95, + "learning_rate": 6.705783606796199e-06, + "loss": 0.1516, + "step": 11819 + }, + { + "epoch": 4.95, + "learning_rate": 6.704182878044732e-06, + "loss": 0.1333, + "step": 11820 + }, + { + "epoch": 4.95, + "learning_rate": 6.702582244028605e-06, + "loss": 0.1902, + "step": 11821 + }, + { + "epoch": 4.95, + "learning_rate": 6.700981704793828e-06, + "loss": 0.1359, + "step": 11822 + }, + { + "epoch": 4.95, + "learning_rate": 6.6993812603864e-06, + "loss": 0.1323, + "step": 11823 + }, + { + "epoch": 4.95, + "learning_rate": 6.6977809108523295e-06, + "loss": 0.1735, + "step": 11824 + }, + { + "epoch": 4.95, + "learning_rate": 6.696180656237616e-06, + "loss": 0.1605, + "step": 11825 + }, + { + "epoch": 4.95, + "learning_rate": 6.69458049658826e-06, + "loss": 0.1059, + "step": 11826 + }, + { + "epoch": 4.95, + "learning_rate": 6.6929804319502515e-06, + "loss": 0.1623, + "step": 11827 + }, + { + "epoch": 4.95, + "learning_rate": 6.691380462369588e-06, + "loss": 0.1609, + "step": 11828 + }, + { + "epoch": 4.95, + "learning_rate": 6.689780587892256e-06, + "loss": 0.1591, + "step": 11829 + }, + { + "epoch": 4.95, + "learning_rate": 6.688180808564244e-06, + "loss": 0.1501, + "step": 11830 + }, + { + "epoch": 4.95, + "learning_rate": 6.686581124431535e-06, + "loss": 0.1616, + "step": 11831 + }, + { + "epoch": 4.95, + "learning_rate": 6.684981535540114e-06, + "loss": 0.1992, + "step": 11832 + }, + { + "epoch": 4.95, + "learning_rate": 6.6833820419359526e-06, + "loss": 0.1278, + "step": 11833 + }, + { + "epoch": 4.95, + "learning_rate": 6.6817826436650335e-06, + "loss": 0.1619, + "step": 11834 + }, + { + "epoch": 4.95, + "learning_rate": 6.680183340773329e-06, + "loss": 0.1707, + "step": 11835 + }, + { + "epoch": 4.96, + "learning_rate": 6.678584133306808e-06, + "loss": 0.1649, + "step": 11836 + }, + { + "epoch": 4.96, + "learning_rate": 6.6769850213114375e-06, + "loss": 0.1744, + "step": 11837 + }, + { + "epoch": 4.96, + "learning_rate": 6.675386004833185e-06, + "loss": 0.1598, + "step": 11838 + }, + { + "epoch": 4.96, + "learning_rate": 6.673787083918011e-06, + "loss": 0.1677, + "step": 11839 + }, + { + "epoch": 4.96, + "learning_rate": 6.672188258611874e-06, + "loss": 0.1296, + "step": 11840 + }, + { + "epoch": 4.96, + "learning_rate": 6.670589528960734e-06, + "loss": 0.1734, + "step": 11841 + }, + { + "epoch": 4.96, + "learning_rate": 6.668990895010545e-06, + "loss": 0.1718, + "step": 11842 + }, + { + "epoch": 4.96, + "learning_rate": 6.667392356807255e-06, + "loss": 0.1761, + "step": 11843 + }, + { + "epoch": 4.96, + "learning_rate": 6.665793914396818e-06, + "loss": 0.1362, + "step": 11844 + }, + { + "epoch": 4.96, + "learning_rate": 6.664195567825174e-06, + "loss": 0.1253, + "step": 11845 + }, + { + "epoch": 4.96, + "learning_rate": 6.662597317138268e-06, + "loss": 0.1808, + "step": 11846 + }, + { + "epoch": 4.96, + "learning_rate": 6.660999162382038e-06, + "loss": 0.1541, + "step": 11847 + }, + { + "epoch": 4.96, + "learning_rate": 6.6594011036024255e-06, + "loss": 0.1515, + "step": 11848 + }, + { + "epoch": 4.96, + "learning_rate": 6.657803140845367e-06, + "loss": 0.1437, + "step": 11849 + }, + { + "epoch": 4.96, + "learning_rate": 6.656205274156788e-06, + "loss": 0.1805, + "step": 11850 + }, + { + "epoch": 4.96, + "learning_rate": 6.654607503582621e-06, + "loss": 0.1519, + "step": 11851 + }, + { + "epoch": 4.96, + "learning_rate": 6.653009829168795e-06, + "loss": 0.1387, + "step": 11852 + }, + { + "epoch": 4.96, + "learning_rate": 6.651412250961229e-06, + "loss": 0.1141, + "step": 11853 + }, + { + "epoch": 4.96, + "learning_rate": 6.649814769005847e-06, + "loss": 0.1398, + "step": 11854 + }, + { + "epoch": 4.96, + "learning_rate": 6.648217383348565e-06, + "loss": 0.1622, + "step": 11855 + }, + { + "epoch": 4.96, + "learning_rate": 6.646620094035305e-06, + "loss": 0.1627, + "step": 11856 + }, + { + "epoch": 4.96, + "learning_rate": 6.645022901111971e-06, + "loss": 0.1522, + "step": 11857 + }, + { + "epoch": 4.96, + "learning_rate": 6.643425804624476e-06, + "loss": 0.1588, + "step": 11858 + }, + { + "epoch": 4.97, + "learning_rate": 6.641828804618731e-06, + "loss": 0.1015, + "step": 11859 + }, + { + "epoch": 4.97, + "learning_rate": 6.640231901140633e-06, + "loss": 0.1449, + "step": 11860 + }, + { + "epoch": 4.97, + "learning_rate": 6.638635094236089e-06, + "loss": 0.1349, + "step": 11861 + }, + { + "epoch": 4.97, + "learning_rate": 6.637038383950998e-06, + "loss": 0.1586, + "step": 11862 + }, + { + "epoch": 4.97, + "learning_rate": 6.635441770331252e-06, + "loss": 0.1046, + "step": 11863 + }, + { + "epoch": 4.97, + "learning_rate": 6.633845253422748e-06, + "loss": 0.1767, + "step": 11864 + }, + { + "epoch": 4.97, + "learning_rate": 6.632248833271376e-06, + "loss": 0.1598, + "step": 11865 + }, + { + "epoch": 4.97, + "learning_rate": 6.6306525099230214e-06, + "loss": 0.1807, + "step": 11866 + }, + { + "epoch": 4.97, + "learning_rate": 6.629056283423574e-06, + "loss": 0.1622, + "step": 11867 + }, + { + "epoch": 4.97, + "learning_rate": 6.62746015381891e-06, + "loss": 0.2002, + "step": 11868 + }, + { + "epoch": 4.97, + "learning_rate": 6.625864121154908e-06, + "loss": 0.1498, + "step": 11869 + }, + { + "epoch": 4.97, + "learning_rate": 6.624268185477452e-06, + "loss": 0.1557, + "step": 11870 + }, + { + "epoch": 4.97, + "learning_rate": 6.622672346832409e-06, + "loss": 0.1275, + "step": 11871 + }, + { + "epoch": 4.97, + "learning_rate": 6.621076605265653e-06, + "loss": 0.1179, + "step": 11872 + }, + { + "epoch": 4.97, + "learning_rate": 6.619480960823053e-06, + "loss": 0.1549, + "step": 11873 + }, + { + "epoch": 4.97, + "learning_rate": 6.617885413550471e-06, + "loss": 0.1529, + "step": 11874 + }, + { + "epoch": 4.97, + "learning_rate": 6.616289963493771e-06, + "loss": 0.1529, + "step": 11875 + }, + { + "epoch": 4.97, + "learning_rate": 6.614694610698815e-06, + "loss": 0.1736, + "step": 11876 + }, + { + "epoch": 4.97, + "learning_rate": 6.613099355211457e-06, + "loss": 0.186, + "step": 11877 + }, + { + "epoch": 4.97, + "learning_rate": 6.611504197077554e-06, + "loss": 0.1384, + "step": 11878 + }, + { + "epoch": 4.97, + "learning_rate": 6.609909136342956e-06, + "loss": 0.1638, + "step": 11879 + }, + { + "epoch": 4.97, + "learning_rate": 6.608314173053509e-06, + "loss": 0.1224, + "step": 11880 + }, + { + "epoch": 4.97, + "learning_rate": 6.606719307255064e-06, + "loss": 0.1773, + "step": 11881 + }, + { + "epoch": 4.97, + "learning_rate": 6.605124538993462e-06, + "loss": 0.1682, + "step": 11882 + }, + { + "epoch": 4.98, + "learning_rate": 6.60352986831454e-06, + "loss": 0.1524, + "step": 11883 + }, + { + "epoch": 4.98, + "learning_rate": 6.601935295264139e-06, + "loss": 0.138, + "step": 11884 + }, + { + "epoch": 4.98, + "learning_rate": 6.600340819888091e-06, + "loss": 0.1323, + "step": 11885 + }, + { + "epoch": 4.98, + "learning_rate": 6.598746442232233e-06, + "loss": 0.1436, + "step": 11886 + }, + { + "epoch": 4.98, + "learning_rate": 6.597152162342388e-06, + "loss": 0.1648, + "step": 11887 + }, + { + "epoch": 4.98, + "learning_rate": 6.595557980264384e-06, + "loss": 0.1288, + "step": 11888 + }, + { + "epoch": 4.98, + "learning_rate": 6.593963896044047e-06, + "loss": 0.1178, + "step": 11889 + }, + { + "epoch": 4.98, + "learning_rate": 6.592369909727196e-06, + "loss": 0.1467, + "step": 11890 + }, + { + "epoch": 4.98, + "learning_rate": 6.590776021359646e-06, + "loss": 0.1334, + "step": 11891 + }, + { + "epoch": 4.98, + "learning_rate": 6.589182230987213e-06, + "loss": 0.1008, + "step": 11892 + }, + { + "epoch": 4.98, + "learning_rate": 6.587588538655714e-06, + "loss": 0.148, + "step": 11893 + }, + { + "epoch": 4.98, + "learning_rate": 6.5859949444109495e-06, + "loss": 0.1416, + "step": 11894 + }, + { + "epoch": 4.98, + "learning_rate": 6.584401448298732e-06, + "loss": 0.0951, + "step": 11895 + }, + { + "epoch": 4.98, + "learning_rate": 6.582808050364864e-06, + "loss": 0.137, + "step": 11896 + }, + { + "epoch": 4.98, + "learning_rate": 6.581214750655148e-06, + "loss": 0.1516, + "step": 11897 + }, + { + "epoch": 4.98, + "learning_rate": 6.579621549215378e-06, + "loss": 0.1675, + "step": 11898 + }, + { + "epoch": 4.98, + "learning_rate": 6.578028446091351e-06, + "loss": 0.0996, + "step": 11899 + }, + { + "epoch": 4.98, + "learning_rate": 6.5764354413288635e-06, + "loss": 0.1206, + "step": 11900 + }, + { + "epoch": 4.98, + "learning_rate": 6.574842534973698e-06, + "loss": 0.1579, + "step": 11901 + }, + { + "epoch": 4.98, + "learning_rate": 6.573249727071644e-06, + "loss": 0.1271, + "step": 11902 + }, + { + "epoch": 4.98, + "learning_rate": 6.571657017668489e-06, + "loss": 0.1465, + "step": 11903 + }, + { + "epoch": 4.98, + "learning_rate": 6.570064406810008e-06, + "loss": 0.1497, + "step": 11904 + }, + { + "epoch": 4.98, + "learning_rate": 6.568471894541982e-06, + "loss": 0.1262, + "step": 11905 + }, + { + "epoch": 4.98, + "learning_rate": 6.5668794809101884e-06, + "loss": 0.1509, + "step": 11906 + }, + { + "epoch": 4.99, + "learning_rate": 6.565287165960396e-06, + "loss": 0.1165, + "step": 11907 + }, + { + "epoch": 4.99, + "learning_rate": 6.563694949738375e-06, + "loss": 0.1142, + "step": 11908 + }, + { + "epoch": 4.99, + "learning_rate": 6.562102832289896e-06, + "loss": 0.1539, + "step": 11909 + }, + { + "epoch": 4.99, + "learning_rate": 6.560510813660719e-06, + "loss": 0.1892, + "step": 11910 + }, + { + "epoch": 4.99, + "learning_rate": 6.558918893896606e-06, + "loss": 0.1619, + "step": 11911 + }, + { + "epoch": 4.99, + "learning_rate": 6.55732707304332e-06, + "loss": 0.1679, + "step": 11912 + }, + { + "epoch": 4.99, + "learning_rate": 6.5557353511466085e-06, + "loss": 0.1434, + "step": 11913 + }, + { + "epoch": 4.99, + "learning_rate": 6.554143728252229e-06, + "loss": 0.1606, + "step": 11914 + }, + { + "epoch": 4.99, + "learning_rate": 6.552552204405928e-06, + "loss": 0.126, + "step": 11915 + }, + { + "epoch": 4.99, + "learning_rate": 6.550960779653456e-06, + "loss": 0.1764, + "step": 11916 + }, + { + "epoch": 4.99, + "learning_rate": 6.5493694540405575e-06, + "loss": 0.158, + "step": 11917 + }, + { + "epoch": 4.99, + "learning_rate": 6.547778227612969e-06, + "loss": 0.1304, + "step": 11918 + }, + { + "epoch": 4.99, + "learning_rate": 6.546187100416432e-06, + "loss": 0.1372, + "step": 11919 + }, + { + "epoch": 4.99, + "learning_rate": 6.544596072496685e-06, + "loss": 0.1006, + "step": 11920 + }, + { + "epoch": 4.99, + "learning_rate": 6.5430051438994545e-06, + "loss": 0.1562, + "step": 11921 + }, + { + "epoch": 4.99, + "learning_rate": 6.541414314670475e-06, + "loss": 0.1908, + "step": 11922 + }, + { + "epoch": 4.99, + "learning_rate": 6.539823584855472e-06, + "loss": 0.179, + "step": 11923 + }, + { + "epoch": 4.99, + "learning_rate": 6.53823295450017e-06, + "loss": 0.1202, + "step": 11924 + }, + { + "epoch": 4.99, + "learning_rate": 6.536642423650288e-06, + "loss": 0.1464, + "step": 11925 + }, + { + "epoch": 4.99, + "learning_rate": 6.5350519923515464e-06, + "loss": 0.1533, + "step": 11926 + }, + { + "epoch": 4.99, + "learning_rate": 6.533461660649663e-06, + "loss": 0.1619, + "step": 11927 + }, + { + "epoch": 4.99, + "learning_rate": 6.531871428590345e-06, + "loss": 0.1269, + "step": 11928 + }, + { + "epoch": 4.99, + "learning_rate": 6.530281296219305e-06, + "loss": 0.1226, + "step": 11929 + }, + { + "epoch": 4.99, + "learning_rate": 6.528691263582254e-06, + "loss": 0.1563, + "step": 11930 + }, + { + "epoch": 5.0, + "learning_rate": 6.52710133072489e-06, + "loss": 0.1348, + "step": 11931 + }, + { + "epoch": 5.0, + "learning_rate": 6.5255114976929154e-06, + "loss": 0.1795, + "step": 11932 + }, + { + "epoch": 5.0, + "learning_rate": 6.523921764532032e-06, + "loss": 0.1407, + "step": 11933 + }, + { + "epoch": 5.0, + "learning_rate": 6.522332131287931e-06, + "loss": 0.1487, + "step": 11934 + }, + { + "epoch": 5.0, + "learning_rate": 6.520742598006311e-06, + "loss": 0.1426, + "step": 11935 + }, + { + "epoch": 5.0, + "learning_rate": 6.519153164732852e-06, + "loss": 0.1117, + "step": 11936 + }, + { + "epoch": 5.0, + "learning_rate": 6.5175638315132484e-06, + "loss": 0.1438, + "step": 11937 + }, + { + "epoch": 5.0, + "learning_rate": 6.5159745983931846e-06, + "loss": 0.1838, + "step": 11938 + }, + { + "epoch": 5.0, + "learning_rate": 6.514385465418336e-06, + "loss": 0.1861, + "step": 11939 + }, + { + "epoch": 5.0, + "learning_rate": 6.512796432634383e-06, + "loss": 0.2126, + "step": 11940 + }, + { + "epoch": 5.0, + "learning_rate": 6.5112075000870065e-06, + "loss": 0.137, + "step": 11941 + }, + { + "epoch": 5.0, + "learning_rate": 6.5096186678218705e-06, + "loss": 0.1794, + "step": 11942 + }, + { + "epoch": 5.0, + "learning_rate": 6.50802993588465e-06, + "loss": 0.1587, + "step": 11943 + }, + { + "epoch": 5.0, + "learning_rate": 6.506441304321012e-06, + "loss": 0.1663, + "step": 11944 + }, + { + "epoch": 5.0, + "learning_rate": 6.5048527731766155e-06, + "loss": 0.1213, + "step": 11945 + }, + { + "epoch": 5.0, + "learning_rate": 6.503264342497127e-06, + "loss": 0.1663, + "step": 11946 + }, + { + "epoch": 5.0, + "learning_rate": 6.501676012328202e-06, + "loss": 0.1341, + "step": 11947 + }, + { + "epoch": 5.0, + "learning_rate": 6.500087782715494e-06, + "loss": 0.1334, + "step": 11948 + }, + { + "epoch": 5.0, + "learning_rate": 6.498499653704656e-06, + "loss": 0.096, + "step": 11949 + }, + { + "epoch": 5.0, + "learning_rate": 6.496911625341342e-06, + "loss": 0.1041, + "step": 11950 + }, + { + "epoch": 5.0, + "learning_rate": 6.495323697671192e-06, + "loss": 0.1372, + "step": 11951 + }, + { + "epoch": 5.0, + "learning_rate": 6.493735870739852e-06, + "loss": 0.1214, + "step": 11952 + }, + { + "epoch": 5.0, + "learning_rate": 6.492148144592962e-06, + "loss": 0.1367, + "step": 11953 + }, + { + "epoch": 5.0, + "learning_rate": 6.4905605192761656e-06, + "loss": 0.1148, + "step": 11954 + }, + { + "epoch": 5.01, + "learning_rate": 6.488972994835089e-06, + "loss": 0.144, + "step": 11955 + }, + { + "epoch": 5.01, + "learning_rate": 6.4873855713153675e-06, + "loss": 0.1576, + "step": 11956 + }, + { + "epoch": 5.01, + "learning_rate": 6.485798248762634e-06, + "loss": 0.1223, + "step": 11957 + }, + { + "epoch": 5.01, + "learning_rate": 6.484211027222513e-06, + "loss": 0.1417, + "step": 11958 + }, + { + "epoch": 5.01, + "learning_rate": 6.4826239067406216e-06, + "loss": 0.1227, + "step": 11959 + }, + { + "epoch": 5.01, + "learning_rate": 6.481036887362586e-06, + "loss": 0.1256, + "step": 11960 + }, + { + "epoch": 5.01, + "learning_rate": 6.479449969134024e-06, + "loss": 0.1219, + "step": 11961 + }, + { + "epoch": 5.01, + "learning_rate": 6.477863152100545e-06, + "loss": 0.128, + "step": 11962 + }, + { + "epoch": 5.01, + "learning_rate": 6.476276436307767e-06, + "loss": 0.1176, + "step": 11963 + }, + { + "epoch": 5.01, + "learning_rate": 6.474689821801295e-06, + "loss": 0.1425, + "step": 11964 + }, + { + "epoch": 5.01, + "learning_rate": 6.473103308626736e-06, + "loss": 0.0925, + "step": 11965 + }, + { + "epoch": 5.01, + "learning_rate": 6.471516896829692e-06, + "loss": 0.1364, + "step": 11966 + }, + { + "epoch": 5.01, + "learning_rate": 6.469930586455764e-06, + "loss": 0.1362, + "step": 11967 + }, + { + "epoch": 5.01, + "learning_rate": 6.468344377550549e-06, + "loss": 0.1267, + "step": 11968 + }, + { + "epoch": 5.01, + "learning_rate": 6.466758270159641e-06, + "loss": 0.156, + "step": 11969 + }, + { + "epoch": 5.01, + "learning_rate": 6.46517226432863e-06, + "loss": 0.1238, + "step": 11970 + }, + { + "epoch": 5.01, + "learning_rate": 6.4635863601031075e-06, + "loss": 0.1058, + "step": 11971 + }, + { + "epoch": 5.01, + "learning_rate": 6.462000557528656e-06, + "loss": 0.1324, + "step": 11972 + }, + { + "epoch": 5.01, + "learning_rate": 6.460414856650858e-06, + "loss": 0.0875, + "step": 11973 + }, + { + "epoch": 5.01, + "learning_rate": 6.4588292575152955e-06, + "loss": 0.1066, + "step": 11974 + }, + { + "epoch": 5.01, + "learning_rate": 6.4572437601675415e-06, + "loss": 0.1413, + "step": 11975 + }, + { + "epoch": 5.01, + "learning_rate": 6.455658364653174e-06, + "loss": 0.1057, + "step": 11976 + }, + { + "epoch": 5.01, + "learning_rate": 6.4540730710177614e-06, + "loss": 0.1148, + "step": 11977 + }, + { + "epoch": 5.01, + "learning_rate": 6.452487879306871e-06, + "loss": 0.1001, + "step": 11978 + }, + { + "epoch": 5.02, + "learning_rate": 6.450902789566068e-06, + "loss": 0.1373, + "step": 11979 + }, + { + "epoch": 5.02, + "learning_rate": 6.44931780184092e-06, + "loss": 0.107, + "step": 11980 + }, + { + "epoch": 5.02, + "learning_rate": 6.447732916176975e-06, + "loss": 0.1356, + "step": 11981 + }, + { + "epoch": 5.02, + "learning_rate": 6.446148132619798e-06, + "loss": 0.1585, + "step": 11982 + }, + { + "epoch": 5.02, + "learning_rate": 6.444563451214938e-06, + "loss": 0.1395, + "step": 11983 + }, + { + "epoch": 5.02, + "learning_rate": 6.442978872007946e-06, + "loss": 0.1106, + "step": 11984 + }, + { + "epoch": 5.02, + "learning_rate": 6.441394395044372e-06, + "loss": 0.1381, + "step": 11985 + }, + { + "epoch": 5.02, + "learning_rate": 6.439810020369755e-06, + "loss": 0.0928, + "step": 11986 + }, + { + "epoch": 5.02, + "learning_rate": 6.4382257480296405e-06, + "loss": 0.1299, + "step": 11987 + }, + { + "epoch": 5.02, + "learning_rate": 6.436641578069569e-06, + "loss": 0.1425, + "step": 11988 + }, + { + "epoch": 5.02, + "learning_rate": 6.435057510535071e-06, + "loss": 0.1285, + "step": 11989 + }, + { + "epoch": 5.02, + "learning_rate": 6.433473545471681e-06, + "loss": 0.1113, + "step": 11990 + }, + { + "epoch": 5.02, + "learning_rate": 6.431889682924932e-06, + "loss": 0.1415, + "step": 11991 + }, + { + "epoch": 5.02, + "learning_rate": 6.430305922940346e-06, + "loss": 0.1447, + "step": 11992 + }, + { + "epoch": 5.02, + "learning_rate": 6.428722265563446e-06, + "loss": 0.1038, + "step": 11993 + }, + { + "epoch": 5.02, + "learning_rate": 6.42713871083976e-06, + "loss": 0.1033, + "step": 11994 + }, + { + "epoch": 5.02, + "learning_rate": 6.4255552588147975e-06, + "loss": 0.1294, + "step": 11995 + }, + { + "epoch": 5.02, + "learning_rate": 6.423971909534077e-06, + "loss": 0.0996, + "step": 11996 + }, + { + "epoch": 5.02, + "learning_rate": 6.422388663043112e-06, + "loss": 0.1159, + "step": 11997 + }, + { + "epoch": 5.02, + "learning_rate": 6.420805519387412e-06, + "loss": 0.1138, + "step": 11998 + }, + { + "epoch": 5.02, + "learning_rate": 6.419222478612479e-06, + "loss": 0.0632, + "step": 11999 + }, + { + "epoch": 5.02, + "learning_rate": 6.417639540763818e-06, + "loss": 0.1196, + "step": 12000 + }, + { + "epoch": 5.02, + "learning_rate": 6.416056705886932e-06, + "loss": 0.1107, + "step": 12001 + }, + { + "epoch": 5.02, + "learning_rate": 6.414473974027313e-06, + "loss": 0.1428, + "step": 12002 + }, + { + "epoch": 5.03, + "learning_rate": 6.412891345230462e-06, + "loss": 0.1136, + "step": 12003 + }, + { + "epoch": 5.03, + "learning_rate": 6.411308819541863e-06, + "loss": 0.1047, + "step": 12004 + }, + { + "epoch": 5.03, + "learning_rate": 6.409726397007008e-06, + "loss": 0.1103, + "step": 12005 + }, + { + "epoch": 5.03, + "learning_rate": 6.408144077671381e-06, + "loss": 0.1787, + "step": 12006 + }, + { + "epoch": 5.03, + "learning_rate": 6.406561861580465e-06, + "loss": 0.0751, + "step": 12007 + }, + { + "epoch": 5.03, + "learning_rate": 6.404979748779738e-06, + "loss": 0.1224, + "step": 12008 + }, + { + "epoch": 5.03, + "learning_rate": 6.403397739314682e-06, + "loss": 0.1314, + "step": 12009 + }, + { + "epoch": 5.03, + "learning_rate": 6.4018158332307635e-06, + "loss": 0.1615, + "step": 12010 + }, + { + "epoch": 5.03, + "learning_rate": 6.400234030573456e-06, + "loss": 0.1406, + "step": 12011 + }, + { + "epoch": 5.03, + "learning_rate": 6.3986523313882285e-06, + "loss": 0.0791, + "step": 12012 + }, + { + "epoch": 5.03, + "learning_rate": 6.397070735720543e-06, + "loss": 0.1291, + "step": 12013 + }, + { + "epoch": 5.03, + "learning_rate": 6.395489243615861e-06, + "loss": 0.1109, + "step": 12014 + }, + { + "epoch": 5.03, + "learning_rate": 6.393907855119645e-06, + "loss": 0.1138, + "step": 12015 + }, + { + "epoch": 5.03, + "learning_rate": 6.392326570277345e-06, + "loss": 0.136, + "step": 12016 + }, + { + "epoch": 5.03, + "learning_rate": 6.390745389134416e-06, + "loss": 0.1366, + "step": 12017 + }, + { + "epoch": 5.03, + "learning_rate": 6.389164311736311e-06, + "loss": 0.1381, + "step": 12018 + }, + { + "epoch": 5.03, + "learning_rate": 6.387583338128471e-06, + "loss": 0.1289, + "step": 12019 + }, + { + "epoch": 5.03, + "learning_rate": 6.386002468356343e-06, + "loss": 0.1132, + "step": 12020 + }, + { + "epoch": 5.03, + "learning_rate": 6.384421702465369e-06, + "loss": 0.1303, + "step": 12021 + }, + { + "epoch": 5.03, + "learning_rate": 6.382841040500982e-06, + "loss": 0.1251, + "step": 12022 + }, + { + "epoch": 5.03, + "learning_rate": 6.38126048250862e-06, + "loss": 0.1139, + "step": 12023 + }, + { + "epoch": 5.03, + "learning_rate": 6.379680028533714e-06, + "loss": 0.0976, + "step": 12024 + }, + { + "epoch": 5.03, + "learning_rate": 6.378099678621697e-06, + "loss": 0.1316, + "step": 12025 + }, + { + "epoch": 5.03, + "learning_rate": 6.376519432817989e-06, + "loss": 0.1245, + "step": 12026 + }, + { + "epoch": 5.04, + "learning_rate": 6.374939291168012e-06, + "loss": 0.1231, + "step": 12027 + }, + { + "epoch": 5.04, + "learning_rate": 6.373359253717188e-06, + "loss": 0.1713, + "step": 12028 + }, + { + "epoch": 5.04, + "learning_rate": 6.3717793205109354e-06, + "loss": 0.1427, + "step": 12029 + }, + { + "epoch": 5.04, + "learning_rate": 6.370199491594667e-06, + "loss": 0.1467, + "step": 12030 + }, + { + "epoch": 5.04, + "learning_rate": 6.368619767013792e-06, + "loss": 0.1034, + "step": 12031 + }, + { + "epoch": 5.04, + "learning_rate": 6.367040146813721e-06, + "loss": 0.1073, + "step": 12032 + }, + { + "epoch": 5.04, + "learning_rate": 6.365460631039855e-06, + "loss": 0.1462, + "step": 12033 + }, + { + "epoch": 5.04, + "learning_rate": 6.363881219737599e-06, + "loss": 0.1324, + "step": 12034 + }, + { + "epoch": 5.04, + "learning_rate": 6.362301912952351e-06, + "loss": 0.1383, + "step": 12035 + }, + { + "epoch": 5.04, + "learning_rate": 6.360722710729506e-06, + "loss": 0.1589, + "step": 12036 + }, + { + "epoch": 5.04, + "learning_rate": 6.3591436131144565e-06, + "loss": 0.1377, + "step": 12037 + }, + { + "epoch": 5.04, + "learning_rate": 6.357564620152595e-06, + "loss": 0.1345, + "step": 12038 + }, + { + "epoch": 5.04, + "learning_rate": 6.355985731889305e-06, + "loss": 0.1272, + "step": 12039 + }, + { + "epoch": 5.04, + "learning_rate": 6.354406948369971e-06, + "loss": 0.1493, + "step": 12040 + }, + { + "epoch": 5.04, + "learning_rate": 6.352828269639974e-06, + "loss": 0.1311, + "step": 12041 + }, + { + "epoch": 5.04, + "learning_rate": 6.351249695744694e-06, + "loss": 0.1302, + "step": 12042 + }, + { + "epoch": 5.04, + "learning_rate": 6.349671226729502e-06, + "loss": 0.1074, + "step": 12043 + }, + { + "epoch": 5.04, + "learning_rate": 6.348092862639772e-06, + "loss": 0.1288, + "step": 12044 + }, + { + "epoch": 5.04, + "learning_rate": 6.346514603520873e-06, + "loss": 0.1185, + "step": 12045 + }, + { + "epoch": 5.04, + "learning_rate": 6.344936449418168e-06, + "loss": 0.1462, + "step": 12046 + }, + { + "epoch": 5.04, + "learning_rate": 6.343358400377021e-06, + "loss": 0.1116, + "step": 12047 + }, + { + "epoch": 5.04, + "learning_rate": 6.341780456442797e-06, + "loss": 0.1073, + "step": 12048 + }, + { + "epoch": 5.04, + "learning_rate": 6.340202617660842e-06, + "loss": 0.1362, + "step": 12049 + }, + { + "epoch": 5.05, + "learning_rate": 6.338624884076517e-06, + "loss": 0.0908, + "step": 12050 + }, + { + "epoch": 5.05, + "learning_rate": 6.337047255735169e-06, + "loss": 0.0934, + "step": 12051 + }, + { + "epoch": 5.05, + "learning_rate": 6.335469732682149e-06, + "loss": 0.1395, + "step": 12052 + }, + { + "epoch": 5.05, + "learning_rate": 6.3338923149628e-06, + "loss": 0.1196, + "step": 12053 + }, + { + "epoch": 5.05, + "learning_rate": 6.332315002622461e-06, + "loss": 0.1087, + "step": 12054 + }, + { + "epoch": 5.05, + "learning_rate": 6.330737795706474e-06, + "loss": 0.1016, + "step": 12055 + }, + { + "epoch": 5.05, + "learning_rate": 6.329160694260173e-06, + "loss": 0.1073, + "step": 12056 + }, + { + "epoch": 5.05, + "learning_rate": 6.3275836983288895e-06, + "loss": 0.1342, + "step": 12057 + }, + { + "epoch": 5.05, + "learning_rate": 6.326006807957954e-06, + "loss": 0.1287, + "step": 12058 + }, + { + "epoch": 5.05, + "learning_rate": 6.324430023192695e-06, + "loss": 0.1116, + "step": 12059 + }, + { + "epoch": 5.05, + "learning_rate": 6.32285334407843e-06, + "loss": 0.11, + "step": 12060 + }, + { + "epoch": 5.05, + "learning_rate": 6.3212767706604826e-06, + "loss": 0.1296, + "step": 12061 + }, + { + "epoch": 5.05, + "learning_rate": 6.319700302984173e-06, + "loss": 0.1233, + "step": 12062 + }, + { + "epoch": 5.05, + "learning_rate": 6.31812394109481e-06, + "loss": 0.0853, + "step": 12063 + }, + { + "epoch": 5.05, + "learning_rate": 6.316547685037707e-06, + "loss": 0.0856, + "step": 12064 + }, + { + "epoch": 5.05, + "learning_rate": 6.314971534858174e-06, + "loss": 0.1416, + "step": 12065 + }, + { + "epoch": 5.05, + "learning_rate": 6.313395490601513e-06, + "loss": 0.172, + "step": 12066 + }, + { + "epoch": 5.05, + "learning_rate": 6.311819552313026e-06, + "loss": 0.1376, + "step": 12067 + }, + { + "epoch": 5.05, + "learning_rate": 6.310243720038014e-06, + "loss": 0.1347, + "step": 12068 + }, + { + "epoch": 5.05, + "learning_rate": 6.308667993821773e-06, + "loss": 0.1733, + "step": 12069 + }, + { + "epoch": 5.05, + "learning_rate": 6.3070923737095946e-06, + "loss": 0.155, + "step": 12070 + }, + { + "epoch": 5.05, + "learning_rate": 6.305516859746772e-06, + "loss": 0.0885, + "step": 12071 + }, + { + "epoch": 5.05, + "learning_rate": 6.303941451978584e-06, + "loss": 0.1177, + "step": 12072 + }, + { + "epoch": 5.05, + "learning_rate": 6.302366150450323e-06, + "loss": 0.1401, + "step": 12073 + }, + { + "epoch": 5.06, + "learning_rate": 6.300790955207262e-06, + "loss": 0.1523, + "step": 12074 + }, + { + "epoch": 5.06, + "learning_rate": 6.299215866294683e-06, + "loss": 0.128, + "step": 12075 + }, + { + "epoch": 5.06, + "learning_rate": 6.297640883757863e-06, + "loss": 0.1039, + "step": 12076 + }, + { + "epoch": 5.06, + "learning_rate": 6.296066007642067e-06, + "loss": 0.1119, + "step": 12077 + }, + { + "epoch": 5.06, + "learning_rate": 6.294491237992566e-06, + "loss": 0.1343, + "step": 12078 + }, + { + "epoch": 5.06, + "learning_rate": 6.292916574854628e-06, + "loss": 0.1404, + "step": 12079 + }, + { + "epoch": 5.06, + "learning_rate": 6.291342018273514e-06, + "loss": 0.1355, + "step": 12080 + }, + { + "epoch": 5.06, + "learning_rate": 6.2897675682944824e-06, + "loss": 0.0829, + "step": 12081 + }, + { + "epoch": 5.06, + "learning_rate": 6.288193224962789e-06, + "loss": 0.0669, + "step": 12082 + }, + { + "epoch": 5.06, + "learning_rate": 6.2866189883236885e-06, + "loss": 0.1222, + "step": 12083 + }, + { + "epoch": 5.06, + "learning_rate": 6.285044858422429e-06, + "loss": 0.1228, + "step": 12084 + }, + { + "epoch": 5.06, + "learning_rate": 6.283470835304257e-06, + "loss": 0.116, + "step": 12085 + }, + { + "epoch": 5.06, + "learning_rate": 6.281896919014422e-06, + "loss": 0.12, + "step": 12086 + }, + { + "epoch": 5.06, + "learning_rate": 6.280323109598157e-06, + "loss": 0.138, + "step": 12087 + }, + { + "epoch": 5.06, + "learning_rate": 6.278749407100704e-06, + "loss": 0.1307, + "step": 12088 + }, + { + "epoch": 5.06, + "learning_rate": 6.2771758115673e-06, + "loss": 0.0881, + "step": 12089 + }, + { + "epoch": 5.06, + "learning_rate": 6.2756023230431705e-06, + "loss": 0.1232, + "step": 12090 + }, + { + "epoch": 5.06, + "learning_rate": 6.2740289415735474e-06, + "loss": 0.1266, + "step": 12091 + }, + { + "epoch": 5.06, + "learning_rate": 6.272455667203659e-06, + "loss": 0.1358, + "step": 12092 + }, + { + "epoch": 5.06, + "learning_rate": 6.270882499978723e-06, + "loss": 0.1208, + "step": 12093 + }, + { + "epoch": 5.06, + "learning_rate": 6.269309439943963e-06, + "loss": 0.1235, + "step": 12094 + }, + { + "epoch": 5.06, + "learning_rate": 6.2677364871445886e-06, + "loss": 0.1502, + "step": 12095 + }, + { + "epoch": 5.06, + "learning_rate": 6.2661636416258175e-06, + "loss": 0.1322, + "step": 12096 + }, + { + "epoch": 5.06, + "learning_rate": 6.264590903432861e-06, + "loss": 0.0914, + "step": 12097 + }, + { + "epoch": 5.07, + "learning_rate": 6.263018272610921e-06, + "loss": 0.123, + "step": 12098 + }, + { + "epoch": 5.07, + "learning_rate": 6.261445749205206e-06, + "loss": 0.1693, + "step": 12099 + }, + { + "epoch": 5.07, + "learning_rate": 6.259873333260917e-06, + "loss": 0.1072, + "step": 12100 + }, + { + "epoch": 5.07, + "learning_rate": 6.258301024823249e-06, + "loss": 0.0645, + "step": 12101 + }, + { + "epoch": 5.07, + "learning_rate": 6.256728823937397e-06, + "loss": 0.1359, + "step": 12102 + }, + { + "epoch": 5.07, + "learning_rate": 6.255156730648555e-06, + "loss": 0.1382, + "step": 12103 + }, + { + "epoch": 5.07, + "learning_rate": 6.253584745001908e-06, + "loss": 0.1039, + "step": 12104 + }, + { + "epoch": 5.07, + "learning_rate": 6.252012867042643e-06, + "loss": 0.1308, + "step": 12105 + }, + { + "epoch": 5.07, + "learning_rate": 6.250441096815945e-06, + "loss": 0.1262, + "step": 12106 + }, + { + "epoch": 5.07, + "learning_rate": 6.248869434366987e-06, + "loss": 0.1414, + "step": 12107 + }, + { + "epoch": 5.07, + "learning_rate": 6.247297879740948e-06, + "loss": 0.1307, + "step": 12108 + }, + { + "epoch": 5.07, + "learning_rate": 6.245726432983003e-06, + "loss": 0.1036, + "step": 12109 + }, + { + "epoch": 5.07, + "learning_rate": 6.244155094138322e-06, + "loss": 0.152, + "step": 12110 + }, + { + "epoch": 5.07, + "learning_rate": 6.2425838632520674e-06, + "loss": 0.1469, + "step": 12111 + }, + { + "epoch": 5.07, + "learning_rate": 6.2410127403694056e-06, + "loss": 0.1288, + "step": 12112 + }, + { + "epoch": 5.07, + "learning_rate": 6.239441725535498e-06, + "loss": 0.0963, + "step": 12113 + }, + { + "epoch": 5.07, + "learning_rate": 6.237870818795499e-06, + "loss": 0.0882, + "step": 12114 + }, + { + "epoch": 5.07, + "learning_rate": 6.2363000201945656e-06, + "loss": 0.1074, + "step": 12115 + }, + { + "epoch": 5.07, + "learning_rate": 6.234729329777853e-06, + "loss": 0.1028, + "step": 12116 + }, + { + "epoch": 5.07, + "learning_rate": 6.233158747590502e-06, + "loss": 0.1171, + "step": 12117 + }, + { + "epoch": 5.07, + "learning_rate": 6.231588273677658e-06, + "loss": 0.1217, + "step": 12118 + }, + { + "epoch": 5.07, + "learning_rate": 6.230017908084465e-06, + "loss": 0.1087, + "step": 12119 + }, + { + "epoch": 5.07, + "learning_rate": 6.228447650856062e-06, + "loss": 0.1036, + "step": 12120 + }, + { + "epoch": 5.07, + "learning_rate": 6.226877502037585e-06, + "loss": 0.0812, + "step": 12121 + }, + { + "epoch": 5.08, + "learning_rate": 6.225307461674166e-06, + "loss": 0.1108, + "step": 12122 + }, + { + "epoch": 5.08, + "learning_rate": 6.2237375298109336e-06, + "loss": 0.1215, + "step": 12123 + }, + { + "epoch": 5.08, + "learning_rate": 6.222167706493016e-06, + "loss": 0.1526, + "step": 12124 + }, + { + "epoch": 5.08, + "learning_rate": 6.220597991765536e-06, + "loss": 0.1058, + "step": 12125 + }, + { + "epoch": 5.08, + "learning_rate": 6.2190283856736115e-06, + "loss": 0.1068, + "step": 12126 + }, + { + "epoch": 5.08, + "learning_rate": 6.217458888262362e-06, + "loss": 0.1346, + "step": 12127 + }, + { + "epoch": 5.08, + "learning_rate": 6.215889499576898e-06, + "loss": 0.1367, + "step": 12128 + }, + { + "epoch": 5.08, + "learning_rate": 6.214320219662333e-06, + "loss": 0.1279, + "step": 12129 + }, + { + "epoch": 5.08, + "learning_rate": 6.212751048563776e-06, + "loss": 0.1141, + "step": 12130 + }, + { + "epoch": 5.08, + "learning_rate": 6.2111819863263275e-06, + "loss": 0.1321, + "step": 12131 + }, + { + "epoch": 5.08, + "learning_rate": 6.209613032995091e-06, + "loss": 0.1796, + "step": 12132 + }, + { + "epoch": 5.08, + "learning_rate": 6.208044188615166e-06, + "loss": 0.1399, + "step": 12133 + }, + { + "epoch": 5.08, + "learning_rate": 6.206475453231644e-06, + "loss": 0.1253, + "step": 12134 + }, + { + "epoch": 5.08, + "learning_rate": 6.204906826889618e-06, + "loss": 0.1111, + "step": 12135 + }, + { + "epoch": 5.08, + "learning_rate": 6.203338309634179e-06, + "loss": 0.1165, + "step": 12136 + }, + { + "epoch": 5.08, + "learning_rate": 6.201769901510412e-06, + "loss": 0.1071, + "step": 12137 + }, + { + "epoch": 5.08, + "learning_rate": 6.200201602563397e-06, + "loss": 0.1172, + "step": 12138 + }, + { + "epoch": 5.08, + "learning_rate": 6.1986334128382195e-06, + "loss": 0.1291, + "step": 12139 + }, + { + "epoch": 5.08, + "learning_rate": 6.197065332379947e-06, + "loss": 0.117, + "step": 12140 + }, + { + "epoch": 5.08, + "learning_rate": 6.195497361233658e-06, + "loss": 0.1428, + "step": 12141 + }, + { + "epoch": 5.08, + "learning_rate": 6.19392949944442e-06, + "loss": 0.1521, + "step": 12142 + }, + { + "epoch": 5.08, + "learning_rate": 6.192361747057301e-06, + "loss": 0.128, + "step": 12143 + }, + { + "epoch": 5.08, + "learning_rate": 6.190794104117368e-06, + "loss": 0.1004, + "step": 12144 + }, + { + "epoch": 5.08, + "learning_rate": 6.189226570669675e-06, + "loss": 0.1502, + "step": 12145 + }, + { + "epoch": 5.09, + "learning_rate": 6.187659146759282e-06, + "loss": 0.1702, + "step": 12146 + }, + { + "epoch": 5.09, + "learning_rate": 6.186091832431247e-06, + "loss": 0.1418, + "step": 12147 + }, + { + "epoch": 5.09, + "learning_rate": 6.1845246277306145e-06, + "loss": 0.1337, + "step": 12148 + }, + { + "epoch": 5.09, + "learning_rate": 6.182957532702437e-06, + "loss": 0.1371, + "step": 12149 + }, + { + "epoch": 5.09, + "learning_rate": 6.181390547391758e-06, + "loss": 0.1352, + "step": 12150 + }, + { + "epoch": 5.09, + "learning_rate": 6.179823671843621e-06, + "loss": 0.138, + "step": 12151 + }, + { + "epoch": 5.09, + "learning_rate": 6.178256906103061e-06, + "loss": 0.1226, + "step": 12152 + }, + { + "epoch": 5.09, + "learning_rate": 6.176690250215114e-06, + "loss": 0.1264, + "step": 12153 + }, + { + "epoch": 5.09, + "learning_rate": 6.175123704224814e-06, + "loss": 0.1309, + "step": 12154 + }, + { + "epoch": 5.09, + "learning_rate": 6.173557268177189e-06, + "loss": 0.135, + "step": 12155 + }, + { + "epoch": 5.09, + "learning_rate": 6.171990942117265e-06, + "loss": 0.1154, + "step": 12156 + }, + { + "epoch": 5.09, + "learning_rate": 6.170424726090065e-06, + "loss": 0.1521, + "step": 12157 + }, + { + "epoch": 5.09, + "learning_rate": 6.168858620140606e-06, + "loss": 0.1361, + "step": 12158 + }, + { + "epoch": 5.09, + "learning_rate": 6.167292624313908e-06, + "loss": 0.1125, + "step": 12159 + }, + { + "epoch": 5.09, + "learning_rate": 6.165726738654982e-06, + "loss": 0.0856, + "step": 12160 + }, + { + "epoch": 5.09, + "learning_rate": 6.16416096320884e-06, + "loss": 0.1175, + "step": 12161 + }, + { + "epoch": 5.09, + "learning_rate": 6.162595298020487e-06, + "loss": 0.0999, + "step": 12162 + }, + { + "epoch": 5.09, + "learning_rate": 6.1610297431349255e-06, + "loss": 0.1059, + "step": 12163 + }, + { + "epoch": 5.09, + "learning_rate": 6.1594642985971565e-06, + "loss": 0.1038, + "step": 12164 + }, + { + "epoch": 5.09, + "learning_rate": 6.15789896445218e-06, + "loss": 0.1068, + "step": 12165 + }, + { + "epoch": 5.09, + "learning_rate": 6.1563337407449865e-06, + "loss": 0.1114, + "step": 12166 + }, + { + "epoch": 5.09, + "learning_rate": 6.15476862752057e-06, + "loss": 0.1109, + "step": 12167 + }, + { + "epoch": 5.09, + "learning_rate": 6.153203624823918e-06, + "loss": 0.1219, + "step": 12168 + }, + { + "epoch": 5.09, + "learning_rate": 6.151638732700012e-06, + "loss": 0.1144, + "step": 12169 + }, + { + "epoch": 5.1, + "learning_rate": 6.150073951193836e-06, + "loss": 0.1246, + "step": 12170 + }, + { + "epoch": 5.1, + "learning_rate": 6.1485092803503695e-06, + "loss": 0.1424, + "step": 12171 + }, + { + "epoch": 5.1, + "learning_rate": 6.146944720214585e-06, + "loss": 0.1515, + "step": 12172 + }, + { + "epoch": 5.1, + "learning_rate": 6.145380270831455e-06, + "loss": 0.1316, + "step": 12173 + }, + { + "epoch": 5.1, + "learning_rate": 6.143815932245951e-06, + "loss": 0.1069, + "step": 12174 + }, + { + "epoch": 5.1, + "learning_rate": 6.142251704503033e-06, + "loss": 0.1006, + "step": 12175 + }, + { + "epoch": 5.1, + "learning_rate": 6.140687587647667e-06, + "loss": 0.0906, + "step": 12176 + }, + { + "epoch": 5.1, + "learning_rate": 6.139123581724813e-06, + "loss": 0.0751, + "step": 12177 + }, + { + "epoch": 5.1, + "learning_rate": 6.1375596867794245e-06, + "loss": 0.1319, + "step": 12178 + }, + { + "epoch": 5.1, + "learning_rate": 6.135995902856455e-06, + "loss": 0.0637, + "step": 12179 + }, + { + "epoch": 5.1, + "learning_rate": 6.134432230000854e-06, + "loss": 0.1416, + "step": 12180 + }, + { + "epoch": 5.1, + "learning_rate": 6.132868668257571e-06, + "loss": 0.1236, + "step": 12181 + }, + { + "epoch": 5.1, + "learning_rate": 6.131305217671544e-06, + "loss": 0.0791, + "step": 12182 + }, + { + "epoch": 5.1, + "learning_rate": 6.129741878287716e-06, + "loss": 0.106, + "step": 12183 + }, + { + "epoch": 5.1, + "learning_rate": 6.128178650151026e-06, + "loss": 0.1149, + "step": 12184 + }, + { + "epoch": 5.1, + "learning_rate": 6.126615533306405e-06, + "loss": 0.131, + "step": 12185 + }, + { + "epoch": 5.1, + "learning_rate": 6.12505252779878e-06, + "loss": 0.1268, + "step": 12186 + }, + { + "epoch": 5.1, + "learning_rate": 6.123489633673082e-06, + "loss": 0.0994, + "step": 12187 + }, + { + "epoch": 5.1, + "learning_rate": 6.121926850974236e-06, + "loss": 0.1267, + "step": 12188 + }, + { + "epoch": 5.1, + "learning_rate": 6.120364179747161e-06, + "loss": 0.1149, + "step": 12189 + }, + { + "epoch": 5.1, + "learning_rate": 6.1188016200367715e-06, + "loss": 0.1255, + "step": 12190 + }, + { + "epoch": 5.1, + "learning_rate": 6.117239171887987e-06, + "loss": 0.1498, + "step": 12191 + }, + { + "epoch": 5.1, + "learning_rate": 6.11567683534572e-06, + "loss": 0.106, + "step": 12192 + }, + { + "epoch": 5.1, + "learning_rate": 6.1141146104548735e-06, + "loss": 0.1239, + "step": 12193 + }, + { + "epoch": 5.11, + "learning_rate": 6.112552497260353e-06, + "loss": 0.0943, + "step": 12194 + }, + { + "epoch": 5.11, + "learning_rate": 6.110990495807064e-06, + "loss": 0.1374, + "step": 12195 + }, + { + "epoch": 5.11, + "learning_rate": 6.1094286061399e-06, + "loss": 0.0907, + "step": 12196 + }, + { + "epoch": 5.11, + "learning_rate": 6.107866828303759e-06, + "loss": 0.1415, + "step": 12197 + }, + { + "epoch": 5.11, + "learning_rate": 6.106305162343533e-06, + "loss": 0.0777, + "step": 12198 + }, + { + "epoch": 5.11, + "learning_rate": 6.104743608304109e-06, + "loss": 0.1269, + "step": 12199 + }, + { + "epoch": 5.11, + "learning_rate": 6.103182166230373e-06, + "loss": 0.1295, + "step": 12200 + }, + { + "epoch": 5.11, + "learning_rate": 6.101620836167208e-06, + "loss": 0.1536, + "step": 12201 + }, + { + "epoch": 5.11, + "learning_rate": 6.100059618159493e-06, + "loss": 0.1431, + "step": 12202 + }, + { + "epoch": 5.11, + "learning_rate": 6.098498512252102e-06, + "loss": 0.0914, + "step": 12203 + }, + { + "epoch": 5.11, + "learning_rate": 6.096937518489912e-06, + "loss": 0.1229, + "step": 12204 + }, + { + "epoch": 5.11, + "learning_rate": 6.095376636917787e-06, + "loss": 0.124, + "step": 12205 + }, + { + "epoch": 5.11, + "learning_rate": 6.093815867580597e-06, + "loss": 0.1116, + "step": 12206 + }, + { + "epoch": 5.11, + "learning_rate": 6.0922552105232055e-06, + "loss": 0.1509, + "step": 12207 + }, + { + "epoch": 5.11, + "learning_rate": 6.0906946657904685e-06, + "loss": 0.1288, + "step": 12208 + }, + { + "epoch": 5.11, + "learning_rate": 6.089134233427245e-06, + "loss": 0.1412, + "step": 12209 + }, + { + "epoch": 5.11, + "learning_rate": 6.087573913478386e-06, + "loss": 0.1191, + "step": 12210 + }, + { + "epoch": 5.11, + "learning_rate": 6.086013705988742e-06, + "loss": 0.1093, + "step": 12211 + }, + { + "epoch": 5.11, + "learning_rate": 6.084453611003164e-06, + "loss": 0.0971, + "step": 12212 + }, + { + "epoch": 5.11, + "learning_rate": 6.08289362856649e-06, + "loss": 0.0934, + "step": 12213 + }, + { + "epoch": 5.11, + "learning_rate": 6.081333758723562e-06, + "loss": 0.1281, + "step": 12214 + }, + { + "epoch": 5.11, + "learning_rate": 6.079774001519221e-06, + "loss": 0.1253, + "step": 12215 + }, + { + "epoch": 5.11, + "learning_rate": 6.078214356998294e-06, + "loss": 0.1323, + "step": 12216 + }, + { + "epoch": 5.11, + "learning_rate": 6.076654825205617e-06, + "loss": 0.154, + "step": 12217 + }, + { + "epoch": 5.12, + "learning_rate": 6.075095406186015e-06, + "loss": 0.0858, + "step": 12218 + }, + { + "epoch": 5.12, + "learning_rate": 6.073536099984314e-06, + "loss": 0.1258, + "step": 12219 + }, + { + "epoch": 5.12, + "learning_rate": 6.071976906645332e-06, + "loss": 0.1146, + "step": 12220 + }, + { + "epoch": 5.12, + "learning_rate": 6.070417826213888e-06, + "loss": 0.1333, + "step": 12221 + }, + { + "epoch": 5.12, + "learning_rate": 6.068858858734799e-06, + "loss": 0.096, + "step": 12222 + }, + { + "epoch": 5.12, + "learning_rate": 6.06730000425287e-06, + "loss": 0.1071, + "step": 12223 + }, + { + "epoch": 5.12, + "learning_rate": 6.065741262812914e-06, + "loss": 0.1269, + "step": 12224 + }, + { + "epoch": 5.12, + "learning_rate": 6.064182634459735e-06, + "loss": 0.1314, + "step": 12225 + }, + { + "epoch": 5.12, + "learning_rate": 6.062624119238133e-06, + "loss": 0.1292, + "step": 12226 + }, + { + "epoch": 5.12, + "learning_rate": 6.061065717192907e-06, + "loss": 0.087, + "step": 12227 + }, + { + "epoch": 5.12, + "learning_rate": 6.059507428368852e-06, + "loss": 0.1242, + "step": 12228 + }, + { + "epoch": 5.12, + "learning_rate": 6.057949252810762e-06, + "loss": 0.1421, + "step": 12229 + }, + { + "epoch": 5.12, + "learning_rate": 6.056391190563419e-06, + "loss": 0.1274, + "step": 12230 + }, + { + "epoch": 5.12, + "learning_rate": 6.054833241671611e-06, + "loss": 0.1439, + "step": 12231 + }, + { + "epoch": 5.12, + "learning_rate": 6.053275406180122e-06, + "loss": 0.1193, + "step": 12232 + }, + { + "epoch": 5.12, + "learning_rate": 6.051717684133729e-06, + "loss": 0.1221, + "step": 12233 + }, + { + "epoch": 5.12, + "learning_rate": 6.0501600755772074e-06, + "loss": 0.1642, + "step": 12234 + }, + { + "epoch": 5.12, + "learning_rate": 6.048602580555328e-06, + "loss": 0.1479, + "step": 12235 + }, + { + "epoch": 5.12, + "learning_rate": 6.047045199112865e-06, + "loss": 0.1157, + "step": 12236 + }, + { + "epoch": 5.12, + "learning_rate": 6.0454879312945755e-06, + "loss": 0.1253, + "step": 12237 + }, + { + "epoch": 5.12, + "learning_rate": 6.043930777145228e-06, + "loss": 0.1043, + "step": 12238 + }, + { + "epoch": 5.12, + "learning_rate": 6.042373736709581e-06, + "loss": 0.0997, + "step": 12239 + }, + { + "epoch": 5.12, + "learning_rate": 6.040816810032387e-06, + "loss": 0.1033, + "step": 12240 + }, + { + "epoch": 5.12, + "learning_rate": 6.039259997158401e-06, + "loss": 0.1221, + "step": 12241 + }, + { + "epoch": 5.13, + "learning_rate": 6.037703298132373e-06, + "loss": 0.1313, + "step": 12242 + }, + { + "epoch": 5.13, + "learning_rate": 6.036146712999047e-06, + "loss": 0.1127, + "step": 12243 + }, + { + "epoch": 5.13, + "learning_rate": 6.0345902418031645e-06, + "loss": 0.0897, + "step": 12244 + }, + { + "epoch": 5.13, + "learning_rate": 6.03303388458947e-06, + "loss": 0.1243, + "step": 12245 + }, + { + "epoch": 5.13, + "learning_rate": 6.031477641402693e-06, + "loss": 0.1178, + "step": 12246 + }, + { + "epoch": 5.13, + "learning_rate": 6.029921512287571e-06, + "loss": 0.0931, + "step": 12247 + }, + { + "epoch": 5.13, + "learning_rate": 6.028365497288831e-06, + "loss": 0.1232, + "step": 12248 + }, + { + "epoch": 5.13, + "learning_rate": 6.026809596451203e-06, + "loss": 0.099, + "step": 12249 + }, + { + "epoch": 5.13, + "learning_rate": 6.025253809819405e-06, + "loss": 0.0836, + "step": 12250 + }, + { + "epoch": 5.13, + "learning_rate": 6.023698137438159e-06, + "loss": 0.1112, + "step": 12251 + }, + { + "epoch": 5.13, + "learning_rate": 6.022142579352186e-06, + "loss": 0.0906, + "step": 12252 + }, + { + "epoch": 5.13, + "learning_rate": 6.020587135606193e-06, + "loss": 0.1082, + "step": 12253 + }, + { + "epoch": 5.13, + "learning_rate": 6.019031806244888e-06, + "loss": 0.0938, + "step": 12254 + }, + { + "epoch": 5.13, + "learning_rate": 6.017476591312983e-06, + "loss": 0.1267, + "step": 12255 + }, + { + "epoch": 5.13, + "learning_rate": 6.015921490855181e-06, + "loss": 0.1103, + "step": 12256 + }, + { + "epoch": 5.13, + "learning_rate": 6.014366504916178e-06, + "loss": 0.1016, + "step": 12257 + }, + { + "epoch": 5.13, + "learning_rate": 6.012811633540672e-06, + "loss": 0.1498, + "step": 12258 + }, + { + "epoch": 5.13, + "learning_rate": 6.011256876773362e-06, + "loss": 0.1537, + "step": 12259 + }, + { + "epoch": 5.13, + "learning_rate": 6.009702234658929e-06, + "loss": 0.1215, + "step": 12260 + }, + { + "epoch": 5.13, + "learning_rate": 6.008147707242065e-06, + "loss": 0.1021, + "step": 12261 + }, + { + "epoch": 5.13, + "learning_rate": 6.0065932945674535e-06, + "loss": 0.1175, + "step": 12262 + }, + { + "epoch": 5.13, + "learning_rate": 6.005038996679775e-06, + "loss": 0.1038, + "step": 12263 + }, + { + "epoch": 5.13, + "learning_rate": 6.0034848136237035e-06, + "loss": 0.118, + "step": 12264 + }, + { + "epoch": 5.14, + "learning_rate": 6.001930745443916e-06, + "loss": 0.1179, + "step": 12265 + }, + { + "epoch": 5.14, + "learning_rate": 6.000376792185082e-06, + "loss": 0.1159, + "step": 12266 + }, + { + "epoch": 5.14, + "learning_rate": 5.998822953891866e-06, + "loss": 0.1377, + "step": 12267 + }, + { + "epoch": 5.14, + "learning_rate": 5.997269230608935e-06, + "loss": 0.167, + "step": 12268 + }, + { + "epoch": 5.14, + "learning_rate": 5.995715622380949e-06, + "loss": 0.1558, + "step": 12269 + }, + { + "epoch": 5.14, + "learning_rate": 5.994162129252561e-06, + "loss": 0.1122, + "step": 12270 + }, + { + "epoch": 5.14, + "learning_rate": 5.992608751268429e-06, + "loss": 0.0987, + "step": 12271 + }, + { + "epoch": 5.14, + "learning_rate": 5.9910554884732056e-06, + "loss": 0.0981, + "step": 12272 + }, + { + "epoch": 5.14, + "learning_rate": 5.989502340911531e-06, + "loss": 0.1281, + "step": 12273 + }, + { + "epoch": 5.14, + "learning_rate": 5.987949308628054e-06, + "loss": 0.1062, + "step": 12274 + }, + { + "epoch": 5.14, + "learning_rate": 5.986396391667416e-06, + "loss": 0.0926, + "step": 12275 + }, + { + "epoch": 5.14, + "learning_rate": 5.98484359007425e-06, + "loss": 0.1085, + "step": 12276 + }, + { + "epoch": 5.14, + "learning_rate": 5.983290903893193e-06, + "loss": 0.1109, + "step": 12277 + }, + { + "epoch": 5.14, + "learning_rate": 5.981738333168873e-06, + "loss": 0.0862, + "step": 12278 + }, + { + "epoch": 5.14, + "learning_rate": 5.980185877945918e-06, + "loss": 0.1289, + "step": 12279 + }, + { + "epoch": 5.14, + "learning_rate": 5.978633538268957e-06, + "loss": 0.1209, + "step": 12280 + }, + { + "epoch": 5.14, + "learning_rate": 5.977081314182602e-06, + "loss": 0.1088, + "step": 12281 + }, + { + "epoch": 5.14, + "learning_rate": 5.975529205731476e-06, + "loss": 0.1414, + "step": 12282 + }, + { + "epoch": 5.14, + "learning_rate": 5.973977212960194e-06, + "loss": 0.0954, + "step": 12283 + }, + { + "epoch": 5.14, + "learning_rate": 5.9724253359133614e-06, + "loss": 0.1335, + "step": 12284 + }, + { + "epoch": 5.14, + "learning_rate": 5.97087357463559e-06, + "loss": 0.1011, + "step": 12285 + }, + { + "epoch": 5.14, + "learning_rate": 5.969321929171482e-06, + "loss": 0.104, + "step": 12286 + }, + { + "epoch": 5.14, + "learning_rate": 5.967770399565638e-06, + "loss": 0.1147, + "step": 12287 + }, + { + "epoch": 5.14, + "learning_rate": 5.966218985862655e-06, + "loss": 0.1452, + "step": 12288 + }, + { + "epoch": 5.15, + "learning_rate": 5.9646676881071305e-06, + "loss": 0.1098, + "step": 12289 + }, + { + "epoch": 5.15, + "learning_rate": 5.963116506343649e-06, + "loss": 0.1594, + "step": 12290 + }, + { + "epoch": 5.15, + "learning_rate": 5.961565440616803e-06, + "loss": 0.1281, + "step": 12291 + }, + { + "epoch": 5.15, + "learning_rate": 5.9600144909711745e-06, + "loss": 0.1219, + "step": 12292 + }, + { + "epoch": 5.15, + "learning_rate": 5.958463657451345e-06, + "loss": 0.1546, + "step": 12293 + }, + { + "epoch": 5.15, + "learning_rate": 5.956912940101891e-06, + "loss": 0.1529, + "step": 12294 + }, + { + "epoch": 5.15, + "learning_rate": 5.9553623389673855e-06, + "loss": 0.126, + "step": 12295 + }, + { + "epoch": 5.15, + "learning_rate": 5.953811854092404e-06, + "loss": 0.1451, + "step": 12296 + }, + { + "epoch": 5.15, + "learning_rate": 5.952261485521511e-06, + "loss": 0.15, + "step": 12297 + }, + { + "epoch": 5.15, + "learning_rate": 5.950711233299265e-06, + "loss": 0.1165, + "step": 12298 + }, + { + "epoch": 5.15, + "learning_rate": 5.949161097470233e-06, + "loss": 0.1291, + "step": 12299 + }, + { + "epoch": 5.15, + "learning_rate": 5.947611078078972e-06, + "loss": 0.1399, + "step": 12300 + }, + { + "epoch": 5.15, + "learning_rate": 5.9460611751700325e-06, + "loss": 0.1133, + "step": 12301 + }, + { + "epoch": 5.15, + "learning_rate": 5.944511388787967e-06, + "loss": 0.1605, + "step": 12302 + }, + { + "epoch": 5.15, + "learning_rate": 5.942961718977323e-06, + "loss": 0.089, + "step": 12303 + }, + { + "epoch": 5.15, + "learning_rate": 5.941412165782645e-06, + "loss": 0.1576, + "step": 12304 + }, + { + "epoch": 5.15, + "learning_rate": 5.939862729248472e-06, + "loss": 0.131, + "step": 12305 + }, + { + "epoch": 5.15, + "learning_rate": 5.9383134094193425e-06, + "loss": 0.0974, + "step": 12306 + }, + { + "epoch": 5.15, + "learning_rate": 5.936764206339791e-06, + "loss": 0.0774, + "step": 12307 + }, + { + "epoch": 5.15, + "learning_rate": 5.935215120054345e-06, + "loss": 0.1228, + "step": 12308 + }, + { + "epoch": 5.15, + "learning_rate": 5.933666150607533e-06, + "loss": 0.1299, + "step": 12309 + }, + { + "epoch": 5.15, + "learning_rate": 5.932117298043882e-06, + "loss": 0.1099, + "step": 12310 + }, + { + "epoch": 5.15, + "learning_rate": 5.930568562407905e-06, + "loss": 0.1544, + "step": 12311 + }, + { + "epoch": 5.15, + "learning_rate": 5.9290199437441256e-06, + "loss": 0.1006, + "step": 12312 + }, + { + "epoch": 5.16, + "learning_rate": 5.927471442097057e-06, + "loss": 0.1432, + "step": 12313 + }, + { + "epoch": 5.16, + "learning_rate": 5.925923057511207e-06, + "loss": 0.1034, + "step": 12314 + }, + { + "epoch": 5.16, + "learning_rate": 5.9243747900310825e-06, + "loss": 0.1427, + "step": 12315 + }, + { + "epoch": 5.16, + "learning_rate": 5.922826639701189e-06, + "loss": 0.1157, + "step": 12316 + }, + { + "epoch": 5.16, + "learning_rate": 5.921278606566026e-06, + "loss": 0.1101, + "step": 12317 + }, + { + "epoch": 5.16, + "learning_rate": 5.9197306906700865e-06, + "loss": 0.1233, + "step": 12318 + }, + { + "epoch": 5.16, + "learning_rate": 5.918182892057871e-06, + "loss": 0.1274, + "step": 12319 + }, + { + "epoch": 5.16, + "learning_rate": 5.916635210773869e-06, + "loss": 0.1298, + "step": 12320 + }, + { + "epoch": 5.16, + "learning_rate": 5.915087646862561e-06, + "loss": 0.0994, + "step": 12321 + }, + { + "epoch": 5.16, + "learning_rate": 5.913540200368434e-06, + "loss": 0.0831, + "step": 12322 + }, + { + "epoch": 5.16, + "learning_rate": 5.911992871335967e-06, + "loss": 0.1441, + "step": 12323 + }, + { + "epoch": 5.16, + "learning_rate": 5.910445659809638e-06, + "loss": 0.0953, + "step": 12324 + }, + { + "epoch": 5.16, + "learning_rate": 5.908898565833918e-06, + "loss": 0.1614, + "step": 12325 + }, + { + "epoch": 5.16, + "learning_rate": 5.907351589453279e-06, + "loss": 0.1289, + "step": 12326 + }, + { + "epoch": 5.16, + "learning_rate": 5.905804730712189e-06, + "loss": 0.1128, + "step": 12327 + }, + { + "epoch": 5.16, + "learning_rate": 5.904257989655107e-06, + "loss": 0.0952, + "step": 12328 + }, + { + "epoch": 5.16, + "learning_rate": 5.902711366326493e-06, + "loss": 0.1338, + "step": 12329 + }, + { + "epoch": 5.16, + "learning_rate": 5.901164860770808e-06, + "loss": 0.1145, + "step": 12330 + }, + { + "epoch": 5.16, + "learning_rate": 5.8996184730325005e-06, + "loss": 0.1594, + "step": 12331 + }, + { + "epoch": 5.16, + "learning_rate": 5.89807220315602e-06, + "loss": 0.1076, + "step": 12332 + }, + { + "epoch": 5.16, + "learning_rate": 5.896526051185814e-06, + "loss": 0.1145, + "step": 12333 + }, + { + "epoch": 5.16, + "learning_rate": 5.894980017166329e-06, + "loss": 0.1147, + "step": 12334 + }, + { + "epoch": 5.16, + "learning_rate": 5.893434101141997e-06, + "loss": 0.1126, + "step": 12335 + }, + { + "epoch": 5.16, + "learning_rate": 5.891888303157258e-06, + "loss": 0.0837, + "step": 12336 + }, + { + "epoch": 5.17, + "learning_rate": 5.890342623256546e-06, + "loss": 0.0819, + "step": 12337 + }, + { + "epoch": 5.17, + "learning_rate": 5.888797061484288e-06, + "loss": 0.1497, + "step": 12338 + }, + { + "epoch": 5.17, + "learning_rate": 5.887251617884909e-06, + "loss": 0.1503, + "step": 12339 + }, + { + "epoch": 5.17, + "learning_rate": 5.885706292502835e-06, + "loss": 0.1024, + "step": 12340 + }, + { + "epoch": 5.17, + "learning_rate": 5.884161085382481e-06, + "loss": 0.0938, + "step": 12341 + }, + { + "epoch": 5.17, + "learning_rate": 5.882615996568264e-06, + "loss": 0.1657, + "step": 12342 + }, + { + "epoch": 5.17, + "learning_rate": 5.8810710261046e-06, + "loss": 0.0848, + "step": 12343 + }, + { + "epoch": 5.17, + "learning_rate": 5.87952617403589e-06, + "loss": 0.1433, + "step": 12344 + }, + { + "epoch": 5.17, + "learning_rate": 5.877981440406547e-06, + "loss": 0.1287, + "step": 12345 + }, + { + "epoch": 5.17, + "learning_rate": 5.876436825260967e-06, + "loss": 0.1218, + "step": 12346 + }, + { + "epoch": 5.17, + "learning_rate": 5.8748923286435515e-06, + "loss": 0.1001, + "step": 12347 + }, + { + "epoch": 5.17, + "learning_rate": 5.873347950598697e-06, + "loss": 0.1375, + "step": 12348 + }, + { + "epoch": 5.17, + "learning_rate": 5.871803691170792e-06, + "loss": 0.1083, + "step": 12349 + }, + { + "epoch": 5.17, + "learning_rate": 5.870259550404227e-06, + "loss": 0.1457, + "step": 12350 + }, + { + "epoch": 5.17, + "learning_rate": 5.868715528343387e-06, + "loss": 0.1202, + "step": 12351 + }, + { + "epoch": 5.17, + "learning_rate": 5.8671716250326525e-06, + "loss": 0.0686, + "step": 12352 + }, + { + "epoch": 5.17, + "learning_rate": 5.865627840516402e-06, + "loss": 0.1238, + "step": 12353 + }, + { + "epoch": 5.17, + "learning_rate": 5.864084174839013e-06, + "loss": 0.0738, + "step": 12354 + }, + { + "epoch": 5.17, + "learning_rate": 5.86254062804485e-06, + "loss": 0.1097, + "step": 12355 + }, + { + "epoch": 5.17, + "learning_rate": 5.860997200178288e-06, + "loss": 0.1317, + "step": 12356 + }, + { + "epoch": 5.17, + "learning_rate": 5.859453891283688e-06, + "loss": 0.1103, + "step": 12357 + }, + { + "epoch": 5.17, + "learning_rate": 5.857910701405411e-06, + "loss": 0.1518, + "step": 12358 + }, + { + "epoch": 5.17, + "learning_rate": 5.8563676305878155e-06, + "loss": 0.1448, + "step": 12359 + }, + { + "epoch": 5.17, + "learning_rate": 5.854824678875256e-06, + "loss": 0.1091, + "step": 12360 + }, + { + "epoch": 5.18, + "learning_rate": 5.853281846312084e-06, + "loss": 0.1521, + "step": 12361 + }, + { + "epoch": 5.18, + "learning_rate": 5.851739132942645e-06, + "loss": 0.1361, + "step": 12362 + }, + { + "epoch": 5.18, + "learning_rate": 5.850196538811282e-06, + "loss": 0.1437, + "step": 12363 + }, + { + "epoch": 5.18, + "learning_rate": 5.848654063962338e-06, + "loss": 0.135, + "step": 12364 + }, + { + "epoch": 5.18, + "learning_rate": 5.847111708440152e-06, + "loss": 0.1703, + "step": 12365 + }, + { + "epoch": 5.18, + "learning_rate": 5.845569472289051e-06, + "loss": 0.1481, + "step": 12366 + }, + { + "epoch": 5.18, + "learning_rate": 5.844027355553369e-06, + "loss": 0.1317, + "step": 12367 + }, + { + "epoch": 5.18, + "learning_rate": 5.842485358277435e-06, + "loss": 0.1071, + "step": 12368 + }, + { + "epoch": 5.18, + "learning_rate": 5.840943480505567e-06, + "loss": 0.1361, + "step": 12369 + }, + { + "epoch": 5.18, + "learning_rate": 5.839401722282089e-06, + "loss": 0.113, + "step": 12370 + }, + { + "epoch": 5.18, + "learning_rate": 5.837860083651316e-06, + "loss": 0.0978, + "step": 12371 + }, + { + "epoch": 5.18, + "learning_rate": 5.836318564657561e-06, + "loss": 0.1384, + "step": 12372 + }, + { + "epoch": 5.18, + "learning_rate": 5.834777165345138e-06, + "loss": 0.0803, + "step": 12373 + }, + { + "epoch": 5.18, + "learning_rate": 5.833235885758346e-06, + "loss": 0.1489, + "step": 12374 + }, + { + "epoch": 5.18, + "learning_rate": 5.831694725941491e-06, + "loss": 0.1497, + "step": 12375 + }, + { + "epoch": 5.18, + "learning_rate": 5.830153685938873e-06, + "loss": 0.17, + "step": 12376 + }, + { + "epoch": 5.18, + "learning_rate": 5.8286127657947854e-06, + "loss": 0.1264, + "step": 12377 + }, + { + "epoch": 5.18, + "learning_rate": 5.827071965553524e-06, + "loss": 0.1041, + "step": 12378 + }, + { + "epoch": 5.18, + "learning_rate": 5.825531285259378e-06, + "loss": 0.1206, + "step": 12379 + }, + { + "epoch": 5.18, + "learning_rate": 5.8239907249566274e-06, + "loss": 0.1378, + "step": 12380 + }, + { + "epoch": 5.18, + "learning_rate": 5.822450284689558e-06, + "loss": 0.128, + "step": 12381 + }, + { + "epoch": 5.18, + "learning_rate": 5.820909964502448e-06, + "loss": 0.1407, + "step": 12382 + }, + { + "epoch": 5.18, + "learning_rate": 5.819369764439573e-06, + "loss": 0.1534, + "step": 12383 + }, + { + "epoch": 5.18, + "learning_rate": 5.817829684545204e-06, + "loss": 0.173, + "step": 12384 + }, + { + "epoch": 5.19, + "learning_rate": 5.816289724863609e-06, + "loss": 0.113, + "step": 12385 + }, + { + "epoch": 5.19, + "learning_rate": 5.814749885439057e-06, + "loss": 0.1181, + "step": 12386 + }, + { + "epoch": 5.19, + "learning_rate": 5.813210166315804e-06, + "loss": 0.1201, + "step": 12387 + }, + { + "epoch": 5.19, + "learning_rate": 5.81167056753811e-06, + "loss": 0.1776, + "step": 12388 + }, + { + "epoch": 5.19, + "learning_rate": 5.810131089150228e-06, + "loss": 0.146, + "step": 12389 + }, + { + "epoch": 5.19, + "learning_rate": 5.808591731196409e-06, + "loss": 0.1464, + "step": 12390 + }, + { + "epoch": 5.19, + "learning_rate": 5.8070524937209e-06, + "loss": 0.169, + "step": 12391 + }, + { + "epoch": 5.19, + "learning_rate": 5.805513376767947e-06, + "loss": 0.1243, + "step": 12392 + }, + { + "epoch": 5.19, + "learning_rate": 5.8039743803817904e-06, + "loss": 0.1317, + "step": 12393 + }, + { + "epoch": 5.19, + "learning_rate": 5.80243550460667e-06, + "loss": 0.1261, + "step": 12394 + }, + { + "epoch": 5.19, + "learning_rate": 5.800896749486813e-06, + "loss": 0.1377, + "step": 12395 + }, + { + "epoch": 5.19, + "learning_rate": 5.799358115066453e-06, + "loss": 0.1073, + "step": 12396 + }, + { + "epoch": 5.19, + "learning_rate": 5.797819601389815e-06, + "loss": 0.0979, + "step": 12397 + }, + { + "epoch": 5.19, + "learning_rate": 5.7962812085011245e-06, + "loss": 0.0694, + "step": 12398 + }, + { + "epoch": 5.19, + "learning_rate": 5.7947429364446e-06, + "loss": 0.1224, + "step": 12399 + }, + { + "epoch": 5.19, + "learning_rate": 5.793204785264461e-06, + "loss": 0.1102, + "step": 12400 + }, + { + "epoch": 5.19, + "learning_rate": 5.791666755004915e-06, + "loss": 0.1489, + "step": 12401 + }, + { + "epoch": 5.19, + "learning_rate": 5.790128845710175e-06, + "loss": 0.1471, + "step": 12402 + }, + { + "epoch": 5.19, + "learning_rate": 5.788591057424443e-06, + "loss": 0.123, + "step": 12403 + }, + { + "epoch": 5.19, + "learning_rate": 5.787053390191925e-06, + "loss": 0.0856, + "step": 12404 + }, + { + "epoch": 5.19, + "learning_rate": 5.785515844056818e-06, + "loss": 0.1065, + "step": 12405 + }, + { + "epoch": 5.19, + "learning_rate": 5.783978419063323e-06, + "loss": 0.1078, + "step": 12406 + }, + { + "epoch": 5.19, + "learning_rate": 5.782441115255624e-06, + "loss": 0.0962, + "step": 12407 + }, + { + "epoch": 5.19, + "learning_rate": 5.78090393267791e-06, + "loss": 0.13, + "step": 12408 + }, + { + "epoch": 5.2, + "learning_rate": 5.77936687137437e-06, + "loss": 0.134, + "step": 12409 + }, + { + "epoch": 5.2, + "learning_rate": 5.777829931389185e-06, + "loss": 0.125, + "step": 12410 + }, + { + "epoch": 5.2, + "learning_rate": 5.776293112766535e-06, + "loss": 0.1014, + "step": 12411 + }, + { + "epoch": 5.2, + "learning_rate": 5.774756415550587e-06, + "loss": 0.1036, + "step": 12412 + }, + { + "epoch": 5.2, + "learning_rate": 5.773219839785517e-06, + "loss": 0.1309, + "step": 12413 + }, + { + "epoch": 5.2, + "learning_rate": 5.771683385515495e-06, + "loss": 0.1354, + "step": 12414 + }, + { + "epoch": 5.2, + "learning_rate": 5.770147052784678e-06, + "loss": 0.1121, + "step": 12415 + }, + { + "epoch": 5.2, + "learning_rate": 5.768610841637232e-06, + "loss": 0.1305, + "step": 12416 + }, + { + "epoch": 5.2, + "learning_rate": 5.767074752117311e-06, + "loss": 0.1492, + "step": 12417 + }, + { + "epoch": 5.2, + "learning_rate": 5.765538784269071e-06, + "loss": 0.0997, + "step": 12418 + }, + { + "epoch": 5.2, + "learning_rate": 5.76400293813666e-06, + "loss": 0.0921, + "step": 12419 + }, + { + "epoch": 5.2, + "learning_rate": 5.762467213764231e-06, + "loss": 0.1206, + "step": 12420 + }, + { + "epoch": 5.2, + "learning_rate": 5.760931611195917e-06, + "loss": 0.1416, + "step": 12421 + }, + { + "epoch": 5.2, + "learning_rate": 5.75939613047586e-06, + "loss": 0.1282, + "step": 12422 + }, + { + "epoch": 5.2, + "learning_rate": 5.757860771648202e-06, + "loss": 0.1128, + "step": 12423 + }, + { + "epoch": 5.2, + "learning_rate": 5.75632553475707e-06, + "loss": 0.1247, + "step": 12424 + }, + { + "epoch": 5.2, + "learning_rate": 5.754790419846595e-06, + "loss": 0.1211, + "step": 12425 + }, + { + "epoch": 5.2, + "learning_rate": 5.753255426960901e-06, + "loss": 0.1361, + "step": 12426 + }, + { + "epoch": 5.2, + "learning_rate": 5.751720556144116e-06, + "loss": 0.1106, + "step": 12427 + }, + { + "epoch": 5.2, + "learning_rate": 5.750185807440349e-06, + "loss": 0.1182, + "step": 12428 + }, + { + "epoch": 5.2, + "learning_rate": 5.748651180893721e-06, + "loss": 0.1056, + "step": 12429 + }, + { + "epoch": 5.2, + "learning_rate": 5.747116676548342e-06, + "loss": 0.0918, + "step": 12430 + }, + { + "epoch": 5.2, + "learning_rate": 5.745582294448319e-06, + "loss": 0.1437, + "step": 12431 + }, + { + "epoch": 5.2, + "learning_rate": 5.744048034637759e-06, + "loss": 0.15, + "step": 12432 + }, + { + "epoch": 5.21, + "learning_rate": 5.742513897160766e-06, + "loss": 0.1019, + "step": 12433 + }, + { + "epoch": 5.21, + "learning_rate": 5.740979882061433e-06, + "loss": 0.0812, + "step": 12434 + }, + { + "epoch": 5.21, + "learning_rate": 5.7394459893838486e-06, + "loss": 0.0771, + "step": 12435 + }, + { + "epoch": 5.21, + "learning_rate": 5.737912219172108e-06, + "loss": 0.115, + "step": 12436 + }, + { + "epoch": 5.21, + "learning_rate": 5.7363785714703e-06, + "loss": 0.1307, + "step": 12437 + }, + { + "epoch": 5.21, + "learning_rate": 5.734845046322506e-06, + "loss": 0.1003, + "step": 12438 + }, + { + "epoch": 5.21, + "learning_rate": 5.733311643772807e-06, + "loss": 0.1297, + "step": 12439 + }, + { + "epoch": 5.21, + "learning_rate": 5.731778363865278e-06, + "loss": 0.1407, + "step": 12440 + }, + { + "epoch": 5.21, + "learning_rate": 5.730245206643996e-06, + "loss": 0.1088, + "step": 12441 + }, + { + "epoch": 5.21, + "learning_rate": 5.728712172153024e-06, + "loss": 0.1324, + "step": 12442 + }, + { + "epoch": 5.21, + "learning_rate": 5.727179260436428e-06, + "loss": 0.1405, + "step": 12443 + }, + { + "epoch": 5.21, + "learning_rate": 5.7256464715382754e-06, + "loss": 0.1619, + "step": 12444 + }, + { + "epoch": 5.21, + "learning_rate": 5.72411380550262e-06, + "loss": 0.088, + "step": 12445 + }, + { + "epoch": 5.21, + "learning_rate": 5.722581262373519e-06, + "loss": 0.1305, + "step": 12446 + }, + { + "epoch": 5.21, + "learning_rate": 5.721048842195028e-06, + "loss": 0.1233, + "step": 12447 + }, + { + "epoch": 5.21, + "learning_rate": 5.719516545011188e-06, + "loss": 0.127, + "step": 12448 + }, + { + "epoch": 5.21, + "learning_rate": 5.7179843708660454e-06, + "loss": 0.1287, + "step": 12449 + }, + { + "epoch": 5.21, + "learning_rate": 5.716452319803643e-06, + "loss": 0.1295, + "step": 12450 + }, + { + "epoch": 5.21, + "learning_rate": 5.7149203918680166e-06, + "loss": 0.1007, + "step": 12451 + }, + { + "epoch": 5.21, + "learning_rate": 5.713388587103201e-06, + "loss": 0.1025, + "step": 12452 + }, + { + "epoch": 5.21, + "learning_rate": 5.711856905553228e-06, + "loss": 0.1114, + "step": 12453 + }, + { + "epoch": 5.21, + "learning_rate": 5.710325347262125e-06, + "loss": 0.1141, + "step": 12454 + }, + { + "epoch": 5.21, + "learning_rate": 5.708793912273911e-06, + "loss": 0.1332, + "step": 12455 + }, + { + "epoch": 5.21, + "learning_rate": 5.707262600632612e-06, + "loss": 0.1265, + "step": 12456 + }, + { + "epoch": 5.22, + "learning_rate": 5.705731412382235e-06, + "loss": 0.1596, + "step": 12457 + }, + { + "epoch": 5.22, + "learning_rate": 5.704200347566798e-06, + "loss": 0.1179, + "step": 12458 + }, + { + "epoch": 5.22, + "learning_rate": 5.7026694062303115e-06, + "loss": 0.0919, + "step": 12459 + }, + { + "epoch": 5.22, + "learning_rate": 5.701138588416778e-06, + "loss": 0.1254, + "step": 12460 + }, + { + "epoch": 5.22, + "learning_rate": 5.699607894170206e-06, + "loss": 0.1102, + "step": 12461 + }, + { + "epoch": 5.22, + "learning_rate": 5.698077323534584e-06, + "loss": 0.1143, + "step": 12462 + }, + { + "epoch": 5.22, + "learning_rate": 5.696546876553912e-06, + "loss": 0.1093, + "step": 12463 + }, + { + "epoch": 5.22, + "learning_rate": 5.695016553272182e-06, + "loss": 0.1028, + "step": 12464 + }, + { + "epoch": 5.22, + "learning_rate": 5.693486353733382e-06, + "loss": 0.1587, + "step": 12465 + }, + { + "epoch": 5.22, + "learning_rate": 5.691956277981493e-06, + "loss": 0.1528, + "step": 12466 + }, + { + "epoch": 5.22, + "learning_rate": 5.690426326060498e-06, + "loss": 0.1301, + "step": 12467 + }, + { + "epoch": 5.22, + "learning_rate": 5.688896498014381e-06, + "loss": 0.1587, + "step": 12468 + }, + { + "epoch": 5.22, + "learning_rate": 5.687366793887101e-06, + "loss": 0.1473, + "step": 12469 + }, + { + "epoch": 5.22, + "learning_rate": 5.685837213722638e-06, + "loss": 0.1648, + "step": 12470 + }, + { + "epoch": 5.22, + "learning_rate": 5.684307757564954e-06, + "loss": 0.116, + "step": 12471 + }, + { + "epoch": 5.22, + "learning_rate": 5.682778425458015e-06, + "loss": 0.0865, + "step": 12472 + }, + { + "epoch": 5.22, + "learning_rate": 5.681249217445779e-06, + "loss": 0.1476, + "step": 12473 + }, + { + "epoch": 5.22, + "learning_rate": 5.6797201335722064e-06, + "loss": 0.142, + "step": 12474 + }, + { + "epoch": 5.22, + "learning_rate": 5.67819117388124e-06, + "loss": 0.1112, + "step": 12475 + }, + { + "epoch": 5.22, + "learning_rate": 5.676662338416833e-06, + "loss": 0.1331, + "step": 12476 + }, + { + "epoch": 5.22, + "learning_rate": 5.675133627222931e-06, + "loss": 0.1341, + "step": 12477 + }, + { + "epoch": 5.22, + "learning_rate": 5.673605040343475e-06, + "loss": 0.1504, + "step": 12478 + }, + { + "epoch": 5.22, + "learning_rate": 5.6720765778224075e-06, + "loss": 0.1181, + "step": 12479 + }, + { + "epoch": 5.23, + "learning_rate": 5.670548239703653e-06, + "loss": 0.1158, + "step": 12480 + }, + { + "epoch": 5.23, + "learning_rate": 5.669020026031148e-06, + "loss": 0.1225, + "step": 12481 + }, + { + "epoch": 5.23, + "learning_rate": 5.6674919368488235e-06, + "loss": 0.1307, + "step": 12482 + }, + { + "epoch": 5.23, + "learning_rate": 5.665963972200594e-06, + "loss": 0.121, + "step": 12483 + }, + { + "epoch": 5.23, + "learning_rate": 5.664436132130384e-06, + "loss": 0.1369, + "step": 12484 + }, + { + "epoch": 5.23, + "learning_rate": 5.662908416682111e-06, + "loss": 0.1592, + "step": 12485 + }, + { + "epoch": 5.23, + "learning_rate": 5.661380825899687e-06, + "loss": 0.1245, + "step": 12486 + }, + { + "epoch": 5.23, + "learning_rate": 5.6598533598270215e-06, + "loss": 0.0663, + "step": 12487 + }, + { + "epoch": 5.23, + "learning_rate": 5.658326018508022e-06, + "loss": 0.0951, + "step": 12488 + }, + { + "epoch": 5.23, + "learning_rate": 5.656798801986585e-06, + "loss": 0.1009, + "step": 12489 + }, + { + "epoch": 5.23, + "learning_rate": 5.655271710306613e-06, + "loss": 0.128, + "step": 12490 + }, + { + "epoch": 5.23, + "learning_rate": 5.653744743512e-06, + "loss": 0.1113, + "step": 12491 + }, + { + "epoch": 5.23, + "learning_rate": 5.652217901646637e-06, + "loss": 0.1389, + "step": 12492 + }, + { + "epoch": 5.23, + "learning_rate": 5.6506911847544134e-06, + "loss": 0.1356, + "step": 12493 + }, + { + "epoch": 5.23, + "learning_rate": 5.649164592879213e-06, + "loss": 0.1161, + "step": 12494 + }, + { + "epoch": 5.23, + "learning_rate": 5.647638126064917e-06, + "loss": 0.1361, + "step": 12495 + }, + { + "epoch": 5.23, + "learning_rate": 5.646111784355398e-06, + "loss": 0.112, + "step": 12496 + }, + { + "epoch": 5.23, + "learning_rate": 5.644585567794531e-06, + "loss": 0.1376, + "step": 12497 + }, + { + "epoch": 5.23, + "learning_rate": 5.643059476426189e-06, + "loss": 0.1104, + "step": 12498 + }, + { + "epoch": 5.23, + "learning_rate": 5.641533510294235e-06, + "loss": 0.115, + "step": 12499 + }, + { + "epoch": 5.23, + "learning_rate": 5.640007669442533e-06, + "loss": 0.0994, + "step": 12500 + }, + { + "epoch": 5.23, + "learning_rate": 5.638481953914945e-06, + "loss": 0.15, + "step": 12501 + }, + { + "epoch": 5.23, + "learning_rate": 5.636956363755324e-06, + "loss": 0.1126, + "step": 12502 + }, + { + "epoch": 5.23, + "learning_rate": 5.635430899007515e-06, + "loss": 0.1103, + "step": 12503 + }, + { + "epoch": 5.24, + "learning_rate": 5.6339055597153734e-06, + "loss": 0.1124, + "step": 12504 + }, + { + "epoch": 5.24, + "learning_rate": 5.632380345922741e-06, + "loss": 0.1223, + "step": 12505 + }, + { + "epoch": 5.24, + "learning_rate": 5.63085525767346e-06, + "loss": 0.1191, + "step": 12506 + }, + { + "epoch": 5.24, + "learning_rate": 5.629330295011367e-06, + "loss": 0.1154, + "step": 12507 + }, + { + "epoch": 5.24, + "learning_rate": 5.627805457980298e-06, + "loss": 0.1415, + "step": 12508 + }, + { + "epoch": 5.24, + "learning_rate": 5.626280746624083e-06, + "loss": 0.1086, + "step": 12509 + }, + { + "epoch": 5.24, + "learning_rate": 5.6247561609865445e-06, + "loss": 0.0688, + "step": 12510 + }, + { + "epoch": 5.24, + "learning_rate": 5.623231701111507e-06, + "loss": 0.1371, + "step": 12511 + }, + { + "epoch": 5.24, + "learning_rate": 5.6217073670427905e-06, + "loss": 0.1134, + "step": 12512 + }, + { + "epoch": 5.24, + "learning_rate": 5.62018315882421e-06, + "loss": 0.1037, + "step": 12513 + }, + { + "epoch": 5.24, + "learning_rate": 5.618659076499578e-06, + "loss": 0.126, + "step": 12514 + }, + { + "epoch": 5.24, + "learning_rate": 5.6171351201127065e-06, + "loss": 0.0937, + "step": 12515 + }, + { + "epoch": 5.24, + "learning_rate": 5.615611289707394e-06, + "loss": 0.1005, + "step": 12516 + }, + { + "epoch": 5.24, + "learning_rate": 5.614087585327444e-06, + "loss": 0.0986, + "step": 12517 + }, + { + "epoch": 5.24, + "learning_rate": 5.612564007016654e-06, + "loss": 0.1368, + "step": 12518 + }, + { + "epoch": 5.24, + "learning_rate": 5.611040554818819e-06, + "loss": 0.1007, + "step": 12519 + }, + { + "epoch": 5.24, + "learning_rate": 5.6095172287777275e-06, + "loss": 0.081, + "step": 12520 + }, + { + "epoch": 5.24, + "learning_rate": 5.607994028937171e-06, + "loss": 0.0976, + "step": 12521 + }, + { + "epoch": 5.24, + "learning_rate": 5.606470955340926e-06, + "loss": 0.1193, + "step": 12522 + }, + { + "epoch": 5.24, + "learning_rate": 5.604948008032774e-06, + "loss": 0.1216, + "step": 12523 + }, + { + "epoch": 5.24, + "learning_rate": 5.603425187056496e-06, + "loss": 0.0846, + "step": 12524 + }, + { + "epoch": 5.24, + "learning_rate": 5.601902492455855e-06, + "loss": 0.0941, + "step": 12525 + }, + { + "epoch": 5.24, + "learning_rate": 5.6003799242746235e-06, + "loss": 0.0941, + "step": 12526 + }, + { + "epoch": 5.24, + "learning_rate": 5.598857482556568e-06, + "loss": 0.107, + "step": 12527 + }, + { + "epoch": 5.25, + "learning_rate": 5.597335167345448e-06, + "loss": 0.1391, + "step": 12528 + }, + { + "epoch": 5.25, + "learning_rate": 5.595812978685027e-06, + "loss": 0.1228, + "step": 12529 + }, + { + "epoch": 5.25, + "learning_rate": 5.594290916619048e-06, + "loss": 0.1307, + "step": 12530 + }, + { + "epoch": 5.25, + "learning_rate": 5.592768981191268e-06, + "loss": 0.1124, + "step": 12531 + }, + { + "epoch": 5.25, + "learning_rate": 5.591247172445432e-06, + "loss": 0.1211, + "step": 12532 + }, + { + "epoch": 5.25, + "learning_rate": 5.589725490425284e-06, + "loss": 0.1426, + "step": 12533 + }, + { + "epoch": 5.25, + "learning_rate": 5.588203935174563e-06, + "loss": 0.1508, + "step": 12534 + }, + { + "epoch": 5.25, + "learning_rate": 5.586682506737007e-06, + "loss": 0.1433, + "step": 12535 + }, + { + "epoch": 5.25, + "learning_rate": 5.585161205156347e-06, + "loss": 0.0949, + "step": 12536 + }, + { + "epoch": 5.25, + "learning_rate": 5.583640030476309e-06, + "loss": 0.1226, + "step": 12537 + }, + { + "epoch": 5.25, + "learning_rate": 5.582118982740618e-06, + "loss": 0.1161, + "step": 12538 + }, + { + "epoch": 5.25, + "learning_rate": 5.580598061992997e-06, + "loss": 0.1382, + "step": 12539 + }, + { + "epoch": 5.25, + "learning_rate": 5.5790772682771625e-06, + "loss": 0.1094, + "step": 12540 + }, + { + "epoch": 5.25, + "learning_rate": 5.577556601636828e-06, + "loss": 0.1252, + "step": 12541 + }, + { + "epoch": 5.25, + "learning_rate": 5.576036062115709e-06, + "loss": 0.1765, + "step": 12542 + }, + { + "epoch": 5.25, + "learning_rate": 5.574515649757505e-06, + "loss": 0.1138, + "step": 12543 + }, + { + "epoch": 5.25, + "learning_rate": 5.572995364605921e-06, + "loss": 0.1526, + "step": 12544 + }, + { + "epoch": 5.25, + "learning_rate": 5.571475206704656e-06, + "loss": 0.1438, + "step": 12545 + }, + { + "epoch": 5.25, + "learning_rate": 5.569955176097408e-06, + "loss": 0.1175, + "step": 12546 + }, + { + "epoch": 5.25, + "learning_rate": 5.5684352728278705e-06, + "loss": 0.1257, + "step": 12547 + }, + { + "epoch": 5.25, + "learning_rate": 5.566915496939723e-06, + "loss": 0.1317, + "step": 12548 + }, + { + "epoch": 5.25, + "learning_rate": 5.565395848476658e-06, + "loss": 0.1251, + "step": 12549 + }, + { + "epoch": 5.25, + "learning_rate": 5.563876327482356e-06, + "loss": 0.1561, + "step": 12550 + }, + { + "epoch": 5.25, + "learning_rate": 5.56235693400049e-06, + "loss": 0.1404, + "step": 12551 + }, + { + "epoch": 5.26, + "learning_rate": 5.5608376680747354e-06, + "loss": 0.144, + "step": 12552 + }, + { + "epoch": 5.26, + "learning_rate": 5.559318529748763e-06, + "loss": 0.1175, + "step": 12553 + }, + { + "epoch": 5.26, + "learning_rate": 5.55779951906624e-06, + "loss": 0.1148, + "step": 12554 + }, + { + "epoch": 5.26, + "learning_rate": 5.556280636070826e-06, + "loss": 0.1124, + "step": 12555 + }, + { + "epoch": 5.26, + "learning_rate": 5.554761880806187e-06, + "loss": 0.102, + "step": 12556 + }, + { + "epoch": 5.26, + "learning_rate": 5.5532432533159676e-06, + "loss": 0.1266, + "step": 12557 + }, + { + "epoch": 5.26, + "learning_rate": 5.551724753643826e-06, + "loss": 0.1217, + "step": 12558 + }, + { + "epoch": 5.26, + "learning_rate": 5.550206381833409e-06, + "loss": 0.1142, + "step": 12559 + }, + { + "epoch": 5.26, + "learning_rate": 5.548688137928362e-06, + "loss": 0.0905, + "step": 12560 + }, + { + "epoch": 5.26, + "learning_rate": 5.547170021972323e-06, + "loss": 0.1191, + "step": 12561 + }, + { + "epoch": 5.26, + "learning_rate": 5.545652034008935e-06, + "loss": 0.1319, + "step": 12562 + }, + { + "epoch": 5.26, + "learning_rate": 5.544134174081821e-06, + "loss": 0.1422, + "step": 12563 + }, + { + "epoch": 5.26, + "learning_rate": 5.542616442234618e-06, + "loss": 0.0982, + "step": 12564 + }, + { + "epoch": 5.26, + "learning_rate": 5.54109883851095e-06, + "loss": 0.1629, + "step": 12565 + }, + { + "epoch": 5.26, + "learning_rate": 5.539581362954438e-06, + "loss": 0.1433, + "step": 12566 + }, + { + "epoch": 5.26, + "learning_rate": 5.5380640156087016e-06, + "loss": 0.0967, + "step": 12567 + }, + { + "epoch": 5.26, + "learning_rate": 5.536546796517357e-06, + "loss": 0.1558, + "step": 12568 + }, + { + "epoch": 5.26, + "learning_rate": 5.535029705724017e-06, + "loss": 0.0975, + "step": 12569 + }, + { + "epoch": 5.26, + "learning_rate": 5.533512743272285e-06, + "loss": 0.1051, + "step": 12570 + }, + { + "epoch": 5.26, + "learning_rate": 5.531995909205763e-06, + "loss": 0.1519, + "step": 12571 + }, + { + "epoch": 5.26, + "learning_rate": 5.530479203568053e-06, + "loss": 0.1319, + "step": 12572 + }, + { + "epoch": 5.26, + "learning_rate": 5.528962626402754e-06, + "loss": 0.1028, + "step": 12573 + }, + { + "epoch": 5.26, + "learning_rate": 5.527446177753455e-06, + "loss": 0.1478, + "step": 12574 + }, + { + "epoch": 5.26, + "learning_rate": 5.525929857663747e-06, + "loss": 0.1575, + "step": 12575 + }, + { + "epoch": 5.27, + "learning_rate": 5.524413666177216e-06, + "loss": 0.116, + "step": 12576 + }, + { + "epoch": 5.27, + "learning_rate": 5.522897603337445e-06, + "loss": 0.1317, + "step": 12577 + }, + { + "epoch": 5.27, + "learning_rate": 5.521381669188006e-06, + "loss": 0.1256, + "step": 12578 + }, + { + "epoch": 5.27, + "learning_rate": 5.519865863772477e-06, + "loss": 0.1148, + "step": 12579 + }, + { + "epoch": 5.27, + "learning_rate": 5.518350187134427e-06, + "loss": 0.1114, + "step": 12580 + }, + { + "epoch": 5.27, + "learning_rate": 5.516834639317426e-06, + "loss": 0.1411, + "step": 12581 + }, + { + "epoch": 5.27, + "learning_rate": 5.515319220365034e-06, + "loss": 0.1377, + "step": 12582 + }, + { + "epoch": 5.27, + "learning_rate": 5.513803930320815e-06, + "loss": 0.079, + "step": 12583 + }, + { + "epoch": 5.27, + "learning_rate": 5.512288769228318e-06, + "loss": 0.1564, + "step": 12584 + }, + { + "epoch": 5.27, + "learning_rate": 5.510773737131097e-06, + "loss": 0.1106, + "step": 12585 + }, + { + "epoch": 5.27, + "learning_rate": 5.509258834072702e-06, + "loss": 0.1236, + "step": 12586 + }, + { + "epoch": 5.27, + "learning_rate": 5.507744060096676e-06, + "loss": 0.1165, + "step": 12587 + }, + { + "epoch": 5.27, + "learning_rate": 5.50622941524656e-06, + "loss": 0.1437, + "step": 12588 + }, + { + "epoch": 5.27, + "learning_rate": 5.5047148995658976e-06, + "loss": 0.1063, + "step": 12589 + }, + { + "epoch": 5.27, + "learning_rate": 5.5032005130982115e-06, + "loss": 0.1413, + "step": 12590 + }, + { + "epoch": 5.27, + "learning_rate": 5.501686255887036e-06, + "loss": 0.1175, + "step": 12591 + }, + { + "epoch": 5.27, + "learning_rate": 5.500172127975903e-06, + "loss": 0.1253, + "step": 12592 + }, + { + "epoch": 5.27, + "learning_rate": 5.498658129408323e-06, + "loss": 0.1528, + "step": 12593 + }, + { + "epoch": 5.27, + "learning_rate": 5.49714426022782e-06, + "loss": 0.104, + "step": 12594 + }, + { + "epoch": 5.27, + "learning_rate": 5.495630520477912e-06, + "loss": 0.1079, + "step": 12595 + }, + { + "epoch": 5.27, + "learning_rate": 5.4941169102021065e-06, + "loss": 0.1383, + "step": 12596 + }, + { + "epoch": 5.27, + "learning_rate": 5.492603429443916e-06, + "loss": 0.1617, + "step": 12597 + }, + { + "epoch": 5.27, + "learning_rate": 5.491090078246836e-06, + "loss": 0.1331, + "step": 12598 + }, + { + "epoch": 5.27, + "learning_rate": 5.489576856654371e-06, + "loss": 0.161, + "step": 12599 + }, + { + "epoch": 5.28, + "learning_rate": 5.488063764710016e-06, + "loss": 0.1442, + "step": 12600 + }, + { + "epoch": 5.28, + "learning_rate": 5.486550802457264e-06, + "loss": 0.1255, + "step": 12601 + }, + { + "epoch": 5.28, + "learning_rate": 5.485037969939605e-06, + "loss": 0.1291, + "step": 12602 + }, + { + "epoch": 5.28, + "learning_rate": 5.483525267200527e-06, + "loss": 0.1409, + "step": 12603 + }, + { + "epoch": 5.28, + "learning_rate": 5.482012694283503e-06, + "loss": 0.1257, + "step": 12604 + }, + { + "epoch": 5.28, + "learning_rate": 5.480500251232015e-06, + "loss": 0.1324, + "step": 12605 + }, + { + "epoch": 5.28, + "learning_rate": 5.478987938089537e-06, + "loss": 0.1447, + "step": 12606 + }, + { + "epoch": 5.28, + "learning_rate": 5.477475754899538e-06, + "loss": 0.1482, + "step": 12607 + }, + { + "epoch": 5.28, + "learning_rate": 5.475963701705487e-06, + "loss": 0.1701, + "step": 12608 + }, + { + "epoch": 5.28, + "learning_rate": 5.4744517785508425e-06, + "loss": 0.1152, + "step": 12609 + }, + { + "epoch": 5.28, + "learning_rate": 5.472939985479071e-06, + "loss": 0.1722, + "step": 12610 + }, + { + "epoch": 5.28, + "learning_rate": 5.471428322533619e-06, + "loss": 0.124, + "step": 12611 + }, + { + "epoch": 5.28, + "learning_rate": 5.469916789757941e-06, + "loss": 0.1133, + "step": 12612 + }, + { + "epoch": 5.28, + "learning_rate": 5.4684053871954835e-06, + "loss": 0.0509, + "step": 12613 + }, + { + "epoch": 5.28, + "learning_rate": 5.466894114889697e-06, + "loss": 0.126, + "step": 12614 + }, + { + "epoch": 5.28, + "learning_rate": 5.4653829728840126e-06, + "loss": 0.117, + "step": 12615 + }, + { + "epoch": 5.28, + "learning_rate": 5.463871961221871e-06, + "loss": 0.115, + "step": 12616 + }, + { + "epoch": 5.28, + "learning_rate": 5.462361079946708e-06, + "loss": 0.1245, + "step": 12617 + }, + { + "epoch": 5.28, + "learning_rate": 5.460850329101944e-06, + "loss": 0.1079, + "step": 12618 + }, + { + "epoch": 5.28, + "learning_rate": 5.459339708731011e-06, + "loss": 0.0993, + "step": 12619 + }, + { + "epoch": 5.28, + "learning_rate": 5.457829218877328e-06, + "loss": 0.1243, + "step": 12620 + }, + { + "epoch": 5.28, + "learning_rate": 5.456318859584314e-06, + "loss": 0.1076, + "step": 12621 + }, + { + "epoch": 5.28, + "learning_rate": 5.454808630895383e-06, + "loss": 0.1082, + "step": 12622 + }, + { + "epoch": 5.28, + "learning_rate": 5.453298532853943e-06, + "loss": 0.1531, + "step": 12623 + }, + { + "epoch": 5.29, + "learning_rate": 5.451788565503406e-06, + "loss": 0.1146, + "step": 12624 + }, + { + "epoch": 5.29, + "learning_rate": 5.450278728887168e-06, + "loss": 0.1133, + "step": 12625 + }, + { + "epoch": 5.29, + "learning_rate": 5.44876902304863e-06, + "loss": 0.1498, + "step": 12626 + }, + { + "epoch": 5.29, + "learning_rate": 5.447259448031188e-06, + "loss": 0.1062, + "step": 12627 + }, + { + "epoch": 5.29, + "learning_rate": 5.445750003878233e-06, + "loss": 0.1089, + "step": 12628 + }, + { + "epoch": 5.29, + "learning_rate": 5.444240690633152e-06, + "loss": 0.1413, + "step": 12629 + }, + { + "epoch": 5.29, + "learning_rate": 5.442731508339334e-06, + "loss": 0.1212, + "step": 12630 + }, + { + "epoch": 5.29, + "learning_rate": 5.441222457040152e-06, + "loss": 0.1449, + "step": 12631 + }, + { + "epoch": 5.29, + "learning_rate": 5.439713536778983e-06, + "loss": 0.1023, + "step": 12632 + }, + { + "epoch": 5.29, + "learning_rate": 5.438204747599204e-06, + "loss": 0.135, + "step": 12633 + }, + { + "epoch": 5.29, + "learning_rate": 5.43669608954418e-06, + "loss": 0.1051, + "step": 12634 + }, + { + "epoch": 5.29, + "learning_rate": 5.435187562657278e-06, + "loss": 0.1177, + "step": 12635 + }, + { + "epoch": 5.29, + "learning_rate": 5.433679166981858e-06, + "loss": 0.1278, + "step": 12636 + }, + { + "epoch": 5.29, + "learning_rate": 5.432170902561282e-06, + "loss": 0.088, + "step": 12637 + }, + { + "epoch": 5.29, + "learning_rate": 5.4306627694389014e-06, + "loss": 0.1628, + "step": 12638 + }, + { + "epoch": 5.29, + "learning_rate": 5.42915476765806e-06, + "loss": 0.0765, + "step": 12639 + }, + { + "epoch": 5.29, + "learning_rate": 5.427646897262109e-06, + "loss": 0.1318, + "step": 12640 + }, + { + "epoch": 5.29, + "learning_rate": 5.42613915829439e-06, + "loss": 0.097, + "step": 12641 + }, + { + "epoch": 5.29, + "learning_rate": 5.424631550798243e-06, + "loss": 0.1231, + "step": 12642 + }, + { + "epoch": 5.29, + "learning_rate": 5.423124074817001e-06, + "loss": 0.1377, + "step": 12643 + }, + { + "epoch": 5.29, + "learning_rate": 5.421616730394e-06, + "loss": 0.1024, + "step": 12644 + }, + { + "epoch": 5.29, + "learning_rate": 5.42010951757256e-06, + "loss": 0.1169, + "step": 12645 + }, + { + "epoch": 5.29, + "learning_rate": 5.418602436396007e-06, + "loss": 0.1132, + "step": 12646 + }, + { + "epoch": 5.29, + "learning_rate": 5.417095486907663e-06, + "loss": 0.1527, + "step": 12647 + }, + { + "epoch": 5.3, + "learning_rate": 5.415588669150842e-06, + "loss": 0.1227, + "step": 12648 + }, + { + "epoch": 5.3, + "learning_rate": 5.414081983168855e-06, + "loss": 0.131, + "step": 12649 + }, + { + "epoch": 5.3, + "learning_rate": 5.412575429005014e-06, + "loss": 0.1329, + "step": 12650 + }, + { + "epoch": 5.3, + "learning_rate": 5.411069006702624e-06, + "loss": 0.1036, + "step": 12651 + }, + { + "epoch": 5.3, + "learning_rate": 5.409562716304979e-06, + "loss": 0.1419, + "step": 12652 + }, + { + "epoch": 5.3, + "learning_rate": 5.408056557855381e-06, + "loss": 0.125, + "step": 12653 + }, + { + "epoch": 5.3, + "learning_rate": 5.406550531397123e-06, + "loss": 0.1224, + "step": 12654 + }, + { + "epoch": 5.3, + "learning_rate": 5.405044636973493e-06, + "loss": 0.1051, + "step": 12655 + }, + { + "epoch": 5.3, + "learning_rate": 5.403538874627777e-06, + "loss": 0.131, + "step": 12656 + }, + { + "epoch": 5.3, + "learning_rate": 5.402033244403262e-06, + "loss": 0.0957, + "step": 12657 + }, + { + "epoch": 5.3, + "learning_rate": 5.400527746343217e-06, + "loss": 0.1377, + "step": 12658 + }, + { + "epoch": 5.3, + "learning_rate": 5.399022380490921e-06, + "loss": 0.086, + "step": 12659 + }, + { + "epoch": 5.3, + "learning_rate": 5.397517146889647e-06, + "loss": 0.0981, + "step": 12660 + }, + { + "epoch": 5.3, + "learning_rate": 5.3960120455826545e-06, + "loss": 0.1452, + "step": 12661 + }, + { + "epoch": 5.3, + "learning_rate": 5.394507076613212e-06, + "loss": 0.1332, + "step": 12662 + }, + { + "epoch": 5.3, + "learning_rate": 5.3930022400245745e-06, + "loss": 0.1038, + "step": 12663 + }, + { + "epoch": 5.3, + "learning_rate": 5.391497535859999e-06, + "loss": 0.1076, + "step": 12664 + }, + { + "epoch": 5.3, + "learning_rate": 5.3899929641627425e-06, + "loss": 0.1232, + "step": 12665 + }, + { + "epoch": 5.3, + "learning_rate": 5.388488524976043e-06, + "loss": 0.1222, + "step": 12666 + }, + { + "epoch": 5.3, + "learning_rate": 5.3869842183431495e-06, + "loss": 0.1345, + "step": 12667 + }, + { + "epoch": 5.3, + "learning_rate": 5.3854800443073006e-06, + "loss": 0.0914, + "step": 12668 + }, + { + "epoch": 5.3, + "learning_rate": 5.383976002911734e-06, + "loss": 0.152, + "step": 12669 + }, + { + "epoch": 5.3, + "learning_rate": 5.38247209419968e-06, + "loss": 0.1065, + "step": 12670 + }, + { + "epoch": 5.31, + "learning_rate": 5.3809683182143725e-06, + "loss": 0.1336, + "step": 12671 + }, + { + "epoch": 5.31, + "learning_rate": 5.3794646749990265e-06, + "loss": 0.1259, + "step": 12672 + }, + { + "epoch": 5.31, + "learning_rate": 5.3779611645968696e-06, + "loss": 0.1357, + "step": 12673 + }, + { + "epoch": 5.31, + "learning_rate": 5.376457787051117e-06, + "loss": 0.1132, + "step": 12674 + }, + { + "epoch": 5.31, + "learning_rate": 5.374954542404982e-06, + "loss": 0.1397, + "step": 12675 + }, + { + "epoch": 5.31, + "learning_rate": 5.373451430701676e-06, + "loss": 0.0993, + "step": 12676 + }, + { + "epoch": 5.31, + "learning_rate": 5.371948451984401e-06, + "loss": 0.0798, + "step": 12677 + }, + { + "epoch": 5.31, + "learning_rate": 5.3704456062963674e-06, + "loss": 0.1446, + "step": 12678 + }, + { + "epoch": 5.31, + "learning_rate": 5.368942893680761e-06, + "loss": 0.1033, + "step": 12679 + }, + { + "epoch": 5.31, + "learning_rate": 5.367440314180782e-06, + "loss": 0.125, + "step": 12680 + }, + { + "epoch": 5.31, + "learning_rate": 5.36593786783962e-06, + "loss": 0.1587, + "step": 12681 + }, + { + "epoch": 5.31, + "learning_rate": 5.3644355547004664e-06, + "loss": 0.1135, + "step": 12682 + }, + { + "epoch": 5.31, + "learning_rate": 5.362933374806494e-06, + "loss": 0.1085, + "step": 12683 + }, + { + "epoch": 5.31, + "learning_rate": 5.361431328200888e-06, + "loss": 0.1236, + "step": 12684 + }, + { + "epoch": 5.31, + "learning_rate": 5.3599294149268255e-06, + "loss": 0.1151, + "step": 12685 + }, + { + "epoch": 5.31, + "learning_rate": 5.3584276350274725e-06, + "loss": 0.1447, + "step": 12686 + }, + { + "epoch": 5.31, + "learning_rate": 5.356925988545996e-06, + "loss": 0.1179, + "step": 12687 + }, + { + "epoch": 5.31, + "learning_rate": 5.355424475525562e-06, + "loss": 0.1462, + "step": 12688 + }, + { + "epoch": 5.31, + "learning_rate": 5.3539230960093305e-06, + "loss": 0.1122, + "step": 12689 + }, + { + "epoch": 5.31, + "learning_rate": 5.352421850040457e-06, + "loss": 0.1016, + "step": 12690 + }, + { + "epoch": 5.31, + "learning_rate": 5.3509207376620935e-06, + "loss": 0.1547, + "step": 12691 + }, + { + "epoch": 5.31, + "learning_rate": 5.349419758917392e-06, + "loss": 0.1373, + "step": 12692 + }, + { + "epoch": 5.31, + "learning_rate": 5.347918913849487e-06, + "loss": 0.1006, + "step": 12693 + }, + { + "epoch": 5.31, + "learning_rate": 5.346418202501526e-06, + "loss": 0.1653, + "step": 12694 + }, + { + "epoch": 5.32, + "learning_rate": 5.344917624916644e-06, + "loss": 0.1047, + "step": 12695 + }, + { + "epoch": 5.32, + "learning_rate": 5.343417181137973e-06, + "loss": 0.1224, + "step": 12696 + }, + { + "epoch": 5.32, + "learning_rate": 5.341916871208644e-06, + "loss": 0.1222, + "step": 12697 + }, + { + "epoch": 5.32, + "learning_rate": 5.340416695171785e-06, + "loss": 0.1651, + "step": 12698 + }, + { + "epoch": 5.32, + "learning_rate": 5.3389166530705096e-06, + "loss": 0.1085, + "step": 12699 + }, + { + "epoch": 5.32, + "learning_rate": 5.337416744947939e-06, + "loss": 0.1256, + "step": 12700 + }, + { + "epoch": 5.32, + "learning_rate": 5.335916970847185e-06, + "loss": 0.0904, + "step": 12701 + }, + { + "epoch": 5.32, + "learning_rate": 5.3344173308113586e-06, + "loss": 0.1208, + "step": 12702 + }, + { + "epoch": 5.32, + "learning_rate": 5.332917824883568e-06, + "loss": 0.0958, + "step": 12703 + }, + { + "epoch": 5.32, + "learning_rate": 5.331418453106916e-06, + "loss": 0.1328, + "step": 12704 + }, + { + "epoch": 5.32, + "learning_rate": 5.329919215524494e-06, + "loss": 0.1195, + "step": 12705 + }, + { + "epoch": 5.32, + "learning_rate": 5.328420112179405e-06, + "loss": 0.094, + "step": 12706 + }, + { + "epoch": 5.32, + "learning_rate": 5.326921143114727e-06, + "loss": 0.1097, + "step": 12707 + }, + { + "epoch": 5.32, + "learning_rate": 5.325422308373557e-06, + "loss": 0.1508, + "step": 12708 + }, + { + "epoch": 5.32, + "learning_rate": 5.3239236079989735e-06, + "loss": 0.0881, + "step": 12709 + }, + { + "epoch": 5.32, + "learning_rate": 5.322425042034056e-06, + "loss": 0.1582, + "step": 12710 + }, + { + "epoch": 5.32, + "learning_rate": 5.3209266105218815e-06, + "loss": 0.1305, + "step": 12711 + }, + { + "epoch": 5.32, + "learning_rate": 5.319428313505523e-06, + "loss": 0.1466, + "step": 12712 + }, + { + "epoch": 5.32, + "learning_rate": 5.317930151028039e-06, + "loss": 0.1536, + "step": 12713 + }, + { + "epoch": 5.32, + "learning_rate": 5.316432123132498e-06, + "loss": 0.1155, + "step": 12714 + }, + { + "epoch": 5.32, + "learning_rate": 5.314934229861961e-06, + "loss": 0.1389, + "step": 12715 + }, + { + "epoch": 5.32, + "learning_rate": 5.3134364712594814e-06, + "loss": 0.1402, + "step": 12716 + }, + { + "epoch": 5.32, + "learning_rate": 5.311938847368111e-06, + "loss": 0.1609, + "step": 12717 + }, + { + "epoch": 5.32, + "learning_rate": 5.310441358230899e-06, + "loss": 0.1318, + "step": 12718 + }, + { + "epoch": 5.33, + "learning_rate": 5.308944003890892e-06, + "loss": 0.1152, + "step": 12719 + }, + { + "epoch": 5.33, + "learning_rate": 5.307446784391124e-06, + "loss": 0.1604, + "step": 12720 + }, + { + "epoch": 5.33, + "learning_rate": 5.305949699774633e-06, + "loss": 0.1279, + "step": 12721 + }, + { + "epoch": 5.33, + "learning_rate": 5.304452750084452e-06, + "loss": 0.1333, + "step": 12722 + }, + { + "epoch": 5.33, + "learning_rate": 5.3029559353636116e-06, + "loss": 0.1252, + "step": 12723 + }, + { + "epoch": 5.33, + "learning_rate": 5.301459255655134e-06, + "loss": 0.1178, + "step": 12724 + }, + { + "epoch": 5.33, + "learning_rate": 5.299962711002045e-06, + "loss": 0.1461, + "step": 12725 + }, + { + "epoch": 5.33, + "learning_rate": 5.298466301447353e-06, + "loss": 0.1093, + "step": 12726 + }, + { + "epoch": 5.33, + "learning_rate": 5.296970027034074e-06, + "loss": 0.1464, + "step": 12727 + }, + { + "epoch": 5.33, + "learning_rate": 5.295473887805223e-06, + "loss": 0.1338, + "step": 12728 + }, + { + "epoch": 5.33, + "learning_rate": 5.293977883803797e-06, + "loss": 0.1077, + "step": 12729 + }, + { + "epoch": 5.33, + "learning_rate": 5.292482015072799e-06, + "loss": 0.1061, + "step": 12730 + }, + { + "epoch": 5.33, + "learning_rate": 5.29098628165523e-06, + "loss": 0.1552, + "step": 12731 + }, + { + "epoch": 5.33, + "learning_rate": 5.289490683594081e-06, + "loss": 0.1318, + "step": 12732 + }, + { + "epoch": 5.33, + "learning_rate": 5.287995220932346e-06, + "loss": 0.1018, + "step": 12733 + }, + { + "epoch": 5.33, + "learning_rate": 5.286499893713003e-06, + "loss": 0.131, + "step": 12734 + }, + { + "epoch": 5.33, + "learning_rate": 5.285004701979037e-06, + "loss": 0.092, + "step": 12735 + }, + { + "epoch": 5.33, + "learning_rate": 5.283509645773429e-06, + "loss": 0.1369, + "step": 12736 + }, + { + "epoch": 5.33, + "learning_rate": 5.282014725139149e-06, + "loss": 0.1574, + "step": 12737 + }, + { + "epoch": 5.33, + "learning_rate": 5.280519940119171e-06, + "loss": 0.1436, + "step": 12738 + }, + { + "epoch": 5.33, + "learning_rate": 5.279025290756461e-06, + "loss": 0.0928, + "step": 12739 + }, + { + "epoch": 5.33, + "learning_rate": 5.2775307770939775e-06, + "loss": 0.1305, + "step": 12740 + }, + { + "epoch": 5.33, + "learning_rate": 5.27603639917468e-06, + "loss": 0.11, + "step": 12741 + }, + { + "epoch": 5.33, + "learning_rate": 5.2745421570415245e-06, + "loss": 0.1133, + "step": 12742 + }, + { + "epoch": 5.34, + "learning_rate": 5.273048050737461e-06, + "loss": 0.0938, + "step": 12743 + }, + { + "epoch": 5.34, + "learning_rate": 5.271554080305439e-06, + "loss": 0.1041, + "step": 12744 + }, + { + "epoch": 5.34, + "learning_rate": 5.270060245788401e-06, + "loss": 0.1204, + "step": 12745 + }, + { + "epoch": 5.34, + "learning_rate": 5.2685665472292805e-06, + "loss": 0.1563, + "step": 12746 + }, + { + "epoch": 5.34, + "learning_rate": 5.2670729846710155e-06, + "loss": 0.1355, + "step": 12747 + }, + { + "epoch": 5.34, + "learning_rate": 5.265579558156537e-06, + "loss": 0.1159, + "step": 12748 + }, + { + "epoch": 5.34, + "learning_rate": 5.264086267728774e-06, + "loss": 0.1063, + "step": 12749 + }, + { + "epoch": 5.34, + "learning_rate": 5.262593113430653e-06, + "loss": 0.1926, + "step": 12750 + }, + { + "epoch": 5.34, + "learning_rate": 5.261100095305085e-06, + "loss": 0.1129, + "step": 12751 + }, + { + "epoch": 5.34, + "learning_rate": 5.259607213394989e-06, + "loss": 0.1329, + "step": 12752 + }, + { + "epoch": 5.34, + "learning_rate": 5.2581144677432805e-06, + "loss": 0.1095, + "step": 12753 + }, + { + "epoch": 5.34, + "learning_rate": 5.256621858392861e-06, + "loss": 0.0983, + "step": 12754 + }, + { + "epoch": 5.34, + "learning_rate": 5.255129385386635e-06, + "loss": 0.1131, + "step": 12755 + }, + { + "epoch": 5.34, + "learning_rate": 5.2536370487675056e-06, + "loss": 0.1182, + "step": 12756 + }, + { + "epoch": 5.34, + "learning_rate": 5.252144848578368e-06, + "loss": 0.101, + "step": 12757 + }, + { + "epoch": 5.34, + "learning_rate": 5.250652784862114e-06, + "loss": 0.1241, + "step": 12758 + }, + { + "epoch": 5.34, + "learning_rate": 5.249160857661628e-06, + "loss": 0.1303, + "step": 12759 + }, + { + "epoch": 5.34, + "learning_rate": 5.247669067019805e-06, + "loss": 0.0913, + "step": 12760 + }, + { + "epoch": 5.34, + "learning_rate": 5.246177412979512e-06, + "loss": 0.0975, + "step": 12761 + }, + { + "epoch": 5.34, + "learning_rate": 5.2446858955836295e-06, + "loss": 0.0928, + "step": 12762 + }, + { + "epoch": 5.34, + "learning_rate": 5.243194514875033e-06, + "loss": 0.1463, + "step": 12763 + }, + { + "epoch": 5.34, + "learning_rate": 5.241703270896589e-06, + "loss": 0.1404, + "step": 12764 + }, + { + "epoch": 5.34, + "learning_rate": 5.240212163691161e-06, + "loss": 0.103, + "step": 12765 + }, + { + "epoch": 5.34, + "learning_rate": 5.2387211933016145e-06, + "loss": 0.1103, + "step": 12766 + }, + { + "epoch": 5.35, + "learning_rate": 5.237230359770799e-06, + "loss": 0.1328, + "step": 12767 + }, + { + "epoch": 5.35, + "learning_rate": 5.235739663141571e-06, + "loss": 0.1178, + "step": 12768 + }, + { + "epoch": 5.35, + "learning_rate": 5.2342491034567775e-06, + "loss": 0.124, + "step": 12769 + }, + { + "epoch": 5.35, + "learning_rate": 5.232758680759264e-06, + "loss": 0.0963, + "step": 12770 + }, + { + "epoch": 5.35, + "learning_rate": 5.231268395091874e-06, + "loss": 0.1234, + "step": 12771 + }, + { + "epoch": 5.35, + "learning_rate": 5.229778246497446e-06, + "loss": 0.117, + "step": 12772 + }, + { + "epoch": 5.35, + "learning_rate": 5.2282882350188035e-06, + "loss": 0.1322, + "step": 12773 + }, + { + "epoch": 5.35, + "learning_rate": 5.226798360698786e-06, + "loss": 0.1263, + "step": 12774 + }, + { + "epoch": 5.35, + "learning_rate": 5.225308623580209e-06, + "loss": 0.1048, + "step": 12775 + }, + { + "epoch": 5.35, + "learning_rate": 5.2238190237059e-06, + "loss": 0.1328, + "step": 12776 + }, + { + "epoch": 5.35, + "learning_rate": 5.222329561118675e-06, + "loss": 0.1141, + "step": 12777 + }, + { + "epoch": 5.35, + "learning_rate": 5.2208402358613465e-06, + "loss": 0.1137, + "step": 12778 + }, + { + "epoch": 5.35, + "learning_rate": 5.219351047976725e-06, + "loss": 0.1288, + "step": 12779 + }, + { + "epoch": 5.35, + "learning_rate": 5.217861997507618e-06, + "loss": 0.1239, + "step": 12780 + }, + { + "epoch": 5.35, + "learning_rate": 5.216373084496822e-06, + "loss": 0.1355, + "step": 12781 + }, + { + "epoch": 5.35, + "learning_rate": 5.214884308987136e-06, + "loss": 0.1093, + "step": 12782 + }, + { + "epoch": 5.35, + "learning_rate": 5.213395671021355e-06, + "loss": 0.1434, + "step": 12783 + }, + { + "epoch": 5.35, + "learning_rate": 5.211907170642267e-06, + "loss": 0.1132, + "step": 12784 + }, + { + "epoch": 5.35, + "learning_rate": 5.210418807892659e-06, + "loss": 0.1423, + "step": 12785 + }, + { + "epoch": 5.35, + "learning_rate": 5.208930582815316e-06, + "loss": 0.1334, + "step": 12786 + }, + { + "epoch": 5.35, + "learning_rate": 5.207442495453008e-06, + "loss": 0.1015, + "step": 12787 + }, + { + "epoch": 5.35, + "learning_rate": 5.205954545848513e-06, + "loss": 0.1216, + "step": 12788 + }, + { + "epoch": 5.35, + "learning_rate": 5.2044667340445985e-06, + "loss": 0.1396, + "step": 12789 + }, + { + "epoch": 5.35, + "learning_rate": 5.202979060084035e-06, + "loss": 0.1208, + "step": 12790 + }, + { + "epoch": 5.36, + "learning_rate": 5.2014915240095785e-06, + "loss": 0.1006, + "step": 12791 + }, + { + "epoch": 5.36, + "learning_rate": 5.200004125863993e-06, + "loss": 0.0823, + "step": 12792 + }, + { + "epoch": 5.36, + "learning_rate": 5.198516865690031e-06, + "loss": 0.1439, + "step": 12793 + }, + { + "epoch": 5.36, + "learning_rate": 5.1970297435304395e-06, + "loss": 0.0832, + "step": 12794 + }, + { + "epoch": 5.36, + "learning_rate": 5.195542759427964e-06, + "loss": 0.1474, + "step": 12795 + }, + { + "epoch": 5.36, + "learning_rate": 5.194055913425353e-06, + "loss": 0.1134, + "step": 12796 + }, + { + "epoch": 5.36, + "learning_rate": 5.192569205565336e-06, + "loss": 0.1257, + "step": 12797 + }, + { + "epoch": 5.36, + "learning_rate": 5.191082635890652e-06, + "loss": 0.0989, + "step": 12798 + }, + { + "epoch": 5.36, + "learning_rate": 5.1895962044440295e-06, + "loss": 0.0935, + "step": 12799 + }, + { + "epoch": 5.36, + "learning_rate": 5.188109911268196e-06, + "loss": 0.1132, + "step": 12800 + }, + { + "epoch": 5.36, + "learning_rate": 5.186623756405875e-06, + "loss": 0.1229, + "step": 12801 + }, + { + "epoch": 5.36, + "learning_rate": 5.185137739899781e-06, + "loss": 0.1153, + "step": 12802 + }, + { + "epoch": 5.36, + "learning_rate": 5.183651861792629e-06, + "loss": 0.125, + "step": 12803 + }, + { + "epoch": 5.36, + "learning_rate": 5.18216612212713e-06, + "loss": 0.1001, + "step": 12804 + }, + { + "epoch": 5.36, + "learning_rate": 5.180680520945992e-06, + "loss": 0.1196, + "step": 12805 + }, + { + "epoch": 5.36, + "learning_rate": 5.179195058291915e-06, + "loss": 0.0949, + "step": 12806 + }, + { + "epoch": 5.36, + "learning_rate": 5.177709734207603e-06, + "loss": 0.1265, + "step": 12807 + }, + { + "epoch": 5.36, + "learning_rate": 5.17622454873574e-06, + "loss": 0.1384, + "step": 12808 + }, + { + "epoch": 5.36, + "learning_rate": 5.174739501919023e-06, + "loss": 0.1639, + "step": 12809 + }, + { + "epoch": 5.36, + "learning_rate": 5.173254593800137e-06, + "loss": 0.1167, + "step": 12810 + }, + { + "epoch": 5.36, + "learning_rate": 5.171769824421764e-06, + "loss": 0.1537, + "step": 12811 + }, + { + "epoch": 5.36, + "learning_rate": 5.170285193826583e-06, + "loss": 0.1342, + "step": 12812 + }, + { + "epoch": 5.36, + "learning_rate": 5.168800702057272e-06, + "loss": 0.1059, + "step": 12813 + }, + { + "epoch": 5.36, + "learning_rate": 5.167316349156495e-06, + "loss": 0.0931, + "step": 12814 + }, + { + "epoch": 5.37, + "learning_rate": 5.16583213516692e-06, + "loss": 0.1204, + "step": 12815 + }, + { + "epoch": 5.37, + "learning_rate": 5.16434806013121e-06, + "loss": 0.1225, + "step": 12816 + }, + { + "epoch": 5.37, + "learning_rate": 5.162864124092024e-06, + "loss": 0.0935, + "step": 12817 + }, + { + "epoch": 5.37, + "learning_rate": 5.161380327092021e-06, + "loss": 0.1368, + "step": 12818 + }, + { + "epoch": 5.37, + "learning_rate": 5.1598966691738425e-06, + "loss": 0.0978, + "step": 12819 + }, + { + "epoch": 5.37, + "learning_rate": 5.158413150380139e-06, + "loss": 0.1169, + "step": 12820 + }, + { + "epoch": 5.37, + "learning_rate": 5.1569297707535585e-06, + "loss": 0.1467, + "step": 12821 + }, + { + "epoch": 5.37, + "learning_rate": 5.155446530336729e-06, + "loss": 0.132, + "step": 12822 + }, + { + "epoch": 5.37, + "learning_rate": 5.15396342917229e-06, + "loss": 0.128, + "step": 12823 + }, + { + "epoch": 5.37, + "learning_rate": 5.1524804673028725e-06, + "loss": 0.1187, + "step": 12824 + }, + { + "epoch": 5.37, + "learning_rate": 5.150997644771102e-06, + "loss": 0.1133, + "step": 12825 + }, + { + "epoch": 5.37, + "learning_rate": 5.149514961619602e-06, + "loss": 0.1066, + "step": 12826 + }, + { + "epoch": 5.37, + "learning_rate": 5.148032417890992e-06, + "loss": 0.1473, + "step": 12827 + }, + { + "epoch": 5.37, + "learning_rate": 5.1465500136278825e-06, + "loss": 0.1187, + "step": 12828 + }, + { + "epoch": 5.37, + "learning_rate": 5.145067748872885e-06, + "loss": 0.1354, + "step": 12829 + }, + { + "epoch": 5.37, + "learning_rate": 5.143585623668607e-06, + "loss": 0.129, + "step": 12830 + }, + { + "epoch": 5.37, + "learning_rate": 5.142103638057652e-06, + "loss": 0.14, + "step": 12831 + }, + { + "epoch": 5.37, + "learning_rate": 5.140621792082616e-06, + "loss": 0.1078, + "step": 12832 + }, + { + "epoch": 5.37, + "learning_rate": 5.139140085786095e-06, + "loss": 0.1107, + "step": 12833 + }, + { + "epoch": 5.37, + "learning_rate": 5.1376585192106824e-06, + "loss": 0.1486, + "step": 12834 + }, + { + "epoch": 5.37, + "learning_rate": 5.136177092398957e-06, + "loss": 0.1297, + "step": 12835 + }, + { + "epoch": 5.37, + "learning_rate": 5.1346958053935045e-06, + "loss": 0.1109, + "step": 12836 + }, + { + "epoch": 5.37, + "learning_rate": 5.133214658236905e-06, + "loss": 0.0966, + "step": 12837 + }, + { + "epoch": 5.37, + "learning_rate": 5.131733650971731e-06, + "loss": 0.1232, + "step": 12838 + }, + { + "epoch": 5.38, + "learning_rate": 5.130252783640553e-06, + "loss": 0.1504, + "step": 12839 + }, + { + "epoch": 5.38, + "learning_rate": 5.12877205628594e-06, + "loss": 0.0922, + "step": 12840 + }, + { + "epoch": 5.38, + "learning_rate": 5.127291468950453e-06, + "loss": 0.1228, + "step": 12841 + }, + { + "epoch": 5.38, + "learning_rate": 5.125811021676643e-06, + "loss": 0.1426, + "step": 12842 + }, + { + "epoch": 5.38, + "learning_rate": 5.124330714507073e-06, + "loss": 0.0989, + "step": 12843 + }, + { + "epoch": 5.38, + "learning_rate": 5.1228505474842885e-06, + "loss": 0.1034, + "step": 12844 + }, + { + "epoch": 5.38, + "learning_rate": 5.121370520650837e-06, + "loss": 0.1275, + "step": 12845 + }, + { + "epoch": 5.38, + "learning_rate": 5.1198906340492605e-06, + "loss": 0.1123, + "step": 12846 + }, + { + "epoch": 5.38, + "learning_rate": 5.118410887722096e-06, + "loss": 0.11, + "step": 12847 + }, + { + "epoch": 5.38, + "learning_rate": 5.116931281711886e-06, + "loss": 0.143, + "step": 12848 + }, + { + "epoch": 5.38, + "learning_rate": 5.115451816061146e-06, + "loss": 0.1328, + "step": 12849 + }, + { + "epoch": 5.38, + "learning_rate": 5.113972490812411e-06, + "loss": 0.1434, + "step": 12850 + }, + { + "epoch": 5.38, + "learning_rate": 5.112493306008199e-06, + "loss": 0.1443, + "step": 12851 + }, + { + "epoch": 5.38, + "learning_rate": 5.111014261691032e-06, + "loss": 0.088, + "step": 12852 + }, + { + "epoch": 5.38, + "learning_rate": 5.109535357903421e-06, + "loss": 0.1044, + "step": 12853 + }, + { + "epoch": 5.38, + "learning_rate": 5.10805659468788e-06, + "loss": 0.1146, + "step": 12854 + }, + { + "epoch": 5.38, + "learning_rate": 5.106577972086907e-06, + "loss": 0.1258, + "step": 12855 + }, + { + "epoch": 5.38, + "learning_rate": 5.105099490143008e-06, + "loss": 0.1459, + "step": 12856 + }, + { + "epoch": 5.38, + "learning_rate": 5.1036211488986806e-06, + "loss": 0.1133, + "step": 12857 + }, + { + "epoch": 5.38, + "learning_rate": 5.102142948396418e-06, + "loss": 0.1259, + "step": 12858 + }, + { + "epoch": 5.38, + "learning_rate": 5.10066488867871e-06, + "loss": 0.0901, + "step": 12859 + }, + { + "epoch": 5.38, + "learning_rate": 5.099186969788044e-06, + "loss": 0.12, + "step": 12860 + }, + { + "epoch": 5.38, + "learning_rate": 5.097709191766902e-06, + "loss": 0.1217, + "step": 12861 + }, + { + "epoch": 5.38, + "learning_rate": 5.096231554657755e-06, + "loss": 0.1196, + "step": 12862 + }, + { + "epoch": 5.39, + "learning_rate": 5.094754058503081e-06, + "loss": 0.1184, + "step": 12863 + }, + { + "epoch": 5.39, + "learning_rate": 5.093276703345351e-06, + "loss": 0.1329, + "step": 12864 + }, + { + "epoch": 5.39, + "learning_rate": 5.091799489227027e-06, + "loss": 0.112, + "step": 12865 + }, + { + "epoch": 5.39, + "learning_rate": 5.09032241619057e-06, + "loss": 0.13, + "step": 12866 + }, + { + "epoch": 5.39, + "learning_rate": 5.088845484278437e-06, + "loss": 0.1355, + "step": 12867 + }, + { + "epoch": 5.39, + "learning_rate": 5.087368693533088e-06, + "loss": 0.1024, + "step": 12868 + }, + { + "epoch": 5.39, + "learning_rate": 5.085892043996962e-06, + "loss": 0.1276, + "step": 12869 + }, + { + "epoch": 5.39, + "learning_rate": 5.084415535712508e-06, + "loss": 0.0943, + "step": 12870 + }, + { + "epoch": 5.39, + "learning_rate": 5.082939168722168e-06, + "loss": 0.1246, + "step": 12871 + }, + { + "epoch": 5.39, + "learning_rate": 5.081462943068378e-06, + "loss": 0.1012, + "step": 12872 + }, + { + "epoch": 5.39, + "learning_rate": 5.079986858793572e-06, + "loss": 0.1261, + "step": 12873 + }, + { + "epoch": 5.39, + "learning_rate": 5.078510915940178e-06, + "loss": 0.1254, + "step": 12874 + }, + { + "epoch": 5.39, + "learning_rate": 5.077035114550624e-06, + "loss": 0.1195, + "step": 12875 + }, + { + "epoch": 5.39, + "learning_rate": 5.075559454667321e-06, + "loss": 0.1738, + "step": 12876 + }, + { + "epoch": 5.39, + "learning_rate": 5.074083936332695e-06, + "loss": 0.132, + "step": 12877 + }, + { + "epoch": 5.39, + "learning_rate": 5.072608559589152e-06, + "loss": 0.1303, + "step": 12878 + }, + { + "epoch": 5.39, + "learning_rate": 5.071133324479105e-06, + "loss": 0.1372, + "step": 12879 + }, + { + "epoch": 5.39, + "learning_rate": 5.069658231044956e-06, + "loss": 0.1013, + "step": 12880 + }, + { + "epoch": 5.39, + "learning_rate": 5.0681832793291105e-06, + "loss": 0.0985, + "step": 12881 + }, + { + "epoch": 5.39, + "learning_rate": 5.066708469373958e-06, + "loss": 0.0946, + "step": 12882 + }, + { + "epoch": 5.39, + "learning_rate": 5.065233801221891e-06, + "loss": 0.1635, + "step": 12883 + }, + { + "epoch": 5.39, + "learning_rate": 5.063759274915299e-06, + "loss": 0.1048, + "step": 12884 + }, + { + "epoch": 5.39, + "learning_rate": 5.062284890496567e-06, + "loss": 0.1156, + "step": 12885 + }, + { + "epoch": 5.4, + "learning_rate": 5.060810648008078e-06, + "loss": 0.0879, + "step": 12886 + }, + { + "epoch": 5.4, + "learning_rate": 5.0593365474921995e-06, + "loss": 0.0964, + "step": 12887 + }, + { + "epoch": 5.4, + "learning_rate": 5.057862588991308e-06, + "loss": 0.1302, + "step": 12888 + }, + { + "epoch": 5.4, + "learning_rate": 5.056388772547775e-06, + "loss": 0.1282, + "step": 12889 + }, + { + "epoch": 5.4, + "learning_rate": 5.054915098203956e-06, + "loss": 0.086, + "step": 12890 + }, + { + "epoch": 5.4, + "learning_rate": 5.053441566002214e-06, + "loss": 0.1227, + "step": 12891 + }, + { + "epoch": 5.4, + "learning_rate": 5.051968175984905e-06, + "loss": 0.1448, + "step": 12892 + }, + { + "epoch": 5.4, + "learning_rate": 5.050494928194379e-06, + "loss": 0.1208, + "step": 12893 + }, + { + "epoch": 5.4, + "learning_rate": 5.0490218226729855e-06, + "loss": 0.1248, + "step": 12894 + }, + { + "epoch": 5.4, + "learning_rate": 5.047548859463069e-06, + "loss": 0.1448, + "step": 12895 + }, + { + "epoch": 5.4, + "learning_rate": 5.046076038606964e-06, + "loss": 0.1254, + "step": 12896 + }, + { + "epoch": 5.4, + "learning_rate": 5.0446033601470045e-06, + "loss": 0.1221, + "step": 12897 + }, + { + "epoch": 5.4, + "learning_rate": 5.043130824125526e-06, + "loss": 0.1134, + "step": 12898 + }, + { + "epoch": 5.4, + "learning_rate": 5.041658430584852e-06, + "loss": 0.1405, + "step": 12899 + }, + { + "epoch": 5.4, + "learning_rate": 5.040186179567308e-06, + "loss": 0.1081, + "step": 12900 + }, + { + "epoch": 5.4, + "learning_rate": 5.03871407111521e-06, + "loss": 0.137, + "step": 12901 + }, + { + "epoch": 5.4, + "learning_rate": 5.037242105270877e-06, + "loss": 0.151, + "step": 12902 + }, + { + "epoch": 5.4, + "learning_rate": 5.035770282076613e-06, + "loss": 0.1329, + "step": 12903 + }, + { + "epoch": 5.4, + "learning_rate": 5.034298601574726e-06, + "loss": 0.117, + "step": 12904 + }, + { + "epoch": 5.4, + "learning_rate": 5.03282706380752e-06, + "loss": 0.091, + "step": 12905 + }, + { + "epoch": 5.4, + "learning_rate": 5.031355668817292e-06, + "loss": 0.1177, + "step": 12906 + }, + { + "epoch": 5.4, + "learning_rate": 5.029884416646335e-06, + "loss": 0.1754, + "step": 12907 + }, + { + "epoch": 5.4, + "learning_rate": 5.0284133073369456e-06, + "loss": 0.122, + "step": 12908 + }, + { + "epoch": 5.4, + "learning_rate": 5.026942340931404e-06, + "loss": 0.1711, + "step": 12909 + }, + { + "epoch": 5.41, + "learning_rate": 5.025471517471987e-06, + "loss": 0.1217, + "step": 12910 + }, + { + "epoch": 5.41, + "learning_rate": 5.0240008370009775e-06, + "loss": 0.1192, + "step": 12911 + }, + { + "epoch": 5.41, + "learning_rate": 5.022530299560648e-06, + "loss": 0.1085, + "step": 12912 + }, + { + "epoch": 5.41, + "learning_rate": 5.021059905193268e-06, + "loss": 0.1179, + "step": 12913 + }, + { + "epoch": 5.41, + "learning_rate": 5.019589653941103e-06, + "loss": 0.1086, + "step": 12914 + }, + { + "epoch": 5.41, + "learning_rate": 5.018119545846414e-06, + "loss": 0.1005, + "step": 12915 + }, + { + "epoch": 5.41, + "learning_rate": 5.016649580951462e-06, + "loss": 0.0988, + "step": 12916 + }, + { + "epoch": 5.41, + "learning_rate": 5.015179759298492e-06, + "loss": 0.135, + "step": 12917 + }, + { + "epoch": 5.41, + "learning_rate": 5.013710080929754e-06, + "loss": 0.1036, + "step": 12918 + }, + { + "epoch": 5.41, + "learning_rate": 5.0122405458874975e-06, + "loss": 0.1485, + "step": 12919 + }, + { + "epoch": 5.41, + "learning_rate": 5.01077115421396e-06, + "loss": 0.0948, + "step": 12920 + }, + { + "epoch": 5.41, + "learning_rate": 5.009301905951378e-06, + "loss": 0.0954, + "step": 12921 + }, + { + "epoch": 5.41, + "learning_rate": 5.007832801141988e-06, + "loss": 0.1289, + "step": 12922 + }, + { + "epoch": 5.41, + "learning_rate": 5.00636383982801e-06, + "loss": 0.1359, + "step": 12923 + }, + { + "epoch": 5.41, + "learning_rate": 5.0048950220516725e-06, + "loss": 0.1199, + "step": 12924 + }, + { + "epoch": 5.41, + "learning_rate": 5.0034263478551935e-06, + "loss": 0.1447, + "step": 12925 + }, + { + "epoch": 5.41, + "learning_rate": 5.001957817280792e-06, + "loss": 0.1321, + "step": 12926 + }, + { + "epoch": 5.41, + "learning_rate": 5.0004894303706775e-06, + "loss": 0.116, + "step": 12927 + }, + { + "epoch": 5.41, + "learning_rate": 4.999021187167061e-06, + "loss": 0.085, + "step": 12928 + }, + { + "epoch": 5.41, + "learning_rate": 4.99755308771214e-06, + "loss": 0.1256, + "step": 12929 + }, + { + "epoch": 5.41, + "learning_rate": 4.996085132048116e-06, + "loss": 0.0945, + "step": 12930 + }, + { + "epoch": 5.41, + "learning_rate": 4.994617320217184e-06, + "loss": 0.1253, + "step": 12931 + }, + { + "epoch": 5.41, + "learning_rate": 4.99314965226154e-06, + "loss": 0.1201, + "step": 12932 + }, + { + "epoch": 5.41, + "learning_rate": 4.991682128223362e-06, + "loss": 0.1346, + "step": 12933 + }, + { + "epoch": 5.42, + "learning_rate": 4.990214748144837e-06, + "loss": 0.1149, + "step": 12934 + }, + { + "epoch": 5.42, + "learning_rate": 4.988747512068143e-06, + "loss": 0.1218, + "step": 12935 + }, + { + "epoch": 5.42, + "learning_rate": 4.9872804200354596e-06, + "loss": 0.0837, + "step": 12936 + }, + { + "epoch": 5.42, + "learning_rate": 4.985813472088947e-06, + "loss": 0.1447, + "step": 12937 + }, + { + "epoch": 5.42, + "learning_rate": 4.984346668270779e-06, + "loss": 0.0828, + "step": 12938 + }, + { + "epoch": 5.42, + "learning_rate": 4.9828800086231135e-06, + "loss": 0.1073, + "step": 12939 + }, + { + "epoch": 5.42, + "learning_rate": 4.98141349318811e-06, + "loss": 0.1124, + "step": 12940 + }, + { + "epoch": 5.42, + "learning_rate": 4.979947122007922e-06, + "loss": 0.115, + "step": 12941 + }, + { + "epoch": 5.42, + "learning_rate": 4.9784808951247e-06, + "loss": 0.0887, + "step": 12942 + }, + { + "epoch": 5.42, + "learning_rate": 4.977014812580593e-06, + "loss": 0.0724, + "step": 12943 + }, + { + "epoch": 5.42, + "learning_rate": 4.975548874417733e-06, + "loss": 0.126, + "step": 12944 + }, + { + "epoch": 5.42, + "learning_rate": 4.9740830806782625e-06, + "loss": 0.1083, + "step": 12945 + }, + { + "epoch": 5.42, + "learning_rate": 4.972617431404313e-06, + "loss": 0.0955, + "step": 12946 + }, + { + "epoch": 5.42, + "learning_rate": 4.971151926638016e-06, + "loss": 0.1092, + "step": 12947 + }, + { + "epoch": 5.42, + "learning_rate": 4.9696865664214936e-06, + "loss": 0.1395, + "step": 12948 + }, + { + "epoch": 5.42, + "learning_rate": 4.96822135079687e-06, + "loss": 0.1324, + "step": 12949 + }, + { + "epoch": 5.42, + "learning_rate": 4.966756279806255e-06, + "loss": 0.0848, + "step": 12950 + }, + { + "epoch": 5.42, + "learning_rate": 4.9652913534917645e-06, + "loss": 0.1021, + "step": 12951 + }, + { + "epoch": 5.42, + "learning_rate": 4.963826571895507e-06, + "loss": 0.11, + "step": 12952 + }, + { + "epoch": 5.42, + "learning_rate": 4.962361935059585e-06, + "loss": 0.1254, + "step": 12953 + }, + { + "epoch": 5.42, + "learning_rate": 4.9608974430261025e-06, + "loss": 0.1299, + "step": 12954 + }, + { + "epoch": 5.42, + "learning_rate": 4.959433095837149e-06, + "loss": 0.1142, + "step": 12955 + }, + { + "epoch": 5.42, + "learning_rate": 4.957968893534816e-06, + "loss": 0.1018, + "step": 12956 + }, + { + "epoch": 5.42, + "learning_rate": 4.956504836161198e-06, + "loss": 0.1073, + "step": 12957 + }, + { + "epoch": 5.43, + "learning_rate": 4.955040923758368e-06, + "loss": 0.1269, + "step": 12958 + }, + { + "epoch": 5.43, + "learning_rate": 4.953577156368412e-06, + "loss": 0.1611, + "step": 12959 + }, + { + "epoch": 5.43, + "learning_rate": 4.952113534033402e-06, + "loss": 0.1238, + "step": 12960 + }, + { + "epoch": 5.43, + "learning_rate": 4.950650056795407e-06, + "loss": 0.122, + "step": 12961 + }, + { + "epoch": 5.43, + "learning_rate": 4.949186724696498e-06, + "loss": 0.1164, + "step": 12962 + }, + { + "epoch": 5.43, + "learning_rate": 4.947723537778737e-06, + "loss": 0.1309, + "step": 12963 + }, + { + "epoch": 5.43, + "learning_rate": 4.946260496084175e-06, + "loss": 0.135, + "step": 12964 + }, + { + "epoch": 5.43, + "learning_rate": 4.9447975996548725e-06, + "loss": 0.1494, + "step": 12965 + }, + { + "epoch": 5.43, + "learning_rate": 4.943334848532876e-06, + "loss": 0.1116, + "step": 12966 + }, + { + "epoch": 5.43, + "learning_rate": 4.9418722427602315e-06, + "loss": 0.126, + "step": 12967 + }, + { + "epoch": 5.43, + "learning_rate": 4.940409782378982e-06, + "loss": 0.1028, + "step": 12968 + }, + { + "epoch": 5.43, + "learning_rate": 4.938947467431167e-06, + "loss": 0.1234, + "step": 12969 + }, + { + "epoch": 5.43, + "learning_rate": 4.937485297958812e-06, + "loss": 0.0987, + "step": 12970 + }, + { + "epoch": 5.43, + "learning_rate": 4.936023274003949e-06, + "loss": 0.1385, + "step": 12971 + }, + { + "epoch": 5.43, + "learning_rate": 4.934561395608603e-06, + "loss": 0.1048, + "step": 12972 + }, + { + "epoch": 5.43, + "learning_rate": 4.933099662814797e-06, + "loss": 0.1374, + "step": 12973 + }, + { + "epoch": 5.43, + "learning_rate": 4.931638075664544e-06, + "loss": 0.1106, + "step": 12974 + }, + { + "epoch": 5.43, + "learning_rate": 4.930176634199856e-06, + "loss": 0.166, + "step": 12975 + }, + { + "epoch": 5.43, + "learning_rate": 4.928715338462746e-06, + "loss": 0.1257, + "step": 12976 + }, + { + "epoch": 5.43, + "learning_rate": 4.927254188495213e-06, + "loss": 0.1514, + "step": 12977 + }, + { + "epoch": 5.43, + "learning_rate": 4.925793184339254e-06, + "loss": 0.1367, + "step": 12978 + }, + { + "epoch": 5.43, + "learning_rate": 4.924332326036867e-06, + "loss": 0.1215, + "step": 12979 + }, + { + "epoch": 5.43, + "learning_rate": 4.922871613630043e-06, + "loss": 0.0959, + "step": 12980 + }, + { + "epoch": 5.43, + "learning_rate": 4.921411047160768e-06, + "loss": 0.1108, + "step": 12981 + }, + { + "epoch": 5.44, + "learning_rate": 4.919950626671028e-06, + "loss": 0.0922, + "step": 12982 + }, + { + "epoch": 5.44, + "learning_rate": 4.918490352202797e-06, + "loss": 0.1245, + "step": 12983 + }, + { + "epoch": 5.44, + "learning_rate": 4.917030223798057e-06, + "loss": 0.1117, + "step": 12984 + }, + { + "epoch": 5.44, + "learning_rate": 4.9155702414987685e-06, + "loss": 0.098, + "step": 12985 + }, + { + "epoch": 5.44, + "learning_rate": 4.914110405346901e-06, + "loss": 0.0783, + "step": 12986 + }, + { + "epoch": 5.44, + "learning_rate": 4.912650715384417e-06, + "loss": 0.1356, + "step": 12987 + }, + { + "epoch": 5.44, + "learning_rate": 4.911191171653273e-06, + "loss": 0.0962, + "step": 12988 + }, + { + "epoch": 5.44, + "learning_rate": 4.909731774195425e-06, + "loss": 0.088, + "step": 12989 + }, + { + "epoch": 5.44, + "learning_rate": 4.908272523052823e-06, + "loss": 0.1381, + "step": 12990 + }, + { + "epoch": 5.44, + "learning_rate": 4.9068134182674044e-06, + "loss": 0.0844, + "step": 12991 + }, + { + "epoch": 5.44, + "learning_rate": 4.905354459881116e-06, + "loss": 0.1077, + "step": 12992 + }, + { + "epoch": 5.44, + "learning_rate": 4.9038956479358915e-06, + "loss": 0.1185, + "step": 12993 + }, + { + "epoch": 5.44, + "learning_rate": 4.9024369824736654e-06, + "loss": 0.1349, + "step": 12994 + }, + { + "epoch": 5.44, + "learning_rate": 4.900978463536363e-06, + "loss": 0.1354, + "step": 12995 + }, + { + "epoch": 5.44, + "learning_rate": 4.8995200911659155e-06, + "loss": 0.0975, + "step": 12996 + }, + { + "epoch": 5.44, + "learning_rate": 4.898061865404232e-06, + "loss": 0.1147, + "step": 12997 + }, + { + "epoch": 5.44, + "learning_rate": 4.896603786293233e-06, + "loss": 0.1106, + "step": 12998 + }, + { + "epoch": 5.44, + "learning_rate": 4.89514585387483e-06, + "loss": 0.0958, + "step": 12999 + }, + { + "epoch": 5.44, + "learning_rate": 4.893688068190933e-06, + "loss": 0.0883, + "step": 13000 + }, + { + "epoch": 5.44, + "learning_rate": 4.892230429283438e-06, + "loss": 0.0716, + "step": 13001 + }, + { + "epoch": 5.44, + "learning_rate": 4.890772937194246e-06, + "loss": 0.1229, + "step": 13002 + }, + { + "epoch": 5.44, + "learning_rate": 4.889315591965251e-06, + "loss": 0.1199, + "step": 13003 + }, + { + "epoch": 5.44, + "learning_rate": 4.88785839363835e-06, + "loss": 0.1117, + "step": 13004 + }, + { + "epoch": 5.44, + "learning_rate": 4.886401342255418e-06, + "loss": 0.1406, + "step": 13005 + }, + { + "epoch": 5.45, + "learning_rate": 4.884944437858342e-06, + "loss": 0.1029, + "step": 13006 + }, + { + "epoch": 5.45, + "learning_rate": 4.883487680489e-06, + "loss": 0.129, + "step": 13007 + }, + { + "epoch": 5.45, + "learning_rate": 4.882031070189262e-06, + "loss": 0.1123, + "step": 13008 + }, + { + "epoch": 5.45, + "learning_rate": 4.880574607001001e-06, + "loss": 0.1319, + "step": 13009 + }, + { + "epoch": 5.45, + "learning_rate": 4.879118290966085e-06, + "loss": 0.0958, + "step": 13010 + }, + { + "epoch": 5.45, + "learning_rate": 4.877662122126364e-06, + "loss": 0.1183, + "step": 13011 + }, + { + "epoch": 5.45, + "learning_rate": 4.876206100523701e-06, + "loss": 0.1054, + "step": 13012 + }, + { + "epoch": 5.45, + "learning_rate": 4.874750226199945e-06, + "loss": 0.0947, + "step": 13013 + }, + { + "epoch": 5.45, + "learning_rate": 4.873294499196946e-06, + "loss": 0.0976, + "step": 13014 + }, + { + "epoch": 5.45, + "learning_rate": 4.871838919556549e-06, + "loss": 0.1035, + "step": 13015 + }, + { + "epoch": 5.45, + "learning_rate": 4.87038348732059e-06, + "loss": 0.1318, + "step": 13016 + }, + { + "epoch": 5.45, + "learning_rate": 4.868928202530911e-06, + "loss": 0.1104, + "step": 13017 + }, + { + "epoch": 5.45, + "learning_rate": 4.867473065229332e-06, + "loss": 0.1334, + "step": 13018 + }, + { + "epoch": 5.45, + "learning_rate": 4.866018075457687e-06, + "loss": 0.0944, + "step": 13019 + }, + { + "epoch": 5.45, + "learning_rate": 4.864563233257795e-06, + "loss": 0.1089, + "step": 13020 + }, + { + "epoch": 5.45, + "learning_rate": 4.863108538671477e-06, + "loss": 0.1105, + "step": 13021 + }, + { + "epoch": 5.45, + "learning_rate": 4.861653991740549e-06, + "loss": 0.0989, + "step": 13022 + }, + { + "epoch": 5.45, + "learning_rate": 4.8601995925068136e-06, + "loss": 0.1159, + "step": 13023 + }, + { + "epoch": 5.45, + "learning_rate": 4.85874534101208e-06, + "loss": 0.1002, + "step": 13024 + }, + { + "epoch": 5.45, + "learning_rate": 4.8572912372981526e-06, + "loss": 0.1181, + "step": 13025 + }, + { + "epoch": 5.45, + "learning_rate": 4.855837281406821e-06, + "loss": 0.1238, + "step": 13026 + }, + { + "epoch": 5.45, + "learning_rate": 4.8543834733798826e-06, + "loss": 0.1285, + "step": 13027 + }, + { + "epoch": 5.45, + "learning_rate": 4.852929813259125e-06, + "loss": 0.1352, + "step": 13028 + }, + { + "epoch": 5.45, + "learning_rate": 4.851476301086333e-06, + "loss": 0.1426, + "step": 13029 + }, + { + "epoch": 5.46, + "learning_rate": 4.850022936903285e-06, + "loss": 0.1417, + "step": 13030 + }, + { + "epoch": 5.46, + "learning_rate": 4.848569720751762e-06, + "loss": 0.1086, + "step": 13031 + }, + { + "epoch": 5.46, + "learning_rate": 4.847116652673527e-06, + "loss": 0.128, + "step": 13032 + }, + { + "epoch": 5.46, + "learning_rate": 4.845663732710352e-06, + "loss": 0.1113, + "step": 13033 + }, + { + "epoch": 5.46, + "learning_rate": 4.844210960903998e-06, + "loss": 0.1092, + "step": 13034 + }, + { + "epoch": 5.46, + "learning_rate": 4.842758337296225e-06, + "loss": 0.0939, + "step": 13035 + }, + { + "epoch": 5.46, + "learning_rate": 4.841305861928787e-06, + "loss": 0.1224, + "step": 13036 + }, + { + "epoch": 5.46, + "learning_rate": 4.839853534843439e-06, + "loss": 0.131, + "step": 13037 + }, + { + "epoch": 5.46, + "learning_rate": 4.838401356081918e-06, + "loss": 0.1202, + "step": 13038 + }, + { + "epoch": 5.46, + "learning_rate": 4.83694932568597e-06, + "loss": 0.0958, + "step": 13039 + }, + { + "epoch": 5.46, + "learning_rate": 4.835497443697332e-06, + "loss": 0.1044, + "step": 13040 + }, + { + "epoch": 5.46, + "learning_rate": 4.834045710157735e-06, + "loss": 0.1014, + "step": 13041 + }, + { + "epoch": 5.46, + "learning_rate": 4.832594125108912e-06, + "loss": 0.1027, + "step": 13042 + }, + { + "epoch": 5.46, + "learning_rate": 4.831142688592585e-06, + "loss": 0.1258, + "step": 13043 + }, + { + "epoch": 5.46, + "learning_rate": 4.829691400650478e-06, + "loss": 0.1244, + "step": 13044 + }, + { + "epoch": 5.46, + "learning_rate": 4.828240261324304e-06, + "loss": 0.0685, + "step": 13045 + }, + { + "epoch": 5.46, + "learning_rate": 4.826789270655771e-06, + "loss": 0.1399, + "step": 13046 + }, + { + "epoch": 5.46, + "learning_rate": 4.825338428686589e-06, + "loss": 0.1374, + "step": 13047 + }, + { + "epoch": 5.46, + "learning_rate": 4.823887735458462e-06, + "loss": 0.1102, + "step": 13048 + }, + { + "epoch": 5.46, + "learning_rate": 4.82243719101309e-06, + "loss": 0.1116, + "step": 13049 + }, + { + "epoch": 5.46, + "learning_rate": 4.820986795392165e-06, + "loss": 0.1159, + "step": 13050 + }, + { + "epoch": 5.46, + "learning_rate": 4.819536548637383e-06, + "loss": 0.1404, + "step": 13051 + }, + { + "epoch": 5.46, + "learning_rate": 4.818086450790423e-06, + "loss": 0.1188, + "step": 13052 + }, + { + "epoch": 5.46, + "learning_rate": 4.816636501892968e-06, + "loss": 0.0973, + "step": 13053 + }, + { + "epoch": 5.47, + "learning_rate": 4.8151867019866985e-06, + "loss": 0.093, + "step": 13054 + }, + { + "epoch": 5.47, + "learning_rate": 4.813737051113286e-06, + "loss": 0.143, + "step": 13055 + }, + { + "epoch": 5.47, + "learning_rate": 4.812287549314399e-06, + "loss": 0.1674, + "step": 13056 + }, + { + "epoch": 5.47, + "learning_rate": 4.810838196631704e-06, + "loss": 0.1127, + "step": 13057 + }, + { + "epoch": 5.47, + "learning_rate": 4.809388993106864e-06, + "loss": 0.1456, + "step": 13058 + }, + { + "epoch": 5.47, + "learning_rate": 4.807939938781526e-06, + "loss": 0.1201, + "step": 13059 + }, + { + "epoch": 5.47, + "learning_rate": 4.806491033697348e-06, + "loss": 0.1025, + "step": 13060 + }, + { + "epoch": 5.47, + "learning_rate": 4.805042277895977e-06, + "loss": 0.1205, + "step": 13061 + }, + { + "epoch": 5.47, + "learning_rate": 4.803593671419056e-06, + "loss": 0.1222, + "step": 13062 + }, + { + "epoch": 5.47, + "learning_rate": 4.802145214308223e-06, + "loss": 0.1502, + "step": 13063 + }, + { + "epoch": 5.47, + "learning_rate": 4.800696906605117e-06, + "loss": 0.1467, + "step": 13064 + }, + { + "epoch": 5.47, + "learning_rate": 4.799248748351362e-06, + "loss": 0.1273, + "step": 13065 + }, + { + "epoch": 5.47, + "learning_rate": 4.797800739588585e-06, + "loss": 0.1242, + "step": 13066 + }, + { + "epoch": 5.47, + "learning_rate": 4.796352880358411e-06, + "loss": 0.1119, + "step": 13067 + }, + { + "epoch": 5.47, + "learning_rate": 4.794905170702458e-06, + "loss": 0.1283, + "step": 13068 + }, + { + "epoch": 5.47, + "learning_rate": 4.793457610662334e-06, + "loss": 0.1104, + "step": 13069 + }, + { + "epoch": 5.47, + "learning_rate": 4.792010200279651e-06, + "loss": 0.0905, + "step": 13070 + }, + { + "epoch": 5.47, + "learning_rate": 4.790562939596013e-06, + "loss": 0.0942, + "step": 13071 + }, + { + "epoch": 5.47, + "learning_rate": 4.789115828653025e-06, + "loss": 0.0938, + "step": 13072 + }, + { + "epoch": 5.47, + "learning_rate": 4.787668867492274e-06, + "loss": 0.112, + "step": 13073 + }, + { + "epoch": 5.47, + "learning_rate": 4.786222056155356e-06, + "loss": 0.1225, + "step": 13074 + }, + { + "epoch": 5.47, + "learning_rate": 4.784775394683859e-06, + "loss": 0.0973, + "step": 13075 + }, + { + "epoch": 5.47, + "learning_rate": 4.783328883119365e-06, + "loss": 0.1067, + "step": 13076 + }, + { + "epoch": 5.47, + "learning_rate": 4.781882521503452e-06, + "loss": 0.1221, + "step": 13077 + }, + { + "epoch": 5.48, + "learning_rate": 4.780436309877701e-06, + "loss": 0.0798, + "step": 13078 + }, + { + "epoch": 5.48, + "learning_rate": 4.778990248283671e-06, + "loss": 0.1351, + "step": 13079 + }, + { + "epoch": 5.48, + "learning_rate": 4.777544336762935e-06, + "loss": 0.1175, + "step": 13080 + }, + { + "epoch": 5.48, + "learning_rate": 4.776098575357052e-06, + "loss": 0.1549, + "step": 13081 + }, + { + "epoch": 5.48, + "learning_rate": 4.7746529641075785e-06, + "loss": 0.1547, + "step": 13082 + }, + { + "epoch": 5.48, + "learning_rate": 4.773207503056071e-06, + "loss": 0.0962, + "step": 13083 + }, + { + "epoch": 5.48, + "learning_rate": 4.771762192244075e-06, + "loss": 0.0888, + "step": 13084 + }, + { + "epoch": 5.48, + "learning_rate": 4.770317031713138e-06, + "loss": 0.1445, + "step": 13085 + }, + { + "epoch": 5.48, + "learning_rate": 4.768872021504795e-06, + "loss": 0.156, + "step": 13086 + }, + { + "epoch": 5.48, + "learning_rate": 4.7674271616605816e-06, + "loss": 0.1104, + "step": 13087 + }, + { + "epoch": 5.48, + "learning_rate": 4.765982452222032e-06, + "loss": 0.1365, + "step": 13088 + }, + { + "epoch": 5.48, + "learning_rate": 4.764537893230673e-06, + "loss": 0.1265, + "step": 13089 + }, + { + "epoch": 5.48, + "learning_rate": 4.76309348472803e-06, + "loss": 0.1443, + "step": 13090 + }, + { + "epoch": 5.48, + "learning_rate": 4.7616492267556134e-06, + "loss": 0.0837, + "step": 13091 + }, + { + "epoch": 5.48, + "learning_rate": 4.760205119354946e-06, + "loss": 0.1026, + "step": 13092 + }, + { + "epoch": 5.48, + "learning_rate": 4.758761162567528e-06, + "loss": 0.141, + "step": 13093 + }, + { + "epoch": 5.48, + "learning_rate": 4.757317356434869e-06, + "loss": 0.1503, + "step": 13094 + }, + { + "epoch": 5.48, + "learning_rate": 4.755873700998471e-06, + "loss": 0.1125, + "step": 13095 + }, + { + "epoch": 5.48, + "learning_rate": 4.75443019629983e-06, + "loss": 0.1134, + "step": 13096 + }, + { + "epoch": 5.48, + "learning_rate": 4.7529868423804385e-06, + "loss": 0.1306, + "step": 13097 + }, + { + "epoch": 5.48, + "learning_rate": 4.751543639281784e-06, + "loss": 0.1281, + "step": 13098 + }, + { + "epoch": 5.48, + "learning_rate": 4.750100587045353e-06, + "loss": 0.1129, + "step": 13099 + }, + { + "epoch": 5.48, + "learning_rate": 4.748657685712619e-06, + "loss": 0.11, + "step": 13100 + }, + { + "epoch": 5.49, + "learning_rate": 4.74721493532506e-06, + "loss": 0.1665, + "step": 13101 + }, + { + "epoch": 5.49, + "learning_rate": 4.745772335924146e-06, + "loss": 0.115, + "step": 13102 + }, + { + "epoch": 5.49, + "learning_rate": 4.744329887551343e-06, + "loss": 0.1227, + "step": 13103 + }, + { + "epoch": 5.49, + "learning_rate": 4.742887590248114e-06, + "loss": 0.1139, + "step": 13104 + }, + { + "epoch": 5.49, + "learning_rate": 4.741445444055921e-06, + "loss": 0.1154, + "step": 13105 + }, + { + "epoch": 5.49, + "learning_rate": 4.740003449016208e-06, + "loss": 0.0991, + "step": 13106 + }, + { + "epoch": 5.49, + "learning_rate": 4.7385616051704275e-06, + "loss": 0.1196, + "step": 13107 + }, + { + "epoch": 5.49, + "learning_rate": 4.737119912560025e-06, + "loss": 0.11, + "step": 13108 + }, + { + "epoch": 5.49, + "learning_rate": 4.7356783712264405e-06, + "loss": 0.117, + "step": 13109 + }, + { + "epoch": 5.49, + "learning_rate": 4.734236981211111e-06, + "loss": 0.0986, + "step": 13110 + }, + { + "epoch": 5.49, + "learning_rate": 4.73279574255547e-06, + "loss": 0.1196, + "step": 13111 + }, + { + "epoch": 5.49, + "learning_rate": 4.731354655300936e-06, + "loss": 0.0981, + "step": 13112 + }, + { + "epoch": 5.49, + "learning_rate": 4.729913719488942e-06, + "loss": 0.0829, + "step": 13113 + }, + { + "epoch": 5.49, + "learning_rate": 4.7284729351608985e-06, + "loss": 0.1425, + "step": 13114 + }, + { + "epoch": 5.49, + "learning_rate": 4.727032302358221e-06, + "loss": 0.1211, + "step": 13115 + }, + { + "epoch": 5.49, + "learning_rate": 4.725591821122322e-06, + "loss": 0.0995, + "step": 13116 + }, + { + "epoch": 5.49, + "learning_rate": 4.724151491494605e-06, + "loss": 0.1103, + "step": 13117 + }, + { + "epoch": 5.49, + "learning_rate": 4.722711313516473e-06, + "loss": 0.1047, + "step": 13118 + }, + { + "epoch": 5.49, + "learning_rate": 4.721271287229323e-06, + "loss": 0.1315, + "step": 13119 + }, + { + "epoch": 5.49, + "learning_rate": 4.7198314126745424e-06, + "loss": 0.1178, + "step": 13120 + }, + { + "epoch": 5.49, + "learning_rate": 4.7183916898935234e-06, + "loss": 0.1067, + "step": 13121 + }, + { + "epoch": 5.49, + "learning_rate": 4.716952118927648e-06, + "loss": 0.1286, + "step": 13122 + }, + { + "epoch": 5.49, + "learning_rate": 4.715512699818295e-06, + "loss": 0.103, + "step": 13123 + }, + { + "epoch": 5.49, + "learning_rate": 4.714073432606841e-06, + "loss": 0.1046, + "step": 13124 + }, + { + "epoch": 5.5, + "learning_rate": 4.712634317334655e-06, + "loss": 0.151, + "step": 13125 + }, + { + "epoch": 5.5, + "learning_rate": 4.711195354043109e-06, + "loss": 0.1192, + "step": 13126 + }, + { + "epoch": 5.5, + "learning_rate": 4.709756542773554e-06, + "loss": 0.125, + "step": 13127 + }, + { + "epoch": 5.5, + "learning_rate": 4.708317883567353e-06, + "loss": 0.1561, + "step": 13128 + }, + { + "epoch": 5.5, + "learning_rate": 4.706879376465858e-06, + "loss": 0.1347, + "step": 13129 + }, + { + "epoch": 5.5, + "learning_rate": 4.70544102151042e-06, + "loss": 0.1131, + "step": 13130 + }, + { + "epoch": 5.5, + "learning_rate": 4.7040028187423795e-06, + "loss": 0.0858, + "step": 13131 + }, + { + "epoch": 5.5, + "learning_rate": 4.702564768203083e-06, + "loss": 0.1196, + "step": 13132 + }, + { + "epoch": 5.5, + "learning_rate": 4.701126869933858e-06, + "loss": 0.1426, + "step": 13133 + }, + { + "epoch": 5.5, + "learning_rate": 4.699689123976039e-06, + "loss": 0.1101, + "step": 13134 + }, + { + "epoch": 5.5, + "learning_rate": 4.698251530370956e-06, + "loss": 0.1094, + "step": 13135 + }, + { + "epoch": 5.5, + "learning_rate": 4.696814089159924e-06, + "loss": 0.1387, + "step": 13136 + }, + { + "epoch": 5.5, + "learning_rate": 4.695376800384266e-06, + "loss": 0.1346, + "step": 13137 + }, + { + "epoch": 5.5, + "learning_rate": 4.6939396640852945e-06, + "loss": 0.1599, + "step": 13138 + }, + { + "epoch": 5.5, + "learning_rate": 4.6925026803043186e-06, + "loss": 0.1209, + "step": 13139 + }, + { + "epoch": 5.5, + "learning_rate": 4.691065849082647e-06, + "loss": 0.1277, + "step": 13140 + }, + { + "epoch": 5.5, + "learning_rate": 4.6896291704615725e-06, + "loss": 0.1685, + "step": 13141 + }, + { + "epoch": 5.5, + "learning_rate": 4.688192644482396e-06, + "loss": 0.1097, + "step": 13142 + }, + { + "epoch": 5.5, + "learning_rate": 4.686756271186407e-06, + "loss": 0.139, + "step": 13143 + }, + { + "epoch": 5.5, + "learning_rate": 4.6853200506148955e-06, + "loss": 0.1138, + "step": 13144 + }, + { + "epoch": 5.5, + "learning_rate": 4.683883982809142e-06, + "loss": 0.1214, + "step": 13145 + }, + { + "epoch": 5.5, + "learning_rate": 4.68244806781043e-06, + "loss": 0.1228, + "step": 13146 + }, + { + "epoch": 5.5, + "learning_rate": 4.681012305660026e-06, + "loss": 0.097, + "step": 13147 + }, + { + "epoch": 5.5, + "learning_rate": 4.679576696399203e-06, + "loss": 0.1354, + "step": 13148 + }, + { + "epoch": 5.51, + "learning_rate": 4.678141240069227e-06, + "loss": 0.1367, + "step": 13149 + }, + { + "epoch": 5.51, + "learning_rate": 4.676705936711358e-06, + "loss": 0.0956, + "step": 13150 + }, + { + "epoch": 5.51, + "learning_rate": 4.675270786366853e-06, + "loss": 0.1226, + "step": 13151 + }, + { + "epoch": 5.51, + "learning_rate": 4.673835789076969e-06, + "loss": 0.1573, + "step": 13152 + }, + { + "epoch": 5.51, + "learning_rate": 4.672400944882944e-06, + "loss": 0.1212, + "step": 13153 + }, + { + "epoch": 5.51, + "learning_rate": 4.670966253826027e-06, + "loss": 0.1723, + "step": 13154 + }, + { + "epoch": 5.51, + "learning_rate": 4.669531715947455e-06, + "loss": 0.1733, + "step": 13155 + }, + { + "epoch": 5.51, + "learning_rate": 4.668097331288465e-06, + "loss": 0.1056, + "step": 13156 + }, + { + "epoch": 5.51, + "learning_rate": 4.666663099890285e-06, + "loss": 0.088, + "step": 13157 + }, + { + "epoch": 5.51, + "learning_rate": 4.665229021794145e-06, + "loss": 0.1706, + "step": 13158 + }, + { + "epoch": 5.51, + "learning_rate": 4.66379509704126e-06, + "loss": 0.0982, + "step": 13159 + }, + { + "epoch": 5.51, + "learning_rate": 4.662361325672853e-06, + "loss": 0.1061, + "step": 13160 + }, + { + "epoch": 5.51, + "learning_rate": 4.660927707730129e-06, + "loss": 0.0863, + "step": 13161 + }, + { + "epoch": 5.51, + "learning_rate": 4.659494243254301e-06, + "loss": 0.1444, + "step": 13162 + }, + { + "epoch": 5.51, + "learning_rate": 4.658060932286571e-06, + "loss": 0.0896, + "step": 13163 + }, + { + "epoch": 5.51, + "learning_rate": 4.65662777486814e-06, + "loss": 0.1124, + "step": 13164 + }, + { + "epoch": 5.51, + "learning_rate": 4.6551947710402016e-06, + "loss": 0.13, + "step": 13165 + }, + { + "epoch": 5.51, + "learning_rate": 4.653761920843946e-06, + "loss": 0.1145, + "step": 13166 + }, + { + "epoch": 5.51, + "learning_rate": 4.652329224320564e-06, + "loss": 0.1004, + "step": 13167 + }, + { + "epoch": 5.51, + "learning_rate": 4.65089668151123e-06, + "loss": 0.0881, + "step": 13168 + }, + { + "epoch": 5.51, + "learning_rate": 4.649464292457123e-06, + "loss": 0.1377, + "step": 13169 + }, + { + "epoch": 5.51, + "learning_rate": 4.648032057199418e-06, + "loss": 0.107, + "step": 13170 + }, + { + "epoch": 5.51, + "learning_rate": 4.6465999757792815e-06, + "loss": 0.1474, + "step": 13171 + }, + { + "epoch": 5.51, + "learning_rate": 4.645168048237878e-06, + "loss": 0.145, + "step": 13172 + }, + { + "epoch": 5.52, + "learning_rate": 4.643736274616372e-06, + "loss": 0.1343, + "step": 13173 + }, + { + "epoch": 5.52, + "learning_rate": 4.6423046549559095e-06, + "loss": 0.1351, + "step": 13174 + }, + { + "epoch": 5.52, + "learning_rate": 4.640873189297645e-06, + "loss": 0.1167, + "step": 13175 + }, + { + "epoch": 5.52, + "learning_rate": 4.639441877682724e-06, + "loss": 0.1504, + "step": 13176 + }, + { + "epoch": 5.52, + "learning_rate": 4.638010720152292e-06, + "loss": 0.1185, + "step": 13177 + }, + { + "epoch": 5.52, + "learning_rate": 4.6365797167474815e-06, + "loss": 0.1556, + "step": 13178 + }, + { + "epoch": 5.52, + "learning_rate": 4.635148867509432e-06, + "loss": 0.1527, + "step": 13179 + }, + { + "epoch": 5.52, + "learning_rate": 4.6337181724792644e-06, + "loss": 0.1212, + "step": 13180 + }, + { + "epoch": 5.52, + "learning_rate": 4.6322876316981095e-06, + "loss": 0.1378, + "step": 13181 + }, + { + "epoch": 5.52, + "learning_rate": 4.630857245207079e-06, + "loss": 0.1086, + "step": 13182 + }, + { + "epoch": 5.52, + "learning_rate": 4.629427013047293e-06, + "loss": 0.1517, + "step": 13183 + }, + { + "epoch": 5.52, + "learning_rate": 4.627996935259862e-06, + "loss": 0.1087, + "step": 13184 + }, + { + "epoch": 5.52, + "learning_rate": 4.626567011885892e-06, + "loss": 0.1032, + "step": 13185 + }, + { + "epoch": 5.52, + "learning_rate": 4.625137242966484e-06, + "loss": 0.0961, + "step": 13186 + }, + { + "epoch": 5.52, + "learning_rate": 4.623707628542742e-06, + "loss": 0.1232, + "step": 13187 + }, + { + "epoch": 5.52, + "learning_rate": 4.6222781686557485e-06, + "loss": 0.1486, + "step": 13188 + }, + { + "epoch": 5.52, + "learning_rate": 4.620848863346596e-06, + "loss": 0.157, + "step": 13189 + }, + { + "epoch": 5.52, + "learning_rate": 4.6194197126563715e-06, + "loss": 0.1104, + "step": 13190 + }, + { + "epoch": 5.52, + "learning_rate": 4.617990716626151e-06, + "loss": 0.0985, + "step": 13191 + }, + { + "epoch": 5.52, + "learning_rate": 4.616561875297012e-06, + "loss": 0.0959, + "step": 13192 + }, + { + "epoch": 5.52, + "learning_rate": 4.615133188710029e-06, + "loss": 0.0781, + "step": 13193 + }, + { + "epoch": 5.52, + "learning_rate": 4.6137046569062595e-06, + "loss": 0.1041, + "step": 13194 + }, + { + "epoch": 5.52, + "learning_rate": 4.612276279926771e-06, + "loss": 0.1304, + "step": 13195 + }, + { + "epoch": 5.52, + "learning_rate": 4.610848057812619e-06, + "loss": 0.1292, + "step": 13196 + }, + { + "epoch": 5.53, + "learning_rate": 4.609419990604858e-06, + "loss": 0.1582, + "step": 13197 + }, + { + "epoch": 5.53, + "learning_rate": 4.607992078344534e-06, + "loss": 0.1267, + "step": 13198 + }, + { + "epoch": 5.53, + "learning_rate": 4.6065643210726955e-06, + "loss": 0.0792, + "step": 13199 + }, + { + "epoch": 5.53, + "learning_rate": 4.605136718830382e-06, + "loss": 0.1392, + "step": 13200 + }, + { + "epoch": 5.53, + "learning_rate": 4.603709271658621e-06, + "loss": 0.1323, + "step": 13201 + }, + { + "epoch": 5.53, + "learning_rate": 4.602281979598451e-06, + "loss": 0.1288, + "step": 13202 + }, + { + "epoch": 5.53, + "learning_rate": 4.600854842690898e-06, + "loss": 0.1464, + "step": 13203 + }, + { + "epoch": 5.53, + "learning_rate": 4.599427860976978e-06, + "loss": 0.092, + "step": 13204 + }, + { + "epoch": 5.53, + "learning_rate": 4.598001034497712e-06, + "loss": 0.1516, + "step": 13205 + }, + { + "epoch": 5.53, + "learning_rate": 4.596574363294113e-06, + "loss": 0.1023, + "step": 13206 + }, + { + "epoch": 5.53, + "learning_rate": 4.595147847407187e-06, + "loss": 0.173, + "step": 13207 + }, + { + "epoch": 5.53, + "learning_rate": 4.593721486877945e-06, + "loss": 0.1068, + "step": 13208 + }, + { + "epoch": 5.53, + "learning_rate": 4.592295281747379e-06, + "loss": 0.1424, + "step": 13209 + }, + { + "epoch": 5.53, + "learning_rate": 4.590869232056485e-06, + "loss": 0.1042, + "step": 13210 + }, + { + "epoch": 5.53, + "learning_rate": 4.589443337846256e-06, + "loss": 0.1413, + "step": 13211 + }, + { + "epoch": 5.53, + "learning_rate": 4.588017599157675e-06, + "loss": 0.1487, + "step": 13212 + }, + { + "epoch": 5.53, + "learning_rate": 4.5865920160317266e-06, + "loss": 0.1182, + "step": 13213 + }, + { + "epoch": 5.53, + "learning_rate": 4.585166588509391e-06, + "loss": 0.1448, + "step": 13214 + }, + { + "epoch": 5.53, + "learning_rate": 4.5837413166316335e-06, + "loss": 0.1197, + "step": 13215 + }, + { + "epoch": 5.53, + "learning_rate": 4.582316200439425e-06, + "loss": 0.1316, + "step": 13216 + }, + { + "epoch": 5.53, + "learning_rate": 4.580891239973729e-06, + "loss": 0.1344, + "step": 13217 + }, + { + "epoch": 5.53, + "learning_rate": 4.579466435275506e-06, + "loss": 0.1287, + "step": 13218 + }, + { + "epoch": 5.53, + "learning_rate": 4.578041786385711e-06, + "loss": 0.1488, + "step": 13219 + }, + { + "epoch": 5.53, + "learning_rate": 4.576617293345296e-06, + "loss": 0.1474, + "step": 13220 + }, + { + "epoch": 5.54, + "learning_rate": 4.575192956195201e-06, + "loss": 0.1238, + "step": 13221 + }, + { + "epoch": 5.54, + "learning_rate": 4.573768774976371e-06, + "loss": 0.1045, + "step": 13222 + }, + { + "epoch": 5.54, + "learning_rate": 4.5723447497297424e-06, + "loss": 0.1188, + "step": 13223 + }, + { + "epoch": 5.54, + "learning_rate": 4.570920880496247e-06, + "loss": 0.1278, + "step": 13224 + }, + { + "epoch": 5.54, + "learning_rate": 4.569497167316813e-06, + "loss": 0.1062, + "step": 13225 + }, + { + "epoch": 5.54, + "learning_rate": 4.568073610232368e-06, + "loss": 0.1161, + "step": 13226 + }, + { + "epoch": 5.54, + "learning_rate": 4.566650209283824e-06, + "loss": 0.1178, + "step": 13227 + }, + { + "epoch": 5.54, + "learning_rate": 4.565226964512102e-06, + "loss": 0.1045, + "step": 13228 + }, + { + "epoch": 5.54, + "learning_rate": 4.563803875958103e-06, + "loss": 0.112, + "step": 13229 + }, + { + "epoch": 5.54, + "learning_rate": 4.562380943662739e-06, + "loss": 0.1019, + "step": 13230 + }, + { + "epoch": 5.54, + "learning_rate": 4.560958167666908e-06, + "loss": 0.1122, + "step": 13231 + }, + { + "epoch": 5.54, + "learning_rate": 4.5595355480115086e-06, + "loss": 0.1071, + "step": 13232 + }, + { + "epoch": 5.54, + "learning_rate": 4.558113084737434e-06, + "loss": 0.1029, + "step": 13233 + }, + { + "epoch": 5.54, + "learning_rate": 4.556690777885571e-06, + "loss": 0.1232, + "step": 13234 + }, + { + "epoch": 5.54, + "learning_rate": 4.5552686274967984e-06, + "loss": 0.1612, + "step": 13235 + }, + { + "epoch": 5.54, + "learning_rate": 4.553846633611997e-06, + "loss": 0.1484, + "step": 13236 + }, + { + "epoch": 5.54, + "learning_rate": 4.552424796272041e-06, + "loss": 0.072, + "step": 13237 + }, + { + "epoch": 5.54, + "learning_rate": 4.551003115517801e-06, + "loss": 0.1069, + "step": 13238 + }, + { + "epoch": 5.54, + "learning_rate": 4.549581591390142e-06, + "loss": 0.1268, + "step": 13239 + }, + { + "epoch": 5.54, + "learning_rate": 4.548160223929923e-06, + "loss": 0.099, + "step": 13240 + }, + { + "epoch": 5.54, + "learning_rate": 4.546739013178003e-06, + "loss": 0.1366, + "step": 13241 + }, + { + "epoch": 5.54, + "learning_rate": 4.545317959175228e-06, + "loss": 0.1154, + "step": 13242 + }, + { + "epoch": 5.54, + "learning_rate": 4.543897061962448e-06, + "loss": 0.1007, + "step": 13243 + }, + { + "epoch": 5.54, + "learning_rate": 4.542476321580505e-06, + "loss": 0.1371, + "step": 13244 + }, + { + "epoch": 5.55, + "learning_rate": 4.541055738070237e-06, + "loss": 0.1588, + "step": 13245 + }, + { + "epoch": 5.55, + "learning_rate": 4.539635311472478e-06, + "loss": 0.154, + "step": 13246 + }, + { + "epoch": 5.55, + "learning_rate": 4.53821504182806e-06, + "loss": 0.0952, + "step": 13247 + }, + { + "epoch": 5.55, + "learning_rate": 4.5367949291778e-06, + "loss": 0.1177, + "step": 13248 + }, + { + "epoch": 5.55, + "learning_rate": 4.535374973562526e-06, + "loss": 0.1216, + "step": 13249 + }, + { + "epoch": 5.55, + "learning_rate": 4.533955175023046e-06, + "loss": 0.1066, + "step": 13250 + }, + { + "epoch": 5.55, + "learning_rate": 4.532535533600173e-06, + "loss": 0.0986, + "step": 13251 + }, + { + "epoch": 5.55, + "learning_rate": 4.531116049334715e-06, + "loss": 0.1444, + "step": 13252 + }, + { + "epoch": 5.55, + "learning_rate": 4.5296967222674724e-06, + "loss": 0.11, + "step": 13253 + }, + { + "epoch": 5.55, + "learning_rate": 4.528277552439244e-06, + "loss": 0.1187, + "step": 13254 + }, + { + "epoch": 5.55, + "learning_rate": 4.526858539890825e-06, + "loss": 0.1006, + "step": 13255 + }, + { + "epoch": 5.55, + "learning_rate": 4.5254396846629975e-06, + "loss": 0.086, + "step": 13256 + }, + { + "epoch": 5.55, + "learning_rate": 4.524020986796548e-06, + "loss": 0.1032, + "step": 13257 + }, + { + "epoch": 5.55, + "learning_rate": 4.522602446332256e-06, + "loss": 0.1266, + "step": 13258 + }, + { + "epoch": 5.55, + "learning_rate": 4.521184063310895e-06, + "loss": 0.1116, + "step": 13259 + }, + { + "epoch": 5.55, + "learning_rate": 4.519765837773238e-06, + "loss": 0.1249, + "step": 13260 + }, + { + "epoch": 5.55, + "learning_rate": 4.518347769760051e-06, + "loss": 0.0959, + "step": 13261 + }, + { + "epoch": 5.55, + "learning_rate": 4.51692985931209e-06, + "loss": 0.1262, + "step": 13262 + }, + { + "epoch": 5.55, + "learning_rate": 4.515512106470115e-06, + "loss": 0.1021, + "step": 13263 + }, + { + "epoch": 5.55, + "learning_rate": 4.514094511274877e-06, + "loss": 0.1451, + "step": 13264 + }, + { + "epoch": 5.55, + "learning_rate": 4.512677073767124e-06, + "loss": 0.1179, + "step": 13265 + }, + { + "epoch": 5.55, + "learning_rate": 4.5112597939876e-06, + "loss": 0.1279, + "step": 13266 + }, + { + "epoch": 5.55, + "learning_rate": 4.509842671977043e-06, + "loss": 0.0967, + "step": 13267 + }, + { + "epoch": 5.55, + "learning_rate": 4.50842570777619e-06, + "loss": 0.0605, + "step": 13268 + }, + { + "epoch": 5.56, + "learning_rate": 4.507008901425762e-06, + "loss": 0.0834, + "step": 13269 + }, + { + "epoch": 5.56, + "learning_rate": 4.50559225296649e-06, + "loss": 0.1333, + "step": 13270 + }, + { + "epoch": 5.56, + "learning_rate": 4.504175762439096e-06, + "loss": 0.1459, + "step": 13271 + }, + { + "epoch": 5.56, + "learning_rate": 4.5027594298842884e-06, + "loss": 0.1377, + "step": 13272 + }, + { + "epoch": 5.56, + "learning_rate": 4.5013432553427835e-06, + "loss": 0.0817, + "step": 13273 + }, + { + "epoch": 5.56, + "learning_rate": 4.499927238855287e-06, + "loss": 0.1335, + "step": 13274 + }, + { + "epoch": 5.56, + "learning_rate": 4.498511380462506e-06, + "loss": 0.1125, + "step": 13275 + }, + { + "epoch": 5.56, + "learning_rate": 4.4970956802051295e-06, + "loss": 0.1297, + "step": 13276 + }, + { + "epoch": 5.56, + "learning_rate": 4.495680138123855e-06, + "loss": 0.1283, + "step": 13277 + }, + { + "epoch": 5.56, + "learning_rate": 4.494264754259368e-06, + "loss": 0.1291, + "step": 13278 + }, + { + "epoch": 5.56, + "learning_rate": 4.492849528652357e-06, + "loss": 0.1326, + "step": 13279 + }, + { + "epoch": 5.56, + "learning_rate": 4.491434461343499e-06, + "loss": 0.109, + "step": 13280 + }, + { + "epoch": 5.56, + "learning_rate": 4.49001955237347e-06, + "loss": 0.1538, + "step": 13281 + }, + { + "epoch": 5.56, + "learning_rate": 4.488604801782942e-06, + "loss": 0.1198, + "step": 13282 + }, + { + "epoch": 5.56, + "learning_rate": 4.4871902096125755e-06, + "loss": 0.1009, + "step": 13283 + }, + { + "epoch": 5.56, + "learning_rate": 4.485775775903033e-06, + "loss": 0.1616, + "step": 13284 + }, + { + "epoch": 5.56, + "learning_rate": 4.484361500694974e-06, + "loss": 0.0977, + "step": 13285 + }, + { + "epoch": 5.56, + "learning_rate": 4.482947384029048e-06, + "loss": 0.1267, + "step": 13286 + }, + { + "epoch": 5.56, + "learning_rate": 4.481533425945904e-06, + "loss": 0.153, + "step": 13287 + }, + { + "epoch": 5.56, + "learning_rate": 4.480119626486189e-06, + "loss": 0.143, + "step": 13288 + }, + { + "epoch": 5.56, + "learning_rate": 4.4787059856905334e-06, + "loss": 0.0814, + "step": 13289 + }, + { + "epoch": 5.56, + "learning_rate": 4.477292503599574e-06, + "loss": 0.1323, + "step": 13290 + }, + { + "epoch": 5.56, + "learning_rate": 4.475879180253941e-06, + "loss": 0.1213, + "step": 13291 + }, + { + "epoch": 5.56, + "learning_rate": 4.47446601569426e-06, + "loss": 0.1197, + "step": 13292 + }, + { + "epoch": 5.57, + "learning_rate": 4.473053009961149e-06, + "loss": 0.1439, + "step": 13293 + }, + { + "epoch": 5.57, + "learning_rate": 4.471640163095229e-06, + "loss": 0.147, + "step": 13294 + }, + { + "epoch": 5.57, + "learning_rate": 4.470227475137102e-06, + "loss": 0.0922, + "step": 13295 + }, + { + "epoch": 5.57, + "learning_rate": 4.4688149461273845e-06, + "loss": 0.1109, + "step": 13296 + }, + { + "epoch": 5.57, + "learning_rate": 4.467402576106668e-06, + "loss": 0.1183, + "step": 13297 + }, + { + "epoch": 5.57, + "learning_rate": 4.465990365115555e-06, + "loss": 0.0781, + "step": 13298 + }, + { + "epoch": 5.57, + "learning_rate": 4.46457831319464e-06, + "loss": 0.1439, + "step": 13299 + }, + { + "epoch": 5.57, + "learning_rate": 4.463166420384507e-06, + "loss": 0.1276, + "step": 13300 + }, + { + "epoch": 5.57, + "learning_rate": 4.461754686725743e-06, + "loss": 0.1152, + "step": 13301 + }, + { + "epoch": 5.57, + "learning_rate": 4.4603431122589295e-06, + "loss": 0.1117, + "step": 13302 + }, + { + "epoch": 5.57, + "learning_rate": 4.458931697024633e-06, + "loss": 0.1098, + "step": 13303 + }, + { + "epoch": 5.57, + "learning_rate": 4.457520441063428e-06, + "loss": 0.1016, + "step": 13304 + }, + { + "epoch": 5.57, + "learning_rate": 4.456109344415881e-06, + "loss": 0.1145, + "step": 13305 + }, + { + "epoch": 5.57, + "learning_rate": 4.454698407122551e-06, + "loss": 0.1367, + "step": 13306 + }, + { + "epoch": 5.57, + "learning_rate": 4.453287629223994e-06, + "loss": 0.1371, + "step": 13307 + }, + { + "epoch": 5.57, + "learning_rate": 4.4518770107607626e-06, + "loss": 0.0942, + "step": 13308 + }, + { + "epoch": 5.57, + "learning_rate": 4.450466551773406e-06, + "loss": 0.114, + "step": 13309 + }, + { + "epoch": 5.57, + "learning_rate": 4.449056252302461e-06, + "loss": 0.1576, + "step": 13310 + }, + { + "epoch": 5.57, + "learning_rate": 4.447646112388467e-06, + "loss": 0.1172, + "step": 13311 + }, + { + "epoch": 5.57, + "learning_rate": 4.44623613207196e-06, + "loss": 0.1272, + "step": 13312 + }, + { + "epoch": 5.57, + "learning_rate": 4.444826311393465e-06, + "loss": 0.1114, + "step": 13313 + }, + { + "epoch": 5.57, + "learning_rate": 4.443416650393509e-06, + "loss": 0.112, + "step": 13314 + }, + { + "epoch": 5.57, + "learning_rate": 4.442007149112615e-06, + "loss": 0.1217, + "step": 13315 + }, + { + "epoch": 5.58, + "learning_rate": 4.44059780759129e-06, + "loss": 0.1172, + "step": 13316 + }, + { + "epoch": 5.58, + "learning_rate": 4.43918862587005e-06, + "loss": 0.1354, + "step": 13317 + }, + { + "epoch": 5.58, + "learning_rate": 4.437779603989395e-06, + "loss": 0.1031, + "step": 13318 + }, + { + "epoch": 5.58, + "learning_rate": 4.436370741989829e-06, + "loss": 0.154, + "step": 13319 + }, + { + "epoch": 5.58, + "learning_rate": 4.4349620399118495e-06, + "loss": 0.1524, + "step": 13320 + }, + { + "epoch": 5.58, + "learning_rate": 4.43355349779595e-06, + "loss": 0.0977, + "step": 13321 + }, + { + "epoch": 5.58, + "learning_rate": 4.432145115682613e-06, + "loss": 0.1257, + "step": 13322 + }, + { + "epoch": 5.58, + "learning_rate": 4.430736893612327e-06, + "loss": 0.1262, + "step": 13323 + }, + { + "epoch": 5.58, + "learning_rate": 4.429328831625565e-06, + "loss": 0.0848, + "step": 13324 + }, + { + "epoch": 5.58, + "learning_rate": 4.427920929762803e-06, + "loss": 0.1323, + "step": 13325 + }, + { + "epoch": 5.58, + "learning_rate": 4.426513188064507e-06, + "loss": 0.1331, + "step": 13326 + }, + { + "epoch": 5.58, + "learning_rate": 4.425105606571145e-06, + "loss": 0.1298, + "step": 13327 + }, + { + "epoch": 5.58, + "learning_rate": 4.423698185323175e-06, + "loss": 0.1379, + "step": 13328 + }, + { + "epoch": 5.58, + "learning_rate": 4.422290924361056e-06, + "loss": 0.1413, + "step": 13329 + }, + { + "epoch": 5.58, + "learning_rate": 4.42088382372523e-06, + "loss": 0.086, + "step": 13330 + }, + { + "epoch": 5.58, + "learning_rate": 4.41947688345615e-06, + "loss": 0.1054, + "step": 13331 + }, + { + "epoch": 5.58, + "learning_rate": 4.418070103594253e-06, + "loss": 0.1087, + "step": 13332 + }, + { + "epoch": 5.58, + "learning_rate": 4.416663484179978e-06, + "loss": 0.138, + "step": 13333 + }, + { + "epoch": 5.58, + "learning_rate": 4.415257025253758e-06, + "loss": 0.1474, + "step": 13334 + }, + { + "epoch": 5.58, + "learning_rate": 4.41385072685602e-06, + "loss": 0.1176, + "step": 13335 + }, + { + "epoch": 5.58, + "learning_rate": 4.412444589027185e-06, + "loss": 0.1425, + "step": 13336 + }, + { + "epoch": 5.58, + "learning_rate": 4.41103861180767e-06, + "loss": 0.1142, + "step": 13337 + }, + { + "epoch": 5.58, + "learning_rate": 4.4096327952378915e-06, + "loss": 0.0912, + "step": 13338 + }, + { + "epoch": 5.58, + "learning_rate": 4.408227139358261e-06, + "loss": 0.1089, + "step": 13339 + }, + { + "epoch": 5.59, + "learning_rate": 4.406821644209175e-06, + "loss": 0.1501, + "step": 13340 + }, + { + "epoch": 5.59, + "learning_rate": 4.4054163098310375e-06, + "loss": 0.1016, + "step": 13341 + }, + { + "epoch": 5.59, + "learning_rate": 4.404011136264245e-06, + "loss": 0.0911, + "step": 13342 + }, + { + "epoch": 5.59, + "learning_rate": 4.402606123549189e-06, + "loss": 0.128, + "step": 13343 + }, + { + "epoch": 5.59, + "learning_rate": 4.40120127172625e-06, + "loss": 0.1497, + "step": 13344 + }, + { + "epoch": 5.59, + "learning_rate": 4.399796580835811e-06, + "loss": 0.1469, + "step": 13345 + }, + { + "epoch": 5.59, + "learning_rate": 4.398392050918251e-06, + "loss": 0.1182, + "step": 13346 + }, + { + "epoch": 5.59, + "learning_rate": 4.39698768201394e-06, + "loss": 0.1152, + "step": 13347 + }, + { + "epoch": 5.59, + "learning_rate": 4.395583474163246e-06, + "loss": 0.1244, + "step": 13348 + }, + { + "epoch": 5.59, + "learning_rate": 4.394179427406533e-06, + "loss": 0.103, + "step": 13349 + }, + { + "epoch": 5.59, + "learning_rate": 4.392775541784161e-06, + "loss": 0.1087, + "step": 13350 + }, + { + "epoch": 5.59, + "learning_rate": 4.391371817336476e-06, + "loss": 0.1513, + "step": 13351 + }, + { + "epoch": 5.59, + "learning_rate": 4.3899682541038316e-06, + "loss": 0.1423, + "step": 13352 + }, + { + "epoch": 5.59, + "learning_rate": 4.388564852126571e-06, + "loss": 0.0978, + "step": 13353 + }, + { + "epoch": 5.59, + "learning_rate": 4.387161611445034e-06, + "loss": 0.1128, + "step": 13354 + }, + { + "epoch": 5.59, + "learning_rate": 4.385758532099556e-06, + "loss": 0.1407, + "step": 13355 + }, + { + "epoch": 5.59, + "learning_rate": 4.384355614130471e-06, + "loss": 0.1338, + "step": 13356 + }, + { + "epoch": 5.59, + "learning_rate": 4.382952857578098e-06, + "loss": 0.1307, + "step": 13357 + }, + { + "epoch": 5.59, + "learning_rate": 4.381550262482759e-06, + "loss": 0.1181, + "step": 13358 + }, + { + "epoch": 5.59, + "learning_rate": 4.380147828884772e-06, + "loss": 0.1042, + "step": 13359 + }, + { + "epoch": 5.59, + "learning_rate": 4.37874555682445e-06, + "loss": 0.1201, + "step": 13360 + }, + { + "epoch": 5.59, + "learning_rate": 4.377343446342097e-06, + "loss": 0.1473, + "step": 13361 + }, + { + "epoch": 5.59, + "learning_rate": 4.375941497478022e-06, + "loss": 0.1154, + "step": 13362 + }, + { + "epoch": 5.59, + "learning_rate": 4.374539710272514e-06, + "loss": 0.1125, + "step": 13363 + }, + { + "epoch": 5.6, + "learning_rate": 4.373138084765875e-06, + "loss": 0.132, + "step": 13364 + }, + { + "epoch": 5.6, + "learning_rate": 4.371736620998383e-06, + "loss": 0.1434, + "step": 13365 + }, + { + "epoch": 5.6, + "learning_rate": 4.370335319010328e-06, + "loss": 0.1287, + "step": 13366 + }, + { + "epoch": 5.6, + "learning_rate": 4.368934178841988e-06, + "loss": 0.1432, + "step": 13367 + }, + { + "epoch": 5.6, + "learning_rate": 4.367533200533639e-06, + "loss": 0.1219, + "step": 13368 + }, + { + "epoch": 5.6, + "learning_rate": 4.36613238412555e-06, + "loss": 0.1447, + "step": 13369 + }, + { + "epoch": 5.6, + "learning_rate": 4.36473172965799e-06, + "loss": 0.1143, + "step": 13370 + }, + { + "epoch": 5.6, + "learning_rate": 4.363331237171211e-06, + "loss": 0.111, + "step": 13371 + }, + { + "epoch": 5.6, + "learning_rate": 4.361930906705475e-06, + "loss": 0.115, + "step": 13372 + }, + { + "epoch": 5.6, + "learning_rate": 4.360530738301031e-06, + "loss": 0.1121, + "step": 13373 + }, + { + "epoch": 5.6, + "learning_rate": 4.359130731998127e-06, + "loss": 0.1524, + "step": 13374 + }, + { + "epoch": 5.6, + "learning_rate": 4.357730887837005e-06, + "loss": 0.1011, + "step": 13375 + }, + { + "epoch": 5.6, + "learning_rate": 4.356331205857904e-06, + "loss": 0.1064, + "step": 13376 + }, + { + "epoch": 5.6, + "learning_rate": 4.354931686101051e-06, + "loss": 0.1177, + "step": 13377 + }, + { + "epoch": 5.6, + "learning_rate": 4.353532328606679e-06, + "loss": 0.0824, + "step": 13378 + }, + { + "epoch": 5.6, + "learning_rate": 4.352133133415009e-06, + "loss": 0.12, + "step": 13379 + }, + { + "epoch": 5.6, + "learning_rate": 4.3507341005662595e-06, + "loss": 0.1201, + "step": 13380 + }, + { + "epoch": 5.6, + "learning_rate": 4.349335230100645e-06, + "loss": 0.0962, + "step": 13381 + }, + { + "epoch": 5.6, + "learning_rate": 4.3479365220583754e-06, + "loss": 0.1817, + "step": 13382 + }, + { + "epoch": 5.6, + "learning_rate": 4.346537976479659e-06, + "loss": 0.143, + "step": 13383 + }, + { + "epoch": 5.6, + "learning_rate": 4.345139593404688e-06, + "loss": 0.1115, + "step": 13384 + }, + { + "epoch": 5.6, + "learning_rate": 4.343741372873665e-06, + "loss": 0.104, + "step": 13385 + }, + { + "epoch": 5.6, + "learning_rate": 4.342343314926773e-06, + "loss": 0.1272, + "step": 13386 + }, + { + "epoch": 5.6, + "learning_rate": 4.340945419604201e-06, + "loss": 0.1128, + "step": 13387 + }, + { + "epoch": 5.61, + "learning_rate": 4.3395476869461325e-06, + "loss": 0.115, + "step": 13388 + }, + { + "epoch": 5.61, + "learning_rate": 4.338150116992743e-06, + "loss": 0.1597, + "step": 13389 + }, + { + "epoch": 5.61, + "learning_rate": 4.3367527097842034e-06, + "loss": 0.1409, + "step": 13390 + }, + { + "epoch": 5.61, + "learning_rate": 4.335355465360686e-06, + "loss": 0.1341, + "step": 13391 + }, + { + "epoch": 5.61, + "learning_rate": 4.333958383762345e-06, + "loss": 0.1257, + "step": 13392 + }, + { + "epoch": 5.61, + "learning_rate": 4.332561465029341e-06, + "loss": 0.1009, + "step": 13393 + }, + { + "epoch": 5.61, + "learning_rate": 4.331164709201829e-06, + "loss": 0.099, + "step": 13394 + }, + { + "epoch": 5.61, + "learning_rate": 4.329768116319958e-06, + "loss": 0.0919, + "step": 13395 + }, + { + "epoch": 5.61, + "learning_rate": 4.328371686423871e-06, + "loss": 0.085, + "step": 13396 + }, + { + "epoch": 5.61, + "learning_rate": 4.32697541955371e-06, + "loss": 0.1586, + "step": 13397 + }, + { + "epoch": 5.61, + "learning_rate": 4.325579315749603e-06, + "loss": 0.1115, + "step": 13398 + }, + { + "epoch": 5.61, + "learning_rate": 4.324183375051683e-06, + "loss": 0.1282, + "step": 13399 + }, + { + "epoch": 5.61, + "learning_rate": 4.322787597500074e-06, + "loss": 0.0924, + "step": 13400 + }, + { + "epoch": 5.61, + "learning_rate": 4.321391983134899e-06, + "loss": 0.1703, + "step": 13401 + }, + { + "epoch": 5.61, + "learning_rate": 4.319996531996272e-06, + "loss": 0.1231, + "step": 13402 + }, + { + "epoch": 5.61, + "learning_rate": 4.318601244124308e-06, + "loss": 0.1272, + "step": 13403 + }, + { + "epoch": 5.61, + "learning_rate": 4.317206119559108e-06, + "loss": 0.114, + "step": 13404 + }, + { + "epoch": 5.61, + "learning_rate": 4.315811158340773e-06, + "loss": 0.11, + "step": 13405 + }, + { + "epoch": 5.61, + "learning_rate": 4.314416360509403e-06, + "loss": 0.1258, + "step": 13406 + }, + { + "epoch": 5.61, + "learning_rate": 4.3130217261050925e-06, + "loss": 0.1358, + "step": 13407 + }, + { + "epoch": 5.61, + "learning_rate": 4.3116272551679215e-06, + "loss": 0.1249, + "step": 13408 + }, + { + "epoch": 5.61, + "learning_rate": 4.310232947737979e-06, + "loss": 0.1491, + "step": 13409 + }, + { + "epoch": 5.61, + "learning_rate": 4.308838803855341e-06, + "loss": 0.1461, + "step": 13410 + }, + { + "epoch": 5.61, + "learning_rate": 4.307444823560084e-06, + "loss": 0.1283, + "step": 13411 + }, + { + "epoch": 5.62, + "learning_rate": 4.306051006892271e-06, + "loss": 0.1214, + "step": 13412 + }, + { + "epoch": 5.62, + "learning_rate": 4.30465735389197e-06, + "loss": 0.1222, + "step": 13413 + }, + { + "epoch": 5.62, + "learning_rate": 4.303263864599239e-06, + "loss": 0.0874, + "step": 13414 + }, + { + "epoch": 5.62, + "learning_rate": 4.3018705390541316e-06, + "loss": 0.1077, + "step": 13415 + }, + { + "epoch": 5.62, + "learning_rate": 4.3004773772967e-06, + "loss": 0.1062, + "step": 13416 + }, + { + "epoch": 5.62, + "learning_rate": 4.2990843793669925e-06, + "loss": 0.0945, + "step": 13417 + }, + { + "epoch": 5.62, + "learning_rate": 4.297691545305042e-06, + "loss": 0.1373, + "step": 13418 + }, + { + "epoch": 5.62, + "learning_rate": 4.296298875150887e-06, + "loss": 0.1218, + "step": 13419 + }, + { + "epoch": 5.62, + "learning_rate": 4.29490636894456e-06, + "loss": 0.1227, + "step": 13420 + }, + { + "epoch": 5.62, + "learning_rate": 4.293514026726088e-06, + "loss": 0.1127, + "step": 13421 + }, + { + "epoch": 5.62, + "learning_rate": 4.29212184853549e-06, + "loss": 0.1294, + "step": 13422 + }, + { + "epoch": 5.62, + "learning_rate": 4.2907298344127844e-06, + "loss": 0.1136, + "step": 13423 + }, + { + "epoch": 5.62, + "learning_rate": 4.289337984397988e-06, + "loss": 0.0996, + "step": 13424 + }, + { + "epoch": 5.62, + "learning_rate": 4.287946298531099e-06, + "loss": 0.107, + "step": 13425 + }, + { + "epoch": 5.62, + "learning_rate": 4.286554776852125e-06, + "loss": 0.1048, + "step": 13426 + }, + { + "epoch": 5.62, + "learning_rate": 4.285163419401065e-06, + "loss": 0.1187, + "step": 13427 + }, + { + "epoch": 5.62, + "learning_rate": 4.2837722262179105e-06, + "loss": 0.1657, + "step": 13428 + }, + { + "epoch": 5.62, + "learning_rate": 4.282381197342651e-06, + "loss": 0.118, + "step": 13429 + }, + { + "epoch": 5.62, + "learning_rate": 4.280990332815274e-06, + "loss": 0.1607, + "step": 13430 + }, + { + "epoch": 5.62, + "learning_rate": 4.27959963267575e-06, + "loss": 0.151, + "step": 13431 + }, + { + "epoch": 5.62, + "learning_rate": 4.278209096964063e-06, + "loss": 0.1083, + "step": 13432 + }, + { + "epoch": 5.62, + "learning_rate": 4.276818725720176e-06, + "loss": 0.0842, + "step": 13433 + }, + { + "epoch": 5.62, + "learning_rate": 4.275428518984054e-06, + "loss": 0.1159, + "step": 13434 + }, + { + "epoch": 5.62, + "learning_rate": 4.27403847679566e-06, + "loss": 0.1504, + "step": 13435 + }, + { + "epoch": 5.63, + "learning_rate": 4.272648599194948e-06, + "loss": 0.1206, + "step": 13436 + }, + { + "epoch": 5.63, + "learning_rate": 4.27125888622187e-06, + "loss": 0.1466, + "step": 13437 + }, + { + "epoch": 5.63, + "learning_rate": 4.269869337916377e-06, + "loss": 0.1112, + "step": 13438 + }, + { + "epoch": 5.63, + "learning_rate": 4.2684799543183995e-06, + "loss": 0.0942, + "step": 13439 + }, + { + "epoch": 5.63, + "learning_rate": 4.26709073546788e-06, + "loss": 0.1187, + "step": 13440 + }, + { + "epoch": 5.63, + "learning_rate": 4.2657016814047515e-06, + "loss": 0.118, + "step": 13441 + }, + { + "epoch": 5.63, + "learning_rate": 4.264312792168938e-06, + "loss": 0.0935, + "step": 13442 + }, + { + "epoch": 5.63, + "learning_rate": 4.262924067800364e-06, + "loss": 0.1004, + "step": 13443 + }, + { + "epoch": 5.63, + "learning_rate": 4.261535508338951e-06, + "loss": 0.113, + "step": 13444 + }, + { + "epoch": 5.63, + "learning_rate": 4.260147113824603e-06, + "loss": 0.1241, + "step": 13445 + }, + { + "epoch": 5.63, + "learning_rate": 4.2587588842972325e-06, + "loss": 0.131, + "step": 13446 + }, + { + "epoch": 5.63, + "learning_rate": 4.257370819796745e-06, + "loss": 0.1177, + "step": 13447 + }, + { + "epoch": 5.63, + "learning_rate": 4.255982920363035e-06, + "loss": 0.1428, + "step": 13448 + }, + { + "epoch": 5.63, + "learning_rate": 4.254595186036001e-06, + "loss": 0.1097, + "step": 13449 + }, + { + "epoch": 5.63, + "learning_rate": 4.253207616855528e-06, + "loss": 0.1172, + "step": 13450 + }, + { + "epoch": 5.63, + "learning_rate": 4.251820212861507e-06, + "loss": 0.1338, + "step": 13451 + }, + { + "epoch": 5.63, + "learning_rate": 4.2504329740938084e-06, + "loss": 0.1281, + "step": 13452 + }, + { + "epoch": 5.63, + "learning_rate": 4.249045900592318e-06, + "loss": 0.1366, + "step": 13453 + }, + { + "epoch": 5.63, + "learning_rate": 4.2476589923968946e-06, + "loss": 0.0745, + "step": 13454 + }, + { + "epoch": 5.63, + "learning_rate": 4.2462722495474085e-06, + "loss": 0.1222, + "step": 13455 + }, + { + "epoch": 5.63, + "learning_rate": 4.244885672083722e-06, + "loss": 0.1208, + "step": 13456 + }, + { + "epoch": 5.63, + "learning_rate": 4.243499260045689e-06, + "loss": 0.1247, + "step": 13457 + }, + { + "epoch": 5.63, + "learning_rate": 4.242113013473166e-06, + "loss": 0.1226, + "step": 13458 + }, + { + "epoch": 5.63, + "learning_rate": 4.240726932405991e-06, + "loss": 0.1737, + "step": 13459 + }, + { + "epoch": 5.64, + "learning_rate": 4.239341016884008e-06, + "loss": 0.1072, + "step": 13460 + }, + { + "epoch": 5.64, + "learning_rate": 4.237955266947058e-06, + "loss": 0.1087, + "step": 13461 + }, + { + "epoch": 5.64, + "learning_rate": 4.23656968263497e-06, + "loss": 0.137, + "step": 13462 + }, + { + "epoch": 5.64, + "learning_rate": 4.235184263987572e-06, + "loss": 0.1212, + "step": 13463 + }, + { + "epoch": 5.64, + "learning_rate": 4.233799011044687e-06, + "loss": 0.1699, + "step": 13464 + }, + { + "epoch": 5.64, + "learning_rate": 4.232413923846135e-06, + "loss": 0.1369, + "step": 13465 + }, + { + "epoch": 5.64, + "learning_rate": 4.231029002431723e-06, + "loss": 0.0873, + "step": 13466 + }, + { + "epoch": 5.64, + "learning_rate": 4.2296442468412646e-06, + "loss": 0.1319, + "step": 13467 + }, + { + "epoch": 5.64, + "learning_rate": 4.2282596571145615e-06, + "loss": 0.1551, + "step": 13468 + }, + { + "epoch": 5.64, + "learning_rate": 4.226875233291412e-06, + "loss": 0.1186, + "step": 13469 + }, + { + "epoch": 5.64, + "learning_rate": 4.225490975411611e-06, + "loss": 0.1651, + "step": 13470 + }, + { + "epoch": 5.64, + "learning_rate": 4.224106883514952e-06, + "loss": 0.1228, + "step": 13471 + }, + { + "epoch": 5.64, + "learning_rate": 4.222722957641212e-06, + "loss": 0.1283, + "step": 13472 + }, + { + "epoch": 5.64, + "learning_rate": 4.221339197830173e-06, + "loss": 0.1121, + "step": 13473 + }, + { + "epoch": 5.64, + "learning_rate": 4.219955604121611e-06, + "loss": 0.1387, + "step": 13474 + }, + { + "epoch": 5.64, + "learning_rate": 4.2185721765553e-06, + "loss": 0.1105, + "step": 13475 + }, + { + "epoch": 5.64, + "learning_rate": 4.217188915170996e-06, + "loss": 0.12, + "step": 13476 + }, + { + "epoch": 5.64, + "learning_rate": 4.215805820008468e-06, + "loss": 0.1138, + "step": 13477 + }, + { + "epoch": 5.64, + "learning_rate": 4.214422891107467e-06, + "loss": 0.1054, + "step": 13478 + }, + { + "epoch": 5.64, + "learning_rate": 4.2130401285077505e-06, + "loss": 0.1142, + "step": 13479 + }, + { + "epoch": 5.64, + "learning_rate": 4.211657532249056e-06, + "loss": 0.1037, + "step": 13480 + }, + { + "epoch": 5.64, + "learning_rate": 4.21027510237113e-06, + "loss": 0.1261, + "step": 13481 + }, + { + "epoch": 5.64, + "learning_rate": 4.208892838913707e-06, + "loss": 0.1181, + "step": 13482 + }, + { + "epoch": 5.64, + "learning_rate": 4.207510741916522e-06, + "loss": 0.1007, + "step": 13483 + }, + { + "epoch": 5.65, + "learning_rate": 4.206128811419299e-06, + "loss": 0.1343, + "step": 13484 + }, + { + "epoch": 5.65, + "learning_rate": 4.204747047461767e-06, + "loss": 0.1377, + "step": 13485 + }, + { + "epoch": 5.65, + "learning_rate": 4.2033654500836344e-06, + "loss": 0.1334, + "step": 13486 + }, + { + "epoch": 5.65, + "learning_rate": 4.201984019324617e-06, + "loss": 0.1247, + "step": 13487 + }, + { + "epoch": 5.65, + "learning_rate": 4.200602755224426e-06, + "loss": 0.1618, + "step": 13488 + }, + { + "epoch": 5.65, + "learning_rate": 4.199221657822761e-06, + "loss": 0.1531, + "step": 13489 + }, + { + "epoch": 5.65, + "learning_rate": 4.1978407271593226e-06, + "loss": 0.1097, + "step": 13490 + }, + { + "epoch": 5.65, + "learning_rate": 4.1964599632738035e-06, + "loss": 0.1293, + "step": 13491 + }, + { + "epoch": 5.65, + "learning_rate": 4.1950793662058965e-06, + "loss": 0.1205, + "step": 13492 + }, + { + "epoch": 5.65, + "learning_rate": 4.1936989359952775e-06, + "loss": 0.1221, + "step": 13493 + }, + { + "epoch": 5.65, + "learning_rate": 4.192318672681631e-06, + "loss": 0.1379, + "step": 13494 + }, + { + "epoch": 5.65, + "learning_rate": 4.1909385763046305e-06, + "loss": 0.1072, + "step": 13495 + }, + { + "epoch": 5.65, + "learning_rate": 4.1895586469039466e-06, + "loss": 0.0995, + "step": 13496 + }, + { + "epoch": 5.65, + "learning_rate": 4.188178884519241e-06, + "loss": 0.1429, + "step": 13497 + }, + { + "epoch": 5.65, + "learning_rate": 4.186799289190182e-06, + "loss": 0.1256, + "step": 13498 + }, + { + "epoch": 5.65, + "learning_rate": 4.185419860956418e-06, + "loss": 0.1429, + "step": 13499 + }, + { + "epoch": 5.65, + "learning_rate": 4.184040599857596e-06, + "loss": 0.1564, + "step": 13500 + }, + { + "epoch": 5.65, + "learning_rate": 4.182661505933366e-06, + "loss": 0.1095, + "step": 13501 + }, + { + "epoch": 5.65, + "learning_rate": 4.18128257922337e-06, + "loss": 0.1098, + "step": 13502 + }, + { + "epoch": 5.65, + "learning_rate": 4.179903819767242e-06, + "loss": 0.1369, + "step": 13503 + }, + { + "epoch": 5.65, + "learning_rate": 4.178525227604615e-06, + "loss": 0.1407, + "step": 13504 + }, + { + "epoch": 5.65, + "learning_rate": 4.177146802775114e-06, + "loss": 0.1111, + "step": 13505 + }, + { + "epoch": 5.65, + "learning_rate": 4.175768545318364e-06, + "loss": 0.1186, + "step": 13506 + }, + { + "epoch": 5.66, + "learning_rate": 4.174390455273975e-06, + "loss": 0.1016, + "step": 13507 + }, + { + "epoch": 5.66, + "learning_rate": 4.173012532681564e-06, + "loss": 0.1468, + "step": 13508 + }, + { + "epoch": 5.66, + "learning_rate": 4.171634777580736e-06, + "loss": 0.1381, + "step": 13509 + }, + { + "epoch": 5.66, + "learning_rate": 4.170257190011096e-06, + "loss": 0.1473, + "step": 13510 + }, + { + "epoch": 5.66, + "learning_rate": 4.168879770012237e-06, + "loss": 0.0879, + "step": 13511 + }, + { + "epoch": 5.66, + "learning_rate": 4.16750251762376e-06, + "loss": 0.0742, + "step": 13512 + }, + { + "epoch": 5.66, + "learning_rate": 4.166125432885244e-06, + "loss": 0.1145, + "step": 13513 + }, + { + "epoch": 5.66, + "learning_rate": 4.164748515836276e-06, + "loss": 0.0923, + "step": 13514 + }, + { + "epoch": 5.66, + "learning_rate": 4.163371766516432e-06, + "loss": 0.1548, + "step": 13515 + }, + { + "epoch": 5.66, + "learning_rate": 4.1619951849652885e-06, + "loss": 0.1205, + "step": 13516 + }, + { + "epoch": 5.66, + "learning_rate": 4.160618771222412e-06, + "loss": 0.1102, + "step": 13517 + }, + { + "epoch": 5.66, + "learning_rate": 4.159242525327371e-06, + "loss": 0.112, + "step": 13518 + }, + { + "epoch": 5.66, + "learning_rate": 4.157866447319718e-06, + "loss": 0.1344, + "step": 13519 + }, + { + "epoch": 5.66, + "learning_rate": 4.15649053723901e-06, + "loss": 0.1181, + "step": 13520 + }, + { + "epoch": 5.66, + "learning_rate": 4.155114795124798e-06, + "loss": 0.1008, + "step": 13521 + }, + { + "epoch": 5.66, + "learning_rate": 4.153739221016622e-06, + "loss": 0.0864, + "step": 13522 + }, + { + "epoch": 5.66, + "learning_rate": 4.152363814954024e-06, + "loss": 0.1235, + "step": 13523 + }, + { + "epoch": 5.66, + "learning_rate": 4.150988576976539e-06, + "loss": 0.1005, + "step": 13524 + }, + { + "epoch": 5.66, + "learning_rate": 4.1496135071236965e-06, + "loss": 0.075, + "step": 13525 + }, + { + "epoch": 5.66, + "learning_rate": 4.1482386054350265e-06, + "loss": 0.1, + "step": 13526 + }, + { + "epoch": 5.66, + "learning_rate": 4.146863871950041e-06, + "loss": 0.0988, + "step": 13527 + }, + { + "epoch": 5.66, + "learning_rate": 4.1454893067082605e-06, + "loss": 0.1228, + "step": 13528 + }, + { + "epoch": 5.66, + "learning_rate": 4.144114909749193e-06, + "loss": 0.127, + "step": 13529 + }, + { + "epoch": 5.66, + "learning_rate": 4.142740681112347e-06, + "loss": 0.1128, + "step": 13530 + }, + { + "epoch": 5.67, + "learning_rate": 4.141366620837223e-06, + "loss": 0.1203, + "step": 13531 + }, + { + "epoch": 5.67, + "learning_rate": 4.139992728963315e-06, + "loss": 0.1196, + "step": 13532 + }, + { + "epoch": 5.67, + "learning_rate": 4.138619005530121e-06, + "loss": 0.1411, + "step": 13533 + }, + { + "epoch": 5.67, + "learning_rate": 4.1372454505771186e-06, + "loss": 0.1377, + "step": 13534 + }, + { + "epoch": 5.67, + "learning_rate": 4.135872064143792e-06, + "loss": 0.1149, + "step": 13535 + }, + { + "epoch": 5.67, + "learning_rate": 4.134498846269621e-06, + "loss": 0.1183, + "step": 13536 + }, + { + "epoch": 5.67, + "learning_rate": 4.133125796994074e-06, + "loss": 0.134, + "step": 13537 + }, + { + "epoch": 5.67, + "learning_rate": 4.131752916356622e-06, + "loss": 0.1001, + "step": 13538 + }, + { + "epoch": 5.67, + "learning_rate": 4.130380204396727e-06, + "loss": 0.1182, + "step": 13539 + }, + { + "epoch": 5.67, + "learning_rate": 4.129007661153841e-06, + "loss": 0.1349, + "step": 13540 + }, + { + "epoch": 5.67, + "learning_rate": 4.12763528666742e-06, + "loss": 0.1489, + "step": 13541 + }, + { + "epoch": 5.67, + "learning_rate": 4.126263080976911e-06, + "loss": 0.1143, + "step": 13542 + }, + { + "epoch": 5.67, + "learning_rate": 4.124891044121761e-06, + "loss": 0.1215, + "step": 13543 + }, + { + "epoch": 5.67, + "learning_rate": 4.1235191761414025e-06, + "loss": 0.0732, + "step": 13544 + }, + { + "epoch": 5.67, + "learning_rate": 4.12214747707527e-06, + "loss": 0.1264, + "step": 13545 + }, + { + "epoch": 5.67, + "learning_rate": 4.120775946962792e-06, + "loss": 0.1171, + "step": 13546 + }, + { + "epoch": 5.67, + "learning_rate": 4.1194045858433965e-06, + "loss": 0.1367, + "step": 13547 + }, + { + "epoch": 5.67, + "learning_rate": 4.118033393756493e-06, + "loss": 0.1333, + "step": 13548 + }, + { + "epoch": 5.67, + "learning_rate": 4.116662370741501e-06, + "loss": 0.0928, + "step": 13549 + }, + { + "epoch": 5.67, + "learning_rate": 4.115291516837827e-06, + "loss": 0.1202, + "step": 13550 + }, + { + "epoch": 5.67, + "learning_rate": 4.113920832084878e-06, + "loss": 0.1247, + "step": 13551 + }, + { + "epoch": 5.67, + "learning_rate": 4.11255031652205e-06, + "loss": 0.1122, + "step": 13552 + }, + { + "epoch": 5.67, + "learning_rate": 4.111179970188745e-06, + "loss": 0.1025, + "step": 13553 + }, + { + "epoch": 5.67, + "learning_rate": 4.10980979312434e-06, + "loss": 0.1193, + "step": 13554 + }, + { + "epoch": 5.68, + "learning_rate": 4.108439785368226e-06, + "loss": 0.1566, + "step": 13555 + }, + { + "epoch": 5.68, + "learning_rate": 4.107069946959783e-06, + "loss": 0.1232, + "step": 13556 + }, + { + "epoch": 5.68, + "learning_rate": 4.105700277938385e-06, + "loss": 0.1144, + "step": 13557 + }, + { + "epoch": 5.68, + "learning_rate": 4.1043307783434e-06, + "loss": 0.1173, + "step": 13558 + }, + { + "epoch": 5.68, + "learning_rate": 4.102961448214201e-06, + "loss": 0.1286, + "step": 13559 + }, + { + "epoch": 5.68, + "learning_rate": 4.101592287590138e-06, + "loss": 0.1327, + "step": 13560 + }, + { + "epoch": 5.68, + "learning_rate": 4.100223296510571e-06, + "loss": 0.1254, + "step": 13561 + }, + { + "epoch": 5.68, + "learning_rate": 4.098854475014849e-06, + "loss": 0.1013, + "step": 13562 + }, + { + "epoch": 5.68, + "learning_rate": 4.097485823142319e-06, + "loss": 0.1239, + "step": 13563 + }, + { + "epoch": 5.68, + "learning_rate": 4.096117340932322e-06, + "loss": 0.1021, + "step": 13564 + }, + { + "epoch": 5.68, + "learning_rate": 4.094749028424192e-06, + "loss": 0.1207, + "step": 13565 + }, + { + "epoch": 5.68, + "learning_rate": 4.093380885657264e-06, + "loss": 0.1143, + "step": 13566 + }, + { + "epoch": 5.68, + "learning_rate": 4.092012912670863e-06, + "loss": 0.1372, + "step": 13567 + }, + { + "epoch": 5.68, + "learning_rate": 4.090645109504303e-06, + "loss": 0.0726, + "step": 13568 + }, + { + "epoch": 5.68, + "learning_rate": 4.0892774761969055e-06, + "loss": 0.1092, + "step": 13569 + }, + { + "epoch": 5.68, + "learning_rate": 4.087910012787983e-06, + "loss": 0.1253, + "step": 13570 + }, + { + "epoch": 5.68, + "learning_rate": 4.08654271931684e-06, + "loss": 0.1181, + "step": 13571 + }, + { + "epoch": 5.68, + "learning_rate": 4.085175595822779e-06, + "loss": 0.1061, + "step": 13572 + }, + { + "epoch": 5.68, + "learning_rate": 4.083808642345098e-06, + "loss": 0.1261, + "step": 13573 + }, + { + "epoch": 5.68, + "learning_rate": 4.082441858923088e-06, + "loss": 0.0955, + "step": 13574 + }, + { + "epoch": 5.68, + "learning_rate": 4.0810752455960356e-06, + "loss": 0.1048, + "step": 13575 + }, + { + "epoch": 5.68, + "learning_rate": 4.07970880240322e-06, + "loss": 0.069, + "step": 13576 + }, + { + "epoch": 5.68, + "learning_rate": 4.0783425293839245e-06, + "loss": 0.1157, + "step": 13577 + }, + { + "epoch": 5.68, + "learning_rate": 4.0769764265774155e-06, + "loss": 0.111, + "step": 13578 + }, + { + "epoch": 5.69, + "learning_rate": 4.0756104940229645e-06, + "loss": 0.1169, + "step": 13579 + }, + { + "epoch": 5.69, + "learning_rate": 4.074244731759836e-06, + "loss": 0.1089, + "step": 13580 + }, + { + "epoch": 5.69, + "learning_rate": 4.07287913982728e-06, + "loss": 0.1629, + "step": 13581 + }, + { + "epoch": 5.69, + "learning_rate": 4.071513718264554e-06, + "loss": 0.1112, + "step": 13582 + }, + { + "epoch": 5.69, + "learning_rate": 4.070148467110906e-06, + "loss": 0.113, + "step": 13583 + }, + { + "epoch": 5.69, + "learning_rate": 4.068783386405577e-06, + "loss": 0.145, + "step": 13584 + }, + { + "epoch": 5.69, + "learning_rate": 4.067418476187807e-06, + "loss": 0.1243, + "step": 13585 + }, + { + "epoch": 5.69, + "learning_rate": 4.066053736496832e-06, + "loss": 0.0941, + "step": 13586 + }, + { + "epoch": 5.69, + "learning_rate": 4.0646891673718725e-06, + "loss": 0.1226, + "step": 13587 + }, + { + "epoch": 5.69, + "learning_rate": 4.063324768852157e-06, + "loss": 0.1108, + "step": 13588 + }, + { + "epoch": 5.69, + "learning_rate": 4.061960540976906e-06, + "loss": 0.1098, + "step": 13589 + }, + { + "epoch": 5.69, + "learning_rate": 4.060596483785326e-06, + "loss": 0.152, + "step": 13590 + }, + { + "epoch": 5.69, + "learning_rate": 4.059232597316629e-06, + "loss": 0.1005, + "step": 13591 + }, + { + "epoch": 5.69, + "learning_rate": 4.057868881610021e-06, + "loss": 0.1304, + "step": 13592 + }, + { + "epoch": 5.69, + "learning_rate": 4.056505336704698e-06, + "loss": 0.0902, + "step": 13593 + }, + { + "epoch": 5.69, + "learning_rate": 4.055141962639858e-06, + "loss": 0.1515, + "step": 13594 + }, + { + "epoch": 5.69, + "learning_rate": 4.053778759454684e-06, + "loss": 0.0776, + "step": 13595 + }, + { + "epoch": 5.69, + "learning_rate": 4.0524157271883635e-06, + "loss": 0.1422, + "step": 13596 + }, + { + "epoch": 5.69, + "learning_rate": 4.051052865880074e-06, + "loss": 0.124, + "step": 13597 + }, + { + "epoch": 5.69, + "learning_rate": 4.049690175568992e-06, + "loss": 0.093, + "step": 13598 + }, + { + "epoch": 5.69, + "learning_rate": 4.048327656294285e-06, + "loss": 0.09, + "step": 13599 + }, + { + "epoch": 5.69, + "learning_rate": 4.046965308095122e-06, + "loss": 0.1262, + "step": 13600 + }, + { + "epoch": 5.69, + "learning_rate": 4.045603131010655e-06, + "loss": 0.1213, + "step": 13601 + }, + { + "epoch": 5.69, + "learning_rate": 4.044241125080041e-06, + "loss": 0.1101, + "step": 13602 + }, + { + "epoch": 5.7, + "learning_rate": 4.042879290342432e-06, + "loss": 0.1136, + "step": 13603 + }, + { + "epoch": 5.7, + "learning_rate": 4.0415176268369714e-06, + "loss": 0.1203, + "step": 13604 + }, + { + "epoch": 5.7, + "learning_rate": 4.0401561346028e-06, + "loss": 0.084, + "step": 13605 + }, + { + "epoch": 5.7, + "learning_rate": 4.03879481367905e-06, + "loss": 0.1028, + "step": 13606 + }, + { + "epoch": 5.7, + "learning_rate": 4.037433664104857e-06, + "loss": 0.1255, + "step": 13607 + }, + { + "epoch": 5.7, + "learning_rate": 4.03607268591934e-06, + "loss": 0.1055, + "step": 13608 + }, + { + "epoch": 5.7, + "learning_rate": 4.03471187916162e-06, + "loss": 0.1279, + "step": 13609 + }, + { + "epoch": 5.7, + "learning_rate": 4.033351243870815e-06, + "loss": 0.1281, + "step": 13610 + }, + { + "epoch": 5.7, + "learning_rate": 4.031990780086037e-06, + "loss": 0.1073, + "step": 13611 + }, + { + "epoch": 5.7, + "learning_rate": 4.0306304878463834e-06, + "loss": 0.1445, + "step": 13612 + }, + { + "epoch": 5.7, + "learning_rate": 4.0292703671909605e-06, + "loss": 0.1192, + "step": 13613 + }, + { + "epoch": 5.7, + "learning_rate": 4.027910418158861e-06, + "loss": 0.1246, + "step": 13614 + }, + { + "epoch": 5.7, + "learning_rate": 4.026550640789183e-06, + "loss": 0.1184, + "step": 13615 + }, + { + "epoch": 5.7, + "learning_rate": 4.025191035121001e-06, + "loss": 0.1196, + "step": 13616 + }, + { + "epoch": 5.7, + "learning_rate": 4.023831601193401e-06, + "loss": 0.1371, + "step": 13617 + }, + { + "epoch": 5.7, + "learning_rate": 4.0224723390454575e-06, + "loss": 0.0976, + "step": 13618 + }, + { + "epoch": 5.7, + "learning_rate": 4.021113248716243e-06, + "loss": 0.0785, + "step": 13619 + }, + { + "epoch": 5.7, + "learning_rate": 4.019754330244822e-06, + "loss": 0.1165, + "step": 13620 + }, + { + "epoch": 5.7, + "learning_rate": 4.0183955836702615e-06, + "loss": 0.1068, + "step": 13621 + }, + { + "epoch": 5.7, + "learning_rate": 4.017037009031607e-06, + "loss": 0.125, + "step": 13622 + }, + { + "epoch": 5.7, + "learning_rate": 4.015678606367916e-06, + "loss": 0.1584, + "step": 13623 + }, + { + "epoch": 5.7, + "learning_rate": 4.01432037571823e-06, + "loss": 0.1086, + "step": 13624 + }, + { + "epoch": 5.7, + "learning_rate": 4.012962317121596e-06, + "loss": 0.1481, + "step": 13625 + }, + { + "epoch": 5.7, + "learning_rate": 4.011604430617046e-06, + "loss": 0.1074, + "step": 13626 + }, + { + "epoch": 5.71, + "learning_rate": 4.010246716243617e-06, + "loss": 0.1154, + "step": 13627 + }, + { + "epoch": 5.71, + "learning_rate": 4.008889174040327e-06, + "loss": 0.1414, + "step": 13628 + }, + { + "epoch": 5.71, + "learning_rate": 4.0075318040462005e-06, + "loss": 0.1165, + "step": 13629 + }, + { + "epoch": 5.71, + "learning_rate": 4.006174606300255e-06, + "loss": 0.0938, + "step": 13630 + }, + { + "epoch": 5.71, + "learning_rate": 4.004817580841503e-06, + "loss": 0.1314, + "step": 13631 + }, + { + "epoch": 5.71, + "learning_rate": 4.003460727708947e-06, + "loss": 0.1182, + "step": 13632 + }, + { + "epoch": 5.71, + "learning_rate": 4.002104046941594e-06, + "loss": 0.1373, + "step": 13633 + }, + { + "epoch": 5.71, + "learning_rate": 4.00074753857844e-06, + "loss": 0.0948, + "step": 13634 + }, + { + "epoch": 5.71, + "learning_rate": 3.9993912026584735e-06, + "loss": 0.1131, + "step": 13635 + }, + { + "epoch": 5.71, + "learning_rate": 3.99803503922068e-06, + "loss": 0.1216, + "step": 13636 + }, + { + "epoch": 5.71, + "learning_rate": 3.996679048304042e-06, + "loss": 0.0913, + "step": 13637 + }, + { + "epoch": 5.71, + "learning_rate": 3.9953232299475385e-06, + "loss": 0.1136, + "step": 13638 + }, + { + "epoch": 5.71, + "learning_rate": 3.993967584190141e-06, + "loss": 0.1164, + "step": 13639 + }, + { + "epoch": 5.71, + "learning_rate": 3.992612111070814e-06, + "loss": 0.0847, + "step": 13640 + }, + { + "epoch": 5.71, + "learning_rate": 3.991256810628525e-06, + "loss": 0.1079, + "step": 13641 + }, + { + "epoch": 5.71, + "learning_rate": 3.989901682902223e-06, + "loss": 0.1372, + "step": 13642 + }, + { + "epoch": 5.71, + "learning_rate": 3.9885467279308644e-06, + "loss": 0.1175, + "step": 13643 + }, + { + "epoch": 5.71, + "learning_rate": 3.987191945753396e-06, + "loss": 0.1287, + "step": 13644 + }, + { + "epoch": 5.71, + "learning_rate": 3.9858373364087586e-06, + "loss": 0.0921, + "step": 13645 + }, + { + "epoch": 5.71, + "learning_rate": 3.98448289993589e-06, + "loss": 0.1228, + "step": 13646 + }, + { + "epoch": 5.71, + "learning_rate": 3.983128636373723e-06, + "loss": 0.1311, + "step": 13647 + }, + { + "epoch": 5.71, + "learning_rate": 3.981774545761188e-06, + "loss": 0.1644, + "step": 13648 + }, + { + "epoch": 5.71, + "learning_rate": 3.980420628137199e-06, + "loss": 0.1409, + "step": 13649 + }, + { + "epoch": 5.71, + "learning_rate": 3.979066883540677e-06, + "loss": 0.115, + "step": 13650 + }, + { + "epoch": 5.72, + "learning_rate": 3.977713312010535e-06, + "loss": 0.1419, + "step": 13651 + }, + { + "epoch": 5.72, + "learning_rate": 3.97635991358568e-06, + "loss": 0.1248, + "step": 13652 + }, + { + "epoch": 5.72, + "learning_rate": 3.975006688305015e-06, + "loss": 0.0984, + "step": 13653 + }, + { + "epoch": 5.72, + "learning_rate": 3.973653636207437e-06, + "loss": 0.0916, + "step": 13654 + }, + { + "epoch": 5.72, + "learning_rate": 3.972300757331836e-06, + "loss": 0.1347, + "step": 13655 + }, + { + "epoch": 5.72, + "learning_rate": 3.970948051717104e-06, + "loss": 0.1198, + "step": 13656 + }, + { + "epoch": 5.72, + "learning_rate": 3.969595519402117e-06, + "loss": 0.1114, + "step": 13657 + }, + { + "epoch": 5.72, + "learning_rate": 3.9682431604257554e-06, + "loss": 0.111, + "step": 13658 + }, + { + "epoch": 5.72, + "learning_rate": 3.966890974826891e-06, + "loss": 0.0948, + "step": 13659 + }, + { + "epoch": 5.72, + "learning_rate": 3.965538962644392e-06, + "loss": 0.0953, + "step": 13660 + }, + { + "epoch": 5.72, + "learning_rate": 3.964187123917122e-06, + "loss": 0.1222, + "step": 13661 + }, + { + "epoch": 5.72, + "learning_rate": 3.962835458683939e-06, + "loss": 0.1081, + "step": 13662 + }, + { + "epoch": 5.72, + "learning_rate": 3.9614839669836905e-06, + "loss": 0.0928, + "step": 13663 + }, + { + "epoch": 5.72, + "learning_rate": 3.960132648855226e-06, + "loss": 0.0574, + "step": 13664 + }, + { + "epoch": 5.72, + "learning_rate": 3.95878150433739e-06, + "loss": 0.1443, + "step": 13665 + }, + { + "epoch": 5.72, + "learning_rate": 3.9574305334690175e-06, + "loss": 0.1078, + "step": 13666 + }, + { + "epoch": 5.72, + "learning_rate": 3.9560797362889436e-06, + "loss": 0.1208, + "step": 13667 + }, + { + "epoch": 5.72, + "learning_rate": 3.954729112835998e-06, + "loss": 0.1301, + "step": 13668 + }, + { + "epoch": 5.72, + "learning_rate": 3.953378663148994e-06, + "loss": 0.1397, + "step": 13669 + }, + { + "epoch": 5.72, + "learning_rate": 3.952028387266757e-06, + "loss": 0.1295, + "step": 13670 + }, + { + "epoch": 5.72, + "learning_rate": 3.950678285228097e-06, + "loss": 0.1382, + "step": 13671 + }, + { + "epoch": 5.72, + "learning_rate": 3.94932835707182e-06, + "loss": 0.1159, + "step": 13672 + }, + { + "epoch": 5.72, + "learning_rate": 3.947978602836731e-06, + "loss": 0.1644, + "step": 13673 + }, + { + "epoch": 5.72, + "learning_rate": 3.946629022561627e-06, + "loss": 0.1093, + "step": 13674 + }, + { + "epoch": 5.73, + "learning_rate": 3.945279616285303e-06, + "loss": 0.117, + "step": 13675 + }, + { + "epoch": 5.73, + "learning_rate": 3.943930384046541e-06, + "loss": 0.1518, + "step": 13676 + }, + { + "epoch": 5.73, + "learning_rate": 3.942581325884124e-06, + "loss": 0.0997, + "step": 13677 + }, + { + "epoch": 5.73, + "learning_rate": 3.941232441836833e-06, + "loss": 0.1424, + "step": 13678 + }, + { + "epoch": 5.73, + "learning_rate": 3.939883731943443e-06, + "loss": 0.1216, + "step": 13679 + }, + { + "epoch": 5.73, + "learning_rate": 3.938535196242713e-06, + "loss": 0.1252, + "step": 13680 + }, + { + "epoch": 5.73, + "learning_rate": 3.93718683477341e-06, + "loss": 0.1371, + "step": 13681 + }, + { + "epoch": 5.73, + "learning_rate": 3.935838647574295e-06, + "loss": 0.098, + "step": 13682 + }, + { + "epoch": 5.73, + "learning_rate": 3.934490634684114e-06, + "loss": 0.1125, + "step": 13683 + }, + { + "epoch": 5.73, + "learning_rate": 3.933142796141616e-06, + "loss": 0.0992, + "step": 13684 + }, + { + "epoch": 5.73, + "learning_rate": 3.931795131985545e-06, + "loss": 0.149, + "step": 13685 + }, + { + "epoch": 5.73, + "learning_rate": 3.930447642254639e-06, + "loss": 0.1305, + "step": 13686 + }, + { + "epoch": 5.73, + "learning_rate": 3.929100326987629e-06, + "loss": 0.1338, + "step": 13687 + }, + { + "epoch": 5.73, + "learning_rate": 3.927753186223243e-06, + "loss": 0.1654, + "step": 13688 + }, + { + "epoch": 5.73, + "learning_rate": 3.926406220000206e-06, + "loss": 0.1465, + "step": 13689 + }, + { + "epoch": 5.73, + "learning_rate": 3.925059428357231e-06, + "loss": 0.1099, + "step": 13690 + }, + { + "epoch": 5.73, + "learning_rate": 3.923712811333031e-06, + "loss": 0.1108, + "step": 13691 + }, + { + "epoch": 5.73, + "learning_rate": 3.922366368966313e-06, + "loss": 0.1221, + "step": 13692 + }, + { + "epoch": 5.73, + "learning_rate": 3.9210201012957816e-06, + "loss": 0.1164, + "step": 13693 + }, + { + "epoch": 5.73, + "learning_rate": 3.919674008360133e-06, + "loss": 0.1125, + "step": 13694 + }, + { + "epoch": 5.73, + "learning_rate": 3.918328090198063e-06, + "loss": 0.1316, + "step": 13695 + }, + { + "epoch": 5.73, + "learning_rate": 3.91698234684825e-06, + "loss": 0.0516, + "step": 13696 + }, + { + "epoch": 5.73, + "learning_rate": 3.915636778349385e-06, + "loss": 0.1248, + "step": 13697 + }, + { + "epoch": 5.73, + "learning_rate": 3.914291384740139e-06, + "loss": 0.1102, + "step": 13698 + }, + { + "epoch": 5.74, + "learning_rate": 3.9129461660591875e-06, + "loss": 0.1044, + "step": 13699 + }, + { + "epoch": 5.74, + "learning_rate": 3.911601122345197e-06, + "loss": 0.0924, + "step": 13700 + }, + { + "epoch": 5.74, + "learning_rate": 3.910256253636832e-06, + "loss": 0.1286, + "step": 13701 + }, + { + "epoch": 5.74, + "learning_rate": 3.9089115599727446e-06, + "loss": 0.135, + "step": 13702 + }, + { + "epoch": 5.74, + "learning_rate": 3.9075670413915925e-06, + "loss": 0.1467, + "step": 13703 + }, + { + "epoch": 5.74, + "learning_rate": 3.906222697932016e-06, + "loss": 0.0969, + "step": 13704 + }, + { + "epoch": 5.74, + "learning_rate": 3.90487852963266e-06, + "loss": 0.1121, + "step": 13705 + }, + { + "epoch": 5.74, + "learning_rate": 3.903534536532163e-06, + "loss": 0.1398, + "step": 13706 + }, + { + "epoch": 5.74, + "learning_rate": 3.902190718669155e-06, + "loss": 0.1083, + "step": 13707 + }, + { + "epoch": 5.74, + "learning_rate": 3.900847076082264e-06, + "loss": 0.1356, + "step": 13708 + }, + { + "epoch": 5.74, + "learning_rate": 3.899503608810114e-06, + "loss": 0.1313, + "step": 13709 + }, + { + "epoch": 5.74, + "learning_rate": 3.898160316891317e-06, + "loss": 0.0883, + "step": 13710 + }, + { + "epoch": 5.74, + "learning_rate": 3.896817200364488e-06, + "loss": 0.1363, + "step": 13711 + }, + { + "epoch": 5.74, + "learning_rate": 3.895474259268232e-06, + "loss": 0.0811, + "step": 13712 + }, + { + "epoch": 5.74, + "learning_rate": 3.89413149364115e-06, + "loss": 0.1213, + "step": 13713 + }, + { + "epoch": 5.74, + "learning_rate": 3.89278890352184e-06, + "loss": 0.1138, + "step": 13714 + }, + { + "epoch": 5.74, + "learning_rate": 3.891446488948894e-06, + "loss": 0.1276, + "step": 13715 + }, + { + "epoch": 5.74, + "learning_rate": 3.890104249960903e-06, + "loss": 0.0994, + "step": 13716 + }, + { + "epoch": 5.74, + "learning_rate": 3.888762186596439e-06, + "loss": 0.1112, + "step": 13717 + }, + { + "epoch": 5.74, + "learning_rate": 3.887420298894082e-06, + "loss": 0.1057, + "step": 13718 + }, + { + "epoch": 5.74, + "learning_rate": 3.886078586892406e-06, + "loss": 0.1357, + "step": 13719 + }, + { + "epoch": 5.74, + "learning_rate": 3.884737050629975e-06, + "loss": 0.136, + "step": 13720 + }, + { + "epoch": 5.74, + "learning_rate": 3.883395690145351e-06, + "loss": 0.1432, + "step": 13721 + }, + { + "epoch": 5.75, + "learning_rate": 3.882054505477092e-06, + "loss": 0.1476, + "step": 13722 + }, + { + "epoch": 5.75, + "learning_rate": 3.880713496663745e-06, + "loss": 0.1501, + "step": 13723 + }, + { + "epoch": 5.75, + "learning_rate": 3.879372663743862e-06, + "loss": 0.1247, + "step": 13724 + }, + { + "epoch": 5.75, + "learning_rate": 3.878032006755977e-06, + "loss": 0.1415, + "step": 13725 + }, + { + "epoch": 5.75, + "learning_rate": 3.87669152573863e-06, + "loss": 0.123, + "step": 13726 + }, + { + "epoch": 5.75, + "learning_rate": 3.87535122073035e-06, + "loss": 0.1068, + "step": 13727 + }, + { + "epoch": 5.75, + "learning_rate": 3.874011091769665e-06, + "loss": 0.1255, + "step": 13728 + }, + { + "epoch": 5.75, + "learning_rate": 3.872671138895095e-06, + "loss": 0.1134, + "step": 13729 + }, + { + "epoch": 5.75, + "learning_rate": 3.871331362145158e-06, + "loss": 0.0983, + "step": 13730 + }, + { + "epoch": 5.75, + "learning_rate": 3.869991761558362e-06, + "loss": 0.1482, + "step": 13731 + }, + { + "epoch": 5.75, + "learning_rate": 3.868652337173211e-06, + "loss": 0.132, + "step": 13732 + }, + { + "epoch": 5.75, + "learning_rate": 3.867313089028208e-06, + "loss": 0.1277, + "step": 13733 + }, + { + "epoch": 5.75, + "learning_rate": 3.8659740171618485e-06, + "loss": 0.1152, + "step": 13734 + }, + { + "epoch": 5.75, + "learning_rate": 3.864635121612622e-06, + "loss": 0.0775, + "step": 13735 + }, + { + "epoch": 5.75, + "learning_rate": 3.863296402419019e-06, + "loss": 0.1493, + "step": 13736 + }, + { + "epoch": 5.75, + "learning_rate": 3.861957859619511e-06, + "loss": 0.125, + "step": 13737 + }, + { + "epoch": 5.75, + "learning_rate": 3.860619493252578e-06, + "loss": 0.1011, + "step": 13738 + }, + { + "epoch": 5.75, + "learning_rate": 3.859281303356689e-06, + "loss": 0.163, + "step": 13739 + }, + { + "epoch": 5.75, + "learning_rate": 3.857943289970311e-06, + "loss": 0.1182, + "step": 13740 + }, + { + "epoch": 5.75, + "learning_rate": 3.856605453131901e-06, + "loss": 0.1114, + "step": 13741 + }, + { + "epoch": 5.75, + "learning_rate": 3.855267792879921e-06, + "loss": 0.1491, + "step": 13742 + }, + { + "epoch": 5.75, + "learning_rate": 3.85393030925281e-06, + "loss": 0.0939, + "step": 13743 + }, + { + "epoch": 5.75, + "learning_rate": 3.85259300228902e-06, + "loss": 0.0992, + "step": 13744 + }, + { + "epoch": 5.75, + "learning_rate": 3.851255872026989e-06, + "loss": 0.1063, + "step": 13745 + }, + { + "epoch": 5.76, + "learning_rate": 3.849918918505151e-06, + "loss": 0.0433, + "step": 13746 + }, + { + "epoch": 5.76, + "learning_rate": 3.848582141761939e-06, + "loss": 0.1548, + "step": 13747 + }, + { + "epoch": 5.76, + "learning_rate": 3.8472455418357735e-06, + "loss": 0.1037, + "step": 13748 + }, + { + "epoch": 5.76, + "learning_rate": 3.845909118765073e-06, + "loss": 0.0993, + "step": 13749 + }, + { + "epoch": 5.76, + "learning_rate": 3.844572872588258e-06, + "loss": 0.1247, + "step": 13750 + }, + { + "epoch": 5.76, + "learning_rate": 3.843236803343731e-06, + "loss": 0.1247, + "step": 13751 + }, + { + "epoch": 5.76, + "learning_rate": 3.841900911069897e-06, + "loss": 0.1141, + "step": 13752 + }, + { + "epoch": 5.76, + "learning_rate": 3.840565195805158e-06, + "loss": 0.0834, + "step": 13753 + }, + { + "epoch": 5.76, + "learning_rate": 3.839229657587906e-06, + "loss": 0.1405, + "step": 13754 + }, + { + "epoch": 5.76, + "learning_rate": 3.83789429645653e-06, + "loss": 0.1399, + "step": 13755 + }, + { + "epoch": 5.76, + "learning_rate": 3.836559112449415e-06, + "loss": 0.1061, + "step": 13756 + }, + { + "epoch": 5.76, + "learning_rate": 3.835224105604942e-06, + "loss": 0.0806, + "step": 13757 + }, + { + "epoch": 5.76, + "learning_rate": 3.833889275961479e-06, + "loss": 0.1568, + "step": 13758 + }, + { + "epoch": 5.76, + "learning_rate": 3.832554623557397e-06, + "loss": 0.1016, + "step": 13759 + }, + { + "epoch": 5.76, + "learning_rate": 3.831220148431058e-06, + "loss": 0.1125, + "step": 13760 + }, + { + "epoch": 5.76, + "learning_rate": 3.829885850620824e-06, + "loss": 0.0723, + "step": 13761 + }, + { + "epoch": 5.76, + "learning_rate": 3.8285517301650445e-06, + "loss": 0.1175, + "step": 13762 + }, + { + "epoch": 5.76, + "learning_rate": 3.827217787102072e-06, + "loss": 0.1329, + "step": 13763 + }, + { + "epoch": 5.76, + "learning_rate": 3.825884021470244e-06, + "loss": 0.1014, + "step": 13764 + }, + { + "epoch": 5.76, + "learning_rate": 3.8245504333079006e-06, + "loss": 0.0892, + "step": 13765 + }, + { + "epoch": 5.76, + "learning_rate": 3.823217022653376e-06, + "loss": 0.1108, + "step": 13766 + }, + { + "epoch": 5.76, + "learning_rate": 3.8218837895449956e-06, + "loss": 0.1202, + "step": 13767 + }, + { + "epoch": 5.76, + "learning_rate": 3.820550734021083e-06, + "loss": 0.1437, + "step": 13768 + }, + { + "epoch": 5.76, + "learning_rate": 3.819217856119961e-06, + "loss": 0.1, + "step": 13769 + }, + { + "epoch": 5.77, + "learning_rate": 3.817885155879933e-06, + "loss": 0.119, + "step": 13770 + }, + { + "epoch": 5.77, + "learning_rate": 3.816552633339313e-06, + "loss": 0.1255, + "step": 13771 + }, + { + "epoch": 5.77, + "learning_rate": 3.815220288536399e-06, + "loss": 0.1156, + "step": 13772 + }, + { + "epoch": 5.77, + "learning_rate": 3.813888121509487e-06, + "loss": 0.1489, + "step": 13773 + }, + { + "epoch": 5.77, + "learning_rate": 3.812556132296873e-06, + "loss": 0.1184, + "step": 13774 + }, + { + "epoch": 5.77, + "learning_rate": 3.8112243209368415e-06, + "loss": 0.1169, + "step": 13775 + }, + { + "epoch": 5.77, + "learning_rate": 3.8098926874676757e-06, + "loss": 0.1359, + "step": 13776 + }, + { + "epoch": 5.77, + "learning_rate": 3.8085612319276556e-06, + "loss": 0.1257, + "step": 13777 + }, + { + "epoch": 5.77, + "learning_rate": 3.8072299543550438e-06, + "loss": 0.1239, + "step": 13778 + }, + { + "epoch": 5.77, + "learning_rate": 3.8058988547881126e-06, + "loss": 0.0871, + "step": 13779 + }, + { + "epoch": 5.77, + "learning_rate": 3.804567933265122e-06, + "loss": 0.0929, + "step": 13780 + }, + { + "epoch": 5.77, + "learning_rate": 3.803237189824328e-06, + "loss": 0.1104, + "step": 13781 + }, + { + "epoch": 5.77, + "learning_rate": 3.8019066245039828e-06, + "loss": 0.0898, + "step": 13782 + }, + { + "epoch": 5.77, + "learning_rate": 3.800576237342335e-06, + "loss": 0.0989, + "step": 13783 + }, + { + "epoch": 5.77, + "learning_rate": 3.799246028377618e-06, + "loss": 0.1339, + "step": 13784 + }, + { + "epoch": 5.77, + "learning_rate": 3.7979159976480717e-06, + "loss": 0.1075, + "step": 13785 + }, + { + "epoch": 5.77, + "learning_rate": 3.796586145191926e-06, + "loss": 0.1306, + "step": 13786 + }, + { + "epoch": 5.77, + "learning_rate": 3.7952564710474072e-06, + "loss": 0.148, + "step": 13787 + }, + { + "epoch": 5.77, + "learning_rate": 3.793926975252735e-06, + "loss": 0.1047, + "step": 13788 + }, + { + "epoch": 5.77, + "learning_rate": 3.7925976578461255e-06, + "loss": 0.143, + "step": 13789 + }, + { + "epoch": 5.77, + "learning_rate": 3.7912685188657905e-06, + "loss": 0.1201, + "step": 13790 + }, + { + "epoch": 5.77, + "learning_rate": 3.789939558349929e-06, + "loss": 0.0978, + "step": 13791 + }, + { + "epoch": 5.77, + "learning_rate": 3.7886107763367474e-06, + "loss": 0.0969, + "step": 13792 + }, + { + "epoch": 5.77, + "learning_rate": 3.787282172864435e-06, + "loss": 0.1366, + "step": 13793 + }, + { + "epoch": 5.78, + "learning_rate": 3.785953747971183e-06, + "loss": 0.1331, + "step": 13794 + }, + { + "epoch": 5.78, + "learning_rate": 3.784625501695175e-06, + "loss": 0.1114, + "step": 13795 + }, + { + "epoch": 5.78, + "learning_rate": 3.783297434074592e-06, + "loss": 0.1316, + "step": 13796 + }, + { + "epoch": 5.78, + "learning_rate": 3.781969545147609e-06, + "loss": 0.1393, + "step": 13797 + }, + { + "epoch": 5.78, + "learning_rate": 3.7806418349523956e-06, + "loss": 0.1353, + "step": 13798 + }, + { + "epoch": 5.78, + "learning_rate": 3.7793143035271106e-06, + "loss": 0.1607, + "step": 13799 + }, + { + "epoch": 5.78, + "learning_rate": 3.7779869509099166e-06, + "loss": 0.0987, + "step": 13800 + }, + { + "epoch": 5.78, + "learning_rate": 3.776659777138966e-06, + "loss": 0.0866, + "step": 13801 + }, + { + "epoch": 5.78, + "learning_rate": 3.7753327822524067e-06, + "loss": 0.1508, + "step": 13802 + }, + { + "epoch": 5.78, + "learning_rate": 3.7740059662883843e-06, + "loss": 0.1343, + "step": 13803 + }, + { + "epoch": 5.78, + "learning_rate": 3.7726793292850382e-06, + "loss": 0.1392, + "step": 13804 + }, + { + "epoch": 5.78, + "learning_rate": 3.771352871280495e-06, + "loss": 0.1397, + "step": 13805 + }, + { + "epoch": 5.78, + "learning_rate": 3.7700265923128866e-06, + "loss": 0.087, + "step": 13806 + }, + { + "epoch": 5.78, + "learning_rate": 3.768700492420335e-06, + "loss": 0.1246, + "step": 13807 + }, + { + "epoch": 5.78, + "learning_rate": 3.7673745716409595e-06, + "loss": 0.1021, + "step": 13808 + }, + { + "epoch": 5.78, + "learning_rate": 3.7660488300128706e-06, + "loss": 0.1268, + "step": 13809 + }, + { + "epoch": 5.78, + "learning_rate": 3.7647232675741794e-06, + "loss": 0.0973, + "step": 13810 + }, + { + "epoch": 5.78, + "learning_rate": 3.7633978843629813e-06, + "loss": 0.0975, + "step": 13811 + }, + { + "epoch": 5.78, + "learning_rate": 3.7620726804173767e-06, + "loss": 0.1035, + "step": 13812 + }, + { + "epoch": 5.78, + "learning_rate": 3.7607476557754574e-06, + "loss": 0.1264, + "step": 13813 + }, + { + "epoch": 5.78, + "learning_rate": 3.7594228104753107e-06, + "loss": 0.1345, + "step": 13814 + }, + { + "epoch": 5.78, + "learning_rate": 3.75809814455502e-06, + "loss": 0.1187, + "step": 13815 + }, + { + "epoch": 5.78, + "learning_rate": 3.756773658052657e-06, + "loss": 0.1339, + "step": 13816 + }, + { + "epoch": 5.78, + "learning_rate": 3.755449351006294e-06, + "loss": 0.1081, + "step": 13817 + }, + { + "epoch": 5.79, + "learning_rate": 3.7541252234540028e-06, + "loss": 0.1106, + "step": 13818 + }, + { + "epoch": 5.79, + "learning_rate": 3.752801275433835e-06, + "loss": 0.1587, + "step": 13819 + }, + { + "epoch": 5.79, + "learning_rate": 3.7514775069838503e-06, + "loss": 0.1327, + "step": 13820 + }, + { + "epoch": 5.79, + "learning_rate": 3.750153918142101e-06, + "loss": 0.1187, + "step": 13821 + }, + { + "epoch": 5.79, + "learning_rate": 3.74883050894663e-06, + "loss": 0.1307, + "step": 13822 + }, + { + "epoch": 5.79, + "learning_rate": 3.7475072794354803e-06, + "loss": 0.0944, + "step": 13823 + }, + { + "epoch": 5.79, + "learning_rate": 3.746184229646689e-06, + "loss": 0.0824, + "step": 13824 + }, + { + "epoch": 5.79, + "learning_rate": 3.744861359618278e-06, + "loss": 0.1315, + "step": 13825 + }, + { + "epoch": 5.79, + "learning_rate": 3.7435386693882767e-06, + "loss": 0.1121, + "step": 13826 + }, + { + "epoch": 5.79, + "learning_rate": 3.742216158994705e-06, + "loss": 0.0785, + "step": 13827 + }, + { + "epoch": 5.79, + "learning_rate": 3.740893828475577e-06, + "loss": 0.1419, + "step": 13828 + }, + { + "epoch": 5.79, + "learning_rate": 3.7395716778689006e-06, + "loss": 0.1205, + "step": 13829 + }, + { + "epoch": 5.79, + "learning_rate": 3.738249707212681e-06, + "loss": 0.1453, + "step": 13830 + }, + { + "epoch": 5.79, + "learning_rate": 3.736927916544921e-06, + "loss": 0.129, + "step": 13831 + }, + { + "epoch": 5.79, + "learning_rate": 3.7356063059036073e-06, + "loss": 0.1384, + "step": 13832 + }, + { + "epoch": 5.79, + "learning_rate": 3.73428487532673e-06, + "loss": 0.1283, + "step": 13833 + }, + { + "epoch": 5.79, + "learning_rate": 3.732963624852275e-06, + "loss": 0.1252, + "step": 13834 + }, + { + "epoch": 5.79, + "learning_rate": 3.731642554518219e-06, + "loss": 0.1, + "step": 13835 + }, + { + "epoch": 5.79, + "learning_rate": 3.7303216643625364e-06, + "loss": 0.1462, + "step": 13836 + }, + { + "epoch": 5.79, + "learning_rate": 3.7290009544231955e-06, + "loss": 0.1298, + "step": 13837 + }, + { + "epoch": 5.79, + "learning_rate": 3.727680424738155e-06, + "loss": 0.1198, + "step": 13838 + }, + { + "epoch": 5.79, + "learning_rate": 3.726360075345378e-06, + "loss": 0.1026, + "step": 13839 + }, + { + "epoch": 5.79, + "learning_rate": 3.7250399062828104e-06, + "loss": 0.1226, + "step": 13840 + }, + { + "epoch": 5.79, + "learning_rate": 3.723719917588402e-06, + "loss": 0.1262, + "step": 13841 + }, + { + "epoch": 5.8, + "learning_rate": 3.722400109300095e-06, + "loss": 0.1533, + "step": 13842 + }, + { + "epoch": 5.8, + "learning_rate": 3.7210804814558253e-06, + "loss": 0.1316, + "step": 13843 + }, + { + "epoch": 5.8, + "learning_rate": 3.7197610340935263e-06, + "loss": 0.1205, + "step": 13844 + }, + { + "epoch": 5.8, + "learning_rate": 3.7184417672511273e-06, + "loss": 0.0987, + "step": 13845 + }, + { + "epoch": 5.8, + "learning_rate": 3.7171226809665417e-06, + "loss": 0.1437, + "step": 13846 + }, + { + "epoch": 5.8, + "learning_rate": 3.71580377527769e-06, + "loss": 0.1369, + "step": 13847 + }, + { + "epoch": 5.8, + "learning_rate": 3.7144850502224816e-06, + "loss": 0.135, + "step": 13848 + }, + { + "epoch": 5.8, + "learning_rate": 3.713166505838822e-06, + "loss": 0.102, + "step": 13849 + }, + { + "epoch": 5.8, + "learning_rate": 3.7118481421646134e-06, + "loss": 0.0825, + "step": 13850 + }, + { + "epoch": 5.8, + "learning_rate": 3.710529959237753e-06, + "loss": 0.0986, + "step": 13851 + }, + { + "epoch": 5.8, + "learning_rate": 3.7092119570961248e-06, + "loss": 0.1431, + "step": 13852 + }, + { + "epoch": 5.8, + "learning_rate": 3.7078941357776153e-06, + "loss": 0.0948, + "step": 13853 + }, + { + "epoch": 5.8, + "learning_rate": 3.7065764953201066e-06, + "loss": 0.1081, + "step": 13854 + }, + { + "epoch": 5.8, + "learning_rate": 3.7052590357614717e-06, + "loss": 0.0667, + "step": 13855 + }, + { + "epoch": 5.8, + "learning_rate": 3.7039417571395796e-06, + "loss": 0.1089, + "step": 13856 + }, + { + "epoch": 5.8, + "learning_rate": 3.7026246594922944e-06, + "loss": 0.1096, + "step": 13857 + }, + { + "epoch": 5.8, + "learning_rate": 3.7013077428574794e-06, + "loss": 0.1008, + "step": 13858 + }, + { + "epoch": 5.8, + "learning_rate": 3.6999910072729808e-06, + "loss": 0.1199, + "step": 13859 + }, + { + "epoch": 5.8, + "learning_rate": 3.698674452776653e-06, + "loss": 0.1097, + "step": 13860 + }, + { + "epoch": 5.8, + "learning_rate": 3.6973580794063323e-06, + "loss": 0.1472, + "step": 13861 + }, + { + "epoch": 5.8, + "learning_rate": 3.6960418871998615e-06, + "loss": 0.1396, + "step": 13862 + }, + { + "epoch": 5.8, + "learning_rate": 3.6947258761950723e-06, + "loss": 0.1408, + "step": 13863 + }, + { + "epoch": 5.8, + "learning_rate": 3.693410046429793e-06, + "loss": 0.0973, + "step": 13864 + }, + { + "epoch": 5.8, + "learning_rate": 3.6920943979418478e-06, + "loss": 0.1614, + "step": 13865 + }, + { + "epoch": 5.81, + "learning_rate": 3.690778930769049e-06, + "loss": 0.1382, + "step": 13866 + }, + { + "epoch": 5.81, + "learning_rate": 3.6894636449492116e-06, + "loss": 0.1073, + "step": 13867 + }, + { + "epoch": 5.81, + "learning_rate": 3.688148540520141e-06, + "loss": 0.1038, + "step": 13868 + }, + { + "epoch": 5.81, + "learning_rate": 3.6868336175196406e-06, + "loss": 0.1121, + "step": 13869 + }, + { + "epoch": 5.81, + "learning_rate": 3.6855188759855043e-06, + "loss": 0.125, + "step": 13870 + }, + { + "epoch": 5.81, + "learning_rate": 3.6842043159555263e-06, + "loss": 0.1256, + "step": 13871 + }, + { + "epoch": 5.81, + "learning_rate": 3.6828899374674933e-06, + "loss": 0.0791, + "step": 13872 + }, + { + "epoch": 5.81, + "learning_rate": 3.681575740559179e-06, + "loss": 0.1311, + "step": 13873 + }, + { + "epoch": 5.81, + "learning_rate": 3.680261725268364e-06, + "loss": 0.122, + "step": 13874 + }, + { + "epoch": 5.81, + "learning_rate": 3.678947891632818e-06, + "loss": 0.1136, + "step": 13875 + }, + { + "epoch": 5.81, + "learning_rate": 3.677634239690305e-06, + "loss": 0.0916, + "step": 13876 + }, + { + "epoch": 5.81, + "learning_rate": 3.676320769478585e-06, + "loss": 0.1307, + "step": 13877 + }, + { + "epoch": 5.81, + "learning_rate": 3.6750074810354173e-06, + "loss": 0.1332, + "step": 13878 + }, + { + "epoch": 5.81, + "learning_rate": 3.673694374398543e-06, + "loss": 0.1622, + "step": 13879 + }, + { + "epoch": 5.81, + "learning_rate": 3.672381449605711e-06, + "loss": 0.1185, + "step": 13880 + }, + { + "epoch": 5.81, + "learning_rate": 3.671068706694658e-06, + "loss": 0.1075, + "step": 13881 + }, + { + "epoch": 5.81, + "learning_rate": 3.6697561457031196e-06, + "loss": 0.0994, + "step": 13882 + }, + { + "epoch": 5.81, + "learning_rate": 3.6684437666688253e-06, + "loss": 0.1085, + "step": 13883 + }, + { + "epoch": 5.81, + "learning_rate": 3.667131569629495e-06, + "loss": 0.1145, + "step": 13884 + }, + { + "epoch": 5.81, + "learning_rate": 3.665819554622847e-06, + "loss": 0.1068, + "step": 13885 + }, + { + "epoch": 5.81, + "learning_rate": 3.6645077216865987e-06, + "loss": 0.0967, + "step": 13886 + }, + { + "epoch": 5.81, + "learning_rate": 3.6631960708584504e-06, + "loss": 0.1445, + "step": 13887 + }, + { + "epoch": 5.81, + "learning_rate": 3.6618846021761078e-06, + "loss": 0.1084, + "step": 13888 + }, + { + "epoch": 5.81, + "learning_rate": 3.660573315677268e-06, + "loss": 0.099, + "step": 13889 + }, + { + "epoch": 5.82, + "learning_rate": 3.6592622113996235e-06, + "loss": 0.1187, + "step": 13890 + }, + { + "epoch": 5.82, + "learning_rate": 3.657951289380859e-06, + "loss": 0.0873, + "step": 13891 + }, + { + "epoch": 5.82, + "learning_rate": 3.656640549658661e-06, + "loss": 0.1233, + "step": 13892 + }, + { + "epoch": 5.82, + "learning_rate": 3.6553299922706977e-06, + "loss": 0.0894, + "step": 13893 + }, + { + "epoch": 5.82, + "learning_rate": 3.654019617254644e-06, + "loss": 0.0926, + "step": 13894 + }, + { + "epoch": 5.82, + "learning_rate": 3.652709424648164e-06, + "loss": 0.0958, + "step": 13895 + }, + { + "epoch": 5.82, + "learning_rate": 3.65139941448892e-06, + "loss": 0.1193, + "step": 13896 + }, + { + "epoch": 5.82, + "learning_rate": 3.650089586814566e-06, + "loss": 0.1366, + "step": 13897 + }, + { + "epoch": 5.82, + "learning_rate": 3.648779941662752e-06, + "loss": 0.124, + "step": 13898 + }, + { + "epoch": 5.82, + "learning_rate": 3.6474704790711258e-06, + "loss": 0.1464, + "step": 13899 + }, + { + "epoch": 5.82, + "learning_rate": 3.6461611990773204e-06, + "loss": 0.1439, + "step": 13900 + }, + { + "epoch": 5.82, + "learning_rate": 3.6448521017189732e-06, + "loss": 0.1262, + "step": 13901 + }, + { + "epoch": 5.82, + "learning_rate": 3.6435431870337123e-06, + "loss": 0.1203, + "step": 13902 + }, + { + "epoch": 5.82, + "learning_rate": 3.6422344550591625e-06, + "loss": 0.1148, + "step": 13903 + }, + { + "epoch": 5.82, + "learning_rate": 3.6409259058329404e-06, + "loss": 0.082, + "step": 13904 + }, + { + "epoch": 5.82, + "learning_rate": 3.6396175393926646e-06, + "loss": 0.095, + "step": 13905 + }, + { + "epoch": 5.82, + "learning_rate": 3.6383093557759386e-06, + "loss": 0.1278, + "step": 13906 + }, + { + "epoch": 5.82, + "learning_rate": 3.637001355020361e-06, + "loss": 0.1405, + "step": 13907 + }, + { + "epoch": 5.82, + "learning_rate": 3.6356935371635336e-06, + "loss": 0.0856, + "step": 13908 + }, + { + "epoch": 5.82, + "learning_rate": 3.6343859022430474e-06, + "loss": 0.1269, + "step": 13909 + }, + { + "epoch": 5.82, + "learning_rate": 3.63307845029649e-06, + "loss": 0.1347, + "step": 13910 + }, + { + "epoch": 5.82, + "learning_rate": 3.631771181361442e-06, + "loss": 0.117, + "step": 13911 + }, + { + "epoch": 5.82, + "learning_rate": 3.630464095475481e-06, + "loss": 0.1103, + "step": 13912 + }, + { + "epoch": 5.82, + "learning_rate": 3.6291571926761805e-06, + "loss": 0.1385, + "step": 13913 + }, + { + "epoch": 5.83, + "learning_rate": 3.6278504730011e-06, + "loss": 0.1068, + "step": 13914 + }, + { + "epoch": 5.83, + "learning_rate": 3.626543936487803e-06, + "loss": 0.0955, + "step": 13915 + }, + { + "epoch": 5.83, + "learning_rate": 3.625237583173844e-06, + "loss": 0.0907, + "step": 13916 + }, + { + "epoch": 5.83, + "learning_rate": 3.6239314130967753e-06, + "loss": 0.1233, + "step": 13917 + }, + { + "epoch": 5.83, + "learning_rate": 3.6226254262941398e-06, + "loss": 0.1122, + "step": 13918 + }, + { + "epoch": 5.83, + "learning_rate": 3.6213196228034796e-06, + "loss": 0.1352, + "step": 13919 + }, + { + "epoch": 5.83, + "learning_rate": 3.6200140026623233e-06, + "loss": 0.1447, + "step": 13920 + }, + { + "epoch": 5.83, + "learning_rate": 3.6187085659082034e-06, + "loss": 0.1134, + "step": 13921 + }, + { + "epoch": 5.83, + "learning_rate": 3.6174033125786434e-06, + "loss": 0.1147, + "step": 13922 + }, + { + "epoch": 5.83, + "learning_rate": 3.6160982427111614e-06, + "loss": 0.0617, + "step": 13923 + }, + { + "epoch": 5.83, + "learning_rate": 3.614793356343269e-06, + "loss": 0.1265, + "step": 13924 + }, + { + "epoch": 5.83, + "learning_rate": 3.6134886535124804e-06, + "loss": 0.0609, + "step": 13925 + }, + { + "epoch": 5.83, + "learning_rate": 3.612184134256289e-06, + "loss": 0.1177, + "step": 13926 + }, + { + "epoch": 5.83, + "learning_rate": 3.6108797986121967e-06, + "loss": 0.1149, + "step": 13927 + }, + { + "epoch": 5.83, + "learning_rate": 3.609575646617699e-06, + "loss": 0.1173, + "step": 13928 + }, + { + "epoch": 5.83, + "learning_rate": 3.6082716783102735e-06, + "loss": 0.0993, + "step": 13929 + }, + { + "epoch": 5.83, + "learning_rate": 3.6069678937274076e-06, + "loss": 0.1415, + "step": 13930 + }, + { + "epoch": 5.83, + "learning_rate": 3.6056642929065765e-06, + "loss": 0.088, + "step": 13931 + }, + { + "epoch": 5.83, + "learning_rate": 3.604360875885251e-06, + "loss": 0.1072, + "step": 13932 + }, + { + "epoch": 5.83, + "learning_rate": 3.6030576427009e-06, + "loss": 0.131, + "step": 13933 + }, + { + "epoch": 5.83, + "learning_rate": 3.601754593390978e-06, + "loss": 0.1314, + "step": 13934 + }, + { + "epoch": 5.83, + "learning_rate": 3.600451727992942e-06, + "loss": 0.1126, + "step": 13935 + }, + { + "epoch": 5.83, + "learning_rate": 3.5991490465442413e-06, + "loss": 0.1463, + "step": 13936 + }, + { + "epoch": 5.84, + "learning_rate": 3.597846549082322e-06, + "loss": 0.1579, + "step": 13937 + }, + { + "epoch": 5.84, + "learning_rate": 3.5965442356446224e-06, + "loss": 0.0985, + "step": 13938 + }, + { + "epoch": 5.84, + "learning_rate": 3.595242106268577e-06, + "loss": 0.1163, + "step": 13939 + }, + { + "epoch": 5.84, + "learning_rate": 3.593940160991617e-06, + "loss": 0.1439, + "step": 13940 + }, + { + "epoch": 5.84, + "learning_rate": 3.592638399851158e-06, + "loss": 0.1152, + "step": 13941 + }, + { + "epoch": 5.84, + "learning_rate": 3.5913368228846236e-06, + "loss": 0.1404, + "step": 13942 + }, + { + "epoch": 5.84, + "learning_rate": 3.5900354301294236e-06, + "loss": 0.1272, + "step": 13943 + }, + { + "epoch": 5.84, + "learning_rate": 3.5887342216229694e-06, + "loss": 0.1131, + "step": 13944 + }, + { + "epoch": 5.84, + "learning_rate": 3.587433197402659e-06, + "loss": 0.1285, + "step": 13945 + }, + { + "epoch": 5.84, + "learning_rate": 3.586132357505895e-06, + "loss": 0.1054, + "step": 13946 + }, + { + "epoch": 5.84, + "learning_rate": 3.5848317019700608e-06, + "loss": 0.09, + "step": 13947 + }, + { + "epoch": 5.84, + "learning_rate": 3.5835312308325475e-06, + "loss": 0.1308, + "step": 13948 + }, + { + "epoch": 5.84, + "learning_rate": 3.5822309441307347e-06, + "loss": 0.1002, + "step": 13949 + }, + { + "epoch": 5.84, + "learning_rate": 3.5809308419019985e-06, + "loss": 0.0906, + "step": 13950 + }, + { + "epoch": 5.84, + "learning_rate": 3.5796309241837125e-06, + "loss": 0.1281, + "step": 13951 + }, + { + "epoch": 5.84, + "learning_rate": 3.5783311910132348e-06, + "loss": 0.1261, + "step": 13952 + }, + { + "epoch": 5.84, + "learning_rate": 3.5770316424279283e-06, + "loss": 0.1357, + "step": 13953 + }, + { + "epoch": 5.84, + "learning_rate": 3.5757322784651506e-06, + "loss": 0.1148, + "step": 13954 + }, + { + "epoch": 5.84, + "learning_rate": 3.5744330991622457e-06, + "loss": 0.0935, + "step": 13955 + }, + { + "epoch": 5.84, + "learning_rate": 3.573134104556558e-06, + "loss": 0.0908, + "step": 13956 + }, + { + "epoch": 5.84, + "learning_rate": 3.5718352946854286e-06, + "loss": 0.1236, + "step": 13957 + }, + { + "epoch": 5.84, + "learning_rate": 3.5705366695861886e-06, + "loss": 0.0759, + "step": 13958 + }, + { + "epoch": 5.84, + "learning_rate": 3.569238229296167e-06, + "loss": 0.1156, + "step": 13959 + }, + { + "epoch": 5.84, + "learning_rate": 3.567939973852689e-06, + "loss": 0.0935, + "step": 13960 + }, + { + "epoch": 5.85, + "learning_rate": 3.5666419032930653e-06, + "loss": 0.1497, + "step": 13961 + }, + { + "epoch": 5.85, + "learning_rate": 3.5653440176546115e-06, + "loss": 0.0916, + "step": 13962 + }, + { + "epoch": 5.85, + "learning_rate": 3.564046316974634e-06, + "loss": 0.0757, + "step": 13963 + }, + { + "epoch": 5.85, + "learning_rate": 3.5627488012904333e-06, + "loss": 0.0908, + "step": 13964 + }, + { + "epoch": 5.85, + "learning_rate": 3.5614514706393066e-06, + "loss": 0.1181, + "step": 13965 + }, + { + "epoch": 5.85, + "learning_rate": 3.560154325058547e-06, + "loss": 0.0886, + "step": 13966 + }, + { + "epoch": 5.85, + "learning_rate": 3.558857364585434e-06, + "loss": 0.1648, + "step": 13967 + }, + { + "epoch": 5.85, + "learning_rate": 3.5575605892572495e-06, + "loss": 0.0831, + "step": 13968 + }, + { + "epoch": 5.85, + "learning_rate": 3.556263999111269e-06, + "loss": 0.1066, + "step": 13969 + }, + { + "epoch": 5.85, + "learning_rate": 3.554967594184762e-06, + "loss": 0.1088, + "step": 13970 + }, + { + "epoch": 5.85, + "learning_rate": 3.553671374514992e-06, + "loss": 0.1059, + "step": 13971 + }, + { + "epoch": 5.85, + "learning_rate": 3.552375340139219e-06, + "loss": 0.1399, + "step": 13972 + }, + { + "epoch": 5.85, + "learning_rate": 3.551079491094698e-06, + "loss": 0.1216, + "step": 13973 + }, + { + "epoch": 5.85, + "learning_rate": 3.5497838274186737e-06, + "loss": 0.151, + "step": 13974 + }, + { + "epoch": 5.85, + "learning_rate": 3.5484883491483866e-06, + "loss": 0.1209, + "step": 13975 + }, + { + "epoch": 5.85, + "learning_rate": 3.5471930563210775e-06, + "loss": 0.14, + "step": 13976 + }, + { + "epoch": 5.85, + "learning_rate": 3.5458979489739778e-06, + "loss": 0.1103, + "step": 13977 + }, + { + "epoch": 5.85, + "learning_rate": 3.5446030271443132e-06, + "loss": 0.1041, + "step": 13978 + }, + { + "epoch": 5.85, + "learning_rate": 3.543308290869307e-06, + "loss": 0.0979, + "step": 13979 + }, + { + "epoch": 5.85, + "learning_rate": 3.5420137401861745e-06, + "loss": 0.1379, + "step": 13980 + }, + { + "epoch": 5.85, + "learning_rate": 3.540719375132129e-06, + "loss": 0.089, + "step": 13981 + }, + { + "epoch": 5.85, + "learning_rate": 3.5394251957443704e-06, + "loss": 0.1059, + "step": 13982 + }, + { + "epoch": 5.85, + "learning_rate": 3.538131202060102e-06, + "loss": 0.1206, + "step": 13983 + }, + { + "epoch": 5.85, + "learning_rate": 3.5368373941165167e-06, + "loss": 0.117, + "step": 13984 + }, + { + "epoch": 5.86, + "learning_rate": 3.5355437719508067e-06, + "loss": 0.0811, + "step": 13985 + }, + { + "epoch": 5.86, + "learning_rate": 3.5342503356001535e-06, + "loss": 0.1076, + "step": 13986 + }, + { + "epoch": 5.86, + "learning_rate": 3.5329570851017405e-06, + "loss": 0.112, + "step": 13987 + }, + { + "epoch": 5.86, + "learning_rate": 3.5316640204927342e-06, + "loss": 0.1049, + "step": 13988 + }, + { + "epoch": 5.86, + "learning_rate": 3.5303711418103047e-06, + "loss": 0.1239, + "step": 13989 + }, + { + "epoch": 5.86, + "learning_rate": 3.5290784490916175e-06, + "loss": 0.139, + "step": 13990 + }, + { + "epoch": 5.86, + "learning_rate": 3.527785942373827e-06, + "loss": 0.1611, + "step": 13991 + }, + { + "epoch": 5.86, + "learning_rate": 3.526493621694086e-06, + "loss": 0.1448, + "step": 13992 + }, + { + "epoch": 5.86, + "learning_rate": 3.5252014870895447e-06, + "loss": 0.1252, + "step": 13993 + }, + { + "epoch": 5.86, + "learning_rate": 3.5239095385973377e-06, + "loss": 0.1505, + "step": 13994 + }, + { + "epoch": 5.86, + "learning_rate": 3.522617776254604e-06, + "loss": 0.1341, + "step": 13995 + }, + { + "epoch": 5.86, + "learning_rate": 3.5213262000984793e-06, + "loss": 0.1043, + "step": 13996 + }, + { + "epoch": 5.86, + "learning_rate": 3.52003481016608e-06, + "loss": 0.1103, + "step": 13997 + }, + { + "epoch": 5.86, + "learning_rate": 3.518743606494529e-06, + "loss": 0.1107, + "step": 13998 + }, + { + "epoch": 5.86, + "learning_rate": 3.517452589120941e-06, + "loss": 0.0967, + "step": 13999 + }, + { + "epoch": 5.86, + "learning_rate": 3.516161758082427e-06, + "loss": 0.1111, + "step": 14000 + } + ], + "max_steps": 19104, + "num_train_epochs": 8, + "total_flos": 2.563773602943992e+18, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-14000/training_args.bin b/checkpoint-14000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..8b1b50208b5713a1848bb03b9cf39bd33c748f72 --- /dev/null +++ b/checkpoint-14000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:457d7e76c0fce8d03d8cfe3fdedf1b89cbf25e0f3a605b6976f387e928f69251 +size 3899 diff --git a/checkpoint-16000/added_tokens.json b/checkpoint-16000/added_tokens.json new file mode 100644 index 0000000000000000000000000000000000000000..301df2d484b892fb69765533defad74b1169af4d --- /dev/null +++ b/checkpoint-16000/added_tokens.json @@ -0,0 +1,6 @@ +{ + "": 32003, + "": 32001, + "": 32002, + "[PAD]": 32000 +} diff --git a/checkpoint-16000/config.json b/checkpoint-16000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..066a0107385020bf48a41a53e63dd916e2f96670 --- /dev/null +++ b/checkpoint-16000/config.json @@ -0,0 +1,30 @@ +{ + "_name_or_path": "/scratch/wzhu59/LLaVA-Med/weights/llava-med-7b-pretrain", + "architectures": [ + "LlavaLlamaForCausalLM" + ], + "bos_token_id": 0, + "eos_token_id": 1, + "freeze_mm_mlp_adapter": false, + "hidden_act": "silu", + "hidden_size": 4096, + "initializer_range": 0.02, + "intermediate_size": 11008, + "max_sequence_length": 2048, + "mm_hidden_size": 1024, + "mm_use_im_start_end": true, + "mm_vision_select_layer": -2, + "mm_vision_tower": "openai/clip-vit-large-patch14", + "model_type": "llava", + "num_attention_heads": 32, + "num_hidden_layers": 32, + "pad_token_id": -1, + "rms_norm_eps": 1e-06, + "tie_word_embeddings": false, + "torch_dtype": "float32", + "transformers_version": "4.28.0.dev0", + "tune_mm_mlp_adapter": false, + "use_cache": false, + "use_mm_proj": true, + "vocab_size": 32004 +} diff --git a/checkpoint-16000/generation_config.json b/checkpoint-16000/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..128b48946713166edcc3e3c62ae5fe285df776be --- /dev/null +++ b/checkpoint-16000/generation_config.json @@ -0,0 +1,7 @@ +{ + "_from_model_config": true, + "bos_token_id": 0, + "eos_token_id": 1, + "pad_token_id": 0, + "transformers_version": "4.28.0.dev0" +} diff --git a/checkpoint-16000/optimizer.pt b/checkpoint-16000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..9dc7c1001ae91c743b68a00a5e727c2180a921a9 --- /dev/null +++ b/checkpoint-16000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7372c20b00294c21c46fcd95f140ab1e6d8a96cf2a119ad5a921b99000796a27 +size 13485322046 diff --git a/checkpoint-16000/pytorch_model-00001-of-00003.bin b/checkpoint-16000/pytorch_model-00001-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..0a12f6c5a9c84bf314271dcf52cd6dbbff2e1e08 --- /dev/null +++ b/checkpoint-16000/pytorch_model-00001-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ff1c4002600493c7a2f5cbef738443099650a1cf1aa57e27c2f1a8a475b7798 +size 9878059026 diff --git a/checkpoint-16000/pytorch_model-00002-of-00003.bin b/checkpoint-16000/pytorch_model-00002-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..82687c348a1342929955e97906bafa4ec2857f0a --- /dev/null +++ b/checkpoint-16000/pytorch_model-00002-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c52805a4972ccee4b8da53a96d71242405f4a04d5ddae98ed0615c340364f906 +size 9894805046 diff --git a/checkpoint-16000/pytorch_model-00003-of-00003.bin b/checkpoint-16000/pytorch_model-00003-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..84b2f94039cb186c6b469e7b0fb15caab550be59 --- /dev/null +++ b/checkpoint-16000/pytorch_model-00003-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86cc06a0e7f946083efe962730f259b1a26f4385538b4eb2030d5301931d20a3 +size 7197853429 diff --git a/checkpoint-16000/pytorch_model.bin.index.json b/checkpoint-16000/pytorch_model.bin.index.json new file mode 100644 index 0000000000000000000000000000000000000000..50d9a764a2282762889d821cfaa49b284963c070 --- /dev/null +++ b/checkpoint-16000/pytorch_model.bin.index.json @@ -0,0 +1,332 @@ +{ + "metadata": { + "total_size": 26970595328 + }, + "weight_map": { + "lm_head.weight": "pytorch_model-00003-of-00003.bin", + "model.embed_tokens.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.12.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.20.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.24.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.30.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.mm_projector.bias": "pytorch_model-00003-of-00003.bin", + "model.mm_projector.weight": "pytorch_model-00003-of-00003.bin", + "model.norm.weight": "pytorch_model-00003-of-00003.bin" + } +} diff --git a/checkpoint-16000/rng_state_0.pth b/checkpoint-16000/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..ff43b898898ffa08528704e9f058d404a42515f1 --- /dev/null +++ b/checkpoint-16000/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00b99c20b2dced2617cd5e70084441a015a660c4bfe1c2d6cb892d22a41e26f6 +size 14583 diff --git a/checkpoint-16000/rng_state_1.pth b/checkpoint-16000/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..0ddfe13712c3095ff016bb6c8cda9bcce48618db --- /dev/null +++ b/checkpoint-16000/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eea1570e194cf8345b3084cb29d7420c0c71ed8a95c001522cc25874e8b94996 +size 14583 diff --git a/checkpoint-16000/rng_state_2.pth b/checkpoint-16000/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..5c2f6cc46f015f9efa7ff0025c1049d9b1d7fd2a --- /dev/null +++ b/checkpoint-16000/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a2afd9d0f0bf8385ff934cee0d18becfbbac667fea3febbb45148df66ed02577 +size 14583 diff --git a/checkpoint-16000/rng_state_3.pth b/checkpoint-16000/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..8d40d430963e04833f39c37b24ebe6a252a13593 --- /dev/null +++ b/checkpoint-16000/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b85338f3ad7c93a51c63bbec639b4f8b73403a235d43a2800f2fd7e6b5adf27 +size 14583 diff --git a/checkpoint-16000/scheduler.pt b/checkpoint-16000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..9f7b2b3683729c6b28c4fad43b282efb41a1e09e --- /dev/null +++ b/checkpoint-16000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1bd78c5a9a49290bc78f9dd846f0b802713ba1386b2c731767dbdca419bf447 +size 627 diff --git a/checkpoint-16000/special_tokens_map.json b/checkpoint-16000/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..5ce7dbc9759ecf5596c5122abbdf0b431355efa4 --- /dev/null +++ b/checkpoint-16000/special_tokens_map.json @@ -0,0 +1,6 @@ +{ + "bos_token": "", + "eos_token": "", + "pad_token": "[PAD]", + "unk_token": "" +} diff --git a/checkpoint-16000/tokenizer.model b/checkpoint-16000/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-16000/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-16000/tokenizer_config.json b/checkpoint-16000/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..22bcdfb351739d00df7b9b12f99757c6e7e0b540 --- /dev/null +++ b/checkpoint-16000/tokenizer_config.json @@ -0,0 +1,32 @@ +{ + "bos_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + }, + "clean_up_tokenization_spaces": false, + "eos_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + }, + "model_max_length": 2048, + "pad_token": null, + "padding_side": "right", + "special_tokens_map_file": "/nobackup/haotian/.cache/huggingface/hub/models--lmsys--vicuna-7b-delta-v0/snapshots/6d2e645f62e734821be8c392c08c8b2098b35434/special_tokens_map.json", + "tokenizer_class": "LlamaTokenizer", + "unk_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-16000/trainer_state.json b/checkpoint-16000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..1c4d8c48bc82bba859b8adb3d0eb53285f60a588 --- /dev/null +++ b/checkpoint-16000/trainer_state.json @@ -0,0 +1,96016 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 6.6987649152187565, + "global_step": 16000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 3.484320557491289e-08, + "loss": 1.487, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 6.968641114982578e-08, + "loss": 1.5418, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 1.045296167247387e-07, + "loss": 1.2413, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 1.3937282229965157e-07, + "loss": 1.1777, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 1.7421602787456448e-07, + "loss": 0.9729, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 2.090592334494774e-07, + "loss": 0.9887, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 2.439024390243903e-07, + "loss": 0.9151, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 2.7874564459930313e-07, + "loss": 0.9341, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 3.1358885017421604e-07, + "loss": 0.9024, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 3.4843205574912896e-07, + "loss": 0.96, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 3.832752613240418e-07, + "loss": 0.9397, + "step": 11 + }, + { + "epoch": 0.01, + "learning_rate": 4.181184668989548e-07, + "loss": 1.0473, + "step": 12 + }, + { + "epoch": 0.01, + "learning_rate": 4.5296167247386764e-07, + "loss": 0.8521, + "step": 13 + }, + { + "epoch": 0.01, + "learning_rate": 4.878048780487805e-07, + "loss": 0.9829, + "step": 14 + }, + { + "epoch": 0.01, + "learning_rate": 5.226480836236935e-07, + "loss": 1.1014, + "step": 15 + }, + { + "epoch": 0.01, + "learning_rate": 5.574912891986063e-07, + "loss": 0.9186, + "step": 16 + }, + { + "epoch": 0.01, + "learning_rate": 5.923344947735192e-07, + "loss": 0.8867, + "step": 17 + }, + { + "epoch": 0.01, + "learning_rate": 6.271777003484321e-07, + "loss": 0.8071, + "step": 18 + }, + { + "epoch": 0.01, + "learning_rate": 6.62020905923345e-07, + "loss": 0.8481, + "step": 19 + }, + { + "epoch": 0.01, + "learning_rate": 6.968641114982579e-07, + "loss": 0.8675, + "step": 20 + }, + { + "epoch": 0.01, + "learning_rate": 7.317073170731707e-07, + "loss": 0.9635, + "step": 21 + }, + { + "epoch": 0.01, + "learning_rate": 7.665505226480836e-07, + "loss": 0.8868, + "step": 22 + }, + { + "epoch": 0.01, + "learning_rate": 8.013937282229965e-07, + "loss": 0.9029, + "step": 23 + }, + { + "epoch": 0.01, + "learning_rate": 8.362369337979096e-07, + "loss": 0.9118, + "step": 24 + }, + { + "epoch": 0.01, + "learning_rate": 8.710801393728225e-07, + "loss": 0.9768, + "step": 25 + }, + { + "epoch": 0.01, + "learning_rate": 9.059233449477353e-07, + "loss": 0.8292, + "step": 26 + }, + { + "epoch": 0.01, + "learning_rate": 9.407665505226482e-07, + "loss": 0.8865, + "step": 27 + }, + { + "epoch": 0.01, + "learning_rate": 9.75609756097561e-07, + "loss": 0.9918, + "step": 28 + }, + { + "epoch": 0.01, + "learning_rate": 1.010452961672474e-06, + "loss": 0.7146, + "step": 29 + }, + { + "epoch": 0.01, + "learning_rate": 1.045296167247387e-06, + "loss": 0.907, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 1.0801393728222997e-06, + "loss": 0.8015, + "step": 31 + }, + { + "epoch": 0.01, + "learning_rate": 1.1149825783972125e-06, + "loss": 0.7597, + "step": 32 + }, + { + "epoch": 0.01, + "learning_rate": 1.1498257839721255e-06, + "loss": 0.8095, + "step": 33 + }, + { + "epoch": 0.01, + "learning_rate": 1.1846689895470384e-06, + "loss": 0.9036, + "step": 34 + }, + { + "epoch": 0.01, + "learning_rate": 1.2195121951219514e-06, + "loss": 0.8162, + "step": 35 + }, + { + "epoch": 0.02, + "learning_rate": 1.2543554006968642e-06, + "loss": 0.9883, + "step": 36 + }, + { + "epoch": 0.02, + "learning_rate": 1.289198606271777e-06, + "loss": 0.8634, + "step": 37 + }, + { + "epoch": 0.02, + "learning_rate": 1.32404181184669e-06, + "loss": 0.7832, + "step": 38 + }, + { + "epoch": 0.02, + "learning_rate": 1.3588850174216028e-06, + "loss": 0.7378, + "step": 39 + }, + { + "epoch": 0.02, + "learning_rate": 1.3937282229965158e-06, + "loss": 0.8348, + "step": 40 + }, + { + "epoch": 0.02, + "learning_rate": 1.4285714285714286e-06, + "loss": 0.6925, + "step": 41 + }, + { + "epoch": 0.02, + "learning_rate": 1.4634146341463414e-06, + "loss": 0.8969, + "step": 42 + }, + { + "epoch": 0.02, + "learning_rate": 1.4982578397212545e-06, + "loss": 0.7099, + "step": 43 + }, + { + "epoch": 0.02, + "learning_rate": 1.5331010452961673e-06, + "loss": 0.8763, + "step": 44 + }, + { + "epoch": 0.02, + "learning_rate": 1.56794425087108e-06, + "loss": 0.7967, + "step": 45 + }, + { + "epoch": 0.02, + "learning_rate": 1.602787456445993e-06, + "loss": 0.7528, + "step": 46 + }, + { + "epoch": 0.02, + "learning_rate": 1.6376306620209059e-06, + "loss": 0.8367, + "step": 47 + }, + { + "epoch": 0.02, + "learning_rate": 1.6724738675958191e-06, + "loss": 0.8172, + "step": 48 + }, + { + "epoch": 0.02, + "learning_rate": 1.707317073170732e-06, + "loss": 0.7745, + "step": 49 + }, + { + "epoch": 0.02, + "learning_rate": 1.742160278745645e-06, + "loss": 0.8643, + "step": 50 + }, + { + "epoch": 0.02, + "learning_rate": 1.7770034843205577e-06, + "loss": 0.7374, + "step": 51 + }, + { + "epoch": 0.02, + "learning_rate": 1.8118466898954705e-06, + "loss": 0.8095, + "step": 52 + }, + { + "epoch": 0.02, + "learning_rate": 1.8466898954703836e-06, + "loss": 0.7755, + "step": 53 + }, + { + "epoch": 0.02, + "learning_rate": 1.8815331010452964e-06, + "loss": 0.8084, + "step": 54 + }, + { + "epoch": 0.02, + "learning_rate": 1.916376306620209e-06, + "loss": 0.7396, + "step": 55 + }, + { + "epoch": 0.02, + "learning_rate": 1.951219512195122e-06, + "loss": 0.7355, + "step": 56 + }, + { + "epoch": 0.02, + "learning_rate": 1.986062717770035e-06, + "loss": 0.7007, + "step": 57 + }, + { + "epoch": 0.02, + "learning_rate": 2.020905923344948e-06, + "loss": 0.8282, + "step": 58 + }, + { + "epoch": 0.02, + "learning_rate": 2.055749128919861e-06, + "loss": 0.7561, + "step": 59 + }, + { + "epoch": 0.03, + "learning_rate": 2.090592334494774e-06, + "loss": 0.8712, + "step": 60 + }, + { + "epoch": 0.03, + "learning_rate": 2.1254355400696864e-06, + "loss": 0.8559, + "step": 61 + }, + { + "epoch": 0.03, + "learning_rate": 2.1602787456445995e-06, + "loss": 0.7767, + "step": 62 + }, + { + "epoch": 0.03, + "learning_rate": 2.1951219512195125e-06, + "loss": 0.81, + "step": 63 + }, + { + "epoch": 0.03, + "learning_rate": 2.229965156794425e-06, + "loss": 0.805, + "step": 64 + }, + { + "epoch": 0.03, + "learning_rate": 2.264808362369338e-06, + "loss": 0.6648, + "step": 65 + }, + { + "epoch": 0.03, + "learning_rate": 2.299651567944251e-06, + "loss": 0.8027, + "step": 66 + }, + { + "epoch": 0.03, + "learning_rate": 2.334494773519164e-06, + "loss": 0.7623, + "step": 67 + }, + { + "epoch": 0.03, + "learning_rate": 2.3693379790940767e-06, + "loss": 0.7903, + "step": 68 + }, + { + "epoch": 0.03, + "learning_rate": 2.4041811846689897e-06, + "loss": 0.6917, + "step": 69 + }, + { + "epoch": 0.03, + "learning_rate": 2.4390243902439027e-06, + "loss": 0.8177, + "step": 70 + }, + { + "epoch": 0.03, + "learning_rate": 2.4738675958188153e-06, + "loss": 0.7348, + "step": 71 + }, + { + "epoch": 0.03, + "learning_rate": 2.5087108013937284e-06, + "loss": 0.6916, + "step": 72 + }, + { + "epoch": 0.03, + "learning_rate": 2.5435540069686414e-06, + "loss": 0.6663, + "step": 73 + }, + { + "epoch": 0.03, + "learning_rate": 2.578397212543554e-06, + "loss": 0.6881, + "step": 74 + }, + { + "epoch": 0.03, + "learning_rate": 2.6132404181184674e-06, + "loss": 0.6372, + "step": 75 + }, + { + "epoch": 0.03, + "learning_rate": 2.64808362369338e-06, + "loss": 0.7268, + "step": 76 + }, + { + "epoch": 0.03, + "learning_rate": 2.682926829268293e-06, + "loss": 0.7138, + "step": 77 + }, + { + "epoch": 0.03, + "learning_rate": 2.7177700348432056e-06, + "loss": 0.7341, + "step": 78 + }, + { + "epoch": 0.03, + "learning_rate": 2.7526132404181186e-06, + "loss": 0.7041, + "step": 79 + }, + { + "epoch": 0.03, + "learning_rate": 2.7874564459930316e-06, + "loss": 0.7239, + "step": 80 + }, + { + "epoch": 0.03, + "learning_rate": 2.8222996515679447e-06, + "loss": 0.6809, + "step": 81 + }, + { + "epoch": 0.03, + "learning_rate": 2.8571428571428573e-06, + "loss": 0.7256, + "step": 82 + }, + { + "epoch": 0.03, + "learning_rate": 2.8919860627177703e-06, + "loss": 0.7205, + "step": 83 + }, + { + "epoch": 0.04, + "learning_rate": 2.926829268292683e-06, + "loss": 0.6596, + "step": 84 + }, + { + "epoch": 0.04, + "learning_rate": 2.9616724738675963e-06, + "loss": 0.657, + "step": 85 + }, + { + "epoch": 0.04, + "learning_rate": 2.996515679442509e-06, + "loss": 0.6502, + "step": 86 + }, + { + "epoch": 0.04, + "learning_rate": 3.031358885017422e-06, + "loss": 0.7095, + "step": 87 + }, + { + "epoch": 0.04, + "learning_rate": 3.0662020905923345e-06, + "loss": 0.6643, + "step": 88 + }, + { + "epoch": 0.04, + "learning_rate": 3.1010452961672475e-06, + "loss": 0.6621, + "step": 89 + }, + { + "epoch": 0.04, + "learning_rate": 3.13588850174216e-06, + "loss": 0.7324, + "step": 90 + }, + { + "epoch": 0.04, + "learning_rate": 3.1707317073170736e-06, + "loss": 0.73, + "step": 91 + }, + { + "epoch": 0.04, + "learning_rate": 3.205574912891986e-06, + "loss": 0.7109, + "step": 92 + }, + { + "epoch": 0.04, + "learning_rate": 3.240418118466899e-06, + "loss": 0.695, + "step": 93 + }, + { + "epoch": 0.04, + "learning_rate": 3.2752613240418118e-06, + "loss": 0.5556, + "step": 94 + }, + { + "epoch": 0.04, + "learning_rate": 3.310104529616725e-06, + "loss": 0.6237, + "step": 95 + }, + { + "epoch": 0.04, + "learning_rate": 3.3449477351916382e-06, + "loss": 0.6923, + "step": 96 + }, + { + "epoch": 0.04, + "learning_rate": 3.379790940766551e-06, + "loss": 0.7276, + "step": 97 + }, + { + "epoch": 0.04, + "learning_rate": 3.414634146341464e-06, + "loss": 0.6814, + "step": 98 + }, + { + "epoch": 0.04, + "learning_rate": 3.4494773519163764e-06, + "loss": 0.7419, + "step": 99 + }, + { + "epoch": 0.04, + "learning_rate": 3.48432055749129e-06, + "loss": 0.6808, + "step": 100 + }, + { + "epoch": 0.04, + "learning_rate": 3.5191637630662025e-06, + "loss": 0.6811, + "step": 101 + }, + { + "epoch": 0.04, + "learning_rate": 3.5540069686411155e-06, + "loss": 0.6577, + "step": 102 + }, + { + "epoch": 0.04, + "learning_rate": 3.588850174216028e-06, + "loss": 0.6782, + "step": 103 + }, + { + "epoch": 0.04, + "learning_rate": 3.623693379790941e-06, + "loss": 0.7048, + "step": 104 + }, + { + "epoch": 0.04, + "learning_rate": 3.6585365853658537e-06, + "loss": 0.6673, + "step": 105 + }, + { + "epoch": 0.04, + "learning_rate": 3.693379790940767e-06, + "loss": 0.6239, + "step": 106 + }, + { + "epoch": 0.04, + "learning_rate": 3.7282229965156797e-06, + "loss": 0.6954, + "step": 107 + }, + { + "epoch": 0.05, + "learning_rate": 3.7630662020905927e-06, + "loss": 0.6656, + "step": 108 + }, + { + "epoch": 0.05, + "learning_rate": 3.7979094076655053e-06, + "loss": 0.6732, + "step": 109 + }, + { + "epoch": 0.05, + "learning_rate": 3.832752613240418e-06, + "loss": 0.7624, + "step": 110 + }, + { + "epoch": 0.05, + "learning_rate": 3.867595818815331e-06, + "loss": 0.7086, + "step": 111 + }, + { + "epoch": 0.05, + "learning_rate": 3.902439024390244e-06, + "loss": 0.6284, + "step": 112 + }, + { + "epoch": 0.05, + "learning_rate": 3.937282229965157e-06, + "loss": 0.6643, + "step": 113 + }, + { + "epoch": 0.05, + "learning_rate": 3.97212543554007e-06, + "loss": 0.6443, + "step": 114 + }, + { + "epoch": 0.05, + "learning_rate": 4.006968641114983e-06, + "loss": 0.6025, + "step": 115 + }, + { + "epoch": 0.05, + "learning_rate": 4.041811846689896e-06, + "loss": 0.6651, + "step": 116 + }, + { + "epoch": 0.05, + "learning_rate": 4.076655052264808e-06, + "loss": 0.6758, + "step": 117 + }, + { + "epoch": 0.05, + "learning_rate": 4.111498257839722e-06, + "loss": 0.6962, + "step": 118 + }, + { + "epoch": 0.05, + "learning_rate": 4.146341463414634e-06, + "loss": 0.6613, + "step": 119 + }, + { + "epoch": 0.05, + "learning_rate": 4.181184668989548e-06, + "loss": 0.6735, + "step": 120 + }, + { + "epoch": 0.05, + "learning_rate": 4.21602787456446e-06, + "loss": 0.5424, + "step": 121 + }, + { + "epoch": 0.05, + "learning_rate": 4.250871080139373e-06, + "loss": 0.7245, + "step": 122 + }, + { + "epoch": 0.05, + "learning_rate": 4.2857142857142855e-06, + "loss": 0.5873, + "step": 123 + }, + { + "epoch": 0.05, + "learning_rate": 4.320557491289199e-06, + "loss": 0.7426, + "step": 124 + }, + { + "epoch": 0.05, + "learning_rate": 4.3554006968641115e-06, + "loss": 0.7477, + "step": 125 + }, + { + "epoch": 0.05, + "learning_rate": 4.390243902439025e-06, + "loss": 0.6925, + "step": 126 + }, + { + "epoch": 0.05, + "learning_rate": 4.4250871080139375e-06, + "loss": 0.7745, + "step": 127 + }, + { + "epoch": 0.05, + "learning_rate": 4.45993031358885e-06, + "loss": 0.6976, + "step": 128 + }, + { + "epoch": 0.05, + "learning_rate": 4.4947735191637636e-06, + "loss": 0.6296, + "step": 129 + }, + { + "epoch": 0.05, + "learning_rate": 4.529616724738676e-06, + "loss": 0.6011, + "step": 130 + }, + { + "epoch": 0.05, + "learning_rate": 4.56445993031359e-06, + "loss": 0.6661, + "step": 131 + }, + { + "epoch": 0.06, + "learning_rate": 4.599303135888502e-06, + "loss": 0.6491, + "step": 132 + }, + { + "epoch": 0.06, + "learning_rate": 4.634146341463416e-06, + "loss": 0.6139, + "step": 133 + }, + { + "epoch": 0.06, + "learning_rate": 4.668989547038328e-06, + "loss": 0.6212, + "step": 134 + }, + { + "epoch": 0.06, + "learning_rate": 4.703832752613241e-06, + "loss": 0.7824, + "step": 135 + }, + { + "epoch": 0.06, + "learning_rate": 4.738675958188153e-06, + "loss": 0.6765, + "step": 136 + }, + { + "epoch": 0.06, + "learning_rate": 4.773519163763067e-06, + "loss": 0.6975, + "step": 137 + }, + { + "epoch": 0.06, + "learning_rate": 4.8083623693379794e-06, + "loss": 0.7251, + "step": 138 + }, + { + "epoch": 0.06, + "learning_rate": 4.843205574912893e-06, + "loss": 0.7232, + "step": 139 + }, + { + "epoch": 0.06, + "learning_rate": 4.8780487804878055e-06, + "loss": 0.6243, + "step": 140 + }, + { + "epoch": 0.06, + "learning_rate": 4.912891986062718e-06, + "loss": 0.6315, + "step": 141 + }, + { + "epoch": 0.06, + "learning_rate": 4.947735191637631e-06, + "loss": 0.632, + "step": 142 + }, + { + "epoch": 0.06, + "learning_rate": 4.982578397212544e-06, + "loss": 0.7962, + "step": 143 + }, + { + "epoch": 0.06, + "learning_rate": 5.017421602787457e-06, + "loss": 0.5554, + "step": 144 + }, + { + "epoch": 0.06, + "learning_rate": 5.052264808362369e-06, + "loss": 0.6004, + "step": 145 + }, + { + "epoch": 0.06, + "learning_rate": 5.087108013937283e-06, + "loss": 0.5713, + "step": 146 + }, + { + "epoch": 0.06, + "learning_rate": 5.121951219512195e-06, + "loss": 0.6607, + "step": 147 + }, + { + "epoch": 0.06, + "learning_rate": 5.156794425087108e-06, + "loss": 0.627, + "step": 148 + }, + { + "epoch": 0.06, + "learning_rate": 5.1916376306620205e-06, + "loss": 0.6944, + "step": 149 + }, + { + "epoch": 0.06, + "learning_rate": 5.226480836236935e-06, + "loss": 0.6617, + "step": 150 + }, + { + "epoch": 0.06, + "learning_rate": 5.261324041811847e-06, + "loss": 0.6516, + "step": 151 + }, + { + "epoch": 0.06, + "learning_rate": 5.29616724738676e-06, + "loss": 0.6271, + "step": 152 + }, + { + "epoch": 0.06, + "learning_rate": 5.331010452961673e-06, + "loss": 0.5682, + "step": 153 + }, + { + "epoch": 0.06, + "learning_rate": 5.365853658536586e-06, + "loss": 0.6439, + "step": 154 + }, + { + "epoch": 0.06, + "learning_rate": 5.400696864111499e-06, + "loss": 0.7387, + "step": 155 + }, + { + "epoch": 0.07, + "learning_rate": 5.435540069686411e-06, + "loss": 0.5682, + "step": 156 + }, + { + "epoch": 0.07, + "learning_rate": 5.470383275261324e-06, + "loss": 0.5933, + "step": 157 + }, + { + "epoch": 0.07, + "learning_rate": 5.505226480836237e-06, + "loss": 0.6817, + "step": 158 + }, + { + "epoch": 0.07, + "learning_rate": 5.540069686411151e-06, + "loss": 0.6364, + "step": 159 + }, + { + "epoch": 0.07, + "learning_rate": 5.574912891986063e-06, + "loss": 0.5878, + "step": 160 + }, + { + "epoch": 0.07, + "learning_rate": 5.609756097560977e-06, + "loss": 0.6615, + "step": 161 + }, + { + "epoch": 0.07, + "learning_rate": 5.644599303135889e-06, + "loss": 0.6586, + "step": 162 + }, + { + "epoch": 0.07, + "learning_rate": 5.679442508710802e-06, + "loss": 0.6446, + "step": 163 + }, + { + "epoch": 0.07, + "learning_rate": 5.7142857142857145e-06, + "loss": 0.733, + "step": 164 + }, + { + "epoch": 0.07, + "learning_rate": 5.749128919860628e-06, + "loss": 0.6642, + "step": 165 + }, + { + "epoch": 0.07, + "learning_rate": 5.7839721254355405e-06, + "loss": 0.6861, + "step": 166 + }, + { + "epoch": 0.07, + "learning_rate": 5.818815331010453e-06, + "loss": 0.623, + "step": 167 + }, + { + "epoch": 0.07, + "learning_rate": 5.853658536585366e-06, + "loss": 0.5951, + "step": 168 + }, + { + "epoch": 0.07, + "learning_rate": 5.88850174216028e-06, + "loss": 0.6887, + "step": 169 + }, + { + "epoch": 0.07, + "learning_rate": 5.923344947735193e-06, + "loss": 0.5783, + "step": 170 + }, + { + "epoch": 0.07, + "learning_rate": 5.958188153310105e-06, + "loss": 0.6036, + "step": 171 + }, + { + "epoch": 0.07, + "learning_rate": 5.993031358885018e-06, + "loss": 0.6031, + "step": 172 + }, + { + "epoch": 0.07, + "learning_rate": 6.027874564459931e-06, + "loss": 0.659, + "step": 173 + }, + { + "epoch": 0.07, + "learning_rate": 6.062717770034844e-06, + "loss": 0.6425, + "step": 174 + }, + { + "epoch": 0.07, + "learning_rate": 6.0975609756097564e-06, + "loss": 0.5897, + "step": 175 + }, + { + "epoch": 0.07, + "learning_rate": 6.132404181184669e-06, + "loss": 0.6508, + "step": 176 + }, + { + "epoch": 0.07, + "learning_rate": 6.1672473867595825e-06, + "loss": 0.6807, + "step": 177 + }, + { + "epoch": 0.07, + "learning_rate": 6.202090592334495e-06, + "loss": 0.5439, + "step": 178 + }, + { + "epoch": 0.07, + "learning_rate": 6.236933797909408e-06, + "loss": 0.5265, + "step": 179 + }, + { + "epoch": 0.08, + "learning_rate": 6.27177700348432e-06, + "loss": 0.6912, + "step": 180 + }, + { + "epoch": 0.08, + "learning_rate": 6.3066202090592345e-06, + "loss": 0.6348, + "step": 181 + }, + { + "epoch": 0.08, + "learning_rate": 6.341463414634147e-06, + "loss": 0.6374, + "step": 182 + }, + { + "epoch": 0.08, + "learning_rate": 6.37630662020906e-06, + "loss": 0.626, + "step": 183 + }, + { + "epoch": 0.08, + "learning_rate": 6.411149825783972e-06, + "loss": 0.6239, + "step": 184 + }, + { + "epoch": 0.08, + "learning_rate": 6.445993031358886e-06, + "loss": 0.6928, + "step": 185 + }, + { + "epoch": 0.08, + "learning_rate": 6.480836236933798e-06, + "loss": 0.6335, + "step": 186 + }, + { + "epoch": 0.08, + "learning_rate": 6.515679442508711e-06, + "loss": 0.5033, + "step": 187 + }, + { + "epoch": 0.08, + "learning_rate": 6.5505226480836235e-06, + "loss": 0.6619, + "step": 188 + }, + { + "epoch": 0.08, + "learning_rate": 6.585365853658538e-06, + "loss": 0.6189, + "step": 189 + }, + { + "epoch": 0.08, + "learning_rate": 6.62020905923345e-06, + "loss": 0.6277, + "step": 190 + }, + { + "epoch": 0.08, + "learning_rate": 6.655052264808363e-06, + "loss": 0.6073, + "step": 191 + }, + { + "epoch": 0.08, + "learning_rate": 6.6898954703832765e-06, + "loss": 0.6097, + "step": 192 + }, + { + "epoch": 0.08, + "learning_rate": 6.724738675958189e-06, + "loss": 0.5943, + "step": 193 + }, + { + "epoch": 0.08, + "learning_rate": 6.759581881533102e-06, + "loss": 0.5979, + "step": 194 + }, + { + "epoch": 0.08, + "learning_rate": 6.794425087108014e-06, + "loss": 0.598, + "step": 195 + }, + { + "epoch": 0.08, + "learning_rate": 6.829268292682928e-06, + "loss": 0.6136, + "step": 196 + }, + { + "epoch": 0.08, + "learning_rate": 6.86411149825784e-06, + "loss": 0.6153, + "step": 197 + }, + { + "epoch": 0.08, + "learning_rate": 6.898954703832753e-06, + "loss": 0.6817, + "step": 198 + }, + { + "epoch": 0.08, + "learning_rate": 6.9337979094076655e-06, + "loss": 0.5501, + "step": 199 + }, + { + "epoch": 0.08, + "learning_rate": 6.96864111498258e-06, + "loss": 0.7031, + "step": 200 + }, + { + "epoch": 0.08, + "learning_rate": 7.003484320557492e-06, + "loss": 0.5989, + "step": 201 + }, + { + "epoch": 0.08, + "learning_rate": 7.038327526132405e-06, + "loss": 0.6308, + "step": 202 + }, + { + "epoch": 0.08, + "learning_rate": 7.0731707317073175e-06, + "loss": 0.5342, + "step": 203 + }, + { + "epoch": 0.09, + "learning_rate": 7.108013937282231e-06, + "loss": 0.6175, + "step": 204 + }, + { + "epoch": 0.09, + "learning_rate": 7.1428571428571436e-06, + "loss": 0.5792, + "step": 205 + }, + { + "epoch": 0.09, + "learning_rate": 7.177700348432056e-06, + "loss": 0.6591, + "step": 206 + }, + { + "epoch": 0.09, + "learning_rate": 7.212543554006969e-06, + "loss": 0.6793, + "step": 207 + }, + { + "epoch": 0.09, + "learning_rate": 7.247386759581882e-06, + "loss": 0.6891, + "step": 208 + }, + { + "epoch": 0.09, + "learning_rate": 7.282229965156795e-06, + "loss": 0.5279, + "step": 209 + }, + { + "epoch": 0.09, + "learning_rate": 7.317073170731707e-06, + "loss": 0.6229, + "step": 210 + }, + { + "epoch": 0.09, + "learning_rate": 7.35191637630662e-06, + "loss": 0.6168, + "step": 211 + }, + { + "epoch": 0.09, + "learning_rate": 7.386759581881534e-06, + "loss": 0.4843, + "step": 212 + }, + { + "epoch": 0.09, + "learning_rate": 7.421602787456447e-06, + "loss": 0.5995, + "step": 213 + }, + { + "epoch": 0.09, + "learning_rate": 7.4564459930313594e-06, + "loss": 0.5968, + "step": 214 + }, + { + "epoch": 0.09, + "learning_rate": 7.491289198606272e-06, + "loss": 0.5597, + "step": 215 + }, + { + "epoch": 0.09, + "learning_rate": 7.5261324041811855e-06, + "loss": 0.6491, + "step": 216 + }, + { + "epoch": 0.09, + "learning_rate": 7.560975609756098e-06, + "loss": 0.643, + "step": 217 + }, + { + "epoch": 0.09, + "learning_rate": 7.595818815331011e-06, + "loss": 0.6927, + "step": 218 + }, + { + "epoch": 0.09, + "learning_rate": 7.630662020905924e-06, + "loss": 0.5916, + "step": 219 + }, + { + "epoch": 0.09, + "learning_rate": 7.665505226480837e-06, + "loss": 0.5445, + "step": 220 + }, + { + "epoch": 0.09, + "learning_rate": 7.70034843205575e-06, + "loss": 0.6517, + "step": 221 + }, + { + "epoch": 0.09, + "learning_rate": 7.735191637630662e-06, + "loss": 0.6, + "step": 222 + }, + { + "epoch": 0.09, + "learning_rate": 7.770034843205574e-06, + "loss": 0.6441, + "step": 223 + }, + { + "epoch": 0.09, + "learning_rate": 7.804878048780489e-06, + "loss": 0.6266, + "step": 224 + }, + { + "epoch": 0.09, + "learning_rate": 7.839721254355401e-06, + "loss": 0.5312, + "step": 225 + }, + { + "epoch": 0.09, + "learning_rate": 7.874564459930314e-06, + "loss": 0.614, + "step": 226 + }, + { + "epoch": 0.1, + "learning_rate": 7.909407665505228e-06, + "loss": 0.6742, + "step": 227 + }, + { + "epoch": 0.1, + "learning_rate": 7.94425087108014e-06, + "loss": 0.588, + "step": 228 + }, + { + "epoch": 0.1, + "learning_rate": 7.979094076655053e-06, + "loss": 0.5782, + "step": 229 + }, + { + "epoch": 0.1, + "learning_rate": 8.013937282229966e-06, + "loss": 0.6051, + "step": 230 + }, + { + "epoch": 0.1, + "learning_rate": 8.048780487804879e-06, + "loss": 0.671, + "step": 231 + }, + { + "epoch": 0.1, + "learning_rate": 8.083623693379791e-06, + "loss": 0.6003, + "step": 232 + }, + { + "epoch": 0.1, + "learning_rate": 8.118466898954704e-06, + "loss": 0.6023, + "step": 233 + }, + { + "epoch": 0.1, + "learning_rate": 8.153310104529616e-06, + "loss": 0.6305, + "step": 234 + }, + { + "epoch": 0.1, + "learning_rate": 8.18815331010453e-06, + "loss": 0.5449, + "step": 235 + }, + { + "epoch": 0.1, + "learning_rate": 8.222996515679443e-06, + "loss": 0.5957, + "step": 236 + }, + { + "epoch": 0.1, + "learning_rate": 8.257839721254356e-06, + "loss": 0.6335, + "step": 237 + }, + { + "epoch": 0.1, + "learning_rate": 8.292682926829268e-06, + "loss": 0.6214, + "step": 238 + }, + { + "epoch": 0.1, + "learning_rate": 8.327526132404183e-06, + "loss": 0.6096, + "step": 239 + }, + { + "epoch": 0.1, + "learning_rate": 8.362369337979095e-06, + "loss": 0.55, + "step": 240 + }, + { + "epoch": 0.1, + "learning_rate": 8.397212543554008e-06, + "loss": 0.641, + "step": 241 + }, + { + "epoch": 0.1, + "learning_rate": 8.43205574912892e-06, + "loss": 0.6294, + "step": 242 + }, + { + "epoch": 0.1, + "learning_rate": 8.466898954703833e-06, + "loss": 0.6216, + "step": 243 + }, + { + "epoch": 0.1, + "learning_rate": 8.501742160278746e-06, + "loss": 0.5949, + "step": 244 + }, + { + "epoch": 0.1, + "learning_rate": 8.536585365853658e-06, + "loss": 0.568, + "step": 245 + }, + { + "epoch": 0.1, + "learning_rate": 8.571428571428571e-06, + "loss": 0.6762, + "step": 246 + }, + { + "epoch": 0.1, + "learning_rate": 8.606271777003485e-06, + "loss": 0.6068, + "step": 247 + }, + { + "epoch": 0.1, + "learning_rate": 8.641114982578398e-06, + "loss": 0.584, + "step": 248 + }, + { + "epoch": 0.1, + "learning_rate": 8.67595818815331e-06, + "loss": 0.5967, + "step": 249 + }, + { + "epoch": 0.1, + "learning_rate": 8.710801393728223e-06, + "loss": 0.5988, + "step": 250 + }, + { + "epoch": 0.11, + "learning_rate": 8.745644599303137e-06, + "loss": 0.611, + "step": 251 + }, + { + "epoch": 0.11, + "learning_rate": 8.78048780487805e-06, + "loss": 0.6672, + "step": 252 + }, + { + "epoch": 0.11, + "learning_rate": 8.815331010452962e-06, + "loss": 0.5487, + "step": 253 + }, + { + "epoch": 0.11, + "learning_rate": 8.850174216027875e-06, + "loss": 0.5397, + "step": 254 + }, + { + "epoch": 0.11, + "learning_rate": 8.885017421602788e-06, + "loss": 0.7101, + "step": 255 + }, + { + "epoch": 0.11, + "learning_rate": 8.9198606271777e-06, + "loss": 0.6165, + "step": 256 + }, + { + "epoch": 0.11, + "learning_rate": 8.954703832752613e-06, + "loss": 0.5816, + "step": 257 + }, + { + "epoch": 0.11, + "learning_rate": 8.989547038327527e-06, + "loss": 0.6, + "step": 258 + }, + { + "epoch": 0.11, + "learning_rate": 9.02439024390244e-06, + "loss": 0.5065, + "step": 259 + }, + { + "epoch": 0.11, + "learning_rate": 9.059233449477352e-06, + "loss": 0.7077, + "step": 260 + }, + { + "epoch": 0.11, + "learning_rate": 9.094076655052265e-06, + "loss": 0.7149, + "step": 261 + }, + { + "epoch": 0.11, + "learning_rate": 9.12891986062718e-06, + "loss": 0.5832, + "step": 262 + }, + { + "epoch": 0.11, + "learning_rate": 9.163763066202092e-06, + "loss": 0.5587, + "step": 263 + }, + { + "epoch": 0.11, + "learning_rate": 9.198606271777004e-06, + "loss": 0.5781, + "step": 264 + }, + { + "epoch": 0.11, + "learning_rate": 9.233449477351917e-06, + "loss": 0.5852, + "step": 265 + }, + { + "epoch": 0.11, + "learning_rate": 9.268292682926831e-06, + "loss": 0.6214, + "step": 266 + }, + { + "epoch": 0.11, + "learning_rate": 9.303135888501744e-06, + "loss": 0.4923, + "step": 267 + }, + { + "epoch": 0.11, + "learning_rate": 9.337979094076656e-06, + "loss": 0.5257, + "step": 268 + }, + { + "epoch": 0.11, + "learning_rate": 9.372822299651569e-06, + "loss": 0.6273, + "step": 269 + }, + { + "epoch": 0.11, + "learning_rate": 9.407665505226482e-06, + "loss": 0.638, + "step": 270 + }, + { + "epoch": 0.11, + "learning_rate": 9.442508710801394e-06, + "loss": 0.5369, + "step": 271 + }, + { + "epoch": 0.11, + "learning_rate": 9.477351916376307e-06, + "loss": 0.6128, + "step": 272 + }, + { + "epoch": 0.11, + "learning_rate": 9.51219512195122e-06, + "loss": 0.61, + "step": 273 + }, + { + "epoch": 0.11, + "learning_rate": 9.547038327526134e-06, + "loss": 0.6117, + "step": 274 + }, + { + "epoch": 0.12, + "learning_rate": 9.581881533101046e-06, + "loss": 0.5191, + "step": 275 + }, + { + "epoch": 0.12, + "learning_rate": 9.616724738675959e-06, + "loss": 0.5381, + "step": 276 + }, + { + "epoch": 0.12, + "learning_rate": 9.651567944250871e-06, + "loss": 0.6686, + "step": 277 + }, + { + "epoch": 0.12, + "learning_rate": 9.686411149825786e-06, + "loss": 0.6415, + "step": 278 + }, + { + "epoch": 0.12, + "learning_rate": 9.721254355400698e-06, + "loss": 0.5333, + "step": 279 + }, + { + "epoch": 0.12, + "learning_rate": 9.756097560975611e-06, + "loss": 0.6254, + "step": 280 + }, + { + "epoch": 0.12, + "learning_rate": 9.790940766550524e-06, + "loss": 0.5899, + "step": 281 + }, + { + "epoch": 0.12, + "learning_rate": 9.825783972125436e-06, + "loss": 0.5635, + "step": 282 + }, + { + "epoch": 0.12, + "learning_rate": 9.860627177700349e-06, + "loss": 0.5338, + "step": 283 + }, + { + "epoch": 0.12, + "learning_rate": 9.895470383275261e-06, + "loss": 0.6047, + "step": 284 + }, + { + "epoch": 0.12, + "learning_rate": 9.930313588850174e-06, + "loss": 0.6124, + "step": 285 + }, + { + "epoch": 0.12, + "learning_rate": 9.965156794425088e-06, + "loss": 0.5827, + "step": 286 + }, + { + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.52, + "step": 287 + }, + { + "epoch": 0.12, + "learning_rate": 1.0034843205574913e-05, + "loss": 0.6033, + "step": 288 + }, + { + "epoch": 0.12, + "learning_rate": 1.0069686411149826e-05, + "loss": 0.5744, + "step": 289 + }, + { + "epoch": 0.12, + "learning_rate": 1.0104529616724739e-05, + "loss": 0.6823, + "step": 290 + }, + { + "epoch": 0.12, + "learning_rate": 1.0139372822299653e-05, + "loss": 0.6204, + "step": 291 + }, + { + "epoch": 0.12, + "learning_rate": 1.0174216027874565e-05, + "loss": 0.5409, + "step": 292 + }, + { + "epoch": 0.12, + "learning_rate": 1.0209059233449478e-05, + "loss": 0.5721, + "step": 293 + }, + { + "epoch": 0.12, + "learning_rate": 1.024390243902439e-05, + "loss": 0.6182, + "step": 294 + }, + { + "epoch": 0.12, + "learning_rate": 1.0278745644599303e-05, + "loss": 0.579, + "step": 295 + }, + { + "epoch": 0.12, + "learning_rate": 1.0313588850174216e-05, + "loss": 0.6389, + "step": 296 + }, + { + "epoch": 0.12, + "learning_rate": 1.0348432055749128e-05, + "loss": 0.6308, + "step": 297 + }, + { + "epoch": 0.12, + "learning_rate": 1.0383275261324041e-05, + "loss": 0.5235, + "step": 298 + }, + { + "epoch": 0.13, + "learning_rate": 1.0418118466898957e-05, + "loss": 0.5905, + "step": 299 + }, + { + "epoch": 0.13, + "learning_rate": 1.045296167247387e-05, + "loss": 0.6244, + "step": 300 + }, + { + "epoch": 0.13, + "learning_rate": 1.0487804878048782e-05, + "loss": 0.692, + "step": 301 + }, + { + "epoch": 0.13, + "learning_rate": 1.0522648083623695e-05, + "loss": 0.5482, + "step": 302 + }, + { + "epoch": 0.13, + "learning_rate": 1.0557491289198607e-05, + "loss": 0.6385, + "step": 303 + }, + { + "epoch": 0.13, + "learning_rate": 1.059233449477352e-05, + "loss": 0.643, + "step": 304 + }, + { + "epoch": 0.13, + "learning_rate": 1.0627177700348433e-05, + "loss": 0.5917, + "step": 305 + }, + { + "epoch": 0.13, + "learning_rate": 1.0662020905923345e-05, + "loss": 0.6183, + "step": 306 + }, + { + "epoch": 0.13, + "learning_rate": 1.069686411149826e-05, + "loss": 0.557, + "step": 307 + }, + { + "epoch": 0.13, + "learning_rate": 1.0731707317073172e-05, + "loss": 0.6811, + "step": 308 + }, + { + "epoch": 0.13, + "learning_rate": 1.0766550522648085e-05, + "loss": 0.6266, + "step": 309 + }, + { + "epoch": 0.13, + "learning_rate": 1.0801393728222997e-05, + "loss": 0.5146, + "step": 310 + }, + { + "epoch": 0.13, + "learning_rate": 1.083623693379791e-05, + "loss": 0.6395, + "step": 311 + }, + { + "epoch": 0.13, + "learning_rate": 1.0871080139372822e-05, + "loss": 0.6394, + "step": 312 + }, + { + "epoch": 0.13, + "learning_rate": 1.0905923344947735e-05, + "loss": 0.5787, + "step": 313 + }, + { + "epoch": 0.13, + "learning_rate": 1.0940766550522648e-05, + "loss": 0.5919, + "step": 314 + }, + { + "epoch": 0.13, + "learning_rate": 1.0975609756097562e-05, + "loss": 0.5283, + "step": 315 + }, + { + "epoch": 0.13, + "learning_rate": 1.1010452961672475e-05, + "loss": 0.5572, + "step": 316 + }, + { + "epoch": 0.13, + "learning_rate": 1.1045296167247389e-05, + "loss": 0.5774, + "step": 317 + }, + { + "epoch": 0.13, + "learning_rate": 1.1080139372822301e-05, + "loss": 0.5864, + "step": 318 + }, + { + "epoch": 0.13, + "learning_rate": 1.1114982578397214e-05, + "loss": 0.5731, + "step": 319 + }, + { + "epoch": 0.13, + "learning_rate": 1.1149825783972127e-05, + "loss": 0.5627, + "step": 320 + }, + { + "epoch": 0.13, + "learning_rate": 1.118466898954704e-05, + "loss": 0.5659, + "step": 321 + }, + { + "epoch": 0.13, + "learning_rate": 1.1219512195121953e-05, + "loss": 0.6295, + "step": 322 + }, + { + "epoch": 0.14, + "learning_rate": 1.1254355400696866e-05, + "loss": 0.5662, + "step": 323 + }, + { + "epoch": 0.14, + "learning_rate": 1.1289198606271779e-05, + "loss": 0.5659, + "step": 324 + }, + { + "epoch": 0.14, + "learning_rate": 1.1324041811846691e-05, + "loss": 0.74, + "step": 325 + }, + { + "epoch": 0.14, + "learning_rate": 1.1358885017421604e-05, + "loss": 0.604, + "step": 326 + }, + { + "epoch": 0.14, + "learning_rate": 1.1393728222996516e-05, + "loss": 0.5228, + "step": 327 + }, + { + "epoch": 0.14, + "learning_rate": 1.1428571428571429e-05, + "loss": 0.5745, + "step": 328 + }, + { + "epoch": 0.14, + "learning_rate": 1.1463414634146342e-05, + "loss": 0.5464, + "step": 329 + }, + { + "epoch": 0.14, + "learning_rate": 1.1498257839721256e-05, + "loss": 0.5523, + "step": 330 + }, + { + "epoch": 0.14, + "learning_rate": 1.1533101045296168e-05, + "loss": 0.5893, + "step": 331 + }, + { + "epoch": 0.14, + "learning_rate": 1.1567944250871081e-05, + "loss": 0.6265, + "step": 332 + }, + { + "epoch": 0.14, + "learning_rate": 1.1602787456445994e-05, + "loss": 0.6626, + "step": 333 + }, + { + "epoch": 0.14, + "learning_rate": 1.1637630662020906e-05, + "loss": 0.6379, + "step": 334 + }, + { + "epoch": 0.14, + "learning_rate": 1.1672473867595819e-05, + "loss": 0.6768, + "step": 335 + }, + { + "epoch": 0.14, + "learning_rate": 1.1707317073170731e-05, + "loss": 0.5926, + "step": 336 + }, + { + "epoch": 0.14, + "learning_rate": 1.1742160278745644e-05, + "loss": 0.5531, + "step": 337 + }, + { + "epoch": 0.14, + "learning_rate": 1.177700348432056e-05, + "loss": 0.5478, + "step": 338 + }, + { + "epoch": 0.14, + "learning_rate": 1.1811846689895473e-05, + "loss": 0.5894, + "step": 339 + }, + { + "epoch": 0.14, + "learning_rate": 1.1846689895470385e-05, + "loss": 0.5819, + "step": 340 + }, + { + "epoch": 0.14, + "learning_rate": 1.1881533101045298e-05, + "loss": 0.4793, + "step": 341 + }, + { + "epoch": 0.14, + "learning_rate": 1.191637630662021e-05, + "loss": 0.6046, + "step": 342 + }, + { + "epoch": 0.14, + "learning_rate": 1.1951219512195123e-05, + "loss": 0.5306, + "step": 343 + }, + { + "epoch": 0.14, + "learning_rate": 1.1986062717770036e-05, + "loss": 0.6153, + "step": 344 + }, + { + "epoch": 0.14, + "learning_rate": 1.2020905923344948e-05, + "loss": 0.6093, + "step": 345 + }, + { + "epoch": 0.14, + "learning_rate": 1.2055749128919862e-05, + "loss": 0.6022, + "step": 346 + }, + { + "epoch": 0.15, + "learning_rate": 1.2090592334494775e-05, + "loss": 0.515, + "step": 347 + }, + { + "epoch": 0.15, + "learning_rate": 1.2125435540069688e-05, + "loss": 0.6889, + "step": 348 + }, + { + "epoch": 0.15, + "learning_rate": 1.21602787456446e-05, + "loss": 0.576, + "step": 349 + }, + { + "epoch": 0.15, + "learning_rate": 1.2195121951219513e-05, + "loss": 0.5454, + "step": 350 + }, + { + "epoch": 0.15, + "learning_rate": 1.2229965156794425e-05, + "loss": 0.4593, + "step": 351 + }, + { + "epoch": 0.15, + "learning_rate": 1.2264808362369338e-05, + "loss": 0.5302, + "step": 352 + }, + { + "epoch": 0.15, + "learning_rate": 1.2299651567944252e-05, + "loss": 0.5591, + "step": 353 + }, + { + "epoch": 0.15, + "learning_rate": 1.2334494773519165e-05, + "loss": 0.6179, + "step": 354 + }, + { + "epoch": 0.15, + "learning_rate": 1.2369337979094078e-05, + "loss": 0.6015, + "step": 355 + }, + { + "epoch": 0.15, + "learning_rate": 1.240418118466899e-05, + "loss": 0.5615, + "step": 356 + }, + { + "epoch": 0.15, + "learning_rate": 1.2439024390243903e-05, + "loss": 0.6798, + "step": 357 + }, + { + "epoch": 0.15, + "learning_rate": 1.2473867595818815e-05, + "loss": 0.578, + "step": 358 + }, + { + "epoch": 0.15, + "learning_rate": 1.2508710801393728e-05, + "loss": 0.629, + "step": 359 + }, + { + "epoch": 0.15, + "learning_rate": 1.254355400696864e-05, + "loss": 0.6236, + "step": 360 + }, + { + "epoch": 0.15, + "learning_rate": 1.2578397212543556e-05, + "loss": 0.577, + "step": 361 + }, + { + "epoch": 0.15, + "learning_rate": 1.2613240418118469e-05, + "loss": 0.5588, + "step": 362 + }, + { + "epoch": 0.15, + "learning_rate": 1.2648083623693382e-05, + "loss": 0.6655, + "step": 363 + }, + { + "epoch": 0.15, + "learning_rate": 1.2682926829268294e-05, + "loss": 0.6124, + "step": 364 + }, + { + "epoch": 0.15, + "learning_rate": 1.2717770034843207e-05, + "loss": 0.4786, + "step": 365 + }, + { + "epoch": 0.15, + "learning_rate": 1.275261324041812e-05, + "loss": 0.566, + "step": 366 + }, + { + "epoch": 0.15, + "learning_rate": 1.2787456445993032e-05, + "loss": 0.5676, + "step": 367 + }, + { + "epoch": 0.15, + "learning_rate": 1.2822299651567945e-05, + "loss": 0.6284, + "step": 368 + }, + { + "epoch": 0.15, + "learning_rate": 1.2857142857142859e-05, + "loss": 0.5828, + "step": 369 + }, + { + "epoch": 0.15, + "learning_rate": 1.2891986062717772e-05, + "loss": 0.6464, + "step": 370 + }, + { + "epoch": 0.16, + "learning_rate": 1.2926829268292684e-05, + "loss": 0.5833, + "step": 371 + }, + { + "epoch": 0.16, + "learning_rate": 1.2961672473867597e-05, + "loss": 0.578, + "step": 372 + }, + { + "epoch": 0.16, + "learning_rate": 1.299651567944251e-05, + "loss": 0.5904, + "step": 373 + }, + { + "epoch": 0.16, + "learning_rate": 1.3031358885017422e-05, + "loss": 0.5454, + "step": 374 + }, + { + "epoch": 0.16, + "learning_rate": 1.3066202090592334e-05, + "loss": 0.5759, + "step": 375 + }, + { + "epoch": 0.16, + "learning_rate": 1.3101045296167247e-05, + "loss": 0.5288, + "step": 376 + }, + { + "epoch": 0.16, + "learning_rate": 1.3135888501742163e-05, + "loss": 0.6116, + "step": 377 + }, + { + "epoch": 0.16, + "learning_rate": 1.3170731707317076e-05, + "loss": 0.5446, + "step": 378 + }, + { + "epoch": 0.16, + "learning_rate": 1.3205574912891988e-05, + "loss": 0.5642, + "step": 379 + }, + { + "epoch": 0.16, + "learning_rate": 1.32404181184669e-05, + "loss": 0.5638, + "step": 380 + }, + { + "epoch": 0.16, + "learning_rate": 1.3275261324041813e-05, + "loss": 0.779, + "step": 381 + }, + { + "epoch": 0.16, + "learning_rate": 1.3310104529616726e-05, + "loss": 0.5757, + "step": 382 + }, + { + "epoch": 0.16, + "learning_rate": 1.3344947735191639e-05, + "loss": 0.6283, + "step": 383 + }, + { + "epoch": 0.16, + "learning_rate": 1.3379790940766553e-05, + "loss": 0.6289, + "step": 384 + }, + { + "epoch": 0.16, + "learning_rate": 1.3414634146341466e-05, + "loss": 0.5548, + "step": 385 + }, + { + "epoch": 0.16, + "learning_rate": 1.3449477351916378e-05, + "loss": 0.5725, + "step": 386 + }, + { + "epoch": 0.16, + "learning_rate": 1.348432055749129e-05, + "loss": 0.6109, + "step": 387 + }, + { + "epoch": 0.16, + "learning_rate": 1.3519163763066203e-05, + "loss": 0.5975, + "step": 388 + }, + { + "epoch": 0.16, + "learning_rate": 1.3554006968641116e-05, + "loss": 0.5972, + "step": 389 + }, + { + "epoch": 0.16, + "learning_rate": 1.3588850174216028e-05, + "loss": 0.6673, + "step": 390 + }, + { + "epoch": 0.16, + "learning_rate": 1.3623693379790941e-05, + "loss": 0.6687, + "step": 391 + }, + { + "epoch": 0.16, + "learning_rate": 1.3658536585365855e-05, + "loss": 0.504, + "step": 392 + }, + { + "epoch": 0.16, + "learning_rate": 1.3693379790940768e-05, + "loss": 0.6478, + "step": 393 + }, + { + "epoch": 0.16, + "learning_rate": 1.372822299651568e-05, + "loss": 0.6042, + "step": 394 + }, + { + "epoch": 0.17, + "learning_rate": 1.3763066202090593e-05, + "loss": 0.6066, + "step": 395 + }, + { + "epoch": 0.17, + "learning_rate": 1.3797909407665506e-05, + "loss": 0.5724, + "step": 396 + }, + { + "epoch": 0.17, + "learning_rate": 1.3832752613240418e-05, + "loss": 0.6821, + "step": 397 + }, + { + "epoch": 0.17, + "learning_rate": 1.3867595818815331e-05, + "loss": 0.544, + "step": 398 + }, + { + "epoch": 0.17, + "learning_rate": 1.3902439024390244e-05, + "loss": 0.5192, + "step": 399 + }, + { + "epoch": 0.17, + "learning_rate": 1.393728222996516e-05, + "loss": 0.6457, + "step": 400 + }, + { + "epoch": 0.17, + "learning_rate": 1.3972125435540072e-05, + "loss": 0.5996, + "step": 401 + }, + { + "epoch": 0.17, + "learning_rate": 1.4006968641114985e-05, + "loss": 0.5737, + "step": 402 + }, + { + "epoch": 0.17, + "learning_rate": 1.4041811846689897e-05, + "loss": 0.6506, + "step": 403 + }, + { + "epoch": 0.17, + "learning_rate": 1.407665505226481e-05, + "loss": 0.5541, + "step": 404 + }, + { + "epoch": 0.17, + "learning_rate": 1.4111498257839722e-05, + "loss": 0.5502, + "step": 405 + }, + { + "epoch": 0.17, + "learning_rate": 1.4146341463414635e-05, + "loss": 0.5909, + "step": 406 + }, + { + "epoch": 0.17, + "learning_rate": 1.4181184668989548e-05, + "loss": 0.5953, + "step": 407 + }, + { + "epoch": 0.17, + "learning_rate": 1.4216027874564462e-05, + "loss": 0.5376, + "step": 408 + }, + { + "epoch": 0.17, + "learning_rate": 1.4250871080139375e-05, + "loss": 0.5377, + "step": 409 + }, + { + "epoch": 0.17, + "learning_rate": 1.4285714285714287e-05, + "loss": 0.6134, + "step": 410 + }, + { + "epoch": 0.17, + "learning_rate": 1.43205574912892e-05, + "loss": 0.6472, + "step": 411 + }, + { + "epoch": 0.17, + "learning_rate": 1.4355400696864112e-05, + "loss": 0.4815, + "step": 412 + }, + { + "epoch": 0.17, + "learning_rate": 1.4390243902439025e-05, + "loss": 0.4832, + "step": 413 + }, + { + "epoch": 0.17, + "learning_rate": 1.4425087108013938e-05, + "loss": 0.5455, + "step": 414 + }, + { + "epoch": 0.17, + "learning_rate": 1.4459930313588852e-05, + "loss": 0.5445, + "step": 415 + }, + { + "epoch": 0.17, + "learning_rate": 1.4494773519163764e-05, + "loss": 0.5713, + "step": 416 + }, + { + "epoch": 0.17, + "learning_rate": 1.4529616724738677e-05, + "loss": 0.5713, + "step": 417 + }, + { + "epoch": 0.18, + "learning_rate": 1.456445993031359e-05, + "loss": 0.6725, + "step": 418 + }, + { + "epoch": 0.18, + "learning_rate": 1.4599303135888502e-05, + "loss": 0.6199, + "step": 419 + }, + { + "epoch": 0.18, + "learning_rate": 1.4634146341463415e-05, + "loss": 0.5049, + "step": 420 + }, + { + "epoch": 0.18, + "learning_rate": 1.4668989547038327e-05, + "loss": 0.5712, + "step": 421 + }, + { + "epoch": 0.18, + "learning_rate": 1.470383275261324e-05, + "loss": 0.6426, + "step": 422 + }, + { + "epoch": 0.18, + "learning_rate": 1.4738675958188156e-05, + "loss": 0.5879, + "step": 423 + }, + { + "epoch": 0.18, + "learning_rate": 1.4773519163763069e-05, + "loss": 0.5529, + "step": 424 + }, + { + "epoch": 0.18, + "learning_rate": 1.4808362369337981e-05, + "loss": 0.6029, + "step": 425 + }, + { + "epoch": 0.18, + "learning_rate": 1.4843205574912894e-05, + "loss": 0.6285, + "step": 426 + }, + { + "epoch": 0.18, + "learning_rate": 1.4878048780487806e-05, + "loss": 0.5832, + "step": 427 + }, + { + "epoch": 0.18, + "learning_rate": 1.4912891986062719e-05, + "loss": 0.506, + "step": 428 + }, + { + "epoch": 0.18, + "learning_rate": 1.4947735191637631e-05, + "loss": 0.6044, + "step": 429 + }, + { + "epoch": 0.18, + "learning_rate": 1.4982578397212544e-05, + "loss": 0.5128, + "step": 430 + }, + { + "epoch": 0.18, + "learning_rate": 1.5017421602787458e-05, + "loss": 0.5998, + "step": 431 + }, + { + "epoch": 0.18, + "learning_rate": 1.5052264808362371e-05, + "loss": 0.6959, + "step": 432 + }, + { + "epoch": 0.18, + "learning_rate": 1.5087108013937284e-05, + "loss": 0.6206, + "step": 433 + }, + { + "epoch": 0.18, + "learning_rate": 1.5121951219512196e-05, + "loss": 0.6337, + "step": 434 + }, + { + "epoch": 0.18, + "learning_rate": 1.5156794425087109e-05, + "loss": 0.5991, + "step": 435 + }, + { + "epoch": 0.18, + "learning_rate": 1.5191637630662021e-05, + "loss": 0.6201, + "step": 436 + }, + { + "epoch": 0.18, + "learning_rate": 1.5226480836236934e-05, + "loss": 0.4951, + "step": 437 + }, + { + "epoch": 0.18, + "learning_rate": 1.5261324041811848e-05, + "loss": 0.5876, + "step": 438 + }, + { + "epoch": 0.18, + "learning_rate": 1.529616724738676e-05, + "loss": 0.5744, + "step": 439 + }, + { + "epoch": 0.18, + "learning_rate": 1.5331010452961673e-05, + "loss": 0.5939, + "step": 440 + }, + { + "epoch": 0.18, + "learning_rate": 1.5365853658536586e-05, + "loss": 0.6046, + "step": 441 + }, + { + "epoch": 0.19, + "learning_rate": 1.54006968641115e-05, + "loss": 0.656, + "step": 442 + }, + { + "epoch": 0.19, + "learning_rate": 1.543554006968641e-05, + "loss": 0.6, + "step": 443 + }, + { + "epoch": 0.19, + "learning_rate": 1.5470383275261324e-05, + "loss": 0.5162, + "step": 444 + }, + { + "epoch": 0.19, + "learning_rate": 1.5505226480836236e-05, + "loss": 0.5459, + "step": 445 + }, + { + "epoch": 0.19, + "learning_rate": 1.554006968641115e-05, + "loss": 0.486, + "step": 446 + }, + { + "epoch": 0.19, + "learning_rate": 1.5574912891986065e-05, + "loss": 0.6013, + "step": 447 + }, + { + "epoch": 0.19, + "learning_rate": 1.5609756097560978e-05, + "loss": 0.6249, + "step": 448 + }, + { + "epoch": 0.19, + "learning_rate": 1.564459930313589e-05, + "loss": 0.545, + "step": 449 + }, + { + "epoch": 0.19, + "learning_rate": 1.5679442508710803e-05, + "loss": 0.6635, + "step": 450 + }, + { + "epoch": 0.19, + "learning_rate": 1.5714285714285715e-05, + "loss": 0.6335, + "step": 451 + }, + { + "epoch": 0.19, + "learning_rate": 1.5749128919860628e-05, + "loss": 0.5401, + "step": 452 + }, + { + "epoch": 0.19, + "learning_rate": 1.578397212543554e-05, + "loss": 0.5021, + "step": 453 + }, + { + "epoch": 0.19, + "learning_rate": 1.5818815331010456e-05, + "loss": 0.5869, + "step": 454 + }, + { + "epoch": 0.19, + "learning_rate": 1.585365853658537e-05, + "loss": 0.6381, + "step": 455 + }, + { + "epoch": 0.19, + "learning_rate": 1.588850174216028e-05, + "loss": 0.4976, + "step": 456 + }, + { + "epoch": 0.19, + "learning_rate": 1.5923344947735194e-05, + "loss": 0.5761, + "step": 457 + }, + { + "epoch": 0.19, + "learning_rate": 1.5958188153310107e-05, + "loss": 0.6614, + "step": 458 + }, + { + "epoch": 0.19, + "learning_rate": 1.599303135888502e-05, + "loss": 0.6228, + "step": 459 + }, + { + "epoch": 0.19, + "learning_rate": 1.6027874564459932e-05, + "loss": 0.5755, + "step": 460 + }, + { + "epoch": 0.19, + "learning_rate": 1.6062717770034845e-05, + "loss": 0.5943, + "step": 461 + }, + { + "epoch": 0.19, + "learning_rate": 1.6097560975609757e-05, + "loss": 0.5631, + "step": 462 + }, + { + "epoch": 0.19, + "learning_rate": 1.613240418118467e-05, + "loss": 0.6802, + "step": 463 + }, + { + "epoch": 0.19, + "learning_rate": 1.6167247386759582e-05, + "loss": 0.5835, + "step": 464 + }, + { + "epoch": 0.19, + "learning_rate": 1.6202090592334495e-05, + "loss": 0.5725, + "step": 465 + }, + { + "epoch": 0.2, + "learning_rate": 1.6236933797909408e-05, + "loss": 0.5487, + "step": 466 + }, + { + "epoch": 0.2, + "learning_rate": 1.627177700348432e-05, + "loss": 0.5889, + "step": 467 + }, + { + "epoch": 0.2, + "learning_rate": 1.6306620209059233e-05, + "loss": 0.5585, + "step": 468 + }, + { + "epoch": 0.2, + "learning_rate": 1.6341463414634145e-05, + "loss": 0.5731, + "step": 469 + }, + { + "epoch": 0.2, + "learning_rate": 1.637630662020906e-05, + "loss": 0.6129, + "step": 470 + }, + { + "epoch": 0.2, + "learning_rate": 1.6411149825783974e-05, + "loss": 0.586, + "step": 471 + }, + { + "epoch": 0.2, + "learning_rate": 1.6445993031358887e-05, + "loss": 0.6418, + "step": 472 + }, + { + "epoch": 0.2, + "learning_rate": 1.64808362369338e-05, + "loss": 0.6442, + "step": 473 + }, + { + "epoch": 0.2, + "learning_rate": 1.6515679442508712e-05, + "loss": 0.5419, + "step": 474 + }, + { + "epoch": 0.2, + "learning_rate": 1.6550522648083624e-05, + "loss": 0.5282, + "step": 475 + }, + { + "epoch": 0.2, + "learning_rate": 1.6585365853658537e-05, + "loss": 0.5588, + "step": 476 + }, + { + "epoch": 0.2, + "learning_rate": 1.662020905923345e-05, + "loss": 0.5536, + "step": 477 + }, + { + "epoch": 0.2, + "learning_rate": 1.6655052264808366e-05, + "loss": 0.5514, + "step": 478 + }, + { + "epoch": 0.2, + "learning_rate": 1.6689895470383278e-05, + "loss": 0.5554, + "step": 479 + }, + { + "epoch": 0.2, + "learning_rate": 1.672473867595819e-05, + "loss": 0.555, + "step": 480 + }, + { + "epoch": 0.2, + "learning_rate": 1.6759581881533103e-05, + "loss": 0.5606, + "step": 481 + }, + { + "epoch": 0.2, + "learning_rate": 1.6794425087108016e-05, + "loss": 0.5154, + "step": 482 + }, + { + "epoch": 0.2, + "learning_rate": 1.682926829268293e-05, + "loss": 0.5407, + "step": 483 + }, + { + "epoch": 0.2, + "learning_rate": 1.686411149825784e-05, + "loss": 0.6047, + "step": 484 + }, + { + "epoch": 0.2, + "learning_rate": 1.6898954703832754e-05, + "loss": 0.5323, + "step": 485 + }, + { + "epoch": 0.2, + "learning_rate": 1.6933797909407666e-05, + "loss": 0.5395, + "step": 486 + }, + { + "epoch": 0.2, + "learning_rate": 1.696864111498258e-05, + "loss": 0.6181, + "step": 487 + }, + { + "epoch": 0.2, + "learning_rate": 1.700348432055749e-05, + "loss": 0.5764, + "step": 488 + }, + { + "epoch": 0.2, + "learning_rate": 1.7038327526132404e-05, + "loss": 0.6155, + "step": 489 + }, + { + "epoch": 0.21, + "learning_rate": 1.7073170731707317e-05, + "loss": 0.5058, + "step": 490 + }, + { + "epoch": 0.21, + "learning_rate": 1.710801393728223e-05, + "loss": 0.6552, + "step": 491 + }, + { + "epoch": 0.21, + "learning_rate": 1.7142857142857142e-05, + "loss": 0.5362, + "step": 492 + }, + { + "epoch": 0.21, + "learning_rate": 1.7177700348432058e-05, + "loss": 0.6479, + "step": 493 + }, + { + "epoch": 0.21, + "learning_rate": 1.721254355400697e-05, + "loss": 0.4939, + "step": 494 + }, + { + "epoch": 0.21, + "learning_rate": 1.7247386759581883e-05, + "loss": 0.6442, + "step": 495 + }, + { + "epoch": 0.21, + "learning_rate": 1.7282229965156796e-05, + "loss": 0.5753, + "step": 496 + }, + { + "epoch": 0.21, + "learning_rate": 1.7317073170731708e-05, + "loss": 0.701, + "step": 497 + }, + { + "epoch": 0.21, + "learning_rate": 1.735191637630662e-05, + "loss": 0.5235, + "step": 498 + }, + { + "epoch": 0.21, + "learning_rate": 1.7386759581881533e-05, + "loss": 0.4857, + "step": 499 + }, + { + "epoch": 0.21, + "learning_rate": 1.7421602787456446e-05, + "loss": 0.5072, + "step": 500 + }, + { + "epoch": 0.21, + "learning_rate": 1.7456445993031362e-05, + "loss": 0.6019, + "step": 501 + }, + { + "epoch": 0.21, + "learning_rate": 1.7491289198606275e-05, + "loss": 0.5263, + "step": 502 + }, + { + "epoch": 0.21, + "learning_rate": 1.7526132404181187e-05, + "loss": 0.5893, + "step": 503 + }, + { + "epoch": 0.21, + "learning_rate": 1.75609756097561e-05, + "loss": 0.6769, + "step": 504 + }, + { + "epoch": 0.21, + "learning_rate": 1.7595818815331012e-05, + "loss": 0.5069, + "step": 505 + }, + { + "epoch": 0.21, + "learning_rate": 1.7630662020905925e-05, + "loss": 0.6241, + "step": 506 + }, + { + "epoch": 0.21, + "learning_rate": 1.7665505226480838e-05, + "loss": 0.5585, + "step": 507 + }, + { + "epoch": 0.21, + "learning_rate": 1.770034843205575e-05, + "loss": 0.5546, + "step": 508 + }, + { + "epoch": 0.21, + "learning_rate": 1.7735191637630663e-05, + "loss": 0.5824, + "step": 509 + }, + { + "epoch": 0.21, + "learning_rate": 1.7770034843205575e-05, + "loss": 0.5741, + "step": 510 + }, + { + "epoch": 0.21, + "learning_rate": 1.7804878048780488e-05, + "loss": 0.5867, + "step": 511 + }, + { + "epoch": 0.21, + "learning_rate": 1.78397212543554e-05, + "loss": 0.5562, + "step": 512 + }, + { + "epoch": 0.21, + "learning_rate": 1.7874564459930313e-05, + "loss": 0.5487, + "step": 513 + }, + { + "epoch": 0.22, + "learning_rate": 1.7909407665505226e-05, + "loss": 0.6476, + "step": 514 + }, + { + "epoch": 0.22, + "learning_rate": 1.7944250871080138e-05, + "loss": 0.5481, + "step": 515 + }, + { + "epoch": 0.22, + "learning_rate": 1.7979094076655054e-05, + "loss": 0.5381, + "step": 516 + }, + { + "epoch": 0.22, + "learning_rate": 1.8013937282229967e-05, + "loss": 0.5591, + "step": 517 + }, + { + "epoch": 0.22, + "learning_rate": 1.804878048780488e-05, + "loss": 0.591, + "step": 518 + }, + { + "epoch": 0.22, + "learning_rate": 1.8083623693379792e-05, + "loss": 0.465, + "step": 519 + }, + { + "epoch": 0.22, + "learning_rate": 1.8118466898954705e-05, + "loss": 0.6022, + "step": 520 + }, + { + "epoch": 0.22, + "learning_rate": 1.8153310104529617e-05, + "loss": 0.4784, + "step": 521 + }, + { + "epoch": 0.22, + "learning_rate": 1.818815331010453e-05, + "loss": 0.5983, + "step": 522 + }, + { + "epoch": 0.22, + "learning_rate": 1.8222996515679442e-05, + "loss": 0.5676, + "step": 523 + }, + { + "epoch": 0.22, + "learning_rate": 1.825783972125436e-05, + "loss": 0.5733, + "step": 524 + }, + { + "epoch": 0.22, + "learning_rate": 1.829268292682927e-05, + "loss": 0.5837, + "step": 525 + }, + { + "epoch": 0.22, + "learning_rate": 1.8327526132404184e-05, + "loss": 0.5893, + "step": 526 + }, + { + "epoch": 0.22, + "learning_rate": 1.8362369337979096e-05, + "loss": 0.6748, + "step": 527 + }, + { + "epoch": 0.22, + "learning_rate": 1.839721254355401e-05, + "loss": 0.5764, + "step": 528 + }, + { + "epoch": 0.22, + "learning_rate": 1.843205574912892e-05, + "loss": 0.6803, + "step": 529 + }, + { + "epoch": 0.22, + "learning_rate": 1.8466898954703834e-05, + "loss": 0.5717, + "step": 530 + }, + { + "epoch": 0.22, + "learning_rate": 1.8501742160278747e-05, + "loss": 0.5755, + "step": 531 + }, + { + "epoch": 0.22, + "learning_rate": 1.8536585365853663e-05, + "loss": 0.5581, + "step": 532 + }, + { + "epoch": 0.22, + "learning_rate": 1.8571428571428575e-05, + "loss": 0.5884, + "step": 533 + }, + { + "epoch": 0.22, + "learning_rate": 1.8606271777003488e-05, + "loss": 0.5363, + "step": 534 + }, + { + "epoch": 0.22, + "learning_rate": 1.86411149825784e-05, + "loss": 0.5798, + "step": 535 + }, + { + "epoch": 0.22, + "learning_rate": 1.8675958188153313e-05, + "loss": 0.5236, + "step": 536 + }, + { + "epoch": 0.22, + "learning_rate": 1.8710801393728226e-05, + "loss": 0.5025, + "step": 537 + }, + { + "epoch": 0.23, + "learning_rate": 1.8745644599303138e-05, + "loss": 0.6242, + "step": 538 + }, + { + "epoch": 0.23, + "learning_rate": 1.878048780487805e-05, + "loss": 0.6023, + "step": 539 + }, + { + "epoch": 0.23, + "learning_rate": 1.8815331010452963e-05, + "loss": 0.6211, + "step": 540 + }, + { + "epoch": 0.23, + "learning_rate": 1.8850174216027876e-05, + "loss": 0.5077, + "step": 541 + }, + { + "epoch": 0.23, + "learning_rate": 1.888501742160279e-05, + "loss": 0.5018, + "step": 542 + }, + { + "epoch": 0.23, + "learning_rate": 1.89198606271777e-05, + "loss": 0.6053, + "step": 543 + }, + { + "epoch": 0.23, + "learning_rate": 1.8954703832752614e-05, + "loss": 0.6225, + "step": 544 + }, + { + "epoch": 0.23, + "learning_rate": 1.8989547038327526e-05, + "loss": 0.6134, + "step": 545 + }, + { + "epoch": 0.23, + "learning_rate": 1.902439024390244e-05, + "loss": 0.5871, + "step": 546 + }, + { + "epoch": 0.23, + "learning_rate": 1.9059233449477355e-05, + "loss": 0.7059, + "step": 547 + }, + { + "epoch": 0.23, + "learning_rate": 1.9094076655052267e-05, + "loss": 0.5859, + "step": 548 + }, + { + "epoch": 0.23, + "learning_rate": 1.912891986062718e-05, + "loss": 0.5957, + "step": 549 + }, + { + "epoch": 0.23, + "learning_rate": 1.9163763066202093e-05, + "loss": 0.5299, + "step": 550 + }, + { + "epoch": 0.23, + "learning_rate": 1.9198606271777005e-05, + "loss": 0.6158, + "step": 551 + }, + { + "epoch": 0.23, + "learning_rate": 1.9233449477351918e-05, + "loss": 0.5808, + "step": 552 + }, + { + "epoch": 0.23, + "learning_rate": 1.926829268292683e-05, + "loss": 0.573, + "step": 553 + }, + { + "epoch": 0.23, + "learning_rate": 1.9303135888501743e-05, + "loss": 0.6362, + "step": 554 + }, + { + "epoch": 0.23, + "learning_rate": 1.933797909407666e-05, + "loss": 0.5059, + "step": 555 + }, + { + "epoch": 0.23, + "learning_rate": 1.937282229965157e-05, + "loss": 0.5784, + "step": 556 + }, + { + "epoch": 0.23, + "learning_rate": 1.9407665505226484e-05, + "loss": 0.649, + "step": 557 + }, + { + "epoch": 0.23, + "learning_rate": 1.9442508710801397e-05, + "loss": 0.6128, + "step": 558 + }, + { + "epoch": 0.23, + "learning_rate": 1.947735191637631e-05, + "loss": 0.595, + "step": 559 + }, + { + "epoch": 0.23, + "learning_rate": 1.9512195121951222e-05, + "loss": 0.5921, + "step": 560 + }, + { + "epoch": 0.23, + "learning_rate": 1.9547038327526135e-05, + "loss": 0.5053, + "step": 561 + }, + { + "epoch": 0.24, + "learning_rate": 1.9581881533101047e-05, + "loss": 0.4499, + "step": 562 + }, + { + "epoch": 0.24, + "learning_rate": 1.961672473867596e-05, + "loss": 0.5918, + "step": 563 + }, + { + "epoch": 0.24, + "learning_rate": 1.9651567944250872e-05, + "loss": 0.5511, + "step": 564 + }, + { + "epoch": 0.24, + "learning_rate": 1.9686411149825785e-05, + "loss": 0.6467, + "step": 565 + }, + { + "epoch": 0.24, + "learning_rate": 1.9721254355400697e-05, + "loss": 0.5508, + "step": 566 + }, + { + "epoch": 0.24, + "learning_rate": 1.975609756097561e-05, + "loss": 0.5592, + "step": 567 + }, + { + "epoch": 0.24, + "learning_rate": 1.9790940766550523e-05, + "loss": 0.5966, + "step": 568 + }, + { + "epoch": 0.24, + "learning_rate": 1.9825783972125435e-05, + "loss": 0.5011, + "step": 569 + }, + { + "epoch": 0.24, + "learning_rate": 1.9860627177700348e-05, + "loss": 0.593, + "step": 570 + }, + { + "epoch": 0.24, + "learning_rate": 1.9895470383275264e-05, + "loss": 0.5659, + "step": 571 + }, + { + "epoch": 0.24, + "learning_rate": 1.9930313588850176e-05, + "loss": 0.5767, + "step": 572 + }, + { + "epoch": 0.24, + "learning_rate": 1.996515679442509e-05, + "loss": 0.5624, + "step": 573 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.4904, + "step": 574 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999985627944e-05, + "loss": 0.5734, + "step": 575 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999999425117754e-05, + "loss": 0.5985, + "step": 576 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999998706514958e-05, + "loss": 0.5729, + "step": 577 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999997700471074e-05, + "loss": 0.558, + "step": 578 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999640698613e-05, + "loss": 0.5446, + "step": 579 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999994826060166e-05, + "loss": 0.5934, + "step": 580 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999295769322e-05, + "loss": 0.6511, + "step": 581 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999990801885353e-05, + "loss": 0.5262, + "step": 582 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999988358636622e-05, + "loss": 0.5777, + "step": 583 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999985627947104e-05, + "loss": 0.4609, + "step": 584 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999982609816867e-05, + "loss": 0.6021, + "step": 585 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999979304246006e-05, + "loss": 0.6274, + "step": 586 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999975711234616e-05, + "loss": 0.517, + "step": 587 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999971830782795e-05, + "loss": 0.562, + "step": 588 + }, + { + "epoch": 0.25, + "learning_rate": 1.999996766289066e-05, + "loss": 0.5583, + "step": 589 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999963207558326e-05, + "loss": 0.5975, + "step": 590 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999958464785925e-05, + "loss": 0.6094, + "step": 591 + }, + { + "epoch": 0.25, + "learning_rate": 1.999995343457359e-05, + "loss": 0.5579, + "step": 592 + }, + { + "epoch": 0.25, + "learning_rate": 1.999994811692147e-05, + "loss": 0.551, + "step": 593 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999942511829714e-05, + "loss": 0.5576, + "step": 594 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999936619298483e-05, + "loss": 0.5486, + "step": 595 + }, + { + "epoch": 0.25, + "learning_rate": 1.999993043932795e-05, + "loss": 0.588, + "step": 596 + }, + { + "epoch": 0.25, + "learning_rate": 1.999992397191829e-05, + "loss": 0.5231, + "step": 597 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999917217069687e-05, + "loss": 0.5773, + "step": 598 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999910174782337e-05, + "loss": 0.5609, + "step": 599 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999902845056444e-05, + "loss": 0.6032, + "step": 600 + }, + { + "epoch": 0.25, + "learning_rate": 1.999989522789222e-05, + "loss": 0.5846, + "step": 601 + }, + { + "epoch": 0.25, + "learning_rate": 1.999988732328988e-05, + "loss": 0.6105, + "step": 602 + }, + { + "epoch": 0.25, + "learning_rate": 1.999987913124965e-05, + "loss": 0.5158, + "step": 603 + }, + { + "epoch": 0.25, + "learning_rate": 1.999987065177177e-05, + "loss": 0.6727, + "step": 604 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999861884856483e-05, + "loss": 0.5327, + "step": 605 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999852830504038e-05, + "loss": 0.5825, + "step": 606 + }, + { + "epoch": 0.25, + "learning_rate": 1.99998434887147e-05, + "loss": 0.5495, + "step": 607 + }, + { + "epoch": 0.25, + "learning_rate": 1.999983385948873e-05, + "loss": 0.542, + "step": 608 + }, + { + "epoch": 0.25, + "learning_rate": 1.999982394282641e-05, + "loss": 0.6037, + "step": 609 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999813738728028e-05, + "loss": 0.6024, + "step": 610 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999803247193873e-05, + "loss": 0.6103, + "step": 611 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999792468224247e-05, + "loss": 0.6136, + "step": 612 + }, + { + "epoch": 0.26, + "learning_rate": 1.999978140181946e-05, + "loss": 0.5756, + "step": 613 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999770047979828e-05, + "loss": 0.6016, + "step": 614 + }, + { + "epoch": 0.26, + "learning_rate": 1.999975840670568e-05, + "loss": 0.6508, + "step": 615 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999746477997352e-05, + "loss": 0.5523, + "step": 616 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999734261855188e-05, + "loss": 0.6448, + "step": 617 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999721758279532e-05, + "loss": 0.5188, + "step": 618 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999708967270747e-05, + "loss": 0.5533, + "step": 619 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999695888829206e-05, + "loss": 0.4754, + "step": 620 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999682522955277e-05, + "loss": 0.5404, + "step": 621 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999668869649347e-05, + "loss": 0.5205, + "step": 622 + }, + { + "epoch": 0.26, + "learning_rate": 1.999965492891181e-05, + "loss": 0.5183, + "step": 623 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999640700743063e-05, + "loss": 0.5141, + "step": 624 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999626185143518e-05, + "loss": 0.5411, + "step": 625 + }, + { + "epoch": 0.26, + "learning_rate": 1.99996113821136e-05, + "loss": 0.5665, + "step": 626 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999596291653715e-05, + "loss": 0.5936, + "step": 627 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999580913764313e-05, + "loss": 0.5419, + "step": 628 + }, + { + "epoch": 0.26, + "learning_rate": 1.999956524844583e-05, + "loss": 0.5021, + "step": 629 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999549295698722e-05, + "loss": 0.5776, + "step": 630 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999533055523437e-05, + "loss": 0.4954, + "step": 631 + }, + { + "epoch": 0.26, + "learning_rate": 1.999951652792045e-05, + "loss": 0.5628, + "step": 632 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999499712890235e-05, + "loss": 0.6356, + "step": 633 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999482610433275e-05, + "loss": 0.5385, + "step": 634 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999465220550057e-05, + "loss": 0.5615, + "step": 635 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999447543241086e-05, + "loss": 0.5947, + "step": 636 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999429578506867e-05, + "loss": 0.6466, + "step": 637 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999411326347924e-05, + "loss": 0.5652, + "step": 638 + }, + { + "epoch": 0.27, + "learning_rate": 1.999939278676477e-05, + "loss": 0.571, + "step": 639 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999373959757945e-05, + "loss": 0.6267, + "step": 640 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999354845327993e-05, + "loss": 0.5371, + "step": 641 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999335443475452e-05, + "loss": 0.5337, + "step": 642 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999315754200892e-05, + "loss": 0.542, + "step": 643 + }, + { + "epoch": 0.27, + "learning_rate": 1.999929577750487e-05, + "loss": 0.5757, + "step": 644 + }, + { + "epoch": 0.27, + "learning_rate": 1.999927551338797e-05, + "loss": 0.5567, + "step": 645 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999254961850765e-05, + "loss": 0.5668, + "step": 646 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999234122893848e-05, + "loss": 0.5143, + "step": 647 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999212996517823e-05, + "loss": 0.5506, + "step": 648 + }, + { + "epoch": 0.27, + "learning_rate": 1.999919158272329e-05, + "loss": 0.6022, + "step": 649 + }, + { + "epoch": 0.27, + "learning_rate": 1.999916988151087e-05, + "loss": 0.5667, + "step": 650 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999147892881182e-05, + "loss": 0.5127, + "step": 651 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999125616834866e-05, + "loss": 0.6013, + "step": 652 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999103053372554e-05, + "loss": 0.4694, + "step": 653 + }, + { + "epoch": 0.27, + "learning_rate": 1.99990802024949e-05, + "loss": 0.5073, + "step": 654 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999057064202555e-05, + "loss": 0.5867, + "step": 655 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999033638496187e-05, + "loss": 0.5578, + "step": 656 + }, + { + "epoch": 0.28, + "learning_rate": 1.9999009925376475e-05, + "loss": 0.588, + "step": 657 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998985924844093e-05, + "loss": 0.5796, + "step": 658 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998961636899736e-05, + "loss": 0.559, + "step": 659 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998937061544096e-05, + "loss": 0.4719, + "step": 660 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998912198777886e-05, + "loss": 0.5707, + "step": 661 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998887048601812e-05, + "loss": 0.5999, + "step": 662 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998861611016608e-05, + "loss": 0.49, + "step": 663 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998835886022997e-05, + "loss": 0.6035, + "step": 664 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998809873621722e-05, + "loss": 0.5609, + "step": 665 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998783573813535e-05, + "loss": 0.5078, + "step": 666 + }, + { + "epoch": 0.28, + "learning_rate": 1.999875698659918e-05, + "loss": 0.5402, + "step": 667 + }, + { + "epoch": 0.28, + "learning_rate": 1.999873011197943e-05, + "loss": 0.5853, + "step": 668 + }, + { + "epoch": 0.28, + "learning_rate": 1.999870294995506e-05, + "loss": 0.4822, + "step": 669 + }, + { + "epoch": 0.28, + "learning_rate": 1.999867550052684e-05, + "loss": 0.6014, + "step": 670 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998647763695568e-05, + "loss": 0.5914, + "step": 671 + }, + { + "epoch": 0.28, + "learning_rate": 1.999861973946204e-05, + "loss": 0.5698, + "step": 672 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998591427827062e-05, + "loss": 0.5685, + "step": 673 + }, + { + "epoch": 0.28, + "learning_rate": 1.999856282879144e-05, + "loss": 0.6114, + "step": 674 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998533942356008e-05, + "loss": 0.5575, + "step": 675 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998504768521588e-05, + "loss": 0.6381, + "step": 676 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998475307289018e-05, + "loss": 0.5428, + "step": 677 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998445558659152e-05, + "loss": 0.5533, + "step": 678 + }, + { + "epoch": 0.28, + "learning_rate": 1.999841552263284e-05, + "loss": 0.4876, + "step": 679 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998385199210945e-05, + "loss": 0.5705, + "step": 680 + }, + { + "epoch": 0.29, + "learning_rate": 1.999835458839434e-05, + "loss": 0.6025, + "step": 681 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998323690183907e-05, + "loss": 0.6213, + "step": 682 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998292504580528e-05, + "loss": 0.5695, + "step": 683 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998261031585104e-05, + "loss": 0.5282, + "step": 684 + }, + { + "epoch": 0.29, + "learning_rate": 1.999822927119854e-05, + "loss": 0.5748, + "step": 685 + }, + { + "epoch": 0.29, + "learning_rate": 1.999819722342175e-05, + "loss": 0.5202, + "step": 686 + }, + { + "epoch": 0.29, + "learning_rate": 1.999816488825565e-05, + "loss": 0.5667, + "step": 687 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998132265701176e-05, + "loss": 0.5365, + "step": 688 + }, + { + "epoch": 0.29, + "learning_rate": 1.999809935575926e-05, + "loss": 0.6048, + "step": 689 + }, + { + "epoch": 0.29, + "learning_rate": 1.999806615843085e-05, + "loss": 0.6033, + "step": 690 + }, + { + "epoch": 0.29, + "learning_rate": 1.99980326737169e-05, + "loss": 0.5729, + "step": 691 + }, + { + "epoch": 0.29, + "learning_rate": 1.999799890161837e-05, + "loss": 0.492, + "step": 692 + }, + { + "epoch": 0.29, + "learning_rate": 1.999796484213624e-05, + "loss": 0.5711, + "step": 693 + }, + { + "epoch": 0.29, + "learning_rate": 1.999793049527148e-05, + "loss": 0.5011, + "step": 694 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997895861025078e-05, + "loss": 0.5375, + "step": 695 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997860939398035e-05, + "loss": 0.5716, + "step": 696 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997825730391345e-05, + "loss": 0.5735, + "step": 697 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997790234006034e-05, + "loss": 0.529, + "step": 698 + }, + { + "epoch": 0.29, + "learning_rate": 1.999775445024311e-05, + "loss": 0.5457, + "step": 699 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997718379103608e-05, + "loss": 0.5196, + "step": 700 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997682020588563e-05, + "loss": 0.5419, + "step": 701 + }, + { + "epoch": 0.29, + "learning_rate": 1.999764537469902e-05, + "loss": 0.4548, + "step": 702 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997608441436033e-05, + "loss": 0.4813, + "step": 703 + }, + { + "epoch": 0.29, + "learning_rate": 1.999757122080066e-05, + "loss": 0.5566, + "step": 704 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997533712793976e-05, + "loss": 0.5593, + "step": 705 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997495917417054e-05, + "loss": 0.5683, + "step": 706 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997457834670988e-05, + "loss": 0.5919, + "step": 707 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997419464556868e-05, + "loss": 0.6129, + "step": 708 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997380807075796e-05, + "loss": 0.4535, + "step": 709 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997341862228886e-05, + "loss": 0.6517, + "step": 710 + }, + { + "epoch": 0.3, + "learning_rate": 1.999730263001725e-05, + "loss": 0.5925, + "step": 711 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997263110442025e-05, + "loss": 0.4815, + "step": 712 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997223303504345e-05, + "loss": 0.5777, + "step": 713 + }, + { + "epoch": 0.3, + "learning_rate": 1.999718320920535e-05, + "loss": 0.4826, + "step": 714 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997142827546196e-05, + "loss": 0.6005, + "step": 715 + }, + { + "epoch": 0.3, + "learning_rate": 1.999710215852804e-05, + "loss": 0.6185, + "step": 716 + }, + { + "epoch": 0.3, + "learning_rate": 1.999706120215206e-05, + "loss": 0.6308, + "step": 717 + }, + { + "epoch": 0.3, + "learning_rate": 1.999701995841942e-05, + "loss": 0.6157, + "step": 718 + }, + { + "epoch": 0.3, + "learning_rate": 1.999697842733132e-05, + "loss": 0.517, + "step": 719 + }, + { + "epoch": 0.3, + "learning_rate": 1.999693660888894e-05, + "loss": 0.5469, + "step": 720 + }, + { + "epoch": 0.3, + "learning_rate": 1.999689450309349e-05, + "loss": 0.5627, + "step": 721 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996852109946178e-05, + "loss": 0.5886, + "step": 722 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996809429448224e-05, + "loss": 0.5238, + "step": 723 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996766461600855e-05, + "loss": 0.6008, + "step": 724 + }, + { + "epoch": 0.3, + "learning_rate": 1.99967232064053e-05, + "loss": 0.4986, + "step": 725 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996679663862815e-05, + "loss": 0.5396, + "step": 726 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996635833974637e-05, + "loss": 0.497, + "step": 727 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996591716742037e-05, + "loss": 0.5982, + "step": 728 + }, + { + "epoch": 0.31, + "learning_rate": 1.999654731216628e-05, + "loss": 0.5701, + "step": 729 + }, + { + "epoch": 0.31, + "learning_rate": 1.999650262024864e-05, + "loss": 0.5235, + "step": 730 + }, + { + "epoch": 0.31, + "learning_rate": 1.99964576409904e-05, + "loss": 0.5917, + "step": 731 + }, + { + "epoch": 0.31, + "learning_rate": 1.999641237439286e-05, + "loss": 0.5691, + "step": 732 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996366820457313e-05, + "loss": 0.6141, + "step": 733 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996320979185074e-05, + "loss": 0.5376, + "step": 734 + }, + { + "epoch": 0.31, + "learning_rate": 1.999627485057746e-05, + "loss": 0.4333, + "step": 735 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996228434635797e-05, + "loss": 0.5657, + "step": 736 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996181731361415e-05, + "loss": 0.5721, + "step": 737 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996134740755662e-05, + "loss": 0.6029, + "step": 738 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996087462819884e-05, + "loss": 0.5711, + "step": 739 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996039897555443e-05, + "loss": 0.5509, + "step": 740 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995992044963705e-05, + "loss": 0.555, + "step": 741 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995943905046048e-05, + "loss": 0.4628, + "step": 742 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995895477803854e-05, + "loss": 0.5218, + "step": 743 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995846763238514e-05, + "loss": 0.4748, + "step": 744 + }, + { + "epoch": 0.31, + "learning_rate": 1.999579776135143e-05, + "loss": 0.6094, + "step": 745 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995748472144005e-05, + "loss": 0.5397, + "step": 746 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995698895617663e-05, + "loss": 0.6416, + "step": 747 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995649031773828e-05, + "loss": 0.5666, + "step": 748 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995598880613928e-05, + "loss": 0.5718, + "step": 749 + }, + { + "epoch": 0.31, + "learning_rate": 1.999554844213941e-05, + "loss": 0.5107, + "step": 750 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995497716351725e-05, + "loss": 0.5182, + "step": 751 + }, + { + "epoch": 0.31, + "learning_rate": 1.999544670325232e-05, + "loss": 0.645, + "step": 752 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995395402842678e-05, + "loss": 0.6027, + "step": 753 + }, + { + "epoch": 0.32, + "learning_rate": 1.999534381512426e-05, + "loss": 0.506, + "step": 754 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995291940098555e-05, + "loss": 0.4984, + "step": 755 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995239777767053e-05, + "loss": 0.4942, + "step": 756 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995187328131252e-05, + "loss": 0.5343, + "step": 757 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995134591192664e-05, + "loss": 0.4752, + "step": 758 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995081566952795e-05, + "loss": 0.4636, + "step": 759 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995028255413184e-05, + "loss": 0.5338, + "step": 760 + }, + { + "epoch": 0.32, + "learning_rate": 1.999497465657535e-05, + "loss": 0.5684, + "step": 761 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994920770440842e-05, + "loss": 0.4523, + "step": 762 + }, + { + "epoch": 0.32, + "learning_rate": 1.99948665970112e-05, + "loss": 0.6039, + "step": 763 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994812136287994e-05, + "loss": 0.4691, + "step": 764 + }, + { + "epoch": 0.32, + "learning_rate": 1.999475738827278e-05, + "loss": 0.5422, + "step": 765 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994702352967132e-05, + "loss": 0.5557, + "step": 766 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994647030372638e-05, + "loss": 0.545, + "step": 767 + }, + { + "epoch": 0.32, + "learning_rate": 1.999459142049088e-05, + "loss": 0.5399, + "step": 768 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994535523323462e-05, + "loss": 0.5359, + "step": 769 + }, + { + "epoch": 0.32, + "learning_rate": 1.999447933887199e-05, + "loss": 0.5282, + "step": 770 + }, + { + "epoch": 0.32, + "learning_rate": 1.999442286713808e-05, + "loss": 0.5645, + "step": 771 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994366108123352e-05, + "loss": 0.5232, + "step": 772 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994309061829438e-05, + "loss": 0.537, + "step": 773 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994251728257978e-05, + "loss": 0.6248, + "step": 774 + }, + { + "epoch": 0.32, + "learning_rate": 1.999419410741062e-05, + "loss": 0.5334, + "step": 775 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994136199289028e-05, + "loss": 0.5189, + "step": 776 + }, + { + "epoch": 0.33, + "learning_rate": 1.9994078003894856e-05, + "loss": 0.64, + "step": 777 + }, + { + "epoch": 0.33, + "learning_rate": 1.999401952122978e-05, + "loss": 0.6003, + "step": 778 + }, + { + "epoch": 0.33, + "learning_rate": 1.999396075129548e-05, + "loss": 0.5584, + "step": 779 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993901694093645e-05, + "loss": 0.4774, + "step": 780 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993842349625978e-05, + "loss": 0.5653, + "step": 781 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993782717894178e-05, + "loss": 0.5828, + "step": 782 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993722798899963e-05, + "loss": 0.6803, + "step": 783 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993662592645057e-05, + "loss": 0.487, + "step": 784 + }, + { + "epoch": 0.33, + "learning_rate": 1.999360209913119e-05, + "loss": 0.5226, + "step": 785 + }, + { + "epoch": 0.33, + "learning_rate": 1.999354131836009e-05, + "loss": 0.5717, + "step": 786 + }, + { + "epoch": 0.33, + "learning_rate": 1.999348025033352e-05, + "loss": 0.5765, + "step": 787 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993418895053227e-05, + "loss": 0.4651, + "step": 788 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993357252520975e-05, + "loss": 0.5336, + "step": 789 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993295322738537e-05, + "loss": 0.5058, + "step": 790 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993233105707693e-05, + "loss": 0.5516, + "step": 791 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993170601430233e-05, + "loss": 0.57, + "step": 792 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993107809907946e-05, + "loss": 0.481, + "step": 793 + }, + { + "epoch": 0.33, + "learning_rate": 1.999304473114265e-05, + "loss": 0.6331, + "step": 794 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992981365136145e-05, + "loss": 0.5095, + "step": 795 + }, + { + "epoch": 0.33, + "learning_rate": 1.999291771189026e-05, + "loss": 0.5001, + "step": 796 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992853771406825e-05, + "loss": 0.5308, + "step": 797 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992789543687673e-05, + "loss": 0.5813, + "step": 798 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992725028734655e-05, + "loss": 0.6555, + "step": 799 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992660226549624e-05, + "loss": 0.4987, + "step": 800 + }, + { + "epoch": 0.34, + "learning_rate": 1.999259513713444e-05, + "loss": 0.5393, + "step": 801 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992529760490976e-05, + "loss": 0.554, + "step": 802 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992464096621113e-05, + "loss": 0.484, + "step": 803 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992398145526738e-05, + "loss": 0.5911, + "step": 804 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992331907209742e-05, + "loss": 0.5786, + "step": 805 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992265381672036e-05, + "loss": 0.5447, + "step": 806 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992198568915526e-05, + "loss": 0.5934, + "step": 807 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992131468942135e-05, + "loss": 0.5529, + "step": 808 + }, + { + "epoch": 0.34, + "learning_rate": 1.999206408175379e-05, + "loss": 0.539, + "step": 809 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991996407352432e-05, + "loss": 0.5555, + "step": 810 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991928445740007e-05, + "loss": 0.5249, + "step": 811 + }, + { + "epoch": 0.34, + "learning_rate": 1.999186019691846e-05, + "loss": 0.5433, + "step": 812 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991791660889762e-05, + "loss": 0.6182, + "step": 813 + }, + { + "epoch": 0.34, + "learning_rate": 1.999172283765588e-05, + "loss": 0.5141, + "step": 814 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991653727218787e-05, + "loss": 0.5161, + "step": 815 + }, + { + "epoch": 0.34, + "learning_rate": 1.999158432958048e-05, + "loss": 0.5049, + "step": 816 + }, + { + "epoch": 0.34, + "learning_rate": 1.999151464474294e-05, + "loss": 0.4883, + "step": 817 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991444672708183e-05, + "loss": 0.5744, + "step": 818 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991374413478214e-05, + "loss": 0.5944, + "step": 819 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991303867055056e-05, + "loss": 0.5067, + "step": 820 + }, + { + "epoch": 0.34, + "learning_rate": 1.999123303344073e-05, + "loss": 0.5539, + "step": 821 + }, + { + "epoch": 0.34, + "learning_rate": 1.999116191263728e-05, + "loss": 0.5636, + "step": 822 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991090504646748e-05, + "loss": 0.63, + "step": 823 + }, + { + "epoch": 0.34, + "learning_rate": 1.999101880947118e-05, + "loss": 0.5853, + "step": 824 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990946827112648e-05, + "loss": 0.5308, + "step": 825 + }, + { + "epoch": 0.35, + "learning_rate": 1.999087455757321e-05, + "loss": 0.5656, + "step": 826 + }, + { + "epoch": 0.35, + "learning_rate": 1.999080200085495e-05, + "loss": 0.5307, + "step": 827 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990729156959956e-05, + "loss": 0.4947, + "step": 828 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990656025890315e-05, + "loss": 0.5177, + "step": 829 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990582607648133e-05, + "loss": 0.5555, + "step": 830 + }, + { + "epoch": 0.35, + "learning_rate": 1.999050890223552e-05, + "loss": 0.4916, + "step": 831 + }, + { + "epoch": 0.35, + "learning_rate": 1.999043490965459e-05, + "loss": 0.5135, + "step": 832 + }, + { + "epoch": 0.35, + "learning_rate": 1.999036062990748e-05, + "loss": 0.63, + "step": 833 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990286062996316e-05, + "loss": 0.5116, + "step": 834 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990211208923242e-05, + "loss": 0.5488, + "step": 835 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990136067690416e-05, + "loss": 0.6538, + "step": 836 + }, + { + "epoch": 0.35, + "learning_rate": 1.999006063929999e-05, + "loss": 0.5465, + "step": 837 + }, + { + "epoch": 0.35, + "learning_rate": 1.998998492375414e-05, + "loss": 0.5929, + "step": 838 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989908921055036e-05, + "loss": 0.5986, + "step": 839 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989832631204863e-05, + "loss": 0.613, + "step": 840 + }, + { + "epoch": 0.35, + "learning_rate": 1.998975605420582e-05, + "loss": 0.5219, + "step": 841 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989679190060104e-05, + "loss": 0.5513, + "step": 842 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989602038769922e-05, + "loss": 0.4865, + "step": 843 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989524600337493e-05, + "loss": 0.6124, + "step": 844 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989446874765048e-05, + "loss": 0.5157, + "step": 845 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989368862054814e-05, + "loss": 0.5596, + "step": 846 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989290562209036e-05, + "loss": 0.5604, + "step": 847 + }, + { + "epoch": 0.36, + "learning_rate": 1.9989211975229966e-05, + "loss": 0.5181, + "step": 848 + }, + { + "epoch": 0.36, + "learning_rate": 1.998913310111986e-05, + "loss": 0.5237, + "step": 849 + }, + { + "epoch": 0.36, + "learning_rate": 1.998905393988099e-05, + "loss": 0.5698, + "step": 850 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988974491515628e-05, + "loss": 0.5563, + "step": 851 + }, + { + "epoch": 0.36, + "learning_rate": 1.998889475602606e-05, + "loss": 0.5638, + "step": 852 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988814733414573e-05, + "loss": 0.5097, + "step": 853 + }, + { + "epoch": 0.36, + "learning_rate": 1.998873442368347e-05, + "loss": 0.4871, + "step": 854 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988653826835062e-05, + "loss": 0.4984, + "step": 855 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988572942871666e-05, + "loss": 0.5244, + "step": 856 + }, + { + "epoch": 0.36, + "learning_rate": 1.99884917717956e-05, + "loss": 0.6335, + "step": 857 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988410313609203e-05, + "loss": 0.5901, + "step": 858 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988328568314818e-05, + "loss": 0.54, + "step": 859 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988246535914787e-05, + "loss": 0.5668, + "step": 860 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988164216411474e-05, + "loss": 0.5125, + "step": 861 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988081609807243e-05, + "loss": 0.5225, + "step": 862 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987998716104473e-05, + "loss": 0.5736, + "step": 863 + }, + { + "epoch": 0.36, + "learning_rate": 1.998791553530554e-05, + "loss": 0.4624, + "step": 864 + }, + { + "epoch": 0.36, + "learning_rate": 1.998783206741284e-05, + "loss": 0.5752, + "step": 865 + }, + { + "epoch": 0.36, + "learning_rate": 1.998774831242877e-05, + "loss": 0.5861, + "step": 866 + }, + { + "epoch": 0.36, + "learning_rate": 1.998766427035574e-05, + "loss": 0.5887, + "step": 867 + }, + { + "epoch": 0.36, + "learning_rate": 1.998757994119616e-05, + "loss": 0.5148, + "step": 868 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987495324952458e-05, + "loss": 0.5131, + "step": 869 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987410421627063e-05, + "loss": 0.5495, + "step": 870 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987325231222423e-05, + "loss": 0.5244, + "step": 871 + }, + { + "epoch": 0.37, + "learning_rate": 1.998723975374098e-05, + "loss": 0.5098, + "step": 872 + }, + { + "epoch": 0.37, + "learning_rate": 1.998715398918519e-05, + "loss": 0.5323, + "step": 873 + }, + { + "epoch": 0.37, + "learning_rate": 1.9987067937557527e-05, + "loss": 0.6008, + "step": 874 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986981598860454e-05, + "loss": 0.613, + "step": 875 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986894973096458e-05, + "loss": 0.5766, + "step": 876 + }, + { + "epoch": 0.37, + "learning_rate": 1.998680806026803e-05, + "loss": 0.5188, + "step": 877 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986720860377664e-05, + "loss": 0.5137, + "step": 878 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986633373427868e-05, + "loss": 0.6744, + "step": 879 + }, + { + "epoch": 0.37, + "learning_rate": 1.998654559942116e-05, + "loss": 0.5557, + "step": 880 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986457538360062e-05, + "loss": 0.5192, + "step": 881 + }, + { + "epoch": 0.37, + "learning_rate": 1.99863691902471e-05, + "loss": 0.5091, + "step": 882 + }, + { + "epoch": 0.37, + "learning_rate": 1.998628055508482e-05, + "loss": 0.568, + "step": 883 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986191632875768e-05, + "loss": 0.5641, + "step": 884 + }, + { + "epoch": 0.37, + "learning_rate": 1.99861024236225e-05, + "loss": 0.5126, + "step": 885 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986012927327575e-05, + "loss": 0.5267, + "step": 886 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985923143993572e-05, + "loss": 0.4799, + "step": 887 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985833073623068e-05, + "loss": 0.5367, + "step": 888 + }, + { + "epoch": 0.37, + "learning_rate": 1.998574271621866e-05, + "loss": 0.5242, + "step": 889 + }, + { + "epoch": 0.37, + "learning_rate": 1.998565207178293e-05, + "loss": 0.5858, + "step": 890 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985561140318496e-05, + "loss": 0.5344, + "step": 891 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985469921827968e-05, + "loss": 0.5244, + "step": 892 + }, + { + "epoch": 0.37, + "learning_rate": 1.998537841631397e-05, + "loss": 0.5289, + "step": 893 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985286623779126e-05, + "loss": 0.5555, + "step": 894 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985194544226082e-05, + "loss": 0.5863, + "step": 895 + }, + { + "epoch": 0.38, + "learning_rate": 1.998510217765748e-05, + "loss": 0.4871, + "step": 896 + }, + { + "epoch": 0.38, + "learning_rate": 1.998500952407598e-05, + "loss": 0.5624, + "step": 897 + }, + { + "epoch": 0.38, + "learning_rate": 1.998491658348424e-05, + "loss": 0.5368, + "step": 898 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984823355884926e-05, + "loss": 0.4979, + "step": 899 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984729841280734e-05, + "loss": 0.6757, + "step": 900 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984636039674342e-05, + "loss": 0.5738, + "step": 901 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984541951068445e-05, + "loss": 0.5586, + "step": 902 + }, + { + "epoch": 0.38, + "learning_rate": 1.998444757546575e-05, + "loss": 0.5576, + "step": 903 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984352912868973e-05, + "loss": 0.5864, + "step": 904 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984257963280823e-05, + "loss": 0.5936, + "step": 905 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984162726704047e-05, + "loss": 0.6112, + "step": 906 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984067203141368e-05, + "loss": 0.5352, + "step": 907 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983971392595537e-05, + "loss": 0.6031, + "step": 908 + }, + { + "epoch": 0.38, + "learning_rate": 1.998387529506931e-05, + "loss": 0.5561, + "step": 909 + }, + { + "epoch": 0.38, + "learning_rate": 1.998377891056545e-05, + "loss": 0.5093, + "step": 910 + }, + { + "epoch": 0.38, + "learning_rate": 1.998368223908672e-05, + "loss": 0.6654, + "step": 911 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983585280635902e-05, + "loss": 0.4926, + "step": 912 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983488035215788e-05, + "loss": 0.5659, + "step": 913 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983390502829168e-05, + "loss": 0.5259, + "step": 914 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983292683478844e-05, + "loss": 0.4566, + "step": 915 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983194577167636e-05, + "loss": 0.524, + "step": 916 + }, + { + "epoch": 0.38, + "learning_rate": 1.998309618389836e-05, + "loss": 0.4984, + "step": 917 + }, + { + "epoch": 0.38, + "learning_rate": 1.9982997503673837e-05, + "loss": 0.5231, + "step": 918 + }, + { + "epoch": 0.38, + "learning_rate": 1.9982898536496912e-05, + "loss": 0.5507, + "step": 919 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982799282370432e-05, + "loss": 0.5801, + "step": 920 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982699741297236e-05, + "loss": 0.5397, + "step": 921 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982599913280202e-05, + "loss": 0.5434, + "step": 922 + }, + { + "epoch": 0.39, + "learning_rate": 1.998249979832219e-05, + "loss": 0.507, + "step": 923 + }, + { + "epoch": 0.39, + "learning_rate": 1.998239939642608e-05, + "loss": 0.5271, + "step": 924 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982298707594757e-05, + "loss": 0.5165, + "step": 925 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982197731831115e-05, + "loss": 0.6061, + "step": 926 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982096469138056e-05, + "loss": 0.6236, + "step": 927 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981994919518496e-05, + "loss": 0.5399, + "step": 928 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981893082975348e-05, + "loss": 0.5696, + "step": 929 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981790959511538e-05, + "loss": 0.5813, + "step": 930 + }, + { + "epoch": 0.39, + "learning_rate": 1.998168854913001e-05, + "loss": 0.5144, + "step": 931 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981585851833697e-05, + "loss": 0.4815, + "step": 932 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981482867625557e-05, + "loss": 0.5817, + "step": 933 + }, + { + "epoch": 0.39, + "learning_rate": 1.998137959650855e-05, + "loss": 0.5457, + "step": 934 + }, + { + "epoch": 0.39, + "learning_rate": 1.998127603848565e-05, + "loss": 0.4847, + "step": 935 + }, + { + "epoch": 0.39, + "learning_rate": 1.998117219355982e-05, + "loss": 0.4814, + "step": 936 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981068061734055e-05, + "loss": 0.5047, + "step": 937 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980963643011346e-05, + "loss": 0.5589, + "step": 938 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980858937394695e-05, + "loss": 0.5293, + "step": 939 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980753944887108e-05, + "loss": 0.5762, + "step": 940 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980648665491606e-05, + "loss": 0.557, + "step": 941 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980543099211215e-05, + "loss": 0.577, + "step": 942 + }, + { + "epoch": 0.39, + "learning_rate": 1.998043724604897e-05, + "loss": 0.5131, + "step": 943 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980331106007912e-05, + "loss": 0.5274, + "step": 944 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980224679091092e-05, + "loss": 0.5165, + "step": 945 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980117965301574e-05, + "loss": 0.5571, + "step": 946 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980010964642417e-05, + "loss": 0.5509, + "step": 947 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979903677116705e-05, + "loss": 0.4886, + "step": 948 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979796102727513e-05, + "loss": 0.5009, + "step": 949 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979688241477938e-05, + "loss": 0.5245, + "step": 950 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979580093371084e-05, + "loss": 0.5203, + "step": 951 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979471658410055e-05, + "loss": 0.5339, + "step": 952 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979362936597967e-05, + "loss": 0.6407, + "step": 953 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979253927937948e-05, + "loss": 0.5562, + "step": 954 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979144632433132e-05, + "loss": 0.5537, + "step": 955 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979035050086656e-05, + "loss": 0.537, + "step": 956 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978925180901675e-05, + "loss": 0.5218, + "step": 957 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978815024881342e-05, + "loss": 0.619, + "step": 958 + }, + { + "epoch": 0.4, + "learning_rate": 1.997870458202883e-05, + "loss": 0.5215, + "step": 959 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978593852347306e-05, + "loss": 0.5597, + "step": 960 + }, + { + "epoch": 0.4, + "learning_rate": 1.997848283583996e-05, + "loss": 0.4835, + "step": 961 + }, + { + "epoch": 0.4, + "learning_rate": 1.997837153250998e-05, + "loss": 0.5569, + "step": 962 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978259942360562e-05, + "loss": 0.5173, + "step": 963 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978148065394917e-05, + "loss": 0.5005, + "step": 964 + }, + { + "epoch": 0.4, + "learning_rate": 1.997803590161626e-05, + "loss": 0.6021, + "step": 965 + }, + { + "epoch": 0.4, + "learning_rate": 1.9977923451027816e-05, + "loss": 0.5151, + "step": 966 + }, + { + "epoch": 0.4, + "learning_rate": 1.9977810713632815e-05, + "loss": 0.5957, + "step": 967 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977697689434504e-05, + "loss": 0.5009, + "step": 968 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977584378436123e-05, + "loss": 0.5259, + "step": 969 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977470780640934e-05, + "loss": 0.4887, + "step": 970 + }, + { + "epoch": 0.41, + "learning_rate": 1.99773568960522e-05, + "loss": 0.5379, + "step": 971 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977242724673198e-05, + "loss": 0.6189, + "step": 972 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977128266507203e-05, + "loss": 0.5615, + "step": 973 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977013521557516e-05, + "loss": 0.5451, + "step": 974 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976898489827423e-05, + "loss": 0.5693, + "step": 975 + }, + { + "epoch": 0.41, + "learning_rate": 1.997678317132024e-05, + "loss": 0.5098, + "step": 976 + }, + { + "epoch": 0.41, + "learning_rate": 1.997666756603928e-05, + "loss": 0.5261, + "step": 977 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976551673987857e-05, + "loss": 0.525, + "step": 978 + }, + { + "epoch": 0.41, + "learning_rate": 1.997643549516931e-05, + "loss": 0.6046, + "step": 979 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976319029586982e-05, + "loss": 0.572, + "step": 980 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976202277244216e-05, + "loss": 0.5908, + "step": 981 + }, + { + "epoch": 0.41, + "learning_rate": 1.997608523814437e-05, + "loss": 0.5843, + "step": 982 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975967912290805e-05, + "loss": 0.5956, + "step": 983 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975850299686894e-05, + "loss": 0.5536, + "step": 984 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975732400336018e-05, + "loss": 0.5441, + "step": 985 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975614214241562e-05, + "loss": 0.5195, + "step": 986 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975495741406934e-05, + "loss": 0.5481, + "step": 987 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975376981835533e-05, + "loss": 0.5622, + "step": 988 + }, + { + "epoch": 0.41, + "learning_rate": 1.997525793553077e-05, + "loss": 0.4714, + "step": 989 + }, + { + "epoch": 0.41, + "learning_rate": 1.997513860249607e-05, + "loss": 0.5424, + "step": 990 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975018982734863e-05, + "loss": 0.5466, + "step": 991 + }, + { + "epoch": 0.42, + "learning_rate": 1.997489907625059e-05, + "loss": 0.5853, + "step": 992 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974778883046688e-05, + "loss": 0.4975, + "step": 993 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974658403126624e-05, + "loss": 0.4859, + "step": 994 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974537636493853e-05, + "loss": 0.4968, + "step": 995 + }, + { + "epoch": 0.42, + "learning_rate": 1.997441658315185e-05, + "loss": 0.5131, + "step": 996 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974295243104093e-05, + "loss": 0.5391, + "step": 997 + }, + { + "epoch": 0.42, + "learning_rate": 1.997417361635407e-05, + "loss": 0.4989, + "step": 998 + }, + { + "epoch": 0.42, + "learning_rate": 1.997405170290528e-05, + "loss": 0.5016, + "step": 999 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973929502761222e-05, + "loss": 0.5196, + "step": 1000 + }, + { + "epoch": 0.42, + "learning_rate": 1.997380701592541e-05, + "loss": 0.5563, + "step": 1001 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973684242401366e-05, + "loss": 0.5229, + "step": 1002 + }, + { + "epoch": 0.42, + "learning_rate": 1.997356118219262e-05, + "loss": 0.4715, + "step": 1003 + }, + { + "epoch": 0.42, + "learning_rate": 1.997343783530271e-05, + "loss": 0.5847, + "step": 1004 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973314201735176e-05, + "loss": 0.6229, + "step": 1005 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973190281493578e-05, + "loss": 0.4641, + "step": 1006 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973066074581475e-05, + "loss": 0.4784, + "step": 1007 + }, + { + "epoch": 0.42, + "learning_rate": 1.997294158100244e-05, + "loss": 0.4818, + "step": 1008 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972816800760046e-05, + "loss": 0.4652, + "step": 1009 + }, + { + "epoch": 0.42, + "learning_rate": 1.997269173385788e-05, + "loss": 0.5223, + "step": 1010 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972566380299544e-05, + "loss": 0.5002, + "step": 1011 + }, + { + "epoch": 0.42, + "learning_rate": 1.997244074008864e-05, + "loss": 0.487, + "step": 1012 + }, + { + "epoch": 0.42, + "learning_rate": 1.997231481322877e-05, + "loss": 0.5619, + "step": 1013 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972188599723564e-05, + "loss": 0.5361, + "step": 1014 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972062099576642e-05, + "loss": 0.524, + "step": 1015 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971935312791646e-05, + "loss": 0.4828, + "step": 1016 + }, + { + "epoch": 0.43, + "learning_rate": 1.997180823937222e-05, + "loss": 0.5571, + "step": 1017 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971680879322012e-05, + "loss": 0.528, + "step": 1018 + }, + { + "epoch": 0.43, + "learning_rate": 1.997155323264469e-05, + "loss": 0.5309, + "step": 1019 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971425299343915e-05, + "loss": 0.4413, + "step": 1020 + }, + { + "epoch": 0.43, + "learning_rate": 1.997129707942337e-05, + "loss": 0.4623, + "step": 1021 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971168572886738e-05, + "loss": 0.4915, + "step": 1022 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971039779737717e-05, + "loss": 0.5633, + "step": 1023 + }, + { + "epoch": 0.43, + "learning_rate": 1.997091069998e-05, + "loss": 0.577, + "step": 1024 + }, + { + "epoch": 0.43, + "learning_rate": 1.9970781333617303e-05, + "loss": 0.5143, + "step": 1025 + }, + { + "epoch": 0.43, + "learning_rate": 1.997065168065335e-05, + "loss": 0.5277, + "step": 1026 + }, + { + "epoch": 0.43, + "learning_rate": 1.997052174109186e-05, + "loss": 0.5045, + "step": 1027 + }, + { + "epoch": 0.43, + "learning_rate": 1.997039151493657e-05, + "loss": 0.5114, + "step": 1028 + }, + { + "epoch": 0.43, + "learning_rate": 1.997026100219122e-05, + "loss": 0.4649, + "step": 1029 + }, + { + "epoch": 0.43, + "learning_rate": 1.997013020285957e-05, + "loss": 0.4951, + "step": 1030 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969999116945374e-05, + "loss": 0.5396, + "step": 1031 + }, + { + "epoch": 0.43, + "learning_rate": 1.99698677444524e-05, + "loss": 0.5606, + "step": 1032 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969736085384422e-05, + "loss": 0.5713, + "step": 1033 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969604139745227e-05, + "loss": 0.5012, + "step": 1034 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969471907538615e-05, + "loss": 0.6019, + "step": 1035 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969339388768372e-05, + "loss": 0.5543, + "step": 1036 + }, + { + "epoch": 0.43, + "learning_rate": 1.996920658343832e-05, + "loss": 0.5305, + "step": 1037 + }, + { + "epoch": 0.43, + "learning_rate": 1.996907349155227e-05, + "loss": 0.5374, + "step": 1038 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968940113114048e-05, + "loss": 0.5252, + "step": 1039 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968806448127488e-05, + "loss": 0.5686, + "step": 1040 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968672496596434e-05, + "loss": 0.5571, + "step": 1041 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968538258524736e-05, + "loss": 0.5308, + "step": 1042 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968403733916246e-05, + "loss": 0.5229, + "step": 1043 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968268922774842e-05, + "loss": 0.5767, + "step": 1044 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968133825104396e-05, + "loss": 0.3965, + "step": 1045 + }, + { + "epoch": 0.44, + "learning_rate": 1.996799844090878e-05, + "loss": 0.6016, + "step": 1046 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967862770191902e-05, + "loss": 0.5493, + "step": 1047 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967726812957652e-05, + "loss": 0.5211, + "step": 1048 + }, + { + "epoch": 0.44, + "learning_rate": 1.996759056920994e-05, + "loss": 0.6184, + "step": 1049 + }, + { + "epoch": 0.44, + "learning_rate": 1.996745403895268e-05, + "loss": 0.5019, + "step": 1050 + }, + { + "epoch": 0.44, + "learning_rate": 1.99673172221898e-05, + "loss": 0.549, + "step": 1051 + }, + { + "epoch": 0.44, + "learning_rate": 1.996718011892523e-05, + "loss": 0.484, + "step": 1052 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967042729162914e-05, + "loss": 0.5575, + "step": 1053 + }, + { + "epoch": 0.44, + "learning_rate": 1.99669050529068e-05, + "loss": 0.5138, + "step": 1054 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966767090160843e-05, + "loss": 0.4945, + "step": 1055 + }, + { + "epoch": 0.44, + "learning_rate": 1.996662884092901e-05, + "loss": 0.4617, + "step": 1056 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966490305215277e-05, + "loss": 0.5356, + "step": 1057 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966351483023627e-05, + "loss": 0.4732, + "step": 1058 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966212374358044e-05, + "loss": 0.4211, + "step": 1059 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966072979222527e-05, + "loss": 0.5779, + "step": 1060 + }, + { + "epoch": 0.44, + "learning_rate": 1.996593329762109e-05, + "loss": 0.5561, + "step": 1061 + }, + { + "epoch": 0.44, + "learning_rate": 1.9965793329557743e-05, + "loss": 0.5145, + "step": 1062 + }, + { + "epoch": 0.45, + "learning_rate": 1.996565307503651e-05, + "loss": 0.5379, + "step": 1063 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965512534061426e-05, + "loss": 0.446, + "step": 1064 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965371706636522e-05, + "loss": 0.5324, + "step": 1065 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965230592765856e-05, + "loss": 0.4506, + "step": 1066 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965089192453475e-05, + "loss": 0.5063, + "step": 1067 + }, + { + "epoch": 0.45, + "learning_rate": 1.996494750570345e-05, + "loss": 0.5347, + "step": 1068 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964805532519852e-05, + "loss": 0.5617, + "step": 1069 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964663272906763e-05, + "loss": 0.4943, + "step": 1070 + }, + { + "epoch": 0.45, + "learning_rate": 1.996452072686827e-05, + "loss": 0.5119, + "step": 1071 + }, + { + "epoch": 0.45, + "learning_rate": 1.996437789440847e-05, + "loss": 0.5418, + "step": 1072 + }, + { + "epoch": 0.45, + "learning_rate": 1.996423477553147e-05, + "loss": 0.5345, + "step": 1073 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964091370241384e-05, + "loss": 0.4945, + "step": 1074 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963947678542335e-05, + "loss": 0.5395, + "step": 1075 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963803700438448e-05, + "loss": 0.5527, + "step": 1076 + }, + { + "epoch": 0.45, + "learning_rate": 1.996365943593387e-05, + "loss": 0.4826, + "step": 1077 + }, + { + "epoch": 0.45, + "learning_rate": 1.996351488503274e-05, + "loss": 0.5785, + "step": 1078 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963370047739218e-05, + "loss": 0.5382, + "step": 1079 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963224924057463e-05, + "loss": 0.5902, + "step": 1080 + }, + { + "epoch": 0.45, + "learning_rate": 1.996307951399165e-05, + "loss": 0.5723, + "step": 1081 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962933817545957e-05, + "loss": 0.5868, + "step": 1082 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962787834724572e-05, + "loss": 0.6129, + "step": 1083 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962641565531694e-05, + "loss": 0.501, + "step": 1084 + }, + { + "epoch": 0.45, + "learning_rate": 1.996249500997152e-05, + "loss": 0.5659, + "step": 1085 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962348168048272e-05, + "loss": 0.556, + "step": 1086 + }, + { + "epoch": 0.46, + "learning_rate": 1.9962201039766163e-05, + "loss": 0.556, + "step": 1087 + }, + { + "epoch": 0.46, + "learning_rate": 1.996205362512943e-05, + "loss": 0.5758, + "step": 1088 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961905924142303e-05, + "loss": 0.4728, + "step": 1089 + }, + { + "epoch": 0.46, + "learning_rate": 1.996175793680903e-05, + "loss": 0.5484, + "step": 1090 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961609663133865e-05, + "loss": 0.5067, + "step": 1091 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961461103121073e-05, + "loss": 0.5013, + "step": 1092 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961312256774915e-05, + "loss": 0.5285, + "step": 1093 + }, + { + "epoch": 0.46, + "learning_rate": 1.996116312409968e-05, + "loss": 0.5439, + "step": 1094 + }, + { + "epoch": 0.46, + "learning_rate": 1.996101370509965e-05, + "loss": 0.5521, + "step": 1095 + }, + { + "epoch": 0.46, + "learning_rate": 1.996086399977912e-05, + "loss": 0.5894, + "step": 1096 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960714008142394e-05, + "loss": 0.4348, + "step": 1097 + }, + { + "epoch": 0.46, + "learning_rate": 1.996056373019378e-05, + "loss": 0.5884, + "step": 1098 + }, + { + "epoch": 0.46, + "learning_rate": 1.99604131659376e-05, + "loss": 0.4819, + "step": 1099 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960262315378188e-05, + "loss": 0.4674, + "step": 1100 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960111178519873e-05, + "loss": 0.5385, + "step": 1101 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959959755366997e-05, + "loss": 0.5105, + "step": 1102 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959808045923917e-05, + "loss": 0.4483, + "step": 1103 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959656050194994e-05, + "loss": 0.5014, + "step": 1104 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959503768184597e-05, + "loss": 0.5339, + "step": 1105 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959351199897102e-05, + "loss": 0.4732, + "step": 1106 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959198345336893e-05, + "loss": 0.5171, + "step": 1107 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959045204508366e-05, + "loss": 0.5483, + "step": 1108 + }, + { + "epoch": 0.46, + "learning_rate": 1.9958891777415917e-05, + "loss": 0.5923, + "step": 1109 + }, + { + "epoch": 0.46, + "learning_rate": 1.9958738064063968e-05, + "loss": 0.5747, + "step": 1110 + }, + { + "epoch": 0.47, + "learning_rate": 1.995858406445693e-05, + "loss": 0.5271, + "step": 1111 + }, + { + "epoch": 0.47, + "learning_rate": 1.9958429778599225e-05, + "loss": 0.5466, + "step": 1112 + }, + { + "epoch": 0.47, + "learning_rate": 1.99582752064953e-05, + "loss": 0.4861, + "step": 1113 + }, + { + "epoch": 0.47, + "learning_rate": 1.9958120348149585e-05, + "loss": 0.5899, + "step": 1114 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957965203566542e-05, + "loss": 0.4815, + "step": 1115 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957809772750623e-05, + "loss": 0.5185, + "step": 1116 + }, + { + "epoch": 0.47, + "learning_rate": 1.99576540557063e-05, + "loss": 0.4802, + "step": 1117 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957498052438046e-05, + "loss": 0.5159, + "step": 1118 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957341762950346e-05, + "loss": 0.6096, + "step": 1119 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957185187247697e-05, + "loss": 0.5007, + "step": 1120 + }, + { + "epoch": 0.47, + "learning_rate": 1.995702832533459e-05, + "loss": 0.5597, + "step": 1121 + }, + { + "epoch": 0.47, + "learning_rate": 1.995687117721554e-05, + "loss": 0.4951, + "step": 1122 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956713742895068e-05, + "loss": 0.4691, + "step": 1123 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956556022377693e-05, + "loss": 0.5219, + "step": 1124 + }, + { + "epoch": 0.47, + "learning_rate": 1.995639801566795e-05, + "loss": 0.545, + "step": 1125 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956239722770382e-05, + "loss": 0.4985, + "step": 1126 + }, + { + "epoch": 0.47, + "learning_rate": 1.995608114368954e-05, + "loss": 0.5149, + "step": 1127 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955922278429975e-05, + "loss": 0.5308, + "step": 1128 + }, + { + "epoch": 0.47, + "learning_rate": 1.995576312699626e-05, + "loss": 0.5482, + "step": 1129 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955603689392974e-05, + "loss": 0.5766, + "step": 1130 + }, + { + "epoch": 0.47, + "learning_rate": 1.995544396562469e-05, + "loss": 0.4821, + "step": 1131 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955283955696004e-05, + "loss": 0.4794, + "step": 1132 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955123659611516e-05, + "loss": 0.5762, + "step": 1133 + }, + { + "epoch": 0.47, + "learning_rate": 1.995496307737583e-05, + "loss": 0.5338, + "step": 1134 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954802208993565e-05, + "loss": 0.574, + "step": 1135 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954641054469347e-05, + "loss": 0.6228, + "step": 1136 + }, + { + "epoch": 0.48, + "learning_rate": 1.99544796138078e-05, + "loss": 0.5616, + "step": 1137 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954317887013576e-05, + "loss": 0.5199, + "step": 1138 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954155874091313e-05, + "loss": 0.5464, + "step": 1139 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953993575045674e-05, + "loss": 0.5484, + "step": 1140 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953830989881324e-05, + "loss": 0.5235, + "step": 1141 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953668118602937e-05, + "loss": 0.5152, + "step": 1142 + }, + { + "epoch": 0.48, + "learning_rate": 1.995350496121519e-05, + "loss": 0.5461, + "step": 1143 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953341517722773e-05, + "loss": 0.4547, + "step": 1144 + }, + { + "epoch": 0.48, + "learning_rate": 1.995317778813039e-05, + "loss": 0.5252, + "step": 1145 + }, + { + "epoch": 0.48, + "learning_rate": 1.995301377244274e-05, + "loss": 0.5046, + "step": 1146 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952849470664547e-05, + "loss": 0.4885, + "step": 1147 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952684882800525e-05, + "loss": 0.5105, + "step": 1148 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952520008855405e-05, + "loss": 0.55, + "step": 1149 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952354848833932e-05, + "loss": 0.5413, + "step": 1150 + }, + { + "epoch": 0.48, + "learning_rate": 1.995218940274085e-05, + "loss": 0.552, + "step": 1151 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952023670580915e-05, + "loss": 0.6011, + "step": 1152 + }, + { + "epoch": 0.48, + "learning_rate": 1.995185765235889e-05, + "loss": 0.5404, + "step": 1153 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951691348079553e-05, + "loss": 0.4995, + "step": 1154 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951524757747675e-05, + "loss": 0.5654, + "step": 1155 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951357881368047e-05, + "loss": 0.5018, + "step": 1156 + }, + { + "epoch": 0.48, + "learning_rate": 1.995119071894547e-05, + "loss": 0.5189, + "step": 1157 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951023270484748e-05, + "loss": 0.5147, + "step": 1158 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950855535990688e-05, + "loss": 0.577, + "step": 1159 + }, + { + "epoch": 0.49, + "learning_rate": 1.995068751546812e-05, + "loss": 0.4954, + "step": 1160 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950519208921866e-05, + "loss": 0.5634, + "step": 1161 + }, + { + "epoch": 0.49, + "learning_rate": 1.995035061635677e-05, + "loss": 0.4313, + "step": 1162 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950181737777673e-05, + "loss": 0.4947, + "step": 1163 + }, + { + "epoch": 0.49, + "learning_rate": 1.995001257318943e-05, + "loss": 0.467, + "step": 1164 + }, + { + "epoch": 0.49, + "learning_rate": 1.994984312259691e-05, + "loss": 0.4954, + "step": 1165 + }, + { + "epoch": 0.49, + "learning_rate": 1.9949673386004975e-05, + "loss": 0.5324, + "step": 1166 + }, + { + "epoch": 0.49, + "learning_rate": 1.994950336341851e-05, + "loss": 0.4987, + "step": 1167 + }, + { + "epoch": 0.49, + "learning_rate": 1.9949333054842398e-05, + "loss": 0.5435, + "step": 1168 + }, + { + "epoch": 0.49, + "learning_rate": 1.994916246028154e-05, + "loss": 0.4749, + "step": 1169 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948991579740832e-05, + "loss": 0.4948, + "step": 1170 + }, + { + "epoch": 0.49, + "learning_rate": 1.994882041322519e-05, + "loss": 0.5437, + "step": 1171 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948648960739536e-05, + "loss": 0.5135, + "step": 1172 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948477222288792e-05, + "loss": 0.5212, + "step": 1173 + }, + { + "epoch": 0.49, + "learning_rate": 1.99483051978779e-05, + "loss": 0.4861, + "step": 1174 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948132887511805e-05, + "loss": 0.5396, + "step": 1175 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947960291195457e-05, + "loss": 0.5566, + "step": 1176 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947787408933816e-05, + "loss": 0.653, + "step": 1177 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947614240731856e-05, + "loss": 0.5085, + "step": 1178 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947440786594552e-05, + "loss": 0.4604, + "step": 1179 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947267046526887e-05, + "loss": 0.538, + "step": 1180 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947093020533863e-05, + "loss": 0.479, + "step": 1181 + }, + { + "epoch": 0.49, + "learning_rate": 1.9946918708620472e-05, + "loss": 0.5215, + "step": 1182 + }, + { + "epoch": 0.5, + "learning_rate": 1.994674411079173e-05, + "loss": 0.49, + "step": 1183 + }, + { + "epoch": 0.5, + "learning_rate": 1.994656922705266e-05, + "loss": 0.5296, + "step": 1184 + }, + { + "epoch": 0.5, + "learning_rate": 1.9946394057408278e-05, + "loss": 0.4532, + "step": 1185 + }, + { + "epoch": 0.5, + "learning_rate": 1.9946218601863626e-05, + "loss": 0.4952, + "step": 1186 + }, + { + "epoch": 0.5, + "learning_rate": 1.994604286042375e-05, + "loss": 0.5172, + "step": 1187 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945866833093694e-05, + "loss": 0.4908, + "step": 1188 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945690519878523e-05, + "loss": 0.6495, + "step": 1189 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945513920783303e-05, + "loss": 0.5633, + "step": 1190 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945337035813114e-05, + "loss": 0.4542, + "step": 1191 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945159864973035e-05, + "loss": 0.5391, + "step": 1192 + }, + { + "epoch": 0.5, + "learning_rate": 1.994498240826816e-05, + "loss": 0.5666, + "step": 1193 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944804665703593e-05, + "loss": 0.5144, + "step": 1194 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944626637284438e-05, + "loss": 0.4959, + "step": 1195 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944448323015818e-05, + "loss": 0.5488, + "step": 1196 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944269722902854e-05, + "loss": 0.4746, + "step": 1197 + }, + { + "epoch": 0.5, + "learning_rate": 1.994409083695068e-05, + "loss": 0.5414, + "step": 1198 + }, + { + "epoch": 0.5, + "learning_rate": 1.994391166516444e-05, + "loss": 0.5357, + "step": 1199 + }, + { + "epoch": 0.5, + "learning_rate": 1.9943732207549287e-05, + "loss": 0.5277, + "step": 1200 + }, + { + "epoch": 0.5, + "learning_rate": 1.994355246411037e-05, + "loss": 0.5926, + "step": 1201 + }, + { + "epoch": 0.5, + "learning_rate": 1.9943372434852864e-05, + "loss": 0.5177, + "step": 1202 + }, + { + "epoch": 0.5, + "learning_rate": 1.994319211978194e-05, + "loss": 0.5034, + "step": 1203 + }, + { + "epoch": 0.5, + "learning_rate": 1.994301151890278e-05, + "loss": 0.5085, + "step": 1204 + }, + { + "epoch": 0.5, + "learning_rate": 1.994283063222058e-05, + "loss": 0.6082, + "step": 1205 + }, + { + "epoch": 0.5, + "learning_rate": 1.9942649459740537e-05, + "loss": 0.4875, + "step": 1206 + }, + { + "epoch": 0.51, + "learning_rate": 1.9942468001467857e-05, + "loss": 0.5117, + "step": 1207 + }, + { + "epoch": 0.51, + "learning_rate": 1.994228625740776e-05, + "loss": 0.453, + "step": 1208 + }, + { + "epoch": 0.51, + "learning_rate": 1.994210422756546e-05, + "loss": 0.509, + "step": 1209 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941921911946202e-05, + "loss": 0.5297, + "step": 1210 + }, + { + "epoch": 0.51, + "learning_rate": 1.994173931055522e-05, + "loss": 0.5244, + "step": 1211 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941556423397762e-05, + "loss": 0.4848, + "step": 1212 + }, + { + "epoch": 0.51, + "learning_rate": 1.994137325047909e-05, + "loss": 0.5528, + "step": 1213 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941189791804466e-05, + "loss": 0.5117, + "step": 1214 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941006047379158e-05, + "loss": 0.4961, + "step": 1215 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940822017208458e-05, + "loss": 0.5069, + "step": 1216 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940637701297647e-05, + "loss": 0.4534, + "step": 1217 + }, + { + "epoch": 0.51, + "learning_rate": 1.994045309965203e-05, + "loss": 0.5828, + "step": 1218 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940268212276904e-05, + "loss": 0.4312, + "step": 1219 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940083039177594e-05, + "loss": 0.4555, + "step": 1220 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939897580359417e-05, + "loss": 0.5016, + "step": 1221 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939711835827703e-05, + "loss": 0.5193, + "step": 1222 + }, + { + "epoch": 0.51, + "learning_rate": 1.993952580558779e-05, + "loss": 0.4985, + "step": 1223 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939339489645034e-05, + "loss": 0.4723, + "step": 1224 + }, + { + "epoch": 0.51, + "learning_rate": 1.993915288800478e-05, + "loss": 0.4698, + "step": 1225 + }, + { + "epoch": 0.51, + "learning_rate": 1.99389660006724e-05, + "loss": 0.4951, + "step": 1226 + }, + { + "epoch": 0.51, + "learning_rate": 1.9938778827653256e-05, + "loss": 0.4973, + "step": 1227 + }, + { + "epoch": 0.51, + "learning_rate": 1.993859136895274e-05, + "loss": 0.5482, + "step": 1228 + }, + { + "epoch": 0.51, + "learning_rate": 1.993840362457623e-05, + "loss": 0.5661, + "step": 1229 + }, + { + "epoch": 0.51, + "learning_rate": 1.993821559452913e-05, + "loss": 0.5084, + "step": 1230 + }, + { + "epoch": 0.52, + "learning_rate": 1.993802727881684e-05, + "loss": 0.6094, + "step": 1231 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937838677444776e-05, + "loss": 0.4842, + "step": 1232 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937649790418357e-05, + "loss": 0.4732, + "step": 1233 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937460617743013e-05, + "loss": 0.5847, + "step": 1234 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937271159424185e-05, + "loss": 0.5418, + "step": 1235 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937081415467313e-05, + "loss": 0.4498, + "step": 1236 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936891385877854e-05, + "loss": 0.5086, + "step": 1237 + }, + { + "epoch": 0.52, + "learning_rate": 1.993670107066127e-05, + "loss": 0.4876, + "step": 1238 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936510469823028e-05, + "loss": 0.5137, + "step": 1239 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936319583368616e-05, + "loss": 0.4546, + "step": 1240 + }, + { + "epoch": 0.52, + "learning_rate": 1.993612841130351e-05, + "loss": 0.5644, + "step": 1241 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935936953633214e-05, + "loss": 0.5763, + "step": 1242 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935745210363224e-05, + "loss": 0.5405, + "step": 1243 + }, + { + "epoch": 0.52, + "learning_rate": 1.993555318149906e-05, + "loss": 0.4923, + "step": 1244 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935360867046228e-05, + "loss": 0.5439, + "step": 1245 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935168267010268e-05, + "loss": 0.4501, + "step": 1246 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934975381396716e-05, + "loss": 0.5177, + "step": 1247 + }, + { + "epoch": 0.52, + "learning_rate": 1.993478221021111e-05, + "loss": 0.5299, + "step": 1248 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934588753459006e-05, + "loss": 0.5357, + "step": 1249 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934395011145967e-05, + "loss": 0.4541, + "step": 1250 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934200983277553e-05, + "loss": 0.464, + "step": 1251 + }, + { + "epoch": 0.52, + "learning_rate": 1.993400666985935e-05, + "loss": 0.5593, + "step": 1252 + }, + { + "epoch": 0.52, + "learning_rate": 1.993381207089694e-05, + "loss": 0.516, + "step": 1253 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933617186395917e-05, + "loss": 0.6846, + "step": 1254 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933422016361886e-05, + "loss": 0.5269, + "step": 1255 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933226560800452e-05, + "loss": 0.5366, + "step": 1256 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933030819717235e-05, + "loss": 0.4933, + "step": 1257 + }, + { + "epoch": 0.53, + "learning_rate": 1.993283479311786e-05, + "loss": 0.5686, + "step": 1258 + }, + { + "epoch": 0.53, + "learning_rate": 1.993263848100796e-05, + "loss": 0.4922, + "step": 1259 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932441883393186e-05, + "loss": 0.546, + "step": 1260 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932245000279184e-05, + "loss": 0.5403, + "step": 1261 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932047831671615e-05, + "loss": 0.4769, + "step": 1262 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931850377576142e-05, + "loss": 0.5271, + "step": 1263 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931652637998443e-05, + "loss": 0.5356, + "step": 1264 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931454612944203e-05, + "loss": 0.5345, + "step": 1265 + }, + { + "epoch": 0.53, + "learning_rate": 1.993125630241911e-05, + "loss": 0.5233, + "step": 1266 + }, + { + "epoch": 0.53, + "learning_rate": 1.993105770642887e-05, + "loss": 0.4715, + "step": 1267 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930858824979192e-05, + "loss": 0.5269, + "step": 1268 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930659658075784e-05, + "loss": 0.5421, + "step": 1269 + }, + { + "epoch": 0.53, + "learning_rate": 1.993046020572438e-05, + "loss": 0.5315, + "step": 1270 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930260467930706e-05, + "loss": 0.6047, + "step": 1271 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930060444700508e-05, + "loss": 0.4419, + "step": 1272 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929860136039537e-05, + "loss": 0.6234, + "step": 1273 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929659541953548e-05, + "loss": 0.4614, + "step": 1274 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929458662448305e-05, + "loss": 0.5827, + "step": 1275 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929257497529584e-05, + "loss": 0.4641, + "step": 1276 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929056047203166e-05, + "loss": 0.4859, + "step": 1277 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928854311474844e-05, + "loss": 0.4513, + "step": 1278 + }, + { + "epoch": 0.54, + "learning_rate": 1.992865229035042e-05, + "loss": 0.6051, + "step": 1279 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928449983835688e-05, + "loss": 0.5138, + "step": 1280 + }, + { + "epoch": 0.54, + "learning_rate": 1.992824739193648e-05, + "loss": 0.5685, + "step": 1281 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928044514658604e-05, + "loss": 0.5332, + "step": 1282 + }, + { + "epoch": 0.54, + "learning_rate": 1.99278413520079e-05, + "loss": 0.5414, + "step": 1283 + }, + { + "epoch": 0.54, + "learning_rate": 1.992763790399021e-05, + "loss": 0.6006, + "step": 1284 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927434170611373e-05, + "loss": 0.519, + "step": 1285 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927230151877253e-05, + "loss": 0.5043, + "step": 1286 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927025847793714e-05, + "loss": 0.5028, + "step": 1287 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926821258366622e-05, + "loss": 0.5702, + "step": 1288 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926616383601862e-05, + "loss": 0.5905, + "step": 1289 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926411223505323e-05, + "loss": 0.5257, + "step": 1290 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926205778082903e-05, + "loss": 0.4561, + "step": 1291 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926000047340507e-05, + "loss": 0.5467, + "step": 1292 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925794031284046e-05, + "loss": 0.5829, + "step": 1293 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925587729919447e-05, + "loss": 0.5134, + "step": 1294 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925381143252633e-05, + "loss": 0.5486, + "step": 1295 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925174271289548e-05, + "loss": 0.5339, + "step": 1296 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924967114036133e-05, + "loss": 0.5039, + "step": 1297 + }, + { + "epoch": 0.54, + "learning_rate": 1.992475967149835e-05, + "loss": 0.5455, + "step": 1298 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924551943682153e-05, + "loss": 0.5704, + "step": 1299 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924343930593516e-05, + "loss": 0.5534, + "step": 1300 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924135632238423e-05, + "loss": 0.4944, + "step": 1301 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923927048622857e-05, + "loss": 0.5032, + "step": 1302 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923718179752815e-05, + "loss": 0.557, + "step": 1303 + }, + { + "epoch": 0.55, + "learning_rate": 1.99235090256343e-05, + "loss": 0.4871, + "step": 1304 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923299586273323e-05, + "loss": 0.5246, + "step": 1305 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923089861675906e-05, + "loss": 0.5811, + "step": 1306 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922879851848077e-05, + "loss": 0.5093, + "step": 1307 + }, + { + "epoch": 0.55, + "learning_rate": 1.992266955679587e-05, + "loss": 0.5393, + "step": 1308 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922458976525335e-05, + "loss": 0.533, + "step": 1309 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922248111042518e-05, + "loss": 0.4694, + "step": 1310 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922036960353486e-05, + "loss": 0.5957, + "step": 1311 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921825524464306e-05, + "loss": 0.5424, + "step": 1312 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921613803381055e-05, + "loss": 0.5007, + "step": 1313 + }, + { + "epoch": 0.55, + "learning_rate": 1.992140179710982e-05, + "loss": 0.4663, + "step": 1314 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921189505656696e-05, + "loss": 0.5117, + "step": 1315 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920976929027782e-05, + "loss": 0.4508, + "step": 1316 + }, + { + "epoch": 0.55, + "learning_rate": 1.992076406722919e-05, + "loss": 0.5698, + "step": 1317 + }, + { + "epoch": 0.55, + "learning_rate": 1.992055092026704e-05, + "loss": 0.5356, + "step": 1318 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920337488147453e-05, + "loss": 0.5155, + "step": 1319 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920123770876573e-05, + "loss": 0.6008, + "step": 1320 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919909768460535e-05, + "loss": 0.5054, + "step": 1321 + }, + { + "epoch": 0.55, + "learning_rate": 1.991969548090549e-05, + "loss": 0.4738, + "step": 1322 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919480908217606e-05, + "loss": 0.4882, + "step": 1323 + }, + { + "epoch": 0.55, + "learning_rate": 1.991926605040305e-05, + "loss": 0.5006, + "step": 1324 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919050907467986e-05, + "loss": 0.5821, + "step": 1325 + }, + { + "epoch": 0.56, + "learning_rate": 1.9918835479418607e-05, + "loss": 0.5611, + "step": 1326 + }, + { + "epoch": 0.56, + "learning_rate": 1.9918619766261106e-05, + "loss": 0.5482, + "step": 1327 + }, + { + "epoch": 0.56, + "learning_rate": 1.991840376800168e-05, + "loss": 0.6395, + "step": 1328 + }, + { + "epoch": 0.56, + "learning_rate": 1.991818748464654e-05, + "loss": 0.5458, + "step": 1329 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917970916201904e-05, + "loss": 0.5189, + "step": 1330 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917754062673992e-05, + "loss": 0.5461, + "step": 1331 + }, + { + "epoch": 0.56, + "learning_rate": 1.991753692406904e-05, + "loss": 0.4995, + "step": 1332 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917319500393294e-05, + "loss": 0.609, + "step": 1333 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917101791652997e-05, + "loss": 0.5375, + "step": 1334 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916883797854407e-05, + "loss": 0.4678, + "step": 1335 + }, + { + "epoch": 0.56, + "learning_rate": 1.991666551900379e-05, + "loss": 0.5375, + "step": 1336 + }, + { + "epoch": 0.56, + "learning_rate": 1.991644695510743e-05, + "loss": 0.441, + "step": 1337 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916228106171596e-05, + "loss": 0.5054, + "step": 1338 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916008972202586e-05, + "loss": 0.5001, + "step": 1339 + }, + { + "epoch": 0.56, + "learning_rate": 1.99157895532067e-05, + "loss": 0.574, + "step": 1340 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915569849190243e-05, + "loss": 0.5413, + "step": 1341 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915349860159528e-05, + "loss": 0.5348, + "step": 1342 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915129586120878e-05, + "loss": 0.4805, + "step": 1343 + }, + { + "epoch": 0.56, + "learning_rate": 1.991490902708063e-05, + "loss": 0.3818, + "step": 1344 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914688183045116e-05, + "loss": 0.5991, + "step": 1345 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914467054020692e-05, + "loss": 0.5398, + "step": 1346 + }, + { + "epoch": 0.56, + "learning_rate": 1.991424564001371e-05, + "loss": 0.5381, + "step": 1347 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914023941030534e-05, + "loss": 0.5163, + "step": 1348 + }, + { + "epoch": 0.56, + "learning_rate": 1.9913801957077535e-05, + "loss": 0.5226, + "step": 1349 + }, + { + "epoch": 0.57, + "learning_rate": 1.99135796881611e-05, + "loss": 0.5332, + "step": 1350 + }, + { + "epoch": 0.57, + "learning_rate": 1.991335713428761e-05, + "loss": 0.5834, + "step": 1351 + }, + { + "epoch": 0.57, + "learning_rate": 1.991313429546347e-05, + "loss": 0.5745, + "step": 1352 + }, + { + "epoch": 0.57, + "learning_rate": 1.991291117169508e-05, + "loss": 0.5629, + "step": 1353 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912687762988855e-05, + "loss": 0.4963, + "step": 1354 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912464069351217e-05, + "loss": 0.4239, + "step": 1355 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912240090788595e-05, + "loss": 0.4935, + "step": 1356 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912015827307426e-05, + "loss": 0.53, + "step": 1357 + }, + { + "epoch": 0.57, + "learning_rate": 1.991179127891416e-05, + "loss": 0.469, + "step": 1358 + }, + { + "epoch": 0.57, + "learning_rate": 1.9911566445615253e-05, + "loss": 0.537, + "step": 1359 + }, + { + "epoch": 0.57, + "learning_rate": 1.9911341327417156e-05, + "loss": 0.5467, + "step": 1360 + }, + { + "epoch": 0.57, + "learning_rate": 1.991111592432635e-05, + "loss": 0.4584, + "step": 1361 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910890236349316e-05, + "loss": 0.6264, + "step": 1362 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910664263492533e-05, + "loss": 0.5039, + "step": 1363 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910438005762503e-05, + "loss": 0.4549, + "step": 1364 + }, + { + "epoch": 0.57, + "learning_rate": 1.991021146316572e-05, + "loss": 0.5509, + "step": 1365 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909984635708714e-05, + "loss": 0.4958, + "step": 1366 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909757523397985e-05, + "loss": 0.5492, + "step": 1367 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909530126240074e-05, + "loss": 0.5841, + "step": 1368 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909302444241508e-05, + "loss": 0.545, + "step": 1369 + }, + { + "epoch": 0.57, + "learning_rate": 1.990907447740884e-05, + "loss": 0.4326, + "step": 1370 + }, + { + "epoch": 0.57, + "learning_rate": 1.9908846225748623e-05, + "loss": 0.5381, + "step": 1371 + }, + { + "epoch": 0.57, + "learning_rate": 1.990861768926741e-05, + "loss": 0.4573, + "step": 1372 + }, + { + "epoch": 0.57, + "learning_rate": 1.9908388867971772e-05, + "loss": 0.4962, + "step": 1373 + }, + { + "epoch": 0.58, + "learning_rate": 1.9908159761868292e-05, + "loss": 0.4974, + "step": 1374 + }, + { + "epoch": 0.58, + "learning_rate": 1.990793037096355e-05, + "loss": 0.4819, + "step": 1375 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907700695264144e-05, + "loss": 0.532, + "step": 1376 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907470734776674e-05, + "loss": 0.4813, + "step": 1377 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907240489507748e-05, + "loss": 0.483, + "step": 1378 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907009959463983e-05, + "loss": 0.6402, + "step": 1379 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906779144652014e-05, + "loss": 0.5279, + "step": 1380 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906548045078464e-05, + "loss": 0.5868, + "step": 1381 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906316660749985e-05, + "loss": 0.4664, + "step": 1382 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906084991673224e-05, + "loss": 0.571, + "step": 1383 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905853037854834e-05, + "loss": 0.5611, + "step": 1384 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905620799301495e-05, + "loss": 0.4523, + "step": 1385 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905388276019873e-05, + "loss": 0.6086, + "step": 1386 + }, + { + "epoch": 0.58, + "learning_rate": 1.990515546801666e-05, + "loss": 0.483, + "step": 1387 + }, + { + "epoch": 0.58, + "learning_rate": 1.990492237529854e-05, + "loss": 0.6169, + "step": 1388 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904688997872215e-05, + "loss": 0.4782, + "step": 1389 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904455335744395e-05, + "loss": 0.5676, + "step": 1390 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904221388921794e-05, + "loss": 0.6075, + "step": 1391 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903987157411138e-05, + "loss": 0.4489, + "step": 1392 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903752641219163e-05, + "loss": 0.5618, + "step": 1393 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903517840352605e-05, + "loss": 0.5132, + "step": 1394 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903282754818216e-05, + "loss": 0.4228, + "step": 1395 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903047384622748e-05, + "loss": 0.513, + "step": 1396 + }, + { + "epoch": 0.58, + "learning_rate": 1.9902811729772974e-05, + "loss": 0.5782, + "step": 1397 + }, + { + "epoch": 0.59, + "learning_rate": 1.990257579027566e-05, + "loss": 0.5138, + "step": 1398 + }, + { + "epoch": 0.59, + "learning_rate": 1.99023395661376e-05, + "loss": 0.5087, + "step": 1399 + }, + { + "epoch": 0.59, + "learning_rate": 1.9902103057365573e-05, + "loss": 0.6183, + "step": 1400 + }, + { + "epoch": 0.59, + "learning_rate": 1.990186626396638e-05, + "loss": 0.5741, + "step": 1401 + }, + { + "epoch": 0.59, + "learning_rate": 1.9901629185946827e-05, + "loss": 0.4439, + "step": 1402 + }, + { + "epoch": 0.59, + "learning_rate": 1.990139182331373e-05, + "loss": 0.6, + "step": 1403 + }, + { + "epoch": 0.59, + "learning_rate": 1.9901154176073914e-05, + "loss": 0.5193, + "step": 1404 + }, + { + "epoch": 0.59, + "learning_rate": 1.9900916244234203e-05, + "loss": 0.6317, + "step": 1405 + }, + { + "epoch": 0.59, + "learning_rate": 1.9900678027801445e-05, + "loss": 0.5093, + "step": 1406 + }, + { + "epoch": 0.59, + "learning_rate": 1.990043952678248e-05, + "loss": 0.4804, + "step": 1407 + }, + { + "epoch": 0.59, + "learning_rate": 1.990020074118417e-05, + "loss": 0.5628, + "step": 1408 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899961671013367e-05, + "loss": 0.5073, + "step": 1409 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899722316276958e-05, + "loss": 0.5252, + "step": 1410 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899482676981812e-05, + "loss": 0.5157, + "step": 1411 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899242753134824e-05, + "loss": 0.5309, + "step": 1412 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899002544742886e-05, + "loss": 0.4717, + "step": 1413 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898762051812906e-05, + "loss": 0.4811, + "step": 1414 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898521274351793e-05, + "loss": 0.496, + "step": 1415 + }, + { + "epoch": 0.59, + "learning_rate": 1.989828021236647e-05, + "loss": 0.4783, + "step": 1416 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898038865863864e-05, + "loss": 0.4625, + "step": 1417 + }, + { + "epoch": 0.59, + "learning_rate": 1.9897797234850916e-05, + "loss": 0.4816, + "step": 1418 + }, + { + "epoch": 0.59, + "learning_rate": 1.989755531933457e-05, + "loss": 0.5341, + "step": 1419 + }, + { + "epoch": 0.59, + "learning_rate": 1.989731311932178e-05, + "loss": 0.5596, + "step": 1420 + }, + { + "epoch": 0.59, + "learning_rate": 1.9897070634819506e-05, + "loss": 0.4642, + "step": 1421 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896827865834717e-05, + "loss": 0.4795, + "step": 1422 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896584812374397e-05, + "loss": 0.4566, + "step": 1423 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896341474445526e-05, + "loss": 0.5784, + "step": 1424 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896097852055103e-05, + "loss": 0.5538, + "step": 1425 + }, + { + "epoch": 0.6, + "learning_rate": 1.989585394521013e-05, + "loss": 0.5432, + "step": 1426 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895609753917614e-05, + "loss": 0.4524, + "step": 1427 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895365278184577e-05, + "loss": 0.4955, + "step": 1428 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895120518018046e-05, + "loss": 0.443, + "step": 1429 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894875473425057e-05, + "loss": 0.5583, + "step": 1430 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894630144412656e-05, + "loss": 0.5227, + "step": 1431 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894384530987885e-05, + "loss": 0.5223, + "step": 1432 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894138633157816e-05, + "loss": 0.4718, + "step": 1433 + }, + { + "epoch": 0.6, + "learning_rate": 1.989389245092951e-05, + "loss": 0.4776, + "step": 1434 + }, + { + "epoch": 0.6, + "learning_rate": 1.9893645984310047e-05, + "loss": 0.4808, + "step": 1435 + }, + { + "epoch": 0.6, + "learning_rate": 1.989339923330651e-05, + "loss": 0.5571, + "step": 1436 + }, + { + "epoch": 0.6, + "learning_rate": 1.989315219792599e-05, + "loss": 0.4956, + "step": 1437 + }, + { + "epoch": 0.6, + "learning_rate": 1.989290487817559e-05, + "loss": 0.5159, + "step": 1438 + }, + { + "epoch": 0.6, + "learning_rate": 1.9892657274062415e-05, + "loss": 0.5385, + "step": 1439 + }, + { + "epoch": 0.6, + "learning_rate": 1.9892409385593586e-05, + "loss": 0.6072, + "step": 1440 + }, + { + "epoch": 0.6, + "learning_rate": 1.989216121277623e-05, + "loss": 0.4643, + "step": 1441 + }, + { + "epoch": 0.6, + "learning_rate": 1.9891912755617477e-05, + "loss": 0.4857, + "step": 1442 + }, + { + "epoch": 0.6, + "learning_rate": 1.989166401412447e-05, + "loss": 0.4994, + "step": 1443 + }, + { + "epoch": 0.6, + "learning_rate": 1.9891414988304357e-05, + "loss": 0.4851, + "step": 1444 + }, + { + "epoch": 0.6, + "learning_rate": 1.98911656781643e-05, + "loss": 0.5248, + "step": 1445 + }, + { + "epoch": 0.61, + "learning_rate": 1.9890916083711463e-05, + "loss": 0.5135, + "step": 1446 + }, + { + "epoch": 0.61, + "learning_rate": 1.9890666204953018e-05, + "loss": 0.5424, + "step": 1447 + }, + { + "epoch": 0.61, + "learning_rate": 1.989041604189615e-05, + "loss": 0.4628, + "step": 1448 + }, + { + "epoch": 0.61, + "learning_rate": 1.989016559454805e-05, + "loss": 0.5106, + "step": 1449 + }, + { + "epoch": 0.61, + "learning_rate": 1.988991486291592e-05, + "loss": 0.5695, + "step": 1450 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889663847006956e-05, + "loss": 0.5715, + "step": 1451 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889412546828386e-05, + "loss": 0.4649, + "step": 1452 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889160962387427e-05, + "loss": 0.5909, + "step": 1453 + }, + { + "epoch": 0.61, + "learning_rate": 1.988890909369131e-05, + "loss": 0.5654, + "step": 1454 + }, + { + "epoch": 0.61, + "learning_rate": 1.988865694074728e-05, + "loss": 0.5696, + "step": 1455 + }, + { + "epoch": 0.61, + "learning_rate": 1.988840450356258e-05, + "loss": 0.486, + "step": 1456 + }, + { + "epoch": 0.61, + "learning_rate": 1.9888151782144463e-05, + "loss": 0.4263, + "step": 1457 + }, + { + "epoch": 0.61, + "learning_rate": 1.9887898776500203e-05, + "loss": 0.4926, + "step": 1458 + }, + { + "epoch": 0.61, + "learning_rate": 1.9887645486637063e-05, + "loss": 0.5234, + "step": 1459 + }, + { + "epoch": 0.61, + "learning_rate": 1.988739191256233e-05, + "loss": 0.4934, + "step": 1460 + }, + { + "epoch": 0.61, + "learning_rate": 1.988713805428329e-05, + "loss": 0.5068, + "step": 1461 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886883911807242e-05, + "loss": 0.5427, + "step": 1462 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886629485141488e-05, + "loss": 0.524, + "step": 1463 + }, + { + "epoch": 0.61, + "learning_rate": 1.988637477429334e-05, + "loss": 0.4719, + "step": 1464 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886119779270126e-05, + "loss": 0.5084, + "step": 1465 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885864500079172e-05, + "loss": 0.4963, + "step": 1466 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885608936727814e-05, + "loss": 0.5211, + "step": 1467 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885353089223397e-05, + "loss": 0.4711, + "step": 1468 + }, + { + "epoch": 0.62, + "learning_rate": 1.988509695757328e-05, + "loss": 0.5843, + "step": 1469 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884840541784824e-05, + "loss": 0.5093, + "step": 1470 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884583841865395e-05, + "loss": 0.4973, + "step": 1471 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884326857822378e-05, + "loss": 0.5495, + "step": 1472 + }, + { + "epoch": 0.62, + "learning_rate": 1.988406958966316e-05, + "loss": 0.6042, + "step": 1473 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883812037395126e-05, + "loss": 0.4894, + "step": 1474 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883554201025687e-05, + "loss": 0.5177, + "step": 1475 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883296080562255e-05, + "loss": 0.4454, + "step": 1476 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883037676012245e-05, + "loss": 0.534, + "step": 1477 + }, + { + "epoch": 0.62, + "learning_rate": 1.988277898738309e-05, + "loss": 0.4713, + "step": 1478 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882520014682218e-05, + "loss": 0.5647, + "step": 1479 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882260757917083e-05, + "loss": 0.5868, + "step": 1480 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882001217095127e-05, + "loss": 0.5463, + "step": 1481 + }, + { + "epoch": 0.62, + "learning_rate": 1.9881741392223822e-05, + "loss": 0.5546, + "step": 1482 + }, + { + "epoch": 0.62, + "learning_rate": 1.9881481283310624e-05, + "loss": 0.5102, + "step": 1483 + }, + { + "epoch": 0.62, + "learning_rate": 1.988122089036302e-05, + "loss": 0.4513, + "step": 1484 + }, + { + "epoch": 0.62, + "learning_rate": 1.9880960213388483e-05, + "loss": 0.5327, + "step": 1485 + }, + { + "epoch": 0.62, + "learning_rate": 1.988069925239452e-05, + "loss": 0.5329, + "step": 1486 + }, + { + "epoch": 0.62, + "learning_rate": 1.988043800738862e-05, + "loss": 0.522, + "step": 1487 + }, + { + "epoch": 0.62, + "learning_rate": 1.98801764783783e-05, + "loss": 0.5105, + "step": 1488 + }, + { + "epoch": 0.62, + "learning_rate": 1.987991466537107e-05, + "loss": 0.5621, + "step": 1489 + }, + { + "epoch": 0.62, + "learning_rate": 1.9879652568374467e-05, + "loss": 0.5259, + "step": 1490 + }, + { + "epoch": 0.62, + "learning_rate": 1.987939018739601e-05, + "loss": 0.464, + "step": 1491 + }, + { + "epoch": 0.62, + "learning_rate": 1.9879127522443255e-05, + "loss": 0.5808, + "step": 1492 + }, + { + "epoch": 0.63, + "learning_rate": 1.9878864573523744e-05, + "loss": 0.5933, + "step": 1493 + }, + { + "epoch": 0.63, + "learning_rate": 1.987860134064504e-05, + "loss": 0.4712, + "step": 1494 + }, + { + "epoch": 0.63, + "learning_rate": 1.98783378238147e-05, + "loss": 0.5199, + "step": 1495 + }, + { + "epoch": 0.63, + "learning_rate": 1.9878074023040313e-05, + "loss": 0.5006, + "step": 1496 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877809938329447e-05, + "loss": 0.5374, + "step": 1497 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877545569689705e-05, + "loss": 0.4978, + "step": 1498 + }, + { + "epoch": 0.63, + "learning_rate": 1.987728091712868e-05, + "loss": 0.4904, + "step": 1499 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877015980653978e-05, + "loss": 0.5521, + "step": 1500 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876750760273218e-05, + "loss": 0.6263, + "step": 1501 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876485255994022e-05, + "loss": 0.5446, + "step": 1502 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876219467824018e-05, + "loss": 0.4413, + "step": 1503 + }, + { + "epoch": 0.63, + "learning_rate": 1.987595339577085e-05, + "loss": 0.4664, + "step": 1504 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875687039842168e-05, + "loss": 0.5025, + "step": 1505 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875420400045625e-05, + "loss": 0.5362, + "step": 1506 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875153476388887e-05, + "loss": 0.5398, + "step": 1507 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874886268879625e-05, + "loss": 0.4848, + "step": 1508 + }, + { + "epoch": 0.63, + "learning_rate": 1.987461877752552e-05, + "loss": 0.4712, + "step": 1509 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874351002334257e-05, + "loss": 0.5298, + "step": 1510 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874082943313543e-05, + "loss": 0.5308, + "step": 1511 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873814600471073e-05, + "loss": 0.5389, + "step": 1512 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873545973814564e-05, + "loss": 0.4738, + "step": 1513 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873277063351737e-05, + "loss": 0.5454, + "step": 1514 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873007869090326e-05, + "loss": 0.5225, + "step": 1515 + }, + { + "epoch": 0.63, + "learning_rate": 1.9872738391038063e-05, + "loss": 0.4892, + "step": 1516 + }, + { + "epoch": 0.64, + "learning_rate": 1.9872468629202696e-05, + "loss": 0.5003, + "step": 1517 + }, + { + "epoch": 0.64, + "learning_rate": 1.9872198583591976e-05, + "loss": 0.5106, + "step": 1518 + }, + { + "epoch": 0.64, + "learning_rate": 1.987192825421367e-05, + "loss": 0.6494, + "step": 1519 + }, + { + "epoch": 0.64, + "learning_rate": 1.987165764107555e-05, + "loss": 0.5218, + "step": 1520 + }, + { + "epoch": 0.64, + "learning_rate": 1.9871386744185386e-05, + "loss": 0.5341, + "step": 1521 + }, + { + "epoch": 0.64, + "learning_rate": 1.9871115563550974e-05, + "loss": 0.4962, + "step": 1522 + }, + { + "epoch": 0.64, + "learning_rate": 1.98708440991801e-05, + "loss": 0.5314, + "step": 1523 + }, + { + "epoch": 0.64, + "learning_rate": 1.9870572351080573e-05, + "loss": 0.5039, + "step": 1524 + }, + { + "epoch": 0.64, + "learning_rate": 1.9870300319260204e-05, + "loss": 0.4734, + "step": 1525 + }, + { + "epoch": 0.64, + "learning_rate": 1.987002800372681e-05, + "loss": 0.565, + "step": 1526 + }, + { + "epoch": 0.64, + "learning_rate": 1.986975540448822e-05, + "loss": 0.5463, + "step": 1527 + }, + { + "epoch": 0.64, + "learning_rate": 1.986948252155227e-05, + "loss": 0.504, + "step": 1528 + }, + { + "epoch": 0.64, + "learning_rate": 1.9869209354926802e-05, + "loss": 0.4311, + "step": 1529 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868935904619673e-05, + "loss": 0.4794, + "step": 1530 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868662170638732e-05, + "loss": 0.5152, + "step": 1531 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868388152991858e-05, + "loss": 0.5112, + "step": 1532 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868113851686925e-05, + "loss": 0.5007, + "step": 1533 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867839266731813e-05, + "loss": 0.5184, + "step": 1534 + }, + { + "epoch": 0.64, + "learning_rate": 1.986756439813442e-05, + "loss": 0.4357, + "step": 1535 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867289245902645e-05, + "loss": 0.532, + "step": 1536 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867013810044394e-05, + "loss": 0.5021, + "step": 1537 + }, + { + "epoch": 0.64, + "learning_rate": 1.9866738090567586e-05, + "loss": 0.4668, + "step": 1538 + }, + { + "epoch": 0.64, + "learning_rate": 1.9866462087480153e-05, + "loss": 0.4493, + "step": 1539 + }, + { + "epoch": 0.64, + "learning_rate": 1.986618580079002e-05, + "loss": 0.503, + "step": 1540 + }, + { + "epoch": 0.65, + "learning_rate": 1.986590923050513e-05, + "loss": 0.5003, + "step": 1541 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865632376633433e-05, + "loss": 0.5519, + "step": 1542 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865355239182892e-05, + "loss": 0.4793, + "step": 1543 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865077818161464e-05, + "loss": 0.4704, + "step": 1544 + }, + { + "epoch": 0.65, + "learning_rate": 1.9864800113577133e-05, + "loss": 0.5161, + "step": 1545 + }, + { + "epoch": 0.65, + "learning_rate": 1.9864522125437874e-05, + "loss": 0.4979, + "step": 1546 + }, + { + "epoch": 0.65, + "learning_rate": 1.986424385375168e-05, + "loss": 0.5197, + "step": 1547 + }, + { + "epoch": 0.65, + "learning_rate": 1.986396529852655e-05, + "loss": 0.4897, + "step": 1548 + }, + { + "epoch": 0.65, + "learning_rate": 1.986368645977049e-05, + "loss": 0.4824, + "step": 1549 + }, + { + "epoch": 0.65, + "learning_rate": 1.9863407337491516e-05, + "loss": 0.565, + "step": 1550 + }, + { + "epoch": 0.65, + "learning_rate": 1.986312793169765e-05, + "loss": 0.5372, + "step": 1551 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862848242396927e-05, + "loss": 0.485, + "step": 1552 + }, + { + "epoch": 0.65, + "learning_rate": 1.986256826959738e-05, + "loss": 0.5229, + "step": 1553 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862288013307062e-05, + "loss": 0.583, + "step": 1554 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862007473534026e-05, + "loss": 0.5444, + "step": 1555 + }, + { + "epoch": 0.65, + "learning_rate": 1.9861726650286338e-05, + "loss": 0.5535, + "step": 1556 + }, + { + "epoch": 0.65, + "learning_rate": 1.986144554357207e-05, + "loss": 0.4921, + "step": 1557 + }, + { + "epoch": 0.65, + "learning_rate": 1.9861164153399297e-05, + "loss": 0.5669, + "step": 1558 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860882479776113e-05, + "loss": 0.5139, + "step": 1559 + }, + { + "epoch": 0.65, + "learning_rate": 1.986060052271061e-05, + "loss": 0.5647, + "step": 1560 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860318282210896e-05, + "loss": 0.6061, + "step": 1561 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860035758285082e-05, + "loss": 0.5984, + "step": 1562 + }, + { + "epoch": 0.65, + "learning_rate": 1.9859752950941292e-05, + "loss": 0.4112, + "step": 1563 + }, + { + "epoch": 0.65, + "learning_rate": 1.9859469860187654e-05, + "loss": 0.4325, + "step": 1564 + }, + { + "epoch": 0.66, + "learning_rate": 1.9859186486032302e-05, + "loss": 0.5618, + "step": 1565 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858902828483382e-05, + "loss": 0.5109, + "step": 1566 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858618887549053e-05, + "loss": 0.5181, + "step": 1567 + }, + { + "epoch": 0.66, + "learning_rate": 1.985833466323747e-05, + "loss": 0.5925, + "step": 1568 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858050155556807e-05, + "loss": 0.5414, + "step": 1569 + }, + { + "epoch": 0.66, + "learning_rate": 1.9857765364515234e-05, + "loss": 0.5568, + "step": 1570 + }, + { + "epoch": 0.66, + "learning_rate": 1.9857480290120947e-05, + "loss": 0.5295, + "step": 1571 + }, + { + "epoch": 0.66, + "learning_rate": 1.985719493238214e-05, + "loss": 0.5144, + "step": 1572 + }, + { + "epoch": 0.66, + "learning_rate": 1.9856909291307008e-05, + "loss": 0.4398, + "step": 1573 + }, + { + "epoch": 0.66, + "learning_rate": 1.9856623366903765e-05, + "loss": 0.5049, + "step": 1574 + }, + { + "epoch": 0.66, + "learning_rate": 1.985633715918063e-05, + "loss": 0.539, + "step": 1575 + }, + { + "epoch": 0.66, + "learning_rate": 1.985605066814583e-05, + "loss": 0.4789, + "step": 1576 + }, + { + "epoch": 0.66, + "learning_rate": 1.9855763893807603e-05, + "loss": 0.4971, + "step": 1577 + }, + { + "epoch": 0.66, + "learning_rate": 1.9855476836174182e-05, + "loss": 0.4828, + "step": 1578 + }, + { + "epoch": 0.66, + "learning_rate": 1.985518949525383e-05, + "loss": 0.4806, + "step": 1579 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854901871054803e-05, + "loss": 0.5103, + "step": 1580 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854613963585362e-05, + "loss": 0.4901, + "step": 1581 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854325772853792e-05, + "loss": 0.5219, + "step": 1582 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854037298868367e-05, + "loss": 0.5073, + "step": 1583 + }, + { + "epoch": 0.66, + "learning_rate": 1.9853748541637387e-05, + "loss": 0.5223, + "step": 1584 + }, + { + "epoch": 0.66, + "learning_rate": 1.985345950116915e-05, + "loss": 0.525, + "step": 1585 + }, + { + "epoch": 0.66, + "learning_rate": 1.9853170177471962e-05, + "loss": 0.4888, + "step": 1586 + }, + { + "epoch": 0.66, + "learning_rate": 1.9852880570554142e-05, + "loss": 0.5425, + "step": 1587 + }, + { + "epoch": 0.66, + "learning_rate": 1.9852590680424012e-05, + "loss": 0.5606, + "step": 1588 + }, + { + "epoch": 0.67, + "learning_rate": 1.9852300507089907e-05, + "loss": 0.5552, + "step": 1589 + }, + { + "epoch": 0.67, + "learning_rate": 1.9852010050560165e-05, + "loss": 0.4529, + "step": 1590 + }, + { + "epoch": 0.67, + "learning_rate": 1.985171931084314e-05, + "loss": 0.3869, + "step": 1591 + }, + { + "epoch": 0.67, + "learning_rate": 1.9851428287947184e-05, + "loss": 0.4744, + "step": 1592 + }, + { + "epoch": 0.67, + "learning_rate": 1.9851136981880662e-05, + "loss": 0.5176, + "step": 1593 + }, + { + "epoch": 0.67, + "learning_rate": 1.985084539265195e-05, + "loss": 0.4943, + "step": 1594 + }, + { + "epoch": 0.67, + "learning_rate": 1.985055352026943e-05, + "loss": 0.4605, + "step": 1595 + }, + { + "epoch": 0.67, + "learning_rate": 1.985026136474149e-05, + "loss": 0.564, + "step": 1596 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849968926076528e-05, + "loss": 0.5434, + "step": 1597 + }, + { + "epoch": 0.67, + "learning_rate": 1.984967620428295e-05, + "loss": 0.5173, + "step": 1598 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849383199369167e-05, + "loss": 0.5304, + "step": 1599 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849089911343608e-05, + "loss": 0.5106, + "step": 1600 + }, + { + "epoch": 0.67, + "learning_rate": 1.98487963402147e-05, + "loss": 0.4964, + "step": 1601 + }, + { + "epoch": 0.67, + "learning_rate": 1.9848502485990876e-05, + "loss": 0.4417, + "step": 1602 + }, + { + "epoch": 0.67, + "learning_rate": 1.9848208348680594e-05, + "loss": 0.4463, + "step": 1603 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847913928292297e-05, + "loss": 0.5096, + "step": 1604 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847619224834455e-05, + "loss": 0.5129, + "step": 1605 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847324238315536e-05, + "loss": 0.4326, + "step": 1606 + }, + { + "epoch": 0.67, + "learning_rate": 1.984702896874402e-05, + "loss": 0.5151, + "step": 1607 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846733416128393e-05, + "loss": 0.6225, + "step": 1608 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846437580477155e-05, + "loss": 0.5483, + "step": 1609 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846141461798804e-05, + "loss": 0.5227, + "step": 1610 + }, + { + "epoch": 0.67, + "learning_rate": 1.984584506010186e-05, + "loss": 0.4964, + "step": 1611 + }, + { + "epoch": 0.67, + "learning_rate": 1.984554837539483e-05, + "loss": 0.5653, + "step": 1612 + }, + { + "epoch": 0.68, + "learning_rate": 1.984525140768625e-05, + "loss": 0.5379, + "step": 1613 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844954156984656e-05, + "loss": 0.5595, + "step": 1614 + }, + { + "epoch": 0.68, + "learning_rate": 1.984465662329859e-05, + "loss": 0.5941, + "step": 1615 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844358806636605e-05, + "loss": 0.4752, + "step": 1616 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844060707007263e-05, + "loss": 0.5019, + "step": 1617 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843762324419132e-05, + "loss": 0.5074, + "step": 1618 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843463658880786e-05, + "loss": 0.628, + "step": 1619 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843164710400815e-05, + "loss": 0.5083, + "step": 1620 + }, + { + "epoch": 0.68, + "learning_rate": 1.9842865478987808e-05, + "loss": 0.4711, + "step": 1621 + }, + { + "epoch": 0.68, + "learning_rate": 1.9842565964650363e-05, + "loss": 0.5057, + "step": 1622 + }, + { + "epoch": 0.68, + "learning_rate": 1.98422661673971e-05, + "loss": 0.5801, + "step": 1623 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841966087236626e-05, + "loss": 0.4771, + "step": 1624 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841665724177572e-05, + "loss": 0.5187, + "step": 1625 + }, + { + "epoch": 0.68, + "learning_rate": 1.984136507822857e-05, + "loss": 0.4719, + "step": 1626 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841064149398262e-05, + "loss": 0.4859, + "step": 1627 + }, + { + "epoch": 0.68, + "learning_rate": 1.9840762937695296e-05, + "loss": 0.4938, + "step": 1628 + }, + { + "epoch": 0.68, + "learning_rate": 1.9840461443128333e-05, + "loss": 0.5628, + "step": 1629 + }, + { + "epoch": 0.68, + "learning_rate": 1.984015966570604e-05, + "loss": 0.5753, + "step": 1630 + }, + { + "epoch": 0.68, + "learning_rate": 1.9839857605437086e-05, + "loss": 0.4204, + "step": 1631 + }, + { + "epoch": 0.68, + "learning_rate": 1.983955526233016e-05, + "loss": 0.4693, + "step": 1632 + }, + { + "epoch": 0.68, + "learning_rate": 1.9839252636393946e-05, + "loss": 0.5352, + "step": 1633 + }, + { + "epoch": 0.68, + "learning_rate": 1.983894972763715e-05, + "loss": 0.5136, + "step": 1634 + }, + { + "epoch": 0.68, + "learning_rate": 1.9838646536068473e-05, + "loss": 0.5446, + "step": 1635 + }, + { + "epoch": 0.68, + "learning_rate": 1.9838343061696633e-05, + "loss": 0.5841, + "step": 1636 + }, + { + "epoch": 0.69, + "learning_rate": 1.983803930453035e-05, + "loss": 0.5229, + "step": 1637 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837735264578358e-05, + "loss": 0.5148, + "step": 1638 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837430941849395e-05, + "loss": 0.4822, + "step": 1639 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837126336352214e-05, + "loss": 0.5169, + "step": 1640 + }, + { + "epoch": 0.69, + "learning_rate": 1.983682144809556e-05, + "loss": 0.4126, + "step": 1641 + }, + { + "epoch": 0.69, + "learning_rate": 1.9836516277088203e-05, + "loss": 0.4774, + "step": 1642 + }, + { + "epoch": 0.69, + "learning_rate": 1.9836210823338917e-05, + "loss": 0.5266, + "step": 1643 + }, + { + "epoch": 0.69, + "learning_rate": 1.983590508685648e-05, + "loss": 0.4671, + "step": 1644 + }, + { + "epoch": 0.69, + "learning_rate": 1.9835599067649676e-05, + "loss": 0.5408, + "step": 1645 + }, + { + "epoch": 0.69, + "learning_rate": 1.983529276572731e-05, + "loss": 0.5083, + "step": 1646 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834986181098176e-05, + "loss": 0.5176, + "step": 1647 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834679313771095e-05, + "loss": 0.5439, + "step": 1648 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834372163754882e-05, + "loss": 0.5107, + "step": 1649 + }, + { + "epoch": 0.69, + "learning_rate": 1.983406473105837e-05, + "loss": 0.5676, + "step": 1650 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833757015690393e-05, + "loss": 0.4696, + "step": 1651 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833449017659796e-05, + "loss": 0.5984, + "step": 1652 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833140736975433e-05, + "loss": 0.5636, + "step": 1653 + }, + { + "epoch": 0.69, + "learning_rate": 1.9832832173646164e-05, + "loss": 0.5231, + "step": 1654 + }, + { + "epoch": 0.69, + "learning_rate": 1.9832523327680865e-05, + "loss": 0.5034, + "step": 1655 + }, + { + "epoch": 0.69, + "learning_rate": 1.98322141990884e-05, + "loss": 0.5617, + "step": 1656 + }, + { + "epoch": 0.69, + "learning_rate": 1.983190478787767e-05, + "loss": 0.4909, + "step": 1657 + }, + { + "epoch": 0.69, + "learning_rate": 1.983159509405756e-05, + "loss": 0.4736, + "step": 1658 + }, + { + "epoch": 0.69, + "learning_rate": 1.9831285117636972e-05, + "loss": 0.592, + "step": 1659 + }, + { + "epoch": 0.69, + "learning_rate": 1.983097485862482e-05, + "loss": 0.5005, + "step": 1660 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830664317030014e-05, + "loss": 0.5292, + "step": 1661 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830353492861493e-05, + "loss": 0.5129, + "step": 1662 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830042386128178e-05, + "loss": 0.575, + "step": 1663 + }, + { + "epoch": 0.7, + "learning_rate": 1.982973099683902e-05, + "loss": 0.4696, + "step": 1664 + }, + { + "epoch": 0.7, + "learning_rate": 1.9829419325002966e-05, + "loss": 0.4746, + "step": 1665 + }, + { + "epoch": 0.7, + "learning_rate": 1.982910737062898e-05, + "loss": 0.4449, + "step": 1666 + }, + { + "epoch": 0.7, + "learning_rate": 1.982879513372602e-05, + "loss": 0.4704, + "step": 1667 + }, + { + "epoch": 0.7, + "learning_rate": 1.9828482614303068e-05, + "loss": 0.5179, + "step": 1668 + }, + { + "epoch": 0.7, + "learning_rate": 1.9828169812369106e-05, + "loss": 0.4781, + "step": 1669 + }, + { + "epoch": 0.7, + "learning_rate": 1.9827856727933122e-05, + "loss": 0.5388, + "step": 1670 + }, + { + "epoch": 0.7, + "learning_rate": 1.9827543361004118e-05, + "loss": 0.5813, + "step": 1671 + }, + { + "epoch": 0.7, + "learning_rate": 1.98272297115911e-05, + "loss": 0.4643, + "step": 1672 + }, + { + "epoch": 0.7, + "learning_rate": 1.9826915779703087e-05, + "loss": 0.4306, + "step": 1673 + }, + { + "epoch": 0.7, + "learning_rate": 1.98266015653491e-05, + "loss": 0.5216, + "step": 1674 + }, + { + "epoch": 0.7, + "learning_rate": 1.9826287068538167e-05, + "loss": 0.5218, + "step": 1675 + }, + { + "epoch": 0.7, + "learning_rate": 1.9825972289279336e-05, + "loss": 0.519, + "step": 1676 + }, + { + "epoch": 0.7, + "learning_rate": 1.982565722758165e-05, + "loss": 0.4881, + "step": 1677 + }, + { + "epoch": 0.7, + "learning_rate": 1.9825341883454165e-05, + "loss": 0.5389, + "step": 1678 + }, + { + "epoch": 0.7, + "learning_rate": 1.982502625690595e-05, + "loss": 0.5541, + "step": 1679 + }, + { + "epoch": 0.7, + "learning_rate": 1.982471034794607e-05, + "loss": 0.4128, + "step": 1680 + }, + { + "epoch": 0.7, + "learning_rate": 1.982439415658361e-05, + "loss": 0.4948, + "step": 1681 + }, + { + "epoch": 0.7, + "learning_rate": 1.9824077682827658e-05, + "loss": 0.4972, + "step": 1682 + }, + { + "epoch": 0.7, + "learning_rate": 1.982376092668731e-05, + "loss": 0.5727, + "step": 1683 + }, + { + "epoch": 0.71, + "learning_rate": 1.9823443888171676e-05, + "loss": 0.5451, + "step": 1684 + }, + { + "epoch": 0.71, + "learning_rate": 1.982312656728986e-05, + "loss": 0.4172, + "step": 1685 + }, + { + "epoch": 0.71, + "learning_rate": 1.9822808964050987e-05, + "loss": 0.5163, + "step": 1686 + }, + { + "epoch": 0.71, + "learning_rate": 1.982249107846419e-05, + "loss": 0.5484, + "step": 1687 + }, + { + "epoch": 0.71, + "learning_rate": 1.98221729105386e-05, + "loss": 0.5193, + "step": 1688 + }, + { + "epoch": 0.71, + "learning_rate": 1.982185446028337e-05, + "loss": 0.483, + "step": 1689 + }, + { + "epoch": 0.71, + "learning_rate": 1.9821535727707647e-05, + "loss": 0.519, + "step": 1690 + }, + { + "epoch": 0.71, + "learning_rate": 1.9821216712820597e-05, + "loss": 0.5161, + "step": 1691 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820897415631383e-05, + "loss": 0.5653, + "step": 1692 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820577836149192e-05, + "loss": 0.4888, + "step": 1693 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820257974383205e-05, + "loss": 0.4977, + "step": 1694 + }, + { + "epoch": 0.71, + "learning_rate": 1.9819937830342618e-05, + "loss": 0.4693, + "step": 1695 + }, + { + "epoch": 0.71, + "learning_rate": 1.981961740403663e-05, + "loss": 0.493, + "step": 1696 + }, + { + "epoch": 0.71, + "learning_rate": 1.9819296695474454e-05, + "loss": 0.5228, + "step": 1697 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818975704665306e-05, + "loss": 0.4938, + "step": 1698 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818654431618416e-05, + "loss": 0.4252, + "step": 1699 + }, + { + "epoch": 0.71, + "learning_rate": 1.981833287634302e-05, + "loss": 0.4584, + "step": 1700 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818011038848355e-05, + "loss": 0.5541, + "step": 1701 + }, + { + "epoch": 0.71, + "learning_rate": 1.9817688919143676e-05, + "loss": 0.412, + "step": 1702 + }, + { + "epoch": 0.71, + "learning_rate": 1.981736651723824e-05, + "loss": 0.5411, + "step": 1703 + }, + { + "epoch": 0.71, + "learning_rate": 1.9817043833141317e-05, + "loss": 0.4207, + "step": 1704 + }, + { + "epoch": 0.71, + "learning_rate": 1.981672086686218e-05, + "loss": 0.502, + "step": 1705 + }, + { + "epoch": 0.71, + "learning_rate": 1.9816397618410112e-05, + "loss": 0.5266, + "step": 1706 + }, + { + "epoch": 0.71, + "learning_rate": 1.9816074087794406e-05, + "loss": 0.5749, + "step": 1707 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815750275024362e-05, + "loss": 0.5609, + "step": 1708 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815426180109286e-05, + "loss": 0.5089, + "step": 1709 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815101803058492e-05, + "loss": 0.5283, + "step": 1710 + }, + { + "epoch": 0.72, + "learning_rate": 1.981477714388131e-05, + "loss": 0.4316, + "step": 1711 + }, + { + "epoch": 0.72, + "learning_rate": 1.9814452202587066e-05, + "loss": 0.4375, + "step": 1712 + }, + { + "epoch": 0.72, + "learning_rate": 1.9814126979185105e-05, + "loss": 0.5121, + "step": 1713 + }, + { + "epoch": 0.72, + "learning_rate": 1.981380147368477e-05, + "loss": 0.4719, + "step": 1714 + }, + { + "epoch": 0.72, + "learning_rate": 1.981347568609542e-05, + "loss": 0.4747, + "step": 1715 + }, + { + "epoch": 0.72, + "learning_rate": 1.9813149616426424e-05, + "loss": 0.5267, + "step": 1716 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812823264687148e-05, + "loss": 0.4687, + "step": 1717 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812496630886974e-05, + "loss": 0.5243, + "step": 1718 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812169715035295e-05, + "loss": 0.5228, + "step": 1719 + }, + { + "epoch": 0.72, + "learning_rate": 1.98118425171415e-05, + "loss": 0.5324, + "step": 1720 + }, + { + "epoch": 0.72, + "learning_rate": 1.9811515037215002e-05, + "loss": 0.5004, + "step": 1721 + }, + { + "epoch": 0.72, + "learning_rate": 1.981118727526521e-05, + "loss": 0.4853, + "step": 1722 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810859231301545e-05, + "loss": 0.551, + "step": 1723 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810530905333437e-05, + "loss": 0.4566, + "step": 1724 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810202297370328e-05, + "loss": 0.4481, + "step": 1725 + }, + { + "epoch": 0.72, + "learning_rate": 1.9809873407421656e-05, + "loss": 0.5405, + "step": 1726 + }, + { + "epoch": 0.72, + "learning_rate": 1.980954423549688e-05, + "loss": 0.5531, + "step": 1727 + }, + { + "epoch": 0.72, + "learning_rate": 1.9809214781605458e-05, + "loss": 0.5072, + "step": 1728 + }, + { + "epoch": 0.72, + "learning_rate": 1.980888504575686e-05, + "loss": 0.6357, + "step": 1729 + }, + { + "epoch": 0.72, + "learning_rate": 1.980855502796057e-05, + "loss": 0.523, + "step": 1730 + }, + { + "epoch": 0.72, + "learning_rate": 1.9808224728226067e-05, + "loss": 0.4832, + "step": 1731 + }, + { + "epoch": 0.73, + "learning_rate": 1.9807894146562847e-05, + "loss": 0.4624, + "step": 1732 + }, + { + "epoch": 0.73, + "learning_rate": 1.9807563282980415e-05, + "loss": 0.444, + "step": 1733 + }, + { + "epoch": 0.73, + "learning_rate": 1.980723213748828e-05, + "loss": 0.485, + "step": 1734 + }, + { + "epoch": 0.73, + "learning_rate": 1.9806900710095958e-05, + "loss": 0.6331, + "step": 1735 + }, + { + "epoch": 0.73, + "learning_rate": 1.980656900081298e-05, + "loss": 0.551, + "step": 1736 + }, + { + "epoch": 0.73, + "learning_rate": 1.9806237009648876e-05, + "loss": 0.4904, + "step": 1737 + }, + { + "epoch": 0.73, + "learning_rate": 1.980590473661319e-05, + "loss": 0.5215, + "step": 1738 + }, + { + "epoch": 0.73, + "learning_rate": 1.9805572181715478e-05, + "loss": 0.5491, + "step": 1739 + }, + { + "epoch": 0.73, + "learning_rate": 1.980523934496529e-05, + "loss": 0.5062, + "step": 1740 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804906226372202e-05, + "loss": 0.5369, + "step": 1741 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804572825945784e-05, + "loss": 0.4987, + "step": 1742 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804239143695622e-05, + "loss": 0.4956, + "step": 1743 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803905179631303e-05, + "loss": 0.4876, + "step": 1744 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803570933762432e-05, + "loss": 0.5259, + "step": 1745 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803236406098612e-05, + "loss": 0.4896, + "step": 1746 + }, + { + "epoch": 0.73, + "learning_rate": 1.9802901596649462e-05, + "loss": 0.4446, + "step": 1747 + }, + { + "epoch": 0.73, + "learning_rate": 1.9802566505424603e-05, + "loss": 0.5677, + "step": 1748 + }, + { + "epoch": 0.73, + "learning_rate": 1.980223113243367e-05, + "loss": 0.5355, + "step": 1749 + }, + { + "epoch": 0.73, + "learning_rate": 1.98018954776863e-05, + "loss": 0.4811, + "step": 1750 + }, + { + "epoch": 0.73, + "learning_rate": 1.9801559541192147e-05, + "loss": 0.4993, + "step": 1751 + }, + { + "epoch": 0.73, + "learning_rate": 1.9801223322960858e-05, + "loss": 0.5684, + "step": 1752 + }, + { + "epoch": 0.73, + "learning_rate": 1.9800886823002105e-05, + "loss": 0.5324, + "step": 1753 + }, + { + "epoch": 0.73, + "learning_rate": 1.9800550041325553e-05, + "loss": 0.5637, + "step": 1754 + }, + { + "epoch": 0.73, + "learning_rate": 1.980021297794089e-05, + "loss": 0.5459, + "step": 1755 + }, + { + "epoch": 0.74, + "learning_rate": 1.9799875632857802e-05, + "loss": 0.5847, + "step": 1756 + }, + { + "epoch": 0.74, + "learning_rate": 1.9799538006085982e-05, + "loss": 0.5407, + "step": 1757 + }, + { + "epoch": 0.74, + "learning_rate": 1.979920009763514e-05, + "loss": 0.4979, + "step": 1758 + }, + { + "epoch": 0.74, + "learning_rate": 1.979886190751499e-05, + "loss": 0.438, + "step": 1759 + }, + { + "epoch": 0.74, + "learning_rate": 1.9798523435735246e-05, + "loss": 0.5051, + "step": 1760 + }, + { + "epoch": 0.74, + "learning_rate": 1.9798184682305645e-05, + "loss": 0.4852, + "step": 1761 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797845647235915e-05, + "loss": 0.5432, + "step": 1762 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797506330535808e-05, + "loss": 0.4505, + "step": 1763 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797166732215078e-05, + "loss": 0.6174, + "step": 1764 + }, + { + "epoch": 0.74, + "learning_rate": 1.979682685228348e-05, + "loss": 0.4877, + "step": 1765 + }, + { + "epoch": 0.74, + "learning_rate": 1.9796486690750792e-05, + "loss": 0.515, + "step": 1766 + }, + { + "epoch": 0.74, + "learning_rate": 1.9796146247626783e-05, + "loss": 0.4858, + "step": 1767 + }, + { + "epoch": 0.74, + "learning_rate": 1.9795805522921246e-05, + "loss": 0.5304, + "step": 1768 + }, + { + "epoch": 0.74, + "learning_rate": 1.979546451664397e-05, + "loss": 0.5342, + "step": 1769 + }, + { + "epoch": 0.74, + "learning_rate": 1.979512322880476e-05, + "loss": 0.5117, + "step": 1770 + }, + { + "epoch": 0.74, + "learning_rate": 1.9794781659413425e-05, + "loss": 0.5361, + "step": 1771 + }, + { + "epoch": 0.74, + "learning_rate": 1.9794439808479783e-05, + "loss": 0.4229, + "step": 1772 + }, + { + "epoch": 0.74, + "learning_rate": 1.979409767601366e-05, + "loss": 0.413, + "step": 1773 + }, + { + "epoch": 0.74, + "learning_rate": 1.979375526202489e-05, + "loss": 0.464, + "step": 1774 + }, + { + "epoch": 0.74, + "learning_rate": 1.9793412566523318e-05, + "loss": 0.5194, + "step": 1775 + }, + { + "epoch": 0.74, + "learning_rate": 1.9793069589518787e-05, + "loss": 0.5144, + "step": 1776 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792726331021166e-05, + "loss": 0.451, + "step": 1777 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792382791040317e-05, + "loss": 0.5394, + "step": 1778 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792038969586113e-05, + "loss": 0.5391, + "step": 1779 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791694866668434e-05, + "loss": 0.5101, + "step": 1780 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791350482297177e-05, + "loss": 0.4901, + "step": 1781 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791005816482242e-05, + "loss": 0.4868, + "step": 1782 + }, + { + "epoch": 0.75, + "learning_rate": 1.9790660869233532e-05, + "loss": 0.4358, + "step": 1783 + }, + { + "epoch": 0.75, + "learning_rate": 1.9790315640560962e-05, + "loss": 0.5186, + "step": 1784 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789970130474455e-05, + "loss": 0.5941, + "step": 1785 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789624338983948e-05, + "loss": 0.4653, + "step": 1786 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789278266099375e-05, + "loss": 0.4051, + "step": 1787 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788931911830688e-05, + "loss": 0.4873, + "step": 1788 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788585276187836e-05, + "loss": 0.5416, + "step": 1789 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788238359180786e-05, + "loss": 0.6702, + "step": 1790 + }, + { + "epoch": 0.75, + "learning_rate": 1.978789116081951e-05, + "loss": 0.5041, + "step": 1791 + }, + { + "epoch": 0.75, + "learning_rate": 1.9787543681113993e-05, + "loss": 0.4556, + "step": 1792 + }, + { + "epoch": 0.75, + "learning_rate": 1.9787195920074213e-05, + "loss": 0.5527, + "step": 1793 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786847877710177e-05, + "loss": 0.5213, + "step": 1794 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786499554031876e-05, + "loss": 0.5291, + "step": 1795 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786150949049335e-05, + "loss": 0.4473, + "step": 1796 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785802062772566e-05, + "loss": 0.5462, + "step": 1797 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785452895211606e-05, + "loss": 0.5014, + "step": 1798 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785103446376483e-05, + "loss": 0.5301, + "step": 1799 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784753716277242e-05, + "loss": 0.5775, + "step": 1800 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784403704923937e-05, + "loss": 0.4802, + "step": 1801 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784053412326637e-05, + "loss": 0.4719, + "step": 1802 + }, + { + "epoch": 0.75, + "learning_rate": 1.97837028384954e-05, + "loss": 0.5638, + "step": 1803 + }, + { + "epoch": 0.76, + "learning_rate": 1.9783351983440306e-05, + "loss": 0.5138, + "step": 1804 + }, + { + "epoch": 0.76, + "learning_rate": 1.978300084717144e-05, + "loss": 0.4597, + "step": 1805 + }, + { + "epoch": 0.76, + "learning_rate": 1.9782649429698896e-05, + "loss": 0.4601, + "step": 1806 + }, + { + "epoch": 0.76, + "learning_rate": 1.9782297731032778e-05, + "loss": 0.5567, + "step": 1807 + }, + { + "epoch": 0.76, + "learning_rate": 1.978194575118319e-05, + "loss": 0.5288, + "step": 1808 + }, + { + "epoch": 0.76, + "learning_rate": 1.9781593490160255e-05, + "loss": 0.5387, + "step": 1809 + }, + { + "epoch": 0.76, + "learning_rate": 1.9781240947974093e-05, + "loss": 0.4789, + "step": 1810 + }, + { + "epoch": 0.76, + "learning_rate": 1.978088812463484e-05, + "loss": 0.4638, + "step": 1811 + }, + { + "epoch": 0.76, + "learning_rate": 1.9780535020152632e-05, + "loss": 0.5389, + "step": 1812 + }, + { + "epoch": 0.76, + "learning_rate": 1.978018163453763e-05, + "loss": 0.4942, + "step": 1813 + }, + { + "epoch": 0.76, + "learning_rate": 1.9779827967799985e-05, + "loss": 0.4926, + "step": 1814 + }, + { + "epoch": 0.76, + "learning_rate": 1.9779474019949864e-05, + "loss": 0.5134, + "step": 1815 + }, + { + "epoch": 0.76, + "learning_rate": 1.977911979099744e-05, + "loss": 0.5047, + "step": 1816 + }, + { + "epoch": 0.76, + "learning_rate": 1.9778765280952896e-05, + "loss": 0.492, + "step": 1817 + }, + { + "epoch": 0.76, + "learning_rate": 1.977841048982642e-05, + "loss": 0.5855, + "step": 1818 + }, + { + "epoch": 0.76, + "learning_rate": 1.977805541762821e-05, + "loss": 0.487, + "step": 1819 + }, + { + "epoch": 0.76, + "learning_rate": 1.9777700064368475e-05, + "loss": 0.562, + "step": 1820 + }, + { + "epoch": 0.76, + "learning_rate": 1.977734443005743e-05, + "loss": 0.4709, + "step": 1821 + }, + { + "epoch": 0.76, + "learning_rate": 1.9776988514705294e-05, + "loss": 0.5106, + "step": 1822 + }, + { + "epoch": 0.76, + "learning_rate": 1.97766323183223e-05, + "loss": 0.5815, + "step": 1823 + }, + { + "epoch": 0.76, + "learning_rate": 1.9776275840918685e-05, + "loss": 0.4957, + "step": 1824 + }, + { + "epoch": 0.76, + "learning_rate": 1.97759190825047e-05, + "loss": 0.4592, + "step": 1825 + }, + { + "epoch": 0.76, + "learning_rate": 1.977556204309059e-05, + "loss": 0.5652, + "step": 1826 + }, + { + "epoch": 0.76, + "learning_rate": 1.9775204722686628e-05, + "loss": 0.4936, + "step": 1827 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774847121303078e-05, + "loss": 0.5499, + "step": 1828 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774489238950222e-05, + "loss": 0.5595, + "step": 1829 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774131075638347e-05, + "loss": 0.5254, + "step": 1830 + }, + { + "epoch": 0.77, + "learning_rate": 1.977377263137775e-05, + "loss": 0.5087, + "step": 1831 + }, + { + "epoch": 0.77, + "learning_rate": 1.977341390617873e-05, + "loss": 0.4864, + "step": 1832 + }, + { + "epoch": 0.77, + "learning_rate": 1.97730549000516e-05, + "loss": 0.5294, + "step": 1833 + }, + { + "epoch": 0.77, + "learning_rate": 1.9772695613006676e-05, + "loss": 0.4775, + "step": 1834 + }, + { + "epoch": 0.77, + "learning_rate": 1.977233604505429e-05, + "loss": 0.5719, + "step": 1835 + }, + { + "epoch": 0.77, + "learning_rate": 1.977197619620478e-05, + "loss": 0.5053, + "step": 1836 + }, + { + "epoch": 0.77, + "learning_rate": 1.9771616066468484e-05, + "loss": 0.4699, + "step": 1837 + }, + { + "epoch": 0.77, + "learning_rate": 1.9771255655855753e-05, + "loss": 0.5384, + "step": 1838 + }, + { + "epoch": 0.77, + "learning_rate": 1.9770894964376956e-05, + "loss": 0.5049, + "step": 1839 + }, + { + "epoch": 0.77, + "learning_rate": 1.9770533992042447e-05, + "loss": 0.5714, + "step": 1840 + }, + { + "epoch": 0.77, + "learning_rate": 1.977017273886261e-05, + "loss": 0.5266, + "step": 1841 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769811204847825e-05, + "loss": 0.5479, + "step": 1842 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769449390008492e-05, + "loss": 0.5449, + "step": 1843 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769087294355006e-05, + "loss": 0.4606, + "step": 1844 + }, + { + "epoch": 0.77, + "learning_rate": 1.976872491789777e-05, + "loss": 0.5291, + "step": 1845 + }, + { + "epoch": 0.77, + "learning_rate": 1.9768362260647205e-05, + "loss": 0.5553, + "step": 1846 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767999322613737e-05, + "loss": 0.5144, + "step": 1847 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767636103807796e-05, + "loss": 0.5458, + "step": 1848 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767272604239823e-05, + "loss": 0.559, + "step": 1849 + }, + { + "epoch": 0.77, + "learning_rate": 1.9766908823920268e-05, + "loss": 0.4688, + "step": 1850 + }, + { + "epoch": 0.77, + "learning_rate": 1.976654476285958e-05, + "loss": 0.558, + "step": 1851 + }, + { + "epoch": 0.78, + "learning_rate": 1.9766180421068236e-05, + "loss": 0.5534, + "step": 1852 + }, + { + "epoch": 0.78, + "learning_rate": 1.97658157985567e-05, + "loss": 0.469, + "step": 1853 + }, + { + "epoch": 0.78, + "learning_rate": 1.9765450895335454e-05, + "loss": 0.4969, + "step": 1854 + }, + { + "epoch": 0.78, + "learning_rate": 1.9765085711414987e-05, + "loss": 0.5451, + "step": 1855 + }, + { + "epoch": 0.78, + "learning_rate": 1.97647202468058e-05, + "loss": 0.5496, + "step": 1856 + }, + { + "epoch": 0.78, + "learning_rate": 1.9764354501518394e-05, + "loss": 0.4641, + "step": 1857 + }, + { + "epoch": 0.78, + "learning_rate": 1.976398847556328e-05, + "loss": 0.513, + "step": 1858 + }, + { + "epoch": 0.78, + "learning_rate": 1.9763622168950982e-05, + "loss": 0.6003, + "step": 1859 + }, + { + "epoch": 0.78, + "learning_rate": 1.9763255581692027e-05, + "loss": 0.5382, + "step": 1860 + }, + { + "epoch": 0.78, + "learning_rate": 1.976288871379696e-05, + "loss": 0.5244, + "step": 1861 + }, + { + "epoch": 0.78, + "learning_rate": 1.9762521565276314e-05, + "loss": 0.5568, + "step": 1862 + }, + { + "epoch": 0.78, + "learning_rate": 1.976215413614065e-05, + "loss": 0.509, + "step": 1863 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761786426400532e-05, + "loss": 0.5482, + "step": 1864 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761418436066523e-05, + "loss": 0.5047, + "step": 1865 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761050165149208e-05, + "loss": 0.4403, + "step": 1866 + }, + { + "epoch": 0.78, + "learning_rate": 1.976068161365916e-05, + "loss": 0.509, + "step": 1867 + }, + { + "epoch": 0.78, + "learning_rate": 1.9760312781606985e-05, + "loss": 0.5031, + "step": 1868 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759943669003282e-05, + "loss": 0.5258, + "step": 1869 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759574275858653e-05, + "loss": 0.5036, + "step": 1870 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759204602183725e-05, + "loss": 0.4541, + "step": 1871 + }, + { + "epoch": 0.78, + "learning_rate": 1.9758834647989123e-05, + "loss": 0.4063, + "step": 1872 + }, + { + "epoch": 0.78, + "learning_rate": 1.9758464413285477e-05, + "loss": 0.5494, + "step": 1873 + }, + { + "epoch": 0.78, + "learning_rate": 1.975809389808343e-05, + "loss": 0.5205, + "step": 1874 + }, + { + "epoch": 0.79, + "learning_rate": 1.9757723102393634e-05, + "loss": 0.4654, + "step": 1875 + }, + { + "epoch": 0.79, + "learning_rate": 1.9757352026226746e-05, + "loss": 0.4682, + "step": 1876 + }, + { + "epoch": 0.79, + "learning_rate": 1.9756980669593432e-05, + "loss": 0.5617, + "step": 1877 + }, + { + "epoch": 0.79, + "learning_rate": 1.975660903250437e-05, + "loss": 0.4518, + "step": 1878 + }, + { + "epoch": 0.79, + "learning_rate": 1.9756237114970235e-05, + "loss": 0.4759, + "step": 1879 + }, + { + "epoch": 0.79, + "learning_rate": 1.9755864917001724e-05, + "loss": 0.6428, + "step": 1880 + }, + { + "epoch": 0.79, + "learning_rate": 1.9755492438609532e-05, + "loss": 0.5162, + "step": 1881 + }, + { + "epoch": 0.79, + "learning_rate": 1.975511967980437e-05, + "loss": 0.4909, + "step": 1882 + }, + { + "epoch": 0.79, + "learning_rate": 1.9754746640596945e-05, + "loss": 0.4749, + "step": 1883 + }, + { + "epoch": 0.79, + "learning_rate": 1.9754373320997987e-05, + "loss": 0.4545, + "step": 1884 + }, + { + "epoch": 0.79, + "learning_rate": 1.975399972101822e-05, + "loss": 0.5069, + "step": 1885 + }, + { + "epoch": 0.79, + "learning_rate": 1.975362584066839e-05, + "loss": 0.4973, + "step": 1886 + }, + { + "epoch": 0.79, + "learning_rate": 1.975325167995924e-05, + "loss": 0.4698, + "step": 1887 + }, + { + "epoch": 0.79, + "learning_rate": 1.9752877238901525e-05, + "loss": 0.4487, + "step": 1888 + }, + { + "epoch": 0.79, + "learning_rate": 1.975250251750601e-05, + "loss": 0.5072, + "step": 1889 + }, + { + "epoch": 0.79, + "learning_rate": 1.9752127515783458e-05, + "loss": 0.5311, + "step": 1890 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751752233744662e-05, + "loss": 0.4764, + "step": 1891 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751376671400398e-05, + "loss": 0.4509, + "step": 1892 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751000828761465e-05, + "loss": 0.5794, + "step": 1893 + }, + { + "epoch": 0.79, + "learning_rate": 1.9750624705838666e-05, + "loss": 0.4593, + "step": 1894 + }, + { + "epoch": 0.79, + "learning_rate": 1.9750248302642813e-05, + "loss": 0.472, + "step": 1895 + }, + { + "epoch": 0.79, + "learning_rate": 1.9749871619184726e-05, + "loss": 0.4706, + "step": 1896 + }, + { + "epoch": 0.79, + "learning_rate": 1.974949465547523e-05, + "loss": 0.5228, + "step": 1897 + }, + { + "epoch": 0.79, + "learning_rate": 1.974911741152516e-05, + "loss": 0.5407, + "step": 1898 + }, + { + "epoch": 0.8, + "learning_rate": 1.9748739887345362e-05, + "loss": 0.6125, + "step": 1899 + }, + { + "epoch": 0.8, + "learning_rate": 1.974836208294669e-05, + "loss": 0.4526, + "step": 1900 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747983998339996e-05, + "loss": 0.4489, + "step": 1901 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747605633536152e-05, + "loss": 0.4619, + "step": 1902 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747226988546038e-05, + "loss": 0.5118, + "step": 1903 + }, + { + "epoch": 0.8, + "learning_rate": 1.974684806338053e-05, + "loss": 0.5502, + "step": 1904 + }, + { + "epoch": 0.8, + "learning_rate": 1.9746468858050526e-05, + "loss": 0.5278, + "step": 1905 + }, + { + "epoch": 0.8, + "learning_rate": 1.9746089372566924e-05, + "loss": 0.4562, + "step": 1906 + }, + { + "epoch": 0.8, + "learning_rate": 1.974570960694063e-05, + "loss": 0.5074, + "step": 1907 + }, + { + "epoch": 0.8, + "learning_rate": 1.974532956118256e-05, + "loss": 0.4635, + "step": 1908 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744949235303644e-05, + "loss": 0.5443, + "step": 1909 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744568629314807e-05, + "loss": 0.4998, + "step": 1910 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744187743226994e-05, + "loss": 0.4901, + "step": 1911 + }, + { + "epoch": 0.8, + "learning_rate": 1.974380657705115e-05, + "loss": 0.5269, + "step": 1912 + }, + { + "epoch": 0.8, + "learning_rate": 1.974342513079823e-05, + "loss": 0.4695, + "step": 1913 + }, + { + "epoch": 0.8, + "learning_rate": 1.9743043404479203e-05, + "loss": 0.4963, + "step": 1914 + }, + { + "epoch": 0.8, + "learning_rate": 1.9742661398105038e-05, + "loss": 0.5258, + "step": 1915 + }, + { + "epoch": 0.8, + "learning_rate": 1.9742279111686717e-05, + "loss": 0.4798, + "step": 1916 + }, + { + "epoch": 0.8, + "learning_rate": 1.974189654523523e-05, + "loss": 0.5146, + "step": 1917 + }, + { + "epoch": 0.8, + "learning_rate": 1.974151369876157e-05, + "loss": 0.4735, + "step": 1918 + }, + { + "epoch": 0.8, + "learning_rate": 1.9741130572276742e-05, + "loss": 0.5074, + "step": 1919 + }, + { + "epoch": 0.8, + "learning_rate": 1.9740747165791762e-05, + "loss": 0.5401, + "step": 1920 + }, + { + "epoch": 0.8, + "learning_rate": 1.974036347931765e-05, + "loss": 0.5378, + "step": 1921 + }, + { + "epoch": 0.8, + "learning_rate": 1.973997951286543e-05, + "loss": 0.5635, + "step": 1922 + }, + { + "epoch": 0.81, + "learning_rate": 1.9739595266446143e-05, + "loss": 0.6037, + "step": 1923 + }, + { + "epoch": 0.81, + "learning_rate": 1.9739210740070833e-05, + "loss": 0.4684, + "step": 1924 + }, + { + "epoch": 0.81, + "learning_rate": 1.973882593375055e-05, + "loss": 0.48, + "step": 1925 + }, + { + "epoch": 0.81, + "learning_rate": 1.973844084749636e-05, + "loss": 0.5285, + "step": 1926 + }, + { + "epoch": 0.81, + "learning_rate": 1.973805548131933e-05, + "loss": 0.4726, + "step": 1927 + }, + { + "epoch": 0.81, + "learning_rate": 1.9737669835230537e-05, + "loss": 0.5762, + "step": 1928 + }, + { + "epoch": 0.81, + "learning_rate": 1.9737283909241065e-05, + "loss": 0.5675, + "step": 1929 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736897703362006e-05, + "loss": 0.5074, + "step": 1930 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736511217604465e-05, + "loss": 0.4967, + "step": 1931 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736124451979547e-05, + "loss": 0.5085, + "step": 1932 + }, + { + "epoch": 0.81, + "learning_rate": 1.9735737406498375e-05, + "loss": 0.5721, + "step": 1933 + }, + { + "epoch": 0.81, + "learning_rate": 1.973535008117207e-05, + "loss": 0.5391, + "step": 1934 + }, + { + "epoch": 0.81, + "learning_rate": 1.9734962476011763e-05, + "loss": 0.4891, + "step": 1935 + }, + { + "epoch": 0.81, + "learning_rate": 1.97345745910286e-05, + "loss": 0.4809, + "step": 1936 + }, + { + "epoch": 0.81, + "learning_rate": 1.9734186426233728e-05, + "loss": 0.4708, + "step": 1937 + }, + { + "epoch": 0.81, + "learning_rate": 1.9733797981638306e-05, + "loss": 0.4815, + "step": 1938 + }, + { + "epoch": 0.81, + "learning_rate": 1.97334092572535e-05, + "loss": 0.5476, + "step": 1939 + }, + { + "epoch": 0.81, + "learning_rate": 1.9733020253090477e-05, + "loss": 0.5123, + "step": 1940 + }, + { + "epoch": 0.81, + "learning_rate": 1.9732630969160428e-05, + "loss": 0.5376, + "step": 1941 + }, + { + "epoch": 0.81, + "learning_rate": 1.9732241405474536e-05, + "loss": 0.5071, + "step": 1942 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731851562044002e-05, + "loss": 0.4348, + "step": 1943 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731461438880034e-05, + "loss": 0.4937, + "step": 1944 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731071035993837e-05, + "loss": 0.5239, + "step": 1945 + }, + { + "epoch": 0.81, + "learning_rate": 1.9730680353396643e-05, + "loss": 0.497, + "step": 1946 + }, + { + "epoch": 0.82, + "learning_rate": 1.9730289391099675e-05, + "loss": 0.563, + "step": 1947 + }, + { + "epoch": 0.82, + "learning_rate": 1.9729898149114174e-05, + "loss": 0.6107, + "step": 1948 + }, + { + "epoch": 0.82, + "learning_rate": 1.9729506627451383e-05, + "loss": 0.5381, + "step": 1949 + }, + { + "epoch": 0.82, + "learning_rate": 1.972911482612256e-05, + "loss": 0.5316, + "step": 1950 + }, + { + "epoch": 0.82, + "learning_rate": 1.972872274513896e-05, + "loss": 0.5669, + "step": 1951 + }, + { + "epoch": 0.82, + "learning_rate": 1.972833038451186e-05, + "loss": 0.5279, + "step": 1952 + }, + { + "epoch": 0.82, + "learning_rate": 1.972793774425254e-05, + "loss": 0.485, + "step": 1953 + }, + { + "epoch": 0.82, + "learning_rate": 1.972754482437228e-05, + "loss": 0.4867, + "step": 1954 + }, + { + "epoch": 0.82, + "learning_rate": 1.972715162488237e-05, + "loss": 0.4395, + "step": 1955 + }, + { + "epoch": 0.82, + "learning_rate": 1.9726758145794127e-05, + "loss": 0.5581, + "step": 1956 + }, + { + "epoch": 0.82, + "learning_rate": 1.9726364387118847e-05, + "loss": 0.4973, + "step": 1957 + }, + { + "epoch": 0.82, + "learning_rate": 1.9725970348867858e-05, + "loss": 0.5836, + "step": 1958 + }, + { + "epoch": 0.82, + "learning_rate": 1.972557603105248e-05, + "loss": 0.461, + "step": 1959 + }, + { + "epoch": 0.82, + "learning_rate": 1.9725181433684047e-05, + "loss": 0.5794, + "step": 1960 + }, + { + "epoch": 0.82, + "learning_rate": 1.9724786556773906e-05, + "loss": 0.4638, + "step": 1961 + }, + { + "epoch": 0.82, + "learning_rate": 1.9724391400333405e-05, + "loss": 0.4794, + "step": 1962 + }, + { + "epoch": 0.82, + "learning_rate": 1.97239959643739e-05, + "loss": 0.4787, + "step": 1963 + }, + { + "epoch": 0.82, + "learning_rate": 1.9723600248906766e-05, + "loss": 0.4904, + "step": 1964 + }, + { + "epoch": 0.82, + "learning_rate": 1.9723204253943366e-05, + "loss": 0.482, + "step": 1965 + }, + { + "epoch": 0.82, + "learning_rate": 1.972280797949509e-05, + "loss": 0.5006, + "step": 1966 + }, + { + "epoch": 0.82, + "learning_rate": 1.9722411425573323e-05, + "loss": 0.506, + "step": 1967 + }, + { + "epoch": 0.82, + "learning_rate": 1.9722014592189472e-05, + "loss": 0.5763, + "step": 1968 + }, + { + "epoch": 0.82, + "learning_rate": 1.9721617479354936e-05, + "loss": 0.5273, + "step": 1969 + }, + { + "epoch": 0.82, + "learning_rate": 1.972122008708113e-05, + "loss": 0.4863, + "step": 1970 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720822415379485e-05, + "loss": 0.4972, + "step": 1971 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720424464261426e-05, + "loss": 0.5359, + "step": 1972 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720026233738388e-05, + "loss": 0.4415, + "step": 1973 + }, + { + "epoch": 0.83, + "learning_rate": 1.9719627723821818e-05, + "loss": 0.5304, + "step": 1974 + }, + { + "epoch": 0.83, + "learning_rate": 1.971922893452318e-05, + "loss": 0.489, + "step": 1975 + }, + { + "epoch": 0.83, + "learning_rate": 1.971882986585393e-05, + "loss": 0.49, + "step": 1976 + }, + { + "epoch": 0.83, + "learning_rate": 1.9718430517825538e-05, + "loss": 0.5563, + "step": 1977 + }, + { + "epoch": 0.83, + "learning_rate": 1.9718030890449483e-05, + "loss": 0.5101, + "step": 1978 + }, + { + "epoch": 0.83, + "learning_rate": 1.9717630983737256e-05, + "loss": 0.5219, + "step": 1979 + }, + { + "epoch": 0.83, + "learning_rate": 1.971723079770035e-05, + "loss": 0.4326, + "step": 1980 + }, + { + "epoch": 0.83, + "learning_rate": 1.971683033235027e-05, + "loss": 0.4955, + "step": 1981 + }, + { + "epoch": 0.83, + "learning_rate": 1.971642958769852e-05, + "loss": 0.5168, + "step": 1982 + }, + { + "epoch": 0.83, + "learning_rate": 1.9716028563756628e-05, + "loss": 0.608, + "step": 1983 + }, + { + "epoch": 0.83, + "learning_rate": 1.971562726053611e-05, + "loss": 0.4624, + "step": 1984 + }, + { + "epoch": 0.83, + "learning_rate": 1.9715225678048515e-05, + "loss": 0.4922, + "step": 1985 + }, + { + "epoch": 0.83, + "learning_rate": 1.9714823816305376e-05, + "loss": 0.6225, + "step": 1986 + }, + { + "epoch": 0.83, + "learning_rate": 1.9714421675318247e-05, + "loss": 0.4787, + "step": 1987 + }, + { + "epoch": 0.83, + "learning_rate": 1.971401925509869e-05, + "loss": 0.4666, + "step": 1988 + }, + { + "epoch": 0.83, + "learning_rate": 1.9713616555658266e-05, + "loss": 0.5071, + "step": 1989 + }, + { + "epoch": 0.83, + "learning_rate": 1.9713213577008556e-05, + "loss": 0.4458, + "step": 1990 + }, + { + "epoch": 0.83, + "learning_rate": 1.971281031916114e-05, + "loss": 0.4941, + "step": 1991 + }, + { + "epoch": 0.83, + "learning_rate": 1.9712406782127612e-05, + "loss": 0.4653, + "step": 1992 + }, + { + "epoch": 0.83, + "learning_rate": 1.9712002965919567e-05, + "loss": 0.5734, + "step": 1993 + }, + { + "epoch": 0.83, + "learning_rate": 1.971159887054862e-05, + "loss": 0.535, + "step": 1994 + }, + { + "epoch": 0.84, + "learning_rate": 1.9711194496026372e-05, + "loss": 0.5193, + "step": 1995 + }, + { + "epoch": 0.84, + "learning_rate": 1.9710789842364465e-05, + "loss": 0.5112, + "step": 1996 + }, + { + "epoch": 0.84, + "learning_rate": 1.971038490957452e-05, + "loss": 0.4812, + "step": 1997 + }, + { + "epoch": 0.84, + "learning_rate": 1.9709979697668172e-05, + "loss": 0.4981, + "step": 1998 + }, + { + "epoch": 0.84, + "learning_rate": 1.970957420665708e-05, + "loss": 0.4752, + "step": 1999 + }, + { + "epoch": 0.84, + "learning_rate": 1.9709168436552887e-05, + "loss": 0.4864, + "step": 2000 + }, + { + "epoch": 0.84, + "learning_rate": 1.9708762387367272e-05, + "loss": 0.4548, + "step": 2001 + }, + { + "epoch": 0.84, + "learning_rate": 1.970835605911189e-05, + "loss": 0.511, + "step": 2002 + }, + { + "epoch": 0.84, + "learning_rate": 1.970794945179843e-05, + "loss": 0.5372, + "step": 2003 + }, + { + "epoch": 0.84, + "learning_rate": 1.970754256543858e-05, + "loss": 0.4788, + "step": 2004 + }, + { + "epoch": 0.84, + "learning_rate": 1.970713540004403e-05, + "loss": 0.5105, + "step": 2005 + }, + { + "epoch": 0.84, + "learning_rate": 1.970672795562649e-05, + "loss": 0.4853, + "step": 2006 + }, + { + "epoch": 0.84, + "learning_rate": 1.970632023219767e-05, + "loss": 0.5353, + "step": 2007 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705912229769285e-05, + "loss": 0.5399, + "step": 2008 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705503948353064e-05, + "loss": 0.4605, + "step": 2009 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705095387960746e-05, + "loss": 0.4799, + "step": 2010 + }, + { + "epoch": 0.84, + "learning_rate": 1.9704686548604076e-05, + "loss": 0.4942, + "step": 2011 + }, + { + "epoch": 0.84, + "learning_rate": 1.9704277430294798e-05, + "loss": 0.5017, + "step": 2012 + }, + { + "epoch": 0.84, + "learning_rate": 1.970386803304468e-05, + "loss": 0.5192, + "step": 2013 + }, + { + "epoch": 0.84, + "learning_rate": 1.9703458356865484e-05, + "loss": 0.4614, + "step": 2014 + }, + { + "epoch": 0.84, + "learning_rate": 1.9703048401768993e-05, + "loss": 0.4899, + "step": 2015 + }, + { + "epoch": 0.84, + "learning_rate": 1.970263816776698e-05, + "loss": 0.4431, + "step": 2016 + }, + { + "epoch": 0.84, + "learning_rate": 1.970222765487125e-05, + "loss": 0.5404, + "step": 2017 + }, + { + "epoch": 0.84, + "learning_rate": 1.970181686309359e-05, + "loss": 0.4927, + "step": 2018 + }, + { + "epoch": 0.85, + "learning_rate": 1.9701405792445815e-05, + "loss": 0.4271, + "step": 2019 + }, + { + "epoch": 0.85, + "learning_rate": 1.970099444293974e-05, + "loss": 0.4929, + "step": 2020 + }, + { + "epoch": 0.85, + "learning_rate": 1.970058281458719e-05, + "loss": 0.4559, + "step": 2021 + }, + { + "epoch": 0.85, + "learning_rate": 1.9700170907399992e-05, + "loss": 0.5275, + "step": 2022 + }, + { + "epoch": 0.85, + "learning_rate": 1.969975872138999e-05, + "loss": 0.5423, + "step": 2023 + }, + { + "epoch": 0.85, + "learning_rate": 1.9699346256569033e-05, + "loss": 0.4918, + "step": 2024 + }, + { + "epoch": 0.85, + "learning_rate": 1.9698933512948973e-05, + "loss": 0.4484, + "step": 2025 + }, + { + "epoch": 0.85, + "learning_rate": 1.9698520490541675e-05, + "loss": 0.4909, + "step": 2026 + }, + { + "epoch": 0.85, + "learning_rate": 1.969810718935901e-05, + "loss": 0.4184, + "step": 2027 + }, + { + "epoch": 0.85, + "learning_rate": 1.9697693609412864e-05, + "loss": 0.516, + "step": 2028 + }, + { + "epoch": 0.85, + "learning_rate": 1.9697279750715118e-05, + "loss": 0.6038, + "step": 2029 + }, + { + "epoch": 0.85, + "learning_rate": 1.9696865613277673e-05, + "loss": 0.5819, + "step": 2030 + }, + { + "epoch": 0.85, + "learning_rate": 1.969645119711243e-05, + "loss": 0.512, + "step": 2031 + }, + { + "epoch": 0.85, + "learning_rate": 1.96960365022313e-05, + "loss": 0.4182, + "step": 2032 + }, + { + "epoch": 0.85, + "learning_rate": 1.969562152864621e-05, + "loss": 0.5777, + "step": 2033 + }, + { + "epoch": 0.85, + "learning_rate": 1.9695206276369077e-05, + "loss": 0.481, + "step": 2034 + }, + { + "epoch": 0.85, + "learning_rate": 1.9694790745411846e-05, + "loss": 0.4749, + "step": 2035 + }, + { + "epoch": 0.85, + "learning_rate": 1.9694374935786457e-05, + "loss": 0.4691, + "step": 2036 + }, + { + "epoch": 0.85, + "learning_rate": 1.9693958847504863e-05, + "loss": 0.5651, + "step": 2037 + }, + { + "epoch": 0.85, + "learning_rate": 1.9693542480579025e-05, + "loss": 0.4447, + "step": 2038 + }, + { + "epoch": 0.85, + "learning_rate": 1.969312583502091e-05, + "loss": 0.4686, + "step": 2039 + }, + { + "epoch": 0.85, + "learning_rate": 1.9692708910842494e-05, + "loss": 0.5077, + "step": 2040 + }, + { + "epoch": 0.85, + "learning_rate": 1.9692291708055762e-05, + "loss": 0.5205, + "step": 2041 + }, + { + "epoch": 0.85, + "learning_rate": 1.9691874226672704e-05, + "loss": 0.4826, + "step": 2042 + }, + { + "epoch": 0.86, + "learning_rate": 1.9691456466705324e-05, + "loss": 0.5561, + "step": 2043 + }, + { + "epoch": 0.86, + "learning_rate": 1.969103842816563e-05, + "loss": 0.4831, + "step": 2044 + }, + { + "epoch": 0.86, + "learning_rate": 1.969062011106563e-05, + "loss": 0.546, + "step": 2045 + }, + { + "epoch": 0.86, + "learning_rate": 1.969020151541736e-05, + "loss": 0.5363, + "step": 2046 + }, + { + "epoch": 0.86, + "learning_rate": 1.968978264123284e-05, + "loss": 0.4719, + "step": 2047 + }, + { + "epoch": 0.86, + "learning_rate": 1.968936348852412e-05, + "loss": 0.5133, + "step": 2048 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688944057303247e-05, + "loss": 0.465, + "step": 2049 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688524347582272e-05, + "loss": 0.5911, + "step": 2050 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688104359373264e-05, + "loss": 0.486, + "step": 2051 + }, + { + "epoch": 0.86, + "learning_rate": 1.9687684092688292e-05, + "loss": 0.4789, + "step": 2052 + }, + { + "epoch": 0.86, + "learning_rate": 1.968726354753944e-05, + "loss": 0.4695, + "step": 2053 + }, + { + "epoch": 0.86, + "learning_rate": 1.968684272393879e-05, + "loss": 0.6171, + "step": 2054 + }, + { + "epoch": 0.86, + "learning_rate": 1.9686421621898443e-05, + "loss": 0.4325, + "step": 2055 + }, + { + "epoch": 0.86, + "learning_rate": 1.9686000241430503e-05, + "loss": 0.513, + "step": 2056 + }, + { + "epoch": 0.86, + "learning_rate": 1.968557858254708e-05, + "loss": 0.4462, + "step": 2057 + }, + { + "epoch": 0.86, + "learning_rate": 1.9685156645260296e-05, + "loss": 0.5465, + "step": 2058 + }, + { + "epoch": 0.86, + "learning_rate": 1.968473442958228e-05, + "loss": 0.4689, + "step": 2059 + }, + { + "epoch": 0.86, + "learning_rate": 1.9684311935525163e-05, + "loss": 0.5581, + "step": 2060 + }, + { + "epoch": 0.86, + "learning_rate": 1.9683889163101096e-05, + "loss": 0.4729, + "step": 2061 + }, + { + "epoch": 0.86, + "learning_rate": 1.968346611232223e-05, + "loss": 0.5558, + "step": 2062 + }, + { + "epoch": 0.86, + "learning_rate": 1.968304278320072e-05, + "loss": 0.5112, + "step": 2063 + }, + { + "epoch": 0.86, + "learning_rate": 1.968261917574874e-05, + "loss": 0.5189, + "step": 2064 + }, + { + "epoch": 0.86, + "learning_rate": 1.9682195289978466e-05, + "loss": 0.5116, + "step": 2065 + }, + { + "epoch": 0.86, + "learning_rate": 1.9681771125902075e-05, + "loss": 0.524, + "step": 2066 + }, + { + "epoch": 0.87, + "learning_rate": 1.9681346683531766e-05, + "loss": 0.5492, + "step": 2067 + }, + { + "epoch": 0.87, + "learning_rate": 1.968092196287974e-05, + "loss": 0.513, + "step": 2068 + }, + { + "epoch": 0.87, + "learning_rate": 1.96804969639582e-05, + "loss": 0.561, + "step": 2069 + }, + { + "epoch": 0.87, + "learning_rate": 1.9680071686779368e-05, + "loss": 0.5189, + "step": 2070 + }, + { + "epoch": 0.87, + "learning_rate": 1.9679646131355463e-05, + "loss": 0.5128, + "step": 2071 + }, + { + "epoch": 0.87, + "learning_rate": 1.9679220297698722e-05, + "loss": 0.45, + "step": 2072 + }, + { + "epoch": 0.87, + "learning_rate": 1.967879418582138e-05, + "loss": 0.5464, + "step": 2073 + }, + { + "epoch": 0.87, + "learning_rate": 1.967836779573569e-05, + "loss": 0.4759, + "step": 2074 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677941127453907e-05, + "loss": 0.5496, + "step": 2075 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677514180988292e-05, + "loss": 0.5275, + "step": 2076 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677086956351118e-05, + "loss": 0.5698, + "step": 2077 + }, + { + "epoch": 0.87, + "learning_rate": 1.967665945355467e-05, + "loss": 0.5322, + "step": 2078 + }, + { + "epoch": 0.87, + "learning_rate": 1.9676231672611233e-05, + "loss": 0.4737, + "step": 2079 + }, + { + "epoch": 0.87, + "learning_rate": 1.96758036135331e-05, + "loss": 0.51, + "step": 2080 + }, + { + "epoch": 0.87, + "learning_rate": 1.9675375276332583e-05, + "loss": 0.4943, + "step": 2081 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674946661021984e-05, + "loss": 0.4394, + "step": 2082 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674517767613633e-05, + "loss": 0.5582, + "step": 2083 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674088596119853e-05, + "loss": 0.4703, + "step": 2084 + }, + { + "epoch": 0.87, + "learning_rate": 1.967365914655298e-05, + "loss": 0.5181, + "step": 2085 + }, + { + "epoch": 0.87, + "learning_rate": 1.9673229418925358e-05, + "loss": 0.4687, + "step": 2086 + }, + { + "epoch": 0.87, + "learning_rate": 1.9672799413249343e-05, + "loss": 0.5428, + "step": 2087 + }, + { + "epoch": 0.87, + "learning_rate": 1.967236912953729e-05, + "loss": 0.5192, + "step": 2088 + }, + { + "epoch": 0.87, + "learning_rate": 1.9671938567801568e-05, + "loss": 0.5889, + "step": 2089 + }, + { + "epoch": 0.88, + "learning_rate": 1.9671507728054558e-05, + "loss": 0.4703, + "step": 2090 + }, + { + "epoch": 0.88, + "learning_rate": 1.9671076610308638e-05, + "loss": 0.5094, + "step": 2091 + }, + { + "epoch": 0.88, + "learning_rate": 1.9670645214576204e-05, + "loss": 0.5278, + "step": 2092 + }, + { + "epoch": 0.88, + "learning_rate": 1.9670213540869655e-05, + "loss": 0.6058, + "step": 2093 + }, + { + "epoch": 0.88, + "learning_rate": 1.96697815892014e-05, + "loss": 0.5246, + "step": 2094 + }, + { + "epoch": 0.88, + "learning_rate": 1.9669349359583852e-05, + "loss": 0.5534, + "step": 2095 + }, + { + "epoch": 0.88, + "learning_rate": 1.966891685202944e-05, + "loss": 0.4617, + "step": 2096 + }, + { + "epoch": 0.88, + "learning_rate": 1.9668484066550592e-05, + "loss": 0.519, + "step": 2097 + }, + { + "epoch": 0.88, + "learning_rate": 1.9668051003159746e-05, + "loss": 0.548, + "step": 2098 + }, + { + "epoch": 0.88, + "learning_rate": 1.9667617661869357e-05, + "loss": 0.4658, + "step": 2099 + }, + { + "epoch": 0.88, + "learning_rate": 1.9667184042691877e-05, + "loss": 0.4929, + "step": 2100 + }, + { + "epoch": 0.88, + "learning_rate": 1.966675014563977e-05, + "loss": 0.4961, + "step": 2101 + }, + { + "epoch": 0.88, + "learning_rate": 1.9666315970725507e-05, + "loss": 0.493, + "step": 2102 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665881517961568e-05, + "loss": 0.4827, + "step": 2103 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665446787360444e-05, + "loss": 0.4241, + "step": 2104 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665011778934628e-05, + "loss": 0.4034, + "step": 2105 + }, + { + "epoch": 0.88, + "learning_rate": 1.9664576492696628e-05, + "loss": 0.463, + "step": 2106 + }, + { + "epoch": 0.88, + "learning_rate": 1.9664140928658947e-05, + "loss": 0.5554, + "step": 2107 + }, + { + "epoch": 0.88, + "learning_rate": 1.9663705086834114e-05, + "loss": 0.481, + "step": 2108 + }, + { + "epoch": 0.88, + "learning_rate": 1.9663268967234655e-05, + "loss": 0.4523, + "step": 2109 + }, + { + "epoch": 0.88, + "learning_rate": 1.9662832569873106e-05, + "loss": 0.5055, + "step": 2110 + }, + { + "epoch": 0.88, + "learning_rate": 1.9662395894762007e-05, + "loss": 0.506, + "step": 2111 + }, + { + "epoch": 0.88, + "learning_rate": 1.9661958941913913e-05, + "loss": 0.4366, + "step": 2112 + }, + { + "epoch": 0.88, + "learning_rate": 1.966152171134138e-05, + "loss": 0.4667, + "step": 2113 + }, + { + "epoch": 0.89, + "learning_rate": 1.9661084203056983e-05, + "loss": 0.4995, + "step": 2114 + }, + { + "epoch": 0.89, + "learning_rate": 1.966064641707329e-05, + "loss": 0.5362, + "step": 2115 + }, + { + "epoch": 0.89, + "learning_rate": 1.9660208353402888e-05, + "loss": 0.5667, + "step": 2116 + }, + { + "epoch": 0.89, + "learning_rate": 1.9659770012058374e-05, + "loss": 0.5167, + "step": 2117 + }, + { + "epoch": 0.89, + "learning_rate": 1.9659331393052342e-05, + "loss": 0.5972, + "step": 2118 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658892496397398e-05, + "loss": 0.4921, + "step": 2119 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658453322106165e-05, + "loss": 0.5383, + "step": 2120 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658013870191258e-05, + "loss": 0.55, + "step": 2121 + }, + { + "epoch": 0.89, + "learning_rate": 1.965757414066531e-05, + "loss": 0.4841, + "step": 2122 + }, + { + "epoch": 0.89, + "learning_rate": 1.9657134133540966e-05, + "loss": 0.4447, + "step": 2123 + }, + { + "epoch": 0.89, + "learning_rate": 1.965669384883087e-05, + "loss": 0.5766, + "step": 2124 + }, + { + "epoch": 0.89, + "learning_rate": 1.9656253286547683e-05, + "loss": 0.53, + "step": 2125 + }, + { + "epoch": 0.89, + "learning_rate": 1.965581244670406e-05, + "loss": 0.5233, + "step": 2126 + }, + { + "epoch": 0.89, + "learning_rate": 1.9655371329312677e-05, + "loss": 0.4755, + "step": 2127 + }, + { + "epoch": 0.89, + "learning_rate": 1.9654929934386215e-05, + "loss": 0.4842, + "step": 2128 + }, + { + "epoch": 0.89, + "learning_rate": 1.965448826193736e-05, + "loss": 0.4566, + "step": 2129 + }, + { + "epoch": 0.89, + "learning_rate": 1.9654046311978803e-05, + "loss": 0.4718, + "step": 2130 + }, + { + "epoch": 0.89, + "learning_rate": 1.9653604084523255e-05, + "loss": 0.4792, + "step": 2131 + }, + { + "epoch": 0.89, + "learning_rate": 1.9653161579583423e-05, + "loss": 0.5052, + "step": 2132 + }, + { + "epoch": 0.89, + "learning_rate": 1.965271879717203e-05, + "loss": 0.5531, + "step": 2133 + }, + { + "epoch": 0.89, + "learning_rate": 1.9652275737301797e-05, + "loss": 0.5135, + "step": 2134 + }, + { + "epoch": 0.89, + "learning_rate": 1.9651832399985466e-05, + "loss": 0.5318, + "step": 2135 + }, + { + "epoch": 0.89, + "learning_rate": 1.9651388785235775e-05, + "loss": 0.548, + "step": 2136 + }, + { + "epoch": 0.89, + "learning_rate": 1.965094489306548e-05, + "loss": 0.4697, + "step": 2137 + }, + { + "epoch": 0.9, + "learning_rate": 1.9650500723487335e-05, + "loss": 0.5317, + "step": 2138 + }, + { + "epoch": 0.9, + "learning_rate": 1.9650056276514115e-05, + "loss": 0.4823, + "step": 2139 + }, + { + "epoch": 0.9, + "learning_rate": 1.964961155215859e-05, + "loss": 0.5023, + "step": 2140 + }, + { + "epoch": 0.9, + "learning_rate": 1.964916655043354e-05, + "loss": 0.4548, + "step": 2141 + }, + { + "epoch": 0.9, + "learning_rate": 1.9648721271351763e-05, + "loss": 0.48, + "step": 2142 + }, + { + "epoch": 0.9, + "learning_rate": 1.9648275714926055e-05, + "loss": 0.5174, + "step": 2143 + }, + { + "epoch": 0.9, + "learning_rate": 1.9647829881169223e-05, + "loss": 0.5626, + "step": 2144 + }, + { + "epoch": 0.9, + "learning_rate": 1.9647383770094083e-05, + "loss": 0.4561, + "step": 2145 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646937381713454e-05, + "loss": 0.4395, + "step": 2146 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646490716040174e-05, + "loss": 0.4942, + "step": 2147 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646043773087078e-05, + "loss": 0.4702, + "step": 2148 + }, + { + "epoch": 0.9, + "learning_rate": 1.9645596552867013e-05, + "loss": 0.4292, + "step": 2149 + }, + { + "epoch": 0.9, + "learning_rate": 1.9645149055392834e-05, + "loss": 0.5833, + "step": 2150 + }, + { + "epoch": 0.9, + "learning_rate": 1.9644701280677406e-05, + "loss": 0.4801, + "step": 2151 + }, + { + "epoch": 0.9, + "learning_rate": 1.9644253228733596e-05, + "loss": 0.4909, + "step": 2152 + }, + { + "epoch": 0.9, + "learning_rate": 1.9643804899574287e-05, + "loss": 0.5129, + "step": 2153 + }, + { + "epoch": 0.9, + "learning_rate": 1.9643356293212363e-05, + "loss": 0.5068, + "step": 2154 + }, + { + "epoch": 0.9, + "learning_rate": 1.964290740966072e-05, + "loss": 0.4858, + "step": 2155 + }, + { + "epoch": 0.9, + "learning_rate": 1.9642458248932262e-05, + "loss": 0.5096, + "step": 2156 + }, + { + "epoch": 0.9, + "learning_rate": 1.9642008811039893e-05, + "loss": 0.5049, + "step": 2157 + }, + { + "epoch": 0.9, + "learning_rate": 1.9641559095996544e-05, + "loss": 0.4885, + "step": 2158 + }, + { + "epoch": 0.9, + "learning_rate": 1.964110910381513e-05, + "loss": 0.4504, + "step": 2159 + }, + { + "epoch": 0.9, + "learning_rate": 1.964065883450859e-05, + "loss": 0.5344, + "step": 2160 + }, + { + "epoch": 0.9, + "learning_rate": 1.9640208288089872e-05, + "loss": 0.4163, + "step": 2161 + }, + { + "epoch": 0.91, + "learning_rate": 1.9639757464571914e-05, + "loss": 0.4949, + "step": 2162 + }, + { + "epoch": 0.91, + "learning_rate": 1.9639306363967686e-05, + "loss": 0.4949, + "step": 2163 + }, + { + "epoch": 0.91, + "learning_rate": 1.963885498629015e-05, + "loss": 0.5139, + "step": 2164 + }, + { + "epoch": 0.91, + "learning_rate": 1.9638403331552277e-05, + "loss": 0.4826, + "step": 2165 + }, + { + "epoch": 0.91, + "learning_rate": 1.963795139976706e-05, + "loss": 0.5396, + "step": 2166 + }, + { + "epoch": 0.91, + "learning_rate": 1.9637499190947478e-05, + "loss": 0.507, + "step": 2167 + }, + { + "epoch": 0.91, + "learning_rate": 1.9637046705106534e-05, + "loss": 0.5509, + "step": 2168 + }, + { + "epoch": 0.91, + "learning_rate": 1.9636593942257237e-05, + "loss": 0.4484, + "step": 2169 + }, + { + "epoch": 0.91, + "learning_rate": 1.9636140902412598e-05, + "loss": 0.4527, + "step": 2170 + }, + { + "epoch": 0.91, + "learning_rate": 1.9635687585585637e-05, + "loss": 0.524, + "step": 2171 + }, + { + "epoch": 0.91, + "learning_rate": 1.963523399178939e-05, + "loss": 0.4281, + "step": 2172 + }, + { + "epoch": 0.91, + "learning_rate": 1.9634780121036887e-05, + "loss": 0.3896, + "step": 2173 + }, + { + "epoch": 0.91, + "learning_rate": 1.9634325973341183e-05, + "loss": 0.5746, + "step": 2174 + }, + { + "epoch": 0.91, + "learning_rate": 1.963387154871533e-05, + "loss": 0.5507, + "step": 2175 + }, + { + "epoch": 0.91, + "learning_rate": 1.9633416847172384e-05, + "loss": 0.4694, + "step": 2176 + }, + { + "epoch": 0.91, + "learning_rate": 1.9632961868725424e-05, + "loss": 0.4982, + "step": 2177 + }, + { + "epoch": 0.91, + "learning_rate": 1.963250661338752e-05, + "loss": 0.4752, + "step": 2178 + }, + { + "epoch": 0.91, + "learning_rate": 1.963205108117176e-05, + "loss": 0.6507, + "step": 2179 + }, + { + "epoch": 0.91, + "learning_rate": 1.963159527209124e-05, + "loss": 0.5155, + "step": 2180 + }, + { + "epoch": 0.91, + "learning_rate": 1.9631139186159064e-05, + "loss": 0.6029, + "step": 2181 + }, + { + "epoch": 0.91, + "learning_rate": 1.9630682823388337e-05, + "loss": 0.5069, + "step": 2182 + }, + { + "epoch": 0.91, + "learning_rate": 1.9630226183792175e-05, + "loss": 0.5005, + "step": 2183 + }, + { + "epoch": 0.91, + "learning_rate": 1.962976926738371e-05, + "loss": 0.5772, + "step": 2184 + }, + { + "epoch": 0.91, + "learning_rate": 1.9629312074176073e-05, + "loss": 0.4924, + "step": 2185 + }, + { + "epoch": 0.92, + "learning_rate": 1.9628854604182403e-05, + "loss": 0.5, + "step": 2186 + }, + { + "epoch": 0.92, + "learning_rate": 1.9628396857415856e-05, + "loss": 0.4859, + "step": 2187 + }, + { + "epoch": 0.92, + "learning_rate": 1.962793883388958e-05, + "loss": 0.4494, + "step": 2188 + }, + { + "epoch": 0.92, + "learning_rate": 1.962748053361675e-05, + "loss": 0.4656, + "step": 2189 + }, + { + "epoch": 0.92, + "learning_rate": 1.9627021956610534e-05, + "loss": 0.475, + "step": 2190 + }, + { + "epoch": 0.92, + "learning_rate": 1.9626563102884117e-05, + "loss": 0.4323, + "step": 2191 + }, + { + "epoch": 0.92, + "learning_rate": 1.9626103972450683e-05, + "loss": 0.578, + "step": 2192 + }, + { + "epoch": 0.92, + "learning_rate": 1.962564456532343e-05, + "loss": 0.4718, + "step": 2193 + }, + { + "epoch": 0.92, + "learning_rate": 1.962518488151557e-05, + "loss": 0.4104, + "step": 2194 + }, + { + "epoch": 0.92, + "learning_rate": 1.962472492104031e-05, + "loss": 0.548, + "step": 2195 + }, + { + "epoch": 0.92, + "learning_rate": 1.9624264683910877e-05, + "loss": 0.555, + "step": 2196 + }, + { + "epoch": 0.92, + "learning_rate": 1.9623804170140494e-05, + "loss": 0.4559, + "step": 2197 + }, + { + "epoch": 0.92, + "learning_rate": 1.9623343379742397e-05, + "loss": 0.5425, + "step": 2198 + }, + { + "epoch": 0.92, + "learning_rate": 1.9622882312729835e-05, + "loss": 0.5496, + "step": 2199 + }, + { + "epoch": 0.92, + "learning_rate": 1.9622420969116062e-05, + "loss": 0.5709, + "step": 2200 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621959348914335e-05, + "loss": 0.5725, + "step": 2201 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621497452137927e-05, + "loss": 0.4704, + "step": 2202 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621035278800115e-05, + "loss": 0.5526, + "step": 2203 + }, + { + "epoch": 0.92, + "learning_rate": 1.9620572828914177e-05, + "loss": 0.5194, + "step": 2204 + }, + { + "epoch": 0.92, + "learning_rate": 1.9620110102493412e-05, + "loss": 0.551, + "step": 2205 + }, + { + "epoch": 0.92, + "learning_rate": 1.9619647099551118e-05, + "loss": 0.53, + "step": 2206 + }, + { + "epoch": 0.92, + "learning_rate": 1.9619183820100603e-05, + "loss": 0.4799, + "step": 2207 + }, + { + "epoch": 0.92, + "learning_rate": 1.9618720264155187e-05, + "loss": 0.4713, + "step": 2208 + }, + { + "epoch": 0.92, + "learning_rate": 1.961825643172819e-05, + "loss": 0.4345, + "step": 2209 + }, + { + "epoch": 0.93, + "learning_rate": 1.9617792322832952e-05, + "loss": 0.5059, + "step": 2210 + }, + { + "epoch": 0.93, + "learning_rate": 1.9617327937482805e-05, + "loss": 0.5279, + "step": 2211 + }, + { + "epoch": 0.93, + "learning_rate": 1.9616863275691102e-05, + "loss": 0.5578, + "step": 2212 + }, + { + "epoch": 0.93, + "learning_rate": 1.9616398337471196e-05, + "loss": 0.5448, + "step": 2213 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615933122836454e-05, + "loss": 0.5377, + "step": 2214 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615467631800248e-05, + "loss": 0.4938, + "step": 2215 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615001864375957e-05, + "loss": 0.4626, + "step": 2216 + }, + { + "epoch": 0.93, + "learning_rate": 1.9614535820576973e-05, + "loss": 0.5208, + "step": 2217 + }, + { + "epoch": 0.93, + "learning_rate": 1.9614069500416685e-05, + "loss": 0.494, + "step": 2218 + }, + { + "epoch": 0.93, + "learning_rate": 1.9613602903908503e-05, + "loss": 0.5347, + "step": 2219 + }, + { + "epoch": 0.93, + "learning_rate": 1.961313603106583e-05, + "loss": 0.6241, + "step": 2220 + }, + { + "epoch": 0.93, + "learning_rate": 1.96126688819021e-05, + "loss": 0.5039, + "step": 2221 + }, + { + "epoch": 0.93, + "learning_rate": 1.9612201456430732e-05, + "loss": 0.4839, + "step": 2222 + }, + { + "epoch": 0.93, + "learning_rate": 1.9611733754665157e-05, + "loss": 0.5787, + "step": 2223 + }, + { + "epoch": 0.93, + "learning_rate": 1.961126577661883e-05, + "loss": 0.5715, + "step": 2224 + }, + { + "epoch": 0.93, + "learning_rate": 1.9610797522305195e-05, + "loss": 0.4819, + "step": 2225 + }, + { + "epoch": 0.93, + "learning_rate": 1.9610328991737716e-05, + "loss": 0.4741, + "step": 2226 + }, + { + "epoch": 0.93, + "learning_rate": 1.9609860184929858e-05, + "loss": 0.4819, + "step": 2227 + }, + { + "epoch": 0.93, + "learning_rate": 1.9609391101895093e-05, + "loss": 0.5646, + "step": 2228 + }, + { + "epoch": 0.93, + "learning_rate": 1.9608921742646913e-05, + "loss": 0.4428, + "step": 2229 + }, + { + "epoch": 0.93, + "learning_rate": 1.9608452107198798e-05, + "loss": 0.4098, + "step": 2230 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607982195564258e-05, + "loss": 0.4311, + "step": 2231 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607512007756793e-05, + "loss": 0.503, + "step": 2232 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607041543789927e-05, + "loss": 0.5069, + "step": 2233 + }, + { + "epoch": 0.94, + "learning_rate": 1.960657080367717e-05, + "loss": 0.4942, + "step": 2234 + }, + { + "epoch": 0.94, + "learning_rate": 1.9606099787432066e-05, + "loss": 0.5272, + "step": 2235 + }, + { + "epoch": 0.94, + "learning_rate": 1.9605628495068143e-05, + "loss": 0.5224, + "step": 2236 + }, + { + "epoch": 0.94, + "learning_rate": 1.9605156926598953e-05, + "loss": 0.6102, + "step": 2237 + }, + { + "epoch": 0.94, + "learning_rate": 1.9604685082038053e-05, + "loss": 0.5718, + "step": 2238 + }, + { + "epoch": 0.94, + "learning_rate": 1.9604212961399004e-05, + "loss": 0.4199, + "step": 2239 + }, + { + "epoch": 0.94, + "learning_rate": 1.960374056469537e-05, + "loss": 0.5303, + "step": 2240 + }, + { + "epoch": 0.94, + "learning_rate": 1.9603267891940742e-05, + "loss": 0.4223, + "step": 2241 + }, + { + "epoch": 0.94, + "learning_rate": 1.9602794943148698e-05, + "loss": 0.6121, + "step": 2242 + }, + { + "epoch": 0.94, + "learning_rate": 1.9602321718332838e-05, + "loss": 0.5694, + "step": 2243 + }, + { + "epoch": 0.94, + "learning_rate": 1.9601848217506755e-05, + "loss": 0.5002, + "step": 2244 + }, + { + "epoch": 0.94, + "learning_rate": 1.960137444068407e-05, + "loss": 0.5088, + "step": 2245 + }, + { + "epoch": 0.94, + "learning_rate": 1.9600900387878397e-05, + "loss": 0.555, + "step": 2246 + }, + { + "epoch": 0.94, + "learning_rate": 1.9600426059103358e-05, + "loss": 0.5601, + "step": 2247 + }, + { + "epoch": 0.94, + "learning_rate": 1.9599951454372594e-05, + "loss": 0.4603, + "step": 2248 + }, + { + "epoch": 0.94, + "learning_rate": 1.9599476573699747e-05, + "loss": 0.5543, + "step": 2249 + }, + { + "epoch": 0.94, + "learning_rate": 1.959900141709846e-05, + "loss": 0.5286, + "step": 2250 + }, + { + "epoch": 0.94, + "learning_rate": 1.9598525984582398e-05, + "loss": 0.494, + "step": 2251 + }, + { + "epoch": 0.94, + "learning_rate": 1.959805027616522e-05, + "loss": 0.4904, + "step": 2252 + }, + { + "epoch": 0.94, + "learning_rate": 1.9597574291860607e-05, + "loss": 0.5076, + "step": 2253 + }, + { + "epoch": 0.94, + "learning_rate": 1.9597098031682238e-05, + "loss": 0.5447, + "step": 2254 + }, + { + "epoch": 0.94, + "learning_rate": 1.9596621495643805e-05, + "loss": 0.4445, + "step": 2255 + }, + { + "epoch": 0.94, + "learning_rate": 1.9596144683759e-05, + "loss": 0.5893, + "step": 2256 + }, + { + "epoch": 0.94, + "learning_rate": 1.959566759604153e-05, + "loss": 0.479, + "step": 2257 + }, + { + "epoch": 0.95, + "learning_rate": 1.959519023250511e-05, + "loss": 0.5299, + "step": 2258 + }, + { + "epoch": 0.95, + "learning_rate": 1.959471259316346e-05, + "loss": 0.4844, + "step": 2259 + }, + { + "epoch": 0.95, + "learning_rate": 1.9594234678030317e-05, + "loss": 0.4575, + "step": 2260 + }, + { + "epoch": 0.95, + "learning_rate": 1.959375648711941e-05, + "loss": 0.538, + "step": 2261 + }, + { + "epoch": 0.95, + "learning_rate": 1.9593278020444483e-05, + "loss": 0.5077, + "step": 2262 + }, + { + "epoch": 0.95, + "learning_rate": 1.9592799278019294e-05, + "loss": 0.5273, + "step": 2263 + }, + { + "epoch": 0.95, + "learning_rate": 1.9592320259857603e-05, + "loss": 0.4944, + "step": 2264 + }, + { + "epoch": 0.95, + "learning_rate": 1.9591840965973175e-05, + "loss": 0.525, + "step": 2265 + }, + { + "epoch": 0.95, + "learning_rate": 1.9591361396379795e-05, + "loss": 0.5158, + "step": 2266 + }, + { + "epoch": 0.95, + "learning_rate": 1.959088155109124e-05, + "loss": 0.5111, + "step": 2267 + }, + { + "epoch": 0.95, + "learning_rate": 1.9590401430121305e-05, + "loss": 0.4892, + "step": 2268 + }, + { + "epoch": 0.95, + "learning_rate": 1.9589921033483793e-05, + "loss": 0.5192, + "step": 2269 + }, + { + "epoch": 0.95, + "learning_rate": 1.958944036119251e-05, + "loss": 0.4518, + "step": 2270 + }, + { + "epoch": 0.95, + "learning_rate": 1.9588959413261272e-05, + "loss": 0.4856, + "step": 2271 + }, + { + "epoch": 0.95, + "learning_rate": 1.9588478189703908e-05, + "loss": 0.4939, + "step": 2272 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587996690534244e-05, + "loss": 0.5192, + "step": 2273 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587514915766124e-05, + "loss": 0.5328, + "step": 2274 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587032865413394e-05, + "loss": 0.4875, + "step": 2275 + }, + { + "epoch": 0.95, + "learning_rate": 1.9586550539489914e-05, + "loss": 0.4893, + "step": 2276 + }, + { + "epoch": 0.95, + "learning_rate": 1.9586067938009546e-05, + "loss": 0.4676, + "step": 2277 + }, + { + "epoch": 0.95, + "learning_rate": 1.9585585060986162e-05, + "loss": 0.4876, + "step": 2278 + }, + { + "epoch": 0.95, + "learning_rate": 1.958510190843364e-05, + "loss": 0.4743, + "step": 2279 + }, + { + "epoch": 0.95, + "learning_rate": 1.958461848036587e-05, + "loss": 0.5614, + "step": 2280 + }, + { + "epoch": 0.95, + "learning_rate": 1.958413477679675e-05, + "loss": 0.4286, + "step": 2281 + }, + { + "epoch": 0.96, + "learning_rate": 1.9583650797740173e-05, + "loss": 0.4811, + "step": 2282 + }, + { + "epoch": 0.96, + "learning_rate": 1.9583166543210065e-05, + "loss": 0.5306, + "step": 2283 + }, + { + "epoch": 0.96, + "learning_rate": 1.9582682013220337e-05, + "loss": 0.4897, + "step": 2284 + }, + { + "epoch": 0.96, + "learning_rate": 1.958219720778492e-05, + "loss": 0.4841, + "step": 2285 + }, + { + "epoch": 0.96, + "learning_rate": 1.9581712126917742e-05, + "loss": 0.5035, + "step": 2286 + }, + { + "epoch": 0.96, + "learning_rate": 1.9581226770632753e-05, + "loss": 0.4354, + "step": 2287 + }, + { + "epoch": 0.96, + "learning_rate": 1.9580741138943906e-05, + "loss": 0.461, + "step": 2288 + }, + { + "epoch": 0.96, + "learning_rate": 1.9580255231865156e-05, + "loss": 0.4559, + "step": 2289 + }, + { + "epoch": 0.96, + "learning_rate": 1.9579769049410468e-05, + "loss": 0.5464, + "step": 2290 + }, + { + "epoch": 0.96, + "learning_rate": 1.957928259159382e-05, + "loss": 0.4297, + "step": 2291 + }, + { + "epoch": 0.96, + "learning_rate": 1.95787958584292e-05, + "loss": 0.5206, + "step": 2292 + }, + { + "epoch": 0.96, + "learning_rate": 1.9578308849930586e-05, + "loss": 0.5657, + "step": 2293 + }, + { + "epoch": 0.96, + "learning_rate": 1.957782156611199e-05, + "loss": 0.4817, + "step": 2294 + }, + { + "epoch": 0.96, + "learning_rate": 1.9577334006987405e-05, + "loss": 0.5052, + "step": 2295 + }, + { + "epoch": 0.96, + "learning_rate": 1.9576846172570855e-05, + "loss": 0.5689, + "step": 2296 + }, + { + "epoch": 0.96, + "learning_rate": 1.9576358062876363e-05, + "loss": 0.5839, + "step": 2297 + }, + { + "epoch": 0.96, + "learning_rate": 1.9575869677917956e-05, + "loss": 0.4988, + "step": 2298 + }, + { + "epoch": 0.96, + "learning_rate": 1.957538101770967e-05, + "loss": 0.4993, + "step": 2299 + }, + { + "epoch": 0.96, + "learning_rate": 1.9574892082265557e-05, + "loss": 0.4793, + "step": 2300 + }, + { + "epoch": 0.96, + "learning_rate": 1.9574402871599666e-05, + "loss": 0.5144, + "step": 2301 + }, + { + "epoch": 0.96, + "learning_rate": 1.957391338572606e-05, + "loss": 0.5921, + "step": 2302 + }, + { + "epoch": 0.96, + "learning_rate": 1.957342362465881e-05, + "loss": 0.5145, + "step": 2303 + }, + { + "epoch": 0.96, + "learning_rate": 1.9572933588411997e-05, + "loss": 0.4551, + "step": 2304 + }, + { + "epoch": 0.97, + "learning_rate": 1.9572443276999696e-05, + "loss": 0.5018, + "step": 2305 + }, + { + "epoch": 0.97, + "learning_rate": 1.957195269043601e-05, + "loss": 0.5091, + "step": 2306 + }, + { + "epoch": 0.97, + "learning_rate": 1.957146182873504e-05, + "loss": 0.5403, + "step": 2307 + }, + { + "epoch": 0.97, + "learning_rate": 1.957097069191089e-05, + "loss": 0.5032, + "step": 2308 + }, + { + "epoch": 0.97, + "learning_rate": 1.9570479279977682e-05, + "loss": 0.5441, + "step": 2309 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569987592949543e-05, + "loss": 0.5361, + "step": 2310 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569495630840597e-05, + "loss": 0.4916, + "step": 2311 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569003393664996e-05, + "loss": 0.4256, + "step": 2312 + }, + { + "epoch": 0.97, + "learning_rate": 1.9568510881436883e-05, + "loss": 0.484, + "step": 2313 + }, + { + "epoch": 0.97, + "learning_rate": 1.9568018094170415e-05, + "loss": 0.4544, + "step": 2314 + }, + { + "epoch": 0.97, + "learning_rate": 1.9567525031879756e-05, + "loss": 0.4739, + "step": 2315 + }, + { + "epoch": 0.97, + "learning_rate": 1.956703169457908e-05, + "loss": 0.4568, + "step": 2316 + }, + { + "epoch": 0.97, + "learning_rate": 1.956653808228257e-05, + "loss": 0.5526, + "step": 2317 + }, + { + "epoch": 0.97, + "learning_rate": 1.956604419500441e-05, + "loss": 0.5471, + "step": 2318 + }, + { + "epoch": 0.97, + "learning_rate": 1.95655500327588e-05, + "loss": 0.4968, + "step": 2319 + }, + { + "epoch": 0.97, + "learning_rate": 1.956505559555994e-05, + "loss": 0.5308, + "step": 2320 + }, + { + "epoch": 0.97, + "learning_rate": 1.9564560883422045e-05, + "loss": 0.6437, + "step": 2321 + }, + { + "epoch": 0.97, + "learning_rate": 1.9564065896359337e-05, + "loss": 0.4779, + "step": 2322 + }, + { + "epoch": 0.97, + "learning_rate": 1.9563570634386042e-05, + "loss": 0.4739, + "step": 2323 + }, + { + "epoch": 0.97, + "learning_rate": 1.9563075097516397e-05, + "loss": 0.4708, + "step": 2324 + }, + { + "epoch": 0.97, + "learning_rate": 1.9562579285764638e-05, + "loss": 0.4772, + "step": 2325 + }, + { + "epoch": 0.97, + "learning_rate": 1.956208319914503e-05, + "loss": 0.5105, + "step": 2326 + }, + { + "epoch": 0.97, + "learning_rate": 1.956158683767182e-05, + "loss": 0.4218, + "step": 2327 + }, + { + "epoch": 0.97, + "learning_rate": 1.9561090201359285e-05, + "loss": 0.5393, + "step": 2328 + }, + { + "epoch": 0.98, + "learning_rate": 1.9560593290221695e-05, + "loss": 0.439, + "step": 2329 + }, + { + "epoch": 0.98, + "learning_rate": 1.9560096104273335e-05, + "loss": 0.4925, + "step": 2330 + }, + { + "epoch": 0.98, + "learning_rate": 1.95595986435285e-05, + "loss": 0.5582, + "step": 2331 + }, + { + "epoch": 0.98, + "learning_rate": 1.9559100908001478e-05, + "loss": 0.4094, + "step": 2332 + }, + { + "epoch": 0.98, + "learning_rate": 1.9558602897706586e-05, + "loss": 0.5477, + "step": 2333 + }, + { + "epoch": 0.98, + "learning_rate": 1.9558104612658137e-05, + "loss": 0.5561, + "step": 2334 + }, + { + "epoch": 0.98, + "learning_rate": 1.9557606052870452e-05, + "loss": 0.4986, + "step": 2335 + }, + { + "epoch": 0.98, + "learning_rate": 1.9557107218357862e-05, + "loss": 0.4391, + "step": 2336 + }, + { + "epoch": 0.98, + "learning_rate": 1.9556608109134707e-05, + "loss": 0.432, + "step": 2337 + }, + { + "epoch": 0.98, + "learning_rate": 1.955610872521533e-05, + "loss": 0.4751, + "step": 2338 + }, + { + "epoch": 0.98, + "learning_rate": 1.955560906661409e-05, + "loss": 0.5742, + "step": 2339 + }, + { + "epoch": 0.98, + "learning_rate": 1.9555109133345347e-05, + "loss": 0.5143, + "step": 2340 + }, + { + "epoch": 0.98, + "learning_rate": 1.9554608925423467e-05, + "loss": 0.4496, + "step": 2341 + }, + { + "epoch": 0.98, + "learning_rate": 1.9554108442862836e-05, + "loss": 0.4794, + "step": 2342 + }, + { + "epoch": 0.98, + "learning_rate": 1.9553607685677834e-05, + "loss": 0.4721, + "step": 2343 + }, + { + "epoch": 0.98, + "learning_rate": 1.955310665388286e-05, + "loss": 0.5335, + "step": 2344 + }, + { + "epoch": 0.98, + "learning_rate": 1.955260534749231e-05, + "loss": 0.4888, + "step": 2345 + }, + { + "epoch": 0.98, + "learning_rate": 1.9552103766520596e-05, + "loss": 0.452, + "step": 2346 + }, + { + "epoch": 0.98, + "learning_rate": 1.9551601910982136e-05, + "loss": 0.5632, + "step": 2347 + }, + { + "epoch": 0.98, + "learning_rate": 1.9551099780891353e-05, + "loss": 0.4978, + "step": 2348 + }, + { + "epoch": 0.98, + "learning_rate": 1.9550597376262685e-05, + "loss": 0.5115, + "step": 2349 + }, + { + "epoch": 0.98, + "learning_rate": 1.955009469711057e-05, + "loss": 0.5622, + "step": 2350 + }, + { + "epoch": 0.98, + "learning_rate": 1.954959174344946e-05, + "loss": 0.5278, + "step": 2351 + }, + { + "epoch": 0.98, + "learning_rate": 1.9549088515293804e-05, + "loss": 0.4838, + "step": 2352 + }, + { + "epoch": 0.99, + "learning_rate": 1.9548585012658074e-05, + "loss": 0.5064, + "step": 2353 + }, + { + "epoch": 0.99, + "learning_rate": 1.954808123555674e-05, + "loss": 0.5339, + "step": 2354 + }, + { + "epoch": 0.99, + "learning_rate": 1.954757718400429e-05, + "loss": 0.4907, + "step": 2355 + }, + { + "epoch": 0.99, + "learning_rate": 1.95470728580152e-05, + "loss": 0.485, + "step": 2356 + }, + { + "epoch": 0.99, + "learning_rate": 1.9546568257603976e-05, + "loss": 0.4973, + "step": 2357 + }, + { + "epoch": 0.99, + "learning_rate": 1.9546063382785117e-05, + "loss": 0.5277, + "step": 2358 + }, + { + "epoch": 0.99, + "learning_rate": 1.9545558233573136e-05, + "loss": 0.4577, + "step": 2359 + }, + { + "epoch": 0.99, + "learning_rate": 1.9545052809982556e-05, + "loss": 0.4347, + "step": 2360 + }, + { + "epoch": 0.99, + "learning_rate": 1.9544547112027904e-05, + "loss": 0.4502, + "step": 2361 + }, + { + "epoch": 0.99, + "learning_rate": 1.9544041139723712e-05, + "loss": 0.446, + "step": 2362 + }, + { + "epoch": 0.99, + "learning_rate": 1.954353489308453e-05, + "loss": 0.618, + "step": 2363 + }, + { + "epoch": 0.99, + "learning_rate": 1.9543028372124903e-05, + "loss": 0.4332, + "step": 2364 + }, + { + "epoch": 0.99, + "learning_rate": 1.9542521576859397e-05, + "loss": 0.4302, + "step": 2365 + }, + { + "epoch": 0.99, + "learning_rate": 1.9542014507302576e-05, + "loss": 0.5073, + "step": 2366 + }, + { + "epoch": 0.99, + "learning_rate": 1.9541507163469012e-05, + "loss": 0.4632, + "step": 2367 + }, + { + "epoch": 0.99, + "learning_rate": 1.9540999545373295e-05, + "loss": 0.5273, + "step": 2368 + }, + { + "epoch": 0.99, + "learning_rate": 1.954049165303001e-05, + "loss": 0.4725, + "step": 2369 + }, + { + "epoch": 0.99, + "learning_rate": 1.9539983486453762e-05, + "loss": 0.4652, + "step": 2370 + }, + { + "epoch": 0.99, + "learning_rate": 1.953947504565915e-05, + "loss": 0.4843, + "step": 2371 + }, + { + "epoch": 0.99, + "learning_rate": 1.9538966330660796e-05, + "loss": 0.511, + "step": 2372 + }, + { + "epoch": 0.99, + "learning_rate": 1.9538457341473318e-05, + "loss": 0.5827, + "step": 2373 + }, + { + "epoch": 0.99, + "learning_rate": 1.9537948078111348e-05, + "loss": 0.4844, + "step": 2374 + }, + { + "epoch": 0.99, + "learning_rate": 1.9537438540589526e-05, + "loss": 0.5425, + "step": 2375 + }, + { + "epoch": 0.99, + "learning_rate": 1.9536928728922496e-05, + "loss": 0.4983, + "step": 2376 + }, + { + "epoch": 1.0, + "learning_rate": 1.953641864312491e-05, + "loss": 0.5085, + "step": 2377 + }, + { + "epoch": 1.0, + "learning_rate": 1.9535908283211435e-05, + "loss": 0.5282, + "step": 2378 + }, + { + "epoch": 1.0, + "learning_rate": 1.9535397649196738e-05, + "loss": 0.4554, + "step": 2379 + }, + { + "epoch": 1.0, + "learning_rate": 1.95348867410955e-05, + "loss": 0.4774, + "step": 2380 + }, + { + "epoch": 1.0, + "learning_rate": 1.95343755589224e-05, + "loss": 0.6091, + "step": 2381 + }, + { + "epoch": 1.0, + "learning_rate": 1.9533864102692134e-05, + "loss": 0.6011, + "step": 2382 + }, + { + "epoch": 1.0, + "learning_rate": 1.9533352372419408e-05, + "loss": 0.4555, + "step": 2383 + }, + { + "epoch": 1.0, + "learning_rate": 1.953284036811893e-05, + "loss": 0.435, + "step": 2384 + }, + { + "epoch": 1.0, + "learning_rate": 1.9532328089805407e-05, + "loss": 0.5451, + "step": 2385 + }, + { + "epoch": 1.0, + "learning_rate": 1.953181553749358e-05, + "loss": 0.4929, + "step": 2386 + }, + { + "epoch": 1.0, + "learning_rate": 1.953130271119817e-05, + "loss": 0.5192, + "step": 2387 + }, + { + "epoch": 1.0, + "learning_rate": 1.953078961093392e-05, + "loss": 0.5065, + "step": 2388 + }, + { + "epoch": 1.0, + "learning_rate": 1.953027623671558e-05, + "loss": 0.4521, + "step": 2389 + }, + { + "epoch": 1.0, + "learning_rate": 1.952976258855791e-05, + "loss": 0.3904, + "step": 2390 + }, + { + "epoch": 1.0, + "learning_rate": 1.9529248666475668e-05, + "loss": 0.3474, + "step": 2391 + }, + { + "epoch": 1.0, + "learning_rate": 1.9528734470483632e-05, + "loss": 0.3506, + "step": 2392 + }, + { + "epoch": 1.0, + "learning_rate": 1.9528220000596577e-05, + "loss": 0.399, + "step": 2393 + }, + { + "epoch": 1.0, + "learning_rate": 1.9527705256829293e-05, + "loss": 0.3954, + "step": 2394 + }, + { + "epoch": 1.0, + "learning_rate": 1.952719023919658e-05, + "loss": 0.3772, + "step": 2395 + }, + { + "epoch": 1.0, + "learning_rate": 1.9526674947713234e-05, + "loss": 0.4016, + "step": 2396 + }, + { + "epoch": 1.0, + "learning_rate": 1.9526159382394068e-05, + "loss": 0.3883, + "step": 2397 + }, + { + "epoch": 1.0, + "learning_rate": 1.952564354325391e-05, + "loss": 0.3803, + "step": 2398 + }, + { + "epoch": 1.0, + "learning_rate": 1.952512743030758e-05, + "loss": 0.3478, + "step": 2399 + }, + { + "epoch": 1.0, + "learning_rate": 1.952461104356991e-05, + "loss": 0.4181, + "step": 2400 + }, + { + "epoch": 1.01, + "learning_rate": 1.952409438305575e-05, + "loss": 0.3764, + "step": 2401 + }, + { + "epoch": 1.01, + "learning_rate": 1.952357744877995e-05, + "loss": 0.412, + "step": 2402 + }, + { + "epoch": 1.01, + "learning_rate": 1.952306024075737e-05, + "loss": 0.3514, + "step": 2403 + }, + { + "epoch": 1.01, + "learning_rate": 1.9522542759002866e-05, + "loss": 0.3896, + "step": 2404 + }, + { + "epoch": 1.01, + "learning_rate": 1.9522025003531327e-05, + "loss": 0.3816, + "step": 2405 + }, + { + "epoch": 1.01, + "learning_rate": 1.9521506974357628e-05, + "loss": 0.4014, + "step": 2406 + }, + { + "epoch": 1.01, + "learning_rate": 1.952098867149666e-05, + "loss": 0.3944, + "step": 2407 + }, + { + "epoch": 1.01, + "learning_rate": 1.952047009496332e-05, + "loss": 0.376, + "step": 2408 + }, + { + "epoch": 1.01, + "learning_rate": 1.9519951244772514e-05, + "loss": 0.363, + "step": 2409 + }, + { + "epoch": 1.01, + "learning_rate": 1.951943212093916e-05, + "loss": 0.4244, + "step": 2410 + }, + { + "epoch": 1.01, + "learning_rate": 1.9518912723478176e-05, + "loss": 0.4064, + "step": 2411 + }, + { + "epoch": 1.01, + "learning_rate": 1.9518393052404495e-05, + "loss": 0.3723, + "step": 2412 + }, + { + "epoch": 1.01, + "learning_rate": 1.951787310773305e-05, + "loss": 0.3509, + "step": 2413 + }, + { + "epoch": 1.01, + "learning_rate": 1.9517352889478787e-05, + "loss": 0.3573, + "step": 2414 + }, + { + "epoch": 1.01, + "learning_rate": 1.9516832397656663e-05, + "loss": 0.3915, + "step": 2415 + }, + { + "epoch": 1.01, + "learning_rate": 1.9516311632281636e-05, + "loss": 0.3676, + "step": 2416 + }, + { + "epoch": 1.01, + "learning_rate": 1.9515790593368677e-05, + "loss": 0.4132, + "step": 2417 + }, + { + "epoch": 1.01, + "learning_rate": 1.951526928093276e-05, + "loss": 0.4165, + "step": 2418 + }, + { + "epoch": 1.01, + "learning_rate": 1.9514747694988873e-05, + "loss": 0.3968, + "step": 2419 + }, + { + "epoch": 1.01, + "learning_rate": 1.9514225835552004e-05, + "loss": 0.3525, + "step": 2420 + }, + { + "epoch": 1.01, + "learning_rate": 1.9513703702637158e-05, + "loss": 0.3924, + "step": 2421 + }, + { + "epoch": 1.01, + "learning_rate": 1.951318129625934e-05, + "loss": 0.3927, + "step": 2422 + }, + { + "epoch": 1.01, + "learning_rate": 1.951265861643357e-05, + "loss": 0.4062, + "step": 2423 + }, + { + "epoch": 1.01, + "learning_rate": 1.951213566317487e-05, + "loss": 0.3723, + "step": 2424 + }, + { + "epoch": 1.02, + "learning_rate": 1.9511612436498267e-05, + "loss": 0.3841, + "step": 2425 + }, + { + "epoch": 1.02, + "learning_rate": 1.9511088936418807e-05, + "loss": 0.3915, + "step": 2426 + }, + { + "epoch": 1.02, + "learning_rate": 1.9510565162951538e-05, + "loss": 0.4367, + "step": 2427 + }, + { + "epoch": 1.02, + "learning_rate": 1.951004111611151e-05, + "loss": 0.3407, + "step": 2428 + }, + { + "epoch": 1.02, + "learning_rate": 1.950951679591379e-05, + "loss": 0.3957, + "step": 2429 + }, + { + "epoch": 1.02, + "learning_rate": 1.9508992202373448e-05, + "loss": 0.3546, + "step": 2430 + }, + { + "epoch": 1.02, + "learning_rate": 1.9508467335505564e-05, + "loss": 0.3823, + "step": 2431 + }, + { + "epoch": 1.02, + "learning_rate": 1.9507942195325223e-05, + "loss": 0.3666, + "step": 2432 + }, + { + "epoch": 1.02, + "learning_rate": 1.9507416781847524e-05, + "loss": 0.4182, + "step": 2433 + }, + { + "epoch": 1.02, + "learning_rate": 1.9506891095087563e-05, + "loss": 0.4023, + "step": 2434 + }, + { + "epoch": 1.02, + "learning_rate": 1.950636513506045e-05, + "loss": 0.4253, + "step": 2435 + }, + { + "epoch": 1.02, + "learning_rate": 1.9505838901781314e-05, + "loss": 0.3536, + "step": 2436 + }, + { + "epoch": 1.02, + "learning_rate": 1.9505312395265273e-05, + "loss": 0.3591, + "step": 2437 + }, + { + "epoch": 1.02, + "learning_rate": 1.950478561552746e-05, + "loss": 0.3788, + "step": 2438 + }, + { + "epoch": 1.02, + "learning_rate": 1.9504258562583018e-05, + "loss": 0.3627, + "step": 2439 + }, + { + "epoch": 1.02, + "learning_rate": 1.95037312364471e-05, + "loss": 0.3955, + "step": 2440 + }, + { + "epoch": 1.02, + "learning_rate": 1.950320363713486e-05, + "loss": 0.3654, + "step": 2441 + }, + { + "epoch": 1.02, + "learning_rate": 1.9502675764661464e-05, + "loss": 0.3787, + "step": 2442 + }, + { + "epoch": 1.02, + "learning_rate": 1.9502147619042086e-05, + "loss": 0.3312, + "step": 2443 + }, + { + "epoch": 1.02, + "learning_rate": 1.950161920029191e-05, + "loss": 0.3557, + "step": 2444 + }, + { + "epoch": 1.02, + "learning_rate": 1.9501090508426117e-05, + "loss": 0.3739, + "step": 2445 + }, + { + "epoch": 1.02, + "learning_rate": 1.950056154345991e-05, + "loss": 0.3747, + "step": 2446 + }, + { + "epoch": 1.02, + "learning_rate": 1.9500032305408494e-05, + "loss": 0.3263, + "step": 2447 + }, + { + "epoch": 1.02, + "learning_rate": 1.9499502794287078e-05, + "loss": 0.3685, + "step": 2448 + }, + { + "epoch": 1.03, + "learning_rate": 1.9498973010110884e-05, + "loss": 0.3698, + "step": 2449 + }, + { + "epoch": 1.03, + "learning_rate": 1.949844295289514e-05, + "loss": 0.3479, + "step": 2450 + }, + { + "epoch": 1.03, + "learning_rate": 1.949791262265508e-05, + "loss": 0.3987, + "step": 2451 + }, + { + "epoch": 1.03, + "learning_rate": 1.949738201940595e-05, + "loss": 0.3598, + "step": 2452 + }, + { + "epoch": 1.03, + "learning_rate": 1.9496851143163004e-05, + "loss": 0.3291, + "step": 2453 + }, + { + "epoch": 1.03, + "learning_rate": 1.9496319993941498e-05, + "loss": 0.4026, + "step": 2454 + }, + { + "epoch": 1.03, + "learning_rate": 1.9495788571756704e-05, + "loss": 0.3992, + "step": 2455 + }, + { + "epoch": 1.03, + "learning_rate": 1.949525687662389e-05, + "loss": 0.3673, + "step": 2456 + }, + { + "epoch": 1.03, + "learning_rate": 1.9494724908558345e-05, + "loss": 0.391, + "step": 2457 + }, + { + "epoch": 1.03, + "learning_rate": 1.9494192667575355e-05, + "loss": 0.3577, + "step": 2458 + }, + { + "epoch": 1.03, + "learning_rate": 1.9493660153690225e-05, + "loss": 0.3866, + "step": 2459 + }, + { + "epoch": 1.03, + "learning_rate": 1.949312736691826e-05, + "loss": 0.3643, + "step": 2460 + }, + { + "epoch": 1.03, + "learning_rate": 1.9492594307274772e-05, + "loss": 0.3734, + "step": 2461 + }, + { + "epoch": 1.03, + "learning_rate": 1.949206097477508e-05, + "loss": 0.4079, + "step": 2462 + }, + { + "epoch": 1.03, + "learning_rate": 1.949152736943452e-05, + "loss": 0.3489, + "step": 2463 + }, + { + "epoch": 1.03, + "learning_rate": 1.949099349126843e-05, + "loss": 0.3724, + "step": 2464 + }, + { + "epoch": 1.03, + "learning_rate": 1.9490459340292153e-05, + "loss": 0.3742, + "step": 2465 + }, + { + "epoch": 1.03, + "learning_rate": 1.9489924916521047e-05, + "loss": 0.3711, + "step": 2466 + }, + { + "epoch": 1.03, + "learning_rate": 1.9489390219970468e-05, + "loss": 0.3826, + "step": 2467 + }, + { + "epoch": 1.03, + "learning_rate": 1.9488855250655785e-05, + "loss": 0.3584, + "step": 2468 + }, + { + "epoch": 1.03, + "learning_rate": 1.948832000859238e-05, + "loss": 0.3569, + "step": 2469 + }, + { + "epoch": 1.03, + "learning_rate": 1.9487784493795637e-05, + "loss": 0.3772, + "step": 2470 + }, + { + "epoch": 1.03, + "learning_rate": 1.948724870628095e-05, + "loss": 0.3452, + "step": 2471 + }, + { + "epoch": 1.03, + "learning_rate": 1.9486712646063715e-05, + "loss": 0.4041, + "step": 2472 + }, + { + "epoch": 1.04, + "learning_rate": 1.9486176313159344e-05, + "loss": 0.3882, + "step": 2473 + }, + { + "epoch": 1.04, + "learning_rate": 1.9485639707583255e-05, + "loss": 0.4075, + "step": 2474 + }, + { + "epoch": 1.04, + "learning_rate": 1.948510282935087e-05, + "loss": 0.355, + "step": 2475 + }, + { + "epoch": 1.04, + "learning_rate": 1.9484565678477616e-05, + "loss": 0.3827, + "step": 2476 + }, + { + "epoch": 1.04, + "learning_rate": 1.9484028254978942e-05, + "loss": 0.3761, + "step": 2477 + }, + { + "epoch": 1.04, + "learning_rate": 1.948349055887029e-05, + "loss": 0.4096, + "step": 2478 + }, + { + "epoch": 1.04, + "learning_rate": 1.9482952590167118e-05, + "loss": 0.3582, + "step": 2479 + }, + { + "epoch": 1.04, + "learning_rate": 1.9482414348884892e-05, + "loss": 0.3936, + "step": 2480 + }, + { + "epoch": 1.04, + "learning_rate": 1.9481875835039075e-05, + "loss": 0.3534, + "step": 2481 + }, + { + "epoch": 1.04, + "learning_rate": 1.9481337048645154e-05, + "loss": 0.3723, + "step": 2482 + }, + { + "epoch": 1.04, + "learning_rate": 1.9480797989718612e-05, + "loss": 0.4351, + "step": 2483 + }, + { + "epoch": 1.04, + "learning_rate": 1.9480258658274947e-05, + "loss": 0.3678, + "step": 2484 + }, + { + "epoch": 1.04, + "learning_rate": 1.9479719054329657e-05, + "loss": 0.3977, + "step": 2485 + }, + { + "epoch": 1.04, + "learning_rate": 1.9479179177898255e-05, + "loss": 0.3488, + "step": 2486 + }, + { + "epoch": 1.04, + "learning_rate": 1.9478639028996262e-05, + "loss": 0.3655, + "step": 2487 + }, + { + "epoch": 1.04, + "learning_rate": 1.9478098607639197e-05, + "loss": 0.4027, + "step": 2488 + }, + { + "epoch": 1.04, + "learning_rate": 1.94775579138426e-05, + "loss": 0.3937, + "step": 2489 + }, + { + "epoch": 1.04, + "learning_rate": 1.9477016947622013e-05, + "loss": 0.354, + "step": 2490 + }, + { + "epoch": 1.04, + "learning_rate": 1.947647570899298e-05, + "loss": 0.4085, + "step": 2491 + }, + { + "epoch": 1.04, + "learning_rate": 1.947593419797107e-05, + "loss": 0.3629, + "step": 2492 + }, + { + "epoch": 1.04, + "learning_rate": 1.947539241457183e-05, + "loss": 0.3365, + "step": 2493 + }, + { + "epoch": 1.04, + "learning_rate": 1.947485035881085e-05, + "loss": 0.3338, + "step": 2494 + }, + { + "epoch": 1.04, + "learning_rate": 1.94743080307037e-05, + "loss": 0.4101, + "step": 2495 + }, + { + "epoch": 1.05, + "learning_rate": 1.947376543026598e-05, + "loss": 0.3708, + "step": 2496 + }, + { + "epoch": 1.05, + "learning_rate": 1.9473222557513273e-05, + "loss": 0.3908, + "step": 2497 + }, + { + "epoch": 1.05, + "learning_rate": 1.947267941246119e-05, + "loss": 0.3852, + "step": 2498 + }, + { + "epoch": 1.05, + "learning_rate": 1.9472135995125344e-05, + "loss": 0.3899, + "step": 2499 + }, + { + "epoch": 1.05, + "learning_rate": 1.9471592305521355e-05, + "loss": 0.4517, + "step": 2500 + }, + { + "epoch": 1.05, + "learning_rate": 1.9471048343664853e-05, + "loss": 0.3975, + "step": 2501 + }, + { + "epoch": 1.05, + "learning_rate": 1.9470504109571467e-05, + "loss": 0.4205, + "step": 2502 + }, + { + "epoch": 1.05, + "learning_rate": 1.9469959603256844e-05, + "loss": 0.3591, + "step": 2503 + }, + { + "epoch": 1.05, + "learning_rate": 1.946941482473664e-05, + "loss": 0.3632, + "step": 2504 + }, + { + "epoch": 1.05, + "learning_rate": 1.9468869774026508e-05, + "loss": 0.4138, + "step": 2505 + }, + { + "epoch": 1.05, + "learning_rate": 1.9468324451142115e-05, + "loss": 0.3705, + "step": 2506 + }, + { + "epoch": 1.05, + "learning_rate": 1.9467778856099134e-05, + "loss": 0.3899, + "step": 2507 + }, + { + "epoch": 1.05, + "learning_rate": 1.946723298891326e-05, + "loss": 0.342, + "step": 2508 + }, + { + "epoch": 1.05, + "learning_rate": 1.9466686849600168e-05, + "loss": 0.4147, + "step": 2509 + }, + { + "epoch": 1.05, + "learning_rate": 1.9466140438175567e-05, + "loss": 0.391, + "step": 2510 + }, + { + "epoch": 1.05, + "learning_rate": 1.9465593754655158e-05, + "loss": 0.3563, + "step": 2511 + }, + { + "epoch": 1.05, + "learning_rate": 1.9465046799054657e-05, + "loss": 0.4002, + "step": 2512 + }, + { + "epoch": 1.05, + "learning_rate": 1.946449957138978e-05, + "loss": 0.4012, + "step": 2513 + }, + { + "epoch": 1.05, + "learning_rate": 1.9463952071676268e-05, + "loss": 0.3771, + "step": 2514 + }, + { + "epoch": 1.05, + "learning_rate": 1.9463404299929852e-05, + "loss": 0.3488, + "step": 2515 + }, + { + "epoch": 1.05, + "learning_rate": 1.9462856256166273e-05, + "loss": 0.379, + "step": 2516 + }, + { + "epoch": 1.05, + "learning_rate": 1.9462307940401288e-05, + "loss": 0.4167, + "step": 2517 + }, + { + "epoch": 1.05, + "learning_rate": 1.9461759352650663e-05, + "loss": 0.3738, + "step": 2518 + }, + { + "epoch": 1.05, + "learning_rate": 1.9461210492930156e-05, + "loss": 0.3786, + "step": 2519 + }, + { + "epoch": 1.06, + "learning_rate": 1.9460661361255552e-05, + "loss": 0.422, + "step": 2520 + }, + { + "epoch": 1.06, + "learning_rate": 1.946011195764263e-05, + "loss": 0.3372, + "step": 2521 + }, + { + "epoch": 1.06, + "learning_rate": 1.9459562282107188e-05, + "loss": 0.3704, + "step": 2522 + }, + { + "epoch": 1.06, + "learning_rate": 1.945901233466502e-05, + "loss": 0.3586, + "step": 2523 + }, + { + "epoch": 1.06, + "learning_rate": 1.9458462115331938e-05, + "loss": 0.3876, + "step": 2524 + }, + { + "epoch": 1.06, + "learning_rate": 1.9457911624123754e-05, + "loss": 0.3911, + "step": 2525 + }, + { + "epoch": 1.06, + "learning_rate": 1.9457360861056294e-05, + "loss": 0.3998, + "step": 2526 + }, + { + "epoch": 1.06, + "learning_rate": 1.9456809826145387e-05, + "loss": 0.3516, + "step": 2527 + }, + { + "epoch": 1.06, + "learning_rate": 1.9456258519406875e-05, + "loss": 0.3591, + "step": 2528 + }, + { + "epoch": 1.06, + "learning_rate": 1.9455706940856602e-05, + "loss": 0.4332, + "step": 2529 + }, + { + "epoch": 1.06, + "learning_rate": 1.9455155090510427e-05, + "loss": 0.4036, + "step": 2530 + }, + { + "epoch": 1.06, + "learning_rate": 1.9454602968384204e-05, + "loss": 0.3703, + "step": 2531 + }, + { + "epoch": 1.06, + "learning_rate": 1.9454050574493815e-05, + "loss": 0.3515, + "step": 2532 + }, + { + "epoch": 1.06, + "learning_rate": 1.9453497908855125e-05, + "loss": 0.4121, + "step": 2533 + }, + { + "epoch": 1.06, + "learning_rate": 1.945294497148403e-05, + "loss": 0.3881, + "step": 2534 + }, + { + "epoch": 1.06, + "learning_rate": 1.945239176239642e-05, + "loss": 0.3798, + "step": 2535 + }, + { + "epoch": 1.06, + "learning_rate": 1.94518382816082e-05, + "loss": 0.3696, + "step": 2536 + }, + { + "epoch": 1.06, + "learning_rate": 1.945128452913527e-05, + "loss": 0.3494, + "step": 2537 + }, + { + "epoch": 1.06, + "learning_rate": 1.9450730504993554e-05, + "loss": 0.4292, + "step": 2538 + }, + { + "epoch": 1.06, + "learning_rate": 1.9450176209198977e-05, + "loss": 0.3643, + "step": 2539 + }, + { + "epoch": 1.06, + "learning_rate": 1.9449621641767474e-05, + "loss": 0.4396, + "step": 2540 + }, + { + "epoch": 1.06, + "learning_rate": 1.944906680271498e-05, + "loss": 0.3905, + "step": 2541 + }, + { + "epoch": 1.06, + "learning_rate": 1.9448511692057446e-05, + "loss": 0.377, + "step": 2542 + }, + { + "epoch": 1.06, + "learning_rate": 1.9447956309810825e-05, + "loss": 0.3605, + "step": 2543 + }, + { + "epoch": 1.07, + "learning_rate": 1.9447400655991086e-05, + "loss": 0.4316, + "step": 2544 + }, + { + "epoch": 1.07, + "learning_rate": 1.94468447306142e-05, + "loss": 0.3723, + "step": 2545 + }, + { + "epoch": 1.07, + "learning_rate": 1.9446288533696145e-05, + "loss": 0.4058, + "step": 2546 + }, + { + "epoch": 1.07, + "learning_rate": 1.9445732065252904e-05, + "loss": 0.379, + "step": 2547 + }, + { + "epoch": 1.07, + "learning_rate": 1.9445175325300482e-05, + "loss": 0.4104, + "step": 2548 + }, + { + "epoch": 1.07, + "learning_rate": 1.9444618313854878e-05, + "loss": 0.4294, + "step": 2549 + }, + { + "epoch": 1.07, + "learning_rate": 1.9444061030932094e-05, + "loss": 0.3712, + "step": 2550 + }, + { + "epoch": 1.07, + "learning_rate": 1.9443503476548162e-05, + "loss": 0.3953, + "step": 2551 + }, + { + "epoch": 1.07, + "learning_rate": 1.9442945650719102e-05, + "loss": 0.3694, + "step": 2552 + }, + { + "epoch": 1.07, + "learning_rate": 1.944238755346095e-05, + "loss": 0.4309, + "step": 2553 + }, + { + "epoch": 1.07, + "learning_rate": 1.9441829184789745e-05, + "loss": 0.3836, + "step": 2554 + }, + { + "epoch": 1.07, + "learning_rate": 1.9441270544721536e-05, + "loss": 0.4283, + "step": 2555 + }, + { + "epoch": 1.07, + "learning_rate": 1.9440711633272387e-05, + "loss": 0.3818, + "step": 2556 + }, + { + "epoch": 1.07, + "learning_rate": 1.944015245045836e-05, + "loss": 0.3676, + "step": 2557 + }, + { + "epoch": 1.07, + "learning_rate": 1.9439592996295525e-05, + "loss": 0.3511, + "step": 2558 + }, + { + "epoch": 1.07, + "learning_rate": 1.943903327079997e-05, + "loss": 0.3633, + "step": 2559 + }, + { + "epoch": 1.07, + "learning_rate": 1.9438473273987775e-05, + "loss": 0.3664, + "step": 2560 + }, + { + "epoch": 1.07, + "learning_rate": 1.9437913005875046e-05, + "loss": 0.4262, + "step": 2561 + }, + { + "epoch": 1.07, + "learning_rate": 1.943735246647788e-05, + "loss": 0.4054, + "step": 2562 + }, + { + "epoch": 1.07, + "learning_rate": 1.9436791655812392e-05, + "loss": 0.4121, + "step": 2563 + }, + { + "epoch": 1.07, + "learning_rate": 1.94362305738947e-05, + "loss": 0.3511, + "step": 2564 + }, + { + "epoch": 1.07, + "learning_rate": 1.9435669220740933e-05, + "loss": 0.3689, + "step": 2565 + }, + { + "epoch": 1.07, + "learning_rate": 1.943510759636723e-05, + "loss": 0.3972, + "step": 2566 + }, + { + "epoch": 1.07, + "learning_rate": 1.9434545700789732e-05, + "loss": 0.404, + "step": 2567 + }, + { + "epoch": 1.08, + "learning_rate": 1.943398353402459e-05, + "loss": 0.3609, + "step": 2568 + }, + { + "epoch": 1.08, + "learning_rate": 1.9433421096087963e-05, + "loss": 0.3381, + "step": 2569 + }, + { + "epoch": 1.08, + "learning_rate": 1.9432858386996018e-05, + "loss": 0.3895, + "step": 2570 + }, + { + "epoch": 1.08, + "learning_rate": 1.9432295406764927e-05, + "loss": 0.3767, + "step": 2571 + }, + { + "epoch": 1.08, + "learning_rate": 1.9431732155410878e-05, + "loss": 0.3764, + "step": 2572 + }, + { + "epoch": 1.08, + "learning_rate": 1.9431168632950053e-05, + "loss": 0.3678, + "step": 2573 + }, + { + "epoch": 1.08, + "learning_rate": 1.9430604839398657e-05, + "loss": 0.3845, + "step": 2574 + }, + { + "epoch": 1.08, + "learning_rate": 1.9430040774772893e-05, + "loss": 0.42, + "step": 2575 + }, + { + "epoch": 1.08, + "learning_rate": 1.9429476439088976e-05, + "loss": 0.385, + "step": 2576 + }, + { + "epoch": 1.08, + "learning_rate": 1.9428911832363126e-05, + "loss": 0.4227, + "step": 2577 + }, + { + "epoch": 1.08, + "learning_rate": 1.9428346954611574e-05, + "loss": 0.3842, + "step": 2578 + }, + { + "epoch": 1.08, + "learning_rate": 1.9427781805850553e-05, + "loss": 0.3735, + "step": 2579 + }, + { + "epoch": 1.08, + "learning_rate": 1.9427216386096313e-05, + "loss": 0.3866, + "step": 2580 + }, + { + "epoch": 1.08, + "learning_rate": 1.94266506953651e-05, + "loss": 0.3834, + "step": 2581 + }, + { + "epoch": 1.08, + "learning_rate": 1.942608473367318e-05, + "loss": 0.4074, + "step": 2582 + }, + { + "epoch": 1.08, + "learning_rate": 1.942551850103682e-05, + "loss": 0.4092, + "step": 2583 + }, + { + "epoch": 1.08, + "learning_rate": 1.942495199747229e-05, + "loss": 0.3959, + "step": 2584 + }, + { + "epoch": 1.08, + "learning_rate": 1.9424385222995883e-05, + "loss": 0.3953, + "step": 2585 + }, + { + "epoch": 1.08, + "learning_rate": 1.9423818177623884e-05, + "loss": 0.3939, + "step": 2586 + }, + { + "epoch": 1.08, + "learning_rate": 1.9423250861372592e-05, + "loss": 0.4049, + "step": 2587 + }, + { + "epoch": 1.08, + "learning_rate": 1.9422683274258318e-05, + "loss": 0.3754, + "step": 2588 + }, + { + "epoch": 1.08, + "learning_rate": 1.9422115416297377e-05, + "loss": 0.4024, + "step": 2589 + }, + { + "epoch": 1.08, + "learning_rate": 1.9421547287506085e-05, + "loss": 0.358, + "step": 2590 + }, + { + "epoch": 1.08, + "learning_rate": 1.942097888790078e-05, + "loss": 0.3966, + "step": 2591 + }, + { + "epoch": 1.09, + "learning_rate": 1.9420410217497795e-05, + "loss": 0.3881, + "step": 2592 + }, + { + "epoch": 1.09, + "learning_rate": 1.941984127631348e-05, + "loss": 0.3901, + "step": 2593 + }, + { + "epoch": 1.09, + "learning_rate": 1.9419272064364186e-05, + "loss": 0.4057, + "step": 2594 + }, + { + "epoch": 1.09, + "learning_rate": 1.9418702581666276e-05, + "loss": 0.351, + "step": 2595 + }, + { + "epoch": 1.09, + "learning_rate": 1.9418132828236114e-05, + "loss": 0.3222, + "step": 2596 + }, + { + "epoch": 1.09, + "learning_rate": 1.9417562804090086e-05, + "loss": 0.3725, + "step": 2597 + }, + { + "epoch": 1.09, + "learning_rate": 1.941699250924457e-05, + "loss": 0.4544, + "step": 2598 + }, + { + "epoch": 1.09, + "learning_rate": 1.941642194371596e-05, + "loss": 0.408, + "step": 2599 + }, + { + "epoch": 1.09, + "learning_rate": 1.941585110752066e-05, + "loss": 0.3905, + "step": 2600 + }, + { + "epoch": 1.09, + "learning_rate": 1.9415280000675074e-05, + "loss": 0.39, + "step": 2601 + }, + { + "epoch": 1.09, + "learning_rate": 1.9414708623195618e-05, + "loss": 0.363, + "step": 2602 + }, + { + "epoch": 1.09, + "learning_rate": 1.9414136975098718e-05, + "loss": 0.3951, + "step": 2603 + }, + { + "epoch": 1.09, + "learning_rate": 1.9413565056400804e-05, + "loss": 0.3983, + "step": 2604 + }, + { + "epoch": 1.09, + "learning_rate": 1.9412992867118318e-05, + "loss": 0.3811, + "step": 2605 + }, + { + "epoch": 1.09, + "learning_rate": 1.9412420407267704e-05, + "loss": 0.3948, + "step": 2606 + }, + { + "epoch": 1.09, + "learning_rate": 1.9411847676865415e-05, + "loss": 0.3897, + "step": 2607 + }, + { + "epoch": 1.09, + "learning_rate": 1.9411274675927922e-05, + "loss": 0.3905, + "step": 2608 + }, + { + "epoch": 1.09, + "learning_rate": 1.9410701404471686e-05, + "loss": 0.4154, + "step": 2609 + }, + { + "epoch": 1.09, + "learning_rate": 1.941012786251319e-05, + "loss": 0.3532, + "step": 2610 + }, + { + "epoch": 1.09, + "learning_rate": 1.9409554050068916e-05, + "loss": 0.3639, + "step": 2611 + }, + { + "epoch": 1.09, + "learning_rate": 1.9408979967155364e-05, + "loss": 0.3617, + "step": 2612 + }, + { + "epoch": 1.09, + "learning_rate": 1.940840561378903e-05, + "loss": 0.3866, + "step": 2613 + }, + { + "epoch": 1.09, + "learning_rate": 1.940783098998643e-05, + "loss": 0.3641, + "step": 2614 + }, + { + "epoch": 1.09, + "learning_rate": 1.9407256095764074e-05, + "loss": 0.3998, + "step": 2615 + }, + { + "epoch": 1.1, + "learning_rate": 1.9406680931138488e-05, + "loss": 0.3803, + "step": 2616 + }, + { + "epoch": 1.1, + "learning_rate": 1.940610549612621e-05, + "loss": 0.3922, + "step": 2617 + }, + { + "epoch": 1.1, + "learning_rate": 1.940552979074377e-05, + "loss": 0.372, + "step": 2618 + }, + { + "epoch": 1.1, + "learning_rate": 1.940495381500773e-05, + "loss": 0.3683, + "step": 2619 + }, + { + "epoch": 1.1, + "learning_rate": 1.9404377568934633e-05, + "loss": 0.4032, + "step": 2620 + }, + { + "epoch": 1.1, + "learning_rate": 1.940380105254105e-05, + "loss": 0.3778, + "step": 2621 + }, + { + "epoch": 1.1, + "learning_rate": 1.9403224265843547e-05, + "loss": 0.3555, + "step": 2622 + }, + { + "epoch": 1.1, + "learning_rate": 1.9402647208858712e-05, + "loss": 0.3768, + "step": 2623 + }, + { + "epoch": 1.1, + "learning_rate": 1.9402069881603123e-05, + "loss": 0.3493, + "step": 2624 + }, + { + "epoch": 1.1, + "learning_rate": 1.9401492284093382e-05, + "loss": 0.4207, + "step": 2625 + }, + { + "epoch": 1.1, + "learning_rate": 1.9400914416346083e-05, + "loss": 0.3887, + "step": 2626 + }, + { + "epoch": 1.1, + "learning_rate": 1.9400336278377845e-05, + "loss": 0.3926, + "step": 2627 + }, + { + "epoch": 1.1, + "learning_rate": 1.939975787020528e-05, + "loss": 0.3902, + "step": 2628 + }, + { + "epoch": 1.1, + "learning_rate": 1.9399179191845022e-05, + "loss": 0.3577, + "step": 2629 + }, + { + "epoch": 1.1, + "learning_rate": 1.9398600243313695e-05, + "loss": 0.3764, + "step": 2630 + }, + { + "epoch": 1.1, + "learning_rate": 1.939802102462794e-05, + "loss": 0.3581, + "step": 2631 + }, + { + "epoch": 1.1, + "learning_rate": 1.9397441535804417e-05, + "loss": 0.3631, + "step": 2632 + }, + { + "epoch": 1.1, + "learning_rate": 1.939686177685977e-05, + "loss": 0.3528, + "step": 2633 + }, + { + "epoch": 1.1, + "learning_rate": 1.939628174781068e-05, + "loss": 0.3927, + "step": 2634 + }, + { + "epoch": 1.1, + "learning_rate": 1.9395701448673797e-05, + "loss": 0.3593, + "step": 2635 + }, + { + "epoch": 1.1, + "learning_rate": 1.939512087946582e-05, + "loss": 0.3985, + "step": 2636 + }, + { + "epoch": 1.1, + "learning_rate": 1.9394540040203428e-05, + "loss": 0.3532, + "step": 2637 + }, + { + "epoch": 1.1, + "learning_rate": 1.9393958930903323e-05, + "loss": 0.3842, + "step": 2638 + }, + { + "epoch": 1.1, + "learning_rate": 1.93933775515822e-05, + "loss": 0.3824, + "step": 2639 + }, + { + "epoch": 1.11, + "learning_rate": 1.939279590225678e-05, + "loss": 0.3778, + "step": 2640 + }, + { + "epoch": 1.11, + "learning_rate": 1.9392213982943774e-05, + "loss": 0.3495, + "step": 2641 + }, + { + "epoch": 1.11, + "learning_rate": 1.9391631793659908e-05, + "loss": 0.3768, + "step": 2642 + }, + { + "epoch": 1.11, + "learning_rate": 1.9391049334421925e-05, + "loss": 0.3629, + "step": 2643 + }, + { + "epoch": 1.11, + "learning_rate": 1.939046660524656e-05, + "loss": 0.375, + "step": 2644 + }, + { + "epoch": 1.11, + "learning_rate": 1.938988360615057e-05, + "loss": 0.3699, + "step": 2645 + }, + { + "epoch": 1.11, + "learning_rate": 1.93893003371507e-05, + "loss": 0.3716, + "step": 2646 + }, + { + "epoch": 1.11, + "learning_rate": 1.9388716798263728e-05, + "loss": 0.4273, + "step": 2647 + }, + { + "epoch": 1.11, + "learning_rate": 1.9388132989506422e-05, + "loss": 0.3377, + "step": 2648 + }, + { + "epoch": 1.11, + "learning_rate": 1.9387548910895567e-05, + "loss": 0.3908, + "step": 2649 + }, + { + "epoch": 1.11, + "learning_rate": 1.9386964562447947e-05, + "loss": 0.3764, + "step": 2650 + }, + { + "epoch": 1.11, + "learning_rate": 1.938637994418036e-05, + "loss": 0.4319, + "step": 2651 + }, + { + "epoch": 1.11, + "learning_rate": 1.9385795056109614e-05, + "loss": 0.3947, + "step": 2652 + }, + { + "epoch": 1.11, + "learning_rate": 1.938520989825251e-05, + "loss": 0.3839, + "step": 2653 + }, + { + "epoch": 1.11, + "learning_rate": 1.9384624470625884e-05, + "loss": 0.4071, + "step": 2654 + }, + { + "epoch": 1.11, + "learning_rate": 1.9384038773246555e-05, + "loss": 0.3845, + "step": 2655 + }, + { + "epoch": 1.11, + "learning_rate": 1.9383452806131356e-05, + "loss": 0.4514, + "step": 2656 + }, + { + "epoch": 1.11, + "learning_rate": 1.9382866569297133e-05, + "loss": 0.34, + "step": 2657 + }, + { + "epoch": 1.11, + "learning_rate": 1.9382280062760738e-05, + "loss": 0.3823, + "step": 2658 + }, + { + "epoch": 1.11, + "learning_rate": 1.9381693286539024e-05, + "loss": 0.381, + "step": 2659 + }, + { + "epoch": 1.11, + "learning_rate": 1.9381106240648867e-05, + "loss": 0.3852, + "step": 2660 + }, + { + "epoch": 1.11, + "learning_rate": 1.9380518925107133e-05, + "loss": 0.4091, + "step": 2661 + }, + { + "epoch": 1.11, + "learning_rate": 1.9379931339930705e-05, + "loss": 0.3806, + "step": 2662 + }, + { + "epoch": 1.11, + "learning_rate": 1.9379343485136478e-05, + "loss": 0.3209, + "step": 2663 + }, + { + "epoch": 1.12, + "learning_rate": 1.9378755360741342e-05, + "loss": 0.3436, + "step": 2664 + }, + { + "epoch": 1.12, + "learning_rate": 1.9378166966762207e-05, + "loss": 0.392, + "step": 2665 + }, + { + "epoch": 1.12, + "learning_rate": 1.9377578303215984e-05, + "loss": 0.3796, + "step": 2666 + }, + { + "epoch": 1.12, + "learning_rate": 1.9376989370119595e-05, + "loss": 0.3759, + "step": 2667 + }, + { + "epoch": 1.12, + "learning_rate": 1.937640016748997e-05, + "loss": 0.4186, + "step": 2668 + }, + { + "epoch": 1.12, + "learning_rate": 1.937581069534404e-05, + "loss": 0.3675, + "step": 2669 + }, + { + "epoch": 1.12, + "learning_rate": 1.937522095369875e-05, + "loss": 0.3504, + "step": 2670 + }, + { + "epoch": 1.12, + "learning_rate": 1.9374630942571055e-05, + "loss": 0.3958, + "step": 2671 + }, + { + "epoch": 1.12, + "learning_rate": 1.9374040661977913e-05, + "loss": 0.3544, + "step": 2672 + }, + { + "epoch": 1.12, + "learning_rate": 1.937345011193629e-05, + "loss": 0.376, + "step": 2673 + }, + { + "epoch": 1.12, + "learning_rate": 1.9372859292463163e-05, + "loss": 0.3849, + "step": 2674 + }, + { + "epoch": 1.12, + "learning_rate": 1.9372268203575512e-05, + "loss": 0.3558, + "step": 2675 + }, + { + "epoch": 1.12, + "learning_rate": 1.9371676845290327e-05, + "loss": 0.3714, + "step": 2676 + }, + { + "epoch": 1.12, + "learning_rate": 1.9371085217624608e-05, + "loss": 0.4111, + "step": 2677 + }, + { + "epoch": 1.12, + "learning_rate": 1.937049332059536e-05, + "loss": 0.3657, + "step": 2678 + }, + { + "epoch": 1.12, + "learning_rate": 1.9369901154219595e-05, + "loss": 0.3718, + "step": 2679 + }, + { + "epoch": 1.12, + "learning_rate": 1.9369308718514342e-05, + "loss": 0.3688, + "step": 2680 + }, + { + "epoch": 1.12, + "learning_rate": 1.9368716013496616e-05, + "loss": 0.4108, + "step": 2681 + }, + { + "epoch": 1.12, + "learning_rate": 1.9368123039183468e-05, + "loss": 0.3761, + "step": 2682 + }, + { + "epoch": 1.12, + "learning_rate": 1.9367529795591935e-05, + "loss": 0.3954, + "step": 2683 + }, + { + "epoch": 1.12, + "learning_rate": 1.9366936282739067e-05, + "loss": 0.3762, + "step": 2684 + }, + { + "epoch": 1.12, + "learning_rate": 1.9366342500641933e-05, + "loss": 0.3663, + "step": 2685 + }, + { + "epoch": 1.12, + "learning_rate": 1.9365748449317593e-05, + "loss": 0.3908, + "step": 2686 + }, + { + "epoch": 1.12, + "learning_rate": 1.9365154128783126e-05, + "loss": 0.3524, + "step": 2687 + }, + { + "epoch": 1.13, + "learning_rate": 1.9364559539055612e-05, + "loss": 0.364, + "step": 2688 + }, + { + "epoch": 1.13, + "learning_rate": 1.9363964680152147e-05, + "loss": 0.3783, + "step": 2689 + }, + { + "epoch": 1.13, + "learning_rate": 1.936336955208982e-05, + "loss": 0.3973, + "step": 2690 + }, + { + "epoch": 1.13, + "learning_rate": 1.936277415488575e-05, + "loss": 0.3912, + "step": 2691 + }, + { + "epoch": 1.13, + "learning_rate": 1.9362178488557043e-05, + "loss": 0.4189, + "step": 2692 + }, + { + "epoch": 1.13, + "learning_rate": 1.9361582553120824e-05, + "loss": 0.3805, + "step": 2693 + }, + { + "epoch": 1.13, + "learning_rate": 1.9360986348594226e-05, + "loss": 0.3507, + "step": 2694 + }, + { + "epoch": 1.13, + "learning_rate": 1.9360389874994374e-05, + "loss": 0.4058, + "step": 2695 + }, + { + "epoch": 1.13, + "learning_rate": 1.9359793132338427e-05, + "loss": 0.3807, + "step": 2696 + }, + { + "epoch": 1.13, + "learning_rate": 1.935919612064353e-05, + "loss": 0.4055, + "step": 2697 + }, + { + "epoch": 1.13, + "learning_rate": 1.9358598839926848e-05, + "loss": 0.3797, + "step": 2698 + }, + { + "epoch": 1.13, + "learning_rate": 1.9358001290205542e-05, + "loss": 0.3917, + "step": 2699 + }, + { + "epoch": 1.13, + "learning_rate": 1.9357403471496797e-05, + "loss": 0.3894, + "step": 2700 + }, + { + "epoch": 1.13, + "learning_rate": 1.9356805383817792e-05, + "loss": 0.3666, + "step": 2701 + }, + { + "epoch": 1.13, + "learning_rate": 1.935620702718572e-05, + "loss": 0.4063, + "step": 2702 + }, + { + "epoch": 1.13, + "learning_rate": 1.9355608401617782e-05, + "loss": 0.3768, + "step": 2703 + }, + { + "epoch": 1.13, + "learning_rate": 1.935500950713118e-05, + "loss": 0.3813, + "step": 2704 + }, + { + "epoch": 1.13, + "learning_rate": 1.9354410343743127e-05, + "loss": 0.3715, + "step": 2705 + }, + { + "epoch": 1.13, + "learning_rate": 1.9353810911470855e-05, + "loss": 0.3931, + "step": 2706 + }, + { + "epoch": 1.13, + "learning_rate": 1.935321121033159e-05, + "loss": 0.41, + "step": 2707 + }, + { + "epoch": 1.13, + "learning_rate": 1.9352611240342565e-05, + "loss": 0.4062, + "step": 2708 + }, + { + "epoch": 1.13, + "learning_rate": 1.935201100152103e-05, + "loss": 0.3632, + "step": 2709 + }, + { + "epoch": 1.13, + "learning_rate": 1.9351410493884234e-05, + "loss": 0.3832, + "step": 2710 + }, + { + "epoch": 1.14, + "learning_rate": 1.9350809717449448e-05, + "loss": 0.3753, + "step": 2711 + }, + { + "epoch": 1.14, + "learning_rate": 1.935020867223393e-05, + "loss": 0.346, + "step": 2712 + }, + { + "epoch": 1.14, + "learning_rate": 1.9349607358254965e-05, + "loss": 0.3275, + "step": 2713 + }, + { + "epoch": 1.14, + "learning_rate": 1.934900577552983e-05, + "loss": 0.4153, + "step": 2714 + }, + { + "epoch": 1.14, + "learning_rate": 1.934840392407582e-05, + "loss": 0.3351, + "step": 2715 + }, + { + "epoch": 1.14, + "learning_rate": 1.9347801803910236e-05, + "loss": 0.4013, + "step": 2716 + }, + { + "epoch": 1.14, + "learning_rate": 1.9347199415050383e-05, + "loss": 0.37, + "step": 2717 + }, + { + "epoch": 1.14, + "learning_rate": 1.934659675751358e-05, + "loss": 0.3452, + "step": 2718 + }, + { + "epoch": 1.14, + "learning_rate": 1.9345993831317147e-05, + "loss": 0.307, + "step": 2719 + }, + { + "epoch": 1.14, + "learning_rate": 1.9345390636478414e-05, + "loss": 0.3732, + "step": 2720 + }, + { + "epoch": 1.14, + "learning_rate": 1.934478717301472e-05, + "loss": 0.4037, + "step": 2721 + }, + { + "epoch": 1.14, + "learning_rate": 1.9344183440943408e-05, + "loss": 0.4228, + "step": 2722 + }, + { + "epoch": 1.14, + "learning_rate": 1.934357944028184e-05, + "loss": 0.3822, + "step": 2723 + }, + { + "epoch": 1.14, + "learning_rate": 1.934297517104737e-05, + "loss": 0.3925, + "step": 2724 + }, + { + "epoch": 1.14, + "learning_rate": 1.9342370633257367e-05, + "loss": 0.3673, + "step": 2725 + }, + { + "epoch": 1.14, + "learning_rate": 1.9341765826929216e-05, + "loss": 0.4204, + "step": 2726 + }, + { + "epoch": 1.14, + "learning_rate": 1.9341160752080294e-05, + "loss": 0.429, + "step": 2727 + }, + { + "epoch": 1.14, + "learning_rate": 1.9340555408727993e-05, + "loss": 0.3993, + "step": 2728 + }, + { + "epoch": 1.14, + "learning_rate": 1.9339949796889717e-05, + "loss": 0.4509, + "step": 2729 + }, + { + "epoch": 1.14, + "learning_rate": 1.933934391658287e-05, + "loss": 0.3768, + "step": 2730 + }, + { + "epoch": 1.14, + "learning_rate": 1.933873776782487e-05, + "loss": 0.3504, + "step": 2731 + }, + { + "epoch": 1.14, + "learning_rate": 1.9338131350633143e-05, + "loss": 0.4084, + "step": 2732 + }, + { + "epoch": 1.14, + "learning_rate": 1.9337524665025114e-05, + "loss": 0.3781, + "step": 2733 + }, + { + "epoch": 1.14, + "learning_rate": 1.9336917711018227e-05, + "loss": 0.4215, + "step": 2734 + }, + { + "epoch": 1.15, + "learning_rate": 1.9336310488629926e-05, + "loss": 0.4066, + "step": 2735 + }, + { + "epoch": 1.15, + "learning_rate": 1.9335702997877664e-05, + "loss": 0.4147, + "step": 2736 + }, + { + "epoch": 1.15, + "learning_rate": 1.9335095238778903e-05, + "loss": 0.3655, + "step": 2737 + }, + { + "epoch": 1.15, + "learning_rate": 1.9334487211351113e-05, + "loss": 0.3868, + "step": 2738 + }, + { + "epoch": 1.15, + "learning_rate": 1.933387891561177e-05, + "loss": 0.3515, + "step": 2739 + }, + { + "epoch": 1.15, + "learning_rate": 1.9333270351578367e-05, + "loss": 0.3973, + "step": 2740 + }, + { + "epoch": 1.15, + "learning_rate": 1.9332661519268384e-05, + "loss": 0.4615, + "step": 2741 + }, + { + "epoch": 1.15, + "learning_rate": 1.933205241869933e-05, + "loss": 0.4009, + "step": 2742 + }, + { + "epoch": 1.15, + "learning_rate": 1.9331443049888706e-05, + "loss": 0.3969, + "step": 2743 + }, + { + "epoch": 1.15, + "learning_rate": 1.9330833412854035e-05, + "loss": 0.3819, + "step": 2744 + }, + { + "epoch": 1.15, + "learning_rate": 1.933022350761284e-05, + "loss": 0.3491, + "step": 2745 + }, + { + "epoch": 1.15, + "learning_rate": 1.9329613334182644e-05, + "loss": 0.3933, + "step": 2746 + }, + { + "epoch": 1.15, + "learning_rate": 1.9329002892580995e-05, + "loss": 0.3607, + "step": 2747 + }, + { + "epoch": 1.15, + "learning_rate": 1.932839218282544e-05, + "loss": 0.3915, + "step": 2748 + }, + { + "epoch": 1.15, + "learning_rate": 1.9327781204933522e-05, + "loss": 0.355, + "step": 2749 + }, + { + "epoch": 1.15, + "learning_rate": 1.9327169958922813e-05, + "loss": 0.4151, + "step": 2750 + }, + { + "epoch": 1.15, + "learning_rate": 1.9326558444810883e-05, + "loss": 0.3827, + "step": 2751 + }, + { + "epoch": 1.15, + "learning_rate": 1.9325946662615304e-05, + "loss": 0.4245, + "step": 2752 + }, + { + "epoch": 1.15, + "learning_rate": 1.932533461235367e-05, + "loss": 0.3551, + "step": 2753 + }, + { + "epoch": 1.15, + "learning_rate": 1.932472229404356e-05, + "loss": 0.3895, + "step": 2754 + }, + { + "epoch": 1.15, + "learning_rate": 1.9324109707702585e-05, + "loss": 0.3821, + "step": 2755 + }, + { + "epoch": 1.15, + "learning_rate": 1.932349685334835e-05, + "loss": 0.3928, + "step": 2756 + }, + { + "epoch": 1.15, + "learning_rate": 1.9322883730998474e-05, + "loss": 0.3673, + "step": 2757 + }, + { + "epoch": 1.15, + "learning_rate": 1.9322270340670574e-05, + "loss": 0.3496, + "step": 2758 + }, + { + "epoch": 1.16, + "learning_rate": 1.9321656682382287e-05, + "loss": 0.3826, + "step": 2759 + }, + { + "epoch": 1.16, + "learning_rate": 1.9321042756151254e-05, + "loss": 0.3732, + "step": 2760 + }, + { + "epoch": 1.16, + "learning_rate": 1.9320428561995114e-05, + "loss": 0.3547, + "step": 2761 + }, + { + "epoch": 1.16, + "learning_rate": 1.931981409993153e-05, + "loss": 0.337, + "step": 2762 + }, + { + "epoch": 1.16, + "learning_rate": 1.9319199369978157e-05, + "loss": 0.3734, + "step": 2763 + }, + { + "epoch": 1.16, + "learning_rate": 1.9318584372152668e-05, + "loss": 0.3379, + "step": 2764 + }, + { + "epoch": 1.16, + "learning_rate": 1.9317969106472742e-05, + "loss": 0.3891, + "step": 2765 + }, + { + "epoch": 1.16, + "learning_rate": 1.9317353572956058e-05, + "loss": 0.4175, + "step": 2766 + }, + { + "epoch": 1.16, + "learning_rate": 1.931673777162032e-05, + "loss": 0.3745, + "step": 2767 + }, + { + "epoch": 1.16, + "learning_rate": 1.9316121702483218e-05, + "loss": 0.421, + "step": 2768 + }, + { + "epoch": 1.16, + "learning_rate": 1.9315505365562464e-05, + "loss": 0.3868, + "step": 2769 + }, + { + "epoch": 1.16, + "learning_rate": 1.931488876087578e-05, + "loss": 0.3765, + "step": 2770 + }, + { + "epoch": 1.16, + "learning_rate": 1.9314271888440878e-05, + "loss": 0.3419, + "step": 2771 + }, + { + "epoch": 1.16, + "learning_rate": 1.9313654748275497e-05, + "loss": 0.4157, + "step": 2772 + }, + { + "epoch": 1.16, + "learning_rate": 1.931303734039738e-05, + "loss": 0.4043, + "step": 2773 + }, + { + "epoch": 1.16, + "learning_rate": 1.9312419664824268e-05, + "loss": 0.4398, + "step": 2774 + }, + { + "epoch": 1.16, + "learning_rate": 1.9311801721573913e-05, + "loss": 0.39, + "step": 2775 + }, + { + "epoch": 1.16, + "learning_rate": 1.931118351066408e-05, + "loss": 0.3564, + "step": 2776 + }, + { + "epoch": 1.16, + "learning_rate": 1.9310565032112542e-05, + "loss": 0.3823, + "step": 2777 + }, + { + "epoch": 1.16, + "learning_rate": 1.9309946285937075e-05, + "loss": 0.3924, + "step": 2778 + }, + { + "epoch": 1.16, + "learning_rate": 1.9309327272155462e-05, + "loss": 0.3895, + "step": 2779 + }, + { + "epoch": 1.16, + "learning_rate": 1.9308707990785497e-05, + "loss": 0.3785, + "step": 2780 + }, + { + "epoch": 1.16, + "learning_rate": 1.9308088441844982e-05, + "loss": 0.3857, + "step": 2781 + }, + { + "epoch": 1.16, + "learning_rate": 1.9307468625351723e-05, + "loss": 0.4335, + "step": 2782 + }, + { + "epoch": 1.17, + "learning_rate": 1.930684854132354e-05, + "loss": 0.3893, + "step": 2783 + }, + { + "epoch": 1.17, + "learning_rate": 1.9306228189778255e-05, + "loss": 0.4093, + "step": 2784 + }, + { + "epoch": 1.17, + "learning_rate": 1.9305607570733696e-05, + "loss": 0.4062, + "step": 2785 + }, + { + "epoch": 1.17, + "learning_rate": 1.930498668420771e-05, + "loss": 0.3321, + "step": 2786 + }, + { + "epoch": 1.17, + "learning_rate": 1.9304365530218134e-05, + "loss": 0.4056, + "step": 2787 + }, + { + "epoch": 1.17, + "learning_rate": 1.930374410878283e-05, + "loss": 0.4217, + "step": 2788 + }, + { + "epoch": 1.17, + "learning_rate": 1.9303122419919655e-05, + "loss": 0.3904, + "step": 2789 + }, + { + "epoch": 1.17, + "learning_rate": 1.9302500463646483e-05, + "loss": 0.3738, + "step": 2790 + }, + { + "epoch": 1.17, + "learning_rate": 1.930187823998119e-05, + "loss": 0.3978, + "step": 2791 + }, + { + "epoch": 1.17, + "learning_rate": 1.930125574894166e-05, + "loss": 0.3531, + "step": 2792 + }, + { + "epoch": 1.17, + "learning_rate": 1.930063299054579e-05, + "loss": 0.3876, + "step": 2793 + }, + { + "epoch": 1.17, + "learning_rate": 1.9300009964811475e-05, + "loss": 0.4144, + "step": 2794 + }, + { + "epoch": 1.17, + "learning_rate": 1.9299386671756628e-05, + "loss": 0.4041, + "step": 2795 + }, + { + "epoch": 1.17, + "learning_rate": 1.9298763111399162e-05, + "loss": 0.4336, + "step": 2796 + }, + { + "epoch": 1.17, + "learning_rate": 1.9298139283757004e-05, + "loss": 0.3942, + "step": 2797 + }, + { + "epoch": 1.17, + "learning_rate": 1.9297515188848083e-05, + "loss": 0.3601, + "step": 2798 + }, + { + "epoch": 1.17, + "learning_rate": 1.9296890826690337e-05, + "loss": 0.3929, + "step": 2799 + }, + { + "epoch": 1.17, + "learning_rate": 1.9296266197301716e-05, + "loss": 0.4062, + "step": 2800 + }, + { + "epoch": 1.17, + "learning_rate": 1.9295641300700173e-05, + "loss": 0.354, + "step": 2801 + }, + { + "epoch": 1.17, + "learning_rate": 1.9295016136903666e-05, + "loss": 0.4031, + "step": 2802 + }, + { + "epoch": 1.17, + "learning_rate": 1.929439070593017e-05, + "loss": 0.4086, + "step": 2803 + }, + { + "epoch": 1.17, + "learning_rate": 1.9293765007797662e-05, + "loss": 0.4178, + "step": 2804 + }, + { + "epoch": 1.17, + "learning_rate": 1.9293139042524124e-05, + "loss": 0.3496, + "step": 2805 + }, + { + "epoch": 1.17, + "learning_rate": 1.9292512810127554e-05, + "loss": 0.397, + "step": 2806 + }, + { + "epoch": 1.18, + "learning_rate": 1.9291886310625948e-05, + "loss": 0.3759, + "step": 2807 + }, + { + "epoch": 1.18, + "learning_rate": 1.9291259544037313e-05, + "loss": 0.3958, + "step": 2808 + }, + { + "epoch": 1.18, + "learning_rate": 1.9290632510379667e-05, + "loss": 0.404, + "step": 2809 + }, + { + "epoch": 1.18, + "learning_rate": 1.9290005209671036e-05, + "loss": 0.394, + "step": 2810 + }, + { + "epoch": 1.18, + "learning_rate": 1.928937764192945e-05, + "loss": 0.4051, + "step": 2811 + }, + { + "epoch": 1.18, + "learning_rate": 1.9288749807172947e-05, + "loss": 0.4203, + "step": 2812 + }, + { + "epoch": 1.18, + "learning_rate": 1.9288121705419572e-05, + "loss": 0.3902, + "step": 2813 + }, + { + "epoch": 1.18, + "learning_rate": 1.928749333668738e-05, + "loss": 0.3657, + "step": 2814 + }, + { + "epoch": 1.18, + "learning_rate": 1.9286864700994433e-05, + "loss": 0.3203, + "step": 2815 + }, + { + "epoch": 1.18, + "learning_rate": 1.9286235798358803e-05, + "loss": 0.3667, + "step": 2816 + }, + { + "epoch": 1.18, + "learning_rate": 1.9285606628798564e-05, + "loss": 0.4075, + "step": 2817 + }, + { + "epoch": 1.18, + "learning_rate": 1.9284977192331807e-05, + "loss": 0.4056, + "step": 2818 + }, + { + "epoch": 1.18, + "learning_rate": 1.9284347488976616e-05, + "loss": 0.4313, + "step": 2819 + }, + { + "epoch": 1.18, + "learning_rate": 1.9283717518751095e-05, + "loss": 0.3834, + "step": 2820 + }, + { + "epoch": 1.18, + "learning_rate": 1.9283087281673352e-05, + "loss": 0.3633, + "step": 2821 + }, + { + "epoch": 1.18, + "learning_rate": 1.9282456777761505e-05, + "loss": 0.3872, + "step": 2822 + }, + { + "epoch": 1.18, + "learning_rate": 1.9281826007033674e-05, + "loss": 0.4124, + "step": 2823 + }, + { + "epoch": 1.18, + "learning_rate": 1.928119496950799e-05, + "loss": 0.4013, + "step": 2824 + }, + { + "epoch": 1.18, + "learning_rate": 1.9280563665202595e-05, + "loss": 0.4354, + "step": 2825 + }, + { + "epoch": 1.18, + "learning_rate": 1.927993209413563e-05, + "loss": 0.3694, + "step": 2826 + }, + { + "epoch": 1.18, + "learning_rate": 1.927930025632525e-05, + "loss": 0.3888, + "step": 2827 + }, + { + "epoch": 1.18, + "learning_rate": 1.9278668151789627e-05, + "loss": 0.4104, + "step": 2828 + }, + { + "epoch": 1.18, + "learning_rate": 1.9278035780546912e-05, + "loss": 0.4169, + "step": 2829 + }, + { + "epoch": 1.18, + "learning_rate": 1.9277403142615297e-05, + "loss": 0.385, + "step": 2830 + }, + { + "epoch": 1.19, + "learning_rate": 1.927677023801296e-05, + "loss": 0.4508, + "step": 2831 + }, + { + "epoch": 1.19, + "learning_rate": 1.9276137066758097e-05, + "loss": 0.4015, + "step": 2832 + }, + { + "epoch": 1.19, + "learning_rate": 1.9275503628868902e-05, + "loss": 0.4012, + "step": 2833 + }, + { + "epoch": 1.19, + "learning_rate": 1.9274869924363585e-05, + "loss": 0.3566, + "step": 2834 + }, + { + "epoch": 1.19, + "learning_rate": 1.9274235953260367e-05, + "loss": 0.3656, + "step": 2835 + }, + { + "epoch": 1.19, + "learning_rate": 1.9273601715577463e-05, + "loss": 0.3389, + "step": 2836 + }, + { + "epoch": 1.19, + "learning_rate": 1.9272967211333107e-05, + "loss": 0.3845, + "step": 2837 + }, + { + "epoch": 1.19, + "learning_rate": 1.927233244054554e-05, + "loss": 0.372, + "step": 2838 + }, + { + "epoch": 1.19, + "learning_rate": 1.9271697403233e-05, + "loss": 0.4459, + "step": 2839 + }, + { + "epoch": 1.19, + "learning_rate": 1.9271062099413752e-05, + "loss": 0.3977, + "step": 2840 + }, + { + "epoch": 1.19, + "learning_rate": 1.9270426529106048e-05, + "loss": 0.3697, + "step": 2841 + }, + { + "epoch": 1.19, + "learning_rate": 1.926979069232816e-05, + "loss": 0.3596, + "step": 2842 + }, + { + "epoch": 1.19, + "learning_rate": 1.9269154589098363e-05, + "loss": 0.388, + "step": 2843 + }, + { + "epoch": 1.19, + "learning_rate": 1.9268518219434943e-05, + "loss": 0.4069, + "step": 2844 + }, + { + "epoch": 1.19, + "learning_rate": 1.9267881583356192e-05, + "loss": 0.367, + "step": 2845 + }, + { + "epoch": 1.19, + "learning_rate": 1.926724468088041e-05, + "loss": 0.3711, + "step": 2846 + }, + { + "epoch": 1.19, + "learning_rate": 1.9266607512025903e-05, + "loss": 0.3994, + "step": 2847 + }, + { + "epoch": 1.19, + "learning_rate": 1.9265970076810987e-05, + "loss": 0.4422, + "step": 2848 + }, + { + "epoch": 1.19, + "learning_rate": 1.9265332375253984e-05, + "loss": 0.3868, + "step": 2849 + }, + { + "epoch": 1.19, + "learning_rate": 1.926469440737322e-05, + "loss": 0.4381, + "step": 2850 + }, + { + "epoch": 1.19, + "learning_rate": 1.9264056173187037e-05, + "loss": 0.4296, + "step": 2851 + }, + { + "epoch": 1.19, + "learning_rate": 1.9263417672713786e-05, + "loss": 0.3709, + "step": 2852 + }, + { + "epoch": 1.19, + "learning_rate": 1.9262778905971808e-05, + "loss": 0.4049, + "step": 2853 + }, + { + "epoch": 1.19, + "learning_rate": 1.9262139872979474e-05, + "loss": 0.3886, + "step": 2854 + }, + { + "epoch": 1.2, + "learning_rate": 1.9261500573755146e-05, + "loss": 0.3822, + "step": 2855 + }, + { + "epoch": 1.2, + "learning_rate": 1.9260861008317203e-05, + "loss": 0.3676, + "step": 2856 + }, + { + "epoch": 1.2, + "learning_rate": 1.9260221176684025e-05, + "loss": 0.386, + "step": 2857 + }, + { + "epoch": 1.2, + "learning_rate": 1.925958107887401e-05, + "loss": 0.3864, + "step": 2858 + }, + { + "epoch": 1.2, + "learning_rate": 1.9258940714905554e-05, + "loss": 0.3853, + "step": 2859 + }, + { + "epoch": 1.2, + "learning_rate": 1.9258300084797063e-05, + "loss": 0.4195, + "step": 2860 + }, + { + "epoch": 1.2, + "learning_rate": 1.925765918856695e-05, + "loss": 0.3893, + "step": 2861 + }, + { + "epoch": 1.2, + "learning_rate": 1.9257018026233642e-05, + "loss": 0.4311, + "step": 2862 + }, + { + "epoch": 1.2, + "learning_rate": 1.9256376597815565e-05, + "loss": 0.3936, + "step": 2863 + }, + { + "epoch": 1.2, + "learning_rate": 1.9255734903331157e-05, + "loss": 0.4135, + "step": 2864 + }, + { + "epoch": 1.2, + "learning_rate": 1.9255092942798858e-05, + "loss": 0.3841, + "step": 2865 + }, + { + "epoch": 1.2, + "learning_rate": 1.9254450716237128e-05, + "loss": 0.394, + "step": 2866 + }, + { + "epoch": 1.2, + "learning_rate": 1.9253808223664427e-05, + "loss": 0.4184, + "step": 2867 + }, + { + "epoch": 1.2, + "learning_rate": 1.9253165465099217e-05, + "loss": 0.3678, + "step": 2868 + }, + { + "epoch": 1.2, + "learning_rate": 1.925252244055998e-05, + "loss": 0.4486, + "step": 2869 + }, + { + "epoch": 1.2, + "learning_rate": 1.9251879150065192e-05, + "loss": 0.3573, + "step": 2870 + }, + { + "epoch": 1.2, + "learning_rate": 1.925123559363335e-05, + "loss": 0.3508, + "step": 2871 + }, + { + "epoch": 1.2, + "learning_rate": 1.925059177128295e-05, + "loss": 0.3864, + "step": 2872 + }, + { + "epoch": 1.2, + "learning_rate": 1.9249947683032497e-05, + "loss": 0.3628, + "step": 2873 + }, + { + "epoch": 1.2, + "learning_rate": 1.9249303328900505e-05, + "loss": 0.3605, + "step": 2874 + }, + { + "epoch": 1.2, + "learning_rate": 1.92486587089055e-05, + "loss": 0.391, + "step": 2875 + }, + { + "epoch": 1.2, + "learning_rate": 1.9248013823066006e-05, + "loss": 0.4173, + "step": 2876 + }, + { + "epoch": 1.2, + "learning_rate": 1.924736867140056e-05, + "loss": 0.413, + "step": 2877 + }, + { + "epoch": 1.2, + "learning_rate": 1.9246723253927714e-05, + "loss": 0.4276, + "step": 2878 + }, + { + "epoch": 1.21, + "learning_rate": 1.9246077570666008e-05, + "loss": 0.4009, + "step": 2879 + }, + { + "epoch": 1.21, + "learning_rate": 1.9245431621634006e-05, + "loss": 0.3918, + "step": 2880 + }, + { + "epoch": 1.21, + "learning_rate": 1.924478540685028e-05, + "loss": 0.3812, + "step": 2881 + }, + { + "epoch": 1.21, + "learning_rate": 1.9244138926333396e-05, + "loss": 0.4158, + "step": 2882 + }, + { + "epoch": 1.21, + "learning_rate": 1.9243492180101947e-05, + "loss": 0.4269, + "step": 2883 + }, + { + "epoch": 1.21, + "learning_rate": 1.9242845168174514e-05, + "loss": 0.3786, + "step": 2884 + }, + { + "epoch": 1.21, + "learning_rate": 1.92421978905697e-05, + "loss": 0.3487, + "step": 2885 + }, + { + "epoch": 1.21, + "learning_rate": 1.924155034730611e-05, + "loss": 0.3736, + "step": 2886 + }, + { + "epoch": 1.21, + "learning_rate": 1.9240902538402354e-05, + "loss": 0.3477, + "step": 2887 + }, + { + "epoch": 1.21, + "learning_rate": 1.9240254463877055e-05, + "loss": 0.3401, + "step": 2888 + }, + { + "epoch": 1.21, + "learning_rate": 1.9239606123748845e-05, + "loss": 0.3752, + "step": 2889 + }, + { + "epoch": 1.21, + "learning_rate": 1.923895751803635e-05, + "loss": 0.37, + "step": 2890 + }, + { + "epoch": 1.21, + "learning_rate": 1.9238308646758224e-05, + "loss": 0.3883, + "step": 2891 + }, + { + "epoch": 1.21, + "learning_rate": 1.9237659509933114e-05, + "loss": 0.3907, + "step": 2892 + }, + { + "epoch": 1.21, + "learning_rate": 1.9237010107579676e-05, + "loss": 0.3955, + "step": 2893 + }, + { + "epoch": 1.21, + "learning_rate": 1.923636043971658e-05, + "loss": 0.3549, + "step": 2894 + }, + { + "epoch": 1.21, + "learning_rate": 1.92357105063625e-05, + "loss": 0.3955, + "step": 2895 + }, + { + "epoch": 1.21, + "learning_rate": 1.9235060307536117e-05, + "loss": 0.3917, + "step": 2896 + }, + { + "epoch": 1.21, + "learning_rate": 1.923440984325612e-05, + "loss": 0.3484, + "step": 2897 + }, + { + "epoch": 1.21, + "learning_rate": 1.923375911354121e-05, + "loss": 0.4145, + "step": 2898 + }, + { + "epoch": 1.21, + "learning_rate": 1.9233108118410082e-05, + "loss": 0.3835, + "step": 2899 + }, + { + "epoch": 1.21, + "learning_rate": 1.923245685788146e-05, + "loss": 0.4056, + "step": 2900 + }, + { + "epoch": 1.21, + "learning_rate": 1.9231805331974053e-05, + "loss": 0.3901, + "step": 2901 + }, + { + "epoch": 1.21, + "learning_rate": 1.9231153540706597e-05, + "loss": 0.4369, + "step": 2902 + }, + { + "epoch": 1.22, + "learning_rate": 1.9230501484097826e-05, + "loss": 0.379, + "step": 2903 + }, + { + "epoch": 1.22, + "learning_rate": 1.9229849162166478e-05, + "loss": 0.3816, + "step": 2904 + }, + { + "epoch": 1.22, + "learning_rate": 1.922919657493131e-05, + "loss": 0.4186, + "step": 2905 + }, + { + "epoch": 1.22, + "learning_rate": 1.922854372241107e-05, + "loss": 0.4063, + "step": 2906 + }, + { + "epoch": 1.22, + "learning_rate": 1.9227890604624534e-05, + "loss": 0.4478, + "step": 2907 + }, + { + "epoch": 1.22, + "learning_rate": 1.9227237221590472e-05, + "loss": 0.4357, + "step": 2908 + }, + { + "epoch": 1.22, + "learning_rate": 1.922658357332766e-05, + "loss": 0.3672, + "step": 2909 + }, + { + "epoch": 1.22, + "learning_rate": 1.9225929659854893e-05, + "loss": 0.3548, + "step": 2910 + }, + { + "epoch": 1.22, + "learning_rate": 1.922527548119097e-05, + "loss": 0.3264, + "step": 2911 + }, + { + "epoch": 1.22, + "learning_rate": 1.922462103735468e-05, + "loss": 0.3426, + "step": 2912 + }, + { + "epoch": 1.22, + "learning_rate": 1.9223966328364853e-05, + "loss": 0.3965, + "step": 2913 + }, + { + "epoch": 1.22, + "learning_rate": 1.922331135424029e-05, + "loss": 0.4179, + "step": 2914 + }, + { + "epoch": 1.22, + "learning_rate": 1.9222656114999835e-05, + "loss": 0.372, + "step": 2915 + }, + { + "epoch": 1.22, + "learning_rate": 1.922200061066231e-05, + "loss": 0.4026, + "step": 2916 + }, + { + "epoch": 1.22, + "learning_rate": 1.9221344841246562e-05, + "loss": 0.356, + "step": 2917 + }, + { + "epoch": 1.22, + "learning_rate": 1.9220688806771443e-05, + "loss": 0.3802, + "step": 2918 + }, + { + "epoch": 1.22, + "learning_rate": 1.92200325072558e-05, + "loss": 0.4339, + "step": 2919 + }, + { + "epoch": 1.22, + "learning_rate": 1.9219375942718508e-05, + "loss": 0.3936, + "step": 2920 + }, + { + "epoch": 1.22, + "learning_rate": 1.9218719113178433e-05, + "loss": 0.3785, + "step": 2921 + }, + { + "epoch": 1.22, + "learning_rate": 1.9218062018654458e-05, + "loss": 0.3854, + "step": 2922 + }, + { + "epoch": 1.22, + "learning_rate": 1.9217404659165476e-05, + "loss": 0.3722, + "step": 2923 + }, + { + "epoch": 1.22, + "learning_rate": 1.921674703473037e-05, + "loss": 0.3839, + "step": 2924 + }, + { + "epoch": 1.22, + "learning_rate": 1.9216089145368052e-05, + "loss": 0.4487, + "step": 2925 + }, + { + "epoch": 1.23, + "learning_rate": 1.921543099109743e-05, + "loss": 0.3922, + "step": 2926 + }, + { + "epoch": 1.23, + "learning_rate": 1.921477257193742e-05, + "loss": 0.393, + "step": 2927 + }, + { + "epoch": 1.23, + "learning_rate": 1.9214113887906947e-05, + "loss": 0.3938, + "step": 2928 + }, + { + "epoch": 1.23, + "learning_rate": 1.9213454939024953e-05, + "loss": 0.3843, + "step": 2929 + }, + { + "epoch": 1.23, + "learning_rate": 1.9212795725310372e-05, + "loss": 0.3286, + "step": 2930 + }, + { + "epoch": 1.23, + "learning_rate": 1.921213624678215e-05, + "loss": 0.3765, + "step": 2931 + }, + { + "epoch": 1.23, + "learning_rate": 1.921147650345925e-05, + "loss": 0.3984, + "step": 2932 + }, + { + "epoch": 1.23, + "learning_rate": 1.9210816495360625e-05, + "loss": 0.4522, + "step": 2933 + }, + { + "epoch": 1.23, + "learning_rate": 1.921015622250526e-05, + "loss": 0.3989, + "step": 2934 + }, + { + "epoch": 1.23, + "learning_rate": 1.920949568491212e-05, + "loss": 0.4184, + "step": 2935 + }, + { + "epoch": 1.23, + "learning_rate": 1.9208834882600207e-05, + "loss": 0.3787, + "step": 2936 + }, + { + "epoch": 1.23, + "learning_rate": 1.9208173815588506e-05, + "loss": 0.4151, + "step": 2937 + }, + { + "epoch": 1.23, + "learning_rate": 1.9207512483896016e-05, + "loss": 0.4006, + "step": 2938 + }, + { + "epoch": 1.23, + "learning_rate": 1.920685088754175e-05, + "loss": 0.3855, + "step": 2939 + }, + { + "epoch": 1.23, + "learning_rate": 1.9206189026544728e-05, + "loss": 0.417, + "step": 2940 + }, + { + "epoch": 1.23, + "learning_rate": 1.920552690092397e-05, + "loss": 0.3611, + "step": 2941 + }, + { + "epoch": 1.23, + "learning_rate": 1.9204864510698513e-05, + "loss": 0.3916, + "step": 2942 + }, + { + "epoch": 1.23, + "learning_rate": 1.9204201855887394e-05, + "loss": 0.3599, + "step": 2943 + }, + { + "epoch": 1.23, + "learning_rate": 1.920353893650966e-05, + "loss": 0.3919, + "step": 2944 + }, + { + "epoch": 1.23, + "learning_rate": 1.9202875752584363e-05, + "loss": 0.3776, + "step": 2945 + }, + { + "epoch": 1.23, + "learning_rate": 1.920221230413057e-05, + "loss": 0.41, + "step": 2946 + }, + { + "epoch": 1.23, + "learning_rate": 1.920154859116735e-05, + "loss": 0.4023, + "step": 2947 + }, + { + "epoch": 1.23, + "learning_rate": 1.9200884613713782e-05, + "loss": 0.3815, + "step": 2948 + }, + { + "epoch": 1.23, + "learning_rate": 1.9200220371788952e-05, + "loss": 0.3629, + "step": 2949 + }, + { + "epoch": 1.24, + "learning_rate": 1.9199555865411948e-05, + "loss": 0.4138, + "step": 2950 + }, + { + "epoch": 1.24, + "learning_rate": 1.9198891094601876e-05, + "loss": 0.3576, + "step": 2951 + }, + { + "epoch": 1.24, + "learning_rate": 1.9198226059377846e-05, + "loss": 0.3736, + "step": 2952 + }, + { + "epoch": 1.24, + "learning_rate": 1.9197560759758963e-05, + "loss": 0.4181, + "step": 2953 + }, + { + "epoch": 1.24, + "learning_rate": 1.9196895195764363e-05, + "loss": 0.3947, + "step": 2954 + }, + { + "epoch": 1.24, + "learning_rate": 1.9196229367413172e-05, + "loss": 0.3668, + "step": 2955 + }, + { + "epoch": 1.24, + "learning_rate": 1.9195563274724523e-05, + "loss": 0.3783, + "step": 2956 + }, + { + "epoch": 1.24, + "learning_rate": 1.9194896917717573e-05, + "loss": 0.3779, + "step": 2957 + }, + { + "epoch": 1.24, + "learning_rate": 1.919423029641147e-05, + "loss": 0.3688, + "step": 2958 + }, + { + "epoch": 1.24, + "learning_rate": 1.9193563410825376e-05, + "loss": 0.3632, + "step": 2959 + }, + { + "epoch": 1.24, + "learning_rate": 1.9192896260978457e-05, + "loss": 0.4211, + "step": 2960 + }, + { + "epoch": 1.24, + "learning_rate": 1.9192228846889896e-05, + "loss": 0.4101, + "step": 2961 + }, + { + "epoch": 1.24, + "learning_rate": 1.9191561168578873e-05, + "loss": 0.3919, + "step": 2962 + }, + { + "epoch": 1.24, + "learning_rate": 1.919089322606458e-05, + "loss": 0.3991, + "step": 2963 + }, + { + "epoch": 1.24, + "learning_rate": 1.919022501936622e-05, + "loss": 0.3567, + "step": 2964 + }, + { + "epoch": 1.24, + "learning_rate": 1.918955654850299e-05, + "loss": 0.3857, + "step": 2965 + }, + { + "epoch": 1.24, + "learning_rate": 1.9188887813494118e-05, + "loss": 0.4223, + "step": 2966 + }, + { + "epoch": 1.24, + "learning_rate": 1.9188218814358817e-05, + "loss": 0.4015, + "step": 2967 + }, + { + "epoch": 1.24, + "learning_rate": 1.9187549551116322e-05, + "loss": 0.4114, + "step": 2968 + }, + { + "epoch": 1.24, + "learning_rate": 1.9186880023785864e-05, + "loss": 0.4167, + "step": 2969 + }, + { + "epoch": 1.24, + "learning_rate": 1.9186210232386694e-05, + "loss": 0.3801, + "step": 2970 + }, + { + "epoch": 1.24, + "learning_rate": 1.9185540176938065e-05, + "loss": 0.3913, + "step": 2971 + }, + { + "epoch": 1.24, + "learning_rate": 1.9184869857459233e-05, + "loss": 0.3867, + "step": 2972 + }, + { + "epoch": 1.24, + "learning_rate": 1.9184199273969465e-05, + "loss": 0.3883, + "step": 2973 + }, + { + "epoch": 1.25, + "learning_rate": 1.918352842648804e-05, + "loss": 0.4711, + "step": 2974 + }, + { + "epoch": 1.25, + "learning_rate": 1.918285731503424e-05, + "loss": 0.4078, + "step": 2975 + }, + { + "epoch": 1.25, + "learning_rate": 1.9182185939627356e-05, + "loss": 0.3993, + "step": 2976 + }, + { + "epoch": 1.25, + "learning_rate": 1.918151430028669e-05, + "loss": 0.3231, + "step": 2977 + }, + { + "epoch": 1.25, + "learning_rate": 1.9180842397031535e-05, + "loss": 0.4097, + "step": 2978 + }, + { + "epoch": 1.25, + "learning_rate": 1.9180170229881216e-05, + "loss": 0.3751, + "step": 2979 + }, + { + "epoch": 1.25, + "learning_rate": 1.9179497798855048e-05, + "loss": 0.3605, + "step": 2980 + }, + { + "epoch": 1.25, + "learning_rate": 1.9178825103972364e-05, + "loss": 0.3395, + "step": 2981 + }, + { + "epoch": 1.25, + "learning_rate": 1.9178152145252497e-05, + "loss": 0.3948, + "step": 2982 + }, + { + "epoch": 1.25, + "learning_rate": 1.917747892271479e-05, + "loss": 0.3671, + "step": 2983 + }, + { + "epoch": 1.25, + "learning_rate": 1.9176805436378597e-05, + "loss": 0.3953, + "step": 2984 + }, + { + "epoch": 1.25, + "learning_rate": 1.9176131686263274e-05, + "loss": 0.4075, + "step": 2985 + }, + { + "epoch": 1.25, + "learning_rate": 1.917545767238819e-05, + "loss": 0.3921, + "step": 2986 + }, + { + "epoch": 1.25, + "learning_rate": 1.9174783394772714e-05, + "loss": 0.4225, + "step": 2987 + }, + { + "epoch": 1.25, + "learning_rate": 1.9174108853436234e-05, + "loss": 0.3768, + "step": 2988 + }, + { + "epoch": 1.25, + "learning_rate": 1.9173434048398133e-05, + "loss": 0.3695, + "step": 2989 + }, + { + "epoch": 1.25, + "learning_rate": 1.917275897967781e-05, + "loss": 0.4141, + "step": 2990 + }, + { + "epoch": 1.25, + "learning_rate": 1.9172083647294676e-05, + "loss": 0.4165, + "step": 2991 + }, + { + "epoch": 1.25, + "learning_rate": 1.9171408051268132e-05, + "loss": 0.4475, + "step": 2992 + }, + { + "epoch": 1.25, + "learning_rate": 1.9170732191617603e-05, + "loss": 0.3717, + "step": 2993 + }, + { + "epoch": 1.25, + "learning_rate": 1.9170056068362516e-05, + "loss": 0.4643, + "step": 2994 + }, + { + "epoch": 1.25, + "learning_rate": 1.9169379681522305e-05, + "loss": 0.3985, + "step": 2995 + }, + { + "epoch": 1.25, + "learning_rate": 1.9168703031116412e-05, + "loss": 0.3602, + "step": 2996 + }, + { + "epoch": 1.25, + "learning_rate": 1.9168026117164285e-05, + "loss": 0.3634, + "step": 2997 + }, + { + "epoch": 1.26, + "learning_rate": 1.916734893968538e-05, + "loss": 0.4244, + "step": 2998 + }, + { + "epoch": 1.26, + "learning_rate": 1.916667149869917e-05, + "loss": 0.3957, + "step": 2999 + }, + { + "epoch": 1.26, + "learning_rate": 1.916599379422512e-05, + "loss": 0.3868, + "step": 3000 + }, + { + "epoch": 1.26, + "learning_rate": 1.9165315826282712e-05, + "loss": 0.3846, + "step": 3001 + }, + { + "epoch": 1.26, + "learning_rate": 1.916463759489143e-05, + "loss": 0.4195, + "step": 3002 + }, + { + "epoch": 1.26, + "learning_rate": 1.916395910007078e-05, + "loss": 0.4048, + "step": 3003 + }, + { + "epoch": 1.26, + "learning_rate": 1.9163280341840253e-05, + "loss": 0.3728, + "step": 3004 + }, + { + "epoch": 1.26, + "learning_rate": 1.9162601320219364e-05, + "loss": 0.419, + "step": 3005 + }, + { + "epoch": 1.26, + "learning_rate": 1.9161922035227633e-05, + "loss": 0.3842, + "step": 3006 + }, + { + "epoch": 1.26, + "learning_rate": 1.9161242486884584e-05, + "loss": 0.4437, + "step": 3007 + }, + { + "epoch": 1.26, + "learning_rate": 1.9160562675209746e-05, + "loss": 0.4448, + "step": 3008 + }, + { + "epoch": 1.26, + "learning_rate": 1.9159882600222664e-05, + "loss": 0.4838, + "step": 3009 + }, + { + "epoch": 1.26, + "learning_rate": 1.9159202261942886e-05, + "loss": 0.3661, + "step": 3010 + }, + { + "epoch": 1.26, + "learning_rate": 1.9158521660389966e-05, + "loss": 0.3588, + "step": 3011 + }, + { + "epoch": 1.26, + "learning_rate": 1.9157840795583466e-05, + "loss": 0.452, + "step": 3012 + }, + { + "epoch": 1.26, + "learning_rate": 1.9157159667542964e-05, + "loss": 0.3669, + "step": 3013 + }, + { + "epoch": 1.26, + "learning_rate": 1.9156478276288027e-05, + "loss": 0.4242, + "step": 3014 + }, + { + "epoch": 1.26, + "learning_rate": 1.9155796621838253e-05, + "loss": 0.3988, + "step": 3015 + }, + { + "epoch": 1.26, + "learning_rate": 1.915511470421323e-05, + "loss": 0.3667, + "step": 3016 + }, + { + "epoch": 1.26, + "learning_rate": 1.9154432523432553e-05, + "loss": 0.4029, + "step": 3017 + }, + { + "epoch": 1.26, + "learning_rate": 1.915375007951584e-05, + "loss": 0.3726, + "step": 3018 + }, + { + "epoch": 1.26, + "learning_rate": 1.9153067372482702e-05, + "loss": 0.4151, + "step": 3019 + }, + { + "epoch": 1.26, + "learning_rate": 1.915238440235277e-05, + "loss": 0.3597, + "step": 3020 + }, + { + "epoch": 1.26, + "learning_rate": 1.9151701169145664e-05, + "loss": 0.3604, + "step": 3021 + }, + { + "epoch": 1.27, + "learning_rate": 1.9151017672881032e-05, + "loss": 0.4356, + "step": 3022 + }, + { + "epoch": 1.27, + "learning_rate": 1.9150333913578518e-05, + "loss": 0.4045, + "step": 3023 + }, + { + "epoch": 1.27, + "learning_rate": 1.9149649891257772e-05, + "loss": 0.379, + "step": 3024 + }, + { + "epoch": 1.27, + "learning_rate": 1.9148965605938463e-05, + "loss": 0.3753, + "step": 3025 + }, + { + "epoch": 1.27, + "learning_rate": 1.9148281057640252e-05, + "loss": 0.4016, + "step": 3026 + }, + { + "epoch": 1.27, + "learning_rate": 1.914759624638282e-05, + "loss": 0.3454, + "step": 3027 + }, + { + "epoch": 1.27, + "learning_rate": 1.914691117218586e-05, + "loss": 0.3758, + "step": 3028 + }, + { + "epoch": 1.27, + "learning_rate": 1.914622583506905e-05, + "loss": 0.449, + "step": 3029 + }, + { + "epoch": 1.27, + "learning_rate": 1.914554023505209e-05, + "loss": 0.3803, + "step": 3030 + }, + { + "epoch": 1.27, + "learning_rate": 1.9144854372154697e-05, + "loss": 0.3536, + "step": 3031 + }, + { + "epoch": 1.27, + "learning_rate": 1.9144168246396576e-05, + "loss": 0.4062, + "step": 3032 + }, + { + "epoch": 1.27, + "learning_rate": 1.914348185779746e-05, + "loss": 0.3956, + "step": 3033 + }, + { + "epoch": 1.27, + "learning_rate": 1.9142795206377066e-05, + "loss": 0.4077, + "step": 3034 + }, + { + "epoch": 1.27, + "learning_rate": 1.914210829215514e-05, + "loss": 0.4146, + "step": 3035 + }, + { + "epoch": 1.27, + "learning_rate": 1.914142111515142e-05, + "loss": 0.4158, + "step": 3036 + }, + { + "epoch": 1.27, + "learning_rate": 1.9140733675385666e-05, + "loss": 0.4014, + "step": 3037 + }, + { + "epoch": 1.27, + "learning_rate": 1.914004597287763e-05, + "loss": 0.3813, + "step": 3038 + }, + { + "epoch": 1.27, + "learning_rate": 1.9139358007647085e-05, + "loss": 0.3681, + "step": 3039 + }, + { + "epoch": 1.27, + "learning_rate": 1.9138669779713807e-05, + "loss": 0.409, + "step": 3040 + }, + { + "epoch": 1.27, + "learning_rate": 1.9137981289097573e-05, + "loss": 0.4381, + "step": 3041 + }, + { + "epoch": 1.27, + "learning_rate": 1.9137292535818174e-05, + "loss": 0.3638, + "step": 3042 + }, + { + "epoch": 1.27, + "learning_rate": 1.913660351989541e-05, + "loss": 0.4291, + "step": 3043 + }, + { + "epoch": 1.27, + "learning_rate": 1.9135914241349088e-05, + "loss": 0.4002, + "step": 3044 + }, + { + "epoch": 1.27, + "learning_rate": 1.9135224700199015e-05, + "loss": 0.3697, + "step": 3045 + }, + { + "epoch": 1.28, + "learning_rate": 1.913453489646502e-05, + "loss": 0.4188, + "step": 3046 + }, + { + "epoch": 1.28, + "learning_rate": 1.913384483016692e-05, + "loss": 0.4102, + "step": 3047 + }, + { + "epoch": 1.28, + "learning_rate": 1.9133154501324557e-05, + "loss": 0.3559, + "step": 3048 + }, + { + "epoch": 1.28, + "learning_rate": 1.9132463909957774e-05, + "loss": 0.4174, + "step": 3049 + }, + { + "epoch": 1.28, + "learning_rate": 1.9131773056086417e-05, + "loss": 0.4116, + "step": 3050 + }, + { + "epoch": 1.28, + "learning_rate": 1.9131081939730348e-05, + "loss": 0.4361, + "step": 3051 + }, + { + "epoch": 1.28, + "learning_rate": 1.913039056090943e-05, + "loss": 0.3952, + "step": 3052 + }, + { + "epoch": 1.28, + "learning_rate": 1.912969891964354e-05, + "loss": 0.3878, + "step": 3053 + }, + { + "epoch": 1.28, + "learning_rate": 1.9129007015952556e-05, + "loss": 0.36, + "step": 3054 + }, + { + "epoch": 1.28, + "learning_rate": 1.9128314849856365e-05, + "loss": 0.3476, + "step": 3055 + }, + { + "epoch": 1.28, + "learning_rate": 1.9127622421374866e-05, + "loss": 0.3948, + "step": 3056 + }, + { + "epoch": 1.28, + "learning_rate": 1.9126929730527957e-05, + "loss": 0.3804, + "step": 3057 + }, + { + "epoch": 1.28, + "learning_rate": 1.9126236777335554e-05, + "loss": 0.4198, + "step": 3058 + }, + { + "epoch": 1.28, + "learning_rate": 1.9125543561817573e-05, + "loss": 0.3692, + "step": 3059 + }, + { + "epoch": 1.28, + "learning_rate": 1.9124850083993942e-05, + "loss": 0.3862, + "step": 3060 + }, + { + "epoch": 1.28, + "learning_rate": 1.912415634388459e-05, + "loss": 0.4109, + "step": 3061 + }, + { + "epoch": 1.28, + "learning_rate": 1.912346234150946e-05, + "loss": 0.3838, + "step": 3062 + }, + { + "epoch": 1.28, + "learning_rate": 1.9122768076888506e-05, + "loss": 0.3874, + "step": 3063 + }, + { + "epoch": 1.28, + "learning_rate": 1.9122073550041675e-05, + "loss": 0.4108, + "step": 3064 + }, + { + "epoch": 1.28, + "learning_rate": 1.9121378760988934e-05, + "loss": 0.3832, + "step": 3065 + }, + { + "epoch": 1.28, + "learning_rate": 1.9120683709750257e-05, + "loss": 0.3902, + "step": 3066 + }, + { + "epoch": 1.28, + "learning_rate": 1.9119988396345615e-05, + "loss": 0.3786, + "step": 3067 + }, + { + "epoch": 1.28, + "learning_rate": 1.9119292820795007e-05, + "loss": 0.4007, + "step": 3068 + }, + { + "epoch": 1.28, + "learning_rate": 1.9118596983118415e-05, + "loss": 0.3548, + "step": 3069 + }, + { + "epoch": 1.29, + "learning_rate": 1.9117900883335846e-05, + "loss": 0.3979, + "step": 3070 + }, + { + "epoch": 1.29, + "learning_rate": 1.9117204521467308e-05, + "loss": 0.3693, + "step": 3071 + }, + { + "epoch": 1.29, + "learning_rate": 1.9116507897532813e-05, + "loss": 0.3875, + "step": 3072 + }, + { + "epoch": 1.29, + "learning_rate": 1.911581101155239e-05, + "loss": 0.3935, + "step": 3073 + }, + { + "epoch": 1.29, + "learning_rate": 1.911511386354607e-05, + "loss": 0.3525, + "step": 3074 + }, + { + "epoch": 1.29, + "learning_rate": 1.911441645353389e-05, + "loss": 0.3911, + "step": 3075 + }, + { + "epoch": 1.29, + "learning_rate": 1.9113718781535896e-05, + "loss": 0.3531, + "step": 3076 + }, + { + "epoch": 1.29, + "learning_rate": 1.9113020847572144e-05, + "loss": 0.4023, + "step": 3077 + }, + { + "epoch": 1.29, + "learning_rate": 1.911232265166269e-05, + "loss": 0.3949, + "step": 3078 + }, + { + "epoch": 1.29, + "learning_rate": 1.9111624193827614e-05, + "loss": 0.3762, + "step": 3079 + }, + { + "epoch": 1.29, + "learning_rate": 1.9110925474086982e-05, + "loss": 0.3759, + "step": 3080 + }, + { + "epoch": 1.29, + "learning_rate": 1.9110226492460886e-05, + "loss": 0.4094, + "step": 3081 + }, + { + "epoch": 1.29, + "learning_rate": 1.9109527248969412e-05, + "loss": 0.4151, + "step": 3082 + }, + { + "epoch": 1.29, + "learning_rate": 1.910882774363266e-05, + "loss": 0.3605, + "step": 3083 + }, + { + "epoch": 1.29, + "learning_rate": 1.9108127976470737e-05, + "loss": 0.3758, + "step": 3084 + }, + { + "epoch": 1.29, + "learning_rate": 1.9107427947503757e-05, + "loss": 0.3833, + "step": 3085 + }, + { + "epoch": 1.29, + "learning_rate": 1.9106727656751846e-05, + "loss": 0.3986, + "step": 3086 + }, + { + "epoch": 1.29, + "learning_rate": 1.9106027104235126e-05, + "loss": 0.3984, + "step": 3087 + }, + { + "epoch": 1.29, + "learning_rate": 1.910532628997374e-05, + "loss": 0.4009, + "step": 3088 + }, + { + "epoch": 1.29, + "learning_rate": 1.910462521398783e-05, + "loss": 0.3672, + "step": 3089 + }, + { + "epoch": 1.29, + "learning_rate": 1.9103923876297544e-05, + "loss": 0.3765, + "step": 3090 + }, + { + "epoch": 1.29, + "learning_rate": 1.9103222276923047e-05, + "loss": 0.3638, + "step": 3091 + }, + { + "epoch": 1.29, + "learning_rate": 1.9102520415884504e-05, + "loss": 0.3996, + "step": 3092 + }, + { + "epoch": 1.29, + "learning_rate": 1.910181829320209e-05, + "loss": 0.3812, + "step": 3093 + }, + { + "epoch": 1.3, + "learning_rate": 1.9101115908895985e-05, + "loss": 0.3676, + "step": 3094 + }, + { + "epoch": 1.3, + "learning_rate": 1.9100413262986383e-05, + "loss": 0.3927, + "step": 3095 + }, + { + "epoch": 1.3, + "learning_rate": 1.909971035549347e-05, + "loss": 0.4282, + "step": 3096 + }, + { + "epoch": 1.3, + "learning_rate": 1.9099007186437465e-05, + "loss": 0.4489, + "step": 3097 + }, + { + "epoch": 1.3, + "learning_rate": 1.9098303755838566e-05, + "loss": 0.3486, + "step": 3098 + }, + { + "epoch": 1.3, + "learning_rate": 1.9097600063717005e-05, + "loss": 0.3687, + "step": 3099 + }, + { + "epoch": 1.3, + "learning_rate": 1.9096896110092998e-05, + "loss": 0.4529, + "step": 3100 + }, + { + "epoch": 1.3, + "learning_rate": 1.909619189498679e-05, + "loss": 0.3496, + "step": 3101 + }, + { + "epoch": 1.3, + "learning_rate": 1.9095487418418614e-05, + "loss": 0.4179, + "step": 3102 + }, + { + "epoch": 1.3, + "learning_rate": 1.9094782680408722e-05, + "loss": 0.4081, + "step": 3103 + }, + { + "epoch": 1.3, + "learning_rate": 1.9094077680977377e-05, + "loss": 0.3881, + "step": 3104 + }, + { + "epoch": 1.3, + "learning_rate": 1.9093372420144835e-05, + "loss": 0.4002, + "step": 3105 + }, + { + "epoch": 1.3, + "learning_rate": 1.9092666897931375e-05, + "loss": 0.3316, + "step": 3106 + }, + { + "epoch": 1.3, + "learning_rate": 1.9091961114357274e-05, + "loss": 0.4296, + "step": 3107 + }, + { + "epoch": 1.3, + "learning_rate": 1.909125506944282e-05, + "loss": 0.3749, + "step": 3108 + }, + { + "epoch": 1.3, + "learning_rate": 1.90905487632083e-05, + "loss": 0.427, + "step": 3109 + }, + { + "epoch": 1.3, + "learning_rate": 1.908984219567403e-05, + "loss": 0.386, + "step": 3110 + }, + { + "epoch": 1.3, + "learning_rate": 1.908913536686031e-05, + "loss": 0.4013, + "step": 3111 + }, + { + "epoch": 1.3, + "learning_rate": 1.9088428276787455e-05, + "loss": 0.3712, + "step": 3112 + }, + { + "epoch": 1.3, + "learning_rate": 1.90877209254758e-05, + "loss": 0.4584, + "step": 3113 + }, + { + "epoch": 1.3, + "learning_rate": 1.9087013312945668e-05, + "loss": 0.41, + "step": 3114 + }, + { + "epoch": 1.3, + "learning_rate": 1.9086305439217402e-05, + "loss": 0.3809, + "step": 3115 + }, + { + "epoch": 1.3, + "learning_rate": 1.908559730431135e-05, + "loss": 0.3908, + "step": 3116 + }, + { + "epoch": 1.31, + "learning_rate": 1.9084888908247864e-05, + "loss": 0.4066, + "step": 3117 + }, + { + "epoch": 1.31, + "learning_rate": 1.908418025104731e-05, + "loss": 0.4241, + "step": 3118 + }, + { + "epoch": 1.31, + "learning_rate": 1.9083471332730058e-05, + "loss": 0.3628, + "step": 3119 + }, + { + "epoch": 1.31, + "learning_rate": 1.908276215331648e-05, + "loss": 0.469, + "step": 3120 + }, + { + "epoch": 1.31, + "learning_rate": 1.9082052712826967e-05, + "loss": 0.3736, + "step": 3121 + }, + { + "epoch": 1.31, + "learning_rate": 1.9081343011281907e-05, + "loss": 0.3628, + "step": 3122 + }, + { + "epoch": 1.31, + "learning_rate": 1.9080633048701696e-05, + "loss": 0.436, + "step": 3123 + }, + { + "epoch": 1.31, + "learning_rate": 1.907992282510675e-05, + "loss": 0.4212, + "step": 3124 + }, + { + "epoch": 1.31, + "learning_rate": 1.907921234051748e-05, + "loss": 0.3986, + "step": 3125 + }, + { + "epoch": 1.31, + "learning_rate": 1.9078501594954305e-05, + "loss": 0.4126, + "step": 3126 + }, + { + "epoch": 1.31, + "learning_rate": 1.9077790588437663e-05, + "loss": 0.3649, + "step": 3127 + }, + { + "epoch": 1.31, + "learning_rate": 1.907707932098798e-05, + "loss": 0.3977, + "step": 3128 + }, + { + "epoch": 1.31, + "learning_rate": 1.9076367792625713e-05, + "loss": 0.368, + "step": 3129 + }, + { + "epoch": 1.31, + "learning_rate": 1.9075656003371304e-05, + "loss": 0.3954, + "step": 3130 + }, + { + "epoch": 1.31, + "learning_rate": 1.907494395324522e-05, + "loss": 0.3712, + "step": 3131 + }, + { + "epoch": 1.31, + "learning_rate": 1.907423164226792e-05, + "loss": 0.3879, + "step": 3132 + }, + { + "epoch": 1.31, + "learning_rate": 1.907351907045989e-05, + "loss": 0.375, + "step": 3133 + }, + { + "epoch": 1.31, + "learning_rate": 1.90728062378416e-05, + "loss": 0.4481, + "step": 3134 + }, + { + "epoch": 1.31, + "learning_rate": 1.907209314443355e-05, + "loss": 0.4373, + "step": 3135 + }, + { + "epoch": 1.31, + "learning_rate": 1.907137979025623e-05, + "loss": 0.4099, + "step": 3136 + }, + { + "epoch": 1.31, + "learning_rate": 1.907066617533015e-05, + "loss": 0.39, + "step": 3137 + }, + { + "epoch": 1.31, + "learning_rate": 1.906995229967582e-05, + "loss": 0.3978, + "step": 3138 + }, + { + "epoch": 1.31, + "learning_rate": 1.9069238163313757e-05, + "loss": 0.4477, + "step": 3139 + }, + { + "epoch": 1.31, + "learning_rate": 1.906852376626449e-05, + "loss": 0.3745, + "step": 3140 + }, + { + "epoch": 1.32, + "learning_rate": 1.906780910854856e-05, + "loss": 0.3677, + "step": 3141 + }, + { + "epoch": 1.32, + "learning_rate": 1.90670941901865e-05, + "loss": 0.359, + "step": 3142 + }, + { + "epoch": 1.32, + "learning_rate": 1.9066379011198863e-05, + "loss": 0.3517, + "step": 3143 + }, + { + "epoch": 1.32, + "learning_rate": 1.906566357160621e-05, + "loss": 0.3351, + "step": 3144 + }, + { + "epoch": 1.32, + "learning_rate": 1.9064947871429102e-05, + "loss": 0.3869, + "step": 3145 + }, + { + "epoch": 1.32, + "learning_rate": 1.9064231910688108e-05, + "loss": 0.4992, + "step": 3146 + }, + { + "epoch": 1.32, + "learning_rate": 1.9063515689403815e-05, + "loss": 0.3779, + "step": 3147 + }, + { + "epoch": 1.32, + "learning_rate": 1.90627992075968e-05, + "loss": 0.3634, + "step": 3148 + }, + { + "epoch": 1.32, + "learning_rate": 1.9062082465287672e-05, + "loss": 0.38, + "step": 3149 + }, + { + "epoch": 1.32, + "learning_rate": 1.9061365462497022e-05, + "loss": 0.3647, + "step": 3150 + }, + { + "epoch": 1.32, + "learning_rate": 1.9060648199245463e-05, + "loss": 0.4137, + "step": 3151 + }, + { + "epoch": 1.32, + "learning_rate": 1.905993067555361e-05, + "loss": 0.392, + "step": 3152 + }, + { + "epoch": 1.32, + "learning_rate": 1.905921289144209e-05, + "loss": 0.3798, + "step": 3153 + }, + { + "epoch": 1.32, + "learning_rate": 1.9058494846931538e-05, + "loss": 0.3981, + "step": 3154 + }, + { + "epoch": 1.32, + "learning_rate": 1.9057776542042587e-05, + "loss": 0.4565, + "step": 3155 + }, + { + "epoch": 1.32, + "learning_rate": 1.9057057976795884e-05, + "loss": 0.3855, + "step": 3156 + }, + { + "epoch": 1.32, + "learning_rate": 1.905633915121209e-05, + "loss": 0.3821, + "step": 3157 + }, + { + "epoch": 1.32, + "learning_rate": 1.905562006531186e-05, + "loss": 0.3874, + "step": 3158 + }, + { + "epoch": 1.32, + "learning_rate": 1.9054900719115874e-05, + "loss": 0.3713, + "step": 3159 + }, + { + "epoch": 1.32, + "learning_rate": 1.9054181112644797e-05, + "loss": 0.3942, + "step": 3160 + }, + { + "epoch": 1.32, + "learning_rate": 1.905346124591932e-05, + "loss": 0.4181, + "step": 3161 + }, + { + "epoch": 1.32, + "learning_rate": 1.9052741118960133e-05, + "loss": 0.405, + "step": 3162 + }, + { + "epoch": 1.32, + "learning_rate": 1.9052020731787936e-05, + "loss": 0.3995, + "step": 3163 + }, + { + "epoch": 1.32, + "learning_rate": 1.9051300084423433e-05, + "loss": 0.4052, + "step": 3164 + }, + { + "epoch": 1.33, + "learning_rate": 1.9050579176887343e-05, + "loss": 0.4243, + "step": 3165 + }, + { + "epoch": 1.33, + "learning_rate": 1.9049858009200388e-05, + "loss": 0.373, + "step": 3166 + }, + { + "epoch": 1.33, + "learning_rate": 1.9049136581383295e-05, + "loss": 0.4055, + "step": 3167 + }, + { + "epoch": 1.33, + "learning_rate": 1.90484148934568e-05, + "loss": 0.3674, + "step": 3168 + }, + { + "epoch": 1.33, + "learning_rate": 1.9047692945441647e-05, + "loss": 0.4224, + "step": 3169 + }, + { + "epoch": 1.33, + "learning_rate": 1.904697073735859e-05, + "loss": 0.4768, + "step": 3170 + }, + { + "epoch": 1.33, + "learning_rate": 1.9046248269228388e-05, + "loss": 0.3627, + "step": 3171 + }, + { + "epoch": 1.33, + "learning_rate": 1.9045525541071807e-05, + "loss": 0.3667, + "step": 3172 + }, + { + "epoch": 1.33, + "learning_rate": 1.904480255290962e-05, + "loss": 0.3859, + "step": 3173 + }, + { + "epoch": 1.33, + "learning_rate": 1.904407930476261e-05, + "loss": 0.3877, + "step": 3174 + }, + { + "epoch": 1.33, + "learning_rate": 1.9043355796651565e-05, + "loss": 0.4244, + "step": 3175 + }, + { + "epoch": 1.33, + "learning_rate": 1.9042632028597286e-05, + "loss": 0.4116, + "step": 3176 + }, + { + "epoch": 1.33, + "learning_rate": 1.904190800062057e-05, + "loss": 0.4461, + "step": 3177 + }, + { + "epoch": 1.33, + "learning_rate": 1.9041183712742233e-05, + "loss": 0.3593, + "step": 3178 + }, + { + "epoch": 1.33, + "learning_rate": 1.9040459164983095e-05, + "loss": 0.3896, + "step": 3179 + }, + { + "epoch": 1.33, + "learning_rate": 1.903973435736398e-05, + "loss": 0.4172, + "step": 3180 + }, + { + "epoch": 1.33, + "learning_rate": 1.903900928990572e-05, + "loss": 0.3866, + "step": 3181 + }, + { + "epoch": 1.33, + "learning_rate": 1.9038283962629164e-05, + "loss": 0.3818, + "step": 3182 + }, + { + "epoch": 1.33, + "learning_rate": 1.903755837555515e-05, + "loss": 0.3322, + "step": 3183 + }, + { + "epoch": 1.33, + "learning_rate": 1.9036832528704547e-05, + "loss": 0.4535, + "step": 3184 + }, + { + "epoch": 1.33, + "learning_rate": 1.9036106422098207e-05, + "loss": 0.4, + "step": 3185 + }, + { + "epoch": 1.33, + "learning_rate": 1.9035380055757007e-05, + "loss": 0.3732, + "step": 3186 + }, + { + "epoch": 1.33, + "learning_rate": 1.9034653429701824e-05, + "loss": 0.3552, + "step": 3187 + }, + { + "epoch": 1.33, + "learning_rate": 1.9033926543953546e-05, + "loss": 0.4101, + "step": 3188 + }, + { + "epoch": 1.34, + "learning_rate": 1.903319939853307e-05, + "loss": 0.409, + "step": 3189 + }, + { + "epoch": 1.34, + "learning_rate": 1.903247199346129e-05, + "loss": 0.3997, + "step": 3190 + }, + { + "epoch": 1.34, + "learning_rate": 1.9031744328759118e-05, + "loss": 0.4128, + "step": 3191 + }, + { + "epoch": 1.34, + "learning_rate": 1.903101640444747e-05, + "loss": 0.4079, + "step": 3192 + }, + { + "epoch": 1.34, + "learning_rate": 1.903028822054727e-05, + "loss": 0.3743, + "step": 3193 + }, + { + "epoch": 1.34, + "learning_rate": 1.902955977707945e-05, + "loss": 0.353, + "step": 3194 + }, + { + "epoch": 1.34, + "learning_rate": 1.9028831074064945e-05, + "loss": 0.3681, + "step": 3195 + }, + { + "epoch": 1.34, + "learning_rate": 1.9028102111524702e-05, + "loss": 0.4156, + "step": 3196 + }, + { + "epoch": 1.34, + "learning_rate": 1.9027372889479676e-05, + "loss": 0.4007, + "step": 3197 + }, + { + "epoch": 1.34, + "learning_rate": 1.9026643407950826e-05, + "loss": 0.3695, + "step": 3198 + }, + { + "epoch": 1.34, + "learning_rate": 1.9025913666959122e-05, + "loss": 0.3651, + "step": 3199 + }, + { + "epoch": 1.34, + "learning_rate": 1.9025183666525543e-05, + "loss": 0.3975, + "step": 3200 + }, + { + "epoch": 1.34, + "learning_rate": 1.9024453406671064e-05, + "loss": 0.4004, + "step": 3201 + }, + { + "epoch": 1.34, + "learning_rate": 1.902372288741668e-05, + "loss": 0.3727, + "step": 3202 + }, + { + "epoch": 1.34, + "learning_rate": 1.902299210878339e-05, + "loss": 0.4255, + "step": 3203 + }, + { + "epoch": 1.34, + "learning_rate": 1.9022261070792197e-05, + "loss": 0.3589, + "step": 3204 + }, + { + "epoch": 1.34, + "learning_rate": 1.902152977346412e-05, + "loss": 0.3909, + "step": 3205 + }, + { + "epoch": 1.34, + "learning_rate": 1.9020798216820174e-05, + "loss": 0.3787, + "step": 3206 + }, + { + "epoch": 1.34, + "learning_rate": 1.9020066400881386e-05, + "loss": 0.3855, + "step": 3207 + }, + { + "epoch": 1.34, + "learning_rate": 1.9019334325668795e-05, + "loss": 0.3937, + "step": 3208 + }, + { + "epoch": 1.34, + "learning_rate": 1.901860199120344e-05, + "loss": 0.448, + "step": 3209 + }, + { + "epoch": 1.34, + "learning_rate": 1.9017869397506374e-05, + "loss": 0.4216, + "step": 3210 + }, + { + "epoch": 1.34, + "learning_rate": 1.9017136544598657e-05, + "loss": 0.3929, + "step": 3211 + }, + { + "epoch": 1.34, + "learning_rate": 1.901640343250135e-05, + "loss": 0.4304, + "step": 3212 + }, + { + "epoch": 1.35, + "learning_rate": 1.9015670061235528e-05, + "loss": 0.3553, + "step": 3213 + }, + { + "epoch": 1.35, + "learning_rate": 1.901493643082227e-05, + "loss": 0.3978, + "step": 3214 + }, + { + "epoch": 1.35, + "learning_rate": 1.9014202541282663e-05, + "loss": 0.425, + "step": 3215 + }, + { + "epoch": 1.35, + "learning_rate": 1.901346839263781e-05, + "loss": 0.4167, + "step": 3216 + }, + { + "epoch": 1.35, + "learning_rate": 1.9012733984908798e-05, + "loss": 0.3343, + "step": 3217 + }, + { + "epoch": 1.35, + "learning_rate": 1.901199931811675e-05, + "loss": 0.3673, + "step": 3218 + }, + { + "epoch": 1.35, + "learning_rate": 1.9011264392282774e-05, + "loss": 0.4347, + "step": 3219 + }, + { + "epoch": 1.35, + "learning_rate": 1.9010529207428004e-05, + "loss": 0.337, + "step": 3220 + }, + { + "epoch": 1.35, + "learning_rate": 1.9009793763573564e-05, + "loss": 0.4128, + "step": 3221 + }, + { + "epoch": 1.35, + "learning_rate": 1.9009058060740603e-05, + "loss": 0.3688, + "step": 3222 + }, + { + "epoch": 1.35, + "learning_rate": 1.900832209895026e-05, + "loss": 0.4163, + "step": 3223 + }, + { + "epoch": 1.35, + "learning_rate": 1.900758587822369e-05, + "loss": 0.417, + "step": 3224 + }, + { + "epoch": 1.35, + "learning_rate": 1.900684939858206e-05, + "loss": 0.3658, + "step": 3225 + }, + { + "epoch": 1.35, + "learning_rate": 1.9006112660046535e-05, + "loss": 0.3795, + "step": 3226 + }, + { + "epoch": 1.35, + "learning_rate": 1.9005375662638294e-05, + "loss": 0.3625, + "step": 3227 + }, + { + "epoch": 1.35, + "learning_rate": 1.900463840637852e-05, + "loss": 0.3675, + "step": 3228 + }, + { + "epoch": 1.35, + "learning_rate": 1.9003900891288406e-05, + "loss": 0.3909, + "step": 3229 + }, + { + "epoch": 1.35, + "learning_rate": 1.900316311738915e-05, + "loss": 0.4061, + "step": 3230 + }, + { + "epoch": 1.35, + "learning_rate": 1.9002425084701958e-05, + "loss": 0.3623, + "step": 3231 + }, + { + "epoch": 1.35, + "learning_rate": 1.900168679324805e-05, + "loss": 0.4009, + "step": 3232 + }, + { + "epoch": 1.35, + "learning_rate": 1.900094824304864e-05, + "loss": 0.373, + "step": 3233 + }, + { + "epoch": 1.35, + "learning_rate": 1.900020943412496e-05, + "loss": 0.3667, + "step": 3234 + }, + { + "epoch": 1.35, + "learning_rate": 1.899947036649825e-05, + "loss": 0.3848, + "step": 3235 + }, + { + "epoch": 1.35, + "learning_rate": 1.8998731040189748e-05, + "loss": 0.4338, + "step": 3236 + }, + { + "epoch": 1.36, + "learning_rate": 1.8997991455220702e-05, + "loss": 0.4409, + "step": 3237 + }, + { + "epoch": 1.36, + "learning_rate": 1.8997251611612385e-05, + "loss": 0.4362, + "step": 3238 + }, + { + "epoch": 1.36, + "learning_rate": 1.899651150938605e-05, + "loss": 0.3413, + "step": 3239 + }, + { + "epoch": 1.36, + "learning_rate": 1.8995771148562978e-05, + "loss": 0.4013, + "step": 3240 + }, + { + "epoch": 1.36, + "learning_rate": 1.8995030529164446e-05, + "loss": 0.3302, + "step": 3241 + }, + { + "epoch": 1.36, + "learning_rate": 1.8994289651211742e-05, + "loss": 0.3261, + "step": 3242 + }, + { + "epoch": 1.36, + "learning_rate": 1.8993548514726162e-05, + "loss": 0.4185, + "step": 3243 + }, + { + "epoch": 1.36, + "learning_rate": 1.8992807119729013e-05, + "loss": 0.4119, + "step": 3244 + }, + { + "epoch": 1.36, + "learning_rate": 1.8992065466241603e-05, + "loss": 0.3402, + "step": 3245 + }, + { + "epoch": 1.36, + "learning_rate": 1.899132355428525e-05, + "loss": 0.3759, + "step": 3246 + }, + { + "epoch": 1.36, + "learning_rate": 1.899058138388128e-05, + "loss": 0.4098, + "step": 3247 + }, + { + "epoch": 1.36, + "learning_rate": 1.8989838955051027e-05, + "loss": 0.4127, + "step": 3248 + }, + { + "epoch": 1.36, + "learning_rate": 1.898909626781583e-05, + "loss": 0.3761, + "step": 3249 + }, + { + "epoch": 1.36, + "learning_rate": 1.898835332219704e-05, + "loss": 0.3541, + "step": 3250 + }, + { + "epoch": 1.36, + "learning_rate": 1.8987610118216006e-05, + "loss": 0.4133, + "step": 3251 + }, + { + "epoch": 1.36, + "learning_rate": 1.8986866655894098e-05, + "loss": 0.4228, + "step": 3252 + }, + { + "epoch": 1.36, + "learning_rate": 1.8986122935252684e-05, + "loss": 0.4267, + "step": 3253 + }, + { + "epoch": 1.36, + "learning_rate": 1.8985378956313133e-05, + "loss": 0.3861, + "step": 3254 + }, + { + "epoch": 1.36, + "learning_rate": 1.8984634719096844e-05, + "loss": 0.3509, + "step": 3255 + }, + { + "epoch": 1.36, + "learning_rate": 1.8983890223625205e-05, + "loss": 0.3746, + "step": 3256 + }, + { + "epoch": 1.36, + "learning_rate": 1.8983145469919606e-05, + "loss": 0.4175, + "step": 3257 + }, + { + "epoch": 1.36, + "learning_rate": 1.8982400458001465e-05, + "loss": 0.4045, + "step": 3258 + }, + { + "epoch": 1.36, + "learning_rate": 1.8981655187892194e-05, + "loss": 0.3661, + "step": 3259 + }, + { + "epoch": 1.36, + "learning_rate": 1.8980909659613217e-05, + "loss": 0.4329, + "step": 3260 + }, + { + "epoch": 1.37, + "learning_rate": 1.898016387318596e-05, + "loss": 0.3702, + "step": 3261 + }, + { + "epoch": 1.37, + "learning_rate": 1.8979417828631858e-05, + "loss": 0.4432, + "step": 3262 + }, + { + "epoch": 1.37, + "learning_rate": 1.8978671525972365e-05, + "loss": 0.3797, + "step": 3263 + }, + { + "epoch": 1.37, + "learning_rate": 1.8977924965228923e-05, + "loss": 0.3628, + "step": 3264 + }, + { + "epoch": 1.37, + "learning_rate": 1.8977178146422992e-05, + "loss": 0.3518, + "step": 3265 + }, + { + "epoch": 1.37, + "learning_rate": 1.8976431069576047e-05, + "loss": 0.3573, + "step": 3266 + }, + { + "epoch": 1.37, + "learning_rate": 1.8975683734709554e-05, + "loss": 0.3757, + "step": 3267 + }, + { + "epoch": 1.37, + "learning_rate": 1.8974936141844997e-05, + "loss": 0.37, + "step": 3268 + }, + { + "epoch": 1.37, + "learning_rate": 1.8974188291003864e-05, + "loss": 0.3833, + "step": 3269 + }, + { + "epoch": 1.37, + "learning_rate": 1.897344018220765e-05, + "loss": 0.3984, + "step": 3270 + }, + { + "epoch": 1.37, + "learning_rate": 1.8972691815477864e-05, + "loss": 0.4031, + "step": 3271 + }, + { + "epoch": 1.37, + "learning_rate": 1.8971943190836013e-05, + "loss": 0.3859, + "step": 3272 + }, + { + "epoch": 1.37, + "learning_rate": 1.8971194308303617e-05, + "loss": 0.3791, + "step": 3273 + }, + { + "epoch": 1.37, + "learning_rate": 1.89704451679022e-05, + "loss": 0.4298, + "step": 3274 + }, + { + "epoch": 1.37, + "learning_rate": 1.89696957696533e-05, + "loss": 0.3776, + "step": 3275 + }, + { + "epoch": 1.37, + "learning_rate": 1.8968946113578455e-05, + "loss": 0.3676, + "step": 3276 + }, + { + "epoch": 1.37, + "learning_rate": 1.896819619969921e-05, + "loss": 0.393, + "step": 3277 + }, + { + "epoch": 1.37, + "learning_rate": 1.8967446028037126e-05, + "loss": 0.372, + "step": 3278 + }, + { + "epoch": 1.37, + "learning_rate": 1.8966695598613758e-05, + "loss": 0.3877, + "step": 3279 + }, + { + "epoch": 1.37, + "learning_rate": 1.8965944911450685e-05, + "loss": 0.3731, + "step": 3280 + }, + { + "epoch": 1.37, + "learning_rate": 1.8965193966569483e-05, + "loss": 0.417, + "step": 3281 + }, + { + "epoch": 1.37, + "learning_rate": 1.8964442763991737e-05, + "loss": 0.4348, + "step": 3282 + }, + { + "epoch": 1.37, + "learning_rate": 1.8963691303739037e-05, + "loss": 0.4124, + "step": 3283 + }, + { + "epoch": 1.37, + "learning_rate": 1.8962939585832985e-05, + "loss": 0.3836, + "step": 3284 + }, + { + "epoch": 1.38, + "learning_rate": 1.8962187610295188e-05, + "loss": 0.4002, + "step": 3285 + }, + { + "epoch": 1.38, + "learning_rate": 1.896143537714726e-05, + "loss": 0.3544, + "step": 3286 + }, + { + "epoch": 1.38, + "learning_rate": 1.8960682886410825e-05, + "loss": 0.3982, + "step": 3287 + }, + { + "epoch": 1.38, + "learning_rate": 1.8959930138107512e-05, + "loss": 0.3865, + "step": 3288 + }, + { + "epoch": 1.38, + "learning_rate": 1.895917713225896e-05, + "loss": 0.3951, + "step": 3289 + }, + { + "epoch": 1.38, + "learning_rate": 1.895842386888681e-05, + "loss": 0.3978, + "step": 3290 + }, + { + "epoch": 1.38, + "learning_rate": 1.8957670348012717e-05, + "loss": 0.3807, + "step": 3291 + }, + { + "epoch": 1.38, + "learning_rate": 1.8956916569658336e-05, + "loss": 0.396, + "step": 3292 + }, + { + "epoch": 1.38, + "learning_rate": 1.895616253384534e-05, + "loss": 0.3715, + "step": 3293 + }, + { + "epoch": 1.38, + "learning_rate": 1.8955408240595396e-05, + "loss": 0.4017, + "step": 3294 + }, + { + "epoch": 1.38, + "learning_rate": 1.895465368993019e-05, + "loss": 0.4133, + "step": 3295 + }, + { + "epoch": 1.38, + "learning_rate": 1.8953898881871412e-05, + "loss": 0.3663, + "step": 3296 + }, + { + "epoch": 1.38, + "learning_rate": 1.895314381644075e-05, + "loss": 0.4083, + "step": 3297 + }, + { + "epoch": 1.38, + "learning_rate": 1.8952388493659918e-05, + "loss": 0.3797, + "step": 3298 + }, + { + "epoch": 1.38, + "learning_rate": 1.8951632913550625e-05, + "loss": 0.3964, + "step": 3299 + }, + { + "epoch": 1.38, + "learning_rate": 1.8950877076134583e-05, + "loss": 0.4279, + "step": 3300 + }, + { + "epoch": 1.38, + "learning_rate": 1.8950120981433527e-05, + "loss": 0.3264, + "step": 3301 + }, + { + "epoch": 1.38, + "learning_rate": 1.8949364629469184e-05, + "loss": 0.3627, + "step": 3302 + }, + { + "epoch": 1.38, + "learning_rate": 1.8948608020263293e-05, + "loss": 0.3707, + "step": 3303 + }, + { + "epoch": 1.38, + "learning_rate": 1.894785115383761e-05, + "loss": 0.3831, + "step": 3304 + }, + { + "epoch": 1.38, + "learning_rate": 1.8947094030213883e-05, + "loss": 0.3846, + "step": 3305 + }, + { + "epoch": 1.38, + "learning_rate": 1.8946336649413883e-05, + "loss": 0.3935, + "step": 3306 + }, + { + "epoch": 1.38, + "learning_rate": 1.8945579011459366e-05, + "loss": 0.4725, + "step": 3307 + }, + { + "epoch": 1.38, + "learning_rate": 1.8944821116372125e-05, + "loss": 0.401, + "step": 3308 + }, + { + "epoch": 1.39, + "learning_rate": 1.8944062964173934e-05, + "loss": 0.446, + "step": 3309 + }, + { + "epoch": 1.39, + "learning_rate": 1.8943304554886594e-05, + "loss": 0.3866, + "step": 3310 + }, + { + "epoch": 1.39, + "learning_rate": 1.8942545888531897e-05, + "loss": 0.4263, + "step": 3311 + }, + { + "epoch": 1.39, + "learning_rate": 1.8941786965131658e-05, + "loss": 0.4343, + "step": 3312 + }, + { + "epoch": 1.39, + "learning_rate": 1.8941027784707685e-05, + "loss": 0.3569, + "step": 3313 + }, + { + "epoch": 1.39, + "learning_rate": 1.8940268347281802e-05, + "loss": 0.3838, + "step": 3314 + }, + { + "epoch": 1.39, + "learning_rate": 1.8939508652875837e-05, + "loss": 0.3983, + "step": 3315 + }, + { + "epoch": 1.39, + "learning_rate": 1.8938748701511633e-05, + "loss": 0.3926, + "step": 3316 + }, + { + "epoch": 1.39, + "learning_rate": 1.8937988493211027e-05, + "loss": 0.3967, + "step": 3317 + }, + { + "epoch": 1.39, + "learning_rate": 1.8937228027995874e-05, + "loss": 0.3469, + "step": 3318 + }, + { + "epoch": 1.39, + "learning_rate": 1.8936467305888033e-05, + "loss": 0.3752, + "step": 3319 + }, + { + "epoch": 1.39, + "learning_rate": 1.893570632690937e-05, + "loss": 0.3537, + "step": 3320 + }, + { + "epoch": 1.39, + "learning_rate": 1.8934945091081755e-05, + "loss": 0.3916, + "step": 3321 + }, + { + "epoch": 1.39, + "learning_rate": 1.8934183598427072e-05, + "loss": 0.3867, + "step": 3322 + }, + { + "epoch": 1.39, + "learning_rate": 1.8933421848967213e-05, + "loss": 0.3915, + "step": 3323 + }, + { + "epoch": 1.39, + "learning_rate": 1.8932659842724067e-05, + "loss": 0.3388, + "step": 3324 + }, + { + "epoch": 1.39, + "learning_rate": 1.8931897579719543e-05, + "loss": 0.4215, + "step": 3325 + }, + { + "epoch": 1.39, + "learning_rate": 1.8931135059975548e-05, + "loss": 0.519, + "step": 3326 + }, + { + "epoch": 1.39, + "learning_rate": 1.8930372283514004e-05, + "loss": 0.3756, + "step": 3327 + }, + { + "epoch": 1.39, + "learning_rate": 1.8929609250356827e-05, + "loss": 0.354, + "step": 3328 + }, + { + "epoch": 1.39, + "learning_rate": 1.8928845960525957e-05, + "loss": 0.3802, + "step": 3329 + }, + { + "epoch": 1.39, + "learning_rate": 1.8928082414043342e-05, + "loss": 0.3898, + "step": 3330 + }, + { + "epoch": 1.39, + "learning_rate": 1.8927318610930912e-05, + "loss": 0.414, + "step": 3331 + }, + { + "epoch": 1.4, + "learning_rate": 1.892655455121063e-05, + "loss": 0.4356, + "step": 3332 + }, + { + "epoch": 1.4, + "learning_rate": 1.8925790234904463e-05, + "loss": 0.3701, + "step": 3333 + }, + { + "epoch": 1.4, + "learning_rate": 1.8925025662034376e-05, + "loss": 0.4053, + "step": 3334 + }, + { + "epoch": 1.4, + "learning_rate": 1.8924260832622347e-05, + "loss": 0.3682, + "step": 3335 + }, + { + "epoch": 1.4, + "learning_rate": 1.8923495746690354e-05, + "loss": 0.3782, + "step": 3336 + }, + { + "epoch": 1.4, + "learning_rate": 1.89227304042604e-05, + "loss": 0.4035, + "step": 3337 + }, + { + "epoch": 1.4, + "learning_rate": 1.8921964805354475e-05, + "loss": 0.3436, + "step": 3338 + }, + { + "epoch": 1.4, + "learning_rate": 1.892119894999459e-05, + "loss": 0.3769, + "step": 3339 + }, + { + "epoch": 1.4, + "learning_rate": 1.892043283820276e-05, + "loss": 0.4044, + "step": 3340 + }, + { + "epoch": 1.4, + "learning_rate": 1.8919666470001003e-05, + "loss": 0.4397, + "step": 3341 + }, + { + "epoch": 1.4, + "learning_rate": 1.8918899845411343e-05, + "loss": 0.3762, + "step": 3342 + }, + { + "epoch": 1.4, + "learning_rate": 1.891813296445583e-05, + "loss": 0.3667, + "step": 3343 + }, + { + "epoch": 1.4, + "learning_rate": 1.8917365827156493e-05, + "loss": 0.4077, + "step": 3344 + }, + { + "epoch": 1.4, + "learning_rate": 1.891659843353539e-05, + "loss": 0.4158, + "step": 3345 + }, + { + "epoch": 1.4, + "learning_rate": 1.8915830783614576e-05, + "loss": 0.3733, + "step": 3346 + }, + { + "epoch": 1.4, + "learning_rate": 1.891506287741612e-05, + "loss": 0.3935, + "step": 3347 + }, + { + "epoch": 1.4, + "learning_rate": 1.8914294714962093e-05, + "loss": 0.3812, + "step": 3348 + }, + { + "epoch": 1.4, + "learning_rate": 1.8913526296274573e-05, + "loss": 0.3571, + "step": 3349 + }, + { + "epoch": 1.4, + "learning_rate": 1.8912757621375654e-05, + "loss": 0.3853, + "step": 3350 + }, + { + "epoch": 1.4, + "learning_rate": 1.8911988690287422e-05, + "loss": 0.412, + "step": 3351 + }, + { + "epoch": 1.4, + "learning_rate": 1.891121950303198e-05, + "loss": 0.4038, + "step": 3352 + }, + { + "epoch": 1.4, + "learning_rate": 1.891045005963145e-05, + "loss": 0.3953, + "step": 3353 + }, + { + "epoch": 1.4, + "learning_rate": 1.890968036010793e-05, + "loss": 0.4022, + "step": 3354 + }, + { + "epoch": 1.4, + "learning_rate": 1.8908910404483564e-05, + "loss": 0.3732, + "step": 3355 + }, + { + "epoch": 1.41, + "learning_rate": 1.8908140192780466e-05, + "loss": 0.3713, + "step": 3356 + }, + { + "epoch": 1.41, + "learning_rate": 1.890736972502079e-05, + "loss": 0.4261, + "step": 3357 + }, + { + "epoch": 1.41, + "learning_rate": 1.890659900122667e-05, + "loss": 0.3903, + "step": 3358 + }, + { + "epoch": 1.41, + "learning_rate": 1.890582802142027e-05, + "loss": 0.4046, + "step": 3359 + }, + { + "epoch": 1.41, + "learning_rate": 1.8905056785623742e-05, + "loss": 0.3788, + "step": 3360 + }, + { + "epoch": 1.41, + "learning_rate": 1.890428529385926e-05, + "loss": 0.3524, + "step": 3361 + }, + { + "epoch": 1.41, + "learning_rate": 1.8903513546149e-05, + "loss": 0.3807, + "step": 3362 + }, + { + "epoch": 1.41, + "learning_rate": 1.8902741542515143e-05, + "loss": 0.33, + "step": 3363 + }, + { + "epoch": 1.41, + "learning_rate": 1.8901969282979882e-05, + "loss": 0.3963, + "step": 3364 + }, + { + "epoch": 1.41, + "learning_rate": 1.890119676756541e-05, + "loss": 0.3798, + "step": 3365 + }, + { + "epoch": 1.41, + "learning_rate": 1.8900423996293938e-05, + "loss": 0.4029, + "step": 3366 + }, + { + "epoch": 1.41, + "learning_rate": 1.889965096918768e-05, + "loss": 0.3793, + "step": 3367 + }, + { + "epoch": 1.41, + "learning_rate": 1.889887768626885e-05, + "loss": 0.3876, + "step": 3368 + }, + { + "epoch": 1.41, + "learning_rate": 1.8898104147559674e-05, + "loss": 0.3914, + "step": 3369 + }, + { + "epoch": 1.41, + "learning_rate": 1.8897330353082392e-05, + "loss": 0.3742, + "step": 3370 + }, + { + "epoch": 1.41, + "learning_rate": 1.8896556302859243e-05, + "loss": 0.4183, + "step": 3371 + }, + { + "epoch": 1.41, + "learning_rate": 1.889578199691248e-05, + "loss": 0.4014, + "step": 3372 + }, + { + "epoch": 1.41, + "learning_rate": 1.889500743526436e-05, + "loss": 0.3522, + "step": 3373 + }, + { + "epoch": 1.41, + "learning_rate": 1.889423261793714e-05, + "loss": 0.4073, + "step": 3374 + }, + { + "epoch": 1.41, + "learning_rate": 1.88934575449531e-05, + "loss": 0.3872, + "step": 3375 + }, + { + "epoch": 1.41, + "learning_rate": 1.889268221633451e-05, + "loss": 0.3361, + "step": 3376 + }, + { + "epoch": 1.41, + "learning_rate": 1.8891906632103664e-05, + "loss": 0.4251, + "step": 3377 + }, + { + "epoch": 1.41, + "learning_rate": 1.889113079228285e-05, + "loss": 0.3913, + "step": 3378 + }, + { + "epoch": 1.41, + "learning_rate": 1.8890354696894374e-05, + "loss": 0.3846, + "step": 3379 + }, + { + "epoch": 1.42, + "learning_rate": 1.888957834596054e-05, + "loss": 0.3444, + "step": 3380 + }, + { + "epoch": 1.42, + "learning_rate": 1.8888801739503667e-05, + "loss": 0.4703, + "step": 3381 + }, + { + "epoch": 1.42, + "learning_rate": 1.8888024877546074e-05, + "loss": 0.3833, + "step": 3382 + }, + { + "epoch": 1.42, + "learning_rate": 1.8887247760110094e-05, + "loss": 0.3652, + "step": 3383 + }, + { + "epoch": 1.42, + "learning_rate": 1.888647038721806e-05, + "loss": 0.4107, + "step": 3384 + }, + { + "epoch": 1.42, + "learning_rate": 1.8885692758892324e-05, + "loss": 0.3597, + "step": 3385 + }, + { + "epoch": 1.42, + "learning_rate": 1.8884914875155235e-05, + "loss": 0.3862, + "step": 3386 + }, + { + "epoch": 1.42, + "learning_rate": 1.8884136736029154e-05, + "loss": 0.4022, + "step": 3387 + }, + { + "epoch": 1.42, + "learning_rate": 1.8883358341536442e-05, + "loss": 0.4315, + "step": 3388 + }, + { + "epoch": 1.42, + "learning_rate": 1.8882579691699475e-05, + "loss": 0.4305, + "step": 3389 + }, + { + "epoch": 1.42, + "learning_rate": 1.8881800786540642e-05, + "loss": 0.3692, + "step": 3390 + }, + { + "epoch": 1.42, + "learning_rate": 1.8881021626082328e-05, + "loss": 0.3415, + "step": 3391 + }, + { + "epoch": 1.42, + "learning_rate": 1.8880242210346925e-05, + "loss": 0.4347, + "step": 3392 + }, + { + "epoch": 1.42, + "learning_rate": 1.8879462539356842e-05, + "loss": 0.3729, + "step": 3393 + }, + { + "epoch": 1.42, + "learning_rate": 1.8878682613134486e-05, + "loss": 0.3826, + "step": 3394 + }, + { + "epoch": 1.42, + "learning_rate": 1.8877902431702274e-05, + "loss": 0.4082, + "step": 3395 + }, + { + "epoch": 1.42, + "learning_rate": 1.8877121995082638e-05, + "loss": 0.3841, + "step": 3396 + }, + { + "epoch": 1.42, + "learning_rate": 1.8876341303298012e-05, + "loss": 0.3887, + "step": 3397 + }, + { + "epoch": 1.42, + "learning_rate": 1.8875560356370827e-05, + "loss": 0.3646, + "step": 3398 + }, + { + "epoch": 1.42, + "learning_rate": 1.8874779154323538e-05, + "loss": 0.3887, + "step": 3399 + }, + { + "epoch": 1.42, + "learning_rate": 1.8873997697178593e-05, + "loss": 0.3638, + "step": 3400 + }, + { + "epoch": 1.42, + "learning_rate": 1.8873215984958465e-05, + "loss": 0.4295, + "step": 3401 + }, + { + "epoch": 1.42, + "learning_rate": 1.8872434017685613e-05, + "loss": 0.3792, + "step": 3402 + }, + { + "epoch": 1.42, + "learning_rate": 1.8871651795382518e-05, + "loss": 0.3996, + "step": 3403 + }, + { + "epoch": 1.43, + "learning_rate": 1.8870869318071667e-05, + "loss": 0.4269, + "step": 3404 + }, + { + "epoch": 1.43, + "learning_rate": 1.8870086585775548e-05, + "loss": 0.3915, + "step": 3405 + }, + { + "epoch": 1.43, + "learning_rate": 1.8869303598516663e-05, + "loss": 0.3593, + "step": 3406 + }, + { + "epoch": 1.43, + "learning_rate": 1.8868520356317517e-05, + "loss": 0.3277, + "step": 3407 + }, + { + "epoch": 1.43, + "learning_rate": 1.886773685920062e-05, + "loss": 0.3957, + "step": 3408 + }, + { + "epoch": 1.43, + "learning_rate": 1.8866953107188498e-05, + "loss": 0.4089, + "step": 3409 + }, + { + "epoch": 1.43, + "learning_rate": 1.8866169100303677e-05, + "loss": 0.368, + "step": 3410 + }, + { + "epoch": 1.43, + "learning_rate": 1.8865384838568693e-05, + "loss": 0.3971, + "step": 3411 + }, + { + "epoch": 1.43, + "learning_rate": 1.886460032200609e-05, + "loss": 0.4052, + "step": 3412 + }, + { + "epoch": 1.43, + "learning_rate": 1.8863815550638414e-05, + "loss": 0.3769, + "step": 3413 + }, + { + "epoch": 1.43, + "learning_rate": 1.886303052448823e-05, + "loss": 0.3886, + "step": 3414 + }, + { + "epoch": 1.43, + "learning_rate": 1.8862245243578097e-05, + "loss": 0.4253, + "step": 3415 + }, + { + "epoch": 1.43, + "learning_rate": 1.8861459707930585e-05, + "loss": 0.3974, + "step": 3416 + }, + { + "epoch": 1.43, + "learning_rate": 1.886067391756828e-05, + "loss": 0.4251, + "step": 3417 + }, + { + "epoch": 1.43, + "learning_rate": 1.885988787251377e-05, + "loss": 0.3961, + "step": 3418 + }, + { + "epoch": 1.43, + "learning_rate": 1.885910157278964e-05, + "loss": 0.3985, + "step": 3419 + }, + { + "epoch": 1.43, + "learning_rate": 1.8858315018418497e-05, + "loss": 0.4277, + "step": 3420 + }, + { + "epoch": 1.43, + "learning_rate": 1.885752820942295e-05, + "loss": 0.4149, + "step": 3421 + }, + { + "epoch": 1.43, + "learning_rate": 1.8856741145825616e-05, + "loss": 0.3769, + "step": 3422 + }, + { + "epoch": 1.43, + "learning_rate": 1.8855953827649115e-05, + "loss": 0.3979, + "step": 3423 + }, + { + "epoch": 1.43, + "learning_rate": 1.8855166254916082e-05, + "loss": 0.398, + "step": 3424 + }, + { + "epoch": 1.43, + "learning_rate": 1.885437842764915e-05, + "loss": 0.3734, + "step": 3425 + }, + { + "epoch": 1.43, + "learning_rate": 1.885359034587097e-05, + "loss": 0.3775, + "step": 3426 + }, + { + "epoch": 1.43, + "learning_rate": 1.885280200960419e-05, + "loss": 0.3443, + "step": 3427 + }, + { + "epoch": 1.44, + "learning_rate": 1.8852013418871474e-05, + "loss": 0.4162, + "step": 3428 + }, + { + "epoch": 1.44, + "learning_rate": 1.885122457369549e-05, + "loss": 0.3529, + "step": 3429 + }, + { + "epoch": 1.44, + "learning_rate": 1.8850435474098903e-05, + "loss": 0.4343, + "step": 3430 + }, + { + "epoch": 1.44, + "learning_rate": 1.884964612010441e-05, + "loss": 0.3688, + "step": 3431 + }, + { + "epoch": 1.44, + "learning_rate": 1.884885651173469e-05, + "loss": 0.4096, + "step": 3432 + }, + { + "epoch": 1.44, + "learning_rate": 1.8848066649012437e-05, + "loss": 0.3725, + "step": 3433 + }, + { + "epoch": 1.44, + "learning_rate": 1.8847276531960363e-05, + "loss": 0.3621, + "step": 3434 + }, + { + "epoch": 1.44, + "learning_rate": 1.8846486160601175e-05, + "loss": 0.3706, + "step": 3435 + }, + { + "epoch": 1.44, + "learning_rate": 1.8845695534957594e-05, + "loss": 0.4113, + "step": 3436 + }, + { + "epoch": 1.44, + "learning_rate": 1.8844904655052344e-05, + "loss": 0.3906, + "step": 3437 + }, + { + "epoch": 1.44, + "learning_rate": 1.884411352090816e-05, + "loss": 0.3552, + "step": 3438 + }, + { + "epoch": 1.44, + "learning_rate": 1.8843322132547778e-05, + "loss": 0.4262, + "step": 3439 + }, + { + "epoch": 1.44, + "learning_rate": 1.884253048999395e-05, + "loss": 0.4011, + "step": 3440 + }, + { + "epoch": 1.44, + "learning_rate": 1.8841738593269432e-05, + "loss": 0.3708, + "step": 3441 + }, + { + "epoch": 1.44, + "learning_rate": 1.884094644239698e-05, + "loss": 0.3951, + "step": 3442 + }, + { + "epoch": 1.44, + "learning_rate": 1.8840154037399368e-05, + "loss": 0.3575, + "step": 3443 + }, + { + "epoch": 1.44, + "learning_rate": 1.8839361378299376e-05, + "loss": 0.4257, + "step": 3444 + }, + { + "epoch": 1.44, + "learning_rate": 1.8838568465119786e-05, + "loss": 0.3638, + "step": 3445 + }, + { + "epoch": 1.44, + "learning_rate": 1.8837775297883384e-05, + "loss": 0.3661, + "step": 3446 + }, + { + "epoch": 1.44, + "learning_rate": 1.883698187661298e-05, + "loss": 0.4143, + "step": 3447 + }, + { + "epoch": 1.44, + "learning_rate": 1.8836188201331368e-05, + "loss": 0.3825, + "step": 3448 + }, + { + "epoch": 1.44, + "learning_rate": 1.8835394272061368e-05, + "loss": 0.3355, + "step": 3449 + }, + { + "epoch": 1.44, + "learning_rate": 1.88346000888258e-05, + "loss": 0.4143, + "step": 3450 + }, + { + "epoch": 1.44, + "learning_rate": 1.8833805651647498e-05, + "loss": 0.428, + "step": 3451 + }, + { + "epoch": 1.45, + "learning_rate": 1.8833010960549286e-05, + "loss": 0.3992, + "step": 3452 + }, + { + "epoch": 1.45, + "learning_rate": 1.883221601555401e-05, + "loss": 0.3523, + "step": 3453 + }, + { + "epoch": 1.45, + "learning_rate": 1.8831420816684528e-05, + "loss": 0.3957, + "step": 3454 + }, + { + "epoch": 1.45, + "learning_rate": 1.8830625363963685e-05, + "loss": 0.366, + "step": 3455 + }, + { + "epoch": 1.45, + "learning_rate": 1.882982965741436e-05, + "loss": 0.3754, + "step": 3456 + }, + { + "epoch": 1.45, + "learning_rate": 1.882903369705941e-05, + "loss": 0.4156, + "step": 3457 + }, + { + "epoch": 1.45, + "learning_rate": 1.8828237482921725e-05, + "loss": 0.4268, + "step": 3458 + }, + { + "epoch": 1.45, + "learning_rate": 1.8827441015024187e-05, + "loss": 0.3932, + "step": 3459 + }, + { + "epoch": 1.45, + "learning_rate": 1.882664429338969e-05, + "loss": 0.4338, + "step": 3460 + }, + { + "epoch": 1.45, + "learning_rate": 1.8825847318041136e-05, + "loss": 0.4078, + "step": 3461 + }, + { + "epoch": 1.45, + "learning_rate": 1.882505008900143e-05, + "loss": 0.4001, + "step": 3462 + }, + { + "epoch": 1.45, + "learning_rate": 1.8824252606293492e-05, + "loss": 0.3735, + "step": 3463 + }, + { + "epoch": 1.45, + "learning_rate": 1.8823454869940243e-05, + "loss": 0.3941, + "step": 3464 + }, + { + "epoch": 1.45, + "learning_rate": 1.8822656879964617e-05, + "loss": 0.3981, + "step": 3465 + }, + { + "epoch": 1.45, + "learning_rate": 1.8821858636389547e-05, + "loss": 0.4284, + "step": 3466 + }, + { + "epoch": 1.45, + "learning_rate": 1.8821060139237975e-05, + "loss": 0.3802, + "step": 3467 + }, + { + "epoch": 1.45, + "learning_rate": 1.8820261388532858e-05, + "loss": 0.3674, + "step": 3468 + }, + { + "epoch": 1.45, + "learning_rate": 1.8819462384297155e-05, + "loss": 0.3818, + "step": 3469 + }, + { + "epoch": 1.45, + "learning_rate": 1.8818663126553833e-05, + "loss": 0.4197, + "step": 3470 + }, + { + "epoch": 1.45, + "learning_rate": 1.8817863615325865e-05, + "loss": 0.425, + "step": 3471 + }, + { + "epoch": 1.45, + "learning_rate": 1.881706385063623e-05, + "loss": 0.3757, + "step": 3472 + }, + { + "epoch": 1.45, + "learning_rate": 1.881626383250792e-05, + "loss": 0.3716, + "step": 3473 + }, + { + "epoch": 1.45, + "learning_rate": 1.8815463560963932e-05, + "loss": 0.392, + "step": 3474 + }, + { + "epoch": 1.45, + "learning_rate": 1.8814663036027262e-05, + "loss": 0.3917, + "step": 3475 + }, + { + "epoch": 1.46, + "learning_rate": 1.881386225772093e-05, + "loss": 0.356, + "step": 3476 + }, + { + "epoch": 1.46, + "learning_rate": 1.8813061226067945e-05, + "loss": 0.3455, + "step": 3477 + }, + { + "epoch": 1.46, + "learning_rate": 1.881225994109134e-05, + "loss": 0.4187, + "step": 3478 + }, + { + "epoch": 1.46, + "learning_rate": 1.881145840281414e-05, + "loss": 0.3953, + "step": 3479 + }, + { + "epoch": 1.46, + "learning_rate": 1.8810656611259387e-05, + "loss": 0.3728, + "step": 3480 + }, + { + "epoch": 1.46, + "learning_rate": 1.8809854566450128e-05, + "loss": 0.3603, + "step": 3481 + }, + { + "epoch": 1.46, + "learning_rate": 1.8809052268409414e-05, + "loss": 0.3356, + "step": 3482 + }, + { + "epoch": 1.46, + "learning_rate": 1.8808249717160318e-05, + "loss": 0.4093, + "step": 3483 + }, + { + "epoch": 1.46, + "learning_rate": 1.8807446912725893e-05, + "loss": 0.4172, + "step": 3484 + }, + { + "epoch": 1.46, + "learning_rate": 1.8806643855129225e-05, + "loss": 0.3647, + "step": 3485 + }, + { + "epoch": 1.46, + "learning_rate": 1.8805840544393393e-05, + "loss": 0.4281, + "step": 3486 + }, + { + "epoch": 1.46, + "learning_rate": 1.880503698054149e-05, + "loss": 0.3747, + "step": 3487 + }, + { + "epoch": 1.46, + "learning_rate": 1.8804233163596613e-05, + "loss": 0.3777, + "step": 3488 + }, + { + "epoch": 1.46, + "learning_rate": 1.8803429093581863e-05, + "loss": 0.3581, + "step": 3489 + }, + { + "epoch": 1.46, + "learning_rate": 1.880262477052036e-05, + "loss": 0.3936, + "step": 3490 + }, + { + "epoch": 1.46, + "learning_rate": 1.8801820194435215e-05, + "loss": 0.3912, + "step": 3491 + }, + { + "epoch": 1.46, + "learning_rate": 1.8801015365349562e-05, + "loss": 0.3921, + "step": 3492 + }, + { + "epoch": 1.46, + "learning_rate": 1.880021028328653e-05, + "loss": 0.358, + "step": 3493 + }, + { + "epoch": 1.46, + "learning_rate": 1.8799404948269267e-05, + "loss": 0.3555, + "step": 3494 + }, + { + "epoch": 1.46, + "learning_rate": 1.879859936032091e-05, + "loss": 0.3767, + "step": 3495 + }, + { + "epoch": 1.46, + "learning_rate": 1.8797793519464628e-05, + "loss": 0.3615, + "step": 3496 + }, + { + "epoch": 1.46, + "learning_rate": 1.8796987425723578e-05, + "loss": 0.3714, + "step": 3497 + }, + { + "epoch": 1.46, + "learning_rate": 1.879618107912093e-05, + "loss": 0.3775, + "step": 3498 + }, + { + "epoch": 1.46, + "learning_rate": 1.8795374479679862e-05, + "loss": 0.4144, + "step": 3499 + }, + { + "epoch": 1.47, + "learning_rate": 1.879456762742356e-05, + "loss": 0.3544, + "step": 3500 + }, + { + "epoch": 1.47, + "learning_rate": 1.8793760522375214e-05, + "loss": 0.3959, + "step": 3501 + }, + { + "epoch": 1.47, + "learning_rate": 1.8792953164558026e-05, + "loss": 0.4117, + "step": 3502 + }, + { + "epoch": 1.47, + "learning_rate": 1.8792145553995203e-05, + "loss": 0.3799, + "step": 3503 + }, + { + "epoch": 1.47, + "learning_rate": 1.879133769070996e-05, + "loss": 0.4163, + "step": 3504 + }, + { + "epoch": 1.47, + "learning_rate": 1.8790529574725514e-05, + "loss": 0.4168, + "step": 3505 + }, + { + "epoch": 1.47, + "learning_rate": 1.8789721206065096e-05, + "loss": 0.3888, + "step": 3506 + }, + { + "epoch": 1.47, + "learning_rate": 1.8788912584751938e-05, + "loss": 0.3918, + "step": 3507 + }, + { + "epoch": 1.47, + "learning_rate": 1.8788103710809292e-05, + "loss": 0.3519, + "step": 3508 + }, + { + "epoch": 1.47, + "learning_rate": 1.8787294584260402e-05, + "loss": 0.377, + "step": 3509 + }, + { + "epoch": 1.47, + "learning_rate": 1.8786485205128526e-05, + "loss": 0.3672, + "step": 3510 + }, + { + "epoch": 1.47, + "learning_rate": 1.878567557343693e-05, + "loss": 0.4139, + "step": 3511 + }, + { + "epoch": 1.47, + "learning_rate": 1.8784865689208888e-05, + "loss": 0.3682, + "step": 3512 + }, + { + "epoch": 1.47, + "learning_rate": 1.8784055552467674e-05, + "loss": 0.4095, + "step": 3513 + }, + { + "epoch": 1.47, + "learning_rate": 1.878324516323658e-05, + "loss": 0.4295, + "step": 3514 + }, + { + "epoch": 1.47, + "learning_rate": 1.87824345215389e-05, + "loss": 0.4504, + "step": 3515 + }, + { + "epoch": 1.47, + "learning_rate": 1.878162362739793e-05, + "loss": 0.3981, + "step": 3516 + }, + { + "epoch": 1.47, + "learning_rate": 1.878081248083698e-05, + "loss": 0.4034, + "step": 3517 + }, + { + "epoch": 1.47, + "learning_rate": 1.878000108187937e-05, + "loss": 0.3617, + "step": 3518 + }, + { + "epoch": 1.47, + "learning_rate": 1.877918943054842e-05, + "loss": 0.3635, + "step": 3519 + }, + { + "epoch": 1.47, + "learning_rate": 1.877837752686746e-05, + "loss": 0.4073, + "step": 3520 + }, + { + "epoch": 1.47, + "learning_rate": 1.8777565370859826e-05, + "loss": 0.3933, + "step": 3521 + }, + { + "epoch": 1.47, + "learning_rate": 1.8776752962548866e-05, + "loss": 0.3756, + "step": 3522 + }, + { + "epoch": 1.47, + "learning_rate": 1.8775940301957932e-05, + "loss": 0.3968, + "step": 3523 + }, + { + "epoch": 1.48, + "learning_rate": 1.877512738911038e-05, + "loss": 0.3968, + "step": 3524 + }, + { + "epoch": 1.48, + "learning_rate": 1.8774314224029578e-05, + "loss": 0.4166, + "step": 3525 + }, + { + "epoch": 1.48, + "learning_rate": 1.87735008067389e-05, + "loss": 0.3903, + "step": 3526 + }, + { + "epoch": 1.48, + "learning_rate": 1.8772687137261728e-05, + "loss": 0.3604, + "step": 3527 + }, + { + "epoch": 1.48, + "learning_rate": 1.8771873215621447e-05, + "loss": 0.3919, + "step": 3528 + }, + { + "epoch": 1.48, + "learning_rate": 1.8771059041841454e-05, + "loss": 0.4071, + "step": 3529 + }, + { + "epoch": 1.48, + "learning_rate": 1.8770244615945153e-05, + "loss": 0.4072, + "step": 3530 + }, + { + "epoch": 1.48, + "learning_rate": 1.8769429937955954e-05, + "loss": 0.3793, + "step": 3531 + }, + { + "epoch": 1.48, + "learning_rate": 1.8768615007897274e-05, + "loss": 0.3788, + "step": 3532 + }, + { + "epoch": 1.48, + "learning_rate": 1.8767799825792533e-05, + "loss": 0.3738, + "step": 3533 + }, + { + "epoch": 1.48, + "learning_rate": 1.8766984391665173e-05, + "loss": 0.3879, + "step": 3534 + }, + { + "epoch": 1.48, + "learning_rate": 1.876616870553862e-05, + "loss": 0.3786, + "step": 3535 + }, + { + "epoch": 1.48, + "learning_rate": 1.876535276743633e-05, + "loss": 0.4138, + "step": 3536 + }, + { + "epoch": 1.48, + "learning_rate": 1.8764536577381748e-05, + "loss": 0.4008, + "step": 3537 + }, + { + "epoch": 1.48, + "learning_rate": 1.8763720135398348e-05, + "loss": 0.3867, + "step": 3538 + }, + { + "epoch": 1.48, + "learning_rate": 1.8762903441509582e-05, + "loss": 0.3631, + "step": 3539 + }, + { + "epoch": 1.48, + "learning_rate": 1.8762086495738937e-05, + "loss": 0.35, + "step": 3540 + }, + { + "epoch": 1.48, + "learning_rate": 1.876126929810989e-05, + "loss": 0.3891, + "step": 3541 + }, + { + "epoch": 1.48, + "learning_rate": 1.8760451848645932e-05, + "loss": 0.3692, + "step": 3542 + }, + { + "epoch": 1.48, + "learning_rate": 1.8759634147370555e-05, + "loss": 0.4024, + "step": 3543 + }, + { + "epoch": 1.48, + "learning_rate": 1.8758816194307272e-05, + "loss": 0.3913, + "step": 3544 + }, + { + "epoch": 1.48, + "learning_rate": 1.875799798947959e-05, + "loss": 0.3868, + "step": 3545 + }, + { + "epoch": 1.48, + "learning_rate": 1.8757179532911026e-05, + "loss": 0.3665, + "step": 3546 + }, + { + "epoch": 1.49, + "learning_rate": 1.8756360824625107e-05, + "loss": 0.4118, + "step": 3547 + }, + { + "epoch": 1.49, + "learning_rate": 1.875554186464537e-05, + "loss": 0.3749, + "step": 3548 + }, + { + "epoch": 1.49, + "learning_rate": 1.8754722652995346e-05, + "loss": 0.4166, + "step": 3549 + }, + { + "epoch": 1.49, + "learning_rate": 1.8753903189698594e-05, + "loss": 0.3962, + "step": 3550 + }, + { + "epoch": 1.49, + "learning_rate": 1.875308347477866e-05, + "loss": 0.4516, + "step": 3551 + }, + { + "epoch": 1.49, + "learning_rate": 1.8752263508259107e-05, + "loss": 0.4046, + "step": 3552 + }, + { + "epoch": 1.49, + "learning_rate": 1.875144329016351e-05, + "loss": 0.4097, + "step": 3553 + }, + { + "epoch": 1.49, + "learning_rate": 1.875062282051544e-05, + "loss": 0.4241, + "step": 3554 + }, + { + "epoch": 1.49, + "learning_rate": 1.874980209933848e-05, + "loss": 0.38, + "step": 3555 + }, + { + "epoch": 1.49, + "learning_rate": 1.8748981126656224e-05, + "loss": 0.3898, + "step": 3556 + }, + { + "epoch": 1.49, + "learning_rate": 1.8748159902492273e-05, + "loss": 0.4338, + "step": 3557 + }, + { + "epoch": 1.49, + "learning_rate": 1.8747338426870222e-05, + "loss": 0.4228, + "step": 3558 + }, + { + "epoch": 1.49, + "learning_rate": 1.8746516699813693e-05, + "loss": 0.35, + "step": 3559 + }, + { + "epoch": 1.49, + "learning_rate": 1.8745694721346305e-05, + "loss": 0.3797, + "step": 3560 + }, + { + "epoch": 1.49, + "learning_rate": 1.874487249149168e-05, + "loss": 0.4204, + "step": 3561 + }, + { + "epoch": 1.49, + "learning_rate": 1.8744050010273458e-05, + "loss": 0.4093, + "step": 3562 + }, + { + "epoch": 1.49, + "learning_rate": 1.874322727771528e-05, + "loss": 0.4614, + "step": 3563 + }, + { + "epoch": 1.49, + "learning_rate": 1.8742404293840784e-05, + "loss": 0.3874, + "step": 3564 + }, + { + "epoch": 1.49, + "learning_rate": 1.8741581058673642e-05, + "loss": 0.3658, + "step": 3565 + }, + { + "epoch": 1.49, + "learning_rate": 1.874075757223751e-05, + "loss": 0.4201, + "step": 3566 + }, + { + "epoch": 1.49, + "learning_rate": 1.8739933834556054e-05, + "loss": 0.3875, + "step": 3567 + }, + { + "epoch": 1.49, + "learning_rate": 1.8739109845652957e-05, + "loss": 0.4209, + "step": 3568 + }, + { + "epoch": 1.49, + "learning_rate": 1.87382856055519e-05, + "loss": 0.347, + "step": 3569 + }, + { + "epoch": 1.49, + "learning_rate": 1.873746111427658e-05, + "loss": 0.4023, + "step": 3570 + }, + { + "epoch": 1.5, + "learning_rate": 1.8736636371850694e-05, + "loss": 0.3956, + "step": 3571 + }, + { + "epoch": 1.5, + "learning_rate": 1.8735811378297946e-05, + "loss": 0.406, + "step": 3572 + }, + { + "epoch": 1.5, + "learning_rate": 1.8734986133642058e-05, + "loss": 0.3692, + "step": 3573 + }, + { + "epoch": 1.5, + "learning_rate": 1.8734160637906738e-05, + "loss": 0.3746, + "step": 3574 + }, + { + "epoch": 1.5, + "learning_rate": 1.8733334891115725e-05, + "loss": 0.3654, + "step": 3575 + }, + { + "epoch": 1.5, + "learning_rate": 1.873250889329275e-05, + "loss": 0.3501, + "step": 3576 + }, + { + "epoch": 1.5, + "learning_rate": 1.8731682644461555e-05, + "loss": 0.4146, + "step": 3577 + }, + { + "epoch": 1.5, + "learning_rate": 1.873085614464589e-05, + "loss": 0.3821, + "step": 3578 + }, + { + "epoch": 1.5, + "learning_rate": 1.8730029393869512e-05, + "loss": 0.3517, + "step": 3579 + }, + { + "epoch": 1.5, + "learning_rate": 1.8729202392156186e-05, + "loss": 0.4147, + "step": 3580 + }, + { + "epoch": 1.5, + "learning_rate": 1.8728375139529688e-05, + "loss": 0.4, + "step": 3581 + }, + { + "epoch": 1.5, + "learning_rate": 1.872754763601379e-05, + "loss": 0.3598, + "step": 3582 + }, + { + "epoch": 1.5, + "learning_rate": 1.8726719881632278e-05, + "loss": 0.405, + "step": 3583 + }, + { + "epoch": 1.5, + "learning_rate": 1.8725891876408946e-05, + "loss": 0.4087, + "step": 3584 + }, + { + "epoch": 1.5, + "learning_rate": 1.8725063620367598e-05, + "loss": 0.4424, + "step": 3585 + }, + { + "epoch": 1.5, + "learning_rate": 1.872423511353204e-05, + "loss": 0.3516, + "step": 3586 + }, + { + "epoch": 1.5, + "learning_rate": 1.8723406355926078e-05, + "loss": 0.4001, + "step": 3587 + }, + { + "epoch": 1.5, + "learning_rate": 1.872257734757355e-05, + "loss": 0.3657, + "step": 3588 + }, + { + "epoch": 1.5, + "learning_rate": 1.8721748088498272e-05, + "loss": 0.3784, + "step": 3589 + }, + { + "epoch": 1.5, + "learning_rate": 1.8720918578724085e-05, + "loss": 0.4014, + "step": 3590 + }, + { + "epoch": 1.5, + "learning_rate": 1.8720088818274835e-05, + "loss": 0.4215, + "step": 3591 + }, + { + "epoch": 1.5, + "learning_rate": 1.8719258807174366e-05, + "loss": 0.3759, + "step": 3592 + }, + { + "epoch": 1.5, + "learning_rate": 1.8718428545446545e-05, + "loss": 0.3367, + "step": 3593 + }, + { + "epoch": 1.5, + "learning_rate": 1.8717598033115224e-05, + "loss": 0.3609, + "step": 3594 + }, + { + "epoch": 1.51, + "learning_rate": 1.871676727020429e-05, + "loss": 0.3814, + "step": 3595 + }, + { + "epoch": 1.51, + "learning_rate": 1.871593625673761e-05, + "loss": 0.4125, + "step": 3596 + }, + { + "epoch": 1.51, + "learning_rate": 1.8715104992739085e-05, + "loss": 0.3841, + "step": 3597 + }, + { + "epoch": 1.51, + "learning_rate": 1.8714273478232597e-05, + "loss": 0.3855, + "step": 3598 + }, + { + "epoch": 1.51, + "learning_rate": 1.871344171324205e-05, + "loss": 0.4007, + "step": 3599 + }, + { + "epoch": 1.51, + "learning_rate": 1.871260969779135e-05, + "loss": 0.3679, + "step": 3600 + }, + { + "epoch": 1.51, + "learning_rate": 1.871177743190442e-05, + "loss": 0.3722, + "step": 3601 + }, + { + "epoch": 1.51, + "learning_rate": 1.871094491560518e-05, + "loss": 0.3753, + "step": 3602 + }, + { + "epoch": 1.51, + "learning_rate": 1.8710112148917555e-05, + "loss": 0.4449, + "step": 3603 + }, + { + "epoch": 1.51, + "learning_rate": 1.8709279131865486e-05, + "loss": 0.4068, + "step": 3604 + }, + { + "epoch": 1.51, + "learning_rate": 1.8708445864472917e-05, + "loss": 0.3895, + "step": 3605 + }, + { + "epoch": 1.51, + "learning_rate": 1.87076123467638e-05, + "loss": 0.3729, + "step": 3606 + }, + { + "epoch": 1.51, + "learning_rate": 1.8706778578762094e-05, + "loss": 0.3501, + "step": 3607 + }, + { + "epoch": 1.51, + "learning_rate": 1.870594456049176e-05, + "loss": 0.337, + "step": 3608 + }, + { + "epoch": 1.51, + "learning_rate": 1.870511029197678e-05, + "loss": 0.3914, + "step": 3609 + }, + { + "epoch": 1.51, + "learning_rate": 1.8704275773241126e-05, + "loss": 0.3952, + "step": 3610 + }, + { + "epoch": 1.51, + "learning_rate": 1.8703441004308792e-05, + "loss": 0.3786, + "step": 3611 + }, + { + "epoch": 1.51, + "learning_rate": 1.8702605985203767e-05, + "loss": 0.38, + "step": 3612 + }, + { + "epoch": 1.51, + "learning_rate": 1.8701770715950058e-05, + "loss": 0.4337, + "step": 3613 + }, + { + "epoch": 1.51, + "learning_rate": 1.8700935196571666e-05, + "loss": 0.3772, + "step": 3614 + }, + { + "epoch": 1.51, + "learning_rate": 1.8700099427092618e-05, + "loss": 0.3557, + "step": 3615 + }, + { + "epoch": 1.51, + "learning_rate": 1.869926340753693e-05, + "loss": 0.363, + "step": 3616 + }, + { + "epoch": 1.51, + "learning_rate": 1.8698427137928637e-05, + "loss": 0.4203, + "step": 3617 + }, + { + "epoch": 1.51, + "learning_rate": 1.8697590618291772e-05, + "loss": 0.4204, + "step": 3618 + }, + { + "epoch": 1.52, + "learning_rate": 1.8696753848650386e-05, + "loss": 0.3818, + "step": 3619 + }, + { + "epoch": 1.52, + "learning_rate": 1.8695916829028525e-05, + "loss": 0.4434, + "step": 3620 + }, + { + "epoch": 1.52, + "learning_rate": 1.8695079559450255e-05, + "loss": 0.4064, + "step": 3621 + }, + { + "epoch": 1.52, + "learning_rate": 1.8694242039939636e-05, + "loss": 0.3619, + "step": 3622 + }, + { + "epoch": 1.52, + "learning_rate": 1.8693404270520745e-05, + "loss": 0.3922, + "step": 3623 + }, + { + "epoch": 1.52, + "learning_rate": 1.8692566251217665e-05, + "loss": 0.3681, + "step": 3624 + }, + { + "epoch": 1.52, + "learning_rate": 1.8691727982054478e-05, + "loss": 0.3653, + "step": 3625 + }, + { + "epoch": 1.52, + "learning_rate": 1.8690889463055285e-05, + "loss": 0.3871, + "step": 3626 + }, + { + "epoch": 1.52, + "learning_rate": 1.8690050694244188e-05, + "loss": 0.3941, + "step": 3627 + }, + { + "epoch": 1.52, + "learning_rate": 1.868921167564529e-05, + "loss": 0.3549, + "step": 3628 + }, + { + "epoch": 1.52, + "learning_rate": 1.8688372407282717e-05, + "loss": 0.3583, + "step": 3629 + }, + { + "epoch": 1.52, + "learning_rate": 1.868753288918059e-05, + "loss": 0.3913, + "step": 3630 + }, + { + "epoch": 1.52, + "learning_rate": 1.868669312136304e-05, + "loss": 0.3775, + "step": 3631 + }, + { + "epoch": 1.52, + "learning_rate": 1.86858531038542e-05, + "loss": 0.3765, + "step": 3632 + }, + { + "epoch": 1.52, + "learning_rate": 1.8685012836678225e-05, + "loss": 0.3596, + "step": 3633 + }, + { + "epoch": 1.52, + "learning_rate": 1.8684172319859258e-05, + "loss": 0.3955, + "step": 3634 + }, + { + "epoch": 1.52, + "learning_rate": 1.8683331553421467e-05, + "loss": 0.468, + "step": 3635 + }, + { + "epoch": 1.52, + "learning_rate": 1.8682490537389017e-05, + "loss": 0.4153, + "step": 3636 + }, + { + "epoch": 1.52, + "learning_rate": 1.868164927178608e-05, + "loss": 0.3604, + "step": 3637 + }, + { + "epoch": 1.52, + "learning_rate": 1.8680807756636837e-05, + "loss": 0.3752, + "step": 3638 + }, + { + "epoch": 1.52, + "learning_rate": 1.8679965991965484e-05, + "loss": 0.3992, + "step": 3639 + }, + { + "epoch": 1.52, + "learning_rate": 1.8679123977796205e-05, + "loss": 0.3463, + "step": 3640 + }, + { + "epoch": 1.52, + "learning_rate": 1.867828171415321e-05, + "loss": 0.412, + "step": 3641 + }, + { + "epoch": 1.52, + "learning_rate": 1.867743920106071e-05, + "loss": 0.3994, + "step": 3642 + }, + { + "epoch": 1.53, + "learning_rate": 1.867659643854292e-05, + "loss": 0.4042, + "step": 3643 + }, + { + "epoch": 1.53, + "learning_rate": 1.867575342662406e-05, + "loss": 0.4215, + "step": 3644 + }, + { + "epoch": 1.53, + "learning_rate": 1.8674910165328374e-05, + "loss": 0.423, + "step": 3645 + }, + { + "epoch": 1.53, + "learning_rate": 1.8674066654680088e-05, + "loss": 0.3863, + "step": 3646 + }, + { + "epoch": 1.53, + "learning_rate": 1.8673222894703456e-05, + "loss": 0.419, + "step": 3647 + }, + { + "epoch": 1.53, + "learning_rate": 1.8672378885422728e-05, + "loss": 0.4253, + "step": 3648 + }, + { + "epoch": 1.53, + "learning_rate": 1.867153462686216e-05, + "loss": 0.3741, + "step": 3649 + }, + { + "epoch": 1.53, + "learning_rate": 1.8670690119046032e-05, + "loss": 0.3937, + "step": 3650 + }, + { + "epoch": 1.53, + "learning_rate": 1.8669845361998607e-05, + "loss": 0.3863, + "step": 3651 + }, + { + "epoch": 1.53, + "learning_rate": 1.8669000355744172e-05, + "loss": 0.4042, + "step": 3652 + }, + { + "epoch": 1.53, + "learning_rate": 1.8668155100307013e-05, + "loss": 0.4318, + "step": 3653 + }, + { + "epoch": 1.53, + "learning_rate": 1.8667309595711426e-05, + "loss": 0.4049, + "step": 3654 + }, + { + "epoch": 1.53, + "learning_rate": 1.8666463841981717e-05, + "loss": 0.4273, + "step": 3655 + }, + { + "epoch": 1.53, + "learning_rate": 1.8665617839142198e-05, + "loss": 0.3994, + "step": 3656 + }, + { + "epoch": 1.53, + "learning_rate": 1.866477158721718e-05, + "loss": 0.3565, + "step": 3657 + }, + { + "epoch": 1.53, + "learning_rate": 1.8663925086230998e-05, + "loss": 0.3925, + "step": 3658 + }, + { + "epoch": 1.53, + "learning_rate": 1.866307833620797e-05, + "loss": 0.446, + "step": 3659 + }, + { + "epoch": 1.53, + "learning_rate": 1.8662231337172446e-05, + "loss": 0.388, + "step": 3660 + }, + { + "epoch": 1.53, + "learning_rate": 1.866138408914877e-05, + "loss": 0.3558, + "step": 3661 + }, + { + "epoch": 1.53, + "learning_rate": 1.8660536592161293e-05, + "loss": 0.3339, + "step": 3662 + }, + { + "epoch": 1.53, + "learning_rate": 1.8659688846234376e-05, + "loss": 0.357, + "step": 3663 + }, + { + "epoch": 1.53, + "learning_rate": 1.8658840851392386e-05, + "loss": 0.4077, + "step": 3664 + }, + { + "epoch": 1.53, + "learning_rate": 1.8657992607659706e-05, + "loss": 0.3655, + "step": 3665 + }, + { + "epoch": 1.53, + "learning_rate": 1.8657144115060707e-05, + "loss": 0.3917, + "step": 3666 + }, + { + "epoch": 1.54, + "learning_rate": 1.8656295373619785e-05, + "loss": 0.4059, + "step": 3667 + }, + { + "epoch": 1.54, + "learning_rate": 1.8655446383361332e-05, + "loss": 0.4232, + "step": 3668 + }, + { + "epoch": 1.54, + "learning_rate": 1.8654597144309753e-05, + "loss": 0.3869, + "step": 3669 + }, + { + "epoch": 1.54, + "learning_rate": 1.865374765648946e-05, + "loss": 0.3615, + "step": 3670 + }, + { + "epoch": 1.54, + "learning_rate": 1.865289791992487e-05, + "loss": 0.368, + "step": 3671 + }, + { + "epoch": 1.54, + "learning_rate": 1.8652047934640405e-05, + "loss": 0.3497, + "step": 3672 + }, + { + "epoch": 1.54, + "learning_rate": 1.86511977006605e-05, + "loss": 0.4531, + "step": 3673 + }, + { + "epoch": 1.54, + "learning_rate": 1.8650347218009596e-05, + "loss": 0.3943, + "step": 3674 + }, + { + "epoch": 1.54, + "learning_rate": 1.864949648671214e-05, + "loss": 0.3597, + "step": 3675 + }, + { + "epoch": 1.54, + "learning_rate": 1.8648645506792578e-05, + "loss": 0.396, + "step": 3676 + }, + { + "epoch": 1.54, + "learning_rate": 1.864779427827538e-05, + "loss": 0.3928, + "step": 3677 + }, + { + "epoch": 1.54, + "learning_rate": 1.8646942801185006e-05, + "loss": 0.3414, + "step": 3678 + }, + { + "epoch": 1.54, + "learning_rate": 1.8646091075545936e-05, + "loss": 0.3588, + "step": 3679 + }, + { + "epoch": 1.54, + "learning_rate": 1.864523910138265e-05, + "loss": 0.4464, + "step": 3680 + }, + { + "epoch": 1.54, + "learning_rate": 1.864438687871964e-05, + "loss": 0.3666, + "step": 3681 + }, + { + "epoch": 1.54, + "learning_rate": 1.8643534407581397e-05, + "loss": 0.4229, + "step": 3682 + }, + { + "epoch": 1.54, + "learning_rate": 1.8642681687992427e-05, + "loss": 0.3988, + "step": 3683 + }, + { + "epoch": 1.54, + "learning_rate": 1.8641828719977244e-05, + "loss": 0.3685, + "step": 3684 + }, + { + "epoch": 1.54, + "learning_rate": 1.864097550356036e-05, + "loss": 0.3398, + "step": 3685 + }, + { + "epoch": 1.54, + "learning_rate": 1.8640122038766307e-05, + "loss": 0.4044, + "step": 3686 + }, + { + "epoch": 1.54, + "learning_rate": 1.863926832561961e-05, + "loss": 0.3999, + "step": 3687 + }, + { + "epoch": 1.54, + "learning_rate": 1.8638414364144813e-05, + "loss": 0.3707, + "step": 3688 + }, + { + "epoch": 1.54, + "learning_rate": 1.863756015436646e-05, + "loss": 0.3947, + "step": 3689 + }, + { + "epoch": 1.54, + "learning_rate": 1.8636705696309104e-05, + "loss": 0.4347, + "step": 3690 + }, + { + "epoch": 1.55, + "learning_rate": 1.8635850989997308e-05, + "loss": 0.3899, + "step": 3691 + }, + { + "epoch": 1.55, + "learning_rate": 1.863499603545564e-05, + "loss": 0.362, + "step": 3692 + }, + { + "epoch": 1.55, + "learning_rate": 1.8634140832708673e-05, + "loss": 0.4191, + "step": 3693 + }, + { + "epoch": 1.55, + "learning_rate": 1.8633285381780987e-05, + "loss": 0.4036, + "step": 3694 + }, + { + "epoch": 1.55, + "learning_rate": 1.8632429682697175e-05, + "loss": 0.4099, + "step": 3695 + }, + { + "epoch": 1.55, + "learning_rate": 1.8631573735481834e-05, + "loss": 0.43, + "step": 3696 + }, + { + "epoch": 1.55, + "learning_rate": 1.8630717540159562e-05, + "loss": 0.3987, + "step": 3697 + }, + { + "epoch": 1.55, + "learning_rate": 1.8629861096754977e-05, + "loss": 0.3406, + "step": 3698 + }, + { + "epoch": 1.55, + "learning_rate": 1.862900440529269e-05, + "loss": 0.3903, + "step": 3699 + }, + { + "epoch": 1.55, + "learning_rate": 1.862814746579733e-05, + "loss": 0.3315, + "step": 3700 + }, + { + "epoch": 1.55, + "learning_rate": 1.8627290278293525e-05, + "loss": 0.3851, + "step": 3701 + }, + { + "epoch": 1.55, + "learning_rate": 1.862643284280592e-05, + "loss": 0.3995, + "step": 3702 + }, + { + "epoch": 1.55, + "learning_rate": 1.8625575159359157e-05, + "loss": 0.393, + "step": 3703 + }, + { + "epoch": 1.55, + "learning_rate": 1.862471722797789e-05, + "loss": 0.405, + "step": 3704 + }, + { + "epoch": 1.55, + "learning_rate": 1.8623859048686777e-05, + "loss": 0.3704, + "step": 3705 + }, + { + "epoch": 1.55, + "learning_rate": 1.862300062151049e-05, + "loss": 0.3837, + "step": 3706 + }, + { + "epoch": 1.55, + "learning_rate": 1.8622141946473705e-05, + "loss": 0.3839, + "step": 3707 + }, + { + "epoch": 1.55, + "learning_rate": 1.8621283023601097e-05, + "loss": 0.4162, + "step": 3708 + }, + { + "epoch": 1.55, + "learning_rate": 1.862042385291736e-05, + "loss": 0.4516, + "step": 3709 + }, + { + "epoch": 1.55, + "learning_rate": 1.8619564434447192e-05, + "loss": 0.3776, + "step": 3710 + }, + { + "epoch": 1.55, + "learning_rate": 1.861870476821529e-05, + "loss": 0.4294, + "step": 3711 + }, + { + "epoch": 1.55, + "learning_rate": 1.8617844854246368e-05, + "loss": 0.3658, + "step": 3712 + }, + { + "epoch": 1.55, + "learning_rate": 1.8616984692565144e-05, + "loss": 0.3766, + "step": 3713 + }, + { + "epoch": 1.55, + "learning_rate": 1.861612428319634e-05, + "loss": 0.3958, + "step": 3714 + }, + { + "epoch": 1.56, + "learning_rate": 1.861526362616469e-05, + "loss": 0.4002, + "step": 3715 + }, + { + "epoch": 1.56, + "learning_rate": 1.8614402721494934e-05, + "loss": 0.3409, + "step": 3716 + }, + { + "epoch": 1.56, + "learning_rate": 1.8613541569211815e-05, + "loss": 0.3693, + "step": 3717 + }, + { + "epoch": 1.56, + "learning_rate": 1.8612680169340085e-05, + "loss": 0.3764, + "step": 3718 + }, + { + "epoch": 1.56, + "learning_rate": 1.861181852190451e-05, + "loss": 0.3525, + "step": 3719 + }, + { + "epoch": 1.56, + "learning_rate": 1.8610956626929852e-05, + "loss": 0.3591, + "step": 3720 + }, + { + "epoch": 1.56, + "learning_rate": 1.8610094484440884e-05, + "loss": 0.3761, + "step": 3721 + }, + { + "epoch": 1.56, + "learning_rate": 1.8609232094462394e-05, + "loss": 0.3549, + "step": 3722 + }, + { + "epoch": 1.56, + "learning_rate": 1.8608369457019166e-05, + "loss": 0.3992, + "step": 3723 + }, + { + "epoch": 1.56, + "learning_rate": 1.8607506572135994e-05, + "loss": 0.3731, + "step": 3724 + }, + { + "epoch": 1.56, + "learning_rate": 1.860664343983769e-05, + "loss": 0.3797, + "step": 3725 + }, + { + "epoch": 1.56, + "learning_rate": 1.8605780060149056e-05, + "loss": 0.3526, + "step": 3726 + }, + { + "epoch": 1.56, + "learning_rate": 1.8604916433094908e-05, + "loss": 0.3517, + "step": 3727 + }, + { + "epoch": 1.56, + "learning_rate": 1.860405255870007e-05, + "loss": 0.3643, + "step": 3728 + }, + { + "epoch": 1.56, + "learning_rate": 1.8603188436989382e-05, + "loss": 0.3997, + "step": 3729 + }, + { + "epoch": 1.56, + "learning_rate": 1.8602324067987678e-05, + "loss": 0.3371, + "step": 3730 + }, + { + "epoch": 1.56, + "learning_rate": 1.86014594517198e-05, + "loss": 0.4331, + "step": 3731 + }, + { + "epoch": 1.56, + "learning_rate": 1.8600594588210602e-05, + "loss": 0.4084, + "step": 3732 + }, + { + "epoch": 1.56, + "learning_rate": 1.8599729477484944e-05, + "loss": 0.3512, + "step": 3733 + }, + { + "epoch": 1.56, + "learning_rate": 1.8598864119567693e-05, + "loss": 0.4022, + "step": 3734 + }, + { + "epoch": 1.56, + "learning_rate": 1.8597998514483724e-05, + "loss": 0.3846, + "step": 3735 + }, + { + "epoch": 1.56, + "learning_rate": 1.859713266225792e-05, + "loss": 0.3828, + "step": 3736 + }, + { + "epoch": 1.56, + "learning_rate": 1.8596266562915163e-05, + "loss": 0.4152, + "step": 3737 + }, + { + "epoch": 1.56, + "learning_rate": 1.8595400216480356e-05, + "loss": 0.347, + "step": 3738 + }, + { + "epoch": 1.57, + "learning_rate": 1.8594533622978394e-05, + "loss": 0.4058, + "step": 3739 + }, + { + "epoch": 1.57, + "learning_rate": 1.859366678243419e-05, + "loss": 0.4159, + "step": 3740 + }, + { + "epoch": 1.57, + "learning_rate": 1.859279969487266e-05, + "loss": 0.3883, + "step": 3741 + }, + { + "epoch": 1.57, + "learning_rate": 1.859193236031873e-05, + "loss": 0.3622, + "step": 3742 + }, + { + "epoch": 1.57, + "learning_rate": 1.859106477879733e-05, + "loss": 0.3714, + "step": 3743 + }, + { + "epoch": 1.57, + "learning_rate": 1.859019695033339e-05, + "loss": 0.3634, + "step": 3744 + }, + { + "epoch": 1.57, + "learning_rate": 1.8589328874951866e-05, + "loss": 0.442, + "step": 3745 + }, + { + "epoch": 1.57, + "learning_rate": 1.8588460552677703e-05, + "loss": 0.426, + "step": 3746 + }, + { + "epoch": 1.57, + "learning_rate": 1.8587591983535868e-05, + "loss": 0.3693, + "step": 3747 + }, + { + "epoch": 1.57, + "learning_rate": 1.8586723167551315e-05, + "loss": 0.4352, + "step": 3748 + }, + { + "epoch": 1.57, + "learning_rate": 1.858585410474903e-05, + "loss": 0.3763, + "step": 3749 + }, + { + "epoch": 1.57, + "learning_rate": 1.8584984795153984e-05, + "loss": 0.4033, + "step": 3750 + }, + { + "epoch": 1.57, + "learning_rate": 1.858411523879117e-05, + "loss": 0.4202, + "step": 3751 + }, + { + "epoch": 1.57, + "learning_rate": 1.858324543568558e-05, + "loss": 0.4004, + "step": 3752 + }, + { + "epoch": 1.57, + "learning_rate": 1.8582375385862215e-05, + "loss": 0.4286, + "step": 3753 + }, + { + "epoch": 1.57, + "learning_rate": 1.858150508934609e-05, + "loss": 0.3823, + "step": 3754 + }, + { + "epoch": 1.57, + "learning_rate": 1.8580634546162212e-05, + "loss": 0.3807, + "step": 3755 + }, + { + "epoch": 1.57, + "learning_rate": 1.857976375633561e-05, + "loss": 0.3462, + "step": 3756 + }, + { + "epoch": 1.57, + "learning_rate": 1.8578892719891314e-05, + "loss": 0.379, + "step": 3757 + }, + { + "epoch": 1.57, + "learning_rate": 1.8578021436854357e-05, + "loss": 0.3501, + "step": 3758 + }, + { + "epoch": 1.57, + "learning_rate": 1.8577149907249786e-05, + "loss": 0.384, + "step": 3759 + }, + { + "epoch": 1.57, + "learning_rate": 1.8576278131102655e-05, + "loss": 0.3863, + "step": 3760 + }, + { + "epoch": 1.57, + "learning_rate": 1.8575406108438017e-05, + "loss": 0.3424, + "step": 3761 + }, + { + "epoch": 1.58, + "learning_rate": 1.857453383928094e-05, + "loss": 0.395, + "step": 3762 + }, + { + "epoch": 1.58, + "learning_rate": 1.8573661323656495e-05, + "loss": 0.3597, + "step": 3763 + }, + { + "epoch": 1.58, + "learning_rate": 1.8572788561589766e-05, + "loss": 0.391, + "step": 3764 + }, + { + "epoch": 1.58, + "learning_rate": 1.857191555310584e-05, + "loss": 0.3818, + "step": 3765 + }, + { + "epoch": 1.58, + "learning_rate": 1.85710422982298e-05, + "loss": 0.4242, + "step": 3766 + }, + { + "epoch": 1.58, + "learning_rate": 1.8570168796986764e-05, + "loss": 0.42, + "step": 3767 + }, + { + "epoch": 1.58, + "learning_rate": 1.8569295049401826e-05, + "loss": 0.3615, + "step": 3768 + }, + { + "epoch": 1.58, + "learning_rate": 1.8568421055500107e-05, + "loss": 0.3709, + "step": 3769 + }, + { + "epoch": 1.58, + "learning_rate": 1.8567546815306726e-05, + "loss": 0.3988, + "step": 3770 + }, + { + "epoch": 1.58, + "learning_rate": 1.8566672328846816e-05, + "loss": 0.3838, + "step": 3771 + }, + { + "epoch": 1.58, + "learning_rate": 1.8565797596145515e-05, + "loss": 0.3679, + "step": 3772 + }, + { + "epoch": 1.58, + "learning_rate": 1.8564922617227958e-05, + "loss": 0.3765, + "step": 3773 + }, + { + "epoch": 1.58, + "learning_rate": 1.8564047392119306e-05, + "loss": 0.4416, + "step": 3774 + }, + { + "epoch": 1.58, + "learning_rate": 1.8563171920844706e-05, + "loss": 0.3616, + "step": 3775 + }, + { + "epoch": 1.58, + "learning_rate": 1.856229620342933e-05, + "loss": 0.3814, + "step": 3776 + }, + { + "epoch": 1.58, + "learning_rate": 1.8561420239898348e-05, + "loss": 0.3511, + "step": 3777 + }, + { + "epoch": 1.58, + "learning_rate": 1.856054403027694e-05, + "loss": 0.3562, + "step": 3778 + }, + { + "epoch": 1.58, + "learning_rate": 1.855966757459029e-05, + "loss": 0.3642, + "step": 3779 + }, + { + "epoch": 1.58, + "learning_rate": 1.855879087286359e-05, + "loss": 0.4004, + "step": 3780 + }, + { + "epoch": 1.58, + "learning_rate": 1.8557913925122042e-05, + "loss": 0.3681, + "step": 3781 + }, + { + "epoch": 1.58, + "learning_rate": 1.8557036731390853e-05, + "loss": 0.3826, + "step": 3782 + }, + { + "epoch": 1.58, + "learning_rate": 1.8556159291695235e-05, + "loss": 0.3754, + "step": 3783 + }, + { + "epoch": 1.58, + "learning_rate": 1.855528160606041e-05, + "loss": 0.3771, + "step": 3784 + }, + { + "epoch": 1.58, + "learning_rate": 1.855440367451161e-05, + "loss": 0.4514, + "step": 3785 + }, + { + "epoch": 1.59, + "learning_rate": 1.8553525497074065e-05, + "loss": 0.5203, + "step": 3786 + }, + { + "epoch": 1.59, + "learning_rate": 1.855264707377302e-05, + "loss": 0.4171, + "step": 3787 + }, + { + "epoch": 1.59, + "learning_rate": 1.8551768404633724e-05, + "loss": 0.3869, + "step": 3788 + }, + { + "epoch": 1.59, + "learning_rate": 1.8550889489681435e-05, + "loss": 0.4159, + "step": 3789 + }, + { + "epoch": 1.59, + "learning_rate": 1.8550010328941417e-05, + "loss": 0.3714, + "step": 3790 + }, + { + "epoch": 1.59, + "learning_rate": 1.8549130922438938e-05, + "loss": 0.375, + "step": 3791 + }, + { + "epoch": 1.59, + "learning_rate": 1.8548251270199274e-05, + "loss": 0.4059, + "step": 3792 + }, + { + "epoch": 1.59, + "learning_rate": 1.854737137224772e-05, + "loss": 0.4218, + "step": 3793 + }, + { + "epoch": 1.59, + "learning_rate": 1.8546491228609552e-05, + "loss": 0.3925, + "step": 3794 + }, + { + "epoch": 1.59, + "learning_rate": 1.8545610839310085e-05, + "loss": 0.3834, + "step": 3795 + }, + { + "epoch": 1.59, + "learning_rate": 1.854473020437461e-05, + "loss": 0.3778, + "step": 3796 + }, + { + "epoch": 1.59, + "learning_rate": 1.8543849323828452e-05, + "loss": 0.427, + "step": 3797 + }, + { + "epoch": 1.59, + "learning_rate": 1.854296819769693e-05, + "loss": 0.3654, + "step": 3798 + }, + { + "epoch": 1.59, + "learning_rate": 1.8542086826005366e-05, + "loss": 0.3874, + "step": 3799 + }, + { + "epoch": 1.59, + "learning_rate": 1.8541205208779095e-05, + "loss": 0.432, + "step": 3800 + }, + { + "epoch": 1.59, + "learning_rate": 1.8540323346043458e-05, + "loss": 0.3539, + "step": 3801 + }, + { + "epoch": 1.59, + "learning_rate": 1.853944123782381e-05, + "loss": 0.421, + "step": 3802 + }, + { + "epoch": 1.59, + "learning_rate": 1.8538558884145497e-05, + "loss": 0.3837, + "step": 3803 + }, + { + "epoch": 1.59, + "learning_rate": 1.8537676285033886e-05, + "loss": 0.3663, + "step": 3804 + }, + { + "epoch": 1.59, + "learning_rate": 1.853679344051435e-05, + "loss": 0.3943, + "step": 3805 + }, + { + "epoch": 1.59, + "learning_rate": 1.8535910350612257e-05, + "loss": 0.3913, + "step": 3806 + }, + { + "epoch": 1.59, + "learning_rate": 1.8535027015353e-05, + "loss": 0.3979, + "step": 3807 + }, + { + "epoch": 1.59, + "learning_rate": 1.8534143434761965e-05, + "loss": 0.4269, + "step": 3808 + }, + { + "epoch": 1.59, + "learning_rate": 1.8533259608864547e-05, + "loss": 0.3765, + "step": 3809 + }, + { + "epoch": 1.6, + "learning_rate": 1.8532375537686155e-05, + "loss": 0.3696, + "step": 3810 + }, + { + "epoch": 1.6, + "learning_rate": 1.85314912212522e-05, + "loss": 0.4306, + "step": 3811 + }, + { + "epoch": 1.6, + "learning_rate": 1.85306066595881e-05, + "loss": 0.3987, + "step": 3812 + }, + { + "epoch": 1.6, + "learning_rate": 1.8529721852719283e-05, + "loss": 0.4022, + "step": 3813 + }, + { + "epoch": 1.6, + "learning_rate": 1.852883680067118e-05, + "loss": 0.4001, + "step": 3814 + }, + { + "epoch": 1.6, + "learning_rate": 1.852795150346923e-05, + "loss": 0.3886, + "step": 3815 + }, + { + "epoch": 1.6, + "learning_rate": 1.8527065961138885e-05, + "loss": 0.339, + "step": 3816 + }, + { + "epoch": 1.6, + "learning_rate": 1.852618017370559e-05, + "loss": 0.3473, + "step": 3817 + }, + { + "epoch": 1.6, + "learning_rate": 1.8525294141194816e-05, + "loss": 0.3534, + "step": 3818 + }, + { + "epoch": 1.6, + "learning_rate": 1.852440786363203e-05, + "loss": 0.3533, + "step": 3819 + }, + { + "epoch": 1.6, + "learning_rate": 1.85235213410427e-05, + "loss": 0.4324, + "step": 3820 + }, + { + "epoch": 1.6, + "learning_rate": 1.8522634573452314e-05, + "loss": 0.3952, + "step": 3821 + }, + { + "epoch": 1.6, + "learning_rate": 1.8521747560886357e-05, + "loss": 0.4197, + "step": 3822 + }, + { + "epoch": 1.6, + "learning_rate": 1.8520860303370333e-05, + "loss": 0.3743, + "step": 3823 + }, + { + "epoch": 1.6, + "learning_rate": 1.851997280092974e-05, + "loss": 0.3709, + "step": 3824 + }, + { + "epoch": 1.6, + "learning_rate": 1.8519085053590083e-05, + "loss": 0.4111, + "step": 3825 + }, + { + "epoch": 1.6, + "learning_rate": 1.8518197061376895e-05, + "loss": 0.4298, + "step": 3826 + }, + { + "epoch": 1.6, + "learning_rate": 1.8517308824315688e-05, + "loss": 0.3054, + "step": 3827 + }, + { + "epoch": 1.6, + "learning_rate": 1.8516420342431994e-05, + "loss": 0.3889, + "step": 3828 + }, + { + "epoch": 1.6, + "learning_rate": 1.8515531615751357e-05, + "loss": 0.387, + "step": 3829 + }, + { + "epoch": 1.6, + "learning_rate": 1.851464264429932e-05, + "loss": 0.3876, + "step": 3830 + }, + { + "epoch": 1.6, + "learning_rate": 1.8513753428101436e-05, + "loss": 0.3743, + "step": 3831 + }, + { + "epoch": 1.6, + "learning_rate": 1.8512863967183265e-05, + "loss": 0.4047, + "step": 3832 + }, + { + "epoch": 1.6, + "learning_rate": 1.8511974261570375e-05, + "loss": 0.3511, + "step": 3833 + }, + { + "epoch": 1.61, + "learning_rate": 1.8511084311288338e-05, + "loss": 0.3873, + "step": 3834 + }, + { + "epoch": 1.61, + "learning_rate": 1.851019411636273e-05, + "loss": 0.3827, + "step": 3835 + }, + { + "epoch": 1.61, + "learning_rate": 1.850930367681915e-05, + "loss": 0.4201, + "step": 3836 + }, + { + "epoch": 1.61, + "learning_rate": 1.8508412992683183e-05, + "loss": 0.3714, + "step": 3837 + }, + { + "epoch": 1.61, + "learning_rate": 1.8507522063980436e-05, + "loss": 0.4159, + "step": 3838 + }, + { + "epoch": 1.61, + "learning_rate": 1.8506630890736515e-05, + "loss": 0.4006, + "step": 3839 + }, + { + "epoch": 1.61, + "learning_rate": 1.8505739472977043e-05, + "loss": 0.3933, + "step": 3840 + }, + { + "epoch": 1.61, + "learning_rate": 1.8504847810727632e-05, + "loss": 0.3986, + "step": 3841 + }, + { + "epoch": 1.61, + "learning_rate": 1.8503955904013922e-05, + "loss": 0.3899, + "step": 3842 + }, + { + "epoch": 1.61, + "learning_rate": 1.8503063752861543e-05, + "loss": 0.4075, + "step": 3843 + }, + { + "epoch": 1.61, + "learning_rate": 1.8502171357296144e-05, + "loss": 0.3603, + "step": 3844 + }, + { + "epoch": 1.61, + "learning_rate": 1.8501278717343372e-05, + "loss": 0.4044, + "step": 3845 + }, + { + "epoch": 1.61, + "learning_rate": 1.850038583302889e-05, + "loss": 0.375, + "step": 3846 + }, + { + "epoch": 1.61, + "learning_rate": 1.8499492704378355e-05, + "loss": 0.3622, + "step": 3847 + }, + { + "epoch": 1.61, + "learning_rate": 1.849859933141745e-05, + "loss": 0.4129, + "step": 3848 + }, + { + "epoch": 1.61, + "learning_rate": 1.8497705714171846e-05, + "loss": 0.4007, + "step": 3849 + }, + { + "epoch": 1.61, + "learning_rate": 1.849681185266723e-05, + "loss": 0.3879, + "step": 3850 + }, + { + "epoch": 1.61, + "learning_rate": 1.8495917746929303e-05, + "loss": 0.4252, + "step": 3851 + }, + { + "epoch": 1.61, + "learning_rate": 1.8495023396983754e-05, + "loss": 0.3669, + "step": 3852 + }, + { + "epoch": 1.61, + "learning_rate": 1.84941288028563e-05, + "loss": 0.3351, + "step": 3853 + }, + { + "epoch": 1.61, + "learning_rate": 1.8493233964572648e-05, + "loss": 0.3886, + "step": 3854 + }, + { + "epoch": 1.61, + "learning_rate": 1.8492338882158526e-05, + "loss": 0.415, + "step": 3855 + }, + { + "epoch": 1.61, + "learning_rate": 1.8491443555639656e-05, + "loss": 0.4072, + "step": 3856 + }, + { + "epoch": 1.61, + "learning_rate": 1.8490547985041774e-05, + "loss": 0.349, + "step": 3857 + }, + { + "epoch": 1.62, + "learning_rate": 1.8489652170390627e-05, + "loss": 0.3835, + "step": 3858 + }, + { + "epoch": 1.62, + "learning_rate": 1.848875611171196e-05, + "loss": 0.4041, + "step": 3859 + }, + { + "epoch": 1.62, + "learning_rate": 1.8487859809031537e-05, + "loss": 0.4144, + "step": 3860 + }, + { + "epoch": 1.62, + "learning_rate": 1.8486963262375113e-05, + "loss": 0.4173, + "step": 3861 + }, + { + "epoch": 1.62, + "learning_rate": 1.8486066471768457e-05, + "loss": 0.329, + "step": 3862 + }, + { + "epoch": 1.62, + "learning_rate": 1.848516943723736e-05, + "loss": 0.4009, + "step": 3863 + }, + { + "epoch": 1.62, + "learning_rate": 1.8484272158807587e-05, + "loss": 0.3595, + "step": 3864 + }, + { + "epoch": 1.62, + "learning_rate": 1.8483374636504943e-05, + "loss": 0.4056, + "step": 3865 + }, + { + "epoch": 1.62, + "learning_rate": 1.8482476870355226e-05, + "loss": 0.3897, + "step": 3866 + }, + { + "epoch": 1.62, + "learning_rate": 1.848157886038424e-05, + "loss": 0.3094, + "step": 3867 + }, + { + "epoch": 1.62, + "learning_rate": 1.848068060661779e-05, + "loss": 0.3604, + "step": 3868 + }, + { + "epoch": 1.62, + "learning_rate": 1.8479782109081702e-05, + "loss": 0.3688, + "step": 3869 + }, + { + "epoch": 1.62, + "learning_rate": 1.8478883367801804e-05, + "loss": 0.3693, + "step": 3870 + }, + { + "epoch": 1.62, + "learning_rate": 1.8477984382803928e-05, + "loss": 0.4919, + "step": 3871 + }, + { + "epoch": 1.62, + "learning_rate": 1.8477085154113913e-05, + "loss": 0.3725, + "step": 3872 + }, + { + "epoch": 1.62, + "learning_rate": 1.847618568175761e-05, + "loss": 0.3605, + "step": 3873 + }, + { + "epoch": 1.62, + "learning_rate": 1.847528596576087e-05, + "loss": 0.4308, + "step": 3874 + }, + { + "epoch": 1.62, + "learning_rate": 1.8474386006149556e-05, + "loss": 0.403, + "step": 3875 + }, + { + "epoch": 1.62, + "learning_rate": 1.847348580294954e-05, + "loss": 0.4061, + "step": 3876 + }, + { + "epoch": 1.62, + "learning_rate": 1.8472585356186688e-05, + "loss": 0.4084, + "step": 3877 + }, + { + "epoch": 1.62, + "learning_rate": 1.847168466588689e-05, + "loss": 0.4078, + "step": 3878 + }, + { + "epoch": 1.62, + "learning_rate": 1.8470783732076034e-05, + "loss": 0.3814, + "step": 3879 + }, + { + "epoch": 1.62, + "learning_rate": 1.8469882554780016e-05, + "loss": 0.4037, + "step": 3880 + }, + { + "epoch": 1.62, + "learning_rate": 1.846898113402474e-05, + "loss": 0.3961, + "step": 3881 + }, + { + "epoch": 1.63, + "learning_rate": 1.8468079469836117e-05, + "loss": 0.3874, + "step": 3882 + }, + { + "epoch": 1.63, + "learning_rate": 1.8467177562240062e-05, + "loss": 0.3401, + "step": 3883 + }, + { + "epoch": 1.63, + "learning_rate": 1.8466275411262504e-05, + "loss": 0.3788, + "step": 3884 + }, + { + "epoch": 1.63, + "learning_rate": 1.8465373016929377e-05, + "loss": 0.3697, + "step": 3885 + }, + { + "epoch": 1.63, + "learning_rate": 1.8464470379266606e-05, + "loss": 0.3385, + "step": 3886 + }, + { + "epoch": 1.63, + "learning_rate": 1.846356749830015e-05, + "loss": 0.4286, + "step": 3887 + }, + { + "epoch": 1.63, + "learning_rate": 1.8462664374055953e-05, + "loss": 0.3876, + "step": 3888 + }, + { + "epoch": 1.63, + "learning_rate": 1.8461761006559982e-05, + "loss": 0.3762, + "step": 3889 + }, + { + "epoch": 1.63, + "learning_rate": 1.84608573958382e-05, + "loss": 0.3909, + "step": 3890 + }, + { + "epoch": 1.63, + "learning_rate": 1.8459953541916576e-05, + "loss": 0.3997, + "step": 3891 + }, + { + "epoch": 1.63, + "learning_rate": 1.84590494448211e-05, + "loss": 0.3785, + "step": 3892 + }, + { + "epoch": 1.63, + "learning_rate": 1.8458145104577752e-05, + "loss": 0.4019, + "step": 3893 + }, + { + "epoch": 1.63, + "learning_rate": 1.8457240521212525e-05, + "loss": 0.3651, + "step": 3894 + }, + { + "epoch": 1.63, + "learning_rate": 1.845633569475143e-05, + "loss": 0.3841, + "step": 3895 + }, + { + "epoch": 1.63, + "learning_rate": 1.8455430625220465e-05, + "loss": 0.4117, + "step": 3896 + }, + { + "epoch": 1.63, + "learning_rate": 1.8454525312645654e-05, + "loss": 0.4056, + "step": 3897 + }, + { + "epoch": 1.63, + "learning_rate": 1.845361975705301e-05, + "loss": 0.379, + "step": 3898 + }, + { + "epoch": 1.63, + "learning_rate": 1.845271395846857e-05, + "loss": 0.4022, + "step": 3899 + }, + { + "epoch": 1.63, + "learning_rate": 1.8451807916918367e-05, + "loss": 0.4109, + "step": 3900 + }, + { + "epoch": 1.63, + "learning_rate": 1.845090163242845e-05, + "loss": 0.3745, + "step": 3901 + }, + { + "epoch": 1.63, + "learning_rate": 1.844999510502486e-05, + "loss": 0.4296, + "step": 3902 + }, + { + "epoch": 1.63, + "learning_rate": 1.844908833473366e-05, + "loss": 0.4323, + "step": 3903 + }, + { + "epoch": 1.63, + "learning_rate": 1.8448181321580916e-05, + "loss": 0.392, + "step": 3904 + }, + { + "epoch": 1.63, + "learning_rate": 1.8447274065592694e-05, + "loss": 0.4409, + "step": 3905 + }, + { + "epoch": 1.64, + "learning_rate": 1.8446366566795075e-05, + "loss": 0.4219, + "step": 3906 + }, + { + "epoch": 1.64, + "learning_rate": 1.8445458825214148e-05, + "loss": 0.4145, + "step": 3907 + }, + { + "epoch": 1.64, + "learning_rate": 1.8444550840875996e-05, + "loss": 0.3887, + "step": 3908 + }, + { + "epoch": 1.64, + "learning_rate": 1.8443642613806727e-05, + "loss": 0.3598, + "step": 3909 + }, + { + "epoch": 1.64, + "learning_rate": 1.8442734144032442e-05, + "loss": 0.3373, + "step": 3910 + }, + { + "epoch": 1.64, + "learning_rate": 1.844182543157926e-05, + "loss": 0.4054, + "step": 3911 + }, + { + "epoch": 1.64, + "learning_rate": 1.844091647647329e-05, + "loss": 0.3754, + "step": 3912 + }, + { + "epoch": 1.64, + "learning_rate": 1.8440007278740673e-05, + "loss": 0.4236, + "step": 3913 + }, + { + "epoch": 1.64, + "learning_rate": 1.8439097838407532e-05, + "loss": 0.3872, + "step": 3914 + }, + { + "epoch": 1.64, + "learning_rate": 1.8438188155500016e-05, + "loss": 0.3718, + "step": 3915 + }, + { + "epoch": 1.64, + "learning_rate": 1.8437278230044266e-05, + "loss": 0.4069, + "step": 3916 + }, + { + "epoch": 1.64, + "learning_rate": 1.8436368062066442e-05, + "loss": 0.4603, + "step": 3917 + }, + { + "epoch": 1.64, + "learning_rate": 1.8435457651592704e-05, + "loss": 0.382, + "step": 3918 + }, + { + "epoch": 1.64, + "learning_rate": 1.843454699864922e-05, + "loss": 0.3786, + "step": 3919 + }, + { + "epoch": 1.64, + "learning_rate": 1.843363610326217e-05, + "loss": 0.3877, + "step": 3920 + }, + { + "epoch": 1.64, + "learning_rate": 1.843272496545773e-05, + "loss": 0.3959, + "step": 3921 + }, + { + "epoch": 1.64, + "learning_rate": 1.8431813585262098e-05, + "loss": 0.3778, + "step": 3922 + }, + { + "epoch": 1.64, + "learning_rate": 1.8430901962701465e-05, + "loss": 0.4074, + "step": 3923 + }, + { + "epoch": 1.64, + "learning_rate": 1.8429990097802037e-05, + "loss": 0.403, + "step": 3924 + }, + { + "epoch": 1.64, + "learning_rate": 1.8429077990590026e-05, + "loss": 0.4485, + "step": 3925 + }, + { + "epoch": 1.64, + "learning_rate": 1.8428165641091643e-05, + "loss": 0.3935, + "step": 3926 + }, + { + "epoch": 1.64, + "learning_rate": 1.8427253049333124e-05, + "loss": 0.3865, + "step": 3927 + }, + { + "epoch": 1.64, + "learning_rate": 1.8426340215340688e-05, + "loss": 0.3605, + "step": 3928 + }, + { + "epoch": 1.64, + "learning_rate": 1.8425427139140586e-05, + "loss": 0.38, + "step": 3929 + }, + { + "epoch": 1.65, + "learning_rate": 1.8424513820759054e-05, + "loss": 0.4266, + "step": 3930 + }, + { + "epoch": 1.65, + "learning_rate": 1.8423600260222346e-05, + "loss": 0.4254, + "step": 3931 + }, + { + "epoch": 1.65, + "learning_rate": 1.8422686457556726e-05, + "loss": 0.398, + "step": 3932 + }, + { + "epoch": 1.65, + "learning_rate": 1.842177241278846e-05, + "loss": 0.3574, + "step": 3933 + }, + { + "epoch": 1.65, + "learning_rate": 1.8420858125943818e-05, + "loss": 0.3633, + "step": 3934 + }, + { + "epoch": 1.65, + "learning_rate": 1.841994359704908e-05, + "loss": 0.4486, + "step": 3935 + }, + { + "epoch": 1.65, + "learning_rate": 1.8419028826130537e-05, + "loss": 0.3329, + "step": 3936 + }, + { + "epoch": 1.65, + "learning_rate": 1.8418113813214477e-05, + "loss": 0.5135, + "step": 3937 + }, + { + "epoch": 1.65, + "learning_rate": 1.841719855832721e-05, + "loss": 0.4119, + "step": 3938 + }, + { + "epoch": 1.65, + "learning_rate": 1.8416283061495038e-05, + "loss": 0.3705, + "step": 3939 + }, + { + "epoch": 1.65, + "learning_rate": 1.841536732274428e-05, + "loss": 0.3832, + "step": 3940 + }, + { + "epoch": 1.65, + "learning_rate": 1.8414451342101253e-05, + "loss": 0.3343, + "step": 3941 + }, + { + "epoch": 1.65, + "learning_rate": 1.8413535119592292e-05, + "loss": 0.364, + "step": 3942 + }, + { + "epoch": 1.65, + "learning_rate": 1.841261865524373e-05, + "loss": 0.3662, + "step": 3943 + }, + { + "epoch": 1.65, + "learning_rate": 1.8411701949081906e-05, + "loss": 0.4828, + "step": 3944 + }, + { + "epoch": 1.65, + "learning_rate": 1.8410785001133177e-05, + "loss": 0.3881, + "step": 3945 + }, + { + "epoch": 1.65, + "learning_rate": 1.8409867811423897e-05, + "loss": 0.4021, + "step": 3946 + }, + { + "epoch": 1.65, + "learning_rate": 1.840895037998043e-05, + "loss": 0.4208, + "step": 3947 + }, + { + "epoch": 1.65, + "learning_rate": 1.840803270682914e-05, + "loss": 0.3583, + "step": 3948 + }, + { + "epoch": 1.65, + "learning_rate": 1.8407114791996416e-05, + "loss": 0.3859, + "step": 3949 + }, + { + "epoch": 1.65, + "learning_rate": 1.840619663550864e-05, + "loss": 0.373, + "step": 3950 + }, + { + "epoch": 1.65, + "learning_rate": 1.84052782373922e-05, + "loss": 0.3767, + "step": 3951 + }, + { + "epoch": 1.65, + "learning_rate": 1.8404359597673492e-05, + "loss": 0.3835, + "step": 3952 + }, + { + "epoch": 1.66, + "learning_rate": 1.840344071637893e-05, + "loss": 0.3904, + "step": 3953 + }, + { + "epoch": 1.66, + "learning_rate": 1.840252159353492e-05, + "loss": 0.3753, + "step": 3954 + }, + { + "epoch": 1.66, + "learning_rate": 1.840160222916788e-05, + "loss": 0.3811, + "step": 3955 + }, + { + "epoch": 1.66, + "learning_rate": 1.8400682623304243e-05, + "loss": 0.362, + "step": 3956 + }, + { + "epoch": 1.66, + "learning_rate": 1.839976277597044e-05, + "loss": 0.4012, + "step": 3957 + }, + { + "epoch": 1.66, + "learning_rate": 1.8398842687192906e-05, + "loss": 0.408, + "step": 3958 + }, + { + "epoch": 1.66, + "learning_rate": 1.839792235699809e-05, + "loss": 0.3913, + "step": 3959 + }, + { + "epoch": 1.66, + "learning_rate": 1.8397001785412455e-05, + "loss": 0.3546, + "step": 3960 + }, + { + "epoch": 1.66, + "learning_rate": 1.839608097246245e-05, + "loss": 0.3846, + "step": 3961 + }, + { + "epoch": 1.66, + "learning_rate": 1.839515991817455e-05, + "loss": 0.3486, + "step": 3962 + }, + { + "epoch": 1.66, + "learning_rate": 1.8394238622575228e-05, + "loss": 0.4618, + "step": 3963 + }, + { + "epoch": 1.66, + "learning_rate": 1.8393317085690964e-05, + "loss": 0.3483, + "step": 3964 + }, + { + "epoch": 1.66, + "learning_rate": 1.839239530754825e-05, + "loss": 0.4279, + "step": 3965 + }, + { + "epoch": 1.66, + "learning_rate": 1.839147328817358e-05, + "loss": 0.4616, + "step": 3966 + }, + { + "epoch": 1.66, + "learning_rate": 1.8390551027593456e-05, + "loss": 0.4158, + "step": 3967 + }, + { + "epoch": 1.66, + "learning_rate": 1.8389628525834385e-05, + "loss": 0.3667, + "step": 3968 + }, + { + "epoch": 1.66, + "learning_rate": 1.8388705782922893e-05, + "loss": 0.4405, + "step": 3969 + }, + { + "epoch": 1.66, + "learning_rate": 1.8387782798885495e-05, + "loss": 0.4028, + "step": 3970 + }, + { + "epoch": 1.66, + "learning_rate": 1.8386859573748724e-05, + "loss": 0.3949, + "step": 3971 + }, + { + "epoch": 1.66, + "learning_rate": 1.8385936107539115e-05, + "loss": 0.3959, + "step": 3972 + }, + { + "epoch": 1.66, + "learning_rate": 1.8385012400283218e-05, + "loss": 0.4073, + "step": 3973 + }, + { + "epoch": 1.66, + "learning_rate": 1.838408845200758e-05, + "loss": 0.3725, + "step": 3974 + }, + { + "epoch": 1.66, + "learning_rate": 1.8383164262738756e-05, + "loss": 0.3761, + "step": 3975 + }, + { + "epoch": 1.66, + "learning_rate": 1.838223983250331e-05, + "loss": 0.3953, + "step": 3976 + }, + { + "epoch": 1.67, + "learning_rate": 1.8381315161327827e-05, + "loss": 0.3925, + "step": 3977 + }, + { + "epoch": 1.67, + "learning_rate": 1.8380390249238876e-05, + "loss": 0.3745, + "step": 3978 + }, + { + "epoch": 1.67, + "learning_rate": 1.837946509626304e-05, + "loss": 0.3934, + "step": 3979 + }, + { + "epoch": 1.67, + "learning_rate": 1.8378539702426917e-05, + "loss": 0.348, + "step": 3980 + }, + { + "epoch": 1.67, + "learning_rate": 1.8377614067757106e-05, + "loss": 0.3714, + "step": 3981 + }, + { + "epoch": 1.67, + "learning_rate": 1.8376688192280213e-05, + "loss": 0.384, + "step": 3982 + }, + { + "epoch": 1.67, + "learning_rate": 1.837576207602285e-05, + "loss": 0.3941, + "step": 3983 + }, + { + "epoch": 1.67, + "learning_rate": 1.837483571901164e-05, + "loss": 0.3607, + "step": 3984 + }, + { + "epoch": 1.67, + "learning_rate": 1.837390912127321e-05, + "loss": 0.3494, + "step": 3985 + }, + { + "epoch": 1.67, + "learning_rate": 1.837298228283419e-05, + "loss": 0.3867, + "step": 3986 + }, + { + "epoch": 1.67, + "learning_rate": 1.837205520372123e-05, + "loss": 0.369, + "step": 3987 + }, + { + "epoch": 1.67, + "learning_rate": 1.8371127883960963e-05, + "loss": 0.3949, + "step": 3988 + }, + { + "epoch": 1.67, + "learning_rate": 1.837020032358006e-05, + "loss": 0.3878, + "step": 3989 + }, + { + "epoch": 1.67, + "learning_rate": 1.8369272522605177e-05, + "loss": 0.4191, + "step": 3990 + }, + { + "epoch": 1.67, + "learning_rate": 1.836834448106298e-05, + "loss": 0.3945, + "step": 3991 + }, + { + "epoch": 1.67, + "learning_rate": 1.836741619898015e-05, + "loss": 0.4298, + "step": 3992 + }, + { + "epoch": 1.67, + "learning_rate": 1.836648767638336e-05, + "loss": 0.3937, + "step": 3993 + }, + { + "epoch": 1.67, + "learning_rate": 1.8365558913299307e-05, + "loss": 0.4084, + "step": 3994 + }, + { + "epoch": 1.67, + "learning_rate": 1.836462990975469e-05, + "loss": 0.3711, + "step": 3995 + }, + { + "epoch": 1.67, + "learning_rate": 1.836370066577621e-05, + "loss": 0.3859, + "step": 3996 + }, + { + "epoch": 1.67, + "learning_rate": 1.836277118139057e-05, + "loss": 0.3787, + "step": 3997 + }, + { + "epoch": 1.67, + "learning_rate": 1.83618414566245e-05, + "loss": 0.387, + "step": 3998 + }, + { + "epoch": 1.67, + "learning_rate": 1.8360911491504714e-05, + "loss": 0.3551, + "step": 3999 + }, + { + "epoch": 1.67, + "learning_rate": 1.835998128605795e-05, + "loss": 0.409, + "step": 4000 + }, + { + "epoch": 1.68, + "learning_rate": 1.8359050840310936e-05, + "loss": 0.3925, + "step": 4001 + }, + { + "epoch": 1.68, + "learning_rate": 1.8358120154290427e-05, + "loss": 0.4395, + "step": 4002 + }, + { + "epoch": 1.68, + "learning_rate": 1.8357189228023174e-05, + "loss": 0.425, + "step": 4003 + }, + { + "epoch": 1.68, + "learning_rate": 1.835625806153593e-05, + "loss": 0.3778, + "step": 4004 + }, + { + "epoch": 1.68, + "learning_rate": 1.835532665485546e-05, + "loss": 0.3671, + "step": 4005 + }, + { + "epoch": 1.68, + "learning_rate": 1.8354395008008545e-05, + "loss": 0.4004, + "step": 4006 + }, + { + "epoch": 1.68, + "learning_rate": 1.835346312102196e-05, + "loss": 0.3547, + "step": 4007 + }, + { + "epoch": 1.68, + "learning_rate": 1.8352530993922483e-05, + "loss": 0.4045, + "step": 4008 + }, + { + "epoch": 1.68, + "learning_rate": 1.835159862673692e-05, + "loss": 0.3707, + "step": 4009 + }, + { + "epoch": 1.68, + "learning_rate": 1.8350666019492065e-05, + "loss": 0.3821, + "step": 4010 + }, + { + "epoch": 1.68, + "learning_rate": 1.8349733172214722e-05, + "loss": 0.4071, + "step": 4011 + }, + { + "epoch": 1.68, + "learning_rate": 1.8348800084931712e-05, + "loss": 0.4002, + "step": 4012 + }, + { + "epoch": 1.68, + "learning_rate": 1.834786675766985e-05, + "loss": 0.3747, + "step": 4013 + }, + { + "epoch": 1.68, + "learning_rate": 1.834693319045597e-05, + "loss": 0.4455, + "step": 4014 + }, + { + "epoch": 1.68, + "learning_rate": 1.8345999383316897e-05, + "loss": 0.3962, + "step": 4015 + }, + { + "epoch": 1.68, + "learning_rate": 1.8345065336279482e-05, + "loss": 0.3184, + "step": 4016 + }, + { + "epoch": 1.68, + "learning_rate": 1.8344131049370566e-05, + "loss": 0.3856, + "step": 4017 + }, + { + "epoch": 1.68, + "learning_rate": 1.834319652261701e-05, + "loss": 0.4466, + "step": 4018 + }, + { + "epoch": 1.68, + "learning_rate": 1.8342261756045673e-05, + "loss": 0.4317, + "step": 4019 + }, + { + "epoch": 1.68, + "learning_rate": 1.834132674968342e-05, + "loss": 0.3718, + "step": 4020 + }, + { + "epoch": 1.68, + "learning_rate": 1.834039150355714e-05, + "loss": 0.3773, + "step": 4021 + }, + { + "epoch": 1.68, + "learning_rate": 1.83394560176937e-05, + "loss": 0.3929, + "step": 4022 + }, + { + "epoch": 1.68, + "learning_rate": 1.8338520292120002e-05, + "loss": 0.4016, + "step": 4023 + }, + { + "epoch": 1.68, + "learning_rate": 1.8337584326862935e-05, + "loss": 0.4165, + "step": 4024 + }, + { + "epoch": 1.69, + "learning_rate": 1.8336648121949405e-05, + "loss": 0.3676, + "step": 4025 + }, + { + "epoch": 1.69, + "learning_rate": 1.8335711677406324e-05, + "loss": 0.3762, + "step": 4026 + }, + { + "epoch": 1.69, + "learning_rate": 1.833477499326061e-05, + "loss": 0.4556, + "step": 4027 + }, + { + "epoch": 1.69, + "learning_rate": 1.833383806953918e-05, + "loss": 0.3767, + "step": 4028 + }, + { + "epoch": 1.69, + "learning_rate": 1.8332900906268972e-05, + "loss": 0.3779, + "step": 4029 + }, + { + "epoch": 1.69, + "learning_rate": 1.8331963503476925e-05, + "loss": 0.3795, + "step": 4030 + }, + { + "epoch": 1.69, + "learning_rate": 1.8331025861189974e-05, + "loss": 0.4227, + "step": 4031 + }, + { + "epoch": 1.69, + "learning_rate": 1.8330087979435083e-05, + "loss": 0.3875, + "step": 4032 + }, + { + "epoch": 1.69, + "learning_rate": 1.8329149858239204e-05, + "loss": 0.4157, + "step": 4033 + }, + { + "epoch": 1.69, + "learning_rate": 1.8328211497629304e-05, + "loss": 0.4011, + "step": 4034 + }, + { + "epoch": 1.69, + "learning_rate": 1.832727289763235e-05, + "loss": 0.3993, + "step": 4035 + }, + { + "epoch": 1.69, + "learning_rate": 1.832633405827533e-05, + "loss": 0.3993, + "step": 4036 + }, + { + "epoch": 1.69, + "learning_rate": 1.8325394979585227e-05, + "loss": 0.4001, + "step": 4037 + }, + { + "epoch": 1.69, + "learning_rate": 1.8324455661589033e-05, + "loss": 0.3964, + "step": 4038 + }, + { + "epoch": 1.69, + "learning_rate": 1.8323516104313746e-05, + "loss": 0.424, + "step": 4039 + }, + { + "epoch": 1.69, + "learning_rate": 1.8322576307786376e-05, + "loss": 0.4006, + "step": 4040 + }, + { + "epoch": 1.69, + "learning_rate": 1.8321636272033934e-05, + "loss": 0.3782, + "step": 4041 + }, + { + "epoch": 1.69, + "learning_rate": 1.8320695997083443e-05, + "loss": 0.3963, + "step": 4042 + }, + { + "epoch": 1.69, + "learning_rate": 1.831975548296193e-05, + "loss": 0.3684, + "step": 4043 + }, + { + "epoch": 1.69, + "learning_rate": 1.8318814729696427e-05, + "loss": 0.4097, + "step": 4044 + }, + { + "epoch": 1.69, + "learning_rate": 1.8317873737313977e-05, + "loss": 0.3876, + "step": 4045 + }, + { + "epoch": 1.69, + "learning_rate": 1.831693250584163e-05, + "loss": 0.4022, + "step": 4046 + }, + { + "epoch": 1.69, + "learning_rate": 1.8315991035306435e-05, + "loss": 0.3946, + "step": 4047 + }, + { + "epoch": 1.69, + "learning_rate": 1.831504932573546e-05, + "loss": 0.4278, + "step": 4048 + }, + { + "epoch": 1.7, + "learning_rate": 1.831410737715577e-05, + "loss": 0.391, + "step": 4049 + }, + { + "epoch": 1.7, + "learning_rate": 1.8313165189594438e-05, + "loss": 0.3522, + "step": 4050 + }, + { + "epoch": 1.7, + "learning_rate": 1.8312222763078552e-05, + "loss": 0.3493, + "step": 4051 + }, + { + "epoch": 1.7, + "learning_rate": 1.83112800976352e-05, + "loss": 0.3808, + "step": 4052 + }, + { + "epoch": 1.7, + "learning_rate": 1.8310337193291473e-05, + "loss": 0.376, + "step": 4053 + }, + { + "epoch": 1.7, + "learning_rate": 1.8309394050074482e-05, + "loss": 0.3485, + "step": 4054 + }, + { + "epoch": 1.7, + "learning_rate": 1.8308450668011334e-05, + "loss": 0.3439, + "step": 4055 + }, + { + "epoch": 1.7, + "learning_rate": 1.830750704712914e-05, + "loss": 0.4055, + "step": 4056 + }, + { + "epoch": 1.7, + "learning_rate": 1.830656318745503e-05, + "loss": 0.4463, + "step": 4057 + }, + { + "epoch": 1.7, + "learning_rate": 1.830561908901613e-05, + "loss": 0.3713, + "step": 4058 + }, + { + "epoch": 1.7, + "learning_rate": 1.8304674751839583e-05, + "loss": 0.3569, + "step": 4059 + }, + { + "epoch": 1.7, + "learning_rate": 1.830373017595253e-05, + "loss": 0.3846, + "step": 4060 + }, + { + "epoch": 1.7, + "learning_rate": 1.8302785361382117e-05, + "loss": 0.4214, + "step": 4061 + }, + { + "epoch": 1.7, + "learning_rate": 1.8301840308155507e-05, + "loss": 0.3752, + "step": 4062 + }, + { + "epoch": 1.7, + "learning_rate": 1.8300895016299868e-05, + "loss": 0.4175, + "step": 4063 + }, + { + "epoch": 1.7, + "learning_rate": 1.8299949485842366e-05, + "loss": 0.4112, + "step": 4064 + }, + { + "epoch": 1.7, + "learning_rate": 1.829900371681018e-05, + "loss": 0.436, + "step": 4065 + }, + { + "epoch": 1.7, + "learning_rate": 1.8298057709230498e-05, + "loss": 0.4509, + "step": 4066 + }, + { + "epoch": 1.7, + "learning_rate": 1.8297111463130512e-05, + "loss": 0.4404, + "step": 4067 + }, + { + "epoch": 1.7, + "learning_rate": 1.8296164978537422e-05, + "loss": 0.3855, + "step": 4068 + }, + { + "epoch": 1.7, + "learning_rate": 1.8295218255478425e-05, + "loss": 0.3585, + "step": 4069 + }, + { + "epoch": 1.7, + "learning_rate": 1.8294271293980744e-05, + "loss": 0.4391, + "step": 4070 + }, + { + "epoch": 1.7, + "learning_rate": 1.8293324094071593e-05, + "loss": 0.34, + "step": 4071 + }, + { + "epoch": 1.7, + "learning_rate": 1.8292376655778202e-05, + "loss": 0.3956, + "step": 4072 + }, + { + "epoch": 1.71, + "learning_rate": 1.8291428979127802e-05, + "loss": 0.3659, + "step": 4073 + }, + { + "epoch": 1.71, + "learning_rate": 1.8290481064147637e-05, + "loss": 0.3756, + "step": 4074 + }, + { + "epoch": 1.71, + "learning_rate": 1.8289532910864943e-05, + "loss": 0.357, + "step": 4075 + }, + { + "epoch": 1.71, + "learning_rate": 1.8288584519306985e-05, + "loss": 0.3498, + "step": 4076 + }, + { + "epoch": 1.71, + "learning_rate": 1.8287635889501024e-05, + "loss": 0.3925, + "step": 4077 + }, + { + "epoch": 1.71, + "learning_rate": 1.828668702147432e-05, + "loss": 0.3581, + "step": 4078 + }, + { + "epoch": 1.71, + "learning_rate": 1.828573791525415e-05, + "loss": 0.372, + "step": 4079 + }, + { + "epoch": 1.71, + "learning_rate": 1.82847885708678e-05, + "loss": 0.388, + "step": 4080 + }, + { + "epoch": 1.71, + "learning_rate": 1.8283838988342553e-05, + "loss": 0.3954, + "step": 4081 + }, + { + "epoch": 1.71, + "learning_rate": 1.8282889167705704e-05, + "loss": 0.3438, + "step": 4082 + }, + { + "epoch": 1.71, + "learning_rate": 1.8281939108984557e-05, + "loss": 0.3651, + "step": 4083 + }, + { + "epoch": 1.71, + "learning_rate": 1.828098881220642e-05, + "loss": 0.3769, + "step": 4084 + }, + { + "epoch": 1.71, + "learning_rate": 1.8280038277398606e-05, + "loss": 0.39, + "step": 4085 + }, + { + "epoch": 1.71, + "learning_rate": 1.827908750458844e-05, + "loss": 0.3994, + "step": 4086 + }, + { + "epoch": 1.71, + "learning_rate": 1.8278136493803253e-05, + "loss": 0.3713, + "step": 4087 + }, + { + "epoch": 1.71, + "learning_rate": 1.8277185245070373e-05, + "loss": 0.4595, + "step": 4088 + }, + { + "epoch": 1.71, + "learning_rate": 1.8276233758417153e-05, + "loss": 0.4077, + "step": 4089 + }, + { + "epoch": 1.71, + "learning_rate": 1.827528203387094e-05, + "loss": 0.4482, + "step": 4090 + }, + { + "epoch": 1.71, + "learning_rate": 1.8274330071459082e-05, + "loss": 0.3704, + "step": 4091 + }, + { + "epoch": 1.71, + "learning_rate": 1.8273377871208952e-05, + "loss": 0.392, + "step": 4092 + }, + { + "epoch": 1.71, + "learning_rate": 1.8272425433147916e-05, + "loss": 0.3883, + "step": 4093 + }, + { + "epoch": 1.71, + "learning_rate": 1.8271472757303352e-05, + "loss": 0.3909, + "step": 4094 + }, + { + "epoch": 1.71, + "learning_rate": 1.827051984370264e-05, + "loss": 0.4504, + "step": 4095 + }, + { + "epoch": 1.71, + "learning_rate": 1.826956669237318e-05, + "loss": 0.3842, + "step": 4096 + }, + { + "epoch": 1.72, + "learning_rate": 1.8268613303342362e-05, + "loss": 0.3769, + "step": 4097 + }, + { + "epoch": 1.72, + "learning_rate": 1.826765967663759e-05, + "loss": 0.3996, + "step": 4098 + }, + { + "epoch": 1.72, + "learning_rate": 1.8266705812286276e-05, + "loss": 0.3751, + "step": 4099 + }, + { + "epoch": 1.72, + "learning_rate": 1.8265751710315843e-05, + "loss": 0.3925, + "step": 4100 + }, + { + "epoch": 1.72, + "learning_rate": 1.826479737075371e-05, + "loss": 0.3749, + "step": 4101 + }, + { + "epoch": 1.72, + "learning_rate": 1.8263842793627314e-05, + "loss": 0.3599, + "step": 4102 + }, + { + "epoch": 1.72, + "learning_rate": 1.8262887978964087e-05, + "loss": 0.4009, + "step": 4103 + }, + { + "epoch": 1.72, + "learning_rate": 1.8261932926791477e-05, + "loss": 0.3794, + "step": 4104 + }, + { + "epoch": 1.72, + "learning_rate": 1.8260977637136937e-05, + "loss": 0.3634, + "step": 4105 + }, + { + "epoch": 1.72, + "learning_rate": 1.8260022110027925e-05, + "loss": 0.3815, + "step": 4106 + }, + { + "epoch": 1.72, + "learning_rate": 1.8259066345491908e-05, + "loss": 0.447, + "step": 4107 + }, + { + "epoch": 1.72, + "learning_rate": 1.8258110343556358e-05, + "loss": 0.4141, + "step": 4108 + }, + { + "epoch": 1.72, + "learning_rate": 1.8257154104248752e-05, + "loss": 0.407, + "step": 4109 + }, + { + "epoch": 1.72, + "learning_rate": 1.8256197627596582e-05, + "loss": 0.395, + "step": 4110 + }, + { + "epoch": 1.72, + "learning_rate": 1.8255240913627335e-05, + "loss": 0.4496, + "step": 4111 + }, + { + "epoch": 1.72, + "learning_rate": 1.8254283962368517e-05, + "loss": 0.366, + "step": 4112 + }, + { + "epoch": 1.72, + "learning_rate": 1.8253326773847628e-05, + "loss": 0.3714, + "step": 4113 + }, + { + "epoch": 1.72, + "learning_rate": 1.8252369348092184e-05, + "loss": 0.3777, + "step": 4114 + }, + { + "epoch": 1.72, + "learning_rate": 1.825141168512971e-05, + "loss": 0.4085, + "step": 4115 + }, + { + "epoch": 1.72, + "learning_rate": 1.825045378498773e-05, + "loss": 0.4169, + "step": 4116 + }, + { + "epoch": 1.72, + "learning_rate": 1.824949564769377e-05, + "loss": 0.4316, + "step": 4117 + }, + { + "epoch": 1.72, + "learning_rate": 1.8248537273275385e-05, + "loss": 0.3995, + "step": 4118 + }, + { + "epoch": 1.72, + "learning_rate": 1.8247578661760114e-05, + "loss": 0.4057, + "step": 4119 + }, + { + "epoch": 1.72, + "learning_rate": 1.824661981317551e-05, + "loss": 0.389, + "step": 4120 + }, + { + "epoch": 1.73, + "learning_rate": 1.8245660727549142e-05, + "loss": 0.4068, + "step": 4121 + }, + { + "epoch": 1.73, + "learning_rate": 1.824470140490857e-05, + "loss": 0.4061, + "step": 4122 + }, + { + "epoch": 1.73, + "learning_rate": 1.8243741845281373e-05, + "loss": 0.4385, + "step": 4123 + }, + { + "epoch": 1.73, + "learning_rate": 1.824278204869513e-05, + "loss": 0.4081, + "step": 4124 + }, + { + "epoch": 1.73, + "learning_rate": 1.8241822015177433e-05, + "loss": 0.4129, + "step": 4125 + }, + { + "epoch": 1.73, + "learning_rate": 1.824086174475588e-05, + "loss": 0.3608, + "step": 4126 + }, + { + "epoch": 1.73, + "learning_rate": 1.8239901237458064e-05, + "loss": 0.4721, + "step": 4127 + }, + { + "epoch": 1.73, + "learning_rate": 1.8238940493311604e-05, + "loss": 0.3642, + "step": 4128 + }, + { + "epoch": 1.73, + "learning_rate": 1.8237979512344107e-05, + "loss": 0.3647, + "step": 4129 + }, + { + "epoch": 1.73, + "learning_rate": 1.8237018294583196e-05, + "loss": 0.4653, + "step": 4130 + }, + { + "epoch": 1.73, + "learning_rate": 1.8236056840056512e-05, + "loss": 0.4066, + "step": 4131 + }, + { + "epoch": 1.73, + "learning_rate": 1.8235095148791675e-05, + "loss": 0.3798, + "step": 4132 + }, + { + "epoch": 1.73, + "learning_rate": 1.823413322081634e-05, + "loss": 0.388, + "step": 4133 + }, + { + "epoch": 1.73, + "learning_rate": 1.823317105615815e-05, + "loss": 0.4043, + "step": 4134 + }, + { + "epoch": 1.73, + "learning_rate": 1.8232208654844766e-05, + "loss": 0.3316, + "step": 4135 + }, + { + "epoch": 1.73, + "learning_rate": 1.823124601690385e-05, + "loss": 0.3562, + "step": 4136 + }, + { + "epoch": 1.73, + "learning_rate": 1.8230283142363068e-05, + "loss": 0.4149, + "step": 4137 + }, + { + "epoch": 1.73, + "learning_rate": 1.8229320031250098e-05, + "loss": 0.3378, + "step": 4138 + }, + { + "epoch": 1.73, + "learning_rate": 1.822835668359263e-05, + "loss": 0.4156, + "step": 4139 + }, + { + "epoch": 1.73, + "learning_rate": 1.8227393099418353e-05, + "loss": 0.3673, + "step": 4140 + }, + { + "epoch": 1.73, + "learning_rate": 1.8226429278754963e-05, + "loss": 0.4593, + "step": 4141 + }, + { + "epoch": 1.73, + "learning_rate": 1.822546522163016e-05, + "loss": 0.3879, + "step": 4142 + }, + { + "epoch": 1.73, + "learning_rate": 1.822450092807166e-05, + "loss": 0.3846, + "step": 4143 + }, + { + "epoch": 1.73, + "learning_rate": 1.8223536398107177e-05, + "loss": 0.3752, + "step": 4144 + }, + { + "epoch": 1.74, + "learning_rate": 1.822257163176444e-05, + "loss": 0.4094, + "step": 4145 + }, + { + "epoch": 1.74, + "learning_rate": 1.8221606629071175e-05, + "loss": 0.3852, + "step": 4146 + }, + { + "epoch": 1.74, + "learning_rate": 1.8220641390055127e-05, + "loss": 0.3741, + "step": 4147 + }, + { + "epoch": 1.74, + "learning_rate": 1.8219675914744035e-05, + "loss": 0.4307, + "step": 4148 + }, + { + "epoch": 1.74, + "learning_rate": 1.8218710203165653e-05, + "loss": 0.3892, + "step": 4149 + }, + { + "epoch": 1.74, + "learning_rate": 1.821774425534774e-05, + "loss": 0.3891, + "step": 4150 + }, + { + "epoch": 1.74, + "learning_rate": 1.821677807131806e-05, + "loss": 0.389, + "step": 4151 + }, + { + "epoch": 1.74, + "learning_rate": 1.8215811651104383e-05, + "loss": 0.3828, + "step": 4152 + }, + { + "epoch": 1.74, + "learning_rate": 1.8214844994734498e-05, + "loss": 0.3996, + "step": 4153 + }, + { + "epoch": 1.74, + "learning_rate": 1.8213878102236177e-05, + "loss": 0.4122, + "step": 4154 + }, + { + "epoch": 1.74, + "learning_rate": 1.821291097363722e-05, + "loss": 0.3949, + "step": 4155 + }, + { + "epoch": 1.74, + "learning_rate": 1.821194360896543e-05, + "loss": 0.3584, + "step": 4156 + }, + { + "epoch": 1.74, + "learning_rate": 1.8210976008248603e-05, + "loss": 0.3853, + "step": 4157 + }, + { + "epoch": 1.74, + "learning_rate": 1.821000817151456e-05, + "loss": 0.4165, + "step": 4158 + }, + { + "epoch": 1.74, + "learning_rate": 1.820904009879111e-05, + "loss": 0.4278, + "step": 4159 + }, + { + "epoch": 1.74, + "learning_rate": 1.8208071790106096e-05, + "loss": 0.3944, + "step": 4160 + }, + { + "epoch": 1.74, + "learning_rate": 1.8207103245487335e-05, + "loss": 0.3955, + "step": 4161 + }, + { + "epoch": 1.74, + "learning_rate": 1.8206134464962678e-05, + "loss": 0.3999, + "step": 4162 + }, + { + "epoch": 1.74, + "learning_rate": 1.820516544855997e-05, + "loss": 0.4308, + "step": 4163 + }, + { + "epoch": 1.74, + "learning_rate": 1.8204196196307058e-05, + "loss": 0.3431, + "step": 4164 + }, + { + "epoch": 1.74, + "learning_rate": 1.8203226708231806e-05, + "loss": 0.3637, + "step": 4165 + }, + { + "epoch": 1.74, + "learning_rate": 1.8202256984362087e-05, + "loss": 0.3876, + "step": 4166 + }, + { + "epoch": 1.74, + "learning_rate": 1.8201287024725765e-05, + "loss": 0.3851, + "step": 4167 + }, + { + "epoch": 1.75, + "learning_rate": 1.8200316829350728e-05, + "loss": 0.3599, + "step": 4168 + }, + { + "epoch": 1.75, + "learning_rate": 1.8199346398264856e-05, + "loss": 0.4277, + "step": 4169 + }, + { + "epoch": 1.75, + "learning_rate": 1.8198375731496054e-05, + "loss": 0.3442, + "step": 4170 + }, + { + "epoch": 1.75, + "learning_rate": 1.8197404829072214e-05, + "loss": 0.4514, + "step": 4171 + }, + { + "epoch": 1.75, + "learning_rate": 1.8196433691021243e-05, + "loss": 0.4127, + "step": 4172 + }, + { + "epoch": 1.75, + "learning_rate": 1.8195462317371063e-05, + "loss": 0.4018, + "step": 4173 + }, + { + "epoch": 1.75, + "learning_rate": 1.819449070814959e-05, + "loss": 0.4155, + "step": 4174 + }, + { + "epoch": 1.75, + "learning_rate": 1.8193518863384753e-05, + "loss": 0.3667, + "step": 4175 + }, + { + "epoch": 1.75, + "learning_rate": 1.819254678310449e-05, + "loss": 0.3539, + "step": 4176 + }, + { + "epoch": 1.75, + "learning_rate": 1.8191574467336737e-05, + "loss": 0.4054, + "step": 4177 + }, + { + "epoch": 1.75, + "learning_rate": 1.8190601916109444e-05, + "loss": 0.4086, + "step": 4178 + }, + { + "epoch": 1.75, + "learning_rate": 1.818962912945057e-05, + "loss": 0.3798, + "step": 4179 + }, + { + "epoch": 1.75, + "learning_rate": 1.818865610738807e-05, + "loss": 0.4052, + "step": 4180 + }, + { + "epoch": 1.75, + "learning_rate": 1.818768284994992e-05, + "loss": 0.3307, + "step": 4181 + }, + { + "epoch": 1.75, + "learning_rate": 1.818670935716409e-05, + "loss": 0.3998, + "step": 4182 + }, + { + "epoch": 1.75, + "learning_rate": 1.8185735629058566e-05, + "loss": 0.3693, + "step": 4183 + }, + { + "epoch": 1.75, + "learning_rate": 1.8184761665661335e-05, + "loss": 0.3871, + "step": 4184 + }, + { + "epoch": 1.75, + "learning_rate": 1.8183787467000392e-05, + "loss": 0.3721, + "step": 4185 + }, + { + "epoch": 1.75, + "learning_rate": 1.8182813033103742e-05, + "loss": 0.3355, + "step": 4186 + }, + { + "epoch": 1.75, + "learning_rate": 1.8181838363999392e-05, + "loss": 0.4374, + "step": 4187 + }, + { + "epoch": 1.75, + "learning_rate": 1.8180863459715358e-05, + "loss": 0.3902, + "step": 4188 + }, + { + "epoch": 1.75, + "learning_rate": 1.8179888320279663e-05, + "loss": 0.4099, + "step": 4189 + }, + { + "epoch": 1.75, + "learning_rate": 1.817891294572034e-05, + "loss": 0.3684, + "step": 4190 + }, + { + "epoch": 1.75, + "learning_rate": 1.817793733606542e-05, + "loss": 0.4386, + "step": 4191 + }, + { + "epoch": 1.76, + "learning_rate": 1.8176961491342947e-05, + "loss": 0.3864, + "step": 4192 + }, + { + "epoch": 1.76, + "learning_rate": 1.8175985411580975e-05, + "loss": 0.4155, + "step": 4193 + }, + { + "epoch": 1.76, + "learning_rate": 1.8175009096807557e-05, + "loss": 0.4204, + "step": 4194 + }, + { + "epoch": 1.76, + "learning_rate": 1.8174032547050754e-05, + "loss": 0.382, + "step": 4195 + }, + { + "epoch": 1.76, + "learning_rate": 1.8173055762338642e-05, + "loss": 0.4276, + "step": 4196 + }, + { + "epoch": 1.76, + "learning_rate": 1.8172078742699293e-05, + "loss": 0.3971, + "step": 4197 + }, + { + "epoch": 1.76, + "learning_rate": 1.817110148816079e-05, + "loss": 0.3628, + "step": 4198 + }, + { + "epoch": 1.76, + "learning_rate": 1.817012399875123e-05, + "loss": 0.4342, + "step": 4199 + }, + { + "epoch": 1.76, + "learning_rate": 1.8169146274498707e-05, + "loss": 0.3692, + "step": 4200 + }, + { + "epoch": 1.76, + "learning_rate": 1.8168168315431318e-05, + "loss": 0.3981, + "step": 4201 + }, + { + "epoch": 1.76, + "learning_rate": 1.8167190121577184e-05, + "loss": 0.4388, + "step": 4202 + }, + { + "epoch": 1.76, + "learning_rate": 1.8166211692964413e-05, + "loss": 0.3119, + "step": 4203 + }, + { + "epoch": 1.76, + "learning_rate": 1.8165233029621136e-05, + "loss": 0.3513, + "step": 4204 + }, + { + "epoch": 1.76, + "learning_rate": 1.816425413157548e-05, + "loss": 0.4071, + "step": 4205 + }, + { + "epoch": 1.76, + "learning_rate": 1.8163274998855585e-05, + "loss": 0.4342, + "step": 4206 + }, + { + "epoch": 1.76, + "learning_rate": 1.8162295631489595e-05, + "loss": 0.418, + "step": 4207 + }, + { + "epoch": 1.76, + "learning_rate": 1.8161316029505656e-05, + "loss": 0.3763, + "step": 4208 + }, + { + "epoch": 1.76, + "learning_rate": 1.8160336192931935e-05, + "loss": 0.4215, + "step": 4209 + }, + { + "epoch": 1.76, + "learning_rate": 1.8159356121796586e-05, + "loss": 0.3893, + "step": 4210 + }, + { + "epoch": 1.76, + "learning_rate": 1.8158375816127792e-05, + "loss": 0.4154, + "step": 4211 + }, + { + "epoch": 1.76, + "learning_rate": 1.8157395275953722e-05, + "loss": 0.375, + "step": 4212 + }, + { + "epoch": 1.76, + "learning_rate": 1.8156414501302565e-05, + "loss": 0.3681, + "step": 4213 + }, + { + "epoch": 1.76, + "learning_rate": 1.8155433492202507e-05, + "loss": 0.3854, + "step": 4214 + }, + { + "epoch": 1.76, + "learning_rate": 1.8154452248681754e-05, + "loss": 0.3947, + "step": 4215 + }, + { + "epoch": 1.77, + "learning_rate": 1.8153470770768506e-05, + "loss": 0.3914, + "step": 4216 + }, + { + "epoch": 1.77, + "learning_rate": 1.8152489058490975e-05, + "loss": 0.3556, + "step": 4217 + }, + { + "epoch": 1.77, + "learning_rate": 1.8151507111877382e-05, + "loss": 0.3644, + "step": 4218 + }, + { + "epoch": 1.77, + "learning_rate": 1.815052493095595e-05, + "loss": 0.3726, + "step": 4219 + }, + { + "epoch": 1.77, + "learning_rate": 1.814954251575491e-05, + "loss": 0.4142, + "step": 4220 + }, + { + "epoch": 1.77, + "learning_rate": 1.8148559866302505e-05, + "loss": 0.3856, + "step": 4221 + }, + { + "epoch": 1.77, + "learning_rate": 1.8147576982626974e-05, + "loss": 0.3581, + "step": 4222 + }, + { + "epoch": 1.77, + "learning_rate": 1.8146593864756577e-05, + "loss": 0.4268, + "step": 4223 + }, + { + "epoch": 1.77, + "learning_rate": 1.8145610512719565e-05, + "loss": 0.3792, + "step": 4224 + }, + { + "epoch": 1.77, + "learning_rate": 1.8144626926544208e-05, + "loss": 0.3814, + "step": 4225 + }, + { + "epoch": 1.77, + "learning_rate": 1.814364310625878e-05, + "loss": 0.3749, + "step": 4226 + }, + { + "epoch": 1.77, + "learning_rate": 1.8142659051891552e-05, + "loss": 0.3667, + "step": 4227 + }, + { + "epoch": 1.77, + "learning_rate": 1.814167476347082e-05, + "loss": 0.3915, + "step": 4228 + }, + { + "epoch": 1.77, + "learning_rate": 1.8140690241024872e-05, + "loss": 0.4039, + "step": 4229 + }, + { + "epoch": 1.77, + "learning_rate": 1.8139705484582005e-05, + "loss": 0.3712, + "step": 4230 + }, + { + "epoch": 1.77, + "learning_rate": 1.813872049417053e-05, + "loss": 0.3634, + "step": 4231 + }, + { + "epoch": 1.77, + "learning_rate": 1.813773526981875e-05, + "loss": 0.4128, + "step": 4232 + }, + { + "epoch": 1.77, + "learning_rate": 1.8136749811554995e-05, + "loss": 0.3902, + "step": 4233 + }, + { + "epoch": 1.77, + "learning_rate": 1.8135764119407585e-05, + "loss": 0.3534, + "step": 4234 + }, + { + "epoch": 1.77, + "learning_rate": 1.8134778193404857e-05, + "loss": 0.3815, + "step": 4235 + }, + { + "epoch": 1.77, + "learning_rate": 1.813379203357515e-05, + "loss": 0.3744, + "step": 4236 + }, + { + "epoch": 1.77, + "learning_rate": 1.8132805639946806e-05, + "loss": 0.3964, + "step": 4237 + }, + { + "epoch": 1.77, + "learning_rate": 1.8131819012548183e-05, + "loss": 0.3992, + "step": 4238 + }, + { + "epoch": 1.77, + "learning_rate": 1.8130832151407634e-05, + "loss": 0.3965, + "step": 4239 + }, + { + "epoch": 1.78, + "learning_rate": 1.8129845056553534e-05, + "loss": 0.4237, + "step": 4240 + }, + { + "epoch": 1.78, + "learning_rate": 1.812885772801425e-05, + "loss": 0.3852, + "step": 4241 + }, + { + "epoch": 1.78, + "learning_rate": 1.8127870165818166e-05, + "loss": 0.4264, + "step": 4242 + }, + { + "epoch": 1.78, + "learning_rate": 1.812688236999367e-05, + "loss": 0.4305, + "step": 4243 + }, + { + "epoch": 1.78, + "learning_rate": 1.8125894340569145e-05, + "loss": 0.3854, + "step": 4244 + }, + { + "epoch": 1.78, + "learning_rate": 1.8124906077573e-05, + "loss": 0.3823, + "step": 4245 + }, + { + "epoch": 1.78, + "learning_rate": 1.8123917581033642e-05, + "loss": 0.3893, + "step": 4246 + }, + { + "epoch": 1.78, + "learning_rate": 1.8122928850979483e-05, + "loss": 0.4558, + "step": 4247 + }, + { + "epoch": 1.78, + "learning_rate": 1.8121939887438936e-05, + "loss": 0.3923, + "step": 4248 + }, + { + "epoch": 1.78, + "learning_rate": 1.812095069044044e-05, + "loss": 0.37, + "step": 4249 + }, + { + "epoch": 1.78, + "learning_rate": 1.8119961260012422e-05, + "loss": 0.423, + "step": 4250 + }, + { + "epoch": 1.78, + "learning_rate": 1.811897159618332e-05, + "loss": 0.3789, + "step": 4251 + }, + { + "epoch": 1.78, + "learning_rate": 1.811798169898159e-05, + "loss": 0.3585, + "step": 4252 + }, + { + "epoch": 1.78, + "learning_rate": 1.8116991568435674e-05, + "loss": 0.3785, + "step": 4253 + }, + { + "epoch": 1.78, + "learning_rate": 1.8116001204574042e-05, + "loss": 0.4106, + "step": 4254 + }, + { + "epoch": 1.78, + "learning_rate": 1.8115010607425154e-05, + "loss": 0.4085, + "step": 4255 + }, + { + "epoch": 1.78, + "learning_rate": 1.811401977701749e-05, + "loss": 0.45, + "step": 4256 + }, + { + "epoch": 1.78, + "learning_rate": 1.8113028713379524e-05, + "loss": 0.4488, + "step": 4257 + }, + { + "epoch": 1.78, + "learning_rate": 1.811203741653975e-05, + "loss": 0.409, + "step": 4258 + }, + { + "epoch": 1.78, + "learning_rate": 1.811104588652666e-05, + "loss": 0.3769, + "step": 4259 + }, + { + "epoch": 1.78, + "learning_rate": 1.8110054123368755e-05, + "loss": 0.3857, + "step": 4260 + }, + { + "epoch": 1.78, + "learning_rate": 1.8109062127094538e-05, + "loss": 0.4303, + "step": 4261 + }, + { + "epoch": 1.78, + "learning_rate": 1.8108069897732527e-05, + "loss": 0.4154, + "step": 4262 + }, + { + "epoch": 1.78, + "learning_rate": 1.810707743531124e-05, + "loss": 0.3739, + "step": 4263 + }, + { + "epoch": 1.79, + "learning_rate": 1.8106084739859206e-05, + "loss": 0.3749, + "step": 4264 + }, + { + "epoch": 1.79, + "learning_rate": 1.8105091811404962e-05, + "loss": 0.4192, + "step": 4265 + }, + { + "epoch": 1.79, + "learning_rate": 1.8104098649977044e-05, + "loss": 0.4326, + "step": 4266 + }, + { + "epoch": 1.79, + "learning_rate": 1.8103105255604003e-05, + "loss": 0.4022, + "step": 4267 + }, + { + "epoch": 1.79, + "learning_rate": 1.8102111628314394e-05, + "loss": 0.4064, + "step": 4268 + }, + { + "epoch": 1.79, + "learning_rate": 1.810111776813677e-05, + "loss": 0.3597, + "step": 4269 + }, + { + "epoch": 1.79, + "learning_rate": 1.810012367509971e-05, + "loss": 0.397, + "step": 4270 + }, + { + "epoch": 1.79, + "learning_rate": 1.8099129349231783e-05, + "loss": 0.4079, + "step": 4271 + }, + { + "epoch": 1.79, + "learning_rate": 1.8098134790561567e-05, + "loss": 0.3914, + "step": 4272 + }, + { + "epoch": 1.79, + "learning_rate": 1.8097139999117655e-05, + "loss": 0.3927, + "step": 4273 + }, + { + "epoch": 1.79, + "learning_rate": 1.8096144974928635e-05, + "loss": 0.4111, + "step": 4274 + }, + { + "epoch": 1.79, + "learning_rate": 1.8095149718023114e-05, + "loss": 0.4026, + "step": 4275 + }, + { + "epoch": 1.79, + "learning_rate": 1.80941542284297e-05, + "loss": 0.3648, + "step": 4276 + }, + { + "epoch": 1.79, + "learning_rate": 1.8093158506177003e-05, + "loss": 0.3546, + "step": 4277 + }, + { + "epoch": 1.79, + "learning_rate": 1.8092162551293646e-05, + "loss": 0.3978, + "step": 4278 + }, + { + "epoch": 1.79, + "learning_rate": 1.8091166363808258e-05, + "loss": 0.3781, + "step": 4279 + }, + { + "epoch": 1.79, + "learning_rate": 1.8090169943749477e-05, + "loss": 0.3416, + "step": 4280 + }, + { + "epoch": 1.79, + "learning_rate": 1.8089173291145935e-05, + "loss": 0.3681, + "step": 4281 + }, + { + "epoch": 1.79, + "learning_rate": 1.808817640602629e-05, + "loss": 0.402, + "step": 4282 + }, + { + "epoch": 1.79, + "learning_rate": 1.8087179288419182e-05, + "loss": 0.4024, + "step": 4283 + }, + { + "epoch": 1.79, + "learning_rate": 1.808618193835329e-05, + "loss": 0.3581, + "step": 4284 + }, + { + "epoch": 1.79, + "learning_rate": 1.808518435585727e-05, + "loss": 0.4094, + "step": 4285 + }, + { + "epoch": 1.79, + "learning_rate": 1.80841865409598e-05, + "loss": 0.3766, + "step": 4286 + }, + { + "epoch": 1.79, + "learning_rate": 1.808318849368957e-05, + "loss": 0.4021, + "step": 4287 + }, + { + "epoch": 1.8, + "learning_rate": 1.8082190214075252e-05, + "loss": 0.3986, + "step": 4288 + }, + { + "epoch": 1.8, + "learning_rate": 1.8081191702145547e-05, + "loss": 0.397, + "step": 4289 + }, + { + "epoch": 1.8, + "learning_rate": 1.8080192957929162e-05, + "loss": 0.3872, + "step": 4290 + }, + { + "epoch": 1.8, + "learning_rate": 1.8079193981454798e-05, + "loss": 0.3809, + "step": 4291 + }, + { + "epoch": 1.8, + "learning_rate": 1.8078194772751173e-05, + "loss": 0.4337, + "step": 4292 + }, + { + "epoch": 1.8, + "learning_rate": 1.8077195331847007e-05, + "loss": 0.4027, + "step": 4293 + }, + { + "epoch": 1.8, + "learning_rate": 1.8076195658771033e-05, + "loss": 0.3748, + "step": 4294 + }, + { + "epoch": 1.8, + "learning_rate": 1.8075195753551974e-05, + "loss": 0.3974, + "step": 4295 + }, + { + "epoch": 1.8, + "learning_rate": 1.8074195616218583e-05, + "loss": 0.3615, + "step": 4296 + }, + { + "epoch": 1.8, + "learning_rate": 1.8073195246799605e-05, + "loss": 0.375, + "step": 4297 + }, + { + "epoch": 1.8, + "learning_rate": 1.8072194645323794e-05, + "loss": 0.449, + "step": 4298 + }, + { + "epoch": 1.8, + "learning_rate": 1.8071193811819912e-05, + "loss": 0.3788, + "step": 4299 + }, + { + "epoch": 1.8, + "learning_rate": 1.807019274631672e-05, + "loss": 0.3921, + "step": 4300 + }, + { + "epoch": 1.8, + "learning_rate": 1.8069191448843005e-05, + "loss": 0.3985, + "step": 4301 + }, + { + "epoch": 1.8, + "learning_rate": 1.806818991942754e-05, + "loss": 0.3922, + "step": 4302 + }, + { + "epoch": 1.8, + "learning_rate": 1.8067188158099112e-05, + "loss": 0.367, + "step": 4303 + }, + { + "epoch": 1.8, + "learning_rate": 1.8066186164886525e-05, + "loss": 0.4072, + "step": 4304 + }, + { + "epoch": 1.8, + "learning_rate": 1.806518393981857e-05, + "loss": 0.3879, + "step": 4305 + }, + { + "epoch": 1.8, + "learning_rate": 1.8064181482924065e-05, + "loss": 0.377, + "step": 4306 + }, + { + "epoch": 1.8, + "learning_rate": 1.8063178794231814e-05, + "loss": 0.4025, + "step": 4307 + }, + { + "epoch": 1.8, + "learning_rate": 1.8062175873770643e-05, + "loss": 0.3922, + "step": 4308 + }, + { + "epoch": 1.8, + "learning_rate": 1.8061172721569383e-05, + "loss": 0.4284, + "step": 4309 + }, + { + "epoch": 1.8, + "learning_rate": 1.8060169337656866e-05, + "loss": 0.4754, + "step": 4310 + }, + { + "epoch": 1.8, + "learning_rate": 1.805916572206194e-05, + "loss": 0.3792, + "step": 4311 + }, + { + "epoch": 1.81, + "learning_rate": 1.805816187481344e-05, + "loss": 0.374, + "step": 4312 + }, + { + "epoch": 1.81, + "learning_rate": 1.8057157795940228e-05, + "loss": 0.4356, + "step": 4313 + }, + { + "epoch": 1.81, + "learning_rate": 1.8056153485471167e-05, + "loss": 0.3891, + "step": 4314 + }, + { + "epoch": 1.81, + "learning_rate": 1.805514894343512e-05, + "loss": 0.3978, + "step": 4315 + }, + { + "epoch": 1.81, + "learning_rate": 1.805414416986097e-05, + "loss": 0.411, + "step": 4316 + }, + { + "epoch": 1.81, + "learning_rate": 1.8053139164777587e-05, + "loss": 0.3439, + "step": 4317 + }, + { + "epoch": 1.81, + "learning_rate": 1.805213392821387e-05, + "loss": 0.42, + "step": 4318 + }, + { + "epoch": 1.81, + "learning_rate": 1.8051128460198703e-05, + "loss": 0.3687, + "step": 4319 + }, + { + "epoch": 1.81, + "learning_rate": 1.8050122760760994e-05, + "loss": 0.3926, + "step": 4320 + }, + { + "epoch": 1.81, + "learning_rate": 1.8049116829929655e-05, + "loss": 0.3792, + "step": 4321 + }, + { + "epoch": 1.81, + "learning_rate": 1.804811066773359e-05, + "loss": 0.3436, + "step": 4322 + }, + { + "epoch": 1.81, + "learning_rate": 1.804710427420173e-05, + "loss": 0.386, + "step": 4323 + }, + { + "epoch": 1.81, + "learning_rate": 1.8046097649362994e-05, + "loss": 0.3816, + "step": 4324 + }, + { + "epoch": 1.81, + "learning_rate": 1.804509079324632e-05, + "loss": 0.3676, + "step": 4325 + }, + { + "epoch": 1.81, + "learning_rate": 1.804408370588066e-05, + "loss": 0.3695, + "step": 4326 + }, + { + "epoch": 1.81, + "learning_rate": 1.8043076387294944e-05, + "loss": 0.4537, + "step": 4327 + }, + { + "epoch": 1.81, + "learning_rate": 1.8042068837518135e-05, + "loss": 0.3724, + "step": 4328 + }, + { + "epoch": 1.81, + "learning_rate": 1.8041061056579194e-05, + "loss": 0.3988, + "step": 4329 + }, + { + "epoch": 1.81, + "learning_rate": 1.8040053044507087e-05, + "loss": 0.3737, + "step": 4330 + }, + { + "epoch": 1.81, + "learning_rate": 1.8039044801330795e-05, + "loss": 0.4418, + "step": 4331 + }, + { + "epoch": 1.81, + "learning_rate": 1.803803632707929e-05, + "loss": 0.3556, + "step": 4332 + }, + { + "epoch": 1.81, + "learning_rate": 1.8037027621781568e-05, + "loss": 0.3917, + "step": 4333 + }, + { + "epoch": 1.81, + "learning_rate": 1.8036018685466614e-05, + "loss": 0.4307, + "step": 4334 + }, + { + "epoch": 1.81, + "learning_rate": 1.8035009518163434e-05, + "loss": 0.4482, + "step": 4335 + }, + { + "epoch": 1.82, + "learning_rate": 1.8034000119901038e-05, + "loss": 0.4407, + "step": 4336 + }, + { + "epoch": 1.82, + "learning_rate": 1.8032990490708434e-05, + "loss": 0.3863, + "step": 4337 + }, + { + "epoch": 1.82, + "learning_rate": 1.8031980630614652e-05, + "loss": 0.3908, + "step": 4338 + }, + { + "epoch": 1.82, + "learning_rate": 1.803097053964871e-05, + "loss": 0.3528, + "step": 4339 + }, + { + "epoch": 1.82, + "learning_rate": 1.8029960217839648e-05, + "loss": 0.3768, + "step": 4340 + }, + { + "epoch": 1.82, + "learning_rate": 1.8028949665216505e-05, + "loss": 0.4015, + "step": 4341 + }, + { + "epoch": 1.82, + "learning_rate": 1.802793888180833e-05, + "loss": 0.4411, + "step": 4342 + }, + { + "epoch": 1.82, + "learning_rate": 1.8026927867644175e-05, + "loss": 0.4535, + "step": 4343 + }, + { + "epoch": 1.82, + "learning_rate": 1.80259166227531e-05, + "loss": 0.3384, + "step": 4344 + }, + { + "epoch": 1.82, + "learning_rate": 1.8024905147164176e-05, + "loss": 0.3852, + "step": 4345 + }, + { + "epoch": 1.82, + "learning_rate": 1.8023893440906476e-05, + "loss": 0.3559, + "step": 4346 + }, + { + "epoch": 1.82, + "learning_rate": 1.8022881504009074e-05, + "loss": 0.4396, + "step": 4347 + }, + { + "epoch": 1.82, + "learning_rate": 1.8021869336501066e-05, + "loss": 0.3677, + "step": 4348 + }, + { + "epoch": 1.82, + "learning_rate": 1.802085693841154e-05, + "loss": 0.3632, + "step": 4349 + }, + { + "epoch": 1.82, + "learning_rate": 1.8019844309769602e-05, + "loss": 0.4229, + "step": 4350 + }, + { + "epoch": 1.82, + "learning_rate": 1.8018831450604356e-05, + "loss": 0.4812, + "step": 4351 + }, + { + "epoch": 1.82, + "learning_rate": 1.801781836094491e-05, + "loss": 0.387, + "step": 4352 + }, + { + "epoch": 1.82, + "learning_rate": 1.80168050408204e-05, + "loss": 0.3763, + "step": 4353 + }, + { + "epoch": 1.82, + "learning_rate": 1.8015791490259932e-05, + "loss": 0.3933, + "step": 4354 + }, + { + "epoch": 1.82, + "learning_rate": 1.801477770929266e-05, + "loss": 0.4103, + "step": 4355 + }, + { + "epoch": 1.82, + "learning_rate": 1.801376369794771e-05, + "loss": 0.4138, + "step": 4356 + }, + { + "epoch": 1.82, + "learning_rate": 1.8012749456254238e-05, + "loss": 0.4028, + "step": 4357 + }, + { + "epoch": 1.82, + "learning_rate": 1.801173498424139e-05, + "loss": 0.3912, + "step": 4358 + }, + { + "epoch": 1.82, + "learning_rate": 1.8010720281938327e-05, + "loss": 0.429, + "step": 4359 + }, + { + "epoch": 1.83, + "learning_rate": 1.800970534937422e-05, + "loss": 0.3638, + "step": 4360 + }, + { + "epoch": 1.83, + "learning_rate": 1.8008690186578243e-05, + "loss": 0.4019, + "step": 4361 + }, + { + "epoch": 1.83, + "learning_rate": 1.800767479357957e-05, + "loss": 0.3711, + "step": 4362 + }, + { + "epoch": 1.83, + "learning_rate": 1.8006659170407396e-05, + "loss": 0.3859, + "step": 4363 + }, + { + "epoch": 1.83, + "learning_rate": 1.80056433170909e-05, + "loss": 0.441, + "step": 4364 + }, + { + "epoch": 1.83, + "learning_rate": 1.80046272336593e-05, + "loss": 0.375, + "step": 4365 + }, + { + "epoch": 1.83, + "learning_rate": 1.800361092014179e-05, + "loss": 0.4191, + "step": 4366 + }, + { + "epoch": 1.83, + "learning_rate": 1.8002594376567585e-05, + "loss": 0.3702, + "step": 4367 + }, + { + "epoch": 1.83, + "learning_rate": 1.8001577602965907e-05, + "loss": 0.4165, + "step": 4368 + }, + { + "epoch": 1.83, + "learning_rate": 1.800056059936598e-05, + "loss": 0.4246, + "step": 4369 + }, + { + "epoch": 1.83, + "learning_rate": 1.7999543365797043e-05, + "loss": 0.4047, + "step": 4370 + }, + { + "epoch": 1.83, + "learning_rate": 1.7998525902288327e-05, + "loss": 0.3736, + "step": 4371 + }, + { + "epoch": 1.83, + "learning_rate": 1.7997508208869082e-05, + "loss": 0.4108, + "step": 4372 + }, + { + "epoch": 1.83, + "learning_rate": 1.799649028556856e-05, + "loss": 0.3926, + "step": 4373 + }, + { + "epoch": 1.83, + "learning_rate": 1.799547213241602e-05, + "loss": 0.403, + "step": 4374 + }, + { + "epoch": 1.83, + "learning_rate": 1.799445374944073e-05, + "loss": 0.3477, + "step": 4375 + }, + { + "epoch": 1.83, + "learning_rate": 1.799343513667196e-05, + "loss": 0.4046, + "step": 4376 + }, + { + "epoch": 1.83, + "learning_rate": 1.7992416294138992e-05, + "loss": 0.3881, + "step": 4377 + }, + { + "epoch": 1.83, + "learning_rate": 1.799139722187111e-05, + "loss": 0.4091, + "step": 4378 + }, + { + "epoch": 1.83, + "learning_rate": 1.7990377919897605e-05, + "loss": 0.399, + "step": 4379 + }, + { + "epoch": 1.83, + "learning_rate": 1.7989358388247776e-05, + "loss": 0.4123, + "step": 4380 + }, + { + "epoch": 1.83, + "learning_rate": 1.798833862695093e-05, + "loss": 0.4001, + "step": 4381 + }, + { + "epoch": 1.83, + "learning_rate": 1.7987318636036382e-05, + "loss": 0.3674, + "step": 4382 + }, + { + "epoch": 1.84, + "learning_rate": 1.7986298415533444e-05, + "loss": 0.3603, + "step": 4383 + }, + { + "epoch": 1.84, + "learning_rate": 1.798527796547145e-05, + "loss": 0.4306, + "step": 4384 + }, + { + "epoch": 1.84, + "learning_rate": 1.7984257285879724e-05, + "loss": 0.3837, + "step": 4385 + }, + { + "epoch": 1.84, + "learning_rate": 1.7983236376787607e-05, + "loss": 0.4104, + "step": 4386 + }, + { + "epoch": 1.84, + "learning_rate": 1.7982215238224445e-05, + "loss": 0.368, + "step": 4387 + }, + { + "epoch": 1.84, + "learning_rate": 1.7981193870219587e-05, + "loss": 0.3786, + "step": 4388 + }, + { + "epoch": 1.84, + "learning_rate": 1.7980172272802398e-05, + "loss": 0.4624, + "step": 4389 + }, + { + "epoch": 1.84, + "learning_rate": 1.7979150446002233e-05, + "loss": 0.4058, + "step": 4390 + }, + { + "epoch": 1.84, + "learning_rate": 1.7978128389848476e-05, + "loss": 0.3636, + "step": 4391 + }, + { + "epoch": 1.84, + "learning_rate": 1.7977106104370493e-05, + "loss": 0.3949, + "step": 4392 + }, + { + "epoch": 1.84, + "learning_rate": 1.7976083589597672e-05, + "loss": 0.3636, + "step": 4393 + }, + { + "epoch": 1.84, + "learning_rate": 1.7975060845559412e-05, + "loss": 0.3717, + "step": 4394 + }, + { + "epoch": 1.84, + "learning_rate": 1.7974037872285102e-05, + "loss": 0.4207, + "step": 4395 + }, + { + "epoch": 1.84, + "learning_rate": 1.797301466980415e-05, + "loss": 0.3758, + "step": 4396 + }, + { + "epoch": 1.84, + "learning_rate": 1.7971991238145965e-05, + "loss": 0.3979, + "step": 4397 + }, + { + "epoch": 1.84, + "learning_rate": 1.7970967577339965e-05, + "loss": 0.4325, + "step": 4398 + }, + { + "epoch": 1.84, + "learning_rate": 1.7969943687415575e-05, + "loss": 0.4123, + "step": 4399 + }, + { + "epoch": 1.84, + "learning_rate": 1.7968919568402234e-05, + "loss": 0.3432, + "step": 4400 + }, + { + "epoch": 1.84, + "learning_rate": 1.7967895220329364e-05, + "loss": 0.3518, + "step": 4401 + }, + { + "epoch": 1.84, + "learning_rate": 1.796687064322642e-05, + "loss": 0.4224, + "step": 4402 + }, + { + "epoch": 1.84, + "learning_rate": 1.7965845837122844e-05, + "loss": 0.3905, + "step": 4403 + }, + { + "epoch": 1.84, + "learning_rate": 1.79648208020481e-05, + "loss": 0.3709, + "step": 4404 + }, + { + "epoch": 1.84, + "learning_rate": 1.796379553803165e-05, + "loss": 0.4077, + "step": 4405 + }, + { + "epoch": 1.84, + "learning_rate": 1.7962770045102968e-05, + "loss": 0.3806, + "step": 4406 + }, + { + "epoch": 1.85, + "learning_rate": 1.7961744323291525e-05, + "loss": 0.3653, + "step": 4407 + }, + { + "epoch": 1.85, + "learning_rate": 1.7960718372626807e-05, + "loss": 0.435, + "step": 4408 + }, + { + "epoch": 1.85, + "learning_rate": 1.7959692193138303e-05, + "loss": 0.3866, + "step": 4409 + }, + { + "epoch": 1.85, + "learning_rate": 1.795866578485551e-05, + "loss": 0.4393, + "step": 4410 + }, + { + "epoch": 1.85, + "learning_rate": 1.795763914780793e-05, + "loss": 0.4025, + "step": 4411 + }, + { + "epoch": 1.85, + "learning_rate": 1.7956612282025078e-05, + "loss": 0.3961, + "step": 4412 + }, + { + "epoch": 1.85, + "learning_rate": 1.7955585187536463e-05, + "loss": 0.4066, + "step": 4413 + }, + { + "epoch": 1.85, + "learning_rate": 1.7954557864371615e-05, + "loss": 0.3567, + "step": 4414 + }, + { + "epoch": 1.85, + "learning_rate": 1.7953530312560056e-05, + "loss": 0.4295, + "step": 4415 + }, + { + "epoch": 1.85, + "learning_rate": 1.7952502532131326e-05, + "loss": 0.3641, + "step": 4416 + }, + { + "epoch": 1.85, + "learning_rate": 1.795147452311497e-05, + "loss": 0.4159, + "step": 4417 + }, + { + "epoch": 1.85, + "learning_rate": 1.7950446285540532e-05, + "loss": 0.3475, + "step": 4418 + }, + { + "epoch": 1.85, + "learning_rate": 1.7949417819437577e-05, + "loss": 0.3599, + "step": 4419 + }, + { + "epoch": 1.85, + "learning_rate": 1.7948389124835654e-05, + "loss": 0.3755, + "step": 4420 + }, + { + "epoch": 1.85, + "learning_rate": 1.794736020176434e-05, + "loss": 0.4222, + "step": 4421 + }, + { + "epoch": 1.85, + "learning_rate": 1.794633105025321e-05, + "loss": 0.3815, + "step": 4422 + }, + { + "epoch": 1.85, + "learning_rate": 1.7945301670331847e-05, + "loss": 0.4236, + "step": 4423 + }, + { + "epoch": 1.85, + "learning_rate": 1.794427206202984e-05, + "loss": 0.3942, + "step": 4424 + }, + { + "epoch": 1.85, + "learning_rate": 1.7943242225376778e-05, + "loss": 0.3716, + "step": 4425 + }, + { + "epoch": 1.85, + "learning_rate": 1.794221216040227e-05, + "loss": 0.3625, + "step": 4426 + }, + { + "epoch": 1.85, + "learning_rate": 1.7941181867135917e-05, + "loss": 0.4135, + "step": 4427 + }, + { + "epoch": 1.85, + "learning_rate": 1.7940151345607343e-05, + "loss": 0.3409, + "step": 4428 + }, + { + "epoch": 1.85, + "learning_rate": 1.793912059584616e-05, + "loss": 0.3957, + "step": 4429 + }, + { + "epoch": 1.85, + "learning_rate": 1.7938089617882004e-05, + "loss": 0.3973, + "step": 4430 + }, + { + "epoch": 1.86, + "learning_rate": 1.7937058411744504e-05, + "loss": 0.4159, + "step": 4431 + }, + { + "epoch": 1.86, + "learning_rate": 1.7936026977463306e-05, + "loss": 0.3604, + "step": 4432 + }, + { + "epoch": 1.86, + "learning_rate": 1.793499531506805e-05, + "loss": 0.3864, + "step": 4433 + }, + { + "epoch": 1.86, + "learning_rate": 1.79339634245884e-05, + "loss": 0.4238, + "step": 4434 + }, + { + "epoch": 1.86, + "learning_rate": 1.7932931306054007e-05, + "loss": 0.3839, + "step": 4435 + }, + { + "epoch": 1.86, + "learning_rate": 1.7931898959494546e-05, + "loss": 0.3724, + "step": 4436 + }, + { + "epoch": 1.86, + "learning_rate": 1.793086638493969e-05, + "loss": 0.383, + "step": 4437 + }, + { + "epoch": 1.86, + "learning_rate": 1.7929833582419115e-05, + "loss": 0.4443, + "step": 4438 + }, + { + "epoch": 1.86, + "learning_rate": 1.792880055196251e-05, + "loss": 0.4123, + "step": 4439 + }, + { + "epoch": 1.86, + "learning_rate": 1.792776729359957e-05, + "loss": 0.3807, + "step": 4440 + }, + { + "epoch": 1.86, + "learning_rate": 1.7926733807359993e-05, + "loss": 0.4245, + "step": 4441 + }, + { + "epoch": 1.86, + "learning_rate": 1.792570009327349e-05, + "loss": 0.3401, + "step": 4442 + }, + { + "epoch": 1.86, + "learning_rate": 1.792466615136977e-05, + "loss": 0.4053, + "step": 4443 + }, + { + "epoch": 1.86, + "learning_rate": 1.7923631981678552e-05, + "loss": 0.4292, + "step": 4444 + }, + { + "epoch": 1.86, + "learning_rate": 1.7922597584229566e-05, + "loss": 0.3729, + "step": 4445 + }, + { + "epoch": 1.86, + "learning_rate": 1.792156295905254e-05, + "loss": 0.4087, + "step": 4446 + }, + { + "epoch": 1.86, + "learning_rate": 1.792052810617722e-05, + "loss": 0.3385, + "step": 4447 + }, + { + "epoch": 1.86, + "learning_rate": 1.7919493025633342e-05, + "loss": 0.4045, + "step": 4448 + }, + { + "epoch": 1.86, + "learning_rate": 1.791845771745067e-05, + "loss": 0.3663, + "step": 4449 + }, + { + "epoch": 1.86, + "learning_rate": 1.7917422181658954e-05, + "loss": 0.4042, + "step": 4450 + }, + { + "epoch": 1.86, + "learning_rate": 1.7916386418287968e-05, + "loss": 0.3809, + "step": 4451 + }, + { + "epoch": 1.86, + "learning_rate": 1.7915350427367476e-05, + "loss": 0.4189, + "step": 4452 + }, + { + "epoch": 1.86, + "learning_rate": 1.791431420892726e-05, + "loss": 0.3534, + "step": 4453 + }, + { + "epoch": 1.86, + "learning_rate": 1.7913277762997105e-05, + "loss": 0.3893, + "step": 4454 + }, + { + "epoch": 1.87, + "learning_rate": 1.7912241089606802e-05, + "loss": 0.3716, + "step": 4455 + }, + { + "epoch": 1.87, + "learning_rate": 1.7911204188786154e-05, + "loss": 0.4082, + "step": 4456 + }, + { + "epoch": 1.87, + "learning_rate": 1.791016706056496e-05, + "loss": 0.3839, + "step": 4457 + }, + { + "epoch": 1.87, + "learning_rate": 1.790912970497303e-05, + "loss": 0.4122, + "step": 4458 + }, + { + "epoch": 1.87, + "learning_rate": 1.790809212204019e-05, + "loss": 0.3856, + "step": 4459 + }, + { + "epoch": 1.87, + "learning_rate": 1.7907054311796254e-05, + "loss": 0.4127, + "step": 4460 + }, + { + "epoch": 1.87, + "learning_rate": 1.7906016274271062e-05, + "loss": 0.3855, + "step": 4461 + }, + { + "epoch": 1.87, + "learning_rate": 1.790497800949445e-05, + "loss": 0.4138, + "step": 4462 + }, + { + "epoch": 1.87, + "learning_rate": 1.7903939517496253e-05, + "loss": 0.4138, + "step": 4463 + }, + { + "epoch": 1.87, + "learning_rate": 1.7902900798306334e-05, + "loss": 0.3514, + "step": 4464 + }, + { + "epoch": 1.87, + "learning_rate": 1.7901861851954543e-05, + "loss": 0.3967, + "step": 4465 + }, + { + "epoch": 1.87, + "learning_rate": 1.7900822678470745e-05, + "loss": 0.4038, + "step": 4466 + }, + { + "epoch": 1.87, + "learning_rate": 1.789978327788481e-05, + "loss": 0.3741, + "step": 4467 + }, + { + "epoch": 1.87, + "learning_rate": 1.7898743650226617e-05, + "loss": 0.4083, + "step": 4468 + }, + { + "epoch": 1.87, + "learning_rate": 1.7897703795526044e-05, + "loss": 0.3672, + "step": 4469 + }, + { + "epoch": 1.87, + "learning_rate": 1.7896663713812984e-05, + "loss": 0.4927, + "step": 4470 + }, + { + "epoch": 1.87, + "learning_rate": 1.7895623405117334e-05, + "loss": 0.4271, + "step": 4471 + }, + { + "epoch": 1.87, + "learning_rate": 1.7894582869468994e-05, + "loss": 0.3796, + "step": 4472 + }, + { + "epoch": 1.87, + "learning_rate": 1.7893542106897878e-05, + "loss": 0.4172, + "step": 4473 + }, + { + "epoch": 1.87, + "learning_rate": 1.7892501117433896e-05, + "loss": 0.3696, + "step": 4474 + }, + { + "epoch": 1.87, + "learning_rate": 1.7891459901106976e-05, + "loss": 0.3844, + "step": 4475 + }, + { + "epoch": 1.87, + "learning_rate": 1.7890418457947038e-05, + "loss": 0.3763, + "step": 4476 + }, + { + "epoch": 1.87, + "learning_rate": 1.788937678798403e-05, + "loss": 0.3572, + "step": 4477 + }, + { + "epoch": 1.87, + "learning_rate": 1.7888334891247883e-05, + "loss": 0.3744, + "step": 4478 + }, + { + "epoch": 1.88, + "learning_rate": 1.788729276776855e-05, + "loss": 0.3865, + "step": 4479 + }, + { + "epoch": 1.88, + "learning_rate": 1.7886250417575986e-05, + "loss": 0.4056, + "step": 4480 + }, + { + "epoch": 1.88, + "learning_rate": 1.7885207840700155e-05, + "loss": 0.4122, + "step": 4481 + }, + { + "epoch": 1.88, + "learning_rate": 1.7884165037171016e-05, + "loss": 0.3764, + "step": 4482 + }, + { + "epoch": 1.88, + "learning_rate": 1.7883122007018554e-05, + "loss": 0.3989, + "step": 4483 + }, + { + "epoch": 1.88, + "learning_rate": 1.788207875027274e-05, + "loss": 0.4024, + "step": 4484 + }, + { + "epoch": 1.88, + "learning_rate": 1.7881035266963575e-05, + "loss": 0.3932, + "step": 4485 + }, + { + "epoch": 1.88, + "learning_rate": 1.787999155712104e-05, + "loss": 0.3933, + "step": 4486 + }, + { + "epoch": 1.88, + "learning_rate": 1.787894762077514e-05, + "loss": 0.3786, + "step": 4487 + }, + { + "epoch": 1.88, + "learning_rate": 1.787790345795588e-05, + "loss": 0.4072, + "step": 4488 + }, + { + "epoch": 1.88, + "learning_rate": 1.787685906869328e-05, + "loss": 0.4105, + "step": 4489 + }, + { + "epoch": 1.88, + "learning_rate": 1.7875814453017355e-05, + "loss": 0.3618, + "step": 4490 + }, + { + "epoch": 1.88, + "learning_rate": 1.787476961095813e-05, + "loss": 0.4214, + "step": 4491 + }, + { + "epoch": 1.88, + "learning_rate": 1.7873724542545647e-05, + "loss": 0.3606, + "step": 4492 + }, + { + "epoch": 1.88, + "learning_rate": 1.7872679247809933e-05, + "loss": 0.397, + "step": 4493 + }, + { + "epoch": 1.88, + "learning_rate": 1.7871633726781042e-05, + "loss": 0.4022, + "step": 4494 + }, + { + "epoch": 1.88, + "learning_rate": 1.787058797948902e-05, + "loss": 0.3629, + "step": 4495 + }, + { + "epoch": 1.88, + "learning_rate": 1.786954200596394e-05, + "loss": 0.427, + "step": 4496 + }, + { + "epoch": 1.88, + "learning_rate": 1.786849580623585e-05, + "loss": 0.3878, + "step": 4497 + }, + { + "epoch": 1.88, + "learning_rate": 1.7867449380334834e-05, + "loss": 0.4738, + "step": 4498 + }, + { + "epoch": 1.88, + "learning_rate": 1.7866402728290967e-05, + "loss": 0.3804, + "step": 4499 + }, + { + "epoch": 1.88, + "learning_rate": 1.786535585013433e-05, + "loss": 0.4011, + "step": 4500 + }, + { + "epoch": 1.88, + "learning_rate": 1.7864308745895024e-05, + "loss": 0.4079, + "step": 4501 + }, + { + "epoch": 1.88, + "learning_rate": 1.7863261415603142e-05, + "loss": 0.3485, + "step": 4502 + }, + { + "epoch": 1.89, + "learning_rate": 1.7862213859288788e-05, + "loss": 0.4219, + "step": 4503 + }, + { + "epoch": 1.89, + "learning_rate": 1.786116607698207e-05, + "loss": 0.3978, + "step": 4504 + }, + { + "epoch": 1.89, + "learning_rate": 1.786011806871311e-05, + "loss": 0.3497, + "step": 4505 + }, + { + "epoch": 1.89, + "learning_rate": 1.7859069834512033e-05, + "loss": 0.354, + "step": 4506 + }, + { + "epoch": 1.89, + "learning_rate": 1.7858021374408968e-05, + "loss": 0.3959, + "step": 4507 + }, + { + "epoch": 1.89, + "learning_rate": 1.785697268843405e-05, + "loss": 0.375, + "step": 4508 + }, + { + "epoch": 1.89, + "learning_rate": 1.7855923776617424e-05, + "loss": 0.3681, + "step": 4509 + }, + { + "epoch": 1.89, + "learning_rate": 1.785487463898924e-05, + "loss": 0.4029, + "step": 4510 + }, + { + "epoch": 1.89, + "learning_rate": 1.7853825275579656e-05, + "loss": 0.4052, + "step": 4511 + }, + { + "epoch": 1.89, + "learning_rate": 1.7852775686418832e-05, + "loss": 0.4151, + "step": 4512 + }, + { + "epoch": 1.89, + "learning_rate": 1.7851725871536945e-05, + "loss": 0.4123, + "step": 4513 + }, + { + "epoch": 1.89, + "learning_rate": 1.785067583096416e-05, + "loss": 0.3862, + "step": 4514 + }, + { + "epoch": 1.89, + "learning_rate": 1.7849625564730667e-05, + "loss": 0.3823, + "step": 4515 + }, + { + "epoch": 1.89, + "learning_rate": 1.784857507286665e-05, + "loss": 0.3856, + "step": 4516 + }, + { + "epoch": 1.89, + "learning_rate": 1.784752435540231e-05, + "loss": 0.3882, + "step": 4517 + }, + { + "epoch": 1.89, + "learning_rate": 1.7846473412367845e-05, + "loss": 0.4147, + "step": 4518 + }, + { + "epoch": 1.89, + "learning_rate": 1.7845422243793464e-05, + "loss": 0.4482, + "step": 4519 + }, + { + "epoch": 1.89, + "learning_rate": 1.784437084970938e-05, + "loss": 0.3736, + "step": 4520 + }, + { + "epoch": 1.89, + "learning_rate": 1.7843319230145822e-05, + "loss": 0.3869, + "step": 4521 + }, + { + "epoch": 1.89, + "learning_rate": 1.7842267385133013e-05, + "loss": 0.4484, + "step": 4522 + }, + { + "epoch": 1.89, + "learning_rate": 1.7841215314701182e-05, + "loss": 0.4245, + "step": 4523 + }, + { + "epoch": 1.89, + "learning_rate": 1.784016301888058e-05, + "loss": 0.3619, + "step": 4524 + }, + { + "epoch": 1.89, + "learning_rate": 1.7839110497701446e-05, + "loss": 0.4076, + "step": 4525 + }, + { + "epoch": 1.89, + "learning_rate": 1.783805775119404e-05, + "loss": 0.4029, + "step": 4526 + }, + { + "epoch": 1.9, + "learning_rate": 1.7837004779388615e-05, + "loss": 0.4297, + "step": 4527 + }, + { + "epoch": 1.9, + "learning_rate": 1.7835951582315445e-05, + "loss": 0.3741, + "step": 4528 + }, + { + "epoch": 1.9, + "learning_rate": 1.78348981600048e-05, + "loss": 0.4124, + "step": 4529 + }, + { + "epoch": 1.9, + "learning_rate": 1.783384451248696e-05, + "loss": 0.3873, + "step": 4530 + }, + { + "epoch": 1.9, + "learning_rate": 1.783279063979221e-05, + "loss": 0.3949, + "step": 4531 + }, + { + "epoch": 1.9, + "learning_rate": 1.7831736541950845e-05, + "loss": 0.399, + "step": 4532 + }, + { + "epoch": 1.9, + "learning_rate": 1.783068221899316e-05, + "loss": 0.4076, + "step": 4533 + }, + { + "epoch": 1.9, + "learning_rate": 1.7829627670949468e-05, + "loss": 0.4031, + "step": 4534 + }, + { + "epoch": 1.9, + "learning_rate": 1.7828572897850076e-05, + "loss": 0.3656, + "step": 4535 + }, + { + "epoch": 1.9, + "learning_rate": 1.78275178997253e-05, + "loss": 0.3846, + "step": 4536 + }, + { + "epoch": 1.9, + "learning_rate": 1.782646267660547e-05, + "loss": 0.409, + "step": 4537 + }, + { + "epoch": 1.9, + "learning_rate": 1.7825407228520917e-05, + "loss": 0.3568, + "step": 4538 + }, + { + "epoch": 1.9, + "learning_rate": 1.7824351555501972e-05, + "loss": 0.4116, + "step": 4539 + }, + { + "epoch": 1.9, + "learning_rate": 1.782329565757899e-05, + "loss": 0.4143, + "step": 4540 + }, + { + "epoch": 1.9, + "learning_rate": 1.7822239534782312e-05, + "loss": 0.3885, + "step": 4541 + }, + { + "epoch": 1.9, + "learning_rate": 1.7821183187142302e-05, + "loss": 0.4098, + "step": 4542 + }, + { + "epoch": 1.9, + "learning_rate": 1.7820126614689326e-05, + "loss": 0.4534, + "step": 4543 + }, + { + "epoch": 1.9, + "learning_rate": 1.7819069817453746e-05, + "loss": 0.3467, + "step": 4544 + }, + { + "epoch": 1.9, + "learning_rate": 1.781801279546594e-05, + "loss": 0.408, + "step": 4545 + }, + { + "epoch": 1.9, + "learning_rate": 1.7816955548756298e-05, + "loss": 0.3673, + "step": 4546 + }, + { + "epoch": 1.9, + "learning_rate": 1.7815898077355204e-05, + "loss": 0.4112, + "step": 4547 + }, + { + "epoch": 1.9, + "learning_rate": 1.7814840381293054e-05, + "loss": 0.3596, + "step": 4548 + }, + { + "epoch": 1.9, + "learning_rate": 1.7813782460600257e-05, + "loss": 0.3993, + "step": 4549 + }, + { + "epoch": 1.9, + "learning_rate": 1.781272431530721e-05, + "loss": 0.3849, + "step": 4550 + }, + { + "epoch": 1.91, + "learning_rate": 1.7811665945444337e-05, + "loss": 0.3657, + "step": 4551 + }, + { + "epoch": 1.91, + "learning_rate": 1.7810607351042062e-05, + "loss": 0.3464, + "step": 4552 + }, + { + "epoch": 1.91, + "learning_rate": 1.7809548532130808e-05, + "loss": 0.3813, + "step": 4553 + }, + { + "epoch": 1.91, + "learning_rate": 1.7808489488741013e-05, + "loss": 0.4107, + "step": 4554 + }, + { + "epoch": 1.91, + "learning_rate": 1.7807430220903113e-05, + "loss": 0.319, + "step": 4555 + }, + { + "epoch": 1.91, + "learning_rate": 1.780637072864756e-05, + "loss": 0.3817, + "step": 4556 + }, + { + "epoch": 1.91, + "learning_rate": 1.7805311012004812e-05, + "loss": 0.36, + "step": 4557 + }, + { + "epoch": 1.91, + "learning_rate": 1.7804251071005323e-05, + "loss": 0.4005, + "step": 4558 + }, + { + "epoch": 1.91, + "learning_rate": 1.7803190905679562e-05, + "loss": 0.3722, + "step": 4559 + }, + { + "epoch": 1.91, + "learning_rate": 1.7802130516058003e-05, + "loss": 0.4129, + "step": 4560 + }, + { + "epoch": 1.91, + "learning_rate": 1.7801069902171123e-05, + "loss": 0.3759, + "step": 4561 + }, + { + "epoch": 1.91, + "learning_rate": 1.7800009064049416e-05, + "loss": 0.414, + "step": 4562 + }, + { + "epoch": 1.91, + "learning_rate": 1.7798948001723368e-05, + "loss": 0.4868, + "step": 4563 + }, + { + "epoch": 1.91, + "learning_rate": 1.779788671522348e-05, + "loss": 0.3889, + "step": 4564 + }, + { + "epoch": 1.91, + "learning_rate": 1.779682520458026e-05, + "loss": 0.3491, + "step": 4565 + }, + { + "epoch": 1.91, + "learning_rate": 1.779576346982421e-05, + "loss": 0.4432, + "step": 4566 + }, + { + "epoch": 1.91, + "learning_rate": 1.7794701510985867e-05, + "loss": 0.3424, + "step": 4567 + }, + { + "epoch": 1.91, + "learning_rate": 1.779363932809574e-05, + "loss": 0.3825, + "step": 4568 + }, + { + "epoch": 1.91, + "learning_rate": 1.7792576921184374e-05, + "loss": 0.3551, + "step": 4569 + }, + { + "epoch": 1.91, + "learning_rate": 1.7791514290282293e-05, + "loss": 0.3705, + "step": 4570 + }, + { + "epoch": 1.91, + "learning_rate": 1.7790451435420046e-05, + "loss": 0.3369, + "step": 4571 + }, + { + "epoch": 1.91, + "learning_rate": 1.778938835662819e-05, + "loss": 0.402, + "step": 4572 + }, + { + "epoch": 1.91, + "learning_rate": 1.7788325053937278e-05, + "loss": 0.3878, + "step": 4573 + }, + { + "epoch": 1.92, + "learning_rate": 1.778726152737787e-05, + "loss": 0.4031, + "step": 4574 + }, + { + "epoch": 1.92, + "learning_rate": 1.7786197776980542e-05, + "loss": 0.384, + "step": 4575 + }, + { + "epoch": 1.92, + "learning_rate": 1.778513380277587e-05, + "loss": 0.3849, + "step": 4576 + }, + { + "epoch": 1.92, + "learning_rate": 1.7784069604794433e-05, + "loss": 0.3563, + "step": 4577 + }, + { + "epoch": 1.92, + "learning_rate": 1.778300518306682e-05, + "loss": 0.4222, + "step": 4578 + }, + { + "epoch": 1.92, + "learning_rate": 1.7781940537623636e-05, + "loss": 0.3769, + "step": 4579 + }, + { + "epoch": 1.92, + "learning_rate": 1.778087566849547e-05, + "loss": 0.3815, + "step": 4580 + }, + { + "epoch": 1.92, + "learning_rate": 1.7779810575712942e-05, + "loss": 0.3749, + "step": 4581 + }, + { + "epoch": 1.92, + "learning_rate": 1.777874525930666e-05, + "loss": 0.3943, + "step": 4582 + }, + { + "epoch": 1.92, + "learning_rate": 1.7777679719307253e-05, + "loss": 0.3639, + "step": 4583 + }, + { + "epoch": 1.92, + "learning_rate": 1.7776613955745337e-05, + "loss": 0.3701, + "step": 4584 + }, + { + "epoch": 1.92, + "learning_rate": 1.7775547968651562e-05, + "loss": 0.3833, + "step": 4585 + }, + { + "epoch": 1.92, + "learning_rate": 1.7774481758056553e-05, + "loss": 0.3607, + "step": 4586 + }, + { + "epoch": 1.92, + "learning_rate": 1.777341532399097e-05, + "loss": 0.3518, + "step": 4587 + }, + { + "epoch": 1.92, + "learning_rate": 1.777234866648546e-05, + "loss": 0.4178, + "step": 4588 + }, + { + "epoch": 1.92, + "learning_rate": 1.7771281785570683e-05, + "loss": 0.4101, + "step": 4589 + }, + { + "epoch": 1.92, + "learning_rate": 1.777021468127731e-05, + "loss": 0.3854, + "step": 4590 + }, + { + "epoch": 1.92, + "learning_rate": 1.776914735363601e-05, + "loss": 0.3975, + "step": 4591 + }, + { + "epoch": 1.92, + "learning_rate": 1.776807980267746e-05, + "loss": 0.409, + "step": 4592 + }, + { + "epoch": 1.92, + "learning_rate": 1.7767012028432352e-05, + "loss": 0.4173, + "step": 4593 + }, + { + "epoch": 1.92, + "learning_rate": 1.776594403093138e-05, + "loss": 0.3405, + "step": 4594 + }, + { + "epoch": 1.92, + "learning_rate": 1.7764875810205234e-05, + "loss": 0.3914, + "step": 4595 + }, + { + "epoch": 1.92, + "learning_rate": 1.776380736628462e-05, + "loss": 0.3719, + "step": 4596 + }, + { + "epoch": 1.92, + "learning_rate": 1.7762738699200256e-05, + "loss": 0.4169, + "step": 4597 + }, + { + "epoch": 1.93, + "learning_rate": 1.7761669808982857e-05, + "loss": 0.4051, + "step": 4598 + }, + { + "epoch": 1.93, + "learning_rate": 1.776060069566315e-05, + "loss": 0.3898, + "step": 4599 + }, + { + "epoch": 1.93, + "learning_rate": 1.7759531359271857e-05, + "loss": 0.3679, + "step": 4600 + }, + { + "epoch": 1.93, + "learning_rate": 1.7758461799839723e-05, + "loss": 0.3402, + "step": 4601 + }, + { + "epoch": 1.93, + "learning_rate": 1.7757392017397493e-05, + "loss": 0.391, + "step": 4602 + }, + { + "epoch": 1.93, + "learning_rate": 1.7756322011975906e-05, + "loss": 0.3498, + "step": 4603 + }, + { + "epoch": 1.93, + "learning_rate": 1.7755251783605733e-05, + "loss": 0.3862, + "step": 4604 + }, + { + "epoch": 1.93, + "learning_rate": 1.7754181332317725e-05, + "loss": 0.4389, + "step": 4605 + }, + { + "epoch": 1.93, + "learning_rate": 1.7753110658142657e-05, + "loss": 0.3772, + "step": 4606 + }, + { + "epoch": 1.93, + "learning_rate": 1.77520397611113e-05, + "loss": 0.4095, + "step": 4607 + }, + { + "epoch": 1.93, + "learning_rate": 1.7750968641254444e-05, + "loss": 0.4164, + "step": 4608 + }, + { + "epoch": 1.93, + "learning_rate": 1.7749897298602867e-05, + "loss": 0.4049, + "step": 4609 + }, + { + "epoch": 1.93, + "learning_rate": 1.7748825733187372e-05, + "loss": 0.3771, + "step": 4610 + }, + { + "epoch": 1.93, + "learning_rate": 1.774775394503876e-05, + "loss": 0.3696, + "step": 4611 + }, + { + "epoch": 1.93, + "learning_rate": 1.7746681934187832e-05, + "loss": 0.3992, + "step": 4612 + }, + { + "epoch": 1.93, + "learning_rate": 1.7745609700665406e-05, + "loss": 0.3612, + "step": 4613 + }, + { + "epoch": 1.93, + "learning_rate": 1.774453724450231e-05, + "loss": 0.3978, + "step": 4614 + }, + { + "epoch": 1.93, + "learning_rate": 1.7743464565729354e-05, + "loss": 0.4378, + "step": 4615 + }, + { + "epoch": 1.93, + "learning_rate": 1.7742391664377384e-05, + "loss": 0.3375, + "step": 4616 + }, + { + "epoch": 1.93, + "learning_rate": 1.7741318540477234e-05, + "loss": 0.4127, + "step": 4617 + }, + { + "epoch": 1.93, + "learning_rate": 1.7740245194059754e-05, + "loss": 0.3915, + "step": 4618 + }, + { + "epoch": 1.93, + "learning_rate": 1.77391716251558e-05, + "loss": 0.3586, + "step": 4619 + }, + { + "epoch": 1.93, + "learning_rate": 1.7738097833796218e-05, + "loss": 0.3378, + "step": 4620 + }, + { + "epoch": 1.93, + "learning_rate": 1.7737023820011884e-05, + "loss": 0.4145, + "step": 4621 + }, + { + "epoch": 1.94, + "learning_rate": 1.7735949583833662e-05, + "loss": 0.4131, + "step": 4622 + }, + { + "epoch": 1.94, + "learning_rate": 1.7734875125292436e-05, + "loss": 0.3654, + "step": 4623 + }, + { + "epoch": 1.94, + "learning_rate": 1.7733800444419088e-05, + "loss": 0.3907, + "step": 4624 + }, + { + "epoch": 1.94, + "learning_rate": 1.773272554124451e-05, + "loss": 0.4214, + "step": 4625 + }, + { + "epoch": 1.94, + "learning_rate": 1.77316504157996e-05, + "loss": 0.3629, + "step": 4626 + }, + { + "epoch": 1.94, + "learning_rate": 1.773057506811526e-05, + "loss": 0.3772, + "step": 4627 + }, + { + "epoch": 1.94, + "learning_rate": 1.7729499498222392e-05, + "loss": 0.4109, + "step": 4628 + }, + { + "epoch": 1.94, + "learning_rate": 1.7728423706151928e-05, + "loss": 0.36, + "step": 4629 + }, + { + "epoch": 1.94, + "learning_rate": 1.772734769193478e-05, + "loss": 0.436, + "step": 4630 + }, + { + "epoch": 1.94, + "learning_rate": 1.772627145560188e-05, + "loss": 0.3648, + "step": 4631 + }, + { + "epoch": 1.94, + "learning_rate": 1.772519499718416e-05, + "loss": 0.4065, + "step": 4632 + }, + { + "epoch": 1.94, + "learning_rate": 1.772411831671257e-05, + "loss": 0.387, + "step": 4633 + }, + { + "epoch": 1.94, + "learning_rate": 1.7723041414218053e-05, + "loss": 0.3432, + "step": 4634 + }, + { + "epoch": 1.94, + "learning_rate": 1.7721964289731562e-05, + "loss": 0.4029, + "step": 4635 + }, + { + "epoch": 1.94, + "learning_rate": 1.772088694328406e-05, + "loss": 0.3698, + "step": 4636 + }, + { + "epoch": 1.94, + "learning_rate": 1.771980937490651e-05, + "loss": 0.4266, + "step": 4637 + }, + { + "epoch": 1.94, + "learning_rate": 1.7718731584629897e-05, + "loss": 0.3723, + "step": 4638 + }, + { + "epoch": 1.94, + "learning_rate": 1.771765357248519e-05, + "loss": 0.3856, + "step": 4639 + }, + { + "epoch": 1.94, + "learning_rate": 1.7716575338503382e-05, + "loss": 0.3812, + "step": 4640 + }, + { + "epoch": 1.94, + "learning_rate": 1.7715496882715463e-05, + "loss": 0.353, + "step": 4641 + }, + { + "epoch": 1.94, + "learning_rate": 1.7714418205152434e-05, + "loss": 0.4068, + "step": 4642 + }, + { + "epoch": 1.94, + "learning_rate": 1.7713339305845293e-05, + "loss": 0.4075, + "step": 4643 + }, + { + "epoch": 1.94, + "learning_rate": 1.7712260184825063e-05, + "loss": 0.3755, + "step": 4644 + }, + { + "epoch": 1.94, + "learning_rate": 1.771118084212276e-05, + "loss": 0.4403, + "step": 4645 + }, + { + "epoch": 1.95, + "learning_rate": 1.7710101277769403e-05, + "loss": 0.4173, + "step": 4646 + }, + { + "epoch": 1.95, + "learning_rate": 1.7709021491796027e-05, + "loss": 0.3637, + "step": 4647 + }, + { + "epoch": 1.95, + "learning_rate": 1.770794148423367e-05, + "loss": 0.4326, + "step": 4648 + }, + { + "epoch": 1.95, + "learning_rate": 1.7706861255113376e-05, + "loss": 0.3747, + "step": 4649 + }, + { + "epoch": 1.95, + "learning_rate": 1.7705780804466192e-05, + "loss": 0.3416, + "step": 4650 + }, + { + "epoch": 1.95, + "learning_rate": 1.7704700132323175e-05, + "loss": 0.3899, + "step": 4651 + }, + { + "epoch": 1.95, + "learning_rate": 1.7703619238715392e-05, + "loss": 0.4293, + "step": 4652 + }, + { + "epoch": 1.95, + "learning_rate": 1.770253812367391e-05, + "loss": 0.4505, + "step": 4653 + }, + { + "epoch": 1.95, + "learning_rate": 1.7701456787229805e-05, + "loss": 0.3799, + "step": 4654 + }, + { + "epoch": 1.95, + "learning_rate": 1.770037522941416e-05, + "loss": 0.3909, + "step": 4655 + }, + { + "epoch": 1.95, + "learning_rate": 1.7699293450258062e-05, + "loss": 0.4465, + "step": 4656 + }, + { + "epoch": 1.95, + "learning_rate": 1.7698211449792604e-05, + "loss": 0.3893, + "step": 4657 + }, + { + "epoch": 1.95, + "learning_rate": 1.769712922804889e-05, + "loss": 0.3736, + "step": 4658 + }, + { + "epoch": 1.95, + "learning_rate": 1.7696046785058024e-05, + "loss": 0.3389, + "step": 4659 + }, + { + "epoch": 1.95, + "learning_rate": 1.7694964120851128e-05, + "loss": 0.4159, + "step": 4660 + }, + { + "epoch": 1.95, + "learning_rate": 1.7693881235459315e-05, + "loss": 0.391, + "step": 4661 + }, + { + "epoch": 1.95, + "learning_rate": 1.7692798128913714e-05, + "loss": 0.3976, + "step": 4662 + }, + { + "epoch": 1.95, + "learning_rate": 1.7691714801245453e-05, + "loss": 0.371, + "step": 4663 + }, + { + "epoch": 1.95, + "learning_rate": 1.7690631252485676e-05, + "loss": 0.3931, + "step": 4664 + }, + { + "epoch": 1.95, + "learning_rate": 1.7689547482665532e-05, + "loss": 0.3735, + "step": 4665 + }, + { + "epoch": 1.95, + "learning_rate": 1.768846349181617e-05, + "loss": 0.4045, + "step": 4666 + }, + { + "epoch": 1.95, + "learning_rate": 1.7687379279968745e-05, + "loss": 0.4001, + "step": 4667 + }, + { + "epoch": 1.95, + "learning_rate": 1.7686294847154423e-05, + "loss": 0.3443, + "step": 4668 + }, + { + "epoch": 1.95, + "learning_rate": 1.7685210193404377e-05, + "loss": 0.4312, + "step": 4669 + }, + { + "epoch": 1.96, + "learning_rate": 1.7684125318749785e-05, + "loss": 0.3976, + "step": 4670 + }, + { + "epoch": 1.96, + "learning_rate": 1.7683040223221828e-05, + "loss": 0.4214, + "step": 4671 + }, + { + "epoch": 1.96, + "learning_rate": 1.76819549068517e-05, + "loss": 0.3772, + "step": 4672 + }, + { + "epoch": 1.96, + "learning_rate": 1.768086936967059e-05, + "loss": 0.3983, + "step": 4673 + }, + { + "epoch": 1.96, + "learning_rate": 1.767978361170971e-05, + "loss": 0.4451, + "step": 4674 + }, + { + "epoch": 1.96, + "learning_rate": 1.7678697633000267e-05, + "loss": 0.3974, + "step": 4675 + }, + { + "epoch": 1.96, + "learning_rate": 1.767761143357347e-05, + "loss": 0.3598, + "step": 4676 + }, + { + "epoch": 1.96, + "learning_rate": 1.7676525013460547e-05, + "loss": 0.3909, + "step": 4677 + }, + { + "epoch": 1.96, + "learning_rate": 1.7675438372692726e-05, + "loss": 0.4153, + "step": 4678 + }, + { + "epoch": 1.96, + "learning_rate": 1.7674351511301238e-05, + "loss": 0.3957, + "step": 4679 + }, + { + "epoch": 1.96, + "learning_rate": 1.767326442931733e-05, + "loss": 0.3571, + "step": 4680 + }, + { + "epoch": 1.96, + "learning_rate": 1.7672177126772237e-05, + "loss": 0.3678, + "step": 4681 + }, + { + "epoch": 1.96, + "learning_rate": 1.7671089603697228e-05, + "loss": 0.4258, + "step": 4682 + }, + { + "epoch": 1.96, + "learning_rate": 1.767000186012355e-05, + "loss": 0.4408, + "step": 4683 + }, + { + "epoch": 1.96, + "learning_rate": 1.766891389608248e-05, + "loss": 0.409, + "step": 4684 + }, + { + "epoch": 1.96, + "learning_rate": 1.7667825711605283e-05, + "loss": 0.4514, + "step": 4685 + }, + { + "epoch": 1.96, + "learning_rate": 1.766673730672324e-05, + "loss": 0.3947, + "step": 4686 + }, + { + "epoch": 1.96, + "learning_rate": 1.766564868146764e-05, + "loss": 0.349, + "step": 4687 + }, + { + "epoch": 1.96, + "learning_rate": 1.7664559835869763e-05, + "loss": 0.4218, + "step": 4688 + }, + { + "epoch": 1.96, + "learning_rate": 1.766347076996092e-05, + "loss": 0.4156, + "step": 4689 + }, + { + "epoch": 1.96, + "learning_rate": 1.7662381483772412e-05, + "loss": 0.3596, + "step": 4690 + }, + { + "epoch": 1.96, + "learning_rate": 1.7661291977335544e-05, + "loss": 0.4368, + "step": 4691 + }, + { + "epoch": 1.96, + "learning_rate": 1.766020225068164e-05, + "loss": 0.3567, + "step": 4692 + }, + { + "epoch": 1.96, + "learning_rate": 1.7659112303842015e-05, + "loss": 0.3661, + "step": 4693 + }, + { + "epoch": 1.97, + "learning_rate": 1.7658022136848008e-05, + "loss": 0.3629, + "step": 4694 + }, + { + "epoch": 1.97, + "learning_rate": 1.7656931749730947e-05, + "loss": 0.3877, + "step": 4695 + }, + { + "epoch": 1.97, + "learning_rate": 1.765584114252218e-05, + "loss": 0.3517, + "step": 4696 + }, + { + "epoch": 1.97, + "learning_rate": 1.7654750315253055e-05, + "loss": 0.3632, + "step": 4697 + }, + { + "epoch": 1.97, + "learning_rate": 1.765365926795492e-05, + "loss": 0.3851, + "step": 4698 + }, + { + "epoch": 1.97, + "learning_rate": 1.7652568000659144e-05, + "loss": 0.4156, + "step": 4699 + }, + { + "epoch": 1.97, + "learning_rate": 1.765147651339709e-05, + "loss": 0.3623, + "step": 4700 + }, + { + "epoch": 1.97, + "learning_rate": 1.7650384806200134e-05, + "loss": 0.4243, + "step": 4701 + }, + { + "epoch": 1.97, + "learning_rate": 1.7649292879099655e-05, + "loss": 0.3677, + "step": 4702 + }, + { + "epoch": 1.97, + "learning_rate": 1.764820073212704e-05, + "loss": 0.4255, + "step": 4703 + }, + { + "epoch": 1.97, + "learning_rate": 1.7647108365313683e-05, + "loss": 0.4099, + "step": 4704 + }, + { + "epoch": 1.97, + "learning_rate": 1.7646015778690983e-05, + "loss": 0.4019, + "step": 4705 + }, + { + "epoch": 1.97, + "learning_rate": 1.7644922972290344e-05, + "loss": 0.4247, + "step": 4706 + }, + { + "epoch": 1.97, + "learning_rate": 1.7643829946143176e-05, + "loss": 0.4158, + "step": 4707 + }, + { + "epoch": 1.97, + "learning_rate": 1.76427367002809e-05, + "loss": 0.354, + "step": 4708 + }, + { + "epoch": 1.97, + "learning_rate": 1.764164323473494e-05, + "loss": 0.4356, + "step": 4709 + }, + { + "epoch": 1.97, + "learning_rate": 1.7640549549536727e-05, + "loss": 0.38, + "step": 4710 + }, + { + "epoch": 1.97, + "learning_rate": 1.7639455644717694e-05, + "loss": 0.3709, + "step": 4711 + }, + { + "epoch": 1.97, + "learning_rate": 1.7638361520309294e-05, + "loss": 0.4102, + "step": 4712 + }, + { + "epoch": 1.97, + "learning_rate": 1.7637267176342964e-05, + "loss": 0.3607, + "step": 4713 + }, + { + "epoch": 1.97, + "learning_rate": 1.763617261285017e-05, + "loss": 0.417, + "step": 4714 + }, + { + "epoch": 1.97, + "learning_rate": 1.763507782986237e-05, + "loss": 0.3558, + "step": 4715 + }, + { + "epoch": 1.97, + "learning_rate": 1.763398282741103e-05, + "loss": 0.381, + "step": 4716 + }, + { + "epoch": 1.97, + "learning_rate": 1.763288760552763e-05, + "loss": 0.3835, + "step": 4717 + }, + { + "epoch": 1.98, + "learning_rate": 1.763179216424365e-05, + "loss": 0.4098, + "step": 4718 + }, + { + "epoch": 1.98, + "learning_rate": 1.7630696503590577e-05, + "loss": 0.3972, + "step": 4719 + }, + { + "epoch": 1.98, + "learning_rate": 1.7629600623599904e-05, + "loss": 0.4298, + "step": 4720 + }, + { + "epoch": 1.98, + "learning_rate": 1.7628504524303128e-05, + "loss": 0.382, + "step": 4721 + }, + { + "epoch": 1.98, + "learning_rate": 1.7627408205731762e-05, + "loss": 0.43, + "step": 4722 + }, + { + "epoch": 1.98, + "learning_rate": 1.7626311667917315e-05, + "loss": 0.3753, + "step": 4723 + }, + { + "epoch": 1.98, + "learning_rate": 1.7625214910891307e-05, + "loss": 0.4242, + "step": 4724 + }, + { + "epoch": 1.98, + "learning_rate": 1.7624117934685264e-05, + "loss": 0.4101, + "step": 4725 + }, + { + "epoch": 1.98, + "learning_rate": 1.7623020739330714e-05, + "loss": 0.3907, + "step": 4726 + }, + { + "epoch": 1.98, + "learning_rate": 1.7621923324859196e-05, + "loss": 0.3705, + "step": 4727 + }, + { + "epoch": 1.98, + "learning_rate": 1.762082569130226e-05, + "loss": 0.3592, + "step": 4728 + }, + { + "epoch": 1.98, + "learning_rate": 1.761972783869145e-05, + "loss": 0.401, + "step": 4729 + }, + { + "epoch": 1.98, + "learning_rate": 1.7618629767058326e-05, + "loss": 0.4288, + "step": 4730 + }, + { + "epoch": 1.98, + "learning_rate": 1.7617531476434447e-05, + "loss": 0.3446, + "step": 4731 + }, + { + "epoch": 1.98, + "learning_rate": 1.7616432966851386e-05, + "loss": 0.3834, + "step": 4732 + }, + { + "epoch": 1.98, + "learning_rate": 1.761533423834072e-05, + "loss": 0.4065, + "step": 4733 + }, + { + "epoch": 1.98, + "learning_rate": 1.761423529093403e-05, + "loss": 0.4022, + "step": 4734 + }, + { + "epoch": 1.98, + "learning_rate": 1.76131361246629e-05, + "loss": 0.3807, + "step": 4735 + }, + { + "epoch": 1.98, + "learning_rate": 1.7612036739558932e-05, + "loss": 0.3898, + "step": 4736 + }, + { + "epoch": 1.98, + "learning_rate": 1.761093713565372e-05, + "loss": 0.3873, + "step": 4737 + }, + { + "epoch": 1.98, + "learning_rate": 1.7609837312978874e-05, + "loss": 0.4046, + "step": 4738 + }, + { + "epoch": 1.98, + "learning_rate": 1.7608737271566004e-05, + "loss": 0.4069, + "step": 4739 + }, + { + "epoch": 1.98, + "learning_rate": 1.7607637011446737e-05, + "loss": 0.3928, + "step": 4740 + }, + { + "epoch": 1.98, + "learning_rate": 1.7606536532652695e-05, + "loss": 0.3691, + "step": 4741 + }, + { + "epoch": 1.99, + "learning_rate": 1.760543583521551e-05, + "loss": 0.4312, + "step": 4742 + }, + { + "epoch": 1.99, + "learning_rate": 1.7604334919166822e-05, + "loss": 0.3853, + "step": 4743 + }, + { + "epoch": 1.99, + "learning_rate": 1.7603233784538275e-05, + "loss": 0.4136, + "step": 4744 + }, + { + "epoch": 1.99, + "learning_rate": 1.7602132431361516e-05, + "loss": 0.4116, + "step": 4745 + }, + { + "epoch": 1.99, + "learning_rate": 1.7601030859668207e-05, + "loss": 0.3834, + "step": 4746 + }, + { + "epoch": 1.99, + "learning_rate": 1.7599929069490014e-05, + "loss": 0.4323, + "step": 4747 + }, + { + "epoch": 1.99, + "learning_rate": 1.7598827060858605e-05, + "loss": 0.4095, + "step": 4748 + }, + { + "epoch": 1.99, + "learning_rate": 1.759772483380565e-05, + "loss": 0.4095, + "step": 4749 + }, + { + "epoch": 1.99, + "learning_rate": 1.759662238836284e-05, + "loss": 0.4272, + "step": 4750 + }, + { + "epoch": 1.99, + "learning_rate": 1.759551972456186e-05, + "loss": 0.405, + "step": 4751 + }, + { + "epoch": 1.99, + "learning_rate": 1.7594416842434405e-05, + "loss": 0.3806, + "step": 4752 + }, + { + "epoch": 1.99, + "learning_rate": 1.759331374201218e-05, + "loss": 0.3747, + "step": 4753 + }, + { + "epoch": 1.99, + "learning_rate": 1.759221042332689e-05, + "loss": 0.3804, + "step": 4754 + }, + { + "epoch": 1.99, + "learning_rate": 1.759110688641024e-05, + "loss": 0.4165, + "step": 4755 + }, + { + "epoch": 1.99, + "learning_rate": 1.7590003131293967e-05, + "loss": 0.3855, + "step": 4756 + }, + { + "epoch": 1.99, + "learning_rate": 1.7588899158009786e-05, + "loss": 0.373, + "step": 4757 + }, + { + "epoch": 1.99, + "learning_rate": 1.7587794966589437e-05, + "loss": 0.3976, + "step": 4758 + }, + { + "epoch": 1.99, + "learning_rate": 1.758669055706465e-05, + "loss": 0.3288, + "step": 4759 + }, + { + "epoch": 1.99, + "learning_rate": 1.758558592946718e-05, + "loss": 0.3985, + "step": 4760 + }, + { + "epoch": 1.99, + "learning_rate": 1.7584481083828766e-05, + "loss": 0.4002, + "step": 4761 + }, + { + "epoch": 1.99, + "learning_rate": 1.7583376020181183e-05, + "loss": 0.3729, + "step": 4762 + }, + { + "epoch": 1.99, + "learning_rate": 1.758227073855618e-05, + "loss": 0.4247, + "step": 4763 + }, + { + "epoch": 1.99, + "learning_rate": 1.7581165238985533e-05, + "loss": 0.4128, + "step": 4764 + }, + { + "epoch": 1.99, + "learning_rate": 1.7580059521501017e-05, + "loss": 0.3545, + "step": 4765 + }, + { + "epoch": 2.0, + "learning_rate": 1.7578953586134423e-05, + "loss": 0.3787, + "step": 4766 + }, + { + "epoch": 2.0, + "learning_rate": 1.7577847432917528e-05, + "loss": 0.3957, + "step": 4767 + }, + { + "epoch": 2.0, + "learning_rate": 1.7576741061882136e-05, + "loss": 0.3975, + "step": 4768 + }, + { + "epoch": 2.0, + "learning_rate": 1.7575634473060045e-05, + "loss": 0.3689, + "step": 4769 + }, + { + "epoch": 2.0, + "learning_rate": 1.757452766648306e-05, + "loss": 0.4133, + "step": 4770 + }, + { + "epoch": 2.0, + "learning_rate": 1.7573420642183004e-05, + "loss": 0.3487, + "step": 4771 + }, + { + "epoch": 2.0, + "learning_rate": 1.757231340019169e-05, + "loss": 0.3746, + "step": 4772 + }, + { + "epoch": 2.0, + "learning_rate": 1.7571205940540945e-05, + "loss": 0.4144, + "step": 4773 + }, + { + "epoch": 2.0, + "learning_rate": 1.7570098263262608e-05, + "loss": 0.3845, + "step": 4774 + }, + { + "epoch": 2.0, + "learning_rate": 1.756899036838851e-05, + "loss": 0.4326, + "step": 4775 + }, + { + "epoch": 2.0, + "learning_rate": 1.7567882255950503e-05, + "loss": 0.3347, + "step": 4776 + }, + { + "epoch": 2.0, + "learning_rate": 1.7566773925980435e-05, + "loss": 0.3743, + "step": 4777 + }, + { + "epoch": 2.0, + "learning_rate": 1.7565665378510165e-05, + "loss": 0.2799, + "step": 4778 + }, + { + "epoch": 2.0, + "learning_rate": 1.756455661357156e-05, + "loss": 0.315, + "step": 4779 + }, + { + "epoch": 2.0, + "learning_rate": 1.7563447631196484e-05, + "loss": 0.2846, + "step": 4780 + }, + { + "epoch": 2.0, + "learning_rate": 1.7562338431416818e-05, + "loss": 0.262, + "step": 4781 + }, + { + "epoch": 2.0, + "learning_rate": 1.7561229014264448e-05, + "loss": 0.2669, + "step": 4782 + }, + { + "epoch": 2.0, + "learning_rate": 1.7560119379771257e-05, + "loss": 0.2785, + "step": 4783 + }, + { + "epoch": 2.0, + "learning_rate": 1.7559009527969145e-05, + "loss": 0.27, + "step": 4784 + }, + { + "epoch": 2.0, + "learning_rate": 1.755789945889001e-05, + "loss": 0.2426, + "step": 4785 + }, + { + "epoch": 2.0, + "learning_rate": 1.7556789172565762e-05, + "loss": 0.266, + "step": 4786 + }, + { + "epoch": 2.0, + "learning_rate": 1.7555678669028315e-05, + "loss": 0.242, + "step": 4787 + }, + { + "epoch": 2.0, + "learning_rate": 1.755456794830959e-05, + "loss": 0.246, + "step": 4788 + }, + { + "epoch": 2.01, + "learning_rate": 1.7553457010441514e-05, + "loss": 0.2788, + "step": 4789 + }, + { + "epoch": 2.01, + "learning_rate": 1.7552345855456017e-05, + "loss": 0.2595, + "step": 4790 + }, + { + "epoch": 2.01, + "learning_rate": 1.755123448338504e-05, + "loss": 0.2886, + "step": 4791 + }, + { + "epoch": 2.01, + "learning_rate": 1.755012289426053e-05, + "loss": 0.2787, + "step": 4792 + }, + { + "epoch": 2.01, + "learning_rate": 1.754901108811444e-05, + "loss": 0.2442, + "step": 4793 + }, + { + "epoch": 2.01, + "learning_rate": 1.754789906497872e-05, + "loss": 0.2656, + "step": 4794 + }, + { + "epoch": 2.01, + "learning_rate": 1.754678682488534e-05, + "loss": 0.2701, + "step": 4795 + }, + { + "epoch": 2.01, + "learning_rate": 1.7545674367866273e-05, + "loss": 0.2597, + "step": 4796 + }, + { + "epoch": 2.01, + "learning_rate": 1.754456169395349e-05, + "loss": 0.2762, + "step": 4797 + }, + { + "epoch": 2.01, + "learning_rate": 1.7543448803178975e-05, + "loss": 0.2363, + "step": 4798 + }, + { + "epoch": 2.01, + "learning_rate": 1.754233569557472e-05, + "loss": 0.2452, + "step": 4799 + }, + { + "epoch": 2.01, + "learning_rate": 1.7541222371172715e-05, + "loss": 0.2096, + "step": 4800 + }, + { + "epoch": 2.01, + "learning_rate": 1.754010883000497e-05, + "loss": 0.264, + "step": 4801 + }, + { + "epoch": 2.01, + "learning_rate": 1.7538995072103485e-05, + "loss": 0.2442, + "step": 4802 + }, + { + "epoch": 2.01, + "learning_rate": 1.7537881097500277e-05, + "loss": 0.2588, + "step": 4803 + }, + { + "epoch": 2.01, + "learning_rate": 1.7536766906227365e-05, + "loss": 0.2637, + "step": 4804 + }, + { + "epoch": 2.01, + "learning_rate": 1.753565249831678e-05, + "loss": 0.255, + "step": 4805 + }, + { + "epoch": 2.01, + "learning_rate": 1.7534537873800544e-05, + "loss": 0.2544, + "step": 4806 + }, + { + "epoch": 2.01, + "learning_rate": 1.7533423032710705e-05, + "loss": 0.2545, + "step": 4807 + }, + { + "epoch": 2.01, + "learning_rate": 1.753230797507931e-05, + "loss": 0.2724, + "step": 4808 + }, + { + "epoch": 2.01, + "learning_rate": 1.7531192700938406e-05, + "loss": 0.2539, + "step": 4809 + }, + { + "epoch": 2.01, + "learning_rate": 1.7530077210320047e-05, + "loss": 0.2681, + "step": 4810 + }, + { + "epoch": 2.01, + "learning_rate": 1.7528961503256302e-05, + "loss": 0.2636, + "step": 4811 + }, + { + "epoch": 2.01, + "learning_rate": 1.752784557977924e-05, + "loss": 0.2572, + "step": 4812 + }, + { + "epoch": 2.02, + "learning_rate": 1.752672943992094e-05, + "loss": 0.27, + "step": 4813 + }, + { + "epoch": 2.02, + "learning_rate": 1.752561308371348e-05, + "loss": 0.2464, + "step": 4814 + }, + { + "epoch": 2.02, + "learning_rate": 1.7524496511188948e-05, + "loss": 0.2508, + "step": 4815 + }, + { + "epoch": 2.02, + "learning_rate": 1.752337972237944e-05, + "loss": 0.2722, + "step": 4816 + }, + { + "epoch": 2.02, + "learning_rate": 1.752226271731706e-05, + "loss": 0.2406, + "step": 4817 + }, + { + "epoch": 2.02, + "learning_rate": 1.7521145496033912e-05, + "loss": 0.258, + "step": 4818 + }, + { + "epoch": 2.02, + "learning_rate": 1.752002805856211e-05, + "loss": 0.2664, + "step": 4819 + }, + { + "epoch": 2.02, + "learning_rate": 1.7518910404933777e-05, + "loss": 0.2426, + "step": 4820 + }, + { + "epoch": 2.02, + "learning_rate": 1.751779253518104e-05, + "loss": 0.2263, + "step": 4821 + }, + { + "epoch": 2.02, + "learning_rate": 1.751667444933602e-05, + "loss": 0.2425, + "step": 4822 + }, + { + "epoch": 2.02, + "learning_rate": 1.7515556147430867e-05, + "loss": 0.2282, + "step": 4823 + }, + { + "epoch": 2.02, + "learning_rate": 1.751443762949772e-05, + "loss": 0.2368, + "step": 4824 + }, + { + "epoch": 2.02, + "learning_rate": 1.7513318895568734e-05, + "loss": 0.2329, + "step": 4825 + }, + { + "epoch": 2.02, + "learning_rate": 1.7512199945676065e-05, + "loss": 0.2456, + "step": 4826 + }, + { + "epoch": 2.02, + "learning_rate": 1.751108077985187e-05, + "loss": 0.2523, + "step": 4827 + }, + { + "epoch": 2.02, + "learning_rate": 1.7509961398128325e-05, + "loss": 0.2661, + "step": 4828 + }, + { + "epoch": 2.02, + "learning_rate": 1.7508841800537608e-05, + "loss": 0.2412, + "step": 4829 + }, + { + "epoch": 2.02, + "learning_rate": 1.750772198711189e-05, + "loss": 0.242, + "step": 4830 + }, + { + "epoch": 2.02, + "learning_rate": 1.750660195788337e-05, + "loss": 0.2648, + "step": 4831 + }, + { + "epoch": 2.02, + "learning_rate": 1.7505481712884235e-05, + "loss": 0.2752, + "step": 4832 + }, + { + "epoch": 2.02, + "learning_rate": 1.750436125214669e-05, + "loss": 0.2557, + "step": 4833 + }, + { + "epoch": 2.02, + "learning_rate": 1.750324057570294e-05, + "loss": 0.2343, + "step": 4834 + }, + { + "epoch": 2.02, + "learning_rate": 1.75021196835852e-05, + "loss": 0.2432, + "step": 4835 + }, + { + "epoch": 2.02, + "learning_rate": 1.750099857582568e-05, + "loss": 0.2425, + "step": 4836 + }, + { + "epoch": 2.03, + "learning_rate": 1.7499877252456624e-05, + "loss": 0.234, + "step": 4837 + }, + { + "epoch": 2.03, + "learning_rate": 1.7498755713510244e-05, + "loss": 0.2568, + "step": 4838 + }, + { + "epoch": 2.03, + "learning_rate": 1.7497633959018784e-05, + "loss": 0.2783, + "step": 4839 + }, + { + "epoch": 2.03, + "learning_rate": 1.749651198901449e-05, + "loss": 0.2693, + "step": 4840 + }, + { + "epoch": 2.03, + "learning_rate": 1.7495389803529615e-05, + "loss": 0.2427, + "step": 4841 + }, + { + "epoch": 2.03, + "learning_rate": 1.749426740259641e-05, + "loss": 0.2645, + "step": 4842 + }, + { + "epoch": 2.03, + "learning_rate": 1.7493144786247136e-05, + "loss": 0.2245, + "step": 4843 + }, + { + "epoch": 2.03, + "learning_rate": 1.7492021954514068e-05, + "loss": 0.2344, + "step": 4844 + }, + { + "epoch": 2.03, + "learning_rate": 1.7490898907429473e-05, + "loss": 0.2408, + "step": 4845 + }, + { + "epoch": 2.03, + "learning_rate": 1.748977564502564e-05, + "loss": 0.2739, + "step": 4846 + }, + { + "epoch": 2.03, + "learning_rate": 1.748865216733485e-05, + "loss": 0.2663, + "step": 4847 + }, + { + "epoch": 2.03, + "learning_rate": 1.7487528474389403e-05, + "loss": 0.2342, + "step": 4848 + }, + { + "epoch": 2.03, + "learning_rate": 1.748640456622159e-05, + "loss": 0.2576, + "step": 4849 + }, + { + "epoch": 2.03, + "learning_rate": 1.7485280442863718e-05, + "loss": 0.2432, + "step": 4850 + }, + { + "epoch": 2.03, + "learning_rate": 1.7484156104348107e-05, + "loss": 0.2371, + "step": 4851 + }, + { + "epoch": 2.03, + "learning_rate": 1.748303155070707e-05, + "loss": 0.2528, + "step": 4852 + }, + { + "epoch": 2.03, + "learning_rate": 1.7481906781972932e-05, + "loss": 0.2573, + "step": 4853 + }, + { + "epoch": 2.03, + "learning_rate": 1.748078179817802e-05, + "loss": 0.2821, + "step": 4854 + }, + { + "epoch": 2.03, + "learning_rate": 1.7479656599354677e-05, + "loss": 0.2348, + "step": 4855 + }, + { + "epoch": 2.03, + "learning_rate": 1.7478531185535235e-05, + "loss": 0.2659, + "step": 4856 + }, + { + "epoch": 2.03, + "learning_rate": 1.747740555675206e-05, + "loss": 0.2252, + "step": 4857 + }, + { + "epoch": 2.03, + "learning_rate": 1.747627971303749e-05, + "loss": 0.2632, + "step": 4858 + }, + { + "epoch": 2.03, + "learning_rate": 1.74751536544239e-05, + "loss": 0.2315, + "step": 4859 + }, + { + "epoch": 2.03, + "learning_rate": 1.747402738094365e-05, + "loss": 0.2415, + "step": 4860 + }, + { + "epoch": 2.04, + "learning_rate": 1.7472900892629114e-05, + "loss": 0.234, + "step": 4861 + }, + { + "epoch": 2.04, + "learning_rate": 1.747177418951267e-05, + "loss": 0.2695, + "step": 4862 + }, + { + "epoch": 2.04, + "learning_rate": 1.7470647271626712e-05, + "loss": 0.2861, + "step": 4863 + }, + { + "epoch": 2.04, + "learning_rate": 1.7469520139003627e-05, + "loss": 0.2673, + "step": 4864 + }, + { + "epoch": 2.04, + "learning_rate": 1.7468392791675812e-05, + "loss": 0.254, + "step": 4865 + }, + { + "epoch": 2.04, + "learning_rate": 1.7467265229675673e-05, + "loss": 0.2536, + "step": 4866 + }, + { + "epoch": 2.04, + "learning_rate": 1.7466137453035626e-05, + "loss": 0.2572, + "step": 4867 + }, + { + "epoch": 2.04, + "learning_rate": 1.746500946178808e-05, + "loss": 0.2533, + "step": 4868 + }, + { + "epoch": 2.04, + "learning_rate": 1.746388125596546e-05, + "loss": 0.2277, + "step": 4869 + }, + { + "epoch": 2.04, + "learning_rate": 1.7462752835600198e-05, + "loss": 0.226, + "step": 4870 + }, + { + "epoch": 2.04, + "learning_rate": 1.746162420072473e-05, + "loss": 0.2779, + "step": 4871 + }, + { + "epoch": 2.04, + "learning_rate": 1.7460495351371492e-05, + "loss": 0.2378, + "step": 4872 + }, + { + "epoch": 2.04, + "learning_rate": 1.745936628757294e-05, + "loss": 0.2827, + "step": 4873 + }, + { + "epoch": 2.04, + "learning_rate": 1.745823700936152e-05, + "loss": 0.2935, + "step": 4874 + }, + { + "epoch": 2.04, + "learning_rate": 1.7457107516769697e-05, + "loss": 0.2524, + "step": 4875 + }, + { + "epoch": 2.04, + "learning_rate": 1.7455977809829933e-05, + "loss": 0.2445, + "step": 4876 + }, + { + "epoch": 2.04, + "learning_rate": 1.7454847888574705e-05, + "loss": 0.2442, + "step": 4877 + }, + { + "epoch": 2.04, + "learning_rate": 1.745371775303649e-05, + "loss": 0.255, + "step": 4878 + }, + { + "epoch": 2.04, + "learning_rate": 1.7452587403247772e-05, + "loss": 0.2344, + "step": 4879 + }, + { + "epoch": 2.04, + "learning_rate": 1.7451456839241044e-05, + "loss": 0.2695, + "step": 4880 + }, + { + "epoch": 2.04, + "learning_rate": 1.7450326061048798e-05, + "loss": 0.2985, + "step": 4881 + }, + { + "epoch": 2.04, + "learning_rate": 1.7449195068703542e-05, + "loss": 0.2883, + "step": 4882 + }, + { + "epoch": 2.04, + "learning_rate": 1.7448063862237784e-05, + "loss": 0.2591, + "step": 4883 + }, + { + "epoch": 2.04, + "learning_rate": 1.7446932441684044e-05, + "loss": 0.2699, + "step": 4884 + }, + { + "epoch": 2.05, + "learning_rate": 1.7445800807074834e-05, + "loss": 0.2448, + "step": 4885 + }, + { + "epoch": 2.05, + "learning_rate": 1.744466895844269e-05, + "loss": 0.2717, + "step": 4886 + }, + { + "epoch": 2.05, + "learning_rate": 1.7443536895820143e-05, + "loss": 0.2549, + "step": 4887 + }, + { + "epoch": 2.05, + "learning_rate": 1.744240461923973e-05, + "loss": 0.261, + "step": 4888 + }, + { + "epoch": 2.05, + "learning_rate": 1.7441272128734004e-05, + "loss": 0.2292, + "step": 4889 + }, + { + "epoch": 2.05, + "learning_rate": 1.7440139424335513e-05, + "loss": 0.2494, + "step": 4890 + }, + { + "epoch": 2.05, + "learning_rate": 1.7439006506076816e-05, + "loss": 0.2344, + "step": 4891 + }, + { + "epoch": 2.05, + "learning_rate": 1.7437873373990478e-05, + "loss": 0.2476, + "step": 4892 + }, + { + "epoch": 2.05, + "learning_rate": 1.743674002810907e-05, + "loss": 0.2553, + "step": 4893 + }, + { + "epoch": 2.05, + "learning_rate": 1.743560646846517e-05, + "loss": 0.2397, + "step": 4894 + }, + { + "epoch": 2.05, + "learning_rate": 1.7434472695091358e-05, + "loss": 0.2433, + "step": 4895 + }, + { + "epoch": 2.05, + "learning_rate": 1.743333870802023e-05, + "loss": 0.2665, + "step": 4896 + }, + { + "epoch": 2.05, + "learning_rate": 1.7432204507284373e-05, + "loss": 0.2498, + "step": 4897 + }, + { + "epoch": 2.05, + "learning_rate": 1.7431070092916395e-05, + "loss": 0.2239, + "step": 4898 + }, + { + "epoch": 2.05, + "learning_rate": 1.74299354649489e-05, + "loss": 0.2373, + "step": 4899 + }, + { + "epoch": 2.05, + "learning_rate": 1.7428800623414502e-05, + "loss": 0.246, + "step": 4900 + }, + { + "epoch": 2.05, + "learning_rate": 1.7427665568345826e-05, + "loss": 0.2617, + "step": 4901 + }, + { + "epoch": 2.05, + "learning_rate": 1.7426530299775492e-05, + "loss": 0.247, + "step": 4902 + }, + { + "epoch": 2.05, + "learning_rate": 1.7425394817736135e-05, + "loss": 0.2407, + "step": 4903 + }, + { + "epoch": 2.05, + "learning_rate": 1.7424259122260396e-05, + "loss": 0.2377, + "step": 4904 + }, + { + "epoch": 2.05, + "learning_rate": 1.7423123213380914e-05, + "loss": 0.2585, + "step": 4905 + }, + { + "epoch": 2.05, + "learning_rate": 1.742198709113034e-05, + "loss": 0.2342, + "step": 4906 + }, + { + "epoch": 2.05, + "learning_rate": 1.7420850755541335e-05, + "loss": 0.2528, + "step": 4907 + }, + { + "epoch": 2.05, + "learning_rate": 1.7419714206646563e-05, + "loss": 0.2572, + "step": 4908 + }, + { + "epoch": 2.06, + "learning_rate": 1.741857744447869e-05, + "loss": 0.2499, + "step": 4909 + }, + { + "epoch": 2.06, + "learning_rate": 1.7417440469070386e-05, + "loss": 0.2417, + "step": 4910 + }, + { + "epoch": 2.06, + "learning_rate": 1.7416303280454343e-05, + "loss": 0.2383, + "step": 4911 + }, + { + "epoch": 2.06, + "learning_rate": 1.741516587866324e-05, + "loss": 0.2331, + "step": 4912 + }, + { + "epoch": 2.06, + "learning_rate": 1.7414028263729776e-05, + "loss": 0.2658, + "step": 4913 + }, + { + "epoch": 2.06, + "learning_rate": 1.741289043568665e-05, + "loss": 0.2369, + "step": 4914 + }, + { + "epoch": 2.06, + "learning_rate": 1.7411752394566565e-05, + "loss": 0.279, + "step": 4915 + }, + { + "epoch": 2.06, + "learning_rate": 1.7410614140402234e-05, + "loss": 0.2687, + "step": 4916 + }, + { + "epoch": 2.06, + "learning_rate": 1.7409475673226376e-05, + "loss": 0.2613, + "step": 4917 + }, + { + "epoch": 2.06, + "learning_rate": 1.7408336993071716e-05, + "loss": 0.2657, + "step": 4918 + }, + { + "epoch": 2.06, + "learning_rate": 1.7407198099970982e-05, + "loss": 0.2621, + "step": 4919 + }, + { + "epoch": 2.06, + "learning_rate": 1.7406058993956908e-05, + "loss": 0.2789, + "step": 4920 + }, + { + "epoch": 2.06, + "learning_rate": 1.7404919675062245e-05, + "loss": 0.2448, + "step": 4921 + }, + { + "epoch": 2.06, + "learning_rate": 1.7403780143319736e-05, + "loss": 0.2405, + "step": 4922 + }, + { + "epoch": 2.06, + "learning_rate": 1.7402640398762137e-05, + "loss": 0.2727, + "step": 4923 + }, + { + "epoch": 2.06, + "learning_rate": 1.740150044142221e-05, + "loss": 0.2417, + "step": 4924 + }, + { + "epoch": 2.06, + "learning_rate": 1.7400360271332718e-05, + "loss": 0.2379, + "step": 4925 + }, + { + "epoch": 2.06, + "learning_rate": 1.7399219888526438e-05, + "loss": 0.2408, + "step": 4926 + }, + { + "epoch": 2.06, + "learning_rate": 1.739807929303615e-05, + "loss": 0.242, + "step": 4927 + }, + { + "epoch": 2.06, + "learning_rate": 1.7396938484894636e-05, + "loss": 0.251, + "step": 4928 + }, + { + "epoch": 2.06, + "learning_rate": 1.739579746413469e-05, + "loss": 0.2409, + "step": 4929 + }, + { + "epoch": 2.06, + "learning_rate": 1.7394656230789106e-05, + "loss": 0.2426, + "step": 4930 + }, + { + "epoch": 2.06, + "learning_rate": 1.7393514784890694e-05, + "loss": 0.2722, + "step": 4931 + }, + { + "epoch": 2.06, + "learning_rate": 1.7392373126472257e-05, + "loss": 0.2441, + "step": 4932 + }, + { + "epoch": 2.07, + "learning_rate": 1.739123125556662e-05, + "loss": 0.2293, + "step": 4933 + }, + { + "epoch": 2.07, + "learning_rate": 1.7390089172206594e-05, + "loss": 0.2598, + "step": 4934 + }, + { + "epoch": 2.07, + "learning_rate": 1.7388946876425016e-05, + "loss": 0.2543, + "step": 4935 + }, + { + "epoch": 2.07, + "learning_rate": 1.7387804368254713e-05, + "loss": 0.2322, + "step": 4936 + }, + { + "epoch": 2.07, + "learning_rate": 1.738666164772853e-05, + "loss": 0.2598, + "step": 4937 + }, + { + "epoch": 2.07, + "learning_rate": 1.7385518714879315e-05, + "loss": 0.2482, + "step": 4938 + }, + { + "epoch": 2.07, + "learning_rate": 1.7384375569739918e-05, + "loss": 0.2783, + "step": 4939 + }, + { + "epoch": 2.07, + "learning_rate": 1.73832322123432e-05, + "loss": 0.2674, + "step": 4940 + }, + { + "epoch": 2.07, + "learning_rate": 1.738208864272202e-05, + "loss": 0.2536, + "step": 4941 + }, + { + "epoch": 2.07, + "learning_rate": 1.7380944860909253e-05, + "loss": 0.2471, + "step": 4942 + }, + { + "epoch": 2.07, + "learning_rate": 1.7379800866937773e-05, + "loss": 0.2384, + "step": 4943 + }, + { + "epoch": 2.07, + "learning_rate": 1.7378656660840475e-05, + "loss": 0.238, + "step": 4944 + }, + { + "epoch": 2.07, + "learning_rate": 1.7377512242650234e-05, + "loss": 0.2664, + "step": 4945 + }, + { + "epoch": 2.07, + "learning_rate": 1.7376367612399947e-05, + "loss": 0.2523, + "step": 4946 + }, + { + "epoch": 2.07, + "learning_rate": 1.737522277012252e-05, + "loss": 0.2584, + "step": 4947 + }, + { + "epoch": 2.07, + "learning_rate": 1.7374077715850865e-05, + "loss": 0.2505, + "step": 4948 + }, + { + "epoch": 2.07, + "learning_rate": 1.7372932449617884e-05, + "loss": 0.2087, + "step": 4949 + }, + { + "epoch": 2.07, + "learning_rate": 1.7371786971456504e-05, + "loss": 0.2631, + "step": 4950 + }, + { + "epoch": 2.07, + "learning_rate": 1.7370641281399647e-05, + "loss": 0.2492, + "step": 4951 + }, + { + "epoch": 2.07, + "learning_rate": 1.736949537948025e-05, + "loss": 0.2934, + "step": 4952 + }, + { + "epoch": 2.07, + "learning_rate": 1.7368349265731246e-05, + "loss": 0.2197, + "step": 4953 + }, + { + "epoch": 2.07, + "learning_rate": 1.7367202940185583e-05, + "loss": 0.2569, + "step": 4954 + }, + { + "epoch": 2.07, + "learning_rate": 1.7366056402876204e-05, + "loss": 0.2391, + "step": 4955 + }, + { + "epoch": 2.07, + "learning_rate": 1.7364909653836074e-05, + "loss": 0.2742, + "step": 4956 + }, + { + "epoch": 2.08, + "learning_rate": 1.7363762693098154e-05, + "loss": 0.2206, + "step": 4957 + }, + { + "epoch": 2.08, + "learning_rate": 1.7362615520695405e-05, + "loss": 0.2756, + "step": 4958 + }, + { + "epoch": 2.08, + "learning_rate": 1.736146813666081e-05, + "loss": 0.2251, + "step": 4959 + }, + { + "epoch": 2.08, + "learning_rate": 1.7360320541027342e-05, + "loss": 0.2713, + "step": 4960 + }, + { + "epoch": 2.08, + "learning_rate": 1.7359172733827995e-05, + "loss": 0.2465, + "step": 4961 + }, + { + "epoch": 2.08, + "learning_rate": 1.735802471509576e-05, + "loss": 0.2497, + "step": 4962 + }, + { + "epoch": 2.08, + "learning_rate": 1.735687648486363e-05, + "loss": 0.2337, + "step": 4963 + }, + { + "epoch": 2.08, + "learning_rate": 1.7355728043164613e-05, + "loss": 0.2509, + "step": 4964 + }, + { + "epoch": 2.08, + "learning_rate": 1.7354579390031724e-05, + "loss": 0.2376, + "step": 4965 + }, + { + "epoch": 2.08, + "learning_rate": 1.7353430525497974e-05, + "loss": 0.2525, + "step": 4966 + }, + { + "epoch": 2.08, + "learning_rate": 1.7352281449596392e-05, + "loss": 0.2847, + "step": 4967 + }, + { + "epoch": 2.08, + "learning_rate": 1.7351132162360002e-05, + "loss": 0.255, + "step": 4968 + }, + { + "epoch": 2.08, + "learning_rate": 1.734998266382184e-05, + "loss": 0.253, + "step": 4969 + }, + { + "epoch": 2.08, + "learning_rate": 1.734883295401495e-05, + "loss": 0.2528, + "step": 4970 + }, + { + "epoch": 2.08, + "learning_rate": 1.7347683032972377e-05, + "loss": 0.2893, + "step": 4971 + }, + { + "epoch": 2.08, + "learning_rate": 1.7346532900727177e-05, + "loss": 0.2915, + "step": 4972 + }, + { + "epoch": 2.08, + "learning_rate": 1.7345382557312408e-05, + "loss": 0.2729, + "step": 4973 + }, + { + "epoch": 2.08, + "learning_rate": 1.7344232002761138e-05, + "loss": 0.2677, + "step": 4974 + }, + { + "epoch": 2.08, + "learning_rate": 1.734308123710643e-05, + "loss": 0.2816, + "step": 4975 + }, + { + "epoch": 2.08, + "learning_rate": 1.734193026038137e-05, + "loss": 0.2447, + "step": 4976 + }, + { + "epoch": 2.08, + "learning_rate": 1.7340779072619043e-05, + "loss": 0.2602, + "step": 4977 + }, + { + "epoch": 2.08, + "learning_rate": 1.7339627673852534e-05, + "loss": 0.2614, + "step": 4978 + }, + { + "epoch": 2.08, + "learning_rate": 1.733847606411494e-05, + "loss": 0.2689, + "step": 4979 + }, + { + "epoch": 2.08, + "learning_rate": 1.7337324243439363e-05, + "loss": 0.2257, + "step": 4980 + }, + { + "epoch": 2.09, + "learning_rate": 1.7336172211858914e-05, + "loss": 0.2506, + "step": 4981 + }, + { + "epoch": 2.09, + "learning_rate": 1.7335019969406702e-05, + "loss": 0.2852, + "step": 4982 + }, + { + "epoch": 2.09, + "learning_rate": 1.7333867516115853e-05, + "loss": 0.226, + "step": 4983 + }, + { + "epoch": 2.09, + "learning_rate": 1.7332714852019487e-05, + "loss": 0.2242, + "step": 4984 + }, + { + "epoch": 2.09, + "learning_rate": 1.733156197715074e-05, + "loss": 0.2465, + "step": 4985 + }, + { + "epoch": 2.09, + "learning_rate": 1.733040889154275e-05, + "loss": 0.2993, + "step": 4986 + }, + { + "epoch": 2.09, + "learning_rate": 1.7329255595228664e-05, + "loss": 0.2579, + "step": 4987 + }, + { + "epoch": 2.09, + "learning_rate": 1.732810208824163e-05, + "loss": 0.2166, + "step": 4988 + }, + { + "epoch": 2.09, + "learning_rate": 1.73269483706148e-05, + "loss": 0.2538, + "step": 4989 + }, + { + "epoch": 2.09, + "learning_rate": 1.7325794442381344e-05, + "loss": 0.2515, + "step": 4990 + }, + { + "epoch": 2.09, + "learning_rate": 1.7324640303574426e-05, + "loss": 0.2579, + "step": 4991 + }, + { + "epoch": 2.09, + "learning_rate": 1.7323485954227223e-05, + "loss": 0.2537, + "step": 4992 + }, + { + "epoch": 2.09, + "learning_rate": 1.7322331394372916e-05, + "loss": 0.2616, + "step": 4993 + }, + { + "epoch": 2.09, + "learning_rate": 1.732117662404469e-05, + "loss": 0.2548, + "step": 4994 + }, + { + "epoch": 2.09, + "learning_rate": 1.7320021643275734e-05, + "loss": 0.2309, + "step": 4995 + }, + { + "epoch": 2.09, + "learning_rate": 1.731886645209926e-05, + "loss": 0.2589, + "step": 4996 + }, + { + "epoch": 2.09, + "learning_rate": 1.731771105054846e-05, + "loss": 0.2616, + "step": 4997 + }, + { + "epoch": 2.09, + "learning_rate": 1.7316555438656547e-05, + "loss": 0.261, + "step": 4998 + }, + { + "epoch": 2.09, + "learning_rate": 1.7315399616456746e-05, + "loss": 0.2548, + "step": 4999 + }, + { + "epoch": 2.09, + "learning_rate": 1.731424358398227e-05, + "loss": 0.2641, + "step": 5000 + }, + { + "epoch": 2.09, + "learning_rate": 1.7313087341266355e-05, + "loss": 0.232, + "step": 5001 + }, + { + "epoch": 2.09, + "learning_rate": 1.7311930888342234e-05, + "loss": 0.2712, + "step": 5002 + }, + { + "epoch": 2.09, + "learning_rate": 1.7310774225243146e-05, + "loss": 0.2451, + "step": 5003 + }, + { + "epoch": 2.1, + "learning_rate": 1.7309617352002343e-05, + "loss": 0.2562, + "step": 5004 + }, + { + "epoch": 2.1, + "learning_rate": 1.7308460268653077e-05, + "loss": 0.2448, + "step": 5005 + }, + { + "epoch": 2.1, + "learning_rate": 1.73073029752286e-05, + "loss": 0.2452, + "step": 5006 + }, + { + "epoch": 2.1, + "learning_rate": 1.730614547176219e-05, + "loss": 0.2327, + "step": 5007 + }, + { + "epoch": 2.1, + "learning_rate": 1.730498775828711e-05, + "loss": 0.2429, + "step": 5008 + }, + { + "epoch": 2.1, + "learning_rate": 1.7303829834836636e-05, + "loss": 0.2622, + "step": 5009 + }, + { + "epoch": 2.1, + "learning_rate": 1.7302671701444057e-05, + "loss": 0.2703, + "step": 5010 + }, + { + "epoch": 2.1, + "learning_rate": 1.730151335814266e-05, + "loss": 0.2536, + "step": 5011 + }, + { + "epoch": 2.1, + "learning_rate": 1.730035480496574e-05, + "loss": 0.2419, + "step": 5012 + }, + { + "epoch": 2.1, + "learning_rate": 1.72991960419466e-05, + "loss": 0.2594, + "step": 5013 + }, + { + "epoch": 2.1, + "learning_rate": 1.7298037069118546e-05, + "loss": 0.2761, + "step": 5014 + }, + { + "epoch": 2.1, + "learning_rate": 1.7296877886514892e-05, + "loss": 0.2803, + "step": 5015 + }, + { + "epoch": 2.1, + "learning_rate": 1.7295718494168963e-05, + "loss": 0.2587, + "step": 5016 + }, + { + "epoch": 2.1, + "learning_rate": 1.7294558892114077e-05, + "loss": 0.2206, + "step": 5017 + }, + { + "epoch": 2.1, + "learning_rate": 1.729339908038357e-05, + "loss": 0.2495, + "step": 5018 + }, + { + "epoch": 2.1, + "learning_rate": 1.7292239059010777e-05, + "loss": 0.2289, + "step": 5019 + }, + { + "epoch": 2.1, + "learning_rate": 1.7291078828029042e-05, + "loss": 0.2807, + "step": 5020 + }, + { + "epoch": 2.1, + "learning_rate": 1.728991838747172e-05, + "loss": 0.242, + "step": 5021 + }, + { + "epoch": 2.1, + "learning_rate": 1.7288757737372157e-05, + "loss": 0.2491, + "step": 5022 + }, + { + "epoch": 2.1, + "learning_rate": 1.7287596877763725e-05, + "loss": 0.2257, + "step": 5023 + }, + { + "epoch": 2.1, + "learning_rate": 1.7286435808679787e-05, + "loss": 0.2492, + "step": 5024 + }, + { + "epoch": 2.1, + "learning_rate": 1.7285274530153718e-05, + "loss": 0.2578, + "step": 5025 + }, + { + "epoch": 2.1, + "learning_rate": 1.7284113042218895e-05, + "loss": 0.26, + "step": 5026 + }, + { + "epoch": 2.1, + "learning_rate": 1.728295134490871e-05, + "loss": 0.2653, + "step": 5027 + }, + { + "epoch": 2.11, + "learning_rate": 1.728178943825655e-05, + "loss": 0.279, + "step": 5028 + }, + { + "epoch": 2.11, + "learning_rate": 1.7280627322295812e-05, + "loss": 0.2164, + "step": 5029 + }, + { + "epoch": 2.11, + "learning_rate": 1.7279464997059905e-05, + "loss": 0.2582, + "step": 5030 + }, + { + "epoch": 2.11, + "learning_rate": 1.727830246258224e-05, + "loss": 0.2465, + "step": 5031 + }, + { + "epoch": 2.11, + "learning_rate": 1.727713971889622e-05, + "loss": 0.2576, + "step": 5032 + }, + { + "epoch": 2.11, + "learning_rate": 1.7275976766035282e-05, + "loss": 0.2423, + "step": 5033 + }, + { + "epoch": 2.11, + "learning_rate": 1.7274813604032847e-05, + "loss": 0.2154, + "step": 5034 + }, + { + "epoch": 2.11, + "learning_rate": 1.727365023292235e-05, + "loss": 0.2747, + "step": 5035 + }, + { + "epoch": 2.11, + "learning_rate": 1.7272486652737236e-05, + "loss": 0.2407, + "step": 5036 + }, + { + "epoch": 2.11, + "learning_rate": 1.7271322863510942e-05, + "loss": 0.2562, + "step": 5037 + }, + { + "epoch": 2.11, + "learning_rate": 1.7270158865276927e-05, + "loss": 0.2562, + "step": 5038 + }, + { + "epoch": 2.11, + "learning_rate": 1.7268994658068645e-05, + "loss": 0.2782, + "step": 5039 + }, + { + "epoch": 2.11, + "learning_rate": 1.7267830241919562e-05, + "loss": 0.2556, + "step": 5040 + }, + { + "epoch": 2.11, + "learning_rate": 1.726666561686315e-05, + "loss": 0.2463, + "step": 5041 + }, + { + "epoch": 2.11, + "learning_rate": 1.7265500782932885e-05, + "loss": 0.2729, + "step": 5042 + }, + { + "epoch": 2.11, + "learning_rate": 1.7264335740162244e-05, + "loss": 0.223, + "step": 5043 + }, + { + "epoch": 2.11, + "learning_rate": 1.7263170488584717e-05, + "loss": 0.2527, + "step": 5044 + }, + { + "epoch": 2.11, + "learning_rate": 1.7262005028233802e-05, + "loss": 0.2513, + "step": 5045 + }, + { + "epoch": 2.11, + "learning_rate": 1.7260839359142995e-05, + "loss": 0.2431, + "step": 5046 + }, + { + "epoch": 2.11, + "learning_rate": 1.7259673481345808e-05, + "loss": 0.2269, + "step": 5047 + }, + { + "epoch": 2.11, + "learning_rate": 1.7258507394875742e-05, + "loss": 0.249, + "step": 5048 + }, + { + "epoch": 2.11, + "learning_rate": 1.725734109976633e-05, + "loss": 0.2325, + "step": 5049 + }, + { + "epoch": 2.11, + "learning_rate": 1.725617459605108e-05, + "loss": 0.2591, + "step": 5050 + }, + { + "epoch": 2.11, + "learning_rate": 1.7255007883763537e-05, + "loss": 0.2405, + "step": 5051 + }, + { + "epoch": 2.12, + "learning_rate": 1.7253840962937226e-05, + "loss": 0.2675, + "step": 5052 + }, + { + "epoch": 2.12, + "learning_rate": 1.7252673833605697e-05, + "loss": 0.2355, + "step": 5053 + }, + { + "epoch": 2.12, + "learning_rate": 1.725150649580249e-05, + "loss": 0.2465, + "step": 5054 + }, + { + "epoch": 2.12, + "learning_rate": 1.7250338949561167e-05, + "loss": 0.2481, + "step": 5055 + }, + { + "epoch": 2.12, + "learning_rate": 1.7249171194915283e-05, + "loss": 0.2396, + "step": 5056 + }, + { + "epoch": 2.12, + "learning_rate": 1.724800323189841e-05, + "loss": 0.2557, + "step": 5057 + }, + { + "epoch": 2.12, + "learning_rate": 1.724683506054411e-05, + "loss": 0.2235, + "step": 5058 + }, + { + "epoch": 2.12, + "learning_rate": 1.7245666680885975e-05, + "loss": 0.2663, + "step": 5059 + }, + { + "epoch": 2.12, + "learning_rate": 1.724449809295757e-05, + "loss": 0.2585, + "step": 5060 + }, + { + "epoch": 2.12, + "learning_rate": 1.7243329296792502e-05, + "loss": 0.2593, + "step": 5061 + }, + { + "epoch": 2.12, + "learning_rate": 1.7242160292424362e-05, + "loss": 0.2311, + "step": 5062 + }, + { + "epoch": 2.12, + "learning_rate": 1.724099107988675e-05, + "loss": 0.2416, + "step": 5063 + }, + { + "epoch": 2.12, + "learning_rate": 1.7239821659213272e-05, + "loss": 0.2705, + "step": 5064 + }, + { + "epoch": 2.12, + "learning_rate": 1.723865203043755e-05, + "loss": 0.2453, + "step": 5065 + }, + { + "epoch": 2.12, + "learning_rate": 1.7237482193593195e-05, + "loss": 0.2731, + "step": 5066 + }, + { + "epoch": 2.12, + "learning_rate": 1.7236312148713838e-05, + "loss": 0.2653, + "step": 5067 + }, + { + "epoch": 2.12, + "learning_rate": 1.723514189583311e-05, + "loss": 0.2623, + "step": 5068 + }, + { + "epoch": 2.12, + "learning_rate": 1.7233971434984648e-05, + "loss": 0.2766, + "step": 5069 + }, + { + "epoch": 2.12, + "learning_rate": 1.72328007662021e-05, + "loss": 0.2531, + "step": 5070 + }, + { + "epoch": 2.12, + "learning_rate": 1.7231629889519106e-05, + "loss": 0.256, + "step": 5071 + }, + { + "epoch": 2.12, + "learning_rate": 1.723045880496933e-05, + "loss": 0.2576, + "step": 5072 + }, + { + "epoch": 2.12, + "learning_rate": 1.7229287512586432e-05, + "loss": 0.2503, + "step": 5073 + }, + { + "epoch": 2.12, + "learning_rate": 1.7228116012404085e-05, + "loss": 0.2751, + "step": 5074 + }, + { + "epoch": 2.12, + "learning_rate": 1.722694430445595e-05, + "loss": 0.258, + "step": 5075 + }, + { + "epoch": 2.13, + "learning_rate": 1.7225772388775716e-05, + "loss": 0.2496, + "step": 5076 + }, + { + "epoch": 2.13, + "learning_rate": 1.722460026539707e-05, + "loss": 0.2667, + "step": 5077 + }, + { + "epoch": 2.13, + "learning_rate": 1.7223427934353695e-05, + "loss": 0.2329, + "step": 5078 + }, + { + "epoch": 2.13, + "learning_rate": 1.7222255395679298e-05, + "loss": 0.2524, + "step": 5079 + }, + { + "epoch": 2.13, + "learning_rate": 1.7221082649407578e-05, + "loss": 0.25, + "step": 5080 + }, + { + "epoch": 2.13, + "learning_rate": 1.7219909695572244e-05, + "loss": 0.2948, + "step": 5081 + }, + { + "epoch": 2.13, + "learning_rate": 1.7218736534207014e-05, + "loss": 0.2579, + "step": 5082 + }, + { + "epoch": 2.13, + "learning_rate": 1.7217563165345607e-05, + "loss": 0.2617, + "step": 5083 + }, + { + "epoch": 2.13, + "learning_rate": 1.721638958902175e-05, + "loss": 0.2608, + "step": 5084 + }, + { + "epoch": 2.13, + "learning_rate": 1.721521580526918e-05, + "loss": 0.2515, + "step": 5085 + }, + { + "epoch": 2.13, + "learning_rate": 1.7214041814121635e-05, + "loss": 0.2472, + "step": 5086 + }, + { + "epoch": 2.13, + "learning_rate": 1.721286761561286e-05, + "loss": 0.2589, + "step": 5087 + }, + { + "epoch": 2.13, + "learning_rate": 1.7211693209776604e-05, + "loss": 0.2718, + "step": 5088 + }, + { + "epoch": 2.13, + "learning_rate": 1.7210518596646627e-05, + "loss": 0.2498, + "step": 5089 + }, + { + "epoch": 2.13, + "learning_rate": 1.720934377625669e-05, + "loss": 0.2506, + "step": 5090 + }, + { + "epoch": 2.13, + "learning_rate": 1.7208168748640565e-05, + "loss": 0.2388, + "step": 5091 + }, + { + "epoch": 2.13, + "learning_rate": 1.7206993513832026e-05, + "loss": 0.2686, + "step": 5092 + }, + { + "epoch": 2.13, + "learning_rate": 1.7205818071864853e-05, + "loss": 0.2474, + "step": 5093 + }, + { + "epoch": 2.13, + "learning_rate": 1.7204642422772836e-05, + "loss": 0.2179, + "step": 5094 + }, + { + "epoch": 2.13, + "learning_rate": 1.7203466566589763e-05, + "loss": 0.257, + "step": 5095 + }, + { + "epoch": 2.13, + "learning_rate": 1.7202290503349436e-05, + "loss": 0.2763, + "step": 5096 + }, + { + "epoch": 2.13, + "learning_rate": 1.7201114233085664e-05, + "loss": 0.2821, + "step": 5097 + }, + { + "epoch": 2.13, + "learning_rate": 1.719993775583225e-05, + "loss": 0.2528, + "step": 5098 + }, + { + "epoch": 2.13, + "learning_rate": 1.7198761071623014e-05, + "loss": 0.2878, + "step": 5099 + }, + { + "epoch": 2.14, + "learning_rate": 1.7197584180491776e-05, + "loss": 0.2428, + "step": 5100 + }, + { + "epoch": 2.14, + "learning_rate": 1.7196407082472375e-05, + "loss": 0.2827, + "step": 5101 + }, + { + "epoch": 2.14, + "learning_rate": 1.7195229777598635e-05, + "loss": 0.2712, + "step": 5102 + }, + { + "epoch": 2.14, + "learning_rate": 1.71940522659044e-05, + "loss": 0.271, + "step": 5103 + }, + { + "epoch": 2.14, + "learning_rate": 1.7192874547423514e-05, + "loss": 0.223, + "step": 5104 + }, + { + "epoch": 2.14, + "learning_rate": 1.7191696622189836e-05, + "loss": 0.2587, + "step": 5105 + }, + { + "epoch": 2.14, + "learning_rate": 1.719051849023722e-05, + "loss": 0.2812, + "step": 5106 + }, + { + "epoch": 2.14, + "learning_rate": 1.718934015159953e-05, + "loss": 0.258, + "step": 5107 + }, + { + "epoch": 2.14, + "learning_rate": 1.7188161606310635e-05, + "loss": 0.2294, + "step": 5108 + }, + { + "epoch": 2.14, + "learning_rate": 1.7186982854404415e-05, + "loss": 0.256, + "step": 5109 + }, + { + "epoch": 2.14, + "learning_rate": 1.718580389591475e-05, + "loss": 0.252, + "step": 5110 + }, + { + "epoch": 2.14, + "learning_rate": 1.7184624730875528e-05, + "loss": 0.2629, + "step": 5111 + }, + { + "epoch": 2.14, + "learning_rate": 1.7183445359320645e-05, + "loss": 0.2399, + "step": 5112 + }, + { + "epoch": 2.14, + "learning_rate": 1.7182265781284e-05, + "loss": 0.257, + "step": 5113 + }, + { + "epoch": 2.14, + "learning_rate": 1.71810859967995e-05, + "loss": 0.2499, + "step": 5114 + }, + { + "epoch": 2.14, + "learning_rate": 1.717990600590105e-05, + "loss": 0.24, + "step": 5115 + }, + { + "epoch": 2.14, + "learning_rate": 1.7178725808622575e-05, + "loss": 0.2659, + "step": 5116 + }, + { + "epoch": 2.14, + "learning_rate": 1.7177545404997998e-05, + "loss": 0.2335, + "step": 5117 + }, + { + "epoch": 2.14, + "learning_rate": 1.717636479506125e-05, + "loss": 0.2443, + "step": 5118 + }, + { + "epoch": 2.14, + "learning_rate": 1.717518397884626e-05, + "loss": 0.2775, + "step": 5119 + }, + { + "epoch": 2.14, + "learning_rate": 1.717400295638698e-05, + "loss": 0.235, + "step": 5120 + }, + { + "epoch": 2.14, + "learning_rate": 1.7172821727717345e-05, + "loss": 0.2412, + "step": 5121 + }, + { + "epoch": 2.14, + "learning_rate": 1.7171640292871314e-05, + "loss": 0.2423, + "step": 5122 + }, + { + "epoch": 2.14, + "learning_rate": 1.7170458651882847e-05, + "loss": 0.2764, + "step": 5123 + }, + { + "epoch": 2.15, + "learning_rate": 1.716927680478591e-05, + "loss": 0.2641, + "step": 5124 + }, + { + "epoch": 2.15, + "learning_rate": 1.716809475161448e-05, + "loss": 0.2793, + "step": 5125 + }, + { + "epoch": 2.15, + "learning_rate": 1.716691249240252e-05, + "loss": 0.2427, + "step": 5126 + }, + { + "epoch": 2.15, + "learning_rate": 1.716573002718402e-05, + "loss": 0.2391, + "step": 5127 + }, + { + "epoch": 2.15, + "learning_rate": 1.7164547355992972e-05, + "loss": 0.2451, + "step": 5128 + }, + { + "epoch": 2.15, + "learning_rate": 1.7163364478863365e-05, + "loss": 0.2525, + "step": 5129 + }, + { + "epoch": 2.15, + "learning_rate": 1.7162181395829204e-05, + "loss": 0.2202, + "step": 5130 + }, + { + "epoch": 2.15, + "learning_rate": 1.7160998106924494e-05, + "loss": 0.2548, + "step": 5131 + }, + { + "epoch": 2.15, + "learning_rate": 1.715981461218325e-05, + "loss": 0.2345, + "step": 5132 + }, + { + "epoch": 2.15, + "learning_rate": 1.7158630911639488e-05, + "loss": 0.2593, + "step": 5133 + }, + { + "epoch": 2.15, + "learning_rate": 1.715744700532723e-05, + "loss": 0.2576, + "step": 5134 + }, + { + "epoch": 2.15, + "learning_rate": 1.7156262893280512e-05, + "loss": 0.2574, + "step": 5135 + }, + { + "epoch": 2.15, + "learning_rate": 1.7155078575533366e-05, + "loss": 0.2137, + "step": 5136 + }, + { + "epoch": 2.15, + "learning_rate": 1.7153894052119837e-05, + "loss": 0.2537, + "step": 5137 + }, + { + "epoch": 2.15, + "learning_rate": 1.7152709323073972e-05, + "loss": 0.2346, + "step": 5138 + }, + { + "epoch": 2.15, + "learning_rate": 1.7151524388429823e-05, + "loss": 0.2037, + "step": 5139 + }, + { + "epoch": 2.15, + "learning_rate": 1.7150339248221454e-05, + "loss": 0.2163, + "step": 5140 + }, + { + "epoch": 2.15, + "learning_rate": 1.7149153902482926e-05, + "loss": 0.2761, + "step": 5141 + }, + { + "epoch": 2.15, + "learning_rate": 1.7147968351248315e-05, + "loss": 0.2492, + "step": 5142 + }, + { + "epoch": 2.15, + "learning_rate": 1.7146782594551698e-05, + "loss": 0.2327, + "step": 5143 + }, + { + "epoch": 2.15, + "learning_rate": 1.714559663242715e-05, + "loss": 0.2507, + "step": 5144 + }, + { + "epoch": 2.15, + "learning_rate": 1.7144410464908774e-05, + "loss": 0.2678, + "step": 5145 + }, + { + "epoch": 2.15, + "learning_rate": 1.714322409203066e-05, + "loss": 0.2523, + "step": 5146 + }, + { + "epoch": 2.15, + "learning_rate": 1.7142037513826904e-05, + "loss": 0.2563, + "step": 5147 + }, + { + "epoch": 2.16, + "learning_rate": 1.714085073033162e-05, + "loss": 0.2264, + "step": 5148 + }, + { + "epoch": 2.16, + "learning_rate": 1.7139663741578917e-05, + "loss": 0.2479, + "step": 5149 + }, + { + "epoch": 2.16, + "learning_rate": 1.7138476547602917e-05, + "loss": 0.2299, + "step": 5150 + }, + { + "epoch": 2.16, + "learning_rate": 1.7137289148437743e-05, + "loss": 0.2625, + "step": 5151 + }, + { + "epoch": 2.16, + "learning_rate": 1.7136101544117526e-05, + "loss": 0.2654, + "step": 5152 + }, + { + "epoch": 2.16, + "learning_rate": 1.7134913734676398e-05, + "loss": 0.2451, + "step": 5153 + }, + { + "epoch": 2.16, + "learning_rate": 1.7133725720148513e-05, + "loss": 0.2287, + "step": 5154 + }, + { + "epoch": 2.16, + "learning_rate": 1.713253750056801e-05, + "loss": 0.2551, + "step": 5155 + }, + { + "epoch": 2.16, + "learning_rate": 1.7131349075969044e-05, + "loss": 0.2604, + "step": 5156 + }, + { + "epoch": 2.16, + "learning_rate": 1.713016044638578e-05, + "loss": 0.2559, + "step": 5157 + }, + { + "epoch": 2.16, + "learning_rate": 1.7128971611852378e-05, + "loss": 0.2613, + "step": 5158 + }, + { + "epoch": 2.16, + "learning_rate": 1.7127782572403015e-05, + "loss": 0.2773, + "step": 5159 + }, + { + "epoch": 2.16, + "learning_rate": 1.7126593328071867e-05, + "loss": 0.2502, + "step": 5160 + }, + { + "epoch": 2.16, + "learning_rate": 1.7125403878893118e-05, + "loss": 0.2821, + "step": 5161 + }, + { + "epoch": 2.16, + "learning_rate": 1.7124214224900955e-05, + "loss": 0.2593, + "step": 5162 + }, + { + "epoch": 2.16, + "learning_rate": 1.712302436612958e-05, + "loss": 0.2377, + "step": 5163 + }, + { + "epoch": 2.16, + "learning_rate": 1.712183430261319e-05, + "loss": 0.2523, + "step": 5164 + }, + { + "epoch": 2.16, + "learning_rate": 1.712064403438599e-05, + "loss": 0.2733, + "step": 5165 + }, + { + "epoch": 2.16, + "learning_rate": 1.71194535614822e-05, + "loss": 0.2376, + "step": 5166 + }, + { + "epoch": 2.16, + "learning_rate": 1.711826288393603e-05, + "loss": 0.2457, + "step": 5167 + }, + { + "epoch": 2.16, + "learning_rate": 1.7117072001781713e-05, + "loss": 0.2349, + "step": 5168 + }, + { + "epoch": 2.16, + "learning_rate": 1.7115880915053476e-05, + "loss": 0.2544, + "step": 5169 + }, + { + "epoch": 2.16, + "learning_rate": 1.7114689623785556e-05, + "loss": 0.2306, + "step": 5170 + }, + { + "epoch": 2.16, + "learning_rate": 1.7113498128012197e-05, + "loss": 0.2666, + "step": 5171 + }, + { + "epoch": 2.17, + "learning_rate": 1.711230642776765e-05, + "loss": 0.2648, + "step": 5172 + }, + { + "epoch": 2.17, + "learning_rate": 1.711111452308616e-05, + "loss": 0.2542, + "step": 5173 + }, + { + "epoch": 2.17, + "learning_rate": 1.7109922414001997e-05, + "loss": 0.2421, + "step": 5174 + }, + { + "epoch": 2.17, + "learning_rate": 1.710873010054942e-05, + "loss": 0.2632, + "step": 5175 + }, + { + "epoch": 2.17, + "learning_rate": 1.710753758276271e-05, + "loss": 0.2471, + "step": 5176 + }, + { + "epoch": 2.17, + "learning_rate": 1.7106344860676137e-05, + "loss": 0.2528, + "step": 5177 + }, + { + "epoch": 2.17, + "learning_rate": 1.7105151934323988e-05, + "loss": 0.2461, + "step": 5178 + }, + { + "epoch": 2.17, + "learning_rate": 1.7103958803740547e-05, + "loss": 0.2155, + "step": 5179 + }, + { + "epoch": 2.17, + "learning_rate": 1.7102765468960116e-05, + "loss": 0.2406, + "step": 5180 + }, + { + "epoch": 2.17, + "learning_rate": 1.7101571930017e-05, + "loss": 0.2621, + "step": 5181 + }, + { + "epoch": 2.17, + "learning_rate": 1.7100378186945494e-05, + "loss": 0.2702, + "step": 5182 + }, + { + "epoch": 2.17, + "learning_rate": 1.709918423977992e-05, + "loss": 0.256, + "step": 5183 + }, + { + "epoch": 2.17, + "learning_rate": 1.7097990088554595e-05, + "loss": 0.2888, + "step": 5184 + }, + { + "epoch": 2.17, + "learning_rate": 1.7096795733303844e-05, + "loss": 0.2608, + "step": 5185 + }, + { + "epoch": 2.17, + "learning_rate": 1.7095601174061997e-05, + "loss": 0.2615, + "step": 5186 + }, + { + "epoch": 2.17, + "learning_rate": 1.709440641086339e-05, + "loss": 0.2309, + "step": 5187 + }, + { + "epoch": 2.17, + "learning_rate": 1.7093211443742368e-05, + "loss": 0.2574, + "step": 5188 + }, + { + "epoch": 2.17, + "learning_rate": 1.7092016272733274e-05, + "loss": 0.2316, + "step": 5189 + }, + { + "epoch": 2.17, + "learning_rate": 1.709082089787047e-05, + "loss": 0.2482, + "step": 5190 + }, + { + "epoch": 2.17, + "learning_rate": 1.7089625319188308e-05, + "loss": 0.264, + "step": 5191 + }, + { + "epoch": 2.17, + "learning_rate": 1.708842953672116e-05, + "loss": 0.2641, + "step": 5192 + }, + { + "epoch": 2.17, + "learning_rate": 1.7087233550503394e-05, + "loss": 0.2737, + "step": 5193 + }, + { + "epoch": 2.17, + "learning_rate": 1.7086037360569386e-05, + "loss": 0.2495, + "step": 5194 + }, + { + "epoch": 2.18, + "learning_rate": 1.7084840966953526e-05, + "loss": 0.2728, + "step": 5195 + }, + { + "epoch": 2.18, + "learning_rate": 1.7083644369690197e-05, + "loss": 0.2599, + "step": 5196 + }, + { + "epoch": 2.18, + "learning_rate": 1.7082447568813798e-05, + "loss": 0.2787, + "step": 5197 + }, + { + "epoch": 2.18, + "learning_rate": 1.708125056435873e-05, + "loss": 0.2641, + "step": 5198 + }, + { + "epoch": 2.18, + "learning_rate": 1.7080053356359396e-05, + "loss": 0.2461, + "step": 5199 + }, + { + "epoch": 2.18, + "learning_rate": 1.707885594485021e-05, + "loss": 0.2538, + "step": 5200 + }, + { + "epoch": 2.18, + "learning_rate": 1.7077658329865593e-05, + "loss": 0.2623, + "step": 5201 + }, + { + "epoch": 2.18, + "learning_rate": 1.7076460511439966e-05, + "loss": 0.2551, + "step": 5202 + }, + { + "epoch": 2.18, + "learning_rate": 1.7075262489607766e-05, + "loss": 0.2842, + "step": 5203 + }, + { + "epoch": 2.18, + "learning_rate": 1.707406426440342e-05, + "loss": 0.2662, + "step": 5204 + }, + { + "epoch": 2.18, + "learning_rate": 1.7072865835861377e-05, + "loss": 0.2783, + "step": 5205 + }, + { + "epoch": 2.18, + "learning_rate": 1.707166720401608e-05, + "loss": 0.2565, + "step": 5206 + }, + { + "epoch": 2.18, + "learning_rate": 1.7070468368901982e-05, + "loss": 0.2397, + "step": 5207 + }, + { + "epoch": 2.18, + "learning_rate": 1.706926933055355e-05, + "loss": 0.2461, + "step": 5208 + }, + { + "epoch": 2.18, + "learning_rate": 1.706807008900524e-05, + "loss": 0.2689, + "step": 5209 + }, + { + "epoch": 2.18, + "learning_rate": 1.7066870644291527e-05, + "loss": 0.2655, + "step": 5210 + }, + { + "epoch": 2.18, + "learning_rate": 1.7065670996446892e-05, + "loss": 0.2626, + "step": 5211 + }, + { + "epoch": 2.18, + "learning_rate": 1.7064471145505813e-05, + "loss": 0.2378, + "step": 5212 + }, + { + "epoch": 2.18, + "learning_rate": 1.706327109150278e-05, + "loss": 0.2428, + "step": 5213 + }, + { + "epoch": 2.18, + "learning_rate": 1.7062070834472287e-05, + "loss": 0.2442, + "step": 5214 + }, + { + "epoch": 2.18, + "learning_rate": 1.7060870374448833e-05, + "loss": 0.2648, + "step": 5215 + }, + { + "epoch": 2.18, + "learning_rate": 1.705966971146693e-05, + "loss": 0.2259, + "step": 5216 + }, + { + "epoch": 2.18, + "learning_rate": 1.7058468845561084e-05, + "loss": 0.2766, + "step": 5217 + }, + { + "epoch": 2.18, + "learning_rate": 1.7057267776765814e-05, + "loss": 0.2512, + "step": 5218 + }, + { + "epoch": 2.19, + "learning_rate": 1.705606650511564e-05, + "loss": 0.2322, + "step": 5219 + }, + { + "epoch": 2.19, + "learning_rate": 1.7054865030645102e-05, + "loss": 0.2896, + "step": 5220 + }, + { + "epoch": 2.19, + "learning_rate": 1.7053663353388727e-05, + "loss": 0.2412, + "step": 5221 + }, + { + "epoch": 2.19, + "learning_rate": 1.7052461473381056e-05, + "loss": 0.2456, + "step": 5222 + }, + { + "epoch": 2.19, + "learning_rate": 1.7051259390656642e-05, + "loss": 0.2572, + "step": 5223 + }, + { + "epoch": 2.19, + "learning_rate": 1.7050057105250034e-05, + "loss": 0.2393, + "step": 5224 + }, + { + "epoch": 2.19, + "learning_rate": 1.7048854617195788e-05, + "loss": 0.2851, + "step": 5225 + }, + { + "epoch": 2.19, + "learning_rate": 1.7047651926528472e-05, + "loss": 0.2603, + "step": 5226 + }, + { + "epoch": 2.19, + "learning_rate": 1.7046449033282656e-05, + "loss": 0.2448, + "step": 5227 + }, + { + "epoch": 2.19, + "learning_rate": 1.7045245937492912e-05, + "loss": 0.2544, + "step": 5228 + }, + { + "epoch": 2.19, + "learning_rate": 1.704404263919383e-05, + "loss": 0.2433, + "step": 5229 + }, + { + "epoch": 2.19, + "learning_rate": 1.7042839138419992e-05, + "loss": 0.2589, + "step": 5230 + }, + { + "epoch": 2.19, + "learning_rate": 1.704163543520599e-05, + "loss": 0.2382, + "step": 5231 + }, + { + "epoch": 2.19, + "learning_rate": 1.7040431529586427e-05, + "loss": 0.2312, + "step": 5232 + }, + { + "epoch": 2.19, + "learning_rate": 1.703922742159591e-05, + "loss": 0.2521, + "step": 5233 + }, + { + "epoch": 2.19, + "learning_rate": 1.7038023111269045e-05, + "loss": 0.258, + "step": 5234 + }, + { + "epoch": 2.19, + "learning_rate": 1.7036818598640456e-05, + "loss": 0.2897, + "step": 5235 + }, + { + "epoch": 2.19, + "learning_rate": 1.7035613883744753e-05, + "loss": 0.2337, + "step": 5236 + }, + { + "epoch": 2.19, + "learning_rate": 1.7034408966616577e-05, + "loss": 0.2502, + "step": 5237 + }, + { + "epoch": 2.19, + "learning_rate": 1.7033203847290553e-05, + "loss": 0.2469, + "step": 5238 + }, + { + "epoch": 2.19, + "learning_rate": 1.703199852580133e-05, + "loss": 0.2517, + "step": 5239 + }, + { + "epoch": 2.19, + "learning_rate": 1.7030793002183546e-05, + "loss": 0.2594, + "step": 5240 + }, + { + "epoch": 2.19, + "learning_rate": 1.702958727647186e-05, + "loss": 0.2306, + "step": 5241 + }, + { + "epoch": 2.19, + "learning_rate": 1.7028381348700926e-05, + "loss": 0.2449, + "step": 5242 + }, + { + "epoch": 2.2, + "learning_rate": 1.70271752189054e-05, + "loss": 0.2469, + "step": 5243 + }, + { + "epoch": 2.2, + "learning_rate": 1.7025968887119962e-05, + "loss": 0.3028, + "step": 5244 + }, + { + "epoch": 2.2, + "learning_rate": 1.702476235337929e-05, + "loss": 0.2763, + "step": 5245 + }, + { + "epoch": 2.2, + "learning_rate": 1.702355561771805e-05, + "loss": 0.2196, + "step": 5246 + }, + { + "epoch": 2.2, + "learning_rate": 1.7022348680170936e-05, + "loss": 0.2704, + "step": 5247 + }, + { + "epoch": 2.2, + "learning_rate": 1.7021141540772645e-05, + "loss": 0.2302, + "step": 5248 + }, + { + "epoch": 2.2, + "learning_rate": 1.7019934199557868e-05, + "loss": 0.2421, + "step": 5249 + }, + { + "epoch": 2.2, + "learning_rate": 1.7018726656561314e-05, + "loss": 0.254, + "step": 5250 + }, + { + "epoch": 2.2, + "learning_rate": 1.7017518911817686e-05, + "loss": 0.251, + "step": 5251 + }, + { + "epoch": 2.2, + "learning_rate": 1.7016310965361708e-05, + "loss": 0.2674, + "step": 5252 + }, + { + "epoch": 2.2, + "learning_rate": 1.70151028172281e-05, + "loss": 0.268, + "step": 5253 + }, + { + "epoch": 2.2, + "learning_rate": 1.701389446745158e-05, + "loss": 0.2271, + "step": 5254 + }, + { + "epoch": 2.2, + "learning_rate": 1.7012685916066894e-05, + "loss": 0.2591, + "step": 5255 + }, + { + "epoch": 2.2, + "learning_rate": 1.7011477163108766e-05, + "loss": 0.2526, + "step": 5256 + }, + { + "epoch": 2.2, + "learning_rate": 1.7010268208611958e-05, + "loss": 0.271, + "step": 5257 + }, + { + "epoch": 2.2, + "learning_rate": 1.7009059052611206e-05, + "loss": 0.238, + "step": 5258 + }, + { + "epoch": 2.2, + "learning_rate": 1.7007849695141275e-05, + "loss": 0.2589, + "step": 5259 + }, + { + "epoch": 2.2, + "learning_rate": 1.700664013623692e-05, + "loss": 0.2659, + "step": 5260 + }, + { + "epoch": 2.2, + "learning_rate": 1.700543037593291e-05, + "loss": 0.2963, + "step": 5261 + }, + { + "epoch": 2.2, + "learning_rate": 1.7004220414264025e-05, + "loss": 0.2727, + "step": 5262 + }, + { + "epoch": 2.2, + "learning_rate": 1.7003010251265036e-05, + "loss": 0.2739, + "step": 5263 + }, + { + "epoch": 2.2, + "learning_rate": 1.7001799886970735e-05, + "loss": 0.2541, + "step": 5264 + }, + { + "epoch": 2.2, + "learning_rate": 1.7000589321415906e-05, + "loss": 0.2445, + "step": 5265 + }, + { + "epoch": 2.2, + "learning_rate": 1.699937855463535e-05, + "loss": 0.2615, + "step": 5266 + }, + { + "epoch": 2.21, + "learning_rate": 1.6998167586663868e-05, + "loss": 0.2552, + "step": 5267 + }, + { + "epoch": 2.21, + "learning_rate": 1.699695641753627e-05, + "loss": 0.2752, + "step": 5268 + }, + { + "epoch": 2.21, + "learning_rate": 1.6995745047287373e-05, + "loss": 0.2623, + "step": 5269 + }, + { + "epoch": 2.21, + "learning_rate": 1.6994533475951985e-05, + "loss": 0.2388, + "step": 5270 + }, + { + "epoch": 2.21, + "learning_rate": 1.6993321703564946e-05, + "loss": 0.2854, + "step": 5271 + }, + { + "epoch": 2.21, + "learning_rate": 1.6992109730161077e-05, + "loss": 0.264, + "step": 5272 + }, + { + "epoch": 2.21, + "learning_rate": 1.699089755577522e-05, + "loss": 0.2667, + "step": 5273 + }, + { + "epoch": 2.21, + "learning_rate": 1.6989685180442212e-05, + "loss": 0.2484, + "step": 5274 + }, + { + "epoch": 2.21, + "learning_rate": 1.6988472604196907e-05, + "loss": 0.2643, + "step": 5275 + }, + { + "epoch": 2.21, + "learning_rate": 1.698725982707416e-05, + "loss": 0.2616, + "step": 5276 + }, + { + "epoch": 2.21, + "learning_rate": 1.6986046849108834e-05, + "loss": 0.2566, + "step": 5277 + }, + { + "epoch": 2.21, + "learning_rate": 1.6984833670335783e-05, + "loss": 0.26, + "step": 5278 + }, + { + "epoch": 2.21, + "learning_rate": 1.698362029078989e-05, + "loss": 0.2434, + "step": 5279 + }, + { + "epoch": 2.21, + "learning_rate": 1.698240671050603e-05, + "loss": 0.2577, + "step": 5280 + }, + { + "epoch": 2.21, + "learning_rate": 1.6981192929519087e-05, + "loss": 0.2326, + "step": 5281 + }, + { + "epoch": 2.21, + "learning_rate": 1.6979978947863946e-05, + "loss": 0.2519, + "step": 5282 + }, + { + "epoch": 2.21, + "learning_rate": 1.69787647655755e-05, + "loss": 0.2774, + "step": 5283 + }, + { + "epoch": 2.21, + "learning_rate": 1.697755038268866e-05, + "loss": 0.2594, + "step": 5284 + }, + { + "epoch": 2.21, + "learning_rate": 1.6976335799238318e-05, + "loss": 0.2421, + "step": 5285 + }, + { + "epoch": 2.21, + "learning_rate": 1.69751210152594e-05, + "loss": 0.2378, + "step": 5286 + }, + { + "epoch": 2.21, + "learning_rate": 1.697390603078682e-05, + "loss": 0.2452, + "step": 5287 + }, + { + "epoch": 2.21, + "learning_rate": 1.6972690845855494e-05, + "loss": 0.2753, + "step": 5288 + }, + { + "epoch": 2.21, + "learning_rate": 1.697147546050036e-05, + "loss": 0.2547, + "step": 5289 + }, + { + "epoch": 2.21, + "learning_rate": 1.6970259874756353e-05, + "loss": 0.2338, + "step": 5290 + }, + { + "epoch": 2.22, + "learning_rate": 1.696904408865841e-05, + "loss": 0.2653, + "step": 5291 + }, + { + "epoch": 2.22, + "learning_rate": 1.6967828102241475e-05, + "loss": 0.2689, + "step": 5292 + }, + { + "epoch": 2.22, + "learning_rate": 1.6966611915540506e-05, + "loss": 0.2678, + "step": 5293 + }, + { + "epoch": 2.22, + "learning_rate": 1.696539552859046e-05, + "loss": 0.2517, + "step": 5294 + }, + { + "epoch": 2.22, + "learning_rate": 1.69641789414263e-05, + "loss": 0.2531, + "step": 5295 + }, + { + "epoch": 2.22, + "learning_rate": 1.6962962154083e-05, + "loss": 0.263, + "step": 5296 + }, + { + "epoch": 2.22, + "learning_rate": 1.6961745166595528e-05, + "loss": 0.26, + "step": 5297 + }, + { + "epoch": 2.22, + "learning_rate": 1.696052797899887e-05, + "loss": 0.2513, + "step": 5298 + }, + { + "epoch": 2.22, + "learning_rate": 1.695931059132801e-05, + "loss": 0.2297, + "step": 5299 + }, + { + "epoch": 2.22, + "learning_rate": 1.6958093003617942e-05, + "loss": 0.2827, + "step": 5300 + }, + { + "epoch": 2.22, + "learning_rate": 1.695687521590367e-05, + "loss": 0.2749, + "step": 5301 + }, + { + "epoch": 2.22, + "learning_rate": 1.695565722822019e-05, + "loss": 0.2556, + "step": 5302 + }, + { + "epoch": 2.22, + "learning_rate": 1.6954439040602516e-05, + "loss": 0.2822, + "step": 5303 + }, + { + "epoch": 2.22, + "learning_rate": 1.6953220653085662e-05, + "loss": 0.2487, + "step": 5304 + }, + { + "epoch": 2.22, + "learning_rate": 1.6952002065704652e-05, + "loss": 0.2533, + "step": 5305 + }, + { + "epoch": 2.22, + "learning_rate": 1.6950783278494513e-05, + "loss": 0.241, + "step": 5306 + }, + { + "epoch": 2.22, + "learning_rate": 1.6949564291490274e-05, + "loss": 0.2709, + "step": 5307 + }, + { + "epoch": 2.22, + "learning_rate": 1.694834510472698e-05, + "loss": 0.2417, + "step": 5308 + }, + { + "epoch": 2.22, + "learning_rate": 1.6947125718239664e-05, + "loss": 0.2468, + "step": 5309 + }, + { + "epoch": 2.22, + "learning_rate": 1.694590613206339e-05, + "loss": 0.2712, + "step": 5310 + }, + { + "epoch": 2.22, + "learning_rate": 1.6944686346233207e-05, + "loss": 0.2649, + "step": 5311 + }, + { + "epoch": 2.22, + "learning_rate": 1.694346636078418e-05, + "loss": 0.25, + "step": 5312 + }, + { + "epoch": 2.22, + "learning_rate": 1.6942246175751372e-05, + "loss": 0.2659, + "step": 5313 + }, + { + "epoch": 2.22, + "learning_rate": 1.6941025791169855e-05, + "loss": 0.2657, + "step": 5314 + }, + { + "epoch": 2.23, + "learning_rate": 1.6939805207074714e-05, + "loss": 0.253, + "step": 5315 + }, + { + "epoch": 2.23, + "learning_rate": 1.693858442350103e-05, + "loss": 0.2573, + "step": 5316 + }, + { + "epoch": 2.23, + "learning_rate": 1.6937363440483892e-05, + "loss": 0.2307, + "step": 5317 + }, + { + "epoch": 2.23, + "learning_rate": 1.6936142258058403e-05, + "loss": 0.2418, + "step": 5318 + }, + { + "epoch": 2.23, + "learning_rate": 1.6934920876259657e-05, + "loss": 0.2507, + "step": 5319 + }, + { + "epoch": 2.23, + "learning_rate": 1.6933699295122766e-05, + "loss": 0.2765, + "step": 5320 + }, + { + "epoch": 2.23, + "learning_rate": 1.6932477514682838e-05, + "loss": 0.2427, + "step": 5321 + }, + { + "epoch": 2.23, + "learning_rate": 1.6931255534974997e-05, + "loss": 0.2684, + "step": 5322 + }, + { + "epoch": 2.23, + "learning_rate": 1.6930033356034368e-05, + "loss": 0.2533, + "step": 5323 + }, + { + "epoch": 2.23, + "learning_rate": 1.692881097789608e-05, + "loss": 0.2579, + "step": 5324 + }, + { + "epoch": 2.23, + "learning_rate": 1.6927588400595265e-05, + "loss": 0.2453, + "step": 5325 + }, + { + "epoch": 2.23, + "learning_rate": 1.6926365624167075e-05, + "loss": 0.2447, + "step": 5326 + }, + { + "epoch": 2.23, + "learning_rate": 1.6925142648646647e-05, + "loss": 0.2478, + "step": 5327 + }, + { + "epoch": 2.23, + "learning_rate": 1.692391947406914e-05, + "loss": 0.2556, + "step": 5328 + }, + { + "epoch": 2.23, + "learning_rate": 1.692269610046971e-05, + "loss": 0.2543, + "step": 5329 + }, + { + "epoch": 2.23, + "learning_rate": 1.6921472527883527e-05, + "loss": 0.2752, + "step": 5330 + }, + { + "epoch": 2.23, + "learning_rate": 1.6920248756345757e-05, + "loss": 0.2648, + "step": 5331 + }, + { + "epoch": 2.23, + "learning_rate": 1.6919024785891576e-05, + "loss": 0.2655, + "step": 5332 + }, + { + "epoch": 2.23, + "learning_rate": 1.691780061655617e-05, + "loss": 0.2572, + "step": 5333 + }, + { + "epoch": 2.23, + "learning_rate": 1.691657624837472e-05, + "loss": 0.2709, + "step": 5334 + }, + { + "epoch": 2.23, + "learning_rate": 1.6915351681382427e-05, + "loss": 0.2539, + "step": 5335 + }, + { + "epoch": 2.23, + "learning_rate": 1.6914126915614483e-05, + "loss": 0.2223, + "step": 5336 + }, + { + "epoch": 2.23, + "learning_rate": 1.69129019511061e-05, + "loss": 0.263, + "step": 5337 + }, + { + "epoch": 2.23, + "learning_rate": 1.6911676787892484e-05, + "loss": 0.2504, + "step": 5338 + }, + { + "epoch": 2.24, + "learning_rate": 1.691045142600885e-05, + "loss": 0.2343, + "step": 5339 + }, + { + "epoch": 2.24, + "learning_rate": 1.6909225865490424e-05, + "loss": 0.2212, + "step": 5340 + }, + { + "epoch": 2.24, + "learning_rate": 1.6908000106372432e-05, + "loss": 0.2872, + "step": 5341 + }, + { + "epoch": 2.24, + "learning_rate": 1.6906774148690106e-05, + "loss": 0.2537, + "step": 5342 + }, + { + "epoch": 2.24, + "learning_rate": 1.6905547992478687e-05, + "loss": 0.2897, + "step": 5343 + }, + { + "epoch": 2.24, + "learning_rate": 1.6904321637773414e-05, + "loss": 0.2736, + "step": 5344 + }, + { + "epoch": 2.24, + "learning_rate": 1.690309508460955e-05, + "loss": 0.2745, + "step": 5345 + }, + { + "epoch": 2.24, + "learning_rate": 1.6901868333022337e-05, + "loss": 0.2602, + "step": 5346 + }, + { + "epoch": 2.24, + "learning_rate": 1.6900641383047046e-05, + "loss": 0.2841, + "step": 5347 + }, + { + "epoch": 2.24, + "learning_rate": 1.6899414234718943e-05, + "loss": 0.2467, + "step": 5348 + }, + { + "epoch": 2.24, + "learning_rate": 1.68981868880733e-05, + "loss": 0.2693, + "step": 5349 + }, + { + "epoch": 2.24, + "learning_rate": 1.6896959343145396e-05, + "loss": 0.267, + "step": 5350 + }, + { + "epoch": 2.24, + "learning_rate": 1.6895731599970516e-05, + "loss": 0.2706, + "step": 5351 + }, + { + "epoch": 2.24, + "learning_rate": 1.689450365858395e-05, + "loss": 0.2446, + "step": 5352 + }, + { + "epoch": 2.24, + "learning_rate": 1.689327551902099e-05, + "loss": 0.2822, + "step": 5353 + }, + { + "epoch": 2.24, + "learning_rate": 1.6892047181316952e-05, + "loss": 0.2618, + "step": 5354 + }, + { + "epoch": 2.24, + "learning_rate": 1.6890818645507126e-05, + "loss": 0.2618, + "step": 5355 + }, + { + "epoch": 2.24, + "learning_rate": 1.6889589911626834e-05, + "loss": 0.2688, + "step": 5356 + }, + { + "epoch": 2.24, + "learning_rate": 1.6888360979711395e-05, + "loss": 0.2756, + "step": 5357 + }, + { + "epoch": 2.24, + "learning_rate": 1.6887131849796132e-05, + "loss": 0.2537, + "step": 5358 + }, + { + "epoch": 2.24, + "learning_rate": 1.6885902521916375e-05, + "loss": 0.2688, + "step": 5359 + }, + { + "epoch": 2.24, + "learning_rate": 1.6884672996107464e-05, + "loss": 0.2811, + "step": 5360 + }, + { + "epoch": 2.24, + "learning_rate": 1.6883443272404735e-05, + "loss": 0.2519, + "step": 5361 + }, + { + "epoch": 2.24, + "learning_rate": 1.688221335084353e-05, + "loss": 0.2722, + "step": 5362 + }, + { + "epoch": 2.25, + "learning_rate": 1.6880983231459217e-05, + "loss": 0.2626, + "step": 5363 + }, + { + "epoch": 2.25, + "learning_rate": 1.687975291428715e-05, + "loss": 0.2585, + "step": 5364 + }, + { + "epoch": 2.25, + "learning_rate": 1.6878522399362683e-05, + "loss": 0.2621, + "step": 5365 + }, + { + "epoch": 2.25, + "learning_rate": 1.68772916867212e-05, + "loss": 0.218, + "step": 5366 + }, + { + "epoch": 2.25, + "learning_rate": 1.6876060776398063e-05, + "loss": 0.2472, + "step": 5367 + }, + { + "epoch": 2.25, + "learning_rate": 1.6874829668428667e-05, + "loss": 0.2508, + "step": 5368 + }, + { + "epoch": 2.25, + "learning_rate": 1.687359836284839e-05, + "loss": 0.2585, + "step": 5369 + }, + { + "epoch": 2.25, + "learning_rate": 1.687236685969263e-05, + "loss": 0.2585, + "step": 5370 + }, + { + "epoch": 2.25, + "learning_rate": 1.6871135158996782e-05, + "loss": 0.2454, + "step": 5371 + }, + { + "epoch": 2.25, + "learning_rate": 1.686990326079625e-05, + "loss": 0.2705, + "step": 5372 + }, + { + "epoch": 2.25, + "learning_rate": 1.6868671165126445e-05, + "loss": 0.299, + "step": 5373 + }, + { + "epoch": 2.25, + "learning_rate": 1.6867438872022782e-05, + "loss": 0.2597, + "step": 5374 + }, + { + "epoch": 2.25, + "learning_rate": 1.6866206381520683e-05, + "loss": 0.2684, + "step": 5375 + }, + { + "epoch": 2.25, + "learning_rate": 1.6864973693655577e-05, + "loss": 0.243, + "step": 5376 + }, + { + "epoch": 2.25, + "learning_rate": 1.686374080846289e-05, + "loss": 0.2693, + "step": 5377 + }, + { + "epoch": 2.25, + "learning_rate": 1.6862507725978067e-05, + "loss": 0.2694, + "step": 5378 + }, + { + "epoch": 2.25, + "learning_rate": 1.686127444623655e-05, + "loss": 0.2304, + "step": 5379 + }, + { + "epoch": 2.25, + "learning_rate": 1.686004096927378e-05, + "loss": 0.2439, + "step": 5380 + }, + { + "epoch": 2.25, + "learning_rate": 1.685880729512523e-05, + "loss": 0.2477, + "step": 5381 + }, + { + "epoch": 2.25, + "learning_rate": 1.6857573423826344e-05, + "loss": 0.264, + "step": 5382 + }, + { + "epoch": 2.25, + "learning_rate": 1.6856339355412602e-05, + "loss": 0.233, + "step": 5383 + }, + { + "epoch": 2.25, + "learning_rate": 1.6855105089919467e-05, + "loss": 0.2768, + "step": 5384 + }, + { + "epoch": 2.25, + "learning_rate": 1.6853870627382416e-05, + "loss": 0.2539, + "step": 5385 + }, + { + "epoch": 2.25, + "learning_rate": 1.6852635967836936e-05, + "loss": 0.2581, + "step": 5386 + }, + { + "epoch": 2.26, + "learning_rate": 1.685140111131852e-05, + "loss": 0.2704, + "step": 5387 + }, + { + "epoch": 2.26, + "learning_rate": 1.6850166057862654e-05, + "loss": 0.2789, + "step": 5388 + }, + { + "epoch": 2.26, + "learning_rate": 1.6848930807504846e-05, + "loss": 0.2843, + "step": 5389 + }, + { + "epoch": 2.26, + "learning_rate": 1.68476953602806e-05, + "loss": 0.2948, + "step": 5390 + }, + { + "epoch": 2.26, + "learning_rate": 1.684645971622543e-05, + "loss": 0.255, + "step": 5391 + }, + { + "epoch": 2.26, + "learning_rate": 1.6845223875374845e-05, + "loss": 0.2635, + "step": 5392 + }, + { + "epoch": 2.26, + "learning_rate": 1.6843987837764375e-05, + "loss": 0.2963, + "step": 5393 + }, + { + "epoch": 2.26, + "learning_rate": 1.684275160342955e-05, + "loss": 0.2592, + "step": 5394 + }, + { + "epoch": 2.26, + "learning_rate": 1.68415151724059e-05, + "loss": 0.2682, + "step": 5395 + }, + { + "epoch": 2.26, + "learning_rate": 1.684027854472897e-05, + "loss": 0.2681, + "step": 5396 + }, + { + "epoch": 2.26, + "learning_rate": 1.6839041720434303e-05, + "loss": 0.2474, + "step": 5397 + }, + { + "epoch": 2.26, + "learning_rate": 1.683780469955745e-05, + "loss": 0.2892, + "step": 5398 + }, + { + "epoch": 2.26, + "learning_rate": 1.683656748213397e-05, + "loss": 0.2526, + "step": 5399 + }, + { + "epoch": 2.26, + "learning_rate": 1.6835330068199424e-05, + "loss": 0.2477, + "step": 5400 + }, + { + "epoch": 2.26, + "learning_rate": 1.683409245778938e-05, + "loss": 0.2851, + "step": 5401 + }, + { + "epoch": 2.26, + "learning_rate": 1.683285465093941e-05, + "loss": 0.2454, + "step": 5402 + }, + { + "epoch": 2.26, + "learning_rate": 1.6831616647685102e-05, + "loss": 0.2666, + "step": 5403 + }, + { + "epoch": 2.26, + "learning_rate": 1.683037844806203e-05, + "loss": 0.2602, + "step": 5404 + }, + { + "epoch": 2.26, + "learning_rate": 1.6829140052105795e-05, + "loss": 0.2423, + "step": 5405 + }, + { + "epoch": 2.26, + "learning_rate": 1.6827901459851986e-05, + "loss": 0.2727, + "step": 5406 + }, + { + "epoch": 2.26, + "learning_rate": 1.6826662671336212e-05, + "loss": 0.2452, + "step": 5407 + }, + { + "epoch": 2.26, + "learning_rate": 1.6825423686594075e-05, + "loss": 0.2498, + "step": 5408 + }, + { + "epoch": 2.26, + "learning_rate": 1.682418450566119e-05, + "loss": 0.2509, + "step": 5409 + }, + { + "epoch": 2.27, + "learning_rate": 1.6822945128573182e-05, + "loss": 0.2815, + "step": 5410 + }, + { + "epoch": 2.27, + "learning_rate": 1.6821705555365665e-05, + "loss": 0.2747, + "step": 5411 + }, + { + "epoch": 2.27, + "learning_rate": 1.6820465786074277e-05, + "loss": 0.2632, + "step": 5412 + }, + { + "epoch": 2.27, + "learning_rate": 1.6819225820734656e-05, + "loss": 0.2268, + "step": 5413 + }, + { + "epoch": 2.27, + "learning_rate": 1.6817985659382437e-05, + "loss": 0.2475, + "step": 5414 + }, + { + "epoch": 2.27, + "learning_rate": 1.681674530205327e-05, + "loss": 0.2406, + "step": 5415 + }, + { + "epoch": 2.27, + "learning_rate": 1.681550474878281e-05, + "loss": 0.2551, + "step": 5416 + }, + { + "epoch": 2.27, + "learning_rate": 1.6814263999606713e-05, + "loss": 0.2745, + "step": 5417 + }, + { + "epoch": 2.27, + "learning_rate": 1.681302305456064e-05, + "loss": 0.2644, + "step": 5418 + }, + { + "epoch": 2.27, + "learning_rate": 1.6811781913680273e-05, + "loss": 0.2814, + "step": 5419 + }, + { + "epoch": 2.27, + "learning_rate": 1.6810540577001277e-05, + "loss": 0.2529, + "step": 5420 + }, + { + "epoch": 2.27, + "learning_rate": 1.6809299044559333e-05, + "loss": 0.2791, + "step": 5421 + }, + { + "epoch": 2.27, + "learning_rate": 1.6808057316390128e-05, + "loss": 0.255, + "step": 5422 + }, + { + "epoch": 2.27, + "learning_rate": 1.6806815392529363e-05, + "loss": 0.26, + "step": 5423 + }, + { + "epoch": 2.27, + "learning_rate": 1.680557327301273e-05, + "loss": 0.256, + "step": 5424 + }, + { + "epoch": 2.27, + "learning_rate": 1.6804330957875926e-05, + "loss": 0.2486, + "step": 5425 + }, + { + "epoch": 2.27, + "learning_rate": 1.6803088447154672e-05, + "loss": 0.2407, + "step": 5426 + }, + { + "epoch": 2.27, + "learning_rate": 1.6801845740884678e-05, + "loss": 0.2346, + "step": 5427 + }, + { + "epoch": 2.27, + "learning_rate": 1.680060283910166e-05, + "loss": 0.2524, + "step": 5428 + }, + { + "epoch": 2.27, + "learning_rate": 1.6799359741841354e-05, + "loss": 0.2708, + "step": 5429 + }, + { + "epoch": 2.27, + "learning_rate": 1.6798116449139483e-05, + "loss": 0.2632, + "step": 5430 + }, + { + "epoch": 2.27, + "learning_rate": 1.6796872961031783e-05, + "loss": 0.2237, + "step": 5431 + }, + { + "epoch": 2.27, + "learning_rate": 1.6795629277554003e-05, + "loss": 0.2382, + "step": 5432 + }, + { + "epoch": 2.27, + "learning_rate": 1.6794385398741894e-05, + "loss": 0.2691, + "step": 5433 + }, + { + "epoch": 2.28, + "learning_rate": 1.67931413246312e-05, + "loss": 0.2636, + "step": 5434 + }, + { + "epoch": 2.28, + "learning_rate": 1.679189705525769e-05, + "loss": 0.2594, + "step": 5435 + }, + { + "epoch": 2.28, + "learning_rate": 1.6790652590657125e-05, + "loss": 0.2543, + "step": 5436 + }, + { + "epoch": 2.28, + "learning_rate": 1.678940793086528e-05, + "loss": 0.2591, + "step": 5437 + }, + { + "epoch": 2.28, + "learning_rate": 1.678816307591793e-05, + "loss": 0.2541, + "step": 5438 + }, + { + "epoch": 2.28, + "learning_rate": 1.6786918025850852e-05, + "loss": 0.2863, + "step": 5439 + }, + { + "epoch": 2.28, + "learning_rate": 1.6785672780699836e-05, + "loss": 0.2611, + "step": 5440 + }, + { + "epoch": 2.28, + "learning_rate": 1.6784427340500684e-05, + "loss": 0.286, + "step": 5441 + }, + { + "epoch": 2.28, + "learning_rate": 1.6783181705289184e-05, + "loss": 0.244, + "step": 5442 + }, + { + "epoch": 2.28, + "learning_rate": 1.6781935875101148e-05, + "loss": 0.2727, + "step": 5443 + }, + { + "epoch": 2.28, + "learning_rate": 1.678068984997238e-05, + "loss": 0.2597, + "step": 5444 + }, + { + "epoch": 2.28, + "learning_rate": 1.67794436299387e-05, + "loss": 0.2972, + "step": 5445 + }, + { + "epoch": 2.28, + "learning_rate": 1.677819721503593e-05, + "loss": 0.2636, + "step": 5446 + }, + { + "epoch": 2.28, + "learning_rate": 1.6776950605299897e-05, + "loss": 0.2327, + "step": 5447 + }, + { + "epoch": 2.28, + "learning_rate": 1.677570380076643e-05, + "loss": 0.275, + "step": 5448 + }, + { + "epoch": 2.28, + "learning_rate": 1.677445680147137e-05, + "loss": 0.2507, + "step": 5449 + }, + { + "epoch": 2.28, + "learning_rate": 1.6773209607450564e-05, + "loss": 0.2583, + "step": 5450 + }, + { + "epoch": 2.28, + "learning_rate": 1.6771962218739856e-05, + "loss": 0.2828, + "step": 5451 + }, + { + "epoch": 2.28, + "learning_rate": 1.6770714635375104e-05, + "loss": 0.2481, + "step": 5452 + }, + { + "epoch": 2.28, + "learning_rate": 1.6769466857392167e-05, + "loss": 0.2325, + "step": 5453 + }, + { + "epoch": 2.28, + "learning_rate": 1.6768218884826916e-05, + "loss": 0.2542, + "step": 5454 + }, + { + "epoch": 2.28, + "learning_rate": 1.6766970717715216e-05, + "loss": 0.2508, + "step": 5455 + }, + { + "epoch": 2.28, + "learning_rate": 1.6765722356092947e-05, + "loss": 0.256, + "step": 5456 + }, + { + "epoch": 2.28, + "learning_rate": 1.6764473799995993e-05, + "loss": 0.261, + "step": 5457 + }, + { + "epoch": 2.29, + "learning_rate": 1.676322504946024e-05, + "loss": 0.2655, + "step": 5458 + }, + { + "epoch": 2.29, + "learning_rate": 1.676197610452159e-05, + "loss": 0.2763, + "step": 5459 + }, + { + "epoch": 2.29, + "learning_rate": 1.6760726965215936e-05, + "loss": 0.2443, + "step": 5460 + }, + { + "epoch": 2.29, + "learning_rate": 1.6759477631579182e-05, + "loss": 0.2783, + "step": 5461 + }, + { + "epoch": 2.29, + "learning_rate": 1.6758228103647246e-05, + "loss": 0.2794, + "step": 5462 + }, + { + "epoch": 2.29, + "learning_rate": 1.6756978381456035e-05, + "loss": 0.2522, + "step": 5463 + }, + { + "epoch": 2.29, + "learning_rate": 1.675572846504148e-05, + "loss": 0.2734, + "step": 5464 + }, + { + "epoch": 2.29, + "learning_rate": 1.6754478354439504e-05, + "loss": 0.2389, + "step": 5465 + }, + { + "epoch": 2.29, + "learning_rate": 1.6753228049686042e-05, + "loss": 0.2517, + "step": 5466 + }, + { + "epoch": 2.29, + "learning_rate": 1.675197755081703e-05, + "loss": 0.2726, + "step": 5467 + }, + { + "epoch": 2.29, + "learning_rate": 1.675072685786842e-05, + "loss": 0.2493, + "step": 5468 + }, + { + "epoch": 2.29, + "learning_rate": 1.674947597087615e-05, + "loss": 0.2413, + "step": 5469 + }, + { + "epoch": 2.29, + "learning_rate": 1.6748224889876188e-05, + "loss": 0.2812, + "step": 5470 + }, + { + "epoch": 2.29, + "learning_rate": 1.674697361490449e-05, + "loss": 0.2636, + "step": 5471 + }, + { + "epoch": 2.29, + "learning_rate": 1.674572214599702e-05, + "loss": 0.2142, + "step": 5472 + }, + { + "epoch": 2.29, + "learning_rate": 1.6744470483189755e-05, + "loss": 0.2527, + "step": 5473 + }, + { + "epoch": 2.29, + "learning_rate": 1.674321862651867e-05, + "loss": 0.2542, + "step": 5474 + }, + { + "epoch": 2.29, + "learning_rate": 1.674196657601975e-05, + "loss": 0.2612, + "step": 5475 + }, + { + "epoch": 2.29, + "learning_rate": 1.6740714331728984e-05, + "loss": 0.2351, + "step": 5476 + }, + { + "epoch": 2.29, + "learning_rate": 1.673946189368236e-05, + "loss": 0.2701, + "step": 5477 + }, + { + "epoch": 2.29, + "learning_rate": 1.673820926191589e-05, + "loss": 0.2584, + "step": 5478 + }, + { + "epoch": 2.29, + "learning_rate": 1.6736956436465573e-05, + "loss": 0.2623, + "step": 5479 + }, + { + "epoch": 2.29, + "learning_rate": 1.6735703417367422e-05, + "loss": 0.2528, + "step": 5480 + }, + { + "epoch": 2.29, + "learning_rate": 1.673445020465745e-05, + "loss": 0.2376, + "step": 5481 + }, + { + "epoch": 2.3, + "learning_rate": 1.673319679837169e-05, + "loss": 0.2837, + "step": 5482 + }, + { + "epoch": 2.3, + "learning_rate": 1.673194319854616e-05, + "loss": 0.261, + "step": 5483 + }, + { + "epoch": 2.3, + "learning_rate": 1.6730689405216895e-05, + "loss": 0.2461, + "step": 5484 + }, + { + "epoch": 2.3, + "learning_rate": 1.6729435418419936e-05, + "loss": 0.2953, + "step": 5485 + }, + { + "epoch": 2.3, + "learning_rate": 1.6728181238191328e-05, + "loss": 0.2853, + "step": 5486 + }, + { + "epoch": 2.3, + "learning_rate": 1.672692686456712e-05, + "loss": 0.2504, + "step": 5487 + }, + { + "epoch": 2.3, + "learning_rate": 1.672567229758337e-05, + "loss": 0.2589, + "step": 5488 + }, + { + "epoch": 2.3, + "learning_rate": 1.6724417537276138e-05, + "loss": 0.2655, + "step": 5489 + }, + { + "epoch": 2.3, + "learning_rate": 1.6723162583681492e-05, + "loss": 0.249, + "step": 5490 + }, + { + "epoch": 2.3, + "learning_rate": 1.67219074368355e-05, + "loss": 0.2526, + "step": 5491 + }, + { + "epoch": 2.3, + "learning_rate": 1.6720652096774245e-05, + "loss": 0.2313, + "step": 5492 + }, + { + "epoch": 2.3, + "learning_rate": 1.6719396563533812e-05, + "loss": 0.2452, + "step": 5493 + }, + { + "epoch": 2.3, + "learning_rate": 1.6718140837150286e-05, + "loss": 0.2643, + "step": 5494 + }, + { + "epoch": 2.3, + "learning_rate": 1.6716884917659763e-05, + "loss": 0.2852, + "step": 5495 + }, + { + "epoch": 2.3, + "learning_rate": 1.671562880509834e-05, + "loss": 0.2577, + "step": 5496 + }, + { + "epoch": 2.3, + "learning_rate": 1.6714372499502133e-05, + "loss": 0.2387, + "step": 5497 + }, + { + "epoch": 2.3, + "learning_rate": 1.671311600090724e-05, + "loss": 0.2763, + "step": 5498 + }, + { + "epoch": 2.3, + "learning_rate": 1.6711859309349788e-05, + "loss": 0.2555, + "step": 5499 + }, + { + "epoch": 2.3, + "learning_rate": 1.67106024248659e-05, + "loss": 0.2557, + "step": 5500 + }, + { + "epoch": 2.3, + "learning_rate": 1.6709345347491695e-05, + "loss": 0.2474, + "step": 5501 + }, + { + "epoch": 2.3, + "learning_rate": 1.670808807726331e-05, + "loss": 0.261, + "step": 5502 + }, + { + "epoch": 2.3, + "learning_rate": 1.670683061421689e-05, + "loss": 0.247, + "step": 5503 + }, + { + "epoch": 2.3, + "learning_rate": 1.6705572958388576e-05, + "loss": 0.26, + "step": 5504 + }, + { + "epoch": 2.3, + "learning_rate": 1.6704315109814514e-05, + "loss": 0.2832, + "step": 5505 + }, + { + "epoch": 2.31, + "learning_rate": 1.670305706853087e-05, + "loss": 0.2386, + "step": 5506 + }, + { + "epoch": 2.31, + "learning_rate": 1.670179883457379e-05, + "loss": 0.286, + "step": 5507 + }, + { + "epoch": 2.31, + "learning_rate": 1.6700540407979457e-05, + "loss": 0.2268, + "step": 5508 + }, + { + "epoch": 2.31, + "learning_rate": 1.669928178878403e-05, + "loss": 0.2397, + "step": 5509 + }, + { + "epoch": 2.31, + "learning_rate": 1.6698022977023697e-05, + "loss": 0.25, + "step": 5510 + }, + { + "epoch": 2.31, + "learning_rate": 1.6696763972734637e-05, + "loss": 0.2595, + "step": 5511 + }, + { + "epoch": 2.31, + "learning_rate": 1.6695504775953036e-05, + "loss": 0.244, + "step": 5512 + }, + { + "epoch": 2.31, + "learning_rate": 1.6694245386715094e-05, + "loss": 0.2475, + "step": 5513 + }, + { + "epoch": 2.31, + "learning_rate": 1.669298580505701e-05, + "loss": 0.253, + "step": 5514 + }, + { + "epoch": 2.31, + "learning_rate": 1.6691726031014985e-05, + "loss": 0.2752, + "step": 5515 + }, + { + "epoch": 2.31, + "learning_rate": 1.669046606462524e-05, + "loss": 0.2726, + "step": 5516 + }, + { + "epoch": 2.31, + "learning_rate": 1.6689205905923977e-05, + "loss": 0.238, + "step": 5517 + }, + { + "epoch": 2.31, + "learning_rate": 1.6687945554947428e-05, + "loss": 0.2627, + "step": 5518 + }, + { + "epoch": 2.31, + "learning_rate": 1.668668501173182e-05, + "loss": 0.2485, + "step": 5519 + }, + { + "epoch": 2.31, + "learning_rate": 1.6685424276313386e-05, + "loss": 0.2596, + "step": 5520 + }, + { + "epoch": 2.31, + "learning_rate": 1.6684163348728364e-05, + "loss": 0.2893, + "step": 5521 + }, + { + "epoch": 2.31, + "learning_rate": 1.6682902229012997e-05, + "loss": 0.2622, + "step": 5522 + }, + { + "epoch": 2.31, + "learning_rate": 1.6681640917203534e-05, + "loss": 0.2604, + "step": 5523 + }, + { + "epoch": 2.31, + "learning_rate": 1.668037941333623e-05, + "loss": 0.2601, + "step": 5524 + }, + { + "epoch": 2.31, + "learning_rate": 1.6679117717447356e-05, + "loss": 0.2353, + "step": 5525 + }, + { + "epoch": 2.31, + "learning_rate": 1.6677855829573164e-05, + "loss": 0.242, + "step": 5526 + }, + { + "epoch": 2.31, + "learning_rate": 1.6676593749749936e-05, + "loss": 0.2716, + "step": 5527 + }, + { + "epoch": 2.31, + "learning_rate": 1.667533147801394e-05, + "loss": 0.258, + "step": 5528 + }, + { + "epoch": 2.31, + "learning_rate": 1.667406901440147e-05, + "loss": 0.273, + "step": 5529 + }, + { + "epoch": 2.32, + "learning_rate": 1.6672806358948803e-05, + "loss": 0.2658, + "step": 5530 + }, + { + "epoch": 2.32, + "learning_rate": 1.667154351169224e-05, + "loss": 0.2419, + "step": 5531 + }, + { + "epoch": 2.32, + "learning_rate": 1.6670280472668082e-05, + "loss": 0.2512, + "step": 5532 + }, + { + "epoch": 2.32, + "learning_rate": 1.666901724191263e-05, + "loss": 0.2694, + "step": 5533 + }, + { + "epoch": 2.32, + "learning_rate": 1.6667753819462196e-05, + "loss": 0.2649, + "step": 5534 + }, + { + "epoch": 2.32, + "learning_rate": 1.6666490205353092e-05, + "loss": 0.2606, + "step": 5535 + }, + { + "epoch": 2.32, + "learning_rate": 1.6665226399621642e-05, + "loss": 0.2565, + "step": 5536 + }, + { + "epoch": 2.32, + "learning_rate": 1.6663962402304178e-05, + "loss": 0.239, + "step": 5537 + }, + { + "epoch": 2.32, + "learning_rate": 1.666269821343703e-05, + "loss": 0.2428, + "step": 5538 + }, + { + "epoch": 2.32, + "learning_rate": 1.6661433833056527e-05, + "loss": 0.2325, + "step": 5539 + }, + { + "epoch": 2.32, + "learning_rate": 1.6660169261199027e-05, + "loss": 0.2501, + "step": 5540 + }, + { + "epoch": 2.32, + "learning_rate": 1.665890449790087e-05, + "loss": 0.2478, + "step": 5541 + }, + { + "epoch": 2.32, + "learning_rate": 1.665763954319841e-05, + "loss": 0.23, + "step": 5542 + }, + { + "epoch": 2.32, + "learning_rate": 1.6656374397128007e-05, + "loss": 0.2581, + "step": 5543 + }, + { + "epoch": 2.32, + "learning_rate": 1.665510905972603e-05, + "loss": 0.2491, + "step": 5544 + }, + { + "epoch": 2.32, + "learning_rate": 1.6653843531028852e-05, + "loss": 0.255, + "step": 5545 + }, + { + "epoch": 2.32, + "learning_rate": 1.6652577811072842e-05, + "loss": 0.2237, + "step": 5546 + }, + { + "epoch": 2.32, + "learning_rate": 1.665131189989439e-05, + "loss": 0.2275, + "step": 5547 + }, + { + "epoch": 2.32, + "learning_rate": 1.665004579752988e-05, + "loss": 0.2532, + "step": 5548 + }, + { + "epoch": 2.32, + "learning_rate": 1.66487795040157e-05, + "loss": 0.2732, + "step": 5549 + }, + { + "epoch": 2.32, + "learning_rate": 1.664751301938826e-05, + "loss": 0.2887, + "step": 5550 + }, + { + "epoch": 2.32, + "learning_rate": 1.6646246343683953e-05, + "loss": 0.2389, + "step": 5551 + }, + { + "epoch": 2.32, + "learning_rate": 1.6644979476939194e-05, + "loss": 0.2584, + "step": 5552 + }, + { + "epoch": 2.32, + "learning_rate": 1.6643712419190396e-05, + "loss": 0.2517, + "step": 5553 + }, + { + "epoch": 2.33, + "learning_rate": 1.664244517047398e-05, + "loss": 0.2563, + "step": 5554 + }, + { + "epoch": 2.33, + "learning_rate": 1.664117773082637e-05, + "loss": 0.2428, + "step": 5555 + }, + { + "epoch": 2.33, + "learning_rate": 1.6639910100284e-05, + "loss": 0.2671, + "step": 5556 + }, + { + "epoch": 2.33, + "learning_rate": 1.663864227888331e-05, + "loss": 0.2196, + "step": 5557 + }, + { + "epoch": 2.33, + "learning_rate": 1.6637374266660734e-05, + "loss": 0.2565, + "step": 5558 + }, + { + "epoch": 2.33, + "learning_rate": 1.6636106063652725e-05, + "loss": 0.2578, + "step": 5559 + }, + { + "epoch": 2.33, + "learning_rate": 1.663483766989574e-05, + "loss": 0.2725, + "step": 5560 + }, + { + "epoch": 2.33, + "learning_rate": 1.663356908542623e-05, + "loss": 0.2808, + "step": 5561 + }, + { + "epoch": 2.33, + "learning_rate": 1.6632300310280664e-05, + "loss": 0.2417, + "step": 5562 + }, + { + "epoch": 2.33, + "learning_rate": 1.663103134449551e-05, + "loss": 0.2653, + "step": 5563 + }, + { + "epoch": 2.33, + "learning_rate": 1.6629762188107248e-05, + "loss": 0.2614, + "step": 5564 + }, + { + "epoch": 2.33, + "learning_rate": 1.6628492841152357e-05, + "loss": 0.2483, + "step": 5565 + }, + { + "epoch": 2.33, + "learning_rate": 1.6627223303667314e-05, + "loss": 0.2439, + "step": 5566 + }, + { + "epoch": 2.33, + "learning_rate": 1.6625953575688624e-05, + "loss": 0.2314, + "step": 5567 + }, + { + "epoch": 2.33, + "learning_rate": 1.6624683657252773e-05, + "loss": 0.216, + "step": 5568 + }, + { + "epoch": 2.33, + "learning_rate": 1.6623413548396272e-05, + "loss": 0.253, + "step": 5569 + }, + { + "epoch": 2.33, + "learning_rate": 1.6622143249155625e-05, + "loss": 0.2446, + "step": 5570 + }, + { + "epoch": 2.33, + "learning_rate": 1.6620872759567345e-05, + "loss": 0.256, + "step": 5571 + }, + { + "epoch": 2.33, + "learning_rate": 1.6619602079667956e-05, + "loss": 0.2411, + "step": 5572 + }, + { + "epoch": 2.33, + "learning_rate": 1.661833120949398e-05, + "loss": 0.2755, + "step": 5573 + }, + { + "epoch": 2.33, + "learning_rate": 1.6617060149081944e-05, + "loss": 0.2411, + "step": 5574 + }, + { + "epoch": 2.33, + "learning_rate": 1.6615788898468384e-05, + "loss": 0.2439, + "step": 5575 + }, + { + "epoch": 2.33, + "learning_rate": 1.661451745768985e-05, + "loss": 0.2606, + "step": 5576 + }, + { + "epoch": 2.33, + "learning_rate": 1.6613245826782875e-05, + "loss": 0.288, + "step": 5577 + }, + { + "epoch": 2.34, + "learning_rate": 1.6611974005784017e-05, + "loss": 0.2454, + "step": 5578 + }, + { + "epoch": 2.34, + "learning_rate": 1.6610701994729838e-05, + "loss": 0.2253, + "step": 5579 + }, + { + "epoch": 2.34, + "learning_rate": 1.660942979365689e-05, + "loss": 0.2392, + "step": 5580 + }, + { + "epoch": 2.34, + "learning_rate": 1.6608157402601753e-05, + "loss": 0.2397, + "step": 5581 + }, + { + "epoch": 2.34, + "learning_rate": 1.6606884821600995e-05, + "loss": 0.2303, + "step": 5582 + }, + { + "epoch": 2.34, + "learning_rate": 1.6605612050691197e-05, + "loss": 0.2431, + "step": 5583 + }, + { + "epoch": 2.34, + "learning_rate": 1.6604339089908935e-05, + "loss": 0.2579, + "step": 5584 + }, + { + "epoch": 2.34, + "learning_rate": 1.6603065939290815e-05, + "loss": 0.2297, + "step": 5585 + }, + { + "epoch": 2.34, + "learning_rate": 1.660179259887342e-05, + "loss": 0.2695, + "step": 5586 + }, + { + "epoch": 2.34, + "learning_rate": 1.6600519068693356e-05, + "loss": 0.2433, + "step": 5587 + }, + { + "epoch": 2.34, + "learning_rate": 1.659924534878723e-05, + "loss": 0.2382, + "step": 5588 + }, + { + "epoch": 2.34, + "learning_rate": 1.659797143919165e-05, + "loss": 0.2639, + "step": 5589 + }, + { + "epoch": 2.34, + "learning_rate": 1.6596697339943234e-05, + "loss": 0.2494, + "step": 5590 + }, + { + "epoch": 2.34, + "learning_rate": 1.659542305107861e-05, + "loss": 0.283, + "step": 5591 + }, + { + "epoch": 2.34, + "learning_rate": 1.6594148572634404e-05, + "loss": 0.2544, + "step": 5592 + }, + { + "epoch": 2.34, + "learning_rate": 1.6592873904647245e-05, + "loss": 0.2498, + "step": 5593 + }, + { + "epoch": 2.34, + "learning_rate": 1.6591599047153778e-05, + "loss": 0.2804, + "step": 5594 + }, + { + "epoch": 2.34, + "learning_rate": 1.6590324000190645e-05, + "loss": 0.2287, + "step": 5595 + }, + { + "epoch": 2.34, + "learning_rate": 1.65890487637945e-05, + "loss": 0.2674, + "step": 5596 + }, + { + "epoch": 2.34, + "learning_rate": 1.6587773338001993e-05, + "loss": 0.2598, + "step": 5597 + }, + { + "epoch": 2.34, + "learning_rate": 1.658649772284979e-05, + "loss": 0.2656, + "step": 5598 + }, + { + "epoch": 2.34, + "learning_rate": 1.658522191837455e-05, + "loss": 0.2451, + "step": 5599 + }, + { + "epoch": 2.34, + "learning_rate": 1.658394592461295e-05, + "loss": 0.2575, + "step": 5600 + }, + { + "epoch": 2.34, + "learning_rate": 1.658266974160167e-05, + "loss": 0.259, + "step": 5601 + }, + { + "epoch": 2.35, + "learning_rate": 1.6581393369377385e-05, + "loss": 0.2611, + "step": 5602 + }, + { + "epoch": 2.35, + "learning_rate": 1.6580116807976795e-05, + "loss": 0.2447, + "step": 5603 + }, + { + "epoch": 2.35, + "learning_rate": 1.657884005743658e-05, + "loss": 0.2716, + "step": 5604 + }, + { + "epoch": 2.35, + "learning_rate": 1.6577563117793447e-05, + "loss": 0.2492, + "step": 5605 + }, + { + "epoch": 2.35, + "learning_rate": 1.65762859890841e-05, + "loss": 0.2713, + "step": 5606 + }, + { + "epoch": 2.35, + "learning_rate": 1.6575008671345248e-05, + "loss": 0.2381, + "step": 5607 + }, + { + "epoch": 2.35, + "learning_rate": 1.6573731164613604e-05, + "loss": 0.2521, + "step": 5608 + }, + { + "epoch": 2.35, + "learning_rate": 1.6572453468925895e-05, + "loss": 0.2688, + "step": 5609 + }, + { + "epoch": 2.35, + "learning_rate": 1.6571175584318845e-05, + "loss": 0.2438, + "step": 5610 + }, + { + "epoch": 2.35, + "learning_rate": 1.656989751082918e-05, + "loss": 0.2655, + "step": 5611 + }, + { + "epoch": 2.35, + "learning_rate": 1.656861924849364e-05, + "loss": 0.2333, + "step": 5612 + }, + { + "epoch": 2.35, + "learning_rate": 1.656734079734897e-05, + "loss": 0.277, + "step": 5613 + }, + { + "epoch": 2.35, + "learning_rate": 1.6566062157431915e-05, + "loss": 0.2572, + "step": 5614 + }, + { + "epoch": 2.35, + "learning_rate": 1.656478332877923e-05, + "loss": 0.2439, + "step": 5615 + }, + { + "epoch": 2.35, + "learning_rate": 1.6563504311427678e-05, + "loss": 0.2518, + "step": 5616 + }, + { + "epoch": 2.35, + "learning_rate": 1.6562225105414016e-05, + "loss": 0.2696, + "step": 5617 + }, + { + "epoch": 2.35, + "learning_rate": 1.656094571077502e-05, + "loss": 0.2628, + "step": 5618 + }, + { + "epoch": 2.35, + "learning_rate": 1.6559666127547455e-05, + "loss": 0.2394, + "step": 5619 + }, + { + "epoch": 2.35, + "learning_rate": 1.655838635576811e-05, + "loss": 0.2871, + "step": 5620 + }, + { + "epoch": 2.35, + "learning_rate": 1.6557106395473775e-05, + "loss": 0.2729, + "step": 5621 + }, + { + "epoch": 2.35, + "learning_rate": 1.655582624670123e-05, + "loss": 0.2643, + "step": 5622 + }, + { + "epoch": 2.35, + "learning_rate": 1.6554545909487277e-05, + "loss": 0.2522, + "step": 5623 + }, + { + "epoch": 2.35, + "learning_rate": 1.655326538386872e-05, + "loss": 0.2616, + "step": 5624 + }, + { + "epoch": 2.36, + "learning_rate": 1.6551984669882362e-05, + "loss": 0.2816, + "step": 5625 + }, + { + "epoch": 2.36, + "learning_rate": 1.655070376756502e-05, + "loss": 0.2558, + "step": 5626 + }, + { + "epoch": 2.36, + "learning_rate": 1.6549422676953513e-05, + "loss": 0.2902, + "step": 5627 + }, + { + "epoch": 2.36, + "learning_rate": 1.6548141398084662e-05, + "loss": 0.2461, + "step": 5628 + }, + { + "epoch": 2.36, + "learning_rate": 1.6546859930995295e-05, + "loss": 0.2601, + "step": 5629 + }, + { + "epoch": 2.36, + "learning_rate": 1.654557827572225e-05, + "loss": 0.2647, + "step": 5630 + }, + { + "epoch": 2.36, + "learning_rate": 1.6544296432302373e-05, + "loss": 0.2559, + "step": 5631 + }, + { + "epoch": 2.36, + "learning_rate": 1.6543014400772496e-05, + "loss": 0.2638, + "step": 5632 + }, + { + "epoch": 2.36, + "learning_rate": 1.6541732181169474e-05, + "loss": 0.2409, + "step": 5633 + }, + { + "epoch": 2.36, + "learning_rate": 1.6540449773530168e-05, + "loss": 0.2524, + "step": 5634 + }, + { + "epoch": 2.36, + "learning_rate": 1.6539167177891436e-05, + "loss": 0.2701, + "step": 5635 + }, + { + "epoch": 2.36, + "learning_rate": 1.653788439429015e-05, + "loss": 0.2518, + "step": 5636 + }, + { + "epoch": 2.36, + "learning_rate": 1.6536601422763173e-05, + "loss": 0.2835, + "step": 5637 + }, + { + "epoch": 2.36, + "learning_rate": 1.6535318263347395e-05, + "loss": 0.2598, + "step": 5638 + }, + { + "epoch": 2.36, + "learning_rate": 1.6534034916079686e-05, + "loss": 0.2588, + "step": 5639 + }, + { + "epoch": 2.36, + "learning_rate": 1.653275138099695e-05, + "loss": 0.222, + "step": 5640 + }, + { + "epoch": 2.36, + "learning_rate": 1.6531467658136064e-05, + "loss": 0.2774, + "step": 5641 + }, + { + "epoch": 2.36, + "learning_rate": 1.6530183747533942e-05, + "loss": 0.2674, + "step": 5642 + }, + { + "epoch": 2.36, + "learning_rate": 1.6528899649227482e-05, + "loss": 0.26, + "step": 5643 + }, + { + "epoch": 2.36, + "learning_rate": 1.652761536325359e-05, + "loss": 0.2622, + "step": 5644 + }, + { + "epoch": 2.36, + "learning_rate": 1.6526330889649194e-05, + "loss": 0.2339, + "step": 5645 + }, + { + "epoch": 2.36, + "learning_rate": 1.6525046228451204e-05, + "loss": 0.279, + "step": 5646 + }, + { + "epoch": 2.36, + "learning_rate": 1.652376137969655e-05, + "loss": 0.2597, + "step": 5647 + }, + { + "epoch": 2.36, + "learning_rate": 1.6522476343422164e-05, + "loss": 0.2777, + "step": 5648 + }, + { + "epoch": 2.37, + "learning_rate": 1.652119111966499e-05, + "loss": 0.2719, + "step": 5649 + }, + { + "epoch": 2.37, + "learning_rate": 1.6519905708461953e-05, + "loss": 0.2532, + "step": 5650 + }, + { + "epoch": 2.37, + "learning_rate": 1.651862010985002e-05, + "loss": 0.2652, + "step": 5651 + }, + { + "epoch": 2.37, + "learning_rate": 1.6517334323866136e-05, + "loss": 0.2313, + "step": 5652 + }, + { + "epoch": 2.37, + "learning_rate": 1.651604835054726e-05, + "loss": 0.253, + "step": 5653 + }, + { + "epoch": 2.37, + "learning_rate": 1.6514762189930353e-05, + "loss": 0.2641, + "step": 5654 + }, + { + "epoch": 2.37, + "learning_rate": 1.651347584205239e-05, + "loss": 0.2449, + "step": 5655 + }, + { + "epoch": 2.37, + "learning_rate": 1.6512189306950345e-05, + "loss": 0.2649, + "step": 5656 + }, + { + "epoch": 2.37, + "learning_rate": 1.6510902584661195e-05, + "loss": 0.2837, + "step": 5657 + }, + { + "epoch": 2.37, + "learning_rate": 1.650961567522193e-05, + "loss": 0.2499, + "step": 5658 + }, + { + "epoch": 2.37, + "learning_rate": 1.6508328578669537e-05, + "loss": 0.2475, + "step": 5659 + }, + { + "epoch": 2.37, + "learning_rate": 1.6507041295041015e-05, + "loss": 0.246, + "step": 5660 + }, + { + "epoch": 2.37, + "learning_rate": 1.6505753824373364e-05, + "loss": 0.2631, + "step": 5661 + }, + { + "epoch": 2.37, + "learning_rate": 1.6504466166703593e-05, + "loss": 0.2594, + "step": 5662 + }, + { + "epoch": 2.37, + "learning_rate": 1.6503178322068718e-05, + "loss": 0.2271, + "step": 5663 + }, + { + "epoch": 2.37, + "learning_rate": 1.650189029050575e-05, + "loss": 0.2303, + "step": 5664 + }, + { + "epoch": 2.37, + "learning_rate": 1.6500602072051714e-05, + "loss": 0.2522, + "step": 5665 + }, + { + "epoch": 2.37, + "learning_rate": 1.6499313666743643e-05, + "loss": 0.2741, + "step": 5666 + }, + { + "epoch": 2.37, + "learning_rate": 1.6498025074618563e-05, + "loss": 0.2291, + "step": 5667 + }, + { + "epoch": 2.37, + "learning_rate": 1.6496736295713526e-05, + "loss": 0.2841, + "step": 5668 + }, + { + "epoch": 2.37, + "learning_rate": 1.649544733006556e-05, + "loss": 0.246, + "step": 5669 + }, + { + "epoch": 2.37, + "learning_rate": 1.649415817771173e-05, + "loss": 0.2773, + "step": 5670 + }, + { + "epoch": 2.37, + "learning_rate": 1.649286883868909e-05, + "loss": 0.2308, + "step": 5671 + }, + { + "epoch": 2.37, + "learning_rate": 1.649157931303469e-05, + "loss": 0.244, + "step": 5672 + }, + { + "epoch": 2.38, + "learning_rate": 1.6490289600785603e-05, + "loss": 0.2649, + "step": 5673 + }, + { + "epoch": 2.38, + "learning_rate": 1.6488999701978905e-05, + "loss": 0.2718, + "step": 5674 + }, + { + "epoch": 2.38, + "learning_rate": 1.6487709616651667e-05, + "loss": 0.2615, + "step": 5675 + }, + { + "epoch": 2.38, + "learning_rate": 1.6486419344840974e-05, + "loss": 0.2741, + "step": 5676 + }, + { + "epoch": 2.38, + "learning_rate": 1.6485128886583913e-05, + "loss": 0.2812, + "step": 5677 + }, + { + "epoch": 2.38, + "learning_rate": 1.6483838241917577e-05, + "loss": 0.2596, + "step": 5678 + }, + { + "epoch": 2.38, + "learning_rate": 1.6482547410879063e-05, + "loss": 0.2359, + "step": 5679 + }, + { + "epoch": 2.38, + "learning_rate": 1.6481256393505477e-05, + "loss": 0.2907, + "step": 5680 + }, + { + "epoch": 2.38, + "learning_rate": 1.6479965189833923e-05, + "loss": 0.2376, + "step": 5681 + }, + { + "epoch": 2.38, + "learning_rate": 1.6478673799901525e-05, + "loss": 0.2418, + "step": 5682 + }, + { + "epoch": 2.38, + "learning_rate": 1.6477382223745396e-05, + "loss": 0.2486, + "step": 5683 + }, + { + "epoch": 2.38, + "learning_rate": 1.6476090461402664e-05, + "loss": 0.2549, + "step": 5684 + }, + { + "epoch": 2.38, + "learning_rate": 1.647479851291046e-05, + "loss": 0.2417, + "step": 5685 + }, + { + "epoch": 2.38, + "learning_rate": 1.6473506378305914e-05, + "loss": 0.2823, + "step": 5686 + }, + { + "epoch": 2.38, + "learning_rate": 1.6472214057626175e-05, + "loss": 0.2542, + "step": 5687 + }, + { + "epoch": 2.38, + "learning_rate": 1.6470921550908385e-05, + "loss": 0.2549, + "step": 5688 + }, + { + "epoch": 2.38, + "learning_rate": 1.6469628858189698e-05, + "loss": 0.2689, + "step": 5689 + }, + { + "epoch": 2.38, + "learning_rate": 1.646833597950727e-05, + "loss": 0.2706, + "step": 5690 + }, + { + "epoch": 2.38, + "learning_rate": 1.6467042914898262e-05, + "loss": 0.2514, + "step": 5691 + }, + { + "epoch": 2.38, + "learning_rate": 1.6465749664399848e-05, + "loss": 0.2491, + "step": 5692 + }, + { + "epoch": 2.38, + "learning_rate": 1.6464456228049195e-05, + "loss": 0.2418, + "step": 5693 + }, + { + "epoch": 2.38, + "learning_rate": 1.6463162605883485e-05, + "loss": 0.2389, + "step": 5694 + }, + { + "epoch": 2.38, + "learning_rate": 1.64618687979399e-05, + "loss": 0.2294, + "step": 5695 + }, + { + "epoch": 2.38, + "learning_rate": 1.6460574804255634e-05, + "loss": 0.2445, + "step": 5696 + }, + { + "epoch": 2.39, + "learning_rate": 1.6459280624867876e-05, + "loss": 0.2614, + "step": 5697 + }, + { + "epoch": 2.39, + "learning_rate": 1.6457986259813827e-05, + "loss": 0.2873, + "step": 5698 + }, + { + "epoch": 2.39, + "learning_rate": 1.6456691709130694e-05, + "loss": 0.251, + "step": 5699 + }, + { + "epoch": 2.39, + "learning_rate": 1.6455396972855688e-05, + "loss": 0.2761, + "step": 5700 + }, + { + "epoch": 2.39, + "learning_rate": 1.6454102051026025e-05, + "loss": 0.2521, + "step": 5701 + }, + { + "epoch": 2.39, + "learning_rate": 1.6452806943678928e-05, + "loss": 0.2566, + "step": 5702 + }, + { + "epoch": 2.39, + "learning_rate": 1.645151165085162e-05, + "loss": 0.2531, + "step": 5703 + }, + { + "epoch": 2.39, + "learning_rate": 1.645021617258133e-05, + "loss": 0.2512, + "step": 5704 + }, + { + "epoch": 2.39, + "learning_rate": 1.6448920508905305e-05, + "loss": 0.239, + "step": 5705 + }, + { + "epoch": 2.39, + "learning_rate": 1.644762465986078e-05, + "loss": 0.2325, + "step": 5706 + }, + { + "epoch": 2.39, + "learning_rate": 1.6446328625485007e-05, + "loss": 0.2521, + "step": 5707 + }, + { + "epoch": 2.39, + "learning_rate": 1.644503240581524e-05, + "loss": 0.2613, + "step": 5708 + }, + { + "epoch": 2.39, + "learning_rate": 1.6443736000888732e-05, + "loss": 0.2535, + "step": 5709 + }, + { + "epoch": 2.39, + "learning_rate": 1.6442439410742753e-05, + "loss": 0.2512, + "step": 5710 + }, + { + "epoch": 2.39, + "learning_rate": 1.6441142635414568e-05, + "loss": 0.2478, + "step": 5711 + }, + { + "epoch": 2.39, + "learning_rate": 1.6439845674941458e-05, + "loss": 0.2688, + "step": 5712 + }, + { + "epoch": 2.39, + "learning_rate": 1.6438548529360698e-05, + "loss": 0.2423, + "step": 5713 + }, + { + "epoch": 2.39, + "learning_rate": 1.6437251198709567e-05, + "loss": 0.2904, + "step": 5714 + }, + { + "epoch": 2.39, + "learning_rate": 1.643595368302537e-05, + "loss": 0.2544, + "step": 5715 + }, + { + "epoch": 2.39, + "learning_rate": 1.643465598234539e-05, + "loss": 0.2706, + "step": 5716 + }, + { + "epoch": 2.39, + "learning_rate": 1.6433358096706937e-05, + "loss": 0.2799, + "step": 5717 + }, + { + "epoch": 2.39, + "learning_rate": 1.6432060026147313e-05, + "loss": 0.2739, + "step": 5718 + }, + { + "epoch": 2.39, + "learning_rate": 1.6430761770703836e-05, + "loss": 0.2693, + "step": 5719 + }, + { + "epoch": 2.39, + "learning_rate": 1.6429463330413812e-05, + "loss": 0.2574, + "step": 5720 + }, + { + "epoch": 2.4, + "learning_rate": 1.6428164705314572e-05, + "loss": 0.2597, + "step": 5721 + }, + { + "epoch": 2.4, + "learning_rate": 1.6426865895443446e-05, + "loss": 0.2549, + "step": 5722 + }, + { + "epoch": 2.4, + "learning_rate": 1.642556690083776e-05, + "loss": 0.2729, + "step": 5723 + }, + { + "epoch": 2.4, + "learning_rate": 1.6424267721534856e-05, + "loss": 0.2518, + "step": 5724 + }, + { + "epoch": 2.4, + "learning_rate": 1.6422968357572076e-05, + "loss": 0.2618, + "step": 5725 + }, + { + "epoch": 2.4, + "learning_rate": 1.642166880898677e-05, + "loss": 0.2654, + "step": 5726 + }, + { + "epoch": 2.4, + "learning_rate": 1.642036907581629e-05, + "loss": 0.2463, + "step": 5727 + }, + { + "epoch": 2.4, + "learning_rate": 1.6419069158098004e-05, + "loss": 0.2646, + "step": 5728 + }, + { + "epoch": 2.4, + "learning_rate": 1.6417769055869266e-05, + "loss": 0.2662, + "step": 5729 + }, + { + "epoch": 2.4, + "learning_rate": 1.641646876916745e-05, + "loss": 0.2697, + "step": 5730 + }, + { + "epoch": 2.4, + "learning_rate": 1.6415168298029943e-05, + "loss": 0.241, + "step": 5731 + }, + { + "epoch": 2.4, + "learning_rate": 1.6413867642494106e-05, + "loss": 0.2836, + "step": 5732 + }, + { + "epoch": 2.4, + "learning_rate": 1.641256680259734e-05, + "loss": 0.2531, + "step": 5733 + }, + { + "epoch": 2.4, + "learning_rate": 1.641126577837703e-05, + "loss": 0.2368, + "step": 5734 + }, + { + "epoch": 2.4, + "learning_rate": 1.640996456987058e-05, + "loss": 0.2644, + "step": 5735 + }, + { + "epoch": 2.4, + "learning_rate": 1.6408663177115382e-05, + "loss": 0.2916, + "step": 5736 + }, + { + "epoch": 2.4, + "learning_rate": 1.6407361600148845e-05, + "loss": 0.2774, + "step": 5737 + }, + { + "epoch": 2.4, + "learning_rate": 1.6406059839008387e-05, + "loss": 0.2308, + "step": 5738 + }, + { + "epoch": 2.4, + "learning_rate": 1.6404757893731424e-05, + "loss": 0.2639, + "step": 5739 + }, + { + "epoch": 2.4, + "learning_rate": 1.640345576435538e-05, + "loss": 0.2969, + "step": 5740 + }, + { + "epoch": 2.4, + "learning_rate": 1.640215345091768e-05, + "loss": 0.2689, + "step": 5741 + }, + { + "epoch": 2.4, + "learning_rate": 1.640085095345576e-05, + "loss": 0.2603, + "step": 5742 + }, + { + "epoch": 2.4, + "learning_rate": 1.6399548272007063e-05, + "loss": 0.2703, + "step": 5743 + }, + { + "epoch": 2.4, + "learning_rate": 1.6398245406609027e-05, + "loss": 0.2651, + "step": 5744 + }, + { + "epoch": 2.41, + "learning_rate": 1.6396942357299104e-05, + "loss": 0.2588, + "step": 5745 + }, + { + "epoch": 2.41, + "learning_rate": 1.639563912411475e-05, + "loss": 0.2697, + "step": 5746 + }, + { + "epoch": 2.41, + "learning_rate": 1.6394335707093423e-05, + "loss": 0.2693, + "step": 5747 + }, + { + "epoch": 2.41, + "learning_rate": 1.6393032106272594e-05, + "loss": 0.2228, + "step": 5748 + }, + { + "epoch": 2.41, + "learning_rate": 1.639172832168973e-05, + "loss": 0.2704, + "step": 5749 + }, + { + "epoch": 2.41, + "learning_rate": 1.639042435338231e-05, + "loss": 0.2572, + "step": 5750 + }, + { + "epoch": 2.41, + "learning_rate": 1.6389120201387806e-05, + "loss": 0.2515, + "step": 5751 + }, + { + "epoch": 2.41, + "learning_rate": 1.6387815865743712e-05, + "loss": 0.2549, + "step": 5752 + }, + { + "epoch": 2.41, + "learning_rate": 1.638651134648752e-05, + "loss": 0.244, + "step": 5753 + }, + { + "epoch": 2.41, + "learning_rate": 1.638520664365673e-05, + "loss": 0.272, + "step": 5754 + }, + { + "epoch": 2.41, + "learning_rate": 1.638390175728884e-05, + "loss": 0.2539, + "step": 5755 + }, + { + "epoch": 2.41, + "learning_rate": 1.638259668742136e-05, + "loss": 0.251, + "step": 5756 + }, + { + "epoch": 2.41, + "learning_rate": 1.63812914340918e-05, + "loss": 0.2558, + "step": 5757 + }, + { + "epoch": 2.41, + "learning_rate": 1.6379985997337678e-05, + "loss": 0.2609, + "step": 5758 + }, + { + "epoch": 2.41, + "learning_rate": 1.6378680377196526e-05, + "loss": 0.2463, + "step": 5759 + }, + { + "epoch": 2.41, + "learning_rate": 1.6377374573705863e-05, + "loss": 0.2639, + "step": 5760 + }, + { + "epoch": 2.41, + "learning_rate": 1.6376068586903227e-05, + "loss": 0.272, + "step": 5761 + }, + { + "epoch": 2.41, + "learning_rate": 1.6374762416826157e-05, + "loss": 0.2646, + "step": 5762 + }, + { + "epoch": 2.41, + "learning_rate": 1.63734560635122e-05, + "loss": 0.251, + "step": 5763 + }, + { + "epoch": 2.41, + "learning_rate": 1.6372149526998905e-05, + "loss": 0.253, + "step": 5764 + }, + { + "epoch": 2.41, + "learning_rate": 1.6370842807323823e-05, + "loss": 0.243, + "step": 5765 + }, + { + "epoch": 2.41, + "learning_rate": 1.636953590452452e-05, + "loss": 0.2471, + "step": 5766 + }, + { + "epoch": 2.41, + "learning_rate": 1.636822881863856e-05, + "loss": 0.2734, + "step": 5767 + }, + { + "epoch": 2.41, + "learning_rate": 1.6366921549703513e-05, + "loss": 0.2282, + "step": 5768 + }, + { + "epoch": 2.42, + "learning_rate": 1.6365614097756956e-05, + "loss": 0.257, + "step": 5769 + }, + { + "epoch": 2.42, + "learning_rate": 1.636430646283647e-05, + "loss": 0.2875, + "step": 5770 + }, + { + "epoch": 2.42, + "learning_rate": 1.6362998644979645e-05, + "loss": 0.2659, + "step": 5771 + }, + { + "epoch": 2.42, + "learning_rate": 1.6361690644224066e-05, + "loss": 0.2742, + "step": 5772 + }, + { + "epoch": 2.42, + "learning_rate": 1.636038246060734e-05, + "loss": 0.2699, + "step": 5773 + }, + { + "epoch": 2.42, + "learning_rate": 1.635907409416706e-05, + "loss": 0.2658, + "step": 5774 + }, + { + "epoch": 2.42, + "learning_rate": 1.6357765544940837e-05, + "loss": 0.2769, + "step": 5775 + }, + { + "epoch": 2.42, + "learning_rate": 1.635645681296629e-05, + "loss": 0.2656, + "step": 5776 + }, + { + "epoch": 2.42, + "learning_rate": 1.6355147898281026e-05, + "loss": 0.2624, + "step": 5777 + }, + { + "epoch": 2.42, + "learning_rate": 1.6353838800922682e-05, + "loss": 0.2353, + "step": 5778 + }, + { + "epoch": 2.42, + "learning_rate": 1.6352529520928877e-05, + "loss": 0.2584, + "step": 5779 + }, + { + "epoch": 2.42, + "learning_rate": 1.6351220058337248e-05, + "loss": 0.2717, + "step": 5780 + }, + { + "epoch": 2.42, + "learning_rate": 1.6349910413185438e-05, + "loss": 0.2829, + "step": 5781 + }, + { + "epoch": 2.42, + "learning_rate": 1.6348600585511082e-05, + "loss": 0.2519, + "step": 5782 + }, + { + "epoch": 2.42, + "learning_rate": 1.634729057535184e-05, + "loss": 0.2674, + "step": 5783 + }, + { + "epoch": 2.42, + "learning_rate": 1.634598038274536e-05, + "loss": 0.2715, + "step": 5784 + }, + { + "epoch": 2.42, + "learning_rate": 1.6344670007729305e-05, + "loss": 0.2771, + "step": 5785 + }, + { + "epoch": 2.42, + "learning_rate": 1.6343359450341344e-05, + "loss": 0.2167, + "step": 5786 + }, + { + "epoch": 2.42, + "learning_rate": 1.6342048710619143e-05, + "loss": 0.2417, + "step": 5787 + }, + { + "epoch": 2.42, + "learning_rate": 1.6340737788600378e-05, + "loss": 0.2649, + "step": 5788 + }, + { + "epoch": 2.42, + "learning_rate": 1.6339426684322735e-05, + "loss": 0.2508, + "step": 5789 + }, + { + "epoch": 2.42, + "learning_rate": 1.6338115397823897e-05, + "loss": 0.2406, + "step": 5790 + }, + { + "epoch": 2.42, + "learning_rate": 1.6336803929141557e-05, + "loss": 0.2676, + "step": 5791 + }, + { + "epoch": 2.42, + "learning_rate": 1.6335492278313405e-05, + "loss": 0.2332, + "step": 5792 + }, + { + "epoch": 2.43, + "learning_rate": 1.6334180445377156e-05, + "loss": 0.251, + "step": 5793 + }, + { + "epoch": 2.43, + "learning_rate": 1.6332868430370508e-05, + "loss": 0.2471, + "step": 5794 + }, + { + "epoch": 2.43, + "learning_rate": 1.6331556233331178e-05, + "loss": 0.2658, + "step": 5795 + }, + { + "epoch": 2.43, + "learning_rate": 1.633024385429688e-05, + "loss": 0.2872, + "step": 5796 + }, + { + "epoch": 2.43, + "learning_rate": 1.6328931293305345e-05, + "loss": 0.2564, + "step": 5797 + }, + { + "epoch": 2.43, + "learning_rate": 1.632761855039429e-05, + "loss": 0.2626, + "step": 5798 + }, + { + "epoch": 2.43, + "learning_rate": 1.632630562560146e-05, + "loss": 0.2719, + "step": 5799 + }, + { + "epoch": 2.43, + "learning_rate": 1.6324992518964586e-05, + "loss": 0.248, + "step": 5800 + }, + { + "epoch": 2.43, + "learning_rate": 1.6323679230521413e-05, + "loss": 0.2723, + "step": 5801 + }, + { + "epoch": 2.43, + "learning_rate": 1.6322365760309698e-05, + "loss": 0.2574, + "step": 5802 + }, + { + "epoch": 2.43, + "learning_rate": 1.6321052108367187e-05, + "loss": 0.2379, + "step": 5803 + }, + { + "epoch": 2.43, + "learning_rate": 1.631973827473164e-05, + "loss": 0.2748, + "step": 5804 + }, + { + "epoch": 2.43, + "learning_rate": 1.6318424259440827e-05, + "loss": 0.2571, + "step": 5805 + }, + { + "epoch": 2.43, + "learning_rate": 1.631711006253251e-05, + "loss": 0.266, + "step": 5806 + }, + { + "epoch": 2.43, + "learning_rate": 1.6315795684044476e-05, + "loss": 0.2561, + "step": 5807 + }, + { + "epoch": 2.43, + "learning_rate": 1.63144811240145e-05, + "loss": 0.2786, + "step": 5808 + }, + { + "epoch": 2.43, + "learning_rate": 1.6313166382480362e-05, + "loss": 0.2648, + "step": 5809 + }, + { + "epoch": 2.43, + "learning_rate": 1.631185145947986e-05, + "loss": 0.2931, + "step": 5810 + }, + { + "epoch": 2.43, + "learning_rate": 1.631053635505079e-05, + "loss": 0.2583, + "step": 5811 + }, + { + "epoch": 2.43, + "learning_rate": 1.6309221069230954e-05, + "loss": 0.2413, + "step": 5812 + }, + { + "epoch": 2.43, + "learning_rate": 1.6307905602058158e-05, + "loss": 0.2587, + "step": 5813 + }, + { + "epoch": 2.43, + "learning_rate": 1.6306589953570207e-05, + "loss": 0.2623, + "step": 5814 + }, + { + "epoch": 2.43, + "learning_rate": 1.6305274123804934e-05, + "loss": 0.2557, + "step": 5815 + }, + { + "epoch": 2.44, + "learning_rate": 1.6303958112800144e-05, + "loss": 0.2457, + "step": 5816 + }, + { + "epoch": 2.44, + "learning_rate": 1.630264192059367e-05, + "loss": 0.2665, + "step": 5817 + }, + { + "epoch": 2.44, + "learning_rate": 1.6301325547223353e-05, + "loss": 0.2545, + "step": 5818 + }, + { + "epoch": 2.44, + "learning_rate": 1.630000899272702e-05, + "loss": 0.2427, + "step": 5819 + }, + { + "epoch": 2.44, + "learning_rate": 1.6298692257142525e-05, + "loss": 0.2687, + "step": 5820 + }, + { + "epoch": 2.44, + "learning_rate": 1.6297375340507705e-05, + "loss": 0.2529, + "step": 5821 + }, + { + "epoch": 2.44, + "learning_rate": 1.629605824286042e-05, + "loss": 0.2595, + "step": 5822 + }, + { + "epoch": 2.44, + "learning_rate": 1.629474096423853e-05, + "loss": 0.2597, + "step": 5823 + }, + { + "epoch": 2.44, + "learning_rate": 1.6293423504679898e-05, + "loss": 0.2469, + "step": 5824 + }, + { + "epoch": 2.44, + "learning_rate": 1.6292105864222386e-05, + "loss": 0.2507, + "step": 5825 + }, + { + "epoch": 2.44, + "learning_rate": 1.629078804290388e-05, + "loss": 0.2638, + "step": 5826 + }, + { + "epoch": 2.44, + "learning_rate": 1.628947004076225e-05, + "loss": 0.2452, + "step": 5827 + }, + { + "epoch": 2.44, + "learning_rate": 1.628815185783539e-05, + "loss": 0.2736, + "step": 5828 + }, + { + "epoch": 2.44, + "learning_rate": 1.628683349416118e-05, + "loss": 0.2595, + "step": 5829 + }, + { + "epoch": 2.44, + "learning_rate": 1.6285514949777522e-05, + "loss": 0.2602, + "step": 5830 + }, + { + "epoch": 2.44, + "learning_rate": 1.628419622472231e-05, + "loss": 0.2559, + "step": 5831 + }, + { + "epoch": 2.44, + "learning_rate": 1.6282877319033462e-05, + "loss": 0.2772, + "step": 5832 + }, + { + "epoch": 2.44, + "learning_rate": 1.6281558232748874e-05, + "loss": 0.2583, + "step": 5833 + }, + { + "epoch": 2.44, + "learning_rate": 1.6280238965906474e-05, + "loss": 0.2928, + "step": 5834 + }, + { + "epoch": 2.44, + "learning_rate": 1.6278919518544175e-05, + "loss": 0.2403, + "step": 5835 + }, + { + "epoch": 2.44, + "learning_rate": 1.627759989069991e-05, + "loss": 0.2579, + "step": 5836 + }, + { + "epoch": 2.44, + "learning_rate": 1.6276280082411604e-05, + "loss": 0.2968, + "step": 5837 + }, + { + "epoch": 2.44, + "learning_rate": 1.6274960093717192e-05, + "loss": 0.2501, + "step": 5838 + }, + { + "epoch": 2.44, + "learning_rate": 1.6273639924654626e-05, + "loss": 0.2598, + "step": 5839 + }, + { + "epoch": 2.45, + "learning_rate": 1.627231957526185e-05, + "loss": 0.2334, + "step": 5840 + }, + { + "epoch": 2.45, + "learning_rate": 1.6270999045576808e-05, + "loss": 0.2504, + "step": 5841 + }, + { + "epoch": 2.45, + "learning_rate": 1.6269678335637464e-05, + "loss": 0.2664, + "step": 5842 + }, + { + "epoch": 2.45, + "learning_rate": 1.626835744548178e-05, + "loss": 0.2308, + "step": 5843 + }, + { + "epoch": 2.45, + "learning_rate": 1.6267036375147728e-05, + "loss": 0.2584, + "step": 5844 + }, + { + "epoch": 2.45, + "learning_rate": 1.6265715124673272e-05, + "loss": 0.2493, + "step": 5845 + }, + { + "epoch": 2.45, + "learning_rate": 1.6264393694096396e-05, + "loss": 0.2382, + "step": 5846 + }, + { + "epoch": 2.45, + "learning_rate": 1.6263072083455084e-05, + "loss": 0.2394, + "step": 5847 + }, + { + "epoch": 2.45, + "learning_rate": 1.626175029278732e-05, + "loss": 0.2658, + "step": 5848 + }, + { + "epoch": 2.45, + "learning_rate": 1.62604283221311e-05, + "loss": 0.2416, + "step": 5849 + }, + { + "epoch": 2.45, + "learning_rate": 1.6259106171524424e-05, + "loss": 0.2642, + "step": 5850 + }, + { + "epoch": 2.45, + "learning_rate": 1.6257783841005297e-05, + "loss": 0.2426, + "step": 5851 + }, + { + "epoch": 2.45, + "learning_rate": 1.6256461330611727e-05, + "loss": 0.2554, + "step": 5852 + }, + { + "epoch": 2.45, + "learning_rate": 1.6255138640381723e-05, + "loss": 0.2516, + "step": 5853 + }, + { + "epoch": 2.45, + "learning_rate": 1.6253815770353316e-05, + "loss": 0.2651, + "step": 5854 + }, + { + "epoch": 2.45, + "learning_rate": 1.625249272056452e-05, + "loss": 0.2797, + "step": 5855 + }, + { + "epoch": 2.45, + "learning_rate": 1.625116949105337e-05, + "loss": 0.2711, + "step": 5856 + }, + { + "epoch": 2.45, + "learning_rate": 1.62498460818579e-05, + "loss": 0.2555, + "step": 5857 + }, + { + "epoch": 2.45, + "learning_rate": 1.6248522493016153e-05, + "loss": 0.2479, + "step": 5858 + }, + { + "epoch": 2.45, + "learning_rate": 1.624719872456617e-05, + "loss": 0.2662, + "step": 5859 + }, + { + "epoch": 2.45, + "learning_rate": 1.6245874776546005e-05, + "loss": 0.2627, + "step": 5860 + }, + { + "epoch": 2.45, + "learning_rate": 1.624455064899371e-05, + "loss": 0.247, + "step": 5861 + }, + { + "epoch": 2.45, + "learning_rate": 1.6243226341947348e-05, + "loss": 0.266, + "step": 5862 + }, + { + "epoch": 2.45, + "learning_rate": 1.624190185544498e-05, + "loss": 0.25, + "step": 5863 + }, + { + "epoch": 2.46, + "learning_rate": 1.6240577189524688e-05, + "loss": 0.2389, + "step": 5864 + }, + { + "epoch": 2.46, + "learning_rate": 1.6239252344224545e-05, + "loss": 0.2448, + "step": 5865 + }, + { + "epoch": 2.46, + "learning_rate": 1.6237927319582624e-05, + "loss": 0.2697, + "step": 5866 + }, + { + "epoch": 2.46, + "learning_rate": 1.623660211563702e-05, + "loss": 0.2607, + "step": 5867 + }, + { + "epoch": 2.46, + "learning_rate": 1.6235276732425823e-05, + "loss": 0.2711, + "step": 5868 + }, + { + "epoch": 2.46, + "learning_rate": 1.623395116998713e-05, + "loss": 0.2854, + "step": 5869 + }, + { + "epoch": 2.46, + "learning_rate": 1.6232625428359042e-05, + "loss": 0.2689, + "step": 5870 + }, + { + "epoch": 2.46, + "learning_rate": 1.6231299507579668e-05, + "loss": 0.2591, + "step": 5871 + }, + { + "epoch": 2.46, + "learning_rate": 1.6229973407687117e-05, + "loss": 0.2603, + "step": 5872 + }, + { + "epoch": 2.46, + "learning_rate": 1.622864712871951e-05, + "loss": 0.2572, + "step": 5873 + }, + { + "epoch": 2.46, + "learning_rate": 1.6227320670714966e-05, + "loss": 0.2879, + "step": 5874 + }, + { + "epoch": 2.46, + "learning_rate": 1.6225994033711617e-05, + "loss": 0.2058, + "step": 5875 + }, + { + "epoch": 2.46, + "learning_rate": 1.6224667217747594e-05, + "loss": 0.26, + "step": 5876 + }, + { + "epoch": 2.46, + "learning_rate": 1.6223340222861037e-05, + "loss": 0.2241, + "step": 5877 + }, + { + "epoch": 2.46, + "learning_rate": 1.6222013049090086e-05, + "loss": 0.2499, + "step": 5878 + }, + { + "epoch": 2.46, + "learning_rate": 1.622068569647289e-05, + "loss": 0.2711, + "step": 5879 + }, + { + "epoch": 2.46, + "learning_rate": 1.6219358165047606e-05, + "loss": 0.2468, + "step": 5880 + }, + { + "epoch": 2.46, + "learning_rate": 1.6218030454852395e-05, + "loss": 0.2711, + "step": 5881 + }, + { + "epoch": 2.46, + "learning_rate": 1.621670256592541e-05, + "loss": 0.2464, + "step": 5882 + }, + { + "epoch": 2.46, + "learning_rate": 1.6215374498304827e-05, + "loss": 0.2739, + "step": 5883 + }, + { + "epoch": 2.46, + "learning_rate": 1.6214046252028823e-05, + "loss": 0.2529, + "step": 5884 + }, + { + "epoch": 2.46, + "learning_rate": 1.621271782713557e-05, + "loss": 0.2508, + "step": 5885 + }, + { + "epoch": 2.46, + "learning_rate": 1.621138922366326e-05, + "loss": 0.245, + "step": 5886 + }, + { + "epoch": 2.46, + "learning_rate": 1.6210060441650073e-05, + "loss": 0.2693, + "step": 5887 + }, + { + "epoch": 2.47, + "learning_rate": 1.6208731481134214e-05, + "loss": 0.2377, + "step": 5888 + }, + { + "epoch": 2.47, + "learning_rate": 1.6207402342153876e-05, + "loss": 0.2442, + "step": 5889 + }, + { + "epoch": 2.47, + "learning_rate": 1.6206073024747264e-05, + "loss": 0.2673, + "step": 5890 + }, + { + "epoch": 2.47, + "learning_rate": 1.6204743528952592e-05, + "loss": 0.2758, + "step": 5891 + }, + { + "epoch": 2.47, + "learning_rate": 1.6203413854808076e-05, + "loss": 0.2612, + "step": 5892 + }, + { + "epoch": 2.47, + "learning_rate": 1.620208400235193e-05, + "loss": 0.2591, + "step": 5893 + }, + { + "epoch": 2.47, + "learning_rate": 1.6200753971622387e-05, + "loss": 0.2512, + "step": 5894 + }, + { + "epoch": 2.47, + "learning_rate": 1.619942376265767e-05, + "loss": 0.2587, + "step": 5895 + }, + { + "epoch": 2.47, + "learning_rate": 1.619809337549602e-05, + "loss": 0.2153, + "step": 5896 + }, + { + "epoch": 2.47, + "learning_rate": 1.6196762810175674e-05, + "loss": 0.2638, + "step": 5897 + }, + { + "epoch": 2.47, + "learning_rate": 1.6195432066734878e-05, + "loss": 0.2553, + "step": 5898 + }, + { + "epoch": 2.47, + "learning_rate": 1.619410114521189e-05, + "loss": 0.2529, + "step": 5899 + }, + { + "epoch": 2.47, + "learning_rate": 1.619277004564496e-05, + "loss": 0.2724, + "step": 5900 + }, + { + "epoch": 2.47, + "learning_rate": 1.619143876807235e-05, + "loss": 0.2458, + "step": 5901 + }, + { + "epoch": 2.47, + "learning_rate": 1.6190107312532327e-05, + "loss": 0.2857, + "step": 5902 + }, + { + "epoch": 2.47, + "learning_rate": 1.618877567906316e-05, + "loss": 0.2431, + "step": 5903 + }, + { + "epoch": 2.47, + "learning_rate": 1.618744386770313e-05, + "loss": 0.2515, + "step": 5904 + }, + { + "epoch": 2.47, + "learning_rate": 1.6186111878490515e-05, + "loss": 0.2522, + "step": 5905 + }, + { + "epoch": 2.47, + "learning_rate": 1.6184779711463607e-05, + "loss": 0.2592, + "step": 5906 + }, + { + "epoch": 2.47, + "learning_rate": 1.618344736666069e-05, + "loss": 0.2663, + "step": 5907 + }, + { + "epoch": 2.47, + "learning_rate": 1.618211484412007e-05, + "loss": 0.2633, + "step": 5908 + }, + { + "epoch": 2.47, + "learning_rate": 1.6180782143880042e-05, + "loss": 0.2422, + "step": 5909 + }, + { + "epoch": 2.47, + "learning_rate": 1.6179449265978913e-05, + "loss": 0.2889, + "step": 5910 + }, + { + "epoch": 2.47, + "learning_rate": 1.6178116210455006e-05, + "loss": 0.3035, + "step": 5911 + }, + { + "epoch": 2.48, + "learning_rate": 1.6176782977346626e-05, + "loss": 0.2714, + "step": 5912 + }, + { + "epoch": 2.48, + "learning_rate": 1.61754495666921e-05, + "loss": 0.254, + "step": 5913 + }, + { + "epoch": 2.48, + "learning_rate": 1.617411597852976e-05, + "loss": 0.2454, + "step": 5914 + }, + { + "epoch": 2.48, + "learning_rate": 1.617278221289793e-05, + "loss": 0.2544, + "step": 5915 + }, + { + "epoch": 2.48, + "learning_rate": 1.6171448269834956e-05, + "loss": 0.2736, + "step": 5916 + }, + { + "epoch": 2.48, + "learning_rate": 1.6170114149379178e-05, + "loss": 0.2428, + "step": 5917 + }, + { + "epoch": 2.48, + "learning_rate": 1.6168779851568947e-05, + "loss": 0.2423, + "step": 5918 + }, + { + "epoch": 2.48, + "learning_rate": 1.6167445376442604e-05, + "loss": 0.2222, + "step": 5919 + }, + { + "epoch": 2.48, + "learning_rate": 1.6166110724038524e-05, + "loss": 0.2606, + "step": 5920 + }, + { + "epoch": 2.48, + "learning_rate": 1.616477589439506e-05, + "loss": 0.2681, + "step": 5921 + }, + { + "epoch": 2.48, + "learning_rate": 1.6163440887550587e-05, + "loss": 0.2595, + "step": 5922 + }, + { + "epoch": 2.48, + "learning_rate": 1.616210570354347e-05, + "loss": 0.2684, + "step": 5923 + }, + { + "epoch": 2.48, + "learning_rate": 1.6160770342412097e-05, + "loss": 0.2561, + "step": 5924 + }, + { + "epoch": 2.48, + "learning_rate": 1.6159434804194845e-05, + "loss": 0.2489, + "step": 5925 + }, + { + "epoch": 2.48, + "learning_rate": 1.615809908893011e-05, + "loss": 0.2594, + "step": 5926 + }, + { + "epoch": 2.48, + "learning_rate": 1.6156763196656276e-05, + "loss": 0.2583, + "step": 5927 + }, + { + "epoch": 2.48, + "learning_rate": 1.6155427127411748e-05, + "loss": 0.2552, + "step": 5928 + }, + { + "epoch": 2.48, + "learning_rate": 1.615409088123493e-05, + "loss": 0.2666, + "step": 5929 + }, + { + "epoch": 2.48, + "learning_rate": 1.615275445816423e-05, + "loss": 0.2609, + "step": 5930 + }, + { + "epoch": 2.48, + "learning_rate": 1.6151417858238064e-05, + "loss": 0.2475, + "step": 5931 + }, + { + "epoch": 2.48, + "learning_rate": 1.615008108149485e-05, + "loss": 0.2644, + "step": 5932 + }, + { + "epoch": 2.48, + "learning_rate": 1.6148744127973014e-05, + "loss": 0.2486, + "step": 5933 + }, + { + "epoch": 2.48, + "learning_rate": 1.614740699771098e-05, + "loss": 0.2463, + "step": 5934 + }, + { + "epoch": 2.48, + "learning_rate": 1.614606969074719e-05, + "loss": 0.2454, + "step": 5935 + }, + { + "epoch": 2.49, + "learning_rate": 1.6144732207120083e-05, + "loss": 0.2456, + "step": 5936 + }, + { + "epoch": 2.49, + "learning_rate": 1.61433945468681e-05, + "loss": 0.2615, + "step": 5937 + }, + { + "epoch": 2.49, + "learning_rate": 1.614205671002969e-05, + "loss": 0.2378, + "step": 5938 + }, + { + "epoch": 2.49, + "learning_rate": 1.6140718696643313e-05, + "loss": 0.2868, + "step": 5939 + }, + { + "epoch": 2.49, + "learning_rate": 1.6139380506747424e-05, + "loss": 0.2496, + "step": 5940 + }, + { + "epoch": 2.49, + "learning_rate": 1.6138042140380493e-05, + "loss": 0.2343, + "step": 5941 + }, + { + "epoch": 2.49, + "learning_rate": 1.6136703597580984e-05, + "loss": 0.2396, + "step": 5942 + }, + { + "epoch": 2.49, + "learning_rate": 1.613536487838738e-05, + "loss": 0.2665, + "step": 5943 + }, + { + "epoch": 2.49, + "learning_rate": 1.6134025982838153e-05, + "loss": 0.2772, + "step": 5944 + }, + { + "epoch": 2.49, + "learning_rate": 1.6132686910971792e-05, + "loss": 0.2544, + "step": 5945 + }, + { + "epoch": 2.49, + "learning_rate": 1.6131347662826793e-05, + "loss": 0.2489, + "step": 5946 + }, + { + "epoch": 2.49, + "learning_rate": 1.613000823844164e-05, + "loss": 0.2798, + "step": 5947 + }, + { + "epoch": 2.49, + "learning_rate": 1.6128668637854847e-05, + "loss": 0.2367, + "step": 5948 + }, + { + "epoch": 2.49, + "learning_rate": 1.6127328861104906e-05, + "loss": 0.2688, + "step": 5949 + }, + { + "epoch": 2.49, + "learning_rate": 1.6125988908230337e-05, + "loss": 0.2514, + "step": 5950 + }, + { + "epoch": 2.49, + "learning_rate": 1.6124648779269654e-05, + "loss": 0.2594, + "step": 5951 + }, + { + "epoch": 2.49, + "learning_rate": 1.6123308474261374e-05, + "loss": 0.2542, + "step": 5952 + }, + { + "epoch": 2.49, + "learning_rate": 1.6121967993244026e-05, + "loss": 0.2548, + "step": 5953 + }, + { + "epoch": 2.49, + "learning_rate": 1.6120627336256142e-05, + "loss": 0.2928, + "step": 5954 + }, + { + "epoch": 2.49, + "learning_rate": 1.6119286503336256e-05, + "loss": 0.2584, + "step": 5955 + }, + { + "epoch": 2.49, + "learning_rate": 1.611794549452291e-05, + "loss": 0.2732, + "step": 5956 + }, + { + "epoch": 2.49, + "learning_rate": 1.611660430985465e-05, + "loss": 0.2218, + "step": 5957 + }, + { + "epoch": 2.49, + "learning_rate": 1.6115262949370025e-05, + "loss": 0.2893, + "step": 5958 + }, + { + "epoch": 2.49, + "learning_rate": 1.6113921413107597e-05, + "loss": 0.2684, + "step": 5959 + }, + { + "epoch": 2.5, + "learning_rate": 1.6112579701105923e-05, + "loss": 0.2812, + "step": 5960 + }, + { + "epoch": 2.5, + "learning_rate": 1.6111237813403568e-05, + "loss": 0.2759, + "step": 5961 + }, + { + "epoch": 2.5, + "learning_rate": 1.61098957500391e-05, + "loss": 0.285, + "step": 5962 + }, + { + "epoch": 2.5, + "learning_rate": 1.6108553511051106e-05, + "loss": 0.2418, + "step": 5963 + }, + { + "epoch": 2.5, + "learning_rate": 1.610721109647816e-05, + "loss": 0.2605, + "step": 5964 + }, + { + "epoch": 2.5, + "learning_rate": 1.610586850635885e-05, + "loss": 0.263, + "step": 5965 + }, + { + "epoch": 2.5, + "learning_rate": 1.6104525740731772e-05, + "loss": 0.28, + "step": 5966 + }, + { + "epoch": 2.5, + "learning_rate": 1.6103182799635514e-05, + "loss": 0.246, + "step": 5967 + }, + { + "epoch": 2.5, + "learning_rate": 1.6101839683108686e-05, + "loss": 0.2725, + "step": 5968 + }, + { + "epoch": 2.5, + "learning_rate": 1.610049639118989e-05, + "loss": 0.2355, + "step": 5969 + }, + { + "epoch": 2.5, + "learning_rate": 1.6099152923917737e-05, + "loss": 0.2961, + "step": 5970 + }, + { + "epoch": 2.5, + "learning_rate": 1.6097809281330848e-05, + "loss": 0.2723, + "step": 5971 + }, + { + "epoch": 2.5, + "learning_rate": 1.609646546346784e-05, + "loss": 0.2695, + "step": 5972 + }, + { + "epoch": 2.5, + "learning_rate": 1.6095121470367345e-05, + "loss": 0.287, + "step": 5973 + }, + { + "epoch": 2.5, + "learning_rate": 1.609377730206799e-05, + "loss": 0.2501, + "step": 5974 + }, + { + "epoch": 2.5, + "learning_rate": 1.6092432958608413e-05, + "loss": 0.2542, + "step": 5975 + }, + { + "epoch": 2.5, + "learning_rate": 1.609108844002726e-05, + "loss": 0.259, + "step": 5976 + }, + { + "epoch": 2.5, + "learning_rate": 1.608974374636317e-05, + "loss": 0.2713, + "step": 5977 + }, + { + "epoch": 2.5, + "learning_rate": 1.6088398877654806e-05, + "loss": 0.2765, + "step": 5978 + }, + { + "epoch": 2.5, + "learning_rate": 1.6087053833940815e-05, + "loss": 0.2254, + "step": 5979 + }, + { + "epoch": 2.5, + "learning_rate": 1.6085708615259863e-05, + "loss": 0.2434, + "step": 5980 + }, + { + "epoch": 2.5, + "learning_rate": 1.608436322165062e-05, + "loss": 0.2638, + "step": 5981 + }, + { + "epoch": 2.5, + "learning_rate": 1.6083017653151752e-05, + "loss": 0.2749, + "step": 5982 + }, + { + "epoch": 2.5, + "learning_rate": 1.6081671909801944e-05, + "loss": 0.2938, + "step": 5983 + }, + { + "epoch": 2.51, + "learning_rate": 1.6080325991639866e-05, + "loss": 0.2419, + "step": 5984 + }, + { + "epoch": 2.51, + "learning_rate": 1.6078979898704223e-05, + "loss": 0.2736, + "step": 5985 + }, + { + "epoch": 2.51, + "learning_rate": 1.607763363103369e-05, + "loss": 0.2498, + "step": 5986 + }, + { + "epoch": 2.51, + "learning_rate": 1.6076287188666972e-05, + "loss": 0.2468, + "step": 5987 + }, + { + "epoch": 2.51, + "learning_rate": 1.6074940571642773e-05, + "loss": 0.2543, + "step": 5988 + }, + { + "epoch": 2.51, + "learning_rate": 1.6073593779999795e-05, + "loss": 0.2549, + "step": 5989 + }, + { + "epoch": 2.51, + "learning_rate": 1.6072246813776758e-05, + "loss": 0.2918, + "step": 5990 + }, + { + "epoch": 2.51, + "learning_rate": 1.6070899673012372e-05, + "loss": 0.2755, + "step": 5991 + }, + { + "epoch": 2.51, + "learning_rate": 1.6069552357745363e-05, + "loss": 0.2557, + "step": 5992 + }, + { + "epoch": 2.51, + "learning_rate": 1.6068204868014455e-05, + "loss": 0.2732, + "step": 5993 + }, + { + "epoch": 2.51, + "learning_rate": 1.6066857203858387e-05, + "loss": 0.2549, + "step": 5994 + }, + { + "epoch": 2.51, + "learning_rate": 1.6065509365315893e-05, + "loss": 0.2497, + "step": 5995 + }, + { + "epoch": 2.51, + "learning_rate": 1.6064161352425708e-05, + "loss": 0.2364, + "step": 5996 + }, + { + "epoch": 2.51, + "learning_rate": 1.6062813165226594e-05, + "loss": 0.2453, + "step": 5997 + }, + { + "epoch": 2.51, + "learning_rate": 1.606146480375729e-05, + "loss": 0.2279, + "step": 5998 + }, + { + "epoch": 2.51, + "learning_rate": 1.606011626805656e-05, + "loss": 0.2577, + "step": 5999 + }, + { + "epoch": 2.51, + "learning_rate": 1.6058767558163168e-05, + "loss": 0.2731, + "step": 6000 + }, + { + "epoch": 2.51, + "learning_rate": 1.6057418674115876e-05, + "loss": 0.2782, + "step": 6001 + }, + { + "epoch": 2.51, + "learning_rate": 1.605606961595346e-05, + "loss": 0.2451, + "step": 6002 + }, + { + "epoch": 2.51, + "learning_rate": 1.60547203837147e-05, + "loss": 0.2513, + "step": 6003 + }, + { + "epoch": 2.51, + "learning_rate": 1.605337097743837e-05, + "loss": 0.2531, + "step": 6004 + }, + { + "epoch": 2.51, + "learning_rate": 1.605202139716327e-05, + "loss": 0.2754, + "step": 6005 + }, + { + "epoch": 2.51, + "learning_rate": 1.605067164292818e-05, + "loss": 0.2522, + "step": 6006 + }, + { + "epoch": 2.51, + "learning_rate": 1.6049321714771907e-05, + "loss": 0.2427, + "step": 6007 + }, + { + "epoch": 2.52, + "learning_rate": 1.6047971612733246e-05, + "loss": 0.2295, + "step": 6008 + }, + { + "epoch": 2.52, + "learning_rate": 1.6046621336851007e-05, + "loss": 0.272, + "step": 6009 + }, + { + "epoch": 2.52, + "learning_rate": 1.6045270887164005e-05, + "loss": 0.2655, + "step": 6010 + }, + { + "epoch": 2.52, + "learning_rate": 1.6043920263711056e-05, + "loss": 0.2788, + "step": 6011 + }, + { + "epoch": 2.52, + "learning_rate": 1.6042569466530985e-05, + "loss": 0.2805, + "step": 6012 + }, + { + "epoch": 2.52, + "learning_rate": 1.604121849566261e-05, + "loss": 0.2553, + "step": 6013 + }, + { + "epoch": 2.52, + "learning_rate": 1.6039867351144778e-05, + "loss": 0.2407, + "step": 6014 + }, + { + "epoch": 2.52, + "learning_rate": 1.6038516033016314e-05, + "loss": 0.2728, + "step": 6015 + }, + { + "epoch": 2.52, + "learning_rate": 1.6037164541316067e-05, + "loss": 0.2442, + "step": 6016 + }, + { + "epoch": 2.52, + "learning_rate": 1.603581287608288e-05, + "loss": 0.2559, + "step": 6017 + }, + { + "epoch": 2.52, + "learning_rate": 1.603446103735561e-05, + "loss": 0.2662, + "step": 6018 + }, + { + "epoch": 2.52, + "learning_rate": 1.6033109025173113e-05, + "loss": 0.2541, + "step": 6019 + }, + { + "epoch": 2.52, + "learning_rate": 1.6031756839574247e-05, + "loss": 0.2909, + "step": 6020 + }, + { + "epoch": 2.52, + "learning_rate": 1.6030404480597888e-05, + "loss": 0.2366, + "step": 6021 + }, + { + "epoch": 2.52, + "learning_rate": 1.60290519482829e-05, + "loss": 0.2578, + "step": 6022 + }, + { + "epoch": 2.52, + "learning_rate": 1.6027699242668166e-05, + "loss": 0.2686, + "step": 6023 + }, + { + "epoch": 2.52, + "learning_rate": 1.6026346363792565e-05, + "loss": 0.2634, + "step": 6024 + }, + { + "epoch": 2.52, + "learning_rate": 1.6024993311694986e-05, + "loss": 0.267, + "step": 6025 + }, + { + "epoch": 2.52, + "learning_rate": 1.6023640086414318e-05, + "loss": 0.2677, + "step": 6026 + }, + { + "epoch": 2.52, + "learning_rate": 1.6022286687989466e-05, + "loss": 0.2588, + "step": 6027 + }, + { + "epoch": 2.52, + "learning_rate": 1.6020933116459324e-05, + "loss": 0.254, + "step": 6028 + }, + { + "epoch": 2.52, + "learning_rate": 1.6019579371862803e-05, + "loss": 0.2767, + "step": 6029 + }, + { + "epoch": 2.52, + "learning_rate": 1.6018225454238816e-05, + "loss": 0.248, + "step": 6030 + }, + { + "epoch": 2.53, + "learning_rate": 1.6016871363626277e-05, + "loss": 0.2448, + "step": 6031 + }, + { + "epoch": 2.53, + "learning_rate": 1.6015517100064114e-05, + "loss": 0.2472, + "step": 6032 + }, + { + "epoch": 2.53, + "learning_rate": 1.601416266359124e-05, + "loss": 0.2665, + "step": 6033 + }, + { + "epoch": 2.53, + "learning_rate": 1.6012808054246608e-05, + "loss": 0.2622, + "step": 6034 + }, + { + "epoch": 2.53, + "learning_rate": 1.601145327206914e-05, + "loss": 0.2483, + "step": 6035 + }, + { + "epoch": 2.53, + "learning_rate": 1.601009831709778e-05, + "loss": 0.2636, + "step": 6036 + }, + { + "epoch": 2.53, + "learning_rate": 1.600874318937148e-05, + "loss": 0.2706, + "step": 6037 + }, + { + "epoch": 2.53, + "learning_rate": 1.600738788892919e-05, + "loss": 0.2803, + "step": 6038 + }, + { + "epoch": 2.53, + "learning_rate": 1.6006032415809866e-05, + "loss": 0.2737, + "step": 6039 + }, + { + "epoch": 2.53, + "learning_rate": 1.6004676770052466e-05, + "loss": 0.2622, + "step": 6040 + }, + { + "epoch": 2.53, + "learning_rate": 1.600332095169596e-05, + "loss": 0.2615, + "step": 6041 + }, + { + "epoch": 2.53, + "learning_rate": 1.6001964960779326e-05, + "loss": 0.2653, + "step": 6042 + }, + { + "epoch": 2.53, + "learning_rate": 1.600060879734153e-05, + "loss": 0.2388, + "step": 6043 + }, + { + "epoch": 2.53, + "learning_rate": 1.5999252461421564e-05, + "loss": 0.2292, + "step": 6044 + }, + { + "epoch": 2.53, + "learning_rate": 1.5997895953058407e-05, + "loss": 0.2427, + "step": 6045 + }, + { + "epoch": 2.53, + "learning_rate": 1.5996539272291052e-05, + "loss": 0.2385, + "step": 6046 + }, + { + "epoch": 2.53, + "learning_rate": 1.59951824191585e-05, + "loss": 0.2729, + "step": 6047 + }, + { + "epoch": 2.53, + "learning_rate": 1.5993825393699746e-05, + "loss": 0.2514, + "step": 6048 + }, + { + "epoch": 2.53, + "learning_rate": 1.59924681959538e-05, + "loss": 0.2577, + "step": 6049 + }, + { + "epoch": 2.53, + "learning_rate": 1.5991110825959676e-05, + "loss": 0.245, + "step": 6050 + }, + { + "epoch": 2.53, + "learning_rate": 1.598975328375639e-05, + "loss": 0.2652, + "step": 6051 + }, + { + "epoch": 2.53, + "learning_rate": 1.5988395569382954e-05, + "loss": 0.2477, + "step": 6052 + }, + { + "epoch": 2.53, + "learning_rate": 1.5987037682878405e-05, + "loss": 0.2296, + "step": 6053 + }, + { + "epoch": 2.53, + "learning_rate": 1.5985679624281772e-05, + "loss": 0.244, + "step": 6054 + }, + { + "epoch": 2.54, + "learning_rate": 1.5984321393632087e-05, + "loss": 0.2329, + "step": 6055 + }, + { + "epoch": 2.54, + "learning_rate": 1.5982962990968397e-05, + "loss": 0.2332, + "step": 6056 + }, + { + "epoch": 2.54, + "learning_rate": 1.5981604416329742e-05, + "loss": 0.243, + "step": 6057 + }, + { + "epoch": 2.54, + "learning_rate": 1.598024566975518e-05, + "loss": 0.2464, + "step": 6058 + }, + { + "epoch": 2.54, + "learning_rate": 1.597888675128376e-05, + "loss": 0.2263, + "step": 6059 + }, + { + "epoch": 2.54, + "learning_rate": 1.5977527660954545e-05, + "loss": 0.2715, + "step": 6060 + }, + { + "epoch": 2.54, + "learning_rate": 1.5976168398806602e-05, + "loss": 0.2625, + "step": 6061 + }, + { + "epoch": 2.54, + "learning_rate": 1.5974808964879005e-05, + "loss": 0.2729, + "step": 6062 + }, + { + "epoch": 2.54, + "learning_rate": 1.5973449359210823e-05, + "loss": 0.2417, + "step": 6063 + }, + { + "epoch": 2.54, + "learning_rate": 1.597208958184114e-05, + "loss": 0.2355, + "step": 6064 + }, + { + "epoch": 2.54, + "learning_rate": 1.5970729632809044e-05, + "loss": 0.2282, + "step": 6065 + }, + { + "epoch": 2.54, + "learning_rate": 1.596936951215362e-05, + "loss": 0.2426, + "step": 6066 + }, + { + "epoch": 2.54, + "learning_rate": 1.5968009219913966e-05, + "loss": 0.2719, + "step": 6067 + }, + { + "epoch": 2.54, + "learning_rate": 1.5966648756129187e-05, + "loss": 0.2581, + "step": 6068 + }, + { + "epoch": 2.54, + "learning_rate": 1.5965288120838383e-05, + "loss": 0.2915, + "step": 6069 + }, + { + "epoch": 2.54, + "learning_rate": 1.596392731408066e-05, + "loss": 0.2598, + "step": 6070 + }, + { + "epoch": 2.54, + "learning_rate": 1.5962566335895146e-05, + "loss": 0.2513, + "step": 6071 + }, + { + "epoch": 2.54, + "learning_rate": 1.5961205186320952e-05, + "loss": 0.2531, + "step": 6072 + }, + { + "epoch": 2.54, + "learning_rate": 1.59598438653972e-05, + "loss": 0.2473, + "step": 6073 + }, + { + "epoch": 2.54, + "learning_rate": 1.595848237316303e-05, + "loss": 0.2286, + "step": 6074 + }, + { + "epoch": 2.54, + "learning_rate": 1.595712070965757e-05, + "loss": 0.2766, + "step": 6075 + }, + { + "epoch": 2.54, + "learning_rate": 1.595575887491996e-05, + "loss": 0.2845, + "step": 6076 + }, + { + "epoch": 2.54, + "learning_rate": 1.595439686898935e-05, + "loss": 0.2496, + "step": 6077 + }, + { + "epoch": 2.54, + "learning_rate": 1.595303469190488e-05, + "loss": 0.2742, + "step": 6078 + }, + { + "epoch": 2.55, + "learning_rate": 1.5951672343705714e-05, + "loss": 0.2465, + "step": 6079 + }, + { + "epoch": 2.55, + "learning_rate": 1.5950309824431008e-05, + "loss": 0.2647, + "step": 6080 + }, + { + "epoch": 2.55, + "learning_rate": 1.5948947134119926e-05, + "loss": 0.2563, + "step": 6081 + }, + { + "epoch": 2.55, + "learning_rate": 1.594758427281164e-05, + "loss": 0.2532, + "step": 6082 + }, + { + "epoch": 2.55, + "learning_rate": 1.594622124054532e-05, + "loss": 0.2513, + "step": 6083 + }, + { + "epoch": 2.55, + "learning_rate": 1.5944858037360145e-05, + "loss": 0.2514, + "step": 6084 + }, + { + "epoch": 2.55, + "learning_rate": 1.5943494663295303e-05, + "loss": 0.2461, + "step": 6085 + }, + { + "epoch": 2.55, + "learning_rate": 1.5942131118389984e-05, + "loss": 0.2538, + "step": 6086 + }, + { + "epoch": 2.55, + "learning_rate": 1.594076740268337e-05, + "loss": 0.2439, + "step": 6087 + }, + { + "epoch": 2.55, + "learning_rate": 1.593940351621468e-05, + "loss": 0.2267, + "step": 6088 + }, + { + "epoch": 2.55, + "learning_rate": 1.5938039459023098e-05, + "loss": 0.2541, + "step": 6089 + }, + { + "epoch": 2.55, + "learning_rate": 1.5936675231147845e-05, + "loss": 0.2191, + "step": 6090 + }, + { + "epoch": 2.55, + "learning_rate": 1.593531083262813e-05, + "loss": 0.268, + "step": 6091 + }, + { + "epoch": 2.55, + "learning_rate": 1.593394626350317e-05, + "loss": 0.2505, + "step": 6092 + }, + { + "epoch": 2.55, + "learning_rate": 1.5932581523812193e-05, + "loss": 0.2636, + "step": 6093 + }, + { + "epoch": 2.55, + "learning_rate": 1.5931216613594426e-05, + "loss": 0.2326, + "step": 6094 + }, + { + "epoch": 2.55, + "learning_rate": 1.5929851532889095e-05, + "loss": 0.2709, + "step": 6095 + }, + { + "epoch": 2.55, + "learning_rate": 1.592848628173545e-05, + "loss": 0.2567, + "step": 6096 + }, + { + "epoch": 2.55, + "learning_rate": 1.5927120860172722e-05, + "loss": 0.2475, + "step": 6097 + }, + { + "epoch": 2.55, + "learning_rate": 1.592575526824017e-05, + "loss": 0.2789, + "step": 6098 + }, + { + "epoch": 2.55, + "learning_rate": 1.5924389505977038e-05, + "loss": 0.2341, + "step": 6099 + }, + { + "epoch": 2.55, + "learning_rate": 1.5923023573422585e-05, + "loss": 0.2605, + "step": 6100 + }, + { + "epoch": 2.55, + "learning_rate": 1.592165747061608e-05, + "loss": 0.2788, + "step": 6101 + }, + { + "epoch": 2.55, + "learning_rate": 1.5920291197596784e-05, + "loss": 0.2772, + "step": 6102 + }, + { + "epoch": 2.56, + "learning_rate": 1.591892475440397e-05, + "loss": 0.232, + "step": 6103 + }, + { + "epoch": 2.56, + "learning_rate": 1.5917558141076915e-05, + "loss": 0.2478, + "step": 6104 + }, + { + "epoch": 2.56, + "learning_rate": 1.5916191357654906e-05, + "loss": 0.2352, + "step": 6105 + }, + { + "epoch": 2.56, + "learning_rate": 1.5914824404177222e-05, + "loss": 0.2803, + "step": 6106 + }, + { + "epoch": 2.56, + "learning_rate": 1.5913457280683163e-05, + "loss": 0.2376, + "step": 6107 + }, + { + "epoch": 2.56, + "learning_rate": 1.591208998721202e-05, + "loss": 0.264, + "step": 6108 + }, + { + "epoch": 2.56, + "learning_rate": 1.5910722523803098e-05, + "loss": 0.2767, + "step": 6109 + }, + { + "epoch": 2.56, + "learning_rate": 1.5909354890495703e-05, + "loss": 0.2921, + "step": 6110 + }, + { + "epoch": 2.56, + "learning_rate": 1.5907987087329143e-05, + "loss": 0.2388, + "step": 6111 + }, + { + "epoch": 2.56, + "learning_rate": 1.590661911434274e-05, + "loss": 0.2408, + "step": 6112 + }, + { + "epoch": 2.56, + "learning_rate": 1.5905250971575808e-05, + "loss": 0.2688, + "step": 6113 + }, + { + "epoch": 2.56, + "learning_rate": 1.590388265906768e-05, + "loss": 0.2435, + "step": 6114 + }, + { + "epoch": 2.56, + "learning_rate": 1.590251417685768e-05, + "loss": 0.2595, + "step": 6115 + }, + { + "epoch": 2.56, + "learning_rate": 1.5901145524985155e-05, + "loss": 0.26, + "step": 6116 + }, + { + "epoch": 2.56, + "learning_rate": 1.5899776703489432e-05, + "loss": 0.304, + "step": 6117 + }, + { + "epoch": 2.56, + "learning_rate": 1.5898407712409863e-05, + "loss": 0.2597, + "step": 6118 + }, + { + "epoch": 2.56, + "learning_rate": 1.58970385517858e-05, + "loss": 0.2655, + "step": 6119 + }, + { + "epoch": 2.56, + "learning_rate": 1.58956692216566e-05, + "loss": 0.2699, + "step": 6120 + }, + { + "epoch": 2.56, + "learning_rate": 1.5894299722061617e-05, + "loss": 0.273, + "step": 6121 + }, + { + "epoch": 2.56, + "learning_rate": 1.5892930053040218e-05, + "loss": 0.2488, + "step": 6122 + }, + { + "epoch": 2.56, + "learning_rate": 1.5891560214631775e-05, + "loss": 0.2928, + "step": 6123 + }, + { + "epoch": 2.56, + "learning_rate": 1.5890190206875663e-05, + "loss": 0.2671, + "step": 6124 + }, + { + "epoch": 2.56, + "learning_rate": 1.588882002981126e-05, + "loss": 0.2488, + "step": 6125 + }, + { + "epoch": 2.56, + "learning_rate": 1.588744968347795e-05, + "loss": 0.2903, + "step": 6126 + }, + { + "epoch": 2.57, + "learning_rate": 1.5886079167915123e-05, + "loss": 0.2526, + "step": 6127 + }, + { + "epoch": 2.57, + "learning_rate": 1.5884708483162175e-05, + "loss": 0.2429, + "step": 6128 + }, + { + "epoch": 2.57, + "learning_rate": 1.5883337629258502e-05, + "loss": 0.2909, + "step": 6129 + }, + { + "epoch": 2.57, + "learning_rate": 1.5881966606243512e-05, + "loss": 0.2554, + "step": 6130 + }, + { + "epoch": 2.57, + "learning_rate": 1.588059541415661e-05, + "loss": 0.2652, + "step": 6131 + }, + { + "epoch": 2.57, + "learning_rate": 1.5879224053037213e-05, + "loss": 0.2657, + "step": 6132 + }, + { + "epoch": 2.57, + "learning_rate": 1.5877852522924733e-05, + "loss": 0.2627, + "step": 6133 + }, + { + "epoch": 2.57, + "learning_rate": 1.58764808238586e-05, + "loss": 0.2544, + "step": 6134 + }, + { + "epoch": 2.57, + "learning_rate": 1.587510895587824e-05, + "loss": 0.2426, + "step": 6135 + }, + { + "epoch": 2.57, + "learning_rate": 1.587373691902309e-05, + "loss": 0.2401, + "step": 6136 + }, + { + "epoch": 2.57, + "learning_rate": 1.5872364713332583e-05, + "loss": 0.2544, + "step": 6137 + }, + { + "epoch": 2.57, + "learning_rate": 1.587099233884616e-05, + "loss": 0.2522, + "step": 6138 + }, + { + "epoch": 2.57, + "learning_rate": 1.586961979560328e-05, + "loss": 0.2804, + "step": 6139 + }, + { + "epoch": 2.57, + "learning_rate": 1.586824708364338e-05, + "loss": 0.2559, + "step": 6140 + }, + { + "epoch": 2.57, + "learning_rate": 1.5866874203005928e-05, + "loss": 0.2641, + "step": 6141 + }, + { + "epoch": 2.57, + "learning_rate": 1.5865501153730383e-05, + "loss": 0.2238, + "step": 6142 + }, + { + "epoch": 2.57, + "learning_rate": 1.586412793585621e-05, + "loss": 0.2681, + "step": 6143 + }, + { + "epoch": 2.57, + "learning_rate": 1.5862754549422884e-05, + "loss": 0.272, + "step": 6144 + }, + { + "epoch": 2.57, + "learning_rate": 1.5861380994469883e-05, + "loss": 0.2693, + "step": 6145 + }, + { + "epoch": 2.57, + "learning_rate": 1.5860007271036682e-05, + "loss": 0.2674, + "step": 6146 + }, + { + "epoch": 2.57, + "learning_rate": 1.585863337916278e-05, + "loss": 0.2741, + "step": 6147 + }, + { + "epoch": 2.57, + "learning_rate": 1.5857259318887653e-05, + "loss": 0.2629, + "step": 6148 + }, + { + "epoch": 2.57, + "learning_rate": 1.585588509025081e-05, + "loss": 0.202, + "step": 6149 + }, + { + "epoch": 2.57, + "learning_rate": 1.5854510693291744e-05, + "loss": 0.26, + "step": 6150 + }, + { + "epoch": 2.58, + "learning_rate": 1.5853136128049963e-05, + "loss": 0.2583, + "step": 6151 + }, + { + "epoch": 2.58, + "learning_rate": 1.585176139456498e-05, + "loss": 0.243, + "step": 6152 + }, + { + "epoch": 2.58, + "learning_rate": 1.5850386492876307e-05, + "loss": 0.2652, + "step": 6153 + }, + { + "epoch": 2.58, + "learning_rate": 1.5849011423023463e-05, + "loss": 0.2457, + "step": 6154 + }, + { + "epoch": 2.58, + "learning_rate": 1.584763618504598e-05, + "loss": 0.2511, + "step": 6155 + }, + { + "epoch": 2.58, + "learning_rate": 1.5846260778983382e-05, + "loss": 0.2497, + "step": 6156 + }, + { + "epoch": 2.58, + "learning_rate": 1.5844885204875207e-05, + "loss": 0.2725, + "step": 6157 + }, + { + "epoch": 2.58, + "learning_rate": 1.584350946276099e-05, + "loss": 0.292, + "step": 6158 + }, + { + "epoch": 2.58, + "learning_rate": 1.5842133552680286e-05, + "loss": 0.2717, + "step": 6159 + }, + { + "epoch": 2.58, + "learning_rate": 1.5840757474672634e-05, + "loss": 0.2678, + "step": 6160 + }, + { + "epoch": 2.58, + "learning_rate": 1.5839381228777586e-05, + "loss": 0.2544, + "step": 6161 + }, + { + "epoch": 2.58, + "learning_rate": 1.583800481503471e-05, + "loss": 0.2599, + "step": 6162 + }, + { + "epoch": 2.58, + "learning_rate": 1.5836628233483568e-05, + "loss": 0.2436, + "step": 6163 + }, + { + "epoch": 2.58, + "learning_rate": 1.5835251484163728e-05, + "loss": 0.2356, + "step": 6164 + }, + { + "epoch": 2.58, + "learning_rate": 1.583387456711476e-05, + "loss": 0.2818, + "step": 6165 + }, + { + "epoch": 2.58, + "learning_rate": 1.5832497482376244e-05, + "loss": 0.2922, + "step": 6166 + }, + { + "epoch": 2.58, + "learning_rate": 1.583112022998776e-05, + "loss": 0.2658, + "step": 6167 + }, + { + "epoch": 2.58, + "learning_rate": 1.5829742809988907e-05, + "loss": 0.2644, + "step": 6168 + }, + { + "epoch": 2.58, + "learning_rate": 1.5828365222419268e-05, + "loss": 0.2485, + "step": 6169 + }, + { + "epoch": 2.58, + "learning_rate": 1.582698746731844e-05, + "loss": 0.2583, + "step": 6170 + }, + { + "epoch": 2.58, + "learning_rate": 1.582560954472603e-05, + "loss": 0.2407, + "step": 6171 + }, + { + "epoch": 2.58, + "learning_rate": 1.582423145468164e-05, + "loss": 0.2561, + "step": 6172 + }, + { + "epoch": 2.58, + "learning_rate": 1.5822853197224888e-05, + "loss": 0.2346, + "step": 6173 + }, + { + "epoch": 2.58, + "learning_rate": 1.5821474772395385e-05, + "loss": 0.2575, + "step": 6174 + }, + { + "epoch": 2.59, + "learning_rate": 1.5820096180232762e-05, + "loss": 0.2469, + "step": 6175 + }, + { + "epoch": 2.59, + "learning_rate": 1.5818717420776635e-05, + "loss": 0.2751, + "step": 6176 + }, + { + "epoch": 2.59, + "learning_rate": 1.5817338494066636e-05, + "loss": 0.251, + "step": 6177 + }, + { + "epoch": 2.59, + "learning_rate": 1.581595940014241e-05, + "loss": 0.2643, + "step": 6178 + }, + { + "epoch": 2.59, + "learning_rate": 1.5814580139043587e-05, + "loss": 0.2536, + "step": 6179 + }, + { + "epoch": 2.59, + "learning_rate": 1.581320071080982e-05, + "loss": 0.2741, + "step": 6180 + }, + { + "epoch": 2.59, + "learning_rate": 1.581182111548076e-05, + "loss": 0.2593, + "step": 6181 + }, + { + "epoch": 2.59, + "learning_rate": 1.5810441353096054e-05, + "loss": 0.2871, + "step": 6182 + }, + { + "epoch": 2.59, + "learning_rate": 1.5809061423695368e-05, + "loss": 0.2528, + "step": 6183 + }, + { + "epoch": 2.59, + "learning_rate": 1.5807681327318372e-05, + "loss": 0.2764, + "step": 6184 + }, + { + "epoch": 2.59, + "learning_rate": 1.5806301064004724e-05, + "loss": 0.2689, + "step": 6185 + }, + { + "epoch": 2.59, + "learning_rate": 1.5804920633794107e-05, + "loss": 0.2624, + "step": 6186 + }, + { + "epoch": 2.59, + "learning_rate": 1.5803540036726198e-05, + "loss": 0.2553, + "step": 6187 + }, + { + "epoch": 2.59, + "learning_rate": 1.5802159272840677e-05, + "loss": 0.24, + "step": 6188 + }, + { + "epoch": 2.59, + "learning_rate": 1.580077834217724e-05, + "loss": 0.2369, + "step": 6189 + }, + { + "epoch": 2.59, + "learning_rate": 1.5799397244775576e-05, + "loss": 0.2515, + "step": 6190 + }, + { + "epoch": 2.59, + "learning_rate": 1.5798015980675384e-05, + "loss": 0.2577, + "step": 6191 + }, + { + "epoch": 2.59, + "learning_rate": 1.579663454991637e-05, + "loss": 0.2721, + "step": 6192 + }, + { + "epoch": 2.59, + "learning_rate": 1.579525295253824e-05, + "loss": 0.2448, + "step": 6193 + }, + { + "epoch": 2.59, + "learning_rate": 1.5793871188580704e-05, + "loss": 0.2729, + "step": 6194 + }, + { + "epoch": 2.59, + "learning_rate": 1.579248925808348e-05, + "loss": 0.2618, + "step": 6195 + }, + { + "epoch": 2.59, + "learning_rate": 1.5791107161086294e-05, + "loss": 0.2722, + "step": 6196 + }, + { + "epoch": 2.59, + "learning_rate": 1.5789724897628873e-05, + "loss": 0.261, + "step": 6197 + }, + { + "epoch": 2.59, + "learning_rate": 1.578834246775095e-05, + "loss": 0.2592, + "step": 6198 + }, + { + "epoch": 2.6, + "learning_rate": 1.5786959871492254e-05, + "loss": 0.2786, + "step": 6199 + }, + { + "epoch": 2.6, + "learning_rate": 1.5785577108892537e-05, + "loss": 0.2406, + "step": 6200 + }, + { + "epoch": 2.6, + "learning_rate": 1.5784194179991534e-05, + "loss": 0.2347, + "step": 6201 + }, + { + "epoch": 2.6, + "learning_rate": 1.5782811084829005e-05, + "loss": 0.2531, + "step": 6202 + }, + { + "epoch": 2.6, + "learning_rate": 1.5781427823444704e-05, + "loss": 0.2519, + "step": 6203 + }, + { + "epoch": 2.6, + "learning_rate": 1.578004439587839e-05, + "loss": 0.2603, + "step": 6204 + }, + { + "epoch": 2.6, + "learning_rate": 1.5778660802169826e-05, + "loss": 0.2648, + "step": 6205 + }, + { + "epoch": 2.6, + "learning_rate": 1.577727704235879e-05, + "loss": 0.2316, + "step": 6206 + }, + { + "epoch": 2.6, + "learning_rate": 1.5775893116485052e-05, + "loss": 0.2395, + "step": 6207 + }, + { + "epoch": 2.6, + "learning_rate": 1.5774509024588388e-05, + "loss": 0.2766, + "step": 6208 + }, + { + "epoch": 2.6, + "learning_rate": 1.5773124766708588e-05, + "loss": 0.2556, + "step": 6209 + }, + { + "epoch": 2.6, + "learning_rate": 1.577174034288544e-05, + "loss": 0.2498, + "step": 6210 + }, + { + "epoch": 2.6, + "learning_rate": 1.5770355753158736e-05, + "loss": 0.2674, + "step": 6211 + }, + { + "epoch": 2.6, + "learning_rate": 1.5768970997568282e-05, + "loss": 0.2553, + "step": 6212 + }, + { + "epoch": 2.6, + "learning_rate": 1.576758607615387e-05, + "loss": 0.2363, + "step": 6213 + }, + { + "epoch": 2.6, + "learning_rate": 1.5766200988955315e-05, + "loss": 0.2762, + "step": 6214 + }, + { + "epoch": 2.6, + "learning_rate": 1.576481573601243e-05, + "loss": 0.2845, + "step": 6215 + }, + { + "epoch": 2.6, + "learning_rate": 1.5763430317365032e-05, + "loss": 0.2653, + "step": 6216 + }, + { + "epoch": 2.6, + "learning_rate": 1.5762044733052944e-05, + "loss": 0.2545, + "step": 6217 + }, + { + "epoch": 2.6, + "learning_rate": 1.576065898311599e-05, + "loss": 0.2495, + "step": 6218 + }, + { + "epoch": 2.6, + "learning_rate": 1.575927306759401e-05, + "loss": 0.2674, + "step": 6219 + }, + { + "epoch": 2.6, + "learning_rate": 1.575788698652684e-05, + "loss": 0.2691, + "step": 6220 + }, + { + "epoch": 2.6, + "learning_rate": 1.5756500739954312e-05, + "loss": 0.2548, + "step": 6221 + }, + { + "epoch": 2.6, + "learning_rate": 1.5755114327916282e-05, + "loss": 0.2469, + "step": 6222 + }, + { + "epoch": 2.61, + "learning_rate": 1.5753727750452596e-05, + "loss": 0.2675, + "step": 6223 + }, + { + "epoch": 2.61, + "learning_rate": 1.575234100760311e-05, + "loss": 0.2719, + "step": 6224 + }, + { + "epoch": 2.61, + "learning_rate": 1.5750954099407688e-05, + "loss": 0.2764, + "step": 6225 + }, + { + "epoch": 2.61, + "learning_rate": 1.5749567025906193e-05, + "loss": 0.2575, + "step": 6226 + }, + { + "epoch": 2.61, + "learning_rate": 1.5748179787138494e-05, + "loss": 0.2601, + "step": 6227 + }, + { + "epoch": 2.61, + "learning_rate": 1.5746792383144472e-05, + "loss": 0.2749, + "step": 6228 + }, + { + "epoch": 2.61, + "learning_rate": 1.5745404813964e-05, + "loss": 0.2885, + "step": 6229 + }, + { + "epoch": 2.61, + "learning_rate": 1.5744017079636967e-05, + "loss": 0.2574, + "step": 6230 + }, + { + "epoch": 2.61, + "learning_rate": 1.5742629180203258e-05, + "loss": 0.2448, + "step": 6231 + }, + { + "epoch": 2.61, + "learning_rate": 1.574124111570277e-05, + "loss": 0.2523, + "step": 6232 + }, + { + "epoch": 2.61, + "learning_rate": 1.57398528861754e-05, + "loss": 0.277, + "step": 6233 + }, + { + "epoch": 2.61, + "learning_rate": 1.5738464491661054e-05, + "loss": 0.2612, + "step": 6234 + }, + { + "epoch": 2.61, + "learning_rate": 1.5737075932199637e-05, + "loss": 0.2514, + "step": 6235 + }, + { + "epoch": 2.61, + "learning_rate": 1.5735687207831064e-05, + "loss": 0.2762, + "step": 6236 + }, + { + "epoch": 2.61, + "learning_rate": 1.5734298318595252e-05, + "loss": 0.266, + "step": 6237 + }, + { + "epoch": 2.61, + "learning_rate": 1.573290926453212e-05, + "loss": 0.2705, + "step": 6238 + }, + { + "epoch": 2.61, + "learning_rate": 1.5731520045681606e-05, + "loss": 0.2766, + "step": 6239 + }, + { + "epoch": 2.61, + "learning_rate": 1.5730130662083628e-05, + "loss": 0.2422, + "step": 6240 + }, + { + "epoch": 2.61, + "learning_rate": 1.572874111377813e-05, + "loss": 0.2674, + "step": 6241 + }, + { + "epoch": 2.61, + "learning_rate": 1.5727351400805054e-05, + "loss": 0.268, + "step": 6242 + }, + { + "epoch": 2.61, + "learning_rate": 1.5725961523204345e-05, + "loss": 0.2462, + "step": 6243 + }, + { + "epoch": 2.61, + "learning_rate": 1.5724571481015953e-05, + "loss": 0.2585, + "step": 6244 + }, + { + "epoch": 2.61, + "learning_rate": 1.572318127427983e-05, + "loss": 0.308, + "step": 6245 + }, + { + "epoch": 2.62, + "learning_rate": 1.5721790903035942e-05, + "loss": 0.2525, + "step": 6246 + }, + { + "epoch": 2.62, + "learning_rate": 1.572040036732425e-05, + "loss": 0.2731, + "step": 6247 + }, + { + "epoch": 2.62, + "learning_rate": 1.571900966718473e-05, + "loss": 0.246, + "step": 6248 + }, + { + "epoch": 2.62, + "learning_rate": 1.571761880265735e-05, + "loss": 0.2731, + "step": 6249 + }, + { + "epoch": 2.62, + "learning_rate": 1.5716227773782088e-05, + "loss": 0.2903, + "step": 6250 + }, + { + "epoch": 2.62, + "learning_rate": 1.5714836580598936e-05, + "loss": 0.2622, + "step": 6251 + }, + { + "epoch": 2.62, + "learning_rate": 1.5713445223147876e-05, + "loss": 0.2476, + "step": 6252 + }, + { + "epoch": 2.62, + "learning_rate": 1.5712053701468902e-05, + "loss": 0.259, + "step": 6253 + }, + { + "epoch": 2.62, + "learning_rate": 1.5710662015602016e-05, + "loss": 0.2912, + "step": 6254 + }, + { + "epoch": 2.62, + "learning_rate": 1.5709270165587216e-05, + "loss": 0.2395, + "step": 6255 + }, + { + "epoch": 2.62, + "learning_rate": 1.570787815146451e-05, + "loss": 0.2619, + "step": 6256 + }, + { + "epoch": 2.62, + "learning_rate": 1.5706485973273916e-05, + "loss": 0.2478, + "step": 6257 + }, + { + "epoch": 2.62, + "learning_rate": 1.570509363105544e-05, + "loss": 0.2812, + "step": 6258 + }, + { + "epoch": 2.62, + "learning_rate": 1.5703701124849115e-05, + "loss": 0.2282, + "step": 6259 + }, + { + "epoch": 2.62, + "learning_rate": 1.5702308454694963e-05, + "loss": 0.2505, + "step": 6260 + }, + { + "epoch": 2.62, + "learning_rate": 1.570091562063301e-05, + "loss": 0.2522, + "step": 6261 + }, + { + "epoch": 2.62, + "learning_rate": 1.5699522622703303e-05, + "loss": 0.2635, + "step": 6262 + }, + { + "epoch": 2.62, + "learning_rate": 1.569812946094587e-05, + "loss": 0.2555, + "step": 6263 + }, + { + "epoch": 2.62, + "learning_rate": 1.5696736135400765e-05, + "loss": 0.2671, + "step": 6264 + }, + { + "epoch": 2.62, + "learning_rate": 1.569534264610803e-05, + "loss": 0.2352, + "step": 6265 + }, + { + "epoch": 2.62, + "learning_rate": 1.5693948993107733e-05, + "loss": 0.2657, + "step": 6266 + }, + { + "epoch": 2.62, + "learning_rate": 1.569255517643992e-05, + "loss": 0.2495, + "step": 6267 + }, + { + "epoch": 2.62, + "learning_rate": 1.569116119614466e-05, + "loss": 0.2657, + "step": 6268 + }, + { + "epoch": 2.62, + "learning_rate": 1.5689767052262028e-05, + "loss": 0.2685, + "step": 6269 + }, + { + "epoch": 2.63, + "learning_rate": 1.568837274483208e-05, + "loss": 0.2572, + "step": 6270 + }, + { + "epoch": 2.63, + "learning_rate": 1.5686978273894914e-05, + "loss": 0.2648, + "step": 6271 + }, + { + "epoch": 2.63, + "learning_rate": 1.5685583639490598e-05, + "loss": 0.2516, + "step": 6272 + }, + { + "epoch": 2.63, + "learning_rate": 1.568418884165923e-05, + "loss": 0.2292, + "step": 6273 + }, + { + "epoch": 2.63, + "learning_rate": 1.5682793880440895e-05, + "loss": 0.2718, + "step": 6274 + }, + { + "epoch": 2.63, + "learning_rate": 1.5681398755875695e-05, + "loss": 0.2613, + "step": 6275 + }, + { + "epoch": 2.63, + "learning_rate": 1.5680003468003725e-05, + "loss": 0.2443, + "step": 6276 + }, + { + "epoch": 2.63, + "learning_rate": 1.56786080168651e-05, + "loss": 0.2538, + "step": 6277 + }, + { + "epoch": 2.63, + "learning_rate": 1.567721240249993e-05, + "loss": 0.2497, + "step": 6278 + }, + { + "epoch": 2.63, + "learning_rate": 1.5675816624948322e-05, + "loss": 0.2681, + "step": 6279 + }, + { + "epoch": 2.63, + "learning_rate": 1.56744206842504e-05, + "loss": 0.2525, + "step": 6280 + }, + { + "epoch": 2.63, + "learning_rate": 1.5673024580446294e-05, + "loss": 0.2496, + "step": 6281 + }, + { + "epoch": 2.63, + "learning_rate": 1.567162831357613e-05, + "loss": 0.2768, + "step": 6282 + }, + { + "epoch": 2.63, + "learning_rate": 1.5670231883680044e-05, + "loss": 0.2721, + "step": 6283 + }, + { + "epoch": 2.63, + "learning_rate": 1.566883529079817e-05, + "loss": 0.274, + "step": 6284 + }, + { + "epoch": 2.63, + "learning_rate": 1.5667438534970664e-05, + "loss": 0.283, + "step": 6285 + }, + { + "epoch": 2.63, + "learning_rate": 1.566604161623766e-05, + "loss": 0.284, + "step": 6286 + }, + { + "epoch": 2.63, + "learning_rate": 1.566464453463932e-05, + "loss": 0.2795, + "step": 6287 + }, + { + "epoch": 2.63, + "learning_rate": 1.5663247290215797e-05, + "loss": 0.2495, + "step": 6288 + }, + { + "epoch": 2.63, + "learning_rate": 1.566184988300726e-05, + "loss": 0.261, + "step": 6289 + }, + { + "epoch": 2.63, + "learning_rate": 1.5660452313053873e-05, + "loss": 0.2663, + "step": 6290 + }, + { + "epoch": 2.63, + "learning_rate": 1.56590545803958e-05, + "loss": 0.2632, + "step": 6291 + }, + { + "epoch": 2.63, + "learning_rate": 1.565765668507323e-05, + "loss": 0.2589, + "step": 6292 + }, + { + "epoch": 2.63, + "learning_rate": 1.565625862712634e-05, + "loss": 0.26, + "step": 6293 + }, + { + "epoch": 2.64, + "learning_rate": 1.5654860406595312e-05, + "loss": 0.2677, + "step": 6294 + }, + { + "epoch": 2.64, + "learning_rate": 1.5653462023520344e-05, + "loss": 0.2423, + "step": 6295 + }, + { + "epoch": 2.64, + "learning_rate": 1.5652063477941625e-05, + "loss": 0.2646, + "step": 6296 + }, + { + "epoch": 2.64, + "learning_rate": 1.5650664769899356e-05, + "loss": 0.2359, + "step": 6297 + }, + { + "epoch": 2.64, + "learning_rate": 1.564926589943374e-05, + "loss": 0.2378, + "step": 6298 + }, + { + "epoch": 2.64, + "learning_rate": 1.5647866866584992e-05, + "loss": 0.2465, + "step": 6299 + }, + { + "epoch": 2.64, + "learning_rate": 1.5646467671393323e-05, + "loss": 0.2394, + "step": 6300 + }, + { + "epoch": 2.64, + "learning_rate": 1.564506831389895e-05, + "loss": 0.2737, + "step": 6301 + }, + { + "epoch": 2.64, + "learning_rate": 1.5643668794142102e-05, + "loss": 0.2833, + "step": 6302 + }, + { + "epoch": 2.64, + "learning_rate": 1.5642269112162998e-05, + "loss": 0.2895, + "step": 6303 + }, + { + "epoch": 2.64, + "learning_rate": 1.5640869268001875e-05, + "loss": 0.2499, + "step": 6304 + }, + { + "epoch": 2.64, + "learning_rate": 1.563946926169897e-05, + "loss": 0.2881, + "step": 6305 + }, + { + "epoch": 2.64, + "learning_rate": 1.5638069093294527e-05, + "loss": 0.271, + "step": 6306 + }, + { + "epoch": 2.64, + "learning_rate": 1.5636668762828793e-05, + "loss": 0.2463, + "step": 6307 + }, + { + "epoch": 2.64, + "learning_rate": 1.5635268270342015e-05, + "loss": 0.2836, + "step": 6308 + }, + { + "epoch": 2.64, + "learning_rate": 1.5633867615874453e-05, + "loss": 0.2646, + "step": 6309 + }, + { + "epoch": 2.64, + "learning_rate": 1.5632466799466362e-05, + "loss": 0.2348, + "step": 6310 + }, + { + "epoch": 2.64, + "learning_rate": 1.5631065821158016e-05, + "loss": 0.2564, + "step": 6311 + }, + { + "epoch": 2.64, + "learning_rate": 1.5629664680989677e-05, + "loss": 0.2469, + "step": 6312 + }, + { + "epoch": 2.64, + "learning_rate": 1.562826337900162e-05, + "loss": 0.2531, + "step": 6313 + }, + { + "epoch": 2.64, + "learning_rate": 1.562686191523413e-05, + "loss": 0.2948, + "step": 6314 + }, + { + "epoch": 2.64, + "learning_rate": 1.562546028972749e-05, + "loss": 0.2857, + "step": 6315 + }, + { + "epoch": 2.64, + "learning_rate": 1.562405850252198e-05, + "loss": 0.2549, + "step": 6316 + }, + { + "epoch": 2.64, + "learning_rate": 1.5622656553657904e-05, + "loss": 0.2772, + "step": 6317 + }, + { + "epoch": 2.65, + "learning_rate": 1.5621254443175553e-05, + "loss": 0.28, + "step": 6318 + }, + { + "epoch": 2.65, + "learning_rate": 1.561985217111523e-05, + "loss": 0.2754, + "step": 6319 + }, + { + "epoch": 2.65, + "learning_rate": 1.5618449737517242e-05, + "loss": 0.2899, + "step": 6320 + }, + { + "epoch": 2.65, + "learning_rate": 1.5617047142421904e-05, + "loss": 0.2448, + "step": 6321 + }, + { + "epoch": 2.65, + "learning_rate": 1.5615644385869532e-05, + "loss": 0.2475, + "step": 6322 + }, + { + "epoch": 2.65, + "learning_rate": 1.5614241467900444e-05, + "loss": 0.276, + "step": 6323 + }, + { + "epoch": 2.65, + "learning_rate": 1.5612838388554966e-05, + "loss": 0.2666, + "step": 6324 + }, + { + "epoch": 2.65, + "learning_rate": 1.561143514787343e-05, + "loss": 0.2741, + "step": 6325 + }, + { + "epoch": 2.65, + "learning_rate": 1.561003174589617e-05, + "loss": 0.2765, + "step": 6326 + }, + { + "epoch": 2.65, + "learning_rate": 1.5608628182663527e-05, + "loss": 0.2711, + "step": 6327 + }, + { + "epoch": 2.65, + "learning_rate": 1.5607224458215843e-05, + "loss": 0.2404, + "step": 6328 + }, + { + "epoch": 2.65, + "learning_rate": 1.560582057259347e-05, + "loss": 0.2638, + "step": 6329 + }, + { + "epoch": 2.65, + "learning_rate": 1.5604416525836756e-05, + "loss": 0.2678, + "step": 6330 + }, + { + "epoch": 2.65, + "learning_rate": 1.560301231798606e-05, + "loss": 0.2781, + "step": 6331 + }, + { + "epoch": 2.65, + "learning_rate": 1.560160794908175e-05, + "loss": 0.2909, + "step": 6332 + }, + { + "epoch": 2.65, + "learning_rate": 1.560020341916419e-05, + "loss": 0.2679, + "step": 6333 + }, + { + "epoch": 2.65, + "learning_rate": 1.5598798728273753e-05, + "loss": 0.251, + "step": 6334 + }, + { + "epoch": 2.65, + "learning_rate": 1.5597393876450813e-05, + "loss": 0.2647, + "step": 6335 + }, + { + "epoch": 2.65, + "learning_rate": 1.5595988863735758e-05, + "loss": 0.2343, + "step": 6336 + }, + { + "epoch": 2.65, + "learning_rate": 1.5594583690168966e-05, + "loss": 0.2831, + "step": 6337 + }, + { + "epoch": 2.65, + "learning_rate": 1.559317835579083e-05, + "loss": 0.2257, + "step": 6338 + }, + { + "epoch": 2.65, + "learning_rate": 1.5591772860641743e-05, + "loss": 0.2514, + "step": 6339 + }, + { + "epoch": 2.65, + "learning_rate": 1.5590367204762113e-05, + "loss": 0.2614, + "step": 6340 + }, + { + "epoch": 2.65, + "learning_rate": 1.5588961388192333e-05, + "loss": 0.2447, + "step": 6341 + }, + { + "epoch": 2.66, + "learning_rate": 1.558755541097282e-05, + "loss": 0.2514, + "step": 6342 + }, + { + "epoch": 2.66, + "learning_rate": 1.5586149273143982e-05, + "loss": 0.2586, + "step": 6343 + }, + { + "epoch": 2.66, + "learning_rate": 1.5584742974746242e-05, + "loss": 0.2454, + "step": 6344 + }, + { + "epoch": 2.66, + "learning_rate": 1.5583336515820024e-05, + "loss": 0.2618, + "step": 6345 + }, + { + "epoch": 2.66, + "learning_rate": 1.558192989640575e-05, + "loss": 0.2594, + "step": 6346 + }, + { + "epoch": 2.66, + "learning_rate": 1.5580523116543853e-05, + "loss": 0.2391, + "step": 6347 + }, + { + "epoch": 2.66, + "learning_rate": 1.557911617627477e-05, + "loss": 0.2645, + "step": 6348 + }, + { + "epoch": 2.66, + "learning_rate": 1.557770907563895e-05, + "loss": 0.2572, + "step": 6349 + }, + { + "epoch": 2.66, + "learning_rate": 1.5576301814676825e-05, + "loss": 0.2587, + "step": 6350 + }, + { + "epoch": 2.66, + "learning_rate": 1.5574894393428856e-05, + "loss": 0.2847, + "step": 6351 + }, + { + "epoch": 2.66, + "learning_rate": 1.5573486811935494e-05, + "loss": 0.2696, + "step": 6352 + }, + { + "epoch": 2.66, + "learning_rate": 1.55720790702372e-05, + "loss": 0.2752, + "step": 6353 + }, + { + "epoch": 2.66, + "learning_rate": 1.557067116837444e-05, + "loss": 0.233, + "step": 6354 + }, + { + "epoch": 2.66, + "learning_rate": 1.5569263106387676e-05, + "loss": 0.2646, + "step": 6355 + }, + { + "epoch": 2.66, + "learning_rate": 1.5567854884317387e-05, + "loss": 0.273, + "step": 6356 + }, + { + "epoch": 2.66, + "learning_rate": 1.5566446502204052e-05, + "loss": 0.2578, + "step": 6357 + }, + { + "epoch": 2.66, + "learning_rate": 1.556503796008815e-05, + "loss": 0.2444, + "step": 6358 + }, + { + "epoch": 2.66, + "learning_rate": 1.5563629258010173e-05, + "loss": 0.2624, + "step": 6359 + }, + { + "epoch": 2.66, + "learning_rate": 1.556222039601061e-05, + "loss": 0.2674, + "step": 6360 + }, + { + "epoch": 2.66, + "learning_rate": 1.5560811374129955e-05, + "loss": 0.2442, + "step": 6361 + }, + { + "epoch": 2.66, + "learning_rate": 1.5559402192408712e-05, + "loss": 0.283, + "step": 6362 + }, + { + "epoch": 2.66, + "learning_rate": 1.5557992850887388e-05, + "loss": 0.2245, + "step": 6363 + }, + { + "epoch": 2.66, + "learning_rate": 1.5556583349606492e-05, + "loss": 0.2622, + "step": 6364 + }, + { + "epoch": 2.66, + "learning_rate": 1.5555173688606537e-05, + "loss": 0.2813, + "step": 6365 + }, + { + "epoch": 2.67, + "learning_rate": 1.5553763867928043e-05, + "loss": 0.2728, + "step": 6366 + }, + { + "epoch": 2.67, + "learning_rate": 1.5552353887611536e-05, + "loss": 0.2805, + "step": 6367 + }, + { + "epoch": 2.67, + "learning_rate": 1.5550943747697546e-05, + "loss": 0.2557, + "step": 6368 + }, + { + "epoch": 2.67, + "learning_rate": 1.5549533448226598e-05, + "loss": 0.2733, + "step": 6369 + }, + { + "epoch": 2.67, + "learning_rate": 1.554812298923924e-05, + "loss": 0.2685, + "step": 6370 + }, + { + "epoch": 2.67, + "learning_rate": 1.5546712370776005e-05, + "loss": 0.2678, + "step": 6371 + }, + { + "epoch": 2.67, + "learning_rate": 1.554530159287745e-05, + "loss": 0.2202, + "step": 6372 + }, + { + "epoch": 2.67, + "learning_rate": 1.5543890655584122e-05, + "loss": 0.2725, + "step": 6373 + }, + { + "epoch": 2.67, + "learning_rate": 1.5542479558936572e-05, + "loss": 0.2627, + "step": 6374 + }, + { + "epoch": 2.67, + "learning_rate": 1.554106830297537e-05, + "loss": 0.2835, + "step": 6375 + }, + { + "epoch": 2.67, + "learning_rate": 1.5539656887741076e-05, + "loss": 0.2782, + "step": 6376 + }, + { + "epoch": 2.67, + "learning_rate": 1.5538245313274258e-05, + "loss": 0.278, + "step": 6377 + }, + { + "epoch": 2.67, + "learning_rate": 1.5536833579615495e-05, + "loss": 0.2575, + "step": 6378 + }, + { + "epoch": 2.67, + "learning_rate": 1.5535421686805367e-05, + "loss": 0.242, + "step": 6379 + }, + { + "epoch": 2.67, + "learning_rate": 1.553400963488445e-05, + "loss": 0.2558, + "step": 6380 + }, + { + "epoch": 2.67, + "learning_rate": 1.553259742389334e-05, + "loss": 0.2637, + "step": 6381 + }, + { + "epoch": 2.67, + "learning_rate": 1.553118505387262e-05, + "loss": 0.2492, + "step": 6382 + }, + { + "epoch": 2.67, + "learning_rate": 1.55297725248629e-05, + "loss": 0.2669, + "step": 6383 + }, + { + "epoch": 2.67, + "learning_rate": 1.5528359836904774e-05, + "loss": 0.2425, + "step": 6384 + }, + { + "epoch": 2.67, + "learning_rate": 1.552694699003885e-05, + "loss": 0.2539, + "step": 6385 + }, + { + "epoch": 2.67, + "learning_rate": 1.5525533984305742e-05, + "loss": 0.2474, + "step": 6386 + }, + { + "epoch": 2.67, + "learning_rate": 1.552412081974606e-05, + "loss": 0.254, + "step": 6387 + }, + { + "epoch": 2.67, + "learning_rate": 1.5522707496400425e-05, + "loss": 0.2662, + "step": 6388 + }, + { + "epoch": 2.67, + "learning_rate": 1.552129401430947e-05, + "loss": 0.2675, + "step": 6389 + }, + { + "epoch": 2.68, + "learning_rate": 1.5519880373513814e-05, + "loss": 0.2582, + "step": 6390 + }, + { + "epoch": 2.68, + "learning_rate": 1.5518466574054097e-05, + "loss": 0.2468, + "step": 6391 + }, + { + "epoch": 2.68, + "learning_rate": 1.5517052615970955e-05, + "loss": 0.2305, + "step": 6392 + }, + { + "epoch": 2.68, + "learning_rate": 1.551563849930503e-05, + "loss": 0.3024, + "step": 6393 + }, + { + "epoch": 2.68, + "learning_rate": 1.551422422409697e-05, + "loss": 0.2718, + "step": 6394 + }, + { + "epoch": 2.68, + "learning_rate": 1.551280979038743e-05, + "loss": 0.2705, + "step": 6395 + }, + { + "epoch": 2.68, + "learning_rate": 1.551139519821706e-05, + "loss": 0.2716, + "step": 6396 + }, + { + "epoch": 2.68, + "learning_rate": 1.5509980447626532e-05, + "loss": 0.2879, + "step": 6397 + }, + { + "epoch": 2.68, + "learning_rate": 1.5508565538656505e-05, + "loss": 0.2648, + "step": 6398 + }, + { + "epoch": 2.68, + "learning_rate": 1.5507150471347644e-05, + "loss": 0.2476, + "step": 6399 + }, + { + "epoch": 2.68, + "learning_rate": 1.5505735245740635e-05, + "loss": 0.2499, + "step": 6400 + }, + { + "epoch": 2.68, + "learning_rate": 1.5504319861876148e-05, + "loss": 0.2537, + "step": 6401 + }, + { + "epoch": 2.68, + "learning_rate": 1.5502904319794875e-05, + "loss": 0.2677, + "step": 6402 + }, + { + "epoch": 2.68, + "learning_rate": 1.55014886195375e-05, + "loss": 0.2583, + "step": 6403 + }, + { + "epoch": 2.68, + "learning_rate": 1.5500072761144712e-05, + "loss": 0.2513, + "step": 6404 + }, + { + "epoch": 2.68, + "learning_rate": 1.5498656744657222e-05, + "loss": 0.2343, + "step": 6405 + }, + { + "epoch": 2.68, + "learning_rate": 1.5497240570115716e-05, + "loss": 0.2468, + "step": 6406 + }, + { + "epoch": 2.68, + "learning_rate": 1.5495824237560907e-05, + "loss": 0.2654, + "step": 6407 + }, + { + "epoch": 2.68, + "learning_rate": 1.5494407747033513e-05, + "loss": 0.2677, + "step": 6408 + }, + { + "epoch": 2.68, + "learning_rate": 1.5492991098574243e-05, + "loss": 0.2488, + "step": 6409 + }, + { + "epoch": 2.68, + "learning_rate": 1.5491574292223814e-05, + "loss": 0.2519, + "step": 6410 + }, + { + "epoch": 2.68, + "learning_rate": 1.5490157328022957e-05, + "loss": 0.2988, + "step": 6411 + }, + { + "epoch": 2.68, + "learning_rate": 1.54887402060124e-05, + "loss": 0.281, + "step": 6412 + }, + { + "epoch": 2.68, + "learning_rate": 1.5487322926232876e-05, + "loss": 0.262, + "step": 6413 + }, + { + "epoch": 2.69, + "learning_rate": 1.5485905488725126e-05, + "loss": 0.2738, + "step": 6414 + }, + { + "epoch": 2.69, + "learning_rate": 1.548448789352989e-05, + "loss": 0.2776, + "step": 6415 + }, + { + "epoch": 2.69, + "learning_rate": 1.5483070140687916e-05, + "loss": 0.2465, + "step": 6416 + }, + { + "epoch": 2.69, + "learning_rate": 1.5481652230239954e-05, + "loss": 0.2877, + "step": 6417 + }, + { + "epoch": 2.69, + "learning_rate": 1.5480234162226763e-05, + "loss": 0.2721, + "step": 6418 + }, + { + "epoch": 2.69, + "learning_rate": 1.5478815936689107e-05, + "loss": 0.2494, + "step": 6419 + }, + { + "epoch": 2.69, + "learning_rate": 1.5477397553667745e-05, + "loss": 0.2774, + "step": 6420 + }, + { + "epoch": 2.69, + "learning_rate": 1.5475979013203456e-05, + "loss": 0.2249, + "step": 6421 + }, + { + "epoch": 2.69, + "learning_rate": 1.5474560315337007e-05, + "loss": 0.2835, + "step": 6422 + }, + { + "epoch": 2.69, + "learning_rate": 1.547314146010918e-05, + "loss": 0.2793, + "step": 6423 + }, + { + "epoch": 2.69, + "learning_rate": 1.5471722447560757e-05, + "loss": 0.2306, + "step": 6424 + }, + { + "epoch": 2.69, + "learning_rate": 1.547030327773253e-05, + "loss": 0.2505, + "step": 6425 + }, + { + "epoch": 2.69, + "learning_rate": 1.546888395066529e-05, + "loss": 0.2627, + "step": 6426 + }, + { + "epoch": 2.69, + "learning_rate": 1.546746446639983e-05, + "loss": 0.2706, + "step": 6427 + }, + { + "epoch": 2.69, + "learning_rate": 1.546604482497696e-05, + "loss": 0.2982, + "step": 6428 + }, + { + "epoch": 2.69, + "learning_rate": 1.546462502643748e-05, + "loss": 0.2484, + "step": 6429 + }, + { + "epoch": 2.69, + "learning_rate": 1.5463205070822202e-05, + "loss": 0.2251, + "step": 6430 + }, + { + "epoch": 2.69, + "learning_rate": 1.5461784958171945e-05, + "loss": 0.2504, + "step": 6431 + }, + { + "epoch": 2.69, + "learning_rate": 1.5460364688527523e-05, + "loss": 0.2677, + "step": 6432 + }, + { + "epoch": 2.69, + "learning_rate": 1.5458944261929763e-05, + "loss": 0.243, + "step": 6433 + }, + { + "epoch": 2.69, + "learning_rate": 1.5457523678419498e-05, + "loss": 0.2672, + "step": 6434 + }, + { + "epoch": 2.69, + "learning_rate": 1.5456102938037555e-05, + "loss": 0.2635, + "step": 6435 + }, + { + "epoch": 2.69, + "learning_rate": 1.5454682040824776e-05, + "loss": 0.2272, + "step": 6436 + }, + { + "epoch": 2.69, + "learning_rate": 1.5453260986822e-05, + "loss": 0.2772, + "step": 6437 + }, + { + "epoch": 2.7, + "learning_rate": 1.545183977607008e-05, + "loss": 0.2579, + "step": 6438 + }, + { + "epoch": 2.7, + "learning_rate": 1.545041840860986e-05, + "loss": 0.2872, + "step": 6439 + }, + { + "epoch": 2.7, + "learning_rate": 1.5448996884482198e-05, + "loss": 0.2657, + "step": 6440 + }, + { + "epoch": 2.7, + "learning_rate": 1.5447575203727964e-05, + "loss": 0.2656, + "step": 6441 + }, + { + "epoch": 2.7, + "learning_rate": 1.5446153366388007e-05, + "loss": 0.2635, + "step": 6442 + }, + { + "epoch": 2.7, + "learning_rate": 1.5444731372503206e-05, + "loss": 0.235, + "step": 6443 + }, + { + "epoch": 2.7, + "learning_rate": 1.5443309222114434e-05, + "loss": 0.2488, + "step": 6444 + }, + { + "epoch": 2.7, + "learning_rate": 1.544188691526257e-05, + "loss": 0.2692, + "step": 6445 + }, + { + "epoch": 2.7, + "learning_rate": 1.5440464451988494e-05, + "loss": 0.2598, + "step": 6446 + }, + { + "epoch": 2.7, + "learning_rate": 1.5439041832333095e-05, + "loss": 0.2663, + "step": 6447 + }, + { + "epoch": 2.7, + "learning_rate": 1.5437619056337267e-05, + "loss": 0.2528, + "step": 6448 + }, + { + "epoch": 2.7, + "learning_rate": 1.5436196124041902e-05, + "loss": 0.2391, + "step": 6449 + }, + { + "epoch": 2.7, + "learning_rate": 1.5434773035487905e-05, + "loss": 0.2803, + "step": 6450 + }, + { + "epoch": 2.7, + "learning_rate": 1.543334979071618e-05, + "loss": 0.2625, + "step": 6451 + }, + { + "epoch": 2.7, + "learning_rate": 1.5431926389767637e-05, + "loss": 0.2697, + "step": 6452 + }, + { + "epoch": 2.7, + "learning_rate": 1.543050283268319e-05, + "loss": 0.2431, + "step": 6453 + }, + { + "epoch": 2.7, + "learning_rate": 1.5429079119503752e-05, + "loss": 0.2628, + "step": 6454 + }, + { + "epoch": 2.7, + "learning_rate": 1.542765525027026e-05, + "loss": 0.266, + "step": 6455 + }, + { + "epoch": 2.7, + "learning_rate": 1.542623122502363e-05, + "loss": 0.2715, + "step": 6456 + }, + { + "epoch": 2.7, + "learning_rate": 1.54248070438048e-05, + "loss": 0.2704, + "step": 6457 + }, + { + "epoch": 2.7, + "learning_rate": 1.5423382706654705e-05, + "loss": 0.2164, + "step": 6458 + }, + { + "epoch": 2.7, + "learning_rate": 1.542195821361429e-05, + "loss": 0.2669, + "step": 6459 + }, + { + "epoch": 2.7, + "learning_rate": 1.5420533564724495e-05, + "loss": 0.2869, + "step": 6460 + }, + { + "epoch": 2.71, + "learning_rate": 1.5419108760026273e-05, + "loss": 0.2427, + "step": 6461 + }, + { + "epoch": 2.71, + "learning_rate": 1.5417683799560576e-05, + "loss": 0.248, + "step": 6462 + }, + { + "epoch": 2.71, + "learning_rate": 1.541625868336837e-05, + "loss": 0.2846, + "step": 6463 + }, + { + "epoch": 2.71, + "learning_rate": 1.5414833411490612e-05, + "loss": 0.2726, + "step": 6464 + }, + { + "epoch": 2.71, + "learning_rate": 1.5413407983968274e-05, + "loss": 0.2504, + "step": 6465 + }, + { + "epoch": 2.71, + "learning_rate": 1.5411982400842328e-05, + "loss": 0.239, + "step": 6466 + }, + { + "epoch": 2.71, + "learning_rate": 1.5410556662153747e-05, + "loss": 0.2547, + "step": 6467 + }, + { + "epoch": 2.71, + "learning_rate": 1.5409130767943516e-05, + "loss": 0.284, + "step": 6468 + }, + { + "epoch": 2.71, + "learning_rate": 1.5407704718252625e-05, + "loss": 0.2341, + "step": 6469 + }, + { + "epoch": 2.71, + "learning_rate": 1.5406278513122058e-05, + "loss": 0.2427, + "step": 6470 + }, + { + "epoch": 2.71, + "learning_rate": 1.5404852152592814e-05, + "loss": 0.2431, + "step": 6471 + }, + { + "epoch": 2.71, + "learning_rate": 1.540342563670589e-05, + "loss": 0.2642, + "step": 6472 + }, + { + "epoch": 2.71, + "learning_rate": 1.5401998965502293e-05, + "loss": 0.2666, + "step": 6473 + }, + { + "epoch": 2.71, + "learning_rate": 1.5400572139023026e-05, + "loss": 0.2834, + "step": 6474 + }, + { + "epoch": 2.71, + "learning_rate": 1.5399145157309105e-05, + "loss": 0.281, + "step": 6475 + }, + { + "epoch": 2.71, + "learning_rate": 1.539771802040155e-05, + "loss": 0.2498, + "step": 6476 + }, + { + "epoch": 2.71, + "learning_rate": 1.539629072834138e-05, + "loss": 0.2607, + "step": 6477 + }, + { + "epoch": 2.71, + "learning_rate": 1.539486328116962e-05, + "loss": 0.2705, + "step": 6478 + }, + { + "epoch": 2.71, + "learning_rate": 1.5393435678927305e-05, + "loss": 0.2686, + "step": 6479 + }, + { + "epoch": 2.71, + "learning_rate": 1.5392007921655466e-05, + "loss": 0.2472, + "step": 6480 + }, + { + "epoch": 2.71, + "learning_rate": 1.5390580009395147e-05, + "loss": 0.2202, + "step": 6481 + }, + { + "epoch": 2.71, + "learning_rate": 1.5389151942187383e-05, + "loss": 0.2606, + "step": 6482 + }, + { + "epoch": 2.71, + "learning_rate": 1.538772372007323e-05, + "loss": 0.2618, + "step": 6483 + }, + { + "epoch": 2.71, + "learning_rate": 1.5386295343093743e-05, + "loss": 0.2498, + "step": 6484 + }, + { + "epoch": 2.72, + "learning_rate": 1.5384866811289976e-05, + "loss": 0.263, + "step": 6485 + }, + { + "epoch": 2.72, + "learning_rate": 1.538343812470299e-05, + "loss": 0.2487, + "step": 6486 + }, + { + "epoch": 2.72, + "learning_rate": 1.538200928337385e-05, + "loss": 0.273, + "step": 6487 + }, + { + "epoch": 2.72, + "learning_rate": 1.538058028734363e-05, + "loss": 0.2388, + "step": 6488 + }, + { + "epoch": 2.72, + "learning_rate": 1.5379151136653407e-05, + "loss": 0.2524, + "step": 6489 + }, + { + "epoch": 2.72, + "learning_rate": 1.5377721831344258e-05, + "loss": 0.244, + "step": 6490 + }, + { + "epoch": 2.72, + "learning_rate": 1.537629237145726e-05, + "loss": 0.2628, + "step": 6491 + }, + { + "epoch": 2.72, + "learning_rate": 1.5374862757033518e-05, + "loss": 0.2961, + "step": 6492 + }, + { + "epoch": 2.72, + "learning_rate": 1.5373432988114113e-05, + "loss": 0.2753, + "step": 6493 + }, + { + "epoch": 2.72, + "learning_rate": 1.537200306474014e-05, + "loss": 0.2629, + "step": 6494 + }, + { + "epoch": 2.72, + "learning_rate": 1.5370572986952712e-05, + "loss": 0.2631, + "step": 6495 + }, + { + "epoch": 2.72, + "learning_rate": 1.5369142754792926e-05, + "loss": 0.2519, + "step": 6496 + }, + { + "epoch": 2.72, + "learning_rate": 1.5367712368301897e-05, + "loss": 0.2696, + "step": 6497 + }, + { + "epoch": 2.72, + "learning_rate": 1.536628182752074e-05, + "loss": 0.2714, + "step": 6498 + }, + { + "epoch": 2.72, + "learning_rate": 1.536485113249057e-05, + "loss": 0.2673, + "step": 6499 + }, + { + "epoch": 2.72, + "learning_rate": 1.5363420283252518e-05, + "loss": 0.2689, + "step": 6500 + }, + { + "epoch": 2.72, + "learning_rate": 1.536198927984771e-05, + "loss": 0.2581, + "step": 6501 + }, + { + "epoch": 2.72, + "learning_rate": 1.5360558122317275e-05, + "loss": 0.2747, + "step": 6502 + }, + { + "epoch": 2.72, + "learning_rate": 1.5359126810702357e-05, + "loss": 0.2599, + "step": 6503 + }, + { + "epoch": 2.72, + "learning_rate": 1.5357695345044096e-05, + "loss": 0.2317, + "step": 6504 + }, + { + "epoch": 2.72, + "learning_rate": 1.5356263725383633e-05, + "loss": 0.264, + "step": 6505 + }, + { + "epoch": 2.72, + "learning_rate": 1.5354831951762122e-05, + "loss": 0.2814, + "step": 6506 + }, + { + "epoch": 2.72, + "learning_rate": 1.535340002422072e-05, + "loss": 0.2611, + "step": 6507 + }, + { + "epoch": 2.72, + "learning_rate": 1.5351967942800584e-05, + "loss": 0.2676, + "step": 6508 + }, + { + "epoch": 2.73, + "learning_rate": 1.535053570754288e-05, + "loss": 0.2563, + "step": 6509 + }, + { + "epoch": 2.73, + "learning_rate": 1.5349103318488772e-05, + "loss": 0.2836, + "step": 6510 + }, + { + "epoch": 2.73, + "learning_rate": 1.534767077567944e-05, + "loss": 0.257, + "step": 6511 + }, + { + "epoch": 2.73, + "learning_rate": 1.5346238079156055e-05, + "loss": 0.2543, + "step": 6512 + }, + { + "epoch": 2.73, + "learning_rate": 1.53448052289598e-05, + "loss": 0.2448, + "step": 6513 + }, + { + "epoch": 2.73, + "learning_rate": 1.5343372225131863e-05, + "loss": 0.2426, + "step": 6514 + }, + { + "epoch": 2.73, + "learning_rate": 1.5341939067713433e-05, + "loss": 0.2579, + "step": 6515 + }, + { + "epoch": 2.73, + "learning_rate": 1.5340505756745703e-05, + "loss": 0.2629, + "step": 6516 + }, + { + "epoch": 2.73, + "learning_rate": 1.5339072292269874e-05, + "loss": 0.2614, + "step": 6517 + }, + { + "epoch": 2.73, + "learning_rate": 1.5337638674327152e-05, + "loss": 0.2702, + "step": 6518 + }, + { + "epoch": 2.73, + "learning_rate": 1.5336204902958743e-05, + "loss": 0.249, + "step": 6519 + }, + { + "epoch": 2.73, + "learning_rate": 1.5334770978205857e-05, + "loss": 0.2522, + "step": 6520 + }, + { + "epoch": 2.73, + "learning_rate": 1.5333336900109715e-05, + "loss": 0.2293, + "step": 6521 + }, + { + "epoch": 2.73, + "learning_rate": 1.5331902668711536e-05, + "loss": 0.2922, + "step": 6522 + }, + { + "epoch": 2.73, + "learning_rate": 1.5330468284052544e-05, + "loss": 0.2786, + "step": 6523 + }, + { + "epoch": 2.73, + "learning_rate": 1.5329033746173975e-05, + "loss": 0.2407, + "step": 6524 + }, + { + "epoch": 2.73, + "learning_rate": 1.5327599055117057e-05, + "loss": 0.2681, + "step": 6525 + }, + { + "epoch": 2.73, + "learning_rate": 1.5326164210923032e-05, + "loss": 0.2567, + "step": 6526 + }, + { + "epoch": 2.73, + "learning_rate": 1.5324729213633146e-05, + "loss": 0.268, + "step": 6527 + }, + { + "epoch": 2.73, + "learning_rate": 1.532329406328864e-05, + "loss": 0.2753, + "step": 6528 + }, + { + "epoch": 2.73, + "learning_rate": 1.5321858759930776e-05, + "loss": 0.256, + "step": 6529 + }, + { + "epoch": 2.73, + "learning_rate": 1.53204233036008e-05, + "loss": 0.2613, + "step": 6530 + }, + { + "epoch": 2.73, + "learning_rate": 1.531898769433998e-05, + "loss": 0.2602, + "step": 6531 + }, + { + "epoch": 2.73, + "learning_rate": 1.5317551932189574e-05, + "loss": 0.2707, + "step": 6532 + }, + { + "epoch": 2.74, + "learning_rate": 1.531611601719086e-05, + "loss": 0.2616, + "step": 6533 + }, + { + "epoch": 2.74, + "learning_rate": 1.5314679949385105e-05, + "loss": 0.2707, + "step": 6534 + }, + { + "epoch": 2.74, + "learning_rate": 1.5313243728813598e-05, + "loss": 0.2467, + "step": 6535 + }, + { + "epoch": 2.74, + "learning_rate": 1.5311807355517608e-05, + "loss": 0.257, + "step": 6536 + }, + { + "epoch": 2.74, + "learning_rate": 1.531037082953843e-05, + "loss": 0.2538, + "step": 6537 + }, + { + "epoch": 2.74, + "learning_rate": 1.530893415091736e-05, + "loss": 0.2585, + "step": 6538 + }, + { + "epoch": 2.74, + "learning_rate": 1.5307497319695683e-05, + "loss": 0.2764, + "step": 6539 + }, + { + "epoch": 2.74, + "learning_rate": 1.5306060335914706e-05, + "loss": 0.2576, + "step": 6540 + }, + { + "epoch": 2.74, + "learning_rate": 1.5304623199615737e-05, + "loss": 0.2602, + "step": 6541 + }, + { + "epoch": 2.74, + "learning_rate": 1.530318591084008e-05, + "loss": 0.2584, + "step": 6542 + }, + { + "epoch": 2.74, + "learning_rate": 1.530174846962905e-05, + "loss": 0.2904, + "step": 6543 + }, + { + "epoch": 2.74, + "learning_rate": 1.530031087602396e-05, + "loss": 0.2618, + "step": 6544 + }, + { + "epoch": 2.74, + "learning_rate": 1.5298873130066144e-05, + "loss": 0.272, + "step": 6545 + }, + { + "epoch": 2.74, + "learning_rate": 1.529743523179692e-05, + "loss": 0.2769, + "step": 6546 + }, + { + "epoch": 2.74, + "learning_rate": 1.529599718125762e-05, + "loss": 0.228, + "step": 6547 + }, + { + "epoch": 2.74, + "learning_rate": 1.5294558978489583e-05, + "loss": 0.2458, + "step": 6548 + }, + { + "epoch": 2.74, + "learning_rate": 1.5293120623534145e-05, + "loss": 0.2771, + "step": 6549 + }, + { + "epoch": 2.74, + "learning_rate": 1.529168211643265e-05, + "loss": 0.244, + "step": 6550 + }, + { + "epoch": 2.74, + "learning_rate": 1.529024345722645e-05, + "loss": 0.2994, + "step": 6551 + }, + { + "epoch": 2.74, + "learning_rate": 1.5288804645956896e-05, + "loss": 0.2486, + "step": 6552 + }, + { + "epoch": 2.74, + "learning_rate": 1.5287365682665346e-05, + "loss": 0.2665, + "step": 6553 + }, + { + "epoch": 2.74, + "learning_rate": 1.5285926567393158e-05, + "loss": 0.2559, + "step": 6554 + }, + { + "epoch": 2.74, + "learning_rate": 1.5284487300181707e-05, + "loss": 0.2747, + "step": 6555 + }, + { + "epoch": 2.74, + "learning_rate": 1.5283047881072357e-05, + "loss": 0.2544, + "step": 6556 + }, + { + "epoch": 2.75, + "learning_rate": 1.5281608310106478e-05, + "loss": 0.2588, + "step": 6557 + }, + { + "epoch": 2.75, + "learning_rate": 1.5280168587325462e-05, + "loss": 0.2612, + "step": 6558 + }, + { + "epoch": 2.75, + "learning_rate": 1.5278728712770683e-05, + "loss": 0.2553, + "step": 6559 + }, + { + "epoch": 2.75, + "learning_rate": 1.5277288686483528e-05, + "loss": 0.2601, + "step": 6560 + }, + { + "epoch": 2.75, + "learning_rate": 1.5275848508505396e-05, + "loss": 0.2463, + "step": 6561 + }, + { + "epoch": 2.75, + "learning_rate": 1.527440817887768e-05, + "loss": 0.2555, + "step": 6562 + }, + { + "epoch": 2.75, + "learning_rate": 1.5272967697641784e-05, + "loss": 0.2678, + "step": 6563 + }, + { + "epoch": 2.75, + "learning_rate": 1.5271527064839107e-05, + "loss": 0.265, + "step": 6564 + }, + { + "epoch": 2.75, + "learning_rate": 1.5270086280511064e-05, + "loss": 0.2693, + "step": 6565 + }, + { + "epoch": 2.75, + "learning_rate": 1.5268645344699066e-05, + "loss": 0.2654, + "step": 6566 + }, + { + "epoch": 2.75, + "learning_rate": 1.5267204257444534e-05, + "loss": 0.2538, + "step": 6567 + }, + { + "epoch": 2.75, + "learning_rate": 1.526576301878889e-05, + "loss": 0.2554, + "step": 6568 + }, + { + "epoch": 2.75, + "learning_rate": 1.526432162877356e-05, + "loss": 0.2549, + "step": 6569 + }, + { + "epoch": 2.75, + "learning_rate": 1.5262880087439975e-05, + "loss": 0.2483, + "step": 6570 + }, + { + "epoch": 2.75, + "learning_rate": 1.526143839482957e-05, + "loss": 0.2568, + "step": 6571 + }, + { + "epoch": 2.75, + "learning_rate": 1.5259996550983797e-05, + "loss": 0.259, + "step": 6572 + }, + { + "epoch": 2.75, + "learning_rate": 1.5258554555944085e-05, + "loss": 0.2885, + "step": 6573 + }, + { + "epoch": 2.75, + "learning_rate": 1.5257112409751886e-05, + "loss": 0.2732, + "step": 6574 + }, + { + "epoch": 2.75, + "learning_rate": 1.5255670112448657e-05, + "loss": 0.2344, + "step": 6575 + }, + { + "epoch": 2.75, + "learning_rate": 1.5254227664075858e-05, + "loss": 0.2616, + "step": 6576 + }, + { + "epoch": 2.75, + "learning_rate": 1.5252785064674945e-05, + "loss": 0.27, + "step": 6577 + }, + { + "epoch": 2.75, + "learning_rate": 1.5251342314287385e-05, + "loss": 0.2484, + "step": 6578 + }, + { + "epoch": 2.75, + "learning_rate": 1.5249899412954652e-05, + "loss": 0.2413, + "step": 6579 + }, + { + "epoch": 2.75, + "learning_rate": 1.5248456360718218e-05, + "loss": 0.2837, + "step": 6580 + }, + { + "epoch": 2.76, + "learning_rate": 1.524701315761956e-05, + "loss": 0.2716, + "step": 6581 + }, + { + "epoch": 2.76, + "learning_rate": 1.5245569803700174e-05, + "loss": 0.2518, + "step": 6582 + }, + { + "epoch": 2.76, + "learning_rate": 1.5244126299001534e-05, + "loss": 0.249, + "step": 6583 + }, + { + "epoch": 2.76, + "learning_rate": 1.5242682643565135e-05, + "loss": 0.2523, + "step": 6584 + }, + { + "epoch": 2.76, + "learning_rate": 1.5241238837432477e-05, + "loss": 0.2322, + "step": 6585 + }, + { + "epoch": 2.76, + "learning_rate": 1.523979488064506e-05, + "loss": 0.2686, + "step": 6586 + }, + { + "epoch": 2.76, + "learning_rate": 1.523835077324439e-05, + "loss": 0.2603, + "step": 6587 + }, + { + "epoch": 2.76, + "learning_rate": 1.5236906515271974e-05, + "loss": 0.2674, + "step": 6588 + }, + { + "epoch": 2.76, + "learning_rate": 1.5235462106769327e-05, + "loss": 0.2468, + "step": 6589 + }, + { + "epoch": 2.76, + "learning_rate": 1.5234017547777969e-05, + "loss": 0.2629, + "step": 6590 + }, + { + "epoch": 2.76, + "learning_rate": 1.5232572838339418e-05, + "loss": 0.2581, + "step": 6591 + }, + { + "epoch": 2.76, + "learning_rate": 1.5231127978495208e-05, + "loss": 0.2561, + "step": 6592 + }, + { + "epoch": 2.76, + "learning_rate": 1.5229682968286862e-05, + "loss": 0.2629, + "step": 6593 + }, + { + "epoch": 2.76, + "learning_rate": 1.5228237807755925e-05, + "loss": 0.254, + "step": 6594 + }, + { + "epoch": 2.76, + "learning_rate": 1.522679249694393e-05, + "loss": 0.2554, + "step": 6595 + }, + { + "epoch": 2.76, + "learning_rate": 1.5225347035892423e-05, + "loss": 0.2422, + "step": 6596 + }, + { + "epoch": 2.76, + "learning_rate": 1.522390142464295e-05, + "loss": 0.2471, + "step": 6597 + }, + { + "epoch": 2.76, + "learning_rate": 1.5222455663237068e-05, + "loss": 0.2608, + "step": 6598 + }, + { + "epoch": 2.76, + "learning_rate": 1.5221009751716333e-05, + "loss": 0.2835, + "step": 6599 + }, + { + "epoch": 2.76, + "learning_rate": 1.5219563690122306e-05, + "loss": 0.2757, + "step": 6600 + }, + { + "epoch": 2.76, + "learning_rate": 1.5218117478496548e-05, + "loss": 0.2538, + "step": 6601 + }, + { + "epoch": 2.76, + "learning_rate": 1.5216671116880637e-05, + "loss": 0.274, + "step": 6602 + }, + { + "epoch": 2.76, + "learning_rate": 1.5215224605316143e-05, + "loss": 0.2488, + "step": 6603 + }, + { + "epoch": 2.76, + "learning_rate": 1.5213777943844648e-05, + "loss": 0.2378, + "step": 6604 + }, + { + "epoch": 2.77, + "learning_rate": 1.521233113250773e-05, + "loss": 0.2754, + "step": 6605 + }, + { + "epoch": 2.77, + "learning_rate": 1.5210884171346983e-05, + "loss": 0.263, + "step": 6606 + }, + { + "epoch": 2.77, + "learning_rate": 1.5209437060403988e-05, + "loss": 0.2472, + "step": 6607 + }, + { + "epoch": 2.77, + "learning_rate": 1.5207989799720352e-05, + "loss": 0.2798, + "step": 6608 + }, + { + "epoch": 2.77, + "learning_rate": 1.520654238933767e-05, + "loss": 0.2764, + "step": 6609 + }, + { + "epoch": 2.77, + "learning_rate": 1.5205094829297546e-05, + "loss": 0.2551, + "step": 6610 + }, + { + "epoch": 2.77, + "learning_rate": 1.5203647119641591e-05, + "loss": 0.2434, + "step": 6611 + }, + { + "epoch": 2.77, + "learning_rate": 1.5202199260411417e-05, + "loss": 0.2527, + "step": 6612 + }, + { + "epoch": 2.77, + "learning_rate": 1.5200751251648641e-05, + "loss": 0.2886, + "step": 6613 + }, + { + "epoch": 2.77, + "learning_rate": 1.5199303093394883e-05, + "loss": 0.2617, + "step": 6614 + }, + { + "epoch": 2.77, + "learning_rate": 1.5197854785691775e-05, + "loss": 0.2527, + "step": 6615 + }, + { + "epoch": 2.77, + "learning_rate": 1.5196406328580943e-05, + "loss": 0.273, + "step": 6616 + }, + { + "epoch": 2.77, + "learning_rate": 1.5194957722104024e-05, + "loss": 0.2436, + "step": 6617 + }, + { + "epoch": 2.77, + "learning_rate": 1.5193508966302655e-05, + "loss": 0.2875, + "step": 6618 + }, + { + "epoch": 2.77, + "learning_rate": 1.5192060061218477e-05, + "loss": 0.2511, + "step": 6619 + }, + { + "epoch": 2.77, + "learning_rate": 1.519061100689314e-05, + "loss": 0.2614, + "step": 6620 + }, + { + "epoch": 2.77, + "learning_rate": 1.5189161803368297e-05, + "loss": 0.2674, + "step": 6621 + }, + { + "epoch": 2.77, + "learning_rate": 1.51877124506856e-05, + "loss": 0.2546, + "step": 6622 + }, + { + "epoch": 2.77, + "learning_rate": 1.5186262948886716e-05, + "loss": 0.2376, + "step": 6623 + }, + { + "epoch": 2.77, + "learning_rate": 1.5184813298013304e-05, + "loss": 0.243, + "step": 6624 + }, + { + "epoch": 2.77, + "learning_rate": 1.5183363498107033e-05, + "loss": 0.2854, + "step": 6625 + }, + { + "epoch": 2.77, + "learning_rate": 1.5181913549209582e-05, + "loss": 0.2663, + "step": 6626 + }, + { + "epoch": 2.77, + "learning_rate": 1.5180463451362621e-05, + "loss": 0.2811, + "step": 6627 + }, + { + "epoch": 2.77, + "learning_rate": 1.5179013204607838e-05, + "loss": 0.2553, + "step": 6628 + }, + { + "epoch": 2.78, + "learning_rate": 1.5177562808986914e-05, + "loss": 0.2549, + "step": 6629 + }, + { + "epoch": 2.78, + "learning_rate": 1.5176112264541542e-05, + "loss": 0.249, + "step": 6630 + }, + { + "epoch": 2.78, + "learning_rate": 1.5174661571313417e-05, + "loss": 0.2608, + "step": 6631 + }, + { + "epoch": 2.78, + "learning_rate": 1.5173210729344234e-05, + "loss": 0.2462, + "step": 6632 + }, + { + "epoch": 2.78, + "learning_rate": 1.5171759738675702e-05, + "loss": 0.2513, + "step": 6633 + }, + { + "epoch": 2.78, + "learning_rate": 1.5170308599349524e-05, + "loss": 0.2768, + "step": 6634 + }, + { + "epoch": 2.78, + "learning_rate": 1.5168857311407411e-05, + "loss": 0.2487, + "step": 6635 + }, + { + "epoch": 2.78, + "learning_rate": 1.516740587489109e-05, + "loss": 0.2677, + "step": 6636 + }, + { + "epoch": 2.78, + "learning_rate": 1.5165954289842267e-05, + "loss": 0.2875, + "step": 6637 + }, + { + "epoch": 2.78, + "learning_rate": 1.516450255630267e-05, + "loss": 0.2793, + "step": 6638 + }, + { + "epoch": 2.78, + "learning_rate": 1.5163050674314033e-05, + "loss": 0.2797, + "step": 6639 + }, + { + "epoch": 2.78, + "learning_rate": 1.5161598643918086e-05, + "loss": 0.2556, + "step": 6640 + }, + { + "epoch": 2.78, + "learning_rate": 1.5160146465156564e-05, + "loss": 0.2648, + "step": 6641 + }, + { + "epoch": 2.78, + "learning_rate": 1.5158694138071213e-05, + "loss": 0.2395, + "step": 6642 + }, + { + "epoch": 2.78, + "learning_rate": 1.5157241662703777e-05, + "loss": 0.2652, + "step": 6643 + }, + { + "epoch": 2.78, + "learning_rate": 1.5155789039096005e-05, + "loss": 0.2498, + "step": 6644 + }, + { + "epoch": 2.78, + "learning_rate": 1.515433626728965e-05, + "loss": 0.2564, + "step": 6645 + }, + { + "epoch": 2.78, + "learning_rate": 1.5152883347326475e-05, + "loss": 0.2747, + "step": 6646 + }, + { + "epoch": 2.78, + "learning_rate": 1.5151430279248242e-05, + "loss": 0.2571, + "step": 6647 + }, + { + "epoch": 2.78, + "learning_rate": 1.5149977063096715e-05, + "loss": 0.2651, + "step": 6648 + }, + { + "epoch": 2.78, + "learning_rate": 1.5148523698913668e-05, + "loss": 0.2875, + "step": 6649 + }, + { + "epoch": 2.78, + "learning_rate": 1.5147070186740877e-05, + "loss": 0.2477, + "step": 6650 + }, + { + "epoch": 2.78, + "learning_rate": 1.514561652662012e-05, + "loss": 0.2882, + "step": 6651 + }, + { + "epoch": 2.79, + "learning_rate": 1.5144162718593183e-05, + "loss": 0.2645, + "step": 6652 + }, + { + "epoch": 2.79, + "learning_rate": 1.5142708762701852e-05, + "loss": 0.279, + "step": 6653 + }, + { + "epoch": 2.79, + "learning_rate": 1.5141254658987922e-05, + "loss": 0.2428, + "step": 6654 + }, + { + "epoch": 2.79, + "learning_rate": 1.5139800407493189e-05, + "loss": 0.2613, + "step": 6655 + }, + { + "epoch": 2.79, + "learning_rate": 1.5138346008259452e-05, + "loss": 0.2361, + "step": 6656 + }, + { + "epoch": 2.79, + "learning_rate": 1.5136891461328525e-05, + "loss": 0.2499, + "step": 6657 + }, + { + "epoch": 2.79, + "learning_rate": 1.5135436766742207e-05, + "loss": 0.2513, + "step": 6658 + }, + { + "epoch": 2.79, + "learning_rate": 1.5133981924542314e-05, + "loss": 0.2727, + "step": 6659 + }, + { + "epoch": 2.79, + "learning_rate": 1.513252693477067e-05, + "loss": 0.2188, + "step": 6660 + }, + { + "epoch": 2.79, + "learning_rate": 1.5131071797469093e-05, + "loss": 0.2486, + "step": 6661 + }, + { + "epoch": 2.79, + "learning_rate": 1.512961651267941e-05, + "loss": 0.2672, + "step": 6662 + }, + { + "epoch": 2.79, + "learning_rate": 1.5128161080443451e-05, + "loss": 0.2455, + "step": 6663 + }, + { + "epoch": 2.79, + "learning_rate": 1.5126705500803055e-05, + "loss": 0.2667, + "step": 6664 + }, + { + "epoch": 2.79, + "learning_rate": 1.5125249773800057e-05, + "loss": 0.2744, + "step": 6665 + }, + { + "epoch": 2.79, + "learning_rate": 1.5123793899476303e-05, + "loss": 0.2415, + "step": 6666 + }, + { + "epoch": 2.79, + "learning_rate": 1.5122337877873641e-05, + "loss": 0.2695, + "step": 6667 + }, + { + "epoch": 2.79, + "learning_rate": 1.5120881709033919e-05, + "loss": 0.2449, + "step": 6668 + }, + { + "epoch": 2.79, + "learning_rate": 1.5119425392999e-05, + "loss": 0.2488, + "step": 6669 + }, + { + "epoch": 2.79, + "learning_rate": 1.511796892981074e-05, + "loss": 0.2597, + "step": 6670 + }, + { + "epoch": 2.79, + "learning_rate": 1.5116512319511003e-05, + "loss": 0.253, + "step": 6671 + }, + { + "epoch": 2.79, + "learning_rate": 1.5115055562141661e-05, + "loss": 0.2759, + "step": 6672 + }, + { + "epoch": 2.79, + "learning_rate": 1.5113598657744585e-05, + "loss": 0.3121, + "step": 6673 + }, + { + "epoch": 2.79, + "learning_rate": 1.5112141606361657e-05, + "loss": 0.2661, + "step": 6674 + }, + { + "epoch": 2.79, + "learning_rate": 1.511068440803475e-05, + "loss": 0.2716, + "step": 6675 + }, + { + "epoch": 2.8, + "learning_rate": 1.5109227062805758e-05, + "loss": 0.2753, + "step": 6676 + }, + { + "epoch": 2.8, + "learning_rate": 1.5107769570716566e-05, + "loss": 0.2742, + "step": 6677 + }, + { + "epoch": 2.8, + "learning_rate": 1.510631193180907e-05, + "loss": 0.2392, + "step": 6678 + }, + { + "epoch": 2.8, + "learning_rate": 1.510485414612517e-05, + "loss": 0.2766, + "step": 6679 + }, + { + "epoch": 2.8, + "learning_rate": 1.510339621370677e-05, + "loss": 0.2514, + "step": 6680 + }, + { + "epoch": 2.8, + "learning_rate": 1.510193813459577e-05, + "loss": 0.2646, + "step": 6681 + }, + { + "epoch": 2.8, + "learning_rate": 1.5100479908834087e-05, + "loss": 0.2824, + "step": 6682 + }, + { + "epoch": 2.8, + "learning_rate": 1.5099021536463636e-05, + "loss": 0.2641, + "step": 6683 + }, + { + "epoch": 2.8, + "learning_rate": 1.5097563017526336e-05, + "loss": 0.2753, + "step": 6684 + }, + { + "epoch": 2.8, + "learning_rate": 1.5096104352064112e-05, + "loss": 0.2636, + "step": 6685 + }, + { + "epoch": 2.8, + "learning_rate": 1.5094645540118888e-05, + "loss": 0.2799, + "step": 6686 + }, + { + "epoch": 2.8, + "learning_rate": 1.5093186581732599e-05, + "loss": 0.2483, + "step": 6687 + }, + { + "epoch": 2.8, + "learning_rate": 1.5091727476947182e-05, + "loss": 0.2439, + "step": 6688 + }, + { + "epoch": 2.8, + "learning_rate": 1.5090268225804577e-05, + "loss": 0.25, + "step": 6689 + }, + { + "epoch": 2.8, + "learning_rate": 1.5088808828346727e-05, + "loss": 0.2747, + "step": 6690 + }, + { + "epoch": 2.8, + "learning_rate": 1.5087349284615584e-05, + "loss": 0.255, + "step": 6691 + }, + { + "epoch": 2.8, + "learning_rate": 1.50858895946531e-05, + "loss": 0.2375, + "step": 6692 + }, + { + "epoch": 2.8, + "learning_rate": 1.5084429758501235e-05, + "loss": 0.2505, + "step": 6693 + }, + { + "epoch": 2.8, + "learning_rate": 1.5082969776201948e-05, + "loss": 0.2722, + "step": 6694 + }, + { + "epoch": 2.8, + "learning_rate": 1.5081509647797204e-05, + "loss": 0.2526, + "step": 6695 + }, + { + "epoch": 2.8, + "learning_rate": 1.5080049373328976e-05, + "loss": 0.263, + "step": 6696 + }, + { + "epoch": 2.8, + "learning_rate": 1.5078588952839235e-05, + "loss": 0.2623, + "step": 6697 + }, + { + "epoch": 2.8, + "learning_rate": 1.507712838636996e-05, + "loss": 0.245, + "step": 6698 + }, + { + "epoch": 2.8, + "learning_rate": 1.5075667673963139e-05, + "loss": 0.2681, + "step": 6699 + }, + { + "epoch": 2.81, + "learning_rate": 1.507420681566075e-05, + "loss": 0.2744, + "step": 6700 + }, + { + "epoch": 2.81, + "learning_rate": 1.5072745811504793e-05, + "loss": 0.2359, + "step": 6701 + }, + { + "epoch": 2.81, + "learning_rate": 1.5071284661537257e-05, + "loss": 0.2707, + "step": 6702 + }, + { + "epoch": 2.81, + "learning_rate": 1.5069823365800146e-05, + "loss": 0.2543, + "step": 6703 + }, + { + "epoch": 2.81, + "learning_rate": 1.5068361924335457e-05, + "loss": 0.2385, + "step": 6704 + }, + { + "epoch": 2.81, + "learning_rate": 1.5066900337185203e-05, + "loss": 0.2427, + "step": 6705 + }, + { + "epoch": 2.81, + "learning_rate": 1.5065438604391398e-05, + "loss": 0.2722, + "step": 6706 + }, + { + "epoch": 2.81, + "learning_rate": 1.5063976725996052e-05, + "loss": 0.2621, + "step": 6707 + }, + { + "epoch": 2.81, + "learning_rate": 1.5062514702041193e-05, + "loss": 0.253, + "step": 6708 + }, + { + "epoch": 2.81, + "learning_rate": 1.5061052532568838e-05, + "loss": 0.2611, + "step": 6709 + }, + { + "epoch": 2.81, + "learning_rate": 1.5059590217621018e-05, + "loss": 0.2644, + "step": 6710 + }, + { + "epoch": 2.81, + "learning_rate": 1.5058127757239772e-05, + "loss": 0.2422, + "step": 6711 + }, + { + "epoch": 2.81, + "learning_rate": 1.5056665151467126e-05, + "loss": 0.2587, + "step": 6712 + }, + { + "epoch": 2.81, + "learning_rate": 1.505520240034513e-05, + "loss": 0.2658, + "step": 6713 + }, + { + "epoch": 2.81, + "learning_rate": 1.505373950391583e-05, + "loss": 0.248, + "step": 6714 + }, + { + "epoch": 2.81, + "learning_rate": 1.5052276462221267e-05, + "loss": 0.3038, + "step": 6715 + }, + { + "epoch": 2.81, + "learning_rate": 1.5050813275303501e-05, + "loss": 0.2474, + "step": 6716 + }, + { + "epoch": 2.81, + "learning_rate": 1.5049349943204595e-05, + "loss": 0.2528, + "step": 6717 + }, + { + "epoch": 2.81, + "learning_rate": 1.50478864659666e-05, + "loss": 0.2764, + "step": 6718 + }, + { + "epoch": 2.81, + "learning_rate": 1.5046422843631594e-05, + "loss": 0.2536, + "step": 6719 + }, + { + "epoch": 2.81, + "learning_rate": 1.5044959076241634e-05, + "loss": 0.246, + "step": 6720 + }, + { + "epoch": 2.81, + "learning_rate": 1.5043495163838809e-05, + "loss": 0.2581, + "step": 6721 + }, + { + "epoch": 2.81, + "learning_rate": 1.5042031106465186e-05, + "loss": 0.2556, + "step": 6722 + }, + { + "epoch": 2.81, + "learning_rate": 1.5040566904162855e-05, + "loss": 0.2449, + "step": 6723 + }, + { + "epoch": 2.82, + "learning_rate": 1.5039102556973899e-05, + "loss": 0.2626, + "step": 6724 + }, + { + "epoch": 2.82, + "learning_rate": 1.5037638064940416e-05, + "loss": 0.2653, + "step": 6725 + }, + { + "epoch": 2.82, + "learning_rate": 1.5036173428104493e-05, + "loss": 0.256, + "step": 6726 + }, + { + "epoch": 2.82, + "learning_rate": 1.5034708646508236e-05, + "loss": 0.2708, + "step": 6727 + }, + { + "epoch": 2.82, + "learning_rate": 1.5033243720193746e-05, + "loss": 0.2576, + "step": 6728 + }, + { + "epoch": 2.82, + "learning_rate": 1.5031778649203133e-05, + "loss": 0.2685, + "step": 6729 + }, + { + "epoch": 2.82, + "learning_rate": 1.5030313433578508e-05, + "loss": 0.2787, + "step": 6730 + }, + { + "epoch": 2.82, + "learning_rate": 1.5028848073361987e-05, + "loss": 0.2648, + "step": 6731 + }, + { + "epoch": 2.82, + "learning_rate": 1.5027382568595688e-05, + "loss": 0.2672, + "step": 6732 + }, + { + "epoch": 2.82, + "learning_rate": 1.5025916919321741e-05, + "loss": 0.2528, + "step": 6733 + }, + { + "epoch": 2.82, + "learning_rate": 1.5024451125582271e-05, + "loss": 0.2618, + "step": 6734 + }, + { + "epoch": 2.82, + "learning_rate": 1.5022985187419411e-05, + "loss": 0.2609, + "step": 6735 + }, + { + "epoch": 2.82, + "learning_rate": 1.50215191048753e-05, + "loss": 0.254, + "step": 6736 + }, + { + "epoch": 2.82, + "learning_rate": 1.5020052877992077e-05, + "loss": 0.2502, + "step": 6737 + }, + { + "epoch": 2.82, + "learning_rate": 1.5018586506811893e-05, + "loss": 0.24, + "step": 6738 + }, + { + "epoch": 2.82, + "learning_rate": 1.501711999137689e-05, + "loss": 0.2637, + "step": 6739 + }, + { + "epoch": 2.82, + "learning_rate": 1.5015653331729225e-05, + "loss": 0.2965, + "step": 6740 + }, + { + "epoch": 2.82, + "learning_rate": 1.5014186527911056e-05, + "loss": 0.2677, + "step": 6741 + }, + { + "epoch": 2.82, + "learning_rate": 1.5012719579964548e-05, + "loss": 0.2725, + "step": 6742 + }, + { + "epoch": 2.82, + "learning_rate": 1.501125248793186e-05, + "loss": 0.2443, + "step": 6743 + }, + { + "epoch": 2.82, + "learning_rate": 1.5009785251855164e-05, + "loss": 0.2601, + "step": 6744 + }, + { + "epoch": 2.82, + "learning_rate": 1.5008317871776643e-05, + "loss": 0.241, + "step": 6745 + }, + { + "epoch": 2.82, + "learning_rate": 1.5006850347738463e-05, + "loss": 0.2925, + "step": 6746 + }, + { + "epoch": 2.82, + "learning_rate": 1.5005382679782817e-05, + "loss": 0.26, + "step": 6747 + }, + { + "epoch": 2.83, + "learning_rate": 1.5003914867951888e-05, + "loss": 0.2512, + "step": 6748 + }, + { + "epoch": 2.83, + "learning_rate": 1.5002446912287862e-05, + "loss": 0.2741, + "step": 6749 + }, + { + "epoch": 2.83, + "learning_rate": 1.5000978812832942e-05, + "loss": 0.2485, + "step": 6750 + }, + { + "epoch": 2.83, + "learning_rate": 1.4999510569629323e-05, + "loss": 0.2567, + "step": 6751 + }, + { + "epoch": 2.83, + "learning_rate": 1.4998042182719207e-05, + "loss": 0.2542, + "step": 6752 + }, + { + "epoch": 2.83, + "learning_rate": 1.4996573652144807e-05, + "loss": 0.2555, + "step": 6753 + }, + { + "epoch": 2.83, + "learning_rate": 1.4995104977948331e-05, + "loss": 0.2351, + "step": 6754 + }, + { + "epoch": 2.83, + "learning_rate": 1.4993636160171994e-05, + "loss": 0.2558, + "step": 6755 + }, + { + "epoch": 2.83, + "learning_rate": 1.4992167198858016e-05, + "loss": 0.2562, + "step": 6756 + }, + { + "epoch": 2.83, + "learning_rate": 1.4990698094048624e-05, + "loss": 0.2462, + "step": 6757 + }, + { + "epoch": 2.83, + "learning_rate": 1.498922884578604e-05, + "loss": 0.2662, + "step": 6758 + }, + { + "epoch": 2.83, + "learning_rate": 1.4987759454112505e-05, + "loss": 0.2444, + "step": 6759 + }, + { + "epoch": 2.83, + "learning_rate": 1.4986289919070247e-05, + "loss": 0.2565, + "step": 6760 + }, + { + "epoch": 2.83, + "learning_rate": 1.4984820240701513e-05, + "loss": 0.2805, + "step": 6761 + }, + { + "epoch": 2.83, + "learning_rate": 1.4983350419048544e-05, + "loss": 0.254, + "step": 6762 + }, + { + "epoch": 2.83, + "learning_rate": 1.4981880454153586e-05, + "loss": 0.2438, + "step": 6763 + }, + { + "epoch": 2.83, + "learning_rate": 1.49804103460589e-05, + "loss": 0.2595, + "step": 6764 + }, + { + "epoch": 2.83, + "learning_rate": 1.4978940094806733e-05, + "loss": 0.254, + "step": 6765 + }, + { + "epoch": 2.83, + "learning_rate": 1.4977469700439358e-05, + "loss": 0.2385, + "step": 6766 + }, + { + "epoch": 2.83, + "learning_rate": 1.4975999162999027e-05, + "loss": 0.2545, + "step": 6767 + }, + { + "epoch": 2.83, + "learning_rate": 1.4974528482528018e-05, + "loss": 0.2445, + "step": 6768 + }, + { + "epoch": 2.83, + "learning_rate": 1.4973057659068601e-05, + "loss": 0.2655, + "step": 6769 + }, + { + "epoch": 2.83, + "learning_rate": 1.4971586692663055e-05, + "loss": 0.2913, + "step": 6770 + }, + { + "epoch": 2.83, + "learning_rate": 1.4970115583353661e-05, + "loss": 0.2701, + "step": 6771 + }, + { + "epoch": 2.84, + "learning_rate": 1.496864433118271e-05, + "loss": 0.263, + "step": 6772 + }, + { + "epoch": 2.84, + "learning_rate": 1.496717293619248e-05, + "loss": 0.2663, + "step": 6773 + }, + { + "epoch": 2.84, + "learning_rate": 1.4965701398425275e-05, + "loss": 0.2575, + "step": 6774 + }, + { + "epoch": 2.84, + "learning_rate": 1.496422971792339e-05, + "loss": 0.2518, + "step": 6775 + }, + { + "epoch": 2.84, + "learning_rate": 1.4962757894729128e-05, + "loss": 0.2548, + "step": 6776 + }, + { + "epoch": 2.84, + "learning_rate": 1.496128592888479e-05, + "loss": 0.2347, + "step": 6777 + }, + { + "epoch": 2.84, + "learning_rate": 1.4959813820432693e-05, + "loss": 0.2599, + "step": 6778 + }, + { + "epoch": 2.84, + "learning_rate": 1.4958341569415149e-05, + "loss": 0.2783, + "step": 6779 + }, + { + "epoch": 2.84, + "learning_rate": 1.4956869175874475e-05, + "loss": 0.2505, + "step": 6780 + }, + { + "epoch": 2.84, + "learning_rate": 1.4955396639852996e-05, + "loss": 0.2879, + "step": 6781 + }, + { + "epoch": 2.84, + "learning_rate": 1.4953923961393043e-05, + "loss": 0.251, + "step": 6782 + }, + { + "epoch": 2.84, + "learning_rate": 1.4952451140536934e-05, + "loss": 0.2481, + "step": 6783 + }, + { + "epoch": 2.84, + "learning_rate": 1.4950978177327016e-05, + "loss": 0.2704, + "step": 6784 + }, + { + "epoch": 2.84, + "learning_rate": 1.4949505071805624e-05, + "loss": 0.2769, + "step": 6785 + }, + { + "epoch": 2.84, + "learning_rate": 1.4948031824015098e-05, + "loss": 0.2836, + "step": 6786 + }, + { + "epoch": 2.84, + "learning_rate": 1.4946558433997792e-05, + "loss": 0.2574, + "step": 6787 + }, + { + "epoch": 2.84, + "learning_rate": 1.4945084901796048e-05, + "loss": 0.2546, + "step": 6788 + }, + { + "epoch": 2.84, + "learning_rate": 1.494361122745223e-05, + "loss": 0.2575, + "step": 6789 + }, + { + "epoch": 2.84, + "learning_rate": 1.4942137411008696e-05, + "loss": 0.2564, + "step": 6790 + }, + { + "epoch": 2.84, + "learning_rate": 1.4940663452507802e-05, + "loss": 0.2227, + "step": 6791 + }, + { + "epoch": 2.84, + "learning_rate": 1.4939189351991928e-05, + "loss": 0.2171, + "step": 6792 + }, + { + "epoch": 2.84, + "learning_rate": 1.4937715109503434e-05, + "loss": 0.2628, + "step": 6793 + }, + { + "epoch": 2.84, + "learning_rate": 1.4936240725084701e-05, + "loss": 0.256, + "step": 6794 + }, + { + "epoch": 2.84, + "learning_rate": 1.4934766198778111e-05, + "loss": 0.2534, + "step": 6795 + }, + { + "epoch": 2.85, + "learning_rate": 1.4933291530626047e-05, + "loss": 0.279, + "step": 6796 + }, + { + "epoch": 2.85, + "learning_rate": 1.493181672067089e-05, + "loss": 0.2554, + "step": 6797 + }, + { + "epoch": 2.85, + "learning_rate": 1.4930341768955043e-05, + "loss": 0.233, + "step": 6798 + }, + { + "epoch": 2.85, + "learning_rate": 1.4928866675520897e-05, + "loss": 0.2738, + "step": 6799 + }, + { + "epoch": 2.85, + "learning_rate": 1.492739144041085e-05, + "loss": 0.2688, + "step": 6800 + }, + { + "epoch": 2.85, + "learning_rate": 1.4925916063667309e-05, + "loss": 0.2855, + "step": 6801 + }, + { + "epoch": 2.85, + "learning_rate": 1.4924440545332681e-05, + "loss": 0.2513, + "step": 6802 + }, + { + "epoch": 2.85, + "learning_rate": 1.4922964885449382e-05, + "loss": 0.2448, + "step": 6803 + }, + { + "epoch": 2.85, + "learning_rate": 1.4921489084059823e-05, + "loss": 0.2518, + "step": 6804 + }, + { + "epoch": 2.85, + "learning_rate": 1.4920013141206428e-05, + "loss": 0.2839, + "step": 6805 + }, + { + "epoch": 2.85, + "learning_rate": 1.4918537056931623e-05, + "loss": 0.2599, + "step": 6806 + }, + { + "epoch": 2.85, + "learning_rate": 1.4917060831277835e-05, + "loss": 0.2528, + "step": 6807 + }, + { + "epoch": 2.85, + "learning_rate": 1.4915584464287494e-05, + "loss": 0.2628, + "step": 6808 + }, + { + "epoch": 2.85, + "learning_rate": 1.4914107956003044e-05, + "loss": 0.2673, + "step": 6809 + }, + { + "epoch": 2.85, + "learning_rate": 1.4912631306466918e-05, + "loss": 0.243, + "step": 6810 + }, + { + "epoch": 2.85, + "learning_rate": 1.4911154515721564e-05, + "loss": 0.2547, + "step": 6811 + }, + { + "epoch": 2.85, + "learning_rate": 1.4909677583809435e-05, + "loss": 0.2695, + "step": 6812 + }, + { + "epoch": 2.85, + "learning_rate": 1.4908200510772979e-05, + "loss": 0.2457, + "step": 6813 + }, + { + "epoch": 2.85, + "learning_rate": 1.4906723296654652e-05, + "loss": 0.2445, + "step": 6814 + }, + { + "epoch": 2.85, + "learning_rate": 1.490524594149692e-05, + "loss": 0.2652, + "step": 6815 + }, + { + "epoch": 2.85, + "learning_rate": 1.4903768445342248e-05, + "loss": 0.2757, + "step": 6816 + }, + { + "epoch": 2.85, + "learning_rate": 1.4902290808233103e-05, + "loss": 0.2322, + "step": 6817 + }, + { + "epoch": 2.85, + "learning_rate": 1.4900813030211956e-05, + "loss": 0.2429, + "step": 6818 + }, + { + "epoch": 2.85, + "learning_rate": 1.4899335111321289e-05, + "loss": 0.2251, + "step": 6819 + }, + { + "epoch": 2.86, + "learning_rate": 1.489785705160358e-05, + "loss": 0.2629, + "step": 6820 + }, + { + "epoch": 2.86, + "learning_rate": 1.4896378851101321e-05, + "loss": 0.2627, + "step": 6821 + }, + { + "epoch": 2.86, + "learning_rate": 1.4894900509856995e-05, + "loss": 0.2647, + "step": 6822 + }, + { + "epoch": 2.86, + "learning_rate": 1.4893422027913099e-05, + "loss": 0.2637, + "step": 6823 + }, + { + "epoch": 2.86, + "learning_rate": 1.4891943405312126e-05, + "loss": 0.25, + "step": 6824 + }, + { + "epoch": 2.86, + "learning_rate": 1.489046464209658e-05, + "loss": 0.2694, + "step": 6825 + }, + { + "epoch": 2.86, + "learning_rate": 1.488898573830897e-05, + "loss": 0.2741, + "step": 6826 + }, + { + "epoch": 2.86, + "learning_rate": 1.4887506693991803e-05, + "loss": 0.269, + "step": 6827 + }, + { + "epoch": 2.86, + "learning_rate": 1.4886027509187593e-05, + "loss": 0.2546, + "step": 6828 + }, + { + "epoch": 2.86, + "learning_rate": 1.4884548183938857e-05, + "loss": 0.2611, + "step": 6829 + }, + { + "epoch": 2.86, + "learning_rate": 1.4883068718288121e-05, + "loss": 0.2603, + "step": 6830 + }, + { + "epoch": 2.86, + "learning_rate": 1.4881589112277903e-05, + "loss": 0.2771, + "step": 6831 + }, + { + "epoch": 2.86, + "learning_rate": 1.4880109365950744e-05, + "loss": 0.2664, + "step": 6832 + }, + { + "epoch": 2.86, + "learning_rate": 1.4878629479349167e-05, + "loss": 0.2743, + "step": 6833 + }, + { + "epoch": 2.86, + "learning_rate": 1.4877149452515718e-05, + "loss": 0.2375, + "step": 6834 + }, + { + "epoch": 2.86, + "learning_rate": 1.4875669285492933e-05, + "loss": 0.2422, + "step": 6835 + }, + { + "epoch": 2.86, + "learning_rate": 1.487418897832336e-05, + "loss": 0.2433, + "step": 6836 + }, + { + "epoch": 2.86, + "learning_rate": 1.4872708531049554e-05, + "loss": 0.2557, + "step": 6837 + }, + { + "epoch": 2.86, + "learning_rate": 1.4871227943714063e-05, + "loss": 0.2563, + "step": 6838 + }, + { + "epoch": 2.86, + "learning_rate": 1.4869747216359446e-05, + "loss": 0.2676, + "step": 6839 + }, + { + "epoch": 2.86, + "learning_rate": 1.486826634902827e-05, + "loss": 0.2543, + "step": 6840 + }, + { + "epoch": 2.86, + "learning_rate": 1.4866785341763097e-05, + "loss": 0.2639, + "step": 6841 + }, + { + "epoch": 2.86, + "learning_rate": 1.4865304194606496e-05, + "loss": 0.2607, + "step": 6842 + }, + { + "epoch": 2.86, + "learning_rate": 1.4863822907601046e-05, + "loss": 0.2511, + "step": 6843 + }, + { + "epoch": 2.87, + "learning_rate": 1.4862341480789322e-05, + "loss": 0.2529, + "step": 6844 + }, + { + "epoch": 2.87, + "learning_rate": 1.4860859914213905e-05, + "loss": 0.2363, + "step": 6845 + }, + { + "epoch": 2.87, + "learning_rate": 1.4859378207917385e-05, + "loss": 0.2423, + "step": 6846 + }, + { + "epoch": 2.87, + "learning_rate": 1.485789636194235e-05, + "loss": 0.2616, + "step": 6847 + }, + { + "epoch": 2.87, + "learning_rate": 1.4856414376331394e-05, + "loss": 0.264, + "step": 6848 + }, + { + "epoch": 2.87, + "learning_rate": 1.4854932251127117e-05, + "loss": 0.2839, + "step": 6849 + }, + { + "epoch": 2.87, + "learning_rate": 1.4853449986372122e-05, + "loss": 0.253, + "step": 6850 + }, + { + "epoch": 2.87, + "learning_rate": 1.4851967582109013e-05, + "loss": 0.288, + "step": 6851 + }, + { + "epoch": 2.87, + "learning_rate": 1.48504850383804e-05, + "loss": 0.2654, + "step": 6852 + }, + { + "epoch": 2.87, + "learning_rate": 1.4849002355228901e-05, + "loss": 0.2839, + "step": 6853 + }, + { + "epoch": 2.87, + "learning_rate": 1.484751953269713e-05, + "loss": 0.2706, + "step": 6854 + }, + { + "epoch": 2.87, + "learning_rate": 1.4846036570827714e-05, + "loss": 0.2648, + "step": 6855 + }, + { + "epoch": 2.87, + "learning_rate": 1.4844553469663274e-05, + "loss": 0.2387, + "step": 6856 + }, + { + "epoch": 2.87, + "learning_rate": 1.4843070229246448e-05, + "loss": 0.2518, + "step": 6857 + }, + { + "epoch": 2.87, + "learning_rate": 1.4841586849619862e-05, + "loss": 0.2505, + "step": 6858 + }, + { + "epoch": 2.87, + "learning_rate": 1.4840103330826158e-05, + "loss": 0.2374, + "step": 6859 + }, + { + "epoch": 2.87, + "learning_rate": 1.4838619672907984e-05, + "loss": 0.2631, + "step": 6860 + }, + { + "epoch": 2.87, + "learning_rate": 1.4837135875907978e-05, + "loss": 0.2686, + "step": 6861 + }, + { + "epoch": 2.87, + "learning_rate": 1.4835651939868792e-05, + "loss": 0.2546, + "step": 6862 + }, + { + "epoch": 2.87, + "learning_rate": 1.4834167864833085e-05, + "loss": 0.2985, + "step": 6863 + }, + { + "epoch": 2.87, + "learning_rate": 1.483268365084351e-05, + "loss": 0.2512, + "step": 6864 + }, + { + "epoch": 2.87, + "learning_rate": 1.4831199297942731e-05, + "loss": 0.2642, + "step": 6865 + }, + { + "epoch": 2.87, + "learning_rate": 1.482971480617342e-05, + "loss": 0.2639, + "step": 6866 + }, + { + "epoch": 2.88, + "learning_rate": 1.482823017557824e-05, + "loss": 0.2675, + "step": 6867 + }, + { + "epoch": 2.88, + "learning_rate": 1.4826745406199865e-05, + "loss": 0.2623, + "step": 6868 + }, + { + "epoch": 2.88, + "learning_rate": 1.482526049808098e-05, + "loss": 0.2545, + "step": 6869 + }, + { + "epoch": 2.88, + "learning_rate": 1.4823775451264262e-05, + "loss": 0.2215, + "step": 6870 + }, + { + "epoch": 2.88, + "learning_rate": 1.4822290265792403e-05, + "loss": 0.229, + "step": 6871 + }, + { + "epoch": 2.88, + "learning_rate": 1.4820804941708086e-05, + "loss": 0.2672, + "step": 6872 + }, + { + "epoch": 2.88, + "learning_rate": 1.4819319479054009e-05, + "loss": 0.2455, + "step": 6873 + }, + { + "epoch": 2.88, + "learning_rate": 1.481783387787287e-05, + "loss": 0.2786, + "step": 6874 + }, + { + "epoch": 2.88, + "learning_rate": 1.4816348138207373e-05, + "loss": 0.2453, + "step": 6875 + }, + { + "epoch": 2.88, + "learning_rate": 1.4814862260100222e-05, + "loss": 0.2477, + "step": 6876 + }, + { + "epoch": 2.88, + "learning_rate": 1.4813376243594128e-05, + "loss": 0.275, + "step": 6877 + }, + { + "epoch": 2.88, + "learning_rate": 1.4811890088731808e-05, + "loss": 0.2415, + "step": 6878 + }, + { + "epoch": 2.88, + "learning_rate": 1.4810403795555975e-05, + "loss": 0.2582, + "step": 6879 + }, + { + "epoch": 2.88, + "learning_rate": 1.4808917364109352e-05, + "loss": 0.2525, + "step": 6880 + }, + { + "epoch": 2.88, + "learning_rate": 1.480743079443467e-05, + "loss": 0.2591, + "step": 6881 + }, + { + "epoch": 2.88, + "learning_rate": 1.4805944086574653e-05, + "loss": 0.2261, + "step": 6882 + }, + { + "epoch": 2.88, + "learning_rate": 1.4804457240572037e-05, + "loss": 0.2787, + "step": 6883 + }, + { + "epoch": 2.88, + "learning_rate": 1.4802970256469564e-05, + "loss": 0.2833, + "step": 6884 + }, + { + "epoch": 2.88, + "learning_rate": 1.4801483134309971e-05, + "loss": 0.2346, + "step": 6885 + }, + { + "epoch": 2.88, + "learning_rate": 1.4799995874136006e-05, + "loss": 0.2583, + "step": 6886 + }, + { + "epoch": 2.88, + "learning_rate": 1.4798508475990421e-05, + "loss": 0.2704, + "step": 6887 + }, + { + "epoch": 2.88, + "learning_rate": 1.4797020939915967e-05, + "loss": 0.2709, + "step": 6888 + }, + { + "epoch": 2.88, + "learning_rate": 1.4795533265955404e-05, + "loss": 0.267, + "step": 6889 + }, + { + "epoch": 2.88, + "learning_rate": 1.4794045454151491e-05, + "loss": 0.2466, + "step": 6890 + }, + { + "epoch": 2.89, + "learning_rate": 1.4792557504546997e-05, + "loss": 0.2482, + "step": 6891 + }, + { + "epoch": 2.89, + "learning_rate": 1.4791069417184689e-05, + "loss": 0.2537, + "step": 6892 + }, + { + "epoch": 2.89, + "learning_rate": 1.4789581192107343e-05, + "loss": 0.2675, + "step": 6893 + }, + { + "epoch": 2.89, + "learning_rate": 1.4788092829357736e-05, + "loss": 0.2571, + "step": 6894 + }, + { + "epoch": 2.89, + "learning_rate": 1.4786604328978648e-05, + "loss": 0.267, + "step": 6895 + }, + { + "epoch": 2.89, + "learning_rate": 1.4785115691012866e-05, + "loss": 0.2748, + "step": 6896 + }, + { + "epoch": 2.89, + "learning_rate": 1.4783626915503182e-05, + "loss": 0.2911, + "step": 6897 + }, + { + "epoch": 2.89, + "learning_rate": 1.4782138002492385e-05, + "loss": 0.2753, + "step": 6898 + }, + { + "epoch": 2.89, + "learning_rate": 1.4780648952023276e-05, + "loss": 0.245, + "step": 6899 + }, + { + "epoch": 2.89, + "learning_rate": 1.4779159764138656e-05, + "loss": 0.2457, + "step": 6900 + }, + { + "epoch": 2.89, + "learning_rate": 1.4777670438881328e-05, + "loss": 0.2733, + "step": 6901 + }, + { + "epoch": 2.89, + "learning_rate": 1.4776180976294105e-05, + "loss": 0.2523, + "step": 6902 + }, + { + "epoch": 2.89, + "learning_rate": 1.4774691376419795e-05, + "loss": 0.2498, + "step": 6903 + }, + { + "epoch": 2.89, + "learning_rate": 1.4773201639301223e-05, + "loss": 0.2705, + "step": 6904 + }, + { + "epoch": 2.89, + "learning_rate": 1.47717117649812e-05, + "loss": 0.2358, + "step": 6905 + }, + { + "epoch": 2.89, + "learning_rate": 1.4770221753502558e-05, + "loss": 0.2867, + "step": 6906 + }, + { + "epoch": 2.89, + "learning_rate": 1.4768731604908125e-05, + "loss": 0.2619, + "step": 6907 + }, + { + "epoch": 2.89, + "learning_rate": 1.4767241319240735e-05, + "loss": 0.2755, + "step": 6908 + }, + { + "epoch": 2.89, + "learning_rate": 1.4765750896543222e-05, + "loss": 0.2694, + "step": 6909 + }, + { + "epoch": 2.89, + "learning_rate": 1.476426033685843e-05, + "loss": 0.294, + "step": 6910 + }, + { + "epoch": 2.89, + "learning_rate": 1.4762769640229202e-05, + "loss": 0.235, + "step": 6911 + }, + { + "epoch": 2.89, + "learning_rate": 1.4761278806698389e-05, + "loss": 0.2529, + "step": 6912 + }, + { + "epoch": 2.89, + "learning_rate": 1.475978783630884e-05, + "loss": 0.2626, + "step": 6913 + }, + { + "epoch": 2.89, + "learning_rate": 1.4758296729103413e-05, + "loss": 0.2572, + "step": 6914 + }, + { + "epoch": 2.9, + "learning_rate": 1.475680548512497e-05, + "loss": 0.2899, + "step": 6915 + }, + { + "epoch": 2.9, + "learning_rate": 1.4755314104416371e-05, + "loss": 0.2459, + "step": 6916 + }, + { + "epoch": 2.9, + "learning_rate": 1.4753822587020491e-05, + "loss": 0.255, + "step": 6917 + }, + { + "epoch": 2.9, + "learning_rate": 1.47523309329802e-05, + "loss": 0.2422, + "step": 6918 + }, + { + "epoch": 2.9, + "learning_rate": 1.4750839142338372e-05, + "loss": 0.2726, + "step": 6919 + }, + { + "epoch": 2.9, + "learning_rate": 1.4749347215137888e-05, + "loss": 0.2683, + "step": 6920 + }, + { + "epoch": 2.9, + "learning_rate": 1.4747855151421636e-05, + "loss": 0.2625, + "step": 6921 + }, + { + "epoch": 2.9, + "learning_rate": 1.4746362951232499e-05, + "loss": 0.253, + "step": 6922 + }, + { + "epoch": 2.9, + "learning_rate": 1.4744870614613369e-05, + "loss": 0.2732, + "step": 6923 + }, + { + "epoch": 2.9, + "learning_rate": 1.4743378141607146e-05, + "loss": 0.2718, + "step": 6924 + }, + { + "epoch": 2.9, + "learning_rate": 1.4741885532256726e-05, + "loss": 0.2532, + "step": 6925 + }, + { + "epoch": 2.9, + "learning_rate": 1.4740392786605014e-05, + "loss": 0.2521, + "step": 6926 + }, + { + "epoch": 2.9, + "learning_rate": 1.473889990469492e-05, + "loss": 0.2389, + "step": 6927 + }, + { + "epoch": 2.9, + "learning_rate": 1.4737406886569352e-05, + "loss": 0.2739, + "step": 6928 + }, + { + "epoch": 2.9, + "learning_rate": 1.4735913732271226e-05, + "loss": 0.2551, + "step": 6929 + }, + { + "epoch": 2.9, + "learning_rate": 1.4734420441843462e-05, + "loss": 0.2653, + "step": 6930 + }, + { + "epoch": 2.9, + "learning_rate": 1.4732927015328989e-05, + "loss": 0.2595, + "step": 6931 + }, + { + "epoch": 2.9, + "learning_rate": 1.4731433452770723e-05, + "loss": 0.2651, + "step": 6932 + }, + { + "epoch": 2.9, + "learning_rate": 1.4729939754211602e-05, + "loss": 0.2988, + "step": 6933 + }, + { + "epoch": 2.9, + "learning_rate": 1.4728445919694563e-05, + "loss": 0.2706, + "step": 6934 + }, + { + "epoch": 2.9, + "learning_rate": 1.472695194926254e-05, + "loss": 0.2431, + "step": 6935 + }, + { + "epoch": 2.9, + "learning_rate": 1.4725457842958477e-05, + "loss": 0.2552, + "step": 6936 + }, + { + "epoch": 2.9, + "learning_rate": 1.4723963600825322e-05, + "loss": 0.2553, + "step": 6937 + }, + { + "epoch": 2.9, + "learning_rate": 1.472246922290603e-05, + "loss": 0.2611, + "step": 6938 + }, + { + "epoch": 2.91, + "learning_rate": 1.4720974709243542e-05, + "loss": 0.2267, + "step": 6939 + }, + { + "epoch": 2.91, + "learning_rate": 1.471948005988083e-05, + "loss": 0.2719, + "step": 6940 + }, + { + "epoch": 2.91, + "learning_rate": 1.471798527486085e-05, + "loss": 0.2622, + "step": 6941 + }, + { + "epoch": 2.91, + "learning_rate": 1.4716490354226573e-05, + "loss": 0.2586, + "step": 6942 + }, + { + "epoch": 2.91, + "learning_rate": 1.4714995298020963e-05, + "loss": 0.2677, + "step": 6943 + }, + { + "epoch": 2.91, + "learning_rate": 1.4713500106287e-05, + "loss": 0.2544, + "step": 6944 + }, + { + "epoch": 2.91, + "learning_rate": 1.4712004779067659e-05, + "loss": 0.2792, + "step": 6945 + }, + { + "epoch": 2.91, + "learning_rate": 1.471050931640592e-05, + "loss": 0.243, + "step": 6946 + }, + { + "epoch": 2.91, + "learning_rate": 1.4709013718344773e-05, + "loss": 0.2542, + "step": 6947 + }, + { + "epoch": 2.91, + "learning_rate": 1.4707517984927201e-05, + "loss": 0.2478, + "step": 6948 + }, + { + "epoch": 2.91, + "learning_rate": 1.4706022116196208e-05, + "loss": 0.2705, + "step": 6949 + }, + { + "epoch": 2.91, + "learning_rate": 1.4704526112194781e-05, + "loss": 0.2619, + "step": 6950 + }, + { + "epoch": 2.91, + "learning_rate": 1.4703029972965927e-05, + "loss": 0.2677, + "step": 6951 + }, + { + "epoch": 2.91, + "learning_rate": 1.470153369855265e-05, + "loss": 0.2332, + "step": 6952 + }, + { + "epoch": 2.91, + "learning_rate": 1.470003728899796e-05, + "loss": 0.2818, + "step": 6953 + }, + { + "epoch": 2.91, + "learning_rate": 1.4698540744344864e-05, + "loss": 0.2309, + "step": 6954 + }, + { + "epoch": 2.91, + "learning_rate": 1.469704406463639e-05, + "loss": 0.2381, + "step": 6955 + }, + { + "epoch": 2.91, + "learning_rate": 1.469554724991555e-05, + "loss": 0.2561, + "step": 6956 + }, + { + "epoch": 2.91, + "learning_rate": 1.469405030022537e-05, + "loss": 0.2776, + "step": 6957 + }, + { + "epoch": 2.91, + "learning_rate": 1.4692553215608878e-05, + "loss": 0.2784, + "step": 6958 + }, + { + "epoch": 2.91, + "learning_rate": 1.4691055996109112e-05, + "loss": 0.265, + "step": 6959 + }, + { + "epoch": 2.91, + "learning_rate": 1.4689558641769101e-05, + "loss": 0.2791, + "step": 6960 + }, + { + "epoch": 2.91, + "learning_rate": 1.4688061152631889e-05, + "loss": 0.2317, + "step": 6961 + }, + { + "epoch": 2.91, + "learning_rate": 1.468656352874052e-05, + "loss": 0.2672, + "step": 6962 + }, + { + "epoch": 2.92, + "learning_rate": 1.4685065770138043e-05, + "loss": 0.2632, + "step": 6963 + }, + { + "epoch": 2.92, + "learning_rate": 1.4683567876867503e-05, + "loss": 0.272, + "step": 6964 + }, + { + "epoch": 2.92, + "learning_rate": 1.4682069848971966e-05, + "loss": 0.2292, + "step": 6965 + }, + { + "epoch": 2.92, + "learning_rate": 1.4680571686494483e-05, + "loss": 0.2913, + "step": 6966 + }, + { + "epoch": 2.92, + "learning_rate": 1.467907338947812e-05, + "loss": 0.2553, + "step": 6967 + }, + { + "epoch": 2.92, + "learning_rate": 1.4677574957965946e-05, + "loss": 0.2323, + "step": 6968 + }, + { + "epoch": 2.92, + "learning_rate": 1.4676076392001032e-05, + "loss": 0.2707, + "step": 6969 + }, + { + "epoch": 2.92, + "learning_rate": 1.4674577691626448e-05, + "loss": 0.2586, + "step": 6970 + }, + { + "epoch": 2.92, + "learning_rate": 1.4673078856885277e-05, + "loss": 0.2988, + "step": 6971 + }, + { + "epoch": 2.92, + "learning_rate": 1.4671579887820602e-05, + "loss": 0.2444, + "step": 6972 + }, + { + "epoch": 2.92, + "learning_rate": 1.467008078447551e-05, + "loss": 0.2504, + "step": 6973 + }, + { + "epoch": 2.92, + "learning_rate": 1.4668581546893086e-05, + "loss": 0.2456, + "step": 6974 + }, + { + "epoch": 2.92, + "learning_rate": 1.4667082175116431e-05, + "loss": 0.2501, + "step": 6975 + }, + { + "epoch": 2.92, + "learning_rate": 1.466558266918864e-05, + "loss": 0.2867, + "step": 6976 + }, + { + "epoch": 2.92, + "learning_rate": 1.4664083029152815e-05, + "loss": 0.2356, + "step": 6977 + }, + { + "epoch": 2.92, + "learning_rate": 1.4662583255052064e-05, + "loss": 0.2852, + "step": 6978 + }, + { + "epoch": 2.92, + "learning_rate": 1.4661083346929492e-05, + "loss": 0.2831, + "step": 6979 + }, + { + "epoch": 2.92, + "learning_rate": 1.465958330482822e-05, + "loss": 0.2642, + "step": 6980 + }, + { + "epoch": 2.92, + "learning_rate": 1.4658083128791356e-05, + "loss": 0.251, + "step": 6981 + }, + { + "epoch": 2.92, + "learning_rate": 1.4656582818862028e-05, + "loss": 0.3004, + "step": 6982 + }, + { + "epoch": 2.92, + "learning_rate": 1.465508237508336e-05, + "loss": 0.2536, + "step": 6983 + }, + { + "epoch": 2.92, + "learning_rate": 1.4653581797498477e-05, + "loss": 0.2684, + "step": 6984 + }, + { + "epoch": 2.92, + "learning_rate": 1.4652081086150516e-05, + "loss": 0.2621, + "step": 6985 + }, + { + "epoch": 2.92, + "learning_rate": 1.4650580241082613e-05, + "loss": 0.2872, + "step": 6986 + }, + { + "epoch": 2.93, + "learning_rate": 1.4649079262337909e-05, + "loss": 0.2488, + "step": 6987 + }, + { + "epoch": 2.93, + "learning_rate": 1.4647578149959545e-05, + "loss": 0.2557, + "step": 6988 + }, + { + "epoch": 2.93, + "learning_rate": 1.4646076903990671e-05, + "loss": 0.2869, + "step": 6989 + }, + { + "epoch": 2.93, + "learning_rate": 1.4644575524474441e-05, + "loss": 0.2733, + "step": 6990 + }, + { + "epoch": 2.93, + "learning_rate": 1.4643074011454007e-05, + "loss": 0.2676, + "step": 6991 + }, + { + "epoch": 2.93, + "learning_rate": 1.4641572364972532e-05, + "loss": 0.2571, + "step": 6992 + }, + { + "epoch": 2.93, + "learning_rate": 1.464007058507318e-05, + "loss": 0.2744, + "step": 6993 + }, + { + "epoch": 2.93, + "learning_rate": 1.4638568671799113e-05, + "loss": 0.2485, + "step": 6994 + }, + { + "epoch": 2.93, + "learning_rate": 1.4637066625193508e-05, + "loss": 0.2437, + "step": 6995 + }, + { + "epoch": 2.93, + "learning_rate": 1.463556444529954e-05, + "loss": 0.2428, + "step": 6996 + }, + { + "epoch": 2.93, + "learning_rate": 1.4634062132160381e-05, + "loss": 0.2526, + "step": 6997 + }, + { + "epoch": 2.93, + "learning_rate": 1.4632559685819219e-05, + "loss": 0.2696, + "step": 6998 + }, + { + "epoch": 2.93, + "learning_rate": 1.4631057106319242e-05, + "loss": 0.2625, + "step": 6999 + }, + { + "epoch": 2.93, + "learning_rate": 1.4629554393703635e-05, + "loss": 0.259, + "step": 7000 + }, + { + "epoch": 2.93, + "learning_rate": 1.4628051548015596e-05, + "loss": 0.2856, + "step": 7001 + }, + { + "epoch": 2.93, + "learning_rate": 1.4626548569298324e-05, + "loss": 0.2357, + "step": 7002 + }, + { + "epoch": 2.93, + "learning_rate": 1.462504545759502e-05, + "loss": 0.2855, + "step": 7003 + }, + { + "epoch": 2.93, + "learning_rate": 1.4623542212948887e-05, + "loss": 0.2517, + "step": 7004 + }, + { + "epoch": 2.93, + "learning_rate": 1.4622038835403135e-05, + "loss": 0.2457, + "step": 7005 + }, + { + "epoch": 2.93, + "learning_rate": 1.4620535325000978e-05, + "loss": 0.2696, + "step": 7006 + }, + { + "epoch": 2.93, + "learning_rate": 1.4619031681785633e-05, + "loss": 0.2551, + "step": 7007 + }, + { + "epoch": 2.93, + "learning_rate": 1.4617527905800321e-05, + "loss": 0.2464, + "step": 7008 + }, + { + "epoch": 2.93, + "learning_rate": 1.461602399708827e-05, + "loss": 0.2891, + "step": 7009 + }, + { + "epoch": 2.93, + "learning_rate": 1.4614519955692701e-05, + "loss": 0.2391, + "step": 7010 + }, + { + "epoch": 2.94, + "learning_rate": 1.4613015781656854e-05, + "loss": 0.2525, + "step": 7011 + }, + { + "epoch": 2.94, + "learning_rate": 1.4611511475023958e-05, + "loss": 0.2371, + "step": 7012 + }, + { + "epoch": 2.94, + "learning_rate": 1.461000703583726e-05, + "loss": 0.2491, + "step": 7013 + }, + { + "epoch": 2.94, + "learning_rate": 1.4608502464140002e-05, + "loss": 0.2961, + "step": 7014 + }, + { + "epoch": 2.94, + "learning_rate": 1.4606997759975427e-05, + "loss": 0.2514, + "step": 7015 + }, + { + "epoch": 2.94, + "learning_rate": 1.4605492923386795e-05, + "loss": 0.2443, + "step": 7016 + }, + { + "epoch": 2.94, + "learning_rate": 1.4603987954417349e-05, + "loss": 0.2637, + "step": 7017 + }, + { + "epoch": 2.94, + "learning_rate": 1.4602482853110358e-05, + "loss": 0.2675, + "step": 7018 + }, + { + "epoch": 2.94, + "learning_rate": 1.460097761950908e-05, + "loss": 0.2693, + "step": 7019 + }, + { + "epoch": 2.94, + "learning_rate": 1.4599472253656786e-05, + "loss": 0.2458, + "step": 7020 + }, + { + "epoch": 2.94, + "learning_rate": 1.459796675559674e-05, + "loss": 0.2845, + "step": 7021 + }, + { + "epoch": 2.94, + "learning_rate": 1.4596461125372221e-05, + "loss": 0.2674, + "step": 7022 + }, + { + "epoch": 2.94, + "learning_rate": 1.4594955363026509e-05, + "loss": 0.2461, + "step": 7023 + }, + { + "epoch": 2.94, + "learning_rate": 1.4593449468602878e-05, + "loss": 0.2608, + "step": 7024 + }, + { + "epoch": 2.94, + "learning_rate": 1.459194344214462e-05, + "loss": 0.25, + "step": 7025 + }, + { + "epoch": 2.94, + "learning_rate": 1.4590437283695022e-05, + "loss": 0.2608, + "step": 7026 + }, + { + "epoch": 2.94, + "learning_rate": 1.4588930993297381e-05, + "loss": 0.2597, + "step": 7027 + }, + { + "epoch": 2.94, + "learning_rate": 1.4587424570994989e-05, + "loss": 0.2485, + "step": 7028 + }, + { + "epoch": 2.94, + "learning_rate": 1.4585918016831144e-05, + "loss": 0.3006, + "step": 7029 + }, + { + "epoch": 2.94, + "learning_rate": 1.4584411330849162e-05, + "loss": 0.2837, + "step": 7030 + }, + { + "epoch": 2.94, + "learning_rate": 1.458290451309234e-05, + "loss": 0.243, + "step": 7031 + }, + { + "epoch": 2.94, + "learning_rate": 1.4581397563603994e-05, + "loss": 0.2504, + "step": 7032 + }, + { + "epoch": 2.94, + "learning_rate": 1.4579890482427446e-05, + "loss": 0.27, + "step": 7033 + }, + { + "epoch": 2.94, + "learning_rate": 1.4578383269606004e-05, + "loss": 0.228, + "step": 7034 + }, + { + "epoch": 2.95, + "learning_rate": 1.4576875925182998e-05, + "loss": 0.2386, + "step": 7035 + }, + { + "epoch": 2.95, + "learning_rate": 1.4575368449201759e-05, + "loss": 0.2873, + "step": 7036 + }, + { + "epoch": 2.95, + "learning_rate": 1.4573860841705613e-05, + "loss": 0.2577, + "step": 7037 + }, + { + "epoch": 2.95, + "learning_rate": 1.4572353102737894e-05, + "loss": 0.2532, + "step": 7038 + }, + { + "epoch": 2.95, + "learning_rate": 1.4570845232341943e-05, + "loss": 0.249, + "step": 7039 + }, + { + "epoch": 2.95, + "learning_rate": 1.4569337230561105e-05, + "loss": 0.2475, + "step": 7040 + }, + { + "epoch": 2.95, + "learning_rate": 1.4567829097438719e-05, + "loss": 0.2634, + "step": 7041 + }, + { + "epoch": 2.95, + "learning_rate": 1.456632083301814e-05, + "loss": 0.2866, + "step": 7042 + }, + { + "epoch": 2.95, + "learning_rate": 1.4564812437342724e-05, + "loss": 0.2312, + "step": 7043 + }, + { + "epoch": 2.95, + "learning_rate": 1.4563303910455821e-05, + "loss": 0.2826, + "step": 7044 + }, + { + "epoch": 2.95, + "learning_rate": 1.4561795252400797e-05, + "loss": 0.2687, + "step": 7045 + }, + { + "epoch": 2.95, + "learning_rate": 1.4560286463221018e-05, + "loss": 0.2409, + "step": 7046 + }, + { + "epoch": 2.95, + "learning_rate": 1.455877754295985e-05, + "loss": 0.2738, + "step": 7047 + }, + { + "epoch": 2.95, + "learning_rate": 1.455726849166067e-05, + "loss": 0.2755, + "step": 7048 + }, + { + "epoch": 2.95, + "learning_rate": 1.4555759309366848e-05, + "loss": 0.2729, + "step": 7049 + }, + { + "epoch": 2.95, + "learning_rate": 1.455424999612177e-05, + "loss": 0.2863, + "step": 7050 + }, + { + "epoch": 2.95, + "learning_rate": 1.4552740551968817e-05, + "loss": 0.2536, + "step": 7051 + }, + { + "epoch": 2.95, + "learning_rate": 1.4551230976951373e-05, + "loss": 0.2727, + "step": 7052 + }, + { + "epoch": 2.95, + "learning_rate": 1.4549721271112835e-05, + "loss": 0.2709, + "step": 7053 + }, + { + "epoch": 2.95, + "learning_rate": 1.4548211434496598e-05, + "loss": 0.2484, + "step": 7054 + }, + { + "epoch": 2.95, + "learning_rate": 1.4546701467146058e-05, + "loss": 0.2855, + "step": 7055 + }, + { + "epoch": 2.95, + "learning_rate": 1.4545191369104618e-05, + "loss": 0.265, + "step": 7056 + }, + { + "epoch": 2.95, + "learning_rate": 1.4543681140415688e-05, + "loss": 0.2499, + "step": 7057 + }, + { + "epoch": 2.95, + "learning_rate": 1.4542170781122673e-05, + "loss": 0.26, + "step": 7058 + }, + { + "epoch": 2.96, + "learning_rate": 1.4540660291268993e-05, + "loss": 0.2607, + "step": 7059 + }, + { + "epoch": 2.96, + "learning_rate": 1.453914967089806e-05, + "loss": 0.2486, + "step": 7060 + }, + { + "epoch": 2.96, + "learning_rate": 1.4537638920053298e-05, + "loss": 0.3044, + "step": 7061 + }, + { + "epoch": 2.96, + "learning_rate": 1.4536128038778132e-05, + "loss": 0.2689, + "step": 7062 + }, + { + "epoch": 2.96, + "learning_rate": 1.453461702711599e-05, + "loss": 0.2797, + "step": 7063 + }, + { + "epoch": 2.96, + "learning_rate": 1.4533105885110307e-05, + "loss": 0.2443, + "step": 7064 + }, + { + "epoch": 2.96, + "learning_rate": 1.4531594612804516e-05, + "loss": 0.2563, + "step": 7065 + }, + { + "epoch": 2.96, + "learning_rate": 1.4530083210242062e-05, + "loss": 0.2787, + "step": 7066 + }, + { + "epoch": 2.96, + "learning_rate": 1.4528571677466383e-05, + "loss": 0.2638, + "step": 7067 + }, + { + "epoch": 2.96, + "learning_rate": 1.4527060014520932e-05, + "loss": 0.2708, + "step": 7068 + }, + { + "epoch": 2.96, + "learning_rate": 1.4525548221449156e-05, + "loss": 0.235, + "step": 7069 + }, + { + "epoch": 2.96, + "learning_rate": 1.4524036298294515e-05, + "loss": 0.2447, + "step": 7070 + }, + { + "epoch": 2.96, + "learning_rate": 1.4522524245100465e-05, + "loss": 0.2411, + "step": 7071 + }, + { + "epoch": 2.96, + "learning_rate": 1.4521012061910466e-05, + "loss": 0.26, + "step": 7072 + }, + { + "epoch": 2.96, + "learning_rate": 1.4519499748767989e-05, + "loss": 0.2866, + "step": 7073 + }, + { + "epoch": 2.96, + "learning_rate": 1.4517987305716502e-05, + "loss": 0.2783, + "step": 7074 + }, + { + "epoch": 2.96, + "learning_rate": 1.4516474732799476e-05, + "loss": 0.2697, + "step": 7075 + }, + { + "epoch": 2.96, + "learning_rate": 1.4514962030060395e-05, + "loss": 0.2484, + "step": 7076 + }, + { + "epoch": 2.96, + "learning_rate": 1.451344919754274e-05, + "loss": 0.2751, + "step": 7077 + }, + { + "epoch": 2.96, + "learning_rate": 1.4511936235289987e-05, + "loss": 0.2818, + "step": 7078 + }, + { + "epoch": 2.96, + "learning_rate": 1.451042314334563e-05, + "loss": 0.2503, + "step": 7079 + }, + { + "epoch": 2.96, + "learning_rate": 1.450890992175317e-05, + "loss": 0.2401, + "step": 7080 + }, + { + "epoch": 2.96, + "learning_rate": 1.450739657055609e-05, + "loss": 0.2721, + "step": 7081 + }, + { + "epoch": 2.97, + "learning_rate": 1.4505883089797896e-05, + "loss": 0.2605, + "step": 7082 + }, + { + "epoch": 2.97, + "learning_rate": 1.4504369479522092e-05, + "loss": 0.2714, + "step": 7083 + }, + { + "epoch": 2.97, + "learning_rate": 1.4502855739772184e-05, + "loss": 0.2754, + "step": 7084 + }, + { + "epoch": 2.97, + "learning_rate": 1.4501341870591683e-05, + "loss": 0.2411, + "step": 7085 + }, + { + "epoch": 2.97, + "learning_rate": 1.4499827872024104e-05, + "loss": 0.2623, + "step": 7086 + }, + { + "epoch": 2.97, + "learning_rate": 1.4498313744112966e-05, + "loss": 0.2632, + "step": 7087 + }, + { + "epoch": 2.97, + "learning_rate": 1.4496799486901791e-05, + "loss": 0.2663, + "step": 7088 + }, + { + "epoch": 2.97, + "learning_rate": 1.4495285100434107e-05, + "loss": 0.2793, + "step": 7089 + }, + { + "epoch": 2.97, + "learning_rate": 1.449377058475344e-05, + "loss": 0.2695, + "step": 7090 + }, + { + "epoch": 2.97, + "learning_rate": 1.4492255939903326e-05, + "loss": 0.2576, + "step": 7091 + }, + { + "epoch": 2.97, + "learning_rate": 1.44907411659273e-05, + "loss": 0.2428, + "step": 7092 + }, + { + "epoch": 2.97, + "learning_rate": 1.4489226262868906e-05, + "loss": 0.2305, + "step": 7093 + }, + { + "epoch": 2.97, + "learning_rate": 1.4487711230771686e-05, + "loss": 0.2953, + "step": 7094 + }, + { + "epoch": 2.97, + "learning_rate": 1.4486196069679191e-05, + "loss": 0.2662, + "step": 7095 + }, + { + "epoch": 2.97, + "learning_rate": 1.4484680779634968e-05, + "loss": 0.2287, + "step": 7096 + }, + { + "epoch": 2.97, + "learning_rate": 1.4483165360682577e-05, + "loss": 0.2736, + "step": 7097 + }, + { + "epoch": 2.97, + "learning_rate": 1.4481649812865574e-05, + "loss": 0.2746, + "step": 7098 + }, + { + "epoch": 2.97, + "learning_rate": 1.4480134136227526e-05, + "loss": 0.2411, + "step": 7099 + }, + { + "epoch": 2.97, + "learning_rate": 1.4478618330811997e-05, + "loss": 0.2587, + "step": 7100 + }, + { + "epoch": 2.97, + "learning_rate": 1.447710239666256e-05, + "loss": 0.2382, + "step": 7101 + }, + { + "epoch": 2.97, + "learning_rate": 1.4475586333822787e-05, + "loss": 0.2577, + "step": 7102 + }, + { + "epoch": 2.97, + "learning_rate": 1.4474070142336254e-05, + "loss": 0.2714, + "step": 7103 + }, + { + "epoch": 2.97, + "learning_rate": 1.4472553822246548e-05, + "loss": 0.2527, + "step": 7104 + }, + { + "epoch": 2.97, + "learning_rate": 1.4471037373597251e-05, + "loss": 0.292, + "step": 7105 + }, + { + "epoch": 2.98, + "learning_rate": 1.446952079643195e-05, + "loss": 0.2405, + "step": 7106 + }, + { + "epoch": 2.98, + "learning_rate": 1.4468004090794241e-05, + "loss": 0.2724, + "step": 7107 + }, + { + "epoch": 2.98, + "learning_rate": 1.4466487256727721e-05, + "loss": 0.2636, + "step": 7108 + }, + { + "epoch": 2.98, + "learning_rate": 1.4464970294275986e-05, + "loss": 0.2888, + "step": 7109 + }, + { + "epoch": 2.98, + "learning_rate": 1.4463453203482643e-05, + "loss": 0.2672, + "step": 7110 + }, + { + "epoch": 2.98, + "learning_rate": 1.44619359843913e-05, + "loss": 0.2623, + "step": 7111 + }, + { + "epoch": 2.98, + "learning_rate": 1.4460418637045564e-05, + "loss": 0.2394, + "step": 7112 + }, + { + "epoch": 2.98, + "learning_rate": 1.4458901161489051e-05, + "loss": 0.2543, + "step": 7113 + }, + { + "epoch": 2.98, + "learning_rate": 1.4457383557765385e-05, + "loss": 0.2451, + "step": 7114 + }, + { + "epoch": 2.98, + "learning_rate": 1.4455865825918182e-05, + "loss": 0.2542, + "step": 7115 + }, + { + "epoch": 2.98, + "learning_rate": 1.445434796599107e-05, + "loss": 0.2598, + "step": 7116 + }, + { + "epoch": 2.98, + "learning_rate": 1.4452829978027678e-05, + "loss": 0.2718, + "step": 7117 + }, + { + "epoch": 2.98, + "learning_rate": 1.4451311862071641e-05, + "loss": 0.2405, + "step": 7118 + }, + { + "epoch": 2.98, + "learning_rate": 1.4449793618166594e-05, + "loss": 0.2497, + "step": 7119 + }, + { + "epoch": 2.98, + "learning_rate": 1.4448275246356175e-05, + "loss": 0.2675, + "step": 7120 + }, + { + "epoch": 2.98, + "learning_rate": 1.4446756746684036e-05, + "loss": 0.2779, + "step": 7121 + }, + { + "epoch": 2.98, + "learning_rate": 1.4445238119193818e-05, + "loss": 0.2645, + "step": 7122 + }, + { + "epoch": 2.98, + "learning_rate": 1.4443719363929173e-05, + "loss": 0.2663, + "step": 7123 + }, + { + "epoch": 2.98, + "learning_rate": 1.4442200480933761e-05, + "loss": 0.2581, + "step": 7124 + }, + { + "epoch": 2.98, + "learning_rate": 1.4440681470251239e-05, + "loss": 0.2753, + "step": 7125 + }, + { + "epoch": 2.98, + "learning_rate": 1.4439162331925268e-05, + "loss": 0.2406, + "step": 7126 + }, + { + "epoch": 2.98, + "learning_rate": 1.4437643065999516e-05, + "loss": 0.2575, + "step": 7127 + }, + { + "epoch": 2.98, + "learning_rate": 1.443612367251765e-05, + "loss": 0.2494, + "step": 7128 + }, + { + "epoch": 2.98, + "learning_rate": 1.4434604151523345e-05, + "loss": 0.2552, + "step": 7129 + }, + { + "epoch": 2.99, + "learning_rate": 1.443308450306028e-05, + "loss": 0.2535, + "step": 7130 + }, + { + "epoch": 2.99, + "learning_rate": 1.4431564727172135e-05, + "loss": 0.2476, + "step": 7131 + }, + { + "epoch": 2.99, + "learning_rate": 1.4430044823902594e-05, + "loss": 0.2381, + "step": 7132 + }, + { + "epoch": 2.99, + "learning_rate": 1.4428524793295344e-05, + "loss": 0.2544, + "step": 7133 + }, + { + "epoch": 2.99, + "learning_rate": 1.442700463539408e-05, + "loss": 0.269, + "step": 7134 + }, + { + "epoch": 2.99, + "learning_rate": 1.4425484350242498e-05, + "loss": 0.24, + "step": 7135 + }, + { + "epoch": 2.99, + "learning_rate": 1.4423963937884293e-05, + "loss": 0.2364, + "step": 7136 + }, + { + "epoch": 2.99, + "learning_rate": 1.442244339836317e-05, + "loss": 0.2322, + "step": 7137 + }, + { + "epoch": 2.99, + "learning_rate": 1.442092273172284e-05, + "loss": 0.2413, + "step": 7138 + }, + { + "epoch": 2.99, + "learning_rate": 1.4419401938007007e-05, + "loss": 0.2707, + "step": 7139 + }, + { + "epoch": 2.99, + "learning_rate": 1.4417881017259385e-05, + "loss": 0.3087, + "step": 7140 + }, + { + "epoch": 2.99, + "learning_rate": 1.4416359969523694e-05, + "loss": 0.265, + "step": 7141 + }, + { + "epoch": 2.99, + "learning_rate": 1.4414838794843658e-05, + "loss": 0.2298, + "step": 7142 + }, + { + "epoch": 2.99, + "learning_rate": 1.4413317493262996e-05, + "loss": 0.2708, + "step": 7143 + }, + { + "epoch": 2.99, + "learning_rate": 1.4411796064825436e-05, + "loss": 0.2349, + "step": 7144 + }, + { + "epoch": 2.99, + "learning_rate": 1.4410274509574718e-05, + "loss": 0.2678, + "step": 7145 + }, + { + "epoch": 2.99, + "learning_rate": 1.4408752827554571e-05, + "loss": 0.2931, + "step": 7146 + }, + { + "epoch": 2.99, + "learning_rate": 1.4407231018808734e-05, + "loss": 0.2673, + "step": 7147 + }, + { + "epoch": 2.99, + "learning_rate": 1.4405709083380957e-05, + "loss": 0.2571, + "step": 7148 + }, + { + "epoch": 2.99, + "learning_rate": 1.4404187021314982e-05, + "loss": 0.2492, + "step": 7149 + }, + { + "epoch": 2.99, + "learning_rate": 1.4402664832654555e-05, + "loss": 0.2863, + "step": 7150 + }, + { + "epoch": 2.99, + "learning_rate": 1.4401142517443434e-05, + "loss": 0.2501, + "step": 7151 + }, + { + "epoch": 2.99, + "learning_rate": 1.439962007572538e-05, + "loss": 0.2647, + "step": 7152 + }, + { + "epoch": 2.99, + "learning_rate": 1.4398097507544152e-05, + "loss": 0.2374, + "step": 7153 + }, + { + "epoch": 3.0, + "learning_rate": 1.439657481294351e-05, + "loss": 0.2462, + "step": 7154 + }, + { + "epoch": 3.0, + "learning_rate": 1.4395051991967228e-05, + "loss": 0.2512, + "step": 7155 + }, + { + "epoch": 3.0, + "learning_rate": 1.4393529044659077e-05, + "loss": 0.2685, + "step": 7156 + }, + { + "epoch": 3.0, + "learning_rate": 1.439200597106283e-05, + "loss": 0.2732, + "step": 7157 + }, + { + "epoch": 3.0, + "learning_rate": 1.4390482771222273e-05, + "loss": 0.2834, + "step": 7158 + }, + { + "epoch": 3.0, + "learning_rate": 1.4388959445181181e-05, + "loss": 0.2718, + "step": 7159 + }, + { + "epoch": 3.0, + "learning_rate": 1.4387435992983347e-05, + "loss": 0.2553, + "step": 7160 + }, + { + "epoch": 3.0, + "learning_rate": 1.4385912414672559e-05, + "loss": 0.23, + "step": 7161 + }, + { + "epoch": 3.0, + "learning_rate": 1.438438871029261e-05, + "loss": 0.2289, + "step": 7162 + }, + { + "epoch": 3.0, + "learning_rate": 1.4382864879887298e-05, + "loss": 0.2558, + "step": 7163 + }, + { + "epoch": 3.0, + "learning_rate": 1.4381340923500422e-05, + "loss": 0.2601, + "step": 7164 + }, + { + "epoch": 3.0, + "learning_rate": 1.437981684117579e-05, + "loss": 0.2692, + "step": 7165 + }, + { + "epoch": 3.0, + "learning_rate": 1.4378292632957214e-05, + "loss": 0.2586, + "step": 7166 + }, + { + "epoch": 3.0, + "learning_rate": 1.4376768298888498e-05, + "loss": 0.1887, + "step": 7167 + }, + { + "epoch": 3.0, + "learning_rate": 1.4375243839013458e-05, + "loss": 0.2059, + "step": 7168 + }, + { + "epoch": 3.0, + "learning_rate": 1.4373719253375921e-05, + "loss": 0.231, + "step": 7169 + }, + { + "epoch": 3.0, + "learning_rate": 1.4372194542019705e-05, + "loss": 0.1802, + "step": 7170 + }, + { + "epoch": 3.0, + "learning_rate": 1.4370669704988634e-05, + "loss": 0.1511, + "step": 7171 + }, + { + "epoch": 3.0, + "learning_rate": 1.4369144742326544e-05, + "loss": 0.1502, + "step": 7172 + }, + { + "epoch": 3.0, + "learning_rate": 1.4367619654077264e-05, + "loss": 0.1817, + "step": 7173 + }, + { + "epoch": 3.0, + "learning_rate": 1.4366094440284632e-05, + "loss": 0.1603, + "step": 7174 + }, + { + "epoch": 3.0, + "learning_rate": 1.4364569100992489e-05, + "loss": 0.2057, + "step": 7175 + }, + { + "epoch": 3.0, + "learning_rate": 1.4363043636244682e-05, + "loss": 0.1826, + "step": 7176 + }, + { + "epoch": 3.0, + "learning_rate": 1.4361518046085057e-05, + "loss": 0.1632, + "step": 7177 + }, + { + "epoch": 3.01, + "learning_rate": 1.4359992330557467e-05, + "loss": 0.1868, + "step": 7178 + }, + { + "epoch": 3.01, + "learning_rate": 1.4358466489705766e-05, + "loss": 0.1837, + "step": 7179 + }, + { + "epoch": 3.01, + "learning_rate": 1.4356940523573812e-05, + "loss": 0.2101, + "step": 7180 + }, + { + "epoch": 3.01, + "learning_rate": 1.4355414432205468e-05, + "loss": 0.1869, + "step": 7181 + }, + { + "epoch": 3.01, + "learning_rate": 1.4353888215644604e-05, + "loss": 0.2015, + "step": 7182 + }, + { + "epoch": 3.01, + "learning_rate": 1.4352361873935088e-05, + "loss": 0.1817, + "step": 7183 + }, + { + "epoch": 3.01, + "learning_rate": 1.435083540712079e-05, + "loss": 0.2035, + "step": 7184 + }, + { + "epoch": 3.01, + "learning_rate": 1.4349308815245586e-05, + "loss": 0.1981, + "step": 7185 + }, + { + "epoch": 3.01, + "learning_rate": 1.4347782098353364e-05, + "loss": 0.1769, + "step": 7186 + }, + { + "epoch": 3.01, + "learning_rate": 1.4346255256488002e-05, + "loss": 0.1934, + "step": 7187 + }, + { + "epoch": 3.01, + "learning_rate": 1.4344728289693391e-05, + "loss": 0.1702, + "step": 7188 + }, + { + "epoch": 3.01, + "learning_rate": 1.434320119801342e-05, + "loss": 0.177, + "step": 7189 + }, + { + "epoch": 3.01, + "learning_rate": 1.4341673981491982e-05, + "loss": 0.1806, + "step": 7190 + }, + { + "epoch": 3.01, + "learning_rate": 1.4340146640172981e-05, + "loss": 0.2014, + "step": 7191 + }, + { + "epoch": 3.01, + "learning_rate": 1.4338619174100315e-05, + "loss": 0.1654, + "step": 7192 + }, + { + "epoch": 3.01, + "learning_rate": 1.433709158331789e-05, + "loss": 0.1999, + "step": 7193 + }, + { + "epoch": 3.01, + "learning_rate": 1.4335563867869618e-05, + "loss": 0.1922, + "step": 7194 + }, + { + "epoch": 3.01, + "learning_rate": 1.433403602779941e-05, + "loss": 0.1618, + "step": 7195 + }, + { + "epoch": 3.01, + "learning_rate": 1.4332508063151184e-05, + "loss": 0.1569, + "step": 7196 + }, + { + "epoch": 3.01, + "learning_rate": 1.4330979973968856e-05, + "loss": 0.2257, + "step": 7197 + }, + { + "epoch": 3.01, + "learning_rate": 1.4329451760296352e-05, + "loss": 0.1733, + "step": 7198 + }, + { + "epoch": 3.01, + "learning_rate": 1.43279234221776e-05, + "loss": 0.2231, + "step": 7199 + }, + { + "epoch": 3.01, + "learning_rate": 1.4326394959656527e-05, + "loss": 0.1967, + "step": 7200 + }, + { + "epoch": 3.01, + "learning_rate": 1.4324866372777072e-05, + "loss": 0.2143, + "step": 7201 + }, + { + "epoch": 3.02, + "learning_rate": 1.4323337661583168e-05, + "loss": 0.1895, + "step": 7202 + }, + { + "epoch": 3.02, + "learning_rate": 1.4321808826118764e-05, + "loss": 0.2105, + "step": 7203 + }, + { + "epoch": 3.02, + "learning_rate": 1.4320279866427798e-05, + "loss": 0.1458, + "step": 7204 + }, + { + "epoch": 3.02, + "learning_rate": 1.431875078255422e-05, + "loss": 0.1854, + "step": 7205 + }, + { + "epoch": 3.02, + "learning_rate": 1.4317221574541985e-05, + "loss": 0.1838, + "step": 7206 + }, + { + "epoch": 3.02, + "learning_rate": 1.4315692242435048e-05, + "loss": 0.1533, + "step": 7207 + }, + { + "epoch": 3.02, + "learning_rate": 1.4314162786277366e-05, + "loss": 0.1994, + "step": 7208 + }, + { + "epoch": 3.02, + "learning_rate": 1.43126332061129e-05, + "loss": 0.2194, + "step": 7209 + }, + { + "epoch": 3.02, + "learning_rate": 1.4311103501985625e-05, + "loss": 0.169, + "step": 7210 + }, + { + "epoch": 3.02, + "learning_rate": 1.4309573673939503e-05, + "loss": 0.2104, + "step": 7211 + }, + { + "epoch": 3.02, + "learning_rate": 1.4308043722018508e-05, + "loss": 0.2042, + "step": 7212 + }, + { + "epoch": 3.02, + "learning_rate": 1.4306513646266621e-05, + "loss": 0.1725, + "step": 7213 + }, + { + "epoch": 3.02, + "learning_rate": 1.430498344672782e-05, + "loss": 0.1701, + "step": 7214 + }, + { + "epoch": 3.02, + "learning_rate": 1.430345312344609e-05, + "loss": 0.1718, + "step": 7215 + }, + { + "epoch": 3.02, + "learning_rate": 1.430192267646542e-05, + "loss": 0.1763, + "step": 7216 + }, + { + "epoch": 3.02, + "learning_rate": 1.4300392105829802e-05, + "loss": 0.17, + "step": 7217 + }, + { + "epoch": 3.02, + "learning_rate": 1.4298861411583224e-05, + "loss": 0.1941, + "step": 7218 + }, + { + "epoch": 3.02, + "learning_rate": 1.4297330593769691e-05, + "loss": 0.2126, + "step": 7219 + }, + { + "epoch": 3.02, + "learning_rate": 1.4295799652433206e-05, + "loss": 0.1735, + "step": 7220 + }, + { + "epoch": 3.02, + "learning_rate": 1.429426858761777e-05, + "loss": 0.1709, + "step": 7221 + }, + { + "epoch": 3.02, + "learning_rate": 1.4292737399367394e-05, + "loss": 0.1972, + "step": 7222 + }, + { + "epoch": 3.02, + "learning_rate": 1.429120608772609e-05, + "loss": 0.19, + "step": 7223 + }, + { + "epoch": 3.02, + "learning_rate": 1.4289674652737877e-05, + "loss": 0.1853, + "step": 7224 + }, + { + "epoch": 3.02, + "learning_rate": 1.4288143094446771e-05, + "loss": 0.1501, + "step": 7225 + }, + { + "epoch": 3.03, + "learning_rate": 1.42866114128968e-05, + "loss": 0.1467, + "step": 7226 + }, + { + "epoch": 3.03, + "learning_rate": 1.4285079608131984e-05, + "loss": 0.148, + "step": 7227 + }, + { + "epoch": 3.03, + "learning_rate": 1.4283547680196359e-05, + "loss": 0.2309, + "step": 7228 + }, + { + "epoch": 3.03, + "learning_rate": 1.4282015629133958e-05, + "loss": 0.2142, + "step": 7229 + }, + { + "epoch": 3.03, + "learning_rate": 1.4280483454988816e-05, + "loss": 0.2085, + "step": 7230 + }, + { + "epoch": 3.03, + "learning_rate": 1.4278951157804977e-05, + "loss": 0.1577, + "step": 7231 + }, + { + "epoch": 3.03, + "learning_rate": 1.4277418737626483e-05, + "loss": 0.1826, + "step": 7232 + }, + { + "epoch": 3.03, + "learning_rate": 1.4275886194497384e-05, + "loss": 0.2294, + "step": 7233 + }, + { + "epoch": 3.03, + "learning_rate": 1.4274353528461728e-05, + "loss": 0.1659, + "step": 7234 + }, + { + "epoch": 3.03, + "learning_rate": 1.4272820739563575e-05, + "loss": 0.226, + "step": 7235 + }, + { + "epoch": 3.03, + "learning_rate": 1.427128782784698e-05, + "loss": 0.1884, + "step": 7236 + }, + { + "epoch": 3.03, + "learning_rate": 1.4269754793356008e-05, + "loss": 0.1751, + "step": 7237 + }, + { + "epoch": 3.03, + "learning_rate": 1.4268221636134722e-05, + "loss": 0.2024, + "step": 7238 + }, + { + "epoch": 3.03, + "learning_rate": 1.4266688356227194e-05, + "loss": 0.1908, + "step": 7239 + }, + { + "epoch": 3.03, + "learning_rate": 1.4265154953677498e-05, + "loss": 0.1581, + "step": 7240 + }, + { + "epoch": 3.03, + "learning_rate": 1.4263621428529703e-05, + "loss": 0.1458, + "step": 7241 + }, + { + "epoch": 3.03, + "learning_rate": 1.4262087780827896e-05, + "loss": 0.18, + "step": 7242 + }, + { + "epoch": 3.03, + "learning_rate": 1.4260554010616156e-05, + "loss": 0.1969, + "step": 7243 + }, + { + "epoch": 3.03, + "learning_rate": 1.4259020117938574e-05, + "loss": 0.1862, + "step": 7244 + }, + { + "epoch": 3.03, + "learning_rate": 1.4257486102839236e-05, + "loss": 0.2063, + "step": 7245 + }, + { + "epoch": 3.03, + "learning_rate": 1.4255951965362238e-05, + "loss": 0.1843, + "step": 7246 + }, + { + "epoch": 3.03, + "learning_rate": 1.4254417705551679e-05, + "loss": 0.1448, + "step": 7247 + }, + { + "epoch": 3.03, + "learning_rate": 1.4252883323451658e-05, + "loss": 0.2114, + "step": 7248 + }, + { + "epoch": 3.03, + "learning_rate": 1.425134881910628e-05, + "loss": 0.2203, + "step": 7249 + }, + { + "epoch": 3.04, + "learning_rate": 1.4249814192559653e-05, + "loss": 0.1884, + "step": 7250 + }, + { + "epoch": 3.04, + "learning_rate": 1.4248279443855891e-05, + "loss": 0.1889, + "step": 7251 + }, + { + "epoch": 3.04, + "learning_rate": 1.42467445730391e-05, + "loss": 0.1804, + "step": 7252 + }, + { + "epoch": 3.04, + "learning_rate": 1.4245209580153408e-05, + "loss": 0.174, + "step": 7253 + }, + { + "epoch": 3.04, + "learning_rate": 1.4243674465242934e-05, + "loss": 0.1691, + "step": 7254 + }, + { + "epoch": 3.04, + "learning_rate": 1.42421392283518e-05, + "loss": 0.1507, + "step": 7255 + }, + { + "epoch": 3.04, + "learning_rate": 1.424060386952414e-05, + "loss": 0.1765, + "step": 7256 + }, + { + "epoch": 3.04, + "learning_rate": 1.4239068388804089e-05, + "loss": 0.1813, + "step": 7257 + }, + { + "epoch": 3.04, + "learning_rate": 1.4237532786235775e-05, + "loss": 0.1951, + "step": 7258 + }, + { + "epoch": 3.04, + "learning_rate": 1.423599706186334e-05, + "loss": 0.2148, + "step": 7259 + }, + { + "epoch": 3.04, + "learning_rate": 1.4234461215730932e-05, + "loss": 0.1641, + "step": 7260 + }, + { + "epoch": 3.04, + "learning_rate": 1.4232925247882692e-05, + "loss": 0.1737, + "step": 7261 + }, + { + "epoch": 3.04, + "learning_rate": 1.4231389158362771e-05, + "loss": 0.1844, + "step": 7262 + }, + { + "epoch": 3.04, + "learning_rate": 1.4229852947215326e-05, + "loss": 0.179, + "step": 7263 + }, + { + "epoch": 3.04, + "learning_rate": 1.4228316614484511e-05, + "loss": 0.1891, + "step": 7264 + }, + { + "epoch": 3.04, + "learning_rate": 1.4226780160214486e-05, + "loss": 0.166, + "step": 7265 + }, + { + "epoch": 3.04, + "learning_rate": 1.4225243584449415e-05, + "loss": 0.1638, + "step": 7266 + }, + { + "epoch": 3.04, + "learning_rate": 1.422370688723347e-05, + "loss": 0.2241, + "step": 7267 + }, + { + "epoch": 3.04, + "learning_rate": 1.4222170068610816e-05, + "loss": 0.2143, + "step": 7268 + }, + { + "epoch": 3.04, + "learning_rate": 1.4220633128625632e-05, + "loss": 0.2018, + "step": 7269 + }, + { + "epoch": 3.04, + "learning_rate": 1.4219096067322094e-05, + "loss": 0.1457, + "step": 7270 + }, + { + "epoch": 3.04, + "learning_rate": 1.421755888474438e-05, + "loss": 0.1974, + "step": 7271 + }, + { + "epoch": 3.04, + "learning_rate": 1.4216021580936681e-05, + "loss": 0.1919, + "step": 7272 + }, + { + "epoch": 3.05, + "learning_rate": 1.421448415594318e-05, + "loss": 0.2016, + "step": 7273 + }, + { + "epoch": 3.05, + "learning_rate": 1.4212946609808079e-05, + "loss": 0.1676, + "step": 7274 + }, + { + "epoch": 3.05, + "learning_rate": 1.421140894257556e-05, + "loss": 0.2124, + "step": 7275 + }, + { + "epoch": 3.05, + "learning_rate": 1.420987115428983e-05, + "loss": 0.2214, + "step": 7276 + }, + { + "epoch": 3.05, + "learning_rate": 1.4208333244995087e-05, + "loss": 0.1987, + "step": 7277 + }, + { + "epoch": 3.05, + "learning_rate": 1.4206795214735543e-05, + "loss": 0.1791, + "step": 7278 + }, + { + "epoch": 3.05, + "learning_rate": 1.42052570635554e-05, + "loss": 0.2003, + "step": 7279 + }, + { + "epoch": 3.05, + "learning_rate": 1.4203718791498877e-05, + "loss": 0.1513, + "step": 7280 + }, + { + "epoch": 3.05, + "learning_rate": 1.4202180398610188e-05, + "loss": 0.1887, + "step": 7281 + }, + { + "epoch": 3.05, + "learning_rate": 1.420064188493355e-05, + "loss": 0.2105, + "step": 7282 + }, + { + "epoch": 3.05, + "learning_rate": 1.4199103250513192e-05, + "loss": 0.1794, + "step": 7283 + }, + { + "epoch": 3.05, + "learning_rate": 1.4197564495393334e-05, + "loss": 0.1558, + "step": 7284 + }, + { + "epoch": 3.05, + "learning_rate": 1.4196025619618212e-05, + "loss": 0.1765, + "step": 7285 + }, + { + "epoch": 3.05, + "learning_rate": 1.4194486623232056e-05, + "loss": 0.1902, + "step": 7286 + }, + { + "epoch": 3.05, + "learning_rate": 1.4192947506279102e-05, + "loss": 0.1742, + "step": 7287 + }, + { + "epoch": 3.05, + "learning_rate": 1.4191408268803596e-05, + "loss": 0.1754, + "step": 7288 + }, + { + "epoch": 3.05, + "learning_rate": 1.4189868910849779e-05, + "loss": 0.1759, + "step": 7289 + }, + { + "epoch": 3.05, + "learning_rate": 1.4188329432461894e-05, + "loss": 0.1869, + "step": 7290 + }, + { + "epoch": 3.05, + "learning_rate": 1.41867898336842e-05, + "loss": 0.1341, + "step": 7291 + }, + { + "epoch": 3.05, + "learning_rate": 1.4185250114560945e-05, + "loss": 0.1637, + "step": 7292 + }, + { + "epoch": 3.05, + "learning_rate": 1.4183710275136391e-05, + "loss": 0.1879, + "step": 7293 + }, + { + "epoch": 3.05, + "learning_rate": 1.4182170315454798e-05, + "loss": 0.1744, + "step": 7294 + }, + { + "epoch": 3.05, + "learning_rate": 1.4180630235560428e-05, + "loss": 0.1949, + "step": 7295 + }, + { + "epoch": 3.05, + "learning_rate": 1.4179090035497554e-05, + "loss": 0.1792, + "step": 7296 + }, + { + "epoch": 3.06, + "learning_rate": 1.4177549715310445e-05, + "loss": 0.169, + "step": 7297 + }, + { + "epoch": 3.06, + "learning_rate": 1.4176009275043376e-05, + "loss": 0.1775, + "step": 7298 + }, + { + "epoch": 3.06, + "learning_rate": 1.4174468714740628e-05, + "loss": 0.1806, + "step": 7299 + }, + { + "epoch": 3.06, + "learning_rate": 1.4172928034446478e-05, + "loss": 0.217, + "step": 7300 + }, + { + "epoch": 3.06, + "learning_rate": 1.4171387234205219e-05, + "loss": 0.1439, + "step": 7301 + }, + { + "epoch": 3.06, + "learning_rate": 1.4169846314061132e-05, + "loss": 0.1562, + "step": 7302 + }, + { + "epoch": 3.06, + "learning_rate": 1.4168305274058511e-05, + "loss": 0.1626, + "step": 7303 + }, + { + "epoch": 3.06, + "learning_rate": 1.4166764114241657e-05, + "loss": 0.1897, + "step": 7304 + }, + { + "epoch": 3.06, + "learning_rate": 1.4165222834654867e-05, + "loss": 0.2121, + "step": 7305 + }, + { + "epoch": 3.06, + "learning_rate": 1.416368143534244e-05, + "loss": 0.1895, + "step": 7306 + }, + { + "epoch": 3.06, + "learning_rate": 1.4162139916348685e-05, + "loss": 0.2021, + "step": 7307 + }, + { + "epoch": 3.06, + "learning_rate": 1.4160598277717916e-05, + "loss": 0.1783, + "step": 7308 + }, + { + "epoch": 3.06, + "learning_rate": 1.4159056519494439e-05, + "loss": 0.2108, + "step": 7309 + }, + { + "epoch": 3.06, + "learning_rate": 1.4157514641722569e-05, + "loss": 0.2187, + "step": 7310 + }, + { + "epoch": 3.06, + "learning_rate": 1.4155972644446635e-05, + "loss": 0.1467, + "step": 7311 + }, + { + "epoch": 3.06, + "learning_rate": 1.4154430527710954e-05, + "loss": 0.1728, + "step": 7312 + }, + { + "epoch": 3.06, + "learning_rate": 1.4152888291559852e-05, + "loss": 0.1972, + "step": 7313 + }, + { + "epoch": 3.06, + "learning_rate": 1.4151345936037664e-05, + "loss": 0.2267, + "step": 7314 + }, + { + "epoch": 3.06, + "learning_rate": 1.414980346118872e-05, + "loss": 0.1653, + "step": 7315 + }, + { + "epoch": 3.06, + "learning_rate": 1.4148260867057357e-05, + "loss": 0.1404, + "step": 7316 + }, + { + "epoch": 3.06, + "learning_rate": 1.4146718153687917e-05, + "loss": 0.1793, + "step": 7317 + }, + { + "epoch": 3.06, + "learning_rate": 1.4145175321124744e-05, + "loss": 0.1372, + "step": 7318 + }, + { + "epoch": 3.06, + "learning_rate": 1.4143632369412185e-05, + "loss": 0.1705, + "step": 7319 + }, + { + "epoch": 3.06, + "learning_rate": 1.414208929859459e-05, + "loss": 0.2271, + "step": 7320 + }, + { + "epoch": 3.07, + "learning_rate": 1.4140546108716313e-05, + "loss": 0.1695, + "step": 7321 + }, + { + "epoch": 3.07, + "learning_rate": 1.4139002799821714e-05, + "loss": 0.1739, + "step": 7322 + }, + { + "epoch": 3.07, + "learning_rate": 1.4137459371955151e-05, + "loss": 0.2014, + "step": 7323 + }, + { + "epoch": 3.07, + "learning_rate": 1.413591582516099e-05, + "loss": 0.1908, + "step": 7324 + }, + { + "epoch": 3.07, + "learning_rate": 1.41343721594836e-05, + "loss": 0.1706, + "step": 7325 + }, + { + "epoch": 3.07, + "learning_rate": 1.4132828374967348e-05, + "loss": 0.1861, + "step": 7326 + }, + { + "epoch": 3.07, + "learning_rate": 1.4131284471656615e-05, + "loss": 0.1634, + "step": 7327 + }, + { + "epoch": 3.07, + "learning_rate": 1.4129740449595777e-05, + "loss": 0.1741, + "step": 7328 + }, + { + "epoch": 3.07, + "learning_rate": 1.4128196308829209e-05, + "loss": 0.1747, + "step": 7329 + }, + { + "epoch": 3.07, + "learning_rate": 1.4126652049401306e-05, + "loss": 0.1682, + "step": 7330 + }, + { + "epoch": 3.07, + "learning_rate": 1.4125107671356451e-05, + "loss": 0.19, + "step": 7331 + }, + { + "epoch": 3.07, + "learning_rate": 1.4123563174739036e-05, + "loss": 0.1916, + "step": 7332 + }, + { + "epoch": 3.07, + "learning_rate": 1.4122018559593457e-05, + "loss": 0.1645, + "step": 7333 + }, + { + "epoch": 3.07, + "learning_rate": 1.4120473825964112e-05, + "loss": 0.1863, + "step": 7334 + }, + { + "epoch": 3.07, + "learning_rate": 1.4118928973895406e-05, + "loss": 0.1731, + "step": 7335 + }, + { + "epoch": 3.07, + "learning_rate": 1.4117384003431738e-05, + "loss": 0.1718, + "step": 7336 + }, + { + "epoch": 3.07, + "learning_rate": 1.4115838914617518e-05, + "loss": 0.2038, + "step": 7337 + }, + { + "epoch": 3.07, + "learning_rate": 1.4114293707497168e-05, + "loss": 0.1613, + "step": 7338 + }, + { + "epoch": 3.07, + "learning_rate": 1.4112748382115091e-05, + "loss": 0.2105, + "step": 7339 + }, + { + "epoch": 3.07, + "learning_rate": 1.4111202938515711e-05, + "loss": 0.1805, + "step": 7340 + }, + { + "epoch": 3.07, + "learning_rate": 1.4109657376743455e-05, + "loss": 0.157, + "step": 7341 + }, + { + "epoch": 3.07, + "learning_rate": 1.4108111696842743e-05, + "loss": 0.2109, + "step": 7342 + }, + { + "epoch": 3.07, + "learning_rate": 1.4106565898858007e-05, + "loss": 0.1733, + "step": 7343 + }, + { + "epoch": 3.07, + "learning_rate": 1.4105019982833674e-05, + "loss": 0.1645, + "step": 7344 + }, + { + "epoch": 3.08, + "learning_rate": 1.4103473948814189e-05, + "loss": 0.1983, + "step": 7345 + }, + { + "epoch": 3.08, + "learning_rate": 1.4101927796843984e-05, + "loss": 0.1839, + "step": 7346 + }, + { + "epoch": 3.08, + "learning_rate": 1.4100381526967502e-05, + "loss": 0.2312, + "step": 7347 + }, + { + "epoch": 3.08, + "learning_rate": 1.4098835139229195e-05, + "loss": 0.1856, + "step": 7348 + }, + { + "epoch": 3.08, + "learning_rate": 1.409728863367351e-05, + "loss": 0.1912, + "step": 7349 + }, + { + "epoch": 3.08, + "learning_rate": 1.4095742010344898e-05, + "loss": 0.1981, + "step": 7350 + }, + { + "epoch": 3.08, + "learning_rate": 1.4094195269287814e-05, + "loss": 0.197, + "step": 7351 + }, + { + "epoch": 3.08, + "learning_rate": 1.4092648410546723e-05, + "loss": 0.1855, + "step": 7352 + }, + { + "epoch": 3.08, + "learning_rate": 1.4091101434166086e-05, + "loss": 0.1751, + "step": 7353 + }, + { + "epoch": 3.08, + "learning_rate": 1.4089554340190365e-05, + "loss": 0.1941, + "step": 7354 + }, + { + "epoch": 3.08, + "learning_rate": 1.4088007128664036e-05, + "loss": 0.1755, + "step": 7355 + }, + { + "epoch": 3.08, + "learning_rate": 1.4086459799631572e-05, + "loss": 0.1973, + "step": 7356 + }, + { + "epoch": 3.08, + "learning_rate": 1.4084912353137444e-05, + "loss": 0.1854, + "step": 7357 + }, + { + "epoch": 3.08, + "learning_rate": 1.4083364789226135e-05, + "loss": 0.2054, + "step": 7358 + }, + { + "epoch": 3.08, + "learning_rate": 1.4081817107942131e-05, + "loss": 0.1948, + "step": 7359 + }, + { + "epoch": 3.08, + "learning_rate": 1.4080269309329912e-05, + "loss": 0.157, + "step": 7360 + }, + { + "epoch": 3.08, + "learning_rate": 1.4078721393433977e-05, + "loss": 0.2108, + "step": 7361 + }, + { + "epoch": 3.08, + "learning_rate": 1.4077173360298812e-05, + "loss": 0.1796, + "step": 7362 + }, + { + "epoch": 3.08, + "learning_rate": 1.407562520996892e-05, + "loss": 0.2147, + "step": 7363 + }, + { + "epoch": 3.08, + "learning_rate": 1.4074076942488794e-05, + "loss": 0.2133, + "step": 7364 + }, + { + "epoch": 3.08, + "learning_rate": 1.4072528557902944e-05, + "loss": 0.1498, + "step": 7365 + }, + { + "epoch": 3.08, + "learning_rate": 1.4070980056255876e-05, + "loss": 0.2077, + "step": 7366 + }, + { + "epoch": 3.08, + "learning_rate": 1.4069431437592095e-05, + "loss": 0.2182, + "step": 7367 + }, + { + "epoch": 3.08, + "learning_rate": 1.4067882701956121e-05, + "loss": 0.2054, + "step": 7368 + }, + { + "epoch": 3.09, + "learning_rate": 1.406633384939247e-05, + "loss": 0.1672, + "step": 7369 + }, + { + "epoch": 3.09, + "learning_rate": 1.4064784879945658e-05, + "loss": 0.1746, + "step": 7370 + }, + { + "epoch": 3.09, + "learning_rate": 1.4063235793660212e-05, + "loss": 0.1901, + "step": 7371 + }, + { + "epoch": 3.09, + "learning_rate": 1.406168659058066e-05, + "loss": 0.1435, + "step": 7372 + }, + { + "epoch": 3.09, + "learning_rate": 1.4060137270751531e-05, + "loss": 0.2062, + "step": 7373 + }, + { + "epoch": 3.09, + "learning_rate": 1.4058587834217356e-05, + "loss": 0.1671, + "step": 7374 + }, + { + "epoch": 3.09, + "learning_rate": 1.405703828102268e-05, + "loss": 0.1781, + "step": 7375 + }, + { + "epoch": 3.09, + "learning_rate": 1.4055488611212038e-05, + "loss": 0.1978, + "step": 7376 + }, + { + "epoch": 3.09, + "learning_rate": 1.405393882482997e-05, + "loss": 0.1653, + "step": 7377 + }, + { + "epoch": 3.09, + "learning_rate": 1.4052388921921033e-05, + "loss": 0.1671, + "step": 7378 + }, + { + "epoch": 3.09, + "learning_rate": 1.405083890252977e-05, + "loss": 0.1926, + "step": 7379 + }, + { + "epoch": 3.09, + "learning_rate": 1.4049288766700739e-05, + "loss": 0.1676, + "step": 7380 + }, + { + "epoch": 3.09, + "learning_rate": 1.4047738514478493e-05, + "loss": 0.1606, + "step": 7381 + }, + { + "epoch": 3.09, + "learning_rate": 1.40461881459076e-05, + "loss": 0.1998, + "step": 7382 + }, + { + "epoch": 3.09, + "learning_rate": 1.4044637661032614e-05, + "loss": 0.2041, + "step": 7383 + }, + { + "epoch": 3.09, + "learning_rate": 1.404308705989811e-05, + "loss": 0.1909, + "step": 7384 + }, + { + "epoch": 3.09, + "learning_rate": 1.4041536342548656e-05, + "loss": 0.1531, + "step": 7385 + }, + { + "epoch": 3.09, + "learning_rate": 1.4039985509028825e-05, + "loss": 0.1778, + "step": 7386 + }, + { + "epoch": 3.09, + "learning_rate": 1.4038434559383198e-05, + "loss": 0.1942, + "step": 7387 + }, + { + "epoch": 3.09, + "learning_rate": 1.4036883493656353e-05, + "loss": 0.2008, + "step": 7388 + }, + { + "epoch": 3.09, + "learning_rate": 1.4035332311892873e-05, + "loss": 0.1685, + "step": 7389 + }, + { + "epoch": 3.09, + "learning_rate": 1.4033781014137348e-05, + "loss": 0.1667, + "step": 7390 + }, + { + "epoch": 3.09, + "learning_rate": 1.4032229600434363e-05, + "loss": 0.1642, + "step": 7391 + }, + { + "epoch": 3.09, + "learning_rate": 1.403067807082852e-05, + "loss": 0.169, + "step": 7392 + }, + { + "epoch": 3.1, + "learning_rate": 1.4029126425364414e-05, + "loss": 0.183, + "step": 7393 + }, + { + "epoch": 3.1, + "learning_rate": 1.4027574664086641e-05, + "loss": 0.1868, + "step": 7394 + }, + { + "epoch": 3.1, + "learning_rate": 1.402602278703981e-05, + "loss": 0.2027, + "step": 7395 + }, + { + "epoch": 3.1, + "learning_rate": 1.4024470794268528e-05, + "loss": 0.2004, + "step": 7396 + }, + { + "epoch": 3.1, + "learning_rate": 1.4022918685817399e-05, + "loss": 0.1987, + "step": 7397 + }, + { + "epoch": 3.1, + "learning_rate": 1.4021366461731047e-05, + "loss": 0.2168, + "step": 7398 + }, + { + "epoch": 3.1, + "learning_rate": 1.4019814122054082e-05, + "loss": 0.1792, + "step": 7399 + }, + { + "epoch": 3.1, + "learning_rate": 1.4018261666831132e-05, + "loss": 0.1826, + "step": 7400 + }, + { + "epoch": 3.1, + "learning_rate": 1.4016709096106813e-05, + "loss": 0.1979, + "step": 7401 + }, + { + "epoch": 3.1, + "learning_rate": 1.4015156409925754e-05, + "loss": 0.1863, + "step": 7402 + }, + { + "epoch": 3.1, + "learning_rate": 1.4013603608332589e-05, + "loss": 0.1953, + "step": 7403 + }, + { + "epoch": 3.1, + "learning_rate": 1.4012050691371948e-05, + "loss": 0.1579, + "step": 7404 + }, + { + "epoch": 3.1, + "learning_rate": 1.4010497659088468e-05, + "loss": 0.2044, + "step": 7405 + }, + { + "epoch": 3.1, + "learning_rate": 1.4008944511526798e-05, + "loss": 0.1969, + "step": 7406 + }, + { + "epoch": 3.1, + "learning_rate": 1.400739124873157e-05, + "loss": 0.1812, + "step": 7407 + }, + { + "epoch": 3.1, + "learning_rate": 1.4005837870747439e-05, + "loss": 0.2105, + "step": 7408 + }, + { + "epoch": 3.1, + "learning_rate": 1.4004284377619053e-05, + "loss": 0.2011, + "step": 7409 + }, + { + "epoch": 3.1, + "learning_rate": 1.4002730769391069e-05, + "loss": 0.1904, + "step": 7410 + }, + { + "epoch": 3.1, + "learning_rate": 1.4001177046108136e-05, + "loss": 0.1919, + "step": 7411 + }, + { + "epoch": 3.1, + "learning_rate": 1.3999623207814921e-05, + "loss": 0.1803, + "step": 7412 + }, + { + "epoch": 3.1, + "learning_rate": 1.3998069254556088e-05, + "loss": 0.1908, + "step": 7413 + }, + { + "epoch": 3.1, + "learning_rate": 1.3996515186376298e-05, + "loss": 0.22, + "step": 7414 + }, + { + "epoch": 3.1, + "learning_rate": 1.399496100332023e-05, + "loss": 0.174, + "step": 7415 + }, + { + "epoch": 3.1, + "learning_rate": 1.399340670543255e-05, + "loss": 0.2192, + "step": 7416 + }, + { + "epoch": 3.11, + "learning_rate": 1.3991852292757937e-05, + "loss": 0.1978, + "step": 7417 + }, + { + "epoch": 3.11, + "learning_rate": 1.3990297765341074e-05, + "loss": 0.1474, + "step": 7418 + }, + { + "epoch": 3.11, + "learning_rate": 1.3988743123226643e-05, + "loss": 0.1784, + "step": 7419 + }, + { + "epoch": 3.11, + "learning_rate": 1.398718836645933e-05, + "loss": 0.177, + "step": 7420 + }, + { + "epoch": 3.11, + "learning_rate": 1.3985633495083828e-05, + "loss": 0.2083, + "step": 7421 + }, + { + "epoch": 3.11, + "learning_rate": 1.3984078509144824e-05, + "loss": 0.2083, + "step": 7422 + }, + { + "epoch": 3.11, + "learning_rate": 1.398252340868702e-05, + "loss": 0.1653, + "step": 7423 + }, + { + "epoch": 3.11, + "learning_rate": 1.3980968193755115e-05, + "loss": 0.1788, + "step": 7424 + }, + { + "epoch": 3.11, + "learning_rate": 1.397941286439381e-05, + "loss": 0.17, + "step": 7425 + }, + { + "epoch": 3.11, + "learning_rate": 1.3977857420647816e-05, + "loss": 0.1635, + "step": 7426 + }, + { + "epoch": 3.11, + "learning_rate": 1.3976301862561841e-05, + "loss": 0.1704, + "step": 7427 + }, + { + "epoch": 3.11, + "learning_rate": 1.3974746190180595e-05, + "loss": 0.1724, + "step": 7428 + }, + { + "epoch": 3.11, + "learning_rate": 1.3973190403548798e-05, + "loss": 0.1812, + "step": 7429 + }, + { + "epoch": 3.11, + "learning_rate": 1.3971634502711169e-05, + "loss": 0.1525, + "step": 7430 + }, + { + "epoch": 3.11, + "learning_rate": 1.397007848771243e-05, + "loss": 0.1913, + "step": 7431 + }, + { + "epoch": 3.11, + "learning_rate": 1.3968522358597307e-05, + "loss": 0.1774, + "step": 7432 + }, + { + "epoch": 3.11, + "learning_rate": 1.3966966115410533e-05, + "loss": 0.1482, + "step": 7433 + }, + { + "epoch": 3.11, + "learning_rate": 1.3965409758196838e-05, + "loss": 0.2149, + "step": 7434 + }, + { + "epoch": 3.11, + "learning_rate": 1.3963853287000956e-05, + "loss": 0.1504, + "step": 7435 + }, + { + "epoch": 3.11, + "learning_rate": 1.396229670186763e-05, + "loss": 0.2362, + "step": 7436 + }, + { + "epoch": 3.11, + "learning_rate": 1.3960740002841602e-05, + "loss": 0.1958, + "step": 7437 + }, + { + "epoch": 3.11, + "learning_rate": 1.3959183189967615e-05, + "loss": 0.1642, + "step": 7438 + }, + { + "epoch": 3.11, + "learning_rate": 1.395762626329042e-05, + "loss": 0.1807, + "step": 7439 + }, + { + "epoch": 3.11, + "learning_rate": 1.3956069222854774e-05, + "loss": 0.1854, + "step": 7440 + }, + { + "epoch": 3.12, + "learning_rate": 1.3954512068705425e-05, + "loss": 0.2109, + "step": 7441 + }, + { + "epoch": 3.12, + "learning_rate": 1.3952954800887137e-05, + "loss": 0.1881, + "step": 7442 + }, + { + "epoch": 3.12, + "learning_rate": 1.3951397419444675e-05, + "loss": 0.1886, + "step": 7443 + }, + { + "epoch": 3.12, + "learning_rate": 1.3949839924422798e-05, + "loss": 0.1795, + "step": 7444 + }, + { + "epoch": 3.12, + "learning_rate": 1.3948282315866274e-05, + "loss": 0.2422, + "step": 7445 + }, + { + "epoch": 3.12, + "learning_rate": 1.394672459381988e-05, + "loss": 0.2021, + "step": 7446 + }, + { + "epoch": 3.12, + "learning_rate": 1.3945166758328394e-05, + "loss": 0.1772, + "step": 7447 + }, + { + "epoch": 3.12, + "learning_rate": 1.3943608809436587e-05, + "loss": 0.2052, + "step": 7448 + }, + { + "epoch": 3.12, + "learning_rate": 1.3942050747189242e-05, + "loss": 0.1767, + "step": 7449 + }, + { + "epoch": 3.12, + "learning_rate": 1.394049257163115e-05, + "loss": 0.1822, + "step": 7450 + }, + { + "epoch": 3.12, + "learning_rate": 1.3938934282807095e-05, + "loss": 0.1777, + "step": 7451 + }, + { + "epoch": 3.12, + "learning_rate": 1.3937375880761868e-05, + "loss": 0.1736, + "step": 7452 + }, + { + "epoch": 3.12, + "learning_rate": 1.3935817365540268e-05, + "loss": 0.1833, + "step": 7453 + }, + { + "epoch": 3.12, + "learning_rate": 1.3934258737187087e-05, + "loss": 0.2147, + "step": 7454 + }, + { + "epoch": 3.12, + "learning_rate": 1.3932699995747134e-05, + "loss": 0.1847, + "step": 7455 + }, + { + "epoch": 3.12, + "learning_rate": 1.3931141141265206e-05, + "loss": 0.1931, + "step": 7456 + }, + { + "epoch": 3.12, + "learning_rate": 1.3929582173786116e-05, + "loss": 0.15, + "step": 7457 + }, + { + "epoch": 3.12, + "learning_rate": 1.3928023093354671e-05, + "loss": 0.2478, + "step": 7458 + }, + { + "epoch": 3.12, + "learning_rate": 1.392646390001569e-05, + "loss": 0.158, + "step": 7459 + }, + { + "epoch": 3.12, + "learning_rate": 1.3924904593813986e-05, + "loss": 0.1903, + "step": 7460 + }, + { + "epoch": 3.12, + "learning_rate": 1.3923345174794386e-05, + "loss": 0.2119, + "step": 7461 + }, + { + "epoch": 3.12, + "learning_rate": 1.3921785643001708e-05, + "loss": 0.165, + "step": 7462 + }, + { + "epoch": 3.12, + "learning_rate": 1.3920225998480782e-05, + "loss": 0.1985, + "step": 7463 + }, + { + "epoch": 3.12, + "learning_rate": 1.3918666241276442e-05, + "loss": 0.1734, + "step": 7464 + }, + { + "epoch": 3.13, + "learning_rate": 1.3917106371433513e-05, + "loss": 0.202, + "step": 7465 + }, + { + "epoch": 3.13, + "learning_rate": 1.3915546388996838e-05, + "loss": 0.1719, + "step": 7466 + }, + { + "epoch": 3.13, + "learning_rate": 1.3913986294011261e-05, + "loss": 0.1945, + "step": 7467 + }, + { + "epoch": 3.13, + "learning_rate": 1.391242608652162e-05, + "loss": 0.1596, + "step": 7468 + }, + { + "epoch": 3.13, + "learning_rate": 1.391086576657276e-05, + "loss": 0.2132, + "step": 7469 + }, + { + "epoch": 3.13, + "learning_rate": 1.3909305334209536e-05, + "loss": 0.1928, + "step": 7470 + }, + { + "epoch": 3.13, + "learning_rate": 1.3907744789476799e-05, + "loss": 0.1837, + "step": 7471 + }, + { + "epoch": 3.13, + "learning_rate": 1.3906184132419405e-05, + "loss": 0.1827, + "step": 7472 + }, + { + "epoch": 3.13, + "learning_rate": 1.3904623363082213e-05, + "loss": 0.192, + "step": 7473 + }, + { + "epoch": 3.13, + "learning_rate": 1.390306248151009e-05, + "loss": 0.1669, + "step": 7474 + }, + { + "epoch": 3.13, + "learning_rate": 1.3901501487747897e-05, + "loss": 0.163, + "step": 7475 + }, + { + "epoch": 3.13, + "learning_rate": 1.389994038184051e-05, + "loss": 0.1953, + "step": 7476 + }, + { + "epoch": 3.13, + "learning_rate": 1.3898379163832794e-05, + "loss": 0.1866, + "step": 7477 + }, + { + "epoch": 3.13, + "learning_rate": 1.3896817833769631e-05, + "loss": 0.1751, + "step": 7478 + }, + { + "epoch": 3.13, + "learning_rate": 1.3895256391695896e-05, + "loss": 0.1909, + "step": 7479 + }, + { + "epoch": 3.13, + "learning_rate": 1.3893694837656471e-05, + "loss": 0.2215, + "step": 7480 + }, + { + "epoch": 3.13, + "learning_rate": 1.3892133171696246e-05, + "loss": 0.1652, + "step": 7481 + }, + { + "epoch": 3.13, + "learning_rate": 1.3890571393860101e-05, + "loss": 0.1398, + "step": 7482 + }, + { + "epoch": 3.13, + "learning_rate": 1.3889009504192938e-05, + "loss": 0.1765, + "step": 7483 + }, + { + "epoch": 3.13, + "learning_rate": 1.3887447502739649e-05, + "loss": 0.1829, + "step": 7484 + }, + { + "epoch": 3.13, + "learning_rate": 1.388588538954513e-05, + "loss": 0.2005, + "step": 7485 + }, + { + "epoch": 3.13, + "learning_rate": 1.3884323164654284e-05, + "loss": 0.1762, + "step": 7486 + }, + { + "epoch": 3.13, + "learning_rate": 1.3882760828112014e-05, + "loss": 0.2078, + "step": 7487 + }, + { + "epoch": 3.14, + "learning_rate": 1.388119837996323e-05, + "loss": 0.1795, + "step": 7488 + }, + { + "epoch": 3.14, + "learning_rate": 1.3879635820252846e-05, + "loss": 0.1307, + "step": 7489 + }, + { + "epoch": 3.14, + "learning_rate": 1.3878073149025768e-05, + "loss": 0.176, + "step": 7490 + }, + { + "epoch": 3.14, + "learning_rate": 1.3876510366326922e-05, + "loss": 0.207, + "step": 7491 + }, + { + "epoch": 3.14, + "learning_rate": 1.3874947472201226e-05, + "loss": 0.1517, + "step": 7492 + }, + { + "epoch": 3.14, + "learning_rate": 1.38733844666936e-05, + "loss": 0.1875, + "step": 7493 + }, + { + "epoch": 3.14, + "learning_rate": 1.3871821349848977e-05, + "loss": 0.1754, + "step": 7494 + }, + { + "epoch": 3.14, + "learning_rate": 1.3870258121712285e-05, + "loss": 0.1777, + "step": 7495 + }, + { + "epoch": 3.14, + "learning_rate": 1.3868694782328456e-05, + "loss": 0.2025, + "step": 7496 + }, + { + "epoch": 3.14, + "learning_rate": 1.3867131331742429e-05, + "loss": 0.1829, + "step": 7497 + }, + { + "epoch": 3.14, + "learning_rate": 1.3865567769999146e-05, + "loss": 0.176, + "step": 7498 + }, + { + "epoch": 3.14, + "learning_rate": 1.3864004097143546e-05, + "loss": 0.1477, + "step": 7499 + }, + { + "epoch": 3.14, + "learning_rate": 1.3862440313220575e-05, + "loss": 0.1657, + "step": 7500 + }, + { + "epoch": 3.14, + "learning_rate": 1.3860876418275189e-05, + "loss": 0.1488, + "step": 7501 + }, + { + "epoch": 3.14, + "learning_rate": 1.3859312412352335e-05, + "loss": 0.193, + "step": 7502 + }, + { + "epoch": 3.14, + "learning_rate": 1.3857748295496972e-05, + "loss": 0.1975, + "step": 7503 + }, + { + "epoch": 3.14, + "learning_rate": 1.3856184067754053e-05, + "loss": 0.1794, + "step": 7504 + }, + { + "epoch": 3.14, + "learning_rate": 1.3854619729168548e-05, + "loss": 0.161, + "step": 7505 + }, + { + "epoch": 3.14, + "learning_rate": 1.3853055279785416e-05, + "loss": 0.1941, + "step": 7506 + }, + { + "epoch": 3.14, + "learning_rate": 1.3851490719649631e-05, + "loss": 0.2179, + "step": 7507 + }, + { + "epoch": 3.14, + "learning_rate": 1.3849926048806166e-05, + "loss": 0.1785, + "step": 7508 + }, + { + "epoch": 3.14, + "learning_rate": 1.384836126729999e-05, + "loss": 0.2004, + "step": 7509 + }, + { + "epoch": 3.14, + "learning_rate": 1.3846796375176083e-05, + "loss": 0.2078, + "step": 7510 + }, + { + "epoch": 3.14, + "learning_rate": 1.3845231372479432e-05, + "loss": 0.1929, + "step": 7511 + }, + { + "epoch": 3.15, + "learning_rate": 1.3843666259255018e-05, + "loss": 0.1956, + "step": 7512 + }, + { + "epoch": 3.15, + "learning_rate": 1.3842101035547827e-05, + "loss": 0.2089, + "step": 7513 + }, + { + "epoch": 3.15, + "learning_rate": 1.3840535701402847e-05, + "loss": 0.184, + "step": 7514 + }, + { + "epoch": 3.15, + "learning_rate": 1.3838970256865078e-05, + "loss": 0.1952, + "step": 7515 + }, + { + "epoch": 3.15, + "learning_rate": 1.3837404701979517e-05, + "loss": 0.1959, + "step": 7516 + }, + { + "epoch": 3.15, + "learning_rate": 1.3835839036791163e-05, + "loss": 0.2267, + "step": 7517 + }, + { + "epoch": 3.15, + "learning_rate": 1.383427326134502e-05, + "loss": 0.2016, + "step": 7518 + }, + { + "epoch": 3.15, + "learning_rate": 1.3832707375686095e-05, + "loss": 0.1755, + "step": 7519 + }, + { + "epoch": 3.15, + "learning_rate": 1.3831141379859394e-05, + "loss": 0.2223, + "step": 7520 + }, + { + "epoch": 3.15, + "learning_rate": 1.3829575273909939e-05, + "loss": 0.1427, + "step": 7521 + }, + { + "epoch": 3.15, + "learning_rate": 1.3828009057882737e-05, + "loss": 0.1875, + "step": 7522 + }, + { + "epoch": 3.15, + "learning_rate": 1.3826442731822814e-05, + "loss": 0.1887, + "step": 7523 + }, + { + "epoch": 3.15, + "learning_rate": 1.3824876295775187e-05, + "loss": 0.1875, + "step": 7524 + }, + { + "epoch": 3.15, + "learning_rate": 1.3823309749784888e-05, + "loss": 0.1904, + "step": 7525 + }, + { + "epoch": 3.15, + "learning_rate": 1.3821743093896942e-05, + "loss": 0.1796, + "step": 7526 + }, + { + "epoch": 3.15, + "learning_rate": 1.3820176328156381e-05, + "loss": 0.1889, + "step": 7527 + }, + { + "epoch": 3.15, + "learning_rate": 1.3818609452608245e-05, + "loss": 0.1661, + "step": 7528 + }, + { + "epoch": 3.15, + "learning_rate": 1.3817042467297565e-05, + "loss": 0.173, + "step": 7529 + }, + { + "epoch": 3.15, + "learning_rate": 1.3815475372269386e-05, + "loss": 0.1865, + "step": 7530 + }, + { + "epoch": 3.15, + "learning_rate": 1.3813908167568757e-05, + "loss": 0.1942, + "step": 7531 + }, + { + "epoch": 3.15, + "learning_rate": 1.381234085324072e-05, + "loss": 0.2108, + "step": 7532 + }, + { + "epoch": 3.15, + "learning_rate": 1.381077342933033e-05, + "loss": 0.1843, + "step": 7533 + }, + { + "epoch": 3.15, + "learning_rate": 1.3809205895882635e-05, + "loss": 0.2048, + "step": 7534 + }, + { + "epoch": 3.15, + "learning_rate": 1.3807638252942704e-05, + "loss": 0.1397, + "step": 7535 + }, + { + "epoch": 3.16, + "learning_rate": 1.3806070500555583e-05, + "loss": 0.2272, + "step": 7536 + }, + { + "epoch": 3.16, + "learning_rate": 1.3804502638766346e-05, + "loss": 0.1862, + "step": 7537 + }, + { + "epoch": 3.16, + "learning_rate": 1.3802934667620055e-05, + "loss": 0.1913, + "step": 7538 + }, + { + "epoch": 3.16, + "learning_rate": 1.3801366587161786e-05, + "loss": 0.1688, + "step": 7539 + }, + { + "epoch": 3.16, + "learning_rate": 1.3799798397436603e-05, + "loss": 0.2065, + "step": 7540 + }, + { + "epoch": 3.16, + "learning_rate": 1.379823009848959e-05, + "loss": 0.1557, + "step": 7541 + }, + { + "epoch": 3.16, + "learning_rate": 1.3796661690365823e-05, + "loss": 0.1426, + "step": 7542 + }, + { + "epoch": 3.16, + "learning_rate": 1.379509317311038e-05, + "loss": 0.2024, + "step": 7543 + }, + { + "epoch": 3.16, + "learning_rate": 1.3793524546768358e-05, + "loss": 0.2058, + "step": 7544 + }, + { + "epoch": 3.16, + "learning_rate": 1.3791955811384837e-05, + "loss": 0.1913, + "step": 7545 + }, + { + "epoch": 3.16, + "learning_rate": 1.3790386967004911e-05, + "loss": 0.1702, + "step": 7546 + }, + { + "epoch": 3.16, + "learning_rate": 1.3788818013673675e-05, + "loss": 0.1924, + "step": 7547 + }, + { + "epoch": 3.16, + "learning_rate": 1.3787248951436225e-05, + "loss": 0.1676, + "step": 7548 + }, + { + "epoch": 3.16, + "learning_rate": 1.3785679780337668e-05, + "loss": 0.1956, + "step": 7549 + }, + { + "epoch": 3.16, + "learning_rate": 1.3784110500423104e-05, + "loss": 0.2423, + "step": 7550 + }, + { + "epoch": 3.16, + "learning_rate": 1.3782541111737642e-05, + "loss": 0.2112, + "step": 7551 + }, + { + "epoch": 3.16, + "learning_rate": 1.3780971614326393e-05, + "loss": 0.1864, + "step": 7552 + }, + { + "epoch": 3.16, + "learning_rate": 1.3779402008234467e-05, + "loss": 0.2137, + "step": 7553 + }, + { + "epoch": 3.16, + "learning_rate": 1.3777832293506985e-05, + "loss": 0.211, + "step": 7554 + }, + { + "epoch": 3.16, + "learning_rate": 1.377626247018907e-05, + "loss": 0.1677, + "step": 7555 + }, + { + "epoch": 3.16, + "learning_rate": 1.3774692538325835e-05, + "loss": 0.2024, + "step": 7556 + }, + { + "epoch": 3.16, + "learning_rate": 1.3773122497962418e-05, + "loss": 0.1875, + "step": 7557 + }, + { + "epoch": 3.16, + "learning_rate": 1.3771552349143942e-05, + "loss": 0.2106, + "step": 7558 + }, + { + "epoch": 3.16, + "learning_rate": 1.3769982091915539e-05, + "loss": 0.1844, + "step": 7559 + }, + { + "epoch": 3.17, + "learning_rate": 1.3768411726322347e-05, + "loss": 0.241, + "step": 7560 + }, + { + "epoch": 3.17, + "learning_rate": 1.3766841252409502e-05, + "loss": 0.1707, + "step": 7561 + }, + { + "epoch": 3.17, + "learning_rate": 1.3765270670222153e-05, + "loss": 0.1793, + "step": 7562 + }, + { + "epoch": 3.17, + "learning_rate": 1.3763699979805434e-05, + "loss": 0.1644, + "step": 7563 + }, + { + "epoch": 3.17, + "learning_rate": 1.3762129181204502e-05, + "loss": 0.2117, + "step": 7564 + }, + { + "epoch": 3.17, + "learning_rate": 1.3760558274464506e-05, + "loss": 0.1701, + "step": 7565 + }, + { + "epoch": 3.17, + "learning_rate": 1.3758987259630597e-05, + "loss": 0.185, + "step": 7566 + }, + { + "epoch": 3.17, + "learning_rate": 1.3757416136747935e-05, + "loss": 0.1732, + "step": 7567 + }, + { + "epoch": 3.17, + "learning_rate": 1.375584490586168e-05, + "loss": 0.218, + "step": 7568 + }, + { + "epoch": 3.17, + "learning_rate": 1.3754273567017e-05, + "loss": 0.1982, + "step": 7569 + }, + { + "epoch": 3.17, + "learning_rate": 1.3752702120259055e-05, + "loss": 0.2061, + "step": 7570 + }, + { + "epoch": 3.17, + "learning_rate": 1.3751130565633015e-05, + "loss": 0.1819, + "step": 7571 + }, + { + "epoch": 3.17, + "learning_rate": 1.3749558903184059e-05, + "loss": 0.1826, + "step": 7572 + }, + { + "epoch": 3.17, + "learning_rate": 1.3747987132957361e-05, + "loss": 0.1503, + "step": 7573 + }, + { + "epoch": 3.17, + "learning_rate": 1.3746415254998095e-05, + "loss": 0.1801, + "step": 7574 + }, + { + "epoch": 3.17, + "learning_rate": 1.3744843269351446e-05, + "loss": 0.1277, + "step": 7575 + }, + { + "epoch": 3.17, + "learning_rate": 1.3743271176062605e-05, + "loss": 0.1731, + "step": 7576 + }, + { + "epoch": 3.17, + "learning_rate": 1.3741698975176754e-05, + "loss": 0.2044, + "step": 7577 + }, + { + "epoch": 3.17, + "learning_rate": 1.3740126666739086e-05, + "loss": 0.1722, + "step": 7578 + }, + { + "epoch": 3.17, + "learning_rate": 1.3738554250794795e-05, + "loss": 0.1917, + "step": 7579 + }, + { + "epoch": 3.17, + "learning_rate": 1.3736981727389083e-05, + "loss": 0.2002, + "step": 7580 + }, + { + "epoch": 3.17, + "learning_rate": 1.3735409096567145e-05, + "loss": 0.1809, + "step": 7581 + }, + { + "epoch": 3.17, + "learning_rate": 1.3733836358374186e-05, + "loss": 0.1463, + "step": 7582 + }, + { + "epoch": 3.17, + "learning_rate": 1.3732263512855416e-05, + "loss": 0.1965, + "step": 7583 + }, + { + "epoch": 3.18, + "learning_rate": 1.3730690560056043e-05, + "loss": 0.1524, + "step": 7584 + }, + { + "epoch": 3.18, + "learning_rate": 1.3729117500021279e-05, + "loss": 0.1706, + "step": 7585 + }, + { + "epoch": 3.18, + "learning_rate": 1.3727544332796344e-05, + "loss": 0.1952, + "step": 7586 + }, + { + "epoch": 3.18, + "learning_rate": 1.3725971058426453e-05, + "loss": 0.1621, + "step": 7587 + }, + { + "epoch": 3.18, + "learning_rate": 1.3724397676956831e-05, + "loss": 0.2063, + "step": 7588 + }, + { + "epoch": 3.18, + "learning_rate": 1.3722824188432702e-05, + "loss": 0.2065, + "step": 7589 + }, + { + "epoch": 3.18, + "learning_rate": 1.3721250592899295e-05, + "loss": 0.2085, + "step": 7590 + }, + { + "epoch": 3.18, + "learning_rate": 1.3719676890401845e-05, + "loss": 0.1861, + "step": 7591 + }, + { + "epoch": 3.18, + "learning_rate": 1.371810308098558e-05, + "loss": 0.1507, + "step": 7592 + }, + { + "epoch": 3.18, + "learning_rate": 1.3716529164695744e-05, + "loss": 0.1772, + "step": 7593 + }, + { + "epoch": 3.18, + "learning_rate": 1.3714955141577573e-05, + "loss": 0.2075, + "step": 7594 + }, + { + "epoch": 3.18, + "learning_rate": 1.3713381011676315e-05, + "loss": 0.2258, + "step": 7595 + }, + { + "epoch": 3.18, + "learning_rate": 1.3711806775037215e-05, + "loss": 0.2036, + "step": 7596 + }, + { + "epoch": 3.18, + "learning_rate": 1.3710232431705521e-05, + "loss": 0.1845, + "step": 7597 + }, + { + "epoch": 3.18, + "learning_rate": 1.3708657981726487e-05, + "loss": 0.1761, + "step": 7598 + }, + { + "epoch": 3.18, + "learning_rate": 1.3707083425145375e-05, + "loss": 0.1685, + "step": 7599 + }, + { + "epoch": 3.18, + "learning_rate": 1.3705508762007436e-05, + "loss": 0.1957, + "step": 7600 + }, + { + "epoch": 3.18, + "learning_rate": 1.3703933992357936e-05, + "loss": 0.1601, + "step": 7601 + }, + { + "epoch": 3.18, + "learning_rate": 1.370235911624214e-05, + "loss": 0.1924, + "step": 7602 + }, + { + "epoch": 3.18, + "learning_rate": 1.370078413370532e-05, + "loss": 0.1813, + "step": 7603 + }, + { + "epoch": 3.18, + "learning_rate": 1.3699209044792743e-05, + "loss": 0.2021, + "step": 7604 + }, + { + "epoch": 3.18, + "learning_rate": 1.3697633849549683e-05, + "loss": 0.1347, + "step": 7605 + }, + { + "epoch": 3.18, + "learning_rate": 1.369605854802142e-05, + "loss": 0.1988, + "step": 7606 + }, + { + "epoch": 3.18, + "learning_rate": 1.3694483140253234e-05, + "loss": 0.1635, + "step": 7607 + }, + { + "epoch": 3.19, + "learning_rate": 1.3692907626290406e-05, + "loss": 0.1745, + "step": 7608 + }, + { + "epoch": 3.19, + "learning_rate": 1.3691332006178226e-05, + "loss": 0.1576, + "step": 7609 + }, + { + "epoch": 3.19, + "learning_rate": 1.3689756279961986e-05, + "loss": 0.1684, + "step": 7610 + }, + { + "epoch": 3.19, + "learning_rate": 1.3688180447686975e-05, + "loss": 0.1723, + "step": 7611 + }, + { + "epoch": 3.19, + "learning_rate": 1.3686604509398489e-05, + "loss": 0.1553, + "step": 7612 + }, + { + "epoch": 3.19, + "learning_rate": 1.3685028465141828e-05, + "loss": 0.1632, + "step": 7613 + }, + { + "epoch": 3.19, + "learning_rate": 1.3683452314962295e-05, + "loss": 0.1662, + "step": 7614 + }, + { + "epoch": 3.19, + "learning_rate": 1.3681876058905193e-05, + "loss": 0.2028, + "step": 7615 + }, + { + "epoch": 3.19, + "learning_rate": 1.368029969701583e-05, + "loss": 0.2133, + "step": 7616 + }, + { + "epoch": 3.19, + "learning_rate": 1.3678723229339518e-05, + "loss": 0.1861, + "step": 7617 + }, + { + "epoch": 3.19, + "learning_rate": 1.3677146655921573e-05, + "loss": 0.2252, + "step": 7618 + }, + { + "epoch": 3.19, + "learning_rate": 1.3675569976807307e-05, + "loss": 0.1522, + "step": 7619 + }, + { + "epoch": 3.19, + "learning_rate": 1.3673993192042048e-05, + "loss": 0.2132, + "step": 7620 + }, + { + "epoch": 3.19, + "learning_rate": 1.3672416301671111e-05, + "loss": 0.1779, + "step": 7621 + }, + { + "epoch": 3.19, + "learning_rate": 1.3670839305739828e-05, + "loss": 0.1832, + "step": 7622 + }, + { + "epoch": 3.19, + "learning_rate": 1.366926220429353e-05, + "loss": 0.1864, + "step": 7623 + }, + { + "epoch": 3.19, + "learning_rate": 1.3667684997377542e-05, + "loss": 0.2032, + "step": 7624 + }, + { + "epoch": 3.19, + "learning_rate": 1.3666107685037204e-05, + "loss": 0.1911, + "step": 7625 + }, + { + "epoch": 3.19, + "learning_rate": 1.3664530267317855e-05, + "loss": 0.1678, + "step": 7626 + }, + { + "epoch": 3.19, + "learning_rate": 1.3662952744264835e-05, + "loss": 0.1906, + "step": 7627 + }, + { + "epoch": 3.19, + "learning_rate": 1.3661375115923487e-05, + "loss": 0.2059, + "step": 7628 + }, + { + "epoch": 3.19, + "learning_rate": 1.3659797382339162e-05, + "loss": 0.1447, + "step": 7629 + }, + { + "epoch": 3.19, + "learning_rate": 1.3658219543557208e-05, + "loss": 0.1612, + "step": 7630 + }, + { + "epoch": 3.19, + "learning_rate": 1.365664159962298e-05, + "loss": 0.1815, + "step": 7631 + }, + { + "epoch": 3.2, + "learning_rate": 1.3655063550581834e-05, + "loss": 0.1874, + "step": 7632 + }, + { + "epoch": 3.2, + "learning_rate": 1.3653485396479129e-05, + "loss": 0.1905, + "step": 7633 + }, + { + "epoch": 3.2, + "learning_rate": 1.365190713736023e-05, + "loss": 0.1837, + "step": 7634 + }, + { + "epoch": 3.2, + "learning_rate": 1.3650328773270498e-05, + "loss": 0.2112, + "step": 7635 + }, + { + "epoch": 3.2, + "learning_rate": 1.3648750304255307e-05, + "loss": 0.1932, + "step": 7636 + }, + { + "epoch": 3.2, + "learning_rate": 1.3647171730360026e-05, + "loss": 0.134, + "step": 7637 + }, + { + "epoch": 3.2, + "learning_rate": 1.3645593051630032e-05, + "loss": 0.18, + "step": 7638 + }, + { + "epoch": 3.2, + "learning_rate": 1.3644014268110697e-05, + "loss": 0.1485, + "step": 7639 + }, + { + "epoch": 3.2, + "learning_rate": 1.364243537984741e-05, + "loss": 0.1891, + "step": 7640 + }, + { + "epoch": 3.2, + "learning_rate": 1.3640856386885546e-05, + "loss": 0.202, + "step": 7641 + }, + { + "epoch": 3.2, + "learning_rate": 1.3639277289270497e-05, + "loss": 0.1828, + "step": 7642 + }, + { + "epoch": 3.2, + "learning_rate": 1.3637698087047651e-05, + "loss": 0.149, + "step": 7643 + }, + { + "epoch": 3.2, + "learning_rate": 1.3636118780262405e-05, + "loss": 0.1346, + "step": 7644 + }, + { + "epoch": 3.2, + "learning_rate": 1.3634539368960147e-05, + "loss": 0.1852, + "step": 7645 + }, + { + "epoch": 3.2, + "learning_rate": 1.363295985318628e-05, + "loss": 0.1923, + "step": 7646 + }, + { + "epoch": 3.2, + "learning_rate": 1.363138023298621e-05, + "loss": 0.182, + "step": 7647 + }, + { + "epoch": 3.2, + "learning_rate": 1.3629800508405338e-05, + "loss": 0.1883, + "step": 7648 + }, + { + "epoch": 3.2, + "learning_rate": 1.3628220679489069e-05, + "loss": 0.2061, + "step": 7649 + }, + { + "epoch": 3.2, + "learning_rate": 1.3626640746282815e-05, + "loss": 0.2, + "step": 7650 + }, + { + "epoch": 3.2, + "learning_rate": 1.3625060708831992e-05, + "loss": 0.1932, + "step": 7651 + }, + { + "epoch": 3.2, + "learning_rate": 1.3623480567182018e-05, + "loss": 0.154, + "step": 7652 + }, + { + "epoch": 3.2, + "learning_rate": 1.3621900321378308e-05, + "loss": 0.1829, + "step": 7653 + }, + { + "epoch": 3.2, + "learning_rate": 1.3620319971466288e-05, + "loss": 0.2042, + "step": 7654 + }, + { + "epoch": 3.2, + "learning_rate": 1.3618739517491381e-05, + "loss": 0.2031, + "step": 7655 + }, + { + "epoch": 3.21, + "learning_rate": 1.361715895949902e-05, + "loss": 0.1759, + "step": 7656 + }, + { + "epoch": 3.21, + "learning_rate": 1.3615578297534635e-05, + "loss": 0.2407, + "step": 7657 + }, + { + "epoch": 3.21, + "learning_rate": 1.3613997531643657e-05, + "loss": 0.1647, + "step": 7658 + }, + { + "epoch": 3.21, + "learning_rate": 1.3612416661871532e-05, + "loss": 0.2102, + "step": 7659 + }, + { + "epoch": 3.21, + "learning_rate": 1.361083568826369e-05, + "loss": 0.1727, + "step": 7660 + }, + { + "epoch": 3.21, + "learning_rate": 1.3609254610865587e-05, + "loss": 0.2066, + "step": 7661 + }, + { + "epoch": 3.21, + "learning_rate": 1.3607673429722658e-05, + "loss": 0.1782, + "step": 7662 + }, + { + "epoch": 3.21, + "learning_rate": 1.3606092144880359e-05, + "loss": 0.1969, + "step": 7663 + }, + { + "epoch": 3.21, + "learning_rate": 1.3604510756384141e-05, + "loss": 0.1575, + "step": 7664 + }, + { + "epoch": 3.21, + "learning_rate": 1.360292926427946e-05, + "loss": 0.1964, + "step": 7665 + }, + { + "epoch": 3.21, + "learning_rate": 1.3601347668611772e-05, + "loss": 0.1888, + "step": 7666 + }, + { + "epoch": 3.21, + "learning_rate": 1.3599765969426546e-05, + "loss": 0.1649, + "step": 7667 + }, + { + "epoch": 3.21, + "learning_rate": 1.359818416676924e-05, + "loss": 0.1653, + "step": 7668 + }, + { + "epoch": 3.21, + "learning_rate": 1.3596602260685321e-05, + "loss": 0.1853, + "step": 7669 + }, + { + "epoch": 3.21, + "learning_rate": 1.3595020251220263e-05, + "loss": 0.1915, + "step": 7670 + }, + { + "epoch": 3.21, + "learning_rate": 1.3593438138419539e-05, + "loss": 0.1966, + "step": 7671 + }, + { + "epoch": 3.21, + "learning_rate": 1.3591855922328624e-05, + "loss": 0.2002, + "step": 7672 + }, + { + "epoch": 3.21, + "learning_rate": 1.3590273602992996e-05, + "loss": 0.1996, + "step": 7673 + }, + { + "epoch": 3.21, + "learning_rate": 1.3588691180458143e-05, + "loss": 0.1883, + "step": 7674 + }, + { + "epoch": 3.21, + "learning_rate": 1.3587108654769542e-05, + "loss": 0.1751, + "step": 7675 + }, + { + "epoch": 3.21, + "learning_rate": 1.3585526025972688e-05, + "loss": 0.1636, + "step": 7676 + }, + { + "epoch": 3.21, + "learning_rate": 1.3583943294113071e-05, + "loss": 0.2001, + "step": 7677 + }, + { + "epoch": 3.21, + "learning_rate": 1.3582360459236183e-05, + "loss": 0.1908, + "step": 7678 + }, + { + "epoch": 3.21, + "learning_rate": 1.3580777521387523e-05, + "loss": 0.1748, + "step": 7679 + }, + { + "epoch": 3.22, + "learning_rate": 1.357919448061259e-05, + "loss": 0.1909, + "step": 7680 + }, + { + "epoch": 3.22, + "learning_rate": 1.3577611336956888e-05, + "loss": 0.1817, + "step": 7681 + }, + { + "epoch": 3.22, + "learning_rate": 1.3576028090465924e-05, + "loss": 0.1583, + "step": 7682 + }, + { + "epoch": 3.22, + "learning_rate": 1.3574444741185207e-05, + "loss": 0.1666, + "step": 7683 + }, + { + "epoch": 3.22, + "learning_rate": 1.3572861289160245e-05, + "loss": 0.2047, + "step": 7684 + }, + { + "epoch": 3.22, + "learning_rate": 1.3571277734436556e-05, + "loss": 0.1697, + "step": 7685 + }, + { + "epoch": 3.22, + "learning_rate": 1.356969407705966e-05, + "loss": 0.1998, + "step": 7686 + }, + { + "epoch": 3.22, + "learning_rate": 1.3568110317075071e-05, + "loss": 0.1905, + "step": 7687 + }, + { + "epoch": 3.22, + "learning_rate": 1.356652645452832e-05, + "loss": 0.1819, + "step": 7688 + }, + { + "epoch": 3.22, + "learning_rate": 1.3564942489464932e-05, + "loss": 0.2108, + "step": 7689 + }, + { + "epoch": 3.22, + "learning_rate": 1.3563358421930433e-05, + "loss": 0.2004, + "step": 7690 + }, + { + "epoch": 3.22, + "learning_rate": 1.356177425197036e-05, + "loss": 0.178, + "step": 7691 + }, + { + "epoch": 3.22, + "learning_rate": 1.3560189979630247e-05, + "loss": 0.2014, + "step": 7692 + }, + { + "epoch": 3.22, + "learning_rate": 1.3558605604955633e-05, + "loss": 0.1777, + "step": 7693 + }, + { + "epoch": 3.22, + "learning_rate": 1.3557021127992055e-05, + "loss": 0.172, + "step": 7694 + }, + { + "epoch": 3.22, + "learning_rate": 1.3555436548785066e-05, + "loss": 0.1939, + "step": 7695 + }, + { + "epoch": 3.22, + "learning_rate": 1.3553851867380205e-05, + "loss": 0.1617, + "step": 7696 + }, + { + "epoch": 3.22, + "learning_rate": 1.3552267083823028e-05, + "loss": 0.1244, + "step": 7697 + }, + { + "epoch": 3.22, + "learning_rate": 1.3550682198159084e-05, + "loss": 0.1537, + "step": 7698 + }, + { + "epoch": 3.22, + "learning_rate": 1.3549097210433932e-05, + "loss": 0.1861, + "step": 7699 + }, + { + "epoch": 3.22, + "learning_rate": 1.354751212069313e-05, + "loss": 0.2075, + "step": 7700 + }, + { + "epoch": 3.22, + "learning_rate": 1.3545926928982241e-05, + "loss": 0.1893, + "step": 7701 + }, + { + "epoch": 3.22, + "learning_rate": 1.3544341635346827e-05, + "loss": 0.215, + "step": 7702 + }, + { + "epoch": 3.23, + "learning_rate": 1.3542756239832458e-05, + "loss": 0.1982, + "step": 7703 + }, + { + "epoch": 3.23, + "learning_rate": 1.3541170742484705e-05, + "loss": 0.2129, + "step": 7704 + }, + { + "epoch": 3.23, + "learning_rate": 1.3539585143349146e-05, + "loss": 0.1981, + "step": 7705 + }, + { + "epoch": 3.23, + "learning_rate": 1.3537999442471349e-05, + "loss": 0.1974, + "step": 7706 + }, + { + "epoch": 3.23, + "learning_rate": 1.3536413639896895e-05, + "loss": 0.176, + "step": 7707 + }, + { + "epoch": 3.23, + "learning_rate": 1.3534827735671373e-05, + "loss": 0.2061, + "step": 7708 + }, + { + "epoch": 3.23, + "learning_rate": 1.3533241729840362e-05, + "loss": 0.1642, + "step": 7709 + }, + { + "epoch": 3.23, + "learning_rate": 1.3531655622449454e-05, + "loss": 0.1684, + "step": 7710 + }, + { + "epoch": 3.23, + "learning_rate": 1.3530069413544241e-05, + "loss": 0.1948, + "step": 7711 + }, + { + "epoch": 3.23, + "learning_rate": 1.3528483103170312e-05, + "loss": 0.1634, + "step": 7712 + }, + { + "epoch": 3.23, + "learning_rate": 1.3526896691373269e-05, + "loss": 0.1526, + "step": 7713 + }, + { + "epoch": 3.23, + "learning_rate": 1.3525310178198707e-05, + "loss": 0.2019, + "step": 7714 + }, + { + "epoch": 3.23, + "learning_rate": 1.3523723563692237e-05, + "loss": 0.2137, + "step": 7715 + }, + { + "epoch": 3.23, + "learning_rate": 1.352213684789946e-05, + "loss": 0.1549, + "step": 7716 + }, + { + "epoch": 3.23, + "learning_rate": 1.3520550030865983e-05, + "loss": 0.1934, + "step": 7717 + }, + { + "epoch": 3.23, + "learning_rate": 1.3518963112637419e-05, + "loss": 0.1872, + "step": 7718 + }, + { + "epoch": 3.23, + "learning_rate": 1.3517376093259383e-05, + "loss": 0.2245, + "step": 7719 + }, + { + "epoch": 3.23, + "learning_rate": 1.3515788972777491e-05, + "loss": 0.1961, + "step": 7720 + }, + { + "epoch": 3.23, + "learning_rate": 1.3514201751237366e-05, + "loss": 0.1684, + "step": 7721 + }, + { + "epoch": 3.23, + "learning_rate": 1.3512614428684633e-05, + "loss": 0.1901, + "step": 7722 + }, + { + "epoch": 3.23, + "learning_rate": 1.3511027005164911e-05, + "loss": 0.1897, + "step": 7723 + }, + { + "epoch": 3.23, + "learning_rate": 1.3509439480723835e-05, + "loss": 0.1878, + "step": 7724 + }, + { + "epoch": 3.23, + "learning_rate": 1.3507851855407038e-05, + "loss": 0.1834, + "step": 7725 + }, + { + "epoch": 3.23, + "learning_rate": 1.350626412926015e-05, + "loss": 0.1637, + "step": 7726 + }, + { + "epoch": 3.24, + "learning_rate": 1.3504676302328813e-05, + "loss": 0.1433, + "step": 7727 + }, + { + "epoch": 3.24, + "learning_rate": 1.3503088374658661e-05, + "loss": 0.2066, + "step": 7728 + }, + { + "epoch": 3.24, + "learning_rate": 1.3501500346295347e-05, + "loss": 0.1597, + "step": 7729 + }, + { + "epoch": 3.24, + "learning_rate": 1.349991221728451e-05, + "loss": 0.201, + "step": 7730 + }, + { + "epoch": 3.24, + "learning_rate": 1.3498323987671801e-05, + "loss": 0.1925, + "step": 7731 + }, + { + "epoch": 3.24, + "learning_rate": 1.3496735657502877e-05, + "loss": 0.1942, + "step": 7732 + }, + { + "epoch": 3.24, + "learning_rate": 1.3495147226823386e-05, + "loss": 0.2095, + "step": 7733 + }, + { + "epoch": 3.24, + "learning_rate": 1.3493558695678992e-05, + "loss": 0.1603, + "step": 7734 + }, + { + "epoch": 3.24, + "learning_rate": 1.3491970064115351e-05, + "loss": 0.17, + "step": 7735 + }, + { + "epoch": 3.24, + "learning_rate": 1.3490381332178131e-05, + "loss": 0.1829, + "step": 7736 + }, + { + "epoch": 3.24, + "learning_rate": 1.3488792499912998e-05, + "loss": 0.2311, + "step": 7737 + }, + { + "epoch": 3.24, + "learning_rate": 1.348720356736562e-05, + "loss": 0.2073, + "step": 7738 + }, + { + "epoch": 3.24, + "learning_rate": 1.348561453458167e-05, + "loss": 0.1692, + "step": 7739 + }, + { + "epoch": 3.24, + "learning_rate": 1.3484025401606821e-05, + "loss": 0.1917, + "step": 7740 + }, + { + "epoch": 3.24, + "learning_rate": 1.3482436168486754e-05, + "loss": 0.1508, + "step": 7741 + }, + { + "epoch": 3.24, + "learning_rate": 1.3480846835267152e-05, + "loss": 0.1405, + "step": 7742 + }, + { + "epoch": 3.24, + "learning_rate": 1.3479257401993696e-05, + "loss": 0.2095, + "step": 7743 + }, + { + "epoch": 3.24, + "learning_rate": 1.347766786871207e-05, + "loss": 0.2075, + "step": 7744 + }, + { + "epoch": 3.24, + "learning_rate": 1.347607823546797e-05, + "loss": 0.1521, + "step": 7745 + }, + { + "epoch": 3.24, + "learning_rate": 1.3474488502307085e-05, + "loss": 0.1693, + "step": 7746 + }, + { + "epoch": 3.24, + "learning_rate": 1.3472898669275112e-05, + "loss": 0.1621, + "step": 7747 + }, + { + "epoch": 3.24, + "learning_rate": 1.347130873641775e-05, + "loss": 0.2106, + "step": 7748 + }, + { + "epoch": 3.24, + "learning_rate": 1.3469718703780694e-05, + "loss": 0.2237, + "step": 7749 + }, + { + "epoch": 3.24, + "learning_rate": 1.3468128571409658e-05, + "loss": 0.1928, + "step": 7750 + }, + { + "epoch": 3.25, + "learning_rate": 1.346653833935034e-05, + "loss": 0.2011, + "step": 7751 + }, + { + "epoch": 3.25, + "learning_rate": 1.3464948007648457e-05, + "loss": 0.1587, + "step": 7752 + }, + { + "epoch": 3.25, + "learning_rate": 1.3463357576349716e-05, + "loss": 0.1522, + "step": 7753 + }, + { + "epoch": 3.25, + "learning_rate": 1.3461767045499834e-05, + "loss": 0.2028, + "step": 7754 + }, + { + "epoch": 3.25, + "learning_rate": 1.346017641514453e-05, + "loss": 0.1807, + "step": 7755 + }, + { + "epoch": 3.25, + "learning_rate": 1.3458585685329529e-05, + "loss": 0.1705, + "step": 7756 + }, + { + "epoch": 3.25, + "learning_rate": 1.3456994856100547e-05, + "loss": 0.2111, + "step": 7757 + }, + { + "epoch": 3.25, + "learning_rate": 1.3455403927503317e-05, + "loss": 0.1834, + "step": 7758 + }, + { + "epoch": 3.25, + "learning_rate": 1.3453812899583572e-05, + "loss": 0.1738, + "step": 7759 + }, + { + "epoch": 3.25, + "learning_rate": 1.3452221772387036e-05, + "loss": 0.1727, + "step": 7760 + }, + { + "epoch": 3.25, + "learning_rate": 1.3450630545959448e-05, + "loss": 0.1906, + "step": 7761 + }, + { + "epoch": 3.25, + "learning_rate": 1.3449039220346547e-05, + "loss": 0.1901, + "step": 7762 + }, + { + "epoch": 3.25, + "learning_rate": 1.3447447795594078e-05, + "loss": 0.1671, + "step": 7763 + }, + { + "epoch": 3.25, + "learning_rate": 1.3445856271747777e-05, + "loss": 0.1833, + "step": 7764 + }, + { + "epoch": 3.25, + "learning_rate": 1.3444264648853395e-05, + "loss": 0.2022, + "step": 7765 + }, + { + "epoch": 3.25, + "learning_rate": 1.3442672926956685e-05, + "loss": 0.1815, + "step": 7766 + }, + { + "epoch": 3.25, + "learning_rate": 1.3441081106103395e-05, + "loss": 0.164, + "step": 7767 + }, + { + "epoch": 3.25, + "learning_rate": 1.3439489186339283e-05, + "loss": 0.208, + "step": 7768 + }, + { + "epoch": 3.25, + "learning_rate": 1.3437897167710105e-05, + "loss": 0.2039, + "step": 7769 + }, + { + "epoch": 3.25, + "learning_rate": 1.3436305050261625e-05, + "loss": 0.1856, + "step": 7770 + }, + { + "epoch": 3.25, + "learning_rate": 1.3434712834039606e-05, + "loss": 0.1896, + "step": 7771 + }, + { + "epoch": 3.25, + "learning_rate": 1.3433120519089813e-05, + "loss": 0.1631, + "step": 7772 + }, + { + "epoch": 3.25, + "learning_rate": 1.3431528105458021e-05, + "loss": 0.2288, + "step": 7773 + }, + { + "epoch": 3.25, + "learning_rate": 1.3429935593189996e-05, + "loss": 0.1845, + "step": 7774 + }, + { + "epoch": 3.26, + "learning_rate": 1.3428342982331515e-05, + "loss": 0.2092, + "step": 7775 + }, + { + "epoch": 3.26, + "learning_rate": 1.3426750272928359e-05, + "loss": 0.1466, + "step": 7776 + }, + { + "epoch": 3.26, + "learning_rate": 1.3425157465026305e-05, + "loss": 0.2149, + "step": 7777 + }, + { + "epoch": 3.26, + "learning_rate": 1.342356455867114e-05, + "loss": 0.2079, + "step": 7778 + }, + { + "epoch": 3.26, + "learning_rate": 1.342197155390865e-05, + "loss": 0.1664, + "step": 7779 + }, + { + "epoch": 3.26, + "learning_rate": 1.3420378450784623e-05, + "loss": 0.1936, + "step": 7780 + }, + { + "epoch": 3.26, + "learning_rate": 1.3418785249344855e-05, + "loss": 0.1982, + "step": 7781 + }, + { + "epoch": 3.26, + "learning_rate": 1.3417191949635137e-05, + "loss": 0.1644, + "step": 7782 + }, + { + "epoch": 3.26, + "learning_rate": 1.341559855170127e-05, + "loss": 0.1735, + "step": 7783 + }, + { + "epoch": 3.26, + "learning_rate": 1.3414005055589057e-05, + "loss": 0.1726, + "step": 7784 + }, + { + "epoch": 3.26, + "learning_rate": 1.3412411461344293e-05, + "loss": 0.1926, + "step": 7785 + }, + { + "epoch": 3.26, + "learning_rate": 1.3410817769012792e-05, + "loss": 0.1661, + "step": 7786 + }, + { + "epoch": 3.26, + "learning_rate": 1.340922397864036e-05, + "loss": 0.2143, + "step": 7787 + }, + { + "epoch": 3.26, + "learning_rate": 1.3407630090272808e-05, + "loss": 0.1829, + "step": 7788 + }, + { + "epoch": 3.26, + "learning_rate": 1.3406036103955957e-05, + "loss": 0.1304, + "step": 7789 + }, + { + "epoch": 3.26, + "learning_rate": 1.3404442019735617e-05, + "loss": 0.1993, + "step": 7790 + }, + { + "epoch": 3.26, + "learning_rate": 1.3402847837657614e-05, + "loss": 0.1636, + "step": 7791 + }, + { + "epoch": 3.26, + "learning_rate": 1.3401253557767768e-05, + "loss": 0.1769, + "step": 7792 + }, + { + "epoch": 3.26, + "learning_rate": 1.339965918011191e-05, + "loss": 0.1884, + "step": 7793 + }, + { + "epoch": 3.26, + "learning_rate": 1.3398064704735862e-05, + "loss": 0.1943, + "step": 7794 + }, + { + "epoch": 3.26, + "learning_rate": 1.3396470131685462e-05, + "loss": 0.2005, + "step": 7795 + }, + { + "epoch": 3.26, + "learning_rate": 1.3394875461006543e-05, + "loss": 0.2422, + "step": 7796 + }, + { + "epoch": 3.26, + "learning_rate": 1.3393280692744938e-05, + "loss": 0.1647, + "step": 7797 + }, + { + "epoch": 3.26, + "learning_rate": 1.3391685826946492e-05, + "loss": 0.2119, + "step": 7798 + }, + { + "epoch": 3.27, + "learning_rate": 1.3390090863657048e-05, + "loss": 0.1554, + "step": 7799 + }, + { + "epoch": 3.27, + "learning_rate": 1.338849580292245e-05, + "loss": 0.1796, + "step": 7800 + }, + { + "epoch": 3.27, + "learning_rate": 1.3386900644788546e-05, + "loss": 0.1523, + "step": 7801 + }, + { + "epoch": 3.27, + "learning_rate": 1.3385305389301187e-05, + "loss": 0.1896, + "step": 7802 + }, + { + "epoch": 3.27, + "learning_rate": 1.3383710036506232e-05, + "loss": 0.2046, + "step": 7803 + }, + { + "epoch": 3.27, + "learning_rate": 1.3382114586449533e-05, + "loss": 0.1643, + "step": 7804 + }, + { + "epoch": 3.27, + "learning_rate": 1.3380519039176951e-05, + "loss": 0.1594, + "step": 7805 + }, + { + "epoch": 3.27, + "learning_rate": 1.337892339473435e-05, + "loss": 0.1781, + "step": 7806 + }, + { + "epoch": 3.27, + "learning_rate": 1.3377327653167596e-05, + "loss": 0.1904, + "step": 7807 + }, + { + "epoch": 3.27, + "learning_rate": 1.3375731814522552e-05, + "loss": 0.1833, + "step": 7808 + }, + { + "epoch": 3.27, + "learning_rate": 1.3374135878845094e-05, + "loss": 0.1631, + "step": 7809 + }, + { + "epoch": 3.27, + "learning_rate": 1.3372539846181096e-05, + "loss": 0.2096, + "step": 7810 + }, + { + "epoch": 3.27, + "learning_rate": 1.337094371657643e-05, + "loss": 0.1732, + "step": 7811 + }, + { + "epoch": 3.27, + "learning_rate": 1.3369347490076981e-05, + "loss": 0.2059, + "step": 7812 + }, + { + "epoch": 3.27, + "learning_rate": 1.3367751166728626e-05, + "loss": 0.1556, + "step": 7813 + }, + { + "epoch": 3.27, + "learning_rate": 1.3366154746577254e-05, + "loss": 0.152, + "step": 7814 + }, + { + "epoch": 3.27, + "learning_rate": 1.3364558229668748e-05, + "loss": 0.1588, + "step": 7815 + }, + { + "epoch": 3.27, + "learning_rate": 1.3362961616049006e-05, + "loss": 0.1508, + "step": 7816 + }, + { + "epoch": 3.27, + "learning_rate": 1.3361364905763912e-05, + "loss": 0.2059, + "step": 7817 + }, + { + "epoch": 3.27, + "learning_rate": 1.335976809885937e-05, + "loss": 0.1803, + "step": 7818 + }, + { + "epoch": 3.27, + "learning_rate": 1.3358171195381272e-05, + "loss": 0.2117, + "step": 7819 + }, + { + "epoch": 3.27, + "learning_rate": 1.3356574195375526e-05, + "loss": 0.118, + "step": 7820 + }, + { + "epoch": 3.27, + "learning_rate": 1.335497709888803e-05, + "loss": 0.1968, + "step": 7821 + }, + { + "epoch": 3.27, + "learning_rate": 1.3353379905964698e-05, + "loss": 0.1957, + "step": 7822 + }, + { + "epoch": 3.28, + "learning_rate": 1.3351782616651436e-05, + "loss": 0.1985, + "step": 7823 + }, + { + "epoch": 3.28, + "learning_rate": 1.3350185230994157e-05, + "loss": 0.227, + "step": 7824 + }, + { + "epoch": 3.28, + "learning_rate": 1.3348587749038773e-05, + "loss": 0.2088, + "step": 7825 + }, + { + "epoch": 3.28, + "learning_rate": 1.334699017083121e-05, + "loss": 0.1898, + "step": 7826 + }, + { + "epoch": 3.28, + "learning_rate": 1.3345392496417381e-05, + "loss": 0.1721, + "step": 7827 + }, + { + "epoch": 3.28, + "learning_rate": 1.3343794725843215e-05, + "loss": 0.1599, + "step": 7828 + }, + { + "epoch": 3.28, + "learning_rate": 1.3342196859154639e-05, + "loss": 0.2064, + "step": 7829 + }, + { + "epoch": 3.28, + "learning_rate": 1.3340598896397579e-05, + "loss": 0.1717, + "step": 7830 + }, + { + "epoch": 3.28, + "learning_rate": 1.3339000837617967e-05, + "loss": 0.161, + "step": 7831 + }, + { + "epoch": 3.28, + "learning_rate": 1.3337402682861737e-05, + "loss": 0.1659, + "step": 7832 + }, + { + "epoch": 3.28, + "learning_rate": 1.333580443217483e-05, + "loss": 0.1689, + "step": 7833 + }, + { + "epoch": 3.28, + "learning_rate": 1.3334206085603186e-05, + "loss": 0.2025, + "step": 7834 + }, + { + "epoch": 3.28, + "learning_rate": 1.3332607643192745e-05, + "loss": 0.1741, + "step": 7835 + }, + { + "epoch": 3.28, + "learning_rate": 1.3331009104989456e-05, + "loss": 0.1395, + "step": 7836 + }, + { + "epoch": 3.28, + "learning_rate": 1.3329410471039266e-05, + "loss": 0.2303, + "step": 7837 + }, + { + "epoch": 3.28, + "learning_rate": 1.3327811741388126e-05, + "loss": 0.1997, + "step": 7838 + }, + { + "epoch": 3.28, + "learning_rate": 1.3326212916081991e-05, + "loss": 0.1643, + "step": 7839 + }, + { + "epoch": 3.28, + "learning_rate": 1.3324613995166818e-05, + "loss": 0.1636, + "step": 7840 + }, + { + "epoch": 3.28, + "learning_rate": 1.3323014978688567e-05, + "loss": 0.1608, + "step": 7841 + }, + { + "epoch": 3.28, + "learning_rate": 1.3321415866693195e-05, + "loss": 0.2052, + "step": 7842 + }, + { + "epoch": 3.28, + "learning_rate": 1.3319816659226672e-05, + "loss": 0.1693, + "step": 7843 + }, + { + "epoch": 3.28, + "learning_rate": 1.3318217356334967e-05, + "loss": 0.1523, + "step": 7844 + }, + { + "epoch": 3.28, + "learning_rate": 1.3316617958064049e-05, + "loss": 0.1823, + "step": 7845 + }, + { + "epoch": 3.28, + "learning_rate": 1.331501846445989e-05, + "loss": 0.2051, + "step": 7846 + }, + { + "epoch": 3.29, + "learning_rate": 1.3313418875568469e-05, + "loss": 0.197, + "step": 7847 + }, + { + "epoch": 3.29, + "learning_rate": 1.331181919143576e-05, + "loss": 0.1993, + "step": 7848 + }, + { + "epoch": 3.29, + "learning_rate": 1.3310219412107745e-05, + "loss": 0.1948, + "step": 7849 + }, + { + "epoch": 3.29, + "learning_rate": 1.3308619537630416e-05, + "loss": 0.1773, + "step": 7850 + }, + { + "epoch": 3.29, + "learning_rate": 1.3307019568049751e-05, + "loss": 0.1666, + "step": 7851 + }, + { + "epoch": 3.29, + "learning_rate": 1.3305419503411745e-05, + "loss": 0.1913, + "step": 7852 + }, + { + "epoch": 3.29, + "learning_rate": 1.3303819343762387e-05, + "loss": 0.2112, + "step": 7853 + }, + { + "epoch": 3.29, + "learning_rate": 1.3302219089147675e-05, + "loss": 0.1746, + "step": 7854 + }, + { + "epoch": 3.29, + "learning_rate": 1.3300618739613605e-05, + "loss": 0.1547, + "step": 7855 + }, + { + "epoch": 3.29, + "learning_rate": 1.3299018295206176e-05, + "loss": 0.1833, + "step": 7856 + }, + { + "epoch": 3.29, + "learning_rate": 1.3297417755971399e-05, + "loss": 0.1845, + "step": 7857 + }, + { + "epoch": 3.29, + "learning_rate": 1.329581712195527e-05, + "loss": 0.1692, + "step": 7858 + }, + { + "epoch": 3.29, + "learning_rate": 1.3294216393203803e-05, + "loss": 0.1781, + "step": 7859 + }, + { + "epoch": 3.29, + "learning_rate": 1.329261556976301e-05, + "loss": 0.1617, + "step": 7860 + }, + { + "epoch": 3.29, + "learning_rate": 1.3291014651678905e-05, + "loss": 0.215, + "step": 7861 + }, + { + "epoch": 3.29, + "learning_rate": 1.3289413638997502e-05, + "loss": 0.1892, + "step": 7862 + }, + { + "epoch": 3.29, + "learning_rate": 1.3287812531764826e-05, + "loss": 0.1876, + "step": 7863 + }, + { + "epoch": 3.29, + "learning_rate": 1.3286211330026894e-05, + "loss": 0.1248, + "step": 7864 + }, + { + "epoch": 3.29, + "learning_rate": 1.3284610033829734e-05, + "loss": 0.2129, + "step": 7865 + }, + { + "epoch": 3.29, + "learning_rate": 1.3283008643219373e-05, + "loss": 0.2075, + "step": 7866 + }, + { + "epoch": 3.29, + "learning_rate": 1.3281407158241842e-05, + "loss": 0.1871, + "step": 7867 + }, + { + "epoch": 3.29, + "learning_rate": 1.3279805578943174e-05, + "loss": 0.1847, + "step": 7868 + }, + { + "epoch": 3.29, + "learning_rate": 1.3278203905369407e-05, + "loss": 0.1944, + "step": 7869 + }, + { + "epoch": 3.29, + "learning_rate": 1.3276602137566575e-05, + "loss": 0.2002, + "step": 7870 + }, + { + "epoch": 3.3, + "learning_rate": 1.3275000275580723e-05, + "loss": 0.1795, + "step": 7871 + }, + { + "epoch": 3.3, + "learning_rate": 1.3273398319457895e-05, + "loss": 0.1858, + "step": 7872 + }, + { + "epoch": 3.3, + "learning_rate": 1.3271796269244135e-05, + "loss": 0.1892, + "step": 7873 + }, + { + "epoch": 3.3, + "learning_rate": 1.3270194124985499e-05, + "loss": 0.1832, + "step": 7874 + }, + { + "epoch": 3.3, + "learning_rate": 1.326859188672803e-05, + "loss": 0.1631, + "step": 7875 + }, + { + "epoch": 3.3, + "learning_rate": 1.3266989554517791e-05, + "loss": 0.2076, + "step": 7876 + }, + { + "epoch": 3.3, + "learning_rate": 1.3265387128400833e-05, + "loss": 0.2149, + "step": 7877 + }, + { + "epoch": 3.3, + "learning_rate": 1.3263784608423225e-05, + "loss": 0.186, + "step": 7878 + }, + { + "epoch": 3.3, + "learning_rate": 1.326218199463102e-05, + "loss": 0.1097, + "step": 7879 + }, + { + "epoch": 3.3, + "learning_rate": 1.3260579287070288e-05, + "loss": 0.1991, + "step": 7880 + }, + { + "epoch": 3.3, + "learning_rate": 1.3258976485787101e-05, + "loss": 0.1648, + "step": 7881 + }, + { + "epoch": 3.3, + "learning_rate": 1.3257373590827525e-05, + "loss": 0.1858, + "step": 7882 + }, + { + "epoch": 3.3, + "learning_rate": 1.3255770602237634e-05, + "loss": 0.2108, + "step": 7883 + }, + { + "epoch": 3.3, + "learning_rate": 1.325416752006351e-05, + "loss": 0.1963, + "step": 7884 + }, + { + "epoch": 3.3, + "learning_rate": 1.3252564344351227e-05, + "loss": 0.2124, + "step": 7885 + }, + { + "epoch": 3.3, + "learning_rate": 1.3250961075146868e-05, + "loss": 0.1754, + "step": 7886 + }, + { + "epoch": 3.3, + "learning_rate": 1.3249357712496518e-05, + "loss": 0.1954, + "step": 7887 + }, + { + "epoch": 3.3, + "learning_rate": 1.3247754256446263e-05, + "loss": 0.1774, + "step": 7888 + }, + { + "epoch": 3.3, + "learning_rate": 1.3246150707042196e-05, + "loss": 0.1806, + "step": 7889 + }, + { + "epoch": 3.3, + "learning_rate": 1.3244547064330404e-05, + "loss": 0.1706, + "step": 7890 + }, + { + "epoch": 3.3, + "learning_rate": 1.3242943328356988e-05, + "loss": 0.1864, + "step": 7891 + }, + { + "epoch": 3.3, + "learning_rate": 1.3241339499168042e-05, + "loss": 0.1964, + "step": 7892 + }, + { + "epoch": 3.3, + "learning_rate": 1.323973557680967e-05, + "loss": 0.1794, + "step": 7893 + }, + { + "epoch": 3.31, + "learning_rate": 1.3238131561327976e-05, + "loss": 0.162, + "step": 7894 + }, + { + "epoch": 3.31, + "learning_rate": 1.3236527452769061e-05, + "loss": 0.1988, + "step": 7895 + }, + { + "epoch": 3.31, + "learning_rate": 1.3234923251179036e-05, + "loss": 0.1611, + "step": 7896 + }, + { + "epoch": 3.31, + "learning_rate": 1.3233318956604013e-05, + "loss": 0.1859, + "step": 7897 + }, + { + "epoch": 3.31, + "learning_rate": 1.3231714569090108e-05, + "loss": 0.1789, + "step": 7898 + }, + { + "epoch": 3.31, + "learning_rate": 1.3230110088683433e-05, + "loss": 0.1955, + "step": 7899 + }, + { + "epoch": 3.31, + "learning_rate": 1.322850551543011e-05, + "loss": 0.2029, + "step": 7900 + }, + { + "epoch": 3.31, + "learning_rate": 1.3226900849376263e-05, + "loss": 0.1774, + "step": 7901 + }, + { + "epoch": 3.31, + "learning_rate": 1.3225296090568011e-05, + "loss": 0.204, + "step": 7902 + }, + { + "epoch": 3.31, + "learning_rate": 1.3223691239051489e-05, + "loss": 0.1998, + "step": 7903 + }, + { + "epoch": 3.31, + "learning_rate": 1.3222086294872818e-05, + "loss": 0.1742, + "step": 7904 + }, + { + "epoch": 3.31, + "learning_rate": 1.3220481258078142e-05, + "loss": 0.2086, + "step": 7905 + }, + { + "epoch": 3.31, + "learning_rate": 1.3218876128713584e-05, + "loss": 0.2133, + "step": 7906 + }, + { + "epoch": 3.31, + "learning_rate": 1.321727090682529e-05, + "loss": 0.2085, + "step": 7907 + }, + { + "epoch": 3.31, + "learning_rate": 1.3215665592459401e-05, + "loss": 0.1646, + "step": 7908 + }, + { + "epoch": 3.31, + "learning_rate": 1.3214060185662058e-05, + "loss": 0.1632, + "step": 7909 + }, + { + "epoch": 3.31, + "learning_rate": 1.3212454686479406e-05, + "loss": 0.1936, + "step": 7910 + }, + { + "epoch": 3.31, + "learning_rate": 1.3210849094957594e-05, + "loss": 0.1792, + "step": 7911 + }, + { + "epoch": 3.31, + "learning_rate": 1.3209243411142776e-05, + "loss": 0.1455, + "step": 7912 + }, + { + "epoch": 3.31, + "learning_rate": 1.32076376350811e-05, + "loss": 0.1829, + "step": 7913 + }, + { + "epoch": 3.31, + "learning_rate": 1.320603176681873e-05, + "loss": 0.2218, + "step": 7914 + }, + { + "epoch": 3.31, + "learning_rate": 1.3204425806401822e-05, + "loss": 0.1586, + "step": 7915 + }, + { + "epoch": 3.31, + "learning_rate": 1.3202819753876535e-05, + "loss": 0.204, + "step": 7916 + }, + { + "epoch": 3.31, + "learning_rate": 1.3201213609289036e-05, + "loss": 0.168, + "step": 7917 + }, + { + "epoch": 3.32, + "learning_rate": 1.3199607372685497e-05, + "loss": 0.1784, + "step": 7918 + }, + { + "epoch": 3.32, + "learning_rate": 1.319800104411208e-05, + "loss": 0.2257, + "step": 7919 + }, + { + "epoch": 3.32, + "learning_rate": 1.3196394623614963e-05, + "loss": 0.1622, + "step": 7920 + }, + { + "epoch": 3.32, + "learning_rate": 1.3194788111240317e-05, + "loss": 0.1386, + "step": 7921 + }, + { + "epoch": 3.32, + "learning_rate": 1.3193181507034323e-05, + "loss": 0.1977, + "step": 7922 + }, + { + "epoch": 3.32, + "learning_rate": 1.3191574811043159e-05, + "loss": 0.2258, + "step": 7923 + }, + { + "epoch": 3.32, + "learning_rate": 1.318996802331301e-05, + "loss": 0.2303, + "step": 7924 + }, + { + "epoch": 3.32, + "learning_rate": 1.318836114389006e-05, + "loss": 0.1982, + "step": 7925 + }, + { + "epoch": 3.32, + "learning_rate": 1.31867541728205e-05, + "loss": 0.1936, + "step": 7926 + }, + { + "epoch": 3.32, + "learning_rate": 1.3185147110150516e-05, + "loss": 0.1944, + "step": 7927 + }, + { + "epoch": 3.32, + "learning_rate": 1.3183539955926307e-05, + "loss": 0.191, + "step": 7928 + }, + { + "epoch": 3.32, + "learning_rate": 1.3181932710194067e-05, + "loss": 0.2047, + "step": 7929 + }, + { + "epoch": 3.32, + "learning_rate": 1.3180325372999994e-05, + "loss": 0.167, + "step": 7930 + }, + { + "epoch": 3.32, + "learning_rate": 1.3178717944390292e-05, + "loss": 0.2035, + "step": 7931 + }, + { + "epoch": 3.32, + "learning_rate": 1.3177110424411164e-05, + "loss": 0.2037, + "step": 7932 + }, + { + "epoch": 3.32, + "learning_rate": 1.3175502813108813e-05, + "loss": 0.2023, + "step": 7933 + }, + { + "epoch": 3.32, + "learning_rate": 1.3173895110529454e-05, + "loss": 0.1541, + "step": 7934 + }, + { + "epoch": 3.32, + "learning_rate": 1.3172287316719297e-05, + "loss": 0.1985, + "step": 7935 + }, + { + "epoch": 3.32, + "learning_rate": 1.3170679431724553e-05, + "loss": 0.1705, + "step": 7936 + }, + { + "epoch": 3.32, + "learning_rate": 1.3169071455591443e-05, + "loss": 0.2345, + "step": 7937 + }, + { + "epoch": 3.32, + "learning_rate": 1.3167463388366186e-05, + "loss": 0.2211, + "step": 7938 + }, + { + "epoch": 3.32, + "learning_rate": 1.3165855230095008e-05, + "loss": 0.2096, + "step": 7939 + }, + { + "epoch": 3.32, + "learning_rate": 1.3164246980824126e-05, + "loss": 0.1247, + "step": 7940 + }, + { + "epoch": 3.32, + "learning_rate": 1.3162638640599777e-05, + "loss": 0.1589, + "step": 7941 + }, + { + "epoch": 3.33, + "learning_rate": 1.3161030209468187e-05, + "loss": 0.1761, + "step": 7942 + }, + { + "epoch": 3.33, + "learning_rate": 1.3159421687475586e-05, + "loss": 0.1564, + "step": 7943 + }, + { + "epoch": 3.33, + "learning_rate": 1.3157813074668213e-05, + "loss": 0.1976, + "step": 7944 + }, + { + "epoch": 3.33, + "learning_rate": 1.3156204371092302e-05, + "loss": 0.1818, + "step": 7945 + }, + { + "epoch": 3.33, + "learning_rate": 1.3154595576794103e-05, + "loss": 0.2022, + "step": 7946 + }, + { + "epoch": 3.33, + "learning_rate": 1.315298669181985e-05, + "loss": 0.1456, + "step": 7947 + }, + { + "epoch": 3.33, + "learning_rate": 1.3151377716215791e-05, + "loss": 0.1878, + "step": 7948 + }, + { + "epoch": 3.33, + "learning_rate": 1.3149768650028181e-05, + "loss": 0.1428, + "step": 7949 + }, + { + "epoch": 3.33, + "learning_rate": 1.3148159493303262e-05, + "loss": 0.1644, + "step": 7950 + }, + { + "epoch": 3.33, + "learning_rate": 1.3146550246087293e-05, + "loss": 0.1733, + "step": 7951 + }, + { + "epoch": 3.33, + "learning_rate": 1.3144940908426532e-05, + "loss": 0.1824, + "step": 7952 + }, + { + "epoch": 3.33, + "learning_rate": 1.314333148036723e-05, + "loss": 0.1695, + "step": 7953 + }, + { + "epoch": 3.33, + "learning_rate": 1.3141721961955658e-05, + "loss": 0.1797, + "step": 7954 + }, + { + "epoch": 3.33, + "learning_rate": 1.3140112353238075e-05, + "loss": 0.1447, + "step": 7955 + }, + { + "epoch": 3.33, + "learning_rate": 1.3138502654260749e-05, + "loss": 0.1697, + "step": 7956 + }, + { + "epoch": 3.33, + "learning_rate": 1.313689286506995e-05, + "loss": 0.1789, + "step": 7957 + }, + { + "epoch": 3.33, + "learning_rate": 1.3135282985711948e-05, + "loss": 0.1959, + "step": 7958 + }, + { + "epoch": 3.33, + "learning_rate": 1.313367301623302e-05, + "loss": 0.1866, + "step": 7959 + }, + { + "epoch": 3.33, + "learning_rate": 1.3132062956679441e-05, + "loss": 0.1782, + "step": 7960 + }, + { + "epoch": 3.33, + "learning_rate": 1.3130452807097491e-05, + "loss": 0.195, + "step": 7961 + }, + { + "epoch": 3.33, + "learning_rate": 1.3128842567533456e-05, + "loss": 0.2005, + "step": 7962 + }, + { + "epoch": 3.33, + "learning_rate": 1.3127232238033613e-05, + "loss": 0.1895, + "step": 7963 + }, + { + "epoch": 3.33, + "learning_rate": 1.3125621818644258e-05, + "loss": 0.2153, + "step": 7964 + }, + { + "epoch": 3.33, + "learning_rate": 1.3124011309411678e-05, + "loss": 0.2004, + "step": 7965 + }, + { + "epoch": 3.34, + "learning_rate": 1.3122400710382164e-05, + "loss": 0.211, + "step": 7966 + }, + { + "epoch": 3.34, + "learning_rate": 1.3120790021602012e-05, + "loss": 0.1833, + "step": 7967 + }, + { + "epoch": 3.34, + "learning_rate": 1.311917924311752e-05, + "loss": 0.2227, + "step": 7968 + }, + { + "epoch": 3.34, + "learning_rate": 1.311756837497499e-05, + "loss": 0.2323, + "step": 7969 + }, + { + "epoch": 3.34, + "learning_rate": 1.3115957417220722e-05, + "loss": 0.1494, + "step": 7970 + }, + { + "epoch": 3.34, + "learning_rate": 1.3114346369901021e-05, + "loss": 0.1479, + "step": 7971 + }, + { + "epoch": 3.34, + "learning_rate": 1.3112735233062201e-05, + "loss": 0.1661, + "step": 7972 + }, + { + "epoch": 3.34, + "learning_rate": 1.311112400675057e-05, + "loss": 0.1722, + "step": 7973 + }, + { + "epoch": 3.34, + "learning_rate": 1.3109512691012434e-05, + "loss": 0.1949, + "step": 7974 + }, + { + "epoch": 3.34, + "learning_rate": 1.3107901285894118e-05, + "loss": 0.2101, + "step": 7975 + }, + { + "epoch": 3.34, + "learning_rate": 1.3106289791441941e-05, + "loss": 0.1923, + "step": 7976 + }, + { + "epoch": 3.34, + "learning_rate": 1.3104678207702218e-05, + "loss": 0.1802, + "step": 7977 + }, + { + "epoch": 3.34, + "learning_rate": 1.3103066534721273e-05, + "loss": 0.2054, + "step": 7978 + }, + { + "epoch": 3.34, + "learning_rate": 1.3101454772545435e-05, + "loss": 0.2031, + "step": 7979 + }, + { + "epoch": 3.34, + "learning_rate": 1.3099842921221034e-05, + "loss": 0.2368, + "step": 7980 + }, + { + "epoch": 3.34, + "learning_rate": 1.3098230980794394e-05, + "loss": 0.1559, + "step": 7981 + }, + { + "epoch": 3.34, + "learning_rate": 1.309661895131186e-05, + "loss": 0.1767, + "step": 7982 + }, + { + "epoch": 3.34, + "learning_rate": 1.3095006832819759e-05, + "loss": 0.1372, + "step": 7983 + }, + { + "epoch": 3.34, + "learning_rate": 1.3093394625364433e-05, + "loss": 0.1748, + "step": 7984 + }, + { + "epoch": 3.34, + "learning_rate": 1.3091782328992225e-05, + "loss": 0.1735, + "step": 7985 + }, + { + "epoch": 3.34, + "learning_rate": 1.3090169943749475e-05, + "loss": 0.1624, + "step": 7986 + }, + { + "epoch": 3.34, + "learning_rate": 1.3088557469682535e-05, + "loss": 0.1809, + "step": 7987 + }, + { + "epoch": 3.34, + "learning_rate": 1.3086944906837751e-05, + "loss": 0.2157, + "step": 7988 + }, + { + "epoch": 3.34, + "learning_rate": 1.3085332255261472e-05, + "loss": 0.1759, + "step": 7989 + }, + { + "epoch": 3.35, + "learning_rate": 1.3083719515000057e-05, + "loss": 0.1399, + "step": 7990 + }, + { + "epoch": 3.35, + "learning_rate": 1.308210668609986e-05, + "loss": 0.1942, + "step": 7991 + }, + { + "epoch": 3.35, + "learning_rate": 1.3080493768607243e-05, + "loss": 0.1867, + "step": 7992 + }, + { + "epoch": 3.35, + "learning_rate": 1.3078880762568566e-05, + "loss": 0.2199, + "step": 7993 + }, + { + "epoch": 3.35, + "learning_rate": 1.3077267668030194e-05, + "loss": 0.2082, + "step": 7994 + }, + { + "epoch": 3.35, + "learning_rate": 1.3075654485038491e-05, + "loss": 0.166, + "step": 7995 + }, + { + "epoch": 3.35, + "learning_rate": 1.3074041213639832e-05, + "loss": 0.1959, + "step": 7996 + }, + { + "epoch": 3.35, + "learning_rate": 1.3072427853880583e-05, + "loss": 0.1902, + "step": 7997 + }, + { + "epoch": 3.35, + "learning_rate": 1.3070814405807124e-05, + "loss": 0.1851, + "step": 7998 + }, + { + "epoch": 3.35, + "learning_rate": 1.3069200869465832e-05, + "loss": 0.1786, + "step": 7999 + }, + { + "epoch": 3.35, + "learning_rate": 1.3067587244903081e-05, + "loss": 0.1853, + "step": 8000 + }, + { + "epoch": 3.35, + "learning_rate": 1.3065973532165259e-05, + "loss": 0.1811, + "step": 8001 + }, + { + "epoch": 3.35, + "learning_rate": 1.3064359731298746e-05, + "loss": 0.2069, + "step": 8002 + }, + { + "epoch": 3.35, + "learning_rate": 1.3062745842349935e-05, + "loss": 0.2145, + "step": 8003 + }, + { + "epoch": 3.35, + "learning_rate": 1.3061131865365211e-05, + "loss": 0.2082, + "step": 8004 + }, + { + "epoch": 3.35, + "learning_rate": 1.3059517800390967e-05, + "loss": 0.1653, + "step": 8005 + }, + { + "epoch": 3.35, + "learning_rate": 1.30579036474736e-05, + "loss": 0.1863, + "step": 8006 + }, + { + "epoch": 3.35, + "learning_rate": 1.3056289406659506e-05, + "loss": 0.172, + "step": 8007 + }, + { + "epoch": 3.35, + "learning_rate": 1.3054675077995082e-05, + "loss": 0.1559, + "step": 8008 + }, + { + "epoch": 3.35, + "learning_rate": 1.3053060661526738e-05, + "loss": 0.202, + "step": 8009 + }, + { + "epoch": 3.35, + "learning_rate": 1.3051446157300875e-05, + "loss": 0.1904, + "step": 8010 + }, + { + "epoch": 3.35, + "learning_rate": 1.3049831565363897e-05, + "loss": 0.2119, + "step": 8011 + }, + { + "epoch": 3.35, + "learning_rate": 1.3048216885762218e-05, + "loss": 0.2065, + "step": 8012 + }, + { + "epoch": 3.35, + "learning_rate": 1.3046602118542251e-05, + "loss": 0.2191, + "step": 8013 + }, + { + "epoch": 3.36, + "learning_rate": 1.3044987263750409e-05, + "loss": 0.1768, + "step": 8014 + }, + { + "epoch": 3.36, + "learning_rate": 1.3043372321433108e-05, + "loss": 0.1888, + "step": 8015 + }, + { + "epoch": 3.36, + "learning_rate": 1.304175729163677e-05, + "loss": 0.1931, + "step": 8016 + }, + { + "epoch": 3.36, + "learning_rate": 1.304014217440782e-05, + "loss": 0.1912, + "step": 8017 + }, + { + "epoch": 3.36, + "learning_rate": 1.303852696979268e-05, + "loss": 0.1944, + "step": 8018 + }, + { + "epoch": 3.36, + "learning_rate": 1.3036911677837778e-05, + "loss": 0.1651, + "step": 8019 + }, + { + "epoch": 3.36, + "learning_rate": 1.3035296298589549e-05, + "loss": 0.2085, + "step": 8020 + }, + { + "epoch": 3.36, + "learning_rate": 1.3033680832094416e-05, + "loss": 0.2003, + "step": 8021 + }, + { + "epoch": 3.36, + "learning_rate": 1.3032065278398823e-05, + "loss": 0.1749, + "step": 8022 + }, + { + "epoch": 3.36, + "learning_rate": 1.3030449637549203e-05, + "loss": 0.1729, + "step": 8023 + }, + { + "epoch": 3.36, + "learning_rate": 1.3028833909591999e-05, + "loss": 0.2078, + "step": 8024 + }, + { + "epoch": 3.36, + "learning_rate": 1.3027218094573648e-05, + "loss": 0.1727, + "step": 8025 + }, + { + "epoch": 3.36, + "learning_rate": 1.30256021925406e-05, + "loss": 0.1386, + "step": 8026 + }, + { + "epoch": 3.36, + "learning_rate": 1.3023986203539304e-05, + "loss": 0.1952, + "step": 8027 + }, + { + "epoch": 3.36, + "learning_rate": 1.3022370127616209e-05, + "loss": 0.1911, + "step": 8028 + }, + { + "epoch": 3.36, + "learning_rate": 1.3020753964817763e-05, + "loss": 0.1608, + "step": 8029 + }, + { + "epoch": 3.36, + "learning_rate": 1.301913771519043e-05, + "loss": 0.225, + "step": 8030 + }, + { + "epoch": 3.36, + "learning_rate": 1.3017521378780655e-05, + "loss": 0.2167, + "step": 8031 + }, + { + "epoch": 3.36, + "learning_rate": 1.301590495563491e-05, + "loss": 0.1651, + "step": 8032 + }, + { + "epoch": 3.36, + "learning_rate": 1.3014288445799656e-05, + "loss": 0.1953, + "step": 8033 + }, + { + "epoch": 3.36, + "learning_rate": 1.3012671849321351e-05, + "loss": 0.2044, + "step": 8034 + }, + { + "epoch": 3.36, + "learning_rate": 1.3011055166246469e-05, + "loss": 0.2039, + "step": 8035 + }, + { + "epoch": 3.36, + "learning_rate": 1.3009438396621479e-05, + "loss": 0.1804, + "step": 8036 + }, + { + "epoch": 3.36, + "learning_rate": 1.3007821540492854e-05, + "loss": 0.1906, + "step": 8037 + }, + { + "epoch": 3.37, + "learning_rate": 1.3006204597907064e-05, + "loss": 0.2217, + "step": 8038 + }, + { + "epoch": 3.37, + "learning_rate": 1.3004587568910592e-05, + "loss": 0.2237, + "step": 8039 + }, + { + "epoch": 3.37, + "learning_rate": 1.3002970453549918e-05, + "loss": 0.1782, + "step": 8040 + }, + { + "epoch": 3.37, + "learning_rate": 1.300135325187152e-05, + "loss": 0.211, + "step": 8041 + }, + { + "epoch": 3.37, + "learning_rate": 1.2999735963921887e-05, + "loss": 0.184, + "step": 8042 + }, + { + "epoch": 3.37, + "learning_rate": 1.2998118589747509e-05, + "loss": 0.1948, + "step": 8043 + }, + { + "epoch": 3.37, + "learning_rate": 1.2996501129394872e-05, + "loss": 0.1965, + "step": 8044 + }, + { + "epoch": 3.37, + "learning_rate": 1.2994883582910465e-05, + "loss": 0.1551, + "step": 8045 + }, + { + "epoch": 3.37, + "learning_rate": 1.299326595034079e-05, + "loss": 0.1572, + "step": 8046 + }, + { + "epoch": 3.37, + "learning_rate": 1.2991648231732342e-05, + "loss": 0.1887, + "step": 8047 + }, + { + "epoch": 3.37, + "learning_rate": 1.2990030427131618e-05, + "loss": 0.1751, + "step": 8048 + }, + { + "epoch": 3.37, + "learning_rate": 1.2988412536585125e-05, + "loss": 0.2004, + "step": 8049 + }, + { + "epoch": 3.37, + "learning_rate": 1.2986794560139362e-05, + "loss": 0.2151, + "step": 8050 + }, + { + "epoch": 3.37, + "learning_rate": 1.2985176497840843e-05, + "loss": 0.2198, + "step": 8051 + }, + { + "epoch": 3.37, + "learning_rate": 1.2983558349736074e-05, + "loss": 0.1641, + "step": 8052 + }, + { + "epoch": 3.37, + "learning_rate": 1.2981940115871566e-05, + "loss": 0.176, + "step": 8053 + }, + { + "epoch": 3.37, + "learning_rate": 1.2980321796293838e-05, + "loss": 0.1961, + "step": 8054 + }, + { + "epoch": 3.37, + "learning_rate": 1.2978703391049402e-05, + "loss": 0.1903, + "step": 8055 + }, + { + "epoch": 3.37, + "learning_rate": 1.2977084900184784e-05, + "loss": 0.1761, + "step": 8056 + }, + { + "epoch": 3.37, + "learning_rate": 1.2975466323746499e-05, + "loss": 0.1296, + "step": 8057 + }, + { + "epoch": 3.37, + "learning_rate": 1.2973847661781078e-05, + "loss": 0.1581, + "step": 8058 + }, + { + "epoch": 3.37, + "learning_rate": 1.2972228914335044e-05, + "loss": 0.1424, + "step": 8059 + }, + { + "epoch": 3.37, + "learning_rate": 1.2970610081454925e-05, + "loss": 0.1769, + "step": 8060 + }, + { + "epoch": 3.37, + "learning_rate": 1.2968991163187256e-05, + "loss": 0.187, + "step": 8061 + }, + { + "epoch": 3.38, + "learning_rate": 1.2967372159578574e-05, + "loss": 0.1795, + "step": 8062 + }, + { + "epoch": 3.38, + "learning_rate": 1.2965753070675406e-05, + "loss": 0.1714, + "step": 8063 + }, + { + "epoch": 3.38, + "learning_rate": 1.2964133896524305e-05, + "loss": 0.1511, + "step": 8064 + }, + { + "epoch": 3.38, + "learning_rate": 1.2962514637171802e-05, + "loss": 0.1504, + "step": 8065 + }, + { + "epoch": 3.38, + "learning_rate": 1.2960895292664443e-05, + "loss": 0.1826, + "step": 8066 + }, + { + "epoch": 3.38, + "learning_rate": 1.295927586304878e-05, + "loss": 0.1782, + "step": 8067 + }, + { + "epoch": 3.38, + "learning_rate": 1.2957656348371357e-05, + "loss": 0.2208, + "step": 8068 + }, + { + "epoch": 3.38, + "learning_rate": 1.2956036748678727e-05, + "loss": 0.1828, + "step": 8069 + }, + { + "epoch": 3.38, + "learning_rate": 1.2954417064017441e-05, + "loss": 0.1838, + "step": 8070 + }, + { + "epoch": 3.38, + "learning_rate": 1.295279729443406e-05, + "loss": 0.1932, + "step": 8071 + }, + { + "epoch": 3.38, + "learning_rate": 1.2951177439975141e-05, + "loss": 0.2389, + "step": 8072 + }, + { + "epoch": 3.38, + "learning_rate": 1.2949557500687244e-05, + "loss": 0.1934, + "step": 8073 + }, + { + "epoch": 3.38, + "learning_rate": 1.2947937476616938e-05, + "loss": 0.1962, + "step": 8074 + }, + { + "epoch": 3.38, + "learning_rate": 1.2946317367810784e-05, + "loss": 0.2111, + "step": 8075 + }, + { + "epoch": 3.38, + "learning_rate": 1.2944697174315349e-05, + "loss": 0.1932, + "step": 8076 + }, + { + "epoch": 3.38, + "learning_rate": 1.2943076896177208e-05, + "loss": 0.1867, + "step": 8077 + }, + { + "epoch": 3.38, + "learning_rate": 1.2941456533442935e-05, + "loss": 0.1605, + "step": 8078 + }, + { + "epoch": 3.38, + "learning_rate": 1.2939836086159102e-05, + "loss": 0.2104, + "step": 8079 + }, + { + "epoch": 3.38, + "learning_rate": 1.2938215554372292e-05, + "loss": 0.1872, + "step": 8080 + }, + { + "epoch": 3.38, + "learning_rate": 1.2936594938129082e-05, + "loss": 0.1978, + "step": 8081 + }, + { + "epoch": 3.38, + "learning_rate": 1.2934974237476054e-05, + "loss": 0.2306, + "step": 8082 + }, + { + "epoch": 3.38, + "learning_rate": 1.2933353452459796e-05, + "loss": 0.1974, + "step": 8083 + }, + { + "epoch": 3.38, + "learning_rate": 1.29317325831269e-05, + "loss": 0.1785, + "step": 8084 + }, + { + "epoch": 3.38, + "learning_rate": 1.293011162952395e-05, + "loss": 0.1864, + "step": 8085 + }, + { + "epoch": 3.39, + "learning_rate": 1.292849059169754e-05, + "loss": 0.2104, + "step": 8086 + }, + { + "epoch": 3.39, + "learning_rate": 1.2926869469694267e-05, + "loss": 0.2014, + "step": 8087 + }, + { + "epoch": 3.39, + "learning_rate": 1.2925248263560733e-05, + "loss": 0.2069, + "step": 8088 + }, + { + "epoch": 3.39, + "learning_rate": 1.2923626973343527e-05, + "loss": 0.1916, + "step": 8089 + }, + { + "epoch": 3.39, + "learning_rate": 1.2922005599089263e-05, + "loss": 0.1521, + "step": 8090 + }, + { + "epoch": 3.39, + "learning_rate": 1.2920384140844539e-05, + "loss": 0.2251, + "step": 8091 + }, + { + "epoch": 3.39, + "learning_rate": 1.2918762598655965e-05, + "loss": 0.2019, + "step": 8092 + }, + { + "epoch": 3.39, + "learning_rate": 1.291714097257015e-05, + "loss": 0.1826, + "step": 8093 + }, + { + "epoch": 3.39, + "learning_rate": 1.2915519262633704e-05, + "loss": 0.2038, + "step": 8094 + }, + { + "epoch": 3.39, + "learning_rate": 1.2913897468893249e-05, + "loss": 0.187, + "step": 8095 + }, + { + "epoch": 3.39, + "learning_rate": 1.2912275591395393e-05, + "loss": 0.2268, + "step": 8096 + }, + { + "epoch": 3.39, + "learning_rate": 1.291065363018676e-05, + "loss": 0.2099, + "step": 8097 + }, + { + "epoch": 3.39, + "learning_rate": 1.290903158531397e-05, + "loss": 0.1623, + "step": 8098 + }, + { + "epoch": 3.39, + "learning_rate": 1.2907409456823651e-05, + "loss": 0.2001, + "step": 8099 + }, + { + "epoch": 3.39, + "learning_rate": 1.2905787244762427e-05, + "loss": 0.1717, + "step": 8100 + }, + { + "epoch": 3.39, + "learning_rate": 1.2904164949176927e-05, + "loss": 0.1823, + "step": 8101 + }, + { + "epoch": 3.39, + "learning_rate": 1.2902542570113786e-05, + "loss": 0.2004, + "step": 8102 + }, + { + "epoch": 3.39, + "learning_rate": 1.2900920107619627e-05, + "loss": 0.1961, + "step": 8103 + }, + { + "epoch": 3.39, + "learning_rate": 1.2899297561741097e-05, + "loss": 0.2076, + "step": 8104 + }, + { + "epoch": 3.39, + "learning_rate": 1.2897674932524832e-05, + "loss": 0.1968, + "step": 8105 + }, + { + "epoch": 3.39, + "learning_rate": 1.2896052220017474e-05, + "loss": 0.1579, + "step": 8106 + }, + { + "epoch": 3.39, + "learning_rate": 1.2894429424265661e-05, + "loss": 0.1819, + "step": 8107 + }, + { + "epoch": 3.39, + "learning_rate": 1.2892806545316048e-05, + "loss": 0.191, + "step": 8108 + }, + { + "epoch": 3.4, + "learning_rate": 1.2891183583215272e-05, + "loss": 0.1599, + "step": 8109 + }, + { + "epoch": 3.4, + "learning_rate": 1.2889560538009993e-05, + "loss": 0.164, + "step": 8110 + }, + { + "epoch": 3.4, + "learning_rate": 1.288793740974686e-05, + "loss": 0.1802, + "step": 8111 + }, + { + "epoch": 3.4, + "learning_rate": 1.288631419847253e-05, + "loss": 0.174, + "step": 8112 + }, + { + "epoch": 3.4, + "learning_rate": 1.2884690904233658e-05, + "loss": 0.2084, + "step": 8113 + }, + { + "epoch": 3.4, + "learning_rate": 1.2883067527076904e-05, + "loss": 0.1475, + "step": 8114 + }, + { + "epoch": 3.4, + "learning_rate": 1.2881444067048936e-05, + "loss": 0.1961, + "step": 8115 + }, + { + "epoch": 3.4, + "learning_rate": 1.2879820524196414e-05, + "loss": 0.1655, + "step": 8116 + }, + { + "epoch": 3.4, + "learning_rate": 1.2878196898566004e-05, + "loss": 0.1942, + "step": 8117 + }, + { + "epoch": 3.4, + "learning_rate": 1.2876573190204381e-05, + "loss": 0.1498, + "step": 8118 + }, + { + "epoch": 3.4, + "learning_rate": 1.2874949399158213e-05, + "loss": 0.2078, + "step": 8119 + }, + { + "epoch": 3.4, + "learning_rate": 1.2873325525474176e-05, + "loss": 0.2005, + "step": 8120 + }, + { + "epoch": 3.4, + "learning_rate": 1.2871701569198946e-05, + "loss": 0.1994, + "step": 8121 + }, + { + "epoch": 3.4, + "learning_rate": 1.2870077530379205e-05, + "loss": 0.1859, + "step": 8122 + }, + { + "epoch": 3.4, + "learning_rate": 1.286845340906163e-05, + "loss": 0.1879, + "step": 8123 + }, + { + "epoch": 3.4, + "learning_rate": 1.286682920529291e-05, + "loss": 0.1767, + "step": 8124 + }, + { + "epoch": 3.4, + "learning_rate": 1.2865204919119726e-05, + "loss": 0.203, + "step": 8125 + }, + { + "epoch": 3.4, + "learning_rate": 1.286358055058877e-05, + "loss": 0.2012, + "step": 8126 + }, + { + "epoch": 3.4, + "learning_rate": 1.2861956099746731e-05, + "loss": 0.1452, + "step": 8127 + }, + { + "epoch": 3.4, + "learning_rate": 1.2860331566640303e-05, + "loss": 0.1434, + "step": 8128 + }, + { + "epoch": 3.4, + "learning_rate": 1.2858706951316184e-05, + "loss": 0.1894, + "step": 8129 + }, + { + "epoch": 3.4, + "learning_rate": 1.285708225382107e-05, + "loss": 0.2196, + "step": 8130 + }, + { + "epoch": 3.4, + "learning_rate": 1.2855457474201663e-05, + "loss": 0.1776, + "step": 8131 + }, + { + "epoch": 3.4, + "learning_rate": 1.2853832612504662e-05, + "loss": 0.1807, + "step": 8132 + }, + { + "epoch": 3.41, + "learning_rate": 1.2852207668776777e-05, + "loss": 0.1557, + "step": 8133 + }, + { + "epoch": 3.41, + "learning_rate": 1.2850582643064713e-05, + "loss": 0.1438, + "step": 8134 + }, + { + "epoch": 3.41, + "learning_rate": 1.2848957535415183e-05, + "loss": 0.179, + "step": 8135 + }, + { + "epoch": 3.41, + "learning_rate": 1.2847332345874895e-05, + "loss": 0.1775, + "step": 8136 + }, + { + "epoch": 3.41, + "learning_rate": 1.2845707074490563e-05, + "loss": 0.1993, + "step": 8137 + }, + { + "epoch": 3.41, + "learning_rate": 1.284408172130891e-05, + "loss": 0.173, + "step": 8138 + }, + { + "epoch": 3.41, + "learning_rate": 1.284245628637665e-05, + "loss": 0.1968, + "step": 8139 + }, + { + "epoch": 3.41, + "learning_rate": 1.2840830769740506e-05, + "loss": 0.1977, + "step": 8140 + }, + { + "epoch": 3.41, + "learning_rate": 1.2839205171447203e-05, + "loss": 0.1816, + "step": 8141 + }, + { + "epoch": 3.41, + "learning_rate": 1.2837579491543468e-05, + "loss": 0.186, + "step": 8142 + }, + { + "epoch": 3.41, + "learning_rate": 1.2835953730076026e-05, + "loss": 0.2, + "step": 8143 + }, + { + "epoch": 3.41, + "learning_rate": 1.2834327887091611e-05, + "loss": 0.2119, + "step": 8144 + }, + { + "epoch": 3.41, + "learning_rate": 1.2832701962636957e-05, + "loss": 0.1799, + "step": 8145 + }, + { + "epoch": 3.41, + "learning_rate": 1.2831075956758801e-05, + "loss": 0.1658, + "step": 8146 + }, + { + "epoch": 3.41, + "learning_rate": 1.2829449869503876e-05, + "loss": 0.1929, + "step": 8147 + }, + { + "epoch": 3.41, + "learning_rate": 1.2827823700918924e-05, + "loss": 0.203, + "step": 8148 + }, + { + "epoch": 3.41, + "learning_rate": 1.2826197451050692e-05, + "loss": 0.1814, + "step": 8149 + }, + { + "epoch": 3.41, + "learning_rate": 1.2824571119945921e-05, + "loss": 0.1706, + "step": 8150 + }, + { + "epoch": 3.41, + "learning_rate": 1.282294470765136e-05, + "loss": 0.1477, + "step": 8151 + }, + { + "epoch": 3.41, + "learning_rate": 1.2821318214213758e-05, + "loss": 0.2028, + "step": 8152 + }, + { + "epoch": 3.41, + "learning_rate": 1.2819691639679867e-05, + "loss": 0.1899, + "step": 8153 + }, + { + "epoch": 3.41, + "learning_rate": 1.2818064984096443e-05, + "loss": 0.2138, + "step": 8154 + }, + { + "epoch": 3.41, + "learning_rate": 1.2816438247510243e-05, + "loss": 0.1841, + "step": 8155 + }, + { + "epoch": 3.41, + "learning_rate": 1.2814811429968022e-05, + "loss": 0.1999, + "step": 8156 + }, + { + "epoch": 3.42, + "learning_rate": 1.2813184531516545e-05, + "loss": 0.1649, + "step": 8157 + }, + { + "epoch": 3.42, + "learning_rate": 1.2811557552202576e-05, + "loss": 0.185, + "step": 8158 + }, + { + "epoch": 3.42, + "learning_rate": 1.2809930492072882e-05, + "loss": 0.1613, + "step": 8159 + }, + { + "epoch": 3.42, + "learning_rate": 1.2808303351174224e-05, + "loss": 0.189, + "step": 8160 + }, + { + "epoch": 3.42, + "learning_rate": 1.2806676129553381e-05, + "loss": 0.2033, + "step": 8161 + }, + { + "epoch": 3.42, + "learning_rate": 1.2805048827257123e-05, + "loss": 0.2193, + "step": 8162 + }, + { + "epoch": 3.42, + "learning_rate": 1.280342144433223e-05, + "loss": 0.2024, + "step": 8163 + }, + { + "epoch": 3.42, + "learning_rate": 1.280179398082547e-05, + "loss": 0.1882, + "step": 8164 + }, + { + "epoch": 3.42, + "learning_rate": 1.2800166436783627e-05, + "loss": 0.1955, + "step": 8165 + }, + { + "epoch": 3.42, + "learning_rate": 1.279853881225349e-05, + "loss": 0.1965, + "step": 8166 + }, + { + "epoch": 3.42, + "learning_rate": 1.2796911107281833e-05, + "loss": 0.1778, + "step": 8167 + }, + { + "epoch": 3.42, + "learning_rate": 1.279528332191545e-05, + "loss": 0.2018, + "step": 8168 + }, + { + "epoch": 3.42, + "learning_rate": 1.279365545620113e-05, + "loss": 0.1709, + "step": 8169 + }, + { + "epoch": 3.42, + "learning_rate": 1.2792027510185665e-05, + "loss": 0.186, + "step": 8170 + }, + { + "epoch": 3.42, + "learning_rate": 1.2790399483915843e-05, + "loss": 0.1931, + "step": 8171 + }, + { + "epoch": 3.42, + "learning_rate": 1.2788771377438465e-05, + "loss": 0.17, + "step": 8172 + }, + { + "epoch": 3.42, + "learning_rate": 1.278714319080033e-05, + "loss": 0.1962, + "step": 8173 + }, + { + "epoch": 3.42, + "learning_rate": 1.2785514924048235e-05, + "loss": 0.1863, + "step": 8174 + }, + { + "epoch": 3.42, + "learning_rate": 1.2783886577228986e-05, + "loss": 0.1965, + "step": 8175 + }, + { + "epoch": 3.42, + "learning_rate": 1.2782258150389387e-05, + "loss": 0.1853, + "step": 8176 + }, + { + "epoch": 3.42, + "learning_rate": 1.2780629643576249e-05, + "loss": 0.2044, + "step": 8177 + }, + { + "epoch": 3.42, + "learning_rate": 1.2779001056836373e-05, + "loss": 0.1349, + "step": 8178 + }, + { + "epoch": 3.42, + "learning_rate": 1.2777372390216585e-05, + "loss": 0.1896, + "step": 8179 + }, + { + "epoch": 3.42, + "learning_rate": 1.2775743643763691e-05, + "loss": 0.1824, + "step": 8180 + }, + { + "epoch": 3.43, + "learning_rate": 1.2774114817524505e-05, + "loss": 0.2179, + "step": 8181 + }, + { + "epoch": 3.43, + "learning_rate": 1.2772485911545854e-05, + "loss": 0.1746, + "step": 8182 + }, + { + "epoch": 3.43, + "learning_rate": 1.2770856925874555e-05, + "loss": 0.1649, + "step": 8183 + }, + { + "epoch": 3.43, + "learning_rate": 1.2769227860557431e-05, + "loss": 0.1964, + "step": 8184 + }, + { + "epoch": 3.43, + "learning_rate": 1.2767598715641309e-05, + "loss": 0.1795, + "step": 8185 + }, + { + "epoch": 3.43, + "learning_rate": 1.2765969491173023e-05, + "loss": 0.1827, + "step": 8186 + }, + { + "epoch": 3.43, + "learning_rate": 1.2764340187199393e-05, + "loss": 0.1434, + "step": 8187 + }, + { + "epoch": 3.43, + "learning_rate": 1.2762710803767258e-05, + "loss": 0.2034, + "step": 8188 + }, + { + "epoch": 3.43, + "learning_rate": 1.2761081340923455e-05, + "loss": 0.1645, + "step": 8189 + }, + { + "epoch": 3.43, + "learning_rate": 1.2759451798714816e-05, + "loss": 0.1894, + "step": 8190 + }, + { + "epoch": 3.43, + "learning_rate": 1.2757822177188187e-05, + "loss": 0.1838, + "step": 8191 + }, + { + "epoch": 3.43, + "learning_rate": 1.2756192476390406e-05, + "loss": 0.2094, + "step": 8192 + }, + { + "epoch": 3.43, + "learning_rate": 1.2754562696368318e-05, + "loss": 0.2115, + "step": 8193 + }, + { + "epoch": 3.43, + "learning_rate": 1.2752932837168768e-05, + "loss": 0.1835, + "step": 8194 + }, + { + "epoch": 3.43, + "learning_rate": 1.2751302898838607e-05, + "loss": 0.1596, + "step": 8195 + }, + { + "epoch": 3.43, + "learning_rate": 1.2749672881424687e-05, + "loss": 0.1958, + "step": 8196 + }, + { + "epoch": 3.43, + "learning_rate": 1.2748042784973857e-05, + "loss": 0.1959, + "step": 8197 + }, + { + "epoch": 3.43, + "learning_rate": 1.274641260953298e-05, + "loss": 0.2248, + "step": 8198 + }, + { + "epoch": 3.43, + "learning_rate": 1.2744782355148905e-05, + "loss": 0.2259, + "step": 8199 + }, + { + "epoch": 3.43, + "learning_rate": 1.27431520218685e-05, + "loss": 0.2132, + "step": 8200 + }, + { + "epoch": 3.43, + "learning_rate": 1.2741521609738623e-05, + "loss": 0.195, + "step": 8201 + }, + { + "epoch": 3.43, + "learning_rate": 1.273989111880614e-05, + "loss": 0.1968, + "step": 8202 + }, + { + "epoch": 3.43, + "learning_rate": 1.273826054911792e-05, + "loss": 0.2022, + "step": 8203 + }, + { + "epoch": 3.43, + "learning_rate": 1.2736629900720832e-05, + "loss": 0.2075, + "step": 8204 + }, + { + "epoch": 3.44, + "learning_rate": 1.2734999173661742e-05, + "loss": 0.173, + "step": 8205 + }, + { + "epoch": 3.44, + "learning_rate": 1.2733368367987528e-05, + "loss": 0.1718, + "step": 8206 + }, + { + "epoch": 3.44, + "learning_rate": 1.2731737483745068e-05, + "loss": 0.1859, + "step": 8207 + }, + { + "epoch": 3.44, + "learning_rate": 1.2730106520981236e-05, + "loss": 0.207, + "step": 8208 + }, + { + "epoch": 3.44, + "learning_rate": 1.2728475479742914e-05, + "loss": 0.1998, + "step": 8209 + }, + { + "epoch": 3.44, + "learning_rate": 1.272684436007699e-05, + "loss": 0.1945, + "step": 8210 + }, + { + "epoch": 3.44, + "learning_rate": 1.272521316203034e-05, + "loss": 0.1999, + "step": 8211 + }, + { + "epoch": 3.44, + "learning_rate": 1.2723581885649858e-05, + "loss": 0.1899, + "step": 8212 + }, + { + "epoch": 3.44, + "learning_rate": 1.2721950530982431e-05, + "loss": 0.1733, + "step": 8213 + }, + { + "epoch": 3.44, + "learning_rate": 1.2720319098074954e-05, + "loss": 0.2122, + "step": 8214 + }, + { + "epoch": 3.44, + "learning_rate": 1.2718687586974317e-05, + "loss": 0.1633, + "step": 8215 + }, + { + "epoch": 3.44, + "learning_rate": 1.2717055997727416e-05, + "loss": 0.1858, + "step": 8216 + }, + { + "epoch": 3.44, + "learning_rate": 1.2715424330381155e-05, + "loss": 0.1657, + "step": 8217 + }, + { + "epoch": 3.44, + "learning_rate": 1.2713792584982426e-05, + "loss": 0.2026, + "step": 8218 + }, + { + "epoch": 3.44, + "learning_rate": 1.2712160761578138e-05, + "loss": 0.1809, + "step": 8219 + }, + { + "epoch": 3.44, + "learning_rate": 1.27105288602152e-05, + "loss": 0.1778, + "step": 8220 + }, + { + "epoch": 3.44, + "learning_rate": 1.2708896880940513e-05, + "loss": 0.2055, + "step": 8221 + }, + { + "epoch": 3.44, + "learning_rate": 1.270726482380099e-05, + "loss": 0.211, + "step": 8222 + }, + { + "epoch": 3.44, + "learning_rate": 1.270563268884354e-05, + "loss": 0.1524, + "step": 8223 + }, + { + "epoch": 3.44, + "learning_rate": 1.2704000476115079e-05, + "loss": 0.1888, + "step": 8224 + }, + { + "epoch": 3.44, + "learning_rate": 1.2702368185662529e-05, + "loss": 0.1963, + "step": 8225 + }, + { + "epoch": 3.44, + "learning_rate": 1.2700735817532796e-05, + "loss": 0.1832, + "step": 8226 + }, + { + "epoch": 3.44, + "learning_rate": 1.2699103371772815e-05, + "loss": 0.1948, + "step": 8227 + }, + { + "epoch": 3.44, + "learning_rate": 1.26974708484295e-05, + "loss": 0.2054, + "step": 8228 + }, + { + "epoch": 3.45, + "learning_rate": 1.2695838247549777e-05, + "loss": 0.1572, + "step": 8229 + }, + { + "epoch": 3.45, + "learning_rate": 1.2694205569180577e-05, + "loss": 0.1622, + "step": 8230 + }, + { + "epoch": 3.45, + "learning_rate": 1.269257281336883e-05, + "loss": 0.1943, + "step": 8231 + }, + { + "epoch": 3.45, + "learning_rate": 1.2690939980161465e-05, + "loss": 0.1662, + "step": 8232 + }, + { + "epoch": 3.45, + "learning_rate": 1.2689307069605418e-05, + "loss": 0.1833, + "step": 8233 + }, + { + "epoch": 3.45, + "learning_rate": 1.268767408174763e-05, + "loss": 0.2034, + "step": 8234 + }, + { + "epoch": 3.45, + "learning_rate": 1.2686041016635031e-05, + "loss": 0.1898, + "step": 8235 + }, + { + "epoch": 3.45, + "learning_rate": 1.2684407874314568e-05, + "loss": 0.1917, + "step": 8236 + }, + { + "epoch": 3.45, + "learning_rate": 1.2682774654833188e-05, + "loss": 0.1471, + "step": 8237 + }, + { + "epoch": 3.45, + "learning_rate": 1.2681141358237823e-05, + "loss": 0.1666, + "step": 8238 + }, + { + "epoch": 3.45, + "learning_rate": 1.2679507984575431e-05, + "loss": 0.162, + "step": 8239 + }, + { + "epoch": 3.45, + "learning_rate": 1.2677874533892963e-05, + "loss": 0.18, + "step": 8240 + }, + { + "epoch": 3.45, + "learning_rate": 1.2676241006237367e-05, + "loss": 0.1862, + "step": 8241 + }, + { + "epoch": 3.45, + "learning_rate": 1.2674607401655595e-05, + "loss": 0.1574, + "step": 8242 + }, + { + "epoch": 3.45, + "learning_rate": 1.2672973720194608e-05, + "loss": 0.1919, + "step": 8243 + }, + { + "epoch": 3.45, + "learning_rate": 1.2671339961901366e-05, + "loss": 0.2196, + "step": 8244 + }, + { + "epoch": 3.45, + "learning_rate": 1.2669706126822823e-05, + "loss": 0.1901, + "step": 8245 + }, + { + "epoch": 3.45, + "learning_rate": 1.2668072215005946e-05, + "loss": 0.1999, + "step": 8246 + }, + { + "epoch": 3.45, + "learning_rate": 1.2666438226497706e-05, + "loss": 0.2016, + "step": 8247 + }, + { + "epoch": 3.45, + "learning_rate": 1.2664804161345064e-05, + "loss": 0.2066, + "step": 8248 + }, + { + "epoch": 3.45, + "learning_rate": 1.2663170019594988e-05, + "loss": 0.1709, + "step": 8249 + }, + { + "epoch": 3.45, + "learning_rate": 1.2661535801294455e-05, + "loss": 0.1545, + "step": 8250 + }, + { + "epoch": 3.45, + "learning_rate": 1.2659901506490437e-05, + "loss": 0.1295, + "step": 8251 + }, + { + "epoch": 3.45, + "learning_rate": 1.2658267135229911e-05, + "loss": 0.1584, + "step": 8252 + }, + { + "epoch": 3.46, + "learning_rate": 1.2656632687559853e-05, + "loss": 0.1862, + "step": 8253 + }, + { + "epoch": 3.46, + "learning_rate": 1.265499816352725e-05, + "loss": 0.1749, + "step": 8254 + }, + { + "epoch": 3.46, + "learning_rate": 1.2653363563179078e-05, + "loss": 0.1752, + "step": 8255 + }, + { + "epoch": 3.46, + "learning_rate": 1.2651728886562324e-05, + "loss": 0.178, + "step": 8256 + }, + { + "epoch": 3.46, + "learning_rate": 1.2650094133723978e-05, + "loss": 0.1874, + "step": 8257 + }, + { + "epoch": 3.46, + "learning_rate": 1.2648459304711026e-05, + "loss": 0.1988, + "step": 8258 + }, + { + "epoch": 3.46, + "learning_rate": 1.2646824399570466e-05, + "loss": 0.1609, + "step": 8259 + }, + { + "epoch": 3.46, + "learning_rate": 1.2645189418349282e-05, + "loss": 0.2107, + "step": 8260 + }, + { + "epoch": 3.46, + "learning_rate": 1.2643554361094478e-05, + "loss": 0.2183, + "step": 8261 + }, + { + "epoch": 3.46, + "learning_rate": 1.264191922785305e-05, + "loss": 0.201, + "step": 8262 + }, + { + "epoch": 3.46, + "learning_rate": 1.2640284018671995e-05, + "loss": 0.1965, + "step": 8263 + }, + { + "epoch": 3.46, + "learning_rate": 1.2638648733598325e-05, + "loss": 0.1527, + "step": 8264 + }, + { + "epoch": 3.46, + "learning_rate": 1.2637013372679032e-05, + "loss": 0.188, + "step": 8265 + }, + { + "epoch": 3.46, + "learning_rate": 1.2635377935961134e-05, + "loss": 0.1662, + "step": 8266 + }, + { + "epoch": 3.46, + "learning_rate": 1.2633742423491636e-05, + "loss": 0.1792, + "step": 8267 + }, + { + "epoch": 3.46, + "learning_rate": 1.2632106835317545e-05, + "loss": 0.1997, + "step": 8268 + }, + { + "epoch": 3.46, + "learning_rate": 1.263047117148588e-05, + "loss": 0.1857, + "step": 8269 + }, + { + "epoch": 3.46, + "learning_rate": 1.2628835432043655e-05, + "loss": 0.2021, + "step": 8270 + }, + { + "epoch": 3.46, + "learning_rate": 1.2627199617037895e-05, + "loss": 0.1827, + "step": 8271 + }, + { + "epoch": 3.46, + "learning_rate": 1.2625563726515609e-05, + "loss": 0.2168, + "step": 8272 + }, + { + "epoch": 3.46, + "learning_rate": 1.2623927760523822e-05, + "loss": 0.2037, + "step": 8273 + }, + { + "epoch": 3.46, + "learning_rate": 1.2622291719109559e-05, + "loss": 0.1773, + "step": 8274 + }, + { + "epoch": 3.46, + "learning_rate": 1.2620655602319851e-05, + "loss": 0.1975, + "step": 8275 + }, + { + "epoch": 3.46, + "learning_rate": 1.261901941020172e-05, + "loss": 0.1958, + "step": 8276 + }, + { + "epoch": 3.47, + "learning_rate": 1.2617383142802202e-05, + "loss": 0.2037, + "step": 8277 + }, + { + "epoch": 3.47, + "learning_rate": 1.261574680016833e-05, + "loss": 0.166, + "step": 8278 + }, + { + "epoch": 3.47, + "learning_rate": 1.2614110382347133e-05, + "loss": 0.2285, + "step": 8279 + }, + { + "epoch": 3.47, + "learning_rate": 1.2612473889385655e-05, + "loss": 0.1713, + "step": 8280 + }, + { + "epoch": 3.47, + "learning_rate": 1.2610837321330935e-05, + "loss": 0.1785, + "step": 8281 + }, + { + "epoch": 3.47, + "learning_rate": 1.2609200678230014e-05, + "loss": 0.1969, + "step": 8282 + }, + { + "epoch": 3.47, + "learning_rate": 1.260756396012993e-05, + "loss": 0.2023, + "step": 8283 + }, + { + "epoch": 3.47, + "learning_rate": 1.2605927167077736e-05, + "loss": 0.2079, + "step": 8284 + }, + { + "epoch": 3.47, + "learning_rate": 1.2604290299120478e-05, + "loss": 0.1478, + "step": 8285 + }, + { + "epoch": 3.47, + "learning_rate": 1.2602653356305207e-05, + "loss": 0.1633, + "step": 8286 + }, + { + "epoch": 3.47, + "learning_rate": 1.2601016338678976e-05, + "loss": 0.2344, + "step": 8287 + }, + { + "epoch": 3.47, + "learning_rate": 1.2599379246288837e-05, + "loss": 0.179, + "step": 8288 + }, + { + "epoch": 3.47, + "learning_rate": 1.2597742079181847e-05, + "loss": 0.2131, + "step": 8289 + }, + { + "epoch": 3.47, + "learning_rate": 1.2596104837405065e-05, + "loss": 0.1779, + "step": 8290 + }, + { + "epoch": 3.47, + "learning_rate": 1.2594467521005557e-05, + "loss": 0.203, + "step": 8291 + }, + { + "epoch": 3.47, + "learning_rate": 1.259283013003038e-05, + "loss": 0.1678, + "step": 8292 + }, + { + "epoch": 3.47, + "learning_rate": 1.2591192664526606e-05, + "loss": 0.1857, + "step": 8293 + }, + { + "epoch": 3.47, + "learning_rate": 1.2589555124541293e-05, + "loss": 0.1401, + "step": 8294 + }, + { + "epoch": 3.47, + "learning_rate": 1.2587917510121518e-05, + "loss": 0.1979, + "step": 8295 + }, + { + "epoch": 3.47, + "learning_rate": 1.2586279821314351e-05, + "loss": 0.1971, + "step": 8296 + }, + { + "epoch": 3.47, + "learning_rate": 1.2584642058166863e-05, + "loss": 0.1511, + "step": 8297 + }, + { + "epoch": 3.47, + "learning_rate": 1.2583004220726136e-05, + "loss": 0.1771, + "step": 8298 + }, + { + "epoch": 3.47, + "learning_rate": 1.2581366309039242e-05, + "loss": 0.1956, + "step": 8299 + }, + { + "epoch": 3.47, + "learning_rate": 1.2579728323153267e-05, + "loss": 0.1676, + "step": 8300 + }, + { + "epoch": 3.48, + "learning_rate": 1.257809026311529e-05, + "loss": 0.1678, + "step": 8301 + }, + { + "epoch": 3.48, + "learning_rate": 1.2576452128972394e-05, + "loss": 0.2065, + "step": 8302 + }, + { + "epoch": 3.48, + "learning_rate": 1.2574813920771667e-05, + "loss": 0.1917, + "step": 8303 + }, + { + "epoch": 3.48, + "learning_rate": 1.25731756385602e-05, + "loss": 0.1817, + "step": 8304 + }, + { + "epoch": 3.48, + "learning_rate": 1.2571537282385087e-05, + "loss": 0.2061, + "step": 8305 + }, + { + "epoch": 3.48, + "learning_rate": 1.2569898852293412e-05, + "loss": 0.1832, + "step": 8306 + }, + { + "epoch": 3.48, + "learning_rate": 1.2568260348332276e-05, + "loss": 0.1655, + "step": 8307 + }, + { + "epoch": 3.48, + "learning_rate": 1.2566621770548777e-05, + "loss": 0.1663, + "step": 8308 + }, + { + "epoch": 3.48, + "learning_rate": 1.256498311899001e-05, + "loss": 0.1934, + "step": 8309 + }, + { + "epoch": 3.48, + "learning_rate": 1.2563344393703081e-05, + "loss": 0.1964, + "step": 8310 + }, + { + "epoch": 3.48, + "learning_rate": 1.256170559473509e-05, + "loss": 0.1858, + "step": 8311 + }, + { + "epoch": 3.48, + "learning_rate": 1.2560066722133146e-05, + "loss": 0.1925, + "step": 8312 + }, + { + "epoch": 3.48, + "learning_rate": 1.2558427775944357e-05, + "loss": 0.2114, + "step": 8313 + }, + { + "epoch": 3.48, + "learning_rate": 1.255678875621583e-05, + "loss": 0.1507, + "step": 8314 + }, + { + "epoch": 3.48, + "learning_rate": 1.2555149662994684e-05, + "loss": 0.1966, + "step": 8315 + }, + { + "epoch": 3.48, + "learning_rate": 1.2553510496328025e-05, + "loss": 0.1924, + "step": 8316 + }, + { + "epoch": 3.48, + "learning_rate": 1.255187125626297e-05, + "loss": 0.1676, + "step": 8317 + }, + { + "epoch": 3.48, + "learning_rate": 1.2550231942846643e-05, + "loss": 0.1972, + "step": 8318 + }, + { + "epoch": 3.48, + "learning_rate": 1.2548592556126163e-05, + "loss": 0.1806, + "step": 8319 + }, + { + "epoch": 3.48, + "learning_rate": 1.2546953096148648e-05, + "loss": 0.18, + "step": 8320 + }, + { + "epoch": 3.48, + "learning_rate": 1.2545313562961228e-05, + "loss": 0.201, + "step": 8321 + }, + { + "epoch": 3.48, + "learning_rate": 1.2543673956611029e-05, + "loss": 0.1876, + "step": 8322 + }, + { + "epoch": 3.48, + "learning_rate": 1.2542034277145178e-05, + "loss": 0.1419, + "step": 8323 + }, + { + "epoch": 3.49, + "learning_rate": 1.2540394524610808e-05, + "loss": 0.1911, + "step": 8324 + }, + { + "epoch": 3.49, + "learning_rate": 1.2538754699055057e-05, + "loss": 0.1678, + "step": 8325 + }, + { + "epoch": 3.49, + "learning_rate": 1.2537114800525047e-05, + "loss": 0.2095, + "step": 8326 + }, + { + "epoch": 3.49, + "learning_rate": 1.2535474829067928e-05, + "loss": 0.2091, + "step": 8327 + }, + { + "epoch": 3.49, + "learning_rate": 1.2533834784730835e-05, + "loss": 0.191, + "step": 8328 + }, + { + "epoch": 3.49, + "learning_rate": 1.2532194667560908e-05, + "loss": 0.2376, + "step": 8329 + }, + { + "epoch": 3.49, + "learning_rate": 1.253055447760529e-05, + "loss": 0.1688, + "step": 8330 + }, + { + "epoch": 3.49, + "learning_rate": 1.2528914214911132e-05, + "loss": 0.2203, + "step": 8331 + }, + { + "epoch": 3.49, + "learning_rate": 1.2527273879525579e-05, + "loss": 0.1921, + "step": 8332 + }, + { + "epoch": 3.49, + "learning_rate": 1.2525633471495782e-05, + "loss": 0.1735, + "step": 8333 + }, + { + "epoch": 3.49, + "learning_rate": 1.252399299086889e-05, + "loss": 0.1715, + "step": 8334 + }, + { + "epoch": 3.49, + "learning_rate": 1.252235243769206e-05, + "loss": 0.2148, + "step": 8335 + }, + { + "epoch": 3.49, + "learning_rate": 1.2520711812012447e-05, + "loss": 0.1942, + "step": 8336 + }, + { + "epoch": 3.49, + "learning_rate": 1.2519071113877211e-05, + "loss": 0.2036, + "step": 8337 + }, + { + "epoch": 3.49, + "learning_rate": 1.2517430343333508e-05, + "loss": 0.1769, + "step": 8338 + }, + { + "epoch": 3.49, + "learning_rate": 1.251578950042851e-05, + "loss": 0.1718, + "step": 8339 + }, + { + "epoch": 3.49, + "learning_rate": 1.2514148585209372e-05, + "loss": 0.1998, + "step": 8340 + }, + { + "epoch": 3.49, + "learning_rate": 1.251250759772326e-05, + "loss": 0.1851, + "step": 8341 + }, + { + "epoch": 3.49, + "learning_rate": 1.2510866538017352e-05, + "loss": 0.2033, + "step": 8342 + }, + { + "epoch": 3.49, + "learning_rate": 1.2509225406138809e-05, + "loss": 0.1818, + "step": 8343 + }, + { + "epoch": 3.49, + "learning_rate": 1.2507584202134812e-05, + "loss": 0.2065, + "step": 8344 + }, + { + "epoch": 3.49, + "learning_rate": 1.250594292605253e-05, + "loss": 0.1722, + "step": 8345 + }, + { + "epoch": 3.49, + "learning_rate": 1.2504301577939146e-05, + "loss": 0.2298, + "step": 8346 + }, + { + "epoch": 3.49, + "learning_rate": 1.250266015784183e-05, + "loss": 0.2453, + "step": 8347 + }, + { + "epoch": 3.5, + "learning_rate": 1.2501018665807772e-05, + "loss": 0.1601, + "step": 8348 + }, + { + "epoch": 3.5, + "learning_rate": 1.2499377101884153e-05, + "loss": 0.1856, + "step": 8349 + }, + { + "epoch": 3.5, + "learning_rate": 1.2497735466118156e-05, + "loss": 0.224, + "step": 8350 + }, + { + "epoch": 3.5, + "learning_rate": 1.249609375855697e-05, + "loss": 0.2147, + "step": 8351 + }, + { + "epoch": 3.5, + "learning_rate": 1.2494451979247782e-05, + "loss": 0.2016, + "step": 8352 + }, + { + "epoch": 3.5, + "learning_rate": 1.2492810128237787e-05, + "loss": 0.1871, + "step": 8353 + }, + { + "epoch": 3.5, + "learning_rate": 1.2491168205574175e-05, + "loss": 0.2007, + "step": 8354 + }, + { + "epoch": 3.5, + "learning_rate": 1.2489526211304143e-05, + "loss": 0.1742, + "step": 8355 + }, + { + "epoch": 3.5, + "learning_rate": 1.2487884145474894e-05, + "loss": 0.1818, + "step": 8356 + }, + { + "epoch": 3.5, + "learning_rate": 1.2486242008133618e-05, + "loss": 0.1848, + "step": 8357 + }, + { + "epoch": 3.5, + "learning_rate": 1.2484599799327524e-05, + "loss": 0.1861, + "step": 8358 + }, + { + "epoch": 3.5, + "learning_rate": 1.2482957519103813e-05, + "loss": 0.1962, + "step": 8359 + }, + { + "epoch": 3.5, + "learning_rate": 1.2481315167509691e-05, + "loss": 0.163, + "step": 8360 + }, + { + "epoch": 3.5, + "learning_rate": 1.2479672744592368e-05, + "loss": 0.1704, + "step": 8361 + }, + { + "epoch": 3.5, + "learning_rate": 1.2478030250399052e-05, + "loss": 0.1996, + "step": 8362 + }, + { + "epoch": 3.5, + "learning_rate": 1.2476387684976956e-05, + "loss": 0.1665, + "step": 8363 + }, + { + "epoch": 3.5, + "learning_rate": 1.2474745048373292e-05, + "loss": 0.1742, + "step": 8364 + }, + { + "epoch": 3.5, + "learning_rate": 1.2473102340635275e-05, + "loss": 0.1708, + "step": 8365 + }, + { + "epoch": 3.5, + "learning_rate": 1.247145956181013e-05, + "loss": 0.2358, + "step": 8366 + }, + { + "epoch": 3.5, + "learning_rate": 1.246981671194507e-05, + "loss": 0.1767, + "step": 8367 + }, + { + "epoch": 3.5, + "learning_rate": 1.2468173791087324e-05, + "loss": 0.1535, + "step": 8368 + }, + { + "epoch": 3.5, + "learning_rate": 1.2466530799284111e-05, + "loss": 0.1793, + "step": 8369 + }, + { + "epoch": 3.5, + "learning_rate": 1.2464887736582659e-05, + "loss": 0.1774, + "step": 8370 + }, + { + "epoch": 3.5, + "learning_rate": 1.2463244603030194e-05, + "loss": 0.1839, + "step": 8371 + }, + { + "epoch": 3.51, + "learning_rate": 1.2461601398673953e-05, + "loss": 0.2015, + "step": 8372 + }, + { + "epoch": 3.51, + "learning_rate": 1.2459958123561163e-05, + "loss": 0.1921, + "step": 8373 + }, + { + "epoch": 3.51, + "learning_rate": 1.2458314777739056e-05, + "loss": 0.1948, + "step": 8374 + }, + { + "epoch": 3.51, + "learning_rate": 1.2456671361254876e-05, + "loss": 0.2294, + "step": 8375 + }, + { + "epoch": 3.51, + "learning_rate": 1.2455027874155856e-05, + "loss": 0.193, + "step": 8376 + }, + { + "epoch": 3.51, + "learning_rate": 1.2453384316489239e-05, + "loss": 0.1784, + "step": 8377 + }, + { + "epoch": 3.51, + "learning_rate": 1.2451740688302265e-05, + "loss": 0.1754, + "step": 8378 + }, + { + "epoch": 3.51, + "learning_rate": 1.2450096989642184e-05, + "loss": 0.1777, + "step": 8379 + }, + { + "epoch": 3.51, + "learning_rate": 1.2448453220556235e-05, + "loss": 0.1634, + "step": 8380 + }, + { + "epoch": 3.51, + "learning_rate": 1.2446809381091674e-05, + "loss": 0.1739, + "step": 8381 + }, + { + "epoch": 3.51, + "learning_rate": 1.2445165471295745e-05, + "loss": 0.1825, + "step": 8382 + }, + { + "epoch": 3.51, + "learning_rate": 1.244352149121571e-05, + "loss": 0.1394, + "step": 8383 + }, + { + "epoch": 3.51, + "learning_rate": 1.2441877440898817e-05, + "loss": 0.1721, + "step": 8384 + }, + { + "epoch": 3.51, + "learning_rate": 1.2440233320392322e-05, + "loss": 0.1992, + "step": 8385 + }, + { + "epoch": 3.51, + "learning_rate": 1.2438589129743484e-05, + "loss": 0.1969, + "step": 8386 + }, + { + "epoch": 3.51, + "learning_rate": 1.243694486899957e-05, + "loss": 0.214, + "step": 8387 + }, + { + "epoch": 3.51, + "learning_rate": 1.2435300538207832e-05, + "loss": 0.1699, + "step": 8388 + }, + { + "epoch": 3.51, + "learning_rate": 1.2433656137415545e-05, + "loss": 0.1888, + "step": 8389 + }, + { + "epoch": 3.51, + "learning_rate": 1.2432011666669972e-05, + "loss": 0.1743, + "step": 8390 + }, + { + "epoch": 3.51, + "learning_rate": 1.243036712601838e-05, + "loss": 0.1822, + "step": 8391 + }, + { + "epoch": 3.51, + "learning_rate": 1.2428722515508043e-05, + "loss": 0.1835, + "step": 8392 + }, + { + "epoch": 3.51, + "learning_rate": 1.2427077835186235e-05, + "loss": 0.2189, + "step": 8393 + }, + { + "epoch": 3.51, + "learning_rate": 1.2425433085100224e-05, + "loss": 0.1872, + "step": 8394 + }, + { + "epoch": 3.51, + "learning_rate": 1.2423788265297294e-05, + "loss": 0.2121, + "step": 8395 + }, + { + "epoch": 3.52, + "learning_rate": 1.242214337582472e-05, + "loss": 0.1972, + "step": 8396 + }, + { + "epoch": 3.52, + "learning_rate": 1.2420498416729785e-05, + "loss": 0.2004, + "step": 8397 + }, + { + "epoch": 3.52, + "learning_rate": 1.2418853388059771e-05, + "loss": 0.1933, + "step": 8398 + }, + { + "epoch": 3.52, + "learning_rate": 1.241720828986196e-05, + "loss": 0.18, + "step": 8399 + }, + { + "epoch": 3.52, + "learning_rate": 1.2415563122183644e-05, + "loss": 0.2336, + "step": 8400 + }, + { + "epoch": 3.52, + "learning_rate": 1.2413917885072109e-05, + "loss": 0.1966, + "step": 8401 + }, + { + "epoch": 3.52, + "learning_rate": 1.2412272578574647e-05, + "loss": 0.2009, + "step": 8402 + }, + { + "epoch": 3.52, + "learning_rate": 1.241062720273855e-05, + "loss": 0.1812, + "step": 8403 + }, + { + "epoch": 3.52, + "learning_rate": 1.2408981757611112e-05, + "loss": 0.2079, + "step": 8404 + }, + { + "epoch": 3.52, + "learning_rate": 1.2407336243239632e-05, + "loss": 0.1831, + "step": 8405 + }, + { + "epoch": 3.52, + "learning_rate": 1.240569065967141e-05, + "loss": 0.2116, + "step": 8406 + }, + { + "epoch": 3.52, + "learning_rate": 1.2404045006953744e-05, + "loss": 0.1673, + "step": 8407 + }, + { + "epoch": 3.52, + "learning_rate": 1.2402399285133935e-05, + "loss": 0.2214, + "step": 8408 + }, + { + "epoch": 3.52, + "learning_rate": 1.2400753494259293e-05, + "loss": 0.1845, + "step": 8409 + }, + { + "epoch": 3.52, + "learning_rate": 1.2399107634377122e-05, + "loss": 0.1506, + "step": 8410 + }, + { + "epoch": 3.52, + "learning_rate": 1.2397461705534727e-05, + "loss": 0.1963, + "step": 8411 + }, + { + "epoch": 3.52, + "learning_rate": 1.2395815707779426e-05, + "loss": 0.2047, + "step": 8412 + }, + { + "epoch": 3.52, + "learning_rate": 1.239416964115853e-05, + "loss": 0.2205, + "step": 8413 + }, + { + "epoch": 3.52, + "learning_rate": 1.2392523505719349e-05, + "loss": 0.1948, + "step": 8414 + }, + { + "epoch": 3.52, + "learning_rate": 1.2390877301509205e-05, + "loss": 0.2205, + "step": 8415 + }, + { + "epoch": 3.52, + "learning_rate": 1.2389231028575412e-05, + "loss": 0.1596, + "step": 8416 + }, + { + "epoch": 3.52, + "learning_rate": 1.2387584686965297e-05, + "loss": 0.2314, + "step": 8417 + }, + { + "epoch": 3.52, + "learning_rate": 1.2385938276726179e-05, + "loss": 0.1764, + "step": 8418 + }, + { + "epoch": 3.52, + "learning_rate": 1.2384291797905382e-05, + "loss": 0.1856, + "step": 8419 + }, + { + "epoch": 3.53, + "learning_rate": 1.2382645250550232e-05, + "loss": 0.2062, + "step": 8420 + }, + { + "epoch": 3.53, + "learning_rate": 1.2380998634708057e-05, + "loss": 0.2007, + "step": 8421 + }, + { + "epoch": 3.53, + "learning_rate": 1.2379351950426188e-05, + "loss": 0.2188, + "step": 8422 + }, + { + "epoch": 3.53, + "learning_rate": 1.2377705197751961e-05, + "loss": 0.212, + "step": 8423 + }, + { + "epoch": 3.53, + "learning_rate": 1.237605837673271e-05, + "loss": 0.2021, + "step": 8424 + }, + { + "epoch": 3.53, + "learning_rate": 1.2374411487415766e-05, + "loss": 0.1813, + "step": 8425 + }, + { + "epoch": 3.53, + "learning_rate": 1.2372764529848471e-05, + "loss": 0.1811, + "step": 8426 + }, + { + "epoch": 3.53, + "learning_rate": 1.237111750407817e-05, + "loss": 0.1925, + "step": 8427 + }, + { + "epoch": 3.53, + "learning_rate": 1.2369470410152195e-05, + "loss": 0.1922, + "step": 8428 + }, + { + "epoch": 3.53, + "learning_rate": 1.2367823248117899e-05, + "loss": 0.2002, + "step": 8429 + }, + { + "epoch": 3.53, + "learning_rate": 1.2366176018022621e-05, + "loss": 0.2103, + "step": 8430 + }, + { + "epoch": 3.53, + "learning_rate": 1.2364528719913717e-05, + "loss": 0.1977, + "step": 8431 + }, + { + "epoch": 3.53, + "learning_rate": 1.2362881353838532e-05, + "loss": 0.1977, + "step": 8432 + }, + { + "epoch": 3.53, + "learning_rate": 1.236123391984442e-05, + "loss": 0.1646, + "step": 8433 + }, + { + "epoch": 3.53, + "learning_rate": 1.2359586417978733e-05, + "loss": 0.1623, + "step": 8434 + }, + { + "epoch": 3.53, + "learning_rate": 1.2357938848288829e-05, + "loss": 0.2135, + "step": 8435 + }, + { + "epoch": 3.53, + "learning_rate": 1.2356291210822063e-05, + "loss": 0.2054, + "step": 8436 + }, + { + "epoch": 3.53, + "learning_rate": 1.23546435056258e-05, + "loss": 0.1851, + "step": 8437 + }, + { + "epoch": 3.53, + "learning_rate": 1.2352995732747398e-05, + "loss": 0.2118, + "step": 8438 + }, + { + "epoch": 3.53, + "learning_rate": 1.235134789223422e-05, + "loss": 0.2034, + "step": 8439 + }, + { + "epoch": 3.53, + "learning_rate": 1.2349699984133636e-05, + "loss": 0.1801, + "step": 8440 + }, + { + "epoch": 3.53, + "learning_rate": 1.234805200849301e-05, + "loss": 0.1968, + "step": 8441 + }, + { + "epoch": 3.53, + "learning_rate": 1.2346403965359712e-05, + "loss": 0.1928, + "step": 8442 + }, + { + "epoch": 3.53, + "learning_rate": 1.2344755854781114e-05, + "loss": 0.1793, + "step": 8443 + }, + { + "epoch": 3.54, + "learning_rate": 1.2343107676804593e-05, + "loss": 0.1609, + "step": 8444 + }, + { + "epoch": 3.54, + "learning_rate": 1.2341459431477516e-05, + "loss": 0.1669, + "step": 8445 + }, + { + "epoch": 3.54, + "learning_rate": 1.2339811118847269e-05, + "loss": 0.1921, + "step": 8446 + }, + { + "epoch": 3.54, + "learning_rate": 1.2338162738961227e-05, + "loss": 0.1854, + "step": 8447 + }, + { + "epoch": 3.54, + "learning_rate": 1.233651429186677e-05, + "loss": 0.1917, + "step": 8448 + }, + { + "epoch": 3.54, + "learning_rate": 1.2334865777611284e-05, + "loss": 0.1943, + "step": 8449 + }, + { + "epoch": 3.54, + "learning_rate": 1.2333217196242152e-05, + "loss": 0.1806, + "step": 8450 + }, + { + "epoch": 3.54, + "learning_rate": 1.2331568547806768e-05, + "loss": 0.1831, + "step": 8451 + }, + { + "epoch": 3.54, + "learning_rate": 1.232991983235251e-05, + "loss": 0.2109, + "step": 8452 + }, + { + "epoch": 3.54, + "learning_rate": 1.2328271049926774e-05, + "loss": 0.213, + "step": 8453 + }, + { + "epoch": 3.54, + "learning_rate": 1.2326622200576956e-05, + "loss": 0.1713, + "step": 8454 + }, + { + "epoch": 3.54, + "learning_rate": 1.2324973284350444e-05, + "loss": 0.2101, + "step": 8455 + }, + { + "epoch": 3.54, + "learning_rate": 1.2323324301294637e-05, + "loss": 0.1893, + "step": 8456 + }, + { + "epoch": 3.54, + "learning_rate": 1.2321675251456937e-05, + "loss": 0.1919, + "step": 8457 + }, + { + "epoch": 3.54, + "learning_rate": 1.2320026134884744e-05, + "loss": 0.1883, + "step": 8458 + }, + { + "epoch": 3.54, + "learning_rate": 1.2318376951625455e-05, + "loss": 0.1351, + "step": 8459 + }, + { + "epoch": 3.54, + "learning_rate": 1.2316727701726479e-05, + "loss": 0.2197, + "step": 8460 + }, + { + "epoch": 3.54, + "learning_rate": 1.2315078385235223e-05, + "loss": 0.1724, + "step": 8461 + }, + { + "epoch": 3.54, + "learning_rate": 1.2313429002199088e-05, + "loss": 0.1657, + "step": 8462 + }, + { + "epoch": 3.54, + "learning_rate": 1.2311779552665495e-05, + "loss": 0.1867, + "step": 8463 + }, + { + "epoch": 3.54, + "learning_rate": 1.2310130036681847e-05, + "loss": 0.1593, + "step": 8464 + }, + { + "epoch": 3.54, + "learning_rate": 1.2308480454295563e-05, + "loss": 0.1857, + "step": 8465 + }, + { + "epoch": 3.54, + "learning_rate": 1.2306830805554055e-05, + "loss": 0.1832, + "step": 8466 + }, + { + "epoch": 3.54, + "learning_rate": 1.2305181090504744e-05, + "loss": 0.1751, + "step": 8467 + }, + { + "epoch": 3.55, + "learning_rate": 1.2303531309195047e-05, + "loss": 0.178, + "step": 8468 + }, + { + "epoch": 3.55, + "learning_rate": 1.2301881461672387e-05, + "loss": 0.1923, + "step": 8469 + }, + { + "epoch": 3.55, + "learning_rate": 1.2300231547984187e-05, + "loss": 0.1603, + "step": 8470 + }, + { + "epoch": 3.55, + "learning_rate": 1.2298581568177874e-05, + "loss": 0.2233, + "step": 8471 + }, + { + "epoch": 3.55, + "learning_rate": 1.229693152230087e-05, + "loss": 0.1936, + "step": 8472 + }, + { + "epoch": 3.55, + "learning_rate": 1.2295281410400611e-05, + "loss": 0.198, + "step": 8473 + }, + { + "epoch": 3.55, + "learning_rate": 1.2293631232524524e-05, + "loss": 0.1754, + "step": 8474 + }, + { + "epoch": 3.55, + "learning_rate": 1.2291980988720042e-05, + "loss": 0.156, + "step": 8475 + }, + { + "epoch": 3.55, + "learning_rate": 1.22903306790346e-05, + "loss": 0.192, + "step": 8476 + }, + { + "epoch": 3.55, + "learning_rate": 1.2288680303515634e-05, + "loss": 0.1959, + "step": 8477 + }, + { + "epoch": 3.55, + "learning_rate": 1.2287029862210587e-05, + "loss": 0.2003, + "step": 8478 + }, + { + "epoch": 3.55, + "learning_rate": 1.2285379355166893e-05, + "loss": 0.1684, + "step": 8479 + }, + { + "epoch": 3.55, + "learning_rate": 1.2283728782431996e-05, + "loss": 0.2383, + "step": 8480 + }, + { + "epoch": 3.55, + "learning_rate": 1.2282078144053347e-05, + "loss": 0.212, + "step": 8481 + }, + { + "epoch": 3.55, + "learning_rate": 1.2280427440078383e-05, + "loss": 0.1748, + "step": 8482 + }, + { + "epoch": 3.55, + "learning_rate": 1.2278776670554555e-05, + "loss": 0.1972, + "step": 8483 + }, + { + "epoch": 3.55, + "learning_rate": 1.2277125835529318e-05, + "loss": 0.1675, + "step": 8484 + }, + { + "epoch": 3.55, + "learning_rate": 1.2275474935050116e-05, + "loss": 0.184, + "step": 8485 + }, + { + "epoch": 3.55, + "learning_rate": 1.227382396916441e-05, + "loss": 0.1557, + "step": 8486 + }, + { + "epoch": 3.55, + "learning_rate": 1.2272172937919646e-05, + "loss": 0.2133, + "step": 8487 + }, + { + "epoch": 3.55, + "learning_rate": 1.2270521841363293e-05, + "loss": 0.2386, + "step": 8488 + }, + { + "epoch": 3.55, + "learning_rate": 1.2268870679542799e-05, + "loss": 0.1974, + "step": 8489 + }, + { + "epoch": 3.55, + "learning_rate": 1.2267219452505635e-05, + "loss": 0.1825, + "step": 8490 + }, + { + "epoch": 3.55, + "learning_rate": 1.226556816029926e-05, + "loss": 0.2028, + "step": 8491 + }, + { + "epoch": 3.56, + "learning_rate": 1.2263916802971134e-05, + "loss": 0.1763, + "step": 8492 + }, + { + "epoch": 3.56, + "learning_rate": 1.2262265380568729e-05, + "loss": 0.1288, + "step": 8493 + }, + { + "epoch": 3.56, + "learning_rate": 1.2260613893139517e-05, + "loss": 0.1676, + "step": 8494 + }, + { + "epoch": 3.56, + "learning_rate": 1.2258962340730965e-05, + "loss": 0.173, + "step": 8495 + }, + { + "epoch": 3.56, + "learning_rate": 1.2257310723390541e-05, + "loss": 0.1756, + "step": 8496 + }, + { + "epoch": 3.56, + "learning_rate": 1.2255659041165726e-05, + "loss": 0.1826, + "step": 8497 + }, + { + "epoch": 3.56, + "learning_rate": 1.2254007294103994e-05, + "loss": 0.2019, + "step": 8498 + }, + { + "epoch": 3.56, + "learning_rate": 1.2252355482252823e-05, + "loss": 0.2117, + "step": 8499 + }, + { + "epoch": 3.56, + "learning_rate": 1.2250703605659688e-05, + "loss": 0.1707, + "step": 8500 + }, + { + "epoch": 3.56, + "learning_rate": 1.2249051664372078e-05, + "loss": 0.1901, + "step": 8501 + }, + { + "epoch": 3.56, + "learning_rate": 1.2247399658437475e-05, + "loss": 0.2087, + "step": 8502 + }, + { + "epoch": 3.56, + "learning_rate": 1.224574758790336e-05, + "loss": 0.2222, + "step": 8503 + }, + { + "epoch": 3.56, + "learning_rate": 1.2244095452817227e-05, + "loss": 0.208, + "step": 8504 + }, + { + "epoch": 3.56, + "learning_rate": 1.2242443253226558e-05, + "loss": 0.2085, + "step": 8505 + }, + { + "epoch": 3.56, + "learning_rate": 1.224079098917885e-05, + "loss": 0.2235, + "step": 8506 + }, + { + "epoch": 3.56, + "learning_rate": 1.2239138660721593e-05, + "loss": 0.1747, + "step": 8507 + }, + { + "epoch": 3.56, + "learning_rate": 1.2237486267902284e-05, + "loss": 0.1916, + "step": 8508 + }, + { + "epoch": 3.56, + "learning_rate": 1.2235833810768417e-05, + "loss": 0.1648, + "step": 8509 + }, + { + "epoch": 3.56, + "learning_rate": 1.2234181289367489e-05, + "loss": 0.2035, + "step": 8510 + }, + { + "epoch": 3.56, + "learning_rate": 1.2232528703747002e-05, + "loss": 0.2268, + "step": 8511 + }, + { + "epoch": 3.56, + "learning_rate": 1.2230876053954462e-05, + "loss": 0.1868, + "step": 8512 + }, + { + "epoch": 3.56, + "learning_rate": 1.2229223340037364e-05, + "loss": 0.1953, + "step": 8513 + }, + { + "epoch": 3.56, + "learning_rate": 1.2227570562043223e-05, + "loss": 0.1992, + "step": 8514 + }, + { + "epoch": 3.56, + "learning_rate": 1.2225917720019544e-05, + "loss": 0.1824, + "step": 8515 + }, + { + "epoch": 3.57, + "learning_rate": 1.2224264814013833e-05, + "loss": 0.1768, + "step": 8516 + }, + { + "epoch": 3.57, + "learning_rate": 1.2222611844073604e-05, + "loss": 0.185, + "step": 8517 + }, + { + "epoch": 3.57, + "learning_rate": 1.2220958810246371e-05, + "loss": 0.1673, + "step": 8518 + }, + { + "epoch": 3.57, + "learning_rate": 1.2219305712579647e-05, + "loss": 0.1732, + "step": 8519 + }, + { + "epoch": 3.57, + "learning_rate": 1.221765255112095e-05, + "loss": 0.2109, + "step": 8520 + }, + { + "epoch": 3.57, + "learning_rate": 1.2215999325917797e-05, + "loss": 0.2072, + "step": 8521 + }, + { + "epoch": 3.57, + "learning_rate": 1.2214346037017715e-05, + "loss": 0.2434, + "step": 8522 + }, + { + "epoch": 3.57, + "learning_rate": 1.2212692684468215e-05, + "loss": 0.1588, + "step": 8523 + }, + { + "epoch": 3.57, + "learning_rate": 1.221103926831683e-05, + "loss": 0.2165, + "step": 8524 + }, + { + "epoch": 3.57, + "learning_rate": 1.2209385788611084e-05, + "loss": 0.2039, + "step": 8525 + }, + { + "epoch": 3.57, + "learning_rate": 1.2207732245398502e-05, + "loss": 0.1687, + "step": 8526 + }, + { + "epoch": 3.57, + "learning_rate": 1.2206078638726615e-05, + "loss": 0.1956, + "step": 8527 + }, + { + "epoch": 3.57, + "learning_rate": 1.2204424968642957e-05, + "loss": 0.2377, + "step": 8528 + }, + { + "epoch": 3.57, + "learning_rate": 1.220277123519506e-05, + "loss": 0.2237, + "step": 8529 + }, + { + "epoch": 3.57, + "learning_rate": 1.2201117438430456e-05, + "loss": 0.1893, + "step": 8530 + }, + { + "epoch": 3.57, + "learning_rate": 1.2199463578396688e-05, + "loss": 0.1917, + "step": 8531 + }, + { + "epoch": 3.57, + "learning_rate": 1.219780965514129e-05, + "loss": 0.2059, + "step": 8532 + }, + { + "epoch": 3.57, + "learning_rate": 1.2196155668711804e-05, + "loss": 0.1924, + "step": 8533 + }, + { + "epoch": 3.57, + "learning_rate": 1.219450161915577e-05, + "loss": 0.1509, + "step": 8534 + }, + { + "epoch": 3.57, + "learning_rate": 1.2192847506520735e-05, + "loss": 0.1846, + "step": 8535 + }, + { + "epoch": 3.57, + "learning_rate": 1.2191193330854246e-05, + "loss": 0.1921, + "step": 8536 + }, + { + "epoch": 3.57, + "learning_rate": 1.2189539092203849e-05, + "loss": 0.1988, + "step": 8537 + }, + { + "epoch": 3.57, + "learning_rate": 1.218788479061709e-05, + "loss": 0.1842, + "step": 8538 + }, + { + "epoch": 3.58, + "learning_rate": 1.2186230426141528e-05, + "loss": 0.1653, + "step": 8539 + }, + { + "epoch": 3.58, + "learning_rate": 1.2184575998824713e-05, + "loss": 0.1869, + "step": 8540 + }, + { + "epoch": 3.58, + "learning_rate": 1.2182921508714196e-05, + "loss": 0.1924, + "step": 8541 + }, + { + "epoch": 3.58, + "learning_rate": 1.218126695585754e-05, + "loss": 0.1826, + "step": 8542 + }, + { + "epoch": 3.58, + "learning_rate": 1.2179612340302302e-05, + "loss": 0.1999, + "step": 8543 + }, + { + "epoch": 3.58, + "learning_rate": 1.2177957662096038e-05, + "loss": 0.217, + "step": 8544 + }, + { + "epoch": 3.58, + "learning_rate": 1.2176302921286316e-05, + "loss": 0.2095, + "step": 8545 + }, + { + "epoch": 3.58, + "learning_rate": 1.2174648117920698e-05, + "loss": 0.202, + "step": 8546 + }, + { + "epoch": 3.58, + "learning_rate": 1.2172993252046747e-05, + "loss": 0.1597, + "step": 8547 + }, + { + "epoch": 3.58, + "learning_rate": 1.2171338323712034e-05, + "loss": 0.1915, + "step": 8548 + }, + { + "epoch": 3.58, + "learning_rate": 1.216968333296413e-05, + "loss": 0.1843, + "step": 8549 + }, + { + "epoch": 3.58, + "learning_rate": 1.2168028279850603e-05, + "loss": 0.2172, + "step": 8550 + }, + { + "epoch": 3.58, + "learning_rate": 1.2166373164419026e-05, + "loss": 0.1981, + "step": 8551 + }, + { + "epoch": 3.58, + "learning_rate": 1.2164717986716977e-05, + "loss": 0.1738, + "step": 8552 + }, + { + "epoch": 3.58, + "learning_rate": 1.2163062746792027e-05, + "loss": 0.1915, + "step": 8553 + }, + { + "epoch": 3.58, + "learning_rate": 1.2161407444691761e-05, + "loss": 0.1887, + "step": 8554 + }, + { + "epoch": 3.58, + "learning_rate": 1.2159752080463753e-05, + "loss": 0.1902, + "step": 8555 + }, + { + "epoch": 3.58, + "learning_rate": 1.2158096654155592e-05, + "loss": 0.1652, + "step": 8556 + }, + { + "epoch": 3.58, + "learning_rate": 1.2156441165814855e-05, + "loss": 0.187, + "step": 8557 + }, + { + "epoch": 3.58, + "learning_rate": 1.2154785615489129e-05, + "loss": 0.1937, + "step": 8558 + }, + { + "epoch": 3.58, + "learning_rate": 1.2153130003226006e-05, + "loss": 0.1814, + "step": 8559 + }, + { + "epoch": 3.58, + "learning_rate": 1.215147432907307e-05, + "loss": 0.1693, + "step": 8560 + }, + { + "epoch": 3.58, + "learning_rate": 1.2149818593077912e-05, + "loss": 0.1944, + "step": 8561 + }, + { + "epoch": 3.58, + "learning_rate": 1.2148162795288127e-05, + "loss": 0.1962, + "step": 8562 + }, + { + "epoch": 3.59, + "learning_rate": 1.214650693575131e-05, + "loss": 0.1754, + "step": 8563 + }, + { + "epoch": 3.59, + "learning_rate": 1.2144851014515055e-05, + "loss": 0.1715, + "step": 8564 + }, + { + "epoch": 3.59, + "learning_rate": 1.2143195031626962e-05, + "loss": 0.1796, + "step": 8565 + }, + { + "epoch": 3.59, + "learning_rate": 1.214153898713463e-05, + "loss": 0.1753, + "step": 8566 + }, + { + "epoch": 3.59, + "learning_rate": 1.2139882881085658e-05, + "loss": 0.1809, + "step": 8567 + }, + { + "epoch": 3.59, + "learning_rate": 1.2138226713527652e-05, + "loss": 0.1781, + "step": 8568 + }, + { + "epoch": 3.59, + "learning_rate": 1.2136570484508217e-05, + "loss": 0.2059, + "step": 8569 + }, + { + "epoch": 3.59, + "learning_rate": 1.213491419407496e-05, + "loss": 0.1986, + "step": 8570 + }, + { + "epoch": 3.59, + "learning_rate": 1.2133257842275487e-05, + "loss": 0.193, + "step": 8571 + }, + { + "epoch": 3.59, + "learning_rate": 1.2131601429157413e-05, + "loss": 0.2063, + "step": 8572 + }, + { + "epoch": 3.59, + "learning_rate": 1.2129944954768346e-05, + "loss": 0.1816, + "step": 8573 + }, + { + "epoch": 3.59, + "learning_rate": 1.2128288419155903e-05, + "loss": 0.2111, + "step": 8574 + }, + { + "epoch": 3.59, + "learning_rate": 1.2126631822367693e-05, + "loss": 0.1456, + "step": 8575 + }, + { + "epoch": 3.59, + "learning_rate": 1.2124975164451345e-05, + "loss": 0.1482, + "step": 8576 + }, + { + "epoch": 3.59, + "learning_rate": 1.212331844545447e-05, + "loss": 0.2008, + "step": 8577 + }, + { + "epoch": 3.59, + "learning_rate": 1.2121661665424688e-05, + "loss": 0.176, + "step": 8578 + }, + { + "epoch": 3.59, + "learning_rate": 1.2120004824409626e-05, + "loss": 0.1779, + "step": 8579 + }, + { + "epoch": 3.59, + "learning_rate": 1.2118347922456905e-05, + "loss": 0.162, + "step": 8580 + }, + { + "epoch": 3.59, + "learning_rate": 1.2116690959614154e-05, + "loss": 0.1954, + "step": 8581 + }, + { + "epoch": 3.59, + "learning_rate": 1.2115033935929e-05, + "loss": 0.183, + "step": 8582 + }, + { + "epoch": 3.59, + "learning_rate": 1.211337685144907e-05, + "loss": 0.1416, + "step": 8583 + }, + { + "epoch": 3.59, + "learning_rate": 1.2111719706222e-05, + "loss": 0.2038, + "step": 8584 + }, + { + "epoch": 3.59, + "learning_rate": 1.2110062500295418e-05, + "loss": 0.1634, + "step": 8585 + }, + { + "epoch": 3.59, + "learning_rate": 1.2108405233716967e-05, + "loss": 0.1703, + "step": 8586 + }, + { + "epoch": 3.6, + "learning_rate": 1.2106747906534275e-05, + "loss": 0.193, + "step": 8587 + }, + { + "epoch": 3.6, + "learning_rate": 1.2105090518794986e-05, + "loss": 0.1664, + "step": 8588 + }, + { + "epoch": 3.6, + "learning_rate": 1.2103433070546737e-05, + "loss": 0.1917, + "step": 8589 + }, + { + "epoch": 3.6, + "learning_rate": 1.2101775561837172e-05, + "loss": 0.1935, + "step": 8590 + }, + { + "epoch": 3.6, + "learning_rate": 1.2100117992713931e-05, + "loss": 0.1685, + "step": 8591 + }, + { + "epoch": 3.6, + "learning_rate": 1.2098460363224662e-05, + "loss": 0.1723, + "step": 8592 + }, + { + "epoch": 3.6, + "learning_rate": 1.2096802673417014e-05, + "loss": 0.1534, + "step": 8593 + }, + { + "epoch": 3.6, + "learning_rate": 1.2095144923338633e-05, + "loss": 0.1846, + "step": 8594 + }, + { + "epoch": 3.6, + "learning_rate": 1.209348711303717e-05, + "loss": 0.217, + "step": 8595 + }, + { + "epoch": 3.6, + "learning_rate": 1.2091829242560278e-05, + "loss": 0.2127, + "step": 8596 + }, + { + "epoch": 3.6, + "learning_rate": 1.209017131195561e-05, + "loss": 0.182, + "step": 8597 + }, + { + "epoch": 3.6, + "learning_rate": 1.2088513321270823e-05, + "loss": 0.2203, + "step": 8598 + }, + { + "epoch": 3.6, + "learning_rate": 1.2086855270553575e-05, + "loss": 0.2172, + "step": 8599 + }, + { + "epoch": 3.6, + "learning_rate": 1.2085197159851525e-05, + "loss": 0.2137, + "step": 8600 + }, + { + "epoch": 3.6, + "learning_rate": 1.208353898921233e-05, + "loss": 0.1924, + "step": 8601 + }, + { + "epoch": 3.6, + "learning_rate": 1.2081880758683656e-05, + "loss": 0.22, + "step": 8602 + }, + { + "epoch": 3.6, + "learning_rate": 1.208022246831317e-05, + "loss": 0.2001, + "step": 8603 + }, + { + "epoch": 3.6, + "learning_rate": 1.2078564118148531e-05, + "loss": 0.2161, + "step": 8604 + }, + { + "epoch": 3.6, + "learning_rate": 1.2076905708237412e-05, + "loss": 0.19, + "step": 8605 + }, + { + "epoch": 3.6, + "learning_rate": 1.2075247238627482e-05, + "loss": 0.1804, + "step": 8606 + }, + { + "epoch": 3.6, + "learning_rate": 1.2073588709366413e-05, + "loss": 0.2071, + "step": 8607 + }, + { + "epoch": 3.6, + "learning_rate": 1.2071930120501874e-05, + "loss": 0.2102, + "step": 8608 + }, + { + "epoch": 3.6, + "learning_rate": 1.2070271472081542e-05, + "loss": 0.1457, + "step": 8609 + }, + { + "epoch": 3.6, + "learning_rate": 1.2068612764153099e-05, + "loss": 0.2246, + "step": 8610 + }, + { + "epoch": 3.61, + "learning_rate": 1.2066953996764215e-05, + "loss": 0.1485, + "step": 8611 + }, + { + "epoch": 3.61, + "learning_rate": 1.2065295169962573e-05, + "loss": 0.1568, + "step": 8612 + }, + { + "epoch": 3.61, + "learning_rate": 1.206363628379585e-05, + "loss": 0.1886, + "step": 8613 + }, + { + "epoch": 3.61, + "learning_rate": 1.2061977338311737e-05, + "loss": 0.1763, + "step": 8614 + }, + { + "epoch": 3.61, + "learning_rate": 1.2060318333557915e-05, + "loss": 0.166, + "step": 8615 + }, + { + "epoch": 3.61, + "learning_rate": 1.2058659269582067e-05, + "loss": 0.1715, + "step": 8616 + }, + { + "epoch": 3.61, + "learning_rate": 1.205700014643189e-05, + "loss": 0.2064, + "step": 8617 + }, + { + "epoch": 3.61, + "learning_rate": 1.2055340964155067e-05, + "loss": 0.2055, + "step": 8618 + }, + { + "epoch": 3.61, + "learning_rate": 1.2053681722799291e-05, + "loss": 0.2265, + "step": 8619 + }, + { + "epoch": 3.61, + "learning_rate": 1.2052022422412259e-05, + "loss": 0.2, + "step": 8620 + }, + { + "epoch": 3.61, + "learning_rate": 1.2050363063041662e-05, + "loss": 0.1541, + "step": 8621 + }, + { + "epoch": 3.61, + "learning_rate": 1.2048703644735198e-05, + "loss": 0.1835, + "step": 8622 + }, + { + "epoch": 3.61, + "learning_rate": 1.2047044167540564e-05, + "loss": 0.1863, + "step": 8623 + }, + { + "epoch": 3.61, + "learning_rate": 1.2045384631505466e-05, + "loss": 0.1417, + "step": 8624 + }, + { + "epoch": 3.61, + "learning_rate": 1.2043725036677599e-05, + "loss": 0.1641, + "step": 8625 + }, + { + "epoch": 3.61, + "learning_rate": 1.204206538310467e-05, + "loss": 0.1823, + "step": 8626 + }, + { + "epoch": 3.61, + "learning_rate": 1.2040405670834385e-05, + "loss": 0.2046, + "step": 8627 + }, + { + "epoch": 3.61, + "learning_rate": 1.2038745899914446e-05, + "loss": 0.1768, + "step": 8628 + }, + { + "epoch": 3.61, + "learning_rate": 1.2037086070392567e-05, + "loss": 0.1889, + "step": 8629 + }, + { + "epoch": 3.61, + "learning_rate": 1.2035426182316459e-05, + "loss": 0.2021, + "step": 8630 + }, + { + "epoch": 3.61, + "learning_rate": 1.2033766235733828e-05, + "loss": 0.2051, + "step": 8631 + }, + { + "epoch": 3.61, + "learning_rate": 1.2032106230692394e-05, + "loss": 0.2008, + "step": 8632 + }, + { + "epoch": 3.61, + "learning_rate": 1.2030446167239869e-05, + "loss": 0.171, + "step": 8633 + }, + { + "epoch": 3.61, + "learning_rate": 1.202878604542397e-05, + "loss": 0.2133, + "step": 8634 + }, + { + "epoch": 3.62, + "learning_rate": 1.2027125865292417e-05, + "loss": 0.1647, + "step": 8635 + }, + { + "epoch": 3.62, + "learning_rate": 1.202546562689293e-05, + "loss": 0.1962, + "step": 8636 + }, + { + "epoch": 3.62, + "learning_rate": 1.202380533027323e-05, + "loss": 0.2258, + "step": 8637 + }, + { + "epoch": 3.62, + "learning_rate": 1.2022144975481043e-05, + "loss": 0.1678, + "step": 8638 + }, + { + "epoch": 3.62, + "learning_rate": 1.202048456256409e-05, + "loss": 0.1997, + "step": 8639 + }, + { + "epoch": 3.62, + "learning_rate": 1.2018824091570103e-05, + "loss": 0.1733, + "step": 8640 + }, + { + "epoch": 3.62, + "learning_rate": 1.201716356254681e-05, + "loss": 0.1818, + "step": 8641 + }, + { + "epoch": 3.62, + "learning_rate": 1.201550297554194e-05, + "loss": 0.1849, + "step": 8642 + }, + { + "epoch": 3.62, + "learning_rate": 1.2013842330603226e-05, + "loss": 0.1728, + "step": 8643 + }, + { + "epoch": 3.62, + "learning_rate": 1.20121816277784e-05, + "loss": 0.1764, + "step": 8644 + }, + { + "epoch": 3.62, + "learning_rate": 1.20105208671152e-05, + "loss": 0.2053, + "step": 8645 + }, + { + "epoch": 3.62, + "learning_rate": 1.2008860048661362e-05, + "loss": 0.1708, + "step": 8646 + }, + { + "epoch": 3.62, + "learning_rate": 1.200719917246462e-05, + "loss": 0.1796, + "step": 8647 + }, + { + "epoch": 3.62, + "learning_rate": 1.2005538238572724e-05, + "loss": 0.1694, + "step": 8648 + }, + { + "epoch": 3.62, + "learning_rate": 1.2003877247033411e-05, + "loss": 0.1919, + "step": 8649 + }, + { + "epoch": 3.62, + "learning_rate": 1.200221619789442e-05, + "loss": 0.242, + "step": 8650 + }, + { + "epoch": 3.62, + "learning_rate": 1.2000555091203507e-05, + "loss": 0.1495, + "step": 8651 + }, + { + "epoch": 3.62, + "learning_rate": 1.1998893927008411e-05, + "loss": 0.1781, + "step": 8652 + }, + { + "epoch": 3.62, + "learning_rate": 1.199723270535688e-05, + "loss": 0.1978, + "step": 8653 + }, + { + "epoch": 3.62, + "learning_rate": 1.1995571426296671e-05, + "loss": 0.2193, + "step": 8654 + }, + { + "epoch": 3.62, + "learning_rate": 1.1993910089875531e-05, + "loss": 0.1623, + "step": 8655 + }, + { + "epoch": 3.62, + "learning_rate": 1.1992248696141215e-05, + "loss": 0.2132, + "step": 8656 + }, + { + "epoch": 3.62, + "learning_rate": 1.1990587245141477e-05, + "loss": 0.1865, + "step": 8657 + }, + { + "epoch": 3.62, + "learning_rate": 1.1988925736924078e-05, + "loss": 0.1778, + "step": 8658 + }, + { + "epoch": 3.63, + "learning_rate": 1.198726417153677e-05, + "loss": 0.1642, + "step": 8659 + }, + { + "epoch": 3.63, + "learning_rate": 1.1985602549027318e-05, + "loss": 0.1997, + "step": 8660 + }, + { + "epoch": 3.63, + "learning_rate": 1.1983940869443484e-05, + "loss": 0.2021, + "step": 8661 + }, + { + "epoch": 3.63, + "learning_rate": 1.1982279132833028e-05, + "loss": 0.1802, + "step": 8662 + }, + { + "epoch": 3.63, + "learning_rate": 1.1980617339243719e-05, + "loss": 0.215, + "step": 8663 + }, + { + "epoch": 3.63, + "learning_rate": 1.1978955488723323e-05, + "loss": 0.1522, + "step": 8664 + }, + { + "epoch": 3.63, + "learning_rate": 1.1977293581319605e-05, + "loss": 0.1754, + "step": 8665 + }, + { + "epoch": 3.63, + "learning_rate": 1.1975631617080339e-05, + "loss": 0.1967, + "step": 8666 + }, + { + "epoch": 3.63, + "learning_rate": 1.1973969596053295e-05, + "loss": 0.1964, + "step": 8667 + }, + { + "epoch": 3.63, + "learning_rate": 1.1972307518286248e-05, + "loss": 0.163, + "step": 8668 + }, + { + "epoch": 3.63, + "learning_rate": 1.1970645383826971e-05, + "loss": 0.2226, + "step": 8669 + }, + { + "epoch": 3.63, + "learning_rate": 1.1968983192723238e-05, + "loss": 0.1526, + "step": 8670 + }, + { + "epoch": 3.63, + "learning_rate": 1.1967320945022836e-05, + "loss": 0.1703, + "step": 8671 + }, + { + "epoch": 3.63, + "learning_rate": 1.1965658640773535e-05, + "loss": 0.1807, + "step": 8672 + }, + { + "epoch": 3.63, + "learning_rate": 1.196399628002312e-05, + "loss": 0.1969, + "step": 8673 + }, + { + "epoch": 3.63, + "learning_rate": 1.1962333862819379e-05, + "loss": 0.1891, + "step": 8674 + }, + { + "epoch": 3.63, + "learning_rate": 1.1960671389210087e-05, + "loss": 0.1824, + "step": 8675 + }, + { + "epoch": 3.63, + "learning_rate": 1.195900885924304e-05, + "loss": 0.1831, + "step": 8676 + }, + { + "epoch": 3.63, + "learning_rate": 1.1957346272966019e-05, + "loss": 0.1629, + "step": 8677 + }, + { + "epoch": 3.63, + "learning_rate": 1.195568363042682e-05, + "loss": 0.187, + "step": 8678 + }, + { + "epoch": 3.63, + "learning_rate": 1.1954020931673229e-05, + "loss": 0.1698, + "step": 8679 + }, + { + "epoch": 3.63, + "learning_rate": 1.1952358176753038e-05, + "loss": 0.1691, + "step": 8680 + }, + { + "epoch": 3.63, + "learning_rate": 1.1950695365714046e-05, + "loss": 0.1688, + "step": 8681 + }, + { + "epoch": 3.63, + "learning_rate": 1.1949032498604047e-05, + "loss": 0.232, + "step": 8682 + }, + { + "epoch": 3.64, + "learning_rate": 1.1947369575470837e-05, + "loss": 0.1872, + "step": 8683 + }, + { + "epoch": 3.64, + "learning_rate": 1.1945706596362216e-05, + "loss": 0.1597, + "step": 8684 + }, + { + "epoch": 3.64, + "learning_rate": 1.1944043561325989e-05, + "loss": 0.2131, + "step": 8685 + }, + { + "epoch": 3.64, + "learning_rate": 1.194238047040995e-05, + "loss": 0.1646, + "step": 8686 + }, + { + "epoch": 3.64, + "learning_rate": 1.194071732366191e-05, + "loss": 0.1981, + "step": 8687 + }, + { + "epoch": 3.64, + "learning_rate": 1.1939054121129677e-05, + "loss": 0.1612, + "step": 8688 + }, + { + "epoch": 3.64, + "learning_rate": 1.1937390862861047e-05, + "loss": 0.1734, + "step": 8689 + }, + { + "epoch": 3.64, + "learning_rate": 1.1935727548903842e-05, + "loss": 0.1866, + "step": 8690 + }, + { + "epoch": 3.64, + "learning_rate": 1.1934064179305861e-05, + "loss": 0.1663, + "step": 8691 + }, + { + "epoch": 3.64, + "learning_rate": 1.1932400754114925e-05, + "loss": 0.1951, + "step": 8692 + }, + { + "epoch": 3.64, + "learning_rate": 1.1930737273378842e-05, + "loss": 0.1756, + "step": 8693 + }, + { + "epoch": 3.64, + "learning_rate": 1.1929073737145432e-05, + "loss": 0.1713, + "step": 8694 + }, + { + "epoch": 3.64, + "learning_rate": 1.1927410145462507e-05, + "loss": 0.2078, + "step": 8695 + }, + { + "epoch": 3.64, + "learning_rate": 1.1925746498377888e-05, + "loss": 0.1786, + "step": 8696 + }, + { + "epoch": 3.64, + "learning_rate": 1.1924082795939394e-05, + "loss": 0.1771, + "step": 8697 + }, + { + "epoch": 3.64, + "learning_rate": 1.192241903819485e-05, + "loss": 0.1918, + "step": 8698 + }, + { + "epoch": 3.64, + "learning_rate": 1.1920755225192074e-05, + "loss": 0.167, + "step": 8699 + }, + { + "epoch": 3.64, + "learning_rate": 1.1919091356978894e-05, + "loss": 0.1738, + "step": 8700 + }, + { + "epoch": 3.64, + "learning_rate": 1.1917427433603138e-05, + "loss": 0.203, + "step": 8701 + }, + { + "epoch": 3.64, + "learning_rate": 1.1915763455112633e-05, + "loss": 0.2271, + "step": 8702 + }, + { + "epoch": 3.64, + "learning_rate": 1.1914099421555203e-05, + "loss": 0.1783, + "step": 8703 + }, + { + "epoch": 3.64, + "learning_rate": 1.1912435332978684e-05, + "loss": 0.1941, + "step": 8704 + }, + { + "epoch": 3.64, + "learning_rate": 1.1910771189430913e-05, + "loss": 0.2174, + "step": 8705 + }, + { + "epoch": 3.64, + "learning_rate": 1.1909106990959716e-05, + "loss": 0.1742, + "step": 8706 + }, + { + "epoch": 3.65, + "learning_rate": 1.1907442737612932e-05, + "loss": 0.1669, + "step": 8707 + }, + { + "epoch": 3.65, + "learning_rate": 1.1905778429438401e-05, + "loss": 0.2056, + "step": 8708 + }, + { + "epoch": 3.65, + "learning_rate": 1.190411406648396e-05, + "loss": 0.2232, + "step": 8709 + }, + { + "epoch": 3.65, + "learning_rate": 1.190244964879745e-05, + "loss": 0.1932, + "step": 8710 + }, + { + "epoch": 3.65, + "learning_rate": 1.190078517642671e-05, + "loss": 0.1979, + "step": 8711 + }, + { + "epoch": 3.65, + "learning_rate": 1.189912064941959e-05, + "loss": 0.1841, + "step": 8712 + }, + { + "epoch": 3.65, + "learning_rate": 1.1897456067823934e-05, + "loss": 0.1931, + "step": 8713 + }, + { + "epoch": 3.65, + "learning_rate": 1.1895791431687585e-05, + "loss": 0.1672, + "step": 8714 + }, + { + "epoch": 3.65, + "learning_rate": 1.1894126741058393e-05, + "loss": 0.1963, + "step": 8715 + }, + { + "epoch": 3.65, + "learning_rate": 1.189246199598421e-05, + "loss": 0.2083, + "step": 8716 + }, + { + "epoch": 3.65, + "learning_rate": 1.1890797196512884e-05, + "loss": 0.218, + "step": 8717 + }, + { + "epoch": 3.65, + "learning_rate": 1.1889132342692272e-05, + "loss": 0.1758, + "step": 8718 + }, + { + "epoch": 3.65, + "learning_rate": 1.1887467434570227e-05, + "loss": 0.1998, + "step": 8719 + }, + { + "epoch": 3.65, + "learning_rate": 1.1885802472194606e-05, + "loss": 0.1669, + "step": 8720 + }, + { + "epoch": 3.65, + "learning_rate": 1.1884137455613266e-05, + "loss": 0.1861, + "step": 8721 + }, + { + "epoch": 3.65, + "learning_rate": 1.1882472384874067e-05, + "loss": 0.2202, + "step": 8722 + }, + { + "epoch": 3.65, + "learning_rate": 1.1880807260024867e-05, + "loss": 0.1837, + "step": 8723 + }, + { + "epoch": 3.65, + "learning_rate": 1.1879142081113535e-05, + "loss": 0.1654, + "step": 8724 + }, + { + "epoch": 3.65, + "learning_rate": 1.187747684818793e-05, + "loss": 0.1711, + "step": 8725 + }, + { + "epoch": 3.65, + "learning_rate": 1.1875811561295921e-05, + "loss": 0.1968, + "step": 8726 + }, + { + "epoch": 3.65, + "learning_rate": 1.187414622048537e-05, + "loss": 0.1816, + "step": 8727 + }, + { + "epoch": 3.65, + "learning_rate": 1.1872480825804149e-05, + "loss": 0.1843, + "step": 8728 + }, + { + "epoch": 3.65, + "learning_rate": 1.1870815377300131e-05, + "loss": 0.171, + "step": 8729 + }, + { + "epoch": 3.66, + "learning_rate": 1.1869149875021184e-05, + "loss": 0.1602, + "step": 8730 + }, + { + "epoch": 3.66, + "learning_rate": 1.186748431901518e-05, + "loss": 0.1338, + "step": 8731 + }, + { + "epoch": 3.66, + "learning_rate": 1.1865818709330001e-05, + "loss": 0.211, + "step": 8732 + }, + { + "epoch": 3.66, + "learning_rate": 1.1864153046013515e-05, + "loss": 0.2019, + "step": 8733 + }, + { + "epoch": 3.66, + "learning_rate": 1.1862487329113606e-05, + "loss": 0.2191, + "step": 8734 + }, + { + "epoch": 3.66, + "learning_rate": 1.1860821558678152e-05, + "loss": 0.1587, + "step": 8735 + }, + { + "epoch": 3.66, + "learning_rate": 1.1859155734755034e-05, + "loss": 0.22, + "step": 8736 + }, + { + "epoch": 3.66, + "learning_rate": 1.1857489857392133e-05, + "loss": 0.1619, + "step": 8737 + }, + { + "epoch": 3.66, + "learning_rate": 1.1855823926637333e-05, + "loss": 0.1612, + "step": 8738 + }, + { + "epoch": 3.66, + "learning_rate": 1.1854157942538522e-05, + "loss": 0.1793, + "step": 8739 + }, + { + "epoch": 3.66, + "learning_rate": 1.1852491905143586e-05, + "loss": 0.1781, + "step": 8740 + }, + { + "epoch": 3.66, + "learning_rate": 1.1850825814500413e-05, + "loss": 0.1956, + "step": 8741 + }, + { + "epoch": 3.66, + "learning_rate": 1.1849159670656897e-05, + "loss": 0.1916, + "step": 8742 + }, + { + "epoch": 3.66, + "learning_rate": 1.1847493473660925e-05, + "loss": 0.1949, + "step": 8743 + }, + { + "epoch": 3.66, + "learning_rate": 1.1845827223560394e-05, + "loss": 0.1847, + "step": 8744 + }, + { + "epoch": 3.66, + "learning_rate": 1.1844160920403197e-05, + "loss": 0.1929, + "step": 8745 + }, + { + "epoch": 3.66, + "learning_rate": 1.1842494564237231e-05, + "loss": 0.131, + "step": 8746 + }, + { + "epoch": 3.66, + "learning_rate": 1.1840828155110395e-05, + "loss": 0.1856, + "step": 8747 + }, + { + "epoch": 3.66, + "learning_rate": 1.1839161693070584e-05, + "loss": 0.1538, + "step": 8748 + }, + { + "epoch": 3.66, + "learning_rate": 1.1837495178165706e-05, + "loss": 0.1963, + "step": 8749 + }, + { + "epoch": 3.66, + "learning_rate": 1.1835828610443656e-05, + "loss": 0.1969, + "step": 8750 + }, + { + "epoch": 3.66, + "learning_rate": 1.1834161989952345e-05, + "loss": 0.1978, + "step": 8751 + }, + { + "epoch": 3.66, + "learning_rate": 1.1832495316739672e-05, + "loss": 0.2095, + "step": 8752 + }, + { + "epoch": 3.66, + "learning_rate": 1.1830828590853548e-05, + "loss": 0.1731, + "step": 8753 + }, + { + "epoch": 3.67, + "learning_rate": 1.1829161812341883e-05, + "loss": 0.1978, + "step": 8754 + }, + { + "epoch": 3.67, + "learning_rate": 1.182749498125258e-05, + "loss": 0.1767, + "step": 8755 + }, + { + "epoch": 3.67, + "learning_rate": 1.1825828097633562e-05, + "loss": 0.1752, + "step": 8756 + }, + { + "epoch": 3.67, + "learning_rate": 1.1824161161532731e-05, + "loss": 0.1937, + "step": 8757 + }, + { + "epoch": 3.67, + "learning_rate": 1.1822494172998008e-05, + "loss": 0.1824, + "step": 8758 + }, + { + "epoch": 3.67, + "learning_rate": 1.1820827132077306e-05, + "loss": 0.1865, + "step": 8759 + }, + { + "epoch": 3.67, + "learning_rate": 1.1819160038818547e-05, + "loss": 0.1787, + "step": 8760 + }, + { + "epoch": 3.67, + "learning_rate": 1.1817492893269645e-05, + "loss": 0.1637, + "step": 8761 + }, + { + "epoch": 3.67, + "learning_rate": 1.181582569547852e-05, + "loss": 0.2163, + "step": 8762 + }, + { + "epoch": 3.67, + "learning_rate": 1.18141584454931e-05, + "loss": 0.1649, + "step": 8763 + }, + { + "epoch": 3.67, + "learning_rate": 1.1812491143361305e-05, + "loss": 0.2028, + "step": 8764 + }, + { + "epoch": 3.67, + "learning_rate": 1.1810823789131059e-05, + "loss": 0.1657, + "step": 8765 + }, + { + "epoch": 3.67, + "learning_rate": 1.1809156382850291e-05, + "loss": 0.1668, + "step": 8766 + }, + { + "epoch": 3.67, + "learning_rate": 1.1807488924566929e-05, + "loss": 0.1856, + "step": 8767 + }, + { + "epoch": 3.67, + "learning_rate": 1.1805821414328897e-05, + "loss": 0.1946, + "step": 8768 + }, + { + "epoch": 3.67, + "learning_rate": 1.1804153852184137e-05, + "loss": 0.2073, + "step": 8769 + }, + { + "epoch": 3.67, + "learning_rate": 1.1802486238180575e-05, + "loss": 0.1951, + "step": 8770 + }, + { + "epoch": 3.67, + "learning_rate": 1.1800818572366144e-05, + "loss": 0.1153, + "step": 8771 + }, + { + "epoch": 3.67, + "learning_rate": 1.179915085478878e-05, + "loss": 0.1781, + "step": 8772 + }, + { + "epoch": 3.67, + "learning_rate": 1.1797483085496424e-05, + "loss": 0.2328, + "step": 8773 + }, + { + "epoch": 3.67, + "learning_rate": 1.1795815264537008e-05, + "loss": 0.2347, + "step": 8774 + }, + { + "epoch": 3.67, + "learning_rate": 1.1794147391958477e-05, + "loss": 0.2036, + "step": 8775 + }, + { + "epoch": 3.67, + "learning_rate": 1.1792479467808774e-05, + "loss": 0.205, + "step": 8776 + }, + { + "epoch": 3.67, + "learning_rate": 1.1790811492135837e-05, + "loss": 0.1933, + "step": 8777 + }, + { + "epoch": 3.68, + "learning_rate": 1.1789143464987611e-05, + "loss": 0.1803, + "step": 8778 + }, + { + "epoch": 3.68, + "learning_rate": 1.1787475386412048e-05, + "loss": 0.2011, + "step": 8779 + }, + { + "epoch": 3.68, + "learning_rate": 1.178580725645709e-05, + "loss": 0.1484, + "step": 8780 + }, + { + "epoch": 3.68, + "learning_rate": 1.1784139075170686e-05, + "loss": 0.1619, + "step": 8781 + }, + { + "epoch": 3.68, + "learning_rate": 1.1782470842600785e-05, + "loss": 0.1636, + "step": 8782 + }, + { + "epoch": 3.68, + "learning_rate": 1.1780802558795346e-05, + "loss": 0.166, + "step": 8783 + }, + { + "epoch": 3.68, + "learning_rate": 1.1779134223802315e-05, + "loss": 0.1686, + "step": 8784 + }, + { + "epoch": 3.68, + "learning_rate": 1.177746583766965e-05, + "loss": 0.224, + "step": 8785 + }, + { + "epoch": 3.68, + "learning_rate": 1.1775797400445307e-05, + "loss": 0.1735, + "step": 8786 + }, + { + "epoch": 3.68, + "learning_rate": 1.1774128912177245e-05, + "loss": 0.1761, + "step": 8787 + }, + { + "epoch": 3.68, + "learning_rate": 1.177246037291342e-05, + "loss": 0.2169, + "step": 8788 + }, + { + "epoch": 3.68, + "learning_rate": 1.1770791782701794e-05, + "loss": 0.1702, + "step": 8789 + }, + { + "epoch": 3.68, + "learning_rate": 1.176912314159033e-05, + "loss": 0.1961, + "step": 8790 + }, + { + "epoch": 3.68, + "learning_rate": 1.1767454449626992e-05, + "loss": 0.2112, + "step": 8791 + }, + { + "epoch": 3.68, + "learning_rate": 1.1765785706859744e-05, + "loss": 0.2216, + "step": 8792 + }, + { + "epoch": 3.68, + "learning_rate": 1.1764116913336552e-05, + "loss": 0.2142, + "step": 8793 + }, + { + "epoch": 3.68, + "learning_rate": 1.1762448069105387e-05, + "loss": 0.2356, + "step": 8794 + }, + { + "epoch": 3.68, + "learning_rate": 1.1760779174214215e-05, + "loss": 0.1855, + "step": 8795 + }, + { + "epoch": 3.68, + "learning_rate": 1.1759110228711006e-05, + "loss": 0.2006, + "step": 8796 + }, + { + "epoch": 3.68, + "learning_rate": 1.1757441232643737e-05, + "loss": 0.2181, + "step": 8797 + }, + { + "epoch": 3.68, + "learning_rate": 1.1755772186060379e-05, + "loss": 0.1845, + "step": 8798 + }, + { + "epoch": 3.68, + "learning_rate": 1.1754103089008905e-05, + "loss": 0.1632, + "step": 8799 + }, + { + "epoch": 3.68, + "learning_rate": 1.1752433941537297e-05, + "loss": 0.1701, + "step": 8800 + }, + { + "epoch": 3.68, + "learning_rate": 1.175076474369353e-05, + "loss": 0.1762, + "step": 8801 + }, + { + "epoch": 3.69, + "learning_rate": 1.1749095495525584e-05, + "loss": 0.2207, + "step": 8802 + }, + { + "epoch": 3.69, + "learning_rate": 1.1747426197081443e-05, + "loss": 0.2055, + "step": 8803 + }, + { + "epoch": 3.69, + "learning_rate": 1.1745756848409083e-05, + "loss": 0.1968, + "step": 8804 + }, + { + "epoch": 3.69, + "learning_rate": 1.1744087449556492e-05, + "loss": 0.164, + "step": 8805 + }, + { + "epoch": 3.69, + "learning_rate": 1.1742418000571653e-05, + "loss": 0.2076, + "step": 8806 + }, + { + "epoch": 3.69, + "learning_rate": 1.1740748501502559e-05, + "loss": 0.1933, + "step": 8807 + }, + { + "epoch": 3.69, + "learning_rate": 1.1739078952397189e-05, + "loss": 0.1662, + "step": 8808 + }, + { + "epoch": 3.69, + "learning_rate": 1.1737409353303541e-05, + "loss": 0.1819, + "step": 8809 + }, + { + "epoch": 3.69, + "learning_rate": 1.1735739704269605e-05, + "loss": 0.1808, + "step": 8810 + }, + { + "epoch": 3.69, + "learning_rate": 1.1734070005343368e-05, + "loss": 0.184, + "step": 8811 + }, + { + "epoch": 3.69, + "learning_rate": 1.1732400256572827e-05, + "loss": 0.1854, + "step": 8812 + }, + { + "epoch": 3.69, + "learning_rate": 1.1730730458005981e-05, + "loss": 0.1776, + "step": 8813 + }, + { + "epoch": 3.69, + "learning_rate": 1.1729060609690824e-05, + "loss": 0.188, + "step": 8814 + }, + { + "epoch": 3.69, + "learning_rate": 1.172739071167535e-05, + "loss": 0.1515, + "step": 8815 + }, + { + "epoch": 3.69, + "learning_rate": 1.1725720764007564e-05, + "loss": 0.2115, + "step": 8816 + }, + { + "epoch": 3.69, + "learning_rate": 1.1724050766735469e-05, + "loss": 0.1856, + "step": 8817 + }, + { + "epoch": 3.69, + "learning_rate": 1.1722380719907063e-05, + "loss": 0.1764, + "step": 8818 + }, + { + "epoch": 3.69, + "learning_rate": 1.172071062357035e-05, + "loss": 0.1929, + "step": 8819 + }, + { + "epoch": 3.69, + "learning_rate": 1.171904047777334e-05, + "loss": 0.1946, + "step": 8820 + }, + { + "epoch": 3.69, + "learning_rate": 1.1717370282564034e-05, + "loss": 0.1833, + "step": 8821 + }, + { + "epoch": 3.69, + "learning_rate": 1.1715700037990443e-05, + "loss": 0.1939, + "step": 8822 + }, + { + "epoch": 3.69, + "learning_rate": 1.171402974410058e-05, + "loss": 0.167, + "step": 8823 + }, + { + "epoch": 3.69, + "learning_rate": 1.171235940094245e-05, + "loss": 0.223, + "step": 8824 + }, + { + "epoch": 3.69, + "learning_rate": 1.1710689008564068e-05, + "loss": 0.1941, + "step": 8825 + }, + { + "epoch": 3.7, + "learning_rate": 1.1709018567013455e-05, + "loss": 0.1697, + "step": 8826 + }, + { + "epoch": 3.7, + "learning_rate": 1.1707348076338616e-05, + "loss": 0.1743, + "step": 8827 + }, + { + "epoch": 3.7, + "learning_rate": 1.170567753658757e-05, + "loss": 0.2077, + "step": 8828 + }, + { + "epoch": 3.7, + "learning_rate": 1.1704006947808338e-05, + "loss": 0.1682, + "step": 8829 + }, + { + "epoch": 3.7, + "learning_rate": 1.1702336310048936e-05, + "loss": 0.1975, + "step": 8830 + }, + { + "epoch": 3.7, + "learning_rate": 1.170066562335739e-05, + "loss": 0.1564, + "step": 8831 + }, + { + "epoch": 3.7, + "learning_rate": 1.1698994887781718e-05, + "loss": 0.1795, + "step": 8832 + }, + { + "epoch": 3.7, + "learning_rate": 1.1697324103369947e-05, + "loss": 0.1962, + "step": 8833 + }, + { + "epoch": 3.7, + "learning_rate": 1.1695653270170102e-05, + "loss": 0.2126, + "step": 8834 + }, + { + "epoch": 3.7, + "learning_rate": 1.1693982388230207e-05, + "loss": 0.2115, + "step": 8835 + }, + { + "epoch": 3.7, + "learning_rate": 1.1692311457598291e-05, + "loss": 0.2035, + "step": 8836 + }, + { + "epoch": 3.7, + "learning_rate": 1.1690640478322386e-05, + "loss": 0.1825, + "step": 8837 + }, + { + "epoch": 3.7, + "learning_rate": 1.168896945045052e-05, + "loss": 0.1644, + "step": 8838 + }, + { + "epoch": 3.7, + "learning_rate": 1.1687298374030724e-05, + "loss": 0.1886, + "step": 8839 + }, + { + "epoch": 3.7, + "learning_rate": 1.1685627249111034e-05, + "loss": 0.1961, + "step": 8840 + }, + { + "epoch": 3.7, + "learning_rate": 1.1683956075739487e-05, + "loss": 0.2219, + "step": 8841 + }, + { + "epoch": 3.7, + "learning_rate": 1.1682284853964114e-05, + "loss": 0.1937, + "step": 8842 + }, + { + "epoch": 3.7, + "learning_rate": 1.1680613583832956e-05, + "loss": 0.1764, + "step": 8843 + }, + { + "epoch": 3.7, + "learning_rate": 1.1678942265394055e-05, + "loss": 0.1845, + "step": 8844 + }, + { + "epoch": 3.7, + "learning_rate": 1.1677270898695445e-05, + "loss": 0.1797, + "step": 8845 + }, + { + "epoch": 3.7, + "learning_rate": 1.1675599483785171e-05, + "loss": 0.1832, + "step": 8846 + }, + { + "epoch": 3.7, + "learning_rate": 1.167392802071128e-05, + "loss": 0.2014, + "step": 8847 + }, + { + "epoch": 3.7, + "learning_rate": 1.1672256509521814e-05, + "loss": 0.2086, + "step": 8848 + }, + { + "epoch": 3.7, + "learning_rate": 1.1670584950264816e-05, + "loss": 0.1592, + "step": 8849 + }, + { + "epoch": 3.71, + "learning_rate": 1.1668913342988335e-05, + "loss": 0.2036, + "step": 8850 + }, + { + "epoch": 3.71, + "learning_rate": 1.1667241687740422e-05, + "loss": 0.1597, + "step": 8851 + }, + { + "epoch": 3.71, + "learning_rate": 1.1665569984569127e-05, + "loss": 0.2258, + "step": 8852 + }, + { + "epoch": 3.71, + "learning_rate": 1.1663898233522499e-05, + "loss": 0.1592, + "step": 8853 + }, + { + "epoch": 3.71, + "learning_rate": 1.1662226434648595e-05, + "loss": 0.1749, + "step": 8854 + }, + { + "epoch": 3.71, + "learning_rate": 1.1660554587995465e-05, + "loss": 0.1716, + "step": 8855 + }, + { + "epoch": 3.71, + "learning_rate": 1.1658882693611169e-05, + "loss": 0.1531, + "step": 8856 + }, + { + "epoch": 3.71, + "learning_rate": 1.165721075154376e-05, + "loss": 0.1848, + "step": 8857 + }, + { + "epoch": 3.71, + "learning_rate": 1.16555387618413e-05, + "loss": 0.1777, + "step": 8858 + }, + { + "epoch": 3.71, + "learning_rate": 1.1653866724551848e-05, + "loss": 0.2096, + "step": 8859 + }, + { + "epoch": 3.71, + "learning_rate": 1.1652194639723465e-05, + "loss": 0.1777, + "step": 8860 + }, + { + "epoch": 3.71, + "learning_rate": 1.1650522507404211e-05, + "loss": 0.2106, + "step": 8861 + }, + { + "epoch": 3.71, + "learning_rate": 1.1648850327642154e-05, + "loss": 0.1917, + "step": 8862 + }, + { + "epoch": 3.71, + "learning_rate": 1.1647178100485354e-05, + "loss": 0.1864, + "step": 8863 + }, + { + "epoch": 3.71, + "learning_rate": 1.1645505825981884e-05, + "loss": 0.1869, + "step": 8864 + }, + { + "epoch": 3.71, + "learning_rate": 1.1643833504179811e-05, + "loss": 0.1971, + "step": 8865 + }, + { + "epoch": 3.71, + "learning_rate": 1.1642161135127198e-05, + "loss": 0.1678, + "step": 8866 + }, + { + "epoch": 3.71, + "learning_rate": 1.1640488718872124e-05, + "loss": 0.1875, + "step": 8867 + }, + { + "epoch": 3.71, + "learning_rate": 1.1638816255462657e-05, + "loss": 0.1905, + "step": 8868 + }, + { + "epoch": 3.71, + "learning_rate": 1.1637143744946871e-05, + "loss": 0.2219, + "step": 8869 + }, + { + "epoch": 3.71, + "learning_rate": 1.163547118737284e-05, + "loss": 0.1915, + "step": 8870 + }, + { + "epoch": 3.71, + "learning_rate": 1.1633798582788642e-05, + "loss": 0.1979, + "step": 8871 + }, + { + "epoch": 3.71, + "learning_rate": 1.1632125931242356e-05, + "loss": 0.2147, + "step": 8872 + }, + { + "epoch": 3.71, + "learning_rate": 1.1630453232782053e-05, + "loss": 0.2346, + "step": 8873 + }, + { + "epoch": 3.72, + "learning_rate": 1.162878048745582e-05, + "loss": 0.1713, + "step": 8874 + }, + { + "epoch": 3.72, + "learning_rate": 1.1627107695311738e-05, + "loss": 0.2076, + "step": 8875 + }, + { + "epoch": 3.72, + "learning_rate": 1.1625434856397891e-05, + "loss": 0.1693, + "step": 8876 + }, + { + "epoch": 3.72, + "learning_rate": 1.1623761970762359e-05, + "loss": 0.1629, + "step": 8877 + }, + { + "epoch": 3.72, + "learning_rate": 1.1622089038453232e-05, + "loss": 0.1809, + "step": 8878 + }, + { + "epoch": 3.72, + "learning_rate": 1.1620416059518593e-05, + "loss": 0.1831, + "step": 8879 + }, + { + "epoch": 3.72, + "learning_rate": 1.1618743034006532e-05, + "loss": 0.1921, + "step": 8880 + }, + { + "epoch": 3.72, + "learning_rate": 1.1617069961965141e-05, + "loss": 0.1877, + "step": 8881 + }, + { + "epoch": 3.72, + "learning_rate": 1.161539684344251e-05, + "loss": 0.2086, + "step": 8882 + }, + { + "epoch": 3.72, + "learning_rate": 1.1613723678486728e-05, + "loss": 0.1888, + "step": 8883 + }, + { + "epoch": 3.72, + "learning_rate": 1.161205046714589e-05, + "loss": 0.2187, + "step": 8884 + }, + { + "epoch": 3.72, + "learning_rate": 1.1610377209468095e-05, + "loss": 0.1716, + "step": 8885 + }, + { + "epoch": 3.72, + "learning_rate": 1.1608703905501432e-05, + "loss": 0.2116, + "step": 8886 + }, + { + "epoch": 3.72, + "learning_rate": 1.1607030555294005e-05, + "loss": 0.2013, + "step": 8887 + }, + { + "epoch": 3.72, + "learning_rate": 1.1605357158893913e-05, + "loss": 0.1809, + "step": 8888 + }, + { + "epoch": 3.72, + "learning_rate": 1.160368371634925e-05, + "loss": 0.196, + "step": 8889 + }, + { + "epoch": 3.72, + "learning_rate": 1.1602010227708122e-05, + "loss": 0.2002, + "step": 8890 + }, + { + "epoch": 3.72, + "learning_rate": 1.1600336693018635e-05, + "loss": 0.1671, + "step": 8891 + }, + { + "epoch": 3.72, + "learning_rate": 1.1598663112328888e-05, + "loss": 0.1837, + "step": 8892 + }, + { + "epoch": 3.72, + "learning_rate": 1.1596989485686987e-05, + "loss": 0.202, + "step": 8893 + }, + { + "epoch": 3.72, + "learning_rate": 1.1595315813141041e-05, + "loss": 0.1697, + "step": 8894 + }, + { + "epoch": 3.72, + "learning_rate": 1.1593642094739161e-05, + "loss": 0.1923, + "step": 8895 + }, + { + "epoch": 3.72, + "learning_rate": 1.159196833052945e-05, + "loss": 0.166, + "step": 8896 + }, + { + "epoch": 3.72, + "learning_rate": 1.1590294520560022e-05, + "loss": 0.2166, + "step": 8897 + }, + { + "epoch": 3.73, + "learning_rate": 1.1588620664878989e-05, + "loss": 0.1934, + "step": 8898 + }, + { + "epoch": 3.73, + "learning_rate": 1.1586946763534465e-05, + "loss": 0.1996, + "step": 8899 + }, + { + "epoch": 3.73, + "learning_rate": 1.1585272816574565e-05, + "loss": 0.1903, + "step": 8900 + }, + { + "epoch": 3.73, + "learning_rate": 1.1583598824047404e-05, + "loss": 0.1923, + "step": 8901 + }, + { + "epoch": 3.73, + "learning_rate": 1.1581924786001102e-05, + "loss": 0.1939, + "step": 8902 + }, + { + "epoch": 3.73, + "learning_rate": 1.1580250702483773e-05, + "loss": 0.1808, + "step": 8903 + }, + { + "epoch": 3.73, + "learning_rate": 1.1578576573543541e-05, + "loss": 0.1719, + "step": 8904 + }, + { + "epoch": 3.73, + "learning_rate": 1.1576902399228527e-05, + "loss": 0.1946, + "step": 8905 + }, + { + "epoch": 3.73, + "learning_rate": 1.1575228179586855e-05, + "loss": 0.1857, + "step": 8906 + }, + { + "epoch": 3.73, + "learning_rate": 1.1573553914666643e-05, + "loss": 0.1814, + "step": 8907 + }, + { + "epoch": 3.73, + "learning_rate": 1.157187960451602e-05, + "loss": 0.1515, + "step": 8908 + }, + { + "epoch": 3.73, + "learning_rate": 1.1570205249183115e-05, + "loss": 0.1651, + "step": 8909 + }, + { + "epoch": 3.73, + "learning_rate": 1.156853084871605e-05, + "loss": 0.1432, + "step": 8910 + }, + { + "epoch": 3.73, + "learning_rate": 1.156685640316296e-05, + "loss": 0.1824, + "step": 8911 + }, + { + "epoch": 3.73, + "learning_rate": 1.1565181912571975e-05, + "loss": 0.1531, + "step": 8912 + }, + { + "epoch": 3.73, + "learning_rate": 1.1563507376991222e-05, + "loss": 0.2062, + "step": 8913 + }, + { + "epoch": 3.73, + "learning_rate": 1.1561832796468837e-05, + "loss": 0.209, + "step": 8914 + }, + { + "epoch": 3.73, + "learning_rate": 1.1560158171052956e-05, + "loss": 0.1206, + "step": 8915 + }, + { + "epoch": 3.73, + "learning_rate": 1.1558483500791713e-05, + "loss": 0.1754, + "step": 8916 + }, + { + "epoch": 3.73, + "learning_rate": 1.1556808785733242e-05, + "loss": 0.2223, + "step": 8917 + }, + { + "epoch": 3.73, + "learning_rate": 1.1555134025925685e-05, + "loss": 0.1631, + "step": 8918 + }, + { + "epoch": 3.73, + "learning_rate": 1.1553459221417182e-05, + "loss": 0.1875, + "step": 8919 + }, + { + "epoch": 3.73, + "learning_rate": 1.155178437225587e-05, + "loss": 0.2032, + "step": 8920 + }, + { + "epoch": 3.73, + "learning_rate": 1.1550109478489896e-05, + "loss": 0.1785, + "step": 8921 + }, + { + "epoch": 3.74, + "learning_rate": 1.1548434540167399e-05, + "loss": 0.1759, + "step": 8922 + }, + { + "epoch": 3.74, + "learning_rate": 1.1546759557336524e-05, + "loss": 0.2091, + "step": 8923 + }, + { + "epoch": 3.74, + "learning_rate": 1.1545084530045419e-05, + "loss": 0.211, + "step": 8924 + }, + { + "epoch": 3.74, + "learning_rate": 1.154340945834223e-05, + "loss": 0.2067, + "step": 8925 + }, + { + "epoch": 3.74, + "learning_rate": 1.1541734342275106e-05, + "loss": 0.2118, + "step": 8926 + }, + { + "epoch": 3.74, + "learning_rate": 1.1540059181892197e-05, + "loss": 0.1759, + "step": 8927 + }, + { + "epoch": 3.74, + "learning_rate": 1.1538383977241654e-05, + "loss": 0.179, + "step": 8928 + }, + { + "epoch": 3.74, + "learning_rate": 1.1536708728371628e-05, + "loss": 0.1841, + "step": 8929 + }, + { + "epoch": 3.74, + "learning_rate": 1.1535033435330272e-05, + "loss": 0.1893, + "step": 8930 + }, + { + "epoch": 3.74, + "learning_rate": 1.1533358098165744e-05, + "loss": 0.2098, + "step": 8931 + }, + { + "epoch": 3.74, + "learning_rate": 1.1531682716926199e-05, + "loss": 0.1792, + "step": 8932 + }, + { + "epoch": 3.74, + "learning_rate": 1.153000729165979e-05, + "loss": 0.1708, + "step": 8933 + }, + { + "epoch": 3.74, + "learning_rate": 1.1528331822414681e-05, + "loss": 0.1958, + "step": 8934 + }, + { + "epoch": 3.74, + "learning_rate": 1.1526656309239028e-05, + "loss": 0.1662, + "step": 8935 + }, + { + "epoch": 3.74, + "learning_rate": 1.1524980752180996e-05, + "loss": 0.208, + "step": 8936 + }, + { + "epoch": 3.74, + "learning_rate": 1.1523305151288747e-05, + "loss": 0.173, + "step": 8937 + }, + { + "epoch": 3.74, + "learning_rate": 1.1521629506610439e-05, + "loss": 0.216, + "step": 8938 + }, + { + "epoch": 3.74, + "learning_rate": 1.1519953818194246e-05, + "loss": 0.1837, + "step": 8939 + }, + { + "epoch": 3.74, + "learning_rate": 1.1518278086088328e-05, + "loss": 0.1763, + "step": 8940 + }, + { + "epoch": 3.74, + "learning_rate": 1.1516602310340852e-05, + "loss": 0.1977, + "step": 8941 + }, + { + "epoch": 3.74, + "learning_rate": 1.151492649099999e-05, + "loss": 0.2517, + "step": 8942 + }, + { + "epoch": 3.74, + "learning_rate": 1.151325062811391e-05, + "loss": 0.234, + "step": 8943 + }, + { + "epoch": 3.74, + "learning_rate": 1.1511574721730781e-05, + "loss": 0.1756, + "step": 8944 + }, + { + "epoch": 3.75, + "learning_rate": 1.150989877189878e-05, + "loss": 0.1934, + "step": 8945 + }, + { + "epoch": 3.75, + "learning_rate": 1.1508222778666078e-05, + "loss": 0.1277, + "step": 8946 + }, + { + "epoch": 3.75, + "learning_rate": 1.1506546742080854e-05, + "loss": 0.1748, + "step": 8947 + }, + { + "epoch": 3.75, + "learning_rate": 1.1504870662191275e-05, + "loss": 0.169, + "step": 8948 + }, + { + "epoch": 3.75, + "learning_rate": 1.1503194539045527e-05, + "loss": 0.1794, + "step": 8949 + }, + { + "epoch": 3.75, + "learning_rate": 1.1501518372691789e-05, + "loss": 0.134, + "step": 8950 + }, + { + "epoch": 3.75, + "learning_rate": 1.1499842163178234e-05, + "loss": 0.1981, + "step": 8951 + }, + { + "epoch": 3.75, + "learning_rate": 1.1498165910553048e-05, + "loss": 0.1996, + "step": 8952 + }, + { + "epoch": 3.75, + "learning_rate": 1.1496489614864412e-05, + "loss": 0.1395, + "step": 8953 + }, + { + "epoch": 3.75, + "learning_rate": 1.1494813276160511e-05, + "loss": 0.1912, + "step": 8954 + }, + { + "epoch": 3.75, + "learning_rate": 1.1493136894489527e-05, + "loss": 0.1953, + "step": 8955 + }, + { + "epoch": 3.75, + "learning_rate": 1.149146046989965e-05, + "loss": 0.1645, + "step": 8956 + }, + { + "epoch": 3.75, + "learning_rate": 1.1489784002439064e-05, + "loss": 0.2151, + "step": 8957 + }, + { + "epoch": 3.75, + "learning_rate": 1.1488107492155958e-05, + "loss": 0.171, + "step": 8958 + }, + { + "epoch": 3.75, + "learning_rate": 1.1486430939098525e-05, + "loss": 0.1875, + "step": 8959 + }, + { + "epoch": 3.75, + "learning_rate": 1.1484754343314952e-05, + "loss": 0.2094, + "step": 8960 + }, + { + "epoch": 3.75, + "learning_rate": 1.1483077704853431e-05, + "loss": 0.1987, + "step": 8961 + }, + { + "epoch": 3.75, + "learning_rate": 1.1481401023762162e-05, + "loss": 0.2137, + "step": 8962 + }, + { + "epoch": 3.75, + "learning_rate": 1.1479724300089335e-05, + "loss": 0.1704, + "step": 8963 + }, + { + "epoch": 3.75, + "learning_rate": 1.1478047533883143e-05, + "loss": 0.1562, + "step": 8964 + }, + { + "epoch": 3.75, + "learning_rate": 1.1476370725191788e-05, + "loss": 0.1986, + "step": 8965 + }, + { + "epoch": 3.75, + "learning_rate": 1.1474693874063467e-05, + "loss": 0.1907, + "step": 8966 + }, + { + "epoch": 3.75, + "learning_rate": 1.1473016980546377e-05, + "loss": 0.1945, + "step": 8967 + }, + { + "epoch": 3.75, + "learning_rate": 1.147134004468872e-05, + "loss": 0.2249, + "step": 8968 + }, + { + "epoch": 3.76, + "learning_rate": 1.1469663066538704e-05, + "loss": 0.1739, + "step": 8969 + }, + { + "epoch": 3.76, + "learning_rate": 1.1467986046144523e-05, + "loss": 0.1978, + "step": 8970 + }, + { + "epoch": 3.76, + "learning_rate": 1.1466308983554386e-05, + "loss": 0.1379, + "step": 8971 + }, + { + "epoch": 3.76, + "learning_rate": 1.1464631878816502e-05, + "loss": 0.1583, + "step": 8972 + }, + { + "epoch": 3.76, + "learning_rate": 1.1462954731979074e-05, + "loss": 0.1796, + "step": 8973 + }, + { + "epoch": 3.76, + "learning_rate": 1.1461277543090308e-05, + "loss": 0.1897, + "step": 8974 + }, + { + "epoch": 3.76, + "learning_rate": 1.1459600312198416e-05, + "loss": 0.2038, + "step": 8975 + }, + { + "epoch": 3.76, + "learning_rate": 1.1457923039351606e-05, + "loss": 0.1697, + "step": 8976 + }, + { + "epoch": 3.76, + "learning_rate": 1.1456245724598096e-05, + "loss": 0.1876, + "step": 8977 + }, + { + "epoch": 3.76, + "learning_rate": 1.1454568367986094e-05, + "loss": 0.19, + "step": 8978 + }, + { + "epoch": 3.76, + "learning_rate": 1.1452890969563813e-05, + "loss": 0.2207, + "step": 8979 + }, + { + "epoch": 3.76, + "learning_rate": 1.1451213529379472e-05, + "loss": 0.1888, + "step": 8980 + }, + { + "epoch": 3.76, + "learning_rate": 1.1449536047481287e-05, + "loss": 0.1905, + "step": 8981 + }, + { + "epoch": 3.76, + "learning_rate": 1.1447858523917472e-05, + "loss": 0.2155, + "step": 8982 + }, + { + "epoch": 3.76, + "learning_rate": 1.144618095873625e-05, + "loss": 0.2026, + "step": 8983 + }, + { + "epoch": 3.76, + "learning_rate": 1.144450335198584e-05, + "loss": 0.1414, + "step": 8984 + }, + { + "epoch": 3.76, + "learning_rate": 1.1442825703714464e-05, + "loss": 0.1805, + "step": 8985 + }, + { + "epoch": 3.76, + "learning_rate": 1.144114801397034e-05, + "loss": 0.1571, + "step": 8986 + }, + { + "epoch": 3.76, + "learning_rate": 1.14394702828017e-05, + "loss": 0.2133, + "step": 8987 + }, + { + "epoch": 3.76, + "learning_rate": 1.143779251025676e-05, + "loss": 0.1764, + "step": 8988 + }, + { + "epoch": 3.76, + "learning_rate": 1.1436114696383749e-05, + "loss": 0.2045, + "step": 8989 + }, + { + "epoch": 3.76, + "learning_rate": 1.14344368412309e-05, + "loss": 0.1676, + "step": 8990 + }, + { + "epoch": 3.76, + "learning_rate": 1.1432758944846433e-05, + "loss": 0.1914, + "step": 8991 + }, + { + "epoch": 3.76, + "learning_rate": 1.1431081007278584e-05, + "loss": 0.1964, + "step": 8992 + }, + { + "epoch": 3.77, + "learning_rate": 1.142940302857558e-05, + "loss": 0.1722, + "step": 8993 + }, + { + "epoch": 3.77, + "learning_rate": 1.1427725008785655e-05, + "loss": 0.173, + "step": 8994 + }, + { + "epoch": 3.77, + "learning_rate": 1.1426046947957041e-05, + "loss": 0.2282, + "step": 8995 + }, + { + "epoch": 3.77, + "learning_rate": 1.1424368846137975e-05, + "loss": 0.1957, + "step": 8996 + }, + { + "epoch": 3.77, + "learning_rate": 1.142269070337669e-05, + "loss": 0.1834, + "step": 8997 + }, + { + "epoch": 3.77, + "learning_rate": 1.1421012519721421e-05, + "loss": 0.1803, + "step": 8998 + }, + { + "epoch": 3.77, + "learning_rate": 1.141933429522041e-05, + "loss": 0.2323, + "step": 8999 + }, + { + "epoch": 3.77, + "learning_rate": 1.1417656029921895e-05, + "loss": 0.2027, + "step": 9000 + }, + { + "epoch": 3.77, + "learning_rate": 1.1415977723874112e-05, + "loss": 0.1768, + "step": 9001 + }, + { + "epoch": 3.77, + "learning_rate": 1.1414299377125308e-05, + "loss": 0.1486, + "step": 9002 + }, + { + "epoch": 3.77, + "learning_rate": 1.1412620989723724e-05, + "loss": 0.1703, + "step": 9003 + }, + { + "epoch": 3.77, + "learning_rate": 1.1410942561717601e-05, + "loss": 0.1961, + "step": 9004 + }, + { + "epoch": 3.77, + "learning_rate": 1.1409264093155189e-05, + "loss": 0.2018, + "step": 9005 + }, + { + "epoch": 3.77, + "learning_rate": 1.1407585584084726e-05, + "loss": 0.1878, + "step": 9006 + }, + { + "epoch": 3.77, + "learning_rate": 1.1405907034554473e-05, + "loss": 0.2096, + "step": 9007 + }, + { + "epoch": 3.77, + "learning_rate": 1.1404228444612664e-05, + "loss": 0.1901, + "step": 9008 + }, + { + "epoch": 3.77, + "learning_rate": 1.1402549814307555e-05, + "loss": 0.1547, + "step": 9009 + }, + { + "epoch": 3.77, + "learning_rate": 1.1400871143687398e-05, + "loss": 0.193, + "step": 9010 + }, + { + "epoch": 3.77, + "learning_rate": 1.1399192432800442e-05, + "loss": 0.1671, + "step": 9011 + }, + { + "epoch": 3.77, + "learning_rate": 1.1397513681694938e-05, + "loss": 0.1926, + "step": 9012 + }, + { + "epoch": 3.77, + "learning_rate": 1.1395834890419147e-05, + "loss": 0.1855, + "step": 9013 + }, + { + "epoch": 3.77, + "learning_rate": 1.139415605902132e-05, + "loss": 0.1667, + "step": 9014 + }, + { + "epoch": 3.77, + "learning_rate": 1.1392477187549714e-05, + "loss": 0.2235, + "step": 9015 + }, + { + "epoch": 3.77, + "learning_rate": 1.1390798276052587e-05, + "loss": 0.206, + "step": 9016 + }, + { + "epoch": 3.78, + "learning_rate": 1.1389119324578197e-05, + "loss": 0.1925, + "step": 9017 + }, + { + "epoch": 3.78, + "learning_rate": 1.1387440333174809e-05, + "loss": 0.1357, + "step": 9018 + }, + { + "epoch": 3.78, + "learning_rate": 1.1385761301890674e-05, + "loss": 0.1902, + "step": 9019 + }, + { + "epoch": 3.78, + "learning_rate": 1.1384082230774063e-05, + "loss": 0.2036, + "step": 9020 + }, + { + "epoch": 3.78, + "learning_rate": 1.1382403119873236e-05, + "loss": 0.1801, + "step": 9021 + }, + { + "epoch": 3.78, + "learning_rate": 1.1380723969236458e-05, + "loss": 0.204, + "step": 9022 + }, + { + "epoch": 3.78, + "learning_rate": 1.1379044778911994e-05, + "loss": 0.2092, + "step": 9023 + }, + { + "epoch": 3.78, + "learning_rate": 1.1377365548948114e-05, + "loss": 0.164, + "step": 9024 + }, + { + "epoch": 3.78, + "learning_rate": 1.137568627939308e-05, + "loss": 0.1681, + "step": 9025 + }, + { + "epoch": 3.78, + "learning_rate": 1.1374006970295167e-05, + "loss": 0.1726, + "step": 9026 + }, + { + "epoch": 3.78, + "learning_rate": 1.137232762170264e-05, + "loss": 0.1698, + "step": 9027 + }, + { + "epoch": 3.78, + "learning_rate": 1.1370648233663776e-05, + "loss": 0.1895, + "step": 9028 + }, + { + "epoch": 3.78, + "learning_rate": 1.1368968806226844e-05, + "loss": 0.195, + "step": 9029 + }, + { + "epoch": 3.78, + "learning_rate": 1.1367289339440123e-05, + "loss": 0.2167, + "step": 9030 + }, + { + "epoch": 3.78, + "learning_rate": 1.1365609833351877e-05, + "loss": 0.232, + "step": 9031 + }, + { + "epoch": 3.78, + "learning_rate": 1.136393028801039e-05, + "loss": 0.1908, + "step": 9032 + }, + { + "epoch": 3.78, + "learning_rate": 1.1362250703463937e-05, + "loss": 0.1637, + "step": 9033 + }, + { + "epoch": 3.78, + "learning_rate": 1.1360571079760798e-05, + "loss": 0.1488, + "step": 9034 + }, + { + "epoch": 3.78, + "learning_rate": 1.1358891416949249e-05, + "loss": 0.2147, + "step": 9035 + }, + { + "epoch": 3.78, + "learning_rate": 1.1357211715077571e-05, + "loss": 0.2032, + "step": 9036 + }, + { + "epoch": 3.78, + "learning_rate": 1.1355531974194051e-05, + "loss": 0.1986, + "step": 9037 + }, + { + "epoch": 3.78, + "learning_rate": 1.1353852194346963e-05, + "loss": 0.1634, + "step": 9038 + }, + { + "epoch": 3.78, + "learning_rate": 1.1352172375584596e-05, + "loss": 0.2022, + "step": 9039 + }, + { + "epoch": 3.78, + "learning_rate": 1.1350492517955234e-05, + "loss": 0.1578, + "step": 9040 + }, + { + "epoch": 3.79, + "learning_rate": 1.1348812621507167e-05, + "loss": 0.1488, + "step": 9041 + }, + { + "epoch": 3.79, + "learning_rate": 1.1347132686288673e-05, + "loss": 0.1921, + "step": 9042 + }, + { + "epoch": 3.79, + "learning_rate": 1.1345452712348048e-05, + "loss": 0.1615, + "step": 9043 + }, + { + "epoch": 3.79, + "learning_rate": 1.1343772699733579e-05, + "loss": 0.2077, + "step": 9044 + }, + { + "epoch": 3.79, + "learning_rate": 1.1342092648493555e-05, + "loss": 0.1969, + "step": 9045 + }, + { + "epoch": 3.79, + "learning_rate": 1.1340412558676266e-05, + "loss": 0.2035, + "step": 9046 + }, + { + "epoch": 3.79, + "learning_rate": 1.1338732430330011e-05, + "loss": 0.1633, + "step": 9047 + }, + { + "epoch": 3.79, + "learning_rate": 1.1337052263503079e-05, + "loss": 0.1723, + "step": 9048 + }, + { + "epoch": 3.79, + "learning_rate": 1.1335372058243768e-05, + "loss": 0.2138, + "step": 9049 + }, + { + "epoch": 3.79, + "learning_rate": 1.1333691814600369e-05, + "loss": 0.1746, + "step": 9050 + }, + { + "epoch": 3.79, + "learning_rate": 1.1332011532621186e-05, + "loss": 0.2055, + "step": 9051 + }, + { + "epoch": 3.79, + "learning_rate": 1.1330331212354513e-05, + "loss": 0.1866, + "step": 9052 + }, + { + "epoch": 3.79, + "learning_rate": 1.1328650853848646e-05, + "loss": 0.1995, + "step": 9053 + }, + { + "epoch": 3.79, + "learning_rate": 1.132697045715189e-05, + "loss": 0.1582, + "step": 9054 + }, + { + "epoch": 3.79, + "learning_rate": 1.132529002231255e-05, + "loss": 0.2029, + "step": 9055 + }, + { + "epoch": 3.79, + "learning_rate": 1.132360954937892e-05, + "loss": 0.1756, + "step": 9056 + }, + { + "epoch": 3.79, + "learning_rate": 1.1321929038399309e-05, + "loss": 0.2151, + "step": 9057 + }, + { + "epoch": 3.79, + "learning_rate": 1.1320248489422024e-05, + "loss": 0.211, + "step": 9058 + }, + { + "epoch": 3.79, + "learning_rate": 1.1318567902495364e-05, + "loss": 0.2204, + "step": 9059 + }, + { + "epoch": 3.79, + "learning_rate": 1.131688727766764e-05, + "loss": 0.1669, + "step": 9060 + }, + { + "epoch": 3.79, + "learning_rate": 1.1315206614987161e-05, + "loss": 0.1861, + "step": 9061 + }, + { + "epoch": 3.79, + "learning_rate": 1.1313525914502235e-05, + "loss": 0.1728, + "step": 9062 + }, + { + "epoch": 3.79, + "learning_rate": 1.1311845176261172e-05, + "loss": 0.1886, + "step": 9063 + }, + { + "epoch": 3.79, + "learning_rate": 1.1310164400312284e-05, + "loss": 0.1917, + "step": 9064 + }, + { + "epoch": 3.8, + "learning_rate": 1.1308483586703885e-05, + "loss": 0.1719, + "step": 9065 + }, + { + "epoch": 3.8, + "learning_rate": 1.1306802735484282e-05, + "loss": 0.1885, + "step": 9066 + }, + { + "epoch": 3.8, + "learning_rate": 1.1305121846701797e-05, + "loss": 0.2081, + "step": 9067 + }, + { + "epoch": 3.8, + "learning_rate": 1.130344092040474e-05, + "loss": 0.2003, + "step": 9068 + }, + { + "epoch": 3.8, + "learning_rate": 1.1301759956641432e-05, + "loss": 0.199, + "step": 9069 + }, + { + "epoch": 3.8, + "learning_rate": 1.1300078955460187e-05, + "loss": 0.2198, + "step": 9070 + }, + { + "epoch": 3.8, + "learning_rate": 1.1298397916909332e-05, + "loss": 0.187, + "step": 9071 + }, + { + "epoch": 3.8, + "learning_rate": 1.1296716841037176e-05, + "loss": 0.2268, + "step": 9072 + }, + { + "epoch": 3.8, + "learning_rate": 1.1295035727892046e-05, + "loss": 0.2108, + "step": 9073 + }, + { + "epoch": 3.8, + "learning_rate": 1.1293354577522264e-05, + "loss": 0.2043, + "step": 9074 + }, + { + "epoch": 3.8, + "learning_rate": 1.1291673389976153e-05, + "loss": 0.1764, + "step": 9075 + }, + { + "epoch": 3.8, + "learning_rate": 1.1289992165302036e-05, + "loss": 0.2057, + "step": 9076 + }, + { + "epoch": 3.8, + "learning_rate": 1.1288310903548236e-05, + "loss": 0.1947, + "step": 9077 + }, + { + "epoch": 3.8, + "learning_rate": 1.1286629604763087e-05, + "loss": 0.2054, + "step": 9078 + }, + { + "epoch": 3.8, + "learning_rate": 1.1284948268994908e-05, + "loss": 0.2173, + "step": 9079 + }, + { + "epoch": 3.8, + "learning_rate": 1.1283266896292033e-05, + "loss": 0.1699, + "step": 9080 + }, + { + "epoch": 3.8, + "learning_rate": 1.1281585486702791e-05, + "loss": 0.1738, + "step": 9081 + }, + { + "epoch": 3.8, + "learning_rate": 1.1279904040275508e-05, + "loss": 0.1923, + "step": 9082 + }, + { + "epoch": 3.8, + "learning_rate": 1.1278222557058521e-05, + "loss": 0.191, + "step": 9083 + }, + { + "epoch": 3.8, + "learning_rate": 1.127654103710016e-05, + "loss": 0.2003, + "step": 9084 + }, + { + "epoch": 3.8, + "learning_rate": 1.127485948044876e-05, + "loss": 0.1608, + "step": 9085 + }, + { + "epoch": 3.8, + "learning_rate": 1.127317788715266e-05, + "loss": 0.1259, + "step": 9086 + }, + { + "epoch": 3.8, + "learning_rate": 1.1271496257260186e-05, + "loss": 0.1972, + "step": 9087 + }, + { + "epoch": 3.8, + "learning_rate": 1.126981459081968e-05, + "loss": 0.1557, + "step": 9088 + }, + { + "epoch": 3.81, + "learning_rate": 1.1268132887879485e-05, + "loss": 0.2127, + "step": 9089 + }, + { + "epoch": 3.81, + "learning_rate": 1.1266451148487934e-05, + "loss": 0.1941, + "step": 9090 + }, + { + "epoch": 3.81, + "learning_rate": 1.1264769372693365e-05, + "loss": 0.1913, + "step": 9091 + }, + { + "epoch": 3.81, + "learning_rate": 1.1263087560544128e-05, + "loss": 0.1689, + "step": 9092 + }, + { + "epoch": 3.81, + "learning_rate": 1.1261405712088556e-05, + "loss": 0.1506, + "step": 9093 + }, + { + "epoch": 3.81, + "learning_rate": 1.1259723827374996e-05, + "loss": 0.2145, + "step": 9094 + }, + { + "epoch": 3.81, + "learning_rate": 1.1258041906451795e-05, + "loss": 0.1616, + "step": 9095 + }, + { + "epoch": 3.81, + "learning_rate": 1.1256359949367293e-05, + "loss": 0.2278, + "step": 9096 + }, + { + "epoch": 3.81, + "learning_rate": 1.1254677956169839e-05, + "loss": 0.1865, + "step": 9097 + }, + { + "epoch": 3.81, + "learning_rate": 1.1252995926907783e-05, + "loss": 0.1616, + "step": 9098 + }, + { + "epoch": 3.81, + "learning_rate": 1.1251313861629474e-05, + "loss": 0.1893, + "step": 9099 + }, + { + "epoch": 3.81, + "learning_rate": 1.1249631760383253e-05, + "loss": 0.2181, + "step": 9100 + }, + { + "epoch": 3.81, + "learning_rate": 1.1247949623217476e-05, + "loss": 0.1697, + "step": 9101 + }, + { + "epoch": 3.81, + "learning_rate": 1.1246267450180497e-05, + "loss": 0.1728, + "step": 9102 + }, + { + "epoch": 3.81, + "learning_rate": 1.1244585241320664e-05, + "loss": 0.2064, + "step": 9103 + }, + { + "epoch": 3.81, + "learning_rate": 1.1242902996686333e-05, + "loss": 0.186, + "step": 9104 + }, + { + "epoch": 3.81, + "learning_rate": 1.1241220716325863e-05, + "loss": 0.1907, + "step": 9105 + }, + { + "epoch": 3.81, + "learning_rate": 1.12395384002876e-05, + "loss": 0.1635, + "step": 9106 + }, + { + "epoch": 3.81, + "learning_rate": 1.1237856048619904e-05, + "loss": 0.1679, + "step": 9107 + }, + { + "epoch": 3.81, + "learning_rate": 1.1236173661371139e-05, + "loss": 0.1374, + "step": 9108 + }, + { + "epoch": 3.81, + "learning_rate": 1.1234491238589658e-05, + "loss": 0.1645, + "step": 9109 + }, + { + "epoch": 3.81, + "learning_rate": 1.1232808780323821e-05, + "loss": 0.2045, + "step": 9110 + }, + { + "epoch": 3.81, + "learning_rate": 1.1231126286621988e-05, + "loss": 0.1994, + "step": 9111 + }, + { + "epoch": 3.81, + "learning_rate": 1.1229443757532526e-05, + "loss": 0.1896, + "step": 9112 + }, + { + "epoch": 3.82, + "learning_rate": 1.122776119310379e-05, + "loss": 0.1959, + "step": 9113 + }, + { + "epoch": 3.82, + "learning_rate": 1.122607859338415e-05, + "loss": 0.2131, + "step": 9114 + }, + { + "epoch": 3.82, + "learning_rate": 1.122439595842197e-05, + "loss": 0.1619, + "step": 9115 + }, + { + "epoch": 3.82, + "learning_rate": 1.1222713288265613e-05, + "loss": 0.1915, + "step": 9116 + }, + { + "epoch": 3.82, + "learning_rate": 1.1221030582963447e-05, + "loss": 0.1749, + "step": 9117 + }, + { + "epoch": 3.82, + "learning_rate": 1.1219347842563838e-05, + "loss": 0.1934, + "step": 9118 + }, + { + "epoch": 3.82, + "learning_rate": 1.1217665067115164e-05, + "loss": 0.1642, + "step": 9119 + }, + { + "epoch": 3.82, + "learning_rate": 1.1215982256665786e-05, + "loss": 0.2118, + "step": 9120 + }, + { + "epoch": 3.82, + "learning_rate": 1.1214299411264075e-05, + "loss": 0.1747, + "step": 9121 + }, + { + "epoch": 3.82, + "learning_rate": 1.1212616530958407e-05, + "loss": 0.1927, + "step": 9122 + }, + { + "epoch": 3.82, + "learning_rate": 1.121093361579715e-05, + "loss": 0.1762, + "step": 9123 + }, + { + "epoch": 3.82, + "learning_rate": 1.1209250665828684e-05, + "loss": 0.1817, + "step": 9124 + }, + { + "epoch": 3.82, + "learning_rate": 1.1207567681101377e-05, + "loss": 0.2285, + "step": 9125 + }, + { + "epoch": 3.82, + "learning_rate": 1.1205884661663614e-05, + "loss": 0.1896, + "step": 9126 + }, + { + "epoch": 3.82, + "learning_rate": 1.1204201607563761e-05, + "loss": 0.2009, + "step": 9127 + }, + { + "epoch": 3.82, + "learning_rate": 1.1202518518850202e-05, + "loss": 0.1918, + "step": 9128 + }, + { + "epoch": 3.82, + "learning_rate": 1.1200835395571319e-05, + "loss": 0.1798, + "step": 9129 + }, + { + "epoch": 3.82, + "learning_rate": 1.1199152237775485e-05, + "loss": 0.1773, + "step": 9130 + }, + { + "epoch": 3.82, + "learning_rate": 1.1197469045511085e-05, + "loss": 0.1865, + "step": 9131 + }, + { + "epoch": 3.82, + "learning_rate": 1.11957858188265e-05, + "loss": 0.1732, + "step": 9132 + }, + { + "epoch": 3.82, + "learning_rate": 1.1194102557770117e-05, + "loss": 0.2204, + "step": 9133 + }, + { + "epoch": 3.82, + "learning_rate": 1.119241926239031e-05, + "loss": 0.1778, + "step": 9134 + }, + { + "epoch": 3.82, + "learning_rate": 1.119073593273547e-05, + "loss": 0.1671, + "step": 9135 + }, + { + "epoch": 3.82, + "learning_rate": 1.1189052568853985e-05, + "loss": 0.2029, + "step": 9136 + }, + { + "epoch": 3.83, + "learning_rate": 1.1187369170794235e-05, + "loss": 0.2035, + "step": 9137 + }, + { + "epoch": 3.83, + "learning_rate": 1.1185685738604615e-05, + "loss": 0.1841, + "step": 9138 + }, + { + "epoch": 3.83, + "learning_rate": 1.1184002272333509e-05, + "loss": 0.1578, + "step": 9139 + }, + { + "epoch": 3.83, + "learning_rate": 1.1182318772029309e-05, + "loss": 0.1937, + "step": 9140 + }, + { + "epoch": 3.83, + "learning_rate": 1.1180635237740405e-05, + "loss": 0.1827, + "step": 9141 + }, + { + "epoch": 3.83, + "learning_rate": 1.117895166951519e-05, + "loss": 0.2154, + "step": 9142 + }, + { + "epoch": 3.83, + "learning_rate": 1.1177268067402057e-05, + "loss": 0.1988, + "step": 9143 + }, + { + "epoch": 3.83, + "learning_rate": 1.1175584431449393e-05, + "loss": 0.1743, + "step": 9144 + }, + { + "epoch": 3.83, + "learning_rate": 1.11739007617056e-05, + "loss": 0.2047, + "step": 9145 + }, + { + "epoch": 3.83, + "learning_rate": 1.1172217058219073e-05, + "loss": 0.1919, + "step": 9146 + }, + { + "epoch": 3.83, + "learning_rate": 1.1170533321038204e-05, + "loss": 0.186, + "step": 9147 + }, + { + "epoch": 3.83, + "learning_rate": 1.1168849550211394e-05, + "loss": 0.1803, + "step": 9148 + }, + { + "epoch": 3.83, + "learning_rate": 1.1167165745787041e-05, + "loss": 0.2222, + "step": 9149 + }, + { + "epoch": 3.83, + "learning_rate": 1.1165481907813544e-05, + "loss": 0.1488, + "step": 9150 + }, + { + "epoch": 3.83, + "learning_rate": 1.1163798036339303e-05, + "loss": 0.1842, + "step": 9151 + }, + { + "epoch": 3.83, + "learning_rate": 1.1162114131412723e-05, + "loss": 0.1747, + "step": 9152 + }, + { + "epoch": 3.83, + "learning_rate": 1.1160430193082203e-05, + "loss": 0.1926, + "step": 9153 + }, + { + "epoch": 3.83, + "learning_rate": 1.1158746221396148e-05, + "loss": 0.1933, + "step": 9154 + }, + { + "epoch": 3.83, + "learning_rate": 1.1157062216402959e-05, + "loss": 0.1534, + "step": 9155 + }, + { + "epoch": 3.83, + "learning_rate": 1.1155378178151044e-05, + "loss": 0.1914, + "step": 9156 + }, + { + "epoch": 3.83, + "learning_rate": 1.1153694106688809e-05, + "loss": 0.1767, + "step": 9157 + }, + { + "epoch": 3.83, + "learning_rate": 1.115201000206466e-05, + "loss": 0.1747, + "step": 9158 + }, + { + "epoch": 3.83, + "learning_rate": 1.1150325864327003e-05, + "loss": 0.193, + "step": 9159 + }, + { + "epoch": 3.84, + "learning_rate": 1.1148641693524254e-05, + "loss": 0.1653, + "step": 9160 + }, + { + "epoch": 3.84, + "learning_rate": 1.1146957489704817e-05, + "loss": 0.1859, + "step": 9161 + }, + { + "epoch": 3.84, + "learning_rate": 1.1145273252917104e-05, + "loss": 0.1975, + "step": 9162 + }, + { + "epoch": 3.84, + "learning_rate": 1.114358898320953e-05, + "loss": 0.1936, + "step": 9163 + }, + { + "epoch": 3.84, + "learning_rate": 1.1141904680630504e-05, + "loss": 0.2159, + "step": 9164 + }, + { + "epoch": 3.84, + "learning_rate": 1.114022034522844e-05, + "loss": 0.1748, + "step": 9165 + }, + { + "epoch": 3.84, + "learning_rate": 1.1138535977051757e-05, + "loss": 0.2104, + "step": 9166 + }, + { + "epoch": 3.84, + "learning_rate": 1.1136851576148867e-05, + "loss": 0.1722, + "step": 9167 + }, + { + "epoch": 3.84, + "learning_rate": 1.1135167142568186e-05, + "loss": 0.2082, + "step": 9168 + }, + { + "epoch": 3.84, + "learning_rate": 1.1133482676358131e-05, + "loss": 0.1616, + "step": 9169 + }, + { + "epoch": 3.84, + "learning_rate": 1.1131798177567129e-05, + "loss": 0.1815, + "step": 9170 + }, + { + "epoch": 3.84, + "learning_rate": 1.1130113646243585e-05, + "loss": 0.2332, + "step": 9171 + }, + { + "epoch": 3.84, + "learning_rate": 1.112842908243593e-05, + "loss": 0.1734, + "step": 9172 + }, + { + "epoch": 3.84, + "learning_rate": 1.1126744486192585e-05, + "loss": 0.1712, + "step": 9173 + }, + { + "epoch": 3.84, + "learning_rate": 1.1125059857561966e-05, + "loss": 0.2049, + "step": 9174 + }, + { + "epoch": 3.84, + "learning_rate": 1.11233751965925e-05, + "loss": 0.2336, + "step": 9175 + }, + { + "epoch": 3.84, + "learning_rate": 1.1121690503332613e-05, + "loss": 0.1608, + "step": 9176 + }, + { + "epoch": 3.84, + "learning_rate": 1.1120005777830729e-05, + "loss": 0.2013, + "step": 9177 + }, + { + "epoch": 3.84, + "learning_rate": 1.111832102013527e-05, + "loss": 0.1843, + "step": 9178 + }, + { + "epoch": 3.84, + "learning_rate": 1.1116636230294663e-05, + "loss": 0.1676, + "step": 9179 + }, + { + "epoch": 3.84, + "learning_rate": 1.1114951408357344e-05, + "loss": 0.1899, + "step": 9180 + }, + { + "epoch": 3.84, + "learning_rate": 1.1113266554371734e-05, + "loss": 0.2414, + "step": 9181 + }, + { + "epoch": 3.84, + "learning_rate": 1.1111581668386262e-05, + "loss": 0.1606, + "step": 9182 + }, + { + "epoch": 3.84, + "learning_rate": 1.1109896750449367e-05, + "loss": 0.1816, + "step": 9183 + }, + { + "epoch": 3.85, + "learning_rate": 1.110821180060947e-05, + "loss": 0.207, + "step": 9184 + }, + { + "epoch": 3.85, + "learning_rate": 1.1106526818915008e-05, + "loss": 0.1648, + "step": 9185 + }, + { + "epoch": 3.85, + "learning_rate": 1.1104841805414418e-05, + "loss": 0.2001, + "step": 9186 + }, + { + "epoch": 3.85, + "learning_rate": 1.1103156760156129e-05, + "loss": 0.154, + "step": 9187 + }, + { + "epoch": 3.85, + "learning_rate": 1.110147168318858e-05, + "loss": 0.183, + "step": 9188 + }, + { + "epoch": 3.85, + "learning_rate": 1.10997865745602e-05, + "loss": 0.2053, + "step": 9189 + }, + { + "epoch": 3.85, + "learning_rate": 1.1098101434319438e-05, + "loss": 0.2275, + "step": 9190 + }, + { + "epoch": 3.85, + "learning_rate": 1.1096416262514717e-05, + "loss": 0.197, + "step": 9191 + }, + { + "epoch": 3.85, + "learning_rate": 1.1094731059194488e-05, + "loss": 0.2249, + "step": 9192 + }, + { + "epoch": 3.85, + "learning_rate": 1.1093045824407186e-05, + "loss": 0.1649, + "step": 9193 + }, + { + "epoch": 3.85, + "learning_rate": 1.1091360558201249e-05, + "loss": 0.205, + "step": 9194 + }, + { + "epoch": 3.85, + "learning_rate": 1.1089675260625124e-05, + "loss": 0.1797, + "step": 9195 + }, + { + "epoch": 3.85, + "learning_rate": 1.1087989931727248e-05, + "loss": 0.1717, + "step": 9196 + }, + { + "epoch": 3.85, + "learning_rate": 1.1086304571556071e-05, + "loss": 0.1843, + "step": 9197 + }, + { + "epoch": 3.85, + "learning_rate": 1.1084619180160028e-05, + "loss": 0.2124, + "step": 9198 + }, + { + "epoch": 3.85, + "learning_rate": 1.1082933757587571e-05, + "loss": 0.1779, + "step": 9199 + }, + { + "epoch": 3.85, + "learning_rate": 1.1081248303887147e-05, + "loss": 0.2332, + "step": 9200 + }, + { + "epoch": 3.85, + "learning_rate": 1.1079562819107198e-05, + "loss": 0.2471, + "step": 9201 + }, + { + "epoch": 3.85, + "learning_rate": 1.1077877303296174e-05, + "loss": 0.1762, + "step": 9202 + }, + { + "epoch": 3.85, + "learning_rate": 1.1076191756502521e-05, + "loss": 0.2314, + "step": 9203 + }, + { + "epoch": 3.85, + "learning_rate": 1.1074506178774696e-05, + "loss": 0.173, + "step": 9204 + }, + { + "epoch": 3.85, + "learning_rate": 1.107282057016114e-05, + "loss": 0.2055, + "step": 9205 + }, + { + "epoch": 3.85, + "learning_rate": 1.1071134930710308e-05, + "loss": 0.1714, + "step": 9206 + }, + { + "epoch": 3.85, + "learning_rate": 1.1069449260470658e-05, + "loss": 0.1473, + "step": 9207 + }, + { + "epoch": 3.86, + "learning_rate": 1.1067763559490632e-05, + "loss": 0.149, + "step": 9208 + }, + { + "epoch": 3.86, + "learning_rate": 1.1066077827818694e-05, + "loss": 0.2025, + "step": 9209 + }, + { + "epoch": 3.86, + "learning_rate": 1.1064392065503294e-05, + "loss": 0.171, + "step": 9210 + }, + { + "epoch": 3.86, + "learning_rate": 1.106270627259289e-05, + "loss": 0.1824, + "step": 9211 + }, + { + "epoch": 3.86, + "learning_rate": 1.1061020449135935e-05, + "loss": 0.186, + "step": 9212 + }, + { + "epoch": 3.86, + "learning_rate": 1.1059334595180888e-05, + "loss": 0.1983, + "step": 9213 + }, + { + "epoch": 3.86, + "learning_rate": 1.1057648710776212e-05, + "loss": 0.1745, + "step": 9214 + }, + { + "epoch": 3.86, + "learning_rate": 1.1055962795970359e-05, + "loss": 0.2321, + "step": 9215 + }, + { + "epoch": 3.86, + "learning_rate": 1.1054276850811791e-05, + "loss": 0.2003, + "step": 9216 + }, + { + "epoch": 3.86, + "learning_rate": 1.1052590875348975e-05, + "loss": 0.2135, + "step": 9217 + }, + { + "epoch": 3.86, + "learning_rate": 1.1050904869630365e-05, + "loss": 0.2082, + "step": 9218 + }, + { + "epoch": 3.86, + "learning_rate": 1.1049218833704427e-05, + "loss": 0.2232, + "step": 9219 + }, + { + "epoch": 3.86, + "learning_rate": 1.1047532767619629e-05, + "loss": 0.2073, + "step": 9220 + }, + { + "epoch": 3.86, + "learning_rate": 1.1045846671424424e-05, + "loss": 0.1981, + "step": 9221 + }, + { + "epoch": 3.86, + "learning_rate": 1.1044160545167294e-05, + "loss": 0.1434, + "step": 9222 + }, + { + "epoch": 3.86, + "learning_rate": 1.1042474388896687e-05, + "loss": 0.1945, + "step": 9223 + }, + { + "epoch": 3.86, + "learning_rate": 1.1040788202661084e-05, + "loss": 0.1881, + "step": 9224 + }, + { + "epoch": 3.86, + "learning_rate": 1.1039101986508945e-05, + "loss": 0.1577, + "step": 9225 + }, + { + "epoch": 3.86, + "learning_rate": 1.1037415740488742e-05, + "loss": 0.1527, + "step": 9226 + }, + { + "epoch": 3.86, + "learning_rate": 1.1035729464648945e-05, + "loss": 0.2051, + "step": 9227 + }, + { + "epoch": 3.86, + "learning_rate": 1.1034043159038021e-05, + "loss": 0.1761, + "step": 9228 + }, + { + "epoch": 3.86, + "learning_rate": 1.1032356823704446e-05, + "loss": 0.2139, + "step": 9229 + }, + { + "epoch": 3.86, + "learning_rate": 1.1030670458696688e-05, + "loss": 0.2011, + "step": 9230 + }, + { + "epoch": 3.86, + "learning_rate": 1.1028984064063226e-05, + "loss": 0.1833, + "step": 9231 + }, + { + "epoch": 3.87, + "learning_rate": 1.1027297639852528e-05, + "loss": 0.1715, + "step": 9232 + }, + { + "epoch": 3.87, + "learning_rate": 1.1025611186113073e-05, + "loss": 0.1462, + "step": 9233 + }, + { + "epoch": 3.87, + "learning_rate": 1.1023924702893334e-05, + "loss": 0.1673, + "step": 9234 + }, + { + "epoch": 3.87, + "learning_rate": 1.1022238190241787e-05, + "loss": 0.1945, + "step": 9235 + }, + { + "epoch": 3.87, + "learning_rate": 1.102055164820691e-05, + "loss": 0.1562, + "step": 9236 + }, + { + "epoch": 3.87, + "learning_rate": 1.1018865076837182e-05, + "loss": 0.1894, + "step": 9237 + }, + { + "epoch": 3.87, + "learning_rate": 1.101717847618108e-05, + "loss": 0.1953, + "step": 9238 + }, + { + "epoch": 3.87, + "learning_rate": 1.1015491846287088e-05, + "loss": 0.2008, + "step": 9239 + }, + { + "epoch": 3.87, + "learning_rate": 1.1013805187203682e-05, + "loss": 0.2021, + "step": 9240 + }, + { + "epoch": 3.87, + "learning_rate": 1.1012118498979347e-05, + "loss": 0.199, + "step": 9241 + }, + { + "epoch": 3.87, + "learning_rate": 1.1010431781662563e-05, + "loss": 0.1611, + "step": 9242 + }, + { + "epoch": 3.87, + "learning_rate": 1.1008745035301814e-05, + "loss": 0.208, + "step": 9243 + }, + { + "epoch": 3.87, + "learning_rate": 1.1007058259945584e-05, + "loss": 0.1939, + "step": 9244 + }, + { + "epoch": 3.87, + "learning_rate": 1.100537145564236e-05, + "loss": 0.1942, + "step": 9245 + }, + { + "epoch": 3.87, + "learning_rate": 1.1003684622440622e-05, + "loss": 0.1912, + "step": 9246 + }, + { + "epoch": 3.87, + "learning_rate": 1.1001997760388862e-05, + "loss": 0.2018, + "step": 9247 + }, + { + "epoch": 3.87, + "learning_rate": 1.100031086953557e-05, + "loss": 0.1898, + "step": 9248 + }, + { + "epoch": 3.87, + "learning_rate": 1.0998623949929223e-05, + "loss": 0.1977, + "step": 9249 + }, + { + "epoch": 3.87, + "learning_rate": 1.0996937001618317e-05, + "loss": 0.1582, + "step": 9250 + }, + { + "epoch": 3.87, + "learning_rate": 1.0995250024651345e-05, + "loss": 0.1945, + "step": 9251 + }, + { + "epoch": 3.87, + "learning_rate": 1.0993563019076794e-05, + "loss": 0.1894, + "step": 9252 + }, + { + "epoch": 3.87, + "learning_rate": 1.0991875984943152e-05, + "loss": 0.1623, + "step": 9253 + }, + { + "epoch": 3.87, + "learning_rate": 1.0990188922298918e-05, + "loss": 0.1477, + "step": 9254 + }, + { + "epoch": 3.87, + "learning_rate": 1.0988501831192582e-05, + "loss": 0.1554, + "step": 9255 + }, + { + "epoch": 3.88, + "learning_rate": 1.098681471167264e-05, + "loss": 0.2056, + "step": 9256 + }, + { + "epoch": 3.88, + "learning_rate": 1.098512756378758e-05, + "loss": 0.2109, + "step": 9257 + }, + { + "epoch": 3.88, + "learning_rate": 1.0983440387585909e-05, + "loss": 0.15, + "step": 9258 + }, + { + "epoch": 3.88, + "learning_rate": 1.0981753183116111e-05, + "loss": 0.2154, + "step": 9259 + }, + { + "epoch": 3.88, + "learning_rate": 1.0980065950426687e-05, + "loss": 0.2, + "step": 9260 + }, + { + "epoch": 3.88, + "learning_rate": 1.0978378689566144e-05, + "loss": 0.217, + "step": 9261 + }, + { + "epoch": 3.88, + "learning_rate": 1.097669140058297e-05, + "loss": 0.2133, + "step": 9262 + }, + { + "epoch": 3.88, + "learning_rate": 1.0975004083525667e-05, + "loss": 0.198, + "step": 9263 + }, + { + "epoch": 3.88, + "learning_rate": 1.0973316738442738e-05, + "loss": 0.2014, + "step": 9264 + }, + { + "epoch": 3.88, + "learning_rate": 1.0971629365382686e-05, + "loss": 0.1903, + "step": 9265 + }, + { + "epoch": 3.88, + "learning_rate": 1.0969941964394006e-05, + "loss": 0.1916, + "step": 9266 + }, + { + "epoch": 3.88, + "learning_rate": 1.0968254535525204e-05, + "loss": 0.2014, + "step": 9267 + }, + { + "epoch": 3.88, + "learning_rate": 1.0966567078824792e-05, + "loss": 0.1985, + "step": 9268 + }, + { + "epoch": 3.88, + "learning_rate": 1.0964879594341262e-05, + "loss": 0.1646, + "step": 9269 + }, + { + "epoch": 3.88, + "learning_rate": 1.0963192082123124e-05, + "loss": 0.1842, + "step": 9270 + }, + { + "epoch": 3.88, + "learning_rate": 1.0961504542218884e-05, + "loss": 0.2025, + "step": 9271 + }, + { + "epoch": 3.88, + "learning_rate": 1.0959816974677052e-05, + "loss": 0.1725, + "step": 9272 + }, + { + "epoch": 3.88, + "learning_rate": 1.095812937954613e-05, + "loss": 0.2005, + "step": 9273 + }, + { + "epoch": 3.88, + "learning_rate": 1.0956441756874628e-05, + "loss": 0.1933, + "step": 9274 + }, + { + "epoch": 3.88, + "learning_rate": 1.095475410671106e-05, + "loss": 0.1843, + "step": 9275 + }, + { + "epoch": 3.88, + "learning_rate": 1.0953066429103929e-05, + "loss": 0.2184, + "step": 9276 + }, + { + "epoch": 3.88, + "learning_rate": 1.0951378724101751e-05, + "loss": 0.2005, + "step": 9277 + }, + { + "epoch": 3.88, + "learning_rate": 1.0949690991753036e-05, + "loss": 0.1757, + "step": 9278 + }, + { + "epoch": 3.88, + "learning_rate": 1.0948003232106298e-05, + "loss": 0.1942, + "step": 9279 + }, + { + "epoch": 3.89, + "learning_rate": 1.0946315445210048e-05, + "loss": 0.1583, + "step": 9280 + }, + { + "epoch": 3.89, + "learning_rate": 1.0944627631112796e-05, + "loss": 0.1767, + "step": 9281 + }, + { + "epoch": 3.89, + "learning_rate": 1.0942939789863065e-05, + "loss": 0.2008, + "step": 9282 + }, + { + "epoch": 3.89, + "learning_rate": 1.0941251921509367e-05, + "loss": 0.1965, + "step": 9283 + }, + { + "epoch": 3.89, + "learning_rate": 1.0939564026100213e-05, + "loss": 0.2036, + "step": 9284 + }, + { + "epoch": 3.89, + "learning_rate": 1.093787610368413e-05, + "loss": 0.1682, + "step": 9285 + }, + { + "epoch": 3.89, + "learning_rate": 1.0936188154309626e-05, + "loss": 0.1771, + "step": 9286 + }, + { + "epoch": 3.89, + "learning_rate": 1.0934500178025228e-05, + "loss": 0.1683, + "step": 9287 + }, + { + "epoch": 3.89, + "learning_rate": 1.093281217487945e-05, + "loss": 0.1783, + "step": 9288 + }, + { + "epoch": 3.89, + "learning_rate": 1.0931124144920813e-05, + "loss": 0.2245, + "step": 9289 + }, + { + "epoch": 3.89, + "learning_rate": 1.0929436088197842e-05, + "loss": 0.2076, + "step": 9290 + }, + { + "epoch": 3.89, + "learning_rate": 1.0927748004759053e-05, + "loss": 0.1446, + "step": 9291 + }, + { + "epoch": 3.89, + "learning_rate": 1.0926059894652972e-05, + "loss": 0.1343, + "step": 9292 + }, + { + "epoch": 3.89, + "learning_rate": 1.0924371757928121e-05, + "loss": 0.1693, + "step": 9293 + }, + { + "epoch": 3.89, + "learning_rate": 1.092268359463302e-05, + "loss": 0.2063, + "step": 9294 + }, + { + "epoch": 3.89, + "learning_rate": 1.0920995404816204e-05, + "loss": 0.1792, + "step": 9295 + }, + { + "epoch": 3.89, + "learning_rate": 1.091930718852619e-05, + "loss": 0.1761, + "step": 9296 + }, + { + "epoch": 3.89, + "learning_rate": 1.0917618945811503e-05, + "loss": 0.1832, + "step": 9297 + }, + { + "epoch": 3.89, + "learning_rate": 1.0915930676720678e-05, + "loss": 0.1779, + "step": 9298 + }, + { + "epoch": 3.89, + "learning_rate": 1.0914242381302238e-05, + "loss": 0.1716, + "step": 9299 + }, + { + "epoch": 3.89, + "learning_rate": 1.091255405960471e-05, + "loss": 0.173, + "step": 9300 + }, + { + "epoch": 3.89, + "learning_rate": 1.0910865711676628e-05, + "loss": 0.1813, + "step": 9301 + }, + { + "epoch": 3.89, + "learning_rate": 1.0909177337566518e-05, + "loss": 0.1746, + "step": 9302 + }, + { + "epoch": 3.89, + "learning_rate": 1.0907488937322915e-05, + "loss": 0.1858, + "step": 9303 + }, + { + "epoch": 3.9, + "learning_rate": 1.0905800510994341e-05, + "loss": 0.2029, + "step": 9304 + }, + { + "epoch": 3.9, + "learning_rate": 1.0904112058629341e-05, + "loss": 0.1727, + "step": 9305 + }, + { + "epoch": 3.9, + "learning_rate": 1.090242358027644e-05, + "loss": 0.1983, + "step": 9306 + }, + { + "epoch": 3.9, + "learning_rate": 1.0900735075984174e-05, + "loss": 0.1724, + "step": 9307 + }, + { + "epoch": 3.9, + "learning_rate": 1.0899046545801078e-05, + "loss": 0.1676, + "step": 9308 + }, + { + "epoch": 3.9, + "learning_rate": 1.0897357989775687e-05, + "loss": 0.1845, + "step": 9309 + }, + { + "epoch": 3.9, + "learning_rate": 1.0895669407956538e-05, + "loss": 0.1733, + "step": 9310 + }, + { + "epoch": 3.9, + "learning_rate": 1.0893980800392164e-05, + "loss": 0.209, + "step": 9311 + }, + { + "epoch": 3.9, + "learning_rate": 1.0892292167131107e-05, + "loss": 0.2143, + "step": 9312 + }, + { + "epoch": 3.9, + "learning_rate": 1.0890603508221905e-05, + "loss": 0.1837, + "step": 9313 + }, + { + "epoch": 3.9, + "learning_rate": 1.0888914823713092e-05, + "loss": 0.2079, + "step": 9314 + }, + { + "epoch": 3.9, + "learning_rate": 1.0887226113653211e-05, + "loss": 0.1714, + "step": 9315 + }, + { + "epoch": 3.9, + "learning_rate": 1.0885537378090807e-05, + "loss": 0.1785, + "step": 9316 + }, + { + "epoch": 3.9, + "learning_rate": 1.0883848617074412e-05, + "loss": 0.1692, + "step": 9317 + }, + { + "epoch": 3.9, + "learning_rate": 1.0882159830652574e-05, + "loss": 0.2079, + "step": 9318 + }, + { + "epoch": 3.9, + "learning_rate": 1.0880471018873836e-05, + "loss": 0.2076, + "step": 9319 + }, + { + "epoch": 3.9, + "learning_rate": 1.087878218178674e-05, + "loss": 0.2011, + "step": 9320 + }, + { + "epoch": 3.9, + "learning_rate": 1.0877093319439828e-05, + "loss": 0.1759, + "step": 9321 + }, + { + "epoch": 3.9, + "learning_rate": 1.0875404431881651e-05, + "loss": 0.2012, + "step": 9322 + }, + { + "epoch": 3.9, + "learning_rate": 1.0873715519160746e-05, + "loss": 0.1937, + "step": 9323 + }, + { + "epoch": 3.9, + "learning_rate": 1.0872026581325667e-05, + "loss": 0.1947, + "step": 9324 + }, + { + "epoch": 3.9, + "learning_rate": 1.0870337618424956e-05, + "loss": 0.1981, + "step": 9325 + }, + { + "epoch": 3.9, + "learning_rate": 1.0868648630507164e-05, + "loss": 0.2003, + "step": 9326 + }, + { + "epoch": 3.9, + "learning_rate": 1.0866959617620835e-05, + "loss": 0.2131, + "step": 9327 + }, + { + "epoch": 3.91, + "learning_rate": 1.0865270579814522e-05, + "loss": 0.1698, + "step": 9328 + }, + { + "epoch": 3.91, + "learning_rate": 1.0863581517136776e-05, + "loss": 0.1846, + "step": 9329 + }, + { + "epoch": 3.91, + "learning_rate": 1.0861892429636143e-05, + "loss": 0.1581, + "step": 9330 + }, + { + "epoch": 3.91, + "learning_rate": 1.0860203317361178e-05, + "loss": 0.1566, + "step": 9331 + }, + { + "epoch": 3.91, + "learning_rate": 1.0858514180360432e-05, + "loss": 0.2144, + "step": 9332 + }, + { + "epoch": 3.91, + "learning_rate": 1.085682501868246e-05, + "loss": 0.1654, + "step": 9333 + }, + { + "epoch": 3.91, + "learning_rate": 1.0855135832375809e-05, + "loss": 0.2098, + "step": 9334 + }, + { + "epoch": 3.91, + "learning_rate": 1.0853446621489042e-05, + "loss": 0.2019, + "step": 9335 + }, + { + "epoch": 3.91, + "learning_rate": 1.0851757386070706e-05, + "loss": 0.1809, + "step": 9336 + }, + { + "epoch": 3.91, + "learning_rate": 1.085006812616936e-05, + "loss": 0.1719, + "step": 9337 + }, + { + "epoch": 3.91, + "learning_rate": 1.084837884183356e-05, + "loss": 0.2099, + "step": 9338 + }, + { + "epoch": 3.91, + "learning_rate": 1.0846689533111865e-05, + "loss": 0.1807, + "step": 9339 + }, + { + "epoch": 3.91, + "learning_rate": 1.0845000200052828e-05, + "loss": 0.198, + "step": 9340 + }, + { + "epoch": 3.91, + "learning_rate": 1.0843310842705011e-05, + "loss": 0.1822, + "step": 9341 + }, + { + "epoch": 3.91, + "learning_rate": 1.0841621461116971e-05, + "loss": 0.2146, + "step": 9342 + }, + { + "epoch": 3.91, + "learning_rate": 1.0839932055337273e-05, + "loss": 0.1954, + "step": 9343 + }, + { + "epoch": 3.91, + "learning_rate": 1.083824262541447e-05, + "loss": 0.1589, + "step": 9344 + }, + { + "epoch": 3.91, + "learning_rate": 1.0836553171397126e-05, + "loss": 0.2323, + "step": 9345 + }, + { + "epoch": 3.91, + "learning_rate": 1.0834863693333805e-05, + "loss": 0.1982, + "step": 9346 + }, + { + "epoch": 3.91, + "learning_rate": 1.083317419127307e-05, + "loss": 0.1637, + "step": 9347 + }, + { + "epoch": 3.91, + "learning_rate": 1.0831484665263479e-05, + "loss": 0.1847, + "step": 9348 + }, + { + "epoch": 3.91, + "learning_rate": 1.0829795115353601e-05, + "loss": 0.1476, + "step": 9349 + }, + { + "epoch": 3.91, + "learning_rate": 1.0828105541591998e-05, + "loss": 0.1757, + "step": 9350 + }, + { + "epoch": 3.92, + "learning_rate": 1.0826415944027235e-05, + "loss": 0.214, + "step": 9351 + }, + { + "epoch": 3.92, + "learning_rate": 1.0824726322707878e-05, + "loss": 0.217, + "step": 9352 + }, + { + "epoch": 3.92, + "learning_rate": 1.0823036677682499e-05, + "loss": 0.1895, + "step": 9353 + }, + { + "epoch": 3.92, + "learning_rate": 1.0821347008999658e-05, + "loss": 0.1615, + "step": 9354 + }, + { + "epoch": 3.92, + "learning_rate": 1.0819657316707926e-05, + "loss": 0.1782, + "step": 9355 + }, + { + "epoch": 3.92, + "learning_rate": 1.0817967600855874e-05, + "loss": 0.1674, + "step": 9356 + }, + { + "epoch": 3.92, + "learning_rate": 1.0816277861492068e-05, + "loss": 0.1784, + "step": 9357 + }, + { + "epoch": 3.92, + "learning_rate": 1.081458809866508e-05, + "loss": 0.1907, + "step": 9358 + }, + { + "epoch": 3.92, + "learning_rate": 1.0812898312423478e-05, + "loss": 0.1914, + "step": 9359 + }, + { + "epoch": 3.92, + "learning_rate": 1.0811208502815838e-05, + "loss": 0.21, + "step": 9360 + }, + { + "epoch": 3.92, + "learning_rate": 1.0809518669890726e-05, + "loss": 0.1671, + "step": 9361 + }, + { + "epoch": 3.92, + "learning_rate": 1.0807828813696719e-05, + "loss": 0.1829, + "step": 9362 + }, + { + "epoch": 3.92, + "learning_rate": 1.0806138934282393e-05, + "loss": 0.2003, + "step": 9363 + }, + { + "epoch": 3.92, + "learning_rate": 1.0804449031696315e-05, + "loss": 0.1604, + "step": 9364 + }, + { + "epoch": 3.92, + "learning_rate": 1.0802759105987065e-05, + "loss": 0.1791, + "step": 9365 + }, + { + "epoch": 3.92, + "learning_rate": 1.0801069157203217e-05, + "loss": 0.1715, + "step": 9366 + }, + { + "epoch": 3.92, + "learning_rate": 1.0799379185393347e-05, + "loss": 0.1648, + "step": 9367 + }, + { + "epoch": 3.92, + "learning_rate": 1.0797689190606028e-05, + "loss": 0.1898, + "step": 9368 + }, + { + "epoch": 3.92, + "learning_rate": 1.0795999172889847e-05, + "loss": 0.1862, + "step": 9369 + }, + { + "epoch": 3.92, + "learning_rate": 1.0794309132293374e-05, + "loss": 0.1888, + "step": 9370 + }, + { + "epoch": 3.92, + "learning_rate": 1.0792619068865189e-05, + "loss": 0.1692, + "step": 9371 + }, + { + "epoch": 3.92, + "learning_rate": 1.0790928982653873e-05, + "loss": 0.2446, + "step": 9372 + }, + { + "epoch": 3.92, + "learning_rate": 1.0789238873708004e-05, + "loss": 0.2284, + "step": 9373 + }, + { + "epoch": 3.92, + "learning_rate": 1.0787548742076164e-05, + "loss": 0.1938, + "step": 9374 + }, + { + "epoch": 3.93, + "learning_rate": 1.0785858587806935e-05, + "loss": 0.1542, + "step": 9375 + }, + { + "epoch": 3.93, + "learning_rate": 1.0784168410948897e-05, + "loss": 0.164, + "step": 9376 + }, + { + "epoch": 3.93, + "learning_rate": 1.0782478211550636e-05, + "loss": 0.1857, + "step": 9377 + }, + { + "epoch": 3.93, + "learning_rate": 1.078078798966073e-05, + "loss": 0.189, + "step": 9378 + }, + { + "epoch": 3.93, + "learning_rate": 1.0779097745327768e-05, + "loss": 0.2207, + "step": 9379 + }, + { + "epoch": 3.93, + "learning_rate": 1.0777407478600334e-05, + "loss": 0.1534, + "step": 9380 + }, + { + "epoch": 3.93, + "learning_rate": 1.0775717189527012e-05, + "loss": 0.2344, + "step": 9381 + }, + { + "epoch": 3.93, + "learning_rate": 1.0774026878156387e-05, + "loss": 0.218, + "step": 9382 + }, + { + "epoch": 3.93, + "learning_rate": 1.0772336544537045e-05, + "loss": 0.2319, + "step": 9383 + }, + { + "epoch": 3.93, + "learning_rate": 1.0770646188717578e-05, + "loss": 0.177, + "step": 9384 + }, + { + "epoch": 3.93, + "learning_rate": 1.0768955810746564e-05, + "loss": 0.1836, + "step": 9385 + }, + { + "epoch": 3.93, + "learning_rate": 1.07672654106726e-05, + "loss": 0.1955, + "step": 9386 + }, + { + "epoch": 3.93, + "learning_rate": 1.0765574988544276e-05, + "loss": 0.1846, + "step": 9387 + }, + { + "epoch": 3.93, + "learning_rate": 1.0763884544410174e-05, + "loss": 0.1924, + "step": 9388 + }, + { + "epoch": 3.93, + "learning_rate": 1.0762194078318891e-05, + "loss": 0.1258, + "step": 9389 + }, + { + "epoch": 3.93, + "learning_rate": 1.0760503590319016e-05, + "loss": 0.1639, + "step": 9390 + }, + { + "epoch": 3.93, + "learning_rate": 1.0758813080459138e-05, + "loss": 0.1734, + "step": 9391 + }, + { + "epoch": 3.93, + "learning_rate": 1.0757122548787855e-05, + "loss": 0.1761, + "step": 9392 + }, + { + "epoch": 3.93, + "learning_rate": 1.0755431995353753e-05, + "loss": 0.199, + "step": 9393 + }, + { + "epoch": 3.93, + "learning_rate": 1.0753741420205431e-05, + "loss": 0.1722, + "step": 9394 + }, + { + "epoch": 3.93, + "learning_rate": 1.0752050823391478e-05, + "loss": 0.1927, + "step": 9395 + }, + { + "epoch": 3.93, + "learning_rate": 1.0750360204960493e-05, + "loss": 0.1614, + "step": 9396 + }, + { + "epoch": 3.93, + "learning_rate": 1.074866956496107e-05, + "loss": 0.2037, + "step": 9397 + }, + { + "epoch": 3.93, + "learning_rate": 1.0746978903441803e-05, + "loss": 0.1731, + "step": 9398 + }, + { + "epoch": 3.94, + "learning_rate": 1.0745288220451291e-05, + "loss": 0.1679, + "step": 9399 + }, + { + "epoch": 3.94, + "learning_rate": 1.0743597516038132e-05, + "loss": 0.1661, + "step": 9400 + }, + { + "epoch": 3.94, + "learning_rate": 1.074190679025092e-05, + "loss": 0.2065, + "step": 9401 + }, + { + "epoch": 3.94, + "learning_rate": 1.0740216043138256e-05, + "loss": 0.1841, + "step": 9402 + }, + { + "epoch": 3.94, + "learning_rate": 1.073852527474874e-05, + "loss": 0.1781, + "step": 9403 + }, + { + "epoch": 3.94, + "learning_rate": 1.0736834485130972e-05, + "loss": 0.1686, + "step": 9404 + }, + { + "epoch": 3.94, + "learning_rate": 1.0735143674333549e-05, + "loss": 0.188, + "step": 9405 + }, + { + "epoch": 3.94, + "learning_rate": 1.073345284240507e-05, + "loss": 0.1495, + "step": 9406 + }, + { + "epoch": 3.94, + "learning_rate": 1.0731761989394145e-05, + "loss": 0.2022, + "step": 9407 + }, + { + "epoch": 3.94, + "learning_rate": 1.0730071115349367e-05, + "loss": 0.2069, + "step": 9408 + }, + { + "epoch": 3.94, + "learning_rate": 1.0728380220319345e-05, + "loss": 0.2076, + "step": 9409 + }, + { + "epoch": 3.94, + "learning_rate": 1.072668930435268e-05, + "loss": 0.2082, + "step": 9410 + }, + { + "epoch": 3.94, + "learning_rate": 1.0724998367497974e-05, + "loss": 0.2061, + "step": 9411 + }, + { + "epoch": 3.94, + "learning_rate": 1.0723307409803834e-05, + "loss": 0.2191, + "step": 9412 + }, + { + "epoch": 3.94, + "learning_rate": 1.0721616431318865e-05, + "loss": 0.1913, + "step": 9413 + }, + { + "epoch": 3.94, + "learning_rate": 1.0719925432091671e-05, + "loss": 0.2349, + "step": 9414 + }, + { + "epoch": 3.94, + "learning_rate": 1.0718234412170862e-05, + "loss": 0.1887, + "step": 9415 + }, + { + "epoch": 3.94, + "learning_rate": 1.0716543371605039e-05, + "loss": 0.1745, + "step": 9416 + }, + { + "epoch": 3.94, + "learning_rate": 1.0714852310442817e-05, + "loss": 0.1596, + "step": 9417 + }, + { + "epoch": 3.94, + "learning_rate": 1.0713161228732795e-05, + "loss": 0.2257, + "step": 9418 + }, + { + "epoch": 3.94, + "learning_rate": 1.0711470126523587e-05, + "loss": 0.2298, + "step": 9419 + }, + { + "epoch": 3.94, + "learning_rate": 1.0709779003863801e-05, + "loss": 0.2116, + "step": 9420 + }, + { + "epoch": 3.94, + "learning_rate": 1.070808786080205e-05, + "loss": 0.1933, + "step": 9421 + }, + { + "epoch": 3.94, + "learning_rate": 1.070639669738694e-05, + "loss": 0.1895, + "step": 9422 + }, + { + "epoch": 3.95, + "learning_rate": 1.0704705513667084e-05, + "loss": 0.2109, + "step": 9423 + }, + { + "epoch": 3.95, + "learning_rate": 1.0703014309691096e-05, + "loss": 0.1643, + "step": 9424 + }, + { + "epoch": 3.95, + "learning_rate": 1.0701323085507583e-05, + "loss": 0.1672, + "step": 9425 + }, + { + "epoch": 3.95, + "learning_rate": 1.0699631841165162e-05, + "loss": 0.1769, + "step": 9426 + }, + { + "epoch": 3.95, + "learning_rate": 1.069794057671244e-05, + "loss": 0.2088, + "step": 9427 + }, + { + "epoch": 3.95, + "learning_rate": 1.069624929219804e-05, + "loss": 0.1518, + "step": 9428 + }, + { + "epoch": 3.95, + "learning_rate": 1.069455798767057e-05, + "loss": 0.1869, + "step": 9429 + }, + { + "epoch": 3.95, + "learning_rate": 1.0692866663178646e-05, + "loss": 0.1764, + "step": 9430 + }, + { + "epoch": 3.95, + "learning_rate": 1.0691175318770887e-05, + "loss": 0.1703, + "step": 9431 + }, + { + "epoch": 3.95, + "learning_rate": 1.0689483954495906e-05, + "loss": 0.1435, + "step": 9432 + }, + { + "epoch": 3.95, + "learning_rate": 1.068779257040232e-05, + "loss": 0.1748, + "step": 9433 + }, + { + "epoch": 3.95, + "learning_rate": 1.068610116653875e-05, + "loss": 0.1518, + "step": 9434 + }, + { + "epoch": 3.95, + "learning_rate": 1.0684409742953805e-05, + "loss": 0.1742, + "step": 9435 + }, + { + "epoch": 3.95, + "learning_rate": 1.0682718299696113e-05, + "loss": 0.1625, + "step": 9436 + }, + { + "epoch": 3.95, + "learning_rate": 1.068102683681429e-05, + "loss": 0.1579, + "step": 9437 + }, + { + "epoch": 3.95, + "learning_rate": 1.0679335354356954e-05, + "loss": 0.2154, + "step": 9438 + }, + { + "epoch": 3.95, + "learning_rate": 1.0677643852372726e-05, + "loss": 0.1967, + "step": 9439 + }, + { + "epoch": 3.95, + "learning_rate": 1.0675952330910224e-05, + "loss": 0.1679, + "step": 9440 + }, + { + "epoch": 3.95, + "learning_rate": 1.0674260790018076e-05, + "loss": 0.1793, + "step": 9441 + }, + { + "epoch": 3.95, + "learning_rate": 1.0672569229744899e-05, + "loss": 0.2155, + "step": 9442 + }, + { + "epoch": 3.95, + "learning_rate": 1.0670877650139316e-05, + "loss": 0.1693, + "step": 9443 + }, + { + "epoch": 3.95, + "learning_rate": 1.0669186051249948e-05, + "loss": 0.1585, + "step": 9444 + }, + { + "epoch": 3.95, + "learning_rate": 1.0667494433125423e-05, + "loss": 0.1796, + "step": 9445 + }, + { + "epoch": 3.95, + "learning_rate": 1.0665802795814362e-05, + "loss": 0.1399, + "step": 9446 + }, + { + "epoch": 3.96, + "learning_rate": 1.0664111139365392e-05, + "loss": 0.1708, + "step": 9447 + }, + { + "epoch": 3.96, + "learning_rate": 1.0662419463827136e-05, + "loss": 0.1761, + "step": 9448 + }, + { + "epoch": 3.96, + "learning_rate": 1.0660727769248221e-05, + "loss": 0.1746, + "step": 9449 + }, + { + "epoch": 3.96, + "learning_rate": 1.0659036055677272e-05, + "loss": 0.1718, + "step": 9450 + }, + { + "epoch": 3.96, + "learning_rate": 1.0657344323162916e-05, + "loss": 0.2022, + "step": 9451 + }, + { + "epoch": 3.96, + "learning_rate": 1.0655652571753781e-05, + "loss": 0.1998, + "step": 9452 + }, + { + "epoch": 3.96, + "learning_rate": 1.0653960801498493e-05, + "loss": 0.1449, + "step": 9453 + }, + { + "epoch": 3.96, + "learning_rate": 1.0652269012445684e-05, + "loss": 0.2073, + "step": 9454 + }, + { + "epoch": 3.96, + "learning_rate": 1.0650577204643981e-05, + "loss": 0.2105, + "step": 9455 + }, + { + "epoch": 3.96, + "learning_rate": 1.0648885378142013e-05, + "loss": 0.159, + "step": 9456 + }, + { + "epoch": 3.96, + "learning_rate": 1.0647193532988409e-05, + "loss": 0.2047, + "step": 9457 + }, + { + "epoch": 3.96, + "learning_rate": 1.0645501669231803e-05, + "loss": 0.1614, + "step": 9458 + }, + { + "epoch": 3.96, + "learning_rate": 1.0643809786920824e-05, + "loss": 0.2312, + "step": 9459 + }, + { + "epoch": 3.96, + "learning_rate": 1.0642117886104106e-05, + "loss": 0.195, + "step": 9460 + }, + { + "epoch": 3.96, + "learning_rate": 1.0640425966830276e-05, + "loss": 0.1706, + "step": 9461 + }, + { + "epoch": 3.96, + "learning_rate": 1.0638734029147974e-05, + "loss": 0.1486, + "step": 9462 + }, + { + "epoch": 3.96, + "learning_rate": 1.0637042073105826e-05, + "loss": 0.1422, + "step": 9463 + }, + { + "epoch": 3.96, + "learning_rate": 1.0635350098752468e-05, + "loss": 0.1756, + "step": 9464 + }, + { + "epoch": 3.96, + "learning_rate": 1.0633658106136538e-05, + "loss": 0.2265, + "step": 9465 + }, + { + "epoch": 3.96, + "learning_rate": 1.0631966095306668e-05, + "loss": 0.2164, + "step": 9466 + }, + { + "epoch": 3.96, + "learning_rate": 1.063027406631149e-05, + "loss": 0.204, + "step": 9467 + }, + { + "epoch": 3.96, + "learning_rate": 1.0628582019199648e-05, + "loss": 0.2023, + "step": 9468 + }, + { + "epoch": 3.96, + "learning_rate": 1.0626889954019768e-05, + "loss": 0.1844, + "step": 9469 + }, + { + "epoch": 3.96, + "learning_rate": 1.0625197870820493e-05, + "loss": 0.1986, + "step": 9470 + }, + { + "epoch": 3.97, + "learning_rate": 1.0623505769650466e-05, + "loss": 0.1838, + "step": 9471 + }, + { + "epoch": 3.97, + "learning_rate": 1.0621813650558317e-05, + "loss": 0.1745, + "step": 9472 + }, + { + "epoch": 3.97, + "learning_rate": 1.0620121513592682e-05, + "loss": 0.1978, + "step": 9473 + }, + { + "epoch": 3.97, + "learning_rate": 1.0618429358802209e-05, + "loss": 0.2103, + "step": 9474 + }, + { + "epoch": 3.97, + "learning_rate": 1.061673718623553e-05, + "loss": 0.1763, + "step": 9475 + }, + { + "epoch": 3.97, + "learning_rate": 1.061504499594129e-05, + "loss": 0.1444, + "step": 9476 + }, + { + "epoch": 3.97, + "learning_rate": 1.0613352787968125e-05, + "loss": 0.1943, + "step": 9477 + }, + { + "epoch": 3.97, + "learning_rate": 1.061166056236468e-05, + "loss": 0.1833, + "step": 9478 + }, + { + "epoch": 3.97, + "learning_rate": 1.0609968319179595e-05, + "loss": 0.1546, + "step": 9479 + }, + { + "epoch": 3.97, + "learning_rate": 1.060827605846151e-05, + "loss": 0.2261, + "step": 9480 + }, + { + "epoch": 3.97, + "learning_rate": 1.0606583780259073e-05, + "loss": 0.1785, + "step": 9481 + }, + { + "epoch": 3.97, + "learning_rate": 1.060489148462092e-05, + "loss": 0.1689, + "step": 9482 + }, + { + "epoch": 3.97, + "learning_rate": 1.0603199171595701e-05, + "loss": 0.1889, + "step": 9483 + }, + { + "epoch": 3.97, + "learning_rate": 1.0601506841232057e-05, + "loss": 0.1917, + "step": 9484 + }, + { + "epoch": 3.97, + "learning_rate": 1.059981449357863e-05, + "loss": 0.2113, + "step": 9485 + }, + { + "epoch": 3.97, + "learning_rate": 1.0598122128684069e-05, + "loss": 0.1792, + "step": 9486 + }, + { + "epoch": 3.97, + "learning_rate": 1.0596429746597017e-05, + "loss": 0.2004, + "step": 9487 + }, + { + "epoch": 3.97, + "learning_rate": 1.0594737347366125e-05, + "loss": 0.166, + "step": 9488 + }, + { + "epoch": 3.97, + "learning_rate": 1.059304493104003e-05, + "loss": 0.1494, + "step": 9489 + }, + { + "epoch": 3.97, + "learning_rate": 1.0591352497667388e-05, + "loss": 0.179, + "step": 9490 + }, + { + "epoch": 3.97, + "learning_rate": 1.0589660047296844e-05, + "loss": 0.1768, + "step": 9491 + }, + { + "epoch": 3.97, + "learning_rate": 1.0587967579977043e-05, + "loss": 0.2044, + "step": 9492 + }, + { + "epoch": 3.97, + "learning_rate": 1.0586275095756638e-05, + "loss": 0.1723, + "step": 9493 + }, + { + "epoch": 3.97, + "learning_rate": 1.0584582594684274e-05, + "loss": 0.1643, + "step": 9494 + }, + { + "epoch": 3.98, + "learning_rate": 1.0582890076808603e-05, + "loss": 0.1814, + "step": 9495 + }, + { + "epoch": 3.98, + "learning_rate": 1.0581197542178274e-05, + "loss": 0.1636, + "step": 9496 + }, + { + "epoch": 3.98, + "learning_rate": 1.0579504990841935e-05, + "loss": 0.2118, + "step": 9497 + }, + { + "epoch": 3.98, + "learning_rate": 1.0577812422848241e-05, + "loss": 0.1787, + "step": 9498 + }, + { + "epoch": 3.98, + "learning_rate": 1.0576119838245843e-05, + "loss": 0.211, + "step": 9499 + }, + { + "epoch": 3.98, + "learning_rate": 1.057442723708339e-05, + "loss": 0.1991, + "step": 9500 + }, + { + "epoch": 3.98, + "learning_rate": 1.0572734619409535e-05, + "loss": 0.1995, + "step": 9501 + }, + { + "epoch": 3.98, + "learning_rate": 1.0571041985272934e-05, + "loss": 0.1729, + "step": 9502 + }, + { + "epoch": 3.98, + "learning_rate": 1.0569349334722236e-05, + "loss": 0.1999, + "step": 9503 + }, + { + "epoch": 3.98, + "learning_rate": 1.0567656667806097e-05, + "loss": 0.192, + "step": 9504 + }, + { + "epoch": 3.98, + "learning_rate": 1.0565963984573174e-05, + "loss": 0.1887, + "step": 9505 + }, + { + "epoch": 3.98, + "learning_rate": 1.0564271285072117e-05, + "loss": 0.1744, + "step": 9506 + }, + { + "epoch": 3.98, + "learning_rate": 1.0562578569351579e-05, + "loss": 0.1877, + "step": 9507 + }, + { + "epoch": 3.98, + "learning_rate": 1.056088583746022e-05, + "loss": 0.1796, + "step": 9508 + }, + { + "epoch": 3.98, + "learning_rate": 1.0559193089446698e-05, + "loss": 0.2141, + "step": 9509 + }, + { + "epoch": 3.98, + "learning_rate": 1.0557500325359665e-05, + "loss": 0.1828, + "step": 9510 + }, + { + "epoch": 3.98, + "learning_rate": 1.055580754524778e-05, + "loss": 0.1784, + "step": 9511 + }, + { + "epoch": 3.98, + "learning_rate": 1.05541147491597e-05, + "loss": 0.1886, + "step": 9512 + }, + { + "epoch": 3.98, + "learning_rate": 1.0552421937144083e-05, + "loss": 0.1875, + "step": 9513 + }, + { + "epoch": 3.98, + "learning_rate": 1.055072910924959e-05, + "loss": 0.1703, + "step": 9514 + }, + { + "epoch": 3.98, + "learning_rate": 1.0549036265524877e-05, + "loss": 0.1968, + "step": 9515 + }, + { + "epoch": 3.98, + "learning_rate": 1.0547343406018602e-05, + "loss": 0.1663, + "step": 9516 + }, + { + "epoch": 3.98, + "learning_rate": 1.0545650530779428e-05, + "loss": 0.1886, + "step": 9517 + }, + { + "epoch": 3.98, + "learning_rate": 1.0543957639856014e-05, + "loss": 0.1585, + "step": 9518 + }, + { + "epoch": 3.99, + "learning_rate": 1.0542264733297021e-05, + "loss": 0.1914, + "step": 9519 + }, + { + "epoch": 3.99, + "learning_rate": 1.0540571811151106e-05, + "loss": 0.163, + "step": 9520 + }, + { + "epoch": 3.99, + "learning_rate": 1.0538878873466936e-05, + "loss": 0.2158, + "step": 9521 + }, + { + "epoch": 3.99, + "learning_rate": 1.0537185920293171e-05, + "loss": 0.1514, + "step": 9522 + }, + { + "epoch": 3.99, + "learning_rate": 1.0535492951678471e-05, + "loss": 0.1868, + "step": 9523 + }, + { + "epoch": 3.99, + "learning_rate": 1.0533799967671503e-05, + "loss": 0.1927, + "step": 9524 + }, + { + "epoch": 3.99, + "learning_rate": 1.053210696832093e-05, + "loss": 0.1469, + "step": 9525 + }, + { + "epoch": 3.99, + "learning_rate": 1.0530413953675414e-05, + "loss": 0.1823, + "step": 9526 + }, + { + "epoch": 3.99, + "learning_rate": 1.0528720923783617e-05, + "loss": 0.2002, + "step": 9527 + }, + { + "epoch": 3.99, + "learning_rate": 1.0527027878694212e-05, + "loss": 0.2211, + "step": 9528 + }, + { + "epoch": 3.99, + "learning_rate": 1.0525334818455855e-05, + "loss": 0.1998, + "step": 9529 + }, + { + "epoch": 3.99, + "learning_rate": 1.0523641743117215e-05, + "loss": 0.1809, + "step": 9530 + }, + { + "epoch": 3.99, + "learning_rate": 1.0521948652726955e-05, + "loss": 0.1798, + "step": 9531 + }, + { + "epoch": 3.99, + "learning_rate": 1.0520255547333744e-05, + "loss": 0.1796, + "step": 9532 + }, + { + "epoch": 3.99, + "learning_rate": 1.0518562426986253e-05, + "loss": 0.202, + "step": 9533 + }, + { + "epoch": 3.99, + "learning_rate": 1.0516869291733142e-05, + "loss": 0.1884, + "step": 9534 + }, + { + "epoch": 3.99, + "learning_rate": 1.0515176141623085e-05, + "loss": 0.1454, + "step": 9535 + }, + { + "epoch": 3.99, + "learning_rate": 1.0513482976704746e-05, + "loss": 0.1957, + "step": 9536 + }, + { + "epoch": 3.99, + "learning_rate": 1.0511789797026792e-05, + "loss": 0.2049, + "step": 9537 + }, + { + "epoch": 3.99, + "learning_rate": 1.0510096602637896e-05, + "loss": 0.179, + "step": 9538 + }, + { + "epoch": 3.99, + "learning_rate": 1.0508403393586728e-05, + "loss": 0.1709, + "step": 9539 + }, + { + "epoch": 3.99, + "learning_rate": 1.0506710169921954e-05, + "loss": 0.1893, + "step": 9540 + }, + { + "epoch": 3.99, + "learning_rate": 1.0505016931692244e-05, + "loss": 0.1522, + "step": 9541 + }, + { + "epoch": 3.99, + "learning_rate": 1.0503323678946272e-05, + "loss": 0.1592, + "step": 9542 + }, + { + "epoch": 4.0, + "learning_rate": 1.0501630411732709e-05, + "loss": 0.1703, + "step": 9543 + }, + { + "epoch": 4.0, + "learning_rate": 1.0499937130100222e-05, + "loss": 0.1883, + "step": 9544 + }, + { + "epoch": 4.0, + "learning_rate": 1.0498243834097485e-05, + "loss": 0.1816, + "step": 9545 + }, + { + "epoch": 4.0, + "learning_rate": 1.0496550523773177e-05, + "loss": 0.1701, + "step": 9546 + }, + { + "epoch": 4.0, + "learning_rate": 1.0494857199175961e-05, + "loss": 0.174, + "step": 9547 + }, + { + "epoch": 4.0, + "learning_rate": 1.0493163860354512e-05, + "loss": 0.1885, + "step": 9548 + }, + { + "epoch": 4.0, + "learning_rate": 1.0491470507357508e-05, + "loss": 0.2248, + "step": 9549 + }, + { + "epoch": 4.0, + "learning_rate": 1.0489777140233619e-05, + "loss": 0.1783, + "step": 9550 + }, + { + "epoch": 4.0, + "learning_rate": 1.0488083759031524e-05, + "loss": 0.1625, + "step": 9551 + }, + { + "epoch": 4.0, + "learning_rate": 1.0486390363799891e-05, + "loss": 0.2177, + "step": 9552 + }, + { + "epoch": 4.0, + "learning_rate": 1.04846969545874e-05, + "loss": 0.1887, + "step": 9553 + }, + { + "epoch": 4.0, + "learning_rate": 1.0483003531442726e-05, + "loss": 0.1784, + "step": 9554 + }, + { + "epoch": 4.0, + "learning_rate": 1.048131009441454e-05, + "loss": 0.1487, + "step": 9555 + }, + { + "epoch": 4.0, + "learning_rate": 1.047961664355153e-05, + "loss": 0.1338, + "step": 9556 + }, + { + "epoch": 4.0, + "learning_rate": 1.0477923178902359e-05, + "loss": 0.1734, + "step": 9557 + }, + { + "epoch": 4.0, + "learning_rate": 1.047622970051571e-05, + "loss": 0.133, + "step": 9558 + }, + { + "epoch": 4.0, + "learning_rate": 1.0474536208440263e-05, + "loss": 0.1673, + "step": 9559 + }, + { + "epoch": 4.0, + "learning_rate": 1.0472842702724695e-05, + "loss": 0.1659, + "step": 9560 + }, + { + "epoch": 4.0, + "learning_rate": 1.047114918341768e-05, + "loss": 0.1201, + "step": 9561 + }, + { + "epoch": 4.0, + "learning_rate": 1.0469455650567903e-05, + "loss": 0.1705, + "step": 9562 + }, + { + "epoch": 4.0, + "learning_rate": 1.0467762104224039e-05, + "loss": 0.164, + "step": 9563 + }, + { + "epoch": 4.0, + "learning_rate": 1.0466068544434766e-05, + "loss": 0.1443, + "step": 9564 + }, + { + "epoch": 4.0, + "learning_rate": 1.046437497124877e-05, + "loss": 0.1925, + "step": 9565 + }, + { + "epoch": 4.01, + "learning_rate": 1.0462681384714723e-05, + "loss": 0.1494, + "step": 9566 + }, + { + "epoch": 4.01, + "learning_rate": 1.0460987784881313e-05, + "loss": 0.1724, + "step": 9567 + }, + { + "epoch": 4.01, + "learning_rate": 1.0459294171797218e-05, + "loss": 0.1702, + "step": 9568 + }, + { + "epoch": 4.01, + "learning_rate": 1.045760054551112e-05, + "loss": 0.1659, + "step": 9569 + }, + { + "epoch": 4.01, + "learning_rate": 1.04559069060717e-05, + "loss": 0.1584, + "step": 9570 + }, + { + "epoch": 4.01, + "learning_rate": 1.045421325352764e-05, + "loss": 0.1523, + "step": 9571 + }, + { + "epoch": 4.01, + "learning_rate": 1.0452519587927623e-05, + "loss": 0.1254, + "step": 9572 + }, + { + "epoch": 4.01, + "learning_rate": 1.0450825909320335e-05, + "loss": 0.136, + "step": 9573 + }, + { + "epoch": 4.01, + "learning_rate": 1.0449132217754455e-05, + "loss": 0.1503, + "step": 9574 + }, + { + "epoch": 4.01, + "learning_rate": 1.0447438513278664e-05, + "loss": 0.1935, + "step": 9575 + }, + { + "epoch": 4.01, + "learning_rate": 1.0445744795941652e-05, + "loss": 0.146, + "step": 9576 + }, + { + "epoch": 4.01, + "learning_rate": 1.0444051065792105e-05, + "loss": 0.192, + "step": 9577 + }, + { + "epoch": 4.01, + "learning_rate": 1.04423573228787e-05, + "loss": 0.0897, + "step": 9578 + }, + { + "epoch": 4.01, + "learning_rate": 1.0440663567250127e-05, + "loss": 0.1551, + "step": 9579 + }, + { + "epoch": 4.01, + "learning_rate": 1.0438969798955072e-05, + "loss": 0.1595, + "step": 9580 + }, + { + "epoch": 4.01, + "learning_rate": 1.0437276018042219e-05, + "loss": 0.1983, + "step": 9581 + }, + { + "epoch": 4.01, + "learning_rate": 1.0435582224560255e-05, + "loss": 0.1687, + "step": 9582 + }, + { + "epoch": 4.01, + "learning_rate": 1.0433888418557867e-05, + "loss": 0.1532, + "step": 9583 + }, + { + "epoch": 4.01, + "learning_rate": 1.043219460008374e-05, + "loss": 0.1476, + "step": 9584 + }, + { + "epoch": 4.01, + "learning_rate": 1.0430500769186565e-05, + "loss": 0.1043, + "step": 9585 + }, + { + "epoch": 4.01, + "learning_rate": 1.0428806925915026e-05, + "loss": 0.1767, + "step": 9586 + }, + { + "epoch": 4.01, + "learning_rate": 1.0427113070317813e-05, + "loss": 0.1923, + "step": 9587 + }, + { + "epoch": 4.01, + "learning_rate": 1.0425419202443614e-05, + "loss": 0.1675, + "step": 9588 + }, + { + "epoch": 4.01, + "learning_rate": 1.0423725322341115e-05, + "loss": 0.1778, + "step": 9589 + }, + { + "epoch": 4.02, + "learning_rate": 1.0422031430059011e-05, + "loss": 0.1971, + "step": 9590 + }, + { + "epoch": 4.02, + "learning_rate": 1.0420337525645984e-05, + "loss": 0.1652, + "step": 9591 + }, + { + "epoch": 4.02, + "learning_rate": 1.0418643609150732e-05, + "loss": 0.1324, + "step": 9592 + }, + { + "epoch": 4.02, + "learning_rate": 1.041694968062194e-05, + "loss": 0.1142, + "step": 9593 + }, + { + "epoch": 4.02, + "learning_rate": 1.0415255740108299e-05, + "loss": 0.1709, + "step": 9594 + }, + { + "epoch": 4.02, + "learning_rate": 1.0413561787658497e-05, + "loss": 0.1278, + "step": 9595 + }, + { + "epoch": 4.02, + "learning_rate": 1.0411867823321233e-05, + "loss": 0.1172, + "step": 9596 + }, + { + "epoch": 4.02, + "learning_rate": 1.0410173847145195e-05, + "loss": 0.1617, + "step": 9597 + }, + { + "epoch": 4.02, + "learning_rate": 1.040847985917907e-05, + "loss": 0.1151, + "step": 9598 + }, + { + "epoch": 4.02, + "learning_rate": 1.0406785859471554e-05, + "loss": 0.1156, + "step": 9599 + }, + { + "epoch": 4.02, + "learning_rate": 1.0405091848071342e-05, + "loss": 0.1262, + "step": 9600 + }, + { + "epoch": 4.02, + "learning_rate": 1.0403397825027122e-05, + "loss": 0.157, + "step": 9601 + }, + { + "epoch": 4.02, + "learning_rate": 1.0401703790387589e-05, + "loss": 0.175, + "step": 9602 + }, + { + "epoch": 4.02, + "learning_rate": 1.0400009744201438e-05, + "loss": 0.1851, + "step": 9603 + }, + { + "epoch": 4.02, + "learning_rate": 1.0398315686517362e-05, + "loss": 0.1766, + "step": 9604 + }, + { + "epoch": 4.02, + "learning_rate": 1.0396621617384055e-05, + "loss": 0.1459, + "step": 9605 + }, + { + "epoch": 4.02, + "learning_rate": 1.039492753685021e-05, + "loss": 0.1324, + "step": 9606 + }, + { + "epoch": 4.02, + "learning_rate": 1.0393233444964528e-05, + "loss": 0.1526, + "step": 9607 + }, + { + "epoch": 4.02, + "learning_rate": 1.0391539341775696e-05, + "loss": 0.1222, + "step": 9608 + }, + { + "epoch": 4.02, + "learning_rate": 1.0389845227332413e-05, + "loss": 0.1661, + "step": 9609 + }, + { + "epoch": 4.02, + "learning_rate": 1.0388151101683375e-05, + "loss": 0.1224, + "step": 9610 + }, + { + "epoch": 4.02, + "learning_rate": 1.0386456964877281e-05, + "loss": 0.154, + "step": 9611 + }, + { + "epoch": 4.02, + "learning_rate": 1.038476281696282e-05, + "loss": 0.1473, + "step": 9612 + }, + { + "epoch": 4.02, + "learning_rate": 1.0383068657988695e-05, + "loss": 0.0998, + "step": 9613 + }, + { + "epoch": 4.03, + "learning_rate": 1.0381374488003603e-05, + "loss": 0.1953, + "step": 9614 + }, + { + "epoch": 4.03, + "learning_rate": 1.0379680307056238e-05, + "loss": 0.1348, + "step": 9615 + }, + { + "epoch": 4.03, + "learning_rate": 1.03779861151953e-05, + "loss": 0.1649, + "step": 9616 + }, + { + "epoch": 4.03, + "learning_rate": 1.0376291912469488e-05, + "loss": 0.1258, + "step": 9617 + }, + { + "epoch": 4.03, + "learning_rate": 1.0374597698927496e-05, + "loss": 0.1482, + "step": 9618 + }, + { + "epoch": 4.03, + "learning_rate": 1.0372903474618027e-05, + "loss": 0.1671, + "step": 9619 + }, + { + "epoch": 4.03, + "learning_rate": 1.0371209239589779e-05, + "loss": 0.1486, + "step": 9620 + }, + { + "epoch": 4.03, + "learning_rate": 1.0369514993891451e-05, + "loss": 0.1308, + "step": 9621 + }, + { + "epoch": 4.03, + "learning_rate": 1.036782073757174e-05, + "loss": 0.1495, + "step": 9622 + }, + { + "epoch": 4.03, + "learning_rate": 1.036612647067935e-05, + "loss": 0.0969, + "step": 9623 + }, + { + "epoch": 4.03, + "learning_rate": 1.036443219326298e-05, + "loss": 0.184, + "step": 9624 + }, + { + "epoch": 4.03, + "learning_rate": 1.036273790537133e-05, + "loss": 0.127, + "step": 9625 + }, + { + "epoch": 4.03, + "learning_rate": 1.0361043607053101e-05, + "loss": 0.1336, + "step": 9626 + }, + { + "epoch": 4.03, + "learning_rate": 1.0359349298356993e-05, + "loss": 0.1505, + "step": 9627 + }, + { + "epoch": 4.03, + "learning_rate": 1.0357654979331708e-05, + "loss": 0.1709, + "step": 9628 + }, + { + "epoch": 4.03, + "learning_rate": 1.0355960650025946e-05, + "loss": 0.1573, + "step": 9629 + }, + { + "epoch": 4.03, + "learning_rate": 1.0354266310488416e-05, + "loss": 0.2003, + "step": 9630 + }, + { + "epoch": 4.03, + "learning_rate": 1.0352571960767814e-05, + "loss": 0.1725, + "step": 9631 + }, + { + "epoch": 4.03, + "learning_rate": 1.0350877600912844e-05, + "loss": 0.1442, + "step": 9632 + }, + { + "epoch": 4.03, + "learning_rate": 1.0349183230972204e-05, + "loss": 0.1851, + "step": 9633 + }, + { + "epoch": 4.03, + "learning_rate": 1.0347488850994608e-05, + "loss": 0.1321, + "step": 9634 + }, + { + "epoch": 4.03, + "learning_rate": 1.034579446102875e-05, + "loss": 0.1416, + "step": 9635 + }, + { + "epoch": 4.03, + "learning_rate": 1.0344100061123337e-05, + "loss": 0.1843, + "step": 9636 + }, + { + "epoch": 4.03, + "learning_rate": 1.0342405651327072e-05, + "loss": 0.1499, + "step": 9637 + }, + { + "epoch": 4.04, + "learning_rate": 1.0340711231688662e-05, + "loss": 0.1437, + "step": 9638 + }, + { + "epoch": 4.04, + "learning_rate": 1.033901680225681e-05, + "loss": 0.1823, + "step": 9639 + }, + { + "epoch": 4.04, + "learning_rate": 1.0337322363080219e-05, + "loss": 0.1039, + "step": 9640 + }, + { + "epoch": 4.04, + "learning_rate": 1.0335627914207602e-05, + "loss": 0.1494, + "step": 9641 + }, + { + "epoch": 4.04, + "learning_rate": 1.0333933455687655e-05, + "loss": 0.166, + "step": 9642 + }, + { + "epoch": 4.04, + "learning_rate": 1.0332238987569086e-05, + "loss": 0.1793, + "step": 9643 + }, + { + "epoch": 4.04, + "learning_rate": 1.0330544509900603e-05, + "loss": 0.1315, + "step": 9644 + }, + { + "epoch": 4.04, + "learning_rate": 1.032885002273091e-05, + "loss": 0.1267, + "step": 9645 + }, + { + "epoch": 4.04, + "learning_rate": 1.0327155526108716e-05, + "loss": 0.1621, + "step": 9646 + }, + { + "epoch": 4.04, + "learning_rate": 1.0325461020082726e-05, + "loss": 0.1321, + "step": 9647 + }, + { + "epoch": 4.04, + "learning_rate": 1.0323766504701649e-05, + "loss": 0.1195, + "step": 9648 + }, + { + "epoch": 4.04, + "learning_rate": 1.032207198001419e-05, + "loss": 0.1324, + "step": 9649 + }, + { + "epoch": 4.04, + "learning_rate": 1.0320377446069059e-05, + "loss": 0.1025, + "step": 9650 + }, + { + "epoch": 4.04, + "learning_rate": 1.0318682902914962e-05, + "loss": 0.1751, + "step": 9651 + }, + { + "epoch": 4.04, + "learning_rate": 1.0316988350600608e-05, + "loss": 0.1372, + "step": 9652 + }, + { + "epoch": 4.04, + "learning_rate": 1.0315293789174705e-05, + "loss": 0.2112, + "step": 9653 + }, + { + "epoch": 4.04, + "learning_rate": 1.0313599218685961e-05, + "loss": 0.1442, + "step": 9654 + }, + { + "epoch": 4.04, + "learning_rate": 1.0311904639183088e-05, + "loss": 0.0953, + "step": 9655 + }, + { + "epoch": 4.04, + "learning_rate": 1.031021005071479e-05, + "loss": 0.1931, + "step": 9656 + }, + { + "epoch": 4.04, + "learning_rate": 1.030851545332978e-05, + "loss": 0.1386, + "step": 9657 + }, + { + "epoch": 4.04, + "learning_rate": 1.0306820847076768e-05, + "loss": 0.1469, + "step": 9658 + }, + { + "epoch": 4.04, + "learning_rate": 1.0305126232004464e-05, + "loss": 0.1834, + "step": 9659 + }, + { + "epoch": 4.04, + "learning_rate": 1.0303431608161574e-05, + "loss": 0.1344, + "step": 9660 + }, + { + "epoch": 4.04, + "learning_rate": 1.0301736975596816e-05, + "loss": 0.1186, + "step": 9661 + }, + { + "epoch": 4.05, + "learning_rate": 1.0300042334358893e-05, + "loss": 0.1086, + "step": 9662 + }, + { + "epoch": 4.05, + "learning_rate": 1.029834768449652e-05, + "loss": 0.1411, + "step": 9663 + }, + { + "epoch": 4.05, + "learning_rate": 1.029665302605841e-05, + "loss": 0.1346, + "step": 9664 + }, + { + "epoch": 4.05, + "learning_rate": 1.029495835909327e-05, + "loss": 0.1431, + "step": 9665 + }, + { + "epoch": 4.05, + "learning_rate": 1.0293263683649813e-05, + "loss": 0.1703, + "step": 9666 + }, + { + "epoch": 4.05, + "learning_rate": 1.0291568999776751e-05, + "loss": 0.164, + "step": 9667 + }, + { + "epoch": 4.05, + "learning_rate": 1.0289874307522801e-05, + "loss": 0.131, + "step": 9668 + }, + { + "epoch": 4.05, + "learning_rate": 1.0288179606936666e-05, + "loss": 0.1263, + "step": 9669 + }, + { + "epoch": 4.05, + "learning_rate": 1.0286484898067067e-05, + "loss": 0.1508, + "step": 9670 + }, + { + "epoch": 4.05, + "learning_rate": 1.0284790180962712e-05, + "loss": 0.1621, + "step": 9671 + }, + { + "epoch": 4.05, + "learning_rate": 1.0283095455672319e-05, + "loss": 0.0984, + "step": 9672 + }, + { + "epoch": 4.05, + "learning_rate": 1.0281400722244594e-05, + "loss": 0.1535, + "step": 9673 + }, + { + "epoch": 4.05, + "learning_rate": 1.0279705980728258e-05, + "loss": 0.1426, + "step": 9674 + }, + { + "epoch": 4.05, + "learning_rate": 1.0278011231172024e-05, + "loss": 0.1577, + "step": 9675 + }, + { + "epoch": 4.05, + "learning_rate": 1.02763164736246e-05, + "loss": 0.1846, + "step": 9676 + }, + { + "epoch": 4.05, + "learning_rate": 1.0274621708134705e-05, + "loss": 0.1882, + "step": 9677 + }, + { + "epoch": 4.05, + "learning_rate": 1.0272926934751053e-05, + "loss": 0.0994, + "step": 9678 + }, + { + "epoch": 4.05, + "learning_rate": 1.0271232153522359e-05, + "loss": 0.193, + "step": 9679 + }, + { + "epoch": 4.05, + "learning_rate": 1.0269537364497337e-05, + "loss": 0.1684, + "step": 9680 + }, + { + "epoch": 4.05, + "learning_rate": 1.0267842567724702e-05, + "loss": 0.1234, + "step": 9681 + }, + { + "epoch": 4.05, + "learning_rate": 1.0266147763253171e-05, + "loss": 0.1486, + "step": 9682 + }, + { + "epoch": 4.05, + "learning_rate": 1.0264452951131459e-05, + "loss": 0.1551, + "step": 9683 + }, + { + "epoch": 4.05, + "learning_rate": 1.026275813140828e-05, + "loss": 0.1232, + "step": 9684 + }, + { + "epoch": 4.05, + "learning_rate": 1.0261063304132353e-05, + "loss": 0.1866, + "step": 9685 + }, + { + "epoch": 4.06, + "learning_rate": 1.025936846935239e-05, + "loss": 0.1786, + "step": 9686 + }, + { + "epoch": 4.06, + "learning_rate": 1.0257673627117115e-05, + "loss": 0.1304, + "step": 9687 + }, + { + "epoch": 4.06, + "learning_rate": 1.0255978777475238e-05, + "loss": 0.1339, + "step": 9688 + }, + { + "epoch": 4.06, + "learning_rate": 1.025428392047548e-05, + "loss": 0.1277, + "step": 9689 + }, + { + "epoch": 4.06, + "learning_rate": 1.0252589056166554e-05, + "loss": 0.1696, + "step": 9690 + }, + { + "epoch": 4.06, + "learning_rate": 1.025089418459718e-05, + "loss": 0.1633, + "step": 9691 + }, + { + "epoch": 4.06, + "learning_rate": 1.0249199305816078e-05, + "loss": 0.179, + "step": 9692 + }, + { + "epoch": 4.06, + "learning_rate": 1.024750441987196e-05, + "loss": 0.1495, + "step": 9693 + }, + { + "epoch": 4.06, + "learning_rate": 1.0245809526813547e-05, + "loss": 0.1376, + "step": 9694 + }, + { + "epoch": 4.06, + "learning_rate": 1.0244114626689558e-05, + "loss": 0.1755, + "step": 9695 + }, + { + "epoch": 4.06, + "learning_rate": 1.0242419719548711e-05, + "loss": 0.0943, + "step": 9696 + }, + { + "epoch": 4.06, + "learning_rate": 1.0240724805439724e-05, + "loss": 0.1556, + "step": 9697 + }, + { + "epoch": 4.06, + "learning_rate": 1.0239029884411318e-05, + "loss": 0.1504, + "step": 9698 + }, + { + "epoch": 4.06, + "learning_rate": 1.0237334956512209e-05, + "loss": 0.1953, + "step": 9699 + }, + { + "epoch": 4.06, + "learning_rate": 1.0235640021791115e-05, + "loss": 0.1541, + "step": 9700 + }, + { + "epoch": 4.06, + "learning_rate": 1.0233945080296758e-05, + "loss": 0.1627, + "step": 9701 + }, + { + "epoch": 4.06, + "learning_rate": 1.0232250132077861e-05, + "loss": 0.1356, + "step": 9702 + }, + { + "epoch": 4.06, + "learning_rate": 1.0230555177183136e-05, + "loss": 0.1696, + "step": 9703 + }, + { + "epoch": 4.06, + "learning_rate": 1.0228860215661309e-05, + "loss": 0.1229, + "step": 9704 + }, + { + "epoch": 4.06, + "learning_rate": 1.02271652475611e-05, + "loss": 0.1677, + "step": 9705 + }, + { + "epoch": 4.06, + "learning_rate": 1.0225470272931224e-05, + "loss": 0.0818, + "step": 9706 + }, + { + "epoch": 4.06, + "learning_rate": 1.0223775291820406e-05, + "loss": 0.163, + "step": 9707 + }, + { + "epoch": 4.06, + "learning_rate": 1.0222080304277366e-05, + "loss": 0.1689, + "step": 9708 + }, + { + "epoch": 4.06, + "learning_rate": 1.0220385310350829e-05, + "loss": 0.1454, + "step": 9709 + }, + { + "epoch": 4.07, + "learning_rate": 1.0218690310089508e-05, + "loss": 0.1118, + "step": 9710 + }, + { + "epoch": 4.07, + "learning_rate": 1.0216995303542126e-05, + "loss": 0.1471, + "step": 9711 + }, + { + "epoch": 4.07, + "learning_rate": 1.021530029075741e-05, + "loss": 0.1493, + "step": 9712 + }, + { + "epoch": 4.07, + "learning_rate": 1.0213605271784076e-05, + "loss": 0.156, + "step": 9713 + }, + { + "epoch": 4.07, + "learning_rate": 1.0211910246670848e-05, + "loss": 0.1622, + "step": 9714 + }, + { + "epoch": 4.07, + "learning_rate": 1.0210215215466447e-05, + "loss": 0.1617, + "step": 9715 + }, + { + "epoch": 4.07, + "learning_rate": 1.0208520178219602e-05, + "loss": 0.1603, + "step": 9716 + }, + { + "epoch": 4.07, + "learning_rate": 1.0206825134979024e-05, + "loss": 0.1312, + "step": 9717 + }, + { + "epoch": 4.07, + "learning_rate": 1.020513008579344e-05, + "loss": 0.1757, + "step": 9718 + }, + { + "epoch": 4.07, + "learning_rate": 1.0203435030711577e-05, + "loss": 0.1677, + "step": 9719 + }, + { + "epoch": 4.07, + "learning_rate": 1.0201739969782154e-05, + "loss": 0.1854, + "step": 9720 + }, + { + "epoch": 4.07, + "learning_rate": 1.0200044903053896e-05, + "loss": 0.1221, + "step": 9721 + }, + { + "epoch": 4.07, + "learning_rate": 1.0198349830575522e-05, + "loss": 0.1636, + "step": 9722 + }, + { + "epoch": 4.07, + "learning_rate": 1.0196654752395761e-05, + "loss": 0.1262, + "step": 9723 + }, + { + "epoch": 4.07, + "learning_rate": 1.0194959668563332e-05, + "loss": 0.1673, + "step": 9724 + }, + { + "epoch": 4.07, + "learning_rate": 1.0193264579126957e-05, + "loss": 0.1389, + "step": 9725 + }, + { + "epoch": 4.07, + "learning_rate": 1.0191569484135368e-05, + "loss": 0.1372, + "step": 9726 + }, + { + "epoch": 4.07, + "learning_rate": 1.0189874383637284e-05, + "loss": 0.149, + "step": 9727 + }, + { + "epoch": 4.07, + "learning_rate": 1.0188179277681426e-05, + "loss": 0.1391, + "step": 9728 + }, + { + "epoch": 4.07, + "learning_rate": 1.0186484166316526e-05, + "loss": 0.1838, + "step": 9729 + }, + { + "epoch": 4.07, + "learning_rate": 1.01847890495913e-05, + "loss": 0.1652, + "step": 9730 + }, + { + "epoch": 4.07, + "learning_rate": 1.0183093927554478e-05, + "loss": 0.1445, + "step": 9731 + }, + { + "epoch": 4.07, + "learning_rate": 1.0181398800254785e-05, + "loss": 0.1057, + "step": 9732 + }, + { + "epoch": 4.07, + "learning_rate": 1.0179703667740946e-05, + "loss": 0.1414, + "step": 9733 + }, + { + "epoch": 4.08, + "learning_rate": 1.017800853006168e-05, + "loss": 0.1351, + "step": 9734 + }, + { + "epoch": 4.08, + "learning_rate": 1.0176313387265717e-05, + "loss": 0.1709, + "step": 9735 + }, + { + "epoch": 4.08, + "learning_rate": 1.0174618239401786e-05, + "loss": 0.1435, + "step": 9736 + }, + { + "epoch": 4.08, + "learning_rate": 1.0172923086518607e-05, + "loss": 0.1688, + "step": 9737 + }, + { + "epoch": 4.08, + "learning_rate": 1.0171227928664905e-05, + "loss": 0.1496, + "step": 9738 + }, + { + "epoch": 4.08, + "learning_rate": 1.016953276588941e-05, + "loss": 0.1329, + "step": 9739 + }, + { + "epoch": 4.08, + "learning_rate": 1.0167837598240847e-05, + "loss": 0.1073, + "step": 9740 + }, + { + "epoch": 4.08, + "learning_rate": 1.016614242576794e-05, + "loss": 0.106, + "step": 9741 + }, + { + "epoch": 4.08, + "learning_rate": 1.0164447248519415e-05, + "loss": 0.1252, + "step": 9742 + }, + { + "epoch": 4.08, + "learning_rate": 1.0162752066544007e-05, + "loss": 0.1417, + "step": 9743 + }, + { + "epoch": 4.08, + "learning_rate": 1.0161056879890427e-05, + "loss": 0.1441, + "step": 9744 + }, + { + "epoch": 4.08, + "learning_rate": 1.0159361688607411e-05, + "loss": 0.1247, + "step": 9745 + }, + { + "epoch": 4.08, + "learning_rate": 1.0157666492743687e-05, + "loss": 0.2127, + "step": 9746 + }, + { + "epoch": 4.08, + "learning_rate": 1.0155971292347978e-05, + "loss": 0.1592, + "step": 9747 + }, + { + "epoch": 4.08, + "learning_rate": 1.0154276087469015e-05, + "loss": 0.0748, + "step": 9748 + }, + { + "epoch": 4.08, + "learning_rate": 1.015258087815552e-05, + "loss": 0.1466, + "step": 9749 + }, + { + "epoch": 4.08, + "learning_rate": 1.0150885664456224e-05, + "loss": 0.1454, + "step": 9750 + }, + { + "epoch": 4.08, + "learning_rate": 1.0149190446419854e-05, + "loss": 0.1237, + "step": 9751 + }, + { + "epoch": 4.08, + "learning_rate": 1.0147495224095136e-05, + "loss": 0.1437, + "step": 9752 + }, + { + "epoch": 4.08, + "learning_rate": 1.0145799997530801e-05, + "loss": 0.1919, + "step": 9753 + }, + { + "epoch": 4.08, + "learning_rate": 1.0144104766775574e-05, + "loss": 0.1484, + "step": 9754 + }, + { + "epoch": 4.08, + "learning_rate": 1.014240953187818e-05, + "loss": 0.0954, + "step": 9755 + }, + { + "epoch": 4.08, + "learning_rate": 1.0140714292887355e-05, + "loss": 0.1143, + "step": 9756 + }, + { + "epoch": 4.08, + "learning_rate": 1.0139019049851818e-05, + "loss": 0.1436, + "step": 9757 + }, + { + "epoch": 4.09, + "learning_rate": 1.0137323802820304e-05, + "loss": 0.1065, + "step": 9758 + }, + { + "epoch": 4.09, + "learning_rate": 1.0135628551841538e-05, + "loss": 0.1562, + "step": 9759 + }, + { + "epoch": 4.09, + "learning_rate": 1.0133933296964254e-05, + "loss": 0.1523, + "step": 9760 + }, + { + "epoch": 4.09, + "learning_rate": 1.0132238038237173e-05, + "loss": 0.1694, + "step": 9761 + }, + { + "epoch": 4.09, + "learning_rate": 1.0130542775709025e-05, + "loss": 0.1982, + "step": 9762 + }, + { + "epoch": 4.09, + "learning_rate": 1.0128847509428543e-05, + "loss": 0.1591, + "step": 9763 + }, + { + "epoch": 4.09, + "learning_rate": 1.0127152239444453e-05, + "loss": 0.1483, + "step": 9764 + }, + { + "epoch": 4.09, + "learning_rate": 1.0125456965805485e-05, + "loss": 0.1438, + "step": 9765 + }, + { + "epoch": 4.09, + "learning_rate": 1.012376168856037e-05, + "loss": 0.1265, + "step": 9766 + }, + { + "epoch": 4.09, + "learning_rate": 1.0122066407757833e-05, + "loss": 0.1673, + "step": 9767 + }, + { + "epoch": 4.09, + "learning_rate": 1.0120371123446603e-05, + "loss": 0.1653, + "step": 9768 + }, + { + "epoch": 4.09, + "learning_rate": 1.0118675835675416e-05, + "loss": 0.1294, + "step": 9769 + }, + { + "epoch": 4.09, + "learning_rate": 1.0116980544492996e-05, + "loss": 0.1593, + "step": 9770 + }, + { + "epoch": 4.09, + "learning_rate": 1.0115285249948073e-05, + "loss": 0.1596, + "step": 9771 + }, + { + "epoch": 4.09, + "learning_rate": 1.0113589952089377e-05, + "loss": 0.1486, + "step": 9772 + }, + { + "epoch": 4.09, + "learning_rate": 1.0111894650965644e-05, + "loss": 0.2042, + "step": 9773 + }, + { + "epoch": 4.09, + "learning_rate": 1.0110199346625593e-05, + "loss": 0.1378, + "step": 9774 + }, + { + "epoch": 4.09, + "learning_rate": 1.010850403911796e-05, + "loss": 0.1603, + "step": 9775 + }, + { + "epoch": 4.09, + "learning_rate": 1.0106808728491477e-05, + "loss": 0.1564, + "step": 9776 + }, + { + "epoch": 4.09, + "learning_rate": 1.0105113414794872e-05, + "loss": 0.1183, + "step": 9777 + }, + { + "epoch": 4.09, + "learning_rate": 1.0103418098076874e-05, + "loss": 0.1432, + "step": 9778 + }, + { + "epoch": 4.09, + "learning_rate": 1.0101722778386214e-05, + "loss": 0.1461, + "step": 9779 + }, + { + "epoch": 4.09, + "learning_rate": 1.0100027455771624e-05, + "loss": 0.1482, + "step": 9780 + }, + { + "epoch": 4.1, + "learning_rate": 1.0098332130281831e-05, + "loss": 0.156, + "step": 9781 + }, + { + "epoch": 4.1, + "learning_rate": 1.0096636801965571e-05, + "loss": 0.1696, + "step": 9782 + }, + { + "epoch": 4.1, + "learning_rate": 1.0094941470871571e-05, + "loss": 0.1552, + "step": 9783 + }, + { + "epoch": 4.1, + "learning_rate": 1.0093246137048563e-05, + "loss": 0.1347, + "step": 9784 + }, + { + "epoch": 4.1, + "learning_rate": 1.0091550800545279e-05, + "loss": 0.1917, + "step": 9785 + }, + { + "epoch": 4.1, + "learning_rate": 1.0089855461410446e-05, + "loss": 0.135, + "step": 9786 + }, + { + "epoch": 4.1, + "learning_rate": 1.0088160119692798e-05, + "loss": 0.1596, + "step": 9787 + }, + { + "epoch": 4.1, + "learning_rate": 1.0086464775441064e-05, + "loss": 0.126, + "step": 9788 + }, + { + "epoch": 4.1, + "learning_rate": 1.008476942870398e-05, + "loss": 0.1321, + "step": 9789 + }, + { + "epoch": 4.1, + "learning_rate": 1.0083074079530272e-05, + "loss": 0.1722, + "step": 9790 + }, + { + "epoch": 4.1, + "learning_rate": 1.0081378727968676e-05, + "loss": 0.1506, + "step": 9791 + }, + { + "epoch": 4.1, + "learning_rate": 1.007968337406792e-05, + "loss": 0.1526, + "step": 9792 + }, + { + "epoch": 4.1, + "learning_rate": 1.0077988017876732e-05, + "loss": 0.147, + "step": 9793 + }, + { + "epoch": 4.1, + "learning_rate": 1.0076292659443852e-05, + "loss": 0.1657, + "step": 9794 + }, + { + "epoch": 4.1, + "learning_rate": 1.0074597298818005e-05, + "loss": 0.1669, + "step": 9795 + }, + { + "epoch": 4.1, + "learning_rate": 1.0072901936047923e-05, + "loss": 0.1401, + "step": 9796 + }, + { + "epoch": 4.1, + "learning_rate": 1.0071206571182344e-05, + "loss": 0.1403, + "step": 9797 + }, + { + "epoch": 4.1, + "learning_rate": 1.0069511204269992e-05, + "loss": 0.1263, + "step": 9798 + }, + { + "epoch": 4.1, + "learning_rate": 1.0067815835359605e-05, + "loss": 0.1474, + "step": 9799 + }, + { + "epoch": 4.1, + "learning_rate": 1.0066120464499912e-05, + "loss": 0.1316, + "step": 9800 + }, + { + "epoch": 4.1, + "learning_rate": 1.0064425091739643e-05, + "loss": 0.156, + "step": 9801 + }, + { + "epoch": 4.1, + "learning_rate": 1.0062729717127531e-05, + "loss": 0.1534, + "step": 9802 + }, + { + "epoch": 4.1, + "learning_rate": 1.006103434071231e-05, + "loss": 0.1773, + "step": 9803 + }, + { + "epoch": 4.1, + "learning_rate": 1.0059338962542713e-05, + "loss": 0.1738, + "step": 9804 + }, + { + "epoch": 4.11, + "learning_rate": 1.0057643582667467e-05, + "loss": 0.1704, + "step": 9805 + }, + { + "epoch": 4.11, + "learning_rate": 1.0055948201135308e-05, + "loss": 0.1433, + "step": 9806 + }, + { + "epoch": 4.11, + "learning_rate": 1.0054252817994969e-05, + "loss": 0.1176, + "step": 9807 + }, + { + "epoch": 4.11, + "learning_rate": 1.0052557433295181e-05, + "loss": 0.1655, + "step": 9808 + }, + { + "epoch": 4.11, + "learning_rate": 1.0050862047084673e-05, + "loss": 0.1327, + "step": 9809 + }, + { + "epoch": 4.11, + "learning_rate": 1.0049166659412185e-05, + "loss": 0.1708, + "step": 9810 + }, + { + "epoch": 4.11, + "learning_rate": 1.0047471270326446e-05, + "loss": 0.1619, + "step": 9811 + }, + { + "epoch": 4.11, + "learning_rate": 1.0045775879876183e-05, + "loss": 0.1612, + "step": 9812 + }, + { + "epoch": 4.11, + "learning_rate": 1.0044080488110133e-05, + "loss": 0.1654, + "step": 9813 + }, + { + "epoch": 4.11, + "learning_rate": 1.0042385095077032e-05, + "loss": 0.1458, + "step": 9814 + }, + { + "epoch": 4.11, + "learning_rate": 1.0040689700825606e-05, + "loss": 0.1526, + "step": 9815 + }, + { + "epoch": 4.11, + "learning_rate": 1.003899430540459e-05, + "loss": 0.18, + "step": 9816 + }, + { + "epoch": 4.11, + "learning_rate": 1.0037298908862721e-05, + "loss": 0.1752, + "step": 9817 + }, + { + "epoch": 4.11, + "learning_rate": 1.0035603511248723e-05, + "loss": 0.1416, + "step": 9818 + }, + { + "epoch": 4.11, + "learning_rate": 1.0033908112611337e-05, + "loss": 0.141, + "step": 9819 + }, + { + "epoch": 4.11, + "learning_rate": 1.0032212712999292e-05, + "loss": 0.1333, + "step": 9820 + }, + { + "epoch": 4.11, + "learning_rate": 1.0030517312461323e-05, + "loss": 0.1397, + "step": 9821 + }, + { + "epoch": 4.11, + "learning_rate": 1.0028821911046158e-05, + "loss": 0.1191, + "step": 9822 + }, + { + "epoch": 4.11, + "learning_rate": 1.0027126508802534e-05, + "loss": 0.1196, + "step": 9823 + }, + { + "epoch": 4.11, + "learning_rate": 1.0025431105779184e-05, + "loss": 0.1096, + "step": 9824 + }, + { + "epoch": 4.11, + "learning_rate": 1.0023735702024838e-05, + "loss": 0.1517, + "step": 9825 + }, + { + "epoch": 4.11, + "learning_rate": 1.0022040297588229e-05, + "loss": 0.147, + "step": 9826 + }, + { + "epoch": 4.11, + "learning_rate": 1.0020344892518093e-05, + "loss": 0.1235, + "step": 9827 + }, + { + "epoch": 4.11, + "learning_rate": 1.001864948686316e-05, + "loss": 0.1635, + "step": 9828 + }, + { + "epoch": 4.12, + "learning_rate": 1.0016954080672163e-05, + "loss": 0.1218, + "step": 9829 + }, + { + "epoch": 4.12, + "learning_rate": 1.0015258673993837e-05, + "loss": 0.1867, + "step": 9830 + }, + { + "epoch": 4.12, + "learning_rate": 1.0013563266876916e-05, + "loss": 0.1583, + "step": 9831 + }, + { + "epoch": 4.12, + "learning_rate": 1.001186785937013e-05, + "loss": 0.1475, + "step": 9832 + }, + { + "epoch": 4.12, + "learning_rate": 1.0010172451522211e-05, + "loss": 0.1283, + "step": 9833 + }, + { + "epoch": 4.12, + "learning_rate": 1.0008477043381896e-05, + "loss": 0.1581, + "step": 9834 + }, + { + "epoch": 4.12, + "learning_rate": 1.0006781634997917e-05, + "loss": 0.1534, + "step": 9835 + }, + { + "epoch": 4.12, + "learning_rate": 1.0005086226419003e-05, + "loss": 0.127, + "step": 9836 + }, + { + "epoch": 4.12, + "learning_rate": 1.000339081769389e-05, + "loss": 0.1332, + "step": 9837 + }, + { + "epoch": 4.12, + "learning_rate": 1.0001695408871312e-05, + "loss": 0.142, + "step": 9838 + }, + { + "epoch": 4.12, + "learning_rate": 1e-05, + "loss": 0.1187, + "step": 9839 + }, + { + "epoch": 4.12, + "learning_rate": 9.99830459112869e-06, + "loss": 0.1665, + "step": 9840 + }, + { + "epoch": 4.12, + "learning_rate": 9.996609182306113e-06, + "loss": 0.1124, + "step": 9841 + }, + { + "epoch": 4.12, + "learning_rate": 9.994913773581e-06, + "loss": 0.1864, + "step": 9842 + }, + { + "epoch": 4.12, + "learning_rate": 9.993218365002086e-06, + "loss": 0.153, + "step": 9843 + }, + { + "epoch": 4.12, + "learning_rate": 9.991522956618107e-06, + "loss": 0.1748, + "step": 9844 + }, + { + "epoch": 4.12, + "learning_rate": 9.98982754847779e-06, + "loss": 0.1491, + "step": 9845 + }, + { + "epoch": 4.12, + "learning_rate": 9.988132140629871e-06, + "loss": 0.1691, + "step": 9846 + }, + { + "epoch": 4.12, + "learning_rate": 9.986436733123086e-06, + "loss": 0.1863, + "step": 9847 + }, + { + "epoch": 4.12, + "learning_rate": 9.984741326006163e-06, + "loss": 0.1493, + "step": 9848 + }, + { + "epoch": 4.12, + "learning_rate": 9.983045919327837e-06, + "loss": 0.1843, + "step": 9849 + }, + { + "epoch": 4.12, + "learning_rate": 9.981350513136841e-06, + "loss": 0.1709, + "step": 9850 + }, + { + "epoch": 4.12, + "learning_rate": 9.979655107481912e-06, + "loss": 0.1936, + "step": 9851 + }, + { + "epoch": 4.12, + "learning_rate": 9.977959702411776e-06, + "loss": 0.1036, + "step": 9852 + }, + { + "epoch": 4.13, + "learning_rate": 9.976264297975167e-06, + "loss": 0.1825, + "step": 9853 + }, + { + "epoch": 4.13, + "learning_rate": 9.974568894220821e-06, + "loss": 0.1521, + "step": 9854 + }, + { + "epoch": 4.13, + "learning_rate": 9.97287349119747e-06, + "loss": 0.1488, + "step": 9855 + }, + { + "epoch": 4.13, + "learning_rate": 9.971178088953845e-06, + "loss": 0.1464, + "step": 9856 + }, + { + "epoch": 4.13, + "learning_rate": 9.96948268753868e-06, + "loss": 0.1594, + "step": 9857 + }, + { + "epoch": 4.13, + "learning_rate": 9.967787287000711e-06, + "loss": 0.1306, + "step": 9858 + }, + { + "epoch": 4.13, + "learning_rate": 9.966091887388665e-06, + "loss": 0.1487, + "step": 9859 + }, + { + "epoch": 4.13, + "learning_rate": 9.964396488751278e-06, + "loss": 0.1637, + "step": 9860 + }, + { + "epoch": 4.13, + "learning_rate": 9.962701091137282e-06, + "loss": 0.1266, + "step": 9861 + }, + { + "epoch": 4.13, + "learning_rate": 9.961005694595413e-06, + "loss": 0.1867, + "step": 9862 + }, + { + "epoch": 4.13, + "learning_rate": 9.959310299174397e-06, + "loss": 0.1562, + "step": 9863 + }, + { + "epoch": 4.13, + "learning_rate": 9.957614904922971e-06, + "loss": 0.1554, + "step": 9864 + }, + { + "epoch": 4.13, + "learning_rate": 9.95591951188987e-06, + "loss": 0.1215, + "step": 9865 + }, + { + "epoch": 4.13, + "learning_rate": 9.954224120123822e-06, + "loss": 0.1551, + "step": 9866 + }, + { + "epoch": 4.13, + "learning_rate": 9.95252872967356e-06, + "loss": 0.1345, + "step": 9867 + }, + { + "epoch": 4.13, + "learning_rate": 9.950833340587818e-06, + "loss": 0.2001, + "step": 9868 + }, + { + "epoch": 4.13, + "learning_rate": 9.949137952915327e-06, + "loss": 0.1969, + "step": 9869 + }, + { + "epoch": 4.13, + "learning_rate": 9.94744256670482e-06, + "loss": 0.1802, + "step": 9870 + }, + { + "epoch": 4.13, + "learning_rate": 9.945747182005033e-06, + "loss": 0.1551, + "step": 9871 + }, + { + "epoch": 4.13, + "learning_rate": 9.944051798864692e-06, + "loss": 0.1237, + "step": 9872 + }, + { + "epoch": 4.13, + "learning_rate": 9.942356417332538e-06, + "loss": 0.1763, + "step": 9873 + }, + { + "epoch": 4.13, + "learning_rate": 9.940661037457294e-06, + "loss": 0.1686, + "step": 9874 + }, + { + "epoch": 4.13, + "learning_rate": 9.938965659287693e-06, + "loss": 0.1286, + "step": 9875 + }, + { + "epoch": 4.13, + "learning_rate": 9.937270282872474e-06, + "loss": 0.1735, + "step": 9876 + }, + { + "epoch": 4.14, + "learning_rate": 9.935574908260362e-06, + "loss": 0.128, + "step": 9877 + }, + { + "epoch": 4.14, + "learning_rate": 9.933879535500093e-06, + "loss": 0.1503, + "step": 9878 + }, + { + "epoch": 4.14, + "learning_rate": 9.932184164640399e-06, + "loss": 0.1645, + "step": 9879 + }, + { + "epoch": 4.14, + "learning_rate": 9.93048879573001e-06, + "loss": 0.1779, + "step": 9880 + }, + { + "epoch": 4.14, + "learning_rate": 9.928793428817658e-06, + "loss": 0.1424, + "step": 9881 + }, + { + "epoch": 4.14, + "learning_rate": 9.927098063952078e-06, + "loss": 0.1156, + "step": 9882 + }, + { + "epoch": 4.14, + "learning_rate": 9.925402701181999e-06, + "loss": 0.1213, + "step": 9883 + }, + { + "epoch": 4.14, + "learning_rate": 9.923707340556151e-06, + "loss": 0.1282, + "step": 9884 + }, + { + "epoch": 4.14, + "learning_rate": 9.92201198212327e-06, + "loss": 0.1577, + "step": 9885 + }, + { + "epoch": 4.14, + "learning_rate": 9.920316625932085e-06, + "loss": 0.1339, + "step": 9886 + }, + { + "epoch": 4.14, + "learning_rate": 9.918621272031327e-06, + "loss": 0.1632, + "step": 9887 + }, + { + "epoch": 4.14, + "learning_rate": 9.91692592046973e-06, + "loss": 0.1693, + "step": 9888 + }, + { + "epoch": 4.14, + "learning_rate": 9.915230571296021e-06, + "loss": 0.1508, + "step": 9889 + }, + { + "epoch": 4.14, + "learning_rate": 9.913535224558936e-06, + "loss": 0.1149, + "step": 9890 + }, + { + "epoch": 4.14, + "learning_rate": 9.911839880307202e-06, + "loss": 0.1464, + "step": 9891 + }, + { + "epoch": 4.14, + "learning_rate": 9.910144538589556e-06, + "loss": 0.1719, + "step": 9892 + }, + { + "epoch": 4.14, + "learning_rate": 9.908449199454723e-06, + "loss": 0.1044, + "step": 9893 + }, + { + "epoch": 4.14, + "learning_rate": 9.906753862951437e-06, + "loss": 0.1697, + "step": 9894 + }, + { + "epoch": 4.14, + "learning_rate": 9.90505852912843e-06, + "loss": 0.173, + "step": 9895 + }, + { + "epoch": 4.14, + "learning_rate": 9.903363198034434e-06, + "loss": 0.158, + "step": 9896 + }, + { + "epoch": 4.14, + "learning_rate": 9.901667869718172e-06, + "loss": 0.1206, + "step": 9897 + }, + { + "epoch": 4.14, + "learning_rate": 9.89997254422838e-06, + "loss": 0.1652, + "step": 9898 + }, + { + "epoch": 4.14, + "learning_rate": 9.89827722161379e-06, + "loss": 0.1206, + "step": 9899 + }, + { + "epoch": 4.14, + "learning_rate": 9.896581901923131e-06, + "loss": 0.1813, + "step": 9900 + }, + { + "epoch": 4.15, + "learning_rate": 9.894886585205132e-06, + "loss": 0.1526, + "step": 9901 + }, + { + "epoch": 4.15, + "learning_rate": 9.893191271508526e-06, + "loss": 0.1389, + "step": 9902 + }, + { + "epoch": 4.15, + "learning_rate": 9.891495960882042e-06, + "loss": 0.1889, + "step": 9903 + }, + { + "epoch": 4.15, + "learning_rate": 9.88980065337441e-06, + "loss": 0.1551, + "step": 9904 + }, + { + "epoch": 4.15, + "learning_rate": 9.88810534903436e-06, + "loss": 0.1207, + "step": 9905 + }, + { + "epoch": 4.15, + "learning_rate": 9.886410047910624e-06, + "loss": 0.1304, + "step": 9906 + }, + { + "epoch": 4.15, + "learning_rate": 9.884714750051929e-06, + "loss": 0.1537, + "step": 9907 + }, + { + "epoch": 4.15, + "learning_rate": 9.883019455507008e-06, + "loss": 0.1206, + "step": 9908 + }, + { + "epoch": 4.15, + "learning_rate": 9.881324164324587e-06, + "loss": 0.14, + "step": 9909 + }, + { + "epoch": 4.15, + "learning_rate": 9.879628876553398e-06, + "loss": 0.1357, + "step": 9910 + }, + { + "epoch": 4.15, + "learning_rate": 9.87793359224217e-06, + "loss": 0.1855, + "step": 9911 + }, + { + "epoch": 4.15, + "learning_rate": 9.876238311439635e-06, + "loss": 0.1721, + "step": 9912 + }, + { + "epoch": 4.15, + "learning_rate": 9.874543034194516e-06, + "loss": 0.172, + "step": 9913 + }, + { + "epoch": 4.15, + "learning_rate": 9.872847760555547e-06, + "loss": 0.1699, + "step": 9914 + }, + { + "epoch": 4.15, + "learning_rate": 9.871152490571459e-06, + "loss": 0.1347, + "step": 9915 + }, + { + "epoch": 4.15, + "learning_rate": 9.869457224290975e-06, + "loss": 0.1038, + "step": 9916 + }, + { + "epoch": 4.15, + "learning_rate": 9.867761961762829e-06, + "loss": 0.1226, + "step": 9917 + }, + { + "epoch": 4.15, + "learning_rate": 9.86606670303575e-06, + "loss": 0.1654, + "step": 9918 + }, + { + "epoch": 4.15, + "learning_rate": 9.864371448158463e-06, + "loss": 0.1471, + "step": 9919 + }, + { + "epoch": 4.15, + "learning_rate": 9.8626761971797e-06, + "loss": 0.1598, + "step": 9920 + }, + { + "epoch": 4.15, + "learning_rate": 9.860980950148185e-06, + "loss": 0.185, + "step": 9921 + }, + { + "epoch": 4.15, + "learning_rate": 9.859285707112648e-06, + "loss": 0.1481, + "step": 9922 + }, + { + "epoch": 4.15, + "learning_rate": 9.857590468121823e-06, + "loss": 0.1243, + "step": 9923 + }, + { + "epoch": 4.15, + "learning_rate": 9.855895233224431e-06, + "loss": 0.1544, + "step": 9924 + }, + { + "epoch": 4.16, + "learning_rate": 9.854200002469202e-06, + "loss": 0.094, + "step": 9925 + }, + { + "epoch": 4.16, + "learning_rate": 9.852504775904867e-06, + "loss": 0.1216, + "step": 9926 + }, + { + "epoch": 4.16, + "learning_rate": 9.850809553580148e-06, + "loss": 0.1627, + "step": 9927 + }, + { + "epoch": 4.16, + "learning_rate": 9.849114335543778e-06, + "loss": 0.1341, + "step": 9928 + }, + { + "epoch": 4.16, + "learning_rate": 9.847419121844484e-06, + "loss": 0.1723, + "step": 9929 + }, + { + "epoch": 4.16, + "learning_rate": 9.845723912530989e-06, + "loss": 0.1964, + "step": 9930 + }, + { + "epoch": 4.16, + "learning_rate": 9.844028707652023e-06, + "loss": 0.1807, + "step": 9931 + }, + { + "epoch": 4.16, + "learning_rate": 9.842333507256315e-06, + "loss": 0.19, + "step": 9932 + }, + { + "epoch": 4.16, + "learning_rate": 9.84063831139259e-06, + "loss": 0.1478, + "step": 9933 + }, + { + "epoch": 4.16, + "learning_rate": 9.838943120109576e-06, + "loss": 0.1816, + "step": 9934 + }, + { + "epoch": 4.16, + "learning_rate": 9.837247933455998e-06, + "loss": 0.1445, + "step": 9935 + }, + { + "epoch": 4.16, + "learning_rate": 9.835552751480587e-06, + "loss": 0.1875, + "step": 9936 + }, + { + "epoch": 4.16, + "learning_rate": 9.833857574232061e-06, + "loss": 0.1343, + "step": 9937 + }, + { + "epoch": 4.16, + "learning_rate": 9.832162401759155e-06, + "loss": 0.1191, + "step": 9938 + }, + { + "epoch": 4.16, + "learning_rate": 9.83046723411059e-06, + "loss": 0.1396, + "step": 9939 + }, + { + "epoch": 4.16, + "learning_rate": 9.828772071335095e-06, + "loss": 0.148, + "step": 9940 + }, + { + "epoch": 4.16, + "learning_rate": 9.8270769134814e-06, + "loss": 0.1541, + "step": 9941 + }, + { + "epoch": 4.16, + "learning_rate": 9.82538176059822e-06, + "loss": 0.1429, + "step": 9942 + }, + { + "epoch": 4.16, + "learning_rate": 9.823686612734285e-06, + "loss": 0.1397, + "step": 9943 + }, + { + "epoch": 4.16, + "learning_rate": 9.821991469938325e-06, + "loss": 0.1751, + "step": 9944 + }, + { + "epoch": 4.16, + "learning_rate": 9.82029633225906e-06, + "loss": 0.1865, + "step": 9945 + }, + { + "epoch": 4.16, + "learning_rate": 9.818601199745218e-06, + "loss": 0.101, + "step": 9946 + }, + { + "epoch": 4.16, + "learning_rate": 9.816906072445526e-06, + "loss": 0.1478, + "step": 9947 + }, + { + "epoch": 4.16, + "learning_rate": 9.815210950408703e-06, + "loss": 0.1287, + "step": 9948 + }, + { + "epoch": 4.17, + "learning_rate": 9.813515833683477e-06, + "loss": 0.168, + "step": 9949 + }, + { + "epoch": 4.17, + "learning_rate": 9.811820722318577e-06, + "loss": 0.1253, + "step": 9950 + }, + { + "epoch": 4.17, + "learning_rate": 9.81012561636272e-06, + "loss": 0.1594, + "step": 9951 + }, + { + "epoch": 4.17, + "learning_rate": 9.808430515864633e-06, + "loss": 0.1339, + "step": 9952 + }, + { + "epoch": 4.17, + "learning_rate": 9.806735420873044e-06, + "loss": 0.1805, + "step": 9953 + }, + { + "epoch": 4.17, + "learning_rate": 9.805040331436672e-06, + "loss": 0.1188, + "step": 9954 + }, + { + "epoch": 4.17, + "learning_rate": 9.803345247604242e-06, + "loss": 0.1549, + "step": 9955 + }, + { + "epoch": 4.17, + "learning_rate": 9.80165016942448e-06, + "loss": 0.1802, + "step": 9956 + }, + { + "epoch": 4.17, + "learning_rate": 9.799955096946107e-06, + "loss": 0.1507, + "step": 9957 + }, + { + "epoch": 4.17, + "learning_rate": 9.798260030217846e-06, + "loss": 0.112, + "step": 9958 + }, + { + "epoch": 4.17, + "learning_rate": 9.796564969288425e-06, + "loss": 0.1529, + "step": 9959 + }, + { + "epoch": 4.17, + "learning_rate": 9.79486991420656e-06, + "loss": 0.1983, + "step": 9960 + }, + { + "epoch": 4.17, + "learning_rate": 9.793174865020978e-06, + "loss": 0.124, + "step": 9961 + }, + { + "epoch": 4.17, + "learning_rate": 9.7914798217804e-06, + "loss": 0.1237, + "step": 9962 + }, + { + "epoch": 4.17, + "learning_rate": 9.789784784533553e-06, + "loss": 0.1132, + "step": 9963 + }, + { + "epoch": 4.17, + "learning_rate": 9.788089753329157e-06, + "loss": 0.0795, + "step": 9964 + }, + { + "epoch": 4.17, + "learning_rate": 9.78639472821593e-06, + "loss": 0.1575, + "step": 9965 + }, + { + "epoch": 4.17, + "learning_rate": 9.784699709242595e-06, + "loss": 0.1362, + "step": 9966 + }, + { + "epoch": 4.17, + "learning_rate": 9.78300469645788e-06, + "loss": 0.1231, + "step": 9967 + }, + { + "epoch": 4.17, + "learning_rate": 9.7813096899105e-06, + "loss": 0.1493, + "step": 9968 + }, + { + "epoch": 4.17, + "learning_rate": 9.779614689649176e-06, + "loss": 0.1198, + "step": 9969 + }, + { + "epoch": 4.17, + "learning_rate": 9.777919695722638e-06, + "loss": 0.1976, + "step": 9970 + }, + { + "epoch": 4.17, + "learning_rate": 9.776224708179597e-06, + "loss": 0.1473, + "step": 9971 + }, + { + "epoch": 4.18, + "learning_rate": 9.774529727068778e-06, + "loss": 0.1737, + "step": 9972 + }, + { + "epoch": 4.18, + "learning_rate": 9.772834752438903e-06, + "loss": 0.1593, + "step": 9973 + }, + { + "epoch": 4.18, + "learning_rate": 9.771139784338693e-06, + "loss": 0.1489, + "step": 9974 + }, + { + "epoch": 4.18, + "learning_rate": 9.769444822816865e-06, + "loss": 0.174, + "step": 9975 + }, + { + "epoch": 4.18, + "learning_rate": 9.76774986792214e-06, + "loss": 0.1863, + "step": 9976 + }, + { + "epoch": 4.18, + "learning_rate": 9.766054919703243e-06, + "loss": 0.1683, + "step": 9977 + }, + { + "epoch": 4.18, + "learning_rate": 9.764359978208886e-06, + "loss": 0.1814, + "step": 9978 + }, + { + "epoch": 4.18, + "learning_rate": 9.762665043487793e-06, + "loss": 0.1657, + "step": 9979 + }, + { + "epoch": 4.18, + "learning_rate": 9.760970115588686e-06, + "loss": 0.1244, + "step": 9980 + }, + { + "epoch": 4.18, + "learning_rate": 9.759275194560278e-06, + "loss": 0.1577, + "step": 9981 + }, + { + "epoch": 4.18, + "learning_rate": 9.757580280451289e-06, + "loss": 0.1613, + "step": 9982 + }, + { + "epoch": 4.18, + "learning_rate": 9.755885373310444e-06, + "loss": 0.1368, + "step": 9983 + }, + { + "epoch": 4.18, + "learning_rate": 9.754190473186455e-06, + "loss": 0.1572, + "step": 9984 + }, + { + "epoch": 4.18, + "learning_rate": 9.752495580128042e-06, + "loss": 0.1111, + "step": 9985 + }, + { + "epoch": 4.18, + "learning_rate": 9.750800694183926e-06, + "loss": 0.1778, + "step": 9986 + }, + { + "epoch": 4.18, + "learning_rate": 9.749105815402823e-06, + "loss": 0.1719, + "step": 9987 + }, + { + "epoch": 4.18, + "learning_rate": 9.747410943833451e-06, + "loss": 0.1589, + "step": 9988 + }, + { + "epoch": 4.18, + "learning_rate": 9.745716079524526e-06, + "loss": 0.1702, + "step": 9989 + }, + { + "epoch": 4.18, + "learning_rate": 9.744021222524764e-06, + "loss": 0.1895, + "step": 9990 + }, + { + "epoch": 4.18, + "learning_rate": 9.74232637288289e-06, + "loss": 0.143, + "step": 9991 + }, + { + "epoch": 4.18, + "learning_rate": 9.740631530647611e-06, + "loss": 0.0912, + "step": 9992 + }, + { + "epoch": 4.18, + "learning_rate": 9.73893669586765e-06, + "loss": 0.1575, + "step": 9993 + }, + { + "epoch": 4.18, + "learning_rate": 9.737241868591724e-06, + "loss": 0.1241, + "step": 9994 + }, + { + "epoch": 4.18, + "learning_rate": 9.735547048868545e-06, + "loss": 0.1368, + "step": 9995 + }, + { + "epoch": 4.19, + "learning_rate": 9.733852236746832e-06, + "loss": 0.1415, + "step": 9996 + }, + { + "epoch": 4.19, + "learning_rate": 9.732157432275301e-06, + "loss": 0.1183, + "step": 9997 + }, + { + "epoch": 4.19, + "learning_rate": 9.730462635502666e-06, + "loss": 0.1642, + "step": 9998 + }, + { + "epoch": 4.19, + "learning_rate": 9.728767846477643e-06, + "loss": 0.153, + "step": 9999 + }, + { + "epoch": 4.19, + "learning_rate": 9.72707306524895e-06, + "loss": 0.1753, + "step": 10000 + }, + { + "epoch": 4.19, + "learning_rate": 9.725378291865297e-06, + "loss": 0.1548, + "step": 10001 + }, + { + "epoch": 4.19, + "learning_rate": 9.723683526375403e-06, + "loss": 0.1527, + "step": 10002 + }, + { + "epoch": 4.19, + "learning_rate": 9.721988768827977e-06, + "loss": 0.1509, + "step": 10003 + }, + { + "epoch": 4.19, + "learning_rate": 9.720294019271744e-06, + "loss": 0.1601, + "step": 10004 + }, + { + "epoch": 4.19, + "learning_rate": 9.718599277755406e-06, + "loss": 0.1228, + "step": 10005 + }, + { + "epoch": 4.19, + "learning_rate": 9.716904544327683e-06, + "loss": 0.1562, + "step": 10006 + }, + { + "epoch": 4.19, + "learning_rate": 9.71520981903729e-06, + "loss": 0.1742, + "step": 10007 + }, + { + "epoch": 4.19, + "learning_rate": 9.713515101932935e-06, + "loss": 0.1533, + "step": 10008 + }, + { + "epoch": 4.19, + "learning_rate": 9.711820393063337e-06, + "loss": 0.1423, + "step": 10009 + }, + { + "epoch": 4.19, + "learning_rate": 9.710125692477204e-06, + "loss": 0.1037, + "step": 10010 + }, + { + "epoch": 4.19, + "learning_rate": 9.708431000223252e-06, + "loss": 0.18, + "step": 10011 + }, + { + "epoch": 4.19, + "learning_rate": 9.706736316350192e-06, + "loss": 0.1598, + "step": 10012 + }, + { + "epoch": 4.19, + "learning_rate": 9.705041640906736e-06, + "loss": 0.1305, + "step": 10013 + }, + { + "epoch": 4.19, + "learning_rate": 9.703346973941595e-06, + "loss": 0.1635, + "step": 10014 + }, + { + "epoch": 4.19, + "learning_rate": 9.701652315503484e-06, + "loss": 0.1586, + "step": 10015 + }, + { + "epoch": 4.19, + "learning_rate": 9.69995766564111e-06, + "loss": 0.1495, + "step": 10016 + }, + { + "epoch": 4.19, + "learning_rate": 9.698263024403187e-06, + "loss": 0.1283, + "step": 10017 + }, + { + "epoch": 4.19, + "learning_rate": 9.696568391838428e-06, + "loss": 0.1545, + "step": 10018 + }, + { + "epoch": 4.19, + "learning_rate": 9.69487376799554e-06, + "loss": 0.1423, + "step": 10019 + }, + { + "epoch": 4.2, + "learning_rate": 9.693179152923233e-06, + "loss": 0.1542, + "step": 10020 + }, + { + "epoch": 4.2, + "learning_rate": 9.691484546670222e-06, + "loss": 0.1238, + "step": 10021 + }, + { + "epoch": 4.2, + "learning_rate": 9.689789949285213e-06, + "loss": 0.1384, + "step": 10022 + }, + { + "epoch": 4.2, + "learning_rate": 9.688095360816916e-06, + "loss": 0.1603, + "step": 10023 + }, + { + "epoch": 4.2, + "learning_rate": 9.686400781314042e-06, + "loss": 0.1407, + "step": 10024 + }, + { + "epoch": 4.2, + "learning_rate": 9.684706210825298e-06, + "loss": 0.156, + "step": 10025 + }, + { + "epoch": 4.2, + "learning_rate": 9.683011649399393e-06, + "loss": 0.1402, + "step": 10026 + }, + { + "epoch": 4.2, + "learning_rate": 9.681317097085041e-06, + "loss": 0.1731, + "step": 10027 + }, + { + "epoch": 4.2, + "learning_rate": 9.679622553930943e-06, + "loss": 0.1694, + "step": 10028 + }, + { + "epoch": 4.2, + "learning_rate": 9.677928019985811e-06, + "loss": 0.179, + "step": 10029 + }, + { + "epoch": 4.2, + "learning_rate": 9.676233495298353e-06, + "loss": 0.1417, + "step": 10030 + }, + { + "epoch": 4.2, + "learning_rate": 9.674538979917274e-06, + "loss": 0.1471, + "step": 10031 + }, + { + "epoch": 4.2, + "learning_rate": 9.67284447389129e-06, + "loss": 0.127, + "step": 10032 + }, + { + "epoch": 4.2, + "learning_rate": 9.671149977269095e-06, + "loss": 0.1081, + "step": 10033 + }, + { + "epoch": 4.2, + "learning_rate": 9.669455490099402e-06, + "loss": 0.1672, + "step": 10034 + }, + { + "epoch": 4.2, + "learning_rate": 9.667761012430919e-06, + "loss": 0.2019, + "step": 10035 + }, + { + "epoch": 4.2, + "learning_rate": 9.66606654431235e-06, + "loss": 0.144, + "step": 10036 + }, + { + "epoch": 4.2, + "learning_rate": 9.664372085792401e-06, + "loss": 0.1272, + "step": 10037 + }, + { + "epoch": 4.2, + "learning_rate": 9.662677636919783e-06, + "loss": 0.1226, + "step": 10038 + }, + { + "epoch": 4.2, + "learning_rate": 9.660983197743192e-06, + "loss": 0.1758, + "step": 10039 + }, + { + "epoch": 4.2, + "learning_rate": 9.65928876831134e-06, + "loss": 0.1523, + "step": 10040 + }, + { + "epoch": 4.2, + "learning_rate": 9.657594348672931e-06, + "loss": 0.1713, + "step": 10041 + }, + { + "epoch": 4.2, + "learning_rate": 9.655899938876668e-06, + "loss": 0.1666, + "step": 10042 + }, + { + "epoch": 4.2, + "learning_rate": 9.654205538971254e-06, + "loss": 0.1977, + "step": 10043 + }, + { + "epoch": 4.21, + "learning_rate": 9.652511149005396e-06, + "loss": 0.1409, + "step": 10044 + }, + { + "epoch": 4.21, + "learning_rate": 9.650816769027797e-06, + "loss": 0.1365, + "step": 10045 + }, + { + "epoch": 4.21, + "learning_rate": 9.649122399087162e-06, + "loss": 0.1512, + "step": 10046 + }, + { + "epoch": 4.21, + "learning_rate": 9.647428039232188e-06, + "loss": 0.1452, + "step": 10047 + }, + { + "epoch": 4.21, + "learning_rate": 9.645733689511585e-06, + "loss": 0.148, + "step": 10048 + }, + { + "epoch": 4.21, + "learning_rate": 9.644039349974054e-06, + "loss": 0.1768, + "step": 10049 + }, + { + "epoch": 4.21, + "learning_rate": 9.642345020668294e-06, + "loss": 0.1461, + "step": 10050 + }, + { + "epoch": 4.21, + "learning_rate": 9.64065070164301e-06, + "loss": 0.1729, + "step": 10051 + }, + { + "epoch": 4.21, + "learning_rate": 9.6389563929469e-06, + "loss": 0.1695, + "step": 10052 + }, + { + "epoch": 4.21, + "learning_rate": 9.63726209462867e-06, + "loss": 0.1668, + "step": 10053 + }, + { + "epoch": 4.21, + "learning_rate": 9.635567806737021e-06, + "loss": 0.1861, + "step": 10054 + }, + { + "epoch": 4.21, + "learning_rate": 9.633873529320652e-06, + "loss": 0.1445, + "step": 10055 + }, + { + "epoch": 4.21, + "learning_rate": 9.632179262428264e-06, + "loss": 0.1459, + "step": 10056 + }, + { + "epoch": 4.21, + "learning_rate": 9.630485006108554e-06, + "loss": 0.2078, + "step": 10057 + }, + { + "epoch": 4.21, + "learning_rate": 9.628790760410224e-06, + "loss": 0.2002, + "step": 10058 + }, + { + "epoch": 4.21, + "learning_rate": 9.627096525381978e-06, + "loss": 0.1857, + "step": 10059 + }, + { + "epoch": 4.21, + "learning_rate": 9.625402301072508e-06, + "loss": 0.1554, + "step": 10060 + }, + { + "epoch": 4.21, + "learning_rate": 9.623708087530515e-06, + "loss": 0.1844, + "step": 10061 + }, + { + "epoch": 4.21, + "learning_rate": 9.622013884804704e-06, + "loss": 0.1169, + "step": 10062 + }, + { + "epoch": 4.21, + "learning_rate": 9.620319692943765e-06, + "loss": 0.1747, + "step": 10063 + }, + { + "epoch": 4.21, + "learning_rate": 9.618625511996398e-06, + "loss": 0.1651, + "step": 10064 + }, + { + "epoch": 4.21, + "learning_rate": 9.616931342011307e-06, + "loss": 0.1641, + "step": 10065 + }, + { + "epoch": 4.21, + "learning_rate": 9.615237183037181e-06, + "loss": 0.1532, + "step": 10066 + }, + { + "epoch": 4.21, + "learning_rate": 9.613543035122722e-06, + "loss": 0.1131, + "step": 10067 + }, + { + "epoch": 4.22, + "learning_rate": 9.611848898316626e-06, + "loss": 0.1774, + "step": 10068 + }, + { + "epoch": 4.22, + "learning_rate": 9.610154772667589e-06, + "loss": 0.1478, + "step": 10069 + }, + { + "epoch": 4.22, + "learning_rate": 9.608460658224306e-06, + "loss": 0.1161, + "step": 10070 + }, + { + "epoch": 4.22, + "learning_rate": 9.606766555035475e-06, + "loss": 0.1271, + "step": 10071 + }, + { + "epoch": 4.22, + "learning_rate": 9.60507246314979e-06, + "loss": 0.1846, + "step": 10072 + }, + { + "epoch": 4.22, + "learning_rate": 9.603378382615946e-06, + "loss": 0.0966, + "step": 10073 + }, + { + "epoch": 4.22, + "learning_rate": 9.601684313482638e-06, + "loss": 0.1844, + "step": 10074 + }, + { + "epoch": 4.22, + "learning_rate": 9.599990255798564e-06, + "loss": 0.1447, + "step": 10075 + }, + { + "epoch": 4.22, + "learning_rate": 9.598296209612413e-06, + "loss": 0.1451, + "step": 10076 + }, + { + "epoch": 4.22, + "learning_rate": 9.596602174972883e-06, + "loss": 0.1175, + "step": 10077 + }, + { + "epoch": 4.22, + "learning_rate": 9.594908151928663e-06, + "loss": 0.1086, + "step": 10078 + }, + { + "epoch": 4.22, + "learning_rate": 9.593214140528451e-06, + "loss": 0.1838, + "step": 10079 + }, + { + "epoch": 4.22, + "learning_rate": 9.591520140820936e-06, + "loss": 0.1618, + "step": 10080 + }, + { + "epoch": 4.22, + "learning_rate": 9.58982615285481e-06, + "loss": 0.1676, + "step": 10081 + }, + { + "epoch": 4.22, + "learning_rate": 9.588132176678772e-06, + "loss": 0.1582, + "step": 10082 + }, + { + "epoch": 4.22, + "learning_rate": 9.586438212341505e-06, + "loss": 0.1614, + "step": 10083 + }, + { + "epoch": 4.22, + "learning_rate": 9.584744259891706e-06, + "loss": 0.1732, + "step": 10084 + }, + { + "epoch": 4.22, + "learning_rate": 9.583050319378063e-06, + "loss": 0.1284, + "step": 10085 + }, + { + "epoch": 4.22, + "learning_rate": 9.581356390849272e-06, + "loss": 0.1317, + "step": 10086 + }, + { + "epoch": 4.22, + "learning_rate": 9.579662474354017e-06, + "loss": 0.1301, + "step": 10087 + }, + { + "epoch": 4.22, + "learning_rate": 9.577968569940992e-06, + "loss": 0.1623, + "step": 10088 + }, + { + "epoch": 4.22, + "learning_rate": 9.576274677658888e-06, + "loss": 0.1329, + "step": 10089 + }, + { + "epoch": 4.22, + "learning_rate": 9.574580797556389e-06, + "loss": 0.162, + "step": 10090 + }, + { + "epoch": 4.22, + "learning_rate": 9.57288692968219e-06, + "loss": 0.1418, + "step": 10091 + }, + { + "epoch": 4.23, + "learning_rate": 9.571193074084976e-06, + "loss": 0.1598, + "step": 10092 + }, + { + "epoch": 4.23, + "learning_rate": 9.569499230813437e-06, + "loss": 0.14, + "step": 10093 + }, + { + "epoch": 4.23, + "learning_rate": 9.56780539991626e-06, + "loss": 0.1781, + "step": 10094 + }, + { + "epoch": 4.23, + "learning_rate": 9.566111581442136e-06, + "loss": 0.0888, + "step": 10095 + }, + { + "epoch": 4.23, + "learning_rate": 9.564417775439747e-06, + "loss": 0.1739, + "step": 10096 + }, + { + "epoch": 4.23, + "learning_rate": 9.562723981957781e-06, + "loss": 0.1608, + "step": 10097 + }, + { + "epoch": 4.23, + "learning_rate": 9.56103020104493e-06, + "loss": 0.1476, + "step": 10098 + }, + { + "epoch": 4.23, + "learning_rate": 9.559336432749873e-06, + "loss": 0.1451, + "step": 10099 + }, + { + "epoch": 4.23, + "learning_rate": 9.557642677121306e-06, + "loss": 0.1617, + "step": 10100 + }, + { + "epoch": 4.23, + "learning_rate": 9.5559489342079e-06, + "loss": 0.2092, + "step": 10101 + }, + { + "epoch": 4.23, + "learning_rate": 9.55425520405835e-06, + "loss": 0.1794, + "step": 10102 + }, + { + "epoch": 4.23, + "learning_rate": 9.55256148672134e-06, + "loss": 0.1469, + "step": 10103 + }, + { + "epoch": 4.23, + "learning_rate": 9.55086778224555e-06, + "loss": 0.1448, + "step": 10104 + }, + { + "epoch": 4.23, + "learning_rate": 9.54917409067967e-06, + "loss": 0.1527, + "step": 10105 + }, + { + "epoch": 4.23, + "learning_rate": 9.547480412072379e-06, + "loss": 0.1258, + "step": 10106 + }, + { + "epoch": 4.23, + "learning_rate": 9.545786746472363e-06, + "loss": 0.1697, + "step": 10107 + }, + { + "epoch": 4.23, + "learning_rate": 9.544093093928302e-06, + "loss": 0.1573, + "step": 10108 + }, + { + "epoch": 4.23, + "learning_rate": 9.542399454488883e-06, + "loss": 0.161, + "step": 10109 + }, + { + "epoch": 4.23, + "learning_rate": 9.540705828202783e-06, + "loss": 0.1356, + "step": 10110 + }, + { + "epoch": 4.23, + "learning_rate": 9.539012215118688e-06, + "loss": 0.1737, + "step": 10111 + }, + { + "epoch": 4.23, + "learning_rate": 9.537318615285279e-06, + "loss": 0.1357, + "step": 10112 + }, + { + "epoch": 4.23, + "learning_rate": 9.535625028751235e-06, + "loss": 0.1329, + "step": 10113 + }, + { + "epoch": 4.23, + "learning_rate": 9.533931455565236e-06, + "loss": 0.1777, + "step": 10114 + }, + { + "epoch": 4.23, + "learning_rate": 9.532237895775963e-06, + "loss": 0.1467, + "step": 10115 + }, + { + "epoch": 4.24, + "learning_rate": 9.5305443494321e-06, + "loss": 0.1216, + "step": 10116 + }, + { + "epoch": 4.24, + "learning_rate": 9.52885081658232e-06, + "loss": 0.1411, + "step": 10117 + }, + { + "epoch": 4.24, + "learning_rate": 9.527157297275306e-06, + "loss": 0.1181, + "step": 10118 + }, + { + "epoch": 4.24, + "learning_rate": 9.525463791559737e-06, + "loss": 0.1709, + "step": 10119 + }, + { + "epoch": 4.24, + "learning_rate": 9.52377029948429e-06, + "loss": 0.1502, + "step": 10120 + }, + { + "epoch": 4.24, + "learning_rate": 9.522076821097643e-06, + "loss": 0.1715, + "step": 10121 + }, + { + "epoch": 4.24, + "learning_rate": 9.520383356448476e-06, + "loss": 0.1614, + "step": 10122 + }, + { + "epoch": 4.24, + "learning_rate": 9.518689905585463e-06, + "loss": 0.1279, + "step": 10123 + }, + { + "epoch": 4.24, + "learning_rate": 9.51699646855728e-06, + "loss": 0.1557, + "step": 10124 + }, + { + "epoch": 4.24, + "learning_rate": 9.515303045412604e-06, + "loss": 0.1346, + "step": 10125 + }, + { + "epoch": 4.24, + "learning_rate": 9.513609636200112e-06, + "loss": 0.1711, + "step": 10126 + }, + { + "epoch": 4.24, + "learning_rate": 9.511916240968481e-06, + "loss": 0.1474, + "step": 10127 + }, + { + "epoch": 4.24, + "learning_rate": 9.510222859766383e-06, + "loss": 0.1593, + "step": 10128 + }, + { + "epoch": 4.24, + "learning_rate": 9.508529492642493e-06, + "loss": 0.1478, + "step": 10129 + }, + { + "epoch": 4.24, + "learning_rate": 9.506836139645491e-06, + "loss": 0.175, + "step": 10130 + }, + { + "epoch": 4.24, + "learning_rate": 9.505142800824044e-06, + "loss": 0.1369, + "step": 10131 + }, + { + "epoch": 4.24, + "learning_rate": 9.503449476226826e-06, + "loss": 0.1794, + "step": 10132 + }, + { + "epoch": 4.24, + "learning_rate": 9.501756165902517e-06, + "loss": 0.1383, + "step": 10133 + }, + { + "epoch": 4.24, + "learning_rate": 9.50006286989978e-06, + "loss": 0.1675, + "step": 10134 + }, + { + "epoch": 4.24, + "learning_rate": 9.498369588267295e-06, + "loss": 0.1657, + "step": 10135 + }, + { + "epoch": 4.24, + "learning_rate": 9.49667632105373e-06, + "loss": 0.1763, + "step": 10136 + }, + { + "epoch": 4.24, + "learning_rate": 9.494983068307759e-06, + "loss": 0.1615, + "step": 10137 + }, + { + "epoch": 4.24, + "learning_rate": 9.493289830078049e-06, + "loss": 0.1786, + "step": 10138 + }, + { + "epoch": 4.24, + "learning_rate": 9.491596606413275e-06, + "loss": 0.1827, + "step": 10139 + }, + { + "epoch": 4.25, + "learning_rate": 9.489903397362104e-06, + "loss": 0.146, + "step": 10140 + }, + { + "epoch": 4.25, + "learning_rate": 9.488210202973208e-06, + "loss": 0.1449, + "step": 10141 + }, + { + "epoch": 4.25, + "learning_rate": 9.486517023295257e-06, + "loss": 0.1465, + "step": 10142 + }, + { + "epoch": 4.25, + "learning_rate": 9.484823858376916e-06, + "loss": 0.1521, + "step": 10143 + }, + { + "epoch": 4.25, + "learning_rate": 9.483130708266856e-06, + "loss": 0.1594, + "step": 10144 + }, + { + "epoch": 4.25, + "learning_rate": 9.48143757301375e-06, + "loss": 0.1279, + "step": 10145 + }, + { + "epoch": 4.25, + "learning_rate": 9.479744452666257e-06, + "loss": 0.2213, + "step": 10146 + }, + { + "epoch": 4.25, + "learning_rate": 9.47805134727305e-06, + "loss": 0.133, + "step": 10147 + }, + { + "epoch": 4.25, + "learning_rate": 9.476358256882792e-06, + "loss": 0.1305, + "step": 10148 + }, + { + "epoch": 4.25, + "learning_rate": 9.47466518154415e-06, + "loss": 0.1897, + "step": 10149 + }, + { + "epoch": 4.25, + "learning_rate": 9.472972121305795e-06, + "loss": 0.1382, + "step": 10150 + }, + { + "epoch": 4.25, + "learning_rate": 9.471279076216385e-06, + "loss": 0.1311, + "step": 10151 + }, + { + "epoch": 4.25, + "learning_rate": 9.46958604632459e-06, + "loss": 0.1497, + "step": 10152 + }, + { + "epoch": 4.25, + "learning_rate": 9.467893031679073e-06, + "loss": 0.1205, + "step": 10153 + }, + { + "epoch": 4.25, + "learning_rate": 9.466200032328499e-06, + "loss": 0.162, + "step": 10154 + }, + { + "epoch": 4.25, + "learning_rate": 9.46450704832153e-06, + "loss": 0.1396, + "step": 10155 + }, + { + "epoch": 4.25, + "learning_rate": 9.462814079706832e-06, + "loss": 0.1599, + "step": 10156 + }, + { + "epoch": 4.25, + "learning_rate": 9.461121126533069e-06, + "loss": 0.1311, + "step": 10157 + }, + { + "epoch": 4.25, + "learning_rate": 9.459428188848897e-06, + "loss": 0.1662, + "step": 10158 + }, + { + "epoch": 4.25, + "learning_rate": 9.457735266702984e-06, + "loss": 0.1174, + "step": 10159 + }, + { + "epoch": 4.25, + "learning_rate": 9.45604236014399e-06, + "loss": 0.1673, + "step": 10160 + }, + { + "epoch": 4.25, + "learning_rate": 9.454349469220573e-06, + "loss": 0.1501, + "step": 10161 + }, + { + "epoch": 4.25, + "learning_rate": 9.452656593981398e-06, + "loss": 0.1671, + "step": 10162 + }, + { + "epoch": 4.25, + "learning_rate": 9.450963734475125e-06, + "loss": 0.1354, + "step": 10163 + }, + { + "epoch": 4.26, + "learning_rate": 9.44927089075041e-06, + "loss": 0.2105, + "step": 10164 + }, + { + "epoch": 4.26, + "learning_rate": 9.447578062855915e-06, + "loss": 0.1904, + "step": 10165 + }, + { + "epoch": 4.26, + "learning_rate": 9.445885250840301e-06, + "loss": 0.1371, + "step": 10166 + }, + { + "epoch": 4.26, + "learning_rate": 9.444192454752221e-06, + "loss": 0.1447, + "step": 10167 + }, + { + "epoch": 4.26, + "learning_rate": 9.44249967464034e-06, + "loss": 0.1443, + "step": 10168 + }, + { + "epoch": 4.26, + "learning_rate": 9.440806910553308e-06, + "loss": 0.1216, + "step": 10169 + }, + { + "epoch": 4.26, + "learning_rate": 9.439114162539784e-06, + "loss": 0.138, + "step": 10170 + }, + { + "epoch": 4.26, + "learning_rate": 9.437421430648426e-06, + "loss": 0.1481, + "step": 10171 + }, + { + "epoch": 4.26, + "learning_rate": 9.43572871492789e-06, + "loss": 0.152, + "step": 10172 + }, + { + "epoch": 4.26, + "learning_rate": 9.434036015426831e-06, + "loss": 0.1472, + "step": 10173 + }, + { + "epoch": 4.26, + "learning_rate": 9.432343332193907e-06, + "loss": 0.1496, + "step": 10174 + }, + { + "epoch": 4.26, + "learning_rate": 9.430650665277767e-06, + "loss": 0.1479, + "step": 10175 + }, + { + "epoch": 4.26, + "learning_rate": 9.428958014727068e-06, + "loss": 0.1076, + "step": 10176 + }, + { + "epoch": 4.26, + "learning_rate": 9.427265380590466e-06, + "loss": 0.1564, + "step": 10177 + }, + { + "epoch": 4.26, + "learning_rate": 9.425572762916613e-06, + "loss": 0.1509, + "step": 10178 + }, + { + "epoch": 4.26, + "learning_rate": 9.423880161754158e-06, + "loss": 0.1453, + "step": 10179 + }, + { + "epoch": 4.26, + "learning_rate": 9.42218757715176e-06, + "loss": 0.1317, + "step": 10180 + }, + { + "epoch": 4.26, + "learning_rate": 9.420495009158066e-06, + "loss": 0.1405, + "step": 10181 + }, + { + "epoch": 4.26, + "learning_rate": 9.418802457821729e-06, + "loss": 0.1627, + "step": 10182 + }, + { + "epoch": 4.26, + "learning_rate": 9.4171099231914e-06, + "loss": 0.1302, + "step": 10183 + }, + { + "epoch": 4.26, + "learning_rate": 9.415417405315727e-06, + "loss": 0.1721, + "step": 10184 + }, + { + "epoch": 4.26, + "learning_rate": 9.413724904243363e-06, + "loss": 0.164, + "step": 10185 + }, + { + "epoch": 4.26, + "learning_rate": 9.412032420022956e-06, + "loss": 0.1351, + "step": 10186 + }, + { + "epoch": 4.27, + "learning_rate": 9.410339952703157e-06, + "loss": 0.116, + "step": 10187 + }, + { + "epoch": 4.27, + "learning_rate": 9.408647502332612e-06, + "loss": 0.1532, + "step": 10188 + }, + { + "epoch": 4.27, + "learning_rate": 9.40695506895997e-06, + "loss": 0.1497, + "step": 10189 + }, + { + "epoch": 4.27, + "learning_rate": 9.405262652633879e-06, + "loss": 0.1906, + "step": 10190 + }, + { + "epoch": 4.27, + "learning_rate": 9.403570253402987e-06, + "loss": 0.1771, + "step": 10191 + }, + { + "epoch": 4.27, + "learning_rate": 9.401877871315936e-06, + "loss": 0.1526, + "step": 10192 + }, + { + "epoch": 4.27, + "learning_rate": 9.400185506421373e-06, + "loss": 0.1485, + "step": 10193 + }, + { + "epoch": 4.27, + "learning_rate": 9.39849315876795e-06, + "loss": 0.0871, + "step": 10194 + }, + { + "epoch": 4.27, + "learning_rate": 9.396800828404304e-06, + "loss": 0.1601, + "step": 10195 + }, + { + "epoch": 4.27, + "learning_rate": 9.395108515379082e-06, + "loss": 0.1222, + "step": 10196 + }, + { + "epoch": 4.27, + "learning_rate": 9.393416219740929e-06, + "loss": 0.1923, + "step": 10197 + }, + { + "epoch": 4.27, + "learning_rate": 9.391723941538492e-06, + "loss": 0.1477, + "step": 10198 + }, + { + "epoch": 4.27, + "learning_rate": 9.390031680820409e-06, + "loss": 0.1737, + "step": 10199 + }, + { + "epoch": 4.27, + "learning_rate": 9.388339437635321e-06, + "loss": 0.1635, + "step": 10200 + }, + { + "epoch": 4.27, + "learning_rate": 9.386647212031877e-06, + "loss": 0.1653, + "step": 10201 + }, + { + "epoch": 4.27, + "learning_rate": 9.384955004058714e-06, + "loss": 0.1863, + "step": 10202 + }, + { + "epoch": 4.27, + "learning_rate": 9.383262813764471e-06, + "loss": 0.1611, + "step": 10203 + }, + { + "epoch": 4.27, + "learning_rate": 9.381570641197795e-06, + "loss": 0.1518, + "step": 10204 + }, + { + "epoch": 4.27, + "learning_rate": 9.37987848640732e-06, + "loss": 0.183, + "step": 10205 + }, + { + "epoch": 4.27, + "learning_rate": 9.378186349441685e-06, + "loss": 0.1258, + "step": 10206 + }, + { + "epoch": 4.27, + "learning_rate": 9.376494230349537e-06, + "loss": 0.1622, + "step": 10207 + }, + { + "epoch": 4.27, + "learning_rate": 9.374802129179505e-06, + "loss": 0.1771, + "step": 10208 + }, + { + "epoch": 4.27, + "learning_rate": 9.373110045980232e-06, + "loss": 0.1735, + "step": 10209 + }, + { + "epoch": 4.27, + "learning_rate": 9.371417980800357e-06, + "loss": 0.1723, + "step": 10210 + }, + { + "epoch": 4.28, + "learning_rate": 9.369725933688512e-06, + "loss": 0.1516, + "step": 10211 + }, + { + "epoch": 4.28, + "learning_rate": 9.368033904693334e-06, + "loss": 0.1744, + "step": 10212 + }, + { + "epoch": 4.28, + "learning_rate": 9.366341893863467e-06, + "loss": 0.1628, + "step": 10213 + }, + { + "epoch": 4.28, + "learning_rate": 9.364649901247534e-06, + "loss": 0.1779, + "step": 10214 + }, + { + "epoch": 4.28, + "learning_rate": 9.362957926894179e-06, + "loss": 0.1468, + "step": 10215 + }, + { + "epoch": 4.28, + "learning_rate": 9.361265970852031e-06, + "loss": 0.1671, + "step": 10216 + }, + { + "epoch": 4.28, + "learning_rate": 9.359574033169725e-06, + "loss": 0.1538, + "step": 10217 + }, + { + "epoch": 4.28, + "learning_rate": 9.357882113895899e-06, + "loss": 0.1673, + "step": 10218 + }, + { + "epoch": 4.28, + "learning_rate": 9.356190213079178e-06, + "loss": 0.1574, + "step": 10219 + }, + { + "epoch": 4.28, + "learning_rate": 9.354498330768198e-06, + "loss": 0.15, + "step": 10220 + }, + { + "epoch": 4.28, + "learning_rate": 9.352806467011595e-06, + "loss": 0.133, + "step": 10221 + }, + { + "epoch": 4.28, + "learning_rate": 9.35111462185799e-06, + "loss": 0.158, + "step": 10222 + }, + { + "epoch": 4.28, + "learning_rate": 9.349422795356022e-06, + "loss": 0.1445, + "step": 10223 + }, + { + "epoch": 4.28, + "learning_rate": 9.34773098755432e-06, + "loss": 0.2036, + "step": 10224 + }, + { + "epoch": 4.28, + "learning_rate": 9.34603919850151e-06, + "loss": 0.1615, + "step": 10225 + }, + { + "epoch": 4.28, + "learning_rate": 9.344347428246222e-06, + "loss": 0.1625, + "step": 10226 + }, + { + "epoch": 4.28, + "learning_rate": 9.342655676837085e-06, + "loss": 0.1402, + "step": 10227 + }, + { + "epoch": 4.28, + "learning_rate": 9.340963944322731e-06, + "loss": 0.185, + "step": 10228 + }, + { + "epoch": 4.28, + "learning_rate": 9.33927223075178e-06, + "loss": 0.1517, + "step": 10229 + }, + { + "epoch": 4.28, + "learning_rate": 9.337580536172864e-06, + "loss": 0.1605, + "step": 10230 + }, + { + "epoch": 4.28, + "learning_rate": 9.33588886063461e-06, + "loss": 0.1153, + "step": 10231 + }, + { + "epoch": 4.28, + "learning_rate": 9.334197204185638e-06, + "loss": 0.1657, + "step": 10232 + }, + { + "epoch": 4.28, + "learning_rate": 9.332505566874577e-06, + "loss": 0.1653, + "step": 10233 + }, + { + "epoch": 4.28, + "learning_rate": 9.330813948750053e-06, + "loss": 0.1501, + "step": 10234 + }, + { + "epoch": 4.29, + "learning_rate": 9.329122349860686e-06, + "loss": 0.1349, + "step": 10235 + }, + { + "epoch": 4.29, + "learning_rate": 9.327430770255106e-06, + "loss": 0.1312, + "step": 10236 + }, + { + "epoch": 4.29, + "learning_rate": 9.325739209981929e-06, + "loss": 0.155, + "step": 10237 + }, + { + "epoch": 4.29, + "learning_rate": 9.324047669089778e-06, + "loss": 0.1361, + "step": 10238 + }, + { + "epoch": 4.29, + "learning_rate": 9.32235614762728e-06, + "loss": 0.1375, + "step": 10239 + }, + { + "epoch": 4.29, + "learning_rate": 9.320664645643051e-06, + "loss": 0.1078, + "step": 10240 + }, + { + "epoch": 4.29, + "learning_rate": 9.318973163185713e-06, + "loss": 0.1587, + "step": 10241 + }, + { + "epoch": 4.29, + "learning_rate": 9.31728170030389e-06, + "loss": 0.172, + "step": 10242 + }, + { + "epoch": 4.29, + "learning_rate": 9.315590257046197e-06, + "loss": 0.1524, + "step": 10243 + }, + { + "epoch": 4.29, + "learning_rate": 9.313898833461255e-06, + "loss": 0.1059, + "step": 10244 + }, + { + "epoch": 4.29, + "learning_rate": 9.312207429597684e-06, + "loss": 0.1627, + "step": 10245 + }, + { + "epoch": 4.29, + "learning_rate": 9.310516045504096e-06, + "loss": 0.1688, + "step": 10246 + }, + { + "epoch": 4.29, + "learning_rate": 9.308824681229115e-06, + "loss": 0.1488, + "step": 10247 + }, + { + "epoch": 4.29, + "learning_rate": 9.307133336821357e-06, + "loss": 0.1414, + "step": 10248 + }, + { + "epoch": 4.29, + "learning_rate": 9.305442012329433e-06, + "loss": 0.1393, + "step": 10249 + }, + { + "epoch": 4.29, + "learning_rate": 9.303750707801963e-06, + "loss": 0.1709, + "step": 10250 + }, + { + "epoch": 4.29, + "learning_rate": 9.302059423287563e-06, + "loss": 0.1623, + "step": 10251 + }, + { + "epoch": 4.29, + "learning_rate": 9.300368158834843e-06, + "loss": 0.1603, + "step": 10252 + }, + { + "epoch": 4.29, + "learning_rate": 9.298676914492419e-06, + "loss": 0.174, + "step": 10253 + }, + { + "epoch": 4.29, + "learning_rate": 9.29698569030891e-06, + "loss": 0.1618, + "step": 10254 + }, + { + "epoch": 4.29, + "learning_rate": 9.295294486332916e-06, + "loss": 0.1749, + "step": 10255 + }, + { + "epoch": 4.29, + "learning_rate": 9.29360330261306e-06, + "loss": 0.175, + "step": 10256 + }, + { + "epoch": 4.29, + "learning_rate": 9.29191213919795e-06, + "loss": 0.1675, + "step": 10257 + }, + { + "epoch": 4.29, + "learning_rate": 9.290220996136199e-06, + "loss": 0.1704, + "step": 10258 + }, + { + "epoch": 4.3, + "learning_rate": 9.288529873476418e-06, + "loss": 0.1415, + "step": 10259 + }, + { + "epoch": 4.3, + "learning_rate": 9.28683877126721e-06, + "loss": 0.1458, + "step": 10260 + }, + { + "epoch": 4.3, + "learning_rate": 9.285147689557188e-06, + "loss": 0.1277, + "step": 10261 + }, + { + "epoch": 4.3, + "learning_rate": 9.283456628394966e-06, + "loss": 0.1414, + "step": 10262 + }, + { + "epoch": 4.3, + "learning_rate": 9.281765587829143e-06, + "loss": 0.1546, + "step": 10263 + }, + { + "epoch": 4.3, + "learning_rate": 9.28007456790833e-06, + "loss": 0.1535, + "step": 10264 + }, + { + "epoch": 4.3, + "learning_rate": 9.278383568681139e-06, + "loss": 0.161, + "step": 10265 + }, + { + "epoch": 4.3, + "learning_rate": 9.27669259019617e-06, + "loss": 0.1447, + "step": 10266 + }, + { + "epoch": 4.3, + "learning_rate": 9.275001632502029e-06, + "loss": 0.1506, + "step": 10267 + }, + { + "epoch": 4.3, + "learning_rate": 9.273310695647323e-06, + "loss": 0.1517, + "step": 10268 + }, + { + "epoch": 4.3, + "learning_rate": 9.271619779680658e-06, + "loss": 0.1645, + "step": 10269 + }, + { + "epoch": 4.3, + "learning_rate": 9.269928884650635e-06, + "loss": 0.1579, + "step": 10270 + }, + { + "epoch": 4.3, + "learning_rate": 9.268238010605858e-06, + "loss": 0.1617, + "step": 10271 + }, + { + "epoch": 4.3, + "learning_rate": 9.266547157594932e-06, + "loss": 0.1459, + "step": 10272 + }, + { + "epoch": 4.3, + "learning_rate": 9.264856325666455e-06, + "loss": 0.1122, + "step": 10273 + }, + { + "epoch": 4.3, + "learning_rate": 9.26316551486903e-06, + "loss": 0.1507, + "step": 10274 + }, + { + "epoch": 4.3, + "learning_rate": 9.261474725251261e-06, + "loss": 0.1768, + "step": 10275 + }, + { + "epoch": 4.3, + "learning_rate": 9.259783956861744e-06, + "loss": 0.1558, + "step": 10276 + }, + { + "epoch": 4.3, + "learning_rate": 9.258093209749081e-06, + "loss": 0.1558, + "step": 10277 + }, + { + "epoch": 4.3, + "learning_rate": 9.25640248396187e-06, + "loss": 0.1407, + "step": 10278 + }, + { + "epoch": 4.3, + "learning_rate": 9.254711779548709e-06, + "loss": 0.1367, + "step": 10279 + }, + { + "epoch": 4.3, + "learning_rate": 9.253021096558197e-06, + "loss": 0.1778, + "step": 10280 + }, + { + "epoch": 4.3, + "learning_rate": 9.251330435038935e-06, + "loss": 0.1635, + "step": 10281 + }, + { + "epoch": 4.3, + "learning_rate": 9.24963979503951e-06, + "loss": 0.111, + "step": 10282 + }, + { + "epoch": 4.31, + "learning_rate": 9.247949176608527e-06, + "loss": 0.1337, + "step": 10283 + }, + { + "epoch": 4.31, + "learning_rate": 9.246258579794574e-06, + "loss": 0.1602, + "step": 10284 + }, + { + "epoch": 4.31, + "learning_rate": 9.244568004646249e-06, + "loss": 0.1061, + "step": 10285 + }, + { + "epoch": 4.31, + "learning_rate": 9.24287745121215e-06, + "loss": 0.1671, + "step": 10286 + }, + { + "epoch": 4.31, + "learning_rate": 9.241186919540864e-06, + "loss": 0.202, + "step": 10287 + }, + { + "epoch": 4.31, + "learning_rate": 9.239496409680986e-06, + "loss": 0.1655, + "step": 10288 + }, + { + "epoch": 4.31, + "learning_rate": 9.237805921681112e-06, + "loss": 0.1347, + "step": 10289 + }, + { + "epoch": 4.31, + "learning_rate": 9.236115455589827e-06, + "loss": 0.1724, + "step": 10290 + }, + { + "epoch": 4.31, + "learning_rate": 9.234425011455727e-06, + "loss": 0.1192, + "step": 10291 + }, + { + "epoch": 4.31, + "learning_rate": 9.232734589327402e-06, + "loss": 0.1302, + "step": 10292 + }, + { + "epoch": 4.31, + "learning_rate": 9.231044189253438e-06, + "loss": 0.144, + "step": 10293 + }, + { + "epoch": 4.31, + "learning_rate": 9.229353811282427e-06, + "loss": 0.1895, + "step": 10294 + }, + { + "epoch": 4.31, + "learning_rate": 9.227663455462957e-06, + "loss": 0.1672, + "step": 10295 + }, + { + "epoch": 4.31, + "learning_rate": 9.225973121843615e-06, + "loss": 0.1519, + "step": 10296 + }, + { + "epoch": 4.31, + "learning_rate": 9.22428281047299e-06, + "loss": 0.1131, + "step": 10297 + }, + { + "epoch": 4.31, + "learning_rate": 9.222592521399666e-06, + "loss": 0.1864, + "step": 10298 + }, + { + "epoch": 4.31, + "learning_rate": 9.220902254672232e-06, + "loss": 0.1448, + "step": 10299 + }, + { + "epoch": 4.31, + "learning_rate": 9.21921201033927e-06, + "loss": 0.1724, + "step": 10300 + }, + { + "epoch": 4.31, + "learning_rate": 9.217521788449365e-06, + "loss": 0.1048, + "step": 10301 + }, + { + "epoch": 4.31, + "learning_rate": 9.215831589051103e-06, + "loss": 0.141, + "step": 10302 + }, + { + "epoch": 4.31, + "learning_rate": 9.214141412193065e-06, + "loss": 0.1361, + "step": 10303 + }, + { + "epoch": 4.31, + "learning_rate": 9.212451257923841e-06, + "loss": 0.1285, + "step": 10304 + }, + { + "epoch": 4.31, + "learning_rate": 9.210761126292e-06, + "loss": 0.1157, + "step": 10305 + }, + { + "epoch": 4.31, + "learning_rate": 9.209071017346134e-06, + "loss": 0.1308, + "step": 10306 + }, + { + "epoch": 4.32, + "learning_rate": 9.207380931134816e-06, + "loss": 0.1388, + "step": 10307 + }, + { + "epoch": 4.32, + "learning_rate": 9.205690867706631e-06, + "loss": 0.1371, + "step": 10308 + }, + { + "epoch": 4.32, + "learning_rate": 9.204000827110157e-06, + "loss": 0.1566, + "step": 10309 + }, + { + "epoch": 4.32, + "learning_rate": 9.202310809393973e-06, + "loss": 0.1379, + "step": 10310 + }, + { + "epoch": 4.32, + "learning_rate": 9.200620814606659e-06, + "loss": 0.1757, + "step": 10311 + }, + { + "epoch": 4.32, + "learning_rate": 9.198930842796786e-06, + "loss": 0.1404, + "step": 10312 + }, + { + "epoch": 4.32, + "learning_rate": 9.197240894012939e-06, + "loss": 0.1312, + "step": 10313 + }, + { + "epoch": 4.32, + "learning_rate": 9.195550968303688e-06, + "loss": 0.1814, + "step": 10314 + }, + { + "epoch": 4.32, + "learning_rate": 9.19386106571761e-06, + "loss": 0.1651, + "step": 10315 + }, + { + "epoch": 4.32, + "learning_rate": 9.192171186303283e-06, + "loss": 0.1172, + "step": 10316 + }, + { + "epoch": 4.32, + "learning_rate": 9.190481330109276e-06, + "loss": 0.1769, + "step": 10317 + }, + { + "epoch": 4.32, + "learning_rate": 9.188791497184165e-06, + "loss": 0.1593, + "step": 10318 + }, + { + "epoch": 4.32, + "learning_rate": 9.187101687576525e-06, + "loss": 0.1895, + "step": 10319 + }, + { + "epoch": 4.32, + "learning_rate": 9.185411901334924e-06, + "loss": 0.1344, + "step": 10320 + }, + { + "epoch": 4.32, + "learning_rate": 9.183722138507934e-06, + "loss": 0.1776, + "step": 10321 + }, + { + "epoch": 4.32, + "learning_rate": 9.18203239914413e-06, + "loss": 0.2088, + "step": 10322 + }, + { + "epoch": 4.32, + "learning_rate": 9.180342683292074e-06, + "loss": 0.1396, + "step": 10323 + }, + { + "epoch": 4.32, + "learning_rate": 9.178652991000342e-06, + "loss": 0.1484, + "step": 10324 + }, + { + "epoch": 4.32, + "learning_rate": 9.176963322317503e-06, + "loss": 0.1141, + "step": 10325 + }, + { + "epoch": 4.32, + "learning_rate": 9.175273677292122e-06, + "loss": 0.1434, + "step": 10326 + }, + { + "epoch": 4.32, + "learning_rate": 9.17358405597277e-06, + "loss": 0.1797, + "step": 10327 + }, + { + "epoch": 4.32, + "learning_rate": 9.171894458408007e-06, + "loss": 0.1074, + "step": 10328 + }, + { + "epoch": 4.32, + "learning_rate": 9.170204884646402e-06, + "loss": 0.1534, + "step": 10329 + }, + { + "epoch": 4.32, + "learning_rate": 9.168515334736526e-06, + "loss": 0.1112, + "step": 10330 + }, + { + "epoch": 4.33, + "learning_rate": 9.166825808726934e-06, + "loss": 0.1436, + "step": 10331 + }, + { + "epoch": 4.33, + "learning_rate": 9.165136306666196e-06, + "loss": 0.1286, + "step": 10332 + }, + { + "epoch": 4.33, + "learning_rate": 9.163446828602876e-06, + "loss": 0.1368, + "step": 10333 + }, + { + "epoch": 4.33, + "learning_rate": 9.161757374585533e-06, + "loss": 0.1154, + "step": 10334 + }, + { + "epoch": 4.33, + "learning_rate": 9.16006794466273e-06, + "loss": 0.1594, + "step": 10335 + }, + { + "epoch": 4.33, + "learning_rate": 9.158378538883032e-06, + "loss": 0.1529, + "step": 10336 + }, + { + "epoch": 4.33, + "learning_rate": 9.156689157294992e-06, + "loss": 0.1316, + "step": 10337 + }, + { + "epoch": 4.33, + "learning_rate": 9.154999799947174e-06, + "loss": 0.1473, + "step": 10338 + }, + { + "epoch": 4.33, + "learning_rate": 9.153310466888138e-06, + "loss": 0.1211, + "step": 10339 + }, + { + "epoch": 4.33, + "learning_rate": 9.151621158166443e-06, + "loss": 0.1527, + "step": 10340 + }, + { + "epoch": 4.33, + "learning_rate": 9.149931873830642e-06, + "loss": 0.1493, + "step": 10341 + }, + { + "epoch": 4.33, + "learning_rate": 9.148242613929295e-06, + "loss": 0.1741, + "step": 10342 + }, + { + "epoch": 4.33, + "learning_rate": 9.146553378510961e-06, + "loss": 0.1669, + "step": 10343 + }, + { + "epoch": 4.33, + "learning_rate": 9.144864167624191e-06, + "loss": 0.1436, + "step": 10344 + }, + { + "epoch": 4.33, + "learning_rate": 9.143174981317542e-06, + "loss": 0.1587, + "step": 10345 + }, + { + "epoch": 4.33, + "learning_rate": 9.141485819639569e-06, + "loss": 0.2177, + "step": 10346 + }, + { + "epoch": 4.33, + "learning_rate": 9.139796682638824e-06, + "loss": 0.1276, + "step": 10347 + }, + { + "epoch": 4.33, + "learning_rate": 9.138107570363857e-06, + "loss": 0.1801, + "step": 10348 + }, + { + "epoch": 4.33, + "learning_rate": 9.13641848286323e-06, + "loss": 0.1598, + "step": 10349 + }, + { + "epoch": 4.33, + "learning_rate": 9.134729420185481e-06, + "loss": 0.1672, + "step": 10350 + }, + { + "epoch": 4.33, + "learning_rate": 9.13304038237917e-06, + "loss": 0.1669, + "step": 10351 + }, + { + "epoch": 4.33, + "learning_rate": 9.13135136949284e-06, + "loss": 0.1917, + "step": 10352 + }, + { + "epoch": 4.33, + "learning_rate": 9.129662381575047e-06, + "loss": 0.1498, + "step": 10353 + }, + { + "epoch": 4.33, + "learning_rate": 9.127973418674338e-06, + "loss": 0.1808, + "step": 10354 + }, + { + "epoch": 4.34, + "learning_rate": 9.126284480839256e-06, + "loss": 0.1474, + "step": 10355 + }, + { + "epoch": 4.34, + "learning_rate": 9.124595568118352e-06, + "loss": 0.1349, + "step": 10356 + }, + { + "epoch": 4.34, + "learning_rate": 9.122906680560173e-06, + "loss": 0.1752, + "step": 10357 + }, + { + "epoch": 4.34, + "learning_rate": 9.121217818213263e-06, + "loss": 0.126, + "step": 10358 + }, + { + "epoch": 4.34, + "learning_rate": 9.119528981126165e-06, + "loss": 0.1602, + "step": 10359 + }, + { + "epoch": 4.34, + "learning_rate": 9.117840169347428e-06, + "loss": 0.1327, + "step": 10360 + }, + { + "epoch": 4.34, + "learning_rate": 9.11615138292559e-06, + "loss": 0.1333, + "step": 10361 + }, + { + "epoch": 4.34, + "learning_rate": 9.114462621909196e-06, + "loss": 0.1549, + "step": 10362 + }, + { + "epoch": 4.34, + "learning_rate": 9.11277388634679e-06, + "loss": 0.2197, + "step": 10363 + }, + { + "epoch": 4.34, + "learning_rate": 9.111085176286912e-06, + "loss": 0.1698, + "step": 10364 + }, + { + "epoch": 4.34, + "learning_rate": 9.109396491778099e-06, + "loss": 0.1588, + "step": 10365 + }, + { + "epoch": 4.34, + "learning_rate": 9.107707832868896e-06, + "loss": 0.1549, + "step": 10366 + }, + { + "epoch": 4.34, + "learning_rate": 9.106019199607838e-06, + "loss": 0.1317, + "step": 10367 + }, + { + "epoch": 4.34, + "learning_rate": 9.104330592043464e-06, + "loss": 0.1306, + "step": 10368 + }, + { + "epoch": 4.34, + "learning_rate": 9.102642010224313e-06, + "loss": 0.1434, + "step": 10369 + }, + { + "epoch": 4.34, + "learning_rate": 9.100953454198924e-06, + "loss": 0.1234, + "step": 10370 + }, + { + "epoch": 4.34, + "learning_rate": 9.099264924015826e-06, + "loss": 0.1644, + "step": 10371 + }, + { + "epoch": 4.34, + "learning_rate": 9.097576419723564e-06, + "loss": 0.1702, + "step": 10372 + }, + { + "epoch": 4.34, + "learning_rate": 9.095887941370662e-06, + "loss": 0.1733, + "step": 10373 + }, + { + "epoch": 4.34, + "learning_rate": 9.094199489005664e-06, + "loss": 0.0816, + "step": 10374 + }, + { + "epoch": 4.34, + "learning_rate": 9.092511062677092e-06, + "loss": 0.1213, + "step": 10375 + }, + { + "epoch": 4.34, + "learning_rate": 9.090822662433485e-06, + "loss": 0.1588, + "step": 10376 + }, + { + "epoch": 4.34, + "learning_rate": 9.089134288323377e-06, + "loss": 0.1595, + "step": 10377 + }, + { + "epoch": 4.34, + "learning_rate": 9.087445940395292e-06, + "loss": 0.1875, + "step": 10378 + }, + { + "epoch": 4.35, + "learning_rate": 9.085757618697765e-06, + "loss": 0.1474, + "step": 10379 + }, + { + "epoch": 4.35, + "learning_rate": 9.084069323279325e-06, + "loss": 0.1392, + "step": 10380 + }, + { + "epoch": 4.35, + "learning_rate": 9.0823810541885e-06, + "loss": 0.1543, + "step": 10381 + }, + { + "epoch": 4.35, + "learning_rate": 9.080692811473815e-06, + "loss": 0.1155, + "step": 10382 + }, + { + "epoch": 4.35, + "learning_rate": 9.079004595183799e-06, + "loss": 0.1661, + "step": 10383 + }, + { + "epoch": 4.35, + "learning_rate": 9.07731640536698e-06, + "loss": 0.1738, + "step": 10384 + }, + { + "epoch": 4.35, + "learning_rate": 9.075628242071884e-06, + "loss": 0.1154, + "step": 10385 + }, + { + "epoch": 4.35, + "learning_rate": 9.073940105347032e-06, + "loss": 0.1177, + "step": 10386 + }, + { + "epoch": 4.35, + "learning_rate": 9.07225199524095e-06, + "loss": 0.1713, + "step": 10387 + }, + { + "epoch": 4.35, + "learning_rate": 9.07056391180216e-06, + "loss": 0.1511, + "step": 10388 + }, + { + "epoch": 4.35, + "learning_rate": 9.068875855079186e-06, + "loss": 0.1077, + "step": 10389 + }, + { + "epoch": 4.35, + "learning_rate": 9.067187825120552e-06, + "loss": 0.1106, + "step": 10390 + }, + { + "epoch": 4.35, + "learning_rate": 9.065499821974774e-06, + "loss": 0.1994, + "step": 10391 + }, + { + "epoch": 4.35, + "learning_rate": 9.063811845690374e-06, + "loss": 0.1749, + "step": 10392 + }, + { + "epoch": 4.35, + "learning_rate": 9.062123896315872e-06, + "loss": 0.1925, + "step": 10393 + }, + { + "epoch": 4.35, + "learning_rate": 9.06043597389979e-06, + "loss": 0.1688, + "step": 10394 + }, + { + "epoch": 4.35, + "learning_rate": 9.05874807849064e-06, + "loss": 0.1216, + "step": 10395 + }, + { + "epoch": 4.35, + "learning_rate": 9.05706021013694e-06, + "loss": 0.1676, + "step": 10396 + }, + { + "epoch": 4.35, + "learning_rate": 9.055372368887207e-06, + "loss": 0.1076, + "step": 10397 + }, + { + "epoch": 4.35, + "learning_rate": 9.053684554789959e-06, + "loss": 0.1369, + "step": 10398 + }, + { + "epoch": 4.35, + "learning_rate": 9.051996767893707e-06, + "loss": 0.1862, + "step": 10399 + }, + { + "epoch": 4.35, + "learning_rate": 9.050309008246965e-06, + "loss": 0.1327, + "step": 10400 + }, + { + "epoch": 4.35, + "learning_rate": 9.048621275898252e-06, + "loss": 0.1802, + "step": 10401 + }, + { + "epoch": 4.36, + "learning_rate": 9.046933570896073e-06, + "loss": 0.1715, + "step": 10402 + }, + { + "epoch": 4.36, + "learning_rate": 9.045245893288943e-06, + "loss": 0.1338, + "step": 10403 + }, + { + "epoch": 4.36, + "learning_rate": 9.043558243125374e-06, + "loss": 0.145, + "step": 10404 + }, + { + "epoch": 4.36, + "learning_rate": 9.041870620453874e-06, + "loss": 0.1623, + "step": 10405 + }, + { + "epoch": 4.36, + "learning_rate": 9.040183025322952e-06, + "loss": 0.199, + "step": 10406 + }, + { + "epoch": 4.36, + "learning_rate": 9.03849545778112e-06, + "loss": 0.149, + "step": 10407 + }, + { + "epoch": 4.36, + "learning_rate": 9.03680791787688e-06, + "loss": 0.1117, + "step": 10408 + }, + { + "epoch": 4.36, + "learning_rate": 9.035120405658741e-06, + "loss": 0.1546, + "step": 10409 + }, + { + "epoch": 4.36, + "learning_rate": 9.03343292117521e-06, + "loss": 0.1345, + "step": 10410 + }, + { + "epoch": 4.36, + "learning_rate": 9.031745464474794e-06, + "loss": 0.131, + "step": 10411 + }, + { + "epoch": 4.36, + "learning_rate": 9.030058035605996e-06, + "loss": 0.1347, + "step": 10412 + }, + { + "epoch": 4.36, + "learning_rate": 9.028370634617316e-06, + "loss": 0.1506, + "step": 10413 + }, + { + "epoch": 4.36, + "learning_rate": 9.026683261557263e-06, + "loss": 0.1459, + "step": 10414 + }, + { + "epoch": 4.36, + "learning_rate": 9.024995916474333e-06, + "loss": 0.1284, + "step": 10415 + }, + { + "epoch": 4.36, + "learning_rate": 9.023308599417032e-06, + "loss": 0.1399, + "step": 10416 + }, + { + "epoch": 4.36, + "learning_rate": 9.021621310433861e-06, + "loss": 0.164, + "step": 10417 + }, + { + "epoch": 4.36, + "learning_rate": 9.019934049573316e-06, + "loss": 0.1458, + "step": 10418 + }, + { + "epoch": 4.36, + "learning_rate": 9.018246816883894e-06, + "loss": 0.1446, + "step": 10419 + }, + { + "epoch": 4.36, + "learning_rate": 9.016559612414098e-06, + "loss": 0.151, + "step": 10420 + }, + { + "epoch": 4.36, + "learning_rate": 9.014872436212423e-06, + "loss": 0.1738, + "step": 10421 + }, + { + "epoch": 4.36, + "learning_rate": 9.013185288327365e-06, + "loss": 0.1546, + "step": 10422 + }, + { + "epoch": 4.36, + "learning_rate": 9.01149816880742e-06, + "loss": 0.182, + "step": 10423 + }, + { + "epoch": 4.36, + "learning_rate": 9.009811077701083e-06, + "loss": 0.1314, + "step": 10424 + }, + { + "epoch": 4.36, + "learning_rate": 9.008124015056851e-06, + "loss": 0.1467, + "step": 10425 + }, + { + "epoch": 4.37, + "learning_rate": 9.00643698092321e-06, + "loss": 0.1292, + "step": 10426 + }, + { + "epoch": 4.37, + "learning_rate": 9.004749975348657e-06, + "loss": 0.0717, + "step": 10427 + }, + { + "epoch": 4.37, + "learning_rate": 9.003062998381685e-06, + "loss": 0.1891, + "step": 10428 + }, + { + "epoch": 4.37, + "learning_rate": 9.00137605007078e-06, + "loss": 0.1756, + "step": 10429 + }, + { + "epoch": 4.37, + "learning_rate": 8.999689130464436e-06, + "loss": 0.1271, + "step": 10430 + }, + { + "epoch": 4.37, + "learning_rate": 8.99800223961114e-06, + "loss": 0.156, + "step": 10431 + }, + { + "epoch": 4.37, + "learning_rate": 8.99631537755938e-06, + "loss": 0.1119, + "step": 10432 + }, + { + "epoch": 4.37, + "learning_rate": 8.994628544357641e-06, + "loss": 0.1482, + "step": 10433 + }, + { + "epoch": 4.37, + "learning_rate": 8.992941740054418e-06, + "loss": 0.1023, + "step": 10434 + }, + { + "epoch": 4.37, + "learning_rate": 8.991254964698188e-06, + "loss": 0.1573, + "step": 10435 + }, + { + "epoch": 4.37, + "learning_rate": 8.989568218337438e-06, + "loss": 0.1465, + "step": 10436 + }, + { + "epoch": 4.37, + "learning_rate": 8.987881501020656e-06, + "loss": 0.155, + "step": 10437 + }, + { + "epoch": 4.37, + "learning_rate": 8.98619481279632e-06, + "loss": 0.1424, + "step": 10438 + }, + { + "epoch": 4.37, + "learning_rate": 8.984508153712912e-06, + "loss": 0.0995, + "step": 10439 + }, + { + "epoch": 4.37, + "learning_rate": 8.982821523818923e-06, + "loss": 0.1584, + "step": 10440 + }, + { + "epoch": 4.37, + "learning_rate": 8.981134923162823e-06, + "loss": 0.1327, + "step": 10441 + }, + { + "epoch": 4.37, + "learning_rate": 8.979448351793095e-06, + "loss": 0.1538, + "step": 10442 + }, + { + "epoch": 4.37, + "learning_rate": 8.977761809758218e-06, + "loss": 0.1266, + "step": 10443 + }, + { + "epoch": 4.37, + "learning_rate": 8.97607529710667e-06, + "loss": 0.1584, + "step": 10444 + }, + { + "epoch": 4.37, + "learning_rate": 8.974388813886932e-06, + "loss": 0.1403, + "step": 10445 + }, + { + "epoch": 4.37, + "learning_rate": 8.972702360147474e-06, + "loss": 0.1293, + "step": 10446 + }, + { + "epoch": 4.37, + "learning_rate": 8.971015935936777e-06, + "loss": 0.142, + "step": 10447 + }, + { + "epoch": 4.37, + "learning_rate": 8.969329541303315e-06, + "loss": 0.1632, + "step": 10448 + }, + { + "epoch": 4.37, + "learning_rate": 8.967643176295557e-06, + "loss": 0.1862, + "step": 10449 + }, + { + "epoch": 4.38, + "learning_rate": 8.96595684096198e-06, + "loss": 0.1808, + "step": 10450 + }, + { + "epoch": 4.38, + "learning_rate": 8.964270535351058e-06, + "loss": 0.1826, + "step": 10451 + }, + { + "epoch": 4.38, + "learning_rate": 8.962584259511262e-06, + "loss": 0.1768, + "step": 10452 + }, + { + "epoch": 4.38, + "learning_rate": 8.960898013491058e-06, + "loss": 0.1202, + "step": 10453 + }, + { + "epoch": 4.38, + "learning_rate": 8.959211797338919e-06, + "loss": 0.1497, + "step": 10454 + }, + { + "epoch": 4.38, + "learning_rate": 8.957525611103316e-06, + "loss": 0.1282, + "step": 10455 + }, + { + "epoch": 4.38, + "learning_rate": 8.955839454832711e-06, + "loss": 0.148, + "step": 10456 + }, + { + "epoch": 4.38, + "learning_rate": 8.954153328575574e-06, + "loss": 0.1241, + "step": 10457 + }, + { + "epoch": 4.38, + "learning_rate": 8.952467232380375e-06, + "loss": 0.1631, + "step": 10458 + }, + { + "epoch": 4.38, + "learning_rate": 8.950781166295573e-06, + "loss": 0.1293, + "step": 10459 + }, + { + "epoch": 4.38, + "learning_rate": 8.949095130369636e-06, + "loss": 0.1547, + "step": 10460 + }, + { + "epoch": 4.38, + "learning_rate": 8.947409124651027e-06, + "loss": 0.1538, + "step": 10461 + }, + { + "epoch": 4.38, + "learning_rate": 8.94572314918821e-06, + "loss": 0.1362, + "step": 10462 + }, + { + "epoch": 4.38, + "learning_rate": 8.944037204029646e-06, + "loss": 0.1879, + "step": 10463 + }, + { + "epoch": 4.38, + "learning_rate": 8.942351289223795e-06, + "loss": 0.1804, + "step": 10464 + }, + { + "epoch": 4.38, + "learning_rate": 8.940665404819114e-06, + "loss": 0.1195, + "step": 10465 + }, + { + "epoch": 4.38, + "learning_rate": 8.93897955086407e-06, + "loss": 0.1211, + "step": 10466 + }, + { + "epoch": 4.38, + "learning_rate": 8.937293727407115e-06, + "loss": 0.1555, + "step": 10467 + }, + { + "epoch": 4.38, + "learning_rate": 8.935607934496708e-06, + "loss": 0.1263, + "step": 10468 + }, + { + "epoch": 4.38, + "learning_rate": 8.93392217218131e-06, + "loss": 0.184, + "step": 10469 + }, + { + "epoch": 4.38, + "learning_rate": 8.93223644050937e-06, + "loss": 0.1298, + "step": 10470 + }, + { + "epoch": 4.38, + "learning_rate": 8.930550739529345e-06, + "loss": 0.13, + "step": 10471 + }, + { + "epoch": 4.38, + "learning_rate": 8.928865069289693e-06, + "loss": 0.134, + "step": 10472 + }, + { + "epoch": 4.38, + "learning_rate": 8.927179429838864e-06, + "loss": 0.145, + "step": 10473 + }, + { + "epoch": 4.39, + "learning_rate": 8.925493821225309e-06, + "loss": 0.1614, + "step": 10474 + }, + { + "epoch": 4.39, + "learning_rate": 8.923808243497482e-06, + "loss": 0.1361, + "step": 10475 + }, + { + "epoch": 4.39, + "learning_rate": 8.92212269670383e-06, + "loss": 0.1655, + "step": 10476 + }, + { + "epoch": 4.39, + "learning_rate": 8.920437180892803e-06, + "loss": 0.1206, + "step": 10477 + }, + { + "epoch": 4.39, + "learning_rate": 8.918751696112856e-06, + "loss": 0.1356, + "step": 10478 + }, + { + "epoch": 4.39, + "learning_rate": 8.917066242412429e-06, + "loss": 0.1348, + "step": 10479 + }, + { + "epoch": 4.39, + "learning_rate": 8.915380819839972e-06, + "loss": 0.1483, + "step": 10480 + }, + { + "epoch": 4.39, + "learning_rate": 8.91369542844393e-06, + "loss": 0.1389, + "step": 10481 + }, + { + "epoch": 4.39, + "learning_rate": 8.912010068272752e-06, + "loss": 0.1384, + "step": 10482 + }, + { + "epoch": 4.39, + "learning_rate": 8.910324739374877e-06, + "loss": 0.1575, + "step": 10483 + }, + { + "epoch": 4.39, + "learning_rate": 8.908639441798751e-06, + "loss": 0.1206, + "step": 10484 + }, + { + "epoch": 4.39, + "learning_rate": 8.906954175592819e-06, + "loss": 0.1597, + "step": 10485 + }, + { + "epoch": 4.39, + "learning_rate": 8.905268940805519e-06, + "loss": 0.1651, + "step": 10486 + }, + { + "epoch": 4.39, + "learning_rate": 8.903583737485286e-06, + "loss": 0.1341, + "step": 10487 + }, + { + "epoch": 4.39, + "learning_rate": 8.901898565680567e-06, + "loss": 0.1445, + "step": 10488 + }, + { + "epoch": 4.39, + "learning_rate": 8.900213425439803e-06, + "loss": 0.1532, + "step": 10489 + }, + { + "epoch": 4.39, + "learning_rate": 8.898528316811425e-06, + "loss": 0.1891, + "step": 10490 + }, + { + "epoch": 4.39, + "learning_rate": 8.896843239843873e-06, + "loss": 0.1499, + "step": 10491 + }, + { + "epoch": 4.39, + "learning_rate": 8.895158194585583e-06, + "loss": 0.1366, + "step": 10492 + }, + { + "epoch": 4.39, + "learning_rate": 8.893473181084993e-06, + "loss": 0.1487, + "step": 10493 + }, + { + "epoch": 4.39, + "learning_rate": 8.891788199390533e-06, + "loss": 0.1514, + "step": 10494 + }, + { + "epoch": 4.39, + "learning_rate": 8.890103249550638e-06, + "loss": 0.1496, + "step": 10495 + }, + { + "epoch": 4.39, + "learning_rate": 8.88841833161374e-06, + "loss": 0.1665, + "step": 10496 + }, + { + "epoch": 4.39, + "learning_rate": 8.88673344562827e-06, + "loss": 0.1539, + "step": 10497 + }, + { + "epoch": 4.4, + "learning_rate": 8.88504859164266e-06, + "loss": 0.1255, + "step": 10498 + }, + { + "epoch": 4.4, + "learning_rate": 8.883363769705338e-06, + "loss": 0.1813, + "step": 10499 + }, + { + "epoch": 4.4, + "learning_rate": 8.881678979864734e-06, + "loss": 0.1768, + "step": 10500 + }, + { + "epoch": 4.4, + "learning_rate": 8.879994222169275e-06, + "loss": 0.1486, + "step": 10501 + }, + { + "epoch": 4.4, + "learning_rate": 8.87830949666739e-06, + "loss": 0.1229, + "step": 10502 + }, + { + "epoch": 4.4, + "learning_rate": 8.876624803407501e-06, + "loss": 0.1755, + "step": 10503 + }, + { + "epoch": 4.4, + "learning_rate": 8.874940142438036e-06, + "loss": 0.1787, + "step": 10504 + }, + { + "epoch": 4.4, + "learning_rate": 8.873255513807418e-06, + "loss": 0.1251, + "step": 10505 + }, + { + "epoch": 4.4, + "learning_rate": 8.87157091756407e-06, + "loss": 0.1068, + "step": 10506 + }, + { + "epoch": 4.4, + "learning_rate": 8.869886353756415e-06, + "loss": 0.1293, + "step": 10507 + }, + { + "epoch": 4.4, + "learning_rate": 8.868201822432878e-06, + "loss": 0.1115, + "step": 10508 + }, + { + "epoch": 4.4, + "learning_rate": 8.86651732364187e-06, + "loss": 0.1715, + "step": 10509 + }, + { + "epoch": 4.4, + "learning_rate": 8.86483285743182e-06, + "loss": 0.1266, + "step": 10510 + }, + { + "epoch": 4.4, + "learning_rate": 8.863148423851138e-06, + "loss": 0.1839, + "step": 10511 + }, + { + "epoch": 4.4, + "learning_rate": 8.861464022948247e-06, + "loss": 0.1993, + "step": 10512 + }, + { + "epoch": 4.4, + "learning_rate": 8.859779654771563e-06, + "loss": 0.163, + "step": 10513 + }, + { + "epoch": 4.4, + "learning_rate": 8.858095319369499e-06, + "loss": 0.1587, + "step": 10514 + }, + { + "epoch": 4.4, + "learning_rate": 8.856411016790472e-06, + "loss": 0.1455, + "step": 10515 + }, + { + "epoch": 4.4, + "learning_rate": 8.8547267470829e-06, + "loss": 0.1502, + "step": 10516 + }, + { + "epoch": 4.4, + "learning_rate": 8.853042510295187e-06, + "loss": 0.1293, + "step": 10517 + }, + { + "epoch": 4.4, + "learning_rate": 8.851358306475748e-06, + "loss": 0.1647, + "step": 10518 + }, + { + "epoch": 4.4, + "learning_rate": 8.849674135672999e-06, + "loss": 0.1932, + "step": 10519 + }, + { + "epoch": 4.4, + "learning_rate": 8.847989997935345e-06, + "loss": 0.1486, + "step": 10520 + }, + { + "epoch": 4.4, + "learning_rate": 8.846305893311194e-06, + "loss": 0.1692, + "step": 10521 + }, + { + "epoch": 4.41, + "learning_rate": 8.844621821848958e-06, + "loss": 0.1121, + "step": 10522 + }, + { + "epoch": 4.41, + "learning_rate": 8.842937783597045e-06, + "loss": 0.1358, + "step": 10523 + }, + { + "epoch": 4.41, + "learning_rate": 8.841253778603855e-06, + "loss": 0.1654, + "step": 10524 + }, + { + "epoch": 4.41, + "learning_rate": 8.839569806917797e-06, + "loss": 0.1473, + "step": 10525 + }, + { + "epoch": 4.41, + "learning_rate": 8.837885868587279e-06, + "loss": 0.1257, + "step": 10526 + }, + { + "epoch": 4.41, + "learning_rate": 8.836201963660695e-06, + "loss": 0.1375, + "step": 10527 + }, + { + "epoch": 4.41, + "learning_rate": 8.834518092186456e-06, + "loss": 0.1417, + "step": 10528 + }, + { + "epoch": 4.41, + "learning_rate": 8.83283425421296e-06, + "loss": 0.139, + "step": 10529 + }, + { + "epoch": 4.41, + "learning_rate": 8.831150449788613e-06, + "loss": 0.1568, + "step": 10530 + }, + { + "epoch": 4.41, + "learning_rate": 8.829466678961803e-06, + "loss": 0.189, + "step": 10531 + }, + { + "epoch": 4.41, + "learning_rate": 8.827782941780934e-06, + "loss": 0.1277, + "step": 10532 + }, + { + "epoch": 4.41, + "learning_rate": 8.826099238294404e-06, + "loss": 0.1296, + "step": 10533 + }, + { + "epoch": 4.41, + "learning_rate": 8.824415568550612e-06, + "loss": 0.1962, + "step": 10534 + }, + { + "epoch": 4.41, + "learning_rate": 8.82273193259795e-06, + "loss": 0.1299, + "step": 10535 + }, + { + "epoch": 4.41, + "learning_rate": 8.821048330484814e-06, + "loss": 0.0992, + "step": 10536 + }, + { + "epoch": 4.41, + "learning_rate": 8.819364762259598e-06, + "loss": 0.1378, + "step": 10537 + }, + { + "epoch": 4.41, + "learning_rate": 8.817681227970693e-06, + "loss": 0.1181, + "step": 10538 + }, + { + "epoch": 4.41, + "learning_rate": 8.815997727666493e-06, + "loss": 0.1476, + "step": 10539 + }, + { + "epoch": 4.41, + "learning_rate": 8.814314261395389e-06, + "loss": 0.1612, + "step": 10540 + }, + { + "epoch": 4.41, + "learning_rate": 8.812630829205766e-06, + "loss": 0.1735, + "step": 10541 + }, + { + "epoch": 4.41, + "learning_rate": 8.810947431146019e-06, + "loss": 0.1622, + "step": 10542 + }, + { + "epoch": 4.41, + "learning_rate": 8.809264067264533e-06, + "loss": 0.1423, + "step": 10543 + }, + { + "epoch": 4.41, + "learning_rate": 8.807580737609693e-06, + "loss": 0.151, + "step": 10544 + }, + { + "epoch": 4.41, + "learning_rate": 8.805897442229888e-06, + "loss": 0.1505, + "step": 10545 + }, + { + "epoch": 4.42, + "learning_rate": 8.804214181173501e-06, + "loss": 0.1616, + "step": 10546 + }, + { + "epoch": 4.42, + "learning_rate": 8.802530954488915e-06, + "loss": 0.1944, + "step": 10547 + }, + { + "epoch": 4.42, + "learning_rate": 8.800847762224514e-06, + "loss": 0.1288, + "step": 10548 + }, + { + "epoch": 4.42, + "learning_rate": 8.799164604428683e-06, + "loss": 0.123, + "step": 10549 + }, + { + "epoch": 4.42, + "learning_rate": 8.797481481149798e-06, + "loss": 0.1643, + "step": 10550 + }, + { + "epoch": 4.42, + "learning_rate": 8.795798392436239e-06, + "loss": 0.1086, + "step": 10551 + }, + { + "epoch": 4.42, + "learning_rate": 8.794115338336388e-06, + "loss": 0.1179, + "step": 10552 + }, + { + "epoch": 4.42, + "learning_rate": 8.792432318898624e-06, + "loss": 0.1645, + "step": 10553 + }, + { + "epoch": 4.42, + "learning_rate": 8.790749334171323e-06, + "loss": 0.1417, + "step": 10554 + }, + { + "epoch": 4.42, + "learning_rate": 8.789066384202853e-06, + "loss": 0.1391, + "step": 10555 + }, + { + "epoch": 4.42, + "learning_rate": 8.787383469041598e-06, + "loss": 0.1525, + "step": 10556 + }, + { + "epoch": 4.42, + "learning_rate": 8.78570058873593e-06, + "loss": 0.1494, + "step": 10557 + }, + { + "epoch": 4.42, + "learning_rate": 8.78401774333422e-06, + "loss": 0.1371, + "step": 10558 + }, + { + "epoch": 4.42, + "learning_rate": 8.78233493288484e-06, + "loss": 0.1782, + "step": 10559 + }, + { + "epoch": 4.42, + "learning_rate": 8.780652157436163e-06, + "loss": 0.1544, + "step": 10560 + }, + { + "epoch": 4.42, + "learning_rate": 8.778969417036557e-06, + "loss": 0.1573, + "step": 10561 + }, + { + "epoch": 4.42, + "learning_rate": 8.77728671173439e-06, + "loss": 0.1482, + "step": 10562 + }, + { + "epoch": 4.42, + "learning_rate": 8.775604041578034e-06, + "loss": 0.1629, + "step": 10563 + }, + { + "epoch": 4.42, + "learning_rate": 8.773921406615854e-06, + "loss": 0.1126, + "step": 10564 + }, + { + "epoch": 4.42, + "learning_rate": 8.772238806896213e-06, + "loss": 0.1316, + "step": 10565 + }, + { + "epoch": 4.42, + "learning_rate": 8.770556242467479e-06, + "loss": 0.1752, + "step": 10566 + }, + { + "epoch": 4.42, + "learning_rate": 8.768873713378015e-06, + "loss": 0.1287, + "step": 10567 + }, + { + "epoch": 4.42, + "learning_rate": 8.767191219676184e-06, + "loss": 0.1922, + "step": 10568 + }, + { + "epoch": 4.42, + "learning_rate": 8.765508761410345e-06, + "loss": 0.1686, + "step": 10569 + }, + { + "epoch": 4.43, + "learning_rate": 8.763826338628865e-06, + "loss": 0.1586, + "step": 10570 + }, + { + "epoch": 4.43, + "learning_rate": 8.762143951380096e-06, + "loss": 0.168, + "step": 10571 + }, + { + "epoch": 4.43, + "learning_rate": 8.760461599712402e-06, + "loss": 0.1332, + "step": 10572 + }, + { + "epoch": 4.43, + "learning_rate": 8.758779283674142e-06, + "loss": 0.1216, + "step": 10573 + }, + { + "epoch": 4.43, + "learning_rate": 8.757097003313665e-06, + "loss": 0.1546, + "step": 10574 + }, + { + "epoch": 4.43, + "learning_rate": 8.755414758679336e-06, + "loss": 0.1556, + "step": 10575 + }, + { + "epoch": 4.43, + "learning_rate": 8.753732549819508e-06, + "loss": 0.1918, + "step": 10576 + }, + { + "epoch": 4.43, + "learning_rate": 8.752050376782525e-06, + "loss": 0.141, + "step": 10577 + }, + { + "epoch": 4.43, + "learning_rate": 8.750368239616752e-06, + "loss": 0.1161, + "step": 10578 + }, + { + "epoch": 4.43, + "learning_rate": 8.748686138370533e-06, + "loss": 0.1649, + "step": 10579 + }, + { + "epoch": 4.43, + "learning_rate": 8.747004073092218e-06, + "loss": 0.1614, + "step": 10580 + }, + { + "epoch": 4.43, + "learning_rate": 8.745322043830163e-06, + "loss": 0.1292, + "step": 10581 + }, + { + "epoch": 4.43, + "learning_rate": 8.743640050632709e-06, + "loss": 0.1331, + "step": 10582 + }, + { + "epoch": 4.43, + "learning_rate": 8.741958093548208e-06, + "loss": 0.166, + "step": 10583 + }, + { + "epoch": 4.43, + "learning_rate": 8.740276172625007e-06, + "loss": 0.1588, + "step": 10584 + }, + { + "epoch": 4.43, + "learning_rate": 8.738594287911447e-06, + "loss": 0.1624, + "step": 10585 + }, + { + "epoch": 4.43, + "learning_rate": 8.736912439455876e-06, + "loss": 0.1911, + "step": 10586 + }, + { + "epoch": 4.43, + "learning_rate": 8.735230627306637e-06, + "loss": 0.1964, + "step": 10587 + }, + { + "epoch": 4.43, + "learning_rate": 8.733548851512071e-06, + "loss": 0.1723, + "step": 10588 + }, + { + "epoch": 4.43, + "learning_rate": 8.731867112120516e-06, + "loss": 0.1644, + "step": 10589 + }, + { + "epoch": 4.43, + "learning_rate": 8.730185409180321e-06, + "loss": 0.1857, + "step": 10590 + }, + { + "epoch": 4.43, + "learning_rate": 8.728503742739816e-06, + "loss": 0.1473, + "step": 10591 + }, + { + "epoch": 4.43, + "learning_rate": 8.726822112847343e-06, + "loss": 0.1511, + "step": 10592 + }, + { + "epoch": 4.44, + "learning_rate": 8.725140519551238e-06, + "loss": 0.1551, + "step": 10593 + }, + { + "epoch": 4.44, + "learning_rate": 8.72345896289984e-06, + "loss": 0.1422, + "step": 10594 + }, + { + "epoch": 4.44, + "learning_rate": 8.721777442941479e-06, + "loss": 0.145, + "step": 10595 + }, + { + "epoch": 4.44, + "learning_rate": 8.720095959724492e-06, + "loss": 0.1595, + "step": 10596 + }, + { + "epoch": 4.44, + "learning_rate": 8.718414513297214e-06, + "loss": 0.1792, + "step": 10597 + }, + { + "epoch": 4.44, + "learning_rate": 8.716733103707972e-06, + "loss": 0.1171, + "step": 10598 + }, + { + "epoch": 4.44, + "learning_rate": 8.715051731005097e-06, + "loss": 0.1347, + "step": 10599 + }, + { + "epoch": 4.44, + "learning_rate": 8.713370395236918e-06, + "loss": 0.1383, + "step": 10600 + }, + { + "epoch": 4.44, + "learning_rate": 8.711689096451769e-06, + "loss": 0.1394, + "step": 10601 + }, + { + "epoch": 4.44, + "learning_rate": 8.71000783469797e-06, + "loss": 0.1451, + "step": 10602 + }, + { + "epoch": 4.44, + "learning_rate": 8.708326610023852e-06, + "loss": 0.1374, + "step": 10603 + }, + { + "epoch": 4.44, + "learning_rate": 8.706645422477739e-06, + "loss": 0.1479, + "step": 10604 + }, + { + "epoch": 4.44, + "learning_rate": 8.704964272107957e-06, + "loss": 0.1753, + "step": 10605 + }, + { + "epoch": 4.44, + "learning_rate": 8.703283158962828e-06, + "loss": 0.1383, + "step": 10606 + }, + { + "epoch": 4.44, + "learning_rate": 8.701602083090673e-06, + "loss": 0.1204, + "step": 10607 + }, + { + "epoch": 4.44, + "learning_rate": 8.699921044539815e-06, + "loss": 0.1408, + "step": 10608 + }, + { + "epoch": 4.44, + "learning_rate": 8.69824004335857e-06, + "loss": 0.1552, + "step": 10609 + }, + { + "epoch": 4.44, + "learning_rate": 8.696559079595261e-06, + "loss": 0.1682, + "step": 10610 + }, + { + "epoch": 4.44, + "learning_rate": 8.694878153298207e-06, + "loss": 0.1135, + "step": 10611 + }, + { + "epoch": 4.44, + "learning_rate": 8.69319726451572e-06, + "loss": 0.1273, + "step": 10612 + }, + { + "epoch": 4.44, + "learning_rate": 8.691516413296118e-06, + "loss": 0.0887, + "step": 10613 + }, + { + "epoch": 4.44, + "learning_rate": 8.689835599687717e-06, + "loss": 0.136, + "step": 10614 + }, + { + "epoch": 4.44, + "learning_rate": 8.68815482373883e-06, + "loss": 0.1311, + "step": 10615 + }, + { + "epoch": 4.44, + "learning_rate": 8.686474085497765e-06, + "loss": 0.1339, + "step": 10616 + }, + { + "epoch": 4.45, + "learning_rate": 8.68479338501284e-06, + "loss": 0.1429, + "step": 10617 + }, + { + "epoch": 4.45, + "learning_rate": 8.683112722332361e-06, + "loss": 0.1564, + "step": 10618 + }, + { + "epoch": 4.45, + "learning_rate": 8.681432097504636e-06, + "loss": 0.1621, + "step": 10619 + }, + { + "epoch": 4.45, + "learning_rate": 8.67975151057798e-06, + "loss": 0.1507, + "step": 10620 + }, + { + "epoch": 4.45, + "learning_rate": 8.678070961600693e-06, + "loss": 0.1407, + "step": 10621 + }, + { + "epoch": 4.45, + "learning_rate": 8.676390450621084e-06, + "loss": 0.1741, + "step": 10622 + }, + { + "epoch": 4.45, + "learning_rate": 8.674709977687456e-06, + "loss": 0.1286, + "step": 10623 + }, + { + "epoch": 4.45, + "learning_rate": 8.67302954284811e-06, + "loss": 0.1865, + "step": 10624 + }, + { + "epoch": 4.45, + "learning_rate": 8.67134914615136e-06, + "loss": 0.1679, + "step": 10625 + }, + { + "epoch": 4.45, + "learning_rate": 8.669668787645494e-06, + "loss": 0.1714, + "step": 10626 + }, + { + "epoch": 4.45, + "learning_rate": 8.667988467378819e-06, + "loss": 0.1482, + "step": 10627 + }, + { + "epoch": 4.45, + "learning_rate": 8.666308185399634e-06, + "loss": 0.2011, + "step": 10628 + }, + { + "epoch": 4.45, + "learning_rate": 8.664627941756236e-06, + "loss": 0.1634, + "step": 10629 + }, + { + "epoch": 4.45, + "learning_rate": 8.662947736496923e-06, + "loss": 0.1275, + "step": 10630 + }, + { + "epoch": 4.45, + "learning_rate": 8.661267569669994e-06, + "loss": 0.1586, + "step": 10631 + }, + { + "epoch": 4.45, + "learning_rate": 8.659587441323736e-06, + "loss": 0.16, + "step": 10632 + }, + { + "epoch": 4.45, + "learning_rate": 8.65790735150645e-06, + "loss": 0.1889, + "step": 10633 + }, + { + "epoch": 4.45, + "learning_rate": 8.656227300266425e-06, + "loss": 0.2088, + "step": 10634 + }, + { + "epoch": 4.45, + "learning_rate": 8.654547287651955e-06, + "loss": 0.1768, + "step": 10635 + }, + { + "epoch": 4.45, + "learning_rate": 8.652867313711328e-06, + "loss": 0.141, + "step": 10636 + }, + { + "epoch": 4.45, + "learning_rate": 8.651187378492836e-06, + "loss": 0.1424, + "step": 10637 + }, + { + "epoch": 4.45, + "learning_rate": 8.649507482044766e-06, + "loss": 0.1509, + "step": 10638 + }, + { + "epoch": 4.45, + "learning_rate": 8.647827624415403e-06, + "loss": 0.176, + "step": 10639 + }, + { + "epoch": 4.45, + "learning_rate": 8.646147805653037e-06, + "loss": 0.1154, + "step": 10640 + }, + { + "epoch": 4.46, + "learning_rate": 8.644468025805952e-06, + "loss": 0.1167, + "step": 10641 + }, + { + "epoch": 4.46, + "learning_rate": 8.642788284922427e-06, + "loss": 0.149, + "step": 10642 + }, + { + "epoch": 4.46, + "learning_rate": 8.641108583050751e-06, + "loss": 0.19, + "step": 10643 + }, + { + "epoch": 4.46, + "learning_rate": 8.639428920239205e-06, + "loss": 0.1419, + "step": 10644 + }, + { + "epoch": 4.46, + "learning_rate": 8.637749296536065e-06, + "loss": 0.2231, + "step": 10645 + }, + { + "epoch": 4.46, + "learning_rate": 8.636069711989614e-06, + "loss": 0.1218, + "step": 10646 + }, + { + "epoch": 4.46, + "learning_rate": 8.634390166648126e-06, + "loss": 0.1739, + "step": 10647 + }, + { + "epoch": 4.46, + "learning_rate": 8.632710660559883e-06, + "loss": 0.1368, + "step": 10648 + }, + { + "epoch": 4.46, + "learning_rate": 8.631031193773158e-06, + "loss": 0.1488, + "step": 10649 + }, + { + "epoch": 4.46, + "learning_rate": 8.629351766336225e-06, + "loss": 0.1991, + "step": 10650 + }, + { + "epoch": 4.46, + "learning_rate": 8.627672378297361e-06, + "loss": 0.154, + "step": 10651 + }, + { + "epoch": 4.46, + "learning_rate": 8.625993029704838e-06, + "loss": 0.1503, + "step": 10652 + }, + { + "epoch": 4.46, + "learning_rate": 8.624313720606922e-06, + "loss": 0.1851, + "step": 10653 + }, + { + "epoch": 4.46, + "learning_rate": 8.62263445105189e-06, + "loss": 0.1845, + "step": 10654 + }, + { + "epoch": 4.46, + "learning_rate": 8.62095522108801e-06, + "loss": 0.1492, + "step": 10655 + }, + { + "epoch": 4.46, + "learning_rate": 8.619276030763545e-06, + "loss": 0.218, + "step": 10656 + }, + { + "epoch": 4.46, + "learning_rate": 8.617596880126766e-06, + "loss": 0.1528, + "step": 10657 + }, + { + "epoch": 4.46, + "learning_rate": 8.61591776922594e-06, + "loss": 0.1055, + "step": 10658 + }, + { + "epoch": 4.46, + "learning_rate": 8.614238698109329e-06, + "loss": 0.1446, + "step": 10659 + }, + { + "epoch": 4.46, + "learning_rate": 8.612559666825196e-06, + "loss": 0.1775, + "step": 10660 + }, + { + "epoch": 4.46, + "learning_rate": 8.610880675421805e-06, + "loss": 0.1524, + "step": 10661 + }, + { + "epoch": 4.46, + "learning_rate": 8.609201723947415e-06, + "loss": 0.1336, + "step": 10662 + }, + { + "epoch": 4.46, + "learning_rate": 8.607522812450288e-06, + "loss": 0.1588, + "step": 10663 + }, + { + "epoch": 4.46, + "learning_rate": 8.60584394097868e-06, + "loss": 0.1851, + "step": 10664 + }, + { + "epoch": 4.47, + "learning_rate": 8.604165109580855e-06, + "loss": 0.1002, + "step": 10665 + }, + { + "epoch": 4.47, + "learning_rate": 8.602486318305067e-06, + "loss": 0.172, + "step": 10666 + }, + { + "epoch": 4.47, + "learning_rate": 8.600807567199565e-06, + "loss": 0.1638, + "step": 10667 + }, + { + "epoch": 4.47, + "learning_rate": 8.599128856312607e-06, + "loss": 0.1558, + "step": 10668 + }, + { + "epoch": 4.47, + "learning_rate": 8.597450185692451e-06, + "loss": 0.1578, + "step": 10669 + }, + { + "epoch": 4.47, + "learning_rate": 8.59577155538734e-06, + "loss": 0.1578, + "step": 10670 + }, + { + "epoch": 4.47, + "learning_rate": 8.594092965445534e-06, + "loss": 0.1351, + "step": 10671 + }, + { + "epoch": 4.47, + "learning_rate": 8.592414415915275e-06, + "loss": 0.1154, + "step": 10672 + }, + { + "epoch": 4.47, + "learning_rate": 8.590735906844816e-06, + "loss": 0.1396, + "step": 10673 + }, + { + "epoch": 4.47, + "learning_rate": 8.5890574382824e-06, + "loss": 0.1288, + "step": 10674 + }, + { + "epoch": 4.47, + "learning_rate": 8.58737901027628e-06, + "loss": 0.154, + "step": 10675 + }, + { + "epoch": 4.47, + "learning_rate": 8.585700622874695e-06, + "loss": 0.1194, + "step": 10676 + }, + { + "epoch": 4.47, + "learning_rate": 8.584022276125891e-06, + "loss": 0.1551, + "step": 10677 + }, + { + "epoch": 4.47, + "learning_rate": 8.582343970078108e-06, + "loss": 0.1489, + "step": 10678 + }, + { + "epoch": 4.47, + "learning_rate": 8.580665704779593e-06, + "loss": 0.1997, + "step": 10679 + }, + { + "epoch": 4.47, + "learning_rate": 8.57898748027858e-06, + "loss": 0.1953, + "step": 10680 + }, + { + "epoch": 4.47, + "learning_rate": 8.577309296623311e-06, + "loss": 0.1389, + "step": 10681 + }, + { + "epoch": 4.47, + "learning_rate": 8.575631153862027e-06, + "loss": 0.1535, + "step": 10682 + }, + { + "epoch": 4.47, + "learning_rate": 8.573953052042959e-06, + "loss": 0.147, + "step": 10683 + }, + { + "epoch": 4.47, + "learning_rate": 8.572274991214345e-06, + "loss": 0.1321, + "step": 10684 + }, + { + "epoch": 4.47, + "learning_rate": 8.570596971424421e-06, + "loss": 0.1436, + "step": 10685 + }, + { + "epoch": 4.47, + "learning_rate": 8.568918992721418e-06, + "loss": 0.121, + "step": 10686 + }, + { + "epoch": 4.47, + "learning_rate": 8.567241055153567e-06, + "loss": 0.137, + "step": 10687 + }, + { + "epoch": 4.47, + "learning_rate": 8.565563158769103e-06, + "loss": 0.1576, + "step": 10688 + }, + { + "epoch": 4.48, + "learning_rate": 8.563885303616253e-06, + "loss": 0.1183, + "step": 10689 + }, + { + "epoch": 4.48, + "learning_rate": 8.562207489743245e-06, + "loss": 0.1324, + "step": 10690 + }, + { + "epoch": 4.48, + "learning_rate": 8.560529717198305e-06, + "loss": 0.1612, + "step": 10691 + }, + { + "epoch": 4.48, + "learning_rate": 8.558851986029663e-06, + "loss": 0.1455, + "step": 10692 + }, + { + "epoch": 4.48, + "learning_rate": 8.557174296285542e-06, + "loss": 0.1424, + "step": 10693 + }, + { + "epoch": 4.48, + "learning_rate": 8.555496648014163e-06, + "loss": 0.1679, + "step": 10694 + }, + { + "epoch": 4.48, + "learning_rate": 8.553819041263752e-06, + "loss": 0.15, + "step": 10695 + }, + { + "epoch": 4.48, + "learning_rate": 8.552141476082531e-06, + "loss": 0.1507, + "step": 10696 + }, + { + "epoch": 4.48, + "learning_rate": 8.550463952518716e-06, + "loss": 0.1299, + "step": 10697 + }, + { + "epoch": 4.48, + "learning_rate": 8.54878647062053e-06, + "loss": 0.1896, + "step": 10698 + }, + { + "epoch": 4.48, + "learning_rate": 8.547109030436189e-06, + "loss": 0.1801, + "step": 10699 + }, + { + "epoch": 4.48, + "learning_rate": 8.545431632013909e-06, + "loss": 0.1252, + "step": 10700 + }, + { + "epoch": 4.48, + "learning_rate": 8.543754275401906e-06, + "loss": 0.1409, + "step": 10701 + }, + { + "epoch": 4.48, + "learning_rate": 8.542076960648395e-06, + "loss": 0.1873, + "step": 10702 + }, + { + "epoch": 4.48, + "learning_rate": 8.540399687801588e-06, + "loss": 0.1732, + "step": 10703 + }, + { + "epoch": 4.48, + "learning_rate": 8.538722456909694e-06, + "loss": 0.1139, + "step": 10704 + }, + { + "epoch": 4.48, + "learning_rate": 8.53704526802093e-06, + "loss": 0.157, + "step": 10705 + }, + { + "epoch": 4.48, + "learning_rate": 8.5353681211835e-06, + "loss": 0.1486, + "step": 10706 + }, + { + "epoch": 4.48, + "learning_rate": 8.533691016445614e-06, + "loss": 0.1798, + "step": 10707 + }, + { + "epoch": 4.48, + "learning_rate": 8.532013953855477e-06, + "loss": 0.1528, + "step": 10708 + }, + { + "epoch": 4.48, + "learning_rate": 8.530336933461298e-06, + "loss": 0.1536, + "step": 10709 + }, + { + "epoch": 4.48, + "learning_rate": 8.52865995531128e-06, + "loss": 0.1856, + "step": 10710 + }, + { + "epoch": 4.48, + "learning_rate": 8.526983019453624e-06, + "loss": 0.1538, + "step": 10711 + }, + { + "epoch": 4.48, + "learning_rate": 8.52530612593654e-06, + "loss": 0.1494, + "step": 10712 + }, + { + "epoch": 4.49, + "learning_rate": 8.523629274808219e-06, + "loss": 0.1811, + "step": 10713 + }, + { + "epoch": 4.49, + "learning_rate": 8.521952466116862e-06, + "loss": 0.2089, + "step": 10714 + }, + { + "epoch": 4.49, + "learning_rate": 8.52027569991067e-06, + "loss": 0.1166, + "step": 10715 + }, + { + "epoch": 4.49, + "learning_rate": 8.51859897623784e-06, + "loss": 0.1852, + "step": 10716 + }, + { + "epoch": 4.49, + "learning_rate": 8.51692229514657e-06, + "loss": 0.1098, + "step": 10717 + }, + { + "epoch": 4.49, + "learning_rate": 8.515245656685051e-06, + "loss": 0.1062, + "step": 10718 + }, + { + "epoch": 4.49, + "learning_rate": 8.513569060901477e-06, + "loss": 0.1651, + "step": 10719 + }, + { + "epoch": 4.49, + "learning_rate": 8.511892507844045e-06, + "loss": 0.1643, + "step": 10720 + }, + { + "epoch": 4.49, + "learning_rate": 8.510215997560939e-06, + "loss": 0.1554, + "step": 10721 + }, + { + "epoch": 4.49, + "learning_rate": 8.508539530100352e-06, + "loss": 0.1887, + "step": 10722 + }, + { + "epoch": 4.49, + "learning_rate": 8.506863105510476e-06, + "loss": 0.1091, + "step": 10723 + }, + { + "epoch": 4.49, + "learning_rate": 8.505186723839492e-06, + "loss": 0.1308, + "step": 10724 + }, + { + "epoch": 4.49, + "learning_rate": 8.50351038513559e-06, + "loss": 0.1375, + "step": 10725 + }, + { + "epoch": 4.49, + "learning_rate": 8.501834089446956e-06, + "loss": 0.1762, + "step": 10726 + }, + { + "epoch": 4.49, + "learning_rate": 8.500157836821768e-06, + "loss": 0.1493, + "step": 10727 + }, + { + "epoch": 4.49, + "learning_rate": 8.498481627308215e-06, + "loss": 0.1416, + "step": 10728 + }, + { + "epoch": 4.49, + "learning_rate": 8.496805460954474e-06, + "loss": 0.1904, + "step": 10729 + }, + { + "epoch": 4.49, + "learning_rate": 8.495129337808725e-06, + "loss": 0.1275, + "step": 10730 + }, + { + "epoch": 4.49, + "learning_rate": 8.493453257919148e-06, + "loss": 0.1303, + "step": 10731 + }, + { + "epoch": 4.49, + "learning_rate": 8.491777221333923e-06, + "loss": 0.1428, + "step": 10732 + }, + { + "epoch": 4.49, + "learning_rate": 8.490101228101222e-06, + "loss": 0.1747, + "step": 10733 + }, + { + "epoch": 4.49, + "learning_rate": 8.488425278269224e-06, + "loss": 0.1431, + "step": 10734 + }, + { + "epoch": 4.49, + "learning_rate": 8.486749371886096e-06, + "loss": 0.1246, + "step": 10735 + }, + { + "epoch": 4.49, + "learning_rate": 8.485073509000015e-06, + "loss": 0.136, + "step": 10736 + }, + { + "epoch": 4.5, + "learning_rate": 8.483397689659153e-06, + "loss": 0.1523, + "step": 10737 + }, + { + "epoch": 4.5, + "learning_rate": 8.481721913911677e-06, + "loss": 0.1385, + "step": 10738 + }, + { + "epoch": 4.5, + "learning_rate": 8.480046181805757e-06, + "loss": 0.1603, + "step": 10739 + }, + { + "epoch": 4.5, + "learning_rate": 8.478370493389563e-06, + "loss": 0.1634, + "step": 10740 + }, + { + "epoch": 4.5, + "learning_rate": 8.476694848711258e-06, + "loss": 0.1531, + "step": 10741 + }, + { + "epoch": 4.5, + "learning_rate": 8.475019247819006e-06, + "loss": 0.1607, + "step": 10742 + }, + { + "epoch": 4.5, + "learning_rate": 8.473343690760976e-06, + "loss": 0.1643, + "step": 10743 + }, + { + "epoch": 4.5, + "learning_rate": 8.471668177585324e-06, + "loss": 0.1783, + "step": 10744 + }, + { + "epoch": 4.5, + "learning_rate": 8.469992708340213e-06, + "loss": 0.1064, + "step": 10745 + }, + { + "epoch": 4.5, + "learning_rate": 8.468317283073805e-06, + "loss": 0.1407, + "step": 10746 + }, + { + "epoch": 4.5, + "learning_rate": 8.46664190183426e-06, + "loss": 0.1761, + "step": 10747 + }, + { + "epoch": 4.5, + "learning_rate": 8.46496656466973e-06, + "loss": 0.143, + "step": 10748 + }, + { + "epoch": 4.5, + "learning_rate": 8.463291271628375e-06, + "loss": 0.1405, + "step": 10749 + }, + { + "epoch": 4.5, + "learning_rate": 8.46161602275835e-06, + "loss": 0.1192, + "step": 10750 + }, + { + "epoch": 4.5, + "learning_rate": 8.459940818107805e-06, + "loss": 0.1546, + "step": 10751 + }, + { + "epoch": 4.5, + "learning_rate": 8.458265657724894e-06, + "loss": 0.1785, + "step": 10752 + }, + { + "epoch": 4.5, + "learning_rate": 8.456590541657772e-06, + "loss": 0.1608, + "step": 10753 + }, + { + "epoch": 4.5, + "learning_rate": 8.454915469954583e-06, + "loss": 0.1609, + "step": 10754 + }, + { + "epoch": 4.5, + "learning_rate": 8.453240442663477e-06, + "loss": 0.1414, + "step": 10755 + }, + { + "epoch": 4.5, + "learning_rate": 8.451565459832604e-06, + "loss": 0.1398, + "step": 10756 + }, + { + "epoch": 4.5, + "learning_rate": 8.44989052151011e-06, + "loss": 0.1418, + "step": 10757 + }, + { + "epoch": 4.5, + "learning_rate": 8.448215627744134e-06, + "loss": 0.1435, + "step": 10758 + }, + { + "epoch": 4.5, + "learning_rate": 8.446540778582823e-06, + "loss": 0.1296, + "step": 10759 + }, + { + "epoch": 4.5, + "learning_rate": 8.444865974074319e-06, + "loss": 0.2264, + "step": 10760 + }, + { + "epoch": 4.51, + "learning_rate": 8.443191214266763e-06, + "loss": 0.1684, + "step": 10761 + }, + { + "epoch": 4.51, + "learning_rate": 8.441516499208292e-06, + "loss": 0.1324, + "step": 10762 + }, + { + "epoch": 4.51, + "learning_rate": 8.439841828947048e-06, + "loss": 0.1965, + "step": 10763 + }, + { + "epoch": 4.51, + "learning_rate": 8.438167203531166e-06, + "loss": 0.1786, + "step": 10764 + }, + { + "epoch": 4.51, + "learning_rate": 8.436492623008782e-06, + "loss": 0.1435, + "step": 10765 + }, + { + "epoch": 4.51, + "learning_rate": 8.434818087428029e-06, + "loss": 0.1287, + "step": 10766 + }, + { + "epoch": 4.51, + "learning_rate": 8.433143596837043e-06, + "loss": 0.1772, + "step": 10767 + }, + { + "epoch": 4.51, + "learning_rate": 8.431469151283953e-06, + "loss": 0.1788, + "step": 10768 + }, + { + "epoch": 4.51, + "learning_rate": 8.429794750816888e-06, + "loss": 0.1539, + "step": 10769 + }, + { + "epoch": 4.51, + "learning_rate": 8.428120395483982e-06, + "loss": 0.14, + "step": 10770 + }, + { + "epoch": 4.51, + "learning_rate": 8.42644608533336e-06, + "loss": 0.1312, + "step": 10771 + }, + { + "epoch": 4.51, + "learning_rate": 8.42477182041315e-06, + "loss": 0.1656, + "step": 10772 + }, + { + "epoch": 4.51, + "learning_rate": 8.423097600771476e-06, + "loss": 0.1824, + "step": 10773 + }, + { + "epoch": 4.51, + "learning_rate": 8.42142342645646e-06, + "loss": 0.1685, + "step": 10774 + }, + { + "epoch": 4.51, + "learning_rate": 8.419749297516227e-06, + "loss": 0.1578, + "step": 10775 + }, + { + "epoch": 4.51, + "learning_rate": 8.4180752139989e-06, + "loss": 0.1387, + "step": 10776 + }, + { + "epoch": 4.51, + "learning_rate": 8.416401175952598e-06, + "loss": 0.1427, + "step": 10777 + }, + { + "epoch": 4.51, + "learning_rate": 8.414727183425437e-06, + "loss": 0.1592, + "step": 10778 + }, + { + "epoch": 4.51, + "learning_rate": 8.413053236465535e-06, + "loss": 0.1332, + "step": 10779 + }, + { + "epoch": 4.51, + "learning_rate": 8.411379335121014e-06, + "loss": 0.1034, + "step": 10780 + }, + { + "epoch": 4.51, + "learning_rate": 8.409705479439985e-06, + "loss": 0.185, + "step": 10781 + }, + { + "epoch": 4.51, + "learning_rate": 8.408031669470555e-06, + "loss": 0.1347, + "step": 10782 + }, + { + "epoch": 4.51, + "learning_rate": 8.406357905260845e-06, + "loss": 0.1821, + "step": 10783 + }, + { + "epoch": 4.51, + "learning_rate": 8.404684186858962e-06, + "loss": 0.1364, + "step": 10784 + }, + { + "epoch": 4.52, + "learning_rate": 8.403010514313016e-06, + "loss": 0.1522, + "step": 10785 + }, + { + "epoch": 4.52, + "learning_rate": 8.401336887671117e-06, + "loss": 0.1601, + "step": 10786 + }, + { + "epoch": 4.52, + "learning_rate": 8.399663306981368e-06, + "loss": 0.1565, + "step": 10787 + }, + { + "epoch": 4.52, + "learning_rate": 8.39798977229188e-06, + "loss": 0.1307, + "step": 10788 + }, + { + "epoch": 4.52, + "learning_rate": 8.396316283650753e-06, + "loss": 0.1278, + "step": 10789 + }, + { + "epoch": 4.52, + "learning_rate": 8.39464284110609e-06, + "loss": 0.1808, + "step": 10790 + }, + { + "epoch": 4.52, + "learning_rate": 8.392969444705998e-06, + "loss": 0.1575, + "step": 10791 + }, + { + "epoch": 4.52, + "learning_rate": 8.39129609449857e-06, + "loss": 0.163, + "step": 10792 + }, + { + "epoch": 4.52, + "learning_rate": 8.389622790531908e-06, + "loss": 0.1786, + "step": 10793 + }, + { + "epoch": 4.52, + "learning_rate": 8.387949532854113e-06, + "loss": 0.1419, + "step": 10794 + }, + { + "epoch": 4.52, + "learning_rate": 8.386276321513275e-06, + "loss": 0.1721, + "step": 10795 + }, + { + "epoch": 4.52, + "learning_rate": 8.384603156557492e-06, + "loss": 0.145, + "step": 10796 + }, + { + "epoch": 4.52, + "learning_rate": 8.38293003803486e-06, + "loss": 0.1321, + "step": 10797 + }, + { + "epoch": 4.52, + "learning_rate": 8.381256965993468e-06, + "loss": 0.1727, + "step": 10798 + }, + { + "epoch": 4.52, + "learning_rate": 8.379583940481407e-06, + "loss": 0.1064, + "step": 10799 + }, + { + "epoch": 4.52, + "learning_rate": 8.377910961546771e-06, + "loss": 0.1787, + "step": 10800 + }, + { + "epoch": 4.52, + "learning_rate": 8.376238029237643e-06, + "loss": 0.1235, + "step": 10801 + }, + { + "epoch": 4.52, + "learning_rate": 8.374565143602114e-06, + "loss": 0.1486, + "step": 10802 + }, + { + "epoch": 4.52, + "learning_rate": 8.372892304688265e-06, + "loss": 0.1193, + "step": 10803 + }, + { + "epoch": 4.52, + "learning_rate": 8.371219512544182e-06, + "loss": 0.1624, + "step": 10804 + }, + { + "epoch": 4.52, + "learning_rate": 8.369546767217952e-06, + "loss": 0.1455, + "step": 10805 + }, + { + "epoch": 4.52, + "learning_rate": 8.367874068757651e-06, + "loss": 0.1264, + "step": 10806 + }, + { + "epoch": 4.52, + "learning_rate": 8.366201417211361e-06, + "loss": 0.171, + "step": 10807 + }, + { + "epoch": 4.53, + "learning_rate": 8.364528812627164e-06, + "loss": 0.1191, + "step": 10808 + }, + { + "epoch": 4.53, + "learning_rate": 8.362856255053132e-06, + "loss": 0.1946, + "step": 10809 + }, + { + "epoch": 4.53, + "learning_rate": 8.361183744537345e-06, + "loss": 0.1596, + "step": 10810 + }, + { + "epoch": 4.53, + "learning_rate": 8.359511281127879e-06, + "loss": 0.2023, + "step": 10811 + }, + { + "epoch": 4.53, + "learning_rate": 8.357838864872803e-06, + "loss": 0.1452, + "step": 10812 + }, + { + "epoch": 4.53, + "learning_rate": 8.356166495820192e-06, + "loss": 0.1487, + "step": 10813 + }, + { + "epoch": 4.53, + "learning_rate": 8.354494174018118e-06, + "loss": 0.1538, + "step": 10814 + }, + { + "epoch": 4.53, + "learning_rate": 8.352821899514648e-06, + "loss": 0.1443, + "step": 10815 + }, + { + "epoch": 4.53, + "learning_rate": 8.35114967235785e-06, + "loss": 0.1427, + "step": 10816 + }, + { + "epoch": 4.53, + "learning_rate": 8.34947749259579e-06, + "loss": 0.1366, + "step": 10817 + }, + { + "epoch": 4.53, + "learning_rate": 8.347805360276538e-06, + "loss": 0.157, + "step": 10818 + }, + { + "epoch": 4.53, + "learning_rate": 8.346133275448153e-06, + "loss": 0.1599, + "step": 10819 + }, + { + "epoch": 4.53, + "learning_rate": 8.3444612381587e-06, + "loss": 0.1406, + "step": 10820 + }, + { + "epoch": 4.53, + "learning_rate": 8.342789248456242e-06, + "loss": 0.1796, + "step": 10821 + }, + { + "epoch": 4.53, + "learning_rate": 8.341117306388833e-06, + "loss": 0.15, + "step": 10822 + }, + { + "epoch": 4.53, + "learning_rate": 8.339445412004533e-06, + "loss": 0.1105, + "step": 10823 + }, + { + "epoch": 4.53, + "learning_rate": 8.337773565351407e-06, + "loss": 0.153, + "step": 10824 + }, + { + "epoch": 4.53, + "learning_rate": 8.336101766477506e-06, + "loss": 0.1288, + "step": 10825 + }, + { + "epoch": 4.53, + "learning_rate": 8.334430015430878e-06, + "loss": 0.1934, + "step": 10826 + }, + { + "epoch": 4.53, + "learning_rate": 8.332758312259583e-06, + "loss": 0.1263, + "step": 10827 + }, + { + "epoch": 4.53, + "learning_rate": 8.331086657011669e-06, + "loss": 0.1337, + "step": 10828 + }, + { + "epoch": 4.53, + "learning_rate": 8.32941504973519e-06, + "loss": 0.1528, + "step": 10829 + }, + { + "epoch": 4.53, + "learning_rate": 8.327743490478192e-06, + "loss": 0.1801, + "step": 10830 + }, + { + "epoch": 4.53, + "learning_rate": 8.326071979288723e-06, + "loss": 0.1432, + "step": 10831 + }, + { + "epoch": 4.54, + "learning_rate": 8.32440051621483e-06, + "loss": 0.1296, + "step": 10832 + }, + { + "epoch": 4.54, + "learning_rate": 8.322729101304558e-06, + "loss": 0.1795, + "step": 10833 + }, + { + "epoch": 4.54, + "learning_rate": 8.321057734605949e-06, + "loss": 0.1599, + "step": 10834 + }, + { + "epoch": 4.54, + "learning_rate": 8.319386416167047e-06, + "loss": 0.1965, + "step": 10835 + }, + { + "epoch": 4.54, + "learning_rate": 8.31771514603589e-06, + "loss": 0.1628, + "step": 10836 + }, + { + "epoch": 4.54, + "learning_rate": 8.316043924260514e-06, + "loss": 0.1537, + "step": 10837 + }, + { + "epoch": 4.54, + "learning_rate": 8.314372750888968e-06, + "loss": 0.1814, + "step": 10838 + }, + { + "epoch": 4.54, + "learning_rate": 8.312701625969278e-06, + "loss": 0.1303, + "step": 10839 + }, + { + "epoch": 4.54, + "learning_rate": 8.311030549549483e-06, + "loss": 0.1532, + "step": 10840 + }, + { + "epoch": 4.54, + "learning_rate": 8.309359521677617e-06, + "loss": 0.1477, + "step": 10841 + }, + { + "epoch": 4.54, + "learning_rate": 8.307688542401709e-06, + "loss": 0.1307, + "step": 10842 + }, + { + "epoch": 4.54, + "learning_rate": 8.306017611769795e-06, + "loss": 0.122, + "step": 10843 + }, + { + "epoch": 4.54, + "learning_rate": 8.3043467298299e-06, + "loss": 0.1751, + "step": 10844 + }, + { + "epoch": 4.54, + "learning_rate": 8.302675896630054e-06, + "loss": 0.1655, + "step": 10845 + }, + { + "epoch": 4.54, + "learning_rate": 8.301005112218282e-06, + "loss": 0.1444, + "step": 10846 + }, + { + "epoch": 4.54, + "learning_rate": 8.299334376642609e-06, + "loss": 0.1179, + "step": 10847 + }, + { + "epoch": 4.54, + "learning_rate": 8.297663689951067e-06, + "loss": 0.1252, + "step": 10848 + }, + { + "epoch": 4.54, + "learning_rate": 8.295993052191669e-06, + "loss": 0.1143, + "step": 10849 + }, + { + "epoch": 4.54, + "learning_rate": 8.294322463412435e-06, + "loss": 0.132, + "step": 10850 + }, + { + "epoch": 4.54, + "learning_rate": 8.292651923661391e-06, + "loss": 0.1182, + "step": 10851 + }, + { + "epoch": 4.54, + "learning_rate": 8.290981432986551e-06, + "loss": 0.1625, + "step": 10852 + }, + { + "epoch": 4.54, + "learning_rate": 8.289310991435933e-06, + "loss": 0.1747, + "step": 10853 + }, + { + "epoch": 4.54, + "learning_rate": 8.287640599057551e-06, + "loss": 0.162, + "step": 10854 + }, + { + "epoch": 4.54, + "learning_rate": 8.285970255899425e-06, + "loss": 0.1723, + "step": 10855 + }, + { + "epoch": 4.55, + "learning_rate": 8.284299962009559e-06, + "loss": 0.1552, + "step": 10856 + }, + { + "epoch": 4.55, + "learning_rate": 8.28262971743597e-06, + "loss": 0.1625, + "step": 10857 + }, + { + "epoch": 4.55, + "learning_rate": 8.280959522226663e-06, + "loss": 0.1036, + "step": 10858 + }, + { + "epoch": 4.55, + "learning_rate": 8.279289376429653e-06, + "loss": 0.1547, + "step": 10859 + }, + { + "epoch": 4.55, + "learning_rate": 8.27761928009294e-06, + "loss": 0.1245, + "step": 10860 + }, + { + "epoch": 4.55, + "learning_rate": 8.275949233264534e-06, + "loss": 0.1447, + "step": 10861 + }, + { + "epoch": 4.55, + "learning_rate": 8.274279235992438e-06, + "loss": 0.1469, + "step": 10862 + }, + { + "epoch": 4.55, + "learning_rate": 8.272609288324653e-06, + "loss": 0.1636, + "step": 10863 + }, + { + "epoch": 4.55, + "learning_rate": 8.270939390309181e-06, + "loss": 0.2064, + "step": 10864 + }, + { + "epoch": 4.55, + "learning_rate": 8.269269541994022e-06, + "loss": 0.1293, + "step": 10865 + }, + { + "epoch": 4.55, + "learning_rate": 8.267599743427173e-06, + "loss": 0.126, + "step": 10866 + }, + { + "epoch": 4.55, + "learning_rate": 8.265929994656634e-06, + "loss": 0.1713, + "step": 10867 + }, + { + "epoch": 4.55, + "learning_rate": 8.2642602957304e-06, + "loss": 0.1381, + "step": 10868 + }, + { + "epoch": 4.55, + "learning_rate": 8.26259064669646e-06, + "loss": 0.0982, + "step": 10869 + }, + { + "epoch": 4.55, + "learning_rate": 8.260921047602814e-06, + "loss": 0.1539, + "step": 10870 + }, + { + "epoch": 4.55, + "learning_rate": 8.259251498497446e-06, + "loss": 0.1549, + "step": 10871 + }, + { + "epoch": 4.55, + "learning_rate": 8.25758199942835e-06, + "loss": 0.1301, + "step": 10872 + }, + { + "epoch": 4.55, + "learning_rate": 8.255912550443515e-06, + "loss": 0.133, + "step": 10873 + }, + { + "epoch": 4.55, + "learning_rate": 8.254243151590922e-06, + "loss": 0.195, + "step": 10874 + }, + { + "epoch": 4.55, + "learning_rate": 8.252573802918562e-06, + "loss": 0.1498, + "step": 10875 + }, + { + "epoch": 4.55, + "learning_rate": 8.25090450447442e-06, + "loss": 0.173, + "step": 10876 + }, + { + "epoch": 4.55, + "learning_rate": 8.249235256306472e-06, + "loss": 0.1879, + "step": 10877 + }, + { + "epoch": 4.55, + "learning_rate": 8.247566058462704e-06, + "loss": 0.1443, + "step": 10878 + }, + { + "epoch": 4.55, + "learning_rate": 8.245896910991096e-06, + "loss": 0.1435, + "step": 10879 + }, + { + "epoch": 4.56, + "learning_rate": 8.244227813939624e-06, + "loss": 0.1807, + "step": 10880 + }, + { + "epoch": 4.56, + "learning_rate": 8.242558767356266e-06, + "loss": 0.1714, + "step": 10881 + }, + { + "epoch": 4.56, + "learning_rate": 8.240889771288997e-06, + "loss": 0.1307, + "step": 10882 + }, + { + "epoch": 4.56, + "learning_rate": 8.239220825785789e-06, + "loss": 0.1448, + "step": 10883 + }, + { + "epoch": 4.56, + "learning_rate": 8.237551930894616e-06, + "loss": 0.1597, + "step": 10884 + }, + { + "epoch": 4.56, + "learning_rate": 8.23588308666345e-06, + "loss": 0.1435, + "step": 10885 + }, + { + "epoch": 4.56, + "learning_rate": 8.234214293140258e-06, + "loss": 0.1298, + "step": 10886 + }, + { + "epoch": 4.56, + "learning_rate": 8.23254555037301e-06, + "loss": 0.1396, + "step": 10887 + }, + { + "epoch": 4.56, + "learning_rate": 8.230876858409669e-06, + "loss": 0.1867, + "step": 10888 + }, + { + "epoch": 4.56, + "learning_rate": 8.229208217298208e-06, + "loss": 0.149, + "step": 10889 + }, + { + "epoch": 4.56, + "learning_rate": 8.227539627086582e-06, + "loss": 0.1266, + "step": 10890 + }, + { + "epoch": 4.56, + "learning_rate": 8.225871087822757e-06, + "loss": 0.1307, + "step": 10891 + }, + { + "epoch": 4.56, + "learning_rate": 8.224202599554695e-06, + "loss": 0.1753, + "step": 10892 + }, + { + "epoch": 4.56, + "learning_rate": 8.222534162330355e-06, + "loss": 0.1343, + "step": 10893 + }, + { + "epoch": 4.56, + "learning_rate": 8.22086577619769e-06, + "loss": 0.145, + "step": 10894 + }, + { + "epoch": 4.56, + "learning_rate": 8.219197441204659e-06, + "loss": 0.1279, + "step": 10895 + }, + { + "epoch": 4.56, + "learning_rate": 8.21752915739922e-06, + "loss": 0.1412, + "step": 10896 + }, + { + "epoch": 4.56, + "learning_rate": 8.21586092482932e-06, + "loss": 0.1737, + "step": 10897 + }, + { + "epoch": 4.56, + "learning_rate": 8.214192743542914e-06, + "loss": 0.1382, + "step": 10898 + }, + { + "epoch": 4.56, + "learning_rate": 8.212524613587955e-06, + "loss": 0.14, + "step": 10899 + }, + { + "epoch": 4.56, + "learning_rate": 8.210856535012392e-06, + "loss": 0.1652, + "step": 10900 + }, + { + "epoch": 4.56, + "learning_rate": 8.209188507864167e-06, + "loss": 0.1024, + "step": 10901 + }, + { + "epoch": 4.56, + "learning_rate": 8.20752053219123e-06, + "loss": 0.2096, + "step": 10902 + }, + { + "epoch": 4.56, + "learning_rate": 8.205852608041526e-06, + "loss": 0.1255, + "step": 10903 + }, + { + "epoch": 4.57, + "learning_rate": 8.204184735462994e-06, + "loss": 0.1135, + "step": 10904 + }, + { + "epoch": 4.57, + "learning_rate": 8.20251691450358e-06, + "loss": 0.1712, + "step": 10905 + }, + { + "epoch": 4.57, + "learning_rate": 8.200849145211224e-06, + "loss": 0.1354, + "step": 10906 + }, + { + "epoch": 4.57, + "learning_rate": 8.19918142763386e-06, + "loss": 0.14, + "step": 10907 + }, + { + "epoch": 4.57, + "learning_rate": 8.197513761819428e-06, + "loss": 0.1466, + "step": 10908 + }, + { + "epoch": 4.57, + "learning_rate": 8.195846147815865e-06, + "loss": 0.1407, + "step": 10909 + }, + { + "epoch": 4.57, + "learning_rate": 8.194178585671102e-06, + "loss": 0.1501, + "step": 10910 + }, + { + "epoch": 4.57, + "learning_rate": 8.192511075433073e-06, + "loss": 0.1264, + "step": 10911 + }, + { + "epoch": 4.57, + "learning_rate": 8.190843617149712e-06, + "loss": 0.181, + "step": 10912 + }, + { + "epoch": 4.57, + "learning_rate": 8.189176210868942e-06, + "loss": 0.1677, + "step": 10913 + }, + { + "epoch": 4.57, + "learning_rate": 8.187508856638697e-06, + "loss": 0.174, + "step": 10914 + }, + { + "epoch": 4.57, + "learning_rate": 8.185841554506901e-06, + "loss": 0.1589, + "step": 10915 + }, + { + "epoch": 4.57, + "learning_rate": 8.184174304521483e-06, + "loss": 0.1424, + "step": 10916 + }, + { + "epoch": 4.57, + "learning_rate": 8.18250710673036e-06, + "loss": 0.1473, + "step": 10917 + }, + { + "epoch": 4.57, + "learning_rate": 8.180839961181458e-06, + "loss": 0.1487, + "step": 10918 + }, + { + "epoch": 4.57, + "learning_rate": 8.179172867922695e-06, + "loss": 0.1064, + "step": 10919 + }, + { + "epoch": 4.57, + "learning_rate": 8.177505827001997e-06, + "loss": 0.1629, + "step": 10920 + }, + { + "epoch": 4.57, + "learning_rate": 8.175838838467272e-06, + "loss": 0.168, + "step": 10921 + }, + { + "epoch": 4.57, + "learning_rate": 8.174171902366441e-06, + "loss": 0.1765, + "step": 10922 + }, + { + "epoch": 4.57, + "learning_rate": 8.172505018747421e-06, + "loss": 0.1488, + "step": 10923 + }, + { + "epoch": 4.57, + "learning_rate": 8.17083818765812e-06, + "loss": 0.1875, + "step": 10924 + }, + { + "epoch": 4.57, + "learning_rate": 8.169171409146453e-06, + "loss": 0.1606, + "step": 10925 + }, + { + "epoch": 4.57, + "learning_rate": 8.167504683260332e-06, + "loss": 0.1725, + "step": 10926 + }, + { + "epoch": 4.57, + "learning_rate": 8.16583801004766e-06, + "loss": 0.129, + "step": 10927 + }, + { + "epoch": 4.58, + "learning_rate": 8.164171389556346e-06, + "loss": 0.1683, + "step": 10928 + }, + { + "epoch": 4.58, + "learning_rate": 8.162504821834296e-06, + "loss": 0.1483, + "step": 10929 + }, + { + "epoch": 4.58, + "learning_rate": 8.160838306929418e-06, + "loss": 0.1661, + "step": 10930 + }, + { + "epoch": 4.58, + "learning_rate": 8.159171844889609e-06, + "loss": 0.149, + "step": 10931 + }, + { + "epoch": 4.58, + "learning_rate": 8.157505435762769e-06, + "loss": 0.161, + "step": 10932 + }, + { + "epoch": 4.58, + "learning_rate": 8.155839079596805e-06, + "loss": 0.1306, + "step": 10933 + }, + { + "epoch": 4.58, + "learning_rate": 8.154172776439608e-06, + "loss": 0.1119, + "step": 10934 + }, + { + "epoch": 4.58, + "learning_rate": 8.152506526339075e-06, + "loss": 0.1548, + "step": 10935 + }, + { + "epoch": 4.58, + "learning_rate": 8.150840329343104e-06, + "loss": 0.1768, + "step": 10936 + }, + { + "epoch": 4.58, + "learning_rate": 8.149174185499587e-06, + "loss": 0.1677, + "step": 10937 + }, + { + "epoch": 4.58, + "learning_rate": 8.14750809485642e-06, + "loss": 0.127, + "step": 10938 + }, + { + "epoch": 4.58, + "learning_rate": 8.145842057461483e-06, + "loss": 0.1572, + "step": 10939 + }, + { + "epoch": 4.58, + "learning_rate": 8.14417607336267e-06, + "loss": 0.1449, + "step": 10940 + }, + { + "epoch": 4.58, + "learning_rate": 8.142510142607874e-06, + "loss": 0.1592, + "step": 10941 + }, + { + "epoch": 4.58, + "learning_rate": 8.140844265244971e-06, + "loss": 0.1251, + "step": 10942 + }, + { + "epoch": 4.58, + "learning_rate": 8.139178441321851e-06, + "loss": 0.1163, + "step": 10943 + }, + { + "epoch": 4.58, + "learning_rate": 8.137512670886397e-06, + "loss": 0.1183, + "step": 10944 + }, + { + "epoch": 4.58, + "learning_rate": 8.135846953986486e-06, + "loss": 0.1977, + "step": 10945 + }, + { + "epoch": 4.58, + "learning_rate": 8.134181290670002e-06, + "loss": 0.1227, + "step": 10946 + }, + { + "epoch": 4.58, + "learning_rate": 8.132515680984821e-06, + "loss": 0.1084, + "step": 10947 + }, + { + "epoch": 4.58, + "learning_rate": 8.130850124978819e-06, + "loss": 0.163, + "step": 10948 + }, + { + "epoch": 4.58, + "learning_rate": 8.12918462269987e-06, + "loss": 0.1608, + "step": 10949 + }, + { + "epoch": 4.58, + "learning_rate": 8.127519174195853e-06, + "loss": 0.1399, + "step": 10950 + }, + { + "epoch": 4.58, + "learning_rate": 8.125853779514632e-06, + "loss": 0.1631, + "step": 10951 + }, + { + "epoch": 4.59, + "learning_rate": 8.124188438704082e-06, + "loss": 0.1099, + "step": 10952 + }, + { + "epoch": 4.59, + "learning_rate": 8.122523151812072e-06, + "loss": 0.1351, + "step": 10953 + }, + { + "epoch": 4.59, + "learning_rate": 8.120857918886467e-06, + "loss": 0.1401, + "step": 10954 + }, + { + "epoch": 4.59, + "learning_rate": 8.119192739975133e-06, + "loss": 0.1397, + "step": 10955 + }, + { + "epoch": 4.59, + "learning_rate": 8.117527615125938e-06, + "loss": 0.1302, + "step": 10956 + }, + { + "epoch": 4.59, + "learning_rate": 8.115862544386736e-06, + "loss": 0.1344, + "step": 10957 + }, + { + "epoch": 4.59, + "learning_rate": 8.114197527805396e-06, + "loss": 0.1636, + "step": 10958 + }, + { + "epoch": 4.59, + "learning_rate": 8.112532565429774e-06, + "loss": 0.1649, + "step": 10959 + }, + { + "epoch": 4.59, + "learning_rate": 8.11086765730773e-06, + "loss": 0.1666, + "step": 10960 + }, + { + "epoch": 4.59, + "learning_rate": 8.109202803487121e-06, + "loss": 0.1521, + "step": 10961 + }, + { + "epoch": 4.59, + "learning_rate": 8.107538004015796e-06, + "loss": 0.1471, + "step": 10962 + }, + { + "epoch": 4.59, + "learning_rate": 8.105873258941612e-06, + "loss": 0.1451, + "step": 10963 + }, + { + "epoch": 4.59, + "learning_rate": 8.10420856831242e-06, + "loss": 0.1674, + "step": 10964 + }, + { + "epoch": 4.59, + "learning_rate": 8.10254393217607e-06, + "loss": 0.1801, + "step": 10965 + }, + { + "epoch": 4.59, + "learning_rate": 8.100879350580411e-06, + "loss": 0.1484, + "step": 10966 + }, + { + "epoch": 4.59, + "learning_rate": 8.099214823573293e-06, + "loss": 0.1657, + "step": 10967 + }, + { + "epoch": 4.59, + "learning_rate": 8.097550351202554e-06, + "loss": 0.179, + "step": 10968 + }, + { + "epoch": 4.59, + "learning_rate": 8.095885933516043e-06, + "loss": 0.1169, + "step": 10969 + }, + { + "epoch": 4.59, + "learning_rate": 8.094221570561602e-06, + "loss": 0.1453, + "step": 10970 + }, + { + "epoch": 4.59, + "learning_rate": 8.09255726238707e-06, + "loss": 0.1403, + "step": 10971 + }, + { + "epoch": 4.59, + "learning_rate": 8.090893009040288e-06, + "loss": 0.1554, + "step": 10972 + }, + { + "epoch": 4.59, + "learning_rate": 8.08922881056909e-06, + "loss": 0.1699, + "step": 10973 + }, + { + "epoch": 4.59, + "learning_rate": 8.087564667021317e-06, + "loss": 0.1783, + "step": 10974 + }, + { + "epoch": 4.59, + "learning_rate": 8.0859005784448e-06, + "loss": 0.1717, + "step": 10975 + }, + { + "epoch": 4.6, + "learning_rate": 8.084236544887372e-06, + "loss": 0.1571, + "step": 10976 + }, + { + "epoch": 4.6, + "learning_rate": 8.082572566396864e-06, + "loss": 0.1587, + "step": 10977 + }, + { + "epoch": 4.6, + "learning_rate": 8.080908643021107e-06, + "loss": 0.1206, + "step": 10978 + }, + { + "epoch": 4.6, + "learning_rate": 8.079244774807926e-06, + "loss": 0.1667, + "step": 10979 + }, + { + "epoch": 4.6, + "learning_rate": 8.077580961805153e-06, + "loss": 0.1552, + "step": 10980 + }, + { + "epoch": 4.6, + "learning_rate": 8.075917204060606e-06, + "loss": 0.1245, + "step": 10981 + }, + { + "epoch": 4.6, + "learning_rate": 8.074253501622113e-06, + "loss": 0.1779, + "step": 10982 + }, + { + "epoch": 4.6, + "learning_rate": 8.072589854537498e-06, + "loss": 0.1806, + "step": 10983 + }, + { + "epoch": 4.6, + "learning_rate": 8.070926262854573e-06, + "loss": 0.1765, + "step": 10984 + }, + { + "epoch": 4.6, + "learning_rate": 8.069262726621163e-06, + "loss": 0.1469, + "step": 10985 + }, + { + "epoch": 4.6, + "learning_rate": 8.06759924588508e-06, + "loss": 0.1461, + "step": 10986 + }, + { + "epoch": 4.6, + "learning_rate": 8.065935820694142e-06, + "loss": 0.1394, + "step": 10987 + }, + { + "epoch": 4.6, + "learning_rate": 8.064272451096165e-06, + "loss": 0.1073, + "step": 10988 + }, + { + "epoch": 4.6, + "learning_rate": 8.062609137138955e-06, + "loss": 0.126, + "step": 10989 + }, + { + "epoch": 4.6, + "learning_rate": 8.060945878870328e-06, + "loss": 0.1782, + "step": 10990 + }, + { + "epoch": 4.6, + "learning_rate": 8.059282676338092e-06, + "loss": 0.1596, + "step": 10991 + }, + { + "epoch": 4.6, + "learning_rate": 8.057619529590051e-06, + "loss": 0.138, + "step": 10992 + }, + { + "epoch": 4.6, + "learning_rate": 8.055956438674015e-06, + "loss": 0.1796, + "step": 10993 + }, + { + "epoch": 4.6, + "learning_rate": 8.054293403637787e-06, + "loss": 0.153, + "step": 10994 + }, + { + "epoch": 4.6, + "learning_rate": 8.052630424529166e-06, + "loss": 0.1554, + "step": 10995 + }, + { + "epoch": 4.6, + "learning_rate": 8.050967501395955e-06, + "loss": 0.1505, + "step": 10996 + }, + { + "epoch": 4.6, + "learning_rate": 8.049304634285957e-06, + "loss": 0.1279, + "step": 10997 + }, + { + "epoch": 4.6, + "learning_rate": 8.047641823246964e-06, + "loss": 0.1643, + "step": 10998 + }, + { + "epoch": 4.6, + "learning_rate": 8.045979068326773e-06, + "loss": 0.1534, + "step": 10999 + }, + { + "epoch": 4.61, + "learning_rate": 8.044316369573181e-06, + "loss": 0.1411, + "step": 11000 + }, + { + "epoch": 4.61, + "learning_rate": 8.042653727033981e-06, + "loss": 0.1839, + "step": 11001 + }, + { + "epoch": 4.61, + "learning_rate": 8.04099114075696e-06, + "loss": 0.158, + "step": 11002 + }, + { + "epoch": 4.61, + "learning_rate": 8.039328610789911e-06, + "loss": 0.1694, + "step": 11003 + }, + { + "epoch": 4.61, + "learning_rate": 8.037666137180624e-06, + "loss": 0.1096, + "step": 11004 + }, + { + "epoch": 4.61, + "learning_rate": 8.03600371997688e-06, + "loss": 0.1871, + "step": 11005 + }, + { + "epoch": 4.61, + "learning_rate": 8.03434135922647e-06, + "loss": 0.12, + "step": 11006 + }, + { + "epoch": 4.61, + "learning_rate": 8.032679054977171e-06, + "loss": 0.1471, + "step": 11007 + }, + { + "epoch": 4.61, + "learning_rate": 8.031016807276765e-06, + "loss": 0.1441, + "step": 11008 + }, + { + "epoch": 4.61, + "learning_rate": 8.029354616173036e-06, + "loss": 0.1719, + "step": 11009 + }, + { + "epoch": 4.61, + "learning_rate": 8.027692481713757e-06, + "loss": 0.1434, + "step": 11010 + }, + { + "epoch": 4.61, + "learning_rate": 8.026030403946707e-06, + "loss": 0.1538, + "step": 11011 + }, + { + "epoch": 4.61, + "learning_rate": 8.024368382919665e-06, + "loss": 0.1316, + "step": 11012 + }, + { + "epoch": 4.61, + "learning_rate": 8.022706418680398e-06, + "loss": 0.1592, + "step": 11013 + }, + { + "epoch": 4.61, + "learning_rate": 8.021044511276682e-06, + "loss": 0.1634, + "step": 11014 + }, + { + "epoch": 4.61, + "learning_rate": 8.019382660756284e-06, + "loss": 0.1346, + "step": 11015 + }, + { + "epoch": 4.61, + "learning_rate": 8.017720867166976e-06, + "loss": 0.142, + "step": 11016 + }, + { + "epoch": 4.61, + "learning_rate": 8.01605913055652e-06, + "loss": 0.1682, + "step": 11017 + }, + { + "epoch": 4.61, + "learning_rate": 8.014397450972685e-06, + "loss": 0.156, + "step": 11018 + }, + { + "epoch": 4.61, + "learning_rate": 8.012735828463233e-06, + "loss": 0.1653, + "step": 11019 + }, + { + "epoch": 4.61, + "learning_rate": 8.011074263075925e-06, + "loss": 0.1269, + "step": 11020 + }, + { + "epoch": 4.61, + "learning_rate": 8.009412754858525e-06, + "loss": 0.1328, + "step": 11021 + }, + { + "epoch": 4.61, + "learning_rate": 8.007751303858788e-06, + "loss": 0.1473, + "step": 11022 + }, + { + "epoch": 4.62, + "learning_rate": 8.00608991012447e-06, + "loss": 0.108, + "step": 11023 + }, + { + "epoch": 4.62, + "learning_rate": 8.00442857370333e-06, + "loss": 0.1977, + "step": 11024 + }, + { + "epoch": 4.62, + "learning_rate": 8.00276729464312e-06, + "loss": 0.1484, + "step": 11025 + }, + { + "epoch": 4.62, + "learning_rate": 8.00110607299159e-06, + "loss": 0.1721, + "step": 11026 + }, + { + "epoch": 4.62, + "learning_rate": 7.999444908796493e-06, + "loss": 0.1437, + "step": 11027 + }, + { + "epoch": 4.62, + "learning_rate": 7.99778380210558e-06, + "loss": 0.132, + "step": 11028 + }, + { + "epoch": 4.62, + "learning_rate": 7.996122752966596e-06, + "loss": 0.148, + "step": 11029 + }, + { + "epoch": 4.62, + "learning_rate": 7.99446176142728e-06, + "loss": 0.1464, + "step": 11030 + }, + { + "epoch": 4.62, + "learning_rate": 7.992800827535381e-06, + "loss": 0.1883, + "step": 11031 + }, + { + "epoch": 4.62, + "learning_rate": 7.991139951338645e-06, + "loss": 0.1545, + "step": 11032 + }, + { + "epoch": 4.62, + "learning_rate": 7.989479132884803e-06, + "loss": 0.1339, + "step": 11033 + }, + { + "epoch": 4.62, + "learning_rate": 7.987818372221603e-06, + "loss": 0.1387, + "step": 11034 + }, + { + "epoch": 4.62, + "learning_rate": 7.98615766939678e-06, + "loss": 0.1584, + "step": 11035 + }, + { + "epoch": 4.62, + "learning_rate": 7.984497024458062e-06, + "loss": 0.128, + "step": 11036 + }, + { + "epoch": 4.62, + "learning_rate": 7.982836437453192e-06, + "loss": 0.1636, + "step": 11037 + }, + { + "epoch": 4.62, + "learning_rate": 7.9811759084299e-06, + "loss": 0.1627, + "step": 11038 + }, + { + "epoch": 4.62, + "learning_rate": 7.979515437435913e-06, + "loss": 0.1573, + "step": 11039 + }, + { + "epoch": 4.62, + "learning_rate": 7.977855024518962e-06, + "loss": 0.1787, + "step": 11040 + }, + { + "epoch": 4.62, + "learning_rate": 7.976194669726773e-06, + "loss": 0.1583, + "step": 11041 + }, + { + "epoch": 4.62, + "learning_rate": 7.974534373107074e-06, + "loss": 0.1613, + "step": 11042 + }, + { + "epoch": 4.62, + "learning_rate": 7.972874134707585e-06, + "loss": 0.1165, + "step": 11043 + }, + { + "epoch": 4.62, + "learning_rate": 7.971213954576031e-06, + "loss": 0.1546, + "step": 11044 + }, + { + "epoch": 4.62, + "learning_rate": 7.969553832760134e-06, + "loss": 0.1246, + "step": 11045 + }, + { + "epoch": 4.62, + "learning_rate": 7.967893769307608e-06, + "loss": 0.1284, + "step": 11046 + }, + { + "epoch": 4.63, + "learning_rate": 7.966233764266172e-06, + "loss": 0.1343, + "step": 11047 + }, + { + "epoch": 4.63, + "learning_rate": 7.964573817683545e-06, + "loss": 0.1521, + "step": 11048 + }, + { + "epoch": 4.63, + "learning_rate": 7.962913929607433e-06, + "loss": 0.1585, + "step": 11049 + }, + { + "epoch": 4.63, + "learning_rate": 7.961254100085554e-06, + "loss": 0.132, + "step": 11050 + }, + { + "epoch": 4.63, + "learning_rate": 7.95959432916562e-06, + "loss": 0.1382, + "step": 11051 + }, + { + "epoch": 4.63, + "learning_rate": 7.957934616895332e-06, + "loss": 0.1273, + "step": 11052 + }, + { + "epoch": 4.63, + "learning_rate": 7.956274963322406e-06, + "loss": 0.1409, + "step": 11053 + }, + { + "epoch": 4.63, + "learning_rate": 7.954615368494538e-06, + "loss": 0.1615, + "step": 11054 + }, + { + "epoch": 4.63, + "learning_rate": 7.952955832459437e-06, + "loss": 0.1566, + "step": 11055 + }, + { + "epoch": 4.63, + "learning_rate": 7.951296355264805e-06, + "loss": 0.1366, + "step": 11056 + }, + { + "epoch": 4.63, + "learning_rate": 7.949636936958342e-06, + "loss": 0.1468, + "step": 11057 + }, + { + "epoch": 4.63, + "learning_rate": 7.947977577587745e-06, + "loss": 0.1938, + "step": 11058 + }, + { + "epoch": 4.63, + "learning_rate": 7.946318277200712e-06, + "loss": 0.1244, + "step": 11059 + }, + { + "epoch": 4.63, + "learning_rate": 7.944659035844937e-06, + "loss": 0.1601, + "step": 11060 + }, + { + "epoch": 4.63, + "learning_rate": 7.942999853568112e-06, + "loss": 0.1841, + "step": 11061 + }, + { + "epoch": 4.63, + "learning_rate": 7.941340730417934e-06, + "loss": 0.1518, + "step": 11062 + }, + { + "epoch": 4.63, + "learning_rate": 7.93968166644209e-06, + "loss": 0.1456, + "step": 11063 + }, + { + "epoch": 4.63, + "learning_rate": 7.938022661688265e-06, + "loss": 0.137, + "step": 11064 + }, + { + "epoch": 4.63, + "learning_rate": 7.936363716204153e-06, + "loss": 0.1501, + "step": 11065 + }, + { + "epoch": 4.63, + "learning_rate": 7.934704830037432e-06, + "loss": 0.1474, + "step": 11066 + }, + { + "epoch": 4.63, + "learning_rate": 7.933046003235789e-06, + "loss": 0.1521, + "step": 11067 + }, + { + "epoch": 4.63, + "learning_rate": 7.931387235846905e-06, + "loss": 0.1458, + "step": 11068 + }, + { + "epoch": 4.63, + "learning_rate": 7.929728527918456e-06, + "loss": 0.1276, + "step": 11069 + }, + { + "epoch": 4.63, + "learning_rate": 7.928069879498126e-06, + "loss": 0.1618, + "step": 11070 + }, + { + "epoch": 4.64, + "learning_rate": 7.926411290633588e-06, + "loss": 0.1723, + "step": 11071 + }, + { + "epoch": 4.64, + "learning_rate": 7.92475276137252e-06, + "loss": 0.151, + "step": 11072 + }, + { + "epoch": 4.64, + "learning_rate": 7.923094291762588e-06, + "loss": 0.1688, + "step": 11073 + }, + { + "epoch": 4.64, + "learning_rate": 7.921435881851474e-06, + "loss": 0.1447, + "step": 11074 + }, + { + "epoch": 4.64, + "learning_rate": 7.919777531686837e-06, + "loss": 0.1317, + "step": 11075 + }, + { + "epoch": 4.64, + "learning_rate": 7.91811924131635e-06, + "loss": 0.1612, + "step": 11076 + }, + { + "epoch": 4.64, + "learning_rate": 7.916461010787676e-06, + "loss": 0.1556, + "step": 11077 + }, + { + "epoch": 4.64, + "learning_rate": 7.914802840148482e-06, + "loss": 0.1761, + "step": 11078 + }, + { + "epoch": 4.64, + "learning_rate": 7.91314472944643e-06, + "loss": 0.1232, + "step": 11079 + }, + { + "epoch": 4.64, + "learning_rate": 7.91148667872918e-06, + "loss": 0.1865, + "step": 11080 + }, + { + "epoch": 4.64, + "learning_rate": 7.909828688044392e-06, + "loss": 0.1404, + "step": 11081 + }, + { + "epoch": 4.64, + "learning_rate": 7.908170757439724e-06, + "loss": 0.148, + "step": 11082 + }, + { + "epoch": 4.64, + "learning_rate": 7.906512886962833e-06, + "loss": 0.1539, + "step": 11083 + }, + { + "epoch": 4.64, + "learning_rate": 7.90485507666137e-06, + "loss": 0.1607, + "step": 11084 + }, + { + "epoch": 4.64, + "learning_rate": 7.903197326582988e-06, + "loss": 0.1351, + "step": 11085 + }, + { + "epoch": 4.64, + "learning_rate": 7.90153963677534e-06, + "loss": 0.1095, + "step": 11086 + }, + { + "epoch": 4.64, + "learning_rate": 7.899882007286074e-06, + "loss": 0.1682, + "step": 11087 + }, + { + "epoch": 4.64, + "learning_rate": 7.898224438162831e-06, + "loss": 0.1204, + "step": 11088 + }, + { + "epoch": 4.64, + "learning_rate": 7.896566929453267e-06, + "loss": 0.166, + "step": 11089 + }, + { + "epoch": 4.64, + "learning_rate": 7.894909481205017e-06, + "loss": 0.118, + "step": 11090 + }, + { + "epoch": 4.64, + "learning_rate": 7.893252093465725e-06, + "loss": 0.133, + "step": 11091 + }, + { + "epoch": 4.64, + "learning_rate": 7.891594766283034e-06, + "loss": 0.1369, + "step": 11092 + }, + { + "epoch": 4.64, + "learning_rate": 7.88993749970458e-06, + "loss": 0.1728, + "step": 11093 + }, + { + "epoch": 4.64, + "learning_rate": 7.888280293778001e-06, + "loss": 0.1538, + "step": 11094 + }, + { + "epoch": 4.65, + "learning_rate": 7.886623148550932e-06, + "loss": 0.1335, + "step": 11095 + }, + { + "epoch": 4.65, + "learning_rate": 7.884966064071002e-06, + "loss": 0.1191, + "step": 11096 + }, + { + "epoch": 4.65, + "learning_rate": 7.88330904038585e-06, + "loss": 0.1459, + "step": 11097 + }, + { + "epoch": 4.65, + "learning_rate": 7.881652077543098e-06, + "loss": 0.1629, + "step": 11098 + }, + { + "epoch": 4.65, + "learning_rate": 7.879995175590378e-06, + "loss": 0.1501, + "step": 11099 + }, + { + "epoch": 4.65, + "learning_rate": 7.878338334575317e-06, + "loss": 0.189, + "step": 11100 + }, + { + "epoch": 4.65, + "learning_rate": 7.876681554545535e-06, + "loss": 0.1861, + "step": 11101 + }, + { + "epoch": 4.65, + "learning_rate": 7.875024835548658e-06, + "loss": 0.1758, + "step": 11102 + }, + { + "epoch": 4.65, + "learning_rate": 7.873368177632308e-06, + "loss": 0.1761, + "step": 11103 + }, + { + "epoch": 4.65, + "learning_rate": 7.871711580844102e-06, + "loss": 0.159, + "step": 11104 + }, + { + "epoch": 4.65, + "learning_rate": 7.870055045231656e-06, + "loss": 0.1334, + "step": 11105 + }, + { + "epoch": 4.65, + "learning_rate": 7.86839857084259e-06, + "loss": 0.1364, + "step": 11106 + }, + { + "epoch": 4.65, + "learning_rate": 7.866742157724515e-06, + "loss": 0.1134, + "step": 11107 + }, + { + "epoch": 4.65, + "learning_rate": 7.865085805925042e-06, + "loss": 0.1967, + "step": 11108 + }, + { + "epoch": 4.65, + "learning_rate": 7.863429515491786e-06, + "loss": 0.1544, + "step": 11109 + }, + { + "epoch": 4.65, + "learning_rate": 7.861773286472351e-06, + "loss": 0.1231, + "step": 11110 + }, + { + "epoch": 4.65, + "learning_rate": 7.860117118914345e-06, + "loss": 0.1251, + "step": 11111 + }, + { + "epoch": 4.65, + "learning_rate": 7.858461012865373e-06, + "loss": 0.1512, + "step": 11112 + }, + { + "epoch": 4.65, + "learning_rate": 7.856804968373041e-06, + "loss": 0.2215, + "step": 11113 + }, + { + "epoch": 4.65, + "learning_rate": 7.855148985484946e-06, + "loss": 0.135, + "step": 11114 + }, + { + "epoch": 4.65, + "learning_rate": 7.85349306424869e-06, + "loss": 0.1436, + "step": 11115 + }, + { + "epoch": 4.65, + "learning_rate": 7.851837204711874e-06, + "loss": 0.2166, + "step": 11116 + }, + { + "epoch": 4.65, + "learning_rate": 7.85018140692209e-06, + "loss": 0.1807, + "step": 11117 + }, + { + "epoch": 4.65, + "learning_rate": 7.848525670926932e-06, + "loss": 0.1591, + "step": 11118 + }, + { + "epoch": 4.66, + "learning_rate": 7.846869996774e-06, + "loss": 0.1464, + "step": 11119 + }, + { + "epoch": 4.66, + "learning_rate": 7.845214384510875e-06, + "loss": 0.1911, + "step": 11120 + }, + { + "epoch": 4.66, + "learning_rate": 7.84355883418515e-06, + "loss": 0.1695, + "step": 11121 + }, + { + "epoch": 4.66, + "learning_rate": 7.841903345844413e-06, + "loss": 0.1471, + "step": 11122 + }, + { + "epoch": 4.66, + "learning_rate": 7.840247919536249e-06, + "loss": 0.1358, + "step": 11123 + }, + { + "epoch": 4.66, + "learning_rate": 7.838592555308244e-06, + "loss": 0.1547, + "step": 11124 + }, + { + "epoch": 4.66, + "learning_rate": 7.836937253207976e-06, + "loss": 0.1534, + "step": 11125 + }, + { + "epoch": 4.66, + "learning_rate": 7.835282013283027e-06, + "loss": 0.1508, + "step": 11126 + }, + { + "epoch": 4.66, + "learning_rate": 7.833626835580977e-06, + "loss": 0.1305, + "step": 11127 + }, + { + "epoch": 4.66, + "learning_rate": 7.8319717201494e-06, + "loss": 0.1499, + "step": 11128 + }, + { + "epoch": 4.66, + "learning_rate": 7.830316667035872e-06, + "loss": 0.1538, + "step": 11129 + }, + { + "epoch": 4.66, + "learning_rate": 7.828661676287968e-06, + "loss": 0.1356, + "step": 11130 + }, + { + "epoch": 4.66, + "learning_rate": 7.827006747953255e-06, + "loss": 0.1237, + "step": 11131 + }, + { + "epoch": 4.66, + "learning_rate": 7.825351882079305e-06, + "loss": 0.1834, + "step": 11132 + }, + { + "epoch": 4.66, + "learning_rate": 7.823697078713689e-06, + "loss": 0.1415, + "step": 11133 + }, + { + "epoch": 4.66, + "learning_rate": 7.822042337903964e-06, + "loss": 0.1404, + "step": 11134 + }, + { + "epoch": 4.66, + "learning_rate": 7.820387659697701e-06, + "loss": 0.1441, + "step": 11135 + }, + { + "epoch": 4.66, + "learning_rate": 7.818733044142462e-06, + "loss": 0.1424, + "step": 11136 + }, + { + "epoch": 4.66, + "learning_rate": 7.817078491285806e-06, + "loss": 0.1405, + "step": 11137 + }, + { + "epoch": 4.66, + "learning_rate": 7.815424001175289e-06, + "loss": 0.1463, + "step": 11138 + }, + { + "epoch": 4.66, + "learning_rate": 7.813769573858474e-06, + "loss": 0.1954, + "step": 11139 + }, + { + "epoch": 4.66, + "learning_rate": 7.81211520938291e-06, + "loss": 0.1358, + "step": 11140 + }, + { + "epoch": 4.66, + "learning_rate": 7.810460907796153e-06, + "loss": 0.1541, + "step": 11141 + }, + { + "epoch": 4.66, + "learning_rate": 7.80880666914576e-06, + "loss": 0.1566, + "step": 11142 + }, + { + "epoch": 4.67, + "learning_rate": 7.807152493479268e-06, + "loss": 0.1288, + "step": 11143 + }, + { + "epoch": 4.67, + "learning_rate": 7.805498380844236e-06, + "loss": 0.1411, + "step": 11144 + }, + { + "epoch": 4.67, + "learning_rate": 7.803844331288201e-06, + "loss": 0.1424, + "step": 11145 + }, + { + "epoch": 4.67, + "learning_rate": 7.802190344858715e-06, + "loss": 0.1793, + "step": 11146 + }, + { + "epoch": 4.67, + "learning_rate": 7.800536421603317e-06, + "loss": 0.1316, + "step": 11147 + }, + { + "epoch": 4.67, + "learning_rate": 7.798882561569546e-06, + "loss": 0.1403, + "step": 11148 + }, + { + "epoch": 4.67, + "learning_rate": 7.797228764804942e-06, + "loss": 0.1246, + "step": 11149 + }, + { + "epoch": 4.67, + "learning_rate": 7.795575031357047e-06, + "loss": 0.1353, + "step": 11150 + }, + { + "epoch": 4.67, + "learning_rate": 7.793921361273388e-06, + "loss": 0.183, + "step": 11151 + }, + { + "epoch": 4.67, + "learning_rate": 7.792267754601501e-06, + "loss": 0.1039, + "step": 11152 + }, + { + "epoch": 4.67, + "learning_rate": 7.79061421138892e-06, + "loss": 0.1485, + "step": 11153 + }, + { + "epoch": 4.67, + "learning_rate": 7.788960731683174e-06, + "loss": 0.1523, + "step": 11154 + }, + { + "epoch": 4.67, + "learning_rate": 7.787307315531787e-06, + "loss": 0.1523, + "step": 11155 + }, + { + "epoch": 4.67, + "learning_rate": 7.785653962982288e-06, + "loss": 0.1329, + "step": 11156 + }, + { + "epoch": 4.67, + "learning_rate": 7.784000674082205e-06, + "loss": 0.1695, + "step": 11157 + }, + { + "epoch": 4.67, + "learning_rate": 7.782347448879051e-06, + "loss": 0.1421, + "step": 11158 + }, + { + "epoch": 4.67, + "learning_rate": 7.780694287420355e-06, + "loss": 0.1418, + "step": 11159 + }, + { + "epoch": 4.67, + "learning_rate": 7.77904118975363e-06, + "loss": 0.1341, + "step": 11160 + }, + { + "epoch": 4.67, + "learning_rate": 7.777388155926397e-06, + "loss": 0.16, + "step": 11161 + }, + { + "epoch": 4.67, + "learning_rate": 7.775735185986168e-06, + "loss": 0.1736, + "step": 11162 + }, + { + "epoch": 4.67, + "learning_rate": 7.774082279980459e-06, + "loss": 0.1503, + "step": 11163 + }, + { + "epoch": 4.67, + "learning_rate": 7.772429437956777e-06, + "loss": 0.1389, + "step": 11164 + }, + { + "epoch": 4.67, + "learning_rate": 7.77077665996264e-06, + "loss": 0.1582, + "step": 11165 + }, + { + "epoch": 4.67, + "learning_rate": 7.769123946045545e-06, + "loss": 0.1867, + "step": 11166 + }, + { + "epoch": 4.68, + "learning_rate": 7.767471296253003e-06, + "loss": 0.1555, + "step": 11167 + }, + { + "epoch": 4.68, + "learning_rate": 7.765818710632516e-06, + "loss": 0.1818, + "step": 11168 + }, + { + "epoch": 4.68, + "learning_rate": 7.764166189231588e-06, + "loss": 0.1169, + "step": 11169 + }, + { + "epoch": 4.68, + "learning_rate": 7.762513732097721e-06, + "loss": 0.1792, + "step": 11170 + }, + { + "epoch": 4.68, + "learning_rate": 7.76086133927841e-06, + "loss": 0.1554, + "step": 11171 + }, + { + "epoch": 4.68, + "learning_rate": 7.759209010821152e-06, + "loss": 0.1676, + "step": 11172 + }, + { + "epoch": 4.68, + "learning_rate": 7.757556746773443e-06, + "loss": 0.1391, + "step": 11173 + }, + { + "epoch": 4.68, + "learning_rate": 7.755904547182778e-06, + "loss": 0.1281, + "step": 11174 + }, + { + "epoch": 4.68, + "learning_rate": 7.754252412096642e-06, + "loss": 0.1502, + "step": 11175 + }, + { + "epoch": 4.68, + "learning_rate": 7.752600341562527e-06, + "loss": 0.152, + "step": 11176 + }, + { + "epoch": 4.68, + "learning_rate": 7.750948335627925e-06, + "loss": 0.1148, + "step": 11177 + }, + { + "epoch": 4.68, + "learning_rate": 7.749296394340313e-06, + "loss": 0.1783, + "step": 11178 + }, + { + "epoch": 4.68, + "learning_rate": 7.747644517747182e-06, + "loss": 0.1686, + "step": 11179 + }, + { + "epoch": 4.68, + "learning_rate": 7.74599270589601e-06, + "loss": 0.1007, + "step": 11180 + }, + { + "epoch": 4.68, + "learning_rate": 7.744340958834275e-06, + "loss": 0.1236, + "step": 11181 + }, + { + "epoch": 4.68, + "learning_rate": 7.742689276609459e-06, + "loss": 0.1305, + "step": 11182 + }, + { + "epoch": 4.68, + "learning_rate": 7.741037659269037e-06, + "loss": 0.1455, + "step": 11183 + }, + { + "epoch": 4.68, + "learning_rate": 7.739386106860485e-06, + "loss": 0.1433, + "step": 11184 + }, + { + "epoch": 4.68, + "learning_rate": 7.73773461943127e-06, + "loss": 0.1277, + "step": 11185 + }, + { + "epoch": 4.68, + "learning_rate": 7.736083197028866e-06, + "loss": 0.2145, + "step": 11186 + }, + { + "epoch": 4.68, + "learning_rate": 7.734431839700747e-06, + "loss": 0.1557, + "step": 11187 + }, + { + "epoch": 4.68, + "learning_rate": 7.732780547494372e-06, + "loss": 0.1628, + "step": 11188 + }, + { + "epoch": 4.68, + "learning_rate": 7.731129320457205e-06, + "loss": 0.1719, + "step": 11189 + }, + { + "epoch": 4.68, + "learning_rate": 7.729478158636712e-06, + "loss": 0.1289, + "step": 11190 + }, + { + "epoch": 4.69, + "learning_rate": 7.727827062080359e-06, + "loss": 0.1182, + "step": 11191 + }, + { + "epoch": 4.69, + "learning_rate": 7.726176030835597e-06, + "loss": 0.1314, + "step": 11192 + }, + { + "epoch": 4.69, + "learning_rate": 7.724525064949887e-06, + "loss": 0.1119, + "step": 11193 + }, + { + "epoch": 4.69, + "learning_rate": 7.722874164470685e-06, + "loss": 0.1527, + "step": 11194 + }, + { + "epoch": 4.69, + "learning_rate": 7.721223329445448e-06, + "loss": 0.1525, + "step": 11195 + }, + { + "epoch": 4.69, + "learning_rate": 7.719572559921619e-06, + "loss": 0.177, + "step": 11196 + }, + { + "epoch": 4.69, + "learning_rate": 7.717921855946656e-06, + "loss": 0.1636, + "step": 11197 + }, + { + "epoch": 4.69, + "learning_rate": 7.716271217568005e-06, + "loss": 0.1738, + "step": 11198 + }, + { + "epoch": 4.69, + "learning_rate": 7.71462064483311e-06, + "loss": 0.1189, + "step": 11199 + }, + { + "epoch": 4.69, + "learning_rate": 7.712970137789416e-06, + "loss": 0.1319, + "step": 11200 + }, + { + "epoch": 4.69, + "learning_rate": 7.711319696484368e-06, + "loss": 0.1306, + "step": 11201 + }, + { + "epoch": 4.69, + "learning_rate": 7.709669320965402e-06, + "loss": 0.1446, + "step": 11202 + }, + { + "epoch": 4.69, + "learning_rate": 7.70801901127996e-06, + "loss": 0.1587, + "step": 11203 + }, + { + "epoch": 4.69, + "learning_rate": 7.70636876747548e-06, + "loss": 0.1619, + "step": 11204 + }, + { + "epoch": 4.69, + "learning_rate": 7.70471858959939e-06, + "loss": 0.1482, + "step": 11205 + }, + { + "epoch": 4.69, + "learning_rate": 7.70306847769913e-06, + "loss": 0.1366, + "step": 11206 + }, + { + "epoch": 4.69, + "learning_rate": 7.70141843182213e-06, + "loss": 0.1257, + "step": 11207 + }, + { + "epoch": 4.69, + "learning_rate": 7.699768452015815e-06, + "loss": 0.151, + "step": 11208 + }, + { + "epoch": 4.69, + "learning_rate": 7.698118538327613e-06, + "loss": 0.1567, + "step": 11209 + }, + { + "epoch": 4.69, + "learning_rate": 7.696468690804956e-06, + "loss": 0.1596, + "step": 11210 + }, + { + "epoch": 4.69, + "learning_rate": 7.69481890949526e-06, + "loss": 0.1584, + "step": 11211 + }, + { + "epoch": 4.69, + "learning_rate": 7.69316919444595e-06, + "loss": 0.1776, + "step": 11212 + }, + { + "epoch": 4.69, + "learning_rate": 7.691519545704442e-06, + "loss": 0.1568, + "step": 11213 + }, + { + "epoch": 4.69, + "learning_rate": 7.689869963318154e-06, + "loss": 0.1598, + "step": 11214 + }, + { + "epoch": 4.7, + "learning_rate": 7.688220447334508e-06, + "loss": 0.1739, + "step": 11215 + }, + { + "epoch": 4.7, + "learning_rate": 7.686570997800914e-06, + "loss": 0.1726, + "step": 11216 + }, + { + "epoch": 4.7, + "learning_rate": 7.68492161476478e-06, + "loss": 0.1606, + "step": 11217 + }, + { + "epoch": 4.7, + "learning_rate": 7.683272298273525e-06, + "loss": 0.1754, + "step": 11218 + }, + { + "epoch": 4.7, + "learning_rate": 7.681623048374549e-06, + "loss": 0.1424, + "step": 11219 + }, + { + "epoch": 4.7, + "learning_rate": 7.67997386511526e-06, + "loss": 0.2035, + "step": 11220 + }, + { + "epoch": 4.7, + "learning_rate": 7.678324748543066e-06, + "loss": 0.2215, + "step": 11221 + }, + { + "epoch": 4.7, + "learning_rate": 7.676675698705364e-06, + "loss": 0.1417, + "step": 11222 + }, + { + "epoch": 4.7, + "learning_rate": 7.67502671564956e-06, + "loss": 0.1331, + "step": 11223 + }, + { + "epoch": 4.7, + "learning_rate": 7.673377799423047e-06, + "loss": 0.1634, + "step": 11224 + }, + { + "epoch": 4.7, + "learning_rate": 7.671728950073228e-06, + "loss": 0.1274, + "step": 11225 + }, + { + "epoch": 4.7, + "learning_rate": 7.670080167647492e-06, + "loss": 0.1679, + "step": 11226 + }, + { + "epoch": 4.7, + "learning_rate": 7.668431452193234e-06, + "loss": 0.1328, + "step": 11227 + }, + { + "epoch": 4.7, + "learning_rate": 7.666782803757848e-06, + "loss": 0.1429, + "step": 11228 + }, + { + "epoch": 4.7, + "learning_rate": 7.665134222388717e-06, + "loss": 0.1673, + "step": 11229 + }, + { + "epoch": 4.7, + "learning_rate": 7.66348570813323e-06, + "loss": 0.1549, + "step": 11230 + }, + { + "epoch": 4.7, + "learning_rate": 7.661837261038776e-06, + "loss": 0.1568, + "step": 11231 + }, + { + "epoch": 4.7, + "learning_rate": 7.660188881152732e-06, + "loss": 0.1671, + "step": 11232 + }, + { + "epoch": 4.7, + "learning_rate": 7.658540568522487e-06, + "loss": 0.1755, + "step": 11233 + }, + { + "epoch": 4.7, + "learning_rate": 7.656892323195414e-06, + "loss": 0.1713, + "step": 11234 + }, + { + "epoch": 4.7, + "learning_rate": 7.655244145218887e-06, + "loss": 0.1667, + "step": 11235 + }, + { + "epoch": 4.7, + "learning_rate": 7.653596034640291e-06, + "loss": 0.166, + "step": 11236 + }, + { + "epoch": 4.7, + "learning_rate": 7.651947991506994e-06, + "loss": 0.1592, + "step": 11237 + }, + { + "epoch": 4.71, + "learning_rate": 7.650300015866367e-06, + "loss": 0.1462, + "step": 11238 + }, + { + "epoch": 4.71, + "learning_rate": 7.648652107765783e-06, + "loss": 0.1501, + "step": 11239 + }, + { + "epoch": 4.71, + "learning_rate": 7.647004267252606e-06, + "loss": 0.1555, + "step": 11240 + }, + { + "epoch": 4.71, + "learning_rate": 7.645356494374203e-06, + "loss": 0.1119, + "step": 11241 + }, + { + "epoch": 4.71, + "learning_rate": 7.643708789177939e-06, + "loss": 0.1433, + "step": 11242 + }, + { + "epoch": 4.71, + "learning_rate": 7.642061151711176e-06, + "loss": 0.1535, + "step": 11243 + }, + { + "epoch": 4.71, + "learning_rate": 7.64041358202127e-06, + "loss": 0.1613, + "step": 11244 + }, + { + "epoch": 4.71, + "learning_rate": 7.638766080155584e-06, + "loss": 0.1381, + "step": 11245 + }, + { + "epoch": 4.71, + "learning_rate": 7.63711864616147e-06, + "loss": 0.1383, + "step": 11246 + }, + { + "epoch": 4.71, + "learning_rate": 7.635471280086285e-06, + "loss": 0.1832, + "step": 11247 + }, + { + "epoch": 4.71, + "learning_rate": 7.63382398197738e-06, + "loss": 0.1316, + "step": 11248 + }, + { + "epoch": 4.71, + "learning_rate": 7.632176751882103e-06, + "loss": 0.1394, + "step": 11249 + }, + { + "epoch": 4.71, + "learning_rate": 7.630529589847807e-06, + "loss": 0.1282, + "step": 11250 + }, + { + "epoch": 4.71, + "learning_rate": 7.628882495921834e-06, + "loss": 0.1677, + "step": 11251 + }, + { + "epoch": 4.71, + "learning_rate": 7.627235470151528e-06, + "loss": 0.2038, + "step": 11252 + }, + { + "epoch": 4.71, + "learning_rate": 7.625588512584234e-06, + "loss": 0.1632, + "step": 11253 + }, + { + "epoch": 4.71, + "learning_rate": 7.6239416232672904e-06, + "loss": 0.1553, + "step": 11254 + }, + { + "epoch": 4.71, + "learning_rate": 7.622294802248041e-06, + "loss": 0.1566, + "step": 11255 + }, + { + "epoch": 4.71, + "learning_rate": 7.620648049573815e-06, + "loss": 0.1448, + "step": 11256 + }, + { + "epoch": 4.71, + "learning_rate": 7.619001365291948e-06, + "loss": 0.141, + "step": 11257 + }, + { + "epoch": 4.71, + "learning_rate": 7.617354749449774e-06, + "loss": 0.1419, + "step": 11258 + }, + { + "epoch": 4.71, + "learning_rate": 7.615708202094625e-06, + "loss": 0.1363, + "step": 11259 + }, + { + "epoch": 4.71, + "learning_rate": 7.614061723273825e-06, + "loss": 0.1554, + "step": 11260 + }, + { + "epoch": 4.71, + "learning_rate": 7.612415313034706e-06, + "loss": 0.1467, + "step": 11261 + }, + { + "epoch": 4.72, + "learning_rate": 7.61076897142459e-06, + "loss": 0.1473, + "step": 11262 + }, + { + "epoch": 4.72, + "learning_rate": 7.6091226984907985e-06, + "loss": 0.1502, + "step": 11263 + }, + { + "epoch": 4.72, + "learning_rate": 7.607476494280653e-06, + "loss": 0.1061, + "step": 11264 + }, + { + "epoch": 4.72, + "learning_rate": 7.6058303588414725e-06, + "loss": 0.2, + "step": 11265 + }, + { + "epoch": 4.72, + "learning_rate": 7.604184292220577e-06, + "loss": 0.1315, + "step": 11266 + }, + { + "epoch": 4.72, + "learning_rate": 7.602538294465273e-06, + "loss": 0.1253, + "step": 11267 + }, + { + "epoch": 4.72, + "learning_rate": 7.6008923656228814e-06, + "loss": 0.1226, + "step": 11268 + }, + { + "epoch": 4.72, + "learning_rate": 7.599246505740711e-06, + "loss": 0.1397, + "step": 11269 + }, + { + "epoch": 4.72, + "learning_rate": 7.597600714866067e-06, + "loss": 0.1627, + "step": 11270 + }, + { + "epoch": 4.72, + "learning_rate": 7.595954993046259e-06, + "loss": 0.1419, + "step": 11271 + }, + { + "epoch": 4.72, + "learning_rate": 7.594309340328594e-06, + "loss": 0.1498, + "step": 11272 + }, + { + "epoch": 4.72, + "learning_rate": 7.592663756760369e-06, + "loss": 0.1064, + "step": 11273 + }, + { + "epoch": 4.72, + "learning_rate": 7.591018242388888e-06, + "loss": 0.1151, + "step": 11274 + }, + { + "epoch": 4.72, + "learning_rate": 7.589372797261453e-06, + "loss": 0.1374, + "step": 11275 + }, + { + "epoch": 4.72, + "learning_rate": 7.587727421425355e-06, + "loss": 0.1407, + "step": 11276 + }, + { + "epoch": 4.72, + "learning_rate": 7.586082114927892e-06, + "loss": 0.1609, + "step": 11277 + }, + { + "epoch": 4.72, + "learning_rate": 7.58443687781636e-06, + "loss": 0.1706, + "step": 11278 + }, + { + "epoch": 4.72, + "learning_rate": 7.582791710138043e-06, + "loss": 0.1377, + "step": 11279 + }, + { + "epoch": 4.72, + "learning_rate": 7.581146611940235e-06, + "loss": 0.1213, + "step": 11280 + }, + { + "epoch": 4.72, + "learning_rate": 7.57950158327022e-06, + "loss": 0.1416, + "step": 11281 + }, + { + "epoch": 4.72, + "learning_rate": 7.577856624175283e-06, + "loss": 0.1573, + "step": 11282 + }, + { + "epoch": 4.72, + "learning_rate": 7.576211734702711e-06, + "loss": 0.155, + "step": 11283 + }, + { + "epoch": 4.72, + "learning_rate": 7.574566914899779e-06, + "loss": 0.0967, + "step": 11284 + }, + { + "epoch": 4.72, + "learning_rate": 7.5729221648137686e-06, + "loss": 0.1409, + "step": 11285 + }, + { + "epoch": 4.73, + "learning_rate": 7.571277484491959e-06, + "loss": 0.1336, + "step": 11286 + }, + { + "epoch": 4.73, + "learning_rate": 7.569632873981622e-06, + "loss": 0.1512, + "step": 11287 + }, + { + "epoch": 4.73, + "learning_rate": 7.56798833333003e-06, + "loss": 0.1534, + "step": 11288 + }, + { + "epoch": 4.73, + "learning_rate": 7.566343862584458e-06, + "loss": 0.1724, + "step": 11289 + }, + { + "epoch": 4.73, + "learning_rate": 7.56469946179217e-06, + "loss": 0.129, + "step": 11290 + }, + { + "epoch": 4.73, + "learning_rate": 7.563055131000434e-06, + "loss": 0.1654, + "step": 11291 + }, + { + "epoch": 4.73, + "learning_rate": 7.561410870256518e-06, + "loss": 0.1414, + "step": 11292 + }, + { + "epoch": 4.73, + "learning_rate": 7.559766679607682e-06, + "loss": 0.1662, + "step": 11293 + }, + { + "epoch": 4.73, + "learning_rate": 7.5581225591011864e-06, + "loss": 0.1258, + "step": 11294 + }, + { + "epoch": 4.73, + "learning_rate": 7.556478508784291e-06, + "loss": 0.1049, + "step": 11295 + }, + { + "epoch": 4.73, + "learning_rate": 7.554834528704254e-06, + "loss": 0.1295, + "step": 11296 + }, + { + "epoch": 4.73, + "learning_rate": 7.553190618908327e-06, + "loss": 0.0941, + "step": 11297 + }, + { + "epoch": 4.73, + "learning_rate": 7.551546779443765e-06, + "loss": 0.0995, + "step": 11298 + }, + { + "epoch": 4.73, + "learning_rate": 7.54990301035782e-06, + "loss": 0.1157, + "step": 11299 + }, + { + "epoch": 4.73, + "learning_rate": 7.5482593116977355e-06, + "loss": 0.1149, + "step": 11300 + }, + { + "epoch": 4.73, + "learning_rate": 7.546615683510767e-06, + "loss": 0.1697, + "step": 11301 + }, + { + "epoch": 4.73, + "learning_rate": 7.544972125844149e-06, + "loss": 0.1343, + "step": 11302 + }, + { + "epoch": 4.73, + "learning_rate": 7.543328638745131e-06, + "loss": 0.099, + "step": 11303 + }, + { + "epoch": 4.73, + "learning_rate": 7.541685222260948e-06, + "loss": 0.1647, + "step": 11304 + }, + { + "epoch": 4.73, + "learning_rate": 7.540041876438842e-06, + "loss": 0.1305, + "step": 11305 + }, + { + "epoch": 4.73, + "learning_rate": 7.538398601326051e-06, + "loss": 0.1737, + "step": 11306 + }, + { + "epoch": 4.73, + "learning_rate": 7.5367553969698095e-06, + "loss": 0.1746, + "step": 11307 + }, + { + "epoch": 4.73, + "learning_rate": 7.535112263417346e-06, + "loss": 0.1647, + "step": 11308 + }, + { + "epoch": 4.73, + "learning_rate": 7.533469200715892e-06, + "loss": 0.1599, + "step": 11309 + }, + { + "epoch": 4.74, + "learning_rate": 7.53182620891268e-06, + "loss": 0.1466, + "step": 11310 + }, + { + "epoch": 4.74, + "learning_rate": 7.530183288054931e-06, + "loss": 0.1535, + "step": 11311 + }, + { + "epoch": 4.74, + "learning_rate": 7.528540438189872e-06, + "loss": 0.1489, + "step": 11312 + }, + { + "epoch": 4.74, + "learning_rate": 7.526897659364727e-06, + "loss": 0.1699, + "step": 11313 + }, + { + "epoch": 4.74, + "learning_rate": 7.525254951626712e-06, + "loss": 0.106, + "step": 11314 + }, + { + "epoch": 4.74, + "learning_rate": 7.523612315023048e-06, + "loss": 0.1492, + "step": 11315 + }, + { + "epoch": 4.74, + "learning_rate": 7.5219697496009525e-06, + "loss": 0.1654, + "step": 11316 + }, + { + "epoch": 4.74, + "learning_rate": 7.520327255407635e-06, + "loss": 0.1983, + "step": 11317 + }, + { + "epoch": 4.74, + "learning_rate": 7.518684832490311e-06, + "loss": 0.167, + "step": 11318 + }, + { + "epoch": 4.74, + "learning_rate": 7.51704248089619e-06, + "loss": 0.1525, + "step": 11319 + }, + { + "epoch": 4.74, + "learning_rate": 7.515400200672478e-06, + "loss": 0.1334, + "step": 11320 + }, + { + "epoch": 4.74, + "learning_rate": 7.513757991866383e-06, + "loss": 0.1492, + "step": 11321 + }, + { + "epoch": 4.74, + "learning_rate": 7.5121158545251085e-06, + "loss": 0.1698, + "step": 11322 + }, + { + "epoch": 4.74, + "learning_rate": 7.51047378869586e-06, + "loss": 0.1589, + "step": 11323 + }, + { + "epoch": 4.74, + "learning_rate": 7.5088317944258305e-06, + "loss": 0.1427, + "step": 11324 + }, + { + "epoch": 4.74, + "learning_rate": 7.50718987176222e-06, + "loss": 0.1504, + "step": 11325 + }, + { + "epoch": 4.74, + "learning_rate": 7.5055480207522225e-06, + "loss": 0.1366, + "step": 11326 + }, + { + "epoch": 4.74, + "learning_rate": 7.503906241443037e-06, + "loss": 0.1122, + "step": 11327 + }, + { + "epoch": 4.74, + "learning_rate": 7.502264533881849e-06, + "loss": 0.1707, + "step": 11328 + }, + { + "epoch": 4.74, + "learning_rate": 7.50062289811585e-06, + "loss": 0.1653, + "step": 11329 + }, + { + "epoch": 4.74, + "learning_rate": 7.49898133419223e-06, + "loss": 0.1242, + "step": 11330 + }, + { + "epoch": 4.74, + "learning_rate": 7.497339842158171e-06, + "loss": 0.1779, + "step": 11331 + }, + { + "epoch": 4.74, + "learning_rate": 7.495698422060857e-06, + "loss": 0.1762, + "step": 11332 + }, + { + "epoch": 4.74, + "learning_rate": 7.494057073947473e-06, + "loss": 0.1771, + "step": 11333 + }, + { + "epoch": 4.75, + "learning_rate": 7.492415797865191e-06, + "loss": 0.1462, + "step": 11334 + }, + { + "epoch": 4.75, + "learning_rate": 7.490774593861192e-06, + "loss": 0.1265, + "step": 11335 + }, + { + "epoch": 4.75, + "learning_rate": 7.489133461982651e-06, + "loss": 0.1422, + "step": 11336 + }, + { + "epoch": 4.75, + "learning_rate": 7.487492402276742e-06, + "loss": 0.1407, + "step": 11337 + }, + { + "epoch": 4.75, + "learning_rate": 7.485851414790632e-06, + "loss": 0.1442, + "step": 11338 + }, + { + "epoch": 4.75, + "learning_rate": 7.4842104995714935e-06, + "loss": 0.2042, + "step": 11339 + }, + { + "epoch": 4.75, + "learning_rate": 7.482569656666493e-06, + "loss": 0.1364, + "step": 11340 + }, + { + "epoch": 4.75, + "learning_rate": 7.480928886122792e-06, + "loss": 0.1347, + "step": 11341 + }, + { + "epoch": 4.75, + "learning_rate": 7.479288187987554e-06, + "loss": 0.1625, + "step": 11342 + }, + { + "epoch": 4.75, + "learning_rate": 7.477647562307943e-06, + "loss": 0.147, + "step": 11343 + }, + { + "epoch": 4.75, + "learning_rate": 7.476007009131112e-06, + "loss": 0.1096, + "step": 11344 + }, + { + "epoch": 4.75, + "learning_rate": 7.47436652850422e-06, + "loss": 0.1525, + "step": 11345 + }, + { + "epoch": 4.75, + "learning_rate": 7.472726120474425e-06, + "loss": 0.1605, + "step": 11346 + }, + { + "epoch": 4.75, + "learning_rate": 7.471085785088871e-06, + "loss": 0.1622, + "step": 11347 + }, + { + "epoch": 4.75, + "learning_rate": 7.469445522394714e-06, + "loss": 0.16, + "step": 11348 + }, + { + "epoch": 4.75, + "learning_rate": 7.467805332439098e-06, + "loss": 0.1244, + "step": 11349 + }, + { + "epoch": 4.75, + "learning_rate": 7.46616521526917e-06, + "loss": 0.1675, + "step": 11350 + }, + { + "epoch": 4.75, + "learning_rate": 7.4645251709320775e-06, + "loss": 0.1349, + "step": 11351 + }, + { + "epoch": 4.75, + "learning_rate": 7.462885199474956e-06, + "loss": 0.164, + "step": 11352 + }, + { + "epoch": 4.75, + "learning_rate": 7.4612453009449485e-06, + "loss": 0.1653, + "step": 11353 + }, + { + "epoch": 4.75, + "learning_rate": 7.459605475389193e-06, + "loss": 0.1166, + "step": 11354 + }, + { + "epoch": 4.75, + "learning_rate": 7.457965722854823e-06, + "loss": 0.1068, + "step": 11355 + }, + { + "epoch": 4.75, + "learning_rate": 7.4563260433889726e-06, + "loss": 0.1797, + "step": 11356 + }, + { + "epoch": 4.75, + "learning_rate": 7.454686437038775e-06, + "loss": 0.1814, + "step": 11357 + }, + { + "epoch": 4.76, + "learning_rate": 7.4530469038513545e-06, + "loss": 0.1696, + "step": 11358 + }, + { + "epoch": 4.76, + "learning_rate": 7.45140744387384e-06, + "loss": 0.1331, + "step": 11359 + }, + { + "epoch": 4.76, + "learning_rate": 7.44976805715336e-06, + "loss": 0.1058, + "step": 11360 + }, + { + "epoch": 4.76, + "learning_rate": 7.448128743737032e-06, + "loss": 0.1342, + "step": 11361 + }, + { + "epoch": 4.76, + "learning_rate": 7.446489503671979e-06, + "loss": 0.1599, + "step": 11362 + }, + { + "epoch": 4.76, + "learning_rate": 7.4448503370053204e-06, + "loss": 0.1291, + "step": 11363 + }, + { + "epoch": 4.76, + "learning_rate": 7.44321124378417e-06, + "loss": 0.1425, + "step": 11364 + }, + { + "epoch": 4.76, + "learning_rate": 7.441572224055644e-06, + "loss": 0.186, + "step": 11365 + }, + { + "epoch": 4.76, + "learning_rate": 7.4399332778668535e-06, + "loss": 0.1956, + "step": 11366 + }, + { + "epoch": 4.76, + "learning_rate": 7.438294405264912e-06, + "loss": 0.2012, + "step": 11367 + }, + { + "epoch": 4.76, + "learning_rate": 7.436655606296921e-06, + "loss": 0.1178, + "step": 11368 + }, + { + "epoch": 4.76, + "learning_rate": 7.4350168810099955e-06, + "loss": 0.1288, + "step": 11369 + }, + { + "epoch": 4.76, + "learning_rate": 7.433378229451229e-06, + "loss": 0.1542, + "step": 11370 + }, + { + "epoch": 4.76, + "learning_rate": 7.431739651667729e-06, + "loss": 0.1506, + "step": 11371 + }, + { + "epoch": 4.76, + "learning_rate": 7.430101147706592e-06, + "loss": 0.2029, + "step": 11372 + }, + { + "epoch": 4.76, + "learning_rate": 7.428462717614917e-06, + "loss": 0.1316, + "step": 11373 + }, + { + "epoch": 4.76, + "learning_rate": 7.426824361439803e-06, + "loss": 0.1284, + "step": 11374 + }, + { + "epoch": 4.76, + "learning_rate": 7.425186079228335e-06, + "loss": 0.1285, + "step": 11375 + }, + { + "epoch": 4.76, + "learning_rate": 7.42354787102761e-06, + "loss": 0.1689, + "step": 11376 + }, + { + "epoch": 4.76, + "learning_rate": 7.421909736884714e-06, + "loss": 0.1667, + "step": 11377 + }, + { + "epoch": 4.76, + "learning_rate": 7.420271676846737e-06, + "loss": 0.1947, + "step": 11378 + }, + { + "epoch": 4.76, + "learning_rate": 7.4186336909607594e-06, + "loss": 0.1408, + "step": 11379 + }, + { + "epoch": 4.76, + "learning_rate": 7.416995779273866e-06, + "loss": 0.1398, + "step": 11380 + }, + { + "epoch": 4.76, + "learning_rate": 7.415357941833139e-06, + "loss": 0.1497, + "step": 11381 + }, + { + "epoch": 4.77, + "learning_rate": 7.413720178685653e-06, + "loss": 0.1695, + "step": 11382 + }, + { + "epoch": 4.77, + "learning_rate": 7.412082489878484e-06, + "loss": 0.1638, + "step": 11383 + }, + { + "epoch": 4.77, + "learning_rate": 7.41044487545871e-06, + "loss": 0.1852, + "step": 11384 + }, + { + "epoch": 4.77, + "learning_rate": 7.408807335473398e-06, + "loss": 0.1481, + "step": 11385 + }, + { + "epoch": 4.77, + "learning_rate": 7.40716986996962e-06, + "loss": 0.1608, + "step": 11386 + }, + { + "epoch": 4.77, + "learning_rate": 7.4055324789944446e-06, + "loss": 0.1231, + "step": 11387 + }, + { + "epoch": 4.77, + "learning_rate": 7.403895162594934e-06, + "loss": 0.1494, + "step": 11388 + }, + { + "epoch": 4.77, + "learning_rate": 7.402257920818154e-06, + "loss": 0.1702, + "step": 11389 + }, + { + "epoch": 4.77, + "learning_rate": 7.400620753711167e-06, + "loss": 0.1696, + "step": 11390 + }, + { + "epoch": 4.77, + "learning_rate": 7.398983661321029e-06, + "loss": 0.1542, + "step": 11391 + }, + { + "epoch": 4.77, + "learning_rate": 7.3973466436947985e-06, + "loss": 0.1124, + "step": 11392 + }, + { + "epoch": 4.77, + "learning_rate": 7.395709700879525e-06, + "loss": 0.1556, + "step": 11393 + }, + { + "epoch": 4.77, + "learning_rate": 7.394072832922268e-06, + "loss": 0.1859, + "step": 11394 + }, + { + "epoch": 4.77, + "learning_rate": 7.3924360398700745e-06, + "loss": 0.1514, + "step": 11395 + }, + { + "epoch": 4.77, + "learning_rate": 7.390799321769993e-06, + "loss": 0.1577, + "step": 11396 + }, + { + "epoch": 4.77, + "learning_rate": 7.389162678669069e-06, + "loss": 0.1659, + "step": 11397 + }, + { + "epoch": 4.77, + "learning_rate": 7.387526110614347e-06, + "loss": 0.1558, + "step": 11398 + }, + { + "epoch": 4.77, + "learning_rate": 7.385889617652868e-06, + "loss": 0.1791, + "step": 11399 + }, + { + "epoch": 4.77, + "learning_rate": 7.384253199831674e-06, + "loss": 0.1584, + "step": 11400 + }, + { + "epoch": 4.77, + "learning_rate": 7.382616857197801e-06, + "loss": 0.1717, + "step": 11401 + }, + { + "epoch": 4.77, + "learning_rate": 7.380980589798281e-06, + "loss": 0.1607, + "step": 11402 + }, + { + "epoch": 4.77, + "learning_rate": 7.379344397680151e-06, + "loss": 0.1706, + "step": 11403 + }, + { + "epoch": 4.77, + "learning_rate": 7.3777082808904435e-06, + "loss": 0.1739, + "step": 11404 + }, + { + "epoch": 4.77, + "learning_rate": 7.376072239476182e-06, + "loss": 0.1403, + "step": 11405 + }, + { + "epoch": 4.78, + "learning_rate": 7.374436273484395e-06, + "loss": 0.1492, + "step": 11406 + }, + { + "epoch": 4.78, + "learning_rate": 7.372800382962107e-06, + "loss": 0.1787, + "step": 11407 + }, + { + "epoch": 4.78, + "learning_rate": 7.371164567956344e-06, + "loss": 0.155, + "step": 11408 + }, + { + "epoch": 4.78, + "learning_rate": 7.36952882851412e-06, + "loss": 0.1437, + "step": 11409 + }, + { + "epoch": 4.78, + "learning_rate": 7.367893164682456e-06, + "loss": 0.1398, + "step": 11410 + }, + { + "epoch": 4.78, + "learning_rate": 7.366257576508369e-06, + "loss": 0.1301, + "step": 11411 + }, + { + "epoch": 4.78, + "learning_rate": 7.364622064038868e-06, + "loss": 0.1722, + "step": 11412 + }, + { + "epoch": 4.78, + "learning_rate": 7.362986627320967e-06, + "loss": 0.1482, + "step": 11413 + }, + { + "epoch": 4.78, + "learning_rate": 7.361351266401682e-06, + "loss": 0.1344, + "step": 11414 + }, + { + "epoch": 4.78, + "learning_rate": 7.359715981328009e-06, + "loss": 0.1519, + "step": 11415 + }, + { + "epoch": 4.78, + "learning_rate": 7.358080772146956e-06, + "loss": 0.1482, + "step": 11416 + }, + { + "epoch": 4.78, + "learning_rate": 7.356445638905526e-06, + "loss": 0.1776, + "step": 11417 + }, + { + "epoch": 4.78, + "learning_rate": 7.354810581650721e-06, + "loss": 0.1388, + "step": 11418 + }, + { + "epoch": 4.78, + "learning_rate": 7.35317560042954e-06, + "loss": 0.1726, + "step": 11419 + }, + { + "epoch": 4.78, + "learning_rate": 7.351540695288977e-06, + "loss": 0.1506, + "step": 11420 + }, + { + "epoch": 4.78, + "learning_rate": 7.349905866276025e-06, + "loss": 0.1139, + "step": 11421 + }, + { + "epoch": 4.78, + "learning_rate": 7.348271113437679e-06, + "loss": 0.1037, + "step": 11422 + }, + { + "epoch": 4.78, + "learning_rate": 7.346636436820926e-06, + "loss": 0.1677, + "step": 11423 + }, + { + "epoch": 4.78, + "learning_rate": 7.345001836472754e-06, + "loss": 0.1564, + "step": 11424 + }, + { + "epoch": 4.78, + "learning_rate": 7.3433673124401495e-06, + "loss": 0.1574, + "step": 11425 + }, + { + "epoch": 4.78, + "learning_rate": 7.341732864770093e-06, + "loss": 0.1255, + "step": 11426 + }, + { + "epoch": 4.78, + "learning_rate": 7.340098493509565e-06, + "loss": 0.1382, + "step": 11427 + }, + { + "epoch": 4.78, + "learning_rate": 7.338464198705548e-06, + "loss": 0.1807, + "step": 11428 + }, + { + "epoch": 4.79, + "learning_rate": 7.336829980405014e-06, + "loss": 0.1531, + "step": 11429 + }, + { + "epoch": 4.79, + "learning_rate": 7.335195838654939e-06, + "loss": 0.1694, + "step": 11430 + }, + { + "epoch": 4.79, + "learning_rate": 7.333561773502297e-06, + "loss": 0.1758, + "step": 11431 + }, + { + "epoch": 4.79, + "learning_rate": 7.3319277849940525e-06, + "loss": 0.1775, + "step": 11432 + }, + { + "epoch": 4.79, + "learning_rate": 7.330293873177179e-06, + "loss": 0.1232, + "step": 11433 + }, + { + "epoch": 4.79, + "learning_rate": 7.328660038098636e-06, + "loss": 0.153, + "step": 11434 + }, + { + "epoch": 4.79, + "learning_rate": 7.327026279805394e-06, + "loss": 0.1797, + "step": 11435 + }, + { + "epoch": 4.79, + "learning_rate": 7.325392598344406e-06, + "loss": 0.1842, + "step": 11436 + }, + { + "epoch": 4.79, + "learning_rate": 7.3237589937626395e-06, + "loss": 0.1748, + "step": 11437 + }, + { + "epoch": 4.79, + "learning_rate": 7.322125466107042e-06, + "loss": 0.1729, + "step": 11438 + }, + { + "epoch": 4.79, + "learning_rate": 7.320492015424572e-06, + "loss": 0.1727, + "step": 11439 + }, + { + "epoch": 4.79, + "learning_rate": 7.318858641762181e-06, + "loss": 0.1761, + "step": 11440 + }, + { + "epoch": 4.79, + "learning_rate": 7.317225345166819e-06, + "loss": 0.1366, + "step": 11441 + }, + { + "epoch": 4.79, + "learning_rate": 7.315592125685434e-06, + "loss": 0.1397, + "step": 11442 + }, + { + "epoch": 4.79, + "learning_rate": 7.31395898336497e-06, + "loss": 0.107, + "step": 11443 + }, + { + "epoch": 4.79, + "learning_rate": 7.312325918252373e-06, + "loss": 0.1468, + "step": 11444 + }, + { + "epoch": 4.79, + "learning_rate": 7.310692930394583e-06, + "loss": 0.1194, + "step": 11445 + }, + { + "epoch": 4.79, + "learning_rate": 7.309060019838536e-06, + "loss": 0.1522, + "step": 11446 + }, + { + "epoch": 4.79, + "learning_rate": 7.307427186631173e-06, + "loss": 0.0958, + "step": 11447 + }, + { + "epoch": 4.79, + "learning_rate": 7.305794430819424e-06, + "loss": 0.1547, + "step": 11448 + }, + { + "epoch": 4.79, + "learning_rate": 7.304161752450227e-06, + "loss": 0.1457, + "step": 11449 + }, + { + "epoch": 4.79, + "learning_rate": 7.302529151570506e-06, + "loss": 0.1221, + "step": 11450 + }, + { + "epoch": 4.79, + "learning_rate": 7.3008966282271885e-06, + "loss": 0.1433, + "step": 11451 + }, + { + "epoch": 4.79, + "learning_rate": 7.299264182467206e-06, + "loss": 0.1421, + "step": 11452 + }, + { + "epoch": 4.8, + "learning_rate": 7.2976318143374755e-06, + "loss": 0.174, + "step": 11453 + }, + { + "epoch": 4.8, + "learning_rate": 7.295999523884921e-06, + "loss": 0.1891, + "step": 11454 + }, + { + "epoch": 4.8, + "learning_rate": 7.294367311156463e-06, + "loss": 0.1325, + "step": 11455 + }, + { + "epoch": 4.8, + "learning_rate": 7.292735176199013e-06, + "loss": 0.1817, + "step": 11456 + }, + { + "epoch": 4.8, + "learning_rate": 7.291103119059488e-06, + "loss": 0.1543, + "step": 11457 + }, + { + "epoch": 4.8, + "learning_rate": 7.289471139784803e-06, + "loss": 0.1808, + "step": 11458 + }, + { + "epoch": 4.8, + "learning_rate": 7.287839238421863e-06, + "loss": 0.1374, + "step": 11459 + }, + { + "epoch": 4.8, + "learning_rate": 7.286207415017579e-06, + "loss": 0.1189, + "step": 11460 + }, + { + "epoch": 4.8, + "learning_rate": 7.284575669618852e-06, + "loss": 0.1296, + "step": 11461 + }, + { + "epoch": 4.8, + "learning_rate": 7.282944002272587e-06, + "loss": 0.1559, + "step": 11462 + }, + { + "epoch": 4.8, + "learning_rate": 7.28131241302569e-06, + "loss": 0.1616, + "step": 11463 + }, + { + "epoch": 4.8, + "learning_rate": 7.279680901925051e-06, + "loss": 0.1827, + "step": 11464 + }, + { + "epoch": 4.8, + "learning_rate": 7.278049469017569e-06, + "loss": 0.1433, + "step": 11465 + }, + { + "epoch": 4.8, + "learning_rate": 7.276418114350145e-06, + "loss": 0.133, + "step": 11466 + }, + { + "epoch": 4.8, + "learning_rate": 7.274786837969663e-06, + "loss": 0.169, + "step": 11467 + }, + { + "epoch": 4.8, + "learning_rate": 7.273155639923013e-06, + "loss": 0.1406, + "step": 11468 + }, + { + "epoch": 4.8, + "learning_rate": 7.271524520257087e-06, + "loss": 0.1203, + "step": 11469 + }, + { + "epoch": 4.8, + "learning_rate": 7.269893479018766e-06, + "loss": 0.1757, + "step": 11470 + }, + { + "epoch": 4.8, + "learning_rate": 7.268262516254934e-06, + "loss": 0.1237, + "step": 11471 + }, + { + "epoch": 4.8, + "learning_rate": 7.2666316320124755e-06, + "loss": 0.1722, + "step": 11472 + }, + { + "epoch": 4.8, + "learning_rate": 7.265000826338261e-06, + "loss": 0.1364, + "step": 11473 + }, + { + "epoch": 4.8, + "learning_rate": 7.263370099279173e-06, + "loss": 0.1386, + "step": 11474 + }, + { + "epoch": 4.8, + "learning_rate": 7.2617394508820815e-06, + "loss": 0.1343, + "step": 11475 + }, + { + "epoch": 4.8, + "learning_rate": 7.26010888119386e-06, + "loss": 0.1714, + "step": 11476 + }, + { + "epoch": 4.81, + "learning_rate": 7.258478390261377e-06, + "loss": 0.156, + "step": 11477 + }, + { + "epoch": 4.81, + "learning_rate": 7.2568479781315e-06, + "loss": 0.1641, + "step": 11478 + }, + { + "epoch": 4.81, + "learning_rate": 7.255217644851096e-06, + "loss": 0.1244, + "step": 11479 + }, + { + "epoch": 4.81, + "learning_rate": 7.253587390467023e-06, + "loss": 0.1458, + "step": 11480 + }, + { + "epoch": 4.81, + "learning_rate": 7.251957215026142e-06, + "loss": 0.1585, + "step": 11481 + }, + { + "epoch": 4.81, + "learning_rate": 7.250327118575318e-06, + "loss": 0.1757, + "step": 11482 + }, + { + "epoch": 4.81, + "learning_rate": 7.2486971011614e-06, + "loss": 0.1324, + "step": 11483 + }, + { + "epoch": 4.81, + "learning_rate": 7.247067162831237e-06, + "loss": 0.1536, + "step": 11484 + }, + { + "epoch": 4.81, + "learning_rate": 7.245437303631687e-06, + "loss": 0.2021, + "step": 11485 + }, + { + "epoch": 4.81, + "learning_rate": 7.243807523609601e-06, + "loss": 0.1557, + "step": 11486 + }, + { + "epoch": 4.81, + "learning_rate": 7.242177822811818e-06, + "loss": 0.1699, + "step": 11487 + }, + { + "epoch": 4.81, + "learning_rate": 7.240548201285186e-06, + "loss": 0.1632, + "step": 11488 + }, + { + "epoch": 4.81, + "learning_rate": 7.238918659076548e-06, + "loss": 0.1633, + "step": 11489 + }, + { + "epoch": 4.81, + "learning_rate": 7.237289196232744e-06, + "loss": 0.1728, + "step": 11490 + }, + { + "epoch": 4.81, + "learning_rate": 7.235659812800611e-06, + "loss": 0.1672, + "step": 11491 + }, + { + "epoch": 4.81, + "learning_rate": 7.234030508826981e-06, + "loss": 0.1507, + "step": 11492 + }, + { + "epoch": 4.81, + "learning_rate": 7.232401284358692e-06, + "loss": 0.174, + "step": 11493 + }, + { + "epoch": 4.81, + "learning_rate": 7.230772139442572e-06, + "loss": 0.0923, + "step": 11494 + }, + { + "epoch": 4.81, + "learning_rate": 7.229143074125447e-06, + "loss": 0.1453, + "step": 11495 + }, + { + "epoch": 4.81, + "learning_rate": 7.2275140884541495e-06, + "loss": 0.1484, + "step": 11496 + }, + { + "epoch": 4.81, + "learning_rate": 7.225885182475496e-06, + "loss": 0.138, + "step": 11497 + }, + { + "epoch": 4.81, + "learning_rate": 7.224256356236312e-06, + "loss": 0.1439, + "step": 11498 + }, + { + "epoch": 4.81, + "learning_rate": 7.222627609783418e-06, + "loss": 0.1385, + "step": 11499 + }, + { + "epoch": 4.81, + "learning_rate": 7.2209989431636265e-06, + "loss": 0.1192, + "step": 11500 + }, + { + "epoch": 4.82, + "learning_rate": 7.219370356423754e-06, + "loss": 0.1667, + "step": 11501 + }, + { + "epoch": 4.82, + "learning_rate": 7.217741849610616e-06, + "loss": 0.1536, + "step": 11502 + }, + { + "epoch": 4.82, + "learning_rate": 7.216113422771016e-06, + "loss": 0.142, + "step": 11503 + }, + { + "epoch": 4.82, + "learning_rate": 7.214485075951771e-06, + "loss": 0.1855, + "step": 11504 + }, + { + "epoch": 4.82, + "learning_rate": 7.212856809199676e-06, + "loss": 0.1364, + "step": 11505 + }, + { + "epoch": 4.82, + "learning_rate": 7.211228622561538e-06, + "loss": 0.1593, + "step": 11506 + }, + { + "epoch": 4.82, + "learning_rate": 7.2096005160841605e-06, + "loss": 0.153, + "step": 11507 + }, + { + "epoch": 4.82, + "learning_rate": 7.207972489814339e-06, + "loss": 0.1313, + "step": 11508 + }, + { + "epoch": 4.82, + "learning_rate": 7.206344543798871e-06, + "loss": 0.1861, + "step": 11509 + }, + { + "epoch": 4.82, + "learning_rate": 7.204716678084551e-06, + "loss": 0.17, + "step": 11510 + }, + { + "epoch": 4.82, + "learning_rate": 7.203088892718168e-06, + "loss": 0.1709, + "step": 11511 + }, + { + "epoch": 4.82, + "learning_rate": 7.201461187746513e-06, + "loss": 0.0981, + "step": 11512 + }, + { + "epoch": 4.82, + "learning_rate": 7.199833563216375e-06, + "loss": 0.1039, + "step": 11513 + }, + { + "epoch": 4.82, + "learning_rate": 7.198206019174534e-06, + "loss": 0.176, + "step": 11514 + }, + { + "epoch": 4.82, + "learning_rate": 7.196578555667775e-06, + "loss": 0.1032, + "step": 11515 + }, + { + "epoch": 4.82, + "learning_rate": 7.194951172742879e-06, + "loss": 0.1818, + "step": 11516 + }, + { + "epoch": 4.82, + "learning_rate": 7.19332387044662e-06, + "loss": 0.1415, + "step": 11517 + }, + { + "epoch": 4.82, + "learning_rate": 7.191696648825778e-06, + "loss": 0.1941, + "step": 11518 + }, + { + "epoch": 4.82, + "learning_rate": 7.190069507927122e-06, + "loss": 0.1256, + "step": 11519 + }, + { + "epoch": 4.82, + "learning_rate": 7.1884424477974265e-06, + "loss": 0.1336, + "step": 11520 + }, + { + "epoch": 4.82, + "learning_rate": 7.186815468483457e-06, + "loss": 0.1809, + "step": 11521 + }, + { + "epoch": 4.82, + "learning_rate": 7.185188570031979e-06, + "loss": 0.1438, + "step": 11522 + }, + { + "epoch": 4.82, + "learning_rate": 7.183561752489761e-06, + "loss": 0.1717, + "step": 11523 + }, + { + "epoch": 4.82, + "learning_rate": 7.181935015903558e-06, + "loss": 0.1621, + "step": 11524 + }, + { + "epoch": 4.83, + "learning_rate": 7.180308360320134e-06, + "loss": 0.1906, + "step": 11525 + }, + { + "epoch": 4.83, + "learning_rate": 7.178681785786245e-06, + "loss": 0.1621, + "step": 11526 + }, + { + "epoch": 4.83, + "learning_rate": 7.177055292348646e-06, + "loss": 0.1375, + "step": 11527 + }, + { + "epoch": 4.83, + "learning_rate": 7.1754288800540836e-06, + "loss": 0.1286, + "step": 11528 + }, + { + "epoch": 4.83, + "learning_rate": 7.173802548949312e-06, + "loss": 0.1323, + "step": 11529 + }, + { + "epoch": 4.83, + "learning_rate": 7.172176299081078e-06, + "loss": 0.1918, + "step": 11530 + }, + { + "epoch": 4.83, + "learning_rate": 7.17055013049613e-06, + "loss": 0.1785, + "step": 11531 + }, + { + "epoch": 4.83, + "learning_rate": 7.168924043241205e-06, + "loss": 0.1526, + "step": 11532 + }, + { + "epoch": 4.83, + "learning_rate": 7.1672980373630455e-06, + "loss": 0.1177, + "step": 11533 + }, + { + "epoch": 4.83, + "learning_rate": 7.165672112908393e-06, + "loss": 0.0849, + "step": 11534 + }, + { + "epoch": 4.83, + "learning_rate": 7.164046269923978e-06, + "loss": 0.2089, + "step": 11535 + }, + { + "epoch": 4.83, + "learning_rate": 7.162420508456536e-06, + "loss": 0.1855, + "step": 11536 + }, + { + "epoch": 4.83, + "learning_rate": 7.160794828552802e-06, + "loss": 0.1502, + "step": 11537 + }, + { + "epoch": 4.83, + "learning_rate": 7.159169230259497e-06, + "loss": 0.1467, + "step": 11538 + }, + { + "epoch": 4.83, + "learning_rate": 7.157543713623353e-06, + "loss": 0.1252, + "step": 11539 + }, + { + "epoch": 4.83, + "learning_rate": 7.155918278691094e-06, + "loss": 0.1651, + "step": 11540 + }, + { + "epoch": 4.83, + "learning_rate": 7.1542929255094386e-06, + "loss": 0.1559, + "step": 11541 + }, + { + "epoch": 4.83, + "learning_rate": 7.1526676541251075e-06, + "loss": 0.1774, + "step": 11542 + }, + { + "epoch": 4.83, + "learning_rate": 7.151042464584821e-06, + "loss": 0.1794, + "step": 11543 + }, + { + "epoch": 4.83, + "learning_rate": 7.149417356935289e-06, + "loss": 0.1324, + "step": 11544 + }, + { + "epoch": 4.83, + "learning_rate": 7.1477923312232235e-06, + "loss": 0.1458, + "step": 11545 + }, + { + "epoch": 4.83, + "learning_rate": 7.14616738749534e-06, + "loss": 0.1399, + "step": 11546 + }, + { + "epoch": 4.83, + "learning_rate": 7.1445425257983395e-06, + "loss": 0.15, + "step": 11547 + }, + { + "epoch": 4.83, + "learning_rate": 7.14291774617893e-06, + "loss": 0.1499, + "step": 11548 + }, + { + "epoch": 4.84, + "learning_rate": 7.141293048683816e-06, + "loss": 0.1482, + "step": 11549 + }, + { + "epoch": 4.84, + "learning_rate": 7.1396684333597e-06, + "loss": 0.1563, + "step": 11550 + }, + { + "epoch": 4.84, + "learning_rate": 7.138043900253275e-06, + "loss": 0.1151, + "step": 11551 + }, + { + "epoch": 4.84, + "learning_rate": 7.136419449411236e-06, + "loss": 0.1529, + "step": 11552 + }, + { + "epoch": 4.84, + "learning_rate": 7.134795080880278e-06, + "loss": 0.1641, + "step": 11553 + }, + { + "epoch": 4.84, + "learning_rate": 7.133170794707096e-06, + "loss": 0.1526, + "step": 11554 + }, + { + "epoch": 4.84, + "learning_rate": 7.131546590938373e-06, + "loss": 0.1107, + "step": 11555 + }, + { + "epoch": 4.84, + "learning_rate": 7.129922469620798e-06, + "loss": 0.1722, + "step": 11556 + }, + { + "epoch": 4.84, + "learning_rate": 7.128298430801057e-06, + "loss": 0.1138, + "step": 11557 + }, + { + "epoch": 4.84, + "learning_rate": 7.1266744745258264e-06, + "loss": 0.1476, + "step": 11558 + }, + { + "epoch": 4.84, + "learning_rate": 7.12505060084179e-06, + "loss": 0.1776, + "step": 11559 + }, + { + "epoch": 4.84, + "learning_rate": 7.123426809795622e-06, + "loss": 0.1628, + "step": 11560 + }, + { + "epoch": 4.84, + "learning_rate": 7.121803101433998e-06, + "loss": 0.1727, + "step": 11561 + }, + { + "epoch": 4.84, + "learning_rate": 7.1201794758035905e-06, + "loss": 0.1536, + "step": 11562 + }, + { + "epoch": 4.84, + "learning_rate": 7.118555932951067e-06, + "loss": 0.1326, + "step": 11563 + }, + { + "epoch": 4.84, + "learning_rate": 7.116932472923098e-06, + "loss": 0.1376, + "step": 11564 + }, + { + "epoch": 4.84, + "learning_rate": 7.115309095766345e-06, + "loss": 0.1419, + "step": 11565 + }, + { + "epoch": 4.84, + "learning_rate": 7.113685801527473e-06, + "loss": 0.111, + "step": 11566 + }, + { + "epoch": 4.84, + "learning_rate": 7.112062590253143e-06, + "loss": 0.1201, + "step": 11567 + }, + { + "epoch": 4.84, + "learning_rate": 7.110439461990008e-06, + "loss": 0.1694, + "step": 11568 + }, + { + "epoch": 4.84, + "learning_rate": 7.108816416784728e-06, + "loss": 0.1499, + "step": 11569 + }, + { + "epoch": 4.84, + "learning_rate": 7.107193454683956e-06, + "loss": 0.1669, + "step": 11570 + }, + { + "epoch": 4.84, + "learning_rate": 7.105570575734339e-06, + "loss": 0.1315, + "step": 11571 + }, + { + "epoch": 4.84, + "learning_rate": 7.103947779982532e-06, + "loss": 0.1531, + "step": 11572 + }, + { + "epoch": 4.85, + "learning_rate": 7.102325067475171e-06, + "loss": 0.13, + "step": 11573 + }, + { + "epoch": 4.85, + "learning_rate": 7.100702438258905e-06, + "loss": 0.1609, + "step": 11574 + }, + { + "epoch": 4.85, + "learning_rate": 7.099079892380376e-06, + "loss": 0.1277, + "step": 11575 + }, + { + "epoch": 4.85, + "learning_rate": 7.097457429886221e-06, + "loss": 0.1312, + "step": 11576 + }, + { + "epoch": 4.85, + "learning_rate": 7.095835050823074e-06, + "loss": 0.1627, + "step": 11577 + }, + { + "epoch": 4.85, + "learning_rate": 7.094212755237576e-06, + "loss": 0.1633, + "step": 11578 + }, + { + "epoch": 4.85, + "learning_rate": 7.092590543176351e-06, + "loss": 0.156, + "step": 11579 + }, + { + "epoch": 4.85, + "learning_rate": 7.09096841468603e-06, + "loss": 0.1543, + "step": 11580 + }, + { + "epoch": 4.85, + "learning_rate": 7.089346369813244e-06, + "loss": 0.1305, + "step": 11581 + }, + { + "epoch": 4.85, + "learning_rate": 7.087724408604611e-06, + "loss": 0.15, + "step": 11582 + }, + { + "epoch": 4.85, + "learning_rate": 7.086102531106755e-06, + "loss": 0.1845, + "step": 11583 + }, + { + "epoch": 4.85, + "learning_rate": 7.084480737366297e-06, + "loss": 0.1297, + "step": 11584 + }, + { + "epoch": 4.85, + "learning_rate": 7.082859027429853e-06, + "loss": 0.1236, + "step": 11585 + }, + { + "epoch": 4.85, + "learning_rate": 7.0812374013440365e-06, + "loss": 0.1535, + "step": 11586 + }, + { + "epoch": 4.85, + "learning_rate": 7.079615859155464e-06, + "loss": 0.1349, + "step": 11587 + }, + { + "epoch": 4.85, + "learning_rate": 7.077994400910739e-06, + "loss": 0.1956, + "step": 11588 + }, + { + "epoch": 4.85, + "learning_rate": 7.0763730266564725e-06, + "loss": 0.1305, + "step": 11589 + }, + { + "epoch": 4.85, + "learning_rate": 7.0747517364392694e-06, + "loss": 0.1581, + "step": 11590 + }, + { + "epoch": 4.85, + "learning_rate": 7.073130530305733e-06, + "loss": 0.149, + "step": 11591 + }, + { + "epoch": 4.85, + "learning_rate": 7.0715094083024604e-06, + "loss": 0.1399, + "step": 11592 + }, + { + "epoch": 4.85, + "learning_rate": 7.069888370476051e-06, + "loss": 0.174, + "step": 11593 + }, + { + "epoch": 4.85, + "learning_rate": 7.068267416873102e-06, + "loss": 0.1423, + "step": 11594 + }, + { + "epoch": 4.85, + "learning_rate": 7.066646547540208e-06, + "loss": 0.148, + "step": 11595 + }, + { + "epoch": 4.85, + "learning_rate": 7.065025762523952e-06, + "loss": 0.174, + "step": 11596 + }, + { + "epoch": 4.86, + "learning_rate": 7.063405061870925e-06, + "loss": 0.1655, + "step": 11597 + }, + { + "epoch": 4.86, + "learning_rate": 7.061784445627715e-06, + "loss": 0.1518, + "step": 11598 + }, + { + "epoch": 4.86, + "learning_rate": 7.060163913840903e-06, + "loss": 0.1504, + "step": 11599 + }, + { + "epoch": 4.86, + "learning_rate": 7.058543466557069e-06, + "loss": 0.1705, + "step": 11600 + }, + { + "epoch": 4.86, + "learning_rate": 7.056923103822795e-06, + "loss": 0.1463, + "step": 11601 + }, + { + "epoch": 4.86, + "learning_rate": 7.0553028256846555e-06, + "loss": 0.1493, + "step": 11602 + }, + { + "epoch": 4.86, + "learning_rate": 7.053682632189221e-06, + "loss": 0.1341, + "step": 11603 + }, + { + "epoch": 4.86, + "learning_rate": 7.052062523383065e-06, + "loss": 0.1391, + "step": 11604 + }, + { + "epoch": 4.86, + "learning_rate": 7.050442499312757e-06, + "loss": 0.1184, + "step": 11605 + }, + { + "epoch": 4.86, + "learning_rate": 7.048822560024862e-06, + "loss": 0.1176, + "step": 11606 + }, + { + "epoch": 4.86, + "learning_rate": 7.047202705565942e-06, + "loss": 0.165, + "step": 11607 + }, + { + "epoch": 4.86, + "learning_rate": 7.045582935982562e-06, + "loss": 0.1549, + "step": 11608 + }, + { + "epoch": 4.86, + "learning_rate": 7.0439632513212775e-06, + "loss": 0.153, + "step": 11609 + }, + { + "epoch": 4.86, + "learning_rate": 7.042343651628646e-06, + "loss": 0.1586, + "step": 11610 + }, + { + "epoch": 4.86, + "learning_rate": 7.040724136951224e-06, + "loss": 0.1513, + "step": 11611 + }, + { + "epoch": 4.86, + "learning_rate": 7.0391047073355575e-06, + "loss": 0.1322, + "step": 11612 + }, + { + "epoch": 4.86, + "learning_rate": 7.0374853628282e-06, + "loss": 0.1278, + "step": 11613 + }, + { + "epoch": 4.86, + "learning_rate": 7.035866103475698e-06, + "loss": 0.1393, + "step": 11614 + }, + { + "epoch": 4.86, + "learning_rate": 7.034246929324593e-06, + "loss": 0.1838, + "step": 11615 + }, + { + "epoch": 4.86, + "learning_rate": 7.032627840421429e-06, + "loss": 0.1764, + "step": 11616 + }, + { + "epoch": 4.86, + "learning_rate": 7.031008836812743e-06, + "loss": 0.1596, + "step": 11617 + }, + { + "epoch": 4.86, + "learning_rate": 7.029389918545078e-06, + "loss": 0.1471, + "step": 11618 + }, + { + "epoch": 4.86, + "learning_rate": 7.0277710856649614e-06, + "loss": 0.1407, + "step": 11619 + }, + { + "epoch": 4.86, + "learning_rate": 7.026152338218925e-06, + "loss": 0.1156, + "step": 11620 + }, + { + "epoch": 4.87, + "learning_rate": 7.024533676253502e-06, + "loss": 0.1556, + "step": 11621 + }, + { + "epoch": 4.87, + "learning_rate": 7.022915099815219e-06, + "loss": 0.1736, + "step": 11622 + }, + { + "epoch": 4.87, + "learning_rate": 7.021296608950599e-06, + "loss": 0.1521, + "step": 11623 + }, + { + "epoch": 4.87, + "learning_rate": 7.019678203706164e-06, + "loss": 0.1597, + "step": 11624 + }, + { + "epoch": 4.87, + "learning_rate": 7.018059884128437e-06, + "loss": 0.1699, + "step": 11625 + }, + { + "epoch": 4.87, + "learning_rate": 7.01644165026393e-06, + "loss": 0.1904, + "step": 11626 + }, + { + "epoch": 4.87, + "learning_rate": 7.014823502159159e-06, + "loss": 0.1579, + "step": 11627 + }, + { + "epoch": 4.87, + "learning_rate": 7.0132054398606405e-06, + "loss": 0.1672, + "step": 11628 + }, + { + "epoch": 4.87, + "learning_rate": 7.011587463414879e-06, + "loss": 0.1292, + "step": 11629 + }, + { + "epoch": 4.87, + "learning_rate": 7.009969572868385e-06, + "loss": 0.1356, + "step": 11630 + }, + { + "epoch": 4.87, + "learning_rate": 7.008351768267661e-06, + "loss": 0.1077, + "step": 11631 + }, + { + "epoch": 4.87, + "learning_rate": 7.006734049659213e-06, + "loss": 0.1405, + "step": 11632 + }, + { + "epoch": 4.87, + "learning_rate": 7.005116417089537e-06, + "loss": 0.1182, + "step": 11633 + }, + { + "epoch": 4.87, + "learning_rate": 7.0034988706051325e-06, + "loss": 0.169, + "step": 11634 + }, + { + "epoch": 4.87, + "learning_rate": 7.001881410252494e-06, + "loss": 0.1308, + "step": 11635 + }, + { + "epoch": 4.87, + "learning_rate": 7.000264036078113e-06, + "loss": 0.1397, + "step": 11636 + }, + { + "epoch": 4.87, + "learning_rate": 6.9986467481284805e-06, + "loss": 0.1342, + "step": 11637 + }, + { + "epoch": 4.87, + "learning_rate": 6.997029546450085e-06, + "loss": 0.112, + "step": 11638 + }, + { + "epoch": 4.87, + "learning_rate": 6.99541243108941e-06, + "loss": 0.1643, + "step": 11639 + }, + { + "epoch": 4.87, + "learning_rate": 6.9937954020929414e-06, + "loss": 0.1515, + "step": 11640 + }, + { + "epoch": 4.87, + "learning_rate": 6.992178459507153e-06, + "loss": 0.1187, + "step": 11641 + }, + { + "epoch": 4.87, + "learning_rate": 6.990561603378524e-06, + "loss": 0.144, + "step": 11642 + }, + { + "epoch": 4.87, + "learning_rate": 6.9889448337535345e-06, + "loss": 0.1462, + "step": 11643 + }, + { + "epoch": 4.88, + "learning_rate": 6.987328150678652e-06, + "loss": 0.1524, + "step": 11644 + }, + { + "epoch": 4.88, + "learning_rate": 6.985711554200347e-06, + "loss": 0.1162, + "step": 11645 + }, + { + "epoch": 4.88, + "learning_rate": 6.984095044365092e-06, + "loss": 0.154, + "step": 11646 + }, + { + "epoch": 4.88, + "learning_rate": 6.982478621219346e-06, + "loss": 0.1642, + "step": 11647 + }, + { + "epoch": 4.88, + "learning_rate": 6.980862284809575e-06, + "loss": 0.1614, + "step": 11648 + }, + { + "epoch": 4.88, + "learning_rate": 6.97924603518224e-06, + "loss": 0.1626, + "step": 11649 + }, + { + "epoch": 4.88, + "learning_rate": 6.9776298723837956e-06, + "loss": 0.1456, + "step": 11650 + }, + { + "epoch": 4.88, + "learning_rate": 6.976013796460698e-06, + "loss": 0.1292, + "step": 11651 + }, + { + "epoch": 4.88, + "learning_rate": 6.974397807459402e-06, + "loss": 0.1422, + "step": 11652 + }, + { + "epoch": 4.88, + "learning_rate": 6.972781905426354e-06, + "loss": 0.133, + "step": 11653 + }, + { + "epoch": 4.88, + "learning_rate": 6.9711660904080055e-06, + "loss": 0.154, + "step": 11654 + }, + { + "epoch": 4.88, + "learning_rate": 6.9695503624508005e-06, + "loss": 0.1427, + "step": 11655 + }, + { + "epoch": 4.88, + "learning_rate": 6.96793472160118e-06, + "loss": 0.1513, + "step": 11656 + }, + { + "epoch": 4.88, + "learning_rate": 6.966319167905584e-06, + "loss": 0.1249, + "step": 11657 + }, + { + "epoch": 4.88, + "learning_rate": 6.964703701410455e-06, + "loss": 0.1561, + "step": 11658 + }, + { + "epoch": 4.88, + "learning_rate": 6.963088322162221e-06, + "loss": 0.139, + "step": 11659 + }, + { + "epoch": 4.88, + "learning_rate": 6.96147303020732e-06, + "loss": 0.194, + "step": 11660 + }, + { + "epoch": 4.88, + "learning_rate": 6.95985782559218e-06, + "loss": 0.1512, + "step": 11661 + }, + { + "epoch": 4.88, + "learning_rate": 6.958242708363231e-06, + "loss": 0.2009, + "step": 11662 + }, + { + "epoch": 4.88, + "learning_rate": 6.956627678566898e-06, + "loss": 0.1388, + "step": 11663 + }, + { + "epoch": 4.88, + "learning_rate": 6.955012736249598e-06, + "loss": 0.1439, + "step": 11664 + }, + { + "epoch": 4.88, + "learning_rate": 6.953397881457754e-06, + "loss": 0.1572, + "step": 11665 + }, + { + "epoch": 4.88, + "learning_rate": 6.951783114237786e-06, + "loss": 0.1202, + "step": 11666 + }, + { + "epoch": 4.88, + "learning_rate": 6.950168434636106e-06, + "loss": 0.1251, + "step": 11667 + }, + { + "epoch": 4.89, + "learning_rate": 6.94855384269913e-06, + "loss": 0.1462, + "step": 11668 + }, + { + "epoch": 4.89, + "learning_rate": 6.946939338473267e-06, + "loss": 0.1682, + "step": 11669 + }, + { + "epoch": 4.89, + "learning_rate": 6.945324922004919e-06, + "loss": 0.1335, + "step": 11670 + }, + { + "epoch": 4.89, + "learning_rate": 6.943710593340498e-06, + "loss": 0.1708, + "step": 11671 + }, + { + "epoch": 4.89, + "learning_rate": 6.9420963525264025e-06, + "loss": 0.1784, + "step": 11672 + }, + { + "epoch": 4.89, + "learning_rate": 6.940482199609036e-06, + "loss": 0.1363, + "step": 11673 + }, + { + "epoch": 4.89, + "learning_rate": 6.938868134634793e-06, + "loss": 0.1453, + "step": 11674 + }, + { + "epoch": 4.89, + "learning_rate": 6.937254157650068e-06, + "loss": 0.1673, + "step": 11675 + }, + { + "epoch": 4.89, + "learning_rate": 6.935640268701256e-06, + "loss": 0.1203, + "step": 11676 + }, + { + "epoch": 4.89, + "learning_rate": 6.934026467834745e-06, + "loss": 0.1493, + "step": 11677 + }, + { + "epoch": 4.89, + "learning_rate": 6.93241275509692e-06, + "loss": 0.1621, + "step": 11678 + }, + { + "epoch": 4.89, + "learning_rate": 6.930799130534172e-06, + "loss": 0.1438, + "step": 11679 + }, + { + "epoch": 4.89, + "learning_rate": 6.929185594192877e-06, + "loss": 0.1704, + "step": 11680 + }, + { + "epoch": 4.89, + "learning_rate": 6.927572146119416e-06, + "loss": 0.1914, + "step": 11681 + }, + { + "epoch": 4.89, + "learning_rate": 6.925958786360172e-06, + "loss": 0.1893, + "step": 11682 + }, + { + "epoch": 4.89, + "learning_rate": 6.92434551496151e-06, + "loss": 0.115, + "step": 11683 + }, + { + "epoch": 4.89, + "learning_rate": 6.922732331969808e-06, + "loss": 0.1342, + "step": 11684 + }, + { + "epoch": 4.89, + "learning_rate": 6.921119237431436e-06, + "loss": 0.1299, + "step": 11685 + }, + { + "epoch": 4.89, + "learning_rate": 6.9195062313927605e-06, + "loss": 0.1314, + "step": 11686 + }, + { + "epoch": 4.89, + "learning_rate": 6.917893313900144e-06, + "loss": 0.1386, + "step": 11687 + }, + { + "epoch": 4.89, + "learning_rate": 6.916280484999946e-06, + "loss": 0.152, + "step": 11688 + }, + { + "epoch": 4.89, + "learning_rate": 6.914667744738531e-06, + "loss": 0.1339, + "step": 11689 + }, + { + "epoch": 4.89, + "learning_rate": 6.9130550931622555e-06, + "loss": 0.1567, + "step": 11690 + }, + { + "epoch": 4.89, + "learning_rate": 6.911442530317469e-06, + "loss": 0.1512, + "step": 11691 + }, + { + "epoch": 4.9, + "learning_rate": 6.909830056250527e-06, + "loss": 0.1504, + "step": 11692 + }, + { + "epoch": 4.9, + "learning_rate": 6.90821767100778e-06, + "loss": 0.192, + "step": 11693 + }, + { + "epoch": 4.9, + "learning_rate": 6.90660537463557e-06, + "loss": 0.1369, + "step": 11694 + }, + { + "epoch": 4.9, + "learning_rate": 6.904993167180243e-06, + "loss": 0.1093, + "step": 11695 + }, + { + "epoch": 4.9, + "learning_rate": 6.9033810486881445e-06, + "loss": 0.159, + "step": 11696 + }, + { + "epoch": 4.9, + "learning_rate": 6.901769019205606e-06, + "loss": 0.1447, + "step": 11697 + }, + { + "epoch": 4.9, + "learning_rate": 6.900157078778969e-06, + "loss": 0.1699, + "step": 11698 + }, + { + "epoch": 4.9, + "learning_rate": 6.898545227454567e-06, + "loss": 0.1377, + "step": 11699 + }, + { + "epoch": 4.9, + "learning_rate": 6.89693346527873e-06, + "loss": 0.1256, + "step": 11700 + }, + { + "epoch": 4.9, + "learning_rate": 6.895321792297785e-06, + "loss": 0.1856, + "step": 11701 + }, + { + "epoch": 4.9, + "learning_rate": 6.893710208558061e-06, + "loss": 0.1304, + "step": 11702 + }, + { + "epoch": 4.9, + "learning_rate": 6.892098714105882e-06, + "loss": 0.1064, + "step": 11703 + }, + { + "epoch": 4.9, + "learning_rate": 6.890487308987567e-06, + "loss": 0.1524, + "step": 11704 + }, + { + "epoch": 4.9, + "learning_rate": 6.888875993249433e-06, + "loss": 0.1516, + "step": 11705 + }, + { + "epoch": 4.9, + "learning_rate": 6.887264766937801e-06, + "loss": 0.1015, + "step": 11706 + }, + { + "epoch": 4.9, + "learning_rate": 6.885653630098978e-06, + "loss": 0.16, + "step": 11707 + }, + { + "epoch": 4.9, + "learning_rate": 6.884042582779285e-06, + "loss": 0.1747, + "step": 11708 + }, + { + "epoch": 4.9, + "learning_rate": 6.882431625025016e-06, + "loss": 0.1781, + "step": 11709 + }, + { + "epoch": 4.9, + "learning_rate": 6.880820756882486e-06, + "loss": 0.1813, + "step": 11710 + }, + { + "epoch": 4.9, + "learning_rate": 6.879209978397993e-06, + "loss": 0.1389, + "step": 11711 + }, + { + "epoch": 4.9, + "learning_rate": 6.87759928961784e-06, + "loss": 0.1352, + "step": 11712 + }, + { + "epoch": 4.9, + "learning_rate": 6.875988690588325e-06, + "loss": 0.2103, + "step": 11713 + }, + { + "epoch": 4.9, + "learning_rate": 6.8743781813557455e-06, + "loss": 0.1271, + "step": 11714 + }, + { + "epoch": 4.9, + "learning_rate": 6.872767761966388e-06, + "loss": 0.1428, + "step": 11715 + }, + { + "epoch": 4.91, + "learning_rate": 6.8711574324665476e-06, + "loss": 0.1631, + "step": 11716 + }, + { + "epoch": 4.91, + "learning_rate": 6.869547192902511e-06, + "loss": 0.1268, + "step": 11717 + }, + { + "epoch": 4.91, + "learning_rate": 6.867937043320562e-06, + "loss": 0.1482, + "step": 11718 + }, + { + "epoch": 4.91, + "learning_rate": 6.866326983766982e-06, + "loss": 0.1294, + "step": 11719 + }, + { + "epoch": 4.91, + "learning_rate": 6.8647170142880545e-06, + "loss": 0.1384, + "step": 11720 + }, + { + "epoch": 4.91, + "learning_rate": 6.863107134930052e-06, + "loss": 0.13, + "step": 11721 + }, + { + "epoch": 4.91, + "learning_rate": 6.861497345739252e-06, + "loss": 0.111, + "step": 11722 + }, + { + "epoch": 4.91, + "learning_rate": 6.859887646761927e-06, + "loss": 0.152, + "step": 11723 + }, + { + "epoch": 4.91, + "learning_rate": 6.858278038044344e-06, + "loss": 0.1708, + "step": 11724 + }, + { + "epoch": 4.91, + "learning_rate": 6.85666851963277e-06, + "loss": 0.1705, + "step": 11725 + }, + { + "epoch": 4.91, + "learning_rate": 6.855059091573472e-06, + "loss": 0.1319, + "step": 11726 + }, + { + "epoch": 4.91, + "learning_rate": 6.853449753912708e-06, + "loss": 0.1653, + "step": 11727 + }, + { + "epoch": 4.91, + "learning_rate": 6.851840506696738e-06, + "loss": 0.1324, + "step": 11728 + }, + { + "epoch": 4.91, + "learning_rate": 6.8502313499718206e-06, + "loss": 0.1431, + "step": 11729 + }, + { + "epoch": 4.91, + "learning_rate": 6.848622283784209e-06, + "loss": 0.1088, + "step": 11730 + }, + { + "epoch": 4.91, + "learning_rate": 6.847013308180155e-06, + "loss": 0.1375, + "step": 11731 + }, + { + "epoch": 4.91, + "learning_rate": 6.845404423205903e-06, + "loss": 0.1367, + "step": 11732 + }, + { + "epoch": 4.91, + "learning_rate": 6.843795628907699e-06, + "loss": 0.1468, + "step": 11733 + }, + { + "epoch": 4.91, + "learning_rate": 6.842186925331793e-06, + "loss": 0.1308, + "step": 11734 + }, + { + "epoch": 4.91, + "learning_rate": 6.8405783125244196e-06, + "loss": 0.1238, + "step": 11735 + }, + { + "epoch": 4.91, + "learning_rate": 6.838969790531817e-06, + "loss": 0.1603, + "step": 11736 + }, + { + "epoch": 4.91, + "learning_rate": 6.837361359400227e-06, + "loss": 0.1417, + "step": 11737 + }, + { + "epoch": 4.91, + "learning_rate": 6.835753019175875e-06, + "loss": 0.1458, + "step": 11738 + }, + { + "epoch": 4.91, + "learning_rate": 6.834144769904996e-06, + "loss": 0.16, + "step": 11739 + }, + { + "epoch": 4.92, + "learning_rate": 6.832536611633816e-06, + "loss": 0.1393, + "step": 11740 + }, + { + "epoch": 4.92, + "learning_rate": 6.830928544408559e-06, + "loss": 0.1548, + "step": 11741 + }, + { + "epoch": 4.92, + "learning_rate": 6.82932056827545e-06, + "loss": 0.1411, + "step": 11742 + }, + { + "epoch": 4.92, + "learning_rate": 6.827712683280707e-06, + "loss": 0.1808, + "step": 11743 + }, + { + "epoch": 4.92, + "learning_rate": 6.8261048894705505e-06, + "loss": 0.181, + "step": 11744 + }, + { + "epoch": 4.92, + "learning_rate": 6.824497186891189e-06, + "loss": 0.1412, + "step": 11745 + }, + { + "epoch": 4.92, + "learning_rate": 6.822889575588838e-06, + "loss": 0.1402, + "step": 11746 + }, + { + "epoch": 4.92, + "learning_rate": 6.8212820556097105e-06, + "loss": 0.1246, + "step": 11747 + }, + { + "epoch": 4.92, + "learning_rate": 6.819674627000006e-06, + "loss": 0.1237, + "step": 11748 + }, + { + "epoch": 4.92, + "learning_rate": 6.818067289805934e-06, + "loss": 0.1781, + "step": 11749 + }, + { + "epoch": 4.92, + "learning_rate": 6.816460044073694e-06, + "loss": 0.144, + "step": 11750 + }, + { + "epoch": 4.92, + "learning_rate": 6.8148528898494845e-06, + "loss": 0.1312, + "step": 11751 + }, + { + "epoch": 4.92, + "learning_rate": 6.813245827179502e-06, + "loss": 0.1339, + "step": 11752 + }, + { + "epoch": 4.92, + "learning_rate": 6.811638856109941e-06, + "loss": 0.1753, + "step": 11753 + }, + { + "epoch": 4.92, + "learning_rate": 6.810031976686994e-06, + "loss": 0.1361, + "step": 11754 + }, + { + "epoch": 4.92, + "learning_rate": 6.808425188956847e-06, + "loss": 0.1599, + "step": 11755 + }, + { + "epoch": 4.92, + "learning_rate": 6.8068184929656815e-06, + "loss": 0.1234, + "step": 11756 + }, + { + "epoch": 4.92, + "learning_rate": 6.8052118887596865e-06, + "loss": 0.1611, + "step": 11757 + }, + { + "epoch": 4.92, + "learning_rate": 6.803605376385042e-06, + "loss": 0.1567, + "step": 11758 + }, + { + "epoch": 4.92, + "learning_rate": 6.801998955887923e-06, + "loss": 0.1952, + "step": 11759 + }, + { + "epoch": 4.92, + "learning_rate": 6.800392627314506e-06, + "loss": 0.1309, + "step": 11760 + }, + { + "epoch": 4.92, + "learning_rate": 6.7987863907109656e-06, + "loss": 0.1267, + "step": 11761 + }, + { + "epoch": 4.92, + "learning_rate": 6.797180246123467e-06, + "loss": 0.2057, + "step": 11762 + }, + { + "epoch": 4.92, + "learning_rate": 6.795574193598182e-06, + "loss": 0.182, + "step": 11763 + }, + { + "epoch": 4.93, + "learning_rate": 6.793968233181274e-06, + "loss": 0.1429, + "step": 11764 + }, + { + "epoch": 4.93, + "learning_rate": 6.792362364918902e-06, + "loss": 0.1341, + "step": 11765 + }, + { + "epoch": 4.93, + "learning_rate": 6.790756588857228e-06, + "loss": 0.1332, + "step": 11766 + }, + { + "epoch": 4.93, + "learning_rate": 6.7891509050424095e-06, + "loss": 0.146, + "step": 11767 + }, + { + "epoch": 4.93, + "learning_rate": 6.787545313520597e-06, + "loss": 0.1507, + "step": 11768 + }, + { + "epoch": 4.93, + "learning_rate": 6.785939814337944e-06, + "loss": 0.1471, + "step": 11769 + }, + { + "epoch": 4.93, + "learning_rate": 6.7843344075406005e-06, + "loss": 0.1545, + "step": 11770 + }, + { + "epoch": 4.93, + "learning_rate": 6.782729093174709e-06, + "loss": 0.1753, + "step": 11771 + }, + { + "epoch": 4.93, + "learning_rate": 6.781123871286416e-06, + "loss": 0.1701, + "step": 11772 + }, + { + "epoch": 4.93, + "learning_rate": 6.77951874192186e-06, + "loss": 0.1585, + "step": 11773 + }, + { + "epoch": 4.93, + "learning_rate": 6.777913705127181e-06, + "loss": 0.1698, + "step": 11774 + }, + { + "epoch": 4.93, + "learning_rate": 6.776308760948514e-06, + "loss": 0.1194, + "step": 11775 + }, + { + "epoch": 4.93, + "learning_rate": 6.774703909431992e-06, + "loss": 0.1379, + "step": 11776 + }, + { + "epoch": 4.93, + "learning_rate": 6.773099150623742e-06, + "loss": 0.1164, + "step": 11777 + }, + { + "epoch": 4.93, + "learning_rate": 6.771494484569895e-06, + "loss": 0.1853, + "step": 11778 + }, + { + "epoch": 4.93, + "learning_rate": 6.7698899113165715e-06, + "loss": 0.1434, + "step": 11779 + }, + { + "epoch": 4.93, + "learning_rate": 6.768285430909896e-06, + "loss": 0.1482, + "step": 11780 + }, + { + "epoch": 4.93, + "learning_rate": 6.76668104339599e-06, + "loss": 0.1466, + "step": 11781 + }, + { + "epoch": 4.93, + "learning_rate": 6.765076748820968e-06, + "loss": 0.1943, + "step": 11782 + }, + { + "epoch": 4.93, + "learning_rate": 6.763472547230943e-06, + "loss": 0.1793, + "step": 11783 + }, + { + "epoch": 4.93, + "learning_rate": 6.761868438672028e-06, + "loss": 0.1735, + "step": 11784 + }, + { + "epoch": 4.93, + "learning_rate": 6.760264423190332e-06, + "loss": 0.1779, + "step": 11785 + }, + { + "epoch": 4.93, + "learning_rate": 6.75866050083196e-06, + "loss": 0.1505, + "step": 11786 + }, + { + "epoch": 4.93, + "learning_rate": 6.757056671643014e-06, + "loss": 0.1513, + "step": 11787 + }, + { + "epoch": 4.94, + "learning_rate": 6.7554529356696e-06, + "loss": 0.1411, + "step": 11788 + }, + { + "epoch": 4.94, + "learning_rate": 6.753849292957808e-06, + "loss": 0.1627, + "step": 11789 + }, + { + "epoch": 4.94, + "learning_rate": 6.752245743553738e-06, + "loss": 0.1364, + "step": 11790 + }, + { + "epoch": 4.94, + "learning_rate": 6.750642287503486e-06, + "loss": 0.1319, + "step": 11791 + }, + { + "epoch": 4.94, + "learning_rate": 6.749038924853134e-06, + "loss": 0.1318, + "step": 11792 + }, + { + "epoch": 4.94, + "learning_rate": 6.747435655648774e-06, + "loss": 0.1795, + "step": 11793 + }, + { + "epoch": 4.94, + "learning_rate": 6.745832479936492e-06, + "loss": 0.1398, + "step": 11794 + }, + { + "epoch": 4.94, + "learning_rate": 6.744229397762365e-06, + "loss": 0.1839, + "step": 11795 + }, + { + "epoch": 4.94, + "learning_rate": 6.742626409172475e-06, + "loss": 0.1448, + "step": 11796 + }, + { + "epoch": 4.94, + "learning_rate": 6.741023514212901e-06, + "loss": 0.1557, + "step": 11797 + }, + { + "epoch": 4.94, + "learning_rate": 6.739420712929712e-06, + "loss": 0.139, + "step": 11798 + }, + { + "epoch": 4.94, + "learning_rate": 6.7378180053689854e-06, + "loss": 0.1092, + "step": 11799 + }, + { + "epoch": 4.94, + "learning_rate": 6.736215391576782e-06, + "loss": 0.1896, + "step": 11800 + }, + { + "epoch": 4.94, + "learning_rate": 6.734612871599169e-06, + "loss": 0.1533, + "step": 11801 + }, + { + "epoch": 4.94, + "learning_rate": 6.733010445482214e-06, + "loss": 0.1236, + "step": 11802 + }, + { + "epoch": 4.94, + "learning_rate": 6.7314081132719735e-06, + "loss": 0.1565, + "step": 11803 + }, + { + "epoch": 4.94, + "learning_rate": 6.729805875014506e-06, + "loss": 0.1567, + "step": 11804 + }, + { + "epoch": 4.94, + "learning_rate": 6.728203730755868e-06, + "loss": 0.1762, + "step": 11805 + }, + { + "epoch": 4.94, + "learning_rate": 6.726601680542109e-06, + "loss": 0.1186, + "step": 11806 + }, + { + "epoch": 4.94, + "learning_rate": 6.724999724419279e-06, + "loss": 0.1542, + "step": 11807 + }, + { + "epoch": 4.94, + "learning_rate": 6.723397862433429e-06, + "loss": 0.1656, + "step": 11808 + }, + { + "epoch": 4.94, + "learning_rate": 6.721796094630597e-06, + "loss": 0.1587, + "step": 11809 + }, + { + "epoch": 4.94, + "learning_rate": 6.720194421056828e-06, + "loss": 0.1193, + "step": 11810 + }, + { + "epoch": 4.94, + "learning_rate": 6.7185928417581615e-06, + "loss": 0.1541, + "step": 11811 + }, + { + "epoch": 4.95, + "learning_rate": 6.71699135678063e-06, + "loss": 0.1437, + "step": 11812 + }, + { + "epoch": 4.95, + "learning_rate": 6.715389966170269e-06, + "loss": 0.1602, + "step": 11813 + }, + { + "epoch": 4.95, + "learning_rate": 6.713788669973108e-06, + "loss": 0.1463, + "step": 11814 + }, + { + "epoch": 4.95, + "learning_rate": 6.7121874682351784e-06, + "loss": 0.1272, + "step": 11815 + }, + { + "epoch": 4.95, + "learning_rate": 6.710586361002499e-06, + "loss": 0.1211, + "step": 11816 + }, + { + "epoch": 4.95, + "learning_rate": 6.708985348321098e-06, + "loss": 0.1686, + "step": 11817 + }, + { + "epoch": 4.95, + "learning_rate": 6.707384430236991e-06, + "loss": 0.1725, + "step": 11818 + }, + { + "epoch": 4.95, + "learning_rate": 6.705783606796199e-06, + "loss": 0.1516, + "step": 11819 + }, + { + "epoch": 4.95, + "learning_rate": 6.704182878044732e-06, + "loss": 0.1333, + "step": 11820 + }, + { + "epoch": 4.95, + "learning_rate": 6.702582244028605e-06, + "loss": 0.1902, + "step": 11821 + }, + { + "epoch": 4.95, + "learning_rate": 6.700981704793828e-06, + "loss": 0.1359, + "step": 11822 + }, + { + "epoch": 4.95, + "learning_rate": 6.6993812603864e-06, + "loss": 0.1323, + "step": 11823 + }, + { + "epoch": 4.95, + "learning_rate": 6.6977809108523295e-06, + "loss": 0.1735, + "step": 11824 + }, + { + "epoch": 4.95, + "learning_rate": 6.696180656237616e-06, + "loss": 0.1605, + "step": 11825 + }, + { + "epoch": 4.95, + "learning_rate": 6.69458049658826e-06, + "loss": 0.1059, + "step": 11826 + }, + { + "epoch": 4.95, + "learning_rate": 6.6929804319502515e-06, + "loss": 0.1623, + "step": 11827 + }, + { + "epoch": 4.95, + "learning_rate": 6.691380462369588e-06, + "loss": 0.1609, + "step": 11828 + }, + { + "epoch": 4.95, + "learning_rate": 6.689780587892256e-06, + "loss": 0.1591, + "step": 11829 + }, + { + "epoch": 4.95, + "learning_rate": 6.688180808564244e-06, + "loss": 0.1501, + "step": 11830 + }, + { + "epoch": 4.95, + "learning_rate": 6.686581124431535e-06, + "loss": 0.1616, + "step": 11831 + }, + { + "epoch": 4.95, + "learning_rate": 6.684981535540114e-06, + "loss": 0.1992, + "step": 11832 + }, + { + "epoch": 4.95, + "learning_rate": 6.6833820419359526e-06, + "loss": 0.1278, + "step": 11833 + }, + { + "epoch": 4.95, + "learning_rate": 6.6817826436650335e-06, + "loss": 0.1619, + "step": 11834 + }, + { + "epoch": 4.95, + "learning_rate": 6.680183340773329e-06, + "loss": 0.1707, + "step": 11835 + }, + { + "epoch": 4.96, + "learning_rate": 6.678584133306808e-06, + "loss": 0.1649, + "step": 11836 + }, + { + "epoch": 4.96, + "learning_rate": 6.6769850213114375e-06, + "loss": 0.1744, + "step": 11837 + }, + { + "epoch": 4.96, + "learning_rate": 6.675386004833185e-06, + "loss": 0.1598, + "step": 11838 + }, + { + "epoch": 4.96, + "learning_rate": 6.673787083918011e-06, + "loss": 0.1677, + "step": 11839 + }, + { + "epoch": 4.96, + "learning_rate": 6.672188258611874e-06, + "loss": 0.1296, + "step": 11840 + }, + { + "epoch": 4.96, + "learning_rate": 6.670589528960734e-06, + "loss": 0.1734, + "step": 11841 + }, + { + "epoch": 4.96, + "learning_rate": 6.668990895010545e-06, + "loss": 0.1718, + "step": 11842 + }, + { + "epoch": 4.96, + "learning_rate": 6.667392356807255e-06, + "loss": 0.1761, + "step": 11843 + }, + { + "epoch": 4.96, + "learning_rate": 6.665793914396818e-06, + "loss": 0.1362, + "step": 11844 + }, + { + "epoch": 4.96, + "learning_rate": 6.664195567825174e-06, + "loss": 0.1253, + "step": 11845 + }, + { + "epoch": 4.96, + "learning_rate": 6.662597317138268e-06, + "loss": 0.1808, + "step": 11846 + }, + { + "epoch": 4.96, + "learning_rate": 6.660999162382038e-06, + "loss": 0.1541, + "step": 11847 + }, + { + "epoch": 4.96, + "learning_rate": 6.6594011036024255e-06, + "loss": 0.1515, + "step": 11848 + }, + { + "epoch": 4.96, + "learning_rate": 6.657803140845367e-06, + "loss": 0.1437, + "step": 11849 + }, + { + "epoch": 4.96, + "learning_rate": 6.656205274156788e-06, + "loss": 0.1805, + "step": 11850 + }, + { + "epoch": 4.96, + "learning_rate": 6.654607503582621e-06, + "loss": 0.1519, + "step": 11851 + }, + { + "epoch": 4.96, + "learning_rate": 6.653009829168795e-06, + "loss": 0.1387, + "step": 11852 + }, + { + "epoch": 4.96, + "learning_rate": 6.651412250961229e-06, + "loss": 0.1141, + "step": 11853 + }, + { + "epoch": 4.96, + "learning_rate": 6.649814769005847e-06, + "loss": 0.1398, + "step": 11854 + }, + { + "epoch": 4.96, + "learning_rate": 6.648217383348565e-06, + "loss": 0.1622, + "step": 11855 + }, + { + "epoch": 4.96, + "learning_rate": 6.646620094035305e-06, + "loss": 0.1627, + "step": 11856 + }, + { + "epoch": 4.96, + "learning_rate": 6.645022901111971e-06, + "loss": 0.1522, + "step": 11857 + }, + { + "epoch": 4.96, + "learning_rate": 6.643425804624476e-06, + "loss": 0.1588, + "step": 11858 + }, + { + "epoch": 4.97, + "learning_rate": 6.641828804618731e-06, + "loss": 0.1015, + "step": 11859 + }, + { + "epoch": 4.97, + "learning_rate": 6.640231901140633e-06, + "loss": 0.1449, + "step": 11860 + }, + { + "epoch": 4.97, + "learning_rate": 6.638635094236089e-06, + "loss": 0.1349, + "step": 11861 + }, + { + "epoch": 4.97, + "learning_rate": 6.637038383950998e-06, + "loss": 0.1586, + "step": 11862 + }, + { + "epoch": 4.97, + "learning_rate": 6.635441770331252e-06, + "loss": 0.1046, + "step": 11863 + }, + { + "epoch": 4.97, + "learning_rate": 6.633845253422748e-06, + "loss": 0.1767, + "step": 11864 + }, + { + "epoch": 4.97, + "learning_rate": 6.632248833271376e-06, + "loss": 0.1598, + "step": 11865 + }, + { + "epoch": 4.97, + "learning_rate": 6.6306525099230214e-06, + "loss": 0.1807, + "step": 11866 + }, + { + "epoch": 4.97, + "learning_rate": 6.629056283423574e-06, + "loss": 0.1622, + "step": 11867 + }, + { + "epoch": 4.97, + "learning_rate": 6.62746015381891e-06, + "loss": 0.2002, + "step": 11868 + }, + { + "epoch": 4.97, + "learning_rate": 6.625864121154908e-06, + "loss": 0.1498, + "step": 11869 + }, + { + "epoch": 4.97, + "learning_rate": 6.624268185477452e-06, + "loss": 0.1557, + "step": 11870 + }, + { + "epoch": 4.97, + "learning_rate": 6.622672346832409e-06, + "loss": 0.1275, + "step": 11871 + }, + { + "epoch": 4.97, + "learning_rate": 6.621076605265653e-06, + "loss": 0.1179, + "step": 11872 + }, + { + "epoch": 4.97, + "learning_rate": 6.619480960823053e-06, + "loss": 0.1549, + "step": 11873 + }, + { + "epoch": 4.97, + "learning_rate": 6.617885413550471e-06, + "loss": 0.1529, + "step": 11874 + }, + { + "epoch": 4.97, + "learning_rate": 6.616289963493771e-06, + "loss": 0.1529, + "step": 11875 + }, + { + "epoch": 4.97, + "learning_rate": 6.614694610698815e-06, + "loss": 0.1736, + "step": 11876 + }, + { + "epoch": 4.97, + "learning_rate": 6.613099355211457e-06, + "loss": 0.186, + "step": 11877 + }, + { + "epoch": 4.97, + "learning_rate": 6.611504197077554e-06, + "loss": 0.1384, + "step": 11878 + }, + { + "epoch": 4.97, + "learning_rate": 6.609909136342956e-06, + "loss": 0.1638, + "step": 11879 + }, + { + "epoch": 4.97, + "learning_rate": 6.608314173053509e-06, + "loss": 0.1224, + "step": 11880 + }, + { + "epoch": 4.97, + "learning_rate": 6.606719307255064e-06, + "loss": 0.1773, + "step": 11881 + }, + { + "epoch": 4.97, + "learning_rate": 6.605124538993462e-06, + "loss": 0.1682, + "step": 11882 + }, + { + "epoch": 4.98, + "learning_rate": 6.60352986831454e-06, + "loss": 0.1524, + "step": 11883 + }, + { + "epoch": 4.98, + "learning_rate": 6.601935295264139e-06, + "loss": 0.138, + "step": 11884 + }, + { + "epoch": 4.98, + "learning_rate": 6.600340819888091e-06, + "loss": 0.1323, + "step": 11885 + }, + { + "epoch": 4.98, + "learning_rate": 6.598746442232233e-06, + "loss": 0.1436, + "step": 11886 + }, + { + "epoch": 4.98, + "learning_rate": 6.597152162342388e-06, + "loss": 0.1648, + "step": 11887 + }, + { + "epoch": 4.98, + "learning_rate": 6.595557980264384e-06, + "loss": 0.1288, + "step": 11888 + }, + { + "epoch": 4.98, + "learning_rate": 6.593963896044047e-06, + "loss": 0.1178, + "step": 11889 + }, + { + "epoch": 4.98, + "learning_rate": 6.592369909727196e-06, + "loss": 0.1467, + "step": 11890 + }, + { + "epoch": 4.98, + "learning_rate": 6.590776021359646e-06, + "loss": 0.1334, + "step": 11891 + }, + { + "epoch": 4.98, + "learning_rate": 6.589182230987213e-06, + "loss": 0.1008, + "step": 11892 + }, + { + "epoch": 4.98, + "learning_rate": 6.587588538655714e-06, + "loss": 0.148, + "step": 11893 + }, + { + "epoch": 4.98, + "learning_rate": 6.5859949444109495e-06, + "loss": 0.1416, + "step": 11894 + }, + { + "epoch": 4.98, + "learning_rate": 6.584401448298732e-06, + "loss": 0.0951, + "step": 11895 + }, + { + "epoch": 4.98, + "learning_rate": 6.582808050364864e-06, + "loss": 0.137, + "step": 11896 + }, + { + "epoch": 4.98, + "learning_rate": 6.581214750655148e-06, + "loss": 0.1516, + "step": 11897 + }, + { + "epoch": 4.98, + "learning_rate": 6.579621549215378e-06, + "loss": 0.1675, + "step": 11898 + }, + { + "epoch": 4.98, + "learning_rate": 6.578028446091351e-06, + "loss": 0.0996, + "step": 11899 + }, + { + "epoch": 4.98, + "learning_rate": 6.5764354413288635e-06, + "loss": 0.1206, + "step": 11900 + }, + { + "epoch": 4.98, + "learning_rate": 6.574842534973698e-06, + "loss": 0.1579, + "step": 11901 + }, + { + "epoch": 4.98, + "learning_rate": 6.573249727071644e-06, + "loss": 0.1271, + "step": 11902 + }, + { + "epoch": 4.98, + "learning_rate": 6.571657017668489e-06, + "loss": 0.1465, + "step": 11903 + }, + { + "epoch": 4.98, + "learning_rate": 6.570064406810008e-06, + "loss": 0.1497, + "step": 11904 + }, + { + "epoch": 4.98, + "learning_rate": 6.568471894541982e-06, + "loss": 0.1262, + "step": 11905 + }, + { + "epoch": 4.98, + "learning_rate": 6.5668794809101884e-06, + "loss": 0.1509, + "step": 11906 + }, + { + "epoch": 4.99, + "learning_rate": 6.565287165960396e-06, + "loss": 0.1165, + "step": 11907 + }, + { + "epoch": 4.99, + "learning_rate": 6.563694949738375e-06, + "loss": 0.1142, + "step": 11908 + }, + { + "epoch": 4.99, + "learning_rate": 6.562102832289896e-06, + "loss": 0.1539, + "step": 11909 + }, + { + "epoch": 4.99, + "learning_rate": 6.560510813660719e-06, + "loss": 0.1892, + "step": 11910 + }, + { + "epoch": 4.99, + "learning_rate": 6.558918893896606e-06, + "loss": 0.1619, + "step": 11911 + }, + { + "epoch": 4.99, + "learning_rate": 6.55732707304332e-06, + "loss": 0.1679, + "step": 11912 + }, + { + "epoch": 4.99, + "learning_rate": 6.5557353511466085e-06, + "loss": 0.1434, + "step": 11913 + }, + { + "epoch": 4.99, + "learning_rate": 6.554143728252229e-06, + "loss": 0.1606, + "step": 11914 + }, + { + "epoch": 4.99, + "learning_rate": 6.552552204405928e-06, + "loss": 0.126, + "step": 11915 + }, + { + "epoch": 4.99, + "learning_rate": 6.550960779653456e-06, + "loss": 0.1764, + "step": 11916 + }, + { + "epoch": 4.99, + "learning_rate": 6.5493694540405575e-06, + "loss": 0.158, + "step": 11917 + }, + { + "epoch": 4.99, + "learning_rate": 6.547778227612969e-06, + "loss": 0.1304, + "step": 11918 + }, + { + "epoch": 4.99, + "learning_rate": 6.546187100416432e-06, + "loss": 0.1372, + "step": 11919 + }, + { + "epoch": 4.99, + "learning_rate": 6.544596072496685e-06, + "loss": 0.1006, + "step": 11920 + }, + { + "epoch": 4.99, + "learning_rate": 6.5430051438994545e-06, + "loss": 0.1562, + "step": 11921 + }, + { + "epoch": 4.99, + "learning_rate": 6.541414314670475e-06, + "loss": 0.1908, + "step": 11922 + }, + { + "epoch": 4.99, + "learning_rate": 6.539823584855472e-06, + "loss": 0.179, + "step": 11923 + }, + { + "epoch": 4.99, + "learning_rate": 6.53823295450017e-06, + "loss": 0.1202, + "step": 11924 + }, + { + "epoch": 4.99, + "learning_rate": 6.536642423650288e-06, + "loss": 0.1464, + "step": 11925 + }, + { + "epoch": 4.99, + "learning_rate": 6.5350519923515464e-06, + "loss": 0.1533, + "step": 11926 + }, + { + "epoch": 4.99, + "learning_rate": 6.533461660649663e-06, + "loss": 0.1619, + "step": 11927 + }, + { + "epoch": 4.99, + "learning_rate": 6.531871428590345e-06, + "loss": 0.1269, + "step": 11928 + }, + { + "epoch": 4.99, + "learning_rate": 6.530281296219305e-06, + "loss": 0.1226, + "step": 11929 + }, + { + "epoch": 4.99, + "learning_rate": 6.528691263582254e-06, + "loss": 0.1563, + "step": 11930 + }, + { + "epoch": 5.0, + "learning_rate": 6.52710133072489e-06, + "loss": 0.1348, + "step": 11931 + }, + { + "epoch": 5.0, + "learning_rate": 6.5255114976929154e-06, + "loss": 0.1795, + "step": 11932 + }, + { + "epoch": 5.0, + "learning_rate": 6.523921764532032e-06, + "loss": 0.1407, + "step": 11933 + }, + { + "epoch": 5.0, + "learning_rate": 6.522332131287931e-06, + "loss": 0.1487, + "step": 11934 + }, + { + "epoch": 5.0, + "learning_rate": 6.520742598006311e-06, + "loss": 0.1426, + "step": 11935 + }, + { + "epoch": 5.0, + "learning_rate": 6.519153164732852e-06, + "loss": 0.1117, + "step": 11936 + }, + { + "epoch": 5.0, + "learning_rate": 6.5175638315132484e-06, + "loss": 0.1438, + "step": 11937 + }, + { + "epoch": 5.0, + "learning_rate": 6.5159745983931846e-06, + "loss": 0.1838, + "step": 11938 + }, + { + "epoch": 5.0, + "learning_rate": 6.514385465418336e-06, + "loss": 0.1861, + "step": 11939 + }, + { + "epoch": 5.0, + "learning_rate": 6.512796432634383e-06, + "loss": 0.2126, + "step": 11940 + }, + { + "epoch": 5.0, + "learning_rate": 6.5112075000870065e-06, + "loss": 0.137, + "step": 11941 + }, + { + "epoch": 5.0, + "learning_rate": 6.5096186678218705e-06, + "loss": 0.1794, + "step": 11942 + }, + { + "epoch": 5.0, + "learning_rate": 6.50802993588465e-06, + "loss": 0.1587, + "step": 11943 + }, + { + "epoch": 5.0, + "learning_rate": 6.506441304321012e-06, + "loss": 0.1663, + "step": 11944 + }, + { + "epoch": 5.0, + "learning_rate": 6.5048527731766155e-06, + "loss": 0.1213, + "step": 11945 + }, + { + "epoch": 5.0, + "learning_rate": 6.503264342497127e-06, + "loss": 0.1663, + "step": 11946 + }, + { + "epoch": 5.0, + "learning_rate": 6.501676012328202e-06, + "loss": 0.1341, + "step": 11947 + }, + { + "epoch": 5.0, + "learning_rate": 6.500087782715494e-06, + "loss": 0.1334, + "step": 11948 + }, + { + "epoch": 5.0, + "learning_rate": 6.498499653704656e-06, + "loss": 0.096, + "step": 11949 + }, + { + "epoch": 5.0, + "learning_rate": 6.496911625341342e-06, + "loss": 0.1041, + "step": 11950 + }, + { + "epoch": 5.0, + "learning_rate": 6.495323697671192e-06, + "loss": 0.1372, + "step": 11951 + }, + { + "epoch": 5.0, + "learning_rate": 6.493735870739852e-06, + "loss": 0.1214, + "step": 11952 + }, + { + "epoch": 5.0, + "learning_rate": 6.492148144592962e-06, + "loss": 0.1367, + "step": 11953 + }, + { + "epoch": 5.0, + "learning_rate": 6.4905605192761656e-06, + "loss": 0.1148, + "step": 11954 + }, + { + "epoch": 5.01, + "learning_rate": 6.488972994835089e-06, + "loss": 0.144, + "step": 11955 + }, + { + "epoch": 5.01, + "learning_rate": 6.4873855713153675e-06, + "loss": 0.1576, + "step": 11956 + }, + { + "epoch": 5.01, + "learning_rate": 6.485798248762634e-06, + "loss": 0.1223, + "step": 11957 + }, + { + "epoch": 5.01, + "learning_rate": 6.484211027222513e-06, + "loss": 0.1417, + "step": 11958 + }, + { + "epoch": 5.01, + "learning_rate": 6.4826239067406216e-06, + "loss": 0.1227, + "step": 11959 + }, + { + "epoch": 5.01, + "learning_rate": 6.481036887362586e-06, + "loss": 0.1256, + "step": 11960 + }, + { + "epoch": 5.01, + "learning_rate": 6.479449969134024e-06, + "loss": 0.1219, + "step": 11961 + }, + { + "epoch": 5.01, + "learning_rate": 6.477863152100545e-06, + "loss": 0.128, + "step": 11962 + }, + { + "epoch": 5.01, + "learning_rate": 6.476276436307767e-06, + "loss": 0.1176, + "step": 11963 + }, + { + "epoch": 5.01, + "learning_rate": 6.474689821801295e-06, + "loss": 0.1425, + "step": 11964 + }, + { + "epoch": 5.01, + "learning_rate": 6.473103308626736e-06, + "loss": 0.0925, + "step": 11965 + }, + { + "epoch": 5.01, + "learning_rate": 6.471516896829692e-06, + "loss": 0.1364, + "step": 11966 + }, + { + "epoch": 5.01, + "learning_rate": 6.469930586455764e-06, + "loss": 0.1362, + "step": 11967 + }, + { + "epoch": 5.01, + "learning_rate": 6.468344377550549e-06, + "loss": 0.1267, + "step": 11968 + }, + { + "epoch": 5.01, + "learning_rate": 6.466758270159641e-06, + "loss": 0.156, + "step": 11969 + }, + { + "epoch": 5.01, + "learning_rate": 6.46517226432863e-06, + "loss": 0.1238, + "step": 11970 + }, + { + "epoch": 5.01, + "learning_rate": 6.4635863601031075e-06, + "loss": 0.1058, + "step": 11971 + }, + { + "epoch": 5.01, + "learning_rate": 6.462000557528656e-06, + "loss": 0.1324, + "step": 11972 + }, + { + "epoch": 5.01, + "learning_rate": 6.460414856650858e-06, + "loss": 0.0875, + "step": 11973 + }, + { + "epoch": 5.01, + "learning_rate": 6.4588292575152955e-06, + "loss": 0.1066, + "step": 11974 + }, + { + "epoch": 5.01, + "learning_rate": 6.4572437601675415e-06, + "loss": 0.1413, + "step": 11975 + }, + { + "epoch": 5.01, + "learning_rate": 6.455658364653174e-06, + "loss": 0.1057, + "step": 11976 + }, + { + "epoch": 5.01, + "learning_rate": 6.4540730710177614e-06, + "loss": 0.1148, + "step": 11977 + }, + { + "epoch": 5.01, + "learning_rate": 6.452487879306871e-06, + "loss": 0.1001, + "step": 11978 + }, + { + "epoch": 5.02, + "learning_rate": 6.450902789566068e-06, + "loss": 0.1373, + "step": 11979 + }, + { + "epoch": 5.02, + "learning_rate": 6.44931780184092e-06, + "loss": 0.107, + "step": 11980 + }, + { + "epoch": 5.02, + "learning_rate": 6.447732916176975e-06, + "loss": 0.1356, + "step": 11981 + }, + { + "epoch": 5.02, + "learning_rate": 6.446148132619798e-06, + "loss": 0.1585, + "step": 11982 + }, + { + "epoch": 5.02, + "learning_rate": 6.444563451214938e-06, + "loss": 0.1395, + "step": 11983 + }, + { + "epoch": 5.02, + "learning_rate": 6.442978872007946e-06, + "loss": 0.1106, + "step": 11984 + }, + { + "epoch": 5.02, + "learning_rate": 6.441394395044372e-06, + "loss": 0.1381, + "step": 11985 + }, + { + "epoch": 5.02, + "learning_rate": 6.439810020369755e-06, + "loss": 0.0928, + "step": 11986 + }, + { + "epoch": 5.02, + "learning_rate": 6.4382257480296405e-06, + "loss": 0.1299, + "step": 11987 + }, + { + "epoch": 5.02, + "learning_rate": 6.436641578069569e-06, + "loss": 0.1425, + "step": 11988 + }, + { + "epoch": 5.02, + "learning_rate": 6.435057510535071e-06, + "loss": 0.1285, + "step": 11989 + }, + { + "epoch": 5.02, + "learning_rate": 6.433473545471681e-06, + "loss": 0.1113, + "step": 11990 + }, + { + "epoch": 5.02, + "learning_rate": 6.431889682924932e-06, + "loss": 0.1415, + "step": 11991 + }, + { + "epoch": 5.02, + "learning_rate": 6.430305922940346e-06, + "loss": 0.1447, + "step": 11992 + }, + { + "epoch": 5.02, + "learning_rate": 6.428722265563446e-06, + "loss": 0.1038, + "step": 11993 + }, + { + "epoch": 5.02, + "learning_rate": 6.42713871083976e-06, + "loss": 0.1033, + "step": 11994 + }, + { + "epoch": 5.02, + "learning_rate": 6.4255552588147975e-06, + "loss": 0.1294, + "step": 11995 + }, + { + "epoch": 5.02, + "learning_rate": 6.423971909534077e-06, + "loss": 0.0996, + "step": 11996 + }, + { + "epoch": 5.02, + "learning_rate": 6.422388663043112e-06, + "loss": 0.1159, + "step": 11997 + }, + { + "epoch": 5.02, + "learning_rate": 6.420805519387412e-06, + "loss": 0.1138, + "step": 11998 + }, + { + "epoch": 5.02, + "learning_rate": 6.419222478612479e-06, + "loss": 0.0632, + "step": 11999 + }, + { + "epoch": 5.02, + "learning_rate": 6.417639540763818e-06, + "loss": 0.1196, + "step": 12000 + }, + { + "epoch": 5.02, + "learning_rate": 6.416056705886932e-06, + "loss": 0.1107, + "step": 12001 + }, + { + "epoch": 5.02, + "learning_rate": 6.414473974027313e-06, + "loss": 0.1428, + "step": 12002 + }, + { + "epoch": 5.03, + "learning_rate": 6.412891345230462e-06, + "loss": 0.1136, + "step": 12003 + }, + { + "epoch": 5.03, + "learning_rate": 6.411308819541863e-06, + "loss": 0.1047, + "step": 12004 + }, + { + "epoch": 5.03, + "learning_rate": 6.409726397007008e-06, + "loss": 0.1103, + "step": 12005 + }, + { + "epoch": 5.03, + "learning_rate": 6.408144077671381e-06, + "loss": 0.1787, + "step": 12006 + }, + { + "epoch": 5.03, + "learning_rate": 6.406561861580465e-06, + "loss": 0.0751, + "step": 12007 + }, + { + "epoch": 5.03, + "learning_rate": 6.404979748779738e-06, + "loss": 0.1224, + "step": 12008 + }, + { + "epoch": 5.03, + "learning_rate": 6.403397739314682e-06, + "loss": 0.1314, + "step": 12009 + }, + { + "epoch": 5.03, + "learning_rate": 6.4018158332307635e-06, + "loss": 0.1615, + "step": 12010 + }, + { + "epoch": 5.03, + "learning_rate": 6.400234030573456e-06, + "loss": 0.1406, + "step": 12011 + }, + { + "epoch": 5.03, + "learning_rate": 6.3986523313882285e-06, + "loss": 0.0791, + "step": 12012 + }, + { + "epoch": 5.03, + "learning_rate": 6.397070735720543e-06, + "loss": 0.1291, + "step": 12013 + }, + { + "epoch": 5.03, + "learning_rate": 6.395489243615861e-06, + "loss": 0.1109, + "step": 12014 + }, + { + "epoch": 5.03, + "learning_rate": 6.393907855119645e-06, + "loss": 0.1138, + "step": 12015 + }, + { + "epoch": 5.03, + "learning_rate": 6.392326570277345e-06, + "loss": 0.136, + "step": 12016 + }, + { + "epoch": 5.03, + "learning_rate": 6.390745389134416e-06, + "loss": 0.1366, + "step": 12017 + }, + { + "epoch": 5.03, + "learning_rate": 6.389164311736311e-06, + "loss": 0.1381, + "step": 12018 + }, + { + "epoch": 5.03, + "learning_rate": 6.387583338128471e-06, + "loss": 0.1289, + "step": 12019 + }, + { + "epoch": 5.03, + "learning_rate": 6.386002468356343e-06, + "loss": 0.1132, + "step": 12020 + }, + { + "epoch": 5.03, + "learning_rate": 6.384421702465369e-06, + "loss": 0.1303, + "step": 12021 + }, + { + "epoch": 5.03, + "learning_rate": 6.382841040500982e-06, + "loss": 0.1251, + "step": 12022 + }, + { + "epoch": 5.03, + "learning_rate": 6.38126048250862e-06, + "loss": 0.1139, + "step": 12023 + }, + { + "epoch": 5.03, + "learning_rate": 6.379680028533714e-06, + "loss": 0.0976, + "step": 12024 + }, + { + "epoch": 5.03, + "learning_rate": 6.378099678621697e-06, + "loss": 0.1316, + "step": 12025 + }, + { + "epoch": 5.03, + "learning_rate": 6.376519432817989e-06, + "loss": 0.1245, + "step": 12026 + }, + { + "epoch": 5.04, + "learning_rate": 6.374939291168012e-06, + "loss": 0.1231, + "step": 12027 + }, + { + "epoch": 5.04, + "learning_rate": 6.373359253717188e-06, + "loss": 0.1713, + "step": 12028 + }, + { + "epoch": 5.04, + "learning_rate": 6.3717793205109354e-06, + "loss": 0.1427, + "step": 12029 + }, + { + "epoch": 5.04, + "learning_rate": 6.370199491594667e-06, + "loss": 0.1467, + "step": 12030 + }, + { + "epoch": 5.04, + "learning_rate": 6.368619767013792e-06, + "loss": 0.1034, + "step": 12031 + }, + { + "epoch": 5.04, + "learning_rate": 6.367040146813721e-06, + "loss": 0.1073, + "step": 12032 + }, + { + "epoch": 5.04, + "learning_rate": 6.365460631039855e-06, + "loss": 0.1462, + "step": 12033 + }, + { + "epoch": 5.04, + "learning_rate": 6.363881219737599e-06, + "loss": 0.1324, + "step": 12034 + }, + { + "epoch": 5.04, + "learning_rate": 6.362301912952351e-06, + "loss": 0.1383, + "step": 12035 + }, + { + "epoch": 5.04, + "learning_rate": 6.360722710729506e-06, + "loss": 0.1589, + "step": 12036 + }, + { + "epoch": 5.04, + "learning_rate": 6.3591436131144565e-06, + "loss": 0.1377, + "step": 12037 + }, + { + "epoch": 5.04, + "learning_rate": 6.357564620152595e-06, + "loss": 0.1345, + "step": 12038 + }, + { + "epoch": 5.04, + "learning_rate": 6.355985731889305e-06, + "loss": 0.1272, + "step": 12039 + }, + { + "epoch": 5.04, + "learning_rate": 6.354406948369971e-06, + "loss": 0.1493, + "step": 12040 + }, + { + "epoch": 5.04, + "learning_rate": 6.352828269639974e-06, + "loss": 0.1311, + "step": 12041 + }, + { + "epoch": 5.04, + "learning_rate": 6.351249695744694e-06, + "loss": 0.1302, + "step": 12042 + }, + { + "epoch": 5.04, + "learning_rate": 6.349671226729502e-06, + "loss": 0.1074, + "step": 12043 + }, + { + "epoch": 5.04, + "learning_rate": 6.348092862639772e-06, + "loss": 0.1288, + "step": 12044 + }, + { + "epoch": 5.04, + "learning_rate": 6.346514603520873e-06, + "loss": 0.1185, + "step": 12045 + }, + { + "epoch": 5.04, + "learning_rate": 6.344936449418168e-06, + "loss": 0.1462, + "step": 12046 + }, + { + "epoch": 5.04, + "learning_rate": 6.343358400377021e-06, + "loss": 0.1116, + "step": 12047 + }, + { + "epoch": 5.04, + "learning_rate": 6.341780456442797e-06, + "loss": 0.1073, + "step": 12048 + }, + { + "epoch": 5.04, + "learning_rate": 6.340202617660842e-06, + "loss": 0.1362, + "step": 12049 + }, + { + "epoch": 5.05, + "learning_rate": 6.338624884076517e-06, + "loss": 0.0908, + "step": 12050 + }, + { + "epoch": 5.05, + "learning_rate": 6.337047255735169e-06, + "loss": 0.0934, + "step": 12051 + }, + { + "epoch": 5.05, + "learning_rate": 6.335469732682149e-06, + "loss": 0.1395, + "step": 12052 + }, + { + "epoch": 5.05, + "learning_rate": 6.3338923149628e-06, + "loss": 0.1196, + "step": 12053 + }, + { + "epoch": 5.05, + "learning_rate": 6.332315002622461e-06, + "loss": 0.1087, + "step": 12054 + }, + { + "epoch": 5.05, + "learning_rate": 6.330737795706474e-06, + "loss": 0.1016, + "step": 12055 + }, + { + "epoch": 5.05, + "learning_rate": 6.329160694260173e-06, + "loss": 0.1073, + "step": 12056 + }, + { + "epoch": 5.05, + "learning_rate": 6.3275836983288895e-06, + "loss": 0.1342, + "step": 12057 + }, + { + "epoch": 5.05, + "learning_rate": 6.326006807957954e-06, + "loss": 0.1287, + "step": 12058 + }, + { + "epoch": 5.05, + "learning_rate": 6.324430023192695e-06, + "loss": 0.1116, + "step": 12059 + }, + { + "epoch": 5.05, + "learning_rate": 6.32285334407843e-06, + "loss": 0.11, + "step": 12060 + }, + { + "epoch": 5.05, + "learning_rate": 6.3212767706604826e-06, + "loss": 0.1296, + "step": 12061 + }, + { + "epoch": 5.05, + "learning_rate": 6.319700302984173e-06, + "loss": 0.1233, + "step": 12062 + }, + { + "epoch": 5.05, + "learning_rate": 6.31812394109481e-06, + "loss": 0.0853, + "step": 12063 + }, + { + "epoch": 5.05, + "learning_rate": 6.316547685037707e-06, + "loss": 0.0856, + "step": 12064 + }, + { + "epoch": 5.05, + "learning_rate": 6.314971534858174e-06, + "loss": 0.1416, + "step": 12065 + }, + { + "epoch": 5.05, + "learning_rate": 6.313395490601513e-06, + "loss": 0.172, + "step": 12066 + }, + { + "epoch": 5.05, + "learning_rate": 6.311819552313026e-06, + "loss": 0.1376, + "step": 12067 + }, + { + "epoch": 5.05, + "learning_rate": 6.310243720038014e-06, + "loss": 0.1347, + "step": 12068 + }, + { + "epoch": 5.05, + "learning_rate": 6.308667993821773e-06, + "loss": 0.1733, + "step": 12069 + }, + { + "epoch": 5.05, + "learning_rate": 6.3070923737095946e-06, + "loss": 0.155, + "step": 12070 + }, + { + "epoch": 5.05, + "learning_rate": 6.305516859746772e-06, + "loss": 0.0885, + "step": 12071 + }, + { + "epoch": 5.05, + "learning_rate": 6.303941451978584e-06, + "loss": 0.1177, + "step": 12072 + }, + { + "epoch": 5.05, + "learning_rate": 6.302366150450323e-06, + "loss": 0.1401, + "step": 12073 + }, + { + "epoch": 5.06, + "learning_rate": 6.300790955207262e-06, + "loss": 0.1523, + "step": 12074 + }, + { + "epoch": 5.06, + "learning_rate": 6.299215866294683e-06, + "loss": 0.128, + "step": 12075 + }, + { + "epoch": 5.06, + "learning_rate": 6.297640883757863e-06, + "loss": 0.1039, + "step": 12076 + }, + { + "epoch": 5.06, + "learning_rate": 6.296066007642067e-06, + "loss": 0.1119, + "step": 12077 + }, + { + "epoch": 5.06, + "learning_rate": 6.294491237992566e-06, + "loss": 0.1343, + "step": 12078 + }, + { + "epoch": 5.06, + "learning_rate": 6.292916574854628e-06, + "loss": 0.1404, + "step": 12079 + }, + { + "epoch": 5.06, + "learning_rate": 6.291342018273514e-06, + "loss": 0.1355, + "step": 12080 + }, + { + "epoch": 5.06, + "learning_rate": 6.2897675682944824e-06, + "loss": 0.0829, + "step": 12081 + }, + { + "epoch": 5.06, + "learning_rate": 6.288193224962789e-06, + "loss": 0.0669, + "step": 12082 + }, + { + "epoch": 5.06, + "learning_rate": 6.2866189883236885e-06, + "loss": 0.1222, + "step": 12083 + }, + { + "epoch": 5.06, + "learning_rate": 6.285044858422429e-06, + "loss": 0.1228, + "step": 12084 + }, + { + "epoch": 5.06, + "learning_rate": 6.283470835304257e-06, + "loss": 0.116, + "step": 12085 + }, + { + "epoch": 5.06, + "learning_rate": 6.281896919014422e-06, + "loss": 0.12, + "step": 12086 + }, + { + "epoch": 5.06, + "learning_rate": 6.280323109598157e-06, + "loss": 0.138, + "step": 12087 + }, + { + "epoch": 5.06, + "learning_rate": 6.278749407100704e-06, + "loss": 0.1307, + "step": 12088 + }, + { + "epoch": 5.06, + "learning_rate": 6.2771758115673e-06, + "loss": 0.0881, + "step": 12089 + }, + { + "epoch": 5.06, + "learning_rate": 6.2756023230431705e-06, + "loss": 0.1232, + "step": 12090 + }, + { + "epoch": 5.06, + "learning_rate": 6.2740289415735474e-06, + "loss": 0.1266, + "step": 12091 + }, + { + "epoch": 5.06, + "learning_rate": 6.272455667203659e-06, + "loss": 0.1358, + "step": 12092 + }, + { + "epoch": 5.06, + "learning_rate": 6.270882499978723e-06, + "loss": 0.1208, + "step": 12093 + }, + { + "epoch": 5.06, + "learning_rate": 6.269309439943963e-06, + "loss": 0.1235, + "step": 12094 + }, + { + "epoch": 5.06, + "learning_rate": 6.2677364871445886e-06, + "loss": 0.1502, + "step": 12095 + }, + { + "epoch": 5.06, + "learning_rate": 6.2661636416258175e-06, + "loss": 0.1322, + "step": 12096 + }, + { + "epoch": 5.06, + "learning_rate": 6.264590903432861e-06, + "loss": 0.0914, + "step": 12097 + }, + { + "epoch": 5.07, + "learning_rate": 6.263018272610921e-06, + "loss": 0.123, + "step": 12098 + }, + { + "epoch": 5.07, + "learning_rate": 6.261445749205206e-06, + "loss": 0.1693, + "step": 12099 + }, + { + "epoch": 5.07, + "learning_rate": 6.259873333260917e-06, + "loss": 0.1072, + "step": 12100 + }, + { + "epoch": 5.07, + "learning_rate": 6.258301024823249e-06, + "loss": 0.0645, + "step": 12101 + }, + { + "epoch": 5.07, + "learning_rate": 6.256728823937397e-06, + "loss": 0.1359, + "step": 12102 + }, + { + "epoch": 5.07, + "learning_rate": 6.255156730648555e-06, + "loss": 0.1382, + "step": 12103 + }, + { + "epoch": 5.07, + "learning_rate": 6.253584745001908e-06, + "loss": 0.1039, + "step": 12104 + }, + { + "epoch": 5.07, + "learning_rate": 6.252012867042643e-06, + "loss": 0.1308, + "step": 12105 + }, + { + "epoch": 5.07, + "learning_rate": 6.250441096815945e-06, + "loss": 0.1262, + "step": 12106 + }, + { + "epoch": 5.07, + "learning_rate": 6.248869434366987e-06, + "loss": 0.1414, + "step": 12107 + }, + { + "epoch": 5.07, + "learning_rate": 6.247297879740948e-06, + "loss": 0.1307, + "step": 12108 + }, + { + "epoch": 5.07, + "learning_rate": 6.245726432983003e-06, + "loss": 0.1036, + "step": 12109 + }, + { + "epoch": 5.07, + "learning_rate": 6.244155094138322e-06, + "loss": 0.152, + "step": 12110 + }, + { + "epoch": 5.07, + "learning_rate": 6.2425838632520674e-06, + "loss": 0.1469, + "step": 12111 + }, + { + "epoch": 5.07, + "learning_rate": 6.2410127403694056e-06, + "loss": 0.1288, + "step": 12112 + }, + { + "epoch": 5.07, + "learning_rate": 6.239441725535498e-06, + "loss": 0.0963, + "step": 12113 + }, + { + "epoch": 5.07, + "learning_rate": 6.237870818795499e-06, + "loss": 0.0882, + "step": 12114 + }, + { + "epoch": 5.07, + "learning_rate": 6.2363000201945656e-06, + "loss": 0.1074, + "step": 12115 + }, + { + "epoch": 5.07, + "learning_rate": 6.234729329777853e-06, + "loss": 0.1028, + "step": 12116 + }, + { + "epoch": 5.07, + "learning_rate": 6.233158747590502e-06, + "loss": 0.1171, + "step": 12117 + }, + { + "epoch": 5.07, + "learning_rate": 6.231588273677658e-06, + "loss": 0.1217, + "step": 12118 + }, + { + "epoch": 5.07, + "learning_rate": 6.230017908084465e-06, + "loss": 0.1087, + "step": 12119 + }, + { + "epoch": 5.07, + "learning_rate": 6.228447650856062e-06, + "loss": 0.1036, + "step": 12120 + }, + { + "epoch": 5.07, + "learning_rate": 6.226877502037585e-06, + "loss": 0.0812, + "step": 12121 + }, + { + "epoch": 5.08, + "learning_rate": 6.225307461674166e-06, + "loss": 0.1108, + "step": 12122 + }, + { + "epoch": 5.08, + "learning_rate": 6.2237375298109336e-06, + "loss": 0.1215, + "step": 12123 + }, + { + "epoch": 5.08, + "learning_rate": 6.222167706493016e-06, + "loss": 0.1526, + "step": 12124 + }, + { + "epoch": 5.08, + "learning_rate": 6.220597991765536e-06, + "loss": 0.1058, + "step": 12125 + }, + { + "epoch": 5.08, + "learning_rate": 6.2190283856736115e-06, + "loss": 0.1068, + "step": 12126 + }, + { + "epoch": 5.08, + "learning_rate": 6.217458888262362e-06, + "loss": 0.1346, + "step": 12127 + }, + { + "epoch": 5.08, + "learning_rate": 6.215889499576898e-06, + "loss": 0.1367, + "step": 12128 + }, + { + "epoch": 5.08, + "learning_rate": 6.214320219662333e-06, + "loss": 0.1279, + "step": 12129 + }, + { + "epoch": 5.08, + "learning_rate": 6.212751048563776e-06, + "loss": 0.1141, + "step": 12130 + }, + { + "epoch": 5.08, + "learning_rate": 6.2111819863263275e-06, + "loss": 0.1321, + "step": 12131 + }, + { + "epoch": 5.08, + "learning_rate": 6.209613032995091e-06, + "loss": 0.1796, + "step": 12132 + }, + { + "epoch": 5.08, + "learning_rate": 6.208044188615166e-06, + "loss": 0.1399, + "step": 12133 + }, + { + "epoch": 5.08, + "learning_rate": 6.206475453231644e-06, + "loss": 0.1253, + "step": 12134 + }, + { + "epoch": 5.08, + "learning_rate": 6.204906826889618e-06, + "loss": 0.1111, + "step": 12135 + }, + { + "epoch": 5.08, + "learning_rate": 6.203338309634179e-06, + "loss": 0.1165, + "step": 12136 + }, + { + "epoch": 5.08, + "learning_rate": 6.201769901510412e-06, + "loss": 0.1071, + "step": 12137 + }, + { + "epoch": 5.08, + "learning_rate": 6.200201602563397e-06, + "loss": 0.1172, + "step": 12138 + }, + { + "epoch": 5.08, + "learning_rate": 6.1986334128382195e-06, + "loss": 0.1291, + "step": 12139 + }, + { + "epoch": 5.08, + "learning_rate": 6.197065332379947e-06, + "loss": 0.117, + "step": 12140 + }, + { + "epoch": 5.08, + "learning_rate": 6.195497361233658e-06, + "loss": 0.1428, + "step": 12141 + }, + { + "epoch": 5.08, + "learning_rate": 6.19392949944442e-06, + "loss": 0.1521, + "step": 12142 + }, + { + "epoch": 5.08, + "learning_rate": 6.192361747057301e-06, + "loss": 0.128, + "step": 12143 + }, + { + "epoch": 5.08, + "learning_rate": 6.190794104117368e-06, + "loss": 0.1004, + "step": 12144 + }, + { + "epoch": 5.08, + "learning_rate": 6.189226570669675e-06, + "loss": 0.1502, + "step": 12145 + }, + { + "epoch": 5.09, + "learning_rate": 6.187659146759282e-06, + "loss": 0.1702, + "step": 12146 + }, + { + "epoch": 5.09, + "learning_rate": 6.186091832431247e-06, + "loss": 0.1418, + "step": 12147 + }, + { + "epoch": 5.09, + "learning_rate": 6.1845246277306145e-06, + "loss": 0.1337, + "step": 12148 + }, + { + "epoch": 5.09, + "learning_rate": 6.182957532702437e-06, + "loss": 0.1371, + "step": 12149 + }, + { + "epoch": 5.09, + "learning_rate": 6.181390547391758e-06, + "loss": 0.1352, + "step": 12150 + }, + { + "epoch": 5.09, + "learning_rate": 6.179823671843621e-06, + "loss": 0.138, + "step": 12151 + }, + { + "epoch": 5.09, + "learning_rate": 6.178256906103061e-06, + "loss": 0.1226, + "step": 12152 + }, + { + "epoch": 5.09, + "learning_rate": 6.176690250215114e-06, + "loss": 0.1264, + "step": 12153 + }, + { + "epoch": 5.09, + "learning_rate": 6.175123704224814e-06, + "loss": 0.1309, + "step": 12154 + }, + { + "epoch": 5.09, + "learning_rate": 6.173557268177189e-06, + "loss": 0.135, + "step": 12155 + }, + { + "epoch": 5.09, + "learning_rate": 6.171990942117265e-06, + "loss": 0.1154, + "step": 12156 + }, + { + "epoch": 5.09, + "learning_rate": 6.170424726090065e-06, + "loss": 0.1521, + "step": 12157 + }, + { + "epoch": 5.09, + "learning_rate": 6.168858620140606e-06, + "loss": 0.1361, + "step": 12158 + }, + { + "epoch": 5.09, + "learning_rate": 6.167292624313908e-06, + "loss": 0.1125, + "step": 12159 + }, + { + "epoch": 5.09, + "learning_rate": 6.165726738654982e-06, + "loss": 0.0856, + "step": 12160 + }, + { + "epoch": 5.09, + "learning_rate": 6.16416096320884e-06, + "loss": 0.1175, + "step": 12161 + }, + { + "epoch": 5.09, + "learning_rate": 6.162595298020487e-06, + "loss": 0.0999, + "step": 12162 + }, + { + "epoch": 5.09, + "learning_rate": 6.1610297431349255e-06, + "loss": 0.1059, + "step": 12163 + }, + { + "epoch": 5.09, + "learning_rate": 6.1594642985971565e-06, + "loss": 0.1038, + "step": 12164 + }, + { + "epoch": 5.09, + "learning_rate": 6.15789896445218e-06, + "loss": 0.1068, + "step": 12165 + }, + { + "epoch": 5.09, + "learning_rate": 6.1563337407449865e-06, + "loss": 0.1114, + "step": 12166 + }, + { + "epoch": 5.09, + "learning_rate": 6.15476862752057e-06, + "loss": 0.1109, + "step": 12167 + }, + { + "epoch": 5.09, + "learning_rate": 6.153203624823918e-06, + "loss": 0.1219, + "step": 12168 + }, + { + "epoch": 5.09, + "learning_rate": 6.151638732700012e-06, + "loss": 0.1144, + "step": 12169 + }, + { + "epoch": 5.1, + "learning_rate": 6.150073951193836e-06, + "loss": 0.1246, + "step": 12170 + }, + { + "epoch": 5.1, + "learning_rate": 6.1485092803503695e-06, + "loss": 0.1424, + "step": 12171 + }, + { + "epoch": 5.1, + "learning_rate": 6.146944720214585e-06, + "loss": 0.1515, + "step": 12172 + }, + { + "epoch": 5.1, + "learning_rate": 6.145380270831455e-06, + "loss": 0.1316, + "step": 12173 + }, + { + "epoch": 5.1, + "learning_rate": 6.143815932245951e-06, + "loss": 0.1069, + "step": 12174 + }, + { + "epoch": 5.1, + "learning_rate": 6.142251704503033e-06, + "loss": 0.1006, + "step": 12175 + }, + { + "epoch": 5.1, + "learning_rate": 6.140687587647667e-06, + "loss": 0.0906, + "step": 12176 + }, + { + "epoch": 5.1, + "learning_rate": 6.139123581724813e-06, + "loss": 0.0751, + "step": 12177 + }, + { + "epoch": 5.1, + "learning_rate": 6.1375596867794245e-06, + "loss": 0.1319, + "step": 12178 + }, + { + "epoch": 5.1, + "learning_rate": 6.135995902856455e-06, + "loss": 0.0637, + "step": 12179 + }, + { + "epoch": 5.1, + "learning_rate": 6.134432230000854e-06, + "loss": 0.1416, + "step": 12180 + }, + { + "epoch": 5.1, + "learning_rate": 6.132868668257571e-06, + "loss": 0.1236, + "step": 12181 + }, + { + "epoch": 5.1, + "learning_rate": 6.131305217671544e-06, + "loss": 0.0791, + "step": 12182 + }, + { + "epoch": 5.1, + "learning_rate": 6.129741878287716e-06, + "loss": 0.106, + "step": 12183 + }, + { + "epoch": 5.1, + "learning_rate": 6.128178650151026e-06, + "loss": 0.1149, + "step": 12184 + }, + { + "epoch": 5.1, + "learning_rate": 6.126615533306405e-06, + "loss": 0.131, + "step": 12185 + }, + { + "epoch": 5.1, + "learning_rate": 6.12505252779878e-06, + "loss": 0.1268, + "step": 12186 + }, + { + "epoch": 5.1, + "learning_rate": 6.123489633673082e-06, + "loss": 0.0994, + "step": 12187 + }, + { + "epoch": 5.1, + "learning_rate": 6.121926850974236e-06, + "loss": 0.1267, + "step": 12188 + }, + { + "epoch": 5.1, + "learning_rate": 6.120364179747161e-06, + "loss": 0.1149, + "step": 12189 + }, + { + "epoch": 5.1, + "learning_rate": 6.1188016200367715e-06, + "loss": 0.1255, + "step": 12190 + }, + { + "epoch": 5.1, + "learning_rate": 6.117239171887987e-06, + "loss": 0.1498, + "step": 12191 + }, + { + "epoch": 5.1, + "learning_rate": 6.11567683534572e-06, + "loss": 0.106, + "step": 12192 + }, + { + "epoch": 5.1, + "learning_rate": 6.1141146104548735e-06, + "loss": 0.1239, + "step": 12193 + }, + { + "epoch": 5.11, + "learning_rate": 6.112552497260353e-06, + "loss": 0.0943, + "step": 12194 + }, + { + "epoch": 5.11, + "learning_rate": 6.110990495807064e-06, + "loss": 0.1374, + "step": 12195 + }, + { + "epoch": 5.11, + "learning_rate": 6.1094286061399e-06, + "loss": 0.0907, + "step": 12196 + }, + { + "epoch": 5.11, + "learning_rate": 6.107866828303759e-06, + "loss": 0.1415, + "step": 12197 + }, + { + "epoch": 5.11, + "learning_rate": 6.106305162343533e-06, + "loss": 0.0777, + "step": 12198 + }, + { + "epoch": 5.11, + "learning_rate": 6.104743608304109e-06, + "loss": 0.1269, + "step": 12199 + }, + { + "epoch": 5.11, + "learning_rate": 6.103182166230373e-06, + "loss": 0.1295, + "step": 12200 + }, + { + "epoch": 5.11, + "learning_rate": 6.101620836167208e-06, + "loss": 0.1536, + "step": 12201 + }, + { + "epoch": 5.11, + "learning_rate": 6.100059618159493e-06, + "loss": 0.1431, + "step": 12202 + }, + { + "epoch": 5.11, + "learning_rate": 6.098498512252102e-06, + "loss": 0.0914, + "step": 12203 + }, + { + "epoch": 5.11, + "learning_rate": 6.096937518489912e-06, + "loss": 0.1229, + "step": 12204 + }, + { + "epoch": 5.11, + "learning_rate": 6.095376636917787e-06, + "loss": 0.124, + "step": 12205 + }, + { + "epoch": 5.11, + "learning_rate": 6.093815867580597e-06, + "loss": 0.1116, + "step": 12206 + }, + { + "epoch": 5.11, + "learning_rate": 6.0922552105232055e-06, + "loss": 0.1509, + "step": 12207 + }, + { + "epoch": 5.11, + "learning_rate": 6.0906946657904685e-06, + "loss": 0.1288, + "step": 12208 + }, + { + "epoch": 5.11, + "learning_rate": 6.089134233427245e-06, + "loss": 0.1412, + "step": 12209 + }, + { + "epoch": 5.11, + "learning_rate": 6.087573913478386e-06, + "loss": 0.1191, + "step": 12210 + }, + { + "epoch": 5.11, + "learning_rate": 6.086013705988742e-06, + "loss": 0.1093, + "step": 12211 + }, + { + "epoch": 5.11, + "learning_rate": 6.084453611003164e-06, + "loss": 0.0971, + "step": 12212 + }, + { + "epoch": 5.11, + "learning_rate": 6.08289362856649e-06, + "loss": 0.0934, + "step": 12213 + }, + { + "epoch": 5.11, + "learning_rate": 6.081333758723562e-06, + "loss": 0.1281, + "step": 12214 + }, + { + "epoch": 5.11, + "learning_rate": 6.079774001519221e-06, + "loss": 0.1253, + "step": 12215 + }, + { + "epoch": 5.11, + "learning_rate": 6.078214356998294e-06, + "loss": 0.1323, + "step": 12216 + }, + { + "epoch": 5.11, + "learning_rate": 6.076654825205617e-06, + "loss": 0.154, + "step": 12217 + }, + { + "epoch": 5.12, + "learning_rate": 6.075095406186015e-06, + "loss": 0.0858, + "step": 12218 + }, + { + "epoch": 5.12, + "learning_rate": 6.073536099984314e-06, + "loss": 0.1258, + "step": 12219 + }, + { + "epoch": 5.12, + "learning_rate": 6.071976906645332e-06, + "loss": 0.1146, + "step": 12220 + }, + { + "epoch": 5.12, + "learning_rate": 6.070417826213888e-06, + "loss": 0.1333, + "step": 12221 + }, + { + "epoch": 5.12, + "learning_rate": 6.068858858734799e-06, + "loss": 0.096, + "step": 12222 + }, + { + "epoch": 5.12, + "learning_rate": 6.06730000425287e-06, + "loss": 0.1071, + "step": 12223 + }, + { + "epoch": 5.12, + "learning_rate": 6.065741262812914e-06, + "loss": 0.1269, + "step": 12224 + }, + { + "epoch": 5.12, + "learning_rate": 6.064182634459735e-06, + "loss": 0.1314, + "step": 12225 + }, + { + "epoch": 5.12, + "learning_rate": 6.062624119238133e-06, + "loss": 0.1292, + "step": 12226 + }, + { + "epoch": 5.12, + "learning_rate": 6.061065717192907e-06, + "loss": 0.087, + "step": 12227 + }, + { + "epoch": 5.12, + "learning_rate": 6.059507428368852e-06, + "loss": 0.1242, + "step": 12228 + }, + { + "epoch": 5.12, + "learning_rate": 6.057949252810762e-06, + "loss": 0.1421, + "step": 12229 + }, + { + "epoch": 5.12, + "learning_rate": 6.056391190563419e-06, + "loss": 0.1274, + "step": 12230 + }, + { + "epoch": 5.12, + "learning_rate": 6.054833241671611e-06, + "loss": 0.1439, + "step": 12231 + }, + { + "epoch": 5.12, + "learning_rate": 6.053275406180122e-06, + "loss": 0.1193, + "step": 12232 + }, + { + "epoch": 5.12, + "learning_rate": 6.051717684133729e-06, + "loss": 0.1221, + "step": 12233 + }, + { + "epoch": 5.12, + "learning_rate": 6.0501600755772074e-06, + "loss": 0.1642, + "step": 12234 + }, + { + "epoch": 5.12, + "learning_rate": 6.048602580555328e-06, + "loss": 0.1479, + "step": 12235 + }, + { + "epoch": 5.12, + "learning_rate": 6.047045199112865e-06, + "loss": 0.1157, + "step": 12236 + }, + { + "epoch": 5.12, + "learning_rate": 6.0454879312945755e-06, + "loss": 0.1253, + "step": 12237 + }, + { + "epoch": 5.12, + "learning_rate": 6.043930777145228e-06, + "loss": 0.1043, + "step": 12238 + }, + { + "epoch": 5.12, + "learning_rate": 6.042373736709581e-06, + "loss": 0.0997, + "step": 12239 + }, + { + "epoch": 5.12, + "learning_rate": 6.040816810032387e-06, + "loss": 0.1033, + "step": 12240 + }, + { + "epoch": 5.12, + "learning_rate": 6.039259997158401e-06, + "loss": 0.1221, + "step": 12241 + }, + { + "epoch": 5.13, + "learning_rate": 6.037703298132373e-06, + "loss": 0.1313, + "step": 12242 + }, + { + "epoch": 5.13, + "learning_rate": 6.036146712999047e-06, + "loss": 0.1127, + "step": 12243 + }, + { + "epoch": 5.13, + "learning_rate": 6.0345902418031645e-06, + "loss": 0.0897, + "step": 12244 + }, + { + "epoch": 5.13, + "learning_rate": 6.03303388458947e-06, + "loss": 0.1243, + "step": 12245 + }, + { + "epoch": 5.13, + "learning_rate": 6.031477641402693e-06, + "loss": 0.1178, + "step": 12246 + }, + { + "epoch": 5.13, + "learning_rate": 6.029921512287571e-06, + "loss": 0.0931, + "step": 12247 + }, + { + "epoch": 5.13, + "learning_rate": 6.028365497288831e-06, + "loss": 0.1232, + "step": 12248 + }, + { + "epoch": 5.13, + "learning_rate": 6.026809596451203e-06, + "loss": 0.099, + "step": 12249 + }, + { + "epoch": 5.13, + "learning_rate": 6.025253809819405e-06, + "loss": 0.0836, + "step": 12250 + }, + { + "epoch": 5.13, + "learning_rate": 6.023698137438159e-06, + "loss": 0.1112, + "step": 12251 + }, + { + "epoch": 5.13, + "learning_rate": 6.022142579352186e-06, + "loss": 0.0906, + "step": 12252 + }, + { + "epoch": 5.13, + "learning_rate": 6.020587135606193e-06, + "loss": 0.1082, + "step": 12253 + }, + { + "epoch": 5.13, + "learning_rate": 6.019031806244888e-06, + "loss": 0.0938, + "step": 12254 + }, + { + "epoch": 5.13, + "learning_rate": 6.017476591312983e-06, + "loss": 0.1267, + "step": 12255 + }, + { + "epoch": 5.13, + "learning_rate": 6.015921490855181e-06, + "loss": 0.1103, + "step": 12256 + }, + { + "epoch": 5.13, + "learning_rate": 6.014366504916178e-06, + "loss": 0.1016, + "step": 12257 + }, + { + "epoch": 5.13, + "learning_rate": 6.012811633540672e-06, + "loss": 0.1498, + "step": 12258 + }, + { + "epoch": 5.13, + "learning_rate": 6.011256876773362e-06, + "loss": 0.1537, + "step": 12259 + }, + { + "epoch": 5.13, + "learning_rate": 6.009702234658929e-06, + "loss": 0.1215, + "step": 12260 + }, + { + "epoch": 5.13, + "learning_rate": 6.008147707242065e-06, + "loss": 0.1021, + "step": 12261 + }, + { + "epoch": 5.13, + "learning_rate": 6.0065932945674535e-06, + "loss": 0.1175, + "step": 12262 + }, + { + "epoch": 5.13, + "learning_rate": 6.005038996679775e-06, + "loss": 0.1038, + "step": 12263 + }, + { + "epoch": 5.13, + "learning_rate": 6.0034848136237035e-06, + "loss": 0.118, + "step": 12264 + }, + { + "epoch": 5.14, + "learning_rate": 6.001930745443916e-06, + "loss": 0.1179, + "step": 12265 + }, + { + "epoch": 5.14, + "learning_rate": 6.000376792185082e-06, + "loss": 0.1159, + "step": 12266 + }, + { + "epoch": 5.14, + "learning_rate": 5.998822953891866e-06, + "loss": 0.1377, + "step": 12267 + }, + { + "epoch": 5.14, + "learning_rate": 5.997269230608935e-06, + "loss": 0.167, + "step": 12268 + }, + { + "epoch": 5.14, + "learning_rate": 5.995715622380949e-06, + "loss": 0.1558, + "step": 12269 + }, + { + "epoch": 5.14, + "learning_rate": 5.994162129252561e-06, + "loss": 0.1122, + "step": 12270 + }, + { + "epoch": 5.14, + "learning_rate": 5.992608751268429e-06, + "loss": 0.0987, + "step": 12271 + }, + { + "epoch": 5.14, + "learning_rate": 5.9910554884732056e-06, + "loss": 0.0981, + "step": 12272 + }, + { + "epoch": 5.14, + "learning_rate": 5.989502340911531e-06, + "loss": 0.1281, + "step": 12273 + }, + { + "epoch": 5.14, + "learning_rate": 5.987949308628054e-06, + "loss": 0.1062, + "step": 12274 + }, + { + "epoch": 5.14, + "learning_rate": 5.986396391667416e-06, + "loss": 0.0926, + "step": 12275 + }, + { + "epoch": 5.14, + "learning_rate": 5.98484359007425e-06, + "loss": 0.1085, + "step": 12276 + }, + { + "epoch": 5.14, + "learning_rate": 5.983290903893193e-06, + "loss": 0.1109, + "step": 12277 + }, + { + "epoch": 5.14, + "learning_rate": 5.981738333168873e-06, + "loss": 0.0862, + "step": 12278 + }, + { + "epoch": 5.14, + "learning_rate": 5.980185877945918e-06, + "loss": 0.1289, + "step": 12279 + }, + { + "epoch": 5.14, + "learning_rate": 5.978633538268957e-06, + "loss": 0.1209, + "step": 12280 + }, + { + "epoch": 5.14, + "learning_rate": 5.977081314182602e-06, + "loss": 0.1088, + "step": 12281 + }, + { + "epoch": 5.14, + "learning_rate": 5.975529205731476e-06, + "loss": 0.1414, + "step": 12282 + }, + { + "epoch": 5.14, + "learning_rate": 5.973977212960194e-06, + "loss": 0.0954, + "step": 12283 + }, + { + "epoch": 5.14, + "learning_rate": 5.9724253359133614e-06, + "loss": 0.1335, + "step": 12284 + }, + { + "epoch": 5.14, + "learning_rate": 5.97087357463559e-06, + "loss": 0.1011, + "step": 12285 + }, + { + "epoch": 5.14, + "learning_rate": 5.969321929171482e-06, + "loss": 0.104, + "step": 12286 + }, + { + "epoch": 5.14, + "learning_rate": 5.967770399565638e-06, + "loss": 0.1147, + "step": 12287 + }, + { + "epoch": 5.14, + "learning_rate": 5.966218985862655e-06, + "loss": 0.1452, + "step": 12288 + }, + { + "epoch": 5.15, + "learning_rate": 5.9646676881071305e-06, + "loss": 0.1098, + "step": 12289 + }, + { + "epoch": 5.15, + "learning_rate": 5.963116506343649e-06, + "loss": 0.1594, + "step": 12290 + }, + { + "epoch": 5.15, + "learning_rate": 5.961565440616803e-06, + "loss": 0.1281, + "step": 12291 + }, + { + "epoch": 5.15, + "learning_rate": 5.9600144909711745e-06, + "loss": 0.1219, + "step": 12292 + }, + { + "epoch": 5.15, + "learning_rate": 5.958463657451345e-06, + "loss": 0.1546, + "step": 12293 + }, + { + "epoch": 5.15, + "learning_rate": 5.956912940101891e-06, + "loss": 0.1529, + "step": 12294 + }, + { + "epoch": 5.15, + "learning_rate": 5.9553623389673855e-06, + "loss": 0.126, + "step": 12295 + }, + { + "epoch": 5.15, + "learning_rate": 5.953811854092404e-06, + "loss": 0.1451, + "step": 12296 + }, + { + "epoch": 5.15, + "learning_rate": 5.952261485521511e-06, + "loss": 0.15, + "step": 12297 + }, + { + "epoch": 5.15, + "learning_rate": 5.950711233299265e-06, + "loss": 0.1165, + "step": 12298 + }, + { + "epoch": 5.15, + "learning_rate": 5.949161097470233e-06, + "loss": 0.1291, + "step": 12299 + }, + { + "epoch": 5.15, + "learning_rate": 5.947611078078972e-06, + "loss": 0.1399, + "step": 12300 + }, + { + "epoch": 5.15, + "learning_rate": 5.9460611751700325e-06, + "loss": 0.1133, + "step": 12301 + }, + { + "epoch": 5.15, + "learning_rate": 5.944511388787967e-06, + "loss": 0.1605, + "step": 12302 + }, + { + "epoch": 5.15, + "learning_rate": 5.942961718977323e-06, + "loss": 0.089, + "step": 12303 + }, + { + "epoch": 5.15, + "learning_rate": 5.941412165782645e-06, + "loss": 0.1576, + "step": 12304 + }, + { + "epoch": 5.15, + "learning_rate": 5.939862729248472e-06, + "loss": 0.131, + "step": 12305 + }, + { + "epoch": 5.15, + "learning_rate": 5.9383134094193425e-06, + "loss": 0.0974, + "step": 12306 + }, + { + "epoch": 5.15, + "learning_rate": 5.936764206339791e-06, + "loss": 0.0774, + "step": 12307 + }, + { + "epoch": 5.15, + "learning_rate": 5.935215120054345e-06, + "loss": 0.1228, + "step": 12308 + }, + { + "epoch": 5.15, + "learning_rate": 5.933666150607533e-06, + "loss": 0.1299, + "step": 12309 + }, + { + "epoch": 5.15, + "learning_rate": 5.932117298043882e-06, + "loss": 0.1099, + "step": 12310 + }, + { + "epoch": 5.15, + "learning_rate": 5.930568562407905e-06, + "loss": 0.1544, + "step": 12311 + }, + { + "epoch": 5.15, + "learning_rate": 5.9290199437441256e-06, + "loss": 0.1006, + "step": 12312 + }, + { + "epoch": 5.16, + "learning_rate": 5.927471442097057e-06, + "loss": 0.1432, + "step": 12313 + }, + { + "epoch": 5.16, + "learning_rate": 5.925923057511207e-06, + "loss": 0.1034, + "step": 12314 + }, + { + "epoch": 5.16, + "learning_rate": 5.9243747900310825e-06, + "loss": 0.1427, + "step": 12315 + }, + { + "epoch": 5.16, + "learning_rate": 5.922826639701189e-06, + "loss": 0.1157, + "step": 12316 + }, + { + "epoch": 5.16, + "learning_rate": 5.921278606566026e-06, + "loss": 0.1101, + "step": 12317 + }, + { + "epoch": 5.16, + "learning_rate": 5.9197306906700865e-06, + "loss": 0.1233, + "step": 12318 + }, + { + "epoch": 5.16, + "learning_rate": 5.918182892057871e-06, + "loss": 0.1274, + "step": 12319 + }, + { + "epoch": 5.16, + "learning_rate": 5.916635210773869e-06, + "loss": 0.1298, + "step": 12320 + }, + { + "epoch": 5.16, + "learning_rate": 5.915087646862561e-06, + "loss": 0.0994, + "step": 12321 + }, + { + "epoch": 5.16, + "learning_rate": 5.913540200368434e-06, + "loss": 0.0831, + "step": 12322 + }, + { + "epoch": 5.16, + "learning_rate": 5.911992871335967e-06, + "loss": 0.1441, + "step": 12323 + }, + { + "epoch": 5.16, + "learning_rate": 5.910445659809638e-06, + "loss": 0.0953, + "step": 12324 + }, + { + "epoch": 5.16, + "learning_rate": 5.908898565833918e-06, + "loss": 0.1614, + "step": 12325 + }, + { + "epoch": 5.16, + "learning_rate": 5.907351589453279e-06, + "loss": 0.1289, + "step": 12326 + }, + { + "epoch": 5.16, + "learning_rate": 5.905804730712189e-06, + "loss": 0.1128, + "step": 12327 + }, + { + "epoch": 5.16, + "learning_rate": 5.904257989655107e-06, + "loss": 0.0952, + "step": 12328 + }, + { + "epoch": 5.16, + "learning_rate": 5.902711366326493e-06, + "loss": 0.1338, + "step": 12329 + }, + { + "epoch": 5.16, + "learning_rate": 5.901164860770808e-06, + "loss": 0.1145, + "step": 12330 + }, + { + "epoch": 5.16, + "learning_rate": 5.8996184730325005e-06, + "loss": 0.1594, + "step": 12331 + }, + { + "epoch": 5.16, + "learning_rate": 5.89807220315602e-06, + "loss": 0.1076, + "step": 12332 + }, + { + "epoch": 5.16, + "learning_rate": 5.896526051185814e-06, + "loss": 0.1145, + "step": 12333 + }, + { + "epoch": 5.16, + "learning_rate": 5.894980017166329e-06, + "loss": 0.1147, + "step": 12334 + }, + { + "epoch": 5.16, + "learning_rate": 5.893434101141997e-06, + "loss": 0.1126, + "step": 12335 + }, + { + "epoch": 5.16, + "learning_rate": 5.891888303157258e-06, + "loss": 0.0837, + "step": 12336 + }, + { + "epoch": 5.17, + "learning_rate": 5.890342623256546e-06, + "loss": 0.0819, + "step": 12337 + }, + { + "epoch": 5.17, + "learning_rate": 5.888797061484288e-06, + "loss": 0.1497, + "step": 12338 + }, + { + "epoch": 5.17, + "learning_rate": 5.887251617884909e-06, + "loss": 0.1503, + "step": 12339 + }, + { + "epoch": 5.17, + "learning_rate": 5.885706292502835e-06, + "loss": 0.1024, + "step": 12340 + }, + { + "epoch": 5.17, + "learning_rate": 5.884161085382481e-06, + "loss": 0.0938, + "step": 12341 + }, + { + "epoch": 5.17, + "learning_rate": 5.882615996568264e-06, + "loss": 0.1657, + "step": 12342 + }, + { + "epoch": 5.17, + "learning_rate": 5.8810710261046e-06, + "loss": 0.0848, + "step": 12343 + }, + { + "epoch": 5.17, + "learning_rate": 5.87952617403589e-06, + "loss": 0.1433, + "step": 12344 + }, + { + "epoch": 5.17, + "learning_rate": 5.877981440406547e-06, + "loss": 0.1287, + "step": 12345 + }, + { + "epoch": 5.17, + "learning_rate": 5.876436825260967e-06, + "loss": 0.1218, + "step": 12346 + }, + { + "epoch": 5.17, + "learning_rate": 5.8748923286435515e-06, + "loss": 0.1001, + "step": 12347 + }, + { + "epoch": 5.17, + "learning_rate": 5.873347950598697e-06, + "loss": 0.1375, + "step": 12348 + }, + { + "epoch": 5.17, + "learning_rate": 5.871803691170792e-06, + "loss": 0.1083, + "step": 12349 + }, + { + "epoch": 5.17, + "learning_rate": 5.870259550404227e-06, + "loss": 0.1457, + "step": 12350 + }, + { + "epoch": 5.17, + "learning_rate": 5.868715528343387e-06, + "loss": 0.1202, + "step": 12351 + }, + { + "epoch": 5.17, + "learning_rate": 5.8671716250326525e-06, + "loss": 0.0686, + "step": 12352 + }, + { + "epoch": 5.17, + "learning_rate": 5.865627840516402e-06, + "loss": 0.1238, + "step": 12353 + }, + { + "epoch": 5.17, + "learning_rate": 5.864084174839013e-06, + "loss": 0.0738, + "step": 12354 + }, + { + "epoch": 5.17, + "learning_rate": 5.86254062804485e-06, + "loss": 0.1097, + "step": 12355 + }, + { + "epoch": 5.17, + "learning_rate": 5.860997200178288e-06, + "loss": 0.1317, + "step": 12356 + }, + { + "epoch": 5.17, + "learning_rate": 5.859453891283688e-06, + "loss": 0.1103, + "step": 12357 + }, + { + "epoch": 5.17, + "learning_rate": 5.857910701405411e-06, + "loss": 0.1518, + "step": 12358 + }, + { + "epoch": 5.17, + "learning_rate": 5.8563676305878155e-06, + "loss": 0.1448, + "step": 12359 + }, + { + "epoch": 5.17, + "learning_rate": 5.854824678875256e-06, + "loss": 0.1091, + "step": 12360 + }, + { + "epoch": 5.18, + "learning_rate": 5.853281846312084e-06, + "loss": 0.1521, + "step": 12361 + }, + { + "epoch": 5.18, + "learning_rate": 5.851739132942645e-06, + "loss": 0.1361, + "step": 12362 + }, + { + "epoch": 5.18, + "learning_rate": 5.850196538811282e-06, + "loss": 0.1437, + "step": 12363 + }, + { + "epoch": 5.18, + "learning_rate": 5.848654063962338e-06, + "loss": 0.135, + "step": 12364 + }, + { + "epoch": 5.18, + "learning_rate": 5.847111708440152e-06, + "loss": 0.1703, + "step": 12365 + }, + { + "epoch": 5.18, + "learning_rate": 5.845569472289051e-06, + "loss": 0.1481, + "step": 12366 + }, + { + "epoch": 5.18, + "learning_rate": 5.844027355553369e-06, + "loss": 0.1317, + "step": 12367 + }, + { + "epoch": 5.18, + "learning_rate": 5.842485358277435e-06, + "loss": 0.1071, + "step": 12368 + }, + { + "epoch": 5.18, + "learning_rate": 5.840943480505567e-06, + "loss": 0.1361, + "step": 12369 + }, + { + "epoch": 5.18, + "learning_rate": 5.839401722282089e-06, + "loss": 0.113, + "step": 12370 + }, + { + "epoch": 5.18, + "learning_rate": 5.837860083651316e-06, + "loss": 0.0978, + "step": 12371 + }, + { + "epoch": 5.18, + "learning_rate": 5.836318564657561e-06, + "loss": 0.1384, + "step": 12372 + }, + { + "epoch": 5.18, + "learning_rate": 5.834777165345138e-06, + "loss": 0.0803, + "step": 12373 + }, + { + "epoch": 5.18, + "learning_rate": 5.833235885758346e-06, + "loss": 0.1489, + "step": 12374 + }, + { + "epoch": 5.18, + "learning_rate": 5.831694725941491e-06, + "loss": 0.1497, + "step": 12375 + }, + { + "epoch": 5.18, + "learning_rate": 5.830153685938873e-06, + "loss": 0.17, + "step": 12376 + }, + { + "epoch": 5.18, + "learning_rate": 5.8286127657947854e-06, + "loss": 0.1264, + "step": 12377 + }, + { + "epoch": 5.18, + "learning_rate": 5.827071965553524e-06, + "loss": 0.1041, + "step": 12378 + }, + { + "epoch": 5.18, + "learning_rate": 5.825531285259378e-06, + "loss": 0.1206, + "step": 12379 + }, + { + "epoch": 5.18, + "learning_rate": 5.8239907249566274e-06, + "loss": 0.1378, + "step": 12380 + }, + { + "epoch": 5.18, + "learning_rate": 5.822450284689558e-06, + "loss": 0.128, + "step": 12381 + }, + { + "epoch": 5.18, + "learning_rate": 5.820909964502448e-06, + "loss": 0.1407, + "step": 12382 + }, + { + "epoch": 5.18, + "learning_rate": 5.819369764439573e-06, + "loss": 0.1534, + "step": 12383 + }, + { + "epoch": 5.18, + "learning_rate": 5.817829684545204e-06, + "loss": 0.173, + "step": 12384 + }, + { + "epoch": 5.19, + "learning_rate": 5.816289724863609e-06, + "loss": 0.113, + "step": 12385 + }, + { + "epoch": 5.19, + "learning_rate": 5.814749885439057e-06, + "loss": 0.1181, + "step": 12386 + }, + { + "epoch": 5.19, + "learning_rate": 5.813210166315804e-06, + "loss": 0.1201, + "step": 12387 + }, + { + "epoch": 5.19, + "learning_rate": 5.81167056753811e-06, + "loss": 0.1776, + "step": 12388 + }, + { + "epoch": 5.19, + "learning_rate": 5.810131089150228e-06, + "loss": 0.146, + "step": 12389 + }, + { + "epoch": 5.19, + "learning_rate": 5.808591731196409e-06, + "loss": 0.1464, + "step": 12390 + }, + { + "epoch": 5.19, + "learning_rate": 5.8070524937209e-06, + "loss": 0.169, + "step": 12391 + }, + { + "epoch": 5.19, + "learning_rate": 5.805513376767947e-06, + "loss": 0.1243, + "step": 12392 + }, + { + "epoch": 5.19, + "learning_rate": 5.8039743803817904e-06, + "loss": 0.1317, + "step": 12393 + }, + { + "epoch": 5.19, + "learning_rate": 5.80243550460667e-06, + "loss": 0.1261, + "step": 12394 + }, + { + "epoch": 5.19, + "learning_rate": 5.800896749486813e-06, + "loss": 0.1377, + "step": 12395 + }, + { + "epoch": 5.19, + "learning_rate": 5.799358115066453e-06, + "loss": 0.1073, + "step": 12396 + }, + { + "epoch": 5.19, + "learning_rate": 5.797819601389815e-06, + "loss": 0.0979, + "step": 12397 + }, + { + "epoch": 5.19, + "learning_rate": 5.7962812085011245e-06, + "loss": 0.0694, + "step": 12398 + }, + { + "epoch": 5.19, + "learning_rate": 5.7947429364446e-06, + "loss": 0.1224, + "step": 12399 + }, + { + "epoch": 5.19, + "learning_rate": 5.793204785264461e-06, + "loss": 0.1102, + "step": 12400 + }, + { + "epoch": 5.19, + "learning_rate": 5.791666755004915e-06, + "loss": 0.1489, + "step": 12401 + }, + { + "epoch": 5.19, + "learning_rate": 5.790128845710175e-06, + "loss": 0.1471, + "step": 12402 + }, + { + "epoch": 5.19, + "learning_rate": 5.788591057424443e-06, + "loss": 0.123, + "step": 12403 + }, + { + "epoch": 5.19, + "learning_rate": 5.787053390191925e-06, + "loss": 0.0856, + "step": 12404 + }, + { + "epoch": 5.19, + "learning_rate": 5.785515844056818e-06, + "loss": 0.1065, + "step": 12405 + }, + { + "epoch": 5.19, + "learning_rate": 5.783978419063323e-06, + "loss": 0.1078, + "step": 12406 + }, + { + "epoch": 5.19, + "learning_rate": 5.782441115255624e-06, + "loss": 0.0962, + "step": 12407 + }, + { + "epoch": 5.19, + "learning_rate": 5.78090393267791e-06, + "loss": 0.13, + "step": 12408 + }, + { + "epoch": 5.2, + "learning_rate": 5.77936687137437e-06, + "loss": 0.134, + "step": 12409 + }, + { + "epoch": 5.2, + "learning_rate": 5.777829931389185e-06, + "loss": 0.125, + "step": 12410 + }, + { + "epoch": 5.2, + "learning_rate": 5.776293112766535e-06, + "loss": 0.1014, + "step": 12411 + }, + { + "epoch": 5.2, + "learning_rate": 5.774756415550587e-06, + "loss": 0.1036, + "step": 12412 + }, + { + "epoch": 5.2, + "learning_rate": 5.773219839785517e-06, + "loss": 0.1309, + "step": 12413 + }, + { + "epoch": 5.2, + "learning_rate": 5.771683385515495e-06, + "loss": 0.1354, + "step": 12414 + }, + { + "epoch": 5.2, + "learning_rate": 5.770147052784678e-06, + "loss": 0.1121, + "step": 12415 + }, + { + "epoch": 5.2, + "learning_rate": 5.768610841637232e-06, + "loss": 0.1305, + "step": 12416 + }, + { + "epoch": 5.2, + "learning_rate": 5.767074752117311e-06, + "loss": 0.1492, + "step": 12417 + }, + { + "epoch": 5.2, + "learning_rate": 5.765538784269071e-06, + "loss": 0.0997, + "step": 12418 + }, + { + "epoch": 5.2, + "learning_rate": 5.76400293813666e-06, + "loss": 0.0921, + "step": 12419 + }, + { + "epoch": 5.2, + "learning_rate": 5.762467213764231e-06, + "loss": 0.1206, + "step": 12420 + }, + { + "epoch": 5.2, + "learning_rate": 5.760931611195917e-06, + "loss": 0.1416, + "step": 12421 + }, + { + "epoch": 5.2, + "learning_rate": 5.75939613047586e-06, + "loss": 0.1282, + "step": 12422 + }, + { + "epoch": 5.2, + "learning_rate": 5.757860771648202e-06, + "loss": 0.1128, + "step": 12423 + }, + { + "epoch": 5.2, + "learning_rate": 5.75632553475707e-06, + "loss": 0.1247, + "step": 12424 + }, + { + "epoch": 5.2, + "learning_rate": 5.754790419846595e-06, + "loss": 0.1211, + "step": 12425 + }, + { + "epoch": 5.2, + "learning_rate": 5.753255426960901e-06, + "loss": 0.1361, + "step": 12426 + }, + { + "epoch": 5.2, + "learning_rate": 5.751720556144116e-06, + "loss": 0.1106, + "step": 12427 + }, + { + "epoch": 5.2, + "learning_rate": 5.750185807440349e-06, + "loss": 0.1182, + "step": 12428 + }, + { + "epoch": 5.2, + "learning_rate": 5.748651180893721e-06, + "loss": 0.1056, + "step": 12429 + }, + { + "epoch": 5.2, + "learning_rate": 5.747116676548342e-06, + "loss": 0.0918, + "step": 12430 + }, + { + "epoch": 5.2, + "learning_rate": 5.745582294448319e-06, + "loss": 0.1437, + "step": 12431 + }, + { + "epoch": 5.2, + "learning_rate": 5.744048034637759e-06, + "loss": 0.15, + "step": 12432 + }, + { + "epoch": 5.21, + "learning_rate": 5.742513897160766e-06, + "loss": 0.1019, + "step": 12433 + }, + { + "epoch": 5.21, + "learning_rate": 5.740979882061433e-06, + "loss": 0.0812, + "step": 12434 + }, + { + "epoch": 5.21, + "learning_rate": 5.7394459893838486e-06, + "loss": 0.0771, + "step": 12435 + }, + { + "epoch": 5.21, + "learning_rate": 5.737912219172108e-06, + "loss": 0.115, + "step": 12436 + }, + { + "epoch": 5.21, + "learning_rate": 5.7363785714703e-06, + "loss": 0.1307, + "step": 12437 + }, + { + "epoch": 5.21, + "learning_rate": 5.734845046322506e-06, + "loss": 0.1003, + "step": 12438 + }, + { + "epoch": 5.21, + "learning_rate": 5.733311643772807e-06, + "loss": 0.1297, + "step": 12439 + }, + { + "epoch": 5.21, + "learning_rate": 5.731778363865278e-06, + "loss": 0.1407, + "step": 12440 + }, + { + "epoch": 5.21, + "learning_rate": 5.730245206643996e-06, + "loss": 0.1088, + "step": 12441 + }, + { + "epoch": 5.21, + "learning_rate": 5.728712172153024e-06, + "loss": 0.1324, + "step": 12442 + }, + { + "epoch": 5.21, + "learning_rate": 5.727179260436428e-06, + "loss": 0.1405, + "step": 12443 + }, + { + "epoch": 5.21, + "learning_rate": 5.7256464715382754e-06, + "loss": 0.1619, + "step": 12444 + }, + { + "epoch": 5.21, + "learning_rate": 5.72411380550262e-06, + "loss": 0.088, + "step": 12445 + }, + { + "epoch": 5.21, + "learning_rate": 5.722581262373519e-06, + "loss": 0.1305, + "step": 12446 + }, + { + "epoch": 5.21, + "learning_rate": 5.721048842195028e-06, + "loss": 0.1233, + "step": 12447 + }, + { + "epoch": 5.21, + "learning_rate": 5.719516545011188e-06, + "loss": 0.127, + "step": 12448 + }, + { + "epoch": 5.21, + "learning_rate": 5.7179843708660454e-06, + "loss": 0.1287, + "step": 12449 + }, + { + "epoch": 5.21, + "learning_rate": 5.716452319803643e-06, + "loss": 0.1295, + "step": 12450 + }, + { + "epoch": 5.21, + "learning_rate": 5.7149203918680166e-06, + "loss": 0.1007, + "step": 12451 + }, + { + "epoch": 5.21, + "learning_rate": 5.713388587103201e-06, + "loss": 0.1025, + "step": 12452 + }, + { + "epoch": 5.21, + "learning_rate": 5.711856905553228e-06, + "loss": 0.1114, + "step": 12453 + }, + { + "epoch": 5.21, + "learning_rate": 5.710325347262125e-06, + "loss": 0.1141, + "step": 12454 + }, + { + "epoch": 5.21, + "learning_rate": 5.708793912273911e-06, + "loss": 0.1332, + "step": 12455 + }, + { + "epoch": 5.21, + "learning_rate": 5.707262600632612e-06, + "loss": 0.1265, + "step": 12456 + }, + { + "epoch": 5.22, + "learning_rate": 5.705731412382235e-06, + "loss": 0.1596, + "step": 12457 + }, + { + "epoch": 5.22, + "learning_rate": 5.704200347566798e-06, + "loss": 0.1179, + "step": 12458 + }, + { + "epoch": 5.22, + "learning_rate": 5.7026694062303115e-06, + "loss": 0.0919, + "step": 12459 + }, + { + "epoch": 5.22, + "learning_rate": 5.701138588416778e-06, + "loss": 0.1254, + "step": 12460 + }, + { + "epoch": 5.22, + "learning_rate": 5.699607894170206e-06, + "loss": 0.1102, + "step": 12461 + }, + { + "epoch": 5.22, + "learning_rate": 5.698077323534584e-06, + "loss": 0.1143, + "step": 12462 + }, + { + "epoch": 5.22, + "learning_rate": 5.696546876553912e-06, + "loss": 0.1093, + "step": 12463 + }, + { + "epoch": 5.22, + "learning_rate": 5.695016553272182e-06, + "loss": 0.1028, + "step": 12464 + }, + { + "epoch": 5.22, + "learning_rate": 5.693486353733382e-06, + "loss": 0.1587, + "step": 12465 + }, + { + "epoch": 5.22, + "learning_rate": 5.691956277981493e-06, + "loss": 0.1528, + "step": 12466 + }, + { + "epoch": 5.22, + "learning_rate": 5.690426326060498e-06, + "loss": 0.1301, + "step": 12467 + }, + { + "epoch": 5.22, + "learning_rate": 5.688896498014381e-06, + "loss": 0.1587, + "step": 12468 + }, + { + "epoch": 5.22, + "learning_rate": 5.687366793887101e-06, + "loss": 0.1473, + "step": 12469 + }, + { + "epoch": 5.22, + "learning_rate": 5.685837213722638e-06, + "loss": 0.1648, + "step": 12470 + }, + { + "epoch": 5.22, + "learning_rate": 5.684307757564954e-06, + "loss": 0.116, + "step": 12471 + }, + { + "epoch": 5.22, + "learning_rate": 5.682778425458015e-06, + "loss": 0.0865, + "step": 12472 + }, + { + "epoch": 5.22, + "learning_rate": 5.681249217445779e-06, + "loss": 0.1476, + "step": 12473 + }, + { + "epoch": 5.22, + "learning_rate": 5.6797201335722064e-06, + "loss": 0.142, + "step": 12474 + }, + { + "epoch": 5.22, + "learning_rate": 5.67819117388124e-06, + "loss": 0.1112, + "step": 12475 + }, + { + "epoch": 5.22, + "learning_rate": 5.676662338416833e-06, + "loss": 0.1331, + "step": 12476 + }, + { + "epoch": 5.22, + "learning_rate": 5.675133627222931e-06, + "loss": 0.1341, + "step": 12477 + }, + { + "epoch": 5.22, + "learning_rate": 5.673605040343475e-06, + "loss": 0.1504, + "step": 12478 + }, + { + "epoch": 5.22, + "learning_rate": 5.6720765778224075e-06, + "loss": 0.1181, + "step": 12479 + }, + { + "epoch": 5.23, + "learning_rate": 5.670548239703653e-06, + "loss": 0.1158, + "step": 12480 + }, + { + "epoch": 5.23, + "learning_rate": 5.669020026031148e-06, + "loss": 0.1225, + "step": 12481 + }, + { + "epoch": 5.23, + "learning_rate": 5.6674919368488235e-06, + "loss": 0.1307, + "step": 12482 + }, + { + "epoch": 5.23, + "learning_rate": 5.665963972200594e-06, + "loss": 0.121, + "step": 12483 + }, + { + "epoch": 5.23, + "learning_rate": 5.664436132130384e-06, + "loss": 0.1369, + "step": 12484 + }, + { + "epoch": 5.23, + "learning_rate": 5.662908416682111e-06, + "loss": 0.1592, + "step": 12485 + }, + { + "epoch": 5.23, + "learning_rate": 5.661380825899687e-06, + "loss": 0.1245, + "step": 12486 + }, + { + "epoch": 5.23, + "learning_rate": 5.6598533598270215e-06, + "loss": 0.0663, + "step": 12487 + }, + { + "epoch": 5.23, + "learning_rate": 5.658326018508022e-06, + "loss": 0.0951, + "step": 12488 + }, + { + "epoch": 5.23, + "learning_rate": 5.656798801986585e-06, + "loss": 0.1009, + "step": 12489 + }, + { + "epoch": 5.23, + "learning_rate": 5.655271710306613e-06, + "loss": 0.128, + "step": 12490 + }, + { + "epoch": 5.23, + "learning_rate": 5.653744743512e-06, + "loss": 0.1113, + "step": 12491 + }, + { + "epoch": 5.23, + "learning_rate": 5.652217901646637e-06, + "loss": 0.1389, + "step": 12492 + }, + { + "epoch": 5.23, + "learning_rate": 5.6506911847544134e-06, + "loss": 0.1356, + "step": 12493 + }, + { + "epoch": 5.23, + "learning_rate": 5.649164592879213e-06, + "loss": 0.1161, + "step": 12494 + }, + { + "epoch": 5.23, + "learning_rate": 5.647638126064917e-06, + "loss": 0.1361, + "step": 12495 + }, + { + "epoch": 5.23, + "learning_rate": 5.646111784355398e-06, + "loss": 0.112, + "step": 12496 + }, + { + "epoch": 5.23, + "learning_rate": 5.644585567794531e-06, + "loss": 0.1376, + "step": 12497 + }, + { + "epoch": 5.23, + "learning_rate": 5.643059476426189e-06, + "loss": 0.1104, + "step": 12498 + }, + { + "epoch": 5.23, + "learning_rate": 5.641533510294235e-06, + "loss": 0.115, + "step": 12499 + }, + { + "epoch": 5.23, + "learning_rate": 5.640007669442533e-06, + "loss": 0.0994, + "step": 12500 + }, + { + "epoch": 5.23, + "learning_rate": 5.638481953914945e-06, + "loss": 0.15, + "step": 12501 + }, + { + "epoch": 5.23, + "learning_rate": 5.636956363755324e-06, + "loss": 0.1126, + "step": 12502 + }, + { + "epoch": 5.23, + "learning_rate": 5.635430899007515e-06, + "loss": 0.1103, + "step": 12503 + }, + { + "epoch": 5.24, + "learning_rate": 5.6339055597153734e-06, + "loss": 0.1124, + "step": 12504 + }, + { + "epoch": 5.24, + "learning_rate": 5.632380345922741e-06, + "loss": 0.1223, + "step": 12505 + }, + { + "epoch": 5.24, + "learning_rate": 5.63085525767346e-06, + "loss": 0.1191, + "step": 12506 + }, + { + "epoch": 5.24, + "learning_rate": 5.629330295011367e-06, + "loss": 0.1154, + "step": 12507 + }, + { + "epoch": 5.24, + "learning_rate": 5.627805457980298e-06, + "loss": 0.1415, + "step": 12508 + }, + { + "epoch": 5.24, + "learning_rate": 5.626280746624083e-06, + "loss": 0.1086, + "step": 12509 + }, + { + "epoch": 5.24, + "learning_rate": 5.6247561609865445e-06, + "loss": 0.0688, + "step": 12510 + }, + { + "epoch": 5.24, + "learning_rate": 5.623231701111507e-06, + "loss": 0.1371, + "step": 12511 + }, + { + "epoch": 5.24, + "learning_rate": 5.6217073670427905e-06, + "loss": 0.1134, + "step": 12512 + }, + { + "epoch": 5.24, + "learning_rate": 5.62018315882421e-06, + "loss": 0.1037, + "step": 12513 + }, + { + "epoch": 5.24, + "learning_rate": 5.618659076499578e-06, + "loss": 0.126, + "step": 12514 + }, + { + "epoch": 5.24, + "learning_rate": 5.6171351201127065e-06, + "loss": 0.0937, + "step": 12515 + }, + { + "epoch": 5.24, + "learning_rate": 5.615611289707394e-06, + "loss": 0.1005, + "step": 12516 + }, + { + "epoch": 5.24, + "learning_rate": 5.614087585327444e-06, + "loss": 0.0986, + "step": 12517 + }, + { + "epoch": 5.24, + "learning_rate": 5.612564007016654e-06, + "loss": 0.1368, + "step": 12518 + }, + { + "epoch": 5.24, + "learning_rate": 5.611040554818819e-06, + "loss": 0.1007, + "step": 12519 + }, + { + "epoch": 5.24, + "learning_rate": 5.6095172287777275e-06, + "loss": 0.081, + "step": 12520 + }, + { + "epoch": 5.24, + "learning_rate": 5.607994028937171e-06, + "loss": 0.0976, + "step": 12521 + }, + { + "epoch": 5.24, + "learning_rate": 5.606470955340926e-06, + "loss": 0.1193, + "step": 12522 + }, + { + "epoch": 5.24, + "learning_rate": 5.604948008032774e-06, + "loss": 0.1216, + "step": 12523 + }, + { + "epoch": 5.24, + "learning_rate": 5.603425187056496e-06, + "loss": 0.0846, + "step": 12524 + }, + { + "epoch": 5.24, + "learning_rate": 5.601902492455855e-06, + "loss": 0.0941, + "step": 12525 + }, + { + "epoch": 5.24, + "learning_rate": 5.6003799242746235e-06, + "loss": 0.0941, + "step": 12526 + }, + { + "epoch": 5.24, + "learning_rate": 5.598857482556568e-06, + "loss": 0.107, + "step": 12527 + }, + { + "epoch": 5.25, + "learning_rate": 5.597335167345448e-06, + "loss": 0.1391, + "step": 12528 + }, + { + "epoch": 5.25, + "learning_rate": 5.595812978685027e-06, + "loss": 0.1228, + "step": 12529 + }, + { + "epoch": 5.25, + "learning_rate": 5.594290916619048e-06, + "loss": 0.1307, + "step": 12530 + }, + { + "epoch": 5.25, + "learning_rate": 5.592768981191268e-06, + "loss": 0.1124, + "step": 12531 + }, + { + "epoch": 5.25, + "learning_rate": 5.591247172445432e-06, + "loss": 0.1211, + "step": 12532 + }, + { + "epoch": 5.25, + "learning_rate": 5.589725490425284e-06, + "loss": 0.1426, + "step": 12533 + }, + { + "epoch": 5.25, + "learning_rate": 5.588203935174563e-06, + "loss": 0.1508, + "step": 12534 + }, + { + "epoch": 5.25, + "learning_rate": 5.586682506737007e-06, + "loss": 0.1433, + "step": 12535 + }, + { + "epoch": 5.25, + "learning_rate": 5.585161205156347e-06, + "loss": 0.0949, + "step": 12536 + }, + { + "epoch": 5.25, + "learning_rate": 5.583640030476309e-06, + "loss": 0.1226, + "step": 12537 + }, + { + "epoch": 5.25, + "learning_rate": 5.582118982740618e-06, + "loss": 0.1161, + "step": 12538 + }, + { + "epoch": 5.25, + "learning_rate": 5.580598061992997e-06, + "loss": 0.1382, + "step": 12539 + }, + { + "epoch": 5.25, + "learning_rate": 5.5790772682771625e-06, + "loss": 0.1094, + "step": 12540 + }, + { + "epoch": 5.25, + "learning_rate": 5.577556601636828e-06, + "loss": 0.1252, + "step": 12541 + }, + { + "epoch": 5.25, + "learning_rate": 5.576036062115709e-06, + "loss": 0.1765, + "step": 12542 + }, + { + "epoch": 5.25, + "learning_rate": 5.574515649757505e-06, + "loss": 0.1138, + "step": 12543 + }, + { + "epoch": 5.25, + "learning_rate": 5.572995364605921e-06, + "loss": 0.1526, + "step": 12544 + }, + { + "epoch": 5.25, + "learning_rate": 5.571475206704656e-06, + "loss": 0.1438, + "step": 12545 + }, + { + "epoch": 5.25, + "learning_rate": 5.569955176097408e-06, + "loss": 0.1175, + "step": 12546 + }, + { + "epoch": 5.25, + "learning_rate": 5.5684352728278705e-06, + "loss": 0.1257, + "step": 12547 + }, + { + "epoch": 5.25, + "learning_rate": 5.566915496939723e-06, + "loss": 0.1317, + "step": 12548 + }, + { + "epoch": 5.25, + "learning_rate": 5.565395848476658e-06, + "loss": 0.1251, + "step": 12549 + }, + { + "epoch": 5.25, + "learning_rate": 5.563876327482356e-06, + "loss": 0.1561, + "step": 12550 + }, + { + "epoch": 5.25, + "learning_rate": 5.56235693400049e-06, + "loss": 0.1404, + "step": 12551 + }, + { + "epoch": 5.26, + "learning_rate": 5.5608376680747354e-06, + "loss": 0.144, + "step": 12552 + }, + { + "epoch": 5.26, + "learning_rate": 5.559318529748763e-06, + "loss": 0.1175, + "step": 12553 + }, + { + "epoch": 5.26, + "learning_rate": 5.55779951906624e-06, + "loss": 0.1148, + "step": 12554 + }, + { + "epoch": 5.26, + "learning_rate": 5.556280636070826e-06, + "loss": 0.1124, + "step": 12555 + }, + { + "epoch": 5.26, + "learning_rate": 5.554761880806187e-06, + "loss": 0.102, + "step": 12556 + }, + { + "epoch": 5.26, + "learning_rate": 5.5532432533159676e-06, + "loss": 0.1266, + "step": 12557 + }, + { + "epoch": 5.26, + "learning_rate": 5.551724753643826e-06, + "loss": 0.1217, + "step": 12558 + }, + { + "epoch": 5.26, + "learning_rate": 5.550206381833409e-06, + "loss": 0.1142, + "step": 12559 + }, + { + "epoch": 5.26, + "learning_rate": 5.548688137928362e-06, + "loss": 0.0905, + "step": 12560 + }, + { + "epoch": 5.26, + "learning_rate": 5.547170021972323e-06, + "loss": 0.1191, + "step": 12561 + }, + { + "epoch": 5.26, + "learning_rate": 5.545652034008935e-06, + "loss": 0.1319, + "step": 12562 + }, + { + "epoch": 5.26, + "learning_rate": 5.544134174081821e-06, + "loss": 0.1422, + "step": 12563 + }, + { + "epoch": 5.26, + "learning_rate": 5.542616442234618e-06, + "loss": 0.0982, + "step": 12564 + }, + { + "epoch": 5.26, + "learning_rate": 5.54109883851095e-06, + "loss": 0.1629, + "step": 12565 + }, + { + "epoch": 5.26, + "learning_rate": 5.539581362954438e-06, + "loss": 0.1433, + "step": 12566 + }, + { + "epoch": 5.26, + "learning_rate": 5.5380640156087016e-06, + "loss": 0.0967, + "step": 12567 + }, + { + "epoch": 5.26, + "learning_rate": 5.536546796517357e-06, + "loss": 0.1558, + "step": 12568 + }, + { + "epoch": 5.26, + "learning_rate": 5.535029705724017e-06, + "loss": 0.0975, + "step": 12569 + }, + { + "epoch": 5.26, + "learning_rate": 5.533512743272285e-06, + "loss": 0.1051, + "step": 12570 + }, + { + "epoch": 5.26, + "learning_rate": 5.531995909205763e-06, + "loss": 0.1519, + "step": 12571 + }, + { + "epoch": 5.26, + "learning_rate": 5.530479203568053e-06, + "loss": 0.1319, + "step": 12572 + }, + { + "epoch": 5.26, + "learning_rate": 5.528962626402754e-06, + "loss": 0.1028, + "step": 12573 + }, + { + "epoch": 5.26, + "learning_rate": 5.527446177753455e-06, + "loss": 0.1478, + "step": 12574 + }, + { + "epoch": 5.26, + "learning_rate": 5.525929857663747e-06, + "loss": 0.1575, + "step": 12575 + }, + { + "epoch": 5.27, + "learning_rate": 5.524413666177216e-06, + "loss": 0.116, + "step": 12576 + }, + { + "epoch": 5.27, + "learning_rate": 5.522897603337445e-06, + "loss": 0.1317, + "step": 12577 + }, + { + "epoch": 5.27, + "learning_rate": 5.521381669188006e-06, + "loss": 0.1256, + "step": 12578 + }, + { + "epoch": 5.27, + "learning_rate": 5.519865863772477e-06, + "loss": 0.1148, + "step": 12579 + }, + { + "epoch": 5.27, + "learning_rate": 5.518350187134427e-06, + "loss": 0.1114, + "step": 12580 + }, + { + "epoch": 5.27, + "learning_rate": 5.516834639317426e-06, + "loss": 0.1411, + "step": 12581 + }, + { + "epoch": 5.27, + "learning_rate": 5.515319220365034e-06, + "loss": 0.1377, + "step": 12582 + }, + { + "epoch": 5.27, + "learning_rate": 5.513803930320815e-06, + "loss": 0.079, + "step": 12583 + }, + { + "epoch": 5.27, + "learning_rate": 5.512288769228318e-06, + "loss": 0.1564, + "step": 12584 + }, + { + "epoch": 5.27, + "learning_rate": 5.510773737131097e-06, + "loss": 0.1106, + "step": 12585 + }, + { + "epoch": 5.27, + "learning_rate": 5.509258834072702e-06, + "loss": 0.1236, + "step": 12586 + }, + { + "epoch": 5.27, + "learning_rate": 5.507744060096676e-06, + "loss": 0.1165, + "step": 12587 + }, + { + "epoch": 5.27, + "learning_rate": 5.50622941524656e-06, + "loss": 0.1437, + "step": 12588 + }, + { + "epoch": 5.27, + "learning_rate": 5.5047148995658976e-06, + "loss": 0.1063, + "step": 12589 + }, + { + "epoch": 5.27, + "learning_rate": 5.5032005130982115e-06, + "loss": 0.1413, + "step": 12590 + }, + { + "epoch": 5.27, + "learning_rate": 5.501686255887036e-06, + "loss": 0.1175, + "step": 12591 + }, + { + "epoch": 5.27, + "learning_rate": 5.500172127975903e-06, + "loss": 0.1253, + "step": 12592 + }, + { + "epoch": 5.27, + "learning_rate": 5.498658129408323e-06, + "loss": 0.1528, + "step": 12593 + }, + { + "epoch": 5.27, + "learning_rate": 5.49714426022782e-06, + "loss": 0.104, + "step": 12594 + }, + { + "epoch": 5.27, + "learning_rate": 5.495630520477912e-06, + "loss": 0.1079, + "step": 12595 + }, + { + "epoch": 5.27, + "learning_rate": 5.4941169102021065e-06, + "loss": 0.1383, + "step": 12596 + }, + { + "epoch": 5.27, + "learning_rate": 5.492603429443916e-06, + "loss": 0.1617, + "step": 12597 + }, + { + "epoch": 5.27, + "learning_rate": 5.491090078246836e-06, + "loss": 0.1331, + "step": 12598 + }, + { + "epoch": 5.27, + "learning_rate": 5.489576856654371e-06, + "loss": 0.161, + "step": 12599 + }, + { + "epoch": 5.28, + "learning_rate": 5.488063764710016e-06, + "loss": 0.1442, + "step": 12600 + }, + { + "epoch": 5.28, + "learning_rate": 5.486550802457264e-06, + "loss": 0.1255, + "step": 12601 + }, + { + "epoch": 5.28, + "learning_rate": 5.485037969939605e-06, + "loss": 0.1291, + "step": 12602 + }, + { + "epoch": 5.28, + "learning_rate": 5.483525267200527e-06, + "loss": 0.1409, + "step": 12603 + }, + { + "epoch": 5.28, + "learning_rate": 5.482012694283503e-06, + "loss": 0.1257, + "step": 12604 + }, + { + "epoch": 5.28, + "learning_rate": 5.480500251232015e-06, + "loss": 0.1324, + "step": 12605 + }, + { + "epoch": 5.28, + "learning_rate": 5.478987938089537e-06, + "loss": 0.1447, + "step": 12606 + }, + { + "epoch": 5.28, + "learning_rate": 5.477475754899538e-06, + "loss": 0.1482, + "step": 12607 + }, + { + "epoch": 5.28, + "learning_rate": 5.475963701705487e-06, + "loss": 0.1701, + "step": 12608 + }, + { + "epoch": 5.28, + "learning_rate": 5.4744517785508425e-06, + "loss": 0.1152, + "step": 12609 + }, + { + "epoch": 5.28, + "learning_rate": 5.472939985479071e-06, + "loss": 0.1722, + "step": 12610 + }, + { + "epoch": 5.28, + "learning_rate": 5.471428322533619e-06, + "loss": 0.124, + "step": 12611 + }, + { + "epoch": 5.28, + "learning_rate": 5.469916789757941e-06, + "loss": 0.1133, + "step": 12612 + }, + { + "epoch": 5.28, + "learning_rate": 5.4684053871954835e-06, + "loss": 0.0509, + "step": 12613 + }, + { + "epoch": 5.28, + "learning_rate": 5.466894114889697e-06, + "loss": 0.126, + "step": 12614 + }, + { + "epoch": 5.28, + "learning_rate": 5.4653829728840126e-06, + "loss": 0.117, + "step": 12615 + }, + { + "epoch": 5.28, + "learning_rate": 5.463871961221871e-06, + "loss": 0.115, + "step": 12616 + }, + { + "epoch": 5.28, + "learning_rate": 5.462361079946708e-06, + "loss": 0.1245, + "step": 12617 + }, + { + "epoch": 5.28, + "learning_rate": 5.460850329101944e-06, + "loss": 0.1079, + "step": 12618 + }, + { + "epoch": 5.28, + "learning_rate": 5.459339708731011e-06, + "loss": 0.0993, + "step": 12619 + }, + { + "epoch": 5.28, + "learning_rate": 5.457829218877328e-06, + "loss": 0.1243, + "step": 12620 + }, + { + "epoch": 5.28, + "learning_rate": 5.456318859584314e-06, + "loss": 0.1076, + "step": 12621 + }, + { + "epoch": 5.28, + "learning_rate": 5.454808630895383e-06, + "loss": 0.1082, + "step": 12622 + }, + { + "epoch": 5.28, + "learning_rate": 5.453298532853943e-06, + "loss": 0.1531, + "step": 12623 + }, + { + "epoch": 5.29, + "learning_rate": 5.451788565503406e-06, + "loss": 0.1146, + "step": 12624 + }, + { + "epoch": 5.29, + "learning_rate": 5.450278728887168e-06, + "loss": 0.1133, + "step": 12625 + }, + { + "epoch": 5.29, + "learning_rate": 5.44876902304863e-06, + "loss": 0.1498, + "step": 12626 + }, + { + "epoch": 5.29, + "learning_rate": 5.447259448031188e-06, + "loss": 0.1062, + "step": 12627 + }, + { + "epoch": 5.29, + "learning_rate": 5.445750003878233e-06, + "loss": 0.1089, + "step": 12628 + }, + { + "epoch": 5.29, + "learning_rate": 5.444240690633152e-06, + "loss": 0.1413, + "step": 12629 + }, + { + "epoch": 5.29, + "learning_rate": 5.442731508339334e-06, + "loss": 0.1212, + "step": 12630 + }, + { + "epoch": 5.29, + "learning_rate": 5.441222457040152e-06, + "loss": 0.1449, + "step": 12631 + }, + { + "epoch": 5.29, + "learning_rate": 5.439713536778983e-06, + "loss": 0.1023, + "step": 12632 + }, + { + "epoch": 5.29, + "learning_rate": 5.438204747599204e-06, + "loss": 0.135, + "step": 12633 + }, + { + "epoch": 5.29, + "learning_rate": 5.43669608954418e-06, + "loss": 0.1051, + "step": 12634 + }, + { + "epoch": 5.29, + "learning_rate": 5.435187562657278e-06, + "loss": 0.1177, + "step": 12635 + }, + { + "epoch": 5.29, + "learning_rate": 5.433679166981858e-06, + "loss": 0.1278, + "step": 12636 + }, + { + "epoch": 5.29, + "learning_rate": 5.432170902561282e-06, + "loss": 0.088, + "step": 12637 + }, + { + "epoch": 5.29, + "learning_rate": 5.4306627694389014e-06, + "loss": 0.1628, + "step": 12638 + }, + { + "epoch": 5.29, + "learning_rate": 5.42915476765806e-06, + "loss": 0.0765, + "step": 12639 + }, + { + "epoch": 5.29, + "learning_rate": 5.427646897262109e-06, + "loss": 0.1318, + "step": 12640 + }, + { + "epoch": 5.29, + "learning_rate": 5.42613915829439e-06, + "loss": 0.097, + "step": 12641 + }, + { + "epoch": 5.29, + "learning_rate": 5.424631550798243e-06, + "loss": 0.1231, + "step": 12642 + }, + { + "epoch": 5.29, + "learning_rate": 5.423124074817001e-06, + "loss": 0.1377, + "step": 12643 + }, + { + "epoch": 5.29, + "learning_rate": 5.421616730394e-06, + "loss": 0.1024, + "step": 12644 + }, + { + "epoch": 5.29, + "learning_rate": 5.42010951757256e-06, + "loss": 0.1169, + "step": 12645 + }, + { + "epoch": 5.29, + "learning_rate": 5.418602436396007e-06, + "loss": 0.1132, + "step": 12646 + }, + { + "epoch": 5.29, + "learning_rate": 5.417095486907663e-06, + "loss": 0.1527, + "step": 12647 + }, + { + "epoch": 5.3, + "learning_rate": 5.415588669150842e-06, + "loss": 0.1227, + "step": 12648 + }, + { + "epoch": 5.3, + "learning_rate": 5.414081983168855e-06, + "loss": 0.131, + "step": 12649 + }, + { + "epoch": 5.3, + "learning_rate": 5.412575429005014e-06, + "loss": 0.1329, + "step": 12650 + }, + { + "epoch": 5.3, + "learning_rate": 5.411069006702624e-06, + "loss": 0.1036, + "step": 12651 + }, + { + "epoch": 5.3, + "learning_rate": 5.409562716304979e-06, + "loss": 0.1419, + "step": 12652 + }, + { + "epoch": 5.3, + "learning_rate": 5.408056557855381e-06, + "loss": 0.125, + "step": 12653 + }, + { + "epoch": 5.3, + "learning_rate": 5.406550531397123e-06, + "loss": 0.1224, + "step": 12654 + }, + { + "epoch": 5.3, + "learning_rate": 5.405044636973493e-06, + "loss": 0.1051, + "step": 12655 + }, + { + "epoch": 5.3, + "learning_rate": 5.403538874627777e-06, + "loss": 0.131, + "step": 12656 + }, + { + "epoch": 5.3, + "learning_rate": 5.402033244403262e-06, + "loss": 0.0957, + "step": 12657 + }, + { + "epoch": 5.3, + "learning_rate": 5.400527746343217e-06, + "loss": 0.1377, + "step": 12658 + }, + { + "epoch": 5.3, + "learning_rate": 5.399022380490921e-06, + "loss": 0.086, + "step": 12659 + }, + { + "epoch": 5.3, + "learning_rate": 5.397517146889647e-06, + "loss": 0.0981, + "step": 12660 + }, + { + "epoch": 5.3, + "learning_rate": 5.3960120455826545e-06, + "loss": 0.1452, + "step": 12661 + }, + { + "epoch": 5.3, + "learning_rate": 5.394507076613212e-06, + "loss": 0.1332, + "step": 12662 + }, + { + "epoch": 5.3, + "learning_rate": 5.3930022400245745e-06, + "loss": 0.1038, + "step": 12663 + }, + { + "epoch": 5.3, + "learning_rate": 5.391497535859999e-06, + "loss": 0.1076, + "step": 12664 + }, + { + "epoch": 5.3, + "learning_rate": 5.3899929641627425e-06, + "loss": 0.1232, + "step": 12665 + }, + { + "epoch": 5.3, + "learning_rate": 5.388488524976043e-06, + "loss": 0.1222, + "step": 12666 + }, + { + "epoch": 5.3, + "learning_rate": 5.3869842183431495e-06, + "loss": 0.1345, + "step": 12667 + }, + { + "epoch": 5.3, + "learning_rate": 5.3854800443073006e-06, + "loss": 0.0914, + "step": 12668 + }, + { + "epoch": 5.3, + "learning_rate": 5.383976002911734e-06, + "loss": 0.152, + "step": 12669 + }, + { + "epoch": 5.3, + "learning_rate": 5.38247209419968e-06, + "loss": 0.1065, + "step": 12670 + }, + { + "epoch": 5.31, + "learning_rate": 5.3809683182143725e-06, + "loss": 0.1336, + "step": 12671 + }, + { + "epoch": 5.31, + "learning_rate": 5.3794646749990265e-06, + "loss": 0.1259, + "step": 12672 + }, + { + "epoch": 5.31, + "learning_rate": 5.3779611645968696e-06, + "loss": 0.1357, + "step": 12673 + }, + { + "epoch": 5.31, + "learning_rate": 5.376457787051117e-06, + "loss": 0.1132, + "step": 12674 + }, + { + "epoch": 5.31, + "learning_rate": 5.374954542404982e-06, + "loss": 0.1397, + "step": 12675 + }, + { + "epoch": 5.31, + "learning_rate": 5.373451430701676e-06, + "loss": 0.0993, + "step": 12676 + }, + { + "epoch": 5.31, + "learning_rate": 5.371948451984401e-06, + "loss": 0.0798, + "step": 12677 + }, + { + "epoch": 5.31, + "learning_rate": 5.3704456062963674e-06, + "loss": 0.1446, + "step": 12678 + }, + { + "epoch": 5.31, + "learning_rate": 5.368942893680761e-06, + "loss": 0.1033, + "step": 12679 + }, + { + "epoch": 5.31, + "learning_rate": 5.367440314180782e-06, + "loss": 0.125, + "step": 12680 + }, + { + "epoch": 5.31, + "learning_rate": 5.36593786783962e-06, + "loss": 0.1587, + "step": 12681 + }, + { + "epoch": 5.31, + "learning_rate": 5.3644355547004664e-06, + "loss": 0.1135, + "step": 12682 + }, + { + "epoch": 5.31, + "learning_rate": 5.362933374806494e-06, + "loss": 0.1085, + "step": 12683 + }, + { + "epoch": 5.31, + "learning_rate": 5.361431328200888e-06, + "loss": 0.1236, + "step": 12684 + }, + { + "epoch": 5.31, + "learning_rate": 5.3599294149268255e-06, + "loss": 0.1151, + "step": 12685 + }, + { + "epoch": 5.31, + "learning_rate": 5.3584276350274725e-06, + "loss": 0.1447, + "step": 12686 + }, + { + "epoch": 5.31, + "learning_rate": 5.356925988545996e-06, + "loss": 0.1179, + "step": 12687 + }, + { + "epoch": 5.31, + "learning_rate": 5.355424475525562e-06, + "loss": 0.1462, + "step": 12688 + }, + { + "epoch": 5.31, + "learning_rate": 5.3539230960093305e-06, + "loss": 0.1122, + "step": 12689 + }, + { + "epoch": 5.31, + "learning_rate": 5.352421850040457e-06, + "loss": 0.1016, + "step": 12690 + }, + { + "epoch": 5.31, + "learning_rate": 5.3509207376620935e-06, + "loss": 0.1547, + "step": 12691 + }, + { + "epoch": 5.31, + "learning_rate": 5.349419758917392e-06, + "loss": 0.1373, + "step": 12692 + }, + { + "epoch": 5.31, + "learning_rate": 5.347918913849487e-06, + "loss": 0.1006, + "step": 12693 + }, + { + "epoch": 5.31, + "learning_rate": 5.346418202501526e-06, + "loss": 0.1653, + "step": 12694 + }, + { + "epoch": 5.32, + "learning_rate": 5.344917624916644e-06, + "loss": 0.1047, + "step": 12695 + }, + { + "epoch": 5.32, + "learning_rate": 5.343417181137973e-06, + "loss": 0.1224, + "step": 12696 + }, + { + "epoch": 5.32, + "learning_rate": 5.341916871208644e-06, + "loss": 0.1222, + "step": 12697 + }, + { + "epoch": 5.32, + "learning_rate": 5.340416695171785e-06, + "loss": 0.1651, + "step": 12698 + }, + { + "epoch": 5.32, + "learning_rate": 5.3389166530705096e-06, + "loss": 0.1085, + "step": 12699 + }, + { + "epoch": 5.32, + "learning_rate": 5.337416744947939e-06, + "loss": 0.1256, + "step": 12700 + }, + { + "epoch": 5.32, + "learning_rate": 5.335916970847185e-06, + "loss": 0.0904, + "step": 12701 + }, + { + "epoch": 5.32, + "learning_rate": 5.3344173308113586e-06, + "loss": 0.1208, + "step": 12702 + }, + { + "epoch": 5.32, + "learning_rate": 5.332917824883568e-06, + "loss": 0.0958, + "step": 12703 + }, + { + "epoch": 5.32, + "learning_rate": 5.331418453106916e-06, + "loss": 0.1328, + "step": 12704 + }, + { + "epoch": 5.32, + "learning_rate": 5.329919215524494e-06, + "loss": 0.1195, + "step": 12705 + }, + { + "epoch": 5.32, + "learning_rate": 5.328420112179405e-06, + "loss": 0.094, + "step": 12706 + }, + { + "epoch": 5.32, + "learning_rate": 5.326921143114727e-06, + "loss": 0.1097, + "step": 12707 + }, + { + "epoch": 5.32, + "learning_rate": 5.325422308373557e-06, + "loss": 0.1508, + "step": 12708 + }, + { + "epoch": 5.32, + "learning_rate": 5.3239236079989735e-06, + "loss": 0.0881, + "step": 12709 + }, + { + "epoch": 5.32, + "learning_rate": 5.322425042034056e-06, + "loss": 0.1582, + "step": 12710 + }, + { + "epoch": 5.32, + "learning_rate": 5.3209266105218815e-06, + "loss": 0.1305, + "step": 12711 + }, + { + "epoch": 5.32, + "learning_rate": 5.319428313505523e-06, + "loss": 0.1466, + "step": 12712 + }, + { + "epoch": 5.32, + "learning_rate": 5.317930151028039e-06, + "loss": 0.1536, + "step": 12713 + }, + { + "epoch": 5.32, + "learning_rate": 5.316432123132498e-06, + "loss": 0.1155, + "step": 12714 + }, + { + "epoch": 5.32, + "learning_rate": 5.314934229861961e-06, + "loss": 0.1389, + "step": 12715 + }, + { + "epoch": 5.32, + "learning_rate": 5.3134364712594814e-06, + "loss": 0.1402, + "step": 12716 + }, + { + "epoch": 5.32, + "learning_rate": 5.311938847368111e-06, + "loss": 0.1609, + "step": 12717 + }, + { + "epoch": 5.32, + "learning_rate": 5.310441358230899e-06, + "loss": 0.1318, + "step": 12718 + }, + { + "epoch": 5.33, + "learning_rate": 5.308944003890892e-06, + "loss": 0.1152, + "step": 12719 + }, + { + "epoch": 5.33, + "learning_rate": 5.307446784391124e-06, + "loss": 0.1604, + "step": 12720 + }, + { + "epoch": 5.33, + "learning_rate": 5.305949699774633e-06, + "loss": 0.1279, + "step": 12721 + }, + { + "epoch": 5.33, + "learning_rate": 5.304452750084452e-06, + "loss": 0.1333, + "step": 12722 + }, + { + "epoch": 5.33, + "learning_rate": 5.3029559353636116e-06, + "loss": 0.1252, + "step": 12723 + }, + { + "epoch": 5.33, + "learning_rate": 5.301459255655134e-06, + "loss": 0.1178, + "step": 12724 + }, + { + "epoch": 5.33, + "learning_rate": 5.299962711002045e-06, + "loss": 0.1461, + "step": 12725 + }, + { + "epoch": 5.33, + "learning_rate": 5.298466301447353e-06, + "loss": 0.1093, + "step": 12726 + }, + { + "epoch": 5.33, + "learning_rate": 5.296970027034074e-06, + "loss": 0.1464, + "step": 12727 + }, + { + "epoch": 5.33, + "learning_rate": 5.295473887805223e-06, + "loss": 0.1338, + "step": 12728 + }, + { + "epoch": 5.33, + "learning_rate": 5.293977883803797e-06, + "loss": 0.1077, + "step": 12729 + }, + { + "epoch": 5.33, + "learning_rate": 5.292482015072799e-06, + "loss": 0.1061, + "step": 12730 + }, + { + "epoch": 5.33, + "learning_rate": 5.29098628165523e-06, + "loss": 0.1552, + "step": 12731 + }, + { + "epoch": 5.33, + "learning_rate": 5.289490683594081e-06, + "loss": 0.1318, + "step": 12732 + }, + { + "epoch": 5.33, + "learning_rate": 5.287995220932346e-06, + "loss": 0.1018, + "step": 12733 + }, + { + "epoch": 5.33, + "learning_rate": 5.286499893713003e-06, + "loss": 0.131, + "step": 12734 + }, + { + "epoch": 5.33, + "learning_rate": 5.285004701979037e-06, + "loss": 0.092, + "step": 12735 + }, + { + "epoch": 5.33, + "learning_rate": 5.283509645773429e-06, + "loss": 0.1369, + "step": 12736 + }, + { + "epoch": 5.33, + "learning_rate": 5.282014725139149e-06, + "loss": 0.1574, + "step": 12737 + }, + { + "epoch": 5.33, + "learning_rate": 5.280519940119171e-06, + "loss": 0.1436, + "step": 12738 + }, + { + "epoch": 5.33, + "learning_rate": 5.279025290756461e-06, + "loss": 0.0928, + "step": 12739 + }, + { + "epoch": 5.33, + "learning_rate": 5.2775307770939775e-06, + "loss": 0.1305, + "step": 12740 + }, + { + "epoch": 5.33, + "learning_rate": 5.27603639917468e-06, + "loss": 0.11, + "step": 12741 + }, + { + "epoch": 5.33, + "learning_rate": 5.2745421570415245e-06, + "loss": 0.1133, + "step": 12742 + }, + { + "epoch": 5.34, + "learning_rate": 5.273048050737461e-06, + "loss": 0.0938, + "step": 12743 + }, + { + "epoch": 5.34, + "learning_rate": 5.271554080305439e-06, + "loss": 0.1041, + "step": 12744 + }, + { + "epoch": 5.34, + "learning_rate": 5.270060245788401e-06, + "loss": 0.1204, + "step": 12745 + }, + { + "epoch": 5.34, + "learning_rate": 5.2685665472292805e-06, + "loss": 0.1563, + "step": 12746 + }, + { + "epoch": 5.34, + "learning_rate": 5.2670729846710155e-06, + "loss": 0.1355, + "step": 12747 + }, + { + "epoch": 5.34, + "learning_rate": 5.265579558156537e-06, + "loss": 0.1159, + "step": 12748 + }, + { + "epoch": 5.34, + "learning_rate": 5.264086267728774e-06, + "loss": 0.1063, + "step": 12749 + }, + { + "epoch": 5.34, + "learning_rate": 5.262593113430653e-06, + "loss": 0.1926, + "step": 12750 + }, + { + "epoch": 5.34, + "learning_rate": 5.261100095305085e-06, + "loss": 0.1129, + "step": 12751 + }, + { + "epoch": 5.34, + "learning_rate": 5.259607213394989e-06, + "loss": 0.1329, + "step": 12752 + }, + { + "epoch": 5.34, + "learning_rate": 5.2581144677432805e-06, + "loss": 0.1095, + "step": 12753 + }, + { + "epoch": 5.34, + "learning_rate": 5.256621858392861e-06, + "loss": 0.0983, + "step": 12754 + }, + { + "epoch": 5.34, + "learning_rate": 5.255129385386635e-06, + "loss": 0.1131, + "step": 12755 + }, + { + "epoch": 5.34, + "learning_rate": 5.2536370487675056e-06, + "loss": 0.1182, + "step": 12756 + }, + { + "epoch": 5.34, + "learning_rate": 5.252144848578368e-06, + "loss": 0.101, + "step": 12757 + }, + { + "epoch": 5.34, + "learning_rate": 5.250652784862114e-06, + "loss": 0.1241, + "step": 12758 + }, + { + "epoch": 5.34, + "learning_rate": 5.249160857661628e-06, + "loss": 0.1303, + "step": 12759 + }, + { + "epoch": 5.34, + "learning_rate": 5.247669067019805e-06, + "loss": 0.0913, + "step": 12760 + }, + { + "epoch": 5.34, + "learning_rate": 5.246177412979512e-06, + "loss": 0.0975, + "step": 12761 + }, + { + "epoch": 5.34, + "learning_rate": 5.2446858955836295e-06, + "loss": 0.0928, + "step": 12762 + }, + { + "epoch": 5.34, + "learning_rate": 5.243194514875033e-06, + "loss": 0.1463, + "step": 12763 + }, + { + "epoch": 5.34, + "learning_rate": 5.241703270896589e-06, + "loss": 0.1404, + "step": 12764 + }, + { + "epoch": 5.34, + "learning_rate": 5.240212163691161e-06, + "loss": 0.103, + "step": 12765 + }, + { + "epoch": 5.34, + "learning_rate": 5.2387211933016145e-06, + "loss": 0.1103, + "step": 12766 + }, + { + "epoch": 5.35, + "learning_rate": 5.237230359770799e-06, + "loss": 0.1328, + "step": 12767 + }, + { + "epoch": 5.35, + "learning_rate": 5.235739663141571e-06, + "loss": 0.1178, + "step": 12768 + }, + { + "epoch": 5.35, + "learning_rate": 5.2342491034567775e-06, + "loss": 0.124, + "step": 12769 + }, + { + "epoch": 5.35, + "learning_rate": 5.232758680759264e-06, + "loss": 0.0963, + "step": 12770 + }, + { + "epoch": 5.35, + "learning_rate": 5.231268395091874e-06, + "loss": 0.1234, + "step": 12771 + }, + { + "epoch": 5.35, + "learning_rate": 5.229778246497446e-06, + "loss": 0.117, + "step": 12772 + }, + { + "epoch": 5.35, + "learning_rate": 5.2282882350188035e-06, + "loss": 0.1322, + "step": 12773 + }, + { + "epoch": 5.35, + "learning_rate": 5.226798360698786e-06, + "loss": 0.1263, + "step": 12774 + }, + { + "epoch": 5.35, + "learning_rate": 5.225308623580209e-06, + "loss": 0.1048, + "step": 12775 + }, + { + "epoch": 5.35, + "learning_rate": 5.2238190237059e-06, + "loss": 0.1328, + "step": 12776 + }, + { + "epoch": 5.35, + "learning_rate": 5.222329561118675e-06, + "loss": 0.1141, + "step": 12777 + }, + { + "epoch": 5.35, + "learning_rate": 5.2208402358613465e-06, + "loss": 0.1137, + "step": 12778 + }, + { + "epoch": 5.35, + "learning_rate": 5.219351047976725e-06, + "loss": 0.1288, + "step": 12779 + }, + { + "epoch": 5.35, + "learning_rate": 5.217861997507618e-06, + "loss": 0.1239, + "step": 12780 + }, + { + "epoch": 5.35, + "learning_rate": 5.216373084496822e-06, + "loss": 0.1355, + "step": 12781 + }, + { + "epoch": 5.35, + "learning_rate": 5.214884308987136e-06, + "loss": 0.1093, + "step": 12782 + }, + { + "epoch": 5.35, + "learning_rate": 5.213395671021355e-06, + "loss": 0.1434, + "step": 12783 + }, + { + "epoch": 5.35, + "learning_rate": 5.211907170642267e-06, + "loss": 0.1132, + "step": 12784 + }, + { + "epoch": 5.35, + "learning_rate": 5.210418807892659e-06, + "loss": 0.1423, + "step": 12785 + }, + { + "epoch": 5.35, + "learning_rate": 5.208930582815316e-06, + "loss": 0.1334, + "step": 12786 + }, + { + "epoch": 5.35, + "learning_rate": 5.207442495453008e-06, + "loss": 0.1015, + "step": 12787 + }, + { + "epoch": 5.35, + "learning_rate": 5.205954545848513e-06, + "loss": 0.1216, + "step": 12788 + }, + { + "epoch": 5.35, + "learning_rate": 5.2044667340445985e-06, + "loss": 0.1396, + "step": 12789 + }, + { + "epoch": 5.35, + "learning_rate": 5.202979060084035e-06, + "loss": 0.1208, + "step": 12790 + }, + { + "epoch": 5.36, + "learning_rate": 5.2014915240095785e-06, + "loss": 0.1006, + "step": 12791 + }, + { + "epoch": 5.36, + "learning_rate": 5.200004125863993e-06, + "loss": 0.0823, + "step": 12792 + }, + { + "epoch": 5.36, + "learning_rate": 5.198516865690031e-06, + "loss": 0.1439, + "step": 12793 + }, + { + "epoch": 5.36, + "learning_rate": 5.1970297435304395e-06, + "loss": 0.0832, + "step": 12794 + }, + { + "epoch": 5.36, + "learning_rate": 5.195542759427964e-06, + "loss": 0.1474, + "step": 12795 + }, + { + "epoch": 5.36, + "learning_rate": 5.194055913425353e-06, + "loss": 0.1134, + "step": 12796 + }, + { + "epoch": 5.36, + "learning_rate": 5.192569205565336e-06, + "loss": 0.1257, + "step": 12797 + }, + { + "epoch": 5.36, + "learning_rate": 5.191082635890652e-06, + "loss": 0.0989, + "step": 12798 + }, + { + "epoch": 5.36, + "learning_rate": 5.1895962044440295e-06, + "loss": 0.0935, + "step": 12799 + }, + { + "epoch": 5.36, + "learning_rate": 5.188109911268196e-06, + "loss": 0.1132, + "step": 12800 + }, + { + "epoch": 5.36, + "learning_rate": 5.186623756405875e-06, + "loss": 0.1229, + "step": 12801 + }, + { + "epoch": 5.36, + "learning_rate": 5.185137739899781e-06, + "loss": 0.1153, + "step": 12802 + }, + { + "epoch": 5.36, + "learning_rate": 5.183651861792629e-06, + "loss": 0.125, + "step": 12803 + }, + { + "epoch": 5.36, + "learning_rate": 5.18216612212713e-06, + "loss": 0.1001, + "step": 12804 + }, + { + "epoch": 5.36, + "learning_rate": 5.180680520945992e-06, + "loss": 0.1196, + "step": 12805 + }, + { + "epoch": 5.36, + "learning_rate": 5.179195058291915e-06, + "loss": 0.0949, + "step": 12806 + }, + { + "epoch": 5.36, + "learning_rate": 5.177709734207603e-06, + "loss": 0.1265, + "step": 12807 + }, + { + "epoch": 5.36, + "learning_rate": 5.17622454873574e-06, + "loss": 0.1384, + "step": 12808 + }, + { + "epoch": 5.36, + "learning_rate": 5.174739501919023e-06, + "loss": 0.1639, + "step": 12809 + }, + { + "epoch": 5.36, + "learning_rate": 5.173254593800137e-06, + "loss": 0.1167, + "step": 12810 + }, + { + "epoch": 5.36, + "learning_rate": 5.171769824421764e-06, + "loss": 0.1537, + "step": 12811 + }, + { + "epoch": 5.36, + "learning_rate": 5.170285193826583e-06, + "loss": 0.1342, + "step": 12812 + }, + { + "epoch": 5.36, + "learning_rate": 5.168800702057272e-06, + "loss": 0.1059, + "step": 12813 + }, + { + "epoch": 5.36, + "learning_rate": 5.167316349156495e-06, + "loss": 0.0931, + "step": 12814 + }, + { + "epoch": 5.37, + "learning_rate": 5.16583213516692e-06, + "loss": 0.1204, + "step": 12815 + }, + { + "epoch": 5.37, + "learning_rate": 5.16434806013121e-06, + "loss": 0.1225, + "step": 12816 + }, + { + "epoch": 5.37, + "learning_rate": 5.162864124092024e-06, + "loss": 0.0935, + "step": 12817 + }, + { + "epoch": 5.37, + "learning_rate": 5.161380327092021e-06, + "loss": 0.1368, + "step": 12818 + }, + { + "epoch": 5.37, + "learning_rate": 5.1598966691738425e-06, + "loss": 0.0978, + "step": 12819 + }, + { + "epoch": 5.37, + "learning_rate": 5.158413150380139e-06, + "loss": 0.1169, + "step": 12820 + }, + { + "epoch": 5.37, + "learning_rate": 5.1569297707535585e-06, + "loss": 0.1467, + "step": 12821 + }, + { + "epoch": 5.37, + "learning_rate": 5.155446530336729e-06, + "loss": 0.132, + "step": 12822 + }, + { + "epoch": 5.37, + "learning_rate": 5.15396342917229e-06, + "loss": 0.128, + "step": 12823 + }, + { + "epoch": 5.37, + "learning_rate": 5.1524804673028725e-06, + "loss": 0.1187, + "step": 12824 + }, + { + "epoch": 5.37, + "learning_rate": 5.150997644771102e-06, + "loss": 0.1133, + "step": 12825 + }, + { + "epoch": 5.37, + "learning_rate": 5.149514961619602e-06, + "loss": 0.1066, + "step": 12826 + }, + { + "epoch": 5.37, + "learning_rate": 5.148032417890992e-06, + "loss": 0.1473, + "step": 12827 + }, + { + "epoch": 5.37, + "learning_rate": 5.1465500136278825e-06, + "loss": 0.1187, + "step": 12828 + }, + { + "epoch": 5.37, + "learning_rate": 5.145067748872885e-06, + "loss": 0.1354, + "step": 12829 + }, + { + "epoch": 5.37, + "learning_rate": 5.143585623668607e-06, + "loss": 0.129, + "step": 12830 + }, + { + "epoch": 5.37, + "learning_rate": 5.142103638057652e-06, + "loss": 0.14, + "step": 12831 + }, + { + "epoch": 5.37, + "learning_rate": 5.140621792082616e-06, + "loss": 0.1078, + "step": 12832 + }, + { + "epoch": 5.37, + "learning_rate": 5.139140085786095e-06, + "loss": 0.1107, + "step": 12833 + }, + { + "epoch": 5.37, + "learning_rate": 5.1376585192106824e-06, + "loss": 0.1486, + "step": 12834 + }, + { + "epoch": 5.37, + "learning_rate": 5.136177092398957e-06, + "loss": 0.1297, + "step": 12835 + }, + { + "epoch": 5.37, + "learning_rate": 5.1346958053935045e-06, + "loss": 0.1109, + "step": 12836 + }, + { + "epoch": 5.37, + "learning_rate": 5.133214658236905e-06, + "loss": 0.0966, + "step": 12837 + }, + { + "epoch": 5.37, + "learning_rate": 5.131733650971731e-06, + "loss": 0.1232, + "step": 12838 + }, + { + "epoch": 5.38, + "learning_rate": 5.130252783640553e-06, + "loss": 0.1504, + "step": 12839 + }, + { + "epoch": 5.38, + "learning_rate": 5.12877205628594e-06, + "loss": 0.0922, + "step": 12840 + }, + { + "epoch": 5.38, + "learning_rate": 5.127291468950453e-06, + "loss": 0.1228, + "step": 12841 + }, + { + "epoch": 5.38, + "learning_rate": 5.125811021676643e-06, + "loss": 0.1426, + "step": 12842 + }, + { + "epoch": 5.38, + "learning_rate": 5.124330714507073e-06, + "loss": 0.0989, + "step": 12843 + }, + { + "epoch": 5.38, + "learning_rate": 5.1228505474842885e-06, + "loss": 0.1034, + "step": 12844 + }, + { + "epoch": 5.38, + "learning_rate": 5.121370520650837e-06, + "loss": 0.1275, + "step": 12845 + }, + { + "epoch": 5.38, + "learning_rate": 5.1198906340492605e-06, + "loss": 0.1123, + "step": 12846 + }, + { + "epoch": 5.38, + "learning_rate": 5.118410887722096e-06, + "loss": 0.11, + "step": 12847 + }, + { + "epoch": 5.38, + "learning_rate": 5.116931281711886e-06, + "loss": 0.143, + "step": 12848 + }, + { + "epoch": 5.38, + "learning_rate": 5.115451816061146e-06, + "loss": 0.1328, + "step": 12849 + }, + { + "epoch": 5.38, + "learning_rate": 5.113972490812411e-06, + "loss": 0.1434, + "step": 12850 + }, + { + "epoch": 5.38, + "learning_rate": 5.112493306008199e-06, + "loss": 0.1443, + "step": 12851 + }, + { + "epoch": 5.38, + "learning_rate": 5.111014261691032e-06, + "loss": 0.088, + "step": 12852 + }, + { + "epoch": 5.38, + "learning_rate": 5.109535357903421e-06, + "loss": 0.1044, + "step": 12853 + }, + { + "epoch": 5.38, + "learning_rate": 5.10805659468788e-06, + "loss": 0.1146, + "step": 12854 + }, + { + "epoch": 5.38, + "learning_rate": 5.106577972086907e-06, + "loss": 0.1258, + "step": 12855 + }, + { + "epoch": 5.38, + "learning_rate": 5.105099490143008e-06, + "loss": 0.1459, + "step": 12856 + }, + { + "epoch": 5.38, + "learning_rate": 5.1036211488986806e-06, + "loss": 0.1133, + "step": 12857 + }, + { + "epoch": 5.38, + "learning_rate": 5.102142948396418e-06, + "loss": 0.1259, + "step": 12858 + }, + { + "epoch": 5.38, + "learning_rate": 5.10066488867871e-06, + "loss": 0.0901, + "step": 12859 + }, + { + "epoch": 5.38, + "learning_rate": 5.099186969788044e-06, + "loss": 0.12, + "step": 12860 + }, + { + "epoch": 5.38, + "learning_rate": 5.097709191766902e-06, + "loss": 0.1217, + "step": 12861 + }, + { + "epoch": 5.38, + "learning_rate": 5.096231554657755e-06, + "loss": 0.1196, + "step": 12862 + }, + { + "epoch": 5.39, + "learning_rate": 5.094754058503081e-06, + "loss": 0.1184, + "step": 12863 + }, + { + "epoch": 5.39, + "learning_rate": 5.093276703345351e-06, + "loss": 0.1329, + "step": 12864 + }, + { + "epoch": 5.39, + "learning_rate": 5.091799489227027e-06, + "loss": 0.112, + "step": 12865 + }, + { + "epoch": 5.39, + "learning_rate": 5.09032241619057e-06, + "loss": 0.13, + "step": 12866 + }, + { + "epoch": 5.39, + "learning_rate": 5.088845484278437e-06, + "loss": 0.1355, + "step": 12867 + }, + { + "epoch": 5.39, + "learning_rate": 5.087368693533088e-06, + "loss": 0.1024, + "step": 12868 + }, + { + "epoch": 5.39, + "learning_rate": 5.085892043996962e-06, + "loss": 0.1276, + "step": 12869 + }, + { + "epoch": 5.39, + "learning_rate": 5.084415535712508e-06, + "loss": 0.0943, + "step": 12870 + }, + { + "epoch": 5.39, + "learning_rate": 5.082939168722168e-06, + "loss": 0.1246, + "step": 12871 + }, + { + "epoch": 5.39, + "learning_rate": 5.081462943068378e-06, + "loss": 0.1012, + "step": 12872 + }, + { + "epoch": 5.39, + "learning_rate": 5.079986858793572e-06, + "loss": 0.1261, + "step": 12873 + }, + { + "epoch": 5.39, + "learning_rate": 5.078510915940178e-06, + "loss": 0.1254, + "step": 12874 + }, + { + "epoch": 5.39, + "learning_rate": 5.077035114550624e-06, + "loss": 0.1195, + "step": 12875 + }, + { + "epoch": 5.39, + "learning_rate": 5.075559454667321e-06, + "loss": 0.1738, + "step": 12876 + }, + { + "epoch": 5.39, + "learning_rate": 5.074083936332695e-06, + "loss": 0.132, + "step": 12877 + }, + { + "epoch": 5.39, + "learning_rate": 5.072608559589152e-06, + "loss": 0.1303, + "step": 12878 + }, + { + "epoch": 5.39, + "learning_rate": 5.071133324479105e-06, + "loss": 0.1372, + "step": 12879 + }, + { + "epoch": 5.39, + "learning_rate": 5.069658231044956e-06, + "loss": 0.1013, + "step": 12880 + }, + { + "epoch": 5.39, + "learning_rate": 5.0681832793291105e-06, + "loss": 0.0985, + "step": 12881 + }, + { + "epoch": 5.39, + "learning_rate": 5.066708469373958e-06, + "loss": 0.0946, + "step": 12882 + }, + { + "epoch": 5.39, + "learning_rate": 5.065233801221891e-06, + "loss": 0.1635, + "step": 12883 + }, + { + "epoch": 5.39, + "learning_rate": 5.063759274915299e-06, + "loss": 0.1048, + "step": 12884 + }, + { + "epoch": 5.39, + "learning_rate": 5.062284890496567e-06, + "loss": 0.1156, + "step": 12885 + }, + { + "epoch": 5.4, + "learning_rate": 5.060810648008078e-06, + "loss": 0.0879, + "step": 12886 + }, + { + "epoch": 5.4, + "learning_rate": 5.0593365474921995e-06, + "loss": 0.0964, + "step": 12887 + }, + { + "epoch": 5.4, + "learning_rate": 5.057862588991308e-06, + "loss": 0.1302, + "step": 12888 + }, + { + "epoch": 5.4, + "learning_rate": 5.056388772547775e-06, + "loss": 0.1282, + "step": 12889 + }, + { + "epoch": 5.4, + "learning_rate": 5.054915098203956e-06, + "loss": 0.086, + "step": 12890 + }, + { + "epoch": 5.4, + "learning_rate": 5.053441566002214e-06, + "loss": 0.1227, + "step": 12891 + }, + { + "epoch": 5.4, + "learning_rate": 5.051968175984905e-06, + "loss": 0.1448, + "step": 12892 + }, + { + "epoch": 5.4, + "learning_rate": 5.050494928194379e-06, + "loss": 0.1208, + "step": 12893 + }, + { + "epoch": 5.4, + "learning_rate": 5.0490218226729855e-06, + "loss": 0.1248, + "step": 12894 + }, + { + "epoch": 5.4, + "learning_rate": 5.047548859463069e-06, + "loss": 0.1448, + "step": 12895 + }, + { + "epoch": 5.4, + "learning_rate": 5.046076038606964e-06, + "loss": 0.1254, + "step": 12896 + }, + { + "epoch": 5.4, + "learning_rate": 5.0446033601470045e-06, + "loss": 0.1221, + "step": 12897 + }, + { + "epoch": 5.4, + "learning_rate": 5.043130824125526e-06, + "loss": 0.1134, + "step": 12898 + }, + { + "epoch": 5.4, + "learning_rate": 5.041658430584852e-06, + "loss": 0.1405, + "step": 12899 + }, + { + "epoch": 5.4, + "learning_rate": 5.040186179567308e-06, + "loss": 0.1081, + "step": 12900 + }, + { + "epoch": 5.4, + "learning_rate": 5.03871407111521e-06, + "loss": 0.137, + "step": 12901 + }, + { + "epoch": 5.4, + "learning_rate": 5.037242105270877e-06, + "loss": 0.151, + "step": 12902 + }, + { + "epoch": 5.4, + "learning_rate": 5.035770282076613e-06, + "loss": 0.1329, + "step": 12903 + }, + { + "epoch": 5.4, + "learning_rate": 5.034298601574726e-06, + "loss": 0.117, + "step": 12904 + }, + { + "epoch": 5.4, + "learning_rate": 5.03282706380752e-06, + "loss": 0.091, + "step": 12905 + }, + { + "epoch": 5.4, + "learning_rate": 5.031355668817292e-06, + "loss": 0.1177, + "step": 12906 + }, + { + "epoch": 5.4, + "learning_rate": 5.029884416646335e-06, + "loss": 0.1754, + "step": 12907 + }, + { + "epoch": 5.4, + "learning_rate": 5.0284133073369456e-06, + "loss": 0.122, + "step": 12908 + }, + { + "epoch": 5.4, + "learning_rate": 5.026942340931404e-06, + "loss": 0.1711, + "step": 12909 + }, + { + "epoch": 5.41, + "learning_rate": 5.025471517471987e-06, + "loss": 0.1217, + "step": 12910 + }, + { + "epoch": 5.41, + "learning_rate": 5.0240008370009775e-06, + "loss": 0.1192, + "step": 12911 + }, + { + "epoch": 5.41, + "learning_rate": 5.022530299560648e-06, + "loss": 0.1085, + "step": 12912 + }, + { + "epoch": 5.41, + "learning_rate": 5.021059905193268e-06, + "loss": 0.1179, + "step": 12913 + }, + { + "epoch": 5.41, + "learning_rate": 5.019589653941103e-06, + "loss": 0.1086, + "step": 12914 + }, + { + "epoch": 5.41, + "learning_rate": 5.018119545846414e-06, + "loss": 0.1005, + "step": 12915 + }, + { + "epoch": 5.41, + "learning_rate": 5.016649580951462e-06, + "loss": 0.0988, + "step": 12916 + }, + { + "epoch": 5.41, + "learning_rate": 5.015179759298492e-06, + "loss": 0.135, + "step": 12917 + }, + { + "epoch": 5.41, + "learning_rate": 5.013710080929754e-06, + "loss": 0.1036, + "step": 12918 + }, + { + "epoch": 5.41, + "learning_rate": 5.0122405458874975e-06, + "loss": 0.1485, + "step": 12919 + }, + { + "epoch": 5.41, + "learning_rate": 5.01077115421396e-06, + "loss": 0.0948, + "step": 12920 + }, + { + "epoch": 5.41, + "learning_rate": 5.009301905951378e-06, + "loss": 0.0954, + "step": 12921 + }, + { + "epoch": 5.41, + "learning_rate": 5.007832801141988e-06, + "loss": 0.1289, + "step": 12922 + }, + { + "epoch": 5.41, + "learning_rate": 5.00636383982801e-06, + "loss": 0.1359, + "step": 12923 + }, + { + "epoch": 5.41, + "learning_rate": 5.0048950220516725e-06, + "loss": 0.1199, + "step": 12924 + }, + { + "epoch": 5.41, + "learning_rate": 5.0034263478551935e-06, + "loss": 0.1447, + "step": 12925 + }, + { + "epoch": 5.41, + "learning_rate": 5.001957817280792e-06, + "loss": 0.1321, + "step": 12926 + }, + { + "epoch": 5.41, + "learning_rate": 5.0004894303706775e-06, + "loss": 0.116, + "step": 12927 + }, + { + "epoch": 5.41, + "learning_rate": 4.999021187167061e-06, + "loss": 0.085, + "step": 12928 + }, + { + "epoch": 5.41, + "learning_rate": 4.99755308771214e-06, + "loss": 0.1256, + "step": 12929 + }, + { + "epoch": 5.41, + "learning_rate": 4.996085132048116e-06, + "loss": 0.0945, + "step": 12930 + }, + { + "epoch": 5.41, + "learning_rate": 4.994617320217184e-06, + "loss": 0.1253, + "step": 12931 + }, + { + "epoch": 5.41, + "learning_rate": 4.99314965226154e-06, + "loss": 0.1201, + "step": 12932 + }, + { + "epoch": 5.41, + "learning_rate": 4.991682128223362e-06, + "loss": 0.1346, + "step": 12933 + }, + { + "epoch": 5.42, + "learning_rate": 4.990214748144837e-06, + "loss": 0.1149, + "step": 12934 + }, + { + "epoch": 5.42, + "learning_rate": 4.988747512068143e-06, + "loss": 0.1218, + "step": 12935 + }, + { + "epoch": 5.42, + "learning_rate": 4.9872804200354596e-06, + "loss": 0.0837, + "step": 12936 + }, + { + "epoch": 5.42, + "learning_rate": 4.985813472088947e-06, + "loss": 0.1447, + "step": 12937 + }, + { + "epoch": 5.42, + "learning_rate": 4.984346668270779e-06, + "loss": 0.0828, + "step": 12938 + }, + { + "epoch": 5.42, + "learning_rate": 4.9828800086231135e-06, + "loss": 0.1073, + "step": 12939 + }, + { + "epoch": 5.42, + "learning_rate": 4.98141349318811e-06, + "loss": 0.1124, + "step": 12940 + }, + { + "epoch": 5.42, + "learning_rate": 4.979947122007922e-06, + "loss": 0.115, + "step": 12941 + }, + { + "epoch": 5.42, + "learning_rate": 4.9784808951247e-06, + "loss": 0.0887, + "step": 12942 + }, + { + "epoch": 5.42, + "learning_rate": 4.977014812580593e-06, + "loss": 0.0724, + "step": 12943 + }, + { + "epoch": 5.42, + "learning_rate": 4.975548874417733e-06, + "loss": 0.126, + "step": 12944 + }, + { + "epoch": 5.42, + "learning_rate": 4.9740830806782625e-06, + "loss": 0.1083, + "step": 12945 + }, + { + "epoch": 5.42, + "learning_rate": 4.972617431404313e-06, + "loss": 0.0955, + "step": 12946 + }, + { + "epoch": 5.42, + "learning_rate": 4.971151926638016e-06, + "loss": 0.1092, + "step": 12947 + }, + { + "epoch": 5.42, + "learning_rate": 4.9696865664214936e-06, + "loss": 0.1395, + "step": 12948 + }, + { + "epoch": 5.42, + "learning_rate": 4.96822135079687e-06, + "loss": 0.1324, + "step": 12949 + }, + { + "epoch": 5.42, + "learning_rate": 4.966756279806255e-06, + "loss": 0.0848, + "step": 12950 + }, + { + "epoch": 5.42, + "learning_rate": 4.9652913534917645e-06, + "loss": 0.1021, + "step": 12951 + }, + { + "epoch": 5.42, + "learning_rate": 4.963826571895507e-06, + "loss": 0.11, + "step": 12952 + }, + { + "epoch": 5.42, + "learning_rate": 4.962361935059585e-06, + "loss": 0.1254, + "step": 12953 + }, + { + "epoch": 5.42, + "learning_rate": 4.9608974430261025e-06, + "loss": 0.1299, + "step": 12954 + }, + { + "epoch": 5.42, + "learning_rate": 4.959433095837149e-06, + "loss": 0.1142, + "step": 12955 + }, + { + "epoch": 5.42, + "learning_rate": 4.957968893534816e-06, + "loss": 0.1018, + "step": 12956 + }, + { + "epoch": 5.42, + "learning_rate": 4.956504836161198e-06, + "loss": 0.1073, + "step": 12957 + }, + { + "epoch": 5.43, + "learning_rate": 4.955040923758368e-06, + "loss": 0.1269, + "step": 12958 + }, + { + "epoch": 5.43, + "learning_rate": 4.953577156368412e-06, + "loss": 0.1611, + "step": 12959 + }, + { + "epoch": 5.43, + "learning_rate": 4.952113534033402e-06, + "loss": 0.1238, + "step": 12960 + }, + { + "epoch": 5.43, + "learning_rate": 4.950650056795407e-06, + "loss": 0.122, + "step": 12961 + }, + { + "epoch": 5.43, + "learning_rate": 4.949186724696498e-06, + "loss": 0.1164, + "step": 12962 + }, + { + "epoch": 5.43, + "learning_rate": 4.947723537778737e-06, + "loss": 0.1309, + "step": 12963 + }, + { + "epoch": 5.43, + "learning_rate": 4.946260496084175e-06, + "loss": 0.135, + "step": 12964 + }, + { + "epoch": 5.43, + "learning_rate": 4.9447975996548725e-06, + "loss": 0.1494, + "step": 12965 + }, + { + "epoch": 5.43, + "learning_rate": 4.943334848532876e-06, + "loss": 0.1116, + "step": 12966 + }, + { + "epoch": 5.43, + "learning_rate": 4.9418722427602315e-06, + "loss": 0.126, + "step": 12967 + }, + { + "epoch": 5.43, + "learning_rate": 4.940409782378982e-06, + "loss": 0.1028, + "step": 12968 + }, + { + "epoch": 5.43, + "learning_rate": 4.938947467431167e-06, + "loss": 0.1234, + "step": 12969 + }, + { + "epoch": 5.43, + "learning_rate": 4.937485297958812e-06, + "loss": 0.0987, + "step": 12970 + }, + { + "epoch": 5.43, + "learning_rate": 4.936023274003949e-06, + "loss": 0.1385, + "step": 12971 + }, + { + "epoch": 5.43, + "learning_rate": 4.934561395608603e-06, + "loss": 0.1048, + "step": 12972 + }, + { + "epoch": 5.43, + "learning_rate": 4.933099662814797e-06, + "loss": 0.1374, + "step": 12973 + }, + { + "epoch": 5.43, + "learning_rate": 4.931638075664544e-06, + "loss": 0.1106, + "step": 12974 + }, + { + "epoch": 5.43, + "learning_rate": 4.930176634199856e-06, + "loss": 0.166, + "step": 12975 + }, + { + "epoch": 5.43, + "learning_rate": 4.928715338462746e-06, + "loss": 0.1257, + "step": 12976 + }, + { + "epoch": 5.43, + "learning_rate": 4.927254188495213e-06, + "loss": 0.1514, + "step": 12977 + }, + { + "epoch": 5.43, + "learning_rate": 4.925793184339254e-06, + "loss": 0.1367, + "step": 12978 + }, + { + "epoch": 5.43, + "learning_rate": 4.924332326036867e-06, + "loss": 0.1215, + "step": 12979 + }, + { + "epoch": 5.43, + "learning_rate": 4.922871613630043e-06, + "loss": 0.0959, + "step": 12980 + }, + { + "epoch": 5.43, + "learning_rate": 4.921411047160768e-06, + "loss": 0.1108, + "step": 12981 + }, + { + "epoch": 5.44, + "learning_rate": 4.919950626671028e-06, + "loss": 0.0922, + "step": 12982 + }, + { + "epoch": 5.44, + "learning_rate": 4.918490352202797e-06, + "loss": 0.1245, + "step": 12983 + }, + { + "epoch": 5.44, + "learning_rate": 4.917030223798057e-06, + "loss": 0.1117, + "step": 12984 + }, + { + "epoch": 5.44, + "learning_rate": 4.9155702414987685e-06, + "loss": 0.098, + "step": 12985 + }, + { + "epoch": 5.44, + "learning_rate": 4.914110405346901e-06, + "loss": 0.0783, + "step": 12986 + }, + { + "epoch": 5.44, + "learning_rate": 4.912650715384417e-06, + "loss": 0.1356, + "step": 12987 + }, + { + "epoch": 5.44, + "learning_rate": 4.911191171653273e-06, + "loss": 0.0962, + "step": 12988 + }, + { + "epoch": 5.44, + "learning_rate": 4.909731774195425e-06, + "loss": 0.088, + "step": 12989 + }, + { + "epoch": 5.44, + "learning_rate": 4.908272523052823e-06, + "loss": 0.1381, + "step": 12990 + }, + { + "epoch": 5.44, + "learning_rate": 4.9068134182674044e-06, + "loss": 0.0844, + "step": 12991 + }, + { + "epoch": 5.44, + "learning_rate": 4.905354459881116e-06, + "loss": 0.1077, + "step": 12992 + }, + { + "epoch": 5.44, + "learning_rate": 4.9038956479358915e-06, + "loss": 0.1185, + "step": 12993 + }, + { + "epoch": 5.44, + "learning_rate": 4.9024369824736654e-06, + "loss": 0.1349, + "step": 12994 + }, + { + "epoch": 5.44, + "learning_rate": 4.900978463536363e-06, + "loss": 0.1354, + "step": 12995 + }, + { + "epoch": 5.44, + "learning_rate": 4.8995200911659155e-06, + "loss": 0.0975, + "step": 12996 + }, + { + "epoch": 5.44, + "learning_rate": 4.898061865404232e-06, + "loss": 0.1147, + "step": 12997 + }, + { + "epoch": 5.44, + "learning_rate": 4.896603786293233e-06, + "loss": 0.1106, + "step": 12998 + }, + { + "epoch": 5.44, + "learning_rate": 4.89514585387483e-06, + "loss": 0.0958, + "step": 12999 + }, + { + "epoch": 5.44, + "learning_rate": 4.893688068190933e-06, + "loss": 0.0883, + "step": 13000 + }, + { + "epoch": 5.44, + "learning_rate": 4.892230429283438e-06, + "loss": 0.0716, + "step": 13001 + }, + { + "epoch": 5.44, + "learning_rate": 4.890772937194246e-06, + "loss": 0.1229, + "step": 13002 + }, + { + "epoch": 5.44, + "learning_rate": 4.889315591965251e-06, + "loss": 0.1199, + "step": 13003 + }, + { + "epoch": 5.44, + "learning_rate": 4.88785839363835e-06, + "loss": 0.1117, + "step": 13004 + }, + { + "epoch": 5.44, + "learning_rate": 4.886401342255418e-06, + "loss": 0.1406, + "step": 13005 + }, + { + "epoch": 5.45, + "learning_rate": 4.884944437858342e-06, + "loss": 0.1029, + "step": 13006 + }, + { + "epoch": 5.45, + "learning_rate": 4.883487680489e-06, + "loss": 0.129, + "step": 13007 + }, + { + "epoch": 5.45, + "learning_rate": 4.882031070189262e-06, + "loss": 0.1123, + "step": 13008 + }, + { + "epoch": 5.45, + "learning_rate": 4.880574607001001e-06, + "loss": 0.1319, + "step": 13009 + }, + { + "epoch": 5.45, + "learning_rate": 4.879118290966085e-06, + "loss": 0.0958, + "step": 13010 + }, + { + "epoch": 5.45, + "learning_rate": 4.877662122126364e-06, + "loss": 0.1183, + "step": 13011 + }, + { + "epoch": 5.45, + "learning_rate": 4.876206100523701e-06, + "loss": 0.1054, + "step": 13012 + }, + { + "epoch": 5.45, + "learning_rate": 4.874750226199945e-06, + "loss": 0.0947, + "step": 13013 + }, + { + "epoch": 5.45, + "learning_rate": 4.873294499196946e-06, + "loss": 0.0976, + "step": 13014 + }, + { + "epoch": 5.45, + "learning_rate": 4.871838919556549e-06, + "loss": 0.1035, + "step": 13015 + }, + { + "epoch": 5.45, + "learning_rate": 4.87038348732059e-06, + "loss": 0.1318, + "step": 13016 + }, + { + "epoch": 5.45, + "learning_rate": 4.868928202530911e-06, + "loss": 0.1104, + "step": 13017 + }, + { + "epoch": 5.45, + "learning_rate": 4.867473065229332e-06, + "loss": 0.1334, + "step": 13018 + }, + { + "epoch": 5.45, + "learning_rate": 4.866018075457687e-06, + "loss": 0.0944, + "step": 13019 + }, + { + "epoch": 5.45, + "learning_rate": 4.864563233257795e-06, + "loss": 0.1089, + "step": 13020 + }, + { + "epoch": 5.45, + "learning_rate": 4.863108538671477e-06, + "loss": 0.1105, + "step": 13021 + }, + { + "epoch": 5.45, + "learning_rate": 4.861653991740549e-06, + "loss": 0.0989, + "step": 13022 + }, + { + "epoch": 5.45, + "learning_rate": 4.8601995925068136e-06, + "loss": 0.1159, + "step": 13023 + }, + { + "epoch": 5.45, + "learning_rate": 4.85874534101208e-06, + "loss": 0.1002, + "step": 13024 + }, + { + "epoch": 5.45, + "learning_rate": 4.8572912372981526e-06, + "loss": 0.1181, + "step": 13025 + }, + { + "epoch": 5.45, + "learning_rate": 4.855837281406821e-06, + "loss": 0.1238, + "step": 13026 + }, + { + "epoch": 5.45, + "learning_rate": 4.8543834733798826e-06, + "loss": 0.1285, + "step": 13027 + }, + { + "epoch": 5.45, + "learning_rate": 4.852929813259125e-06, + "loss": 0.1352, + "step": 13028 + }, + { + "epoch": 5.45, + "learning_rate": 4.851476301086333e-06, + "loss": 0.1426, + "step": 13029 + }, + { + "epoch": 5.46, + "learning_rate": 4.850022936903285e-06, + "loss": 0.1417, + "step": 13030 + }, + { + "epoch": 5.46, + "learning_rate": 4.848569720751762e-06, + "loss": 0.1086, + "step": 13031 + }, + { + "epoch": 5.46, + "learning_rate": 4.847116652673527e-06, + "loss": 0.128, + "step": 13032 + }, + { + "epoch": 5.46, + "learning_rate": 4.845663732710352e-06, + "loss": 0.1113, + "step": 13033 + }, + { + "epoch": 5.46, + "learning_rate": 4.844210960903998e-06, + "loss": 0.1092, + "step": 13034 + }, + { + "epoch": 5.46, + "learning_rate": 4.842758337296225e-06, + "loss": 0.0939, + "step": 13035 + }, + { + "epoch": 5.46, + "learning_rate": 4.841305861928787e-06, + "loss": 0.1224, + "step": 13036 + }, + { + "epoch": 5.46, + "learning_rate": 4.839853534843439e-06, + "loss": 0.131, + "step": 13037 + }, + { + "epoch": 5.46, + "learning_rate": 4.838401356081918e-06, + "loss": 0.1202, + "step": 13038 + }, + { + "epoch": 5.46, + "learning_rate": 4.83694932568597e-06, + "loss": 0.0958, + "step": 13039 + }, + { + "epoch": 5.46, + "learning_rate": 4.835497443697332e-06, + "loss": 0.1044, + "step": 13040 + }, + { + "epoch": 5.46, + "learning_rate": 4.834045710157735e-06, + "loss": 0.1014, + "step": 13041 + }, + { + "epoch": 5.46, + "learning_rate": 4.832594125108912e-06, + "loss": 0.1027, + "step": 13042 + }, + { + "epoch": 5.46, + "learning_rate": 4.831142688592585e-06, + "loss": 0.1258, + "step": 13043 + }, + { + "epoch": 5.46, + "learning_rate": 4.829691400650478e-06, + "loss": 0.1244, + "step": 13044 + }, + { + "epoch": 5.46, + "learning_rate": 4.828240261324304e-06, + "loss": 0.0685, + "step": 13045 + }, + { + "epoch": 5.46, + "learning_rate": 4.826789270655771e-06, + "loss": 0.1399, + "step": 13046 + }, + { + "epoch": 5.46, + "learning_rate": 4.825338428686589e-06, + "loss": 0.1374, + "step": 13047 + }, + { + "epoch": 5.46, + "learning_rate": 4.823887735458462e-06, + "loss": 0.1102, + "step": 13048 + }, + { + "epoch": 5.46, + "learning_rate": 4.82243719101309e-06, + "loss": 0.1116, + "step": 13049 + }, + { + "epoch": 5.46, + "learning_rate": 4.820986795392165e-06, + "loss": 0.1159, + "step": 13050 + }, + { + "epoch": 5.46, + "learning_rate": 4.819536548637383e-06, + "loss": 0.1404, + "step": 13051 + }, + { + "epoch": 5.46, + "learning_rate": 4.818086450790423e-06, + "loss": 0.1188, + "step": 13052 + }, + { + "epoch": 5.46, + "learning_rate": 4.816636501892968e-06, + "loss": 0.0973, + "step": 13053 + }, + { + "epoch": 5.47, + "learning_rate": 4.8151867019866985e-06, + "loss": 0.093, + "step": 13054 + }, + { + "epoch": 5.47, + "learning_rate": 4.813737051113286e-06, + "loss": 0.143, + "step": 13055 + }, + { + "epoch": 5.47, + "learning_rate": 4.812287549314399e-06, + "loss": 0.1674, + "step": 13056 + }, + { + "epoch": 5.47, + "learning_rate": 4.810838196631704e-06, + "loss": 0.1127, + "step": 13057 + }, + { + "epoch": 5.47, + "learning_rate": 4.809388993106864e-06, + "loss": 0.1456, + "step": 13058 + }, + { + "epoch": 5.47, + "learning_rate": 4.807939938781526e-06, + "loss": 0.1201, + "step": 13059 + }, + { + "epoch": 5.47, + "learning_rate": 4.806491033697348e-06, + "loss": 0.1025, + "step": 13060 + }, + { + "epoch": 5.47, + "learning_rate": 4.805042277895977e-06, + "loss": 0.1205, + "step": 13061 + }, + { + "epoch": 5.47, + "learning_rate": 4.803593671419056e-06, + "loss": 0.1222, + "step": 13062 + }, + { + "epoch": 5.47, + "learning_rate": 4.802145214308223e-06, + "loss": 0.1502, + "step": 13063 + }, + { + "epoch": 5.47, + "learning_rate": 4.800696906605117e-06, + "loss": 0.1467, + "step": 13064 + }, + { + "epoch": 5.47, + "learning_rate": 4.799248748351362e-06, + "loss": 0.1273, + "step": 13065 + }, + { + "epoch": 5.47, + "learning_rate": 4.797800739588585e-06, + "loss": 0.1242, + "step": 13066 + }, + { + "epoch": 5.47, + "learning_rate": 4.796352880358411e-06, + "loss": 0.1119, + "step": 13067 + }, + { + "epoch": 5.47, + "learning_rate": 4.794905170702458e-06, + "loss": 0.1283, + "step": 13068 + }, + { + "epoch": 5.47, + "learning_rate": 4.793457610662334e-06, + "loss": 0.1104, + "step": 13069 + }, + { + "epoch": 5.47, + "learning_rate": 4.792010200279651e-06, + "loss": 0.0905, + "step": 13070 + }, + { + "epoch": 5.47, + "learning_rate": 4.790562939596013e-06, + "loss": 0.0942, + "step": 13071 + }, + { + "epoch": 5.47, + "learning_rate": 4.789115828653025e-06, + "loss": 0.0938, + "step": 13072 + }, + { + "epoch": 5.47, + "learning_rate": 4.787668867492274e-06, + "loss": 0.112, + "step": 13073 + }, + { + "epoch": 5.47, + "learning_rate": 4.786222056155356e-06, + "loss": 0.1225, + "step": 13074 + }, + { + "epoch": 5.47, + "learning_rate": 4.784775394683859e-06, + "loss": 0.0973, + "step": 13075 + }, + { + "epoch": 5.47, + "learning_rate": 4.783328883119365e-06, + "loss": 0.1067, + "step": 13076 + }, + { + "epoch": 5.47, + "learning_rate": 4.781882521503452e-06, + "loss": 0.1221, + "step": 13077 + }, + { + "epoch": 5.48, + "learning_rate": 4.780436309877701e-06, + "loss": 0.0798, + "step": 13078 + }, + { + "epoch": 5.48, + "learning_rate": 4.778990248283671e-06, + "loss": 0.1351, + "step": 13079 + }, + { + "epoch": 5.48, + "learning_rate": 4.777544336762935e-06, + "loss": 0.1175, + "step": 13080 + }, + { + "epoch": 5.48, + "learning_rate": 4.776098575357052e-06, + "loss": 0.1549, + "step": 13081 + }, + { + "epoch": 5.48, + "learning_rate": 4.7746529641075785e-06, + "loss": 0.1547, + "step": 13082 + }, + { + "epoch": 5.48, + "learning_rate": 4.773207503056071e-06, + "loss": 0.0962, + "step": 13083 + }, + { + "epoch": 5.48, + "learning_rate": 4.771762192244075e-06, + "loss": 0.0888, + "step": 13084 + }, + { + "epoch": 5.48, + "learning_rate": 4.770317031713138e-06, + "loss": 0.1445, + "step": 13085 + }, + { + "epoch": 5.48, + "learning_rate": 4.768872021504795e-06, + "loss": 0.156, + "step": 13086 + }, + { + "epoch": 5.48, + "learning_rate": 4.7674271616605816e-06, + "loss": 0.1104, + "step": 13087 + }, + { + "epoch": 5.48, + "learning_rate": 4.765982452222032e-06, + "loss": 0.1365, + "step": 13088 + }, + { + "epoch": 5.48, + "learning_rate": 4.764537893230673e-06, + "loss": 0.1265, + "step": 13089 + }, + { + "epoch": 5.48, + "learning_rate": 4.76309348472803e-06, + "loss": 0.1443, + "step": 13090 + }, + { + "epoch": 5.48, + "learning_rate": 4.7616492267556134e-06, + "loss": 0.0837, + "step": 13091 + }, + { + "epoch": 5.48, + "learning_rate": 4.760205119354946e-06, + "loss": 0.1026, + "step": 13092 + }, + { + "epoch": 5.48, + "learning_rate": 4.758761162567528e-06, + "loss": 0.141, + "step": 13093 + }, + { + "epoch": 5.48, + "learning_rate": 4.757317356434869e-06, + "loss": 0.1503, + "step": 13094 + }, + { + "epoch": 5.48, + "learning_rate": 4.755873700998471e-06, + "loss": 0.1125, + "step": 13095 + }, + { + "epoch": 5.48, + "learning_rate": 4.75443019629983e-06, + "loss": 0.1134, + "step": 13096 + }, + { + "epoch": 5.48, + "learning_rate": 4.7529868423804385e-06, + "loss": 0.1306, + "step": 13097 + }, + { + "epoch": 5.48, + "learning_rate": 4.751543639281784e-06, + "loss": 0.1281, + "step": 13098 + }, + { + "epoch": 5.48, + "learning_rate": 4.750100587045353e-06, + "loss": 0.1129, + "step": 13099 + }, + { + "epoch": 5.48, + "learning_rate": 4.748657685712619e-06, + "loss": 0.11, + "step": 13100 + }, + { + "epoch": 5.49, + "learning_rate": 4.74721493532506e-06, + "loss": 0.1665, + "step": 13101 + }, + { + "epoch": 5.49, + "learning_rate": 4.745772335924146e-06, + "loss": 0.115, + "step": 13102 + }, + { + "epoch": 5.49, + "learning_rate": 4.744329887551343e-06, + "loss": 0.1227, + "step": 13103 + }, + { + "epoch": 5.49, + "learning_rate": 4.742887590248114e-06, + "loss": 0.1139, + "step": 13104 + }, + { + "epoch": 5.49, + "learning_rate": 4.741445444055921e-06, + "loss": 0.1154, + "step": 13105 + }, + { + "epoch": 5.49, + "learning_rate": 4.740003449016208e-06, + "loss": 0.0991, + "step": 13106 + }, + { + "epoch": 5.49, + "learning_rate": 4.7385616051704275e-06, + "loss": 0.1196, + "step": 13107 + }, + { + "epoch": 5.49, + "learning_rate": 4.737119912560025e-06, + "loss": 0.11, + "step": 13108 + }, + { + "epoch": 5.49, + "learning_rate": 4.7356783712264405e-06, + "loss": 0.117, + "step": 13109 + }, + { + "epoch": 5.49, + "learning_rate": 4.734236981211111e-06, + "loss": 0.0986, + "step": 13110 + }, + { + "epoch": 5.49, + "learning_rate": 4.73279574255547e-06, + "loss": 0.1196, + "step": 13111 + }, + { + "epoch": 5.49, + "learning_rate": 4.731354655300936e-06, + "loss": 0.0981, + "step": 13112 + }, + { + "epoch": 5.49, + "learning_rate": 4.729913719488942e-06, + "loss": 0.0829, + "step": 13113 + }, + { + "epoch": 5.49, + "learning_rate": 4.7284729351608985e-06, + "loss": 0.1425, + "step": 13114 + }, + { + "epoch": 5.49, + "learning_rate": 4.727032302358221e-06, + "loss": 0.1211, + "step": 13115 + }, + { + "epoch": 5.49, + "learning_rate": 4.725591821122322e-06, + "loss": 0.0995, + "step": 13116 + }, + { + "epoch": 5.49, + "learning_rate": 4.724151491494605e-06, + "loss": 0.1103, + "step": 13117 + }, + { + "epoch": 5.49, + "learning_rate": 4.722711313516473e-06, + "loss": 0.1047, + "step": 13118 + }, + { + "epoch": 5.49, + "learning_rate": 4.721271287229323e-06, + "loss": 0.1315, + "step": 13119 + }, + { + "epoch": 5.49, + "learning_rate": 4.7198314126745424e-06, + "loss": 0.1178, + "step": 13120 + }, + { + "epoch": 5.49, + "learning_rate": 4.7183916898935234e-06, + "loss": 0.1067, + "step": 13121 + }, + { + "epoch": 5.49, + "learning_rate": 4.716952118927648e-06, + "loss": 0.1286, + "step": 13122 + }, + { + "epoch": 5.49, + "learning_rate": 4.715512699818295e-06, + "loss": 0.103, + "step": 13123 + }, + { + "epoch": 5.49, + "learning_rate": 4.714073432606841e-06, + "loss": 0.1046, + "step": 13124 + }, + { + "epoch": 5.5, + "learning_rate": 4.712634317334655e-06, + "loss": 0.151, + "step": 13125 + }, + { + "epoch": 5.5, + "learning_rate": 4.711195354043109e-06, + "loss": 0.1192, + "step": 13126 + }, + { + "epoch": 5.5, + "learning_rate": 4.709756542773554e-06, + "loss": 0.125, + "step": 13127 + }, + { + "epoch": 5.5, + "learning_rate": 4.708317883567353e-06, + "loss": 0.1561, + "step": 13128 + }, + { + "epoch": 5.5, + "learning_rate": 4.706879376465858e-06, + "loss": 0.1347, + "step": 13129 + }, + { + "epoch": 5.5, + "learning_rate": 4.70544102151042e-06, + "loss": 0.1131, + "step": 13130 + }, + { + "epoch": 5.5, + "learning_rate": 4.7040028187423795e-06, + "loss": 0.0858, + "step": 13131 + }, + { + "epoch": 5.5, + "learning_rate": 4.702564768203083e-06, + "loss": 0.1196, + "step": 13132 + }, + { + "epoch": 5.5, + "learning_rate": 4.701126869933858e-06, + "loss": 0.1426, + "step": 13133 + }, + { + "epoch": 5.5, + "learning_rate": 4.699689123976039e-06, + "loss": 0.1101, + "step": 13134 + }, + { + "epoch": 5.5, + "learning_rate": 4.698251530370956e-06, + "loss": 0.1094, + "step": 13135 + }, + { + "epoch": 5.5, + "learning_rate": 4.696814089159924e-06, + "loss": 0.1387, + "step": 13136 + }, + { + "epoch": 5.5, + "learning_rate": 4.695376800384266e-06, + "loss": 0.1346, + "step": 13137 + }, + { + "epoch": 5.5, + "learning_rate": 4.6939396640852945e-06, + "loss": 0.1599, + "step": 13138 + }, + { + "epoch": 5.5, + "learning_rate": 4.6925026803043186e-06, + "loss": 0.1209, + "step": 13139 + }, + { + "epoch": 5.5, + "learning_rate": 4.691065849082647e-06, + "loss": 0.1277, + "step": 13140 + }, + { + "epoch": 5.5, + "learning_rate": 4.6896291704615725e-06, + "loss": 0.1685, + "step": 13141 + }, + { + "epoch": 5.5, + "learning_rate": 4.688192644482396e-06, + "loss": 0.1097, + "step": 13142 + }, + { + "epoch": 5.5, + "learning_rate": 4.686756271186407e-06, + "loss": 0.139, + "step": 13143 + }, + { + "epoch": 5.5, + "learning_rate": 4.6853200506148955e-06, + "loss": 0.1138, + "step": 13144 + }, + { + "epoch": 5.5, + "learning_rate": 4.683883982809142e-06, + "loss": 0.1214, + "step": 13145 + }, + { + "epoch": 5.5, + "learning_rate": 4.68244806781043e-06, + "loss": 0.1228, + "step": 13146 + }, + { + "epoch": 5.5, + "learning_rate": 4.681012305660026e-06, + "loss": 0.097, + "step": 13147 + }, + { + "epoch": 5.5, + "learning_rate": 4.679576696399203e-06, + "loss": 0.1354, + "step": 13148 + }, + { + "epoch": 5.51, + "learning_rate": 4.678141240069227e-06, + "loss": 0.1367, + "step": 13149 + }, + { + "epoch": 5.51, + "learning_rate": 4.676705936711358e-06, + "loss": 0.0956, + "step": 13150 + }, + { + "epoch": 5.51, + "learning_rate": 4.675270786366853e-06, + "loss": 0.1226, + "step": 13151 + }, + { + "epoch": 5.51, + "learning_rate": 4.673835789076969e-06, + "loss": 0.1573, + "step": 13152 + }, + { + "epoch": 5.51, + "learning_rate": 4.672400944882944e-06, + "loss": 0.1212, + "step": 13153 + }, + { + "epoch": 5.51, + "learning_rate": 4.670966253826027e-06, + "loss": 0.1723, + "step": 13154 + }, + { + "epoch": 5.51, + "learning_rate": 4.669531715947455e-06, + "loss": 0.1733, + "step": 13155 + }, + { + "epoch": 5.51, + "learning_rate": 4.668097331288465e-06, + "loss": 0.1056, + "step": 13156 + }, + { + "epoch": 5.51, + "learning_rate": 4.666663099890285e-06, + "loss": 0.088, + "step": 13157 + }, + { + "epoch": 5.51, + "learning_rate": 4.665229021794145e-06, + "loss": 0.1706, + "step": 13158 + }, + { + "epoch": 5.51, + "learning_rate": 4.66379509704126e-06, + "loss": 0.0982, + "step": 13159 + }, + { + "epoch": 5.51, + "learning_rate": 4.662361325672853e-06, + "loss": 0.1061, + "step": 13160 + }, + { + "epoch": 5.51, + "learning_rate": 4.660927707730129e-06, + "loss": 0.0863, + "step": 13161 + }, + { + "epoch": 5.51, + "learning_rate": 4.659494243254301e-06, + "loss": 0.1444, + "step": 13162 + }, + { + "epoch": 5.51, + "learning_rate": 4.658060932286571e-06, + "loss": 0.0896, + "step": 13163 + }, + { + "epoch": 5.51, + "learning_rate": 4.65662777486814e-06, + "loss": 0.1124, + "step": 13164 + }, + { + "epoch": 5.51, + "learning_rate": 4.6551947710402016e-06, + "loss": 0.13, + "step": 13165 + }, + { + "epoch": 5.51, + "learning_rate": 4.653761920843946e-06, + "loss": 0.1145, + "step": 13166 + }, + { + "epoch": 5.51, + "learning_rate": 4.652329224320564e-06, + "loss": 0.1004, + "step": 13167 + }, + { + "epoch": 5.51, + "learning_rate": 4.65089668151123e-06, + "loss": 0.0881, + "step": 13168 + }, + { + "epoch": 5.51, + "learning_rate": 4.649464292457123e-06, + "loss": 0.1377, + "step": 13169 + }, + { + "epoch": 5.51, + "learning_rate": 4.648032057199418e-06, + "loss": 0.107, + "step": 13170 + }, + { + "epoch": 5.51, + "learning_rate": 4.6465999757792815e-06, + "loss": 0.1474, + "step": 13171 + }, + { + "epoch": 5.51, + "learning_rate": 4.645168048237878e-06, + "loss": 0.145, + "step": 13172 + }, + { + "epoch": 5.52, + "learning_rate": 4.643736274616372e-06, + "loss": 0.1343, + "step": 13173 + }, + { + "epoch": 5.52, + "learning_rate": 4.6423046549559095e-06, + "loss": 0.1351, + "step": 13174 + }, + { + "epoch": 5.52, + "learning_rate": 4.640873189297645e-06, + "loss": 0.1167, + "step": 13175 + }, + { + "epoch": 5.52, + "learning_rate": 4.639441877682724e-06, + "loss": 0.1504, + "step": 13176 + }, + { + "epoch": 5.52, + "learning_rate": 4.638010720152292e-06, + "loss": 0.1185, + "step": 13177 + }, + { + "epoch": 5.52, + "learning_rate": 4.6365797167474815e-06, + "loss": 0.1556, + "step": 13178 + }, + { + "epoch": 5.52, + "learning_rate": 4.635148867509432e-06, + "loss": 0.1527, + "step": 13179 + }, + { + "epoch": 5.52, + "learning_rate": 4.6337181724792644e-06, + "loss": 0.1212, + "step": 13180 + }, + { + "epoch": 5.52, + "learning_rate": 4.6322876316981095e-06, + "loss": 0.1378, + "step": 13181 + }, + { + "epoch": 5.52, + "learning_rate": 4.630857245207079e-06, + "loss": 0.1086, + "step": 13182 + }, + { + "epoch": 5.52, + "learning_rate": 4.629427013047293e-06, + "loss": 0.1517, + "step": 13183 + }, + { + "epoch": 5.52, + "learning_rate": 4.627996935259862e-06, + "loss": 0.1087, + "step": 13184 + }, + { + "epoch": 5.52, + "learning_rate": 4.626567011885892e-06, + "loss": 0.1032, + "step": 13185 + }, + { + "epoch": 5.52, + "learning_rate": 4.625137242966484e-06, + "loss": 0.0961, + "step": 13186 + }, + { + "epoch": 5.52, + "learning_rate": 4.623707628542742e-06, + "loss": 0.1232, + "step": 13187 + }, + { + "epoch": 5.52, + "learning_rate": 4.6222781686557485e-06, + "loss": 0.1486, + "step": 13188 + }, + { + "epoch": 5.52, + "learning_rate": 4.620848863346596e-06, + "loss": 0.157, + "step": 13189 + }, + { + "epoch": 5.52, + "learning_rate": 4.6194197126563715e-06, + "loss": 0.1104, + "step": 13190 + }, + { + "epoch": 5.52, + "learning_rate": 4.617990716626151e-06, + "loss": 0.0985, + "step": 13191 + }, + { + "epoch": 5.52, + "learning_rate": 4.616561875297012e-06, + "loss": 0.0959, + "step": 13192 + }, + { + "epoch": 5.52, + "learning_rate": 4.615133188710029e-06, + "loss": 0.0781, + "step": 13193 + }, + { + "epoch": 5.52, + "learning_rate": 4.6137046569062595e-06, + "loss": 0.1041, + "step": 13194 + }, + { + "epoch": 5.52, + "learning_rate": 4.612276279926771e-06, + "loss": 0.1304, + "step": 13195 + }, + { + "epoch": 5.52, + "learning_rate": 4.610848057812619e-06, + "loss": 0.1292, + "step": 13196 + }, + { + "epoch": 5.53, + "learning_rate": 4.609419990604858e-06, + "loss": 0.1582, + "step": 13197 + }, + { + "epoch": 5.53, + "learning_rate": 4.607992078344534e-06, + "loss": 0.1267, + "step": 13198 + }, + { + "epoch": 5.53, + "learning_rate": 4.6065643210726955e-06, + "loss": 0.0792, + "step": 13199 + }, + { + "epoch": 5.53, + "learning_rate": 4.605136718830382e-06, + "loss": 0.1392, + "step": 13200 + }, + { + "epoch": 5.53, + "learning_rate": 4.603709271658621e-06, + "loss": 0.1323, + "step": 13201 + }, + { + "epoch": 5.53, + "learning_rate": 4.602281979598451e-06, + "loss": 0.1288, + "step": 13202 + }, + { + "epoch": 5.53, + "learning_rate": 4.600854842690898e-06, + "loss": 0.1464, + "step": 13203 + }, + { + "epoch": 5.53, + "learning_rate": 4.599427860976978e-06, + "loss": 0.092, + "step": 13204 + }, + { + "epoch": 5.53, + "learning_rate": 4.598001034497712e-06, + "loss": 0.1516, + "step": 13205 + }, + { + "epoch": 5.53, + "learning_rate": 4.596574363294113e-06, + "loss": 0.1023, + "step": 13206 + }, + { + "epoch": 5.53, + "learning_rate": 4.595147847407187e-06, + "loss": 0.173, + "step": 13207 + }, + { + "epoch": 5.53, + "learning_rate": 4.593721486877945e-06, + "loss": 0.1068, + "step": 13208 + }, + { + "epoch": 5.53, + "learning_rate": 4.592295281747379e-06, + "loss": 0.1424, + "step": 13209 + }, + { + "epoch": 5.53, + "learning_rate": 4.590869232056485e-06, + "loss": 0.1042, + "step": 13210 + }, + { + "epoch": 5.53, + "learning_rate": 4.589443337846256e-06, + "loss": 0.1413, + "step": 13211 + }, + { + "epoch": 5.53, + "learning_rate": 4.588017599157675e-06, + "loss": 0.1487, + "step": 13212 + }, + { + "epoch": 5.53, + "learning_rate": 4.5865920160317266e-06, + "loss": 0.1182, + "step": 13213 + }, + { + "epoch": 5.53, + "learning_rate": 4.585166588509391e-06, + "loss": 0.1448, + "step": 13214 + }, + { + "epoch": 5.53, + "learning_rate": 4.5837413166316335e-06, + "loss": 0.1197, + "step": 13215 + }, + { + "epoch": 5.53, + "learning_rate": 4.582316200439425e-06, + "loss": 0.1316, + "step": 13216 + }, + { + "epoch": 5.53, + "learning_rate": 4.580891239973729e-06, + "loss": 0.1344, + "step": 13217 + }, + { + "epoch": 5.53, + "learning_rate": 4.579466435275506e-06, + "loss": 0.1287, + "step": 13218 + }, + { + "epoch": 5.53, + "learning_rate": 4.578041786385711e-06, + "loss": 0.1488, + "step": 13219 + }, + { + "epoch": 5.53, + "learning_rate": 4.576617293345296e-06, + "loss": 0.1474, + "step": 13220 + }, + { + "epoch": 5.54, + "learning_rate": 4.575192956195201e-06, + "loss": 0.1238, + "step": 13221 + }, + { + "epoch": 5.54, + "learning_rate": 4.573768774976371e-06, + "loss": 0.1045, + "step": 13222 + }, + { + "epoch": 5.54, + "learning_rate": 4.5723447497297424e-06, + "loss": 0.1188, + "step": 13223 + }, + { + "epoch": 5.54, + "learning_rate": 4.570920880496247e-06, + "loss": 0.1278, + "step": 13224 + }, + { + "epoch": 5.54, + "learning_rate": 4.569497167316813e-06, + "loss": 0.1062, + "step": 13225 + }, + { + "epoch": 5.54, + "learning_rate": 4.568073610232368e-06, + "loss": 0.1161, + "step": 13226 + }, + { + "epoch": 5.54, + "learning_rate": 4.566650209283824e-06, + "loss": 0.1178, + "step": 13227 + }, + { + "epoch": 5.54, + "learning_rate": 4.565226964512102e-06, + "loss": 0.1045, + "step": 13228 + }, + { + "epoch": 5.54, + "learning_rate": 4.563803875958103e-06, + "loss": 0.112, + "step": 13229 + }, + { + "epoch": 5.54, + "learning_rate": 4.562380943662739e-06, + "loss": 0.1019, + "step": 13230 + }, + { + "epoch": 5.54, + "learning_rate": 4.560958167666908e-06, + "loss": 0.1122, + "step": 13231 + }, + { + "epoch": 5.54, + "learning_rate": 4.5595355480115086e-06, + "loss": 0.1071, + "step": 13232 + }, + { + "epoch": 5.54, + "learning_rate": 4.558113084737434e-06, + "loss": 0.1029, + "step": 13233 + }, + { + "epoch": 5.54, + "learning_rate": 4.556690777885571e-06, + "loss": 0.1232, + "step": 13234 + }, + { + "epoch": 5.54, + "learning_rate": 4.5552686274967984e-06, + "loss": 0.1612, + "step": 13235 + }, + { + "epoch": 5.54, + "learning_rate": 4.553846633611997e-06, + "loss": 0.1484, + "step": 13236 + }, + { + "epoch": 5.54, + "learning_rate": 4.552424796272041e-06, + "loss": 0.072, + "step": 13237 + }, + { + "epoch": 5.54, + "learning_rate": 4.551003115517801e-06, + "loss": 0.1069, + "step": 13238 + }, + { + "epoch": 5.54, + "learning_rate": 4.549581591390142e-06, + "loss": 0.1268, + "step": 13239 + }, + { + "epoch": 5.54, + "learning_rate": 4.548160223929923e-06, + "loss": 0.099, + "step": 13240 + }, + { + "epoch": 5.54, + "learning_rate": 4.546739013178003e-06, + "loss": 0.1366, + "step": 13241 + }, + { + "epoch": 5.54, + "learning_rate": 4.545317959175228e-06, + "loss": 0.1154, + "step": 13242 + }, + { + "epoch": 5.54, + "learning_rate": 4.543897061962448e-06, + "loss": 0.1007, + "step": 13243 + }, + { + "epoch": 5.54, + "learning_rate": 4.542476321580505e-06, + "loss": 0.1371, + "step": 13244 + }, + { + "epoch": 5.55, + "learning_rate": 4.541055738070237e-06, + "loss": 0.1588, + "step": 13245 + }, + { + "epoch": 5.55, + "learning_rate": 4.539635311472478e-06, + "loss": 0.154, + "step": 13246 + }, + { + "epoch": 5.55, + "learning_rate": 4.53821504182806e-06, + "loss": 0.0952, + "step": 13247 + }, + { + "epoch": 5.55, + "learning_rate": 4.5367949291778e-06, + "loss": 0.1177, + "step": 13248 + }, + { + "epoch": 5.55, + "learning_rate": 4.535374973562526e-06, + "loss": 0.1216, + "step": 13249 + }, + { + "epoch": 5.55, + "learning_rate": 4.533955175023046e-06, + "loss": 0.1066, + "step": 13250 + }, + { + "epoch": 5.55, + "learning_rate": 4.532535533600173e-06, + "loss": 0.0986, + "step": 13251 + }, + { + "epoch": 5.55, + "learning_rate": 4.531116049334715e-06, + "loss": 0.1444, + "step": 13252 + }, + { + "epoch": 5.55, + "learning_rate": 4.5296967222674724e-06, + "loss": 0.11, + "step": 13253 + }, + { + "epoch": 5.55, + "learning_rate": 4.528277552439244e-06, + "loss": 0.1187, + "step": 13254 + }, + { + "epoch": 5.55, + "learning_rate": 4.526858539890825e-06, + "loss": 0.1006, + "step": 13255 + }, + { + "epoch": 5.55, + "learning_rate": 4.5254396846629975e-06, + "loss": 0.086, + "step": 13256 + }, + { + "epoch": 5.55, + "learning_rate": 4.524020986796548e-06, + "loss": 0.1032, + "step": 13257 + }, + { + "epoch": 5.55, + "learning_rate": 4.522602446332256e-06, + "loss": 0.1266, + "step": 13258 + }, + { + "epoch": 5.55, + "learning_rate": 4.521184063310895e-06, + "loss": 0.1116, + "step": 13259 + }, + { + "epoch": 5.55, + "learning_rate": 4.519765837773238e-06, + "loss": 0.1249, + "step": 13260 + }, + { + "epoch": 5.55, + "learning_rate": 4.518347769760051e-06, + "loss": 0.0959, + "step": 13261 + }, + { + "epoch": 5.55, + "learning_rate": 4.51692985931209e-06, + "loss": 0.1262, + "step": 13262 + }, + { + "epoch": 5.55, + "learning_rate": 4.515512106470115e-06, + "loss": 0.1021, + "step": 13263 + }, + { + "epoch": 5.55, + "learning_rate": 4.514094511274877e-06, + "loss": 0.1451, + "step": 13264 + }, + { + "epoch": 5.55, + "learning_rate": 4.512677073767124e-06, + "loss": 0.1179, + "step": 13265 + }, + { + "epoch": 5.55, + "learning_rate": 4.5112597939876e-06, + "loss": 0.1279, + "step": 13266 + }, + { + "epoch": 5.55, + "learning_rate": 4.509842671977043e-06, + "loss": 0.0967, + "step": 13267 + }, + { + "epoch": 5.55, + "learning_rate": 4.50842570777619e-06, + "loss": 0.0605, + "step": 13268 + }, + { + "epoch": 5.56, + "learning_rate": 4.507008901425762e-06, + "loss": 0.0834, + "step": 13269 + }, + { + "epoch": 5.56, + "learning_rate": 4.50559225296649e-06, + "loss": 0.1333, + "step": 13270 + }, + { + "epoch": 5.56, + "learning_rate": 4.504175762439096e-06, + "loss": 0.1459, + "step": 13271 + }, + { + "epoch": 5.56, + "learning_rate": 4.5027594298842884e-06, + "loss": 0.1377, + "step": 13272 + }, + { + "epoch": 5.56, + "learning_rate": 4.5013432553427835e-06, + "loss": 0.0817, + "step": 13273 + }, + { + "epoch": 5.56, + "learning_rate": 4.499927238855287e-06, + "loss": 0.1335, + "step": 13274 + }, + { + "epoch": 5.56, + "learning_rate": 4.498511380462506e-06, + "loss": 0.1125, + "step": 13275 + }, + { + "epoch": 5.56, + "learning_rate": 4.4970956802051295e-06, + "loss": 0.1297, + "step": 13276 + }, + { + "epoch": 5.56, + "learning_rate": 4.495680138123855e-06, + "loss": 0.1283, + "step": 13277 + }, + { + "epoch": 5.56, + "learning_rate": 4.494264754259368e-06, + "loss": 0.1291, + "step": 13278 + }, + { + "epoch": 5.56, + "learning_rate": 4.492849528652357e-06, + "loss": 0.1326, + "step": 13279 + }, + { + "epoch": 5.56, + "learning_rate": 4.491434461343499e-06, + "loss": 0.109, + "step": 13280 + }, + { + "epoch": 5.56, + "learning_rate": 4.49001955237347e-06, + "loss": 0.1538, + "step": 13281 + }, + { + "epoch": 5.56, + "learning_rate": 4.488604801782942e-06, + "loss": 0.1198, + "step": 13282 + }, + { + "epoch": 5.56, + "learning_rate": 4.4871902096125755e-06, + "loss": 0.1009, + "step": 13283 + }, + { + "epoch": 5.56, + "learning_rate": 4.485775775903033e-06, + "loss": 0.1616, + "step": 13284 + }, + { + "epoch": 5.56, + "learning_rate": 4.484361500694974e-06, + "loss": 0.0977, + "step": 13285 + }, + { + "epoch": 5.56, + "learning_rate": 4.482947384029048e-06, + "loss": 0.1267, + "step": 13286 + }, + { + "epoch": 5.56, + "learning_rate": 4.481533425945904e-06, + "loss": 0.153, + "step": 13287 + }, + { + "epoch": 5.56, + "learning_rate": 4.480119626486189e-06, + "loss": 0.143, + "step": 13288 + }, + { + "epoch": 5.56, + "learning_rate": 4.4787059856905334e-06, + "loss": 0.0814, + "step": 13289 + }, + { + "epoch": 5.56, + "learning_rate": 4.477292503599574e-06, + "loss": 0.1323, + "step": 13290 + }, + { + "epoch": 5.56, + "learning_rate": 4.475879180253941e-06, + "loss": 0.1213, + "step": 13291 + }, + { + "epoch": 5.56, + "learning_rate": 4.47446601569426e-06, + "loss": 0.1197, + "step": 13292 + }, + { + "epoch": 5.57, + "learning_rate": 4.473053009961149e-06, + "loss": 0.1439, + "step": 13293 + }, + { + "epoch": 5.57, + "learning_rate": 4.471640163095229e-06, + "loss": 0.147, + "step": 13294 + }, + { + "epoch": 5.57, + "learning_rate": 4.470227475137102e-06, + "loss": 0.0922, + "step": 13295 + }, + { + "epoch": 5.57, + "learning_rate": 4.4688149461273845e-06, + "loss": 0.1109, + "step": 13296 + }, + { + "epoch": 5.57, + "learning_rate": 4.467402576106668e-06, + "loss": 0.1183, + "step": 13297 + }, + { + "epoch": 5.57, + "learning_rate": 4.465990365115555e-06, + "loss": 0.0781, + "step": 13298 + }, + { + "epoch": 5.57, + "learning_rate": 4.46457831319464e-06, + "loss": 0.1439, + "step": 13299 + }, + { + "epoch": 5.57, + "learning_rate": 4.463166420384507e-06, + "loss": 0.1276, + "step": 13300 + }, + { + "epoch": 5.57, + "learning_rate": 4.461754686725743e-06, + "loss": 0.1152, + "step": 13301 + }, + { + "epoch": 5.57, + "learning_rate": 4.4603431122589295e-06, + "loss": 0.1117, + "step": 13302 + }, + { + "epoch": 5.57, + "learning_rate": 4.458931697024633e-06, + "loss": 0.1098, + "step": 13303 + }, + { + "epoch": 5.57, + "learning_rate": 4.457520441063428e-06, + "loss": 0.1016, + "step": 13304 + }, + { + "epoch": 5.57, + "learning_rate": 4.456109344415881e-06, + "loss": 0.1145, + "step": 13305 + }, + { + "epoch": 5.57, + "learning_rate": 4.454698407122551e-06, + "loss": 0.1367, + "step": 13306 + }, + { + "epoch": 5.57, + "learning_rate": 4.453287629223994e-06, + "loss": 0.1371, + "step": 13307 + }, + { + "epoch": 5.57, + "learning_rate": 4.4518770107607626e-06, + "loss": 0.0942, + "step": 13308 + }, + { + "epoch": 5.57, + "learning_rate": 4.450466551773406e-06, + "loss": 0.114, + "step": 13309 + }, + { + "epoch": 5.57, + "learning_rate": 4.449056252302461e-06, + "loss": 0.1576, + "step": 13310 + }, + { + "epoch": 5.57, + "learning_rate": 4.447646112388467e-06, + "loss": 0.1172, + "step": 13311 + }, + { + "epoch": 5.57, + "learning_rate": 4.44623613207196e-06, + "loss": 0.1272, + "step": 13312 + }, + { + "epoch": 5.57, + "learning_rate": 4.444826311393465e-06, + "loss": 0.1114, + "step": 13313 + }, + { + "epoch": 5.57, + "learning_rate": 4.443416650393509e-06, + "loss": 0.112, + "step": 13314 + }, + { + "epoch": 5.57, + "learning_rate": 4.442007149112615e-06, + "loss": 0.1217, + "step": 13315 + }, + { + "epoch": 5.58, + "learning_rate": 4.44059780759129e-06, + "loss": 0.1172, + "step": 13316 + }, + { + "epoch": 5.58, + "learning_rate": 4.43918862587005e-06, + "loss": 0.1354, + "step": 13317 + }, + { + "epoch": 5.58, + "learning_rate": 4.437779603989395e-06, + "loss": 0.1031, + "step": 13318 + }, + { + "epoch": 5.58, + "learning_rate": 4.436370741989829e-06, + "loss": 0.154, + "step": 13319 + }, + { + "epoch": 5.58, + "learning_rate": 4.4349620399118495e-06, + "loss": 0.1524, + "step": 13320 + }, + { + "epoch": 5.58, + "learning_rate": 4.43355349779595e-06, + "loss": 0.0977, + "step": 13321 + }, + { + "epoch": 5.58, + "learning_rate": 4.432145115682613e-06, + "loss": 0.1257, + "step": 13322 + }, + { + "epoch": 5.58, + "learning_rate": 4.430736893612327e-06, + "loss": 0.1262, + "step": 13323 + }, + { + "epoch": 5.58, + "learning_rate": 4.429328831625565e-06, + "loss": 0.0848, + "step": 13324 + }, + { + "epoch": 5.58, + "learning_rate": 4.427920929762803e-06, + "loss": 0.1323, + "step": 13325 + }, + { + "epoch": 5.58, + "learning_rate": 4.426513188064507e-06, + "loss": 0.1331, + "step": 13326 + }, + { + "epoch": 5.58, + "learning_rate": 4.425105606571145e-06, + "loss": 0.1298, + "step": 13327 + }, + { + "epoch": 5.58, + "learning_rate": 4.423698185323175e-06, + "loss": 0.1379, + "step": 13328 + }, + { + "epoch": 5.58, + "learning_rate": 4.422290924361056e-06, + "loss": 0.1413, + "step": 13329 + }, + { + "epoch": 5.58, + "learning_rate": 4.42088382372523e-06, + "loss": 0.086, + "step": 13330 + }, + { + "epoch": 5.58, + "learning_rate": 4.41947688345615e-06, + "loss": 0.1054, + "step": 13331 + }, + { + "epoch": 5.58, + "learning_rate": 4.418070103594253e-06, + "loss": 0.1087, + "step": 13332 + }, + { + "epoch": 5.58, + "learning_rate": 4.416663484179978e-06, + "loss": 0.138, + "step": 13333 + }, + { + "epoch": 5.58, + "learning_rate": 4.415257025253758e-06, + "loss": 0.1474, + "step": 13334 + }, + { + "epoch": 5.58, + "learning_rate": 4.41385072685602e-06, + "loss": 0.1176, + "step": 13335 + }, + { + "epoch": 5.58, + "learning_rate": 4.412444589027185e-06, + "loss": 0.1425, + "step": 13336 + }, + { + "epoch": 5.58, + "learning_rate": 4.41103861180767e-06, + "loss": 0.1142, + "step": 13337 + }, + { + "epoch": 5.58, + "learning_rate": 4.4096327952378915e-06, + "loss": 0.0912, + "step": 13338 + }, + { + "epoch": 5.58, + "learning_rate": 4.408227139358261e-06, + "loss": 0.1089, + "step": 13339 + }, + { + "epoch": 5.59, + "learning_rate": 4.406821644209175e-06, + "loss": 0.1501, + "step": 13340 + }, + { + "epoch": 5.59, + "learning_rate": 4.4054163098310375e-06, + "loss": 0.1016, + "step": 13341 + }, + { + "epoch": 5.59, + "learning_rate": 4.404011136264245e-06, + "loss": 0.0911, + "step": 13342 + }, + { + "epoch": 5.59, + "learning_rate": 4.402606123549189e-06, + "loss": 0.128, + "step": 13343 + }, + { + "epoch": 5.59, + "learning_rate": 4.40120127172625e-06, + "loss": 0.1497, + "step": 13344 + }, + { + "epoch": 5.59, + "learning_rate": 4.399796580835811e-06, + "loss": 0.1469, + "step": 13345 + }, + { + "epoch": 5.59, + "learning_rate": 4.398392050918251e-06, + "loss": 0.1182, + "step": 13346 + }, + { + "epoch": 5.59, + "learning_rate": 4.39698768201394e-06, + "loss": 0.1152, + "step": 13347 + }, + { + "epoch": 5.59, + "learning_rate": 4.395583474163246e-06, + "loss": 0.1244, + "step": 13348 + }, + { + "epoch": 5.59, + "learning_rate": 4.394179427406533e-06, + "loss": 0.103, + "step": 13349 + }, + { + "epoch": 5.59, + "learning_rate": 4.392775541784161e-06, + "loss": 0.1087, + "step": 13350 + }, + { + "epoch": 5.59, + "learning_rate": 4.391371817336476e-06, + "loss": 0.1513, + "step": 13351 + }, + { + "epoch": 5.59, + "learning_rate": 4.3899682541038316e-06, + "loss": 0.1423, + "step": 13352 + }, + { + "epoch": 5.59, + "learning_rate": 4.388564852126571e-06, + "loss": 0.0978, + "step": 13353 + }, + { + "epoch": 5.59, + "learning_rate": 4.387161611445034e-06, + "loss": 0.1128, + "step": 13354 + }, + { + "epoch": 5.59, + "learning_rate": 4.385758532099556e-06, + "loss": 0.1407, + "step": 13355 + }, + { + "epoch": 5.59, + "learning_rate": 4.384355614130471e-06, + "loss": 0.1338, + "step": 13356 + }, + { + "epoch": 5.59, + "learning_rate": 4.382952857578098e-06, + "loss": 0.1307, + "step": 13357 + }, + { + "epoch": 5.59, + "learning_rate": 4.381550262482759e-06, + "loss": 0.1181, + "step": 13358 + }, + { + "epoch": 5.59, + "learning_rate": 4.380147828884772e-06, + "loss": 0.1042, + "step": 13359 + }, + { + "epoch": 5.59, + "learning_rate": 4.37874555682445e-06, + "loss": 0.1201, + "step": 13360 + }, + { + "epoch": 5.59, + "learning_rate": 4.377343446342097e-06, + "loss": 0.1473, + "step": 13361 + }, + { + "epoch": 5.59, + "learning_rate": 4.375941497478022e-06, + "loss": 0.1154, + "step": 13362 + }, + { + "epoch": 5.59, + "learning_rate": 4.374539710272514e-06, + "loss": 0.1125, + "step": 13363 + }, + { + "epoch": 5.6, + "learning_rate": 4.373138084765875e-06, + "loss": 0.132, + "step": 13364 + }, + { + "epoch": 5.6, + "learning_rate": 4.371736620998383e-06, + "loss": 0.1434, + "step": 13365 + }, + { + "epoch": 5.6, + "learning_rate": 4.370335319010328e-06, + "loss": 0.1287, + "step": 13366 + }, + { + "epoch": 5.6, + "learning_rate": 4.368934178841988e-06, + "loss": 0.1432, + "step": 13367 + }, + { + "epoch": 5.6, + "learning_rate": 4.367533200533639e-06, + "loss": 0.1219, + "step": 13368 + }, + { + "epoch": 5.6, + "learning_rate": 4.36613238412555e-06, + "loss": 0.1447, + "step": 13369 + }, + { + "epoch": 5.6, + "learning_rate": 4.36473172965799e-06, + "loss": 0.1143, + "step": 13370 + }, + { + "epoch": 5.6, + "learning_rate": 4.363331237171211e-06, + "loss": 0.111, + "step": 13371 + }, + { + "epoch": 5.6, + "learning_rate": 4.361930906705475e-06, + "loss": 0.115, + "step": 13372 + }, + { + "epoch": 5.6, + "learning_rate": 4.360530738301031e-06, + "loss": 0.1121, + "step": 13373 + }, + { + "epoch": 5.6, + "learning_rate": 4.359130731998127e-06, + "loss": 0.1524, + "step": 13374 + }, + { + "epoch": 5.6, + "learning_rate": 4.357730887837005e-06, + "loss": 0.1011, + "step": 13375 + }, + { + "epoch": 5.6, + "learning_rate": 4.356331205857904e-06, + "loss": 0.1064, + "step": 13376 + }, + { + "epoch": 5.6, + "learning_rate": 4.354931686101051e-06, + "loss": 0.1177, + "step": 13377 + }, + { + "epoch": 5.6, + "learning_rate": 4.353532328606679e-06, + "loss": 0.0824, + "step": 13378 + }, + { + "epoch": 5.6, + "learning_rate": 4.352133133415009e-06, + "loss": 0.12, + "step": 13379 + }, + { + "epoch": 5.6, + "learning_rate": 4.3507341005662595e-06, + "loss": 0.1201, + "step": 13380 + }, + { + "epoch": 5.6, + "learning_rate": 4.349335230100645e-06, + "loss": 0.0962, + "step": 13381 + }, + { + "epoch": 5.6, + "learning_rate": 4.3479365220583754e-06, + "loss": 0.1817, + "step": 13382 + }, + { + "epoch": 5.6, + "learning_rate": 4.346537976479659e-06, + "loss": 0.143, + "step": 13383 + }, + { + "epoch": 5.6, + "learning_rate": 4.345139593404688e-06, + "loss": 0.1115, + "step": 13384 + }, + { + "epoch": 5.6, + "learning_rate": 4.343741372873665e-06, + "loss": 0.104, + "step": 13385 + }, + { + "epoch": 5.6, + "learning_rate": 4.342343314926773e-06, + "loss": 0.1272, + "step": 13386 + }, + { + "epoch": 5.6, + "learning_rate": 4.340945419604201e-06, + "loss": 0.1128, + "step": 13387 + }, + { + "epoch": 5.61, + "learning_rate": 4.3395476869461325e-06, + "loss": 0.115, + "step": 13388 + }, + { + "epoch": 5.61, + "learning_rate": 4.338150116992743e-06, + "loss": 0.1597, + "step": 13389 + }, + { + "epoch": 5.61, + "learning_rate": 4.3367527097842034e-06, + "loss": 0.1409, + "step": 13390 + }, + { + "epoch": 5.61, + "learning_rate": 4.335355465360686e-06, + "loss": 0.1341, + "step": 13391 + }, + { + "epoch": 5.61, + "learning_rate": 4.333958383762345e-06, + "loss": 0.1257, + "step": 13392 + }, + { + "epoch": 5.61, + "learning_rate": 4.332561465029341e-06, + "loss": 0.1009, + "step": 13393 + }, + { + "epoch": 5.61, + "learning_rate": 4.331164709201829e-06, + "loss": 0.099, + "step": 13394 + }, + { + "epoch": 5.61, + "learning_rate": 4.329768116319958e-06, + "loss": 0.0919, + "step": 13395 + }, + { + "epoch": 5.61, + "learning_rate": 4.328371686423871e-06, + "loss": 0.085, + "step": 13396 + }, + { + "epoch": 5.61, + "learning_rate": 4.32697541955371e-06, + "loss": 0.1586, + "step": 13397 + }, + { + "epoch": 5.61, + "learning_rate": 4.325579315749603e-06, + "loss": 0.1115, + "step": 13398 + }, + { + "epoch": 5.61, + "learning_rate": 4.324183375051683e-06, + "loss": 0.1282, + "step": 13399 + }, + { + "epoch": 5.61, + "learning_rate": 4.322787597500074e-06, + "loss": 0.0924, + "step": 13400 + }, + { + "epoch": 5.61, + "learning_rate": 4.321391983134899e-06, + "loss": 0.1703, + "step": 13401 + }, + { + "epoch": 5.61, + "learning_rate": 4.319996531996272e-06, + "loss": 0.1231, + "step": 13402 + }, + { + "epoch": 5.61, + "learning_rate": 4.318601244124308e-06, + "loss": 0.1272, + "step": 13403 + }, + { + "epoch": 5.61, + "learning_rate": 4.317206119559108e-06, + "loss": 0.114, + "step": 13404 + }, + { + "epoch": 5.61, + "learning_rate": 4.315811158340773e-06, + "loss": 0.11, + "step": 13405 + }, + { + "epoch": 5.61, + "learning_rate": 4.314416360509403e-06, + "loss": 0.1258, + "step": 13406 + }, + { + "epoch": 5.61, + "learning_rate": 4.3130217261050925e-06, + "loss": 0.1358, + "step": 13407 + }, + { + "epoch": 5.61, + "learning_rate": 4.3116272551679215e-06, + "loss": 0.1249, + "step": 13408 + }, + { + "epoch": 5.61, + "learning_rate": 4.310232947737979e-06, + "loss": 0.1491, + "step": 13409 + }, + { + "epoch": 5.61, + "learning_rate": 4.308838803855341e-06, + "loss": 0.1461, + "step": 13410 + }, + { + "epoch": 5.61, + "learning_rate": 4.307444823560084e-06, + "loss": 0.1283, + "step": 13411 + }, + { + "epoch": 5.62, + "learning_rate": 4.306051006892271e-06, + "loss": 0.1214, + "step": 13412 + }, + { + "epoch": 5.62, + "learning_rate": 4.30465735389197e-06, + "loss": 0.1222, + "step": 13413 + }, + { + "epoch": 5.62, + "learning_rate": 4.303263864599239e-06, + "loss": 0.0874, + "step": 13414 + }, + { + "epoch": 5.62, + "learning_rate": 4.3018705390541316e-06, + "loss": 0.1077, + "step": 13415 + }, + { + "epoch": 5.62, + "learning_rate": 4.3004773772967e-06, + "loss": 0.1062, + "step": 13416 + }, + { + "epoch": 5.62, + "learning_rate": 4.2990843793669925e-06, + "loss": 0.0945, + "step": 13417 + }, + { + "epoch": 5.62, + "learning_rate": 4.297691545305042e-06, + "loss": 0.1373, + "step": 13418 + }, + { + "epoch": 5.62, + "learning_rate": 4.296298875150887e-06, + "loss": 0.1218, + "step": 13419 + }, + { + "epoch": 5.62, + "learning_rate": 4.29490636894456e-06, + "loss": 0.1227, + "step": 13420 + }, + { + "epoch": 5.62, + "learning_rate": 4.293514026726088e-06, + "loss": 0.1127, + "step": 13421 + }, + { + "epoch": 5.62, + "learning_rate": 4.29212184853549e-06, + "loss": 0.1294, + "step": 13422 + }, + { + "epoch": 5.62, + "learning_rate": 4.2907298344127844e-06, + "loss": 0.1136, + "step": 13423 + }, + { + "epoch": 5.62, + "learning_rate": 4.289337984397988e-06, + "loss": 0.0996, + "step": 13424 + }, + { + "epoch": 5.62, + "learning_rate": 4.287946298531099e-06, + "loss": 0.107, + "step": 13425 + }, + { + "epoch": 5.62, + "learning_rate": 4.286554776852125e-06, + "loss": 0.1048, + "step": 13426 + }, + { + "epoch": 5.62, + "learning_rate": 4.285163419401065e-06, + "loss": 0.1187, + "step": 13427 + }, + { + "epoch": 5.62, + "learning_rate": 4.2837722262179105e-06, + "loss": 0.1657, + "step": 13428 + }, + { + "epoch": 5.62, + "learning_rate": 4.282381197342651e-06, + "loss": 0.118, + "step": 13429 + }, + { + "epoch": 5.62, + "learning_rate": 4.280990332815274e-06, + "loss": 0.1607, + "step": 13430 + }, + { + "epoch": 5.62, + "learning_rate": 4.27959963267575e-06, + "loss": 0.151, + "step": 13431 + }, + { + "epoch": 5.62, + "learning_rate": 4.278209096964063e-06, + "loss": 0.1083, + "step": 13432 + }, + { + "epoch": 5.62, + "learning_rate": 4.276818725720176e-06, + "loss": 0.0842, + "step": 13433 + }, + { + "epoch": 5.62, + "learning_rate": 4.275428518984054e-06, + "loss": 0.1159, + "step": 13434 + }, + { + "epoch": 5.62, + "learning_rate": 4.27403847679566e-06, + "loss": 0.1504, + "step": 13435 + }, + { + "epoch": 5.63, + "learning_rate": 4.272648599194948e-06, + "loss": 0.1206, + "step": 13436 + }, + { + "epoch": 5.63, + "learning_rate": 4.27125888622187e-06, + "loss": 0.1466, + "step": 13437 + }, + { + "epoch": 5.63, + "learning_rate": 4.269869337916377e-06, + "loss": 0.1112, + "step": 13438 + }, + { + "epoch": 5.63, + "learning_rate": 4.2684799543183995e-06, + "loss": 0.0942, + "step": 13439 + }, + { + "epoch": 5.63, + "learning_rate": 4.26709073546788e-06, + "loss": 0.1187, + "step": 13440 + }, + { + "epoch": 5.63, + "learning_rate": 4.2657016814047515e-06, + "loss": 0.118, + "step": 13441 + }, + { + "epoch": 5.63, + "learning_rate": 4.264312792168938e-06, + "loss": 0.0935, + "step": 13442 + }, + { + "epoch": 5.63, + "learning_rate": 4.262924067800364e-06, + "loss": 0.1004, + "step": 13443 + }, + { + "epoch": 5.63, + "learning_rate": 4.261535508338951e-06, + "loss": 0.113, + "step": 13444 + }, + { + "epoch": 5.63, + "learning_rate": 4.260147113824603e-06, + "loss": 0.1241, + "step": 13445 + }, + { + "epoch": 5.63, + "learning_rate": 4.2587588842972325e-06, + "loss": 0.131, + "step": 13446 + }, + { + "epoch": 5.63, + "learning_rate": 4.257370819796745e-06, + "loss": 0.1177, + "step": 13447 + }, + { + "epoch": 5.63, + "learning_rate": 4.255982920363035e-06, + "loss": 0.1428, + "step": 13448 + }, + { + "epoch": 5.63, + "learning_rate": 4.254595186036001e-06, + "loss": 0.1097, + "step": 13449 + }, + { + "epoch": 5.63, + "learning_rate": 4.253207616855528e-06, + "loss": 0.1172, + "step": 13450 + }, + { + "epoch": 5.63, + "learning_rate": 4.251820212861507e-06, + "loss": 0.1338, + "step": 13451 + }, + { + "epoch": 5.63, + "learning_rate": 4.2504329740938084e-06, + "loss": 0.1281, + "step": 13452 + }, + { + "epoch": 5.63, + "learning_rate": 4.249045900592318e-06, + "loss": 0.1366, + "step": 13453 + }, + { + "epoch": 5.63, + "learning_rate": 4.2476589923968946e-06, + "loss": 0.0745, + "step": 13454 + }, + { + "epoch": 5.63, + "learning_rate": 4.2462722495474085e-06, + "loss": 0.1222, + "step": 13455 + }, + { + "epoch": 5.63, + "learning_rate": 4.244885672083722e-06, + "loss": 0.1208, + "step": 13456 + }, + { + "epoch": 5.63, + "learning_rate": 4.243499260045689e-06, + "loss": 0.1247, + "step": 13457 + }, + { + "epoch": 5.63, + "learning_rate": 4.242113013473166e-06, + "loss": 0.1226, + "step": 13458 + }, + { + "epoch": 5.63, + "learning_rate": 4.240726932405991e-06, + "loss": 0.1737, + "step": 13459 + }, + { + "epoch": 5.64, + "learning_rate": 4.239341016884008e-06, + "loss": 0.1072, + "step": 13460 + }, + { + "epoch": 5.64, + "learning_rate": 4.237955266947058e-06, + "loss": 0.1087, + "step": 13461 + }, + { + "epoch": 5.64, + "learning_rate": 4.23656968263497e-06, + "loss": 0.137, + "step": 13462 + }, + { + "epoch": 5.64, + "learning_rate": 4.235184263987572e-06, + "loss": 0.1212, + "step": 13463 + }, + { + "epoch": 5.64, + "learning_rate": 4.233799011044687e-06, + "loss": 0.1699, + "step": 13464 + }, + { + "epoch": 5.64, + "learning_rate": 4.232413923846135e-06, + "loss": 0.1369, + "step": 13465 + }, + { + "epoch": 5.64, + "learning_rate": 4.231029002431723e-06, + "loss": 0.0873, + "step": 13466 + }, + { + "epoch": 5.64, + "learning_rate": 4.2296442468412646e-06, + "loss": 0.1319, + "step": 13467 + }, + { + "epoch": 5.64, + "learning_rate": 4.2282596571145615e-06, + "loss": 0.1551, + "step": 13468 + }, + { + "epoch": 5.64, + "learning_rate": 4.226875233291412e-06, + "loss": 0.1186, + "step": 13469 + }, + { + "epoch": 5.64, + "learning_rate": 4.225490975411611e-06, + "loss": 0.1651, + "step": 13470 + }, + { + "epoch": 5.64, + "learning_rate": 4.224106883514952e-06, + "loss": 0.1228, + "step": 13471 + }, + { + "epoch": 5.64, + "learning_rate": 4.222722957641212e-06, + "loss": 0.1283, + "step": 13472 + }, + { + "epoch": 5.64, + "learning_rate": 4.221339197830173e-06, + "loss": 0.1121, + "step": 13473 + }, + { + "epoch": 5.64, + "learning_rate": 4.219955604121611e-06, + "loss": 0.1387, + "step": 13474 + }, + { + "epoch": 5.64, + "learning_rate": 4.2185721765553e-06, + "loss": 0.1105, + "step": 13475 + }, + { + "epoch": 5.64, + "learning_rate": 4.217188915170996e-06, + "loss": 0.12, + "step": 13476 + }, + { + "epoch": 5.64, + "learning_rate": 4.215805820008468e-06, + "loss": 0.1138, + "step": 13477 + }, + { + "epoch": 5.64, + "learning_rate": 4.214422891107467e-06, + "loss": 0.1054, + "step": 13478 + }, + { + "epoch": 5.64, + "learning_rate": 4.2130401285077505e-06, + "loss": 0.1142, + "step": 13479 + }, + { + "epoch": 5.64, + "learning_rate": 4.211657532249056e-06, + "loss": 0.1037, + "step": 13480 + }, + { + "epoch": 5.64, + "learning_rate": 4.21027510237113e-06, + "loss": 0.1261, + "step": 13481 + }, + { + "epoch": 5.64, + "learning_rate": 4.208892838913707e-06, + "loss": 0.1181, + "step": 13482 + }, + { + "epoch": 5.64, + "learning_rate": 4.207510741916522e-06, + "loss": 0.1007, + "step": 13483 + }, + { + "epoch": 5.65, + "learning_rate": 4.206128811419299e-06, + "loss": 0.1343, + "step": 13484 + }, + { + "epoch": 5.65, + "learning_rate": 4.204747047461767e-06, + "loss": 0.1377, + "step": 13485 + }, + { + "epoch": 5.65, + "learning_rate": 4.2033654500836344e-06, + "loss": 0.1334, + "step": 13486 + }, + { + "epoch": 5.65, + "learning_rate": 4.201984019324617e-06, + "loss": 0.1247, + "step": 13487 + }, + { + "epoch": 5.65, + "learning_rate": 4.200602755224426e-06, + "loss": 0.1618, + "step": 13488 + }, + { + "epoch": 5.65, + "learning_rate": 4.199221657822761e-06, + "loss": 0.1531, + "step": 13489 + }, + { + "epoch": 5.65, + "learning_rate": 4.1978407271593226e-06, + "loss": 0.1097, + "step": 13490 + }, + { + "epoch": 5.65, + "learning_rate": 4.1964599632738035e-06, + "loss": 0.1293, + "step": 13491 + }, + { + "epoch": 5.65, + "learning_rate": 4.1950793662058965e-06, + "loss": 0.1205, + "step": 13492 + }, + { + "epoch": 5.65, + "learning_rate": 4.1936989359952775e-06, + "loss": 0.1221, + "step": 13493 + }, + { + "epoch": 5.65, + "learning_rate": 4.192318672681631e-06, + "loss": 0.1379, + "step": 13494 + }, + { + "epoch": 5.65, + "learning_rate": 4.1909385763046305e-06, + "loss": 0.1072, + "step": 13495 + }, + { + "epoch": 5.65, + "learning_rate": 4.1895586469039466e-06, + "loss": 0.0995, + "step": 13496 + }, + { + "epoch": 5.65, + "learning_rate": 4.188178884519241e-06, + "loss": 0.1429, + "step": 13497 + }, + { + "epoch": 5.65, + "learning_rate": 4.186799289190182e-06, + "loss": 0.1256, + "step": 13498 + }, + { + "epoch": 5.65, + "learning_rate": 4.185419860956418e-06, + "loss": 0.1429, + "step": 13499 + }, + { + "epoch": 5.65, + "learning_rate": 4.184040599857596e-06, + "loss": 0.1564, + "step": 13500 + }, + { + "epoch": 5.65, + "learning_rate": 4.182661505933366e-06, + "loss": 0.1095, + "step": 13501 + }, + { + "epoch": 5.65, + "learning_rate": 4.18128257922337e-06, + "loss": 0.1098, + "step": 13502 + }, + { + "epoch": 5.65, + "learning_rate": 4.179903819767242e-06, + "loss": 0.1369, + "step": 13503 + }, + { + "epoch": 5.65, + "learning_rate": 4.178525227604615e-06, + "loss": 0.1407, + "step": 13504 + }, + { + "epoch": 5.65, + "learning_rate": 4.177146802775114e-06, + "loss": 0.1111, + "step": 13505 + }, + { + "epoch": 5.65, + "learning_rate": 4.175768545318364e-06, + "loss": 0.1186, + "step": 13506 + }, + { + "epoch": 5.66, + "learning_rate": 4.174390455273975e-06, + "loss": 0.1016, + "step": 13507 + }, + { + "epoch": 5.66, + "learning_rate": 4.173012532681564e-06, + "loss": 0.1468, + "step": 13508 + }, + { + "epoch": 5.66, + "learning_rate": 4.171634777580736e-06, + "loss": 0.1381, + "step": 13509 + }, + { + "epoch": 5.66, + "learning_rate": 4.170257190011096e-06, + "loss": 0.1473, + "step": 13510 + }, + { + "epoch": 5.66, + "learning_rate": 4.168879770012237e-06, + "loss": 0.0879, + "step": 13511 + }, + { + "epoch": 5.66, + "learning_rate": 4.16750251762376e-06, + "loss": 0.0742, + "step": 13512 + }, + { + "epoch": 5.66, + "learning_rate": 4.166125432885244e-06, + "loss": 0.1145, + "step": 13513 + }, + { + "epoch": 5.66, + "learning_rate": 4.164748515836276e-06, + "loss": 0.0923, + "step": 13514 + }, + { + "epoch": 5.66, + "learning_rate": 4.163371766516432e-06, + "loss": 0.1548, + "step": 13515 + }, + { + "epoch": 5.66, + "learning_rate": 4.1619951849652885e-06, + "loss": 0.1205, + "step": 13516 + }, + { + "epoch": 5.66, + "learning_rate": 4.160618771222412e-06, + "loss": 0.1102, + "step": 13517 + }, + { + "epoch": 5.66, + "learning_rate": 4.159242525327371e-06, + "loss": 0.112, + "step": 13518 + }, + { + "epoch": 5.66, + "learning_rate": 4.157866447319718e-06, + "loss": 0.1344, + "step": 13519 + }, + { + "epoch": 5.66, + "learning_rate": 4.15649053723901e-06, + "loss": 0.1181, + "step": 13520 + }, + { + "epoch": 5.66, + "learning_rate": 4.155114795124798e-06, + "loss": 0.1008, + "step": 13521 + }, + { + "epoch": 5.66, + "learning_rate": 4.153739221016622e-06, + "loss": 0.0864, + "step": 13522 + }, + { + "epoch": 5.66, + "learning_rate": 4.152363814954024e-06, + "loss": 0.1235, + "step": 13523 + }, + { + "epoch": 5.66, + "learning_rate": 4.150988576976539e-06, + "loss": 0.1005, + "step": 13524 + }, + { + "epoch": 5.66, + "learning_rate": 4.1496135071236965e-06, + "loss": 0.075, + "step": 13525 + }, + { + "epoch": 5.66, + "learning_rate": 4.1482386054350265e-06, + "loss": 0.1, + "step": 13526 + }, + { + "epoch": 5.66, + "learning_rate": 4.146863871950041e-06, + "loss": 0.0988, + "step": 13527 + }, + { + "epoch": 5.66, + "learning_rate": 4.1454893067082605e-06, + "loss": 0.1228, + "step": 13528 + }, + { + "epoch": 5.66, + "learning_rate": 4.144114909749193e-06, + "loss": 0.127, + "step": 13529 + }, + { + "epoch": 5.66, + "learning_rate": 4.142740681112347e-06, + "loss": 0.1128, + "step": 13530 + }, + { + "epoch": 5.67, + "learning_rate": 4.141366620837223e-06, + "loss": 0.1203, + "step": 13531 + }, + { + "epoch": 5.67, + "learning_rate": 4.139992728963315e-06, + "loss": 0.1196, + "step": 13532 + }, + { + "epoch": 5.67, + "learning_rate": 4.138619005530121e-06, + "loss": 0.1411, + "step": 13533 + }, + { + "epoch": 5.67, + "learning_rate": 4.1372454505771186e-06, + "loss": 0.1377, + "step": 13534 + }, + { + "epoch": 5.67, + "learning_rate": 4.135872064143792e-06, + "loss": 0.1149, + "step": 13535 + }, + { + "epoch": 5.67, + "learning_rate": 4.134498846269621e-06, + "loss": 0.1183, + "step": 13536 + }, + { + "epoch": 5.67, + "learning_rate": 4.133125796994074e-06, + "loss": 0.134, + "step": 13537 + }, + { + "epoch": 5.67, + "learning_rate": 4.131752916356622e-06, + "loss": 0.1001, + "step": 13538 + }, + { + "epoch": 5.67, + "learning_rate": 4.130380204396727e-06, + "loss": 0.1182, + "step": 13539 + }, + { + "epoch": 5.67, + "learning_rate": 4.129007661153841e-06, + "loss": 0.1349, + "step": 13540 + }, + { + "epoch": 5.67, + "learning_rate": 4.12763528666742e-06, + "loss": 0.1489, + "step": 13541 + }, + { + "epoch": 5.67, + "learning_rate": 4.126263080976911e-06, + "loss": 0.1143, + "step": 13542 + }, + { + "epoch": 5.67, + "learning_rate": 4.124891044121761e-06, + "loss": 0.1215, + "step": 13543 + }, + { + "epoch": 5.67, + "learning_rate": 4.1235191761414025e-06, + "loss": 0.0732, + "step": 13544 + }, + { + "epoch": 5.67, + "learning_rate": 4.12214747707527e-06, + "loss": 0.1264, + "step": 13545 + }, + { + "epoch": 5.67, + "learning_rate": 4.120775946962792e-06, + "loss": 0.1171, + "step": 13546 + }, + { + "epoch": 5.67, + "learning_rate": 4.1194045858433965e-06, + "loss": 0.1367, + "step": 13547 + }, + { + "epoch": 5.67, + "learning_rate": 4.118033393756493e-06, + "loss": 0.1333, + "step": 13548 + }, + { + "epoch": 5.67, + "learning_rate": 4.116662370741501e-06, + "loss": 0.0928, + "step": 13549 + }, + { + "epoch": 5.67, + "learning_rate": 4.115291516837827e-06, + "loss": 0.1202, + "step": 13550 + }, + { + "epoch": 5.67, + "learning_rate": 4.113920832084878e-06, + "loss": 0.1247, + "step": 13551 + }, + { + "epoch": 5.67, + "learning_rate": 4.11255031652205e-06, + "loss": 0.1122, + "step": 13552 + }, + { + "epoch": 5.67, + "learning_rate": 4.111179970188745e-06, + "loss": 0.1025, + "step": 13553 + }, + { + "epoch": 5.67, + "learning_rate": 4.10980979312434e-06, + "loss": 0.1193, + "step": 13554 + }, + { + "epoch": 5.68, + "learning_rate": 4.108439785368226e-06, + "loss": 0.1566, + "step": 13555 + }, + { + "epoch": 5.68, + "learning_rate": 4.107069946959783e-06, + "loss": 0.1232, + "step": 13556 + }, + { + "epoch": 5.68, + "learning_rate": 4.105700277938385e-06, + "loss": 0.1144, + "step": 13557 + }, + { + "epoch": 5.68, + "learning_rate": 4.1043307783434e-06, + "loss": 0.1173, + "step": 13558 + }, + { + "epoch": 5.68, + "learning_rate": 4.102961448214201e-06, + "loss": 0.1286, + "step": 13559 + }, + { + "epoch": 5.68, + "learning_rate": 4.101592287590138e-06, + "loss": 0.1327, + "step": 13560 + }, + { + "epoch": 5.68, + "learning_rate": 4.100223296510571e-06, + "loss": 0.1254, + "step": 13561 + }, + { + "epoch": 5.68, + "learning_rate": 4.098854475014849e-06, + "loss": 0.1013, + "step": 13562 + }, + { + "epoch": 5.68, + "learning_rate": 4.097485823142319e-06, + "loss": 0.1239, + "step": 13563 + }, + { + "epoch": 5.68, + "learning_rate": 4.096117340932322e-06, + "loss": 0.1021, + "step": 13564 + }, + { + "epoch": 5.68, + "learning_rate": 4.094749028424192e-06, + "loss": 0.1207, + "step": 13565 + }, + { + "epoch": 5.68, + "learning_rate": 4.093380885657264e-06, + "loss": 0.1143, + "step": 13566 + }, + { + "epoch": 5.68, + "learning_rate": 4.092012912670863e-06, + "loss": 0.1372, + "step": 13567 + }, + { + "epoch": 5.68, + "learning_rate": 4.090645109504303e-06, + "loss": 0.0726, + "step": 13568 + }, + { + "epoch": 5.68, + "learning_rate": 4.0892774761969055e-06, + "loss": 0.1092, + "step": 13569 + }, + { + "epoch": 5.68, + "learning_rate": 4.087910012787983e-06, + "loss": 0.1253, + "step": 13570 + }, + { + "epoch": 5.68, + "learning_rate": 4.08654271931684e-06, + "loss": 0.1181, + "step": 13571 + }, + { + "epoch": 5.68, + "learning_rate": 4.085175595822779e-06, + "loss": 0.1061, + "step": 13572 + }, + { + "epoch": 5.68, + "learning_rate": 4.083808642345098e-06, + "loss": 0.1261, + "step": 13573 + }, + { + "epoch": 5.68, + "learning_rate": 4.082441858923088e-06, + "loss": 0.0955, + "step": 13574 + }, + { + "epoch": 5.68, + "learning_rate": 4.0810752455960356e-06, + "loss": 0.1048, + "step": 13575 + }, + { + "epoch": 5.68, + "learning_rate": 4.07970880240322e-06, + "loss": 0.069, + "step": 13576 + }, + { + "epoch": 5.68, + "learning_rate": 4.0783425293839245e-06, + "loss": 0.1157, + "step": 13577 + }, + { + "epoch": 5.68, + "learning_rate": 4.0769764265774155e-06, + "loss": 0.111, + "step": 13578 + }, + { + "epoch": 5.69, + "learning_rate": 4.0756104940229645e-06, + "loss": 0.1169, + "step": 13579 + }, + { + "epoch": 5.69, + "learning_rate": 4.074244731759836e-06, + "loss": 0.1089, + "step": 13580 + }, + { + "epoch": 5.69, + "learning_rate": 4.07287913982728e-06, + "loss": 0.1629, + "step": 13581 + }, + { + "epoch": 5.69, + "learning_rate": 4.071513718264554e-06, + "loss": 0.1112, + "step": 13582 + }, + { + "epoch": 5.69, + "learning_rate": 4.070148467110906e-06, + "loss": 0.113, + "step": 13583 + }, + { + "epoch": 5.69, + "learning_rate": 4.068783386405577e-06, + "loss": 0.145, + "step": 13584 + }, + { + "epoch": 5.69, + "learning_rate": 4.067418476187807e-06, + "loss": 0.1243, + "step": 13585 + }, + { + "epoch": 5.69, + "learning_rate": 4.066053736496832e-06, + "loss": 0.0941, + "step": 13586 + }, + { + "epoch": 5.69, + "learning_rate": 4.0646891673718725e-06, + "loss": 0.1226, + "step": 13587 + }, + { + "epoch": 5.69, + "learning_rate": 4.063324768852157e-06, + "loss": 0.1108, + "step": 13588 + }, + { + "epoch": 5.69, + "learning_rate": 4.061960540976906e-06, + "loss": 0.1098, + "step": 13589 + }, + { + "epoch": 5.69, + "learning_rate": 4.060596483785326e-06, + "loss": 0.152, + "step": 13590 + }, + { + "epoch": 5.69, + "learning_rate": 4.059232597316629e-06, + "loss": 0.1005, + "step": 13591 + }, + { + "epoch": 5.69, + "learning_rate": 4.057868881610021e-06, + "loss": 0.1304, + "step": 13592 + }, + { + "epoch": 5.69, + "learning_rate": 4.056505336704698e-06, + "loss": 0.0902, + "step": 13593 + }, + { + "epoch": 5.69, + "learning_rate": 4.055141962639858e-06, + "loss": 0.1515, + "step": 13594 + }, + { + "epoch": 5.69, + "learning_rate": 4.053778759454684e-06, + "loss": 0.0776, + "step": 13595 + }, + { + "epoch": 5.69, + "learning_rate": 4.0524157271883635e-06, + "loss": 0.1422, + "step": 13596 + }, + { + "epoch": 5.69, + "learning_rate": 4.051052865880074e-06, + "loss": 0.124, + "step": 13597 + }, + { + "epoch": 5.69, + "learning_rate": 4.049690175568992e-06, + "loss": 0.093, + "step": 13598 + }, + { + "epoch": 5.69, + "learning_rate": 4.048327656294285e-06, + "loss": 0.09, + "step": 13599 + }, + { + "epoch": 5.69, + "learning_rate": 4.046965308095122e-06, + "loss": 0.1262, + "step": 13600 + }, + { + "epoch": 5.69, + "learning_rate": 4.045603131010655e-06, + "loss": 0.1213, + "step": 13601 + }, + { + "epoch": 5.69, + "learning_rate": 4.044241125080041e-06, + "loss": 0.1101, + "step": 13602 + }, + { + "epoch": 5.7, + "learning_rate": 4.042879290342432e-06, + "loss": 0.1136, + "step": 13603 + }, + { + "epoch": 5.7, + "learning_rate": 4.0415176268369714e-06, + "loss": 0.1203, + "step": 13604 + }, + { + "epoch": 5.7, + "learning_rate": 4.0401561346028e-06, + "loss": 0.084, + "step": 13605 + }, + { + "epoch": 5.7, + "learning_rate": 4.03879481367905e-06, + "loss": 0.1028, + "step": 13606 + }, + { + "epoch": 5.7, + "learning_rate": 4.037433664104857e-06, + "loss": 0.1255, + "step": 13607 + }, + { + "epoch": 5.7, + "learning_rate": 4.03607268591934e-06, + "loss": 0.1055, + "step": 13608 + }, + { + "epoch": 5.7, + "learning_rate": 4.03471187916162e-06, + "loss": 0.1279, + "step": 13609 + }, + { + "epoch": 5.7, + "learning_rate": 4.033351243870815e-06, + "loss": 0.1281, + "step": 13610 + }, + { + "epoch": 5.7, + "learning_rate": 4.031990780086037e-06, + "loss": 0.1073, + "step": 13611 + }, + { + "epoch": 5.7, + "learning_rate": 4.0306304878463834e-06, + "loss": 0.1445, + "step": 13612 + }, + { + "epoch": 5.7, + "learning_rate": 4.0292703671909605e-06, + "loss": 0.1192, + "step": 13613 + }, + { + "epoch": 5.7, + "learning_rate": 4.027910418158861e-06, + "loss": 0.1246, + "step": 13614 + }, + { + "epoch": 5.7, + "learning_rate": 4.026550640789183e-06, + "loss": 0.1184, + "step": 13615 + }, + { + "epoch": 5.7, + "learning_rate": 4.025191035121001e-06, + "loss": 0.1196, + "step": 13616 + }, + { + "epoch": 5.7, + "learning_rate": 4.023831601193401e-06, + "loss": 0.1371, + "step": 13617 + }, + { + "epoch": 5.7, + "learning_rate": 4.0224723390454575e-06, + "loss": 0.0976, + "step": 13618 + }, + { + "epoch": 5.7, + "learning_rate": 4.021113248716243e-06, + "loss": 0.0785, + "step": 13619 + }, + { + "epoch": 5.7, + "learning_rate": 4.019754330244822e-06, + "loss": 0.1165, + "step": 13620 + }, + { + "epoch": 5.7, + "learning_rate": 4.0183955836702615e-06, + "loss": 0.1068, + "step": 13621 + }, + { + "epoch": 5.7, + "learning_rate": 4.017037009031607e-06, + "loss": 0.125, + "step": 13622 + }, + { + "epoch": 5.7, + "learning_rate": 4.015678606367916e-06, + "loss": 0.1584, + "step": 13623 + }, + { + "epoch": 5.7, + "learning_rate": 4.01432037571823e-06, + "loss": 0.1086, + "step": 13624 + }, + { + "epoch": 5.7, + "learning_rate": 4.012962317121596e-06, + "loss": 0.1481, + "step": 13625 + }, + { + "epoch": 5.7, + "learning_rate": 4.011604430617046e-06, + "loss": 0.1074, + "step": 13626 + }, + { + "epoch": 5.71, + "learning_rate": 4.010246716243617e-06, + "loss": 0.1154, + "step": 13627 + }, + { + "epoch": 5.71, + "learning_rate": 4.008889174040327e-06, + "loss": 0.1414, + "step": 13628 + }, + { + "epoch": 5.71, + "learning_rate": 4.0075318040462005e-06, + "loss": 0.1165, + "step": 13629 + }, + { + "epoch": 5.71, + "learning_rate": 4.006174606300255e-06, + "loss": 0.0938, + "step": 13630 + }, + { + "epoch": 5.71, + "learning_rate": 4.004817580841503e-06, + "loss": 0.1314, + "step": 13631 + }, + { + "epoch": 5.71, + "learning_rate": 4.003460727708947e-06, + "loss": 0.1182, + "step": 13632 + }, + { + "epoch": 5.71, + "learning_rate": 4.002104046941594e-06, + "loss": 0.1373, + "step": 13633 + }, + { + "epoch": 5.71, + "learning_rate": 4.00074753857844e-06, + "loss": 0.0948, + "step": 13634 + }, + { + "epoch": 5.71, + "learning_rate": 3.9993912026584735e-06, + "loss": 0.1131, + "step": 13635 + }, + { + "epoch": 5.71, + "learning_rate": 3.99803503922068e-06, + "loss": 0.1216, + "step": 13636 + }, + { + "epoch": 5.71, + "learning_rate": 3.996679048304042e-06, + "loss": 0.0913, + "step": 13637 + }, + { + "epoch": 5.71, + "learning_rate": 3.9953232299475385e-06, + "loss": 0.1136, + "step": 13638 + }, + { + "epoch": 5.71, + "learning_rate": 3.993967584190141e-06, + "loss": 0.1164, + "step": 13639 + }, + { + "epoch": 5.71, + "learning_rate": 3.992612111070814e-06, + "loss": 0.0847, + "step": 13640 + }, + { + "epoch": 5.71, + "learning_rate": 3.991256810628525e-06, + "loss": 0.1079, + "step": 13641 + }, + { + "epoch": 5.71, + "learning_rate": 3.989901682902223e-06, + "loss": 0.1372, + "step": 13642 + }, + { + "epoch": 5.71, + "learning_rate": 3.9885467279308644e-06, + "loss": 0.1175, + "step": 13643 + }, + { + "epoch": 5.71, + "learning_rate": 3.987191945753396e-06, + "loss": 0.1287, + "step": 13644 + }, + { + "epoch": 5.71, + "learning_rate": 3.9858373364087586e-06, + "loss": 0.0921, + "step": 13645 + }, + { + "epoch": 5.71, + "learning_rate": 3.98448289993589e-06, + "loss": 0.1228, + "step": 13646 + }, + { + "epoch": 5.71, + "learning_rate": 3.983128636373723e-06, + "loss": 0.1311, + "step": 13647 + }, + { + "epoch": 5.71, + "learning_rate": 3.981774545761188e-06, + "loss": 0.1644, + "step": 13648 + }, + { + "epoch": 5.71, + "learning_rate": 3.980420628137199e-06, + "loss": 0.1409, + "step": 13649 + }, + { + "epoch": 5.71, + "learning_rate": 3.979066883540677e-06, + "loss": 0.115, + "step": 13650 + }, + { + "epoch": 5.72, + "learning_rate": 3.977713312010535e-06, + "loss": 0.1419, + "step": 13651 + }, + { + "epoch": 5.72, + "learning_rate": 3.97635991358568e-06, + "loss": 0.1248, + "step": 13652 + }, + { + "epoch": 5.72, + "learning_rate": 3.975006688305015e-06, + "loss": 0.0984, + "step": 13653 + }, + { + "epoch": 5.72, + "learning_rate": 3.973653636207437e-06, + "loss": 0.0916, + "step": 13654 + }, + { + "epoch": 5.72, + "learning_rate": 3.972300757331836e-06, + "loss": 0.1347, + "step": 13655 + }, + { + "epoch": 5.72, + "learning_rate": 3.970948051717104e-06, + "loss": 0.1198, + "step": 13656 + }, + { + "epoch": 5.72, + "learning_rate": 3.969595519402117e-06, + "loss": 0.1114, + "step": 13657 + }, + { + "epoch": 5.72, + "learning_rate": 3.9682431604257554e-06, + "loss": 0.111, + "step": 13658 + }, + { + "epoch": 5.72, + "learning_rate": 3.966890974826891e-06, + "loss": 0.0948, + "step": 13659 + }, + { + "epoch": 5.72, + "learning_rate": 3.965538962644392e-06, + "loss": 0.0953, + "step": 13660 + }, + { + "epoch": 5.72, + "learning_rate": 3.964187123917122e-06, + "loss": 0.1222, + "step": 13661 + }, + { + "epoch": 5.72, + "learning_rate": 3.962835458683939e-06, + "loss": 0.1081, + "step": 13662 + }, + { + "epoch": 5.72, + "learning_rate": 3.9614839669836905e-06, + "loss": 0.0928, + "step": 13663 + }, + { + "epoch": 5.72, + "learning_rate": 3.960132648855226e-06, + "loss": 0.0574, + "step": 13664 + }, + { + "epoch": 5.72, + "learning_rate": 3.95878150433739e-06, + "loss": 0.1443, + "step": 13665 + }, + { + "epoch": 5.72, + "learning_rate": 3.9574305334690175e-06, + "loss": 0.1078, + "step": 13666 + }, + { + "epoch": 5.72, + "learning_rate": 3.9560797362889436e-06, + "loss": 0.1208, + "step": 13667 + }, + { + "epoch": 5.72, + "learning_rate": 3.954729112835998e-06, + "loss": 0.1301, + "step": 13668 + }, + { + "epoch": 5.72, + "learning_rate": 3.953378663148994e-06, + "loss": 0.1397, + "step": 13669 + }, + { + "epoch": 5.72, + "learning_rate": 3.952028387266757e-06, + "loss": 0.1295, + "step": 13670 + }, + { + "epoch": 5.72, + "learning_rate": 3.950678285228097e-06, + "loss": 0.1382, + "step": 13671 + }, + { + "epoch": 5.72, + "learning_rate": 3.94932835707182e-06, + "loss": 0.1159, + "step": 13672 + }, + { + "epoch": 5.72, + "learning_rate": 3.947978602836731e-06, + "loss": 0.1644, + "step": 13673 + }, + { + "epoch": 5.72, + "learning_rate": 3.946629022561627e-06, + "loss": 0.1093, + "step": 13674 + }, + { + "epoch": 5.73, + "learning_rate": 3.945279616285303e-06, + "loss": 0.117, + "step": 13675 + }, + { + "epoch": 5.73, + "learning_rate": 3.943930384046541e-06, + "loss": 0.1518, + "step": 13676 + }, + { + "epoch": 5.73, + "learning_rate": 3.942581325884124e-06, + "loss": 0.0997, + "step": 13677 + }, + { + "epoch": 5.73, + "learning_rate": 3.941232441836833e-06, + "loss": 0.1424, + "step": 13678 + }, + { + "epoch": 5.73, + "learning_rate": 3.939883731943443e-06, + "loss": 0.1216, + "step": 13679 + }, + { + "epoch": 5.73, + "learning_rate": 3.938535196242713e-06, + "loss": 0.1252, + "step": 13680 + }, + { + "epoch": 5.73, + "learning_rate": 3.93718683477341e-06, + "loss": 0.1371, + "step": 13681 + }, + { + "epoch": 5.73, + "learning_rate": 3.935838647574295e-06, + "loss": 0.098, + "step": 13682 + }, + { + "epoch": 5.73, + "learning_rate": 3.934490634684114e-06, + "loss": 0.1125, + "step": 13683 + }, + { + "epoch": 5.73, + "learning_rate": 3.933142796141616e-06, + "loss": 0.0992, + "step": 13684 + }, + { + "epoch": 5.73, + "learning_rate": 3.931795131985545e-06, + "loss": 0.149, + "step": 13685 + }, + { + "epoch": 5.73, + "learning_rate": 3.930447642254639e-06, + "loss": 0.1305, + "step": 13686 + }, + { + "epoch": 5.73, + "learning_rate": 3.929100326987629e-06, + "loss": 0.1338, + "step": 13687 + }, + { + "epoch": 5.73, + "learning_rate": 3.927753186223243e-06, + "loss": 0.1654, + "step": 13688 + }, + { + "epoch": 5.73, + "learning_rate": 3.926406220000206e-06, + "loss": 0.1465, + "step": 13689 + }, + { + "epoch": 5.73, + "learning_rate": 3.925059428357231e-06, + "loss": 0.1099, + "step": 13690 + }, + { + "epoch": 5.73, + "learning_rate": 3.923712811333031e-06, + "loss": 0.1108, + "step": 13691 + }, + { + "epoch": 5.73, + "learning_rate": 3.922366368966313e-06, + "loss": 0.1221, + "step": 13692 + }, + { + "epoch": 5.73, + "learning_rate": 3.9210201012957816e-06, + "loss": 0.1164, + "step": 13693 + }, + { + "epoch": 5.73, + "learning_rate": 3.919674008360133e-06, + "loss": 0.1125, + "step": 13694 + }, + { + "epoch": 5.73, + "learning_rate": 3.918328090198063e-06, + "loss": 0.1316, + "step": 13695 + }, + { + "epoch": 5.73, + "learning_rate": 3.91698234684825e-06, + "loss": 0.0516, + "step": 13696 + }, + { + "epoch": 5.73, + "learning_rate": 3.915636778349385e-06, + "loss": 0.1248, + "step": 13697 + }, + { + "epoch": 5.73, + "learning_rate": 3.914291384740139e-06, + "loss": 0.1102, + "step": 13698 + }, + { + "epoch": 5.74, + "learning_rate": 3.9129461660591875e-06, + "loss": 0.1044, + "step": 13699 + }, + { + "epoch": 5.74, + "learning_rate": 3.911601122345197e-06, + "loss": 0.0924, + "step": 13700 + }, + { + "epoch": 5.74, + "learning_rate": 3.910256253636832e-06, + "loss": 0.1286, + "step": 13701 + }, + { + "epoch": 5.74, + "learning_rate": 3.9089115599727446e-06, + "loss": 0.135, + "step": 13702 + }, + { + "epoch": 5.74, + "learning_rate": 3.9075670413915925e-06, + "loss": 0.1467, + "step": 13703 + }, + { + "epoch": 5.74, + "learning_rate": 3.906222697932016e-06, + "loss": 0.0969, + "step": 13704 + }, + { + "epoch": 5.74, + "learning_rate": 3.90487852963266e-06, + "loss": 0.1121, + "step": 13705 + }, + { + "epoch": 5.74, + "learning_rate": 3.903534536532163e-06, + "loss": 0.1398, + "step": 13706 + }, + { + "epoch": 5.74, + "learning_rate": 3.902190718669155e-06, + "loss": 0.1083, + "step": 13707 + }, + { + "epoch": 5.74, + "learning_rate": 3.900847076082264e-06, + "loss": 0.1356, + "step": 13708 + }, + { + "epoch": 5.74, + "learning_rate": 3.899503608810114e-06, + "loss": 0.1313, + "step": 13709 + }, + { + "epoch": 5.74, + "learning_rate": 3.898160316891317e-06, + "loss": 0.0883, + "step": 13710 + }, + { + "epoch": 5.74, + "learning_rate": 3.896817200364488e-06, + "loss": 0.1363, + "step": 13711 + }, + { + "epoch": 5.74, + "learning_rate": 3.895474259268232e-06, + "loss": 0.0811, + "step": 13712 + }, + { + "epoch": 5.74, + "learning_rate": 3.89413149364115e-06, + "loss": 0.1213, + "step": 13713 + }, + { + "epoch": 5.74, + "learning_rate": 3.89278890352184e-06, + "loss": 0.1138, + "step": 13714 + }, + { + "epoch": 5.74, + "learning_rate": 3.891446488948894e-06, + "loss": 0.1276, + "step": 13715 + }, + { + "epoch": 5.74, + "learning_rate": 3.890104249960903e-06, + "loss": 0.0994, + "step": 13716 + }, + { + "epoch": 5.74, + "learning_rate": 3.888762186596439e-06, + "loss": 0.1112, + "step": 13717 + }, + { + "epoch": 5.74, + "learning_rate": 3.887420298894082e-06, + "loss": 0.1057, + "step": 13718 + }, + { + "epoch": 5.74, + "learning_rate": 3.886078586892406e-06, + "loss": 0.1357, + "step": 13719 + }, + { + "epoch": 5.74, + "learning_rate": 3.884737050629975e-06, + "loss": 0.136, + "step": 13720 + }, + { + "epoch": 5.74, + "learning_rate": 3.883395690145351e-06, + "loss": 0.1432, + "step": 13721 + }, + { + "epoch": 5.75, + "learning_rate": 3.882054505477092e-06, + "loss": 0.1476, + "step": 13722 + }, + { + "epoch": 5.75, + "learning_rate": 3.880713496663745e-06, + "loss": 0.1501, + "step": 13723 + }, + { + "epoch": 5.75, + "learning_rate": 3.879372663743862e-06, + "loss": 0.1247, + "step": 13724 + }, + { + "epoch": 5.75, + "learning_rate": 3.878032006755977e-06, + "loss": 0.1415, + "step": 13725 + }, + { + "epoch": 5.75, + "learning_rate": 3.87669152573863e-06, + "loss": 0.123, + "step": 13726 + }, + { + "epoch": 5.75, + "learning_rate": 3.87535122073035e-06, + "loss": 0.1068, + "step": 13727 + }, + { + "epoch": 5.75, + "learning_rate": 3.874011091769665e-06, + "loss": 0.1255, + "step": 13728 + }, + { + "epoch": 5.75, + "learning_rate": 3.872671138895095e-06, + "loss": 0.1134, + "step": 13729 + }, + { + "epoch": 5.75, + "learning_rate": 3.871331362145158e-06, + "loss": 0.0983, + "step": 13730 + }, + { + "epoch": 5.75, + "learning_rate": 3.869991761558362e-06, + "loss": 0.1482, + "step": 13731 + }, + { + "epoch": 5.75, + "learning_rate": 3.868652337173211e-06, + "loss": 0.132, + "step": 13732 + }, + { + "epoch": 5.75, + "learning_rate": 3.867313089028208e-06, + "loss": 0.1277, + "step": 13733 + }, + { + "epoch": 5.75, + "learning_rate": 3.8659740171618485e-06, + "loss": 0.1152, + "step": 13734 + }, + { + "epoch": 5.75, + "learning_rate": 3.864635121612622e-06, + "loss": 0.0775, + "step": 13735 + }, + { + "epoch": 5.75, + "learning_rate": 3.863296402419019e-06, + "loss": 0.1493, + "step": 13736 + }, + { + "epoch": 5.75, + "learning_rate": 3.861957859619511e-06, + "loss": 0.125, + "step": 13737 + }, + { + "epoch": 5.75, + "learning_rate": 3.860619493252578e-06, + "loss": 0.1011, + "step": 13738 + }, + { + "epoch": 5.75, + "learning_rate": 3.859281303356689e-06, + "loss": 0.163, + "step": 13739 + }, + { + "epoch": 5.75, + "learning_rate": 3.857943289970311e-06, + "loss": 0.1182, + "step": 13740 + }, + { + "epoch": 5.75, + "learning_rate": 3.856605453131901e-06, + "loss": 0.1114, + "step": 13741 + }, + { + "epoch": 5.75, + "learning_rate": 3.855267792879921e-06, + "loss": 0.1491, + "step": 13742 + }, + { + "epoch": 5.75, + "learning_rate": 3.85393030925281e-06, + "loss": 0.0939, + "step": 13743 + }, + { + "epoch": 5.75, + "learning_rate": 3.85259300228902e-06, + "loss": 0.0992, + "step": 13744 + }, + { + "epoch": 5.75, + "learning_rate": 3.851255872026989e-06, + "loss": 0.1063, + "step": 13745 + }, + { + "epoch": 5.76, + "learning_rate": 3.849918918505151e-06, + "loss": 0.0433, + "step": 13746 + }, + { + "epoch": 5.76, + "learning_rate": 3.848582141761939e-06, + "loss": 0.1548, + "step": 13747 + }, + { + "epoch": 5.76, + "learning_rate": 3.8472455418357735e-06, + "loss": 0.1037, + "step": 13748 + }, + { + "epoch": 5.76, + "learning_rate": 3.845909118765073e-06, + "loss": 0.0993, + "step": 13749 + }, + { + "epoch": 5.76, + "learning_rate": 3.844572872588258e-06, + "loss": 0.1247, + "step": 13750 + }, + { + "epoch": 5.76, + "learning_rate": 3.843236803343731e-06, + "loss": 0.1247, + "step": 13751 + }, + { + "epoch": 5.76, + "learning_rate": 3.841900911069897e-06, + "loss": 0.1141, + "step": 13752 + }, + { + "epoch": 5.76, + "learning_rate": 3.840565195805158e-06, + "loss": 0.0834, + "step": 13753 + }, + { + "epoch": 5.76, + "learning_rate": 3.839229657587906e-06, + "loss": 0.1405, + "step": 13754 + }, + { + "epoch": 5.76, + "learning_rate": 3.83789429645653e-06, + "loss": 0.1399, + "step": 13755 + }, + { + "epoch": 5.76, + "learning_rate": 3.836559112449415e-06, + "loss": 0.1061, + "step": 13756 + }, + { + "epoch": 5.76, + "learning_rate": 3.835224105604942e-06, + "loss": 0.0806, + "step": 13757 + }, + { + "epoch": 5.76, + "learning_rate": 3.833889275961479e-06, + "loss": 0.1568, + "step": 13758 + }, + { + "epoch": 5.76, + "learning_rate": 3.832554623557397e-06, + "loss": 0.1016, + "step": 13759 + }, + { + "epoch": 5.76, + "learning_rate": 3.831220148431058e-06, + "loss": 0.1125, + "step": 13760 + }, + { + "epoch": 5.76, + "learning_rate": 3.829885850620824e-06, + "loss": 0.0723, + "step": 13761 + }, + { + "epoch": 5.76, + "learning_rate": 3.8285517301650445e-06, + "loss": 0.1175, + "step": 13762 + }, + { + "epoch": 5.76, + "learning_rate": 3.827217787102072e-06, + "loss": 0.1329, + "step": 13763 + }, + { + "epoch": 5.76, + "learning_rate": 3.825884021470244e-06, + "loss": 0.1014, + "step": 13764 + }, + { + "epoch": 5.76, + "learning_rate": 3.8245504333079006e-06, + "loss": 0.0892, + "step": 13765 + }, + { + "epoch": 5.76, + "learning_rate": 3.823217022653376e-06, + "loss": 0.1108, + "step": 13766 + }, + { + "epoch": 5.76, + "learning_rate": 3.8218837895449956e-06, + "loss": 0.1202, + "step": 13767 + }, + { + "epoch": 5.76, + "learning_rate": 3.820550734021083e-06, + "loss": 0.1437, + "step": 13768 + }, + { + "epoch": 5.76, + "learning_rate": 3.819217856119961e-06, + "loss": 0.1, + "step": 13769 + }, + { + "epoch": 5.77, + "learning_rate": 3.817885155879933e-06, + "loss": 0.119, + "step": 13770 + }, + { + "epoch": 5.77, + "learning_rate": 3.816552633339313e-06, + "loss": 0.1255, + "step": 13771 + }, + { + "epoch": 5.77, + "learning_rate": 3.815220288536399e-06, + "loss": 0.1156, + "step": 13772 + }, + { + "epoch": 5.77, + "learning_rate": 3.813888121509487e-06, + "loss": 0.1489, + "step": 13773 + }, + { + "epoch": 5.77, + "learning_rate": 3.812556132296873e-06, + "loss": 0.1184, + "step": 13774 + }, + { + "epoch": 5.77, + "learning_rate": 3.8112243209368415e-06, + "loss": 0.1169, + "step": 13775 + }, + { + "epoch": 5.77, + "learning_rate": 3.8098926874676757e-06, + "loss": 0.1359, + "step": 13776 + }, + { + "epoch": 5.77, + "learning_rate": 3.8085612319276556e-06, + "loss": 0.1257, + "step": 13777 + }, + { + "epoch": 5.77, + "learning_rate": 3.8072299543550438e-06, + "loss": 0.1239, + "step": 13778 + }, + { + "epoch": 5.77, + "learning_rate": 3.8058988547881126e-06, + "loss": 0.0871, + "step": 13779 + }, + { + "epoch": 5.77, + "learning_rate": 3.804567933265122e-06, + "loss": 0.0929, + "step": 13780 + }, + { + "epoch": 5.77, + "learning_rate": 3.803237189824328e-06, + "loss": 0.1104, + "step": 13781 + }, + { + "epoch": 5.77, + "learning_rate": 3.8019066245039828e-06, + "loss": 0.0898, + "step": 13782 + }, + { + "epoch": 5.77, + "learning_rate": 3.800576237342335e-06, + "loss": 0.0989, + "step": 13783 + }, + { + "epoch": 5.77, + "learning_rate": 3.799246028377618e-06, + "loss": 0.1339, + "step": 13784 + }, + { + "epoch": 5.77, + "learning_rate": 3.7979159976480717e-06, + "loss": 0.1075, + "step": 13785 + }, + { + "epoch": 5.77, + "learning_rate": 3.796586145191926e-06, + "loss": 0.1306, + "step": 13786 + }, + { + "epoch": 5.77, + "learning_rate": 3.7952564710474072e-06, + "loss": 0.148, + "step": 13787 + }, + { + "epoch": 5.77, + "learning_rate": 3.793926975252735e-06, + "loss": 0.1047, + "step": 13788 + }, + { + "epoch": 5.77, + "learning_rate": 3.7925976578461255e-06, + "loss": 0.143, + "step": 13789 + }, + { + "epoch": 5.77, + "learning_rate": 3.7912685188657905e-06, + "loss": 0.1201, + "step": 13790 + }, + { + "epoch": 5.77, + "learning_rate": 3.789939558349929e-06, + "loss": 0.0978, + "step": 13791 + }, + { + "epoch": 5.77, + "learning_rate": 3.7886107763367474e-06, + "loss": 0.0969, + "step": 13792 + }, + { + "epoch": 5.77, + "learning_rate": 3.787282172864435e-06, + "loss": 0.1366, + "step": 13793 + }, + { + "epoch": 5.78, + "learning_rate": 3.785953747971183e-06, + "loss": 0.1331, + "step": 13794 + }, + { + "epoch": 5.78, + "learning_rate": 3.784625501695175e-06, + "loss": 0.1114, + "step": 13795 + }, + { + "epoch": 5.78, + "learning_rate": 3.783297434074592e-06, + "loss": 0.1316, + "step": 13796 + }, + { + "epoch": 5.78, + "learning_rate": 3.781969545147609e-06, + "loss": 0.1393, + "step": 13797 + }, + { + "epoch": 5.78, + "learning_rate": 3.7806418349523956e-06, + "loss": 0.1353, + "step": 13798 + }, + { + "epoch": 5.78, + "learning_rate": 3.7793143035271106e-06, + "loss": 0.1607, + "step": 13799 + }, + { + "epoch": 5.78, + "learning_rate": 3.7779869509099166e-06, + "loss": 0.0987, + "step": 13800 + }, + { + "epoch": 5.78, + "learning_rate": 3.776659777138966e-06, + "loss": 0.0866, + "step": 13801 + }, + { + "epoch": 5.78, + "learning_rate": 3.7753327822524067e-06, + "loss": 0.1508, + "step": 13802 + }, + { + "epoch": 5.78, + "learning_rate": 3.7740059662883843e-06, + "loss": 0.1343, + "step": 13803 + }, + { + "epoch": 5.78, + "learning_rate": 3.7726793292850382e-06, + "loss": 0.1392, + "step": 13804 + }, + { + "epoch": 5.78, + "learning_rate": 3.771352871280495e-06, + "loss": 0.1397, + "step": 13805 + }, + { + "epoch": 5.78, + "learning_rate": 3.7700265923128866e-06, + "loss": 0.087, + "step": 13806 + }, + { + "epoch": 5.78, + "learning_rate": 3.768700492420335e-06, + "loss": 0.1246, + "step": 13807 + }, + { + "epoch": 5.78, + "learning_rate": 3.7673745716409595e-06, + "loss": 0.1021, + "step": 13808 + }, + { + "epoch": 5.78, + "learning_rate": 3.7660488300128706e-06, + "loss": 0.1268, + "step": 13809 + }, + { + "epoch": 5.78, + "learning_rate": 3.7647232675741794e-06, + "loss": 0.0973, + "step": 13810 + }, + { + "epoch": 5.78, + "learning_rate": 3.7633978843629813e-06, + "loss": 0.0975, + "step": 13811 + }, + { + "epoch": 5.78, + "learning_rate": 3.7620726804173767e-06, + "loss": 0.1035, + "step": 13812 + }, + { + "epoch": 5.78, + "learning_rate": 3.7607476557754574e-06, + "loss": 0.1264, + "step": 13813 + }, + { + "epoch": 5.78, + "learning_rate": 3.7594228104753107e-06, + "loss": 0.1345, + "step": 13814 + }, + { + "epoch": 5.78, + "learning_rate": 3.75809814455502e-06, + "loss": 0.1187, + "step": 13815 + }, + { + "epoch": 5.78, + "learning_rate": 3.756773658052657e-06, + "loss": 0.1339, + "step": 13816 + }, + { + "epoch": 5.78, + "learning_rate": 3.755449351006294e-06, + "loss": 0.1081, + "step": 13817 + }, + { + "epoch": 5.79, + "learning_rate": 3.7541252234540028e-06, + "loss": 0.1106, + "step": 13818 + }, + { + "epoch": 5.79, + "learning_rate": 3.752801275433835e-06, + "loss": 0.1587, + "step": 13819 + }, + { + "epoch": 5.79, + "learning_rate": 3.7514775069838503e-06, + "loss": 0.1327, + "step": 13820 + }, + { + "epoch": 5.79, + "learning_rate": 3.750153918142101e-06, + "loss": 0.1187, + "step": 13821 + }, + { + "epoch": 5.79, + "learning_rate": 3.74883050894663e-06, + "loss": 0.1307, + "step": 13822 + }, + { + "epoch": 5.79, + "learning_rate": 3.7475072794354803e-06, + "loss": 0.0944, + "step": 13823 + }, + { + "epoch": 5.79, + "learning_rate": 3.746184229646689e-06, + "loss": 0.0824, + "step": 13824 + }, + { + "epoch": 5.79, + "learning_rate": 3.744861359618278e-06, + "loss": 0.1315, + "step": 13825 + }, + { + "epoch": 5.79, + "learning_rate": 3.7435386693882767e-06, + "loss": 0.1121, + "step": 13826 + }, + { + "epoch": 5.79, + "learning_rate": 3.742216158994705e-06, + "loss": 0.0785, + "step": 13827 + }, + { + "epoch": 5.79, + "learning_rate": 3.740893828475577e-06, + "loss": 0.1419, + "step": 13828 + }, + { + "epoch": 5.79, + "learning_rate": 3.7395716778689006e-06, + "loss": 0.1205, + "step": 13829 + }, + { + "epoch": 5.79, + "learning_rate": 3.738249707212681e-06, + "loss": 0.1453, + "step": 13830 + }, + { + "epoch": 5.79, + "learning_rate": 3.736927916544921e-06, + "loss": 0.129, + "step": 13831 + }, + { + "epoch": 5.79, + "learning_rate": 3.7356063059036073e-06, + "loss": 0.1384, + "step": 13832 + }, + { + "epoch": 5.79, + "learning_rate": 3.73428487532673e-06, + "loss": 0.1283, + "step": 13833 + }, + { + "epoch": 5.79, + "learning_rate": 3.732963624852275e-06, + "loss": 0.1252, + "step": 13834 + }, + { + "epoch": 5.79, + "learning_rate": 3.731642554518219e-06, + "loss": 0.1, + "step": 13835 + }, + { + "epoch": 5.79, + "learning_rate": 3.7303216643625364e-06, + "loss": 0.1462, + "step": 13836 + }, + { + "epoch": 5.79, + "learning_rate": 3.7290009544231955e-06, + "loss": 0.1298, + "step": 13837 + }, + { + "epoch": 5.79, + "learning_rate": 3.727680424738155e-06, + "loss": 0.1198, + "step": 13838 + }, + { + "epoch": 5.79, + "learning_rate": 3.726360075345378e-06, + "loss": 0.1026, + "step": 13839 + }, + { + "epoch": 5.79, + "learning_rate": 3.7250399062828104e-06, + "loss": 0.1226, + "step": 13840 + }, + { + "epoch": 5.79, + "learning_rate": 3.723719917588402e-06, + "loss": 0.1262, + "step": 13841 + }, + { + "epoch": 5.8, + "learning_rate": 3.722400109300095e-06, + "loss": 0.1533, + "step": 13842 + }, + { + "epoch": 5.8, + "learning_rate": 3.7210804814558253e-06, + "loss": 0.1316, + "step": 13843 + }, + { + "epoch": 5.8, + "learning_rate": 3.7197610340935263e-06, + "loss": 0.1205, + "step": 13844 + }, + { + "epoch": 5.8, + "learning_rate": 3.7184417672511273e-06, + "loss": 0.0987, + "step": 13845 + }, + { + "epoch": 5.8, + "learning_rate": 3.7171226809665417e-06, + "loss": 0.1437, + "step": 13846 + }, + { + "epoch": 5.8, + "learning_rate": 3.71580377527769e-06, + "loss": 0.1369, + "step": 13847 + }, + { + "epoch": 5.8, + "learning_rate": 3.7144850502224816e-06, + "loss": 0.135, + "step": 13848 + }, + { + "epoch": 5.8, + "learning_rate": 3.713166505838822e-06, + "loss": 0.102, + "step": 13849 + }, + { + "epoch": 5.8, + "learning_rate": 3.7118481421646134e-06, + "loss": 0.0825, + "step": 13850 + }, + { + "epoch": 5.8, + "learning_rate": 3.710529959237753e-06, + "loss": 0.0986, + "step": 13851 + }, + { + "epoch": 5.8, + "learning_rate": 3.7092119570961248e-06, + "loss": 0.1431, + "step": 13852 + }, + { + "epoch": 5.8, + "learning_rate": 3.7078941357776153e-06, + "loss": 0.0948, + "step": 13853 + }, + { + "epoch": 5.8, + "learning_rate": 3.7065764953201066e-06, + "loss": 0.1081, + "step": 13854 + }, + { + "epoch": 5.8, + "learning_rate": 3.7052590357614717e-06, + "loss": 0.0667, + "step": 13855 + }, + { + "epoch": 5.8, + "learning_rate": 3.7039417571395796e-06, + "loss": 0.1089, + "step": 13856 + }, + { + "epoch": 5.8, + "learning_rate": 3.7026246594922944e-06, + "loss": 0.1096, + "step": 13857 + }, + { + "epoch": 5.8, + "learning_rate": 3.7013077428574794e-06, + "loss": 0.1008, + "step": 13858 + }, + { + "epoch": 5.8, + "learning_rate": 3.6999910072729808e-06, + "loss": 0.1199, + "step": 13859 + }, + { + "epoch": 5.8, + "learning_rate": 3.698674452776653e-06, + "loss": 0.1097, + "step": 13860 + }, + { + "epoch": 5.8, + "learning_rate": 3.6973580794063323e-06, + "loss": 0.1472, + "step": 13861 + }, + { + "epoch": 5.8, + "learning_rate": 3.6960418871998615e-06, + "loss": 0.1396, + "step": 13862 + }, + { + "epoch": 5.8, + "learning_rate": 3.6947258761950723e-06, + "loss": 0.1408, + "step": 13863 + }, + { + "epoch": 5.8, + "learning_rate": 3.693410046429793e-06, + "loss": 0.0973, + "step": 13864 + }, + { + "epoch": 5.8, + "learning_rate": 3.6920943979418478e-06, + "loss": 0.1614, + "step": 13865 + }, + { + "epoch": 5.81, + "learning_rate": 3.690778930769049e-06, + "loss": 0.1382, + "step": 13866 + }, + { + "epoch": 5.81, + "learning_rate": 3.6894636449492116e-06, + "loss": 0.1073, + "step": 13867 + }, + { + "epoch": 5.81, + "learning_rate": 3.688148540520141e-06, + "loss": 0.1038, + "step": 13868 + }, + { + "epoch": 5.81, + "learning_rate": 3.6868336175196406e-06, + "loss": 0.1121, + "step": 13869 + }, + { + "epoch": 5.81, + "learning_rate": 3.6855188759855043e-06, + "loss": 0.125, + "step": 13870 + }, + { + "epoch": 5.81, + "learning_rate": 3.6842043159555263e-06, + "loss": 0.1256, + "step": 13871 + }, + { + "epoch": 5.81, + "learning_rate": 3.6828899374674933e-06, + "loss": 0.0791, + "step": 13872 + }, + { + "epoch": 5.81, + "learning_rate": 3.681575740559179e-06, + "loss": 0.1311, + "step": 13873 + }, + { + "epoch": 5.81, + "learning_rate": 3.680261725268364e-06, + "loss": 0.122, + "step": 13874 + }, + { + "epoch": 5.81, + "learning_rate": 3.678947891632818e-06, + "loss": 0.1136, + "step": 13875 + }, + { + "epoch": 5.81, + "learning_rate": 3.677634239690305e-06, + "loss": 0.0916, + "step": 13876 + }, + { + "epoch": 5.81, + "learning_rate": 3.676320769478585e-06, + "loss": 0.1307, + "step": 13877 + }, + { + "epoch": 5.81, + "learning_rate": 3.6750074810354173e-06, + "loss": 0.1332, + "step": 13878 + }, + { + "epoch": 5.81, + "learning_rate": 3.673694374398543e-06, + "loss": 0.1622, + "step": 13879 + }, + { + "epoch": 5.81, + "learning_rate": 3.672381449605711e-06, + "loss": 0.1185, + "step": 13880 + }, + { + "epoch": 5.81, + "learning_rate": 3.671068706694658e-06, + "loss": 0.1075, + "step": 13881 + }, + { + "epoch": 5.81, + "learning_rate": 3.6697561457031196e-06, + "loss": 0.0994, + "step": 13882 + }, + { + "epoch": 5.81, + "learning_rate": 3.6684437666688253e-06, + "loss": 0.1085, + "step": 13883 + }, + { + "epoch": 5.81, + "learning_rate": 3.667131569629495e-06, + "loss": 0.1145, + "step": 13884 + }, + { + "epoch": 5.81, + "learning_rate": 3.665819554622847e-06, + "loss": 0.1068, + "step": 13885 + }, + { + "epoch": 5.81, + "learning_rate": 3.6645077216865987e-06, + "loss": 0.0967, + "step": 13886 + }, + { + "epoch": 5.81, + "learning_rate": 3.6631960708584504e-06, + "loss": 0.1445, + "step": 13887 + }, + { + "epoch": 5.81, + "learning_rate": 3.6618846021761078e-06, + "loss": 0.1084, + "step": 13888 + }, + { + "epoch": 5.81, + "learning_rate": 3.660573315677268e-06, + "loss": 0.099, + "step": 13889 + }, + { + "epoch": 5.82, + "learning_rate": 3.6592622113996235e-06, + "loss": 0.1187, + "step": 13890 + }, + { + "epoch": 5.82, + "learning_rate": 3.657951289380859e-06, + "loss": 0.0873, + "step": 13891 + }, + { + "epoch": 5.82, + "learning_rate": 3.656640549658661e-06, + "loss": 0.1233, + "step": 13892 + }, + { + "epoch": 5.82, + "learning_rate": 3.6553299922706977e-06, + "loss": 0.0894, + "step": 13893 + }, + { + "epoch": 5.82, + "learning_rate": 3.654019617254644e-06, + "loss": 0.0926, + "step": 13894 + }, + { + "epoch": 5.82, + "learning_rate": 3.652709424648164e-06, + "loss": 0.0958, + "step": 13895 + }, + { + "epoch": 5.82, + "learning_rate": 3.65139941448892e-06, + "loss": 0.1193, + "step": 13896 + }, + { + "epoch": 5.82, + "learning_rate": 3.650089586814566e-06, + "loss": 0.1366, + "step": 13897 + }, + { + "epoch": 5.82, + "learning_rate": 3.648779941662752e-06, + "loss": 0.124, + "step": 13898 + }, + { + "epoch": 5.82, + "learning_rate": 3.6474704790711258e-06, + "loss": 0.1464, + "step": 13899 + }, + { + "epoch": 5.82, + "learning_rate": 3.6461611990773204e-06, + "loss": 0.1439, + "step": 13900 + }, + { + "epoch": 5.82, + "learning_rate": 3.6448521017189732e-06, + "loss": 0.1262, + "step": 13901 + }, + { + "epoch": 5.82, + "learning_rate": 3.6435431870337123e-06, + "loss": 0.1203, + "step": 13902 + }, + { + "epoch": 5.82, + "learning_rate": 3.6422344550591625e-06, + "loss": 0.1148, + "step": 13903 + }, + { + "epoch": 5.82, + "learning_rate": 3.6409259058329404e-06, + "loss": 0.082, + "step": 13904 + }, + { + "epoch": 5.82, + "learning_rate": 3.6396175393926646e-06, + "loss": 0.095, + "step": 13905 + }, + { + "epoch": 5.82, + "learning_rate": 3.6383093557759386e-06, + "loss": 0.1278, + "step": 13906 + }, + { + "epoch": 5.82, + "learning_rate": 3.637001355020361e-06, + "loss": 0.1405, + "step": 13907 + }, + { + "epoch": 5.82, + "learning_rate": 3.6356935371635336e-06, + "loss": 0.0856, + "step": 13908 + }, + { + "epoch": 5.82, + "learning_rate": 3.6343859022430474e-06, + "loss": 0.1269, + "step": 13909 + }, + { + "epoch": 5.82, + "learning_rate": 3.63307845029649e-06, + "loss": 0.1347, + "step": 13910 + }, + { + "epoch": 5.82, + "learning_rate": 3.631771181361442e-06, + "loss": 0.117, + "step": 13911 + }, + { + "epoch": 5.82, + "learning_rate": 3.630464095475481e-06, + "loss": 0.1103, + "step": 13912 + }, + { + "epoch": 5.82, + "learning_rate": 3.6291571926761805e-06, + "loss": 0.1385, + "step": 13913 + }, + { + "epoch": 5.83, + "learning_rate": 3.6278504730011e-06, + "loss": 0.1068, + "step": 13914 + }, + { + "epoch": 5.83, + "learning_rate": 3.626543936487803e-06, + "loss": 0.0955, + "step": 13915 + }, + { + "epoch": 5.83, + "learning_rate": 3.625237583173844e-06, + "loss": 0.0907, + "step": 13916 + }, + { + "epoch": 5.83, + "learning_rate": 3.6239314130967753e-06, + "loss": 0.1233, + "step": 13917 + }, + { + "epoch": 5.83, + "learning_rate": 3.6226254262941398e-06, + "loss": 0.1122, + "step": 13918 + }, + { + "epoch": 5.83, + "learning_rate": 3.6213196228034796e-06, + "loss": 0.1352, + "step": 13919 + }, + { + "epoch": 5.83, + "learning_rate": 3.6200140026623233e-06, + "loss": 0.1447, + "step": 13920 + }, + { + "epoch": 5.83, + "learning_rate": 3.6187085659082034e-06, + "loss": 0.1134, + "step": 13921 + }, + { + "epoch": 5.83, + "learning_rate": 3.6174033125786434e-06, + "loss": 0.1147, + "step": 13922 + }, + { + "epoch": 5.83, + "learning_rate": 3.6160982427111614e-06, + "loss": 0.0617, + "step": 13923 + }, + { + "epoch": 5.83, + "learning_rate": 3.614793356343269e-06, + "loss": 0.1265, + "step": 13924 + }, + { + "epoch": 5.83, + "learning_rate": 3.6134886535124804e-06, + "loss": 0.0609, + "step": 13925 + }, + { + "epoch": 5.83, + "learning_rate": 3.612184134256289e-06, + "loss": 0.1177, + "step": 13926 + }, + { + "epoch": 5.83, + "learning_rate": 3.6108797986121967e-06, + "loss": 0.1149, + "step": 13927 + }, + { + "epoch": 5.83, + "learning_rate": 3.609575646617699e-06, + "loss": 0.1173, + "step": 13928 + }, + { + "epoch": 5.83, + "learning_rate": 3.6082716783102735e-06, + "loss": 0.0993, + "step": 13929 + }, + { + "epoch": 5.83, + "learning_rate": 3.6069678937274076e-06, + "loss": 0.1415, + "step": 13930 + }, + { + "epoch": 5.83, + "learning_rate": 3.6056642929065765e-06, + "loss": 0.088, + "step": 13931 + }, + { + "epoch": 5.83, + "learning_rate": 3.604360875885251e-06, + "loss": 0.1072, + "step": 13932 + }, + { + "epoch": 5.83, + "learning_rate": 3.6030576427009e-06, + "loss": 0.131, + "step": 13933 + }, + { + "epoch": 5.83, + "learning_rate": 3.601754593390978e-06, + "loss": 0.1314, + "step": 13934 + }, + { + "epoch": 5.83, + "learning_rate": 3.600451727992942e-06, + "loss": 0.1126, + "step": 13935 + }, + { + "epoch": 5.83, + "learning_rate": 3.5991490465442413e-06, + "loss": 0.1463, + "step": 13936 + }, + { + "epoch": 5.84, + "learning_rate": 3.597846549082322e-06, + "loss": 0.1579, + "step": 13937 + }, + { + "epoch": 5.84, + "learning_rate": 3.5965442356446224e-06, + "loss": 0.0985, + "step": 13938 + }, + { + "epoch": 5.84, + "learning_rate": 3.595242106268577e-06, + "loss": 0.1163, + "step": 13939 + }, + { + "epoch": 5.84, + "learning_rate": 3.593940160991617e-06, + "loss": 0.1439, + "step": 13940 + }, + { + "epoch": 5.84, + "learning_rate": 3.592638399851158e-06, + "loss": 0.1152, + "step": 13941 + }, + { + "epoch": 5.84, + "learning_rate": 3.5913368228846236e-06, + "loss": 0.1404, + "step": 13942 + }, + { + "epoch": 5.84, + "learning_rate": 3.5900354301294236e-06, + "loss": 0.1272, + "step": 13943 + }, + { + "epoch": 5.84, + "learning_rate": 3.5887342216229694e-06, + "loss": 0.1131, + "step": 13944 + }, + { + "epoch": 5.84, + "learning_rate": 3.587433197402659e-06, + "loss": 0.1285, + "step": 13945 + }, + { + "epoch": 5.84, + "learning_rate": 3.586132357505895e-06, + "loss": 0.1054, + "step": 13946 + }, + { + "epoch": 5.84, + "learning_rate": 3.5848317019700608e-06, + "loss": 0.09, + "step": 13947 + }, + { + "epoch": 5.84, + "learning_rate": 3.5835312308325475e-06, + "loss": 0.1308, + "step": 13948 + }, + { + "epoch": 5.84, + "learning_rate": 3.5822309441307347e-06, + "loss": 0.1002, + "step": 13949 + }, + { + "epoch": 5.84, + "learning_rate": 3.5809308419019985e-06, + "loss": 0.0906, + "step": 13950 + }, + { + "epoch": 5.84, + "learning_rate": 3.5796309241837125e-06, + "loss": 0.1281, + "step": 13951 + }, + { + "epoch": 5.84, + "learning_rate": 3.5783311910132348e-06, + "loss": 0.1261, + "step": 13952 + }, + { + "epoch": 5.84, + "learning_rate": 3.5770316424279283e-06, + "loss": 0.1357, + "step": 13953 + }, + { + "epoch": 5.84, + "learning_rate": 3.5757322784651506e-06, + "loss": 0.1148, + "step": 13954 + }, + { + "epoch": 5.84, + "learning_rate": 3.5744330991622457e-06, + "loss": 0.0935, + "step": 13955 + }, + { + "epoch": 5.84, + "learning_rate": 3.573134104556558e-06, + "loss": 0.0908, + "step": 13956 + }, + { + "epoch": 5.84, + "learning_rate": 3.5718352946854286e-06, + "loss": 0.1236, + "step": 13957 + }, + { + "epoch": 5.84, + "learning_rate": 3.5705366695861886e-06, + "loss": 0.0759, + "step": 13958 + }, + { + "epoch": 5.84, + "learning_rate": 3.569238229296167e-06, + "loss": 0.1156, + "step": 13959 + }, + { + "epoch": 5.84, + "learning_rate": 3.567939973852689e-06, + "loss": 0.0935, + "step": 13960 + }, + { + "epoch": 5.85, + "learning_rate": 3.5666419032930653e-06, + "loss": 0.1497, + "step": 13961 + }, + { + "epoch": 5.85, + "learning_rate": 3.5653440176546115e-06, + "loss": 0.0916, + "step": 13962 + }, + { + "epoch": 5.85, + "learning_rate": 3.564046316974634e-06, + "loss": 0.0757, + "step": 13963 + }, + { + "epoch": 5.85, + "learning_rate": 3.5627488012904333e-06, + "loss": 0.0908, + "step": 13964 + }, + { + "epoch": 5.85, + "learning_rate": 3.5614514706393066e-06, + "loss": 0.1181, + "step": 13965 + }, + { + "epoch": 5.85, + "learning_rate": 3.560154325058547e-06, + "loss": 0.0886, + "step": 13966 + }, + { + "epoch": 5.85, + "learning_rate": 3.558857364585434e-06, + "loss": 0.1648, + "step": 13967 + }, + { + "epoch": 5.85, + "learning_rate": 3.5575605892572495e-06, + "loss": 0.0831, + "step": 13968 + }, + { + "epoch": 5.85, + "learning_rate": 3.556263999111269e-06, + "loss": 0.1066, + "step": 13969 + }, + { + "epoch": 5.85, + "learning_rate": 3.554967594184762e-06, + "loss": 0.1088, + "step": 13970 + }, + { + "epoch": 5.85, + "learning_rate": 3.553671374514992e-06, + "loss": 0.1059, + "step": 13971 + }, + { + "epoch": 5.85, + "learning_rate": 3.552375340139219e-06, + "loss": 0.1399, + "step": 13972 + }, + { + "epoch": 5.85, + "learning_rate": 3.551079491094698e-06, + "loss": 0.1216, + "step": 13973 + }, + { + "epoch": 5.85, + "learning_rate": 3.5497838274186737e-06, + "loss": 0.151, + "step": 13974 + }, + { + "epoch": 5.85, + "learning_rate": 3.5484883491483866e-06, + "loss": 0.1209, + "step": 13975 + }, + { + "epoch": 5.85, + "learning_rate": 3.5471930563210775e-06, + "loss": 0.14, + "step": 13976 + }, + { + "epoch": 5.85, + "learning_rate": 3.5458979489739778e-06, + "loss": 0.1103, + "step": 13977 + }, + { + "epoch": 5.85, + "learning_rate": 3.5446030271443132e-06, + "loss": 0.1041, + "step": 13978 + }, + { + "epoch": 5.85, + "learning_rate": 3.543308290869307e-06, + "loss": 0.0979, + "step": 13979 + }, + { + "epoch": 5.85, + "learning_rate": 3.5420137401861745e-06, + "loss": 0.1379, + "step": 13980 + }, + { + "epoch": 5.85, + "learning_rate": 3.540719375132129e-06, + "loss": 0.089, + "step": 13981 + }, + { + "epoch": 5.85, + "learning_rate": 3.5394251957443704e-06, + "loss": 0.1059, + "step": 13982 + }, + { + "epoch": 5.85, + "learning_rate": 3.538131202060102e-06, + "loss": 0.1206, + "step": 13983 + }, + { + "epoch": 5.85, + "learning_rate": 3.5368373941165167e-06, + "loss": 0.117, + "step": 13984 + }, + { + "epoch": 5.86, + "learning_rate": 3.5355437719508067e-06, + "loss": 0.0811, + "step": 13985 + }, + { + "epoch": 5.86, + "learning_rate": 3.5342503356001535e-06, + "loss": 0.1076, + "step": 13986 + }, + { + "epoch": 5.86, + "learning_rate": 3.5329570851017405e-06, + "loss": 0.112, + "step": 13987 + }, + { + "epoch": 5.86, + "learning_rate": 3.5316640204927342e-06, + "loss": 0.1049, + "step": 13988 + }, + { + "epoch": 5.86, + "learning_rate": 3.5303711418103047e-06, + "loss": 0.1239, + "step": 13989 + }, + { + "epoch": 5.86, + "learning_rate": 3.5290784490916175e-06, + "loss": 0.139, + "step": 13990 + }, + { + "epoch": 5.86, + "learning_rate": 3.527785942373827e-06, + "loss": 0.1611, + "step": 13991 + }, + { + "epoch": 5.86, + "learning_rate": 3.526493621694086e-06, + "loss": 0.1448, + "step": 13992 + }, + { + "epoch": 5.86, + "learning_rate": 3.5252014870895447e-06, + "loss": 0.1252, + "step": 13993 + }, + { + "epoch": 5.86, + "learning_rate": 3.5239095385973377e-06, + "loss": 0.1505, + "step": 13994 + }, + { + "epoch": 5.86, + "learning_rate": 3.522617776254604e-06, + "loss": 0.1341, + "step": 13995 + }, + { + "epoch": 5.86, + "learning_rate": 3.5213262000984793e-06, + "loss": 0.1043, + "step": 13996 + }, + { + "epoch": 5.86, + "learning_rate": 3.52003481016608e-06, + "loss": 0.1103, + "step": 13997 + }, + { + "epoch": 5.86, + "learning_rate": 3.518743606494529e-06, + "loss": 0.1107, + "step": 13998 + }, + { + "epoch": 5.86, + "learning_rate": 3.517452589120941e-06, + "loss": 0.0967, + "step": 13999 + }, + { + "epoch": 5.86, + "learning_rate": 3.516161758082427e-06, + "loss": 0.1111, + "step": 14000 + }, + { + "epoch": 5.86, + "learning_rate": 3.514871113416093e-06, + "loss": 0.0789, + "step": 14001 + }, + { + "epoch": 5.86, + "learning_rate": 3.51358065515903e-06, + "loss": 0.1114, + "step": 14002 + }, + { + "epoch": 5.86, + "learning_rate": 3.5122903833483357e-06, + "loss": 0.1079, + "step": 14003 + }, + { + "epoch": 5.86, + "learning_rate": 3.511000298021098e-06, + "loss": 0.118, + "step": 14004 + }, + { + "epoch": 5.86, + "learning_rate": 3.5097103992143978e-06, + "loss": 0.1236, + "step": 14005 + }, + { + "epoch": 5.86, + "learning_rate": 3.5084206869653126e-06, + "loss": 0.1465, + "step": 14006 + }, + { + "epoch": 5.86, + "learning_rate": 3.507131161310917e-06, + "loss": 0.1022, + "step": 14007 + }, + { + "epoch": 5.86, + "learning_rate": 3.505841822288272e-06, + "loss": 0.0969, + "step": 14008 + }, + { + "epoch": 5.87, + "learning_rate": 3.5045526699344413e-06, + "loss": 0.1046, + "step": 14009 + }, + { + "epoch": 5.87, + "learning_rate": 3.5032637042864793e-06, + "loss": 0.1009, + "step": 14010 + }, + { + "epoch": 5.87, + "learning_rate": 3.5019749253814372e-06, + "loss": 0.1093, + "step": 14011 + }, + { + "epoch": 5.87, + "learning_rate": 3.50068633325636e-06, + "loss": 0.1327, + "step": 14012 + }, + { + "epoch": 5.87, + "learning_rate": 3.499397927948286e-06, + "loss": 0.0909, + "step": 14013 + }, + { + "epoch": 5.87, + "learning_rate": 3.498109709494254e-06, + "loss": 0.1392, + "step": 14014 + }, + { + "epoch": 5.87, + "learning_rate": 3.496821677931286e-06, + "loss": 0.1085, + "step": 14015 + }, + { + "epoch": 5.87, + "learning_rate": 3.495533833296406e-06, + "loss": 0.1361, + "step": 14016 + }, + { + "epoch": 5.87, + "learning_rate": 3.4942461756266356e-06, + "loss": 0.1297, + "step": 14017 + }, + { + "epoch": 5.87, + "learning_rate": 3.4929587049589853e-06, + "loss": 0.1175, + "step": 14018 + }, + { + "epoch": 5.87, + "learning_rate": 3.491671421330467e-06, + "loss": 0.0825, + "step": 14019 + }, + { + "epoch": 5.87, + "learning_rate": 3.4903843247780743e-06, + "loss": 0.1452, + "step": 14020 + }, + { + "epoch": 5.87, + "learning_rate": 3.489097415338808e-06, + "loss": 0.1271, + "step": 14021 + }, + { + "epoch": 5.87, + "learning_rate": 3.4878106930496616e-06, + "loss": 0.1284, + "step": 14022 + }, + { + "epoch": 5.87, + "learning_rate": 3.486524157947615e-06, + "loss": 0.0917, + "step": 14023 + }, + { + "epoch": 5.87, + "learning_rate": 3.4852378100696505e-06, + "loss": 0.1334, + "step": 14024 + }, + { + "epoch": 5.87, + "learning_rate": 3.4839516494527457e-06, + "loss": 0.1086, + "step": 14025 + }, + { + "epoch": 5.87, + "learning_rate": 3.4826656761338673e-06, + "loss": 0.1463, + "step": 14026 + }, + { + "epoch": 5.87, + "learning_rate": 3.481379890149982e-06, + "loss": 0.1149, + "step": 14027 + }, + { + "epoch": 5.87, + "learning_rate": 3.4800942915380485e-06, + "loss": 0.127, + "step": 14028 + }, + { + "epoch": 5.87, + "learning_rate": 3.4788088803350173e-06, + "loss": 0.111, + "step": 14029 + }, + { + "epoch": 5.87, + "learning_rate": 3.477523656577837e-06, + "loss": 0.1165, + "step": 14030 + }, + { + "epoch": 5.87, + "learning_rate": 3.4762386203034527e-06, + "loss": 0.1217, + "step": 14031 + }, + { + "epoch": 5.87, + "learning_rate": 3.474953771548799e-06, + "loss": 0.1365, + "step": 14032 + }, + { + "epoch": 5.88, + "learning_rate": 3.473669110350809e-06, + "loss": 0.1499, + "step": 14033 + }, + { + "epoch": 5.88, + "learning_rate": 3.472384636746412e-06, + "loss": 0.1611, + "step": 14034 + }, + { + "epoch": 5.88, + "learning_rate": 3.471100350772524e-06, + "loss": 0.0953, + "step": 14035 + }, + { + "epoch": 5.88, + "learning_rate": 3.4698162524660615e-06, + "loss": 0.1255, + "step": 14036 + }, + { + "epoch": 5.88, + "learning_rate": 3.468532341863937e-06, + "loss": 0.1541, + "step": 14037 + }, + { + "epoch": 5.88, + "learning_rate": 3.4672486190030543e-06, + "loss": 0.1318, + "step": 14038 + }, + { + "epoch": 5.88, + "learning_rate": 3.4659650839203127e-06, + "loss": 0.1182, + "step": 14039 + }, + { + "epoch": 5.88, + "learning_rate": 3.4646817366526066e-06, + "loss": 0.1264, + "step": 14040 + }, + { + "epoch": 5.88, + "learning_rate": 3.4633985772368274e-06, + "loss": 0.1143, + "step": 14041 + }, + { + "epoch": 5.88, + "learning_rate": 3.4621156057098573e-06, + "loss": 0.1463, + "step": 14042 + }, + { + "epoch": 5.88, + "learning_rate": 3.4608328221085674e-06, + "loss": 0.1425, + "step": 14043 + }, + { + "epoch": 5.88, + "learning_rate": 3.4595502264698356e-06, + "loss": 0.1706, + "step": 14044 + }, + { + "epoch": 5.88, + "learning_rate": 3.4582678188305297e-06, + "loss": 0.1091, + "step": 14045 + }, + { + "epoch": 5.88, + "learning_rate": 3.4569855992275093e-06, + "loss": 0.1409, + "step": 14046 + }, + { + "epoch": 5.88, + "learning_rate": 3.4557035676976313e-06, + "loss": 0.0968, + "step": 14047 + }, + { + "epoch": 5.88, + "learning_rate": 3.454421724277751e-06, + "loss": 0.1727, + "step": 14048 + }, + { + "epoch": 5.88, + "learning_rate": 3.453140069004707e-06, + "loss": 0.1531, + "step": 14049 + }, + { + "epoch": 5.88, + "learning_rate": 3.451858601915341e-06, + "loss": 0.121, + "step": 14050 + }, + { + "epoch": 5.88, + "learning_rate": 3.45057732304649e-06, + "loss": 0.1237, + "step": 14051 + }, + { + "epoch": 5.88, + "learning_rate": 3.449296232434981e-06, + "loss": 0.0952, + "step": 14052 + }, + { + "epoch": 5.88, + "learning_rate": 3.448015330117639e-06, + "loss": 0.1005, + "step": 14053 + }, + { + "epoch": 5.88, + "learning_rate": 3.446734616131282e-06, + "loss": 0.1103, + "step": 14054 + }, + { + "epoch": 5.88, + "learning_rate": 3.4454540905127277e-06, + "loss": 0.1124, + "step": 14055 + }, + { + "epoch": 5.88, + "learning_rate": 3.444173753298775e-06, + "loss": 0.114, + "step": 14056 + }, + { + "epoch": 5.89, + "learning_rate": 3.4428936045262296e-06, + "loss": 0.0952, + "step": 14057 + }, + { + "epoch": 5.89, + "learning_rate": 3.4416136442318894e-06, + "loss": 0.1132, + "step": 14058 + }, + { + "epoch": 5.89, + "learning_rate": 3.440333872452546e-06, + "loss": 0.1363, + "step": 14059 + }, + { + "epoch": 5.89, + "learning_rate": 3.4390542892249833e-06, + "loss": 0.0637, + "step": 14060 + }, + { + "epoch": 5.89, + "learning_rate": 3.437774894585987e-06, + "loss": 0.1097, + "step": 14061 + }, + { + "epoch": 5.89, + "learning_rate": 3.4364956885723246e-06, + "loss": 0.1029, + "step": 14062 + }, + { + "epoch": 5.89, + "learning_rate": 3.435216671220769e-06, + "loss": 0.1439, + "step": 14063 + }, + { + "epoch": 5.89, + "learning_rate": 3.433937842568088e-06, + "loss": 0.1027, + "step": 14064 + }, + { + "epoch": 5.89, + "learning_rate": 3.432659202651034e-06, + "loss": 0.0999, + "step": 14065 + }, + { + "epoch": 5.89, + "learning_rate": 3.4313807515063635e-06, + "loss": 0.1, + "step": 14066 + }, + { + "epoch": 5.89, + "learning_rate": 3.4301024891708245e-06, + "loss": 0.1165, + "step": 14067 + }, + { + "epoch": 5.89, + "learning_rate": 3.42882441568116e-06, + "loss": 0.1132, + "step": 14068 + }, + { + "epoch": 5.89, + "learning_rate": 3.4275465310741083e-06, + "loss": 0.1187, + "step": 14069 + }, + { + "epoch": 5.89, + "learning_rate": 3.426268835386397e-06, + "loss": 0.0904, + "step": 14070 + }, + { + "epoch": 5.89, + "learning_rate": 3.4249913286547553e-06, + "loss": 0.1018, + "step": 14071 + }, + { + "epoch": 5.89, + "learning_rate": 3.4237140109159015e-06, + "loss": 0.1082, + "step": 14072 + }, + { + "epoch": 5.89, + "learning_rate": 3.4224368822065547e-06, + "loss": 0.1081, + "step": 14073 + }, + { + "epoch": 5.89, + "learning_rate": 3.421159942563421e-06, + "loss": 0.1229, + "step": 14074 + }, + { + "epoch": 5.89, + "learning_rate": 3.4198831920232113e-06, + "loss": 0.0917, + "step": 14075 + }, + { + "epoch": 5.89, + "learning_rate": 3.4186066306226174e-06, + "loss": 0.0752, + "step": 14076 + }, + { + "epoch": 5.89, + "learning_rate": 3.4173302583983335e-06, + "loss": 0.1235, + "step": 14077 + }, + { + "epoch": 5.89, + "learning_rate": 3.416054075387052e-06, + "loss": 0.0999, + "step": 14078 + }, + { + "epoch": 5.89, + "learning_rate": 3.414778081625453e-06, + "loss": 0.1152, + "step": 14079 + }, + { + "epoch": 5.89, + "learning_rate": 3.4135022771502146e-06, + "loss": 0.0914, + "step": 14080 + }, + { + "epoch": 5.9, + "learning_rate": 3.4122266619980083e-06, + "loss": 0.0757, + "step": 14081 + }, + { + "epoch": 5.9, + "learning_rate": 3.410951236205503e-06, + "loss": 0.0896, + "step": 14082 + }, + { + "epoch": 5.9, + "learning_rate": 3.4096759998093552e-06, + "loss": 0.1502, + "step": 14083 + }, + { + "epoch": 5.9, + "learning_rate": 3.4084009528462227e-06, + "loss": 0.1271, + "step": 14084 + }, + { + "epoch": 5.9, + "learning_rate": 3.407126095352755e-06, + "loss": 0.1263, + "step": 14085 + }, + { + "epoch": 5.9, + "learning_rate": 3.4058514273655984e-06, + "loss": 0.1307, + "step": 14086 + }, + { + "epoch": 5.9, + "learning_rate": 3.404576948921393e-06, + "loss": 0.1591, + "step": 14087 + }, + { + "epoch": 5.9, + "learning_rate": 3.4033026600567677e-06, + "loss": 0.1583, + "step": 14088 + }, + { + "epoch": 5.9, + "learning_rate": 3.402028560808357e-06, + "loss": 0.1016, + "step": 14089 + }, + { + "epoch": 5.9, + "learning_rate": 3.4007546512127764e-06, + "loss": 0.1096, + "step": 14090 + }, + { + "epoch": 5.9, + "learning_rate": 3.3994809313066477e-06, + "loss": 0.1148, + "step": 14091 + }, + { + "epoch": 5.9, + "learning_rate": 3.398207401126583e-06, + "loss": 0.1304, + "step": 14092 + }, + { + "epoch": 5.9, + "learning_rate": 3.3969340607091884e-06, + "loss": 0.0871, + "step": 14093 + }, + { + "epoch": 5.9, + "learning_rate": 3.3956609100910643e-06, + "loss": 0.1377, + "step": 14094 + }, + { + "epoch": 5.9, + "learning_rate": 3.394387949308806e-06, + "loss": 0.1212, + "step": 14095 + }, + { + "epoch": 5.9, + "learning_rate": 3.3931151783990092e-06, + "loss": 0.1206, + "step": 14096 + }, + { + "epoch": 5.9, + "learning_rate": 3.3918425973982496e-06, + "loss": 0.0933, + "step": 14097 + }, + { + "epoch": 5.9, + "learning_rate": 3.39057020634311e-06, + "loss": 0.0626, + "step": 14098 + }, + { + "epoch": 5.9, + "learning_rate": 3.389298005270165e-06, + "loss": 0.1276, + "step": 14099 + }, + { + "epoch": 5.9, + "learning_rate": 3.3880259942159833e-06, + "loss": 0.1148, + "step": 14100 + }, + { + "epoch": 5.9, + "learning_rate": 3.386754173217126e-06, + "loss": 0.1493, + "step": 14101 + }, + { + "epoch": 5.9, + "learning_rate": 3.3854825423101555e-06, + "loss": 0.1155, + "step": 14102 + }, + { + "epoch": 5.9, + "learning_rate": 3.384211101531617e-06, + "loss": 0.0956, + "step": 14103 + }, + { + "epoch": 5.9, + "learning_rate": 3.382939850918059e-06, + "loss": 0.0938, + "step": 14104 + }, + { + "epoch": 5.91, + "learning_rate": 3.3816687905060232e-06, + "loss": 0.1378, + "step": 14105 + }, + { + "epoch": 5.91, + "learning_rate": 3.380397920332045e-06, + "loss": 0.1013, + "step": 14106 + }, + { + "epoch": 5.91, + "learning_rate": 3.3791272404326535e-06, + "loss": 0.1205, + "step": 14107 + }, + { + "epoch": 5.91, + "learning_rate": 3.3778567508443782e-06, + "loss": 0.106, + "step": 14108 + }, + { + "epoch": 5.91, + "learning_rate": 3.3765864516037317e-06, + "loss": 0.1332, + "step": 14109 + }, + { + "epoch": 5.91, + "learning_rate": 3.3753163427472324e-06, + "loss": 0.1093, + "step": 14110 + }, + { + "epoch": 5.91, + "learning_rate": 3.3740464243113834e-06, + "loss": 0.1576, + "step": 14111 + }, + { + "epoch": 5.91, + "learning_rate": 3.3727766963326903e-06, + "loss": 0.1264, + "step": 14112 + }, + { + "epoch": 5.91, + "learning_rate": 3.37150715884765e-06, + "loss": 0.0976, + "step": 14113 + }, + { + "epoch": 5.91, + "learning_rate": 3.3702378118927537e-06, + "loss": 0.1251, + "step": 14114 + }, + { + "epoch": 5.91, + "learning_rate": 3.3689686555044888e-06, + "loss": 0.1625, + "step": 14115 + }, + { + "epoch": 5.91, + "learning_rate": 3.3676996897193393e-06, + "loss": 0.1327, + "step": 14116 + }, + { + "epoch": 5.91, + "learning_rate": 3.366430914573774e-06, + "loss": 0.1531, + "step": 14117 + }, + { + "epoch": 5.91, + "learning_rate": 3.3651623301042646e-06, + "loss": 0.1045, + "step": 14118 + }, + { + "epoch": 5.91, + "learning_rate": 3.363893936347277e-06, + "loss": 0.1105, + "step": 14119 + }, + { + "epoch": 5.91, + "learning_rate": 3.3626257333392687e-06, + "loss": 0.1239, + "step": 14120 + }, + { + "epoch": 5.91, + "learning_rate": 3.3613577211166936e-06, + "loss": 0.0995, + "step": 14121 + }, + { + "epoch": 5.91, + "learning_rate": 3.3600898997160005e-06, + "loss": 0.1189, + "step": 14122 + }, + { + "epoch": 5.91, + "learning_rate": 3.358822269173634e-06, + "loss": 0.1224, + "step": 14123 + }, + { + "epoch": 5.91, + "learning_rate": 3.357554829526025e-06, + "loss": 0.1318, + "step": 14124 + }, + { + "epoch": 5.91, + "learning_rate": 3.3562875808096083e-06, + "loss": 0.1149, + "step": 14125 + }, + { + "epoch": 5.91, + "learning_rate": 3.355020523060809e-06, + "loss": 0.1103, + "step": 14126 + }, + { + "epoch": 5.91, + "learning_rate": 3.3537536563160487e-06, + "loss": 0.1313, + "step": 14127 + }, + { + "epoch": 5.92, + "learning_rate": 3.352486980611741e-06, + "loss": 0.1419, + "step": 14128 + }, + { + "epoch": 5.92, + "learning_rate": 3.3512204959843e-06, + "loss": 0.1164, + "step": 14129 + }, + { + "epoch": 5.92, + "learning_rate": 3.3499542024701226e-06, + "loss": 0.1058, + "step": 14130 + }, + { + "epoch": 5.92, + "learning_rate": 3.3486881001056104e-06, + "loss": 0.1403, + "step": 14131 + }, + { + "epoch": 5.92, + "learning_rate": 3.3474221889271597e-06, + "loss": 0.0979, + "step": 14132 + }, + { + "epoch": 5.92, + "learning_rate": 3.346156468971151e-06, + "loss": 0.1352, + "step": 14133 + }, + { + "epoch": 5.92, + "learning_rate": 3.3448909402739716e-06, + "loss": 0.1098, + "step": 14134 + }, + { + "epoch": 5.92, + "learning_rate": 3.343625602871995e-06, + "loss": 0.1001, + "step": 14135 + }, + { + "epoch": 5.92, + "learning_rate": 3.3423604568015944e-06, + "loss": 0.139, + "step": 14136 + }, + { + "epoch": 5.92, + "learning_rate": 3.3410955020991377e-06, + "loss": 0.1335, + "step": 14137 + }, + { + "epoch": 5.92, + "learning_rate": 3.3398307388009786e-06, + "loss": 0.1197, + "step": 14138 + }, + { + "epoch": 5.92, + "learning_rate": 3.3385661669434744e-06, + "loss": 0.094, + "step": 14139 + }, + { + "epoch": 5.92, + "learning_rate": 3.3373017865629742e-06, + "loss": 0.1103, + "step": 14140 + }, + { + "epoch": 5.92, + "learning_rate": 3.3360375976958226e-06, + "loss": 0.1332, + "step": 14141 + }, + { + "epoch": 5.92, + "learning_rate": 3.334773600378357e-06, + "loss": 0.1125, + "step": 14142 + }, + { + "epoch": 5.92, + "learning_rate": 3.3335097946469118e-06, + "loss": 0.1256, + "step": 14143 + }, + { + "epoch": 5.92, + "learning_rate": 3.3322461805378094e-06, + "loss": 0.1456, + "step": 14144 + }, + { + "epoch": 5.92, + "learning_rate": 3.3309827580873734e-06, + "loss": 0.1428, + "step": 14145 + }, + { + "epoch": 5.92, + "learning_rate": 3.3297195273319195e-06, + "loss": 0.131, + "step": 14146 + }, + { + "epoch": 5.92, + "learning_rate": 3.32845648830776e-06, + "loss": 0.122, + "step": 14147 + }, + { + "epoch": 5.92, + "learning_rate": 3.3271936410511974e-06, + "loss": 0.1043, + "step": 14148 + }, + { + "epoch": 5.92, + "learning_rate": 3.3259309855985353e-06, + "loss": 0.1327, + "step": 14149 + }, + { + "epoch": 5.92, + "learning_rate": 3.324668521986062e-06, + "loss": 0.1611, + "step": 14150 + }, + { + "epoch": 5.92, + "learning_rate": 3.3234062502500685e-06, + "loss": 0.0954, + "step": 14151 + }, + { + "epoch": 5.93, + "learning_rate": 3.3221441704268375e-06, + "loss": 0.1467, + "step": 14152 + }, + { + "epoch": 5.93, + "learning_rate": 3.3208822825526464e-06, + "loss": 0.0981, + "step": 14153 + }, + { + "epoch": 5.93, + "learning_rate": 3.3196205866637667e-06, + "loss": 0.114, + "step": 14154 + }, + { + "epoch": 5.93, + "learning_rate": 3.3183590827964696e-06, + "loss": 0.1119, + "step": 14155 + }, + { + "epoch": 5.93, + "learning_rate": 3.3170977709870066e-06, + "loss": 0.12, + "step": 14156 + }, + { + "epoch": 5.93, + "learning_rate": 3.3158366512716424e-06, + "loss": 0.1203, + "step": 14157 + }, + { + "epoch": 5.93, + "learning_rate": 3.314575723686618e-06, + "loss": 0.1199, + "step": 14158 + }, + { + "epoch": 5.93, + "learning_rate": 3.3133149882681836e-06, + "loss": 0.1252, + "step": 14159 + }, + { + "epoch": 5.93, + "learning_rate": 3.3120544450525747e-06, + "loss": 0.095, + "step": 14160 + }, + { + "epoch": 5.93, + "learning_rate": 3.3107940940760264e-06, + "loss": 0.1227, + "step": 14161 + }, + { + "epoch": 5.93, + "learning_rate": 3.3095339353747657e-06, + "loss": 0.0897, + "step": 14162 + }, + { + "epoch": 5.93, + "learning_rate": 3.308273968985015e-06, + "loss": 0.1321, + "step": 14163 + }, + { + "epoch": 5.93, + "learning_rate": 3.3070141949429945e-06, + "loss": 0.0816, + "step": 14164 + }, + { + "epoch": 5.93, + "learning_rate": 3.3057546132849095e-06, + "loss": 0.1089, + "step": 14165 + }, + { + "epoch": 5.93, + "learning_rate": 3.304495224046966e-06, + "loss": 0.1081, + "step": 14166 + }, + { + "epoch": 5.93, + "learning_rate": 3.3032360272653664e-06, + "loss": 0.1332, + "step": 14167 + }, + { + "epoch": 5.93, + "learning_rate": 3.3019770229763048e-06, + "loss": 0.1229, + "step": 14168 + }, + { + "epoch": 5.93, + "learning_rate": 3.30071821121597e-06, + "loss": 0.1548, + "step": 14169 + }, + { + "epoch": 5.93, + "learning_rate": 3.2994595920205485e-06, + "loss": 0.1346, + "step": 14170 + }, + { + "epoch": 5.93, + "learning_rate": 3.2982011654262112e-06, + "loss": 0.1042, + "step": 14171 + }, + { + "epoch": 5.93, + "learning_rate": 3.296942931469136e-06, + "loss": 0.1263, + "step": 14172 + }, + { + "epoch": 5.93, + "learning_rate": 3.2956848901854863e-06, + "loss": 0.1211, + "step": 14173 + }, + { + "epoch": 5.93, + "learning_rate": 3.2944270416114256e-06, + "loss": 0.1317, + "step": 14174 + }, + { + "epoch": 5.93, + "learning_rate": 3.2931693857831094e-06, + "loss": 0.1582, + "step": 14175 + }, + { + "epoch": 5.94, + "learning_rate": 3.2919119227366913e-06, + "loss": 0.1371, + "step": 14176 + }, + { + "epoch": 5.94, + "learning_rate": 3.2906546525083082e-06, + "loss": 0.14, + "step": 14177 + }, + { + "epoch": 5.94, + "learning_rate": 3.2893975751341078e-06, + "loss": 0.1377, + "step": 14178 + }, + { + "epoch": 5.94, + "learning_rate": 3.2881406906502146e-06, + "loss": 0.134, + "step": 14179 + }, + { + "epoch": 5.94, + "learning_rate": 3.2868839990927627e-06, + "loss": 0.1461, + "step": 14180 + }, + { + "epoch": 5.94, + "learning_rate": 3.285627500497872e-06, + "loss": 0.1224, + "step": 14181 + }, + { + "epoch": 5.94, + "learning_rate": 3.284371194901661e-06, + "loss": 0.1106, + "step": 14182 + }, + { + "epoch": 5.94, + "learning_rate": 3.2831150823402403e-06, + "loss": 0.1034, + "step": 14183 + }, + { + "epoch": 5.94, + "learning_rate": 3.28185916284972e-06, + "loss": 0.1049, + "step": 14184 + }, + { + "epoch": 5.94, + "learning_rate": 3.280603436466192e-06, + "loss": 0.0994, + "step": 14185 + }, + { + "epoch": 5.94, + "learning_rate": 3.279347903225757e-06, + "loss": 0.1359, + "step": 14186 + }, + { + "epoch": 5.94, + "learning_rate": 3.278092563164502e-06, + "loss": 0.1302, + "step": 14187 + }, + { + "epoch": 5.94, + "learning_rate": 3.2768374163185114e-06, + "loss": 0.0997, + "step": 14188 + }, + { + "epoch": 5.94, + "learning_rate": 3.275582462723863e-06, + "loss": 0.1485, + "step": 14189 + }, + { + "epoch": 5.94, + "learning_rate": 3.274327702416633e-06, + "loss": 0.1328, + "step": 14190 + }, + { + "epoch": 5.94, + "learning_rate": 3.2730731354328827e-06, + "loss": 0.0914, + "step": 14191 + }, + { + "epoch": 5.94, + "learning_rate": 3.2718187618086737e-06, + "loss": 0.1373, + "step": 14192 + }, + { + "epoch": 5.94, + "learning_rate": 3.2705645815800658e-06, + "loss": 0.1274, + "step": 14193 + }, + { + "epoch": 5.94, + "learning_rate": 3.269310594783106e-06, + "loss": 0.0989, + "step": 14194 + }, + { + "epoch": 5.94, + "learning_rate": 3.2680568014538415e-06, + "loss": 0.1325, + "step": 14195 + }, + { + "epoch": 5.94, + "learning_rate": 3.2668032016283103e-06, + "loss": 0.1238, + "step": 14196 + }, + { + "epoch": 5.94, + "learning_rate": 3.2655497953425484e-06, + "loss": 0.1115, + "step": 14197 + }, + { + "epoch": 5.94, + "learning_rate": 3.26429658263258e-06, + "loss": 0.1297, + "step": 14198 + }, + { + "epoch": 5.94, + "learning_rate": 3.2630435635344283e-06, + "loss": 0.0975, + "step": 14199 + }, + { + "epoch": 5.95, + "learning_rate": 3.261790738084114e-06, + "loss": 0.1321, + "step": 14200 + }, + { + "epoch": 5.95, + "learning_rate": 3.2605381063176423e-06, + "loss": 0.0961, + "step": 14201 + }, + { + "epoch": 5.95, + "learning_rate": 3.259285668271023e-06, + "loss": 0.1038, + "step": 14202 + }, + { + "epoch": 5.95, + "learning_rate": 3.2580334239802545e-06, + "loss": 0.1285, + "step": 14203 + }, + { + "epoch": 5.95, + "learning_rate": 3.2567813734813336e-06, + "loss": 0.0986, + "step": 14204 + }, + { + "epoch": 5.95, + "learning_rate": 3.255529516810251e-06, + "loss": 0.128, + "step": 14205 + }, + { + "epoch": 5.95, + "learning_rate": 3.254277854002984e-06, + "loss": 0.0905, + "step": 14206 + }, + { + "epoch": 5.95, + "learning_rate": 3.2530263850955147e-06, + "loss": 0.1159, + "step": 14207 + }, + { + "epoch": 5.95, + "learning_rate": 3.251775110123814e-06, + "loss": 0.1275, + "step": 14208 + }, + { + "epoch": 5.95, + "learning_rate": 3.2505240291238505e-06, + "loss": 0.1211, + "step": 14209 + }, + { + "epoch": 5.95, + "learning_rate": 3.2492731421315838e-06, + "loss": 0.0921, + "step": 14210 + }, + { + "epoch": 5.95, + "learning_rate": 3.2480224491829728e-06, + "loss": 0.1079, + "step": 14211 + }, + { + "epoch": 5.95, + "learning_rate": 3.2467719503139626e-06, + "loss": 0.1004, + "step": 14212 + }, + { + "epoch": 5.95, + "learning_rate": 3.2455216455605e-06, + "loss": 0.1377, + "step": 14213 + }, + { + "epoch": 5.95, + "learning_rate": 3.244271534958523e-06, + "loss": 0.1096, + "step": 14214 + }, + { + "epoch": 5.95, + "learning_rate": 3.2430216185439667e-06, + "loss": 0.102, + "step": 14215 + }, + { + "epoch": 5.95, + "learning_rate": 3.2417718963527576e-06, + "loss": 0.1079, + "step": 14216 + }, + { + "epoch": 5.95, + "learning_rate": 3.2405223684208208e-06, + "loss": 0.12, + "step": 14217 + }, + { + "epoch": 5.95, + "learning_rate": 3.2392730347840686e-06, + "loss": 0.1327, + "step": 14218 + }, + { + "epoch": 5.95, + "learning_rate": 3.2380238954784115e-06, + "loss": 0.1363, + "step": 14219 + }, + { + "epoch": 5.95, + "learning_rate": 3.2367749505397582e-06, + "loss": 0.1274, + "step": 14220 + }, + { + "epoch": 5.95, + "learning_rate": 3.2355262000040077e-06, + "loss": 0.1217, + "step": 14221 + }, + { + "epoch": 5.95, + "learning_rate": 3.2342776439070545e-06, + "loss": 0.087, + "step": 14222 + }, + { + "epoch": 5.95, + "learning_rate": 3.2330292822847885e-06, + "loss": 0.1326, + "step": 14223 + }, + { + "epoch": 5.96, + "learning_rate": 3.231781115173088e-06, + "loss": 0.0972, + "step": 14224 + }, + { + "epoch": 5.96, + "learning_rate": 3.2305331426078367e-06, + "loss": 0.1557, + "step": 14225 + }, + { + "epoch": 5.96, + "learning_rate": 3.2292853646249002e-06, + "loss": 0.1304, + "step": 14226 + }, + { + "epoch": 5.96, + "learning_rate": 3.228037781260147e-06, + "loss": 0.1231, + "step": 14227 + }, + { + "epoch": 5.96, + "learning_rate": 3.2267903925494394e-06, + "loss": 0.1325, + "step": 14228 + }, + { + "epoch": 5.96, + "learning_rate": 3.22554319852863e-06, + "loss": 0.1139, + "step": 14229 + }, + { + "epoch": 5.96, + "learning_rate": 3.2242961992335707e-06, + "loss": 0.0909, + "step": 14230 + }, + { + "epoch": 5.96, + "learning_rate": 3.2230493947001084e-06, + "loss": 0.1008, + "step": 14231 + }, + { + "epoch": 5.96, + "learning_rate": 3.221802784964073e-06, + "loss": 0.1182, + "step": 14232 + }, + { + "epoch": 5.96, + "learning_rate": 3.2205563700613028e-06, + "loss": 0.123, + "step": 14233 + }, + { + "epoch": 5.96, + "learning_rate": 3.2193101500276226e-06, + "loss": 0.0822, + "step": 14234 + }, + { + "epoch": 5.96, + "learning_rate": 3.2180641248988566e-06, + "loss": 0.1333, + "step": 14235 + }, + { + "epoch": 5.96, + "learning_rate": 3.2168182947108173e-06, + "loss": 0.0668, + "step": 14236 + }, + { + "epoch": 5.96, + "learning_rate": 3.2155726594993177e-06, + "loss": 0.1355, + "step": 14237 + }, + { + "epoch": 5.96, + "learning_rate": 3.2143272193001652e-06, + "loss": 0.1102, + "step": 14238 + }, + { + "epoch": 5.96, + "learning_rate": 3.2130819741491513e-06, + "loss": 0.1288, + "step": 14239 + }, + { + "epoch": 5.96, + "learning_rate": 3.2118369240820745e-06, + "loss": 0.1122, + "step": 14240 + }, + { + "epoch": 5.96, + "learning_rate": 3.2105920691347216e-06, + "loss": 0.1103, + "step": 14241 + }, + { + "epoch": 5.96, + "learning_rate": 3.2093474093428733e-06, + "loss": 0.118, + "step": 14242 + }, + { + "epoch": 5.96, + "learning_rate": 3.2081029447423095e-06, + "loss": 0.1368, + "step": 14243 + }, + { + "epoch": 5.96, + "learning_rate": 3.2068586753688014e-06, + "loss": 0.1325, + "step": 14244 + }, + { + "epoch": 5.96, + "learning_rate": 3.20561460125811e-06, + "loss": 0.0986, + "step": 14245 + }, + { + "epoch": 5.96, + "learning_rate": 3.2043707224460006e-06, + "loss": 0.1024, + "step": 14246 + }, + { + "epoch": 5.96, + "learning_rate": 3.203127038968222e-06, + "loss": 0.1025, + "step": 14247 + }, + { + "epoch": 5.97, + "learning_rate": 3.201883550860525e-06, + "loss": 0.1007, + "step": 14248 + }, + { + "epoch": 5.97, + "learning_rate": 3.2006402581586527e-06, + "loss": 0.0998, + "step": 14249 + }, + { + "epoch": 5.97, + "learning_rate": 3.199397160898342e-06, + "loss": 0.1018, + "step": 14250 + }, + { + "epoch": 5.97, + "learning_rate": 3.1981542591153247e-06, + "loss": 0.1279, + "step": 14251 + }, + { + "epoch": 5.97, + "learning_rate": 3.196911552845332e-06, + "loss": 0.1581, + "step": 14252 + }, + { + "epoch": 5.97, + "learning_rate": 3.1956690421240754e-06, + "loss": 0.1524, + "step": 14253 + }, + { + "epoch": 5.97, + "learning_rate": 3.1944267269872754e-06, + "loss": 0.147, + "step": 14254 + }, + { + "epoch": 5.97, + "learning_rate": 3.1931846074706386e-06, + "loss": 0.1009, + "step": 14255 + }, + { + "epoch": 5.97, + "learning_rate": 3.191942683609871e-06, + "loss": 0.108, + "step": 14256 + }, + { + "epoch": 5.97, + "learning_rate": 3.190700955440669e-06, + "loss": 0.1121, + "step": 14257 + }, + { + "epoch": 5.97, + "learning_rate": 3.18945942299873e-06, + "loss": 0.1387, + "step": 14258 + }, + { + "epoch": 5.97, + "learning_rate": 3.1882180863197308e-06, + "loss": 0.121, + "step": 14259 + }, + { + "epoch": 5.97, + "learning_rate": 3.1869769454393596e-06, + "loss": 0.1329, + "step": 14260 + }, + { + "epoch": 5.97, + "learning_rate": 3.18573600039329e-06, + "loss": 0.1233, + "step": 14261 + }, + { + "epoch": 5.97, + "learning_rate": 3.1844952512171924e-06, + "loss": 0.094, + "step": 14262 + }, + { + "epoch": 5.97, + "learning_rate": 3.1832546979467304e-06, + "loss": 0.1106, + "step": 14263 + }, + { + "epoch": 5.97, + "learning_rate": 3.1820143406175642e-06, + "loss": 0.1171, + "step": 14264 + }, + { + "epoch": 5.97, + "learning_rate": 3.180774179265348e-06, + "loss": 0.1091, + "step": 14265 + }, + { + "epoch": 5.97, + "learning_rate": 3.1795342139257233e-06, + "loss": 0.1443, + "step": 14266 + }, + { + "epoch": 5.97, + "learning_rate": 3.1782944446343357e-06, + "loss": 0.1002, + "step": 14267 + }, + { + "epoch": 5.97, + "learning_rate": 3.1770548714268236e-06, + "loss": 0.0809, + "step": 14268 + }, + { + "epoch": 5.97, + "learning_rate": 3.175815494338811e-06, + "loss": 0.1243, + "step": 14269 + }, + { + "epoch": 5.97, + "learning_rate": 3.174576313405927e-06, + "loss": 0.1191, + "step": 14270 + }, + { + "epoch": 5.97, + "learning_rate": 3.17333732866379e-06, + "loss": 0.1145, + "step": 14271 + }, + { + "epoch": 5.98, + "learning_rate": 3.1720985401480176e-06, + "loss": 0.1007, + "step": 14272 + }, + { + "epoch": 5.98, + "learning_rate": 3.1708599478942094e-06, + "loss": 0.1246, + "step": 14273 + }, + { + "epoch": 5.98, + "learning_rate": 3.1696215519379726e-06, + "loss": 0.1157, + "step": 14274 + }, + { + "epoch": 5.98, + "learning_rate": 3.168383352314902e-06, + "loss": 0.1044, + "step": 14275 + }, + { + "epoch": 5.98, + "learning_rate": 3.1671453490605906e-06, + "loss": 0.1279, + "step": 14276 + }, + { + "epoch": 5.98, + "learning_rate": 3.1659075422106233e-06, + "loss": 0.1473, + "step": 14277 + }, + { + "epoch": 5.98, + "learning_rate": 3.164669931800578e-06, + "loss": 0.1236, + "step": 14278 + }, + { + "epoch": 5.98, + "learning_rate": 3.163432517866034e-06, + "loss": 0.1409, + "step": 14279 + }, + { + "epoch": 5.98, + "learning_rate": 3.162195300442552e-06, + "loss": 0.1223, + "step": 14280 + }, + { + "epoch": 5.98, + "learning_rate": 3.160958279565699e-06, + "loss": 0.1252, + "step": 14281 + }, + { + "epoch": 5.98, + "learning_rate": 3.15972145527103e-06, + "loss": 0.1619, + "step": 14282 + }, + { + "epoch": 5.98, + "learning_rate": 3.158484827594099e-06, + "loss": 0.0831, + "step": 14283 + }, + { + "epoch": 5.98, + "learning_rate": 3.1572483965704503e-06, + "loss": 0.1201, + "step": 14284 + }, + { + "epoch": 5.98, + "learning_rate": 3.156012162235628e-06, + "loss": 0.1578, + "step": 14285 + }, + { + "epoch": 5.98, + "learning_rate": 3.1547761246251587e-06, + "loss": 0.1137, + "step": 14286 + }, + { + "epoch": 5.98, + "learning_rate": 3.1535402837745753e-06, + "loss": 0.1081, + "step": 14287 + }, + { + "epoch": 5.98, + "learning_rate": 3.1523046397194015e-06, + "loss": 0.1271, + "step": 14288 + }, + { + "epoch": 5.98, + "learning_rate": 3.151069192495154e-06, + "loss": 0.1111, + "step": 14289 + }, + { + "epoch": 5.98, + "learning_rate": 3.149833942137346e-06, + "loss": 0.1529, + "step": 14290 + }, + { + "epoch": 5.98, + "learning_rate": 3.1485988886814846e-06, + "loss": 0.1303, + "step": 14291 + }, + { + "epoch": 5.98, + "learning_rate": 3.147364032163065e-06, + "loss": 0.1105, + "step": 14292 + }, + { + "epoch": 5.98, + "learning_rate": 3.146129372617589e-06, + "loss": 0.1075, + "step": 14293 + }, + { + "epoch": 5.98, + "learning_rate": 3.1448949100805394e-06, + "loss": 0.107, + "step": 14294 + }, + { + "epoch": 5.98, + "learning_rate": 3.1436606445874017e-06, + "loss": 0.1458, + "step": 14295 + }, + { + "epoch": 5.99, + "learning_rate": 3.142426576173655e-06, + "loss": 0.1445, + "step": 14296 + }, + { + "epoch": 5.99, + "learning_rate": 3.1411927048747715e-06, + "loss": 0.1194, + "step": 14297 + }, + { + "epoch": 5.99, + "learning_rate": 3.1399590307262175e-06, + "loss": 0.1093, + "step": 14298 + }, + { + "epoch": 5.99, + "learning_rate": 3.1387255537634553e-06, + "loss": 0.1034, + "step": 14299 + }, + { + "epoch": 5.99, + "learning_rate": 3.137492274021937e-06, + "loss": 0.0787, + "step": 14300 + }, + { + "epoch": 5.99, + "learning_rate": 3.1362591915371123e-06, + "loss": 0.1017, + "step": 14301 + }, + { + "epoch": 5.99, + "learning_rate": 3.1350263063444264e-06, + "loss": 0.1271, + "step": 14302 + }, + { + "epoch": 5.99, + "learning_rate": 3.1337936184793185e-06, + "loss": 0.1039, + "step": 14303 + }, + { + "epoch": 5.99, + "learning_rate": 3.1325611279772194e-06, + "loss": 0.0875, + "step": 14304 + }, + { + "epoch": 5.99, + "learning_rate": 3.131328834873557e-06, + "loss": 0.1423, + "step": 14305 + }, + { + "epoch": 5.99, + "learning_rate": 3.130096739203755e-06, + "loss": 0.1137, + "step": 14306 + }, + { + "epoch": 5.99, + "learning_rate": 3.128864841003223e-06, + "loss": 0.107, + "step": 14307 + }, + { + "epoch": 5.99, + "learning_rate": 3.1276331403073733e-06, + "loss": 0.1402, + "step": 14308 + }, + { + "epoch": 5.99, + "learning_rate": 3.1264016371516106e-06, + "loss": 0.1349, + "step": 14309 + }, + { + "epoch": 5.99, + "learning_rate": 3.1251703315713333e-06, + "loss": 0.1357, + "step": 14310 + }, + { + "epoch": 5.99, + "learning_rate": 3.123939223601935e-06, + "loss": 0.156, + "step": 14311 + }, + { + "epoch": 5.99, + "learning_rate": 3.122708313278805e-06, + "loss": 0.1171, + "step": 14312 + }, + { + "epoch": 5.99, + "learning_rate": 3.1214776006373205e-06, + "loss": 0.1377, + "step": 14313 + }, + { + "epoch": 5.99, + "learning_rate": 3.1202470857128564e-06, + "loss": 0.1145, + "step": 14314 + }, + { + "epoch": 5.99, + "learning_rate": 3.1190167685407848e-06, + "loss": 0.0946, + "step": 14315 + }, + { + "epoch": 5.99, + "learning_rate": 3.11778664915647e-06, + "loss": 0.1007, + "step": 14316 + }, + { + "epoch": 5.99, + "learning_rate": 3.1165567275952712e-06, + "loss": 0.1388, + "step": 14317 + }, + { + "epoch": 5.99, + "learning_rate": 3.1153270038925397e-06, + "loss": 0.1399, + "step": 14318 + }, + { + "epoch": 5.99, + "learning_rate": 3.1140974780836254e-06, + "loss": 0.1044, + "step": 14319 + }, + { + "epoch": 6.0, + "learning_rate": 3.1128681502038717e-06, + "loss": 0.1562, + "step": 14320 + }, + { + "epoch": 6.0, + "learning_rate": 3.1116390202886083e-06, + "loss": 0.1052, + "step": 14321 + }, + { + "epoch": 6.0, + "learning_rate": 3.110410088373168e-06, + "loss": 0.1, + "step": 14322 + }, + { + "epoch": 6.0, + "learning_rate": 3.1091813544928763e-06, + "loss": 0.1175, + "step": 14323 + }, + { + "epoch": 6.0, + "learning_rate": 3.107952818683052e-06, + "loss": 0.1078, + "step": 14324 + }, + { + "epoch": 6.0, + "learning_rate": 3.1067244809790076e-06, + "loss": 0.1231, + "step": 14325 + }, + { + "epoch": 6.0, + "learning_rate": 3.1054963414160545e-06, + "loss": 0.1265, + "step": 14326 + }, + { + "epoch": 6.0, + "learning_rate": 3.1042684000294876e-06, + "loss": 0.118, + "step": 14327 + }, + { + "epoch": 6.0, + "learning_rate": 3.1030406568546066e-06, + "loss": 0.1147, + "step": 14328 + }, + { + "epoch": 6.0, + "learning_rate": 3.1018131119267014e-06, + "loss": 0.0994, + "step": 14329 + }, + { + "epoch": 6.0, + "learning_rate": 3.100585765281057e-06, + "loss": 0.1185, + "step": 14330 + }, + { + "epoch": 6.0, + "learning_rate": 3.099358616952952e-06, + "loss": 0.1292, + "step": 14331 + }, + { + "epoch": 6.0, + "learning_rate": 3.098131666977664e-06, + "loss": 0.1093, + "step": 14332 + }, + { + "epoch": 6.0, + "learning_rate": 3.096904915390454e-06, + "loss": 0.0678, + "step": 14333 + }, + { + "epoch": 6.0, + "learning_rate": 3.095678362226585e-06, + "loss": 0.1091, + "step": 14334 + }, + { + "epoch": 6.0, + "learning_rate": 3.094452007521316e-06, + "loss": 0.1122, + "step": 14335 + }, + { + "epoch": 6.0, + "learning_rate": 3.0932258513098978e-06, + "loss": 0.0605, + "step": 14336 + }, + { + "epoch": 6.0, + "learning_rate": 3.091999893627572e-06, + "loss": 0.071, + "step": 14337 + }, + { + "epoch": 6.0, + "learning_rate": 3.090774134509579e-06, + "loss": 0.0975, + "step": 14338 + }, + { + "epoch": 6.0, + "learning_rate": 3.089548573991151e-06, + "loss": 0.0612, + "step": 14339 + }, + { + "epoch": 6.0, + "learning_rate": 3.088323212107521e-06, + "loss": 0.1026, + "step": 14340 + }, + { + "epoch": 6.0, + "learning_rate": 3.087098048893904e-06, + "loss": 0.0785, + "step": 14341 + }, + { + "epoch": 6.0, + "learning_rate": 3.0858730843855187e-06, + "loss": 0.0928, + "step": 14342 + }, + { + "epoch": 6.01, + "learning_rate": 3.084648318617577e-06, + "loss": 0.0686, + "step": 14343 + }, + { + "epoch": 6.01, + "learning_rate": 3.0834237516252817e-06, + "loss": 0.0737, + "step": 14344 + }, + { + "epoch": 6.01, + "learning_rate": 3.0821993834438337e-06, + "loss": 0.1265, + "step": 14345 + }, + { + "epoch": 6.01, + "learning_rate": 3.0809752141084248e-06, + "loss": 0.0865, + "step": 14346 + }, + { + "epoch": 6.01, + "learning_rate": 3.079751243654248e-06, + "loss": 0.0877, + "step": 14347 + }, + { + "epoch": 6.01, + "learning_rate": 3.078527472116477e-06, + "loss": 0.0758, + "step": 14348 + }, + { + "epoch": 6.01, + "learning_rate": 3.0773038995302918e-06, + "loss": 0.0773, + "step": 14349 + }, + { + "epoch": 6.01, + "learning_rate": 3.0760805259308626e-06, + "loss": 0.0597, + "step": 14350 + }, + { + "epoch": 6.01, + "learning_rate": 3.0748573513533554e-06, + "loss": 0.1065, + "step": 14351 + }, + { + "epoch": 6.01, + "learning_rate": 3.0736343758329277e-06, + "loss": 0.0947, + "step": 14352 + }, + { + "epoch": 6.01, + "learning_rate": 3.0724115994047367e-06, + "loss": 0.0626, + "step": 14353 + }, + { + "epoch": 6.01, + "learning_rate": 3.071189022103923e-06, + "loss": 0.0927, + "step": 14354 + }, + { + "epoch": 6.01, + "learning_rate": 3.069966643965634e-06, + "loss": 0.077, + "step": 14355 + }, + { + "epoch": 6.01, + "learning_rate": 3.068744465025003e-06, + "loss": 0.1038, + "step": 14356 + }, + { + "epoch": 6.01, + "learning_rate": 3.0675224853171613e-06, + "loss": 0.0589, + "step": 14357 + }, + { + "epoch": 6.01, + "learning_rate": 3.0663007048772354e-06, + "loss": 0.0928, + "step": 14358 + }, + { + "epoch": 6.01, + "learning_rate": 3.0650791237403456e-06, + "loss": 0.0821, + "step": 14359 + }, + { + "epoch": 6.01, + "learning_rate": 3.063857741941598e-06, + "loss": 0.066, + "step": 14360 + }, + { + "epoch": 6.01, + "learning_rate": 3.06263655951611e-06, + "loss": 0.0888, + "step": 14361 + }, + { + "epoch": 6.01, + "learning_rate": 3.061415576498973e-06, + "loss": 0.0679, + "step": 14362 + }, + { + "epoch": 6.01, + "learning_rate": 3.0601947929252883e-06, + "loss": 0.0956, + "step": 14363 + }, + { + "epoch": 6.01, + "learning_rate": 3.058974208830147e-06, + "loss": 0.0821, + "step": 14364 + }, + { + "epoch": 6.01, + "learning_rate": 3.0577538242486317e-06, + "loss": 0.0883, + "step": 14365 + }, + { + "epoch": 6.01, + "learning_rate": 3.0565336392158228e-06, + "loss": 0.0893, + "step": 14366 + }, + { + "epoch": 6.02, + "learning_rate": 3.055313653766796e-06, + "loss": 0.0802, + "step": 14367 + }, + { + "epoch": 6.02, + "learning_rate": 3.0540938679366126e-06, + "loss": 0.0919, + "step": 14368 + }, + { + "epoch": 6.02, + "learning_rate": 3.0528742817603364e-06, + "loss": 0.0956, + "step": 14369 + }, + { + "epoch": 6.02, + "learning_rate": 3.0516548952730253e-06, + "loss": 0.0819, + "step": 14370 + }, + { + "epoch": 6.02, + "learning_rate": 3.050435708509728e-06, + "loss": 0.0625, + "step": 14371 + }, + { + "epoch": 6.02, + "learning_rate": 3.049216721505489e-06, + "loss": 0.0944, + "step": 14372 + }, + { + "epoch": 6.02, + "learning_rate": 3.0479979342953504e-06, + "loss": 0.0734, + "step": 14373 + }, + { + "epoch": 6.02, + "learning_rate": 3.046779346914339e-06, + "loss": 0.0737, + "step": 14374 + }, + { + "epoch": 6.02, + "learning_rate": 3.0455609593974855e-06, + "loss": 0.0908, + "step": 14375 + }, + { + "epoch": 6.02, + "learning_rate": 3.044342771779811e-06, + "loss": 0.0815, + "step": 14376 + }, + { + "epoch": 6.02, + "learning_rate": 3.0431247840963316e-06, + "loss": 0.1037, + "step": 14377 + }, + { + "epoch": 6.02, + "learning_rate": 3.041906996382056e-06, + "loss": 0.0726, + "step": 14378 + }, + { + "epoch": 6.02, + "learning_rate": 3.0406894086719894e-06, + "loss": 0.1095, + "step": 14379 + }, + { + "epoch": 6.02, + "learning_rate": 3.039472021001134e-06, + "loss": 0.0771, + "step": 14380 + }, + { + "epoch": 6.02, + "learning_rate": 3.0382548334044783e-06, + "loss": 0.0466, + "step": 14381 + }, + { + "epoch": 6.02, + "learning_rate": 3.0370378459170058e-06, + "loss": 0.0773, + "step": 14382 + }, + { + "epoch": 6.02, + "learning_rate": 3.0358210585737024e-06, + "loss": 0.0787, + "step": 14383 + }, + { + "epoch": 6.02, + "learning_rate": 3.0346044714095425e-06, + "loss": 0.0997, + "step": 14384 + }, + { + "epoch": 6.02, + "learning_rate": 3.0333880844594963e-06, + "loss": 0.0779, + "step": 14385 + }, + { + "epoch": 6.02, + "learning_rate": 3.032171897758527e-06, + "loss": 0.0939, + "step": 14386 + }, + { + "epoch": 6.02, + "learning_rate": 3.030955911341594e-06, + "loss": 0.1008, + "step": 14387 + }, + { + "epoch": 6.02, + "learning_rate": 3.029740125243652e-06, + "loss": 0.0828, + "step": 14388 + }, + { + "epoch": 6.02, + "learning_rate": 3.028524539499642e-06, + "loss": 0.1049, + "step": 14389 + }, + { + "epoch": 6.02, + "learning_rate": 3.0273091541445065e-06, + "loss": 0.0757, + "step": 14390 + }, + { + "epoch": 6.03, + "learning_rate": 3.026093969213183e-06, + "loss": 0.0831, + "step": 14391 + }, + { + "epoch": 6.03, + "learning_rate": 3.0248789847406003e-06, + "loss": 0.0728, + "step": 14392 + }, + { + "epoch": 6.03, + "learning_rate": 3.02366420076168e-06, + "loss": 0.0453, + "step": 14393 + }, + { + "epoch": 6.03, + "learning_rate": 3.022449617311346e-06, + "loss": 0.089, + "step": 14394 + }, + { + "epoch": 6.03, + "learning_rate": 3.021235234424502e-06, + "loss": 0.0904, + "step": 14395 + }, + { + "epoch": 6.03, + "learning_rate": 3.0200210521360598e-06, + "loss": 0.0671, + "step": 14396 + }, + { + "epoch": 6.03, + "learning_rate": 3.0188070704809167e-06, + "loss": 0.1163, + "step": 14397 + }, + { + "epoch": 6.03, + "learning_rate": 3.017593289493971e-06, + "loss": 0.0829, + "step": 14398 + }, + { + "epoch": 6.03, + "learning_rate": 3.0163797092101087e-06, + "loss": 0.0714, + "step": 14399 + }, + { + "epoch": 6.03, + "learning_rate": 3.0151663296642188e-06, + "loss": 0.09, + "step": 14400 + }, + { + "epoch": 6.03, + "learning_rate": 3.01395315089117e-06, + "loss": 0.0753, + "step": 14401 + }, + { + "epoch": 6.03, + "learning_rate": 3.01274017292584e-06, + "loss": 0.0752, + "step": 14402 + }, + { + "epoch": 6.03, + "learning_rate": 3.011527395803092e-06, + "loss": 0.0562, + "step": 14403 + }, + { + "epoch": 6.03, + "learning_rate": 3.0103148195577914e-06, + "loss": 0.0999, + "step": 14404 + }, + { + "epoch": 6.03, + "learning_rate": 3.009102444224786e-06, + "loss": 0.0673, + "step": 14405 + }, + { + "epoch": 6.03, + "learning_rate": 3.0078902698389266e-06, + "loss": 0.0899, + "step": 14406 + }, + { + "epoch": 6.03, + "learning_rate": 3.0066782964350573e-06, + "loss": 0.1033, + "step": 14407 + }, + { + "epoch": 6.03, + "learning_rate": 3.0054665240480174e-06, + "loss": 0.0715, + "step": 14408 + }, + { + "epoch": 6.03, + "learning_rate": 3.004254952712632e-06, + "loss": 0.0692, + "step": 14409 + }, + { + "epoch": 6.03, + "learning_rate": 3.0030435824637304e-06, + "loss": 0.0857, + "step": 14410 + }, + { + "epoch": 6.03, + "learning_rate": 3.0018324133361323e-06, + "loss": 0.0605, + "step": 14411 + }, + { + "epoch": 6.03, + "learning_rate": 3.000621445364651e-06, + "loss": 0.0519, + "step": 14412 + }, + { + "epoch": 6.03, + "learning_rate": 2.999410678584096e-06, + "loss": 0.0537, + "step": 14413 + }, + { + "epoch": 6.03, + "learning_rate": 2.9982001130292703e-06, + "loss": 0.0751, + "step": 14414 + }, + { + "epoch": 6.04, + "learning_rate": 2.9969897487349675e-06, + "loss": 0.083, + "step": 14415 + }, + { + "epoch": 6.04, + "learning_rate": 2.9957795857359794e-06, + "loss": 0.0699, + "step": 14416 + }, + { + "epoch": 6.04, + "learning_rate": 2.9945696240670905e-06, + "loss": 0.1112, + "step": 14417 + }, + { + "epoch": 6.04, + "learning_rate": 2.9933598637630833e-06, + "loss": 0.0777, + "step": 14418 + }, + { + "epoch": 6.04, + "learning_rate": 2.9921503048587284e-06, + "loss": 0.0872, + "step": 14419 + }, + { + "epoch": 6.04, + "learning_rate": 2.990940947388794e-06, + "loss": 0.1089, + "step": 14420 + }, + { + "epoch": 6.04, + "learning_rate": 2.9897317913880454e-06, + "loss": 0.0951, + "step": 14421 + }, + { + "epoch": 6.04, + "learning_rate": 2.9885228368912333e-06, + "loss": 0.052, + "step": 14422 + }, + { + "epoch": 6.04, + "learning_rate": 2.9873140839331094e-06, + "loss": 0.0862, + "step": 14423 + }, + { + "epoch": 6.04, + "learning_rate": 2.98610553254842e-06, + "loss": 0.0954, + "step": 14424 + }, + { + "epoch": 6.04, + "learning_rate": 2.984897182771903e-06, + "loss": 0.061, + "step": 14425 + }, + { + "epoch": 6.04, + "learning_rate": 2.9836890346382907e-06, + "loss": 0.0772, + "step": 14426 + }, + { + "epoch": 6.04, + "learning_rate": 2.982481088182315e-06, + "loss": 0.0885, + "step": 14427 + }, + { + "epoch": 6.04, + "learning_rate": 2.98127334343869e-06, + "loss": 0.1007, + "step": 14428 + }, + { + "epoch": 6.04, + "learning_rate": 2.980065800442137e-06, + "loss": 0.046, + "step": 14429 + }, + { + "epoch": 6.04, + "learning_rate": 2.9788584592273595e-06, + "loss": 0.068, + "step": 14430 + }, + { + "epoch": 6.04, + "learning_rate": 2.9776513198290657e-06, + "loss": 0.0835, + "step": 14431 + }, + { + "epoch": 6.04, + "learning_rate": 2.9764443822819534e-06, + "loss": 0.0601, + "step": 14432 + }, + { + "epoch": 6.04, + "learning_rate": 2.9752376466207154e-06, + "loss": 0.0832, + "step": 14433 + }, + { + "epoch": 6.04, + "learning_rate": 2.974031112880037e-06, + "loss": 0.1016, + "step": 14434 + }, + { + "epoch": 6.04, + "learning_rate": 2.972824781094602e-06, + "loss": 0.0758, + "step": 14435 + }, + { + "epoch": 6.04, + "learning_rate": 2.9716186512990806e-06, + "loss": 0.062, + "step": 14436 + }, + { + "epoch": 6.04, + "learning_rate": 2.9704127235281433e-06, + "loss": 0.0975, + "step": 14437 + }, + { + "epoch": 6.04, + "learning_rate": 2.9692069978164552e-06, + "loss": 0.0696, + "step": 14438 + }, + { + "epoch": 6.05, + "learning_rate": 2.968001474198673e-06, + "loss": 0.0852, + "step": 14439 + }, + { + "epoch": 6.05, + "learning_rate": 2.966796152709447e-06, + "loss": 0.1034, + "step": 14440 + }, + { + "epoch": 6.05, + "learning_rate": 2.9655910333834294e-06, + "loss": 0.0885, + "step": 14441 + }, + { + "epoch": 6.05, + "learning_rate": 2.9643861162552512e-06, + "loss": 0.0735, + "step": 14442 + }, + { + "epoch": 6.05, + "learning_rate": 2.9631814013595507e-06, + "loss": 0.114, + "step": 14443 + }, + { + "epoch": 6.05, + "learning_rate": 2.961976888730956e-06, + "loss": 0.0802, + "step": 14444 + }, + { + "epoch": 6.05, + "learning_rate": 2.960772578404091e-06, + "loss": 0.0662, + "step": 14445 + }, + { + "epoch": 6.05, + "learning_rate": 2.959568470413572e-06, + "loss": 0.0662, + "step": 14446 + }, + { + "epoch": 6.05, + "learning_rate": 2.9583645647940097e-06, + "loss": 0.0978, + "step": 14447 + }, + { + "epoch": 6.05, + "learning_rate": 2.9571608615800118e-06, + "loss": 0.0529, + "step": 14448 + }, + { + "epoch": 6.05, + "learning_rate": 2.9559573608061755e-06, + "loss": 0.0754, + "step": 14449 + }, + { + "epoch": 6.05, + "learning_rate": 2.9547540625070903e-06, + "loss": 0.0878, + "step": 14450 + }, + { + "epoch": 6.05, + "learning_rate": 2.953550966717349e-06, + "loss": 0.0838, + "step": 14451 + }, + { + "epoch": 6.05, + "learning_rate": 2.9523480734715317e-06, + "loss": 0.0953, + "step": 14452 + }, + { + "epoch": 6.05, + "learning_rate": 2.9511453828042147e-06, + "loss": 0.1005, + "step": 14453 + }, + { + "epoch": 6.05, + "learning_rate": 2.9499428947499696e-06, + "loss": 0.0766, + "step": 14454 + }, + { + "epoch": 6.05, + "learning_rate": 2.9487406093433622e-06, + "loss": 0.0949, + "step": 14455 + }, + { + "epoch": 6.05, + "learning_rate": 2.947538526618946e-06, + "loss": 0.0926, + "step": 14456 + }, + { + "epoch": 6.05, + "learning_rate": 2.9463366466112764e-06, + "loss": 0.1011, + "step": 14457 + }, + { + "epoch": 6.05, + "learning_rate": 2.9451349693549e-06, + "loss": 0.0735, + "step": 14458 + }, + { + "epoch": 6.05, + "learning_rate": 2.9439334948843588e-06, + "loss": 0.1279, + "step": 14459 + }, + { + "epoch": 6.05, + "learning_rate": 2.9427322232341893e-06, + "loss": 0.067, + "step": 14460 + }, + { + "epoch": 6.05, + "learning_rate": 2.941531154438918e-06, + "loss": 0.064, + "step": 14461 + }, + { + "epoch": 6.05, + "learning_rate": 2.940330288533074e-06, + "loss": 0.0995, + "step": 14462 + }, + { + "epoch": 6.06, + "learning_rate": 2.9391296255511683e-06, + "loss": 0.1056, + "step": 14463 + }, + { + "epoch": 6.06, + "learning_rate": 2.9379291655277143e-06, + "loss": 0.0957, + "step": 14464 + }, + { + "epoch": 6.06, + "learning_rate": 2.9367289084972206e-06, + "loss": 0.0597, + "step": 14465 + }, + { + "epoch": 6.06, + "learning_rate": 2.9355288544941873e-06, + "loss": 0.0803, + "step": 14466 + }, + { + "epoch": 6.06, + "learning_rate": 2.934329003553107e-06, + "loss": 0.0615, + "step": 14467 + }, + { + "epoch": 6.06, + "learning_rate": 2.933129355708473e-06, + "loss": 0.0744, + "step": 14468 + }, + { + "epoch": 6.06, + "learning_rate": 2.931929910994762e-06, + "loss": 0.0773, + "step": 14469 + }, + { + "epoch": 6.06, + "learning_rate": 2.930730669446453e-06, + "loss": 0.0954, + "step": 14470 + }, + { + "epoch": 6.06, + "learning_rate": 2.9295316310980184e-06, + "loss": 0.0626, + "step": 14471 + }, + { + "epoch": 6.06, + "learning_rate": 2.928332795983926e-06, + "loss": 0.0806, + "step": 14472 + }, + { + "epoch": 6.06, + "learning_rate": 2.9271341641386285e-06, + "loss": 0.0626, + "step": 14473 + }, + { + "epoch": 6.06, + "learning_rate": 2.925935735596582e-06, + "loss": 0.0804, + "step": 14474 + }, + { + "epoch": 6.06, + "learning_rate": 2.9247375103922373e-06, + "loss": 0.0862, + "step": 14475 + }, + { + "epoch": 6.06, + "learning_rate": 2.9235394885600367e-06, + "loss": 0.0685, + "step": 14476 + }, + { + "epoch": 6.06, + "learning_rate": 2.922341670134411e-06, + "loss": 0.1191, + "step": 14477 + }, + { + "epoch": 6.06, + "learning_rate": 2.921144055149793e-06, + "loss": 0.0693, + "step": 14478 + }, + { + "epoch": 6.06, + "learning_rate": 2.9199466436406077e-06, + "loss": 0.0933, + "step": 14479 + }, + { + "epoch": 6.06, + "learning_rate": 2.918749435641274e-06, + "loss": 0.1152, + "step": 14480 + }, + { + "epoch": 6.06, + "learning_rate": 2.9175524311862034e-06, + "loss": 0.103, + "step": 14481 + }, + { + "epoch": 6.06, + "learning_rate": 2.916355630309806e-06, + "loss": 0.0681, + "step": 14482 + }, + { + "epoch": 6.06, + "learning_rate": 2.9151590330464775e-06, + "loss": 0.084, + "step": 14483 + }, + { + "epoch": 6.06, + "learning_rate": 2.913962639430615e-06, + "loss": 0.0755, + "step": 14484 + }, + { + "epoch": 6.06, + "learning_rate": 2.91276644949661e-06, + "loss": 0.0906, + "step": 14485 + }, + { + "epoch": 6.06, + "learning_rate": 2.911570463278843e-06, + "loss": 0.0835, + "step": 14486 + }, + { + "epoch": 6.07, + "learning_rate": 2.910374680811693e-06, + "loss": 0.0773, + "step": 14487 + }, + { + "epoch": 6.07, + "learning_rate": 2.909179102129531e-06, + "loss": 0.0837, + "step": 14488 + }, + { + "epoch": 6.07, + "learning_rate": 2.9079837272667276e-06, + "loss": 0.0662, + "step": 14489 + }, + { + "epoch": 6.07, + "learning_rate": 2.906788556257636e-06, + "loss": 0.077, + "step": 14490 + }, + { + "epoch": 6.07, + "learning_rate": 2.9055935891366117e-06, + "loss": 0.089, + "step": 14491 + }, + { + "epoch": 6.07, + "learning_rate": 2.904398825938005e-06, + "loss": 0.0903, + "step": 14492 + }, + { + "epoch": 6.07, + "learning_rate": 2.903204266696157e-06, + "loss": 0.0882, + "step": 14493 + }, + { + "epoch": 6.07, + "learning_rate": 2.9020099114454046e-06, + "loss": 0.0891, + "step": 14494 + }, + { + "epoch": 6.07, + "learning_rate": 2.9008157602200827e-06, + "loss": 0.0848, + "step": 14495 + }, + { + "epoch": 6.07, + "learning_rate": 2.8996218130545108e-06, + "loss": 0.1225, + "step": 14496 + }, + { + "epoch": 6.07, + "learning_rate": 2.898428069983007e-06, + "loss": 0.0876, + "step": 14497 + }, + { + "epoch": 6.07, + "learning_rate": 2.8972345310398863e-06, + "loss": 0.0819, + "step": 14498 + }, + { + "epoch": 6.07, + "learning_rate": 2.8960411962594558e-06, + "loss": 0.0752, + "step": 14499 + }, + { + "epoch": 6.07, + "learning_rate": 2.894848065676017e-06, + "loss": 0.0973, + "step": 14500 + }, + { + "epoch": 6.07, + "learning_rate": 2.8936551393238664e-06, + "loss": 0.0801, + "step": 14501 + }, + { + "epoch": 6.07, + "learning_rate": 2.892462417237292e-06, + "loss": 0.0798, + "step": 14502 + }, + { + "epoch": 6.07, + "learning_rate": 2.8912698994505816e-06, + "loss": 0.0809, + "step": 14503 + }, + { + "epoch": 6.07, + "learning_rate": 2.890077585998007e-06, + "loss": 0.0906, + "step": 14504 + }, + { + "epoch": 6.07, + "learning_rate": 2.888885476913842e-06, + "loss": 0.0601, + "step": 14505 + }, + { + "epoch": 6.07, + "learning_rate": 2.887693572232355e-06, + "loss": 0.0999, + "step": 14506 + }, + { + "epoch": 6.07, + "learning_rate": 2.8865018719878034e-06, + "loss": 0.0891, + "step": 14507 + }, + { + "epoch": 6.07, + "learning_rate": 2.8853103762144443e-06, + "loss": 0.0517, + "step": 14508 + }, + { + "epoch": 6.07, + "learning_rate": 2.8841190849465282e-06, + "loss": 0.0721, + "step": 14509 + }, + { + "epoch": 6.07, + "learning_rate": 2.882927998218291e-06, + "loss": 0.0974, + "step": 14510 + }, + { + "epoch": 6.08, + "learning_rate": 2.881737116063973e-06, + "loss": 0.0599, + "step": 14511 + }, + { + "epoch": 6.08, + "learning_rate": 2.880546438517804e-06, + "loss": 0.0839, + "step": 14512 + }, + { + "epoch": 6.08, + "learning_rate": 2.8793559656140114e-06, + "loss": 0.1268, + "step": 14513 + }, + { + "epoch": 6.08, + "learning_rate": 2.878165697386812e-06, + "loss": 0.0802, + "step": 14514 + }, + { + "epoch": 6.08, + "learning_rate": 2.8769756338704235e-06, + "loss": 0.0703, + "step": 14515 + }, + { + "epoch": 6.08, + "learning_rate": 2.875785775099046e-06, + "loss": 0.0828, + "step": 14516 + }, + { + "epoch": 6.08, + "learning_rate": 2.8745961211068873e-06, + "loss": 0.1026, + "step": 14517 + }, + { + "epoch": 6.08, + "learning_rate": 2.8734066719281372e-06, + "loss": 0.0841, + "step": 14518 + }, + { + "epoch": 6.08, + "learning_rate": 2.8722174275969884e-06, + "loss": 0.1171, + "step": 14519 + }, + { + "epoch": 6.08, + "learning_rate": 2.871028388147624e-06, + "loss": 0.1032, + "step": 14520 + }, + { + "epoch": 6.08, + "learning_rate": 2.8698395536142232e-06, + "loss": 0.0571, + "step": 14521 + }, + { + "epoch": 6.08, + "learning_rate": 2.868650924030958e-06, + "loss": 0.084, + "step": 14522 + }, + { + "epoch": 6.08, + "learning_rate": 2.8674624994319954e-06, + "loss": 0.0753, + "step": 14523 + }, + { + "epoch": 6.08, + "learning_rate": 2.866274279851491e-06, + "loss": 0.0792, + "step": 14524 + }, + { + "epoch": 6.08, + "learning_rate": 2.8650862653236034e-06, + "loss": 0.097, + "step": 14525 + }, + { + "epoch": 6.08, + "learning_rate": 2.8638984558824777e-06, + "loss": 0.0608, + "step": 14526 + }, + { + "epoch": 6.08, + "learning_rate": 2.8627108515622603e-06, + "loss": 0.1011, + "step": 14527 + }, + { + "epoch": 6.08, + "learning_rate": 2.861523452397085e-06, + "loss": 0.0821, + "step": 14528 + }, + { + "epoch": 6.08, + "learning_rate": 2.8603362584210838e-06, + "loss": 0.0893, + "step": 14529 + }, + { + "epoch": 6.08, + "learning_rate": 2.8591492696683833e-06, + "loss": 0.0995, + "step": 14530 + }, + { + "epoch": 6.08, + "learning_rate": 2.857962486173098e-06, + "loss": 0.0804, + "step": 14531 + }, + { + "epoch": 6.08, + "learning_rate": 2.8567759079693435e-06, + "loss": 0.101, + "step": 14532 + }, + { + "epoch": 6.08, + "learning_rate": 2.855589535091227e-06, + "loss": 0.0949, + "step": 14533 + }, + { + "epoch": 6.08, + "learning_rate": 2.854403367572849e-06, + "loss": 0.0694, + "step": 14534 + }, + { + "epoch": 6.09, + "learning_rate": 2.8532174054483053e-06, + "loss": 0.0773, + "step": 14535 + }, + { + "epoch": 6.09, + "learning_rate": 2.8520316487516884e-06, + "loss": 0.069, + "step": 14536 + }, + { + "epoch": 6.09, + "learning_rate": 2.850846097517076e-06, + "loss": 0.079, + "step": 14537 + }, + { + "epoch": 6.09, + "learning_rate": 2.849660751778548e-06, + "loss": 0.0623, + "step": 14538 + }, + { + "epoch": 6.09, + "learning_rate": 2.8484756115701773e-06, + "loss": 0.0909, + "step": 14539 + }, + { + "epoch": 6.09, + "learning_rate": 2.847290676926031e-06, + "loss": 0.0821, + "step": 14540 + }, + { + "epoch": 6.09, + "learning_rate": 2.8461059478801657e-06, + "loss": 0.1235, + "step": 14541 + }, + { + "epoch": 6.09, + "learning_rate": 2.8449214244666345e-06, + "loss": 0.0907, + "step": 14542 + }, + { + "epoch": 6.09, + "learning_rate": 2.84373710671949e-06, + "loss": 0.0718, + "step": 14543 + }, + { + "epoch": 6.09, + "learning_rate": 2.8425529946727725e-06, + "loss": 0.0996, + "step": 14544 + }, + { + "epoch": 6.09, + "learning_rate": 2.841369088360516e-06, + "loss": 0.0839, + "step": 14545 + }, + { + "epoch": 6.09, + "learning_rate": 2.8401853878167528e-06, + "loss": 0.1058, + "step": 14546 + }, + { + "epoch": 6.09, + "learning_rate": 2.839001893075507e-06, + "loss": 0.0762, + "step": 14547 + }, + { + "epoch": 6.09, + "learning_rate": 2.8378186041707977e-06, + "loss": 0.0689, + "step": 14548 + }, + { + "epoch": 6.09, + "learning_rate": 2.836635521136636e-06, + "loss": 0.0865, + "step": 14549 + }, + { + "epoch": 6.09, + "learning_rate": 2.835452644007033e-06, + "loss": 0.0619, + "step": 14550 + }, + { + "epoch": 6.09, + "learning_rate": 2.834269972815983e-06, + "loss": 0.0879, + "step": 14551 + }, + { + "epoch": 6.09, + "learning_rate": 2.833087507597484e-06, + "loss": 0.0751, + "step": 14552 + }, + { + "epoch": 6.09, + "learning_rate": 2.831905248385525e-06, + "loss": 0.0685, + "step": 14553 + }, + { + "epoch": 6.09, + "learning_rate": 2.830723195214088e-06, + "loss": 0.0894, + "step": 14554 + }, + { + "epoch": 6.09, + "learning_rate": 2.8295413481171517e-06, + "loss": 0.0732, + "step": 14555 + }, + { + "epoch": 6.09, + "learning_rate": 2.8283597071286897e-06, + "loss": 0.0846, + "step": 14556 + }, + { + "epoch": 6.09, + "learning_rate": 2.82717827228266e-06, + "loss": 0.0812, + "step": 14557 + }, + { + "epoch": 6.1, + "learning_rate": 2.825997043613026e-06, + "loss": 0.1015, + "step": 14558 + }, + { + "epoch": 6.1, + "learning_rate": 2.824816021153741e-06, + "loss": 0.0906, + "step": 14559 + }, + { + "epoch": 6.1, + "learning_rate": 2.8236352049387515e-06, + "loss": 0.0706, + "step": 14560 + }, + { + "epoch": 6.1, + "learning_rate": 2.822454595002001e-06, + "loss": 0.11, + "step": 14561 + }, + { + "epoch": 6.1, + "learning_rate": 2.821274191377423e-06, + "loss": 0.0869, + "step": 14562 + }, + { + "epoch": 6.1, + "learning_rate": 2.8200939940989524e-06, + "loss": 0.0942, + "step": 14563 + }, + { + "epoch": 6.1, + "learning_rate": 2.818914003200508e-06, + "loss": 0.0914, + "step": 14564 + }, + { + "epoch": 6.1, + "learning_rate": 2.817734218716005e-06, + "loss": 0.0794, + "step": 14565 + }, + { + "epoch": 6.1, + "learning_rate": 2.8165546406793586e-06, + "loss": 0.0576, + "step": 14566 + }, + { + "epoch": 6.1, + "learning_rate": 2.815375269124474e-06, + "loss": 0.0971, + "step": 14567 + }, + { + "epoch": 6.1, + "learning_rate": 2.814196104085253e-06, + "loss": 0.0714, + "step": 14568 + }, + { + "epoch": 6.1, + "learning_rate": 2.813017145595588e-06, + "loss": 0.0884, + "step": 14569 + }, + { + "epoch": 6.1, + "learning_rate": 2.811838393689367e-06, + "loss": 0.0818, + "step": 14570 + }, + { + "epoch": 6.1, + "learning_rate": 2.8106598484004756e-06, + "loss": 0.1187, + "step": 14571 + }, + { + "epoch": 6.1, + "learning_rate": 2.8094815097627846e-06, + "loss": 0.0948, + "step": 14572 + }, + { + "epoch": 6.1, + "learning_rate": 2.8083033778101664e-06, + "loss": 0.0907, + "step": 14573 + }, + { + "epoch": 6.1, + "learning_rate": 2.807125452576487e-06, + "loss": 0.1035, + "step": 14574 + }, + { + "epoch": 6.1, + "learning_rate": 2.805947734095602e-06, + "loss": 0.0851, + "step": 14575 + }, + { + "epoch": 6.1, + "learning_rate": 2.804770222401366e-06, + "loss": 0.1148, + "step": 14576 + }, + { + "epoch": 6.1, + "learning_rate": 2.803592917527629e-06, + "loss": 0.0831, + "step": 14577 + }, + { + "epoch": 6.1, + "learning_rate": 2.802415819508223e-06, + "loss": 0.0842, + "step": 14578 + }, + { + "epoch": 6.1, + "learning_rate": 2.8012389283769883e-06, + "loss": 0.0846, + "step": 14579 + }, + { + "epoch": 6.1, + "learning_rate": 2.8000622441677527e-06, + "loss": 0.0884, + "step": 14580 + }, + { + "epoch": 6.1, + "learning_rate": 2.798885766914339e-06, + "loss": 0.0936, + "step": 14581 + }, + { + "epoch": 6.11, + "learning_rate": 2.7977094966505624e-06, + "loss": 0.081, + "step": 14582 + }, + { + "epoch": 6.11, + "learning_rate": 2.796533433410239e-06, + "loss": 0.0873, + "step": 14583 + }, + { + "epoch": 6.11, + "learning_rate": 2.7953575772271667e-06, + "loss": 0.0961, + "step": 14584 + }, + { + "epoch": 6.11, + "learning_rate": 2.7941819281351513e-06, + "loss": 0.0902, + "step": 14585 + }, + { + "epoch": 6.11, + "learning_rate": 2.7930064861679783e-06, + "loss": 0.0795, + "step": 14586 + }, + { + "epoch": 6.11, + "learning_rate": 2.791831251359438e-06, + "loss": 0.0775, + "step": 14587 + }, + { + "epoch": 6.11, + "learning_rate": 2.790656223743312e-06, + "loss": 0.0974, + "step": 14588 + }, + { + "epoch": 6.11, + "learning_rate": 2.7894814033533768e-06, + "loss": 0.0759, + "step": 14589 + }, + { + "epoch": 6.11, + "learning_rate": 2.7883067902233985e-06, + "loss": 0.1007, + "step": 14590 + }, + { + "epoch": 6.11, + "learning_rate": 2.7871323843871467e-06, + "loss": 0.0767, + "step": 14591 + }, + { + "epoch": 6.11, + "learning_rate": 2.7859581858783693e-06, + "loss": 0.0675, + "step": 14592 + }, + { + "epoch": 6.11, + "learning_rate": 2.784784194730823e-06, + "loss": 0.1, + "step": 14593 + }, + { + "epoch": 6.11, + "learning_rate": 2.7836104109782525e-06, + "loss": 0.1084, + "step": 14594 + }, + { + "epoch": 6.11, + "learning_rate": 2.7824368346543963e-06, + "loss": 0.0986, + "step": 14595 + }, + { + "epoch": 6.11, + "learning_rate": 2.781263465792989e-06, + "loss": 0.078, + "step": 14596 + }, + { + "epoch": 6.11, + "learning_rate": 2.7800903044277604e-06, + "loss": 0.08, + "step": 14597 + }, + { + "epoch": 6.11, + "learning_rate": 2.7789173505924274e-06, + "loss": 0.0671, + "step": 14598 + }, + { + "epoch": 6.11, + "learning_rate": 2.777744604320706e-06, + "loss": 0.0647, + "step": 14599 + }, + { + "epoch": 6.11, + "learning_rate": 2.776572065646307e-06, + "loss": 0.0704, + "step": 14600 + }, + { + "epoch": 6.11, + "learning_rate": 2.775399734602934e-06, + "loss": 0.1009, + "step": 14601 + }, + { + "epoch": 6.11, + "learning_rate": 2.774227611224285e-06, + "loss": 0.1078, + "step": 14602 + }, + { + "epoch": 6.11, + "learning_rate": 2.7730556955440515e-06, + "loss": 0.0757, + "step": 14603 + }, + { + "epoch": 6.11, + "learning_rate": 2.7718839875959213e-06, + "loss": 0.0806, + "step": 14604 + }, + { + "epoch": 6.11, + "learning_rate": 2.7707124874135694e-06, + "loss": 0.0631, + "step": 14605 + }, + { + "epoch": 6.12, + "learning_rate": 2.7695411950306714e-06, + "loss": 0.1014, + "step": 14606 + }, + { + "epoch": 6.12, + "learning_rate": 2.7683701104808956e-06, + "loss": 0.0764, + "step": 14607 + }, + { + "epoch": 6.12, + "learning_rate": 2.767199233797907e-06, + "loss": 0.1016, + "step": 14608 + }, + { + "epoch": 6.12, + "learning_rate": 2.766028565015355e-06, + "loss": 0.0935, + "step": 14609 + }, + { + "epoch": 6.12, + "learning_rate": 2.7648581041668922e-06, + "loss": 0.0877, + "step": 14610 + }, + { + "epoch": 6.12, + "learning_rate": 2.7636878512861633e-06, + "loss": 0.0711, + "step": 14611 + }, + { + "epoch": 6.12, + "learning_rate": 2.7625178064068094e-06, + "loss": 0.0836, + "step": 14612 + }, + { + "epoch": 6.12, + "learning_rate": 2.7613479695624544e-06, + "loss": 0.0938, + "step": 14613 + }, + { + "epoch": 6.12, + "learning_rate": 2.760178340786729e-06, + "loss": 0.0893, + "step": 14614 + }, + { + "epoch": 6.12, + "learning_rate": 2.759008920113253e-06, + "loss": 0.096, + "step": 14615 + }, + { + "epoch": 6.12, + "learning_rate": 2.7578397075756404e-06, + "loss": 0.0646, + "step": 14616 + }, + { + "epoch": 6.12, + "learning_rate": 2.756670703207498e-06, + "loss": 0.0842, + "step": 14617 + }, + { + "epoch": 6.12, + "learning_rate": 2.7555019070424316e-06, + "loss": 0.0547, + "step": 14618 + }, + { + "epoch": 6.12, + "learning_rate": 2.754333319114032e-06, + "loss": 0.0824, + "step": 14619 + }, + { + "epoch": 6.12, + "learning_rate": 2.753164939455891e-06, + "loss": 0.0972, + "step": 14620 + }, + { + "epoch": 6.12, + "learning_rate": 2.7519967681015933e-06, + "loss": 0.0783, + "step": 14621 + }, + { + "epoch": 6.12, + "learning_rate": 2.7508288050847166e-06, + "loss": 0.1054, + "step": 14622 + }, + { + "epoch": 6.12, + "learning_rate": 2.7496610504388334e-06, + "loss": 0.0799, + "step": 14623 + }, + { + "epoch": 6.12, + "learning_rate": 2.7484935041975113e-06, + "loss": 0.0862, + "step": 14624 + }, + { + "epoch": 6.12, + "learning_rate": 2.7473261663943063e-06, + "loss": 0.1045, + "step": 14625 + }, + { + "epoch": 6.12, + "learning_rate": 2.7461590370627753e-06, + "loss": 0.0807, + "step": 14626 + }, + { + "epoch": 6.12, + "learning_rate": 2.7449921162364647e-06, + "loss": 0.0825, + "step": 14627 + }, + { + "epoch": 6.12, + "learning_rate": 2.7438254039489185e-06, + "loss": 0.0802, + "step": 14628 + }, + { + "epoch": 6.12, + "learning_rate": 2.7426589002336723e-06, + "loss": 0.0778, + "step": 14629 + }, + { + "epoch": 6.13, + "learning_rate": 2.741492605124255e-06, + "loss": 0.0931, + "step": 14630 + }, + { + "epoch": 6.13, + "learning_rate": 2.7403265186541963e-06, + "loss": 0.0809, + "step": 14631 + }, + { + "epoch": 6.13, + "learning_rate": 2.7391606408570084e-06, + "loss": 0.0576, + "step": 14632 + }, + { + "epoch": 6.13, + "learning_rate": 2.7379949717662014e-06, + "loss": 0.0951, + "step": 14633 + }, + { + "epoch": 6.13, + "learning_rate": 2.7368295114152852e-06, + "loss": 0.1012, + "step": 14634 + }, + { + "epoch": 6.13, + "learning_rate": 2.7356642598377604e-06, + "loss": 0.0901, + "step": 14635 + }, + { + "epoch": 6.13, + "learning_rate": 2.734499217067119e-06, + "loss": 0.0891, + "step": 14636 + }, + { + "epoch": 6.13, + "learning_rate": 2.733334383136851e-06, + "loss": 0.0668, + "step": 14637 + }, + { + "epoch": 6.13, + "learning_rate": 2.7321697580804407e-06, + "loss": 0.0831, + "step": 14638 + }, + { + "epoch": 6.13, + "learning_rate": 2.7310053419313576e-06, + "loss": 0.0706, + "step": 14639 + }, + { + "epoch": 6.13, + "learning_rate": 2.729841134723077e-06, + "loss": 0.0909, + "step": 14640 + }, + { + "epoch": 6.13, + "learning_rate": 2.7286771364890606e-06, + "loss": 0.0715, + "step": 14641 + }, + { + "epoch": 6.13, + "learning_rate": 2.727513347262767e-06, + "loss": 0.0648, + "step": 14642 + }, + { + "epoch": 6.13, + "learning_rate": 2.7263497670776495e-06, + "loss": 0.103, + "step": 14643 + }, + { + "epoch": 6.13, + "learning_rate": 2.7251863959671533e-06, + "loss": 0.0876, + "step": 14644 + }, + { + "epoch": 6.13, + "learning_rate": 2.724023233964721e-06, + "loss": 0.0984, + "step": 14645 + }, + { + "epoch": 6.13, + "learning_rate": 2.722860281103782e-06, + "loss": 0.0727, + "step": 14646 + }, + { + "epoch": 6.13, + "learning_rate": 2.721697537417767e-06, + "loss": 0.1052, + "step": 14647 + }, + { + "epoch": 6.13, + "learning_rate": 2.7205350029400966e-06, + "loss": 0.0902, + "step": 14648 + }, + { + "epoch": 6.13, + "learning_rate": 2.7193726777041885e-06, + "loss": 0.0884, + "step": 14649 + }, + { + "epoch": 6.13, + "learning_rate": 2.7182105617434516e-06, + "loss": 0.1211, + "step": 14650 + }, + { + "epoch": 6.13, + "learning_rate": 2.717048655091293e-06, + "loss": 0.0977, + "step": 14651 + }, + { + "epoch": 6.13, + "learning_rate": 2.7158869577811064e-06, + "loss": 0.0991, + "step": 14652 + }, + { + "epoch": 6.13, + "learning_rate": 2.7147254698462875e-06, + "loss": 0.0861, + "step": 14653 + }, + { + "epoch": 6.14, + "learning_rate": 2.713564191320217e-06, + "loss": 0.0808, + "step": 14654 + }, + { + "epoch": 6.14, + "learning_rate": 2.712403122236279e-06, + "loss": 0.0791, + "step": 14655 + }, + { + "epoch": 6.14, + "learning_rate": 2.7112422626278456e-06, + "loss": 0.0886, + "step": 14656 + }, + { + "epoch": 6.14, + "learning_rate": 2.7100816125282846e-06, + "loss": 0.0898, + "step": 14657 + }, + { + "epoch": 6.14, + "learning_rate": 2.7089211719709596e-06, + "loss": 0.0485, + "step": 14658 + }, + { + "epoch": 6.14, + "learning_rate": 2.707760940989229e-06, + "loss": 0.0637, + "step": 14659 + }, + { + "epoch": 6.14, + "learning_rate": 2.7066009196164355e-06, + "loss": 0.0994, + "step": 14660 + }, + { + "epoch": 6.14, + "learning_rate": 2.7054411078859267e-06, + "loss": 0.1317, + "step": 14661 + }, + { + "epoch": 6.14, + "learning_rate": 2.704281505831039e-06, + "loss": 0.0502, + "step": 14662 + }, + { + "epoch": 6.14, + "learning_rate": 2.7031221134851073e-06, + "loss": 0.0918, + "step": 14663 + }, + { + "epoch": 6.14, + "learning_rate": 2.7019629308814544e-06, + "loss": 0.081, + "step": 14664 + }, + { + "epoch": 6.14, + "learning_rate": 2.700803958053403e-06, + "loss": 0.1076, + "step": 14665 + }, + { + "epoch": 6.14, + "learning_rate": 2.699645195034263e-06, + "loss": 0.0651, + "step": 14666 + }, + { + "epoch": 6.14, + "learning_rate": 2.698486641857343e-06, + "loss": 0.0671, + "step": 14667 + }, + { + "epoch": 6.14, + "learning_rate": 2.6973282985559458e-06, + "loss": 0.1349, + "step": 14668 + }, + { + "epoch": 6.14, + "learning_rate": 2.6961701651633653e-06, + "loss": 0.0904, + "step": 14669 + }, + { + "epoch": 6.14, + "learning_rate": 2.6950122417128933e-06, + "loss": 0.0759, + "step": 14670 + }, + { + "epoch": 6.14, + "learning_rate": 2.6938545282378117e-06, + "loss": 0.0651, + "step": 14671 + }, + { + "epoch": 6.14, + "learning_rate": 2.6926970247714e-06, + "loss": 0.0691, + "step": 14672 + }, + { + "epoch": 6.14, + "learning_rate": 2.691539731346927e-06, + "loss": 0.1042, + "step": 14673 + }, + { + "epoch": 6.14, + "learning_rate": 2.6903826479976582e-06, + "loss": 0.0805, + "step": 14674 + }, + { + "epoch": 6.14, + "learning_rate": 2.689225774756853e-06, + "loss": 0.0479, + "step": 14675 + }, + { + "epoch": 6.14, + "learning_rate": 2.6880691116577694e-06, + "loss": 0.1131, + "step": 14676 + }, + { + "epoch": 6.14, + "learning_rate": 2.686912658733647e-06, + "loss": 0.1021, + "step": 14677 + }, + { + "epoch": 6.15, + "learning_rate": 2.685756416017732e-06, + "loss": 0.1051, + "step": 14678 + }, + { + "epoch": 6.15, + "learning_rate": 2.6846003835432597e-06, + "loss": 0.0603, + "step": 14679 + }, + { + "epoch": 6.15, + "learning_rate": 2.6834445613434555e-06, + "loss": 0.0844, + "step": 14680 + }, + { + "epoch": 6.15, + "learning_rate": 2.6822889494515445e-06, + "loss": 0.109, + "step": 14681 + }, + { + "epoch": 6.15, + "learning_rate": 2.681133547900745e-06, + "loss": 0.1006, + "step": 14682 + }, + { + "epoch": 6.15, + "learning_rate": 2.6799783567242653e-06, + "loss": 0.0806, + "step": 14683 + }, + { + "epoch": 6.15, + "learning_rate": 2.678823375955314e-06, + "loss": 0.0697, + "step": 14684 + }, + { + "epoch": 6.15, + "learning_rate": 2.677668605627086e-06, + "loss": 0.1013, + "step": 14685 + }, + { + "epoch": 6.15, + "learning_rate": 2.6765140457727813e-06, + "loss": 0.0662, + "step": 14686 + }, + { + "epoch": 6.15, + "learning_rate": 2.675359696425578e-06, + "loss": 0.0848, + "step": 14687 + }, + { + "epoch": 6.15, + "learning_rate": 2.67420555761866e-06, + "loss": 0.0681, + "step": 14688 + }, + { + "epoch": 6.15, + "learning_rate": 2.6730516293852014e-06, + "loss": 0.093, + "step": 14689 + }, + { + "epoch": 6.15, + "learning_rate": 2.671897911758373e-06, + "loss": 0.0732, + "step": 14690 + }, + { + "epoch": 6.15, + "learning_rate": 2.6707444047713373e-06, + "loss": 0.1054, + "step": 14691 + }, + { + "epoch": 6.15, + "learning_rate": 2.6695911084572514e-06, + "loss": 0.0781, + "step": 14692 + }, + { + "epoch": 6.15, + "learning_rate": 2.668438022849262e-06, + "loss": 0.0715, + "step": 14693 + }, + { + "epoch": 6.15, + "learning_rate": 2.667285147980515e-06, + "loss": 0.0876, + "step": 14694 + }, + { + "epoch": 6.15, + "learning_rate": 2.6661324838841495e-06, + "loss": 0.0982, + "step": 14695 + }, + { + "epoch": 6.15, + "learning_rate": 2.6649800305932984e-06, + "loss": 0.1052, + "step": 14696 + }, + { + "epoch": 6.15, + "learning_rate": 2.663827788141088e-06, + "loss": 0.0998, + "step": 14697 + }, + { + "epoch": 6.15, + "learning_rate": 2.6626757565606398e-06, + "loss": 0.0784, + "step": 14698 + }, + { + "epoch": 6.15, + "learning_rate": 2.661523935885063e-06, + "loss": 0.0928, + "step": 14699 + }, + { + "epoch": 6.15, + "learning_rate": 2.660372326147471e-06, + "loss": 0.0695, + "step": 14700 + }, + { + "epoch": 6.15, + "learning_rate": 2.659220927380961e-06, + "loss": 0.0439, + "step": 14701 + }, + { + "epoch": 6.16, + "learning_rate": 2.658069739618632e-06, + "loss": 0.0846, + "step": 14702 + }, + { + "epoch": 6.16, + "learning_rate": 2.6569187628935723e-06, + "loss": 0.072, + "step": 14703 + }, + { + "epoch": 6.16, + "learning_rate": 2.6557679972388674e-06, + "loss": 0.0804, + "step": 14704 + }, + { + "epoch": 6.16, + "learning_rate": 2.6546174426875937e-06, + "loss": 0.0963, + "step": 14705 + }, + { + "epoch": 6.16, + "learning_rate": 2.653467099272826e-06, + "loss": 0.0867, + "step": 14706 + }, + { + "epoch": 6.16, + "learning_rate": 2.652316967027625e-06, + "loss": 0.0503, + "step": 14707 + }, + { + "epoch": 6.16, + "learning_rate": 2.6511670459850526e-06, + "loss": 0.0855, + "step": 14708 + }, + { + "epoch": 6.16, + "learning_rate": 2.650017336178162e-06, + "loss": 0.0748, + "step": 14709 + }, + { + "epoch": 6.16, + "learning_rate": 2.648867837640001e-06, + "loss": 0.0822, + "step": 14710 + }, + { + "epoch": 6.16, + "learning_rate": 2.6477185504036106e-06, + "loss": 0.0679, + "step": 14711 + }, + { + "epoch": 6.16, + "learning_rate": 2.6465694745020265e-06, + "loss": 0.0653, + "step": 14712 + }, + { + "epoch": 6.16, + "learning_rate": 2.64542060996828e-06, + "loss": 0.0884, + "step": 14713 + }, + { + "epoch": 6.16, + "learning_rate": 2.6442719568353894e-06, + "loss": 0.0766, + "step": 14714 + }, + { + "epoch": 6.16, + "learning_rate": 2.643123515136373e-06, + "loss": 0.0957, + "step": 14715 + }, + { + "epoch": 6.16, + "learning_rate": 2.6419752849042437e-06, + "loss": 0.0947, + "step": 14716 + }, + { + "epoch": 6.16, + "learning_rate": 2.640827266172006e-06, + "loss": 0.1111, + "step": 14717 + }, + { + "epoch": 6.16, + "learning_rate": 2.639679458972657e-06, + "loss": 0.0927, + "step": 14718 + }, + { + "epoch": 6.16, + "learning_rate": 2.6385318633391934e-06, + "loss": 0.04, + "step": 14719 + }, + { + "epoch": 6.16, + "learning_rate": 2.6373844793045965e-06, + "loss": 0.0662, + "step": 14720 + }, + { + "epoch": 6.16, + "learning_rate": 2.636237306901852e-06, + "loss": 0.0734, + "step": 14721 + }, + { + "epoch": 6.16, + "learning_rate": 2.6350903461639287e-06, + "loss": 0.0683, + "step": 14722 + }, + { + "epoch": 6.16, + "learning_rate": 2.633943597123798e-06, + "loss": 0.0921, + "step": 14723 + }, + { + "epoch": 6.16, + "learning_rate": 2.6327970598144214e-06, + "loss": 0.0751, + "step": 14724 + }, + { + "epoch": 6.16, + "learning_rate": 2.631650734268757e-06, + "loss": 0.0901, + "step": 14725 + }, + { + "epoch": 6.17, + "learning_rate": 2.630504620519753e-06, + "loss": 0.095, + "step": 14726 + }, + { + "epoch": 6.17, + "learning_rate": 2.6293587186003568e-06, + "loss": 0.082, + "step": 14727 + }, + { + "epoch": 6.17, + "learning_rate": 2.6282130285435016e-06, + "loss": 0.1297, + "step": 14728 + }, + { + "epoch": 6.17, + "learning_rate": 2.6270675503821207e-06, + "loss": 0.0839, + "step": 14729 + }, + { + "epoch": 6.17, + "learning_rate": 2.6259222841491406e-06, + "loss": 0.0984, + "step": 14730 + }, + { + "epoch": 6.17, + "learning_rate": 2.6247772298774797e-06, + "loss": 0.0922, + "step": 14731 + }, + { + "epoch": 6.17, + "learning_rate": 2.6236323876000547e-06, + "loss": 0.0765, + "step": 14732 + }, + { + "epoch": 6.17, + "learning_rate": 2.622487757349772e-06, + "loss": 0.09, + "step": 14733 + }, + { + "epoch": 6.17, + "learning_rate": 2.62134333915953e-06, + "loss": 0.092, + "step": 14734 + }, + { + "epoch": 6.17, + "learning_rate": 2.620199133062227e-06, + "loss": 0.1081, + "step": 14735 + }, + { + "epoch": 6.17, + "learning_rate": 2.6190551390907495e-06, + "loss": 0.0826, + "step": 14736 + }, + { + "epoch": 6.17, + "learning_rate": 2.6179113572779835e-06, + "loss": 0.0785, + "step": 14737 + }, + { + "epoch": 6.17, + "learning_rate": 2.6167677876568044e-06, + "loss": 0.1054, + "step": 14738 + }, + { + "epoch": 6.17, + "learning_rate": 2.6156244302600864e-06, + "loss": 0.0771, + "step": 14739 + }, + { + "epoch": 6.17, + "learning_rate": 2.614481285120688e-06, + "loss": 0.0614, + "step": 14740 + }, + { + "epoch": 6.17, + "learning_rate": 2.6133383522714718e-06, + "loss": 0.0825, + "step": 14741 + }, + { + "epoch": 6.17, + "learning_rate": 2.612195631745289e-06, + "loss": 0.1003, + "step": 14742 + }, + { + "epoch": 6.17, + "learning_rate": 2.611053123574988e-06, + "loss": 0.0942, + "step": 14743 + }, + { + "epoch": 6.17, + "learning_rate": 2.6099108277934105e-06, + "loss": 0.1015, + "step": 14744 + }, + { + "epoch": 6.17, + "learning_rate": 2.6087687444333853e-06, + "loss": 0.0999, + "step": 14745 + }, + { + "epoch": 6.17, + "learning_rate": 2.6076268735277445e-06, + "loss": 0.068, + "step": 14746 + }, + { + "epoch": 6.17, + "learning_rate": 2.6064852151093124e-06, + "loss": 0.115, + "step": 14747 + }, + { + "epoch": 6.17, + "learning_rate": 2.6053437692108974e-06, + "loss": 0.0829, + "step": 14748 + }, + { + "epoch": 6.18, + "learning_rate": 2.6042025358653157e-06, + "loss": 0.0689, + "step": 14749 + }, + { + "epoch": 6.18, + "learning_rate": 2.6030615151053675e-06, + "loss": 0.0944, + "step": 14750 + }, + { + "epoch": 6.18, + "learning_rate": 2.601920706963853e-06, + "loss": 0.0837, + "step": 14751 + }, + { + "epoch": 6.18, + "learning_rate": 2.600780111473563e-06, + "loss": 0.0939, + "step": 14752 + }, + { + "epoch": 6.18, + "learning_rate": 2.5996397286672826e-06, + "loss": 0.1301, + "step": 14753 + }, + { + "epoch": 6.18, + "learning_rate": 2.598499558577794e-06, + "loss": 0.0731, + "step": 14754 + }, + { + "epoch": 6.18, + "learning_rate": 2.5973596012378644e-06, + "loss": 0.099, + "step": 14755 + }, + { + "epoch": 6.18, + "learning_rate": 2.5962198566802643e-06, + "loss": 0.0624, + "step": 14756 + }, + { + "epoch": 6.18, + "learning_rate": 2.5950803249377554e-06, + "loss": 0.0775, + "step": 14757 + }, + { + "epoch": 6.18, + "learning_rate": 2.5939410060430904e-06, + "loss": 0.0849, + "step": 14758 + }, + { + "epoch": 6.18, + "learning_rate": 2.5928019000290194e-06, + "loss": 0.0922, + "step": 14759 + }, + { + "epoch": 6.18, + "learning_rate": 2.591663006928288e-06, + "loss": 0.045, + "step": 14760 + }, + { + "epoch": 6.18, + "learning_rate": 2.5905243267736256e-06, + "loss": 0.1222, + "step": 14761 + }, + { + "epoch": 6.18, + "learning_rate": 2.589385859597767e-06, + "loss": 0.0858, + "step": 14762 + }, + { + "epoch": 6.18, + "learning_rate": 2.588247605433436e-06, + "loss": 0.0735, + "step": 14763 + }, + { + "epoch": 6.18, + "learning_rate": 2.5871095643133502e-06, + "loss": 0.0729, + "step": 14764 + }, + { + "epoch": 6.18, + "learning_rate": 2.585971736270223e-06, + "loss": 0.0955, + "step": 14765 + }, + { + "epoch": 6.18, + "learning_rate": 2.584834121336761e-06, + "loss": 0.0926, + "step": 14766 + }, + { + "epoch": 6.18, + "learning_rate": 2.5836967195456586e-06, + "loss": 0.0997, + "step": 14767 + }, + { + "epoch": 6.18, + "learning_rate": 2.582559530929617e-06, + "loss": 0.0763, + "step": 14768 + }, + { + "epoch": 6.18, + "learning_rate": 2.581422555521316e-06, + "loss": 0.0615, + "step": 14769 + }, + { + "epoch": 6.18, + "learning_rate": 2.5802857933534407e-06, + "loss": 0.0617, + "step": 14770 + }, + { + "epoch": 6.18, + "learning_rate": 2.579149244458666e-06, + "loss": 0.0791, + "step": 14771 + }, + { + "epoch": 6.18, + "learning_rate": 2.5780129088696614e-06, + "loss": 0.0943, + "step": 14772 + }, + { + "epoch": 6.19, + "learning_rate": 2.5768767866190893e-06, + "loss": 0.1239, + "step": 14773 + }, + { + "epoch": 6.19, + "learning_rate": 2.57574087773961e-06, + "loss": 0.101, + "step": 14774 + }, + { + "epoch": 6.19, + "learning_rate": 2.5746051822638675e-06, + "loss": 0.0708, + "step": 14775 + }, + { + "epoch": 6.19, + "learning_rate": 2.57346970022451e-06, + "loss": 0.0843, + "step": 14776 + }, + { + "epoch": 6.19, + "learning_rate": 2.5723344316541767e-06, + "loss": 0.0627, + "step": 14777 + }, + { + "epoch": 6.19, + "learning_rate": 2.5711993765854983e-06, + "loss": 0.0937, + "step": 14778 + }, + { + "epoch": 6.19, + "learning_rate": 2.570064535051101e-06, + "loss": 0.1021, + "step": 14779 + }, + { + "epoch": 6.19, + "learning_rate": 2.568929907083609e-06, + "loss": 0.0855, + "step": 14780 + }, + { + "epoch": 6.19, + "learning_rate": 2.56779549271563e-06, + "loss": 0.084, + "step": 14781 + }, + { + "epoch": 6.19, + "learning_rate": 2.566661291979774e-06, + "loss": 0.0663, + "step": 14782 + }, + { + "epoch": 6.19, + "learning_rate": 2.5655273049086425e-06, + "loss": 0.0879, + "step": 14783 + }, + { + "epoch": 6.19, + "learning_rate": 2.564393531534832e-06, + "loss": 0.0717, + "step": 14784 + }, + { + "epoch": 6.19, + "learning_rate": 2.563259971890931e-06, + "loss": 0.0717, + "step": 14785 + }, + { + "epoch": 6.19, + "learning_rate": 2.562126626009522e-06, + "loss": 0.1152, + "step": 14786 + }, + { + "epoch": 6.19, + "learning_rate": 2.5609934939231873e-06, + "loss": 0.0707, + "step": 14787 + }, + { + "epoch": 6.19, + "learning_rate": 2.5598605756644902e-06, + "loss": 0.0919, + "step": 14788 + }, + { + "epoch": 6.19, + "learning_rate": 2.558727871266001e-06, + "loss": 0.0781, + "step": 14789 + }, + { + "epoch": 6.19, + "learning_rate": 2.5575953807602727e-06, + "loss": 0.0687, + "step": 14790 + }, + { + "epoch": 6.19, + "learning_rate": 2.5564631041798615e-06, + "loss": 0.0992, + "step": 14791 + }, + { + "epoch": 6.19, + "learning_rate": 2.555331041557313e-06, + "loss": 0.1118, + "step": 14792 + }, + { + "epoch": 6.19, + "learning_rate": 2.5541991929251676e-06, + "loss": 0.0891, + "step": 14793 + }, + { + "epoch": 6.19, + "learning_rate": 2.553067558315958e-06, + "loss": 0.1247, + "step": 14794 + }, + { + "epoch": 6.19, + "learning_rate": 2.5519361377622176e-06, + "loss": 0.0741, + "step": 14795 + }, + { + "epoch": 6.19, + "learning_rate": 2.55080493129646e-06, + "loss": 0.0669, + "step": 14796 + }, + { + "epoch": 6.2, + "learning_rate": 2.5496739389512036e-06, + "loss": 0.0552, + "step": 14797 + }, + { + "epoch": 6.2, + "learning_rate": 2.5485431607589594e-06, + "loss": 0.0481, + "step": 14798 + }, + { + "epoch": 6.2, + "learning_rate": 2.54741259675223e-06, + "loss": 0.0748, + "step": 14799 + }, + { + "epoch": 6.2, + "learning_rate": 2.5462822469635117e-06, + "loss": 0.0843, + "step": 14800 + }, + { + "epoch": 6.2, + "learning_rate": 2.5451521114252985e-06, + "loss": 0.0868, + "step": 14801 + }, + { + "epoch": 6.2, + "learning_rate": 2.54402219017007e-06, + "loss": 0.0869, + "step": 14802 + }, + { + "epoch": 6.2, + "learning_rate": 2.542892483230308e-06, + "loss": 0.0633, + "step": 14803 + }, + { + "epoch": 6.2, + "learning_rate": 2.5417629906384834e-06, + "loss": 0.1067, + "step": 14804 + }, + { + "epoch": 6.2, + "learning_rate": 2.540633712427063e-06, + "loss": 0.0969, + "step": 14805 + }, + { + "epoch": 6.2, + "learning_rate": 2.539504648628508e-06, + "loss": 0.0993, + "step": 14806 + }, + { + "epoch": 6.2, + "learning_rate": 2.5383757992752744e-06, + "loss": 0.1001, + "step": 14807 + }, + { + "epoch": 6.2, + "learning_rate": 2.5372471643998042e-06, + "loss": 0.1042, + "step": 14808 + }, + { + "epoch": 6.2, + "learning_rate": 2.5361187440345416e-06, + "loss": 0.0951, + "step": 14809 + }, + { + "epoch": 6.2, + "learning_rate": 2.534990538211922e-06, + "loss": 0.0607, + "step": 14810 + }, + { + "epoch": 6.2, + "learning_rate": 2.533862546964376e-06, + "loss": 0.0593, + "step": 14811 + }, + { + "epoch": 6.2, + "learning_rate": 2.532734770324329e-06, + "loss": 0.1005, + "step": 14812 + }, + { + "epoch": 6.2, + "learning_rate": 2.5316072083241904e-06, + "loss": 0.0799, + "step": 14813 + }, + { + "epoch": 6.2, + "learning_rate": 2.5304798609963754e-06, + "loss": 0.0865, + "step": 14814 + }, + { + "epoch": 6.2, + "learning_rate": 2.529352728373292e-06, + "loss": 0.0834, + "step": 14815 + }, + { + "epoch": 6.2, + "learning_rate": 2.5282258104873324e-06, + "loss": 0.0644, + "step": 14816 + }, + { + "epoch": 6.2, + "learning_rate": 2.5270991073708905e-06, + "loss": 0.1018, + "step": 14817 + }, + { + "epoch": 6.2, + "learning_rate": 2.525972619056355e-06, + "loss": 0.0698, + "step": 14818 + }, + { + "epoch": 6.2, + "learning_rate": 2.524846345576103e-06, + "loss": 0.0751, + "step": 14819 + }, + { + "epoch": 6.2, + "learning_rate": 2.52372028696251e-06, + "loss": 0.062, + "step": 14820 + }, + { + "epoch": 6.21, + "learning_rate": 2.5225944432479452e-06, + "loss": 0.1017, + "step": 14821 + }, + { + "epoch": 6.21, + "learning_rate": 2.5214688144647658e-06, + "loss": 0.1301, + "step": 14822 + }, + { + "epoch": 6.21, + "learning_rate": 2.5203434006453287e-06, + "loss": 0.1092, + "step": 14823 + }, + { + "epoch": 6.21, + "learning_rate": 2.5192182018219826e-06, + "loss": 0.0819, + "step": 14824 + }, + { + "epoch": 6.21, + "learning_rate": 2.518093218027071e-06, + "loss": 0.0859, + "step": 14825 + }, + { + "epoch": 6.21, + "learning_rate": 2.5169684492929315e-06, + "loss": 0.0949, + "step": 14826 + }, + { + "epoch": 6.21, + "learning_rate": 2.515843895651893e-06, + "loss": 0.1155, + "step": 14827 + }, + { + "epoch": 6.21, + "learning_rate": 2.5147195571362837e-06, + "loss": 0.0719, + "step": 14828 + }, + { + "epoch": 6.21, + "learning_rate": 2.513595433778415e-06, + "loss": 0.0965, + "step": 14829 + }, + { + "epoch": 6.21, + "learning_rate": 2.5124715256106024e-06, + "loss": 0.0823, + "step": 14830 + }, + { + "epoch": 6.21, + "learning_rate": 2.511347832665152e-06, + "loss": 0.0948, + "step": 14831 + }, + { + "epoch": 6.21, + "learning_rate": 2.510224354974362e-06, + "loss": 0.1077, + "step": 14832 + }, + { + "epoch": 6.21, + "learning_rate": 2.509101092570527e-06, + "loss": 0.0804, + "step": 14833 + }, + { + "epoch": 6.21, + "learning_rate": 2.507978045485937e-06, + "loss": 0.0918, + "step": 14834 + }, + { + "epoch": 6.21, + "learning_rate": 2.5068552137528667e-06, + "loss": 0.0652, + "step": 14835 + }, + { + "epoch": 6.21, + "learning_rate": 2.5057325974035964e-06, + "loss": 0.0755, + "step": 14836 + }, + { + "epoch": 6.21, + "learning_rate": 2.5046101964703894e-06, + "loss": 0.0666, + "step": 14837 + }, + { + "epoch": 6.21, + "learning_rate": 2.503488010985512e-06, + "loss": 0.0726, + "step": 14838 + }, + { + "epoch": 6.21, + "learning_rate": 2.502366040981218e-06, + "loss": 0.0622, + "step": 14839 + }, + { + "epoch": 6.21, + "learning_rate": 2.50124428648976e-06, + "loss": 0.0765, + "step": 14840 + }, + { + "epoch": 6.21, + "learning_rate": 2.5001227475433797e-06, + "loss": 0.1044, + "step": 14841 + }, + { + "epoch": 6.21, + "learning_rate": 2.49900142417432e-06, + "loss": 0.0872, + "step": 14842 + }, + { + "epoch": 6.21, + "learning_rate": 2.4978803164148047e-06, + "loss": 0.1062, + "step": 14843 + }, + { + "epoch": 6.21, + "learning_rate": 2.496759424297062e-06, + "loss": 0.0653, + "step": 14844 + }, + { + "epoch": 6.22, + "learning_rate": 2.495638747853312e-06, + "loss": 0.0966, + "step": 14845 + }, + { + "epoch": 6.22, + "learning_rate": 2.4945182871157656e-06, + "loss": 0.1056, + "step": 14846 + }, + { + "epoch": 6.22, + "learning_rate": 2.4933980421166326e-06, + "loss": 0.1049, + "step": 14847 + }, + { + "epoch": 6.22, + "learning_rate": 2.492278012888113e-06, + "loss": 0.0643, + "step": 14848 + }, + { + "epoch": 6.22, + "learning_rate": 2.491158199462398e-06, + "loss": 0.0765, + "step": 14849 + }, + { + "epoch": 6.22, + "learning_rate": 2.490038601871676e-06, + "loss": 0.0877, + "step": 14850 + }, + { + "epoch": 6.22, + "learning_rate": 2.4889192201481315e-06, + "loss": 0.0766, + "step": 14851 + }, + { + "epoch": 6.22, + "learning_rate": 2.4878000543239377e-06, + "loss": 0.0746, + "step": 14852 + }, + { + "epoch": 6.22, + "learning_rate": 2.4866811044312667e-06, + "loss": 0.0629, + "step": 14853 + }, + { + "epoch": 6.22, + "learning_rate": 2.485562370502279e-06, + "loss": 0.0779, + "step": 14854 + }, + { + "epoch": 6.22, + "learning_rate": 2.484443852569135e-06, + "loss": 0.0831, + "step": 14855 + }, + { + "epoch": 6.22, + "learning_rate": 2.483325550663982e-06, + "loss": 0.0806, + "step": 14856 + }, + { + "epoch": 6.22, + "learning_rate": 2.4822074648189675e-06, + "loss": 0.0728, + "step": 14857 + }, + { + "epoch": 6.22, + "learning_rate": 2.481089595066226e-06, + "loss": 0.0995, + "step": 14858 + }, + { + "epoch": 6.22, + "learning_rate": 2.479971941437892e-06, + "loss": 0.0717, + "step": 14859 + }, + { + "epoch": 6.22, + "learning_rate": 2.4788545039660905e-06, + "loss": 0.082, + "step": 14860 + }, + { + "epoch": 6.22, + "learning_rate": 2.4777372826829427e-06, + "loss": 0.1138, + "step": 14861 + }, + { + "epoch": 6.22, + "learning_rate": 2.476620277620565e-06, + "loss": 0.0985, + "step": 14862 + }, + { + "epoch": 6.22, + "learning_rate": 2.4755034888110576e-06, + "loss": 0.1266, + "step": 14863 + }, + { + "epoch": 6.22, + "learning_rate": 2.474386916286526e-06, + "loss": 0.0751, + "step": 14864 + }, + { + "epoch": 6.22, + "learning_rate": 2.473270560079064e-06, + "loss": 0.0774, + "step": 14865 + }, + { + "epoch": 6.22, + "learning_rate": 2.47215442022076e-06, + "loss": 0.101, + "step": 14866 + }, + { + "epoch": 6.22, + "learning_rate": 2.4710384967436983e-06, + "loss": 0.0881, + "step": 14867 + }, + { + "epoch": 6.22, + "learning_rate": 2.469922789679954e-06, + "loss": 0.0825, + "step": 14868 + }, + { + "epoch": 6.23, + "learning_rate": 2.4688072990615987e-06, + "loss": 0.0701, + "step": 14869 + }, + { + "epoch": 6.23, + "learning_rate": 2.4676920249206926e-06, + "loss": 0.1069, + "step": 14870 + }, + { + "epoch": 6.23, + "learning_rate": 2.4665769672892946e-06, + "loss": 0.0739, + "step": 14871 + }, + { + "epoch": 6.23, + "learning_rate": 2.465462126199457e-06, + "loss": 0.0832, + "step": 14872 + }, + { + "epoch": 6.23, + "learning_rate": 2.464347501683224e-06, + "loss": 0.0882, + "step": 14873 + }, + { + "epoch": 6.23, + "learning_rate": 2.4632330937726356e-06, + "loss": 0.0955, + "step": 14874 + }, + { + "epoch": 6.23, + "learning_rate": 2.4621189024997273e-06, + "loss": 0.1045, + "step": 14875 + }, + { + "epoch": 6.23, + "learning_rate": 2.4610049278965175e-06, + "loss": 0.104, + "step": 14876 + }, + { + "epoch": 6.23, + "learning_rate": 2.4598911699950325e-06, + "loss": 0.0882, + "step": 14877 + }, + { + "epoch": 6.23, + "learning_rate": 2.4587776288272846e-06, + "loss": 0.0646, + "step": 14878 + }, + { + "epoch": 6.23, + "learning_rate": 2.4576643044252824e-06, + "loss": 0.0777, + "step": 14879 + }, + { + "epoch": 6.23, + "learning_rate": 2.4565511968210285e-06, + "loss": 0.0748, + "step": 14880 + }, + { + "epoch": 6.23, + "learning_rate": 2.455438306046515e-06, + "loss": 0.1052, + "step": 14881 + }, + { + "epoch": 6.23, + "learning_rate": 2.4543256321337306e-06, + "loss": 0.0863, + "step": 14882 + }, + { + "epoch": 6.23, + "learning_rate": 2.453213175114664e-06, + "loss": 0.0721, + "step": 14883 + }, + { + "epoch": 6.23, + "learning_rate": 2.4521009350212854e-06, + "loss": 0.0985, + "step": 14884 + }, + { + "epoch": 6.23, + "learning_rate": 2.4509889118855666e-06, + "loss": 0.0605, + "step": 14885 + }, + { + "epoch": 6.23, + "learning_rate": 2.449877105739473e-06, + "loss": 0.0552, + "step": 14886 + }, + { + "epoch": 6.23, + "learning_rate": 2.4487655166149625e-06, + "loss": 0.0864, + "step": 14887 + }, + { + "epoch": 6.23, + "learning_rate": 2.447654144543986e-06, + "loss": 0.0859, + "step": 14888 + }, + { + "epoch": 6.23, + "learning_rate": 2.4465429895584914e-06, + "loss": 0.0513, + "step": 14889 + }, + { + "epoch": 6.23, + "learning_rate": 2.445432051690414e-06, + "loss": 0.0702, + "step": 14890 + }, + { + "epoch": 6.23, + "learning_rate": 2.4443213309716884e-06, + "loss": 0.0878, + "step": 14891 + }, + { + "epoch": 6.23, + "learning_rate": 2.4432108274342405e-06, + "loss": 0.0621, + "step": 14892 + }, + { + "epoch": 6.24, + "learning_rate": 2.4421005411099928e-06, + "loss": 0.0659, + "step": 14893 + }, + { + "epoch": 6.24, + "learning_rate": 2.4409904720308576e-06, + "loss": 0.0877, + "step": 14894 + }, + { + "epoch": 6.24, + "learning_rate": 2.4398806202287427e-06, + "loss": 0.1058, + "step": 14895 + }, + { + "epoch": 6.24, + "learning_rate": 2.4387709857355545e-06, + "loss": 0.0755, + "step": 14896 + }, + { + "epoch": 6.24, + "learning_rate": 2.4376615685831817e-06, + "loss": 0.1033, + "step": 14897 + }, + { + "epoch": 6.24, + "learning_rate": 2.4365523688035163e-06, + "loss": 0.0995, + "step": 14898 + }, + { + "epoch": 6.24, + "learning_rate": 2.435443386428442e-06, + "loss": 0.0744, + "step": 14899 + }, + { + "epoch": 6.24, + "learning_rate": 2.4343346214898343e-06, + "loss": 0.0632, + "step": 14900 + }, + { + "epoch": 6.24, + "learning_rate": 2.4332260740195645e-06, + "loss": 0.0854, + "step": 14901 + }, + { + "epoch": 6.24, + "learning_rate": 2.4321177440494993e-06, + "loss": 0.0875, + "step": 14902 + }, + { + "epoch": 6.24, + "learning_rate": 2.431009631611493e-06, + "loss": 0.0932, + "step": 14903 + }, + { + "epoch": 6.24, + "learning_rate": 2.4299017367373966e-06, + "loss": 0.0912, + "step": 14904 + }, + { + "epoch": 6.24, + "learning_rate": 2.4287940594590564e-06, + "loss": 0.0737, + "step": 14905 + }, + { + "epoch": 6.24, + "learning_rate": 2.4276865998083132e-06, + "loss": 0.0779, + "step": 14906 + }, + { + "epoch": 6.24, + "learning_rate": 2.426579357816998e-06, + "loss": 0.0774, + "step": 14907 + }, + { + "epoch": 6.24, + "learning_rate": 2.4254723335169394e-06, + "loss": 0.0942, + "step": 14908 + }, + { + "epoch": 6.24, + "learning_rate": 2.424365526939958e-06, + "loss": 0.0811, + "step": 14909 + }, + { + "epoch": 6.24, + "learning_rate": 2.4232589381178685e-06, + "loss": 0.0895, + "step": 14910 + }, + { + "epoch": 6.24, + "learning_rate": 2.422152567082474e-06, + "loss": 0.1295, + "step": 14911 + }, + { + "epoch": 6.24, + "learning_rate": 2.4210464138655807e-06, + "loss": 0.1035, + "step": 14912 + }, + { + "epoch": 6.24, + "learning_rate": 2.4199404784989833e-06, + "loss": 0.0901, + "step": 14913 + }, + { + "epoch": 6.24, + "learning_rate": 2.4188347610144693e-06, + "loss": 0.0943, + "step": 14914 + }, + { + "epoch": 6.24, + "learning_rate": 2.417729261443823e-06, + "loss": 0.07, + "step": 14915 + }, + { + "epoch": 6.24, + "learning_rate": 2.416623979818823e-06, + "loss": 0.0666, + "step": 14916 + }, + { + "epoch": 6.25, + "learning_rate": 2.4155189161712358e-06, + "loss": 0.0687, + "step": 14917 + }, + { + "epoch": 6.25, + "learning_rate": 2.4144140705328258e-06, + "loss": 0.0905, + "step": 14918 + }, + { + "epoch": 6.25, + "learning_rate": 2.4133094429353522e-06, + "loss": 0.0994, + "step": 14919 + }, + { + "epoch": 6.25, + "learning_rate": 2.412205033410566e-06, + "loss": 0.0748, + "step": 14920 + }, + { + "epoch": 6.25, + "learning_rate": 2.4111008419902138e-06, + "loss": 0.1008, + "step": 14921 + }, + { + "epoch": 6.25, + "learning_rate": 2.409996868706036e-06, + "loss": 0.1057, + "step": 14922 + }, + { + "epoch": 6.25, + "learning_rate": 2.4088931135897596e-06, + "loss": 0.0828, + "step": 14923 + }, + { + "epoch": 6.25, + "learning_rate": 2.407789576673115e-06, + "loss": 0.0655, + "step": 14924 + }, + { + "epoch": 6.25, + "learning_rate": 2.406686257987825e-06, + "loss": 0.0944, + "step": 14925 + }, + { + "epoch": 6.25, + "learning_rate": 2.4055831575655975e-06, + "loss": 0.0802, + "step": 14926 + }, + { + "epoch": 6.25, + "learning_rate": 2.4044802754381423e-06, + "loss": 0.0785, + "step": 14927 + }, + { + "epoch": 6.25, + "learning_rate": 2.4033776116371623e-06, + "loss": 0.0997, + "step": 14928 + }, + { + "epoch": 6.25, + "learning_rate": 2.402275166194351e-06, + "loss": 0.084, + "step": 14929 + }, + { + "epoch": 6.25, + "learning_rate": 2.401172939141401e-06, + "loss": 0.0743, + "step": 14930 + }, + { + "epoch": 6.25, + "learning_rate": 2.400070930509989e-06, + "loss": 0.0961, + "step": 14931 + }, + { + "epoch": 6.25, + "learning_rate": 2.398969140331794e-06, + "loss": 0.0833, + "step": 14932 + }, + { + "epoch": 6.25, + "learning_rate": 2.3978675686384854e-06, + "loss": 0.0987, + "step": 14933 + }, + { + "epoch": 6.25, + "learning_rate": 2.396766215461728e-06, + "loss": 0.0845, + "step": 14934 + }, + { + "epoch": 6.25, + "learning_rate": 2.3956650808331795e-06, + "loss": 0.0589, + "step": 14935 + }, + { + "epoch": 6.25, + "learning_rate": 2.394564164784489e-06, + "loss": 0.1022, + "step": 14936 + }, + { + "epoch": 6.25, + "learning_rate": 2.3934634673473066e-06, + "loss": 0.0674, + "step": 14937 + }, + { + "epoch": 6.25, + "learning_rate": 2.392362988553263e-06, + "loss": 0.0875, + "step": 14938 + }, + { + "epoch": 6.25, + "learning_rate": 2.391262728433995e-06, + "loss": 0.084, + "step": 14939 + }, + { + "epoch": 6.25, + "learning_rate": 2.3901626870211282e-06, + "loss": 0.0814, + "step": 14940 + }, + { + "epoch": 6.26, + "learning_rate": 2.389062864346282e-06, + "loss": 0.0893, + "step": 14941 + }, + { + "epoch": 6.26, + "learning_rate": 2.3879632604410695e-06, + "loss": 0.101, + "step": 14942 + }, + { + "epoch": 6.26, + "learning_rate": 2.386863875337102e-06, + "loss": 0.0771, + "step": 14943 + }, + { + "epoch": 6.26, + "learning_rate": 2.385764709065973e-06, + "loss": 0.09, + "step": 14944 + }, + { + "epoch": 6.26, + "learning_rate": 2.3846657616592805e-06, + "loss": 0.0615, + "step": 14945 + }, + { + "epoch": 6.26, + "learning_rate": 2.3835670331486137e-06, + "loss": 0.094, + "step": 14946 + }, + { + "epoch": 6.26, + "learning_rate": 2.3824685235655532e-06, + "loss": 0.083, + "step": 14947 + }, + { + "epoch": 6.26, + "learning_rate": 2.381370232941679e-06, + "loss": 0.0625, + "step": 14948 + }, + { + "epoch": 6.26, + "learning_rate": 2.3802721613085534e-06, + "loss": 0.1044, + "step": 14949 + }, + { + "epoch": 6.26, + "learning_rate": 2.3791743086977425e-06, + "loss": 0.0795, + "step": 14950 + }, + { + "epoch": 6.26, + "learning_rate": 2.378076675140807e-06, + "loss": 0.0791, + "step": 14951 + }, + { + "epoch": 6.26, + "learning_rate": 2.376979260669291e-06, + "loss": 0.0932, + "step": 14952 + }, + { + "epoch": 6.26, + "learning_rate": 2.3758820653147405e-06, + "loss": 0.0702, + "step": 14953 + }, + { + "epoch": 6.26, + "learning_rate": 2.3747850891086956e-06, + "loss": 0.0805, + "step": 14954 + }, + { + "epoch": 6.26, + "learning_rate": 2.3736883320826864e-06, + "loss": 0.0756, + "step": 14955 + }, + { + "epoch": 6.26, + "learning_rate": 2.3725917942682397e-06, + "loss": 0.099, + "step": 14956 + }, + { + "epoch": 6.26, + "learning_rate": 2.3714954756968746e-06, + "loss": 0.0872, + "step": 14957 + }, + { + "epoch": 6.26, + "learning_rate": 2.370399376400101e-06, + "loss": 0.0862, + "step": 14958 + }, + { + "epoch": 6.26, + "learning_rate": 2.3693034964094264e-06, + "loss": 0.0518, + "step": 14959 + }, + { + "epoch": 6.26, + "learning_rate": 2.368207835756352e-06, + "loss": 0.0863, + "step": 14960 + }, + { + "epoch": 6.26, + "learning_rate": 2.367112394472371e-06, + "loss": 0.0785, + "step": 14961 + }, + { + "epoch": 6.26, + "learning_rate": 2.3660171725889703e-06, + "loss": 0.1134, + "step": 14962 + }, + { + "epoch": 6.26, + "learning_rate": 2.3649221701376346e-06, + "loss": 0.0889, + "step": 14963 + }, + { + "epoch": 6.27, + "learning_rate": 2.363827387149833e-06, + "loss": 0.0916, + "step": 14964 + }, + { + "epoch": 6.27, + "learning_rate": 2.3627328236570367e-06, + "loss": 0.087, + "step": 14965 + }, + { + "epoch": 6.27, + "learning_rate": 2.3616384796907087e-06, + "loss": 0.0704, + "step": 14966 + }, + { + "epoch": 6.27, + "learning_rate": 2.360544355282305e-06, + "loss": 0.0926, + "step": 14967 + }, + { + "epoch": 6.27, + "learning_rate": 2.359450450463274e-06, + "loss": 0.0874, + "step": 14968 + }, + { + "epoch": 6.27, + "learning_rate": 2.35835676526506e-06, + "loss": 0.1054, + "step": 14969 + }, + { + "epoch": 6.27, + "learning_rate": 2.3572632997191016e-06, + "loss": 0.0871, + "step": 14970 + }, + { + "epoch": 6.27, + "learning_rate": 2.356170053856829e-06, + "loss": 0.0815, + "step": 14971 + }, + { + "epoch": 6.27, + "learning_rate": 2.3550770277096615e-06, + "loss": 0.0962, + "step": 14972 + }, + { + "epoch": 6.27, + "learning_rate": 2.35398422130902e-06, + "loss": 0.0683, + "step": 14973 + }, + { + "epoch": 6.27, + "learning_rate": 2.3528916346863185e-06, + "loss": 0.0937, + "step": 14974 + }, + { + "epoch": 6.27, + "learning_rate": 2.3517992678729607e-06, + "loss": 0.0538, + "step": 14975 + }, + { + "epoch": 6.27, + "learning_rate": 2.3507071209003463e-06, + "loss": 0.0896, + "step": 14976 + }, + { + "epoch": 6.27, + "learning_rate": 2.349615193799868e-06, + "loss": 0.0762, + "step": 14977 + }, + { + "epoch": 6.27, + "learning_rate": 2.3485234866029137e-06, + "loss": 0.1219, + "step": 14978 + }, + { + "epoch": 6.27, + "learning_rate": 2.34743199934086e-06, + "loss": 0.1079, + "step": 14979 + }, + { + "epoch": 6.27, + "learning_rate": 2.3463407320450827e-06, + "loss": 0.0814, + "step": 14980 + }, + { + "epoch": 6.27, + "learning_rate": 2.34524968474695e-06, + "loss": 0.1027, + "step": 14981 + }, + { + "epoch": 6.27, + "learning_rate": 2.3441588574778207e-06, + "loss": 0.0672, + "step": 14982 + }, + { + "epoch": 6.27, + "learning_rate": 2.3430682502690527e-06, + "loss": 0.0835, + "step": 14983 + }, + { + "epoch": 6.27, + "learning_rate": 2.3419778631519963e-06, + "loss": 0.079, + "step": 14984 + }, + { + "epoch": 6.27, + "learning_rate": 2.340887696157986e-06, + "loss": 0.0962, + "step": 14985 + }, + { + "epoch": 6.27, + "learning_rate": 2.3397977493183643e-06, + "loss": 0.1165, + "step": 14986 + }, + { + "epoch": 6.27, + "learning_rate": 2.3387080226644578e-06, + "loss": 0.0685, + "step": 14987 + }, + { + "epoch": 6.28, + "learning_rate": 2.337618516227591e-06, + "loss": 0.058, + "step": 14988 + }, + { + "epoch": 6.28, + "learning_rate": 2.3365292300390795e-06, + "loss": 0.0789, + "step": 14989 + }, + { + "epoch": 6.28, + "learning_rate": 2.3354401641302395e-06, + "loss": 0.0779, + "step": 14990 + }, + { + "epoch": 6.28, + "learning_rate": 2.3343513185323664e-06, + "loss": 0.0699, + "step": 14991 + }, + { + "epoch": 6.28, + "learning_rate": 2.333262693276762e-06, + "loss": 0.0783, + "step": 14992 + }, + { + "epoch": 6.28, + "learning_rate": 2.332174288394722e-06, + "loss": 0.0981, + "step": 14993 + }, + { + "epoch": 6.28, + "learning_rate": 2.331086103917525e-06, + "loss": 0.0759, + "step": 14994 + }, + { + "epoch": 6.28, + "learning_rate": 2.3299981398764515e-06, + "loss": 0.0885, + "step": 14995 + }, + { + "epoch": 6.28, + "learning_rate": 2.328910396302776e-06, + "loss": 0.0812, + "step": 14996 + }, + { + "epoch": 6.28, + "learning_rate": 2.327822873227764e-06, + "loss": 0.0845, + "step": 14997 + }, + { + "epoch": 6.28, + "learning_rate": 2.326735570682678e-06, + "loss": 0.0802, + "step": 14998 + }, + { + "epoch": 6.28, + "learning_rate": 2.325648488698766e-06, + "loss": 0.0752, + "step": 14999 + }, + { + "epoch": 6.28, + "learning_rate": 2.324561627307278e-06, + "loss": 0.0878, + "step": 15000 + }, + { + "epoch": 6.28, + "learning_rate": 2.3234749865394545e-06, + "loss": 0.0812, + "step": 15001 + }, + { + "epoch": 6.28, + "learning_rate": 2.322388566426532e-06, + "loss": 0.055, + "step": 15002 + }, + { + "epoch": 6.28, + "learning_rate": 2.321302366999736e-06, + "loss": 0.0815, + "step": 15003 + }, + { + "epoch": 6.28, + "learning_rate": 2.320216388290293e-06, + "loss": 0.0931, + "step": 15004 + }, + { + "epoch": 6.28, + "learning_rate": 2.3191306303294114e-06, + "loss": 0.0878, + "step": 15005 + }, + { + "epoch": 6.28, + "learning_rate": 2.3180450931483045e-06, + "loss": 0.0696, + "step": 15006 + }, + { + "epoch": 6.28, + "learning_rate": 2.3169597767781736e-06, + "loss": 0.0859, + "step": 15007 + }, + { + "epoch": 6.28, + "learning_rate": 2.3158746812502174e-06, + "loss": 0.0825, + "step": 15008 + }, + { + "epoch": 6.28, + "learning_rate": 2.314789806595624e-06, + "loss": 0.1027, + "step": 15009 + }, + { + "epoch": 6.28, + "learning_rate": 2.3137051528455778e-06, + "loss": 0.1111, + "step": 15010 + }, + { + "epoch": 6.28, + "learning_rate": 2.31262072003126e-06, + "loss": 0.0816, + "step": 15011 + }, + { + "epoch": 6.29, + "learning_rate": 2.3115365081838346e-06, + "loss": 0.0815, + "step": 15012 + }, + { + "epoch": 6.29, + "learning_rate": 2.3104525173344693e-06, + "loss": 0.0544, + "step": 15013 + }, + { + "epoch": 6.29, + "learning_rate": 2.3093687475143236e-06, + "loss": 0.0848, + "step": 15014 + }, + { + "epoch": 6.29, + "learning_rate": 2.3082851987545474e-06, + "loss": 0.1051, + "step": 15015 + }, + { + "epoch": 6.29, + "learning_rate": 2.307201871086292e-06, + "loss": 0.0788, + "step": 15016 + }, + { + "epoch": 6.29, + "learning_rate": 2.306118764540688e-06, + "loss": 0.1185, + "step": 15017 + }, + { + "epoch": 6.29, + "learning_rate": 2.3050358791488734e-06, + "loss": 0.1118, + "step": 15018 + }, + { + "epoch": 6.29, + "learning_rate": 2.3039532149419775e-06, + "loss": 0.1037, + "step": 15019 + }, + { + "epoch": 6.29, + "learning_rate": 2.3028707719511134e-06, + "loss": 0.1002, + "step": 15020 + }, + { + "epoch": 6.29, + "learning_rate": 2.301788550207399e-06, + "loss": 0.0603, + "step": 15021 + }, + { + "epoch": 6.29, + "learning_rate": 2.3007065497419413e-06, + "loss": 0.1137, + "step": 15022 + }, + { + "epoch": 6.29, + "learning_rate": 2.299624770585842e-06, + "loss": 0.1029, + "step": 15023 + }, + { + "epoch": 6.29, + "learning_rate": 2.2985432127701945e-06, + "loss": 0.1103, + "step": 15024 + }, + { + "epoch": 6.29, + "learning_rate": 2.2974618763260927e-06, + "loss": 0.0929, + "step": 15025 + }, + { + "epoch": 6.29, + "learning_rate": 2.29638076128461e-06, + "loss": 0.0851, + "step": 15026 + }, + { + "epoch": 6.29, + "learning_rate": 2.2952998676768267e-06, + "loss": 0.0539, + "step": 15027 + }, + { + "epoch": 6.29, + "learning_rate": 2.294219195533811e-06, + "loss": 0.1102, + "step": 15028 + }, + { + "epoch": 6.29, + "learning_rate": 2.293138744886627e-06, + "loss": 0.0641, + "step": 15029 + }, + { + "epoch": 6.29, + "learning_rate": 2.292058515766331e-06, + "loss": 0.0838, + "step": 15030 + }, + { + "epoch": 6.29, + "learning_rate": 2.290978508203976e-06, + "loss": 0.0982, + "step": 15031 + }, + { + "epoch": 6.29, + "learning_rate": 2.2898987222306e-06, + "loss": 0.0812, + "step": 15032 + }, + { + "epoch": 6.29, + "learning_rate": 2.2888191578772436e-06, + "loss": 0.0796, + "step": 15033 + }, + { + "epoch": 6.29, + "learning_rate": 2.2877398151749375e-06, + "loss": 0.0761, + "step": 15034 + }, + { + "epoch": 6.29, + "learning_rate": 2.2866606941547063e-06, + "loss": 0.0802, + "step": 15035 + }, + { + "epoch": 6.3, + "learning_rate": 2.285581794847569e-06, + "loss": 0.0624, + "step": 15036 + }, + { + "epoch": 6.3, + "learning_rate": 2.284503117284538e-06, + "loss": 0.0575, + "step": 15037 + }, + { + "epoch": 6.3, + "learning_rate": 2.2834246614966192e-06, + "loss": 0.0717, + "step": 15038 + }, + { + "epoch": 6.3, + "learning_rate": 2.282346427514813e-06, + "loss": 0.0925, + "step": 15039 + }, + { + "epoch": 6.3, + "learning_rate": 2.281268415370107e-06, + "loss": 0.1024, + "step": 15040 + }, + { + "epoch": 6.3, + "learning_rate": 2.2801906250934904e-06, + "loss": 0.0787, + "step": 15041 + }, + { + "epoch": 6.3, + "learning_rate": 2.2791130567159434e-06, + "loss": 0.0818, + "step": 15042 + }, + { + "epoch": 6.3, + "learning_rate": 2.2780357102684415e-06, + "loss": 0.0743, + "step": 15043 + }, + { + "epoch": 6.3, + "learning_rate": 2.27695858578195e-06, + "loss": 0.0911, + "step": 15044 + }, + { + "epoch": 6.3, + "learning_rate": 2.2758816832874333e-06, + "loss": 0.0787, + "step": 15045 + }, + { + "epoch": 6.3, + "learning_rate": 2.274805002815841e-06, + "loss": 0.0562, + "step": 15046 + }, + { + "epoch": 6.3, + "learning_rate": 2.2737285443981226e-06, + "loss": 0.0959, + "step": 15047 + }, + { + "epoch": 6.3, + "learning_rate": 2.272652308065222e-06, + "loss": 0.0979, + "step": 15048 + }, + { + "epoch": 6.3, + "learning_rate": 2.2715762938480734e-06, + "loss": 0.0647, + "step": 15049 + }, + { + "epoch": 6.3, + "learning_rate": 2.2705005017776073e-06, + "loss": 0.0847, + "step": 15050 + }, + { + "epoch": 6.3, + "learning_rate": 2.269424931884744e-06, + "loss": 0.0836, + "step": 15051 + }, + { + "epoch": 6.3, + "learning_rate": 2.268349584200403e-06, + "loss": 0.0797, + "step": 15052 + }, + { + "epoch": 6.3, + "learning_rate": 2.2672744587554917e-06, + "loss": 0.0747, + "step": 15053 + }, + { + "epoch": 6.3, + "learning_rate": 2.2661995555809123e-06, + "loss": 0.0945, + "step": 15054 + }, + { + "epoch": 6.3, + "learning_rate": 2.265124874707565e-06, + "loss": 0.0886, + "step": 15055 + }, + { + "epoch": 6.3, + "learning_rate": 2.264050416166339e-06, + "loss": 0.1056, + "step": 15056 + }, + { + "epoch": 6.3, + "learning_rate": 2.262976179988119e-06, + "loss": 0.0627, + "step": 15057 + }, + { + "epoch": 6.3, + "learning_rate": 2.2619021662037855e-06, + "loss": 0.0828, + "step": 15058 + }, + { + "epoch": 6.3, + "learning_rate": 2.260828374844205e-06, + "loss": 0.0694, + "step": 15059 + }, + { + "epoch": 6.31, + "learning_rate": 2.2597548059402453e-06, + "loss": 0.0854, + "step": 15060 + }, + { + "epoch": 6.31, + "learning_rate": 2.258681459522768e-06, + "loss": 0.0818, + "step": 15061 + }, + { + "epoch": 6.31, + "learning_rate": 2.2576083356226197e-06, + "loss": 0.0873, + "step": 15062 + }, + { + "epoch": 6.31, + "learning_rate": 2.256535434270649e-06, + "loss": 0.0569, + "step": 15063 + }, + { + "epoch": 6.31, + "learning_rate": 2.255462755497696e-06, + "loss": 0.06, + "step": 15064 + }, + { + "epoch": 6.31, + "learning_rate": 2.2543902993345946e-06, + "loss": 0.0988, + "step": 15065 + }, + { + "epoch": 6.31, + "learning_rate": 2.2533180658121723e-06, + "loss": 0.0793, + "step": 15066 + }, + { + "epoch": 6.31, + "learning_rate": 2.252246054961246e-06, + "loss": 0.0841, + "step": 15067 + }, + { + "epoch": 6.31, + "learning_rate": 2.25117426681263e-06, + "loss": 0.0698, + "step": 15068 + }, + { + "epoch": 6.31, + "learning_rate": 2.2501027013971354e-06, + "loss": 0.1104, + "step": 15069 + }, + { + "epoch": 6.31, + "learning_rate": 2.2490313587455605e-06, + "loss": 0.0768, + "step": 15070 + }, + { + "epoch": 6.31, + "learning_rate": 2.2479602388887013e-06, + "loss": 0.0616, + "step": 15071 + }, + { + "epoch": 6.31, + "learning_rate": 2.2468893418573488e-06, + "loss": 0.0998, + "step": 15072 + }, + { + "epoch": 6.31, + "learning_rate": 2.24581866768228e-06, + "loss": 0.0846, + "step": 15073 + }, + { + "epoch": 6.31, + "learning_rate": 2.2447482163942725e-06, + "loss": 0.0625, + "step": 15074 + }, + { + "epoch": 6.31, + "learning_rate": 2.2436779880240943e-06, + "loss": 0.0746, + "step": 15075 + }, + { + "epoch": 6.31, + "learning_rate": 2.2426079826025118e-06, + "loss": 0.0756, + "step": 15076 + }, + { + "epoch": 6.31, + "learning_rate": 2.241538200160277e-06, + "loss": 0.0823, + "step": 15077 + }, + { + "epoch": 6.31, + "learning_rate": 2.2404686407281427e-06, + "loss": 0.0996, + "step": 15078 + }, + { + "epoch": 6.31, + "learning_rate": 2.2393993043368545e-06, + "loss": 0.0844, + "step": 15079 + }, + { + "epoch": 6.31, + "learning_rate": 2.238330191017144e-06, + "loss": 0.1074, + "step": 15080 + }, + { + "epoch": 6.31, + "learning_rate": 2.2372613007997435e-06, + "loss": 0.0777, + "step": 15081 + }, + { + "epoch": 6.31, + "learning_rate": 2.2361926337153796e-06, + "loss": 0.0737, + "step": 15082 + }, + { + "epoch": 6.31, + "learning_rate": 2.2351241897947684e-06, + "loss": 0.0865, + "step": 15083 + }, + { + "epoch": 6.32, + "learning_rate": 2.2340559690686247e-06, + "loss": 0.094, + "step": 15084 + }, + { + "epoch": 6.32, + "learning_rate": 2.232987971567648e-06, + "loss": 0.0923, + "step": 15085 + }, + { + "epoch": 6.32, + "learning_rate": 2.2319201973225434e-06, + "loss": 0.0568, + "step": 15086 + }, + { + "epoch": 6.32, + "learning_rate": 2.2308526463639955e-06, + "loss": 0.1045, + "step": 15087 + }, + { + "epoch": 6.32, + "learning_rate": 2.2297853187226946e-06, + "loss": 0.0798, + "step": 15088 + }, + { + "epoch": 6.32, + "learning_rate": 2.2287182144293196e-06, + "loss": 0.0816, + "step": 15089 + }, + { + "epoch": 6.32, + "learning_rate": 2.227651333514543e-06, + "loss": 0.0923, + "step": 15090 + }, + { + "epoch": 6.32, + "learning_rate": 2.2265846760090326e-06, + "loss": 0.0729, + "step": 15091 + }, + { + "epoch": 6.32, + "learning_rate": 2.225518241943446e-06, + "loss": 0.0955, + "step": 15092 + }, + { + "epoch": 6.32, + "learning_rate": 2.224452031348443e-06, + "loss": 0.0795, + "step": 15093 + }, + { + "epoch": 6.32, + "learning_rate": 2.223386044254663e-06, + "loss": 0.0862, + "step": 15094 + }, + { + "epoch": 6.32, + "learning_rate": 2.222320280692751e-06, + "loss": 0.0762, + "step": 15095 + }, + { + "epoch": 6.32, + "learning_rate": 2.2212547406933405e-06, + "loss": 0.0771, + "step": 15096 + }, + { + "epoch": 6.32, + "learning_rate": 2.220189424287059e-06, + "loss": 0.0832, + "step": 15097 + }, + { + "epoch": 6.32, + "learning_rate": 2.2191243315045296e-06, + "loss": 0.0759, + "step": 15098 + }, + { + "epoch": 6.32, + "learning_rate": 2.2180594623763686e-06, + "loss": 0.066, + "step": 15099 + }, + { + "epoch": 6.32, + "learning_rate": 2.2169948169331812e-06, + "loss": 0.0649, + "step": 15100 + }, + { + "epoch": 6.32, + "learning_rate": 2.21593039520557e-06, + "loss": 0.0836, + "step": 15101 + }, + { + "epoch": 6.32, + "learning_rate": 2.2148661972241324e-06, + "loss": 0.0737, + "step": 15102 + }, + { + "epoch": 6.32, + "learning_rate": 2.213802223019459e-06, + "loss": 0.0903, + "step": 15103 + }, + { + "epoch": 6.32, + "learning_rate": 2.2127384726221303e-06, + "loss": 0.0921, + "step": 15104 + }, + { + "epoch": 6.32, + "learning_rate": 2.211674946062726e-06, + "loss": 0.0945, + "step": 15105 + }, + { + "epoch": 6.32, + "learning_rate": 2.2106116433718115e-06, + "loss": 0.0758, + "step": 15106 + }, + { + "epoch": 6.32, + "learning_rate": 2.2095485645799564e-06, + "loss": 0.0526, + "step": 15107 + }, + { + "epoch": 6.33, + "learning_rate": 2.2084857097177125e-06, + "loss": 0.1023, + "step": 15108 + }, + { + "epoch": 6.33, + "learning_rate": 2.2074230788156325e-06, + "loss": 0.0932, + "step": 15109 + }, + { + "epoch": 6.33, + "learning_rate": 2.2063606719042605e-06, + "loss": 0.0918, + "step": 15110 + }, + { + "epoch": 6.33, + "learning_rate": 2.205298489014135e-06, + "loss": 0.0743, + "step": 15111 + }, + { + "epoch": 6.33, + "learning_rate": 2.2042365301757883e-06, + "loss": 0.0802, + "step": 15112 + }, + { + "epoch": 6.33, + "learning_rate": 2.203174795419747e-06, + "loss": 0.105, + "step": 15113 + }, + { + "epoch": 6.33, + "learning_rate": 2.2021132847765246e-06, + "loss": 0.1102, + "step": 15114 + }, + { + "epoch": 6.33, + "learning_rate": 2.2010519982766364e-06, + "loss": 0.0819, + "step": 15115 + }, + { + "epoch": 6.33, + "learning_rate": 2.1999909359505878e-06, + "loss": 0.0604, + "step": 15116 + }, + { + "epoch": 6.33, + "learning_rate": 2.1989300978288773e-06, + "loss": 0.0653, + "step": 15117 + }, + { + "epoch": 6.33, + "learning_rate": 2.197869483942e-06, + "loss": 0.0892, + "step": 15118 + }, + { + "epoch": 6.33, + "learning_rate": 2.1968090943204402e-06, + "loss": 0.0818, + "step": 15119 + }, + { + "epoch": 6.33, + "learning_rate": 2.195748928994681e-06, + "loss": 0.1011, + "step": 15120 + }, + { + "epoch": 6.33, + "learning_rate": 2.1946889879951903e-06, + "loss": 0.1059, + "step": 15121 + }, + { + "epoch": 6.33, + "learning_rate": 2.19362927135244e-06, + "loss": 0.0953, + "step": 15122 + }, + { + "epoch": 6.33, + "learning_rate": 2.192569779096888e-06, + "loss": 0.0665, + "step": 15123 + }, + { + "epoch": 6.33, + "learning_rate": 2.1915105112589894e-06, + "loss": 0.0824, + "step": 15124 + }, + { + "epoch": 6.33, + "learning_rate": 2.1904514678691924e-06, + "loss": 0.1006, + "step": 15125 + }, + { + "epoch": 6.33, + "learning_rate": 2.18939264895794e-06, + "loss": 0.0923, + "step": 15126 + }, + { + "epoch": 6.33, + "learning_rate": 2.188334054555663e-06, + "loss": 0.1008, + "step": 15127 + }, + { + "epoch": 6.33, + "learning_rate": 2.1872756846927913e-06, + "loss": 0.1174, + "step": 15128 + }, + { + "epoch": 6.33, + "learning_rate": 2.1862175393997497e-06, + "loss": 0.1017, + "step": 15129 + }, + { + "epoch": 6.33, + "learning_rate": 2.1851596187069478e-06, + "loss": 0.0738, + "step": 15130 + }, + { + "epoch": 6.33, + "learning_rate": 2.184101922644799e-06, + "loss": 0.086, + "step": 15131 + }, + { + "epoch": 6.34, + "learning_rate": 2.183044451243704e-06, + "loss": 0.0789, + "step": 15132 + }, + { + "epoch": 6.34, + "learning_rate": 2.181987204534061e-06, + "loss": 0.1091, + "step": 15133 + }, + { + "epoch": 6.34, + "learning_rate": 2.1809301825462593e-06, + "loss": 0.0802, + "step": 15134 + }, + { + "epoch": 6.34, + "learning_rate": 2.179873385310679e-06, + "loss": 0.1005, + "step": 15135 + }, + { + "epoch": 6.34, + "learning_rate": 2.178816812857698e-06, + "loss": 0.0425, + "step": 15136 + }, + { + "epoch": 6.34, + "learning_rate": 2.1777604652176885e-06, + "loss": 0.0622, + "step": 15137 + }, + { + "epoch": 6.34, + "learning_rate": 2.176704342421012e-06, + "loss": 0.0391, + "step": 15138 + }, + { + "epoch": 6.34, + "learning_rate": 2.1756484444980274e-06, + "loss": 0.0774, + "step": 15139 + }, + { + "epoch": 6.34, + "learning_rate": 2.1745927714790882e-06, + "loss": 0.0738, + "step": 15140 + }, + { + "epoch": 6.34, + "learning_rate": 2.1735373233945323e-06, + "loss": 0.079, + "step": 15141 + }, + { + "epoch": 6.34, + "learning_rate": 2.172482100274702e-06, + "loss": 0.0918, + "step": 15142 + }, + { + "epoch": 6.34, + "learning_rate": 2.171427102149927e-06, + "loss": 0.0805, + "step": 15143 + }, + { + "epoch": 6.34, + "learning_rate": 2.170372329050533e-06, + "loss": 0.057, + "step": 15144 + }, + { + "epoch": 6.34, + "learning_rate": 2.169317781006838e-06, + "loss": 0.0869, + "step": 15145 + }, + { + "epoch": 6.34, + "learning_rate": 2.168263458049158e-06, + "loss": 0.0805, + "step": 15146 + }, + { + "epoch": 6.34, + "learning_rate": 2.1672093602077914e-06, + "loss": 0.0829, + "step": 15147 + }, + { + "epoch": 6.34, + "learning_rate": 2.166155487513042e-06, + "loss": 0.0954, + "step": 15148 + }, + { + "epoch": 6.34, + "learning_rate": 2.1651018399952017e-06, + "loss": 0.0794, + "step": 15149 + }, + { + "epoch": 6.34, + "learning_rate": 2.1640484176845557e-06, + "loss": 0.1011, + "step": 15150 + }, + { + "epoch": 6.34, + "learning_rate": 2.1629952206113845e-06, + "loss": 0.0983, + "step": 15151 + }, + { + "epoch": 6.34, + "learning_rate": 2.161942248805965e-06, + "loss": 0.099, + "step": 15152 + }, + { + "epoch": 6.34, + "learning_rate": 2.1608895022985564e-06, + "loss": 0.0934, + "step": 15153 + }, + { + "epoch": 6.34, + "learning_rate": 2.159836981119425e-06, + "loss": 0.1135, + "step": 15154 + }, + { + "epoch": 6.34, + "learning_rate": 2.1587846852988203e-06, + "loss": 0.0888, + "step": 15155 + }, + { + "epoch": 6.35, + "learning_rate": 2.157732614866992e-06, + "loss": 0.0539, + "step": 15156 + }, + { + "epoch": 6.35, + "learning_rate": 2.1566807698541803e-06, + "loss": 0.0841, + "step": 15157 + }, + { + "epoch": 6.35, + "learning_rate": 2.15562915029062e-06, + "loss": 0.0645, + "step": 15158 + }, + { + "epoch": 6.35, + "learning_rate": 2.1545777562065383e-06, + "loss": 0.0846, + "step": 15159 + }, + { + "epoch": 6.35, + "learning_rate": 2.1535265876321574e-06, + "loss": 0.0775, + "step": 15160 + }, + { + "epoch": 6.35, + "learning_rate": 2.1524756445976947e-06, + "loss": 0.0767, + "step": 15161 + }, + { + "epoch": 6.35, + "learning_rate": 2.1514249271333532e-06, + "loss": 0.0852, + "step": 15162 + }, + { + "epoch": 6.35, + "learning_rate": 2.1503744352693377e-06, + "loss": 0.0796, + "step": 15163 + }, + { + "epoch": 6.35, + "learning_rate": 2.1493241690358437e-06, + "loss": 0.0639, + "step": 15164 + }, + { + "epoch": 6.35, + "learning_rate": 2.148274128463058e-06, + "loss": 0.0731, + "step": 15165 + }, + { + "epoch": 6.35, + "learning_rate": 2.1472243135811666e-06, + "loss": 0.0772, + "step": 15166 + }, + { + "epoch": 6.35, + "learning_rate": 2.146174724420347e-06, + "loss": 0.0946, + "step": 15167 + }, + { + "epoch": 6.35, + "learning_rate": 2.1451253610107616e-06, + "loss": 0.0783, + "step": 15168 + }, + { + "epoch": 6.35, + "learning_rate": 2.1440762233825774e-06, + "loss": 0.0823, + "step": 15169 + }, + { + "epoch": 6.35, + "learning_rate": 2.1430273115659516e-06, + "loss": 0.0751, + "step": 15170 + }, + { + "epoch": 6.35, + "learning_rate": 2.1419786255910336e-06, + "loss": 0.0774, + "step": 15171 + }, + { + "epoch": 6.35, + "learning_rate": 2.1409301654879665e-06, + "loss": 0.1268, + "step": 15172 + }, + { + "epoch": 6.35, + "learning_rate": 2.139881931286891e-06, + "loss": 0.1119, + "step": 15173 + }, + { + "epoch": 6.35, + "learning_rate": 2.138833923017931e-06, + "loss": 0.092, + "step": 15174 + }, + { + "epoch": 6.35, + "learning_rate": 2.1377861407112178e-06, + "loss": 0.09, + "step": 15175 + }, + { + "epoch": 6.35, + "learning_rate": 2.136738584396861e-06, + "loss": 0.0835, + "step": 15176 + }, + { + "epoch": 6.35, + "learning_rate": 2.135691254104977e-06, + "loss": 0.1081, + "step": 15177 + }, + { + "epoch": 6.35, + "learning_rate": 2.1346441498656688e-06, + "loss": 0.0438, + "step": 15178 + }, + { + "epoch": 6.36, + "learning_rate": 2.1335972717090358e-06, + "loss": 0.068, + "step": 15179 + }, + { + "epoch": 6.36, + "learning_rate": 2.132550619665168e-06, + "loss": 0.0942, + "step": 15180 + }, + { + "epoch": 6.36, + "learning_rate": 2.1315041937641534e-06, + "loss": 0.0728, + "step": 15181 + }, + { + "epoch": 6.36, + "learning_rate": 2.130457994036066e-06, + "loss": 0.1139, + "step": 15182 + }, + { + "epoch": 6.36, + "learning_rate": 2.12941202051098e-06, + "loss": 0.0853, + "step": 15183 + }, + { + "epoch": 6.36, + "learning_rate": 2.1283662732189624e-06, + "loss": 0.092, + "step": 15184 + }, + { + "epoch": 6.36, + "learning_rate": 2.12732075219007e-06, + "loss": 0.0503, + "step": 15185 + }, + { + "epoch": 6.36, + "learning_rate": 2.1262754574543566e-06, + "loss": 0.0582, + "step": 15186 + }, + { + "epoch": 6.36, + "learning_rate": 2.1252303890418714e-06, + "loss": 0.0838, + "step": 15187 + }, + { + "epoch": 6.36, + "learning_rate": 2.124185546982648e-06, + "loss": 0.0767, + "step": 15188 + }, + { + "epoch": 6.36, + "learning_rate": 2.123140931306722e-06, + "loss": 0.08, + "step": 15189 + }, + { + "epoch": 6.36, + "learning_rate": 2.1220965420441198e-06, + "loss": 0.0973, + "step": 15190 + }, + { + "epoch": 6.36, + "learning_rate": 2.1210523792248615e-06, + "loss": 0.0849, + "step": 15191 + }, + { + "epoch": 6.36, + "learning_rate": 2.1200084428789614e-06, + "loss": 0.0809, + "step": 15192 + }, + { + "epoch": 6.36, + "learning_rate": 2.118964733036426e-06, + "loss": 0.1049, + "step": 15193 + }, + { + "epoch": 6.36, + "learning_rate": 2.1179212497272582e-06, + "loss": 0.076, + "step": 15194 + }, + { + "epoch": 6.36, + "learning_rate": 2.1168779929814476e-06, + "loss": 0.0701, + "step": 15195 + }, + { + "epoch": 6.36, + "learning_rate": 2.1158349628289832e-06, + "loss": 0.0752, + "step": 15196 + }, + { + "epoch": 6.36, + "learning_rate": 2.1147921592998498e-06, + "loss": 0.0687, + "step": 15197 + }, + { + "epoch": 6.36, + "learning_rate": 2.1137495824240152e-06, + "loss": 0.0641, + "step": 15198 + }, + { + "epoch": 6.36, + "learning_rate": 2.1127072322314514e-06, + "loss": 0.0852, + "step": 15199 + }, + { + "epoch": 6.36, + "learning_rate": 2.1116651087521188e-06, + "loss": 0.0573, + "step": 15200 + }, + { + "epoch": 6.36, + "learning_rate": 2.110623212015972e-06, + "loss": 0.0811, + "step": 15201 + }, + { + "epoch": 6.36, + "learning_rate": 2.109581542052964e-06, + "loss": 0.0704, + "step": 15202 + }, + { + "epoch": 6.37, + "learning_rate": 2.108540098893029e-06, + "loss": 0.074, + "step": 15203 + }, + { + "epoch": 6.37, + "learning_rate": 2.107498882566107e-06, + "loss": 0.0836, + "step": 15204 + }, + { + "epoch": 6.37, + "learning_rate": 2.106457893102125e-06, + "loss": 0.0861, + "step": 15205 + }, + { + "epoch": 6.37, + "learning_rate": 2.1054171305310065e-06, + "loss": 0.0707, + "step": 15206 + }, + { + "epoch": 6.37, + "learning_rate": 2.1043765948826678e-06, + "loss": 0.0875, + "step": 15207 + }, + { + "epoch": 6.37, + "learning_rate": 2.1033362861870198e-06, + "loss": 0.083, + "step": 15208 + }, + { + "epoch": 6.37, + "learning_rate": 2.1022962044739602e-06, + "loss": 0.0617, + "step": 15209 + }, + { + "epoch": 6.37, + "learning_rate": 2.1012563497733873e-06, + "loss": 0.0751, + "step": 15210 + }, + { + "epoch": 6.37, + "learning_rate": 2.1002167221151915e-06, + "loss": 0.0753, + "step": 15211 + }, + { + "epoch": 6.37, + "learning_rate": 2.0991773215292566e-06, + "loss": 0.0736, + "step": 15212 + }, + { + "epoch": 6.37, + "learning_rate": 2.0981381480454575e-06, + "loss": 0.0571, + "step": 15213 + }, + { + "epoch": 6.37, + "learning_rate": 2.0970992016936685e-06, + "loss": 0.057, + "step": 15214 + }, + { + "epoch": 6.37, + "learning_rate": 2.096060482503748e-06, + "loss": 0.0662, + "step": 15215 + }, + { + "epoch": 6.37, + "learning_rate": 2.0950219905055548e-06, + "loss": 0.1205, + "step": 15216 + }, + { + "epoch": 6.37, + "learning_rate": 2.0939837257289395e-06, + "loss": 0.0759, + "step": 15217 + }, + { + "epoch": 6.37, + "learning_rate": 2.0929456882037458e-06, + "loss": 0.0748, + "step": 15218 + }, + { + "epoch": 6.37, + "learning_rate": 2.0919078779598123e-06, + "loss": 0.0819, + "step": 15219 + }, + { + "epoch": 6.37, + "learning_rate": 2.090870295026972e-06, + "loss": 0.0789, + "step": 15220 + }, + { + "epoch": 6.37, + "learning_rate": 2.0898329394350446e-06, + "loss": 0.0746, + "step": 15221 + }, + { + "epoch": 6.37, + "learning_rate": 2.0887958112138516e-06, + "loss": 0.0714, + "step": 15222 + }, + { + "epoch": 6.37, + "learning_rate": 2.0877589103932007e-06, + "loss": 0.0954, + "step": 15223 + }, + { + "epoch": 6.37, + "learning_rate": 2.0867222370028985e-06, + "loss": 0.1011, + "step": 15224 + }, + { + "epoch": 6.37, + "learning_rate": 2.0856857910727436e-06, + "loss": 0.0701, + "step": 15225 + }, + { + "epoch": 6.37, + "learning_rate": 2.0846495726325267e-06, + "loss": 0.1062, + "step": 15226 + }, + { + "epoch": 6.38, + "learning_rate": 2.0836135817120352e-06, + "loss": 0.0459, + "step": 15227 + }, + { + "epoch": 6.38, + "learning_rate": 2.0825778183410485e-06, + "loss": 0.0785, + "step": 15228 + }, + { + "epoch": 6.38, + "learning_rate": 2.0815422825493336e-06, + "loss": 0.1151, + "step": 15229 + }, + { + "epoch": 6.38, + "learning_rate": 2.0805069743666593e-06, + "loss": 0.0823, + "step": 15230 + }, + { + "epoch": 6.38, + "learning_rate": 2.0794718938227855e-06, + "loss": 0.0946, + "step": 15231 + }, + { + "epoch": 6.38, + "learning_rate": 2.078437040947462e-06, + "loss": 0.0926, + "step": 15232 + }, + { + "epoch": 6.38, + "learning_rate": 2.0774024157704364e-06, + "loss": 0.0794, + "step": 15233 + }, + { + "epoch": 6.38, + "learning_rate": 2.0763680183214485e-06, + "loss": 0.1051, + "step": 15234 + }, + { + "epoch": 6.38, + "learning_rate": 2.0753338486302334e-06, + "loss": 0.0725, + "step": 15235 + }, + { + "epoch": 6.38, + "learning_rate": 2.0742999067265125e-06, + "loss": 0.091, + "step": 15236 + }, + { + "epoch": 6.38, + "learning_rate": 2.0732661926400065e-06, + "loss": 0.0525, + "step": 15237 + }, + { + "epoch": 6.38, + "learning_rate": 2.07223270640043e-06, + "loss": 0.0917, + "step": 15238 + }, + { + "epoch": 6.38, + "learning_rate": 2.0711994480374895e-06, + "loss": 0.0621, + "step": 15239 + }, + { + "epoch": 6.38, + "learning_rate": 2.0701664175808843e-06, + "loss": 0.0761, + "step": 15240 + }, + { + "epoch": 6.38, + "learning_rate": 2.0691336150603124e-06, + "loss": 0.1114, + "step": 15241 + }, + { + "epoch": 6.38, + "learning_rate": 2.0681010405054535e-06, + "loss": 0.0802, + "step": 15242 + }, + { + "epoch": 6.38, + "learning_rate": 2.067068693945995e-06, + "loss": 0.1087, + "step": 15243 + }, + { + "epoch": 6.38, + "learning_rate": 2.0660365754116042e-06, + "loss": 0.0917, + "step": 15244 + }, + { + "epoch": 6.38, + "learning_rate": 2.0650046849319516e-06, + "loss": 0.0681, + "step": 15245 + }, + { + "epoch": 6.38, + "learning_rate": 2.063973022536698e-06, + "loss": 0.0731, + "step": 15246 + }, + { + "epoch": 6.38, + "learning_rate": 2.062941588255498e-06, + "loss": 0.081, + "step": 15247 + }, + { + "epoch": 6.38, + "learning_rate": 2.0619103821179975e-06, + "loss": 0.0566, + "step": 15248 + }, + { + "epoch": 6.38, + "learning_rate": 2.0608794041538426e-06, + "loss": 0.0903, + "step": 15249 + }, + { + "epoch": 6.38, + "learning_rate": 2.0598486543926623e-06, + "loss": 0.1197, + "step": 15250 + }, + { + "epoch": 6.39, + "learning_rate": 2.0588181328640856e-06, + "loss": 0.0987, + "step": 15251 + }, + { + "epoch": 6.39, + "learning_rate": 2.0577878395977347e-06, + "loss": 0.0553, + "step": 15252 + }, + { + "epoch": 6.39, + "learning_rate": 2.056757774623224e-06, + "loss": 0.077, + "step": 15253 + }, + { + "epoch": 6.39, + "learning_rate": 2.0557279379701632e-06, + "loss": 0.0711, + "step": 15254 + }, + { + "epoch": 6.39, + "learning_rate": 2.0546983296681554e-06, + "loss": 0.0822, + "step": 15255 + }, + { + "epoch": 6.39, + "learning_rate": 2.053668949746791e-06, + "loss": 0.1005, + "step": 15256 + }, + { + "epoch": 6.39, + "learning_rate": 2.0526397982356617e-06, + "loss": 0.0839, + "step": 15257 + }, + { + "epoch": 6.39, + "learning_rate": 2.0516108751643483e-06, + "loss": 0.1148, + "step": 15258 + }, + { + "epoch": 6.39, + "learning_rate": 2.050582180562427e-06, + "loss": 0.1279, + "step": 15259 + }, + { + "epoch": 6.39, + "learning_rate": 2.0495537144594678e-06, + "loss": 0.0718, + "step": 15260 + }, + { + "epoch": 6.39, + "learning_rate": 2.0485254768850317e-06, + "loss": 0.0816, + "step": 15261 + }, + { + "epoch": 6.39, + "learning_rate": 2.0474974678686764e-06, + "loss": 0.0784, + "step": 15262 + }, + { + "epoch": 6.39, + "learning_rate": 2.046469687439947e-06, + "loss": 0.075, + "step": 15263 + }, + { + "epoch": 6.39, + "learning_rate": 2.045442135628389e-06, + "loss": 0.0981, + "step": 15264 + }, + { + "epoch": 6.39, + "learning_rate": 2.044414812463541e-06, + "loss": 0.0727, + "step": 15265 + }, + { + "epoch": 6.39, + "learning_rate": 2.0433877179749273e-06, + "loss": 0.0525, + "step": 15266 + }, + { + "epoch": 6.39, + "learning_rate": 2.0423608521920723e-06, + "loss": 0.0993, + "step": 15267 + }, + { + "epoch": 6.39, + "learning_rate": 2.0413342151444928e-06, + "loss": 0.1002, + "step": 15268 + }, + { + "epoch": 6.39, + "learning_rate": 2.040307806861702e-06, + "loss": 0.065, + "step": 15269 + }, + { + "epoch": 6.39, + "learning_rate": 2.0392816273731973e-06, + "loss": 0.056, + "step": 15270 + }, + { + "epoch": 6.39, + "learning_rate": 2.0382556767084784e-06, + "loss": 0.0821, + "step": 15271 + }, + { + "epoch": 6.39, + "learning_rate": 2.0372299548970343e-06, + "loss": 0.0702, + "step": 15272 + }, + { + "epoch": 6.39, + "learning_rate": 2.0362044619683485e-06, + "loss": 0.1083, + "step": 15273 + }, + { + "epoch": 6.39, + "learning_rate": 2.0351791979518997e-06, + "loss": 0.0707, + "step": 15274 + }, + { + "epoch": 6.4, + "learning_rate": 2.0341541628771567e-06, + "loss": 0.0821, + "step": 15275 + }, + { + "epoch": 6.4, + "learning_rate": 2.0331293567735865e-06, + "loss": 0.055, + "step": 15276 + }, + { + "epoch": 6.4, + "learning_rate": 2.0321047796706393e-06, + "loss": 0.0862, + "step": 15277 + }, + { + "epoch": 6.4, + "learning_rate": 2.0310804315977717e-06, + "loss": 0.1026, + "step": 15278 + }, + { + "epoch": 6.4, + "learning_rate": 2.030056312584424e-06, + "loss": 0.0964, + "step": 15279 + }, + { + "epoch": 6.4, + "learning_rate": 2.0290324226600357e-06, + "loss": 0.104, + "step": 15280 + }, + { + "epoch": 6.4, + "learning_rate": 2.028008761854037e-06, + "loss": 0.101, + "step": 15281 + }, + { + "epoch": 6.4, + "learning_rate": 2.0269853301958554e-06, + "loss": 0.0903, + "step": 15282 + }, + { + "epoch": 6.4, + "learning_rate": 2.0259621277149023e-06, + "loss": 0.077, + "step": 15283 + }, + { + "epoch": 6.4, + "learning_rate": 2.0249391544405916e-06, + "loss": 0.1145, + "step": 15284 + }, + { + "epoch": 6.4, + "learning_rate": 2.0239164104023278e-06, + "loss": 0.0936, + "step": 15285 + }, + { + "epoch": 6.4, + "learning_rate": 2.0228938956295087e-06, + "loss": 0.0487, + "step": 15286 + }, + { + "epoch": 6.4, + "learning_rate": 2.0218716101515268e-06, + "loss": 0.1228, + "step": 15287 + }, + { + "epoch": 6.4, + "learning_rate": 2.020849553997768e-06, + "loss": 0.1073, + "step": 15288 + }, + { + "epoch": 6.4, + "learning_rate": 2.019827727197605e-06, + "loss": 0.0721, + "step": 15289 + }, + { + "epoch": 6.4, + "learning_rate": 2.018806129780416e-06, + "loss": 0.1021, + "step": 15290 + }, + { + "epoch": 6.4, + "learning_rate": 2.0177847617755595e-06, + "loss": 0.0533, + "step": 15291 + }, + { + "epoch": 6.4, + "learning_rate": 2.0167636232123967e-06, + "loss": 0.0427, + "step": 15292 + }, + { + "epoch": 6.4, + "learning_rate": 2.015742714120279e-06, + "loss": 0.082, + "step": 15293 + }, + { + "epoch": 6.4, + "learning_rate": 2.0147220345285524e-06, + "loss": 0.0755, + "step": 15294 + }, + { + "epoch": 6.4, + "learning_rate": 2.013701584466555e-06, + "loss": 0.1105, + "step": 15295 + }, + { + "epoch": 6.4, + "learning_rate": 2.012681363963621e-06, + "loss": 0.0404, + "step": 15296 + }, + { + "epoch": 6.4, + "learning_rate": 2.011661373049071e-06, + "loss": 0.1166, + "step": 15297 + }, + { + "epoch": 6.4, + "learning_rate": 2.0106416117522255e-06, + "loss": 0.0799, + "step": 15298 + }, + { + "epoch": 6.41, + "learning_rate": 2.009622080102398e-06, + "loss": 0.09, + "step": 15299 + }, + { + "epoch": 6.41, + "learning_rate": 2.008602778128892e-06, + "loss": 0.0517, + "step": 15300 + }, + { + "epoch": 6.41, + "learning_rate": 2.007583705861009e-06, + "loss": 0.1018, + "step": 15301 + }, + { + "epoch": 6.41, + "learning_rate": 2.00656486332804e-06, + "loss": 0.0705, + "step": 15302 + }, + { + "epoch": 6.41, + "learning_rate": 2.0055462505592725e-06, + "loss": 0.0839, + "step": 15303 + }, + { + "epoch": 6.41, + "learning_rate": 2.004527867583982e-06, + "loss": 0.0649, + "step": 15304 + }, + { + "epoch": 6.41, + "learning_rate": 2.003509714431442e-06, + "loss": 0.0991, + "step": 15305 + }, + { + "epoch": 6.41, + "learning_rate": 2.0024917911309207e-06, + "loss": 0.0896, + "step": 15306 + }, + { + "epoch": 6.41, + "learning_rate": 2.0014740977116743e-06, + "loss": 0.0785, + "step": 15307 + }, + { + "epoch": 6.41, + "learning_rate": 2.0004566342029586e-06, + "loss": 0.0783, + "step": 15308 + }, + { + "epoch": 6.41, + "learning_rate": 1.99943940063402e-06, + "loss": 0.0617, + "step": 15309 + }, + { + "epoch": 6.41, + "learning_rate": 1.9984223970340965e-06, + "loss": 0.1016, + "step": 15310 + }, + { + "epoch": 6.41, + "learning_rate": 1.997405623432419e-06, + "loss": 0.0764, + "step": 15311 + }, + { + "epoch": 6.41, + "learning_rate": 1.996389079858214e-06, + "loss": 0.0764, + "step": 15312 + }, + { + "epoch": 6.41, + "learning_rate": 1.9953727663407043e-06, + "loss": 0.0679, + "step": 15313 + }, + { + "epoch": 6.41, + "learning_rate": 1.9943566829091e-06, + "loss": 0.0698, + "step": 15314 + }, + { + "epoch": 6.41, + "learning_rate": 1.9933408295926095e-06, + "loss": 0.0677, + "step": 15315 + }, + { + "epoch": 6.41, + "learning_rate": 1.9923252064204313e-06, + "loss": 0.0525, + "step": 15316 + }, + { + "epoch": 6.41, + "learning_rate": 1.9913098134217612e-06, + "loss": 0.0881, + "step": 15317 + }, + { + "epoch": 6.41, + "learning_rate": 1.9902946506257825e-06, + "loss": 0.0696, + "step": 15318 + }, + { + "epoch": 6.41, + "learning_rate": 1.989279718061675e-06, + "loss": 0.095, + "step": 15319 + }, + { + "epoch": 6.41, + "learning_rate": 1.988265015758615e-06, + "loss": 0.0556, + "step": 15320 + }, + { + "epoch": 6.41, + "learning_rate": 1.9872505437457666e-06, + "loss": 0.1042, + "step": 15321 + }, + { + "epoch": 6.41, + "learning_rate": 1.986236302052291e-06, + "loss": 0.0944, + "step": 15322 + }, + { + "epoch": 6.42, + "learning_rate": 1.985222290707344e-06, + "loss": 0.1087, + "step": 15323 + }, + { + "epoch": 6.42, + "learning_rate": 1.9842085097400678e-06, + "loss": 0.0571, + "step": 15324 + }, + { + "epoch": 6.42, + "learning_rate": 1.9831949591796052e-06, + "loss": 0.1036, + "step": 15325 + }, + { + "epoch": 6.42, + "learning_rate": 1.9821816390550887e-06, + "loss": 0.0771, + "step": 15326 + }, + { + "epoch": 6.42, + "learning_rate": 1.981168549395647e-06, + "loss": 0.1008, + "step": 15327 + }, + { + "epoch": 6.42, + "learning_rate": 1.9801556902303986e-06, + "loss": 0.0746, + "step": 15328 + }, + { + "epoch": 6.42, + "learning_rate": 1.9791430615884622e-06, + "loss": 0.087, + "step": 15329 + }, + { + "epoch": 6.42, + "learning_rate": 1.9781306634989373e-06, + "loss": 0.0939, + "step": 15330 + }, + { + "epoch": 6.42, + "learning_rate": 1.9771184959909273e-06, + "loss": 0.083, + "step": 15331 + }, + { + "epoch": 6.42, + "learning_rate": 1.976106559093528e-06, + "loss": 0.0843, + "step": 15332 + }, + { + "epoch": 6.42, + "learning_rate": 1.9750948528358283e-06, + "loss": 0.0991, + "step": 15333 + }, + { + "epoch": 6.42, + "learning_rate": 1.9740833772469025e-06, + "loss": 0.0816, + "step": 15334 + }, + { + "epoch": 6.42, + "learning_rate": 1.973072132355828e-06, + "loss": 0.0678, + "step": 15335 + }, + { + "epoch": 6.42, + "learning_rate": 1.9720611181916716e-06, + "loss": 0.0686, + "step": 15336 + }, + { + "epoch": 6.42, + "learning_rate": 1.9710503347834975e-06, + "loss": 0.0766, + "step": 15337 + }, + { + "epoch": 6.42, + "learning_rate": 1.9700397821603546e-06, + "loss": 0.1118, + "step": 15338 + }, + { + "epoch": 6.42, + "learning_rate": 1.9690294603512915e-06, + "loss": 0.0656, + "step": 15339 + }, + { + "epoch": 6.42, + "learning_rate": 1.968019369385351e-06, + "loss": 0.0869, + "step": 15340 + }, + { + "epoch": 6.42, + "learning_rate": 1.9670095092915666e-06, + "loss": 0.0868, + "step": 15341 + }, + { + "epoch": 6.42, + "learning_rate": 1.9659998800989642e-06, + "loss": 0.0864, + "step": 15342 + }, + { + "epoch": 6.42, + "learning_rate": 1.9649904818365673e-06, + "loss": 0.096, + "step": 15343 + }, + { + "epoch": 6.42, + "learning_rate": 1.963981314533391e-06, + "loss": 0.0787, + "step": 15344 + }, + { + "epoch": 6.42, + "learning_rate": 1.962972378218437e-06, + "loss": 0.0652, + "step": 15345 + }, + { + "epoch": 6.42, + "learning_rate": 1.9619636729207116e-06, + "loss": 0.0911, + "step": 15346 + }, + { + "epoch": 6.43, + "learning_rate": 1.9609551986692067e-06, + "loss": 0.0688, + "step": 15347 + }, + { + "epoch": 6.43, + "learning_rate": 1.9599469554929116e-06, + "loss": 0.0916, + "step": 15348 + }, + { + "epoch": 6.43, + "learning_rate": 1.9589389434208063e-06, + "loss": 0.1026, + "step": 15349 + }, + { + "epoch": 6.43, + "learning_rate": 1.9579311624818674e-06, + "loss": 0.1018, + "step": 15350 + }, + { + "epoch": 6.43, + "learning_rate": 1.9569236127050596e-06, + "loss": 0.0661, + "step": 15351 + }, + { + "epoch": 6.43, + "learning_rate": 1.955916294119344e-06, + "loss": 0.0718, + "step": 15352 + }, + { + "epoch": 6.43, + "learning_rate": 1.954909206753678e-06, + "loss": 0.0781, + "step": 15353 + }, + { + "epoch": 6.43, + "learning_rate": 1.953902350637006e-06, + "loss": 0.0836, + "step": 15354 + }, + { + "epoch": 6.43, + "learning_rate": 1.9528957257982752e-06, + "loss": 0.0657, + "step": 15355 + }, + { + "epoch": 6.43, + "learning_rate": 1.9518893322664124e-06, + "loss": 0.0784, + "step": 15356 + }, + { + "epoch": 6.43, + "learning_rate": 1.9508831700703477e-06, + "loss": 0.0803, + "step": 15357 + }, + { + "epoch": 6.43, + "learning_rate": 1.9498772392390084e-06, + "loss": 0.0647, + "step": 15358 + }, + { + "epoch": 6.43, + "learning_rate": 1.9488715398013002e-06, + "loss": 0.0873, + "step": 15359 + }, + { + "epoch": 6.43, + "learning_rate": 1.9478660717861362e-06, + "loss": 0.0837, + "step": 15360 + }, + { + "epoch": 6.43, + "learning_rate": 1.946860835222416e-06, + "loss": 0.0908, + "step": 15361 + }, + { + "epoch": 6.43, + "learning_rate": 1.9458558301390353e-06, + "loss": 0.0487, + "step": 15362 + }, + { + "epoch": 6.43, + "learning_rate": 1.9448510565648813e-06, + "loss": 0.1196, + "step": 15363 + }, + { + "epoch": 6.43, + "learning_rate": 1.9438465145288377e-06, + "loss": 0.0769, + "step": 15364 + }, + { + "epoch": 6.43, + "learning_rate": 1.942842204059775e-06, + "loss": 0.0767, + "step": 15365 + }, + { + "epoch": 6.43, + "learning_rate": 1.9418381251865636e-06, + "loss": 0.0949, + "step": 15366 + }, + { + "epoch": 6.43, + "learning_rate": 1.940834277938065e-06, + "loss": 0.0808, + "step": 15367 + }, + { + "epoch": 6.43, + "learning_rate": 1.9398306623431327e-06, + "loss": 0.1119, + "step": 15368 + }, + { + "epoch": 6.43, + "learning_rate": 1.9388272784306163e-06, + "loss": 0.0912, + "step": 15369 + }, + { + "epoch": 6.44, + "learning_rate": 1.9378241262293583e-06, + "loss": 0.0776, + "step": 15370 + }, + { + "epoch": 6.44, + "learning_rate": 1.936821205768189e-06, + "loss": 0.062, + "step": 15371 + }, + { + "epoch": 6.44, + "learning_rate": 1.9358185170759394e-06, + "loss": 0.0768, + "step": 15372 + }, + { + "epoch": 6.44, + "learning_rate": 1.93481606018143e-06, + "loss": 0.095, + "step": 15373 + }, + { + "epoch": 6.44, + "learning_rate": 1.9338138351134773e-06, + "loss": 0.0929, + "step": 15374 + }, + { + "epoch": 6.44, + "learning_rate": 1.932811841900887e-06, + "loss": 0.0751, + "step": 15375 + }, + { + "epoch": 6.44, + "learning_rate": 1.9318100805724617e-06, + "loss": 0.0926, + "step": 15376 + }, + { + "epoch": 6.44, + "learning_rate": 1.9308085511569984e-06, + "loss": 0.073, + "step": 15377 + }, + { + "epoch": 6.44, + "learning_rate": 1.9298072536832834e-06, + "loss": 0.0421, + "step": 15378 + }, + { + "epoch": 6.44, + "learning_rate": 1.9288061881800945e-06, + "loss": 0.0576, + "step": 15379 + }, + { + "epoch": 6.44, + "learning_rate": 1.92780535467621e-06, + "loss": 0.106, + "step": 15380 + }, + { + "epoch": 6.44, + "learning_rate": 1.9268047532003962e-06, + "loss": 0.0936, + "step": 15381 + }, + { + "epoch": 6.44, + "learning_rate": 1.9258043837814164e-06, + "loss": 0.0728, + "step": 15382 + }, + { + "epoch": 6.44, + "learning_rate": 1.9248042464480255e-06, + "loss": 0.0798, + "step": 15383 + }, + { + "epoch": 6.44, + "learning_rate": 1.9238043412289696e-06, + "loss": 0.088, + "step": 15384 + }, + { + "epoch": 6.44, + "learning_rate": 1.9228046681529954e-06, + "loss": 0.1014, + "step": 15385 + }, + { + "epoch": 6.44, + "learning_rate": 1.92180522724883e-06, + "loss": 0.0671, + "step": 15386 + }, + { + "epoch": 6.44, + "learning_rate": 1.9208060185452037e-06, + "loss": 0.082, + "step": 15387 + }, + { + "epoch": 6.44, + "learning_rate": 1.9198070420708405e-06, + "loss": 0.0795, + "step": 15388 + }, + { + "epoch": 6.44, + "learning_rate": 1.918808297854453e-06, + "loss": 0.0988, + "step": 15389 + }, + { + "epoch": 6.44, + "learning_rate": 1.9178097859247502e-06, + "loss": 0.0596, + "step": 15390 + }, + { + "epoch": 6.44, + "learning_rate": 1.9168115063104355e-06, + "loss": 0.0862, + "step": 15391 + }, + { + "epoch": 6.44, + "learning_rate": 1.915813459040199e-06, + "loss": 0.089, + "step": 15392 + }, + { + "epoch": 6.44, + "learning_rate": 1.91481564414273e-06, + "loss": 0.113, + "step": 15393 + }, + { + "epoch": 6.45, + "learning_rate": 1.9138180616467105e-06, + "loss": 0.0811, + "step": 15394 + }, + { + "epoch": 6.45, + "learning_rate": 1.912820711580816e-06, + "loss": 0.0709, + "step": 15395 + }, + { + "epoch": 6.45, + "learning_rate": 1.911823593973714e-06, + "loss": 0.0679, + "step": 15396 + }, + { + "epoch": 6.45, + "learning_rate": 1.910826708854068e-06, + "loss": 0.0947, + "step": 15397 + }, + { + "epoch": 6.45, + "learning_rate": 1.9098300562505266e-06, + "loss": 0.1242, + "step": 15398 + }, + { + "epoch": 6.45, + "learning_rate": 1.9088336361917415e-06, + "loss": 0.0767, + "step": 15399 + }, + { + "epoch": 6.45, + "learning_rate": 1.9078374487063536e-06, + "loss": 0.0597, + "step": 15400 + }, + { + "epoch": 6.45, + "learning_rate": 1.9068414938229996e-06, + "loss": 0.0804, + "step": 15401 + }, + { + "epoch": 6.45, + "learning_rate": 1.9058457715703038e-06, + "loss": 0.0633, + "step": 15402 + }, + { + "epoch": 6.45, + "learning_rate": 1.9048502819768877e-06, + "loss": 0.0782, + "step": 15403 + }, + { + "epoch": 6.45, + "learning_rate": 1.903855025071366e-06, + "loss": 0.0845, + "step": 15404 + }, + { + "epoch": 6.45, + "learning_rate": 1.9028600008823517e-06, + "loss": 0.09, + "step": 15405 + }, + { + "epoch": 6.45, + "learning_rate": 1.901865209438437e-06, + "loss": 0.0917, + "step": 15406 + }, + { + "epoch": 6.45, + "learning_rate": 1.9008706507682217e-06, + "loss": 0.0871, + "step": 15407 + }, + { + "epoch": 6.45, + "learning_rate": 1.8998763249002928e-06, + "loss": 0.0657, + "step": 15408 + }, + { + "epoch": 6.45, + "learning_rate": 1.8988822318632305e-06, + "loss": 0.0579, + "step": 15409 + }, + { + "epoch": 6.45, + "learning_rate": 1.897888371685609e-06, + "loss": 0.0778, + "step": 15410 + }, + { + "epoch": 6.45, + "learning_rate": 1.8968947443959995e-06, + "loss": 0.0683, + "step": 15411 + }, + { + "epoch": 6.45, + "learning_rate": 1.8959013500229584e-06, + "loss": 0.0723, + "step": 15412 + }, + { + "epoch": 6.45, + "learning_rate": 1.8949081885950403e-06, + "loss": 0.0686, + "step": 15413 + }, + { + "epoch": 6.45, + "learning_rate": 1.8939152601407952e-06, + "loss": 0.0635, + "step": 15414 + }, + { + "epoch": 6.45, + "learning_rate": 1.8929225646887617e-06, + "loss": 0.0956, + "step": 15415 + }, + { + "epoch": 6.45, + "learning_rate": 1.8919301022674752e-06, + "loss": 0.1139, + "step": 15416 + }, + { + "epoch": 6.45, + "learning_rate": 1.890937872905463e-06, + "loss": 0.0914, + "step": 15417 + }, + { + "epoch": 6.46, + "learning_rate": 1.8899458766312484e-06, + "loss": 0.1168, + "step": 15418 + }, + { + "epoch": 6.46, + "learning_rate": 1.8889541134733401e-06, + "loss": 0.0744, + "step": 15419 + }, + { + "epoch": 6.46, + "learning_rate": 1.8879625834602488e-06, + "loss": 0.0642, + "step": 15420 + }, + { + "epoch": 6.46, + "learning_rate": 1.8869712866204748e-06, + "loss": 0.0822, + "step": 15421 + }, + { + "epoch": 6.46, + "learning_rate": 1.8859802229825109e-06, + "loss": 0.0684, + "step": 15422 + }, + { + "epoch": 6.46, + "learning_rate": 1.8849893925748474e-06, + "loss": 0.0957, + "step": 15423 + }, + { + "epoch": 6.46, + "learning_rate": 1.8839987954259609e-06, + "loss": 0.078, + "step": 15424 + }, + { + "epoch": 6.46, + "learning_rate": 1.883008431564327e-06, + "loss": 0.0764, + "step": 15425 + }, + { + "epoch": 6.46, + "learning_rate": 1.8820183010184157e-06, + "loss": 0.1057, + "step": 15426 + }, + { + "epoch": 6.46, + "learning_rate": 1.8810284038166815e-06, + "loss": 0.0593, + "step": 15427 + }, + { + "epoch": 6.46, + "learning_rate": 1.880038739987582e-06, + "loss": 0.0686, + "step": 15428 + }, + { + "epoch": 6.46, + "learning_rate": 1.8790493095595618e-06, + "loss": 0.0886, + "step": 15429 + }, + { + "epoch": 6.46, + "learning_rate": 1.8780601125610643e-06, + "loss": 0.0795, + "step": 15430 + }, + { + "epoch": 6.46, + "learning_rate": 1.8770711490205207e-06, + "loss": 0.0824, + "step": 15431 + }, + { + "epoch": 6.46, + "learning_rate": 1.8760824189663618e-06, + "loss": 0.0775, + "step": 15432 + }, + { + "epoch": 6.46, + "learning_rate": 1.8750939224270014e-06, + "loss": 0.0763, + "step": 15433 + }, + { + "epoch": 6.46, + "learning_rate": 1.8741056594308571e-06, + "loss": 0.0712, + "step": 15434 + }, + { + "epoch": 6.46, + "learning_rate": 1.8731176300063347e-06, + "loss": 0.0806, + "step": 15435 + }, + { + "epoch": 6.46, + "learning_rate": 1.8721298341818339e-06, + "loss": 0.0542, + "step": 15436 + }, + { + "epoch": 6.46, + "learning_rate": 1.8711422719857487e-06, + "loss": 0.07, + "step": 15437 + }, + { + "epoch": 6.46, + "learning_rate": 1.8701549434464684e-06, + "loss": 0.0852, + "step": 15438 + }, + { + "epoch": 6.46, + "learning_rate": 1.869167848592367e-06, + "loss": 0.1039, + "step": 15439 + }, + { + "epoch": 6.46, + "learning_rate": 1.8681809874518208e-06, + "loss": 0.0915, + "step": 15440 + }, + { + "epoch": 6.46, + "learning_rate": 1.8671943600531962e-06, + "loss": 0.0983, + "step": 15441 + }, + { + "epoch": 6.47, + "learning_rate": 1.8662079664248522e-06, + "loss": 0.047, + "step": 15442 + }, + { + "epoch": 6.47, + "learning_rate": 1.8652218065951432e-06, + "loss": 0.1026, + "step": 15443 + }, + { + "epoch": 6.47, + "learning_rate": 1.8642358805924144e-06, + "loss": 0.0711, + "step": 15444 + }, + { + "epoch": 6.47, + "learning_rate": 1.8632501884450082e-06, + "loss": 0.1029, + "step": 15445 + }, + { + "epoch": 6.47, + "learning_rate": 1.8622647301812546e-06, + "loss": 0.1079, + "step": 15446 + }, + { + "epoch": 6.47, + "learning_rate": 1.8612795058294775e-06, + "loss": 0.0836, + "step": 15447 + }, + { + "epoch": 6.47, + "learning_rate": 1.8602945154179997e-06, + "loss": 0.0754, + "step": 15448 + }, + { + "epoch": 6.47, + "learning_rate": 1.8593097589751318e-06, + "loss": 0.0884, + "step": 15449 + }, + { + "epoch": 6.47, + "learning_rate": 1.8583252365291814e-06, + "loss": 0.0766, + "step": 15450 + }, + { + "epoch": 6.47, + "learning_rate": 1.857340948108448e-06, + "loss": 0.0849, + "step": 15451 + }, + { + "epoch": 6.47, + "learning_rate": 1.8563568937412247e-06, + "loss": 0.0885, + "step": 15452 + }, + { + "epoch": 6.47, + "learning_rate": 1.855373073455794e-06, + "loss": 0.0845, + "step": 15453 + }, + { + "epoch": 6.47, + "learning_rate": 1.8543894872804368e-06, + "loss": 0.0834, + "step": 15454 + }, + { + "epoch": 6.47, + "learning_rate": 1.8534061352434263e-06, + "loss": 0.084, + "step": 15455 + }, + { + "epoch": 6.47, + "learning_rate": 1.8524230173730261e-06, + "loss": 0.0745, + "step": 15456 + }, + { + "epoch": 6.47, + "learning_rate": 1.8514401336974975e-06, + "loss": 0.1124, + "step": 15457 + }, + { + "epoch": 6.47, + "learning_rate": 1.8504574842450907e-06, + "loss": 0.0824, + "step": 15458 + }, + { + "epoch": 6.47, + "learning_rate": 1.8494750690440533e-06, + "loss": 0.0709, + "step": 15459 + }, + { + "epoch": 6.47, + "learning_rate": 1.8484928881226205e-06, + "loss": 0.0754, + "step": 15460 + }, + { + "epoch": 6.47, + "learning_rate": 1.8475109415090264e-06, + "loss": 0.0705, + "step": 15461 + }, + { + "epoch": 6.47, + "learning_rate": 1.8465292292314963e-06, + "loss": 0.0823, + "step": 15462 + }, + { + "epoch": 6.47, + "learning_rate": 1.8455477513182473e-06, + "loss": 0.089, + "step": 15463 + }, + { + "epoch": 6.47, + "learning_rate": 1.8445665077974928e-06, + "loss": 0.1006, + "step": 15464 + }, + { + "epoch": 6.47, + "learning_rate": 1.8435854986974399e-06, + "loss": 0.084, + "step": 15465 + }, + { + "epoch": 6.48, + "learning_rate": 1.8426047240462807e-06, + "loss": 0.1041, + "step": 15466 + }, + { + "epoch": 6.48, + "learning_rate": 1.8416241838722093e-06, + "loss": 0.0706, + "step": 15467 + }, + { + "epoch": 6.48, + "learning_rate": 1.840643878203412e-06, + "loss": 0.0696, + "step": 15468 + }, + { + "epoch": 6.48, + "learning_rate": 1.8396638070680685e-06, + "loss": 0.0779, + "step": 15469 + }, + { + "epoch": 6.48, + "learning_rate": 1.8386839704943449e-06, + "loss": 0.0984, + "step": 15470 + }, + { + "epoch": 6.48, + "learning_rate": 1.837704368510409e-06, + "loss": 0.0896, + "step": 15471 + }, + { + "epoch": 6.48, + "learning_rate": 1.836725001144417e-06, + "loss": 0.0775, + "step": 15472 + }, + { + "epoch": 6.48, + "learning_rate": 1.8357458684245234e-06, + "loss": 0.0882, + "step": 15473 + }, + { + "epoch": 6.48, + "learning_rate": 1.8347669703788683e-06, + "loss": 0.0753, + "step": 15474 + }, + { + "epoch": 6.48, + "learning_rate": 1.83378830703559e-06, + "loss": 0.0791, + "step": 15475 + }, + { + "epoch": 6.48, + "learning_rate": 1.832809878422821e-06, + "loss": 0.0847, + "step": 15476 + }, + { + "epoch": 6.48, + "learning_rate": 1.8318316845686834e-06, + "loss": 0.0809, + "step": 15477 + }, + { + "epoch": 6.48, + "learning_rate": 1.830853725501297e-06, + "loss": 0.0602, + "step": 15478 + }, + { + "epoch": 6.48, + "learning_rate": 1.8298760012487726e-06, + "loss": 0.0987, + "step": 15479 + }, + { + "epoch": 6.48, + "learning_rate": 1.828898511839211e-06, + "loss": 0.0938, + "step": 15480 + }, + { + "epoch": 6.48, + "learning_rate": 1.8279212573007099e-06, + "loss": 0.1222, + "step": 15481 + }, + { + "epoch": 6.48, + "learning_rate": 1.826944237661361e-06, + "loss": 0.0844, + "step": 15482 + }, + { + "epoch": 6.48, + "learning_rate": 1.8259674529492476e-06, + "loss": 0.0666, + "step": 15483 + }, + { + "epoch": 6.48, + "learning_rate": 1.824990903192445e-06, + "loss": 0.073, + "step": 15484 + }, + { + "epoch": 6.48, + "learning_rate": 1.8240145884190264e-06, + "loss": 0.0588, + "step": 15485 + }, + { + "epoch": 6.48, + "learning_rate": 1.8230385086570546e-06, + "loss": 0.0886, + "step": 15486 + }, + { + "epoch": 6.48, + "learning_rate": 1.8220626639345817e-06, + "loss": 0.0772, + "step": 15487 + }, + { + "epoch": 6.48, + "learning_rate": 1.8210870542796622e-06, + "loss": 0.1025, + "step": 15488 + }, + { + "epoch": 6.48, + "learning_rate": 1.8201116797203355e-06, + "loss": 0.0705, + "step": 15489 + }, + { + "epoch": 6.49, + "learning_rate": 1.8191365402846417e-06, + "loss": 0.0535, + "step": 15490 + }, + { + "epoch": 6.49, + "learning_rate": 1.8181616360006094e-06, + "loss": 0.0975, + "step": 15491 + }, + { + "epoch": 6.49, + "learning_rate": 1.8171869668962594e-06, + "loss": 0.0777, + "step": 15492 + }, + { + "epoch": 6.49, + "learning_rate": 1.8162125329996105e-06, + "loss": 0.0518, + "step": 15493 + }, + { + "epoch": 6.49, + "learning_rate": 1.815238334338667e-06, + "loss": 0.0665, + "step": 15494 + }, + { + "epoch": 6.49, + "learning_rate": 1.8142643709414355e-06, + "loss": 0.0883, + "step": 15495 + }, + { + "epoch": 6.49, + "learning_rate": 1.813290642835911e-06, + "loss": 0.0653, + "step": 15496 + }, + { + "epoch": 6.49, + "learning_rate": 1.8123171500500814e-06, + "loss": 0.0458, + "step": 15497 + }, + { + "epoch": 6.49, + "learning_rate": 1.8113438926119297e-06, + "loss": 0.0869, + "step": 15498 + }, + { + "epoch": 6.49, + "learning_rate": 1.8103708705494316e-06, + "loss": 0.092, + "step": 15499 + }, + { + "epoch": 6.49, + "learning_rate": 1.809398083890559e-06, + "loss": 0.0797, + "step": 15500 + }, + { + "epoch": 6.49, + "learning_rate": 1.808425532663266e-06, + "loss": 0.084, + "step": 15501 + }, + { + "epoch": 6.49, + "learning_rate": 1.8074532168955128e-06, + "loss": 0.0594, + "step": 15502 + }, + { + "epoch": 6.49, + "learning_rate": 1.8064811366152468e-06, + "loss": 0.0855, + "step": 15503 + }, + { + "epoch": 6.49, + "learning_rate": 1.8055092918504103e-06, + "loss": 0.0568, + "step": 15504 + }, + { + "epoch": 6.49, + "learning_rate": 1.8045376826289374e-06, + "loss": 0.0883, + "step": 15505 + }, + { + "epoch": 6.49, + "learning_rate": 1.8035663089787582e-06, + "loss": 0.1022, + "step": 15506 + }, + { + "epoch": 6.49, + "learning_rate": 1.80259517092779e-06, + "loss": 0.0616, + "step": 15507 + }, + { + "epoch": 6.49, + "learning_rate": 1.8016242685039486e-06, + "loss": 0.079, + "step": 15508 + }, + { + "epoch": 6.49, + "learning_rate": 1.8006536017351438e-06, + "loss": 0.084, + "step": 15509 + }, + { + "epoch": 6.49, + "learning_rate": 1.7996831706492745e-06, + "loss": 0.0961, + "step": 15510 + }, + { + "epoch": 6.49, + "learning_rate": 1.798712975274236e-06, + "loss": 0.0689, + "step": 15511 + }, + { + "epoch": 6.49, + "learning_rate": 1.7977430156379173e-06, + "loss": 0.0799, + "step": 15512 + }, + { + "epoch": 6.49, + "learning_rate": 1.7967732917681946e-06, + "loss": 0.0747, + "step": 15513 + }, + { + "epoch": 6.5, + "learning_rate": 1.795803803692947e-06, + "loss": 0.0868, + "step": 15514 + }, + { + "epoch": 6.5, + "learning_rate": 1.7948345514400356e-06, + "loss": 0.0968, + "step": 15515 + }, + { + "epoch": 6.5, + "learning_rate": 1.7938655350373246e-06, + "loss": 0.0793, + "step": 15516 + }, + { + "epoch": 6.5, + "learning_rate": 1.7928967545126674e-06, + "loss": 0.0978, + "step": 15517 + }, + { + "epoch": 6.5, + "learning_rate": 1.7919282098939095e-06, + "loss": 0.0584, + "step": 15518 + }, + { + "epoch": 6.5, + "learning_rate": 1.7909599012088908e-06, + "loss": 0.0498, + "step": 15519 + }, + { + "epoch": 6.5, + "learning_rate": 1.789991828485449e-06, + "loss": 0.0937, + "step": 15520 + }, + { + "epoch": 6.5, + "learning_rate": 1.789023991751403e-06, + "loss": 0.0835, + "step": 15521 + }, + { + "epoch": 6.5, + "learning_rate": 1.788056391034576e-06, + "loss": 0.0685, + "step": 15522 + }, + { + "epoch": 6.5, + "learning_rate": 1.7870890263627817e-06, + "loss": 0.0886, + "step": 15523 + }, + { + "epoch": 6.5, + "learning_rate": 1.7861218977638252e-06, + "loss": 0.0806, + "step": 15524 + }, + { + "epoch": 6.5, + "learning_rate": 1.7851550052655052e-06, + "loss": 0.0801, + "step": 15525 + }, + { + "epoch": 6.5, + "learning_rate": 1.7841883488956157e-06, + "loss": 0.0688, + "step": 15526 + }, + { + "epoch": 6.5, + "learning_rate": 1.7832219286819442e-06, + "loss": 0.1044, + "step": 15527 + }, + { + "epoch": 6.5, + "learning_rate": 1.7822557446522637e-06, + "loss": 0.0741, + "step": 15528 + }, + { + "epoch": 6.5, + "learning_rate": 1.781289796834349e-06, + "loss": 0.0708, + "step": 15529 + }, + { + "epoch": 6.5, + "learning_rate": 1.7803240852559667e-06, + "loss": 0.0867, + "step": 15530 + }, + { + "epoch": 6.5, + "learning_rate": 1.7793586099448745e-06, + "loss": 0.0651, + "step": 15531 + }, + { + "epoch": 6.5, + "learning_rate": 1.7783933709288249e-06, + "loss": 0.0988, + "step": 15532 + }, + { + "epoch": 6.5, + "learning_rate": 1.7774283682355642e-06, + "loss": 0.0839, + "step": 15533 + }, + { + "epoch": 6.5, + "learning_rate": 1.7764636018928249e-06, + "loss": 0.055, + "step": 15534 + }, + { + "epoch": 6.5, + "learning_rate": 1.7754990719283427e-06, + "loss": 0.083, + "step": 15535 + }, + { + "epoch": 6.5, + "learning_rate": 1.774534778369842e-06, + "loss": 0.0905, + "step": 15536 + }, + { + "epoch": 6.5, + "learning_rate": 1.7735707212450414e-06, + "loss": 0.0762, + "step": 15537 + }, + { + "epoch": 6.51, + "learning_rate": 1.7726069005816482e-06, + "loss": 0.0912, + "step": 15538 + }, + { + "epoch": 6.51, + "learning_rate": 1.7716433164073698e-06, + "loss": 0.0708, + "step": 15539 + }, + { + "epoch": 6.51, + "learning_rate": 1.770679968749901e-06, + "loss": 0.0814, + "step": 15540 + }, + { + "epoch": 6.51, + "learning_rate": 1.7697168576369373e-06, + "loss": 0.0873, + "step": 15541 + }, + { + "epoch": 6.51, + "learning_rate": 1.7687539830961564e-06, + "loss": 0.0748, + "step": 15542 + }, + { + "epoch": 6.51, + "learning_rate": 1.7677913451552386e-06, + "loss": 0.0954, + "step": 15543 + }, + { + "epoch": 6.51, + "learning_rate": 1.766828943841853e-06, + "loss": 0.0713, + "step": 15544 + }, + { + "epoch": 6.51, + "learning_rate": 1.7658667791836626e-06, + "loss": 0.0869, + "step": 15545 + }, + { + "epoch": 6.51, + "learning_rate": 1.7649048512083256e-06, + "loss": 0.0974, + "step": 15546 + }, + { + "epoch": 6.51, + "learning_rate": 1.763943159943493e-06, + "loss": 0.0788, + "step": 15547 + }, + { + "epoch": 6.51, + "learning_rate": 1.762981705416804e-06, + "loss": 0.0978, + "step": 15548 + }, + { + "epoch": 6.51, + "learning_rate": 1.7620204876558967e-06, + "loss": 0.0823, + "step": 15549 + }, + { + "epoch": 6.51, + "learning_rate": 1.7610595066883996e-06, + "loss": 0.0808, + "step": 15550 + }, + { + "epoch": 6.51, + "learning_rate": 1.7600987625419353e-06, + "loss": 0.0848, + "step": 15551 + }, + { + "epoch": 6.51, + "learning_rate": 1.7591382552441206e-06, + "loss": 0.1244, + "step": 15552 + }, + { + "epoch": 6.51, + "learning_rate": 1.7581779848225665e-06, + "loss": 0.0953, + "step": 15553 + }, + { + "epoch": 6.51, + "learning_rate": 1.7572179513048703e-06, + "loss": 0.0913, + "step": 15554 + }, + { + "epoch": 6.51, + "learning_rate": 1.7562581547186298e-06, + "loss": 0.1055, + "step": 15555 + }, + { + "epoch": 6.51, + "learning_rate": 1.7552985950914326e-06, + "loss": 0.0888, + "step": 15556 + }, + { + "epoch": 6.51, + "learning_rate": 1.7543392724508613e-06, + "loss": 0.0913, + "step": 15557 + }, + { + "epoch": 6.51, + "learning_rate": 1.7533801868244904e-06, + "loss": 0.0821, + "step": 15558 + }, + { + "epoch": 6.51, + "learning_rate": 1.7524213382398902e-06, + "loss": 0.1092, + "step": 15559 + }, + { + "epoch": 6.51, + "learning_rate": 1.7514627267246187e-06, + "loss": 0.0841, + "step": 15560 + }, + { + "epoch": 6.51, + "learning_rate": 1.7505043523062326e-06, + "loss": 0.0623, + "step": 15561 + }, + { + "epoch": 6.52, + "learning_rate": 1.7495462150122767e-06, + "loss": 0.0702, + "step": 15562 + }, + { + "epoch": 6.52, + "learning_rate": 1.7485883148702931e-06, + "loss": 0.0709, + "step": 15563 + }, + { + "epoch": 6.52, + "learning_rate": 1.7476306519078168e-06, + "loss": 0.1001, + "step": 15564 + }, + { + "epoch": 6.52, + "learning_rate": 1.7466732261523745e-06, + "loss": 0.079, + "step": 15565 + }, + { + "epoch": 6.52, + "learning_rate": 1.7457160376314852e-06, + "loss": 0.0705, + "step": 15566 + }, + { + "epoch": 6.52, + "learning_rate": 1.7447590863726649e-06, + "loss": 0.0983, + "step": 15567 + }, + { + "epoch": 6.52, + "learning_rate": 1.7438023724034215e-06, + "loss": 0.0806, + "step": 15568 + }, + { + "epoch": 6.52, + "learning_rate": 1.7428458957512496e-06, + "loss": 0.0615, + "step": 15569 + }, + { + "epoch": 6.52, + "learning_rate": 1.741889656443645e-06, + "loss": 0.0832, + "step": 15570 + }, + { + "epoch": 6.52, + "learning_rate": 1.7409336545080947e-06, + "loss": 0.0716, + "step": 15571 + }, + { + "epoch": 6.52, + "learning_rate": 1.7399778899720764e-06, + "loss": 0.1212, + "step": 15572 + }, + { + "epoch": 6.52, + "learning_rate": 1.7390223628630643e-06, + "loss": 0.1216, + "step": 15573 + }, + { + "epoch": 6.52, + "learning_rate": 1.738067073208527e-06, + "loss": 0.0805, + "step": 15574 + }, + { + "epoch": 6.52, + "learning_rate": 1.737112021035917e-06, + "loss": 0.0785, + "step": 15575 + }, + { + "epoch": 6.52, + "learning_rate": 1.7361572063726894e-06, + "loss": 0.1026, + "step": 15576 + }, + { + "epoch": 6.52, + "learning_rate": 1.735202629246291e-06, + "loss": 0.0766, + "step": 15577 + }, + { + "epoch": 6.52, + "learning_rate": 1.7342482896841572e-06, + "loss": 0.0695, + "step": 15578 + }, + { + "epoch": 6.52, + "learning_rate": 1.7332941877137233e-06, + "loss": 0.0673, + "step": 15579 + }, + { + "epoch": 6.52, + "learning_rate": 1.7323403233624136e-06, + "loss": 0.0663, + "step": 15580 + }, + { + "epoch": 6.52, + "learning_rate": 1.731386696657642e-06, + "loss": 0.0911, + "step": 15581 + }, + { + "epoch": 6.52, + "learning_rate": 1.7304333076268253e-06, + "loss": 0.0878, + "step": 15582 + }, + { + "epoch": 6.52, + "learning_rate": 1.7294801562973619e-06, + "loss": 0.0663, + "step": 15583 + }, + { + "epoch": 6.52, + "learning_rate": 1.7285272426966526e-06, + "loss": 0.0778, + "step": 15584 + }, + { + "epoch": 6.53, + "learning_rate": 1.7275745668520883e-06, + "loss": 0.0794, + "step": 15585 + }, + { + "epoch": 6.53, + "learning_rate": 1.7266221287910512e-06, + "loss": 0.0696, + "step": 15586 + }, + { + "epoch": 6.53, + "learning_rate": 1.7256699285409206e-06, + "loss": 0.0906, + "step": 15587 + }, + { + "epoch": 6.53, + "learning_rate": 1.724717966129067e-06, + "loss": 0.0507, + "step": 15588 + }, + { + "epoch": 6.53, + "learning_rate": 1.7237662415828493e-06, + "loss": 0.0617, + "step": 15589 + }, + { + "epoch": 6.53, + "learning_rate": 1.7228147549296271e-06, + "loss": 0.0886, + "step": 15590 + }, + { + "epoch": 6.53, + "learning_rate": 1.7218635061967505e-06, + "loss": 0.0585, + "step": 15591 + }, + { + "epoch": 6.53, + "learning_rate": 1.7209124954115608e-06, + "loss": 0.086, + "step": 15592 + }, + { + "epoch": 6.53, + "learning_rate": 1.7199617226013943e-06, + "loss": 0.0943, + "step": 15593 + }, + { + "epoch": 6.53, + "learning_rate": 1.7190111877935833e-06, + "loss": 0.0768, + "step": 15594 + }, + { + "epoch": 6.53, + "learning_rate": 1.718060891015445e-06, + "loss": 0.0918, + "step": 15595 + }, + { + "epoch": 6.53, + "learning_rate": 1.7171108322942975e-06, + "loss": 0.0945, + "step": 15596 + }, + { + "epoch": 6.53, + "learning_rate": 1.716161011657449e-06, + "loss": 0.0979, + "step": 15597 + }, + { + "epoch": 6.53, + "learning_rate": 1.7152114291322008e-06, + "loss": 0.1001, + "step": 15598 + }, + { + "epoch": 6.53, + "learning_rate": 1.714262084745848e-06, + "loss": 0.0842, + "step": 15599 + }, + { + "epoch": 6.53, + "learning_rate": 1.7133129785256797e-06, + "loss": 0.0716, + "step": 15600 + }, + { + "epoch": 6.53, + "learning_rate": 1.7123641104989786e-06, + "loss": 0.0983, + "step": 15601 + }, + { + "epoch": 6.53, + "learning_rate": 1.7114154806930138e-06, + "loss": 0.0482, + "step": 15602 + }, + { + "epoch": 6.53, + "learning_rate": 1.7104670891350572e-06, + "loss": 0.0885, + "step": 15603 + }, + { + "epoch": 6.53, + "learning_rate": 1.7095189358523668e-06, + "loss": 0.0932, + "step": 15604 + }, + { + "epoch": 6.53, + "learning_rate": 1.7085710208722006e-06, + "loss": 0.073, + "step": 15605 + }, + { + "epoch": 6.53, + "learning_rate": 1.7076233442218004e-06, + "loss": 0.0746, + "step": 15606 + }, + { + "epoch": 6.53, + "learning_rate": 1.7066759059284087e-06, + "loss": 0.0675, + "step": 15607 + }, + { + "epoch": 6.53, + "learning_rate": 1.7057287060192585e-06, + "loss": 0.0975, + "step": 15608 + }, + { + "epoch": 6.54, + "learning_rate": 1.7047817445215786e-06, + "loss": 0.0812, + "step": 15609 + }, + { + "epoch": 6.54, + "learning_rate": 1.7038350214625844e-06, + "loss": 0.1039, + "step": 15610 + }, + { + "epoch": 6.54, + "learning_rate": 1.7028885368694903e-06, + "loss": 0.0927, + "step": 15611 + }, + { + "epoch": 6.54, + "learning_rate": 1.7019422907695026e-06, + "loss": 0.0954, + "step": 15612 + }, + { + "epoch": 6.54, + "learning_rate": 1.7009962831898208e-06, + "loss": 0.0851, + "step": 15613 + }, + { + "epoch": 6.54, + "learning_rate": 1.700050514157635e-06, + "loss": 0.0976, + "step": 15614 + }, + { + "epoch": 6.54, + "learning_rate": 1.6991049837001362e-06, + "loss": 0.1027, + "step": 15615 + }, + { + "epoch": 6.54, + "learning_rate": 1.6981596918444953e-06, + "loss": 0.103, + "step": 15616 + }, + { + "epoch": 6.54, + "learning_rate": 1.6972146386178868e-06, + "loss": 0.0748, + "step": 15617 + }, + { + "epoch": 6.54, + "learning_rate": 1.6962698240474762e-06, + "loss": 0.092, + "step": 15618 + }, + { + "epoch": 6.54, + "learning_rate": 1.6953252481604198e-06, + "loss": 0.0901, + "step": 15619 + }, + { + "epoch": 6.54, + "learning_rate": 1.6943809109838716e-06, + "loss": 0.1037, + "step": 15620 + }, + { + "epoch": 6.54, + "learning_rate": 1.6934368125449752e-06, + "loss": 0.1248, + "step": 15621 + }, + { + "epoch": 6.54, + "learning_rate": 1.6924929528708633e-06, + "loss": 0.072, + "step": 15622 + }, + { + "epoch": 6.54, + "learning_rate": 1.6915493319886712e-06, + "loss": 0.0888, + "step": 15623 + }, + { + "epoch": 6.54, + "learning_rate": 1.690605949925519e-06, + "loss": 0.0678, + "step": 15624 + }, + { + "epoch": 6.54, + "learning_rate": 1.6896628067085264e-06, + "loss": 0.0959, + "step": 15625 + }, + { + "epoch": 6.54, + "learning_rate": 1.6887199023648016e-06, + "loss": 0.0664, + "step": 15626 + }, + { + "epoch": 6.54, + "learning_rate": 1.6877772369214507e-06, + "loss": 0.0693, + "step": 15627 + }, + { + "epoch": 6.54, + "learning_rate": 1.6868348104055642e-06, + "loss": 0.0777, + "step": 15628 + }, + { + "epoch": 6.54, + "learning_rate": 1.6858926228442362e-06, + "loss": 0.1035, + "step": 15629 + }, + { + "epoch": 6.54, + "learning_rate": 1.6849506742645449e-06, + "loss": 0.0704, + "step": 15630 + }, + { + "epoch": 6.54, + "learning_rate": 1.6840089646935686e-06, + "loss": 0.1098, + "step": 15631 + }, + { + "epoch": 6.54, + "learning_rate": 1.6830674941583736e-06, + "loss": 0.0978, + "step": 15632 + }, + { + "epoch": 6.55, + "learning_rate": 1.6821262626860246e-06, + "loss": 0.094, + "step": 15633 + }, + { + "epoch": 6.55, + "learning_rate": 1.6811852703035737e-06, + "loss": 0.0567, + "step": 15634 + }, + { + "epoch": 6.55, + "learning_rate": 1.6802445170380733e-06, + "loss": 0.098, + "step": 15635 + }, + { + "epoch": 6.55, + "learning_rate": 1.6793040029165596e-06, + "loss": 0.0864, + "step": 15636 + }, + { + "epoch": 6.55, + "learning_rate": 1.678363727966068e-06, + "loss": 0.0934, + "step": 15637 + }, + { + "epoch": 6.55, + "learning_rate": 1.6774236922136266e-06, + "loss": 0.0623, + "step": 15638 + }, + { + "epoch": 6.55, + "learning_rate": 1.676483895686255e-06, + "loss": 0.0902, + "step": 15639 + }, + { + "epoch": 6.55, + "learning_rate": 1.6755443384109694e-06, + "loss": 0.0686, + "step": 15640 + }, + { + "epoch": 6.55, + "learning_rate": 1.6746050204147734e-06, + "loss": 0.0762, + "step": 15641 + }, + { + "epoch": 6.55, + "learning_rate": 1.6736659417246714e-06, + "loss": 0.0639, + "step": 15642 + }, + { + "epoch": 6.55, + "learning_rate": 1.6727271023676506e-06, + "loss": 0.0844, + "step": 15643 + }, + { + "epoch": 6.55, + "learning_rate": 1.6717885023707004e-06, + "loss": 0.0582, + "step": 15644 + }, + { + "epoch": 6.55, + "learning_rate": 1.670850141760798e-06, + "loss": 0.1034, + "step": 15645 + }, + { + "epoch": 6.55, + "learning_rate": 1.6699120205649177e-06, + "loss": 0.0958, + "step": 15646 + }, + { + "epoch": 6.55, + "learning_rate": 1.6689741388100244e-06, + "loss": 0.086, + "step": 15647 + }, + { + "epoch": 6.55, + "learning_rate": 1.66803649652308e-06, + "loss": 0.08, + "step": 15648 + }, + { + "epoch": 6.55, + "learning_rate": 1.6670990937310293e-06, + "loss": 0.098, + "step": 15649 + }, + { + "epoch": 6.55, + "learning_rate": 1.6661619304608234e-06, + "loss": 0.0737, + "step": 15650 + }, + { + "epoch": 6.55, + "learning_rate": 1.665225006739396e-06, + "loss": 0.0956, + "step": 15651 + }, + { + "epoch": 6.55, + "learning_rate": 1.6642883225936779e-06, + "loss": 0.0917, + "step": 15652 + }, + { + "epoch": 6.55, + "learning_rate": 1.6633518780505964e-06, + "loss": 0.098, + "step": 15653 + }, + { + "epoch": 6.55, + "learning_rate": 1.6624156731370666e-06, + "loss": 0.0646, + "step": 15654 + }, + { + "epoch": 6.55, + "learning_rate": 1.6614797078799993e-06, + "loss": 0.0836, + "step": 15655 + }, + { + "epoch": 6.55, + "learning_rate": 1.6605439823063018e-06, + "loss": 0.0809, + "step": 15656 + }, + { + "epoch": 6.56, + "learning_rate": 1.659608496442865e-06, + "loss": 0.0779, + "step": 15657 + }, + { + "epoch": 6.56, + "learning_rate": 1.6586732503165793e-06, + "loss": 0.0667, + "step": 15658 + }, + { + "epoch": 6.56, + "learning_rate": 1.65773824395433e-06, + "loss": 0.1243, + "step": 15659 + }, + { + "epoch": 6.56, + "learning_rate": 1.6568034773829923e-06, + "loss": 0.0629, + "step": 15660 + }, + { + "epoch": 6.56, + "learning_rate": 1.6558689506294345e-06, + "loss": 0.1078, + "step": 15661 + }, + { + "epoch": 6.56, + "learning_rate": 1.654934663720522e-06, + "loss": 0.0776, + "step": 15662 + }, + { + "epoch": 6.56, + "learning_rate": 1.6540006166831046e-06, + "loss": 0.104, + "step": 15663 + }, + { + "epoch": 6.56, + "learning_rate": 1.6530668095440339e-06, + "loss": 0.0857, + "step": 15664 + }, + { + "epoch": 6.56, + "learning_rate": 1.6521332423301507e-06, + "loss": 0.091, + "step": 15665 + }, + { + "epoch": 6.56, + "learning_rate": 1.6511999150682888e-06, + "loss": 0.0714, + "step": 15666 + }, + { + "epoch": 6.56, + "learning_rate": 1.6502668277852774e-06, + "loss": 0.0698, + "step": 15667 + }, + { + "epoch": 6.56, + "learning_rate": 1.6493339805079367e-06, + "loss": 0.0865, + "step": 15668 + }, + { + "epoch": 6.56, + "learning_rate": 1.6484013732630833e-06, + "loss": 0.0685, + "step": 15669 + }, + { + "epoch": 6.56, + "learning_rate": 1.6474690060775178e-06, + "loss": 0.0712, + "step": 15670 + }, + { + "epoch": 6.56, + "learning_rate": 1.6465368789780446e-06, + "loss": 0.049, + "step": 15671 + }, + { + "epoch": 6.56, + "learning_rate": 1.6456049919914563e-06, + "loss": 0.0795, + "step": 15672 + }, + { + "epoch": 6.56, + "learning_rate": 1.6446733451445408e-06, + "loss": 0.0751, + "step": 15673 + }, + { + "epoch": 6.56, + "learning_rate": 1.643741938464073e-06, + "loss": 0.0589, + "step": 15674 + }, + { + "epoch": 6.56, + "learning_rate": 1.6428107719768293e-06, + "loss": 0.0733, + "step": 15675 + }, + { + "epoch": 6.56, + "learning_rate": 1.6418798457095753e-06, + "loss": 0.0658, + "step": 15676 + }, + { + "epoch": 6.56, + "learning_rate": 1.640949159689066e-06, + "loss": 0.0452, + "step": 15677 + }, + { + "epoch": 6.56, + "learning_rate": 1.6400187139420554e-06, + "loss": 0.0795, + "step": 15678 + }, + { + "epoch": 6.56, + "learning_rate": 1.6390885084952878e-06, + "loss": 0.1011, + "step": 15679 + }, + { + "epoch": 6.56, + "learning_rate": 1.638158543375502e-06, + "loss": 0.0588, + "step": 15680 + }, + { + "epoch": 6.57, + "learning_rate": 1.6372288186094287e-06, + "loss": 0.0611, + "step": 15681 + }, + { + "epoch": 6.57, + "learning_rate": 1.6362993342237931e-06, + "loss": 0.0871, + "step": 15682 + }, + { + "epoch": 6.57, + "learning_rate": 1.635370090245313e-06, + "loss": 0.1033, + "step": 15683 + }, + { + "epoch": 6.57, + "learning_rate": 1.6344410867006944e-06, + "loss": 0.0919, + "step": 15684 + }, + { + "epoch": 6.57, + "learning_rate": 1.6335123236166428e-06, + "loss": 0.0885, + "step": 15685 + }, + { + "epoch": 6.57, + "learning_rate": 1.6325838010198558e-06, + "loss": 0.0697, + "step": 15686 + }, + { + "epoch": 6.57, + "learning_rate": 1.631655518937022e-06, + "loss": 0.0664, + "step": 15687 + }, + { + "epoch": 6.57, + "learning_rate": 1.6307274773948244e-06, + "loss": 0.0457, + "step": 15688 + }, + { + "epoch": 6.57, + "learning_rate": 1.629799676419942e-06, + "loss": 0.0925, + "step": 15689 + }, + { + "epoch": 6.57, + "learning_rate": 1.6288721160390375e-06, + "loss": 0.0883, + "step": 15690 + }, + { + "epoch": 6.57, + "learning_rate": 1.6279447962787753e-06, + "loss": 0.0826, + "step": 15691 + }, + { + "epoch": 6.57, + "learning_rate": 1.6270177171658108e-06, + "loss": 0.0764, + "step": 15692 + }, + { + "epoch": 6.57, + "learning_rate": 1.6260908787267915e-06, + "loss": 0.0912, + "step": 15693 + }, + { + "epoch": 6.57, + "learning_rate": 1.6251642809883595e-06, + "loss": 0.0812, + "step": 15694 + }, + { + "epoch": 6.57, + "learning_rate": 1.624237923977151e-06, + "loss": 0.0999, + "step": 15695 + }, + { + "epoch": 6.57, + "learning_rate": 1.6233118077197884e-06, + "loss": 0.0605, + "step": 15696 + }, + { + "epoch": 6.57, + "learning_rate": 1.622385932242897e-06, + "loss": 0.0817, + "step": 15697 + }, + { + "epoch": 6.57, + "learning_rate": 1.6214602975730853e-06, + "loss": 0.0605, + "step": 15698 + }, + { + "epoch": 6.57, + "learning_rate": 1.6205349037369621e-06, + "loss": 0.101, + "step": 15699 + }, + { + "epoch": 6.57, + "learning_rate": 1.6196097507611274e-06, + "loss": 0.0774, + "step": 15700 + }, + { + "epoch": 6.57, + "learning_rate": 1.6186848386721743e-06, + "loss": 0.1025, + "step": 15701 + }, + { + "epoch": 6.57, + "learning_rate": 1.617760167496688e-06, + "loss": 0.0637, + "step": 15702 + }, + { + "epoch": 6.57, + "learning_rate": 1.6168357372612498e-06, + "loss": 0.046, + "step": 15703 + }, + { + "epoch": 6.57, + "learning_rate": 1.6159115479924259e-06, + "loss": 0.0925, + "step": 15704 + }, + { + "epoch": 6.58, + "learning_rate": 1.6149875997167851e-06, + "loss": 0.1007, + "step": 15705 + }, + { + "epoch": 6.58, + "learning_rate": 1.6140638924608853e-06, + "loss": 0.0874, + "step": 15706 + }, + { + "epoch": 6.58, + "learning_rate": 1.6131404262512774e-06, + "loss": 0.0554, + "step": 15707 + }, + { + "epoch": 6.58, + "learning_rate": 1.6122172011145053e-06, + "loss": 0.0789, + "step": 15708 + }, + { + "epoch": 6.58, + "learning_rate": 1.6112942170771073e-06, + "loss": 0.0864, + "step": 15709 + }, + { + "epoch": 6.58, + "learning_rate": 1.6103714741656152e-06, + "loss": 0.0571, + "step": 15710 + }, + { + "epoch": 6.58, + "learning_rate": 1.6094489724065476e-06, + "loss": 0.1119, + "step": 15711 + }, + { + "epoch": 6.58, + "learning_rate": 1.6085267118264235e-06, + "loss": 0.0649, + "step": 15712 + }, + { + "epoch": 6.58, + "learning_rate": 1.6076046924517518e-06, + "loss": 0.0511, + "step": 15713 + }, + { + "epoch": 6.58, + "learning_rate": 1.606682914309038e-06, + "loss": 0.0763, + "step": 15714 + }, + { + "epoch": 6.58, + "learning_rate": 1.6057613774247738e-06, + "loss": 0.0917, + "step": 15715 + }, + { + "epoch": 6.58, + "learning_rate": 1.6048400818254528e-06, + "loss": 0.0586, + "step": 15716 + }, + { + "epoch": 6.58, + "learning_rate": 1.6039190275375526e-06, + "loss": 0.0785, + "step": 15717 + }, + { + "epoch": 6.58, + "learning_rate": 1.60299821458755e-06, + "loss": 0.0894, + "step": 15718 + }, + { + "epoch": 6.58, + "learning_rate": 1.6020776430019113e-06, + "loss": 0.0793, + "step": 15719 + }, + { + "epoch": 6.58, + "learning_rate": 1.6011573128070978e-06, + "loss": 0.0858, + "step": 15720 + }, + { + "epoch": 6.58, + "learning_rate": 1.600237224029565e-06, + "loss": 0.0924, + "step": 15721 + }, + { + "epoch": 6.58, + "learning_rate": 1.5993173766957592e-06, + "loss": 0.0849, + "step": 15722 + }, + { + "epoch": 6.58, + "learning_rate": 1.5983977708321207e-06, + "loss": 0.0839, + "step": 15723 + }, + { + "epoch": 6.58, + "learning_rate": 1.597478406465085e-06, + "loss": 0.0739, + "step": 15724 + }, + { + "epoch": 6.58, + "learning_rate": 1.596559283621074e-06, + "loss": 0.0767, + "step": 15725 + }, + { + "epoch": 6.58, + "learning_rate": 1.5956404023265092e-06, + "loss": 0.0593, + "step": 15726 + }, + { + "epoch": 6.58, + "learning_rate": 1.5947217626078039e-06, + "loss": 0.0842, + "step": 15727 + }, + { + "epoch": 6.58, + "learning_rate": 1.593803364491362e-06, + "loss": 0.0605, + "step": 15728 + }, + { + "epoch": 6.59, + "learning_rate": 1.592885208003583e-06, + "loss": 0.0713, + "step": 15729 + }, + { + "epoch": 6.59, + "learning_rate": 1.5919672931708607e-06, + "loss": 0.1021, + "step": 15730 + }, + { + "epoch": 6.59, + "learning_rate": 1.5910496200195757e-06, + "loss": 0.0756, + "step": 15731 + }, + { + "epoch": 6.59, + "learning_rate": 1.5901321885761068e-06, + "loss": 0.0888, + "step": 15732 + }, + { + "epoch": 6.59, + "learning_rate": 1.5892149988668249e-06, + "loss": 0.0864, + "step": 15733 + }, + { + "epoch": 6.59, + "learning_rate": 1.5882980509180956e-06, + "loss": 0.083, + "step": 15734 + }, + { + "epoch": 6.59, + "learning_rate": 1.5873813447562735e-06, + "loss": 0.0715, + "step": 15735 + }, + { + "epoch": 6.59, + "learning_rate": 1.5864648804077122e-06, + "loss": 0.053, + "step": 15736 + }, + { + "epoch": 6.59, + "learning_rate": 1.5855486578987489e-06, + "loss": 0.0777, + "step": 15737 + }, + { + "epoch": 6.59, + "learning_rate": 1.584632677255723e-06, + "loss": 0.0909, + "step": 15738 + }, + { + "epoch": 6.59, + "learning_rate": 1.5837169385049623e-06, + "loss": 0.0655, + "step": 15739 + }, + { + "epoch": 6.59, + "learning_rate": 1.582801441672791e-06, + "loss": 0.0622, + "step": 15740 + }, + { + "epoch": 6.59, + "learning_rate": 1.5818861867855239e-06, + "loss": 0.0597, + "step": 15741 + }, + { + "epoch": 6.59, + "learning_rate": 1.5809711738694665e-06, + "loss": 0.0651, + "step": 15742 + }, + { + "epoch": 6.59, + "learning_rate": 1.5800564029509225e-06, + "loss": 0.0885, + "step": 15743 + }, + { + "epoch": 6.59, + "learning_rate": 1.579141874056187e-06, + "loss": 0.0718, + "step": 15744 + }, + { + "epoch": 6.59, + "learning_rate": 1.578227587211545e-06, + "loss": 0.0479, + "step": 15745 + }, + { + "epoch": 6.59, + "learning_rate": 1.5773135424432762e-06, + "loss": 0.0664, + "step": 15746 + }, + { + "epoch": 6.59, + "learning_rate": 1.5763997397776553e-06, + "loss": 0.0864, + "step": 15747 + }, + { + "epoch": 6.59, + "learning_rate": 1.5754861792409493e-06, + "loss": 0.081, + "step": 15748 + }, + { + "epoch": 6.59, + "learning_rate": 1.574572860859417e-06, + "loss": 0.0448, + "step": 15749 + }, + { + "epoch": 6.59, + "learning_rate": 1.5736597846593115e-06, + "loss": 0.0616, + "step": 15750 + }, + { + "epoch": 6.59, + "learning_rate": 1.5727469506668814e-06, + "loss": 0.0791, + "step": 15751 + }, + { + "epoch": 6.59, + "learning_rate": 1.5718343589083584e-06, + "loss": 0.0777, + "step": 15752 + }, + { + "epoch": 6.6, + "learning_rate": 1.5709220094099775e-06, + "loss": 0.087, + "step": 15753 + }, + { + "epoch": 6.6, + "learning_rate": 1.5700099021979642e-06, + "loss": 0.0565, + "step": 15754 + }, + { + "epoch": 6.6, + "learning_rate": 1.5690980372985354e-06, + "loss": 0.0675, + "step": 15755 + }, + { + "epoch": 6.6, + "learning_rate": 1.5681864147379022e-06, + "loss": 0.0759, + "step": 15756 + }, + { + "epoch": 6.6, + "learning_rate": 1.5672750345422716e-06, + "loss": 0.0632, + "step": 15757 + }, + { + "epoch": 6.6, + "learning_rate": 1.5663638967378337e-06, + "loss": 0.0824, + "step": 15758 + }, + { + "epoch": 6.6, + "learning_rate": 1.565453001350782e-06, + "loss": 0.0609, + "step": 15759 + }, + { + "epoch": 6.6, + "learning_rate": 1.5645423484072987e-06, + "loss": 0.0871, + "step": 15760 + }, + { + "epoch": 6.6, + "learning_rate": 1.5636319379335596e-06, + "loss": 0.103, + "step": 15761 + }, + { + "epoch": 6.6, + "learning_rate": 1.5627217699557352e-06, + "loss": 0.0919, + "step": 15762 + }, + { + "epoch": 6.6, + "learning_rate": 1.5618118444999885e-06, + "loss": 0.0721, + "step": 15763 + }, + { + "epoch": 6.6, + "learning_rate": 1.5609021615924702e-06, + "loss": 0.0606, + "step": 15764 + }, + { + "epoch": 6.6, + "learning_rate": 1.5599927212593325e-06, + "loss": 0.0896, + "step": 15765 + }, + { + "epoch": 6.6, + "learning_rate": 1.559083523526712e-06, + "loss": 0.1032, + "step": 15766 + }, + { + "epoch": 6.6, + "learning_rate": 1.558174568420746e-06, + "loss": 0.1055, + "step": 15767 + }, + { + "epoch": 6.6, + "learning_rate": 1.5572658559675602e-06, + "loss": 0.0868, + "step": 15768 + }, + { + "epoch": 6.6, + "learning_rate": 1.5563573861932756e-06, + "loss": 0.075, + "step": 15769 + }, + { + "epoch": 6.6, + "learning_rate": 1.555449159124005e-06, + "loss": 0.1048, + "step": 15770 + }, + { + "epoch": 6.6, + "learning_rate": 1.5545411747858574e-06, + "loss": 0.0805, + "step": 15771 + }, + { + "epoch": 6.6, + "learning_rate": 1.5536334332049274e-06, + "loss": 0.0666, + "step": 15772 + }, + { + "epoch": 6.6, + "learning_rate": 1.5527259344073087e-06, + "loss": 0.0646, + "step": 15773 + }, + { + "epoch": 6.6, + "learning_rate": 1.551818678419087e-06, + "loss": 0.1058, + "step": 15774 + }, + { + "epoch": 6.6, + "learning_rate": 1.5509116652663403e-06, + "loss": 0.0872, + "step": 15775 + }, + { + "epoch": 6.6, + "learning_rate": 1.5500048949751411e-06, + "loss": 0.0849, + "step": 15776 + }, + { + "epoch": 6.61, + "learning_rate": 1.5490983675715544e-06, + "loss": 0.0757, + "step": 15777 + }, + { + "epoch": 6.61, + "learning_rate": 1.5481920830816343e-06, + "loss": 0.107, + "step": 15778 + }, + { + "epoch": 6.61, + "learning_rate": 1.5472860415314317e-06, + "loss": 0.0731, + "step": 15779 + }, + { + "epoch": 6.61, + "learning_rate": 1.546380242946992e-06, + "loss": 0.1056, + "step": 15780 + }, + { + "epoch": 6.61, + "learning_rate": 1.5454746873543503e-06, + "loss": 0.078, + "step": 15781 + }, + { + "epoch": 6.61, + "learning_rate": 1.5445693747795366e-06, + "loss": 0.0975, + "step": 15782 + }, + { + "epoch": 6.61, + "learning_rate": 1.5436643052485723e-06, + "loss": 0.0566, + "step": 15783 + }, + { + "epoch": 6.61, + "learning_rate": 1.5427594787874766e-06, + "loss": 0.0795, + "step": 15784 + }, + { + "epoch": 6.61, + "learning_rate": 1.5418548954222512e-06, + "loss": 0.1206, + "step": 15785 + }, + { + "epoch": 6.61, + "learning_rate": 1.540950555178905e-06, + "loss": 0.0608, + "step": 15786 + }, + { + "epoch": 6.61, + "learning_rate": 1.5400464580834262e-06, + "loss": 0.0807, + "step": 15787 + }, + { + "epoch": 6.61, + "learning_rate": 1.539142604161804e-06, + "loss": 0.066, + "step": 15788 + }, + { + "epoch": 6.61, + "learning_rate": 1.5382389934400199e-06, + "loss": 0.0738, + "step": 15789 + }, + { + "epoch": 6.61, + "learning_rate": 1.5373356259440475e-06, + "loss": 0.0766, + "step": 15790 + }, + { + "epoch": 6.61, + "learning_rate": 1.5364325016998528e-06, + "loss": 0.0858, + "step": 15791 + }, + { + "epoch": 6.61, + "learning_rate": 1.535529620733397e-06, + "loss": 0.0799, + "step": 15792 + }, + { + "epoch": 6.61, + "learning_rate": 1.5346269830706295e-06, + "loss": 0.0814, + "step": 15793 + }, + { + "epoch": 6.61, + "learning_rate": 1.5337245887374962e-06, + "loss": 0.0911, + "step": 15794 + }, + { + "epoch": 6.61, + "learning_rate": 1.5328224377599376e-06, + "loss": 0.0756, + "step": 15795 + }, + { + "epoch": 6.61, + "learning_rate": 1.531920530163885e-06, + "loss": 0.1003, + "step": 15796 + }, + { + "epoch": 6.61, + "learning_rate": 1.5310188659752612e-06, + "loss": 0.0924, + "step": 15797 + }, + { + "epoch": 6.61, + "learning_rate": 1.5301174452199874e-06, + "loss": 0.0671, + "step": 15798 + }, + { + "epoch": 6.61, + "learning_rate": 1.5292162679239697e-06, + "loss": 0.0975, + "step": 15799 + }, + { + "epoch": 6.62, + "learning_rate": 1.528315334113114e-06, + "loss": 0.0744, + "step": 15800 + }, + { + "epoch": 6.62, + "learning_rate": 1.5274146438133154e-06, + "loss": 0.0805, + "step": 15801 + }, + { + "epoch": 6.62, + "learning_rate": 1.5265141970504648e-06, + "loss": 0.0738, + "step": 15802 + }, + { + "epoch": 6.62, + "learning_rate": 1.5256139938504445e-06, + "loss": 0.1022, + "step": 15803 + }, + { + "epoch": 6.62, + "learning_rate": 1.524714034239132e-06, + "loss": 0.0975, + "step": 15804 + }, + { + "epoch": 6.62, + "learning_rate": 1.5238143182423914e-06, + "loss": 0.0859, + "step": 15805 + }, + { + "epoch": 6.62, + "learning_rate": 1.5229148458860865e-06, + "loss": 0.0904, + "step": 15806 + }, + { + "epoch": 6.62, + "learning_rate": 1.522015617196072e-06, + "loss": 0.0979, + "step": 15807 + }, + { + "epoch": 6.62, + "learning_rate": 1.5211166321981963e-06, + "loss": 0.0959, + "step": 15808 + }, + { + "epoch": 6.62, + "learning_rate": 1.5202178909182997e-06, + "loss": 0.0619, + "step": 15809 + }, + { + "epoch": 6.62, + "learning_rate": 1.5193193933822136e-06, + "loss": 0.0899, + "step": 15810 + }, + { + "epoch": 6.62, + "learning_rate": 1.5184211396157655e-06, + "loss": 0.073, + "step": 15811 + }, + { + "epoch": 6.62, + "learning_rate": 1.5175231296447778e-06, + "loss": 0.0746, + "step": 15812 + }, + { + "epoch": 6.62, + "learning_rate": 1.516625363495059e-06, + "loss": 0.0975, + "step": 15813 + }, + { + "epoch": 6.62, + "learning_rate": 1.5157278411924148e-06, + "loss": 0.0644, + "step": 15814 + }, + { + "epoch": 6.62, + "learning_rate": 1.5148305627626458e-06, + "loss": 0.0867, + "step": 15815 + }, + { + "epoch": 6.62, + "learning_rate": 1.5139335282315427e-06, + "loss": 0.089, + "step": 15816 + }, + { + "epoch": 6.62, + "learning_rate": 1.51303673762489e-06, + "loss": 0.0824, + "step": 15817 + }, + { + "epoch": 6.62, + "learning_rate": 1.5121401909684663e-06, + "loss": 0.0783, + "step": 15818 + }, + { + "epoch": 6.62, + "learning_rate": 1.51124388828804e-06, + "loss": 0.0422, + "step": 15819 + }, + { + "epoch": 6.62, + "learning_rate": 1.5103478296093743e-06, + "loss": 0.0688, + "step": 15820 + }, + { + "epoch": 6.62, + "learning_rate": 1.509452014958227e-06, + "loss": 0.0618, + "step": 15821 + }, + { + "epoch": 6.62, + "learning_rate": 1.5085564443603473e-06, + "loss": 0.0895, + "step": 15822 + }, + { + "epoch": 6.62, + "learning_rate": 1.507661117841477e-06, + "loss": 0.0749, + "step": 15823 + }, + { + "epoch": 6.63, + "learning_rate": 1.5067660354273527e-06, + "loss": 0.0942, + "step": 15824 + }, + { + "epoch": 6.63, + "learning_rate": 1.5058711971437035e-06, + "loss": 0.0595, + "step": 15825 + }, + { + "epoch": 6.63, + "learning_rate": 1.504976603016247e-06, + "loss": 0.1109, + "step": 15826 + }, + { + "epoch": 6.63, + "learning_rate": 1.5040822530707e-06, + "loss": 0.0918, + "step": 15827 + }, + { + "epoch": 6.63, + "learning_rate": 1.5031881473327691e-06, + "loss": 0.0847, + "step": 15828 + }, + { + "epoch": 6.63, + "learning_rate": 1.5022942858281553e-06, + "loss": 0.0581, + "step": 15829 + }, + { + "epoch": 6.63, + "learning_rate": 1.5014006685825512e-06, + "loss": 0.0948, + "step": 15830 + }, + { + "epoch": 6.63, + "learning_rate": 1.500507295621646e-06, + "loss": 0.0823, + "step": 15831 + }, + { + "epoch": 6.63, + "learning_rate": 1.499614166971114e-06, + "loss": 0.0959, + "step": 15832 + }, + { + "epoch": 6.63, + "learning_rate": 1.4987212826566323e-06, + "loss": 0.0836, + "step": 15833 + }, + { + "epoch": 6.63, + "learning_rate": 1.4978286427038602e-06, + "loss": 0.0649, + "step": 15834 + }, + { + "epoch": 6.63, + "learning_rate": 1.4969362471384597e-06, + "loss": 0.1003, + "step": 15835 + }, + { + "epoch": 6.63, + "learning_rate": 1.4960440959860812e-06, + "loss": 0.0916, + "step": 15836 + }, + { + "epoch": 6.63, + "learning_rate": 1.4951521892723685e-06, + "loss": 0.0738, + "step": 15837 + }, + { + "epoch": 6.63, + "learning_rate": 1.4942605270229594e-06, + "loss": 0.0764, + "step": 15838 + }, + { + "epoch": 6.63, + "learning_rate": 1.4933691092634862e-06, + "loss": 0.0718, + "step": 15839 + }, + { + "epoch": 6.63, + "learning_rate": 1.4924779360195662e-06, + "loss": 0.0773, + "step": 15840 + }, + { + "epoch": 6.63, + "learning_rate": 1.4915870073168194e-06, + "loss": 0.0804, + "step": 15841 + }, + { + "epoch": 6.63, + "learning_rate": 1.4906963231808524e-06, + "loss": 0.0988, + "step": 15842 + }, + { + "epoch": 6.63, + "learning_rate": 1.4898058836372697e-06, + "loss": 0.0759, + "step": 15843 + }, + { + "epoch": 6.63, + "learning_rate": 1.4889156887116652e-06, + "loss": 0.0621, + "step": 15844 + }, + { + "epoch": 6.63, + "learning_rate": 1.4880257384296282e-06, + "loss": 0.0546, + "step": 15845 + }, + { + "epoch": 6.63, + "learning_rate": 1.487136032816736e-06, + "loss": 0.0796, + "step": 15846 + }, + { + "epoch": 6.63, + "learning_rate": 1.4862465718985653e-06, + "loss": 0.0723, + "step": 15847 + }, + { + "epoch": 6.64, + "learning_rate": 1.4853573557006807e-06, + "loss": 0.0789, + "step": 15848 + }, + { + "epoch": 6.64, + "learning_rate": 1.484468384248643e-06, + "loss": 0.0721, + "step": 15849 + }, + { + "epoch": 6.64, + "learning_rate": 1.4835796575680062e-06, + "loss": 0.0997, + "step": 15850 + }, + { + "epoch": 6.64, + "learning_rate": 1.4826911756843144e-06, + "loss": 0.095, + "step": 15851 + }, + { + "epoch": 6.64, + "learning_rate": 1.4818029386231082e-06, + "loss": 0.0576, + "step": 15852 + }, + { + "epoch": 6.64, + "learning_rate": 1.480914946409916e-06, + "loss": 0.0996, + "step": 15853 + }, + { + "epoch": 6.64, + "learning_rate": 1.4800271990702664e-06, + "loss": 0.0596, + "step": 15854 + }, + { + "epoch": 6.64, + "learning_rate": 1.4791396966296711e-06, + "loss": 0.0884, + "step": 15855 + }, + { + "epoch": 6.64, + "learning_rate": 1.4782524391136443e-06, + "loss": 0.0684, + "step": 15856 + }, + { + "epoch": 6.64, + "learning_rate": 1.47736542654769e-06, + "loss": 0.0636, + "step": 15857 + }, + { + "epoch": 6.64, + "learning_rate": 1.4764786589573033e-06, + "loss": 0.0788, + "step": 15858 + }, + { + "epoch": 6.64, + "learning_rate": 1.4755921363679758e-06, + "loss": 0.0973, + "step": 15859 + }, + { + "epoch": 6.64, + "learning_rate": 1.4747058588051854e-06, + "loss": 0.0763, + "step": 15860 + }, + { + "epoch": 6.64, + "learning_rate": 1.4738198262944103e-06, + "loss": 0.0712, + "step": 15861 + }, + { + "epoch": 6.64, + "learning_rate": 1.472934038861119e-06, + "loss": 0.0674, + "step": 15862 + }, + { + "epoch": 6.64, + "learning_rate": 1.4720484965307714e-06, + "loss": 0.1018, + "step": 15863 + }, + { + "epoch": 6.64, + "learning_rate": 1.4711631993288222e-06, + "loss": 0.1098, + "step": 15864 + }, + { + "epoch": 6.64, + "learning_rate": 1.4702781472807181e-06, + "loss": 0.0737, + "step": 15865 + }, + { + "epoch": 6.64, + "learning_rate": 1.469393340411902e-06, + "loss": 0.0964, + "step": 15866 + }, + { + "epoch": 6.64, + "learning_rate": 1.4685087787478025e-06, + "loss": 0.1032, + "step": 15867 + }, + { + "epoch": 6.64, + "learning_rate": 1.4676244623138468e-06, + "loss": 0.0856, + "step": 15868 + }, + { + "epoch": 6.64, + "learning_rate": 1.466740391135455e-06, + "loss": 0.1217, + "step": 15869 + }, + { + "epoch": 6.64, + "learning_rate": 1.4658565652380373e-06, + "loss": 0.0789, + "step": 15870 + }, + { + "epoch": 6.64, + "learning_rate": 1.4649729846470006e-06, + "loss": 0.073, + "step": 15871 + }, + { + "epoch": 6.65, + "learning_rate": 1.4640896493877443e-06, + "loss": 0.0972, + "step": 15872 + }, + { + "epoch": 6.65, + "learning_rate": 1.4632065594856527e-06, + "loss": 0.0651, + "step": 15873 + }, + { + "epoch": 6.65, + "learning_rate": 1.462323714966114e-06, + "loss": 0.1199, + "step": 15874 + }, + { + "epoch": 6.65, + "learning_rate": 1.4614411158545038e-06, + "loss": 0.1062, + "step": 15875 + }, + { + "epoch": 6.65, + "learning_rate": 1.460558762176194e-06, + "loss": 0.0922, + "step": 15876 + }, + { + "epoch": 6.65, + "learning_rate": 1.4596766539565432e-06, + "loss": 0.0987, + "step": 15877 + }, + { + "epoch": 6.65, + "learning_rate": 1.4587947912209076e-06, + "loss": 0.0748, + "step": 15878 + }, + { + "epoch": 6.65, + "learning_rate": 1.4579131739946361e-06, + "loss": 0.0895, + "step": 15879 + }, + { + "epoch": 6.65, + "learning_rate": 1.4570318023030738e-06, + "loss": 0.0804, + "step": 15880 + }, + { + "epoch": 6.65, + "learning_rate": 1.4561506761715482e-06, + "loss": 0.0968, + "step": 15881 + }, + { + "epoch": 6.65, + "learning_rate": 1.4552697956253903e-06, + "loss": 0.0778, + "step": 15882 + }, + { + "epoch": 6.65, + "learning_rate": 1.4543891606899196e-06, + "loss": 0.0658, + "step": 15883 + }, + { + "epoch": 6.65, + "learning_rate": 1.4535087713904484e-06, + "loss": 0.1075, + "step": 15884 + }, + { + "epoch": 6.65, + "learning_rate": 1.452628627752285e-06, + "loss": 0.0736, + "step": 15885 + }, + { + "epoch": 6.65, + "learning_rate": 1.4517487298007283e-06, + "loss": 0.086, + "step": 15886 + }, + { + "epoch": 6.65, + "learning_rate": 1.4508690775610656e-06, + "loss": 0.0852, + "step": 15887 + }, + { + "epoch": 6.65, + "learning_rate": 1.4499896710585859e-06, + "loss": 0.0803, + "step": 15888 + }, + { + "epoch": 6.65, + "learning_rate": 1.4491105103185666e-06, + "loss": 0.095, + "step": 15889 + }, + { + "epoch": 6.65, + "learning_rate": 1.4482315953662762e-06, + "loss": 0.1004, + "step": 15890 + }, + { + "epoch": 6.65, + "learning_rate": 1.4473529262269814e-06, + "loss": 0.0824, + "step": 15891 + }, + { + "epoch": 6.65, + "learning_rate": 1.4464745029259365e-06, + "loss": 0.0721, + "step": 15892 + }, + { + "epoch": 6.65, + "learning_rate": 1.4455963254883942e-06, + "loss": 0.0821, + "step": 15893 + }, + { + "epoch": 6.65, + "learning_rate": 1.4447183939395925e-06, + "loss": 0.0898, + "step": 15894 + }, + { + "epoch": 6.65, + "learning_rate": 1.4438407083047679e-06, + "loss": 0.092, + "step": 15895 + }, + { + "epoch": 6.66, + "learning_rate": 1.4429632686091498e-06, + "loss": 0.0992, + "step": 15896 + }, + { + "epoch": 6.66, + "learning_rate": 1.4420860748779596e-06, + "loss": 0.0756, + "step": 15897 + }, + { + "epoch": 6.66, + "learning_rate": 1.4412091271364105e-06, + "loss": 0.0674, + "step": 15898 + }, + { + "epoch": 6.66, + "learning_rate": 1.4403324254097128e-06, + "loss": 0.0817, + "step": 15899 + }, + { + "epoch": 6.66, + "learning_rate": 1.4394559697230637e-06, + "loss": 0.0912, + "step": 15900 + }, + { + "epoch": 6.66, + "learning_rate": 1.4385797601016538e-06, + "loss": 0.0836, + "step": 15901 + }, + { + "epoch": 6.66, + "learning_rate": 1.437703796570672e-06, + "loss": 0.0662, + "step": 15902 + }, + { + "epoch": 6.66, + "learning_rate": 1.4368280791552957e-06, + "loss": 0.1097, + "step": 15903 + }, + { + "epoch": 6.66, + "learning_rate": 1.4359526078806985e-06, + "loss": 0.0821, + "step": 15904 + }, + { + "epoch": 6.66, + "learning_rate": 1.4350773827720432e-06, + "loss": 0.0611, + "step": 15905 + }, + { + "epoch": 6.66, + "learning_rate": 1.4342024038544878e-06, + "loss": 0.1006, + "step": 15906 + }, + { + "epoch": 6.66, + "learning_rate": 1.4333276711531863e-06, + "loss": 0.0588, + "step": 15907 + }, + { + "epoch": 6.66, + "learning_rate": 1.4324531846932766e-06, + "loss": 0.0845, + "step": 15908 + }, + { + "epoch": 6.66, + "learning_rate": 1.4315789444998973e-06, + "loss": 0.0851, + "step": 15909 + }, + { + "epoch": 6.66, + "learning_rate": 1.4307049505981773e-06, + "loss": 0.0803, + "step": 15910 + }, + { + "epoch": 6.66, + "learning_rate": 1.4298312030132388e-06, + "loss": 0.0909, + "step": 15911 + }, + { + "epoch": 6.66, + "learning_rate": 1.4289577017701984e-06, + "loss": 0.1119, + "step": 15912 + }, + { + "epoch": 6.66, + "learning_rate": 1.4280844468941647e-06, + "loss": 0.0861, + "step": 15913 + }, + { + "epoch": 6.66, + "learning_rate": 1.4272114384102353e-06, + "loss": 0.1005, + "step": 15914 + }, + { + "epoch": 6.66, + "learning_rate": 1.426338676343505e-06, + "loss": 0.0662, + "step": 15915 + }, + { + "epoch": 6.66, + "learning_rate": 1.4254661607190624e-06, + "loss": 0.0832, + "step": 15916 + }, + { + "epoch": 6.66, + "learning_rate": 1.424593891561985e-06, + "loss": 0.0695, + "step": 15917 + }, + { + "epoch": 6.66, + "learning_rate": 1.4237218688973475e-06, + "loss": 0.0699, + "step": 15918 + }, + { + "epoch": 6.66, + "learning_rate": 1.4228500927502164e-06, + "loss": 0.0661, + "step": 15919 + }, + { + "epoch": 6.67, + "learning_rate": 1.4219785631456452e-06, + "loss": 0.0687, + "step": 15920 + }, + { + "epoch": 6.67, + "learning_rate": 1.421107280108689e-06, + "loss": 0.066, + "step": 15921 + }, + { + "epoch": 6.67, + "learning_rate": 1.4202362436643924e-06, + "loss": 0.1135, + "step": 15922 + }, + { + "epoch": 6.67, + "learning_rate": 1.41936545383779e-06, + "loss": 0.0978, + "step": 15923 + }, + { + "epoch": 6.67, + "learning_rate": 1.4184949106539137e-06, + "loss": 0.0523, + "step": 15924 + }, + { + "epoch": 6.67, + "learning_rate": 1.417624614137786e-06, + "loss": 0.105, + "step": 15925 + }, + { + "epoch": 6.67, + "learning_rate": 1.416754564314422e-06, + "loss": 0.085, + "step": 15926 + }, + { + "epoch": 6.67, + "learning_rate": 1.4158847612088344e-06, + "loss": 0.0774, + "step": 15927 + }, + { + "epoch": 6.67, + "learning_rate": 1.4150152048460198e-06, + "loss": 0.0513, + "step": 15928 + }, + { + "epoch": 6.67, + "learning_rate": 1.4141458952509745e-06, + "loss": 0.086, + "step": 15929 + }, + { + "epoch": 6.67, + "learning_rate": 1.4132768324486868e-06, + "loss": 0.0827, + "step": 15930 + }, + { + "epoch": 6.67, + "learning_rate": 1.4124080164641362e-06, + "loss": 0.0576, + "step": 15931 + }, + { + "epoch": 6.67, + "learning_rate": 1.4115394473222977e-06, + "loss": 0.052, + "step": 15932 + }, + { + "epoch": 6.67, + "learning_rate": 1.4106711250481353e-06, + "loss": 0.0713, + "step": 15933 + }, + { + "epoch": 6.67, + "learning_rate": 1.4098030496666126e-06, + "loss": 0.086, + "step": 15934 + }, + { + "epoch": 6.67, + "learning_rate": 1.4089352212026752e-06, + "loss": 0.0926, + "step": 15935 + }, + { + "epoch": 6.67, + "learning_rate": 1.408067639681272e-06, + "loss": 0.1034, + "step": 15936 + }, + { + "epoch": 6.67, + "learning_rate": 1.4072003051273408e-06, + "loss": 0.0879, + "step": 15937 + }, + { + "epoch": 6.67, + "learning_rate": 1.4063332175658107e-06, + "loss": 0.0597, + "step": 15938 + }, + { + "epoch": 6.67, + "learning_rate": 1.405466377021607e-06, + "loss": 0.0823, + "step": 15939 + }, + { + "epoch": 6.67, + "learning_rate": 1.404599783519648e-06, + "loss": 0.0752, + "step": 15940 + }, + { + "epoch": 6.67, + "learning_rate": 1.4037334370848376e-06, + "loss": 0.0663, + "step": 15941 + }, + { + "epoch": 6.67, + "learning_rate": 1.4028673377420821e-06, + "loss": 0.0863, + "step": 15942 + }, + { + "epoch": 6.67, + "learning_rate": 1.4020014855162755e-06, + "loss": 0.0547, + "step": 15943 + }, + { + "epoch": 6.68, + "learning_rate": 1.4011358804323095e-06, + "loss": 0.0713, + "step": 15944 + }, + { + "epoch": 6.68, + "learning_rate": 1.4002705225150593e-06, + "loss": 0.0967, + "step": 15945 + }, + { + "epoch": 6.68, + "learning_rate": 1.3994054117894018e-06, + "loss": 0.0731, + "step": 15946 + }, + { + "epoch": 6.68, + "learning_rate": 1.398540548280203e-06, + "loss": 0.0999, + "step": 15947 + }, + { + "epoch": 6.68, + "learning_rate": 1.3976759320123268e-06, + "loss": 0.0824, + "step": 15948 + }, + { + "epoch": 6.68, + "learning_rate": 1.3968115630106193e-06, + "loss": 0.0903, + "step": 15949 + }, + { + "epoch": 6.68, + "learning_rate": 1.3959474412999296e-06, + "loss": 0.0937, + "step": 15950 + }, + { + "epoch": 6.68, + "learning_rate": 1.3950835669050955e-06, + "loss": 0.0559, + "step": 15951 + }, + { + "epoch": 6.68, + "learning_rate": 1.394219939850948e-06, + "loss": 0.1039, + "step": 15952 + }, + { + "epoch": 6.68, + "learning_rate": 1.3933565601623123e-06, + "loss": 0.0779, + "step": 15953 + }, + { + "epoch": 6.68, + "learning_rate": 1.3924934278640068e-06, + "loss": 0.0767, + "step": 15954 + }, + { + "epoch": 6.68, + "learning_rate": 1.3916305429808375e-06, + "loss": 0.0572, + "step": 15955 + }, + { + "epoch": 6.68, + "learning_rate": 1.3907679055376089e-06, + "loss": 0.0967, + "step": 15956 + }, + { + "epoch": 6.68, + "learning_rate": 1.3899055155591167e-06, + "loss": 0.0821, + "step": 15957 + }, + { + "epoch": 6.68, + "learning_rate": 1.3890433730701513e-06, + "loss": 0.0777, + "step": 15958 + }, + { + "epoch": 6.68, + "learning_rate": 1.3881814780954916e-06, + "loss": 0.0568, + "step": 15959 + }, + { + "epoch": 6.68, + "learning_rate": 1.3873198306599168e-06, + "loss": 0.0594, + "step": 15960 + }, + { + "epoch": 6.68, + "learning_rate": 1.3864584307881878e-06, + "loss": 0.0879, + "step": 15961 + }, + { + "epoch": 6.68, + "learning_rate": 1.3855972785050675e-06, + "loss": 0.0866, + "step": 15962 + }, + { + "epoch": 6.68, + "learning_rate": 1.3847363738353091e-06, + "loss": 0.0754, + "step": 15963 + }, + { + "epoch": 6.68, + "learning_rate": 1.3838757168036598e-06, + "loss": 0.0802, + "step": 15964 + }, + { + "epoch": 6.68, + "learning_rate": 1.3830153074348562e-06, + "loss": 0.1218, + "step": 15965 + }, + { + "epoch": 6.68, + "learning_rate": 1.382155145753631e-06, + "loss": 0.0689, + "step": 15966 + }, + { + "epoch": 6.68, + "learning_rate": 1.3812952317847116e-06, + "loss": 0.0817, + "step": 15967 + }, + { + "epoch": 6.69, + "learning_rate": 1.3804355655528123e-06, + "loss": 0.0786, + "step": 15968 + }, + { + "epoch": 6.69, + "learning_rate": 1.3795761470826408e-06, + "loss": 0.0705, + "step": 15969 + }, + { + "epoch": 6.69, + "learning_rate": 1.3787169763989038e-06, + "loss": 0.0808, + "step": 15970 + }, + { + "epoch": 6.69, + "learning_rate": 1.3778580535262975e-06, + "loss": 0.0696, + "step": 15971 + }, + { + "epoch": 6.69, + "learning_rate": 1.3769993784895096e-06, + "loss": 0.0934, + "step": 15972 + }, + { + "epoch": 6.69, + "learning_rate": 1.376140951313224e-06, + "loss": 0.0787, + "step": 15973 + }, + { + "epoch": 6.69, + "learning_rate": 1.3752827720221128e-06, + "loss": 0.0771, + "step": 15974 + }, + { + "epoch": 6.69, + "learning_rate": 1.3744248406408467e-06, + "loss": 0.0788, + "step": 15975 + }, + { + "epoch": 6.69, + "learning_rate": 1.3735671571940835e-06, + "loss": 0.0711, + "step": 15976 + }, + { + "epoch": 6.69, + "learning_rate": 1.3727097217064767e-06, + "loss": 0.0902, + "step": 15977 + }, + { + "epoch": 6.69, + "learning_rate": 1.3718525342026733e-06, + "loss": 0.0761, + "step": 15978 + }, + { + "epoch": 6.69, + "learning_rate": 1.3709955947073117e-06, + "loss": 0.0823, + "step": 15979 + }, + { + "epoch": 6.69, + "learning_rate": 1.3701389032450252e-06, + "loss": 0.0721, + "step": 15980 + }, + { + "epoch": 6.69, + "learning_rate": 1.3692824598404398e-06, + "loss": 0.0955, + "step": 15981 + }, + { + "epoch": 6.69, + "learning_rate": 1.3684262645181702e-06, + "loss": 0.0907, + "step": 15982 + }, + { + "epoch": 6.69, + "learning_rate": 1.3675703173028266e-06, + "loss": 0.0557, + "step": 15983 + }, + { + "epoch": 6.69, + "learning_rate": 1.366714618219015e-06, + "loss": 0.0809, + "step": 15984 + }, + { + "epoch": 6.69, + "learning_rate": 1.3658591672913301e-06, + "loss": 0.0915, + "step": 15985 + }, + { + "epoch": 6.69, + "learning_rate": 1.3650039645443614e-06, + "loss": 0.0763, + "step": 15986 + }, + { + "epoch": 6.69, + "learning_rate": 1.3641490100026943e-06, + "loss": 0.0912, + "step": 15987 + }, + { + "epoch": 6.69, + "learning_rate": 1.3632943036908974e-06, + "loss": 0.0533, + "step": 15988 + }, + { + "epoch": 6.69, + "learning_rate": 1.362439845633542e-06, + "loss": 0.1142, + "step": 15989 + }, + { + "epoch": 6.69, + "learning_rate": 1.3615856358551904e-06, + "loss": 0.0713, + "step": 15990 + }, + { + "epoch": 6.69, + "learning_rate": 1.3607316743803922e-06, + "loss": 0.0435, + "step": 15991 + }, + { + "epoch": 6.7, + "learning_rate": 1.3598779612336965e-06, + "loss": 0.0534, + "step": 15992 + }, + { + "epoch": 6.7, + "learning_rate": 1.3590244964396403e-06, + "loss": 0.0946, + "step": 15993 + }, + { + "epoch": 6.7, + "learning_rate": 1.3581712800227576e-06, + "loss": 0.0693, + "step": 15994 + }, + { + "epoch": 6.7, + "learning_rate": 1.357318312007575e-06, + "loss": 0.0973, + "step": 15995 + }, + { + "epoch": 6.7, + "learning_rate": 1.3564655924186066e-06, + "loss": 0.0871, + "step": 15996 + }, + { + "epoch": 6.7, + "learning_rate": 1.3556131212803635e-06, + "loss": 0.0774, + "step": 15997 + }, + { + "epoch": 6.7, + "learning_rate": 1.354760898617351e-06, + "loss": 0.074, + "step": 15998 + }, + { + "epoch": 6.7, + "learning_rate": 1.3539089244540648e-06, + "loss": 0.0737, + "step": 15999 + }, + { + "epoch": 6.7, + "learning_rate": 1.3530571988149943e-06, + "loss": 0.0847, + "step": 16000 + } + ], + "max_steps": 19104, + "num_train_epochs": 8, + "total_flos": 2.930250359618142e+18, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-16000/training_args.bin b/checkpoint-16000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..8b1b50208b5713a1848bb03b9cf39bd33c748f72 --- /dev/null +++ b/checkpoint-16000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:457d7e76c0fce8d03d8cfe3fdedf1b89cbf25e0f3a605b6976f387e928f69251 +size 3899 diff --git a/checkpoint-18000/added_tokens.json b/checkpoint-18000/added_tokens.json new file mode 100644 index 0000000000000000000000000000000000000000..301df2d484b892fb69765533defad74b1169af4d --- /dev/null +++ b/checkpoint-18000/added_tokens.json @@ -0,0 +1,6 @@ +{ + "": 32003, + "": 32001, + "": 32002, + "[PAD]": 32000 +} diff --git a/checkpoint-18000/config.json b/checkpoint-18000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..066a0107385020bf48a41a53e63dd916e2f96670 --- /dev/null +++ b/checkpoint-18000/config.json @@ -0,0 +1,30 @@ +{ + "_name_or_path": "/scratch/wzhu59/LLaVA-Med/weights/llava-med-7b-pretrain", + "architectures": [ + "LlavaLlamaForCausalLM" + ], + "bos_token_id": 0, + "eos_token_id": 1, + "freeze_mm_mlp_adapter": false, + "hidden_act": "silu", + "hidden_size": 4096, + "initializer_range": 0.02, + "intermediate_size": 11008, + "max_sequence_length": 2048, + "mm_hidden_size": 1024, + "mm_use_im_start_end": true, + "mm_vision_select_layer": -2, + "mm_vision_tower": "openai/clip-vit-large-patch14", + "model_type": "llava", + "num_attention_heads": 32, + "num_hidden_layers": 32, + "pad_token_id": -1, + "rms_norm_eps": 1e-06, + "tie_word_embeddings": false, + "torch_dtype": "float32", + "transformers_version": "4.28.0.dev0", + "tune_mm_mlp_adapter": false, + "use_cache": false, + "use_mm_proj": true, + "vocab_size": 32004 +} diff --git a/checkpoint-18000/generation_config.json b/checkpoint-18000/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..128b48946713166edcc3e3c62ae5fe285df776be --- /dev/null +++ b/checkpoint-18000/generation_config.json @@ -0,0 +1,7 @@ +{ + "_from_model_config": true, + "bos_token_id": 0, + "eos_token_id": 1, + "pad_token_id": 0, + "transformers_version": "4.28.0.dev0" +} diff --git a/checkpoint-18000/optimizer.pt b/checkpoint-18000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..92f4e6c67a793a5e6a0d3693a563f7a509e1788b --- /dev/null +++ b/checkpoint-18000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d35b9fb4343a76fb82cb0e67467cf79fc33fa1d65c0a0d4280c70c28f3df68a6 +size 13485322046 diff --git a/checkpoint-18000/pytorch_model-00001-of-00003.bin b/checkpoint-18000/pytorch_model-00001-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..0366de5388d4689bdc40867fc56a618347e296cd --- /dev/null +++ b/checkpoint-18000/pytorch_model-00001-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:64ff76b8a6e29d28cab7958bb6bb55b63e16b1d4e6f3e51935989bf1ec4f435b +size 9878059026 diff --git a/checkpoint-18000/pytorch_model-00002-of-00003.bin b/checkpoint-18000/pytorch_model-00002-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..afe8b2de2fa8037c7aa39e448a568746a5610ee2 --- /dev/null +++ b/checkpoint-18000/pytorch_model-00002-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc7960f5a1c45f8dd3e70c404ac4abd6365eec84c6553c26198e1695bfba7a94 +size 9894805046 diff --git a/checkpoint-18000/pytorch_model-00003-of-00003.bin b/checkpoint-18000/pytorch_model-00003-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..000a8a81791ec07c46565db7a81a9e63a8ef8e26 --- /dev/null +++ b/checkpoint-18000/pytorch_model-00003-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd23917ebca0f2ac9bfe1c1a467b7014255ced53cb0d5909ffd05aa5b2323950 +size 7197853429 diff --git a/checkpoint-18000/pytorch_model.bin.index.json b/checkpoint-18000/pytorch_model.bin.index.json new file mode 100644 index 0000000000000000000000000000000000000000..50d9a764a2282762889d821cfaa49b284963c070 --- /dev/null +++ b/checkpoint-18000/pytorch_model.bin.index.json @@ -0,0 +1,332 @@ +{ + "metadata": { + "total_size": 26970595328 + }, + "weight_map": { + "lm_head.weight": "pytorch_model-00003-of-00003.bin", + "model.embed_tokens.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.12.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.20.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.24.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.30.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.mm_projector.bias": "pytorch_model-00003-of-00003.bin", + "model.mm_projector.weight": "pytorch_model-00003-of-00003.bin", + "model.norm.weight": "pytorch_model-00003-of-00003.bin" + } +} diff --git a/checkpoint-18000/rng_state_0.pth b/checkpoint-18000/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..dd23e37561ea1206776c76959b7a70b759644b71 --- /dev/null +++ b/checkpoint-18000/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e3c0dadac97bbc4a6befe64dc8f31e929c3d48028d2c139d7972f341b2b1ebb +size 14583 diff --git a/checkpoint-18000/rng_state_1.pth b/checkpoint-18000/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..f488eb6d92a8608ca0038762493264e0e7acca40 --- /dev/null +++ b/checkpoint-18000/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c078143a87d02d736f8bbe878cb2707daf9377fb68ff52177f7c985aeb5a4328 +size 14583 diff --git a/checkpoint-18000/rng_state_2.pth b/checkpoint-18000/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..a1e8f7326d2a54bd48cb4401b2be24f71d3db6d7 --- /dev/null +++ b/checkpoint-18000/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b4b8b94e8e8206ca3e20e413e982ca61d670f2e81126e1303ea53eb8be6524e +size 14583 diff --git a/checkpoint-18000/rng_state_3.pth b/checkpoint-18000/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..e34b76bc1227ce30f968c4fb2238340a8e806cce --- /dev/null +++ b/checkpoint-18000/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08d9f22a14f53b2bcae6ccbd924524f0c71c5a17f162b4399e78f4f0772de1eb +size 14583 diff --git a/checkpoint-18000/scheduler.pt b/checkpoint-18000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..c9744f086cefab7e68b94d8ab42f07e453ecbcfe --- /dev/null +++ b/checkpoint-18000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4d7ce3f4754a372bef7a3b061e927f43fae84a907c9394f329b63759599bef1b +size 627 diff --git a/checkpoint-18000/special_tokens_map.json b/checkpoint-18000/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..5ce7dbc9759ecf5596c5122abbdf0b431355efa4 --- /dev/null +++ b/checkpoint-18000/special_tokens_map.json @@ -0,0 +1,6 @@ +{ + "bos_token": "", + "eos_token": "", + "pad_token": "[PAD]", + "unk_token": "" +} diff --git a/checkpoint-18000/tokenizer.model b/checkpoint-18000/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-18000/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-18000/tokenizer_config.json b/checkpoint-18000/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..22bcdfb351739d00df7b9b12f99757c6e7e0b540 --- /dev/null +++ b/checkpoint-18000/tokenizer_config.json @@ -0,0 +1,32 @@ +{ + "bos_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + }, + "clean_up_tokenization_spaces": false, + "eos_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + }, + "model_max_length": 2048, + "pad_token": null, + "padding_side": "right", + "special_tokens_map_file": "/nobackup/haotian/.cache/huggingface/hub/models--lmsys--vicuna-7b-delta-v0/snapshots/6d2e645f62e734821be8c392c08c8b2098b35434/special_tokens_map.json", + "tokenizer_class": "LlamaTokenizer", + "unk_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-18000/trainer_state.json b/checkpoint-18000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..4479d58ab32ad6a23c1f7b352080196b41f9e1ce --- /dev/null +++ b/checkpoint-18000/trainer_state.json @@ -0,0 +1,108016 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 7.5361105296211015, + "global_step": 18000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 3.484320557491289e-08, + "loss": 1.487, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 6.968641114982578e-08, + "loss": 1.5418, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 1.045296167247387e-07, + "loss": 1.2413, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 1.3937282229965157e-07, + "loss": 1.1777, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 1.7421602787456448e-07, + "loss": 0.9729, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 2.090592334494774e-07, + "loss": 0.9887, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 2.439024390243903e-07, + "loss": 0.9151, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 2.7874564459930313e-07, + "loss": 0.9341, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 3.1358885017421604e-07, + "loss": 0.9024, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 3.4843205574912896e-07, + "loss": 0.96, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 3.832752613240418e-07, + "loss": 0.9397, + "step": 11 + }, + { + "epoch": 0.01, + "learning_rate": 4.181184668989548e-07, + "loss": 1.0473, + "step": 12 + }, + { + "epoch": 0.01, + "learning_rate": 4.5296167247386764e-07, + "loss": 0.8521, + "step": 13 + }, + { + "epoch": 0.01, + "learning_rate": 4.878048780487805e-07, + "loss": 0.9829, + "step": 14 + }, + { + "epoch": 0.01, + "learning_rate": 5.226480836236935e-07, + "loss": 1.1014, + "step": 15 + }, + { + "epoch": 0.01, + "learning_rate": 5.574912891986063e-07, + "loss": 0.9186, + "step": 16 + }, + { + "epoch": 0.01, + "learning_rate": 5.923344947735192e-07, + "loss": 0.8867, + "step": 17 + }, + { + "epoch": 0.01, + "learning_rate": 6.271777003484321e-07, + "loss": 0.8071, + "step": 18 + }, + { + "epoch": 0.01, + "learning_rate": 6.62020905923345e-07, + "loss": 0.8481, + "step": 19 + }, + { + "epoch": 0.01, + "learning_rate": 6.968641114982579e-07, + "loss": 0.8675, + "step": 20 + }, + { + "epoch": 0.01, + "learning_rate": 7.317073170731707e-07, + "loss": 0.9635, + "step": 21 + }, + { + "epoch": 0.01, + "learning_rate": 7.665505226480836e-07, + "loss": 0.8868, + "step": 22 + }, + { + "epoch": 0.01, + "learning_rate": 8.013937282229965e-07, + "loss": 0.9029, + "step": 23 + }, + { + "epoch": 0.01, + "learning_rate": 8.362369337979096e-07, + "loss": 0.9118, + "step": 24 + }, + { + "epoch": 0.01, + "learning_rate": 8.710801393728225e-07, + "loss": 0.9768, + "step": 25 + }, + { + "epoch": 0.01, + "learning_rate": 9.059233449477353e-07, + "loss": 0.8292, + "step": 26 + }, + { + "epoch": 0.01, + "learning_rate": 9.407665505226482e-07, + "loss": 0.8865, + "step": 27 + }, + { + "epoch": 0.01, + "learning_rate": 9.75609756097561e-07, + "loss": 0.9918, + "step": 28 + }, + { + "epoch": 0.01, + "learning_rate": 1.010452961672474e-06, + "loss": 0.7146, + "step": 29 + }, + { + "epoch": 0.01, + "learning_rate": 1.045296167247387e-06, + "loss": 0.907, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 1.0801393728222997e-06, + "loss": 0.8015, + "step": 31 + }, + { + "epoch": 0.01, + "learning_rate": 1.1149825783972125e-06, + "loss": 0.7597, + "step": 32 + }, + { + "epoch": 0.01, + "learning_rate": 1.1498257839721255e-06, + "loss": 0.8095, + "step": 33 + }, + { + "epoch": 0.01, + "learning_rate": 1.1846689895470384e-06, + "loss": 0.9036, + "step": 34 + }, + { + "epoch": 0.01, + "learning_rate": 1.2195121951219514e-06, + "loss": 0.8162, + "step": 35 + }, + { + "epoch": 0.02, + "learning_rate": 1.2543554006968642e-06, + "loss": 0.9883, + "step": 36 + }, + { + "epoch": 0.02, + "learning_rate": 1.289198606271777e-06, + "loss": 0.8634, + "step": 37 + }, + { + "epoch": 0.02, + "learning_rate": 1.32404181184669e-06, + "loss": 0.7832, + "step": 38 + }, + { + "epoch": 0.02, + "learning_rate": 1.3588850174216028e-06, + "loss": 0.7378, + "step": 39 + }, + { + "epoch": 0.02, + "learning_rate": 1.3937282229965158e-06, + "loss": 0.8348, + "step": 40 + }, + { + "epoch": 0.02, + "learning_rate": 1.4285714285714286e-06, + "loss": 0.6925, + "step": 41 + }, + { + "epoch": 0.02, + "learning_rate": 1.4634146341463414e-06, + "loss": 0.8969, + "step": 42 + }, + { + "epoch": 0.02, + "learning_rate": 1.4982578397212545e-06, + "loss": 0.7099, + "step": 43 + }, + { + "epoch": 0.02, + "learning_rate": 1.5331010452961673e-06, + "loss": 0.8763, + "step": 44 + }, + { + "epoch": 0.02, + "learning_rate": 1.56794425087108e-06, + "loss": 0.7967, + "step": 45 + }, + { + "epoch": 0.02, + "learning_rate": 1.602787456445993e-06, + "loss": 0.7528, + "step": 46 + }, + { + "epoch": 0.02, + "learning_rate": 1.6376306620209059e-06, + "loss": 0.8367, + "step": 47 + }, + { + "epoch": 0.02, + "learning_rate": 1.6724738675958191e-06, + "loss": 0.8172, + "step": 48 + }, + { + "epoch": 0.02, + "learning_rate": 1.707317073170732e-06, + "loss": 0.7745, + "step": 49 + }, + { + "epoch": 0.02, + "learning_rate": 1.742160278745645e-06, + "loss": 0.8643, + "step": 50 + }, + { + "epoch": 0.02, + "learning_rate": 1.7770034843205577e-06, + "loss": 0.7374, + "step": 51 + }, + { + "epoch": 0.02, + "learning_rate": 1.8118466898954705e-06, + "loss": 0.8095, + "step": 52 + }, + { + "epoch": 0.02, + "learning_rate": 1.8466898954703836e-06, + "loss": 0.7755, + "step": 53 + }, + { + "epoch": 0.02, + "learning_rate": 1.8815331010452964e-06, + "loss": 0.8084, + "step": 54 + }, + { + "epoch": 0.02, + "learning_rate": 1.916376306620209e-06, + "loss": 0.7396, + "step": 55 + }, + { + "epoch": 0.02, + "learning_rate": 1.951219512195122e-06, + "loss": 0.7355, + "step": 56 + }, + { + "epoch": 0.02, + "learning_rate": 1.986062717770035e-06, + "loss": 0.7007, + "step": 57 + }, + { + "epoch": 0.02, + "learning_rate": 2.020905923344948e-06, + "loss": 0.8282, + "step": 58 + }, + { + "epoch": 0.02, + "learning_rate": 2.055749128919861e-06, + "loss": 0.7561, + "step": 59 + }, + { + "epoch": 0.03, + "learning_rate": 2.090592334494774e-06, + "loss": 0.8712, + "step": 60 + }, + { + "epoch": 0.03, + "learning_rate": 2.1254355400696864e-06, + "loss": 0.8559, + "step": 61 + }, + { + "epoch": 0.03, + "learning_rate": 2.1602787456445995e-06, + "loss": 0.7767, + "step": 62 + }, + { + "epoch": 0.03, + "learning_rate": 2.1951219512195125e-06, + "loss": 0.81, + "step": 63 + }, + { + "epoch": 0.03, + "learning_rate": 2.229965156794425e-06, + "loss": 0.805, + "step": 64 + }, + { + "epoch": 0.03, + "learning_rate": 2.264808362369338e-06, + "loss": 0.6648, + "step": 65 + }, + { + "epoch": 0.03, + "learning_rate": 2.299651567944251e-06, + "loss": 0.8027, + "step": 66 + }, + { + "epoch": 0.03, + "learning_rate": 2.334494773519164e-06, + "loss": 0.7623, + "step": 67 + }, + { + "epoch": 0.03, + "learning_rate": 2.3693379790940767e-06, + "loss": 0.7903, + "step": 68 + }, + { + "epoch": 0.03, + "learning_rate": 2.4041811846689897e-06, + "loss": 0.6917, + "step": 69 + }, + { + "epoch": 0.03, + "learning_rate": 2.4390243902439027e-06, + "loss": 0.8177, + "step": 70 + }, + { + "epoch": 0.03, + "learning_rate": 2.4738675958188153e-06, + "loss": 0.7348, + "step": 71 + }, + { + "epoch": 0.03, + "learning_rate": 2.5087108013937284e-06, + "loss": 0.6916, + "step": 72 + }, + { + "epoch": 0.03, + "learning_rate": 2.5435540069686414e-06, + "loss": 0.6663, + "step": 73 + }, + { + "epoch": 0.03, + "learning_rate": 2.578397212543554e-06, + "loss": 0.6881, + "step": 74 + }, + { + "epoch": 0.03, + "learning_rate": 2.6132404181184674e-06, + "loss": 0.6372, + "step": 75 + }, + { + "epoch": 0.03, + "learning_rate": 2.64808362369338e-06, + "loss": 0.7268, + "step": 76 + }, + { + "epoch": 0.03, + "learning_rate": 2.682926829268293e-06, + "loss": 0.7138, + "step": 77 + }, + { + "epoch": 0.03, + "learning_rate": 2.7177700348432056e-06, + "loss": 0.7341, + "step": 78 + }, + { + "epoch": 0.03, + "learning_rate": 2.7526132404181186e-06, + "loss": 0.7041, + "step": 79 + }, + { + "epoch": 0.03, + "learning_rate": 2.7874564459930316e-06, + "loss": 0.7239, + "step": 80 + }, + { + "epoch": 0.03, + "learning_rate": 2.8222996515679447e-06, + "loss": 0.6809, + "step": 81 + }, + { + "epoch": 0.03, + "learning_rate": 2.8571428571428573e-06, + "loss": 0.7256, + "step": 82 + }, + { + "epoch": 0.03, + "learning_rate": 2.8919860627177703e-06, + "loss": 0.7205, + "step": 83 + }, + { + "epoch": 0.04, + "learning_rate": 2.926829268292683e-06, + "loss": 0.6596, + "step": 84 + }, + { + "epoch": 0.04, + "learning_rate": 2.9616724738675963e-06, + "loss": 0.657, + "step": 85 + }, + { + "epoch": 0.04, + "learning_rate": 2.996515679442509e-06, + "loss": 0.6502, + "step": 86 + }, + { + "epoch": 0.04, + "learning_rate": 3.031358885017422e-06, + "loss": 0.7095, + "step": 87 + }, + { + "epoch": 0.04, + "learning_rate": 3.0662020905923345e-06, + "loss": 0.6643, + "step": 88 + }, + { + "epoch": 0.04, + "learning_rate": 3.1010452961672475e-06, + "loss": 0.6621, + "step": 89 + }, + { + "epoch": 0.04, + "learning_rate": 3.13588850174216e-06, + "loss": 0.7324, + "step": 90 + }, + { + "epoch": 0.04, + "learning_rate": 3.1707317073170736e-06, + "loss": 0.73, + "step": 91 + }, + { + "epoch": 0.04, + "learning_rate": 3.205574912891986e-06, + "loss": 0.7109, + "step": 92 + }, + { + "epoch": 0.04, + "learning_rate": 3.240418118466899e-06, + "loss": 0.695, + "step": 93 + }, + { + "epoch": 0.04, + "learning_rate": 3.2752613240418118e-06, + "loss": 0.5556, + "step": 94 + }, + { + "epoch": 0.04, + "learning_rate": 3.310104529616725e-06, + "loss": 0.6237, + "step": 95 + }, + { + "epoch": 0.04, + "learning_rate": 3.3449477351916382e-06, + "loss": 0.6923, + "step": 96 + }, + { + "epoch": 0.04, + "learning_rate": 3.379790940766551e-06, + "loss": 0.7276, + "step": 97 + }, + { + "epoch": 0.04, + "learning_rate": 3.414634146341464e-06, + "loss": 0.6814, + "step": 98 + }, + { + "epoch": 0.04, + "learning_rate": 3.4494773519163764e-06, + "loss": 0.7419, + "step": 99 + }, + { + "epoch": 0.04, + "learning_rate": 3.48432055749129e-06, + "loss": 0.6808, + "step": 100 + }, + { + "epoch": 0.04, + "learning_rate": 3.5191637630662025e-06, + "loss": 0.6811, + "step": 101 + }, + { + "epoch": 0.04, + "learning_rate": 3.5540069686411155e-06, + "loss": 0.6577, + "step": 102 + }, + { + "epoch": 0.04, + "learning_rate": 3.588850174216028e-06, + "loss": 0.6782, + "step": 103 + }, + { + "epoch": 0.04, + "learning_rate": 3.623693379790941e-06, + "loss": 0.7048, + "step": 104 + }, + { + "epoch": 0.04, + "learning_rate": 3.6585365853658537e-06, + "loss": 0.6673, + "step": 105 + }, + { + "epoch": 0.04, + "learning_rate": 3.693379790940767e-06, + "loss": 0.6239, + "step": 106 + }, + { + "epoch": 0.04, + "learning_rate": 3.7282229965156797e-06, + "loss": 0.6954, + "step": 107 + }, + { + "epoch": 0.05, + "learning_rate": 3.7630662020905927e-06, + "loss": 0.6656, + "step": 108 + }, + { + "epoch": 0.05, + "learning_rate": 3.7979094076655053e-06, + "loss": 0.6732, + "step": 109 + }, + { + "epoch": 0.05, + "learning_rate": 3.832752613240418e-06, + "loss": 0.7624, + "step": 110 + }, + { + "epoch": 0.05, + "learning_rate": 3.867595818815331e-06, + "loss": 0.7086, + "step": 111 + }, + { + "epoch": 0.05, + "learning_rate": 3.902439024390244e-06, + "loss": 0.6284, + "step": 112 + }, + { + "epoch": 0.05, + "learning_rate": 3.937282229965157e-06, + "loss": 0.6643, + "step": 113 + }, + { + "epoch": 0.05, + "learning_rate": 3.97212543554007e-06, + "loss": 0.6443, + "step": 114 + }, + { + "epoch": 0.05, + "learning_rate": 4.006968641114983e-06, + "loss": 0.6025, + "step": 115 + }, + { + "epoch": 0.05, + "learning_rate": 4.041811846689896e-06, + "loss": 0.6651, + "step": 116 + }, + { + "epoch": 0.05, + "learning_rate": 4.076655052264808e-06, + "loss": 0.6758, + "step": 117 + }, + { + "epoch": 0.05, + "learning_rate": 4.111498257839722e-06, + "loss": 0.6962, + "step": 118 + }, + { + "epoch": 0.05, + "learning_rate": 4.146341463414634e-06, + "loss": 0.6613, + "step": 119 + }, + { + "epoch": 0.05, + "learning_rate": 4.181184668989548e-06, + "loss": 0.6735, + "step": 120 + }, + { + "epoch": 0.05, + "learning_rate": 4.21602787456446e-06, + "loss": 0.5424, + "step": 121 + }, + { + "epoch": 0.05, + "learning_rate": 4.250871080139373e-06, + "loss": 0.7245, + "step": 122 + }, + { + "epoch": 0.05, + "learning_rate": 4.2857142857142855e-06, + "loss": 0.5873, + "step": 123 + }, + { + "epoch": 0.05, + "learning_rate": 4.320557491289199e-06, + "loss": 0.7426, + "step": 124 + }, + { + "epoch": 0.05, + "learning_rate": 4.3554006968641115e-06, + "loss": 0.7477, + "step": 125 + }, + { + "epoch": 0.05, + "learning_rate": 4.390243902439025e-06, + "loss": 0.6925, + "step": 126 + }, + { + "epoch": 0.05, + "learning_rate": 4.4250871080139375e-06, + "loss": 0.7745, + "step": 127 + }, + { + "epoch": 0.05, + "learning_rate": 4.45993031358885e-06, + "loss": 0.6976, + "step": 128 + }, + { + "epoch": 0.05, + "learning_rate": 4.4947735191637636e-06, + "loss": 0.6296, + "step": 129 + }, + { + "epoch": 0.05, + "learning_rate": 4.529616724738676e-06, + "loss": 0.6011, + "step": 130 + }, + { + "epoch": 0.05, + "learning_rate": 4.56445993031359e-06, + "loss": 0.6661, + "step": 131 + }, + { + "epoch": 0.06, + "learning_rate": 4.599303135888502e-06, + "loss": 0.6491, + "step": 132 + }, + { + "epoch": 0.06, + "learning_rate": 4.634146341463416e-06, + "loss": 0.6139, + "step": 133 + }, + { + "epoch": 0.06, + "learning_rate": 4.668989547038328e-06, + "loss": 0.6212, + "step": 134 + }, + { + "epoch": 0.06, + "learning_rate": 4.703832752613241e-06, + "loss": 0.7824, + "step": 135 + }, + { + "epoch": 0.06, + "learning_rate": 4.738675958188153e-06, + "loss": 0.6765, + "step": 136 + }, + { + "epoch": 0.06, + "learning_rate": 4.773519163763067e-06, + "loss": 0.6975, + "step": 137 + }, + { + "epoch": 0.06, + "learning_rate": 4.8083623693379794e-06, + "loss": 0.7251, + "step": 138 + }, + { + "epoch": 0.06, + "learning_rate": 4.843205574912893e-06, + "loss": 0.7232, + "step": 139 + }, + { + "epoch": 0.06, + "learning_rate": 4.8780487804878055e-06, + "loss": 0.6243, + "step": 140 + }, + { + "epoch": 0.06, + "learning_rate": 4.912891986062718e-06, + "loss": 0.6315, + "step": 141 + }, + { + "epoch": 0.06, + "learning_rate": 4.947735191637631e-06, + "loss": 0.632, + "step": 142 + }, + { + "epoch": 0.06, + "learning_rate": 4.982578397212544e-06, + "loss": 0.7962, + "step": 143 + }, + { + "epoch": 0.06, + "learning_rate": 5.017421602787457e-06, + "loss": 0.5554, + "step": 144 + }, + { + "epoch": 0.06, + "learning_rate": 5.052264808362369e-06, + "loss": 0.6004, + "step": 145 + }, + { + "epoch": 0.06, + "learning_rate": 5.087108013937283e-06, + "loss": 0.5713, + "step": 146 + }, + { + "epoch": 0.06, + "learning_rate": 5.121951219512195e-06, + "loss": 0.6607, + "step": 147 + }, + { + "epoch": 0.06, + "learning_rate": 5.156794425087108e-06, + "loss": 0.627, + "step": 148 + }, + { + "epoch": 0.06, + "learning_rate": 5.1916376306620205e-06, + "loss": 0.6944, + "step": 149 + }, + { + "epoch": 0.06, + "learning_rate": 5.226480836236935e-06, + "loss": 0.6617, + "step": 150 + }, + { + "epoch": 0.06, + "learning_rate": 5.261324041811847e-06, + "loss": 0.6516, + "step": 151 + }, + { + "epoch": 0.06, + "learning_rate": 5.29616724738676e-06, + "loss": 0.6271, + "step": 152 + }, + { + "epoch": 0.06, + "learning_rate": 5.331010452961673e-06, + "loss": 0.5682, + "step": 153 + }, + { + "epoch": 0.06, + "learning_rate": 5.365853658536586e-06, + "loss": 0.6439, + "step": 154 + }, + { + "epoch": 0.06, + "learning_rate": 5.400696864111499e-06, + "loss": 0.7387, + "step": 155 + }, + { + "epoch": 0.07, + "learning_rate": 5.435540069686411e-06, + "loss": 0.5682, + "step": 156 + }, + { + "epoch": 0.07, + "learning_rate": 5.470383275261324e-06, + "loss": 0.5933, + "step": 157 + }, + { + "epoch": 0.07, + "learning_rate": 5.505226480836237e-06, + "loss": 0.6817, + "step": 158 + }, + { + "epoch": 0.07, + "learning_rate": 5.540069686411151e-06, + "loss": 0.6364, + "step": 159 + }, + { + "epoch": 0.07, + "learning_rate": 5.574912891986063e-06, + "loss": 0.5878, + "step": 160 + }, + { + "epoch": 0.07, + "learning_rate": 5.609756097560977e-06, + "loss": 0.6615, + "step": 161 + }, + { + "epoch": 0.07, + "learning_rate": 5.644599303135889e-06, + "loss": 0.6586, + "step": 162 + }, + { + "epoch": 0.07, + "learning_rate": 5.679442508710802e-06, + "loss": 0.6446, + "step": 163 + }, + { + "epoch": 0.07, + "learning_rate": 5.7142857142857145e-06, + "loss": 0.733, + "step": 164 + }, + { + "epoch": 0.07, + "learning_rate": 5.749128919860628e-06, + "loss": 0.6642, + "step": 165 + }, + { + "epoch": 0.07, + "learning_rate": 5.7839721254355405e-06, + "loss": 0.6861, + "step": 166 + }, + { + "epoch": 0.07, + "learning_rate": 5.818815331010453e-06, + "loss": 0.623, + "step": 167 + }, + { + "epoch": 0.07, + "learning_rate": 5.853658536585366e-06, + "loss": 0.5951, + "step": 168 + }, + { + "epoch": 0.07, + "learning_rate": 5.88850174216028e-06, + "loss": 0.6887, + "step": 169 + }, + { + "epoch": 0.07, + "learning_rate": 5.923344947735193e-06, + "loss": 0.5783, + "step": 170 + }, + { + "epoch": 0.07, + "learning_rate": 5.958188153310105e-06, + "loss": 0.6036, + "step": 171 + }, + { + "epoch": 0.07, + "learning_rate": 5.993031358885018e-06, + "loss": 0.6031, + "step": 172 + }, + { + "epoch": 0.07, + "learning_rate": 6.027874564459931e-06, + "loss": 0.659, + "step": 173 + }, + { + "epoch": 0.07, + "learning_rate": 6.062717770034844e-06, + "loss": 0.6425, + "step": 174 + }, + { + "epoch": 0.07, + "learning_rate": 6.0975609756097564e-06, + "loss": 0.5897, + "step": 175 + }, + { + "epoch": 0.07, + "learning_rate": 6.132404181184669e-06, + "loss": 0.6508, + "step": 176 + }, + { + "epoch": 0.07, + "learning_rate": 6.1672473867595825e-06, + "loss": 0.6807, + "step": 177 + }, + { + "epoch": 0.07, + "learning_rate": 6.202090592334495e-06, + "loss": 0.5439, + "step": 178 + }, + { + "epoch": 0.07, + "learning_rate": 6.236933797909408e-06, + "loss": 0.5265, + "step": 179 + }, + { + "epoch": 0.08, + "learning_rate": 6.27177700348432e-06, + "loss": 0.6912, + "step": 180 + }, + { + "epoch": 0.08, + "learning_rate": 6.3066202090592345e-06, + "loss": 0.6348, + "step": 181 + }, + { + "epoch": 0.08, + "learning_rate": 6.341463414634147e-06, + "loss": 0.6374, + "step": 182 + }, + { + "epoch": 0.08, + "learning_rate": 6.37630662020906e-06, + "loss": 0.626, + "step": 183 + }, + { + "epoch": 0.08, + "learning_rate": 6.411149825783972e-06, + "loss": 0.6239, + "step": 184 + }, + { + "epoch": 0.08, + "learning_rate": 6.445993031358886e-06, + "loss": 0.6928, + "step": 185 + }, + { + "epoch": 0.08, + "learning_rate": 6.480836236933798e-06, + "loss": 0.6335, + "step": 186 + }, + { + "epoch": 0.08, + "learning_rate": 6.515679442508711e-06, + "loss": 0.5033, + "step": 187 + }, + { + "epoch": 0.08, + "learning_rate": 6.5505226480836235e-06, + "loss": 0.6619, + "step": 188 + }, + { + "epoch": 0.08, + "learning_rate": 6.585365853658538e-06, + "loss": 0.6189, + "step": 189 + }, + { + "epoch": 0.08, + "learning_rate": 6.62020905923345e-06, + "loss": 0.6277, + "step": 190 + }, + { + "epoch": 0.08, + "learning_rate": 6.655052264808363e-06, + "loss": 0.6073, + "step": 191 + }, + { + "epoch": 0.08, + "learning_rate": 6.6898954703832765e-06, + "loss": 0.6097, + "step": 192 + }, + { + "epoch": 0.08, + "learning_rate": 6.724738675958189e-06, + "loss": 0.5943, + "step": 193 + }, + { + "epoch": 0.08, + "learning_rate": 6.759581881533102e-06, + "loss": 0.5979, + "step": 194 + }, + { + "epoch": 0.08, + "learning_rate": 6.794425087108014e-06, + "loss": 0.598, + "step": 195 + }, + { + "epoch": 0.08, + "learning_rate": 6.829268292682928e-06, + "loss": 0.6136, + "step": 196 + }, + { + "epoch": 0.08, + "learning_rate": 6.86411149825784e-06, + "loss": 0.6153, + "step": 197 + }, + { + "epoch": 0.08, + "learning_rate": 6.898954703832753e-06, + "loss": 0.6817, + "step": 198 + }, + { + "epoch": 0.08, + "learning_rate": 6.9337979094076655e-06, + "loss": 0.5501, + "step": 199 + }, + { + "epoch": 0.08, + "learning_rate": 6.96864111498258e-06, + "loss": 0.7031, + "step": 200 + }, + { + "epoch": 0.08, + "learning_rate": 7.003484320557492e-06, + "loss": 0.5989, + "step": 201 + }, + { + "epoch": 0.08, + "learning_rate": 7.038327526132405e-06, + "loss": 0.6308, + "step": 202 + }, + { + "epoch": 0.08, + "learning_rate": 7.0731707317073175e-06, + "loss": 0.5342, + "step": 203 + }, + { + "epoch": 0.09, + "learning_rate": 7.108013937282231e-06, + "loss": 0.6175, + "step": 204 + }, + { + "epoch": 0.09, + "learning_rate": 7.1428571428571436e-06, + "loss": 0.5792, + "step": 205 + }, + { + "epoch": 0.09, + "learning_rate": 7.177700348432056e-06, + "loss": 0.6591, + "step": 206 + }, + { + "epoch": 0.09, + "learning_rate": 7.212543554006969e-06, + "loss": 0.6793, + "step": 207 + }, + { + "epoch": 0.09, + "learning_rate": 7.247386759581882e-06, + "loss": 0.6891, + "step": 208 + }, + { + "epoch": 0.09, + "learning_rate": 7.282229965156795e-06, + "loss": 0.5279, + "step": 209 + }, + { + "epoch": 0.09, + "learning_rate": 7.317073170731707e-06, + "loss": 0.6229, + "step": 210 + }, + { + "epoch": 0.09, + "learning_rate": 7.35191637630662e-06, + "loss": 0.6168, + "step": 211 + }, + { + "epoch": 0.09, + "learning_rate": 7.386759581881534e-06, + "loss": 0.4843, + "step": 212 + }, + { + "epoch": 0.09, + "learning_rate": 7.421602787456447e-06, + "loss": 0.5995, + "step": 213 + }, + { + "epoch": 0.09, + "learning_rate": 7.4564459930313594e-06, + "loss": 0.5968, + "step": 214 + }, + { + "epoch": 0.09, + "learning_rate": 7.491289198606272e-06, + "loss": 0.5597, + "step": 215 + }, + { + "epoch": 0.09, + "learning_rate": 7.5261324041811855e-06, + "loss": 0.6491, + "step": 216 + }, + { + "epoch": 0.09, + "learning_rate": 7.560975609756098e-06, + "loss": 0.643, + "step": 217 + }, + { + "epoch": 0.09, + "learning_rate": 7.595818815331011e-06, + "loss": 0.6927, + "step": 218 + }, + { + "epoch": 0.09, + "learning_rate": 7.630662020905924e-06, + "loss": 0.5916, + "step": 219 + }, + { + "epoch": 0.09, + "learning_rate": 7.665505226480837e-06, + "loss": 0.5445, + "step": 220 + }, + { + "epoch": 0.09, + "learning_rate": 7.70034843205575e-06, + "loss": 0.6517, + "step": 221 + }, + { + "epoch": 0.09, + "learning_rate": 7.735191637630662e-06, + "loss": 0.6, + "step": 222 + }, + { + "epoch": 0.09, + "learning_rate": 7.770034843205574e-06, + "loss": 0.6441, + "step": 223 + }, + { + "epoch": 0.09, + "learning_rate": 7.804878048780489e-06, + "loss": 0.6266, + "step": 224 + }, + { + "epoch": 0.09, + "learning_rate": 7.839721254355401e-06, + "loss": 0.5312, + "step": 225 + }, + { + "epoch": 0.09, + "learning_rate": 7.874564459930314e-06, + "loss": 0.614, + "step": 226 + }, + { + "epoch": 0.1, + "learning_rate": 7.909407665505228e-06, + "loss": 0.6742, + "step": 227 + }, + { + "epoch": 0.1, + "learning_rate": 7.94425087108014e-06, + "loss": 0.588, + "step": 228 + }, + { + "epoch": 0.1, + "learning_rate": 7.979094076655053e-06, + "loss": 0.5782, + "step": 229 + }, + { + "epoch": 0.1, + "learning_rate": 8.013937282229966e-06, + "loss": 0.6051, + "step": 230 + }, + { + "epoch": 0.1, + "learning_rate": 8.048780487804879e-06, + "loss": 0.671, + "step": 231 + }, + { + "epoch": 0.1, + "learning_rate": 8.083623693379791e-06, + "loss": 0.6003, + "step": 232 + }, + { + "epoch": 0.1, + "learning_rate": 8.118466898954704e-06, + "loss": 0.6023, + "step": 233 + }, + { + "epoch": 0.1, + "learning_rate": 8.153310104529616e-06, + "loss": 0.6305, + "step": 234 + }, + { + "epoch": 0.1, + "learning_rate": 8.18815331010453e-06, + "loss": 0.5449, + "step": 235 + }, + { + "epoch": 0.1, + "learning_rate": 8.222996515679443e-06, + "loss": 0.5957, + "step": 236 + }, + { + "epoch": 0.1, + "learning_rate": 8.257839721254356e-06, + "loss": 0.6335, + "step": 237 + }, + { + "epoch": 0.1, + "learning_rate": 8.292682926829268e-06, + "loss": 0.6214, + "step": 238 + }, + { + "epoch": 0.1, + "learning_rate": 8.327526132404183e-06, + "loss": 0.6096, + "step": 239 + }, + { + "epoch": 0.1, + "learning_rate": 8.362369337979095e-06, + "loss": 0.55, + "step": 240 + }, + { + "epoch": 0.1, + "learning_rate": 8.397212543554008e-06, + "loss": 0.641, + "step": 241 + }, + { + "epoch": 0.1, + "learning_rate": 8.43205574912892e-06, + "loss": 0.6294, + "step": 242 + }, + { + "epoch": 0.1, + "learning_rate": 8.466898954703833e-06, + "loss": 0.6216, + "step": 243 + }, + { + "epoch": 0.1, + "learning_rate": 8.501742160278746e-06, + "loss": 0.5949, + "step": 244 + }, + { + "epoch": 0.1, + "learning_rate": 8.536585365853658e-06, + "loss": 0.568, + "step": 245 + }, + { + "epoch": 0.1, + "learning_rate": 8.571428571428571e-06, + "loss": 0.6762, + "step": 246 + }, + { + "epoch": 0.1, + "learning_rate": 8.606271777003485e-06, + "loss": 0.6068, + "step": 247 + }, + { + "epoch": 0.1, + "learning_rate": 8.641114982578398e-06, + "loss": 0.584, + "step": 248 + }, + { + "epoch": 0.1, + "learning_rate": 8.67595818815331e-06, + "loss": 0.5967, + "step": 249 + }, + { + "epoch": 0.1, + "learning_rate": 8.710801393728223e-06, + "loss": 0.5988, + "step": 250 + }, + { + "epoch": 0.11, + "learning_rate": 8.745644599303137e-06, + "loss": 0.611, + "step": 251 + }, + { + "epoch": 0.11, + "learning_rate": 8.78048780487805e-06, + "loss": 0.6672, + "step": 252 + }, + { + "epoch": 0.11, + "learning_rate": 8.815331010452962e-06, + "loss": 0.5487, + "step": 253 + }, + { + "epoch": 0.11, + "learning_rate": 8.850174216027875e-06, + "loss": 0.5397, + "step": 254 + }, + { + "epoch": 0.11, + "learning_rate": 8.885017421602788e-06, + "loss": 0.7101, + "step": 255 + }, + { + "epoch": 0.11, + "learning_rate": 8.9198606271777e-06, + "loss": 0.6165, + "step": 256 + }, + { + "epoch": 0.11, + "learning_rate": 8.954703832752613e-06, + "loss": 0.5816, + "step": 257 + }, + { + "epoch": 0.11, + "learning_rate": 8.989547038327527e-06, + "loss": 0.6, + "step": 258 + }, + { + "epoch": 0.11, + "learning_rate": 9.02439024390244e-06, + "loss": 0.5065, + "step": 259 + }, + { + "epoch": 0.11, + "learning_rate": 9.059233449477352e-06, + "loss": 0.7077, + "step": 260 + }, + { + "epoch": 0.11, + "learning_rate": 9.094076655052265e-06, + "loss": 0.7149, + "step": 261 + }, + { + "epoch": 0.11, + "learning_rate": 9.12891986062718e-06, + "loss": 0.5832, + "step": 262 + }, + { + "epoch": 0.11, + "learning_rate": 9.163763066202092e-06, + "loss": 0.5587, + "step": 263 + }, + { + "epoch": 0.11, + "learning_rate": 9.198606271777004e-06, + "loss": 0.5781, + "step": 264 + }, + { + "epoch": 0.11, + "learning_rate": 9.233449477351917e-06, + "loss": 0.5852, + "step": 265 + }, + { + "epoch": 0.11, + "learning_rate": 9.268292682926831e-06, + "loss": 0.6214, + "step": 266 + }, + { + "epoch": 0.11, + "learning_rate": 9.303135888501744e-06, + "loss": 0.4923, + "step": 267 + }, + { + "epoch": 0.11, + "learning_rate": 9.337979094076656e-06, + "loss": 0.5257, + "step": 268 + }, + { + "epoch": 0.11, + "learning_rate": 9.372822299651569e-06, + "loss": 0.6273, + "step": 269 + }, + { + "epoch": 0.11, + "learning_rate": 9.407665505226482e-06, + "loss": 0.638, + "step": 270 + }, + { + "epoch": 0.11, + "learning_rate": 9.442508710801394e-06, + "loss": 0.5369, + "step": 271 + }, + { + "epoch": 0.11, + "learning_rate": 9.477351916376307e-06, + "loss": 0.6128, + "step": 272 + }, + { + "epoch": 0.11, + "learning_rate": 9.51219512195122e-06, + "loss": 0.61, + "step": 273 + }, + { + "epoch": 0.11, + "learning_rate": 9.547038327526134e-06, + "loss": 0.6117, + "step": 274 + }, + { + "epoch": 0.12, + "learning_rate": 9.581881533101046e-06, + "loss": 0.5191, + "step": 275 + }, + { + "epoch": 0.12, + "learning_rate": 9.616724738675959e-06, + "loss": 0.5381, + "step": 276 + }, + { + "epoch": 0.12, + "learning_rate": 9.651567944250871e-06, + "loss": 0.6686, + "step": 277 + }, + { + "epoch": 0.12, + "learning_rate": 9.686411149825786e-06, + "loss": 0.6415, + "step": 278 + }, + { + "epoch": 0.12, + "learning_rate": 9.721254355400698e-06, + "loss": 0.5333, + "step": 279 + }, + { + "epoch": 0.12, + "learning_rate": 9.756097560975611e-06, + "loss": 0.6254, + "step": 280 + }, + { + "epoch": 0.12, + "learning_rate": 9.790940766550524e-06, + "loss": 0.5899, + "step": 281 + }, + { + "epoch": 0.12, + "learning_rate": 9.825783972125436e-06, + "loss": 0.5635, + "step": 282 + }, + { + "epoch": 0.12, + "learning_rate": 9.860627177700349e-06, + "loss": 0.5338, + "step": 283 + }, + { + "epoch": 0.12, + "learning_rate": 9.895470383275261e-06, + "loss": 0.6047, + "step": 284 + }, + { + "epoch": 0.12, + "learning_rate": 9.930313588850174e-06, + "loss": 0.6124, + "step": 285 + }, + { + "epoch": 0.12, + "learning_rate": 9.965156794425088e-06, + "loss": 0.5827, + "step": 286 + }, + { + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.52, + "step": 287 + }, + { + "epoch": 0.12, + "learning_rate": 1.0034843205574913e-05, + "loss": 0.6033, + "step": 288 + }, + { + "epoch": 0.12, + "learning_rate": 1.0069686411149826e-05, + "loss": 0.5744, + "step": 289 + }, + { + "epoch": 0.12, + "learning_rate": 1.0104529616724739e-05, + "loss": 0.6823, + "step": 290 + }, + { + "epoch": 0.12, + "learning_rate": 1.0139372822299653e-05, + "loss": 0.6204, + "step": 291 + }, + { + "epoch": 0.12, + "learning_rate": 1.0174216027874565e-05, + "loss": 0.5409, + "step": 292 + }, + { + "epoch": 0.12, + "learning_rate": 1.0209059233449478e-05, + "loss": 0.5721, + "step": 293 + }, + { + "epoch": 0.12, + "learning_rate": 1.024390243902439e-05, + "loss": 0.6182, + "step": 294 + }, + { + "epoch": 0.12, + "learning_rate": 1.0278745644599303e-05, + "loss": 0.579, + "step": 295 + }, + { + "epoch": 0.12, + "learning_rate": 1.0313588850174216e-05, + "loss": 0.6389, + "step": 296 + }, + { + "epoch": 0.12, + "learning_rate": 1.0348432055749128e-05, + "loss": 0.6308, + "step": 297 + }, + { + "epoch": 0.12, + "learning_rate": 1.0383275261324041e-05, + "loss": 0.5235, + "step": 298 + }, + { + "epoch": 0.13, + "learning_rate": 1.0418118466898957e-05, + "loss": 0.5905, + "step": 299 + }, + { + "epoch": 0.13, + "learning_rate": 1.045296167247387e-05, + "loss": 0.6244, + "step": 300 + }, + { + "epoch": 0.13, + "learning_rate": 1.0487804878048782e-05, + "loss": 0.692, + "step": 301 + }, + { + "epoch": 0.13, + "learning_rate": 1.0522648083623695e-05, + "loss": 0.5482, + "step": 302 + }, + { + "epoch": 0.13, + "learning_rate": 1.0557491289198607e-05, + "loss": 0.6385, + "step": 303 + }, + { + "epoch": 0.13, + "learning_rate": 1.059233449477352e-05, + "loss": 0.643, + "step": 304 + }, + { + "epoch": 0.13, + "learning_rate": 1.0627177700348433e-05, + "loss": 0.5917, + "step": 305 + }, + { + "epoch": 0.13, + "learning_rate": 1.0662020905923345e-05, + "loss": 0.6183, + "step": 306 + }, + { + "epoch": 0.13, + "learning_rate": 1.069686411149826e-05, + "loss": 0.557, + "step": 307 + }, + { + "epoch": 0.13, + "learning_rate": 1.0731707317073172e-05, + "loss": 0.6811, + "step": 308 + }, + { + "epoch": 0.13, + "learning_rate": 1.0766550522648085e-05, + "loss": 0.6266, + "step": 309 + }, + { + "epoch": 0.13, + "learning_rate": 1.0801393728222997e-05, + "loss": 0.5146, + "step": 310 + }, + { + "epoch": 0.13, + "learning_rate": 1.083623693379791e-05, + "loss": 0.6395, + "step": 311 + }, + { + "epoch": 0.13, + "learning_rate": 1.0871080139372822e-05, + "loss": 0.6394, + "step": 312 + }, + { + "epoch": 0.13, + "learning_rate": 1.0905923344947735e-05, + "loss": 0.5787, + "step": 313 + }, + { + "epoch": 0.13, + "learning_rate": 1.0940766550522648e-05, + "loss": 0.5919, + "step": 314 + }, + { + "epoch": 0.13, + "learning_rate": 1.0975609756097562e-05, + "loss": 0.5283, + "step": 315 + }, + { + "epoch": 0.13, + "learning_rate": 1.1010452961672475e-05, + "loss": 0.5572, + "step": 316 + }, + { + "epoch": 0.13, + "learning_rate": 1.1045296167247389e-05, + "loss": 0.5774, + "step": 317 + }, + { + "epoch": 0.13, + "learning_rate": 1.1080139372822301e-05, + "loss": 0.5864, + "step": 318 + }, + { + "epoch": 0.13, + "learning_rate": 1.1114982578397214e-05, + "loss": 0.5731, + "step": 319 + }, + { + "epoch": 0.13, + "learning_rate": 1.1149825783972127e-05, + "loss": 0.5627, + "step": 320 + }, + { + "epoch": 0.13, + "learning_rate": 1.118466898954704e-05, + "loss": 0.5659, + "step": 321 + }, + { + "epoch": 0.13, + "learning_rate": 1.1219512195121953e-05, + "loss": 0.6295, + "step": 322 + }, + { + "epoch": 0.14, + "learning_rate": 1.1254355400696866e-05, + "loss": 0.5662, + "step": 323 + }, + { + "epoch": 0.14, + "learning_rate": 1.1289198606271779e-05, + "loss": 0.5659, + "step": 324 + }, + { + "epoch": 0.14, + "learning_rate": 1.1324041811846691e-05, + "loss": 0.74, + "step": 325 + }, + { + "epoch": 0.14, + "learning_rate": 1.1358885017421604e-05, + "loss": 0.604, + "step": 326 + }, + { + "epoch": 0.14, + "learning_rate": 1.1393728222996516e-05, + "loss": 0.5228, + "step": 327 + }, + { + "epoch": 0.14, + "learning_rate": 1.1428571428571429e-05, + "loss": 0.5745, + "step": 328 + }, + { + "epoch": 0.14, + "learning_rate": 1.1463414634146342e-05, + "loss": 0.5464, + "step": 329 + }, + { + "epoch": 0.14, + "learning_rate": 1.1498257839721256e-05, + "loss": 0.5523, + "step": 330 + }, + { + "epoch": 0.14, + "learning_rate": 1.1533101045296168e-05, + "loss": 0.5893, + "step": 331 + }, + { + "epoch": 0.14, + "learning_rate": 1.1567944250871081e-05, + "loss": 0.6265, + "step": 332 + }, + { + "epoch": 0.14, + "learning_rate": 1.1602787456445994e-05, + "loss": 0.6626, + "step": 333 + }, + { + "epoch": 0.14, + "learning_rate": 1.1637630662020906e-05, + "loss": 0.6379, + "step": 334 + }, + { + "epoch": 0.14, + "learning_rate": 1.1672473867595819e-05, + "loss": 0.6768, + "step": 335 + }, + { + "epoch": 0.14, + "learning_rate": 1.1707317073170731e-05, + "loss": 0.5926, + "step": 336 + }, + { + "epoch": 0.14, + "learning_rate": 1.1742160278745644e-05, + "loss": 0.5531, + "step": 337 + }, + { + "epoch": 0.14, + "learning_rate": 1.177700348432056e-05, + "loss": 0.5478, + "step": 338 + }, + { + "epoch": 0.14, + "learning_rate": 1.1811846689895473e-05, + "loss": 0.5894, + "step": 339 + }, + { + "epoch": 0.14, + "learning_rate": 1.1846689895470385e-05, + "loss": 0.5819, + "step": 340 + }, + { + "epoch": 0.14, + "learning_rate": 1.1881533101045298e-05, + "loss": 0.4793, + "step": 341 + }, + { + "epoch": 0.14, + "learning_rate": 1.191637630662021e-05, + "loss": 0.6046, + "step": 342 + }, + { + "epoch": 0.14, + "learning_rate": 1.1951219512195123e-05, + "loss": 0.5306, + "step": 343 + }, + { + "epoch": 0.14, + "learning_rate": 1.1986062717770036e-05, + "loss": 0.6153, + "step": 344 + }, + { + "epoch": 0.14, + "learning_rate": 1.2020905923344948e-05, + "loss": 0.6093, + "step": 345 + }, + { + "epoch": 0.14, + "learning_rate": 1.2055749128919862e-05, + "loss": 0.6022, + "step": 346 + }, + { + "epoch": 0.15, + "learning_rate": 1.2090592334494775e-05, + "loss": 0.515, + "step": 347 + }, + { + "epoch": 0.15, + "learning_rate": 1.2125435540069688e-05, + "loss": 0.6889, + "step": 348 + }, + { + "epoch": 0.15, + "learning_rate": 1.21602787456446e-05, + "loss": 0.576, + "step": 349 + }, + { + "epoch": 0.15, + "learning_rate": 1.2195121951219513e-05, + "loss": 0.5454, + "step": 350 + }, + { + "epoch": 0.15, + "learning_rate": 1.2229965156794425e-05, + "loss": 0.4593, + "step": 351 + }, + { + "epoch": 0.15, + "learning_rate": 1.2264808362369338e-05, + "loss": 0.5302, + "step": 352 + }, + { + "epoch": 0.15, + "learning_rate": 1.2299651567944252e-05, + "loss": 0.5591, + "step": 353 + }, + { + "epoch": 0.15, + "learning_rate": 1.2334494773519165e-05, + "loss": 0.6179, + "step": 354 + }, + { + "epoch": 0.15, + "learning_rate": 1.2369337979094078e-05, + "loss": 0.6015, + "step": 355 + }, + { + "epoch": 0.15, + "learning_rate": 1.240418118466899e-05, + "loss": 0.5615, + "step": 356 + }, + { + "epoch": 0.15, + "learning_rate": 1.2439024390243903e-05, + "loss": 0.6798, + "step": 357 + }, + { + "epoch": 0.15, + "learning_rate": 1.2473867595818815e-05, + "loss": 0.578, + "step": 358 + }, + { + "epoch": 0.15, + "learning_rate": 1.2508710801393728e-05, + "loss": 0.629, + "step": 359 + }, + { + "epoch": 0.15, + "learning_rate": 1.254355400696864e-05, + "loss": 0.6236, + "step": 360 + }, + { + "epoch": 0.15, + "learning_rate": 1.2578397212543556e-05, + "loss": 0.577, + "step": 361 + }, + { + "epoch": 0.15, + "learning_rate": 1.2613240418118469e-05, + "loss": 0.5588, + "step": 362 + }, + { + "epoch": 0.15, + "learning_rate": 1.2648083623693382e-05, + "loss": 0.6655, + "step": 363 + }, + { + "epoch": 0.15, + "learning_rate": 1.2682926829268294e-05, + "loss": 0.6124, + "step": 364 + }, + { + "epoch": 0.15, + "learning_rate": 1.2717770034843207e-05, + "loss": 0.4786, + "step": 365 + }, + { + "epoch": 0.15, + "learning_rate": 1.275261324041812e-05, + "loss": 0.566, + "step": 366 + }, + { + "epoch": 0.15, + "learning_rate": 1.2787456445993032e-05, + "loss": 0.5676, + "step": 367 + }, + { + "epoch": 0.15, + "learning_rate": 1.2822299651567945e-05, + "loss": 0.6284, + "step": 368 + }, + { + "epoch": 0.15, + "learning_rate": 1.2857142857142859e-05, + "loss": 0.5828, + "step": 369 + }, + { + "epoch": 0.15, + "learning_rate": 1.2891986062717772e-05, + "loss": 0.6464, + "step": 370 + }, + { + "epoch": 0.16, + "learning_rate": 1.2926829268292684e-05, + "loss": 0.5833, + "step": 371 + }, + { + "epoch": 0.16, + "learning_rate": 1.2961672473867597e-05, + "loss": 0.578, + "step": 372 + }, + { + "epoch": 0.16, + "learning_rate": 1.299651567944251e-05, + "loss": 0.5904, + "step": 373 + }, + { + "epoch": 0.16, + "learning_rate": 1.3031358885017422e-05, + "loss": 0.5454, + "step": 374 + }, + { + "epoch": 0.16, + "learning_rate": 1.3066202090592334e-05, + "loss": 0.5759, + "step": 375 + }, + { + "epoch": 0.16, + "learning_rate": 1.3101045296167247e-05, + "loss": 0.5288, + "step": 376 + }, + { + "epoch": 0.16, + "learning_rate": 1.3135888501742163e-05, + "loss": 0.6116, + "step": 377 + }, + { + "epoch": 0.16, + "learning_rate": 1.3170731707317076e-05, + "loss": 0.5446, + "step": 378 + }, + { + "epoch": 0.16, + "learning_rate": 1.3205574912891988e-05, + "loss": 0.5642, + "step": 379 + }, + { + "epoch": 0.16, + "learning_rate": 1.32404181184669e-05, + "loss": 0.5638, + "step": 380 + }, + { + "epoch": 0.16, + "learning_rate": 1.3275261324041813e-05, + "loss": 0.779, + "step": 381 + }, + { + "epoch": 0.16, + "learning_rate": 1.3310104529616726e-05, + "loss": 0.5757, + "step": 382 + }, + { + "epoch": 0.16, + "learning_rate": 1.3344947735191639e-05, + "loss": 0.6283, + "step": 383 + }, + { + "epoch": 0.16, + "learning_rate": 1.3379790940766553e-05, + "loss": 0.6289, + "step": 384 + }, + { + "epoch": 0.16, + "learning_rate": 1.3414634146341466e-05, + "loss": 0.5548, + "step": 385 + }, + { + "epoch": 0.16, + "learning_rate": 1.3449477351916378e-05, + "loss": 0.5725, + "step": 386 + }, + { + "epoch": 0.16, + "learning_rate": 1.348432055749129e-05, + "loss": 0.6109, + "step": 387 + }, + { + "epoch": 0.16, + "learning_rate": 1.3519163763066203e-05, + "loss": 0.5975, + "step": 388 + }, + { + "epoch": 0.16, + "learning_rate": 1.3554006968641116e-05, + "loss": 0.5972, + "step": 389 + }, + { + "epoch": 0.16, + "learning_rate": 1.3588850174216028e-05, + "loss": 0.6673, + "step": 390 + }, + { + "epoch": 0.16, + "learning_rate": 1.3623693379790941e-05, + "loss": 0.6687, + "step": 391 + }, + { + "epoch": 0.16, + "learning_rate": 1.3658536585365855e-05, + "loss": 0.504, + "step": 392 + }, + { + "epoch": 0.16, + "learning_rate": 1.3693379790940768e-05, + "loss": 0.6478, + "step": 393 + }, + { + "epoch": 0.16, + "learning_rate": 1.372822299651568e-05, + "loss": 0.6042, + "step": 394 + }, + { + "epoch": 0.17, + "learning_rate": 1.3763066202090593e-05, + "loss": 0.6066, + "step": 395 + }, + { + "epoch": 0.17, + "learning_rate": 1.3797909407665506e-05, + "loss": 0.5724, + "step": 396 + }, + { + "epoch": 0.17, + "learning_rate": 1.3832752613240418e-05, + "loss": 0.6821, + "step": 397 + }, + { + "epoch": 0.17, + "learning_rate": 1.3867595818815331e-05, + "loss": 0.544, + "step": 398 + }, + { + "epoch": 0.17, + "learning_rate": 1.3902439024390244e-05, + "loss": 0.5192, + "step": 399 + }, + { + "epoch": 0.17, + "learning_rate": 1.393728222996516e-05, + "loss": 0.6457, + "step": 400 + }, + { + "epoch": 0.17, + "learning_rate": 1.3972125435540072e-05, + "loss": 0.5996, + "step": 401 + }, + { + "epoch": 0.17, + "learning_rate": 1.4006968641114985e-05, + "loss": 0.5737, + "step": 402 + }, + { + "epoch": 0.17, + "learning_rate": 1.4041811846689897e-05, + "loss": 0.6506, + "step": 403 + }, + { + "epoch": 0.17, + "learning_rate": 1.407665505226481e-05, + "loss": 0.5541, + "step": 404 + }, + { + "epoch": 0.17, + "learning_rate": 1.4111498257839722e-05, + "loss": 0.5502, + "step": 405 + }, + { + "epoch": 0.17, + "learning_rate": 1.4146341463414635e-05, + "loss": 0.5909, + "step": 406 + }, + { + "epoch": 0.17, + "learning_rate": 1.4181184668989548e-05, + "loss": 0.5953, + "step": 407 + }, + { + "epoch": 0.17, + "learning_rate": 1.4216027874564462e-05, + "loss": 0.5376, + "step": 408 + }, + { + "epoch": 0.17, + "learning_rate": 1.4250871080139375e-05, + "loss": 0.5377, + "step": 409 + }, + { + "epoch": 0.17, + "learning_rate": 1.4285714285714287e-05, + "loss": 0.6134, + "step": 410 + }, + { + "epoch": 0.17, + "learning_rate": 1.43205574912892e-05, + "loss": 0.6472, + "step": 411 + }, + { + "epoch": 0.17, + "learning_rate": 1.4355400696864112e-05, + "loss": 0.4815, + "step": 412 + }, + { + "epoch": 0.17, + "learning_rate": 1.4390243902439025e-05, + "loss": 0.4832, + "step": 413 + }, + { + "epoch": 0.17, + "learning_rate": 1.4425087108013938e-05, + "loss": 0.5455, + "step": 414 + }, + { + "epoch": 0.17, + "learning_rate": 1.4459930313588852e-05, + "loss": 0.5445, + "step": 415 + }, + { + "epoch": 0.17, + "learning_rate": 1.4494773519163764e-05, + "loss": 0.5713, + "step": 416 + }, + { + "epoch": 0.17, + "learning_rate": 1.4529616724738677e-05, + "loss": 0.5713, + "step": 417 + }, + { + "epoch": 0.18, + "learning_rate": 1.456445993031359e-05, + "loss": 0.6725, + "step": 418 + }, + { + "epoch": 0.18, + "learning_rate": 1.4599303135888502e-05, + "loss": 0.6199, + "step": 419 + }, + { + "epoch": 0.18, + "learning_rate": 1.4634146341463415e-05, + "loss": 0.5049, + "step": 420 + }, + { + "epoch": 0.18, + "learning_rate": 1.4668989547038327e-05, + "loss": 0.5712, + "step": 421 + }, + { + "epoch": 0.18, + "learning_rate": 1.470383275261324e-05, + "loss": 0.6426, + "step": 422 + }, + { + "epoch": 0.18, + "learning_rate": 1.4738675958188156e-05, + "loss": 0.5879, + "step": 423 + }, + { + "epoch": 0.18, + "learning_rate": 1.4773519163763069e-05, + "loss": 0.5529, + "step": 424 + }, + { + "epoch": 0.18, + "learning_rate": 1.4808362369337981e-05, + "loss": 0.6029, + "step": 425 + }, + { + "epoch": 0.18, + "learning_rate": 1.4843205574912894e-05, + "loss": 0.6285, + "step": 426 + }, + { + "epoch": 0.18, + "learning_rate": 1.4878048780487806e-05, + "loss": 0.5832, + "step": 427 + }, + { + "epoch": 0.18, + "learning_rate": 1.4912891986062719e-05, + "loss": 0.506, + "step": 428 + }, + { + "epoch": 0.18, + "learning_rate": 1.4947735191637631e-05, + "loss": 0.6044, + "step": 429 + }, + { + "epoch": 0.18, + "learning_rate": 1.4982578397212544e-05, + "loss": 0.5128, + "step": 430 + }, + { + "epoch": 0.18, + "learning_rate": 1.5017421602787458e-05, + "loss": 0.5998, + "step": 431 + }, + { + "epoch": 0.18, + "learning_rate": 1.5052264808362371e-05, + "loss": 0.6959, + "step": 432 + }, + { + "epoch": 0.18, + "learning_rate": 1.5087108013937284e-05, + "loss": 0.6206, + "step": 433 + }, + { + "epoch": 0.18, + "learning_rate": 1.5121951219512196e-05, + "loss": 0.6337, + "step": 434 + }, + { + "epoch": 0.18, + "learning_rate": 1.5156794425087109e-05, + "loss": 0.5991, + "step": 435 + }, + { + "epoch": 0.18, + "learning_rate": 1.5191637630662021e-05, + "loss": 0.6201, + "step": 436 + }, + { + "epoch": 0.18, + "learning_rate": 1.5226480836236934e-05, + "loss": 0.4951, + "step": 437 + }, + { + "epoch": 0.18, + "learning_rate": 1.5261324041811848e-05, + "loss": 0.5876, + "step": 438 + }, + { + "epoch": 0.18, + "learning_rate": 1.529616724738676e-05, + "loss": 0.5744, + "step": 439 + }, + { + "epoch": 0.18, + "learning_rate": 1.5331010452961673e-05, + "loss": 0.5939, + "step": 440 + }, + { + "epoch": 0.18, + "learning_rate": 1.5365853658536586e-05, + "loss": 0.6046, + "step": 441 + }, + { + "epoch": 0.19, + "learning_rate": 1.54006968641115e-05, + "loss": 0.656, + "step": 442 + }, + { + "epoch": 0.19, + "learning_rate": 1.543554006968641e-05, + "loss": 0.6, + "step": 443 + }, + { + "epoch": 0.19, + "learning_rate": 1.5470383275261324e-05, + "loss": 0.5162, + "step": 444 + }, + { + "epoch": 0.19, + "learning_rate": 1.5505226480836236e-05, + "loss": 0.5459, + "step": 445 + }, + { + "epoch": 0.19, + "learning_rate": 1.554006968641115e-05, + "loss": 0.486, + "step": 446 + }, + { + "epoch": 0.19, + "learning_rate": 1.5574912891986065e-05, + "loss": 0.6013, + "step": 447 + }, + { + "epoch": 0.19, + "learning_rate": 1.5609756097560978e-05, + "loss": 0.6249, + "step": 448 + }, + { + "epoch": 0.19, + "learning_rate": 1.564459930313589e-05, + "loss": 0.545, + "step": 449 + }, + { + "epoch": 0.19, + "learning_rate": 1.5679442508710803e-05, + "loss": 0.6635, + "step": 450 + }, + { + "epoch": 0.19, + "learning_rate": 1.5714285714285715e-05, + "loss": 0.6335, + "step": 451 + }, + { + "epoch": 0.19, + "learning_rate": 1.5749128919860628e-05, + "loss": 0.5401, + "step": 452 + }, + { + "epoch": 0.19, + "learning_rate": 1.578397212543554e-05, + "loss": 0.5021, + "step": 453 + }, + { + "epoch": 0.19, + "learning_rate": 1.5818815331010456e-05, + "loss": 0.5869, + "step": 454 + }, + { + "epoch": 0.19, + "learning_rate": 1.585365853658537e-05, + "loss": 0.6381, + "step": 455 + }, + { + "epoch": 0.19, + "learning_rate": 1.588850174216028e-05, + "loss": 0.4976, + "step": 456 + }, + { + "epoch": 0.19, + "learning_rate": 1.5923344947735194e-05, + "loss": 0.5761, + "step": 457 + }, + { + "epoch": 0.19, + "learning_rate": 1.5958188153310107e-05, + "loss": 0.6614, + "step": 458 + }, + { + "epoch": 0.19, + "learning_rate": 1.599303135888502e-05, + "loss": 0.6228, + "step": 459 + }, + { + "epoch": 0.19, + "learning_rate": 1.6027874564459932e-05, + "loss": 0.5755, + "step": 460 + }, + { + "epoch": 0.19, + "learning_rate": 1.6062717770034845e-05, + "loss": 0.5943, + "step": 461 + }, + { + "epoch": 0.19, + "learning_rate": 1.6097560975609757e-05, + "loss": 0.5631, + "step": 462 + }, + { + "epoch": 0.19, + "learning_rate": 1.613240418118467e-05, + "loss": 0.6802, + "step": 463 + }, + { + "epoch": 0.19, + "learning_rate": 1.6167247386759582e-05, + "loss": 0.5835, + "step": 464 + }, + { + "epoch": 0.19, + "learning_rate": 1.6202090592334495e-05, + "loss": 0.5725, + "step": 465 + }, + { + "epoch": 0.2, + "learning_rate": 1.6236933797909408e-05, + "loss": 0.5487, + "step": 466 + }, + { + "epoch": 0.2, + "learning_rate": 1.627177700348432e-05, + "loss": 0.5889, + "step": 467 + }, + { + "epoch": 0.2, + "learning_rate": 1.6306620209059233e-05, + "loss": 0.5585, + "step": 468 + }, + { + "epoch": 0.2, + "learning_rate": 1.6341463414634145e-05, + "loss": 0.5731, + "step": 469 + }, + { + "epoch": 0.2, + "learning_rate": 1.637630662020906e-05, + "loss": 0.6129, + "step": 470 + }, + { + "epoch": 0.2, + "learning_rate": 1.6411149825783974e-05, + "loss": 0.586, + "step": 471 + }, + { + "epoch": 0.2, + "learning_rate": 1.6445993031358887e-05, + "loss": 0.6418, + "step": 472 + }, + { + "epoch": 0.2, + "learning_rate": 1.64808362369338e-05, + "loss": 0.6442, + "step": 473 + }, + { + "epoch": 0.2, + "learning_rate": 1.6515679442508712e-05, + "loss": 0.5419, + "step": 474 + }, + { + "epoch": 0.2, + "learning_rate": 1.6550522648083624e-05, + "loss": 0.5282, + "step": 475 + }, + { + "epoch": 0.2, + "learning_rate": 1.6585365853658537e-05, + "loss": 0.5588, + "step": 476 + }, + { + "epoch": 0.2, + "learning_rate": 1.662020905923345e-05, + "loss": 0.5536, + "step": 477 + }, + { + "epoch": 0.2, + "learning_rate": 1.6655052264808366e-05, + "loss": 0.5514, + "step": 478 + }, + { + "epoch": 0.2, + "learning_rate": 1.6689895470383278e-05, + "loss": 0.5554, + "step": 479 + }, + { + "epoch": 0.2, + "learning_rate": 1.672473867595819e-05, + "loss": 0.555, + "step": 480 + }, + { + "epoch": 0.2, + "learning_rate": 1.6759581881533103e-05, + "loss": 0.5606, + "step": 481 + }, + { + "epoch": 0.2, + "learning_rate": 1.6794425087108016e-05, + "loss": 0.5154, + "step": 482 + }, + { + "epoch": 0.2, + "learning_rate": 1.682926829268293e-05, + "loss": 0.5407, + "step": 483 + }, + { + "epoch": 0.2, + "learning_rate": 1.686411149825784e-05, + "loss": 0.6047, + "step": 484 + }, + { + "epoch": 0.2, + "learning_rate": 1.6898954703832754e-05, + "loss": 0.5323, + "step": 485 + }, + { + "epoch": 0.2, + "learning_rate": 1.6933797909407666e-05, + "loss": 0.5395, + "step": 486 + }, + { + "epoch": 0.2, + "learning_rate": 1.696864111498258e-05, + "loss": 0.6181, + "step": 487 + }, + { + "epoch": 0.2, + "learning_rate": 1.700348432055749e-05, + "loss": 0.5764, + "step": 488 + }, + { + "epoch": 0.2, + "learning_rate": 1.7038327526132404e-05, + "loss": 0.6155, + "step": 489 + }, + { + "epoch": 0.21, + "learning_rate": 1.7073170731707317e-05, + "loss": 0.5058, + "step": 490 + }, + { + "epoch": 0.21, + "learning_rate": 1.710801393728223e-05, + "loss": 0.6552, + "step": 491 + }, + { + "epoch": 0.21, + "learning_rate": 1.7142857142857142e-05, + "loss": 0.5362, + "step": 492 + }, + { + "epoch": 0.21, + "learning_rate": 1.7177700348432058e-05, + "loss": 0.6479, + "step": 493 + }, + { + "epoch": 0.21, + "learning_rate": 1.721254355400697e-05, + "loss": 0.4939, + "step": 494 + }, + { + "epoch": 0.21, + "learning_rate": 1.7247386759581883e-05, + "loss": 0.6442, + "step": 495 + }, + { + "epoch": 0.21, + "learning_rate": 1.7282229965156796e-05, + "loss": 0.5753, + "step": 496 + }, + { + "epoch": 0.21, + "learning_rate": 1.7317073170731708e-05, + "loss": 0.701, + "step": 497 + }, + { + "epoch": 0.21, + "learning_rate": 1.735191637630662e-05, + "loss": 0.5235, + "step": 498 + }, + { + "epoch": 0.21, + "learning_rate": 1.7386759581881533e-05, + "loss": 0.4857, + "step": 499 + }, + { + "epoch": 0.21, + "learning_rate": 1.7421602787456446e-05, + "loss": 0.5072, + "step": 500 + }, + { + "epoch": 0.21, + "learning_rate": 1.7456445993031362e-05, + "loss": 0.6019, + "step": 501 + }, + { + "epoch": 0.21, + "learning_rate": 1.7491289198606275e-05, + "loss": 0.5263, + "step": 502 + }, + { + "epoch": 0.21, + "learning_rate": 1.7526132404181187e-05, + "loss": 0.5893, + "step": 503 + }, + { + "epoch": 0.21, + "learning_rate": 1.75609756097561e-05, + "loss": 0.6769, + "step": 504 + }, + { + "epoch": 0.21, + "learning_rate": 1.7595818815331012e-05, + "loss": 0.5069, + "step": 505 + }, + { + "epoch": 0.21, + "learning_rate": 1.7630662020905925e-05, + "loss": 0.6241, + "step": 506 + }, + { + "epoch": 0.21, + "learning_rate": 1.7665505226480838e-05, + "loss": 0.5585, + "step": 507 + }, + { + "epoch": 0.21, + "learning_rate": 1.770034843205575e-05, + "loss": 0.5546, + "step": 508 + }, + { + "epoch": 0.21, + "learning_rate": 1.7735191637630663e-05, + "loss": 0.5824, + "step": 509 + }, + { + "epoch": 0.21, + "learning_rate": 1.7770034843205575e-05, + "loss": 0.5741, + "step": 510 + }, + { + "epoch": 0.21, + "learning_rate": 1.7804878048780488e-05, + "loss": 0.5867, + "step": 511 + }, + { + "epoch": 0.21, + "learning_rate": 1.78397212543554e-05, + "loss": 0.5562, + "step": 512 + }, + { + "epoch": 0.21, + "learning_rate": 1.7874564459930313e-05, + "loss": 0.5487, + "step": 513 + }, + { + "epoch": 0.22, + "learning_rate": 1.7909407665505226e-05, + "loss": 0.6476, + "step": 514 + }, + { + "epoch": 0.22, + "learning_rate": 1.7944250871080138e-05, + "loss": 0.5481, + "step": 515 + }, + { + "epoch": 0.22, + "learning_rate": 1.7979094076655054e-05, + "loss": 0.5381, + "step": 516 + }, + { + "epoch": 0.22, + "learning_rate": 1.8013937282229967e-05, + "loss": 0.5591, + "step": 517 + }, + { + "epoch": 0.22, + "learning_rate": 1.804878048780488e-05, + "loss": 0.591, + "step": 518 + }, + { + "epoch": 0.22, + "learning_rate": 1.8083623693379792e-05, + "loss": 0.465, + "step": 519 + }, + { + "epoch": 0.22, + "learning_rate": 1.8118466898954705e-05, + "loss": 0.6022, + "step": 520 + }, + { + "epoch": 0.22, + "learning_rate": 1.8153310104529617e-05, + "loss": 0.4784, + "step": 521 + }, + { + "epoch": 0.22, + "learning_rate": 1.818815331010453e-05, + "loss": 0.5983, + "step": 522 + }, + { + "epoch": 0.22, + "learning_rate": 1.8222996515679442e-05, + "loss": 0.5676, + "step": 523 + }, + { + "epoch": 0.22, + "learning_rate": 1.825783972125436e-05, + "loss": 0.5733, + "step": 524 + }, + { + "epoch": 0.22, + "learning_rate": 1.829268292682927e-05, + "loss": 0.5837, + "step": 525 + }, + { + "epoch": 0.22, + "learning_rate": 1.8327526132404184e-05, + "loss": 0.5893, + "step": 526 + }, + { + "epoch": 0.22, + "learning_rate": 1.8362369337979096e-05, + "loss": 0.6748, + "step": 527 + }, + { + "epoch": 0.22, + "learning_rate": 1.839721254355401e-05, + "loss": 0.5764, + "step": 528 + }, + { + "epoch": 0.22, + "learning_rate": 1.843205574912892e-05, + "loss": 0.6803, + "step": 529 + }, + { + "epoch": 0.22, + "learning_rate": 1.8466898954703834e-05, + "loss": 0.5717, + "step": 530 + }, + { + "epoch": 0.22, + "learning_rate": 1.8501742160278747e-05, + "loss": 0.5755, + "step": 531 + }, + { + "epoch": 0.22, + "learning_rate": 1.8536585365853663e-05, + "loss": 0.5581, + "step": 532 + }, + { + "epoch": 0.22, + "learning_rate": 1.8571428571428575e-05, + "loss": 0.5884, + "step": 533 + }, + { + "epoch": 0.22, + "learning_rate": 1.8606271777003488e-05, + "loss": 0.5363, + "step": 534 + }, + { + "epoch": 0.22, + "learning_rate": 1.86411149825784e-05, + "loss": 0.5798, + "step": 535 + }, + { + "epoch": 0.22, + "learning_rate": 1.8675958188153313e-05, + "loss": 0.5236, + "step": 536 + }, + { + "epoch": 0.22, + "learning_rate": 1.8710801393728226e-05, + "loss": 0.5025, + "step": 537 + }, + { + "epoch": 0.23, + "learning_rate": 1.8745644599303138e-05, + "loss": 0.6242, + "step": 538 + }, + { + "epoch": 0.23, + "learning_rate": 1.878048780487805e-05, + "loss": 0.6023, + "step": 539 + }, + { + "epoch": 0.23, + "learning_rate": 1.8815331010452963e-05, + "loss": 0.6211, + "step": 540 + }, + { + "epoch": 0.23, + "learning_rate": 1.8850174216027876e-05, + "loss": 0.5077, + "step": 541 + }, + { + "epoch": 0.23, + "learning_rate": 1.888501742160279e-05, + "loss": 0.5018, + "step": 542 + }, + { + "epoch": 0.23, + "learning_rate": 1.89198606271777e-05, + "loss": 0.6053, + "step": 543 + }, + { + "epoch": 0.23, + "learning_rate": 1.8954703832752614e-05, + "loss": 0.6225, + "step": 544 + }, + { + "epoch": 0.23, + "learning_rate": 1.8989547038327526e-05, + "loss": 0.6134, + "step": 545 + }, + { + "epoch": 0.23, + "learning_rate": 1.902439024390244e-05, + "loss": 0.5871, + "step": 546 + }, + { + "epoch": 0.23, + "learning_rate": 1.9059233449477355e-05, + "loss": 0.7059, + "step": 547 + }, + { + "epoch": 0.23, + "learning_rate": 1.9094076655052267e-05, + "loss": 0.5859, + "step": 548 + }, + { + "epoch": 0.23, + "learning_rate": 1.912891986062718e-05, + "loss": 0.5957, + "step": 549 + }, + { + "epoch": 0.23, + "learning_rate": 1.9163763066202093e-05, + "loss": 0.5299, + "step": 550 + }, + { + "epoch": 0.23, + "learning_rate": 1.9198606271777005e-05, + "loss": 0.6158, + "step": 551 + }, + { + "epoch": 0.23, + "learning_rate": 1.9233449477351918e-05, + "loss": 0.5808, + "step": 552 + }, + { + "epoch": 0.23, + "learning_rate": 1.926829268292683e-05, + "loss": 0.573, + "step": 553 + }, + { + "epoch": 0.23, + "learning_rate": 1.9303135888501743e-05, + "loss": 0.6362, + "step": 554 + }, + { + "epoch": 0.23, + "learning_rate": 1.933797909407666e-05, + "loss": 0.5059, + "step": 555 + }, + { + "epoch": 0.23, + "learning_rate": 1.937282229965157e-05, + "loss": 0.5784, + "step": 556 + }, + { + "epoch": 0.23, + "learning_rate": 1.9407665505226484e-05, + "loss": 0.649, + "step": 557 + }, + { + "epoch": 0.23, + "learning_rate": 1.9442508710801397e-05, + "loss": 0.6128, + "step": 558 + }, + { + "epoch": 0.23, + "learning_rate": 1.947735191637631e-05, + "loss": 0.595, + "step": 559 + }, + { + "epoch": 0.23, + "learning_rate": 1.9512195121951222e-05, + "loss": 0.5921, + "step": 560 + }, + { + "epoch": 0.23, + "learning_rate": 1.9547038327526135e-05, + "loss": 0.5053, + "step": 561 + }, + { + "epoch": 0.24, + "learning_rate": 1.9581881533101047e-05, + "loss": 0.4499, + "step": 562 + }, + { + "epoch": 0.24, + "learning_rate": 1.961672473867596e-05, + "loss": 0.5918, + "step": 563 + }, + { + "epoch": 0.24, + "learning_rate": 1.9651567944250872e-05, + "loss": 0.5511, + "step": 564 + }, + { + "epoch": 0.24, + "learning_rate": 1.9686411149825785e-05, + "loss": 0.6467, + "step": 565 + }, + { + "epoch": 0.24, + "learning_rate": 1.9721254355400697e-05, + "loss": 0.5508, + "step": 566 + }, + { + "epoch": 0.24, + "learning_rate": 1.975609756097561e-05, + "loss": 0.5592, + "step": 567 + }, + { + "epoch": 0.24, + "learning_rate": 1.9790940766550523e-05, + "loss": 0.5966, + "step": 568 + }, + { + "epoch": 0.24, + "learning_rate": 1.9825783972125435e-05, + "loss": 0.5011, + "step": 569 + }, + { + "epoch": 0.24, + "learning_rate": 1.9860627177700348e-05, + "loss": 0.593, + "step": 570 + }, + { + "epoch": 0.24, + "learning_rate": 1.9895470383275264e-05, + "loss": 0.5659, + "step": 571 + }, + { + "epoch": 0.24, + "learning_rate": 1.9930313588850176e-05, + "loss": 0.5767, + "step": 572 + }, + { + "epoch": 0.24, + "learning_rate": 1.996515679442509e-05, + "loss": 0.5624, + "step": 573 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.4904, + "step": 574 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999985627944e-05, + "loss": 0.5734, + "step": 575 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999999425117754e-05, + "loss": 0.5985, + "step": 576 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999998706514958e-05, + "loss": 0.5729, + "step": 577 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999997700471074e-05, + "loss": 0.558, + "step": 578 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999640698613e-05, + "loss": 0.5446, + "step": 579 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999994826060166e-05, + "loss": 0.5934, + "step": 580 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999295769322e-05, + "loss": 0.6511, + "step": 581 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999990801885353e-05, + "loss": 0.5262, + "step": 582 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999988358636622e-05, + "loss": 0.5777, + "step": 583 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999985627947104e-05, + "loss": 0.4609, + "step": 584 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999982609816867e-05, + "loss": 0.6021, + "step": 585 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999979304246006e-05, + "loss": 0.6274, + "step": 586 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999975711234616e-05, + "loss": 0.517, + "step": 587 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999971830782795e-05, + "loss": 0.562, + "step": 588 + }, + { + "epoch": 0.25, + "learning_rate": 1.999996766289066e-05, + "loss": 0.5583, + "step": 589 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999963207558326e-05, + "loss": 0.5975, + "step": 590 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999958464785925e-05, + "loss": 0.6094, + "step": 591 + }, + { + "epoch": 0.25, + "learning_rate": 1.999995343457359e-05, + "loss": 0.5579, + "step": 592 + }, + { + "epoch": 0.25, + "learning_rate": 1.999994811692147e-05, + "loss": 0.551, + "step": 593 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999942511829714e-05, + "loss": 0.5576, + "step": 594 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999936619298483e-05, + "loss": 0.5486, + "step": 595 + }, + { + "epoch": 0.25, + "learning_rate": 1.999993043932795e-05, + "loss": 0.588, + "step": 596 + }, + { + "epoch": 0.25, + "learning_rate": 1.999992397191829e-05, + "loss": 0.5231, + "step": 597 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999917217069687e-05, + "loss": 0.5773, + "step": 598 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999910174782337e-05, + "loss": 0.5609, + "step": 599 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999902845056444e-05, + "loss": 0.6032, + "step": 600 + }, + { + "epoch": 0.25, + "learning_rate": 1.999989522789222e-05, + "loss": 0.5846, + "step": 601 + }, + { + "epoch": 0.25, + "learning_rate": 1.999988732328988e-05, + "loss": 0.6105, + "step": 602 + }, + { + "epoch": 0.25, + "learning_rate": 1.999987913124965e-05, + "loss": 0.5158, + "step": 603 + }, + { + "epoch": 0.25, + "learning_rate": 1.999987065177177e-05, + "loss": 0.6727, + "step": 604 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999861884856483e-05, + "loss": 0.5327, + "step": 605 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999852830504038e-05, + "loss": 0.5825, + "step": 606 + }, + { + "epoch": 0.25, + "learning_rate": 1.99998434887147e-05, + "loss": 0.5495, + "step": 607 + }, + { + "epoch": 0.25, + "learning_rate": 1.999983385948873e-05, + "loss": 0.542, + "step": 608 + }, + { + "epoch": 0.25, + "learning_rate": 1.999982394282641e-05, + "loss": 0.6037, + "step": 609 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999813738728028e-05, + "loss": 0.6024, + "step": 610 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999803247193873e-05, + "loss": 0.6103, + "step": 611 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999792468224247e-05, + "loss": 0.6136, + "step": 612 + }, + { + "epoch": 0.26, + "learning_rate": 1.999978140181946e-05, + "loss": 0.5756, + "step": 613 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999770047979828e-05, + "loss": 0.6016, + "step": 614 + }, + { + "epoch": 0.26, + "learning_rate": 1.999975840670568e-05, + "loss": 0.6508, + "step": 615 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999746477997352e-05, + "loss": 0.5523, + "step": 616 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999734261855188e-05, + "loss": 0.6448, + "step": 617 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999721758279532e-05, + "loss": 0.5188, + "step": 618 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999708967270747e-05, + "loss": 0.5533, + "step": 619 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999695888829206e-05, + "loss": 0.4754, + "step": 620 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999682522955277e-05, + "loss": 0.5404, + "step": 621 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999668869649347e-05, + "loss": 0.5205, + "step": 622 + }, + { + "epoch": 0.26, + "learning_rate": 1.999965492891181e-05, + "loss": 0.5183, + "step": 623 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999640700743063e-05, + "loss": 0.5141, + "step": 624 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999626185143518e-05, + "loss": 0.5411, + "step": 625 + }, + { + "epoch": 0.26, + "learning_rate": 1.99996113821136e-05, + "loss": 0.5665, + "step": 626 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999596291653715e-05, + "loss": 0.5936, + "step": 627 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999580913764313e-05, + "loss": 0.5419, + "step": 628 + }, + { + "epoch": 0.26, + "learning_rate": 1.999956524844583e-05, + "loss": 0.5021, + "step": 629 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999549295698722e-05, + "loss": 0.5776, + "step": 630 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999533055523437e-05, + "loss": 0.4954, + "step": 631 + }, + { + "epoch": 0.26, + "learning_rate": 1.999951652792045e-05, + "loss": 0.5628, + "step": 632 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999499712890235e-05, + "loss": 0.6356, + "step": 633 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999482610433275e-05, + "loss": 0.5385, + "step": 634 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999465220550057e-05, + "loss": 0.5615, + "step": 635 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999447543241086e-05, + "loss": 0.5947, + "step": 636 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999429578506867e-05, + "loss": 0.6466, + "step": 637 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999411326347924e-05, + "loss": 0.5652, + "step": 638 + }, + { + "epoch": 0.27, + "learning_rate": 1.999939278676477e-05, + "loss": 0.571, + "step": 639 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999373959757945e-05, + "loss": 0.6267, + "step": 640 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999354845327993e-05, + "loss": 0.5371, + "step": 641 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999335443475452e-05, + "loss": 0.5337, + "step": 642 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999315754200892e-05, + "loss": 0.542, + "step": 643 + }, + { + "epoch": 0.27, + "learning_rate": 1.999929577750487e-05, + "loss": 0.5757, + "step": 644 + }, + { + "epoch": 0.27, + "learning_rate": 1.999927551338797e-05, + "loss": 0.5567, + "step": 645 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999254961850765e-05, + "loss": 0.5668, + "step": 646 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999234122893848e-05, + "loss": 0.5143, + "step": 647 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999212996517823e-05, + "loss": 0.5506, + "step": 648 + }, + { + "epoch": 0.27, + "learning_rate": 1.999919158272329e-05, + "loss": 0.6022, + "step": 649 + }, + { + "epoch": 0.27, + "learning_rate": 1.999916988151087e-05, + "loss": 0.5667, + "step": 650 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999147892881182e-05, + "loss": 0.5127, + "step": 651 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999125616834866e-05, + "loss": 0.6013, + "step": 652 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999103053372554e-05, + "loss": 0.4694, + "step": 653 + }, + { + "epoch": 0.27, + "learning_rate": 1.99990802024949e-05, + "loss": 0.5073, + "step": 654 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999057064202555e-05, + "loss": 0.5867, + "step": 655 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999033638496187e-05, + "loss": 0.5578, + "step": 656 + }, + { + "epoch": 0.28, + "learning_rate": 1.9999009925376475e-05, + "loss": 0.588, + "step": 657 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998985924844093e-05, + "loss": 0.5796, + "step": 658 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998961636899736e-05, + "loss": 0.559, + "step": 659 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998937061544096e-05, + "loss": 0.4719, + "step": 660 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998912198777886e-05, + "loss": 0.5707, + "step": 661 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998887048601812e-05, + "loss": 0.5999, + "step": 662 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998861611016608e-05, + "loss": 0.49, + "step": 663 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998835886022997e-05, + "loss": 0.6035, + "step": 664 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998809873621722e-05, + "loss": 0.5609, + "step": 665 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998783573813535e-05, + "loss": 0.5078, + "step": 666 + }, + { + "epoch": 0.28, + "learning_rate": 1.999875698659918e-05, + "loss": 0.5402, + "step": 667 + }, + { + "epoch": 0.28, + "learning_rate": 1.999873011197943e-05, + "loss": 0.5853, + "step": 668 + }, + { + "epoch": 0.28, + "learning_rate": 1.999870294995506e-05, + "loss": 0.4822, + "step": 669 + }, + { + "epoch": 0.28, + "learning_rate": 1.999867550052684e-05, + "loss": 0.6014, + "step": 670 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998647763695568e-05, + "loss": 0.5914, + "step": 671 + }, + { + "epoch": 0.28, + "learning_rate": 1.999861973946204e-05, + "loss": 0.5698, + "step": 672 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998591427827062e-05, + "loss": 0.5685, + "step": 673 + }, + { + "epoch": 0.28, + "learning_rate": 1.999856282879144e-05, + "loss": 0.6114, + "step": 674 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998533942356008e-05, + "loss": 0.5575, + "step": 675 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998504768521588e-05, + "loss": 0.6381, + "step": 676 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998475307289018e-05, + "loss": 0.5428, + "step": 677 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998445558659152e-05, + "loss": 0.5533, + "step": 678 + }, + { + "epoch": 0.28, + "learning_rate": 1.999841552263284e-05, + "loss": 0.4876, + "step": 679 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998385199210945e-05, + "loss": 0.5705, + "step": 680 + }, + { + "epoch": 0.29, + "learning_rate": 1.999835458839434e-05, + "loss": 0.6025, + "step": 681 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998323690183907e-05, + "loss": 0.6213, + "step": 682 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998292504580528e-05, + "loss": 0.5695, + "step": 683 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998261031585104e-05, + "loss": 0.5282, + "step": 684 + }, + { + "epoch": 0.29, + "learning_rate": 1.999822927119854e-05, + "loss": 0.5748, + "step": 685 + }, + { + "epoch": 0.29, + "learning_rate": 1.999819722342175e-05, + "loss": 0.5202, + "step": 686 + }, + { + "epoch": 0.29, + "learning_rate": 1.999816488825565e-05, + "loss": 0.5667, + "step": 687 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998132265701176e-05, + "loss": 0.5365, + "step": 688 + }, + { + "epoch": 0.29, + "learning_rate": 1.999809935575926e-05, + "loss": 0.6048, + "step": 689 + }, + { + "epoch": 0.29, + "learning_rate": 1.999806615843085e-05, + "loss": 0.6033, + "step": 690 + }, + { + "epoch": 0.29, + "learning_rate": 1.99980326737169e-05, + "loss": 0.5729, + "step": 691 + }, + { + "epoch": 0.29, + "learning_rate": 1.999799890161837e-05, + "loss": 0.492, + "step": 692 + }, + { + "epoch": 0.29, + "learning_rate": 1.999796484213624e-05, + "loss": 0.5711, + "step": 693 + }, + { + "epoch": 0.29, + "learning_rate": 1.999793049527148e-05, + "loss": 0.5011, + "step": 694 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997895861025078e-05, + "loss": 0.5375, + "step": 695 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997860939398035e-05, + "loss": 0.5716, + "step": 696 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997825730391345e-05, + "loss": 0.5735, + "step": 697 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997790234006034e-05, + "loss": 0.529, + "step": 698 + }, + { + "epoch": 0.29, + "learning_rate": 1.999775445024311e-05, + "loss": 0.5457, + "step": 699 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997718379103608e-05, + "loss": 0.5196, + "step": 700 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997682020588563e-05, + "loss": 0.5419, + "step": 701 + }, + { + "epoch": 0.29, + "learning_rate": 1.999764537469902e-05, + "loss": 0.4548, + "step": 702 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997608441436033e-05, + "loss": 0.4813, + "step": 703 + }, + { + "epoch": 0.29, + "learning_rate": 1.999757122080066e-05, + "loss": 0.5566, + "step": 704 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997533712793976e-05, + "loss": 0.5593, + "step": 705 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997495917417054e-05, + "loss": 0.5683, + "step": 706 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997457834670988e-05, + "loss": 0.5919, + "step": 707 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997419464556868e-05, + "loss": 0.6129, + "step": 708 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997380807075796e-05, + "loss": 0.4535, + "step": 709 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997341862228886e-05, + "loss": 0.6517, + "step": 710 + }, + { + "epoch": 0.3, + "learning_rate": 1.999730263001725e-05, + "loss": 0.5925, + "step": 711 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997263110442025e-05, + "loss": 0.4815, + "step": 712 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997223303504345e-05, + "loss": 0.5777, + "step": 713 + }, + { + "epoch": 0.3, + "learning_rate": 1.999718320920535e-05, + "loss": 0.4826, + "step": 714 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997142827546196e-05, + "loss": 0.6005, + "step": 715 + }, + { + "epoch": 0.3, + "learning_rate": 1.999710215852804e-05, + "loss": 0.6185, + "step": 716 + }, + { + "epoch": 0.3, + "learning_rate": 1.999706120215206e-05, + "loss": 0.6308, + "step": 717 + }, + { + "epoch": 0.3, + "learning_rate": 1.999701995841942e-05, + "loss": 0.6157, + "step": 718 + }, + { + "epoch": 0.3, + "learning_rate": 1.999697842733132e-05, + "loss": 0.517, + "step": 719 + }, + { + "epoch": 0.3, + "learning_rate": 1.999693660888894e-05, + "loss": 0.5469, + "step": 720 + }, + { + "epoch": 0.3, + "learning_rate": 1.999689450309349e-05, + "loss": 0.5627, + "step": 721 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996852109946178e-05, + "loss": 0.5886, + "step": 722 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996809429448224e-05, + "loss": 0.5238, + "step": 723 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996766461600855e-05, + "loss": 0.6008, + "step": 724 + }, + { + "epoch": 0.3, + "learning_rate": 1.99967232064053e-05, + "loss": 0.4986, + "step": 725 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996679663862815e-05, + "loss": 0.5396, + "step": 726 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996635833974637e-05, + "loss": 0.497, + "step": 727 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996591716742037e-05, + "loss": 0.5982, + "step": 728 + }, + { + "epoch": 0.31, + "learning_rate": 1.999654731216628e-05, + "loss": 0.5701, + "step": 729 + }, + { + "epoch": 0.31, + "learning_rate": 1.999650262024864e-05, + "loss": 0.5235, + "step": 730 + }, + { + "epoch": 0.31, + "learning_rate": 1.99964576409904e-05, + "loss": 0.5917, + "step": 731 + }, + { + "epoch": 0.31, + "learning_rate": 1.999641237439286e-05, + "loss": 0.5691, + "step": 732 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996366820457313e-05, + "loss": 0.6141, + "step": 733 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996320979185074e-05, + "loss": 0.5376, + "step": 734 + }, + { + "epoch": 0.31, + "learning_rate": 1.999627485057746e-05, + "loss": 0.4333, + "step": 735 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996228434635797e-05, + "loss": 0.5657, + "step": 736 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996181731361415e-05, + "loss": 0.5721, + "step": 737 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996134740755662e-05, + "loss": 0.6029, + "step": 738 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996087462819884e-05, + "loss": 0.5711, + "step": 739 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996039897555443e-05, + "loss": 0.5509, + "step": 740 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995992044963705e-05, + "loss": 0.555, + "step": 741 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995943905046048e-05, + "loss": 0.4628, + "step": 742 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995895477803854e-05, + "loss": 0.5218, + "step": 743 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995846763238514e-05, + "loss": 0.4748, + "step": 744 + }, + { + "epoch": 0.31, + "learning_rate": 1.999579776135143e-05, + "loss": 0.6094, + "step": 745 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995748472144005e-05, + "loss": 0.5397, + "step": 746 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995698895617663e-05, + "loss": 0.6416, + "step": 747 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995649031773828e-05, + "loss": 0.5666, + "step": 748 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995598880613928e-05, + "loss": 0.5718, + "step": 749 + }, + { + "epoch": 0.31, + "learning_rate": 1.999554844213941e-05, + "loss": 0.5107, + "step": 750 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995497716351725e-05, + "loss": 0.5182, + "step": 751 + }, + { + "epoch": 0.31, + "learning_rate": 1.999544670325232e-05, + "loss": 0.645, + "step": 752 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995395402842678e-05, + "loss": 0.6027, + "step": 753 + }, + { + "epoch": 0.32, + "learning_rate": 1.999534381512426e-05, + "loss": 0.506, + "step": 754 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995291940098555e-05, + "loss": 0.4984, + "step": 755 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995239777767053e-05, + "loss": 0.4942, + "step": 756 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995187328131252e-05, + "loss": 0.5343, + "step": 757 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995134591192664e-05, + "loss": 0.4752, + "step": 758 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995081566952795e-05, + "loss": 0.4636, + "step": 759 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995028255413184e-05, + "loss": 0.5338, + "step": 760 + }, + { + "epoch": 0.32, + "learning_rate": 1.999497465657535e-05, + "loss": 0.5684, + "step": 761 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994920770440842e-05, + "loss": 0.4523, + "step": 762 + }, + { + "epoch": 0.32, + "learning_rate": 1.99948665970112e-05, + "loss": 0.6039, + "step": 763 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994812136287994e-05, + "loss": 0.4691, + "step": 764 + }, + { + "epoch": 0.32, + "learning_rate": 1.999475738827278e-05, + "loss": 0.5422, + "step": 765 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994702352967132e-05, + "loss": 0.5557, + "step": 766 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994647030372638e-05, + "loss": 0.545, + "step": 767 + }, + { + "epoch": 0.32, + "learning_rate": 1.999459142049088e-05, + "loss": 0.5399, + "step": 768 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994535523323462e-05, + "loss": 0.5359, + "step": 769 + }, + { + "epoch": 0.32, + "learning_rate": 1.999447933887199e-05, + "loss": 0.5282, + "step": 770 + }, + { + "epoch": 0.32, + "learning_rate": 1.999442286713808e-05, + "loss": 0.5645, + "step": 771 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994366108123352e-05, + "loss": 0.5232, + "step": 772 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994309061829438e-05, + "loss": 0.537, + "step": 773 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994251728257978e-05, + "loss": 0.6248, + "step": 774 + }, + { + "epoch": 0.32, + "learning_rate": 1.999419410741062e-05, + "loss": 0.5334, + "step": 775 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994136199289028e-05, + "loss": 0.5189, + "step": 776 + }, + { + "epoch": 0.33, + "learning_rate": 1.9994078003894856e-05, + "loss": 0.64, + "step": 777 + }, + { + "epoch": 0.33, + "learning_rate": 1.999401952122978e-05, + "loss": 0.6003, + "step": 778 + }, + { + "epoch": 0.33, + "learning_rate": 1.999396075129548e-05, + "loss": 0.5584, + "step": 779 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993901694093645e-05, + "loss": 0.4774, + "step": 780 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993842349625978e-05, + "loss": 0.5653, + "step": 781 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993782717894178e-05, + "loss": 0.5828, + "step": 782 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993722798899963e-05, + "loss": 0.6803, + "step": 783 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993662592645057e-05, + "loss": 0.487, + "step": 784 + }, + { + "epoch": 0.33, + "learning_rate": 1.999360209913119e-05, + "loss": 0.5226, + "step": 785 + }, + { + "epoch": 0.33, + "learning_rate": 1.999354131836009e-05, + "loss": 0.5717, + "step": 786 + }, + { + "epoch": 0.33, + "learning_rate": 1.999348025033352e-05, + "loss": 0.5765, + "step": 787 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993418895053227e-05, + "loss": 0.4651, + "step": 788 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993357252520975e-05, + "loss": 0.5336, + "step": 789 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993295322738537e-05, + "loss": 0.5058, + "step": 790 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993233105707693e-05, + "loss": 0.5516, + "step": 791 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993170601430233e-05, + "loss": 0.57, + "step": 792 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993107809907946e-05, + "loss": 0.481, + "step": 793 + }, + { + "epoch": 0.33, + "learning_rate": 1.999304473114265e-05, + "loss": 0.6331, + "step": 794 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992981365136145e-05, + "loss": 0.5095, + "step": 795 + }, + { + "epoch": 0.33, + "learning_rate": 1.999291771189026e-05, + "loss": 0.5001, + "step": 796 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992853771406825e-05, + "loss": 0.5308, + "step": 797 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992789543687673e-05, + "loss": 0.5813, + "step": 798 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992725028734655e-05, + "loss": 0.6555, + "step": 799 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992660226549624e-05, + "loss": 0.4987, + "step": 800 + }, + { + "epoch": 0.34, + "learning_rate": 1.999259513713444e-05, + "loss": 0.5393, + "step": 801 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992529760490976e-05, + "loss": 0.554, + "step": 802 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992464096621113e-05, + "loss": 0.484, + "step": 803 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992398145526738e-05, + "loss": 0.5911, + "step": 804 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992331907209742e-05, + "loss": 0.5786, + "step": 805 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992265381672036e-05, + "loss": 0.5447, + "step": 806 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992198568915526e-05, + "loss": 0.5934, + "step": 807 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992131468942135e-05, + "loss": 0.5529, + "step": 808 + }, + { + "epoch": 0.34, + "learning_rate": 1.999206408175379e-05, + "loss": 0.539, + "step": 809 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991996407352432e-05, + "loss": 0.5555, + "step": 810 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991928445740007e-05, + "loss": 0.5249, + "step": 811 + }, + { + "epoch": 0.34, + "learning_rate": 1.999186019691846e-05, + "loss": 0.5433, + "step": 812 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991791660889762e-05, + "loss": 0.6182, + "step": 813 + }, + { + "epoch": 0.34, + "learning_rate": 1.999172283765588e-05, + "loss": 0.5141, + "step": 814 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991653727218787e-05, + "loss": 0.5161, + "step": 815 + }, + { + "epoch": 0.34, + "learning_rate": 1.999158432958048e-05, + "loss": 0.5049, + "step": 816 + }, + { + "epoch": 0.34, + "learning_rate": 1.999151464474294e-05, + "loss": 0.4883, + "step": 817 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991444672708183e-05, + "loss": 0.5744, + "step": 818 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991374413478214e-05, + "loss": 0.5944, + "step": 819 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991303867055056e-05, + "loss": 0.5067, + "step": 820 + }, + { + "epoch": 0.34, + "learning_rate": 1.999123303344073e-05, + "loss": 0.5539, + "step": 821 + }, + { + "epoch": 0.34, + "learning_rate": 1.999116191263728e-05, + "loss": 0.5636, + "step": 822 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991090504646748e-05, + "loss": 0.63, + "step": 823 + }, + { + "epoch": 0.34, + "learning_rate": 1.999101880947118e-05, + "loss": 0.5853, + "step": 824 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990946827112648e-05, + "loss": 0.5308, + "step": 825 + }, + { + "epoch": 0.35, + "learning_rate": 1.999087455757321e-05, + "loss": 0.5656, + "step": 826 + }, + { + "epoch": 0.35, + "learning_rate": 1.999080200085495e-05, + "loss": 0.5307, + "step": 827 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990729156959956e-05, + "loss": 0.4947, + "step": 828 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990656025890315e-05, + "loss": 0.5177, + "step": 829 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990582607648133e-05, + "loss": 0.5555, + "step": 830 + }, + { + "epoch": 0.35, + "learning_rate": 1.999050890223552e-05, + "loss": 0.4916, + "step": 831 + }, + { + "epoch": 0.35, + "learning_rate": 1.999043490965459e-05, + "loss": 0.5135, + "step": 832 + }, + { + "epoch": 0.35, + "learning_rate": 1.999036062990748e-05, + "loss": 0.63, + "step": 833 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990286062996316e-05, + "loss": 0.5116, + "step": 834 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990211208923242e-05, + "loss": 0.5488, + "step": 835 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990136067690416e-05, + "loss": 0.6538, + "step": 836 + }, + { + "epoch": 0.35, + "learning_rate": 1.999006063929999e-05, + "loss": 0.5465, + "step": 837 + }, + { + "epoch": 0.35, + "learning_rate": 1.998998492375414e-05, + "loss": 0.5929, + "step": 838 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989908921055036e-05, + "loss": 0.5986, + "step": 839 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989832631204863e-05, + "loss": 0.613, + "step": 840 + }, + { + "epoch": 0.35, + "learning_rate": 1.998975605420582e-05, + "loss": 0.5219, + "step": 841 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989679190060104e-05, + "loss": 0.5513, + "step": 842 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989602038769922e-05, + "loss": 0.4865, + "step": 843 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989524600337493e-05, + "loss": 0.6124, + "step": 844 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989446874765048e-05, + "loss": 0.5157, + "step": 845 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989368862054814e-05, + "loss": 0.5596, + "step": 846 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989290562209036e-05, + "loss": 0.5604, + "step": 847 + }, + { + "epoch": 0.36, + "learning_rate": 1.9989211975229966e-05, + "loss": 0.5181, + "step": 848 + }, + { + "epoch": 0.36, + "learning_rate": 1.998913310111986e-05, + "loss": 0.5237, + "step": 849 + }, + { + "epoch": 0.36, + "learning_rate": 1.998905393988099e-05, + "loss": 0.5698, + "step": 850 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988974491515628e-05, + "loss": 0.5563, + "step": 851 + }, + { + "epoch": 0.36, + "learning_rate": 1.998889475602606e-05, + "loss": 0.5638, + "step": 852 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988814733414573e-05, + "loss": 0.5097, + "step": 853 + }, + { + "epoch": 0.36, + "learning_rate": 1.998873442368347e-05, + "loss": 0.4871, + "step": 854 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988653826835062e-05, + "loss": 0.4984, + "step": 855 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988572942871666e-05, + "loss": 0.5244, + "step": 856 + }, + { + "epoch": 0.36, + "learning_rate": 1.99884917717956e-05, + "loss": 0.6335, + "step": 857 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988410313609203e-05, + "loss": 0.5901, + "step": 858 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988328568314818e-05, + "loss": 0.54, + "step": 859 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988246535914787e-05, + "loss": 0.5668, + "step": 860 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988164216411474e-05, + "loss": 0.5125, + "step": 861 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988081609807243e-05, + "loss": 0.5225, + "step": 862 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987998716104473e-05, + "loss": 0.5736, + "step": 863 + }, + { + "epoch": 0.36, + "learning_rate": 1.998791553530554e-05, + "loss": 0.4624, + "step": 864 + }, + { + "epoch": 0.36, + "learning_rate": 1.998783206741284e-05, + "loss": 0.5752, + "step": 865 + }, + { + "epoch": 0.36, + "learning_rate": 1.998774831242877e-05, + "loss": 0.5861, + "step": 866 + }, + { + "epoch": 0.36, + "learning_rate": 1.998766427035574e-05, + "loss": 0.5887, + "step": 867 + }, + { + "epoch": 0.36, + "learning_rate": 1.998757994119616e-05, + "loss": 0.5148, + "step": 868 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987495324952458e-05, + "loss": 0.5131, + "step": 869 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987410421627063e-05, + "loss": 0.5495, + "step": 870 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987325231222423e-05, + "loss": 0.5244, + "step": 871 + }, + { + "epoch": 0.37, + "learning_rate": 1.998723975374098e-05, + "loss": 0.5098, + "step": 872 + }, + { + "epoch": 0.37, + "learning_rate": 1.998715398918519e-05, + "loss": 0.5323, + "step": 873 + }, + { + "epoch": 0.37, + "learning_rate": 1.9987067937557527e-05, + "loss": 0.6008, + "step": 874 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986981598860454e-05, + "loss": 0.613, + "step": 875 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986894973096458e-05, + "loss": 0.5766, + "step": 876 + }, + { + "epoch": 0.37, + "learning_rate": 1.998680806026803e-05, + "loss": 0.5188, + "step": 877 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986720860377664e-05, + "loss": 0.5137, + "step": 878 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986633373427868e-05, + "loss": 0.6744, + "step": 879 + }, + { + "epoch": 0.37, + "learning_rate": 1.998654559942116e-05, + "loss": 0.5557, + "step": 880 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986457538360062e-05, + "loss": 0.5192, + "step": 881 + }, + { + "epoch": 0.37, + "learning_rate": 1.99863691902471e-05, + "loss": 0.5091, + "step": 882 + }, + { + "epoch": 0.37, + "learning_rate": 1.998628055508482e-05, + "loss": 0.568, + "step": 883 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986191632875768e-05, + "loss": 0.5641, + "step": 884 + }, + { + "epoch": 0.37, + "learning_rate": 1.99861024236225e-05, + "loss": 0.5126, + "step": 885 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986012927327575e-05, + "loss": 0.5267, + "step": 886 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985923143993572e-05, + "loss": 0.4799, + "step": 887 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985833073623068e-05, + "loss": 0.5367, + "step": 888 + }, + { + "epoch": 0.37, + "learning_rate": 1.998574271621866e-05, + "loss": 0.5242, + "step": 889 + }, + { + "epoch": 0.37, + "learning_rate": 1.998565207178293e-05, + "loss": 0.5858, + "step": 890 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985561140318496e-05, + "loss": 0.5344, + "step": 891 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985469921827968e-05, + "loss": 0.5244, + "step": 892 + }, + { + "epoch": 0.37, + "learning_rate": 1.998537841631397e-05, + "loss": 0.5289, + "step": 893 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985286623779126e-05, + "loss": 0.5555, + "step": 894 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985194544226082e-05, + "loss": 0.5863, + "step": 895 + }, + { + "epoch": 0.38, + "learning_rate": 1.998510217765748e-05, + "loss": 0.4871, + "step": 896 + }, + { + "epoch": 0.38, + "learning_rate": 1.998500952407598e-05, + "loss": 0.5624, + "step": 897 + }, + { + "epoch": 0.38, + "learning_rate": 1.998491658348424e-05, + "loss": 0.5368, + "step": 898 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984823355884926e-05, + "loss": 0.4979, + "step": 899 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984729841280734e-05, + "loss": 0.6757, + "step": 900 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984636039674342e-05, + "loss": 0.5738, + "step": 901 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984541951068445e-05, + "loss": 0.5586, + "step": 902 + }, + { + "epoch": 0.38, + "learning_rate": 1.998444757546575e-05, + "loss": 0.5576, + "step": 903 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984352912868973e-05, + "loss": 0.5864, + "step": 904 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984257963280823e-05, + "loss": 0.5936, + "step": 905 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984162726704047e-05, + "loss": 0.6112, + "step": 906 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984067203141368e-05, + "loss": 0.5352, + "step": 907 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983971392595537e-05, + "loss": 0.6031, + "step": 908 + }, + { + "epoch": 0.38, + "learning_rate": 1.998387529506931e-05, + "loss": 0.5561, + "step": 909 + }, + { + "epoch": 0.38, + "learning_rate": 1.998377891056545e-05, + "loss": 0.5093, + "step": 910 + }, + { + "epoch": 0.38, + "learning_rate": 1.998368223908672e-05, + "loss": 0.6654, + "step": 911 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983585280635902e-05, + "loss": 0.4926, + "step": 912 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983488035215788e-05, + "loss": 0.5659, + "step": 913 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983390502829168e-05, + "loss": 0.5259, + "step": 914 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983292683478844e-05, + "loss": 0.4566, + "step": 915 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983194577167636e-05, + "loss": 0.524, + "step": 916 + }, + { + "epoch": 0.38, + "learning_rate": 1.998309618389836e-05, + "loss": 0.4984, + "step": 917 + }, + { + "epoch": 0.38, + "learning_rate": 1.9982997503673837e-05, + "loss": 0.5231, + "step": 918 + }, + { + "epoch": 0.38, + "learning_rate": 1.9982898536496912e-05, + "loss": 0.5507, + "step": 919 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982799282370432e-05, + "loss": 0.5801, + "step": 920 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982699741297236e-05, + "loss": 0.5397, + "step": 921 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982599913280202e-05, + "loss": 0.5434, + "step": 922 + }, + { + "epoch": 0.39, + "learning_rate": 1.998249979832219e-05, + "loss": 0.507, + "step": 923 + }, + { + "epoch": 0.39, + "learning_rate": 1.998239939642608e-05, + "loss": 0.5271, + "step": 924 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982298707594757e-05, + "loss": 0.5165, + "step": 925 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982197731831115e-05, + "loss": 0.6061, + "step": 926 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982096469138056e-05, + "loss": 0.6236, + "step": 927 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981994919518496e-05, + "loss": 0.5399, + "step": 928 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981893082975348e-05, + "loss": 0.5696, + "step": 929 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981790959511538e-05, + "loss": 0.5813, + "step": 930 + }, + { + "epoch": 0.39, + "learning_rate": 1.998168854913001e-05, + "loss": 0.5144, + "step": 931 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981585851833697e-05, + "loss": 0.4815, + "step": 932 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981482867625557e-05, + "loss": 0.5817, + "step": 933 + }, + { + "epoch": 0.39, + "learning_rate": 1.998137959650855e-05, + "loss": 0.5457, + "step": 934 + }, + { + "epoch": 0.39, + "learning_rate": 1.998127603848565e-05, + "loss": 0.4847, + "step": 935 + }, + { + "epoch": 0.39, + "learning_rate": 1.998117219355982e-05, + "loss": 0.4814, + "step": 936 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981068061734055e-05, + "loss": 0.5047, + "step": 937 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980963643011346e-05, + "loss": 0.5589, + "step": 938 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980858937394695e-05, + "loss": 0.5293, + "step": 939 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980753944887108e-05, + "loss": 0.5762, + "step": 940 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980648665491606e-05, + "loss": 0.557, + "step": 941 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980543099211215e-05, + "loss": 0.577, + "step": 942 + }, + { + "epoch": 0.39, + "learning_rate": 1.998043724604897e-05, + "loss": 0.5131, + "step": 943 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980331106007912e-05, + "loss": 0.5274, + "step": 944 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980224679091092e-05, + "loss": 0.5165, + "step": 945 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980117965301574e-05, + "loss": 0.5571, + "step": 946 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980010964642417e-05, + "loss": 0.5509, + "step": 947 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979903677116705e-05, + "loss": 0.4886, + "step": 948 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979796102727513e-05, + "loss": 0.5009, + "step": 949 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979688241477938e-05, + "loss": 0.5245, + "step": 950 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979580093371084e-05, + "loss": 0.5203, + "step": 951 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979471658410055e-05, + "loss": 0.5339, + "step": 952 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979362936597967e-05, + "loss": 0.6407, + "step": 953 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979253927937948e-05, + "loss": 0.5562, + "step": 954 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979144632433132e-05, + "loss": 0.5537, + "step": 955 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979035050086656e-05, + "loss": 0.537, + "step": 956 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978925180901675e-05, + "loss": 0.5218, + "step": 957 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978815024881342e-05, + "loss": 0.619, + "step": 958 + }, + { + "epoch": 0.4, + "learning_rate": 1.997870458202883e-05, + "loss": 0.5215, + "step": 959 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978593852347306e-05, + "loss": 0.5597, + "step": 960 + }, + { + "epoch": 0.4, + "learning_rate": 1.997848283583996e-05, + "loss": 0.4835, + "step": 961 + }, + { + "epoch": 0.4, + "learning_rate": 1.997837153250998e-05, + "loss": 0.5569, + "step": 962 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978259942360562e-05, + "loss": 0.5173, + "step": 963 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978148065394917e-05, + "loss": 0.5005, + "step": 964 + }, + { + "epoch": 0.4, + "learning_rate": 1.997803590161626e-05, + "loss": 0.6021, + "step": 965 + }, + { + "epoch": 0.4, + "learning_rate": 1.9977923451027816e-05, + "loss": 0.5151, + "step": 966 + }, + { + "epoch": 0.4, + "learning_rate": 1.9977810713632815e-05, + "loss": 0.5957, + "step": 967 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977697689434504e-05, + "loss": 0.5009, + "step": 968 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977584378436123e-05, + "loss": 0.5259, + "step": 969 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977470780640934e-05, + "loss": 0.4887, + "step": 970 + }, + { + "epoch": 0.41, + "learning_rate": 1.99773568960522e-05, + "loss": 0.5379, + "step": 971 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977242724673198e-05, + "loss": 0.6189, + "step": 972 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977128266507203e-05, + "loss": 0.5615, + "step": 973 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977013521557516e-05, + "loss": 0.5451, + "step": 974 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976898489827423e-05, + "loss": 0.5693, + "step": 975 + }, + { + "epoch": 0.41, + "learning_rate": 1.997678317132024e-05, + "loss": 0.5098, + "step": 976 + }, + { + "epoch": 0.41, + "learning_rate": 1.997666756603928e-05, + "loss": 0.5261, + "step": 977 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976551673987857e-05, + "loss": 0.525, + "step": 978 + }, + { + "epoch": 0.41, + "learning_rate": 1.997643549516931e-05, + "loss": 0.6046, + "step": 979 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976319029586982e-05, + "loss": 0.572, + "step": 980 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976202277244216e-05, + "loss": 0.5908, + "step": 981 + }, + { + "epoch": 0.41, + "learning_rate": 1.997608523814437e-05, + "loss": 0.5843, + "step": 982 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975967912290805e-05, + "loss": 0.5956, + "step": 983 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975850299686894e-05, + "loss": 0.5536, + "step": 984 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975732400336018e-05, + "loss": 0.5441, + "step": 985 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975614214241562e-05, + "loss": 0.5195, + "step": 986 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975495741406934e-05, + "loss": 0.5481, + "step": 987 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975376981835533e-05, + "loss": 0.5622, + "step": 988 + }, + { + "epoch": 0.41, + "learning_rate": 1.997525793553077e-05, + "loss": 0.4714, + "step": 989 + }, + { + "epoch": 0.41, + "learning_rate": 1.997513860249607e-05, + "loss": 0.5424, + "step": 990 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975018982734863e-05, + "loss": 0.5466, + "step": 991 + }, + { + "epoch": 0.42, + "learning_rate": 1.997489907625059e-05, + "loss": 0.5853, + "step": 992 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974778883046688e-05, + "loss": 0.4975, + "step": 993 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974658403126624e-05, + "loss": 0.4859, + "step": 994 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974537636493853e-05, + "loss": 0.4968, + "step": 995 + }, + { + "epoch": 0.42, + "learning_rate": 1.997441658315185e-05, + "loss": 0.5131, + "step": 996 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974295243104093e-05, + "loss": 0.5391, + "step": 997 + }, + { + "epoch": 0.42, + "learning_rate": 1.997417361635407e-05, + "loss": 0.4989, + "step": 998 + }, + { + "epoch": 0.42, + "learning_rate": 1.997405170290528e-05, + "loss": 0.5016, + "step": 999 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973929502761222e-05, + "loss": 0.5196, + "step": 1000 + }, + { + "epoch": 0.42, + "learning_rate": 1.997380701592541e-05, + "loss": 0.5563, + "step": 1001 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973684242401366e-05, + "loss": 0.5229, + "step": 1002 + }, + { + "epoch": 0.42, + "learning_rate": 1.997356118219262e-05, + "loss": 0.4715, + "step": 1003 + }, + { + "epoch": 0.42, + "learning_rate": 1.997343783530271e-05, + "loss": 0.5847, + "step": 1004 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973314201735176e-05, + "loss": 0.6229, + "step": 1005 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973190281493578e-05, + "loss": 0.4641, + "step": 1006 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973066074581475e-05, + "loss": 0.4784, + "step": 1007 + }, + { + "epoch": 0.42, + "learning_rate": 1.997294158100244e-05, + "loss": 0.4818, + "step": 1008 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972816800760046e-05, + "loss": 0.4652, + "step": 1009 + }, + { + "epoch": 0.42, + "learning_rate": 1.997269173385788e-05, + "loss": 0.5223, + "step": 1010 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972566380299544e-05, + "loss": 0.5002, + "step": 1011 + }, + { + "epoch": 0.42, + "learning_rate": 1.997244074008864e-05, + "loss": 0.487, + "step": 1012 + }, + { + "epoch": 0.42, + "learning_rate": 1.997231481322877e-05, + "loss": 0.5619, + "step": 1013 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972188599723564e-05, + "loss": 0.5361, + "step": 1014 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972062099576642e-05, + "loss": 0.524, + "step": 1015 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971935312791646e-05, + "loss": 0.4828, + "step": 1016 + }, + { + "epoch": 0.43, + "learning_rate": 1.997180823937222e-05, + "loss": 0.5571, + "step": 1017 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971680879322012e-05, + "loss": 0.528, + "step": 1018 + }, + { + "epoch": 0.43, + "learning_rate": 1.997155323264469e-05, + "loss": 0.5309, + "step": 1019 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971425299343915e-05, + "loss": 0.4413, + "step": 1020 + }, + { + "epoch": 0.43, + "learning_rate": 1.997129707942337e-05, + "loss": 0.4623, + "step": 1021 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971168572886738e-05, + "loss": 0.4915, + "step": 1022 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971039779737717e-05, + "loss": 0.5633, + "step": 1023 + }, + { + "epoch": 0.43, + "learning_rate": 1.997091069998e-05, + "loss": 0.577, + "step": 1024 + }, + { + "epoch": 0.43, + "learning_rate": 1.9970781333617303e-05, + "loss": 0.5143, + "step": 1025 + }, + { + "epoch": 0.43, + "learning_rate": 1.997065168065335e-05, + "loss": 0.5277, + "step": 1026 + }, + { + "epoch": 0.43, + "learning_rate": 1.997052174109186e-05, + "loss": 0.5045, + "step": 1027 + }, + { + "epoch": 0.43, + "learning_rate": 1.997039151493657e-05, + "loss": 0.5114, + "step": 1028 + }, + { + "epoch": 0.43, + "learning_rate": 1.997026100219122e-05, + "loss": 0.4649, + "step": 1029 + }, + { + "epoch": 0.43, + "learning_rate": 1.997013020285957e-05, + "loss": 0.4951, + "step": 1030 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969999116945374e-05, + "loss": 0.5396, + "step": 1031 + }, + { + "epoch": 0.43, + "learning_rate": 1.99698677444524e-05, + "loss": 0.5606, + "step": 1032 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969736085384422e-05, + "loss": 0.5713, + "step": 1033 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969604139745227e-05, + "loss": 0.5012, + "step": 1034 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969471907538615e-05, + "loss": 0.6019, + "step": 1035 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969339388768372e-05, + "loss": 0.5543, + "step": 1036 + }, + { + "epoch": 0.43, + "learning_rate": 1.996920658343832e-05, + "loss": 0.5305, + "step": 1037 + }, + { + "epoch": 0.43, + "learning_rate": 1.996907349155227e-05, + "loss": 0.5374, + "step": 1038 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968940113114048e-05, + "loss": 0.5252, + "step": 1039 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968806448127488e-05, + "loss": 0.5686, + "step": 1040 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968672496596434e-05, + "loss": 0.5571, + "step": 1041 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968538258524736e-05, + "loss": 0.5308, + "step": 1042 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968403733916246e-05, + "loss": 0.5229, + "step": 1043 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968268922774842e-05, + "loss": 0.5767, + "step": 1044 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968133825104396e-05, + "loss": 0.3965, + "step": 1045 + }, + { + "epoch": 0.44, + "learning_rate": 1.996799844090878e-05, + "loss": 0.6016, + "step": 1046 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967862770191902e-05, + "loss": 0.5493, + "step": 1047 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967726812957652e-05, + "loss": 0.5211, + "step": 1048 + }, + { + "epoch": 0.44, + "learning_rate": 1.996759056920994e-05, + "loss": 0.6184, + "step": 1049 + }, + { + "epoch": 0.44, + "learning_rate": 1.996745403895268e-05, + "loss": 0.5019, + "step": 1050 + }, + { + "epoch": 0.44, + "learning_rate": 1.99673172221898e-05, + "loss": 0.549, + "step": 1051 + }, + { + "epoch": 0.44, + "learning_rate": 1.996718011892523e-05, + "loss": 0.484, + "step": 1052 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967042729162914e-05, + "loss": 0.5575, + "step": 1053 + }, + { + "epoch": 0.44, + "learning_rate": 1.99669050529068e-05, + "loss": 0.5138, + "step": 1054 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966767090160843e-05, + "loss": 0.4945, + "step": 1055 + }, + { + "epoch": 0.44, + "learning_rate": 1.996662884092901e-05, + "loss": 0.4617, + "step": 1056 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966490305215277e-05, + "loss": 0.5356, + "step": 1057 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966351483023627e-05, + "loss": 0.4732, + "step": 1058 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966212374358044e-05, + "loss": 0.4211, + "step": 1059 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966072979222527e-05, + "loss": 0.5779, + "step": 1060 + }, + { + "epoch": 0.44, + "learning_rate": 1.996593329762109e-05, + "loss": 0.5561, + "step": 1061 + }, + { + "epoch": 0.44, + "learning_rate": 1.9965793329557743e-05, + "loss": 0.5145, + "step": 1062 + }, + { + "epoch": 0.45, + "learning_rate": 1.996565307503651e-05, + "loss": 0.5379, + "step": 1063 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965512534061426e-05, + "loss": 0.446, + "step": 1064 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965371706636522e-05, + "loss": 0.5324, + "step": 1065 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965230592765856e-05, + "loss": 0.4506, + "step": 1066 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965089192453475e-05, + "loss": 0.5063, + "step": 1067 + }, + { + "epoch": 0.45, + "learning_rate": 1.996494750570345e-05, + "loss": 0.5347, + "step": 1068 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964805532519852e-05, + "loss": 0.5617, + "step": 1069 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964663272906763e-05, + "loss": 0.4943, + "step": 1070 + }, + { + "epoch": 0.45, + "learning_rate": 1.996452072686827e-05, + "loss": 0.5119, + "step": 1071 + }, + { + "epoch": 0.45, + "learning_rate": 1.996437789440847e-05, + "loss": 0.5418, + "step": 1072 + }, + { + "epoch": 0.45, + "learning_rate": 1.996423477553147e-05, + "loss": 0.5345, + "step": 1073 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964091370241384e-05, + "loss": 0.4945, + "step": 1074 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963947678542335e-05, + "loss": 0.5395, + "step": 1075 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963803700438448e-05, + "loss": 0.5527, + "step": 1076 + }, + { + "epoch": 0.45, + "learning_rate": 1.996365943593387e-05, + "loss": 0.4826, + "step": 1077 + }, + { + "epoch": 0.45, + "learning_rate": 1.996351488503274e-05, + "loss": 0.5785, + "step": 1078 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963370047739218e-05, + "loss": 0.5382, + "step": 1079 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963224924057463e-05, + "loss": 0.5902, + "step": 1080 + }, + { + "epoch": 0.45, + "learning_rate": 1.996307951399165e-05, + "loss": 0.5723, + "step": 1081 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962933817545957e-05, + "loss": 0.5868, + "step": 1082 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962787834724572e-05, + "loss": 0.6129, + "step": 1083 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962641565531694e-05, + "loss": 0.501, + "step": 1084 + }, + { + "epoch": 0.45, + "learning_rate": 1.996249500997152e-05, + "loss": 0.5659, + "step": 1085 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962348168048272e-05, + "loss": 0.556, + "step": 1086 + }, + { + "epoch": 0.46, + "learning_rate": 1.9962201039766163e-05, + "loss": 0.556, + "step": 1087 + }, + { + "epoch": 0.46, + "learning_rate": 1.996205362512943e-05, + "loss": 0.5758, + "step": 1088 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961905924142303e-05, + "loss": 0.4728, + "step": 1089 + }, + { + "epoch": 0.46, + "learning_rate": 1.996175793680903e-05, + "loss": 0.5484, + "step": 1090 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961609663133865e-05, + "loss": 0.5067, + "step": 1091 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961461103121073e-05, + "loss": 0.5013, + "step": 1092 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961312256774915e-05, + "loss": 0.5285, + "step": 1093 + }, + { + "epoch": 0.46, + "learning_rate": 1.996116312409968e-05, + "loss": 0.5439, + "step": 1094 + }, + { + "epoch": 0.46, + "learning_rate": 1.996101370509965e-05, + "loss": 0.5521, + "step": 1095 + }, + { + "epoch": 0.46, + "learning_rate": 1.996086399977912e-05, + "loss": 0.5894, + "step": 1096 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960714008142394e-05, + "loss": 0.4348, + "step": 1097 + }, + { + "epoch": 0.46, + "learning_rate": 1.996056373019378e-05, + "loss": 0.5884, + "step": 1098 + }, + { + "epoch": 0.46, + "learning_rate": 1.99604131659376e-05, + "loss": 0.4819, + "step": 1099 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960262315378188e-05, + "loss": 0.4674, + "step": 1100 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960111178519873e-05, + "loss": 0.5385, + "step": 1101 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959959755366997e-05, + "loss": 0.5105, + "step": 1102 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959808045923917e-05, + "loss": 0.4483, + "step": 1103 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959656050194994e-05, + "loss": 0.5014, + "step": 1104 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959503768184597e-05, + "loss": 0.5339, + "step": 1105 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959351199897102e-05, + "loss": 0.4732, + "step": 1106 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959198345336893e-05, + "loss": 0.5171, + "step": 1107 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959045204508366e-05, + "loss": 0.5483, + "step": 1108 + }, + { + "epoch": 0.46, + "learning_rate": 1.9958891777415917e-05, + "loss": 0.5923, + "step": 1109 + }, + { + "epoch": 0.46, + "learning_rate": 1.9958738064063968e-05, + "loss": 0.5747, + "step": 1110 + }, + { + "epoch": 0.47, + "learning_rate": 1.995858406445693e-05, + "loss": 0.5271, + "step": 1111 + }, + { + "epoch": 0.47, + "learning_rate": 1.9958429778599225e-05, + "loss": 0.5466, + "step": 1112 + }, + { + "epoch": 0.47, + "learning_rate": 1.99582752064953e-05, + "loss": 0.4861, + "step": 1113 + }, + { + "epoch": 0.47, + "learning_rate": 1.9958120348149585e-05, + "loss": 0.5899, + "step": 1114 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957965203566542e-05, + "loss": 0.4815, + "step": 1115 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957809772750623e-05, + "loss": 0.5185, + "step": 1116 + }, + { + "epoch": 0.47, + "learning_rate": 1.99576540557063e-05, + "loss": 0.4802, + "step": 1117 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957498052438046e-05, + "loss": 0.5159, + "step": 1118 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957341762950346e-05, + "loss": 0.6096, + "step": 1119 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957185187247697e-05, + "loss": 0.5007, + "step": 1120 + }, + { + "epoch": 0.47, + "learning_rate": 1.995702832533459e-05, + "loss": 0.5597, + "step": 1121 + }, + { + "epoch": 0.47, + "learning_rate": 1.995687117721554e-05, + "loss": 0.4951, + "step": 1122 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956713742895068e-05, + "loss": 0.4691, + "step": 1123 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956556022377693e-05, + "loss": 0.5219, + "step": 1124 + }, + { + "epoch": 0.47, + "learning_rate": 1.995639801566795e-05, + "loss": 0.545, + "step": 1125 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956239722770382e-05, + "loss": 0.4985, + "step": 1126 + }, + { + "epoch": 0.47, + "learning_rate": 1.995608114368954e-05, + "loss": 0.5149, + "step": 1127 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955922278429975e-05, + "loss": 0.5308, + "step": 1128 + }, + { + "epoch": 0.47, + "learning_rate": 1.995576312699626e-05, + "loss": 0.5482, + "step": 1129 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955603689392974e-05, + "loss": 0.5766, + "step": 1130 + }, + { + "epoch": 0.47, + "learning_rate": 1.995544396562469e-05, + "loss": 0.4821, + "step": 1131 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955283955696004e-05, + "loss": 0.4794, + "step": 1132 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955123659611516e-05, + "loss": 0.5762, + "step": 1133 + }, + { + "epoch": 0.47, + "learning_rate": 1.995496307737583e-05, + "loss": 0.5338, + "step": 1134 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954802208993565e-05, + "loss": 0.574, + "step": 1135 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954641054469347e-05, + "loss": 0.6228, + "step": 1136 + }, + { + "epoch": 0.48, + "learning_rate": 1.99544796138078e-05, + "loss": 0.5616, + "step": 1137 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954317887013576e-05, + "loss": 0.5199, + "step": 1138 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954155874091313e-05, + "loss": 0.5464, + "step": 1139 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953993575045674e-05, + "loss": 0.5484, + "step": 1140 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953830989881324e-05, + "loss": 0.5235, + "step": 1141 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953668118602937e-05, + "loss": 0.5152, + "step": 1142 + }, + { + "epoch": 0.48, + "learning_rate": 1.995350496121519e-05, + "loss": 0.5461, + "step": 1143 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953341517722773e-05, + "loss": 0.4547, + "step": 1144 + }, + { + "epoch": 0.48, + "learning_rate": 1.995317778813039e-05, + "loss": 0.5252, + "step": 1145 + }, + { + "epoch": 0.48, + "learning_rate": 1.995301377244274e-05, + "loss": 0.5046, + "step": 1146 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952849470664547e-05, + "loss": 0.4885, + "step": 1147 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952684882800525e-05, + "loss": 0.5105, + "step": 1148 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952520008855405e-05, + "loss": 0.55, + "step": 1149 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952354848833932e-05, + "loss": 0.5413, + "step": 1150 + }, + { + "epoch": 0.48, + "learning_rate": 1.995218940274085e-05, + "loss": 0.552, + "step": 1151 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952023670580915e-05, + "loss": 0.6011, + "step": 1152 + }, + { + "epoch": 0.48, + "learning_rate": 1.995185765235889e-05, + "loss": 0.5404, + "step": 1153 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951691348079553e-05, + "loss": 0.4995, + "step": 1154 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951524757747675e-05, + "loss": 0.5654, + "step": 1155 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951357881368047e-05, + "loss": 0.5018, + "step": 1156 + }, + { + "epoch": 0.48, + "learning_rate": 1.995119071894547e-05, + "loss": 0.5189, + "step": 1157 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951023270484748e-05, + "loss": 0.5147, + "step": 1158 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950855535990688e-05, + "loss": 0.577, + "step": 1159 + }, + { + "epoch": 0.49, + "learning_rate": 1.995068751546812e-05, + "loss": 0.4954, + "step": 1160 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950519208921866e-05, + "loss": 0.5634, + "step": 1161 + }, + { + "epoch": 0.49, + "learning_rate": 1.995035061635677e-05, + "loss": 0.4313, + "step": 1162 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950181737777673e-05, + "loss": 0.4947, + "step": 1163 + }, + { + "epoch": 0.49, + "learning_rate": 1.995001257318943e-05, + "loss": 0.467, + "step": 1164 + }, + { + "epoch": 0.49, + "learning_rate": 1.994984312259691e-05, + "loss": 0.4954, + "step": 1165 + }, + { + "epoch": 0.49, + "learning_rate": 1.9949673386004975e-05, + "loss": 0.5324, + "step": 1166 + }, + { + "epoch": 0.49, + "learning_rate": 1.994950336341851e-05, + "loss": 0.4987, + "step": 1167 + }, + { + "epoch": 0.49, + "learning_rate": 1.9949333054842398e-05, + "loss": 0.5435, + "step": 1168 + }, + { + "epoch": 0.49, + "learning_rate": 1.994916246028154e-05, + "loss": 0.4749, + "step": 1169 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948991579740832e-05, + "loss": 0.4948, + "step": 1170 + }, + { + "epoch": 0.49, + "learning_rate": 1.994882041322519e-05, + "loss": 0.5437, + "step": 1171 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948648960739536e-05, + "loss": 0.5135, + "step": 1172 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948477222288792e-05, + "loss": 0.5212, + "step": 1173 + }, + { + "epoch": 0.49, + "learning_rate": 1.99483051978779e-05, + "loss": 0.4861, + "step": 1174 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948132887511805e-05, + "loss": 0.5396, + "step": 1175 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947960291195457e-05, + "loss": 0.5566, + "step": 1176 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947787408933816e-05, + "loss": 0.653, + "step": 1177 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947614240731856e-05, + "loss": 0.5085, + "step": 1178 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947440786594552e-05, + "loss": 0.4604, + "step": 1179 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947267046526887e-05, + "loss": 0.538, + "step": 1180 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947093020533863e-05, + "loss": 0.479, + "step": 1181 + }, + { + "epoch": 0.49, + "learning_rate": 1.9946918708620472e-05, + "loss": 0.5215, + "step": 1182 + }, + { + "epoch": 0.5, + "learning_rate": 1.994674411079173e-05, + "loss": 0.49, + "step": 1183 + }, + { + "epoch": 0.5, + "learning_rate": 1.994656922705266e-05, + "loss": 0.5296, + "step": 1184 + }, + { + "epoch": 0.5, + "learning_rate": 1.9946394057408278e-05, + "loss": 0.4532, + "step": 1185 + }, + { + "epoch": 0.5, + "learning_rate": 1.9946218601863626e-05, + "loss": 0.4952, + "step": 1186 + }, + { + "epoch": 0.5, + "learning_rate": 1.994604286042375e-05, + "loss": 0.5172, + "step": 1187 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945866833093694e-05, + "loss": 0.4908, + "step": 1188 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945690519878523e-05, + "loss": 0.6495, + "step": 1189 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945513920783303e-05, + "loss": 0.5633, + "step": 1190 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945337035813114e-05, + "loss": 0.4542, + "step": 1191 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945159864973035e-05, + "loss": 0.5391, + "step": 1192 + }, + { + "epoch": 0.5, + "learning_rate": 1.994498240826816e-05, + "loss": 0.5666, + "step": 1193 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944804665703593e-05, + "loss": 0.5144, + "step": 1194 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944626637284438e-05, + "loss": 0.4959, + "step": 1195 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944448323015818e-05, + "loss": 0.5488, + "step": 1196 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944269722902854e-05, + "loss": 0.4746, + "step": 1197 + }, + { + "epoch": 0.5, + "learning_rate": 1.994409083695068e-05, + "loss": 0.5414, + "step": 1198 + }, + { + "epoch": 0.5, + "learning_rate": 1.994391166516444e-05, + "loss": 0.5357, + "step": 1199 + }, + { + "epoch": 0.5, + "learning_rate": 1.9943732207549287e-05, + "loss": 0.5277, + "step": 1200 + }, + { + "epoch": 0.5, + "learning_rate": 1.994355246411037e-05, + "loss": 0.5926, + "step": 1201 + }, + { + "epoch": 0.5, + "learning_rate": 1.9943372434852864e-05, + "loss": 0.5177, + "step": 1202 + }, + { + "epoch": 0.5, + "learning_rate": 1.994319211978194e-05, + "loss": 0.5034, + "step": 1203 + }, + { + "epoch": 0.5, + "learning_rate": 1.994301151890278e-05, + "loss": 0.5085, + "step": 1204 + }, + { + "epoch": 0.5, + "learning_rate": 1.994283063222058e-05, + "loss": 0.6082, + "step": 1205 + }, + { + "epoch": 0.5, + "learning_rate": 1.9942649459740537e-05, + "loss": 0.4875, + "step": 1206 + }, + { + "epoch": 0.51, + "learning_rate": 1.9942468001467857e-05, + "loss": 0.5117, + "step": 1207 + }, + { + "epoch": 0.51, + "learning_rate": 1.994228625740776e-05, + "loss": 0.453, + "step": 1208 + }, + { + "epoch": 0.51, + "learning_rate": 1.994210422756546e-05, + "loss": 0.509, + "step": 1209 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941921911946202e-05, + "loss": 0.5297, + "step": 1210 + }, + { + "epoch": 0.51, + "learning_rate": 1.994173931055522e-05, + "loss": 0.5244, + "step": 1211 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941556423397762e-05, + "loss": 0.4848, + "step": 1212 + }, + { + "epoch": 0.51, + "learning_rate": 1.994137325047909e-05, + "loss": 0.5528, + "step": 1213 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941189791804466e-05, + "loss": 0.5117, + "step": 1214 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941006047379158e-05, + "loss": 0.4961, + "step": 1215 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940822017208458e-05, + "loss": 0.5069, + "step": 1216 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940637701297647e-05, + "loss": 0.4534, + "step": 1217 + }, + { + "epoch": 0.51, + "learning_rate": 1.994045309965203e-05, + "loss": 0.5828, + "step": 1218 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940268212276904e-05, + "loss": 0.4312, + "step": 1219 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940083039177594e-05, + "loss": 0.4555, + "step": 1220 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939897580359417e-05, + "loss": 0.5016, + "step": 1221 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939711835827703e-05, + "loss": 0.5193, + "step": 1222 + }, + { + "epoch": 0.51, + "learning_rate": 1.993952580558779e-05, + "loss": 0.4985, + "step": 1223 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939339489645034e-05, + "loss": 0.4723, + "step": 1224 + }, + { + "epoch": 0.51, + "learning_rate": 1.993915288800478e-05, + "loss": 0.4698, + "step": 1225 + }, + { + "epoch": 0.51, + "learning_rate": 1.99389660006724e-05, + "loss": 0.4951, + "step": 1226 + }, + { + "epoch": 0.51, + "learning_rate": 1.9938778827653256e-05, + "loss": 0.4973, + "step": 1227 + }, + { + "epoch": 0.51, + "learning_rate": 1.993859136895274e-05, + "loss": 0.5482, + "step": 1228 + }, + { + "epoch": 0.51, + "learning_rate": 1.993840362457623e-05, + "loss": 0.5661, + "step": 1229 + }, + { + "epoch": 0.51, + "learning_rate": 1.993821559452913e-05, + "loss": 0.5084, + "step": 1230 + }, + { + "epoch": 0.52, + "learning_rate": 1.993802727881684e-05, + "loss": 0.6094, + "step": 1231 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937838677444776e-05, + "loss": 0.4842, + "step": 1232 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937649790418357e-05, + "loss": 0.4732, + "step": 1233 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937460617743013e-05, + "loss": 0.5847, + "step": 1234 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937271159424185e-05, + "loss": 0.5418, + "step": 1235 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937081415467313e-05, + "loss": 0.4498, + "step": 1236 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936891385877854e-05, + "loss": 0.5086, + "step": 1237 + }, + { + "epoch": 0.52, + "learning_rate": 1.993670107066127e-05, + "loss": 0.4876, + "step": 1238 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936510469823028e-05, + "loss": 0.5137, + "step": 1239 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936319583368616e-05, + "loss": 0.4546, + "step": 1240 + }, + { + "epoch": 0.52, + "learning_rate": 1.993612841130351e-05, + "loss": 0.5644, + "step": 1241 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935936953633214e-05, + "loss": 0.5763, + "step": 1242 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935745210363224e-05, + "loss": 0.5405, + "step": 1243 + }, + { + "epoch": 0.52, + "learning_rate": 1.993555318149906e-05, + "loss": 0.4923, + "step": 1244 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935360867046228e-05, + "loss": 0.5439, + "step": 1245 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935168267010268e-05, + "loss": 0.4501, + "step": 1246 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934975381396716e-05, + "loss": 0.5177, + "step": 1247 + }, + { + "epoch": 0.52, + "learning_rate": 1.993478221021111e-05, + "loss": 0.5299, + "step": 1248 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934588753459006e-05, + "loss": 0.5357, + "step": 1249 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934395011145967e-05, + "loss": 0.4541, + "step": 1250 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934200983277553e-05, + "loss": 0.464, + "step": 1251 + }, + { + "epoch": 0.52, + "learning_rate": 1.993400666985935e-05, + "loss": 0.5593, + "step": 1252 + }, + { + "epoch": 0.52, + "learning_rate": 1.993381207089694e-05, + "loss": 0.516, + "step": 1253 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933617186395917e-05, + "loss": 0.6846, + "step": 1254 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933422016361886e-05, + "loss": 0.5269, + "step": 1255 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933226560800452e-05, + "loss": 0.5366, + "step": 1256 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933030819717235e-05, + "loss": 0.4933, + "step": 1257 + }, + { + "epoch": 0.53, + "learning_rate": 1.993283479311786e-05, + "loss": 0.5686, + "step": 1258 + }, + { + "epoch": 0.53, + "learning_rate": 1.993263848100796e-05, + "loss": 0.4922, + "step": 1259 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932441883393186e-05, + "loss": 0.546, + "step": 1260 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932245000279184e-05, + "loss": 0.5403, + "step": 1261 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932047831671615e-05, + "loss": 0.4769, + "step": 1262 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931850377576142e-05, + "loss": 0.5271, + "step": 1263 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931652637998443e-05, + "loss": 0.5356, + "step": 1264 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931454612944203e-05, + "loss": 0.5345, + "step": 1265 + }, + { + "epoch": 0.53, + "learning_rate": 1.993125630241911e-05, + "loss": 0.5233, + "step": 1266 + }, + { + "epoch": 0.53, + "learning_rate": 1.993105770642887e-05, + "loss": 0.4715, + "step": 1267 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930858824979192e-05, + "loss": 0.5269, + "step": 1268 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930659658075784e-05, + "loss": 0.5421, + "step": 1269 + }, + { + "epoch": 0.53, + "learning_rate": 1.993046020572438e-05, + "loss": 0.5315, + "step": 1270 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930260467930706e-05, + "loss": 0.6047, + "step": 1271 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930060444700508e-05, + "loss": 0.4419, + "step": 1272 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929860136039537e-05, + "loss": 0.6234, + "step": 1273 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929659541953548e-05, + "loss": 0.4614, + "step": 1274 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929458662448305e-05, + "loss": 0.5827, + "step": 1275 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929257497529584e-05, + "loss": 0.4641, + "step": 1276 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929056047203166e-05, + "loss": 0.4859, + "step": 1277 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928854311474844e-05, + "loss": 0.4513, + "step": 1278 + }, + { + "epoch": 0.54, + "learning_rate": 1.992865229035042e-05, + "loss": 0.6051, + "step": 1279 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928449983835688e-05, + "loss": 0.5138, + "step": 1280 + }, + { + "epoch": 0.54, + "learning_rate": 1.992824739193648e-05, + "loss": 0.5685, + "step": 1281 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928044514658604e-05, + "loss": 0.5332, + "step": 1282 + }, + { + "epoch": 0.54, + "learning_rate": 1.99278413520079e-05, + "loss": 0.5414, + "step": 1283 + }, + { + "epoch": 0.54, + "learning_rate": 1.992763790399021e-05, + "loss": 0.6006, + "step": 1284 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927434170611373e-05, + "loss": 0.519, + "step": 1285 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927230151877253e-05, + "loss": 0.5043, + "step": 1286 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927025847793714e-05, + "loss": 0.5028, + "step": 1287 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926821258366622e-05, + "loss": 0.5702, + "step": 1288 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926616383601862e-05, + "loss": 0.5905, + "step": 1289 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926411223505323e-05, + "loss": 0.5257, + "step": 1290 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926205778082903e-05, + "loss": 0.4561, + "step": 1291 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926000047340507e-05, + "loss": 0.5467, + "step": 1292 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925794031284046e-05, + "loss": 0.5829, + "step": 1293 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925587729919447e-05, + "loss": 0.5134, + "step": 1294 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925381143252633e-05, + "loss": 0.5486, + "step": 1295 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925174271289548e-05, + "loss": 0.5339, + "step": 1296 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924967114036133e-05, + "loss": 0.5039, + "step": 1297 + }, + { + "epoch": 0.54, + "learning_rate": 1.992475967149835e-05, + "loss": 0.5455, + "step": 1298 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924551943682153e-05, + "loss": 0.5704, + "step": 1299 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924343930593516e-05, + "loss": 0.5534, + "step": 1300 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924135632238423e-05, + "loss": 0.4944, + "step": 1301 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923927048622857e-05, + "loss": 0.5032, + "step": 1302 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923718179752815e-05, + "loss": 0.557, + "step": 1303 + }, + { + "epoch": 0.55, + "learning_rate": 1.99235090256343e-05, + "loss": 0.4871, + "step": 1304 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923299586273323e-05, + "loss": 0.5246, + "step": 1305 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923089861675906e-05, + "loss": 0.5811, + "step": 1306 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922879851848077e-05, + "loss": 0.5093, + "step": 1307 + }, + { + "epoch": 0.55, + "learning_rate": 1.992266955679587e-05, + "loss": 0.5393, + "step": 1308 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922458976525335e-05, + "loss": 0.533, + "step": 1309 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922248111042518e-05, + "loss": 0.4694, + "step": 1310 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922036960353486e-05, + "loss": 0.5957, + "step": 1311 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921825524464306e-05, + "loss": 0.5424, + "step": 1312 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921613803381055e-05, + "loss": 0.5007, + "step": 1313 + }, + { + "epoch": 0.55, + "learning_rate": 1.992140179710982e-05, + "loss": 0.4663, + "step": 1314 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921189505656696e-05, + "loss": 0.5117, + "step": 1315 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920976929027782e-05, + "loss": 0.4508, + "step": 1316 + }, + { + "epoch": 0.55, + "learning_rate": 1.992076406722919e-05, + "loss": 0.5698, + "step": 1317 + }, + { + "epoch": 0.55, + "learning_rate": 1.992055092026704e-05, + "loss": 0.5356, + "step": 1318 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920337488147453e-05, + "loss": 0.5155, + "step": 1319 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920123770876573e-05, + "loss": 0.6008, + "step": 1320 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919909768460535e-05, + "loss": 0.5054, + "step": 1321 + }, + { + "epoch": 0.55, + "learning_rate": 1.991969548090549e-05, + "loss": 0.4738, + "step": 1322 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919480908217606e-05, + "loss": 0.4882, + "step": 1323 + }, + { + "epoch": 0.55, + "learning_rate": 1.991926605040305e-05, + "loss": 0.5006, + "step": 1324 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919050907467986e-05, + "loss": 0.5821, + "step": 1325 + }, + { + "epoch": 0.56, + "learning_rate": 1.9918835479418607e-05, + "loss": 0.5611, + "step": 1326 + }, + { + "epoch": 0.56, + "learning_rate": 1.9918619766261106e-05, + "loss": 0.5482, + "step": 1327 + }, + { + "epoch": 0.56, + "learning_rate": 1.991840376800168e-05, + "loss": 0.6395, + "step": 1328 + }, + { + "epoch": 0.56, + "learning_rate": 1.991818748464654e-05, + "loss": 0.5458, + "step": 1329 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917970916201904e-05, + "loss": 0.5189, + "step": 1330 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917754062673992e-05, + "loss": 0.5461, + "step": 1331 + }, + { + "epoch": 0.56, + "learning_rate": 1.991753692406904e-05, + "loss": 0.4995, + "step": 1332 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917319500393294e-05, + "loss": 0.609, + "step": 1333 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917101791652997e-05, + "loss": 0.5375, + "step": 1334 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916883797854407e-05, + "loss": 0.4678, + "step": 1335 + }, + { + "epoch": 0.56, + "learning_rate": 1.991666551900379e-05, + "loss": 0.5375, + "step": 1336 + }, + { + "epoch": 0.56, + "learning_rate": 1.991644695510743e-05, + "loss": 0.441, + "step": 1337 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916228106171596e-05, + "loss": 0.5054, + "step": 1338 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916008972202586e-05, + "loss": 0.5001, + "step": 1339 + }, + { + "epoch": 0.56, + "learning_rate": 1.99157895532067e-05, + "loss": 0.574, + "step": 1340 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915569849190243e-05, + "loss": 0.5413, + "step": 1341 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915349860159528e-05, + "loss": 0.5348, + "step": 1342 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915129586120878e-05, + "loss": 0.4805, + "step": 1343 + }, + { + "epoch": 0.56, + "learning_rate": 1.991490902708063e-05, + "loss": 0.3818, + "step": 1344 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914688183045116e-05, + "loss": 0.5991, + "step": 1345 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914467054020692e-05, + "loss": 0.5398, + "step": 1346 + }, + { + "epoch": 0.56, + "learning_rate": 1.991424564001371e-05, + "loss": 0.5381, + "step": 1347 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914023941030534e-05, + "loss": 0.5163, + "step": 1348 + }, + { + "epoch": 0.56, + "learning_rate": 1.9913801957077535e-05, + "loss": 0.5226, + "step": 1349 + }, + { + "epoch": 0.57, + "learning_rate": 1.99135796881611e-05, + "loss": 0.5332, + "step": 1350 + }, + { + "epoch": 0.57, + "learning_rate": 1.991335713428761e-05, + "loss": 0.5834, + "step": 1351 + }, + { + "epoch": 0.57, + "learning_rate": 1.991313429546347e-05, + "loss": 0.5745, + "step": 1352 + }, + { + "epoch": 0.57, + "learning_rate": 1.991291117169508e-05, + "loss": 0.5629, + "step": 1353 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912687762988855e-05, + "loss": 0.4963, + "step": 1354 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912464069351217e-05, + "loss": 0.4239, + "step": 1355 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912240090788595e-05, + "loss": 0.4935, + "step": 1356 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912015827307426e-05, + "loss": 0.53, + "step": 1357 + }, + { + "epoch": 0.57, + "learning_rate": 1.991179127891416e-05, + "loss": 0.469, + "step": 1358 + }, + { + "epoch": 0.57, + "learning_rate": 1.9911566445615253e-05, + "loss": 0.537, + "step": 1359 + }, + { + "epoch": 0.57, + "learning_rate": 1.9911341327417156e-05, + "loss": 0.5467, + "step": 1360 + }, + { + "epoch": 0.57, + "learning_rate": 1.991111592432635e-05, + "loss": 0.4584, + "step": 1361 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910890236349316e-05, + "loss": 0.6264, + "step": 1362 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910664263492533e-05, + "loss": 0.5039, + "step": 1363 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910438005762503e-05, + "loss": 0.4549, + "step": 1364 + }, + { + "epoch": 0.57, + "learning_rate": 1.991021146316572e-05, + "loss": 0.5509, + "step": 1365 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909984635708714e-05, + "loss": 0.4958, + "step": 1366 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909757523397985e-05, + "loss": 0.5492, + "step": 1367 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909530126240074e-05, + "loss": 0.5841, + "step": 1368 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909302444241508e-05, + "loss": 0.545, + "step": 1369 + }, + { + "epoch": 0.57, + "learning_rate": 1.990907447740884e-05, + "loss": 0.4326, + "step": 1370 + }, + { + "epoch": 0.57, + "learning_rate": 1.9908846225748623e-05, + "loss": 0.5381, + "step": 1371 + }, + { + "epoch": 0.57, + "learning_rate": 1.990861768926741e-05, + "loss": 0.4573, + "step": 1372 + }, + { + "epoch": 0.57, + "learning_rate": 1.9908388867971772e-05, + "loss": 0.4962, + "step": 1373 + }, + { + "epoch": 0.58, + "learning_rate": 1.9908159761868292e-05, + "loss": 0.4974, + "step": 1374 + }, + { + "epoch": 0.58, + "learning_rate": 1.990793037096355e-05, + "loss": 0.4819, + "step": 1375 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907700695264144e-05, + "loss": 0.532, + "step": 1376 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907470734776674e-05, + "loss": 0.4813, + "step": 1377 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907240489507748e-05, + "loss": 0.483, + "step": 1378 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907009959463983e-05, + "loss": 0.6402, + "step": 1379 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906779144652014e-05, + "loss": 0.5279, + "step": 1380 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906548045078464e-05, + "loss": 0.5868, + "step": 1381 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906316660749985e-05, + "loss": 0.4664, + "step": 1382 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906084991673224e-05, + "loss": 0.571, + "step": 1383 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905853037854834e-05, + "loss": 0.5611, + "step": 1384 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905620799301495e-05, + "loss": 0.4523, + "step": 1385 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905388276019873e-05, + "loss": 0.6086, + "step": 1386 + }, + { + "epoch": 0.58, + "learning_rate": 1.990515546801666e-05, + "loss": 0.483, + "step": 1387 + }, + { + "epoch": 0.58, + "learning_rate": 1.990492237529854e-05, + "loss": 0.6169, + "step": 1388 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904688997872215e-05, + "loss": 0.4782, + "step": 1389 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904455335744395e-05, + "loss": 0.5676, + "step": 1390 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904221388921794e-05, + "loss": 0.6075, + "step": 1391 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903987157411138e-05, + "loss": 0.4489, + "step": 1392 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903752641219163e-05, + "loss": 0.5618, + "step": 1393 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903517840352605e-05, + "loss": 0.5132, + "step": 1394 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903282754818216e-05, + "loss": 0.4228, + "step": 1395 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903047384622748e-05, + "loss": 0.513, + "step": 1396 + }, + { + "epoch": 0.58, + "learning_rate": 1.9902811729772974e-05, + "loss": 0.5782, + "step": 1397 + }, + { + "epoch": 0.59, + "learning_rate": 1.990257579027566e-05, + "loss": 0.5138, + "step": 1398 + }, + { + "epoch": 0.59, + "learning_rate": 1.99023395661376e-05, + "loss": 0.5087, + "step": 1399 + }, + { + "epoch": 0.59, + "learning_rate": 1.9902103057365573e-05, + "loss": 0.6183, + "step": 1400 + }, + { + "epoch": 0.59, + "learning_rate": 1.990186626396638e-05, + "loss": 0.5741, + "step": 1401 + }, + { + "epoch": 0.59, + "learning_rate": 1.9901629185946827e-05, + "loss": 0.4439, + "step": 1402 + }, + { + "epoch": 0.59, + "learning_rate": 1.990139182331373e-05, + "loss": 0.6, + "step": 1403 + }, + { + "epoch": 0.59, + "learning_rate": 1.9901154176073914e-05, + "loss": 0.5193, + "step": 1404 + }, + { + "epoch": 0.59, + "learning_rate": 1.9900916244234203e-05, + "loss": 0.6317, + "step": 1405 + }, + { + "epoch": 0.59, + "learning_rate": 1.9900678027801445e-05, + "loss": 0.5093, + "step": 1406 + }, + { + "epoch": 0.59, + "learning_rate": 1.990043952678248e-05, + "loss": 0.4804, + "step": 1407 + }, + { + "epoch": 0.59, + "learning_rate": 1.990020074118417e-05, + "loss": 0.5628, + "step": 1408 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899961671013367e-05, + "loss": 0.5073, + "step": 1409 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899722316276958e-05, + "loss": 0.5252, + "step": 1410 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899482676981812e-05, + "loss": 0.5157, + "step": 1411 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899242753134824e-05, + "loss": 0.5309, + "step": 1412 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899002544742886e-05, + "loss": 0.4717, + "step": 1413 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898762051812906e-05, + "loss": 0.4811, + "step": 1414 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898521274351793e-05, + "loss": 0.496, + "step": 1415 + }, + { + "epoch": 0.59, + "learning_rate": 1.989828021236647e-05, + "loss": 0.4783, + "step": 1416 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898038865863864e-05, + "loss": 0.4625, + "step": 1417 + }, + { + "epoch": 0.59, + "learning_rate": 1.9897797234850916e-05, + "loss": 0.4816, + "step": 1418 + }, + { + "epoch": 0.59, + "learning_rate": 1.989755531933457e-05, + "loss": 0.5341, + "step": 1419 + }, + { + "epoch": 0.59, + "learning_rate": 1.989731311932178e-05, + "loss": 0.5596, + "step": 1420 + }, + { + "epoch": 0.59, + "learning_rate": 1.9897070634819506e-05, + "loss": 0.4642, + "step": 1421 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896827865834717e-05, + "loss": 0.4795, + "step": 1422 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896584812374397e-05, + "loss": 0.4566, + "step": 1423 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896341474445526e-05, + "loss": 0.5784, + "step": 1424 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896097852055103e-05, + "loss": 0.5538, + "step": 1425 + }, + { + "epoch": 0.6, + "learning_rate": 1.989585394521013e-05, + "loss": 0.5432, + "step": 1426 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895609753917614e-05, + "loss": 0.4524, + "step": 1427 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895365278184577e-05, + "loss": 0.4955, + "step": 1428 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895120518018046e-05, + "loss": 0.443, + "step": 1429 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894875473425057e-05, + "loss": 0.5583, + "step": 1430 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894630144412656e-05, + "loss": 0.5227, + "step": 1431 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894384530987885e-05, + "loss": 0.5223, + "step": 1432 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894138633157816e-05, + "loss": 0.4718, + "step": 1433 + }, + { + "epoch": 0.6, + "learning_rate": 1.989389245092951e-05, + "loss": 0.4776, + "step": 1434 + }, + { + "epoch": 0.6, + "learning_rate": 1.9893645984310047e-05, + "loss": 0.4808, + "step": 1435 + }, + { + "epoch": 0.6, + "learning_rate": 1.989339923330651e-05, + "loss": 0.5571, + "step": 1436 + }, + { + "epoch": 0.6, + "learning_rate": 1.989315219792599e-05, + "loss": 0.4956, + "step": 1437 + }, + { + "epoch": 0.6, + "learning_rate": 1.989290487817559e-05, + "loss": 0.5159, + "step": 1438 + }, + { + "epoch": 0.6, + "learning_rate": 1.9892657274062415e-05, + "loss": 0.5385, + "step": 1439 + }, + { + "epoch": 0.6, + "learning_rate": 1.9892409385593586e-05, + "loss": 0.6072, + "step": 1440 + }, + { + "epoch": 0.6, + "learning_rate": 1.989216121277623e-05, + "loss": 0.4643, + "step": 1441 + }, + { + "epoch": 0.6, + "learning_rate": 1.9891912755617477e-05, + "loss": 0.4857, + "step": 1442 + }, + { + "epoch": 0.6, + "learning_rate": 1.989166401412447e-05, + "loss": 0.4994, + "step": 1443 + }, + { + "epoch": 0.6, + "learning_rate": 1.9891414988304357e-05, + "loss": 0.4851, + "step": 1444 + }, + { + "epoch": 0.6, + "learning_rate": 1.98911656781643e-05, + "loss": 0.5248, + "step": 1445 + }, + { + "epoch": 0.61, + "learning_rate": 1.9890916083711463e-05, + "loss": 0.5135, + "step": 1446 + }, + { + "epoch": 0.61, + "learning_rate": 1.9890666204953018e-05, + "loss": 0.5424, + "step": 1447 + }, + { + "epoch": 0.61, + "learning_rate": 1.989041604189615e-05, + "loss": 0.4628, + "step": 1448 + }, + { + "epoch": 0.61, + "learning_rate": 1.989016559454805e-05, + "loss": 0.5106, + "step": 1449 + }, + { + "epoch": 0.61, + "learning_rate": 1.988991486291592e-05, + "loss": 0.5695, + "step": 1450 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889663847006956e-05, + "loss": 0.5715, + "step": 1451 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889412546828386e-05, + "loss": 0.4649, + "step": 1452 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889160962387427e-05, + "loss": 0.5909, + "step": 1453 + }, + { + "epoch": 0.61, + "learning_rate": 1.988890909369131e-05, + "loss": 0.5654, + "step": 1454 + }, + { + "epoch": 0.61, + "learning_rate": 1.988865694074728e-05, + "loss": 0.5696, + "step": 1455 + }, + { + "epoch": 0.61, + "learning_rate": 1.988840450356258e-05, + "loss": 0.486, + "step": 1456 + }, + { + "epoch": 0.61, + "learning_rate": 1.9888151782144463e-05, + "loss": 0.4263, + "step": 1457 + }, + { + "epoch": 0.61, + "learning_rate": 1.9887898776500203e-05, + "loss": 0.4926, + "step": 1458 + }, + { + "epoch": 0.61, + "learning_rate": 1.9887645486637063e-05, + "loss": 0.5234, + "step": 1459 + }, + { + "epoch": 0.61, + "learning_rate": 1.988739191256233e-05, + "loss": 0.4934, + "step": 1460 + }, + { + "epoch": 0.61, + "learning_rate": 1.988713805428329e-05, + "loss": 0.5068, + "step": 1461 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886883911807242e-05, + "loss": 0.5427, + "step": 1462 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886629485141488e-05, + "loss": 0.524, + "step": 1463 + }, + { + "epoch": 0.61, + "learning_rate": 1.988637477429334e-05, + "loss": 0.4719, + "step": 1464 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886119779270126e-05, + "loss": 0.5084, + "step": 1465 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885864500079172e-05, + "loss": 0.4963, + "step": 1466 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885608936727814e-05, + "loss": 0.5211, + "step": 1467 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885353089223397e-05, + "loss": 0.4711, + "step": 1468 + }, + { + "epoch": 0.62, + "learning_rate": 1.988509695757328e-05, + "loss": 0.5843, + "step": 1469 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884840541784824e-05, + "loss": 0.5093, + "step": 1470 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884583841865395e-05, + "loss": 0.4973, + "step": 1471 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884326857822378e-05, + "loss": 0.5495, + "step": 1472 + }, + { + "epoch": 0.62, + "learning_rate": 1.988406958966316e-05, + "loss": 0.6042, + "step": 1473 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883812037395126e-05, + "loss": 0.4894, + "step": 1474 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883554201025687e-05, + "loss": 0.5177, + "step": 1475 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883296080562255e-05, + "loss": 0.4454, + "step": 1476 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883037676012245e-05, + "loss": 0.534, + "step": 1477 + }, + { + "epoch": 0.62, + "learning_rate": 1.988277898738309e-05, + "loss": 0.4713, + "step": 1478 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882520014682218e-05, + "loss": 0.5647, + "step": 1479 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882260757917083e-05, + "loss": 0.5868, + "step": 1480 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882001217095127e-05, + "loss": 0.5463, + "step": 1481 + }, + { + "epoch": 0.62, + "learning_rate": 1.9881741392223822e-05, + "loss": 0.5546, + "step": 1482 + }, + { + "epoch": 0.62, + "learning_rate": 1.9881481283310624e-05, + "loss": 0.5102, + "step": 1483 + }, + { + "epoch": 0.62, + "learning_rate": 1.988122089036302e-05, + "loss": 0.4513, + "step": 1484 + }, + { + "epoch": 0.62, + "learning_rate": 1.9880960213388483e-05, + "loss": 0.5327, + "step": 1485 + }, + { + "epoch": 0.62, + "learning_rate": 1.988069925239452e-05, + "loss": 0.5329, + "step": 1486 + }, + { + "epoch": 0.62, + "learning_rate": 1.988043800738862e-05, + "loss": 0.522, + "step": 1487 + }, + { + "epoch": 0.62, + "learning_rate": 1.98801764783783e-05, + "loss": 0.5105, + "step": 1488 + }, + { + "epoch": 0.62, + "learning_rate": 1.987991466537107e-05, + "loss": 0.5621, + "step": 1489 + }, + { + "epoch": 0.62, + "learning_rate": 1.9879652568374467e-05, + "loss": 0.5259, + "step": 1490 + }, + { + "epoch": 0.62, + "learning_rate": 1.987939018739601e-05, + "loss": 0.464, + "step": 1491 + }, + { + "epoch": 0.62, + "learning_rate": 1.9879127522443255e-05, + "loss": 0.5808, + "step": 1492 + }, + { + "epoch": 0.63, + "learning_rate": 1.9878864573523744e-05, + "loss": 0.5933, + "step": 1493 + }, + { + "epoch": 0.63, + "learning_rate": 1.987860134064504e-05, + "loss": 0.4712, + "step": 1494 + }, + { + "epoch": 0.63, + "learning_rate": 1.98783378238147e-05, + "loss": 0.5199, + "step": 1495 + }, + { + "epoch": 0.63, + "learning_rate": 1.9878074023040313e-05, + "loss": 0.5006, + "step": 1496 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877809938329447e-05, + "loss": 0.5374, + "step": 1497 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877545569689705e-05, + "loss": 0.4978, + "step": 1498 + }, + { + "epoch": 0.63, + "learning_rate": 1.987728091712868e-05, + "loss": 0.4904, + "step": 1499 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877015980653978e-05, + "loss": 0.5521, + "step": 1500 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876750760273218e-05, + "loss": 0.6263, + "step": 1501 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876485255994022e-05, + "loss": 0.5446, + "step": 1502 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876219467824018e-05, + "loss": 0.4413, + "step": 1503 + }, + { + "epoch": 0.63, + "learning_rate": 1.987595339577085e-05, + "loss": 0.4664, + "step": 1504 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875687039842168e-05, + "loss": 0.5025, + "step": 1505 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875420400045625e-05, + "loss": 0.5362, + "step": 1506 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875153476388887e-05, + "loss": 0.5398, + "step": 1507 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874886268879625e-05, + "loss": 0.4848, + "step": 1508 + }, + { + "epoch": 0.63, + "learning_rate": 1.987461877752552e-05, + "loss": 0.4712, + "step": 1509 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874351002334257e-05, + "loss": 0.5298, + "step": 1510 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874082943313543e-05, + "loss": 0.5308, + "step": 1511 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873814600471073e-05, + "loss": 0.5389, + "step": 1512 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873545973814564e-05, + "loss": 0.4738, + "step": 1513 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873277063351737e-05, + "loss": 0.5454, + "step": 1514 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873007869090326e-05, + "loss": 0.5225, + "step": 1515 + }, + { + "epoch": 0.63, + "learning_rate": 1.9872738391038063e-05, + "loss": 0.4892, + "step": 1516 + }, + { + "epoch": 0.64, + "learning_rate": 1.9872468629202696e-05, + "loss": 0.5003, + "step": 1517 + }, + { + "epoch": 0.64, + "learning_rate": 1.9872198583591976e-05, + "loss": 0.5106, + "step": 1518 + }, + { + "epoch": 0.64, + "learning_rate": 1.987192825421367e-05, + "loss": 0.6494, + "step": 1519 + }, + { + "epoch": 0.64, + "learning_rate": 1.987165764107555e-05, + "loss": 0.5218, + "step": 1520 + }, + { + "epoch": 0.64, + "learning_rate": 1.9871386744185386e-05, + "loss": 0.5341, + "step": 1521 + }, + { + "epoch": 0.64, + "learning_rate": 1.9871115563550974e-05, + "loss": 0.4962, + "step": 1522 + }, + { + "epoch": 0.64, + "learning_rate": 1.98708440991801e-05, + "loss": 0.5314, + "step": 1523 + }, + { + "epoch": 0.64, + "learning_rate": 1.9870572351080573e-05, + "loss": 0.5039, + "step": 1524 + }, + { + "epoch": 0.64, + "learning_rate": 1.9870300319260204e-05, + "loss": 0.4734, + "step": 1525 + }, + { + "epoch": 0.64, + "learning_rate": 1.987002800372681e-05, + "loss": 0.565, + "step": 1526 + }, + { + "epoch": 0.64, + "learning_rate": 1.986975540448822e-05, + "loss": 0.5463, + "step": 1527 + }, + { + "epoch": 0.64, + "learning_rate": 1.986948252155227e-05, + "loss": 0.504, + "step": 1528 + }, + { + "epoch": 0.64, + "learning_rate": 1.9869209354926802e-05, + "loss": 0.4311, + "step": 1529 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868935904619673e-05, + "loss": 0.4794, + "step": 1530 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868662170638732e-05, + "loss": 0.5152, + "step": 1531 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868388152991858e-05, + "loss": 0.5112, + "step": 1532 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868113851686925e-05, + "loss": 0.5007, + "step": 1533 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867839266731813e-05, + "loss": 0.5184, + "step": 1534 + }, + { + "epoch": 0.64, + "learning_rate": 1.986756439813442e-05, + "loss": 0.4357, + "step": 1535 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867289245902645e-05, + "loss": 0.532, + "step": 1536 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867013810044394e-05, + "loss": 0.5021, + "step": 1537 + }, + { + "epoch": 0.64, + "learning_rate": 1.9866738090567586e-05, + "loss": 0.4668, + "step": 1538 + }, + { + "epoch": 0.64, + "learning_rate": 1.9866462087480153e-05, + "loss": 0.4493, + "step": 1539 + }, + { + "epoch": 0.64, + "learning_rate": 1.986618580079002e-05, + "loss": 0.503, + "step": 1540 + }, + { + "epoch": 0.65, + "learning_rate": 1.986590923050513e-05, + "loss": 0.5003, + "step": 1541 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865632376633433e-05, + "loss": 0.5519, + "step": 1542 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865355239182892e-05, + "loss": 0.4793, + "step": 1543 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865077818161464e-05, + "loss": 0.4704, + "step": 1544 + }, + { + "epoch": 0.65, + "learning_rate": 1.9864800113577133e-05, + "loss": 0.5161, + "step": 1545 + }, + { + "epoch": 0.65, + "learning_rate": 1.9864522125437874e-05, + "loss": 0.4979, + "step": 1546 + }, + { + "epoch": 0.65, + "learning_rate": 1.986424385375168e-05, + "loss": 0.5197, + "step": 1547 + }, + { + "epoch": 0.65, + "learning_rate": 1.986396529852655e-05, + "loss": 0.4897, + "step": 1548 + }, + { + "epoch": 0.65, + "learning_rate": 1.986368645977049e-05, + "loss": 0.4824, + "step": 1549 + }, + { + "epoch": 0.65, + "learning_rate": 1.9863407337491516e-05, + "loss": 0.565, + "step": 1550 + }, + { + "epoch": 0.65, + "learning_rate": 1.986312793169765e-05, + "loss": 0.5372, + "step": 1551 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862848242396927e-05, + "loss": 0.485, + "step": 1552 + }, + { + "epoch": 0.65, + "learning_rate": 1.986256826959738e-05, + "loss": 0.5229, + "step": 1553 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862288013307062e-05, + "loss": 0.583, + "step": 1554 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862007473534026e-05, + "loss": 0.5444, + "step": 1555 + }, + { + "epoch": 0.65, + "learning_rate": 1.9861726650286338e-05, + "loss": 0.5535, + "step": 1556 + }, + { + "epoch": 0.65, + "learning_rate": 1.986144554357207e-05, + "loss": 0.4921, + "step": 1557 + }, + { + "epoch": 0.65, + "learning_rate": 1.9861164153399297e-05, + "loss": 0.5669, + "step": 1558 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860882479776113e-05, + "loss": 0.5139, + "step": 1559 + }, + { + "epoch": 0.65, + "learning_rate": 1.986060052271061e-05, + "loss": 0.5647, + "step": 1560 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860318282210896e-05, + "loss": 0.6061, + "step": 1561 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860035758285082e-05, + "loss": 0.5984, + "step": 1562 + }, + { + "epoch": 0.65, + "learning_rate": 1.9859752950941292e-05, + "loss": 0.4112, + "step": 1563 + }, + { + "epoch": 0.65, + "learning_rate": 1.9859469860187654e-05, + "loss": 0.4325, + "step": 1564 + }, + { + "epoch": 0.66, + "learning_rate": 1.9859186486032302e-05, + "loss": 0.5618, + "step": 1565 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858902828483382e-05, + "loss": 0.5109, + "step": 1566 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858618887549053e-05, + "loss": 0.5181, + "step": 1567 + }, + { + "epoch": 0.66, + "learning_rate": 1.985833466323747e-05, + "loss": 0.5925, + "step": 1568 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858050155556807e-05, + "loss": 0.5414, + "step": 1569 + }, + { + "epoch": 0.66, + "learning_rate": 1.9857765364515234e-05, + "loss": 0.5568, + "step": 1570 + }, + { + "epoch": 0.66, + "learning_rate": 1.9857480290120947e-05, + "loss": 0.5295, + "step": 1571 + }, + { + "epoch": 0.66, + "learning_rate": 1.985719493238214e-05, + "loss": 0.5144, + "step": 1572 + }, + { + "epoch": 0.66, + "learning_rate": 1.9856909291307008e-05, + "loss": 0.4398, + "step": 1573 + }, + { + "epoch": 0.66, + "learning_rate": 1.9856623366903765e-05, + "loss": 0.5049, + "step": 1574 + }, + { + "epoch": 0.66, + "learning_rate": 1.985633715918063e-05, + "loss": 0.539, + "step": 1575 + }, + { + "epoch": 0.66, + "learning_rate": 1.985605066814583e-05, + "loss": 0.4789, + "step": 1576 + }, + { + "epoch": 0.66, + "learning_rate": 1.9855763893807603e-05, + "loss": 0.4971, + "step": 1577 + }, + { + "epoch": 0.66, + "learning_rate": 1.9855476836174182e-05, + "loss": 0.4828, + "step": 1578 + }, + { + "epoch": 0.66, + "learning_rate": 1.985518949525383e-05, + "loss": 0.4806, + "step": 1579 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854901871054803e-05, + "loss": 0.5103, + "step": 1580 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854613963585362e-05, + "loss": 0.4901, + "step": 1581 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854325772853792e-05, + "loss": 0.5219, + "step": 1582 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854037298868367e-05, + "loss": 0.5073, + "step": 1583 + }, + { + "epoch": 0.66, + "learning_rate": 1.9853748541637387e-05, + "loss": 0.5223, + "step": 1584 + }, + { + "epoch": 0.66, + "learning_rate": 1.985345950116915e-05, + "loss": 0.525, + "step": 1585 + }, + { + "epoch": 0.66, + "learning_rate": 1.9853170177471962e-05, + "loss": 0.4888, + "step": 1586 + }, + { + "epoch": 0.66, + "learning_rate": 1.9852880570554142e-05, + "loss": 0.5425, + "step": 1587 + }, + { + "epoch": 0.66, + "learning_rate": 1.9852590680424012e-05, + "loss": 0.5606, + "step": 1588 + }, + { + "epoch": 0.67, + "learning_rate": 1.9852300507089907e-05, + "loss": 0.5552, + "step": 1589 + }, + { + "epoch": 0.67, + "learning_rate": 1.9852010050560165e-05, + "loss": 0.4529, + "step": 1590 + }, + { + "epoch": 0.67, + "learning_rate": 1.985171931084314e-05, + "loss": 0.3869, + "step": 1591 + }, + { + "epoch": 0.67, + "learning_rate": 1.9851428287947184e-05, + "loss": 0.4744, + "step": 1592 + }, + { + "epoch": 0.67, + "learning_rate": 1.9851136981880662e-05, + "loss": 0.5176, + "step": 1593 + }, + { + "epoch": 0.67, + "learning_rate": 1.985084539265195e-05, + "loss": 0.4943, + "step": 1594 + }, + { + "epoch": 0.67, + "learning_rate": 1.985055352026943e-05, + "loss": 0.4605, + "step": 1595 + }, + { + "epoch": 0.67, + "learning_rate": 1.985026136474149e-05, + "loss": 0.564, + "step": 1596 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849968926076528e-05, + "loss": 0.5434, + "step": 1597 + }, + { + "epoch": 0.67, + "learning_rate": 1.984967620428295e-05, + "loss": 0.5173, + "step": 1598 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849383199369167e-05, + "loss": 0.5304, + "step": 1599 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849089911343608e-05, + "loss": 0.5106, + "step": 1600 + }, + { + "epoch": 0.67, + "learning_rate": 1.98487963402147e-05, + "loss": 0.4964, + "step": 1601 + }, + { + "epoch": 0.67, + "learning_rate": 1.9848502485990876e-05, + "loss": 0.4417, + "step": 1602 + }, + { + "epoch": 0.67, + "learning_rate": 1.9848208348680594e-05, + "loss": 0.4463, + "step": 1603 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847913928292297e-05, + "loss": 0.5096, + "step": 1604 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847619224834455e-05, + "loss": 0.5129, + "step": 1605 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847324238315536e-05, + "loss": 0.4326, + "step": 1606 + }, + { + "epoch": 0.67, + "learning_rate": 1.984702896874402e-05, + "loss": 0.5151, + "step": 1607 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846733416128393e-05, + "loss": 0.6225, + "step": 1608 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846437580477155e-05, + "loss": 0.5483, + "step": 1609 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846141461798804e-05, + "loss": 0.5227, + "step": 1610 + }, + { + "epoch": 0.67, + "learning_rate": 1.984584506010186e-05, + "loss": 0.4964, + "step": 1611 + }, + { + "epoch": 0.67, + "learning_rate": 1.984554837539483e-05, + "loss": 0.5653, + "step": 1612 + }, + { + "epoch": 0.68, + "learning_rate": 1.984525140768625e-05, + "loss": 0.5379, + "step": 1613 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844954156984656e-05, + "loss": 0.5595, + "step": 1614 + }, + { + "epoch": 0.68, + "learning_rate": 1.984465662329859e-05, + "loss": 0.5941, + "step": 1615 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844358806636605e-05, + "loss": 0.4752, + "step": 1616 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844060707007263e-05, + "loss": 0.5019, + "step": 1617 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843762324419132e-05, + "loss": 0.5074, + "step": 1618 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843463658880786e-05, + "loss": 0.628, + "step": 1619 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843164710400815e-05, + "loss": 0.5083, + "step": 1620 + }, + { + "epoch": 0.68, + "learning_rate": 1.9842865478987808e-05, + "loss": 0.4711, + "step": 1621 + }, + { + "epoch": 0.68, + "learning_rate": 1.9842565964650363e-05, + "loss": 0.5057, + "step": 1622 + }, + { + "epoch": 0.68, + "learning_rate": 1.98422661673971e-05, + "loss": 0.5801, + "step": 1623 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841966087236626e-05, + "loss": 0.4771, + "step": 1624 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841665724177572e-05, + "loss": 0.5187, + "step": 1625 + }, + { + "epoch": 0.68, + "learning_rate": 1.984136507822857e-05, + "loss": 0.4719, + "step": 1626 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841064149398262e-05, + "loss": 0.4859, + "step": 1627 + }, + { + "epoch": 0.68, + "learning_rate": 1.9840762937695296e-05, + "loss": 0.4938, + "step": 1628 + }, + { + "epoch": 0.68, + "learning_rate": 1.9840461443128333e-05, + "loss": 0.5628, + "step": 1629 + }, + { + "epoch": 0.68, + "learning_rate": 1.984015966570604e-05, + "loss": 0.5753, + "step": 1630 + }, + { + "epoch": 0.68, + "learning_rate": 1.9839857605437086e-05, + "loss": 0.4204, + "step": 1631 + }, + { + "epoch": 0.68, + "learning_rate": 1.983955526233016e-05, + "loss": 0.4693, + "step": 1632 + }, + { + "epoch": 0.68, + "learning_rate": 1.9839252636393946e-05, + "loss": 0.5352, + "step": 1633 + }, + { + "epoch": 0.68, + "learning_rate": 1.983894972763715e-05, + "loss": 0.5136, + "step": 1634 + }, + { + "epoch": 0.68, + "learning_rate": 1.9838646536068473e-05, + "loss": 0.5446, + "step": 1635 + }, + { + "epoch": 0.68, + "learning_rate": 1.9838343061696633e-05, + "loss": 0.5841, + "step": 1636 + }, + { + "epoch": 0.69, + "learning_rate": 1.983803930453035e-05, + "loss": 0.5229, + "step": 1637 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837735264578358e-05, + "loss": 0.5148, + "step": 1638 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837430941849395e-05, + "loss": 0.4822, + "step": 1639 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837126336352214e-05, + "loss": 0.5169, + "step": 1640 + }, + { + "epoch": 0.69, + "learning_rate": 1.983682144809556e-05, + "loss": 0.4126, + "step": 1641 + }, + { + "epoch": 0.69, + "learning_rate": 1.9836516277088203e-05, + "loss": 0.4774, + "step": 1642 + }, + { + "epoch": 0.69, + "learning_rate": 1.9836210823338917e-05, + "loss": 0.5266, + "step": 1643 + }, + { + "epoch": 0.69, + "learning_rate": 1.983590508685648e-05, + "loss": 0.4671, + "step": 1644 + }, + { + "epoch": 0.69, + "learning_rate": 1.9835599067649676e-05, + "loss": 0.5408, + "step": 1645 + }, + { + "epoch": 0.69, + "learning_rate": 1.983529276572731e-05, + "loss": 0.5083, + "step": 1646 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834986181098176e-05, + "loss": 0.5176, + "step": 1647 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834679313771095e-05, + "loss": 0.5439, + "step": 1648 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834372163754882e-05, + "loss": 0.5107, + "step": 1649 + }, + { + "epoch": 0.69, + "learning_rate": 1.983406473105837e-05, + "loss": 0.5676, + "step": 1650 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833757015690393e-05, + "loss": 0.4696, + "step": 1651 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833449017659796e-05, + "loss": 0.5984, + "step": 1652 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833140736975433e-05, + "loss": 0.5636, + "step": 1653 + }, + { + "epoch": 0.69, + "learning_rate": 1.9832832173646164e-05, + "loss": 0.5231, + "step": 1654 + }, + { + "epoch": 0.69, + "learning_rate": 1.9832523327680865e-05, + "loss": 0.5034, + "step": 1655 + }, + { + "epoch": 0.69, + "learning_rate": 1.98322141990884e-05, + "loss": 0.5617, + "step": 1656 + }, + { + "epoch": 0.69, + "learning_rate": 1.983190478787767e-05, + "loss": 0.4909, + "step": 1657 + }, + { + "epoch": 0.69, + "learning_rate": 1.983159509405756e-05, + "loss": 0.4736, + "step": 1658 + }, + { + "epoch": 0.69, + "learning_rate": 1.9831285117636972e-05, + "loss": 0.592, + "step": 1659 + }, + { + "epoch": 0.69, + "learning_rate": 1.983097485862482e-05, + "loss": 0.5005, + "step": 1660 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830664317030014e-05, + "loss": 0.5292, + "step": 1661 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830353492861493e-05, + "loss": 0.5129, + "step": 1662 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830042386128178e-05, + "loss": 0.575, + "step": 1663 + }, + { + "epoch": 0.7, + "learning_rate": 1.982973099683902e-05, + "loss": 0.4696, + "step": 1664 + }, + { + "epoch": 0.7, + "learning_rate": 1.9829419325002966e-05, + "loss": 0.4746, + "step": 1665 + }, + { + "epoch": 0.7, + "learning_rate": 1.982910737062898e-05, + "loss": 0.4449, + "step": 1666 + }, + { + "epoch": 0.7, + "learning_rate": 1.982879513372602e-05, + "loss": 0.4704, + "step": 1667 + }, + { + "epoch": 0.7, + "learning_rate": 1.9828482614303068e-05, + "loss": 0.5179, + "step": 1668 + }, + { + "epoch": 0.7, + "learning_rate": 1.9828169812369106e-05, + "loss": 0.4781, + "step": 1669 + }, + { + "epoch": 0.7, + "learning_rate": 1.9827856727933122e-05, + "loss": 0.5388, + "step": 1670 + }, + { + "epoch": 0.7, + "learning_rate": 1.9827543361004118e-05, + "loss": 0.5813, + "step": 1671 + }, + { + "epoch": 0.7, + "learning_rate": 1.98272297115911e-05, + "loss": 0.4643, + "step": 1672 + }, + { + "epoch": 0.7, + "learning_rate": 1.9826915779703087e-05, + "loss": 0.4306, + "step": 1673 + }, + { + "epoch": 0.7, + "learning_rate": 1.98266015653491e-05, + "loss": 0.5216, + "step": 1674 + }, + { + "epoch": 0.7, + "learning_rate": 1.9826287068538167e-05, + "loss": 0.5218, + "step": 1675 + }, + { + "epoch": 0.7, + "learning_rate": 1.9825972289279336e-05, + "loss": 0.519, + "step": 1676 + }, + { + "epoch": 0.7, + "learning_rate": 1.982565722758165e-05, + "loss": 0.4881, + "step": 1677 + }, + { + "epoch": 0.7, + "learning_rate": 1.9825341883454165e-05, + "loss": 0.5389, + "step": 1678 + }, + { + "epoch": 0.7, + "learning_rate": 1.982502625690595e-05, + "loss": 0.5541, + "step": 1679 + }, + { + "epoch": 0.7, + "learning_rate": 1.982471034794607e-05, + "loss": 0.4128, + "step": 1680 + }, + { + "epoch": 0.7, + "learning_rate": 1.982439415658361e-05, + "loss": 0.4948, + "step": 1681 + }, + { + "epoch": 0.7, + "learning_rate": 1.9824077682827658e-05, + "loss": 0.4972, + "step": 1682 + }, + { + "epoch": 0.7, + "learning_rate": 1.982376092668731e-05, + "loss": 0.5727, + "step": 1683 + }, + { + "epoch": 0.71, + "learning_rate": 1.9823443888171676e-05, + "loss": 0.5451, + "step": 1684 + }, + { + "epoch": 0.71, + "learning_rate": 1.982312656728986e-05, + "loss": 0.4172, + "step": 1685 + }, + { + "epoch": 0.71, + "learning_rate": 1.9822808964050987e-05, + "loss": 0.5163, + "step": 1686 + }, + { + "epoch": 0.71, + "learning_rate": 1.982249107846419e-05, + "loss": 0.5484, + "step": 1687 + }, + { + "epoch": 0.71, + "learning_rate": 1.98221729105386e-05, + "loss": 0.5193, + "step": 1688 + }, + { + "epoch": 0.71, + "learning_rate": 1.982185446028337e-05, + "loss": 0.483, + "step": 1689 + }, + { + "epoch": 0.71, + "learning_rate": 1.9821535727707647e-05, + "loss": 0.519, + "step": 1690 + }, + { + "epoch": 0.71, + "learning_rate": 1.9821216712820597e-05, + "loss": 0.5161, + "step": 1691 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820897415631383e-05, + "loss": 0.5653, + "step": 1692 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820577836149192e-05, + "loss": 0.4888, + "step": 1693 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820257974383205e-05, + "loss": 0.4977, + "step": 1694 + }, + { + "epoch": 0.71, + "learning_rate": 1.9819937830342618e-05, + "loss": 0.4693, + "step": 1695 + }, + { + "epoch": 0.71, + "learning_rate": 1.981961740403663e-05, + "loss": 0.493, + "step": 1696 + }, + { + "epoch": 0.71, + "learning_rate": 1.9819296695474454e-05, + "loss": 0.5228, + "step": 1697 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818975704665306e-05, + "loss": 0.4938, + "step": 1698 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818654431618416e-05, + "loss": 0.4252, + "step": 1699 + }, + { + "epoch": 0.71, + "learning_rate": 1.981833287634302e-05, + "loss": 0.4584, + "step": 1700 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818011038848355e-05, + "loss": 0.5541, + "step": 1701 + }, + { + "epoch": 0.71, + "learning_rate": 1.9817688919143676e-05, + "loss": 0.412, + "step": 1702 + }, + { + "epoch": 0.71, + "learning_rate": 1.981736651723824e-05, + "loss": 0.5411, + "step": 1703 + }, + { + "epoch": 0.71, + "learning_rate": 1.9817043833141317e-05, + "loss": 0.4207, + "step": 1704 + }, + { + "epoch": 0.71, + "learning_rate": 1.981672086686218e-05, + "loss": 0.502, + "step": 1705 + }, + { + "epoch": 0.71, + "learning_rate": 1.9816397618410112e-05, + "loss": 0.5266, + "step": 1706 + }, + { + "epoch": 0.71, + "learning_rate": 1.9816074087794406e-05, + "loss": 0.5749, + "step": 1707 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815750275024362e-05, + "loss": 0.5609, + "step": 1708 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815426180109286e-05, + "loss": 0.5089, + "step": 1709 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815101803058492e-05, + "loss": 0.5283, + "step": 1710 + }, + { + "epoch": 0.72, + "learning_rate": 1.981477714388131e-05, + "loss": 0.4316, + "step": 1711 + }, + { + "epoch": 0.72, + "learning_rate": 1.9814452202587066e-05, + "loss": 0.4375, + "step": 1712 + }, + { + "epoch": 0.72, + "learning_rate": 1.9814126979185105e-05, + "loss": 0.5121, + "step": 1713 + }, + { + "epoch": 0.72, + "learning_rate": 1.981380147368477e-05, + "loss": 0.4719, + "step": 1714 + }, + { + "epoch": 0.72, + "learning_rate": 1.981347568609542e-05, + "loss": 0.4747, + "step": 1715 + }, + { + "epoch": 0.72, + "learning_rate": 1.9813149616426424e-05, + "loss": 0.5267, + "step": 1716 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812823264687148e-05, + "loss": 0.4687, + "step": 1717 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812496630886974e-05, + "loss": 0.5243, + "step": 1718 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812169715035295e-05, + "loss": 0.5228, + "step": 1719 + }, + { + "epoch": 0.72, + "learning_rate": 1.98118425171415e-05, + "loss": 0.5324, + "step": 1720 + }, + { + "epoch": 0.72, + "learning_rate": 1.9811515037215002e-05, + "loss": 0.5004, + "step": 1721 + }, + { + "epoch": 0.72, + "learning_rate": 1.981118727526521e-05, + "loss": 0.4853, + "step": 1722 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810859231301545e-05, + "loss": 0.551, + "step": 1723 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810530905333437e-05, + "loss": 0.4566, + "step": 1724 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810202297370328e-05, + "loss": 0.4481, + "step": 1725 + }, + { + "epoch": 0.72, + "learning_rate": 1.9809873407421656e-05, + "loss": 0.5405, + "step": 1726 + }, + { + "epoch": 0.72, + "learning_rate": 1.980954423549688e-05, + "loss": 0.5531, + "step": 1727 + }, + { + "epoch": 0.72, + "learning_rate": 1.9809214781605458e-05, + "loss": 0.5072, + "step": 1728 + }, + { + "epoch": 0.72, + "learning_rate": 1.980888504575686e-05, + "loss": 0.6357, + "step": 1729 + }, + { + "epoch": 0.72, + "learning_rate": 1.980855502796057e-05, + "loss": 0.523, + "step": 1730 + }, + { + "epoch": 0.72, + "learning_rate": 1.9808224728226067e-05, + "loss": 0.4832, + "step": 1731 + }, + { + "epoch": 0.73, + "learning_rate": 1.9807894146562847e-05, + "loss": 0.4624, + "step": 1732 + }, + { + "epoch": 0.73, + "learning_rate": 1.9807563282980415e-05, + "loss": 0.444, + "step": 1733 + }, + { + "epoch": 0.73, + "learning_rate": 1.980723213748828e-05, + "loss": 0.485, + "step": 1734 + }, + { + "epoch": 0.73, + "learning_rate": 1.9806900710095958e-05, + "loss": 0.6331, + "step": 1735 + }, + { + "epoch": 0.73, + "learning_rate": 1.980656900081298e-05, + "loss": 0.551, + "step": 1736 + }, + { + "epoch": 0.73, + "learning_rate": 1.9806237009648876e-05, + "loss": 0.4904, + "step": 1737 + }, + { + "epoch": 0.73, + "learning_rate": 1.980590473661319e-05, + "loss": 0.5215, + "step": 1738 + }, + { + "epoch": 0.73, + "learning_rate": 1.9805572181715478e-05, + "loss": 0.5491, + "step": 1739 + }, + { + "epoch": 0.73, + "learning_rate": 1.980523934496529e-05, + "loss": 0.5062, + "step": 1740 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804906226372202e-05, + "loss": 0.5369, + "step": 1741 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804572825945784e-05, + "loss": 0.4987, + "step": 1742 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804239143695622e-05, + "loss": 0.4956, + "step": 1743 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803905179631303e-05, + "loss": 0.4876, + "step": 1744 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803570933762432e-05, + "loss": 0.5259, + "step": 1745 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803236406098612e-05, + "loss": 0.4896, + "step": 1746 + }, + { + "epoch": 0.73, + "learning_rate": 1.9802901596649462e-05, + "loss": 0.4446, + "step": 1747 + }, + { + "epoch": 0.73, + "learning_rate": 1.9802566505424603e-05, + "loss": 0.5677, + "step": 1748 + }, + { + "epoch": 0.73, + "learning_rate": 1.980223113243367e-05, + "loss": 0.5355, + "step": 1749 + }, + { + "epoch": 0.73, + "learning_rate": 1.98018954776863e-05, + "loss": 0.4811, + "step": 1750 + }, + { + "epoch": 0.73, + "learning_rate": 1.9801559541192147e-05, + "loss": 0.4993, + "step": 1751 + }, + { + "epoch": 0.73, + "learning_rate": 1.9801223322960858e-05, + "loss": 0.5684, + "step": 1752 + }, + { + "epoch": 0.73, + "learning_rate": 1.9800886823002105e-05, + "loss": 0.5324, + "step": 1753 + }, + { + "epoch": 0.73, + "learning_rate": 1.9800550041325553e-05, + "loss": 0.5637, + "step": 1754 + }, + { + "epoch": 0.73, + "learning_rate": 1.980021297794089e-05, + "loss": 0.5459, + "step": 1755 + }, + { + "epoch": 0.74, + "learning_rate": 1.9799875632857802e-05, + "loss": 0.5847, + "step": 1756 + }, + { + "epoch": 0.74, + "learning_rate": 1.9799538006085982e-05, + "loss": 0.5407, + "step": 1757 + }, + { + "epoch": 0.74, + "learning_rate": 1.979920009763514e-05, + "loss": 0.4979, + "step": 1758 + }, + { + "epoch": 0.74, + "learning_rate": 1.979886190751499e-05, + "loss": 0.438, + "step": 1759 + }, + { + "epoch": 0.74, + "learning_rate": 1.9798523435735246e-05, + "loss": 0.5051, + "step": 1760 + }, + { + "epoch": 0.74, + "learning_rate": 1.9798184682305645e-05, + "loss": 0.4852, + "step": 1761 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797845647235915e-05, + "loss": 0.5432, + "step": 1762 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797506330535808e-05, + "loss": 0.4505, + "step": 1763 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797166732215078e-05, + "loss": 0.6174, + "step": 1764 + }, + { + "epoch": 0.74, + "learning_rate": 1.979682685228348e-05, + "loss": 0.4877, + "step": 1765 + }, + { + "epoch": 0.74, + "learning_rate": 1.9796486690750792e-05, + "loss": 0.515, + "step": 1766 + }, + { + "epoch": 0.74, + "learning_rate": 1.9796146247626783e-05, + "loss": 0.4858, + "step": 1767 + }, + { + "epoch": 0.74, + "learning_rate": 1.9795805522921246e-05, + "loss": 0.5304, + "step": 1768 + }, + { + "epoch": 0.74, + "learning_rate": 1.979546451664397e-05, + "loss": 0.5342, + "step": 1769 + }, + { + "epoch": 0.74, + "learning_rate": 1.979512322880476e-05, + "loss": 0.5117, + "step": 1770 + }, + { + "epoch": 0.74, + "learning_rate": 1.9794781659413425e-05, + "loss": 0.5361, + "step": 1771 + }, + { + "epoch": 0.74, + "learning_rate": 1.9794439808479783e-05, + "loss": 0.4229, + "step": 1772 + }, + { + "epoch": 0.74, + "learning_rate": 1.979409767601366e-05, + "loss": 0.413, + "step": 1773 + }, + { + "epoch": 0.74, + "learning_rate": 1.979375526202489e-05, + "loss": 0.464, + "step": 1774 + }, + { + "epoch": 0.74, + "learning_rate": 1.9793412566523318e-05, + "loss": 0.5194, + "step": 1775 + }, + { + "epoch": 0.74, + "learning_rate": 1.9793069589518787e-05, + "loss": 0.5144, + "step": 1776 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792726331021166e-05, + "loss": 0.451, + "step": 1777 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792382791040317e-05, + "loss": 0.5394, + "step": 1778 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792038969586113e-05, + "loss": 0.5391, + "step": 1779 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791694866668434e-05, + "loss": 0.5101, + "step": 1780 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791350482297177e-05, + "loss": 0.4901, + "step": 1781 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791005816482242e-05, + "loss": 0.4868, + "step": 1782 + }, + { + "epoch": 0.75, + "learning_rate": 1.9790660869233532e-05, + "loss": 0.4358, + "step": 1783 + }, + { + "epoch": 0.75, + "learning_rate": 1.9790315640560962e-05, + "loss": 0.5186, + "step": 1784 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789970130474455e-05, + "loss": 0.5941, + "step": 1785 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789624338983948e-05, + "loss": 0.4653, + "step": 1786 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789278266099375e-05, + "loss": 0.4051, + "step": 1787 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788931911830688e-05, + "loss": 0.4873, + "step": 1788 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788585276187836e-05, + "loss": 0.5416, + "step": 1789 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788238359180786e-05, + "loss": 0.6702, + "step": 1790 + }, + { + "epoch": 0.75, + "learning_rate": 1.978789116081951e-05, + "loss": 0.5041, + "step": 1791 + }, + { + "epoch": 0.75, + "learning_rate": 1.9787543681113993e-05, + "loss": 0.4556, + "step": 1792 + }, + { + "epoch": 0.75, + "learning_rate": 1.9787195920074213e-05, + "loss": 0.5527, + "step": 1793 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786847877710177e-05, + "loss": 0.5213, + "step": 1794 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786499554031876e-05, + "loss": 0.5291, + "step": 1795 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786150949049335e-05, + "loss": 0.4473, + "step": 1796 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785802062772566e-05, + "loss": 0.5462, + "step": 1797 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785452895211606e-05, + "loss": 0.5014, + "step": 1798 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785103446376483e-05, + "loss": 0.5301, + "step": 1799 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784753716277242e-05, + "loss": 0.5775, + "step": 1800 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784403704923937e-05, + "loss": 0.4802, + "step": 1801 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784053412326637e-05, + "loss": 0.4719, + "step": 1802 + }, + { + "epoch": 0.75, + "learning_rate": 1.97837028384954e-05, + "loss": 0.5638, + "step": 1803 + }, + { + "epoch": 0.76, + "learning_rate": 1.9783351983440306e-05, + "loss": 0.5138, + "step": 1804 + }, + { + "epoch": 0.76, + "learning_rate": 1.978300084717144e-05, + "loss": 0.4597, + "step": 1805 + }, + { + "epoch": 0.76, + "learning_rate": 1.9782649429698896e-05, + "loss": 0.4601, + "step": 1806 + }, + { + "epoch": 0.76, + "learning_rate": 1.9782297731032778e-05, + "loss": 0.5567, + "step": 1807 + }, + { + "epoch": 0.76, + "learning_rate": 1.978194575118319e-05, + "loss": 0.5288, + "step": 1808 + }, + { + "epoch": 0.76, + "learning_rate": 1.9781593490160255e-05, + "loss": 0.5387, + "step": 1809 + }, + { + "epoch": 0.76, + "learning_rate": 1.9781240947974093e-05, + "loss": 0.4789, + "step": 1810 + }, + { + "epoch": 0.76, + "learning_rate": 1.978088812463484e-05, + "loss": 0.4638, + "step": 1811 + }, + { + "epoch": 0.76, + "learning_rate": 1.9780535020152632e-05, + "loss": 0.5389, + "step": 1812 + }, + { + "epoch": 0.76, + "learning_rate": 1.978018163453763e-05, + "loss": 0.4942, + "step": 1813 + }, + { + "epoch": 0.76, + "learning_rate": 1.9779827967799985e-05, + "loss": 0.4926, + "step": 1814 + }, + { + "epoch": 0.76, + "learning_rate": 1.9779474019949864e-05, + "loss": 0.5134, + "step": 1815 + }, + { + "epoch": 0.76, + "learning_rate": 1.977911979099744e-05, + "loss": 0.5047, + "step": 1816 + }, + { + "epoch": 0.76, + "learning_rate": 1.9778765280952896e-05, + "loss": 0.492, + "step": 1817 + }, + { + "epoch": 0.76, + "learning_rate": 1.977841048982642e-05, + "loss": 0.5855, + "step": 1818 + }, + { + "epoch": 0.76, + "learning_rate": 1.977805541762821e-05, + "loss": 0.487, + "step": 1819 + }, + { + "epoch": 0.76, + "learning_rate": 1.9777700064368475e-05, + "loss": 0.562, + "step": 1820 + }, + { + "epoch": 0.76, + "learning_rate": 1.977734443005743e-05, + "loss": 0.4709, + "step": 1821 + }, + { + "epoch": 0.76, + "learning_rate": 1.9776988514705294e-05, + "loss": 0.5106, + "step": 1822 + }, + { + "epoch": 0.76, + "learning_rate": 1.97766323183223e-05, + "loss": 0.5815, + "step": 1823 + }, + { + "epoch": 0.76, + "learning_rate": 1.9776275840918685e-05, + "loss": 0.4957, + "step": 1824 + }, + { + "epoch": 0.76, + "learning_rate": 1.97759190825047e-05, + "loss": 0.4592, + "step": 1825 + }, + { + "epoch": 0.76, + "learning_rate": 1.977556204309059e-05, + "loss": 0.5652, + "step": 1826 + }, + { + "epoch": 0.76, + "learning_rate": 1.9775204722686628e-05, + "loss": 0.4936, + "step": 1827 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774847121303078e-05, + "loss": 0.5499, + "step": 1828 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774489238950222e-05, + "loss": 0.5595, + "step": 1829 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774131075638347e-05, + "loss": 0.5254, + "step": 1830 + }, + { + "epoch": 0.77, + "learning_rate": 1.977377263137775e-05, + "loss": 0.5087, + "step": 1831 + }, + { + "epoch": 0.77, + "learning_rate": 1.977341390617873e-05, + "loss": 0.4864, + "step": 1832 + }, + { + "epoch": 0.77, + "learning_rate": 1.97730549000516e-05, + "loss": 0.5294, + "step": 1833 + }, + { + "epoch": 0.77, + "learning_rate": 1.9772695613006676e-05, + "loss": 0.4775, + "step": 1834 + }, + { + "epoch": 0.77, + "learning_rate": 1.977233604505429e-05, + "loss": 0.5719, + "step": 1835 + }, + { + "epoch": 0.77, + "learning_rate": 1.977197619620478e-05, + "loss": 0.5053, + "step": 1836 + }, + { + "epoch": 0.77, + "learning_rate": 1.9771616066468484e-05, + "loss": 0.4699, + "step": 1837 + }, + { + "epoch": 0.77, + "learning_rate": 1.9771255655855753e-05, + "loss": 0.5384, + "step": 1838 + }, + { + "epoch": 0.77, + "learning_rate": 1.9770894964376956e-05, + "loss": 0.5049, + "step": 1839 + }, + { + "epoch": 0.77, + "learning_rate": 1.9770533992042447e-05, + "loss": 0.5714, + "step": 1840 + }, + { + "epoch": 0.77, + "learning_rate": 1.977017273886261e-05, + "loss": 0.5266, + "step": 1841 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769811204847825e-05, + "loss": 0.5479, + "step": 1842 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769449390008492e-05, + "loss": 0.5449, + "step": 1843 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769087294355006e-05, + "loss": 0.4606, + "step": 1844 + }, + { + "epoch": 0.77, + "learning_rate": 1.976872491789777e-05, + "loss": 0.5291, + "step": 1845 + }, + { + "epoch": 0.77, + "learning_rate": 1.9768362260647205e-05, + "loss": 0.5553, + "step": 1846 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767999322613737e-05, + "loss": 0.5144, + "step": 1847 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767636103807796e-05, + "loss": 0.5458, + "step": 1848 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767272604239823e-05, + "loss": 0.559, + "step": 1849 + }, + { + "epoch": 0.77, + "learning_rate": 1.9766908823920268e-05, + "loss": 0.4688, + "step": 1850 + }, + { + "epoch": 0.77, + "learning_rate": 1.976654476285958e-05, + "loss": 0.558, + "step": 1851 + }, + { + "epoch": 0.78, + "learning_rate": 1.9766180421068236e-05, + "loss": 0.5534, + "step": 1852 + }, + { + "epoch": 0.78, + "learning_rate": 1.97658157985567e-05, + "loss": 0.469, + "step": 1853 + }, + { + "epoch": 0.78, + "learning_rate": 1.9765450895335454e-05, + "loss": 0.4969, + "step": 1854 + }, + { + "epoch": 0.78, + "learning_rate": 1.9765085711414987e-05, + "loss": 0.5451, + "step": 1855 + }, + { + "epoch": 0.78, + "learning_rate": 1.97647202468058e-05, + "loss": 0.5496, + "step": 1856 + }, + { + "epoch": 0.78, + "learning_rate": 1.9764354501518394e-05, + "loss": 0.4641, + "step": 1857 + }, + { + "epoch": 0.78, + "learning_rate": 1.976398847556328e-05, + "loss": 0.513, + "step": 1858 + }, + { + "epoch": 0.78, + "learning_rate": 1.9763622168950982e-05, + "loss": 0.6003, + "step": 1859 + }, + { + "epoch": 0.78, + "learning_rate": 1.9763255581692027e-05, + "loss": 0.5382, + "step": 1860 + }, + { + "epoch": 0.78, + "learning_rate": 1.976288871379696e-05, + "loss": 0.5244, + "step": 1861 + }, + { + "epoch": 0.78, + "learning_rate": 1.9762521565276314e-05, + "loss": 0.5568, + "step": 1862 + }, + { + "epoch": 0.78, + "learning_rate": 1.976215413614065e-05, + "loss": 0.509, + "step": 1863 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761786426400532e-05, + "loss": 0.5482, + "step": 1864 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761418436066523e-05, + "loss": 0.5047, + "step": 1865 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761050165149208e-05, + "loss": 0.4403, + "step": 1866 + }, + { + "epoch": 0.78, + "learning_rate": 1.976068161365916e-05, + "loss": 0.509, + "step": 1867 + }, + { + "epoch": 0.78, + "learning_rate": 1.9760312781606985e-05, + "loss": 0.5031, + "step": 1868 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759943669003282e-05, + "loss": 0.5258, + "step": 1869 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759574275858653e-05, + "loss": 0.5036, + "step": 1870 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759204602183725e-05, + "loss": 0.4541, + "step": 1871 + }, + { + "epoch": 0.78, + "learning_rate": 1.9758834647989123e-05, + "loss": 0.4063, + "step": 1872 + }, + { + "epoch": 0.78, + "learning_rate": 1.9758464413285477e-05, + "loss": 0.5494, + "step": 1873 + }, + { + "epoch": 0.78, + "learning_rate": 1.975809389808343e-05, + "loss": 0.5205, + "step": 1874 + }, + { + "epoch": 0.79, + "learning_rate": 1.9757723102393634e-05, + "loss": 0.4654, + "step": 1875 + }, + { + "epoch": 0.79, + "learning_rate": 1.9757352026226746e-05, + "loss": 0.4682, + "step": 1876 + }, + { + "epoch": 0.79, + "learning_rate": 1.9756980669593432e-05, + "loss": 0.5617, + "step": 1877 + }, + { + "epoch": 0.79, + "learning_rate": 1.975660903250437e-05, + "loss": 0.4518, + "step": 1878 + }, + { + "epoch": 0.79, + "learning_rate": 1.9756237114970235e-05, + "loss": 0.4759, + "step": 1879 + }, + { + "epoch": 0.79, + "learning_rate": 1.9755864917001724e-05, + "loss": 0.6428, + "step": 1880 + }, + { + "epoch": 0.79, + "learning_rate": 1.9755492438609532e-05, + "loss": 0.5162, + "step": 1881 + }, + { + "epoch": 0.79, + "learning_rate": 1.975511967980437e-05, + "loss": 0.4909, + "step": 1882 + }, + { + "epoch": 0.79, + "learning_rate": 1.9754746640596945e-05, + "loss": 0.4749, + "step": 1883 + }, + { + "epoch": 0.79, + "learning_rate": 1.9754373320997987e-05, + "loss": 0.4545, + "step": 1884 + }, + { + "epoch": 0.79, + "learning_rate": 1.975399972101822e-05, + "loss": 0.5069, + "step": 1885 + }, + { + "epoch": 0.79, + "learning_rate": 1.975362584066839e-05, + "loss": 0.4973, + "step": 1886 + }, + { + "epoch": 0.79, + "learning_rate": 1.975325167995924e-05, + "loss": 0.4698, + "step": 1887 + }, + { + "epoch": 0.79, + "learning_rate": 1.9752877238901525e-05, + "loss": 0.4487, + "step": 1888 + }, + { + "epoch": 0.79, + "learning_rate": 1.975250251750601e-05, + "loss": 0.5072, + "step": 1889 + }, + { + "epoch": 0.79, + "learning_rate": 1.9752127515783458e-05, + "loss": 0.5311, + "step": 1890 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751752233744662e-05, + "loss": 0.4764, + "step": 1891 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751376671400398e-05, + "loss": 0.4509, + "step": 1892 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751000828761465e-05, + "loss": 0.5794, + "step": 1893 + }, + { + "epoch": 0.79, + "learning_rate": 1.9750624705838666e-05, + "loss": 0.4593, + "step": 1894 + }, + { + "epoch": 0.79, + "learning_rate": 1.9750248302642813e-05, + "loss": 0.472, + "step": 1895 + }, + { + "epoch": 0.79, + "learning_rate": 1.9749871619184726e-05, + "loss": 0.4706, + "step": 1896 + }, + { + "epoch": 0.79, + "learning_rate": 1.974949465547523e-05, + "loss": 0.5228, + "step": 1897 + }, + { + "epoch": 0.79, + "learning_rate": 1.974911741152516e-05, + "loss": 0.5407, + "step": 1898 + }, + { + "epoch": 0.8, + "learning_rate": 1.9748739887345362e-05, + "loss": 0.6125, + "step": 1899 + }, + { + "epoch": 0.8, + "learning_rate": 1.974836208294669e-05, + "loss": 0.4526, + "step": 1900 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747983998339996e-05, + "loss": 0.4489, + "step": 1901 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747605633536152e-05, + "loss": 0.4619, + "step": 1902 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747226988546038e-05, + "loss": 0.5118, + "step": 1903 + }, + { + "epoch": 0.8, + "learning_rate": 1.974684806338053e-05, + "loss": 0.5502, + "step": 1904 + }, + { + "epoch": 0.8, + "learning_rate": 1.9746468858050526e-05, + "loss": 0.5278, + "step": 1905 + }, + { + "epoch": 0.8, + "learning_rate": 1.9746089372566924e-05, + "loss": 0.4562, + "step": 1906 + }, + { + "epoch": 0.8, + "learning_rate": 1.974570960694063e-05, + "loss": 0.5074, + "step": 1907 + }, + { + "epoch": 0.8, + "learning_rate": 1.974532956118256e-05, + "loss": 0.4635, + "step": 1908 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744949235303644e-05, + "loss": 0.5443, + "step": 1909 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744568629314807e-05, + "loss": 0.4998, + "step": 1910 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744187743226994e-05, + "loss": 0.4901, + "step": 1911 + }, + { + "epoch": 0.8, + "learning_rate": 1.974380657705115e-05, + "loss": 0.5269, + "step": 1912 + }, + { + "epoch": 0.8, + "learning_rate": 1.974342513079823e-05, + "loss": 0.4695, + "step": 1913 + }, + { + "epoch": 0.8, + "learning_rate": 1.9743043404479203e-05, + "loss": 0.4963, + "step": 1914 + }, + { + "epoch": 0.8, + "learning_rate": 1.9742661398105038e-05, + "loss": 0.5258, + "step": 1915 + }, + { + "epoch": 0.8, + "learning_rate": 1.9742279111686717e-05, + "loss": 0.4798, + "step": 1916 + }, + { + "epoch": 0.8, + "learning_rate": 1.974189654523523e-05, + "loss": 0.5146, + "step": 1917 + }, + { + "epoch": 0.8, + "learning_rate": 1.974151369876157e-05, + "loss": 0.4735, + "step": 1918 + }, + { + "epoch": 0.8, + "learning_rate": 1.9741130572276742e-05, + "loss": 0.5074, + "step": 1919 + }, + { + "epoch": 0.8, + "learning_rate": 1.9740747165791762e-05, + "loss": 0.5401, + "step": 1920 + }, + { + "epoch": 0.8, + "learning_rate": 1.974036347931765e-05, + "loss": 0.5378, + "step": 1921 + }, + { + "epoch": 0.8, + "learning_rate": 1.973997951286543e-05, + "loss": 0.5635, + "step": 1922 + }, + { + "epoch": 0.81, + "learning_rate": 1.9739595266446143e-05, + "loss": 0.6037, + "step": 1923 + }, + { + "epoch": 0.81, + "learning_rate": 1.9739210740070833e-05, + "loss": 0.4684, + "step": 1924 + }, + { + "epoch": 0.81, + "learning_rate": 1.973882593375055e-05, + "loss": 0.48, + "step": 1925 + }, + { + "epoch": 0.81, + "learning_rate": 1.973844084749636e-05, + "loss": 0.5285, + "step": 1926 + }, + { + "epoch": 0.81, + "learning_rate": 1.973805548131933e-05, + "loss": 0.4726, + "step": 1927 + }, + { + "epoch": 0.81, + "learning_rate": 1.9737669835230537e-05, + "loss": 0.5762, + "step": 1928 + }, + { + "epoch": 0.81, + "learning_rate": 1.9737283909241065e-05, + "loss": 0.5675, + "step": 1929 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736897703362006e-05, + "loss": 0.5074, + "step": 1930 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736511217604465e-05, + "loss": 0.4967, + "step": 1931 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736124451979547e-05, + "loss": 0.5085, + "step": 1932 + }, + { + "epoch": 0.81, + "learning_rate": 1.9735737406498375e-05, + "loss": 0.5721, + "step": 1933 + }, + { + "epoch": 0.81, + "learning_rate": 1.973535008117207e-05, + "loss": 0.5391, + "step": 1934 + }, + { + "epoch": 0.81, + "learning_rate": 1.9734962476011763e-05, + "loss": 0.4891, + "step": 1935 + }, + { + "epoch": 0.81, + "learning_rate": 1.97345745910286e-05, + "loss": 0.4809, + "step": 1936 + }, + { + "epoch": 0.81, + "learning_rate": 1.9734186426233728e-05, + "loss": 0.4708, + "step": 1937 + }, + { + "epoch": 0.81, + "learning_rate": 1.9733797981638306e-05, + "loss": 0.4815, + "step": 1938 + }, + { + "epoch": 0.81, + "learning_rate": 1.97334092572535e-05, + "loss": 0.5476, + "step": 1939 + }, + { + "epoch": 0.81, + "learning_rate": 1.9733020253090477e-05, + "loss": 0.5123, + "step": 1940 + }, + { + "epoch": 0.81, + "learning_rate": 1.9732630969160428e-05, + "loss": 0.5376, + "step": 1941 + }, + { + "epoch": 0.81, + "learning_rate": 1.9732241405474536e-05, + "loss": 0.5071, + "step": 1942 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731851562044002e-05, + "loss": 0.4348, + "step": 1943 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731461438880034e-05, + "loss": 0.4937, + "step": 1944 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731071035993837e-05, + "loss": 0.5239, + "step": 1945 + }, + { + "epoch": 0.81, + "learning_rate": 1.9730680353396643e-05, + "loss": 0.497, + "step": 1946 + }, + { + "epoch": 0.82, + "learning_rate": 1.9730289391099675e-05, + "loss": 0.563, + "step": 1947 + }, + { + "epoch": 0.82, + "learning_rate": 1.9729898149114174e-05, + "loss": 0.6107, + "step": 1948 + }, + { + "epoch": 0.82, + "learning_rate": 1.9729506627451383e-05, + "loss": 0.5381, + "step": 1949 + }, + { + "epoch": 0.82, + "learning_rate": 1.972911482612256e-05, + "loss": 0.5316, + "step": 1950 + }, + { + "epoch": 0.82, + "learning_rate": 1.972872274513896e-05, + "loss": 0.5669, + "step": 1951 + }, + { + "epoch": 0.82, + "learning_rate": 1.972833038451186e-05, + "loss": 0.5279, + "step": 1952 + }, + { + "epoch": 0.82, + "learning_rate": 1.972793774425254e-05, + "loss": 0.485, + "step": 1953 + }, + { + "epoch": 0.82, + "learning_rate": 1.972754482437228e-05, + "loss": 0.4867, + "step": 1954 + }, + { + "epoch": 0.82, + "learning_rate": 1.972715162488237e-05, + "loss": 0.4395, + "step": 1955 + }, + { + "epoch": 0.82, + "learning_rate": 1.9726758145794127e-05, + "loss": 0.5581, + "step": 1956 + }, + { + "epoch": 0.82, + "learning_rate": 1.9726364387118847e-05, + "loss": 0.4973, + "step": 1957 + }, + { + "epoch": 0.82, + "learning_rate": 1.9725970348867858e-05, + "loss": 0.5836, + "step": 1958 + }, + { + "epoch": 0.82, + "learning_rate": 1.972557603105248e-05, + "loss": 0.461, + "step": 1959 + }, + { + "epoch": 0.82, + "learning_rate": 1.9725181433684047e-05, + "loss": 0.5794, + "step": 1960 + }, + { + "epoch": 0.82, + "learning_rate": 1.9724786556773906e-05, + "loss": 0.4638, + "step": 1961 + }, + { + "epoch": 0.82, + "learning_rate": 1.9724391400333405e-05, + "loss": 0.4794, + "step": 1962 + }, + { + "epoch": 0.82, + "learning_rate": 1.97239959643739e-05, + "loss": 0.4787, + "step": 1963 + }, + { + "epoch": 0.82, + "learning_rate": 1.9723600248906766e-05, + "loss": 0.4904, + "step": 1964 + }, + { + "epoch": 0.82, + "learning_rate": 1.9723204253943366e-05, + "loss": 0.482, + "step": 1965 + }, + { + "epoch": 0.82, + "learning_rate": 1.972280797949509e-05, + "loss": 0.5006, + "step": 1966 + }, + { + "epoch": 0.82, + "learning_rate": 1.9722411425573323e-05, + "loss": 0.506, + "step": 1967 + }, + { + "epoch": 0.82, + "learning_rate": 1.9722014592189472e-05, + "loss": 0.5763, + "step": 1968 + }, + { + "epoch": 0.82, + "learning_rate": 1.9721617479354936e-05, + "loss": 0.5273, + "step": 1969 + }, + { + "epoch": 0.82, + "learning_rate": 1.972122008708113e-05, + "loss": 0.4863, + "step": 1970 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720822415379485e-05, + "loss": 0.4972, + "step": 1971 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720424464261426e-05, + "loss": 0.5359, + "step": 1972 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720026233738388e-05, + "loss": 0.4415, + "step": 1973 + }, + { + "epoch": 0.83, + "learning_rate": 1.9719627723821818e-05, + "loss": 0.5304, + "step": 1974 + }, + { + "epoch": 0.83, + "learning_rate": 1.971922893452318e-05, + "loss": 0.489, + "step": 1975 + }, + { + "epoch": 0.83, + "learning_rate": 1.971882986585393e-05, + "loss": 0.49, + "step": 1976 + }, + { + "epoch": 0.83, + "learning_rate": 1.9718430517825538e-05, + "loss": 0.5563, + "step": 1977 + }, + { + "epoch": 0.83, + "learning_rate": 1.9718030890449483e-05, + "loss": 0.5101, + "step": 1978 + }, + { + "epoch": 0.83, + "learning_rate": 1.9717630983737256e-05, + "loss": 0.5219, + "step": 1979 + }, + { + "epoch": 0.83, + "learning_rate": 1.971723079770035e-05, + "loss": 0.4326, + "step": 1980 + }, + { + "epoch": 0.83, + "learning_rate": 1.971683033235027e-05, + "loss": 0.4955, + "step": 1981 + }, + { + "epoch": 0.83, + "learning_rate": 1.971642958769852e-05, + "loss": 0.5168, + "step": 1982 + }, + { + "epoch": 0.83, + "learning_rate": 1.9716028563756628e-05, + "loss": 0.608, + "step": 1983 + }, + { + "epoch": 0.83, + "learning_rate": 1.971562726053611e-05, + "loss": 0.4624, + "step": 1984 + }, + { + "epoch": 0.83, + "learning_rate": 1.9715225678048515e-05, + "loss": 0.4922, + "step": 1985 + }, + { + "epoch": 0.83, + "learning_rate": 1.9714823816305376e-05, + "loss": 0.6225, + "step": 1986 + }, + { + "epoch": 0.83, + "learning_rate": 1.9714421675318247e-05, + "loss": 0.4787, + "step": 1987 + }, + { + "epoch": 0.83, + "learning_rate": 1.971401925509869e-05, + "loss": 0.4666, + "step": 1988 + }, + { + "epoch": 0.83, + "learning_rate": 1.9713616555658266e-05, + "loss": 0.5071, + "step": 1989 + }, + { + "epoch": 0.83, + "learning_rate": 1.9713213577008556e-05, + "loss": 0.4458, + "step": 1990 + }, + { + "epoch": 0.83, + "learning_rate": 1.971281031916114e-05, + "loss": 0.4941, + "step": 1991 + }, + { + "epoch": 0.83, + "learning_rate": 1.9712406782127612e-05, + "loss": 0.4653, + "step": 1992 + }, + { + "epoch": 0.83, + "learning_rate": 1.9712002965919567e-05, + "loss": 0.5734, + "step": 1993 + }, + { + "epoch": 0.83, + "learning_rate": 1.971159887054862e-05, + "loss": 0.535, + "step": 1994 + }, + { + "epoch": 0.84, + "learning_rate": 1.9711194496026372e-05, + "loss": 0.5193, + "step": 1995 + }, + { + "epoch": 0.84, + "learning_rate": 1.9710789842364465e-05, + "loss": 0.5112, + "step": 1996 + }, + { + "epoch": 0.84, + "learning_rate": 1.971038490957452e-05, + "loss": 0.4812, + "step": 1997 + }, + { + "epoch": 0.84, + "learning_rate": 1.9709979697668172e-05, + "loss": 0.4981, + "step": 1998 + }, + { + "epoch": 0.84, + "learning_rate": 1.970957420665708e-05, + "loss": 0.4752, + "step": 1999 + }, + { + "epoch": 0.84, + "learning_rate": 1.9709168436552887e-05, + "loss": 0.4864, + "step": 2000 + }, + { + "epoch": 0.84, + "learning_rate": 1.9708762387367272e-05, + "loss": 0.4548, + "step": 2001 + }, + { + "epoch": 0.84, + "learning_rate": 1.970835605911189e-05, + "loss": 0.511, + "step": 2002 + }, + { + "epoch": 0.84, + "learning_rate": 1.970794945179843e-05, + "loss": 0.5372, + "step": 2003 + }, + { + "epoch": 0.84, + "learning_rate": 1.970754256543858e-05, + "loss": 0.4788, + "step": 2004 + }, + { + "epoch": 0.84, + "learning_rate": 1.970713540004403e-05, + "loss": 0.5105, + "step": 2005 + }, + { + "epoch": 0.84, + "learning_rate": 1.970672795562649e-05, + "loss": 0.4853, + "step": 2006 + }, + { + "epoch": 0.84, + "learning_rate": 1.970632023219767e-05, + "loss": 0.5353, + "step": 2007 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705912229769285e-05, + "loss": 0.5399, + "step": 2008 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705503948353064e-05, + "loss": 0.4605, + "step": 2009 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705095387960746e-05, + "loss": 0.4799, + "step": 2010 + }, + { + "epoch": 0.84, + "learning_rate": 1.9704686548604076e-05, + "loss": 0.4942, + "step": 2011 + }, + { + "epoch": 0.84, + "learning_rate": 1.9704277430294798e-05, + "loss": 0.5017, + "step": 2012 + }, + { + "epoch": 0.84, + "learning_rate": 1.970386803304468e-05, + "loss": 0.5192, + "step": 2013 + }, + { + "epoch": 0.84, + "learning_rate": 1.9703458356865484e-05, + "loss": 0.4614, + "step": 2014 + }, + { + "epoch": 0.84, + "learning_rate": 1.9703048401768993e-05, + "loss": 0.4899, + "step": 2015 + }, + { + "epoch": 0.84, + "learning_rate": 1.970263816776698e-05, + "loss": 0.4431, + "step": 2016 + }, + { + "epoch": 0.84, + "learning_rate": 1.970222765487125e-05, + "loss": 0.5404, + "step": 2017 + }, + { + "epoch": 0.84, + "learning_rate": 1.970181686309359e-05, + "loss": 0.4927, + "step": 2018 + }, + { + "epoch": 0.85, + "learning_rate": 1.9701405792445815e-05, + "loss": 0.4271, + "step": 2019 + }, + { + "epoch": 0.85, + "learning_rate": 1.970099444293974e-05, + "loss": 0.4929, + "step": 2020 + }, + { + "epoch": 0.85, + "learning_rate": 1.970058281458719e-05, + "loss": 0.4559, + "step": 2021 + }, + { + "epoch": 0.85, + "learning_rate": 1.9700170907399992e-05, + "loss": 0.5275, + "step": 2022 + }, + { + "epoch": 0.85, + "learning_rate": 1.969975872138999e-05, + "loss": 0.5423, + "step": 2023 + }, + { + "epoch": 0.85, + "learning_rate": 1.9699346256569033e-05, + "loss": 0.4918, + "step": 2024 + }, + { + "epoch": 0.85, + "learning_rate": 1.9698933512948973e-05, + "loss": 0.4484, + "step": 2025 + }, + { + "epoch": 0.85, + "learning_rate": 1.9698520490541675e-05, + "loss": 0.4909, + "step": 2026 + }, + { + "epoch": 0.85, + "learning_rate": 1.969810718935901e-05, + "loss": 0.4184, + "step": 2027 + }, + { + "epoch": 0.85, + "learning_rate": 1.9697693609412864e-05, + "loss": 0.516, + "step": 2028 + }, + { + "epoch": 0.85, + "learning_rate": 1.9697279750715118e-05, + "loss": 0.6038, + "step": 2029 + }, + { + "epoch": 0.85, + "learning_rate": 1.9696865613277673e-05, + "loss": 0.5819, + "step": 2030 + }, + { + "epoch": 0.85, + "learning_rate": 1.969645119711243e-05, + "loss": 0.512, + "step": 2031 + }, + { + "epoch": 0.85, + "learning_rate": 1.96960365022313e-05, + "loss": 0.4182, + "step": 2032 + }, + { + "epoch": 0.85, + "learning_rate": 1.969562152864621e-05, + "loss": 0.5777, + "step": 2033 + }, + { + "epoch": 0.85, + "learning_rate": 1.9695206276369077e-05, + "loss": 0.481, + "step": 2034 + }, + { + "epoch": 0.85, + "learning_rate": 1.9694790745411846e-05, + "loss": 0.4749, + "step": 2035 + }, + { + "epoch": 0.85, + "learning_rate": 1.9694374935786457e-05, + "loss": 0.4691, + "step": 2036 + }, + { + "epoch": 0.85, + "learning_rate": 1.9693958847504863e-05, + "loss": 0.5651, + "step": 2037 + }, + { + "epoch": 0.85, + "learning_rate": 1.9693542480579025e-05, + "loss": 0.4447, + "step": 2038 + }, + { + "epoch": 0.85, + "learning_rate": 1.969312583502091e-05, + "loss": 0.4686, + "step": 2039 + }, + { + "epoch": 0.85, + "learning_rate": 1.9692708910842494e-05, + "loss": 0.5077, + "step": 2040 + }, + { + "epoch": 0.85, + "learning_rate": 1.9692291708055762e-05, + "loss": 0.5205, + "step": 2041 + }, + { + "epoch": 0.85, + "learning_rate": 1.9691874226672704e-05, + "loss": 0.4826, + "step": 2042 + }, + { + "epoch": 0.86, + "learning_rate": 1.9691456466705324e-05, + "loss": 0.5561, + "step": 2043 + }, + { + "epoch": 0.86, + "learning_rate": 1.969103842816563e-05, + "loss": 0.4831, + "step": 2044 + }, + { + "epoch": 0.86, + "learning_rate": 1.969062011106563e-05, + "loss": 0.546, + "step": 2045 + }, + { + "epoch": 0.86, + "learning_rate": 1.969020151541736e-05, + "loss": 0.5363, + "step": 2046 + }, + { + "epoch": 0.86, + "learning_rate": 1.968978264123284e-05, + "loss": 0.4719, + "step": 2047 + }, + { + "epoch": 0.86, + "learning_rate": 1.968936348852412e-05, + "loss": 0.5133, + "step": 2048 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688944057303247e-05, + "loss": 0.465, + "step": 2049 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688524347582272e-05, + "loss": 0.5911, + "step": 2050 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688104359373264e-05, + "loss": 0.486, + "step": 2051 + }, + { + "epoch": 0.86, + "learning_rate": 1.9687684092688292e-05, + "loss": 0.4789, + "step": 2052 + }, + { + "epoch": 0.86, + "learning_rate": 1.968726354753944e-05, + "loss": 0.4695, + "step": 2053 + }, + { + "epoch": 0.86, + "learning_rate": 1.968684272393879e-05, + "loss": 0.6171, + "step": 2054 + }, + { + "epoch": 0.86, + "learning_rate": 1.9686421621898443e-05, + "loss": 0.4325, + "step": 2055 + }, + { + "epoch": 0.86, + "learning_rate": 1.9686000241430503e-05, + "loss": 0.513, + "step": 2056 + }, + { + "epoch": 0.86, + "learning_rate": 1.968557858254708e-05, + "loss": 0.4462, + "step": 2057 + }, + { + "epoch": 0.86, + "learning_rate": 1.9685156645260296e-05, + "loss": 0.5465, + "step": 2058 + }, + { + "epoch": 0.86, + "learning_rate": 1.968473442958228e-05, + "loss": 0.4689, + "step": 2059 + }, + { + "epoch": 0.86, + "learning_rate": 1.9684311935525163e-05, + "loss": 0.5581, + "step": 2060 + }, + { + "epoch": 0.86, + "learning_rate": 1.9683889163101096e-05, + "loss": 0.4729, + "step": 2061 + }, + { + "epoch": 0.86, + "learning_rate": 1.968346611232223e-05, + "loss": 0.5558, + "step": 2062 + }, + { + "epoch": 0.86, + "learning_rate": 1.968304278320072e-05, + "loss": 0.5112, + "step": 2063 + }, + { + "epoch": 0.86, + "learning_rate": 1.968261917574874e-05, + "loss": 0.5189, + "step": 2064 + }, + { + "epoch": 0.86, + "learning_rate": 1.9682195289978466e-05, + "loss": 0.5116, + "step": 2065 + }, + { + "epoch": 0.86, + "learning_rate": 1.9681771125902075e-05, + "loss": 0.524, + "step": 2066 + }, + { + "epoch": 0.87, + "learning_rate": 1.9681346683531766e-05, + "loss": 0.5492, + "step": 2067 + }, + { + "epoch": 0.87, + "learning_rate": 1.968092196287974e-05, + "loss": 0.513, + "step": 2068 + }, + { + "epoch": 0.87, + "learning_rate": 1.96804969639582e-05, + "loss": 0.561, + "step": 2069 + }, + { + "epoch": 0.87, + "learning_rate": 1.9680071686779368e-05, + "loss": 0.5189, + "step": 2070 + }, + { + "epoch": 0.87, + "learning_rate": 1.9679646131355463e-05, + "loss": 0.5128, + "step": 2071 + }, + { + "epoch": 0.87, + "learning_rate": 1.9679220297698722e-05, + "loss": 0.45, + "step": 2072 + }, + { + "epoch": 0.87, + "learning_rate": 1.967879418582138e-05, + "loss": 0.5464, + "step": 2073 + }, + { + "epoch": 0.87, + "learning_rate": 1.967836779573569e-05, + "loss": 0.4759, + "step": 2074 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677941127453907e-05, + "loss": 0.5496, + "step": 2075 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677514180988292e-05, + "loss": 0.5275, + "step": 2076 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677086956351118e-05, + "loss": 0.5698, + "step": 2077 + }, + { + "epoch": 0.87, + "learning_rate": 1.967665945355467e-05, + "loss": 0.5322, + "step": 2078 + }, + { + "epoch": 0.87, + "learning_rate": 1.9676231672611233e-05, + "loss": 0.4737, + "step": 2079 + }, + { + "epoch": 0.87, + "learning_rate": 1.96758036135331e-05, + "loss": 0.51, + "step": 2080 + }, + { + "epoch": 0.87, + "learning_rate": 1.9675375276332583e-05, + "loss": 0.4943, + "step": 2081 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674946661021984e-05, + "loss": 0.4394, + "step": 2082 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674517767613633e-05, + "loss": 0.5582, + "step": 2083 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674088596119853e-05, + "loss": 0.4703, + "step": 2084 + }, + { + "epoch": 0.87, + "learning_rate": 1.967365914655298e-05, + "loss": 0.5181, + "step": 2085 + }, + { + "epoch": 0.87, + "learning_rate": 1.9673229418925358e-05, + "loss": 0.4687, + "step": 2086 + }, + { + "epoch": 0.87, + "learning_rate": 1.9672799413249343e-05, + "loss": 0.5428, + "step": 2087 + }, + { + "epoch": 0.87, + "learning_rate": 1.967236912953729e-05, + "loss": 0.5192, + "step": 2088 + }, + { + "epoch": 0.87, + "learning_rate": 1.9671938567801568e-05, + "loss": 0.5889, + "step": 2089 + }, + { + "epoch": 0.88, + "learning_rate": 1.9671507728054558e-05, + "loss": 0.4703, + "step": 2090 + }, + { + "epoch": 0.88, + "learning_rate": 1.9671076610308638e-05, + "loss": 0.5094, + "step": 2091 + }, + { + "epoch": 0.88, + "learning_rate": 1.9670645214576204e-05, + "loss": 0.5278, + "step": 2092 + }, + { + "epoch": 0.88, + "learning_rate": 1.9670213540869655e-05, + "loss": 0.6058, + "step": 2093 + }, + { + "epoch": 0.88, + "learning_rate": 1.96697815892014e-05, + "loss": 0.5246, + "step": 2094 + }, + { + "epoch": 0.88, + "learning_rate": 1.9669349359583852e-05, + "loss": 0.5534, + "step": 2095 + }, + { + "epoch": 0.88, + "learning_rate": 1.966891685202944e-05, + "loss": 0.4617, + "step": 2096 + }, + { + "epoch": 0.88, + "learning_rate": 1.9668484066550592e-05, + "loss": 0.519, + "step": 2097 + }, + { + "epoch": 0.88, + "learning_rate": 1.9668051003159746e-05, + "loss": 0.548, + "step": 2098 + }, + { + "epoch": 0.88, + "learning_rate": 1.9667617661869357e-05, + "loss": 0.4658, + "step": 2099 + }, + { + "epoch": 0.88, + "learning_rate": 1.9667184042691877e-05, + "loss": 0.4929, + "step": 2100 + }, + { + "epoch": 0.88, + "learning_rate": 1.966675014563977e-05, + "loss": 0.4961, + "step": 2101 + }, + { + "epoch": 0.88, + "learning_rate": 1.9666315970725507e-05, + "loss": 0.493, + "step": 2102 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665881517961568e-05, + "loss": 0.4827, + "step": 2103 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665446787360444e-05, + "loss": 0.4241, + "step": 2104 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665011778934628e-05, + "loss": 0.4034, + "step": 2105 + }, + { + "epoch": 0.88, + "learning_rate": 1.9664576492696628e-05, + "loss": 0.463, + "step": 2106 + }, + { + "epoch": 0.88, + "learning_rate": 1.9664140928658947e-05, + "loss": 0.5554, + "step": 2107 + }, + { + "epoch": 0.88, + "learning_rate": 1.9663705086834114e-05, + "loss": 0.481, + "step": 2108 + }, + { + "epoch": 0.88, + "learning_rate": 1.9663268967234655e-05, + "loss": 0.4523, + "step": 2109 + }, + { + "epoch": 0.88, + "learning_rate": 1.9662832569873106e-05, + "loss": 0.5055, + "step": 2110 + }, + { + "epoch": 0.88, + "learning_rate": 1.9662395894762007e-05, + "loss": 0.506, + "step": 2111 + }, + { + "epoch": 0.88, + "learning_rate": 1.9661958941913913e-05, + "loss": 0.4366, + "step": 2112 + }, + { + "epoch": 0.88, + "learning_rate": 1.966152171134138e-05, + "loss": 0.4667, + "step": 2113 + }, + { + "epoch": 0.89, + "learning_rate": 1.9661084203056983e-05, + "loss": 0.4995, + "step": 2114 + }, + { + "epoch": 0.89, + "learning_rate": 1.966064641707329e-05, + "loss": 0.5362, + "step": 2115 + }, + { + "epoch": 0.89, + "learning_rate": 1.9660208353402888e-05, + "loss": 0.5667, + "step": 2116 + }, + { + "epoch": 0.89, + "learning_rate": 1.9659770012058374e-05, + "loss": 0.5167, + "step": 2117 + }, + { + "epoch": 0.89, + "learning_rate": 1.9659331393052342e-05, + "loss": 0.5972, + "step": 2118 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658892496397398e-05, + "loss": 0.4921, + "step": 2119 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658453322106165e-05, + "loss": 0.5383, + "step": 2120 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658013870191258e-05, + "loss": 0.55, + "step": 2121 + }, + { + "epoch": 0.89, + "learning_rate": 1.965757414066531e-05, + "loss": 0.4841, + "step": 2122 + }, + { + "epoch": 0.89, + "learning_rate": 1.9657134133540966e-05, + "loss": 0.4447, + "step": 2123 + }, + { + "epoch": 0.89, + "learning_rate": 1.965669384883087e-05, + "loss": 0.5766, + "step": 2124 + }, + { + "epoch": 0.89, + "learning_rate": 1.9656253286547683e-05, + "loss": 0.53, + "step": 2125 + }, + { + "epoch": 0.89, + "learning_rate": 1.965581244670406e-05, + "loss": 0.5233, + "step": 2126 + }, + { + "epoch": 0.89, + "learning_rate": 1.9655371329312677e-05, + "loss": 0.4755, + "step": 2127 + }, + { + "epoch": 0.89, + "learning_rate": 1.9654929934386215e-05, + "loss": 0.4842, + "step": 2128 + }, + { + "epoch": 0.89, + "learning_rate": 1.965448826193736e-05, + "loss": 0.4566, + "step": 2129 + }, + { + "epoch": 0.89, + "learning_rate": 1.9654046311978803e-05, + "loss": 0.4718, + "step": 2130 + }, + { + "epoch": 0.89, + "learning_rate": 1.9653604084523255e-05, + "loss": 0.4792, + "step": 2131 + }, + { + "epoch": 0.89, + "learning_rate": 1.9653161579583423e-05, + "loss": 0.5052, + "step": 2132 + }, + { + "epoch": 0.89, + "learning_rate": 1.965271879717203e-05, + "loss": 0.5531, + "step": 2133 + }, + { + "epoch": 0.89, + "learning_rate": 1.9652275737301797e-05, + "loss": 0.5135, + "step": 2134 + }, + { + "epoch": 0.89, + "learning_rate": 1.9651832399985466e-05, + "loss": 0.5318, + "step": 2135 + }, + { + "epoch": 0.89, + "learning_rate": 1.9651388785235775e-05, + "loss": 0.548, + "step": 2136 + }, + { + "epoch": 0.89, + "learning_rate": 1.965094489306548e-05, + "loss": 0.4697, + "step": 2137 + }, + { + "epoch": 0.9, + "learning_rate": 1.9650500723487335e-05, + "loss": 0.5317, + "step": 2138 + }, + { + "epoch": 0.9, + "learning_rate": 1.9650056276514115e-05, + "loss": 0.4823, + "step": 2139 + }, + { + "epoch": 0.9, + "learning_rate": 1.964961155215859e-05, + "loss": 0.5023, + "step": 2140 + }, + { + "epoch": 0.9, + "learning_rate": 1.964916655043354e-05, + "loss": 0.4548, + "step": 2141 + }, + { + "epoch": 0.9, + "learning_rate": 1.9648721271351763e-05, + "loss": 0.48, + "step": 2142 + }, + { + "epoch": 0.9, + "learning_rate": 1.9648275714926055e-05, + "loss": 0.5174, + "step": 2143 + }, + { + "epoch": 0.9, + "learning_rate": 1.9647829881169223e-05, + "loss": 0.5626, + "step": 2144 + }, + { + "epoch": 0.9, + "learning_rate": 1.9647383770094083e-05, + "loss": 0.4561, + "step": 2145 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646937381713454e-05, + "loss": 0.4395, + "step": 2146 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646490716040174e-05, + "loss": 0.4942, + "step": 2147 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646043773087078e-05, + "loss": 0.4702, + "step": 2148 + }, + { + "epoch": 0.9, + "learning_rate": 1.9645596552867013e-05, + "loss": 0.4292, + "step": 2149 + }, + { + "epoch": 0.9, + "learning_rate": 1.9645149055392834e-05, + "loss": 0.5833, + "step": 2150 + }, + { + "epoch": 0.9, + "learning_rate": 1.9644701280677406e-05, + "loss": 0.4801, + "step": 2151 + }, + { + "epoch": 0.9, + "learning_rate": 1.9644253228733596e-05, + "loss": 0.4909, + "step": 2152 + }, + { + "epoch": 0.9, + "learning_rate": 1.9643804899574287e-05, + "loss": 0.5129, + "step": 2153 + }, + { + "epoch": 0.9, + "learning_rate": 1.9643356293212363e-05, + "loss": 0.5068, + "step": 2154 + }, + { + "epoch": 0.9, + "learning_rate": 1.964290740966072e-05, + "loss": 0.4858, + "step": 2155 + }, + { + "epoch": 0.9, + "learning_rate": 1.9642458248932262e-05, + "loss": 0.5096, + "step": 2156 + }, + { + "epoch": 0.9, + "learning_rate": 1.9642008811039893e-05, + "loss": 0.5049, + "step": 2157 + }, + { + "epoch": 0.9, + "learning_rate": 1.9641559095996544e-05, + "loss": 0.4885, + "step": 2158 + }, + { + "epoch": 0.9, + "learning_rate": 1.964110910381513e-05, + "loss": 0.4504, + "step": 2159 + }, + { + "epoch": 0.9, + "learning_rate": 1.964065883450859e-05, + "loss": 0.5344, + "step": 2160 + }, + { + "epoch": 0.9, + "learning_rate": 1.9640208288089872e-05, + "loss": 0.4163, + "step": 2161 + }, + { + "epoch": 0.91, + "learning_rate": 1.9639757464571914e-05, + "loss": 0.4949, + "step": 2162 + }, + { + "epoch": 0.91, + "learning_rate": 1.9639306363967686e-05, + "loss": 0.4949, + "step": 2163 + }, + { + "epoch": 0.91, + "learning_rate": 1.963885498629015e-05, + "loss": 0.5139, + "step": 2164 + }, + { + "epoch": 0.91, + "learning_rate": 1.9638403331552277e-05, + "loss": 0.4826, + "step": 2165 + }, + { + "epoch": 0.91, + "learning_rate": 1.963795139976706e-05, + "loss": 0.5396, + "step": 2166 + }, + { + "epoch": 0.91, + "learning_rate": 1.9637499190947478e-05, + "loss": 0.507, + "step": 2167 + }, + { + "epoch": 0.91, + "learning_rate": 1.9637046705106534e-05, + "loss": 0.5509, + "step": 2168 + }, + { + "epoch": 0.91, + "learning_rate": 1.9636593942257237e-05, + "loss": 0.4484, + "step": 2169 + }, + { + "epoch": 0.91, + "learning_rate": 1.9636140902412598e-05, + "loss": 0.4527, + "step": 2170 + }, + { + "epoch": 0.91, + "learning_rate": 1.9635687585585637e-05, + "loss": 0.524, + "step": 2171 + }, + { + "epoch": 0.91, + "learning_rate": 1.963523399178939e-05, + "loss": 0.4281, + "step": 2172 + }, + { + "epoch": 0.91, + "learning_rate": 1.9634780121036887e-05, + "loss": 0.3896, + "step": 2173 + }, + { + "epoch": 0.91, + "learning_rate": 1.9634325973341183e-05, + "loss": 0.5746, + "step": 2174 + }, + { + "epoch": 0.91, + "learning_rate": 1.963387154871533e-05, + "loss": 0.5507, + "step": 2175 + }, + { + "epoch": 0.91, + "learning_rate": 1.9633416847172384e-05, + "loss": 0.4694, + "step": 2176 + }, + { + "epoch": 0.91, + "learning_rate": 1.9632961868725424e-05, + "loss": 0.4982, + "step": 2177 + }, + { + "epoch": 0.91, + "learning_rate": 1.963250661338752e-05, + "loss": 0.4752, + "step": 2178 + }, + { + "epoch": 0.91, + "learning_rate": 1.963205108117176e-05, + "loss": 0.6507, + "step": 2179 + }, + { + "epoch": 0.91, + "learning_rate": 1.963159527209124e-05, + "loss": 0.5155, + "step": 2180 + }, + { + "epoch": 0.91, + "learning_rate": 1.9631139186159064e-05, + "loss": 0.6029, + "step": 2181 + }, + { + "epoch": 0.91, + "learning_rate": 1.9630682823388337e-05, + "loss": 0.5069, + "step": 2182 + }, + { + "epoch": 0.91, + "learning_rate": 1.9630226183792175e-05, + "loss": 0.5005, + "step": 2183 + }, + { + "epoch": 0.91, + "learning_rate": 1.962976926738371e-05, + "loss": 0.5772, + "step": 2184 + }, + { + "epoch": 0.91, + "learning_rate": 1.9629312074176073e-05, + "loss": 0.4924, + "step": 2185 + }, + { + "epoch": 0.92, + "learning_rate": 1.9628854604182403e-05, + "loss": 0.5, + "step": 2186 + }, + { + "epoch": 0.92, + "learning_rate": 1.9628396857415856e-05, + "loss": 0.4859, + "step": 2187 + }, + { + "epoch": 0.92, + "learning_rate": 1.962793883388958e-05, + "loss": 0.4494, + "step": 2188 + }, + { + "epoch": 0.92, + "learning_rate": 1.962748053361675e-05, + "loss": 0.4656, + "step": 2189 + }, + { + "epoch": 0.92, + "learning_rate": 1.9627021956610534e-05, + "loss": 0.475, + "step": 2190 + }, + { + "epoch": 0.92, + "learning_rate": 1.9626563102884117e-05, + "loss": 0.4323, + "step": 2191 + }, + { + "epoch": 0.92, + "learning_rate": 1.9626103972450683e-05, + "loss": 0.578, + "step": 2192 + }, + { + "epoch": 0.92, + "learning_rate": 1.962564456532343e-05, + "loss": 0.4718, + "step": 2193 + }, + { + "epoch": 0.92, + "learning_rate": 1.962518488151557e-05, + "loss": 0.4104, + "step": 2194 + }, + { + "epoch": 0.92, + "learning_rate": 1.962472492104031e-05, + "loss": 0.548, + "step": 2195 + }, + { + "epoch": 0.92, + "learning_rate": 1.9624264683910877e-05, + "loss": 0.555, + "step": 2196 + }, + { + "epoch": 0.92, + "learning_rate": 1.9623804170140494e-05, + "loss": 0.4559, + "step": 2197 + }, + { + "epoch": 0.92, + "learning_rate": 1.9623343379742397e-05, + "loss": 0.5425, + "step": 2198 + }, + { + "epoch": 0.92, + "learning_rate": 1.9622882312729835e-05, + "loss": 0.5496, + "step": 2199 + }, + { + "epoch": 0.92, + "learning_rate": 1.9622420969116062e-05, + "loss": 0.5709, + "step": 2200 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621959348914335e-05, + "loss": 0.5725, + "step": 2201 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621497452137927e-05, + "loss": 0.4704, + "step": 2202 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621035278800115e-05, + "loss": 0.5526, + "step": 2203 + }, + { + "epoch": 0.92, + "learning_rate": 1.9620572828914177e-05, + "loss": 0.5194, + "step": 2204 + }, + { + "epoch": 0.92, + "learning_rate": 1.9620110102493412e-05, + "loss": 0.551, + "step": 2205 + }, + { + "epoch": 0.92, + "learning_rate": 1.9619647099551118e-05, + "loss": 0.53, + "step": 2206 + }, + { + "epoch": 0.92, + "learning_rate": 1.9619183820100603e-05, + "loss": 0.4799, + "step": 2207 + }, + { + "epoch": 0.92, + "learning_rate": 1.9618720264155187e-05, + "loss": 0.4713, + "step": 2208 + }, + { + "epoch": 0.92, + "learning_rate": 1.961825643172819e-05, + "loss": 0.4345, + "step": 2209 + }, + { + "epoch": 0.93, + "learning_rate": 1.9617792322832952e-05, + "loss": 0.5059, + "step": 2210 + }, + { + "epoch": 0.93, + "learning_rate": 1.9617327937482805e-05, + "loss": 0.5279, + "step": 2211 + }, + { + "epoch": 0.93, + "learning_rate": 1.9616863275691102e-05, + "loss": 0.5578, + "step": 2212 + }, + { + "epoch": 0.93, + "learning_rate": 1.9616398337471196e-05, + "loss": 0.5448, + "step": 2213 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615933122836454e-05, + "loss": 0.5377, + "step": 2214 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615467631800248e-05, + "loss": 0.4938, + "step": 2215 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615001864375957e-05, + "loss": 0.4626, + "step": 2216 + }, + { + "epoch": 0.93, + "learning_rate": 1.9614535820576973e-05, + "loss": 0.5208, + "step": 2217 + }, + { + "epoch": 0.93, + "learning_rate": 1.9614069500416685e-05, + "loss": 0.494, + "step": 2218 + }, + { + "epoch": 0.93, + "learning_rate": 1.9613602903908503e-05, + "loss": 0.5347, + "step": 2219 + }, + { + "epoch": 0.93, + "learning_rate": 1.961313603106583e-05, + "loss": 0.6241, + "step": 2220 + }, + { + "epoch": 0.93, + "learning_rate": 1.96126688819021e-05, + "loss": 0.5039, + "step": 2221 + }, + { + "epoch": 0.93, + "learning_rate": 1.9612201456430732e-05, + "loss": 0.4839, + "step": 2222 + }, + { + "epoch": 0.93, + "learning_rate": 1.9611733754665157e-05, + "loss": 0.5787, + "step": 2223 + }, + { + "epoch": 0.93, + "learning_rate": 1.961126577661883e-05, + "loss": 0.5715, + "step": 2224 + }, + { + "epoch": 0.93, + "learning_rate": 1.9610797522305195e-05, + "loss": 0.4819, + "step": 2225 + }, + { + "epoch": 0.93, + "learning_rate": 1.9610328991737716e-05, + "loss": 0.4741, + "step": 2226 + }, + { + "epoch": 0.93, + "learning_rate": 1.9609860184929858e-05, + "loss": 0.4819, + "step": 2227 + }, + { + "epoch": 0.93, + "learning_rate": 1.9609391101895093e-05, + "loss": 0.5646, + "step": 2228 + }, + { + "epoch": 0.93, + "learning_rate": 1.9608921742646913e-05, + "loss": 0.4428, + "step": 2229 + }, + { + "epoch": 0.93, + "learning_rate": 1.9608452107198798e-05, + "loss": 0.4098, + "step": 2230 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607982195564258e-05, + "loss": 0.4311, + "step": 2231 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607512007756793e-05, + "loss": 0.503, + "step": 2232 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607041543789927e-05, + "loss": 0.5069, + "step": 2233 + }, + { + "epoch": 0.94, + "learning_rate": 1.960657080367717e-05, + "loss": 0.4942, + "step": 2234 + }, + { + "epoch": 0.94, + "learning_rate": 1.9606099787432066e-05, + "loss": 0.5272, + "step": 2235 + }, + { + "epoch": 0.94, + "learning_rate": 1.9605628495068143e-05, + "loss": 0.5224, + "step": 2236 + }, + { + "epoch": 0.94, + "learning_rate": 1.9605156926598953e-05, + "loss": 0.6102, + "step": 2237 + }, + { + "epoch": 0.94, + "learning_rate": 1.9604685082038053e-05, + "loss": 0.5718, + "step": 2238 + }, + { + "epoch": 0.94, + "learning_rate": 1.9604212961399004e-05, + "loss": 0.4199, + "step": 2239 + }, + { + "epoch": 0.94, + "learning_rate": 1.960374056469537e-05, + "loss": 0.5303, + "step": 2240 + }, + { + "epoch": 0.94, + "learning_rate": 1.9603267891940742e-05, + "loss": 0.4223, + "step": 2241 + }, + { + "epoch": 0.94, + "learning_rate": 1.9602794943148698e-05, + "loss": 0.6121, + "step": 2242 + }, + { + "epoch": 0.94, + "learning_rate": 1.9602321718332838e-05, + "loss": 0.5694, + "step": 2243 + }, + { + "epoch": 0.94, + "learning_rate": 1.9601848217506755e-05, + "loss": 0.5002, + "step": 2244 + }, + { + "epoch": 0.94, + "learning_rate": 1.960137444068407e-05, + "loss": 0.5088, + "step": 2245 + }, + { + "epoch": 0.94, + "learning_rate": 1.9600900387878397e-05, + "loss": 0.555, + "step": 2246 + }, + { + "epoch": 0.94, + "learning_rate": 1.9600426059103358e-05, + "loss": 0.5601, + "step": 2247 + }, + { + "epoch": 0.94, + "learning_rate": 1.9599951454372594e-05, + "loss": 0.4603, + "step": 2248 + }, + { + "epoch": 0.94, + "learning_rate": 1.9599476573699747e-05, + "loss": 0.5543, + "step": 2249 + }, + { + "epoch": 0.94, + "learning_rate": 1.959900141709846e-05, + "loss": 0.5286, + "step": 2250 + }, + { + "epoch": 0.94, + "learning_rate": 1.9598525984582398e-05, + "loss": 0.494, + "step": 2251 + }, + { + "epoch": 0.94, + "learning_rate": 1.959805027616522e-05, + "loss": 0.4904, + "step": 2252 + }, + { + "epoch": 0.94, + "learning_rate": 1.9597574291860607e-05, + "loss": 0.5076, + "step": 2253 + }, + { + "epoch": 0.94, + "learning_rate": 1.9597098031682238e-05, + "loss": 0.5447, + "step": 2254 + }, + { + "epoch": 0.94, + "learning_rate": 1.9596621495643805e-05, + "loss": 0.4445, + "step": 2255 + }, + { + "epoch": 0.94, + "learning_rate": 1.9596144683759e-05, + "loss": 0.5893, + "step": 2256 + }, + { + "epoch": 0.94, + "learning_rate": 1.959566759604153e-05, + "loss": 0.479, + "step": 2257 + }, + { + "epoch": 0.95, + "learning_rate": 1.959519023250511e-05, + "loss": 0.5299, + "step": 2258 + }, + { + "epoch": 0.95, + "learning_rate": 1.959471259316346e-05, + "loss": 0.4844, + "step": 2259 + }, + { + "epoch": 0.95, + "learning_rate": 1.9594234678030317e-05, + "loss": 0.4575, + "step": 2260 + }, + { + "epoch": 0.95, + "learning_rate": 1.959375648711941e-05, + "loss": 0.538, + "step": 2261 + }, + { + "epoch": 0.95, + "learning_rate": 1.9593278020444483e-05, + "loss": 0.5077, + "step": 2262 + }, + { + "epoch": 0.95, + "learning_rate": 1.9592799278019294e-05, + "loss": 0.5273, + "step": 2263 + }, + { + "epoch": 0.95, + "learning_rate": 1.9592320259857603e-05, + "loss": 0.4944, + "step": 2264 + }, + { + "epoch": 0.95, + "learning_rate": 1.9591840965973175e-05, + "loss": 0.525, + "step": 2265 + }, + { + "epoch": 0.95, + "learning_rate": 1.9591361396379795e-05, + "loss": 0.5158, + "step": 2266 + }, + { + "epoch": 0.95, + "learning_rate": 1.959088155109124e-05, + "loss": 0.5111, + "step": 2267 + }, + { + "epoch": 0.95, + "learning_rate": 1.9590401430121305e-05, + "loss": 0.4892, + "step": 2268 + }, + { + "epoch": 0.95, + "learning_rate": 1.9589921033483793e-05, + "loss": 0.5192, + "step": 2269 + }, + { + "epoch": 0.95, + "learning_rate": 1.958944036119251e-05, + "loss": 0.4518, + "step": 2270 + }, + { + "epoch": 0.95, + "learning_rate": 1.9588959413261272e-05, + "loss": 0.4856, + "step": 2271 + }, + { + "epoch": 0.95, + "learning_rate": 1.9588478189703908e-05, + "loss": 0.4939, + "step": 2272 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587996690534244e-05, + "loss": 0.5192, + "step": 2273 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587514915766124e-05, + "loss": 0.5328, + "step": 2274 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587032865413394e-05, + "loss": 0.4875, + "step": 2275 + }, + { + "epoch": 0.95, + "learning_rate": 1.9586550539489914e-05, + "loss": 0.4893, + "step": 2276 + }, + { + "epoch": 0.95, + "learning_rate": 1.9586067938009546e-05, + "loss": 0.4676, + "step": 2277 + }, + { + "epoch": 0.95, + "learning_rate": 1.9585585060986162e-05, + "loss": 0.4876, + "step": 2278 + }, + { + "epoch": 0.95, + "learning_rate": 1.958510190843364e-05, + "loss": 0.4743, + "step": 2279 + }, + { + "epoch": 0.95, + "learning_rate": 1.958461848036587e-05, + "loss": 0.5614, + "step": 2280 + }, + { + "epoch": 0.95, + "learning_rate": 1.958413477679675e-05, + "loss": 0.4286, + "step": 2281 + }, + { + "epoch": 0.96, + "learning_rate": 1.9583650797740173e-05, + "loss": 0.4811, + "step": 2282 + }, + { + "epoch": 0.96, + "learning_rate": 1.9583166543210065e-05, + "loss": 0.5306, + "step": 2283 + }, + { + "epoch": 0.96, + "learning_rate": 1.9582682013220337e-05, + "loss": 0.4897, + "step": 2284 + }, + { + "epoch": 0.96, + "learning_rate": 1.958219720778492e-05, + "loss": 0.4841, + "step": 2285 + }, + { + "epoch": 0.96, + "learning_rate": 1.9581712126917742e-05, + "loss": 0.5035, + "step": 2286 + }, + { + "epoch": 0.96, + "learning_rate": 1.9581226770632753e-05, + "loss": 0.4354, + "step": 2287 + }, + { + "epoch": 0.96, + "learning_rate": 1.9580741138943906e-05, + "loss": 0.461, + "step": 2288 + }, + { + "epoch": 0.96, + "learning_rate": 1.9580255231865156e-05, + "loss": 0.4559, + "step": 2289 + }, + { + "epoch": 0.96, + "learning_rate": 1.9579769049410468e-05, + "loss": 0.5464, + "step": 2290 + }, + { + "epoch": 0.96, + "learning_rate": 1.957928259159382e-05, + "loss": 0.4297, + "step": 2291 + }, + { + "epoch": 0.96, + "learning_rate": 1.95787958584292e-05, + "loss": 0.5206, + "step": 2292 + }, + { + "epoch": 0.96, + "learning_rate": 1.9578308849930586e-05, + "loss": 0.5657, + "step": 2293 + }, + { + "epoch": 0.96, + "learning_rate": 1.957782156611199e-05, + "loss": 0.4817, + "step": 2294 + }, + { + "epoch": 0.96, + "learning_rate": 1.9577334006987405e-05, + "loss": 0.5052, + "step": 2295 + }, + { + "epoch": 0.96, + "learning_rate": 1.9576846172570855e-05, + "loss": 0.5689, + "step": 2296 + }, + { + "epoch": 0.96, + "learning_rate": 1.9576358062876363e-05, + "loss": 0.5839, + "step": 2297 + }, + { + "epoch": 0.96, + "learning_rate": 1.9575869677917956e-05, + "loss": 0.4988, + "step": 2298 + }, + { + "epoch": 0.96, + "learning_rate": 1.957538101770967e-05, + "loss": 0.4993, + "step": 2299 + }, + { + "epoch": 0.96, + "learning_rate": 1.9574892082265557e-05, + "loss": 0.4793, + "step": 2300 + }, + { + "epoch": 0.96, + "learning_rate": 1.9574402871599666e-05, + "loss": 0.5144, + "step": 2301 + }, + { + "epoch": 0.96, + "learning_rate": 1.957391338572606e-05, + "loss": 0.5921, + "step": 2302 + }, + { + "epoch": 0.96, + "learning_rate": 1.957342362465881e-05, + "loss": 0.5145, + "step": 2303 + }, + { + "epoch": 0.96, + "learning_rate": 1.9572933588411997e-05, + "loss": 0.4551, + "step": 2304 + }, + { + "epoch": 0.97, + "learning_rate": 1.9572443276999696e-05, + "loss": 0.5018, + "step": 2305 + }, + { + "epoch": 0.97, + "learning_rate": 1.957195269043601e-05, + "loss": 0.5091, + "step": 2306 + }, + { + "epoch": 0.97, + "learning_rate": 1.957146182873504e-05, + "loss": 0.5403, + "step": 2307 + }, + { + "epoch": 0.97, + "learning_rate": 1.957097069191089e-05, + "loss": 0.5032, + "step": 2308 + }, + { + "epoch": 0.97, + "learning_rate": 1.9570479279977682e-05, + "loss": 0.5441, + "step": 2309 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569987592949543e-05, + "loss": 0.5361, + "step": 2310 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569495630840597e-05, + "loss": 0.4916, + "step": 2311 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569003393664996e-05, + "loss": 0.4256, + "step": 2312 + }, + { + "epoch": 0.97, + "learning_rate": 1.9568510881436883e-05, + "loss": 0.484, + "step": 2313 + }, + { + "epoch": 0.97, + "learning_rate": 1.9568018094170415e-05, + "loss": 0.4544, + "step": 2314 + }, + { + "epoch": 0.97, + "learning_rate": 1.9567525031879756e-05, + "loss": 0.4739, + "step": 2315 + }, + { + "epoch": 0.97, + "learning_rate": 1.956703169457908e-05, + "loss": 0.4568, + "step": 2316 + }, + { + "epoch": 0.97, + "learning_rate": 1.956653808228257e-05, + "loss": 0.5526, + "step": 2317 + }, + { + "epoch": 0.97, + "learning_rate": 1.956604419500441e-05, + "loss": 0.5471, + "step": 2318 + }, + { + "epoch": 0.97, + "learning_rate": 1.95655500327588e-05, + "loss": 0.4968, + "step": 2319 + }, + { + "epoch": 0.97, + "learning_rate": 1.956505559555994e-05, + "loss": 0.5308, + "step": 2320 + }, + { + "epoch": 0.97, + "learning_rate": 1.9564560883422045e-05, + "loss": 0.6437, + "step": 2321 + }, + { + "epoch": 0.97, + "learning_rate": 1.9564065896359337e-05, + "loss": 0.4779, + "step": 2322 + }, + { + "epoch": 0.97, + "learning_rate": 1.9563570634386042e-05, + "loss": 0.4739, + "step": 2323 + }, + { + "epoch": 0.97, + "learning_rate": 1.9563075097516397e-05, + "loss": 0.4708, + "step": 2324 + }, + { + "epoch": 0.97, + "learning_rate": 1.9562579285764638e-05, + "loss": 0.4772, + "step": 2325 + }, + { + "epoch": 0.97, + "learning_rate": 1.956208319914503e-05, + "loss": 0.5105, + "step": 2326 + }, + { + "epoch": 0.97, + "learning_rate": 1.956158683767182e-05, + "loss": 0.4218, + "step": 2327 + }, + { + "epoch": 0.97, + "learning_rate": 1.9561090201359285e-05, + "loss": 0.5393, + "step": 2328 + }, + { + "epoch": 0.98, + "learning_rate": 1.9560593290221695e-05, + "loss": 0.439, + "step": 2329 + }, + { + "epoch": 0.98, + "learning_rate": 1.9560096104273335e-05, + "loss": 0.4925, + "step": 2330 + }, + { + "epoch": 0.98, + "learning_rate": 1.95595986435285e-05, + "loss": 0.5582, + "step": 2331 + }, + { + "epoch": 0.98, + "learning_rate": 1.9559100908001478e-05, + "loss": 0.4094, + "step": 2332 + }, + { + "epoch": 0.98, + "learning_rate": 1.9558602897706586e-05, + "loss": 0.5477, + "step": 2333 + }, + { + "epoch": 0.98, + "learning_rate": 1.9558104612658137e-05, + "loss": 0.5561, + "step": 2334 + }, + { + "epoch": 0.98, + "learning_rate": 1.9557606052870452e-05, + "loss": 0.4986, + "step": 2335 + }, + { + "epoch": 0.98, + "learning_rate": 1.9557107218357862e-05, + "loss": 0.4391, + "step": 2336 + }, + { + "epoch": 0.98, + "learning_rate": 1.9556608109134707e-05, + "loss": 0.432, + "step": 2337 + }, + { + "epoch": 0.98, + "learning_rate": 1.955610872521533e-05, + "loss": 0.4751, + "step": 2338 + }, + { + "epoch": 0.98, + "learning_rate": 1.955560906661409e-05, + "loss": 0.5742, + "step": 2339 + }, + { + "epoch": 0.98, + "learning_rate": 1.9555109133345347e-05, + "loss": 0.5143, + "step": 2340 + }, + { + "epoch": 0.98, + "learning_rate": 1.9554608925423467e-05, + "loss": 0.4496, + "step": 2341 + }, + { + "epoch": 0.98, + "learning_rate": 1.9554108442862836e-05, + "loss": 0.4794, + "step": 2342 + }, + { + "epoch": 0.98, + "learning_rate": 1.9553607685677834e-05, + "loss": 0.4721, + "step": 2343 + }, + { + "epoch": 0.98, + "learning_rate": 1.955310665388286e-05, + "loss": 0.5335, + "step": 2344 + }, + { + "epoch": 0.98, + "learning_rate": 1.955260534749231e-05, + "loss": 0.4888, + "step": 2345 + }, + { + "epoch": 0.98, + "learning_rate": 1.9552103766520596e-05, + "loss": 0.452, + "step": 2346 + }, + { + "epoch": 0.98, + "learning_rate": 1.9551601910982136e-05, + "loss": 0.5632, + "step": 2347 + }, + { + "epoch": 0.98, + "learning_rate": 1.9551099780891353e-05, + "loss": 0.4978, + "step": 2348 + }, + { + "epoch": 0.98, + "learning_rate": 1.9550597376262685e-05, + "loss": 0.5115, + "step": 2349 + }, + { + "epoch": 0.98, + "learning_rate": 1.955009469711057e-05, + "loss": 0.5622, + "step": 2350 + }, + { + "epoch": 0.98, + "learning_rate": 1.954959174344946e-05, + "loss": 0.5278, + "step": 2351 + }, + { + "epoch": 0.98, + "learning_rate": 1.9549088515293804e-05, + "loss": 0.4838, + "step": 2352 + }, + { + "epoch": 0.99, + "learning_rate": 1.9548585012658074e-05, + "loss": 0.5064, + "step": 2353 + }, + { + "epoch": 0.99, + "learning_rate": 1.954808123555674e-05, + "loss": 0.5339, + "step": 2354 + }, + { + "epoch": 0.99, + "learning_rate": 1.954757718400429e-05, + "loss": 0.4907, + "step": 2355 + }, + { + "epoch": 0.99, + "learning_rate": 1.95470728580152e-05, + "loss": 0.485, + "step": 2356 + }, + { + "epoch": 0.99, + "learning_rate": 1.9546568257603976e-05, + "loss": 0.4973, + "step": 2357 + }, + { + "epoch": 0.99, + "learning_rate": 1.9546063382785117e-05, + "loss": 0.5277, + "step": 2358 + }, + { + "epoch": 0.99, + "learning_rate": 1.9545558233573136e-05, + "loss": 0.4577, + "step": 2359 + }, + { + "epoch": 0.99, + "learning_rate": 1.9545052809982556e-05, + "loss": 0.4347, + "step": 2360 + }, + { + "epoch": 0.99, + "learning_rate": 1.9544547112027904e-05, + "loss": 0.4502, + "step": 2361 + }, + { + "epoch": 0.99, + "learning_rate": 1.9544041139723712e-05, + "loss": 0.446, + "step": 2362 + }, + { + "epoch": 0.99, + "learning_rate": 1.954353489308453e-05, + "loss": 0.618, + "step": 2363 + }, + { + "epoch": 0.99, + "learning_rate": 1.9543028372124903e-05, + "loss": 0.4332, + "step": 2364 + }, + { + "epoch": 0.99, + "learning_rate": 1.9542521576859397e-05, + "loss": 0.4302, + "step": 2365 + }, + { + "epoch": 0.99, + "learning_rate": 1.9542014507302576e-05, + "loss": 0.5073, + "step": 2366 + }, + { + "epoch": 0.99, + "learning_rate": 1.9541507163469012e-05, + "loss": 0.4632, + "step": 2367 + }, + { + "epoch": 0.99, + "learning_rate": 1.9540999545373295e-05, + "loss": 0.5273, + "step": 2368 + }, + { + "epoch": 0.99, + "learning_rate": 1.954049165303001e-05, + "loss": 0.4725, + "step": 2369 + }, + { + "epoch": 0.99, + "learning_rate": 1.9539983486453762e-05, + "loss": 0.4652, + "step": 2370 + }, + { + "epoch": 0.99, + "learning_rate": 1.953947504565915e-05, + "loss": 0.4843, + "step": 2371 + }, + { + "epoch": 0.99, + "learning_rate": 1.9538966330660796e-05, + "loss": 0.511, + "step": 2372 + }, + { + "epoch": 0.99, + "learning_rate": 1.9538457341473318e-05, + "loss": 0.5827, + "step": 2373 + }, + { + "epoch": 0.99, + "learning_rate": 1.9537948078111348e-05, + "loss": 0.4844, + "step": 2374 + }, + { + "epoch": 0.99, + "learning_rate": 1.9537438540589526e-05, + "loss": 0.5425, + "step": 2375 + }, + { + "epoch": 0.99, + "learning_rate": 1.9536928728922496e-05, + "loss": 0.4983, + "step": 2376 + }, + { + "epoch": 1.0, + "learning_rate": 1.953641864312491e-05, + "loss": 0.5085, + "step": 2377 + }, + { + "epoch": 1.0, + "learning_rate": 1.9535908283211435e-05, + "loss": 0.5282, + "step": 2378 + }, + { + "epoch": 1.0, + "learning_rate": 1.9535397649196738e-05, + "loss": 0.4554, + "step": 2379 + }, + { + "epoch": 1.0, + "learning_rate": 1.95348867410955e-05, + "loss": 0.4774, + "step": 2380 + }, + { + "epoch": 1.0, + "learning_rate": 1.95343755589224e-05, + "loss": 0.6091, + "step": 2381 + }, + { + "epoch": 1.0, + "learning_rate": 1.9533864102692134e-05, + "loss": 0.6011, + "step": 2382 + }, + { + "epoch": 1.0, + "learning_rate": 1.9533352372419408e-05, + "loss": 0.4555, + "step": 2383 + }, + { + "epoch": 1.0, + "learning_rate": 1.953284036811893e-05, + "loss": 0.435, + "step": 2384 + }, + { + "epoch": 1.0, + "learning_rate": 1.9532328089805407e-05, + "loss": 0.5451, + "step": 2385 + }, + { + "epoch": 1.0, + "learning_rate": 1.953181553749358e-05, + "loss": 0.4929, + "step": 2386 + }, + { + "epoch": 1.0, + "learning_rate": 1.953130271119817e-05, + "loss": 0.5192, + "step": 2387 + }, + { + "epoch": 1.0, + "learning_rate": 1.953078961093392e-05, + "loss": 0.5065, + "step": 2388 + }, + { + "epoch": 1.0, + "learning_rate": 1.953027623671558e-05, + "loss": 0.4521, + "step": 2389 + }, + { + "epoch": 1.0, + "learning_rate": 1.952976258855791e-05, + "loss": 0.3904, + "step": 2390 + }, + { + "epoch": 1.0, + "learning_rate": 1.9529248666475668e-05, + "loss": 0.3474, + "step": 2391 + }, + { + "epoch": 1.0, + "learning_rate": 1.9528734470483632e-05, + "loss": 0.3506, + "step": 2392 + }, + { + "epoch": 1.0, + "learning_rate": 1.9528220000596577e-05, + "loss": 0.399, + "step": 2393 + }, + { + "epoch": 1.0, + "learning_rate": 1.9527705256829293e-05, + "loss": 0.3954, + "step": 2394 + }, + { + "epoch": 1.0, + "learning_rate": 1.952719023919658e-05, + "loss": 0.3772, + "step": 2395 + }, + { + "epoch": 1.0, + "learning_rate": 1.9526674947713234e-05, + "loss": 0.4016, + "step": 2396 + }, + { + "epoch": 1.0, + "learning_rate": 1.9526159382394068e-05, + "loss": 0.3883, + "step": 2397 + }, + { + "epoch": 1.0, + "learning_rate": 1.952564354325391e-05, + "loss": 0.3803, + "step": 2398 + }, + { + "epoch": 1.0, + "learning_rate": 1.952512743030758e-05, + "loss": 0.3478, + "step": 2399 + }, + { + "epoch": 1.0, + "learning_rate": 1.952461104356991e-05, + "loss": 0.4181, + "step": 2400 + }, + { + "epoch": 1.01, + "learning_rate": 1.952409438305575e-05, + "loss": 0.3764, + "step": 2401 + }, + { + "epoch": 1.01, + "learning_rate": 1.952357744877995e-05, + "loss": 0.412, + "step": 2402 + }, + { + "epoch": 1.01, + "learning_rate": 1.952306024075737e-05, + "loss": 0.3514, + "step": 2403 + }, + { + "epoch": 1.01, + "learning_rate": 1.9522542759002866e-05, + "loss": 0.3896, + "step": 2404 + }, + { + "epoch": 1.01, + "learning_rate": 1.9522025003531327e-05, + "loss": 0.3816, + "step": 2405 + }, + { + "epoch": 1.01, + "learning_rate": 1.9521506974357628e-05, + "loss": 0.4014, + "step": 2406 + }, + { + "epoch": 1.01, + "learning_rate": 1.952098867149666e-05, + "loss": 0.3944, + "step": 2407 + }, + { + "epoch": 1.01, + "learning_rate": 1.952047009496332e-05, + "loss": 0.376, + "step": 2408 + }, + { + "epoch": 1.01, + "learning_rate": 1.9519951244772514e-05, + "loss": 0.363, + "step": 2409 + }, + { + "epoch": 1.01, + "learning_rate": 1.951943212093916e-05, + "loss": 0.4244, + "step": 2410 + }, + { + "epoch": 1.01, + "learning_rate": 1.9518912723478176e-05, + "loss": 0.4064, + "step": 2411 + }, + { + "epoch": 1.01, + "learning_rate": 1.9518393052404495e-05, + "loss": 0.3723, + "step": 2412 + }, + { + "epoch": 1.01, + "learning_rate": 1.951787310773305e-05, + "loss": 0.3509, + "step": 2413 + }, + { + "epoch": 1.01, + "learning_rate": 1.9517352889478787e-05, + "loss": 0.3573, + "step": 2414 + }, + { + "epoch": 1.01, + "learning_rate": 1.9516832397656663e-05, + "loss": 0.3915, + "step": 2415 + }, + { + "epoch": 1.01, + "learning_rate": 1.9516311632281636e-05, + "loss": 0.3676, + "step": 2416 + }, + { + "epoch": 1.01, + "learning_rate": 1.9515790593368677e-05, + "loss": 0.4132, + "step": 2417 + }, + { + "epoch": 1.01, + "learning_rate": 1.951526928093276e-05, + "loss": 0.4165, + "step": 2418 + }, + { + "epoch": 1.01, + "learning_rate": 1.9514747694988873e-05, + "loss": 0.3968, + "step": 2419 + }, + { + "epoch": 1.01, + "learning_rate": 1.9514225835552004e-05, + "loss": 0.3525, + "step": 2420 + }, + { + "epoch": 1.01, + "learning_rate": 1.9513703702637158e-05, + "loss": 0.3924, + "step": 2421 + }, + { + "epoch": 1.01, + "learning_rate": 1.951318129625934e-05, + "loss": 0.3927, + "step": 2422 + }, + { + "epoch": 1.01, + "learning_rate": 1.951265861643357e-05, + "loss": 0.4062, + "step": 2423 + }, + { + "epoch": 1.01, + "learning_rate": 1.951213566317487e-05, + "loss": 0.3723, + "step": 2424 + }, + { + "epoch": 1.02, + "learning_rate": 1.9511612436498267e-05, + "loss": 0.3841, + "step": 2425 + }, + { + "epoch": 1.02, + "learning_rate": 1.9511088936418807e-05, + "loss": 0.3915, + "step": 2426 + }, + { + "epoch": 1.02, + "learning_rate": 1.9510565162951538e-05, + "loss": 0.4367, + "step": 2427 + }, + { + "epoch": 1.02, + "learning_rate": 1.951004111611151e-05, + "loss": 0.3407, + "step": 2428 + }, + { + "epoch": 1.02, + "learning_rate": 1.950951679591379e-05, + "loss": 0.3957, + "step": 2429 + }, + { + "epoch": 1.02, + "learning_rate": 1.9508992202373448e-05, + "loss": 0.3546, + "step": 2430 + }, + { + "epoch": 1.02, + "learning_rate": 1.9508467335505564e-05, + "loss": 0.3823, + "step": 2431 + }, + { + "epoch": 1.02, + "learning_rate": 1.9507942195325223e-05, + "loss": 0.3666, + "step": 2432 + }, + { + "epoch": 1.02, + "learning_rate": 1.9507416781847524e-05, + "loss": 0.4182, + "step": 2433 + }, + { + "epoch": 1.02, + "learning_rate": 1.9506891095087563e-05, + "loss": 0.4023, + "step": 2434 + }, + { + "epoch": 1.02, + "learning_rate": 1.950636513506045e-05, + "loss": 0.4253, + "step": 2435 + }, + { + "epoch": 1.02, + "learning_rate": 1.9505838901781314e-05, + "loss": 0.3536, + "step": 2436 + }, + { + "epoch": 1.02, + "learning_rate": 1.9505312395265273e-05, + "loss": 0.3591, + "step": 2437 + }, + { + "epoch": 1.02, + "learning_rate": 1.950478561552746e-05, + "loss": 0.3788, + "step": 2438 + }, + { + "epoch": 1.02, + "learning_rate": 1.9504258562583018e-05, + "loss": 0.3627, + "step": 2439 + }, + { + "epoch": 1.02, + "learning_rate": 1.95037312364471e-05, + "loss": 0.3955, + "step": 2440 + }, + { + "epoch": 1.02, + "learning_rate": 1.950320363713486e-05, + "loss": 0.3654, + "step": 2441 + }, + { + "epoch": 1.02, + "learning_rate": 1.9502675764661464e-05, + "loss": 0.3787, + "step": 2442 + }, + { + "epoch": 1.02, + "learning_rate": 1.9502147619042086e-05, + "loss": 0.3312, + "step": 2443 + }, + { + "epoch": 1.02, + "learning_rate": 1.950161920029191e-05, + "loss": 0.3557, + "step": 2444 + }, + { + "epoch": 1.02, + "learning_rate": 1.9501090508426117e-05, + "loss": 0.3739, + "step": 2445 + }, + { + "epoch": 1.02, + "learning_rate": 1.950056154345991e-05, + "loss": 0.3747, + "step": 2446 + }, + { + "epoch": 1.02, + "learning_rate": 1.9500032305408494e-05, + "loss": 0.3263, + "step": 2447 + }, + { + "epoch": 1.02, + "learning_rate": 1.9499502794287078e-05, + "loss": 0.3685, + "step": 2448 + }, + { + "epoch": 1.03, + "learning_rate": 1.9498973010110884e-05, + "loss": 0.3698, + "step": 2449 + }, + { + "epoch": 1.03, + "learning_rate": 1.949844295289514e-05, + "loss": 0.3479, + "step": 2450 + }, + { + "epoch": 1.03, + "learning_rate": 1.949791262265508e-05, + "loss": 0.3987, + "step": 2451 + }, + { + "epoch": 1.03, + "learning_rate": 1.949738201940595e-05, + "loss": 0.3598, + "step": 2452 + }, + { + "epoch": 1.03, + "learning_rate": 1.9496851143163004e-05, + "loss": 0.3291, + "step": 2453 + }, + { + "epoch": 1.03, + "learning_rate": 1.9496319993941498e-05, + "loss": 0.4026, + "step": 2454 + }, + { + "epoch": 1.03, + "learning_rate": 1.9495788571756704e-05, + "loss": 0.3992, + "step": 2455 + }, + { + "epoch": 1.03, + "learning_rate": 1.949525687662389e-05, + "loss": 0.3673, + "step": 2456 + }, + { + "epoch": 1.03, + "learning_rate": 1.9494724908558345e-05, + "loss": 0.391, + "step": 2457 + }, + { + "epoch": 1.03, + "learning_rate": 1.9494192667575355e-05, + "loss": 0.3577, + "step": 2458 + }, + { + "epoch": 1.03, + "learning_rate": 1.9493660153690225e-05, + "loss": 0.3866, + "step": 2459 + }, + { + "epoch": 1.03, + "learning_rate": 1.949312736691826e-05, + "loss": 0.3643, + "step": 2460 + }, + { + "epoch": 1.03, + "learning_rate": 1.9492594307274772e-05, + "loss": 0.3734, + "step": 2461 + }, + { + "epoch": 1.03, + "learning_rate": 1.949206097477508e-05, + "loss": 0.4079, + "step": 2462 + }, + { + "epoch": 1.03, + "learning_rate": 1.949152736943452e-05, + "loss": 0.3489, + "step": 2463 + }, + { + "epoch": 1.03, + "learning_rate": 1.949099349126843e-05, + "loss": 0.3724, + "step": 2464 + }, + { + "epoch": 1.03, + "learning_rate": 1.9490459340292153e-05, + "loss": 0.3742, + "step": 2465 + }, + { + "epoch": 1.03, + "learning_rate": 1.9489924916521047e-05, + "loss": 0.3711, + "step": 2466 + }, + { + "epoch": 1.03, + "learning_rate": 1.9489390219970468e-05, + "loss": 0.3826, + "step": 2467 + }, + { + "epoch": 1.03, + "learning_rate": 1.9488855250655785e-05, + "loss": 0.3584, + "step": 2468 + }, + { + "epoch": 1.03, + "learning_rate": 1.948832000859238e-05, + "loss": 0.3569, + "step": 2469 + }, + { + "epoch": 1.03, + "learning_rate": 1.9487784493795637e-05, + "loss": 0.3772, + "step": 2470 + }, + { + "epoch": 1.03, + "learning_rate": 1.948724870628095e-05, + "loss": 0.3452, + "step": 2471 + }, + { + "epoch": 1.03, + "learning_rate": 1.9486712646063715e-05, + "loss": 0.4041, + "step": 2472 + }, + { + "epoch": 1.04, + "learning_rate": 1.9486176313159344e-05, + "loss": 0.3882, + "step": 2473 + }, + { + "epoch": 1.04, + "learning_rate": 1.9485639707583255e-05, + "loss": 0.4075, + "step": 2474 + }, + { + "epoch": 1.04, + "learning_rate": 1.948510282935087e-05, + "loss": 0.355, + "step": 2475 + }, + { + "epoch": 1.04, + "learning_rate": 1.9484565678477616e-05, + "loss": 0.3827, + "step": 2476 + }, + { + "epoch": 1.04, + "learning_rate": 1.9484028254978942e-05, + "loss": 0.3761, + "step": 2477 + }, + { + "epoch": 1.04, + "learning_rate": 1.948349055887029e-05, + "loss": 0.4096, + "step": 2478 + }, + { + "epoch": 1.04, + "learning_rate": 1.9482952590167118e-05, + "loss": 0.3582, + "step": 2479 + }, + { + "epoch": 1.04, + "learning_rate": 1.9482414348884892e-05, + "loss": 0.3936, + "step": 2480 + }, + { + "epoch": 1.04, + "learning_rate": 1.9481875835039075e-05, + "loss": 0.3534, + "step": 2481 + }, + { + "epoch": 1.04, + "learning_rate": 1.9481337048645154e-05, + "loss": 0.3723, + "step": 2482 + }, + { + "epoch": 1.04, + "learning_rate": 1.9480797989718612e-05, + "loss": 0.4351, + "step": 2483 + }, + { + "epoch": 1.04, + "learning_rate": 1.9480258658274947e-05, + "loss": 0.3678, + "step": 2484 + }, + { + "epoch": 1.04, + "learning_rate": 1.9479719054329657e-05, + "loss": 0.3977, + "step": 2485 + }, + { + "epoch": 1.04, + "learning_rate": 1.9479179177898255e-05, + "loss": 0.3488, + "step": 2486 + }, + { + "epoch": 1.04, + "learning_rate": 1.9478639028996262e-05, + "loss": 0.3655, + "step": 2487 + }, + { + "epoch": 1.04, + "learning_rate": 1.9478098607639197e-05, + "loss": 0.4027, + "step": 2488 + }, + { + "epoch": 1.04, + "learning_rate": 1.94775579138426e-05, + "loss": 0.3937, + "step": 2489 + }, + { + "epoch": 1.04, + "learning_rate": 1.9477016947622013e-05, + "loss": 0.354, + "step": 2490 + }, + { + "epoch": 1.04, + "learning_rate": 1.947647570899298e-05, + "loss": 0.4085, + "step": 2491 + }, + { + "epoch": 1.04, + "learning_rate": 1.947593419797107e-05, + "loss": 0.3629, + "step": 2492 + }, + { + "epoch": 1.04, + "learning_rate": 1.947539241457183e-05, + "loss": 0.3365, + "step": 2493 + }, + { + "epoch": 1.04, + "learning_rate": 1.947485035881085e-05, + "loss": 0.3338, + "step": 2494 + }, + { + "epoch": 1.04, + "learning_rate": 1.94743080307037e-05, + "loss": 0.4101, + "step": 2495 + }, + { + "epoch": 1.05, + "learning_rate": 1.947376543026598e-05, + "loss": 0.3708, + "step": 2496 + }, + { + "epoch": 1.05, + "learning_rate": 1.9473222557513273e-05, + "loss": 0.3908, + "step": 2497 + }, + { + "epoch": 1.05, + "learning_rate": 1.947267941246119e-05, + "loss": 0.3852, + "step": 2498 + }, + { + "epoch": 1.05, + "learning_rate": 1.9472135995125344e-05, + "loss": 0.3899, + "step": 2499 + }, + { + "epoch": 1.05, + "learning_rate": 1.9471592305521355e-05, + "loss": 0.4517, + "step": 2500 + }, + { + "epoch": 1.05, + "learning_rate": 1.9471048343664853e-05, + "loss": 0.3975, + "step": 2501 + }, + { + "epoch": 1.05, + "learning_rate": 1.9470504109571467e-05, + "loss": 0.4205, + "step": 2502 + }, + { + "epoch": 1.05, + "learning_rate": 1.9469959603256844e-05, + "loss": 0.3591, + "step": 2503 + }, + { + "epoch": 1.05, + "learning_rate": 1.946941482473664e-05, + "loss": 0.3632, + "step": 2504 + }, + { + "epoch": 1.05, + "learning_rate": 1.9468869774026508e-05, + "loss": 0.4138, + "step": 2505 + }, + { + "epoch": 1.05, + "learning_rate": 1.9468324451142115e-05, + "loss": 0.3705, + "step": 2506 + }, + { + "epoch": 1.05, + "learning_rate": 1.9467778856099134e-05, + "loss": 0.3899, + "step": 2507 + }, + { + "epoch": 1.05, + "learning_rate": 1.946723298891326e-05, + "loss": 0.342, + "step": 2508 + }, + { + "epoch": 1.05, + "learning_rate": 1.9466686849600168e-05, + "loss": 0.4147, + "step": 2509 + }, + { + "epoch": 1.05, + "learning_rate": 1.9466140438175567e-05, + "loss": 0.391, + "step": 2510 + }, + { + "epoch": 1.05, + "learning_rate": 1.9465593754655158e-05, + "loss": 0.3563, + "step": 2511 + }, + { + "epoch": 1.05, + "learning_rate": 1.9465046799054657e-05, + "loss": 0.4002, + "step": 2512 + }, + { + "epoch": 1.05, + "learning_rate": 1.946449957138978e-05, + "loss": 0.4012, + "step": 2513 + }, + { + "epoch": 1.05, + "learning_rate": 1.9463952071676268e-05, + "loss": 0.3771, + "step": 2514 + }, + { + "epoch": 1.05, + "learning_rate": 1.9463404299929852e-05, + "loss": 0.3488, + "step": 2515 + }, + { + "epoch": 1.05, + "learning_rate": 1.9462856256166273e-05, + "loss": 0.379, + "step": 2516 + }, + { + "epoch": 1.05, + "learning_rate": 1.9462307940401288e-05, + "loss": 0.4167, + "step": 2517 + }, + { + "epoch": 1.05, + "learning_rate": 1.9461759352650663e-05, + "loss": 0.3738, + "step": 2518 + }, + { + "epoch": 1.05, + "learning_rate": 1.9461210492930156e-05, + "loss": 0.3786, + "step": 2519 + }, + { + "epoch": 1.06, + "learning_rate": 1.9460661361255552e-05, + "loss": 0.422, + "step": 2520 + }, + { + "epoch": 1.06, + "learning_rate": 1.946011195764263e-05, + "loss": 0.3372, + "step": 2521 + }, + { + "epoch": 1.06, + "learning_rate": 1.9459562282107188e-05, + "loss": 0.3704, + "step": 2522 + }, + { + "epoch": 1.06, + "learning_rate": 1.945901233466502e-05, + "loss": 0.3586, + "step": 2523 + }, + { + "epoch": 1.06, + "learning_rate": 1.9458462115331938e-05, + "loss": 0.3876, + "step": 2524 + }, + { + "epoch": 1.06, + "learning_rate": 1.9457911624123754e-05, + "loss": 0.3911, + "step": 2525 + }, + { + "epoch": 1.06, + "learning_rate": 1.9457360861056294e-05, + "loss": 0.3998, + "step": 2526 + }, + { + "epoch": 1.06, + "learning_rate": 1.9456809826145387e-05, + "loss": 0.3516, + "step": 2527 + }, + { + "epoch": 1.06, + "learning_rate": 1.9456258519406875e-05, + "loss": 0.3591, + "step": 2528 + }, + { + "epoch": 1.06, + "learning_rate": 1.9455706940856602e-05, + "loss": 0.4332, + "step": 2529 + }, + { + "epoch": 1.06, + "learning_rate": 1.9455155090510427e-05, + "loss": 0.4036, + "step": 2530 + }, + { + "epoch": 1.06, + "learning_rate": 1.9454602968384204e-05, + "loss": 0.3703, + "step": 2531 + }, + { + "epoch": 1.06, + "learning_rate": 1.9454050574493815e-05, + "loss": 0.3515, + "step": 2532 + }, + { + "epoch": 1.06, + "learning_rate": 1.9453497908855125e-05, + "loss": 0.4121, + "step": 2533 + }, + { + "epoch": 1.06, + "learning_rate": 1.945294497148403e-05, + "loss": 0.3881, + "step": 2534 + }, + { + "epoch": 1.06, + "learning_rate": 1.945239176239642e-05, + "loss": 0.3798, + "step": 2535 + }, + { + "epoch": 1.06, + "learning_rate": 1.94518382816082e-05, + "loss": 0.3696, + "step": 2536 + }, + { + "epoch": 1.06, + "learning_rate": 1.945128452913527e-05, + "loss": 0.3494, + "step": 2537 + }, + { + "epoch": 1.06, + "learning_rate": 1.9450730504993554e-05, + "loss": 0.4292, + "step": 2538 + }, + { + "epoch": 1.06, + "learning_rate": 1.9450176209198977e-05, + "loss": 0.3643, + "step": 2539 + }, + { + "epoch": 1.06, + "learning_rate": 1.9449621641767474e-05, + "loss": 0.4396, + "step": 2540 + }, + { + "epoch": 1.06, + "learning_rate": 1.944906680271498e-05, + "loss": 0.3905, + "step": 2541 + }, + { + "epoch": 1.06, + "learning_rate": 1.9448511692057446e-05, + "loss": 0.377, + "step": 2542 + }, + { + "epoch": 1.06, + "learning_rate": 1.9447956309810825e-05, + "loss": 0.3605, + "step": 2543 + }, + { + "epoch": 1.07, + "learning_rate": 1.9447400655991086e-05, + "loss": 0.4316, + "step": 2544 + }, + { + "epoch": 1.07, + "learning_rate": 1.94468447306142e-05, + "loss": 0.3723, + "step": 2545 + }, + { + "epoch": 1.07, + "learning_rate": 1.9446288533696145e-05, + "loss": 0.4058, + "step": 2546 + }, + { + "epoch": 1.07, + "learning_rate": 1.9445732065252904e-05, + "loss": 0.379, + "step": 2547 + }, + { + "epoch": 1.07, + "learning_rate": 1.9445175325300482e-05, + "loss": 0.4104, + "step": 2548 + }, + { + "epoch": 1.07, + "learning_rate": 1.9444618313854878e-05, + "loss": 0.4294, + "step": 2549 + }, + { + "epoch": 1.07, + "learning_rate": 1.9444061030932094e-05, + "loss": 0.3712, + "step": 2550 + }, + { + "epoch": 1.07, + "learning_rate": 1.9443503476548162e-05, + "loss": 0.3953, + "step": 2551 + }, + { + "epoch": 1.07, + "learning_rate": 1.9442945650719102e-05, + "loss": 0.3694, + "step": 2552 + }, + { + "epoch": 1.07, + "learning_rate": 1.944238755346095e-05, + "loss": 0.4309, + "step": 2553 + }, + { + "epoch": 1.07, + "learning_rate": 1.9441829184789745e-05, + "loss": 0.3836, + "step": 2554 + }, + { + "epoch": 1.07, + "learning_rate": 1.9441270544721536e-05, + "loss": 0.4283, + "step": 2555 + }, + { + "epoch": 1.07, + "learning_rate": 1.9440711633272387e-05, + "loss": 0.3818, + "step": 2556 + }, + { + "epoch": 1.07, + "learning_rate": 1.944015245045836e-05, + "loss": 0.3676, + "step": 2557 + }, + { + "epoch": 1.07, + "learning_rate": 1.9439592996295525e-05, + "loss": 0.3511, + "step": 2558 + }, + { + "epoch": 1.07, + "learning_rate": 1.943903327079997e-05, + "loss": 0.3633, + "step": 2559 + }, + { + "epoch": 1.07, + "learning_rate": 1.9438473273987775e-05, + "loss": 0.3664, + "step": 2560 + }, + { + "epoch": 1.07, + "learning_rate": 1.9437913005875046e-05, + "loss": 0.4262, + "step": 2561 + }, + { + "epoch": 1.07, + "learning_rate": 1.943735246647788e-05, + "loss": 0.4054, + "step": 2562 + }, + { + "epoch": 1.07, + "learning_rate": 1.9436791655812392e-05, + "loss": 0.4121, + "step": 2563 + }, + { + "epoch": 1.07, + "learning_rate": 1.94362305738947e-05, + "loss": 0.3511, + "step": 2564 + }, + { + "epoch": 1.07, + "learning_rate": 1.9435669220740933e-05, + "loss": 0.3689, + "step": 2565 + }, + { + "epoch": 1.07, + "learning_rate": 1.943510759636723e-05, + "loss": 0.3972, + "step": 2566 + }, + { + "epoch": 1.07, + "learning_rate": 1.9434545700789732e-05, + "loss": 0.404, + "step": 2567 + }, + { + "epoch": 1.08, + "learning_rate": 1.943398353402459e-05, + "loss": 0.3609, + "step": 2568 + }, + { + "epoch": 1.08, + "learning_rate": 1.9433421096087963e-05, + "loss": 0.3381, + "step": 2569 + }, + { + "epoch": 1.08, + "learning_rate": 1.9432858386996018e-05, + "loss": 0.3895, + "step": 2570 + }, + { + "epoch": 1.08, + "learning_rate": 1.9432295406764927e-05, + "loss": 0.3767, + "step": 2571 + }, + { + "epoch": 1.08, + "learning_rate": 1.9431732155410878e-05, + "loss": 0.3764, + "step": 2572 + }, + { + "epoch": 1.08, + "learning_rate": 1.9431168632950053e-05, + "loss": 0.3678, + "step": 2573 + }, + { + "epoch": 1.08, + "learning_rate": 1.9430604839398657e-05, + "loss": 0.3845, + "step": 2574 + }, + { + "epoch": 1.08, + "learning_rate": 1.9430040774772893e-05, + "loss": 0.42, + "step": 2575 + }, + { + "epoch": 1.08, + "learning_rate": 1.9429476439088976e-05, + "loss": 0.385, + "step": 2576 + }, + { + "epoch": 1.08, + "learning_rate": 1.9428911832363126e-05, + "loss": 0.4227, + "step": 2577 + }, + { + "epoch": 1.08, + "learning_rate": 1.9428346954611574e-05, + "loss": 0.3842, + "step": 2578 + }, + { + "epoch": 1.08, + "learning_rate": 1.9427781805850553e-05, + "loss": 0.3735, + "step": 2579 + }, + { + "epoch": 1.08, + "learning_rate": 1.9427216386096313e-05, + "loss": 0.3866, + "step": 2580 + }, + { + "epoch": 1.08, + "learning_rate": 1.94266506953651e-05, + "loss": 0.3834, + "step": 2581 + }, + { + "epoch": 1.08, + "learning_rate": 1.942608473367318e-05, + "loss": 0.4074, + "step": 2582 + }, + { + "epoch": 1.08, + "learning_rate": 1.942551850103682e-05, + "loss": 0.4092, + "step": 2583 + }, + { + "epoch": 1.08, + "learning_rate": 1.942495199747229e-05, + "loss": 0.3959, + "step": 2584 + }, + { + "epoch": 1.08, + "learning_rate": 1.9424385222995883e-05, + "loss": 0.3953, + "step": 2585 + }, + { + "epoch": 1.08, + "learning_rate": 1.9423818177623884e-05, + "loss": 0.3939, + "step": 2586 + }, + { + "epoch": 1.08, + "learning_rate": 1.9423250861372592e-05, + "loss": 0.4049, + "step": 2587 + }, + { + "epoch": 1.08, + "learning_rate": 1.9422683274258318e-05, + "loss": 0.3754, + "step": 2588 + }, + { + "epoch": 1.08, + "learning_rate": 1.9422115416297377e-05, + "loss": 0.4024, + "step": 2589 + }, + { + "epoch": 1.08, + "learning_rate": 1.9421547287506085e-05, + "loss": 0.358, + "step": 2590 + }, + { + "epoch": 1.08, + "learning_rate": 1.942097888790078e-05, + "loss": 0.3966, + "step": 2591 + }, + { + "epoch": 1.09, + "learning_rate": 1.9420410217497795e-05, + "loss": 0.3881, + "step": 2592 + }, + { + "epoch": 1.09, + "learning_rate": 1.941984127631348e-05, + "loss": 0.3901, + "step": 2593 + }, + { + "epoch": 1.09, + "learning_rate": 1.9419272064364186e-05, + "loss": 0.4057, + "step": 2594 + }, + { + "epoch": 1.09, + "learning_rate": 1.9418702581666276e-05, + "loss": 0.351, + "step": 2595 + }, + { + "epoch": 1.09, + "learning_rate": 1.9418132828236114e-05, + "loss": 0.3222, + "step": 2596 + }, + { + "epoch": 1.09, + "learning_rate": 1.9417562804090086e-05, + "loss": 0.3725, + "step": 2597 + }, + { + "epoch": 1.09, + "learning_rate": 1.941699250924457e-05, + "loss": 0.4544, + "step": 2598 + }, + { + "epoch": 1.09, + "learning_rate": 1.941642194371596e-05, + "loss": 0.408, + "step": 2599 + }, + { + "epoch": 1.09, + "learning_rate": 1.941585110752066e-05, + "loss": 0.3905, + "step": 2600 + }, + { + "epoch": 1.09, + "learning_rate": 1.9415280000675074e-05, + "loss": 0.39, + "step": 2601 + }, + { + "epoch": 1.09, + "learning_rate": 1.9414708623195618e-05, + "loss": 0.363, + "step": 2602 + }, + { + "epoch": 1.09, + "learning_rate": 1.9414136975098718e-05, + "loss": 0.3951, + "step": 2603 + }, + { + "epoch": 1.09, + "learning_rate": 1.9413565056400804e-05, + "loss": 0.3983, + "step": 2604 + }, + { + "epoch": 1.09, + "learning_rate": 1.9412992867118318e-05, + "loss": 0.3811, + "step": 2605 + }, + { + "epoch": 1.09, + "learning_rate": 1.9412420407267704e-05, + "loss": 0.3948, + "step": 2606 + }, + { + "epoch": 1.09, + "learning_rate": 1.9411847676865415e-05, + "loss": 0.3897, + "step": 2607 + }, + { + "epoch": 1.09, + "learning_rate": 1.9411274675927922e-05, + "loss": 0.3905, + "step": 2608 + }, + { + "epoch": 1.09, + "learning_rate": 1.9410701404471686e-05, + "loss": 0.4154, + "step": 2609 + }, + { + "epoch": 1.09, + "learning_rate": 1.941012786251319e-05, + "loss": 0.3532, + "step": 2610 + }, + { + "epoch": 1.09, + "learning_rate": 1.9409554050068916e-05, + "loss": 0.3639, + "step": 2611 + }, + { + "epoch": 1.09, + "learning_rate": 1.9408979967155364e-05, + "loss": 0.3617, + "step": 2612 + }, + { + "epoch": 1.09, + "learning_rate": 1.940840561378903e-05, + "loss": 0.3866, + "step": 2613 + }, + { + "epoch": 1.09, + "learning_rate": 1.940783098998643e-05, + "loss": 0.3641, + "step": 2614 + }, + { + "epoch": 1.09, + "learning_rate": 1.9407256095764074e-05, + "loss": 0.3998, + "step": 2615 + }, + { + "epoch": 1.1, + "learning_rate": 1.9406680931138488e-05, + "loss": 0.3803, + "step": 2616 + }, + { + "epoch": 1.1, + "learning_rate": 1.940610549612621e-05, + "loss": 0.3922, + "step": 2617 + }, + { + "epoch": 1.1, + "learning_rate": 1.940552979074377e-05, + "loss": 0.372, + "step": 2618 + }, + { + "epoch": 1.1, + "learning_rate": 1.940495381500773e-05, + "loss": 0.3683, + "step": 2619 + }, + { + "epoch": 1.1, + "learning_rate": 1.9404377568934633e-05, + "loss": 0.4032, + "step": 2620 + }, + { + "epoch": 1.1, + "learning_rate": 1.940380105254105e-05, + "loss": 0.3778, + "step": 2621 + }, + { + "epoch": 1.1, + "learning_rate": 1.9403224265843547e-05, + "loss": 0.3555, + "step": 2622 + }, + { + "epoch": 1.1, + "learning_rate": 1.9402647208858712e-05, + "loss": 0.3768, + "step": 2623 + }, + { + "epoch": 1.1, + "learning_rate": 1.9402069881603123e-05, + "loss": 0.3493, + "step": 2624 + }, + { + "epoch": 1.1, + "learning_rate": 1.9401492284093382e-05, + "loss": 0.4207, + "step": 2625 + }, + { + "epoch": 1.1, + "learning_rate": 1.9400914416346083e-05, + "loss": 0.3887, + "step": 2626 + }, + { + "epoch": 1.1, + "learning_rate": 1.9400336278377845e-05, + "loss": 0.3926, + "step": 2627 + }, + { + "epoch": 1.1, + "learning_rate": 1.939975787020528e-05, + "loss": 0.3902, + "step": 2628 + }, + { + "epoch": 1.1, + "learning_rate": 1.9399179191845022e-05, + "loss": 0.3577, + "step": 2629 + }, + { + "epoch": 1.1, + "learning_rate": 1.9398600243313695e-05, + "loss": 0.3764, + "step": 2630 + }, + { + "epoch": 1.1, + "learning_rate": 1.939802102462794e-05, + "loss": 0.3581, + "step": 2631 + }, + { + "epoch": 1.1, + "learning_rate": 1.9397441535804417e-05, + "loss": 0.3631, + "step": 2632 + }, + { + "epoch": 1.1, + "learning_rate": 1.939686177685977e-05, + "loss": 0.3528, + "step": 2633 + }, + { + "epoch": 1.1, + "learning_rate": 1.939628174781068e-05, + "loss": 0.3927, + "step": 2634 + }, + { + "epoch": 1.1, + "learning_rate": 1.9395701448673797e-05, + "loss": 0.3593, + "step": 2635 + }, + { + "epoch": 1.1, + "learning_rate": 1.939512087946582e-05, + "loss": 0.3985, + "step": 2636 + }, + { + "epoch": 1.1, + "learning_rate": 1.9394540040203428e-05, + "loss": 0.3532, + "step": 2637 + }, + { + "epoch": 1.1, + "learning_rate": 1.9393958930903323e-05, + "loss": 0.3842, + "step": 2638 + }, + { + "epoch": 1.1, + "learning_rate": 1.93933775515822e-05, + "loss": 0.3824, + "step": 2639 + }, + { + "epoch": 1.11, + "learning_rate": 1.939279590225678e-05, + "loss": 0.3778, + "step": 2640 + }, + { + "epoch": 1.11, + "learning_rate": 1.9392213982943774e-05, + "loss": 0.3495, + "step": 2641 + }, + { + "epoch": 1.11, + "learning_rate": 1.9391631793659908e-05, + "loss": 0.3768, + "step": 2642 + }, + { + "epoch": 1.11, + "learning_rate": 1.9391049334421925e-05, + "loss": 0.3629, + "step": 2643 + }, + { + "epoch": 1.11, + "learning_rate": 1.939046660524656e-05, + "loss": 0.375, + "step": 2644 + }, + { + "epoch": 1.11, + "learning_rate": 1.938988360615057e-05, + "loss": 0.3699, + "step": 2645 + }, + { + "epoch": 1.11, + "learning_rate": 1.93893003371507e-05, + "loss": 0.3716, + "step": 2646 + }, + { + "epoch": 1.11, + "learning_rate": 1.9388716798263728e-05, + "loss": 0.4273, + "step": 2647 + }, + { + "epoch": 1.11, + "learning_rate": 1.9388132989506422e-05, + "loss": 0.3377, + "step": 2648 + }, + { + "epoch": 1.11, + "learning_rate": 1.9387548910895567e-05, + "loss": 0.3908, + "step": 2649 + }, + { + "epoch": 1.11, + "learning_rate": 1.9386964562447947e-05, + "loss": 0.3764, + "step": 2650 + }, + { + "epoch": 1.11, + "learning_rate": 1.938637994418036e-05, + "loss": 0.4319, + "step": 2651 + }, + { + "epoch": 1.11, + "learning_rate": 1.9385795056109614e-05, + "loss": 0.3947, + "step": 2652 + }, + { + "epoch": 1.11, + "learning_rate": 1.938520989825251e-05, + "loss": 0.3839, + "step": 2653 + }, + { + "epoch": 1.11, + "learning_rate": 1.9384624470625884e-05, + "loss": 0.4071, + "step": 2654 + }, + { + "epoch": 1.11, + "learning_rate": 1.9384038773246555e-05, + "loss": 0.3845, + "step": 2655 + }, + { + "epoch": 1.11, + "learning_rate": 1.9383452806131356e-05, + "loss": 0.4514, + "step": 2656 + }, + { + "epoch": 1.11, + "learning_rate": 1.9382866569297133e-05, + "loss": 0.34, + "step": 2657 + }, + { + "epoch": 1.11, + "learning_rate": 1.9382280062760738e-05, + "loss": 0.3823, + "step": 2658 + }, + { + "epoch": 1.11, + "learning_rate": 1.9381693286539024e-05, + "loss": 0.381, + "step": 2659 + }, + { + "epoch": 1.11, + "learning_rate": 1.9381106240648867e-05, + "loss": 0.3852, + "step": 2660 + }, + { + "epoch": 1.11, + "learning_rate": 1.9380518925107133e-05, + "loss": 0.4091, + "step": 2661 + }, + { + "epoch": 1.11, + "learning_rate": 1.9379931339930705e-05, + "loss": 0.3806, + "step": 2662 + }, + { + "epoch": 1.11, + "learning_rate": 1.9379343485136478e-05, + "loss": 0.3209, + "step": 2663 + }, + { + "epoch": 1.12, + "learning_rate": 1.9378755360741342e-05, + "loss": 0.3436, + "step": 2664 + }, + { + "epoch": 1.12, + "learning_rate": 1.9378166966762207e-05, + "loss": 0.392, + "step": 2665 + }, + { + "epoch": 1.12, + "learning_rate": 1.9377578303215984e-05, + "loss": 0.3796, + "step": 2666 + }, + { + "epoch": 1.12, + "learning_rate": 1.9376989370119595e-05, + "loss": 0.3759, + "step": 2667 + }, + { + "epoch": 1.12, + "learning_rate": 1.937640016748997e-05, + "loss": 0.4186, + "step": 2668 + }, + { + "epoch": 1.12, + "learning_rate": 1.937581069534404e-05, + "loss": 0.3675, + "step": 2669 + }, + { + "epoch": 1.12, + "learning_rate": 1.937522095369875e-05, + "loss": 0.3504, + "step": 2670 + }, + { + "epoch": 1.12, + "learning_rate": 1.9374630942571055e-05, + "loss": 0.3958, + "step": 2671 + }, + { + "epoch": 1.12, + "learning_rate": 1.9374040661977913e-05, + "loss": 0.3544, + "step": 2672 + }, + { + "epoch": 1.12, + "learning_rate": 1.937345011193629e-05, + "loss": 0.376, + "step": 2673 + }, + { + "epoch": 1.12, + "learning_rate": 1.9372859292463163e-05, + "loss": 0.3849, + "step": 2674 + }, + { + "epoch": 1.12, + "learning_rate": 1.9372268203575512e-05, + "loss": 0.3558, + "step": 2675 + }, + { + "epoch": 1.12, + "learning_rate": 1.9371676845290327e-05, + "loss": 0.3714, + "step": 2676 + }, + { + "epoch": 1.12, + "learning_rate": 1.9371085217624608e-05, + "loss": 0.4111, + "step": 2677 + }, + { + "epoch": 1.12, + "learning_rate": 1.937049332059536e-05, + "loss": 0.3657, + "step": 2678 + }, + { + "epoch": 1.12, + "learning_rate": 1.9369901154219595e-05, + "loss": 0.3718, + "step": 2679 + }, + { + "epoch": 1.12, + "learning_rate": 1.9369308718514342e-05, + "loss": 0.3688, + "step": 2680 + }, + { + "epoch": 1.12, + "learning_rate": 1.9368716013496616e-05, + "loss": 0.4108, + "step": 2681 + }, + { + "epoch": 1.12, + "learning_rate": 1.9368123039183468e-05, + "loss": 0.3761, + "step": 2682 + }, + { + "epoch": 1.12, + "learning_rate": 1.9367529795591935e-05, + "loss": 0.3954, + "step": 2683 + }, + { + "epoch": 1.12, + "learning_rate": 1.9366936282739067e-05, + "loss": 0.3762, + "step": 2684 + }, + { + "epoch": 1.12, + "learning_rate": 1.9366342500641933e-05, + "loss": 0.3663, + "step": 2685 + }, + { + "epoch": 1.12, + "learning_rate": 1.9365748449317593e-05, + "loss": 0.3908, + "step": 2686 + }, + { + "epoch": 1.12, + "learning_rate": 1.9365154128783126e-05, + "loss": 0.3524, + "step": 2687 + }, + { + "epoch": 1.13, + "learning_rate": 1.9364559539055612e-05, + "loss": 0.364, + "step": 2688 + }, + { + "epoch": 1.13, + "learning_rate": 1.9363964680152147e-05, + "loss": 0.3783, + "step": 2689 + }, + { + "epoch": 1.13, + "learning_rate": 1.936336955208982e-05, + "loss": 0.3973, + "step": 2690 + }, + { + "epoch": 1.13, + "learning_rate": 1.936277415488575e-05, + "loss": 0.3912, + "step": 2691 + }, + { + "epoch": 1.13, + "learning_rate": 1.9362178488557043e-05, + "loss": 0.4189, + "step": 2692 + }, + { + "epoch": 1.13, + "learning_rate": 1.9361582553120824e-05, + "loss": 0.3805, + "step": 2693 + }, + { + "epoch": 1.13, + "learning_rate": 1.9360986348594226e-05, + "loss": 0.3507, + "step": 2694 + }, + { + "epoch": 1.13, + "learning_rate": 1.9360389874994374e-05, + "loss": 0.4058, + "step": 2695 + }, + { + "epoch": 1.13, + "learning_rate": 1.9359793132338427e-05, + "loss": 0.3807, + "step": 2696 + }, + { + "epoch": 1.13, + "learning_rate": 1.935919612064353e-05, + "loss": 0.4055, + "step": 2697 + }, + { + "epoch": 1.13, + "learning_rate": 1.9358598839926848e-05, + "loss": 0.3797, + "step": 2698 + }, + { + "epoch": 1.13, + "learning_rate": 1.9358001290205542e-05, + "loss": 0.3917, + "step": 2699 + }, + { + "epoch": 1.13, + "learning_rate": 1.9357403471496797e-05, + "loss": 0.3894, + "step": 2700 + }, + { + "epoch": 1.13, + "learning_rate": 1.9356805383817792e-05, + "loss": 0.3666, + "step": 2701 + }, + { + "epoch": 1.13, + "learning_rate": 1.935620702718572e-05, + "loss": 0.4063, + "step": 2702 + }, + { + "epoch": 1.13, + "learning_rate": 1.9355608401617782e-05, + "loss": 0.3768, + "step": 2703 + }, + { + "epoch": 1.13, + "learning_rate": 1.935500950713118e-05, + "loss": 0.3813, + "step": 2704 + }, + { + "epoch": 1.13, + "learning_rate": 1.9354410343743127e-05, + "loss": 0.3715, + "step": 2705 + }, + { + "epoch": 1.13, + "learning_rate": 1.9353810911470855e-05, + "loss": 0.3931, + "step": 2706 + }, + { + "epoch": 1.13, + "learning_rate": 1.935321121033159e-05, + "loss": 0.41, + "step": 2707 + }, + { + "epoch": 1.13, + "learning_rate": 1.9352611240342565e-05, + "loss": 0.4062, + "step": 2708 + }, + { + "epoch": 1.13, + "learning_rate": 1.935201100152103e-05, + "loss": 0.3632, + "step": 2709 + }, + { + "epoch": 1.13, + "learning_rate": 1.9351410493884234e-05, + "loss": 0.3832, + "step": 2710 + }, + { + "epoch": 1.14, + "learning_rate": 1.9350809717449448e-05, + "loss": 0.3753, + "step": 2711 + }, + { + "epoch": 1.14, + "learning_rate": 1.935020867223393e-05, + "loss": 0.346, + "step": 2712 + }, + { + "epoch": 1.14, + "learning_rate": 1.9349607358254965e-05, + "loss": 0.3275, + "step": 2713 + }, + { + "epoch": 1.14, + "learning_rate": 1.934900577552983e-05, + "loss": 0.4153, + "step": 2714 + }, + { + "epoch": 1.14, + "learning_rate": 1.934840392407582e-05, + "loss": 0.3351, + "step": 2715 + }, + { + "epoch": 1.14, + "learning_rate": 1.9347801803910236e-05, + "loss": 0.4013, + "step": 2716 + }, + { + "epoch": 1.14, + "learning_rate": 1.9347199415050383e-05, + "loss": 0.37, + "step": 2717 + }, + { + "epoch": 1.14, + "learning_rate": 1.934659675751358e-05, + "loss": 0.3452, + "step": 2718 + }, + { + "epoch": 1.14, + "learning_rate": 1.9345993831317147e-05, + "loss": 0.307, + "step": 2719 + }, + { + "epoch": 1.14, + "learning_rate": 1.9345390636478414e-05, + "loss": 0.3732, + "step": 2720 + }, + { + "epoch": 1.14, + "learning_rate": 1.934478717301472e-05, + "loss": 0.4037, + "step": 2721 + }, + { + "epoch": 1.14, + "learning_rate": 1.9344183440943408e-05, + "loss": 0.4228, + "step": 2722 + }, + { + "epoch": 1.14, + "learning_rate": 1.934357944028184e-05, + "loss": 0.3822, + "step": 2723 + }, + { + "epoch": 1.14, + "learning_rate": 1.934297517104737e-05, + "loss": 0.3925, + "step": 2724 + }, + { + "epoch": 1.14, + "learning_rate": 1.9342370633257367e-05, + "loss": 0.3673, + "step": 2725 + }, + { + "epoch": 1.14, + "learning_rate": 1.9341765826929216e-05, + "loss": 0.4204, + "step": 2726 + }, + { + "epoch": 1.14, + "learning_rate": 1.9341160752080294e-05, + "loss": 0.429, + "step": 2727 + }, + { + "epoch": 1.14, + "learning_rate": 1.9340555408727993e-05, + "loss": 0.3993, + "step": 2728 + }, + { + "epoch": 1.14, + "learning_rate": 1.9339949796889717e-05, + "loss": 0.4509, + "step": 2729 + }, + { + "epoch": 1.14, + "learning_rate": 1.933934391658287e-05, + "loss": 0.3768, + "step": 2730 + }, + { + "epoch": 1.14, + "learning_rate": 1.933873776782487e-05, + "loss": 0.3504, + "step": 2731 + }, + { + "epoch": 1.14, + "learning_rate": 1.9338131350633143e-05, + "loss": 0.4084, + "step": 2732 + }, + { + "epoch": 1.14, + "learning_rate": 1.9337524665025114e-05, + "loss": 0.3781, + "step": 2733 + }, + { + "epoch": 1.14, + "learning_rate": 1.9336917711018227e-05, + "loss": 0.4215, + "step": 2734 + }, + { + "epoch": 1.15, + "learning_rate": 1.9336310488629926e-05, + "loss": 0.4066, + "step": 2735 + }, + { + "epoch": 1.15, + "learning_rate": 1.9335702997877664e-05, + "loss": 0.4147, + "step": 2736 + }, + { + "epoch": 1.15, + "learning_rate": 1.9335095238778903e-05, + "loss": 0.3655, + "step": 2737 + }, + { + "epoch": 1.15, + "learning_rate": 1.9334487211351113e-05, + "loss": 0.3868, + "step": 2738 + }, + { + "epoch": 1.15, + "learning_rate": 1.933387891561177e-05, + "loss": 0.3515, + "step": 2739 + }, + { + "epoch": 1.15, + "learning_rate": 1.9333270351578367e-05, + "loss": 0.3973, + "step": 2740 + }, + { + "epoch": 1.15, + "learning_rate": 1.9332661519268384e-05, + "loss": 0.4615, + "step": 2741 + }, + { + "epoch": 1.15, + "learning_rate": 1.933205241869933e-05, + "loss": 0.4009, + "step": 2742 + }, + { + "epoch": 1.15, + "learning_rate": 1.9331443049888706e-05, + "loss": 0.3969, + "step": 2743 + }, + { + "epoch": 1.15, + "learning_rate": 1.9330833412854035e-05, + "loss": 0.3819, + "step": 2744 + }, + { + "epoch": 1.15, + "learning_rate": 1.933022350761284e-05, + "loss": 0.3491, + "step": 2745 + }, + { + "epoch": 1.15, + "learning_rate": 1.9329613334182644e-05, + "loss": 0.3933, + "step": 2746 + }, + { + "epoch": 1.15, + "learning_rate": 1.9329002892580995e-05, + "loss": 0.3607, + "step": 2747 + }, + { + "epoch": 1.15, + "learning_rate": 1.932839218282544e-05, + "loss": 0.3915, + "step": 2748 + }, + { + "epoch": 1.15, + "learning_rate": 1.9327781204933522e-05, + "loss": 0.355, + "step": 2749 + }, + { + "epoch": 1.15, + "learning_rate": 1.9327169958922813e-05, + "loss": 0.4151, + "step": 2750 + }, + { + "epoch": 1.15, + "learning_rate": 1.9326558444810883e-05, + "loss": 0.3827, + "step": 2751 + }, + { + "epoch": 1.15, + "learning_rate": 1.9325946662615304e-05, + "loss": 0.4245, + "step": 2752 + }, + { + "epoch": 1.15, + "learning_rate": 1.932533461235367e-05, + "loss": 0.3551, + "step": 2753 + }, + { + "epoch": 1.15, + "learning_rate": 1.932472229404356e-05, + "loss": 0.3895, + "step": 2754 + }, + { + "epoch": 1.15, + "learning_rate": 1.9324109707702585e-05, + "loss": 0.3821, + "step": 2755 + }, + { + "epoch": 1.15, + "learning_rate": 1.932349685334835e-05, + "loss": 0.3928, + "step": 2756 + }, + { + "epoch": 1.15, + "learning_rate": 1.9322883730998474e-05, + "loss": 0.3673, + "step": 2757 + }, + { + "epoch": 1.15, + "learning_rate": 1.9322270340670574e-05, + "loss": 0.3496, + "step": 2758 + }, + { + "epoch": 1.16, + "learning_rate": 1.9321656682382287e-05, + "loss": 0.3826, + "step": 2759 + }, + { + "epoch": 1.16, + "learning_rate": 1.9321042756151254e-05, + "loss": 0.3732, + "step": 2760 + }, + { + "epoch": 1.16, + "learning_rate": 1.9320428561995114e-05, + "loss": 0.3547, + "step": 2761 + }, + { + "epoch": 1.16, + "learning_rate": 1.931981409993153e-05, + "loss": 0.337, + "step": 2762 + }, + { + "epoch": 1.16, + "learning_rate": 1.9319199369978157e-05, + "loss": 0.3734, + "step": 2763 + }, + { + "epoch": 1.16, + "learning_rate": 1.9318584372152668e-05, + "loss": 0.3379, + "step": 2764 + }, + { + "epoch": 1.16, + "learning_rate": 1.9317969106472742e-05, + "loss": 0.3891, + "step": 2765 + }, + { + "epoch": 1.16, + "learning_rate": 1.9317353572956058e-05, + "loss": 0.4175, + "step": 2766 + }, + { + "epoch": 1.16, + "learning_rate": 1.931673777162032e-05, + "loss": 0.3745, + "step": 2767 + }, + { + "epoch": 1.16, + "learning_rate": 1.9316121702483218e-05, + "loss": 0.421, + "step": 2768 + }, + { + "epoch": 1.16, + "learning_rate": 1.9315505365562464e-05, + "loss": 0.3868, + "step": 2769 + }, + { + "epoch": 1.16, + "learning_rate": 1.931488876087578e-05, + "loss": 0.3765, + "step": 2770 + }, + { + "epoch": 1.16, + "learning_rate": 1.9314271888440878e-05, + "loss": 0.3419, + "step": 2771 + }, + { + "epoch": 1.16, + "learning_rate": 1.9313654748275497e-05, + "loss": 0.4157, + "step": 2772 + }, + { + "epoch": 1.16, + "learning_rate": 1.931303734039738e-05, + "loss": 0.4043, + "step": 2773 + }, + { + "epoch": 1.16, + "learning_rate": 1.9312419664824268e-05, + "loss": 0.4398, + "step": 2774 + }, + { + "epoch": 1.16, + "learning_rate": 1.9311801721573913e-05, + "loss": 0.39, + "step": 2775 + }, + { + "epoch": 1.16, + "learning_rate": 1.931118351066408e-05, + "loss": 0.3564, + "step": 2776 + }, + { + "epoch": 1.16, + "learning_rate": 1.9310565032112542e-05, + "loss": 0.3823, + "step": 2777 + }, + { + "epoch": 1.16, + "learning_rate": 1.9309946285937075e-05, + "loss": 0.3924, + "step": 2778 + }, + { + "epoch": 1.16, + "learning_rate": 1.9309327272155462e-05, + "loss": 0.3895, + "step": 2779 + }, + { + "epoch": 1.16, + "learning_rate": 1.9308707990785497e-05, + "loss": 0.3785, + "step": 2780 + }, + { + "epoch": 1.16, + "learning_rate": 1.9308088441844982e-05, + "loss": 0.3857, + "step": 2781 + }, + { + "epoch": 1.16, + "learning_rate": 1.9307468625351723e-05, + "loss": 0.4335, + "step": 2782 + }, + { + "epoch": 1.17, + "learning_rate": 1.930684854132354e-05, + "loss": 0.3893, + "step": 2783 + }, + { + "epoch": 1.17, + "learning_rate": 1.9306228189778255e-05, + "loss": 0.4093, + "step": 2784 + }, + { + "epoch": 1.17, + "learning_rate": 1.9305607570733696e-05, + "loss": 0.4062, + "step": 2785 + }, + { + "epoch": 1.17, + "learning_rate": 1.930498668420771e-05, + "loss": 0.3321, + "step": 2786 + }, + { + "epoch": 1.17, + "learning_rate": 1.9304365530218134e-05, + "loss": 0.4056, + "step": 2787 + }, + { + "epoch": 1.17, + "learning_rate": 1.930374410878283e-05, + "loss": 0.4217, + "step": 2788 + }, + { + "epoch": 1.17, + "learning_rate": 1.9303122419919655e-05, + "loss": 0.3904, + "step": 2789 + }, + { + "epoch": 1.17, + "learning_rate": 1.9302500463646483e-05, + "loss": 0.3738, + "step": 2790 + }, + { + "epoch": 1.17, + "learning_rate": 1.930187823998119e-05, + "loss": 0.3978, + "step": 2791 + }, + { + "epoch": 1.17, + "learning_rate": 1.930125574894166e-05, + "loss": 0.3531, + "step": 2792 + }, + { + "epoch": 1.17, + "learning_rate": 1.930063299054579e-05, + "loss": 0.3876, + "step": 2793 + }, + { + "epoch": 1.17, + "learning_rate": 1.9300009964811475e-05, + "loss": 0.4144, + "step": 2794 + }, + { + "epoch": 1.17, + "learning_rate": 1.9299386671756628e-05, + "loss": 0.4041, + "step": 2795 + }, + { + "epoch": 1.17, + "learning_rate": 1.9298763111399162e-05, + "loss": 0.4336, + "step": 2796 + }, + { + "epoch": 1.17, + "learning_rate": 1.9298139283757004e-05, + "loss": 0.3942, + "step": 2797 + }, + { + "epoch": 1.17, + "learning_rate": 1.9297515188848083e-05, + "loss": 0.3601, + "step": 2798 + }, + { + "epoch": 1.17, + "learning_rate": 1.9296890826690337e-05, + "loss": 0.3929, + "step": 2799 + }, + { + "epoch": 1.17, + "learning_rate": 1.9296266197301716e-05, + "loss": 0.4062, + "step": 2800 + }, + { + "epoch": 1.17, + "learning_rate": 1.9295641300700173e-05, + "loss": 0.354, + "step": 2801 + }, + { + "epoch": 1.17, + "learning_rate": 1.9295016136903666e-05, + "loss": 0.4031, + "step": 2802 + }, + { + "epoch": 1.17, + "learning_rate": 1.929439070593017e-05, + "loss": 0.4086, + "step": 2803 + }, + { + "epoch": 1.17, + "learning_rate": 1.9293765007797662e-05, + "loss": 0.4178, + "step": 2804 + }, + { + "epoch": 1.17, + "learning_rate": 1.9293139042524124e-05, + "loss": 0.3496, + "step": 2805 + }, + { + "epoch": 1.17, + "learning_rate": 1.9292512810127554e-05, + "loss": 0.397, + "step": 2806 + }, + { + "epoch": 1.18, + "learning_rate": 1.9291886310625948e-05, + "loss": 0.3759, + "step": 2807 + }, + { + "epoch": 1.18, + "learning_rate": 1.9291259544037313e-05, + "loss": 0.3958, + "step": 2808 + }, + { + "epoch": 1.18, + "learning_rate": 1.9290632510379667e-05, + "loss": 0.404, + "step": 2809 + }, + { + "epoch": 1.18, + "learning_rate": 1.9290005209671036e-05, + "loss": 0.394, + "step": 2810 + }, + { + "epoch": 1.18, + "learning_rate": 1.928937764192945e-05, + "loss": 0.4051, + "step": 2811 + }, + { + "epoch": 1.18, + "learning_rate": 1.9288749807172947e-05, + "loss": 0.4203, + "step": 2812 + }, + { + "epoch": 1.18, + "learning_rate": 1.9288121705419572e-05, + "loss": 0.3902, + "step": 2813 + }, + { + "epoch": 1.18, + "learning_rate": 1.928749333668738e-05, + "loss": 0.3657, + "step": 2814 + }, + { + "epoch": 1.18, + "learning_rate": 1.9286864700994433e-05, + "loss": 0.3203, + "step": 2815 + }, + { + "epoch": 1.18, + "learning_rate": 1.9286235798358803e-05, + "loss": 0.3667, + "step": 2816 + }, + { + "epoch": 1.18, + "learning_rate": 1.9285606628798564e-05, + "loss": 0.4075, + "step": 2817 + }, + { + "epoch": 1.18, + "learning_rate": 1.9284977192331807e-05, + "loss": 0.4056, + "step": 2818 + }, + { + "epoch": 1.18, + "learning_rate": 1.9284347488976616e-05, + "loss": 0.4313, + "step": 2819 + }, + { + "epoch": 1.18, + "learning_rate": 1.9283717518751095e-05, + "loss": 0.3834, + "step": 2820 + }, + { + "epoch": 1.18, + "learning_rate": 1.9283087281673352e-05, + "loss": 0.3633, + "step": 2821 + }, + { + "epoch": 1.18, + "learning_rate": 1.9282456777761505e-05, + "loss": 0.3872, + "step": 2822 + }, + { + "epoch": 1.18, + "learning_rate": 1.9281826007033674e-05, + "loss": 0.4124, + "step": 2823 + }, + { + "epoch": 1.18, + "learning_rate": 1.928119496950799e-05, + "loss": 0.4013, + "step": 2824 + }, + { + "epoch": 1.18, + "learning_rate": 1.9280563665202595e-05, + "loss": 0.4354, + "step": 2825 + }, + { + "epoch": 1.18, + "learning_rate": 1.927993209413563e-05, + "loss": 0.3694, + "step": 2826 + }, + { + "epoch": 1.18, + "learning_rate": 1.927930025632525e-05, + "loss": 0.3888, + "step": 2827 + }, + { + "epoch": 1.18, + "learning_rate": 1.9278668151789627e-05, + "loss": 0.4104, + "step": 2828 + }, + { + "epoch": 1.18, + "learning_rate": 1.9278035780546912e-05, + "loss": 0.4169, + "step": 2829 + }, + { + "epoch": 1.18, + "learning_rate": 1.9277403142615297e-05, + "loss": 0.385, + "step": 2830 + }, + { + "epoch": 1.19, + "learning_rate": 1.927677023801296e-05, + "loss": 0.4508, + "step": 2831 + }, + { + "epoch": 1.19, + "learning_rate": 1.9276137066758097e-05, + "loss": 0.4015, + "step": 2832 + }, + { + "epoch": 1.19, + "learning_rate": 1.9275503628868902e-05, + "loss": 0.4012, + "step": 2833 + }, + { + "epoch": 1.19, + "learning_rate": 1.9274869924363585e-05, + "loss": 0.3566, + "step": 2834 + }, + { + "epoch": 1.19, + "learning_rate": 1.9274235953260367e-05, + "loss": 0.3656, + "step": 2835 + }, + { + "epoch": 1.19, + "learning_rate": 1.9273601715577463e-05, + "loss": 0.3389, + "step": 2836 + }, + { + "epoch": 1.19, + "learning_rate": 1.9272967211333107e-05, + "loss": 0.3845, + "step": 2837 + }, + { + "epoch": 1.19, + "learning_rate": 1.927233244054554e-05, + "loss": 0.372, + "step": 2838 + }, + { + "epoch": 1.19, + "learning_rate": 1.9271697403233e-05, + "loss": 0.4459, + "step": 2839 + }, + { + "epoch": 1.19, + "learning_rate": 1.9271062099413752e-05, + "loss": 0.3977, + "step": 2840 + }, + { + "epoch": 1.19, + "learning_rate": 1.9270426529106048e-05, + "loss": 0.3697, + "step": 2841 + }, + { + "epoch": 1.19, + "learning_rate": 1.926979069232816e-05, + "loss": 0.3596, + "step": 2842 + }, + { + "epoch": 1.19, + "learning_rate": 1.9269154589098363e-05, + "loss": 0.388, + "step": 2843 + }, + { + "epoch": 1.19, + "learning_rate": 1.9268518219434943e-05, + "loss": 0.4069, + "step": 2844 + }, + { + "epoch": 1.19, + "learning_rate": 1.9267881583356192e-05, + "loss": 0.367, + "step": 2845 + }, + { + "epoch": 1.19, + "learning_rate": 1.926724468088041e-05, + "loss": 0.3711, + "step": 2846 + }, + { + "epoch": 1.19, + "learning_rate": 1.9266607512025903e-05, + "loss": 0.3994, + "step": 2847 + }, + { + "epoch": 1.19, + "learning_rate": 1.9265970076810987e-05, + "loss": 0.4422, + "step": 2848 + }, + { + "epoch": 1.19, + "learning_rate": 1.9265332375253984e-05, + "loss": 0.3868, + "step": 2849 + }, + { + "epoch": 1.19, + "learning_rate": 1.926469440737322e-05, + "loss": 0.4381, + "step": 2850 + }, + { + "epoch": 1.19, + "learning_rate": 1.9264056173187037e-05, + "loss": 0.4296, + "step": 2851 + }, + { + "epoch": 1.19, + "learning_rate": 1.9263417672713786e-05, + "loss": 0.3709, + "step": 2852 + }, + { + "epoch": 1.19, + "learning_rate": 1.9262778905971808e-05, + "loss": 0.4049, + "step": 2853 + }, + { + "epoch": 1.19, + "learning_rate": 1.9262139872979474e-05, + "loss": 0.3886, + "step": 2854 + }, + { + "epoch": 1.2, + "learning_rate": 1.9261500573755146e-05, + "loss": 0.3822, + "step": 2855 + }, + { + "epoch": 1.2, + "learning_rate": 1.9260861008317203e-05, + "loss": 0.3676, + "step": 2856 + }, + { + "epoch": 1.2, + "learning_rate": 1.9260221176684025e-05, + "loss": 0.386, + "step": 2857 + }, + { + "epoch": 1.2, + "learning_rate": 1.925958107887401e-05, + "loss": 0.3864, + "step": 2858 + }, + { + "epoch": 1.2, + "learning_rate": 1.9258940714905554e-05, + "loss": 0.3853, + "step": 2859 + }, + { + "epoch": 1.2, + "learning_rate": 1.9258300084797063e-05, + "loss": 0.4195, + "step": 2860 + }, + { + "epoch": 1.2, + "learning_rate": 1.925765918856695e-05, + "loss": 0.3893, + "step": 2861 + }, + { + "epoch": 1.2, + "learning_rate": 1.9257018026233642e-05, + "loss": 0.4311, + "step": 2862 + }, + { + "epoch": 1.2, + "learning_rate": 1.9256376597815565e-05, + "loss": 0.3936, + "step": 2863 + }, + { + "epoch": 1.2, + "learning_rate": 1.9255734903331157e-05, + "loss": 0.4135, + "step": 2864 + }, + { + "epoch": 1.2, + "learning_rate": 1.9255092942798858e-05, + "loss": 0.3841, + "step": 2865 + }, + { + "epoch": 1.2, + "learning_rate": 1.9254450716237128e-05, + "loss": 0.394, + "step": 2866 + }, + { + "epoch": 1.2, + "learning_rate": 1.9253808223664427e-05, + "loss": 0.4184, + "step": 2867 + }, + { + "epoch": 1.2, + "learning_rate": 1.9253165465099217e-05, + "loss": 0.3678, + "step": 2868 + }, + { + "epoch": 1.2, + "learning_rate": 1.925252244055998e-05, + "loss": 0.4486, + "step": 2869 + }, + { + "epoch": 1.2, + "learning_rate": 1.9251879150065192e-05, + "loss": 0.3573, + "step": 2870 + }, + { + "epoch": 1.2, + "learning_rate": 1.925123559363335e-05, + "loss": 0.3508, + "step": 2871 + }, + { + "epoch": 1.2, + "learning_rate": 1.925059177128295e-05, + "loss": 0.3864, + "step": 2872 + }, + { + "epoch": 1.2, + "learning_rate": 1.9249947683032497e-05, + "loss": 0.3628, + "step": 2873 + }, + { + "epoch": 1.2, + "learning_rate": 1.9249303328900505e-05, + "loss": 0.3605, + "step": 2874 + }, + { + "epoch": 1.2, + "learning_rate": 1.92486587089055e-05, + "loss": 0.391, + "step": 2875 + }, + { + "epoch": 1.2, + "learning_rate": 1.9248013823066006e-05, + "loss": 0.4173, + "step": 2876 + }, + { + "epoch": 1.2, + "learning_rate": 1.924736867140056e-05, + "loss": 0.413, + "step": 2877 + }, + { + "epoch": 1.2, + "learning_rate": 1.9246723253927714e-05, + "loss": 0.4276, + "step": 2878 + }, + { + "epoch": 1.21, + "learning_rate": 1.9246077570666008e-05, + "loss": 0.4009, + "step": 2879 + }, + { + "epoch": 1.21, + "learning_rate": 1.9245431621634006e-05, + "loss": 0.3918, + "step": 2880 + }, + { + "epoch": 1.21, + "learning_rate": 1.924478540685028e-05, + "loss": 0.3812, + "step": 2881 + }, + { + "epoch": 1.21, + "learning_rate": 1.9244138926333396e-05, + "loss": 0.4158, + "step": 2882 + }, + { + "epoch": 1.21, + "learning_rate": 1.9243492180101947e-05, + "loss": 0.4269, + "step": 2883 + }, + { + "epoch": 1.21, + "learning_rate": 1.9242845168174514e-05, + "loss": 0.3786, + "step": 2884 + }, + { + "epoch": 1.21, + "learning_rate": 1.92421978905697e-05, + "loss": 0.3487, + "step": 2885 + }, + { + "epoch": 1.21, + "learning_rate": 1.924155034730611e-05, + "loss": 0.3736, + "step": 2886 + }, + { + "epoch": 1.21, + "learning_rate": 1.9240902538402354e-05, + "loss": 0.3477, + "step": 2887 + }, + { + "epoch": 1.21, + "learning_rate": 1.9240254463877055e-05, + "loss": 0.3401, + "step": 2888 + }, + { + "epoch": 1.21, + "learning_rate": 1.9239606123748845e-05, + "loss": 0.3752, + "step": 2889 + }, + { + "epoch": 1.21, + "learning_rate": 1.923895751803635e-05, + "loss": 0.37, + "step": 2890 + }, + { + "epoch": 1.21, + "learning_rate": 1.9238308646758224e-05, + "loss": 0.3883, + "step": 2891 + }, + { + "epoch": 1.21, + "learning_rate": 1.9237659509933114e-05, + "loss": 0.3907, + "step": 2892 + }, + { + "epoch": 1.21, + "learning_rate": 1.9237010107579676e-05, + "loss": 0.3955, + "step": 2893 + }, + { + "epoch": 1.21, + "learning_rate": 1.923636043971658e-05, + "loss": 0.3549, + "step": 2894 + }, + { + "epoch": 1.21, + "learning_rate": 1.92357105063625e-05, + "loss": 0.3955, + "step": 2895 + }, + { + "epoch": 1.21, + "learning_rate": 1.9235060307536117e-05, + "loss": 0.3917, + "step": 2896 + }, + { + "epoch": 1.21, + "learning_rate": 1.923440984325612e-05, + "loss": 0.3484, + "step": 2897 + }, + { + "epoch": 1.21, + "learning_rate": 1.923375911354121e-05, + "loss": 0.4145, + "step": 2898 + }, + { + "epoch": 1.21, + "learning_rate": 1.9233108118410082e-05, + "loss": 0.3835, + "step": 2899 + }, + { + "epoch": 1.21, + "learning_rate": 1.923245685788146e-05, + "loss": 0.4056, + "step": 2900 + }, + { + "epoch": 1.21, + "learning_rate": 1.9231805331974053e-05, + "loss": 0.3901, + "step": 2901 + }, + { + "epoch": 1.21, + "learning_rate": 1.9231153540706597e-05, + "loss": 0.4369, + "step": 2902 + }, + { + "epoch": 1.22, + "learning_rate": 1.9230501484097826e-05, + "loss": 0.379, + "step": 2903 + }, + { + "epoch": 1.22, + "learning_rate": 1.9229849162166478e-05, + "loss": 0.3816, + "step": 2904 + }, + { + "epoch": 1.22, + "learning_rate": 1.922919657493131e-05, + "loss": 0.4186, + "step": 2905 + }, + { + "epoch": 1.22, + "learning_rate": 1.922854372241107e-05, + "loss": 0.4063, + "step": 2906 + }, + { + "epoch": 1.22, + "learning_rate": 1.9227890604624534e-05, + "loss": 0.4478, + "step": 2907 + }, + { + "epoch": 1.22, + "learning_rate": 1.9227237221590472e-05, + "loss": 0.4357, + "step": 2908 + }, + { + "epoch": 1.22, + "learning_rate": 1.922658357332766e-05, + "loss": 0.3672, + "step": 2909 + }, + { + "epoch": 1.22, + "learning_rate": 1.9225929659854893e-05, + "loss": 0.3548, + "step": 2910 + }, + { + "epoch": 1.22, + "learning_rate": 1.922527548119097e-05, + "loss": 0.3264, + "step": 2911 + }, + { + "epoch": 1.22, + "learning_rate": 1.922462103735468e-05, + "loss": 0.3426, + "step": 2912 + }, + { + "epoch": 1.22, + "learning_rate": 1.9223966328364853e-05, + "loss": 0.3965, + "step": 2913 + }, + { + "epoch": 1.22, + "learning_rate": 1.922331135424029e-05, + "loss": 0.4179, + "step": 2914 + }, + { + "epoch": 1.22, + "learning_rate": 1.9222656114999835e-05, + "loss": 0.372, + "step": 2915 + }, + { + "epoch": 1.22, + "learning_rate": 1.922200061066231e-05, + "loss": 0.4026, + "step": 2916 + }, + { + "epoch": 1.22, + "learning_rate": 1.9221344841246562e-05, + "loss": 0.356, + "step": 2917 + }, + { + "epoch": 1.22, + "learning_rate": 1.9220688806771443e-05, + "loss": 0.3802, + "step": 2918 + }, + { + "epoch": 1.22, + "learning_rate": 1.92200325072558e-05, + "loss": 0.4339, + "step": 2919 + }, + { + "epoch": 1.22, + "learning_rate": 1.9219375942718508e-05, + "loss": 0.3936, + "step": 2920 + }, + { + "epoch": 1.22, + "learning_rate": 1.9218719113178433e-05, + "loss": 0.3785, + "step": 2921 + }, + { + "epoch": 1.22, + "learning_rate": 1.9218062018654458e-05, + "loss": 0.3854, + "step": 2922 + }, + { + "epoch": 1.22, + "learning_rate": 1.9217404659165476e-05, + "loss": 0.3722, + "step": 2923 + }, + { + "epoch": 1.22, + "learning_rate": 1.921674703473037e-05, + "loss": 0.3839, + "step": 2924 + }, + { + "epoch": 1.22, + "learning_rate": 1.9216089145368052e-05, + "loss": 0.4487, + "step": 2925 + }, + { + "epoch": 1.23, + "learning_rate": 1.921543099109743e-05, + "loss": 0.3922, + "step": 2926 + }, + { + "epoch": 1.23, + "learning_rate": 1.921477257193742e-05, + "loss": 0.393, + "step": 2927 + }, + { + "epoch": 1.23, + "learning_rate": 1.9214113887906947e-05, + "loss": 0.3938, + "step": 2928 + }, + { + "epoch": 1.23, + "learning_rate": 1.9213454939024953e-05, + "loss": 0.3843, + "step": 2929 + }, + { + "epoch": 1.23, + "learning_rate": 1.9212795725310372e-05, + "loss": 0.3286, + "step": 2930 + }, + { + "epoch": 1.23, + "learning_rate": 1.921213624678215e-05, + "loss": 0.3765, + "step": 2931 + }, + { + "epoch": 1.23, + "learning_rate": 1.921147650345925e-05, + "loss": 0.3984, + "step": 2932 + }, + { + "epoch": 1.23, + "learning_rate": 1.9210816495360625e-05, + "loss": 0.4522, + "step": 2933 + }, + { + "epoch": 1.23, + "learning_rate": 1.921015622250526e-05, + "loss": 0.3989, + "step": 2934 + }, + { + "epoch": 1.23, + "learning_rate": 1.920949568491212e-05, + "loss": 0.4184, + "step": 2935 + }, + { + "epoch": 1.23, + "learning_rate": 1.9208834882600207e-05, + "loss": 0.3787, + "step": 2936 + }, + { + "epoch": 1.23, + "learning_rate": 1.9208173815588506e-05, + "loss": 0.4151, + "step": 2937 + }, + { + "epoch": 1.23, + "learning_rate": 1.9207512483896016e-05, + "loss": 0.4006, + "step": 2938 + }, + { + "epoch": 1.23, + "learning_rate": 1.920685088754175e-05, + "loss": 0.3855, + "step": 2939 + }, + { + "epoch": 1.23, + "learning_rate": 1.9206189026544728e-05, + "loss": 0.417, + "step": 2940 + }, + { + "epoch": 1.23, + "learning_rate": 1.920552690092397e-05, + "loss": 0.3611, + "step": 2941 + }, + { + "epoch": 1.23, + "learning_rate": 1.9204864510698513e-05, + "loss": 0.3916, + "step": 2942 + }, + { + "epoch": 1.23, + "learning_rate": 1.9204201855887394e-05, + "loss": 0.3599, + "step": 2943 + }, + { + "epoch": 1.23, + "learning_rate": 1.920353893650966e-05, + "loss": 0.3919, + "step": 2944 + }, + { + "epoch": 1.23, + "learning_rate": 1.9202875752584363e-05, + "loss": 0.3776, + "step": 2945 + }, + { + "epoch": 1.23, + "learning_rate": 1.920221230413057e-05, + "loss": 0.41, + "step": 2946 + }, + { + "epoch": 1.23, + "learning_rate": 1.920154859116735e-05, + "loss": 0.4023, + "step": 2947 + }, + { + "epoch": 1.23, + "learning_rate": 1.9200884613713782e-05, + "loss": 0.3815, + "step": 2948 + }, + { + "epoch": 1.23, + "learning_rate": 1.9200220371788952e-05, + "loss": 0.3629, + "step": 2949 + }, + { + "epoch": 1.24, + "learning_rate": 1.9199555865411948e-05, + "loss": 0.4138, + "step": 2950 + }, + { + "epoch": 1.24, + "learning_rate": 1.9198891094601876e-05, + "loss": 0.3576, + "step": 2951 + }, + { + "epoch": 1.24, + "learning_rate": 1.9198226059377846e-05, + "loss": 0.3736, + "step": 2952 + }, + { + "epoch": 1.24, + "learning_rate": 1.9197560759758963e-05, + "loss": 0.4181, + "step": 2953 + }, + { + "epoch": 1.24, + "learning_rate": 1.9196895195764363e-05, + "loss": 0.3947, + "step": 2954 + }, + { + "epoch": 1.24, + "learning_rate": 1.9196229367413172e-05, + "loss": 0.3668, + "step": 2955 + }, + { + "epoch": 1.24, + "learning_rate": 1.9195563274724523e-05, + "loss": 0.3783, + "step": 2956 + }, + { + "epoch": 1.24, + "learning_rate": 1.9194896917717573e-05, + "loss": 0.3779, + "step": 2957 + }, + { + "epoch": 1.24, + "learning_rate": 1.919423029641147e-05, + "loss": 0.3688, + "step": 2958 + }, + { + "epoch": 1.24, + "learning_rate": 1.9193563410825376e-05, + "loss": 0.3632, + "step": 2959 + }, + { + "epoch": 1.24, + "learning_rate": 1.9192896260978457e-05, + "loss": 0.4211, + "step": 2960 + }, + { + "epoch": 1.24, + "learning_rate": 1.9192228846889896e-05, + "loss": 0.4101, + "step": 2961 + }, + { + "epoch": 1.24, + "learning_rate": 1.9191561168578873e-05, + "loss": 0.3919, + "step": 2962 + }, + { + "epoch": 1.24, + "learning_rate": 1.919089322606458e-05, + "loss": 0.3991, + "step": 2963 + }, + { + "epoch": 1.24, + "learning_rate": 1.919022501936622e-05, + "loss": 0.3567, + "step": 2964 + }, + { + "epoch": 1.24, + "learning_rate": 1.918955654850299e-05, + "loss": 0.3857, + "step": 2965 + }, + { + "epoch": 1.24, + "learning_rate": 1.9188887813494118e-05, + "loss": 0.4223, + "step": 2966 + }, + { + "epoch": 1.24, + "learning_rate": 1.9188218814358817e-05, + "loss": 0.4015, + "step": 2967 + }, + { + "epoch": 1.24, + "learning_rate": 1.9187549551116322e-05, + "loss": 0.4114, + "step": 2968 + }, + { + "epoch": 1.24, + "learning_rate": 1.9186880023785864e-05, + "loss": 0.4167, + "step": 2969 + }, + { + "epoch": 1.24, + "learning_rate": 1.9186210232386694e-05, + "loss": 0.3801, + "step": 2970 + }, + { + "epoch": 1.24, + "learning_rate": 1.9185540176938065e-05, + "loss": 0.3913, + "step": 2971 + }, + { + "epoch": 1.24, + "learning_rate": 1.9184869857459233e-05, + "loss": 0.3867, + "step": 2972 + }, + { + "epoch": 1.24, + "learning_rate": 1.9184199273969465e-05, + "loss": 0.3883, + "step": 2973 + }, + { + "epoch": 1.25, + "learning_rate": 1.918352842648804e-05, + "loss": 0.4711, + "step": 2974 + }, + { + "epoch": 1.25, + "learning_rate": 1.918285731503424e-05, + "loss": 0.4078, + "step": 2975 + }, + { + "epoch": 1.25, + "learning_rate": 1.9182185939627356e-05, + "loss": 0.3993, + "step": 2976 + }, + { + "epoch": 1.25, + "learning_rate": 1.918151430028669e-05, + "loss": 0.3231, + "step": 2977 + }, + { + "epoch": 1.25, + "learning_rate": 1.9180842397031535e-05, + "loss": 0.4097, + "step": 2978 + }, + { + "epoch": 1.25, + "learning_rate": 1.9180170229881216e-05, + "loss": 0.3751, + "step": 2979 + }, + { + "epoch": 1.25, + "learning_rate": 1.9179497798855048e-05, + "loss": 0.3605, + "step": 2980 + }, + { + "epoch": 1.25, + "learning_rate": 1.9178825103972364e-05, + "loss": 0.3395, + "step": 2981 + }, + { + "epoch": 1.25, + "learning_rate": 1.9178152145252497e-05, + "loss": 0.3948, + "step": 2982 + }, + { + "epoch": 1.25, + "learning_rate": 1.917747892271479e-05, + "loss": 0.3671, + "step": 2983 + }, + { + "epoch": 1.25, + "learning_rate": 1.9176805436378597e-05, + "loss": 0.3953, + "step": 2984 + }, + { + "epoch": 1.25, + "learning_rate": 1.9176131686263274e-05, + "loss": 0.4075, + "step": 2985 + }, + { + "epoch": 1.25, + "learning_rate": 1.917545767238819e-05, + "loss": 0.3921, + "step": 2986 + }, + { + "epoch": 1.25, + "learning_rate": 1.9174783394772714e-05, + "loss": 0.4225, + "step": 2987 + }, + { + "epoch": 1.25, + "learning_rate": 1.9174108853436234e-05, + "loss": 0.3768, + "step": 2988 + }, + { + "epoch": 1.25, + "learning_rate": 1.9173434048398133e-05, + "loss": 0.3695, + "step": 2989 + }, + { + "epoch": 1.25, + "learning_rate": 1.917275897967781e-05, + "loss": 0.4141, + "step": 2990 + }, + { + "epoch": 1.25, + "learning_rate": 1.9172083647294676e-05, + "loss": 0.4165, + "step": 2991 + }, + { + "epoch": 1.25, + "learning_rate": 1.9171408051268132e-05, + "loss": 0.4475, + "step": 2992 + }, + { + "epoch": 1.25, + "learning_rate": 1.9170732191617603e-05, + "loss": 0.3717, + "step": 2993 + }, + { + "epoch": 1.25, + "learning_rate": 1.9170056068362516e-05, + "loss": 0.4643, + "step": 2994 + }, + { + "epoch": 1.25, + "learning_rate": 1.9169379681522305e-05, + "loss": 0.3985, + "step": 2995 + }, + { + "epoch": 1.25, + "learning_rate": 1.9168703031116412e-05, + "loss": 0.3602, + "step": 2996 + }, + { + "epoch": 1.25, + "learning_rate": 1.9168026117164285e-05, + "loss": 0.3634, + "step": 2997 + }, + { + "epoch": 1.26, + "learning_rate": 1.916734893968538e-05, + "loss": 0.4244, + "step": 2998 + }, + { + "epoch": 1.26, + "learning_rate": 1.916667149869917e-05, + "loss": 0.3957, + "step": 2999 + }, + { + "epoch": 1.26, + "learning_rate": 1.916599379422512e-05, + "loss": 0.3868, + "step": 3000 + }, + { + "epoch": 1.26, + "learning_rate": 1.9165315826282712e-05, + "loss": 0.3846, + "step": 3001 + }, + { + "epoch": 1.26, + "learning_rate": 1.916463759489143e-05, + "loss": 0.4195, + "step": 3002 + }, + { + "epoch": 1.26, + "learning_rate": 1.916395910007078e-05, + "loss": 0.4048, + "step": 3003 + }, + { + "epoch": 1.26, + "learning_rate": 1.9163280341840253e-05, + "loss": 0.3728, + "step": 3004 + }, + { + "epoch": 1.26, + "learning_rate": 1.9162601320219364e-05, + "loss": 0.419, + "step": 3005 + }, + { + "epoch": 1.26, + "learning_rate": 1.9161922035227633e-05, + "loss": 0.3842, + "step": 3006 + }, + { + "epoch": 1.26, + "learning_rate": 1.9161242486884584e-05, + "loss": 0.4437, + "step": 3007 + }, + { + "epoch": 1.26, + "learning_rate": 1.9160562675209746e-05, + "loss": 0.4448, + "step": 3008 + }, + { + "epoch": 1.26, + "learning_rate": 1.9159882600222664e-05, + "loss": 0.4838, + "step": 3009 + }, + { + "epoch": 1.26, + "learning_rate": 1.9159202261942886e-05, + "loss": 0.3661, + "step": 3010 + }, + { + "epoch": 1.26, + "learning_rate": 1.9158521660389966e-05, + "loss": 0.3588, + "step": 3011 + }, + { + "epoch": 1.26, + "learning_rate": 1.9157840795583466e-05, + "loss": 0.452, + "step": 3012 + }, + { + "epoch": 1.26, + "learning_rate": 1.9157159667542964e-05, + "loss": 0.3669, + "step": 3013 + }, + { + "epoch": 1.26, + "learning_rate": 1.9156478276288027e-05, + "loss": 0.4242, + "step": 3014 + }, + { + "epoch": 1.26, + "learning_rate": 1.9155796621838253e-05, + "loss": 0.3988, + "step": 3015 + }, + { + "epoch": 1.26, + "learning_rate": 1.915511470421323e-05, + "loss": 0.3667, + "step": 3016 + }, + { + "epoch": 1.26, + "learning_rate": 1.9154432523432553e-05, + "loss": 0.4029, + "step": 3017 + }, + { + "epoch": 1.26, + "learning_rate": 1.915375007951584e-05, + "loss": 0.3726, + "step": 3018 + }, + { + "epoch": 1.26, + "learning_rate": 1.9153067372482702e-05, + "loss": 0.4151, + "step": 3019 + }, + { + "epoch": 1.26, + "learning_rate": 1.915238440235277e-05, + "loss": 0.3597, + "step": 3020 + }, + { + "epoch": 1.26, + "learning_rate": 1.9151701169145664e-05, + "loss": 0.3604, + "step": 3021 + }, + { + "epoch": 1.27, + "learning_rate": 1.9151017672881032e-05, + "loss": 0.4356, + "step": 3022 + }, + { + "epoch": 1.27, + "learning_rate": 1.9150333913578518e-05, + "loss": 0.4045, + "step": 3023 + }, + { + "epoch": 1.27, + "learning_rate": 1.9149649891257772e-05, + "loss": 0.379, + "step": 3024 + }, + { + "epoch": 1.27, + "learning_rate": 1.9148965605938463e-05, + "loss": 0.3753, + "step": 3025 + }, + { + "epoch": 1.27, + "learning_rate": 1.9148281057640252e-05, + "loss": 0.4016, + "step": 3026 + }, + { + "epoch": 1.27, + "learning_rate": 1.914759624638282e-05, + "loss": 0.3454, + "step": 3027 + }, + { + "epoch": 1.27, + "learning_rate": 1.914691117218586e-05, + "loss": 0.3758, + "step": 3028 + }, + { + "epoch": 1.27, + "learning_rate": 1.914622583506905e-05, + "loss": 0.449, + "step": 3029 + }, + { + "epoch": 1.27, + "learning_rate": 1.914554023505209e-05, + "loss": 0.3803, + "step": 3030 + }, + { + "epoch": 1.27, + "learning_rate": 1.9144854372154697e-05, + "loss": 0.3536, + "step": 3031 + }, + { + "epoch": 1.27, + "learning_rate": 1.9144168246396576e-05, + "loss": 0.4062, + "step": 3032 + }, + { + "epoch": 1.27, + "learning_rate": 1.914348185779746e-05, + "loss": 0.3956, + "step": 3033 + }, + { + "epoch": 1.27, + "learning_rate": 1.9142795206377066e-05, + "loss": 0.4077, + "step": 3034 + }, + { + "epoch": 1.27, + "learning_rate": 1.914210829215514e-05, + "loss": 0.4146, + "step": 3035 + }, + { + "epoch": 1.27, + "learning_rate": 1.914142111515142e-05, + "loss": 0.4158, + "step": 3036 + }, + { + "epoch": 1.27, + "learning_rate": 1.9140733675385666e-05, + "loss": 0.4014, + "step": 3037 + }, + { + "epoch": 1.27, + "learning_rate": 1.914004597287763e-05, + "loss": 0.3813, + "step": 3038 + }, + { + "epoch": 1.27, + "learning_rate": 1.9139358007647085e-05, + "loss": 0.3681, + "step": 3039 + }, + { + "epoch": 1.27, + "learning_rate": 1.9138669779713807e-05, + "loss": 0.409, + "step": 3040 + }, + { + "epoch": 1.27, + "learning_rate": 1.9137981289097573e-05, + "loss": 0.4381, + "step": 3041 + }, + { + "epoch": 1.27, + "learning_rate": 1.9137292535818174e-05, + "loss": 0.3638, + "step": 3042 + }, + { + "epoch": 1.27, + "learning_rate": 1.913660351989541e-05, + "loss": 0.4291, + "step": 3043 + }, + { + "epoch": 1.27, + "learning_rate": 1.9135914241349088e-05, + "loss": 0.4002, + "step": 3044 + }, + { + "epoch": 1.27, + "learning_rate": 1.9135224700199015e-05, + "loss": 0.3697, + "step": 3045 + }, + { + "epoch": 1.28, + "learning_rate": 1.913453489646502e-05, + "loss": 0.4188, + "step": 3046 + }, + { + "epoch": 1.28, + "learning_rate": 1.913384483016692e-05, + "loss": 0.4102, + "step": 3047 + }, + { + "epoch": 1.28, + "learning_rate": 1.9133154501324557e-05, + "loss": 0.3559, + "step": 3048 + }, + { + "epoch": 1.28, + "learning_rate": 1.9132463909957774e-05, + "loss": 0.4174, + "step": 3049 + }, + { + "epoch": 1.28, + "learning_rate": 1.9131773056086417e-05, + "loss": 0.4116, + "step": 3050 + }, + { + "epoch": 1.28, + "learning_rate": 1.9131081939730348e-05, + "loss": 0.4361, + "step": 3051 + }, + { + "epoch": 1.28, + "learning_rate": 1.913039056090943e-05, + "loss": 0.3952, + "step": 3052 + }, + { + "epoch": 1.28, + "learning_rate": 1.912969891964354e-05, + "loss": 0.3878, + "step": 3053 + }, + { + "epoch": 1.28, + "learning_rate": 1.9129007015952556e-05, + "loss": 0.36, + "step": 3054 + }, + { + "epoch": 1.28, + "learning_rate": 1.9128314849856365e-05, + "loss": 0.3476, + "step": 3055 + }, + { + "epoch": 1.28, + "learning_rate": 1.9127622421374866e-05, + "loss": 0.3948, + "step": 3056 + }, + { + "epoch": 1.28, + "learning_rate": 1.9126929730527957e-05, + "loss": 0.3804, + "step": 3057 + }, + { + "epoch": 1.28, + "learning_rate": 1.9126236777335554e-05, + "loss": 0.4198, + "step": 3058 + }, + { + "epoch": 1.28, + "learning_rate": 1.9125543561817573e-05, + "loss": 0.3692, + "step": 3059 + }, + { + "epoch": 1.28, + "learning_rate": 1.9124850083993942e-05, + "loss": 0.3862, + "step": 3060 + }, + { + "epoch": 1.28, + "learning_rate": 1.912415634388459e-05, + "loss": 0.4109, + "step": 3061 + }, + { + "epoch": 1.28, + "learning_rate": 1.912346234150946e-05, + "loss": 0.3838, + "step": 3062 + }, + { + "epoch": 1.28, + "learning_rate": 1.9122768076888506e-05, + "loss": 0.3874, + "step": 3063 + }, + { + "epoch": 1.28, + "learning_rate": 1.9122073550041675e-05, + "loss": 0.4108, + "step": 3064 + }, + { + "epoch": 1.28, + "learning_rate": 1.9121378760988934e-05, + "loss": 0.3832, + "step": 3065 + }, + { + "epoch": 1.28, + "learning_rate": 1.9120683709750257e-05, + "loss": 0.3902, + "step": 3066 + }, + { + "epoch": 1.28, + "learning_rate": 1.9119988396345615e-05, + "loss": 0.3786, + "step": 3067 + }, + { + "epoch": 1.28, + "learning_rate": 1.9119292820795007e-05, + "loss": 0.4007, + "step": 3068 + }, + { + "epoch": 1.28, + "learning_rate": 1.9118596983118415e-05, + "loss": 0.3548, + "step": 3069 + }, + { + "epoch": 1.29, + "learning_rate": 1.9117900883335846e-05, + "loss": 0.3979, + "step": 3070 + }, + { + "epoch": 1.29, + "learning_rate": 1.9117204521467308e-05, + "loss": 0.3693, + "step": 3071 + }, + { + "epoch": 1.29, + "learning_rate": 1.9116507897532813e-05, + "loss": 0.3875, + "step": 3072 + }, + { + "epoch": 1.29, + "learning_rate": 1.911581101155239e-05, + "loss": 0.3935, + "step": 3073 + }, + { + "epoch": 1.29, + "learning_rate": 1.911511386354607e-05, + "loss": 0.3525, + "step": 3074 + }, + { + "epoch": 1.29, + "learning_rate": 1.911441645353389e-05, + "loss": 0.3911, + "step": 3075 + }, + { + "epoch": 1.29, + "learning_rate": 1.9113718781535896e-05, + "loss": 0.3531, + "step": 3076 + }, + { + "epoch": 1.29, + "learning_rate": 1.9113020847572144e-05, + "loss": 0.4023, + "step": 3077 + }, + { + "epoch": 1.29, + "learning_rate": 1.911232265166269e-05, + "loss": 0.3949, + "step": 3078 + }, + { + "epoch": 1.29, + "learning_rate": 1.9111624193827614e-05, + "loss": 0.3762, + "step": 3079 + }, + { + "epoch": 1.29, + "learning_rate": 1.9110925474086982e-05, + "loss": 0.3759, + "step": 3080 + }, + { + "epoch": 1.29, + "learning_rate": 1.9110226492460886e-05, + "loss": 0.4094, + "step": 3081 + }, + { + "epoch": 1.29, + "learning_rate": 1.9109527248969412e-05, + "loss": 0.4151, + "step": 3082 + }, + { + "epoch": 1.29, + "learning_rate": 1.910882774363266e-05, + "loss": 0.3605, + "step": 3083 + }, + { + "epoch": 1.29, + "learning_rate": 1.9108127976470737e-05, + "loss": 0.3758, + "step": 3084 + }, + { + "epoch": 1.29, + "learning_rate": 1.9107427947503757e-05, + "loss": 0.3833, + "step": 3085 + }, + { + "epoch": 1.29, + "learning_rate": 1.9106727656751846e-05, + "loss": 0.3986, + "step": 3086 + }, + { + "epoch": 1.29, + "learning_rate": 1.9106027104235126e-05, + "loss": 0.3984, + "step": 3087 + }, + { + "epoch": 1.29, + "learning_rate": 1.910532628997374e-05, + "loss": 0.4009, + "step": 3088 + }, + { + "epoch": 1.29, + "learning_rate": 1.910462521398783e-05, + "loss": 0.3672, + "step": 3089 + }, + { + "epoch": 1.29, + "learning_rate": 1.9103923876297544e-05, + "loss": 0.3765, + "step": 3090 + }, + { + "epoch": 1.29, + "learning_rate": 1.9103222276923047e-05, + "loss": 0.3638, + "step": 3091 + }, + { + "epoch": 1.29, + "learning_rate": 1.9102520415884504e-05, + "loss": 0.3996, + "step": 3092 + }, + { + "epoch": 1.29, + "learning_rate": 1.910181829320209e-05, + "loss": 0.3812, + "step": 3093 + }, + { + "epoch": 1.3, + "learning_rate": 1.9101115908895985e-05, + "loss": 0.3676, + "step": 3094 + }, + { + "epoch": 1.3, + "learning_rate": 1.9100413262986383e-05, + "loss": 0.3927, + "step": 3095 + }, + { + "epoch": 1.3, + "learning_rate": 1.909971035549347e-05, + "loss": 0.4282, + "step": 3096 + }, + { + "epoch": 1.3, + "learning_rate": 1.9099007186437465e-05, + "loss": 0.4489, + "step": 3097 + }, + { + "epoch": 1.3, + "learning_rate": 1.9098303755838566e-05, + "loss": 0.3486, + "step": 3098 + }, + { + "epoch": 1.3, + "learning_rate": 1.9097600063717005e-05, + "loss": 0.3687, + "step": 3099 + }, + { + "epoch": 1.3, + "learning_rate": 1.9096896110092998e-05, + "loss": 0.4529, + "step": 3100 + }, + { + "epoch": 1.3, + "learning_rate": 1.909619189498679e-05, + "loss": 0.3496, + "step": 3101 + }, + { + "epoch": 1.3, + "learning_rate": 1.9095487418418614e-05, + "loss": 0.4179, + "step": 3102 + }, + { + "epoch": 1.3, + "learning_rate": 1.9094782680408722e-05, + "loss": 0.4081, + "step": 3103 + }, + { + "epoch": 1.3, + "learning_rate": 1.9094077680977377e-05, + "loss": 0.3881, + "step": 3104 + }, + { + "epoch": 1.3, + "learning_rate": 1.9093372420144835e-05, + "loss": 0.4002, + "step": 3105 + }, + { + "epoch": 1.3, + "learning_rate": 1.9092666897931375e-05, + "loss": 0.3316, + "step": 3106 + }, + { + "epoch": 1.3, + "learning_rate": 1.9091961114357274e-05, + "loss": 0.4296, + "step": 3107 + }, + { + "epoch": 1.3, + "learning_rate": 1.909125506944282e-05, + "loss": 0.3749, + "step": 3108 + }, + { + "epoch": 1.3, + "learning_rate": 1.90905487632083e-05, + "loss": 0.427, + "step": 3109 + }, + { + "epoch": 1.3, + "learning_rate": 1.908984219567403e-05, + "loss": 0.386, + "step": 3110 + }, + { + "epoch": 1.3, + "learning_rate": 1.908913536686031e-05, + "loss": 0.4013, + "step": 3111 + }, + { + "epoch": 1.3, + "learning_rate": 1.9088428276787455e-05, + "loss": 0.3712, + "step": 3112 + }, + { + "epoch": 1.3, + "learning_rate": 1.90877209254758e-05, + "loss": 0.4584, + "step": 3113 + }, + { + "epoch": 1.3, + "learning_rate": 1.9087013312945668e-05, + "loss": 0.41, + "step": 3114 + }, + { + "epoch": 1.3, + "learning_rate": 1.9086305439217402e-05, + "loss": 0.3809, + "step": 3115 + }, + { + "epoch": 1.3, + "learning_rate": 1.908559730431135e-05, + "loss": 0.3908, + "step": 3116 + }, + { + "epoch": 1.31, + "learning_rate": 1.9084888908247864e-05, + "loss": 0.4066, + "step": 3117 + }, + { + "epoch": 1.31, + "learning_rate": 1.908418025104731e-05, + "loss": 0.4241, + "step": 3118 + }, + { + "epoch": 1.31, + "learning_rate": 1.9083471332730058e-05, + "loss": 0.3628, + "step": 3119 + }, + { + "epoch": 1.31, + "learning_rate": 1.908276215331648e-05, + "loss": 0.469, + "step": 3120 + }, + { + "epoch": 1.31, + "learning_rate": 1.9082052712826967e-05, + "loss": 0.3736, + "step": 3121 + }, + { + "epoch": 1.31, + "learning_rate": 1.9081343011281907e-05, + "loss": 0.3628, + "step": 3122 + }, + { + "epoch": 1.31, + "learning_rate": 1.9080633048701696e-05, + "loss": 0.436, + "step": 3123 + }, + { + "epoch": 1.31, + "learning_rate": 1.907992282510675e-05, + "loss": 0.4212, + "step": 3124 + }, + { + "epoch": 1.31, + "learning_rate": 1.907921234051748e-05, + "loss": 0.3986, + "step": 3125 + }, + { + "epoch": 1.31, + "learning_rate": 1.9078501594954305e-05, + "loss": 0.4126, + "step": 3126 + }, + { + "epoch": 1.31, + "learning_rate": 1.9077790588437663e-05, + "loss": 0.3649, + "step": 3127 + }, + { + "epoch": 1.31, + "learning_rate": 1.907707932098798e-05, + "loss": 0.3977, + "step": 3128 + }, + { + "epoch": 1.31, + "learning_rate": 1.9076367792625713e-05, + "loss": 0.368, + "step": 3129 + }, + { + "epoch": 1.31, + "learning_rate": 1.9075656003371304e-05, + "loss": 0.3954, + "step": 3130 + }, + { + "epoch": 1.31, + "learning_rate": 1.907494395324522e-05, + "loss": 0.3712, + "step": 3131 + }, + { + "epoch": 1.31, + "learning_rate": 1.907423164226792e-05, + "loss": 0.3879, + "step": 3132 + }, + { + "epoch": 1.31, + "learning_rate": 1.907351907045989e-05, + "loss": 0.375, + "step": 3133 + }, + { + "epoch": 1.31, + "learning_rate": 1.90728062378416e-05, + "loss": 0.4481, + "step": 3134 + }, + { + "epoch": 1.31, + "learning_rate": 1.907209314443355e-05, + "loss": 0.4373, + "step": 3135 + }, + { + "epoch": 1.31, + "learning_rate": 1.907137979025623e-05, + "loss": 0.4099, + "step": 3136 + }, + { + "epoch": 1.31, + "learning_rate": 1.907066617533015e-05, + "loss": 0.39, + "step": 3137 + }, + { + "epoch": 1.31, + "learning_rate": 1.906995229967582e-05, + "loss": 0.3978, + "step": 3138 + }, + { + "epoch": 1.31, + "learning_rate": 1.9069238163313757e-05, + "loss": 0.4477, + "step": 3139 + }, + { + "epoch": 1.31, + "learning_rate": 1.906852376626449e-05, + "loss": 0.3745, + "step": 3140 + }, + { + "epoch": 1.32, + "learning_rate": 1.906780910854856e-05, + "loss": 0.3677, + "step": 3141 + }, + { + "epoch": 1.32, + "learning_rate": 1.90670941901865e-05, + "loss": 0.359, + "step": 3142 + }, + { + "epoch": 1.32, + "learning_rate": 1.9066379011198863e-05, + "loss": 0.3517, + "step": 3143 + }, + { + "epoch": 1.32, + "learning_rate": 1.906566357160621e-05, + "loss": 0.3351, + "step": 3144 + }, + { + "epoch": 1.32, + "learning_rate": 1.9064947871429102e-05, + "loss": 0.3869, + "step": 3145 + }, + { + "epoch": 1.32, + "learning_rate": 1.9064231910688108e-05, + "loss": 0.4992, + "step": 3146 + }, + { + "epoch": 1.32, + "learning_rate": 1.9063515689403815e-05, + "loss": 0.3779, + "step": 3147 + }, + { + "epoch": 1.32, + "learning_rate": 1.90627992075968e-05, + "loss": 0.3634, + "step": 3148 + }, + { + "epoch": 1.32, + "learning_rate": 1.9062082465287672e-05, + "loss": 0.38, + "step": 3149 + }, + { + "epoch": 1.32, + "learning_rate": 1.9061365462497022e-05, + "loss": 0.3647, + "step": 3150 + }, + { + "epoch": 1.32, + "learning_rate": 1.9060648199245463e-05, + "loss": 0.4137, + "step": 3151 + }, + { + "epoch": 1.32, + "learning_rate": 1.905993067555361e-05, + "loss": 0.392, + "step": 3152 + }, + { + "epoch": 1.32, + "learning_rate": 1.905921289144209e-05, + "loss": 0.3798, + "step": 3153 + }, + { + "epoch": 1.32, + "learning_rate": 1.9058494846931538e-05, + "loss": 0.3981, + "step": 3154 + }, + { + "epoch": 1.32, + "learning_rate": 1.9057776542042587e-05, + "loss": 0.4565, + "step": 3155 + }, + { + "epoch": 1.32, + "learning_rate": 1.9057057976795884e-05, + "loss": 0.3855, + "step": 3156 + }, + { + "epoch": 1.32, + "learning_rate": 1.905633915121209e-05, + "loss": 0.3821, + "step": 3157 + }, + { + "epoch": 1.32, + "learning_rate": 1.905562006531186e-05, + "loss": 0.3874, + "step": 3158 + }, + { + "epoch": 1.32, + "learning_rate": 1.9054900719115874e-05, + "loss": 0.3713, + "step": 3159 + }, + { + "epoch": 1.32, + "learning_rate": 1.9054181112644797e-05, + "loss": 0.3942, + "step": 3160 + }, + { + "epoch": 1.32, + "learning_rate": 1.905346124591932e-05, + "loss": 0.4181, + "step": 3161 + }, + { + "epoch": 1.32, + "learning_rate": 1.9052741118960133e-05, + "loss": 0.405, + "step": 3162 + }, + { + "epoch": 1.32, + "learning_rate": 1.9052020731787936e-05, + "loss": 0.3995, + "step": 3163 + }, + { + "epoch": 1.32, + "learning_rate": 1.9051300084423433e-05, + "loss": 0.4052, + "step": 3164 + }, + { + "epoch": 1.33, + "learning_rate": 1.9050579176887343e-05, + "loss": 0.4243, + "step": 3165 + }, + { + "epoch": 1.33, + "learning_rate": 1.9049858009200388e-05, + "loss": 0.373, + "step": 3166 + }, + { + "epoch": 1.33, + "learning_rate": 1.9049136581383295e-05, + "loss": 0.4055, + "step": 3167 + }, + { + "epoch": 1.33, + "learning_rate": 1.90484148934568e-05, + "loss": 0.3674, + "step": 3168 + }, + { + "epoch": 1.33, + "learning_rate": 1.9047692945441647e-05, + "loss": 0.4224, + "step": 3169 + }, + { + "epoch": 1.33, + "learning_rate": 1.904697073735859e-05, + "loss": 0.4768, + "step": 3170 + }, + { + "epoch": 1.33, + "learning_rate": 1.9046248269228388e-05, + "loss": 0.3627, + "step": 3171 + }, + { + "epoch": 1.33, + "learning_rate": 1.9045525541071807e-05, + "loss": 0.3667, + "step": 3172 + }, + { + "epoch": 1.33, + "learning_rate": 1.904480255290962e-05, + "loss": 0.3859, + "step": 3173 + }, + { + "epoch": 1.33, + "learning_rate": 1.904407930476261e-05, + "loss": 0.3877, + "step": 3174 + }, + { + "epoch": 1.33, + "learning_rate": 1.9043355796651565e-05, + "loss": 0.4244, + "step": 3175 + }, + { + "epoch": 1.33, + "learning_rate": 1.9042632028597286e-05, + "loss": 0.4116, + "step": 3176 + }, + { + "epoch": 1.33, + "learning_rate": 1.904190800062057e-05, + "loss": 0.4461, + "step": 3177 + }, + { + "epoch": 1.33, + "learning_rate": 1.9041183712742233e-05, + "loss": 0.3593, + "step": 3178 + }, + { + "epoch": 1.33, + "learning_rate": 1.9040459164983095e-05, + "loss": 0.3896, + "step": 3179 + }, + { + "epoch": 1.33, + "learning_rate": 1.903973435736398e-05, + "loss": 0.4172, + "step": 3180 + }, + { + "epoch": 1.33, + "learning_rate": 1.903900928990572e-05, + "loss": 0.3866, + "step": 3181 + }, + { + "epoch": 1.33, + "learning_rate": 1.9038283962629164e-05, + "loss": 0.3818, + "step": 3182 + }, + { + "epoch": 1.33, + "learning_rate": 1.903755837555515e-05, + "loss": 0.3322, + "step": 3183 + }, + { + "epoch": 1.33, + "learning_rate": 1.9036832528704547e-05, + "loss": 0.4535, + "step": 3184 + }, + { + "epoch": 1.33, + "learning_rate": 1.9036106422098207e-05, + "loss": 0.4, + "step": 3185 + }, + { + "epoch": 1.33, + "learning_rate": 1.9035380055757007e-05, + "loss": 0.3732, + "step": 3186 + }, + { + "epoch": 1.33, + "learning_rate": 1.9034653429701824e-05, + "loss": 0.3552, + "step": 3187 + }, + { + "epoch": 1.33, + "learning_rate": 1.9033926543953546e-05, + "loss": 0.4101, + "step": 3188 + }, + { + "epoch": 1.34, + "learning_rate": 1.903319939853307e-05, + "loss": 0.409, + "step": 3189 + }, + { + "epoch": 1.34, + "learning_rate": 1.903247199346129e-05, + "loss": 0.3997, + "step": 3190 + }, + { + "epoch": 1.34, + "learning_rate": 1.9031744328759118e-05, + "loss": 0.4128, + "step": 3191 + }, + { + "epoch": 1.34, + "learning_rate": 1.903101640444747e-05, + "loss": 0.4079, + "step": 3192 + }, + { + "epoch": 1.34, + "learning_rate": 1.903028822054727e-05, + "loss": 0.3743, + "step": 3193 + }, + { + "epoch": 1.34, + "learning_rate": 1.902955977707945e-05, + "loss": 0.353, + "step": 3194 + }, + { + "epoch": 1.34, + "learning_rate": 1.9028831074064945e-05, + "loss": 0.3681, + "step": 3195 + }, + { + "epoch": 1.34, + "learning_rate": 1.9028102111524702e-05, + "loss": 0.4156, + "step": 3196 + }, + { + "epoch": 1.34, + "learning_rate": 1.9027372889479676e-05, + "loss": 0.4007, + "step": 3197 + }, + { + "epoch": 1.34, + "learning_rate": 1.9026643407950826e-05, + "loss": 0.3695, + "step": 3198 + }, + { + "epoch": 1.34, + "learning_rate": 1.9025913666959122e-05, + "loss": 0.3651, + "step": 3199 + }, + { + "epoch": 1.34, + "learning_rate": 1.9025183666525543e-05, + "loss": 0.3975, + "step": 3200 + }, + { + "epoch": 1.34, + "learning_rate": 1.9024453406671064e-05, + "loss": 0.4004, + "step": 3201 + }, + { + "epoch": 1.34, + "learning_rate": 1.902372288741668e-05, + "loss": 0.3727, + "step": 3202 + }, + { + "epoch": 1.34, + "learning_rate": 1.902299210878339e-05, + "loss": 0.4255, + "step": 3203 + }, + { + "epoch": 1.34, + "learning_rate": 1.9022261070792197e-05, + "loss": 0.3589, + "step": 3204 + }, + { + "epoch": 1.34, + "learning_rate": 1.902152977346412e-05, + "loss": 0.3909, + "step": 3205 + }, + { + "epoch": 1.34, + "learning_rate": 1.9020798216820174e-05, + "loss": 0.3787, + "step": 3206 + }, + { + "epoch": 1.34, + "learning_rate": 1.9020066400881386e-05, + "loss": 0.3855, + "step": 3207 + }, + { + "epoch": 1.34, + "learning_rate": 1.9019334325668795e-05, + "loss": 0.3937, + "step": 3208 + }, + { + "epoch": 1.34, + "learning_rate": 1.901860199120344e-05, + "loss": 0.448, + "step": 3209 + }, + { + "epoch": 1.34, + "learning_rate": 1.9017869397506374e-05, + "loss": 0.4216, + "step": 3210 + }, + { + "epoch": 1.34, + "learning_rate": 1.9017136544598657e-05, + "loss": 0.3929, + "step": 3211 + }, + { + "epoch": 1.34, + "learning_rate": 1.901640343250135e-05, + "loss": 0.4304, + "step": 3212 + }, + { + "epoch": 1.35, + "learning_rate": 1.9015670061235528e-05, + "loss": 0.3553, + "step": 3213 + }, + { + "epoch": 1.35, + "learning_rate": 1.901493643082227e-05, + "loss": 0.3978, + "step": 3214 + }, + { + "epoch": 1.35, + "learning_rate": 1.9014202541282663e-05, + "loss": 0.425, + "step": 3215 + }, + { + "epoch": 1.35, + "learning_rate": 1.901346839263781e-05, + "loss": 0.4167, + "step": 3216 + }, + { + "epoch": 1.35, + "learning_rate": 1.9012733984908798e-05, + "loss": 0.3343, + "step": 3217 + }, + { + "epoch": 1.35, + "learning_rate": 1.901199931811675e-05, + "loss": 0.3673, + "step": 3218 + }, + { + "epoch": 1.35, + "learning_rate": 1.9011264392282774e-05, + "loss": 0.4347, + "step": 3219 + }, + { + "epoch": 1.35, + "learning_rate": 1.9010529207428004e-05, + "loss": 0.337, + "step": 3220 + }, + { + "epoch": 1.35, + "learning_rate": 1.9009793763573564e-05, + "loss": 0.4128, + "step": 3221 + }, + { + "epoch": 1.35, + "learning_rate": 1.9009058060740603e-05, + "loss": 0.3688, + "step": 3222 + }, + { + "epoch": 1.35, + "learning_rate": 1.900832209895026e-05, + "loss": 0.4163, + "step": 3223 + }, + { + "epoch": 1.35, + "learning_rate": 1.900758587822369e-05, + "loss": 0.417, + "step": 3224 + }, + { + "epoch": 1.35, + "learning_rate": 1.900684939858206e-05, + "loss": 0.3658, + "step": 3225 + }, + { + "epoch": 1.35, + "learning_rate": 1.9006112660046535e-05, + "loss": 0.3795, + "step": 3226 + }, + { + "epoch": 1.35, + "learning_rate": 1.9005375662638294e-05, + "loss": 0.3625, + "step": 3227 + }, + { + "epoch": 1.35, + "learning_rate": 1.900463840637852e-05, + "loss": 0.3675, + "step": 3228 + }, + { + "epoch": 1.35, + "learning_rate": 1.9003900891288406e-05, + "loss": 0.3909, + "step": 3229 + }, + { + "epoch": 1.35, + "learning_rate": 1.900316311738915e-05, + "loss": 0.4061, + "step": 3230 + }, + { + "epoch": 1.35, + "learning_rate": 1.9002425084701958e-05, + "loss": 0.3623, + "step": 3231 + }, + { + "epoch": 1.35, + "learning_rate": 1.900168679324805e-05, + "loss": 0.4009, + "step": 3232 + }, + { + "epoch": 1.35, + "learning_rate": 1.900094824304864e-05, + "loss": 0.373, + "step": 3233 + }, + { + "epoch": 1.35, + "learning_rate": 1.900020943412496e-05, + "loss": 0.3667, + "step": 3234 + }, + { + "epoch": 1.35, + "learning_rate": 1.899947036649825e-05, + "loss": 0.3848, + "step": 3235 + }, + { + "epoch": 1.35, + "learning_rate": 1.8998731040189748e-05, + "loss": 0.4338, + "step": 3236 + }, + { + "epoch": 1.36, + "learning_rate": 1.8997991455220702e-05, + "loss": 0.4409, + "step": 3237 + }, + { + "epoch": 1.36, + "learning_rate": 1.8997251611612385e-05, + "loss": 0.4362, + "step": 3238 + }, + { + "epoch": 1.36, + "learning_rate": 1.899651150938605e-05, + "loss": 0.3413, + "step": 3239 + }, + { + "epoch": 1.36, + "learning_rate": 1.8995771148562978e-05, + "loss": 0.4013, + "step": 3240 + }, + { + "epoch": 1.36, + "learning_rate": 1.8995030529164446e-05, + "loss": 0.3302, + "step": 3241 + }, + { + "epoch": 1.36, + "learning_rate": 1.8994289651211742e-05, + "loss": 0.3261, + "step": 3242 + }, + { + "epoch": 1.36, + "learning_rate": 1.8993548514726162e-05, + "loss": 0.4185, + "step": 3243 + }, + { + "epoch": 1.36, + "learning_rate": 1.8992807119729013e-05, + "loss": 0.4119, + "step": 3244 + }, + { + "epoch": 1.36, + "learning_rate": 1.8992065466241603e-05, + "loss": 0.3402, + "step": 3245 + }, + { + "epoch": 1.36, + "learning_rate": 1.899132355428525e-05, + "loss": 0.3759, + "step": 3246 + }, + { + "epoch": 1.36, + "learning_rate": 1.899058138388128e-05, + "loss": 0.4098, + "step": 3247 + }, + { + "epoch": 1.36, + "learning_rate": 1.8989838955051027e-05, + "loss": 0.4127, + "step": 3248 + }, + { + "epoch": 1.36, + "learning_rate": 1.898909626781583e-05, + "loss": 0.3761, + "step": 3249 + }, + { + "epoch": 1.36, + "learning_rate": 1.898835332219704e-05, + "loss": 0.3541, + "step": 3250 + }, + { + "epoch": 1.36, + "learning_rate": 1.8987610118216006e-05, + "loss": 0.4133, + "step": 3251 + }, + { + "epoch": 1.36, + "learning_rate": 1.8986866655894098e-05, + "loss": 0.4228, + "step": 3252 + }, + { + "epoch": 1.36, + "learning_rate": 1.8986122935252684e-05, + "loss": 0.4267, + "step": 3253 + }, + { + "epoch": 1.36, + "learning_rate": 1.8985378956313133e-05, + "loss": 0.3861, + "step": 3254 + }, + { + "epoch": 1.36, + "learning_rate": 1.8984634719096844e-05, + "loss": 0.3509, + "step": 3255 + }, + { + "epoch": 1.36, + "learning_rate": 1.8983890223625205e-05, + "loss": 0.3746, + "step": 3256 + }, + { + "epoch": 1.36, + "learning_rate": 1.8983145469919606e-05, + "loss": 0.4175, + "step": 3257 + }, + { + "epoch": 1.36, + "learning_rate": 1.8982400458001465e-05, + "loss": 0.4045, + "step": 3258 + }, + { + "epoch": 1.36, + "learning_rate": 1.8981655187892194e-05, + "loss": 0.3661, + "step": 3259 + }, + { + "epoch": 1.36, + "learning_rate": 1.8980909659613217e-05, + "loss": 0.4329, + "step": 3260 + }, + { + "epoch": 1.37, + "learning_rate": 1.898016387318596e-05, + "loss": 0.3702, + "step": 3261 + }, + { + "epoch": 1.37, + "learning_rate": 1.8979417828631858e-05, + "loss": 0.4432, + "step": 3262 + }, + { + "epoch": 1.37, + "learning_rate": 1.8978671525972365e-05, + "loss": 0.3797, + "step": 3263 + }, + { + "epoch": 1.37, + "learning_rate": 1.8977924965228923e-05, + "loss": 0.3628, + "step": 3264 + }, + { + "epoch": 1.37, + "learning_rate": 1.8977178146422992e-05, + "loss": 0.3518, + "step": 3265 + }, + { + "epoch": 1.37, + "learning_rate": 1.8976431069576047e-05, + "loss": 0.3573, + "step": 3266 + }, + { + "epoch": 1.37, + "learning_rate": 1.8975683734709554e-05, + "loss": 0.3757, + "step": 3267 + }, + { + "epoch": 1.37, + "learning_rate": 1.8974936141844997e-05, + "loss": 0.37, + "step": 3268 + }, + { + "epoch": 1.37, + "learning_rate": 1.8974188291003864e-05, + "loss": 0.3833, + "step": 3269 + }, + { + "epoch": 1.37, + "learning_rate": 1.897344018220765e-05, + "loss": 0.3984, + "step": 3270 + }, + { + "epoch": 1.37, + "learning_rate": 1.8972691815477864e-05, + "loss": 0.4031, + "step": 3271 + }, + { + "epoch": 1.37, + "learning_rate": 1.8971943190836013e-05, + "loss": 0.3859, + "step": 3272 + }, + { + "epoch": 1.37, + "learning_rate": 1.8971194308303617e-05, + "loss": 0.3791, + "step": 3273 + }, + { + "epoch": 1.37, + "learning_rate": 1.89704451679022e-05, + "loss": 0.4298, + "step": 3274 + }, + { + "epoch": 1.37, + "learning_rate": 1.89696957696533e-05, + "loss": 0.3776, + "step": 3275 + }, + { + "epoch": 1.37, + "learning_rate": 1.8968946113578455e-05, + "loss": 0.3676, + "step": 3276 + }, + { + "epoch": 1.37, + "learning_rate": 1.896819619969921e-05, + "loss": 0.393, + "step": 3277 + }, + { + "epoch": 1.37, + "learning_rate": 1.8967446028037126e-05, + "loss": 0.372, + "step": 3278 + }, + { + "epoch": 1.37, + "learning_rate": 1.8966695598613758e-05, + "loss": 0.3877, + "step": 3279 + }, + { + "epoch": 1.37, + "learning_rate": 1.8965944911450685e-05, + "loss": 0.3731, + "step": 3280 + }, + { + "epoch": 1.37, + "learning_rate": 1.8965193966569483e-05, + "loss": 0.417, + "step": 3281 + }, + { + "epoch": 1.37, + "learning_rate": 1.8964442763991737e-05, + "loss": 0.4348, + "step": 3282 + }, + { + "epoch": 1.37, + "learning_rate": 1.8963691303739037e-05, + "loss": 0.4124, + "step": 3283 + }, + { + "epoch": 1.37, + "learning_rate": 1.8962939585832985e-05, + "loss": 0.3836, + "step": 3284 + }, + { + "epoch": 1.38, + "learning_rate": 1.8962187610295188e-05, + "loss": 0.4002, + "step": 3285 + }, + { + "epoch": 1.38, + "learning_rate": 1.896143537714726e-05, + "loss": 0.3544, + "step": 3286 + }, + { + "epoch": 1.38, + "learning_rate": 1.8960682886410825e-05, + "loss": 0.3982, + "step": 3287 + }, + { + "epoch": 1.38, + "learning_rate": 1.8959930138107512e-05, + "loss": 0.3865, + "step": 3288 + }, + { + "epoch": 1.38, + "learning_rate": 1.895917713225896e-05, + "loss": 0.3951, + "step": 3289 + }, + { + "epoch": 1.38, + "learning_rate": 1.895842386888681e-05, + "loss": 0.3978, + "step": 3290 + }, + { + "epoch": 1.38, + "learning_rate": 1.8957670348012717e-05, + "loss": 0.3807, + "step": 3291 + }, + { + "epoch": 1.38, + "learning_rate": 1.8956916569658336e-05, + "loss": 0.396, + "step": 3292 + }, + { + "epoch": 1.38, + "learning_rate": 1.895616253384534e-05, + "loss": 0.3715, + "step": 3293 + }, + { + "epoch": 1.38, + "learning_rate": 1.8955408240595396e-05, + "loss": 0.4017, + "step": 3294 + }, + { + "epoch": 1.38, + "learning_rate": 1.895465368993019e-05, + "loss": 0.4133, + "step": 3295 + }, + { + "epoch": 1.38, + "learning_rate": 1.8953898881871412e-05, + "loss": 0.3663, + "step": 3296 + }, + { + "epoch": 1.38, + "learning_rate": 1.895314381644075e-05, + "loss": 0.4083, + "step": 3297 + }, + { + "epoch": 1.38, + "learning_rate": 1.8952388493659918e-05, + "loss": 0.3797, + "step": 3298 + }, + { + "epoch": 1.38, + "learning_rate": 1.8951632913550625e-05, + "loss": 0.3964, + "step": 3299 + }, + { + "epoch": 1.38, + "learning_rate": 1.8950877076134583e-05, + "loss": 0.4279, + "step": 3300 + }, + { + "epoch": 1.38, + "learning_rate": 1.8950120981433527e-05, + "loss": 0.3264, + "step": 3301 + }, + { + "epoch": 1.38, + "learning_rate": 1.8949364629469184e-05, + "loss": 0.3627, + "step": 3302 + }, + { + "epoch": 1.38, + "learning_rate": 1.8948608020263293e-05, + "loss": 0.3707, + "step": 3303 + }, + { + "epoch": 1.38, + "learning_rate": 1.894785115383761e-05, + "loss": 0.3831, + "step": 3304 + }, + { + "epoch": 1.38, + "learning_rate": 1.8947094030213883e-05, + "loss": 0.3846, + "step": 3305 + }, + { + "epoch": 1.38, + "learning_rate": 1.8946336649413883e-05, + "loss": 0.3935, + "step": 3306 + }, + { + "epoch": 1.38, + "learning_rate": 1.8945579011459366e-05, + "loss": 0.4725, + "step": 3307 + }, + { + "epoch": 1.38, + "learning_rate": 1.8944821116372125e-05, + "loss": 0.401, + "step": 3308 + }, + { + "epoch": 1.39, + "learning_rate": 1.8944062964173934e-05, + "loss": 0.446, + "step": 3309 + }, + { + "epoch": 1.39, + "learning_rate": 1.8943304554886594e-05, + "loss": 0.3866, + "step": 3310 + }, + { + "epoch": 1.39, + "learning_rate": 1.8942545888531897e-05, + "loss": 0.4263, + "step": 3311 + }, + { + "epoch": 1.39, + "learning_rate": 1.8941786965131658e-05, + "loss": 0.4343, + "step": 3312 + }, + { + "epoch": 1.39, + "learning_rate": 1.8941027784707685e-05, + "loss": 0.3569, + "step": 3313 + }, + { + "epoch": 1.39, + "learning_rate": 1.8940268347281802e-05, + "loss": 0.3838, + "step": 3314 + }, + { + "epoch": 1.39, + "learning_rate": 1.8939508652875837e-05, + "loss": 0.3983, + "step": 3315 + }, + { + "epoch": 1.39, + "learning_rate": 1.8938748701511633e-05, + "loss": 0.3926, + "step": 3316 + }, + { + "epoch": 1.39, + "learning_rate": 1.8937988493211027e-05, + "loss": 0.3967, + "step": 3317 + }, + { + "epoch": 1.39, + "learning_rate": 1.8937228027995874e-05, + "loss": 0.3469, + "step": 3318 + }, + { + "epoch": 1.39, + "learning_rate": 1.8936467305888033e-05, + "loss": 0.3752, + "step": 3319 + }, + { + "epoch": 1.39, + "learning_rate": 1.893570632690937e-05, + "loss": 0.3537, + "step": 3320 + }, + { + "epoch": 1.39, + "learning_rate": 1.8934945091081755e-05, + "loss": 0.3916, + "step": 3321 + }, + { + "epoch": 1.39, + "learning_rate": 1.8934183598427072e-05, + "loss": 0.3867, + "step": 3322 + }, + { + "epoch": 1.39, + "learning_rate": 1.8933421848967213e-05, + "loss": 0.3915, + "step": 3323 + }, + { + "epoch": 1.39, + "learning_rate": 1.8932659842724067e-05, + "loss": 0.3388, + "step": 3324 + }, + { + "epoch": 1.39, + "learning_rate": 1.8931897579719543e-05, + "loss": 0.4215, + "step": 3325 + }, + { + "epoch": 1.39, + "learning_rate": 1.8931135059975548e-05, + "loss": 0.519, + "step": 3326 + }, + { + "epoch": 1.39, + "learning_rate": 1.8930372283514004e-05, + "loss": 0.3756, + "step": 3327 + }, + { + "epoch": 1.39, + "learning_rate": 1.8929609250356827e-05, + "loss": 0.354, + "step": 3328 + }, + { + "epoch": 1.39, + "learning_rate": 1.8928845960525957e-05, + "loss": 0.3802, + "step": 3329 + }, + { + "epoch": 1.39, + "learning_rate": 1.8928082414043342e-05, + "loss": 0.3898, + "step": 3330 + }, + { + "epoch": 1.39, + "learning_rate": 1.8927318610930912e-05, + "loss": 0.414, + "step": 3331 + }, + { + "epoch": 1.4, + "learning_rate": 1.892655455121063e-05, + "loss": 0.4356, + "step": 3332 + }, + { + "epoch": 1.4, + "learning_rate": 1.8925790234904463e-05, + "loss": 0.3701, + "step": 3333 + }, + { + "epoch": 1.4, + "learning_rate": 1.8925025662034376e-05, + "loss": 0.4053, + "step": 3334 + }, + { + "epoch": 1.4, + "learning_rate": 1.8924260832622347e-05, + "loss": 0.3682, + "step": 3335 + }, + { + "epoch": 1.4, + "learning_rate": 1.8923495746690354e-05, + "loss": 0.3782, + "step": 3336 + }, + { + "epoch": 1.4, + "learning_rate": 1.89227304042604e-05, + "loss": 0.4035, + "step": 3337 + }, + { + "epoch": 1.4, + "learning_rate": 1.8921964805354475e-05, + "loss": 0.3436, + "step": 3338 + }, + { + "epoch": 1.4, + "learning_rate": 1.892119894999459e-05, + "loss": 0.3769, + "step": 3339 + }, + { + "epoch": 1.4, + "learning_rate": 1.892043283820276e-05, + "loss": 0.4044, + "step": 3340 + }, + { + "epoch": 1.4, + "learning_rate": 1.8919666470001003e-05, + "loss": 0.4397, + "step": 3341 + }, + { + "epoch": 1.4, + "learning_rate": 1.8918899845411343e-05, + "loss": 0.3762, + "step": 3342 + }, + { + "epoch": 1.4, + "learning_rate": 1.891813296445583e-05, + "loss": 0.3667, + "step": 3343 + }, + { + "epoch": 1.4, + "learning_rate": 1.8917365827156493e-05, + "loss": 0.4077, + "step": 3344 + }, + { + "epoch": 1.4, + "learning_rate": 1.891659843353539e-05, + "loss": 0.4158, + "step": 3345 + }, + { + "epoch": 1.4, + "learning_rate": 1.8915830783614576e-05, + "loss": 0.3733, + "step": 3346 + }, + { + "epoch": 1.4, + "learning_rate": 1.891506287741612e-05, + "loss": 0.3935, + "step": 3347 + }, + { + "epoch": 1.4, + "learning_rate": 1.8914294714962093e-05, + "loss": 0.3812, + "step": 3348 + }, + { + "epoch": 1.4, + "learning_rate": 1.8913526296274573e-05, + "loss": 0.3571, + "step": 3349 + }, + { + "epoch": 1.4, + "learning_rate": 1.8912757621375654e-05, + "loss": 0.3853, + "step": 3350 + }, + { + "epoch": 1.4, + "learning_rate": 1.8911988690287422e-05, + "loss": 0.412, + "step": 3351 + }, + { + "epoch": 1.4, + "learning_rate": 1.891121950303198e-05, + "loss": 0.4038, + "step": 3352 + }, + { + "epoch": 1.4, + "learning_rate": 1.891045005963145e-05, + "loss": 0.3953, + "step": 3353 + }, + { + "epoch": 1.4, + "learning_rate": 1.890968036010793e-05, + "loss": 0.4022, + "step": 3354 + }, + { + "epoch": 1.4, + "learning_rate": 1.8908910404483564e-05, + "loss": 0.3732, + "step": 3355 + }, + { + "epoch": 1.41, + "learning_rate": 1.8908140192780466e-05, + "loss": 0.3713, + "step": 3356 + }, + { + "epoch": 1.41, + "learning_rate": 1.890736972502079e-05, + "loss": 0.4261, + "step": 3357 + }, + { + "epoch": 1.41, + "learning_rate": 1.890659900122667e-05, + "loss": 0.3903, + "step": 3358 + }, + { + "epoch": 1.41, + "learning_rate": 1.890582802142027e-05, + "loss": 0.4046, + "step": 3359 + }, + { + "epoch": 1.41, + "learning_rate": 1.8905056785623742e-05, + "loss": 0.3788, + "step": 3360 + }, + { + "epoch": 1.41, + "learning_rate": 1.890428529385926e-05, + "loss": 0.3524, + "step": 3361 + }, + { + "epoch": 1.41, + "learning_rate": 1.8903513546149e-05, + "loss": 0.3807, + "step": 3362 + }, + { + "epoch": 1.41, + "learning_rate": 1.8902741542515143e-05, + "loss": 0.33, + "step": 3363 + }, + { + "epoch": 1.41, + "learning_rate": 1.8901969282979882e-05, + "loss": 0.3963, + "step": 3364 + }, + { + "epoch": 1.41, + "learning_rate": 1.890119676756541e-05, + "loss": 0.3798, + "step": 3365 + }, + { + "epoch": 1.41, + "learning_rate": 1.8900423996293938e-05, + "loss": 0.4029, + "step": 3366 + }, + { + "epoch": 1.41, + "learning_rate": 1.889965096918768e-05, + "loss": 0.3793, + "step": 3367 + }, + { + "epoch": 1.41, + "learning_rate": 1.889887768626885e-05, + "loss": 0.3876, + "step": 3368 + }, + { + "epoch": 1.41, + "learning_rate": 1.8898104147559674e-05, + "loss": 0.3914, + "step": 3369 + }, + { + "epoch": 1.41, + "learning_rate": 1.8897330353082392e-05, + "loss": 0.3742, + "step": 3370 + }, + { + "epoch": 1.41, + "learning_rate": 1.8896556302859243e-05, + "loss": 0.4183, + "step": 3371 + }, + { + "epoch": 1.41, + "learning_rate": 1.889578199691248e-05, + "loss": 0.4014, + "step": 3372 + }, + { + "epoch": 1.41, + "learning_rate": 1.889500743526436e-05, + "loss": 0.3522, + "step": 3373 + }, + { + "epoch": 1.41, + "learning_rate": 1.889423261793714e-05, + "loss": 0.4073, + "step": 3374 + }, + { + "epoch": 1.41, + "learning_rate": 1.88934575449531e-05, + "loss": 0.3872, + "step": 3375 + }, + { + "epoch": 1.41, + "learning_rate": 1.889268221633451e-05, + "loss": 0.3361, + "step": 3376 + }, + { + "epoch": 1.41, + "learning_rate": 1.8891906632103664e-05, + "loss": 0.4251, + "step": 3377 + }, + { + "epoch": 1.41, + "learning_rate": 1.889113079228285e-05, + "loss": 0.3913, + "step": 3378 + }, + { + "epoch": 1.41, + "learning_rate": 1.8890354696894374e-05, + "loss": 0.3846, + "step": 3379 + }, + { + "epoch": 1.42, + "learning_rate": 1.888957834596054e-05, + "loss": 0.3444, + "step": 3380 + }, + { + "epoch": 1.42, + "learning_rate": 1.8888801739503667e-05, + "loss": 0.4703, + "step": 3381 + }, + { + "epoch": 1.42, + "learning_rate": 1.8888024877546074e-05, + "loss": 0.3833, + "step": 3382 + }, + { + "epoch": 1.42, + "learning_rate": 1.8887247760110094e-05, + "loss": 0.3652, + "step": 3383 + }, + { + "epoch": 1.42, + "learning_rate": 1.888647038721806e-05, + "loss": 0.4107, + "step": 3384 + }, + { + "epoch": 1.42, + "learning_rate": 1.8885692758892324e-05, + "loss": 0.3597, + "step": 3385 + }, + { + "epoch": 1.42, + "learning_rate": 1.8884914875155235e-05, + "loss": 0.3862, + "step": 3386 + }, + { + "epoch": 1.42, + "learning_rate": 1.8884136736029154e-05, + "loss": 0.4022, + "step": 3387 + }, + { + "epoch": 1.42, + "learning_rate": 1.8883358341536442e-05, + "loss": 0.4315, + "step": 3388 + }, + { + "epoch": 1.42, + "learning_rate": 1.8882579691699475e-05, + "loss": 0.4305, + "step": 3389 + }, + { + "epoch": 1.42, + "learning_rate": 1.8881800786540642e-05, + "loss": 0.3692, + "step": 3390 + }, + { + "epoch": 1.42, + "learning_rate": 1.8881021626082328e-05, + "loss": 0.3415, + "step": 3391 + }, + { + "epoch": 1.42, + "learning_rate": 1.8880242210346925e-05, + "loss": 0.4347, + "step": 3392 + }, + { + "epoch": 1.42, + "learning_rate": 1.8879462539356842e-05, + "loss": 0.3729, + "step": 3393 + }, + { + "epoch": 1.42, + "learning_rate": 1.8878682613134486e-05, + "loss": 0.3826, + "step": 3394 + }, + { + "epoch": 1.42, + "learning_rate": 1.8877902431702274e-05, + "loss": 0.4082, + "step": 3395 + }, + { + "epoch": 1.42, + "learning_rate": 1.8877121995082638e-05, + "loss": 0.3841, + "step": 3396 + }, + { + "epoch": 1.42, + "learning_rate": 1.8876341303298012e-05, + "loss": 0.3887, + "step": 3397 + }, + { + "epoch": 1.42, + "learning_rate": 1.8875560356370827e-05, + "loss": 0.3646, + "step": 3398 + }, + { + "epoch": 1.42, + "learning_rate": 1.8874779154323538e-05, + "loss": 0.3887, + "step": 3399 + }, + { + "epoch": 1.42, + "learning_rate": 1.8873997697178593e-05, + "loss": 0.3638, + "step": 3400 + }, + { + "epoch": 1.42, + "learning_rate": 1.8873215984958465e-05, + "loss": 0.4295, + "step": 3401 + }, + { + "epoch": 1.42, + "learning_rate": 1.8872434017685613e-05, + "loss": 0.3792, + "step": 3402 + }, + { + "epoch": 1.42, + "learning_rate": 1.8871651795382518e-05, + "loss": 0.3996, + "step": 3403 + }, + { + "epoch": 1.43, + "learning_rate": 1.8870869318071667e-05, + "loss": 0.4269, + "step": 3404 + }, + { + "epoch": 1.43, + "learning_rate": 1.8870086585775548e-05, + "loss": 0.3915, + "step": 3405 + }, + { + "epoch": 1.43, + "learning_rate": 1.8869303598516663e-05, + "loss": 0.3593, + "step": 3406 + }, + { + "epoch": 1.43, + "learning_rate": 1.8868520356317517e-05, + "loss": 0.3277, + "step": 3407 + }, + { + "epoch": 1.43, + "learning_rate": 1.886773685920062e-05, + "loss": 0.3957, + "step": 3408 + }, + { + "epoch": 1.43, + "learning_rate": 1.8866953107188498e-05, + "loss": 0.4089, + "step": 3409 + }, + { + "epoch": 1.43, + "learning_rate": 1.8866169100303677e-05, + "loss": 0.368, + "step": 3410 + }, + { + "epoch": 1.43, + "learning_rate": 1.8865384838568693e-05, + "loss": 0.3971, + "step": 3411 + }, + { + "epoch": 1.43, + "learning_rate": 1.886460032200609e-05, + "loss": 0.4052, + "step": 3412 + }, + { + "epoch": 1.43, + "learning_rate": 1.8863815550638414e-05, + "loss": 0.3769, + "step": 3413 + }, + { + "epoch": 1.43, + "learning_rate": 1.886303052448823e-05, + "loss": 0.3886, + "step": 3414 + }, + { + "epoch": 1.43, + "learning_rate": 1.8862245243578097e-05, + "loss": 0.4253, + "step": 3415 + }, + { + "epoch": 1.43, + "learning_rate": 1.8861459707930585e-05, + "loss": 0.3974, + "step": 3416 + }, + { + "epoch": 1.43, + "learning_rate": 1.886067391756828e-05, + "loss": 0.4251, + "step": 3417 + }, + { + "epoch": 1.43, + "learning_rate": 1.885988787251377e-05, + "loss": 0.3961, + "step": 3418 + }, + { + "epoch": 1.43, + "learning_rate": 1.885910157278964e-05, + "loss": 0.3985, + "step": 3419 + }, + { + "epoch": 1.43, + "learning_rate": 1.8858315018418497e-05, + "loss": 0.4277, + "step": 3420 + }, + { + "epoch": 1.43, + "learning_rate": 1.885752820942295e-05, + "loss": 0.4149, + "step": 3421 + }, + { + "epoch": 1.43, + "learning_rate": 1.8856741145825616e-05, + "loss": 0.3769, + "step": 3422 + }, + { + "epoch": 1.43, + "learning_rate": 1.8855953827649115e-05, + "loss": 0.3979, + "step": 3423 + }, + { + "epoch": 1.43, + "learning_rate": 1.8855166254916082e-05, + "loss": 0.398, + "step": 3424 + }, + { + "epoch": 1.43, + "learning_rate": 1.885437842764915e-05, + "loss": 0.3734, + "step": 3425 + }, + { + "epoch": 1.43, + "learning_rate": 1.885359034587097e-05, + "loss": 0.3775, + "step": 3426 + }, + { + "epoch": 1.43, + "learning_rate": 1.885280200960419e-05, + "loss": 0.3443, + "step": 3427 + }, + { + "epoch": 1.44, + "learning_rate": 1.8852013418871474e-05, + "loss": 0.4162, + "step": 3428 + }, + { + "epoch": 1.44, + "learning_rate": 1.885122457369549e-05, + "loss": 0.3529, + "step": 3429 + }, + { + "epoch": 1.44, + "learning_rate": 1.8850435474098903e-05, + "loss": 0.4343, + "step": 3430 + }, + { + "epoch": 1.44, + "learning_rate": 1.884964612010441e-05, + "loss": 0.3688, + "step": 3431 + }, + { + "epoch": 1.44, + "learning_rate": 1.884885651173469e-05, + "loss": 0.4096, + "step": 3432 + }, + { + "epoch": 1.44, + "learning_rate": 1.8848066649012437e-05, + "loss": 0.3725, + "step": 3433 + }, + { + "epoch": 1.44, + "learning_rate": 1.8847276531960363e-05, + "loss": 0.3621, + "step": 3434 + }, + { + "epoch": 1.44, + "learning_rate": 1.8846486160601175e-05, + "loss": 0.3706, + "step": 3435 + }, + { + "epoch": 1.44, + "learning_rate": 1.8845695534957594e-05, + "loss": 0.4113, + "step": 3436 + }, + { + "epoch": 1.44, + "learning_rate": 1.8844904655052344e-05, + "loss": 0.3906, + "step": 3437 + }, + { + "epoch": 1.44, + "learning_rate": 1.884411352090816e-05, + "loss": 0.3552, + "step": 3438 + }, + { + "epoch": 1.44, + "learning_rate": 1.8843322132547778e-05, + "loss": 0.4262, + "step": 3439 + }, + { + "epoch": 1.44, + "learning_rate": 1.884253048999395e-05, + "loss": 0.4011, + "step": 3440 + }, + { + "epoch": 1.44, + "learning_rate": 1.8841738593269432e-05, + "loss": 0.3708, + "step": 3441 + }, + { + "epoch": 1.44, + "learning_rate": 1.884094644239698e-05, + "loss": 0.3951, + "step": 3442 + }, + { + "epoch": 1.44, + "learning_rate": 1.8840154037399368e-05, + "loss": 0.3575, + "step": 3443 + }, + { + "epoch": 1.44, + "learning_rate": 1.8839361378299376e-05, + "loss": 0.4257, + "step": 3444 + }, + { + "epoch": 1.44, + "learning_rate": 1.8838568465119786e-05, + "loss": 0.3638, + "step": 3445 + }, + { + "epoch": 1.44, + "learning_rate": 1.8837775297883384e-05, + "loss": 0.3661, + "step": 3446 + }, + { + "epoch": 1.44, + "learning_rate": 1.883698187661298e-05, + "loss": 0.4143, + "step": 3447 + }, + { + "epoch": 1.44, + "learning_rate": 1.8836188201331368e-05, + "loss": 0.3825, + "step": 3448 + }, + { + "epoch": 1.44, + "learning_rate": 1.8835394272061368e-05, + "loss": 0.3355, + "step": 3449 + }, + { + "epoch": 1.44, + "learning_rate": 1.88346000888258e-05, + "loss": 0.4143, + "step": 3450 + }, + { + "epoch": 1.44, + "learning_rate": 1.8833805651647498e-05, + "loss": 0.428, + "step": 3451 + }, + { + "epoch": 1.45, + "learning_rate": 1.8833010960549286e-05, + "loss": 0.3992, + "step": 3452 + }, + { + "epoch": 1.45, + "learning_rate": 1.883221601555401e-05, + "loss": 0.3523, + "step": 3453 + }, + { + "epoch": 1.45, + "learning_rate": 1.8831420816684528e-05, + "loss": 0.3957, + "step": 3454 + }, + { + "epoch": 1.45, + "learning_rate": 1.8830625363963685e-05, + "loss": 0.366, + "step": 3455 + }, + { + "epoch": 1.45, + "learning_rate": 1.882982965741436e-05, + "loss": 0.3754, + "step": 3456 + }, + { + "epoch": 1.45, + "learning_rate": 1.882903369705941e-05, + "loss": 0.4156, + "step": 3457 + }, + { + "epoch": 1.45, + "learning_rate": 1.8828237482921725e-05, + "loss": 0.4268, + "step": 3458 + }, + { + "epoch": 1.45, + "learning_rate": 1.8827441015024187e-05, + "loss": 0.3932, + "step": 3459 + }, + { + "epoch": 1.45, + "learning_rate": 1.882664429338969e-05, + "loss": 0.4338, + "step": 3460 + }, + { + "epoch": 1.45, + "learning_rate": 1.8825847318041136e-05, + "loss": 0.4078, + "step": 3461 + }, + { + "epoch": 1.45, + "learning_rate": 1.882505008900143e-05, + "loss": 0.4001, + "step": 3462 + }, + { + "epoch": 1.45, + "learning_rate": 1.8824252606293492e-05, + "loss": 0.3735, + "step": 3463 + }, + { + "epoch": 1.45, + "learning_rate": 1.8823454869940243e-05, + "loss": 0.3941, + "step": 3464 + }, + { + "epoch": 1.45, + "learning_rate": 1.8822656879964617e-05, + "loss": 0.3981, + "step": 3465 + }, + { + "epoch": 1.45, + "learning_rate": 1.8821858636389547e-05, + "loss": 0.4284, + "step": 3466 + }, + { + "epoch": 1.45, + "learning_rate": 1.8821060139237975e-05, + "loss": 0.3802, + "step": 3467 + }, + { + "epoch": 1.45, + "learning_rate": 1.8820261388532858e-05, + "loss": 0.3674, + "step": 3468 + }, + { + "epoch": 1.45, + "learning_rate": 1.8819462384297155e-05, + "loss": 0.3818, + "step": 3469 + }, + { + "epoch": 1.45, + "learning_rate": 1.8818663126553833e-05, + "loss": 0.4197, + "step": 3470 + }, + { + "epoch": 1.45, + "learning_rate": 1.8817863615325865e-05, + "loss": 0.425, + "step": 3471 + }, + { + "epoch": 1.45, + "learning_rate": 1.881706385063623e-05, + "loss": 0.3757, + "step": 3472 + }, + { + "epoch": 1.45, + "learning_rate": 1.881626383250792e-05, + "loss": 0.3716, + "step": 3473 + }, + { + "epoch": 1.45, + "learning_rate": 1.8815463560963932e-05, + "loss": 0.392, + "step": 3474 + }, + { + "epoch": 1.45, + "learning_rate": 1.8814663036027262e-05, + "loss": 0.3917, + "step": 3475 + }, + { + "epoch": 1.46, + "learning_rate": 1.881386225772093e-05, + "loss": 0.356, + "step": 3476 + }, + { + "epoch": 1.46, + "learning_rate": 1.8813061226067945e-05, + "loss": 0.3455, + "step": 3477 + }, + { + "epoch": 1.46, + "learning_rate": 1.881225994109134e-05, + "loss": 0.4187, + "step": 3478 + }, + { + "epoch": 1.46, + "learning_rate": 1.881145840281414e-05, + "loss": 0.3953, + "step": 3479 + }, + { + "epoch": 1.46, + "learning_rate": 1.8810656611259387e-05, + "loss": 0.3728, + "step": 3480 + }, + { + "epoch": 1.46, + "learning_rate": 1.8809854566450128e-05, + "loss": 0.3603, + "step": 3481 + }, + { + "epoch": 1.46, + "learning_rate": 1.8809052268409414e-05, + "loss": 0.3356, + "step": 3482 + }, + { + "epoch": 1.46, + "learning_rate": 1.8808249717160318e-05, + "loss": 0.4093, + "step": 3483 + }, + { + "epoch": 1.46, + "learning_rate": 1.8807446912725893e-05, + "loss": 0.4172, + "step": 3484 + }, + { + "epoch": 1.46, + "learning_rate": 1.8806643855129225e-05, + "loss": 0.3647, + "step": 3485 + }, + { + "epoch": 1.46, + "learning_rate": 1.8805840544393393e-05, + "loss": 0.4281, + "step": 3486 + }, + { + "epoch": 1.46, + "learning_rate": 1.880503698054149e-05, + "loss": 0.3747, + "step": 3487 + }, + { + "epoch": 1.46, + "learning_rate": 1.8804233163596613e-05, + "loss": 0.3777, + "step": 3488 + }, + { + "epoch": 1.46, + "learning_rate": 1.8803429093581863e-05, + "loss": 0.3581, + "step": 3489 + }, + { + "epoch": 1.46, + "learning_rate": 1.880262477052036e-05, + "loss": 0.3936, + "step": 3490 + }, + { + "epoch": 1.46, + "learning_rate": 1.8801820194435215e-05, + "loss": 0.3912, + "step": 3491 + }, + { + "epoch": 1.46, + "learning_rate": 1.8801015365349562e-05, + "loss": 0.3921, + "step": 3492 + }, + { + "epoch": 1.46, + "learning_rate": 1.880021028328653e-05, + "loss": 0.358, + "step": 3493 + }, + { + "epoch": 1.46, + "learning_rate": 1.8799404948269267e-05, + "loss": 0.3555, + "step": 3494 + }, + { + "epoch": 1.46, + "learning_rate": 1.879859936032091e-05, + "loss": 0.3767, + "step": 3495 + }, + { + "epoch": 1.46, + "learning_rate": 1.8797793519464628e-05, + "loss": 0.3615, + "step": 3496 + }, + { + "epoch": 1.46, + "learning_rate": 1.8796987425723578e-05, + "loss": 0.3714, + "step": 3497 + }, + { + "epoch": 1.46, + "learning_rate": 1.879618107912093e-05, + "loss": 0.3775, + "step": 3498 + }, + { + "epoch": 1.46, + "learning_rate": 1.8795374479679862e-05, + "loss": 0.4144, + "step": 3499 + }, + { + "epoch": 1.47, + "learning_rate": 1.879456762742356e-05, + "loss": 0.3544, + "step": 3500 + }, + { + "epoch": 1.47, + "learning_rate": 1.8793760522375214e-05, + "loss": 0.3959, + "step": 3501 + }, + { + "epoch": 1.47, + "learning_rate": 1.8792953164558026e-05, + "loss": 0.4117, + "step": 3502 + }, + { + "epoch": 1.47, + "learning_rate": 1.8792145553995203e-05, + "loss": 0.3799, + "step": 3503 + }, + { + "epoch": 1.47, + "learning_rate": 1.879133769070996e-05, + "loss": 0.4163, + "step": 3504 + }, + { + "epoch": 1.47, + "learning_rate": 1.8790529574725514e-05, + "loss": 0.4168, + "step": 3505 + }, + { + "epoch": 1.47, + "learning_rate": 1.8789721206065096e-05, + "loss": 0.3888, + "step": 3506 + }, + { + "epoch": 1.47, + "learning_rate": 1.8788912584751938e-05, + "loss": 0.3918, + "step": 3507 + }, + { + "epoch": 1.47, + "learning_rate": 1.8788103710809292e-05, + "loss": 0.3519, + "step": 3508 + }, + { + "epoch": 1.47, + "learning_rate": 1.8787294584260402e-05, + "loss": 0.377, + "step": 3509 + }, + { + "epoch": 1.47, + "learning_rate": 1.8786485205128526e-05, + "loss": 0.3672, + "step": 3510 + }, + { + "epoch": 1.47, + "learning_rate": 1.878567557343693e-05, + "loss": 0.4139, + "step": 3511 + }, + { + "epoch": 1.47, + "learning_rate": 1.8784865689208888e-05, + "loss": 0.3682, + "step": 3512 + }, + { + "epoch": 1.47, + "learning_rate": 1.8784055552467674e-05, + "loss": 0.4095, + "step": 3513 + }, + { + "epoch": 1.47, + "learning_rate": 1.878324516323658e-05, + "loss": 0.4295, + "step": 3514 + }, + { + "epoch": 1.47, + "learning_rate": 1.87824345215389e-05, + "loss": 0.4504, + "step": 3515 + }, + { + "epoch": 1.47, + "learning_rate": 1.878162362739793e-05, + "loss": 0.3981, + "step": 3516 + }, + { + "epoch": 1.47, + "learning_rate": 1.878081248083698e-05, + "loss": 0.4034, + "step": 3517 + }, + { + "epoch": 1.47, + "learning_rate": 1.878000108187937e-05, + "loss": 0.3617, + "step": 3518 + }, + { + "epoch": 1.47, + "learning_rate": 1.877918943054842e-05, + "loss": 0.3635, + "step": 3519 + }, + { + "epoch": 1.47, + "learning_rate": 1.877837752686746e-05, + "loss": 0.4073, + "step": 3520 + }, + { + "epoch": 1.47, + "learning_rate": 1.8777565370859826e-05, + "loss": 0.3933, + "step": 3521 + }, + { + "epoch": 1.47, + "learning_rate": 1.8776752962548866e-05, + "loss": 0.3756, + "step": 3522 + }, + { + "epoch": 1.47, + "learning_rate": 1.8775940301957932e-05, + "loss": 0.3968, + "step": 3523 + }, + { + "epoch": 1.48, + "learning_rate": 1.877512738911038e-05, + "loss": 0.3968, + "step": 3524 + }, + { + "epoch": 1.48, + "learning_rate": 1.8774314224029578e-05, + "loss": 0.4166, + "step": 3525 + }, + { + "epoch": 1.48, + "learning_rate": 1.87735008067389e-05, + "loss": 0.3903, + "step": 3526 + }, + { + "epoch": 1.48, + "learning_rate": 1.8772687137261728e-05, + "loss": 0.3604, + "step": 3527 + }, + { + "epoch": 1.48, + "learning_rate": 1.8771873215621447e-05, + "loss": 0.3919, + "step": 3528 + }, + { + "epoch": 1.48, + "learning_rate": 1.8771059041841454e-05, + "loss": 0.4071, + "step": 3529 + }, + { + "epoch": 1.48, + "learning_rate": 1.8770244615945153e-05, + "loss": 0.4072, + "step": 3530 + }, + { + "epoch": 1.48, + "learning_rate": 1.8769429937955954e-05, + "loss": 0.3793, + "step": 3531 + }, + { + "epoch": 1.48, + "learning_rate": 1.8768615007897274e-05, + "loss": 0.3788, + "step": 3532 + }, + { + "epoch": 1.48, + "learning_rate": 1.8767799825792533e-05, + "loss": 0.3738, + "step": 3533 + }, + { + "epoch": 1.48, + "learning_rate": 1.8766984391665173e-05, + "loss": 0.3879, + "step": 3534 + }, + { + "epoch": 1.48, + "learning_rate": 1.876616870553862e-05, + "loss": 0.3786, + "step": 3535 + }, + { + "epoch": 1.48, + "learning_rate": 1.876535276743633e-05, + "loss": 0.4138, + "step": 3536 + }, + { + "epoch": 1.48, + "learning_rate": 1.8764536577381748e-05, + "loss": 0.4008, + "step": 3537 + }, + { + "epoch": 1.48, + "learning_rate": 1.8763720135398348e-05, + "loss": 0.3867, + "step": 3538 + }, + { + "epoch": 1.48, + "learning_rate": 1.8762903441509582e-05, + "loss": 0.3631, + "step": 3539 + }, + { + "epoch": 1.48, + "learning_rate": 1.8762086495738937e-05, + "loss": 0.35, + "step": 3540 + }, + { + "epoch": 1.48, + "learning_rate": 1.876126929810989e-05, + "loss": 0.3891, + "step": 3541 + }, + { + "epoch": 1.48, + "learning_rate": 1.8760451848645932e-05, + "loss": 0.3692, + "step": 3542 + }, + { + "epoch": 1.48, + "learning_rate": 1.8759634147370555e-05, + "loss": 0.4024, + "step": 3543 + }, + { + "epoch": 1.48, + "learning_rate": 1.8758816194307272e-05, + "loss": 0.3913, + "step": 3544 + }, + { + "epoch": 1.48, + "learning_rate": 1.875799798947959e-05, + "loss": 0.3868, + "step": 3545 + }, + { + "epoch": 1.48, + "learning_rate": 1.8757179532911026e-05, + "loss": 0.3665, + "step": 3546 + }, + { + "epoch": 1.49, + "learning_rate": 1.8756360824625107e-05, + "loss": 0.4118, + "step": 3547 + }, + { + "epoch": 1.49, + "learning_rate": 1.875554186464537e-05, + "loss": 0.3749, + "step": 3548 + }, + { + "epoch": 1.49, + "learning_rate": 1.8754722652995346e-05, + "loss": 0.4166, + "step": 3549 + }, + { + "epoch": 1.49, + "learning_rate": 1.8753903189698594e-05, + "loss": 0.3962, + "step": 3550 + }, + { + "epoch": 1.49, + "learning_rate": 1.875308347477866e-05, + "loss": 0.4516, + "step": 3551 + }, + { + "epoch": 1.49, + "learning_rate": 1.8752263508259107e-05, + "loss": 0.4046, + "step": 3552 + }, + { + "epoch": 1.49, + "learning_rate": 1.875144329016351e-05, + "loss": 0.4097, + "step": 3553 + }, + { + "epoch": 1.49, + "learning_rate": 1.875062282051544e-05, + "loss": 0.4241, + "step": 3554 + }, + { + "epoch": 1.49, + "learning_rate": 1.874980209933848e-05, + "loss": 0.38, + "step": 3555 + }, + { + "epoch": 1.49, + "learning_rate": 1.8748981126656224e-05, + "loss": 0.3898, + "step": 3556 + }, + { + "epoch": 1.49, + "learning_rate": 1.8748159902492273e-05, + "loss": 0.4338, + "step": 3557 + }, + { + "epoch": 1.49, + "learning_rate": 1.8747338426870222e-05, + "loss": 0.4228, + "step": 3558 + }, + { + "epoch": 1.49, + "learning_rate": 1.8746516699813693e-05, + "loss": 0.35, + "step": 3559 + }, + { + "epoch": 1.49, + "learning_rate": 1.8745694721346305e-05, + "loss": 0.3797, + "step": 3560 + }, + { + "epoch": 1.49, + "learning_rate": 1.874487249149168e-05, + "loss": 0.4204, + "step": 3561 + }, + { + "epoch": 1.49, + "learning_rate": 1.8744050010273458e-05, + "loss": 0.4093, + "step": 3562 + }, + { + "epoch": 1.49, + "learning_rate": 1.874322727771528e-05, + "loss": 0.4614, + "step": 3563 + }, + { + "epoch": 1.49, + "learning_rate": 1.8742404293840784e-05, + "loss": 0.3874, + "step": 3564 + }, + { + "epoch": 1.49, + "learning_rate": 1.8741581058673642e-05, + "loss": 0.3658, + "step": 3565 + }, + { + "epoch": 1.49, + "learning_rate": 1.874075757223751e-05, + "loss": 0.4201, + "step": 3566 + }, + { + "epoch": 1.49, + "learning_rate": 1.8739933834556054e-05, + "loss": 0.3875, + "step": 3567 + }, + { + "epoch": 1.49, + "learning_rate": 1.8739109845652957e-05, + "loss": 0.4209, + "step": 3568 + }, + { + "epoch": 1.49, + "learning_rate": 1.87382856055519e-05, + "loss": 0.347, + "step": 3569 + }, + { + "epoch": 1.49, + "learning_rate": 1.873746111427658e-05, + "loss": 0.4023, + "step": 3570 + }, + { + "epoch": 1.5, + "learning_rate": 1.8736636371850694e-05, + "loss": 0.3956, + "step": 3571 + }, + { + "epoch": 1.5, + "learning_rate": 1.8735811378297946e-05, + "loss": 0.406, + "step": 3572 + }, + { + "epoch": 1.5, + "learning_rate": 1.8734986133642058e-05, + "loss": 0.3692, + "step": 3573 + }, + { + "epoch": 1.5, + "learning_rate": 1.8734160637906738e-05, + "loss": 0.3746, + "step": 3574 + }, + { + "epoch": 1.5, + "learning_rate": 1.8733334891115725e-05, + "loss": 0.3654, + "step": 3575 + }, + { + "epoch": 1.5, + "learning_rate": 1.873250889329275e-05, + "loss": 0.3501, + "step": 3576 + }, + { + "epoch": 1.5, + "learning_rate": 1.8731682644461555e-05, + "loss": 0.4146, + "step": 3577 + }, + { + "epoch": 1.5, + "learning_rate": 1.873085614464589e-05, + "loss": 0.3821, + "step": 3578 + }, + { + "epoch": 1.5, + "learning_rate": 1.8730029393869512e-05, + "loss": 0.3517, + "step": 3579 + }, + { + "epoch": 1.5, + "learning_rate": 1.8729202392156186e-05, + "loss": 0.4147, + "step": 3580 + }, + { + "epoch": 1.5, + "learning_rate": 1.8728375139529688e-05, + "loss": 0.4, + "step": 3581 + }, + { + "epoch": 1.5, + "learning_rate": 1.872754763601379e-05, + "loss": 0.3598, + "step": 3582 + }, + { + "epoch": 1.5, + "learning_rate": 1.8726719881632278e-05, + "loss": 0.405, + "step": 3583 + }, + { + "epoch": 1.5, + "learning_rate": 1.8725891876408946e-05, + "loss": 0.4087, + "step": 3584 + }, + { + "epoch": 1.5, + "learning_rate": 1.8725063620367598e-05, + "loss": 0.4424, + "step": 3585 + }, + { + "epoch": 1.5, + "learning_rate": 1.872423511353204e-05, + "loss": 0.3516, + "step": 3586 + }, + { + "epoch": 1.5, + "learning_rate": 1.8723406355926078e-05, + "loss": 0.4001, + "step": 3587 + }, + { + "epoch": 1.5, + "learning_rate": 1.872257734757355e-05, + "loss": 0.3657, + "step": 3588 + }, + { + "epoch": 1.5, + "learning_rate": 1.8721748088498272e-05, + "loss": 0.3784, + "step": 3589 + }, + { + "epoch": 1.5, + "learning_rate": 1.8720918578724085e-05, + "loss": 0.4014, + "step": 3590 + }, + { + "epoch": 1.5, + "learning_rate": 1.8720088818274835e-05, + "loss": 0.4215, + "step": 3591 + }, + { + "epoch": 1.5, + "learning_rate": 1.8719258807174366e-05, + "loss": 0.3759, + "step": 3592 + }, + { + "epoch": 1.5, + "learning_rate": 1.8718428545446545e-05, + "loss": 0.3367, + "step": 3593 + }, + { + "epoch": 1.5, + "learning_rate": 1.8717598033115224e-05, + "loss": 0.3609, + "step": 3594 + }, + { + "epoch": 1.51, + "learning_rate": 1.871676727020429e-05, + "loss": 0.3814, + "step": 3595 + }, + { + "epoch": 1.51, + "learning_rate": 1.871593625673761e-05, + "loss": 0.4125, + "step": 3596 + }, + { + "epoch": 1.51, + "learning_rate": 1.8715104992739085e-05, + "loss": 0.3841, + "step": 3597 + }, + { + "epoch": 1.51, + "learning_rate": 1.8714273478232597e-05, + "loss": 0.3855, + "step": 3598 + }, + { + "epoch": 1.51, + "learning_rate": 1.871344171324205e-05, + "loss": 0.4007, + "step": 3599 + }, + { + "epoch": 1.51, + "learning_rate": 1.871260969779135e-05, + "loss": 0.3679, + "step": 3600 + }, + { + "epoch": 1.51, + "learning_rate": 1.871177743190442e-05, + "loss": 0.3722, + "step": 3601 + }, + { + "epoch": 1.51, + "learning_rate": 1.871094491560518e-05, + "loss": 0.3753, + "step": 3602 + }, + { + "epoch": 1.51, + "learning_rate": 1.8710112148917555e-05, + "loss": 0.4449, + "step": 3603 + }, + { + "epoch": 1.51, + "learning_rate": 1.8709279131865486e-05, + "loss": 0.4068, + "step": 3604 + }, + { + "epoch": 1.51, + "learning_rate": 1.8708445864472917e-05, + "loss": 0.3895, + "step": 3605 + }, + { + "epoch": 1.51, + "learning_rate": 1.87076123467638e-05, + "loss": 0.3729, + "step": 3606 + }, + { + "epoch": 1.51, + "learning_rate": 1.8706778578762094e-05, + "loss": 0.3501, + "step": 3607 + }, + { + "epoch": 1.51, + "learning_rate": 1.870594456049176e-05, + "loss": 0.337, + "step": 3608 + }, + { + "epoch": 1.51, + "learning_rate": 1.870511029197678e-05, + "loss": 0.3914, + "step": 3609 + }, + { + "epoch": 1.51, + "learning_rate": 1.8704275773241126e-05, + "loss": 0.3952, + "step": 3610 + }, + { + "epoch": 1.51, + "learning_rate": 1.8703441004308792e-05, + "loss": 0.3786, + "step": 3611 + }, + { + "epoch": 1.51, + "learning_rate": 1.8702605985203767e-05, + "loss": 0.38, + "step": 3612 + }, + { + "epoch": 1.51, + "learning_rate": 1.8701770715950058e-05, + "loss": 0.4337, + "step": 3613 + }, + { + "epoch": 1.51, + "learning_rate": 1.8700935196571666e-05, + "loss": 0.3772, + "step": 3614 + }, + { + "epoch": 1.51, + "learning_rate": 1.8700099427092618e-05, + "loss": 0.3557, + "step": 3615 + }, + { + "epoch": 1.51, + "learning_rate": 1.869926340753693e-05, + "loss": 0.363, + "step": 3616 + }, + { + "epoch": 1.51, + "learning_rate": 1.8698427137928637e-05, + "loss": 0.4203, + "step": 3617 + }, + { + "epoch": 1.51, + "learning_rate": 1.8697590618291772e-05, + "loss": 0.4204, + "step": 3618 + }, + { + "epoch": 1.52, + "learning_rate": 1.8696753848650386e-05, + "loss": 0.3818, + "step": 3619 + }, + { + "epoch": 1.52, + "learning_rate": 1.8695916829028525e-05, + "loss": 0.4434, + "step": 3620 + }, + { + "epoch": 1.52, + "learning_rate": 1.8695079559450255e-05, + "loss": 0.4064, + "step": 3621 + }, + { + "epoch": 1.52, + "learning_rate": 1.8694242039939636e-05, + "loss": 0.3619, + "step": 3622 + }, + { + "epoch": 1.52, + "learning_rate": 1.8693404270520745e-05, + "loss": 0.3922, + "step": 3623 + }, + { + "epoch": 1.52, + "learning_rate": 1.8692566251217665e-05, + "loss": 0.3681, + "step": 3624 + }, + { + "epoch": 1.52, + "learning_rate": 1.8691727982054478e-05, + "loss": 0.3653, + "step": 3625 + }, + { + "epoch": 1.52, + "learning_rate": 1.8690889463055285e-05, + "loss": 0.3871, + "step": 3626 + }, + { + "epoch": 1.52, + "learning_rate": 1.8690050694244188e-05, + "loss": 0.3941, + "step": 3627 + }, + { + "epoch": 1.52, + "learning_rate": 1.868921167564529e-05, + "loss": 0.3549, + "step": 3628 + }, + { + "epoch": 1.52, + "learning_rate": 1.8688372407282717e-05, + "loss": 0.3583, + "step": 3629 + }, + { + "epoch": 1.52, + "learning_rate": 1.868753288918059e-05, + "loss": 0.3913, + "step": 3630 + }, + { + "epoch": 1.52, + "learning_rate": 1.868669312136304e-05, + "loss": 0.3775, + "step": 3631 + }, + { + "epoch": 1.52, + "learning_rate": 1.86858531038542e-05, + "loss": 0.3765, + "step": 3632 + }, + { + "epoch": 1.52, + "learning_rate": 1.8685012836678225e-05, + "loss": 0.3596, + "step": 3633 + }, + { + "epoch": 1.52, + "learning_rate": 1.8684172319859258e-05, + "loss": 0.3955, + "step": 3634 + }, + { + "epoch": 1.52, + "learning_rate": 1.8683331553421467e-05, + "loss": 0.468, + "step": 3635 + }, + { + "epoch": 1.52, + "learning_rate": 1.8682490537389017e-05, + "loss": 0.4153, + "step": 3636 + }, + { + "epoch": 1.52, + "learning_rate": 1.868164927178608e-05, + "loss": 0.3604, + "step": 3637 + }, + { + "epoch": 1.52, + "learning_rate": 1.8680807756636837e-05, + "loss": 0.3752, + "step": 3638 + }, + { + "epoch": 1.52, + "learning_rate": 1.8679965991965484e-05, + "loss": 0.3992, + "step": 3639 + }, + { + "epoch": 1.52, + "learning_rate": 1.8679123977796205e-05, + "loss": 0.3463, + "step": 3640 + }, + { + "epoch": 1.52, + "learning_rate": 1.867828171415321e-05, + "loss": 0.412, + "step": 3641 + }, + { + "epoch": 1.52, + "learning_rate": 1.867743920106071e-05, + "loss": 0.3994, + "step": 3642 + }, + { + "epoch": 1.53, + "learning_rate": 1.867659643854292e-05, + "loss": 0.4042, + "step": 3643 + }, + { + "epoch": 1.53, + "learning_rate": 1.867575342662406e-05, + "loss": 0.4215, + "step": 3644 + }, + { + "epoch": 1.53, + "learning_rate": 1.8674910165328374e-05, + "loss": 0.423, + "step": 3645 + }, + { + "epoch": 1.53, + "learning_rate": 1.8674066654680088e-05, + "loss": 0.3863, + "step": 3646 + }, + { + "epoch": 1.53, + "learning_rate": 1.8673222894703456e-05, + "loss": 0.419, + "step": 3647 + }, + { + "epoch": 1.53, + "learning_rate": 1.8672378885422728e-05, + "loss": 0.4253, + "step": 3648 + }, + { + "epoch": 1.53, + "learning_rate": 1.867153462686216e-05, + "loss": 0.3741, + "step": 3649 + }, + { + "epoch": 1.53, + "learning_rate": 1.8670690119046032e-05, + "loss": 0.3937, + "step": 3650 + }, + { + "epoch": 1.53, + "learning_rate": 1.8669845361998607e-05, + "loss": 0.3863, + "step": 3651 + }, + { + "epoch": 1.53, + "learning_rate": 1.8669000355744172e-05, + "loss": 0.4042, + "step": 3652 + }, + { + "epoch": 1.53, + "learning_rate": 1.8668155100307013e-05, + "loss": 0.4318, + "step": 3653 + }, + { + "epoch": 1.53, + "learning_rate": 1.8667309595711426e-05, + "loss": 0.4049, + "step": 3654 + }, + { + "epoch": 1.53, + "learning_rate": 1.8666463841981717e-05, + "loss": 0.4273, + "step": 3655 + }, + { + "epoch": 1.53, + "learning_rate": 1.8665617839142198e-05, + "loss": 0.3994, + "step": 3656 + }, + { + "epoch": 1.53, + "learning_rate": 1.866477158721718e-05, + "loss": 0.3565, + "step": 3657 + }, + { + "epoch": 1.53, + "learning_rate": 1.8663925086230998e-05, + "loss": 0.3925, + "step": 3658 + }, + { + "epoch": 1.53, + "learning_rate": 1.866307833620797e-05, + "loss": 0.446, + "step": 3659 + }, + { + "epoch": 1.53, + "learning_rate": 1.8662231337172446e-05, + "loss": 0.388, + "step": 3660 + }, + { + "epoch": 1.53, + "learning_rate": 1.866138408914877e-05, + "loss": 0.3558, + "step": 3661 + }, + { + "epoch": 1.53, + "learning_rate": 1.8660536592161293e-05, + "loss": 0.3339, + "step": 3662 + }, + { + "epoch": 1.53, + "learning_rate": 1.8659688846234376e-05, + "loss": 0.357, + "step": 3663 + }, + { + "epoch": 1.53, + "learning_rate": 1.8658840851392386e-05, + "loss": 0.4077, + "step": 3664 + }, + { + "epoch": 1.53, + "learning_rate": 1.8657992607659706e-05, + "loss": 0.3655, + "step": 3665 + }, + { + "epoch": 1.53, + "learning_rate": 1.8657144115060707e-05, + "loss": 0.3917, + "step": 3666 + }, + { + "epoch": 1.54, + "learning_rate": 1.8656295373619785e-05, + "loss": 0.4059, + "step": 3667 + }, + { + "epoch": 1.54, + "learning_rate": 1.8655446383361332e-05, + "loss": 0.4232, + "step": 3668 + }, + { + "epoch": 1.54, + "learning_rate": 1.8654597144309753e-05, + "loss": 0.3869, + "step": 3669 + }, + { + "epoch": 1.54, + "learning_rate": 1.865374765648946e-05, + "loss": 0.3615, + "step": 3670 + }, + { + "epoch": 1.54, + "learning_rate": 1.865289791992487e-05, + "loss": 0.368, + "step": 3671 + }, + { + "epoch": 1.54, + "learning_rate": 1.8652047934640405e-05, + "loss": 0.3497, + "step": 3672 + }, + { + "epoch": 1.54, + "learning_rate": 1.86511977006605e-05, + "loss": 0.4531, + "step": 3673 + }, + { + "epoch": 1.54, + "learning_rate": 1.8650347218009596e-05, + "loss": 0.3943, + "step": 3674 + }, + { + "epoch": 1.54, + "learning_rate": 1.864949648671214e-05, + "loss": 0.3597, + "step": 3675 + }, + { + "epoch": 1.54, + "learning_rate": 1.8648645506792578e-05, + "loss": 0.396, + "step": 3676 + }, + { + "epoch": 1.54, + "learning_rate": 1.864779427827538e-05, + "loss": 0.3928, + "step": 3677 + }, + { + "epoch": 1.54, + "learning_rate": 1.8646942801185006e-05, + "loss": 0.3414, + "step": 3678 + }, + { + "epoch": 1.54, + "learning_rate": 1.8646091075545936e-05, + "loss": 0.3588, + "step": 3679 + }, + { + "epoch": 1.54, + "learning_rate": 1.864523910138265e-05, + "loss": 0.4464, + "step": 3680 + }, + { + "epoch": 1.54, + "learning_rate": 1.864438687871964e-05, + "loss": 0.3666, + "step": 3681 + }, + { + "epoch": 1.54, + "learning_rate": 1.8643534407581397e-05, + "loss": 0.4229, + "step": 3682 + }, + { + "epoch": 1.54, + "learning_rate": 1.8642681687992427e-05, + "loss": 0.3988, + "step": 3683 + }, + { + "epoch": 1.54, + "learning_rate": 1.8641828719977244e-05, + "loss": 0.3685, + "step": 3684 + }, + { + "epoch": 1.54, + "learning_rate": 1.864097550356036e-05, + "loss": 0.3398, + "step": 3685 + }, + { + "epoch": 1.54, + "learning_rate": 1.8640122038766307e-05, + "loss": 0.4044, + "step": 3686 + }, + { + "epoch": 1.54, + "learning_rate": 1.863926832561961e-05, + "loss": 0.3999, + "step": 3687 + }, + { + "epoch": 1.54, + "learning_rate": 1.8638414364144813e-05, + "loss": 0.3707, + "step": 3688 + }, + { + "epoch": 1.54, + "learning_rate": 1.863756015436646e-05, + "loss": 0.3947, + "step": 3689 + }, + { + "epoch": 1.54, + "learning_rate": 1.8636705696309104e-05, + "loss": 0.4347, + "step": 3690 + }, + { + "epoch": 1.55, + "learning_rate": 1.8635850989997308e-05, + "loss": 0.3899, + "step": 3691 + }, + { + "epoch": 1.55, + "learning_rate": 1.863499603545564e-05, + "loss": 0.362, + "step": 3692 + }, + { + "epoch": 1.55, + "learning_rate": 1.8634140832708673e-05, + "loss": 0.4191, + "step": 3693 + }, + { + "epoch": 1.55, + "learning_rate": 1.8633285381780987e-05, + "loss": 0.4036, + "step": 3694 + }, + { + "epoch": 1.55, + "learning_rate": 1.8632429682697175e-05, + "loss": 0.4099, + "step": 3695 + }, + { + "epoch": 1.55, + "learning_rate": 1.8631573735481834e-05, + "loss": 0.43, + "step": 3696 + }, + { + "epoch": 1.55, + "learning_rate": 1.8630717540159562e-05, + "loss": 0.3987, + "step": 3697 + }, + { + "epoch": 1.55, + "learning_rate": 1.8629861096754977e-05, + "loss": 0.3406, + "step": 3698 + }, + { + "epoch": 1.55, + "learning_rate": 1.862900440529269e-05, + "loss": 0.3903, + "step": 3699 + }, + { + "epoch": 1.55, + "learning_rate": 1.862814746579733e-05, + "loss": 0.3315, + "step": 3700 + }, + { + "epoch": 1.55, + "learning_rate": 1.8627290278293525e-05, + "loss": 0.3851, + "step": 3701 + }, + { + "epoch": 1.55, + "learning_rate": 1.862643284280592e-05, + "loss": 0.3995, + "step": 3702 + }, + { + "epoch": 1.55, + "learning_rate": 1.8625575159359157e-05, + "loss": 0.393, + "step": 3703 + }, + { + "epoch": 1.55, + "learning_rate": 1.862471722797789e-05, + "loss": 0.405, + "step": 3704 + }, + { + "epoch": 1.55, + "learning_rate": 1.8623859048686777e-05, + "loss": 0.3704, + "step": 3705 + }, + { + "epoch": 1.55, + "learning_rate": 1.862300062151049e-05, + "loss": 0.3837, + "step": 3706 + }, + { + "epoch": 1.55, + "learning_rate": 1.8622141946473705e-05, + "loss": 0.3839, + "step": 3707 + }, + { + "epoch": 1.55, + "learning_rate": 1.8621283023601097e-05, + "loss": 0.4162, + "step": 3708 + }, + { + "epoch": 1.55, + "learning_rate": 1.862042385291736e-05, + "loss": 0.4516, + "step": 3709 + }, + { + "epoch": 1.55, + "learning_rate": 1.8619564434447192e-05, + "loss": 0.3776, + "step": 3710 + }, + { + "epoch": 1.55, + "learning_rate": 1.861870476821529e-05, + "loss": 0.4294, + "step": 3711 + }, + { + "epoch": 1.55, + "learning_rate": 1.8617844854246368e-05, + "loss": 0.3658, + "step": 3712 + }, + { + "epoch": 1.55, + "learning_rate": 1.8616984692565144e-05, + "loss": 0.3766, + "step": 3713 + }, + { + "epoch": 1.55, + "learning_rate": 1.861612428319634e-05, + "loss": 0.3958, + "step": 3714 + }, + { + "epoch": 1.56, + "learning_rate": 1.861526362616469e-05, + "loss": 0.4002, + "step": 3715 + }, + { + "epoch": 1.56, + "learning_rate": 1.8614402721494934e-05, + "loss": 0.3409, + "step": 3716 + }, + { + "epoch": 1.56, + "learning_rate": 1.8613541569211815e-05, + "loss": 0.3693, + "step": 3717 + }, + { + "epoch": 1.56, + "learning_rate": 1.8612680169340085e-05, + "loss": 0.3764, + "step": 3718 + }, + { + "epoch": 1.56, + "learning_rate": 1.861181852190451e-05, + "loss": 0.3525, + "step": 3719 + }, + { + "epoch": 1.56, + "learning_rate": 1.8610956626929852e-05, + "loss": 0.3591, + "step": 3720 + }, + { + "epoch": 1.56, + "learning_rate": 1.8610094484440884e-05, + "loss": 0.3761, + "step": 3721 + }, + { + "epoch": 1.56, + "learning_rate": 1.8609232094462394e-05, + "loss": 0.3549, + "step": 3722 + }, + { + "epoch": 1.56, + "learning_rate": 1.8608369457019166e-05, + "loss": 0.3992, + "step": 3723 + }, + { + "epoch": 1.56, + "learning_rate": 1.8607506572135994e-05, + "loss": 0.3731, + "step": 3724 + }, + { + "epoch": 1.56, + "learning_rate": 1.860664343983769e-05, + "loss": 0.3797, + "step": 3725 + }, + { + "epoch": 1.56, + "learning_rate": 1.8605780060149056e-05, + "loss": 0.3526, + "step": 3726 + }, + { + "epoch": 1.56, + "learning_rate": 1.8604916433094908e-05, + "loss": 0.3517, + "step": 3727 + }, + { + "epoch": 1.56, + "learning_rate": 1.860405255870007e-05, + "loss": 0.3643, + "step": 3728 + }, + { + "epoch": 1.56, + "learning_rate": 1.8603188436989382e-05, + "loss": 0.3997, + "step": 3729 + }, + { + "epoch": 1.56, + "learning_rate": 1.8602324067987678e-05, + "loss": 0.3371, + "step": 3730 + }, + { + "epoch": 1.56, + "learning_rate": 1.86014594517198e-05, + "loss": 0.4331, + "step": 3731 + }, + { + "epoch": 1.56, + "learning_rate": 1.8600594588210602e-05, + "loss": 0.4084, + "step": 3732 + }, + { + "epoch": 1.56, + "learning_rate": 1.8599729477484944e-05, + "loss": 0.3512, + "step": 3733 + }, + { + "epoch": 1.56, + "learning_rate": 1.8598864119567693e-05, + "loss": 0.4022, + "step": 3734 + }, + { + "epoch": 1.56, + "learning_rate": 1.8597998514483724e-05, + "loss": 0.3846, + "step": 3735 + }, + { + "epoch": 1.56, + "learning_rate": 1.859713266225792e-05, + "loss": 0.3828, + "step": 3736 + }, + { + "epoch": 1.56, + "learning_rate": 1.8596266562915163e-05, + "loss": 0.4152, + "step": 3737 + }, + { + "epoch": 1.56, + "learning_rate": 1.8595400216480356e-05, + "loss": 0.347, + "step": 3738 + }, + { + "epoch": 1.57, + "learning_rate": 1.8594533622978394e-05, + "loss": 0.4058, + "step": 3739 + }, + { + "epoch": 1.57, + "learning_rate": 1.859366678243419e-05, + "loss": 0.4159, + "step": 3740 + }, + { + "epoch": 1.57, + "learning_rate": 1.859279969487266e-05, + "loss": 0.3883, + "step": 3741 + }, + { + "epoch": 1.57, + "learning_rate": 1.859193236031873e-05, + "loss": 0.3622, + "step": 3742 + }, + { + "epoch": 1.57, + "learning_rate": 1.859106477879733e-05, + "loss": 0.3714, + "step": 3743 + }, + { + "epoch": 1.57, + "learning_rate": 1.859019695033339e-05, + "loss": 0.3634, + "step": 3744 + }, + { + "epoch": 1.57, + "learning_rate": 1.8589328874951866e-05, + "loss": 0.442, + "step": 3745 + }, + { + "epoch": 1.57, + "learning_rate": 1.8588460552677703e-05, + "loss": 0.426, + "step": 3746 + }, + { + "epoch": 1.57, + "learning_rate": 1.8587591983535868e-05, + "loss": 0.3693, + "step": 3747 + }, + { + "epoch": 1.57, + "learning_rate": 1.8586723167551315e-05, + "loss": 0.4352, + "step": 3748 + }, + { + "epoch": 1.57, + "learning_rate": 1.858585410474903e-05, + "loss": 0.3763, + "step": 3749 + }, + { + "epoch": 1.57, + "learning_rate": 1.8584984795153984e-05, + "loss": 0.4033, + "step": 3750 + }, + { + "epoch": 1.57, + "learning_rate": 1.858411523879117e-05, + "loss": 0.4202, + "step": 3751 + }, + { + "epoch": 1.57, + "learning_rate": 1.858324543568558e-05, + "loss": 0.4004, + "step": 3752 + }, + { + "epoch": 1.57, + "learning_rate": 1.8582375385862215e-05, + "loss": 0.4286, + "step": 3753 + }, + { + "epoch": 1.57, + "learning_rate": 1.858150508934609e-05, + "loss": 0.3823, + "step": 3754 + }, + { + "epoch": 1.57, + "learning_rate": 1.8580634546162212e-05, + "loss": 0.3807, + "step": 3755 + }, + { + "epoch": 1.57, + "learning_rate": 1.857976375633561e-05, + "loss": 0.3462, + "step": 3756 + }, + { + "epoch": 1.57, + "learning_rate": 1.8578892719891314e-05, + "loss": 0.379, + "step": 3757 + }, + { + "epoch": 1.57, + "learning_rate": 1.8578021436854357e-05, + "loss": 0.3501, + "step": 3758 + }, + { + "epoch": 1.57, + "learning_rate": 1.8577149907249786e-05, + "loss": 0.384, + "step": 3759 + }, + { + "epoch": 1.57, + "learning_rate": 1.8576278131102655e-05, + "loss": 0.3863, + "step": 3760 + }, + { + "epoch": 1.57, + "learning_rate": 1.8575406108438017e-05, + "loss": 0.3424, + "step": 3761 + }, + { + "epoch": 1.58, + "learning_rate": 1.857453383928094e-05, + "loss": 0.395, + "step": 3762 + }, + { + "epoch": 1.58, + "learning_rate": 1.8573661323656495e-05, + "loss": 0.3597, + "step": 3763 + }, + { + "epoch": 1.58, + "learning_rate": 1.8572788561589766e-05, + "loss": 0.391, + "step": 3764 + }, + { + "epoch": 1.58, + "learning_rate": 1.857191555310584e-05, + "loss": 0.3818, + "step": 3765 + }, + { + "epoch": 1.58, + "learning_rate": 1.85710422982298e-05, + "loss": 0.4242, + "step": 3766 + }, + { + "epoch": 1.58, + "learning_rate": 1.8570168796986764e-05, + "loss": 0.42, + "step": 3767 + }, + { + "epoch": 1.58, + "learning_rate": 1.8569295049401826e-05, + "loss": 0.3615, + "step": 3768 + }, + { + "epoch": 1.58, + "learning_rate": 1.8568421055500107e-05, + "loss": 0.3709, + "step": 3769 + }, + { + "epoch": 1.58, + "learning_rate": 1.8567546815306726e-05, + "loss": 0.3988, + "step": 3770 + }, + { + "epoch": 1.58, + "learning_rate": 1.8566672328846816e-05, + "loss": 0.3838, + "step": 3771 + }, + { + "epoch": 1.58, + "learning_rate": 1.8565797596145515e-05, + "loss": 0.3679, + "step": 3772 + }, + { + "epoch": 1.58, + "learning_rate": 1.8564922617227958e-05, + "loss": 0.3765, + "step": 3773 + }, + { + "epoch": 1.58, + "learning_rate": 1.8564047392119306e-05, + "loss": 0.4416, + "step": 3774 + }, + { + "epoch": 1.58, + "learning_rate": 1.8563171920844706e-05, + "loss": 0.3616, + "step": 3775 + }, + { + "epoch": 1.58, + "learning_rate": 1.856229620342933e-05, + "loss": 0.3814, + "step": 3776 + }, + { + "epoch": 1.58, + "learning_rate": 1.8561420239898348e-05, + "loss": 0.3511, + "step": 3777 + }, + { + "epoch": 1.58, + "learning_rate": 1.856054403027694e-05, + "loss": 0.3562, + "step": 3778 + }, + { + "epoch": 1.58, + "learning_rate": 1.855966757459029e-05, + "loss": 0.3642, + "step": 3779 + }, + { + "epoch": 1.58, + "learning_rate": 1.855879087286359e-05, + "loss": 0.4004, + "step": 3780 + }, + { + "epoch": 1.58, + "learning_rate": 1.8557913925122042e-05, + "loss": 0.3681, + "step": 3781 + }, + { + "epoch": 1.58, + "learning_rate": 1.8557036731390853e-05, + "loss": 0.3826, + "step": 3782 + }, + { + "epoch": 1.58, + "learning_rate": 1.8556159291695235e-05, + "loss": 0.3754, + "step": 3783 + }, + { + "epoch": 1.58, + "learning_rate": 1.855528160606041e-05, + "loss": 0.3771, + "step": 3784 + }, + { + "epoch": 1.58, + "learning_rate": 1.855440367451161e-05, + "loss": 0.4514, + "step": 3785 + }, + { + "epoch": 1.59, + "learning_rate": 1.8553525497074065e-05, + "loss": 0.5203, + "step": 3786 + }, + { + "epoch": 1.59, + "learning_rate": 1.855264707377302e-05, + "loss": 0.4171, + "step": 3787 + }, + { + "epoch": 1.59, + "learning_rate": 1.8551768404633724e-05, + "loss": 0.3869, + "step": 3788 + }, + { + "epoch": 1.59, + "learning_rate": 1.8550889489681435e-05, + "loss": 0.4159, + "step": 3789 + }, + { + "epoch": 1.59, + "learning_rate": 1.8550010328941417e-05, + "loss": 0.3714, + "step": 3790 + }, + { + "epoch": 1.59, + "learning_rate": 1.8549130922438938e-05, + "loss": 0.375, + "step": 3791 + }, + { + "epoch": 1.59, + "learning_rate": 1.8548251270199274e-05, + "loss": 0.4059, + "step": 3792 + }, + { + "epoch": 1.59, + "learning_rate": 1.854737137224772e-05, + "loss": 0.4218, + "step": 3793 + }, + { + "epoch": 1.59, + "learning_rate": 1.8546491228609552e-05, + "loss": 0.3925, + "step": 3794 + }, + { + "epoch": 1.59, + "learning_rate": 1.8545610839310085e-05, + "loss": 0.3834, + "step": 3795 + }, + { + "epoch": 1.59, + "learning_rate": 1.854473020437461e-05, + "loss": 0.3778, + "step": 3796 + }, + { + "epoch": 1.59, + "learning_rate": 1.8543849323828452e-05, + "loss": 0.427, + "step": 3797 + }, + { + "epoch": 1.59, + "learning_rate": 1.854296819769693e-05, + "loss": 0.3654, + "step": 3798 + }, + { + "epoch": 1.59, + "learning_rate": 1.8542086826005366e-05, + "loss": 0.3874, + "step": 3799 + }, + { + "epoch": 1.59, + "learning_rate": 1.8541205208779095e-05, + "loss": 0.432, + "step": 3800 + }, + { + "epoch": 1.59, + "learning_rate": 1.8540323346043458e-05, + "loss": 0.3539, + "step": 3801 + }, + { + "epoch": 1.59, + "learning_rate": 1.853944123782381e-05, + "loss": 0.421, + "step": 3802 + }, + { + "epoch": 1.59, + "learning_rate": 1.8538558884145497e-05, + "loss": 0.3837, + "step": 3803 + }, + { + "epoch": 1.59, + "learning_rate": 1.8537676285033886e-05, + "loss": 0.3663, + "step": 3804 + }, + { + "epoch": 1.59, + "learning_rate": 1.853679344051435e-05, + "loss": 0.3943, + "step": 3805 + }, + { + "epoch": 1.59, + "learning_rate": 1.8535910350612257e-05, + "loss": 0.3913, + "step": 3806 + }, + { + "epoch": 1.59, + "learning_rate": 1.8535027015353e-05, + "loss": 0.3979, + "step": 3807 + }, + { + "epoch": 1.59, + "learning_rate": 1.8534143434761965e-05, + "loss": 0.4269, + "step": 3808 + }, + { + "epoch": 1.59, + "learning_rate": 1.8533259608864547e-05, + "loss": 0.3765, + "step": 3809 + }, + { + "epoch": 1.6, + "learning_rate": 1.8532375537686155e-05, + "loss": 0.3696, + "step": 3810 + }, + { + "epoch": 1.6, + "learning_rate": 1.85314912212522e-05, + "loss": 0.4306, + "step": 3811 + }, + { + "epoch": 1.6, + "learning_rate": 1.85306066595881e-05, + "loss": 0.3987, + "step": 3812 + }, + { + "epoch": 1.6, + "learning_rate": 1.8529721852719283e-05, + "loss": 0.4022, + "step": 3813 + }, + { + "epoch": 1.6, + "learning_rate": 1.852883680067118e-05, + "loss": 0.4001, + "step": 3814 + }, + { + "epoch": 1.6, + "learning_rate": 1.852795150346923e-05, + "loss": 0.3886, + "step": 3815 + }, + { + "epoch": 1.6, + "learning_rate": 1.8527065961138885e-05, + "loss": 0.339, + "step": 3816 + }, + { + "epoch": 1.6, + "learning_rate": 1.852618017370559e-05, + "loss": 0.3473, + "step": 3817 + }, + { + "epoch": 1.6, + "learning_rate": 1.8525294141194816e-05, + "loss": 0.3534, + "step": 3818 + }, + { + "epoch": 1.6, + "learning_rate": 1.852440786363203e-05, + "loss": 0.3533, + "step": 3819 + }, + { + "epoch": 1.6, + "learning_rate": 1.85235213410427e-05, + "loss": 0.4324, + "step": 3820 + }, + { + "epoch": 1.6, + "learning_rate": 1.8522634573452314e-05, + "loss": 0.3952, + "step": 3821 + }, + { + "epoch": 1.6, + "learning_rate": 1.8521747560886357e-05, + "loss": 0.4197, + "step": 3822 + }, + { + "epoch": 1.6, + "learning_rate": 1.8520860303370333e-05, + "loss": 0.3743, + "step": 3823 + }, + { + "epoch": 1.6, + "learning_rate": 1.851997280092974e-05, + "loss": 0.3709, + "step": 3824 + }, + { + "epoch": 1.6, + "learning_rate": 1.8519085053590083e-05, + "loss": 0.4111, + "step": 3825 + }, + { + "epoch": 1.6, + "learning_rate": 1.8518197061376895e-05, + "loss": 0.4298, + "step": 3826 + }, + { + "epoch": 1.6, + "learning_rate": 1.8517308824315688e-05, + "loss": 0.3054, + "step": 3827 + }, + { + "epoch": 1.6, + "learning_rate": 1.8516420342431994e-05, + "loss": 0.3889, + "step": 3828 + }, + { + "epoch": 1.6, + "learning_rate": 1.8515531615751357e-05, + "loss": 0.387, + "step": 3829 + }, + { + "epoch": 1.6, + "learning_rate": 1.851464264429932e-05, + "loss": 0.3876, + "step": 3830 + }, + { + "epoch": 1.6, + "learning_rate": 1.8513753428101436e-05, + "loss": 0.3743, + "step": 3831 + }, + { + "epoch": 1.6, + "learning_rate": 1.8512863967183265e-05, + "loss": 0.4047, + "step": 3832 + }, + { + "epoch": 1.6, + "learning_rate": 1.8511974261570375e-05, + "loss": 0.3511, + "step": 3833 + }, + { + "epoch": 1.61, + "learning_rate": 1.8511084311288338e-05, + "loss": 0.3873, + "step": 3834 + }, + { + "epoch": 1.61, + "learning_rate": 1.851019411636273e-05, + "loss": 0.3827, + "step": 3835 + }, + { + "epoch": 1.61, + "learning_rate": 1.850930367681915e-05, + "loss": 0.4201, + "step": 3836 + }, + { + "epoch": 1.61, + "learning_rate": 1.8508412992683183e-05, + "loss": 0.3714, + "step": 3837 + }, + { + "epoch": 1.61, + "learning_rate": 1.8507522063980436e-05, + "loss": 0.4159, + "step": 3838 + }, + { + "epoch": 1.61, + "learning_rate": 1.8506630890736515e-05, + "loss": 0.4006, + "step": 3839 + }, + { + "epoch": 1.61, + "learning_rate": 1.8505739472977043e-05, + "loss": 0.3933, + "step": 3840 + }, + { + "epoch": 1.61, + "learning_rate": 1.8504847810727632e-05, + "loss": 0.3986, + "step": 3841 + }, + { + "epoch": 1.61, + "learning_rate": 1.8503955904013922e-05, + "loss": 0.3899, + "step": 3842 + }, + { + "epoch": 1.61, + "learning_rate": 1.8503063752861543e-05, + "loss": 0.4075, + "step": 3843 + }, + { + "epoch": 1.61, + "learning_rate": 1.8502171357296144e-05, + "loss": 0.3603, + "step": 3844 + }, + { + "epoch": 1.61, + "learning_rate": 1.8501278717343372e-05, + "loss": 0.4044, + "step": 3845 + }, + { + "epoch": 1.61, + "learning_rate": 1.850038583302889e-05, + "loss": 0.375, + "step": 3846 + }, + { + "epoch": 1.61, + "learning_rate": 1.8499492704378355e-05, + "loss": 0.3622, + "step": 3847 + }, + { + "epoch": 1.61, + "learning_rate": 1.849859933141745e-05, + "loss": 0.4129, + "step": 3848 + }, + { + "epoch": 1.61, + "learning_rate": 1.8497705714171846e-05, + "loss": 0.4007, + "step": 3849 + }, + { + "epoch": 1.61, + "learning_rate": 1.849681185266723e-05, + "loss": 0.3879, + "step": 3850 + }, + { + "epoch": 1.61, + "learning_rate": 1.8495917746929303e-05, + "loss": 0.4252, + "step": 3851 + }, + { + "epoch": 1.61, + "learning_rate": 1.8495023396983754e-05, + "loss": 0.3669, + "step": 3852 + }, + { + "epoch": 1.61, + "learning_rate": 1.84941288028563e-05, + "loss": 0.3351, + "step": 3853 + }, + { + "epoch": 1.61, + "learning_rate": 1.8493233964572648e-05, + "loss": 0.3886, + "step": 3854 + }, + { + "epoch": 1.61, + "learning_rate": 1.8492338882158526e-05, + "loss": 0.415, + "step": 3855 + }, + { + "epoch": 1.61, + "learning_rate": 1.8491443555639656e-05, + "loss": 0.4072, + "step": 3856 + }, + { + "epoch": 1.61, + "learning_rate": 1.8490547985041774e-05, + "loss": 0.349, + "step": 3857 + }, + { + "epoch": 1.62, + "learning_rate": 1.8489652170390627e-05, + "loss": 0.3835, + "step": 3858 + }, + { + "epoch": 1.62, + "learning_rate": 1.848875611171196e-05, + "loss": 0.4041, + "step": 3859 + }, + { + "epoch": 1.62, + "learning_rate": 1.8487859809031537e-05, + "loss": 0.4144, + "step": 3860 + }, + { + "epoch": 1.62, + "learning_rate": 1.8486963262375113e-05, + "loss": 0.4173, + "step": 3861 + }, + { + "epoch": 1.62, + "learning_rate": 1.8486066471768457e-05, + "loss": 0.329, + "step": 3862 + }, + { + "epoch": 1.62, + "learning_rate": 1.848516943723736e-05, + "loss": 0.4009, + "step": 3863 + }, + { + "epoch": 1.62, + "learning_rate": 1.8484272158807587e-05, + "loss": 0.3595, + "step": 3864 + }, + { + "epoch": 1.62, + "learning_rate": 1.8483374636504943e-05, + "loss": 0.4056, + "step": 3865 + }, + { + "epoch": 1.62, + "learning_rate": 1.8482476870355226e-05, + "loss": 0.3897, + "step": 3866 + }, + { + "epoch": 1.62, + "learning_rate": 1.848157886038424e-05, + "loss": 0.3094, + "step": 3867 + }, + { + "epoch": 1.62, + "learning_rate": 1.848068060661779e-05, + "loss": 0.3604, + "step": 3868 + }, + { + "epoch": 1.62, + "learning_rate": 1.8479782109081702e-05, + "loss": 0.3688, + "step": 3869 + }, + { + "epoch": 1.62, + "learning_rate": 1.8478883367801804e-05, + "loss": 0.3693, + "step": 3870 + }, + { + "epoch": 1.62, + "learning_rate": 1.8477984382803928e-05, + "loss": 0.4919, + "step": 3871 + }, + { + "epoch": 1.62, + "learning_rate": 1.8477085154113913e-05, + "loss": 0.3725, + "step": 3872 + }, + { + "epoch": 1.62, + "learning_rate": 1.847618568175761e-05, + "loss": 0.3605, + "step": 3873 + }, + { + "epoch": 1.62, + "learning_rate": 1.847528596576087e-05, + "loss": 0.4308, + "step": 3874 + }, + { + "epoch": 1.62, + "learning_rate": 1.8474386006149556e-05, + "loss": 0.403, + "step": 3875 + }, + { + "epoch": 1.62, + "learning_rate": 1.847348580294954e-05, + "loss": 0.4061, + "step": 3876 + }, + { + "epoch": 1.62, + "learning_rate": 1.8472585356186688e-05, + "loss": 0.4084, + "step": 3877 + }, + { + "epoch": 1.62, + "learning_rate": 1.847168466588689e-05, + "loss": 0.4078, + "step": 3878 + }, + { + "epoch": 1.62, + "learning_rate": 1.8470783732076034e-05, + "loss": 0.3814, + "step": 3879 + }, + { + "epoch": 1.62, + "learning_rate": 1.8469882554780016e-05, + "loss": 0.4037, + "step": 3880 + }, + { + "epoch": 1.62, + "learning_rate": 1.846898113402474e-05, + "loss": 0.3961, + "step": 3881 + }, + { + "epoch": 1.63, + "learning_rate": 1.8468079469836117e-05, + "loss": 0.3874, + "step": 3882 + }, + { + "epoch": 1.63, + "learning_rate": 1.8467177562240062e-05, + "loss": 0.3401, + "step": 3883 + }, + { + "epoch": 1.63, + "learning_rate": 1.8466275411262504e-05, + "loss": 0.3788, + "step": 3884 + }, + { + "epoch": 1.63, + "learning_rate": 1.8465373016929377e-05, + "loss": 0.3697, + "step": 3885 + }, + { + "epoch": 1.63, + "learning_rate": 1.8464470379266606e-05, + "loss": 0.3385, + "step": 3886 + }, + { + "epoch": 1.63, + "learning_rate": 1.846356749830015e-05, + "loss": 0.4286, + "step": 3887 + }, + { + "epoch": 1.63, + "learning_rate": 1.8462664374055953e-05, + "loss": 0.3876, + "step": 3888 + }, + { + "epoch": 1.63, + "learning_rate": 1.8461761006559982e-05, + "loss": 0.3762, + "step": 3889 + }, + { + "epoch": 1.63, + "learning_rate": 1.84608573958382e-05, + "loss": 0.3909, + "step": 3890 + }, + { + "epoch": 1.63, + "learning_rate": 1.8459953541916576e-05, + "loss": 0.3997, + "step": 3891 + }, + { + "epoch": 1.63, + "learning_rate": 1.84590494448211e-05, + "loss": 0.3785, + "step": 3892 + }, + { + "epoch": 1.63, + "learning_rate": 1.8458145104577752e-05, + "loss": 0.4019, + "step": 3893 + }, + { + "epoch": 1.63, + "learning_rate": 1.8457240521212525e-05, + "loss": 0.3651, + "step": 3894 + }, + { + "epoch": 1.63, + "learning_rate": 1.845633569475143e-05, + "loss": 0.3841, + "step": 3895 + }, + { + "epoch": 1.63, + "learning_rate": 1.8455430625220465e-05, + "loss": 0.4117, + "step": 3896 + }, + { + "epoch": 1.63, + "learning_rate": 1.8454525312645654e-05, + "loss": 0.4056, + "step": 3897 + }, + { + "epoch": 1.63, + "learning_rate": 1.845361975705301e-05, + "loss": 0.379, + "step": 3898 + }, + { + "epoch": 1.63, + "learning_rate": 1.845271395846857e-05, + "loss": 0.4022, + "step": 3899 + }, + { + "epoch": 1.63, + "learning_rate": 1.8451807916918367e-05, + "loss": 0.4109, + "step": 3900 + }, + { + "epoch": 1.63, + "learning_rate": 1.845090163242845e-05, + "loss": 0.3745, + "step": 3901 + }, + { + "epoch": 1.63, + "learning_rate": 1.844999510502486e-05, + "loss": 0.4296, + "step": 3902 + }, + { + "epoch": 1.63, + "learning_rate": 1.844908833473366e-05, + "loss": 0.4323, + "step": 3903 + }, + { + "epoch": 1.63, + "learning_rate": 1.8448181321580916e-05, + "loss": 0.392, + "step": 3904 + }, + { + "epoch": 1.63, + "learning_rate": 1.8447274065592694e-05, + "loss": 0.4409, + "step": 3905 + }, + { + "epoch": 1.64, + "learning_rate": 1.8446366566795075e-05, + "loss": 0.4219, + "step": 3906 + }, + { + "epoch": 1.64, + "learning_rate": 1.8445458825214148e-05, + "loss": 0.4145, + "step": 3907 + }, + { + "epoch": 1.64, + "learning_rate": 1.8444550840875996e-05, + "loss": 0.3887, + "step": 3908 + }, + { + "epoch": 1.64, + "learning_rate": 1.8443642613806727e-05, + "loss": 0.3598, + "step": 3909 + }, + { + "epoch": 1.64, + "learning_rate": 1.8442734144032442e-05, + "loss": 0.3373, + "step": 3910 + }, + { + "epoch": 1.64, + "learning_rate": 1.844182543157926e-05, + "loss": 0.4054, + "step": 3911 + }, + { + "epoch": 1.64, + "learning_rate": 1.844091647647329e-05, + "loss": 0.3754, + "step": 3912 + }, + { + "epoch": 1.64, + "learning_rate": 1.8440007278740673e-05, + "loss": 0.4236, + "step": 3913 + }, + { + "epoch": 1.64, + "learning_rate": 1.8439097838407532e-05, + "loss": 0.3872, + "step": 3914 + }, + { + "epoch": 1.64, + "learning_rate": 1.8438188155500016e-05, + "loss": 0.3718, + "step": 3915 + }, + { + "epoch": 1.64, + "learning_rate": 1.8437278230044266e-05, + "loss": 0.4069, + "step": 3916 + }, + { + "epoch": 1.64, + "learning_rate": 1.8436368062066442e-05, + "loss": 0.4603, + "step": 3917 + }, + { + "epoch": 1.64, + "learning_rate": 1.8435457651592704e-05, + "loss": 0.382, + "step": 3918 + }, + { + "epoch": 1.64, + "learning_rate": 1.843454699864922e-05, + "loss": 0.3786, + "step": 3919 + }, + { + "epoch": 1.64, + "learning_rate": 1.843363610326217e-05, + "loss": 0.3877, + "step": 3920 + }, + { + "epoch": 1.64, + "learning_rate": 1.843272496545773e-05, + "loss": 0.3959, + "step": 3921 + }, + { + "epoch": 1.64, + "learning_rate": 1.8431813585262098e-05, + "loss": 0.3778, + "step": 3922 + }, + { + "epoch": 1.64, + "learning_rate": 1.8430901962701465e-05, + "loss": 0.4074, + "step": 3923 + }, + { + "epoch": 1.64, + "learning_rate": 1.8429990097802037e-05, + "loss": 0.403, + "step": 3924 + }, + { + "epoch": 1.64, + "learning_rate": 1.8429077990590026e-05, + "loss": 0.4485, + "step": 3925 + }, + { + "epoch": 1.64, + "learning_rate": 1.8428165641091643e-05, + "loss": 0.3935, + "step": 3926 + }, + { + "epoch": 1.64, + "learning_rate": 1.8427253049333124e-05, + "loss": 0.3865, + "step": 3927 + }, + { + "epoch": 1.64, + "learning_rate": 1.8426340215340688e-05, + "loss": 0.3605, + "step": 3928 + }, + { + "epoch": 1.64, + "learning_rate": 1.8425427139140586e-05, + "loss": 0.38, + "step": 3929 + }, + { + "epoch": 1.65, + "learning_rate": 1.8424513820759054e-05, + "loss": 0.4266, + "step": 3930 + }, + { + "epoch": 1.65, + "learning_rate": 1.8423600260222346e-05, + "loss": 0.4254, + "step": 3931 + }, + { + "epoch": 1.65, + "learning_rate": 1.8422686457556726e-05, + "loss": 0.398, + "step": 3932 + }, + { + "epoch": 1.65, + "learning_rate": 1.842177241278846e-05, + "loss": 0.3574, + "step": 3933 + }, + { + "epoch": 1.65, + "learning_rate": 1.8420858125943818e-05, + "loss": 0.3633, + "step": 3934 + }, + { + "epoch": 1.65, + "learning_rate": 1.841994359704908e-05, + "loss": 0.4486, + "step": 3935 + }, + { + "epoch": 1.65, + "learning_rate": 1.8419028826130537e-05, + "loss": 0.3329, + "step": 3936 + }, + { + "epoch": 1.65, + "learning_rate": 1.8418113813214477e-05, + "loss": 0.5135, + "step": 3937 + }, + { + "epoch": 1.65, + "learning_rate": 1.841719855832721e-05, + "loss": 0.4119, + "step": 3938 + }, + { + "epoch": 1.65, + "learning_rate": 1.8416283061495038e-05, + "loss": 0.3705, + "step": 3939 + }, + { + "epoch": 1.65, + "learning_rate": 1.841536732274428e-05, + "loss": 0.3832, + "step": 3940 + }, + { + "epoch": 1.65, + "learning_rate": 1.8414451342101253e-05, + "loss": 0.3343, + "step": 3941 + }, + { + "epoch": 1.65, + "learning_rate": 1.8413535119592292e-05, + "loss": 0.364, + "step": 3942 + }, + { + "epoch": 1.65, + "learning_rate": 1.841261865524373e-05, + "loss": 0.3662, + "step": 3943 + }, + { + "epoch": 1.65, + "learning_rate": 1.8411701949081906e-05, + "loss": 0.4828, + "step": 3944 + }, + { + "epoch": 1.65, + "learning_rate": 1.8410785001133177e-05, + "loss": 0.3881, + "step": 3945 + }, + { + "epoch": 1.65, + "learning_rate": 1.8409867811423897e-05, + "loss": 0.4021, + "step": 3946 + }, + { + "epoch": 1.65, + "learning_rate": 1.840895037998043e-05, + "loss": 0.4208, + "step": 3947 + }, + { + "epoch": 1.65, + "learning_rate": 1.840803270682914e-05, + "loss": 0.3583, + "step": 3948 + }, + { + "epoch": 1.65, + "learning_rate": 1.8407114791996416e-05, + "loss": 0.3859, + "step": 3949 + }, + { + "epoch": 1.65, + "learning_rate": 1.840619663550864e-05, + "loss": 0.373, + "step": 3950 + }, + { + "epoch": 1.65, + "learning_rate": 1.84052782373922e-05, + "loss": 0.3767, + "step": 3951 + }, + { + "epoch": 1.65, + "learning_rate": 1.8404359597673492e-05, + "loss": 0.3835, + "step": 3952 + }, + { + "epoch": 1.66, + "learning_rate": 1.840344071637893e-05, + "loss": 0.3904, + "step": 3953 + }, + { + "epoch": 1.66, + "learning_rate": 1.840252159353492e-05, + "loss": 0.3753, + "step": 3954 + }, + { + "epoch": 1.66, + "learning_rate": 1.840160222916788e-05, + "loss": 0.3811, + "step": 3955 + }, + { + "epoch": 1.66, + "learning_rate": 1.8400682623304243e-05, + "loss": 0.362, + "step": 3956 + }, + { + "epoch": 1.66, + "learning_rate": 1.839976277597044e-05, + "loss": 0.4012, + "step": 3957 + }, + { + "epoch": 1.66, + "learning_rate": 1.8398842687192906e-05, + "loss": 0.408, + "step": 3958 + }, + { + "epoch": 1.66, + "learning_rate": 1.839792235699809e-05, + "loss": 0.3913, + "step": 3959 + }, + { + "epoch": 1.66, + "learning_rate": 1.8397001785412455e-05, + "loss": 0.3546, + "step": 3960 + }, + { + "epoch": 1.66, + "learning_rate": 1.839608097246245e-05, + "loss": 0.3846, + "step": 3961 + }, + { + "epoch": 1.66, + "learning_rate": 1.839515991817455e-05, + "loss": 0.3486, + "step": 3962 + }, + { + "epoch": 1.66, + "learning_rate": 1.8394238622575228e-05, + "loss": 0.4618, + "step": 3963 + }, + { + "epoch": 1.66, + "learning_rate": 1.8393317085690964e-05, + "loss": 0.3483, + "step": 3964 + }, + { + "epoch": 1.66, + "learning_rate": 1.839239530754825e-05, + "loss": 0.4279, + "step": 3965 + }, + { + "epoch": 1.66, + "learning_rate": 1.839147328817358e-05, + "loss": 0.4616, + "step": 3966 + }, + { + "epoch": 1.66, + "learning_rate": 1.8390551027593456e-05, + "loss": 0.4158, + "step": 3967 + }, + { + "epoch": 1.66, + "learning_rate": 1.8389628525834385e-05, + "loss": 0.3667, + "step": 3968 + }, + { + "epoch": 1.66, + "learning_rate": 1.8388705782922893e-05, + "loss": 0.4405, + "step": 3969 + }, + { + "epoch": 1.66, + "learning_rate": 1.8387782798885495e-05, + "loss": 0.4028, + "step": 3970 + }, + { + "epoch": 1.66, + "learning_rate": 1.8386859573748724e-05, + "loss": 0.3949, + "step": 3971 + }, + { + "epoch": 1.66, + "learning_rate": 1.8385936107539115e-05, + "loss": 0.3959, + "step": 3972 + }, + { + "epoch": 1.66, + "learning_rate": 1.8385012400283218e-05, + "loss": 0.4073, + "step": 3973 + }, + { + "epoch": 1.66, + "learning_rate": 1.838408845200758e-05, + "loss": 0.3725, + "step": 3974 + }, + { + "epoch": 1.66, + "learning_rate": 1.8383164262738756e-05, + "loss": 0.3761, + "step": 3975 + }, + { + "epoch": 1.66, + "learning_rate": 1.838223983250331e-05, + "loss": 0.3953, + "step": 3976 + }, + { + "epoch": 1.67, + "learning_rate": 1.8381315161327827e-05, + "loss": 0.3925, + "step": 3977 + }, + { + "epoch": 1.67, + "learning_rate": 1.8380390249238876e-05, + "loss": 0.3745, + "step": 3978 + }, + { + "epoch": 1.67, + "learning_rate": 1.837946509626304e-05, + "loss": 0.3934, + "step": 3979 + }, + { + "epoch": 1.67, + "learning_rate": 1.8378539702426917e-05, + "loss": 0.348, + "step": 3980 + }, + { + "epoch": 1.67, + "learning_rate": 1.8377614067757106e-05, + "loss": 0.3714, + "step": 3981 + }, + { + "epoch": 1.67, + "learning_rate": 1.8376688192280213e-05, + "loss": 0.384, + "step": 3982 + }, + { + "epoch": 1.67, + "learning_rate": 1.837576207602285e-05, + "loss": 0.3941, + "step": 3983 + }, + { + "epoch": 1.67, + "learning_rate": 1.837483571901164e-05, + "loss": 0.3607, + "step": 3984 + }, + { + "epoch": 1.67, + "learning_rate": 1.837390912127321e-05, + "loss": 0.3494, + "step": 3985 + }, + { + "epoch": 1.67, + "learning_rate": 1.837298228283419e-05, + "loss": 0.3867, + "step": 3986 + }, + { + "epoch": 1.67, + "learning_rate": 1.837205520372123e-05, + "loss": 0.369, + "step": 3987 + }, + { + "epoch": 1.67, + "learning_rate": 1.8371127883960963e-05, + "loss": 0.3949, + "step": 3988 + }, + { + "epoch": 1.67, + "learning_rate": 1.837020032358006e-05, + "loss": 0.3878, + "step": 3989 + }, + { + "epoch": 1.67, + "learning_rate": 1.8369272522605177e-05, + "loss": 0.4191, + "step": 3990 + }, + { + "epoch": 1.67, + "learning_rate": 1.836834448106298e-05, + "loss": 0.3945, + "step": 3991 + }, + { + "epoch": 1.67, + "learning_rate": 1.836741619898015e-05, + "loss": 0.4298, + "step": 3992 + }, + { + "epoch": 1.67, + "learning_rate": 1.836648767638336e-05, + "loss": 0.3937, + "step": 3993 + }, + { + "epoch": 1.67, + "learning_rate": 1.8365558913299307e-05, + "loss": 0.4084, + "step": 3994 + }, + { + "epoch": 1.67, + "learning_rate": 1.836462990975469e-05, + "loss": 0.3711, + "step": 3995 + }, + { + "epoch": 1.67, + "learning_rate": 1.836370066577621e-05, + "loss": 0.3859, + "step": 3996 + }, + { + "epoch": 1.67, + "learning_rate": 1.836277118139057e-05, + "loss": 0.3787, + "step": 3997 + }, + { + "epoch": 1.67, + "learning_rate": 1.83618414566245e-05, + "loss": 0.387, + "step": 3998 + }, + { + "epoch": 1.67, + "learning_rate": 1.8360911491504714e-05, + "loss": 0.3551, + "step": 3999 + }, + { + "epoch": 1.67, + "learning_rate": 1.835998128605795e-05, + "loss": 0.409, + "step": 4000 + }, + { + "epoch": 1.68, + "learning_rate": 1.8359050840310936e-05, + "loss": 0.3925, + "step": 4001 + }, + { + "epoch": 1.68, + "learning_rate": 1.8358120154290427e-05, + "loss": 0.4395, + "step": 4002 + }, + { + "epoch": 1.68, + "learning_rate": 1.8357189228023174e-05, + "loss": 0.425, + "step": 4003 + }, + { + "epoch": 1.68, + "learning_rate": 1.835625806153593e-05, + "loss": 0.3778, + "step": 4004 + }, + { + "epoch": 1.68, + "learning_rate": 1.835532665485546e-05, + "loss": 0.3671, + "step": 4005 + }, + { + "epoch": 1.68, + "learning_rate": 1.8354395008008545e-05, + "loss": 0.4004, + "step": 4006 + }, + { + "epoch": 1.68, + "learning_rate": 1.835346312102196e-05, + "loss": 0.3547, + "step": 4007 + }, + { + "epoch": 1.68, + "learning_rate": 1.8352530993922483e-05, + "loss": 0.4045, + "step": 4008 + }, + { + "epoch": 1.68, + "learning_rate": 1.835159862673692e-05, + "loss": 0.3707, + "step": 4009 + }, + { + "epoch": 1.68, + "learning_rate": 1.8350666019492065e-05, + "loss": 0.3821, + "step": 4010 + }, + { + "epoch": 1.68, + "learning_rate": 1.8349733172214722e-05, + "loss": 0.4071, + "step": 4011 + }, + { + "epoch": 1.68, + "learning_rate": 1.8348800084931712e-05, + "loss": 0.4002, + "step": 4012 + }, + { + "epoch": 1.68, + "learning_rate": 1.834786675766985e-05, + "loss": 0.3747, + "step": 4013 + }, + { + "epoch": 1.68, + "learning_rate": 1.834693319045597e-05, + "loss": 0.4455, + "step": 4014 + }, + { + "epoch": 1.68, + "learning_rate": 1.8345999383316897e-05, + "loss": 0.3962, + "step": 4015 + }, + { + "epoch": 1.68, + "learning_rate": 1.8345065336279482e-05, + "loss": 0.3184, + "step": 4016 + }, + { + "epoch": 1.68, + "learning_rate": 1.8344131049370566e-05, + "loss": 0.3856, + "step": 4017 + }, + { + "epoch": 1.68, + "learning_rate": 1.834319652261701e-05, + "loss": 0.4466, + "step": 4018 + }, + { + "epoch": 1.68, + "learning_rate": 1.8342261756045673e-05, + "loss": 0.4317, + "step": 4019 + }, + { + "epoch": 1.68, + "learning_rate": 1.834132674968342e-05, + "loss": 0.3718, + "step": 4020 + }, + { + "epoch": 1.68, + "learning_rate": 1.834039150355714e-05, + "loss": 0.3773, + "step": 4021 + }, + { + "epoch": 1.68, + "learning_rate": 1.83394560176937e-05, + "loss": 0.3929, + "step": 4022 + }, + { + "epoch": 1.68, + "learning_rate": 1.8338520292120002e-05, + "loss": 0.4016, + "step": 4023 + }, + { + "epoch": 1.68, + "learning_rate": 1.8337584326862935e-05, + "loss": 0.4165, + "step": 4024 + }, + { + "epoch": 1.69, + "learning_rate": 1.8336648121949405e-05, + "loss": 0.3676, + "step": 4025 + }, + { + "epoch": 1.69, + "learning_rate": 1.8335711677406324e-05, + "loss": 0.3762, + "step": 4026 + }, + { + "epoch": 1.69, + "learning_rate": 1.833477499326061e-05, + "loss": 0.4556, + "step": 4027 + }, + { + "epoch": 1.69, + "learning_rate": 1.833383806953918e-05, + "loss": 0.3767, + "step": 4028 + }, + { + "epoch": 1.69, + "learning_rate": 1.8332900906268972e-05, + "loss": 0.3779, + "step": 4029 + }, + { + "epoch": 1.69, + "learning_rate": 1.8331963503476925e-05, + "loss": 0.3795, + "step": 4030 + }, + { + "epoch": 1.69, + "learning_rate": 1.8331025861189974e-05, + "loss": 0.4227, + "step": 4031 + }, + { + "epoch": 1.69, + "learning_rate": 1.8330087979435083e-05, + "loss": 0.3875, + "step": 4032 + }, + { + "epoch": 1.69, + "learning_rate": 1.8329149858239204e-05, + "loss": 0.4157, + "step": 4033 + }, + { + "epoch": 1.69, + "learning_rate": 1.8328211497629304e-05, + "loss": 0.4011, + "step": 4034 + }, + { + "epoch": 1.69, + "learning_rate": 1.832727289763235e-05, + "loss": 0.3993, + "step": 4035 + }, + { + "epoch": 1.69, + "learning_rate": 1.832633405827533e-05, + "loss": 0.3993, + "step": 4036 + }, + { + "epoch": 1.69, + "learning_rate": 1.8325394979585227e-05, + "loss": 0.4001, + "step": 4037 + }, + { + "epoch": 1.69, + "learning_rate": 1.8324455661589033e-05, + "loss": 0.3964, + "step": 4038 + }, + { + "epoch": 1.69, + "learning_rate": 1.8323516104313746e-05, + "loss": 0.424, + "step": 4039 + }, + { + "epoch": 1.69, + "learning_rate": 1.8322576307786376e-05, + "loss": 0.4006, + "step": 4040 + }, + { + "epoch": 1.69, + "learning_rate": 1.8321636272033934e-05, + "loss": 0.3782, + "step": 4041 + }, + { + "epoch": 1.69, + "learning_rate": 1.8320695997083443e-05, + "loss": 0.3963, + "step": 4042 + }, + { + "epoch": 1.69, + "learning_rate": 1.831975548296193e-05, + "loss": 0.3684, + "step": 4043 + }, + { + "epoch": 1.69, + "learning_rate": 1.8318814729696427e-05, + "loss": 0.4097, + "step": 4044 + }, + { + "epoch": 1.69, + "learning_rate": 1.8317873737313977e-05, + "loss": 0.3876, + "step": 4045 + }, + { + "epoch": 1.69, + "learning_rate": 1.831693250584163e-05, + "loss": 0.4022, + "step": 4046 + }, + { + "epoch": 1.69, + "learning_rate": 1.8315991035306435e-05, + "loss": 0.3946, + "step": 4047 + }, + { + "epoch": 1.69, + "learning_rate": 1.831504932573546e-05, + "loss": 0.4278, + "step": 4048 + }, + { + "epoch": 1.7, + "learning_rate": 1.831410737715577e-05, + "loss": 0.391, + "step": 4049 + }, + { + "epoch": 1.7, + "learning_rate": 1.8313165189594438e-05, + "loss": 0.3522, + "step": 4050 + }, + { + "epoch": 1.7, + "learning_rate": 1.8312222763078552e-05, + "loss": 0.3493, + "step": 4051 + }, + { + "epoch": 1.7, + "learning_rate": 1.83112800976352e-05, + "loss": 0.3808, + "step": 4052 + }, + { + "epoch": 1.7, + "learning_rate": 1.8310337193291473e-05, + "loss": 0.376, + "step": 4053 + }, + { + "epoch": 1.7, + "learning_rate": 1.8309394050074482e-05, + "loss": 0.3485, + "step": 4054 + }, + { + "epoch": 1.7, + "learning_rate": 1.8308450668011334e-05, + "loss": 0.3439, + "step": 4055 + }, + { + "epoch": 1.7, + "learning_rate": 1.830750704712914e-05, + "loss": 0.4055, + "step": 4056 + }, + { + "epoch": 1.7, + "learning_rate": 1.830656318745503e-05, + "loss": 0.4463, + "step": 4057 + }, + { + "epoch": 1.7, + "learning_rate": 1.830561908901613e-05, + "loss": 0.3713, + "step": 4058 + }, + { + "epoch": 1.7, + "learning_rate": 1.8304674751839583e-05, + "loss": 0.3569, + "step": 4059 + }, + { + "epoch": 1.7, + "learning_rate": 1.830373017595253e-05, + "loss": 0.3846, + "step": 4060 + }, + { + "epoch": 1.7, + "learning_rate": 1.8302785361382117e-05, + "loss": 0.4214, + "step": 4061 + }, + { + "epoch": 1.7, + "learning_rate": 1.8301840308155507e-05, + "loss": 0.3752, + "step": 4062 + }, + { + "epoch": 1.7, + "learning_rate": 1.8300895016299868e-05, + "loss": 0.4175, + "step": 4063 + }, + { + "epoch": 1.7, + "learning_rate": 1.8299949485842366e-05, + "loss": 0.4112, + "step": 4064 + }, + { + "epoch": 1.7, + "learning_rate": 1.829900371681018e-05, + "loss": 0.436, + "step": 4065 + }, + { + "epoch": 1.7, + "learning_rate": 1.8298057709230498e-05, + "loss": 0.4509, + "step": 4066 + }, + { + "epoch": 1.7, + "learning_rate": 1.8297111463130512e-05, + "loss": 0.4404, + "step": 4067 + }, + { + "epoch": 1.7, + "learning_rate": 1.8296164978537422e-05, + "loss": 0.3855, + "step": 4068 + }, + { + "epoch": 1.7, + "learning_rate": 1.8295218255478425e-05, + "loss": 0.3585, + "step": 4069 + }, + { + "epoch": 1.7, + "learning_rate": 1.8294271293980744e-05, + "loss": 0.4391, + "step": 4070 + }, + { + "epoch": 1.7, + "learning_rate": 1.8293324094071593e-05, + "loss": 0.34, + "step": 4071 + }, + { + "epoch": 1.7, + "learning_rate": 1.8292376655778202e-05, + "loss": 0.3956, + "step": 4072 + }, + { + "epoch": 1.71, + "learning_rate": 1.8291428979127802e-05, + "loss": 0.3659, + "step": 4073 + }, + { + "epoch": 1.71, + "learning_rate": 1.8290481064147637e-05, + "loss": 0.3756, + "step": 4074 + }, + { + "epoch": 1.71, + "learning_rate": 1.8289532910864943e-05, + "loss": 0.357, + "step": 4075 + }, + { + "epoch": 1.71, + "learning_rate": 1.8288584519306985e-05, + "loss": 0.3498, + "step": 4076 + }, + { + "epoch": 1.71, + "learning_rate": 1.8287635889501024e-05, + "loss": 0.3925, + "step": 4077 + }, + { + "epoch": 1.71, + "learning_rate": 1.828668702147432e-05, + "loss": 0.3581, + "step": 4078 + }, + { + "epoch": 1.71, + "learning_rate": 1.828573791525415e-05, + "loss": 0.372, + "step": 4079 + }, + { + "epoch": 1.71, + "learning_rate": 1.82847885708678e-05, + "loss": 0.388, + "step": 4080 + }, + { + "epoch": 1.71, + "learning_rate": 1.8283838988342553e-05, + "loss": 0.3954, + "step": 4081 + }, + { + "epoch": 1.71, + "learning_rate": 1.8282889167705704e-05, + "loss": 0.3438, + "step": 4082 + }, + { + "epoch": 1.71, + "learning_rate": 1.8281939108984557e-05, + "loss": 0.3651, + "step": 4083 + }, + { + "epoch": 1.71, + "learning_rate": 1.828098881220642e-05, + "loss": 0.3769, + "step": 4084 + }, + { + "epoch": 1.71, + "learning_rate": 1.8280038277398606e-05, + "loss": 0.39, + "step": 4085 + }, + { + "epoch": 1.71, + "learning_rate": 1.827908750458844e-05, + "loss": 0.3994, + "step": 4086 + }, + { + "epoch": 1.71, + "learning_rate": 1.8278136493803253e-05, + "loss": 0.3713, + "step": 4087 + }, + { + "epoch": 1.71, + "learning_rate": 1.8277185245070373e-05, + "loss": 0.4595, + "step": 4088 + }, + { + "epoch": 1.71, + "learning_rate": 1.8276233758417153e-05, + "loss": 0.4077, + "step": 4089 + }, + { + "epoch": 1.71, + "learning_rate": 1.827528203387094e-05, + "loss": 0.4482, + "step": 4090 + }, + { + "epoch": 1.71, + "learning_rate": 1.8274330071459082e-05, + "loss": 0.3704, + "step": 4091 + }, + { + "epoch": 1.71, + "learning_rate": 1.8273377871208952e-05, + "loss": 0.392, + "step": 4092 + }, + { + "epoch": 1.71, + "learning_rate": 1.8272425433147916e-05, + "loss": 0.3883, + "step": 4093 + }, + { + "epoch": 1.71, + "learning_rate": 1.8271472757303352e-05, + "loss": 0.3909, + "step": 4094 + }, + { + "epoch": 1.71, + "learning_rate": 1.827051984370264e-05, + "loss": 0.4504, + "step": 4095 + }, + { + "epoch": 1.71, + "learning_rate": 1.826956669237318e-05, + "loss": 0.3842, + "step": 4096 + }, + { + "epoch": 1.72, + "learning_rate": 1.8268613303342362e-05, + "loss": 0.3769, + "step": 4097 + }, + { + "epoch": 1.72, + "learning_rate": 1.826765967663759e-05, + "loss": 0.3996, + "step": 4098 + }, + { + "epoch": 1.72, + "learning_rate": 1.8266705812286276e-05, + "loss": 0.3751, + "step": 4099 + }, + { + "epoch": 1.72, + "learning_rate": 1.8265751710315843e-05, + "loss": 0.3925, + "step": 4100 + }, + { + "epoch": 1.72, + "learning_rate": 1.826479737075371e-05, + "loss": 0.3749, + "step": 4101 + }, + { + "epoch": 1.72, + "learning_rate": 1.8263842793627314e-05, + "loss": 0.3599, + "step": 4102 + }, + { + "epoch": 1.72, + "learning_rate": 1.8262887978964087e-05, + "loss": 0.4009, + "step": 4103 + }, + { + "epoch": 1.72, + "learning_rate": 1.8261932926791477e-05, + "loss": 0.3794, + "step": 4104 + }, + { + "epoch": 1.72, + "learning_rate": 1.8260977637136937e-05, + "loss": 0.3634, + "step": 4105 + }, + { + "epoch": 1.72, + "learning_rate": 1.8260022110027925e-05, + "loss": 0.3815, + "step": 4106 + }, + { + "epoch": 1.72, + "learning_rate": 1.8259066345491908e-05, + "loss": 0.447, + "step": 4107 + }, + { + "epoch": 1.72, + "learning_rate": 1.8258110343556358e-05, + "loss": 0.4141, + "step": 4108 + }, + { + "epoch": 1.72, + "learning_rate": 1.8257154104248752e-05, + "loss": 0.407, + "step": 4109 + }, + { + "epoch": 1.72, + "learning_rate": 1.8256197627596582e-05, + "loss": 0.395, + "step": 4110 + }, + { + "epoch": 1.72, + "learning_rate": 1.8255240913627335e-05, + "loss": 0.4496, + "step": 4111 + }, + { + "epoch": 1.72, + "learning_rate": 1.8254283962368517e-05, + "loss": 0.366, + "step": 4112 + }, + { + "epoch": 1.72, + "learning_rate": 1.8253326773847628e-05, + "loss": 0.3714, + "step": 4113 + }, + { + "epoch": 1.72, + "learning_rate": 1.8252369348092184e-05, + "loss": 0.3777, + "step": 4114 + }, + { + "epoch": 1.72, + "learning_rate": 1.825141168512971e-05, + "loss": 0.4085, + "step": 4115 + }, + { + "epoch": 1.72, + "learning_rate": 1.825045378498773e-05, + "loss": 0.4169, + "step": 4116 + }, + { + "epoch": 1.72, + "learning_rate": 1.824949564769377e-05, + "loss": 0.4316, + "step": 4117 + }, + { + "epoch": 1.72, + "learning_rate": 1.8248537273275385e-05, + "loss": 0.3995, + "step": 4118 + }, + { + "epoch": 1.72, + "learning_rate": 1.8247578661760114e-05, + "loss": 0.4057, + "step": 4119 + }, + { + "epoch": 1.72, + "learning_rate": 1.824661981317551e-05, + "loss": 0.389, + "step": 4120 + }, + { + "epoch": 1.73, + "learning_rate": 1.8245660727549142e-05, + "loss": 0.4068, + "step": 4121 + }, + { + "epoch": 1.73, + "learning_rate": 1.824470140490857e-05, + "loss": 0.4061, + "step": 4122 + }, + { + "epoch": 1.73, + "learning_rate": 1.8243741845281373e-05, + "loss": 0.4385, + "step": 4123 + }, + { + "epoch": 1.73, + "learning_rate": 1.824278204869513e-05, + "loss": 0.4081, + "step": 4124 + }, + { + "epoch": 1.73, + "learning_rate": 1.8241822015177433e-05, + "loss": 0.4129, + "step": 4125 + }, + { + "epoch": 1.73, + "learning_rate": 1.824086174475588e-05, + "loss": 0.3608, + "step": 4126 + }, + { + "epoch": 1.73, + "learning_rate": 1.8239901237458064e-05, + "loss": 0.4721, + "step": 4127 + }, + { + "epoch": 1.73, + "learning_rate": 1.8238940493311604e-05, + "loss": 0.3642, + "step": 4128 + }, + { + "epoch": 1.73, + "learning_rate": 1.8237979512344107e-05, + "loss": 0.3647, + "step": 4129 + }, + { + "epoch": 1.73, + "learning_rate": 1.8237018294583196e-05, + "loss": 0.4653, + "step": 4130 + }, + { + "epoch": 1.73, + "learning_rate": 1.8236056840056512e-05, + "loss": 0.4066, + "step": 4131 + }, + { + "epoch": 1.73, + "learning_rate": 1.8235095148791675e-05, + "loss": 0.3798, + "step": 4132 + }, + { + "epoch": 1.73, + "learning_rate": 1.823413322081634e-05, + "loss": 0.388, + "step": 4133 + }, + { + "epoch": 1.73, + "learning_rate": 1.823317105615815e-05, + "loss": 0.4043, + "step": 4134 + }, + { + "epoch": 1.73, + "learning_rate": 1.8232208654844766e-05, + "loss": 0.3316, + "step": 4135 + }, + { + "epoch": 1.73, + "learning_rate": 1.823124601690385e-05, + "loss": 0.3562, + "step": 4136 + }, + { + "epoch": 1.73, + "learning_rate": 1.8230283142363068e-05, + "loss": 0.4149, + "step": 4137 + }, + { + "epoch": 1.73, + "learning_rate": 1.8229320031250098e-05, + "loss": 0.3378, + "step": 4138 + }, + { + "epoch": 1.73, + "learning_rate": 1.822835668359263e-05, + "loss": 0.4156, + "step": 4139 + }, + { + "epoch": 1.73, + "learning_rate": 1.8227393099418353e-05, + "loss": 0.3673, + "step": 4140 + }, + { + "epoch": 1.73, + "learning_rate": 1.8226429278754963e-05, + "loss": 0.4593, + "step": 4141 + }, + { + "epoch": 1.73, + "learning_rate": 1.822546522163016e-05, + "loss": 0.3879, + "step": 4142 + }, + { + "epoch": 1.73, + "learning_rate": 1.822450092807166e-05, + "loss": 0.3846, + "step": 4143 + }, + { + "epoch": 1.73, + "learning_rate": 1.8223536398107177e-05, + "loss": 0.3752, + "step": 4144 + }, + { + "epoch": 1.74, + "learning_rate": 1.822257163176444e-05, + "loss": 0.4094, + "step": 4145 + }, + { + "epoch": 1.74, + "learning_rate": 1.8221606629071175e-05, + "loss": 0.3852, + "step": 4146 + }, + { + "epoch": 1.74, + "learning_rate": 1.8220641390055127e-05, + "loss": 0.3741, + "step": 4147 + }, + { + "epoch": 1.74, + "learning_rate": 1.8219675914744035e-05, + "loss": 0.4307, + "step": 4148 + }, + { + "epoch": 1.74, + "learning_rate": 1.8218710203165653e-05, + "loss": 0.3892, + "step": 4149 + }, + { + "epoch": 1.74, + "learning_rate": 1.821774425534774e-05, + "loss": 0.3891, + "step": 4150 + }, + { + "epoch": 1.74, + "learning_rate": 1.821677807131806e-05, + "loss": 0.389, + "step": 4151 + }, + { + "epoch": 1.74, + "learning_rate": 1.8215811651104383e-05, + "loss": 0.3828, + "step": 4152 + }, + { + "epoch": 1.74, + "learning_rate": 1.8214844994734498e-05, + "loss": 0.3996, + "step": 4153 + }, + { + "epoch": 1.74, + "learning_rate": 1.8213878102236177e-05, + "loss": 0.4122, + "step": 4154 + }, + { + "epoch": 1.74, + "learning_rate": 1.821291097363722e-05, + "loss": 0.3949, + "step": 4155 + }, + { + "epoch": 1.74, + "learning_rate": 1.821194360896543e-05, + "loss": 0.3584, + "step": 4156 + }, + { + "epoch": 1.74, + "learning_rate": 1.8210976008248603e-05, + "loss": 0.3853, + "step": 4157 + }, + { + "epoch": 1.74, + "learning_rate": 1.821000817151456e-05, + "loss": 0.4165, + "step": 4158 + }, + { + "epoch": 1.74, + "learning_rate": 1.820904009879111e-05, + "loss": 0.4278, + "step": 4159 + }, + { + "epoch": 1.74, + "learning_rate": 1.8208071790106096e-05, + "loss": 0.3944, + "step": 4160 + }, + { + "epoch": 1.74, + "learning_rate": 1.8207103245487335e-05, + "loss": 0.3955, + "step": 4161 + }, + { + "epoch": 1.74, + "learning_rate": 1.8206134464962678e-05, + "loss": 0.3999, + "step": 4162 + }, + { + "epoch": 1.74, + "learning_rate": 1.820516544855997e-05, + "loss": 0.4308, + "step": 4163 + }, + { + "epoch": 1.74, + "learning_rate": 1.8204196196307058e-05, + "loss": 0.3431, + "step": 4164 + }, + { + "epoch": 1.74, + "learning_rate": 1.8203226708231806e-05, + "loss": 0.3637, + "step": 4165 + }, + { + "epoch": 1.74, + "learning_rate": 1.8202256984362087e-05, + "loss": 0.3876, + "step": 4166 + }, + { + "epoch": 1.74, + "learning_rate": 1.8201287024725765e-05, + "loss": 0.3851, + "step": 4167 + }, + { + "epoch": 1.75, + "learning_rate": 1.8200316829350728e-05, + "loss": 0.3599, + "step": 4168 + }, + { + "epoch": 1.75, + "learning_rate": 1.8199346398264856e-05, + "loss": 0.4277, + "step": 4169 + }, + { + "epoch": 1.75, + "learning_rate": 1.8198375731496054e-05, + "loss": 0.3442, + "step": 4170 + }, + { + "epoch": 1.75, + "learning_rate": 1.8197404829072214e-05, + "loss": 0.4514, + "step": 4171 + }, + { + "epoch": 1.75, + "learning_rate": 1.8196433691021243e-05, + "loss": 0.4127, + "step": 4172 + }, + { + "epoch": 1.75, + "learning_rate": 1.8195462317371063e-05, + "loss": 0.4018, + "step": 4173 + }, + { + "epoch": 1.75, + "learning_rate": 1.819449070814959e-05, + "loss": 0.4155, + "step": 4174 + }, + { + "epoch": 1.75, + "learning_rate": 1.8193518863384753e-05, + "loss": 0.3667, + "step": 4175 + }, + { + "epoch": 1.75, + "learning_rate": 1.819254678310449e-05, + "loss": 0.3539, + "step": 4176 + }, + { + "epoch": 1.75, + "learning_rate": 1.8191574467336737e-05, + "loss": 0.4054, + "step": 4177 + }, + { + "epoch": 1.75, + "learning_rate": 1.8190601916109444e-05, + "loss": 0.4086, + "step": 4178 + }, + { + "epoch": 1.75, + "learning_rate": 1.818962912945057e-05, + "loss": 0.3798, + "step": 4179 + }, + { + "epoch": 1.75, + "learning_rate": 1.818865610738807e-05, + "loss": 0.4052, + "step": 4180 + }, + { + "epoch": 1.75, + "learning_rate": 1.818768284994992e-05, + "loss": 0.3307, + "step": 4181 + }, + { + "epoch": 1.75, + "learning_rate": 1.818670935716409e-05, + "loss": 0.3998, + "step": 4182 + }, + { + "epoch": 1.75, + "learning_rate": 1.8185735629058566e-05, + "loss": 0.3693, + "step": 4183 + }, + { + "epoch": 1.75, + "learning_rate": 1.8184761665661335e-05, + "loss": 0.3871, + "step": 4184 + }, + { + "epoch": 1.75, + "learning_rate": 1.8183787467000392e-05, + "loss": 0.3721, + "step": 4185 + }, + { + "epoch": 1.75, + "learning_rate": 1.8182813033103742e-05, + "loss": 0.3355, + "step": 4186 + }, + { + "epoch": 1.75, + "learning_rate": 1.8181838363999392e-05, + "loss": 0.4374, + "step": 4187 + }, + { + "epoch": 1.75, + "learning_rate": 1.8180863459715358e-05, + "loss": 0.3902, + "step": 4188 + }, + { + "epoch": 1.75, + "learning_rate": 1.8179888320279663e-05, + "loss": 0.4099, + "step": 4189 + }, + { + "epoch": 1.75, + "learning_rate": 1.817891294572034e-05, + "loss": 0.3684, + "step": 4190 + }, + { + "epoch": 1.75, + "learning_rate": 1.817793733606542e-05, + "loss": 0.4386, + "step": 4191 + }, + { + "epoch": 1.76, + "learning_rate": 1.8176961491342947e-05, + "loss": 0.3864, + "step": 4192 + }, + { + "epoch": 1.76, + "learning_rate": 1.8175985411580975e-05, + "loss": 0.4155, + "step": 4193 + }, + { + "epoch": 1.76, + "learning_rate": 1.8175009096807557e-05, + "loss": 0.4204, + "step": 4194 + }, + { + "epoch": 1.76, + "learning_rate": 1.8174032547050754e-05, + "loss": 0.382, + "step": 4195 + }, + { + "epoch": 1.76, + "learning_rate": 1.8173055762338642e-05, + "loss": 0.4276, + "step": 4196 + }, + { + "epoch": 1.76, + "learning_rate": 1.8172078742699293e-05, + "loss": 0.3971, + "step": 4197 + }, + { + "epoch": 1.76, + "learning_rate": 1.817110148816079e-05, + "loss": 0.3628, + "step": 4198 + }, + { + "epoch": 1.76, + "learning_rate": 1.817012399875123e-05, + "loss": 0.4342, + "step": 4199 + }, + { + "epoch": 1.76, + "learning_rate": 1.8169146274498707e-05, + "loss": 0.3692, + "step": 4200 + }, + { + "epoch": 1.76, + "learning_rate": 1.8168168315431318e-05, + "loss": 0.3981, + "step": 4201 + }, + { + "epoch": 1.76, + "learning_rate": 1.8167190121577184e-05, + "loss": 0.4388, + "step": 4202 + }, + { + "epoch": 1.76, + "learning_rate": 1.8166211692964413e-05, + "loss": 0.3119, + "step": 4203 + }, + { + "epoch": 1.76, + "learning_rate": 1.8165233029621136e-05, + "loss": 0.3513, + "step": 4204 + }, + { + "epoch": 1.76, + "learning_rate": 1.816425413157548e-05, + "loss": 0.4071, + "step": 4205 + }, + { + "epoch": 1.76, + "learning_rate": 1.8163274998855585e-05, + "loss": 0.4342, + "step": 4206 + }, + { + "epoch": 1.76, + "learning_rate": 1.8162295631489595e-05, + "loss": 0.418, + "step": 4207 + }, + { + "epoch": 1.76, + "learning_rate": 1.8161316029505656e-05, + "loss": 0.3763, + "step": 4208 + }, + { + "epoch": 1.76, + "learning_rate": 1.8160336192931935e-05, + "loss": 0.4215, + "step": 4209 + }, + { + "epoch": 1.76, + "learning_rate": 1.8159356121796586e-05, + "loss": 0.3893, + "step": 4210 + }, + { + "epoch": 1.76, + "learning_rate": 1.8158375816127792e-05, + "loss": 0.4154, + "step": 4211 + }, + { + "epoch": 1.76, + "learning_rate": 1.8157395275953722e-05, + "loss": 0.375, + "step": 4212 + }, + { + "epoch": 1.76, + "learning_rate": 1.8156414501302565e-05, + "loss": 0.3681, + "step": 4213 + }, + { + "epoch": 1.76, + "learning_rate": 1.8155433492202507e-05, + "loss": 0.3854, + "step": 4214 + }, + { + "epoch": 1.76, + "learning_rate": 1.8154452248681754e-05, + "loss": 0.3947, + "step": 4215 + }, + { + "epoch": 1.77, + "learning_rate": 1.8153470770768506e-05, + "loss": 0.3914, + "step": 4216 + }, + { + "epoch": 1.77, + "learning_rate": 1.8152489058490975e-05, + "loss": 0.3556, + "step": 4217 + }, + { + "epoch": 1.77, + "learning_rate": 1.8151507111877382e-05, + "loss": 0.3644, + "step": 4218 + }, + { + "epoch": 1.77, + "learning_rate": 1.815052493095595e-05, + "loss": 0.3726, + "step": 4219 + }, + { + "epoch": 1.77, + "learning_rate": 1.814954251575491e-05, + "loss": 0.4142, + "step": 4220 + }, + { + "epoch": 1.77, + "learning_rate": 1.8148559866302505e-05, + "loss": 0.3856, + "step": 4221 + }, + { + "epoch": 1.77, + "learning_rate": 1.8147576982626974e-05, + "loss": 0.3581, + "step": 4222 + }, + { + "epoch": 1.77, + "learning_rate": 1.8146593864756577e-05, + "loss": 0.4268, + "step": 4223 + }, + { + "epoch": 1.77, + "learning_rate": 1.8145610512719565e-05, + "loss": 0.3792, + "step": 4224 + }, + { + "epoch": 1.77, + "learning_rate": 1.8144626926544208e-05, + "loss": 0.3814, + "step": 4225 + }, + { + "epoch": 1.77, + "learning_rate": 1.814364310625878e-05, + "loss": 0.3749, + "step": 4226 + }, + { + "epoch": 1.77, + "learning_rate": 1.8142659051891552e-05, + "loss": 0.3667, + "step": 4227 + }, + { + "epoch": 1.77, + "learning_rate": 1.814167476347082e-05, + "loss": 0.3915, + "step": 4228 + }, + { + "epoch": 1.77, + "learning_rate": 1.8140690241024872e-05, + "loss": 0.4039, + "step": 4229 + }, + { + "epoch": 1.77, + "learning_rate": 1.8139705484582005e-05, + "loss": 0.3712, + "step": 4230 + }, + { + "epoch": 1.77, + "learning_rate": 1.813872049417053e-05, + "loss": 0.3634, + "step": 4231 + }, + { + "epoch": 1.77, + "learning_rate": 1.813773526981875e-05, + "loss": 0.4128, + "step": 4232 + }, + { + "epoch": 1.77, + "learning_rate": 1.8136749811554995e-05, + "loss": 0.3902, + "step": 4233 + }, + { + "epoch": 1.77, + "learning_rate": 1.8135764119407585e-05, + "loss": 0.3534, + "step": 4234 + }, + { + "epoch": 1.77, + "learning_rate": 1.8134778193404857e-05, + "loss": 0.3815, + "step": 4235 + }, + { + "epoch": 1.77, + "learning_rate": 1.813379203357515e-05, + "loss": 0.3744, + "step": 4236 + }, + { + "epoch": 1.77, + "learning_rate": 1.8132805639946806e-05, + "loss": 0.3964, + "step": 4237 + }, + { + "epoch": 1.77, + "learning_rate": 1.8131819012548183e-05, + "loss": 0.3992, + "step": 4238 + }, + { + "epoch": 1.77, + "learning_rate": 1.8130832151407634e-05, + "loss": 0.3965, + "step": 4239 + }, + { + "epoch": 1.78, + "learning_rate": 1.8129845056553534e-05, + "loss": 0.4237, + "step": 4240 + }, + { + "epoch": 1.78, + "learning_rate": 1.812885772801425e-05, + "loss": 0.3852, + "step": 4241 + }, + { + "epoch": 1.78, + "learning_rate": 1.8127870165818166e-05, + "loss": 0.4264, + "step": 4242 + }, + { + "epoch": 1.78, + "learning_rate": 1.812688236999367e-05, + "loss": 0.4305, + "step": 4243 + }, + { + "epoch": 1.78, + "learning_rate": 1.8125894340569145e-05, + "loss": 0.3854, + "step": 4244 + }, + { + "epoch": 1.78, + "learning_rate": 1.8124906077573e-05, + "loss": 0.3823, + "step": 4245 + }, + { + "epoch": 1.78, + "learning_rate": 1.8123917581033642e-05, + "loss": 0.3893, + "step": 4246 + }, + { + "epoch": 1.78, + "learning_rate": 1.8122928850979483e-05, + "loss": 0.4558, + "step": 4247 + }, + { + "epoch": 1.78, + "learning_rate": 1.8121939887438936e-05, + "loss": 0.3923, + "step": 4248 + }, + { + "epoch": 1.78, + "learning_rate": 1.812095069044044e-05, + "loss": 0.37, + "step": 4249 + }, + { + "epoch": 1.78, + "learning_rate": 1.8119961260012422e-05, + "loss": 0.423, + "step": 4250 + }, + { + "epoch": 1.78, + "learning_rate": 1.811897159618332e-05, + "loss": 0.3789, + "step": 4251 + }, + { + "epoch": 1.78, + "learning_rate": 1.811798169898159e-05, + "loss": 0.3585, + "step": 4252 + }, + { + "epoch": 1.78, + "learning_rate": 1.8116991568435674e-05, + "loss": 0.3785, + "step": 4253 + }, + { + "epoch": 1.78, + "learning_rate": 1.8116001204574042e-05, + "loss": 0.4106, + "step": 4254 + }, + { + "epoch": 1.78, + "learning_rate": 1.8115010607425154e-05, + "loss": 0.4085, + "step": 4255 + }, + { + "epoch": 1.78, + "learning_rate": 1.811401977701749e-05, + "loss": 0.45, + "step": 4256 + }, + { + "epoch": 1.78, + "learning_rate": 1.8113028713379524e-05, + "loss": 0.4488, + "step": 4257 + }, + { + "epoch": 1.78, + "learning_rate": 1.811203741653975e-05, + "loss": 0.409, + "step": 4258 + }, + { + "epoch": 1.78, + "learning_rate": 1.811104588652666e-05, + "loss": 0.3769, + "step": 4259 + }, + { + "epoch": 1.78, + "learning_rate": 1.8110054123368755e-05, + "loss": 0.3857, + "step": 4260 + }, + { + "epoch": 1.78, + "learning_rate": 1.8109062127094538e-05, + "loss": 0.4303, + "step": 4261 + }, + { + "epoch": 1.78, + "learning_rate": 1.8108069897732527e-05, + "loss": 0.4154, + "step": 4262 + }, + { + "epoch": 1.78, + "learning_rate": 1.810707743531124e-05, + "loss": 0.3739, + "step": 4263 + }, + { + "epoch": 1.79, + "learning_rate": 1.8106084739859206e-05, + "loss": 0.3749, + "step": 4264 + }, + { + "epoch": 1.79, + "learning_rate": 1.8105091811404962e-05, + "loss": 0.4192, + "step": 4265 + }, + { + "epoch": 1.79, + "learning_rate": 1.8104098649977044e-05, + "loss": 0.4326, + "step": 4266 + }, + { + "epoch": 1.79, + "learning_rate": 1.8103105255604003e-05, + "loss": 0.4022, + "step": 4267 + }, + { + "epoch": 1.79, + "learning_rate": 1.8102111628314394e-05, + "loss": 0.4064, + "step": 4268 + }, + { + "epoch": 1.79, + "learning_rate": 1.810111776813677e-05, + "loss": 0.3597, + "step": 4269 + }, + { + "epoch": 1.79, + "learning_rate": 1.810012367509971e-05, + "loss": 0.397, + "step": 4270 + }, + { + "epoch": 1.79, + "learning_rate": 1.8099129349231783e-05, + "loss": 0.4079, + "step": 4271 + }, + { + "epoch": 1.79, + "learning_rate": 1.8098134790561567e-05, + "loss": 0.3914, + "step": 4272 + }, + { + "epoch": 1.79, + "learning_rate": 1.8097139999117655e-05, + "loss": 0.3927, + "step": 4273 + }, + { + "epoch": 1.79, + "learning_rate": 1.8096144974928635e-05, + "loss": 0.4111, + "step": 4274 + }, + { + "epoch": 1.79, + "learning_rate": 1.8095149718023114e-05, + "loss": 0.4026, + "step": 4275 + }, + { + "epoch": 1.79, + "learning_rate": 1.80941542284297e-05, + "loss": 0.3648, + "step": 4276 + }, + { + "epoch": 1.79, + "learning_rate": 1.8093158506177003e-05, + "loss": 0.3546, + "step": 4277 + }, + { + "epoch": 1.79, + "learning_rate": 1.8092162551293646e-05, + "loss": 0.3978, + "step": 4278 + }, + { + "epoch": 1.79, + "learning_rate": 1.8091166363808258e-05, + "loss": 0.3781, + "step": 4279 + }, + { + "epoch": 1.79, + "learning_rate": 1.8090169943749477e-05, + "loss": 0.3416, + "step": 4280 + }, + { + "epoch": 1.79, + "learning_rate": 1.8089173291145935e-05, + "loss": 0.3681, + "step": 4281 + }, + { + "epoch": 1.79, + "learning_rate": 1.808817640602629e-05, + "loss": 0.402, + "step": 4282 + }, + { + "epoch": 1.79, + "learning_rate": 1.8087179288419182e-05, + "loss": 0.4024, + "step": 4283 + }, + { + "epoch": 1.79, + "learning_rate": 1.808618193835329e-05, + "loss": 0.3581, + "step": 4284 + }, + { + "epoch": 1.79, + "learning_rate": 1.808518435585727e-05, + "loss": 0.4094, + "step": 4285 + }, + { + "epoch": 1.79, + "learning_rate": 1.80841865409598e-05, + "loss": 0.3766, + "step": 4286 + }, + { + "epoch": 1.79, + "learning_rate": 1.808318849368957e-05, + "loss": 0.4021, + "step": 4287 + }, + { + "epoch": 1.8, + "learning_rate": 1.8082190214075252e-05, + "loss": 0.3986, + "step": 4288 + }, + { + "epoch": 1.8, + "learning_rate": 1.8081191702145547e-05, + "loss": 0.397, + "step": 4289 + }, + { + "epoch": 1.8, + "learning_rate": 1.8080192957929162e-05, + "loss": 0.3872, + "step": 4290 + }, + { + "epoch": 1.8, + "learning_rate": 1.8079193981454798e-05, + "loss": 0.3809, + "step": 4291 + }, + { + "epoch": 1.8, + "learning_rate": 1.8078194772751173e-05, + "loss": 0.4337, + "step": 4292 + }, + { + "epoch": 1.8, + "learning_rate": 1.8077195331847007e-05, + "loss": 0.4027, + "step": 4293 + }, + { + "epoch": 1.8, + "learning_rate": 1.8076195658771033e-05, + "loss": 0.3748, + "step": 4294 + }, + { + "epoch": 1.8, + "learning_rate": 1.8075195753551974e-05, + "loss": 0.3974, + "step": 4295 + }, + { + "epoch": 1.8, + "learning_rate": 1.8074195616218583e-05, + "loss": 0.3615, + "step": 4296 + }, + { + "epoch": 1.8, + "learning_rate": 1.8073195246799605e-05, + "loss": 0.375, + "step": 4297 + }, + { + "epoch": 1.8, + "learning_rate": 1.8072194645323794e-05, + "loss": 0.449, + "step": 4298 + }, + { + "epoch": 1.8, + "learning_rate": 1.8071193811819912e-05, + "loss": 0.3788, + "step": 4299 + }, + { + "epoch": 1.8, + "learning_rate": 1.807019274631672e-05, + "loss": 0.3921, + "step": 4300 + }, + { + "epoch": 1.8, + "learning_rate": 1.8069191448843005e-05, + "loss": 0.3985, + "step": 4301 + }, + { + "epoch": 1.8, + "learning_rate": 1.806818991942754e-05, + "loss": 0.3922, + "step": 4302 + }, + { + "epoch": 1.8, + "learning_rate": 1.8067188158099112e-05, + "loss": 0.367, + "step": 4303 + }, + { + "epoch": 1.8, + "learning_rate": 1.8066186164886525e-05, + "loss": 0.4072, + "step": 4304 + }, + { + "epoch": 1.8, + "learning_rate": 1.806518393981857e-05, + "loss": 0.3879, + "step": 4305 + }, + { + "epoch": 1.8, + "learning_rate": 1.8064181482924065e-05, + "loss": 0.377, + "step": 4306 + }, + { + "epoch": 1.8, + "learning_rate": 1.8063178794231814e-05, + "loss": 0.4025, + "step": 4307 + }, + { + "epoch": 1.8, + "learning_rate": 1.8062175873770643e-05, + "loss": 0.3922, + "step": 4308 + }, + { + "epoch": 1.8, + "learning_rate": 1.8061172721569383e-05, + "loss": 0.4284, + "step": 4309 + }, + { + "epoch": 1.8, + "learning_rate": 1.8060169337656866e-05, + "loss": 0.4754, + "step": 4310 + }, + { + "epoch": 1.8, + "learning_rate": 1.805916572206194e-05, + "loss": 0.3792, + "step": 4311 + }, + { + "epoch": 1.81, + "learning_rate": 1.805816187481344e-05, + "loss": 0.374, + "step": 4312 + }, + { + "epoch": 1.81, + "learning_rate": 1.8057157795940228e-05, + "loss": 0.4356, + "step": 4313 + }, + { + "epoch": 1.81, + "learning_rate": 1.8056153485471167e-05, + "loss": 0.3891, + "step": 4314 + }, + { + "epoch": 1.81, + "learning_rate": 1.805514894343512e-05, + "loss": 0.3978, + "step": 4315 + }, + { + "epoch": 1.81, + "learning_rate": 1.805414416986097e-05, + "loss": 0.411, + "step": 4316 + }, + { + "epoch": 1.81, + "learning_rate": 1.8053139164777587e-05, + "loss": 0.3439, + "step": 4317 + }, + { + "epoch": 1.81, + "learning_rate": 1.805213392821387e-05, + "loss": 0.42, + "step": 4318 + }, + { + "epoch": 1.81, + "learning_rate": 1.8051128460198703e-05, + "loss": 0.3687, + "step": 4319 + }, + { + "epoch": 1.81, + "learning_rate": 1.8050122760760994e-05, + "loss": 0.3926, + "step": 4320 + }, + { + "epoch": 1.81, + "learning_rate": 1.8049116829929655e-05, + "loss": 0.3792, + "step": 4321 + }, + { + "epoch": 1.81, + "learning_rate": 1.804811066773359e-05, + "loss": 0.3436, + "step": 4322 + }, + { + "epoch": 1.81, + "learning_rate": 1.804710427420173e-05, + "loss": 0.386, + "step": 4323 + }, + { + "epoch": 1.81, + "learning_rate": 1.8046097649362994e-05, + "loss": 0.3816, + "step": 4324 + }, + { + "epoch": 1.81, + "learning_rate": 1.804509079324632e-05, + "loss": 0.3676, + "step": 4325 + }, + { + "epoch": 1.81, + "learning_rate": 1.804408370588066e-05, + "loss": 0.3695, + "step": 4326 + }, + { + "epoch": 1.81, + "learning_rate": 1.8043076387294944e-05, + "loss": 0.4537, + "step": 4327 + }, + { + "epoch": 1.81, + "learning_rate": 1.8042068837518135e-05, + "loss": 0.3724, + "step": 4328 + }, + { + "epoch": 1.81, + "learning_rate": 1.8041061056579194e-05, + "loss": 0.3988, + "step": 4329 + }, + { + "epoch": 1.81, + "learning_rate": 1.8040053044507087e-05, + "loss": 0.3737, + "step": 4330 + }, + { + "epoch": 1.81, + "learning_rate": 1.8039044801330795e-05, + "loss": 0.4418, + "step": 4331 + }, + { + "epoch": 1.81, + "learning_rate": 1.803803632707929e-05, + "loss": 0.3556, + "step": 4332 + }, + { + "epoch": 1.81, + "learning_rate": 1.8037027621781568e-05, + "loss": 0.3917, + "step": 4333 + }, + { + "epoch": 1.81, + "learning_rate": 1.8036018685466614e-05, + "loss": 0.4307, + "step": 4334 + }, + { + "epoch": 1.81, + "learning_rate": 1.8035009518163434e-05, + "loss": 0.4482, + "step": 4335 + }, + { + "epoch": 1.82, + "learning_rate": 1.8034000119901038e-05, + "loss": 0.4407, + "step": 4336 + }, + { + "epoch": 1.82, + "learning_rate": 1.8032990490708434e-05, + "loss": 0.3863, + "step": 4337 + }, + { + "epoch": 1.82, + "learning_rate": 1.8031980630614652e-05, + "loss": 0.3908, + "step": 4338 + }, + { + "epoch": 1.82, + "learning_rate": 1.803097053964871e-05, + "loss": 0.3528, + "step": 4339 + }, + { + "epoch": 1.82, + "learning_rate": 1.8029960217839648e-05, + "loss": 0.3768, + "step": 4340 + }, + { + "epoch": 1.82, + "learning_rate": 1.8028949665216505e-05, + "loss": 0.4015, + "step": 4341 + }, + { + "epoch": 1.82, + "learning_rate": 1.802793888180833e-05, + "loss": 0.4411, + "step": 4342 + }, + { + "epoch": 1.82, + "learning_rate": 1.8026927867644175e-05, + "loss": 0.4535, + "step": 4343 + }, + { + "epoch": 1.82, + "learning_rate": 1.80259166227531e-05, + "loss": 0.3384, + "step": 4344 + }, + { + "epoch": 1.82, + "learning_rate": 1.8024905147164176e-05, + "loss": 0.3852, + "step": 4345 + }, + { + "epoch": 1.82, + "learning_rate": 1.8023893440906476e-05, + "loss": 0.3559, + "step": 4346 + }, + { + "epoch": 1.82, + "learning_rate": 1.8022881504009074e-05, + "loss": 0.4396, + "step": 4347 + }, + { + "epoch": 1.82, + "learning_rate": 1.8021869336501066e-05, + "loss": 0.3677, + "step": 4348 + }, + { + "epoch": 1.82, + "learning_rate": 1.802085693841154e-05, + "loss": 0.3632, + "step": 4349 + }, + { + "epoch": 1.82, + "learning_rate": 1.8019844309769602e-05, + "loss": 0.4229, + "step": 4350 + }, + { + "epoch": 1.82, + "learning_rate": 1.8018831450604356e-05, + "loss": 0.4812, + "step": 4351 + }, + { + "epoch": 1.82, + "learning_rate": 1.801781836094491e-05, + "loss": 0.387, + "step": 4352 + }, + { + "epoch": 1.82, + "learning_rate": 1.80168050408204e-05, + "loss": 0.3763, + "step": 4353 + }, + { + "epoch": 1.82, + "learning_rate": 1.8015791490259932e-05, + "loss": 0.3933, + "step": 4354 + }, + { + "epoch": 1.82, + "learning_rate": 1.801477770929266e-05, + "loss": 0.4103, + "step": 4355 + }, + { + "epoch": 1.82, + "learning_rate": 1.801376369794771e-05, + "loss": 0.4138, + "step": 4356 + }, + { + "epoch": 1.82, + "learning_rate": 1.8012749456254238e-05, + "loss": 0.4028, + "step": 4357 + }, + { + "epoch": 1.82, + "learning_rate": 1.801173498424139e-05, + "loss": 0.3912, + "step": 4358 + }, + { + "epoch": 1.82, + "learning_rate": 1.8010720281938327e-05, + "loss": 0.429, + "step": 4359 + }, + { + "epoch": 1.83, + "learning_rate": 1.800970534937422e-05, + "loss": 0.3638, + "step": 4360 + }, + { + "epoch": 1.83, + "learning_rate": 1.8008690186578243e-05, + "loss": 0.4019, + "step": 4361 + }, + { + "epoch": 1.83, + "learning_rate": 1.800767479357957e-05, + "loss": 0.3711, + "step": 4362 + }, + { + "epoch": 1.83, + "learning_rate": 1.8006659170407396e-05, + "loss": 0.3859, + "step": 4363 + }, + { + "epoch": 1.83, + "learning_rate": 1.80056433170909e-05, + "loss": 0.441, + "step": 4364 + }, + { + "epoch": 1.83, + "learning_rate": 1.80046272336593e-05, + "loss": 0.375, + "step": 4365 + }, + { + "epoch": 1.83, + "learning_rate": 1.800361092014179e-05, + "loss": 0.4191, + "step": 4366 + }, + { + "epoch": 1.83, + "learning_rate": 1.8002594376567585e-05, + "loss": 0.3702, + "step": 4367 + }, + { + "epoch": 1.83, + "learning_rate": 1.8001577602965907e-05, + "loss": 0.4165, + "step": 4368 + }, + { + "epoch": 1.83, + "learning_rate": 1.800056059936598e-05, + "loss": 0.4246, + "step": 4369 + }, + { + "epoch": 1.83, + "learning_rate": 1.7999543365797043e-05, + "loss": 0.4047, + "step": 4370 + }, + { + "epoch": 1.83, + "learning_rate": 1.7998525902288327e-05, + "loss": 0.3736, + "step": 4371 + }, + { + "epoch": 1.83, + "learning_rate": 1.7997508208869082e-05, + "loss": 0.4108, + "step": 4372 + }, + { + "epoch": 1.83, + "learning_rate": 1.799649028556856e-05, + "loss": 0.3926, + "step": 4373 + }, + { + "epoch": 1.83, + "learning_rate": 1.799547213241602e-05, + "loss": 0.403, + "step": 4374 + }, + { + "epoch": 1.83, + "learning_rate": 1.799445374944073e-05, + "loss": 0.3477, + "step": 4375 + }, + { + "epoch": 1.83, + "learning_rate": 1.799343513667196e-05, + "loss": 0.4046, + "step": 4376 + }, + { + "epoch": 1.83, + "learning_rate": 1.7992416294138992e-05, + "loss": 0.3881, + "step": 4377 + }, + { + "epoch": 1.83, + "learning_rate": 1.799139722187111e-05, + "loss": 0.4091, + "step": 4378 + }, + { + "epoch": 1.83, + "learning_rate": 1.7990377919897605e-05, + "loss": 0.399, + "step": 4379 + }, + { + "epoch": 1.83, + "learning_rate": 1.7989358388247776e-05, + "loss": 0.4123, + "step": 4380 + }, + { + "epoch": 1.83, + "learning_rate": 1.798833862695093e-05, + "loss": 0.4001, + "step": 4381 + }, + { + "epoch": 1.83, + "learning_rate": 1.7987318636036382e-05, + "loss": 0.3674, + "step": 4382 + }, + { + "epoch": 1.84, + "learning_rate": 1.7986298415533444e-05, + "loss": 0.3603, + "step": 4383 + }, + { + "epoch": 1.84, + "learning_rate": 1.798527796547145e-05, + "loss": 0.4306, + "step": 4384 + }, + { + "epoch": 1.84, + "learning_rate": 1.7984257285879724e-05, + "loss": 0.3837, + "step": 4385 + }, + { + "epoch": 1.84, + "learning_rate": 1.7983236376787607e-05, + "loss": 0.4104, + "step": 4386 + }, + { + "epoch": 1.84, + "learning_rate": 1.7982215238224445e-05, + "loss": 0.368, + "step": 4387 + }, + { + "epoch": 1.84, + "learning_rate": 1.7981193870219587e-05, + "loss": 0.3786, + "step": 4388 + }, + { + "epoch": 1.84, + "learning_rate": 1.7980172272802398e-05, + "loss": 0.4624, + "step": 4389 + }, + { + "epoch": 1.84, + "learning_rate": 1.7979150446002233e-05, + "loss": 0.4058, + "step": 4390 + }, + { + "epoch": 1.84, + "learning_rate": 1.7978128389848476e-05, + "loss": 0.3636, + "step": 4391 + }, + { + "epoch": 1.84, + "learning_rate": 1.7977106104370493e-05, + "loss": 0.3949, + "step": 4392 + }, + { + "epoch": 1.84, + "learning_rate": 1.7976083589597672e-05, + "loss": 0.3636, + "step": 4393 + }, + { + "epoch": 1.84, + "learning_rate": 1.7975060845559412e-05, + "loss": 0.3717, + "step": 4394 + }, + { + "epoch": 1.84, + "learning_rate": 1.7974037872285102e-05, + "loss": 0.4207, + "step": 4395 + }, + { + "epoch": 1.84, + "learning_rate": 1.797301466980415e-05, + "loss": 0.3758, + "step": 4396 + }, + { + "epoch": 1.84, + "learning_rate": 1.7971991238145965e-05, + "loss": 0.3979, + "step": 4397 + }, + { + "epoch": 1.84, + "learning_rate": 1.7970967577339965e-05, + "loss": 0.4325, + "step": 4398 + }, + { + "epoch": 1.84, + "learning_rate": 1.7969943687415575e-05, + "loss": 0.4123, + "step": 4399 + }, + { + "epoch": 1.84, + "learning_rate": 1.7968919568402234e-05, + "loss": 0.3432, + "step": 4400 + }, + { + "epoch": 1.84, + "learning_rate": 1.7967895220329364e-05, + "loss": 0.3518, + "step": 4401 + }, + { + "epoch": 1.84, + "learning_rate": 1.796687064322642e-05, + "loss": 0.4224, + "step": 4402 + }, + { + "epoch": 1.84, + "learning_rate": 1.7965845837122844e-05, + "loss": 0.3905, + "step": 4403 + }, + { + "epoch": 1.84, + "learning_rate": 1.79648208020481e-05, + "loss": 0.3709, + "step": 4404 + }, + { + "epoch": 1.84, + "learning_rate": 1.796379553803165e-05, + "loss": 0.4077, + "step": 4405 + }, + { + "epoch": 1.84, + "learning_rate": 1.7962770045102968e-05, + "loss": 0.3806, + "step": 4406 + }, + { + "epoch": 1.85, + "learning_rate": 1.7961744323291525e-05, + "loss": 0.3653, + "step": 4407 + }, + { + "epoch": 1.85, + "learning_rate": 1.7960718372626807e-05, + "loss": 0.435, + "step": 4408 + }, + { + "epoch": 1.85, + "learning_rate": 1.7959692193138303e-05, + "loss": 0.3866, + "step": 4409 + }, + { + "epoch": 1.85, + "learning_rate": 1.795866578485551e-05, + "loss": 0.4393, + "step": 4410 + }, + { + "epoch": 1.85, + "learning_rate": 1.795763914780793e-05, + "loss": 0.4025, + "step": 4411 + }, + { + "epoch": 1.85, + "learning_rate": 1.7956612282025078e-05, + "loss": 0.3961, + "step": 4412 + }, + { + "epoch": 1.85, + "learning_rate": 1.7955585187536463e-05, + "loss": 0.4066, + "step": 4413 + }, + { + "epoch": 1.85, + "learning_rate": 1.7954557864371615e-05, + "loss": 0.3567, + "step": 4414 + }, + { + "epoch": 1.85, + "learning_rate": 1.7953530312560056e-05, + "loss": 0.4295, + "step": 4415 + }, + { + "epoch": 1.85, + "learning_rate": 1.7952502532131326e-05, + "loss": 0.3641, + "step": 4416 + }, + { + "epoch": 1.85, + "learning_rate": 1.795147452311497e-05, + "loss": 0.4159, + "step": 4417 + }, + { + "epoch": 1.85, + "learning_rate": 1.7950446285540532e-05, + "loss": 0.3475, + "step": 4418 + }, + { + "epoch": 1.85, + "learning_rate": 1.7949417819437577e-05, + "loss": 0.3599, + "step": 4419 + }, + { + "epoch": 1.85, + "learning_rate": 1.7948389124835654e-05, + "loss": 0.3755, + "step": 4420 + }, + { + "epoch": 1.85, + "learning_rate": 1.794736020176434e-05, + "loss": 0.4222, + "step": 4421 + }, + { + "epoch": 1.85, + "learning_rate": 1.794633105025321e-05, + "loss": 0.3815, + "step": 4422 + }, + { + "epoch": 1.85, + "learning_rate": 1.7945301670331847e-05, + "loss": 0.4236, + "step": 4423 + }, + { + "epoch": 1.85, + "learning_rate": 1.794427206202984e-05, + "loss": 0.3942, + "step": 4424 + }, + { + "epoch": 1.85, + "learning_rate": 1.7943242225376778e-05, + "loss": 0.3716, + "step": 4425 + }, + { + "epoch": 1.85, + "learning_rate": 1.794221216040227e-05, + "loss": 0.3625, + "step": 4426 + }, + { + "epoch": 1.85, + "learning_rate": 1.7941181867135917e-05, + "loss": 0.4135, + "step": 4427 + }, + { + "epoch": 1.85, + "learning_rate": 1.7940151345607343e-05, + "loss": 0.3409, + "step": 4428 + }, + { + "epoch": 1.85, + "learning_rate": 1.793912059584616e-05, + "loss": 0.3957, + "step": 4429 + }, + { + "epoch": 1.85, + "learning_rate": 1.7938089617882004e-05, + "loss": 0.3973, + "step": 4430 + }, + { + "epoch": 1.86, + "learning_rate": 1.7937058411744504e-05, + "loss": 0.4159, + "step": 4431 + }, + { + "epoch": 1.86, + "learning_rate": 1.7936026977463306e-05, + "loss": 0.3604, + "step": 4432 + }, + { + "epoch": 1.86, + "learning_rate": 1.793499531506805e-05, + "loss": 0.3864, + "step": 4433 + }, + { + "epoch": 1.86, + "learning_rate": 1.79339634245884e-05, + "loss": 0.4238, + "step": 4434 + }, + { + "epoch": 1.86, + "learning_rate": 1.7932931306054007e-05, + "loss": 0.3839, + "step": 4435 + }, + { + "epoch": 1.86, + "learning_rate": 1.7931898959494546e-05, + "loss": 0.3724, + "step": 4436 + }, + { + "epoch": 1.86, + "learning_rate": 1.793086638493969e-05, + "loss": 0.383, + "step": 4437 + }, + { + "epoch": 1.86, + "learning_rate": 1.7929833582419115e-05, + "loss": 0.4443, + "step": 4438 + }, + { + "epoch": 1.86, + "learning_rate": 1.792880055196251e-05, + "loss": 0.4123, + "step": 4439 + }, + { + "epoch": 1.86, + "learning_rate": 1.792776729359957e-05, + "loss": 0.3807, + "step": 4440 + }, + { + "epoch": 1.86, + "learning_rate": 1.7926733807359993e-05, + "loss": 0.4245, + "step": 4441 + }, + { + "epoch": 1.86, + "learning_rate": 1.792570009327349e-05, + "loss": 0.3401, + "step": 4442 + }, + { + "epoch": 1.86, + "learning_rate": 1.792466615136977e-05, + "loss": 0.4053, + "step": 4443 + }, + { + "epoch": 1.86, + "learning_rate": 1.7923631981678552e-05, + "loss": 0.4292, + "step": 4444 + }, + { + "epoch": 1.86, + "learning_rate": 1.7922597584229566e-05, + "loss": 0.3729, + "step": 4445 + }, + { + "epoch": 1.86, + "learning_rate": 1.792156295905254e-05, + "loss": 0.4087, + "step": 4446 + }, + { + "epoch": 1.86, + "learning_rate": 1.792052810617722e-05, + "loss": 0.3385, + "step": 4447 + }, + { + "epoch": 1.86, + "learning_rate": 1.7919493025633342e-05, + "loss": 0.4045, + "step": 4448 + }, + { + "epoch": 1.86, + "learning_rate": 1.791845771745067e-05, + "loss": 0.3663, + "step": 4449 + }, + { + "epoch": 1.86, + "learning_rate": 1.7917422181658954e-05, + "loss": 0.4042, + "step": 4450 + }, + { + "epoch": 1.86, + "learning_rate": 1.7916386418287968e-05, + "loss": 0.3809, + "step": 4451 + }, + { + "epoch": 1.86, + "learning_rate": 1.7915350427367476e-05, + "loss": 0.4189, + "step": 4452 + }, + { + "epoch": 1.86, + "learning_rate": 1.791431420892726e-05, + "loss": 0.3534, + "step": 4453 + }, + { + "epoch": 1.86, + "learning_rate": 1.7913277762997105e-05, + "loss": 0.3893, + "step": 4454 + }, + { + "epoch": 1.87, + "learning_rate": 1.7912241089606802e-05, + "loss": 0.3716, + "step": 4455 + }, + { + "epoch": 1.87, + "learning_rate": 1.7911204188786154e-05, + "loss": 0.4082, + "step": 4456 + }, + { + "epoch": 1.87, + "learning_rate": 1.791016706056496e-05, + "loss": 0.3839, + "step": 4457 + }, + { + "epoch": 1.87, + "learning_rate": 1.790912970497303e-05, + "loss": 0.4122, + "step": 4458 + }, + { + "epoch": 1.87, + "learning_rate": 1.790809212204019e-05, + "loss": 0.3856, + "step": 4459 + }, + { + "epoch": 1.87, + "learning_rate": 1.7907054311796254e-05, + "loss": 0.4127, + "step": 4460 + }, + { + "epoch": 1.87, + "learning_rate": 1.7906016274271062e-05, + "loss": 0.3855, + "step": 4461 + }, + { + "epoch": 1.87, + "learning_rate": 1.790497800949445e-05, + "loss": 0.4138, + "step": 4462 + }, + { + "epoch": 1.87, + "learning_rate": 1.7903939517496253e-05, + "loss": 0.4138, + "step": 4463 + }, + { + "epoch": 1.87, + "learning_rate": 1.7902900798306334e-05, + "loss": 0.3514, + "step": 4464 + }, + { + "epoch": 1.87, + "learning_rate": 1.7901861851954543e-05, + "loss": 0.3967, + "step": 4465 + }, + { + "epoch": 1.87, + "learning_rate": 1.7900822678470745e-05, + "loss": 0.4038, + "step": 4466 + }, + { + "epoch": 1.87, + "learning_rate": 1.789978327788481e-05, + "loss": 0.3741, + "step": 4467 + }, + { + "epoch": 1.87, + "learning_rate": 1.7898743650226617e-05, + "loss": 0.4083, + "step": 4468 + }, + { + "epoch": 1.87, + "learning_rate": 1.7897703795526044e-05, + "loss": 0.3672, + "step": 4469 + }, + { + "epoch": 1.87, + "learning_rate": 1.7896663713812984e-05, + "loss": 0.4927, + "step": 4470 + }, + { + "epoch": 1.87, + "learning_rate": 1.7895623405117334e-05, + "loss": 0.4271, + "step": 4471 + }, + { + "epoch": 1.87, + "learning_rate": 1.7894582869468994e-05, + "loss": 0.3796, + "step": 4472 + }, + { + "epoch": 1.87, + "learning_rate": 1.7893542106897878e-05, + "loss": 0.4172, + "step": 4473 + }, + { + "epoch": 1.87, + "learning_rate": 1.7892501117433896e-05, + "loss": 0.3696, + "step": 4474 + }, + { + "epoch": 1.87, + "learning_rate": 1.7891459901106976e-05, + "loss": 0.3844, + "step": 4475 + }, + { + "epoch": 1.87, + "learning_rate": 1.7890418457947038e-05, + "loss": 0.3763, + "step": 4476 + }, + { + "epoch": 1.87, + "learning_rate": 1.788937678798403e-05, + "loss": 0.3572, + "step": 4477 + }, + { + "epoch": 1.87, + "learning_rate": 1.7888334891247883e-05, + "loss": 0.3744, + "step": 4478 + }, + { + "epoch": 1.88, + "learning_rate": 1.788729276776855e-05, + "loss": 0.3865, + "step": 4479 + }, + { + "epoch": 1.88, + "learning_rate": 1.7886250417575986e-05, + "loss": 0.4056, + "step": 4480 + }, + { + "epoch": 1.88, + "learning_rate": 1.7885207840700155e-05, + "loss": 0.4122, + "step": 4481 + }, + { + "epoch": 1.88, + "learning_rate": 1.7884165037171016e-05, + "loss": 0.3764, + "step": 4482 + }, + { + "epoch": 1.88, + "learning_rate": 1.7883122007018554e-05, + "loss": 0.3989, + "step": 4483 + }, + { + "epoch": 1.88, + "learning_rate": 1.788207875027274e-05, + "loss": 0.4024, + "step": 4484 + }, + { + "epoch": 1.88, + "learning_rate": 1.7881035266963575e-05, + "loss": 0.3932, + "step": 4485 + }, + { + "epoch": 1.88, + "learning_rate": 1.787999155712104e-05, + "loss": 0.3933, + "step": 4486 + }, + { + "epoch": 1.88, + "learning_rate": 1.787894762077514e-05, + "loss": 0.3786, + "step": 4487 + }, + { + "epoch": 1.88, + "learning_rate": 1.787790345795588e-05, + "loss": 0.4072, + "step": 4488 + }, + { + "epoch": 1.88, + "learning_rate": 1.787685906869328e-05, + "loss": 0.4105, + "step": 4489 + }, + { + "epoch": 1.88, + "learning_rate": 1.7875814453017355e-05, + "loss": 0.3618, + "step": 4490 + }, + { + "epoch": 1.88, + "learning_rate": 1.787476961095813e-05, + "loss": 0.4214, + "step": 4491 + }, + { + "epoch": 1.88, + "learning_rate": 1.7873724542545647e-05, + "loss": 0.3606, + "step": 4492 + }, + { + "epoch": 1.88, + "learning_rate": 1.7872679247809933e-05, + "loss": 0.397, + "step": 4493 + }, + { + "epoch": 1.88, + "learning_rate": 1.7871633726781042e-05, + "loss": 0.4022, + "step": 4494 + }, + { + "epoch": 1.88, + "learning_rate": 1.787058797948902e-05, + "loss": 0.3629, + "step": 4495 + }, + { + "epoch": 1.88, + "learning_rate": 1.786954200596394e-05, + "loss": 0.427, + "step": 4496 + }, + { + "epoch": 1.88, + "learning_rate": 1.786849580623585e-05, + "loss": 0.3878, + "step": 4497 + }, + { + "epoch": 1.88, + "learning_rate": 1.7867449380334834e-05, + "loss": 0.4738, + "step": 4498 + }, + { + "epoch": 1.88, + "learning_rate": 1.7866402728290967e-05, + "loss": 0.3804, + "step": 4499 + }, + { + "epoch": 1.88, + "learning_rate": 1.786535585013433e-05, + "loss": 0.4011, + "step": 4500 + }, + { + "epoch": 1.88, + "learning_rate": 1.7864308745895024e-05, + "loss": 0.4079, + "step": 4501 + }, + { + "epoch": 1.88, + "learning_rate": 1.7863261415603142e-05, + "loss": 0.3485, + "step": 4502 + }, + { + "epoch": 1.89, + "learning_rate": 1.7862213859288788e-05, + "loss": 0.4219, + "step": 4503 + }, + { + "epoch": 1.89, + "learning_rate": 1.786116607698207e-05, + "loss": 0.3978, + "step": 4504 + }, + { + "epoch": 1.89, + "learning_rate": 1.786011806871311e-05, + "loss": 0.3497, + "step": 4505 + }, + { + "epoch": 1.89, + "learning_rate": 1.7859069834512033e-05, + "loss": 0.354, + "step": 4506 + }, + { + "epoch": 1.89, + "learning_rate": 1.7858021374408968e-05, + "loss": 0.3959, + "step": 4507 + }, + { + "epoch": 1.89, + "learning_rate": 1.785697268843405e-05, + "loss": 0.375, + "step": 4508 + }, + { + "epoch": 1.89, + "learning_rate": 1.7855923776617424e-05, + "loss": 0.3681, + "step": 4509 + }, + { + "epoch": 1.89, + "learning_rate": 1.785487463898924e-05, + "loss": 0.4029, + "step": 4510 + }, + { + "epoch": 1.89, + "learning_rate": 1.7853825275579656e-05, + "loss": 0.4052, + "step": 4511 + }, + { + "epoch": 1.89, + "learning_rate": 1.7852775686418832e-05, + "loss": 0.4151, + "step": 4512 + }, + { + "epoch": 1.89, + "learning_rate": 1.7851725871536945e-05, + "loss": 0.4123, + "step": 4513 + }, + { + "epoch": 1.89, + "learning_rate": 1.785067583096416e-05, + "loss": 0.3862, + "step": 4514 + }, + { + "epoch": 1.89, + "learning_rate": 1.7849625564730667e-05, + "loss": 0.3823, + "step": 4515 + }, + { + "epoch": 1.89, + "learning_rate": 1.784857507286665e-05, + "loss": 0.3856, + "step": 4516 + }, + { + "epoch": 1.89, + "learning_rate": 1.784752435540231e-05, + "loss": 0.3882, + "step": 4517 + }, + { + "epoch": 1.89, + "learning_rate": 1.7846473412367845e-05, + "loss": 0.4147, + "step": 4518 + }, + { + "epoch": 1.89, + "learning_rate": 1.7845422243793464e-05, + "loss": 0.4482, + "step": 4519 + }, + { + "epoch": 1.89, + "learning_rate": 1.784437084970938e-05, + "loss": 0.3736, + "step": 4520 + }, + { + "epoch": 1.89, + "learning_rate": 1.7843319230145822e-05, + "loss": 0.3869, + "step": 4521 + }, + { + "epoch": 1.89, + "learning_rate": 1.7842267385133013e-05, + "loss": 0.4484, + "step": 4522 + }, + { + "epoch": 1.89, + "learning_rate": 1.7841215314701182e-05, + "loss": 0.4245, + "step": 4523 + }, + { + "epoch": 1.89, + "learning_rate": 1.784016301888058e-05, + "loss": 0.3619, + "step": 4524 + }, + { + "epoch": 1.89, + "learning_rate": 1.7839110497701446e-05, + "loss": 0.4076, + "step": 4525 + }, + { + "epoch": 1.89, + "learning_rate": 1.783805775119404e-05, + "loss": 0.4029, + "step": 4526 + }, + { + "epoch": 1.9, + "learning_rate": 1.7837004779388615e-05, + "loss": 0.4297, + "step": 4527 + }, + { + "epoch": 1.9, + "learning_rate": 1.7835951582315445e-05, + "loss": 0.3741, + "step": 4528 + }, + { + "epoch": 1.9, + "learning_rate": 1.78348981600048e-05, + "loss": 0.4124, + "step": 4529 + }, + { + "epoch": 1.9, + "learning_rate": 1.783384451248696e-05, + "loss": 0.3873, + "step": 4530 + }, + { + "epoch": 1.9, + "learning_rate": 1.783279063979221e-05, + "loss": 0.3949, + "step": 4531 + }, + { + "epoch": 1.9, + "learning_rate": 1.7831736541950845e-05, + "loss": 0.399, + "step": 4532 + }, + { + "epoch": 1.9, + "learning_rate": 1.783068221899316e-05, + "loss": 0.4076, + "step": 4533 + }, + { + "epoch": 1.9, + "learning_rate": 1.7829627670949468e-05, + "loss": 0.4031, + "step": 4534 + }, + { + "epoch": 1.9, + "learning_rate": 1.7828572897850076e-05, + "loss": 0.3656, + "step": 4535 + }, + { + "epoch": 1.9, + "learning_rate": 1.78275178997253e-05, + "loss": 0.3846, + "step": 4536 + }, + { + "epoch": 1.9, + "learning_rate": 1.782646267660547e-05, + "loss": 0.409, + "step": 4537 + }, + { + "epoch": 1.9, + "learning_rate": 1.7825407228520917e-05, + "loss": 0.3568, + "step": 4538 + }, + { + "epoch": 1.9, + "learning_rate": 1.7824351555501972e-05, + "loss": 0.4116, + "step": 4539 + }, + { + "epoch": 1.9, + "learning_rate": 1.782329565757899e-05, + "loss": 0.4143, + "step": 4540 + }, + { + "epoch": 1.9, + "learning_rate": 1.7822239534782312e-05, + "loss": 0.3885, + "step": 4541 + }, + { + "epoch": 1.9, + "learning_rate": 1.7821183187142302e-05, + "loss": 0.4098, + "step": 4542 + }, + { + "epoch": 1.9, + "learning_rate": 1.7820126614689326e-05, + "loss": 0.4534, + "step": 4543 + }, + { + "epoch": 1.9, + "learning_rate": 1.7819069817453746e-05, + "loss": 0.3467, + "step": 4544 + }, + { + "epoch": 1.9, + "learning_rate": 1.781801279546594e-05, + "loss": 0.408, + "step": 4545 + }, + { + "epoch": 1.9, + "learning_rate": 1.7816955548756298e-05, + "loss": 0.3673, + "step": 4546 + }, + { + "epoch": 1.9, + "learning_rate": 1.7815898077355204e-05, + "loss": 0.4112, + "step": 4547 + }, + { + "epoch": 1.9, + "learning_rate": 1.7814840381293054e-05, + "loss": 0.3596, + "step": 4548 + }, + { + "epoch": 1.9, + "learning_rate": 1.7813782460600257e-05, + "loss": 0.3993, + "step": 4549 + }, + { + "epoch": 1.9, + "learning_rate": 1.781272431530721e-05, + "loss": 0.3849, + "step": 4550 + }, + { + "epoch": 1.91, + "learning_rate": 1.7811665945444337e-05, + "loss": 0.3657, + "step": 4551 + }, + { + "epoch": 1.91, + "learning_rate": 1.7810607351042062e-05, + "loss": 0.3464, + "step": 4552 + }, + { + "epoch": 1.91, + "learning_rate": 1.7809548532130808e-05, + "loss": 0.3813, + "step": 4553 + }, + { + "epoch": 1.91, + "learning_rate": 1.7808489488741013e-05, + "loss": 0.4107, + "step": 4554 + }, + { + "epoch": 1.91, + "learning_rate": 1.7807430220903113e-05, + "loss": 0.319, + "step": 4555 + }, + { + "epoch": 1.91, + "learning_rate": 1.780637072864756e-05, + "loss": 0.3817, + "step": 4556 + }, + { + "epoch": 1.91, + "learning_rate": 1.7805311012004812e-05, + "loss": 0.36, + "step": 4557 + }, + { + "epoch": 1.91, + "learning_rate": 1.7804251071005323e-05, + "loss": 0.4005, + "step": 4558 + }, + { + "epoch": 1.91, + "learning_rate": 1.7803190905679562e-05, + "loss": 0.3722, + "step": 4559 + }, + { + "epoch": 1.91, + "learning_rate": 1.7802130516058003e-05, + "loss": 0.4129, + "step": 4560 + }, + { + "epoch": 1.91, + "learning_rate": 1.7801069902171123e-05, + "loss": 0.3759, + "step": 4561 + }, + { + "epoch": 1.91, + "learning_rate": 1.7800009064049416e-05, + "loss": 0.414, + "step": 4562 + }, + { + "epoch": 1.91, + "learning_rate": 1.7798948001723368e-05, + "loss": 0.4868, + "step": 4563 + }, + { + "epoch": 1.91, + "learning_rate": 1.779788671522348e-05, + "loss": 0.3889, + "step": 4564 + }, + { + "epoch": 1.91, + "learning_rate": 1.779682520458026e-05, + "loss": 0.3491, + "step": 4565 + }, + { + "epoch": 1.91, + "learning_rate": 1.779576346982421e-05, + "loss": 0.4432, + "step": 4566 + }, + { + "epoch": 1.91, + "learning_rate": 1.7794701510985867e-05, + "loss": 0.3424, + "step": 4567 + }, + { + "epoch": 1.91, + "learning_rate": 1.779363932809574e-05, + "loss": 0.3825, + "step": 4568 + }, + { + "epoch": 1.91, + "learning_rate": 1.7792576921184374e-05, + "loss": 0.3551, + "step": 4569 + }, + { + "epoch": 1.91, + "learning_rate": 1.7791514290282293e-05, + "loss": 0.3705, + "step": 4570 + }, + { + "epoch": 1.91, + "learning_rate": 1.7790451435420046e-05, + "loss": 0.3369, + "step": 4571 + }, + { + "epoch": 1.91, + "learning_rate": 1.778938835662819e-05, + "loss": 0.402, + "step": 4572 + }, + { + "epoch": 1.91, + "learning_rate": 1.7788325053937278e-05, + "loss": 0.3878, + "step": 4573 + }, + { + "epoch": 1.92, + "learning_rate": 1.778726152737787e-05, + "loss": 0.4031, + "step": 4574 + }, + { + "epoch": 1.92, + "learning_rate": 1.7786197776980542e-05, + "loss": 0.384, + "step": 4575 + }, + { + "epoch": 1.92, + "learning_rate": 1.778513380277587e-05, + "loss": 0.3849, + "step": 4576 + }, + { + "epoch": 1.92, + "learning_rate": 1.7784069604794433e-05, + "loss": 0.3563, + "step": 4577 + }, + { + "epoch": 1.92, + "learning_rate": 1.778300518306682e-05, + "loss": 0.4222, + "step": 4578 + }, + { + "epoch": 1.92, + "learning_rate": 1.7781940537623636e-05, + "loss": 0.3769, + "step": 4579 + }, + { + "epoch": 1.92, + "learning_rate": 1.778087566849547e-05, + "loss": 0.3815, + "step": 4580 + }, + { + "epoch": 1.92, + "learning_rate": 1.7779810575712942e-05, + "loss": 0.3749, + "step": 4581 + }, + { + "epoch": 1.92, + "learning_rate": 1.777874525930666e-05, + "loss": 0.3943, + "step": 4582 + }, + { + "epoch": 1.92, + "learning_rate": 1.7777679719307253e-05, + "loss": 0.3639, + "step": 4583 + }, + { + "epoch": 1.92, + "learning_rate": 1.7776613955745337e-05, + "loss": 0.3701, + "step": 4584 + }, + { + "epoch": 1.92, + "learning_rate": 1.7775547968651562e-05, + "loss": 0.3833, + "step": 4585 + }, + { + "epoch": 1.92, + "learning_rate": 1.7774481758056553e-05, + "loss": 0.3607, + "step": 4586 + }, + { + "epoch": 1.92, + "learning_rate": 1.777341532399097e-05, + "loss": 0.3518, + "step": 4587 + }, + { + "epoch": 1.92, + "learning_rate": 1.777234866648546e-05, + "loss": 0.4178, + "step": 4588 + }, + { + "epoch": 1.92, + "learning_rate": 1.7771281785570683e-05, + "loss": 0.4101, + "step": 4589 + }, + { + "epoch": 1.92, + "learning_rate": 1.777021468127731e-05, + "loss": 0.3854, + "step": 4590 + }, + { + "epoch": 1.92, + "learning_rate": 1.776914735363601e-05, + "loss": 0.3975, + "step": 4591 + }, + { + "epoch": 1.92, + "learning_rate": 1.776807980267746e-05, + "loss": 0.409, + "step": 4592 + }, + { + "epoch": 1.92, + "learning_rate": 1.7767012028432352e-05, + "loss": 0.4173, + "step": 4593 + }, + { + "epoch": 1.92, + "learning_rate": 1.776594403093138e-05, + "loss": 0.3405, + "step": 4594 + }, + { + "epoch": 1.92, + "learning_rate": 1.7764875810205234e-05, + "loss": 0.3914, + "step": 4595 + }, + { + "epoch": 1.92, + "learning_rate": 1.776380736628462e-05, + "loss": 0.3719, + "step": 4596 + }, + { + "epoch": 1.92, + "learning_rate": 1.7762738699200256e-05, + "loss": 0.4169, + "step": 4597 + }, + { + "epoch": 1.93, + "learning_rate": 1.7761669808982857e-05, + "loss": 0.4051, + "step": 4598 + }, + { + "epoch": 1.93, + "learning_rate": 1.776060069566315e-05, + "loss": 0.3898, + "step": 4599 + }, + { + "epoch": 1.93, + "learning_rate": 1.7759531359271857e-05, + "loss": 0.3679, + "step": 4600 + }, + { + "epoch": 1.93, + "learning_rate": 1.7758461799839723e-05, + "loss": 0.3402, + "step": 4601 + }, + { + "epoch": 1.93, + "learning_rate": 1.7757392017397493e-05, + "loss": 0.391, + "step": 4602 + }, + { + "epoch": 1.93, + "learning_rate": 1.7756322011975906e-05, + "loss": 0.3498, + "step": 4603 + }, + { + "epoch": 1.93, + "learning_rate": 1.7755251783605733e-05, + "loss": 0.3862, + "step": 4604 + }, + { + "epoch": 1.93, + "learning_rate": 1.7754181332317725e-05, + "loss": 0.4389, + "step": 4605 + }, + { + "epoch": 1.93, + "learning_rate": 1.7753110658142657e-05, + "loss": 0.3772, + "step": 4606 + }, + { + "epoch": 1.93, + "learning_rate": 1.77520397611113e-05, + "loss": 0.4095, + "step": 4607 + }, + { + "epoch": 1.93, + "learning_rate": 1.7750968641254444e-05, + "loss": 0.4164, + "step": 4608 + }, + { + "epoch": 1.93, + "learning_rate": 1.7749897298602867e-05, + "loss": 0.4049, + "step": 4609 + }, + { + "epoch": 1.93, + "learning_rate": 1.7748825733187372e-05, + "loss": 0.3771, + "step": 4610 + }, + { + "epoch": 1.93, + "learning_rate": 1.774775394503876e-05, + "loss": 0.3696, + "step": 4611 + }, + { + "epoch": 1.93, + "learning_rate": 1.7746681934187832e-05, + "loss": 0.3992, + "step": 4612 + }, + { + "epoch": 1.93, + "learning_rate": 1.7745609700665406e-05, + "loss": 0.3612, + "step": 4613 + }, + { + "epoch": 1.93, + "learning_rate": 1.774453724450231e-05, + "loss": 0.3978, + "step": 4614 + }, + { + "epoch": 1.93, + "learning_rate": 1.7743464565729354e-05, + "loss": 0.4378, + "step": 4615 + }, + { + "epoch": 1.93, + "learning_rate": 1.7742391664377384e-05, + "loss": 0.3375, + "step": 4616 + }, + { + "epoch": 1.93, + "learning_rate": 1.7741318540477234e-05, + "loss": 0.4127, + "step": 4617 + }, + { + "epoch": 1.93, + "learning_rate": 1.7740245194059754e-05, + "loss": 0.3915, + "step": 4618 + }, + { + "epoch": 1.93, + "learning_rate": 1.77391716251558e-05, + "loss": 0.3586, + "step": 4619 + }, + { + "epoch": 1.93, + "learning_rate": 1.7738097833796218e-05, + "loss": 0.3378, + "step": 4620 + }, + { + "epoch": 1.93, + "learning_rate": 1.7737023820011884e-05, + "loss": 0.4145, + "step": 4621 + }, + { + "epoch": 1.94, + "learning_rate": 1.7735949583833662e-05, + "loss": 0.4131, + "step": 4622 + }, + { + "epoch": 1.94, + "learning_rate": 1.7734875125292436e-05, + "loss": 0.3654, + "step": 4623 + }, + { + "epoch": 1.94, + "learning_rate": 1.7733800444419088e-05, + "loss": 0.3907, + "step": 4624 + }, + { + "epoch": 1.94, + "learning_rate": 1.773272554124451e-05, + "loss": 0.4214, + "step": 4625 + }, + { + "epoch": 1.94, + "learning_rate": 1.77316504157996e-05, + "loss": 0.3629, + "step": 4626 + }, + { + "epoch": 1.94, + "learning_rate": 1.773057506811526e-05, + "loss": 0.3772, + "step": 4627 + }, + { + "epoch": 1.94, + "learning_rate": 1.7729499498222392e-05, + "loss": 0.4109, + "step": 4628 + }, + { + "epoch": 1.94, + "learning_rate": 1.7728423706151928e-05, + "loss": 0.36, + "step": 4629 + }, + { + "epoch": 1.94, + "learning_rate": 1.772734769193478e-05, + "loss": 0.436, + "step": 4630 + }, + { + "epoch": 1.94, + "learning_rate": 1.772627145560188e-05, + "loss": 0.3648, + "step": 4631 + }, + { + "epoch": 1.94, + "learning_rate": 1.772519499718416e-05, + "loss": 0.4065, + "step": 4632 + }, + { + "epoch": 1.94, + "learning_rate": 1.772411831671257e-05, + "loss": 0.387, + "step": 4633 + }, + { + "epoch": 1.94, + "learning_rate": 1.7723041414218053e-05, + "loss": 0.3432, + "step": 4634 + }, + { + "epoch": 1.94, + "learning_rate": 1.7721964289731562e-05, + "loss": 0.4029, + "step": 4635 + }, + { + "epoch": 1.94, + "learning_rate": 1.772088694328406e-05, + "loss": 0.3698, + "step": 4636 + }, + { + "epoch": 1.94, + "learning_rate": 1.771980937490651e-05, + "loss": 0.4266, + "step": 4637 + }, + { + "epoch": 1.94, + "learning_rate": 1.7718731584629897e-05, + "loss": 0.3723, + "step": 4638 + }, + { + "epoch": 1.94, + "learning_rate": 1.771765357248519e-05, + "loss": 0.3856, + "step": 4639 + }, + { + "epoch": 1.94, + "learning_rate": 1.7716575338503382e-05, + "loss": 0.3812, + "step": 4640 + }, + { + "epoch": 1.94, + "learning_rate": 1.7715496882715463e-05, + "loss": 0.353, + "step": 4641 + }, + { + "epoch": 1.94, + "learning_rate": 1.7714418205152434e-05, + "loss": 0.4068, + "step": 4642 + }, + { + "epoch": 1.94, + "learning_rate": 1.7713339305845293e-05, + "loss": 0.4075, + "step": 4643 + }, + { + "epoch": 1.94, + "learning_rate": 1.7712260184825063e-05, + "loss": 0.3755, + "step": 4644 + }, + { + "epoch": 1.94, + "learning_rate": 1.771118084212276e-05, + "loss": 0.4403, + "step": 4645 + }, + { + "epoch": 1.95, + "learning_rate": 1.7710101277769403e-05, + "loss": 0.4173, + "step": 4646 + }, + { + "epoch": 1.95, + "learning_rate": 1.7709021491796027e-05, + "loss": 0.3637, + "step": 4647 + }, + { + "epoch": 1.95, + "learning_rate": 1.770794148423367e-05, + "loss": 0.4326, + "step": 4648 + }, + { + "epoch": 1.95, + "learning_rate": 1.7706861255113376e-05, + "loss": 0.3747, + "step": 4649 + }, + { + "epoch": 1.95, + "learning_rate": 1.7705780804466192e-05, + "loss": 0.3416, + "step": 4650 + }, + { + "epoch": 1.95, + "learning_rate": 1.7704700132323175e-05, + "loss": 0.3899, + "step": 4651 + }, + { + "epoch": 1.95, + "learning_rate": 1.7703619238715392e-05, + "loss": 0.4293, + "step": 4652 + }, + { + "epoch": 1.95, + "learning_rate": 1.770253812367391e-05, + "loss": 0.4505, + "step": 4653 + }, + { + "epoch": 1.95, + "learning_rate": 1.7701456787229805e-05, + "loss": 0.3799, + "step": 4654 + }, + { + "epoch": 1.95, + "learning_rate": 1.770037522941416e-05, + "loss": 0.3909, + "step": 4655 + }, + { + "epoch": 1.95, + "learning_rate": 1.7699293450258062e-05, + "loss": 0.4465, + "step": 4656 + }, + { + "epoch": 1.95, + "learning_rate": 1.7698211449792604e-05, + "loss": 0.3893, + "step": 4657 + }, + { + "epoch": 1.95, + "learning_rate": 1.769712922804889e-05, + "loss": 0.3736, + "step": 4658 + }, + { + "epoch": 1.95, + "learning_rate": 1.7696046785058024e-05, + "loss": 0.3389, + "step": 4659 + }, + { + "epoch": 1.95, + "learning_rate": 1.7694964120851128e-05, + "loss": 0.4159, + "step": 4660 + }, + { + "epoch": 1.95, + "learning_rate": 1.7693881235459315e-05, + "loss": 0.391, + "step": 4661 + }, + { + "epoch": 1.95, + "learning_rate": 1.7692798128913714e-05, + "loss": 0.3976, + "step": 4662 + }, + { + "epoch": 1.95, + "learning_rate": 1.7691714801245453e-05, + "loss": 0.371, + "step": 4663 + }, + { + "epoch": 1.95, + "learning_rate": 1.7690631252485676e-05, + "loss": 0.3931, + "step": 4664 + }, + { + "epoch": 1.95, + "learning_rate": 1.7689547482665532e-05, + "loss": 0.3735, + "step": 4665 + }, + { + "epoch": 1.95, + "learning_rate": 1.768846349181617e-05, + "loss": 0.4045, + "step": 4666 + }, + { + "epoch": 1.95, + "learning_rate": 1.7687379279968745e-05, + "loss": 0.4001, + "step": 4667 + }, + { + "epoch": 1.95, + "learning_rate": 1.7686294847154423e-05, + "loss": 0.3443, + "step": 4668 + }, + { + "epoch": 1.95, + "learning_rate": 1.7685210193404377e-05, + "loss": 0.4312, + "step": 4669 + }, + { + "epoch": 1.96, + "learning_rate": 1.7684125318749785e-05, + "loss": 0.3976, + "step": 4670 + }, + { + "epoch": 1.96, + "learning_rate": 1.7683040223221828e-05, + "loss": 0.4214, + "step": 4671 + }, + { + "epoch": 1.96, + "learning_rate": 1.76819549068517e-05, + "loss": 0.3772, + "step": 4672 + }, + { + "epoch": 1.96, + "learning_rate": 1.768086936967059e-05, + "loss": 0.3983, + "step": 4673 + }, + { + "epoch": 1.96, + "learning_rate": 1.767978361170971e-05, + "loss": 0.4451, + "step": 4674 + }, + { + "epoch": 1.96, + "learning_rate": 1.7678697633000267e-05, + "loss": 0.3974, + "step": 4675 + }, + { + "epoch": 1.96, + "learning_rate": 1.767761143357347e-05, + "loss": 0.3598, + "step": 4676 + }, + { + "epoch": 1.96, + "learning_rate": 1.7676525013460547e-05, + "loss": 0.3909, + "step": 4677 + }, + { + "epoch": 1.96, + "learning_rate": 1.7675438372692726e-05, + "loss": 0.4153, + "step": 4678 + }, + { + "epoch": 1.96, + "learning_rate": 1.7674351511301238e-05, + "loss": 0.3957, + "step": 4679 + }, + { + "epoch": 1.96, + "learning_rate": 1.767326442931733e-05, + "loss": 0.3571, + "step": 4680 + }, + { + "epoch": 1.96, + "learning_rate": 1.7672177126772237e-05, + "loss": 0.3678, + "step": 4681 + }, + { + "epoch": 1.96, + "learning_rate": 1.7671089603697228e-05, + "loss": 0.4258, + "step": 4682 + }, + { + "epoch": 1.96, + "learning_rate": 1.767000186012355e-05, + "loss": 0.4408, + "step": 4683 + }, + { + "epoch": 1.96, + "learning_rate": 1.766891389608248e-05, + "loss": 0.409, + "step": 4684 + }, + { + "epoch": 1.96, + "learning_rate": 1.7667825711605283e-05, + "loss": 0.4514, + "step": 4685 + }, + { + "epoch": 1.96, + "learning_rate": 1.766673730672324e-05, + "loss": 0.3947, + "step": 4686 + }, + { + "epoch": 1.96, + "learning_rate": 1.766564868146764e-05, + "loss": 0.349, + "step": 4687 + }, + { + "epoch": 1.96, + "learning_rate": 1.7664559835869763e-05, + "loss": 0.4218, + "step": 4688 + }, + { + "epoch": 1.96, + "learning_rate": 1.766347076996092e-05, + "loss": 0.4156, + "step": 4689 + }, + { + "epoch": 1.96, + "learning_rate": 1.7662381483772412e-05, + "loss": 0.3596, + "step": 4690 + }, + { + "epoch": 1.96, + "learning_rate": 1.7661291977335544e-05, + "loss": 0.4368, + "step": 4691 + }, + { + "epoch": 1.96, + "learning_rate": 1.766020225068164e-05, + "loss": 0.3567, + "step": 4692 + }, + { + "epoch": 1.96, + "learning_rate": 1.7659112303842015e-05, + "loss": 0.3661, + "step": 4693 + }, + { + "epoch": 1.97, + "learning_rate": 1.7658022136848008e-05, + "loss": 0.3629, + "step": 4694 + }, + { + "epoch": 1.97, + "learning_rate": 1.7656931749730947e-05, + "loss": 0.3877, + "step": 4695 + }, + { + "epoch": 1.97, + "learning_rate": 1.765584114252218e-05, + "loss": 0.3517, + "step": 4696 + }, + { + "epoch": 1.97, + "learning_rate": 1.7654750315253055e-05, + "loss": 0.3632, + "step": 4697 + }, + { + "epoch": 1.97, + "learning_rate": 1.765365926795492e-05, + "loss": 0.3851, + "step": 4698 + }, + { + "epoch": 1.97, + "learning_rate": 1.7652568000659144e-05, + "loss": 0.4156, + "step": 4699 + }, + { + "epoch": 1.97, + "learning_rate": 1.765147651339709e-05, + "loss": 0.3623, + "step": 4700 + }, + { + "epoch": 1.97, + "learning_rate": 1.7650384806200134e-05, + "loss": 0.4243, + "step": 4701 + }, + { + "epoch": 1.97, + "learning_rate": 1.7649292879099655e-05, + "loss": 0.3677, + "step": 4702 + }, + { + "epoch": 1.97, + "learning_rate": 1.764820073212704e-05, + "loss": 0.4255, + "step": 4703 + }, + { + "epoch": 1.97, + "learning_rate": 1.7647108365313683e-05, + "loss": 0.4099, + "step": 4704 + }, + { + "epoch": 1.97, + "learning_rate": 1.7646015778690983e-05, + "loss": 0.4019, + "step": 4705 + }, + { + "epoch": 1.97, + "learning_rate": 1.7644922972290344e-05, + "loss": 0.4247, + "step": 4706 + }, + { + "epoch": 1.97, + "learning_rate": 1.7643829946143176e-05, + "loss": 0.4158, + "step": 4707 + }, + { + "epoch": 1.97, + "learning_rate": 1.76427367002809e-05, + "loss": 0.354, + "step": 4708 + }, + { + "epoch": 1.97, + "learning_rate": 1.764164323473494e-05, + "loss": 0.4356, + "step": 4709 + }, + { + "epoch": 1.97, + "learning_rate": 1.7640549549536727e-05, + "loss": 0.38, + "step": 4710 + }, + { + "epoch": 1.97, + "learning_rate": 1.7639455644717694e-05, + "loss": 0.3709, + "step": 4711 + }, + { + "epoch": 1.97, + "learning_rate": 1.7638361520309294e-05, + "loss": 0.4102, + "step": 4712 + }, + { + "epoch": 1.97, + "learning_rate": 1.7637267176342964e-05, + "loss": 0.3607, + "step": 4713 + }, + { + "epoch": 1.97, + "learning_rate": 1.763617261285017e-05, + "loss": 0.417, + "step": 4714 + }, + { + "epoch": 1.97, + "learning_rate": 1.763507782986237e-05, + "loss": 0.3558, + "step": 4715 + }, + { + "epoch": 1.97, + "learning_rate": 1.763398282741103e-05, + "loss": 0.381, + "step": 4716 + }, + { + "epoch": 1.97, + "learning_rate": 1.763288760552763e-05, + "loss": 0.3835, + "step": 4717 + }, + { + "epoch": 1.98, + "learning_rate": 1.763179216424365e-05, + "loss": 0.4098, + "step": 4718 + }, + { + "epoch": 1.98, + "learning_rate": 1.7630696503590577e-05, + "loss": 0.3972, + "step": 4719 + }, + { + "epoch": 1.98, + "learning_rate": 1.7629600623599904e-05, + "loss": 0.4298, + "step": 4720 + }, + { + "epoch": 1.98, + "learning_rate": 1.7628504524303128e-05, + "loss": 0.382, + "step": 4721 + }, + { + "epoch": 1.98, + "learning_rate": 1.7627408205731762e-05, + "loss": 0.43, + "step": 4722 + }, + { + "epoch": 1.98, + "learning_rate": 1.7626311667917315e-05, + "loss": 0.3753, + "step": 4723 + }, + { + "epoch": 1.98, + "learning_rate": 1.7625214910891307e-05, + "loss": 0.4242, + "step": 4724 + }, + { + "epoch": 1.98, + "learning_rate": 1.7624117934685264e-05, + "loss": 0.4101, + "step": 4725 + }, + { + "epoch": 1.98, + "learning_rate": 1.7623020739330714e-05, + "loss": 0.3907, + "step": 4726 + }, + { + "epoch": 1.98, + "learning_rate": 1.7621923324859196e-05, + "loss": 0.3705, + "step": 4727 + }, + { + "epoch": 1.98, + "learning_rate": 1.762082569130226e-05, + "loss": 0.3592, + "step": 4728 + }, + { + "epoch": 1.98, + "learning_rate": 1.761972783869145e-05, + "loss": 0.401, + "step": 4729 + }, + { + "epoch": 1.98, + "learning_rate": 1.7618629767058326e-05, + "loss": 0.4288, + "step": 4730 + }, + { + "epoch": 1.98, + "learning_rate": 1.7617531476434447e-05, + "loss": 0.3446, + "step": 4731 + }, + { + "epoch": 1.98, + "learning_rate": 1.7616432966851386e-05, + "loss": 0.3834, + "step": 4732 + }, + { + "epoch": 1.98, + "learning_rate": 1.761533423834072e-05, + "loss": 0.4065, + "step": 4733 + }, + { + "epoch": 1.98, + "learning_rate": 1.761423529093403e-05, + "loss": 0.4022, + "step": 4734 + }, + { + "epoch": 1.98, + "learning_rate": 1.76131361246629e-05, + "loss": 0.3807, + "step": 4735 + }, + { + "epoch": 1.98, + "learning_rate": 1.7612036739558932e-05, + "loss": 0.3898, + "step": 4736 + }, + { + "epoch": 1.98, + "learning_rate": 1.761093713565372e-05, + "loss": 0.3873, + "step": 4737 + }, + { + "epoch": 1.98, + "learning_rate": 1.7609837312978874e-05, + "loss": 0.4046, + "step": 4738 + }, + { + "epoch": 1.98, + "learning_rate": 1.7608737271566004e-05, + "loss": 0.4069, + "step": 4739 + }, + { + "epoch": 1.98, + "learning_rate": 1.7607637011446737e-05, + "loss": 0.3928, + "step": 4740 + }, + { + "epoch": 1.98, + "learning_rate": 1.7606536532652695e-05, + "loss": 0.3691, + "step": 4741 + }, + { + "epoch": 1.99, + "learning_rate": 1.760543583521551e-05, + "loss": 0.4312, + "step": 4742 + }, + { + "epoch": 1.99, + "learning_rate": 1.7604334919166822e-05, + "loss": 0.3853, + "step": 4743 + }, + { + "epoch": 1.99, + "learning_rate": 1.7603233784538275e-05, + "loss": 0.4136, + "step": 4744 + }, + { + "epoch": 1.99, + "learning_rate": 1.7602132431361516e-05, + "loss": 0.4116, + "step": 4745 + }, + { + "epoch": 1.99, + "learning_rate": 1.7601030859668207e-05, + "loss": 0.3834, + "step": 4746 + }, + { + "epoch": 1.99, + "learning_rate": 1.7599929069490014e-05, + "loss": 0.4323, + "step": 4747 + }, + { + "epoch": 1.99, + "learning_rate": 1.7598827060858605e-05, + "loss": 0.4095, + "step": 4748 + }, + { + "epoch": 1.99, + "learning_rate": 1.759772483380565e-05, + "loss": 0.4095, + "step": 4749 + }, + { + "epoch": 1.99, + "learning_rate": 1.759662238836284e-05, + "loss": 0.4272, + "step": 4750 + }, + { + "epoch": 1.99, + "learning_rate": 1.759551972456186e-05, + "loss": 0.405, + "step": 4751 + }, + { + "epoch": 1.99, + "learning_rate": 1.7594416842434405e-05, + "loss": 0.3806, + "step": 4752 + }, + { + "epoch": 1.99, + "learning_rate": 1.759331374201218e-05, + "loss": 0.3747, + "step": 4753 + }, + { + "epoch": 1.99, + "learning_rate": 1.759221042332689e-05, + "loss": 0.3804, + "step": 4754 + }, + { + "epoch": 1.99, + "learning_rate": 1.759110688641024e-05, + "loss": 0.4165, + "step": 4755 + }, + { + "epoch": 1.99, + "learning_rate": 1.7590003131293967e-05, + "loss": 0.3855, + "step": 4756 + }, + { + "epoch": 1.99, + "learning_rate": 1.7588899158009786e-05, + "loss": 0.373, + "step": 4757 + }, + { + "epoch": 1.99, + "learning_rate": 1.7587794966589437e-05, + "loss": 0.3976, + "step": 4758 + }, + { + "epoch": 1.99, + "learning_rate": 1.758669055706465e-05, + "loss": 0.3288, + "step": 4759 + }, + { + "epoch": 1.99, + "learning_rate": 1.758558592946718e-05, + "loss": 0.3985, + "step": 4760 + }, + { + "epoch": 1.99, + "learning_rate": 1.7584481083828766e-05, + "loss": 0.4002, + "step": 4761 + }, + { + "epoch": 1.99, + "learning_rate": 1.7583376020181183e-05, + "loss": 0.3729, + "step": 4762 + }, + { + "epoch": 1.99, + "learning_rate": 1.758227073855618e-05, + "loss": 0.4247, + "step": 4763 + }, + { + "epoch": 1.99, + "learning_rate": 1.7581165238985533e-05, + "loss": 0.4128, + "step": 4764 + }, + { + "epoch": 1.99, + "learning_rate": 1.7580059521501017e-05, + "loss": 0.3545, + "step": 4765 + }, + { + "epoch": 2.0, + "learning_rate": 1.7578953586134423e-05, + "loss": 0.3787, + "step": 4766 + }, + { + "epoch": 2.0, + "learning_rate": 1.7577847432917528e-05, + "loss": 0.3957, + "step": 4767 + }, + { + "epoch": 2.0, + "learning_rate": 1.7576741061882136e-05, + "loss": 0.3975, + "step": 4768 + }, + { + "epoch": 2.0, + "learning_rate": 1.7575634473060045e-05, + "loss": 0.3689, + "step": 4769 + }, + { + "epoch": 2.0, + "learning_rate": 1.757452766648306e-05, + "loss": 0.4133, + "step": 4770 + }, + { + "epoch": 2.0, + "learning_rate": 1.7573420642183004e-05, + "loss": 0.3487, + "step": 4771 + }, + { + "epoch": 2.0, + "learning_rate": 1.757231340019169e-05, + "loss": 0.3746, + "step": 4772 + }, + { + "epoch": 2.0, + "learning_rate": 1.7571205940540945e-05, + "loss": 0.4144, + "step": 4773 + }, + { + "epoch": 2.0, + "learning_rate": 1.7570098263262608e-05, + "loss": 0.3845, + "step": 4774 + }, + { + "epoch": 2.0, + "learning_rate": 1.756899036838851e-05, + "loss": 0.4326, + "step": 4775 + }, + { + "epoch": 2.0, + "learning_rate": 1.7567882255950503e-05, + "loss": 0.3347, + "step": 4776 + }, + { + "epoch": 2.0, + "learning_rate": 1.7566773925980435e-05, + "loss": 0.3743, + "step": 4777 + }, + { + "epoch": 2.0, + "learning_rate": 1.7565665378510165e-05, + "loss": 0.2799, + "step": 4778 + }, + { + "epoch": 2.0, + "learning_rate": 1.756455661357156e-05, + "loss": 0.315, + "step": 4779 + }, + { + "epoch": 2.0, + "learning_rate": 1.7563447631196484e-05, + "loss": 0.2846, + "step": 4780 + }, + { + "epoch": 2.0, + "learning_rate": 1.7562338431416818e-05, + "loss": 0.262, + "step": 4781 + }, + { + "epoch": 2.0, + "learning_rate": 1.7561229014264448e-05, + "loss": 0.2669, + "step": 4782 + }, + { + "epoch": 2.0, + "learning_rate": 1.7560119379771257e-05, + "loss": 0.2785, + "step": 4783 + }, + { + "epoch": 2.0, + "learning_rate": 1.7559009527969145e-05, + "loss": 0.27, + "step": 4784 + }, + { + "epoch": 2.0, + "learning_rate": 1.755789945889001e-05, + "loss": 0.2426, + "step": 4785 + }, + { + "epoch": 2.0, + "learning_rate": 1.7556789172565762e-05, + "loss": 0.266, + "step": 4786 + }, + { + "epoch": 2.0, + "learning_rate": 1.7555678669028315e-05, + "loss": 0.242, + "step": 4787 + }, + { + "epoch": 2.0, + "learning_rate": 1.755456794830959e-05, + "loss": 0.246, + "step": 4788 + }, + { + "epoch": 2.01, + "learning_rate": 1.7553457010441514e-05, + "loss": 0.2788, + "step": 4789 + }, + { + "epoch": 2.01, + "learning_rate": 1.7552345855456017e-05, + "loss": 0.2595, + "step": 4790 + }, + { + "epoch": 2.01, + "learning_rate": 1.755123448338504e-05, + "loss": 0.2886, + "step": 4791 + }, + { + "epoch": 2.01, + "learning_rate": 1.755012289426053e-05, + "loss": 0.2787, + "step": 4792 + }, + { + "epoch": 2.01, + "learning_rate": 1.754901108811444e-05, + "loss": 0.2442, + "step": 4793 + }, + { + "epoch": 2.01, + "learning_rate": 1.754789906497872e-05, + "loss": 0.2656, + "step": 4794 + }, + { + "epoch": 2.01, + "learning_rate": 1.754678682488534e-05, + "loss": 0.2701, + "step": 4795 + }, + { + "epoch": 2.01, + "learning_rate": 1.7545674367866273e-05, + "loss": 0.2597, + "step": 4796 + }, + { + "epoch": 2.01, + "learning_rate": 1.754456169395349e-05, + "loss": 0.2762, + "step": 4797 + }, + { + "epoch": 2.01, + "learning_rate": 1.7543448803178975e-05, + "loss": 0.2363, + "step": 4798 + }, + { + "epoch": 2.01, + "learning_rate": 1.754233569557472e-05, + "loss": 0.2452, + "step": 4799 + }, + { + "epoch": 2.01, + "learning_rate": 1.7541222371172715e-05, + "loss": 0.2096, + "step": 4800 + }, + { + "epoch": 2.01, + "learning_rate": 1.754010883000497e-05, + "loss": 0.264, + "step": 4801 + }, + { + "epoch": 2.01, + "learning_rate": 1.7538995072103485e-05, + "loss": 0.2442, + "step": 4802 + }, + { + "epoch": 2.01, + "learning_rate": 1.7537881097500277e-05, + "loss": 0.2588, + "step": 4803 + }, + { + "epoch": 2.01, + "learning_rate": 1.7536766906227365e-05, + "loss": 0.2637, + "step": 4804 + }, + { + "epoch": 2.01, + "learning_rate": 1.753565249831678e-05, + "loss": 0.255, + "step": 4805 + }, + { + "epoch": 2.01, + "learning_rate": 1.7534537873800544e-05, + "loss": 0.2544, + "step": 4806 + }, + { + "epoch": 2.01, + "learning_rate": 1.7533423032710705e-05, + "loss": 0.2545, + "step": 4807 + }, + { + "epoch": 2.01, + "learning_rate": 1.753230797507931e-05, + "loss": 0.2724, + "step": 4808 + }, + { + "epoch": 2.01, + "learning_rate": 1.7531192700938406e-05, + "loss": 0.2539, + "step": 4809 + }, + { + "epoch": 2.01, + "learning_rate": 1.7530077210320047e-05, + "loss": 0.2681, + "step": 4810 + }, + { + "epoch": 2.01, + "learning_rate": 1.7528961503256302e-05, + "loss": 0.2636, + "step": 4811 + }, + { + "epoch": 2.01, + "learning_rate": 1.752784557977924e-05, + "loss": 0.2572, + "step": 4812 + }, + { + "epoch": 2.02, + "learning_rate": 1.752672943992094e-05, + "loss": 0.27, + "step": 4813 + }, + { + "epoch": 2.02, + "learning_rate": 1.752561308371348e-05, + "loss": 0.2464, + "step": 4814 + }, + { + "epoch": 2.02, + "learning_rate": 1.7524496511188948e-05, + "loss": 0.2508, + "step": 4815 + }, + { + "epoch": 2.02, + "learning_rate": 1.752337972237944e-05, + "loss": 0.2722, + "step": 4816 + }, + { + "epoch": 2.02, + "learning_rate": 1.752226271731706e-05, + "loss": 0.2406, + "step": 4817 + }, + { + "epoch": 2.02, + "learning_rate": 1.7521145496033912e-05, + "loss": 0.258, + "step": 4818 + }, + { + "epoch": 2.02, + "learning_rate": 1.752002805856211e-05, + "loss": 0.2664, + "step": 4819 + }, + { + "epoch": 2.02, + "learning_rate": 1.7518910404933777e-05, + "loss": 0.2426, + "step": 4820 + }, + { + "epoch": 2.02, + "learning_rate": 1.751779253518104e-05, + "loss": 0.2263, + "step": 4821 + }, + { + "epoch": 2.02, + "learning_rate": 1.751667444933602e-05, + "loss": 0.2425, + "step": 4822 + }, + { + "epoch": 2.02, + "learning_rate": 1.7515556147430867e-05, + "loss": 0.2282, + "step": 4823 + }, + { + "epoch": 2.02, + "learning_rate": 1.751443762949772e-05, + "loss": 0.2368, + "step": 4824 + }, + { + "epoch": 2.02, + "learning_rate": 1.7513318895568734e-05, + "loss": 0.2329, + "step": 4825 + }, + { + "epoch": 2.02, + "learning_rate": 1.7512199945676065e-05, + "loss": 0.2456, + "step": 4826 + }, + { + "epoch": 2.02, + "learning_rate": 1.751108077985187e-05, + "loss": 0.2523, + "step": 4827 + }, + { + "epoch": 2.02, + "learning_rate": 1.7509961398128325e-05, + "loss": 0.2661, + "step": 4828 + }, + { + "epoch": 2.02, + "learning_rate": 1.7508841800537608e-05, + "loss": 0.2412, + "step": 4829 + }, + { + "epoch": 2.02, + "learning_rate": 1.750772198711189e-05, + "loss": 0.242, + "step": 4830 + }, + { + "epoch": 2.02, + "learning_rate": 1.750660195788337e-05, + "loss": 0.2648, + "step": 4831 + }, + { + "epoch": 2.02, + "learning_rate": 1.7505481712884235e-05, + "loss": 0.2752, + "step": 4832 + }, + { + "epoch": 2.02, + "learning_rate": 1.750436125214669e-05, + "loss": 0.2557, + "step": 4833 + }, + { + "epoch": 2.02, + "learning_rate": 1.750324057570294e-05, + "loss": 0.2343, + "step": 4834 + }, + { + "epoch": 2.02, + "learning_rate": 1.75021196835852e-05, + "loss": 0.2432, + "step": 4835 + }, + { + "epoch": 2.02, + "learning_rate": 1.750099857582568e-05, + "loss": 0.2425, + "step": 4836 + }, + { + "epoch": 2.03, + "learning_rate": 1.7499877252456624e-05, + "loss": 0.234, + "step": 4837 + }, + { + "epoch": 2.03, + "learning_rate": 1.7498755713510244e-05, + "loss": 0.2568, + "step": 4838 + }, + { + "epoch": 2.03, + "learning_rate": 1.7497633959018784e-05, + "loss": 0.2783, + "step": 4839 + }, + { + "epoch": 2.03, + "learning_rate": 1.749651198901449e-05, + "loss": 0.2693, + "step": 4840 + }, + { + "epoch": 2.03, + "learning_rate": 1.7495389803529615e-05, + "loss": 0.2427, + "step": 4841 + }, + { + "epoch": 2.03, + "learning_rate": 1.749426740259641e-05, + "loss": 0.2645, + "step": 4842 + }, + { + "epoch": 2.03, + "learning_rate": 1.7493144786247136e-05, + "loss": 0.2245, + "step": 4843 + }, + { + "epoch": 2.03, + "learning_rate": 1.7492021954514068e-05, + "loss": 0.2344, + "step": 4844 + }, + { + "epoch": 2.03, + "learning_rate": 1.7490898907429473e-05, + "loss": 0.2408, + "step": 4845 + }, + { + "epoch": 2.03, + "learning_rate": 1.748977564502564e-05, + "loss": 0.2739, + "step": 4846 + }, + { + "epoch": 2.03, + "learning_rate": 1.748865216733485e-05, + "loss": 0.2663, + "step": 4847 + }, + { + "epoch": 2.03, + "learning_rate": 1.7487528474389403e-05, + "loss": 0.2342, + "step": 4848 + }, + { + "epoch": 2.03, + "learning_rate": 1.748640456622159e-05, + "loss": 0.2576, + "step": 4849 + }, + { + "epoch": 2.03, + "learning_rate": 1.7485280442863718e-05, + "loss": 0.2432, + "step": 4850 + }, + { + "epoch": 2.03, + "learning_rate": 1.7484156104348107e-05, + "loss": 0.2371, + "step": 4851 + }, + { + "epoch": 2.03, + "learning_rate": 1.748303155070707e-05, + "loss": 0.2528, + "step": 4852 + }, + { + "epoch": 2.03, + "learning_rate": 1.7481906781972932e-05, + "loss": 0.2573, + "step": 4853 + }, + { + "epoch": 2.03, + "learning_rate": 1.748078179817802e-05, + "loss": 0.2821, + "step": 4854 + }, + { + "epoch": 2.03, + "learning_rate": 1.7479656599354677e-05, + "loss": 0.2348, + "step": 4855 + }, + { + "epoch": 2.03, + "learning_rate": 1.7478531185535235e-05, + "loss": 0.2659, + "step": 4856 + }, + { + "epoch": 2.03, + "learning_rate": 1.747740555675206e-05, + "loss": 0.2252, + "step": 4857 + }, + { + "epoch": 2.03, + "learning_rate": 1.747627971303749e-05, + "loss": 0.2632, + "step": 4858 + }, + { + "epoch": 2.03, + "learning_rate": 1.74751536544239e-05, + "loss": 0.2315, + "step": 4859 + }, + { + "epoch": 2.03, + "learning_rate": 1.747402738094365e-05, + "loss": 0.2415, + "step": 4860 + }, + { + "epoch": 2.04, + "learning_rate": 1.7472900892629114e-05, + "loss": 0.234, + "step": 4861 + }, + { + "epoch": 2.04, + "learning_rate": 1.747177418951267e-05, + "loss": 0.2695, + "step": 4862 + }, + { + "epoch": 2.04, + "learning_rate": 1.7470647271626712e-05, + "loss": 0.2861, + "step": 4863 + }, + { + "epoch": 2.04, + "learning_rate": 1.7469520139003627e-05, + "loss": 0.2673, + "step": 4864 + }, + { + "epoch": 2.04, + "learning_rate": 1.7468392791675812e-05, + "loss": 0.254, + "step": 4865 + }, + { + "epoch": 2.04, + "learning_rate": 1.7467265229675673e-05, + "loss": 0.2536, + "step": 4866 + }, + { + "epoch": 2.04, + "learning_rate": 1.7466137453035626e-05, + "loss": 0.2572, + "step": 4867 + }, + { + "epoch": 2.04, + "learning_rate": 1.746500946178808e-05, + "loss": 0.2533, + "step": 4868 + }, + { + "epoch": 2.04, + "learning_rate": 1.746388125596546e-05, + "loss": 0.2277, + "step": 4869 + }, + { + "epoch": 2.04, + "learning_rate": 1.7462752835600198e-05, + "loss": 0.226, + "step": 4870 + }, + { + "epoch": 2.04, + "learning_rate": 1.746162420072473e-05, + "loss": 0.2779, + "step": 4871 + }, + { + "epoch": 2.04, + "learning_rate": 1.7460495351371492e-05, + "loss": 0.2378, + "step": 4872 + }, + { + "epoch": 2.04, + "learning_rate": 1.745936628757294e-05, + "loss": 0.2827, + "step": 4873 + }, + { + "epoch": 2.04, + "learning_rate": 1.745823700936152e-05, + "loss": 0.2935, + "step": 4874 + }, + { + "epoch": 2.04, + "learning_rate": 1.7457107516769697e-05, + "loss": 0.2524, + "step": 4875 + }, + { + "epoch": 2.04, + "learning_rate": 1.7455977809829933e-05, + "loss": 0.2445, + "step": 4876 + }, + { + "epoch": 2.04, + "learning_rate": 1.7454847888574705e-05, + "loss": 0.2442, + "step": 4877 + }, + { + "epoch": 2.04, + "learning_rate": 1.745371775303649e-05, + "loss": 0.255, + "step": 4878 + }, + { + "epoch": 2.04, + "learning_rate": 1.7452587403247772e-05, + "loss": 0.2344, + "step": 4879 + }, + { + "epoch": 2.04, + "learning_rate": 1.7451456839241044e-05, + "loss": 0.2695, + "step": 4880 + }, + { + "epoch": 2.04, + "learning_rate": 1.7450326061048798e-05, + "loss": 0.2985, + "step": 4881 + }, + { + "epoch": 2.04, + "learning_rate": 1.7449195068703542e-05, + "loss": 0.2883, + "step": 4882 + }, + { + "epoch": 2.04, + "learning_rate": 1.7448063862237784e-05, + "loss": 0.2591, + "step": 4883 + }, + { + "epoch": 2.04, + "learning_rate": 1.7446932441684044e-05, + "loss": 0.2699, + "step": 4884 + }, + { + "epoch": 2.05, + "learning_rate": 1.7445800807074834e-05, + "loss": 0.2448, + "step": 4885 + }, + { + "epoch": 2.05, + "learning_rate": 1.744466895844269e-05, + "loss": 0.2717, + "step": 4886 + }, + { + "epoch": 2.05, + "learning_rate": 1.7443536895820143e-05, + "loss": 0.2549, + "step": 4887 + }, + { + "epoch": 2.05, + "learning_rate": 1.744240461923973e-05, + "loss": 0.261, + "step": 4888 + }, + { + "epoch": 2.05, + "learning_rate": 1.7441272128734004e-05, + "loss": 0.2292, + "step": 4889 + }, + { + "epoch": 2.05, + "learning_rate": 1.7440139424335513e-05, + "loss": 0.2494, + "step": 4890 + }, + { + "epoch": 2.05, + "learning_rate": 1.7439006506076816e-05, + "loss": 0.2344, + "step": 4891 + }, + { + "epoch": 2.05, + "learning_rate": 1.7437873373990478e-05, + "loss": 0.2476, + "step": 4892 + }, + { + "epoch": 2.05, + "learning_rate": 1.743674002810907e-05, + "loss": 0.2553, + "step": 4893 + }, + { + "epoch": 2.05, + "learning_rate": 1.743560646846517e-05, + "loss": 0.2397, + "step": 4894 + }, + { + "epoch": 2.05, + "learning_rate": 1.7434472695091358e-05, + "loss": 0.2433, + "step": 4895 + }, + { + "epoch": 2.05, + "learning_rate": 1.743333870802023e-05, + "loss": 0.2665, + "step": 4896 + }, + { + "epoch": 2.05, + "learning_rate": 1.7432204507284373e-05, + "loss": 0.2498, + "step": 4897 + }, + { + "epoch": 2.05, + "learning_rate": 1.7431070092916395e-05, + "loss": 0.2239, + "step": 4898 + }, + { + "epoch": 2.05, + "learning_rate": 1.74299354649489e-05, + "loss": 0.2373, + "step": 4899 + }, + { + "epoch": 2.05, + "learning_rate": 1.7428800623414502e-05, + "loss": 0.246, + "step": 4900 + }, + { + "epoch": 2.05, + "learning_rate": 1.7427665568345826e-05, + "loss": 0.2617, + "step": 4901 + }, + { + "epoch": 2.05, + "learning_rate": 1.7426530299775492e-05, + "loss": 0.247, + "step": 4902 + }, + { + "epoch": 2.05, + "learning_rate": 1.7425394817736135e-05, + "loss": 0.2407, + "step": 4903 + }, + { + "epoch": 2.05, + "learning_rate": 1.7424259122260396e-05, + "loss": 0.2377, + "step": 4904 + }, + { + "epoch": 2.05, + "learning_rate": 1.7423123213380914e-05, + "loss": 0.2585, + "step": 4905 + }, + { + "epoch": 2.05, + "learning_rate": 1.742198709113034e-05, + "loss": 0.2342, + "step": 4906 + }, + { + "epoch": 2.05, + "learning_rate": 1.7420850755541335e-05, + "loss": 0.2528, + "step": 4907 + }, + { + "epoch": 2.05, + "learning_rate": 1.7419714206646563e-05, + "loss": 0.2572, + "step": 4908 + }, + { + "epoch": 2.06, + "learning_rate": 1.741857744447869e-05, + "loss": 0.2499, + "step": 4909 + }, + { + "epoch": 2.06, + "learning_rate": 1.7417440469070386e-05, + "loss": 0.2417, + "step": 4910 + }, + { + "epoch": 2.06, + "learning_rate": 1.7416303280454343e-05, + "loss": 0.2383, + "step": 4911 + }, + { + "epoch": 2.06, + "learning_rate": 1.741516587866324e-05, + "loss": 0.2331, + "step": 4912 + }, + { + "epoch": 2.06, + "learning_rate": 1.7414028263729776e-05, + "loss": 0.2658, + "step": 4913 + }, + { + "epoch": 2.06, + "learning_rate": 1.741289043568665e-05, + "loss": 0.2369, + "step": 4914 + }, + { + "epoch": 2.06, + "learning_rate": 1.7411752394566565e-05, + "loss": 0.279, + "step": 4915 + }, + { + "epoch": 2.06, + "learning_rate": 1.7410614140402234e-05, + "loss": 0.2687, + "step": 4916 + }, + { + "epoch": 2.06, + "learning_rate": 1.7409475673226376e-05, + "loss": 0.2613, + "step": 4917 + }, + { + "epoch": 2.06, + "learning_rate": 1.7408336993071716e-05, + "loss": 0.2657, + "step": 4918 + }, + { + "epoch": 2.06, + "learning_rate": 1.7407198099970982e-05, + "loss": 0.2621, + "step": 4919 + }, + { + "epoch": 2.06, + "learning_rate": 1.7406058993956908e-05, + "loss": 0.2789, + "step": 4920 + }, + { + "epoch": 2.06, + "learning_rate": 1.7404919675062245e-05, + "loss": 0.2448, + "step": 4921 + }, + { + "epoch": 2.06, + "learning_rate": 1.7403780143319736e-05, + "loss": 0.2405, + "step": 4922 + }, + { + "epoch": 2.06, + "learning_rate": 1.7402640398762137e-05, + "loss": 0.2727, + "step": 4923 + }, + { + "epoch": 2.06, + "learning_rate": 1.740150044142221e-05, + "loss": 0.2417, + "step": 4924 + }, + { + "epoch": 2.06, + "learning_rate": 1.7400360271332718e-05, + "loss": 0.2379, + "step": 4925 + }, + { + "epoch": 2.06, + "learning_rate": 1.7399219888526438e-05, + "loss": 0.2408, + "step": 4926 + }, + { + "epoch": 2.06, + "learning_rate": 1.739807929303615e-05, + "loss": 0.242, + "step": 4927 + }, + { + "epoch": 2.06, + "learning_rate": 1.7396938484894636e-05, + "loss": 0.251, + "step": 4928 + }, + { + "epoch": 2.06, + "learning_rate": 1.739579746413469e-05, + "loss": 0.2409, + "step": 4929 + }, + { + "epoch": 2.06, + "learning_rate": 1.7394656230789106e-05, + "loss": 0.2426, + "step": 4930 + }, + { + "epoch": 2.06, + "learning_rate": 1.7393514784890694e-05, + "loss": 0.2722, + "step": 4931 + }, + { + "epoch": 2.06, + "learning_rate": 1.7392373126472257e-05, + "loss": 0.2441, + "step": 4932 + }, + { + "epoch": 2.07, + "learning_rate": 1.739123125556662e-05, + "loss": 0.2293, + "step": 4933 + }, + { + "epoch": 2.07, + "learning_rate": 1.7390089172206594e-05, + "loss": 0.2598, + "step": 4934 + }, + { + "epoch": 2.07, + "learning_rate": 1.7388946876425016e-05, + "loss": 0.2543, + "step": 4935 + }, + { + "epoch": 2.07, + "learning_rate": 1.7387804368254713e-05, + "loss": 0.2322, + "step": 4936 + }, + { + "epoch": 2.07, + "learning_rate": 1.738666164772853e-05, + "loss": 0.2598, + "step": 4937 + }, + { + "epoch": 2.07, + "learning_rate": 1.7385518714879315e-05, + "loss": 0.2482, + "step": 4938 + }, + { + "epoch": 2.07, + "learning_rate": 1.7384375569739918e-05, + "loss": 0.2783, + "step": 4939 + }, + { + "epoch": 2.07, + "learning_rate": 1.73832322123432e-05, + "loss": 0.2674, + "step": 4940 + }, + { + "epoch": 2.07, + "learning_rate": 1.738208864272202e-05, + "loss": 0.2536, + "step": 4941 + }, + { + "epoch": 2.07, + "learning_rate": 1.7380944860909253e-05, + "loss": 0.2471, + "step": 4942 + }, + { + "epoch": 2.07, + "learning_rate": 1.7379800866937773e-05, + "loss": 0.2384, + "step": 4943 + }, + { + "epoch": 2.07, + "learning_rate": 1.7378656660840475e-05, + "loss": 0.238, + "step": 4944 + }, + { + "epoch": 2.07, + "learning_rate": 1.7377512242650234e-05, + "loss": 0.2664, + "step": 4945 + }, + { + "epoch": 2.07, + "learning_rate": 1.7376367612399947e-05, + "loss": 0.2523, + "step": 4946 + }, + { + "epoch": 2.07, + "learning_rate": 1.737522277012252e-05, + "loss": 0.2584, + "step": 4947 + }, + { + "epoch": 2.07, + "learning_rate": 1.7374077715850865e-05, + "loss": 0.2505, + "step": 4948 + }, + { + "epoch": 2.07, + "learning_rate": 1.7372932449617884e-05, + "loss": 0.2087, + "step": 4949 + }, + { + "epoch": 2.07, + "learning_rate": 1.7371786971456504e-05, + "loss": 0.2631, + "step": 4950 + }, + { + "epoch": 2.07, + "learning_rate": 1.7370641281399647e-05, + "loss": 0.2492, + "step": 4951 + }, + { + "epoch": 2.07, + "learning_rate": 1.736949537948025e-05, + "loss": 0.2934, + "step": 4952 + }, + { + "epoch": 2.07, + "learning_rate": 1.7368349265731246e-05, + "loss": 0.2197, + "step": 4953 + }, + { + "epoch": 2.07, + "learning_rate": 1.7367202940185583e-05, + "loss": 0.2569, + "step": 4954 + }, + { + "epoch": 2.07, + "learning_rate": 1.7366056402876204e-05, + "loss": 0.2391, + "step": 4955 + }, + { + "epoch": 2.07, + "learning_rate": 1.7364909653836074e-05, + "loss": 0.2742, + "step": 4956 + }, + { + "epoch": 2.08, + "learning_rate": 1.7363762693098154e-05, + "loss": 0.2206, + "step": 4957 + }, + { + "epoch": 2.08, + "learning_rate": 1.7362615520695405e-05, + "loss": 0.2756, + "step": 4958 + }, + { + "epoch": 2.08, + "learning_rate": 1.736146813666081e-05, + "loss": 0.2251, + "step": 4959 + }, + { + "epoch": 2.08, + "learning_rate": 1.7360320541027342e-05, + "loss": 0.2713, + "step": 4960 + }, + { + "epoch": 2.08, + "learning_rate": 1.7359172733827995e-05, + "loss": 0.2465, + "step": 4961 + }, + { + "epoch": 2.08, + "learning_rate": 1.735802471509576e-05, + "loss": 0.2497, + "step": 4962 + }, + { + "epoch": 2.08, + "learning_rate": 1.735687648486363e-05, + "loss": 0.2337, + "step": 4963 + }, + { + "epoch": 2.08, + "learning_rate": 1.7355728043164613e-05, + "loss": 0.2509, + "step": 4964 + }, + { + "epoch": 2.08, + "learning_rate": 1.7354579390031724e-05, + "loss": 0.2376, + "step": 4965 + }, + { + "epoch": 2.08, + "learning_rate": 1.7353430525497974e-05, + "loss": 0.2525, + "step": 4966 + }, + { + "epoch": 2.08, + "learning_rate": 1.7352281449596392e-05, + "loss": 0.2847, + "step": 4967 + }, + { + "epoch": 2.08, + "learning_rate": 1.7351132162360002e-05, + "loss": 0.255, + "step": 4968 + }, + { + "epoch": 2.08, + "learning_rate": 1.734998266382184e-05, + "loss": 0.253, + "step": 4969 + }, + { + "epoch": 2.08, + "learning_rate": 1.734883295401495e-05, + "loss": 0.2528, + "step": 4970 + }, + { + "epoch": 2.08, + "learning_rate": 1.7347683032972377e-05, + "loss": 0.2893, + "step": 4971 + }, + { + "epoch": 2.08, + "learning_rate": 1.7346532900727177e-05, + "loss": 0.2915, + "step": 4972 + }, + { + "epoch": 2.08, + "learning_rate": 1.7345382557312408e-05, + "loss": 0.2729, + "step": 4973 + }, + { + "epoch": 2.08, + "learning_rate": 1.7344232002761138e-05, + "loss": 0.2677, + "step": 4974 + }, + { + "epoch": 2.08, + "learning_rate": 1.734308123710643e-05, + "loss": 0.2816, + "step": 4975 + }, + { + "epoch": 2.08, + "learning_rate": 1.734193026038137e-05, + "loss": 0.2447, + "step": 4976 + }, + { + "epoch": 2.08, + "learning_rate": 1.7340779072619043e-05, + "loss": 0.2602, + "step": 4977 + }, + { + "epoch": 2.08, + "learning_rate": 1.7339627673852534e-05, + "loss": 0.2614, + "step": 4978 + }, + { + "epoch": 2.08, + "learning_rate": 1.733847606411494e-05, + "loss": 0.2689, + "step": 4979 + }, + { + "epoch": 2.08, + "learning_rate": 1.7337324243439363e-05, + "loss": 0.2257, + "step": 4980 + }, + { + "epoch": 2.09, + "learning_rate": 1.7336172211858914e-05, + "loss": 0.2506, + "step": 4981 + }, + { + "epoch": 2.09, + "learning_rate": 1.7335019969406702e-05, + "loss": 0.2852, + "step": 4982 + }, + { + "epoch": 2.09, + "learning_rate": 1.7333867516115853e-05, + "loss": 0.226, + "step": 4983 + }, + { + "epoch": 2.09, + "learning_rate": 1.7332714852019487e-05, + "loss": 0.2242, + "step": 4984 + }, + { + "epoch": 2.09, + "learning_rate": 1.733156197715074e-05, + "loss": 0.2465, + "step": 4985 + }, + { + "epoch": 2.09, + "learning_rate": 1.733040889154275e-05, + "loss": 0.2993, + "step": 4986 + }, + { + "epoch": 2.09, + "learning_rate": 1.7329255595228664e-05, + "loss": 0.2579, + "step": 4987 + }, + { + "epoch": 2.09, + "learning_rate": 1.732810208824163e-05, + "loss": 0.2166, + "step": 4988 + }, + { + "epoch": 2.09, + "learning_rate": 1.73269483706148e-05, + "loss": 0.2538, + "step": 4989 + }, + { + "epoch": 2.09, + "learning_rate": 1.7325794442381344e-05, + "loss": 0.2515, + "step": 4990 + }, + { + "epoch": 2.09, + "learning_rate": 1.7324640303574426e-05, + "loss": 0.2579, + "step": 4991 + }, + { + "epoch": 2.09, + "learning_rate": 1.7323485954227223e-05, + "loss": 0.2537, + "step": 4992 + }, + { + "epoch": 2.09, + "learning_rate": 1.7322331394372916e-05, + "loss": 0.2616, + "step": 4993 + }, + { + "epoch": 2.09, + "learning_rate": 1.732117662404469e-05, + "loss": 0.2548, + "step": 4994 + }, + { + "epoch": 2.09, + "learning_rate": 1.7320021643275734e-05, + "loss": 0.2309, + "step": 4995 + }, + { + "epoch": 2.09, + "learning_rate": 1.731886645209926e-05, + "loss": 0.2589, + "step": 4996 + }, + { + "epoch": 2.09, + "learning_rate": 1.731771105054846e-05, + "loss": 0.2616, + "step": 4997 + }, + { + "epoch": 2.09, + "learning_rate": 1.7316555438656547e-05, + "loss": 0.261, + "step": 4998 + }, + { + "epoch": 2.09, + "learning_rate": 1.7315399616456746e-05, + "loss": 0.2548, + "step": 4999 + }, + { + "epoch": 2.09, + "learning_rate": 1.731424358398227e-05, + "loss": 0.2641, + "step": 5000 + }, + { + "epoch": 2.09, + "learning_rate": 1.7313087341266355e-05, + "loss": 0.232, + "step": 5001 + }, + { + "epoch": 2.09, + "learning_rate": 1.7311930888342234e-05, + "loss": 0.2712, + "step": 5002 + }, + { + "epoch": 2.09, + "learning_rate": 1.7310774225243146e-05, + "loss": 0.2451, + "step": 5003 + }, + { + "epoch": 2.1, + "learning_rate": 1.7309617352002343e-05, + "loss": 0.2562, + "step": 5004 + }, + { + "epoch": 2.1, + "learning_rate": 1.7308460268653077e-05, + "loss": 0.2448, + "step": 5005 + }, + { + "epoch": 2.1, + "learning_rate": 1.73073029752286e-05, + "loss": 0.2452, + "step": 5006 + }, + { + "epoch": 2.1, + "learning_rate": 1.730614547176219e-05, + "loss": 0.2327, + "step": 5007 + }, + { + "epoch": 2.1, + "learning_rate": 1.730498775828711e-05, + "loss": 0.2429, + "step": 5008 + }, + { + "epoch": 2.1, + "learning_rate": 1.7303829834836636e-05, + "loss": 0.2622, + "step": 5009 + }, + { + "epoch": 2.1, + "learning_rate": 1.7302671701444057e-05, + "loss": 0.2703, + "step": 5010 + }, + { + "epoch": 2.1, + "learning_rate": 1.730151335814266e-05, + "loss": 0.2536, + "step": 5011 + }, + { + "epoch": 2.1, + "learning_rate": 1.730035480496574e-05, + "loss": 0.2419, + "step": 5012 + }, + { + "epoch": 2.1, + "learning_rate": 1.72991960419466e-05, + "loss": 0.2594, + "step": 5013 + }, + { + "epoch": 2.1, + "learning_rate": 1.7298037069118546e-05, + "loss": 0.2761, + "step": 5014 + }, + { + "epoch": 2.1, + "learning_rate": 1.7296877886514892e-05, + "loss": 0.2803, + "step": 5015 + }, + { + "epoch": 2.1, + "learning_rate": 1.7295718494168963e-05, + "loss": 0.2587, + "step": 5016 + }, + { + "epoch": 2.1, + "learning_rate": 1.7294558892114077e-05, + "loss": 0.2206, + "step": 5017 + }, + { + "epoch": 2.1, + "learning_rate": 1.729339908038357e-05, + "loss": 0.2495, + "step": 5018 + }, + { + "epoch": 2.1, + "learning_rate": 1.7292239059010777e-05, + "loss": 0.2289, + "step": 5019 + }, + { + "epoch": 2.1, + "learning_rate": 1.7291078828029042e-05, + "loss": 0.2807, + "step": 5020 + }, + { + "epoch": 2.1, + "learning_rate": 1.728991838747172e-05, + "loss": 0.242, + "step": 5021 + }, + { + "epoch": 2.1, + "learning_rate": 1.7288757737372157e-05, + "loss": 0.2491, + "step": 5022 + }, + { + "epoch": 2.1, + "learning_rate": 1.7287596877763725e-05, + "loss": 0.2257, + "step": 5023 + }, + { + "epoch": 2.1, + "learning_rate": 1.7286435808679787e-05, + "loss": 0.2492, + "step": 5024 + }, + { + "epoch": 2.1, + "learning_rate": 1.7285274530153718e-05, + "loss": 0.2578, + "step": 5025 + }, + { + "epoch": 2.1, + "learning_rate": 1.7284113042218895e-05, + "loss": 0.26, + "step": 5026 + }, + { + "epoch": 2.1, + "learning_rate": 1.728295134490871e-05, + "loss": 0.2653, + "step": 5027 + }, + { + "epoch": 2.11, + "learning_rate": 1.728178943825655e-05, + "loss": 0.279, + "step": 5028 + }, + { + "epoch": 2.11, + "learning_rate": 1.7280627322295812e-05, + "loss": 0.2164, + "step": 5029 + }, + { + "epoch": 2.11, + "learning_rate": 1.7279464997059905e-05, + "loss": 0.2582, + "step": 5030 + }, + { + "epoch": 2.11, + "learning_rate": 1.727830246258224e-05, + "loss": 0.2465, + "step": 5031 + }, + { + "epoch": 2.11, + "learning_rate": 1.727713971889622e-05, + "loss": 0.2576, + "step": 5032 + }, + { + "epoch": 2.11, + "learning_rate": 1.7275976766035282e-05, + "loss": 0.2423, + "step": 5033 + }, + { + "epoch": 2.11, + "learning_rate": 1.7274813604032847e-05, + "loss": 0.2154, + "step": 5034 + }, + { + "epoch": 2.11, + "learning_rate": 1.727365023292235e-05, + "loss": 0.2747, + "step": 5035 + }, + { + "epoch": 2.11, + "learning_rate": 1.7272486652737236e-05, + "loss": 0.2407, + "step": 5036 + }, + { + "epoch": 2.11, + "learning_rate": 1.7271322863510942e-05, + "loss": 0.2562, + "step": 5037 + }, + { + "epoch": 2.11, + "learning_rate": 1.7270158865276927e-05, + "loss": 0.2562, + "step": 5038 + }, + { + "epoch": 2.11, + "learning_rate": 1.7268994658068645e-05, + "loss": 0.2782, + "step": 5039 + }, + { + "epoch": 2.11, + "learning_rate": 1.7267830241919562e-05, + "loss": 0.2556, + "step": 5040 + }, + { + "epoch": 2.11, + "learning_rate": 1.726666561686315e-05, + "loss": 0.2463, + "step": 5041 + }, + { + "epoch": 2.11, + "learning_rate": 1.7265500782932885e-05, + "loss": 0.2729, + "step": 5042 + }, + { + "epoch": 2.11, + "learning_rate": 1.7264335740162244e-05, + "loss": 0.223, + "step": 5043 + }, + { + "epoch": 2.11, + "learning_rate": 1.7263170488584717e-05, + "loss": 0.2527, + "step": 5044 + }, + { + "epoch": 2.11, + "learning_rate": 1.7262005028233802e-05, + "loss": 0.2513, + "step": 5045 + }, + { + "epoch": 2.11, + "learning_rate": 1.7260839359142995e-05, + "loss": 0.2431, + "step": 5046 + }, + { + "epoch": 2.11, + "learning_rate": 1.7259673481345808e-05, + "loss": 0.2269, + "step": 5047 + }, + { + "epoch": 2.11, + "learning_rate": 1.7258507394875742e-05, + "loss": 0.249, + "step": 5048 + }, + { + "epoch": 2.11, + "learning_rate": 1.725734109976633e-05, + "loss": 0.2325, + "step": 5049 + }, + { + "epoch": 2.11, + "learning_rate": 1.725617459605108e-05, + "loss": 0.2591, + "step": 5050 + }, + { + "epoch": 2.11, + "learning_rate": 1.7255007883763537e-05, + "loss": 0.2405, + "step": 5051 + }, + { + "epoch": 2.12, + "learning_rate": 1.7253840962937226e-05, + "loss": 0.2675, + "step": 5052 + }, + { + "epoch": 2.12, + "learning_rate": 1.7252673833605697e-05, + "loss": 0.2355, + "step": 5053 + }, + { + "epoch": 2.12, + "learning_rate": 1.725150649580249e-05, + "loss": 0.2465, + "step": 5054 + }, + { + "epoch": 2.12, + "learning_rate": 1.7250338949561167e-05, + "loss": 0.2481, + "step": 5055 + }, + { + "epoch": 2.12, + "learning_rate": 1.7249171194915283e-05, + "loss": 0.2396, + "step": 5056 + }, + { + "epoch": 2.12, + "learning_rate": 1.724800323189841e-05, + "loss": 0.2557, + "step": 5057 + }, + { + "epoch": 2.12, + "learning_rate": 1.724683506054411e-05, + "loss": 0.2235, + "step": 5058 + }, + { + "epoch": 2.12, + "learning_rate": 1.7245666680885975e-05, + "loss": 0.2663, + "step": 5059 + }, + { + "epoch": 2.12, + "learning_rate": 1.724449809295757e-05, + "loss": 0.2585, + "step": 5060 + }, + { + "epoch": 2.12, + "learning_rate": 1.7243329296792502e-05, + "loss": 0.2593, + "step": 5061 + }, + { + "epoch": 2.12, + "learning_rate": 1.7242160292424362e-05, + "loss": 0.2311, + "step": 5062 + }, + { + "epoch": 2.12, + "learning_rate": 1.724099107988675e-05, + "loss": 0.2416, + "step": 5063 + }, + { + "epoch": 2.12, + "learning_rate": 1.7239821659213272e-05, + "loss": 0.2705, + "step": 5064 + }, + { + "epoch": 2.12, + "learning_rate": 1.723865203043755e-05, + "loss": 0.2453, + "step": 5065 + }, + { + "epoch": 2.12, + "learning_rate": 1.7237482193593195e-05, + "loss": 0.2731, + "step": 5066 + }, + { + "epoch": 2.12, + "learning_rate": 1.7236312148713838e-05, + "loss": 0.2653, + "step": 5067 + }, + { + "epoch": 2.12, + "learning_rate": 1.723514189583311e-05, + "loss": 0.2623, + "step": 5068 + }, + { + "epoch": 2.12, + "learning_rate": 1.7233971434984648e-05, + "loss": 0.2766, + "step": 5069 + }, + { + "epoch": 2.12, + "learning_rate": 1.72328007662021e-05, + "loss": 0.2531, + "step": 5070 + }, + { + "epoch": 2.12, + "learning_rate": 1.7231629889519106e-05, + "loss": 0.256, + "step": 5071 + }, + { + "epoch": 2.12, + "learning_rate": 1.723045880496933e-05, + "loss": 0.2576, + "step": 5072 + }, + { + "epoch": 2.12, + "learning_rate": 1.7229287512586432e-05, + "loss": 0.2503, + "step": 5073 + }, + { + "epoch": 2.12, + "learning_rate": 1.7228116012404085e-05, + "loss": 0.2751, + "step": 5074 + }, + { + "epoch": 2.12, + "learning_rate": 1.722694430445595e-05, + "loss": 0.258, + "step": 5075 + }, + { + "epoch": 2.13, + "learning_rate": 1.7225772388775716e-05, + "loss": 0.2496, + "step": 5076 + }, + { + "epoch": 2.13, + "learning_rate": 1.722460026539707e-05, + "loss": 0.2667, + "step": 5077 + }, + { + "epoch": 2.13, + "learning_rate": 1.7223427934353695e-05, + "loss": 0.2329, + "step": 5078 + }, + { + "epoch": 2.13, + "learning_rate": 1.7222255395679298e-05, + "loss": 0.2524, + "step": 5079 + }, + { + "epoch": 2.13, + "learning_rate": 1.7221082649407578e-05, + "loss": 0.25, + "step": 5080 + }, + { + "epoch": 2.13, + "learning_rate": 1.7219909695572244e-05, + "loss": 0.2948, + "step": 5081 + }, + { + "epoch": 2.13, + "learning_rate": 1.7218736534207014e-05, + "loss": 0.2579, + "step": 5082 + }, + { + "epoch": 2.13, + "learning_rate": 1.7217563165345607e-05, + "loss": 0.2617, + "step": 5083 + }, + { + "epoch": 2.13, + "learning_rate": 1.721638958902175e-05, + "loss": 0.2608, + "step": 5084 + }, + { + "epoch": 2.13, + "learning_rate": 1.721521580526918e-05, + "loss": 0.2515, + "step": 5085 + }, + { + "epoch": 2.13, + "learning_rate": 1.7214041814121635e-05, + "loss": 0.2472, + "step": 5086 + }, + { + "epoch": 2.13, + "learning_rate": 1.721286761561286e-05, + "loss": 0.2589, + "step": 5087 + }, + { + "epoch": 2.13, + "learning_rate": 1.7211693209776604e-05, + "loss": 0.2718, + "step": 5088 + }, + { + "epoch": 2.13, + "learning_rate": 1.7210518596646627e-05, + "loss": 0.2498, + "step": 5089 + }, + { + "epoch": 2.13, + "learning_rate": 1.720934377625669e-05, + "loss": 0.2506, + "step": 5090 + }, + { + "epoch": 2.13, + "learning_rate": 1.7208168748640565e-05, + "loss": 0.2388, + "step": 5091 + }, + { + "epoch": 2.13, + "learning_rate": 1.7206993513832026e-05, + "loss": 0.2686, + "step": 5092 + }, + { + "epoch": 2.13, + "learning_rate": 1.7205818071864853e-05, + "loss": 0.2474, + "step": 5093 + }, + { + "epoch": 2.13, + "learning_rate": 1.7204642422772836e-05, + "loss": 0.2179, + "step": 5094 + }, + { + "epoch": 2.13, + "learning_rate": 1.7203466566589763e-05, + "loss": 0.257, + "step": 5095 + }, + { + "epoch": 2.13, + "learning_rate": 1.7202290503349436e-05, + "loss": 0.2763, + "step": 5096 + }, + { + "epoch": 2.13, + "learning_rate": 1.7201114233085664e-05, + "loss": 0.2821, + "step": 5097 + }, + { + "epoch": 2.13, + "learning_rate": 1.719993775583225e-05, + "loss": 0.2528, + "step": 5098 + }, + { + "epoch": 2.13, + "learning_rate": 1.7198761071623014e-05, + "loss": 0.2878, + "step": 5099 + }, + { + "epoch": 2.14, + "learning_rate": 1.7197584180491776e-05, + "loss": 0.2428, + "step": 5100 + }, + { + "epoch": 2.14, + "learning_rate": 1.7196407082472375e-05, + "loss": 0.2827, + "step": 5101 + }, + { + "epoch": 2.14, + "learning_rate": 1.7195229777598635e-05, + "loss": 0.2712, + "step": 5102 + }, + { + "epoch": 2.14, + "learning_rate": 1.71940522659044e-05, + "loss": 0.271, + "step": 5103 + }, + { + "epoch": 2.14, + "learning_rate": 1.7192874547423514e-05, + "loss": 0.223, + "step": 5104 + }, + { + "epoch": 2.14, + "learning_rate": 1.7191696622189836e-05, + "loss": 0.2587, + "step": 5105 + }, + { + "epoch": 2.14, + "learning_rate": 1.719051849023722e-05, + "loss": 0.2812, + "step": 5106 + }, + { + "epoch": 2.14, + "learning_rate": 1.718934015159953e-05, + "loss": 0.258, + "step": 5107 + }, + { + "epoch": 2.14, + "learning_rate": 1.7188161606310635e-05, + "loss": 0.2294, + "step": 5108 + }, + { + "epoch": 2.14, + "learning_rate": 1.7186982854404415e-05, + "loss": 0.256, + "step": 5109 + }, + { + "epoch": 2.14, + "learning_rate": 1.718580389591475e-05, + "loss": 0.252, + "step": 5110 + }, + { + "epoch": 2.14, + "learning_rate": 1.7184624730875528e-05, + "loss": 0.2629, + "step": 5111 + }, + { + "epoch": 2.14, + "learning_rate": 1.7183445359320645e-05, + "loss": 0.2399, + "step": 5112 + }, + { + "epoch": 2.14, + "learning_rate": 1.7182265781284e-05, + "loss": 0.257, + "step": 5113 + }, + { + "epoch": 2.14, + "learning_rate": 1.71810859967995e-05, + "loss": 0.2499, + "step": 5114 + }, + { + "epoch": 2.14, + "learning_rate": 1.717990600590105e-05, + "loss": 0.24, + "step": 5115 + }, + { + "epoch": 2.14, + "learning_rate": 1.7178725808622575e-05, + "loss": 0.2659, + "step": 5116 + }, + { + "epoch": 2.14, + "learning_rate": 1.7177545404997998e-05, + "loss": 0.2335, + "step": 5117 + }, + { + "epoch": 2.14, + "learning_rate": 1.717636479506125e-05, + "loss": 0.2443, + "step": 5118 + }, + { + "epoch": 2.14, + "learning_rate": 1.717518397884626e-05, + "loss": 0.2775, + "step": 5119 + }, + { + "epoch": 2.14, + "learning_rate": 1.717400295638698e-05, + "loss": 0.235, + "step": 5120 + }, + { + "epoch": 2.14, + "learning_rate": 1.7172821727717345e-05, + "loss": 0.2412, + "step": 5121 + }, + { + "epoch": 2.14, + "learning_rate": 1.7171640292871314e-05, + "loss": 0.2423, + "step": 5122 + }, + { + "epoch": 2.14, + "learning_rate": 1.7170458651882847e-05, + "loss": 0.2764, + "step": 5123 + }, + { + "epoch": 2.15, + "learning_rate": 1.716927680478591e-05, + "loss": 0.2641, + "step": 5124 + }, + { + "epoch": 2.15, + "learning_rate": 1.716809475161448e-05, + "loss": 0.2793, + "step": 5125 + }, + { + "epoch": 2.15, + "learning_rate": 1.716691249240252e-05, + "loss": 0.2427, + "step": 5126 + }, + { + "epoch": 2.15, + "learning_rate": 1.716573002718402e-05, + "loss": 0.2391, + "step": 5127 + }, + { + "epoch": 2.15, + "learning_rate": 1.7164547355992972e-05, + "loss": 0.2451, + "step": 5128 + }, + { + "epoch": 2.15, + "learning_rate": 1.7163364478863365e-05, + "loss": 0.2525, + "step": 5129 + }, + { + "epoch": 2.15, + "learning_rate": 1.7162181395829204e-05, + "loss": 0.2202, + "step": 5130 + }, + { + "epoch": 2.15, + "learning_rate": 1.7160998106924494e-05, + "loss": 0.2548, + "step": 5131 + }, + { + "epoch": 2.15, + "learning_rate": 1.715981461218325e-05, + "loss": 0.2345, + "step": 5132 + }, + { + "epoch": 2.15, + "learning_rate": 1.7158630911639488e-05, + "loss": 0.2593, + "step": 5133 + }, + { + "epoch": 2.15, + "learning_rate": 1.715744700532723e-05, + "loss": 0.2576, + "step": 5134 + }, + { + "epoch": 2.15, + "learning_rate": 1.7156262893280512e-05, + "loss": 0.2574, + "step": 5135 + }, + { + "epoch": 2.15, + "learning_rate": 1.7155078575533366e-05, + "loss": 0.2137, + "step": 5136 + }, + { + "epoch": 2.15, + "learning_rate": 1.7153894052119837e-05, + "loss": 0.2537, + "step": 5137 + }, + { + "epoch": 2.15, + "learning_rate": 1.7152709323073972e-05, + "loss": 0.2346, + "step": 5138 + }, + { + "epoch": 2.15, + "learning_rate": 1.7151524388429823e-05, + "loss": 0.2037, + "step": 5139 + }, + { + "epoch": 2.15, + "learning_rate": 1.7150339248221454e-05, + "loss": 0.2163, + "step": 5140 + }, + { + "epoch": 2.15, + "learning_rate": 1.7149153902482926e-05, + "loss": 0.2761, + "step": 5141 + }, + { + "epoch": 2.15, + "learning_rate": 1.7147968351248315e-05, + "loss": 0.2492, + "step": 5142 + }, + { + "epoch": 2.15, + "learning_rate": 1.7146782594551698e-05, + "loss": 0.2327, + "step": 5143 + }, + { + "epoch": 2.15, + "learning_rate": 1.714559663242715e-05, + "loss": 0.2507, + "step": 5144 + }, + { + "epoch": 2.15, + "learning_rate": 1.7144410464908774e-05, + "loss": 0.2678, + "step": 5145 + }, + { + "epoch": 2.15, + "learning_rate": 1.714322409203066e-05, + "loss": 0.2523, + "step": 5146 + }, + { + "epoch": 2.15, + "learning_rate": 1.7142037513826904e-05, + "loss": 0.2563, + "step": 5147 + }, + { + "epoch": 2.16, + "learning_rate": 1.714085073033162e-05, + "loss": 0.2264, + "step": 5148 + }, + { + "epoch": 2.16, + "learning_rate": 1.7139663741578917e-05, + "loss": 0.2479, + "step": 5149 + }, + { + "epoch": 2.16, + "learning_rate": 1.7138476547602917e-05, + "loss": 0.2299, + "step": 5150 + }, + { + "epoch": 2.16, + "learning_rate": 1.7137289148437743e-05, + "loss": 0.2625, + "step": 5151 + }, + { + "epoch": 2.16, + "learning_rate": 1.7136101544117526e-05, + "loss": 0.2654, + "step": 5152 + }, + { + "epoch": 2.16, + "learning_rate": 1.7134913734676398e-05, + "loss": 0.2451, + "step": 5153 + }, + { + "epoch": 2.16, + "learning_rate": 1.7133725720148513e-05, + "loss": 0.2287, + "step": 5154 + }, + { + "epoch": 2.16, + "learning_rate": 1.713253750056801e-05, + "loss": 0.2551, + "step": 5155 + }, + { + "epoch": 2.16, + "learning_rate": 1.7131349075969044e-05, + "loss": 0.2604, + "step": 5156 + }, + { + "epoch": 2.16, + "learning_rate": 1.713016044638578e-05, + "loss": 0.2559, + "step": 5157 + }, + { + "epoch": 2.16, + "learning_rate": 1.7128971611852378e-05, + "loss": 0.2613, + "step": 5158 + }, + { + "epoch": 2.16, + "learning_rate": 1.7127782572403015e-05, + "loss": 0.2773, + "step": 5159 + }, + { + "epoch": 2.16, + "learning_rate": 1.7126593328071867e-05, + "loss": 0.2502, + "step": 5160 + }, + { + "epoch": 2.16, + "learning_rate": 1.7125403878893118e-05, + "loss": 0.2821, + "step": 5161 + }, + { + "epoch": 2.16, + "learning_rate": 1.7124214224900955e-05, + "loss": 0.2593, + "step": 5162 + }, + { + "epoch": 2.16, + "learning_rate": 1.712302436612958e-05, + "loss": 0.2377, + "step": 5163 + }, + { + "epoch": 2.16, + "learning_rate": 1.712183430261319e-05, + "loss": 0.2523, + "step": 5164 + }, + { + "epoch": 2.16, + "learning_rate": 1.712064403438599e-05, + "loss": 0.2733, + "step": 5165 + }, + { + "epoch": 2.16, + "learning_rate": 1.71194535614822e-05, + "loss": 0.2376, + "step": 5166 + }, + { + "epoch": 2.16, + "learning_rate": 1.711826288393603e-05, + "loss": 0.2457, + "step": 5167 + }, + { + "epoch": 2.16, + "learning_rate": 1.7117072001781713e-05, + "loss": 0.2349, + "step": 5168 + }, + { + "epoch": 2.16, + "learning_rate": 1.7115880915053476e-05, + "loss": 0.2544, + "step": 5169 + }, + { + "epoch": 2.16, + "learning_rate": 1.7114689623785556e-05, + "loss": 0.2306, + "step": 5170 + }, + { + "epoch": 2.16, + "learning_rate": 1.7113498128012197e-05, + "loss": 0.2666, + "step": 5171 + }, + { + "epoch": 2.17, + "learning_rate": 1.711230642776765e-05, + "loss": 0.2648, + "step": 5172 + }, + { + "epoch": 2.17, + "learning_rate": 1.711111452308616e-05, + "loss": 0.2542, + "step": 5173 + }, + { + "epoch": 2.17, + "learning_rate": 1.7109922414001997e-05, + "loss": 0.2421, + "step": 5174 + }, + { + "epoch": 2.17, + "learning_rate": 1.710873010054942e-05, + "loss": 0.2632, + "step": 5175 + }, + { + "epoch": 2.17, + "learning_rate": 1.710753758276271e-05, + "loss": 0.2471, + "step": 5176 + }, + { + "epoch": 2.17, + "learning_rate": 1.7106344860676137e-05, + "loss": 0.2528, + "step": 5177 + }, + { + "epoch": 2.17, + "learning_rate": 1.7105151934323988e-05, + "loss": 0.2461, + "step": 5178 + }, + { + "epoch": 2.17, + "learning_rate": 1.7103958803740547e-05, + "loss": 0.2155, + "step": 5179 + }, + { + "epoch": 2.17, + "learning_rate": 1.7102765468960116e-05, + "loss": 0.2406, + "step": 5180 + }, + { + "epoch": 2.17, + "learning_rate": 1.7101571930017e-05, + "loss": 0.2621, + "step": 5181 + }, + { + "epoch": 2.17, + "learning_rate": 1.7100378186945494e-05, + "loss": 0.2702, + "step": 5182 + }, + { + "epoch": 2.17, + "learning_rate": 1.709918423977992e-05, + "loss": 0.256, + "step": 5183 + }, + { + "epoch": 2.17, + "learning_rate": 1.7097990088554595e-05, + "loss": 0.2888, + "step": 5184 + }, + { + "epoch": 2.17, + "learning_rate": 1.7096795733303844e-05, + "loss": 0.2608, + "step": 5185 + }, + { + "epoch": 2.17, + "learning_rate": 1.7095601174061997e-05, + "loss": 0.2615, + "step": 5186 + }, + { + "epoch": 2.17, + "learning_rate": 1.709440641086339e-05, + "loss": 0.2309, + "step": 5187 + }, + { + "epoch": 2.17, + "learning_rate": 1.7093211443742368e-05, + "loss": 0.2574, + "step": 5188 + }, + { + "epoch": 2.17, + "learning_rate": 1.7092016272733274e-05, + "loss": 0.2316, + "step": 5189 + }, + { + "epoch": 2.17, + "learning_rate": 1.709082089787047e-05, + "loss": 0.2482, + "step": 5190 + }, + { + "epoch": 2.17, + "learning_rate": 1.7089625319188308e-05, + "loss": 0.264, + "step": 5191 + }, + { + "epoch": 2.17, + "learning_rate": 1.708842953672116e-05, + "loss": 0.2641, + "step": 5192 + }, + { + "epoch": 2.17, + "learning_rate": 1.7087233550503394e-05, + "loss": 0.2737, + "step": 5193 + }, + { + "epoch": 2.17, + "learning_rate": 1.7086037360569386e-05, + "loss": 0.2495, + "step": 5194 + }, + { + "epoch": 2.18, + "learning_rate": 1.7084840966953526e-05, + "loss": 0.2728, + "step": 5195 + }, + { + "epoch": 2.18, + "learning_rate": 1.7083644369690197e-05, + "loss": 0.2599, + "step": 5196 + }, + { + "epoch": 2.18, + "learning_rate": 1.7082447568813798e-05, + "loss": 0.2787, + "step": 5197 + }, + { + "epoch": 2.18, + "learning_rate": 1.708125056435873e-05, + "loss": 0.2641, + "step": 5198 + }, + { + "epoch": 2.18, + "learning_rate": 1.7080053356359396e-05, + "loss": 0.2461, + "step": 5199 + }, + { + "epoch": 2.18, + "learning_rate": 1.707885594485021e-05, + "loss": 0.2538, + "step": 5200 + }, + { + "epoch": 2.18, + "learning_rate": 1.7077658329865593e-05, + "loss": 0.2623, + "step": 5201 + }, + { + "epoch": 2.18, + "learning_rate": 1.7076460511439966e-05, + "loss": 0.2551, + "step": 5202 + }, + { + "epoch": 2.18, + "learning_rate": 1.7075262489607766e-05, + "loss": 0.2842, + "step": 5203 + }, + { + "epoch": 2.18, + "learning_rate": 1.707406426440342e-05, + "loss": 0.2662, + "step": 5204 + }, + { + "epoch": 2.18, + "learning_rate": 1.7072865835861377e-05, + "loss": 0.2783, + "step": 5205 + }, + { + "epoch": 2.18, + "learning_rate": 1.707166720401608e-05, + "loss": 0.2565, + "step": 5206 + }, + { + "epoch": 2.18, + "learning_rate": 1.7070468368901982e-05, + "loss": 0.2397, + "step": 5207 + }, + { + "epoch": 2.18, + "learning_rate": 1.706926933055355e-05, + "loss": 0.2461, + "step": 5208 + }, + { + "epoch": 2.18, + "learning_rate": 1.706807008900524e-05, + "loss": 0.2689, + "step": 5209 + }, + { + "epoch": 2.18, + "learning_rate": 1.7066870644291527e-05, + "loss": 0.2655, + "step": 5210 + }, + { + "epoch": 2.18, + "learning_rate": 1.7065670996446892e-05, + "loss": 0.2626, + "step": 5211 + }, + { + "epoch": 2.18, + "learning_rate": 1.7064471145505813e-05, + "loss": 0.2378, + "step": 5212 + }, + { + "epoch": 2.18, + "learning_rate": 1.706327109150278e-05, + "loss": 0.2428, + "step": 5213 + }, + { + "epoch": 2.18, + "learning_rate": 1.7062070834472287e-05, + "loss": 0.2442, + "step": 5214 + }, + { + "epoch": 2.18, + "learning_rate": 1.7060870374448833e-05, + "loss": 0.2648, + "step": 5215 + }, + { + "epoch": 2.18, + "learning_rate": 1.705966971146693e-05, + "loss": 0.2259, + "step": 5216 + }, + { + "epoch": 2.18, + "learning_rate": 1.7058468845561084e-05, + "loss": 0.2766, + "step": 5217 + }, + { + "epoch": 2.18, + "learning_rate": 1.7057267776765814e-05, + "loss": 0.2512, + "step": 5218 + }, + { + "epoch": 2.19, + "learning_rate": 1.705606650511564e-05, + "loss": 0.2322, + "step": 5219 + }, + { + "epoch": 2.19, + "learning_rate": 1.7054865030645102e-05, + "loss": 0.2896, + "step": 5220 + }, + { + "epoch": 2.19, + "learning_rate": 1.7053663353388727e-05, + "loss": 0.2412, + "step": 5221 + }, + { + "epoch": 2.19, + "learning_rate": 1.7052461473381056e-05, + "loss": 0.2456, + "step": 5222 + }, + { + "epoch": 2.19, + "learning_rate": 1.7051259390656642e-05, + "loss": 0.2572, + "step": 5223 + }, + { + "epoch": 2.19, + "learning_rate": 1.7050057105250034e-05, + "loss": 0.2393, + "step": 5224 + }, + { + "epoch": 2.19, + "learning_rate": 1.7048854617195788e-05, + "loss": 0.2851, + "step": 5225 + }, + { + "epoch": 2.19, + "learning_rate": 1.7047651926528472e-05, + "loss": 0.2603, + "step": 5226 + }, + { + "epoch": 2.19, + "learning_rate": 1.7046449033282656e-05, + "loss": 0.2448, + "step": 5227 + }, + { + "epoch": 2.19, + "learning_rate": 1.7045245937492912e-05, + "loss": 0.2544, + "step": 5228 + }, + { + "epoch": 2.19, + "learning_rate": 1.704404263919383e-05, + "loss": 0.2433, + "step": 5229 + }, + { + "epoch": 2.19, + "learning_rate": 1.7042839138419992e-05, + "loss": 0.2589, + "step": 5230 + }, + { + "epoch": 2.19, + "learning_rate": 1.704163543520599e-05, + "loss": 0.2382, + "step": 5231 + }, + { + "epoch": 2.19, + "learning_rate": 1.7040431529586427e-05, + "loss": 0.2312, + "step": 5232 + }, + { + "epoch": 2.19, + "learning_rate": 1.703922742159591e-05, + "loss": 0.2521, + "step": 5233 + }, + { + "epoch": 2.19, + "learning_rate": 1.7038023111269045e-05, + "loss": 0.258, + "step": 5234 + }, + { + "epoch": 2.19, + "learning_rate": 1.7036818598640456e-05, + "loss": 0.2897, + "step": 5235 + }, + { + "epoch": 2.19, + "learning_rate": 1.7035613883744753e-05, + "loss": 0.2337, + "step": 5236 + }, + { + "epoch": 2.19, + "learning_rate": 1.7034408966616577e-05, + "loss": 0.2502, + "step": 5237 + }, + { + "epoch": 2.19, + "learning_rate": 1.7033203847290553e-05, + "loss": 0.2469, + "step": 5238 + }, + { + "epoch": 2.19, + "learning_rate": 1.703199852580133e-05, + "loss": 0.2517, + "step": 5239 + }, + { + "epoch": 2.19, + "learning_rate": 1.7030793002183546e-05, + "loss": 0.2594, + "step": 5240 + }, + { + "epoch": 2.19, + "learning_rate": 1.702958727647186e-05, + "loss": 0.2306, + "step": 5241 + }, + { + "epoch": 2.19, + "learning_rate": 1.7028381348700926e-05, + "loss": 0.2449, + "step": 5242 + }, + { + "epoch": 2.2, + "learning_rate": 1.70271752189054e-05, + "loss": 0.2469, + "step": 5243 + }, + { + "epoch": 2.2, + "learning_rate": 1.7025968887119962e-05, + "loss": 0.3028, + "step": 5244 + }, + { + "epoch": 2.2, + "learning_rate": 1.702476235337929e-05, + "loss": 0.2763, + "step": 5245 + }, + { + "epoch": 2.2, + "learning_rate": 1.702355561771805e-05, + "loss": 0.2196, + "step": 5246 + }, + { + "epoch": 2.2, + "learning_rate": 1.7022348680170936e-05, + "loss": 0.2704, + "step": 5247 + }, + { + "epoch": 2.2, + "learning_rate": 1.7021141540772645e-05, + "loss": 0.2302, + "step": 5248 + }, + { + "epoch": 2.2, + "learning_rate": 1.7019934199557868e-05, + "loss": 0.2421, + "step": 5249 + }, + { + "epoch": 2.2, + "learning_rate": 1.7018726656561314e-05, + "loss": 0.254, + "step": 5250 + }, + { + "epoch": 2.2, + "learning_rate": 1.7017518911817686e-05, + "loss": 0.251, + "step": 5251 + }, + { + "epoch": 2.2, + "learning_rate": 1.7016310965361708e-05, + "loss": 0.2674, + "step": 5252 + }, + { + "epoch": 2.2, + "learning_rate": 1.70151028172281e-05, + "loss": 0.268, + "step": 5253 + }, + { + "epoch": 2.2, + "learning_rate": 1.701389446745158e-05, + "loss": 0.2271, + "step": 5254 + }, + { + "epoch": 2.2, + "learning_rate": 1.7012685916066894e-05, + "loss": 0.2591, + "step": 5255 + }, + { + "epoch": 2.2, + "learning_rate": 1.7011477163108766e-05, + "loss": 0.2526, + "step": 5256 + }, + { + "epoch": 2.2, + "learning_rate": 1.7010268208611958e-05, + "loss": 0.271, + "step": 5257 + }, + { + "epoch": 2.2, + "learning_rate": 1.7009059052611206e-05, + "loss": 0.238, + "step": 5258 + }, + { + "epoch": 2.2, + "learning_rate": 1.7007849695141275e-05, + "loss": 0.2589, + "step": 5259 + }, + { + "epoch": 2.2, + "learning_rate": 1.700664013623692e-05, + "loss": 0.2659, + "step": 5260 + }, + { + "epoch": 2.2, + "learning_rate": 1.700543037593291e-05, + "loss": 0.2963, + "step": 5261 + }, + { + "epoch": 2.2, + "learning_rate": 1.7004220414264025e-05, + "loss": 0.2727, + "step": 5262 + }, + { + "epoch": 2.2, + "learning_rate": 1.7003010251265036e-05, + "loss": 0.2739, + "step": 5263 + }, + { + "epoch": 2.2, + "learning_rate": 1.7001799886970735e-05, + "loss": 0.2541, + "step": 5264 + }, + { + "epoch": 2.2, + "learning_rate": 1.7000589321415906e-05, + "loss": 0.2445, + "step": 5265 + }, + { + "epoch": 2.2, + "learning_rate": 1.699937855463535e-05, + "loss": 0.2615, + "step": 5266 + }, + { + "epoch": 2.21, + "learning_rate": 1.6998167586663868e-05, + "loss": 0.2552, + "step": 5267 + }, + { + "epoch": 2.21, + "learning_rate": 1.699695641753627e-05, + "loss": 0.2752, + "step": 5268 + }, + { + "epoch": 2.21, + "learning_rate": 1.6995745047287373e-05, + "loss": 0.2623, + "step": 5269 + }, + { + "epoch": 2.21, + "learning_rate": 1.6994533475951985e-05, + "loss": 0.2388, + "step": 5270 + }, + { + "epoch": 2.21, + "learning_rate": 1.6993321703564946e-05, + "loss": 0.2854, + "step": 5271 + }, + { + "epoch": 2.21, + "learning_rate": 1.6992109730161077e-05, + "loss": 0.264, + "step": 5272 + }, + { + "epoch": 2.21, + "learning_rate": 1.699089755577522e-05, + "loss": 0.2667, + "step": 5273 + }, + { + "epoch": 2.21, + "learning_rate": 1.6989685180442212e-05, + "loss": 0.2484, + "step": 5274 + }, + { + "epoch": 2.21, + "learning_rate": 1.6988472604196907e-05, + "loss": 0.2643, + "step": 5275 + }, + { + "epoch": 2.21, + "learning_rate": 1.698725982707416e-05, + "loss": 0.2616, + "step": 5276 + }, + { + "epoch": 2.21, + "learning_rate": 1.6986046849108834e-05, + "loss": 0.2566, + "step": 5277 + }, + { + "epoch": 2.21, + "learning_rate": 1.6984833670335783e-05, + "loss": 0.26, + "step": 5278 + }, + { + "epoch": 2.21, + "learning_rate": 1.698362029078989e-05, + "loss": 0.2434, + "step": 5279 + }, + { + "epoch": 2.21, + "learning_rate": 1.698240671050603e-05, + "loss": 0.2577, + "step": 5280 + }, + { + "epoch": 2.21, + "learning_rate": 1.6981192929519087e-05, + "loss": 0.2326, + "step": 5281 + }, + { + "epoch": 2.21, + "learning_rate": 1.6979978947863946e-05, + "loss": 0.2519, + "step": 5282 + }, + { + "epoch": 2.21, + "learning_rate": 1.69787647655755e-05, + "loss": 0.2774, + "step": 5283 + }, + { + "epoch": 2.21, + "learning_rate": 1.697755038268866e-05, + "loss": 0.2594, + "step": 5284 + }, + { + "epoch": 2.21, + "learning_rate": 1.6976335799238318e-05, + "loss": 0.2421, + "step": 5285 + }, + { + "epoch": 2.21, + "learning_rate": 1.69751210152594e-05, + "loss": 0.2378, + "step": 5286 + }, + { + "epoch": 2.21, + "learning_rate": 1.697390603078682e-05, + "loss": 0.2452, + "step": 5287 + }, + { + "epoch": 2.21, + "learning_rate": 1.6972690845855494e-05, + "loss": 0.2753, + "step": 5288 + }, + { + "epoch": 2.21, + "learning_rate": 1.697147546050036e-05, + "loss": 0.2547, + "step": 5289 + }, + { + "epoch": 2.21, + "learning_rate": 1.6970259874756353e-05, + "loss": 0.2338, + "step": 5290 + }, + { + "epoch": 2.22, + "learning_rate": 1.696904408865841e-05, + "loss": 0.2653, + "step": 5291 + }, + { + "epoch": 2.22, + "learning_rate": 1.6967828102241475e-05, + "loss": 0.2689, + "step": 5292 + }, + { + "epoch": 2.22, + "learning_rate": 1.6966611915540506e-05, + "loss": 0.2678, + "step": 5293 + }, + { + "epoch": 2.22, + "learning_rate": 1.696539552859046e-05, + "loss": 0.2517, + "step": 5294 + }, + { + "epoch": 2.22, + "learning_rate": 1.69641789414263e-05, + "loss": 0.2531, + "step": 5295 + }, + { + "epoch": 2.22, + "learning_rate": 1.6962962154083e-05, + "loss": 0.263, + "step": 5296 + }, + { + "epoch": 2.22, + "learning_rate": 1.6961745166595528e-05, + "loss": 0.26, + "step": 5297 + }, + { + "epoch": 2.22, + "learning_rate": 1.696052797899887e-05, + "loss": 0.2513, + "step": 5298 + }, + { + "epoch": 2.22, + "learning_rate": 1.695931059132801e-05, + "loss": 0.2297, + "step": 5299 + }, + { + "epoch": 2.22, + "learning_rate": 1.6958093003617942e-05, + "loss": 0.2827, + "step": 5300 + }, + { + "epoch": 2.22, + "learning_rate": 1.695687521590367e-05, + "loss": 0.2749, + "step": 5301 + }, + { + "epoch": 2.22, + "learning_rate": 1.695565722822019e-05, + "loss": 0.2556, + "step": 5302 + }, + { + "epoch": 2.22, + "learning_rate": 1.6954439040602516e-05, + "loss": 0.2822, + "step": 5303 + }, + { + "epoch": 2.22, + "learning_rate": 1.6953220653085662e-05, + "loss": 0.2487, + "step": 5304 + }, + { + "epoch": 2.22, + "learning_rate": 1.6952002065704652e-05, + "loss": 0.2533, + "step": 5305 + }, + { + "epoch": 2.22, + "learning_rate": 1.6950783278494513e-05, + "loss": 0.241, + "step": 5306 + }, + { + "epoch": 2.22, + "learning_rate": 1.6949564291490274e-05, + "loss": 0.2709, + "step": 5307 + }, + { + "epoch": 2.22, + "learning_rate": 1.694834510472698e-05, + "loss": 0.2417, + "step": 5308 + }, + { + "epoch": 2.22, + "learning_rate": 1.6947125718239664e-05, + "loss": 0.2468, + "step": 5309 + }, + { + "epoch": 2.22, + "learning_rate": 1.694590613206339e-05, + "loss": 0.2712, + "step": 5310 + }, + { + "epoch": 2.22, + "learning_rate": 1.6944686346233207e-05, + "loss": 0.2649, + "step": 5311 + }, + { + "epoch": 2.22, + "learning_rate": 1.694346636078418e-05, + "loss": 0.25, + "step": 5312 + }, + { + "epoch": 2.22, + "learning_rate": 1.6942246175751372e-05, + "loss": 0.2659, + "step": 5313 + }, + { + "epoch": 2.22, + "learning_rate": 1.6941025791169855e-05, + "loss": 0.2657, + "step": 5314 + }, + { + "epoch": 2.23, + "learning_rate": 1.6939805207074714e-05, + "loss": 0.253, + "step": 5315 + }, + { + "epoch": 2.23, + "learning_rate": 1.693858442350103e-05, + "loss": 0.2573, + "step": 5316 + }, + { + "epoch": 2.23, + "learning_rate": 1.6937363440483892e-05, + "loss": 0.2307, + "step": 5317 + }, + { + "epoch": 2.23, + "learning_rate": 1.6936142258058403e-05, + "loss": 0.2418, + "step": 5318 + }, + { + "epoch": 2.23, + "learning_rate": 1.6934920876259657e-05, + "loss": 0.2507, + "step": 5319 + }, + { + "epoch": 2.23, + "learning_rate": 1.6933699295122766e-05, + "loss": 0.2765, + "step": 5320 + }, + { + "epoch": 2.23, + "learning_rate": 1.6932477514682838e-05, + "loss": 0.2427, + "step": 5321 + }, + { + "epoch": 2.23, + "learning_rate": 1.6931255534974997e-05, + "loss": 0.2684, + "step": 5322 + }, + { + "epoch": 2.23, + "learning_rate": 1.6930033356034368e-05, + "loss": 0.2533, + "step": 5323 + }, + { + "epoch": 2.23, + "learning_rate": 1.692881097789608e-05, + "loss": 0.2579, + "step": 5324 + }, + { + "epoch": 2.23, + "learning_rate": 1.6927588400595265e-05, + "loss": 0.2453, + "step": 5325 + }, + { + "epoch": 2.23, + "learning_rate": 1.6926365624167075e-05, + "loss": 0.2447, + "step": 5326 + }, + { + "epoch": 2.23, + "learning_rate": 1.6925142648646647e-05, + "loss": 0.2478, + "step": 5327 + }, + { + "epoch": 2.23, + "learning_rate": 1.692391947406914e-05, + "loss": 0.2556, + "step": 5328 + }, + { + "epoch": 2.23, + "learning_rate": 1.692269610046971e-05, + "loss": 0.2543, + "step": 5329 + }, + { + "epoch": 2.23, + "learning_rate": 1.6921472527883527e-05, + "loss": 0.2752, + "step": 5330 + }, + { + "epoch": 2.23, + "learning_rate": 1.6920248756345757e-05, + "loss": 0.2648, + "step": 5331 + }, + { + "epoch": 2.23, + "learning_rate": 1.6919024785891576e-05, + "loss": 0.2655, + "step": 5332 + }, + { + "epoch": 2.23, + "learning_rate": 1.691780061655617e-05, + "loss": 0.2572, + "step": 5333 + }, + { + "epoch": 2.23, + "learning_rate": 1.691657624837472e-05, + "loss": 0.2709, + "step": 5334 + }, + { + "epoch": 2.23, + "learning_rate": 1.6915351681382427e-05, + "loss": 0.2539, + "step": 5335 + }, + { + "epoch": 2.23, + "learning_rate": 1.6914126915614483e-05, + "loss": 0.2223, + "step": 5336 + }, + { + "epoch": 2.23, + "learning_rate": 1.69129019511061e-05, + "loss": 0.263, + "step": 5337 + }, + { + "epoch": 2.23, + "learning_rate": 1.6911676787892484e-05, + "loss": 0.2504, + "step": 5338 + }, + { + "epoch": 2.24, + "learning_rate": 1.691045142600885e-05, + "loss": 0.2343, + "step": 5339 + }, + { + "epoch": 2.24, + "learning_rate": 1.6909225865490424e-05, + "loss": 0.2212, + "step": 5340 + }, + { + "epoch": 2.24, + "learning_rate": 1.6908000106372432e-05, + "loss": 0.2872, + "step": 5341 + }, + { + "epoch": 2.24, + "learning_rate": 1.6906774148690106e-05, + "loss": 0.2537, + "step": 5342 + }, + { + "epoch": 2.24, + "learning_rate": 1.6905547992478687e-05, + "loss": 0.2897, + "step": 5343 + }, + { + "epoch": 2.24, + "learning_rate": 1.6904321637773414e-05, + "loss": 0.2736, + "step": 5344 + }, + { + "epoch": 2.24, + "learning_rate": 1.690309508460955e-05, + "loss": 0.2745, + "step": 5345 + }, + { + "epoch": 2.24, + "learning_rate": 1.6901868333022337e-05, + "loss": 0.2602, + "step": 5346 + }, + { + "epoch": 2.24, + "learning_rate": 1.6900641383047046e-05, + "loss": 0.2841, + "step": 5347 + }, + { + "epoch": 2.24, + "learning_rate": 1.6899414234718943e-05, + "loss": 0.2467, + "step": 5348 + }, + { + "epoch": 2.24, + "learning_rate": 1.68981868880733e-05, + "loss": 0.2693, + "step": 5349 + }, + { + "epoch": 2.24, + "learning_rate": 1.6896959343145396e-05, + "loss": 0.267, + "step": 5350 + }, + { + "epoch": 2.24, + "learning_rate": 1.6895731599970516e-05, + "loss": 0.2706, + "step": 5351 + }, + { + "epoch": 2.24, + "learning_rate": 1.689450365858395e-05, + "loss": 0.2446, + "step": 5352 + }, + { + "epoch": 2.24, + "learning_rate": 1.689327551902099e-05, + "loss": 0.2822, + "step": 5353 + }, + { + "epoch": 2.24, + "learning_rate": 1.6892047181316952e-05, + "loss": 0.2618, + "step": 5354 + }, + { + "epoch": 2.24, + "learning_rate": 1.6890818645507126e-05, + "loss": 0.2618, + "step": 5355 + }, + { + "epoch": 2.24, + "learning_rate": 1.6889589911626834e-05, + "loss": 0.2688, + "step": 5356 + }, + { + "epoch": 2.24, + "learning_rate": 1.6888360979711395e-05, + "loss": 0.2756, + "step": 5357 + }, + { + "epoch": 2.24, + "learning_rate": 1.6887131849796132e-05, + "loss": 0.2537, + "step": 5358 + }, + { + "epoch": 2.24, + "learning_rate": 1.6885902521916375e-05, + "loss": 0.2688, + "step": 5359 + }, + { + "epoch": 2.24, + "learning_rate": 1.6884672996107464e-05, + "loss": 0.2811, + "step": 5360 + }, + { + "epoch": 2.24, + "learning_rate": 1.6883443272404735e-05, + "loss": 0.2519, + "step": 5361 + }, + { + "epoch": 2.24, + "learning_rate": 1.688221335084353e-05, + "loss": 0.2722, + "step": 5362 + }, + { + "epoch": 2.25, + "learning_rate": 1.6880983231459217e-05, + "loss": 0.2626, + "step": 5363 + }, + { + "epoch": 2.25, + "learning_rate": 1.687975291428715e-05, + "loss": 0.2585, + "step": 5364 + }, + { + "epoch": 2.25, + "learning_rate": 1.6878522399362683e-05, + "loss": 0.2621, + "step": 5365 + }, + { + "epoch": 2.25, + "learning_rate": 1.68772916867212e-05, + "loss": 0.218, + "step": 5366 + }, + { + "epoch": 2.25, + "learning_rate": 1.6876060776398063e-05, + "loss": 0.2472, + "step": 5367 + }, + { + "epoch": 2.25, + "learning_rate": 1.6874829668428667e-05, + "loss": 0.2508, + "step": 5368 + }, + { + "epoch": 2.25, + "learning_rate": 1.687359836284839e-05, + "loss": 0.2585, + "step": 5369 + }, + { + "epoch": 2.25, + "learning_rate": 1.687236685969263e-05, + "loss": 0.2585, + "step": 5370 + }, + { + "epoch": 2.25, + "learning_rate": 1.6871135158996782e-05, + "loss": 0.2454, + "step": 5371 + }, + { + "epoch": 2.25, + "learning_rate": 1.686990326079625e-05, + "loss": 0.2705, + "step": 5372 + }, + { + "epoch": 2.25, + "learning_rate": 1.6868671165126445e-05, + "loss": 0.299, + "step": 5373 + }, + { + "epoch": 2.25, + "learning_rate": 1.6867438872022782e-05, + "loss": 0.2597, + "step": 5374 + }, + { + "epoch": 2.25, + "learning_rate": 1.6866206381520683e-05, + "loss": 0.2684, + "step": 5375 + }, + { + "epoch": 2.25, + "learning_rate": 1.6864973693655577e-05, + "loss": 0.243, + "step": 5376 + }, + { + "epoch": 2.25, + "learning_rate": 1.686374080846289e-05, + "loss": 0.2693, + "step": 5377 + }, + { + "epoch": 2.25, + "learning_rate": 1.6862507725978067e-05, + "loss": 0.2694, + "step": 5378 + }, + { + "epoch": 2.25, + "learning_rate": 1.686127444623655e-05, + "loss": 0.2304, + "step": 5379 + }, + { + "epoch": 2.25, + "learning_rate": 1.686004096927378e-05, + "loss": 0.2439, + "step": 5380 + }, + { + "epoch": 2.25, + "learning_rate": 1.685880729512523e-05, + "loss": 0.2477, + "step": 5381 + }, + { + "epoch": 2.25, + "learning_rate": 1.6857573423826344e-05, + "loss": 0.264, + "step": 5382 + }, + { + "epoch": 2.25, + "learning_rate": 1.6856339355412602e-05, + "loss": 0.233, + "step": 5383 + }, + { + "epoch": 2.25, + "learning_rate": 1.6855105089919467e-05, + "loss": 0.2768, + "step": 5384 + }, + { + "epoch": 2.25, + "learning_rate": 1.6853870627382416e-05, + "loss": 0.2539, + "step": 5385 + }, + { + "epoch": 2.25, + "learning_rate": 1.6852635967836936e-05, + "loss": 0.2581, + "step": 5386 + }, + { + "epoch": 2.26, + "learning_rate": 1.685140111131852e-05, + "loss": 0.2704, + "step": 5387 + }, + { + "epoch": 2.26, + "learning_rate": 1.6850166057862654e-05, + "loss": 0.2789, + "step": 5388 + }, + { + "epoch": 2.26, + "learning_rate": 1.6848930807504846e-05, + "loss": 0.2843, + "step": 5389 + }, + { + "epoch": 2.26, + "learning_rate": 1.68476953602806e-05, + "loss": 0.2948, + "step": 5390 + }, + { + "epoch": 2.26, + "learning_rate": 1.684645971622543e-05, + "loss": 0.255, + "step": 5391 + }, + { + "epoch": 2.26, + "learning_rate": 1.6845223875374845e-05, + "loss": 0.2635, + "step": 5392 + }, + { + "epoch": 2.26, + "learning_rate": 1.6843987837764375e-05, + "loss": 0.2963, + "step": 5393 + }, + { + "epoch": 2.26, + "learning_rate": 1.684275160342955e-05, + "loss": 0.2592, + "step": 5394 + }, + { + "epoch": 2.26, + "learning_rate": 1.68415151724059e-05, + "loss": 0.2682, + "step": 5395 + }, + { + "epoch": 2.26, + "learning_rate": 1.684027854472897e-05, + "loss": 0.2681, + "step": 5396 + }, + { + "epoch": 2.26, + "learning_rate": 1.6839041720434303e-05, + "loss": 0.2474, + "step": 5397 + }, + { + "epoch": 2.26, + "learning_rate": 1.683780469955745e-05, + "loss": 0.2892, + "step": 5398 + }, + { + "epoch": 2.26, + "learning_rate": 1.683656748213397e-05, + "loss": 0.2526, + "step": 5399 + }, + { + "epoch": 2.26, + "learning_rate": 1.6835330068199424e-05, + "loss": 0.2477, + "step": 5400 + }, + { + "epoch": 2.26, + "learning_rate": 1.683409245778938e-05, + "loss": 0.2851, + "step": 5401 + }, + { + "epoch": 2.26, + "learning_rate": 1.683285465093941e-05, + "loss": 0.2454, + "step": 5402 + }, + { + "epoch": 2.26, + "learning_rate": 1.6831616647685102e-05, + "loss": 0.2666, + "step": 5403 + }, + { + "epoch": 2.26, + "learning_rate": 1.683037844806203e-05, + "loss": 0.2602, + "step": 5404 + }, + { + "epoch": 2.26, + "learning_rate": 1.6829140052105795e-05, + "loss": 0.2423, + "step": 5405 + }, + { + "epoch": 2.26, + "learning_rate": 1.6827901459851986e-05, + "loss": 0.2727, + "step": 5406 + }, + { + "epoch": 2.26, + "learning_rate": 1.6826662671336212e-05, + "loss": 0.2452, + "step": 5407 + }, + { + "epoch": 2.26, + "learning_rate": 1.6825423686594075e-05, + "loss": 0.2498, + "step": 5408 + }, + { + "epoch": 2.26, + "learning_rate": 1.682418450566119e-05, + "loss": 0.2509, + "step": 5409 + }, + { + "epoch": 2.27, + "learning_rate": 1.6822945128573182e-05, + "loss": 0.2815, + "step": 5410 + }, + { + "epoch": 2.27, + "learning_rate": 1.6821705555365665e-05, + "loss": 0.2747, + "step": 5411 + }, + { + "epoch": 2.27, + "learning_rate": 1.6820465786074277e-05, + "loss": 0.2632, + "step": 5412 + }, + { + "epoch": 2.27, + "learning_rate": 1.6819225820734656e-05, + "loss": 0.2268, + "step": 5413 + }, + { + "epoch": 2.27, + "learning_rate": 1.6817985659382437e-05, + "loss": 0.2475, + "step": 5414 + }, + { + "epoch": 2.27, + "learning_rate": 1.681674530205327e-05, + "loss": 0.2406, + "step": 5415 + }, + { + "epoch": 2.27, + "learning_rate": 1.681550474878281e-05, + "loss": 0.2551, + "step": 5416 + }, + { + "epoch": 2.27, + "learning_rate": 1.6814263999606713e-05, + "loss": 0.2745, + "step": 5417 + }, + { + "epoch": 2.27, + "learning_rate": 1.681302305456064e-05, + "loss": 0.2644, + "step": 5418 + }, + { + "epoch": 2.27, + "learning_rate": 1.6811781913680273e-05, + "loss": 0.2814, + "step": 5419 + }, + { + "epoch": 2.27, + "learning_rate": 1.6810540577001277e-05, + "loss": 0.2529, + "step": 5420 + }, + { + "epoch": 2.27, + "learning_rate": 1.6809299044559333e-05, + "loss": 0.2791, + "step": 5421 + }, + { + "epoch": 2.27, + "learning_rate": 1.6808057316390128e-05, + "loss": 0.255, + "step": 5422 + }, + { + "epoch": 2.27, + "learning_rate": 1.6806815392529363e-05, + "loss": 0.26, + "step": 5423 + }, + { + "epoch": 2.27, + "learning_rate": 1.680557327301273e-05, + "loss": 0.256, + "step": 5424 + }, + { + "epoch": 2.27, + "learning_rate": 1.6804330957875926e-05, + "loss": 0.2486, + "step": 5425 + }, + { + "epoch": 2.27, + "learning_rate": 1.6803088447154672e-05, + "loss": 0.2407, + "step": 5426 + }, + { + "epoch": 2.27, + "learning_rate": 1.6801845740884678e-05, + "loss": 0.2346, + "step": 5427 + }, + { + "epoch": 2.27, + "learning_rate": 1.680060283910166e-05, + "loss": 0.2524, + "step": 5428 + }, + { + "epoch": 2.27, + "learning_rate": 1.6799359741841354e-05, + "loss": 0.2708, + "step": 5429 + }, + { + "epoch": 2.27, + "learning_rate": 1.6798116449139483e-05, + "loss": 0.2632, + "step": 5430 + }, + { + "epoch": 2.27, + "learning_rate": 1.6796872961031783e-05, + "loss": 0.2237, + "step": 5431 + }, + { + "epoch": 2.27, + "learning_rate": 1.6795629277554003e-05, + "loss": 0.2382, + "step": 5432 + }, + { + "epoch": 2.27, + "learning_rate": 1.6794385398741894e-05, + "loss": 0.2691, + "step": 5433 + }, + { + "epoch": 2.28, + "learning_rate": 1.67931413246312e-05, + "loss": 0.2636, + "step": 5434 + }, + { + "epoch": 2.28, + "learning_rate": 1.679189705525769e-05, + "loss": 0.2594, + "step": 5435 + }, + { + "epoch": 2.28, + "learning_rate": 1.6790652590657125e-05, + "loss": 0.2543, + "step": 5436 + }, + { + "epoch": 2.28, + "learning_rate": 1.678940793086528e-05, + "loss": 0.2591, + "step": 5437 + }, + { + "epoch": 2.28, + "learning_rate": 1.678816307591793e-05, + "loss": 0.2541, + "step": 5438 + }, + { + "epoch": 2.28, + "learning_rate": 1.6786918025850852e-05, + "loss": 0.2863, + "step": 5439 + }, + { + "epoch": 2.28, + "learning_rate": 1.6785672780699836e-05, + "loss": 0.2611, + "step": 5440 + }, + { + "epoch": 2.28, + "learning_rate": 1.6784427340500684e-05, + "loss": 0.286, + "step": 5441 + }, + { + "epoch": 2.28, + "learning_rate": 1.6783181705289184e-05, + "loss": 0.244, + "step": 5442 + }, + { + "epoch": 2.28, + "learning_rate": 1.6781935875101148e-05, + "loss": 0.2727, + "step": 5443 + }, + { + "epoch": 2.28, + "learning_rate": 1.678068984997238e-05, + "loss": 0.2597, + "step": 5444 + }, + { + "epoch": 2.28, + "learning_rate": 1.67794436299387e-05, + "loss": 0.2972, + "step": 5445 + }, + { + "epoch": 2.28, + "learning_rate": 1.677819721503593e-05, + "loss": 0.2636, + "step": 5446 + }, + { + "epoch": 2.28, + "learning_rate": 1.6776950605299897e-05, + "loss": 0.2327, + "step": 5447 + }, + { + "epoch": 2.28, + "learning_rate": 1.677570380076643e-05, + "loss": 0.275, + "step": 5448 + }, + { + "epoch": 2.28, + "learning_rate": 1.677445680147137e-05, + "loss": 0.2507, + "step": 5449 + }, + { + "epoch": 2.28, + "learning_rate": 1.6773209607450564e-05, + "loss": 0.2583, + "step": 5450 + }, + { + "epoch": 2.28, + "learning_rate": 1.6771962218739856e-05, + "loss": 0.2828, + "step": 5451 + }, + { + "epoch": 2.28, + "learning_rate": 1.6770714635375104e-05, + "loss": 0.2481, + "step": 5452 + }, + { + "epoch": 2.28, + "learning_rate": 1.6769466857392167e-05, + "loss": 0.2325, + "step": 5453 + }, + { + "epoch": 2.28, + "learning_rate": 1.6768218884826916e-05, + "loss": 0.2542, + "step": 5454 + }, + { + "epoch": 2.28, + "learning_rate": 1.6766970717715216e-05, + "loss": 0.2508, + "step": 5455 + }, + { + "epoch": 2.28, + "learning_rate": 1.6765722356092947e-05, + "loss": 0.256, + "step": 5456 + }, + { + "epoch": 2.28, + "learning_rate": 1.6764473799995993e-05, + "loss": 0.261, + "step": 5457 + }, + { + "epoch": 2.29, + "learning_rate": 1.676322504946024e-05, + "loss": 0.2655, + "step": 5458 + }, + { + "epoch": 2.29, + "learning_rate": 1.676197610452159e-05, + "loss": 0.2763, + "step": 5459 + }, + { + "epoch": 2.29, + "learning_rate": 1.6760726965215936e-05, + "loss": 0.2443, + "step": 5460 + }, + { + "epoch": 2.29, + "learning_rate": 1.6759477631579182e-05, + "loss": 0.2783, + "step": 5461 + }, + { + "epoch": 2.29, + "learning_rate": 1.6758228103647246e-05, + "loss": 0.2794, + "step": 5462 + }, + { + "epoch": 2.29, + "learning_rate": 1.6756978381456035e-05, + "loss": 0.2522, + "step": 5463 + }, + { + "epoch": 2.29, + "learning_rate": 1.675572846504148e-05, + "loss": 0.2734, + "step": 5464 + }, + { + "epoch": 2.29, + "learning_rate": 1.6754478354439504e-05, + "loss": 0.2389, + "step": 5465 + }, + { + "epoch": 2.29, + "learning_rate": 1.6753228049686042e-05, + "loss": 0.2517, + "step": 5466 + }, + { + "epoch": 2.29, + "learning_rate": 1.675197755081703e-05, + "loss": 0.2726, + "step": 5467 + }, + { + "epoch": 2.29, + "learning_rate": 1.675072685786842e-05, + "loss": 0.2493, + "step": 5468 + }, + { + "epoch": 2.29, + "learning_rate": 1.674947597087615e-05, + "loss": 0.2413, + "step": 5469 + }, + { + "epoch": 2.29, + "learning_rate": 1.6748224889876188e-05, + "loss": 0.2812, + "step": 5470 + }, + { + "epoch": 2.29, + "learning_rate": 1.674697361490449e-05, + "loss": 0.2636, + "step": 5471 + }, + { + "epoch": 2.29, + "learning_rate": 1.674572214599702e-05, + "loss": 0.2142, + "step": 5472 + }, + { + "epoch": 2.29, + "learning_rate": 1.6744470483189755e-05, + "loss": 0.2527, + "step": 5473 + }, + { + "epoch": 2.29, + "learning_rate": 1.674321862651867e-05, + "loss": 0.2542, + "step": 5474 + }, + { + "epoch": 2.29, + "learning_rate": 1.674196657601975e-05, + "loss": 0.2612, + "step": 5475 + }, + { + "epoch": 2.29, + "learning_rate": 1.6740714331728984e-05, + "loss": 0.2351, + "step": 5476 + }, + { + "epoch": 2.29, + "learning_rate": 1.673946189368236e-05, + "loss": 0.2701, + "step": 5477 + }, + { + "epoch": 2.29, + "learning_rate": 1.673820926191589e-05, + "loss": 0.2584, + "step": 5478 + }, + { + "epoch": 2.29, + "learning_rate": 1.6736956436465573e-05, + "loss": 0.2623, + "step": 5479 + }, + { + "epoch": 2.29, + "learning_rate": 1.6735703417367422e-05, + "loss": 0.2528, + "step": 5480 + }, + { + "epoch": 2.29, + "learning_rate": 1.673445020465745e-05, + "loss": 0.2376, + "step": 5481 + }, + { + "epoch": 2.3, + "learning_rate": 1.673319679837169e-05, + "loss": 0.2837, + "step": 5482 + }, + { + "epoch": 2.3, + "learning_rate": 1.673194319854616e-05, + "loss": 0.261, + "step": 5483 + }, + { + "epoch": 2.3, + "learning_rate": 1.6730689405216895e-05, + "loss": 0.2461, + "step": 5484 + }, + { + "epoch": 2.3, + "learning_rate": 1.6729435418419936e-05, + "loss": 0.2953, + "step": 5485 + }, + { + "epoch": 2.3, + "learning_rate": 1.6728181238191328e-05, + "loss": 0.2853, + "step": 5486 + }, + { + "epoch": 2.3, + "learning_rate": 1.672692686456712e-05, + "loss": 0.2504, + "step": 5487 + }, + { + "epoch": 2.3, + "learning_rate": 1.672567229758337e-05, + "loss": 0.2589, + "step": 5488 + }, + { + "epoch": 2.3, + "learning_rate": 1.6724417537276138e-05, + "loss": 0.2655, + "step": 5489 + }, + { + "epoch": 2.3, + "learning_rate": 1.6723162583681492e-05, + "loss": 0.249, + "step": 5490 + }, + { + "epoch": 2.3, + "learning_rate": 1.67219074368355e-05, + "loss": 0.2526, + "step": 5491 + }, + { + "epoch": 2.3, + "learning_rate": 1.6720652096774245e-05, + "loss": 0.2313, + "step": 5492 + }, + { + "epoch": 2.3, + "learning_rate": 1.6719396563533812e-05, + "loss": 0.2452, + "step": 5493 + }, + { + "epoch": 2.3, + "learning_rate": 1.6718140837150286e-05, + "loss": 0.2643, + "step": 5494 + }, + { + "epoch": 2.3, + "learning_rate": 1.6716884917659763e-05, + "loss": 0.2852, + "step": 5495 + }, + { + "epoch": 2.3, + "learning_rate": 1.671562880509834e-05, + "loss": 0.2577, + "step": 5496 + }, + { + "epoch": 2.3, + "learning_rate": 1.6714372499502133e-05, + "loss": 0.2387, + "step": 5497 + }, + { + "epoch": 2.3, + "learning_rate": 1.671311600090724e-05, + "loss": 0.2763, + "step": 5498 + }, + { + "epoch": 2.3, + "learning_rate": 1.6711859309349788e-05, + "loss": 0.2555, + "step": 5499 + }, + { + "epoch": 2.3, + "learning_rate": 1.67106024248659e-05, + "loss": 0.2557, + "step": 5500 + }, + { + "epoch": 2.3, + "learning_rate": 1.6709345347491695e-05, + "loss": 0.2474, + "step": 5501 + }, + { + "epoch": 2.3, + "learning_rate": 1.670808807726331e-05, + "loss": 0.261, + "step": 5502 + }, + { + "epoch": 2.3, + "learning_rate": 1.670683061421689e-05, + "loss": 0.247, + "step": 5503 + }, + { + "epoch": 2.3, + "learning_rate": 1.6705572958388576e-05, + "loss": 0.26, + "step": 5504 + }, + { + "epoch": 2.3, + "learning_rate": 1.6704315109814514e-05, + "loss": 0.2832, + "step": 5505 + }, + { + "epoch": 2.31, + "learning_rate": 1.670305706853087e-05, + "loss": 0.2386, + "step": 5506 + }, + { + "epoch": 2.31, + "learning_rate": 1.670179883457379e-05, + "loss": 0.286, + "step": 5507 + }, + { + "epoch": 2.31, + "learning_rate": 1.6700540407979457e-05, + "loss": 0.2268, + "step": 5508 + }, + { + "epoch": 2.31, + "learning_rate": 1.669928178878403e-05, + "loss": 0.2397, + "step": 5509 + }, + { + "epoch": 2.31, + "learning_rate": 1.6698022977023697e-05, + "loss": 0.25, + "step": 5510 + }, + { + "epoch": 2.31, + "learning_rate": 1.6696763972734637e-05, + "loss": 0.2595, + "step": 5511 + }, + { + "epoch": 2.31, + "learning_rate": 1.6695504775953036e-05, + "loss": 0.244, + "step": 5512 + }, + { + "epoch": 2.31, + "learning_rate": 1.6694245386715094e-05, + "loss": 0.2475, + "step": 5513 + }, + { + "epoch": 2.31, + "learning_rate": 1.669298580505701e-05, + "loss": 0.253, + "step": 5514 + }, + { + "epoch": 2.31, + "learning_rate": 1.6691726031014985e-05, + "loss": 0.2752, + "step": 5515 + }, + { + "epoch": 2.31, + "learning_rate": 1.669046606462524e-05, + "loss": 0.2726, + "step": 5516 + }, + { + "epoch": 2.31, + "learning_rate": 1.6689205905923977e-05, + "loss": 0.238, + "step": 5517 + }, + { + "epoch": 2.31, + "learning_rate": 1.6687945554947428e-05, + "loss": 0.2627, + "step": 5518 + }, + { + "epoch": 2.31, + "learning_rate": 1.668668501173182e-05, + "loss": 0.2485, + "step": 5519 + }, + { + "epoch": 2.31, + "learning_rate": 1.6685424276313386e-05, + "loss": 0.2596, + "step": 5520 + }, + { + "epoch": 2.31, + "learning_rate": 1.6684163348728364e-05, + "loss": 0.2893, + "step": 5521 + }, + { + "epoch": 2.31, + "learning_rate": 1.6682902229012997e-05, + "loss": 0.2622, + "step": 5522 + }, + { + "epoch": 2.31, + "learning_rate": 1.6681640917203534e-05, + "loss": 0.2604, + "step": 5523 + }, + { + "epoch": 2.31, + "learning_rate": 1.668037941333623e-05, + "loss": 0.2601, + "step": 5524 + }, + { + "epoch": 2.31, + "learning_rate": 1.6679117717447356e-05, + "loss": 0.2353, + "step": 5525 + }, + { + "epoch": 2.31, + "learning_rate": 1.6677855829573164e-05, + "loss": 0.242, + "step": 5526 + }, + { + "epoch": 2.31, + "learning_rate": 1.6676593749749936e-05, + "loss": 0.2716, + "step": 5527 + }, + { + "epoch": 2.31, + "learning_rate": 1.667533147801394e-05, + "loss": 0.258, + "step": 5528 + }, + { + "epoch": 2.31, + "learning_rate": 1.667406901440147e-05, + "loss": 0.273, + "step": 5529 + }, + { + "epoch": 2.32, + "learning_rate": 1.6672806358948803e-05, + "loss": 0.2658, + "step": 5530 + }, + { + "epoch": 2.32, + "learning_rate": 1.667154351169224e-05, + "loss": 0.2419, + "step": 5531 + }, + { + "epoch": 2.32, + "learning_rate": 1.6670280472668082e-05, + "loss": 0.2512, + "step": 5532 + }, + { + "epoch": 2.32, + "learning_rate": 1.666901724191263e-05, + "loss": 0.2694, + "step": 5533 + }, + { + "epoch": 2.32, + "learning_rate": 1.6667753819462196e-05, + "loss": 0.2649, + "step": 5534 + }, + { + "epoch": 2.32, + "learning_rate": 1.6666490205353092e-05, + "loss": 0.2606, + "step": 5535 + }, + { + "epoch": 2.32, + "learning_rate": 1.6665226399621642e-05, + "loss": 0.2565, + "step": 5536 + }, + { + "epoch": 2.32, + "learning_rate": 1.6663962402304178e-05, + "loss": 0.239, + "step": 5537 + }, + { + "epoch": 2.32, + "learning_rate": 1.666269821343703e-05, + "loss": 0.2428, + "step": 5538 + }, + { + "epoch": 2.32, + "learning_rate": 1.6661433833056527e-05, + "loss": 0.2325, + "step": 5539 + }, + { + "epoch": 2.32, + "learning_rate": 1.6660169261199027e-05, + "loss": 0.2501, + "step": 5540 + }, + { + "epoch": 2.32, + "learning_rate": 1.665890449790087e-05, + "loss": 0.2478, + "step": 5541 + }, + { + "epoch": 2.32, + "learning_rate": 1.665763954319841e-05, + "loss": 0.23, + "step": 5542 + }, + { + "epoch": 2.32, + "learning_rate": 1.6656374397128007e-05, + "loss": 0.2581, + "step": 5543 + }, + { + "epoch": 2.32, + "learning_rate": 1.665510905972603e-05, + "loss": 0.2491, + "step": 5544 + }, + { + "epoch": 2.32, + "learning_rate": 1.6653843531028852e-05, + "loss": 0.255, + "step": 5545 + }, + { + "epoch": 2.32, + "learning_rate": 1.6652577811072842e-05, + "loss": 0.2237, + "step": 5546 + }, + { + "epoch": 2.32, + "learning_rate": 1.665131189989439e-05, + "loss": 0.2275, + "step": 5547 + }, + { + "epoch": 2.32, + "learning_rate": 1.665004579752988e-05, + "loss": 0.2532, + "step": 5548 + }, + { + "epoch": 2.32, + "learning_rate": 1.66487795040157e-05, + "loss": 0.2732, + "step": 5549 + }, + { + "epoch": 2.32, + "learning_rate": 1.664751301938826e-05, + "loss": 0.2887, + "step": 5550 + }, + { + "epoch": 2.32, + "learning_rate": 1.6646246343683953e-05, + "loss": 0.2389, + "step": 5551 + }, + { + "epoch": 2.32, + "learning_rate": 1.6644979476939194e-05, + "loss": 0.2584, + "step": 5552 + }, + { + "epoch": 2.32, + "learning_rate": 1.6643712419190396e-05, + "loss": 0.2517, + "step": 5553 + }, + { + "epoch": 2.33, + "learning_rate": 1.664244517047398e-05, + "loss": 0.2563, + "step": 5554 + }, + { + "epoch": 2.33, + "learning_rate": 1.664117773082637e-05, + "loss": 0.2428, + "step": 5555 + }, + { + "epoch": 2.33, + "learning_rate": 1.6639910100284e-05, + "loss": 0.2671, + "step": 5556 + }, + { + "epoch": 2.33, + "learning_rate": 1.663864227888331e-05, + "loss": 0.2196, + "step": 5557 + }, + { + "epoch": 2.33, + "learning_rate": 1.6637374266660734e-05, + "loss": 0.2565, + "step": 5558 + }, + { + "epoch": 2.33, + "learning_rate": 1.6636106063652725e-05, + "loss": 0.2578, + "step": 5559 + }, + { + "epoch": 2.33, + "learning_rate": 1.663483766989574e-05, + "loss": 0.2725, + "step": 5560 + }, + { + "epoch": 2.33, + "learning_rate": 1.663356908542623e-05, + "loss": 0.2808, + "step": 5561 + }, + { + "epoch": 2.33, + "learning_rate": 1.6632300310280664e-05, + "loss": 0.2417, + "step": 5562 + }, + { + "epoch": 2.33, + "learning_rate": 1.663103134449551e-05, + "loss": 0.2653, + "step": 5563 + }, + { + "epoch": 2.33, + "learning_rate": 1.6629762188107248e-05, + "loss": 0.2614, + "step": 5564 + }, + { + "epoch": 2.33, + "learning_rate": 1.6628492841152357e-05, + "loss": 0.2483, + "step": 5565 + }, + { + "epoch": 2.33, + "learning_rate": 1.6627223303667314e-05, + "loss": 0.2439, + "step": 5566 + }, + { + "epoch": 2.33, + "learning_rate": 1.6625953575688624e-05, + "loss": 0.2314, + "step": 5567 + }, + { + "epoch": 2.33, + "learning_rate": 1.6624683657252773e-05, + "loss": 0.216, + "step": 5568 + }, + { + "epoch": 2.33, + "learning_rate": 1.6623413548396272e-05, + "loss": 0.253, + "step": 5569 + }, + { + "epoch": 2.33, + "learning_rate": 1.6622143249155625e-05, + "loss": 0.2446, + "step": 5570 + }, + { + "epoch": 2.33, + "learning_rate": 1.6620872759567345e-05, + "loss": 0.256, + "step": 5571 + }, + { + "epoch": 2.33, + "learning_rate": 1.6619602079667956e-05, + "loss": 0.2411, + "step": 5572 + }, + { + "epoch": 2.33, + "learning_rate": 1.661833120949398e-05, + "loss": 0.2755, + "step": 5573 + }, + { + "epoch": 2.33, + "learning_rate": 1.6617060149081944e-05, + "loss": 0.2411, + "step": 5574 + }, + { + "epoch": 2.33, + "learning_rate": 1.6615788898468384e-05, + "loss": 0.2439, + "step": 5575 + }, + { + "epoch": 2.33, + "learning_rate": 1.661451745768985e-05, + "loss": 0.2606, + "step": 5576 + }, + { + "epoch": 2.33, + "learning_rate": 1.6613245826782875e-05, + "loss": 0.288, + "step": 5577 + }, + { + "epoch": 2.34, + "learning_rate": 1.6611974005784017e-05, + "loss": 0.2454, + "step": 5578 + }, + { + "epoch": 2.34, + "learning_rate": 1.6610701994729838e-05, + "loss": 0.2253, + "step": 5579 + }, + { + "epoch": 2.34, + "learning_rate": 1.660942979365689e-05, + "loss": 0.2392, + "step": 5580 + }, + { + "epoch": 2.34, + "learning_rate": 1.6608157402601753e-05, + "loss": 0.2397, + "step": 5581 + }, + { + "epoch": 2.34, + "learning_rate": 1.6606884821600995e-05, + "loss": 0.2303, + "step": 5582 + }, + { + "epoch": 2.34, + "learning_rate": 1.6605612050691197e-05, + "loss": 0.2431, + "step": 5583 + }, + { + "epoch": 2.34, + "learning_rate": 1.6604339089908935e-05, + "loss": 0.2579, + "step": 5584 + }, + { + "epoch": 2.34, + "learning_rate": 1.6603065939290815e-05, + "loss": 0.2297, + "step": 5585 + }, + { + "epoch": 2.34, + "learning_rate": 1.660179259887342e-05, + "loss": 0.2695, + "step": 5586 + }, + { + "epoch": 2.34, + "learning_rate": 1.6600519068693356e-05, + "loss": 0.2433, + "step": 5587 + }, + { + "epoch": 2.34, + "learning_rate": 1.659924534878723e-05, + "loss": 0.2382, + "step": 5588 + }, + { + "epoch": 2.34, + "learning_rate": 1.659797143919165e-05, + "loss": 0.2639, + "step": 5589 + }, + { + "epoch": 2.34, + "learning_rate": 1.6596697339943234e-05, + "loss": 0.2494, + "step": 5590 + }, + { + "epoch": 2.34, + "learning_rate": 1.659542305107861e-05, + "loss": 0.283, + "step": 5591 + }, + { + "epoch": 2.34, + "learning_rate": 1.6594148572634404e-05, + "loss": 0.2544, + "step": 5592 + }, + { + "epoch": 2.34, + "learning_rate": 1.6592873904647245e-05, + "loss": 0.2498, + "step": 5593 + }, + { + "epoch": 2.34, + "learning_rate": 1.6591599047153778e-05, + "loss": 0.2804, + "step": 5594 + }, + { + "epoch": 2.34, + "learning_rate": 1.6590324000190645e-05, + "loss": 0.2287, + "step": 5595 + }, + { + "epoch": 2.34, + "learning_rate": 1.65890487637945e-05, + "loss": 0.2674, + "step": 5596 + }, + { + "epoch": 2.34, + "learning_rate": 1.6587773338001993e-05, + "loss": 0.2598, + "step": 5597 + }, + { + "epoch": 2.34, + "learning_rate": 1.658649772284979e-05, + "loss": 0.2656, + "step": 5598 + }, + { + "epoch": 2.34, + "learning_rate": 1.658522191837455e-05, + "loss": 0.2451, + "step": 5599 + }, + { + "epoch": 2.34, + "learning_rate": 1.658394592461295e-05, + "loss": 0.2575, + "step": 5600 + }, + { + "epoch": 2.34, + "learning_rate": 1.658266974160167e-05, + "loss": 0.259, + "step": 5601 + }, + { + "epoch": 2.35, + "learning_rate": 1.6581393369377385e-05, + "loss": 0.2611, + "step": 5602 + }, + { + "epoch": 2.35, + "learning_rate": 1.6580116807976795e-05, + "loss": 0.2447, + "step": 5603 + }, + { + "epoch": 2.35, + "learning_rate": 1.657884005743658e-05, + "loss": 0.2716, + "step": 5604 + }, + { + "epoch": 2.35, + "learning_rate": 1.6577563117793447e-05, + "loss": 0.2492, + "step": 5605 + }, + { + "epoch": 2.35, + "learning_rate": 1.65762859890841e-05, + "loss": 0.2713, + "step": 5606 + }, + { + "epoch": 2.35, + "learning_rate": 1.6575008671345248e-05, + "loss": 0.2381, + "step": 5607 + }, + { + "epoch": 2.35, + "learning_rate": 1.6573731164613604e-05, + "loss": 0.2521, + "step": 5608 + }, + { + "epoch": 2.35, + "learning_rate": 1.6572453468925895e-05, + "loss": 0.2688, + "step": 5609 + }, + { + "epoch": 2.35, + "learning_rate": 1.6571175584318845e-05, + "loss": 0.2438, + "step": 5610 + }, + { + "epoch": 2.35, + "learning_rate": 1.656989751082918e-05, + "loss": 0.2655, + "step": 5611 + }, + { + "epoch": 2.35, + "learning_rate": 1.656861924849364e-05, + "loss": 0.2333, + "step": 5612 + }, + { + "epoch": 2.35, + "learning_rate": 1.656734079734897e-05, + "loss": 0.277, + "step": 5613 + }, + { + "epoch": 2.35, + "learning_rate": 1.6566062157431915e-05, + "loss": 0.2572, + "step": 5614 + }, + { + "epoch": 2.35, + "learning_rate": 1.656478332877923e-05, + "loss": 0.2439, + "step": 5615 + }, + { + "epoch": 2.35, + "learning_rate": 1.6563504311427678e-05, + "loss": 0.2518, + "step": 5616 + }, + { + "epoch": 2.35, + "learning_rate": 1.6562225105414016e-05, + "loss": 0.2696, + "step": 5617 + }, + { + "epoch": 2.35, + "learning_rate": 1.656094571077502e-05, + "loss": 0.2628, + "step": 5618 + }, + { + "epoch": 2.35, + "learning_rate": 1.6559666127547455e-05, + "loss": 0.2394, + "step": 5619 + }, + { + "epoch": 2.35, + "learning_rate": 1.655838635576811e-05, + "loss": 0.2871, + "step": 5620 + }, + { + "epoch": 2.35, + "learning_rate": 1.6557106395473775e-05, + "loss": 0.2729, + "step": 5621 + }, + { + "epoch": 2.35, + "learning_rate": 1.655582624670123e-05, + "loss": 0.2643, + "step": 5622 + }, + { + "epoch": 2.35, + "learning_rate": 1.6554545909487277e-05, + "loss": 0.2522, + "step": 5623 + }, + { + "epoch": 2.35, + "learning_rate": 1.655326538386872e-05, + "loss": 0.2616, + "step": 5624 + }, + { + "epoch": 2.36, + "learning_rate": 1.6551984669882362e-05, + "loss": 0.2816, + "step": 5625 + }, + { + "epoch": 2.36, + "learning_rate": 1.655070376756502e-05, + "loss": 0.2558, + "step": 5626 + }, + { + "epoch": 2.36, + "learning_rate": 1.6549422676953513e-05, + "loss": 0.2902, + "step": 5627 + }, + { + "epoch": 2.36, + "learning_rate": 1.6548141398084662e-05, + "loss": 0.2461, + "step": 5628 + }, + { + "epoch": 2.36, + "learning_rate": 1.6546859930995295e-05, + "loss": 0.2601, + "step": 5629 + }, + { + "epoch": 2.36, + "learning_rate": 1.654557827572225e-05, + "loss": 0.2647, + "step": 5630 + }, + { + "epoch": 2.36, + "learning_rate": 1.6544296432302373e-05, + "loss": 0.2559, + "step": 5631 + }, + { + "epoch": 2.36, + "learning_rate": 1.6543014400772496e-05, + "loss": 0.2638, + "step": 5632 + }, + { + "epoch": 2.36, + "learning_rate": 1.6541732181169474e-05, + "loss": 0.2409, + "step": 5633 + }, + { + "epoch": 2.36, + "learning_rate": 1.6540449773530168e-05, + "loss": 0.2524, + "step": 5634 + }, + { + "epoch": 2.36, + "learning_rate": 1.6539167177891436e-05, + "loss": 0.2701, + "step": 5635 + }, + { + "epoch": 2.36, + "learning_rate": 1.653788439429015e-05, + "loss": 0.2518, + "step": 5636 + }, + { + "epoch": 2.36, + "learning_rate": 1.6536601422763173e-05, + "loss": 0.2835, + "step": 5637 + }, + { + "epoch": 2.36, + "learning_rate": 1.6535318263347395e-05, + "loss": 0.2598, + "step": 5638 + }, + { + "epoch": 2.36, + "learning_rate": 1.6534034916079686e-05, + "loss": 0.2588, + "step": 5639 + }, + { + "epoch": 2.36, + "learning_rate": 1.653275138099695e-05, + "loss": 0.222, + "step": 5640 + }, + { + "epoch": 2.36, + "learning_rate": 1.6531467658136064e-05, + "loss": 0.2774, + "step": 5641 + }, + { + "epoch": 2.36, + "learning_rate": 1.6530183747533942e-05, + "loss": 0.2674, + "step": 5642 + }, + { + "epoch": 2.36, + "learning_rate": 1.6528899649227482e-05, + "loss": 0.26, + "step": 5643 + }, + { + "epoch": 2.36, + "learning_rate": 1.652761536325359e-05, + "loss": 0.2622, + "step": 5644 + }, + { + "epoch": 2.36, + "learning_rate": 1.6526330889649194e-05, + "loss": 0.2339, + "step": 5645 + }, + { + "epoch": 2.36, + "learning_rate": 1.6525046228451204e-05, + "loss": 0.279, + "step": 5646 + }, + { + "epoch": 2.36, + "learning_rate": 1.652376137969655e-05, + "loss": 0.2597, + "step": 5647 + }, + { + "epoch": 2.36, + "learning_rate": 1.6522476343422164e-05, + "loss": 0.2777, + "step": 5648 + }, + { + "epoch": 2.37, + "learning_rate": 1.652119111966499e-05, + "loss": 0.2719, + "step": 5649 + }, + { + "epoch": 2.37, + "learning_rate": 1.6519905708461953e-05, + "loss": 0.2532, + "step": 5650 + }, + { + "epoch": 2.37, + "learning_rate": 1.651862010985002e-05, + "loss": 0.2652, + "step": 5651 + }, + { + "epoch": 2.37, + "learning_rate": 1.6517334323866136e-05, + "loss": 0.2313, + "step": 5652 + }, + { + "epoch": 2.37, + "learning_rate": 1.651604835054726e-05, + "loss": 0.253, + "step": 5653 + }, + { + "epoch": 2.37, + "learning_rate": 1.6514762189930353e-05, + "loss": 0.2641, + "step": 5654 + }, + { + "epoch": 2.37, + "learning_rate": 1.651347584205239e-05, + "loss": 0.2449, + "step": 5655 + }, + { + "epoch": 2.37, + "learning_rate": 1.6512189306950345e-05, + "loss": 0.2649, + "step": 5656 + }, + { + "epoch": 2.37, + "learning_rate": 1.6510902584661195e-05, + "loss": 0.2837, + "step": 5657 + }, + { + "epoch": 2.37, + "learning_rate": 1.650961567522193e-05, + "loss": 0.2499, + "step": 5658 + }, + { + "epoch": 2.37, + "learning_rate": 1.6508328578669537e-05, + "loss": 0.2475, + "step": 5659 + }, + { + "epoch": 2.37, + "learning_rate": 1.6507041295041015e-05, + "loss": 0.246, + "step": 5660 + }, + { + "epoch": 2.37, + "learning_rate": 1.6505753824373364e-05, + "loss": 0.2631, + "step": 5661 + }, + { + "epoch": 2.37, + "learning_rate": 1.6504466166703593e-05, + "loss": 0.2594, + "step": 5662 + }, + { + "epoch": 2.37, + "learning_rate": 1.6503178322068718e-05, + "loss": 0.2271, + "step": 5663 + }, + { + "epoch": 2.37, + "learning_rate": 1.650189029050575e-05, + "loss": 0.2303, + "step": 5664 + }, + { + "epoch": 2.37, + "learning_rate": 1.6500602072051714e-05, + "loss": 0.2522, + "step": 5665 + }, + { + "epoch": 2.37, + "learning_rate": 1.6499313666743643e-05, + "loss": 0.2741, + "step": 5666 + }, + { + "epoch": 2.37, + "learning_rate": 1.6498025074618563e-05, + "loss": 0.2291, + "step": 5667 + }, + { + "epoch": 2.37, + "learning_rate": 1.6496736295713526e-05, + "loss": 0.2841, + "step": 5668 + }, + { + "epoch": 2.37, + "learning_rate": 1.649544733006556e-05, + "loss": 0.246, + "step": 5669 + }, + { + "epoch": 2.37, + "learning_rate": 1.649415817771173e-05, + "loss": 0.2773, + "step": 5670 + }, + { + "epoch": 2.37, + "learning_rate": 1.649286883868909e-05, + "loss": 0.2308, + "step": 5671 + }, + { + "epoch": 2.37, + "learning_rate": 1.649157931303469e-05, + "loss": 0.244, + "step": 5672 + }, + { + "epoch": 2.38, + "learning_rate": 1.6490289600785603e-05, + "loss": 0.2649, + "step": 5673 + }, + { + "epoch": 2.38, + "learning_rate": 1.6488999701978905e-05, + "loss": 0.2718, + "step": 5674 + }, + { + "epoch": 2.38, + "learning_rate": 1.6487709616651667e-05, + "loss": 0.2615, + "step": 5675 + }, + { + "epoch": 2.38, + "learning_rate": 1.6486419344840974e-05, + "loss": 0.2741, + "step": 5676 + }, + { + "epoch": 2.38, + "learning_rate": 1.6485128886583913e-05, + "loss": 0.2812, + "step": 5677 + }, + { + "epoch": 2.38, + "learning_rate": 1.6483838241917577e-05, + "loss": 0.2596, + "step": 5678 + }, + { + "epoch": 2.38, + "learning_rate": 1.6482547410879063e-05, + "loss": 0.2359, + "step": 5679 + }, + { + "epoch": 2.38, + "learning_rate": 1.6481256393505477e-05, + "loss": 0.2907, + "step": 5680 + }, + { + "epoch": 2.38, + "learning_rate": 1.6479965189833923e-05, + "loss": 0.2376, + "step": 5681 + }, + { + "epoch": 2.38, + "learning_rate": 1.6478673799901525e-05, + "loss": 0.2418, + "step": 5682 + }, + { + "epoch": 2.38, + "learning_rate": 1.6477382223745396e-05, + "loss": 0.2486, + "step": 5683 + }, + { + "epoch": 2.38, + "learning_rate": 1.6476090461402664e-05, + "loss": 0.2549, + "step": 5684 + }, + { + "epoch": 2.38, + "learning_rate": 1.647479851291046e-05, + "loss": 0.2417, + "step": 5685 + }, + { + "epoch": 2.38, + "learning_rate": 1.6473506378305914e-05, + "loss": 0.2823, + "step": 5686 + }, + { + "epoch": 2.38, + "learning_rate": 1.6472214057626175e-05, + "loss": 0.2542, + "step": 5687 + }, + { + "epoch": 2.38, + "learning_rate": 1.6470921550908385e-05, + "loss": 0.2549, + "step": 5688 + }, + { + "epoch": 2.38, + "learning_rate": 1.6469628858189698e-05, + "loss": 0.2689, + "step": 5689 + }, + { + "epoch": 2.38, + "learning_rate": 1.646833597950727e-05, + "loss": 0.2706, + "step": 5690 + }, + { + "epoch": 2.38, + "learning_rate": 1.6467042914898262e-05, + "loss": 0.2514, + "step": 5691 + }, + { + "epoch": 2.38, + "learning_rate": 1.6465749664399848e-05, + "loss": 0.2491, + "step": 5692 + }, + { + "epoch": 2.38, + "learning_rate": 1.6464456228049195e-05, + "loss": 0.2418, + "step": 5693 + }, + { + "epoch": 2.38, + "learning_rate": 1.6463162605883485e-05, + "loss": 0.2389, + "step": 5694 + }, + { + "epoch": 2.38, + "learning_rate": 1.64618687979399e-05, + "loss": 0.2294, + "step": 5695 + }, + { + "epoch": 2.38, + "learning_rate": 1.6460574804255634e-05, + "loss": 0.2445, + "step": 5696 + }, + { + "epoch": 2.39, + "learning_rate": 1.6459280624867876e-05, + "loss": 0.2614, + "step": 5697 + }, + { + "epoch": 2.39, + "learning_rate": 1.6457986259813827e-05, + "loss": 0.2873, + "step": 5698 + }, + { + "epoch": 2.39, + "learning_rate": 1.6456691709130694e-05, + "loss": 0.251, + "step": 5699 + }, + { + "epoch": 2.39, + "learning_rate": 1.6455396972855688e-05, + "loss": 0.2761, + "step": 5700 + }, + { + "epoch": 2.39, + "learning_rate": 1.6454102051026025e-05, + "loss": 0.2521, + "step": 5701 + }, + { + "epoch": 2.39, + "learning_rate": 1.6452806943678928e-05, + "loss": 0.2566, + "step": 5702 + }, + { + "epoch": 2.39, + "learning_rate": 1.645151165085162e-05, + "loss": 0.2531, + "step": 5703 + }, + { + "epoch": 2.39, + "learning_rate": 1.645021617258133e-05, + "loss": 0.2512, + "step": 5704 + }, + { + "epoch": 2.39, + "learning_rate": 1.6448920508905305e-05, + "loss": 0.239, + "step": 5705 + }, + { + "epoch": 2.39, + "learning_rate": 1.644762465986078e-05, + "loss": 0.2325, + "step": 5706 + }, + { + "epoch": 2.39, + "learning_rate": 1.6446328625485007e-05, + "loss": 0.2521, + "step": 5707 + }, + { + "epoch": 2.39, + "learning_rate": 1.644503240581524e-05, + "loss": 0.2613, + "step": 5708 + }, + { + "epoch": 2.39, + "learning_rate": 1.6443736000888732e-05, + "loss": 0.2535, + "step": 5709 + }, + { + "epoch": 2.39, + "learning_rate": 1.6442439410742753e-05, + "loss": 0.2512, + "step": 5710 + }, + { + "epoch": 2.39, + "learning_rate": 1.6441142635414568e-05, + "loss": 0.2478, + "step": 5711 + }, + { + "epoch": 2.39, + "learning_rate": 1.6439845674941458e-05, + "loss": 0.2688, + "step": 5712 + }, + { + "epoch": 2.39, + "learning_rate": 1.6438548529360698e-05, + "loss": 0.2423, + "step": 5713 + }, + { + "epoch": 2.39, + "learning_rate": 1.6437251198709567e-05, + "loss": 0.2904, + "step": 5714 + }, + { + "epoch": 2.39, + "learning_rate": 1.643595368302537e-05, + "loss": 0.2544, + "step": 5715 + }, + { + "epoch": 2.39, + "learning_rate": 1.643465598234539e-05, + "loss": 0.2706, + "step": 5716 + }, + { + "epoch": 2.39, + "learning_rate": 1.6433358096706937e-05, + "loss": 0.2799, + "step": 5717 + }, + { + "epoch": 2.39, + "learning_rate": 1.6432060026147313e-05, + "loss": 0.2739, + "step": 5718 + }, + { + "epoch": 2.39, + "learning_rate": 1.6430761770703836e-05, + "loss": 0.2693, + "step": 5719 + }, + { + "epoch": 2.39, + "learning_rate": 1.6429463330413812e-05, + "loss": 0.2574, + "step": 5720 + }, + { + "epoch": 2.4, + "learning_rate": 1.6428164705314572e-05, + "loss": 0.2597, + "step": 5721 + }, + { + "epoch": 2.4, + "learning_rate": 1.6426865895443446e-05, + "loss": 0.2549, + "step": 5722 + }, + { + "epoch": 2.4, + "learning_rate": 1.642556690083776e-05, + "loss": 0.2729, + "step": 5723 + }, + { + "epoch": 2.4, + "learning_rate": 1.6424267721534856e-05, + "loss": 0.2518, + "step": 5724 + }, + { + "epoch": 2.4, + "learning_rate": 1.6422968357572076e-05, + "loss": 0.2618, + "step": 5725 + }, + { + "epoch": 2.4, + "learning_rate": 1.642166880898677e-05, + "loss": 0.2654, + "step": 5726 + }, + { + "epoch": 2.4, + "learning_rate": 1.642036907581629e-05, + "loss": 0.2463, + "step": 5727 + }, + { + "epoch": 2.4, + "learning_rate": 1.6419069158098004e-05, + "loss": 0.2646, + "step": 5728 + }, + { + "epoch": 2.4, + "learning_rate": 1.6417769055869266e-05, + "loss": 0.2662, + "step": 5729 + }, + { + "epoch": 2.4, + "learning_rate": 1.641646876916745e-05, + "loss": 0.2697, + "step": 5730 + }, + { + "epoch": 2.4, + "learning_rate": 1.6415168298029943e-05, + "loss": 0.241, + "step": 5731 + }, + { + "epoch": 2.4, + "learning_rate": 1.6413867642494106e-05, + "loss": 0.2836, + "step": 5732 + }, + { + "epoch": 2.4, + "learning_rate": 1.641256680259734e-05, + "loss": 0.2531, + "step": 5733 + }, + { + "epoch": 2.4, + "learning_rate": 1.641126577837703e-05, + "loss": 0.2368, + "step": 5734 + }, + { + "epoch": 2.4, + "learning_rate": 1.640996456987058e-05, + "loss": 0.2644, + "step": 5735 + }, + { + "epoch": 2.4, + "learning_rate": 1.6408663177115382e-05, + "loss": 0.2916, + "step": 5736 + }, + { + "epoch": 2.4, + "learning_rate": 1.6407361600148845e-05, + "loss": 0.2774, + "step": 5737 + }, + { + "epoch": 2.4, + "learning_rate": 1.6406059839008387e-05, + "loss": 0.2308, + "step": 5738 + }, + { + "epoch": 2.4, + "learning_rate": 1.6404757893731424e-05, + "loss": 0.2639, + "step": 5739 + }, + { + "epoch": 2.4, + "learning_rate": 1.640345576435538e-05, + "loss": 0.2969, + "step": 5740 + }, + { + "epoch": 2.4, + "learning_rate": 1.640215345091768e-05, + "loss": 0.2689, + "step": 5741 + }, + { + "epoch": 2.4, + "learning_rate": 1.640085095345576e-05, + "loss": 0.2603, + "step": 5742 + }, + { + "epoch": 2.4, + "learning_rate": 1.6399548272007063e-05, + "loss": 0.2703, + "step": 5743 + }, + { + "epoch": 2.4, + "learning_rate": 1.6398245406609027e-05, + "loss": 0.2651, + "step": 5744 + }, + { + "epoch": 2.41, + "learning_rate": 1.6396942357299104e-05, + "loss": 0.2588, + "step": 5745 + }, + { + "epoch": 2.41, + "learning_rate": 1.639563912411475e-05, + "loss": 0.2697, + "step": 5746 + }, + { + "epoch": 2.41, + "learning_rate": 1.6394335707093423e-05, + "loss": 0.2693, + "step": 5747 + }, + { + "epoch": 2.41, + "learning_rate": 1.6393032106272594e-05, + "loss": 0.2228, + "step": 5748 + }, + { + "epoch": 2.41, + "learning_rate": 1.639172832168973e-05, + "loss": 0.2704, + "step": 5749 + }, + { + "epoch": 2.41, + "learning_rate": 1.639042435338231e-05, + "loss": 0.2572, + "step": 5750 + }, + { + "epoch": 2.41, + "learning_rate": 1.6389120201387806e-05, + "loss": 0.2515, + "step": 5751 + }, + { + "epoch": 2.41, + "learning_rate": 1.6387815865743712e-05, + "loss": 0.2549, + "step": 5752 + }, + { + "epoch": 2.41, + "learning_rate": 1.638651134648752e-05, + "loss": 0.244, + "step": 5753 + }, + { + "epoch": 2.41, + "learning_rate": 1.638520664365673e-05, + "loss": 0.272, + "step": 5754 + }, + { + "epoch": 2.41, + "learning_rate": 1.638390175728884e-05, + "loss": 0.2539, + "step": 5755 + }, + { + "epoch": 2.41, + "learning_rate": 1.638259668742136e-05, + "loss": 0.251, + "step": 5756 + }, + { + "epoch": 2.41, + "learning_rate": 1.63812914340918e-05, + "loss": 0.2558, + "step": 5757 + }, + { + "epoch": 2.41, + "learning_rate": 1.6379985997337678e-05, + "loss": 0.2609, + "step": 5758 + }, + { + "epoch": 2.41, + "learning_rate": 1.6378680377196526e-05, + "loss": 0.2463, + "step": 5759 + }, + { + "epoch": 2.41, + "learning_rate": 1.6377374573705863e-05, + "loss": 0.2639, + "step": 5760 + }, + { + "epoch": 2.41, + "learning_rate": 1.6376068586903227e-05, + "loss": 0.272, + "step": 5761 + }, + { + "epoch": 2.41, + "learning_rate": 1.6374762416826157e-05, + "loss": 0.2646, + "step": 5762 + }, + { + "epoch": 2.41, + "learning_rate": 1.63734560635122e-05, + "loss": 0.251, + "step": 5763 + }, + { + "epoch": 2.41, + "learning_rate": 1.6372149526998905e-05, + "loss": 0.253, + "step": 5764 + }, + { + "epoch": 2.41, + "learning_rate": 1.6370842807323823e-05, + "loss": 0.243, + "step": 5765 + }, + { + "epoch": 2.41, + "learning_rate": 1.636953590452452e-05, + "loss": 0.2471, + "step": 5766 + }, + { + "epoch": 2.41, + "learning_rate": 1.636822881863856e-05, + "loss": 0.2734, + "step": 5767 + }, + { + "epoch": 2.41, + "learning_rate": 1.6366921549703513e-05, + "loss": 0.2282, + "step": 5768 + }, + { + "epoch": 2.42, + "learning_rate": 1.6365614097756956e-05, + "loss": 0.257, + "step": 5769 + }, + { + "epoch": 2.42, + "learning_rate": 1.636430646283647e-05, + "loss": 0.2875, + "step": 5770 + }, + { + "epoch": 2.42, + "learning_rate": 1.6362998644979645e-05, + "loss": 0.2659, + "step": 5771 + }, + { + "epoch": 2.42, + "learning_rate": 1.6361690644224066e-05, + "loss": 0.2742, + "step": 5772 + }, + { + "epoch": 2.42, + "learning_rate": 1.636038246060734e-05, + "loss": 0.2699, + "step": 5773 + }, + { + "epoch": 2.42, + "learning_rate": 1.635907409416706e-05, + "loss": 0.2658, + "step": 5774 + }, + { + "epoch": 2.42, + "learning_rate": 1.6357765544940837e-05, + "loss": 0.2769, + "step": 5775 + }, + { + "epoch": 2.42, + "learning_rate": 1.635645681296629e-05, + "loss": 0.2656, + "step": 5776 + }, + { + "epoch": 2.42, + "learning_rate": 1.6355147898281026e-05, + "loss": 0.2624, + "step": 5777 + }, + { + "epoch": 2.42, + "learning_rate": 1.6353838800922682e-05, + "loss": 0.2353, + "step": 5778 + }, + { + "epoch": 2.42, + "learning_rate": 1.6352529520928877e-05, + "loss": 0.2584, + "step": 5779 + }, + { + "epoch": 2.42, + "learning_rate": 1.6351220058337248e-05, + "loss": 0.2717, + "step": 5780 + }, + { + "epoch": 2.42, + "learning_rate": 1.6349910413185438e-05, + "loss": 0.2829, + "step": 5781 + }, + { + "epoch": 2.42, + "learning_rate": 1.6348600585511082e-05, + "loss": 0.2519, + "step": 5782 + }, + { + "epoch": 2.42, + "learning_rate": 1.634729057535184e-05, + "loss": 0.2674, + "step": 5783 + }, + { + "epoch": 2.42, + "learning_rate": 1.634598038274536e-05, + "loss": 0.2715, + "step": 5784 + }, + { + "epoch": 2.42, + "learning_rate": 1.6344670007729305e-05, + "loss": 0.2771, + "step": 5785 + }, + { + "epoch": 2.42, + "learning_rate": 1.6343359450341344e-05, + "loss": 0.2167, + "step": 5786 + }, + { + "epoch": 2.42, + "learning_rate": 1.6342048710619143e-05, + "loss": 0.2417, + "step": 5787 + }, + { + "epoch": 2.42, + "learning_rate": 1.6340737788600378e-05, + "loss": 0.2649, + "step": 5788 + }, + { + "epoch": 2.42, + "learning_rate": 1.6339426684322735e-05, + "loss": 0.2508, + "step": 5789 + }, + { + "epoch": 2.42, + "learning_rate": 1.6338115397823897e-05, + "loss": 0.2406, + "step": 5790 + }, + { + "epoch": 2.42, + "learning_rate": 1.6336803929141557e-05, + "loss": 0.2676, + "step": 5791 + }, + { + "epoch": 2.42, + "learning_rate": 1.6335492278313405e-05, + "loss": 0.2332, + "step": 5792 + }, + { + "epoch": 2.43, + "learning_rate": 1.6334180445377156e-05, + "loss": 0.251, + "step": 5793 + }, + { + "epoch": 2.43, + "learning_rate": 1.6332868430370508e-05, + "loss": 0.2471, + "step": 5794 + }, + { + "epoch": 2.43, + "learning_rate": 1.6331556233331178e-05, + "loss": 0.2658, + "step": 5795 + }, + { + "epoch": 2.43, + "learning_rate": 1.633024385429688e-05, + "loss": 0.2872, + "step": 5796 + }, + { + "epoch": 2.43, + "learning_rate": 1.6328931293305345e-05, + "loss": 0.2564, + "step": 5797 + }, + { + "epoch": 2.43, + "learning_rate": 1.632761855039429e-05, + "loss": 0.2626, + "step": 5798 + }, + { + "epoch": 2.43, + "learning_rate": 1.632630562560146e-05, + "loss": 0.2719, + "step": 5799 + }, + { + "epoch": 2.43, + "learning_rate": 1.6324992518964586e-05, + "loss": 0.248, + "step": 5800 + }, + { + "epoch": 2.43, + "learning_rate": 1.6323679230521413e-05, + "loss": 0.2723, + "step": 5801 + }, + { + "epoch": 2.43, + "learning_rate": 1.6322365760309698e-05, + "loss": 0.2574, + "step": 5802 + }, + { + "epoch": 2.43, + "learning_rate": 1.6321052108367187e-05, + "loss": 0.2379, + "step": 5803 + }, + { + "epoch": 2.43, + "learning_rate": 1.631973827473164e-05, + "loss": 0.2748, + "step": 5804 + }, + { + "epoch": 2.43, + "learning_rate": 1.6318424259440827e-05, + "loss": 0.2571, + "step": 5805 + }, + { + "epoch": 2.43, + "learning_rate": 1.631711006253251e-05, + "loss": 0.266, + "step": 5806 + }, + { + "epoch": 2.43, + "learning_rate": 1.6315795684044476e-05, + "loss": 0.2561, + "step": 5807 + }, + { + "epoch": 2.43, + "learning_rate": 1.63144811240145e-05, + "loss": 0.2786, + "step": 5808 + }, + { + "epoch": 2.43, + "learning_rate": 1.6313166382480362e-05, + "loss": 0.2648, + "step": 5809 + }, + { + "epoch": 2.43, + "learning_rate": 1.631185145947986e-05, + "loss": 0.2931, + "step": 5810 + }, + { + "epoch": 2.43, + "learning_rate": 1.631053635505079e-05, + "loss": 0.2583, + "step": 5811 + }, + { + "epoch": 2.43, + "learning_rate": 1.6309221069230954e-05, + "loss": 0.2413, + "step": 5812 + }, + { + "epoch": 2.43, + "learning_rate": 1.6307905602058158e-05, + "loss": 0.2587, + "step": 5813 + }, + { + "epoch": 2.43, + "learning_rate": 1.6306589953570207e-05, + "loss": 0.2623, + "step": 5814 + }, + { + "epoch": 2.43, + "learning_rate": 1.6305274123804934e-05, + "loss": 0.2557, + "step": 5815 + }, + { + "epoch": 2.44, + "learning_rate": 1.6303958112800144e-05, + "loss": 0.2457, + "step": 5816 + }, + { + "epoch": 2.44, + "learning_rate": 1.630264192059367e-05, + "loss": 0.2665, + "step": 5817 + }, + { + "epoch": 2.44, + "learning_rate": 1.6301325547223353e-05, + "loss": 0.2545, + "step": 5818 + }, + { + "epoch": 2.44, + "learning_rate": 1.630000899272702e-05, + "loss": 0.2427, + "step": 5819 + }, + { + "epoch": 2.44, + "learning_rate": 1.6298692257142525e-05, + "loss": 0.2687, + "step": 5820 + }, + { + "epoch": 2.44, + "learning_rate": 1.6297375340507705e-05, + "loss": 0.2529, + "step": 5821 + }, + { + "epoch": 2.44, + "learning_rate": 1.629605824286042e-05, + "loss": 0.2595, + "step": 5822 + }, + { + "epoch": 2.44, + "learning_rate": 1.629474096423853e-05, + "loss": 0.2597, + "step": 5823 + }, + { + "epoch": 2.44, + "learning_rate": 1.6293423504679898e-05, + "loss": 0.2469, + "step": 5824 + }, + { + "epoch": 2.44, + "learning_rate": 1.6292105864222386e-05, + "loss": 0.2507, + "step": 5825 + }, + { + "epoch": 2.44, + "learning_rate": 1.629078804290388e-05, + "loss": 0.2638, + "step": 5826 + }, + { + "epoch": 2.44, + "learning_rate": 1.628947004076225e-05, + "loss": 0.2452, + "step": 5827 + }, + { + "epoch": 2.44, + "learning_rate": 1.628815185783539e-05, + "loss": 0.2736, + "step": 5828 + }, + { + "epoch": 2.44, + "learning_rate": 1.628683349416118e-05, + "loss": 0.2595, + "step": 5829 + }, + { + "epoch": 2.44, + "learning_rate": 1.6285514949777522e-05, + "loss": 0.2602, + "step": 5830 + }, + { + "epoch": 2.44, + "learning_rate": 1.628419622472231e-05, + "loss": 0.2559, + "step": 5831 + }, + { + "epoch": 2.44, + "learning_rate": 1.6282877319033462e-05, + "loss": 0.2772, + "step": 5832 + }, + { + "epoch": 2.44, + "learning_rate": 1.6281558232748874e-05, + "loss": 0.2583, + "step": 5833 + }, + { + "epoch": 2.44, + "learning_rate": 1.6280238965906474e-05, + "loss": 0.2928, + "step": 5834 + }, + { + "epoch": 2.44, + "learning_rate": 1.6278919518544175e-05, + "loss": 0.2403, + "step": 5835 + }, + { + "epoch": 2.44, + "learning_rate": 1.627759989069991e-05, + "loss": 0.2579, + "step": 5836 + }, + { + "epoch": 2.44, + "learning_rate": 1.6276280082411604e-05, + "loss": 0.2968, + "step": 5837 + }, + { + "epoch": 2.44, + "learning_rate": 1.6274960093717192e-05, + "loss": 0.2501, + "step": 5838 + }, + { + "epoch": 2.44, + "learning_rate": 1.6273639924654626e-05, + "loss": 0.2598, + "step": 5839 + }, + { + "epoch": 2.45, + "learning_rate": 1.627231957526185e-05, + "loss": 0.2334, + "step": 5840 + }, + { + "epoch": 2.45, + "learning_rate": 1.6270999045576808e-05, + "loss": 0.2504, + "step": 5841 + }, + { + "epoch": 2.45, + "learning_rate": 1.6269678335637464e-05, + "loss": 0.2664, + "step": 5842 + }, + { + "epoch": 2.45, + "learning_rate": 1.626835744548178e-05, + "loss": 0.2308, + "step": 5843 + }, + { + "epoch": 2.45, + "learning_rate": 1.6267036375147728e-05, + "loss": 0.2584, + "step": 5844 + }, + { + "epoch": 2.45, + "learning_rate": 1.6265715124673272e-05, + "loss": 0.2493, + "step": 5845 + }, + { + "epoch": 2.45, + "learning_rate": 1.6264393694096396e-05, + "loss": 0.2382, + "step": 5846 + }, + { + "epoch": 2.45, + "learning_rate": 1.6263072083455084e-05, + "loss": 0.2394, + "step": 5847 + }, + { + "epoch": 2.45, + "learning_rate": 1.626175029278732e-05, + "loss": 0.2658, + "step": 5848 + }, + { + "epoch": 2.45, + "learning_rate": 1.62604283221311e-05, + "loss": 0.2416, + "step": 5849 + }, + { + "epoch": 2.45, + "learning_rate": 1.6259106171524424e-05, + "loss": 0.2642, + "step": 5850 + }, + { + "epoch": 2.45, + "learning_rate": 1.6257783841005297e-05, + "loss": 0.2426, + "step": 5851 + }, + { + "epoch": 2.45, + "learning_rate": 1.6256461330611727e-05, + "loss": 0.2554, + "step": 5852 + }, + { + "epoch": 2.45, + "learning_rate": 1.6255138640381723e-05, + "loss": 0.2516, + "step": 5853 + }, + { + "epoch": 2.45, + "learning_rate": 1.6253815770353316e-05, + "loss": 0.2651, + "step": 5854 + }, + { + "epoch": 2.45, + "learning_rate": 1.625249272056452e-05, + "loss": 0.2797, + "step": 5855 + }, + { + "epoch": 2.45, + "learning_rate": 1.625116949105337e-05, + "loss": 0.2711, + "step": 5856 + }, + { + "epoch": 2.45, + "learning_rate": 1.62498460818579e-05, + "loss": 0.2555, + "step": 5857 + }, + { + "epoch": 2.45, + "learning_rate": 1.6248522493016153e-05, + "loss": 0.2479, + "step": 5858 + }, + { + "epoch": 2.45, + "learning_rate": 1.624719872456617e-05, + "loss": 0.2662, + "step": 5859 + }, + { + "epoch": 2.45, + "learning_rate": 1.6245874776546005e-05, + "loss": 0.2627, + "step": 5860 + }, + { + "epoch": 2.45, + "learning_rate": 1.624455064899371e-05, + "loss": 0.247, + "step": 5861 + }, + { + "epoch": 2.45, + "learning_rate": 1.6243226341947348e-05, + "loss": 0.266, + "step": 5862 + }, + { + "epoch": 2.45, + "learning_rate": 1.624190185544498e-05, + "loss": 0.25, + "step": 5863 + }, + { + "epoch": 2.46, + "learning_rate": 1.6240577189524688e-05, + "loss": 0.2389, + "step": 5864 + }, + { + "epoch": 2.46, + "learning_rate": 1.6239252344224545e-05, + "loss": 0.2448, + "step": 5865 + }, + { + "epoch": 2.46, + "learning_rate": 1.6237927319582624e-05, + "loss": 0.2697, + "step": 5866 + }, + { + "epoch": 2.46, + "learning_rate": 1.623660211563702e-05, + "loss": 0.2607, + "step": 5867 + }, + { + "epoch": 2.46, + "learning_rate": 1.6235276732425823e-05, + "loss": 0.2711, + "step": 5868 + }, + { + "epoch": 2.46, + "learning_rate": 1.623395116998713e-05, + "loss": 0.2854, + "step": 5869 + }, + { + "epoch": 2.46, + "learning_rate": 1.6232625428359042e-05, + "loss": 0.2689, + "step": 5870 + }, + { + "epoch": 2.46, + "learning_rate": 1.6231299507579668e-05, + "loss": 0.2591, + "step": 5871 + }, + { + "epoch": 2.46, + "learning_rate": 1.6229973407687117e-05, + "loss": 0.2603, + "step": 5872 + }, + { + "epoch": 2.46, + "learning_rate": 1.622864712871951e-05, + "loss": 0.2572, + "step": 5873 + }, + { + "epoch": 2.46, + "learning_rate": 1.6227320670714966e-05, + "loss": 0.2879, + "step": 5874 + }, + { + "epoch": 2.46, + "learning_rate": 1.6225994033711617e-05, + "loss": 0.2058, + "step": 5875 + }, + { + "epoch": 2.46, + "learning_rate": 1.6224667217747594e-05, + "loss": 0.26, + "step": 5876 + }, + { + "epoch": 2.46, + "learning_rate": 1.6223340222861037e-05, + "loss": 0.2241, + "step": 5877 + }, + { + "epoch": 2.46, + "learning_rate": 1.6222013049090086e-05, + "loss": 0.2499, + "step": 5878 + }, + { + "epoch": 2.46, + "learning_rate": 1.622068569647289e-05, + "loss": 0.2711, + "step": 5879 + }, + { + "epoch": 2.46, + "learning_rate": 1.6219358165047606e-05, + "loss": 0.2468, + "step": 5880 + }, + { + "epoch": 2.46, + "learning_rate": 1.6218030454852395e-05, + "loss": 0.2711, + "step": 5881 + }, + { + "epoch": 2.46, + "learning_rate": 1.621670256592541e-05, + "loss": 0.2464, + "step": 5882 + }, + { + "epoch": 2.46, + "learning_rate": 1.6215374498304827e-05, + "loss": 0.2739, + "step": 5883 + }, + { + "epoch": 2.46, + "learning_rate": 1.6214046252028823e-05, + "loss": 0.2529, + "step": 5884 + }, + { + "epoch": 2.46, + "learning_rate": 1.621271782713557e-05, + "loss": 0.2508, + "step": 5885 + }, + { + "epoch": 2.46, + "learning_rate": 1.621138922366326e-05, + "loss": 0.245, + "step": 5886 + }, + { + "epoch": 2.46, + "learning_rate": 1.6210060441650073e-05, + "loss": 0.2693, + "step": 5887 + }, + { + "epoch": 2.47, + "learning_rate": 1.6208731481134214e-05, + "loss": 0.2377, + "step": 5888 + }, + { + "epoch": 2.47, + "learning_rate": 1.6207402342153876e-05, + "loss": 0.2442, + "step": 5889 + }, + { + "epoch": 2.47, + "learning_rate": 1.6206073024747264e-05, + "loss": 0.2673, + "step": 5890 + }, + { + "epoch": 2.47, + "learning_rate": 1.6204743528952592e-05, + "loss": 0.2758, + "step": 5891 + }, + { + "epoch": 2.47, + "learning_rate": 1.6203413854808076e-05, + "loss": 0.2612, + "step": 5892 + }, + { + "epoch": 2.47, + "learning_rate": 1.620208400235193e-05, + "loss": 0.2591, + "step": 5893 + }, + { + "epoch": 2.47, + "learning_rate": 1.6200753971622387e-05, + "loss": 0.2512, + "step": 5894 + }, + { + "epoch": 2.47, + "learning_rate": 1.619942376265767e-05, + "loss": 0.2587, + "step": 5895 + }, + { + "epoch": 2.47, + "learning_rate": 1.619809337549602e-05, + "loss": 0.2153, + "step": 5896 + }, + { + "epoch": 2.47, + "learning_rate": 1.6196762810175674e-05, + "loss": 0.2638, + "step": 5897 + }, + { + "epoch": 2.47, + "learning_rate": 1.6195432066734878e-05, + "loss": 0.2553, + "step": 5898 + }, + { + "epoch": 2.47, + "learning_rate": 1.619410114521189e-05, + "loss": 0.2529, + "step": 5899 + }, + { + "epoch": 2.47, + "learning_rate": 1.619277004564496e-05, + "loss": 0.2724, + "step": 5900 + }, + { + "epoch": 2.47, + "learning_rate": 1.619143876807235e-05, + "loss": 0.2458, + "step": 5901 + }, + { + "epoch": 2.47, + "learning_rate": 1.6190107312532327e-05, + "loss": 0.2857, + "step": 5902 + }, + { + "epoch": 2.47, + "learning_rate": 1.618877567906316e-05, + "loss": 0.2431, + "step": 5903 + }, + { + "epoch": 2.47, + "learning_rate": 1.618744386770313e-05, + "loss": 0.2515, + "step": 5904 + }, + { + "epoch": 2.47, + "learning_rate": 1.6186111878490515e-05, + "loss": 0.2522, + "step": 5905 + }, + { + "epoch": 2.47, + "learning_rate": 1.6184779711463607e-05, + "loss": 0.2592, + "step": 5906 + }, + { + "epoch": 2.47, + "learning_rate": 1.618344736666069e-05, + "loss": 0.2663, + "step": 5907 + }, + { + "epoch": 2.47, + "learning_rate": 1.618211484412007e-05, + "loss": 0.2633, + "step": 5908 + }, + { + "epoch": 2.47, + "learning_rate": 1.6180782143880042e-05, + "loss": 0.2422, + "step": 5909 + }, + { + "epoch": 2.47, + "learning_rate": 1.6179449265978913e-05, + "loss": 0.2889, + "step": 5910 + }, + { + "epoch": 2.47, + "learning_rate": 1.6178116210455006e-05, + "loss": 0.3035, + "step": 5911 + }, + { + "epoch": 2.48, + "learning_rate": 1.6176782977346626e-05, + "loss": 0.2714, + "step": 5912 + }, + { + "epoch": 2.48, + "learning_rate": 1.61754495666921e-05, + "loss": 0.254, + "step": 5913 + }, + { + "epoch": 2.48, + "learning_rate": 1.617411597852976e-05, + "loss": 0.2454, + "step": 5914 + }, + { + "epoch": 2.48, + "learning_rate": 1.617278221289793e-05, + "loss": 0.2544, + "step": 5915 + }, + { + "epoch": 2.48, + "learning_rate": 1.6171448269834956e-05, + "loss": 0.2736, + "step": 5916 + }, + { + "epoch": 2.48, + "learning_rate": 1.6170114149379178e-05, + "loss": 0.2428, + "step": 5917 + }, + { + "epoch": 2.48, + "learning_rate": 1.6168779851568947e-05, + "loss": 0.2423, + "step": 5918 + }, + { + "epoch": 2.48, + "learning_rate": 1.6167445376442604e-05, + "loss": 0.2222, + "step": 5919 + }, + { + "epoch": 2.48, + "learning_rate": 1.6166110724038524e-05, + "loss": 0.2606, + "step": 5920 + }, + { + "epoch": 2.48, + "learning_rate": 1.616477589439506e-05, + "loss": 0.2681, + "step": 5921 + }, + { + "epoch": 2.48, + "learning_rate": 1.6163440887550587e-05, + "loss": 0.2595, + "step": 5922 + }, + { + "epoch": 2.48, + "learning_rate": 1.616210570354347e-05, + "loss": 0.2684, + "step": 5923 + }, + { + "epoch": 2.48, + "learning_rate": 1.6160770342412097e-05, + "loss": 0.2561, + "step": 5924 + }, + { + "epoch": 2.48, + "learning_rate": 1.6159434804194845e-05, + "loss": 0.2489, + "step": 5925 + }, + { + "epoch": 2.48, + "learning_rate": 1.615809908893011e-05, + "loss": 0.2594, + "step": 5926 + }, + { + "epoch": 2.48, + "learning_rate": 1.6156763196656276e-05, + "loss": 0.2583, + "step": 5927 + }, + { + "epoch": 2.48, + "learning_rate": 1.6155427127411748e-05, + "loss": 0.2552, + "step": 5928 + }, + { + "epoch": 2.48, + "learning_rate": 1.615409088123493e-05, + "loss": 0.2666, + "step": 5929 + }, + { + "epoch": 2.48, + "learning_rate": 1.615275445816423e-05, + "loss": 0.2609, + "step": 5930 + }, + { + "epoch": 2.48, + "learning_rate": 1.6151417858238064e-05, + "loss": 0.2475, + "step": 5931 + }, + { + "epoch": 2.48, + "learning_rate": 1.615008108149485e-05, + "loss": 0.2644, + "step": 5932 + }, + { + "epoch": 2.48, + "learning_rate": 1.6148744127973014e-05, + "loss": 0.2486, + "step": 5933 + }, + { + "epoch": 2.48, + "learning_rate": 1.614740699771098e-05, + "loss": 0.2463, + "step": 5934 + }, + { + "epoch": 2.48, + "learning_rate": 1.614606969074719e-05, + "loss": 0.2454, + "step": 5935 + }, + { + "epoch": 2.49, + "learning_rate": 1.6144732207120083e-05, + "loss": 0.2456, + "step": 5936 + }, + { + "epoch": 2.49, + "learning_rate": 1.61433945468681e-05, + "loss": 0.2615, + "step": 5937 + }, + { + "epoch": 2.49, + "learning_rate": 1.614205671002969e-05, + "loss": 0.2378, + "step": 5938 + }, + { + "epoch": 2.49, + "learning_rate": 1.6140718696643313e-05, + "loss": 0.2868, + "step": 5939 + }, + { + "epoch": 2.49, + "learning_rate": 1.6139380506747424e-05, + "loss": 0.2496, + "step": 5940 + }, + { + "epoch": 2.49, + "learning_rate": 1.6138042140380493e-05, + "loss": 0.2343, + "step": 5941 + }, + { + "epoch": 2.49, + "learning_rate": 1.6136703597580984e-05, + "loss": 0.2396, + "step": 5942 + }, + { + "epoch": 2.49, + "learning_rate": 1.613536487838738e-05, + "loss": 0.2665, + "step": 5943 + }, + { + "epoch": 2.49, + "learning_rate": 1.6134025982838153e-05, + "loss": 0.2772, + "step": 5944 + }, + { + "epoch": 2.49, + "learning_rate": 1.6132686910971792e-05, + "loss": 0.2544, + "step": 5945 + }, + { + "epoch": 2.49, + "learning_rate": 1.6131347662826793e-05, + "loss": 0.2489, + "step": 5946 + }, + { + "epoch": 2.49, + "learning_rate": 1.613000823844164e-05, + "loss": 0.2798, + "step": 5947 + }, + { + "epoch": 2.49, + "learning_rate": 1.6128668637854847e-05, + "loss": 0.2367, + "step": 5948 + }, + { + "epoch": 2.49, + "learning_rate": 1.6127328861104906e-05, + "loss": 0.2688, + "step": 5949 + }, + { + "epoch": 2.49, + "learning_rate": 1.6125988908230337e-05, + "loss": 0.2514, + "step": 5950 + }, + { + "epoch": 2.49, + "learning_rate": 1.6124648779269654e-05, + "loss": 0.2594, + "step": 5951 + }, + { + "epoch": 2.49, + "learning_rate": 1.6123308474261374e-05, + "loss": 0.2542, + "step": 5952 + }, + { + "epoch": 2.49, + "learning_rate": 1.6121967993244026e-05, + "loss": 0.2548, + "step": 5953 + }, + { + "epoch": 2.49, + "learning_rate": 1.6120627336256142e-05, + "loss": 0.2928, + "step": 5954 + }, + { + "epoch": 2.49, + "learning_rate": 1.6119286503336256e-05, + "loss": 0.2584, + "step": 5955 + }, + { + "epoch": 2.49, + "learning_rate": 1.611794549452291e-05, + "loss": 0.2732, + "step": 5956 + }, + { + "epoch": 2.49, + "learning_rate": 1.611660430985465e-05, + "loss": 0.2218, + "step": 5957 + }, + { + "epoch": 2.49, + "learning_rate": 1.6115262949370025e-05, + "loss": 0.2893, + "step": 5958 + }, + { + "epoch": 2.49, + "learning_rate": 1.6113921413107597e-05, + "loss": 0.2684, + "step": 5959 + }, + { + "epoch": 2.5, + "learning_rate": 1.6112579701105923e-05, + "loss": 0.2812, + "step": 5960 + }, + { + "epoch": 2.5, + "learning_rate": 1.6111237813403568e-05, + "loss": 0.2759, + "step": 5961 + }, + { + "epoch": 2.5, + "learning_rate": 1.61098957500391e-05, + "loss": 0.285, + "step": 5962 + }, + { + "epoch": 2.5, + "learning_rate": 1.6108553511051106e-05, + "loss": 0.2418, + "step": 5963 + }, + { + "epoch": 2.5, + "learning_rate": 1.610721109647816e-05, + "loss": 0.2605, + "step": 5964 + }, + { + "epoch": 2.5, + "learning_rate": 1.610586850635885e-05, + "loss": 0.263, + "step": 5965 + }, + { + "epoch": 2.5, + "learning_rate": 1.6104525740731772e-05, + "loss": 0.28, + "step": 5966 + }, + { + "epoch": 2.5, + "learning_rate": 1.6103182799635514e-05, + "loss": 0.246, + "step": 5967 + }, + { + "epoch": 2.5, + "learning_rate": 1.6101839683108686e-05, + "loss": 0.2725, + "step": 5968 + }, + { + "epoch": 2.5, + "learning_rate": 1.610049639118989e-05, + "loss": 0.2355, + "step": 5969 + }, + { + "epoch": 2.5, + "learning_rate": 1.6099152923917737e-05, + "loss": 0.2961, + "step": 5970 + }, + { + "epoch": 2.5, + "learning_rate": 1.6097809281330848e-05, + "loss": 0.2723, + "step": 5971 + }, + { + "epoch": 2.5, + "learning_rate": 1.609646546346784e-05, + "loss": 0.2695, + "step": 5972 + }, + { + "epoch": 2.5, + "learning_rate": 1.6095121470367345e-05, + "loss": 0.287, + "step": 5973 + }, + { + "epoch": 2.5, + "learning_rate": 1.609377730206799e-05, + "loss": 0.2501, + "step": 5974 + }, + { + "epoch": 2.5, + "learning_rate": 1.6092432958608413e-05, + "loss": 0.2542, + "step": 5975 + }, + { + "epoch": 2.5, + "learning_rate": 1.609108844002726e-05, + "loss": 0.259, + "step": 5976 + }, + { + "epoch": 2.5, + "learning_rate": 1.608974374636317e-05, + "loss": 0.2713, + "step": 5977 + }, + { + "epoch": 2.5, + "learning_rate": 1.6088398877654806e-05, + "loss": 0.2765, + "step": 5978 + }, + { + "epoch": 2.5, + "learning_rate": 1.6087053833940815e-05, + "loss": 0.2254, + "step": 5979 + }, + { + "epoch": 2.5, + "learning_rate": 1.6085708615259863e-05, + "loss": 0.2434, + "step": 5980 + }, + { + "epoch": 2.5, + "learning_rate": 1.608436322165062e-05, + "loss": 0.2638, + "step": 5981 + }, + { + "epoch": 2.5, + "learning_rate": 1.6083017653151752e-05, + "loss": 0.2749, + "step": 5982 + }, + { + "epoch": 2.5, + "learning_rate": 1.6081671909801944e-05, + "loss": 0.2938, + "step": 5983 + }, + { + "epoch": 2.51, + "learning_rate": 1.6080325991639866e-05, + "loss": 0.2419, + "step": 5984 + }, + { + "epoch": 2.51, + "learning_rate": 1.6078979898704223e-05, + "loss": 0.2736, + "step": 5985 + }, + { + "epoch": 2.51, + "learning_rate": 1.607763363103369e-05, + "loss": 0.2498, + "step": 5986 + }, + { + "epoch": 2.51, + "learning_rate": 1.6076287188666972e-05, + "loss": 0.2468, + "step": 5987 + }, + { + "epoch": 2.51, + "learning_rate": 1.6074940571642773e-05, + "loss": 0.2543, + "step": 5988 + }, + { + "epoch": 2.51, + "learning_rate": 1.6073593779999795e-05, + "loss": 0.2549, + "step": 5989 + }, + { + "epoch": 2.51, + "learning_rate": 1.6072246813776758e-05, + "loss": 0.2918, + "step": 5990 + }, + { + "epoch": 2.51, + "learning_rate": 1.6070899673012372e-05, + "loss": 0.2755, + "step": 5991 + }, + { + "epoch": 2.51, + "learning_rate": 1.6069552357745363e-05, + "loss": 0.2557, + "step": 5992 + }, + { + "epoch": 2.51, + "learning_rate": 1.6068204868014455e-05, + "loss": 0.2732, + "step": 5993 + }, + { + "epoch": 2.51, + "learning_rate": 1.6066857203858387e-05, + "loss": 0.2549, + "step": 5994 + }, + { + "epoch": 2.51, + "learning_rate": 1.6065509365315893e-05, + "loss": 0.2497, + "step": 5995 + }, + { + "epoch": 2.51, + "learning_rate": 1.6064161352425708e-05, + "loss": 0.2364, + "step": 5996 + }, + { + "epoch": 2.51, + "learning_rate": 1.6062813165226594e-05, + "loss": 0.2453, + "step": 5997 + }, + { + "epoch": 2.51, + "learning_rate": 1.606146480375729e-05, + "loss": 0.2279, + "step": 5998 + }, + { + "epoch": 2.51, + "learning_rate": 1.606011626805656e-05, + "loss": 0.2577, + "step": 5999 + }, + { + "epoch": 2.51, + "learning_rate": 1.6058767558163168e-05, + "loss": 0.2731, + "step": 6000 + }, + { + "epoch": 2.51, + "learning_rate": 1.6057418674115876e-05, + "loss": 0.2782, + "step": 6001 + }, + { + "epoch": 2.51, + "learning_rate": 1.605606961595346e-05, + "loss": 0.2451, + "step": 6002 + }, + { + "epoch": 2.51, + "learning_rate": 1.60547203837147e-05, + "loss": 0.2513, + "step": 6003 + }, + { + "epoch": 2.51, + "learning_rate": 1.605337097743837e-05, + "loss": 0.2531, + "step": 6004 + }, + { + "epoch": 2.51, + "learning_rate": 1.605202139716327e-05, + "loss": 0.2754, + "step": 6005 + }, + { + "epoch": 2.51, + "learning_rate": 1.605067164292818e-05, + "loss": 0.2522, + "step": 6006 + }, + { + "epoch": 2.51, + "learning_rate": 1.6049321714771907e-05, + "loss": 0.2427, + "step": 6007 + }, + { + "epoch": 2.52, + "learning_rate": 1.6047971612733246e-05, + "loss": 0.2295, + "step": 6008 + }, + { + "epoch": 2.52, + "learning_rate": 1.6046621336851007e-05, + "loss": 0.272, + "step": 6009 + }, + { + "epoch": 2.52, + "learning_rate": 1.6045270887164005e-05, + "loss": 0.2655, + "step": 6010 + }, + { + "epoch": 2.52, + "learning_rate": 1.6043920263711056e-05, + "loss": 0.2788, + "step": 6011 + }, + { + "epoch": 2.52, + "learning_rate": 1.6042569466530985e-05, + "loss": 0.2805, + "step": 6012 + }, + { + "epoch": 2.52, + "learning_rate": 1.604121849566261e-05, + "loss": 0.2553, + "step": 6013 + }, + { + "epoch": 2.52, + "learning_rate": 1.6039867351144778e-05, + "loss": 0.2407, + "step": 6014 + }, + { + "epoch": 2.52, + "learning_rate": 1.6038516033016314e-05, + "loss": 0.2728, + "step": 6015 + }, + { + "epoch": 2.52, + "learning_rate": 1.6037164541316067e-05, + "loss": 0.2442, + "step": 6016 + }, + { + "epoch": 2.52, + "learning_rate": 1.603581287608288e-05, + "loss": 0.2559, + "step": 6017 + }, + { + "epoch": 2.52, + "learning_rate": 1.603446103735561e-05, + "loss": 0.2662, + "step": 6018 + }, + { + "epoch": 2.52, + "learning_rate": 1.6033109025173113e-05, + "loss": 0.2541, + "step": 6019 + }, + { + "epoch": 2.52, + "learning_rate": 1.6031756839574247e-05, + "loss": 0.2909, + "step": 6020 + }, + { + "epoch": 2.52, + "learning_rate": 1.6030404480597888e-05, + "loss": 0.2366, + "step": 6021 + }, + { + "epoch": 2.52, + "learning_rate": 1.60290519482829e-05, + "loss": 0.2578, + "step": 6022 + }, + { + "epoch": 2.52, + "learning_rate": 1.6027699242668166e-05, + "loss": 0.2686, + "step": 6023 + }, + { + "epoch": 2.52, + "learning_rate": 1.6026346363792565e-05, + "loss": 0.2634, + "step": 6024 + }, + { + "epoch": 2.52, + "learning_rate": 1.6024993311694986e-05, + "loss": 0.267, + "step": 6025 + }, + { + "epoch": 2.52, + "learning_rate": 1.6023640086414318e-05, + "loss": 0.2677, + "step": 6026 + }, + { + "epoch": 2.52, + "learning_rate": 1.6022286687989466e-05, + "loss": 0.2588, + "step": 6027 + }, + { + "epoch": 2.52, + "learning_rate": 1.6020933116459324e-05, + "loss": 0.254, + "step": 6028 + }, + { + "epoch": 2.52, + "learning_rate": 1.6019579371862803e-05, + "loss": 0.2767, + "step": 6029 + }, + { + "epoch": 2.52, + "learning_rate": 1.6018225454238816e-05, + "loss": 0.248, + "step": 6030 + }, + { + "epoch": 2.53, + "learning_rate": 1.6016871363626277e-05, + "loss": 0.2448, + "step": 6031 + }, + { + "epoch": 2.53, + "learning_rate": 1.6015517100064114e-05, + "loss": 0.2472, + "step": 6032 + }, + { + "epoch": 2.53, + "learning_rate": 1.601416266359124e-05, + "loss": 0.2665, + "step": 6033 + }, + { + "epoch": 2.53, + "learning_rate": 1.6012808054246608e-05, + "loss": 0.2622, + "step": 6034 + }, + { + "epoch": 2.53, + "learning_rate": 1.601145327206914e-05, + "loss": 0.2483, + "step": 6035 + }, + { + "epoch": 2.53, + "learning_rate": 1.601009831709778e-05, + "loss": 0.2636, + "step": 6036 + }, + { + "epoch": 2.53, + "learning_rate": 1.600874318937148e-05, + "loss": 0.2706, + "step": 6037 + }, + { + "epoch": 2.53, + "learning_rate": 1.600738788892919e-05, + "loss": 0.2803, + "step": 6038 + }, + { + "epoch": 2.53, + "learning_rate": 1.6006032415809866e-05, + "loss": 0.2737, + "step": 6039 + }, + { + "epoch": 2.53, + "learning_rate": 1.6004676770052466e-05, + "loss": 0.2622, + "step": 6040 + }, + { + "epoch": 2.53, + "learning_rate": 1.600332095169596e-05, + "loss": 0.2615, + "step": 6041 + }, + { + "epoch": 2.53, + "learning_rate": 1.6001964960779326e-05, + "loss": 0.2653, + "step": 6042 + }, + { + "epoch": 2.53, + "learning_rate": 1.600060879734153e-05, + "loss": 0.2388, + "step": 6043 + }, + { + "epoch": 2.53, + "learning_rate": 1.5999252461421564e-05, + "loss": 0.2292, + "step": 6044 + }, + { + "epoch": 2.53, + "learning_rate": 1.5997895953058407e-05, + "loss": 0.2427, + "step": 6045 + }, + { + "epoch": 2.53, + "learning_rate": 1.5996539272291052e-05, + "loss": 0.2385, + "step": 6046 + }, + { + "epoch": 2.53, + "learning_rate": 1.59951824191585e-05, + "loss": 0.2729, + "step": 6047 + }, + { + "epoch": 2.53, + "learning_rate": 1.5993825393699746e-05, + "loss": 0.2514, + "step": 6048 + }, + { + "epoch": 2.53, + "learning_rate": 1.59924681959538e-05, + "loss": 0.2577, + "step": 6049 + }, + { + "epoch": 2.53, + "learning_rate": 1.5991110825959676e-05, + "loss": 0.245, + "step": 6050 + }, + { + "epoch": 2.53, + "learning_rate": 1.598975328375639e-05, + "loss": 0.2652, + "step": 6051 + }, + { + "epoch": 2.53, + "learning_rate": 1.5988395569382954e-05, + "loss": 0.2477, + "step": 6052 + }, + { + "epoch": 2.53, + "learning_rate": 1.5987037682878405e-05, + "loss": 0.2296, + "step": 6053 + }, + { + "epoch": 2.53, + "learning_rate": 1.5985679624281772e-05, + "loss": 0.244, + "step": 6054 + }, + { + "epoch": 2.54, + "learning_rate": 1.5984321393632087e-05, + "loss": 0.2329, + "step": 6055 + }, + { + "epoch": 2.54, + "learning_rate": 1.5982962990968397e-05, + "loss": 0.2332, + "step": 6056 + }, + { + "epoch": 2.54, + "learning_rate": 1.5981604416329742e-05, + "loss": 0.243, + "step": 6057 + }, + { + "epoch": 2.54, + "learning_rate": 1.598024566975518e-05, + "loss": 0.2464, + "step": 6058 + }, + { + "epoch": 2.54, + "learning_rate": 1.597888675128376e-05, + "loss": 0.2263, + "step": 6059 + }, + { + "epoch": 2.54, + "learning_rate": 1.5977527660954545e-05, + "loss": 0.2715, + "step": 6060 + }, + { + "epoch": 2.54, + "learning_rate": 1.5976168398806602e-05, + "loss": 0.2625, + "step": 6061 + }, + { + "epoch": 2.54, + "learning_rate": 1.5974808964879005e-05, + "loss": 0.2729, + "step": 6062 + }, + { + "epoch": 2.54, + "learning_rate": 1.5973449359210823e-05, + "loss": 0.2417, + "step": 6063 + }, + { + "epoch": 2.54, + "learning_rate": 1.597208958184114e-05, + "loss": 0.2355, + "step": 6064 + }, + { + "epoch": 2.54, + "learning_rate": 1.5970729632809044e-05, + "loss": 0.2282, + "step": 6065 + }, + { + "epoch": 2.54, + "learning_rate": 1.596936951215362e-05, + "loss": 0.2426, + "step": 6066 + }, + { + "epoch": 2.54, + "learning_rate": 1.5968009219913966e-05, + "loss": 0.2719, + "step": 6067 + }, + { + "epoch": 2.54, + "learning_rate": 1.5966648756129187e-05, + "loss": 0.2581, + "step": 6068 + }, + { + "epoch": 2.54, + "learning_rate": 1.5965288120838383e-05, + "loss": 0.2915, + "step": 6069 + }, + { + "epoch": 2.54, + "learning_rate": 1.596392731408066e-05, + "loss": 0.2598, + "step": 6070 + }, + { + "epoch": 2.54, + "learning_rate": 1.5962566335895146e-05, + "loss": 0.2513, + "step": 6071 + }, + { + "epoch": 2.54, + "learning_rate": 1.5961205186320952e-05, + "loss": 0.2531, + "step": 6072 + }, + { + "epoch": 2.54, + "learning_rate": 1.59598438653972e-05, + "loss": 0.2473, + "step": 6073 + }, + { + "epoch": 2.54, + "learning_rate": 1.595848237316303e-05, + "loss": 0.2286, + "step": 6074 + }, + { + "epoch": 2.54, + "learning_rate": 1.595712070965757e-05, + "loss": 0.2766, + "step": 6075 + }, + { + "epoch": 2.54, + "learning_rate": 1.595575887491996e-05, + "loss": 0.2845, + "step": 6076 + }, + { + "epoch": 2.54, + "learning_rate": 1.595439686898935e-05, + "loss": 0.2496, + "step": 6077 + }, + { + "epoch": 2.54, + "learning_rate": 1.595303469190488e-05, + "loss": 0.2742, + "step": 6078 + }, + { + "epoch": 2.55, + "learning_rate": 1.5951672343705714e-05, + "loss": 0.2465, + "step": 6079 + }, + { + "epoch": 2.55, + "learning_rate": 1.5950309824431008e-05, + "loss": 0.2647, + "step": 6080 + }, + { + "epoch": 2.55, + "learning_rate": 1.5948947134119926e-05, + "loss": 0.2563, + "step": 6081 + }, + { + "epoch": 2.55, + "learning_rate": 1.594758427281164e-05, + "loss": 0.2532, + "step": 6082 + }, + { + "epoch": 2.55, + "learning_rate": 1.594622124054532e-05, + "loss": 0.2513, + "step": 6083 + }, + { + "epoch": 2.55, + "learning_rate": 1.5944858037360145e-05, + "loss": 0.2514, + "step": 6084 + }, + { + "epoch": 2.55, + "learning_rate": 1.5943494663295303e-05, + "loss": 0.2461, + "step": 6085 + }, + { + "epoch": 2.55, + "learning_rate": 1.5942131118389984e-05, + "loss": 0.2538, + "step": 6086 + }, + { + "epoch": 2.55, + "learning_rate": 1.594076740268337e-05, + "loss": 0.2439, + "step": 6087 + }, + { + "epoch": 2.55, + "learning_rate": 1.593940351621468e-05, + "loss": 0.2267, + "step": 6088 + }, + { + "epoch": 2.55, + "learning_rate": 1.5938039459023098e-05, + "loss": 0.2541, + "step": 6089 + }, + { + "epoch": 2.55, + "learning_rate": 1.5936675231147845e-05, + "loss": 0.2191, + "step": 6090 + }, + { + "epoch": 2.55, + "learning_rate": 1.593531083262813e-05, + "loss": 0.268, + "step": 6091 + }, + { + "epoch": 2.55, + "learning_rate": 1.593394626350317e-05, + "loss": 0.2505, + "step": 6092 + }, + { + "epoch": 2.55, + "learning_rate": 1.5932581523812193e-05, + "loss": 0.2636, + "step": 6093 + }, + { + "epoch": 2.55, + "learning_rate": 1.5931216613594426e-05, + "loss": 0.2326, + "step": 6094 + }, + { + "epoch": 2.55, + "learning_rate": 1.5929851532889095e-05, + "loss": 0.2709, + "step": 6095 + }, + { + "epoch": 2.55, + "learning_rate": 1.592848628173545e-05, + "loss": 0.2567, + "step": 6096 + }, + { + "epoch": 2.55, + "learning_rate": 1.5927120860172722e-05, + "loss": 0.2475, + "step": 6097 + }, + { + "epoch": 2.55, + "learning_rate": 1.592575526824017e-05, + "loss": 0.2789, + "step": 6098 + }, + { + "epoch": 2.55, + "learning_rate": 1.5924389505977038e-05, + "loss": 0.2341, + "step": 6099 + }, + { + "epoch": 2.55, + "learning_rate": 1.5923023573422585e-05, + "loss": 0.2605, + "step": 6100 + }, + { + "epoch": 2.55, + "learning_rate": 1.592165747061608e-05, + "loss": 0.2788, + "step": 6101 + }, + { + "epoch": 2.55, + "learning_rate": 1.5920291197596784e-05, + "loss": 0.2772, + "step": 6102 + }, + { + "epoch": 2.56, + "learning_rate": 1.591892475440397e-05, + "loss": 0.232, + "step": 6103 + }, + { + "epoch": 2.56, + "learning_rate": 1.5917558141076915e-05, + "loss": 0.2478, + "step": 6104 + }, + { + "epoch": 2.56, + "learning_rate": 1.5916191357654906e-05, + "loss": 0.2352, + "step": 6105 + }, + { + "epoch": 2.56, + "learning_rate": 1.5914824404177222e-05, + "loss": 0.2803, + "step": 6106 + }, + { + "epoch": 2.56, + "learning_rate": 1.5913457280683163e-05, + "loss": 0.2376, + "step": 6107 + }, + { + "epoch": 2.56, + "learning_rate": 1.591208998721202e-05, + "loss": 0.264, + "step": 6108 + }, + { + "epoch": 2.56, + "learning_rate": 1.5910722523803098e-05, + "loss": 0.2767, + "step": 6109 + }, + { + "epoch": 2.56, + "learning_rate": 1.5909354890495703e-05, + "loss": 0.2921, + "step": 6110 + }, + { + "epoch": 2.56, + "learning_rate": 1.5907987087329143e-05, + "loss": 0.2388, + "step": 6111 + }, + { + "epoch": 2.56, + "learning_rate": 1.590661911434274e-05, + "loss": 0.2408, + "step": 6112 + }, + { + "epoch": 2.56, + "learning_rate": 1.5905250971575808e-05, + "loss": 0.2688, + "step": 6113 + }, + { + "epoch": 2.56, + "learning_rate": 1.590388265906768e-05, + "loss": 0.2435, + "step": 6114 + }, + { + "epoch": 2.56, + "learning_rate": 1.590251417685768e-05, + "loss": 0.2595, + "step": 6115 + }, + { + "epoch": 2.56, + "learning_rate": 1.5901145524985155e-05, + "loss": 0.26, + "step": 6116 + }, + { + "epoch": 2.56, + "learning_rate": 1.5899776703489432e-05, + "loss": 0.304, + "step": 6117 + }, + { + "epoch": 2.56, + "learning_rate": 1.5898407712409863e-05, + "loss": 0.2597, + "step": 6118 + }, + { + "epoch": 2.56, + "learning_rate": 1.58970385517858e-05, + "loss": 0.2655, + "step": 6119 + }, + { + "epoch": 2.56, + "learning_rate": 1.58956692216566e-05, + "loss": 0.2699, + "step": 6120 + }, + { + "epoch": 2.56, + "learning_rate": 1.5894299722061617e-05, + "loss": 0.273, + "step": 6121 + }, + { + "epoch": 2.56, + "learning_rate": 1.5892930053040218e-05, + "loss": 0.2488, + "step": 6122 + }, + { + "epoch": 2.56, + "learning_rate": 1.5891560214631775e-05, + "loss": 0.2928, + "step": 6123 + }, + { + "epoch": 2.56, + "learning_rate": 1.5890190206875663e-05, + "loss": 0.2671, + "step": 6124 + }, + { + "epoch": 2.56, + "learning_rate": 1.588882002981126e-05, + "loss": 0.2488, + "step": 6125 + }, + { + "epoch": 2.56, + "learning_rate": 1.588744968347795e-05, + "loss": 0.2903, + "step": 6126 + }, + { + "epoch": 2.57, + "learning_rate": 1.5886079167915123e-05, + "loss": 0.2526, + "step": 6127 + }, + { + "epoch": 2.57, + "learning_rate": 1.5884708483162175e-05, + "loss": 0.2429, + "step": 6128 + }, + { + "epoch": 2.57, + "learning_rate": 1.5883337629258502e-05, + "loss": 0.2909, + "step": 6129 + }, + { + "epoch": 2.57, + "learning_rate": 1.5881966606243512e-05, + "loss": 0.2554, + "step": 6130 + }, + { + "epoch": 2.57, + "learning_rate": 1.588059541415661e-05, + "loss": 0.2652, + "step": 6131 + }, + { + "epoch": 2.57, + "learning_rate": 1.5879224053037213e-05, + "loss": 0.2657, + "step": 6132 + }, + { + "epoch": 2.57, + "learning_rate": 1.5877852522924733e-05, + "loss": 0.2627, + "step": 6133 + }, + { + "epoch": 2.57, + "learning_rate": 1.58764808238586e-05, + "loss": 0.2544, + "step": 6134 + }, + { + "epoch": 2.57, + "learning_rate": 1.587510895587824e-05, + "loss": 0.2426, + "step": 6135 + }, + { + "epoch": 2.57, + "learning_rate": 1.587373691902309e-05, + "loss": 0.2401, + "step": 6136 + }, + { + "epoch": 2.57, + "learning_rate": 1.5872364713332583e-05, + "loss": 0.2544, + "step": 6137 + }, + { + "epoch": 2.57, + "learning_rate": 1.587099233884616e-05, + "loss": 0.2522, + "step": 6138 + }, + { + "epoch": 2.57, + "learning_rate": 1.586961979560328e-05, + "loss": 0.2804, + "step": 6139 + }, + { + "epoch": 2.57, + "learning_rate": 1.586824708364338e-05, + "loss": 0.2559, + "step": 6140 + }, + { + "epoch": 2.57, + "learning_rate": 1.5866874203005928e-05, + "loss": 0.2641, + "step": 6141 + }, + { + "epoch": 2.57, + "learning_rate": 1.5865501153730383e-05, + "loss": 0.2238, + "step": 6142 + }, + { + "epoch": 2.57, + "learning_rate": 1.586412793585621e-05, + "loss": 0.2681, + "step": 6143 + }, + { + "epoch": 2.57, + "learning_rate": 1.5862754549422884e-05, + "loss": 0.272, + "step": 6144 + }, + { + "epoch": 2.57, + "learning_rate": 1.5861380994469883e-05, + "loss": 0.2693, + "step": 6145 + }, + { + "epoch": 2.57, + "learning_rate": 1.5860007271036682e-05, + "loss": 0.2674, + "step": 6146 + }, + { + "epoch": 2.57, + "learning_rate": 1.585863337916278e-05, + "loss": 0.2741, + "step": 6147 + }, + { + "epoch": 2.57, + "learning_rate": 1.5857259318887653e-05, + "loss": 0.2629, + "step": 6148 + }, + { + "epoch": 2.57, + "learning_rate": 1.585588509025081e-05, + "loss": 0.202, + "step": 6149 + }, + { + "epoch": 2.57, + "learning_rate": 1.5854510693291744e-05, + "loss": 0.26, + "step": 6150 + }, + { + "epoch": 2.58, + "learning_rate": 1.5853136128049963e-05, + "loss": 0.2583, + "step": 6151 + }, + { + "epoch": 2.58, + "learning_rate": 1.585176139456498e-05, + "loss": 0.243, + "step": 6152 + }, + { + "epoch": 2.58, + "learning_rate": 1.5850386492876307e-05, + "loss": 0.2652, + "step": 6153 + }, + { + "epoch": 2.58, + "learning_rate": 1.5849011423023463e-05, + "loss": 0.2457, + "step": 6154 + }, + { + "epoch": 2.58, + "learning_rate": 1.584763618504598e-05, + "loss": 0.2511, + "step": 6155 + }, + { + "epoch": 2.58, + "learning_rate": 1.5846260778983382e-05, + "loss": 0.2497, + "step": 6156 + }, + { + "epoch": 2.58, + "learning_rate": 1.5844885204875207e-05, + "loss": 0.2725, + "step": 6157 + }, + { + "epoch": 2.58, + "learning_rate": 1.584350946276099e-05, + "loss": 0.292, + "step": 6158 + }, + { + "epoch": 2.58, + "learning_rate": 1.5842133552680286e-05, + "loss": 0.2717, + "step": 6159 + }, + { + "epoch": 2.58, + "learning_rate": 1.5840757474672634e-05, + "loss": 0.2678, + "step": 6160 + }, + { + "epoch": 2.58, + "learning_rate": 1.5839381228777586e-05, + "loss": 0.2544, + "step": 6161 + }, + { + "epoch": 2.58, + "learning_rate": 1.583800481503471e-05, + "loss": 0.2599, + "step": 6162 + }, + { + "epoch": 2.58, + "learning_rate": 1.5836628233483568e-05, + "loss": 0.2436, + "step": 6163 + }, + { + "epoch": 2.58, + "learning_rate": 1.5835251484163728e-05, + "loss": 0.2356, + "step": 6164 + }, + { + "epoch": 2.58, + "learning_rate": 1.583387456711476e-05, + "loss": 0.2818, + "step": 6165 + }, + { + "epoch": 2.58, + "learning_rate": 1.5832497482376244e-05, + "loss": 0.2922, + "step": 6166 + }, + { + "epoch": 2.58, + "learning_rate": 1.583112022998776e-05, + "loss": 0.2658, + "step": 6167 + }, + { + "epoch": 2.58, + "learning_rate": 1.5829742809988907e-05, + "loss": 0.2644, + "step": 6168 + }, + { + "epoch": 2.58, + "learning_rate": 1.5828365222419268e-05, + "loss": 0.2485, + "step": 6169 + }, + { + "epoch": 2.58, + "learning_rate": 1.582698746731844e-05, + "loss": 0.2583, + "step": 6170 + }, + { + "epoch": 2.58, + "learning_rate": 1.582560954472603e-05, + "loss": 0.2407, + "step": 6171 + }, + { + "epoch": 2.58, + "learning_rate": 1.582423145468164e-05, + "loss": 0.2561, + "step": 6172 + }, + { + "epoch": 2.58, + "learning_rate": 1.5822853197224888e-05, + "loss": 0.2346, + "step": 6173 + }, + { + "epoch": 2.58, + "learning_rate": 1.5821474772395385e-05, + "loss": 0.2575, + "step": 6174 + }, + { + "epoch": 2.59, + "learning_rate": 1.5820096180232762e-05, + "loss": 0.2469, + "step": 6175 + }, + { + "epoch": 2.59, + "learning_rate": 1.5818717420776635e-05, + "loss": 0.2751, + "step": 6176 + }, + { + "epoch": 2.59, + "learning_rate": 1.5817338494066636e-05, + "loss": 0.251, + "step": 6177 + }, + { + "epoch": 2.59, + "learning_rate": 1.581595940014241e-05, + "loss": 0.2643, + "step": 6178 + }, + { + "epoch": 2.59, + "learning_rate": 1.5814580139043587e-05, + "loss": 0.2536, + "step": 6179 + }, + { + "epoch": 2.59, + "learning_rate": 1.581320071080982e-05, + "loss": 0.2741, + "step": 6180 + }, + { + "epoch": 2.59, + "learning_rate": 1.581182111548076e-05, + "loss": 0.2593, + "step": 6181 + }, + { + "epoch": 2.59, + "learning_rate": 1.5810441353096054e-05, + "loss": 0.2871, + "step": 6182 + }, + { + "epoch": 2.59, + "learning_rate": 1.5809061423695368e-05, + "loss": 0.2528, + "step": 6183 + }, + { + "epoch": 2.59, + "learning_rate": 1.5807681327318372e-05, + "loss": 0.2764, + "step": 6184 + }, + { + "epoch": 2.59, + "learning_rate": 1.5806301064004724e-05, + "loss": 0.2689, + "step": 6185 + }, + { + "epoch": 2.59, + "learning_rate": 1.5804920633794107e-05, + "loss": 0.2624, + "step": 6186 + }, + { + "epoch": 2.59, + "learning_rate": 1.5803540036726198e-05, + "loss": 0.2553, + "step": 6187 + }, + { + "epoch": 2.59, + "learning_rate": 1.5802159272840677e-05, + "loss": 0.24, + "step": 6188 + }, + { + "epoch": 2.59, + "learning_rate": 1.580077834217724e-05, + "loss": 0.2369, + "step": 6189 + }, + { + "epoch": 2.59, + "learning_rate": 1.5799397244775576e-05, + "loss": 0.2515, + "step": 6190 + }, + { + "epoch": 2.59, + "learning_rate": 1.5798015980675384e-05, + "loss": 0.2577, + "step": 6191 + }, + { + "epoch": 2.59, + "learning_rate": 1.579663454991637e-05, + "loss": 0.2721, + "step": 6192 + }, + { + "epoch": 2.59, + "learning_rate": 1.579525295253824e-05, + "loss": 0.2448, + "step": 6193 + }, + { + "epoch": 2.59, + "learning_rate": 1.5793871188580704e-05, + "loss": 0.2729, + "step": 6194 + }, + { + "epoch": 2.59, + "learning_rate": 1.579248925808348e-05, + "loss": 0.2618, + "step": 6195 + }, + { + "epoch": 2.59, + "learning_rate": 1.5791107161086294e-05, + "loss": 0.2722, + "step": 6196 + }, + { + "epoch": 2.59, + "learning_rate": 1.5789724897628873e-05, + "loss": 0.261, + "step": 6197 + }, + { + "epoch": 2.59, + "learning_rate": 1.578834246775095e-05, + "loss": 0.2592, + "step": 6198 + }, + { + "epoch": 2.6, + "learning_rate": 1.5786959871492254e-05, + "loss": 0.2786, + "step": 6199 + }, + { + "epoch": 2.6, + "learning_rate": 1.5785577108892537e-05, + "loss": 0.2406, + "step": 6200 + }, + { + "epoch": 2.6, + "learning_rate": 1.5784194179991534e-05, + "loss": 0.2347, + "step": 6201 + }, + { + "epoch": 2.6, + "learning_rate": 1.5782811084829005e-05, + "loss": 0.2531, + "step": 6202 + }, + { + "epoch": 2.6, + "learning_rate": 1.5781427823444704e-05, + "loss": 0.2519, + "step": 6203 + }, + { + "epoch": 2.6, + "learning_rate": 1.578004439587839e-05, + "loss": 0.2603, + "step": 6204 + }, + { + "epoch": 2.6, + "learning_rate": 1.5778660802169826e-05, + "loss": 0.2648, + "step": 6205 + }, + { + "epoch": 2.6, + "learning_rate": 1.577727704235879e-05, + "loss": 0.2316, + "step": 6206 + }, + { + "epoch": 2.6, + "learning_rate": 1.5775893116485052e-05, + "loss": 0.2395, + "step": 6207 + }, + { + "epoch": 2.6, + "learning_rate": 1.5774509024588388e-05, + "loss": 0.2766, + "step": 6208 + }, + { + "epoch": 2.6, + "learning_rate": 1.5773124766708588e-05, + "loss": 0.2556, + "step": 6209 + }, + { + "epoch": 2.6, + "learning_rate": 1.577174034288544e-05, + "loss": 0.2498, + "step": 6210 + }, + { + "epoch": 2.6, + "learning_rate": 1.5770355753158736e-05, + "loss": 0.2674, + "step": 6211 + }, + { + "epoch": 2.6, + "learning_rate": 1.5768970997568282e-05, + "loss": 0.2553, + "step": 6212 + }, + { + "epoch": 2.6, + "learning_rate": 1.576758607615387e-05, + "loss": 0.2363, + "step": 6213 + }, + { + "epoch": 2.6, + "learning_rate": 1.5766200988955315e-05, + "loss": 0.2762, + "step": 6214 + }, + { + "epoch": 2.6, + "learning_rate": 1.576481573601243e-05, + "loss": 0.2845, + "step": 6215 + }, + { + "epoch": 2.6, + "learning_rate": 1.5763430317365032e-05, + "loss": 0.2653, + "step": 6216 + }, + { + "epoch": 2.6, + "learning_rate": 1.5762044733052944e-05, + "loss": 0.2545, + "step": 6217 + }, + { + "epoch": 2.6, + "learning_rate": 1.576065898311599e-05, + "loss": 0.2495, + "step": 6218 + }, + { + "epoch": 2.6, + "learning_rate": 1.575927306759401e-05, + "loss": 0.2674, + "step": 6219 + }, + { + "epoch": 2.6, + "learning_rate": 1.575788698652684e-05, + "loss": 0.2691, + "step": 6220 + }, + { + "epoch": 2.6, + "learning_rate": 1.5756500739954312e-05, + "loss": 0.2548, + "step": 6221 + }, + { + "epoch": 2.6, + "learning_rate": 1.5755114327916282e-05, + "loss": 0.2469, + "step": 6222 + }, + { + "epoch": 2.61, + "learning_rate": 1.5753727750452596e-05, + "loss": 0.2675, + "step": 6223 + }, + { + "epoch": 2.61, + "learning_rate": 1.575234100760311e-05, + "loss": 0.2719, + "step": 6224 + }, + { + "epoch": 2.61, + "learning_rate": 1.5750954099407688e-05, + "loss": 0.2764, + "step": 6225 + }, + { + "epoch": 2.61, + "learning_rate": 1.5749567025906193e-05, + "loss": 0.2575, + "step": 6226 + }, + { + "epoch": 2.61, + "learning_rate": 1.5748179787138494e-05, + "loss": 0.2601, + "step": 6227 + }, + { + "epoch": 2.61, + "learning_rate": 1.5746792383144472e-05, + "loss": 0.2749, + "step": 6228 + }, + { + "epoch": 2.61, + "learning_rate": 1.5745404813964e-05, + "loss": 0.2885, + "step": 6229 + }, + { + "epoch": 2.61, + "learning_rate": 1.5744017079636967e-05, + "loss": 0.2574, + "step": 6230 + }, + { + "epoch": 2.61, + "learning_rate": 1.5742629180203258e-05, + "loss": 0.2448, + "step": 6231 + }, + { + "epoch": 2.61, + "learning_rate": 1.574124111570277e-05, + "loss": 0.2523, + "step": 6232 + }, + { + "epoch": 2.61, + "learning_rate": 1.57398528861754e-05, + "loss": 0.277, + "step": 6233 + }, + { + "epoch": 2.61, + "learning_rate": 1.5738464491661054e-05, + "loss": 0.2612, + "step": 6234 + }, + { + "epoch": 2.61, + "learning_rate": 1.5737075932199637e-05, + "loss": 0.2514, + "step": 6235 + }, + { + "epoch": 2.61, + "learning_rate": 1.5735687207831064e-05, + "loss": 0.2762, + "step": 6236 + }, + { + "epoch": 2.61, + "learning_rate": 1.5734298318595252e-05, + "loss": 0.266, + "step": 6237 + }, + { + "epoch": 2.61, + "learning_rate": 1.573290926453212e-05, + "loss": 0.2705, + "step": 6238 + }, + { + "epoch": 2.61, + "learning_rate": 1.5731520045681606e-05, + "loss": 0.2766, + "step": 6239 + }, + { + "epoch": 2.61, + "learning_rate": 1.5730130662083628e-05, + "loss": 0.2422, + "step": 6240 + }, + { + "epoch": 2.61, + "learning_rate": 1.572874111377813e-05, + "loss": 0.2674, + "step": 6241 + }, + { + "epoch": 2.61, + "learning_rate": 1.5727351400805054e-05, + "loss": 0.268, + "step": 6242 + }, + { + "epoch": 2.61, + "learning_rate": 1.5725961523204345e-05, + "loss": 0.2462, + "step": 6243 + }, + { + "epoch": 2.61, + "learning_rate": 1.5724571481015953e-05, + "loss": 0.2585, + "step": 6244 + }, + { + "epoch": 2.61, + "learning_rate": 1.572318127427983e-05, + "loss": 0.308, + "step": 6245 + }, + { + "epoch": 2.62, + "learning_rate": 1.5721790903035942e-05, + "loss": 0.2525, + "step": 6246 + }, + { + "epoch": 2.62, + "learning_rate": 1.572040036732425e-05, + "loss": 0.2731, + "step": 6247 + }, + { + "epoch": 2.62, + "learning_rate": 1.571900966718473e-05, + "loss": 0.246, + "step": 6248 + }, + { + "epoch": 2.62, + "learning_rate": 1.571761880265735e-05, + "loss": 0.2731, + "step": 6249 + }, + { + "epoch": 2.62, + "learning_rate": 1.5716227773782088e-05, + "loss": 0.2903, + "step": 6250 + }, + { + "epoch": 2.62, + "learning_rate": 1.5714836580598936e-05, + "loss": 0.2622, + "step": 6251 + }, + { + "epoch": 2.62, + "learning_rate": 1.5713445223147876e-05, + "loss": 0.2476, + "step": 6252 + }, + { + "epoch": 2.62, + "learning_rate": 1.5712053701468902e-05, + "loss": 0.259, + "step": 6253 + }, + { + "epoch": 2.62, + "learning_rate": 1.5710662015602016e-05, + "loss": 0.2912, + "step": 6254 + }, + { + "epoch": 2.62, + "learning_rate": 1.5709270165587216e-05, + "loss": 0.2395, + "step": 6255 + }, + { + "epoch": 2.62, + "learning_rate": 1.570787815146451e-05, + "loss": 0.2619, + "step": 6256 + }, + { + "epoch": 2.62, + "learning_rate": 1.5706485973273916e-05, + "loss": 0.2478, + "step": 6257 + }, + { + "epoch": 2.62, + "learning_rate": 1.570509363105544e-05, + "loss": 0.2812, + "step": 6258 + }, + { + "epoch": 2.62, + "learning_rate": 1.5703701124849115e-05, + "loss": 0.2282, + "step": 6259 + }, + { + "epoch": 2.62, + "learning_rate": 1.5702308454694963e-05, + "loss": 0.2505, + "step": 6260 + }, + { + "epoch": 2.62, + "learning_rate": 1.570091562063301e-05, + "loss": 0.2522, + "step": 6261 + }, + { + "epoch": 2.62, + "learning_rate": 1.5699522622703303e-05, + "loss": 0.2635, + "step": 6262 + }, + { + "epoch": 2.62, + "learning_rate": 1.569812946094587e-05, + "loss": 0.2555, + "step": 6263 + }, + { + "epoch": 2.62, + "learning_rate": 1.5696736135400765e-05, + "loss": 0.2671, + "step": 6264 + }, + { + "epoch": 2.62, + "learning_rate": 1.569534264610803e-05, + "loss": 0.2352, + "step": 6265 + }, + { + "epoch": 2.62, + "learning_rate": 1.5693948993107733e-05, + "loss": 0.2657, + "step": 6266 + }, + { + "epoch": 2.62, + "learning_rate": 1.569255517643992e-05, + "loss": 0.2495, + "step": 6267 + }, + { + "epoch": 2.62, + "learning_rate": 1.569116119614466e-05, + "loss": 0.2657, + "step": 6268 + }, + { + "epoch": 2.62, + "learning_rate": 1.5689767052262028e-05, + "loss": 0.2685, + "step": 6269 + }, + { + "epoch": 2.63, + "learning_rate": 1.568837274483208e-05, + "loss": 0.2572, + "step": 6270 + }, + { + "epoch": 2.63, + "learning_rate": 1.5686978273894914e-05, + "loss": 0.2648, + "step": 6271 + }, + { + "epoch": 2.63, + "learning_rate": 1.5685583639490598e-05, + "loss": 0.2516, + "step": 6272 + }, + { + "epoch": 2.63, + "learning_rate": 1.568418884165923e-05, + "loss": 0.2292, + "step": 6273 + }, + { + "epoch": 2.63, + "learning_rate": 1.5682793880440895e-05, + "loss": 0.2718, + "step": 6274 + }, + { + "epoch": 2.63, + "learning_rate": 1.5681398755875695e-05, + "loss": 0.2613, + "step": 6275 + }, + { + "epoch": 2.63, + "learning_rate": 1.5680003468003725e-05, + "loss": 0.2443, + "step": 6276 + }, + { + "epoch": 2.63, + "learning_rate": 1.56786080168651e-05, + "loss": 0.2538, + "step": 6277 + }, + { + "epoch": 2.63, + "learning_rate": 1.567721240249993e-05, + "loss": 0.2497, + "step": 6278 + }, + { + "epoch": 2.63, + "learning_rate": 1.5675816624948322e-05, + "loss": 0.2681, + "step": 6279 + }, + { + "epoch": 2.63, + "learning_rate": 1.56744206842504e-05, + "loss": 0.2525, + "step": 6280 + }, + { + "epoch": 2.63, + "learning_rate": 1.5673024580446294e-05, + "loss": 0.2496, + "step": 6281 + }, + { + "epoch": 2.63, + "learning_rate": 1.567162831357613e-05, + "loss": 0.2768, + "step": 6282 + }, + { + "epoch": 2.63, + "learning_rate": 1.5670231883680044e-05, + "loss": 0.2721, + "step": 6283 + }, + { + "epoch": 2.63, + "learning_rate": 1.566883529079817e-05, + "loss": 0.274, + "step": 6284 + }, + { + "epoch": 2.63, + "learning_rate": 1.5667438534970664e-05, + "loss": 0.283, + "step": 6285 + }, + { + "epoch": 2.63, + "learning_rate": 1.566604161623766e-05, + "loss": 0.284, + "step": 6286 + }, + { + "epoch": 2.63, + "learning_rate": 1.566464453463932e-05, + "loss": 0.2795, + "step": 6287 + }, + { + "epoch": 2.63, + "learning_rate": 1.5663247290215797e-05, + "loss": 0.2495, + "step": 6288 + }, + { + "epoch": 2.63, + "learning_rate": 1.566184988300726e-05, + "loss": 0.261, + "step": 6289 + }, + { + "epoch": 2.63, + "learning_rate": 1.5660452313053873e-05, + "loss": 0.2663, + "step": 6290 + }, + { + "epoch": 2.63, + "learning_rate": 1.56590545803958e-05, + "loss": 0.2632, + "step": 6291 + }, + { + "epoch": 2.63, + "learning_rate": 1.565765668507323e-05, + "loss": 0.2589, + "step": 6292 + }, + { + "epoch": 2.63, + "learning_rate": 1.565625862712634e-05, + "loss": 0.26, + "step": 6293 + }, + { + "epoch": 2.64, + "learning_rate": 1.5654860406595312e-05, + "loss": 0.2677, + "step": 6294 + }, + { + "epoch": 2.64, + "learning_rate": 1.5653462023520344e-05, + "loss": 0.2423, + "step": 6295 + }, + { + "epoch": 2.64, + "learning_rate": 1.5652063477941625e-05, + "loss": 0.2646, + "step": 6296 + }, + { + "epoch": 2.64, + "learning_rate": 1.5650664769899356e-05, + "loss": 0.2359, + "step": 6297 + }, + { + "epoch": 2.64, + "learning_rate": 1.564926589943374e-05, + "loss": 0.2378, + "step": 6298 + }, + { + "epoch": 2.64, + "learning_rate": 1.5647866866584992e-05, + "loss": 0.2465, + "step": 6299 + }, + { + "epoch": 2.64, + "learning_rate": 1.5646467671393323e-05, + "loss": 0.2394, + "step": 6300 + }, + { + "epoch": 2.64, + "learning_rate": 1.564506831389895e-05, + "loss": 0.2737, + "step": 6301 + }, + { + "epoch": 2.64, + "learning_rate": 1.5643668794142102e-05, + "loss": 0.2833, + "step": 6302 + }, + { + "epoch": 2.64, + "learning_rate": 1.5642269112162998e-05, + "loss": 0.2895, + "step": 6303 + }, + { + "epoch": 2.64, + "learning_rate": 1.5640869268001875e-05, + "loss": 0.2499, + "step": 6304 + }, + { + "epoch": 2.64, + "learning_rate": 1.563946926169897e-05, + "loss": 0.2881, + "step": 6305 + }, + { + "epoch": 2.64, + "learning_rate": 1.5638069093294527e-05, + "loss": 0.271, + "step": 6306 + }, + { + "epoch": 2.64, + "learning_rate": 1.5636668762828793e-05, + "loss": 0.2463, + "step": 6307 + }, + { + "epoch": 2.64, + "learning_rate": 1.5635268270342015e-05, + "loss": 0.2836, + "step": 6308 + }, + { + "epoch": 2.64, + "learning_rate": 1.5633867615874453e-05, + "loss": 0.2646, + "step": 6309 + }, + { + "epoch": 2.64, + "learning_rate": 1.5632466799466362e-05, + "loss": 0.2348, + "step": 6310 + }, + { + "epoch": 2.64, + "learning_rate": 1.5631065821158016e-05, + "loss": 0.2564, + "step": 6311 + }, + { + "epoch": 2.64, + "learning_rate": 1.5629664680989677e-05, + "loss": 0.2469, + "step": 6312 + }, + { + "epoch": 2.64, + "learning_rate": 1.562826337900162e-05, + "loss": 0.2531, + "step": 6313 + }, + { + "epoch": 2.64, + "learning_rate": 1.562686191523413e-05, + "loss": 0.2948, + "step": 6314 + }, + { + "epoch": 2.64, + "learning_rate": 1.562546028972749e-05, + "loss": 0.2857, + "step": 6315 + }, + { + "epoch": 2.64, + "learning_rate": 1.562405850252198e-05, + "loss": 0.2549, + "step": 6316 + }, + { + "epoch": 2.64, + "learning_rate": 1.5622656553657904e-05, + "loss": 0.2772, + "step": 6317 + }, + { + "epoch": 2.65, + "learning_rate": 1.5621254443175553e-05, + "loss": 0.28, + "step": 6318 + }, + { + "epoch": 2.65, + "learning_rate": 1.561985217111523e-05, + "loss": 0.2754, + "step": 6319 + }, + { + "epoch": 2.65, + "learning_rate": 1.5618449737517242e-05, + "loss": 0.2899, + "step": 6320 + }, + { + "epoch": 2.65, + "learning_rate": 1.5617047142421904e-05, + "loss": 0.2448, + "step": 6321 + }, + { + "epoch": 2.65, + "learning_rate": 1.5615644385869532e-05, + "loss": 0.2475, + "step": 6322 + }, + { + "epoch": 2.65, + "learning_rate": 1.5614241467900444e-05, + "loss": 0.276, + "step": 6323 + }, + { + "epoch": 2.65, + "learning_rate": 1.5612838388554966e-05, + "loss": 0.2666, + "step": 6324 + }, + { + "epoch": 2.65, + "learning_rate": 1.561143514787343e-05, + "loss": 0.2741, + "step": 6325 + }, + { + "epoch": 2.65, + "learning_rate": 1.561003174589617e-05, + "loss": 0.2765, + "step": 6326 + }, + { + "epoch": 2.65, + "learning_rate": 1.5608628182663527e-05, + "loss": 0.2711, + "step": 6327 + }, + { + "epoch": 2.65, + "learning_rate": 1.5607224458215843e-05, + "loss": 0.2404, + "step": 6328 + }, + { + "epoch": 2.65, + "learning_rate": 1.560582057259347e-05, + "loss": 0.2638, + "step": 6329 + }, + { + "epoch": 2.65, + "learning_rate": 1.5604416525836756e-05, + "loss": 0.2678, + "step": 6330 + }, + { + "epoch": 2.65, + "learning_rate": 1.560301231798606e-05, + "loss": 0.2781, + "step": 6331 + }, + { + "epoch": 2.65, + "learning_rate": 1.560160794908175e-05, + "loss": 0.2909, + "step": 6332 + }, + { + "epoch": 2.65, + "learning_rate": 1.560020341916419e-05, + "loss": 0.2679, + "step": 6333 + }, + { + "epoch": 2.65, + "learning_rate": 1.5598798728273753e-05, + "loss": 0.251, + "step": 6334 + }, + { + "epoch": 2.65, + "learning_rate": 1.5597393876450813e-05, + "loss": 0.2647, + "step": 6335 + }, + { + "epoch": 2.65, + "learning_rate": 1.5595988863735758e-05, + "loss": 0.2343, + "step": 6336 + }, + { + "epoch": 2.65, + "learning_rate": 1.5594583690168966e-05, + "loss": 0.2831, + "step": 6337 + }, + { + "epoch": 2.65, + "learning_rate": 1.559317835579083e-05, + "loss": 0.2257, + "step": 6338 + }, + { + "epoch": 2.65, + "learning_rate": 1.5591772860641743e-05, + "loss": 0.2514, + "step": 6339 + }, + { + "epoch": 2.65, + "learning_rate": 1.5590367204762113e-05, + "loss": 0.2614, + "step": 6340 + }, + { + "epoch": 2.65, + "learning_rate": 1.5588961388192333e-05, + "loss": 0.2447, + "step": 6341 + }, + { + "epoch": 2.66, + "learning_rate": 1.558755541097282e-05, + "loss": 0.2514, + "step": 6342 + }, + { + "epoch": 2.66, + "learning_rate": 1.5586149273143982e-05, + "loss": 0.2586, + "step": 6343 + }, + { + "epoch": 2.66, + "learning_rate": 1.5584742974746242e-05, + "loss": 0.2454, + "step": 6344 + }, + { + "epoch": 2.66, + "learning_rate": 1.5583336515820024e-05, + "loss": 0.2618, + "step": 6345 + }, + { + "epoch": 2.66, + "learning_rate": 1.558192989640575e-05, + "loss": 0.2594, + "step": 6346 + }, + { + "epoch": 2.66, + "learning_rate": 1.5580523116543853e-05, + "loss": 0.2391, + "step": 6347 + }, + { + "epoch": 2.66, + "learning_rate": 1.557911617627477e-05, + "loss": 0.2645, + "step": 6348 + }, + { + "epoch": 2.66, + "learning_rate": 1.557770907563895e-05, + "loss": 0.2572, + "step": 6349 + }, + { + "epoch": 2.66, + "learning_rate": 1.5576301814676825e-05, + "loss": 0.2587, + "step": 6350 + }, + { + "epoch": 2.66, + "learning_rate": 1.5574894393428856e-05, + "loss": 0.2847, + "step": 6351 + }, + { + "epoch": 2.66, + "learning_rate": 1.5573486811935494e-05, + "loss": 0.2696, + "step": 6352 + }, + { + "epoch": 2.66, + "learning_rate": 1.55720790702372e-05, + "loss": 0.2752, + "step": 6353 + }, + { + "epoch": 2.66, + "learning_rate": 1.557067116837444e-05, + "loss": 0.233, + "step": 6354 + }, + { + "epoch": 2.66, + "learning_rate": 1.5569263106387676e-05, + "loss": 0.2646, + "step": 6355 + }, + { + "epoch": 2.66, + "learning_rate": 1.5567854884317387e-05, + "loss": 0.273, + "step": 6356 + }, + { + "epoch": 2.66, + "learning_rate": 1.5566446502204052e-05, + "loss": 0.2578, + "step": 6357 + }, + { + "epoch": 2.66, + "learning_rate": 1.556503796008815e-05, + "loss": 0.2444, + "step": 6358 + }, + { + "epoch": 2.66, + "learning_rate": 1.5563629258010173e-05, + "loss": 0.2624, + "step": 6359 + }, + { + "epoch": 2.66, + "learning_rate": 1.556222039601061e-05, + "loss": 0.2674, + "step": 6360 + }, + { + "epoch": 2.66, + "learning_rate": 1.5560811374129955e-05, + "loss": 0.2442, + "step": 6361 + }, + { + "epoch": 2.66, + "learning_rate": 1.5559402192408712e-05, + "loss": 0.283, + "step": 6362 + }, + { + "epoch": 2.66, + "learning_rate": 1.5557992850887388e-05, + "loss": 0.2245, + "step": 6363 + }, + { + "epoch": 2.66, + "learning_rate": 1.5556583349606492e-05, + "loss": 0.2622, + "step": 6364 + }, + { + "epoch": 2.66, + "learning_rate": 1.5555173688606537e-05, + "loss": 0.2813, + "step": 6365 + }, + { + "epoch": 2.67, + "learning_rate": 1.5553763867928043e-05, + "loss": 0.2728, + "step": 6366 + }, + { + "epoch": 2.67, + "learning_rate": 1.5552353887611536e-05, + "loss": 0.2805, + "step": 6367 + }, + { + "epoch": 2.67, + "learning_rate": 1.5550943747697546e-05, + "loss": 0.2557, + "step": 6368 + }, + { + "epoch": 2.67, + "learning_rate": 1.5549533448226598e-05, + "loss": 0.2733, + "step": 6369 + }, + { + "epoch": 2.67, + "learning_rate": 1.554812298923924e-05, + "loss": 0.2685, + "step": 6370 + }, + { + "epoch": 2.67, + "learning_rate": 1.5546712370776005e-05, + "loss": 0.2678, + "step": 6371 + }, + { + "epoch": 2.67, + "learning_rate": 1.554530159287745e-05, + "loss": 0.2202, + "step": 6372 + }, + { + "epoch": 2.67, + "learning_rate": 1.5543890655584122e-05, + "loss": 0.2725, + "step": 6373 + }, + { + "epoch": 2.67, + "learning_rate": 1.5542479558936572e-05, + "loss": 0.2627, + "step": 6374 + }, + { + "epoch": 2.67, + "learning_rate": 1.554106830297537e-05, + "loss": 0.2835, + "step": 6375 + }, + { + "epoch": 2.67, + "learning_rate": 1.5539656887741076e-05, + "loss": 0.2782, + "step": 6376 + }, + { + "epoch": 2.67, + "learning_rate": 1.5538245313274258e-05, + "loss": 0.278, + "step": 6377 + }, + { + "epoch": 2.67, + "learning_rate": 1.5536833579615495e-05, + "loss": 0.2575, + "step": 6378 + }, + { + "epoch": 2.67, + "learning_rate": 1.5535421686805367e-05, + "loss": 0.242, + "step": 6379 + }, + { + "epoch": 2.67, + "learning_rate": 1.553400963488445e-05, + "loss": 0.2558, + "step": 6380 + }, + { + "epoch": 2.67, + "learning_rate": 1.553259742389334e-05, + "loss": 0.2637, + "step": 6381 + }, + { + "epoch": 2.67, + "learning_rate": 1.553118505387262e-05, + "loss": 0.2492, + "step": 6382 + }, + { + "epoch": 2.67, + "learning_rate": 1.55297725248629e-05, + "loss": 0.2669, + "step": 6383 + }, + { + "epoch": 2.67, + "learning_rate": 1.5528359836904774e-05, + "loss": 0.2425, + "step": 6384 + }, + { + "epoch": 2.67, + "learning_rate": 1.552694699003885e-05, + "loss": 0.2539, + "step": 6385 + }, + { + "epoch": 2.67, + "learning_rate": 1.5525533984305742e-05, + "loss": 0.2474, + "step": 6386 + }, + { + "epoch": 2.67, + "learning_rate": 1.552412081974606e-05, + "loss": 0.254, + "step": 6387 + }, + { + "epoch": 2.67, + "learning_rate": 1.5522707496400425e-05, + "loss": 0.2662, + "step": 6388 + }, + { + "epoch": 2.67, + "learning_rate": 1.552129401430947e-05, + "loss": 0.2675, + "step": 6389 + }, + { + "epoch": 2.68, + "learning_rate": 1.5519880373513814e-05, + "loss": 0.2582, + "step": 6390 + }, + { + "epoch": 2.68, + "learning_rate": 1.5518466574054097e-05, + "loss": 0.2468, + "step": 6391 + }, + { + "epoch": 2.68, + "learning_rate": 1.5517052615970955e-05, + "loss": 0.2305, + "step": 6392 + }, + { + "epoch": 2.68, + "learning_rate": 1.551563849930503e-05, + "loss": 0.3024, + "step": 6393 + }, + { + "epoch": 2.68, + "learning_rate": 1.551422422409697e-05, + "loss": 0.2718, + "step": 6394 + }, + { + "epoch": 2.68, + "learning_rate": 1.551280979038743e-05, + "loss": 0.2705, + "step": 6395 + }, + { + "epoch": 2.68, + "learning_rate": 1.551139519821706e-05, + "loss": 0.2716, + "step": 6396 + }, + { + "epoch": 2.68, + "learning_rate": 1.5509980447626532e-05, + "loss": 0.2879, + "step": 6397 + }, + { + "epoch": 2.68, + "learning_rate": 1.5508565538656505e-05, + "loss": 0.2648, + "step": 6398 + }, + { + "epoch": 2.68, + "learning_rate": 1.5507150471347644e-05, + "loss": 0.2476, + "step": 6399 + }, + { + "epoch": 2.68, + "learning_rate": 1.5505735245740635e-05, + "loss": 0.2499, + "step": 6400 + }, + { + "epoch": 2.68, + "learning_rate": 1.5504319861876148e-05, + "loss": 0.2537, + "step": 6401 + }, + { + "epoch": 2.68, + "learning_rate": 1.5502904319794875e-05, + "loss": 0.2677, + "step": 6402 + }, + { + "epoch": 2.68, + "learning_rate": 1.55014886195375e-05, + "loss": 0.2583, + "step": 6403 + }, + { + "epoch": 2.68, + "learning_rate": 1.5500072761144712e-05, + "loss": 0.2513, + "step": 6404 + }, + { + "epoch": 2.68, + "learning_rate": 1.5498656744657222e-05, + "loss": 0.2343, + "step": 6405 + }, + { + "epoch": 2.68, + "learning_rate": 1.5497240570115716e-05, + "loss": 0.2468, + "step": 6406 + }, + { + "epoch": 2.68, + "learning_rate": 1.5495824237560907e-05, + "loss": 0.2654, + "step": 6407 + }, + { + "epoch": 2.68, + "learning_rate": 1.5494407747033513e-05, + "loss": 0.2677, + "step": 6408 + }, + { + "epoch": 2.68, + "learning_rate": 1.5492991098574243e-05, + "loss": 0.2488, + "step": 6409 + }, + { + "epoch": 2.68, + "learning_rate": 1.5491574292223814e-05, + "loss": 0.2519, + "step": 6410 + }, + { + "epoch": 2.68, + "learning_rate": 1.5490157328022957e-05, + "loss": 0.2988, + "step": 6411 + }, + { + "epoch": 2.68, + "learning_rate": 1.54887402060124e-05, + "loss": 0.281, + "step": 6412 + }, + { + "epoch": 2.68, + "learning_rate": 1.5487322926232876e-05, + "loss": 0.262, + "step": 6413 + }, + { + "epoch": 2.69, + "learning_rate": 1.5485905488725126e-05, + "loss": 0.2738, + "step": 6414 + }, + { + "epoch": 2.69, + "learning_rate": 1.548448789352989e-05, + "loss": 0.2776, + "step": 6415 + }, + { + "epoch": 2.69, + "learning_rate": 1.5483070140687916e-05, + "loss": 0.2465, + "step": 6416 + }, + { + "epoch": 2.69, + "learning_rate": 1.5481652230239954e-05, + "loss": 0.2877, + "step": 6417 + }, + { + "epoch": 2.69, + "learning_rate": 1.5480234162226763e-05, + "loss": 0.2721, + "step": 6418 + }, + { + "epoch": 2.69, + "learning_rate": 1.5478815936689107e-05, + "loss": 0.2494, + "step": 6419 + }, + { + "epoch": 2.69, + "learning_rate": 1.5477397553667745e-05, + "loss": 0.2774, + "step": 6420 + }, + { + "epoch": 2.69, + "learning_rate": 1.5475979013203456e-05, + "loss": 0.2249, + "step": 6421 + }, + { + "epoch": 2.69, + "learning_rate": 1.5474560315337007e-05, + "loss": 0.2835, + "step": 6422 + }, + { + "epoch": 2.69, + "learning_rate": 1.547314146010918e-05, + "loss": 0.2793, + "step": 6423 + }, + { + "epoch": 2.69, + "learning_rate": 1.5471722447560757e-05, + "loss": 0.2306, + "step": 6424 + }, + { + "epoch": 2.69, + "learning_rate": 1.547030327773253e-05, + "loss": 0.2505, + "step": 6425 + }, + { + "epoch": 2.69, + "learning_rate": 1.546888395066529e-05, + "loss": 0.2627, + "step": 6426 + }, + { + "epoch": 2.69, + "learning_rate": 1.546746446639983e-05, + "loss": 0.2706, + "step": 6427 + }, + { + "epoch": 2.69, + "learning_rate": 1.546604482497696e-05, + "loss": 0.2982, + "step": 6428 + }, + { + "epoch": 2.69, + "learning_rate": 1.546462502643748e-05, + "loss": 0.2484, + "step": 6429 + }, + { + "epoch": 2.69, + "learning_rate": 1.5463205070822202e-05, + "loss": 0.2251, + "step": 6430 + }, + { + "epoch": 2.69, + "learning_rate": 1.5461784958171945e-05, + "loss": 0.2504, + "step": 6431 + }, + { + "epoch": 2.69, + "learning_rate": 1.5460364688527523e-05, + "loss": 0.2677, + "step": 6432 + }, + { + "epoch": 2.69, + "learning_rate": 1.5458944261929763e-05, + "loss": 0.243, + "step": 6433 + }, + { + "epoch": 2.69, + "learning_rate": 1.5457523678419498e-05, + "loss": 0.2672, + "step": 6434 + }, + { + "epoch": 2.69, + "learning_rate": 1.5456102938037555e-05, + "loss": 0.2635, + "step": 6435 + }, + { + "epoch": 2.69, + "learning_rate": 1.5454682040824776e-05, + "loss": 0.2272, + "step": 6436 + }, + { + "epoch": 2.69, + "learning_rate": 1.5453260986822e-05, + "loss": 0.2772, + "step": 6437 + }, + { + "epoch": 2.7, + "learning_rate": 1.545183977607008e-05, + "loss": 0.2579, + "step": 6438 + }, + { + "epoch": 2.7, + "learning_rate": 1.545041840860986e-05, + "loss": 0.2872, + "step": 6439 + }, + { + "epoch": 2.7, + "learning_rate": 1.5448996884482198e-05, + "loss": 0.2657, + "step": 6440 + }, + { + "epoch": 2.7, + "learning_rate": 1.5447575203727964e-05, + "loss": 0.2656, + "step": 6441 + }, + { + "epoch": 2.7, + "learning_rate": 1.5446153366388007e-05, + "loss": 0.2635, + "step": 6442 + }, + { + "epoch": 2.7, + "learning_rate": 1.5444731372503206e-05, + "loss": 0.235, + "step": 6443 + }, + { + "epoch": 2.7, + "learning_rate": 1.5443309222114434e-05, + "loss": 0.2488, + "step": 6444 + }, + { + "epoch": 2.7, + "learning_rate": 1.544188691526257e-05, + "loss": 0.2692, + "step": 6445 + }, + { + "epoch": 2.7, + "learning_rate": 1.5440464451988494e-05, + "loss": 0.2598, + "step": 6446 + }, + { + "epoch": 2.7, + "learning_rate": 1.5439041832333095e-05, + "loss": 0.2663, + "step": 6447 + }, + { + "epoch": 2.7, + "learning_rate": 1.5437619056337267e-05, + "loss": 0.2528, + "step": 6448 + }, + { + "epoch": 2.7, + "learning_rate": 1.5436196124041902e-05, + "loss": 0.2391, + "step": 6449 + }, + { + "epoch": 2.7, + "learning_rate": 1.5434773035487905e-05, + "loss": 0.2803, + "step": 6450 + }, + { + "epoch": 2.7, + "learning_rate": 1.543334979071618e-05, + "loss": 0.2625, + "step": 6451 + }, + { + "epoch": 2.7, + "learning_rate": 1.5431926389767637e-05, + "loss": 0.2697, + "step": 6452 + }, + { + "epoch": 2.7, + "learning_rate": 1.543050283268319e-05, + "loss": 0.2431, + "step": 6453 + }, + { + "epoch": 2.7, + "learning_rate": 1.5429079119503752e-05, + "loss": 0.2628, + "step": 6454 + }, + { + "epoch": 2.7, + "learning_rate": 1.542765525027026e-05, + "loss": 0.266, + "step": 6455 + }, + { + "epoch": 2.7, + "learning_rate": 1.542623122502363e-05, + "loss": 0.2715, + "step": 6456 + }, + { + "epoch": 2.7, + "learning_rate": 1.54248070438048e-05, + "loss": 0.2704, + "step": 6457 + }, + { + "epoch": 2.7, + "learning_rate": 1.5423382706654705e-05, + "loss": 0.2164, + "step": 6458 + }, + { + "epoch": 2.7, + "learning_rate": 1.542195821361429e-05, + "loss": 0.2669, + "step": 6459 + }, + { + "epoch": 2.7, + "learning_rate": 1.5420533564724495e-05, + "loss": 0.2869, + "step": 6460 + }, + { + "epoch": 2.71, + "learning_rate": 1.5419108760026273e-05, + "loss": 0.2427, + "step": 6461 + }, + { + "epoch": 2.71, + "learning_rate": 1.5417683799560576e-05, + "loss": 0.248, + "step": 6462 + }, + { + "epoch": 2.71, + "learning_rate": 1.541625868336837e-05, + "loss": 0.2846, + "step": 6463 + }, + { + "epoch": 2.71, + "learning_rate": 1.5414833411490612e-05, + "loss": 0.2726, + "step": 6464 + }, + { + "epoch": 2.71, + "learning_rate": 1.5413407983968274e-05, + "loss": 0.2504, + "step": 6465 + }, + { + "epoch": 2.71, + "learning_rate": 1.5411982400842328e-05, + "loss": 0.239, + "step": 6466 + }, + { + "epoch": 2.71, + "learning_rate": 1.5410556662153747e-05, + "loss": 0.2547, + "step": 6467 + }, + { + "epoch": 2.71, + "learning_rate": 1.5409130767943516e-05, + "loss": 0.284, + "step": 6468 + }, + { + "epoch": 2.71, + "learning_rate": 1.5407704718252625e-05, + "loss": 0.2341, + "step": 6469 + }, + { + "epoch": 2.71, + "learning_rate": 1.5406278513122058e-05, + "loss": 0.2427, + "step": 6470 + }, + { + "epoch": 2.71, + "learning_rate": 1.5404852152592814e-05, + "loss": 0.2431, + "step": 6471 + }, + { + "epoch": 2.71, + "learning_rate": 1.540342563670589e-05, + "loss": 0.2642, + "step": 6472 + }, + { + "epoch": 2.71, + "learning_rate": 1.5401998965502293e-05, + "loss": 0.2666, + "step": 6473 + }, + { + "epoch": 2.71, + "learning_rate": 1.5400572139023026e-05, + "loss": 0.2834, + "step": 6474 + }, + { + "epoch": 2.71, + "learning_rate": 1.5399145157309105e-05, + "loss": 0.281, + "step": 6475 + }, + { + "epoch": 2.71, + "learning_rate": 1.539771802040155e-05, + "loss": 0.2498, + "step": 6476 + }, + { + "epoch": 2.71, + "learning_rate": 1.539629072834138e-05, + "loss": 0.2607, + "step": 6477 + }, + { + "epoch": 2.71, + "learning_rate": 1.539486328116962e-05, + "loss": 0.2705, + "step": 6478 + }, + { + "epoch": 2.71, + "learning_rate": 1.5393435678927305e-05, + "loss": 0.2686, + "step": 6479 + }, + { + "epoch": 2.71, + "learning_rate": 1.5392007921655466e-05, + "loss": 0.2472, + "step": 6480 + }, + { + "epoch": 2.71, + "learning_rate": 1.5390580009395147e-05, + "loss": 0.2202, + "step": 6481 + }, + { + "epoch": 2.71, + "learning_rate": 1.5389151942187383e-05, + "loss": 0.2606, + "step": 6482 + }, + { + "epoch": 2.71, + "learning_rate": 1.538772372007323e-05, + "loss": 0.2618, + "step": 6483 + }, + { + "epoch": 2.71, + "learning_rate": 1.5386295343093743e-05, + "loss": 0.2498, + "step": 6484 + }, + { + "epoch": 2.72, + "learning_rate": 1.5384866811289976e-05, + "loss": 0.263, + "step": 6485 + }, + { + "epoch": 2.72, + "learning_rate": 1.538343812470299e-05, + "loss": 0.2487, + "step": 6486 + }, + { + "epoch": 2.72, + "learning_rate": 1.538200928337385e-05, + "loss": 0.273, + "step": 6487 + }, + { + "epoch": 2.72, + "learning_rate": 1.538058028734363e-05, + "loss": 0.2388, + "step": 6488 + }, + { + "epoch": 2.72, + "learning_rate": 1.5379151136653407e-05, + "loss": 0.2524, + "step": 6489 + }, + { + "epoch": 2.72, + "learning_rate": 1.5377721831344258e-05, + "loss": 0.244, + "step": 6490 + }, + { + "epoch": 2.72, + "learning_rate": 1.537629237145726e-05, + "loss": 0.2628, + "step": 6491 + }, + { + "epoch": 2.72, + "learning_rate": 1.5374862757033518e-05, + "loss": 0.2961, + "step": 6492 + }, + { + "epoch": 2.72, + "learning_rate": 1.5373432988114113e-05, + "loss": 0.2753, + "step": 6493 + }, + { + "epoch": 2.72, + "learning_rate": 1.537200306474014e-05, + "loss": 0.2629, + "step": 6494 + }, + { + "epoch": 2.72, + "learning_rate": 1.5370572986952712e-05, + "loss": 0.2631, + "step": 6495 + }, + { + "epoch": 2.72, + "learning_rate": 1.5369142754792926e-05, + "loss": 0.2519, + "step": 6496 + }, + { + "epoch": 2.72, + "learning_rate": 1.5367712368301897e-05, + "loss": 0.2696, + "step": 6497 + }, + { + "epoch": 2.72, + "learning_rate": 1.536628182752074e-05, + "loss": 0.2714, + "step": 6498 + }, + { + "epoch": 2.72, + "learning_rate": 1.536485113249057e-05, + "loss": 0.2673, + "step": 6499 + }, + { + "epoch": 2.72, + "learning_rate": 1.5363420283252518e-05, + "loss": 0.2689, + "step": 6500 + }, + { + "epoch": 2.72, + "learning_rate": 1.536198927984771e-05, + "loss": 0.2581, + "step": 6501 + }, + { + "epoch": 2.72, + "learning_rate": 1.5360558122317275e-05, + "loss": 0.2747, + "step": 6502 + }, + { + "epoch": 2.72, + "learning_rate": 1.5359126810702357e-05, + "loss": 0.2599, + "step": 6503 + }, + { + "epoch": 2.72, + "learning_rate": 1.5357695345044096e-05, + "loss": 0.2317, + "step": 6504 + }, + { + "epoch": 2.72, + "learning_rate": 1.5356263725383633e-05, + "loss": 0.264, + "step": 6505 + }, + { + "epoch": 2.72, + "learning_rate": 1.5354831951762122e-05, + "loss": 0.2814, + "step": 6506 + }, + { + "epoch": 2.72, + "learning_rate": 1.535340002422072e-05, + "loss": 0.2611, + "step": 6507 + }, + { + "epoch": 2.72, + "learning_rate": 1.5351967942800584e-05, + "loss": 0.2676, + "step": 6508 + }, + { + "epoch": 2.73, + "learning_rate": 1.535053570754288e-05, + "loss": 0.2563, + "step": 6509 + }, + { + "epoch": 2.73, + "learning_rate": 1.5349103318488772e-05, + "loss": 0.2836, + "step": 6510 + }, + { + "epoch": 2.73, + "learning_rate": 1.534767077567944e-05, + "loss": 0.257, + "step": 6511 + }, + { + "epoch": 2.73, + "learning_rate": 1.5346238079156055e-05, + "loss": 0.2543, + "step": 6512 + }, + { + "epoch": 2.73, + "learning_rate": 1.53448052289598e-05, + "loss": 0.2448, + "step": 6513 + }, + { + "epoch": 2.73, + "learning_rate": 1.5343372225131863e-05, + "loss": 0.2426, + "step": 6514 + }, + { + "epoch": 2.73, + "learning_rate": 1.5341939067713433e-05, + "loss": 0.2579, + "step": 6515 + }, + { + "epoch": 2.73, + "learning_rate": 1.5340505756745703e-05, + "loss": 0.2629, + "step": 6516 + }, + { + "epoch": 2.73, + "learning_rate": 1.5339072292269874e-05, + "loss": 0.2614, + "step": 6517 + }, + { + "epoch": 2.73, + "learning_rate": 1.5337638674327152e-05, + "loss": 0.2702, + "step": 6518 + }, + { + "epoch": 2.73, + "learning_rate": 1.5336204902958743e-05, + "loss": 0.249, + "step": 6519 + }, + { + "epoch": 2.73, + "learning_rate": 1.5334770978205857e-05, + "loss": 0.2522, + "step": 6520 + }, + { + "epoch": 2.73, + "learning_rate": 1.5333336900109715e-05, + "loss": 0.2293, + "step": 6521 + }, + { + "epoch": 2.73, + "learning_rate": 1.5331902668711536e-05, + "loss": 0.2922, + "step": 6522 + }, + { + "epoch": 2.73, + "learning_rate": 1.5330468284052544e-05, + "loss": 0.2786, + "step": 6523 + }, + { + "epoch": 2.73, + "learning_rate": 1.5329033746173975e-05, + "loss": 0.2407, + "step": 6524 + }, + { + "epoch": 2.73, + "learning_rate": 1.5327599055117057e-05, + "loss": 0.2681, + "step": 6525 + }, + { + "epoch": 2.73, + "learning_rate": 1.5326164210923032e-05, + "loss": 0.2567, + "step": 6526 + }, + { + "epoch": 2.73, + "learning_rate": 1.5324729213633146e-05, + "loss": 0.268, + "step": 6527 + }, + { + "epoch": 2.73, + "learning_rate": 1.532329406328864e-05, + "loss": 0.2753, + "step": 6528 + }, + { + "epoch": 2.73, + "learning_rate": 1.5321858759930776e-05, + "loss": 0.256, + "step": 6529 + }, + { + "epoch": 2.73, + "learning_rate": 1.53204233036008e-05, + "loss": 0.2613, + "step": 6530 + }, + { + "epoch": 2.73, + "learning_rate": 1.531898769433998e-05, + "loss": 0.2602, + "step": 6531 + }, + { + "epoch": 2.73, + "learning_rate": 1.5317551932189574e-05, + "loss": 0.2707, + "step": 6532 + }, + { + "epoch": 2.74, + "learning_rate": 1.531611601719086e-05, + "loss": 0.2616, + "step": 6533 + }, + { + "epoch": 2.74, + "learning_rate": 1.5314679949385105e-05, + "loss": 0.2707, + "step": 6534 + }, + { + "epoch": 2.74, + "learning_rate": 1.5313243728813598e-05, + "loss": 0.2467, + "step": 6535 + }, + { + "epoch": 2.74, + "learning_rate": 1.5311807355517608e-05, + "loss": 0.257, + "step": 6536 + }, + { + "epoch": 2.74, + "learning_rate": 1.531037082953843e-05, + "loss": 0.2538, + "step": 6537 + }, + { + "epoch": 2.74, + "learning_rate": 1.530893415091736e-05, + "loss": 0.2585, + "step": 6538 + }, + { + "epoch": 2.74, + "learning_rate": 1.5307497319695683e-05, + "loss": 0.2764, + "step": 6539 + }, + { + "epoch": 2.74, + "learning_rate": 1.5306060335914706e-05, + "loss": 0.2576, + "step": 6540 + }, + { + "epoch": 2.74, + "learning_rate": 1.5304623199615737e-05, + "loss": 0.2602, + "step": 6541 + }, + { + "epoch": 2.74, + "learning_rate": 1.530318591084008e-05, + "loss": 0.2584, + "step": 6542 + }, + { + "epoch": 2.74, + "learning_rate": 1.530174846962905e-05, + "loss": 0.2904, + "step": 6543 + }, + { + "epoch": 2.74, + "learning_rate": 1.530031087602396e-05, + "loss": 0.2618, + "step": 6544 + }, + { + "epoch": 2.74, + "learning_rate": 1.5298873130066144e-05, + "loss": 0.272, + "step": 6545 + }, + { + "epoch": 2.74, + "learning_rate": 1.529743523179692e-05, + "loss": 0.2769, + "step": 6546 + }, + { + "epoch": 2.74, + "learning_rate": 1.529599718125762e-05, + "loss": 0.228, + "step": 6547 + }, + { + "epoch": 2.74, + "learning_rate": 1.5294558978489583e-05, + "loss": 0.2458, + "step": 6548 + }, + { + "epoch": 2.74, + "learning_rate": 1.5293120623534145e-05, + "loss": 0.2771, + "step": 6549 + }, + { + "epoch": 2.74, + "learning_rate": 1.529168211643265e-05, + "loss": 0.244, + "step": 6550 + }, + { + "epoch": 2.74, + "learning_rate": 1.529024345722645e-05, + "loss": 0.2994, + "step": 6551 + }, + { + "epoch": 2.74, + "learning_rate": 1.5288804645956896e-05, + "loss": 0.2486, + "step": 6552 + }, + { + "epoch": 2.74, + "learning_rate": 1.5287365682665346e-05, + "loss": 0.2665, + "step": 6553 + }, + { + "epoch": 2.74, + "learning_rate": 1.5285926567393158e-05, + "loss": 0.2559, + "step": 6554 + }, + { + "epoch": 2.74, + "learning_rate": 1.5284487300181707e-05, + "loss": 0.2747, + "step": 6555 + }, + { + "epoch": 2.74, + "learning_rate": 1.5283047881072357e-05, + "loss": 0.2544, + "step": 6556 + }, + { + "epoch": 2.75, + "learning_rate": 1.5281608310106478e-05, + "loss": 0.2588, + "step": 6557 + }, + { + "epoch": 2.75, + "learning_rate": 1.5280168587325462e-05, + "loss": 0.2612, + "step": 6558 + }, + { + "epoch": 2.75, + "learning_rate": 1.5278728712770683e-05, + "loss": 0.2553, + "step": 6559 + }, + { + "epoch": 2.75, + "learning_rate": 1.5277288686483528e-05, + "loss": 0.2601, + "step": 6560 + }, + { + "epoch": 2.75, + "learning_rate": 1.5275848508505396e-05, + "loss": 0.2463, + "step": 6561 + }, + { + "epoch": 2.75, + "learning_rate": 1.527440817887768e-05, + "loss": 0.2555, + "step": 6562 + }, + { + "epoch": 2.75, + "learning_rate": 1.5272967697641784e-05, + "loss": 0.2678, + "step": 6563 + }, + { + "epoch": 2.75, + "learning_rate": 1.5271527064839107e-05, + "loss": 0.265, + "step": 6564 + }, + { + "epoch": 2.75, + "learning_rate": 1.5270086280511064e-05, + "loss": 0.2693, + "step": 6565 + }, + { + "epoch": 2.75, + "learning_rate": 1.5268645344699066e-05, + "loss": 0.2654, + "step": 6566 + }, + { + "epoch": 2.75, + "learning_rate": 1.5267204257444534e-05, + "loss": 0.2538, + "step": 6567 + }, + { + "epoch": 2.75, + "learning_rate": 1.526576301878889e-05, + "loss": 0.2554, + "step": 6568 + }, + { + "epoch": 2.75, + "learning_rate": 1.526432162877356e-05, + "loss": 0.2549, + "step": 6569 + }, + { + "epoch": 2.75, + "learning_rate": 1.5262880087439975e-05, + "loss": 0.2483, + "step": 6570 + }, + { + "epoch": 2.75, + "learning_rate": 1.526143839482957e-05, + "loss": 0.2568, + "step": 6571 + }, + { + "epoch": 2.75, + "learning_rate": 1.5259996550983797e-05, + "loss": 0.259, + "step": 6572 + }, + { + "epoch": 2.75, + "learning_rate": 1.5258554555944085e-05, + "loss": 0.2885, + "step": 6573 + }, + { + "epoch": 2.75, + "learning_rate": 1.5257112409751886e-05, + "loss": 0.2732, + "step": 6574 + }, + { + "epoch": 2.75, + "learning_rate": 1.5255670112448657e-05, + "loss": 0.2344, + "step": 6575 + }, + { + "epoch": 2.75, + "learning_rate": 1.5254227664075858e-05, + "loss": 0.2616, + "step": 6576 + }, + { + "epoch": 2.75, + "learning_rate": 1.5252785064674945e-05, + "loss": 0.27, + "step": 6577 + }, + { + "epoch": 2.75, + "learning_rate": 1.5251342314287385e-05, + "loss": 0.2484, + "step": 6578 + }, + { + "epoch": 2.75, + "learning_rate": 1.5249899412954652e-05, + "loss": 0.2413, + "step": 6579 + }, + { + "epoch": 2.75, + "learning_rate": 1.5248456360718218e-05, + "loss": 0.2837, + "step": 6580 + }, + { + "epoch": 2.76, + "learning_rate": 1.524701315761956e-05, + "loss": 0.2716, + "step": 6581 + }, + { + "epoch": 2.76, + "learning_rate": 1.5245569803700174e-05, + "loss": 0.2518, + "step": 6582 + }, + { + "epoch": 2.76, + "learning_rate": 1.5244126299001534e-05, + "loss": 0.249, + "step": 6583 + }, + { + "epoch": 2.76, + "learning_rate": 1.5242682643565135e-05, + "loss": 0.2523, + "step": 6584 + }, + { + "epoch": 2.76, + "learning_rate": 1.5241238837432477e-05, + "loss": 0.2322, + "step": 6585 + }, + { + "epoch": 2.76, + "learning_rate": 1.523979488064506e-05, + "loss": 0.2686, + "step": 6586 + }, + { + "epoch": 2.76, + "learning_rate": 1.523835077324439e-05, + "loss": 0.2603, + "step": 6587 + }, + { + "epoch": 2.76, + "learning_rate": 1.5236906515271974e-05, + "loss": 0.2674, + "step": 6588 + }, + { + "epoch": 2.76, + "learning_rate": 1.5235462106769327e-05, + "loss": 0.2468, + "step": 6589 + }, + { + "epoch": 2.76, + "learning_rate": 1.5234017547777969e-05, + "loss": 0.2629, + "step": 6590 + }, + { + "epoch": 2.76, + "learning_rate": 1.5232572838339418e-05, + "loss": 0.2581, + "step": 6591 + }, + { + "epoch": 2.76, + "learning_rate": 1.5231127978495208e-05, + "loss": 0.2561, + "step": 6592 + }, + { + "epoch": 2.76, + "learning_rate": 1.5229682968286862e-05, + "loss": 0.2629, + "step": 6593 + }, + { + "epoch": 2.76, + "learning_rate": 1.5228237807755925e-05, + "loss": 0.254, + "step": 6594 + }, + { + "epoch": 2.76, + "learning_rate": 1.522679249694393e-05, + "loss": 0.2554, + "step": 6595 + }, + { + "epoch": 2.76, + "learning_rate": 1.5225347035892423e-05, + "loss": 0.2422, + "step": 6596 + }, + { + "epoch": 2.76, + "learning_rate": 1.522390142464295e-05, + "loss": 0.2471, + "step": 6597 + }, + { + "epoch": 2.76, + "learning_rate": 1.5222455663237068e-05, + "loss": 0.2608, + "step": 6598 + }, + { + "epoch": 2.76, + "learning_rate": 1.5221009751716333e-05, + "loss": 0.2835, + "step": 6599 + }, + { + "epoch": 2.76, + "learning_rate": 1.5219563690122306e-05, + "loss": 0.2757, + "step": 6600 + }, + { + "epoch": 2.76, + "learning_rate": 1.5218117478496548e-05, + "loss": 0.2538, + "step": 6601 + }, + { + "epoch": 2.76, + "learning_rate": 1.5216671116880637e-05, + "loss": 0.274, + "step": 6602 + }, + { + "epoch": 2.76, + "learning_rate": 1.5215224605316143e-05, + "loss": 0.2488, + "step": 6603 + }, + { + "epoch": 2.76, + "learning_rate": 1.5213777943844648e-05, + "loss": 0.2378, + "step": 6604 + }, + { + "epoch": 2.77, + "learning_rate": 1.521233113250773e-05, + "loss": 0.2754, + "step": 6605 + }, + { + "epoch": 2.77, + "learning_rate": 1.5210884171346983e-05, + "loss": 0.263, + "step": 6606 + }, + { + "epoch": 2.77, + "learning_rate": 1.5209437060403988e-05, + "loss": 0.2472, + "step": 6607 + }, + { + "epoch": 2.77, + "learning_rate": 1.5207989799720352e-05, + "loss": 0.2798, + "step": 6608 + }, + { + "epoch": 2.77, + "learning_rate": 1.520654238933767e-05, + "loss": 0.2764, + "step": 6609 + }, + { + "epoch": 2.77, + "learning_rate": 1.5205094829297546e-05, + "loss": 0.2551, + "step": 6610 + }, + { + "epoch": 2.77, + "learning_rate": 1.5203647119641591e-05, + "loss": 0.2434, + "step": 6611 + }, + { + "epoch": 2.77, + "learning_rate": 1.5202199260411417e-05, + "loss": 0.2527, + "step": 6612 + }, + { + "epoch": 2.77, + "learning_rate": 1.5200751251648641e-05, + "loss": 0.2886, + "step": 6613 + }, + { + "epoch": 2.77, + "learning_rate": 1.5199303093394883e-05, + "loss": 0.2617, + "step": 6614 + }, + { + "epoch": 2.77, + "learning_rate": 1.5197854785691775e-05, + "loss": 0.2527, + "step": 6615 + }, + { + "epoch": 2.77, + "learning_rate": 1.5196406328580943e-05, + "loss": 0.273, + "step": 6616 + }, + { + "epoch": 2.77, + "learning_rate": 1.5194957722104024e-05, + "loss": 0.2436, + "step": 6617 + }, + { + "epoch": 2.77, + "learning_rate": 1.5193508966302655e-05, + "loss": 0.2875, + "step": 6618 + }, + { + "epoch": 2.77, + "learning_rate": 1.5192060061218477e-05, + "loss": 0.2511, + "step": 6619 + }, + { + "epoch": 2.77, + "learning_rate": 1.519061100689314e-05, + "loss": 0.2614, + "step": 6620 + }, + { + "epoch": 2.77, + "learning_rate": 1.5189161803368297e-05, + "loss": 0.2674, + "step": 6621 + }, + { + "epoch": 2.77, + "learning_rate": 1.51877124506856e-05, + "loss": 0.2546, + "step": 6622 + }, + { + "epoch": 2.77, + "learning_rate": 1.5186262948886716e-05, + "loss": 0.2376, + "step": 6623 + }, + { + "epoch": 2.77, + "learning_rate": 1.5184813298013304e-05, + "loss": 0.243, + "step": 6624 + }, + { + "epoch": 2.77, + "learning_rate": 1.5183363498107033e-05, + "loss": 0.2854, + "step": 6625 + }, + { + "epoch": 2.77, + "learning_rate": 1.5181913549209582e-05, + "loss": 0.2663, + "step": 6626 + }, + { + "epoch": 2.77, + "learning_rate": 1.5180463451362621e-05, + "loss": 0.2811, + "step": 6627 + }, + { + "epoch": 2.77, + "learning_rate": 1.5179013204607838e-05, + "loss": 0.2553, + "step": 6628 + }, + { + "epoch": 2.78, + "learning_rate": 1.5177562808986914e-05, + "loss": 0.2549, + "step": 6629 + }, + { + "epoch": 2.78, + "learning_rate": 1.5176112264541542e-05, + "loss": 0.249, + "step": 6630 + }, + { + "epoch": 2.78, + "learning_rate": 1.5174661571313417e-05, + "loss": 0.2608, + "step": 6631 + }, + { + "epoch": 2.78, + "learning_rate": 1.5173210729344234e-05, + "loss": 0.2462, + "step": 6632 + }, + { + "epoch": 2.78, + "learning_rate": 1.5171759738675702e-05, + "loss": 0.2513, + "step": 6633 + }, + { + "epoch": 2.78, + "learning_rate": 1.5170308599349524e-05, + "loss": 0.2768, + "step": 6634 + }, + { + "epoch": 2.78, + "learning_rate": 1.5168857311407411e-05, + "loss": 0.2487, + "step": 6635 + }, + { + "epoch": 2.78, + "learning_rate": 1.516740587489109e-05, + "loss": 0.2677, + "step": 6636 + }, + { + "epoch": 2.78, + "learning_rate": 1.5165954289842267e-05, + "loss": 0.2875, + "step": 6637 + }, + { + "epoch": 2.78, + "learning_rate": 1.516450255630267e-05, + "loss": 0.2793, + "step": 6638 + }, + { + "epoch": 2.78, + "learning_rate": 1.5163050674314033e-05, + "loss": 0.2797, + "step": 6639 + }, + { + "epoch": 2.78, + "learning_rate": 1.5161598643918086e-05, + "loss": 0.2556, + "step": 6640 + }, + { + "epoch": 2.78, + "learning_rate": 1.5160146465156564e-05, + "loss": 0.2648, + "step": 6641 + }, + { + "epoch": 2.78, + "learning_rate": 1.5158694138071213e-05, + "loss": 0.2395, + "step": 6642 + }, + { + "epoch": 2.78, + "learning_rate": 1.5157241662703777e-05, + "loss": 0.2652, + "step": 6643 + }, + { + "epoch": 2.78, + "learning_rate": 1.5155789039096005e-05, + "loss": 0.2498, + "step": 6644 + }, + { + "epoch": 2.78, + "learning_rate": 1.515433626728965e-05, + "loss": 0.2564, + "step": 6645 + }, + { + "epoch": 2.78, + "learning_rate": 1.5152883347326475e-05, + "loss": 0.2747, + "step": 6646 + }, + { + "epoch": 2.78, + "learning_rate": 1.5151430279248242e-05, + "loss": 0.2571, + "step": 6647 + }, + { + "epoch": 2.78, + "learning_rate": 1.5149977063096715e-05, + "loss": 0.2651, + "step": 6648 + }, + { + "epoch": 2.78, + "learning_rate": 1.5148523698913668e-05, + "loss": 0.2875, + "step": 6649 + }, + { + "epoch": 2.78, + "learning_rate": 1.5147070186740877e-05, + "loss": 0.2477, + "step": 6650 + }, + { + "epoch": 2.78, + "learning_rate": 1.514561652662012e-05, + "loss": 0.2882, + "step": 6651 + }, + { + "epoch": 2.79, + "learning_rate": 1.5144162718593183e-05, + "loss": 0.2645, + "step": 6652 + }, + { + "epoch": 2.79, + "learning_rate": 1.5142708762701852e-05, + "loss": 0.279, + "step": 6653 + }, + { + "epoch": 2.79, + "learning_rate": 1.5141254658987922e-05, + "loss": 0.2428, + "step": 6654 + }, + { + "epoch": 2.79, + "learning_rate": 1.5139800407493189e-05, + "loss": 0.2613, + "step": 6655 + }, + { + "epoch": 2.79, + "learning_rate": 1.5138346008259452e-05, + "loss": 0.2361, + "step": 6656 + }, + { + "epoch": 2.79, + "learning_rate": 1.5136891461328525e-05, + "loss": 0.2499, + "step": 6657 + }, + { + "epoch": 2.79, + "learning_rate": 1.5135436766742207e-05, + "loss": 0.2513, + "step": 6658 + }, + { + "epoch": 2.79, + "learning_rate": 1.5133981924542314e-05, + "loss": 0.2727, + "step": 6659 + }, + { + "epoch": 2.79, + "learning_rate": 1.513252693477067e-05, + "loss": 0.2188, + "step": 6660 + }, + { + "epoch": 2.79, + "learning_rate": 1.5131071797469093e-05, + "loss": 0.2486, + "step": 6661 + }, + { + "epoch": 2.79, + "learning_rate": 1.512961651267941e-05, + "loss": 0.2672, + "step": 6662 + }, + { + "epoch": 2.79, + "learning_rate": 1.5128161080443451e-05, + "loss": 0.2455, + "step": 6663 + }, + { + "epoch": 2.79, + "learning_rate": 1.5126705500803055e-05, + "loss": 0.2667, + "step": 6664 + }, + { + "epoch": 2.79, + "learning_rate": 1.5125249773800057e-05, + "loss": 0.2744, + "step": 6665 + }, + { + "epoch": 2.79, + "learning_rate": 1.5123793899476303e-05, + "loss": 0.2415, + "step": 6666 + }, + { + "epoch": 2.79, + "learning_rate": 1.5122337877873641e-05, + "loss": 0.2695, + "step": 6667 + }, + { + "epoch": 2.79, + "learning_rate": 1.5120881709033919e-05, + "loss": 0.2449, + "step": 6668 + }, + { + "epoch": 2.79, + "learning_rate": 1.5119425392999e-05, + "loss": 0.2488, + "step": 6669 + }, + { + "epoch": 2.79, + "learning_rate": 1.511796892981074e-05, + "loss": 0.2597, + "step": 6670 + }, + { + "epoch": 2.79, + "learning_rate": 1.5116512319511003e-05, + "loss": 0.253, + "step": 6671 + }, + { + "epoch": 2.79, + "learning_rate": 1.5115055562141661e-05, + "loss": 0.2759, + "step": 6672 + }, + { + "epoch": 2.79, + "learning_rate": 1.5113598657744585e-05, + "loss": 0.3121, + "step": 6673 + }, + { + "epoch": 2.79, + "learning_rate": 1.5112141606361657e-05, + "loss": 0.2661, + "step": 6674 + }, + { + "epoch": 2.79, + "learning_rate": 1.511068440803475e-05, + "loss": 0.2716, + "step": 6675 + }, + { + "epoch": 2.8, + "learning_rate": 1.5109227062805758e-05, + "loss": 0.2753, + "step": 6676 + }, + { + "epoch": 2.8, + "learning_rate": 1.5107769570716566e-05, + "loss": 0.2742, + "step": 6677 + }, + { + "epoch": 2.8, + "learning_rate": 1.510631193180907e-05, + "loss": 0.2392, + "step": 6678 + }, + { + "epoch": 2.8, + "learning_rate": 1.510485414612517e-05, + "loss": 0.2766, + "step": 6679 + }, + { + "epoch": 2.8, + "learning_rate": 1.510339621370677e-05, + "loss": 0.2514, + "step": 6680 + }, + { + "epoch": 2.8, + "learning_rate": 1.510193813459577e-05, + "loss": 0.2646, + "step": 6681 + }, + { + "epoch": 2.8, + "learning_rate": 1.5100479908834087e-05, + "loss": 0.2824, + "step": 6682 + }, + { + "epoch": 2.8, + "learning_rate": 1.5099021536463636e-05, + "loss": 0.2641, + "step": 6683 + }, + { + "epoch": 2.8, + "learning_rate": 1.5097563017526336e-05, + "loss": 0.2753, + "step": 6684 + }, + { + "epoch": 2.8, + "learning_rate": 1.5096104352064112e-05, + "loss": 0.2636, + "step": 6685 + }, + { + "epoch": 2.8, + "learning_rate": 1.5094645540118888e-05, + "loss": 0.2799, + "step": 6686 + }, + { + "epoch": 2.8, + "learning_rate": 1.5093186581732599e-05, + "loss": 0.2483, + "step": 6687 + }, + { + "epoch": 2.8, + "learning_rate": 1.5091727476947182e-05, + "loss": 0.2439, + "step": 6688 + }, + { + "epoch": 2.8, + "learning_rate": 1.5090268225804577e-05, + "loss": 0.25, + "step": 6689 + }, + { + "epoch": 2.8, + "learning_rate": 1.5088808828346727e-05, + "loss": 0.2747, + "step": 6690 + }, + { + "epoch": 2.8, + "learning_rate": 1.5087349284615584e-05, + "loss": 0.255, + "step": 6691 + }, + { + "epoch": 2.8, + "learning_rate": 1.50858895946531e-05, + "loss": 0.2375, + "step": 6692 + }, + { + "epoch": 2.8, + "learning_rate": 1.5084429758501235e-05, + "loss": 0.2505, + "step": 6693 + }, + { + "epoch": 2.8, + "learning_rate": 1.5082969776201948e-05, + "loss": 0.2722, + "step": 6694 + }, + { + "epoch": 2.8, + "learning_rate": 1.5081509647797204e-05, + "loss": 0.2526, + "step": 6695 + }, + { + "epoch": 2.8, + "learning_rate": 1.5080049373328976e-05, + "loss": 0.263, + "step": 6696 + }, + { + "epoch": 2.8, + "learning_rate": 1.5078588952839235e-05, + "loss": 0.2623, + "step": 6697 + }, + { + "epoch": 2.8, + "learning_rate": 1.507712838636996e-05, + "loss": 0.245, + "step": 6698 + }, + { + "epoch": 2.8, + "learning_rate": 1.5075667673963139e-05, + "loss": 0.2681, + "step": 6699 + }, + { + "epoch": 2.81, + "learning_rate": 1.507420681566075e-05, + "loss": 0.2744, + "step": 6700 + }, + { + "epoch": 2.81, + "learning_rate": 1.5072745811504793e-05, + "loss": 0.2359, + "step": 6701 + }, + { + "epoch": 2.81, + "learning_rate": 1.5071284661537257e-05, + "loss": 0.2707, + "step": 6702 + }, + { + "epoch": 2.81, + "learning_rate": 1.5069823365800146e-05, + "loss": 0.2543, + "step": 6703 + }, + { + "epoch": 2.81, + "learning_rate": 1.5068361924335457e-05, + "loss": 0.2385, + "step": 6704 + }, + { + "epoch": 2.81, + "learning_rate": 1.5066900337185203e-05, + "loss": 0.2427, + "step": 6705 + }, + { + "epoch": 2.81, + "learning_rate": 1.5065438604391398e-05, + "loss": 0.2722, + "step": 6706 + }, + { + "epoch": 2.81, + "learning_rate": 1.5063976725996052e-05, + "loss": 0.2621, + "step": 6707 + }, + { + "epoch": 2.81, + "learning_rate": 1.5062514702041193e-05, + "loss": 0.253, + "step": 6708 + }, + { + "epoch": 2.81, + "learning_rate": 1.5061052532568838e-05, + "loss": 0.2611, + "step": 6709 + }, + { + "epoch": 2.81, + "learning_rate": 1.5059590217621018e-05, + "loss": 0.2644, + "step": 6710 + }, + { + "epoch": 2.81, + "learning_rate": 1.5058127757239772e-05, + "loss": 0.2422, + "step": 6711 + }, + { + "epoch": 2.81, + "learning_rate": 1.5056665151467126e-05, + "loss": 0.2587, + "step": 6712 + }, + { + "epoch": 2.81, + "learning_rate": 1.505520240034513e-05, + "loss": 0.2658, + "step": 6713 + }, + { + "epoch": 2.81, + "learning_rate": 1.505373950391583e-05, + "loss": 0.248, + "step": 6714 + }, + { + "epoch": 2.81, + "learning_rate": 1.5052276462221267e-05, + "loss": 0.3038, + "step": 6715 + }, + { + "epoch": 2.81, + "learning_rate": 1.5050813275303501e-05, + "loss": 0.2474, + "step": 6716 + }, + { + "epoch": 2.81, + "learning_rate": 1.5049349943204595e-05, + "loss": 0.2528, + "step": 6717 + }, + { + "epoch": 2.81, + "learning_rate": 1.50478864659666e-05, + "loss": 0.2764, + "step": 6718 + }, + { + "epoch": 2.81, + "learning_rate": 1.5046422843631594e-05, + "loss": 0.2536, + "step": 6719 + }, + { + "epoch": 2.81, + "learning_rate": 1.5044959076241634e-05, + "loss": 0.246, + "step": 6720 + }, + { + "epoch": 2.81, + "learning_rate": 1.5043495163838809e-05, + "loss": 0.2581, + "step": 6721 + }, + { + "epoch": 2.81, + "learning_rate": 1.5042031106465186e-05, + "loss": 0.2556, + "step": 6722 + }, + { + "epoch": 2.81, + "learning_rate": 1.5040566904162855e-05, + "loss": 0.2449, + "step": 6723 + }, + { + "epoch": 2.82, + "learning_rate": 1.5039102556973899e-05, + "loss": 0.2626, + "step": 6724 + }, + { + "epoch": 2.82, + "learning_rate": 1.5037638064940416e-05, + "loss": 0.2653, + "step": 6725 + }, + { + "epoch": 2.82, + "learning_rate": 1.5036173428104493e-05, + "loss": 0.256, + "step": 6726 + }, + { + "epoch": 2.82, + "learning_rate": 1.5034708646508236e-05, + "loss": 0.2708, + "step": 6727 + }, + { + "epoch": 2.82, + "learning_rate": 1.5033243720193746e-05, + "loss": 0.2576, + "step": 6728 + }, + { + "epoch": 2.82, + "learning_rate": 1.5031778649203133e-05, + "loss": 0.2685, + "step": 6729 + }, + { + "epoch": 2.82, + "learning_rate": 1.5030313433578508e-05, + "loss": 0.2787, + "step": 6730 + }, + { + "epoch": 2.82, + "learning_rate": 1.5028848073361987e-05, + "loss": 0.2648, + "step": 6731 + }, + { + "epoch": 2.82, + "learning_rate": 1.5027382568595688e-05, + "loss": 0.2672, + "step": 6732 + }, + { + "epoch": 2.82, + "learning_rate": 1.5025916919321741e-05, + "loss": 0.2528, + "step": 6733 + }, + { + "epoch": 2.82, + "learning_rate": 1.5024451125582271e-05, + "loss": 0.2618, + "step": 6734 + }, + { + "epoch": 2.82, + "learning_rate": 1.5022985187419411e-05, + "loss": 0.2609, + "step": 6735 + }, + { + "epoch": 2.82, + "learning_rate": 1.50215191048753e-05, + "loss": 0.254, + "step": 6736 + }, + { + "epoch": 2.82, + "learning_rate": 1.5020052877992077e-05, + "loss": 0.2502, + "step": 6737 + }, + { + "epoch": 2.82, + "learning_rate": 1.5018586506811893e-05, + "loss": 0.24, + "step": 6738 + }, + { + "epoch": 2.82, + "learning_rate": 1.501711999137689e-05, + "loss": 0.2637, + "step": 6739 + }, + { + "epoch": 2.82, + "learning_rate": 1.5015653331729225e-05, + "loss": 0.2965, + "step": 6740 + }, + { + "epoch": 2.82, + "learning_rate": 1.5014186527911056e-05, + "loss": 0.2677, + "step": 6741 + }, + { + "epoch": 2.82, + "learning_rate": 1.5012719579964548e-05, + "loss": 0.2725, + "step": 6742 + }, + { + "epoch": 2.82, + "learning_rate": 1.501125248793186e-05, + "loss": 0.2443, + "step": 6743 + }, + { + "epoch": 2.82, + "learning_rate": 1.5009785251855164e-05, + "loss": 0.2601, + "step": 6744 + }, + { + "epoch": 2.82, + "learning_rate": 1.5008317871776643e-05, + "loss": 0.241, + "step": 6745 + }, + { + "epoch": 2.82, + "learning_rate": 1.5006850347738463e-05, + "loss": 0.2925, + "step": 6746 + }, + { + "epoch": 2.82, + "learning_rate": 1.5005382679782817e-05, + "loss": 0.26, + "step": 6747 + }, + { + "epoch": 2.83, + "learning_rate": 1.5003914867951888e-05, + "loss": 0.2512, + "step": 6748 + }, + { + "epoch": 2.83, + "learning_rate": 1.5002446912287862e-05, + "loss": 0.2741, + "step": 6749 + }, + { + "epoch": 2.83, + "learning_rate": 1.5000978812832942e-05, + "loss": 0.2485, + "step": 6750 + }, + { + "epoch": 2.83, + "learning_rate": 1.4999510569629323e-05, + "loss": 0.2567, + "step": 6751 + }, + { + "epoch": 2.83, + "learning_rate": 1.4998042182719207e-05, + "loss": 0.2542, + "step": 6752 + }, + { + "epoch": 2.83, + "learning_rate": 1.4996573652144807e-05, + "loss": 0.2555, + "step": 6753 + }, + { + "epoch": 2.83, + "learning_rate": 1.4995104977948331e-05, + "loss": 0.2351, + "step": 6754 + }, + { + "epoch": 2.83, + "learning_rate": 1.4993636160171994e-05, + "loss": 0.2558, + "step": 6755 + }, + { + "epoch": 2.83, + "learning_rate": 1.4992167198858016e-05, + "loss": 0.2562, + "step": 6756 + }, + { + "epoch": 2.83, + "learning_rate": 1.4990698094048624e-05, + "loss": 0.2462, + "step": 6757 + }, + { + "epoch": 2.83, + "learning_rate": 1.498922884578604e-05, + "loss": 0.2662, + "step": 6758 + }, + { + "epoch": 2.83, + "learning_rate": 1.4987759454112505e-05, + "loss": 0.2444, + "step": 6759 + }, + { + "epoch": 2.83, + "learning_rate": 1.4986289919070247e-05, + "loss": 0.2565, + "step": 6760 + }, + { + "epoch": 2.83, + "learning_rate": 1.4984820240701513e-05, + "loss": 0.2805, + "step": 6761 + }, + { + "epoch": 2.83, + "learning_rate": 1.4983350419048544e-05, + "loss": 0.254, + "step": 6762 + }, + { + "epoch": 2.83, + "learning_rate": 1.4981880454153586e-05, + "loss": 0.2438, + "step": 6763 + }, + { + "epoch": 2.83, + "learning_rate": 1.49804103460589e-05, + "loss": 0.2595, + "step": 6764 + }, + { + "epoch": 2.83, + "learning_rate": 1.4978940094806733e-05, + "loss": 0.254, + "step": 6765 + }, + { + "epoch": 2.83, + "learning_rate": 1.4977469700439358e-05, + "loss": 0.2385, + "step": 6766 + }, + { + "epoch": 2.83, + "learning_rate": 1.4975999162999027e-05, + "loss": 0.2545, + "step": 6767 + }, + { + "epoch": 2.83, + "learning_rate": 1.4974528482528018e-05, + "loss": 0.2445, + "step": 6768 + }, + { + "epoch": 2.83, + "learning_rate": 1.4973057659068601e-05, + "loss": 0.2655, + "step": 6769 + }, + { + "epoch": 2.83, + "learning_rate": 1.4971586692663055e-05, + "loss": 0.2913, + "step": 6770 + }, + { + "epoch": 2.83, + "learning_rate": 1.4970115583353661e-05, + "loss": 0.2701, + "step": 6771 + }, + { + "epoch": 2.84, + "learning_rate": 1.496864433118271e-05, + "loss": 0.263, + "step": 6772 + }, + { + "epoch": 2.84, + "learning_rate": 1.496717293619248e-05, + "loss": 0.2663, + "step": 6773 + }, + { + "epoch": 2.84, + "learning_rate": 1.4965701398425275e-05, + "loss": 0.2575, + "step": 6774 + }, + { + "epoch": 2.84, + "learning_rate": 1.496422971792339e-05, + "loss": 0.2518, + "step": 6775 + }, + { + "epoch": 2.84, + "learning_rate": 1.4962757894729128e-05, + "loss": 0.2548, + "step": 6776 + }, + { + "epoch": 2.84, + "learning_rate": 1.496128592888479e-05, + "loss": 0.2347, + "step": 6777 + }, + { + "epoch": 2.84, + "learning_rate": 1.4959813820432693e-05, + "loss": 0.2599, + "step": 6778 + }, + { + "epoch": 2.84, + "learning_rate": 1.4958341569415149e-05, + "loss": 0.2783, + "step": 6779 + }, + { + "epoch": 2.84, + "learning_rate": 1.4956869175874475e-05, + "loss": 0.2505, + "step": 6780 + }, + { + "epoch": 2.84, + "learning_rate": 1.4955396639852996e-05, + "loss": 0.2879, + "step": 6781 + }, + { + "epoch": 2.84, + "learning_rate": 1.4953923961393043e-05, + "loss": 0.251, + "step": 6782 + }, + { + "epoch": 2.84, + "learning_rate": 1.4952451140536934e-05, + "loss": 0.2481, + "step": 6783 + }, + { + "epoch": 2.84, + "learning_rate": 1.4950978177327016e-05, + "loss": 0.2704, + "step": 6784 + }, + { + "epoch": 2.84, + "learning_rate": 1.4949505071805624e-05, + "loss": 0.2769, + "step": 6785 + }, + { + "epoch": 2.84, + "learning_rate": 1.4948031824015098e-05, + "loss": 0.2836, + "step": 6786 + }, + { + "epoch": 2.84, + "learning_rate": 1.4946558433997792e-05, + "loss": 0.2574, + "step": 6787 + }, + { + "epoch": 2.84, + "learning_rate": 1.4945084901796048e-05, + "loss": 0.2546, + "step": 6788 + }, + { + "epoch": 2.84, + "learning_rate": 1.494361122745223e-05, + "loss": 0.2575, + "step": 6789 + }, + { + "epoch": 2.84, + "learning_rate": 1.4942137411008696e-05, + "loss": 0.2564, + "step": 6790 + }, + { + "epoch": 2.84, + "learning_rate": 1.4940663452507802e-05, + "loss": 0.2227, + "step": 6791 + }, + { + "epoch": 2.84, + "learning_rate": 1.4939189351991928e-05, + "loss": 0.2171, + "step": 6792 + }, + { + "epoch": 2.84, + "learning_rate": 1.4937715109503434e-05, + "loss": 0.2628, + "step": 6793 + }, + { + "epoch": 2.84, + "learning_rate": 1.4936240725084701e-05, + "loss": 0.256, + "step": 6794 + }, + { + "epoch": 2.84, + "learning_rate": 1.4934766198778111e-05, + "loss": 0.2534, + "step": 6795 + }, + { + "epoch": 2.85, + "learning_rate": 1.4933291530626047e-05, + "loss": 0.279, + "step": 6796 + }, + { + "epoch": 2.85, + "learning_rate": 1.493181672067089e-05, + "loss": 0.2554, + "step": 6797 + }, + { + "epoch": 2.85, + "learning_rate": 1.4930341768955043e-05, + "loss": 0.233, + "step": 6798 + }, + { + "epoch": 2.85, + "learning_rate": 1.4928866675520897e-05, + "loss": 0.2738, + "step": 6799 + }, + { + "epoch": 2.85, + "learning_rate": 1.492739144041085e-05, + "loss": 0.2688, + "step": 6800 + }, + { + "epoch": 2.85, + "learning_rate": 1.4925916063667309e-05, + "loss": 0.2855, + "step": 6801 + }, + { + "epoch": 2.85, + "learning_rate": 1.4924440545332681e-05, + "loss": 0.2513, + "step": 6802 + }, + { + "epoch": 2.85, + "learning_rate": 1.4922964885449382e-05, + "loss": 0.2448, + "step": 6803 + }, + { + "epoch": 2.85, + "learning_rate": 1.4921489084059823e-05, + "loss": 0.2518, + "step": 6804 + }, + { + "epoch": 2.85, + "learning_rate": 1.4920013141206428e-05, + "loss": 0.2839, + "step": 6805 + }, + { + "epoch": 2.85, + "learning_rate": 1.4918537056931623e-05, + "loss": 0.2599, + "step": 6806 + }, + { + "epoch": 2.85, + "learning_rate": 1.4917060831277835e-05, + "loss": 0.2528, + "step": 6807 + }, + { + "epoch": 2.85, + "learning_rate": 1.4915584464287494e-05, + "loss": 0.2628, + "step": 6808 + }, + { + "epoch": 2.85, + "learning_rate": 1.4914107956003044e-05, + "loss": 0.2673, + "step": 6809 + }, + { + "epoch": 2.85, + "learning_rate": 1.4912631306466918e-05, + "loss": 0.243, + "step": 6810 + }, + { + "epoch": 2.85, + "learning_rate": 1.4911154515721564e-05, + "loss": 0.2547, + "step": 6811 + }, + { + "epoch": 2.85, + "learning_rate": 1.4909677583809435e-05, + "loss": 0.2695, + "step": 6812 + }, + { + "epoch": 2.85, + "learning_rate": 1.4908200510772979e-05, + "loss": 0.2457, + "step": 6813 + }, + { + "epoch": 2.85, + "learning_rate": 1.4906723296654652e-05, + "loss": 0.2445, + "step": 6814 + }, + { + "epoch": 2.85, + "learning_rate": 1.490524594149692e-05, + "loss": 0.2652, + "step": 6815 + }, + { + "epoch": 2.85, + "learning_rate": 1.4903768445342248e-05, + "loss": 0.2757, + "step": 6816 + }, + { + "epoch": 2.85, + "learning_rate": 1.4902290808233103e-05, + "loss": 0.2322, + "step": 6817 + }, + { + "epoch": 2.85, + "learning_rate": 1.4900813030211956e-05, + "loss": 0.2429, + "step": 6818 + }, + { + "epoch": 2.85, + "learning_rate": 1.4899335111321289e-05, + "loss": 0.2251, + "step": 6819 + }, + { + "epoch": 2.86, + "learning_rate": 1.489785705160358e-05, + "loss": 0.2629, + "step": 6820 + }, + { + "epoch": 2.86, + "learning_rate": 1.4896378851101321e-05, + "loss": 0.2627, + "step": 6821 + }, + { + "epoch": 2.86, + "learning_rate": 1.4894900509856995e-05, + "loss": 0.2647, + "step": 6822 + }, + { + "epoch": 2.86, + "learning_rate": 1.4893422027913099e-05, + "loss": 0.2637, + "step": 6823 + }, + { + "epoch": 2.86, + "learning_rate": 1.4891943405312126e-05, + "loss": 0.25, + "step": 6824 + }, + { + "epoch": 2.86, + "learning_rate": 1.489046464209658e-05, + "loss": 0.2694, + "step": 6825 + }, + { + "epoch": 2.86, + "learning_rate": 1.488898573830897e-05, + "loss": 0.2741, + "step": 6826 + }, + { + "epoch": 2.86, + "learning_rate": 1.4887506693991803e-05, + "loss": 0.269, + "step": 6827 + }, + { + "epoch": 2.86, + "learning_rate": 1.4886027509187593e-05, + "loss": 0.2546, + "step": 6828 + }, + { + "epoch": 2.86, + "learning_rate": 1.4884548183938857e-05, + "loss": 0.2611, + "step": 6829 + }, + { + "epoch": 2.86, + "learning_rate": 1.4883068718288121e-05, + "loss": 0.2603, + "step": 6830 + }, + { + "epoch": 2.86, + "learning_rate": 1.4881589112277903e-05, + "loss": 0.2771, + "step": 6831 + }, + { + "epoch": 2.86, + "learning_rate": 1.4880109365950744e-05, + "loss": 0.2664, + "step": 6832 + }, + { + "epoch": 2.86, + "learning_rate": 1.4878629479349167e-05, + "loss": 0.2743, + "step": 6833 + }, + { + "epoch": 2.86, + "learning_rate": 1.4877149452515718e-05, + "loss": 0.2375, + "step": 6834 + }, + { + "epoch": 2.86, + "learning_rate": 1.4875669285492933e-05, + "loss": 0.2422, + "step": 6835 + }, + { + "epoch": 2.86, + "learning_rate": 1.487418897832336e-05, + "loss": 0.2433, + "step": 6836 + }, + { + "epoch": 2.86, + "learning_rate": 1.4872708531049554e-05, + "loss": 0.2557, + "step": 6837 + }, + { + "epoch": 2.86, + "learning_rate": 1.4871227943714063e-05, + "loss": 0.2563, + "step": 6838 + }, + { + "epoch": 2.86, + "learning_rate": 1.4869747216359446e-05, + "loss": 0.2676, + "step": 6839 + }, + { + "epoch": 2.86, + "learning_rate": 1.486826634902827e-05, + "loss": 0.2543, + "step": 6840 + }, + { + "epoch": 2.86, + "learning_rate": 1.4866785341763097e-05, + "loss": 0.2639, + "step": 6841 + }, + { + "epoch": 2.86, + "learning_rate": 1.4865304194606496e-05, + "loss": 0.2607, + "step": 6842 + }, + { + "epoch": 2.86, + "learning_rate": 1.4863822907601046e-05, + "loss": 0.2511, + "step": 6843 + }, + { + "epoch": 2.87, + "learning_rate": 1.4862341480789322e-05, + "loss": 0.2529, + "step": 6844 + }, + { + "epoch": 2.87, + "learning_rate": 1.4860859914213905e-05, + "loss": 0.2363, + "step": 6845 + }, + { + "epoch": 2.87, + "learning_rate": 1.4859378207917385e-05, + "loss": 0.2423, + "step": 6846 + }, + { + "epoch": 2.87, + "learning_rate": 1.485789636194235e-05, + "loss": 0.2616, + "step": 6847 + }, + { + "epoch": 2.87, + "learning_rate": 1.4856414376331394e-05, + "loss": 0.264, + "step": 6848 + }, + { + "epoch": 2.87, + "learning_rate": 1.4854932251127117e-05, + "loss": 0.2839, + "step": 6849 + }, + { + "epoch": 2.87, + "learning_rate": 1.4853449986372122e-05, + "loss": 0.253, + "step": 6850 + }, + { + "epoch": 2.87, + "learning_rate": 1.4851967582109013e-05, + "loss": 0.288, + "step": 6851 + }, + { + "epoch": 2.87, + "learning_rate": 1.48504850383804e-05, + "loss": 0.2654, + "step": 6852 + }, + { + "epoch": 2.87, + "learning_rate": 1.4849002355228901e-05, + "loss": 0.2839, + "step": 6853 + }, + { + "epoch": 2.87, + "learning_rate": 1.484751953269713e-05, + "loss": 0.2706, + "step": 6854 + }, + { + "epoch": 2.87, + "learning_rate": 1.4846036570827714e-05, + "loss": 0.2648, + "step": 6855 + }, + { + "epoch": 2.87, + "learning_rate": 1.4844553469663274e-05, + "loss": 0.2387, + "step": 6856 + }, + { + "epoch": 2.87, + "learning_rate": 1.4843070229246448e-05, + "loss": 0.2518, + "step": 6857 + }, + { + "epoch": 2.87, + "learning_rate": 1.4841586849619862e-05, + "loss": 0.2505, + "step": 6858 + }, + { + "epoch": 2.87, + "learning_rate": 1.4840103330826158e-05, + "loss": 0.2374, + "step": 6859 + }, + { + "epoch": 2.87, + "learning_rate": 1.4838619672907984e-05, + "loss": 0.2631, + "step": 6860 + }, + { + "epoch": 2.87, + "learning_rate": 1.4837135875907978e-05, + "loss": 0.2686, + "step": 6861 + }, + { + "epoch": 2.87, + "learning_rate": 1.4835651939868792e-05, + "loss": 0.2546, + "step": 6862 + }, + { + "epoch": 2.87, + "learning_rate": 1.4834167864833085e-05, + "loss": 0.2985, + "step": 6863 + }, + { + "epoch": 2.87, + "learning_rate": 1.483268365084351e-05, + "loss": 0.2512, + "step": 6864 + }, + { + "epoch": 2.87, + "learning_rate": 1.4831199297942731e-05, + "loss": 0.2642, + "step": 6865 + }, + { + "epoch": 2.87, + "learning_rate": 1.482971480617342e-05, + "loss": 0.2639, + "step": 6866 + }, + { + "epoch": 2.88, + "learning_rate": 1.482823017557824e-05, + "loss": 0.2675, + "step": 6867 + }, + { + "epoch": 2.88, + "learning_rate": 1.4826745406199865e-05, + "loss": 0.2623, + "step": 6868 + }, + { + "epoch": 2.88, + "learning_rate": 1.482526049808098e-05, + "loss": 0.2545, + "step": 6869 + }, + { + "epoch": 2.88, + "learning_rate": 1.4823775451264262e-05, + "loss": 0.2215, + "step": 6870 + }, + { + "epoch": 2.88, + "learning_rate": 1.4822290265792403e-05, + "loss": 0.229, + "step": 6871 + }, + { + "epoch": 2.88, + "learning_rate": 1.4820804941708086e-05, + "loss": 0.2672, + "step": 6872 + }, + { + "epoch": 2.88, + "learning_rate": 1.4819319479054009e-05, + "loss": 0.2455, + "step": 6873 + }, + { + "epoch": 2.88, + "learning_rate": 1.481783387787287e-05, + "loss": 0.2786, + "step": 6874 + }, + { + "epoch": 2.88, + "learning_rate": 1.4816348138207373e-05, + "loss": 0.2453, + "step": 6875 + }, + { + "epoch": 2.88, + "learning_rate": 1.4814862260100222e-05, + "loss": 0.2477, + "step": 6876 + }, + { + "epoch": 2.88, + "learning_rate": 1.4813376243594128e-05, + "loss": 0.275, + "step": 6877 + }, + { + "epoch": 2.88, + "learning_rate": 1.4811890088731808e-05, + "loss": 0.2415, + "step": 6878 + }, + { + "epoch": 2.88, + "learning_rate": 1.4810403795555975e-05, + "loss": 0.2582, + "step": 6879 + }, + { + "epoch": 2.88, + "learning_rate": 1.4808917364109352e-05, + "loss": 0.2525, + "step": 6880 + }, + { + "epoch": 2.88, + "learning_rate": 1.480743079443467e-05, + "loss": 0.2591, + "step": 6881 + }, + { + "epoch": 2.88, + "learning_rate": 1.4805944086574653e-05, + "loss": 0.2261, + "step": 6882 + }, + { + "epoch": 2.88, + "learning_rate": 1.4804457240572037e-05, + "loss": 0.2787, + "step": 6883 + }, + { + "epoch": 2.88, + "learning_rate": 1.4802970256469564e-05, + "loss": 0.2833, + "step": 6884 + }, + { + "epoch": 2.88, + "learning_rate": 1.4801483134309971e-05, + "loss": 0.2346, + "step": 6885 + }, + { + "epoch": 2.88, + "learning_rate": 1.4799995874136006e-05, + "loss": 0.2583, + "step": 6886 + }, + { + "epoch": 2.88, + "learning_rate": 1.4798508475990421e-05, + "loss": 0.2704, + "step": 6887 + }, + { + "epoch": 2.88, + "learning_rate": 1.4797020939915967e-05, + "loss": 0.2709, + "step": 6888 + }, + { + "epoch": 2.88, + "learning_rate": 1.4795533265955404e-05, + "loss": 0.267, + "step": 6889 + }, + { + "epoch": 2.88, + "learning_rate": 1.4794045454151491e-05, + "loss": 0.2466, + "step": 6890 + }, + { + "epoch": 2.89, + "learning_rate": 1.4792557504546997e-05, + "loss": 0.2482, + "step": 6891 + }, + { + "epoch": 2.89, + "learning_rate": 1.4791069417184689e-05, + "loss": 0.2537, + "step": 6892 + }, + { + "epoch": 2.89, + "learning_rate": 1.4789581192107343e-05, + "loss": 0.2675, + "step": 6893 + }, + { + "epoch": 2.89, + "learning_rate": 1.4788092829357736e-05, + "loss": 0.2571, + "step": 6894 + }, + { + "epoch": 2.89, + "learning_rate": 1.4786604328978648e-05, + "loss": 0.267, + "step": 6895 + }, + { + "epoch": 2.89, + "learning_rate": 1.4785115691012866e-05, + "loss": 0.2748, + "step": 6896 + }, + { + "epoch": 2.89, + "learning_rate": 1.4783626915503182e-05, + "loss": 0.2911, + "step": 6897 + }, + { + "epoch": 2.89, + "learning_rate": 1.4782138002492385e-05, + "loss": 0.2753, + "step": 6898 + }, + { + "epoch": 2.89, + "learning_rate": 1.4780648952023276e-05, + "loss": 0.245, + "step": 6899 + }, + { + "epoch": 2.89, + "learning_rate": 1.4779159764138656e-05, + "loss": 0.2457, + "step": 6900 + }, + { + "epoch": 2.89, + "learning_rate": 1.4777670438881328e-05, + "loss": 0.2733, + "step": 6901 + }, + { + "epoch": 2.89, + "learning_rate": 1.4776180976294105e-05, + "loss": 0.2523, + "step": 6902 + }, + { + "epoch": 2.89, + "learning_rate": 1.4774691376419795e-05, + "loss": 0.2498, + "step": 6903 + }, + { + "epoch": 2.89, + "learning_rate": 1.4773201639301223e-05, + "loss": 0.2705, + "step": 6904 + }, + { + "epoch": 2.89, + "learning_rate": 1.47717117649812e-05, + "loss": 0.2358, + "step": 6905 + }, + { + "epoch": 2.89, + "learning_rate": 1.4770221753502558e-05, + "loss": 0.2867, + "step": 6906 + }, + { + "epoch": 2.89, + "learning_rate": 1.4768731604908125e-05, + "loss": 0.2619, + "step": 6907 + }, + { + "epoch": 2.89, + "learning_rate": 1.4767241319240735e-05, + "loss": 0.2755, + "step": 6908 + }, + { + "epoch": 2.89, + "learning_rate": 1.4765750896543222e-05, + "loss": 0.2694, + "step": 6909 + }, + { + "epoch": 2.89, + "learning_rate": 1.476426033685843e-05, + "loss": 0.294, + "step": 6910 + }, + { + "epoch": 2.89, + "learning_rate": 1.4762769640229202e-05, + "loss": 0.235, + "step": 6911 + }, + { + "epoch": 2.89, + "learning_rate": 1.4761278806698389e-05, + "loss": 0.2529, + "step": 6912 + }, + { + "epoch": 2.89, + "learning_rate": 1.475978783630884e-05, + "loss": 0.2626, + "step": 6913 + }, + { + "epoch": 2.89, + "learning_rate": 1.4758296729103413e-05, + "loss": 0.2572, + "step": 6914 + }, + { + "epoch": 2.9, + "learning_rate": 1.475680548512497e-05, + "loss": 0.2899, + "step": 6915 + }, + { + "epoch": 2.9, + "learning_rate": 1.4755314104416371e-05, + "loss": 0.2459, + "step": 6916 + }, + { + "epoch": 2.9, + "learning_rate": 1.4753822587020491e-05, + "loss": 0.255, + "step": 6917 + }, + { + "epoch": 2.9, + "learning_rate": 1.47523309329802e-05, + "loss": 0.2422, + "step": 6918 + }, + { + "epoch": 2.9, + "learning_rate": 1.4750839142338372e-05, + "loss": 0.2726, + "step": 6919 + }, + { + "epoch": 2.9, + "learning_rate": 1.4749347215137888e-05, + "loss": 0.2683, + "step": 6920 + }, + { + "epoch": 2.9, + "learning_rate": 1.4747855151421636e-05, + "loss": 0.2625, + "step": 6921 + }, + { + "epoch": 2.9, + "learning_rate": 1.4746362951232499e-05, + "loss": 0.253, + "step": 6922 + }, + { + "epoch": 2.9, + "learning_rate": 1.4744870614613369e-05, + "loss": 0.2732, + "step": 6923 + }, + { + "epoch": 2.9, + "learning_rate": 1.4743378141607146e-05, + "loss": 0.2718, + "step": 6924 + }, + { + "epoch": 2.9, + "learning_rate": 1.4741885532256726e-05, + "loss": 0.2532, + "step": 6925 + }, + { + "epoch": 2.9, + "learning_rate": 1.4740392786605014e-05, + "loss": 0.2521, + "step": 6926 + }, + { + "epoch": 2.9, + "learning_rate": 1.473889990469492e-05, + "loss": 0.2389, + "step": 6927 + }, + { + "epoch": 2.9, + "learning_rate": 1.4737406886569352e-05, + "loss": 0.2739, + "step": 6928 + }, + { + "epoch": 2.9, + "learning_rate": 1.4735913732271226e-05, + "loss": 0.2551, + "step": 6929 + }, + { + "epoch": 2.9, + "learning_rate": 1.4734420441843462e-05, + "loss": 0.2653, + "step": 6930 + }, + { + "epoch": 2.9, + "learning_rate": 1.4732927015328989e-05, + "loss": 0.2595, + "step": 6931 + }, + { + "epoch": 2.9, + "learning_rate": 1.4731433452770723e-05, + "loss": 0.2651, + "step": 6932 + }, + { + "epoch": 2.9, + "learning_rate": 1.4729939754211602e-05, + "loss": 0.2988, + "step": 6933 + }, + { + "epoch": 2.9, + "learning_rate": 1.4728445919694563e-05, + "loss": 0.2706, + "step": 6934 + }, + { + "epoch": 2.9, + "learning_rate": 1.472695194926254e-05, + "loss": 0.2431, + "step": 6935 + }, + { + "epoch": 2.9, + "learning_rate": 1.4725457842958477e-05, + "loss": 0.2552, + "step": 6936 + }, + { + "epoch": 2.9, + "learning_rate": 1.4723963600825322e-05, + "loss": 0.2553, + "step": 6937 + }, + { + "epoch": 2.9, + "learning_rate": 1.472246922290603e-05, + "loss": 0.2611, + "step": 6938 + }, + { + "epoch": 2.91, + "learning_rate": 1.4720974709243542e-05, + "loss": 0.2267, + "step": 6939 + }, + { + "epoch": 2.91, + "learning_rate": 1.471948005988083e-05, + "loss": 0.2719, + "step": 6940 + }, + { + "epoch": 2.91, + "learning_rate": 1.471798527486085e-05, + "loss": 0.2622, + "step": 6941 + }, + { + "epoch": 2.91, + "learning_rate": 1.4716490354226573e-05, + "loss": 0.2586, + "step": 6942 + }, + { + "epoch": 2.91, + "learning_rate": 1.4714995298020963e-05, + "loss": 0.2677, + "step": 6943 + }, + { + "epoch": 2.91, + "learning_rate": 1.4713500106287e-05, + "loss": 0.2544, + "step": 6944 + }, + { + "epoch": 2.91, + "learning_rate": 1.4712004779067659e-05, + "loss": 0.2792, + "step": 6945 + }, + { + "epoch": 2.91, + "learning_rate": 1.471050931640592e-05, + "loss": 0.243, + "step": 6946 + }, + { + "epoch": 2.91, + "learning_rate": 1.4709013718344773e-05, + "loss": 0.2542, + "step": 6947 + }, + { + "epoch": 2.91, + "learning_rate": 1.4707517984927201e-05, + "loss": 0.2478, + "step": 6948 + }, + { + "epoch": 2.91, + "learning_rate": 1.4706022116196208e-05, + "loss": 0.2705, + "step": 6949 + }, + { + "epoch": 2.91, + "learning_rate": 1.4704526112194781e-05, + "loss": 0.2619, + "step": 6950 + }, + { + "epoch": 2.91, + "learning_rate": 1.4703029972965927e-05, + "loss": 0.2677, + "step": 6951 + }, + { + "epoch": 2.91, + "learning_rate": 1.470153369855265e-05, + "loss": 0.2332, + "step": 6952 + }, + { + "epoch": 2.91, + "learning_rate": 1.470003728899796e-05, + "loss": 0.2818, + "step": 6953 + }, + { + "epoch": 2.91, + "learning_rate": 1.4698540744344864e-05, + "loss": 0.2309, + "step": 6954 + }, + { + "epoch": 2.91, + "learning_rate": 1.469704406463639e-05, + "loss": 0.2381, + "step": 6955 + }, + { + "epoch": 2.91, + "learning_rate": 1.469554724991555e-05, + "loss": 0.2561, + "step": 6956 + }, + { + "epoch": 2.91, + "learning_rate": 1.469405030022537e-05, + "loss": 0.2776, + "step": 6957 + }, + { + "epoch": 2.91, + "learning_rate": 1.4692553215608878e-05, + "loss": 0.2784, + "step": 6958 + }, + { + "epoch": 2.91, + "learning_rate": 1.4691055996109112e-05, + "loss": 0.265, + "step": 6959 + }, + { + "epoch": 2.91, + "learning_rate": 1.4689558641769101e-05, + "loss": 0.2791, + "step": 6960 + }, + { + "epoch": 2.91, + "learning_rate": 1.4688061152631889e-05, + "loss": 0.2317, + "step": 6961 + }, + { + "epoch": 2.91, + "learning_rate": 1.468656352874052e-05, + "loss": 0.2672, + "step": 6962 + }, + { + "epoch": 2.92, + "learning_rate": 1.4685065770138043e-05, + "loss": 0.2632, + "step": 6963 + }, + { + "epoch": 2.92, + "learning_rate": 1.4683567876867503e-05, + "loss": 0.272, + "step": 6964 + }, + { + "epoch": 2.92, + "learning_rate": 1.4682069848971966e-05, + "loss": 0.2292, + "step": 6965 + }, + { + "epoch": 2.92, + "learning_rate": 1.4680571686494483e-05, + "loss": 0.2913, + "step": 6966 + }, + { + "epoch": 2.92, + "learning_rate": 1.467907338947812e-05, + "loss": 0.2553, + "step": 6967 + }, + { + "epoch": 2.92, + "learning_rate": 1.4677574957965946e-05, + "loss": 0.2323, + "step": 6968 + }, + { + "epoch": 2.92, + "learning_rate": 1.4676076392001032e-05, + "loss": 0.2707, + "step": 6969 + }, + { + "epoch": 2.92, + "learning_rate": 1.4674577691626448e-05, + "loss": 0.2586, + "step": 6970 + }, + { + "epoch": 2.92, + "learning_rate": 1.4673078856885277e-05, + "loss": 0.2988, + "step": 6971 + }, + { + "epoch": 2.92, + "learning_rate": 1.4671579887820602e-05, + "loss": 0.2444, + "step": 6972 + }, + { + "epoch": 2.92, + "learning_rate": 1.467008078447551e-05, + "loss": 0.2504, + "step": 6973 + }, + { + "epoch": 2.92, + "learning_rate": 1.4668581546893086e-05, + "loss": 0.2456, + "step": 6974 + }, + { + "epoch": 2.92, + "learning_rate": 1.4667082175116431e-05, + "loss": 0.2501, + "step": 6975 + }, + { + "epoch": 2.92, + "learning_rate": 1.466558266918864e-05, + "loss": 0.2867, + "step": 6976 + }, + { + "epoch": 2.92, + "learning_rate": 1.4664083029152815e-05, + "loss": 0.2356, + "step": 6977 + }, + { + "epoch": 2.92, + "learning_rate": 1.4662583255052064e-05, + "loss": 0.2852, + "step": 6978 + }, + { + "epoch": 2.92, + "learning_rate": 1.4661083346929492e-05, + "loss": 0.2831, + "step": 6979 + }, + { + "epoch": 2.92, + "learning_rate": 1.465958330482822e-05, + "loss": 0.2642, + "step": 6980 + }, + { + "epoch": 2.92, + "learning_rate": 1.4658083128791356e-05, + "loss": 0.251, + "step": 6981 + }, + { + "epoch": 2.92, + "learning_rate": 1.4656582818862028e-05, + "loss": 0.3004, + "step": 6982 + }, + { + "epoch": 2.92, + "learning_rate": 1.465508237508336e-05, + "loss": 0.2536, + "step": 6983 + }, + { + "epoch": 2.92, + "learning_rate": 1.4653581797498477e-05, + "loss": 0.2684, + "step": 6984 + }, + { + "epoch": 2.92, + "learning_rate": 1.4652081086150516e-05, + "loss": 0.2621, + "step": 6985 + }, + { + "epoch": 2.92, + "learning_rate": 1.4650580241082613e-05, + "loss": 0.2872, + "step": 6986 + }, + { + "epoch": 2.93, + "learning_rate": 1.4649079262337909e-05, + "loss": 0.2488, + "step": 6987 + }, + { + "epoch": 2.93, + "learning_rate": 1.4647578149959545e-05, + "loss": 0.2557, + "step": 6988 + }, + { + "epoch": 2.93, + "learning_rate": 1.4646076903990671e-05, + "loss": 0.2869, + "step": 6989 + }, + { + "epoch": 2.93, + "learning_rate": 1.4644575524474441e-05, + "loss": 0.2733, + "step": 6990 + }, + { + "epoch": 2.93, + "learning_rate": 1.4643074011454007e-05, + "loss": 0.2676, + "step": 6991 + }, + { + "epoch": 2.93, + "learning_rate": 1.4641572364972532e-05, + "loss": 0.2571, + "step": 6992 + }, + { + "epoch": 2.93, + "learning_rate": 1.464007058507318e-05, + "loss": 0.2744, + "step": 6993 + }, + { + "epoch": 2.93, + "learning_rate": 1.4638568671799113e-05, + "loss": 0.2485, + "step": 6994 + }, + { + "epoch": 2.93, + "learning_rate": 1.4637066625193508e-05, + "loss": 0.2437, + "step": 6995 + }, + { + "epoch": 2.93, + "learning_rate": 1.463556444529954e-05, + "loss": 0.2428, + "step": 6996 + }, + { + "epoch": 2.93, + "learning_rate": 1.4634062132160381e-05, + "loss": 0.2526, + "step": 6997 + }, + { + "epoch": 2.93, + "learning_rate": 1.4632559685819219e-05, + "loss": 0.2696, + "step": 6998 + }, + { + "epoch": 2.93, + "learning_rate": 1.4631057106319242e-05, + "loss": 0.2625, + "step": 6999 + }, + { + "epoch": 2.93, + "learning_rate": 1.4629554393703635e-05, + "loss": 0.259, + "step": 7000 + }, + { + "epoch": 2.93, + "learning_rate": 1.4628051548015596e-05, + "loss": 0.2856, + "step": 7001 + }, + { + "epoch": 2.93, + "learning_rate": 1.4626548569298324e-05, + "loss": 0.2357, + "step": 7002 + }, + { + "epoch": 2.93, + "learning_rate": 1.462504545759502e-05, + "loss": 0.2855, + "step": 7003 + }, + { + "epoch": 2.93, + "learning_rate": 1.4623542212948887e-05, + "loss": 0.2517, + "step": 7004 + }, + { + "epoch": 2.93, + "learning_rate": 1.4622038835403135e-05, + "loss": 0.2457, + "step": 7005 + }, + { + "epoch": 2.93, + "learning_rate": 1.4620535325000978e-05, + "loss": 0.2696, + "step": 7006 + }, + { + "epoch": 2.93, + "learning_rate": 1.4619031681785633e-05, + "loss": 0.2551, + "step": 7007 + }, + { + "epoch": 2.93, + "learning_rate": 1.4617527905800321e-05, + "loss": 0.2464, + "step": 7008 + }, + { + "epoch": 2.93, + "learning_rate": 1.461602399708827e-05, + "loss": 0.2891, + "step": 7009 + }, + { + "epoch": 2.93, + "learning_rate": 1.4614519955692701e-05, + "loss": 0.2391, + "step": 7010 + }, + { + "epoch": 2.94, + "learning_rate": 1.4613015781656854e-05, + "loss": 0.2525, + "step": 7011 + }, + { + "epoch": 2.94, + "learning_rate": 1.4611511475023958e-05, + "loss": 0.2371, + "step": 7012 + }, + { + "epoch": 2.94, + "learning_rate": 1.461000703583726e-05, + "loss": 0.2491, + "step": 7013 + }, + { + "epoch": 2.94, + "learning_rate": 1.4608502464140002e-05, + "loss": 0.2961, + "step": 7014 + }, + { + "epoch": 2.94, + "learning_rate": 1.4606997759975427e-05, + "loss": 0.2514, + "step": 7015 + }, + { + "epoch": 2.94, + "learning_rate": 1.4605492923386795e-05, + "loss": 0.2443, + "step": 7016 + }, + { + "epoch": 2.94, + "learning_rate": 1.4603987954417349e-05, + "loss": 0.2637, + "step": 7017 + }, + { + "epoch": 2.94, + "learning_rate": 1.4602482853110358e-05, + "loss": 0.2675, + "step": 7018 + }, + { + "epoch": 2.94, + "learning_rate": 1.460097761950908e-05, + "loss": 0.2693, + "step": 7019 + }, + { + "epoch": 2.94, + "learning_rate": 1.4599472253656786e-05, + "loss": 0.2458, + "step": 7020 + }, + { + "epoch": 2.94, + "learning_rate": 1.459796675559674e-05, + "loss": 0.2845, + "step": 7021 + }, + { + "epoch": 2.94, + "learning_rate": 1.4596461125372221e-05, + "loss": 0.2674, + "step": 7022 + }, + { + "epoch": 2.94, + "learning_rate": 1.4594955363026509e-05, + "loss": 0.2461, + "step": 7023 + }, + { + "epoch": 2.94, + "learning_rate": 1.4593449468602878e-05, + "loss": 0.2608, + "step": 7024 + }, + { + "epoch": 2.94, + "learning_rate": 1.459194344214462e-05, + "loss": 0.25, + "step": 7025 + }, + { + "epoch": 2.94, + "learning_rate": 1.4590437283695022e-05, + "loss": 0.2608, + "step": 7026 + }, + { + "epoch": 2.94, + "learning_rate": 1.4588930993297381e-05, + "loss": 0.2597, + "step": 7027 + }, + { + "epoch": 2.94, + "learning_rate": 1.4587424570994989e-05, + "loss": 0.2485, + "step": 7028 + }, + { + "epoch": 2.94, + "learning_rate": 1.4585918016831144e-05, + "loss": 0.3006, + "step": 7029 + }, + { + "epoch": 2.94, + "learning_rate": 1.4584411330849162e-05, + "loss": 0.2837, + "step": 7030 + }, + { + "epoch": 2.94, + "learning_rate": 1.458290451309234e-05, + "loss": 0.243, + "step": 7031 + }, + { + "epoch": 2.94, + "learning_rate": 1.4581397563603994e-05, + "loss": 0.2504, + "step": 7032 + }, + { + "epoch": 2.94, + "learning_rate": 1.4579890482427446e-05, + "loss": 0.27, + "step": 7033 + }, + { + "epoch": 2.94, + "learning_rate": 1.4578383269606004e-05, + "loss": 0.228, + "step": 7034 + }, + { + "epoch": 2.95, + "learning_rate": 1.4576875925182998e-05, + "loss": 0.2386, + "step": 7035 + }, + { + "epoch": 2.95, + "learning_rate": 1.4575368449201759e-05, + "loss": 0.2873, + "step": 7036 + }, + { + "epoch": 2.95, + "learning_rate": 1.4573860841705613e-05, + "loss": 0.2577, + "step": 7037 + }, + { + "epoch": 2.95, + "learning_rate": 1.4572353102737894e-05, + "loss": 0.2532, + "step": 7038 + }, + { + "epoch": 2.95, + "learning_rate": 1.4570845232341943e-05, + "loss": 0.249, + "step": 7039 + }, + { + "epoch": 2.95, + "learning_rate": 1.4569337230561105e-05, + "loss": 0.2475, + "step": 7040 + }, + { + "epoch": 2.95, + "learning_rate": 1.4567829097438719e-05, + "loss": 0.2634, + "step": 7041 + }, + { + "epoch": 2.95, + "learning_rate": 1.456632083301814e-05, + "loss": 0.2866, + "step": 7042 + }, + { + "epoch": 2.95, + "learning_rate": 1.4564812437342724e-05, + "loss": 0.2312, + "step": 7043 + }, + { + "epoch": 2.95, + "learning_rate": 1.4563303910455821e-05, + "loss": 0.2826, + "step": 7044 + }, + { + "epoch": 2.95, + "learning_rate": 1.4561795252400797e-05, + "loss": 0.2687, + "step": 7045 + }, + { + "epoch": 2.95, + "learning_rate": 1.4560286463221018e-05, + "loss": 0.2409, + "step": 7046 + }, + { + "epoch": 2.95, + "learning_rate": 1.455877754295985e-05, + "loss": 0.2738, + "step": 7047 + }, + { + "epoch": 2.95, + "learning_rate": 1.455726849166067e-05, + "loss": 0.2755, + "step": 7048 + }, + { + "epoch": 2.95, + "learning_rate": 1.4555759309366848e-05, + "loss": 0.2729, + "step": 7049 + }, + { + "epoch": 2.95, + "learning_rate": 1.455424999612177e-05, + "loss": 0.2863, + "step": 7050 + }, + { + "epoch": 2.95, + "learning_rate": 1.4552740551968817e-05, + "loss": 0.2536, + "step": 7051 + }, + { + "epoch": 2.95, + "learning_rate": 1.4551230976951373e-05, + "loss": 0.2727, + "step": 7052 + }, + { + "epoch": 2.95, + "learning_rate": 1.4549721271112835e-05, + "loss": 0.2709, + "step": 7053 + }, + { + "epoch": 2.95, + "learning_rate": 1.4548211434496598e-05, + "loss": 0.2484, + "step": 7054 + }, + { + "epoch": 2.95, + "learning_rate": 1.4546701467146058e-05, + "loss": 0.2855, + "step": 7055 + }, + { + "epoch": 2.95, + "learning_rate": 1.4545191369104618e-05, + "loss": 0.265, + "step": 7056 + }, + { + "epoch": 2.95, + "learning_rate": 1.4543681140415688e-05, + "loss": 0.2499, + "step": 7057 + }, + { + "epoch": 2.95, + "learning_rate": 1.4542170781122673e-05, + "loss": 0.26, + "step": 7058 + }, + { + "epoch": 2.96, + "learning_rate": 1.4540660291268993e-05, + "loss": 0.2607, + "step": 7059 + }, + { + "epoch": 2.96, + "learning_rate": 1.453914967089806e-05, + "loss": 0.2486, + "step": 7060 + }, + { + "epoch": 2.96, + "learning_rate": 1.4537638920053298e-05, + "loss": 0.3044, + "step": 7061 + }, + { + "epoch": 2.96, + "learning_rate": 1.4536128038778132e-05, + "loss": 0.2689, + "step": 7062 + }, + { + "epoch": 2.96, + "learning_rate": 1.453461702711599e-05, + "loss": 0.2797, + "step": 7063 + }, + { + "epoch": 2.96, + "learning_rate": 1.4533105885110307e-05, + "loss": 0.2443, + "step": 7064 + }, + { + "epoch": 2.96, + "learning_rate": 1.4531594612804516e-05, + "loss": 0.2563, + "step": 7065 + }, + { + "epoch": 2.96, + "learning_rate": 1.4530083210242062e-05, + "loss": 0.2787, + "step": 7066 + }, + { + "epoch": 2.96, + "learning_rate": 1.4528571677466383e-05, + "loss": 0.2638, + "step": 7067 + }, + { + "epoch": 2.96, + "learning_rate": 1.4527060014520932e-05, + "loss": 0.2708, + "step": 7068 + }, + { + "epoch": 2.96, + "learning_rate": 1.4525548221449156e-05, + "loss": 0.235, + "step": 7069 + }, + { + "epoch": 2.96, + "learning_rate": 1.4524036298294515e-05, + "loss": 0.2447, + "step": 7070 + }, + { + "epoch": 2.96, + "learning_rate": 1.4522524245100465e-05, + "loss": 0.2411, + "step": 7071 + }, + { + "epoch": 2.96, + "learning_rate": 1.4521012061910466e-05, + "loss": 0.26, + "step": 7072 + }, + { + "epoch": 2.96, + "learning_rate": 1.4519499748767989e-05, + "loss": 0.2866, + "step": 7073 + }, + { + "epoch": 2.96, + "learning_rate": 1.4517987305716502e-05, + "loss": 0.2783, + "step": 7074 + }, + { + "epoch": 2.96, + "learning_rate": 1.4516474732799476e-05, + "loss": 0.2697, + "step": 7075 + }, + { + "epoch": 2.96, + "learning_rate": 1.4514962030060395e-05, + "loss": 0.2484, + "step": 7076 + }, + { + "epoch": 2.96, + "learning_rate": 1.451344919754274e-05, + "loss": 0.2751, + "step": 7077 + }, + { + "epoch": 2.96, + "learning_rate": 1.4511936235289987e-05, + "loss": 0.2818, + "step": 7078 + }, + { + "epoch": 2.96, + "learning_rate": 1.451042314334563e-05, + "loss": 0.2503, + "step": 7079 + }, + { + "epoch": 2.96, + "learning_rate": 1.450890992175317e-05, + "loss": 0.2401, + "step": 7080 + }, + { + "epoch": 2.96, + "learning_rate": 1.450739657055609e-05, + "loss": 0.2721, + "step": 7081 + }, + { + "epoch": 2.97, + "learning_rate": 1.4505883089797896e-05, + "loss": 0.2605, + "step": 7082 + }, + { + "epoch": 2.97, + "learning_rate": 1.4504369479522092e-05, + "loss": 0.2714, + "step": 7083 + }, + { + "epoch": 2.97, + "learning_rate": 1.4502855739772184e-05, + "loss": 0.2754, + "step": 7084 + }, + { + "epoch": 2.97, + "learning_rate": 1.4501341870591683e-05, + "loss": 0.2411, + "step": 7085 + }, + { + "epoch": 2.97, + "learning_rate": 1.4499827872024104e-05, + "loss": 0.2623, + "step": 7086 + }, + { + "epoch": 2.97, + "learning_rate": 1.4498313744112966e-05, + "loss": 0.2632, + "step": 7087 + }, + { + "epoch": 2.97, + "learning_rate": 1.4496799486901791e-05, + "loss": 0.2663, + "step": 7088 + }, + { + "epoch": 2.97, + "learning_rate": 1.4495285100434107e-05, + "loss": 0.2793, + "step": 7089 + }, + { + "epoch": 2.97, + "learning_rate": 1.449377058475344e-05, + "loss": 0.2695, + "step": 7090 + }, + { + "epoch": 2.97, + "learning_rate": 1.4492255939903326e-05, + "loss": 0.2576, + "step": 7091 + }, + { + "epoch": 2.97, + "learning_rate": 1.44907411659273e-05, + "loss": 0.2428, + "step": 7092 + }, + { + "epoch": 2.97, + "learning_rate": 1.4489226262868906e-05, + "loss": 0.2305, + "step": 7093 + }, + { + "epoch": 2.97, + "learning_rate": 1.4487711230771686e-05, + "loss": 0.2953, + "step": 7094 + }, + { + "epoch": 2.97, + "learning_rate": 1.4486196069679191e-05, + "loss": 0.2662, + "step": 7095 + }, + { + "epoch": 2.97, + "learning_rate": 1.4484680779634968e-05, + "loss": 0.2287, + "step": 7096 + }, + { + "epoch": 2.97, + "learning_rate": 1.4483165360682577e-05, + "loss": 0.2736, + "step": 7097 + }, + { + "epoch": 2.97, + "learning_rate": 1.4481649812865574e-05, + "loss": 0.2746, + "step": 7098 + }, + { + "epoch": 2.97, + "learning_rate": 1.4480134136227526e-05, + "loss": 0.2411, + "step": 7099 + }, + { + "epoch": 2.97, + "learning_rate": 1.4478618330811997e-05, + "loss": 0.2587, + "step": 7100 + }, + { + "epoch": 2.97, + "learning_rate": 1.447710239666256e-05, + "loss": 0.2382, + "step": 7101 + }, + { + "epoch": 2.97, + "learning_rate": 1.4475586333822787e-05, + "loss": 0.2577, + "step": 7102 + }, + { + "epoch": 2.97, + "learning_rate": 1.4474070142336254e-05, + "loss": 0.2714, + "step": 7103 + }, + { + "epoch": 2.97, + "learning_rate": 1.4472553822246548e-05, + "loss": 0.2527, + "step": 7104 + }, + { + "epoch": 2.97, + "learning_rate": 1.4471037373597251e-05, + "loss": 0.292, + "step": 7105 + }, + { + "epoch": 2.98, + "learning_rate": 1.446952079643195e-05, + "loss": 0.2405, + "step": 7106 + }, + { + "epoch": 2.98, + "learning_rate": 1.4468004090794241e-05, + "loss": 0.2724, + "step": 7107 + }, + { + "epoch": 2.98, + "learning_rate": 1.4466487256727721e-05, + "loss": 0.2636, + "step": 7108 + }, + { + "epoch": 2.98, + "learning_rate": 1.4464970294275986e-05, + "loss": 0.2888, + "step": 7109 + }, + { + "epoch": 2.98, + "learning_rate": 1.4463453203482643e-05, + "loss": 0.2672, + "step": 7110 + }, + { + "epoch": 2.98, + "learning_rate": 1.44619359843913e-05, + "loss": 0.2623, + "step": 7111 + }, + { + "epoch": 2.98, + "learning_rate": 1.4460418637045564e-05, + "loss": 0.2394, + "step": 7112 + }, + { + "epoch": 2.98, + "learning_rate": 1.4458901161489051e-05, + "loss": 0.2543, + "step": 7113 + }, + { + "epoch": 2.98, + "learning_rate": 1.4457383557765385e-05, + "loss": 0.2451, + "step": 7114 + }, + { + "epoch": 2.98, + "learning_rate": 1.4455865825918182e-05, + "loss": 0.2542, + "step": 7115 + }, + { + "epoch": 2.98, + "learning_rate": 1.445434796599107e-05, + "loss": 0.2598, + "step": 7116 + }, + { + "epoch": 2.98, + "learning_rate": 1.4452829978027678e-05, + "loss": 0.2718, + "step": 7117 + }, + { + "epoch": 2.98, + "learning_rate": 1.4451311862071641e-05, + "loss": 0.2405, + "step": 7118 + }, + { + "epoch": 2.98, + "learning_rate": 1.4449793618166594e-05, + "loss": 0.2497, + "step": 7119 + }, + { + "epoch": 2.98, + "learning_rate": 1.4448275246356175e-05, + "loss": 0.2675, + "step": 7120 + }, + { + "epoch": 2.98, + "learning_rate": 1.4446756746684036e-05, + "loss": 0.2779, + "step": 7121 + }, + { + "epoch": 2.98, + "learning_rate": 1.4445238119193818e-05, + "loss": 0.2645, + "step": 7122 + }, + { + "epoch": 2.98, + "learning_rate": 1.4443719363929173e-05, + "loss": 0.2663, + "step": 7123 + }, + { + "epoch": 2.98, + "learning_rate": 1.4442200480933761e-05, + "loss": 0.2581, + "step": 7124 + }, + { + "epoch": 2.98, + "learning_rate": 1.4440681470251239e-05, + "loss": 0.2753, + "step": 7125 + }, + { + "epoch": 2.98, + "learning_rate": 1.4439162331925268e-05, + "loss": 0.2406, + "step": 7126 + }, + { + "epoch": 2.98, + "learning_rate": 1.4437643065999516e-05, + "loss": 0.2575, + "step": 7127 + }, + { + "epoch": 2.98, + "learning_rate": 1.443612367251765e-05, + "loss": 0.2494, + "step": 7128 + }, + { + "epoch": 2.98, + "learning_rate": 1.4434604151523345e-05, + "loss": 0.2552, + "step": 7129 + }, + { + "epoch": 2.99, + "learning_rate": 1.443308450306028e-05, + "loss": 0.2535, + "step": 7130 + }, + { + "epoch": 2.99, + "learning_rate": 1.4431564727172135e-05, + "loss": 0.2476, + "step": 7131 + }, + { + "epoch": 2.99, + "learning_rate": 1.4430044823902594e-05, + "loss": 0.2381, + "step": 7132 + }, + { + "epoch": 2.99, + "learning_rate": 1.4428524793295344e-05, + "loss": 0.2544, + "step": 7133 + }, + { + "epoch": 2.99, + "learning_rate": 1.442700463539408e-05, + "loss": 0.269, + "step": 7134 + }, + { + "epoch": 2.99, + "learning_rate": 1.4425484350242498e-05, + "loss": 0.24, + "step": 7135 + }, + { + "epoch": 2.99, + "learning_rate": 1.4423963937884293e-05, + "loss": 0.2364, + "step": 7136 + }, + { + "epoch": 2.99, + "learning_rate": 1.442244339836317e-05, + "loss": 0.2322, + "step": 7137 + }, + { + "epoch": 2.99, + "learning_rate": 1.442092273172284e-05, + "loss": 0.2413, + "step": 7138 + }, + { + "epoch": 2.99, + "learning_rate": 1.4419401938007007e-05, + "loss": 0.2707, + "step": 7139 + }, + { + "epoch": 2.99, + "learning_rate": 1.4417881017259385e-05, + "loss": 0.3087, + "step": 7140 + }, + { + "epoch": 2.99, + "learning_rate": 1.4416359969523694e-05, + "loss": 0.265, + "step": 7141 + }, + { + "epoch": 2.99, + "learning_rate": 1.4414838794843658e-05, + "loss": 0.2298, + "step": 7142 + }, + { + "epoch": 2.99, + "learning_rate": 1.4413317493262996e-05, + "loss": 0.2708, + "step": 7143 + }, + { + "epoch": 2.99, + "learning_rate": 1.4411796064825436e-05, + "loss": 0.2349, + "step": 7144 + }, + { + "epoch": 2.99, + "learning_rate": 1.4410274509574718e-05, + "loss": 0.2678, + "step": 7145 + }, + { + "epoch": 2.99, + "learning_rate": 1.4408752827554571e-05, + "loss": 0.2931, + "step": 7146 + }, + { + "epoch": 2.99, + "learning_rate": 1.4407231018808734e-05, + "loss": 0.2673, + "step": 7147 + }, + { + "epoch": 2.99, + "learning_rate": 1.4405709083380957e-05, + "loss": 0.2571, + "step": 7148 + }, + { + "epoch": 2.99, + "learning_rate": 1.4404187021314982e-05, + "loss": 0.2492, + "step": 7149 + }, + { + "epoch": 2.99, + "learning_rate": 1.4402664832654555e-05, + "loss": 0.2863, + "step": 7150 + }, + { + "epoch": 2.99, + "learning_rate": 1.4401142517443434e-05, + "loss": 0.2501, + "step": 7151 + }, + { + "epoch": 2.99, + "learning_rate": 1.439962007572538e-05, + "loss": 0.2647, + "step": 7152 + }, + { + "epoch": 2.99, + "learning_rate": 1.4398097507544152e-05, + "loss": 0.2374, + "step": 7153 + }, + { + "epoch": 3.0, + "learning_rate": 1.439657481294351e-05, + "loss": 0.2462, + "step": 7154 + }, + { + "epoch": 3.0, + "learning_rate": 1.4395051991967228e-05, + "loss": 0.2512, + "step": 7155 + }, + { + "epoch": 3.0, + "learning_rate": 1.4393529044659077e-05, + "loss": 0.2685, + "step": 7156 + }, + { + "epoch": 3.0, + "learning_rate": 1.439200597106283e-05, + "loss": 0.2732, + "step": 7157 + }, + { + "epoch": 3.0, + "learning_rate": 1.4390482771222273e-05, + "loss": 0.2834, + "step": 7158 + }, + { + "epoch": 3.0, + "learning_rate": 1.4388959445181181e-05, + "loss": 0.2718, + "step": 7159 + }, + { + "epoch": 3.0, + "learning_rate": 1.4387435992983347e-05, + "loss": 0.2553, + "step": 7160 + }, + { + "epoch": 3.0, + "learning_rate": 1.4385912414672559e-05, + "loss": 0.23, + "step": 7161 + }, + { + "epoch": 3.0, + "learning_rate": 1.438438871029261e-05, + "loss": 0.2289, + "step": 7162 + }, + { + "epoch": 3.0, + "learning_rate": 1.4382864879887298e-05, + "loss": 0.2558, + "step": 7163 + }, + { + "epoch": 3.0, + "learning_rate": 1.4381340923500422e-05, + "loss": 0.2601, + "step": 7164 + }, + { + "epoch": 3.0, + "learning_rate": 1.437981684117579e-05, + "loss": 0.2692, + "step": 7165 + }, + { + "epoch": 3.0, + "learning_rate": 1.4378292632957214e-05, + "loss": 0.2586, + "step": 7166 + }, + { + "epoch": 3.0, + "learning_rate": 1.4376768298888498e-05, + "loss": 0.1887, + "step": 7167 + }, + { + "epoch": 3.0, + "learning_rate": 1.4375243839013458e-05, + "loss": 0.2059, + "step": 7168 + }, + { + "epoch": 3.0, + "learning_rate": 1.4373719253375921e-05, + "loss": 0.231, + "step": 7169 + }, + { + "epoch": 3.0, + "learning_rate": 1.4372194542019705e-05, + "loss": 0.1802, + "step": 7170 + }, + { + "epoch": 3.0, + "learning_rate": 1.4370669704988634e-05, + "loss": 0.1511, + "step": 7171 + }, + { + "epoch": 3.0, + "learning_rate": 1.4369144742326544e-05, + "loss": 0.1502, + "step": 7172 + }, + { + "epoch": 3.0, + "learning_rate": 1.4367619654077264e-05, + "loss": 0.1817, + "step": 7173 + }, + { + "epoch": 3.0, + "learning_rate": 1.4366094440284632e-05, + "loss": 0.1603, + "step": 7174 + }, + { + "epoch": 3.0, + "learning_rate": 1.4364569100992489e-05, + "loss": 0.2057, + "step": 7175 + }, + { + "epoch": 3.0, + "learning_rate": 1.4363043636244682e-05, + "loss": 0.1826, + "step": 7176 + }, + { + "epoch": 3.0, + "learning_rate": 1.4361518046085057e-05, + "loss": 0.1632, + "step": 7177 + }, + { + "epoch": 3.01, + "learning_rate": 1.4359992330557467e-05, + "loss": 0.1868, + "step": 7178 + }, + { + "epoch": 3.01, + "learning_rate": 1.4358466489705766e-05, + "loss": 0.1837, + "step": 7179 + }, + { + "epoch": 3.01, + "learning_rate": 1.4356940523573812e-05, + "loss": 0.2101, + "step": 7180 + }, + { + "epoch": 3.01, + "learning_rate": 1.4355414432205468e-05, + "loss": 0.1869, + "step": 7181 + }, + { + "epoch": 3.01, + "learning_rate": 1.4353888215644604e-05, + "loss": 0.2015, + "step": 7182 + }, + { + "epoch": 3.01, + "learning_rate": 1.4352361873935088e-05, + "loss": 0.1817, + "step": 7183 + }, + { + "epoch": 3.01, + "learning_rate": 1.435083540712079e-05, + "loss": 0.2035, + "step": 7184 + }, + { + "epoch": 3.01, + "learning_rate": 1.4349308815245586e-05, + "loss": 0.1981, + "step": 7185 + }, + { + "epoch": 3.01, + "learning_rate": 1.4347782098353364e-05, + "loss": 0.1769, + "step": 7186 + }, + { + "epoch": 3.01, + "learning_rate": 1.4346255256488002e-05, + "loss": 0.1934, + "step": 7187 + }, + { + "epoch": 3.01, + "learning_rate": 1.4344728289693391e-05, + "loss": 0.1702, + "step": 7188 + }, + { + "epoch": 3.01, + "learning_rate": 1.434320119801342e-05, + "loss": 0.177, + "step": 7189 + }, + { + "epoch": 3.01, + "learning_rate": 1.4341673981491982e-05, + "loss": 0.1806, + "step": 7190 + }, + { + "epoch": 3.01, + "learning_rate": 1.4340146640172981e-05, + "loss": 0.2014, + "step": 7191 + }, + { + "epoch": 3.01, + "learning_rate": 1.4338619174100315e-05, + "loss": 0.1654, + "step": 7192 + }, + { + "epoch": 3.01, + "learning_rate": 1.433709158331789e-05, + "loss": 0.1999, + "step": 7193 + }, + { + "epoch": 3.01, + "learning_rate": 1.4335563867869618e-05, + "loss": 0.1922, + "step": 7194 + }, + { + "epoch": 3.01, + "learning_rate": 1.433403602779941e-05, + "loss": 0.1618, + "step": 7195 + }, + { + "epoch": 3.01, + "learning_rate": 1.4332508063151184e-05, + "loss": 0.1569, + "step": 7196 + }, + { + "epoch": 3.01, + "learning_rate": 1.4330979973968856e-05, + "loss": 0.2257, + "step": 7197 + }, + { + "epoch": 3.01, + "learning_rate": 1.4329451760296352e-05, + "loss": 0.1733, + "step": 7198 + }, + { + "epoch": 3.01, + "learning_rate": 1.43279234221776e-05, + "loss": 0.2231, + "step": 7199 + }, + { + "epoch": 3.01, + "learning_rate": 1.4326394959656527e-05, + "loss": 0.1967, + "step": 7200 + }, + { + "epoch": 3.01, + "learning_rate": 1.4324866372777072e-05, + "loss": 0.2143, + "step": 7201 + }, + { + "epoch": 3.02, + "learning_rate": 1.4323337661583168e-05, + "loss": 0.1895, + "step": 7202 + }, + { + "epoch": 3.02, + "learning_rate": 1.4321808826118764e-05, + "loss": 0.2105, + "step": 7203 + }, + { + "epoch": 3.02, + "learning_rate": 1.4320279866427798e-05, + "loss": 0.1458, + "step": 7204 + }, + { + "epoch": 3.02, + "learning_rate": 1.431875078255422e-05, + "loss": 0.1854, + "step": 7205 + }, + { + "epoch": 3.02, + "learning_rate": 1.4317221574541985e-05, + "loss": 0.1838, + "step": 7206 + }, + { + "epoch": 3.02, + "learning_rate": 1.4315692242435048e-05, + "loss": 0.1533, + "step": 7207 + }, + { + "epoch": 3.02, + "learning_rate": 1.4314162786277366e-05, + "loss": 0.1994, + "step": 7208 + }, + { + "epoch": 3.02, + "learning_rate": 1.43126332061129e-05, + "loss": 0.2194, + "step": 7209 + }, + { + "epoch": 3.02, + "learning_rate": 1.4311103501985625e-05, + "loss": 0.169, + "step": 7210 + }, + { + "epoch": 3.02, + "learning_rate": 1.4309573673939503e-05, + "loss": 0.2104, + "step": 7211 + }, + { + "epoch": 3.02, + "learning_rate": 1.4308043722018508e-05, + "loss": 0.2042, + "step": 7212 + }, + { + "epoch": 3.02, + "learning_rate": 1.4306513646266621e-05, + "loss": 0.1725, + "step": 7213 + }, + { + "epoch": 3.02, + "learning_rate": 1.430498344672782e-05, + "loss": 0.1701, + "step": 7214 + }, + { + "epoch": 3.02, + "learning_rate": 1.430345312344609e-05, + "loss": 0.1718, + "step": 7215 + }, + { + "epoch": 3.02, + "learning_rate": 1.430192267646542e-05, + "loss": 0.1763, + "step": 7216 + }, + { + "epoch": 3.02, + "learning_rate": 1.4300392105829802e-05, + "loss": 0.17, + "step": 7217 + }, + { + "epoch": 3.02, + "learning_rate": 1.4298861411583224e-05, + "loss": 0.1941, + "step": 7218 + }, + { + "epoch": 3.02, + "learning_rate": 1.4297330593769691e-05, + "loss": 0.2126, + "step": 7219 + }, + { + "epoch": 3.02, + "learning_rate": 1.4295799652433206e-05, + "loss": 0.1735, + "step": 7220 + }, + { + "epoch": 3.02, + "learning_rate": 1.429426858761777e-05, + "loss": 0.1709, + "step": 7221 + }, + { + "epoch": 3.02, + "learning_rate": 1.4292737399367394e-05, + "loss": 0.1972, + "step": 7222 + }, + { + "epoch": 3.02, + "learning_rate": 1.429120608772609e-05, + "loss": 0.19, + "step": 7223 + }, + { + "epoch": 3.02, + "learning_rate": 1.4289674652737877e-05, + "loss": 0.1853, + "step": 7224 + }, + { + "epoch": 3.02, + "learning_rate": 1.4288143094446771e-05, + "loss": 0.1501, + "step": 7225 + }, + { + "epoch": 3.03, + "learning_rate": 1.42866114128968e-05, + "loss": 0.1467, + "step": 7226 + }, + { + "epoch": 3.03, + "learning_rate": 1.4285079608131984e-05, + "loss": 0.148, + "step": 7227 + }, + { + "epoch": 3.03, + "learning_rate": 1.4283547680196359e-05, + "loss": 0.2309, + "step": 7228 + }, + { + "epoch": 3.03, + "learning_rate": 1.4282015629133958e-05, + "loss": 0.2142, + "step": 7229 + }, + { + "epoch": 3.03, + "learning_rate": 1.4280483454988816e-05, + "loss": 0.2085, + "step": 7230 + }, + { + "epoch": 3.03, + "learning_rate": 1.4278951157804977e-05, + "loss": 0.1577, + "step": 7231 + }, + { + "epoch": 3.03, + "learning_rate": 1.4277418737626483e-05, + "loss": 0.1826, + "step": 7232 + }, + { + "epoch": 3.03, + "learning_rate": 1.4275886194497384e-05, + "loss": 0.2294, + "step": 7233 + }, + { + "epoch": 3.03, + "learning_rate": 1.4274353528461728e-05, + "loss": 0.1659, + "step": 7234 + }, + { + "epoch": 3.03, + "learning_rate": 1.4272820739563575e-05, + "loss": 0.226, + "step": 7235 + }, + { + "epoch": 3.03, + "learning_rate": 1.427128782784698e-05, + "loss": 0.1884, + "step": 7236 + }, + { + "epoch": 3.03, + "learning_rate": 1.4269754793356008e-05, + "loss": 0.1751, + "step": 7237 + }, + { + "epoch": 3.03, + "learning_rate": 1.4268221636134722e-05, + "loss": 0.2024, + "step": 7238 + }, + { + "epoch": 3.03, + "learning_rate": 1.4266688356227194e-05, + "loss": 0.1908, + "step": 7239 + }, + { + "epoch": 3.03, + "learning_rate": 1.4265154953677498e-05, + "loss": 0.1581, + "step": 7240 + }, + { + "epoch": 3.03, + "learning_rate": 1.4263621428529703e-05, + "loss": 0.1458, + "step": 7241 + }, + { + "epoch": 3.03, + "learning_rate": 1.4262087780827896e-05, + "loss": 0.18, + "step": 7242 + }, + { + "epoch": 3.03, + "learning_rate": 1.4260554010616156e-05, + "loss": 0.1969, + "step": 7243 + }, + { + "epoch": 3.03, + "learning_rate": 1.4259020117938574e-05, + "loss": 0.1862, + "step": 7244 + }, + { + "epoch": 3.03, + "learning_rate": 1.4257486102839236e-05, + "loss": 0.2063, + "step": 7245 + }, + { + "epoch": 3.03, + "learning_rate": 1.4255951965362238e-05, + "loss": 0.1843, + "step": 7246 + }, + { + "epoch": 3.03, + "learning_rate": 1.4254417705551679e-05, + "loss": 0.1448, + "step": 7247 + }, + { + "epoch": 3.03, + "learning_rate": 1.4252883323451658e-05, + "loss": 0.2114, + "step": 7248 + }, + { + "epoch": 3.03, + "learning_rate": 1.425134881910628e-05, + "loss": 0.2203, + "step": 7249 + }, + { + "epoch": 3.04, + "learning_rate": 1.4249814192559653e-05, + "loss": 0.1884, + "step": 7250 + }, + { + "epoch": 3.04, + "learning_rate": 1.4248279443855891e-05, + "loss": 0.1889, + "step": 7251 + }, + { + "epoch": 3.04, + "learning_rate": 1.42467445730391e-05, + "loss": 0.1804, + "step": 7252 + }, + { + "epoch": 3.04, + "learning_rate": 1.4245209580153408e-05, + "loss": 0.174, + "step": 7253 + }, + { + "epoch": 3.04, + "learning_rate": 1.4243674465242934e-05, + "loss": 0.1691, + "step": 7254 + }, + { + "epoch": 3.04, + "learning_rate": 1.42421392283518e-05, + "loss": 0.1507, + "step": 7255 + }, + { + "epoch": 3.04, + "learning_rate": 1.424060386952414e-05, + "loss": 0.1765, + "step": 7256 + }, + { + "epoch": 3.04, + "learning_rate": 1.4239068388804089e-05, + "loss": 0.1813, + "step": 7257 + }, + { + "epoch": 3.04, + "learning_rate": 1.4237532786235775e-05, + "loss": 0.1951, + "step": 7258 + }, + { + "epoch": 3.04, + "learning_rate": 1.423599706186334e-05, + "loss": 0.2148, + "step": 7259 + }, + { + "epoch": 3.04, + "learning_rate": 1.4234461215730932e-05, + "loss": 0.1641, + "step": 7260 + }, + { + "epoch": 3.04, + "learning_rate": 1.4232925247882692e-05, + "loss": 0.1737, + "step": 7261 + }, + { + "epoch": 3.04, + "learning_rate": 1.4231389158362771e-05, + "loss": 0.1844, + "step": 7262 + }, + { + "epoch": 3.04, + "learning_rate": 1.4229852947215326e-05, + "loss": 0.179, + "step": 7263 + }, + { + "epoch": 3.04, + "learning_rate": 1.4228316614484511e-05, + "loss": 0.1891, + "step": 7264 + }, + { + "epoch": 3.04, + "learning_rate": 1.4226780160214486e-05, + "loss": 0.166, + "step": 7265 + }, + { + "epoch": 3.04, + "learning_rate": 1.4225243584449415e-05, + "loss": 0.1638, + "step": 7266 + }, + { + "epoch": 3.04, + "learning_rate": 1.422370688723347e-05, + "loss": 0.2241, + "step": 7267 + }, + { + "epoch": 3.04, + "learning_rate": 1.4222170068610816e-05, + "loss": 0.2143, + "step": 7268 + }, + { + "epoch": 3.04, + "learning_rate": 1.4220633128625632e-05, + "loss": 0.2018, + "step": 7269 + }, + { + "epoch": 3.04, + "learning_rate": 1.4219096067322094e-05, + "loss": 0.1457, + "step": 7270 + }, + { + "epoch": 3.04, + "learning_rate": 1.421755888474438e-05, + "loss": 0.1974, + "step": 7271 + }, + { + "epoch": 3.04, + "learning_rate": 1.4216021580936681e-05, + "loss": 0.1919, + "step": 7272 + }, + { + "epoch": 3.05, + "learning_rate": 1.421448415594318e-05, + "loss": 0.2016, + "step": 7273 + }, + { + "epoch": 3.05, + "learning_rate": 1.4212946609808079e-05, + "loss": 0.1676, + "step": 7274 + }, + { + "epoch": 3.05, + "learning_rate": 1.421140894257556e-05, + "loss": 0.2124, + "step": 7275 + }, + { + "epoch": 3.05, + "learning_rate": 1.420987115428983e-05, + "loss": 0.2214, + "step": 7276 + }, + { + "epoch": 3.05, + "learning_rate": 1.4208333244995087e-05, + "loss": 0.1987, + "step": 7277 + }, + { + "epoch": 3.05, + "learning_rate": 1.4206795214735543e-05, + "loss": 0.1791, + "step": 7278 + }, + { + "epoch": 3.05, + "learning_rate": 1.42052570635554e-05, + "loss": 0.2003, + "step": 7279 + }, + { + "epoch": 3.05, + "learning_rate": 1.4203718791498877e-05, + "loss": 0.1513, + "step": 7280 + }, + { + "epoch": 3.05, + "learning_rate": 1.4202180398610188e-05, + "loss": 0.1887, + "step": 7281 + }, + { + "epoch": 3.05, + "learning_rate": 1.420064188493355e-05, + "loss": 0.2105, + "step": 7282 + }, + { + "epoch": 3.05, + "learning_rate": 1.4199103250513192e-05, + "loss": 0.1794, + "step": 7283 + }, + { + "epoch": 3.05, + "learning_rate": 1.4197564495393334e-05, + "loss": 0.1558, + "step": 7284 + }, + { + "epoch": 3.05, + "learning_rate": 1.4196025619618212e-05, + "loss": 0.1765, + "step": 7285 + }, + { + "epoch": 3.05, + "learning_rate": 1.4194486623232056e-05, + "loss": 0.1902, + "step": 7286 + }, + { + "epoch": 3.05, + "learning_rate": 1.4192947506279102e-05, + "loss": 0.1742, + "step": 7287 + }, + { + "epoch": 3.05, + "learning_rate": 1.4191408268803596e-05, + "loss": 0.1754, + "step": 7288 + }, + { + "epoch": 3.05, + "learning_rate": 1.4189868910849779e-05, + "loss": 0.1759, + "step": 7289 + }, + { + "epoch": 3.05, + "learning_rate": 1.4188329432461894e-05, + "loss": 0.1869, + "step": 7290 + }, + { + "epoch": 3.05, + "learning_rate": 1.41867898336842e-05, + "loss": 0.1341, + "step": 7291 + }, + { + "epoch": 3.05, + "learning_rate": 1.4185250114560945e-05, + "loss": 0.1637, + "step": 7292 + }, + { + "epoch": 3.05, + "learning_rate": 1.4183710275136391e-05, + "loss": 0.1879, + "step": 7293 + }, + { + "epoch": 3.05, + "learning_rate": 1.4182170315454798e-05, + "loss": 0.1744, + "step": 7294 + }, + { + "epoch": 3.05, + "learning_rate": 1.4180630235560428e-05, + "loss": 0.1949, + "step": 7295 + }, + { + "epoch": 3.05, + "learning_rate": 1.4179090035497554e-05, + "loss": 0.1792, + "step": 7296 + }, + { + "epoch": 3.06, + "learning_rate": 1.4177549715310445e-05, + "loss": 0.169, + "step": 7297 + }, + { + "epoch": 3.06, + "learning_rate": 1.4176009275043376e-05, + "loss": 0.1775, + "step": 7298 + }, + { + "epoch": 3.06, + "learning_rate": 1.4174468714740628e-05, + "loss": 0.1806, + "step": 7299 + }, + { + "epoch": 3.06, + "learning_rate": 1.4172928034446478e-05, + "loss": 0.217, + "step": 7300 + }, + { + "epoch": 3.06, + "learning_rate": 1.4171387234205219e-05, + "loss": 0.1439, + "step": 7301 + }, + { + "epoch": 3.06, + "learning_rate": 1.4169846314061132e-05, + "loss": 0.1562, + "step": 7302 + }, + { + "epoch": 3.06, + "learning_rate": 1.4168305274058511e-05, + "loss": 0.1626, + "step": 7303 + }, + { + "epoch": 3.06, + "learning_rate": 1.4166764114241657e-05, + "loss": 0.1897, + "step": 7304 + }, + { + "epoch": 3.06, + "learning_rate": 1.4165222834654867e-05, + "loss": 0.2121, + "step": 7305 + }, + { + "epoch": 3.06, + "learning_rate": 1.416368143534244e-05, + "loss": 0.1895, + "step": 7306 + }, + { + "epoch": 3.06, + "learning_rate": 1.4162139916348685e-05, + "loss": 0.2021, + "step": 7307 + }, + { + "epoch": 3.06, + "learning_rate": 1.4160598277717916e-05, + "loss": 0.1783, + "step": 7308 + }, + { + "epoch": 3.06, + "learning_rate": 1.4159056519494439e-05, + "loss": 0.2108, + "step": 7309 + }, + { + "epoch": 3.06, + "learning_rate": 1.4157514641722569e-05, + "loss": 0.2187, + "step": 7310 + }, + { + "epoch": 3.06, + "learning_rate": 1.4155972644446635e-05, + "loss": 0.1467, + "step": 7311 + }, + { + "epoch": 3.06, + "learning_rate": 1.4154430527710954e-05, + "loss": 0.1728, + "step": 7312 + }, + { + "epoch": 3.06, + "learning_rate": 1.4152888291559852e-05, + "loss": 0.1972, + "step": 7313 + }, + { + "epoch": 3.06, + "learning_rate": 1.4151345936037664e-05, + "loss": 0.2267, + "step": 7314 + }, + { + "epoch": 3.06, + "learning_rate": 1.414980346118872e-05, + "loss": 0.1653, + "step": 7315 + }, + { + "epoch": 3.06, + "learning_rate": 1.4148260867057357e-05, + "loss": 0.1404, + "step": 7316 + }, + { + "epoch": 3.06, + "learning_rate": 1.4146718153687917e-05, + "loss": 0.1793, + "step": 7317 + }, + { + "epoch": 3.06, + "learning_rate": 1.4145175321124744e-05, + "loss": 0.1372, + "step": 7318 + }, + { + "epoch": 3.06, + "learning_rate": 1.4143632369412185e-05, + "loss": 0.1705, + "step": 7319 + }, + { + "epoch": 3.06, + "learning_rate": 1.414208929859459e-05, + "loss": 0.2271, + "step": 7320 + }, + { + "epoch": 3.07, + "learning_rate": 1.4140546108716313e-05, + "loss": 0.1695, + "step": 7321 + }, + { + "epoch": 3.07, + "learning_rate": 1.4139002799821714e-05, + "loss": 0.1739, + "step": 7322 + }, + { + "epoch": 3.07, + "learning_rate": 1.4137459371955151e-05, + "loss": 0.2014, + "step": 7323 + }, + { + "epoch": 3.07, + "learning_rate": 1.413591582516099e-05, + "loss": 0.1908, + "step": 7324 + }, + { + "epoch": 3.07, + "learning_rate": 1.41343721594836e-05, + "loss": 0.1706, + "step": 7325 + }, + { + "epoch": 3.07, + "learning_rate": 1.4132828374967348e-05, + "loss": 0.1861, + "step": 7326 + }, + { + "epoch": 3.07, + "learning_rate": 1.4131284471656615e-05, + "loss": 0.1634, + "step": 7327 + }, + { + "epoch": 3.07, + "learning_rate": 1.4129740449595777e-05, + "loss": 0.1741, + "step": 7328 + }, + { + "epoch": 3.07, + "learning_rate": 1.4128196308829209e-05, + "loss": 0.1747, + "step": 7329 + }, + { + "epoch": 3.07, + "learning_rate": 1.4126652049401306e-05, + "loss": 0.1682, + "step": 7330 + }, + { + "epoch": 3.07, + "learning_rate": 1.4125107671356451e-05, + "loss": 0.19, + "step": 7331 + }, + { + "epoch": 3.07, + "learning_rate": 1.4123563174739036e-05, + "loss": 0.1916, + "step": 7332 + }, + { + "epoch": 3.07, + "learning_rate": 1.4122018559593457e-05, + "loss": 0.1645, + "step": 7333 + }, + { + "epoch": 3.07, + "learning_rate": 1.4120473825964112e-05, + "loss": 0.1863, + "step": 7334 + }, + { + "epoch": 3.07, + "learning_rate": 1.4118928973895406e-05, + "loss": 0.1731, + "step": 7335 + }, + { + "epoch": 3.07, + "learning_rate": 1.4117384003431738e-05, + "loss": 0.1718, + "step": 7336 + }, + { + "epoch": 3.07, + "learning_rate": 1.4115838914617518e-05, + "loss": 0.2038, + "step": 7337 + }, + { + "epoch": 3.07, + "learning_rate": 1.4114293707497168e-05, + "loss": 0.1613, + "step": 7338 + }, + { + "epoch": 3.07, + "learning_rate": 1.4112748382115091e-05, + "loss": 0.2105, + "step": 7339 + }, + { + "epoch": 3.07, + "learning_rate": 1.4111202938515711e-05, + "loss": 0.1805, + "step": 7340 + }, + { + "epoch": 3.07, + "learning_rate": 1.4109657376743455e-05, + "loss": 0.157, + "step": 7341 + }, + { + "epoch": 3.07, + "learning_rate": 1.4108111696842743e-05, + "loss": 0.2109, + "step": 7342 + }, + { + "epoch": 3.07, + "learning_rate": 1.4106565898858007e-05, + "loss": 0.1733, + "step": 7343 + }, + { + "epoch": 3.07, + "learning_rate": 1.4105019982833674e-05, + "loss": 0.1645, + "step": 7344 + }, + { + "epoch": 3.08, + "learning_rate": 1.4103473948814189e-05, + "loss": 0.1983, + "step": 7345 + }, + { + "epoch": 3.08, + "learning_rate": 1.4101927796843984e-05, + "loss": 0.1839, + "step": 7346 + }, + { + "epoch": 3.08, + "learning_rate": 1.4100381526967502e-05, + "loss": 0.2312, + "step": 7347 + }, + { + "epoch": 3.08, + "learning_rate": 1.4098835139229195e-05, + "loss": 0.1856, + "step": 7348 + }, + { + "epoch": 3.08, + "learning_rate": 1.409728863367351e-05, + "loss": 0.1912, + "step": 7349 + }, + { + "epoch": 3.08, + "learning_rate": 1.4095742010344898e-05, + "loss": 0.1981, + "step": 7350 + }, + { + "epoch": 3.08, + "learning_rate": 1.4094195269287814e-05, + "loss": 0.197, + "step": 7351 + }, + { + "epoch": 3.08, + "learning_rate": 1.4092648410546723e-05, + "loss": 0.1855, + "step": 7352 + }, + { + "epoch": 3.08, + "learning_rate": 1.4091101434166086e-05, + "loss": 0.1751, + "step": 7353 + }, + { + "epoch": 3.08, + "learning_rate": 1.4089554340190365e-05, + "loss": 0.1941, + "step": 7354 + }, + { + "epoch": 3.08, + "learning_rate": 1.4088007128664036e-05, + "loss": 0.1755, + "step": 7355 + }, + { + "epoch": 3.08, + "learning_rate": 1.4086459799631572e-05, + "loss": 0.1973, + "step": 7356 + }, + { + "epoch": 3.08, + "learning_rate": 1.4084912353137444e-05, + "loss": 0.1854, + "step": 7357 + }, + { + "epoch": 3.08, + "learning_rate": 1.4083364789226135e-05, + "loss": 0.2054, + "step": 7358 + }, + { + "epoch": 3.08, + "learning_rate": 1.4081817107942131e-05, + "loss": 0.1948, + "step": 7359 + }, + { + "epoch": 3.08, + "learning_rate": 1.4080269309329912e-05, + "loss": 0.157, + "step": 7360 + }, + { + "epoch": 3.08, + "learning_rate": 1.4078721393433977e-05, + "loss": 0.2108, + "step": 7361 + }, + { + "epoch": 3.08, + "learning_rate": 1.4077173360298812e-05, + "loss": 0.1796, + "step": 7362 + }, + { + "epoch": 3.08, + "learning_rate": 1.407562520996892e-05, + "loss": 0.2147, + "step": 7363 + }, + { + "epoch": 3.08, + "learning_rate": 1.4074076942488794e-05, + "loss": 0.2133, + "step": 7364 + }, + { + "epoch": 3.08, + "learning_rate": 1.4072528557902944e-05, + "loss": 0.1498, + "step": 7365 + }, + { + "epoch": 3.08, + "learning_rate": 1.4070980056255876e-05, + "loss": 0.2077, + "step": 7366 + }, + { + "epoch": 3.08, + "learning_rate": 1.4069431437592095e-05, + "loss": 0.2182, + "step": 7367 + }, + { + "epoch": 3.08, + "learning_rate": 1.4067882701956121e-05, + "loss": 0.2054, + "step": 7368 + }, + { + "epoch": 3.09, + "learning_rate": 1.406633384939247e-05, + "loss": 0.1672, + "step": 7369 + }, + { + "epoch": 3.09, + "learning_rate": 1.4064784879945658e-05, + "loss": 0.1746, + "step": 7370 + }, + { + "epoch": 3.09, + "learning_rate": 1.4063235793660212e-05, + "loss": 0.1901, + "step": 7371 + }, + { + "epoch": 3.09, + "learning_rate": 1.406168659058066e-05, + "loss": 0.1435, + "step": 7372 + }, + { + "epoch": 3.09, + "learning_rate": 1.4060137270751531e-05, + "loss": 0.2062, + "step": 7373 + }, + { + "epoch": 3.09, + "learning_rate": 1.4058587834217356e-05, + "loss": 0.1671, + "step": 7374 + }, + { + "epoch": 3.09, + "learning_rate": 1.405703828102268e-05, + "loss": 0.1781, + "step": 7375 + }, + { + "epoch": 3.09, + "learning_rate": 1.4055488611212038e-05, + "loss": 0.1978, + "step": 7376 + }, + { + "epoch": 3.09, + "learning_rate": 1.405393882482997e-05, + "loss": 0.1653, + "step": 7377 + }, + { + "epoch": 3.09, + "learning_rate": 1.4052388921921033e-05, + "loss": 0.1671, + "step": 7378 + }, + { + "epoch": 3.09, + "learning_rate": 1.405083890252977e-05, + "loss": 0.1926, + "step": 7379 + }, + { + "epoch": 3.09, + "learning_rate": 1.4049288766700739e-05, + "loss": 0.1676, + "step": 7380 + }, + { + "epoch": 3.09, + "learning_rate": 1.4047738514478493e-05, + "loss": 0.1606, + "step": 7381 + }, + { + "epoch": 3.09, + "learning_rate": 1.40461881459076e-05, + "loss": 0.1998, + "step": 7382 + }, + { + "epoch": 3.09, + "learning_rate": 1.4044637661032614e-05, + "loss": 0.2041, + "step": 7383 + }, + { + "epoch": 3.09, + "learning_rate": 1.404308705989811e-05, + "loss": 0.1909, + "step": 7384 + }, + { + "epoch": 3.09, + "learning_rate": 1.4041536342548656e-05, + "loss": 0.1531, + "step": 7385 + }, + { + "epoch": 3.09, + "learning_rate": 1.4039985509028825e-05, + "loss": 0.1778, + "step": 7386 + }, + { + "epoch": 3.09, + "learning_rate": 1.4038434559383198e-05, + "loss": 0.1942, + "step": 7387 + }, + { + "epoch": 3.09, + "learning_rate": 1.4036883493656353e-05, + "loss": 0.2008, + "step": 7388 + }, + { + "epoch": 3.09, + "learning_rate": 1.4035332311892873e-05, + "loss": 0.1685, + "step": 7389 + }, + { + "epoch": 3.09, + "learning_rate": 1.4033781014137348e-05, + "loss": 0.1667, + "step": 7390 + }, + { + "epoch": 3.09, + "learning_rate": 1.4032229600434363e-05, + "loss": 0.1642, + "step": 7391 + }, + { + "epoch": 3.09, + "learning_rate": 1.403067807082852e-05, + "loss": 0.169, + "step": 7392 + }, + { + "epoch": 3.1, + "learning_rate": 1.4029126425364414e-05, + "loss": 0.183, + "step": 7393 + }, + { + "epoch": 3.1, + "learning_rate": 1.4027574664086641e-05, + "loss": 0.1868, + "step": 7394 + }, + { + "epoch": 3.1, + "learning_rate": 1.402602278703981e-05, + "loss": 0.2027, + "step": 7395 + }, + { + "epoch": 3.1, + "learning_rate": 1.4024470794268528e-05, + "loss": 0.2004, + "step": 7396 + }, + { + "epoch": 3.1, + "learning_rate": 1.4022918685817399e-05, + "loss": 0.1987, + "step": 7397 + }, + { + "epoch": 3.1, + "learning_rate": 1.4021366461731047e-05, + "loss": 0.2168, + "step": 7398 + }, + { + "epoch": 3.1, + "learning_rate": 1.4019814122054082e-05, + "loss": 0.1792, + "step": 7399 + }, + { + "epoch": 3.1, + "learning_rate": 1.4018261666831132e-05, + "loss": 0.1826, + "step": 7400 + }, + { + "epoch": 3.1, + "learning_rate": 1.4016709096106813e-05, + "loss": 0.1979, + "step": 7401 + }, + { + "epoch": 3.1, + "learning_rate": 1.4015156409925754e-05, + "loss": 0.1863, + "step": 7402 + }, + { + "epoch": 3.1, + "learning_rate": 1.4013603608332589e-05, + "loss": 0.1953, + "step": 7403 + }, + { + "epoch": 3.1, + "learning_rate": 1.4012050691371948e-05, + "loss": 0.1579, + "step": 7404 + }, + { + "epoch": 3.1, + "learning_rate": 1.4010497659088468e-05, + "loss": 0.2044, + "step": 7405 + }, + { + "epoch": 3.1, + "learning_rate": 1.4008944511526798e-05, + "loss": 0.1969, + "step": 7406 + }, + { + "epoch": 3.1, + "learning_rate": 1.400739124873157e-05, + "loss": 0.1812, + "step": 7407 + }, + { + "epoch": 3.1, + "learning_rate": 1.4005837870747439e-05, + "loss": 0.2105, + "step": 7408 + }, + { + "epoch": 3.1, + "learning_rate": 1.4004284377619053e-05, + "loss": 0.2011, + "step": 7409 + }, + { + "epoch": 3.1, + "learning_rate": 1.4002730769391069e-05, + "loss": 0.1904, + "step": 7410 + }, + { + "epoch": 3.1, + "learning_rate": 1.4001177046108136e-05, + "loss": 0.1919, + "step": 7411 + }, + { + "epoch": 3.1, + "learning_rate": 1.3999623207814921e-05, + "loss": 0.1803, + "step": 7412 + }, + { + "epoch": 3.1, + "learning_rate": 1.3998069254556088e-05, + "loss": 0.1908, + "step": 7413 + }, + { + "epoch": 3.1, + "learning_rate": 1.3996515186376298e-05, + "loss": 0.22, + "step": 7414 + }, + { + "epoch": 3.1, + "learning_rate": 1.399496100332023e-05, + "loss": 0.174, + "step": 7415 + }, + { + "epoch": 3.1, + "learning_rate": 1.399340670543255e-05, + "loss": 0.2192, + "step": 7416 + }, + { + "epoch": 3.11, + "learning_rate": 1.3991852292757937e-05, + "loss": 0.1978, + "step": 7417 + }, + { + "epoch": 3.11, + "learning_rate": 1.3990297765341074e-05, + "loss": 0.1474, + "step": 7418 + }, + { + "epoch": 3.11, + "learning_rate": 1.3988743123226643e-05, + "loss": 0.1784, + "step": 7419 + }, + { + "epoch": 3.11, + "learning_rate": 1.398718836645933e-05, + "loss": 0.177, + "step": 7420 + }, + { + "epoch": 3.11, + "learning_rate": 1.3985633495083828e-05, + "loss": 0.2083, + "step": 7421 + }, + { + "epoch": 3.11, + "learning_rate": 1.3984078509144824e-05, + "loss": 0.2083, + "step": 7422 + }, + { + "epoch": 3.11, + "learning_rate": 1.398252340868702e-05, + "loss": 0.1653, + "step": 7423 + }, + { + "epoch": 3.11, + "learning_rate": 1.3980968193755115e-05, + "loss": 0.1788, + "step": 7424 + }, + { + "epoch": 3.11, + "learning_rate": 1.397941286439381e-05, + "loss": 0.17, + "step": 7425 + }, + { + "epoch": 3.11, + "learning_rate": 1.3977857420647816e-05, + "loss": 0.1635, + "step": 7426 + }, + { + "epoch": 3.11, + "learning_rate": 1.3976301862561841e-05, + "loss": 0.1704, + "step": 7427 + }, + { + "epoch": 3.11, + "learning_rate": 1.3974746190180595e-05, + "loss": 0.1724, + "step": 7428 + }, + { + "epoch": 3.11, + "learning_rate": 1.3973190403548798e-05, + "loss": 0.1812, + "step": 7429 + }, + { + "epoch": 3.11, + "learning_rate": 1.3971634502711169e-05, + "loss": 0.1525, + "step": 7430 + }, + { + "epoch": 3.11, + "learning_rate": 1.397007848771243e-05, + "loss": 0.1913, + "step": 7431 + }, + { + "epoch": 3.11, + "learning_rate": 1.3968522358597307e-05, + "loss": 0.1774, + "step": 7432 + }, + { + "epoch": 3.11, + "learning_rate": 1.3966966115410533e-05, + "loss": 0.1482, + "step": 7433 + }, + { + "epoch": 3.11, + "learning_rate": 1.3965409758196838e-05, + "loss": 0.2149, + "step": 7434 + }, + { + "epoch": 3.11, + "learning_rate": 1.3963853287000956e-05, + "loss": 0.1504, + "step": 7435 + }, + { + "epoch": 3.11, + "learning_rate": 1.396229670186763e-05, + "loss": 0.2362, + "step": 7436 + }, + { + "epoch": 3.11, + "learning_rate": 1.3960740002841602e-05, + "loss": 0.1958, + "step": 7437 + }, + { + "epoch": 3.11, + "learning_rate": 1.3959183189967615e-05, + "loss": 0.1642, + "step": 7438 + }, + { + "epoch": 3.11, + "learning_rate": 1.395762626329042e-05, + "loss": 0.1807, + "step": 7439 + }, + { + "epoch": 3.11, + "learning_rate": 1.3956069222854774e-05, + "loss": 0.1854, + "step": 7440 + }, + { + "epoch": 3.12, + "learning_rate": 1.3954512068705425e-05, + "loss": 0.2109, + "step": 7441 + }, + { + "epoch": 3.12, + "learning_rate": 1.3952954800887137e-05, + "loss": 0.1881, + "step": 7442 + }, + { + "epoch": 3.12, + "learning_rate": 1.3951397419444675e-05, + "loss": 0.1886, + "step": 7443 + }, + { + "epoch": 3.12, + "learning_rate": 1.3949839924422798e-05, + "loss": 0.1795, + "step": 7444 + }, + { + "epoch": 3.12, + "learning_rate": 1.3948282315866274e-05, + "loss": 0.2422, + "step": 7445 + }, + { + "epoch": 3.12, + "learning_rate": 1.394672459381988e-05, + "loss": 0.2021, + "step": 7446 + }, + { + "epoch": 3.12, + "learning_rate": 1.3945166758328394e-05, + "loss": 0.1772, + "step": 7447 + }, + { + "epoch": 3.12, + "learning_rate": 1.3943608809436587e-05, + "loss": 0.2052, + "step": 7448 + }, + { + "epoch": 3.12, + "learning_rate": 1.3942050747189242e-05, + "loss": 0.1767, + "step": 7449 + }, + { + "epoch": 3.12, + "learning_rate": 1.394049257163115e-05, + "loss": 0.1822, + "step": 7450 + }, + { + "epoch": 3.12, + "learning_rate": 1.3938934282807095e-05, + "loss": 0.1777, + "step": 7451 + }, + { + "epoch": 3.12, + "learning_rate": 1.3937375880761868e-05, + "loss": 0.1736, + "step": 7452 + }, + { + "epoch": 3.12, + "learning_rate": 1.3935817365540268e-05, + "loss": 0.1833, + "step": 7453 + }, + { + "epoch": 3.12, + "learning_rate": 1.3934258737187087e-05, + "loss": 0.2147, + "step": 7454 + }, + { + "epoch": 3.12, + "learning_rate": 1.3932699995747134e-05, + "loss": 0.1847, + "step": 7455 + }, + { + "epoch": 3.12, + "learning_rate": 1.3931141141265206e-05, + "loss": 0.1931, + "step": 7456 + }, + { + "epoch": 3.12, + "learning_rate": 1.3929582173786116e-05, + "loss": 0.15, + "step": 7457 + }, + { + "epoch": 3.12, + "learning_rate": 1.3928023093354671e-05, + "loss": 0.2478, + "step": 7458 + }, + { + "epoch": 3.12, + "learning_rate": 1.392646390001569e-05, + "loss": 0.158, + "step": 7459 + }, + { + "epoch": 3.12, + "learning_rate": 1.3924904593813986e-05, + "loss": 0.1903, + "step": 7460 + }, + { + "epoch": 3.12, + "learning_rate": 1.3923345174794386e-05, + "loss": 0.2119, + "step": 7461 + }, + { + "epoch": 3.12, + "learning_rate": 1.3921785643001708e-05, + "loss": 0.165, + "step": 7462 + }, + { + "epoch": 3.12, + "learning_rate": 1.3920225998480782e-05, + "loss": 0.1985, + "step": 7463 + }, + { + "epoch": 3.12, + "learning_rate": 1.3918666241276442e-05, + "loss": 0.1734, + "step": 7464 + }, + { + "epoch": 3.13, + "learning_rate": 1.3917106371433513e-05, + "loss": 0.202, + "step": 7465 + }, + { + "epoch": 3.13, + "learning_rate": 1.3915546388996838e-05, + "loss": 0.1719, + "step": 7466 + }, + { + "epoch": 3.13, + "learning_rate": 1.3913986294011261e-05, + "loss": 0.1945, + "step": 7467 + }, + { + "epoch": 3.13, + "learning_rate": 1.391242608652162e-05, + "loss": 0.1596, + "step": 7468 + }, + { + "epoch": 3.13, + "learning_rate": 1.391086576657276e-05, + "loss": 0.2132, + "step": 7469 + }, + { + "epoch": 3.13, + "learning_rate": 1.3909305334209536e-05, + "loss": 0.1928, + "step": 7470 + }, + { + "epoch": 3.13, + "learning_rate": 1.3907744789476799e-05, + "loss": 0.1837, + "step": 7471 + }, + { + "epoch": 3.13, + "learning_rate": 1.3906184132419405e-05, + "loss": 0.1827, + "step": 7472 + }, + { + "epoch": 3.13, + "learning_rate": 1.3904623363082213e-05, + "loss": 0.192, + "step": 7473 + }, + { + "epoch": 3.13, + "learning_rate": 1.390306248151009e-05, + "loss": 0.1669, + "step": 7474 + }, + { + "epoch": 3.13, + "learning_rate": 1.3901501487747897e-05, + "loss": 0.163, + "step": 7475 + }, + { + "epoch": 3.13, + "learning_rate": 1.389994038184051e-05, + "loss": 0.1953, + "step": 7476 + }, + { + "epoch": 3.13, + "learning_rate": 1.3898379163832794e-05, + "loss": 0.1866, + "step": 7477 + }, + { + "epoch": 3.13, + "learning_rate": 1.3896817833769631e-05, + "loss": 0.1751, + "step": 7478 + }, + { + "epoch": 3.13, + "learning_rate": 1.3895256391695896e-05, + "loss": 0.1909, + "step": 7479 + }, + { + "epoch": 3.13, + "learning_rate": 1.3893694837656471e-05, + "loss": 0.2215, + "step": 7480 + }, + { + "epoch": 3.13, + "learning_rate": 1.3892133171696246e-05, + "loss": 0.1652, + "step": 7481 + }, + { + "epoch": 3.13, + "learning_rate": 1.3890571393860101e-05, + "loss": 0.1398, + "step": 7482 + }, + { + "epoch": 3.13, + "learning_rate": 1.3889009504192938e-05, + "loss": 0.1765, + "step": 7483 + }, + { + "epoch": 3.13, + "learning_rate": 1.3887447502739649e-05, + "loss": 0.1829, + "step": 7484 + }, + { + "epoch": 3.13, + "learning_rate": 1.388588538954513e-05, + "loss": 0.2005, + "step": 7485 + }, + { + "epoch": 3.13, + "learning_rate": 1.3884323164654284e-05, + "loss": 0.1762, + "step": 7486 + }, + { + "epoch": 3.13, + "learning_rate": 1.3882760828112014e-05, + "loss": 0.2078, + "step": 7487 + }, + { + "epoch": 3.14, + "learning_rate": 1.388119837996323e-05, + "loss": 0.1795, + "step": 7488 + }, + { + "epoch": 3.14, + "learning_rate": 1.3879635820252846e-05, + "loss": 0.1307, + "step": 7489 + }, + { + "epoch": 3.14, + "learning_rate": 1.3878073149025768e-05, + "loss": 0.176, + "step": 7490 + }, + { + "epoch": 3.14, + "learning_rate": 1.3876510366326922e-05, + "loss": 0.207, + "step": 7491 + }, + { + "epoch": 3.14, + "learning_rate": 1.3874947472201226e-05, + "loss": 0.1517, + "step": 7492 + }, + { + "epoch": 3.14, + "learning_rate": 1.38733844666936e-05, + "loss": 0.1875, + "step": 7493 + }, + { + "epoch": 3.14, + "learning_rate": 1.3871821349848977e-05, + "loss": 0.1754, + "step": 7494 + }, + { + "epoch": 3.14, + "learning_rate": 1.3870258121712285e-05, + "loss": 0.1777, + "step": 7495 + }, + { + "epoch": 3.14, + "learning_rate": 1.3868694782328456e-05, + "loss": 0.2025, + "step": 7496 + }, + { + "epoch": 3.14, + "learning_rate": 1.3867131331742429e-05, + "loss": 0.1829, + "step": 7497 + }, + { + "epoch": 3.14, + "learning_rate": 1.3865567769999146e-05, + "loss": 0.176, + "step": 7498 + }, + { + "epoch": 3.14, + "learning_rate": 1.3864004097143546e-05, + "loss": 0.1477, + "step": 7499 + }, + { + "epoch": 3.14, + "learning_rate": 1.3862440313220575e-05, + "loss": 0.1657, + "step": 7500 + }, + { + "epoch": 3.14, + "learning_rate": 1.3860876418275189e-05, + "loss": 0.1488, + "step": 7501 + }, + { + "epoch": 3.14, + "learning_rate": 1.3859312412352335e-05, + "loss": 0.193, + "step": 7502 + }, + { + "epoch": 3.14, + "learning_rate": 1.3857748295496972e-05, + "loss": 0.1975, + "step": 7503 + }, + { + "epoch": 3.14, + "learning_rate": 1.3856184067754053e-05, + "loss": 0.1794, + "step": 7504 + }, + { + "epoch": 3.14, + "learning_rate": 1.3854619729168548e-05, + "loss": 0.161, + "step": 7505 + }, + { + "epoch": 3.14, + "learning_rate": 1.3853055279785416e-05, + "loss": 0.1941, + "step": 7506 + }, + { + "epoch": 3.14, + "learning_rate": 1.3851490719649631e-05, + "loss": 0.2179, + "step": 7507 + }, + { + "epoch": 3.14, + "learning_rate": 1.3849926048806166e-05, + "loss": 0.1785, + "step": 7508 + }, + { + "epoch": 3.14, + "learning_rate": 1.384836126729999e-05, + "loss": 0.2004, + "step": 7509 + }, + { + "epoch": 3.14, + "learning_rate": 1.3846796375176083e-05, + "loss": 0.2078, + "step": 7510 + }, + { + "epoch": 3.14, + "learning_rate": 1.3845231372479432e-05, + "loss": 0.1929, + "step": 7511 + }, + { + "epoch": 3.15, + "learning_rate": 1.3843666259255018e-05, + "loss": 0.1956, + "step": 7512 + }, + { + "epoch": 3.15, + "learning_rate": 1.3842101035547827e-05, + "loss": 0.2089, + "step": 7513 + }, + { + "epoch": 3.15, + "learning_rate": 1.3840535701402847e-05, + "loss": 0.184, + "step": 7514 + }, + { + "epoch": 3.15, + "learning_rate": 1.3838970256865078e-05, + "loss": 0.1952, + "step": 7515 + }, + { + "epoch": 3.15, + "learning_rate": 1.3837404701979517e-05, + "loss": 0.1959, + "step": 7516 + }, + { + "epoch": 3.15, + "learning_rate": 1.3835839036791163e-05, + "loss": 0.2267, + "step": 7517 + }, + { + "epoch": 3.15, + "learning_rate": 1.383427326134502e-05, + "loss": 0.2016, + "step": 7518 + }, + { + "epoch": 3.15, + "learning_rate": 1.3832707375686095e-05, + "loss": 0.1755, + "step": 7519 + }, + { + "epoch": 3.15, + "learning_rate": 1.3831141379859394e-05, + "loss": 0.2223, + "step": 7520 + }, + { + "epoch": 3.15, + "learning_rate": 1.3829575273909939e-05, + "loss": 0.1427, + "step": 7521 + }, + { + "epoch": 3.15, + "learning_rate": 1.3828009057882737e-05, + "loss": 0.1875, + "step": 7522 + }, + { + "epoch": 3.15, + "learning_rate": 1.3826442731822814e-05, + "loss": 0.1887, + "step": 7523 + }, + { + "epoch": 3.15, + "learning_rate": 1.3824876295775187e-05, + "loss": 0.1875, + "step": 7524 + }, + { + "epoch": 3.15, + "learning_rate": 1.3823309749784888e-05, + "loss": 0.1904, + "step": 7525 + }, + { + "epoch": 3.15, + "learning_rate": 1.3821743093896942e-05, + "loss": 0.1796, + "step": 7526 + }, + { + "epoch": 3.15, + "learning_rate": 1.3820176328156381e-05, + "loss": 0.1889, + "step": 7527 + }, + { + "epoch": 3.15, + "learning_rate": 1.3818609452608245e-05, + "loss": 0.1661, + "step": 7528 + }, + { + "epoch": 3.15, + "learning_rate": 1.3817042467297565e-05, + "loss": 0.173, + "step": 7529 + }, + { + "epoch": 3.15, + "learning_rate": 1.3815475372269386e-05, + "loss": 0.1865, + "step": 7530 + }, + { + "epoch": 3.15, + "learning_rate": 1.3813908167568757e-05, + "loss": 0.1942, + "step": 7531 + }, + { + "epoch": 3.15, + "learning_rate": 1.381234085324072e-05, + "loss": 0.2108, + "step": 7532 + }, + { + "epoch": 3.15, + "learning_rate": 1.381077342933033e-05, + "loss": 0.1843, + "step": 7533 + }, + { + "epoch": 3.15, + "learning_rate": 1.3809205895882635e-05, + "loss": 0.2048, + "step": 7534 + }, + { + "epoch": 3.15, + "learning_rate": 1.3807638252942704e-05, + "loss": 0.1397, + "step": 7535 + }, + { + "epoch": 3.16, + "learning_rate": 1.3806070500555583e-05, + "loss": 0.2272, + "step": 7536 + }, + { + "epoch": 3.16, + "learning_rate": 1.3804502638766346e-05, + "loss": 0.1862, + "step": 7537 + }, + { + "epoch": 3.16, + "learning_rate": 1.3802934667620055e-05, + "loss": 0.1913, + "step": 7538 + }, + { + "epoch": 3.16, + "learning_rate": 1.3801366587161786e-05, + "loss": 0.1688, + "step": 7539 + }, + { + "epoch": 3.16, + "learning_rate": 1.3799798397436603e-05, + "loss": 0.2065, + "step": 7540 + }, + { + "epoch": 3.16, + "learning_rate": 1.379823009848959e-05, + "loss": 0.1557, + "step": 7541 + }, + { + "epoch": 3.16, + "learning_rate": 1.3796661690365823e-05, + "loss": 0.1426, + "step": 7542 + }, + { + "epoch": 3.16, + "learning_rate": 1.379509317311038e-05, + "loss": 0.2024, + "step": 7543 + }, + { + "epoch": 3.16, + "learning_rate": 1.3793524546768358e-05, + "loss": 0.2058, + "step": 7544 + }, + { + "epoch": 3.16, + "learning_rate": 1.3791955811384837e-05, + "loss": 0.1913, + "step": 7545 + }, + { + "epoch": 3.16, + "learning_rate": 1.3790386967004911e-05, + "loss": 0.1702, + "step": 7546 + }, + { + "epoch": 3.16, + "learning_rate": 1.3788818013673675e-05, + "loss": 0.1924, + "step": 7547 + }, + { + "epoch": 3.16, + "learning_rate": 1.3787248951436225e-05, + "loss": 0.1676, + "step": 7548 + }, + { + "epoch": 3.16, + "learning_rate": 1.3785679780337668e-05, + "loss": 0.1956, + "step": 7549 + }, + { + "epoch": 3.16, + "learning_rate": 1.3784110500423104e-05, + "loss": 0.2423, + "step": 7550 + }, + { + "epoch": 3.16, + "learning_rate": 1.3782541111737642e-05, + "loss": 0.2112, + "step": 7551 + }, + { + "epoch": 3.16, + "learning_rate": 1.3780971614326393e-05, + "loss": 0.1864, + "step": 7552 + }, + { + "epoch": 3.16, + "learning_rate": 1.3779402008234467e-05, + "loss": 0.2137, + "step": 7553 + }, + { + "epoch": 3.16, + "learning_rate": 1.3777832293506985e-05, + "loss": 0.211, + "step": 7554 + }, + { + "epoch": 3.16, + "learning_rate": 1.377626247018907e-05, + "loss": 0.1677, + "step": 7555 + }, + { + "epoch": 3.16, + "learning_rate": 1.3774692538325835e-05, + "loss": 0.2024, + "step": 7556 + }, + { + "epoch": 3.16, + "learning_rate": 1.3773122497962418e-05, + "loss": 0.1875, + "step": 7557 + }, + { + "epoch": 3.16, + "learning_rate": 1.3771552349143942e-05, + "loss": 0.2106, + "step": 7558 + }, + { + "epoch": 3.16, + "learning_rate": 1.3769982091915539e-05, + "loss": 0.1844, + "step": 7559 + }, + { + "epoch": 3.17, + "learning_rate": 1.3768411726322347e-05, + "loss": 0.241, + "step": 7560 + }, + { + "epoch": 3.17, + "learning_rate": 1.3766841252409502e-05, + "loss": 0.1707, + "step": 7561 + }, + { + "epoch": 3.17, + "learning_rate": 1.3765270670222153e-05, + "loss": 0.1793, + "step": 7562 + }, + { + "epoch": 3.17, + "learning_rate": 1.3763699979805434e-05, + "loss": 0.1644, + "step": 7563 + }, + { + "epoch": 3.17, + "learning_rate": 1.3762129181204502e-05, + "loss": 0.2117, + "step": 7564 + }, + { + "epoch": 3.17, + "learning_rate": 1.3760558274464506e-05, + "loss": 0.1701, + "step": 7565 + }, + { + "epoch": 3.17, + "learning_rate": 1.3758987259630597e-05, + "loss": 0.185, + "step": 7566 + }, + { + "epoch": 3.17, + "learning_rate": 1.3757416136747935e-05, + "loss": 0.1732, + "step": 7567 + }, + { + "epoch": 3.17, + "learning_rate": 1.375584490586168e-05, + "loss": 0.218, + "step": 7568 + }, + { + "epoch": 3.17, + "learning_rate": 1.3754273567017e-05, + "loss": 0.1982, + "step": 7569 + }, + { + "epoch": 3.17, + "learning_rate": 1.3752702120259055e-05, + "loss": 0.2061, + "step": 7570 + }, + { + "epoch": 3.17, + "learning_rate": 1.3751130565633015e-05, + "loss": 0.1819, + "step": 7571 + }, + { + "epoch": 3.17, + "learning_rate": 1.3749558903184059e-05, + "loss": 0.1826, + "step": 7572 + }, + { + "epoch": 3.17, + "learning_rate": 1.3747987132957361e-05, + "loss": 0.1503, + "step": 7573 + }, + { + "epoch": 3.17, + "learning_rate": 1.3746415254998095e-05, + "loss": 0.1801, + "step": 7574 + }, + { + "epoch": 3.17, + "learning_rate": 1.3744843269351446e-05, + "loss": 0.1277, + "step": 7575 + }, + { + "epoch": 3.17, + "learning_rate": 1.3743271176062605e-05, + "loss": 0.1731, + "step": 7576 + }, + { + "epoch": 3.17, + "learning_rate": 1.3741698975176754e-05, + "loss": 0.2044, + "step": 7577 + }, + { + "epoch": 3.17, + "learning_rate": 1.3740126666739086e-05, + "loss": 0.1722, + "step": 7578 + }, + { + "epoch": 3.17, + "learning_rate": 1.3738554250794795e-05, + "loss": 0.1917, + "step": 7579 + }, + { + "epoch": 3.17, + "learning_rate": 1.3736981727389083e-05, + "loss": 0.2002, + "step": 7580 + }, + { + "epoch": 3.17, + "learning_rate": 1.3735409096567145e-05, + "loss": 0.1809, + "step": 7581 + }, + { + "epoch": 3.17, + "learning_rate": 1.3733836358374186e-05, + "loss": 0.1463, + "step": 7582 + }, + { + "epoch": 3.17, + "learning_rate": 1.3732263512855416e-05, + "loss": 0.1965, + "step": 7583 + }, + { + "epoch": 3.18, + "learning_rate": 1.3730690560056043e-05, + "loss": 0.1524, + "step": 7584 + }, + { + "epoch": 3.18, + "learning_rate": 1.3729117500021279e-05, + "loss": 0.1706, + "step": 7585 + }, + { + "epoch": 3.18, + "learning_rate": 1.3727544332796344e-05, + "loss": 0.1952, + "step": 7586 + }, + { + "epoch": 3.18, + "learning_rate": 1.3725971058426453e-05, + "loss": 0.1621, + "step": 7587 + }, + { + "epoch": 3.18, + "learning_rate": 1.3724397676956831e-05, + "loss": 0.2063, + "step": 7588 + }, + { + "epoch": 3.18, + "learning_rate": 1.3722824188432702e-05, + "loss": 0.2065, + "step": 7589 + }, + { + "epoch": 3.18, + "learning_rate": 1.3721250592899295e-05, + "loss": 0.2085, + "step": 7590 + }, + { + "epoch": 3.18, + "learning_rate": 1.3719676890401845e-05, + "loss": 0.1861, + "step": 7591 + }, + { + "epoch": 3.18, + "learning_rate": 1.371810308098558e-05, + "loss": 0.1507, + "step": 7592 + }, + { + "epoch": 3.18, + "learning_rate": 1.3716529164695744e-05, + "loss": 0.1772, + "step": 7593 + }, + { + "epoch": 3.18, + "learning_rate": 1.3714955141577573e-05, + "loss": 0.2075, + "step": 7594 + }, + { + "epoch": 3.18, + "learning_rate": 1.3713381011676315e-05, + "loss": 0.2258, + "step": 7595 + }, + { + "epoch": 3.18, + "learning_rate": 1.3711806775037215e-05, + "loss": 0.2036, + "step": 7596 + }, + { + "epoch": 3.18, + "learning_rate": 1.3710232431705521e-05, + "loss": 0.1845, + "step": 7597 + }, + { + "epoch": 3.18, + "learning_rate": 1.3708657981726487e-05, + "loss": 0.1761, + "step": 7598 + }, + { + "epoch": 3.18, + "learning_rate": 1.3707083425145375e-05, + "loss": 0.1685, + "step": 7599 + }, + { + "epoch": 3.18, + "learning_rate": 1.3705508762007436e-05, + "loss": 0.1957, + "step": 7600 + }, + { + "epoch": 3.18, + "learning_rate": 1.3703933992357936e-05, + "loss": 0.1601, + "step": 7601 + }, + { + "epoch": 3.18, + "learning_rate": 1.370235911624214e-05, + "loss": 0.1924, + "step": 7602 + }, + { + "epoch": 3.18, + "learning_rate": 1.370078413370532e-05, + "loss": 0.1813, + "step": 7603 + }, + { + "epoch": 3.18, + "learning_rate": 1.3699209044792743e-05, + "loss": 0.2021, + "step": 7604 + }, + { + "epoch": 3.18, + "learning_rate": 1.3697633849549683e-05, + "loss": 0.1347, + "step": 7605 + }, + { + "epoch": 3.18, + "learning_rate": 1.369605854802142e-05, + "loss": 0.1988, + "step": 7606 + }, + { + "epoch": 3.18, + "learning_rate": 1.3694483140253234e-05, + "loss": 0.1635, + "step": 7607 + }, + { + "epoch": 3.19, + "learning_rate": 1.3692907626290406e-05, + "loss": 0.1745, + "step": 7608 + }, + { + "epoch": 3.19, + "learning_rate": 1.3691332006178226e-05, + "loss": 0.1576, + "step": 7609 + }, + { + "epoch": 3.19, + "learning_rate": 1.3689756279961986e-05, + "loss": 0.1684, + "step": 7610 + }, + { + "epoch": 3.19, + "learning_rate": 1.3688180447686975e-05, + "loss": 0.1723, + "step": 7611 + }, + { + "epoch": 3.19, + "learning_rate": 1.3686604509398489e-05, + "loss": 0.1553, + "step": 7612 + }, + { + "epoch": 3.19, + "learning_rate": 1.3685028465141828e-05, + "loss": 0.1632, + "step": 7613 + }, + { + "epoch": 3.19, + "learning_rate": 1.3683452314962295e-05, + "loss": 0.1662, + "step": 7614 + }, + { + "epoch": 3.19, + "learning_rate": 1.3681876058905193e-05, + "loss": 0.2028, + "step": 7615 + }, + { + "epoch": 3.19, + "learning_rate": 1.368029969701583e-05, + "loss": 0.2133, + "step": 7616 + }, + { + "epoch": 3.19, + "learning_rate": 1.3678723229339518e-05, + "loss": 0.1861, + "step": 7617 + }, + { + "epoch": 3.19, + "learning_rate": 1.3677146655921573e-05, + "loss": 0.2252, + "step": 7618 + }, + { + "epoch": 3.19, + "learning_rate": 1.3675569976807307e-05, + "loss": 0.1522, + "step": 7619 + }, + { + "epoch": 3.19, + "learning_rate": 1.3673993192042048e-05, + "loss": 0.2132, + "step": 7620 + }, + { + "epoch": 3.19, + "learning_rate": 1.3672416301671111e-05, + "loss": 0.1779, + "step": 7621 + }, + { + "epoch": 3.19, + "learning_rate": 1.3670839305739828e-05, + "loss": 0.1832, + "step": 7622 + }, + { + "epoch": 3.19, + "learning_rate": 1.366926220429353e-05, + "loss": 0.1864, + "step": 7623 + }, + { + "epoch": 3.19, + "learning_rate": 1.3667684997377542e-05, + "loss": 0.2032, + "step": 7624 + }, + { + "epoch": 3.19, + "learning_rate": 1.3666107685037204e-05, + "loss": 0.1911, + "step": 7625 + }, + { + "epoch": 3.19, + "learning_rate": 1.3664530267317855e-05, + "loss": 0.1678, + "step": 7626 + }, + { + "epoch": 3.19, + "learning_rate": 1.3662952744264835e-05, + "loss": 0.1906, + "step": 7627 + }, + { + "epoch": 3.19, + "learning_rate": 1.3661375115923487e-05, + "loss": 0.2059, + "step": 7628 + }, + { + "epoch": 3.19, + "learning_rate": 1.3659797382339162e-05, + "loss": 0.1447, + "step": 7629 + }, + { + "epoch": 3.19, + "learning_rate": 1.3658219543557208e-05, + "loss": 0.1612, + "step": 7630 + }, + { + "epoch": 3.19, + "learning_rate": 1.365664159962298e-05, + "loss": 0.1815, + "step": 7631 + }, + { + "epoch": 3.2, + "learning_rate": 1.3655063550581834e-05, + "loss": 0.1874, + "step": 7632 + }, + { + "epoch": 3.2, + "learning_rate": 1.3653485396479129e-05, + "loss": 0.1905, + "step": 7633 + }, + { + "epoch": 3.2, + "learning_rate": 1.365190713736023e-05, + "loss": 0.1837, + "step": 7634 + }, + { + "epoch": 3.2, + "learning_rate": 1.3650328773270498e-05, + "loss": 0.2112, + "step": 7635 + }, + { + "epoch": 3.2, + "learning_rate": 1.3648750304255307e-05, + "loss": 0.1932, + "step": 7636 + }, + { + "epoch": 3.2, + "learning_rate": 1.3647171730360026e-05, + "loss": 0.134, + "step": 7637 + }, + { + "epoch": 3.2, + "learning_rate": 1.3645593051630032e-05, + "loss": 0.18, + "step": 7638 + }, + { + "epoch": 3.2, + "learning_rate": 1.3644014268110697e-05, + "loss": 0.1485, + "step": 7639 + }, + { + "epoch": 3.2, + "learning_rate": 1.364243537984741e-05, + "loss": 0.1891, + "step": 7640 + }, + { + "epoch": 3.2, + "learning_rate": 1.3640856386885546e-05, + "loss": 0.202, + "step": 7641 + }, + { + "epoch": 3.2, + "learning_rate": 1.3639277289270497e-05, + "loss": 0.1828, + "step": 7642 + }, + { + "epoch": 3.2, + "learning_rate": 1.3637698087047651e-05, + "loss": 0.149, + "step": 7643 + }, + { + "epoch": 3.2, + "learning_rate": 1.3636118780262405e-05, + "loss": 0.1346, + "step": 7644 + }, + { + "epoch": 3.2, + "learning_rate": 1.3634539368960147e-05, + "loss": 0.1852, + "step": 7645 + }, + { + "epoch": 3.2, + "learning_rate": 1.363295985318628e-05, + "loss": 0.1923, + "step": 7646 + }, + { + "epoch": 3.2, + "learning_rate": 1.363138023298621e-05, + "loss": 0.182, + "step": 7647 + }, + { + "epoch": 3.2, + "learning_rate": 1.3629800508405338e-05, + "loss": 0.1883, + "step": 7648 + }, + { + "epoch": 3.2, + "learning_rate": 1.3628220679489069e-05, + "loss": 0.2061, + "step": 7649 + }, + { + "epoch": 3.2, + "learning_rate": 1.3626640746282815e-05, + "loss": 0.2, + "step": 7650 + }, + { + "epoch": 3.2, + "learning_rate": 1.3625060708831992e-05, + "loss": 0.1932, + "step": 7651 + }, + { + "epoch": 3.2, + "learning_rate": 1.3623480567182018e-05, + "loss": 0.154, + "step": 7652 + }, + { + "epoch": 3.2, + "learning_rate": 1.3621900321378308e-05, + "loss": 0.1829, + "step": 7653 + }, + { + "epoch": 3.2, + "learning_rate": 1.3620319971466288e-05, + "loss": 0.2042, + "step": 7654 + }, + { + "epoch": 3.2, + "learning_rate": 1.3618739517491381e-05, + "loss": 0.2031, + "step": 7655 + }, + { + "epoch": 3.21, + "learning_rate": 1.361715895949902e-05, + "loss": 0.1759, + "step": 7656 + }, + { + "epoch": 3.21, + "learning_rate": 1.3615578297534635e-05, + "loss": 0.2407, + "step": 7657 + }, + { + "epoch": 3.21, + "learning_rate": 1.3613997531643657e-05, + "loss": 0.1647, + "step": 7658 + }, + { + "epoch": 3.21, + "learning_rate": 1.3612416661871532e-05, + "loss": 0.2102, + "step": 7659 + }, + { + "epoch": 3.21, + "learning_rate": 1.361083568826369e-05, + "loss": 0.1727, + "step": 7660 + }, + { + "epoch": 3.21, + "learning_rate": 1.3609254610865587e-05, + "loss": 0.2066, + "step": 7661 + }, + { + "epoch": 3.21, + "learning_rate": 1.3607673429722658e-05, + "loss": 0.1782, + "step": 7662 + }, + { + "epoch": 3.21, + "learning_rate": 1.3606092144880359e-05, + "loss": 0.1969, + "step": 7663 + }, + { + "epoch": 3.21, + "learning_rate": 1.3604510756384141e-05, + "loss": 0.1575, + "step": 7664 + }, + { + "epoch": 3.21, + "learning_rate": 1.360292926427946e-05, + "loss": 0.1964, + "step": 7665 + }, + { + "epoch": 3.21, + "learning_rate": 1.3601347668611772e-05, + "loss": 0.1888, + "step": 7666 + }, + { + "epoch": 3.21, + "learning_rate": 1.3599765969426546e-05, + "loss": 0.1649, + "step": 7667 + }, + { + "epoch": 3.21, + "learning_rate": 1.359818416676924e-05, + "loss": 0.1653, + "step": 7668 + }, + { + "epoch": 3.21, + "learning_rate": 1.3596602260685321e-05, + "loss": 0.1853, + "step": 7669 + }, + { + "epoch": 3.21, + "learning_rate": 1.3595020251220263e-05, + "loss": 0.1915, + "step": 7670 + }, + { + "epoch": 3.21, + "learning_rate": 1.3593438138419539e-05, + "loss": 0.1966, + "step": 7671 + }, + { + "epoch": 3.21, + "learning_rate": 1.3591855922328624e-05, + "loss": 0.2002, + "step": 7672 + }, + { + "epoch": 3.21, + "learning_rate": 1.3590273602992996e-05, + "loss": 0.1996, + "step": 7673 + }, + { + "epoch": 3.21, + "learning_rate": 1.3588691180458143e-05, + "loss": 0.1883, + "step": 7674 + }, + { + "epoch": 3.21, + "learning_rate": 1.3587108654769542e-05, + "loss": 0.1751, + "step": 7675 + }, + { + "epoch": 3.21, + "learning_rate": 1.3585526025972688e-05, + "loss": 0.1636, + "step": 7676 + }, + { + "epoch": 3.21, + "learning_rate": 1.3583943294113071e-05, + "loss": 0.2001, + "step": 7677 + }, + { + "epoch": 3.21, + "learning_rate": 1.3582360459236183e-05, + "loss": 0.1908, + "step": 7678 + }, + { + "epoch": 3.21, + "learning_rate": 1.3580777521387523e-05, + "loss": 0.1748, + "step": 7679 + }, + { + "epoch": 3.22, + "learning_rate": 1.357919448061259e-05, + "loss": 0.1909, + "step": 7680 + }, + { + "epoch": 3.22, + "learning_rate": 1.3577611336956888e-05, + "loss": 0.1817, + "step": 7681 + }, + { + "epoch": 3.22, + "learning_rate": 1.3576028090465924e-05, + "loss": 0.1583, + "step": 7682 + }, + { + "epoch": 3.22, + "learning_rate": 1.3574444741185207e-05, + "loss": 0.1666, + "step": 7683 + }, + { + "epoch": 3.22, + "learning_rate": 1.3572861289160245e-05, + "loss": 0.2047, + "step": 7684 + }, + { + "epoch": 3.22, + "learning_rate": 1.3571277734436556e-05, + "loss": 0.1697, + "step": 7685 + }, + { + "epoch": 3.22, + "learning_rate": 1.356969407705966e-05, + "loss": 0.1998, + "step": 7686 + }, + { + "epoch": 3.22, + "learning_rate": 1.3568110317075071e-05, + "loss": 0.1905, + "step": 7687 + }, + { + "epoch": 3.22, + "learning_rate": 1.356652645452832e-05, + "loss": 0.1819, + "step": 7688 + }, + { + "epoch": 3.22, + "learning_rate": 1.3564942489464932e-05, + "loss": 0.2108, + "step": 7689 + }, + { + "epoch": 3.22, + "learning_rate": 1.3563358421930433e-05, + "loss": 0.2004, + "step": 7690 + }, + { + "epoch": 3.22, + "learning_rate": 1.356177425197036e-05, + "loss": 0.178, + "step": 7691 + }, + { + "epoch": 3.22, + "learning_rate": 1.3560189979630247e-05, + "loss": 0.2014, + "step": 7692 + }, + { + "epoch": 3.22, + "learning_rate": 1.3558605604955633e-05, + "loss": 0.1777, + "step": 7693 + }, + { + "epoch": 3.22, + "learning_rate": 1.3557021127992055e-05, + "loss": 0.172, + "step": 7694 + }, + { + "epoch": 3.22, + "learning_rate": 1.3555436548785066e-05, + "loss": 0.1939, + "step": 7695 + }, + { + "epoch": 3.22, + "learning_rate": 1.3553851867380205e-05, + "loss": 0.1617, + "step": 7696 + }, + { + "epoch": 3.22, + "learning_rate": 1.3552267083823028e-05, + "loss": 0.1244, + "step": 7697 + }, + { + "epoch": 3.22, + "learning_rate": 1.3550682198159084e-05, + "loss": 0.1537, + "step": 7698 + }, + { + "epoch": 3.22, + "learning_rate": 1.3549097210433932e-05, + "loss": 0.1861, + "step": 7699 + }, + { + "epoch": 3.22, + "learning_rate": 1.354751212069313e-05, + "loss": 0.2075, + "step": 7700 + }, + { + "epoch": 3.22, + "learning_rate": 1.3545926928982241e-05, + "loss": 0.1893, + "step": 7701 + }, + { + "epoch": 3.22, + "learning_rate": 1.3544341635346827e-05, + "loss": 0.215, + "step": 7702 + }, + { + "epoch": 3.23, + "learning_rate": 1.3542756239832458e-05, + "loss": 0.1982, + "step": 7703 + }, + { + "epoch": 3.23, + "learning_rate": 1.3541170742484705e-05, + "loss": 0.2129, + "step": 7704 + }, + { + "epoch": 3.23, + "learning_rate": 1.3539585143349146e-05, + "loss": 0.1981, + "step": 7705 + }, + { + "epoch": 3.23, + "learning_rate": 1.3537999442471349e-05, + "loss": 0.1974, + "step": 7706 + }, + { + "epoch": 3.23, + "learning_rate": 1.3536413639896895e-05, + "loss": 0.176, + "step": 7707 + }, + { + "epoch": 3.23, + "learning_rate": 1.3534827735671373e-05, + "loss": 0.2061, + "step": 7708 + }, + { + "epoch": 3.23, + "learning_rate": 1.3533241729840362e-05, + "loss": 0.1642, + "step": 7709 + }, + { + "epoch": 3.23, + "learning_rate": 1.3531655622449454e-05, + "loss": 0.1684, + "step": 7710 + }, + { + "epoch": 3.23, + "learning_rate": 1.3530069413544241e-05, + "loss": 0.1948, + "step": 7711 + }, + { + "epoch": 3.23, + "learning_rate": 1.3528483103170312e-05, + "loss": 0.1634, + "step": 7712 + }, + { + "epoch": 3.23, + "learning_rate": 1.3526896691373269e-05, + "loss": 0.1526, + "step": 7713 + }, + { + "epoch": 3.23, + "learning_rate": 1.3525310178198707e-05, + "loss": 0.2019, + "step": 7714 + }, + { + "epoch": 3.23, + "learning_rate": 1.3523723563692237e-05, + "loss": 0.2137, + "step": 7715 + }, + { + "epoch": 3.23, + "learning_rate": 1.352213684789946e-05, + "loss": 0.1549, + "step": 7716 + }, + { + "epoch": 3.23, + "learning_rate": 1.3520550030865983e-05, + "loss": 0.1934, + "step": 7717 + }, + { + "epoch": 3.23, + "learning_rate": 1.3518963112637419e-05, + "loss": 0.1872, + "step": 7718 + }, + { + "epoch": 3.23, + "learning_rate": 1.3517376093259383e-05, + "loss": 0.2245, + "step": 7719 + }, + { + "epoch": 3.23, + "learning_rate": 1.3515788972777491e-05, + "loss": 0.1961, + "step": 7720 + }, + { + "epoch": 3.23, + "learning_rate": 1.3514201751237366e-05, + "loss": 0.1684, + "step": 7721 + }, + { + "epoch": 3.23, + "learning_rate": 1.3512614428684633e-05, + "loss": 0.1901, + "step": 7722 + }, + { + "epoch": 3.23, + "learning_rate": 1.3511027005164911e-05, + "loss": 0.1897, + "step": 7723 + }, + { + "epoch": 3.23, + "learning_rate": 1.3509439480723835e-05, + "loss": 0.1878, + "step": 7724 + }, + { + "epoch": 3.23, + "learning_rate": 1.3507851855407038e-05, + "loss": 0.1834, + "step": 7725 + }, + { + "epoch": 3.23, + "learning_rate": 1.350626412926015e-05, + "loss": 0.1637, + "step": 7726 + }, + { + "epoch": 3.24, + "learning_rate": 1.3504676302328813e-05, + "loss": 0.1433, + "step": 7727 + }, + { + "epoch": 3.24, + "learning_rate": 1.3503088374658661e-05, + "loss": 0.2066, + "step": 7728 + }, + { + "epoch": 3.24, + "learning_rate": 1.3501500346295347e-05, + "loss": 0.1597, + "step": 7729 + }, + { + "epoch": 3.24, + "learning_rate": 1.349991221728451e-05, + "loss": 0.201, + "step": 7730 + }, + { + "epoch": 3.24, + "learning_rate": 1.3498323987671801e-05, + "loss": 0.1925, + "step": 7731 + }, + { + "epoch": 3.24, + "learning_rate": 1.3496735657502877e-05, + "loss": 0.1942, + "step": 7732 + }, + { + "epoch": 3.24, + "learning_rate": 1.3495147226823386e-05, + "loss": 0.2095, + "step": 7733 + }, + { + "epoch": 3.24, + "learning_rate": 1.3493558695678992e-05, + "loss": 0.1603, + "step": 7734 + }, + { + "epoch": 3.24, + "learning_rate": 1.3491970064115351e-05, + "loss": 0.17, + "step": 7735 + }, + { + "epoch": 3.24, + "learning_rate": 1.3490381332178131e-05, + "loss": 0.1829, + "step": 7736 + }, + { + "epoch": 3.24, + "learning_rate": 1.3488792499912998e-05, + "loss": 0.2311, + "step": 7737 + }, + { + "epoch": 3.24, + "learning_rate": 1.348720356736562e-05, + "loss": 0.2073, + "step": 7738 + }, + { + "epoch": 3.24, + "learning_rate": 1.348561453458167e-05, + "loss": 0.1692, + "step": 7739 + }, + { + "epoch": 3.24, + "learning_rate": 1.3484025401606821e-05, + "loss": 0.1917, + "step": 7740 + }, + { + "epoch": 3.24, + "learning_rate": 1.3482436168486754e-05, + "loss": 0.1508, + "step": 7741 + }, + { + "epoch": 3.24, + "learning_rate": 1.3480846835267152e-05, + "loss": 0.1405, + "step": 7742 + }, + { + "epoch": 3.24, + "learning_rate": 1.3479257401993696e-05, + "loss": 0.2095, + "step": 7743 + }, + { + "epoch": 3.24, + "learning_rate": 1.347766786871207e-05, + "loss": 0.2075, + "step": 7744 + }, + { + "epoch": 3.24, + "learning_rate": 1.347607823546797e-05, + "loss": 0.1521, + "step": 7745 + }, + { + "epoch": 3.24, + "learning_rate": 1.3474488502307085e-05, + "loss": 0.1693, + "step": 7746 + }, + { + "epoch": 3.24, + "learning_rate": 1.3472898669275112e-05, + "loss": 0.1621, + "step": 7747 + }, + { + "epoch": 3.24, + "learning_rate": 1.347130873641775e-05, + "loss": 0.2106, + "step": 7748 + }, + { + "epoch": 3.24, + "learning_rate": 1.3469718703780694e-05, + "loss": 0.2237, + "step": 7749 + }, + { + "epoch": 3.24, + "learning_rate": 1.3468128571409658e-05, + "loss": 0.1928, + "step": 7750 + }, + { + "epoch": 3.25, + "learning_rate": 1.346653833935034e-05, + "loss": 0.2011, + "step": 7751 + }, + { + "epoch": 3.25, + "learning_rate": 1.3464948007648457e-05, + "loss": 0.1587, + "step": 7752 + }, + { + "epoch": 3.25, + "learning_rate": 1.3463357576349716e-05, + "loss": 0.1522, + "step": 7753 + }, + { + "epoch": 3.25, + "learning_rate": 1.3461767045499834e-05, + "loss": 0.2028, + "step": 7754 + }, + { + "epoch": 3.25, + "learning_rate": 1.346017641514453e-05, + "loss": 0.1807, + "step": 7755 + }, + { + "epoch": 3.25, + "learning_rate": 1.3458585685329529e-05, + "loss": 0.1705, + "step": 7756 + }, + { + "epoch": 3.25, + "learning_rate": 1.3456994856100547e-05, + "loss": 0.2111, + "step": 7757 + }, + { + "epoch": 3.25, + "learning_rate": 1.3455403927503317e-05, + "loss": 0.1834, + "step": 7758 + }, + { + "epoch": 3.25, + "learning_rate": 1.3453812899583572e-05, + "loss": 0.1738, + "step": 7759 + }, + { + "epoch": 3.25, + "learning_rate": 1.3452221772387036e-05, + "loss": 0.1727, + "step": 7760 + }, + { + "epoch": 3.25, + "learning_rate": 1.3450630545959448e-05, + "loss": 0.1906, + "step": 7761 + }, + { + "epoch": 3.25, + "learning_rate": 1.3449039220346547e-05, + "loss": 0.1901, + "step": 7762 + }, + { + "epoch": 3.25, + "learning_rate": 1.3447447795594078e-05, + "loss": 0.1671, + "step": 7763 + }, + { + "epoch": 3.25, + "learning_rate": 1.3445856271747777e-05, + "loss": 0.1833, + "step": 7764 + }, + { + "epoch": 3.25, + "learning_rate": 1.3444264648853395e-05, + "loss": 0.2022, + "step": 7765 + }, + { + "epoch": 3.25, + "learning_rate": 1.3442672926956685e-05, + "loss": 0.1815, + "step": 7766 + }, + { + "epoch": 3.25, + "learning_rate": 1.3441081106103395e-05, + "loss": 0.164, + "step": 7767 + }, + { + "epoch": 3.25, + "learning_rate": 1.3439489186339283e-05, + "loss": 0.208, + "step": 7768 + }, + { + "epoch": 3.25, + "learning_rate": 1.3437897167710105e-05, + "loss": 0.2039, + "step": 7769 + }, + { + "epoch": 3.25, + "learning_rate": 1.3436305050261625e-05, + "loss": 0.1856, + "step": 7770 + }, + { + "epoch": 3.25, + "learning_rate": 1.3434712834039606e-05, + "loss": 0.1896, + "step": 7771 + }, + { + "epoch": 3.25, + "learning_rate": 1.3433120519089813e-05, + "loss": 0.1631, + "step": 7772 + }, + { + "epoch": 3.25, + "learning_rate": 1.3431528105458021e-05, + "loss": 0.2288, + "step": 7773 + }, + { + "epoch": 3.25, + "learning_rate": 1.3429935593189996e-05, + "loss": 0.1845, + "step": 7774 + }, + { + "epoch": 3.26, + "learning_rate": 1.3428342982331515e-05, + "loss": 0.2092, + "step": 7775 + }, + { + "epoch": 3.26, + "learning_rate": 1.3426750272928359e-05, + "loss": 0.1466, + "step": 7776 + }, + { + "epoch": 3.26, + "learning_rate": 1.3425157465026305e-05, + "loss": 0.2149, + "step": 7777 + }, + { + "epoch": 3.26, + "learning_rate": 1.342356455867114e-05, + "loss": 0.2079, + "step": 7778 + }, + { + "epoch": 3.26, + "learning_rate": 1.342197155390865e-05, + "loss": 0.1664, + "step": 7779 + }, + { + "epoch": 3.26, + "learning_rate": 1.3420378450784623e-05, + "loss": 0.1936, + "step": 7780 + }, + { + "epoch": 3.26, + "learning_rate": 1.3418785249344855e-05, + "loss": 0.1982, + "step": 7781 + }, + { + "epoch": 3.26, + "learning_rate": 1.3417191949635137e-05, + "loss": 0.1644, + "step": 7782 + }, + { + "epoch": 3.26, + "learning_rate": 1.341559855170127e-05, + "loss": 0.1735, + "step": 7783 + }, + { + "epoch": 3.26, + "learning_rate": 1.3414005055589057e-05, + "loss": 0.1726, + "step": 7784 + }, + { + "epoch": 3.26, + "learning_rate": 1.3412411461344293e-05, + "loss": 0.1926, + "step": 7785 + }, + { + "epoch": 3.26, + "learning_rate": 1.3410817769012792e-05, + "loss": 0.1661, + "step": 7786 + }, + { + "epoch": 3.26, + "learning_rate": 1.340922397864036e-05, + "loss": 0.2143, + "step": 7787 + }, + { + "epoch": 3.26, + "learning_rate": 1.3407630090272808e-05, + "loss": 0.1829, + "step": 7788 + }, + { + "epoch": 3.26, + "learning_rate": 1.3406036103955957e-05, + "loss": 0.1304, + "step": 7789 + }, + { + "epoch": 3.26, + "learning_rate": 1.3404442019735617e-05, + "loss": 0.1993, + "step": 7790 + }, + { + "epoch": 3.26, + "learning_rate": 1.3402847837657614e-05, + "loss": 0.1636, + "step": 7791 + }, + { + "epoch": 3.26, + "learning_rate": 1.3401253557767768e-05, + "loss": 0.1769, + "step": 7792 + }, + { + "epoch": 3.26, + "learning_rate": 1.339965918011191e-05, + "loss": 0.1884, + "step": 7793 + }, + { + "epoch": 3.26, + "learning_rate": 1.3398064704735862e-05, + "loss": 0.1943, + "step": 7794 + }, + { + "epoch": 3.26, + "learning_rate": 1.3396470131685462e-05, + "loss": 0.2005, + "step": 7795 + }, + { + "epoch": 3.26, + "learning_rate": 1.3394875461006543e-05, + "loss": 0.2422, + "step": 7796 + }, + { + "epoch": 3.26, + "learning_rate": 1.3393280692744938e-05, + "loss": 0.1647, + "step": 7797 + }, + { + "epoch": 3.26, + "learning_rate": 1.3391685826946492e-05, + "loss": 0.2119, + "step": 7798 + }, + { + "epoch": 3.27, + "learning_rate": 1.3390090863657048e-05, + "loss": 0.1554, + "step": 7799 + }, + { + "epoch": 3.27, + "learning_rate": 1.338849580292245e-05, + "loss": 0.1796, + "step": 7800 + }, + { + "epoch": 3.27, + "learning_rate": 1.3386900644788546e-05, + "loss": 0.1523, + "step": 7801 + }, + { + "epoch": 3.27, + "learning_rate": 1.3385305389301187e-05, + "loss": 0.1896, + "step": 7802 + }, + { + "epoch": 3.27, + "learning_rate": 1.3383710036506232e-05, + "loss": 0.2046, + "step": 7803 + }, + { + "epoch": 3.27, + "learning_rate": 1.3382114586449533e-05, + "loss": 0.1643, + "step": 7804 + }, + { + "epoch": 3.27, + "learning_rate": 1.3380519039176951e-05, + "loss": 0.1594, + "step": 7805 + }, + { + "epoch": 3.27, + "learning_rate": 1.337892339473435e-05, + "loss": 0.1781, + "step": 7806 + }, + { + "epoch": 3.27, + "learning_rate": 1.3377327653167596e-05, + "loss": 0.1904, + "step": 7807 + }, + { + "epoch": 3.27, + "learning_rate": 1.3375731814522552e-05, + "loss": 0.1833, + "step": 7808 + }, + { + "epoch": 3.27, + "learning_rate": 1.3374135878845094e-05, + "loss": 0.1631, + "step": 7809 + }, + { + "epoch": 3.27, + "learning_rate": 1.3372539846181096e-05, + "loss": 0.2096, + "step": 7810 + }, + { + "epoch": 3.27, + "learning_rate": 1.337094371657643e-05, + "loss": 0.1732, + "step": 7811 + }, + { + "epoch": 3.27, + "learning_rate": 1.3369347490076981e-05, + "loss": 0.2059, + "step": 7812 + }, + { + "epoch": 3.27, + "learning_rate": 1.3367751166728626e-05, + "loss": 0.1556, + "step": 7813 + }, + { + "epoch": 3.27, + "learning_rate": 1.3366154746577254e-05, + "loss": 0.152, + "step": 7814 + }, + { + "epoch": 3.27, + "learning_rate": 1.3364558229668748e-05, + "loss": 0.1588, + "step": 7815 + }, + { + "epoch": 3.27, + "learning_rate": 1.3362961616049006e-05, + "loss": 0.1508, + "step": 7816 + }, + { + "epoch": 3.27, + "learning_rate": 1.3361364905763912e-05, + "loss": 0.2059, + "step": 7817 + }, + { + "epoch": 3.27, + "learning_rate": 1.335976809885937e-05, + "loss": 0.1803, + "step": 7818 + }, + { + "epoch": 3.27, + "learning_rate": 1.3358171195381272e-05, + "loss": 0.2117, + "step": 7819 + }, + { + "epoch": 3.27, + "learning_rate": 1.3356574195375526e-05, + "loss": 0.118, + "step": 7820 + }, + { + "epoch": 3.27, + "learning_rate": 1.335497709888803e-05, + "loss": 0.1968, + "step": 7821 + }, + { + "epoch": 3.27, + "learning_rate": 1.3353379905964698e-05, + "loss": 0.1957, + "step": 7822 + }, + { + "epoch": 3.28, + "learning_rate": 1.3351782616651436e-05, + "loss": 0.1985, + "step": 7823 + }, + { + "epoch": 3.28, + "learning_rate": 1.3350185230994157e-05, + "loss": 0.227, + "step": 7824 + }, + { + "epoch": 3.28, + "learning_rate": 1.3348587749038773e-05, + "loss": 0.2088, + "step": 7825 + }, + { + "epoch": 3.28, + "learning_rate": 1.334699017083121e-05, + "loss": 0.1898, + "step": 7826 + }, + { + "epoch": 3.28, + "learning_rate": 1.3345392496417381e-05, + "loss": 0.1721, + "step": 7827 + }, + { + "epoch": 3.28, + "learning_rate": 1.3343794725843215e-05, + "loss": 0.1599, + "step": 7828 + }, + { + "epoch": 3.28, + "learning_rate": 1.3342196859154639e-05, + "loss": 0.2064, + "step": 7829 + }, + { + "epoch": 3.28, + "learning_rate": 1.3340598896397579e-05, + "loss": 0.1717, + "step": 7830 + }, + { + "epoch": 3.28, + "learning_rate": 1.3339000837617967e-05, + "loss": 0.161, + "step": 7831 + }, + { + "epoch": 3.28, + "learning_rate": 1.3337402682861737e-05, + "loss": 0.1659, + "step": 7832 + }, + { + "epoch": 3.28, + "learning_rate": 1.333580443217483e-05, + "loss": 0.1689, + "step": 7833 + }, + { + "epoch": 3.28, + "learning_rate": 1.3334206085603186e-05, + "loss": 0.2025, + "step": 7834 + }, + { + "epoch": 3.28, + "learning_rate": 1.3332607643192745e-05, + "loss": 0.1741, + "step": 7835 + }, + { + "epoch": 3.28, + "learning_rate": 1.3331009104989456e-05, + "loss": 0.1395, + "step": 7836 + }, + { + "epoch": 3.28, + "learning_rate": 1.3329410471039266e-05, + "loss": 0.2303, + "step": 7837 + }, + { + "epoch": 3.28, + "learning_rate": 1.3327811741388126e-05, + "loss": 0.1997, + "step": 7838 + }, + { + "epoch": 3.28, + "learning_rate": 1.3326212916081991e-05, + "loss": 0.1643, + "step": 7839 + }, + { + "epoch": 3.28, + "learning_rate": 1.3324613995166818e-05, + "loss": 0.1636, + "step": 7840 + }, + { + "epoch": 3.28, + "learning_rate": 1.3323014978688567e-05, + "loss": 0.1608, + "step": 7841 + }, + { + "epoch": 3.28, + "learning_rate": 1.3321415866693195e-05, + "loss": 0.2052, + "step": 7842 + }, + { + "epoch": 3.28, + "learning_rate": 1.3319816659226672e-05, + "loss": 0.1693, + "step": 7843 + }, + { + "epoch": 3.28, + "learning_rate": 1.3318217356334967e-05, + "loss": 0.1523, + "step": 7844 + }, + { + "epoch": 3.28, + "learning_rate": 1.3316617958064049e-05, + "loss": 0.1823, + "step": 7845 + }, + { + "epoch": 3.28, + "learning_rate": 1.331501846445989e-05, + "loss": 0.2051, + "step": 7846 + }, + { + "epoch": 3.29, + "learning_rate": 1.3313418875568469e-05, + "loss": 0.197, + "step": 7847 + }, + { + "epoch": 3.29, + "learning_rate": 1.331181919143576e-05, + "loss": 0.1993, + "step": 7848 + }, + { + "epoch": 3.29, + "learning_rate": 1.3310219412107745e-05, + "loss": 0.1948, + "step": 7849 + }, + { + "epoch": 3.29, + "learning_rate": 1.3308619537630416e-05, + "loss": 0.1773, + "step": 7850 + }, + { + "epoch": 3.29, + "learning_rate": 1.3307019568049751e-05, + "loss": 0.1666, + "step": 7851 + }, + { + "epoch": 3.29, + "learning_rate": 1.3305419503411745e-05, + "loss": 0.1913, + "step": 7852 + }, + { + "epoch": 3.29, + "learning_rate": 1.3303819343762387e-05, + "loss": 0.2112, + "step": 7853 + }, + { + "epoch": 3.29, + "learning_rate": 1.3302219089147675e-05, + "loss": 0.1746, + "step": 7854 + }, + { + "epoch": 3.29, + "learning_rate": 1.3300618739613605e-05, + "loss": 0.1547, + "step": 7855 + }, + { + "epoch": 3.29, + "learning_rate": 1.3299018295206176e-05, + "loss": 0.1833, + "step": 7856 + }, + { + "epoch": 3.29, + "learning_rate": 1.3297417755971399e-05, + "loss": 0.1845, + "step": 7857 + }, + { + "epoch": 3.29, + "learning_rate": 1.329581712195527e-05, + "loss": 0.1692, + "step": 7858 + }, + { + "epoch": 3.29, + "learning_rate": 1.3294216393203803e-05, + "loss": 0.1781, + "step": 7859 + }, + { + "epoch": 3.29, + "learning_rate": 1.329261556976301e-05, + "loss": 0.1617, + "step": 7860 + }, + { + "epoch": 3.29, + "learning_rate": 1.3291014651678905e-05, + "loss": 0.215, + "step": 7861 + }, + { + "epoch": 3.29, + "learning_rate": 1.3289413638997502e-05, + "loss": 0.1892, + "step": 7862 + }, + { + "epoch": 3.29, + "learning_rate": 1.3287812531764826e-05, + "loss": 0.1876, + "step": 7863 + }, + { + "epoch": 3.29, + "learning_rate": 1.3286211330026894e-05, + "loss": 0.1248, + "step": 7864 + }, + { + "epoch": 3.29, + "learning_rate": 1.3284610033829734e-05, + "loss": 0.2129, + "step": 7865 + }, + { + "epoch": 3.29, + "learning_rate": 1.3283008643219373e-05, + "loss": 0.2075, + "step": 7866 + }, + { + "epoch": 3.29, + "learning_rate": 1.3281407158241842e-05, + "loss": 0.1871, + "step": 7867 + }, + { + "epoch": 3.29, + "learning_rate": 1.3279805578943174e-05, + "loss": 0.1847, + "step": 7868 + }, + { + "epoch": 3.29, + "learning_rate": 1.3278203905369407e-05, + "loss": 0.1944, + "step": 7869 + }, + { + "epoch": 3.29, + "learning_rate": 1.3276602137566575e-05, + "loss": 0.2002, + "step": 7870 + }, + { + "epoch": 3.3, + "learning_rate": 1.3275000275580723e-05, + "loss": 0.1795, + "step": 7871 + }, + { + "epoch": 3.3, + "learning_rate": 1.3273398319457895e-05, + "loss": 0.1858, + "step": 7872 + }, + { + "epoch": 3.3, + "learning_rate": 1.3271796269244135e-05, + "loss": 0.1892, + "step": 7873 + }, + { + "epoch": 3.3, + "learning_rate": 1.3270194124985499e-05, + "loss": 0.1832, + "step": 7874 + }, + { + "epoch": 3.3, + "learning_rate": 1.326859188672803e-05, + "loss": 0.1631, + "step": 7875 + }, + { + "epoch": 3.3, + "learning_rate": 1.3266989554517791e-05, + "loss": 0.2076, + "step": 7876 + }, + { + "epoch": 3.3, + "learning_rate": 1.3265387128400833e-05, + "loss": 0.2149, + "step": 7877 + }, + { + "epoch": 3.3, + "learning_rate": 1.3263784608423225e-05, + "loss": 0.186, + "step": 7878 + }, + { + "epoch": 3.3, + "learning_rate": 1.326218199463102e-05, + "loss": 0.1097, + "step": 7879 + }, + { + "epoch": 3.3, + "learning_rate": 1.3260579287070288e-05, + "loss": 0.1991, + "step": 7880 + }, + { + "epoch": 3.3, + "learning_rate": 1.3258976485787101e-05, + "loss": 0.1648, + "step": 7881 + }, + { + "epoch": 3.3, + "learning_rate": 1.3257373590827525e-05, + "loss": 0.1858, + "step": 7882 + }, + { + "epoch": 3.3, + "learning_rate": 1.3255770602237634e-05, + "loss": 0.2108, + "step": 7883 + }, + { + "epoch": 3.3, + "learning_rate": 1.325416752006351e-05, + "loss": 0.1963, + "step": 7884 + }, + { + "epoch": 3.3, + "learning_rate": 1.3252564344351227e-05, + "loss": 0.2124, + "step": 7885 + }, + { + "epoch": 3.3, + "learning_rate": 1.3250961075146868e-05, + "loss": 0.1754, + "step": 7886 + }, + { + "epoch": 3.3, + "learning_rate": 1.3249357712496518e-05, + "loss": 0.1954, + "step": 7887 + }, + { + "epoch": 3.3, + "learning_rate": 1.3247754256446263e-05, + "loss": 0.1774, + "step": 7888 + }, + { + "epoch": 3.3, + "learning_rate": 1.3246150707042196e-05, + "loss": 0.1806, + "step": 7889 + }, + { + "epoch": 3.3, + "learning_rate": 1.3244547064330404e-05, + "loss": 0.1706, + "step": 7890 + }, + { + "epoch": 3.3, + "learning_rate": 1.3242943328356988e-05, + "loss": 0.1864, + "step": 7891 + }, + { + "epoch": 3.3, + "learning_rate": 1.3241339499168042e-05, + "loss": 0.1964, + "step": 7892 + }, + { + "epoch": 3.3, + "learning_rate": 1.323973557680967e-05, + "loss": 0.1794, + "step": 7893 + }, + { + "epoch": 3.31, + "learning_rate": 1.3238131561327976e-05, + "loss": 0.162, + "step": 7894 + }, + { + "epoch": 3.31, + "learning_rate": 1.3236527452769061e-05, + "loss": 0.1988, + "step": 7895 + }, + { + "epoch": 3.31, + "learning_rate": 1.3234923251179036e-05, + "loss": 0.1611, + "step": 7896 + }, + { + "epoch": 3.31, + "learning_rate": 1.3233318956604013e-05, + "loss": 0.1859, + "step": 7897 + }, + { + "epoch": 3.31, + "learning_rate": 1.3231714569090108e-05, + "loss": 0.1789, + "step": 7898 + }, + { + "epoch": 3.31, + "learning_rate": 1.3230110088683433e-05, + "loss": 0.1955, + "step": 7899 + }, + { + "epoch": 3.31, + "learning_rate": 1.322850551543011e-05, + "loss": 0.2029, + "step": 7900 + }, + { + "epoch": 3.31, + "learning_rate": 1.3226900849376263e-05, + "loss": 0.1774, + "step": 7901 + }, + { + "epoch": 3.31, + "learning_rate": 1.3225296090568011e-05, + "loss": 0.204, + "step": 7902 + }, + { + "epoch": 3.31, + "learning_rate": 1.3223691239051489e-05, + "loss": 0.1998, + "step": 7903 + }, + { + "epoch": 3.31, + "learning_rate": 1.3222086294872818e-05, + "loss": 0.1742, + "step": 7904 + }, + { + "epoch": 3.31, + "learning_rate": 1.3220481258078142e-05, + "loss": 0.2086, + "step": 7905 + }, + { + "epoch": 3.31, + "learning_rate": 1.3218876128713584e-05, + "loss": 0.2133, + "step": 7906 + }, + { + "epoch": 3.31, + "learning_rate": 1.321727090682529e-05, + "loss": 0.2085, + "step": 7907 + }, + { + "epoch": 3.31, + "learning_rate": 1.3215665592459401e-05, + "loss": 0.1646, + "step": 7908 + }, + { + "epoch": 3.31, + "learning_rate": 1.3214060185662058e-05, + "loss": 0.1632, + "step": 7909 + }, + { + "epoch": 3.31, + "learning_rate": 1.3212454686479406e-05, + "loss": 0.1936, + "step": 7910 + }, + { + "epoch": 3.31, + "learning_rate": 1.3210849094957594e-05, + "loss": 0.1792, + "step": 7911 + }, + { + "epoch": 3.31, + "learning_rate": 1.3209243411142776e-05, + "loss": 0.1455, + "step": 7912 + }, + { + "epoch": 3.31, + "learning_rate": 1.32076376350811e-05, + "loss": 0.1829, + "step": 7913 + }, + { + "epoch": 3.31, + "learning_rate": 1.320603176681873e-05, + "loss": 0.2218, + "step": 7914 + }, + { + "epoch": 3.31, + "learning_rate": 1.3204425806401822e-05, + "loss": 0.1586, + "step": 7915 + }, + { + "epoch": 3.31, + "learning_rate": 1.3202819753876535e-05, + "loss": 0.204, + "step": 7916 + }, + { + "epoch": 3.31, + "learning_rate": 1.3201213609289036e-05, + "loss": 0.168, + "step": 7917 + }, + { + "epoch": 3.32, + "learning_rate": 1.3199607372685497e-05, + "loss": 0.1784, + "step": 7918 + }, + { + "epoch": 3.32, + "learning_rate": 1.319800104411208e-05, + "loss": 0.2257, + "step": 7919 + }, + { + "epoch": 3.32, + "learning_rate": 1.3196394623614963e-05, + "loss": 0.1622, + "step": 7920 + }, + { + "epoch": 3.32, + "learning_rate": 1.3194788111240317e-05, + "loss": 0.1386, + "step": 7921 + }, + { + "epoch": 3.32, + "learning_rate": 1.3193181507034323e-05, + "loss": 0.1977, + "step": 7922 + }, + { + "epoch": 3.32, + "learning_rate": 1.3191574811043159e-05, + "loss": 0.2258, + "step": 7923 + }, + { + "epoch": 3.32, + "learning_rate": 1.318996802331301e-05, + "loss": 0.2303, + "step": 7924 + }, + { + "epoch": 3.32, + "learning_rate": 1.318836114389006e-05, + "loss": 0.1982, + "step": 7925 + }, + { + "epoch": 3.32, + "learning_rate": 1.31867541728205e-05, + "loss": 0.1936, + "step": 7926 + }, + { + "epoch": 3.32, + "learning_rate": 1.3185147110150516e-05, + "loss": 0.1944, + "step": 7927 + }, + { + "epoch": 3.32, + "learning_rate": 1.3183539955926307e-05, + "loss": 0.191, + "step": 7928 + }, + { + "epoch": 3.32, + "learning_rate": 1.3181932710194067e-05, + "loss": 0.2047, + "step": 7929 + }, + { + "epoch": 3.32, + "learning_rate": 1.3180325372999994e-05, + "loss": 0.167, + "step": 7930 + }, + { + "epoch": 3.32, + "learning_rate": 1.3178717944390292e-05, + "loss": 0.2035, + "step": 7931 + }, + { + "epoch": 3.32, + "learning_rate": 1.3177110424411164e-05, + "loss": 0.2037, + "step": 7932 + }, + { + "epoch": 3.32, + "learning_rate": 1.3175502813108813e-05, + "loss": 0.2023, + "step": 7933 + }, + { + "epoch": 3.32, + "learning_rate": 1.3173895110529454e-05, + "loss": 0.1541, + "step": 7934 + }, + { + "epoch": 3.32, + "learning_rate": 1.3172287316719297e-05, + "loss": 0.1985, + "step": 7935 + }, + { + "epoch": 3.32, + "learning_rate": 1.3170679431724553e-05, + "loss": 0.1705, + "step": 7936 + }, + { + "epoch": 3.32, + "learning_rate": 1.3169071455591443e-05, + "loss": 0.2345, + "step": 7937 + }, + { + "epoch": 3.32, + "learning_rate": 1.3167463388366186e-05, + "loss": 0.2211, + "step": 7938 + }, + { + "epoch": 3.32, + "learning_rate": 1.3165855230095008e-05, + "loss": 0.2096, + "step": 7939 + }, + { + "epoch": 3.32, + "learning_rate": 1.3164246980824126e-05, + "loss": 0.1247, + "step": 7940 + }, + { + "epoch": 3.32, + "learning_rate": 1.3162638640599777e-05, + "loss": 0.1589, + "step": 7941 + }, + { + "epoch": 3.33, + "learning_rate": 1.3161030209468187e-05, + "loss": 0.1761, + "step": 7942 + }, + { + "epoch": 3.33, + "learning_rate": 1.3159421687475586e-05, + "loss": 0.1564, + "step": 7943 + }, + { + "epoch": 3.33, + "learning_rate": 1.3157813074668213e-05, + "loss": 0.1976, + "step": 7944 + }, + { + "epoch": 3.33, + "learning_rate": 1.3156204371092302e-05, + "loss": 0.1818, + "step": 7945 + }, + { + "epoch": 3.33, + "learning_rate": 1.3154595576794103e-05, + "loss": 0.2022, + "step": 7946 + }, + { + "epoch": 3.33, + "learning_rate": 1.315298669181985e-05, + "loss": 0.1456, + "step": 7947 + }, + { + "epoch": 3.33, + "learning_rate": 1.3151377716215791e-05, + "loss": 0.1878, + "step": 7948 + }, + { + "epoch": 3.33, + "learning_rate": 1.3149768650028181e-05, + "loss": 0.1428, + "step": 7949 + }, + { + "epoch": 3.33, + "learning_rate": 1.3148159493303262e-05, + "loss": 0.1644, + "step": 7950 + }, + { + "epoch": 3.33, + "learning_rate": 1.3146550246087293e-05, + "loss": 0.1733, + "step": 7951 + }, + { + "epoch": 3.33, + "learning_rate": 1.3144940908426532e-05, + "loss": 0.1824, + "step": 7952 + }, + { + "epoch": 3.33, + "learning_rate": 1.314333148036723e-05, + "loss": 0.1695, + "step": 7953 + }, + { + "epoch": 3.33, + "learning_rate": 1.3141721961955658e-05, + "loss": 0.1797, + "step": 7954 + }, + { + "epoch": 3.33, + "learning_rate": 1.3140112353238075e-05, + "loss": 0.1447, + "step": 7955 + }, + { + "epoch": 3.33, + "learning_rate": 1.3138502654260749e-05, + "loss": 0.1697, + "step": 7956 + }, + { + "epoch": 3.33, + "learning_rate": 1.313689286506995e-05, + "loss": 0.1789, + "step": 7957 + }, + { + "epoch": 3.33, + "learning_rate": 1.3135282985711948e-05, + "loss": 0.1959, + "step": 7958 + }, + { + "epoch": 3.33, + "learning_rate": 1.313367301623302e-05, + "loss": 0.1866, + "step": 7959 + }, + { + "epoch": 3.33, + "learning_rate": 1.3132062956679441e-05, + "loss": 0.1782, + "step": 7960 + }, + { + "epoch": 3.33, + "learning_rate": 1.3130452807097491e-05, + "loss": 0.195, + "step": 7961 + }, + { + "epoch": 3.33, + "learning_rate": 1.3128842567533456e-05, + "loss": 0.2005, + "step": 7962 + }, + { + "epoch": 3.33, + "learning_rate": 1.3127232238033613e-05, + "loss": 0.1895, + "step": 7963 + }, + { + "epoch": 3.33, + "learning_rate": 1.3125621818644258e-05, + "loss": 0.2153, + "step": 7964 + }, + { + "epoch": 3.33, + "learning_rate": 1.3124011309411678e-05, + "loss": 0.2004, + "step": 7965 + }, + { + "epoch": 3.34, + "learning_rate": 1.3122400710382164e-05, + "loss": 0.211, + "step": 7966 + }, + { + "epoch": 3.34, + "learning_rate": 1.3120790021602012e-05, + "loss": 0.1833, + "step": 7967 + }, + { + "epoch": 3.34, + "learning_rate": 1.311917924311752e-05, + "loss": 0.2227, + "step": 7968 + }, + { + "epoch": 3.34, + "learning_rate": 1.311756837497499e-05, + "loss": 0.2323, + "step": 7969 + }, + { + "epoch": 3.34, + "learning_rate": 1.3115957417220722e-05, + "loss": 0.1494, + "step": 7970 + }, + { + "epoch": 3.34, + "learning_rate": 1.3114346369901021e-05, + "loss": 0.1479, + "step": 7971 + }, + { + "epoch": 3.34, + "learning_rate": 1.3112735233062201e-05, + "loss": 0.1661, + "step": 7972 + }, + { + "epoch": 3.34, + "learning_rate": 1.311112400675057e-05, + "loss": 0.1722, + "step": 7973 + }, + { + "epoch": 3.34, + "learning_rate": 1.3109512691012434e-05, + "loss": 0.1949, + "step": 7974 + }, + { + "epoch": 3.34, + "learning_rate": 1.3107901285894118e-05, + "loss": 0.2101, + "step": 7975 + }, + { + "epoch": 3.34, + "learning_rate": 1.3106289791441941e-05, + "loss": 0.1923, + "step": 7976 + }, + { + "epoch": 3.34, + "learning_rate": 1.3104678207702218e-05, + "loss": 0.1802, + "step": 7977 + }, + { + "epoch": 3.34, + "learning_rate": 1.3103066534721273e-05, + "loss": 0.2054, + "step": 7978 + }, + { + "epoch": 3.34, + "learning_rate": 1.3101454772545435e-05, + "loss": 0.2031, + "step": 7979 + }, + { + "epoch": 3.34, + "learning_rate": 1.3099842921221034e-05, + "loss": 0.2368, + "step": 7980 + }, + { + "epoch": 3.34, + "learning_rate": 1.3098230980794394e-05, + "loss": 0.1559, + "step": 7981 + }, + { + "epoch": 3.34, + "learning_rate": 1.309661895131186e-05, + "loss": 0.1767, + "step": 7982 + }, + { + "epoch": 3.34, + "learning_rate": 1.3095006832819759e-05, + "loss": 0.1372, + "step": 7983 + }, + { + "epoch": 3.34, + "learning_rate": 1.3093394625364433e-05, + "loss": 0.1748, + "step": 7984 + }, + { + "epoch": 3.34, + "learning_rate": 1.3091782328992225e-05, + "loss": 0.1735, + "step": 7985 + }, + { + "epoch": 3.34, + "learning_rate": 1.3090169943749475e-05, + "loss": 0.1624, + "step": 7986 + }, + { + "epoch": 3.34, + "learning_rate": 1.3088557469682535e-05, + "loss": 0.1809, + "step": 7987 + }, + { + "epoch": 3.34, + "learning_rate": 1.3086944906837751e-05, + "loss": 0.2157, + "step": 7988 + }, + { + "epoch": 3.34, + "learning_rate": 1.3085332255261472e-05, + "loss": 0.1759, + "step": 7989 + }, + { + "epoch": 3.35, + "learning_rate": 1.3083719515000057e-05, + "loss": 0.1399, + "step": 7990 + }, + { + "epoch": 3.35, + "learning_rate": 1.308210668609986e-05, + "loss": 0.1942, + "step": 7991 + }, + { + "epoch": 3.35, + "learning_rate": 1.3080493768607243e-05, + "loss": 0.1867, + "step": 7992 + }, + { + "epoch": 3.35, + "learning_rate": 1.3078880762568566e-05, + "loss": 0.2199, + "step": 7993 + }, + { + "epoch": 3.35, + "learning_rate": 1.3077267668030194e-05, + "loss": 0.2082, + "step": 7994 + }, + { + "epoch": 3.35, + "learning_rate": 1.3075654485038491e-05, + "loss": 0.166, + "step": 7995 + }, + { + "epoch": 3.35, + "learning_rate": 1.3074041213639832e-05, + "loss": 0.1959, + "step": 7996 + }, + { + "epoch": 3.35, + "learning_rate": 1.3072427853880583e-05, + "loss": 0.1902, + "step": 7997 + }, + { + "epoch": 3.35, + "learning_rate": 1.3070814405807124e-05, + "loss": 0.1851, + "step": 7998 + }, + { + "epoch": 3.35, + "learning_rate": 1.3069200869465832e-05, + "loss": 0.1786, + "step": 7999 + }, + { + "epoch": 3.35, + "learning_rate": 1.3067587244903081e-05, + "loss": 0.1853, + "step": 8000 + }, + { + "epoch": 3.35, + "learning_rate": 1.3065973532165259e-05, + "loss": 0.1811, + "step": 8001 + }, + { + "epoch": 3.35, + "learning_rate": 1.3064359731298746e-05, + "loss": 0.2069, + "step": 8002 + }, + { + "epoch": 3.35, + "learning_rate": 1.3062745842349935e-05, + "loss": 0.2145, + "step": 8003 + }, + { + "epoch": 3.35, + "learning_rate": 1.3061131865365211e-05, + "loss": 0.2082, + "step": 8004 + }, + { + "epoch": 3.35, + "learning_rate": 1.3059517800390967e-05, + "loss": 0.1653, + "step": 8005 + }, + { + "epoch": 3.35, + "learning_rate": 1.30579036474736e-05, + "loss": 0.1863, + "step": 8006 + }, + { + "epoch": 3.35, + "learning_rate": 1.3056289406659506e-05, + "loss": 0.172, + "step": 8007 + }, + { + "epoch": 3.35, + "learning_rate": 1.3054675077995082e-05, + "loss": 0.1559, + "step": 8008 + }, + { + "epoch": 3.35, + "learning_rate": 1.3053060661526738e-05, + "loss": 0.202, + "step": 8009 + }, + { + "epoch": 3.35, + "learning_rate": 1.3051446157300875e-05, + "loss": 0.1904, + "step": 8010 + }, + { + "epoch": 3.35, + "learning_rate": 1.3049831565363897e-05, + "loss": 0.2119, + "step": 8011 + }, + { + "epoch": 3.35, + "learning_rate": 1.3048216885762218e-05, + "loss": 0.2065, + "step": 8012 + }, + { + "epoch": 3.35, + "learning_rate": 1.3046602118542251e-05, + "loss": 0.2191, + "step": 8013 + }, + { + "epoch": 3.36, + "learning_rate": 1.3044987263750409e-05, + "loss": 0.1768, + "step": 8014 + }, + { + "epoch": 3.36, + "learning_rate": 1.3043372321433108e-05, + "loss": 0.1888, + "step": 8015 + }, + { + "epoch": 3.36, + "learning_rate": 1.304175729163677e-05, + "loss": 0.1931, + "step": 8016 + }, + { + "epoch": 3.36, + "learning_rate": 1.304014217440782e-05, + "loss": 0.1912, + "step": 8017 + }, + { + "epoch": 3.36, + "learning_rate": 1.303852696979268e-05, + "loss": 0.1944, + "step": 8018 + }, + { + "epoch": 3.36, + "learning_rate": 1.3036911677837778e-05, + "loss": 0.1651, + "step": 8019 + }, + { + "epoch": 3.36, + "learning_rate": 1.3035296298589549e-05, + "loss": 0.2085, + "step": 8020 + }, + { + "epoch": 3.36, + "learning_rate": 1.3033680832094416e-05, + "loss": 0.2003, + "step": 8021 + }, + { + "epoch": 3.36, + "learning_rate": 1.3032065278398823e-05, + "loss": 0.1749, + "step": 8022 + }, + { + "epoch": 3.36, + "learning_rate": 1.3030449637549203e-05, + "loss": 0.1729, + "step": 8023 + }, + { + "epoch": 3.36, + "learning_rate": 1.3028833909591999e-05, + "loss": 0.2078, + "step": 8024 + }, + { + "epoch": 3.36, + "learning_rate": 1.3027218094573648e-05, + "loss": 0.1727, + "step": 8025 + }, + { + "epoch": 3.36, + "learning_rate": 1.30256021925406e-05, + "loss": 0.1386, + "step": 8026 + }, + { + "epoch": 3.36, + "learning_rate": 1.3023986203539304e-05, + "loss": 0.1952, + "step": 8027 + }, + { + "epoch": 3.36, + "learning_rate": 1.3022370127616209e-05, + "loss": 0.1911, + "step": 8028 + }, + { + "epoch": 3.36, + "learning_rate": 1.3020753964817763e-05, + "loss": 0.1608, + "step": 8029 + }, + { + "epoch": 3.36, + "learning_rate": 1.301913771519043e-05, + "loss": 0.225, + "step": 8030 + }, + { + "epoch": 3.36, + "learning_rate": 1.3017521378780655e-05, + "loss": 0.2167, + "step": 8031 + }, + { + "epoch": 3.36, + "learning_rate": 1.301590495563491e-05, + "loss": 0.1651, + "step": 8032 + }, + { + "epoch": 3.36, + "learning_rate": 1.3014288445799656e-05, + "loss": 0.1953, + "step": 8033 + }, + { + "epoch": 3.36, + "learning_rate": 1.3012671849321351e-05, + "loss": 0.2044, + "step": 8034 + }, + { + "epoch": 3.36, + "learning_rate": 1.3011055166246469e-05, + "loss": 0.2039, + "step": 8035 + }, + { + "epoch": 3.36, + "learning_rate": 1.3009438396621479e-05, + "loss": 0.1804, + "step": 8036 + }, + { + "epoch": 3.36, + "learning_rate": 1.3007821540492854e-05, + "loss": 0.1906, + "step": 8037 + }, + { + "epoch": 3.37, + "learning_rate": 1.3006204597907064e-05, + "loss": 0.2217, + "step": 8038 + }, + { + "epoch": 3.37, + "learning_rate": 1.3004587568910592e-05, + "loss": 0.2237, + "step": 8039 + }, + { + "epoch": 3.37, + "learning_rate": 1.3002970453549918e-05, + "loss": 0.1782, + "step": 8040 + }, + { + "epoch": 3.37, + "learning_rate": 1.300135325187152e-05, + "loss": 0.211, + "step": 8041 + }, + { + "epoch": 3.37, + "learning_rate": 1.2999735963921887e-05, + "loss": 0.184, + "step": 8042 + }, + { + "epoch": 3.37, + "learning_rate": 1.2998118589747509e-05, + "loss": 0.1948, + "step": 8043 + }, + { + "epoch": 3.37, + "learning_rate": 1.2996501129394872e-05, + "loss": 0.1965, + "step": 8044 + }, + { + "epoch": 3.37, + "learning_rate": 1.2994883582910465e-05, + "loss": 0.1551, + "step": 8045 + }, + { + "epoch": 3.37, + "learning_rate": 1.299326595034079e-05, + "loss": 0.1572, + "step": 8046 + }, + { + "epoch": 3.37, + "learning_rate": 1.2991648231732342e-05, + "loss": 0.1887, + "step": 8047 + }, + { + "epoch": 3.37, + "learning_rate": 1.2990030427131618e-05, + "loss": 0.1751, + "step": 8048 + }, + { + "epoch": 3.37, + "learning_rate": 1.2988412536585125e-05, + "loss": 0.2004, + "step": 8049 + }, + { + "epoch": 3.37, + "learning_rate": 1.2986794560139362e-05, + "loss": 0.2151, + "step": 8050 + }, + { + "epoch": 3.37, + "learning_rate": 1.2985176497840843e-05, + "loss": 0.2198, + "step": 8051 + }, + { + "epoch": 3.37, + "learning_rate": 1.2983558349736074e-05, + "loss": 0.1641, + "step": 8052 + }, + { + "epoch": 3.37, + "learning_rate": 1.2981940115871566e-05, + "loss": 0.176, + "step": 8053 + }, + { + "epoch": 3.37, + "learning_rate": 1.2980321796293838e-05, + "loss": 0.1961, + "step": 8054 + }, + { + "epoch": 3.37, + "learning_rate": 1.2978703391049402e-05, + "loss": 0.1903, + "step": 8055 + }, + { + "epoch": 3.37, + "learning_rate": 1.2977084900184784e-05, + "loss": 0.1761, + "step": 8056 + }, + { + "epoch": 3.37, + "learning_rate": 1.2975466323746499e-05, + "loss": 0.1296, + "step": 8057 + }, + { + "epoch": 3.37, + "learning_rate": 1.2973847661781078e-05, + "loss": 0.1581, + "step": 8058 + }, + { + "epoch": 3.37, + "learning_rate": 1.2972228914335044e-05, + "loss": 0.1424, + "step": 8059 + }, + { + "epoch": 3.37, + "learning_rate": 1.2970610081454925e-05, + "loss": 0.1769, + "step": 8060 + }, + { + "epoch": 3.37, + "learning_rate": 1.2968991163187256e-05, + "loss": 0.187, + "step": 8061 + }, + { + "epoch": 3.38, + "learning_rate": 1.2967372159578574e-05, + "loss": 0.1795, + "step": 8062 + }, + { + "epoch": 3.38, + "learning_rate": 1.2965753070675406e-05, + "loss": 0.1714, + "step": 8063 + }, + { + "epoch": 3.38, + "learning_rate": 1.2964133896524305e-05, + "loss": 0.1511, + "step": 8064 + }, + { + "epoch": 3.38, + "learning_rate": 1.2962514637171802e-05, + "loss": 0.1504, + "step": 8065 + }, + { + "epoch": 3.38, + "learning_rate": 1.2960895292664443e-05, + "loss": 0.1826, + "step": 8066 + }, + { + "epoch": 3.38, + "learning_rate": 1.295927586304878e-05, + "loss": 0.1782, + "step": 8067 + }, + { + "epoch": 3.38, + "learning_rate": 1.2957656348371357e-05, + "loss": 0.2208, + "step": 8068 + }, + { + "epoch": 3.38, + "learning_rate": 1.2956036748678727e-05, + "loss": 0.1828, + "step": 8069 + }, + { + "epoch": 3.38, + "learning_rate": 1.2954417064017441e-05, + "loss": 0.1838, + "step": 8070 + }, + { + "epoch": 3.38, + "learning_rate": 1.295279729443406e-05, + "loss": 0.1932, + "step": 8071 + }, + { + "epoch": 3.38, + "learning_rate": 1.2951177439975141e-05, + "loss": 0.2389, + "step": 8072 + }, + { + "epoch": 3.38, + "learning_rate": 1.2949557500687244e-05, + "loss": 0.1934, + "step": 8073 + }, + { + "epoch": 3.38, + "learning_rate": 1.2947937476616938e-05, + "loss": 0.1962, + "step": 8074 + }, + { + "epoch": 3.38, + "learning_rate": 1.2946317367810784e-05, + "loss": 0.2111, + "step": 8075 + }, + { + "epoch": 3.38, + "learning_rate": 1.2944697174315349e-05, + "loss": 0.1932, + "step": 8076 + }, + { + "epoch": 3.38, + "learning_rate": 1.2943076896177208e-05, + "loss": 0.1867, + "step": 8077 + }, + { + "epoch": 3.38, + "learning_rate": 1.2941456533442935e-05, + "loss": 0.1605, + "step": 8078 + }, + { + "epoch": 3.38, + "learning_rate": 1.2939836086159102e-05, + "loss": 0.2104, + "step": 8079 + }, + { + "epoch": 3.38, + "learning_rate": 1.2938215554372292e-05, + "loss": 0.1872, + "step": 8080 + }, + { + "epoch": 3.38, + "learning_rate": 1.2936594938129082e-05, + "loss": 0.1978, + "step": 8081 + }, + { + "epoch": 3.38, + "learning_rate": 1.2934974237476054e-05, + "loss": 0.2306, + "step": 8082 + }, + { + "epoch": 3.38, + "learning_rate": 1.2933353452459796e-05, + "loss": 0.1974, + "step": 8083 + }, + { + "epoch": 3.38, + "learning_rate": 1.29317325831269e-05, + "loss": 0.1785, + "step": 8084 + }, + { + "epoch": 3.38, + "learning_rate": 1.293011162952395e-05, + "loss": 0.1864, + "step": 8085 + }, + { + "epoch": 3.39, + "learning_rate": 1.292849059169754e-05, + "loss": 0.2104, + "step": 8086 + }, + { + "epoch": 3.39, + "learning_rate": 1.2926869469694267e-05, + "loss": 0.2014, + "step": 8087 + }, + { + "epoch": 3.39, + "learning_rate": 1.2925248263560733e-05, + "loss": 0.2069, + "step": 8088 + }, + { + "epoch": 3.39, + "learning_rate": 1.2923626973343527e-05, + "loss": 0.1916, + "step": 8089 + }, + { + "epoch": 3.39, + "learning_rate": 1.2922005599089263e-05, + "loss": 0.1521, + "step": 8090 + }, + { + "epoch": 3.39, + "learning_rate": 1.2920384140844539e-05, + "loss": 0.2251, + "step": 8091 + }, + { + "epoch": 3.39, + "learning_rate": 1.2918762598655965e-05, + "loss": 0.2019, + "step": 8092 + }, + { + "epoch": 3.39, + "learning_rate": 1.291714097257015e-05, + "loss": 0.1826, + "step": 8093 + }, + { + "epoch": 3.39, + "learning_rate": 1.2915519262633704e-05, + "loss": 0.2038, + "step": 8094 + }, + { + "epoch": 3.39, + "learning_rate": 1.2913897468893249e-05, + "loss": 0.187, + "step": 8095 + }, + { + "epoch": 3.39, + "learning_rate": 1.2912275591395393e-05, + "loss": 0.2268, + "step": 8096 + }, + { + "epoch": 3.39, + "learning_rate": 1.291065363018676e-05, + "loss": 0.2099, + "step": 8097 + }, + { + "epoch": 3.39, + "learning_rate": 1.290903158531397e-05, + "loss": 0.1623, + "step": 8098 + }, + { + "epoch": 3.39, + "learning_rate": 1.2907409456823651e-05, + "loss": 0.2001, + "step": 8099 + }, + { + "epoch": 3.39, + "learning_rate": 1.2905787244762427e-05, + "loss": 0.1717, + "step": 8100 + }, + { + "epoch": 3.39, + "learning_rate": 1.2904164949176927e-05, + "loss": 0.1823, + "step": 8101 + }, + { + "epoch": 3.39, + "learning_rate": 1.2902542570113786e-05, + "loss": 0.2004, + "step": 8102 + }, + { + "epoch": 3.39, + "learning_rate": 1.2900920107619627e-05, + "loss": 0.1961, + "step": 8103 + }, + { + "epoch": 3.39, + "learning_rate": 1.2899297561741097e-05, + "loss": 0.2076, + "step": 8104 + }, + { + "epoch": 3.39, + "learning_rate": 1.2897674932524832e-05, + "loss": 0.1968, + "step": 8105 + }, + { + "epoch": 3.39, + "learning_rate": 1.2896052220017474e-05, + "loss": 0.1579, + "step": 8106 + }, + { + "epoch": 3.39, + "learning_rate": 1.2894429424265661e-05, + "loss": 0.1819, + "step": 8107 + }, + { + "epoch": 3.39, + "learning_rate": 1.2892806545316048e-05, + "loss": 0.191, + "step": 8108 + }, + { + "epoch": 3.4, + "learning_rate": 1.2891183583215272e-05, + "loss": 0.1599, + "step": 8109 + }, + { + "epoch": 3.4, + "learning_rate": 1.2889560538009993e-05, + "loss": 0.164, + "step": 8110 + }, + { + "epoch": 3.4, + "learning_rate": 1.288793740974686e-05, + "loss": 0.1802, + "step": 8111 + }, + { + "epoch": 3.4, + "learning_rate": 1.288631419847253e-05, + "loss": 0.174, + "step": 8112 + }, + { + "epoch": 3.4, + "learning_rate": 1.2884690904233658e-05, + "loss": 0.2084, + "step": 8113 + }, + { + "epoch": 3.4, + "learning_rate": 1.2883067527076904e-05, + "loss": 0.1475, + "step": 8114 + }, + { + "epoch": 3.4, + "learning_rate": 1.2881444067048936e-05, + "loss": 0.1961, + "step": 8115 + }, + { + "epoch": 3.4, + "learning_rate": 1.2879820524196414e-05, + "loss": 0.1655, + "step": 8116 + }, + { + "epoch": 3.4, + "learning_rate": 1.2878196898566004e-05, + "loss": 0.1942, + "step": 8117 + }, + { + "epoch": 3.4, + "learning_rate": 1.2876573190204381e-05, + "loss": 0.1498, + "step": 8118 + }, + { + "epoch": 3.4, + "learning_rate": 1.2874949399158213e-05, + "loss": 0.2078, + "step": 8119 + }, + { + "epoch": 3.4, + "learning_rate": 1.2873325525474176e-05, + "loss": 0.2005, + "step": 8120 + }, + { + "epoch": 3.4, + "learning_rate": 1.2871701569198946e-05, + "loss": 0.1994, + "step": 8121 + }, + { + "epoch": 3.4, + "learning_rate": 1.2870077530379205e-05, + "loss": 0.1859, + "step": 8122 + }, + { + "epoch": 3.4, + "learning_rate": 1.286845340906163e-05, + "loss": 0.1879, + "step": 8123 + }, + { + "epoch": 3.4, + "learning_rate": 1.286682920529291e-05, + "loss": 0.1767, + "step": 8124 + }, + { + "epoch": 3.4, + "learning_rate": 1.2865204919119726e-05, + "loss": 0.203, + "step": 8125 + }, + { + "epoch": 3.4, + "learning_rate": 1.286358055058877e-05, + "loss": 0.2012, + "step": 8126 + }, + { + "epoch": 3.4, + "learning_rate": 1.2861956099746731e-05, + "loss": 0.1452, + "step": 8127 + }, + { + "epoch": 3.4, + "learning_rate": 1.2860331566640303e-05, + "loss": 0.1434, + "step": 8128 + }, + { + "epoch": 3.4, + "learning_rate": 1.2858706951316184e-05, + "loss": 0.1894, + "step": 8129 + }, + { + "epoch": 3.4, + "learning_rate": 1.285708225382107e-05, + "loss": 0.2196, + "step": 8130 + }, + { + "epoch": 3.4, + "learning_rate": 1.2855457474201663e-05, + "loss": 0.1776, + "step": 8131 + }, + { + "epoch": 3.4, + "learning_rate": 1.2853832612504662e-05, + "loss": 0.1807, + "step": 8132 + }, + { + "epoch": 3.41, + "learning_rate": 1.2852207668776777e-05, + "loss": 0.1557, + "step": 8133 + }, + { + "epoch": 3.41, + "learning_rate": 1.2850582643064713e-05, + "loss": 0.1438, + "step": 8134 + }, + { + "epoch": 3.41, + "learning_rate": 1.2848957535415183e-05, + "loss": 0.179, + "step": 8135 + }, + { + "epoch": 3.41, + "learning_rate": 1.2847332345874895e-05, + "loss": 0.1775, + "step": 8136 + }, + { + "epoch": 3.41, + "learning_rate": 1.2845707074490563e-05, + "loss": 0.1993, + "step": 8137 + }, + { + "epoch": 3.41, + "learning_rate": 1.284408172130891e-05, + "loss": 0.173, + "step": 8138 + }, + { + "epoch": 3.41, + "learning_rate": 1.284245628637665e-05, + "loss": 0.1968, + "step": 8139 + }, + { + "epoch": 3.41, + "learning_rate": 1.2840830769740506e-05, + "loss": 0.1977, + "step": 8140 + }, + { + "epoch": 3.41, + "learning_rate": 1.2839205171447203e-05, + "loss": 0.1816, + "step": 8141 + }, + { + "epoch": 3.41, + "learning_rate": 1.2837579491543468e-05, + "loss": 0.186, + "step": 8142 + }, + { + "epoch": 3.41, + "learning_rate": 1.2835953730076026e-05, + "loss": 0.2, + "step": 8143 + }, + { + "epoch": 3.41, + "learning_rate": 1.2834327887091611e-05, + "loss": 0.2119, + "step": 8144 + }, + { + "epoch": 3.41, + "learning_rate": 1.2832701962636957e-05, + "loss": 0.1799, + "step": 8145 + }, + { + "epoch": 3.41, + "learning_rate": 1.2831075956758801e-05, + "loss": 0.1658, + "step": 8146 + }, + { + "epoch": 3.41, + "learning_rate": 1.2829449869503876e-05, + "loss": 0.1929, + "step": 8147 + }, + { + "epoch": 3.41, + "learning_rate": 1.2827823700918924e-05, + "loss": 0.203, + "step": 8148 + }, + { + "epoch": 3.41, + "learning_rate": 1.2826197451050692e-05, + "loss": 0.1814, + "step": 8149 + }, + { + "epoch": 3.41, + "learning_rate": 1.2824571119945921e-05, + "loss": 0.1706, + "step": 8150 + }, + { + "epoch": 3.41, + "learning_rate": 1.282294470765136e-05, + "loss": 0.1477, + "step": 8151 + }, + { + "epoch": 3.41, + "learning_rate": 1.2821318214213758e-05, + "loss": 0.2028, + "step": 8152 + }, + { + "epoch": 3.41, + "learning_rate": 1.2819691639679867e-05, + "loss": 0.1899, + "step": 8153 + }, + { + "epoch": 3.41, + "learning_rate": 1.2818064984096443e-05, + "loss": 0.2138, + "step": 8154 + }, + { + "epoch": 3.41, + "learning_rate": 1.2816438247510243e-05, + "loss": 0.1841, + "step": 8155 + }, + { + "epoch": 3.41, + "learning_rate": 1.2814811429968022e-05, + "loss": 0.1999, + "step": 8156 + }, + { + "epoch": 3.42, + "learning_rate": 1.2813184531516545e-05, + "loss": 0.1649, + "step": 8157 + }, + { + "epoch": 3.42, + "learning_rate": 1.2811557552202576e-05, + "loss": 0.185, + "step": 8158 + }, + { + "epoch": 3.42, + "learning_rate": 1.2809930492072882e-05, + "loss": 0.1613, + "step": 8159 + }, + { + "epoch": 3.42, + "learning_rate": 1.2808303351174224e-05, + "loss": 0.189, + "step": 8160 + }, + { + "epoch": 3.42, + "learning_rate": 1.2806676129553381e-05, + "loss": 0.2033, + "step": 8161 + }, + { + "epoch": 3.42, + "learning_rate": 1.2805048827257123e-05, + "loss": 0.2193, + "step": 8162 + }, + { + "epoch": 3.42, + "learning_rate": 1.280342144433223e-05, + "loss": 0.2024, + "step": 8163 + }, + { + "epoch": 3.42, + "learning_rate": 1.280179398082547e-05, + "loss": 0.1882, + "step": 8164 + }, + { + "epoch": 3.42, + "learning_rate": 1.2800166436783627e-05, + "loss": 0.1955, + "step": 8165 + }, + { + "epoch": 3.42, + "learning_rate": 1.279853881225349e-05, + "loss": 0.1965, + "step": 8166 + }, + { + "epoch": 3.42, + "learning_rate": 1.2796911107281833e-05, + "loss": 0.1778, + "step": 8167 + }, + { + "epoch": 3.42, + "learning_rate": 1.279528332191545e-05, + "loss": 0.2018, + "step": 8168 + }, + { + "epoch": 3.42, + "learning_rate": 1.279365545620113e-05, + "loss": 0.1709, + "step": 8169 + }, + { + "epoch": 3.42, + "learning_rate": 1.2792027510185665e-05, + "loss": 0.186, + "step": 8170 + }, + { + "epoch": 3.42, + "learning_rate": 1.2790399483915843e-05, + "loss": 0.1931, + "step": 8171 + }, + { + "epoch": 3.42, + "learning_rate": 1.2788771377438465e-05, + "loss": 0.17, + "step": 8172 + }, + { + "epoch": 3.42, + "learning_rate": 1.278714319080033e-05, + "loss": 0.1962, + "step": 8173 + }, + { + "epoch": 3.42, + "learning_rate": 1.2785514924048235e-05, + "loss": 0.1863, + "step": 8174 + }, + { + "epoch": 3.42, + "learning_rate": 1.2783886577228986e-05, + "loss": 0.1965, + "step": 8175 + }, + { + "epoch": 3.42, + "learning_rate": 1.2782258150389387e-05, + "loss": 0.1853, + "step": 8176 + }, + { + "epoch": 3.42, + "learning_rate": 1.2780629643576249e-05, + "loss": 0.2044, + "step": 8177 + }, + { + "epoch": 3.42, + "learning_rate": 1.2779001056836373e-05, + "loss": 0.1349, + "step": 8178 + }, + { + "epoch": 3.42, + "learning_rate": 1.2777372390216585e-05, + "loss": 0.1896, + "step": 8179 + }, + { + "epoch": 3.42, + "learning_rate": 1.2775743643763691e-05, + "loss": 0.1824, + "step": 8180 + }, + { + "epoch": 3.43, + "learning_rate": 1.2774114817524505e-05, + "loss": 0.2179, + "step": 8181 + }, + { + "epoch": 3.43, + "learning_rate": 1.2772485911545854e-05, + "loss": 0.1746, + "step": 8182 + }, + { + "epoch": 3.43, + "learning_rate": 1.2770856925874555e-05, + "loss": 0.1649, + "step": 8183 + }, + { + "epoch": 3.43, + "learning_rate": 1.2769227860557431e-05, + "loss": 0.1964, + "step": 8184 + }, + { + "epoch": 3.43, + "learning_rate": 1.2767598715641309e-05, + "loss": 0.1795, + "step": 8185 + }, + { + "epoch": 3.43, + "learning_rate": 1.2765969491173023e-05, + "loss": 0.1827, + "step": 8186 + }, + { + "epoch": 3.43, + "learning_rate": 1.2764340187199393e-05, + "loss": 0.1434, + "step": 8187 + }, + { + "epoch": 3.43, + "learning_rate": 1.2762710803767258e-05, + "loss": 0.2034, + "step": 8188 + }, + { + "epoch": 3.43, + "learning_rate": 1.2761081340923455e-05, + "loss": 0.1645, + "step": 8189 + }, + { + "epoch": 3.43, + "learning_rate": 1.2759451798714816e-05, + "loss": 0.1894, + "step": 8190 + }, + { + "epoch": 3.43, + "learning_rate": 1.2757822177188187e-05, + "loss": 0.1838, + "step": 8191 + }, + { + "epoch": 3.43, + "learning_rate": 1.2756192476390406e-05, + "loss": 0.2094, + "step": 8192 + }, + { + "epoch": 3.43, + "learning_rate": 1.2754562696368318e-05, + "loss": 0.2115, + "step": 8193 + }, + { + "epoch": 3.43, + "learning_rate": 1.2752932837168768e-05, + "loss": 0.1835, + "step": 8194 + }, + { + "epoch": 3.43, + "learning_rate": 1.2751302898838607e-05, + "loss": 0.1596, + "step": 8195 + }, + { + "epoch": 3.43, + "learning_rate": 1.2749672881424687e-05, + "loss": 0.1958, + "step": 8196 + }, + { + "epoch": 3.43, + "learning_rate": 1.2748042784973857e-05, + "loss": 0.1959, + "step": 8197 + }, + { + "epoch": 3.43, + "learning_rate": 1.274641260953298e-05, + "loss": 0.2248, + "step": 8198 + }, + { + "epoch": 3.43, + "learning_rate": 1.2744782355148905e-05, + "loss": 0.2259, + "step": 8199 + }, + { + "epoch": 3.43, + "learning_rate": 1.27431520218685e-05, + "loss": 0.2132, + "step": 8200 + }, + { + "epoch": 3.43, + "learning_rate": 1.2741521609738623e-05, + "loss": 0.195, + "step": 8201 + }, + { + "epoch": 3.43, + "learning_rate": 1.273989111880614e-05, + "loss": 0.1968, + "step": 8202 + }, + { + "epoch": 3.43, + "learning_rate": 1.273826054911792e-05, + "loss": 0.2022, + "step": 8203 + }, + { + "epoch": 3.43, + "learning_rate": 1.2736629900720832e-05, + "loss": 0.2075, + "step": 8204 + }, + { + "epoch": 3.44, + "learning_rate": 1.2734999173661742e-05, + "loss": 0.173, + "step": 8205 + }, + { + "epoch": 3.44, + "learning_rate": 1.2733368367987528e-05, + "loss": 0.1718, + "step": 8206 + }, + { + "epoch": 3.44, + "learning_rate": 1.2731737483745068e-05, + "loss": 0.1859, + "step": 8207 + }, + { + "epoch": 3.44, + "learning_rate": 1.2730106520981236e-05, + "loss": 0.207, + "step": 8208 + }, + { + "epoch": 3.44, + "learning_rate": 1.2728475479742914e-05, + "loss": 0.1998, + "step": 8209 + }, + { + "epoch": 3.44, + "learning_rate": 1.272684436007699e-05, + "loss": 0.1945, + "step": 8210 + }, + { + "epoch": 3.44, + "learning_rate": 1.272521316203034e-05, + "loss": 0.1999, + "step": 8211 + }, + { + "epoch": 3.44, + "learning_rate": 1.2723581885649858e-05, + "loss": 0.1899, + "step": 8212 + }, + { + "epoch": 3.44, + "learning_rate": 1.2721950530982431e-05, + "loss": 0.1733, + "step": 8213 + }, + { + "epoch": 3.44, + "learning_rate": 1.2720319098074954e-05, + "loss": 0.2122, + "step": 8214 + }, + { + "epoch": 3.44, + "learning_rate": 1.2718687586974317e-05, + "loss": 0.1633, + "step": 8215 + }, + { + "epoch": 3.44, + "learning_rate": 1.2717055997727416e-05, + "loss": 0.1858, + "step": 8216 + }, + { + "epoch": 3.44, + "learning_rate": 1.2715424330381155e-05, + "loss": 0.1657, + "step": 8217 + }, + { + "epoch": 3.44, + "learning_rate": 1.2713792584982426e-05, + "loss": 0.2026, + "step": 8218 + }, + { + "epoch": 3.44, + "learning_rate": 1.2712160761578138e-05, + "loss": 0.1809, + "step": 8219 + }, + { + "epoch": 3.44, + "learning_rate": 1.27105288602152e-05, + "loss": 0.1778, + "step": 8220 + }, + { + "epoch": 3.44, + "learning_rate": 1.2708896880940513e-05, + "loss": 0.2055, + "step": 8221 + }, + { + "epoch": 3.44, + "learning_rate": 1.270726482380099e-05, + "loss": 0.211, + "step": 8222 + }, + { + "epoch": 3.44, + "learning_rate": 1.270563268884354e-05, + "loss": 0.1524, + "step": 8223 + }, + { + "epoch": 3.44, + "learning_rate": 1.2704000476115079e-05, + "loss": 0.1888, + "step": 8224 + }, + { + "epoch": 3.44, + "learning_rate": 1.2702368185662529e-05, + "loss": 0.1963, + "step": 8225 + }, + { + "epoch": 3.44, + "learning_rate": 1.2700735817532796e-05, + "loss": 0.1832, + "step": 8226 + }, + { + "epoch": 3.44, + "learning_rate": 1.2699103371772815e-05, + "loss": 0.1948, + "step": 8227 + }, + { + "epoch": 3.44, + "learning_rate": 1.26974708484295e-05, + "loss": 0.2054, + "step": 8228 + }, + { + "epoch": 3.45, + "learning_rate": 1.2695838247549777e-05, + "loss": 0.1572, + "step": 8229 + }, + { + "epoch": 3.45, + "learning_rate": 1.2694205569180577e-05, + "loss": 0.1622, + "step": 8230 + }, + { + "epoch": 3.45, + "learning_rate": 1.269257281336883e-05, + "loss": 0.1943, + "step": 8231 + }, + { + "epoch": 3.45, + "learning_rate": 1.2690939980161465e-05, + "loss": 0.1662, + "step": 8232 + }, + { + "epoch": 3.45, + "learning_rate": 1.2689307069605418e-05, + "loss": 0.1833, + "step": 8233 + }, + { + "epoch": 3.45, + "learning_rate": 1.268767408174763e-05, + "loss": 0.2034, + "step": 8234 + }, + { + "epoch": 3.45, + "learning_rate": 1.2686041016635031e-05, + "loss": 0.1898, + "step": 8235 + }, + { + "epoch": 3.45, + "learning_rate": 1.2684407874314568e-05, + "loss": 0.1917, + "step": 8236 + }, + { + "epoch": 3.45, + "learning_rate": 1.2682774654833188e-05, + "loss": 0.1471, + "step": 8237 + }, + { + "epoch": 3.45, + "learning_rate": 1.2681141358237823e-05, + "loss": 0.1666, + "step": 8238 + }, + { + "epoch": 3.45, + "learning_rate": 1.2679507984575431e-05, + "loss": 0.162, + "step": 8239 + }, + { + "epoch": 3.45, + "learning_rate": 1.2677874533892963e-05, + "loss": 0.18, + "step": 8240 + }, + { + "epoch": 3.45, + "learning_rate": 1.2676241006237367e-05, + "loss": 0.1862, + "step": 8241 + }, + { + "epoch": 3.45, + "learning_rate": 1.2674607401655595e-05, + "loss": 0.1574, + "step": 8242 + }, + { + "epoch": 3.45, + "learning_rate": 1.2672973720194608e-05, + "loss": 0.1919, + "step": 8243 + }, + { + "epoch": 3.45, + "learning_rate": 1.2671339961901366e-05, + "loss": 0.2196, + "step": 8244 + }, + { + "epoch": 3.45, + "learning_rate": 1.2669706126822823e-05, + "loss": 0.1901, + "step": 8245 + }, + { + "epoch": 3.45, + "learning_rate": 1.2668072215005946e-05, + "loss": 0.1999, + "step": 8246 + }, + { + "epoch": 3.45, + "learning_rate": 1.2666438226497706e-05, + "loss": 0.2016, + "step": 8247 + }, + { + "epoch": 3.45, + "learning_rate": 1.2664804161345064e-05, + "loss": 0.2066, + "step": 8248 + }, + { + "epoch": 3.45, + "learning_rate": 1.2663170019594988e-05, + "loss": 0.1709, + "step": 8249 + }, + { + "epoch": 3.45, + "learning_rate": 1.2661535801294455e-05, + "loss": 0.1545, + "step": 8250 + }, + { + "epoch": 3.45, + "learning_rate": 1.2659901506490437e-05, + "loss": 0.1295, + "step": 8251 + }, + { + "epoch": 3.45, + "learning_rate": 1.2658267135229911e-05, + "loss": 0.1584, + "step": 8252 + }, + { + "epoch": 3.46, + "learning_rate": 1.2656632687559853e-05, + "loss": 0.1862, + "step": 8253 + }, + { + "epoch": 3.46, + "learning_rate": 1.265499816352725e-05, + "loss": 0.1749, + "step": 8254 + }, + { + "epoch": 3.46, + "learning_rate": 1.2653363563179078e-05, + "loss": 0.1752, + "step": 8255 + }, + { + "epoch": 3.46, + "learning_rate": 1.2651728886562324e-05, + "loss": 0.178, + "step": 8256 + }, + { + "epoch": 3.46, + "learning_rate": 1.2650094133723978e-05, + "loss": 0.1874, + "step": 8257 + }, + { + "epoch": 3.46, + "learning_rate": 1.2648459304711026e-05, + "loss": 0.1988, + "step": 8258 + }, + { + "epoch": 3.46, + "learning_rate": 1.2646824399570466e-05, + "loss": 0.1609, + "step": 8259 + }, + { + "epoch": 3.46, + "learning_rate": 1.2645189418349282e-05, + "loss": 0.2107, + "step": 8260 + }, + { + "epoch": 3.46, + "learning_rate": 1.2643554361094478e-05, + "loss": 0.2183, + "step": 8261 + }, + { + "epoch": 3.46, + "learning_rate": 1.264191922785305e-05, + "loss": 0.201, + "step": 8262 + }, + { + "epoch": 3.46, + "learning_rate": 1.2640284018671995e-05, + "loss": 0.1965, + "step": 8263 + }, + { + "epoch": 3.46, + "learning_rate": 1.2638648733598325e-05, + "loss": 0.1527, + "step": 8264 + }, + { + "epoch": 3.46, + "learning_rate": 1.2637013372679032e-05, + "loss": 0.188, + "step": 8265 + }, + { + "epoch": 3.46, + "learning_rate": 1.2635377935961134e-05, + "loss": 0.1662, + "step": 8266 + }, + { + "epoch": 3.46, + "learning_rate": 1.2633742423491636e-05, + "loss": 0.1792, + "step": 8267 + }, + { + "epoch": 3.46, + "learning_rate": 1.2632106835317545e-05, + "loss": 0.1997, + "step": 8268 + }, + { + "epoch": 3.46, + "learning_rate": 1.263047117148588e-05, + "loss": 0.1857, + "step": 8269 + }, + { + "epoch": 3.46, + "learning_rate": 1.2628835432043655e-05, + "loss": 0.2021, + "step": 8270 + }, + { + "epoch": 3.46, + "learning_rate": 1.2627199617037895e-05, + "loss": 0.1827, + "step": 8271 + }, + { + "epoch": 3.46, + "learning_rate": 1.2625563726515609e-05, + "loss": 0.2168, + "step": 8272 + }, + { + "epoch": 3.46, + "learning_rate": 1.2623927760523822e-05, + "loss": 0.2037, + "step": 8273 + }, + { + "epoch": 3.46, + "learning_rate": 1.2622291719109559e-05, + "loss": 0.1773, + "step": 8274 + }, + { + "epoch": 3.46, + "learning_rate": 1.2620655602319851e-05, + "loss": 0.1975, + "step": 8275 + }, + { + "epoch": 3.46, + "learning_rate": 1.261901941020172e-05, + "loss": 0.1958, + "step": 8276 + }, + { + "epoch": 3.47, + "learning_rate": 1.2617383142802202e-05, + "loss": 0.2037, + "step": 8277 + }, + { + "epoch": 3.47, + "learning_rate": 1.261574680016833e-05, + "loss": 0.166, + "step": 8278 + }, + { + "epoch": 3.47, + "learning_rate": 1.2614110382347133e-05, + "loss": 0.2285, + "step": 8279 + }, + { + "epoch": 3.47, + "learning_rate": 1.2612473889385655e-05, + "loss": 0.1713, + "step": 8280 + }, + { + "epoch": 3.47, + "learning_rate": 1.2610837321330935e-05, + "loss": 0.1785, + "step": 8281 + }, + { + "epoch": 3.47, + "learning_rate": 1.2609200678230014e-05, + "loss": 0.1969, + "step": 8282 + }, + { + "epoch": 3.47, + "learning_rate": 1.260756396012993e-05, + "loss": 0.2023, + "step": 8283 + }, + { + "epoch": 3.47, + "learning_rate": 1.2605927167077736e-05, + "loss": 0.2079, + "step": 8284 + }, + { + "epoch": 3.47, + "learning_rate": 1.2604290299120478e-05, + "loss": 0.1478, + "step": 8285 + }, + { + "epoch": 3.47, + "learning_rate": 1.2602653356305207e-05, + "loss": 0.1633, + "step": 8286 + }, + { + "epoch": 3.47, + "learning_rate": 1.2601016338678976e-05, + "loss": 0.2344, + "step": 8287 + }, + { + "epoch": 3.47, + "learning_rate": 1.2599379246288837e-05, + "loss": 0.179, + "step": 8288 + }, + { + "epoch": 3.47, + "learning_rate": 1.2597742079181847e-05, + "loss": 0.2131, + "step": 8289 + }, + { + "epoch": 3.47, + "learning_rate": 1.2596104837405065e-05, + "loss": 0.1779, + "step": 8290 + }, + { + "epoch": 3.47, + "learning_rate": 1.2594467521005557e-05, + "loss": 0.203, + "step": 8291 + }, + { + "epoch": 3.47, + "learning_rate": 1.259283013003038e-05, + "loss": 0.1678, + "step": 8292 + }, + { + "epoch": 3.47, + "learning_rate": 1.2591192664526606e-05, + "loss": 0.1857, + "step": 8293 + }, + { + "epoch": 3.47, + "learning_rate": 1.2589555124541293e-05, + "loss": 0.1401, + "step": 8294 + }, + { + "epoch": 3.47, + "learning_rate": 1.2587917510121518e-05, + "loss": 0.1979, + "step": 8295 + }, + { + "epoch": 3.47, + "learning_rate": 1.2586279821314351e-05, + "loss": 0.1971, + "step": 8296 + }, + { + "epoch": 3.47, + "learning_rate": 1.2584642058166863e-05, + "loss": 0.1511, + "step": 8297 + }, + { + "epoch": 3.47, + "learning_rate": 1.2583004220726136e-05, + "loss": 0.1771, + "step": 8298 + }, + { + "epoch": 3.47, + "learning_rate": 1.2581366309039242e-05, + "loss": 0.1956, + "step": 8299 + }, + { + "epoch": 3.47, + "learning_rate": 1.2579728323153267e-05, + "loss": 0.1676, + "step": 8300 + }, + { + "epoch": 3.48, + "learning_rate": 1.257809026311529e-05, + "loss": 0.1678, + "step": 8301 + }, + { + "epoch": 3.48, + "learning_rate": 1.2576452128972394e-05, + "loss": 0.2065, + "step": 8302 + }, + { + "epoch": 3.48, + "learning_rate": 1.2574813920771667e-05, + "loss": 0.1917, + "step": 8303 + }, + { + "epoch": 3.48, + "learning_rate": 1.25731756385602e-05, + "loss": 0.1817, + "step": 8304 + }, + { + "epoch": 3.48, + "learning_rate": 1.2571537282385087e-05, + "loss": 0.2061, + "step": 8305 + }, + { + "epoch": 3.48, + "learning_rate": 1.2569898852293412e-05, + "loss": 0.1832, + "step": 8306 + }, + { + "epoch": 3.48, + "learning_rate": 1.2568260348332276e-05, + "loss": 0.1655, + "step": 8307 + }, + { + "epoch": 3.48, + "learning_rate": 1.2566621770548777e-05, + "loss": 0.1663, + "step": 8308 + }, + { + "epoch": 3.48, + "learning_rate": 1.256498311899001e-05, + "loss": 0.1934, + "step": 8309 + }, + { + "epoch": 3.48, + "learning_rate": 1.2563344393703081e-05, + "loss": 0.1964, + "step": 8310 + }, + { + "epoch": 3.48, + "learning_rate": 1.256170559473509e-05, + "loss": 0.1858, + "step": 8311 + }, + { + "epoch": 3.48, + "learning_rate": 1.2560066722133146e-05, + "loss": 0.1925, + "step": 8312 + }, + { + "epoch": 3.48, + "learning_rate": 1.2558427775944357e-05, + "loss": 0.2114, + "step": 8313 + }, + { + "epoch": 3.48, + "learning_rate": 1.255678875621583e-05, + "loss": 0.1507, + "step": 8314 + }, + { + "epoch": 3.48, + "learning_rate": 1.2555149662994684e-05, + "loss": 0.1966, + "step": 8315 + }, + { + "epoch": 3.48, + "learning_rate": 1.2553510496328025e-05, + "loss": 0.1924, + "step": 8316 + }, + { + "epoch": 3.48, + "learning_rate": 1.255187125626297e-05, + "loss": 0.1676, + "step": 8317 + }, + { + "epoch": 3.48, + "learning_rate": 1.2550231942846643e-05, + "loss": 0.1972, + "step": 8318 + }, + { + "epoch": 3.48, + "learning_rate": 1.2548592556126163e-05, + "loss": 0.1806, + "step": 8319 + }, + { + "epoch": 3.48, + "learning_rate": 1.2546953096148648e-05, + "loss": 0.18, + "step": 8320 + }, + { + "epoch": 3.48, + "learning_rate": 1.2545313562961228e-05, + "loss": 0.201, + "step": 8321 + }, + { + "epoch": 3.48, + "learning_rate": 1.2543673956611029e-05, + "loss": 0.1876, + "step": 8322 + }, + { + "epoch": 3.48, + "learning_rate": 1.2542034277145178e-05, + "loss": 0.1419, + "step": 8323 + }, + { + "epoch": 3.49, + "learning_rate": 1.2540394524610808e-05, + "loss": 0.1911, + "step": 8324 + }, + { + "epoch": 3.49, + "learning_rate": 1.2538754699055057e-05, + "loss": 0.1678, + "step": 8325 + }, + { + "epoch": 3.49, + "learning_rate": 1.2537114800525047e-05, + "loss": 0.2095, + "step": 8326 + }, + { + "epoch": 3.49, + "learning_rate": 1.2535474829067928e-05, + "loss": 0.2091, + "step": 8327 + }, + { + "epoch": 3.49, + "learning_rate": 1.2533834784730835e-05, + "loss": 0.191, + "step": 8328 + }, + { + "epoch": 3.49, + "learning_rate": 1.2532194667560908e-05, + "loss": 0.2376, + "step": 8329 + }, + { + "epoch": 3.49, + "learning_rate": 1.253055447760529e-05, + "loss": 0.1688, + "step": 8330 + }, + { + "epoch": 3.49, + "learning_rate": 1.2528914214911132e-05, + "loss": 0.2203, + "step": 8331 + }, + { + "epoch": 3.49, + "learning_rate": 1.2527273879525579e-05, + "loss": 0.1921, + "step": 8332 + }, + { + "epoch": 3.49, + "learning_rate": 1.2525633471495782e-05, + "loss": 0.1735, + "step": 8333 + }, + { + "epoch": 3.49, + "learning_rate": 1.252399299086889e-05, + "loss": 0.1715, + "step": 8334 + }, + { + "epoch": 3.49, + "learning_rate": 1.252235243769206e-05, + "loss": 0.2148, + "step": 8335 + }, + { + "epoch": 3.49, + "learning_rate": 1.2520711812012447e-05, + "loss": 0.1942, + "step": 8336 + }, + { + "epoch": 3.49, + "learning_rate": 1.2519071113877211e-05, + "loss": 0.2036, + "step": 8337 + }, + { + "epoch": 3.49, + "learning_rate": 1.2517430343333508e-05, + "loss": 0.1769, + "step": 8338 + }, + { + "epoch": 3.49, + "learning_rate": 1.251578950042851e-05, + "loss": 0.1718, + "step": 8339 + }, + { + "epoch": 3.49, + "learning_rate": 1.2514148585209372e-05, + "loss": 0.1998, + "step": 8340 + }, + { + "epoch": 3.49, + "learning_rate": 1.251250759772326e-05, + "loss": 0.1851, + "step": 8341 + }, + { + "epoch": 3.49, + "learning_rate": 1.2510866538017352e-05, + "loss": 0.2033, + "step": 8342 + }, + { + "epoch": 3.49, + "learning_rate": 1.2509225406138809e-05, + "loss": 0.1818, + "step": 8343 + }, + { + "epoch": 3.49, + "learning_rate": 1.2507584202134812e-05, + "loss": 0.2065, + "step": 8344 + }, + { + "epoch": 3.49, + "learning_rate": 1.250594292605253e-05, + "loss": 0.1722, + "step": 8345 + }, + { + "epoch": 3.49, + "learning_rate": 1.2504301577939146e-05, + "loss": 0.2298, + "step": 8346 + }, + { + "epoch": 3.49, + "learning_rate": 1.250266015784183e-05, + "loss": 0.2453, + "step": 8347 + }, + { + "epoch": 3.5, + "learning_rate": 1.2501018665807772e-05, + "loss": 0.1601, + "step": 8348 + }, + { + "epoch": 3.5, + "learning_rate": 1.2499377101884153e-05, + "loss": 0.1856, + "step": 8349 + }, + { + "epoch": 3.5, + "learning_rate": 1.2497735466118156e-05, + "loss": 0.224, + "step": 8350 + }, + { + "epoch": 3.5, + "learning_rate": 1.249609375855697e-05, + "loss": 0.2147, + "step": 8351 + }, + { + "epoch": 3.5, + "learning_rate": 1.2494451979247782e-05, + "loss": 0.2016, + "step": 8352 + }, + { + "epoch": 3.5, + "learning_rate": 1.2492810128237787e-05, + "loss": 0.1871, + "step": 8353 + }, + { + "epoch": 3.5, + "learning_rate": 1.2491168205574175e-05, + "loss": 0.2007, + "step": 8354 + }, + { + "epoch": 3.5, + "learning_rate": 1.2489526211304143e-05, + "loss": 0.1742, + "step": 8355 + }, + { + "epoch": 3.5, + "learning_rate": 1.2487884145474894e-05, + "loss": 0.1818, + "step": 8356 + }, + { + "epoch": 3.5, + "learning_rate": 1.2486242008133618e-05, + "loss": 0.1848, + "step": 8357 + }, + { + "epoch": 3.5, + "learning_rate": 1.2484599799327524e-05, + "loss": 0.1861, + "step": 8358 + }, + { + "epoch": 3.5, + "learning_rate": 1.2482957519103813e-05, + "loss": 0.1962, + "step": 8359 + }, + { + "epoch": 3.5, + "learning_rate": 1.2481315167509691e-05, + "loss": 0.163, + "step": 8360 + }, + { + "epoch": 3.5, + "learning_rate": 1.2479672744592368e-05, + "loss": 0.1704, + "step": 8361 + }, + { + "epoch": 3.5, + "learning_rate": 1.2478030250399052e-05, + "loss": 0.1996, + "step": 8362 + }, + { + "epoch": 3.5, + "learning_rate": 1.2476387684976956e-05, + "loss": 0.1665, + "step": 8363 + }, + { + "epoch": 3.5, + "learning_rate": 1.2474745048373292e-05, + "loss": 0.1742, + "step": 8364 + }, + { + "epoch": 3.5, + "learning_rate": 1.2473102340635275e-05, + "loss": 0.1708, + "step": 8365 + }, + { + "epoch": 3.5, + "learning_rate": 1.247145956181013e-05, + "loss": 0.2358, + "step": 8366 + }, + { + "epoch": 3.5, + "learning_rate": 1.246981671194507e-05, + "loss": 0.1767, + "step": 8367 + }, + { + "epoch": 3.5, + "learning_rate": 1.2468173791087324e-05, + "loss": 0.1535, + "step": 8368 + }, + { + "epoch": 3.5, + "learning_rate": 1.2466530799284111e-05, + "loss": 0.1793, + "step": 8369 + }, + { + "epoch": 3.5, + "learning_rate": 1.2464887736582659e-05, + "loss": 0.1774, + "step": 8370 + }, + { + "epoch": 3.5, + "learning_rate": 1.2463244603030194e-05, + "loss": 0.1839, + "step": 8371 + }, + { + "epoch": 3.51, + "learning_rate": 1.2461601398673953e-05, + "loss": 0.2015, + "step": 8372 + }, + { + "epoch": 3.51, + "learning_rate": 1.2459958123561163e-05, + "loss": 0.1921, + "step": 8373 + }, + { + "epoch": 3.51, + "learning_rate": 1.2458314777739056e-05, + "loss": 0.1948, + "step": 8374 + }, + { + "epoch": 3.51, + "learning_rate": 1.2456671361254876e-05, + "loss": 0.2294, + "step": 8375 + }, + { + "epoch": 3.51, + "learning_rate": 1.2455027874155856e-05, + "loss": 0.193, + "step": 8376 + }, + { + "epoch": 3.51, + "learning_rate": 1.2453384316489239e-05, + "loss": 0.1784, + "step": 8377 + }, + { + "epoch": 3.51, + "learning_rate": 1.2451740688302265e-05, + "loss": 0.1754, + "step": 8378 + }, + { + "epoch": 3.51, + "learning_rate": 1.2450096989642184e-05, + "loss": 0.1777, + "step": 8379 + }, + { + "epoch": 3.51, + "learning_rate": 1.2448453220556235e-05, + "loss": 0.1634, + "step": 8380 + }, + { + "epoch": 3.51, + "learning_rate": 1.2446809381091674e-05, + "loss": 0.1739, + "step": 8381 + }, + { + "epoch": 3.51, + "learning_rate": 1.2445165471295745e-05, + "loss": 0.1825, + "step": 8382 + }, + { + "epoch": 3.51, + "learning_rate": 1.244352149121571e-05, + "loss": 0.1394, + "step": 8383 + }, + { + "epoch": 3.51, + "learning_rate": 1.2441877440898817e-05, + "loss": 0.1721, + "step": 8384 + }, + { + "epoch": 3.51, + "learning_rate": 1.2440233320392322e-05, + "loss": 0.1992, + "step": 8385 + }, + { + "epoch": 3.51, + "learning_rate": 1.2438589129743484e-05, + "loss": 0.1969, + "step": 8386 + }, + { + "epoch": 3.51, + "learning_rate": 1.243694486899957e-05, + "loss": 0.214, + "step": 8387 + }, + { + "epoch": 3.51, + "learning_rate": 1.2435300538207832e-05, + "loss": 0.1699, + "step": 8388 + }, + { + "epoch": 3.51, + "learning_rate": 1.2433656137415545e-05, + "loss": 0.1888, + "step": 8389 + }, + { + "epoch": 3.51, + "learning_rate": 1.2432011666669972e-05, + "loss": 0.1743, + "step": 8390 + }, + { + "epoch": 3.51, + "learning_rate": 1.243036712601838e-05, + "loss": 0.1822, + "step": 8391 + }, + { + "epoch": 3.51, + "learning_rate": 1.2428722515508043e-05, + "loss": 0.1835, + "step": 8392 + }, + { + "epoch": 3.51, + "learning_rate": 1.2427077835186235e-05, + "loss": 0.2189, + "step": 8393 + }, + { + "epoch": 3.51, + "learning_rate": 1.2425433085100224e-05, + "loss": 0.1872, + "step": 8394 + }, + { + "epoch": 3.51, + "learning_rate": 1.2423788265297294e-05, + "loss": 0.2121, + "step": 8395 + }, + { + "epoch": 3.52, + "learning_rate": 1.242214337582472e-05, + "loss": 0.1972, + "step": 8396 + }, + { + "epoch": 3.52, + "learning_rate": 1.2420498416729785e-05, + "loss": 0.2004, + "step": 8397 + }, + { + "epoch": 3.52, + "learning_rate": 1.2418853388059771e-05, + "loss": 0.1933, + "step": 8398 + }, + { + "epoch": 3.52, + "learning_rate": 1.241720828986196e-05, + "loss": 0.18, + "step": 8399 + }, + { + "epoch": 3.52, + "learning_rate": 1.2415563122183644e-05, + "loss": 0.2336, + "step": 8400 + }, + { + "epoch": 3.52, + "learning_rate": 1.2413917885072109e-05, + "loss": 0.1966, + "step": 8401 + }, + { + "epoch": 3.52, + "learning_rate": 1.2412272578574647e-05, + "loss": 0.2009, + "step": 8402 + }, + { + "epoch": 3.52, + "learning_rate": 1.241062720273855e-05, + "loss": 0.1812, + "step": 8403 + }, + { + "epoch": 3.52, + "learning_rate": 1.2408981757611112e-05, + "loss": 0.2079, + "step": 8404 + }, + { + "epoch": 3.52, + "learning_rate": 1.2407336243239632e-05, + "loss": 0.1831, + "step": 8405 + }, + { + "epoch": 3.52, + "learning_rate": 1.240569065967141e-05, + "loss": 0.2116, + "step": 8406 + }, + { + "epoch": 3.52, + "learning_rate": 1.2404045006953744e-05, + "loss": 0.1673, + "step": 8407 + }, + { + "epoch": 3.52, + "learning_rate": 1.2402399285133935e-05, + "loss": 0.2214, + "step": 8408 + }, + { + "epoch": 3.52, + "learning_rate": 1.2400753494259293e-05, + "loss": 0.1845, + "step": 8409 + }, + { + "epoch": 3.52, + "learning_rate": 1.2399107634377122e-05, + "loss": 0.1506, + "step": 8410 + }, + { + "epoch": 3.52, + "learning_rate": 1.2397461705534727e-05, + "loss": 0.1963, + "step": 8411 + }, + { + "epoch": 3.52, + "learning_rate": 1.2395815707779426e-05, + "loss": 0.2047, + "step": 8412 + }, + { + "epoch": 3.52, + "learning_rate": 1.239416964115853e-05, + "loss": 0.2205, + "step": 8413 + }, + { + "epoch": 3.52, + "learning_rate": 1.2392523505719349e-05, + "loss": 0.1948, + "step": 8414 + }, + { + "epoch": 3.52, + "learning_rate": 1.2390877301509205e-05, + "loss": 0.2205, + "step": 8415 + }, + { + "epoch": 3.52, + "learning_rate": 1.2389231028575412e-05, + "loss": 0.1596, + "step": 8416 + }, + { + "epoch": 3.52, + "learning_rate": 1.2387584686965297e-05, + "loss": 0.2314, + "step": 8417 + }, + { + "epoch": 3.52, + "learning_rate": 1.2385938276726179e-05, + "loss": 0.1764, + "step": 8418 + }, + { + "epoch": 3.52, + "learning_rate": 1.2384291797905382e-05, + "loss": 0.1856, + "step": 8419 + }, + { + "epoch": 3.53, + "learning_rate": 1.2382645250550232e-05, + "loss": 0.2062, + "step": 8420 + }, + { + "epoch": 3.53, + "learning_rate": 1.2380998634708057e-05, + "loss": 0.2007, + "step": 8421 + }, + { + "epoch": 3.53, + "learning_rate": 1.2379351950426188e-05, + "loss": 0.2188, + "step": 8422 + }, + { + "epoch": 3.53, + "learning_rate": 1.2377705197751961e-05, + "loss": 0.212, + "step": 8423 + }, + { + "epoch": 3.53, + "learning_rate": 1.237605837673271e-05, + "loss": 0.2021, + "step": 8424 + }, + { + "epoch": 3.53, + "learning_rate": 1.2374411487415766e-05, + "loss": 0.1813, + "step": 8425 + }, + { + "epoch": 3.53, + "learning_rate": 1.2372764529848471e-05, + "loss": 0.1811, + "step": 8426 + }, + { + "epoch": 3.53, + "learning_rate": 1.237111750407817e-05, + "loss": 0.1925, + "step": 8427 + }, + { + "epoch": 3.53, + "learning_rate": 1.2369470410152195e-05, + "loss": 0.1922, + "step": 8428 + }, + { + "epoch": 3.53, + "learning_rate": 1.2367823248117899e-05, + "loss": 0.2002, + "step": 8429 + }, + { + "epoch": 3.53, + "learning_rate": 1.2366176018022621e-05, + "loss": 0.2103, + "step": 8430 + }, + { + "epoch": 3.53, + "learning_rate": 1.2364528719913717e-05, + "loss": 0.1977, + "step": 8431 + }, + { + "epoch": 3.53, + "learning_rate": 1.2362881353838532e-05, + "loss": 0.1977, + "step": 8432 + }, + { + "epoch": 3.53, + "learning_rate": 1.236123391984442e-05, + "loss": 0.1646, + "step": 8433 + }, + { + "epoch": 3.53, + "learning_rate": 1.2359586417978733e-05, + "loss": 0.1623, + "step": 8434 + }, + { + "epoch": 3.53, + "learning_rate": 1.2357938848288829e-05, + "loss": 0.2135, + "step": 8435 + }, + { + "epoch": 3.53, + "learning_rate": 1.2356291210822063e-05, + "loss": 0.2054, + "step": 8436 + }, + { + "epoch": 3.53, + "learning_rate": 1.23546435056258e-05, + "loss": 0.1851, + "step": 8437 + }, + { + "epoch": 3.53, + "learning_rate": 1.2352995732747398e-05, + "loss": 0.2118, + "step": 8438 + }, + { + "epoch": 3.53, + "learning_rate": 1.235134789223422e-05, + "loss": 0.2034, + "step": 8439 + }, + { + "epoch": 3.53, + "learning_rate": 1.2349699984133636e-05, + "loss": 0.1801, + "step": 8440 + }, + { + "epoch": 3.53, + "learning_rate": 1.234805200849301e-05, + "loss": 0.1968, + "step": 8441 + }, + { + "epoch": 3.53, + "learning_rate": 1.2346403965359712e-05, + "loss": 0.1928, + "step": 8442 + }, + { + "epoch": 3.53, + "learning_rate": 1.2344755854781114e-05, + "loss": 0.1793, + "step": 8443 + }, + { + "epoch": 3.54, + "learning_rate": 1.2343107676804593e-05, + "loss": 0.1609, + "step": 8444 + }, + { + "epoch": 3.54, + "learning_rate": 1.2341459431477516e-05, + "loss": 0.1669, + "step": 8445 + }, + { + "epoch": 3.54, + "learning_rate": 1.2339811118847269e-05, + "loss": 0.1921, + "step": 8446 + }, + { + "epoch": 3.54, + "learning_rate": 1.2338162738961227e-05, + "loss": 0.1854, + "step": 8447 + }, + { + "epoch": 3.54, + "learning_rate": 1.233651429186677e-05, + "loss": 0.1917, + "step": 8448 + }, + { + "epoch": 3.54, + "learning_rate": 1.2334865777611284e-05, + "loss": 0.1943, + "step": 8449 + }, + { + "epoch": 3.54, + "learning_rate": 1.2333217196242152e-05, + "loss": 0.1806, + "step": 8450 + }, + { + "epoch": 3.54, + "learning_rate": 1.2331568547806768e-05, + "loss": 0.1831, + "step": 8451 + }, + { + "epoch": 3.54, + "learning_rate": 1.232991983235251e-05, + "loss": 0.2109, + "step": 8452 + }, + { + "epoch": 3.54, + "learning_rate": 1.2328271049926774e-05, + "loss": 0.213, + "step": 8453 + }, + { + "epoch": 3.54, + "learning_rate": 1.2326622200576956e-05, + "loss": 0.1713, + "step": 8454 + }, + { + "epoch": 3.54, + "learning_rate": 1.2324973284350444e-05, + "loss": 0.2101, + "step": 8455 + }, + { + "epoch": 3.54, + "learning_rate": 1.2323324301294637e-05, + "loss": 0.1893, + "step": 8456 + }, + { + "epoch": 3.54, + "learning_rate": 1.2321675251456937e-05, + "loss": 0.1919, + "step": 8457 + }, + { + "epoch": 3.54, + "learning_rate": 1.2320026134884744e-05, + "loss": 0.1883, + "step": 8458 + }, + { + "epoch": 3.54, + "learning_rate": 1.2318376951625455e-05, + "loss": 0.1351, + "step": 8459 + }, + { + "epoch": 3.54, + "learning_rate": 1.2316727701726479e-05, + "loss": 0.2197, + "step": 8460 + }, + { + "epoch": 3.54, + "learning_rate": 1.2315078385235223e-05, + "loss": 0.1724, + "step": 8461 + }, + { + "epoch": 3.54, + "learning_rate": 1.2313429002199088e-05, + "loss": 0.1657, + "step": 8462 + }, + { + "epoch": 3.54, + "learning_rate": 1.2311779552665495e-05, + "loss": 0.1867, + "step": 8463 + }, + { + "epoch": 3.54, + "learning_rate": 1.2310130036681847e-05, + "loss": 0.1593, + "step": 8464 + }, + { + "epoch": 3.54, + "learning_rate": 1.2308480454295563e-05, + "loss": 0.1857, + "step": 8465 + }, + { + "epoch": 3.54, + "learning_rate": 1.2306830805554055e-05, + "loss": 0.1832, + "step": 8466 + }, + { + "epoch": 3.54, + "learning_rate": 1.2305181090504744e-05, + "loss": 0.1751, + "step": 8467 + }, + { + "epoch": 3.55, + "learning_rate": 1.2303531309195047e-05, + "loss": 0.178, + "step": 8468 + }, + { + "epoch": 3.55, + "learning_rate": 1.2301881461672387e-05, + "loss": 0.1923, + "step": 8469 + }, + { + "epoch": 3.55, + "learning_rate": 1.2300231547984187e-05, + "loss": 0.1603, + "step": 8470 + }, + { + "epoch": 3.55, + "learning_rate": 1.2298581568177874e-05, + "loss": 0.2233, + "step": 8471 + }, + { + "epoch": 3.55, + "learning_rate": 1.229693152230087e-05, + "loss": 0.1936, + "step": 8472 + }, + { + "epoch": 3.55, + "learning_rate": 1.2295281410400611e-05, + "loss": 0.198, + "step": 8473 + }, + { + "epoch": 3.55, + "learning_rate": 1.2293631232524524e-05, + "loss": 0.1754, + "step": 8474 + }, + { + "epoch": 3.55, + "learning_rate": 1.2291980988720042e-05, + "loss": 0.156, + "step": 8475 + }, + { + "epoch": 3.55, + "learning_rate": 1.22903306790346e-05, + "loss": 0.192, + "step": 8476 + }, + { + "epoch": 3.55, + "learning_rate": 1.2288680303515634e-05, + "loss": 0.1959, + "step": 8477 + }, + { + "epoch": 3.55, + "learning_rate": 1.2287029862210587e-05, + "loss": 0.2003, + "step": 8478 + }, + { + "epoch": 3.55, + "learning_rate": 1.2285379355166893e-05, + "loss": 0.1684, + "step": 8479 + }, + { + "epoch": 3.55, + "learning_rate": 1.2283728782431996e-05, + "loss": 0.2383, + "step": 8480 + }, + { + "epoch": 3.55, + "learning_rate": 1.2282078144053347e-05, + "loss": 0.212, + "step": 8481 + }, + { + "epoch": 3.55, + "learning_rate": 1.2280427440078383e-05, + "loss": 0.1748, + "step": 8482 + }, + { + "epoch": 3.55, + "learning_rate": 1.2278776670554555e-05, + "loss": 0.1972, + "step": 8483 + }, + { + "epoch": 3.55, + "learning_rate": 1.2277125835529318e-05, + "loss": 0.1675, + "step": 8484 + }, + { + "epoch": 3.55, + "learning_rate": 1.2275474935050116e-05, + "loss": 0.184, + "step": 8485 + }, + { + "epoch": 3.55, + "learning_rate": 1.227382396916441e-05, + "loss": 0.1557, + "step": 8486 + }, + { + "epoch": 3.55, + "learning_rate": 1.2272172937919646e-05, + "loss": 0.2133, + "step": 8487 + }, + { + "epoch": 3.55, + "learning_rate": 1.2270521841363293e-05, + "loss": 0.2386, + "step": 8488 + }, + { + "epoch": 3.55, + "learning_rate": 1.2268870679542799e-05, + "loss": 0.1974, + "step": 8489 + }, + { + "epoch": 3.55, + "learning_rate": 1.2267219452505635e-05, + "loss": 0.1825, + "step": 8490 + }, + { + "epoch": 3.55, + "learning_rate": 1.226556816029926e-05, + "loss": 0.2028, + "step": 8491 + }, + { + "epoch": 3.56, + "learning_rate": 1.2263916802971134e-05, + "loss": 0.1763, + "step": 8492 + }, + { + "epoch": 3.56, + "learning_rate": 1.2262265380568729e-05, + "loss": 0.1288, + "step": 8493 + }, + { + "epoch": 3.56, + "learning_rate": 1.2260613893139517e-05, + "loss": 0.1676, + "step": 8494 + }, + { + "epoch": 3.56, + "learning_rate": 1.2258962340730965e-05, + "loss": 0.173, + "step": 8495 + }, + { + "epoch": 3.56, + "learning_rate": 1.2257310723390541e-05, + "loss": 0.1756, + "step": 8496 + }, + { + "epoch": 3.56, + "learning_rate": 1.2255659041165726e-05, + "loss": 0.1826, + "step": 8497 + }, + { + "epoch": 3.56, + "learning_rate": 1.2254007294103994e-05, + "loss": 0.2019, + "step": 8498 + }, + { + "epoch": 3.56, + "learning_rate": 1.2252355482252823e-05, + "loss": 0.2117, + "step": 8499 + }, + { + "epoch": 3.56, + "learning_rate": 1.2250703605659688e-05, + "loss": 0.1707, + "step": 8500 + }, + { + "epoch": 3.56, + "learning_rate": 1.2249051664372078e-05, + "loss": 0.1901, + "step": 8501 + }, + { + "epoch": 3.56, + "learning_rate": 1.2247399658437475e-05, + "loss": 0.2087, + "step": 8502 + }, + { + "epoch": 3.56, + "learning_rate": 1.224574758790336e-05, + "loss": 0.2222, + "step": 8503 + }, + { + "epoch": 3.56, + "learning_rate": 1.2244095452817227e-05, + "loss": 0.208, + "step": 8504 + }, + { + "epoch": 3.56, + "learning_rate": 1.2242443253226558e-05, + "loss": 0.2085, + "step": 8505 + }, + { + "epoch": 3.56, + "learning_rate": 1.224079098917885e-05, + "loss": 0.2235, + "step": 8506 + }, + { + "epoch": 3.56, + "learning_rate": 1.2239138660721593e-05, + "loss": 0.1747, + "step": 8507 + }, + { + "epoch": 3.56, + "learning_rate": 1.2237486267902284e-05, + "loss": 0.1916, + "step": 8508 + }, + { + "epoch": 3.56, + "learning_rate": 1.2235833810768417e-05, + "loss": 0.1648, + "step": 8509 + }, + { + "epoch": 3.56, + "learning_rate": 1.2234181289367489e-05, + "loss": 0.2035, + "step": 8510 + }, + { + "epoch": 3.56, + "learning_rate": 1.2232528703747002e-05, + "loss": 0.2268, + "step": 8511 + }, + { + "epoch": 3.56, + "learning_rate": 1.2230876053954462e-05, + "loss": 0.1868, + "step": 8512 + }, + { + "epoch": 3.56, + "learning_rate": 1.2229223340037364e-05, + "loss": 0.1953, + "step": 8513 + }, + { + "epoch": 3.56, + "learning_rate": 1.2227570562043223e-05, + "loss": 0.1992, + "step": 8514 + }, + { + "epoch": 3.56, + "learning_rate": 1.2225917720019544e-05, + "loss": 0.1824, + "step": 8515 + }, + { + "epoch": 3.57, + "learning_rate": 1.2224264814013833e-05, + "loss": 0.1768, + "step": 8516 + }, + { + "epoch": 3.57, + "learning_rate": 1.2222611844073604e-05, + "loss": 0.185, + "step": 8517 + }, + { + "epoch": 3.57, + "learning_rate": 1.2220958810246371e-05, + "loss": 0.1673, + "step": 8518 + }, + { + "epoch": 3.57, + "learning_rate": 1.2219305712579647e-05, + "loss": 0.1732, + "step": 8519 + }, + { + "epoch": 3.57, + "learning_rate": 1.221765255112095e-05, + "loss": 0.2109, + "step": 8520 + }, + { + "epoch": 3.57, + "learning_rate": 1.2215999325917797e-05, + "loss": 0.2072, + "step": 8521 + }, + { + "epoch": 3.57, + "learning_rate": 1.2214346037017715e-05, + "loss": 0.2434, + "step": 8522 + }, + { + "epoch": 3.57, + "learning_rate": 1.2212692684468215e-05, + "loss": 0.1588, + "step": 8523 + }, + { + "epoch": 3.57, + "learning_rate": 1.221103926831683e-05, + "loss": 0.2165, + "step": 8524 + }, + { + "epoch": 3.57, + "learning_rate": 1.2209385788611084e-05, + "loss": 0.2039, + "step": 8525 + }, + { + "epoch": 3.57, + "learning_rate": 1.2207732245398502e-05, + "loss": 0.1687, + "step": 8526 + }, + { + "epoch": 3.57, + "learning_rate": 1.2206078638726615e-05, + "loss": 0.1956, + "step": 8527 + }, + { + "epoch": 3.57, + "learning_rate": 1.2204424968642957e-05, + "loss": 0.2377, + "step": 8528 + }, + { + "epoch": 3.57, + "learning_rate": 1.220277123519506e-05, + "loss": 0.2237, + "step": 8529 + }, + { + "epoch": 3.57, + "learning_rate": 1.2201117438430456e-05, + "loss": 0.1893, + "step": 8530 + }, + { + "epoch": 3.57, + "learning_rate": 1.2199463578396688e-05, + "loss": 0.1917, + "step": 8531 + }, + { + "epoch": 3.57, + "learning_rate": 1.219780965514129e-05, + "loss": 0.2059, + "step": 8532 + }, + { + "epoch": 3.57, + "learning_rate": 1.2196155668711804e-05, + "loss": 0.1924, + "step": 8533 + }, + { + "epoch": 3.57, + "learning_rate": 1.219450161915577e-05, + "loss": 0.1509, + "step": 8534 + }, + { + "epoch": 3.57, + "learning_rate": 1.2192847506520735e-05, + "loss": 0.1846, + "step": 8535 + }, + { + "epoch": 3.57, + "learning_rate": 1.2191193330854246e-05, + "loss": 0.1921, + "step": 8536 + }, + { + "epoch": 3.57, + "learning_rate": 1.2189539092203849e-05, + "loss": 0.1988, + "step": 8537 + }, + { + "epoch": 3.57, + "learning_rate": 1.218788479061709e-05, + "loss": 0.1842, + "step": 8538 + }, + { + "epoch": 3.58, + "learning_rate": 1.2186230426141528e-05, + "loss": 0.1653, + "step": 8539 + }, + { + "epoch": 3.58, + "learning_rate": 1.2184575998824713e-05, + "loss": 0.1869, + "step": 8540 + }, + { + "epoch": 3.58, + "learning_rate": 1.2182921508714196e-05, + "loss": 0.1924, + "step": 8541 + }, + { + "epoch": 3.58, + "learning_rate": 1.218126695585754e-05, + "loss": 0.1826, + "step": 8542 + }, + { + "epoch": 3.58, + "learning_rate": 1.2179612340302302e-05, + "loss": 0.1999, + "step": 8543 + }, + { + "epoch": 3.58, + "learning_rate": 1.2177957662096038e-05, + "loss": 0.217, + "step": 8544 + }, + { + "epoch": 3.58, + "learning_rate": 1.2176302921286316e-05, + "loss": 0.2095, + "step": 8545 + }, + { + "epoch": 3.58, + "learning_rate": 1.2174648117920698e-05, + "loss": 0.202, + "step": 8546 + }, + { + "epoch": 3.58, + "learning_rate": 1.2172993252046747e-05, + "loss": 0.1597, + "step": 8547 + }, + { + "epoch": 3.58, + "learning_rate": 1.2171338323712034e-05, + "loss": 0.1915, + "step": 8548 + }, + { + "epoch": 3.58, + "learning_rate": 1.216968333296413e-05, + "loss": 0.1843, + "step": 8549 + }, + { + "epoch": 3.58, + "learning_rate": 1.2168028279850603e-05, + "loss": 0.2172, + "step": 8550 + }, + { + "epoch": 3.58, + "learning_rate": 1.2166373164419026e-05, + "loss": 0.1981, + "step": 8551 + }, + { + "epoch": 3.58, + "learning_rate": 1.2164717986716977e-05, + "loss": 0.1738, + "step": 8552 + }, + { + "epoch": 3.58, + "learning_rate": 1.2163062746792027e-05, + "loss": 0.1915, + "step": 8553 + }, + { + "epoch": 3.58, + "learning_rate": 1.2161407444691761e-05, + "loss": 0.1887, + "step": 8554 + }, + { + "epoch": 3.58, + "learning_rate": 1.2159752080463753e-05, + "loss": 0.1902, + "step": 8555 + }, + { + "epoch": 3.58, + "learning_rate": 1.2158096654155592e-05, + "loss": 0.1652, + "step": 8556 + }, + { + "epoch": 3.58, + "learning_rate": 1.2156441165814855e-05, + "loss": 0.187, + "step": 8557 + }, + { + "epoch": 3.58, + "learning_rate": 1.2154785615489129e-05, + "loss": 0.1937, + "step": 8558 + }, + { + "epoch": 3.58, + "learning_rate": 1.2153130003226006e-05, + "loss": 0.1814, + "step": 8559 + }, + { + "epoch": 3.58, + "learning_rate": 1.215147432907307e-05, + "loss": 0.1693, + "step": 8560 + }, + { + "epoch": 3.58, + "learning_rate": 1.2149818593077912e-05, + "loss": 0.1944, + "step": 8561 + }, + { + "epoch": 3.58, + "learning_rate": 1.2148162795288127e-05, + "loss": 0.1962, + "step": 8562 + }, + { + "epoch": 3.59, + "learning_rate": 1.214650693575131e-05, + "loss": 0.1754, + "step": 8563 + }, + { + "epoch": 3.59, + "learning_rate": 1.2144851014515055e-05, + "loss": 0.1715, + "step": 8564 + }, + { + "epoch": 3.59, + "learning_rate": 1.2143195031626962e-05, + "loss": 0.1796, + "step": 8565 + }, + { + "epoch": 3.59, + "learning_rate": 1.214153898713463e-05, + "loss": 0.1753, + "step": 8566 + }, + { + "epoch": 3.59, + "learning_rate": 1.2139882881085658e-05, + "loss": 0.1809, + "step": 8567 + }, + { + "epoch": 3.59, + "learning_rate": 1.2138226713527652e-05, + "loss": 0.1781, + "step": 8568 + }, + { + "epoch": 3.59, + "learning_rate": 1.2136570484508217e-05, + "loss": 0.2059, + "step": 8569 + }, + { + "epoch": 3.59, + "learning_rate": 1.213491419407496e-05, + "loss": 0.1986, + "step": 8570 + }, + { + "epoch": 3.59, + "learning_rate": 1.2133257842275487e-05, + "loss": 0.193, + "step": 8571 + }, + { + "epoch": 3.59, + "learning_rate": 1.2131601429157413e-05, + "loss": 0.2063, + "step": 8572 + }, + { + "epoch": 3.59, + "learning_rate": 1.2129944954768346e-05, + "loss": 0.1816, + "step": 8573 + }, + { + "epoch": 3.59, + "learning_rate": 1.2128288419155903e-05, + "loss": 0.2111, + "step": 8574 + }, + { + "epoch": 3.59, + "learning_rate": 1.2126631822367693e-05, + "loss": 0.1456, + "step": 8575 + }, + { + "epoch": 3.59, + "learning_rate": 1.2124975164451345e-05, + "loss": 0.1482, + "step": 8576 + }, + { + "epoch": 3.59, + "learning_rate": 1.212331844545447e-05, + "loss": 0.2008, + "step": 8577 + }, + { + "epoch": 3.59, + "learning_rate": 1.2121661665424688e-05, + "loss": 0.176, + "step": 8578 + }, + { + "epoch": 3.59, + "learning_rate": 1.2120004824409626e-05, + "loss": 0.1779, + "step": 8579 + }, + { + "epoch": 3.59, + "learning_rate": 1.2118347922456905e-05, + "loss": 0.162, + "step": 8580 + }, + { + "epoch": 3.59, + "learning_rate": 1.2116690959614154e-05, + "loss": 0.1954, + "step": 8581 + }, + { + "epoch": 3.59, + "learning_rate": 1.2115033935929e-05, + "loss": 0.183, + "step": 8582 + }, + { + "epoch": 3.59, + "learning_rate": 1.211337685144907e-05, + "loss": 0.1416, + "step": 8583 + }, + { + "epoch": 3.59, + "learning_rate": 1.2111719706222e-05, + "loss": 0.2038, + "step": 8584 + }, + { + "epoch": 3.59, + "learning_rate": 1.2110062500295418e-05, + "loss": 0.1634, + "step": 8585 + }, + { + "epoch": 3.59, + "learning_rate": 1.2108405233716967e-05, + "loss": 0.1703, + "step": 8586 + }, + { + "epoch": 3.6, + "learning_rate": 1.2106747906534275e-05, + "loss": 0.193, + "step": 8587 + }, + { + "epoch": 3.6, + "learning_rate": 1.2105090518794986e-05, + "loss": 0.1664, + "step": 8588 + }, + { + "epoch": 3.6, + "learning_rate": 1.2103433070546737e-05, + "loss": 0.1917, + "step": 8589 + }, + { + "epoch": 3.6, + "learning_rate": 1.2101775561837172e-05, + "loss": 0.1935, + "step": 8590 + }, + { + "epoch": 3.6, + "learning_rate": 1.2100117992713931e-05, + "loss": 0.1685, + "step": 8591 + }, + { + "epoch": 3.6, + "learning_rate": 1.2098460363224662e-05, + "loss": 0.1723, + "step": 8592 + }, + { + "epoch": 3.6, + "learning_rate": 1.2096802673417014e-05, + "loss": 0.1534, + "step": 8593 + }, + { + "epoch": 3.6, + "learning_rate": 1.2095144923338633e-05, + "loss": 0.1846, + "step": 8594 + }, + { + "epoch": 3.6, + "learning_rate": 1.209348711303717e-05, + "loss": 0.217, + "step": 8595 + }, + { + "epoch": 3.6, + "learning_rate": 1.2091829242560278e-05, + "loss": 0.2127, + "step": 8596 + }, + { + "epoch": 3.6, + "learning_rate": 1.209017131195561e-05, + "loss": 0.182, + "step": 8597 + }, + { + "epoch": 3.6, + "learning_rate": 1.2088513321270823e-05, + "loss": 0.2203, + "step": 8598 + }, + { + "epoch": 3.6, + "learning_rate": 1.2086855270553575e-05, + "loss": 0.2172, + "step": 8599 + }, + { + "epoch": 3.6, + "learning_rate": 1.2085197159851525e-05, + "loss": 0.2137, + "step": 8600 + }, + { + "epoch": 3.6, + "learning_rate": 1.208353898921233e-05, + "loss": 0.1924, + "step": 8601 + }, + { + "epoch": 3.6, + "learning_rate": 1.2081880758683656e-05, + "loss": 0.22, + "step": 8602 + }, + { + "epoch": 3.6, + "learning_rate": 1.208022246831317e-05, + "loss": 0.2001, + "step": 8603 + }, + { + "epoch": 3.6, + "learning_rate": 1.2078564118148531e-05, + "loss": 0.2161, + "step": 8604 + }, + { + "epoch": 3.6, + "learning_rate": 1.2076905708237412e-05, + "loss": 0.19, + "step": 8605 + }, + { + "epoch": 3.6, + "learning_rate": 1.2075247238627482e-05, + "loss": 0.1804, + "step": 8606 + }, + { + "epoch": 3.6, + "learning_rate": 1.2073588709366413e-05, + "loss": 0.2071, + "step": 8607 + }, + { + "epoch": 3.6, + "learning_rate": 1.2071930120501874e-05, + "loss": 0.2102, + "step": 8608 + }, + { + "epoch": 3.6, + "learning_rate": 1.2070271472081542e-05, + "loss": 0.1457, + "step": 8609 + }, + { + "epoch": 3.6, + "learning_rate": 1.2068612764153099e-05, + "loss": 0.2246, + "step": 8610 + }, + { + "epoch": 3.61, + "learning_rate": 1.2066953996764215e-05, + "loss": 0.1485, + "step": 8611 + }, + { + "epoch": 3.61, + "learning_rate": 1.2065295169962573e-05, + "loss": 0.1568, + "step": 8612 + }, + { + "epoch": 3.61, + "learning_rate": 1.206363628379585e-05, + "loss": 0.1886, + "step": 8613 + }, + { + "epoch": 3.61, + "learning_rate": 1.2061977338311737e-05, + "loss": 0.1763, + "step": 8614 + }, + { + "epoch": 3.61, + "learning_rate": 1.2060318333557915e-05, + "loss": 0.166, + "step": 8615 + }, + { + "epoch": 3.61, + "learning_rate": 1.2058659269582067e-05, + "loss": 0.1715, + "step": 8616 + }, + { + "epoch": 3.61, + "learning_rate": 1.205700014643189e-05, + "loss": 0.2064, + "step": 8617 + }, + { + "epoch": 3.61, + "learning_rate": 1.2055340964155067e-05, + "loss": 0.2055, + "step": 8618 + }, + { + "epoch": 3.61, + "learning_rate": 1.2053681722799291e-05, + "loss": 0.2265, + "step": 8619 + }, + { + "epoch": 3.61, + "learning_rate": 1.2052022422412259e-05, + "loss": 0.2, + "step": 8620 + }, + { + "epoch": 3.61, + "learning_rate": 1.2050363063041662e-05, + "loss": 0.1541, + "step": 8621 + }, + { + "epoch": 3.61, + "learning_rate": 1.2048703644735198e-05, + "loss": 0.1835, + "step": 8622 + }, + { + "epoch": 3.61, + "learning_rate": 1.2047044167540564e-05, + "loss": 0.1863, + "step": 8623 + }, + { + "epoch": 3.61, + "learning_rate": 1.2045384631505466e-05, + "loss": 0.1417, + "step": 8624 + }, + { + "epoch": 3.61, + "learning_rate": 1.2043725036677599e-05, + "loss": 0.1641, + "step": 8625 + }, + { + "epoch": 3.61, + "learning_rate": 1.204206538310467e-05, + "loss": 0.1823, + "step": 8626 + }, + { + "epoch": 3.61, + "learning_rate": 1.2040405670834385e-05, + "loss": 0.2046, + "step": 8627 + }, + { + "epoch": 3.61, + "learning_rate": 1.2038745899914446e-05, + "loss": 0.1768, + "step": 8628 + }, + { + "epoch": 3.61, + "learning_rate": 1.2037086070392567e-05, + "loss": 0.1889, + "step": 8629 + }, + { + "epoch": 3.61, + "learning_rate": 1.2035426182316459e-05, + "loss": 0.2021, + "step": 8630 + }, + { + "epoch": 3.61, + "learning_rate": 1.2033766235733828e-05, + "loss": 0.2051, + "step": 8631 + }, + { + "epoch": 3.61, + "learning_rate": 1.2032106230692394e-05, + "loss": 0.2008, + "step": 8632 + }, + { + "epoch": 3.61, + "learning_rate": 1.2030446167239869e-05, + "loss": 0.171, + "step": 8633 + }, + { + "epoch": 3.61, + "learning_rate": 1.202878604542397e-05, + "loss": 0.2133, + "step": 8634 + }, + { + "epoch": 3.62, + "learning_rate": 1.2027125865292417e-05, + "loss": 0.1647, + "step": 8635 + }, + { + "epoch": 3.62, + "learning_rate": 1.202546562689293e-05, + "loss": 0.1962, + "step": 8636 + }, + { + "epoch": 3.62, + "learning_rate": 1.202380533027323e-05, + "loss": 0.2258, + "step": 8637 + }, + { + "epoch": 3.62, + "learning_rate": 1.2022144975481043e-05, + "loss": 0.1678, + "step": 8638 + }, + { + "epoch": 3.62, + "learning_rate": 1.202048456256409e-05, + "loss": 0.1997, + "step": 8639 + }, + { + "epoch": 3.62, + "learning_rate": 1.2018824091570103e-05, + "loss": 0.1733, + "step": 8640 + }, + { + "epoch": 3.62, + "learning_rate": 1.201716356254681e-05, + "loss": 0.1818, + "step": 8641 + }, + { + "epoch": 3.62, + "learning_rate": 1.201550297554194e-05, + "loss": 0.1849, + "step": 8642 + }, + { + "epoch": 3.62, + "learning_rate": 1.2013842330603226e-05, + "loss": 0.1728, + "step": 8643 + }, + { + "epoch": 3.62, + "learning_rate": 1.20121816277784e-05, + "loss": 0.1764, + "step": 8644 + }, + { + "epoch": 3.62, + "learning_rate": 1.20105208671152e-05, + "loss": 0.2053, + "step": 8645 + }, + { + "epoch": 3.62, + "learning_rate": 1.2008860048661362e-05, + "loss": 0.1708, + "step": 8646 + }, + { + "epoch": 3.62, + "learning_rate": 1.200719917246462e-05, + "loss": 0.1796, + "step": 8647 + }, + { + "epoch": 3.62, + "learning_rate": 1.2005538238572724e-05, + "loss": 0.1694, + "step": 8648 + }, + { + "epoch": 3.62, + "learning_rate": 1.2003877247033411e-05, + "loss": 0.1919, + "step": 8649 + }, + { + "epoch": 3.62, + "learning_rate": 1.200221619789442e-05, + "loss": 0.242, + "step": 8650 + }, + { + "epoch": 3.62, + "learning_rate": 1.2000555091203507e-05, + "loss": 0.1495, + "step": 8651 + }, + { + "epoch": 3.62, + "learning_rate": 1.1998893927008411e-05, + "loss": 0.1781, + "step": 8652 + }, + { + "epoch": 3.62, + "learning_rate": 1.199723270535688e-05, + "loss": 0.1978, + "step": 8653 + }, + { + "epoch": 3.62, + "learning_rate": 1.1995571426296671e-05, + "loss": 0.2193, + "step": 8654 + }, + { + "epoch": 3.62, + "learning_rate": 1.1993910089875531e-05, + "loss": 0.1623, + "step": 8655 + }, + { + "epoch": 3.62, + "learning_rate": 1.1992248696141215e-05, + "loss": 0.2132, + "step": 8656 + }, + { + "epoch": 3.62, + "learning_rate": 1.1990587245141477e-05, + "loss": 0.1865, + "step": 8657 + }, + { + "epoch": 3.62, + "learning_rate": 1.1988925736924078e-05, + "loss": 0.1778, + "step": 8658 + }, + { + "epoch": 3.63, + "learning_rate": 1.198726417153677e-05, + "loss": 0.1642, + "step": 8659 + }, + { + "epoch": 3.63, + "learning_rate": 1.1985602549027318e-05, + "loss": 0.1997, + "step": 8660 + }, + { + "epoch": 3.63, + "learning_rate": 1.1983940869443484e-05, + "loss": 0.2021, + "step": 8661 + }, + { + "epoch": 3.63, + "learning_rate": 1.1982279132833028e-05, + "loss": 0.1802, + "step": 8662 + }, + { + "epoch": 3.63, + "learning_rate": 1.1980617339243719e-05, + "loss": 0.215, + "step": 8663 + }, + { + "epoch": 3.63, + "learning_rate": 1.1978955488723323e-05, + "loss": 0.1522, + "step": 8664 + }, + { + "epoch": 3.63, + "learning_rate": 1.1977293581319605e-05, + "loss": 0.1754, + "step": 8665 + }, + { + "epoch": 3.63, + "learning_rate": 1.1975631617080339e-05, + "loss": 0.1967, + "step": 8666 + }, + { + "epoch": 3.63, + "learning_rate": 1.1973969596053295e-05, + "loss": 0.1964, + "step": 8667 + }, + { + "epoch": 3.63, + "learning_rate": 1.1972307518286248e-05, + "loss": 0.163, + "step": 8668 + }, + { + "epoch": 3.63, + "learning_rate": 1.1970645383826971e-05, + "loss": 0.2226, + "step": 8669 + }, + { + "epoch": 3.63, + "learning_rate": 1.1968983192723238e-05, + "loss": 0.1526, + "step": 8670 + }, + { + "epoch": 3.63, + "learning_rate": 1.1967320945022836e-05, + "loss": 0.1703, + "step": 8671 + }, + { + "epoch": 3.63, + "learning_rate": 1.1965658640773535e-05, + "loss": 0.1807, + "step": 8672 + }, + { + "epoch": 3.63, + "learning_rate": 1.196399628002312e-05, + "loss": 0.1969, + "step": 8673 + }, + { + "epoch": 3.63, + "learning_rate": 1.1962333862819379e-05, + "loss": 0.1891, + "step": 8674 + }, + { + "epoch": 3.63, + "learning_rate": 1.1960671389210087e-05, + "loss": 0.1824, + "step": 8675 + }, + { + "epoch": 3.63, + "learning_rate": 1.195900885924304e-05, + "loss": 0.1831, + "step": 8676 + }, + { + "epoch": 3.63, + "learning_rate": 1.1957346272966019e-05, + "loss": 0.1629, + "step": 8677 + }, + { + "epoch": 3.63, + "learning_rate": 1.195568363042682e-05, + "loss": 0.187, + "step": 8678 + }, + { + "epoch": 3.63, + "learning_rate": 1.1954020931673229e-05, + "loss": 0.1698, + "step": 8679 + }, + { + "epoch": 3.63, + "learning_rate": 1.1952358176753038e-05, + "loss": 0.1691, + "step": 8680 + }, + { + "epoch": 3.63, + "learning_rate": 1.1950695365714046e-05, + "loss": 0.1688, + "step": 8681 + }, + { + "epoch": 3.63, + "learning_rate": 1.1949032498604047e-05, + "loss": 0.232, + "step": 8682 + }, + { + "epoch": 3.64, + "learning_rate": 1.1947369575470837e-05, + "loss": 0.1872, + "step": 8683 + }, + { + "epoch": 3.64, + "learning_rate": 1.1945706596362216e-05, + "loss": 0.1597, + "step": 8684 + }, + { + "epoch": 3.64, + "learning_rate": 1.1944043561325989e-05, + "loss": 0.2131, + "step": 8685 + }, + { + "epoch": 3.64, + "learning_rate": 1.194238047040995e-05, + "loss": 0.1646, + "step": 8686 + }, + { + "epoch": 3.64, + "learning_rate": 1.194071732366191e-05, + "loss": 0.1981, + "step": 8687 + }, + { + "epoch": 3.64, + "learning_rate": 1.1939054121129677e-05, + "loss": 0.1612, + "step": 8688 + }, + { + "epoch": 3.64, + "learning_rate": 1.1937390862861047e-05, + "loss": 0.1734, + "step": 8689 + }, + { + "epoch": 3.64, + "learning_rate": 1.1935727548903842e-05, + "loss": 0.1866, + "step": 8690 + }, + { + "epoch": 3.64, + "learning_rate": 1.1934064179305861e-05, + "loss": 0.1663, + "step": 8691 + }, + { + "epoch": 3.64, + "learning_rate": 1.1932400754114925e-05, + "loss": 0.1951, + "step": 8692 + }, + { + "epoch": 3.64, + "learning_rate": 1.1930737273378842e-05, + "loss": 0.1756, + "step": 8693 + }, + { + "epoch": 3.64, + "learning_rate": 1.1929073737145432e-05, + "loss": 0.1713, + "step": 8694 + }, + { + "epoch": 3.64, + "learning_rate": 1.1927410145462507e-05, + "loss": 0.2078, + "step": 8695 + }, + { + "epoch": 3.64, + "learning_rate": 1.1925746498377888e-05, + "loss": 0.1786, + "step": 8696 + }, + { + "epoch": 3.64, + "learning_rate": 1.1924082795939394e-05, + "loss": 0.1771, + "step": 8697 + }, + { + "epoch": 3.64, + "learning_rate": 1.192241903819485e-05, + "loss": 0.1918, + "step": 8698 + }, + { + "epoch": 3.64, + "learning_rate": 1.1920755225192074e-05, + "loss": 0.167, + "step": 8699 + }, + { + "epoch": 3.64, + "learning_rate": 1.1919091356978894e-05, + "loss": 0.1738, + "step": 8700 + }, + { + "epoch": 3.64, + "learning_rate": 1.1917427433603138e-05, + "loss": 0.203, + "step": 8701 + }, + { + "epoch": 3.64, + "learning_rate": 1.1915763455112633e-05, + "loss": 0.2271, + "step": 8702 + }, + { + "epoch": 3.64, + "learning_rate": 1.1914099421555203e-05, + "loss": 0.1783, + "step": 8703 + }, + { + "epoch": 3.64, + "learning_rate": 1.1912435332978684e-05, + "loss": 0.1941, + "step": 8704 + }, + { + "epoch": 3.64, + "learning_rate": 1.1910771189430913e-05, + "loss": 0.2174, + "step": 8705 + }, + { + "epoch": 3.64, + "learning_rate": 1.1909106990959716e-05, + "loss": 0.1742, + "step": 8706 + }, + { + "epoch": 3.65, + "learning_rate": 1.1907442737612932e-05, + "loss": 0.1669, + "step": 8707 + }, + { + "epoch": 3.65, + "learning_rate": 1.1905778429438401e-05, + "loss": 0.2056, + "step": 8708 + }, + { + "epoch": 3.65, + "learning_rate": 1.190411406648396e-05, + "loss": 0.2232, + "step": 8709 + }, + { + "epoch": 3.65, + "learning_rate": 1.190244964879745e-05, + "loss": 0.1932, + "step": 8710 + }, + { + "epoch": 3.65, + "learning_rate": 1.190078517642671e-05, + "loss": 0.1979, + "step": 8711 + }, + { + "epoch": 3.65, + "learning_rate": 1.189912064941959e-05, + "loss": 0.1841, + "step": 8712 + }, + { + "epoch": 3.65, + "learning_rate": 1.1897456067823934e-05, + "loss": 0.1931, + "step": 8713 + }, + { + "epoch": 3.65, + "learning_rate": 1.1895791431687585e-05, + "loss": 0.1672, + "step": 8714 + }, + { + "epoch": 3.65, + "learning_rate": 1.1894126741058393e-05, + "loss": 0.1963, + "step": 8715 + }, + { + "epoch": 3.65, + "learning_rate": 1.189246199598421e-05, + "loss": 0.2083, + "step": 8716 + }, + { + "epoch": 3.65, + "learning_rate": 1.1890797196512884e-05, + "loss": 0.218, + "step": 8717 + }, + { + "epoch": 3.65, + "learning_rate": 1.1889132342692272e-05, + "loss": 0.1758, + "step": 8718 + }, + { + "epoch": 3.65, + "learning_rate": 1.1887467434570227e-05, + "loss": 0.1998, + "step": 8719 + }, + { + "epoch": 3.65, + "learning_rate": 1.1885802472194606e-05, + "loss": 0.1669, + "step": 8720 + }, + { + "epoch": 3.65, + "learning_rate": 1.1884137455613266e-05, + "loss": 0.1861, + "step": 8721 + }, + { + "epoch": 3.65, + "learning_rate": 1.1882472384874067e-05, + "loss": 0.2202, + "step": 8722 + }, + { + "epoch": 3.65, + "learning_rate": 1.1880807260024867e-05, + "loss": 0.1837, + "step": 8723 + }, + { + "epoch": 3.65, + "learning_rate": 1.1879142081113535e-05, + "loss": 0.1654, + "step": 8724 + }, + { + "epoch": 3.65, + "learning_rate": 1.187747684818793e-05, + "loss": 0.1711, + "step": 8725 + }, + { + "epoch": 3.65, + "learning_rate": 1.1875811561295921e-05, + "loss": 0.1968, + "step": 8726 + }, + { + "epoch": 3.65, + "learning_rate": 1.187414622048537e-05, + "loss": 0.1816, + "step": 8727 + }, + { + "epoch": 3.65, + "learning_rate": 1.1872480825804149e-05, + "loss": 0.1843, + "step": 8728 + }, + { + "epoch": 3.65, + "learning_rate": 1.1870815377300131e-05, + "loss": 0.171, + "step": 8729 + }, + { + "epoch": 3.66, + "learning_rate": 1.1869149875021184e-05, + "loss": 0.1602, + "step": 8730 + }, + { + "epoch": 3.66, + "learning_rate": 1.186748431901518e-05, + "loss": 0.1338, + "step": 8731 + }, + { + "epoch": 3.66, + "learning_rate": 1.1865818709330001e-05, + "loss": 0.211, + "step": 8732 + }, + { + "epoch": 3.66, + "learning_rate": 1.1864153046013515e-05, + "loss": 0.2019, + "step": 8733 + }, + { + "epoch": 3.66, + "learning_rate": 1.1862487329113606e-05, + "loss": 0.2191, + "step": 8734 + }, + { + "epoch": 3.66, + "learning_rate": 1.1860821558678152e-05, + "loss": 0.1587, + "step": 8735 + }, + { + "epoch": 3.66, + "learning_rate": 1.1859155734755034e-05, + "loss": 0.22, + "step": 8736 + }, + { + "epoch": 3.66, + "learning_rate": 1.1857489857392133e-05, + "loss": 0.1619, + "step": 8737 + }, + { + "epoch": 3.66, + "learning_rate": 1.1855823926637333e-05, + "loss": 0.1612, + "step": 8738 + }, + { + "epoch": 3.66, + "learning_rate": 1.1854157942538522e-05, + "loss": 0.1793, + "step": 8739 + }, + { + "epoch": 3.66, + "learning_rate": 1.1852491905143586e-05, + "loss": 0.1781, + "step": 8740 + }, + { + "epoch": 3.66, + "learning_rate": 1.1850825814500413e-05, + "loss": 0.1956, + "step": 8741 + }, + { + "epoch": 3.66, + "learning_rate": 1.1849159670656897e-05, + "loss": 0.1916, + "step": 8742 + }, + { + "epoch": 3.66, + "learning_rate": 1.1847493473660925e-05, + "loss": 0.1949, + "step": 8743 + }, + { + "epoch": 3.66, + "learning_rate": 1.1845827223560394e-05, + "loss": 0.1847, + "step": 8744 + }, + { + "epoch": 3.66, + "learning_rate": 1.1844160920403197e-05, + "loss": 0.1929, + "step": 8745 + }, + { + "epoch": 3.66, + "learning_rate": 1.1842494564237231e-05, + "loss": 0.131, + "step": 8746 + }, + { + "epoch": 3.66, + "learning_rate": 1.1840828155110395e-05, + "loss": 0.1856, + "step": 8747 + }, + { + "epoch": 3.66, + "learning_rate": 1.1839161693070584e-05, + "loss": 0.1538, + "step": 8748 + }, + { + "epoch": 3.66, + "learning_rate": 1.1837495178165706e-05, + "loss": 0.1963, + "step": 8749 + }, + { + "epoch": 3.66, + "learning_rate": 1.1835828610443656e-05, + "loss": 0.1969, + "step": 8750 + }, + { + "epoch": 3.66, + "learning_rate": 1.1834161989952345e-05, + "loss": 0.1978, + "step": 8751 + }, + { + "epoch": 3.66, + "learning_rate": 1.1832495316739672e-05, + "loss": 0.2095, + "step": 8752 + }, + { + "epoch": 3.66, + "learning_rate": 1.1830828590853548e-05, + "loss": 0.1731, + "step": 8753 + }, + { + "epoch": 3.67, + "learning_rate": 1.1829161812341883e-05, + "loss": 0.1978, + "step": 8754 + }, + { + "epoch": 3.67, + "learning_rate": 1.182749498125258e-05, + "loss": 0.1767, + "step": 8755 + }, + { + "epoch": 3.67, + "learning_rate": 1.1825828097633562e-05, + "loss": 0.1752, + "step": 8756 + }, + { + "epoch": 3.67, + "learning_rate": 1.1824161161532731e-05, + "loss": 0.1937, + "step": 8757 + }, + { + "epoch": 3.67, + "learning_rate": 1.1822494172998008e-05, + "loss": 0.1824, + "step": 8758 + }, + { + "epoch": 3.67, + "learning_rate": 1.1820827132077306e-05, + "loss": 0.1865, + "step": 8759 + }, + { + "epoch": 3.67, + "learning_rate": 1.1819160038818547e-05, + "loss": 0.1787, + "step": 8760 + }, + { + "epoch": 3.67, + "learning_rate": 1.1817492893269645e-05, + "loss": 0.1637, + "step": 8761 + }, + { + "epoch": 3.67, + "learning_rate": 1.181582569547852e-05, + "loss": 0.2163, + "step": 8762 + }, + { + "epoch": 3.67, + "learning_rate": 1.18141584454931e-05, + "loss": 0.1649, + "step": 8763 + }, + { + "epoch": 3.67, + "learning_rate": 1.1812491143361305e-05, + "loss": 0.2028, + "step": 8764 + }, + { + "epoch": 3.67, + "learning_rate": 1.1810823789131059e-05, + "loss": 0.1657, + "step": 8765 + }, + { + "epoch": 3.67, + "learning_rate": 1.1809156382850291e-05, + "loss": 0.1668, + "step": 8766 + }, + { + "epoch": 3.67, + "learning_rate": 1.1807488924566929e-05, + "loss": 0.1856, + "step": 8767 + }, + { + "epoch": 3.67, + "learning_rate": 1.1805821414328897e-05, + "loss": 0.1946, + "step": 8768 + }, + { + "epoch": 3.67, + "learning_rate": 1.1804153852184137e-05, + "loss": 0.2073, + "step": 8769 + }, + { + "epoch": 3.67, + "learning_rate": 1.1802486238180575e-05, + "loss": 0.1951, + "step": 8770 + }, + { + "epoch": 3.67, + "learning_rate": 1.1800818572366144e-05, + "loss": 0.1153, + "step": 8771 + }, + { + "epoch": 3.67, + "learning_rate": 1.179915085478878e-05, + "loss": 0.1781, + "step": 8772 + }, + { + "epoch": 3.67, + "learning_rate": 1.1797483085496424e-05, + "loss": 0.2328, + "step": 8773 + }, + { + "epoch": 3.67, + "learning_rate": 1.1795815264537008e-05, + "loss": 0.2347, + "step": 8774 + }, + { + "epoch": 3.67, + "learning_rate": 1.1794147391958477e-05, + "loss": 0.2036, + "step": 8775 + }, + { + "epoch": 3.67, + "learning_rate": 1.1792479467808774e-05, + "loss": 0.205, + "step": 8776 + }, + { + "epoch": 3.67, + "learning_rate": 1.1790811492135837e-05, + "loss": 0.1933, + "step": 8777 + }, + { + "epoch": 3.68, + "learning_rate": 1.1789143464987611e-05, + "loss": 0.1803, + "step": 8778 + }, + { + "epoch": 3.68, + "learning_rate": 1.1787475386412048e-05, + "loss": 0.2011, + "step": 8779 + }, + { + "epoch": 3.68, + "learning_rate": 1.178580725645709e-05, + "loss": 0.1484, + "step": 8780 + }, + { + "epoch": 3.68, + "learning_rate": 1.1784139075170686e-05, + "loss": 0.1619, + "step": 8781 + }, + { + "epoch": 3.68, + "learning_rate": 1.1782470842600785e-05, + "loss": 0.1636, + "step": 8782 + }, + { + "epoch": 3.68, + "learning_rate": 1.1780802558795346e-05, + "loss": 0.166, + "step": 8783 + }, + { + "epoch": 3.68, + "learning_rate": 1.1779134223802315e-05, + "loss": 0.1686, + "step": 8784 + }, + { + "epoch": 3.68, + "learning_rate": 1.177746583766965e-05, + "loss": 0.224, + "step": 8785 + }, + { + "epoch": 3.68, + "learning_rate": 1.1775797400445307e-05, + "loss": 0.1735, + "step": 8786 + }, + { + "epoch": 3.68, + "learning_rate": 1.1774128912177245e-05, + "loss": 0.1761, + "step": 8787 + }, + { + "epoch": 3.68, + "learning_rate": 1.177246037291342e-05, + "loss": 0.2169, + "step": 8788 + }, + { + "epoch": 3.68, + "learning_rate": 1.1770791782701794e-05, + "loss": 0.1702, + "step": 8789 + }, + { + "epoch": 3.68, + "learning_rate": 1.176912314159033e-05, + "loss": 0.1961, + "step": 8790 + }, + { + "epoch": 3.68, + "learning_rate": 1.1767454449626992e-05, + "loss": 0.2112, + "step": 8791 + }, + { + "epoch": 3.68, + "learning_rate": 1.1765785706859744e-05, + "loss": 0.2216, + "step": 8792 + }, + { + "epoch": 3.68, + "learning_rate": 1.1764116913336552e-05, + "loss": 0.2142, + "step": 8793 + }, + { + "epoch": 3.68, + "learning_rate": 1.1762448069105387e-05, + "loss": 0.2356, + "step": 8794 + }, + { + "epoch": 3.68, + "learning_rate": 1.1760779174214215e-05, + "loss": 0.1855, + "step": 8795 + }, + { + "epoch": 3.68, + "learning_rate": 1.1759110228711006e-05, + "loss": 0.2006, + "step": 8796 + }, + { + "epoch": 3.68, + "learning_rate": 1.1757441232643737e-05, + "loss": 0.2181, + "step": 8797 + }, + { + "epoch": 3.68, + "learning_rate": 1.1755772186060379e-05, + "loss": 0.1845, + "step": 8798 + }, + { + "epoch": 3.68, + "learning_rate": 1.1754103089008905e-05, + "loss": 0.1632, + "step": 8799 + }, + { + "epoch": 3.68, + "learning_rate": 1.1752433941537297e-05, + "loss": 0.1701, + "step": 8800 + }, + { + "epoch": 3.68, + "learning_rate": 1.175076474369353e-05, + "loss": 0.1762, + "step": 8801 + }, + { + "epoch": 3.69, + "learning_rate": 1.1749095495525584e-05, + "loss": 0.2207, + "step": 8802 + }, + { + "epoch": 3.69, + "learning_rate": 1.1747426197081443e-05, + "loss": 0.2055, + "step": 8803 + }, + { + "epoch": 3.69, + "learning_rate": 1.1745756848409083e-05, + "loss": 0.1968, + "step": 8804 + }, + { + "epoch": 3.69, + "learning_rate": 1.1744087449556492e-05, + "loss": 0.164, + "step": 8805 + }, + { + "epoch": 3.69, + "learning_rate": 1.1742418000571653e-05, + "loss": 0.2076, + "step": 8806 + }, + { + "epoch": 3.69, + "learning_rate": 1.1740748501502559e-05, + "loss": 0.1933, + "step": 8807 + }, + { + "epoch": 3.69, + "learning_rate": 1.1739078952397189e-05, + "loss": 0.1662, + "step": 8808 + }, + { + "epoch": 3.69, + "learning_rate": 1.1737409353303541e-05, + "loss": 0.1819, + "step": 8809 + }, + { + "epoch": 3.69, + "learning_rate": 1.1735739704269605e-05, + "loss": 0.1808, + "step": 8810 + }, + { + "epoch": 3.69, + "learning_rate": 1.1734070005343368e-05, + "loss": 0.184, + "step": 8811 + }, + { + "epoch": 3.69, + "learning_rate": 1.1732400256572827e-05, + "loss": 0.1854, + "step": 8812 + }, + { + "epoch": 3.69, + "learning_rate": 1.1730730458005981e-05, + "loss": 0.1776, + "step": 8813 + }, + { + "epoch": 3.69, + "learning_rate": 1.1729060609690824e-05, + "loss": 0.188, + "step": 8814 + }, + { + "epoch": 3.69, + "learning_rate": 1.172739071167535e-05, + "loss": 0.1515, + "step": 8815 + }, + { + "epoch": 3.69, + "learning_rate": 1.1725720764007564e-05, + "loss": 0.2115, + "step": 8816 + }, + { + "epoch": 3.69, + "learning_rate": 1.1724050766735469e-05, + "loss": 0.1856, + "step": 8817 + }, + { + "epoch": 3.69, + "learning_rate": 1.1722380719907063e-05, + "loss": 0.1764, + "step": 8818 + }, + { + "epoch": 3.69, + "learning_rate": 1.172071062357035e-05, + "loss": 0.1929, + "step": 8819 + }, + { + "epoch": 3.69, + "learning_rate": 1.171904047777334e-05, + "loss": 0.1946, + "step": 8820 + }, + { + "epoch": 3.69, + "learning_rate": 1.1717370282564034e-05, + "loss": 0.1833, + "step": 8821 + }, + { + "epoch": 3.69, + "learning_rate": 1.1715700037990443e-05, + "loss": 0.1939, + "step": 8822 + }, + { + "epoch": 3.69, + "learning_rate": 1.171402974410058e-05, + "loss": 0.167, + "step": 8823 + }, + { + "epoch": 3.69, + "learning_rate": 1.171235940094245e-05, + "loss": 0.223, + "step": 8824 + }, + { + "epoch": 3.69, + "learning_rate": 1.1710689008564068e-05, + "loss": 0.1941, + "step": 8825 + }, + { + "epoch": 3.7, + "learning_rate": 1.1709018567013455e-05, + "loss": 0.1697, + "step": 8826 + }, + { + "epoch": 3.7, + "learning_rate": 1.1707348076338616e-05, + "loss": 0.1743, + "step": 8827 + }, + { + "epoch": 3.7, + "learning_rate": 1.170567753658757e-05, + "loss": 0.2077, + "step": 8828 + }, + { + "epoch": 3.7, + "learning_rate": 1.1704006947808338e-05, + "loss": 0.1682, + "step": 8829 + }, + { + "epoch": 3.7, + "learning_rate": 1.1702336310048936e-05, + "loss": 0.1975, + "step": 8830 + }, + { + "epoch": 3.7, + "learning_rate": 1.170066562335739e-05, + "loss": 0.1564, + "step": 8831 + }, + { + "epoch": 3.7, + "learning_rate": 1.1698994887781718e-05, + "loss": 0.1795, + "step": 8832 + }, + { + "epoch": 3.7, + "learning_rate": 1.1697324103369947e-05, + "loss": 0.1962, + "step": 8833 + }, + { + "epoch": 3.7, + "learning_rate": 1.1695653270170102e-05, + "loss": 0.2126, + "step": 8834 + }, + { + "epoch": 3.7, + "learning_rate": 1.1693982388230207e-05, + "loss": 0.2115, + "step": 8835 + }, + { + "epoch": 3.7, + "learning_rate": 1.1692311457598291e-05, + "loss": 0.2035, + "step": 8836 + }, + { + "epoch": 3.7, + "learning_rate": 1.1690640478322386e-05, + "loss": 0.1825, + "step": 8837 + }, + { + "epoch": 3.7, + "learning_rate": 1.168896945045052e-05, + "loss": 0.1644, + "step": 8838 + }, + { + "epoch": 3.7, + "learning_rate": 1.1687298374030724e-05, + "loss": 0.1886, + "step": 8839 + }, + { + "epoch": 3.7, + "learning_rate": 1.1685627249111034e-05, + "loss": 0.1961, + "step": 8840 + }, + { + "epoch": 3.7, + "learning_rate": 1.1683956075739487e-05, + "loss": 0.2219, + "step": 8841 + }, + { + "epoch": 3.7, + "learning_rate": 1.1682284853964114e-05, + "loss": 0.1937, + "step": 8842 + }, + { + "epoch": 3.7, + "learning_rate": 1.1680613583832956e-05, + "loss": 0.1764, + "step": 8843 + }, + { + "epoch": 3.7, + "learning_rate": 1.1678942265394055e-05, + "loss": 0.1845, + "step": 8844 + }, + { + "epoch": 3.7, + "learning_rate": 1.1677270898695445e-05, + "loss": 0.1797, + "step": 8845 + }, + { + "epoch": 3.7, + "learning_rate": 1.1675599483785171e-05, + "loss": 0.1832, + "step": 8846 + }, + { + "epoch": 3.7, + "learning_rate": 1.167392802071128e-05, + "loss": 0.2014, + "step": 8847 + }, + { + "epoch": 3.7, + "learning_rate": 1.1672256509521814e-05, + "loss": 0.2086, + "step": 8848 + }, + { + "epoch": 3.7, + "learning_rate": 1.1670584950264816e-05, + "loss": 0.1592, + "step": 8849 + }, + { + "epoch": 3.71, + "learning_rate": 1.1668913342988335e-05, + "loss": 0.2036, + "step": 8850 + }, + { + "epoch": 3.71, + "learning_rate": 1.1667241687740422e-05, + "loss": 0.1597, + "step": 8851 + }, + { + "epoch": 3.71, + "learning_rate": 1.1665569984569127e-05, + "loss": 0.2258, + "step": 8852 + }, + { + "epoch": 3.71, + "learning_rate": 1.1663898233522499e-05, + "loss": 0.1592, + "step": 8853 + }, + { + "epoch": 3.71, + "learning_rate": 1.1662226434648595e-05, + "loss": 0.1749, + "step": 8854 + }, + { + "epoch": 3.71, + "learning_rate": 1.1660554587995465e-05, + "loss": 0.1716, + "step": 8855 + }, + { + "epoch": 3.71, + "learning_rate": 1.1658882693611169e-05, + "loss": 0.1531, + "step": 8856 + }, + { + "epoch": 3.71, + "learning_rate": 1.165721075154376e-05, + "loss": 0.1848, + "step": 8857 + }, + { + "epoch": 3.71, + "learning_rate": 1.16555387618413e-05, + "loss": 0.1777, + "step": 8858 + }, + { + "epoch": 3.71, + "learning_rate": 1.1653866724551848e-05, + "loss": 0.2096, + "step": 8859 + }, + { + "epoch": 3.71, + "learning_rate": 1.1652194639723465e-05, + "loss": 0.1777, + "step": 8860 + }, + { + "epoch": 3.71, + "learning_rate": 1.1650522507404211e-05, + "loss": 0.2106, + "step": 8861 + }, + { + "epoch": 3.71, + "learning_rate": 1.1648850327642154e-05, + "loss": 0.1917, + "step": 8862 + }, + { + "epoch": 3.71, + "learning_rate": 1.1647178100485354e-05, + "loss": 0.1864, + "step": 8863 + }, + { + "epoch": 3.71, + "learning_rate": 1.1645505825981884e-05, + "loss": 0.1869, + "step": 8864 + }, + { + "epoch": 3.71, + "learning_rate": 1.1643833504179811e-05, + "loss": 0.1971, + "step": 8865 + }, + { + "epoch": 3.71, + "learning_rate": 1.1642161135127198e-05, + "loss": 0.1678, + "step": 8866 + }, + { + "epoch": 3.71, + "learning_rate": 1.1640488718872124e-05, + "loss": 0.1875, + "step": 8867 + }, + { + "epoch": 3.71, + "learning_rate": 1.1638816255462657e-05, + "loss": 0.1905, + "step": 8868 + }, + { + "epoch": 3.71, + "learning_rate": 1.1637143744946871e-05, + "loss": 0.2219, + "step": 8869 + }, + { + "epoch": 3.71, + "learning_rate": 1.163547118737284e-05, + "loss": 0.1915, + "step": 8870 + }, + { + "epoch": 3.71, + "learning_rate": 1.1633798582788642e-05, + "loss": 0.1979, + "step": 8871 + }, + { + "epoch": 3.71, + "learning_rate": 1.1632125931242356e-05, + "loss": 0.2147, + "step": 8872 + }, + { + "epoch": 3.71, + "learning_rate": 1.1630453232782053e-05, + "loss": 0.2346, + "step": 8873 + }, + { + "epoch": 3.72, + "learning_rate": 1.162878048745582e-05, + "loss": 0.1713, + "step": 8874 + }, + { + "epoch": 3.72, + "learning_rate": 1.1627107695311738e-05, + "loss": 0.2076, + "step": 8875 + }, + { + "epoch": 3.72, + "learning_rate": 1.1625434856397891e-05, + "loss": 0.1693, + "step": 8876 + }, + { + "epoch": 3.72, + "learning_rate": 1.1623761970762359e-05, + "loss": 0.1629, + "step": 8877 + }, + { + "epoch": 3.72, + "learning_rate": 1.1622089038453232e-05, + "loss": 0.1809, + "step": 8878 + }, + { + "epoch": 3.72, + "learning_rate": 1.1620416059518593e-05, + "loss": 0.1831, + "step": 8879 + }, + { + "epoch": 3.72, + "learning_rate": 1.1618743034006532e-05, + "loss": 0.1921, + "step": 8880 + }, + { + "epoch": 3.72, + "learning_rate": 1.1617069961965141e-05, + "loss": 0.1877, + "step": 8881 + }, + { + "epoch": 3.72, + "learning_rate": 1.161539684344251e-05, + "loss": 0.2086, + "step": 8882 + }, + { + "epoch": 3.72, + "learning_rate": 1.1613723678486728e-05, + "loss": 0.1888, + "step": 8883 + }, + { + "epoch": 3.72, + "learning_rate": 1.161205046714589e-05, + "loss": 0.2187, + "step": 8884 + }, + { + "epoch": 3.72, + "learning_rate": 1.1610377209468095e-05, + "loss": 0.1716, + "step": 8885 + }, + { + "epoch": 3.72, + "learning_rate": 1.1608703905501432e-05, + "loss": 0.2116, + "step": 8886 + }, + { + "epoch": 3.72, + "learning_rate": 1.1607030555294005e-05, + "loss": 0.2013, + "step": 8887 + }, + { + "epoch": 3.72, + "learning_rate": 1.1605357158893913e-05, + "loss": 0.1809, + "step": 8888 + }, + { + "epoch": 3.72, + "learning_rate": 1.160368371634925e-05, + "loss": 0.196, + "step": 8889 + }, + { + "epoch": 3.72, + "learning_rate": 1.1602010227708122e-05, + "loss": 0.2002, + "step": 8890 + }, + { + "epoch": 3.72, + "learning_rate": 1.1600336693018635e-05, + "loss": 0.1671, + "step": 8891 + }, + { + "epoch": 3.72, + "learning_rate": 1.1598663112328888e-05, + "loss": 0.1837, + "step": 8892 + }, + { + "epoch": 3.72, + "learning_rate": 1.1596989485686987e-05, + "loss": 0.202, + "step": 8893 + }, + { + "epoch": 3.72, + "learning_rate": 1.1595315813141041e-05, + "loss": 0.1697, + "step": 8894 + }, + { + "epoch": 3.72, + "learning_rate": 1.1593642094739161e-05, + "loss": 0.1923, + "step": 8895 + }, + { + "epoch": 3.72, + "learning_rate": 1.159196833052945e-05, + "loss": 0.166, + "step": 8896 + }, + { + "epoch": 3.72, + "learning_rate": 1.1590294520560022e-05, + "loss": 0.2166, + "step": 8897 + }, + { + "epoch": 3.73, + "learning_rate": 1.1588620664878989e-05, + "loss": 0.1934, + "step": 8898 + }, + { + "epoch": 3.73, + "learning_rate": 1.1586946763534465e-05, + "loss": 0.1996, + "step": 8899 + }, + { + "epoch": 3.73, + "learning_rate": 1.1585272816574565e-05, + "loss": 0.1903, + "step": 8900 + }, + { + "epoch": 3.73, + "learning_rate": 1.1583598824047404e-05, + "loss": 0.1923, + "step": 8901 + }, + { + "epoch": 3.73, + "learning_rate": 1.1581924786001102e-05, + "loss": 0.1939, + "step": 8902 + }, + { + "epoch": 3.73, + "learning_rate": 1.1580250702483773e-05, + "loss": 0.1808, + "step": 8903 + }, + { + "epoch": 3.73, + "learning_rate": 1.1578576573543541e-05, + "loss": 0.1719, + "step": 8904 + }, + { + "epoch": 3.73, + "learning_rate": 1.1576902399228527e-05, + "loss": 0.1946, + "step": 8905 + }, + { + "epoch": 3.73, + "learning_rate": 1.1575228179586855e-05, + "loss": 0.1857, + "step": 8906 + }, + { + "epoch": 3.73, + "learning_rate": 1.1573553914666643e-05, + "loss": 0.1814, + "step": 8907 + }, + { + "epoch": 3.73, + "learning_rate": 1.157187960451602e-05, + "loss": 0.1515, + "step": 8908 + }, + { + "epoch": 3.73, + "learning_rate": 1.1570205249183115e-05, + "loss": 0.1651, + "step": 8909 + }, + { + "epoch": 3.73, + "learning_rate": 1.156853084871605e-05, + "loss": 0.1432, + "step": 8910 + }, + { + "epoch": 3.73, + "learning_rate": 1.156685640316296e-05, + "loss": 0.1824, + "step": 8911 + }, + { + "epoch": 3.73, + "learning_rate": 1.1565181912571975e-05, + "loss": 0.1531, + "step": 8912 + }, + { + "epoch": 3.73, + "learning_rate": 1.1563507376991222e-05, + "loss": 0.2062, + "step": 8913 + }, + { + "epoch": 3.73, + "learning_rate": 1.1561832796468837e-05, + "loss": 0.209, + "step": 8914 + }, + { + "epoch": 3.73, + "learning_rate": 1.1560158171052956e-05, + "loss": 0.1206, + "step": 8915 + }, + { + "epoch": 3.73, + "learning_rate": 1.1558483500791713e-05, + "loss": 0.1754, + "step": 8916 + }, + { + "epoch": 3.73, + "learning_rate": 1.1556808785733242e-05, + "loss": 0.2223, + "step": 8917 + }, + { + "epoch": 3.73, + "learning_rate": 1.1555134025925685e-05, + "loss": 0.1631, + "step": 8918 + }, + { + "epoch": 3.73, + "learning_rate": 1.1553459221417182e-05, + "loss": 0.1875, + "step": 8919 + }, + { + "epoch": 3.73, + "learning_rate": 1.155178437225587e-05, + "loss": 0.2032, + "step": 8920 + }, + { + "epoch": 3.73, + "learning_rate": 1.1550109478489896e-05, + "loss": 0.1785, + "step": 8921 + }, + { + "epoch": 3.74, + "learning_rate": 1.1548434540167399e-05, + "loss": 0.1759, + "step": 8922 + }, + { + "epoch": 3.74, + "learning_rate": 1.1546759557336524e-05, + "loss": 0.2091, + "step": 8923 + }, + { + "epoch": 3.74, + "learning_rate": 1.1545084530045419e-05, + "loss": 0.211, + "step": 8924 + }, + { + "epoch": 3.74, + "learning_rate": 1.154340945834223e-05, + "loss": 0.2067, + "step": 8925 + }, + { + "epoch": 3.74, + "learning_rate": 1.1541734342275106e-05, + "loss": 0.2118, + "step": 8926 + }, + { + "epoch": 3.74, + "learning_rate": 1.1540059181892197e-05, + "loss": 0.1759, + "step": 8927 + }, + { + "epoch": 3.74, + "learning_rate": 1.1538383977241654e-05, + "loss": 0.179, + "step": 8928 + }, + { + "epoch": 3.74, + "learning_rate": 1.1536708728371628e-05, + "loss": 0.1841, + "step": 8929 + }, + { + "epoch": 3.74, + "learning_rate": 1.1535033435330272e-05, + "loss": 0.1893, + "step": 8930 + }, + { + "epoch": 3.74, + "learning_rate": 1.1533358098165744e-05, + "loss": 0.2098, + "step": 8931 + }, + { + "epoch": 3.74, + "learning_rate": 1.1531682716926199e-05, + "loss": 0.1792, + "step": 8932 + }, + { + "epoch": 3.74, + "learning_rate": 1.153000729165979e-05, + "loss": 0.1708, + "step": 8933 + }, + { + "epoch": 3.74, + "learning_rate": 1.1528331822414681e-05, + "loss": 0.1958, + "step": 8934 + }, + { + "epoch": 3.74, + "learning_rate": 1.1526656309239028e-05, + "loss": 0.1662, + "step": 8935 + }, + { + "epoch": 3.74, + "learning_rate": 1.1524980752180996e-05, + "loss": 0.208, + "step": 8936 + }, + { + "epoch": 3.74, + "learning_rate": 1.1523305151288747e-05, + "loss": 0.173, + "step": 8937 + }, + { + "epoch": 3.74, + "learning_rate": 1.1521629506610439e-05, + "loss": 0.216, + "step": 8938 + }, + { + "epoch": 3.74, + "learning_rate": 1.1519953818194246e-05, + "loss": 0.1837, + "step": 8939 + }, + { + "epoch": 3.74, + "learning_rate": 1.1518278086088328e-05, + "loss": 0.1763, + "step": 8940 + }, + { + "epoch": 3.74, + "learning_rate": 1.1516602310340852e-05, + "loss": 0.1977, + "step": 8941 + }, + { + "epoch": 3.74, + "learning_rate": 1.151492649099999e-05, + "loss": 0.2517, + "step": 8942 + }, + { + "epoch": 3.74, + "learning_rate": 1.151325062811391e-05, + "loss": 0.234, + "step": 8943 + }, + { + "epoch": 3.74, + "learning_rate": 1.1511574721730781e-05, + "loss": 0.1756, + "step": 8944 + }, + { + "epoch": 3.75, + "learning_rate": 1.150989877189878e-05, + "loss": 0.1934, + "step": 8945 + }, + { + "epoch": 3.75, + "learning_rate": 1.1508222778666078e-05, + "loss": 0.1277, + "step": 8946 + }, + { + "epoch": 3.75, + "learning_rate": 1.1506546742080854e-05, + "loss": 0.1748, + "step": 8947 + }, + { + "epoch": 3.75, + "learning_rate": 1.1504870662191275e-05, + "loss": 0.169, + "step": 8948 + }, + { + "epoch": 3.75, + "learning_rate": 1.1503194539045527e-05, + "loss": 0.1794, + "step": 8949 + }, + { + "epoch": 3.75, + "learning_rate": 1.1501518372691789e-05, + "loss": 0.134, + "step": 8950 + }, + { + "epoch": 3.75, + "learning_rate": 1.1499842163178234e-05, + "loss": 0.1981, + "step": 8951 + }, + { + "epoch": 3.75, + "learning_rate": 1.1498165910553048e-05, + "loss": 0.1996, + "step": 8952 + }, + { + "epoch": 3.75, + "learning_rate": 1.1496489614864412e-05, + "loss": 0.1395, + "step": 8953 + }, + { + "epoch": 3.75, + "learning_rate": 1.1494813276160511e-05, + "loss": 0.1912, + "step": 8954 + }, + { + "epoch": 3.75, + "learning_rate": 1.1493136894489527e-05, + "loss": 0.1953, + "step": 8955 + }, + { + "epoch": 3.75, + "learning_rate": 1.149146046989965e-05, + "loss": 0.1645, + "step": 8956 + }, + { + "epoch": 3.75, + "learning_rate": 1.1489784002439064e-05, + "loss": 0.2151, + "step": 8957 + }, + { + "epoch": 3.75, + "learning_rate": 1.1488107492155958e-05, + "loss": 0.171, + "step": 8958 + }, + { + "epoch": 3.75, + "learning_rate": 1.1486430939098525e-05, + "loss": 0.1875, + "step": 8959 + }, + { + "epoch": 3.75, + "learning_rate": 1.1484754343314952e-05, + "loss": 0.2094, + "step": 8960 + }, + { + "epoch": 3.75, + "learning_rate": 1.1483077704853431e-05, + "loss": 0.1987, + "step": 8961 + }, + { + "epoch": 3.75, + "learning_rate": 1.1481401023762162e-05, + "loss": 0.2137, + "step": 8962 + }, + { + "epoch": 3.75, + "learning_rate": 1.1479724300089335e-05, + "loss": 0.1704, + "step": 8963 + }, + { + "epoch": 3.75, + "learning_rate": 1.1478047533883143e-05, + "loss": 0.1562, + "step": 8964 + }, + { + "epoch": 3.75, + "learning_rate": 1.1476370725191788e-05, + "loss": 0.1986, + "step": 8965 + }, + { + "epoch": 3.75, + "learning_rate": 1.1474693874063467e-05, + "loss": 0.1907, + "step": 8966 + }, + { + "epoch": 3.75, + "learning_rate": 1.1473016980546377e-05, + "loss": 0.1945, + "step": 8967 + }, + { + "epoch": 3.75, + "learning_rate": 1.147134004468872e-05, + "loss": 0.2249, + "step": 8968 + }, + { + "epoch": 3.76, + "learning_rate": 1.1469663066538704e-05, + "loss": 0.1739, + "step": 8969 + }, + { + "epoch": 3.76, + "learning_rate": 1.1467986046144523e-05, + "loss": 0.1978, + "step": 8970 + }, + { + "epoch": 3.76, + "learning_rate": 1.1466308983554386e-05, + "loss": 0.1379, + "step": 8971 + }, + { + "epoch": 3.76, + "learning_rate": 1.1464631878816502e-05, + "loss": 0.1583, + "step": 8972 + }, + { + "epoch": 3.76, + "learning_rate": 1.1462954731979074e-05, + "loss": 0.1796, + "step": 8973 + }, + { + "epoch": 3.76, + "learning_rate": 1.1461277543090308e-05, + "loss": 0.1897, + "step": 8974 + }, + { + "epoch": 3.76, + "learning_rate": 1.1459600312198416e-05, + "loss": 0.2038, + "step": 8975 + }, + { + "epoch": 3.76, + "learning_rate": 1.1457923039351606e-05, + "loss": 0.1697, + "step": 8976 + }, + { + "epoch": 3.76, + "learning_rate": 1.1456245724598096e-05, + "loss": 0.1876, + "step": 8977 + }, + { + "epoch": 3.76, + "learning_rate": 1.1454568367986094e-05, + "loss": 0.19, + "step": 8978 + }, + { + "epoch": 3.76, + "learning_rate": 1.1452890969563813e-05, + "loss": 0.2207, + "step": 8979 + }, + { + "epoch": 3.76, + "learning_rate": 1.1451213529379472e-05, + "loss": 0.1888, + "step": 8980 + }, + { + "epoch": 3.76, + "learning_rate": 1.1449536047481287e-05, + "loss": 0.1905, + "step": 8981 + }, + { + "epoch": 3.76, + "learning_rate": 1.1447858523917472e-05, + "loss": 0.2155, + "step": 8982 + }, + { + "epoch": 3.76, + "learning_rate": 1.144618095873625e-05, + "loss": 0.2026, + "step": 8983 + }, + { + "epoch": 3.76, + "learning_rate": 1.144450335198584e-05, + "loss": 0.1414, + "step": 8984 + }, + { + "epoch": 3.76, + "learning_rate": 1.1442825703714464e-05, + "loss": 0.1805, + "step": 8985 + }, + { + "epoch": 3.76, + "learning_rate": 1.144114801397034e-05, + "loss": 0.1571, + "step": 8986 + }, + { + "epoch": 3.76, + "learning_rate": 1.14394702828017e-05, + "loss": 0.2133, + "step": 8987 + }, + { + "epoch": 3.76, + "learning_rate": 1.143779251025676e-05, + "loss": 0.1764, + "step": 8988 + }, + { + "epoch": 3.76, + "learning_rate": 1.1436114696383749e-05, + "loss": 0.2045, + "step": 8989 + }, + { + "epoch": 3.76, + "learning_rate": 1.14344368412309e-05, + "loss": 0.1676, + "step": 8990 + }, + { + "epoch": 3.76, + "learning_rate": 1.1432758944846433e-05, + "loss": 0.1914, + "step": 8991 + }, + { + "epoch": 3.76, + "learning_rate": 1.1431081007278584e-05, + "loss": 0.1964, + "step": 8992 + }, + { + "epoch": 3.77, + "learning_rate": 1.142940302857558e-05, + "loss": 0.1722, + "step": 8993 + }, + { + "epoch": 3.77, + "learning_rate": 1.1427725008785655e-05, + "loss": 0.173, + "step": 8994 + }, + { + "epoch": 3.77, + "learning_rate": 1.1426046947957041e-05, + "loss": 0.2282, + "step": 8995 + }, + { + "epoch": 3.77, + "learning_rate": 1.1424368846137975e-05, + "loss": 0.1957, + "step": 8996 + }, + { + "epoch": 3.77, + "learning_rate": 1.142269070337669e-05, + "loss": 0.1834, + "step": 8997 + }, + { + "epoch": 3.77, + "learning_rate": 1.1421012519721421e-05, + "loss": 0.1803, + "step": 8998 + }, + { + "epoch": 3.77, + "learning_rate": 1.141933429522041e-05, + "loss": 0.2323, + "step": 8999 + }, + { + "epoch": 3.77, + "learning_rate": 1.1417656029921895e-05, + "loss": 0.2027, + "step": 9000 + }, + { + "epoch": 3.77, + "learning_rate": 1.1415977723874112e-05, + "loss": 0.1768, + "step": 9001 + }, + { + "epoch": 3.77, + "learning_rate": 1.1414299377125308e-05, + "loss": 0.1486, + "step": 9002 + }, + { + "epoch": 3.77, + "learning_rate": 1.1412620989723724e-05, + "loss": 0.1703, + "step": 9003 + }, + { + "epoch": 3.77, + "learning_rate": 1.1410942561717601e-05, + "loss": 0.1961, + "step": 9004 + }, + { + "epoch": 3.77, + "learning_rate": 1.1409264093155189e-05, + "loss": 0.2018, + "step": 9005 + }, + { + "epoch": 3.77, + "learning_rate": 1.1407585584084726e-05, + "loss": 0.1878, + "step": 9006 + }, + { + "epoch": 3.77, + "learning_rate": 1.1405907034554473e-05, + "loss": 0.2096, + "step": 9007 + }, + { + "epoch": 3.77, + "learning_rate": 1.1404228444612664e-05, + "loss": 0.1901, + "step": 9008 + }, + { + "epoch": 3.77, + "learning_rate": 1.1402549814307555e-05, + "loss": 0.1547, + "step": 9009 + }, + { + "epoch": 3.77, + "learning_rate": 1.1400871143687398e-05, + "loss": 0.193, + "step": 9010 + }, + { + "epoch": 3.77, + "learning_rate": 1.1399192432800442e-05, + "loss": 0.1671, + "step": 9011 + }, + { + "epoch": 3.77, + "learning_rate": 1.1397513681694938e-05, + "loss": 0.1926, + "step": 9012 + }, + { + "epoch": 3.77, + "learning_rate": 1.1395834890419147e-05, + "loss": 0.1855, + "step": 9013 + }, + { + "epoch": 3.77, + "learning_rate": 1.139415605902132e-05, + "loss": 0.1667, + "step": 9014 + }, + { + "epoch": 3.77, + "learning_rate": 1.1392477187549714e-05, + "loss": 0.2235, + "step": 9015 + }, + { + "epoch": 3.77, + "learning_rate": 1.1390798276052587e-05, + "loss": 0.206, + "step": 9016 + }, + { + "epoch": 3.78, + "learning_rate": 1.1389119324578197e-05, + "loss": 0.1925, + "step": 9017 + }, + { + "epoch": 3.78, + "learning_rate": 1.1387440333174809e-05, + "loss": 0.1357, + "step": 9018 + }, + { + "epoch": 3.78, + "learning_rate": 1.1385761301890674e-05, + "loss": 0.1902, + "step": 9019 + }, + { + "epoch": 3.78, + "learning_rate": 1.1384082230774063e-05, + "loss": 0.2036, + "step": 9020 + }, + { + "epoch": 3.78, + "learning_rate": 1.1382403119873236e-05, + "loss": 0.1801, + "step": 9021 + }, + { + "epoch": 3.78, + "learning_rate": 1.1380723969236458e-05, + "loss": 0.204, + "step": 9022 + }, + { + "epoch": 3.78, + "learning_rate": 1.1379044778911994e-05, + "loss": 0.2092, + "step": 9023 + }, + { + "epoch": 3.78, + "learning_rate": 1.1377365548948114e-05, + "loss": 0.164, + "step": 9024 + }, + { + "epoch": 3.78, + "learning_rate": 1.137568627939308e-05, + "loss": 0.1681, + "step": 9025 + }, + { + "epoch": 3.78, + "learning_rate": 1.1374006970295167e-05, + "loss": 0.1726, + "step": 9026 + }, + { + "epoch": 3.78, + "learning_rate": 1.137232762170264e-05, + "loss": 0.1698, + "step": 9027 + }, + { + "epoch": 3.78, + "learning_rate": 1.1370648233663776e-05, + "loss": 0.1895, + "step": 9028 + }, + { + "epoch": 3.78, + "learning_rate": 1.1368968806226844e-05, + "loss": 0.195, + "step": 9029 + }, + { + "epoch": 3.78, + "learning_rate": 1.1367289339440123e-05, + "loss": 0.2167, + "step": 9030 + }, + { + "epoch": 3.78, + "learning_rate": 1.1365609833351877e-05, + "loss": 0.232, + "step": 9031 + }, + { + "epoch": 3.78, + "learning_rate": 1.136393028801039e-05, + "loss": 0.1908, + "step": 9032 + }, + { + "epoch": 3.78, + "learning_rate": 1.1362250703463937e-05, + "loss": 0.1637, + "step": 9033 + }, + { + "epoch": 3.78, + "learning_rate": 1.1360571079760798e-05, + "loss": 0.1488, + "step": 9034 + }, + { + "epoch": 3.78, + "learning_rate": 1.1358891416949249e-05, + "loss": 0.2147, + "step": 9035 + }, + { + "epoch": 3.78, + "learning_rate": 1.1357211715077571e-05, + "loss": 0.2032, + "step": 9036 + }, + { + "epoch": 3.78, + "learning_rate": 1.1355531974194051e-05, + "loss": 0.1986, + "step": 9037 + }, + { + "epoch": 3.78, + "learning_rate": 1.1353852194346963e-05, + "loss": 0.1634, + "step": 9038 + }, + { + "epoch": 3.78, + "learning_rate": 1.1352172375584596e-05, + "loss": 0.2022, + "step": 9039 + }, + { + "epoch": 3.78, + "learning_rate": 1.1350492517955234e-05, + "loss": 0.1578, + "step": 9040 + }, + { + "epoch": 3.79, + "learning_rate": 1.1348812621507167e-05, + "loss": 0.1488, + "step": 9041 + }, + { + "epoch": 3.79, + "learning_rate": 1.1347132686288673e-05, + "loss": 0.1921, + "step": 9042 + }, + { + "epoch": 3.79, + "learning_rate": 1.1345452712348048e-05, + "loss": 0.1615, + "step": 9043 + }, + { + "epoch": 3.79, + "learning_rate": 1.1343772699733579e-05, + "loss": 0.2077, + "step": 9044 + }, + { + "epoch": 3.79, + "learning_rate": 1.1342092648493555e-05, + "loss": 0.1969, + "step": 9045 + }, + { + "epoch": 3.79, + "learning_rate": 1.1340412558676266e-05, + "loss": 0.2035, + "step": 9046 + }, + { + "epoch": 3.79, + "learning_rate": 1.1338732430330011e-05, + "loss": 0.1633, + "step": 9047 + }, + { + "epoch": 3.79, + "learning_rate": 1.1337052263503079e-05, + "loss": 0.1723, + "step": 9048 + }, + { + "epoch": 3.79, + "learning_rate": 1.1335372058243768e-05, + "loss": 0.2138, + "step": 9049 + }, + { + "epoch": 3.79, + "learning_rate": 1.1333691814600369e-05, + "loss": 0.1746, + "step": 9050 + }, + { + "epoch": 3.79, + "learning_rate": 1.1332011532621186e-05, + "loss": 0.2055, + "step": 9051 + }, + { + "epoch": 3.79, + "learning_rate": 1.1330331212354513e-05, + "loss": 0.1866, + "step": 9052 + }, + { + "epoch": 3.79, + "learning_rate": 1.1328650853848646e-05, + "loss": 0.1995, + "step": 9053 + }, + { + "epoch": 3.79, + "learning_rate": 1.132697045715189e-05, + "loss": 0.1582, + "step": 9054 + }, + { + "epoch": 3.79, + "learning_rate": 1.132529002231255e-05, + "loss": 0.2029, + "step": 9055 + }, + { + "epoch": 3.79, + "learning_rate": 1.132360954937892e-05, + "loss": 0.1756, + "step": 9056 + }, + { + "epoch": 3.79, + "learning_rate": 1.1321929038399309e-05, + "loss": 0.2151, + "step": 9057 + }, + { + "epoch": 3.79, + "learning_rate": 1.1320248489422024e-05, + "loss": 0.211, + "step": 9058 + }, + { + "epoch": 3.79, + "learning_rate": 1.1318567902495364e-05, + "loss": 0.2204, + "step": 9059 + }, + { + "epoch": 3.79, + "learning_rate": 1.131688727766764e-05, + "loss": 0.1669, + "step": 9060 + }, + { + "epoch": 3.79, + "learning_rate": 1.1315206614987161e-05, + "loss": 0.1861, + "step": 9061 + }, + { + "epoch": 3.79, + "learning_rate": 1.1313525914502235e-05, + "loss": 0.1728, + "step": 9062 + }, + { + "epoch": 3.79, + "learning_rate": 1.1311845176261172e-05, + "loss": 0.1886, + "step": 9063 + }, + { + "epoch": 3.79, + "learning_rate": 1.1310164400312284e-05, + "loss": 0.1917, + "step": 9064 + }, + { + "epoch": 3.8, + "learning_rate": 1.1308483586703885e-05, + "loss": 0.1719, + "step": 9065 + }, + { + "epoch": 3.8, + "learning_rate": 1.1306802735484282e-05, + "loss": 0.1885, + "step": 9066 + }, + { + "epoch": 3.8, + "learning_rate": 1.1305121846701797e-05, + "loss": 0.2081, + "step": 9067 + }, + { + "epoch": 3.8, + "learning_rate": 1.130344092040474e-05, + "loss": 0.2003, + "step": 9068 + }, + { + "epoch": 3.8, + "learning_rate": 1.1301759956641432e-05, + "loss": 0.199, + "step": 9069 + }, + { + "epoch": 3.8, + "learning_rate": 1.1300078955460187e-05, + "loss": 0.2198, + "step": 9070 + }, + { + "epoch": 3.8, + "learning_rate": 1.1298397916909332e-05, + "loss": 0.187, + "step": 9071 + }, + { + "epoch": 3.8, + "learning_rate": 1.1296716841037176e-05, + "loss": 0.2268, + "step": 9072 + }, + { + "epoch": 3.8, + "learning_rate": 1.1295035727892046e-05, + "loss": 0.2108, + "step": 9073 + }, + { + "epoch": 3.8, + "learning_rate": 1.1293354577522264e-05, + "loss": 0.2043, + "step": 9074 + }, + { + "epoch": 3.8, + "learning_rate": 1.1291673389976153e-05, + "loss": 0.1764, + "step": 9075 + }, + { + "epoch": 3.8, + "learning_rate": 1.1289992165302036e-05, + "loss": 0.2057, + "step": 9076 + }, + { + "epoch": 3.8, + "learning_rate": 1.1288310903548236e-05, + "loss": 0.1947, + "step": 9077 + }, + { + "epoch": 3.8, + "learning_rate": 1.1286629604763087e-05, + "loss": 0.2054, + "step": 9078 + }, + { + "epoch": 3.8, + "learning_rate": 1.1284948268994908e-05, + "loss": 0.2173, + "step": 9079 + }, + { + "epoch": 3.8, + "learning_rate": 1.1283266896292033e-05, + "loss": 0.1699, + "step": 9080 + }, + { + "epoch": 3.8, + "learning_rate": 1.1281585486702791e-05, + "loss": 0.1738, + "step": 9081 + }, + { + "epoch": 3.8, + "learning_rate": 1.1279904040275508e-05, + "loss": 0.1923, + "step": 9082 + }, + { + "epoch": 3.8, + "learning_rate": 1.1278222557058521e-05, + "loss": 0.191, + "step": 9083 + }, + { + "epoch": 3.8, + "learning_rate": 1.127654103710016e-05, + "loss": 0.2003, + "step": 9084 + }, + { + "epoch": 3.8, + "learning_rate": 1.127485948044876e-05, + "loss": 0.1608, + "step": 9085 + }, + { + "epoch": 3.8, + "learning_rate": 1.127317788715266e-05, + "loss": 0.1259, + "step": 9086 + }, + { + "epoch": 3.8, + "learning_rate": 1.1271496257260186e-05, + "loss": 0.1972, + "step": 9087 + }, + { + "epoch": 3.8, + "learning_rate": 1.126981459081968e-05, + "loss": 0.1557, + "step": 9088 + }, + { + "epoch": 3.81, + "learning_rate": 1.1268132887879485e-05, + "loss": 0.2127, + "step": 9089 + }, + { + "epoch": 3.81, + "learning_rate": 1.1266451148487934e-05, + "loss": 0.1941, + "step": 9090 + }, + { + "epoch": 3.81, + "learning_rate": 1.1264769372693365e-05, + "loss": 0.1913, + "step": 9091 + }, + { + "epoch": 3.81, + "learning_rate": 1.1263087560544128e-05, + "loss": 0.1689, + "step": 9092 + }, + { + "epoch": 3.81, + "learning_rate": 1.1261405712088556e-05, + "loss": 0.1506, + "step": 9093 + }, + { + "epoch": 3.81, + "learning_rate": 1.1259723827374996e-05, + "loss": 0.2145, + "step": 9094 + }, + { + "epoch": 3.81, + "learning_rate": 1.1258041906451795e-05, + "loss": 0.1616, + "step": 9095 + }, + { + "epoch": 3.81, + "learning_rate": 1.1256359949367293e-05, + "loss": 0.2278, + "step": 9096 + }, + { + "epoch": 3.81, + "learning_rate": 1.1254677956169839e-05, + "loss": 0.1865, + "step": 9097 + }, + { + "epoch": 3.81, + "learning_rate": 1.1252995926907783e-05, + "loss": 0.1616, + "step": 9098 + }, + { + "epoch": 3.81, + "learning_rate": 1.1251313861629474e-05, + "loss": 0.1893, + "step": 9099 + }, + { + "epoch": 3.81, + "learning_rate": 1.1249631760383253e-05, + "loss": 0.2181, + "step": 9100 + }, + { + "epoch": 3.81, + "learning_rate": 1.1247949623217476e-05, + "loss": 0.1697, + "step": 9101 + }, + { + "epoch": 3.81, + "learning_rate": 1.1246267450180497e-05, + "loss": 0.1728, + "step": 9102 + }, + { + "epoch": 3.81, + "learning_rate": 1.1244585241320664e-05, + "loss": 0.2064, + "step": 9103 + }, + { + "epoch": 3.81, + "learning_rate": 1.1242902996686333e-05, + "loss": 0.186, + "step": 9104 + }, + { + "epoch": 3.81, + "learning_rate": 1.1241220716325863e-05, + "loss": 0.1907, + "step": 9105 + }, + { + "epoch": 3.81, + "learning_rate": 1.12395384002876e-05, + "loss": 0.1635, + "step": 9106 + }, + { + "epoch": 3.81, + "learning_rate": 1.1237856048619904e-05, + "loss": 0.1679, + "step": 9107 + }, + { + "epoch": 3.81, + "learning_rate": 1.1236173661371139e-05, + "loss": 0.1374, + "step": 9108 + }, + { + "epoch": 3.81, + "learning_rate": 1.1234491238589658e-05, + "loss": 0.1645, + "step": 9109 + }, + { + "epoch": 3.81, + "learning_rate": 1.1232808780323821e-05, + "loss": 0.2045, + "step": 9110 + }, + { + "epoch": 3.81, + "learning_rate": 1.1231126286621988e-05, + "loss": 0.1994, + "step": 9111 + }, + { + "epoch": 3.81, + "learning_rate": 1.1229443757532526e-05, + "loss": 0.1896, + "step": 9112 + }, + { + "epoch": 3.82, + "learning_rate": 1.122776119310379e-05, + "loss": 0.1959, + "step": 9113 + }, + { + "epoch": 3.82, + "learning_rate": 1.122607859338415e-05, + "loss": 0.2131, + "step": 9114 + }, + { + "epoch": 3.82, + "learning_rate": 1.122439595842197e-05, + "loss": 0.1619, + "step": 9115 + }, + { + "epoch": 3.82, + "learning_rate": 1.1222713288265613e-05, + "loss": 0.1915, + "step": 9116 + }, + { + "epoch": 3.82, + "learning_rate": 1.1221030582963447e-05, + "loss": 0.1749, + "step": 9117 + }, + { + "epoch": 3.82, + "learning_rate": 1.1219347842563838e-05, + "loss": 0.1934, + "step": 9118 + }, + { + "epoch": 3.82, + "learning_rate": 1.1217665067115164e-05, + "loss": 0.1642, + "step": 9119 + }, + { + "epoch": 3.82, + "learning_rate": 1.1215982256665786e-05, + "loss": 0.2118, + "step": 9120 + }, + { + "epoch": 3.82, + "learning_rate": 1.1214299411264075e-05, + "loss": 0.1747, + "step": 9121 + }, + { + "epoch": 3.82, + "learning_rate": 1.1212616530958407e-05, + "loss": 0.1927, + "step": 9122 + }, + { + "epoch": 3.82, + "learning_rate": 1.121093361579715e-05, + "loss": 0.1762, + "step": 9123 + }, + { + "epoch": 3.82, + "learning_rate": 1.1209250665828684e-05, + "loss": 0.1817, + "step": 9124 + }, + { + "epoch": 3.82, + "learning_rate": 1.1207567681101377e-05, + "loss": 0.2285, + "step": 9125 + }, + { + "epoch": 3.82, + "learning_rate": 1.1205884661663614e-05, + "loss": 0.1896, + "step": 9126 + }, + { + "epoch": 3.82, + "learning_rate": 1.1204201607563761e-05, + "loss": 0.2009, + "step": 9127 + }, + { + "epoch": 3.82, + "learning_rate": 1.1202518518850202e-05, + "loss": 0.1918, + "step": 9128 + }, + { + "epoch": 3.82, + "learning_rate": 1.1200835395571319e-05, + "loss": 0.1798, + "step": 9129 + }, + { + "epoch": 3.82, + "learning_rate": 1.1199152237775485e-05, + "loss": 0.1773, + "step": 9130 + }, + { + "epoch": 3.82, + "learning_rate": 1.1197469045511085e-05, + "loss": 0.1865, + "step": 9131 + }, + { + "epoch": 3.82, + "learning_rate": 1.11957858188265e-05, + "loss": 0.1732, + "step": 9132 + }, + { + "epoch": 3.82, + "learning_rate": 1.1194102557770117e-05, + "loss": 0.2204, + "step": 9133 + }, + { + "epoch": 3.82, + "learning_rate": 1.119241926239031e-05, + "loss": 0.1778, + "step": 9134 + }, + { + "epoch": 3.82, + "learning_rate": 1.119073593273547e-05, + "loss": 0.1671, + "step": 9135 + }, + { + "epoch": 3.82, + "learning_rate": 1.1189052568853985e-05, + "loss": 0.2029, + "step": 9136 + }, + { + "epoch": 3.83, + "learning_rate": 1.1187369170794235e-05, + "loss": 0.2035, + "step": 9137 + }, + { + "epoch": 3.83, + "learning_rate": 1.1185685738604615e-05, + "loss": 0.1841, + "step": 9138 + }, + { + "epoch": 3.83, + "learning_rate": 1.1184002272333509e-05, + "loss": 0.1578, + "step": 9139 + }, + { + "epoch": 3.83, + "learning_rate": 1.1182318772029309e-05, + "loss": 0.1937, + "step": 9140 + }, + { + "epoch": 3.83, + "learning_rate": 1.1180635237740405e-05, + "loss": 0.1827, + "step": 9141 + }, + { + "epoch": 3.83, + "learning_rate": 1.117895166951519e-05, + "loss": 0.2154, + "step": 9142 + }, + { + "epoch": 3.83, + "learning_rate": 1.1177268067402057e-05, + "loss": 0.1988, + "step": 9143 + }, + { + "epoch": 3.83, + "learning_rate": 1.1175584431449393e-05, + "loss": 0.1743, + "step": 9144 + }, + { + "epoch": 3.83, + "learning_rate": 1.11739007617056e-05, + "loss": 0.2047, + "step": 9145 + }, + { + "epoch": 3.83, + "learning_rate": 1.1172217058219073e-05, + "loss": 0.1919, + "step": 9146 + }, + { + "epoch": 3.83, + "learning_rate": 1.1170533321038204e-05, + "loss": 0.186, + "step": 9147 + }, + { + "epoch": 3.83, + "learning_rate": 1.1168849550211394e-05, + "loss": 0.1803, + "step": 9148 + }, + { + "epoch": 3.83, + "learning_rate": 1.1167165745787041e-05, + "loss": 0.2222, + "step": 9149 + }, + { + "epoch": 3.83, + "learning_rate": 1.1165481907813544e-05, + "loss": 0.1488, + "step": 9150 + }, + { + "epoch": 3.83, + "learning_rate": 1.1163798036339303e-05, + "loss": 0.1842, + "step": 9151 + }, + { + "epoch": 3.83, + "learning_rate": 1.1162114131412723e-05, + "loss": 0.1747, + "step": 9152 + }, + { + "epoch": 3.83, + "learning_rate": 1.1160430193082203e-05, + "loss": 0.1926, + "step": 9153 + }, + { + "epoch": 3.83, + "learning_rate": 1.1158746221396148e-05, + "loss": 0.1933, + "step": 9154 + }, + { + "epoch": 3.83, + "learning_rate": 1.1157062216402959e-05, + "loss": 0.1534, + "step": 9155 + }, + { + "epoch": 3.83, + "learning_rate": 1.1155378178151044e-05, + "loss": 0.1914, + "step": 9156 + }, + { + "epoch": 3.83, + "learning_rate": 1.1153694106688809e-05, + "loss": 0.1767, + "step": 9157 + }, + { + "epoch": 3.83, + "learning_rate": 1.115201000206466e-05, + "loss": 0.1747, + "step": 9158 + }, + { + "epoch": 3.83, + "learning_rate": 1.1150325864327003e-05, + "loss": 0.193, + "step": 9159 + }, + { + "epoch": 3.84, + "learning_rate": 1.1148641693524254e-05, + "loss": 0.1653, + "step": 9160 + }, + { + "epoch": 3.84, + "learning_rate": 1.1146957489704817e-05, + "loss": 0.1859, + "step": 9161 + }, + { + "epoch": 3.84, + "learning_rate": 1.1145273252917104e-05, + "loss": 0.1975, + "step": 9162 + }, + { + "epoch": 3.84, + "learning_rate": 1.114358898320953e-05, + "loss": 0.1936, + "step": 9163 + }, + { + "epoch": 3.84, + "learning_rate": 1.1141904680630504e-05, + "loss": 0.2159, + "step": 9164 + }, + { + "epoch": 3.84, + "learning_rate": 1.114022034522844e-05, + "loss": 0.1748, + "step": 9165 + }, + { + "epoch": 3.84, + "learning_rate": 1.1138535977051757e-05, + "loss": 0.2104, + "step": 9166 + }, + { + "epoch": 3.84, + "learning_rate": 1.1136851576148867e-05, + "loss": 0.1722, + "step": 9167 + }, + { + "epoch": 3.84, + "learning_rate": 1.1135167142568186e-05, + "loss": 0.2082, + "step": 9168 + }, + { + "epoch": 3.84, + "learning_rate": 1.1133482676358131e-05, + "loss": 0.1616, + "step": 9169 + }, + { + "epoch": 3.84, + "learning_rate": 1.1131798177567129e-05, + "loss": 0.1815, + "step": 9170 + }, + { + "epoch": 3.84, + "learning_rate": 1.1130113646243585e-05, + "loss": 0.2332, + "step": 9171 + }, + { + "epoch": 3.84, + "learning_rate": 1.112842908243593e-05, + "loss": 0.1734, + "step": 9172 + }, + { + "epoch": 3.84, + "learning_rate": 1.1126744486192585e-05, + "loss": 0.1712, + "step": 9173 + }, + { + "epoch": 3.84, + "learning_rate": 1.1125059857561966e-05, + "loss": 0.2049, + "step": 9174 + }, + { + "epoch": 3.84, + "learning_rate": 1.11233751965925e-05, + "loss": 0.2336, + "step": 9175 + }, + { + "epoch": 3.84, + "learning_rate": 1.1121690503332613e-05, + "loss": 0.1608, + "step": 9176 + }, + { + "epoch": 3.84, + "learning_rate": 1.1120005777830729e-05, + "loss": 0.2013, + "step": 9177 + }, + { + "epoch": 3.84, + "learning_rate": 1.111832102013527e-05, + "loss": 0.1843, + "step": 9178 + }, + { + "epoch": 3.84, + "learning_rate": 1.1116636230294663e-05, + "loss": 0.1676, + "step": 9179 + }, + { + "epoch": 3.84, + "learning_rate": 1.1114951408357344e-05, + "loss": 0.1899, + "step": 9180 + }, + { + "epoch": 3.84, + "learning_rate": 1.1113266554371734e-05, + "loss": 0.2414, + "step": 9181 + }, + { + "epoch": 3.84, + "learning_rate": 1.1111581668386262e-05, + "loss": 0.1606, + "step": 9182 + }, + { + "epoch": 3.84, + "learning_rate": 1.1109896750449367e-05, + "loss": 0.1816, + "step": 9183 + }, + { + "epoch": 3.85, + "learning_rate": 1.110821180060947e-05, + "loss": 0.207, + "step": 9184 + }, + { + "epoch": 3.85, + "learning_rate": 1.1106526818915008e-05, + "loss": 0.1648, + "step": 9185 + }, + { + "epoch": 3.85, + "learning_rate": 1.1104841805414418e-05, + "loss": 0.2001, + "step": 9186 + }, + { + "epoch": 3.85, + "learning_rate": 1.1103156760156129e-05, + "loss": 0.154, + "step": 9187 + }, + { + "epoch": 3.85, + "learning_rate": 1.110147168318858e-05, + "loss": 0.183, + "step": 9188 + }, + { + "epoch": 3.85, + "learning_rate": 1.10997865745602e-05, + "loss": 0.2053, + "step": 9189 + }, + { + "epoch": 3.85, + "learning_rate": 1.1098101434319438e-05, + "loss": 0.2275, + "step": 9190 + }, + { + "epoch": 3.85, + "learning_rate": 1.1096416262514717e-05, + "loss": 0.197, + "step": 9191 + }, + { + "epoch": 3.85, + "learning_rate": 1.1094731059194488e-05, + "loss": 0.2249, + "step": 9192 + }, + { + "epoch": 3.85, + "learning_rate": 1.1093045824407186e-05, + "loss": 0.1649, + "step": 9193 + }, + { + "epoch": 3.85, + "learning_rate": 1.1091360558201249e-05, + "loss": 0.205, + "step": 9194 + }, + { + "epoch": 3.85, + "learning_rate": 1.1089675260625124e-05, + "loss": 0.1797, + "step": 9195 + }, + { + "epoch": 3.85, + "learning_rate": 1.1087989931727248e-05, + "loss": 0.1717, + "step": 9196 + }, + { + "epoch": 3.85, + "learning_rate": 1.1086304571556071e-05, + "loss": 0.1843, + "step": 9197 + }, + { + "epoch": 3.85, + "learning_rate": 1.1084619180160028e-05, + "loss": 0.2124, + "step": 9198 + }, + { + "epoch": 3.85, + "learning_rate": 1.1082933757587571e-05, + "loss": 0.1779, + "step": 9199 + }, + { + "epoch": 3.85, + "learning_rate": 1.1081248303887147e-05, + "loss": 0.2332, + "step": 9200 + }, + { + "epoch": 3.85, + "learning_rate": 1.1079562819107198e-05, + "loss": 0.2471, + "step": 9201 + }, + { + "epoch": 3.85, + "learning_rate": 1.1077877303296174e-05, + "loss": 0.1762, + "step": 9202 + }, + { + "epoch": 3.85, + "learning_rate": 1.1076191756502521e-05, + "loss": 0.2314, + "step": 9203 + }, + { + "epoch": 3.85, + "learning_rate": 1.1074506178774696e-05, + "loss": 0.173, + "step": 9204 + }, + { + "epoch": 3.85, + "learning_rate": 1.107282057016114e-05, + "loss": 0.2055, + "step": 9205 + }, + { + "epoch": 3.85, + "learning_rate": 1.1071134930710308e-05, + "loss": 0.1714, + "step": 9206 + }, + { + "epoch": 3.85, + "learning_rate": 1.1069449260470658e-05, + "loss": 0.1473, + "step": 9207 + }, + { + "epoch": 3.86, + "learning_rate": 1.1067763559490632e-05, + "loss": 0.149, + "step": 9208 + }, + { + "epoch": 3.86, + "learning_rate": 1.1066077827818694e-05, + "loss": 0.2025, + "step": 9209 + }, + { + "epoch": 3.86, + "learning_rate": 1.1064392065503294e-05, + "loss": 0.171, + "step": 9210 + }, + { + "epoch": 3.86, + "learning_rate": 1.106270627259289e-05, + "loss": 0.1824, + "step": 9211 + }, + { + "epoch": 3.86, + "learning_rate": 1.1061020449135935e-05, + "loss": 0.186, + "step": 9212 + }, + { + "epoch": 3.86, + "learning_rate": 1.1059334595180888e-05, + "loss": 0.1983, + "step": 9213 + }, + { + "epoch": 3.86, + "learning_rate": 1.1057648710776212e-05, + "loss": 0.1745, + "step": 9214 + }, + { + "epoch": 3.86, + "learning_rate": 1.1055962795970359e-05, + "loss": 0.2321, + "step": 9215 + }, + { + "epoch": 3.86, + "learning_rate": 1.1054276850811791e-05, + "loss": 0.2003, + "step": 9216 + }, + { + "epoch": 3.86, + "learning_rate": 1.1052590875348975e-05, + "loss": 0.2135, + "step": 9217 + }, + { + "epoch": 3.86, + "learning_rate": 1.1050904869630365e-05, + "loss": 0.2082, + "step": 9218 + }, + { + "epoch": 3.86, + "learning_rate": 1.1049218833704427e-05, + "loss": 0.2232, + "step": 9219 + }, + { + "epoch": 3.86, + "learning_rate": 1.1047532767619629e-05, + "loss": 0.2073, + "step": 9220 + }, + { + "epoch": 3.86, + "learning_rate": 1.1045846671424424e-05, + "loss": 0.1981, + "step": 9221 + }, + { + "epoch": 3.86, + "learning_rate": 1.1044160545167294e-05, + "loss": 0.1434, + "step": 9222 + }, + { + "epoch": 3.86, + "learning_rate": 1.1042474388896687e-05, + "loss": 0.1945, + "step": 9223 + }, + { + "epoch": 3.86, + "learning_rate": 1.1040788202661084e-05, + "loss": 0.1881, + "step": 9224 + }, + { + "epoch": 3.86, + "learning_rate": 1.1039101986508945e-05, + "loss": 0.1577, + "step": 9225 + }, + { + "epoch": 3.86, + "learning_rate": 1.1037415740488742e-05, + "loss": 0.1527, + "step": 9226 + }, + { + "epoch": 3.86, + "learning_rate": 1.1035729464648945e-05, + "loss": 0.2051, + "step": 9227 + }, + { + "epoch": 3.86, + "learning_rate": 1.1034043159038021e-05, + "loss": 0.1761, + "step": 9228 + }, + { + "epoch": 3.86, + "learning_rate": 1.1032356823704446e-05, + "loss": 0.2139, + "step": 9229 + }, + { + "epoch": 3.86, + "learning_rate": 1.1030670458696688e-05, + "loss": 0.2011, + "step": 9230 + }, + { + "epoch": 3.86, + "learning_rate": 1.1028984064063226e-05, + "loss": 0.1833, + "step": 9231 + }, + { + "epoch": 3.87, + "learning_rate": 1.1027297639852528e-05, + "loss": 0.1715, + "step": 9232 + }, + { + "epoch": 3.87, + "learning_rate": 1.1025611186113073e-05, + "loss": 0.1462, + "step": 9233 + }, + { + "epoch": 3.87, + "learning_rate": 1.1023924702893334e-05, + "loss": 0.1673, + "step": 9234 + }, + { + "epoch": 3.87, + "learning_rate": 1.1022238190241787e-05, + "loss": 0.1945, + "step": 9235 + }, + { + "epoch": 3.87, + "learning_rate": 1.102055164820691e-05, + "loss": 0.1562, + "step": 9236 + }, + { + "epoch": 3.87, + "learning_rate": 1.1018865076837182e-05, + "loss": 0.1894, + "step": 9237 + }, + { + "epoch": 3.87, + "learning_rate": 1.101717847618108e-05, + "loss": 0.1953, + "step": 9238 + }, + { + "epoch": 3.87, + "learning_rate": 1.1015491846287088e-05, + "loss": 0.2008, + "step": 9239 + }, + { + "epoch": 3.87, + "learning_rate": 1.1013805187203682e-05, + "loss": 0.2021, + "step": 9240 + }, + { + "epoch": 3.87, + "learning_rate": 1.1012118498979347e-05, + "loss": 0.199, + "step": 9241 + }, + { + "epoch": 3.87, + "learning_rate": 1.1010431781662563e-05, + "loss": 0.1611, + "step": 9242 + }, + { + "epoch": 3.87, + "learning_rate": 1.1008745035301814e-05, + "loss": 0.208, + "step": 9243 + }, + { + "epoch": 3.87, + "learning_rate": 1.1007058259945584e-05, + "loss": 0.1939, + "step": 9244 + }, + { + "epoch": 3.87, + "learning_rate": 1.100537145564236e-05, + "loss": 0.1942, + "step": 9245 + }, + { + "epoch": 3.87, + "learning_rate": 1.1003684622440622e-05, + "loss": 0.1912, + "step": 9246 + }, + { + "epoch": 3.87, + "learning_rate": 1.1001997760388862e-05, + "loss": 0.2018, + "step": 9247 + }, + { + "epoch": 3.87, + "learning_rate": 1.100031086953557e-05, + "loss": 0.1898, + "step": 9248 + }, + { + "epoch": 3.87, + "learning_rate": 1.0998623949929223e-05, + "loss": 0.1977, + "step": 9249 + }, + { + "epoch": 3.87, + "learning_rate": 1.0996937001618317e-05, + "loss": 0.1582, + "step": 9250 + }, + { + "epoch": 3.87, + "learning_rate": 1.0995250024651345e-05, + "loss": 0.1945, + "step": 9251 + }, + { + "epoch": 3.87, + "learning_rate": 1.0993563019076794e-05, + "loss": 0.1894, + "step": 9252 + }, + { + "epoch": 3.87, + "learning_rate": 1.0991875984943152e-05, + "loss": 0.1623, + "step": 9253 + }, + { + "epoch": 3.87, + "learning_rate": 1.0990188922298918e-05, + "loss": 0.1477, + "step": 9254 + }, + { + "epoch": 3.87, + "learning_rate": 1.0988501831192582e-05, + "loss": 0.1554, + "step": 9255 + }, + { + "epoch": 3.88, + "learning_rate": 1.098681471167264e-05, + "loss": 0.2056, + "step": 9256 + }, + { + "epoch": 3.88, + "learning_rate": 1.098512756378758e-05, + "loss": 0.2109, + "step": 9257 + }, + { + "epoch": 3.88, + "learning_rate": 1.0983440387585909e-05, + "loss": 0.15, + "step": 9258 + }, + { + "epoch": 3.88, + "learning_rate": 1.0981753183116111e-05, + "loss": 0.2154, + "step": 9259 + }, + { + "epoch": 3.88, + "learning_rate": 1.0980065950426687e-05, + "loss": 0.2, + "step": 9260 + }, + { + "epoch": 3.88, + "learning_rate": 1.0978378689566144e-05, + "loss": 0.217, + "step": 9261 + }, + { + "epoch": 3.88, + "learning_rate": 1.097669140058297e-05, + "loss": 0.2133, + "step": 9262 + }, + { + "epoch": 3.88, + "learning_rate": 1.0975004083525667e-05, + "loss": 0.198, + "step": 9263 + }, + { + "epoch": 3.88, + "learning_rate": 1.0973316738442738e-05, + "loss": 0.2014, + "step": 9264 + }, + { + "epoch": 3.88, + "learning_rate": 1.0971629365382686e-05, + "loss": 0.1903, + "step": 9265 + }, + { + "epoch": 3.88, + "learning_rate": 1.0969941964394006e-05, + "loss": 0.1916, + "step": 9266 + }, + { + "epoch": 3.88, + "learning_rate": 1.0968254535525204e-05, + "loss": 0.2014, + "step": 9267 + }, + { + "epoch": 3.88, + "learning_rate": 1.0966567078824792e-05, + "loss": 0.1985, + "step": 9268 + }, + { + "epoch": 3.88, + "learning_rate": 1.0964879594341262e-05, + "loss": 0.1646, + "step": 9269 + }, + { + "epoch": 3.88, + "learning_rate": 1.0963192082123124e-05, + "loss": 0.1842, + "step": 9270 + }, + { + "epoch": 3.88, + "learning_rate": 1.0961504542218884e-05, + "loss": 0.2025, + "step": 9271 + }, + { + "epoch": 3.88, + "learning_rate": 1.0959816974677052e-05, + "loss": 0.1725, + "step": 9272 + }, + { + "epoch": 3.88, + "learning_rate": 1.095812937954613e-05, + "loss": 0.2005, + "step": 9273 + }, + { + "epoch": 3.88, + "learning_rate": 1.0956441756874628e-05, + "loss": 0.1933, + "step": 9274 + }, + { + "epoch": 3.88, + "learning_rate": 1.095475410671106e-05, + "loss": 0.1843, + "step": 9275 + }, + { + "epoch": 3.88, + "learning_rate": 1.0953066429103929e-05, + "loss": 0.2184, + "step": 9276 + }, + { + "epoch": 3.88, + "learning_rate": 1.0951378724101751e-05, + "loss": 0.2005, + "step": 9277 + }, + { + "epoch": 3.88, + "learning_rate": 1.0949690991753036e-05, + "loss": 0.1757, + "step": 9278 + }, + { + "epoch": 3.88, + "learning_rate": 1.0948003232106298e-05, + "loss": 0.1942, + "step": 9279 + }, + { + "epoch": 3.89, + "learning_rate": 1.0946315445210048e-05, + "loss": 0.1583, + "step": 9280 + }, + { + "epoch": 3.89, + "learning_rate": 1.0944627631112796e-05, + "loss": 0.1767, + "step": 9281 + }, + { + "epoch": 3.89, + "learning_rate": 1.0942939789863065e-05, + "loss": 0.2008, + "step": 9282 + }, + { + "epoch": 3.89, + "learning_rate": 1.0941251921509367e-05, + "loss": 0.1965, + "step": 9283 + }, + { + "epoch": 3.89, + "learning_rate": 1.0939564026100213e-05, + "loss": 0.2036, + "step": 9284 + }, + { + "epoch": 3.89, + "learning_rate": 1.093787610368413e-05, + "loss": 0.1682, + "step": 9285 + }, + { + "epoch": 3.89, + "learning_rate": 1.0936188154309626e-05, + "loss": 0.1771, + "step": 9286 + }, + { + "epoch": 3.89, + "learning_rate": 1.0934500178025228e-05, + "loss": 0.1683, + "step": 9287 + }, + { + "epoch": 3.89, + "learning_rate": 1.093281217487945e-05, + "loss": 0.1783, + "step": 9288 + }, + { + "epoch": 3.89, + "learning_rate": 1.0931124144920813e-05, + "loss": 0.2245, + "step": 9289 + }, + { + "epoch": 3.89, + "learning_rate": 1.0929436088197842e-05, + "loss": 0.2076, + "step": 9290 + }, + { + "epoch": 3.89, + "learning_rate": 1.0927748004759053e-05, + "loss": 0.1446, + "step": 9291 + }, + { + "epoch": 3.89, + "learning_rate": 1.0926059894652972e-05, + "loss": 0.1343, + "step": 9292 + }, + { + "epoch": 3.89, + "learning_rate": 1.0924371757928121e-05, + "loss": 0.1693, + "step": 9293 + }, + { + "epoch": 3.89, + "learning_rate": 1.092268359463302e-05, + "loss": 0.2063, + "step": 9294 + }, + { + "epoch": 3.89, + "learning_rate": 1.0920995404816204e-05, + "loss": 0.1792, + "step": 9295 + }, + { + "epoch": 3.89, + "learning_rate": 1.091930718852619e-05, + "loss": 0.1761, + "step": 9296 + }, + { + "epoch": 3.89, + "learning_rate": 1.0917618945811503e-05, + "loss": 0.1832, + "step": 9297 + }, + { + "epoch": 3.89, + "learning_rate": 1.0915930676720678e-05, + "loss": 0.1779, + "step": 9298 + }, + { + "epoch": 3.89, + "learning_rate": 1.0914242381302238e-05, + "loss": 0.1716, + "step": 9299 + }, + { + "epoch": 3.89, + "learning_rate": 1.091255405960471e-05, + "loss": 0.173, + "step": 9300 + }, + { + "epoch": 3.89, + "learning_rate": 1.0910865711676628e-05, + "loss": 0.1813, + "step": 9301 + }, + { + "epoch": 3.89, + "learning_rate": 1.0909177337566518e-05, + "loss": 0.1746, + "step": 9302 + }, + { + "epoch": 3.89, + "learning_rate": 1.0907488937322915e-05, + "loss": 0.1858, + "step": 9303 + }, + { + "epoch": 3.9, + "learning_rate": 1.0905800510994341e-05, + "loss": 0.2029, + "step": 9304 + }, + { + "epoch": 3.9, + "learning_rate": 1.0904112058629341e-05, + "loss": 0.1727, + "step": 9305 + }, + { + "epoch": 3.9, + "learning_rate": 1.090242358027644e-05, + "loss": 0.1983, + "step": 9306 + }, + { + "epoch": 3.9, + "learning_rate": 1.0900735075984174e-05, + "loss": 0.1724, + "step": 9307 + }, + { + "epoch": 3.9, + "learning_rate": 1.0899046545801078e-05, + "loss": 0.1676, + "step": 9308 + }, + { + "epoch": 3.9, + "learning_rate": 1.0897357989775687e-05, + "loss": 0.1845, + "step": 9309 + }, + { + "epoch": 3.9, + "learning_rate": 1.0895669407956538e-05, + "loss": 0.1733, + "step": 9310 + }, + { + "epoch": 3.9, + "learning_rate": 1.0893980800392164e-05, + "loss": 0.209, + "step": 9311 + }, + { + "epoch": 3.9, + "learning_rate": 1.0892292167131107e-05, + "loss": 0.2143, + "step": 9312 + }, + { + "epoch": 3.9, + "learning_rate": 1.0890603508221905e-05, + "loss": 0.1837, + "step": 9313 + }, + { + "epoch": 3.9, + "learning_rate": 1.0888914823713092e-05, + "loss": 0.2079, + "step": 9314 + }, + { + "epoch": 3.9, + "learning_rate": 1.0887226113653211e-05, + "loss": 0.1714, + "step": 9315 + }, + { + "epoch": 3.9, + "learning_rate": 1.0885537378090807e-05, + "loss": 0.1785, + "step": 9316 + }, + { + "epoch": 3.9, + "learning_rate": 1.0883848617074412e-05, + "loss": 0.1692, + "step": 9317 + }, + { + "epoch": 3.9, + "learning_rate": 1.0882159830652574e-05, + "loss": 0.2079, + "step": 9318 + }, + { + "epoch": 3.9, + "learning_rate": 1.0880471018873836e-05, + "loss": 0.2076, + "step": 9319 + }, + { + "epoch": 3.9, + "learning_rate": 1.087878218178674e-05, + "loss": 0.2011, + "step": 9320 + }, + { + "epoch": 3.9, + "learning_rate": 1.0877093319439828e-05, + "loss": 0.1759, + "step": 9321 + }, + { + "epoch": 3.9, + "learning_rate": 1.0875404431881651e-05, + "loss": 0.2012, + "step": 9322 + }, + { + "epoch": 3.9, + "learning_rate": 1.0873715519160746e-05, + "loss": 0.1937, + "step": 9323 + }, + { + "epoch": 3.9, + "learning_rate": 1.0872026581325667e-05, + "loss": 0.1947, + "step": 9324 + }, + { + "epoch": 3.9, + "learning_rate": 1.0870337618424956e-05, + "loss": 0.1981, + "step": 9325 + }, + { + "epoch": 3.9, + "learning_rate": 1.0868648630507164e-05, + "loss": 0.2003, + "step": 9326 + }, + { + "epoch": 3.9, + "learning_rate": 1.0866959617620835e-05, + "loss": 0.2131, + "step": 9327 + }, + { + "epoch": 3.91, + "learning_rate": 1.0865270579814522e-05, + "loss": 0.1698, + "step": 9328 + }, + { + "epoch": 3.91, + "learning_rate": 1.0863581517136776e-05, + "loss": 0.1846, + "step": 9329 + }, + { + "epoch": 3.91, + "learning_rate": 1.0861892429636143e-05, + "loss": 0.1581, + "step": 9330 + }, + { + "epoch": 3.91, + "learning_rate": 1.0860203317361178e-05, + "loss": 0.1566, + "step": 9331 + }, + { + "epoch": 3.91, + "learning_rate": 1.0858514180360432e-05, + "loss": 0.2144, + "step": 9332 + }, + { + "epoch": 3.91, + "learning_rate": 1.085682501868246e-05, + "loss": 0.1654, + "step": 9333 + }, + { + "epoch": 3.91, + "learning_rate": 1.0855135832375809e-05, + "loss": 0.2098, + "step": 9334 + }, + { + "epoch": 3.91, + "learning_rate": 1.0853446621489042e-05, + "loss": 0.2019, + "step": 9335 + }, + { + "epoch": 3.91, + "learning_rate": 1.0851757386070706e-05, + "loss": 0.1809, + "step": 9336 + }, + { + "epoch": 3.91, + "learning_rate": 1.085006812616936e-05, + "loss": 0.1719, + "step": 9337 + }, + { + "epoch": 3.91, + "learning_rate": 1.084837884183356e-05, + "loss": 0.2099, + "step": 9338 + }, + { + "epoch": 3.91, + "learning_rate": 1.0846689533111865e-05, + "loss": 0.1807, + "step": 9339 + }, + { + "epoch": 3.91, + "learning_rate": 1.0845000200052828e-05, + "loss": 0.198, + "step": 9340 + }, + { + "epoch": 3.91, + "learning_rate": 1.0843310842705011e-05, + "loss": 0.1822, + "step": 9341 + }, + { + "epoch": 3.91, + "learning_rate": 1.0841621461116971e-05, + "loss": 0.2146, + "step": 9342 + }, + { + "epoch": 3.91, + "learning_rate": 1.0839932055337273e-05, + "loss": 0.1954, + "step": 9343 + }, + { + "epoch": 3.91, + "learning_rate": 1.083824262541447e-05, + "loss": 0.1589, + "step": 9344 + }, + { + "epoch": 3.91, + "learning_rate": 1.0836553171397126e-05, + "loss": 0.2323, + "step": 9345 + }, + { + "epoch": 3.91, + "learning_rate": 1.0834863693333805e-05, + "loss": 0.1982, + "step": 9346 + }, + { + "epoch": 3.91, + "learning_rate": 1.083317419127307e-05, + "loss": 0.1637, + "step": 9347 + }, + { + "epoch": 3.91, + "learning_rate": 1.0831484665263479e-05, + "loss": 0.1847, + "step": 9348 + }, + { + "epoch": 3.91, + "learning_rate": 1.0829795115353601e-05, + "loss": 0.1476, + "step": 9349 + }, + { + "epoch": 3.91, + "learning_rate": 1.0828105541591998e-05, + "loss": 0.1757, + "step": 9350 + }, + { + "epoch": 3.92, + "learning_rate": 1.0826415944027235e-05, + "loss": 0.214, + "step": 9351 + }, + { + "epoch": 3.92, + "learning_rate": 1.0824726322707878e-05, + "loss": 0.217, + "step": 9352 + }, + { + "epoch": 3.92, + "learning_rate": 1.0823036677682499e-05, + "loss": 0.1895, + "step": 9353 + }, + { + "epoch": 3.92, + "learning_rate": 1.0821347008999658e-05, + "loss": 0.1615, + "step": 9354 + }, + { + "epoch": 3.92, + "learning_rate": 1.0819657316707926e-05, + "loss": 0.1782, + "step": 9355 + }, + { + "epoch": 3.92, + "learning_rate": 1.0817967600855874e-05, + "loss": 0.1674, + "step": 9356 + }, + { + "epoch": 3.92, + "learning_rate": 1.0816277861492068e-05, + "loss": 0.1784, + "step": 9357 + }, + { + "epoch": 3.92, + "learning_rate": 1.081458809866508e-05, + "loss": 0.1907, + "step": 9358 + }, + { + "epoch": 3.92, + "learning_rate": 1.0812898312423478e-05, + "loss": 0.1914, + "step": 9359 + }, + { + "epoch": 3.92, + "learning_rate": 1.0811208502815838e-05, + "loss": 0.21, + "step": 9360 + }, + { + "epoch": 3.92, + "learning_rate": 1.0809518669890726e-05, + "loss": 0.1671, + "step": 9361 + }, + { + "epoch": 3.92, + "learning_rate": 1.0807828813696719e-05, + "loss": 0.1829, + "step": 9362 + }, + { + "epoch": 3.92, + "learning_rate": 1.0806138934282393e-05, + "loss": 0.2003, + "step": 9363 + }, + { + "epoch": 3.92, + "learning_rate": 1.0804449031696315e-05, + "loss": 0.1604, + "step": 9364 + }, + { + "epoch": 3.92, + "learning_rate": 1.0802759105987065e-05, + "loss": 0.1791, + "step": 9365 + }, + { + "epoch": 3.92, + "learning_rate": 1.0801069157203217e-05, + "loss": 0.1715, + "step": 9366 + }, + { + "epoch": 3.92, + "learning_rate": 1.0799379185393347e-05, + "loss": 0.1648, + "step": 9367 + }, + { + "epoch": 3.92, + "learning_rate": 1.0797689190606028e-05, + "loss": 0.1898, + "step": 9368 + }, + { + "epoch": 3.92, + "learning_rate": 1.0795999172889847e-05, + "loss": 0.1862, + "step": 9369 + }, + { + "epoch": 3.92, + "learning_rate": 1.0794309132293374e-05, + "loss": 0.1888, + "step": 9370 + }, + { + "epoch": 3.92, + "learning_rate": 1.0792619068865189e-05, + "loss": 0.1692, + "step": 9371 + }, + { + "epoch": 3.92, + "learning_rate": 1.0790928982653873e-05, + "loss": 0.2446, + "step": 9372 + }, + { + "epoch": 3.92, + "learning_rate": 1.0789238873708004e-05, + "loss": 0.2284, + "step": 9373 + }, + { + "epoch": 3.92, + "learning_rate": 1.0787548742076164e-05, + "loss": 0.1938, + "step": 9374 + }, + { + "epoch": 3.93, + "learning_rate": 1.0785858587806935e-05, + "loss": 0.1542, + "step": 9375 + }, + { + "epoch": 3.93, + "learning_rate": 1.0784168410948897e-05, + "loss": 0.164, + "step": 9376 + }, + { + "epoch": 3.93, + "learning_rate": 1.0782478211550636e-05, + "loss": 0.1857, + "step": 9377 + }, + { + "epoch": 3.93, + "learning_rate": 1.078078798966073e-05, + "loss": 0.189, + "step": 9378 + }, + { + "epoch": 3.93, + "learning_rate": 1.0779097745327768e-05, + "loss": 0.2207, + "step": 9379 + }, + { + "epoch": 3.93, + "learning_rate": 1.0777407478600334e-05, + "loss": 0.1534, + "step": 9380 + }, + { + "epoch": 3.93, + "learning_rate": 1.0775717189527012e-05, + "loss": 0.2344, + "step": 9381 + }, + { + "epoch": 3.93, + "learning_rate": 1.0774026878156387e-05, + "loss": 0.218, + "step": 9382 + }, + { + "epoch": 3.93, + "learning_rate": 1.0772336544537045e-05, + "loss": 0.2319, + "step": 9383 + }, + { + "epoch": 3.93, + "learning_rate": 1.0770646188717578e-05, + "loss": 0.177, + "step": 9384 + }, + { + "epoch": 3.93, + "learning_rate": 1.0768955810746564e-05, + "loss": 0.1836, + "step": 9385 + }, + { + "epoch": 3.93, + "learning_rate": 1.07672654106726e-05, + "loss": 0.1955, + "step": 9386 + }, + { + "epoch": 3.93, + "learning_rate": 1.0765574988544276e-05, + "loss": 0.1846, + "step": 9387 + }, + { + "epoch": 3.93, + "learning_rate": 1.0763884544410174e-05, + "loss": 0.1924, + "step": 9388 + }, + { + "epoch": 3.93, + "learning_rate": 1.0762194078318891e-05, + "loss": 0.1258, + "step": 9389 + }, + { + "epoch": 3.93, + "learning_rate": 1.0760503590319016e-05, + "loss": 0.1639, + "step": 9390 + }, + { + "epoch": 3.93, + "learning_rate": 1.0758813080459138e-05, + "loss": 0.1734, + "step": 9391 + }, + { + "epoch": 3.93, + "learning_rate": 1.0757122548787855e-05, + "loss": 0.1761, + "step": 9392 + }, + { + "epoch": 3.93, + "learning_rate": 1.0755431995353753e-05, + "loss": 0.199, + "step": 9393 + }, + { + "epoch": 3.93, + "learning_rate": 1.0753741420205431e-05, + "loss": 0.1722, + "step": 9394 + }, + { + "epoch": 3.93, + "learning_rate": 1.0752050823391478e-05, + "loss": 0.1927, + "step": 9395 + }, + { + "epoch": 3.93, + "learning_rate": 1.0750360204960493e-05, + "loss": 0.1614, + "step": 9396 + }, + { + "epoch": 3.93, + "learning_rate": 1.074866956496107e-05, + "loss": 0.2037, + "step": 9397 + }, + { + "epoch": 3.93, + "learning_rate": 1.0746978903441803e-05, + "loss": 0.1731, + "step": 9398 + }, + { + "epoch": 3.94, + "learning_rate": 1.0745288220451291e-05, + "loss": 0.1679, + "step": 9399 + }, + { + "epoch": 3.94, + "learning_rate": 1.0743597516038132e-05, + "loss": 0.1661, + "step": 9400 + }, + { + "epoch": 3.94, + "learning_rate": 1.074190679025092e-05, + "loss": 0.2065, + "step": 9401 + }, + { + "epoch": 3.94, + "learning_rate": 1.0740216043138256e-05, + "loss": 0.1841, + "step": 9402 + }, + { + "epoch": 3.94, + "learning_rate": 1.073852527474874e-05, + "loss": 0.1781, + "step": 9403 + }, + { + "epoch": 3.94, + "learning_rate": 1.0736834485130972e-05, + "loss": 0.1686, + "step": 9404 + }, + { + "epoch": 3.94, + "learning_rate": 1.0735143674333549e-05, + "loss": 0.188, + "step": 9405 + }, + { + "epoch": 3.94, + "learning_rate": 1.073345284240507e-05, + "loss": 0.1495, + "step": 9406 + }, + { + "epoch": 3.94, + "learning_rate": 1.0731761989394145e-05, + "loss": 0.2022, + "step": 9407 + }, + { + "epoch": 3.94, + "learning_rate": 1.0730071115349367e-05, + "loss": 0.2069, + "step": 9408 + }, + { + "epoch": 3.94, + "learning_rate": 1.0728380220319345e-05, + "loss": 0.2076, + "step": 9409 + }, + { + "epoch": 3.94, + "learning_rate": 1.072668930435268e-05, + "loss": 0.2082, + "step": 9410 + }, + { + "epoch": 3.94, + "learning_rate": 1.0724998367497974e-05, + "loss": 0.2061, + "step": 9411 + }, + { + "epoch": 3.94, + "learning_rate": 1.0723307409803834e-05, + "loss": 0.2191, + "step": 9412 + }, + { + "epoch": 3.94, + "learning_rate": 1.0721616431318865e-05, + "loss": 0.1913, + "step": 9413 + }, + { + "epoch": 3.94, + "learning_rate": 1.0719925432091671e-05, + "loss": 0.2349, + "step": 9414 + }, + { + "epoch": 3.94, + "learning_rate": 1.0718234412170862e-05, + "loss": 0.1887, + "step": 9415 + }, + { + "epoch": 3.94, + "learning_rate": 1.0716543371605039e-05, + "loss": 0.1745, + "step": 9416 + }, + { + "epoch": 3.94, + "learning_rate": 1.0714852310442817e-05, + "loss": 0.1596, + "step": 9417 + }, + { + "epoch": 3.94, + "learning_rate": 1.0713161228732795e-05, + "loss": 0.2257, + "step": 9418 + }, + { + "epoch": 3.94, + "learning_rate": 1.0711470126523587e-05, + "loss": 0.2298, + "step": 9419 + }, + { + "epoch": 3.94, + "learning_rate": 1.0709779003863801e-05, + "loss": 0.2116, + "step": 9420 + }, + { + "epoch": 3.94, + "learning_rate": 1.070808786080205e-05, + "loss": 0.1933, + "step": 9421 + }, + { + "epoch": 3.94, + "learning_rate": 1.070639669738694e-05, + "loss": 0.1895, + "step": 9422 + }, + { + "epoch": 3.95, + "learning_rate": 1.0704705513667084e-05, + "loss": 0.2109, + "step": 9423 + }, + { + "epoch": 3.95, + "learning_rate": 1.0703014309691096e-05, + "loss": 0.1643, + "step": 9424 + }, + { + "epoch": 3.95, + "learning_rate": 1.0701323085507583e-05, + "loss": 0.1672, + "step": 9425 + }, + { + "epoch": 3.95, + "learning_rate": 1.0699631841165162e-05, + "loss": 0.1769, + "step": 9426 + }, + { + "epoch": 3.95, + "learning_rate": 1.069794057671244e-05, + "loss": 0.2088, + "step": 9427 + }, + { + "epoch": 3.95, + "learning_rate": 1.069624929219804e-05, + "loss": 0.1518, + "step": 9428 + }, + { + "epoch": 3.95, + "learning_rate": 1.069455798767057e-05, + "loss": 0.1869, + "step": 9429 + }, + { + "epoch": 3.95, + "learning_rate": 1.0692866663178646e-05, + "loss": 0.1764, + "step": 9430 + }, + { + "epoch": 3.95, + "learning_rate": 1.0691175318770887e-05, + "loss": 0.1703, + "step": 9431 + }, + { + "epoch": 3.95, + "learning_rate": 1.0689483954495906e-05, + "loss": 0.1435, + "step": 9432 + }, + { + "epoch": 3.95, + "learning_rate": 1.068779257040232e-05, + "loss": 0.1748, + "step": 9433 + }, + { + "epoch": 3.95, + "learning_rate": 1.068610116653875e-05, + "loss": 0.1518, + "step": 9434 + }, + { + "epoch": 3.95, + "learning_rate": 1.0684409742953805e-05, + "loss": 0.1742, + "step": 9435 + }, + { + "epoch": 3.95, + "learning_rate": 1.0682718299696113e-05, + "loss": 0.1625, + "step": 9436 + }, + { + "epoch": 3.95, + "learning_rate": 1.068102683681429e-05, + "loss": 0.1579, + "step": 9437 + }, + { + "epoch": 3.95, + "learning_rate": 1.0679335354356954e-05, + "loss": 0.2154, + "step": 9438 + }, + { + "epoch": 3.95, + "learning_rate": 1.0677643852372726e-05, + "loss": 0.1967, + "step": 9439 + }, + { + "epoch": 3.95, + "learning_rate": 1.0675952330910224e-05, + "loss": 0.1679, + "step": 9440 + }, + { + "epoch": 3.95, + "learning_rate": 1.0674260790018076e-05, + "loss": 0.1793, + "step": 9441 + }, + { + "epoch": 3.95, + "learning_rate": 1.0672569229744899e-05, + "loss": 0.2155, + "step": 9442 + }, + { + "epoch": 3.95, + "learning_rate": 1.0670877650139316e-05, + "loss": 0.1693, + "step": 9443 + }, + { + "epoch": 3.95, + "learning_rate": 1.0669186051249948e-05, + "loss": 0.1585, + "step": 9444 + }, + { + "epoch": 3.95, + "learning_rate": 1.0667494433125423e-05, + "loss": 0.1796, + "step": 9445 + }, + { + "epoch": 3.95, + "learning_rate": 1.0665802795814362e-05, + "loss": 0.1399, + "step": 9446 + }, + { + "epoch": 3.96, + "learning_rate": 1.0664111139365392e-05, + "loss": 0.1708, + "step": 9447 + }, + { + "epoch": 3.96, + "learning_rate": 1.0662419463827136e-05, + "loss": 0.1761, + "step": 9448 + }, + { + "epoch": 3.96, + "learning_rate": 1.0660727769248221e-05, + "loss": 0.1746, + "step": 9449 + }, + { + "epoch": 3.96, + "learning_rate": 1.0659036055677272e-05, + "loss": 0.1718, + "step": 9450 + }, + { + "epoch": 3.96, + "learning_rate": 1.0657344323162916e-05, + "loss": 0.2022, + "step": 9451 + }, + { + "epoch": 3.96, + "learning_rate": 1.0655652571753781e-05, + "loss": 0.1998, + "step": 9452 + }, + { + "epoch": 3.96, + "learning_rate": 1.0653960801498493e-05, + "loss": 0.1449, + "step": 9453 + }, + { + "epoch": 3.96, + "learning_rate": 1.0652269012445684e-05, + "loss": 0.2073, + "step": 9454 + }, + { + "epoch": 3.96, + "learning_rate": 1.0650577204643981e-05, + "loss": 0.2105, + "step": 9455 + }, + { + "epoch": 3.96, + "learning_rate": 1.0648885378142013e-05, + "loss": 0.159, + "step": 9456 + }, + { + "epoch": 3.96, + "learning_rate": 1.0647193532988409e-05, + "loss": 0.2047, + "step": 9457 + }, + { + "epoch": 3.96, + "learning_rate": 1.0645501669231803e-05, + "loss": 0.1614, + "step": 9458 + }, + { + "epoch": 3.96, + "learning_rate": 1.0643809786920824e-05, + "loss": 0.2312, + "step": 9459 + }, + { + "epoch": 3.96, + "learning_rate": 1.0642117886104106e-05, + "loss": 0.195, + "step": 9460 + }, + { + "epoch": 3.96, + "learning_rate": 1.0640425966830276e-05, + "loss": 0.1706, + "step": 9461 + }, + { + "epoch": 3.96, + "learning_rate": 1.0638734029147974e-05, + "loss": 0.1486, + "step": 9462 + }, + { + "epoch": 3.96, + "learning_rate": 1.0637042073105826e-05, + "loss": 0.1422, + "step": 9463 + }, + { + "epoch": 3.96, + "learning_rate": 1.0635350098752468e-05, + "loss": 0.1756, + "step": 9464 + }, + { + "epoch": 3.96, + "learning_rate": 1.0633658106136538e-05, + "loss": 0.2265, + "step": 9465 + }, + { + "epoch": 3.96, + "learning_rate": 1.0631966095306668e-05, + "loss": 0.2164, + "step": 9466 + }, + { + "epoch": 3.96, + "learning_rate": 1.063027406631149e-05, + "loss": 0.204, + "step": 9467 + }, + { + "epoch": 3.96, + "learning_rate": 1.0628582019199648e-05, + "loss": 0.2023, + "step": 9468 + }, + { + "epoch": 3.96, + "learning_rate": 1.0626889954019768e-05, + "loss": 0.1844, + "step": 9469 + }, + { + "epoch": 3.96, + "learning_rate": 1.0625197870820493e-05, + "loss": 0.1986, + "step": 9470 + }, + { + "epoch": 3.97, + "learning_rate": 1.0623505769650466e-05, + "loss": 0.1838, + "step": 9471 + }, + { + "epoch": 3.97, + "learning_rate": 1.0621813650558317e-05, + "loss": 0.1745, + "step": 9472 + }, + { + "epoch": 3.97, + "learning_rate": 1.0620121513592682e-05, + "loss": 0.1978, + "step": 9473 + }, + { + "epoch": 3.97, + "learning_rate": 1.0618429358802209e-05, + "loss": 0.2103, + "step": 9474 + }, + { + "epoch": 3.97, + "learning_rate": 1.061673718623553e-05, + "loss": 0.1763, + "step": 9475 + }, + { + "epoch": 3.97, + "learning_rate": 1.061504499594129e-05, + "loss": 0.1444, + "step": 9476 + }, + { + "epoch": 3.97, + "learning_rate": 1.0613352787968125e-05, + "loss": 0.1943, + "step": 9477 + }, + { + "epoch": 3.97, + "learning_rate": 1.061166056236468e-05, + "loss": 0.1833, + "step": 9478 + }, + { + "epoch": 3.97, + "learning_rate": 1.0609968319179595e-05, + "loss": 0.1546, + "step": 9479 + }, + { + "epoch": 3.97, + "learning_rate": 1.060827605846151e-05, + "loss": 0.2261, + "step": 9480 + }, + { + "epoch": 3.97, + "learning_rate": 1.0606583780259073e-05, + "loss": 0.1785, + "step": 9481 + }, + { + "epoch": 3.97, + "learning_rate": 1.060489148462092e-05, + "loss": 0.1689, + "step": 9482 + }, + { + "epoch": 3.97, + "learning_rate": 1.0603199171595701e-05, + "loss": 0.1889, + "step": 9483 + }, + { + "epoch": 3.97, + "learning_rate": 1.0601506841232057e-05, + "loss": 0.1917, + "step": 9484 + }, + { + "epoch": 3.97, + "learning_rate": 1.059981449357863e-05, + "loss": 0.2113, + "step": 9485 + }, + { + "epoch": 3.97, + "learning_rate": 1.0598122128684069e-05, + "loss": 0.1792, + "step": 9486 + }, + { + "epoch": 3.97, + "learning_rate": 1.0596429746597017e-05, + "loss": 0.2004, + "step": 9487 + }, + { + "epoch": 3.97, + "learning_rate": 1.0594737347366125e-05, + "loss": 0.166, + "step": 9488 + }, + { + "epoch": 3.97, + "learning_rate": 1.059304493104003e-05, + "loss": 0.1494, + "step": 9489 + }, + { + "epoch": 3.97, + "learning_rate": 1.0591352497667388e-05, + "loss": 0.179, + "step": 9490 + }, + { + "epoch": 3.97, + "learning_rate": 1.0589660047296844e-05, + "loss": 0.1768, + "step": 9491 + }, + { + "epoch": 3.97, + "learning_rate": 1.0587967579977043e-05, + "loss": 0.2044, + "step": 9492 + }, + { + "epoch": 3.97, + "learning_rate": 1.0586275095756638e-05, + "loss": 0.1723, + "step": 9493 + }, + { + "epoch": 3.97, + "learning_rate": 1.0584582594684274e-05, + "loss": 0.1643, + "step": 9494 + }, + { + "epoch": 3.98, + "learning_rate": 1.0582890076808603e-05, + "loss": 0.1814, + "step": 9495 + }, + { + "epoch": 3.98, + "learning_rate": 1.0581197542178274e-05, + "loss": 0.1636, + "step": 9496 + }, + { + "epoch": 3.98, + "learning_rate": 1.0579504990841935e-05, + "loss": 0.2118, + "step": 9497 + }, + { + "epoch": 3.98, + "learning_rate": 1.0577812422848241e-05, + "loss": 0.1787, + "step": 9498 + }, + { + "epoch": 3.98, + "learning_rate": 1.0576119838245843e-05, + "loss": 0.211, + "step": 9499 + }, + { + "epoch": 3.98, + "learning_rate": 1.057442723708339e-05, + "loss": 0.1991, + "step": 9500 + }, + { + "epoch": 3.98, + "learning_rate": 1.0572734619409535e-05, + "loss": 0.1995, + "step": 9501 + }, + { + "epoch": 3.98, + "learning_rate": 1.0571041985272934e-05, + "loss": 0.1729, + "step": 9502 + }, + { + "epoch": 3.98, + "learning_rate": 1.0569349334722236e-05, + "loss": 0.1999, + "step": 9503 + }, + { + "epoch": 3.98, + "learning_rate": 1.0567656667806097e-05, + "loss": 0.192, + "step": 9504 + }, + { + "epoch": 3.98, + "learning_rate": 1.0565963984573174e-05, + "loss": 0.1887, + "step": 9505 + }, + { + "epoch": 3.98, + "learning_rate": 1.0564271285072117e-05, + "loss": 0.1744, + "step": 9506 + }, + { + "epoch": 3.98, + "learning_rate": 1.0562578569351579e-05, + "loss": 0.1877, + "step": 9507 + }, + { + "epoch": 3.98, + "learning_rate": 1.056088583746022e-05, + "loss": 0.1796, + "step": 9508 + }, + { + "epoch": 3.98, + "learning_rate": 1.0559193089446698e-05, + "loss": 0.2141, + "step": 9509 + }, + { + "epoch": 3.98, + "learning_rate": 1.0557500325359665e-05, + "loss": 0.1828, + "step": 9510 + }, + { + "epoch": 3.98, + "learning_rate": 1.055580754524778e-05, + "loss": 0.1784, + "step": 9511 + }, + { + "epoch": 3.98, + "learning_rate": 1.05541147491597e-05, + "loss": 0.1886, + "step": 9512 + }, + { + "epoch": 3.98, + "learning_rate": 1.0552421937144083e-05, + "loss": 0.1875, + "step": 9513 + }, + { + "epoch": 3.98, + "learning_rate": 1.055072910924959e-05, + "loss": 0.1703, + "step": 9514 + }, + { + "epoch": 3.98, + "learning_rate": 1.0549036265524877e-05, + "loss": 0.1968, + "step": 9515 + }, + { + "epoch": 3.98, + "learning_rate": 1.0547343406018602e-05, + "loss": 0.1663, + "step": 9516 + }, + { + "epoch": 3.98, + "learning_rate": 1.0545650530779428e-05, + "loss": 0.1886, + "step": 9517 + }, + { + "epoch": 3.98, + "learning_rate": 1.0543957639856014e-05, + "loss": 0.1585, + "step": 9518 + }, + { + "epoch": 3.99, + "learning_rate": 1.0542264733297021e-05, + "loss": 0.1914, + "step": 9519 + }, + { + "epoch": 3.99, + "learning_rate": 1.0540571811151106e-05, + "loss": 0.163, + "step": 9520 + }, + { + "epoch": 3.99, + "learning_rate": 1.0538878873466936e-05, + "loss": 0.2158, + "step": 9521 + }, + { + "epoch": 3.99, + "learning_rate": 1.0537185920293171e-05, + "loss": 0.1514, + "step": 9522 + }, + { + "epoch": 3.99, + "learning_rate": 1.0535492951678471e-05, + "loss": 0.1868, + "step": 9523 + }, + { + "epoch": 3.99, + "learning_rate": 1.0533799967671503e-05, + "loss": 0.1927, + "step": 9524 + }, + { + "epoch": 3.99, + "learning_rate": 1.053210696832093e-05, + "loss": 0.1469, + "step": 9525 + }, + { + "epoch": 3.99, + "learning_rate": 1.0530413953675414e-05, + "loss": 0.1823, + "step": 9526 + }, + { + "epoch": 3.99, + "learning_rate": 1.0528720923783617e-05, + "loss": 0.2002, + "step": 9527 + }, + { + "epoch": 3.99, + "learning_rate": 1.0527027878694212e-05, + "loss": 0.2211, + "step": 9528 + }, + { + "epoch": 3.99, + "learning_rate": 1.0525334818455855e-05, + "loss": 0.1998, + "step": 9529 + }, + { + "epoch": 3.99, + "learning_rate": 1.0523641743117215e-05, + "loss": 0.1809, + "step": 9530 + }, + { + "epoch": 3.99, + "learning_rate": 1.0521948652726955e-05, + "loss": 0.1798, + "step": 9531 + }, + { + "epoch": 3.99, + "learning_rate": 1.0520255547333744e-05, + "loss": 0.1796, + "step": 9532 + }, + { + "epoch": 3.99, + "learning_rate": 1.0518562426986253e-05, + "loss": 0.202, + "step": 9533 + }, + { + "epoch": 3.99, + "learning_rate": 1.0516869291733142e-05, + "loss": 0.1884, + "step": 9534 + }, + { + "epoch": 3.99, + "learning_rate": 1.0515176141623085e-05, + "loss": 0.1454, + "step": 9535 + }, + { + "epoch": 3.99, + "learning_rate": 1.0513482976704746e-05, + "loss": 0.1957, + "step": 9536 + }, + { + "epoch": 3.99, + "learning_rate": 1.0511789797026792e-05, + "loss": 0.2049, + "step": 9537 + }, + { + "epoch": 3.99, + "learning_rate": 1.0510096602637896e-05, + "loss": 0.179, + "step": 9538 + }, + { + "epoch": 3.99, + "learning_rate": 1.0508403393586728e-05, + "loss": 0.1709, + "step": 9539 + }, + { + "epoch": 3.99, + "learning_rate": 1.0506710169921954e-05, + "loss": 0.1893, + "step": 9540 + }, + { + "epoch": 3.99, + "learning_rate": 1.0505016931692244e-05, + "loss": 0.1522, + "step": 9541 + }, + { + "epoch": 3.99, + "learning_rate": 1.0503323678946272e-05, + "loss": 0.1592, + "step": 9542 + }, + { + "epoch": 4.0, + "learning_rate": 1.0501630411732709e-05, + "loss": 0.1703, + "step": 9543 + }, + { + "epoch": 4.0, + "learning_rate": 1.0499937130100222e-05, + "loss": 0.1883, + "step": 9544 + }, + { + "epoch": 4.0, + "learning_rate": 1.0498243834097485e-05, + "loss": 0.1816, + "step": 9545 + }, + { + "epoch": 4.0, + "learning_rate": 1.0496550523773177e-05, + "loss": 0.1701, + "step": 9546 + }, + { + "epoch": 4.0, + "learning_rate": 1.0494857199175961e-05, + "loss": 0.174, + "step": 9547 + }, + { + "epoch": 4.0, + "learning_rate": 1.0493163860354512e-05, + "loss": 0.1885, + "step": 9548 + }, + { + "epoch": 4.0, + "learning_rate": 1.0491470507357508e-05, + "loss": 0.2248, + "step": 9549 + }, + { + "epoch": 4.0, + "learning_rate": 1.0489777140233619e-05, + "loss": 0.1783, + "step": 9550 + }, + { + "epoch": 4.0, + "learning_rate": 1.0488083759031524e-05, + "loss": 0.1625, + "step": 9551 + }, + { + "epoch": 4.0, + "learning_rate": 1.0486390363799891e-05, + "loss": 0.2177, + "step": 9552 + }, + { + "epoch": 4.0, + "learning_rate": 1.04846969545874e-05, + "loss": 0.1887, + "step": 9553 + }, + { + "epoch": 4.0, + "learning_rate": 1.0483003531442726e-05, + "loss": 0.1784, + "step": 9554 + }, + { + "epoch": 4.0, + "learning_rate": 1.048131009441454e-05, + "loss": 0.1487, + "step": 9555 + }, + { + "epoch": 4.0, + "learning_rate": 1.047961664355153e-05, + "loss": 0.1338, + "step": 9556 + }, + { + "epoch": 4.0, + "learning_rate": 1.0477923178902359e-05, + "loss": 0.1734, + "step": 9557 + }, + { + "epoch": 4.0, + "learning_rate": 1.047622970051571e-05, + "loss": 0.133, + "step": 9558 + }, + { + "epoch": 4.0, + "learning_rate": 1.0474536208440263e-05, + "loss": 0.1673, + "step": 9559 + }, + { + "epoch": 4.0, + "learning_rate": 1.0472842702724695e-05, + "loss": 0.1659, + "step": 9560 + }, + { + "epoch": 4.0, + "learning_rate": 1.047114918341768e-05, + "loss": 0.1201, + "step": 9561 + }, + { + "epoch": 4.0, + "learning_rate": 1.0469455650567903e-05, + "loss": 0.1705, + "step": 9562 + }, + { + "epoch": 4.0, + "learning_rate": 1.0467762104224039e-05, + "loss": 0.164, + "step": 9563 + }, + { + "epoch": 4.0, + "learning_rate": 1.0466068544434766e-05, + "loss": 0.1443, + "step": 9564 + }, + { + "epoch": 4.0, + "learning_rate": 1.046437497124877e-05, + "loss": 0.1925, + "step": 9565 + }, + { + "epoch": 4.01, + "learning_rate": 1.0462681384714723e-05, + "loss": 0.1494, + "step": 9566 + }, + { + "epoch": 4.01, + "learning_rate": 1.0460987784881313e-05, + "loss": 0.1724, + "step": 9567 + }, + { + "epoch": 4.01, + "learning_rate": 1.0459294171797218e-05, + "loss": 0.1702, + "step": 9568 + }, + { + "epoch": 4.01, + "learning_rate": 1.045760054551112e-05, + "loss": 0.1659, + "step": 9569 + }, + { + "epoch": 4.01, + "learning_rate": 1.04559069060717e-05, + "loss": 0.1584, + "step": 9570 + }, + { + "epoch": 4.01, + "learning_rate": 1.045421325352764e-05, + "loss": 0.1523, + "step": 9571 + }, + { + "epoch": 4.01, + "learning_rate": 1.0452519587927623e-05, + "loss": 0.1254, + "step": 9572 + }, + { + "epoch": 4.01, + "learning_rate": 1.0450825909320335e-05, + "loss": 0.136, + "step": 9573 + }, + { + "epoch": 4.01, + "learning_rate": 1.0449132217754455e-05, + "loss": 0.1503, + "step": 9574 + }, + { + "epoch": 4.01, + "learning_rate": 1.0447438513278664e-05, + "loss": 0.1935, + "step": 9575 + }, + { + "epoch": 4.01, + "learning_rate": 1.0445744795941652e-05, + "loss": 0.146, + "step": 9576 + }, + { + "epoch": 4.01, + "learning_rate": 1.0444051065792105e-05, + "loss": 0.192, + "step": 9577 + }, + { + "epoch": 4.01, + "learning_rate": 1.04423573228787e-05, + "loss": 0.0897, + "step": 9578 + }, + { + "epoch": 4.01, + "learning_rate": 1.0440663567250127e-05, + "loss": 0.1551, + "step": 9579 + }, + { + "epoch": 4.01, + "learning_rate": 1.0438969798955072e-05, + "loss": 0.1595, + "step": 9580 + }, + { + "epoch": 4.01, + "learning_rate": 1.0437276018042219e-05, + "loss": 0.1983, + "step": 9581 + }, + { + "epoch": 4.01, + "learning_rate": 1.0435582224560255e-05, + "loss": 0.1687, + "step": 9582 + }, + { + "epoch": 4.01, + "learning_rate": 1.0433888418557867e-05, + "loss": 0.1532, + "step": 9583 + }, + { + "epoch": 4.01, + "learning_rate": 1.043219460008374e-05, + "loss": 0.1476, + "step": 9584 + }, + { + "epoch": 4.01, + "learning_rate": 1.0430500769186565e-05, + "loss": 0.1043, + "step": 9585 + }, + { + "epoch": 4.01, + "learning_rate": 1.0428806925915026e-05, + "loss": 0.1767, + "step": 9586 + }, + { + "epoch": 4.01, + "learning_rate": 1.0427113070317813e-05, + "loss": 0.1923, + "step": 9587 + }, + { + "epoch": 4.01, + "learning_rate": 1.0425419202443614e-05, + "loss": 0.1675, + "step": 9588 + }, + { + "epoch": 4.01, + "learning_rate": 1.0423725322341115e-05, + "loss": 0.1778, + "step": 9589 + }, + { + "epoch": 4.02, + "learning_rate": 1.0422031430059011e-05, + "loss": 0.1971, + "step": 9590 + }, + { + "epoch": 4.02, + "learning_rate": 1.0420337525645984e-05, + "loss": 0.1652, + "step": 9591 + }, + { + "epoch": 4.02, + "learning_rate": 1.0418643609150732e-05, + "loss": 0.1324, + "step": 9592 + }, + { + "epoch": 4.02, + "learning_rate": 1.041694968062194e-05, + "loss": 0.1142, + "step": 9593 + }, + { + "epoch": 4.02, + "learning_rate": 1.0415255740108299e-05, + "loss": 0.1709, + "step": 9594 + }, + { + "epoch": 4.02, + "learning_rate": 1.0413561787658497e-05, + "loss": 0.1278, + "step": 9595 + }, + { + "epoch": 4.02, + "learning_rate": 1.0411867823321233e-05, + "loss": 0.1172, + "step": 9596 + }, + { + "epoch": 4.02, + "learning_rate": 1.0410173847145195e-05, + "loss": 0.1617, + "step": 9597 + }, + { + "epoch": 4.02, + "learning_rate": 1.040847985917907e-05, + "loss": 0.1151, + "step": 9598 + }, + { + "epoch": 4.02, + "learning_rate": 1.0406785859471554e-05, + "loss": 0.1156, + "step": 9599 + }, + { + "epoch": 4.02, + "learning_rate": 1.0405091848071342e-05, + "loss": 0.1262, + "step": 9600 + }, + { + "epoch": 4.02, + "learning_rate": 1.0403397825027122e-05, + "loss": 0.157, + "step": 9601 + }, + { + "epoch": 4.02, + "learning_rate": 1.0401703790387589e-05, + "loss": 0.175, + "step": 9602 + }, + { + "epoch": 4.02, + "learning_rate": 1.0400009744201438e-05, + "loss": 0.1851, + "step": 9603 + }, + { + "epoch": 4.02, + "learning_rate": 1.0398315686517362e-05, + "loss": 0.1766, + "step": 9604 + }, + { + "epoch": 4.02, + "learning_rate": 1.0396621617384055e-05, + "loss": 0.1459, + "step": 9605 + }, + { + "epoch": 4.02, + "learning_rate": 1.039492753685021e-05, + "loss": 0.1324, + "step": 9606 + }, + { + "epoch": 4.02, + "learning_rate": 1.0393233444964528e-05, + "loss": 0.1526, + "step": 9607 + }, + { + "epoch": 4.02, + "learning_rate": 1.0391539341775696e-05, + "loss": 0.1222, + "step": 9608 + }, + { + "epoch": 4.02, + "learning_rate": 1.0389845227332413e-05, + "loss": 0.1661, + "step": 9609 + }, + { + "epoch": 4.02, + "learning_rate": 1.0388151101683375e-05, + "loss": 0.1224, + "step": 9610 + }, + { + "epoch": 4.02, + "learning_rate": 1.0386456964877281e-05, + "loss": 0.154, + "step": 9611 + }, + { + "epoch": 4.02, + "learning_rate": 1.038476281696282e-05, + "loss": 0.1473, + "step": 9612 + }, + { + "epoch": 4.02, + "learning_rate": 1.0383068657988695e-05, + "loss": 0.0998, + "step": 9613 + }, + { + "epoch": 4.03, + "learning_rate": 1.0381374488003603e-05, + "loss": 0.1953, + "step": 9614 + }, + { + "epoch": 4.03, + "learning_rate": 1.0379680307056238e-05, + "loss": 0.1348, + "step": 9615 + }, + { + "epoch": 4.03, + "learning_rate": 1.03779861151953e-05, + "loss": 0.1649, + "step": 9616 + }, + { + "epoch": 4.03, + "learning_rate": 1.0376291912469488e-05, + "loss": 0.1258, + "step": 9617 + }, + { + "epoch": 4.03, + "learning_rate": 1.0374597698927496e-05, + "loss": 0.1482, + "step": 9618 + }, + { + "epoch": 4.03, + "learning_rate": 1.0372903474618027e-05, + "loss": 0.1671, + "step": 9619 + }, + { + "epoch": 4.03, + "learning_rate": 1.0371209239589779e-05, + "loss": 0.1486, + "step": 9620 + }, + { + "epoch": 4.03, + "learning_rate": 1.0369514993891451e-05, + "loss": 0.1308, + "step": 9621 + }, + { + "epoch": 4.03, + "learning_rate": 1.036782073757174e-05, + "loss": 0.1495, + "step": 9622 + }, + { + "epoch": 4.03, + "learning_rate": 1.036612647067935e-05, + "loss": 0.0969, + "step": 9623 + }, + { + "epoch": 4.03, + "learning_rate": 1.036443219326298e-05, + "loss": 0.184, + "step": 9624 + }, + { + "epoch": 4.03, + "learning_rate": 1.036273790537133e-05, + "loss": 0.127, + "step": 9625 + }, + { + "epoch": 4.03, + "learning_rate": 1.0361043607053101e-05, + "loss": 0.1336, + "step": 9626 + }, + { + "epoch": 4.03, + "learning_rate": 1.0359349298356993e-05, + "loss": 0.1505, + "step": 9627 + }, + { + "epoch": 4.03, + "learning_rate": 1.0357654979331708e-05, + "loss": 0.1709, + "step": 9628 + }, + { + "epoch": 4.03, + "learning_rate": 1.0355960650025946e-05, + "loss": 0.1573, + "step": 9629 + }, + { + "epoch": 4.03, + "learning_rate": 1.0354266310488416e-05, + "loss": 0.2003, + "step": 9630 + }, + { + "epoch": 4.03, + "learning_rate": 1.0352571960767814e-05, + "loss": 0.1725, + "step": 9631 + }, + { + "epoch": 4.03, + "learning_rate": 1.0350877600912844e-05, + "loss": 0.1442, + "step": 9632 + }, + { + "epoch": 4.03, + "learning_rate": 1.0349183230972204e-05, + "loss": 0.1851, + "step": 9633 + }, + { + "epoch": 4.03, + "learning_rate": 1.0347488850994608e-05, + "loss": 0.1321, + "step": 9634 + }, + { + "epoch": 4.03, + "learning_rate": 1.034579446102875e-05, + "loss": 0.1416, + "step": 9635 + }, + { + "epoch": 4.03, + "learning_rate": 1.0344100061123337e-05, + "loss": 0.1843, + "step": 9636 + }, + { + "epoch": 4.03, + "learning_rate": 1.0342405651327072e-05, + "loss": 0.1499, + "step": 9637 + }, + { + "epoch": 4.04, + "learning_rate": 1.0340711231688662e-05, + "loss": 0.1437, + "step": 9638 + }, + { + "epoch": 4.04, + "learning_rate": 1.033901680225681e-05, + "loss": 0.1823, + "step": 9639 + }, + { + "epoch": 4.04, + "learning_rate": 1.0337322363080219e-05, + "loss": 0.1039, + "step": 9640 + }, + { + "epoch": 4.04, + "learning_rate": 1.0335627914207602e-05, + "loss": 0.1494, + "step": 9641 + }, + { + "epoch": 4.04, + "learning_rate": 1.0333933455687655e-05, + "loss": 0.166, + "step": 9642 + }, + { + "epoch": 4.04, + "learning_rate": 1.0332238987569086e-05, + "loss": 0.1793, + "step": 9643 + }, + { + "epoch": 4.04, + "learning_rate": 1.0330544509900603e-05, + "loss": 0.1315, + "step": 9644 + }, + { + "epoch": 4.04, + "learning_rate": 1.032885002273091e-05, + "loss": 0.1267, + "step": 9645 + }, + { + "epoch": 4.04, + "learning_rate": 1.0327155526108716e-05, + "loss": 0.1621, + "step": 9646 + }, + { + "epoch": 4.04, + "learning_rate": 1.0325461020082726e-05, + "loss": 0.1321, + "step": 9647 + }, + { + "epoch": 4.04, + "learning_rate": 1.0323766504701649e-05, + "loss": 0.1195, + "step": 9648 + }, + { + "epoch": 4.04, + "learning_rate": 1.032207198001419e-05, + "loss": 0.1324, + "step": 9649 + }, + { + "epoch": 4.04, + "learning_rate": 1.0320377446069059e-05, + "loss": 0.1025, + "step": 9650 + }, + { + "epoch": 4.04, + "learning_rate": 1.0318682902914962e-05, + "loss": 0.1751, + "step": 9651 + }, + { + "epoch": 4.04, + "learning_rate": 1.0316988350600608e-05, + "loss": 0.1372, + "step": 9652 + }, + { + "epoch": 4.04, + "learning_rate": 1.0315293789174705e-05, + "loss": 0.2112, + "step": 9653 + }, + { + "epoch": 4.04, + "learning_rate": 1.0313599218685961e-05, + "loss": 0.1442, + "step": 9654 + }, + { + "epoch": 4.04, + "learning_rate": 1.0311904639183088e-05, + "loss": 0.0953, + "step": 9655 + }, + { + "epoch": 4.04, + "learning_rate": 1.031021005071479e-05, + "loss": 0.1931, + "step": 9656 + }, + { + "epoch": 4.04, + "learning_rate": 1.030851545332978e-05, + "loss": 0.1386, + "step": 9657 + }, + { + "epoch": 4.04, + "learning_rate": 1.0306820847076768e-05, + "loss": 0.1469, + "step": 9658 + }, + { + "epoch": 4.04, + "learning_rate": 1.0305126232004464e-05, + "loss": 0.1834, + "step": 9659 + }, + { + "epoch": 4.04, + "learning_rate": 1.0303431608161574e-05, + "loss": 0.1344, + "step": 9660 + }, + { + "epoch": 4.04, + "learning_rate": 1.0301736975596816e-05, + "loss": 0.1186, + "step": 9661 + }, + { + "epoch": 4.05, + "learning_rate": 1.0300042334358893e-05, + "loss": 0.1086, + "step": 9662 + }, + { + "epoch": 4.05, + "learning_rate": 1.029834768449652e-05, + "loss": 0.1411, + "step": 9663 + }, + { + "epoch": 4.05, + "learning_rate": 1.029665302605841e-05, + "loss": 0.1346, + "step": 9664 + }, + { + "epoch": 4.05, + "learning_rate": 1.029495835909327e-05, + "loss": 0.1431, + "step": 9665 + }, + { + "epoch": 4.05, + "learning_rate": 1.0293263683649813e-05, + "loss": 0.1703, + "step": 9666 + }, + { + "epoch": 4.05, + "learning_rate": 1.0291568999776751e-05, + "loss": 0.164, + "step": 9667 + }, + { + "epoch": 4.05, + "learning_rate": 1.0289874307522801e-05, + "loss": 0.131, + "step": 9668 + }, + { + "epoch": 4.05, + "learning_rate": 1.0288179606936666e-05, + "loss": 0.1263, + "step": 9669 + }, + { + "epoch": 4.05, + "learning_rate": 1.0286484898067067e-05, + "loss": 0.1508, + "step": 9670 + }, + { + "epoch": 4.05, + "learning_rate": 1.0284790180962712e-05, + "loss": 0.1621, + "step": 9671 + }, + { + "epoch": 4.05, + "learning_rate": 1.0283095455672319e-05, + "loss": 0.0984, + "step": 9672 + }, + { + "epoch": 4.05, + "learning_rate": 1.0281400722244594e-05, + "loss": 0.1535, + "step": 9673 + }, + { + "epoch": 4.05, + "learning_rate": 1.0279705980728258e-05, + "loss": 0.1426, + "step": 9674 + }, + { + "epoch": 4.05, + "learning_rate": 1.0278011231172024e-05, + "loss": 0.1577, + "step": 9675 + }, + { + "epoch": 4.05, + "learning_rate": 1.02763164736246e-05, + "loss": 0.1846, + "step": 9676 + }, + { + "epoch": 4.05, + "learning_rate": 1.0274621708134705e-05, + "loss": 0.1882, + "step": 9677 + }, + { + "epoch": 4.05, + "learning_rate": 1.0272926934751053e-05, + "loss": 0.0994, + "step": 9678 + }, + { + "epoch": 4.05, + "learning_rate": 1.0271232153522359e-05, + "loss": 0.193, + "step": 9679 + }, + { + "epoch": 4.05, + "learning_rate": 1.0269537364497337e-05, + "loss": 0.1684, + "step": 9680 + }, + { + "epoch": 4.05, + "learning_rate": 1.0267842567724702e-05, + "loss": 0.1234, + "step": 9681 + }, + { + "epoch": 4.05, + "learning_rate": 1.0266147763253171e-05, + "loss": 0.1486, + "step": 9682 + }, + { + "epoch": 4.05, + "learning_rate": 1.0264452951131459e-05, + "loss": 0.1551, + "step": 9683 + }, + { + "epoch": 4.05, + "learning_rate": 1.026275813140828e-05, + "loss": 0.1232, + "step": 9684 + }, + { + "epoch": 4.05, + "learning_rate": 1.0261063304132353e-05, + "loss": 0.1866, + "step": 9685 + }, + { + "epoch": 4.06, + "learning_rate": 1.025936846935239e-05, + "loss": 0.1786, + "step": 9686 + }, + { + "epoch": 4.06, + "learning_rate": 1.0257673627117115e-05, + "loss": 0.1304, + "step": 9687 + }, + { + "epoch": 4.06, + "learning_rate": 1.0255978777475238e-05, + "loss": 0.1339, + "step": 9688 + }, + { + "epoch": 4.06, + "learning_rate": 1.025428392047548e-05, + "loss": 0.1277, + "step": 9689 + }, + { + "epoch": 4.06, + "learning_rate": 1.0252589056166554e-05, + "loss": 0.1696, + "step": 9690 + }, + { + "epoch": 4.06, + "learning_rate": 1.025089418459718e-05, + "loss": 0.1633, + "step": 9691 + }, + { + "epoch": 4.06, + "learning_rate": 1.0249199305816078e-05, + "loss": 0.179, + "step": 9692 + }, + { + "epoch": 4.06, + "learning_rate": 1.024750441987196e-05, + "loss": 0.1495, + "step": 9693 + }, + { + "epoch": 4.06, + "learning_rate": 1.0245809526813547e-05, + "loss": 0.1376, + "step": 9694 + }, + { + "epoch": 4.06, + "learning_rate": 1.0244114626689558e-05, + "loss": 0.1755, + "step": 9695 + }, + { + "epoch": 4.06, + "learning_rate": 1.0242419719548711e-05, + "loss": 0.0943, + "step": 9696 + }, + { + "epoch": 4.06, + "learning_rate": 1.0240724805439724e-05, + "loss": 0.1556, + "step": 9697 + }, + { + "epoch": 4.06, + "learning_rate": 1.0239029884411318e-05, + "loss": 0.1504, + "step": 9698 + }, + { + "epoch": 4.06, + "learning_rate": 1.0237334956512209e-05, + "loss": 0.1953, + "step": 9699 + }, + { + "epoch": 4.06, + "learning_rate": 1.0235640021791115e-05, + "loss": 0.1541, + "step": 9700 + }, + { + "epoch": 4.06, + "learning_rate": 1.0233945080296758e-05, + "loss": 0.1627, + "step": 9701 + }, + { + "epoch": 4.06, + "learning_rate": 1.0232250132077861e-05, + "loss": 0.1356, + "step": 9702 + }, + { + "epoch": 4.06, + "learning_rate": 1.0230555177183136e-05, + "loss": 0.1696, + "step": 9703 + }, + { + "epoch": 4.06, + "learning_rate": 1.0228860215661309e-05, + "loss": 0.1229, + "step": 9704 + }, + { + "epoch": 4.06, + "learning_rate": 1.02271652475611e-05, + "loss": 0.1677, + "step": 9705 + }, + { + "epoch": 4.06, + "learning_rate": 1.0225470272931224e-05, + "loss": 0.0818, + "step": 9706 + }, + { + "epoch": 4.06, + "learning_rate": 1.0223775291820406e-05, + "loss": 0.163, + "step": 9707 + }, + { + "epoch": 4.06, + "learning_rate": 1.0222080304277366e-05, + "loss": 0.1689, + "step": 9708 + }, + { + "epoch": 4.06, + "learning_rate": 1.0220385310350829e-05, + "loss": 0.1454, + "step": 9709 + }, + { + "epoch": 4.07, + "learning_rate": 1.0218690310089508e-05, + "loss": 0.1118, + "step": 9710 + }, + { + "epoch": 4.07, + "learning_rate": 1.0216995303542126e-05, + "loss": 0.1471, + "step": 9711 + }, + { + "epoch": 4.07, + "learning_rate": 1.021530029075741e-05, + "loss": 0.1493, + "step": 9712 + }, + { + "epoch": 4.07, + "learning_rate": 1.0213605271784076e-05, + "loss": 0.156, + "step": 9713 + }, + { + "epoch": 4.07, + "learning_rate": 1.0211910246670848e-05, + "loss": 0.1622, + "step": 9714 + }, + { + "epoch": 4.07, + "learning_rate": 1.0210215215466447e-05, + "loss": 0.1617, + "step": 9715 + }, + { + "epoch": 4.07, + "learning_rate": 1.0208520178219602e-05, + "loss": 0.1603, + "step": 9716 + }, + { + "epoch": 4.07, + "learning_rate": 1.0206825134979024e-05, + "loss": 0.1312, + "step": 9717 + }, + { + "epoch": 4.07, + "learning_rate": 1.020513008579344e-05, + "loss": 0.1757, + "step": 9718 + }, + { + "epoch": 4.07, + "learning_rate": 1.0203435030711577e-05, + "loss": 0.1677, + "step": 9719 + }, + { + "epoch": 4.07, + "learning_rate": 1.0201739969782154e-05, + "loss": 0.1854, + "step": 9720 + }, + { + "epoch": 4.07, + "learning_rate": 1.0200044903053896e-05, + "loss": 0.1221, + "step": 9721 + }, + { + "epoch": 4.07, + "learning_rate": 1.0198349830575522e-05, + "loss": 0.1636, + "step": 9722 + }, + { + "epoch": 4.07, + "learning_rate": 1.0196654752395761e-05, + "loss": 0.1262, + "step": 9723 + }, + { + "epoch": 4.07, + "learning_rate": 1.0194959668563332e-05, + "loss": 0.1673, + "step": 9724 + }, + { + "epoch": 4.07, + "learning_rate": 1.0193264579126957e-05, + "loss": 0.1389, + "step": 9725 + }, + { + "epoch": 4.07, + "learning_rate": 1.0191569484135368e-05, + "loss": 0.1372, + "step": 9726 + }, + { + "epoch": 4.07, + "learning_rate": 1.0189874383637284e-05, + "loss": 0.149, + "step": 9727 + }, + { + "epoch": 4.07, + "learning_rate": 1.0188179277681426e-05, + "loss": 0.1391, + "step": 9728 + }, + { + "epoch": 4.07, + "learning_rate": 1.0186484166316526e-05, + "loss": 0.1838, + "step": 9729 + }, + { + "epoch": 4.07, + "learning_rate": 1.01847890495913e-05, + "loss": 0.1652, + "step": 9730 + }, + { + "epoch": 4.07, + "learning_rate": 1.0183093927554478e-05, + "loss": 0.1445, + "step": 9731 + }, + { + "epoch": 4.07, + "learning_rate": 1.0181398800254785e-05, + "loss": 0.1057, + "step": 9732 + }, + { + "epoch": 4.07, + "learning_rate": 1.0179703667740946e-05, + "loss": 0.1414, + "step": 9733 + }, + { + "epoch": 4.08, + "learning_rate": 1.017800853006168e-05, + "loss": 0.1351, + "step": 9734 + }, + { + "epoch": 4.08, + "learning_rate": 1.0176313387265717e-05, + "loss": 0.1709, + "step": 9735 + }, + { + "epoch": 4.08, + "learning_rate": 1.0174618239401786e-05, + "loss": 0.1435, + "step": 9736 + }, + { + "epoch": 4.08, + "learning_rate": 1.0172923086518607e-05, + "loss": 0.1688, + "step": 9737 + }, + { + "epoch": 4.08, + "learning_rate": 1.0171227928664905e-05, + "loss": 0.1496, + "step": 9738 + }, + { + "epoch": 4.08, + "learning_rate": 1.016953276588941e-05, + "loss": 0.1329, + "step": 9739 + }, + { + "epoch": 4.08, + "learning_rate": 1.0167837598240847e-05, + "loss": 0.1073, + "step": 9740 + }, + { + "epoch": 4.08, + "learning_rate": 1.016614242576794e-05, + "loss": 0.106, + "step": 9741 + }, + { + "epoch": 4.08, + "learning_rate": 1.0164447248519415e-05, + "loss": 0.1252, + "step": 9742 + }, + { + "epoch": 4.08, + "learning_rate": 1.0162752066544007e-05, + "loss": 0.1417, + "step": 9743 + }, + { + "epoch": 4.08, + "learning_rate": 1.0161056879890427e-05, + "loss": 0.1441, + "step": 9744 + }, + { + "epoch": 4.08, + "learning_rate": 1.0159361688607411e-05, + "loss": 0.1247, + "step": 9745 + }, + { + "epoch": 4.08, + "learning_rate": 1.0157666492743687e-05, + "loss": 0.2127, + "step": 9746 + }, + { + "epoch": 4.08, + "learning_rate": 1.0155971292347978e-05, + "loss": 0.1592, + "step": 9747 + }, + { + "epoch": 4.08, + "learning_rate": 1.0154276087469015e-05, + "loss": 0.0748, + "step": 9748 + }, + { + "epoch": 4.08, + "learning_rate": 1.015258087815552e-05, + "loss": 0.1466, + "step": 9749 + }, + { + "epoch": 4.08, + "learning_rate": 1.0150885664456224e-05, + "loss": 0.1454, + "step": 9750 + }, + { + "epoch": 4.08, + "learning_rate": 1.0149190446419854e-05, + "loss": 0.1237, + "step": 9751 + }, + { + "epoch": 4.08, + "learning_rate": 1.0147495224095136e-05, + "loss": 0.1437, + "step": 9752 + }, + { + "epoch": 4.08, + "learning_rate": 1.0145799997530801e-05, + "loss": 0.1919, + "step": 9753 + }, + { + "epoch": 4.08, + "learning_rate": 1.0144104766775574e-05, + "loss": 0.1484, + "step": 9754 + }, + { + "epoch": 4.08, + "learning_rate": 1.014240953187818e-05, + "loss": 0.0954, + "step": 9755 + }, + { + "epoch": 4.08, + "learning_rate": 1.0140714292887355e-05, + "loss": 0.1143, + "step": 9756 + }, + { + "epoch": 4.08, + "learning_rate": 1.0139019049851818e-05, + "loss": 0.1436, + "step": 9757 + }, + { + "epoch": 4.09, + "learning_rate": 1.0137323802820304e-05, + "loss": 0.1065, + "step": 9758 + }, + { + "epoch": 4.09, + "learning_rate": 1.0135628551841538e-05, + "loss": 0.1562, + "step": 9759 + }, + { + "epoch": 4.09, + "learning_rate": 1.0133933296964254e-05, + "loss": 0.1523, + "step": 9760 + }, + { + "epoch": 4.09, + "learning_rate": 1.0132238038237173e-05, + "loss": 0.1694, + "step": 9761 + }, + { + "epoch": 4.09, + "learning_rate": 1.0130542775709025e-05, + "loss": 0.1982, + "step": 9762 + }, + { + "epoch": 4.09, + "learning_rate": 1.0128847509428543e-05, + "loss": 0.1591, + "step": 9763 + }, + { + "epoch": 4.09, + "learning_rate": 1.0127152239444453e-05, + "loss": 0.1483, + "step": 9764 + }, + { + "epoch": 4.09, + "learning_rate": 1.0125456965805485e-05, + "loss": 0.1438, + "step": 9765 + }, + { + "epoch": 4.09, + "learning_rate": 1.012376168856037e-05, + "loss": 0.1265, + "step": 9766 + }, + { + "epoch": 4.09, + "learning_rate": 1.0122066407757833e-05, + "loss": 0.1673, + "step": 9767 + }, + { + "epoch": 4.09, + "learning_rate": 1.0120371123446603e-05, + "loss": 0.1653, + "step": 9768 + }, + { + "epoch": 4.09, + "learning_rate": 1.0118675835675416e-05, + "loss": 0.1294, + "step": 9769 + }, + { + "epoch": 4.09, + "learning_rate": 1.0116980544492996e-05, + "loss": 0.1593, + "step": 9770 + }, + { + "epoch": 4.09, + "learning_rate": 1.0115285249948073e-05, + "loss": 0.1596, + "step": 9771 + }, + { + "epoch": 4.09, + "learning_rate": 1.0113589952089377e-05, + "loss": 0.1486, + "step": 9772 + }, + { + "epoch": 4.09, + "learning_rate": 1.0111894650965644e-05, + "loss": 0.2042, + "step": 9773 + }, + { + "epoch": 4.09, + "learning_rate": 1.0110199346625593e-05, + "loss": 0.1378, + "step": 9774 + }, + { + "epoch": 4.09, + "learning_rate": 1.010850403911796e-05, + "loss": 0.1603, + "step": 9775 + }, + { + "epoch": 4.09, + "learning_rate": 1.0106808728491477e-05, + "loss": 0.1564, + "step": 9776 + }, + { + "epoch": 4.09, + "learning_rate": 1.0105113414794872e-05, + "loss": 0.1183, + "step": 9777 + }, + { + "epoch": 4.09, + "learning_rate": 1.0103418098076874e-05, + "loss": 0.1432, + "step": 9778 + }, + { + "epoch": 4.09, + "learning_rate": 1.0101722778386214e-05, + "loss": 0.1461, + "step": 9779 + }, + { + "epoch": 4.09, + "learning_rate": 1.0100027455771624e-05, + "loss": 0.1482, + "step": 9780 + }, + { + "epoch": 4.1, + "learning_rate": 1.0098332130281831e-05, + "loss": 0.156, + "step": 9781 + }, + { + "epoch": 4.1, + "learning_rate": 1.0096636801965571e-05, + "loss": 0.1696, + "step": 9782 + }, + { + "epoch": 4.1, + "learning_rate": 1.0094941470871571e-05, + "loss": 0.1552, + "step": 9783 + }, + { + "epoch": 4.1, + "learning_rate": 1.0093246137048563e-05, + "loss": 0.1347, + "step": 9784 + }, + { + "epoch": 4.1, + "learning_rate": 1.0091550800545279e-05, + "loss": 0.1917, + "step": 9785 + }, + { + "epoch": 4.1, + "learning_rate": 1.0089855461410446e-05, + "loss": 0.135, + "step": 9786 + }, + { + "epoch": 4.1, + "learning_rate": 1.0088160119692798e-05, + "loss": 0.1596, + "step": 9787 + }, + { + "epoch": 4.1, + "learning_rate": 1.0086464775441064e-05, + "loss": 0.126, + "step": 9788 + }, + { + "epoch": 4.1, + "learning_rate": 1.008476942870398e-05, + "loss": 0.1321, + "step": 9789 + }, + { + "epoch": 4.1, + "learning_rate": 1.0083074079530272e-05, + "loss": 0.1722, + "step": 9790 + }, + { + "epoch": 4.1, + "learning_rate": 1.0081378727968676e-05, + "loss": 0.1506, + "step": 9791 + }, + { + "epoch": 4.1, + "learning_rate": 1.007968337406792e-05, + "loss": 0.1526, + "step": 9792 + }, + { + "epoch": 4.1, + "learning_rate": 1.0077988017876732e-05, + "loss": 0.147, + "step": 9793 + }, + { + "epoch": 4.1, + "learning_rate": 1.0076292659443852e-05, + "loss": 0.1657, + "step": 9794 + }, + { + "epoch": 4.1, + "learning_rate": 1.0074597298818005e-05, + "loss": 0.1669, + "step": 9795 + }, + { + "epoch": 4.1, + "learning_rate": 1.0072901936047923e-05, + "loss": 0.1401, + "step": 9796 + }, + { + "epoch": 4.1, + "learning_rate": 1.0071206571182344e-05, + "loss": 0.1403, + "step": 9797 + }, + { + "epoch": 4.1, + "learning_rate": 1.0069511204269992e-05, + "loss": 0.1263, + "step": 9798 + }, + { + "epoch": 4.1, + "learning_rate": 1.0067815835359605e-05, + "loss": 0.1474, + "step": 9799 + }, + { + "epoch": 4.1, + "learning_rate": 1.0066120464499912e-05, + "loss": 0.1316, + "step": 9800 + }, + { + "epoch": 4.1, + "learning_rate": 1.0064425091739643e-05, + "loss": 0.156, + "step": 9801 + }, + { + "epoch": 4.1, + "learning_rate": 1.0062729717127531e-05, + "loss": 0.1534, + "step": 9802 + }, + { + "epoch": 4.1, + "learning_rate": 1.006103434071231e-05, + "loss": 0.1773, + "step": 9803 + }, + { + "epoch": 4.1, + "learning_rate": 1.0059338962542713e-05, + "loss": 0.1738, + "step": 9804 + }, + { + "epoch": 4.11, + "learning_rate": 1.0057643582667467e-05, + "loss": 0.1704, + "step": 9805 + }, + { + "epoch": 4.11, + "learning_rate": 1.0055948201135308e-05, + "loss": 0.1433, + "step": 9806 + }, + { + "epoch": 4.11, + "learning_rate": 1.0054252817994969e-05, + "loss": 0.1176, + "step": 9807 + }, + { + "epoch": 4.11, + "learning_rate": 1.0052557433295181e-05, + "loss": 0.1655, + "step": 9808 + }, + { + "epoch": 4.11, + "learning_rate": 1.0050862047084673e-05, + "loss": 0.1327, + "step": 9809 + }, + { + "epoch": 4.11, + "learning_rate": 1.0049166659412185e-05, + "loss": 0.1708, + "step": 9810 + }, + { + "epoch": 4.11, + "learning_rate": 1.0047471270326446e-05, + "loss": 0.1619, + "step": 9811 + }, + { + "epoch": 4.11, + "learning_rate": 1.0045775879876183e-05, + "loss": 0.1612, + "step": 9812 + }, + { + "epoch": 4.11, + "learning_rate": 1.0044080488110133e-05, + "loss": 0.1654, + "step": 9813 + }, + { + "epoch": 4.11, + "learning_rate": 1.0042385095077032e-05, + "loss": 0.1458, + "step": 9814 + }, + { + "epoch": 4.11, + "learning_rate": 1.0040689700825606e-05, + "loss": 0.1526, + "step": 9815 + }, + { + "epoch": 4.11, + "learning_rate": 1.003899430540459e-05, + "loss": 0.18, + "step": 9816 + }, + { + "epoch": 4.11, + "learning_rate": 1.0037298908862721e-05, + "loss": 0.1752, + "step": 9817 + }, + { + "epoch": 4.11, + "learning_rate": 1.0035603511248723e-05, + "loss": 0.1416, + "step": 9818 + }, + { + "epoch": 4.11, + "learning_rate": 1.0033908112611337e-05, + "loss": 0.141, + "step": 9819 + }, + { + "epoch": 4.11, + "learning_rate": 1.0032212712999292e-05, + "loss": 0.1333, + "step": 9820 + }, + { + "epoch": 4.11, + "learning_rate": 1.0030517312461323e-05, + "loss": 0.1397, + "step": 9821 + }, + { + "epoch": 4.11, + "learning_rate": 1.0028821911046158e-05, + "loss": 0.1191, + "step": 9822 + }, + { + "epoch": 4.11, + "learning_rate": 1.0027126508802534e-05, + "loss": 0.1196, + "step": 9823 + }, + { + "epoch": 4.11, + "learning_rate": 1.0025431105779184e-05, + "loss": 0.1096, + "step": 9824 + }, + { + "epoch": 4.11, + "learning_rate": 1.0023735702024838e-05, + "loss": 0.1517, + "step": 9825 + }, + { + "epoch": 4.11, + "learning_rate": 1.0022040297588229e-05, + "loss": 0.147, + "step": 9826 + }, + { + "epoch": 4.11, + "learning_rate": 1.0020344892518093e-05, + "loss": 0.1235, + "step": 9827 + }, + { + "epoch": 4.11, + "learning_rate": 1.001864948686316e-05, + "loss": 0.1635, + "step": 9828 + }, + { + "epoch": 4.12, + "learning_rate": 1.0016954080672163e-05, + "loss": 0.1218, + "step": 9829 + }, + { + "epoch": 4.12, + "learning_rate": 1.0015258673993837e-05, + "loss": 0.1867, + "step": 9830 + }, + { + "epoch": 4.12, + "learning_rate": 1.0013563266876916e-05, + "loss": 0.1583, + "step": 9831 + }, + { + "epoch": 4.12, + "learning_rate": 1.001186785937013e-05, + "loss": 0.1475, + "step": 9832 + }, + { + "epoch": 4.12, + "learning_rate": 1.0010172451522211e-05, + "loss": 0.1283, + "step": 9833 + }, + { + "epoch": 4.12, + "learning_rate": 1.0008477043381896e-05, + "loss": 0.1581, + "step": 9834 + }, + { + "epoch": 4.12, + "learning_rate": 1.0006781634997917e-05, + "loss": 0.1534, + "step": 9835 + }, + { + "epoch": 4.12, + "learning_rate": 1.0005086226419003e-05, + "loss": 0.127, + "step": 9836 + }, + { + "epoch": 4.12, + "learning_rate": 1.000339081769389e-05, + "loss": 0.1332, + "step": 9837 + }, + { + "epoch": 4.12, + "learning_rate": 1.0001695408871312e-05, + "loss": 0.142, + "step": 9838 + }, + { + "epoch": 4.12, + "learning_rate": 1e-05, + "loss": 0.1187, + "step": 9839 + }, + { + "epoch": 4.12, + "learning_rate": 9.99830459112869e-06, + "loss": 0.1665, + "step": 9840 + }, + { + "epoch": 4.12, + "learning_rate": 9.996609182306113e-06, + "loss": 0.1124, + "step": 9841 + }, + { + "epoch": 4.12, + "learning_rate": 9.994913773581e-06, + "loss": 0.1864, + "step": 9842 + }, + { + "epoch": 4.12, + "learning_rate": 9.993218365002086e-06, + "loss": 0.153, + "step": 9843 + }, + { + "epoch": 4.12, + "learning_rate": 9.991522956618107e-06, + "loss": 0.1748, + "step": 9844 + }, + { + "epoch": 4.12, + "learning_rate": 9.98982754847779e-06, + "loss": 0.1491, + "step": 9845 + }, + { + "epoch": 4.12, + "learning_rate": 9.988132140629871e-06, + "loss": 0.1691, + "step": 9846 + }, + { + "epoch": 4.12, + "learning_rate": 9.986436733123086e-06, + "loss": 0.1863, + "step": 9847 + }, + { + "epoch": 4.12, + "learning_rate": 9.984741326006163e-06, + "loss": 0.1493, + "step": 9848 + }, + { + "epoch": 4.12, + "learning_rate": 9.983045919327837e-06, + "loss": 0.1843, + "step": 9849 + }, + { + "epoch": 4.12, + "learning_rate": 9.981350513136841e-06, + "loss": 0.1709, + "step": 9850 + }, + { + "epoch": 4.12, + "learning_rate": 9.979655107481912e-06, + "loss": 0.1936, + "step": 9851 + }, + { + "epoch": 4.12, + "learning_rate": 9.977959702411776e-06, + "loss": 0.1036, + "step": 9852 + }, + { + "epoch": 4.13, + "learning_rate": 9.976264297975167e-06, + "loss": 0.1825, + "step": 9853 + }, + { + "epoch": 4.13, + "learning_rate": 9.974568894220821e-06, + "loss": 0.1521, + "step": 9854 + }, + { + "epoch": 4.13, + "learning_rate": 9.97287349119747e-06, + "loss": 0.1488, + "step": 9855 + }, + { + "epoch": 4.13, + "learning_rate": 9.971178088953845e-06, + "loss": 0.1464, + "step": 9856 + }, + { + "epoch": 4.13, + "learning_rate": 9.96948268753868e-06, + "loss": 0.1594, + "step": 9857 + }, + { + "epoch": 4.13, + "learning_rate": 9.967787287000711e-06, + "loss": 0.1306, + "step": 9858 + }, + { + "epoch": 4.13, + "learning_rate": 9.966091887388665e-06, + "loss": 0.1487, + "step": 9859 + }, + { + "epoch": 4.13, + "learning_rate": 9.964396488751278e-06, + "loss": 0.1637, + "step": 9860 + }, + { + "epoch": 4.13, + "learning_rate": 9.962701091137282e-06, + "loss": 0.1266, + "step": 9861 + }, + { + "epoch": 4.13, + "learning_rate": 9.961005694595413e-06, + "loss": 0.1867, + "step": 9862 + }, + { + "epoch": 4.13, + "learning_rate": 9.959310299174397e-06, + "loss": 0.1562, + "step": 9863 + }, + { + "epoch": 4.13, + "learning_rate": 9.957614904922971e-06, + "loss": 0.1554, + "step": 9864 + }, + { + "epoch": 4.13, + "learning_rate": 9.95591951188987e-06, + "loss": 0.1215, + "step": 9865 + }, + { + "epoch": 4.13, + "learning_rate": 9.954224120123822e-06, + "loss": 0.1551, + "step": 9866 + }, + { + "epoch": 4.13, + "learning_rate": 9.95252872967356e-06, + "loss": 0.1345, + "step": 9867 + }, + { + "epoch": 4.13, + "learning_rate": 9.950833340587818e-06, + "loss": 0.2001, + "step": 9868 + }, + { + "epoch": 4.13, + "learning_rate": 9.949137952915327e-06, + "loss": 0.1969, + "step": 9869 + }, + { + "epoch": 4.13, + "learning_rate": 9.94744256670482e-06, + "loss": 0.1802, + "step": 9870 + }, + { + "epoch": 4.13, + "learning_rate": 9.945747182005033e-06, + "loss": 0.1551, + "step": 9871 + }, + { + "epoch": 4.13, + "learning_rate": 9.944051798864692e-06, + "loss": 0.1237, + "step": 9872 + }, + { + "epoch": 4.13, + "learning_rate": 9.942356417332538e-06, + "loss": 0.1763, + "step": 9873 + }, + { + "epoch": 4.13, + "learning_rate": 9.940661037457294e-06, + "loss": 0.1686, + "step": 9874 + }, + { + "epoch": 4.13, + "learning_rate": 9.938965659287693e-06, + "loss": 0.1286, + "step": 9875 + }, + { + "epoch": 4.13, + "learning_rate": 9.937270282872474e-06, + "loss": 0.1735, + "step": 9876 + }, + { + "epoch": 4.14, + "learning_rate": 9.935574908260362e-06, + "loss": 0.128, + "step": 9877 + }, + { + "epoch": 4.14, + "learning_rate": 9.933879535500093e-06, + "loss": 0.1503, + "step": 9878 + }, + { + "epoch": 4.14, + "learning_rate": 9.932184164640399e-06, + "loss": 0.1645, + "step": 9879 + }, + { + "epoch": 4.14, + "learning_rate": 9.93048879573001e-06, + "loss": 0.1779, + "step": 9880 + }, + { + "epoch": 4.14, + "learning_rate": 9.928793428817658e-06, + "loss": 0.1424, + "step": 9881 + }, + { + "epoch": 4.14, + "learning_rate": 9.927098063952078e-06, + "loss": 0.1156, + "step": 9882 + }, + { + "epoch": 4.14, + "learning_rate": 9.925402701181999e-06, + "loss": 0.1213, + "step": 9883 + }, + { + "epoch": 4.14, + "learning_rate": 9.923707340556151e-06, + "loss": 0.1282, + "step": 9884 + }, + { + "epoch": 4.14, + "learning_rate": 9.92201198212327e-06, + "loss": 0.1577, + "step": 9885 + }, + { + "epoch": 4.14, + "learning_rate": 9.920316625932085e-06, + "loss": 0.1339, + "step": 9886 + }, + { + "epoch": 4.14, + "learning_rate": 9.918621272031327e-06, + "loss": 0.1632, + "step": 9887 + }, + { + "epoch": 4.14, + "learning_rate": 9.91692592046973e-06, + "loss": 0.1693, + "step": 9888 + }, + { + "epoch": 4.14, + "learning_rate": 9.915230571296021e-06, + "loss": 0.1508, + "step": 9889 + }, + { + "epoch": 4.14, + "learning_rate": 9.913535224558936e-06, + "loss": 0.1149, + "step": 9890 + }, + { + "epoch": 4.14, + "learning_rate": 9.911839880307202e-06, + "loss": 0.1464, + "step": 9891 + }, + { + "epoch": 4.14, + "learning_rate": 9.910144538589556e-06, + "loss": 0.1719, + "step": 9892 + }, + { + "epoch": 4.14, + "learning_rate": 9.908449199454723e-06, + "loss": 0.1044, + "step": 9893 + }, + { + "epoch": 4.14, + "learning_rate": 9.906753862951437e-06, + "loss": 0.1697, + "step": 9894 + }, + { + "epoch": 4.14, + "learning_rate": 9.90505852912843e-06, + "loss": 0.173, + "step": 9895 + }, + { + "epoch": 4.14, + "learning_rate": 9.903363198034434e-06, + "loss": 0.158, + "step": 9896 + }, + { + "epoch": 4.14, + "learning_rate": 9.901667869718172e-06, + "loss": 0.1206, + "step": 9897 + }, + { + "epoch": 4.14, + "learning_rate": 9.89997254422838e-06, + "loss": 0.1652, + "step": 9898 + }, + { + "epoch": 4.14, + "learning_rate": 9.89827722161379e-06, + "loss": 0.1206, + "step": 9899 + }, + { + "epoch": 4.14, + "learning_rate": 9.896581901923131e-06, + "loss": 0.1813, + "step": 9900 + }, + { + "epoch": 4.15, + "learning_rate": 9.894886585205132e-06, + "loss": 0.1526, + "step": 9901 + }, + { + "epoch": 4.15, + "learning_rate": 9.893191271508526e-06, + "loss": 0.1389, + "step": 9902 + }, + { + "epoch": 4.15, + "learning_rate": 9.891495960882042e-06, + "loss": 0.1889, + "step": 9903 + }, + { + "epoch": 4.15, + "learning_rate": 9.88980065337441e-06, + "loss": 0.1551, + "step": 9904 + }, + { + "epoch": 4.15, + "learning_rate": 9.88810534903436e-06, + "loss": 0.1207, + "step": 9905 + }, + { + "epoch": 4.15, + "learning_rate": 9.886410047910624e-06, + "loss": 0.1304, + "step": 9906 + }, + { + "epoch": 4.15, + "learning_rate": 9.884714750051929e-06, + "loss": 0.1537, + "step": 9907 + }, + { + "epoch": 4.15, + "learning_rate": 9.883019455507008e-06, + "loss": 0.1206, + "step": 9908 + }, + { + "epoch": 4.15, + "learning_rate": 9.881324164324587e-06, + "loss": 0.14, + "step": 9909 + }, + { + "epoch": 4.15, + "learning_rate": 9.879628876553398e-06, + "loss": 0.1357, + "step": 9910 + }, + { + "epoch": 4.15, + "learning_rate": 9.87793359224217e-06, + "loss": 0.1855, + "step": 9911 + }, + { + "epoch": 4.15, + "learning_rate": 9.876238311439635e-06, + "loss": 0.1721, + "step": 9912 + }, + { + "epoch": 4.15, + "learning_rate": 9.874543034194516e-06, + "loss": 0.172, + "step": 9913 + }, + { + "epoch": 4.15, + "learning_rate": 9.872847760555547e-06, + "loss": 0.1699, + "step": 9914 + }, + { + "epoch": 4.15, + "learning_rate": 9.871152490571459e-06, + "loss": 0.1347, + "step": 9915 + }, + { + "epoch": 4.15, + "learning_rate": 9.869457224290975e-06, + "loss": 0.1038, + "step": 9916 + }, + { + "epoch": 4.15, + "learning_rate": 9.867761961762829e-06, + "loss": 0.1226, + "step": 9917 + }, + { + "epoch": 4.15, + "learning_rate": 9.86606670303575e-06, + "loss": 0.1654, + "step": 9918 + }, + { + "epoch": 4.15, + "learning_rate": 9.864371448158463e-06, + "loss": 0.1471, + "step": 9919 + }, + { + "epoch": 4.15, + "learning_rate": 9.8626761971797e-06, + "loss": 0.1598, + "step": 9920 + }, + { + "epoch": 4.15, + "learning_rate": 9.860980950148185e-06, + "loss": 0.185, + "step": 9921 + }, + { + "epoch": 4.15, + "learning_rate": 9.859285707112648e-06, + "loss": 0.1481, + "step": 9922 + }, + { + "epoch": 4.15, + "learning_rate": 9.857590468121823e-06, + "loss": 0.1243, + "step": 9923 + }, + { + "epoch": 4.15, + "learning_rate": 9.855895233224431e-06, + "loss": 0.1544, + "step": 9924 + }, + { + "epoch": 4.16, + "learning_rate": 9.854200002469202e-06, + "loss": 0.094, + "step": 9925 + }, + { + "epoch": 4.16, + "learning_rate": 9.852504775904867e-06, + "loss": 0.1216, + "step": 9926 + }, + { + "epoch": 4.16, + "learning_rate": 9.850809553580148e-06, + "loss": 0.1627, + "step": 9927 + }, + { + "epoch": 4.16, + "learning_rate": 9.849114335543778e-06, + "loss": 0.1341, + "step": 9928 + }, + { + "epoch": 4.16, + "learning_rate": 9.847419121844484e-06, + "loss": 0.1723, + "step": 9929 + }, + { + "epoch": 4.16, + "learning_rate": 9.845723912530989e-06, + "loss": 0.1964, + "step": 9930 + }, + { + "epoch": 4.16, + "learning_rate": 9.844028707652023e-06, + "loss": 0.1807, + "step": 9931 + }, + { + "epoch": 4.16, + "learning_rate": 9.842333507256315e-06, + "loss": 0.19, + "step": 9932 + }, + { + "epoch": 4.16, + "learning_rate": 9.84063831139259e-06, + "loss": 0.1478, + "step": 9933 + }, + { + "epoch": 4.16, + "learning_rate": 9.838943120109576e-06, + "loss": 0.1816, + "step": 9934 + }, + { + "epoch": 4.16, + "learning_rate": 9.837247933455998e-06, + "loss": 0.1445, + "step": 9935 + }, + { + "epoch": 4.16, + "learning_rate": 9.835552751480587e-06, + "loss": 0.1875, + "step": 9936 + }, + { + "epoch": 4.16, + "learning_rate": 9.833857574232061e-06, + "loss": 0.1343, + "step": 9937 + }, + { + "epoch": 4.16, + "learning_rate": 9.832162401759155e-06, + "loss": 0.1191, + "step": 9938 + }, + { + "epoch": 4.16, + "learning_rate": 9.83046723411059e-06, + "loss": 0.1396, + "step": 9939 + }, + { + "epoch": 4.16, + "learning_rate": 9.828772071335095e-06, + "loss": 0.148, + "step": 9940 + }, + { + "epoch": 4.16, + "learning_rate": 9.8270769134814e-06, + "loss": 0.1541, + "step": 9941 + }, + { + "epoch": 4.16, + "learning_rate": 9.82538176059822e-06, + "loss": 0.1429, + "step": 9942 + }, + { + "epoch": 4.16, + "learning_rate": 9.823686612734285e-06, + "loss": 0.1397, + "step": 9943 + }, + { + "epoch": 4.16, + "learning_rate": 9.821991469938325e-06, + "loss": 0.1751, + "step": 9944 + }, + { + "epoch": 4.16, + "learning_rate": 9.82029633225906e-06, + "loss": 0.1865, + "step": 9945 + }, + { + "epoch": 4.16, + "learning_rate": 9.818601199745218e-06, + "loss": 0.101, + "step": 9946 + }, + { + "epoch": 4.16, + "learning_rate": 9.816906072445526e-06, + "loss": 0.1478, + "step": 9947 + }, + { + "epoch": 4.16, + "learning_rate": 9.815210950408703e-06, + "loss": 0.1287, + "step": 9948 + }, + { + "epoch": 4.17, + "learning_rate": 9.813515833683477e-06, + "loss": 0.168, + "step": 9949 + }, + { + "epoch": 4.17, + "learning_rate": 9.811820722318577e-06, + "loss": 0.1253, + "step": 9950 + }, + { + "epoch": 4.17, + "learning_rate": 9.81012561636272e-06, + "loss": 0.1594, + "step": 9951 + }, + { + "epoch": 4.17, + "learning_rate": 9.808430515864633e-06, + "loss": 0.1339, + "step": 9952 + }, + { + "epoch": 4.17, + "learning_rate": 9.806735420873044e-06, + "loss": 0.1805, + "step": 9953 + }, + { + "epoch": 4.17, + "learning_rate": 9.805040331436672e-06, + "loss": 0.1188, + "step": 9954 + }, + { + "epoch": 4.17, + "learning_rate": 9.803345247604242e-06, + "loss": 0.1549, + "step": 9955 + }, + { + "epoch": 4.17, + "learning_rate": 9.80165016942448e-06, + "loss": 0.1802, + "step": 9956 + }, + { + "epoch": 4.17, + "learning_rate": 9.799955096946107e-06, + "loss": 0.1507, + "step": 9957 + }, + { + "epoch": 4.17, + "learning_rate": 9.798260030217846e-06, + "loss": 0.112, + "step": 9958 + }, + { + "epoch": 4.17, + "learning_rate": 9.796564969288425e-06, + "loss": 0.1529, + "step": 9959 + }, + { + "epoch": 4.17, + "learning_rate": 9.79486991420656e-06, + "loss": 0.1983, + "step": 9960 + }, + { + "epoch": 4.17, + "learning_rate": 9.793174865020978e-06, + "loss": 0.124, + "step": 9961 + }, + { + "epoch": 4.17, + "learning_rate": 9.7914798217804e-06, + "loss": 0.1237, + "step": 9962 + }, + { + "epoch": 4.17, + "learning_rate": 9.789784784533553e-06, + "loss": 0.1132, + "step": 9963 + }, + { + "epoch": 4.17, + "learning_rate": 9.788089753329157e-06, + "loss": 0.0795, + "step": 9964 + }, + { + "epoch": 4.17, + "learning_rate": 9.78639472821593e-06, + "loss": 0.1575, + "step": 9965 + }, + { + "epoch": 4.17, + "learning_rate": 9.784699709242595e-06, + "loss": 0.1362, + "step": 9966 + }, + { + "epoch": 4.17, + "learning_rate": 9.78300469645788e-06, + "loss": 0.1231, + "step": 9967 + }, + { + "epoch": 4.17, + "learning_rate": 9.7813096899105e-06, + "loss": 0.1493, + "step": 9968 + }, + { + "epoch": 4.17, + "learning_rate": 9.779614689649176e-06, + "loss": 0.1198, + "step": 9969 + }, + { + "epoch": 4.17, + "learning_rate": 9.777919695722638e-06, + "loss": 0.1976, + "step": 9970 + }, + { + "epoch": 4.17, + "learning_rate": 9.776224708179597e-06, + "loss": 0.1473, + "step": 9971 + }, + { + "epoch": 4.18, + "learning_rate": 9.774529727068778e-06, + "loss": 0.1737, + "step": 9972 + }, + { + "epoch": 4.18, + "learning_rate": 9.772834752438903e-06, + "loss": 0.1593, + "step": 9973 + }, + { + "epoch": 4.18, + "learning_rate": 9.771139784338693e-06, + "loss": 0.1489, + "step": 9974 + }, + { + "epoch": 4.18, + "learning_rate": 9.769444822816865e-06, + "loss": 0.174, + "step": 9975 + }, + { + "epoch": 4.18, + "learning_rate": 9.76774986792214e-06, + "loss": 0.1863, + "step": 9976 + }, + { + "epoch": 4.18, + "learning_rate": 9.766054919703243e-06, + "loss": 0.1683, + "step": 9977 + }, + { + "epoch": 4.18, + "learning_rate": 9.764359978208886e-06, + "loss": 0.1814, + "step": 9978 + }, + { + "epoch": 4.18, + "learning_rate": 9.762665043487793e-06, + "loss": 0.1657, + "step": 9979 + }, + { + "epoch": 4.18, + "learning_rate": 9.760970115588686e-06, + "loss": 0.1244, + "step": 9980 + }, + { + "epoch": 4.18, + "learning_rate": 9.759275194560278e-06, + "loss": 0.1577, + "step": 9981 + }, + { + "epoch": 4.18, + "learning_rate": 9.757580280451289e-06, + "loss": 0.1613, + "step": 9982 + }, + { + "epoch": 4.18, + "learning_rate": 9.755885373310444e-06, + "loss": 0.1368, + "step": 9983 + }, + { + "epoch": 4.18, + "learning_rate": 9.754190473186455e-06, + "loss": 0.1572, + "step": 9984 + }, + { + "epoch": 4.18, + "learning_rate": 9.752495580128042e-06, + "loss": 0.1111, + "step": 9985 + }, + { + "epoch": 4.18, + "learning_rate": 9.750800694183926e-06, + "loss": 0.1778, + "step": 9986 + }, + { + "epoch": 4.18, + "learning_rate": 9.749105815402823e-06, + "loss": 0.1719, + "step": 9987 + }, + { + "epoch": 4.18, + "learning_rate": 9.747410943833451e-06, + "loss": 0.1589, + "step": 9988 + }, + { + "epoch": 4.18, + "learning_rate": 9.745716079524526e-06, + "loss": 0.1702, + "step": 9989 + }, + { + "epoch": 4.18, + "learning_rate": 9.744021222524764e-06, + "loss": 0.1895, + "step": 9990 + }, + { + "epoch": 4.18, + "learning_rate": 9.74232637288289e-06, + "loss": 0.143, + "step": 9991 + }, + { + "epoch": 4.18, + "learning_rate": 9.740631530647611e-06, + "loss": 0.0912, + "step": 9992 + }, + { + "epoch": 4.18, + "learning_rate": 9.73893669586765e-06, + "loss": 0.1575, + "step": 9993 + }, + { + "epoch": 4.18, + "learning_rate": 9.737241868591724e-06, + "loss": 0.1241, + "step": 9994 + }, + { + "epoch": 4.18, + "learning_rate": 9.735547048868545e-06, + "loss": 0.1368, + "step": 9995 + }, + { + "epoch": 4.19, + "learning_rate": 9.733852236746832e-06, + "loss": 0.1415, + "step": 9996 + }, + { + "epoch": 4.19, + "learning_rate": 9.732157432275301e-06, + "loss": 0.1183, + "step": 9997 + }, + { + "epoch": 4.19, + "learning_rate": 9.730462635502666e-06, + "loss": 0.1642, + "step": 9998 + }, + { + "epoch": 4.19, + "learning_rate": 9.728767846477643e-06, + "loss": 0.153, + "step": 9999 + }, + { + "epoch": 4.19, + "learning_rate": 9.72707306524895e-06, + "loss": 0.1753, + "step": 10000 + }, + { + "epoch": 4.19, + "learning_rate": 9.725378291865297e-06, + "loss": 0.1548, + "step": 10001 + }, + { + "epoch": 4.19, + "learning_rate": 9.723683526375403e-06, + "loss": 0.1527, + "step": 10002 + }, + { + "epoch": 4.19, + "learning_rate": 9.721988768827977e-06, + "loss": 0.1509, + "step": 10003 + }, + { + "epoch": 4.19, + "learning_rate": 9.720294019271744e-06, + "loss": 0.1601, + "step": 10004 + }, + { + "epoch": 4.19, + "learning_rate": 9.718599277755406e-06, + "loss": 0.1228, + "step": 10005 + }, + { + "epoch": 4.19, + "learning_rate": 9.716904544327683e-06, + "loss": 0.1562, + "step": 10006 + }, + { + "epoch": 4.19, + "learning_rate": 9.71520981903729e-06, + "loss": 0.1742, + "step": 10007 + }, + { + "epoch": 4.19, + "learning_rate": 9.713515101932935e-06, + "loss": 0.1533, + "step": 10008 + }, + { + "epoch": 4.19, + "learning_rate": 9.711820393063337e-06, + "loss": 0.1423, + "step": 10009 + }, + { + "epoch": 4.19, + "learning_rate": 9.710125692477204e-06, + "loss": 0.1037, + "step": 10010 + }, + { + "epoch": 4.19, + "learning_rate": 9.708431000223252e-06, + "loss": 0.18, + "step": 10011 + }, + { + "epoch": 4.19, + "learning_rate": 9.706736316350192e-06, + "loss": 0.1598, + "step": 10012 + }, + { + "epoch": 4.19, + "learning_rate": 9.705041640906736e-06, + "loss": 0.1305, + "step": 10013 + }, + { + "epoch": 4.19, + "learning_rate": 9.703346973941595e-06, + "loss": 0.1635, + "step": 10014 + }, + { + "epoch": 4.19, + "learning_rate": 9.701652315503484e-06, + "loss": 0.1586, + "step": 10015 + }, + { + "epoch": 4.19, + "learning_rate": 9.69995766564111e-06, + "loss": 0.1495, + "step": 10016 + }, + { + "epoch": 4.19, + "learning_rate": 9.698263024403187e-06, + "loss": 0.1283, + "step": 10017 + }, + { + "epoch": 4.19, + "learning_rate": 9.696568391838428e-06, + "loss": 0.1545, + "step": 10018 + }, + { + "epoch": 4.19, + "learning_rate": 9.69487376799554e-06, + "loss": 0.1423, + "step": 10019 + }, + { + "epoch": 4.2, + "learning_rate": 9.693179152923233e-06, + "loss": 0.1542, + "step": 10020 + }, + { + "epoch": 4.2, + "learning_rate": 9.691484546670222e-06, + "loss": 0.1238, + "step": 10021 + }, + { + "epoch": 4.2, + "learning_rate": 9.689789949285213e-06, + "loss": 0.1384, + "step": 10022 + }, + { + "epoch": 4.2, + "learning_rate": 9.688095360816916e-06, + "loss": 0.1603, + "step": 10023 + }, + { + "epoch": 4.2, + "learning_rate": 9.686400781314042e-06, + "loss": 0.1407, + "step": 10024 + }, + { + "epoch": 4.2, + "learning_rate": 9.684706210825298e-06, + "loss": 0.156, + "step": 10025 + }, + { + "epoch": 4.2, + "learning_rate": 9.683011649399393e-06, + "loss": 0.1402, + "step": 10026 + }, + { + "epoch": 4.2, + "learning_rate": 9.681317097085041e-06, + "loss": 0.1731, + "step": 10027 + }, + { + "epoch": 4.2, + "learning_rate": 9.679622553930943e-06, + "loss": 0.1694, + "step": 10028 + }, + { + "epoch": 4.2, + "learning_rate": 9.677928019985811e-06, + "loss": 0.179, + "step": 10029 + }, + { + "epoch": 4.2, + "learning_rate": 9.676233495298353e-06, + "loss": 0.1417, + "step": 10030 + }, + { + "epoch": 4.2, + "learning_rate": 9.674538979917274e-06, + "loss": 0.1471, + "step": 10031 + }, + { + "epoch": 4.2, + "learning_rate": 9.67284447389129e-06, + "loss": 0.127, + "step": 10032 + }, + { + "epoch": 4.2, + "learning_rate": 9.671149977269095e-06, + "loss": 0.1081, + "step": 10033 + }, + { + "epoch": 4.2, + "learning_rate": 9.669455490099402e-06, + "loss": 0.1672, + "step": 10034 + }, + { + "epoch": 4.2, + "learning_rate": 9.667761012430919e-06, + "loss": 0.2019, + "step": 10035 + }, + { + "epoch": 4.2, + "learning_rate": 9.66606654431235e-06, + "loss": 0.144, + "step": 10036 + }, + { + "epoch": 4.2, + "learning_rate": 9.664372085792401e-06, + "loss": 0.1272, + "step": 10037 + }, + { + "epoch": 4.2, + "learning_rate": 9.662677636919783e-06, + "loss": 0.1226, + "step": 10038 + }, + { + "epoch": 4.2, + "learning_rate": 9.660983197743192e-06, + "loss": 0.1758, + "step": 10039 + }, + { + "epoch": 4.2, + "learning_rate": 9.65928876831134e-06, + "loss": 0.1523, + "step": 10040 + }, + { + "epoch": 4.2, + "learning_rate": 9.657594348672931e-06, + "loss": 0.1713, + "step": 10041 + }, + { + "epoch": 4.2, + "learning_rate": 9.655899938876668e-06, + "loss": 0.1666, + "step": 10042 + }, + { + "epoch": 4.2, + "learning_rate": 9.654205538971254e-06, + "loss": 0.1977, + "step": 10043 + }, + { + "epoch": 4.21, + "learning_rate": 9.652511149005396e-06, + "loss": 0.1409, + "step": 10044 + }, + { + "epoch": 4.21, + "learning_rate": 9.650816769027797e-06, + "loss": 0.1365, + "step": 10045 + }, + { + "epoch": 4.21, + "learning_rate": 9.649122399087162e-06, + "loss": 0.1512, + "step": 10046 + }, + { + "epoch": 4.21, + "learning_rate": 9.647428039232188e-06, + "loss": 0.1452, + "step": 10047 + }, + { + "epoch": 4.21, + "learning_rate": 9.645733689511585e-06, + "loss": 0.148, + "step": 10048 + }, + { + "epoch": 4.21, + "learning_rate": 9.644039349974054e-06, + "loss": 0.1768, + "step": 10049 + }, + { + "epoch": 4.21, + "learning_rate": 9.642345020668294e-06, + "loss": 0.1461, + "step": 10050 + }, + { + "epoch": 4.21, + "learning_rate": 9.64065070164301e-06, + "loss": 0.1729, + "step": 10051 + }, + { + "epoch": 4.21, + "learning_rate": 9.6389563929469e-06, + "loss": 0.1695, + "step": 10052 + }, + { + "epoch": 4.21, + "learning_rate": 9.63726209462867e-06, + "loss": 0.1668, + "step": 10053 + }, + { + "epoch": 4.21, + "learning_rate": 9.635567806737021e-06, + "loss": 0.1861, + "step": 10054 + }, + { + "epoch": 4.21, + "learning_rate": 9.633873529320652e-06, + "loss": 0.1445, + "step": 10055 + }, + { + "epoch": 4.21, + "learning_rate": 9.632179262428264e-06, + "loss": 0.1459, + "step": 10056 + }, + { + "epoch": 4.21, + "learning_rate": 9.630485006108554e-06, + "loss": 0.2078, + "step": 10057 + }, + { + "epoch": 4.21, + "learning_rate": 9.628790760410224e-06, + "loss": 0.2002, + "step": 10058 + }, + { + "epoch": 4.21, + "learning_rate": 9.627096525381978e-06, + "loss": 0.1857, + "step": 10059 + }, + { + "epoch": 4.21, + "learning_rate": 9.625402301072508e-06, + "loss": 0.1554, + "step": 10060 + }, + { + "epoch": 4.21, + "learning_rate": 9.623708087530515e-06, + "loss": 0.1844, + "step": 10061 + }, + { + "epoch": 4.21, + "learning_rate": 9.622013884804704e-06, + "loss": 0.1169, + "step": 10062 + }, + { + "epoch": 4.21, + "learning_rate": 9.620319692943765e-06, + "loss": 0.1747, + "step": 10063 + }, + { + "epoch": 4.21, + "learning_rate": 9.618625511996398e-06, + "loss": 0.1651, + "step": 10064 + }, + { + "epoch": 4.21, + "learning_rate": 9.616931342011307e-06, + "loss": 0.1641, + "step": 10065 + }, + { + "epoch": 4.21, + "learning_rate": 9.615237183037181e-06, + "loss": 0.1532, + "step": 10066 + }, + { + "epoch": 4.21, + "learning_rate": 9.613543035122722e-06, + "loss": 0.1131, + "step": 10067 + }, + { + "epoch": 4.22, + "learning_rate": 9.611848898316626e-06, + "loss": 0.1774, + "step": 10068 + }, + { + "epoch": 4.22, + "learning_rate": 9.610154772667589e-06, + "loss": 0.1478, + "step": 10069 + }, + { + "epoch": 4.22, + "learning_rate": 9.608460658224306e-06, + "loss": 0.1161, + "step": 10070 + }, + { + "epoch": 4.22, + "learning_rate": 9.606766555035475e-06, + "loss": 0.1271, + "step": 10071 + }, + { + "epoch": 4.22, + "learning_rate": 9.60507246314979e-06, + "loss": 0.1846, + "step": 10072 + }, + { + "epoch": 4.22, + "learning_rate": 9.603378382615946e-06, + "loss": 0.0966, + "step": 10073 + }, + { + "epoch": 4.22, + "learning_rate": 9.601684313482638e-06, + "loss": 0.1844, + "step": 10074 + }, + { + "epoch": 4.22, + "learning_rate": 9.599990255798564e-06, + "loss": 0.1447, + "step": 10075 + }, + { + "epoch": 4.22, + "learning_rate": 9.598296209612413e-06, + "loss": 0.1451, + "step": 10076 + }, + { + "epoch": 4.22, + "learning_rate": 9.596602174972883e-06, + "loss": 0.1175, + "step": 10077 + }, + { + "epoch": 4.22, + "learning_rate": 9.594908151928663e-06, + "loss": 0.1086, + "step": 10078 + }, + { + "epoch": 4.22, + "learning_rate": 9.593214140528451e-06, + "loss": 0.1838, + "step": 10079 + }, + { + "epoch": 4.22, + "learning_rate": 9.591520140820936e-06, + "loss": 0.1618, + "step": 10080 + }, + { + "epoch": 4.22, + "learning_rate": 9.58982615285481e-06, + "loss": 0.1676, + "step": 10081 + }, + { + "epoch": 4.22, + "learning_rate": 9.588132176678772e-06, + "loss": 0.1582, + "step": 10082 + }, + { + "epoch": 4.22, + "learning_rate": 9.586438212341505e-06, + "loss": 0.1614, + "step": 10083 + }, + { + "epoch": 4.22, + "learning_rate": 9.584744259891706e-06, + "loss": 0.1732, + "step": 10084 + }, + { + "epoch": 4.22, + "learning_rate": 9.583050319378063e-06, + "loss": 0.1284, + "step": 10085 + }, + { + "epoch": 4.22, + "learning_rate": 9.581356390849272e-06, + "loss": 0.1317, + "step": 10086 + }, + { + "epoch": 4.22, + "learning_rate": 9.579662474354017e-06, + "loss": 0.1301, + "step": 10087 + }, + { + "epoch": 4.22, + "learning_rate": 9.577968569940992e-06, + "loss": 0.1623, + "step": 10088 + }, + { + "epoch": 4.22, + "learning_rate": 9.576274677658888e-06, + "loss": 0.1329, + "step": 10089 + }, + { + "epoch": 4.22, + "learning_rate": 9.574580797556389e-06, + "loss": 0.162, + "step": 10090 + }, + { + "epoch": 4.22, + "learning_rate": 9.57288692968219e-06, + "loss": 0.1418, + "step": 10091 + }, + { + "epoch": 4.23, + "learning_rate": 9.571193074084976e-06, + "loss": 0.1598, + "step": 10092 + }, + { + "epoch": 4.23, + "learning_rate": 9.569499230813437e-06, + "loss": 0.14, + "step": 10093 + }, + { + "epoch": 4.23, + "learning_rate": 9.56780539991626e-06, + "loss": 0.1781, + "step": 10094 + }, + { + "epoch": 4.23, + "learning_rate": 9.566111581442136e-06, + "loss": 0.0888, + "step": 10095 + }, + { + "epoch": 4.23, + "learning_rate": 9.564417775439747e-06, + "loss": 0.1739, + "step": 10096 + }, + { + "epoch": 4.23, + "learning_rate": 9.562723981957781e-06, + "loss": 0.1608, + "step": 10097 + }, + { + "epoch": 4.23, + "learning_rate": 9.56103020104493e-06, + "loss": 0.1476, + "step": 10098 + }, + { + "epoch": 4.23, + "learning_rate": 9.559336432749873e-06, + "loss": 0.1451, + "step": 10099 + }, + { + "epoch": 4.23, + "learning_rate": 9.557642677121306e-06, + "loss": 0.1617, + "step": 10100 + }, + { + "epoch": 4.23, + "learning_rate": 9.5559489342079e-06, + "loss": 0.2092, + "step": 10101 + }, + { + "epoch": 4.23, + "learning_rate": 9.55425520405835e-06, + "loss": 0.1794, + "step": 10102 + }, + { + "epoch": 4.23, + "learning_rate": 9.55256148672134e-06, + "loss": 0.1469, + "step": 10103 + }, + { + "epoch": 4.23, + "learning_rate": 9.55086778224555e-06, + "loss": 0.1448, + "step": 10104 + }, + { + "epoch": 4.23, + "learning_rate": 9.54917409067967e-06, + "loss": 0.1527, + "step": 10105 + }, + { + "epoch": 4.23, + "learning_rate": 9.547480412072379e-06, + "loss": 0.1258, + "step": 10106 + }, + { + "epoch": 4.23, + "learning_rate": 9.545786746472363e-06, + "loss": 0.1697, + "step": 10107 + }, + { + "epoch": 4.23, + "learning_rate": 9.544093093928302e-06, + "loss": 0.1573, + "step": 10108 + }, + { + "epoch": 4.23, + "learning_rate": 9.542399454488883e-06, + "loss": 0.161, + "step": 10109 + }, + { + "epoch": 4.23, + "learning_rate": 9.540705828202783e-06, + "loss": 0.1356, + "step": 10110 + }, + { + "epoch": 4.23, + "learning_rate": 9.539012215118688e-06, + "loss": 0.1737, + "step": 10111 + }, + { + "epoch": 4.23, + "learning_rate": 9.537318615285279e-06, + "loss": 0.1357, + "step": 10112 + }, + { + "epoch": 4.23, + "learning_rate": 9.535625028751235e-06, + "loss": 0.1329, + "step": 10113 + }, + { + "epoch": 4.23, + "learning_rate": 9.533931455565236e-06, + "loss": 0.1777, + "step": 10114 + }, + { + "epoch": 4.23, + "learning_rate": 9.532237895775963e-06, + "loss": 0.1467, + "step": 10115 + }, + { + "epoch": 4.24, + "learning_rate": 9.5305443494321e-06, + "loss": 0.1216, + "step": 10116 + }, + { + "epoch": 4.24, + "learning_rate": 9.52885081658232e-06, + "loss": 0.1411, + "step": 10117 + }, + { + "epoch": 4.24, + "learning_rate": 9.527157297275306e-06, + "loss": 0.1181, + "step": 10118 + }, + { + "epoch": 4.24, + "learning_rate": 9.525463791559737e-06, + "loss": 0.1709, + "step": 10119 + }, + { + "epoch": 4.24, + "learning_rate": 9.52377029948429e-06, + "loss": 0.1502, + "step": 10120 + }, + { + "epoch": 4.24, + "learning_rate": 9.522076821097643e-06, + "loss": 0.1715, + "step": 10121 + }, + { + "epoch": 4.24, + "learning_rate": 9.520383356448476e-06, + "loss": 0.1614, + "step": 10122 + }, + { + "epoch": 4.24, + "learning_rate": 9.518689905585463e-06, + "loss": 0.1279, + "step": 10123 + }, + { + "epoch": 4.24, + "learning_rate": 9.51699646855728e-06, + "loss": 0.1557, + "step": 10124 + }, + { + "epoch": 4.24, + "learning_rate": 9.515303045412604e-06, + "loss": 0.1346, + "step": 10125 + }, + { + "epoch": 4.24, + "learning_rate": 9.513609636200112e-06, + "loss": 0.1711, + "step": 10126 + }, + { + "epoch": 4.24, + "learning_rate": 9.511916240968481e-06, + "loss": 0.1474, + "step": 10127 + }, + { + "epoch": 4.24, + "learning_rate": 9.510222859766383e-06, + "loss": 0.1593, + "step": 10128 + }, + { + "epoch": 4.24, + "learning_rate": 9.508529492642493e-06, + "loss": 0.1478, + "step": 10129 + }, + { + "epoch": 4.24, + "learning_rate": 9.506836139645491e-06, + "loss": 0.175, + "step": 10130 + }, + { + "epoch": 4.24, + "learning_rate": 9.505142800824044e-06, + "loss": 0.1369, + "step": 10131 + }, + { + "epoch": 4.24, + "learning_rate": 9.503449476226826e-06, + "loss": 0.1794, + "step": 10132 + }, + { + "epoch": 4.24, + "learning_rate": 9.501756165902517e-06, + "loss": 0.1383, + "step": 10133 + }, + { + "epoch": 4.24, + "learning_rate": 9.50006286989978e-06, + "loss": 0.1675, + "step": 10134 + }, + { + "epoch": 4.24, + "learning_rate": 9.498369588267295e-06, + "loss": 0.1657, + "step": 10135 + }, + { + "epoch": 4.24, + "learning_rate": 9.49667632105373e-06, + "loss": 0.1763, + "step": 10136 + }, + { + "epoch": 4.24, + "learning_rate": 9.494983068307759e-06, + "loss": 0.1615, + "step": 10137 + }, + { + "epoch": 4.24, + "learning_rate": 9.493289830078049e-06, + "loss": 0.1786, + "step": 10138 + }, + { + "epoch": 4.24, + "learning_rate": 9.491596606413275e-06, + "loss": 0.1827, + "step": 10139 + }, + { + "epoch": 4.25, + "learning_rate": 9.489903397362104e-06, + "loss": 0.146, + "step": 10140 + }, + { + "epoch": 4.25, + "learning_rate": 9.488210202973208e-06, + "loss": 0.1449, + "step": 10141 + }, + { + "epoch": 4.25, + "learning_rate": 9.486517023295257e-06, + "loss": 0.1465, + "step": 10142 + }, + { + "epoch": 4.25, + "learning_rate": 9.484823858376916e-06, + "loss": 0.1521, + "step": 10143 + }, + { + "epoch": 4.25, + "learning_rate": 9.483130708266856e-06, + "loss": 0.1594, + "step": 10144 + }, + { + "epoch": 4.25, + "learning_rate": 9.48143757301375e-06, + "loss": 0.1279, + "step": 10145 + }, + { + "epoch": 4.25, + "learning_rate": 9.479744452666257e-06, + "loss": 0.2213, + "step": 10146 + }, + { + "epoch": 4.25, + "learning_rate": 9.47805134727305e-06, + "loss": 0.133, + "step": 10147 + }, + { + "epoch": 4.25, + "learning_rate": 9.476358256882792e-06, + "loss": 0.1305, + "step": 10148 + }, + { + "epoch": 4.25, + "learning_rate": 9.47466518154415e-06, + "loss": 0.1897, + "step": 10149 + }, + { + "epoch": 4.25, + "learning_rate": 9.472972121305795e-06, + "loss": 0.1382, + "step": 10150 + }, + { + "epoch": 4.25, + "learning_rate": 9.471279076216385e-06, + "loss": 0.1311, + "step": 10151 + }, + { + "epoch": 4.25, + "learning_rate": 9.46958604632459e-06, + "loss": 0.1497, + "step": 10152 + }, + { + "epoch": 4.25, + "learning_rate": 9.467893031679073e-06, + "loss": 0.1205, + "step": 10153 + }, + { + "epoch": 4.25, + "learning_rate": 9.466200032328499e-06, + "loss": 0.162, + "step": 10154 + }, + { + "epoch": 4.25, + "learning_rate": 9.46450704832153e-06, + "loss": 0.1396, + "step": 10155 + }, + { + "epoch": 4.25, + "learning_rate": 9.462814079706832e-06, + "loss": 0.1599, + "step": 10156 + }, + { + "epoch": 4.25, + "learning_rate": 9.461121126533069e-06, + "loss": 0.1311, + "step": 10157 + }, + { + "epoch": 4.25, + "learning_rate": 9.459428188848897e-06, + "loss": 0.1662, + "step": 10158 + }, + { + "epoch": 4.25, + "learning_rate": 9.457735266702984e-06, + "loss": 0.1174, + "step": 10159 + }, + { + "epoch": 4.25, + "learning_rate": 9.45604236014399e-06, + "loss": 0.1673, + "step": 10160 + }, + { + "epoch": 4.25, + "learning_rate": 9.454349469220573e-06, + "loss": 0.1501, + "step": 10161 + }, + { + "epoch": 4.25, + "learning_rate": 9.452656593981398e-06, + "loss": 0.1671, + "step": 10162 + }, + { + "epoch": 4.25, + "learning_rate": 9.450963734475125e-06, + "loss": 0.1354, + "step": 10163 + }, + { + "epoch": 4.26, + "learning_rate": 9.44927089075041e-06, + "loss": 0.2105, + "step": 10164 + }, + { + "epoch": 4.26, + "learning_rate": 9.447578062855915e-06, + "loss": 0.1904, + "step": 10165 + }, + { + "epoch": 4.26, + "learning_rate": 9.445885250840301e-06, + "loss": 0.1371, + "step": 10166 + }, + { + "epoch": 4.26, + "learning_rate": 9.444192454752221e-06, + "loss": 0.1447, + "step": 10167 + }, + { + "epoch": 4.26, + "learning_rate": 9.44249967464034e-06, + "loss": 0.1443, + "step": 10168 + }, + { + "epoch": 4.26, + "learning_rate": 9.440806910553308e-06, + "loss": 0.1216, + "step": 10169 + }, + { + "epoch": 4.26, + "learning_rate": 9.439114162539784e-06, + "loss": 0.138, + "step": 10170 + }, + { + "epoch": 4.26, + "learning_rate": 9.437421430648426e-06, + "loss": 0.1481, + "step": 10171 + }, + { + "epoch": 4.26, + "learning_rate": 9.43572871492789e-06, + "loss": 0.152, + "step": 10172 + }, + { + "epoch": 4.26, + "learning_rate": 9.434036015426831e-06, + "loss": 0.1472, + "step": 10173 + }, + { + "epoch": 4.26, + "learning_rate": 9.432343332193907e-06, + "loss": 0.1496, + "step": 10174 + }, + { + "epoch": 4.26, + "learning_rate": 9.430650665277767e-06, + "loss": 0.1479, + "step": 10175 + }, + { + "epoch": 4.26, + "learning_rate": 9.428958014727068e-06, + "loss": 0.1076, + "step": 10176 + }, + { + "epoch": 4.26, + "learning_rate": 9.427265380590466e-06, + "loss": 0.1564, + "step": 10177 + }, + { + "epoch": 4.26, + "learning_rate": 9.425572762916613e-06, + "loss": 0.1509, + "step": 10178 + }, + { + "epoch": 4.26, + "learning_rate": 9.423880161754158e-06, + "loss": 0.1453, + "step": 10179 + }, + { + "epoch": 4.26, + "learning_rate": 9.42218757715176e-06, + "loss": 0.1317, + "step": 10180 + }, + { + "epoch": 4.26, + "learning_rate": 9.420495009158066e-06, + "loss": 0.1405, + "step": 10181 + }, + { + "epoch": 4.26, + "learning_rate": 9.418802457821729e-06, + "loss": 0.1627, + "step": 10182 + }, + { + "epoch": 4.26, + "learning_rate": 9.4171099231914e-06, + "loss": 0.1302, + "step": 10183 + }, + { + "epoch": 4.26, + "learning_rate": 9.415417405315727e-06, + "loss": 0.1721, + "step": 10184 + }, + { + "epoch": 4.26, + "learning_rate": 9.413724904243363e-06, + "loss": 0.164, + "step": 10185 + }, + { + "epoch": 4.26, + "learning_rate": 9.412032420022956e-06, + "loss": 0.1351, + "step": 10186 + }, + { + "epoch": 4.27, + "learning_rate": 9.410339952703157e-06, + "loss": 0.116, + "step": 10187 + }, + { + "epoch": 4.27, + "learning_rate": 9.408647502332612e-06, + "loss": 0.1532, + "step": 10188 + }, + { + "epoch": 4.27, + "learning_rate": 9.40695506895997e-06, + "loss": 0.1497, + "step": 10189 + }, + { + "epoch": 4.27, + "learning_rate": 9.405262652633879e-06, + "loss": 0.1906, + "step": 10190 + }, + { + "epoch": 4.27, + "learning_rate": 9.403570253402987e-06, + "loss": 0.1771, + "step": 10191 + }, + { + "epoch": 4.27, + "learning_rate": 9.401877871315936e-06, + "loss": 0.1526, + "step": 10192 + }, + { + "epoch": 4.27, + "learning_rate": 9.400185506421373e-06, + "loss": 0.1485, + "step": 10193 + }, + { + "epoch": 4.27, + "learning_rate": 9.39849315876795e-06, + "loss": 0.0871, + "step": 10194 + }, + { + "epoch": 4.27, + "learning_rate": 9.396800828404304e-06, + "loss": 0.1601, + "step": 10195 + }, + { + "epoch": 4.27, + "learning_rate": 9.395108515379082e-06, + "loss": 0.1222, + "step": 10196 + }, + { + "epoch": 4.27, + "learning_rate": 9.393416219740929e-06, + "loss": 0.1923, + "step": 10197 + }, + { + "epoch": 4.27, + "learning_rate": 9.391723941538492e-06, + "loss": 0.1477, + "step": 10198 + }, + { + "epoch": 4.27, + "learning_rate": 9.390031680820409e-06, + "loss": 0.1737, + "step": 10199 + }, + { + "epoch": 4.27, + "learning_rate": 9.388339437635321e-06, + "loss": 0.1635, + "step": 10200 + }, + { + "epoch": 4.27, + "learning_rate": 9.386647212031877e-06, + "loss": 0.1653, + "step": 10201 + }, + { + "epoch": 4.27, + "learning_rate": 9.384955004058714e-06, + "loss": 0.1863, + "step": 10202 + }, + { + "epoch": 4.27, + "learning_rate": 9.383262813764471e-06, + "loss": 0.1611, + "step": 10203 + }, + { + "epoch": 4.27, + "learning_rate": 9.381570641197795e-06, + "loss": 0.1518, + "step": 10204 + }, + { + "epoch": 4.27, + "learning_rate": 9.37987848640732e-06, + "loss": 0.183, + "step": 10205 + }, + { + "epoch": 4.27, + "learning_rate": 9.378186349441685e-06, + "loss": 0.1258, + "step": 10206 + }, + { + "epoch": 4.27, + "learning_rate": 9.376494230349537e-06, + "loss": 0.1622, + "step": 10207 + }, + { + "epoch": 4.27, + "learning_rate": 9.374802129179505e-06, + "loss": 0.1771, + "step": 10208 + }, + { + "epoch": 4.27, + "learning_rate": 9.373110045980232e-06, + "loss": 0.1735, + "step": 10209 + }, + { + "epoch": 4.27, + "learning_rate": 9.371417980800357e-06, + "loss": 0.1723, + "step": 10210 + }, + { + "epoch": 4.28, + "learning_rate": 9.369725933688512e-06, + "loss": 0.1516, + "step": 10211 + }, + { + "epoch": 4.28, + "learning_rate": 9.368033904693334e-06, + "loss": 0.1744, + "step": 10212 + }, + { + "epoch": 4.28, + "learning_rate": 9.366341893863467e-06, + "loss": 0.1628, + "step": 10213 + }, + { + "epoch": 4.28, + "learning_rate": 9.364649901247534e-06, + "loss": 0.1779, + "step": 10214 + }, + { + "epoch": 4.28, + "learning_rate": 9.362957926894179e-06, + "loss": 0.1468, + "step": 10215 + }, + { + "epoch": 4.28, + "learning_rate": 9.361265970852031e-06, + "loss": 0.1671, + "step": 10216 + }, + { + "epoch": 4.28, + "learning_rate": 9.359574033169725e-06, + "loss": 0.1538, + "step": 10217 + }, + { + "epoch": 4.28, + "learning_rate": 9.357882113895899e-06, + "loss": 0.1673, + "step": 10218 + }, + { + "epoch": 4.28, + "learning_rate": 9.356190213079178e-06, + "loss": 0.1574, + "step": 10219 + }, + { + "epoch": 4.28, + "learning_rate": 9.354498330768198e-06, + "loss": 0.15, + "step": 10220 + }, + { + "epoch": 4.28, + "learning_rate": 9.352806467011595e-06, + "loss": 0.133, + "step": 10221 + }, + { + "epoch": 4.28, + "learning_rate": 9.35111462185799e-06, + "loss": 0.158, + "step": 10222 + }, + { + "epoch": 4.28, + "learning_rate": 9.349422795356022e-06, + "loss": 0.1445, + "step": 10223 + }, + { + "epoch": 4.28, + "learning_rate": 9.34773098755432e-06, + "loss": 0.2036, + "step": 10224 + }, + { + "epoch": 4.28, + "learning_rate": 9.34603919850151e-06, + "loss": 0.1615, + "step": 10225 + }, + { + "epoch": 4.28, + "learning_rate": 9.344347428246222e-06, + "loss": 0.1625, + "step": 10226 + }, + { + "epoch": 4.28, + "learning_rate": 9.342655676837085e-06, + "loss": 0.1402, + "step": 10227 + }, + { + "epoch": 4.28, + "learning_rate": 9.340963944322731e-06, + "loss": 0.185, + "step": 10228 + }, + { + "epoch": 4.28, + "learning_rate": 9.33927223075178e-06, + "loss": 0.1517, + "step": 10229 + }, + { + "epoch": 4.28, + "learning_rate": 9.337580536172864e-06, + "loss": 0.1605, + "step": 10230 + }, + { + "epoch": 4.28, + "learning_rate": 9.33588886063461e-06, + "loss": 0.1153, + "step": 10231 + }, + { + "epoch": 4.28, + "learning_rate": 9.334197204185638e-06, + "loss": 0.1657, + "step": 10232 + }, + { + "epoch": 4.28, + "learning_rate": 9.332505566874577e-06, + "loss": 0.1653, + "step": 10233 + }, + { + "epoch": 4.28, + "learning_rate": 9.330813948750053e-06, + "loss": 0.1501, + "step": 10234 + }, + { + "epoch": 4.29, + "learning_rate": 9.329122349860686e-06, + "loss": 0.1349, + "step": 10235 + }, + { + "epoch": 4.29, + "learning_rate": 9.327430770255106e-06, + "loss": 0.1312, + "step": 10236 + }, + { + "epoch": 4.29, + "learning_rate": 9.325739209981929e-06, + "loss": 0.155, + "step": 10237 + }, + { + "epoch": 4.29, + "learning_rate": 9.324047669089778e-06, + "loss": 0.1361, + "step": 10238 + }, + { + "epoch": 4.29, + "learning_rate": 9.32235614762728e-06, + "loss": 0.1375, + "step": 10239 + }, + { + "epoch": 4.29, + "learning_rate": 9.320664645643051e-06, + "loss": 0.1078, + "step": 10240 + }, + { + "epoch": 4.29, + "learning_rate": 9.318973163185713e-06, + "loss": 0.1587, + "step": 10241 + }, + { + "epoch": 4.29, + "learning_rate": 9.31728170030389e-06, + "loss": 0.172, + "step": 10242 + }, + { + "epoch": 4.29, + "learning_rate": 9.315590257046197e-06, + "loss": 0.1524, + "step": 10243 + }, + { + "epoch": 4.29, + "learning_rate": 9.313898833461255e-06, + "loss": 0.1059, + "step": 10244 + }, + { + "epoch": 4.29, + "learning_rate": 9.312207429597684e-06, + "loss": 0.1627, + "step": 10245 + }, + { + "epoch": 4.29, + "learning_rate": 9.310516045504096e-06, + "loss": 0.1688, + "step": 10246 + }, + { + "epoch": 4.29, + "learning_rate": 9.308824681229115e-06, + "loss": 0.1488, + "step": 10247 + }, + { + "epoch": 4.29, + "learning_rate": 9.307133336821357e-06, + "loss": 0.1414, + "step": 10248 + }, + { + "epoch": 4.29, + "learning_rate": 9.305442012329433e-06, + "loss": 0.1393, + "step": 10249 + }, + { + "epoch": 4.29, + "learning_rate": 9.303750707801963e-06, + "loss": 0.1709, + "step": 10250 + }, + { + "epoch": 4.29, + "learning_rate": 9.302059423287563e-06, + "loss": 0.1623, + "step": 10251 + }, + { + "epoch": 4.29, + "learning_rate": 9.300368158834843e-06, + "loss": 0.1603, + "step": 10252 + }, + { + "epoch": 4.29, + "learning_rate": 9.298676914492419e-06, + "loss": 0.174, + "step": 10253 + }, + { + "epoch": 4.29, + "learning_rate": 9.29698569030891e-06, + "loss": 0.1618, + "step": 10254 + }, + { + "epoch": 4.29, + "learning_rate": 9.295294486332916e-06, + "loss": 0.1749, + "step": 10255 + }, + { + "epoch": 4.29, + "learning_rate": 9.29360330261306e-06, + "loss": 0.175, + "step": 10256 + }, + { + "epoch": 4.29, + "learning_rate": 9.29191213919795e-06, + "loss": 0.1675, + "step": 10257 + }, + { + "epoch": 4.29, + "learning_rate": 9.290220996136199e-06, + "loss": 0.1704, + "step": 10258 + }, + { + "epoch": 4.3, + "learning_rate": 9.288529873476418e-06, + "loss": 0.1415, + "step": 10259 + }, + { + "epoch": 4.3, + "learning_rate": 9.28683877126721e-06, + "loss": 0.1458, + "step": 10260 + }, + { + "epoch": 4.3, + "learning_rate": 9.285147689557188e-06, + "loss": 0.1277, + "step": 10261 + }, + { + "epoch": 4.3, + "learning_rate": 9.283456628394966e-06, + "loss": 0.1414, + "step": 10262 + }, + { + "epoch": 4.3, + "learning_rate": 9.281765587829143e-06, + "loss": 0.1546, + "step": 10263 + }, + { + "epoch": 4.3, + "learning_rate": 9.28007456790833e-06, + "loss": 0.1535, + "step": 10264 + }, + { + "epoch": 4.3, + "learning_rate": 9.278383568681139e-06, + "loss": 0.161, + "step": 10265 + }, + { + "epoch": 4.3, + "learning_rate": 9.27669259019617e-06, + "loss": 0.1447, + "step": 10266 + }, + { + "epoch": 4.3, + "learning_rate": 9.275001632502029e-06, + "loss": 0.1506, + "step": 10267 + }, + { + "epoch": 4.3, + "learning_rate": 9.273310695647323e-06, + "loss": 0.1517, + "step": 10268 + }, + { + "epoch": 4.3, + "learning_rate": 9.271619779680658e-06, + "loss": 0.1645, + "step": 10269 + }, + { + "epoch": 4.3, + "learning_rate": 9.269928884650635e-06, + "loss": 0.1579, + "step": 10270 + }, + { + "epoch": 4.3, + "learning_rate": 9.268238010605858e-06, + "loss": 0.1617, + "step": 10271 + }, + { + "epoch": 4.3, + "learning_rate": 9.266547157594932e-06, + "loss": 0.1459, + "step": 10272 + }, + { + "epoch": 4.3, + "learning_rate": 9.264856325666455e-06, + "loss": 0.1122, + "step": 10273 + }, + { + "epoch": 4.3, + "learning_rate": 9.26316551486903e-06, + "loss": 0.1507, + "step": 10274 + }, + { + "epoch": 4.3, + "learning_rate": 9.261474725251261e-06, + "loss": 0.1768, + "step": 10275 + }, + { + "epoch": 4.3, + "learning_rate": 9.259783956861744e-06, + "loss": 0.1558, + "step": 10276 + }, + { + "epoch": 4.3, + "learning_rate": 9.258093209749081e-06, + "loss": 0.1558, + "step": 10277 + }, + { + "epoch": 4.3, + "learning_rate": 9.25640248396187e-06, + "loss": 0.1407, + "step": 10278 + }, + { + "epoch": 4.3, + "learning_rate": 9.254711779548709e-06, + "loss": 0.1367, + "step": 10279 + }, + { + "epoch": 4.3, + "learning_rate": 9.253021096558197e-06, + "loss": 0.1778, + "step": 10280 + }, + { + "epoch": 4.3, + "learning_rate": 9.251330435038935e-06, + "loss": 0.1635, + "step": 10281 + }, + { + "epoch": 4.3, + "learning_rate": 9.24963979503951e-06, + "loss": 0.111, + "step": 10282 + }, + { + "epoch": 4.31, + "learning_rate": 9.247949176608527e-06, + "loss": 0.1337, + "step": 10283 + }, + { + "epoch": 4.31, + "learning_rate": 9.246258579794574e-06, + "loss": 0.1602, + "step": 10284 + }, + { + "epoch": 4.31, + "learning_rate": 9.244568004646249e-06, + "loss": 0.1061, + "step": 10285 + }, + { + "epoch": 4.31, + "learning_rate": 9.24287745121215e-06, + "loss": 0.1671, + "step": 10286 + }, + { + "epoch": 4.31, + "learning_rate": 9.241186919540864e-06, + "loss": 0.202, + "step": 10287 + }, + { + "epoch": 4.31, + "learning_rate": 9.239496409680986e-06, + "loss": 0.1655, + "step": 10288 + }, + { + "epoch": 4.31, + "learning_rate": 9.237805921681112e-06, + "loss": 0.1347, + "step": 10289 + }, + { + "epoch": 4.31, + "learning_rate": 9.236115455589827e-06, + "loss": 0.1724, + "step": 10290 + }, + { + "epoch": 4.31, + "learning_rate": 9.234425011455727e-06, + "loss": 0.1192, + "step": 10291 + }, + { + "epoch": 4.31, + "learning_rate": 9.232734589327402e-06, + "loss": 0.1302, + "step": 10292 + }, + { + "epoch": 4.31, + "learning_rate": 9.231044189253438e-06, + "loss": 0.144, + "step": 10293 + }, + { + "epoch": 4.31, + "learning_rate": 9.229353811282427e-06, + "loss": 0.1895, + "step": 10294 + }, + { + "epoch": 4.31, + "learning_rate": 9.227663455462957e-06, + "loss": 0.1672, + "step": 10295 + }, + { + "epoch": 4.31, + "learning_rate": 9.225973121843615e-06, + "loss": 0.1519, + "step": 10296 + }, + { + "epoch": 4.31, + "learning_rate": 9.22428281047299e-06, + "loss": 0.1131, + "step": 10297 + }, + { + "epoch": 4.31, + "learning_rate": 9.222592521399666e-06, + "loss": 0.1864, + "step": 10298 + }, + { + "epoch": 4.31, + "learning_rate": 9.220902254672232e-06, + "loss": 0.1448, + "step": 10299 + }, + { + "epoch": 4.31, + "learning_rate": 9.21921201033927e-06, + "loss": 0.1724, + "step": 10300 + }, + { + "epoch": 4.31, + "learning_rate": 9.217521788449365e-06, + "loss": 0.1048, + "step": 10301 + }, + { + "epoch": 4.31, + "learning_rate": 9.215831589051103e-06, + "loss": 0.141, + "step": 10302 + }, + { + "epoch": 4.31, + "learning_rate": 9.214141412193065e-06, + "loss": 0.1361, + "step": 10303 + }, + { + "epoch": 4.31, + "learning_rate": 9.212451257923841e-06, + "loss": 0.1285, + "step": 10304 + }, + { + "epoch": 4.31, + "learning_rate": 9.210761126292e-06, + "loss": 0.1157, + "step": 10305 + }, + { + "epoch": 4.31, + "learning_rate": 9.209071017346134e-06, + "loss": 0.1308, + "step": 10306 + }, + { + "epoch": 4.32, + "learning_rate": 9.207380931134816e-06, + "loss": 0.1388, + "step": 10307 + }, + { + "epoch": 4.32, + "learning_rate": 9.205690867706631e-06, + "loss": 0.1371, + "step": 10308 + }, + { + "epoch": 4.32, + "learning_rate": 9.204000827110157e-06, + "loss": 0.1566, + "step": 10309 + }, + { + "epoch": 4.32, + "learning_rate": 9.202310809393973e-06, + "loss": 0.1379, + "step": 10310 + }, + { + "epoch": 4.32, + "learning_rate": 9.200620814606659e-06, + "loss": 0.1757, + "step": 10311 + }, + { + "epoch": 4.32, + "learning_rate": 9.198930842796786e-06, + "loss": 0.1404, + "step": 10312 + }, + { + "epoch": 4.32, + "learning_rate": 9.197240894012939e-06, + "loss": 0.1312, + "step": 10313 + }, + { + "epoch": 4.32, + "learning_rate": 9.195550968303688e-06, + "loss": 0.1814, + "step": 10314 + }, + { + "epoch": 4.32, + "learning_rate": 9.19386106571761e-06, + "loss": 0.1651, + "step": 10315 + }, + { + "epoch": 4.32, + "learning_rate": 9.192171186303283e-06, + "loss": 0.1172, + "step": 10316 + }, + { + "epoch": 4.32, + "learning_rate": 9.190481330109276e-06, + "loss": 0.1769, + "step": 10317 + }, + { + "epoch": 4.32, + "learning_rate": 9.188791497184165e-06, + "loss": 0.1593, + "step": 10318 + }, + { + "epoch": 4.32, + "learning_rate": 9.187101687576525e-06, + "loss": 0.1895, + "step": 10319 + }, + { + "epoch": 4.32, + "learning_rate": 9.185411901334924e-06, + "loss": 0.1344, + "step": 10320 + }, + { + "epoch": 4.32, + "learning_rate": 9.183722138507934e-06, + "loss": 0.1776, + "step": 10321 + }, + { + "epoch": 4.32, + "learning_rate": 9.18203239914413e-06, + "loss": 0.2088, + "step": 10322 + }, + { + "epoch": 4.32, + "learning_rate": 9.180342683292074e-06, + "loss": 0.1396, + "step": 10323 + }, + { + "epoch": 4.32, + "learning_rate": 9.178652991000342e-06, + "loss": 0.1484, + "step": 10324 + }, + { + "epoch": 4.32, + "learning_rate": 9.176963322317503e-06, + "loss": 0.1141, + "step": 10325 + }, + { + "epoch": 4.32, + "learning_rate": 9.175273677292122e-06, + "loss": 0.1434, + "step": 10326 + }, + { + "epoch": 4.32, + "learning_rate": 9.17358405597277e-06, + "loss": 0.1797, + "step": 10327 + }, + { + "epoch": 4.32, + "learning_rate": 9.171894458408007e-06, + "loss": 0.1074, + "step": 10328 + }, + { + "epoch": 4.32, + "learning_rate": 9.170204884646402e-06, + "loss": 0.1534, + "step": 10329 + }, + { + "epoch": 4.32, + "learning_rate": 9.168515334736526e-06, + "loss": 0.1112, + "step": 10330 + }, + { + "epoch": 4.33, + "learning_rate": 9.166825808726934e-06, + "loss": 0.1436, + "step": 10331 + }, + { + "epoch": 4.33, + "learning_rate": 9.165136306666196e-06, + "loss": 0.1286, + "step": 10332 + }, + { + "epoch": 4.33, + "learning_rate": 9.163446828602876e-06, + "loss": 0.1368, + "step": 10333 + }, + { + "epoch": 4.33, + "learning_rate": 9.161757374585533e-06, + "loss": 0.1154, + "step": 10334 + }, + { + "epoch": 4.33, + "learning_rate": 9.16006794466273e-06, + "loss": 0.1594, + "step": 10335 + }, + { + "epoch": 4.33, + "learning_rate": 9.158378538883032e-06, + "loss": 0.1529, + "step": 10336 + }, + { + "epoch": 4.33, + "learning_rate": 9.156689157294992e-06, + "loss": 0.1316, + "step": 10337 + }, + { + "epoch": 4.33, + "learning_rate": 9.154999799947174e-06, + "loss": 0.1473, + "step": 10338 + }, + { + "epoch": 4.33, + "learning_rate": 9.153310466888138e-06, + "loss": 0.1211, + "step": 10339 + }, + { + "epoch": 4.33, + "learning_rate": 9.151621158166443e-06, + "loss": 0.1527, + "step": 10340 + }, + { + "epoch": 4.33, + "learning_rate": 9.149931873830642e-06, + "loss": 0.1493, + "step": 10341 + }, + { + "epoch": 4.33, + "learning_rate": 9.148242613929295e-06, + "loss": 0.1741, + "step": 10342 + }, + { + "epoch": 4.33, + "learning_rate": 9.146553378510961e-06, + "loss": 0.1669, + "step": 10343 + }, + { + "epoch": 4.33, + "learning_rate": 9.144864167624191e-06, + "loss": 0.1436, + "step": 10344 + }, + { + "epoch": 4.33, + "learning_rate": 9.143174981317542e-06, + "loss": 0.1587, + "step": 10345 + }, + { + "epoch": 4.33, + "learning_rate": 9.141485819639569e-06, + "loss": 0.2177, + "step": 10346 + }, + { + "epoch": 4.33, + "learning_rate": 9.139796682638824e-06, + "loss": 0.1276, + "step": 10347 + }, + { + "epoch": 4.33, + "learning_rate": 9.138107570363857e-06, + "loss": 0.1801, + "step": 10348 + }, + { + "epoch": 4.33, + "learning_rate": 9.13641848286323e-06, + "loss": 0.1598, + "step": 10349 + }, + { + "epoch": 4.33, + "learning_rate": 9.134729420185481e-06, + "loss": 0.1672, + "step": 10350 + }, + { + "epoch": 4.33, + "learning_rate": 9.13304038237917e-06, + "loss": 0.1669, + "step": 10351 + }, + { + "epoch": 4.33, + "learning_rate": 9.13135136949284e-06, + "loss": 0.1917, + "step": 10352 + }, + { + "epoch": 4.33, + "learning_rate": 9.129662381575047e-06, + "loss": 0.1498, + "step": 10353 + }, + { + "epoch": 4.33, + "learning_rate": 9.127973418674338e-06, + "loss": 0.1808, + "step": 10354 + }, + { + "epoch": 4.34, + "learning_rate": 9.126284480839256e-06, + "loss": 0.1474, + "step": 10355 + }, + { + "epoch": 4.34, + "learning_rate": 9.124595568118352e-06, + "loss": 0.1349, + "step": 10356 + }, + { + "epoch": 4.34, + "learning_rate": 9.122906680560173e-06, + "loss": 0.1752, + "step": 10357 + }, + { + "epoch": 4.34, + "learning_rate": 9.121217818213263e-06, + "loss": 0.126, + "step": 10358 + }, + { + "epoch": 4.34, + "learning_rate": 9.119528981126165e-06, + "loss": 0.1602, + "step": 10359 + }, + { + "epoch": 4.34, + "learning_rate": 9.117840169347428e-06, + "loss": 0.1327, + "step": 10360 + }, + { + "epoch": 4.34, + "learning_rate": 9.11615138292559e-06, + "loss": 0.1333, + "step": 10361 + }, + { + "epoch": 4.34, + "learning_rate": 9.114462621909196e-06, + "loss": 0.1549, + "step": 10362 + }, + { + "epoch": 4.34, + "learning_rate": 9.11277388634679e-06, + "loss": 0.2197, + "step": 10363 + }, + { + "epoch": 4.34, + "learning_rate": 9.111085176286912e-06, + "loss": 0.1698, + "step": 10364 + }, + { + "epoch": 4.34, + "learning_rate": 9.109396491778099e-06, + "loss": 0.1588, + "step": 10365 + }, + { + "epoch": 4.34, + "learning_rate": 9.107707832868896e-06, + "loss": 0.1549, + "step": 10366 + }, + { + "epoch": 4.34, + "learning_rate": 9.106019199607838e-06, + "loss": 0.1317, + "step": 10367 + }, + { + "epoch": 4.34, + "learning_rate": 9.104330592043464e-06, + "loss": 0.1306, + "step": 10368 + }, + { + "epoch": 4.34, + "learning_rate": 9.102642010224313e-06, + "loss": 0.1434, + "step": 10369 + }, + { + "epoch": 4.34, + "learning_rate": 9.100953454198924e-06, + "loss": 0.1234, + "step": 10370 + }, + { + "epoch": 4.34, + "learning_rate": 9.099264924015826e-06, + "loss": 0.1644, + "step": 10371 + }, + { + "epoch": 4.34, + "learning_rate": 9.097576419723564e-06, + "loss": 0.1702, + "step": 10372 + }, + { + "epoch": 4.34, + "learning_rate": 9.095887941370662e-06, + "loss": 0.1733, + "step": 10373 + }, + { + "epoch": 4.34, + "learning_rate": 9.094199489005664e-06, + "loss": 0.0816, + "step": 10374 + }, + { + "epoch": 4.34, + "learning_rate": 9.092511062677092e-06, + "loss": 0.1213, + "step": 10375 + }, + { + "epoch": 4.34, + "learning_rate": 9.090822662433485e-06, + "loss": 0.1588, + "step": 10376 + }, + { + "epoch": 4.34, + "learning_rate": 9.089134288323377e-06, + "loss": 0.1595, + "step": 10377 + }, + { + "epoch": 4.34, + "learning_rate": 9.087445940395292e-06, + "loss": 0.1875, + "step": 10378 + }, + { + "epoch": 4.35, + "learning_rate": 9.085757618697765e-06, + "loss": 0.1474, + "step": 10379 + }, + { + "epoch": 4.35, + "learning_rate": 9.084069323279325e-06, + "loss": 0.1392, + "step": 10380 + }, + { + "epoch": 4.35, + "learning_rate": 9.0823810541885e-06, + "loss": 0.1543, + "step": 10381 + }, + { + "epoch": 4.35, + "learning_rate": 9.080692811473815e-06, + "loss": 0.1155, + "step": 10382 + }, + { + "epoch": 4.35, + "learning_rate": 9.079004595183799e-06, + "loss": 0.1661, + "step": 10383 + }, + { + "epoch": 4.35, + "learning_rate": 9.07731640536698e-06, + "loss": 0.1738, + "step": 10384 + }, + { + "epoch": 4.35, + "learning_rate": 9.075628242071884e-06, + "loss": 0.1154, + "step": 10385 + }, + { + "epoch": 4.35, + "learning_rate": 9.073940105347032e-06, + "loss": 0.1177, + "step": 10386 + }, + { + "epoch": 4.35, + "learning_rate": 9.07225199524095e-06, + "loss": 0.1713, + "step": 10387 + }, + { + "epoch": 4.35, + "learning_rate": 9.07056391180216e-06, + "loss": 0.1511, + "step": 10388 + }, + { + "epoch": 4.35, + "learning_rate": 9.068875855079186e-06, + "loss": 0.1077, + "step": 10389 + }, + { + "epoch": 4.35, + "learning_rate": 9.067187825120552e-06, + "loss": 0.1106, + "step": 10390 + }, + { + "epoch": 4.35, + "learning_rate": 9.065499821974774e-06, + "loss": 0.1994, + "step": 10391 + }, + { + "epoch": 4.35, + "learning_rate": 9.063811845690374e-06, + "loss": 0.1749, + "step": 10392 + }, + { + "epoch": 4.35, + "learning_rate": 9.062123896315872e-06, + "loss": 0.1925, + "step": 10393 + }, + { + "epoch": 4.35, + "learning_rate": 9.06043597389979e-06, + "loss": 0.1688, + "step": 10394 + }, + { + "epoch": 4.35, + "learning_rate": 9.05874807849064e-06, + "loss": 0.1216, + "step": 10395 + }, + { + "epoch": 4.35, + "learning_rate": 9.05706021013694e-06, + "loss": 0.1676, + "step": 10396 + }, + { + "epoch": 4.35, + "learning_rate": 9.055372368887207e-06, + "loss": 0.1076, + "step": 10397 + }, + { + "epoch": 4.35, + "learning_rate": 9.053684554789959e-06, + "loss": 0.1369, + "step": 10398 + }, + { + "epoch": 4.35, + "learning_rate": 9.051996767893707e-06, + "loss": 0.1862, + "step": 10399 + }, + { + "epoch": 4.35, + "learning_rate": 9.050309008246965e-06, + "loss": 0.1327, + "step": 10400 + }, + { + "epoch": 4.35, + "learning_rate": 9.048621275898252e-06, + "loss": 0.1802, + "step": 10401 + }, + { + "epoch": 4.36, + "learning_rate": 9.046933570896073e-06, + "loss": 0.1715, + "step": 10402 + }, + { + "epoch": 4.36, + "learning_rate": 9.045245893288943e-06, + "loss": 0.1338, + "step": 10403 + }, + { + "epoch": 4.36, + "learning_rate": 9.043558243125374e-06, + "loss": 0.145, + "step": 10404 + }, + { + "epoch": 4.36, + "learning_rate": 9.041870620453874e-06, + "loss": 0.1623, + "step": 10405 + }, + { + "epoch": 4.36, + "learning_rate": 9.040183025322952e-06, + "loss": 0.199, + "step": 10406 + }, + { + "epoch": 4.36, + "learning_rate": 9.03849545778112e-06, + "loss": 0.149, + "step": 10407 + }, + { + "epoch": 4.36, + "learning_rate": 9.03680791787688e-06, + "loss": 0.1117, + "step": 10408 + }, + { + "epoch": 4.36, + "learning_rate": 9.035120405658741e-06, + "loss": 0.1546, + "step": 10409 + }, + { + "epoch": 4.36, + "learning_rate": 9.03343292117521e-06, + "loss": 0.1345, + "step": 10410 + }, + { + "epoch": 4.36, + "learning_rate": 9.031745464474794e-06, + "loss": 0.131, + "step": 10411 + }, + { + "epoch": 4.36, + "learning_rate": 9.030058035605996e-06, + "loss": 0.1347, + "step": 10412 + }, + { + "epoch": 4.36, + "learning_rate": 9.028370634617316e-06, + "loss": 0.1506, + "step": 10413 + }, + { + "epoch": 4.36, + "learning_rate": 9.026683261557263e-06, + "loss": 0.1459, + "step": 10414 + }, + { + "epoch": 4.36, + "learning_rate": 9.024995916474333e-06, + "loss": 0.1284, + "step": 10415 + }, + { + "epoch": 4.36, + "learning_rate": 9.023308599417032e-06, + "loss": 0.1399, + "step": 10416 + }, + { + "epoch": 4.36, + "learning_rate": 9.021621310433861e-06, + "loss": 0.164, + "step": 10417 + }, + { + "epoch": 4.36, + "learning_rate": 9.019934049573316e-06, + "loss": 0.1458, + "step": 10418 + }, + { + "epoch": 4.36, + "learning_rate": 9.018246816883894e-06, + "loss": 0.1446, + "step": 10419 + }, + { + "epoch": 4.36, + "learning_rate": 9.016559612414098e-06, + "loss": 0.151, + "step": 10420 + }, + { + "epoch": 4.36, + "learning_rate": 9.014872436212423e-06, + "loss": 0.1738, + "step": 10421 + }, + { + "epoch": 4.36, + "learning_rate": 9.013185288327365e-06, + "loss": 0.1546, + "step": 10422 + }, + { + "epoch": 4.36, + "learning_rate": 9.01149816880742e-06, + "loss": 0.182, + "step": 10423 + }, + { + "epoch": 4.36, + "learning_rate": 9.009811077701083e-06, + "loss": 0.1314, + "step": 10424 + }, + { + "epoch": 4.36, + "learning_rate": 9.008124015056851e-06, + "loss": 0.1467, + "step": 10425 + }, + { + "epoch": 4.37, + "learning_rate": 9.00643698092321e-06, + "loss": 0.1292, + "step": 10426 + }, + { + "epoch": 4.37, + "learning_rate": 9.004749975348657e-06, + "loss": 0.0717, + "step": 10427 + }, + { + "epoch": 4.37, + "learning_rate": 9.003062998381685e-06, + "loss": 0.1891, + "step": 10428 + }, + { + "epoch": 4.37, + "learning_rate": 9.00137605007078e-06, + "loss": 0.1756, + "step": 10429 + }, + { + "epoch": 4.37, + "learning_rate": 8.999689130464436e-06, + "loss": 0.1271, + "step": 10430 + }, + { + "epoch": 4.37, + "learning_rate": 8.99800223961114e-06, + "loss": 0.156, + "step": 10431 + }, + { + "epoch": 4.37, + "learning_rate": 8.99631537755938e-06, + "loss": 0.1119, + "step": 10432 + }, + { + "epoch": 4.37, + "learning_rate": 8.994628544357641e-06, + "loss": 0.1482, + "step": 10433 + }, + { + "epoch": 4.37, + "learning_rate": 8.992941740054418e-06, + "loss": 0.1023, + "step": 10434 + }, + { + "epoch": 4.37, + "learning_rate": 8.991254964698188e-06, + "loss": 0.1573, + "step": 10435 + }, + { + "epoch": 4.37, + "learning_rate": 8.989568218337438e-06, + "loss": 0.1465, + "step": 10436 + }, + { + "epoch": 4.37, + "learning_rate": 8.987881501020656e-06, + "loss": 0.155, + "step": 10437 + }, + { + "epoch": 4.37, + "learning_rate": 8.98619481279632e-06, + "loss": 0.1424, + "step": 10438 + }, + { + "epoch": 4.37, + "learning_rate": 8.984508153712912e-06, + "loss": 0.0995, + "step": 10439 + }, + { + "epoch": 4.37, + "learning_rate": 8.982821523818923e-06, + "loss": 0.1584, + "step": 10440 + }, + { + "epoch": 4.37, + "learning_rate": 8.981134923162823e-06, + "loss": 0.1327, + "step": 10441 + }, + { + "epoch": 4.37, + "learning_rate": 8.979448351793095e-06, + "loss": 0.1538, + "step": 10442 + }, + { + "epoch": 4.37, + "learning_rate": 8.977761809758218e-06, + "loss": 0.1266, + "step": 10443 + }, + { + "epoch": 4.37, + "learning_rate": 8.97607529710667e-06, + "loss": 0.1584, + "step": 10444 + }, + { + "epoch": 4.37, + "learning_rate": 8.974388813886932e-06, + "loss": 0.1403, + "step": 10445 + }, + { + "epoch": 4.37, + "learning_rate": 8.972702360147474e-06, + "loss": 0.1293, + "step": 10446 + }, + { + "epoch": 4.37, + "learning_rate": 8.971015935936777e-06, + "loss": 0.142, + "step": 10447 + }, + { + "epoch": 4.37, + "learning_rate": 8.969329541303315e-06, + "loss": 0.1632, + "step": 10448 + }, + { + "epoch": 4.37, + "learning_rate": 8.967643176295557e-06, + "loss": 0.1862, + "step": 10449 + }, + { + "epoch": 4.38, + "learning_rate": 8.96595684096198e-06, + "loss": 0.1808, + "step": 10450 + }, + { + "epoch": 4.38, + "learning_rate": 8.964270535351058e-06, + "loss": 0.1826, + "step": 10451 + }, + { + "epoch": 4.38, + "learning_rate": 8.962584259511262e-06, + "loss": 0.1768, + "step": 10452 + }, + { + "epoch": 4.38, + "learning_rate": 8.960898013491058e-06, + "loss": 0.1202, + "step": 10453 + }, + { + "epoch": 4.38, + "learning_rate": 8.959211797338919e-06, + "loss": 0.1497, + "step": 10454 + }, + { + "epoch": 4.38, + "learning_rate": 8.957525611103316e-06, + "loss": 0.1282, + "step": 10455 + }, + { + "epoch": 4.38, + "learning_rate": 8.955839454832711e-06, + "loss": 0.148, + "step": 10456 + }, + { + "epoch": 4.38, + "learning_rate": 8.954153328575574e-06, + "loss": 0.1241, + "step": 10457 + }, + { + "epoch": 4.38, + "learning_rate": 8.952467232380375e-06, + "loss": 0.1631, + "step": 10458 + }, + { + "epoch": 4.38, + "learning_rate": 8.950781166295573e-06, + "loss": 0.1293, + "step": 10459 + }, + { + "epoch": 4.38, + "learning_rate": 8.949095130369636e-06, + "loss": 0.1547, + "step": 10460 + }, + { + "epoch": 4.38, + "learning_rate": 8.947409124651027e-06, + "loss": 0.1538, + "step": 10461 + }, + { + "epoch": 4.38, + "learning_rate": 8.94572314918821e-06, + "loss": 0.1362, + "step": 10462 + }, + { + "epoch": 4.38, + "learning_rate": 8.944037204029646e-06, + "loss": 0.1879, + "step": 10463 + }, + { + "epoch": 4.38, + "learning_rate": 8.942351289223795e-06, + "loss": 0.1804, + "step": 10464 + }, + { + "epoch": 4.38, + "learning_rate": 8.940665404819114e-06, + "loss": 0.1195, + "step": 10465 + }, + { + "epoch": 4.38, + "learning_rate": 8.93897955086407e-06, + "loss": 0.1211, + "step": 10466 + }, + { + "epoch": 4.38, + "learning_rate": 8.937293727407115e-06, + "loss": 0.1555, + "step": 10467 + }, + { + "epoch": 4.38, + "learning_rate": 8.935607934496708e-06, + "loss": 0.1263, + "step": 10468 + }, + { + "epoch": 4.38, + "learning_rate": 8.93392217218131e-06, + "loss": 0.184, + "step": 10469 + }, + { + "epoch": 4.38, + "learning_rate": 8.93223644050937e-06, + "loss": 0.1298, + "step": 10470 + }, + { + "epoch": 4.38, + "learning_rate": 8.930550739529345e-06, + "loss": 0.13, + "step": 10471 + }, + { + "epoch": 4.38, + "learning_rate": 8.928865069289693e-06, + "loss": 0.134, + "step": 10472 + }, + { + "epoch": 4.38, + "learning_rate": 8.927179429838864e-06, + "loss": 0.145, + "step": 10473 + }, + { + "epoch": 4.39, + "learning_rate": 8.925493821225309e-06, + "loss": 0.1614, + "step": 10474 + }, + { + "epoch": 4.39, + "learning_rate": 8.923808243497482e-06, + "loss": 0.1361, + "step": 10475 + }, + { + "epoch": 4.39, + "learning_rate": 8.92212269670383e-06, + "loss": 0.1655, + "step": 10476 + }, + { + "epoch": 4.39, + "learning_rate": 8.920437180892803e-06, + "loss": 0.1206, + "step": 10477 + }, + { + "epoch": 4.39, + "learning_rate": 8.918751696112856e-06, + "loss": 0.1356, + "step": 10478 + }, + { + "epoch": 4.39, + "learning_rate": 8.917066242412429e-06, + "loss": 0.1348, + "step": 10479 + }, + { + "epoch": 4.39, + "learning_rate": 8.915380819839972e-06, + "loss": 0.1483, + "step": 10480 + }, + { + "epoch": 4.39, + "learning_rate": 8.91369542844393e-06, + "loss": 0.1389, + "step": 10481 + }, + { + "epoch": 4.39, + "learning_rate": 8.912010068272752e-06, + "loss": 0.1384, + "step": 10482 + }, + { + "epoch": 4.39, + "learning_rate": 8.910324739374877e-06, + "loss": 0.1575, + "step": 10483 + }, + { + "epoch": 4.39, + "learning_rate": 8.908639441798751e-06, + "loss": 0.1206, + "step": 10484 + }, + { + "epoch": 4.39, + "learning_rate": 8.906954175592819e-06, + "loss": 0.1597, + "step": 10485 + }, + { + "epoch": 4.39, + "learning_rate": 8.905268940805519e-06, + "loss": 0.1651, + "step": 10486 + }, + { + "epoch": 4.39, + "learning_rate": 8.903583737485286e-06, + "loss": 0.1341, + "step": 10487 + }, + { + "epoch": 4.39, + "learning_rate": 8.901898565680567e-06, + "loss": 0.1445, + "step": 10488 + }, + { + "epoch": 4.39, + "learning_rate": 8.900213425439803e-06, + "loss": 0.1532, + "step": 10489 + }, + { + "epoch": 4.39, + "learning_rate": 8.898528316811425e-06, + "loss": 0.1891, + "step": 10490 + }, + { + "epoch": 4.39, + "learning_rate": 8.896843239843873e-06, + "loss": 0.1499, + "step": 10491 + }, + { + "epoch": 4.39, + "learning_rate": 8.895158194585583e-06, + "loss": 0.1366, + "step": 10492 + }, + { + "epoch": 4.39, + "learning_rate": 8.893473181084993e-06, + "loss": 0.1487, + "step": 10493 + }, + { + "epoch": 4.39, + "learning_rate": 8.891788199390533e-06, + "loss": 0.1514, + "step": 10494 + }, + { + "epoch": 4.39, + "learning_rate": 8.890103249550638e-06, + "loss": 0.1496, + "step": 10495 + }, + { + "epoch": 4.39, + "learning_rate": 8.88841833161374e-06, + "loss": 0.1665, + "step": 10496 + }, + { + "epoch": 4.39, + "learning_rate": 8.88673344562827e-06, + "loss": 0.1539, + "step": 10497 + }, + { + "epoch": 4.4, + "learning_rate": 8.88504859164266e-06, + "loss": 0.1255, + "step": 10498 + }, + { + "epoch": 4.4, + "learning_rate": 8.883363769705338e-06, + "loss": 0.1813, + "step": 10499 + }, + { + "epoch": 4.4, + "learning_rate": 8.881678979864734e-06, + "loss": 0.1768, + "step": 10500 + }, + { + "epoch": 4.4, + "learning_rate": 8.879994222169275e-06, + "loss": 0.1486, + "step": 10501 + }, + { + "epoch": 4.4, + "learning_rate": 8.87830949666739e-06, + "loss": 0.1229, + "step": 10502 + }, + { + "epoch": 4.4, + "learning_rate": 8.876624803407501e-06, + "loss": 0.1755, + "step": 10503 + }, + { + "epoch": 4.4, + "learning_rate": 8.874940142438036e-06, + "loss": 0.1787, + "step": 10504 + }, + { + "epoch": 4.4, + "learning_rate": 8.873255513807418e-06, + "loss": 0.1251, + "step": 10505 + }, + { + "epoch": 4.4, + "learning_rate": 8.87157091756407e-06, + "loss": 0.1068, + "step": 10506 + }, + { + "epoch": 4.4, + "learning_rate": 8.869886353756415e-06, + "loss": 0.1293, + "step": 10507 + }, + { + "epoch": 4.4, + "learning_rate": 8.868201822432878e-06, + "loss": 0.1115, + "step": 10508 + }, + { + "epoch": 4.4, + "learning_rate": 8.86651732364187e-06, + "loss": 0.1715, + "step": 10509 + }, + { + "epoch": 4.4, + "learning_rate": 8.86483285743182e-06, + "loss": 0.1266, + "step": 10510 + }, + { + "epoch": 4.4, + "learning_rate": 8.863148423851138e-06, + "loss": 0.1839, + "step": 10511 + }, + { + "epoch": 4.4, + "learning_rate": 8.861464022948247e-06, + "loss": 0.1993, + "step": 10512 + }, + { + "epoch": 4.4, + "learning_rate": 8.859779654771563e-06, + "loss": 0.163, + "step": 10513 + }, + { + "epoch": 4.4, + "learning_rate": 8.858095319369499e-06, + "loss": 0.1587, + "step": 10514 + }, + { + "epoch": 4.4, + "learning_rate": 8.856411016790472e-06, + "loss": 0.1455, + "step": 10515 + }, + { + "epoch": 4.4, + "learning_rate": 8.8547267470829e-06, + "loss": 0.1502, + "step": 10516 + }, + { + "epoch": 4.4, + "learning_rate": 8.853042510295187e-06, + "loss": 0.1293, + "step": 10517 + }, + { + "epoch": 4.4, + "learning_rate": 8.851358306475748e-06, + "loss": 0.1647, + "step": 10518 + }, + { + "epoch": 4.4, + "learning_rate": 8.849674135672999e-06, + "loss": 0.1932, + "step": 10519 + }, + { + "epoch": 4.4, + "learning_rate": 8.847989997935345e-06, + "loss": 0.1486, + "step": 10520 + }, + { + "epoch": 4.4, + "learning_rate": 8.846305893311194e-06, + "loss": 0.1692, + "step": 10521 + }, + { + "epoch": 4.41, + "learning_rate": 8.844621821848958e-06, + "loss": 0.1121, + "step": 10522 + }, + { + "epoch": 4.41, + "learning_rate": 8.842937783597045e-06, + "loss": 0.1358, + "step": 10523 + }, + { + "epoch": 4.41, + "learning_rate": 8.841253778603855e-06, + "loss": 0.1654, + "step": 10524 + }, + { + "epoch": 4.41, + "learning_rate": 8.839569806917797e-06, + "loss": 0.1473, + "step": 10525 + }, + { + "epoch": 4.41, + "learning_rate": 8.837885868587279e-06, + "loss": 0.1257, + "step": 10526 + }, + { + "epoch": 4.41, + "learning_rate": 8.836201963660695e-06, + "loss": 0.1375, + "step": 10527 + }, + { + "epoch": 4.41, + "learning_rate": 8.834518092186456e-06, + "loss": 0.1417, + "step": 10528 + }, + { + "epoch": 4.41, + "learning_rate": 8.83283425421296e-06, + "loss": 0.139, + "step": 10529 + }, + { + "epoch": 4.41, + "learning_rate": 8.831150449788613e-06, + "loss": 0.1568, + "step": 10530 + }, + { + "epoch": 4.41, + "learning_rate": 8.829466678961803e-06, + "loss": 0.189, + "step": 10531 + }, + { + "epoch": 4.41, + "learning_rate": 8.827782941780934e-06, + "loss": 0.1277, + "step": 10532 + }, + { + "epoch": 4.41, + "learning_rate": 8.826099238294404e-06, + "loss": 0.1296, + "step": 10533 + }, + { + "epoch": 4.41, + "learning_rate": 8.824415568550612e-06, + "loss": 0.1962, + "step": 10534 + }, + { + "epoch": 4.41, + "learning_rate": 8.82273193259795e-06, + "loss": 0.1299, + "step": 10535 + }, + { + "epoch": 4.41, + "learning_rate": 8.821048330484814e-06, + "loss": 0.0992, + "step": 10536 + }, + { + "epoch": 4.41, + "learning_rate": 8.819364762259598e-06, + "loss": 0.1378, + "step": 10537 + }, + { + "epoch": 4.41, + "learning_rate": 8.817681227970693e-06, + "loss": 0.1181, + "step": 10538 + }, + { + "epoch": 4.41, + "learning_rate": 8.815997727666493e-06, + "loss": 0.1476, + "step": 10539 + }, + { + "epoch": 4.41, + "learning_rate": 8.814314261395389e-06, + "loss": 0.1612, + "step": 10540 + }, + { + "epoch": 4.41, + "learning_rate": 8.812630829205766e-06, + "loss": 0.1735, + "step": 10541 + }, + { + "epoch": 4.41, + "learning_rate": 8.810947431146019e-06, + "loss": 0.1622, + "step": 10542 + }, + { + "epoch": 4.41, + "learning_rate": 8.809264067264533e-06, + "loss": 0.1423, + "step": 10543 + }, + { + "epoch": 4.41, + "learning_rate": 8.807580737609693e-06, + "loss": 0.151, + "step": 10544 + }, + { + "epoch": 4.41, + "learning_rate": 8.805897442229888e-06, + "loss": 0.1505, + "step": 10545 + }, + { + "epoch": 4.42, + "learning_rate": 8.804214181173501e-06, + "loss": 0.1616, + "step": 10546 + }, + { + "epoch": 4.42, + "learning_rate": 8.802530954488915e-06, + "loss": 0.1944, + "step": 10547 + }, + { + "epoch": 4.42, + "learning_rate": 8.800847762224514e-06, + "loss": 0.1288, + "step": 10548 + }, + { + "epoch": 4.42, + "learning_rate": 8.799164604428683e-06, + "loss": 0.123, + "step": 10549 + }, + { + "epoch": 4.42, + "learning_rate": 8.797481481149798e-06, + "loss": 0.1643, + "step": 10550 + }, + { + "epoch": 4.42, + "learning_rate": 8.795798392436239e-06, + "loss": 0.1086, + "step": 10551 + }, + { + "epoch": 4.42, + "learning_rate": 8.794115338336388e-06, + "loss": 0.1179, + "step": 10552 + }, + { + "epoch": 4.42, + "learning_rate": 8.792432318898624e-06, + "loss": 0.1645, + "step": 10553 + }, + { + "epoch": 4.42, + "learning_rate": 8.790749334171323e-06, + "loss": 0.1417, + "step": 10554 + }, + { + "epoch": 4.42, + "learning_rate": 8.789066384202853e-06, + "loss": 0.1391, + "step": 10555 + }, + { + "epoch": 4.42, + "learning_rate": 8.787383469041598e-06, + "loss": 0.1525, + "step": 10556 + }, + { + "epoch": 4.42, + "learning_rate": 8.78570058873593e-06, + "loss": 0.1494, + "step": 10557 + }, + { + "epoch": 4.42, + "learning_rate": 8.78401774333422e-06, + "loss": 0.1371, + "step": 10558 + }, + { + "epoch": 4.42, + "learning_rate": 8.78233493288484e-06, + "loss": 0.1782, + "step": 10559 + }, + { + "epoch": 4.42, + "learning_rate": 8.780652157436163e-06, + "loss": 0.1544, + "step": 10560 + }, + { + "epoch": 4.42, + "learning_rate": 8.778969417036557e-06, + "loss": 0.1573, + "step": 10561 + }, + { + "epoch": 4.42, + "learning_rate": 8.77728671173439e-06, + "loss": 0.1482, + "step": 10562 + }, + { + "epoch": 4.42, + "learning_rate": 8.775604041578034e-06, + "loss": 0.1629, + "step": 10563 + }, + { + "epoch": 4.42, + "learning_rate": 8.773921406615854e-06, + "loss": 0.1126, + "step": 10564 + }, + { + "epoch": 4.42, + "learning_rate": 8.772238806896213e-06, + "loss": 0.1316, + "step": 10565 + }, + { + "epoch": 4.42, + "learning_rate": 8.770556242467479e-06, + "loss": 0.1752, + "step": 10566 + }, + { + "epoch": 4.42, + "learning_rate": 8.768873713378015e-06, + "loss": 0.1287, + "step": 10567 + }, + { + "epoch": 4.42, + "learning_rate": 8.767191219676184e-06, + "loss": 0.1922, + "step": 10568 + }, + { + "epoch": 4.42, + "learning_rate": 8.765508761410345e-06, + "loss": 0.1686, + "step": 10569 + }, + { + "epoch": 4.43, + "learning_rate": 8.763826338628865e-06, + "loss": 0.1586, + "step": 10570 + }, + { + "epoch": 4.43, + "learning_rate": 8.762143951380096e-06, + "loss": 0.168, + "step": 10571 + }, + { + "epoch": 4.43, + "learning_rate": 8.760461599712402e-06, + "loss": 0.1332, + "step": 10572 + }, + { + "epoch": 4.43, + "learning_rate": 8.758779283674142e-06, + "loss": 0.1216, + "step": 10573 + }, + { + "epoch": 4.43, + "learning_rate": 8.757097003313665e-06, + "loss": 0.1546, + "step": 10574 + }, + { + "epoch": 4.43, + "learning_rate": 8.755414758679336e-06, + "loss": 0.1556, + "step": 10575 + }, + { + "epoch": 4.43, + "learning_rate": 8.753732549819508e-06, + "loss": 0.1918, + "step": 10576 + }, + { + "epoch": 4.43, + "learning_rate": 8.752050376782525e-06, + "loss": 0.141, + "step": 10577 + }, + { + "epoch": 4.43, + "learning_rate": 8.750368239616752e-06, + "loss": 0.1161, + "step": 10578 + }, + { + "epoch": 4.43, + "learning_rate": 8.748686138370533e-06, + "loss": 0.1649, + "step": 10579 + }, + { + "epoch": 4.43, + "learning_rate": 8.747004073092218e-06, + "loss": 0.1614, + "step": 10580 + }, + { + "epoch": 4.43, + "learning_rate": 8.745322043830163e-06, + "loss": 0.1292, + "step": 10581 + }, + { + "epoch": 4.43, + "learning_rate": 8.743640050632709e-06, + "loss": 0.1331, + "step": 10582 + }, + { + "epoch": 4.43, + "learning_rate": 8.741958093548208e-06, + "loss": 0.166, + "step": 10583 + }, + { + "epoch": 4.43, + "learning_rate": 8.740276172625007e-06, + "loss": 0.1588, + "step": 10584 + }, + { + "epoch": 4.43, + "learning_rate": 8.738594287911447e-06, + "loss": 0.1624, + "step": 10585 + }, + { + "epoch": 4.43, + "learning_rate": 8.736912439455876e-06, + "loss": 0.1911, + "step": 10586 + }, + { + "epoch": 4.43, + "learning_rate": 8.735230627306637e-06, + "loss": 0.1964, + "step": 10587 + }, + { + "epoch": 4.43, + "learning_rate": 8.733548851512071e-06, + "loss": 0.1723, + "step": 10588 + }, + { + "epoch": 4.43, + "learning_rate": 8.731867112120516e-06, + "loss": 0.1644, + "step": 10589 + }, + { + "epoch": 4.43, + "learning_rate": 8.730185409180321e-06, + "loss": 0.1857, + "step": 10590 + }, + { + "epoch": 4.43, + "learning_rate": 8.728503742739816e-06, + "loss": 0.1473, + "step": 10591 + }, + { + "epoch": 4.43, + "learning_rate": 8.726822112847343e-06, + "loss": 0.1511, + "step": 10592 + }, + { + "epoch": 4.44, + "learning_rate": 8.725140519551238e-06, + "loss": 0.1551, + "step": 10593 + }, + { + "epoch": 4.44, + "learning_rate": 8.72345896289984e-06, + "loss": 0.1422, + "step": 10594 + }, + { + "epoch": 4.44, + "learning_rate": 8.721777442941479e-06, + "loss": 0.145, + "step": 10595 + }, + { + "epoch": 4.44, + "learning_rate": 8.720095959724492e-06, + "loss": 0.1595, + "step": 10596 + }, + { + "epoch": 4.44, + "learning_rate": 8.718414513297214e-06, + "loss": 0.1792, + "step": 10597 + }, + { + "epoch": 4.44, + "learning_rate": 8.716733103707972e-06, + "loss": 0.1171, + "step": 10598 + }, + { + "epoch": 4.44, + "learning_rate": 8.715051731005097e-06, + "loss": 0.1347, + "step": 10599 + }, + { + "epoch": 4.44, + "learning_rate": 8.713370395236918e-06, + "loss": 0.1383, + "step": 10600 + }, + { + "epoch": 4.44, + "learning_rate": 8.711689096451769e-06, + "loss": 0.1394, + "step": 10601 + }, + { + "epoch": 4.44, + "learning_rate": 8.71000783469797e-06, + "loss": 0.1451, + "step": 10602 + }, + { + "epoch": 4.44, + "learning_rate": 8.708326610023852e-06, + "loss": 0.1374, + "step": 10603 + }, + { + "epoch": 4.44, + "learning_rate": 8.706645422477739e-06, + "loss": 0.1479, + "step": 10604 + }, + { + "epoch": 4.44, + "learning_rate": 8.704964272107957e-06, + "loss": 0.1753, + "step": 10605 + }, + { + "epoch": 4.44, + "learning_rate": 8.703283158962828e-06, + "loss": 0.1383, + "step": 10606 + }, + { + "epoch": 4.44, + "learning_rate": 8.701602083090673e-06, + "loss": 0.1204, + "step": 10607 + }, + { + "epoch": 4.44, + "learning_rate": 8.699921044539815e-06, + "loss": 0.1408, + "step": 10608 + }, + { + "epoch": 4.44, + "learning_rate": 8.69824004335857e-06, + "loss": 0.1552, + "step": 10609 + }, + { + "epoch": 4.44, + "learning_rate": 8.696559079595261e-06, + "loss": 0.1682, + "step": 10610 + }, + { + "epoch": 4.44, + "learning_rate": 8.694878153298207e-06, + "loss": 0.1135, + "step": 10611 + }, + { + "epoch": 4.44, + "learning_rate": 8.69319726451572e-06, + "loss": 0.1273, + "step": 10612 + }, + { + "epoch": 4.44, + "learning_rate": 8.691516413296118e-06, + "loss": 0.0887, + "step": 10613 + }, + { + "epoch": 4.44, + "learning_rate": 8.689835599687717e-06, + "loss": 0.136, + "step": 10614 + }, + { + "epoch": 4.44, + "learning_rate": 8.68815482373883e-06, + "loss": 0.1311, + "step": 10615 + }, + { + "epoch": 4.44, + "learning_rate": 8.686474085497765e-06, + "loss": 0.1339, + "step": 10616 + }, + { + "epoch": 4.45, + "learning_rate": 8.68479338501284e-06, + "loss": 0.1429, + "step": 10617 + }, + { + "epoch": 4.45, + "learning_rate": 8.683112722332361e-06, + "loss": 0.1564, + "step": 10618 + }, + { + "epoch": 4.45, + "learning_rate": 8.681432097504636e-06, + "loss": 0.1621, + "step": 10619 + }, + { + "epoch": 4.45, + "learning_rate": 8.67975151057798e-06, + "loss": 0.1507, + "step": 10620 + }, + { + "epoch": 4.45, + "learning_rate": 8.678070961600693e-06, + "loss": 0.1407, + "step": 10621 + }, + { + "epoch": 4.45, + "learning_rate": 8.676390450621084e-06, + "loss": 0.1741, + "step": 10622 + }, + { + "epoch": 4.45, + "learning_rate": 8.674709977687456e-06, + "loss": 0.1286, + "step": 10623 + }, + { + "epoch": 4.45, + "learning_rate": 8.67302954284811e-06, + "loss": 0.1865, + "step": 10624 + }, + { + "epoch": 4.45, + "learning_rate": 8.67134914615136e-06, + "loss": 0.1679, + "step": 10625 + }, + { + "epoch": 4.45, + "learning_rate": 8.669668787645494e-06, + "loss": 0.1714, + "step": 10626 + }, + { + "epoch": 4.45, + "learning_rate": 8.667988467378819e-06, + "loss": 0.1482, + "step": 10627 + }, + { + "epoch": 4.45, + "learning_rate": 8.666308185399634e-06, + "loss": 0.2011, + "step": 10628 + }, + { + "epoch": 4.45, + "learning_rate": 8.664627941756236e-06, + "loss": 0.1634, + "step": 10629 + }, + { + "epoch": 4.45, + "learning_rate": 8.662947736496923e-06, + "loss": 0.1275, + "step": 10630 + }, + { + "epoch": 4.45, + "learning_rate": 8.661267569669994e-06, + "loss": 0.1586, + "step": 10631 + }, + { + "epoch": 4.45, + "learning_rate": 8.659587441323736e-06, + "loss": 0.16, + "step": 10632 + }, + { + "epoch": 4.45, + "learning_rate": 8.65790735150645e-06, + "loss": 0.1889, + "step": 10633 + }, + { + "epoch": 4.45, + "learning_rate": 8.656227300266425e-06, + "loss": 0.2088, + "step": 10634 + }, + { + "epoch": 4.45, + "learning_rate": 8.654547287651955e-06, + "loss": 0.1768, + "step": 10635 + }, + { + "epoch": 4.45, + "learning_rate": 8.652867313711328e-06, + "loss": 0.141, + "step": 10636 + }, + { + "epoch": 4.45, + "learning_rate": 8.651187378492836e-06, + "loss": 0.1424, + "step": 10637 + }, + { + "epoch": 4.45, + "learning_rate": 8.649507482044766e-06, + "loss": 0.1509, + "step": 10638 + }, + { + "epoch": 4.45, + "learning_rate": 8.647827624415403e-06, + "loss": 0.176, + "step": 10639 + }, + { + "epoch": 4.45, + "learning_rate": 8.646147805653037e-06, + "loss": 0.1154, + "step": 10640 + }, + { + "epoch": 4.46, + "learning_rate": 8.644468025805952e-06, + "loss": 0.1167, + "step": 10641 + }, + { + "epoch": 4.46, + "learning_rate": 8.642788284922427e-06, + "loss": 0.149, + "step": 10642 + }, + { + "epoch": 4.46, + "learning_rate": 8.641108583050751e-06, + "loss": 0.19, + "step": 10643 + }, + { + "epoch": 4.46, + "learning_rate": 8.639428920239205e-06, + "loss": 0.1419, + "step": 10644 + }, + { + "epoch": 4.46, + "learning_rate": 8.637749296536065e-06, + "loss": 0.2231, + "step": 10645 + }, + { + "epoch": 4.46, + "learning_rate": 8.636069711989614e-06, + "loss": 0.1218, + "step": 10646 + }, + { + "epoch": 4.46, + "learning_rate": 8.634390166648126e-06, + "loss": 0.1739, + "step": 10647 + }, + { + "epoch": 4.46, + "learning_rate": 8.632710660559883e-06, + "loss": 0.1368, + "step": 10648 + }, + { + "epoch": 4.46, + "learning_rate": 8.631031193773158e-06, + "loss": 0.1488, + "step": 10649 + }, + { + "epoch": 4.46, + "learning_rate": 8.629351766336225e-06, + "loss": 0.1991, + "step": 10650 + }, + { + "epoch": 4.46, + "learning_rate": 8.627672378297361e-06, + "loss": 0.154, + "step": 10651 + }, + { + "epoch": 4.46, + "learning_rate": 8.625993029704838e-06, + "loss": 0.1503, + "step": 10652 + }, + { + "epoch": 4.46, + "learning_rate": 8.624313720606922e-06, + "loss": 0.1851, + "step": 10653 + }, + { + "epoch": 4.46, + "learning_rate": 8.62263445105189e-06, + "loss": 0.1845, + "step": 10654 + }, + { + "epoch": 4.46, + "learning_rate": 8.62095522108801e-06, + "loss": 0.1492, + "step": 10655 + }, + { + "epoch": 4.46, + "learning_rate": 8.619276030763545e-06, + "loss": 0.218, + "step": 10656 + }, + { + "epoch": 4.46, + "learning_rate": 8.617596880126766e-06, + "loss": 0.1528, + "step": 10657 + }, + { + "epoch": 4.46, + "learning_rate": 8.61591776922594e-06, + "loss": 0.1055, + "step": 10658 + }, + { + "epoch": 4.46, + "learning_rate": 8.614238698109329e-06, + "loss": 0.1446, + "step": 10659 + }, + { + "epoch": 4.46, + "learning_rate": 8.612559666825196e-06, + "loss": 0.1775, + "step": 10660 + }, + { + "epoch": 4.46, + "learning_rate": 8.610880675421805e-06, + "loss": 0.1524, + "step": 10661 + }, + { + "epoch": 4.46, + "learning_rate": 8.609201723947415e-06, + "loss": 0.1336, + "step": 10662 + }, + { + "epoch": 4.46, + "learning_rate": 8.607522812450288e-06, + "loss": 0.1588, + "step": 10663 + }, + { + "epoch": 4.46, + "learning_rate": 8.60584394097868e-06, + "loss": 0.1851, + "step": 10664 + }, + { + "epoch": 4.47, + "learning_rate": 8.604165109580855e-06, + "loss": 0.1002, + "step": 10665 + }, + { + "epoch": 4.47, + "learning_rate": 8.602486318305067e-06, + "loss": 0.172, + "step": 10666 + }, + { + "epoch": 4.47, + "learning_rate": 8.600807567199565e-06, + "loss": 0.1638, + "step": 10667 + }, + { + "epoch": 4.47, + "learning_rate": 8.599128856312607e-06, + "loss": 0.1558, + "step": 10668 + }, + { + "epoch": 4.47, + "learning_rate": 8.597450185692451e-06, + "loss": 0.1578, + "step": 10669 + }, + { + "epoch": 4.47, + "learning_rate": 8.59577155538734e-06, + "loss": 0.1578, + "step": 10670 + }, + { + "epoch": 4.47, + "learning_rate": 8.594092965445534e-06, + "loss": 0.1351, + "step": 10671 + }, + { + "epoch": 4.47, + "learning_rate": 8.592414415915275e-06, + "loss": 0.1154, + "step": 10672 + }, + { + "epoch": 4.47, + "learning_rate": 8.590735906844816e-06, + "loss": 0.1396, + "step": 10673 + }, + { + "epoch": 4.47, + "learning_rate": 8.5890574382824e-06, + "loss": 0.1288, + "step": 10674 + }, + { + "epoch": 4.47, + "learning_rate": 8.58737901027628e-06, + "loss": 0.154, + "step": 10675 + }, + { + "epoch": 4.47, + "learning_rate": 8.585700622874695e-06, + "loss": 0.1194, + "step": 10676 + }, + { + "epoch": 4.47, + "learning_rate": 8.584022276125891e-06, + "loss": 0.1551, + "step": 10677 + }, + { + "epoch": 4.47, + "learning_rate": 8.582343970078108e-06, + "loss": 0.1489, + "step": 10678 + }, + { + "epoch": 4.47, + "learning_rate": 8.580665704779593e-06, + "loss": 0.1997, + "step": 10679 + }, + { + "epoch": 4.47, + "learning_rate": 8.57898748027858e-06, + "loss": 0.1953, + "step": 10680 + }, + { + "epoch": 4.47, + "learning_rate": 8.577309296623311e-06, + "loss": 0.1389, + "step": 10681 + }, + { + "epoch": 4.47, + "learning_rate": 8.575631153862027e-06, + "loss": 0.1535, + "step": 10682 + }, + { + "epoch": 4.47, + "learning_rate": 8.573953052042959e-06, + "loss": 0.147, + "step": 10683 + }, + { + "epoch": 4.47, + "learning_rate": 8.572274991214345e-06, + "loss": 0.1321, + "step": 10684 + }, + { + "epoch": 4.47, + "learning_rate": 8.570596971424421e-06, + "loss": 0.1436, + "step": 10685 + }, + { + "epoch": 4.47, + "learning_rate": 8.568918992721418e-06, + "loss": 0.121, + "step": 10686 + }, + { + "epoch": 4.47, + "learning_rate": 8.567241055153567e-06, + "loss": 0.137, + "step": 10687 + }, + { + "epoch": 4.47, + "learning_rate": 8.565563158769103e-06, + "loss": 0.1576, + "step": 10688 + }, + { + "epoch": 4.48, + "learning_rate": 8.563885303616253e-06, + "loss": 0.1183, + "step": 10689 + }, + { + "epoch": 4.48, + "learning_rate": 8.562207489743245e-06, + "loss": 0.1324, + "step": 10690 + }, + { + "epoch": 4.48, + "learning_rate": 8.560529717198305e-06, + "loss": 0.1612, + "step": 10691 + }, + { + "epoch": 4.48, + "learning_rate": 8.558851986029663e-06, + "loss": 0.1455, + "step": 10692 + }, + { + "epoch": 4.48, + "learning_rate": 8.557174296285542e-06, + "loss": 0.1424, + "step": 10693 + }, + { + "epoch": 4.48, + "learning_rate": 8.555496648014163e-06, + "loss": 0.1679, + "step": 10694 + }, + { + "epoch": 4.48, + "learning_rate": 8.553819041263752e-06, + "loss": 0.15, + "step": 10695 + }, + { + "epoch": 4.48, + "learning_rate": 8.552141476082531e-06, + "loss": 0.1507, + "step": 10696 + }, + { + "epoch": 4.48, + "learning_rate": 8.550463952518716e-06, + "loss": 0.1299, + "step": 10697 + }, + { + "epoch": 4.48, + "learning_rate": 8.54878647062053e-06, + "loss": 0.1896, + "step": 10698 + }, + { + "epoch": 4.48, + "learning_rate": 8.547109030436189e-06, + "loss": 0.1801, + "step": 10699 + }, + { + "epoch": 4.48, + "learning_rate": 8.545431632013909e-06, + "loss": 0.1252, + "step": 10700 + }, + { + "epoch": 4.48, + "learning_rate": 8.543754275401906e-06, + "loss": 0.1409, + "step": 10701 + }, + { + "epoch": 4.48, + "learning_rate": 8.542076960648395e-06, + "loss": 0.1873, + "step": 10702 + }, + { + "epoch": 4.48, + "learning_rate": 8.540399687801588e-06, + "loss": 0.1732, + "step": 10703 + }, + { + "epoch": 4.48, + "learning_rate": 8.538722456909694e-06, + "loss": 0.1139, + "step": 10704 + }, + { + "epoch": 4.48, + "learning_rate": 8.53704526802093e-06, + "loss": 0.157, + "step": 10705 + }, + { + "epoch": 4.48, + "learning_rate": 8.5353681211835e-06, + "loss": 0.1486, + "step": 10706 + }, + { + "epoch": 4.48, + "learning_rate": 8.533691016445614e-06, + "loss": 0.1798, + "step": 10707 + }, + { + "epoch": 4.48, + "learning_rate": 8.532013953855477e-06, + "loss": 0.1528, + "step": 10708 + }, + { + "epoch": 4.48, + "learning_rate": 8.530336933461298e-06, + "loss": 0.1536, + "step": 10709 + }, + { + "epoch": 4.48, + "learning_rate": 8.52865995531128e-06, + "loss": 0.1856, + "step": 10710 + }, + { + "epoch": 4.48, + "learning_rate": 8.526983019453624e-06, + "loss": 0.1538, + "step": 10711 + }, + { + "epoch": 4.48, + "learning_rate": 8.52530612593654e-06, + "loss": 0.1494, + "step": 10712 + }, + { + "epoch": 4.49, + "learning_rate": 8.523629274808219e-06, + "loss": 0.1811, + "step": 10713 + }, + { + "epoch": 4.49, + "learning_rate": 8.521952466116862e-06, + "loss": 0.2089, + "step": 10714 + }, + { + "epoch": 4.49, + "learning_rate": 8.52027569991067e-06, + "loss": 0.1166, + "step": 10715 + }, + { + "epoch": 4.49, + "learning_rate": 8.51859897623784e-06, + "loss": 0.1852, + "step": 10716 + }, + { + "epoch": 4.49, + "learning_rate": 8.51692229514657e-06, + "loss": 0.1098, + "step": 10717 + }, + { + "epoch": 4.49, + "learning_rate": 8.515245656685051e-06, + "loss": 0.1062, + "step": 10718 + }, + { + "epoch": 4.49, + "learning_rate": 8.513569060901477e-06, + "loss": 0.1651, + "step": 10719 + }, + { + "epoch": 4.49, + "learning_rate": 8.511892507844045e-06, + "loss": 0.1643, + "step": 10720 + }, + { + "epoch": 4.49, + "learning_rate": 8.510215997560939e-06, + "loss": 0.1554, + "step": 10721 + }, + { + "epoch": 4.49, + "learning_rate": 8.508539530100352e-06, + "loss": 0.1887, + "step": 10722 + }, + { + "epoch": 4.49, + "learning_rate": 8.506863105510476e-06, + "loss": 0.1091, + "step": 10723 + }, + { + "epoch": 4.49, + "learning_rate": 8.505186723839492e-06, + "loss": 0.1308, + "step": 10724 + }, + { + "epoch": 4.49, + "learning_rate": 8.50351038513559e-06, + "loss": 0.1375, + "step": 10725 + }, + { + "epoch": 4.49, + "learning_rate": 8.501834089446956e-06, + "loss": 0.1762, + "step": 10726 + }, + { + "epoch": 4.49, + "learning_rate": 8.500157836821768e-06, + "loss": 0.1493, + "step": 10727 + }, + { + "epoch": 4.49, + "learning_rate": 8.498481627308215e-06, + "loss": 0.1416, + "step": 10728 + }, + { + "epoch": 4.49, + "learning_rate": 8.496805460954474e-06, + "loss": 0.1904, + "step": 10729 + }, + { + "epoch": 4.49, + "learning_rate": 8.495129337808725e-06, + "loss": 0.1275, + "step": 10730 + }, + { + "epoch": 4.49, + "learning_rate": 8.493453257919148e-06, + "loss": 0.1303, + "step": 10731 + }, + { + "epoch": 4.49, + "learning_rate": 8.491777221333923e-06, + "loss": 0.1428, + "step": 10732 + }, + { + "epoch": 4.49, + "learning_rate": 8.490101228101222e-06, + "loss": 0.1747, + "step": 10733 + }, + { + "epoch": 4.49, + "learning_rate": 8.488425278269224e-06, + "loss": 0.1431, + "step": 10734 + }, + { + "epoch": 4.49, + "learning_rate": 8.486749371886096e-06, + "loss": 0.1246, + "step": 10735 + }, + { + "epoch": 4.49, + "learning_rate": 8.485073509000015e-06, + "loss": 0.136, + "step": 10736 + }, + { + "epoch": 4.5, + "learning_rate": 8.483397689659153e-06, + "loss": 0.1523, + "step": 10737 + }, + { + "epoch": 4.5, + "learning_rate": 8.481721913911677e-06, + "loss": 0.1385, + "step": 10738 + }, + { + "epoch": 4.5, + "learning_rate": 8.480046181805757e-06, + "loss": 0.1603, + "step": 10739 + }, + { + "epoch": 4.5, + "learning_rate": 8.478370493389563e-06, + "loss": 0.1634, + "step": 10740 + }, + { + "epoch": 4.5, + "learning_rate": 8.476694848711258e-06, + "loss": 0.1531, + "step": 10741 + }, + { + "epoch": 4.5, + "learning_rate": 8.475019247819006e-06, + "loss": 0.1607, + "step": 10742 + }, + { + "epoch": 4.5, + "learning_rate": 8.473343690760976e-06, + "loss": 0.1643, + "step": 10743 + }, + { + "epoch": 4.5, + "learning_rate": 8.471668177585324e-06, + "loss": 0.1783, + "step": 10744 + }, + { + "epoch": 4.5, + "learning_rate": 8.469992708340213e-06, + "loss": 0.1064, + "step": 10745 + }, + { + "epoch": 4.5, + "learning_rate": 8.468317283073805e-06, + "loss": 0.1407, + "step": 10746 + }, + { + "epoch": 4.5, + "learning_rate": 8.46664190183426e-06, + "loss": 0.1761, + "step": 10747 + }, + { + "epoch": 4.5, + "learning_rate": 8.46496656466973e-06, + "loss": 0.143, + "step": 10748 + }, + { + "epoch": 4.5, + "learning_rate": 8.463291271628375e-06, + "loss": 0.1405, + "step": 10749 + }, + { + "epoch": 4.5, + "learning_rate": 8.46161602275835e-06, + "loss": 0.1192, + "step": 10750 + }, + { + "epoch": 4.5, + "learning_rate": 8.459940818107805e-06, + "loss": 0.1546, + "step": 10751 + }, + { + "epoch": 4.5, + "learning_rate": 8.458265657724894e-06, + "loss": 0.1785, + "step": 10752 + }, + { + "epoch": 4.5, + "learning_rate": 8.456590541657772e-06, + "loss": 0.1608, + "step": 10753 + }, + { + "epoch": 4.5, + "learning_rate": 8.454915469954583e-06, + "loss": 0.1609, + "step": 10754 + }, + { + "epoch": 4.5, + "learning_rate": 8.453240442663477e-06, + "loss": 0.1414, + "step": 10755 + }, + { + "epoch": 4.5, + "learning_rate": 8.451565459832604e-06, + "loss": 0.1398, + "step": 10756 + }, + { + "epoch": 4.5, + "learning_rate": 8.44989052151011e-06, + "loss": 0.1418, + "step": 10757 + }, + { + "epoch": 4.5, + "learning_rate": 8.448215627744134e-06, + "loss": 0.1435, + "step": 10758 + }, + { + "epoch": 4.5, + "learning_rate": 8.446540778582823e-06, + "loss": 0.1296, + "step": 10759 + }, + { + "epoch": 4.5, + "learning_rate": 8.444865974074319e-06, + "loss": 0.2264, + "step": 10760 + }, + { + "epoch": 4.51, + "learning_rate": 8.443191214266763e-06, + "loss": 0.1684, + "step": 10761 + }, + { + "epoch": 4.51, + "learning_rate": 8.441516499208292e-06, + "loss": 0.1324, + "step": 10762 + }, + { + "epoch": 4.51, + "learning_rate": 8.439841828947048e-06, + "loss": 0.1965, + "step": 10763 + }, + { + "epoch": 4.51, + "learning_rate": 8.438167203531166e-06, + "loss": 0.1786, + "step": 10764 + }, + { + "epoch": 4.51, + "learning_rate": 8.436492623008782e-06, + "loss": 0.1435, + "step": 10765 + }, + { + "epoch": 4.51, + "learning_rate": 8.434818087428029e-06, + "loss": 0.1287, + "step": 10766 + }, + { + "epoch": 4.51, + "learning_rate": 8.433143596837043e-06, + "loss": 0.1772, + "step": 10767 + }, + { + "epoch": 4.51, + "learning_rate": 8.431469151283953e-06, + "loss": 0.1788, + "step": 10768 + }, + { + "epoch": 4.51, + "learning_rate": 8.429794750816888e-06, + "loss": 0.1539, + "step": 10769 + }, + { + "epoch": 4.51, + "learning_rate": 8.428120395483982e-06, + "loss": 0.14, + "step": 10770 + }, + { + "epoch": 4.51, + "learning_rate": 8.42644608533336e-06, + "loss": 0.1312, + "step": 10771 + }, + { + "epoch": 4.51, + "learning_rate": 8.42477182041315e-06, + "loss": 0.1656, + "step": 10772 + }, + { + "epoch": 4.51, + "learning_rate": 8.423097600771476e-06, + "loss": 0.1824, + "step": 10773 + }, + { + "epoch": 4.51, + "learning_rate": 8.42142342645646e-06, + "loss": 0.1685, + "step": 10774 + }, + { + "epoch": 4.51, + "learning_rate": 8.419749297516227e-06, + "loss": 0.1578, + "step": 10775 + }, + { + "epoch": 4.51, + "learning_rate": 8.4180752139989e-06, + "loss": 0.1387, + "step": 10776 + }, + { + "epoch": 4.51, + "learning_rate": 8.416401175952598e-06, + "loss": 0.1427, + "step": 10777 + }, + { + "epoch": 4.51, + "learning_rate": 8.414727183425437e-06, + "loss": 0.1592, + "step": 10778 + }, + { + "epoch": 4.51, + "learning_rate": 8.413053236465535e-06, + "loss": 0.1332, + "step": 10779 + }, + { + "epoch": 4.51, + "learning_rate": 8.411379335121014e-06, + "loss": 0.1034, + "step": 10780 + }, + { + "epoch": 4.51, + "learning_rate": 8.409705479439985e-06, + "loss": 0.185, + "step": 10781 + }, + { + "epoch": 4.51, + "learning_rate": 8.408031669470555e-06, + "loss": 0.1347, + "step": 10782 + }, + { + "epoch": 4.51, + "learning_rate": 8.406357905260845e-06, + "loss": 0.1821, + "step": 10783 + }, + { + "epoch": 4.51, + "learning_rate": 8.404684186858962e-06, + "loss": 0.1364, + "step": 10784 + }, + { + "epoch": 4.52, + "learning_rate": 8.403010514313016e-06, + "loss": 0.1522, + "step": 10785 + }, + { + "epoch": 4.52, + "learning_rate": 8.401336887671117e-06, + "loss": 0.1601, + "step": 10786 + }, + { + "epoch": 4.52, + "learning_rate": 8.399663306981368e-06, + "loss": 0.1565, + "step": 10787 + }, + { + "epoch": 4.52, + "learning_rate": 8.39798977229188e-06, + "loss": 0.1307, + "step": 10788 + }, + { + "epoch": 4.52, + "learning_rate": 8.396316283650753e-06, + "loss": 0.1278, + "step": 10789 + }, + { + "epoch": 4.52, + "learning_rate": 8.39464284110609e-06, + "loss": 0.1808, + "step": 10790 + }, + { + "epoch": 4.52, + "learning_rate": 8.392969444705998e-06, + "loss": 0.1575, + "step": 10791 + }, + { + "epoch": 4.52, + "learning_rate": 8.39129609449857e-06, + "loss": 0.163, + "step": 10792 + }, + { + "epoch": 4.52, + "learning_rate": 8.389622790531908e-06, + "loss": 0.1786, + "step": 10793 + }, + { + "epoch": 4.52, + "learning_rate": 8.387949532854113e-06, + "loss": 0.1419, + "step": 10794 + }, + { + "epoch": 4.52, + "learning_rate": 8.386276321513275e-06, + "loss": 0.1721, + "step": 10795 + }, + { + "epoch": 4.52, + "learning_rate": 8.384603156557492e-06, + "loss": 0.145, + "step": 10796 + }, + { + "epoch": 4.52, + "learning_rate": 8.38293003803486e-06, + "loss": 0.1321, + "step": 10797 + }, + { + "epoch": 4.52, + "learning_rate": 8.381256965993468e-06, + "loss": 0.1727, + "step": 10798 + }, + { + "epoch": 4.52, + "learning_rate": 8.379583940481407e-06, + "loss": 0.1064, + "step": 10799 + }, + { + "epoch": 4.52, + "learning_rate": 8.377910961546771e-06, + "loss": 0.1787, + "step": 10800 + }, + { + "epoch": 4.52, + "learning_rate": 8.376238029237643e-06, + "loss": 0.1235, + "step": 10801 + }, + { + "epoch": 4.52, + "learning_rate": 8.374565143602114e-06, + "loss": 0.1486, + "step": 10802 + }, + { + "epoch": 4.52, + "learning_rate": 8.372892304688265e-06, + "loss": 0.1193, + "step": 10803 + }, + { + "epoch": 4.52, + "learning_rate": 8.371219512544182e-06, + "loss": 0.1624, + "step": 10804 + }, + { + "epoch": 4.52, + "learning_rate": 8.369546767217952e-06, + "loss": 0.1455, + "step": 10805 + }, + { + "epoch": 4.52, + "learning_rate": 8.367874068757651e-06, + "loss": 0.1264, + "step": 10806 + }, + { + "epoch": 4.52, + "learning_rate": 8.366201417211361e-06, + "loss": 0.171, + "step": 10807 + }, + { + "epoch": 4.53, + "learning_rate": 8.364528812627164e-06, + "loss": 0.1191, + "step": 10808 + }, + { + "epoch": 4.53, + "learning_rate": 8.362856255053132e-06, + "loss": 0.1946, + "step": 10809 + }, + { + "epoch": 4.53, + "learning_rate": 8.361183744537345e-06, + "loss": 0.1596, + "step": 10810 + }, + { + "epoch": 4.53, + "learning_rate": 8.359511281127879e-06, + "loss": 0.2023, + "step": 10811 + }, + { + "epoch": 4.53, + "learning_rate": 8.357838864872803e-06, + "loss": 0.1452, + "step": 10812 + }, + { + "epoch": 4.53, + "learning_rate": 8.356166495820192e-06, + "loss": 0.1487, + "step": 10813 + }, + { + "epoch": 4.53, + "learning_rate": 8.354494174018118e-06, + "loss": 0.1538, + "step": 10814 + }, + { + "epoch": 4.53, + "learning_rate": 8.352821899514648e-06, + "loss": 0.1443, + "step": 10815 + }, + { + "epoch": 4.53, + "learning_rate": 8.35114967235785e-06, + "loss": 0.1427, + "step": 10816 + }, + { + "epoch": 4.53, + "learning_rate": 8.34947749259579e-06, + "loss": 0.1366, + "step": 10817 + }, + { + "epoch": 4.53, + "learning_rate": 8.347805360276538e-06, + "loss": 0.157, + "step": 10818 + }, + { + "epoch": 4.53, + "learning_rate": 8.346133275448153e-06, + "loss": 0.1599, + "step": 10819 + }, + { + "epoch": 4.53, + "learning_rate": 8.3444612381587e-06, + "loss": 0.1406, + "step": 10820 + }, + { + "epoch": 4.53, + "learning_rate": 8.342789248456242e-06, + "loss": 0.1796, + "step": 10821 + }, + { + "epoch": 4.53, + "learning_rate": 8.341117306388833e-06, + "loss": 0.15, + "step": 10822 + }, + { + "epoch": 4.53, + "learning_rate": 8.339445412004533e-06, + "loss": 0.1105, + "step": 10823 + }, + { + "epoch": 4.53, + "learning_rate": 8.337773565351407e-06, + "loss": 0.153, + "step": 10824 + }, + { + "epoch": 4.53, + "learning_rate": 8.336101766477506e-06, + "loss": 0.1288, + "step": 10825 + }, + { + "epoch": 4.53, + "learning_rate": 8.334430015430878e-06, + "loss": 0.1934, + "step": 10826 + }, + { + "epoch": 4.53, + "learning_rate": 8.332758312259583e-06, + "loss": 0.1263, + "step": 10827 + }, + { + "epoch": 4.53, + "learning_rate": 8.331086657011669e-06, + "loss": 0.1337, + "step": 10828 + }, + { + "epoch": 4.53, + "learning_rate": 8.32941504973519e-06, + "loss": 0.1528, + "step": 10829 + }, + { + "epoch": 4.53, + "learning_rate": 8.327743490478192e-06, + "loss": 0.1801, + "step": 10830 + }, + { + "epoch": 4.53, + "learning_rate": 8.326071979288723e-06, + "loss": 0.1432, + "step": 10831 + }, + { + "epoch": 4.54, + "learning_rate": 8.32440051621483e-06, + "loss": 0.1296, + "step": 10832 + }, + { + "epoch": 4.54, + "learning_rate": 8.322729101304558e-06, + "loss": 0.1795, + "step": 10833 + }, + { + "epoch": 4.54, + "learning_rate": 8.321057734605949e-06, + "loss": 0.1599, + "step": 10834 + }, + { + "epoch": 4.54, + "learning_rate": 8.319386416167047e-06, + "loss": 0.1965, + "step": 10835 + }, + { + "epoch": 4.54, + "learning_rate": 8.31771514603589e-06, + "loss": 0.1628, + "step": 10836 + }, + { + "epoch": 4.54, + "learning_rate": 8.316043924260514e-06, + "loss": 0.1537, + "step": 10837 + }, + { + "epoch": 4.54, + "learning_rate": 8.314372750888968e-06, + "loss": 0.1814, + "step": 10838 + }, + { + "epoch": 4.54, + "learning_rate": 8.312701625969278e-06, + "loss": 0.1303, + "step": 10839 + }, + { + "epoch": 4.54, + "learning_rate": 8.311030549549483e-06, + "loss": 0.1532, + "step": 10840 + }, + { + "epoch": 4.54, + "learning_rate": 8.309359521677617e-06, + "loss": 0.1477, + "step": 10841 + }, + { + "epoch": 4.54, + "learning_rate": 8.307688542401709e-06, + "loss": 0.1307, + "step": 10842 + }, + { + "epoch": 4.54, + "learning_rate": 8.306017611769795e-06, + "loss": 0.122, + "step": 10843 + }, + { + "epoch": 4.54, + "learning_rate": 8.3043467298299e-06, + "loss": 0.1751, + "step": 10844 + }, + { + "epoch": 4.54, + "learning_rate": 8.302675896630054e-06, + "loss": 0.1655, + "step": 10845 + }, + { + "epoch": 4.54, + "learning_rate": 8.301005112218282e-06, + "loss": 0.1444, + "step": 10846 + }, + { + "epoch": 4.54, + "learning_rate": 8.299334376642609e-06, + "loss": 0.1179, + "step": 10847 + }, + { + "epoch": 4.54, + "learning_rate": 8.297663689951067e-06, + "loss": 0.1252, + "step": 10848 + }, + { + "epoch": 4.54, + "learning_rate": 8.295993052191669e-06, + "loss": 0.1143, + "step": 10849 + }, + { + "epoch": 4.54, + "learning_rate": 8.294322463412435e-06, + "loss": 0.132, + "step": 10850 + }, + { + "epoch": 4.54, + "learning_rate": 8.292651923661391e-06, + "loss": 0.1182, + "step": 10851 + }, + { + "epoch": 4.54, + "learning_rate": 8.290981432986551e-06, + "loss": 0.1625, + "step": 10852 + }, + { + "epoch": 4.54, + "learning_rate": 8.289310991435933e-06, + "loss": 0.1747, + "step": 10853 + }, + { + "epoch": 4.54, + "learning_rate": 8.287640599057551e-06, + "loss": 0.162, + "step": 10854 + }, + { + "epoch": 4.54, + "learning_rate": 8.285970255899425e-06, + "loss": 0.1723, + "step": 10855 + }, + { + "epoch": 4.55, + "learning_rate": 8.284299962009559e-06, + "loss": 0.1552, + "step": 10856 + }, + { + "epoch": 4.55, + "learning_rate": 8.28262971743597e-06, + "loss": 0.1625, + "step": 10857 + }, + { + "epoch": 4.55, + "learning_rate": 8.280959522226663e-06, + "loss": 0.1036, + "step": 10858 + }, + { + "epoch": 4.55, + "learning_rate": 8.279289376429653e-06, + "loss": 0.1547, + "step": 10859 + }, + { + "epoch": 4.55, + "learning_rate": 8.27761928009294e-06, + "loss": 0.1245, + "step": 10860 + }, + { + "epoch": 4.55, + "learning_rate": 8.275949233264534e-06, + "loss": 0.1447, + "step": 10861 + }, + { + "epoch": 4.55, + "learning_rate": 8.274279235992438e-06, + "loss": 0.1469, + "step": 10862 + }, + { + "epoch": 4.55, + "learning_rate": 8.272609288324653e-06, + "loss": 0.1636, + "step": 10863 + }, + { + "epoch": 4.55, + "learning_rate": 8.270939390309181e-06, + "loss": 0.2064, + "step": 10864 + }, + { + "epoch": 4.55, + "learning_rate": 8.269269541994022e-06, + "loss": 0.1293, + "step": 10865 + }, + { + "epoch": 4.55, + "learning_rate": 8.267599743427173e-06, + "loss": 0.126, + "step": 10866 + }, + { + "epoch": 4.55, + "learning_rate": 8.265929994656634e-06, + "loss": 0.1713, + "step": 10867 + }, + { + "epoch": 4.55, + "learning_rate": 8.2642602957304e-06, + "loss": 0.1381, + "step": 10868 + }, + { + "epoch": 4.55, + "learning_rate": 8.26259064669646e-06, + "loss": 0.0982, + "step": 10869 + }, + { + "epoch": 4.55, + "learning_rate": 8.260921047602814e-06, + "loss": 0.1539, + "step": 10870 + }, + { + "epoch": 4.55, + "learning_rate": 8.259251498497446e-06, + "loss": 0.1549, + "step": 10871 + }, + { + "epoch": 4.55, + "learning_rate": 8.25758199942835e-06, + "loss": 0.1301, + "step": 10872 + }, + { + "epoch": 4.55, + "learning_rate": 8.255912550443515e-06, + "loss": 0.133, + "step": 10873 + }, + { + "epoch": 4.55, + "learning_rate": 8.254243151590922e-06, + "loss": 0.195, + "step": 10874 + }, + { + "epoch": 4.55, + "learning_rate": 8.252573802918562e-06, + "loss": 0.1498, + "step": 10875 + }, + { + "epoch": 4.55, + "learning_rate": 8.25090450447442e-06, + "loss": 0.173, + "step": 10876 + }, + { + "epoch": 4.55, + "learning_rate": 8.249235256306472e-06, + "loss": 0.1879, + "step": 10877 + }, + { + "epoch": 4.55, + "learning_rate": 8.247566058462704e-06, + "loss": 0.1443, + "step": 10878 + }, + { + "epoch": 4.55, + "learning_rate": 8.245896910991096e-06, + "loss": 0.1435, + "step": 10879 + }, + { + "epoch": 4.56, + "learning_rate": 8.244227813939624e-06, + "loss": 0.1807, + "step": 10880 + }, + { + "epoch": 4.56, + "learning_rate": 8.242558767356266e-06, + "loss": 0.1714, + "step": 10881 + }, + { + "epoch": 4.56, + "learning_rate": 8.240889771288997e-06, + "loss": 0.1307, + "step": 10882 + }, + { + "epoch": 4.56, + "learning_rate": 8.239220825785789e-06, + "loss": 0.1448, + "step": 10883 + }, + { + "epoch": 4.56, + "learning_rate": 8.237551930894616e-06, + "loss": 0.1597, + "step": 10884 + }, + { + "epoch": 4.56, + "learning_rate": 8.23588308666345e-06, + "loss": 0.1435, + "step": 10885 + }, + { + "epoch": 4.56, + "learning_rate": 8.234214293140258e-06, + "loss": 0.1298, + "step": 10886 + }, + { + "epoch": 4.56, + "learning_rate": 8.23254555037301e-06, + "loss": 0.1396, + "step": 10887 + }, + { + "epoch": 4.56, + "learning_rate": 8.230876858409669e-06, + "loss": 0.1867, + "step": 10888 + }, + { + "epoch": 4.56, + "learning_rate": 8.229208217298208e-06, + "loss": 0.149, + "step": 10889 + }, + { + "epoch": 4.56, + "learning_rate": 8.227539627086582e-06, + "loss": 0.1266, + "step": 10890 + }, + { + "epoch": 4.56, + "learning_rate": 8.225871087822757e-06, + "loss": 0.1307, + "step": 10891 + }, + { + "epoch": 4.56, + "learning_rate": 8.224202599554695e-06, + "loss": 0.1753, + "step": 10892 + }, + { + "epoch": 4.56, + "learning_rate": 8.222534162330355e-06, + "loss": 0.1343, + "step": 10893 + }, + { + "epoch": 4.56, + "learning_rate": 8.22086577619769e-06, + "loss": 0.145, + "step": 10894 + }, + { + "epoch": 4.56, + "learning_rate": 8.219197441204659e-06, + "loss": 0.1279, + "step": 10895 + }, + { + "epoch": 4.56, + "learning_rate": 8.21752915739922e-06, + "loss": 0.1412, + "step": 10896 + }, + { + "epoch": 4.56, + "learning_rate": 8.21586092482932e-06, + "loss": 0.1737, + "step": 10897 + }, + { + "epoch": 4.56, + "learning_rate": 8.214192743542914e-06, + "loss": 0.1382, + "step": 10898 + }, + { + "epoch": 4.56, + "learning_rate": 8.212524613587955e-06, + "loss": 0.14, + "step": 10899 + }, + { + "epoch": 4.56, + "learning_rate": 8.210856535012392e-06, + "loss": 0.1652, + "step": 10900 + }, + { + "epoch": 4.56, + "learning_rate": 8.209188507864167e-06, + "loss": 0.1024, + "step": 10901 + }, + { + "epoch": 4.56, + "learning_rate": 8.20752053219123e-06, + "loss": 0.2096, + "step": 10902 + }, + { + "epoch": 4.56, + "learning_rate": 8.205852608041526e-06, + "loss": 0.1255, + "step": 10903 + }, + { + "epoch": 4.57, + "learning_rate": 8.204184735462994e-06, + "loss": 0.1135, + "step": 10904 + }, + { + "epoch": 4.57, + "learning_rate": 8.20251691450358e-06, + "loss": 0.1712, + "step": 10905 + }, + { + "epoch": 4.57, + "learning_rate": 8.200849145211224e-06, + "loss": 0.1354, + "step": 10906 + }, + { + "epoch": 4.57, + "learning_rate": 8.19918142763386e-06, + "loss": 0.14, + "step": 10907 + }, + { + "epoch": 4.57, + "learning_rate": 8.197513761819428e-06, + "loss": 0.1466, + "step": 10908 + }, + { + "epoch": 4.57, + "learning_rate": 8.195846147815865e-06, + "loss": 0.1407, + "step": 10909 + }, + { + "epoch": 4.57, + "learning_rate": 8.194178585671102e-06, + "loss": 0.1501, + "step": 10910 + }, + { + "epoch": 4.57, + "learning_rate": 8.192511075433073e-06, + "loss": 0.1264, + "step": 10911 + }, + { + "epoch": 4.57, + "learning_rate": 8.190843617149712e-06, + "loss": 0.181, + "step": 10912 + }, + { + "epoch": 4.57, + "learning_rate": 8.189176210868942e-06, + "loss": 0.1677, + "step": 10913 + }, + { + "epoch": 4.57, + "learning_rate": 8.187508856638697e-06, + "loss": 0.174, + "step": 10914 + }, + { + "epoch": 4.57, + "learning_rate": 8.185841554506901e-06, + "loss": 0.1589, + "step": 10915 + }, + { + "epoch": 4.57, + "learning_rate": 8.184174304521483e-06, + "loss": 0.1424, + "step": 10916 + }, + { + "epoch": 4.57, + "learning_rate": 8.18250710673036e-06, + "loss": 0.1473, + "step": 10917 + }, + { + "epoch": 4.57, + "learning_rate": 8.180839961181458e-06, + "loss": 0.1487, + "step": 10918 + }, + { + "epoch": 4.57, + "learning_rate": 8.179172867922695e-06, + "loss": 0.1064, + "step": 10919 + }, + { + "epoch": 4.57, + "learning_rate": 8.177505827001997e-06, + "loss": 0.1629, + "step": 10920 + }, + { + "epoch": 4.57, + "learning_rate": 8.175838838467272e-06, + "loss": 0.168, + "step": 10921 + }, + { + "epoch": 4.57, + "learning_rate": 8.174171902366441e-06, + "loss": 0.1765, + "step": 10922 + }, + { + "epoch": 4.57, + "learning_rate": 8.172505018747421e-06, + "loss": 0.1488, + "step": 10923 + }, + { + "epoch": 4.57, + "learning_rate": 8.17083818765812e-06, + "loss": 0.1875, + "step": 10924 + }, + { + "epoch": 4.57, + "learning_rate": 8.169171409146453e-06, + "loss": 0.1606, + "step": 10925 + }, + { + "epoch": 4.57, + "learning_rate": 8.167504683260332e-06, + "loss": 0.1725, + "step": 10926 + }, + { + "epoch": 4.57, + "learning_rate": 8.16583801004766e-06, + "loss": 0.129, + "step": 10927 + }, + { + "epoch": 4.58, + "learning_rate": 8.164171389556346e-06, + "loss": 0.1683, + "step": 10928 + }, + { + "epoch": 4.58, + "learning_rate": 8.162504821834296e-06, + "loss": 0.1483, + "step": 10929 + }, + { + "epoch": 4.58, + "learning_rate": 8.160838306929418e-06, + "loss": 0.1661, + "step": 10930 + }, + { + "epoch": 4.58, + "learning_rate": 8.159171844889609e-06, + "loss": 0.149, + "step": 10931 + }, + { + "epoch": 4.58, + "learning_rate": 8.157505435762769e-06, + "loss": 0.161, + "step": 10932 + }, + { + "epoch": 4.58, + "learning_rate": 8.155839079596805e-06, + "loss": 0.1306, + "step": 10933 + }, + { + "epoch": 4.58, + "learning_rate": 8.154172776439608e-06, + "loss": 0.1119, + "step": 10934 + }, + { + "epoch": 4.58, + "learning_rate": 8.152506526339075e-06, + "loss": 0.1548, + "step": 10935 + }, + { + "epoch": 4.58, + "learning_rate": 8.150840329343104e-06, + "loss": 0.1768, + "step": 10936 + }, + { + "epoch": 4.58, + "learning_rate": 8.149174185499587e-06, + "loss": 0.1677, + "step": 10937 + }, + { + "epoch": 4.58, + "learning_rate": 8.14750809485642e-06, + "loss": 0.127, + "step": 10938 + }, + { + "epoch": 4.58, + "learning_rate": 8.145842057461483e-06, + "loss": 0.1572, + "step": 10939 + }, + { + "epoch": 4.58, + "learning_rate": 8.14417607336267e-06, + "loss": 0.1449, + "step": 10940 + }, + { + "epoch": 4.58, + "learning_rate": 8.142510142607874e-06, + "loss": 0.1592, + "step": 10941 + }, + { + "epoch": 4.58, + "learning_rate": 8.140844265244971e-06, + "loss": 0.1251, + "step": 10942 + }, + { + "epoch": 4.58, + "learning_rate": 8.139178441321851e-06, + "loss": 0.1163, + "step": 10943 + }, + { + "epoch": 4.58, + "learning_rate": 8.137512670886397e-06, + "loss": 0.1183, + "step": 10944 + }, + { + "epoch": 4.58, + "learning_rate": 8.135846953986486e-06, + "loss": 0.1977, + "step": 10945 + }, + { + "epoch": 4.58, + "learning_rate": 8.134181290670002e-06, + "loss": 0.1227, + "step": 10946 + }, + { + "epoch": 4.58, + "learning_rate": 8.132515680984821e-06, + "loss": 0.1084, + "step": 10947 + }, + { + "epoch": 4.58, + "learning_rate": 8.130850124978819e-06, + "loss": 0.163, + "step": 10948 + }, + { + "epoch": 4.58, + "learning_rate": 8.12918462269987e-06, + "loss": 0.1608, + "step": 10949 + }, + { + "epoch": 4.58, + "learning_rate": 8.127519174195853e-06, + "loss": 0.1399, + "step": 10950 + }, + { + "epoch": 4.58, + "learning_rate": 8.125853779514632e-06, + "loss": 0.1631, + "step": 10951 + }, + { + "epoch": 4.59, + "learning_rate": 8.124188438704082e-06, + "loss": 0.1099, + "step": 10952 + }, + { + "epoch": 4.59, + "learning_rate": 8.122523151812072e-06, + "loss": 0.1351, + "step": 10953 + }, + { + "epoch": 4.59, + "learning_rate": 8.120857918886467e-06, + "loss": 0.1401, + "step": 10954 + }, + { + "epoch": 4.59, + "learning_rate": 8.119192739975133e-06, + "loss": 0.1397, + "step": 10955 + }, + { + "epoch": 4.59, + "learning_rate": 8.117527615125938e-06, + "loss": 0.1302, + "step": 10956 + }, + { + "epoch": 4.59, + "learning_rate": 8.115862544386736e-06, + "loss": 0.1344, + "step": 10957 + }, + { + "epoch": 4.59, + "learning_rate": 8.114197527805396e-06, + "loss": 0.1636, + "step": 10958 + }, + { + "epoch": 4.59, + "learning_rate": 8.112532565429774e-06, + "loss": 0.1649, + "step": 10959 + }, + { + "epoch": 4.59, + "learning_rate": 8.11086765730773e-06, + "loss": 0.1666, + "step": 10960 + }, + { + "epoch": 4.59, + "learning_rate": 8.109202803487121e-06, + "loss": 0.1521, + "step": 10961 + }, + { + "epoch": 4.59, + "learning_rate": 8.107538004015796e-06, + "loss": 0.1471, + "step": 10962 + }, + { + "epoch": 4.59, + "learning_rate": 8.105873258941612e-06, + "loss": 0.1451, + "step": 10963 + }, + { + "epoch": 4.59, + "learning_rate": 8.10420856831242e-06, + "loss": 0.1674, + "step": 10964 + }, + { + "epoch": 4.59, + "learning_rate": 8.10254393217607e-06, + "loss": 0.1801, + "step": 10965 + }, + { + "epoch": 4.59, + "learning_rate": 8.100879350580411e-06, + "loss": 0.1484, + "step": 10966 + }, + { + "epoch": 4.59, + "learning_rate": 8.099214823573293e-06, + "loss": 0.1657, + "step": 10967 + }, + { + "epoch": 4.59, + "learning_rate": 8.097550351202554e-06, + "loss": 0.179, + "step": 10968 + }, + { + "epoch": 4.59, + "learning_rate": 8.095885933516043e-06, + "loss": 0.1169, + "step": 10969 + }, + { + "epoch": 4.59, + "learning_rate": 8.094221570561602e-06, + "loss": 0.1453, + "step": 10970 + }, + { + "epoch": 4.59, + "learning_rate": 8.09255726238707e-06, + "loss": 0.1403, + "step": 10971 + }, + { + "epoch": 4.59, + "learning_rate": 8.090893009040288e-06, + "loss": 0.1554, + "step": 10972 + }, + { + "epoch": 4.59, + "learning_rate": 8.08922881056909e-06, + "loss": 0.1699, + "step": 10973 + }, + { + "epoch": 4.59, + "learning_rate": 8.087564667021317e-06, + "loss": 0.1783, + "step": 10974 + }, + { + "epoch": 4.59, + "learning_rate": 8.0859005784448e-06, + "loss": 0.1717, + "step": 10975 + }, + { + "epoch": 4.6, + "learning_rate": 8.084236544887372e-06, + "loss": 0.1571, + "step": 10976 + }, + { + "epoch": 4.6, + "learning_rate": 8.082572566396864e-06, + "loss": 0.1587, + "step": 10977 + }, + { + "epoch": 4.6, + "learning_rate": 8.080908643021107e-06, + "loss": 0.1206, + "step": 10978 + }, + { + "epoch": 4.6, + "learning_rate": 8.079244774807926e-06, + "loss": 0.1667, + "step": 10979 + }, + { + "epoch": 4.6, + "learning_rate": 8.077580961805153e-06, + "loss": 0.1552, + "step": 10980 + }, + { + "epoch": 4.6, + "learning_rate": 8.075917204060606e-06, + "loss": 0.1245, + "step": 10981 + }, + { + "epoch": 4.6, + "learning_rate": 8.074253501622113e-06, + "loss": 0.1779, + "step": 10982 + }, + { + "epoch": 4.6, + "learning_rate": 8.072589854537498e-06, + "loss": 0.1806, + "step": 10983 + }, + { + "epoch": 4.6, + "learning_rate": 8.070926262854573e-06, + "loss": 0.1765, + "step": 10984 + }, + { + "epoch": 4.6, + "learning_rate": 8.069262726621163e-06, + "loss": 0.1469, + "step": 10985 + }, + { + "epoch": 4.6, + "learning_rate": 8.06759924588508e-06, + "loss": 0.1461, + "step": 10986 + }, + { + "epoch": 4.6, + "learning_rate": 8.065935820694142e-06, + "loss": 0.1394, + "step": 10987 + }, + { + "epoch": 4.6, + "learning_rate": 8.064272451096165e-06, + "loss": 0.1073, + "step": 10988 + }, + { + "epoch": 4.6, + "learning_rate": 8.062609137138955e-06, + "loss": 0.126, + "step": 10989 + }, + { + "epoch": 4.6, + "learning_rate": 8.060945878870328e-06, + "loss": 0.1782, + "step": 10990 + }, + { + "epoch": 4.6, + "learning_rate": 8.059282676338092e-06, + "loss": 0.1596, + "step": 10991 + }, + { + "epoch": 4.6, + "learning_rate": 8.057619529590051e-06, + "loss": 0.138, + "step": 10992 + }, + { + "epoch": 4.6, + "learning_rate": 8.055956438674015e-06, + "loss": 0.1796, + "step": 10993 + }, + { + "epoch": 4.6, + "learning_rate": 8.054293403637787e-06, + "loss": 0.153, + "step": 10994 + }, + { + "epoch": 4.6, + "learning_rate": 8.052630424529166e-06, + "loss": 0.1554, + "step": 10995 + }, + { + "epoch": 4.6, + "learning_rate": 8.050967501395955e-06, + "loss": 0.1505, + "step": 10996 + }, + { + "epoch": 4.6, + "learning_rate": 8.049304634285957e-06, + "loss": 0.1279, + "step": 10997 + }, + { + "epoch": 4.6, + "learning_rate": 8.047641823246964e-06, + "loss": 0.1643, + "step": 10998 + }, + { + "epoch": 4.6, + "learning_rate": 8.045979068326773e-06, + "loss": 0.1534, + "step": 10999 + }, + { + "epoch": 4.61, + "learning_rate": 8.044316369573181e-06, + "loss": 0.1411, + "step": 11000 + }, + { + "epoch": 4.61, + "learning_rate": 8.042653727033981e-06, + "loss": 0.1839, + "step": 11001 + }, + { + "epoch": 4.61, + "learning_rate": 8.04099114075696e-06, + "loss": 0.158, + "step": 11002 + }, + { + "epoch": 4.61, + "learning_rate": 8.039328610789911e-06, + "loss": 0.1694, + "step": 11003 + }, + { + "epoch": 4.61, + "learning_rate": 8.037666137180624e-06, + "loss": 0.1096, + "step": 11004 + }, + { + "epoch": 4.61, + "learning_rate": 8.03600371997688e-06, + "loss": 0.1871, + "step": 11005 + }, + { + "epoch": 4.61, + "learning_rate": 8.03434135922647e-06, + "loss": 0.12, + "step": 11006 + }, + { + "epoch": 4.61, + "learning_rate": 8.032679054977171e-06, + "loss": 0.1471, + "step": 11007 + }, + { + "epoch": 4.61, + "learning_rate": 8.031016807276765e-06, + "loss": 0.1441, + "step": 11008 + }, + { + "epoch": 4.61, + "learning_rate": 8.029354616173036e-06, + "loss": 0.1719, + "step": 11009 + }, + { + "epoch": 4.61, + "learning_rate": 8.027692481713757e-06, + "loss": 0.1434, + "step": 11010 + }, + { + "epoch": 4.61, + "learning_rate": 8.026030403946707e-06, + "loss": 0.1538, + "step": 11011 + }, + { + "epoch": 4.61, + "learning_rate": 8.024368382919665e-06, + "loss": 0.1316, + "step": 11012 + }, + { + "epoch": 4.61, + "learning_rate": 8.022706418680398e-06, + "loss": 0.1592, + "step": 11013 + }, + { + "epoch": 4.61, + "learning_rate": 8.021044511276682e-06, + "loss": 0.1634, + "step": 11014 + }, + { + "epoch": 4.61, + "learning_rate": 8.019382660756284e-06, + "loss": 0.1346, + "step": 11015 + }, + { + "epoch": 4.61, + "learning_rate": 8.017720867166976e-06, + "loss": 0.142, + "step": 11016 + }, + { + "epoch": 4.61, + "learning_rate": 8.01605913055652e-06, + "loss": 0.1682, + "step": 11017 + }, + { + "epoch": 4.61, + "learning_rate": 8.014397450972685e-06, + "loss": 0.156, + "step": 11018 + }, + { + "epoch": 4.61, + "learning_rate": 8.012735828463233e-06, + "loss": 0.1653, + "step": 11019 + }, + { + "epoch": 4.61, + "learning_rate": 8.011074263075925e-06, + "loss": 0.1269, + "step": 11020 + }, + { + "epoch": 4.61, + "learning_rate": 8.009412754858525e-06, + "loss": 0.1328, + "step": 11021 + }, + { + "epoch": 4.61, + "learning_rate": 8.007751303858788e-06, + "loss": 0.1473, + "step": 11022 + }, + { + "epoch": 4.62, + "learning_rate": 8.00608991012447e-06, + "loss": 0.108, + "step": 11023 + }, + { + "epoch": 4.62, + "learning_rate": 8.00442857370333e-06, + "loss": 0.1977, + "step": 11024 + }, + { + "epoch": 4.62, + "learning_rate": 8.00276729464312e-06, + "loss": 0.1484, + "step": 11025 + }, + { + "epoch": 4.62, + "learning_rate": 8.00110607299159e-06, + "loss": 0.1721, + "step": 11026 + }, + { + "epoch": 4.62, + "learning_rate": 7.999444908796493e-06, + "loss": 0.1437, + "step": 11027 + }, + { + "epoch": 4.62, + "learning_rate": 7.99778380210558e-06, + "loss": 0.132, + "step": 11028 + }, + { + "epoch": 4.62, + "learning_rate": 7.996122752966596e-06, + "loss": 0.148, + "step": 11029 + }, + { + "epoch": 4.62, + "learning_rate": 7.99446176142728e-06, + "loss": 0.1464, + "step": 11030 + }, + { + "epoch": 4.62, + "learning_rate": 7.992800827535381e-06, + "loss": 0.1883, + "step": 11031 + }, + { + "epoch": 4.62, + "learning_rate": 7.991139951338645e-06, + "loss": 0.1545, + "step": 11032 + }, + { + "epoch": 4.62, + "learning_rate": 7.989479132884803e-06, + "loss": 0.1339, + "step": 11033 + }, + { + "epoch": 4.62, + "learning_rate": 7.987818372221603e-06, + "loss": 0.1387, + "step": 11034 + }, + { + "epoch": 4.62, + "learning_rate": 7.98615766939678e-06, + "loss": 0.1584, + "step": 11035 + }, + { + "epoch": 4.62, + "learning_rate": 7.984497024458062e-06, + "loss": 0.128, + "step": 11036 + }, + { + "epoch": 4.62, + "learning_rate": 7.982836437453192e-06, + "loss": 0.1636, + "step": 11037 + }, + { + "epoch": 4.62, + "learning_rate": 7.9811759084299e-06, + "loss": 0.1627, + "step": 11038 + }, + { + "epoch": 4.62, + "learning_rate": 7.979515437435913e-06, + "loss": 0.1573, + "step": 11039 + }, + { + "epoch": 4.62, + "learning_rate": 7.977855024518962e-06, + "loss": 0.1787, + "step": 11040 + }, + { + "epoch": 4.62, + "learning_rate": 7.976194669726773e-06, + "loss": 0.1583, + "step": 11041 + }, + { + "epoch": 4.62, + "learning_rate": 7.974534373107074e-06, + "loss": 0.1613, + "step": 11042 + }, + { + "epoch": 4.62, + "learning_rate": 7.972874134707585e-06, + "loss": 0.1165, + "step": 11043 + }, + { + "epoch": 4.62, + "learning_rate": 7.971213954576031e-06, + "loss": 0.1546, + "step": 11044 + }, + { + "epoch": 4.62, + "learning_rate": 7.969553832760134e-06, + "loss": 0.1246, + "step": 11045 + }, + { + "epoch": 4.62, + "learning_rate": 7.967893769307608e-06, + "loss": 0.1284, + "step": 11046 + }, + { + "epoch": 4.63, + "learning_rate": 7.966233764266172e-06, + "loss": 0.1343, + "step": 11047 + }, + { + "epoch": 4.63, + "learning_rate": 7.964573817683545e-06, + "loss": 0.1521, + "step": 11048 + }, + { + "epoch": 4.63, + "learning_rate": 7.962913929607433e-06, + "loss": 0.1585, + "step": 11049 + }, + { + "epoch": 4.63, + "learning_rate": 7.961254100085554e-06, + "loss": 0.132, + "step": 11050 + }, + { + "epoch": 4.63, + "learning_rate": 7.95959432916562e-06, + "loss": 0.1382, + "step": 11051 + }, + { + "epoch": 4.63, + "learning_rate": 7.957934616895332e-06, + "loss": 0.1273, + "step": 11052 + }, + { + "epoch": 4.63, + "learning_rate": 7.956274963322406e-06, + "loss": 0.1409, + "step": 11053 + }, + { + "epoch": 4.63, + "learning_rate": 7.954615368494538e-06, + "loss": 0.1615, + "step": 11054 + }, + { + "epoch": 4.63, + "learning_rate": 7.952955832459437e-06, + "loss": 0.1566, + "step": 11055 + }, + { + "epoch": 4.63, + "learning_rate": 7.951296355264805e-06, + "loss": 0.1366, + "step": 11056 + }, + { + "epoch": 4.63, + "learning_rate": 7.949636936958342e-06, + "loss": 0.1468, + "step": 11057 + }, + { + "epoch": 4.63, + "learning_rate": 7.947977577587745e-06, + "loss": 0.1938, + "step": 11058 + }, + { + "epoch": 4.63, + "learning_rate": 7.946318277200712e-06, + "loss": 0.1244, + "step": 11059 + }, + { + "epoch": 4.63, + "learning_rate": 7.944659035844937e-06, + "loss": 0.1601, + "step": 11060 + }, + { + "epoch": 4.63, + "learning_rate": 7.942999853568112e-06, + "loss": 0.1841, + "step": 11061 + }, + { + "epoch": 4.63, + "learning_rate": 7.941340730417934e-06, + "loss": 0.1518, + "step": 11062 + }, + { + "epoch": 4.63, + "learning_rate": 7.93968166644209e-06, + "loss": 0.1456, + "step": 11063 + }, + { + "epoch": 4.63, + "learning_rate": 7.938022661688265e-06, + "loss": 0.137, + "step": 11064 + }, + { + "epoch": 4.63, + "learning_rate": 7.936363716204153e-06, + "loss": 0.1501, + "step": 11065 + }, + { + "epoch": 4.63, + "learning_rate": 7.934704830037432e-06, + "loss": 0.1474, + "step": 11066 + }, + { + "epoch": 4.63, + "learning_rate": 7.933046003235789e-06, + "loss": 0.1521, + "step": 11067 + }, + { + "epoch": 4.63, + "learning_rate": 7.931387235846905e-06, + "loss": 0.1458, + "step": 11068 + }, + { + "epoch": 4.63, + "learning_rate": 7.929728527918456e-06, + "loss": 0.1276, + "step": 11069 + }, + { + "epoch": 4.63, + "learning_rate": 7.928069879498126e-06, + "loss": 0.1618, + "step": 11070 + }, + { + "epoch": 4.64, + "learning_rate": 7.926411290633588e-06, + "loss": 0.1723, + "step": 11071 + }, + { + "epoch": 4.64, + "learning_rate": 7.92475276137252e-06, + "loss": 0.151, + "step": 11072 + }, + { + "epoch": 4.64, + "learning_rate": 7.923094291762588e-06, + "loss": 0.1688, + "step": 11073 + }, + { + "epoch": 4.64, + "learning_rate": 7.921435881851474e-06, + "loss": 0.1447, + "step": 11074 + }, + { + "epoch": 4.64, + "learning_rate": 7.919777531686837e-06, + "loss": 0.1317, + "step": 11075 + }, + { + "epoch": 4.64, + "learning_rate": 7.91811924131635e-06, + "loss": 0.1612, + "step": 11076 + }, + { + "epoch": 4.64, + "learning_rate": 7.916461010787676e-06, + "loss": 0.1556, + "step": 11077 + }, + { + "epoch": 4.64, + "learning_rate": 7.914802840148482e-06, + "loss": 0.1761, + "step": 11078 + }, + { + "epoch": 4.64, + "learning_rate": 7.91314472944643e-06, + "loss": 0.1232, + "step": 11079 + }, + { + "epoch": 4.64, + "learning_rate": 7.91148667872918e-06, + "loss": 0.1865, + "step": 11080 + }, + { + "epoch": 4.64, + "learning_rate": 7.909828688044392e-06, + "loss": 0.1404, + "step": 11081 + }, + { + "epoch": 4.64, + "learning_rate": 7.908170757439724e-06, + "loss": 0.148, + "step": 11082 + }, + { + "epoch": 4.64, + "learning_rate": 7.906512886962833e-06, + "loss": 0.1539, + "step": 11083 + }, + { + "epoch": 4.64, + "learning_rate": 7.90485507666137e-06, + "loss": 0.1607, + "step": 11084 + }, + { + "epoch": 4.64, + "learning_rate": 7.903197326582988e-06, + "loss": 0.1351, + "step": 11085 + }, + { + "epoch": 4.64, + "learning_rate": 7.90153963677534e-06, + "loss": 0.1095, + "step": 11086 + }, + { + "epoch": 4.64, + "learning_rate": 7.899882007286074e-06, + "loss": 0.1682, + "step": 11087 + }, + { + "epoch": 4.64, + "learning_rate": 7.898224438162831e-06, + "loss": 0.1204, + "step": 11088 + }, + { + "epoch": 4.64, + "learning_rate": 7.896566929453267e-06, + "loss": 0.166, + "step": 11089 + }, + { + "epoch": 4.64, + "learning_rate": 7.894909481205017e-06, + "loss": 0.118, + "step": 11090 + }, + { + "epoch": 4.64, + "learning_rate": 7.893252093465725e-06, + "loss": 0.133, + "step": 11091 + }, + { + "epoch": 4.64, + "learning_rate": 7.891594766283034e-06, + "loss": 0.1369, + "step": 11092 + }, + { + "epoch": 4.64, + "learning_rate": 7.88993749970458e-06, + "loss": 0.1728, + "step": 11093 + }, + { + "epoch": 4.64, + "learning_rate": 7.888280293778001e-06, + "loss": 0.1538, + "step": 11094 + }, + { + "epoch": 4.65, + "learning_rate": 7.886623148550932e-06, + "loss": 0.1335, + "step": 11095 + }, + { + "epoch": 4.65, + "learning_rate": 7.884966064071002e-06, + "loss": 0.1191, + "step": 11096 + }, + { + "epoch": 4.65, + "learning_rate": 7.88330904038585e-06, + "loss": 0.1459, + "step": 11097 + }, + { + "epoch": 4.65, + "learning_rate": 7.881652077543098e-06, + "loss": 0.1629, + "step": 11098 + }, + { + "epoch": 4.65, + "learning_rate": 7.879995175590378e-06, + "loss": 0.1501, + "step": 11099 + }, + { + "epoch": 4.65, + "learning_rate": 7.878338334575317e-06, + "loss": 0.189, + "step": 11100 + }, + { + "epoch": 4.65, + "learning_rate": 7.876681554545535e-06, + "loss": 0.1861, + "step": 11101 + }, + { + "epoch": 4.65, + "learning_rate": 7.875024835548658e-06, + "loss": 0.1758, + "step": 11102 + }, + { + "epoch": 4.65, + "learning_rate": 7.873368177632308e-06, + "loss": 0.1761, + "step": 11103 + }, + { + "epoch": 4.65, + "learning_rate": 7.871711580844102e-06, + "loss": 0.159, + "step": 11104 + }, + { + "epoch": 4.65, + "learning_rate": 7.870055045231656e-06, + "loss": 0.1334, + "step": 11105 + }, + { + "epoch": 4.65, + "learning_rate": 7.86839857084259e-06, + "loss": 0.1364, + "step": 11106 + }, + { + "epoch": 4.65, + "learning_rate": 7.866742157724515e-06, + "loss": 0.1134, + "step": 11107 + }, + { + "epoch": 4.65, + "learning_rate": 7.865085805925042e-06, + "loss": 0.1967, + "step": 11108 + }, + { + "epoch": 4.65, + "learning_rate": 7.863429515491786e-06, + "loss": 0.1544, + "step": 11109 + }, + { + "epoch": 4.65, + "learning_rate": 7.861773286472351e-06, + "loss": 0.1231, + "step": 11110 + }, + { + "epoch": 4.65, + "learning_rate": 7.860117118914345e-06, + "loss": 0.1251, + "step": 11111 + }, + { + "epoch": 4.65, + "learning_rate": 7.858461012865373e-06, + "loss": 0.1512, + "step": 11112 + }, + { + "epoch": 4.65, + "learning_rate": 7.856804968373041e-06, + "loss": 0.2215, + "step": 11113 + }, + { + "epoch": 4.65, + "learning_rate": 7.855148985484946e-06, + "loss": 0.135, + "step": 11114 + }, + { + "epoch": 4.65, + "learning_rate": 7.85349306424869e-06, + "loss": 0.1436, + "step": 11115 + }, + { + "epoch": 4.65, + "learning_rate": 7.851837204711874e-06, + "loss": 0.2166, + "step": 11116 + }, + { + "epoch": 4.65, + "learning_rate": 7.85018140692209e-06, + "loss": 0.1807, + "step": 11117 + }, + { + "epoch": 4.65, + "learning_rate": 7.848525670926932e-06, + "loss": 0.1591, + "step": 11118 + }, + { + "epoch": 4.66, + "learning_rate": 7.846869996774e-06, + "loss": 0.1464, + "step": 11119 + }, + { + "epoch": 4.66, + "learning_rate": 7.845214384510875e-06, + "loss": 0.1911, + "step": 11120 + }, + { + "epoch": 4.66, + "learning_rate": 7.84355883418515e-06, + "loss": 0.1695, + "step": 11121 + }, + { + "epoch": 4.66, + "learning_rate": 7.841903345844413e-06, + "loss": 0.1471, + "step": 11122 + }, + { + "epoch": 4.66, + "learning_rate": 7.840247919536249e-06, + "loss": 0.1358, + "step": 11123 + }, + { + "epoch": 4.66, + "learning_rate": 7.838592555308244e-06, + "loss": 0.1547, + "step": 11124 + }, + { + "epoch": 4.66, + "learning_rate": 7.836937253207976e-06, + "loss": 0.1534, + "step": 11125 + }, + { + "epoch": 4.66, + "learning_rate": 7.835282013283027e-06, + "loss": 0.1508, + "step": 11126 + }, + { + "epoch": 4.66, + "learning_rate": 7.833626835580977e-06, + "loss": 0.1305, + "step": 11127 + }, + { + "epoch": 4.66, + "learning_rate": 7.8319717201494e-06, + "loss": 0.1499, + "step": 11128 + }, + { + "epoch": 4.66, + "learning_rate": 7.830316667035872e-06, + "loss": 0.1538, + "step": 11129 + }, + { + "epoch": 4.66, + "learning_rate": 7.828661676287968e-06, + "loss": 0.1356, + "step": 11130 + }, + { + "epoch": 4.66, + "learning_rate": 7.827006747953255e-06, + "loss": 0.1237, + "step": 11131 + }, + { + "epoch": 4.66, + "learning_rate": 7.825351882079305e-06, + "loss": 0.1834, + "step": 11132 + }, + { + "epoch": 4.66, + "learning_rate": 7.823697078713689e-06, + "loss": 0.1415, + "step": 11133 + }, + { + "epoch": 4.66, + "learning_rate": 7.822042337903964e-06, + "loss": 0.1404, + "step": 11134 + }, + { + "epoch": 4.66, + "learning_rate": 7.820387659697701e-06, + "loss": 0.1441, + "step": 11135 + }, + { + "epoch": 4.66, + "learning_rate": 7.818733044142462e-06, + "loss": 0.1424, + "step": 11136 + }, + { + "epoch": 4.66, + "learning_rate": 7.817078491285806e-06, + "loss": 0.1405, + "step": 11137 + }, + { + "epoch": 4.66, + "learning_rate": 7.815424001175289e-06, + "loss": 0.1463, + "step": 11138 + }, + { + "epoch": 4.66, + "learning_rate": 7.813769573858474e-06, + "loss": 0.1954, + "step": 11139 + }, + { + "epoch": 4.66, + "learning_rate": 7.81211520938291e-06, + "loss": 0.1358, + "step": 11140 + }, + { + "epoch": 4.66, + "learning_rate": 7.810460907796153e-06, + "loss": 0.1541, + "step": 11141 + }, + { + "epoch": 4.66, + "learning_rate": 7.80880666914576e-06, + "loss": 0.1566, + "step": 11142 + }, + { + "epoch": 4.67, + "learning_rate": 7.807152493479268e-06, + "loss": 0.1288, + "step": 11143 + }, + { + "epoch": 4.67, + "learning_rate": 7.805498380844236e-06, + "loss": 0.1411, + "step": 11144 + }, + { + "epoch": 4.67, + "learning_rate": 7.803844331288201e-06, + "loss": 0.1424, + "step": 11145 + }, + { + "epoch": 4.67, + "learning_rate": 7.802190344858715e-06, + "loss": 0.1793, + "step": 11146 + }, + { + "epoch": 4.67, + "learning_rate": 7.800536421603317e-06, + "loss": 0.1316, + "step": 11147 + }, + { + "epoch": 4.67, + "learning_rate": 7.798882561569546e-06, + "loss": 0.1403, + "step": 11148 + }, + { + "epoch": 4.67, + "learning_rate": 7.797228764804942e-06, + "loss": 0.1246, + "step": 11149 + }, + { + "epoch": 4.67, + "learning_rate": 7.795575031357047e-06, + "loss": 0.1353, + "step": 11150 + }, + { + "epoch": 4.67, + "learning_rate": 7.793921361273388e-06, + "loss": 0.183, + "step": 11151 + }, + { + "epoch": 4.67, + "learning_rate": 7.792267754601501e-06, + "loss": 0.1039, + "step": 11152 + }, + { + "epoch": 4.67, + "learning_rate": 7.79061421138892e-06, + "loss": 0.1485, + "step": 11153 + }, + { + "epoch": 4.67, + "learning_rate": 7.788960731683174e-06, + "loss": 0.1523, + "step": 11154 + }, + { + "epoch": 4.67, + "learning_rate": 7.787307315531787e-06, + "loss": 0.1523, + "step": 11155 + }, + { + "epoch": 4.67, + "learning_rate": 7.785653962982288e-06, + "loss": 0.1329, + "step": 11156 + }, + { + "epoch": 4.67, + "learning_rate": 7.784000674082205e-06, + "loss": 0.1695, + "step": 11157 + }, + { + "epoch": 4.67, + "learning_rate": 7.782347448879051e-06, + "loss": 0.1421, + "step": 11158 + }, + { + "epoch": 4.67, + "learning_rate": 7.780694287420355e-06, + "loss": 0.1418, + "step": 11159 + }, + { + "epoch": 4.67, + "learning_rate": 7.77904118975363e-06, + "loss": 0.1341, + "step": 11160 + }, + { + "epoch": 4.67, + "learning_rate": 7.777388155926397e-06, + "loss": 0.16, + "step": 11161 + }, + { + "epoch": 4.67, + "learning_rate": 7.775735185986168e-06, + "loss": 0.1736, + "step": 11162 + }, + { + "epoch": 4.67, + "learning_rate": 7.774082279980459e-06, + "loss": 0.1503, + "step": 11163 + }, + { + "epoch": 4.67, + "learning_rate": 7.772429437956777e-06, + "loss": 0.1389, + "step": 11164 + }, + { + "epoch": 4.67, + "learning_rate": 7.77077665996264e-06, + "loss": 0.1582, + "step": 11165 + }, + { + "epoch": 4.67, + "learning_rate": 7.769123946045545e-06, + "loss": 0.1867, + "step": 11166 + }, + { + "epoch": 4.68, + "learning_rate": 7.767471296253003e-06, + "loss": 0.1555, + "step": 11167 + }, + { + "epoch": 4.68, + "learning_rate": 7.765818710632516e-06, + "loss": 0.1818, + "step": 11168 + }, + { + "epoch": 4.68, + "learning_rate": 7.764166189231588e-06, + "loss": 0.1169, + "step": 11169 + }, + { + "epoch": 4.68, + "learning_rate": 7.762513732097721e-06, + "loss": 0.1792, + "step": 11170 + }, + { + "epoch": 4.68, + "learning_rate": 7.76086133927841e-06, + "loss": 0.1554, + "step": 11171 + }, + { + "epoch": 4.68, + "learning_rate": 7.759209010821152e-06, + "loss": 0.1676, + "step": 11172 + }, + { + "epoch": 4.68, + "learning_rate": 7.757556746773443e-06, + "loss": 0.1391, + "step": 11173 + }, + { + "epoch": 4.68, + "learning_rate": 7.755904547182778e-06, + "loss": 0.1281, + "step": 11174 + }, + { + "epoch": 4.68, + "learning_rate": 7.754252412096642e-06, + "loss": 0.1502, + "step": 11175 + }, + { + "epoch": 4.68, + "learning_rate": 7.752600341562527e-06, + "loss": 0.152, + "step": 11176 + }, + { + "epoch": 4.68, + "learning_rate": 7.750948335627925e-06, + "loss": 0.1148, + "step": 11177 + }, + { + "epoch": 4.68, + "learning_rate": 7.749296394340313e-06, + "loss": 0.1783, + "step": 11178 + }, + { + "epoch": 4.68, + "learning_rate": 7.747644517747182e-06, + "loss": 0.1686, + "step": 11179 + }, + { + "epoch": 4.68, + "learning_rate": 7.74599270589601e-06, + "loss": 0.1007, + "step": 11180 + }, + { + "epoch": 4.68, + "learning_rate": 7.744340958834275e-06, + "loss": 0.1236, + "step": 11181 + }, + { + "epoch": 4.68, + "learning_rate": 7.742689276609459e-06, + "loss": 0.1305, + "step": 11182 + }, + { + "epoch": 4.68, + "learning_rate": 7.741037659269037e-06, + "loss": 0.1455, + "step": 11183 + }, + { + "epoch": 4.68, + "learning_rate": 7.739386106860485e-06, + "loss": 0.1433, + "step": 11184 + }, + { + "epoch": 4.68, + "learning_rate": 7.73773461943127e-06, + "loss": 0.1277, + "step": 11185 + }, + { + "epoch": 4.68, + "learning_rate": 7.736083197028866e-06, + "loss": 0.2145, + "step": 11186 + }, + { + "epoch": 4.68, + "learning_rate": 7.734431839700747e-06, + "loss": 0.1557, + "step": 11187 + }, + { + "epoch": 4.68, + "learning_rate": 7.732780547494372e-06, + "loss": 0.1628, + "step": 11188 + }, + { + "epoch": 4.68, + "learning_rate": 7.731129320457205e-06, + "loss": 0.1719, + "step": 11189 + }, + { + "epoch": 4.68, + "learning_rate": 7.729478158636712e-06, + "loss": 0.1289, + "step": 11190 + }, + { + "epoch": 4.69, + "learning_rate": 7.727827062080359e-06, + "loss": 0.1182, + "step": 11191 + }, + { + "epoch": 4.69, + "learning_rate": 7.726176030835597e-06, + "loss": 0.1314, + "step": 11192 + }, + { + "epoch": 4.69, + "learning_rate": 7.724525064949887e-06, + "loss": 0.1119, + "step": 11193 + }, + { + "epoch": 4.69, + "learning_rate": 7.722874164470685e-06, + "loss": 0.1527, + "step": 11194 + }, + { + "epoch": 4.69, + "learning_rate": 7.721223329445448e-06, + "loss": 0.1525, + "step": 11195 + }, + { + "epoch": 4.69, + "learning_rate": 7.719572559921619e-06, + "loss": 0.177, + "step": 11196 + }, + { + "epoch": 4.69, + "learning_rate": 7.717921855946656e-06, + "loss": 0.1636, + "step": 11197 + }, + { + "epoch": 4.69, + "learning_rate": 7.716271217568005e-06, + "loss": 0.1738, + "step": 11198 + }, + { + "epoch": 4.69, + "learning_rate": 7.71462064483311e-06, + "loss": 0.1189, + "step": 11199 + }, + { + "epoch": 4.69, + "learning_rate": 7.712970137789416e-06, + "loss": 0.1319, + "step": 11200 + }, + { + "epoch": 4.69, + "learning_rate": 7.711319696484368e-06, + "loss": 0.1306, + "step": 11201 + }, + { + "epoch": 4.69, + "learning_rate": 7.709669320965402e-06, + "loss": 0.1446, + "step": 11202 + }, + { + "epoch": 4.69, + "learning_rate": 7.70801901127996e-06, + "loss": 0.1587, + "step": 11203 + }, + { + "epoch": 4.69, + "learning_rate": 7.70636876747548e-06, + "loss": 0.1619, + "step": 11204 + }, + { + "epoch": 4.69, + "learning_rate": 7.70471858959939e-06, + "loss": 0.1482, + "step": 11205 + }, + { + "epoch": 4.69, + "learning_rate": 7.70306847769913e-06, + "loss": 0.1366, + "step": 11206 + }, + { + "epoch": 4.69, + "learning_rate": 7.70141843182213e-06, + "loss": 0.1257, + "step": 11207 + }, + { + "epoch": 4.69, + "learning_rate": 7.699768452015815e-06, + "loss": 0.151, + "step": 11208 + }, + { + "epoch": 4.69, + "learning_rate": 7.698118538327613e-06, + "loss": 0.1567, + "step": 11209 + }, + { + "epoch": 4.69, + "learning_rate": 7.696468690804956e-06, + "loss": 0.1596, + "step": 11210 + }, + { + "epoch": 4.69, + "learning_rate": 7.69481890949526e-06, + "loss": 0.1584, + "step": 11211 + }, + { + "epoch": 4.69, + "learning_rate": 7.69316919444595e-06, + "loss": 0.1776, + "step": 11212 + }, + { + "epoch": 4.69, + "learning_rate": 7.691519545704442e-06, + "loss": 0.1568, + "step": 11213 + }, + { + "epoch": 4.69, + "learning_rate": 7.689869963318154e-06, + "loss": 0.1598, + "step": 11214 + }, + { + "epoch": 4.7, + "learning_rate": 7.688220447334508e-06, + "loss": 0.1739, + "step": 11215 + }, + { + "epoch": 4.7, + "learning_rate": 7.686570997800914e-06, + "loss": 0.1726, + "step": 11216 + }, + { + "epoch": 4.7, + "learning_rate": 7.68492161476478e-06, + "loss": 0.1606, + "step": 11217 + }, + { + "epoch": 4.7, + "learning_rate": 7.683272298273525e-06, + "loss": 0.1754, + "step": 11218 + }, + { + "epoch": 4.7, + "learning_rate": 7.681623048374549e-06, + "loss": 0.1424, + "step": 11219 + }, + { + "epoch": 4.7, + "learning_rate": 7.67997386511526e-06, + "loss": 0.2035, + "step": 11220 + }, + { + "epoch": 4.7, + "learning_rate": 7.678324748543066e-06, + "loss": 0.2215, + "step": 11221 + }, + { + "epoch": 4.7, + "learning_rate": 7.676675698705364e-06, + "loss": 0.1417, + "step": 11222 + }, + { + "epoch": 4.7, + "learning_rate": 7.67502671564956e-06, + "loss": 0.1331, + "step": 11223 + }, + { + "epoch": 4.7, + "learning_rate": 7.673377799423047e-06, + "loss": 0.1634, + "step": 11224 + }, + { + "epoch": 4.7, + "learning_rate": 7.671728950073228e-06, + "loss": 0.1274, + "step": 11225 + }, + { + "epoch": 4.7, + "learning_rate": 7.670080167647492e-06, + "loss": 0.1679, + "step": 11226 + }, + { + "epoch": 4.7, + "learning_rate": 7.668431452193234e-06, + "loss": 0.1328, + "step": 11227 + }, + { + "epoch": 4.7, + "learning_rate": 7.666782803757848e-06, + "loss": 0.1429, + "step": 11228 + }, + { + "epoch": 4.7, + "learning_rate": 7.665134222388717e-06, + "loss": 0.1673, + "step": 11229 + }, + { + "epoch": 4.7, + "learning_rate": 7.66348570813323e-06, + "loss": 0.1549, + "step": 11230 + }, + { + "epoch": 4.7, + "learning_rate": 7.661837261038776e-06, + "loss": 0.1568, + "step": 11231 + }, + { + "epoch": 4.7, + "learning_rate": 7.660188881152732e-06, + "loss": 0.1671, + "step": 11232 + }, + { + "epoch": 4.7, + "learning_rate": 7.658540568522487e-06, + "loss": 0.1755, + "step": 11233 + }, + { + "epoch": 4.7, + "learning_rate": 7.656892323195414e-06, + "loss": 0.1713, + "step": 11234 + }, + { + "epoch": 4.7, + "learning_rate": 7.655244145218887e-06, + "loss": 0.1667, + "step": 11235 + }, + { + "epoch": 4.7, + "learning_rate": 7.653596034640291e-06, + "loss": 0.166, + "step": 11236 + }, + { + "epoch": 4.7, + "learning_rate": 7.651947991506994e-06, + "loss": 0.1592, + "step": 11237 + }, + { + "epoch": 4.71, + "learning_rate": 7.650300015866367e-06, + "loss": 0.1462, + "step": 11238 + }, + { + "epoch": 4.71, + "learning_rate": 7.648652107765783e-06, + "loss": 0.1501, + "step": 11239 + }, + { + "epoch": 4.71, + "learning_rate": 7.647004267252606e-06, + "loss": 0.1555, + "step": 11240 + }, + { + "epoch": 4.71, + "learning_rate": 7.645356494374203e-06, + "loss": 0.1119, + "step": 11241 + }, + { + "epoch": 4.71, + "learning_rate": 7.643708789177939e-06, + "loss": 0.1433, + "step": 11242 + }, + { + "epoch": 4.71, + "learning_rate": 7.642061151711176e-06, + "loss": 0.1535, + "step": 11243 + }, + { + "epoch": 4.71, + "learning_rate": 7.64041358202127e-06, + "loss": 0.1613, + "step": 11244 + }, + { + "epoch": 4.71, + "learning_rate": 7.638766080155584e-06, + "loss": 0.1381, + "step": 11245 + }, + { + "epoch": 4.71, + "learning_rate": 7.63711864616147e-06, + "loss": 0.1383, + "step": 11246 + }, + { + "epoch": 4.71, + "learning_rate": 7.635471280086285e-06, + "loss": 0.1832, + "step": 11247 + }, + { + "epoch": 4.71, + "learning_rate": 7.63382398197738e-06, + "loss": 0.1316, + "step": 11248 + }, + { + "epoch": 4.71, + "learning_rate": 7.632176751882103e-06, + "loss": 0.1394, + "step": 11249 + }, + { + "epoch": 4.71, + "learning_rate": 7.630529589847807e-06, + "loss": 0.1282, + "step": 11250 + }, + { + "epoch": 4.71, + "learning_rate": 7.628882495921834e-06, + "loss": 0.1677, + "step": 11251 + }, + { + "epoch": 4.71, + "learning_rate": 7.627235470151528e-06, + "loss": 0.2038, + "step": 11252 + }, + { + "epoch": 4.71, + "learning_rate": 7.625588512584234e-06, + "loss": 0.1632, + "step": 11253 + }, + { + "epoch": 4.71, + "learning_rate": 7.6239416232672904e-06, + "loss": 0.1553, + "step": 11254 + }, + { + "epoch": 4.71, + "learning_rate": 7.622294802248041e-06, + "loss": 0.1566, + "step": 11255 + }, + { + "epoch": 4.71, + "learning_rate": 7.620648049573815e-06, + "loss": 0.1448, + "step": 11256 + }, + { + "epoch": 4.71, + "learning_rate": 7.619001365291948e-06, + "loss": 0.141, + "step": 11257 + }, + { + "epoch": 4.71, + "learning_rate": 7.617354749449774e-06, + "loss": 0.1419, + "step": 11258 + }, + { + "epoch": 4.71, + "learning_rate": 7.615708202094625e-06, + "loss": 0.1363, + "step": 11259 + }, + { + "epoch": 4.71, + "learning_rate": 7.614061723273825e-06, + "loss": 0.1554, + "step": 11260 + }, + { + "epoch": 4.71, + "learning_rate": 7.612415313034706e-06, + "loss": 0.1467, + "step": 11261 + }, + { + "epoch": 4.72, + "learning_rate": 7.61076897142459e-06, + "loss": 0.1473, + "step": 11262 + }, + { + "epoch": 4.72, + "learning_rate": 7.6091226984907985e-06, + "loss": 0.1502, + "step": 11263 + }, + { + "epoch": 4.72, + "learning_rate": 7.607476494280653e-06, + "loss": 0.1061, + "step": 11264 + }, + { + "epoch": 4.72, + "learning_rate": 7.6058303588414725e-06, + "loss": 0.2, + "step": 11265 + }, + { + "epoch": 4.72, + "learning_rate": 7.604184292220577e-06, + "loss": 0.1315, + "step": 11266 + }, + { + "epoch": 4.72, + "learning_rate": 7.602538294465273e-06, + "loss": 0.1253, + "step": 11267 + }, + { + "epoch": 4.72, + "learning_rate": 7.6008923656228814e-06, + "loss": 0.1226, + "step": 11268 + }, + { + "epoch": 4.72, + "learning_rate": 7.599246505740711e-06, + "loss": 0.1397, + "step": 11269 + }, + { + "epoch": 4.72, + "learning_rate": 7.597600714866067e-06, + "loss": 0.1627, + "step": 11270 + }, + { + "epoch": 4.72, + "learning_rate": 7.595954993046259e-06, + "loss": 0.1419, + "step": 11271 + }, + { + "epoch": 4.72, + "learning_rate": 7.594309340328594e-06, + "loss": 0.1498, + "step": 11272 + }, + { + "epoch": 4.72, + "learning_rate": 7.592663756760369e-06, + "loss": 0.1064, + "step": 11273 + }, + { + "epoch": 4.72, + "learning_rate": 7.591018242388888e-06, + "loss": 0.1151, + "step": 11274 + }, + { + "epoch": 4.72, + "learning_rate": 7.589372797261453e-06, + "loss": 0.1374, + "step": 11275 + }, + { + "epoch": 4.72, + "learning_rate": 7.587727421425355e-06, + "loss": 0.1407, + "step": 11276 + }, + { + "epoch": 4.72, + "learning_rate": 7.586082114927892e-06, + "loss": 0.1609, + "step": 11277 + }, + { + "epoch": 4.72, + "learning_rate": 7.58443687781636e-06, + "loss": 0.1706, + "step": 11278 + }, + { + "epoch": 4.72, + "learning_rate": 7.582791710138043e-06, + "loss": 0.1377, + "step": 11279 + }, + { + "epoch": 4.72, + "learning_rate": 7.581146611940235e-06, + "loss": 0.1213, + "step": 11280 + }, + { + "epoch": 4.72, + "learning_rate": 7.57950158327022e-06, + "loss": 0.1416, + "step": 11281 + }, + { + "epoch": 4.72, + "learning_rate": 7.577856624175283e-06, + "loss": 0.1573, + "step": 11282 + }, + { + "epoch": 4.72, + "learning_rate": 7.576211734702711e-06, + "loss": 0.155, + "step": 11283 + }, + { + "epoch": 4.72, + "learning_rate": 7.574566914899779e-06, + "loss": 0.0967, + "step": 11284 + }, + { + "epoch": 4.72, + "learning_rate": 7.5729221648137686e-06, + "loss": 0.1409, + "step": 11285 + }, + { + "epoch": 4.73, + "learning_rate": 7.571277484491959e-06, + "loss": 0.1336, + "step": 11286 + }, + { + "epoch": 4.73, + "learning_rate": 7.569632873981622e-06, + "loss": 0.1512, + "step": 11287 + }, + { + "epoch": 4.73, + "learning_rate": 7.56798833333003e-06, + "loss": 0.1534, + "step": 11288 + }, + { + "epoch": 4.73, + "learning_rate": 7.566343862584458e-06, + "loss": 0.1724, + "step": 11289 + }, + { + "epoch": 4.73, + "learning_rate": 7.56469946179217e-06, + "loss": 0.129, + "step": 11290 + }, + { + "epoch": 4.73, + "learning_rate": 7.563055131000434e-06, + "loss": 0.1654, + "step": 11291 + }, + { + "epoch": 4.73, + "learning_rate": 7.561410870256518e-06, + "loss": 0.1414, + "step": 11292 + }, + { + "epoch": 4.73, + "learning_rate": 7.559766679607682e-06, + "loss": 0.1662, + "step": 11293 + }, + { + "epoch": 4.73, + "learning_rate": 7.5581225591011864e-06, + "loss": 0.1258, + "step": 11294 + }, + { + "epoch": 4.73, + "learning_rate": 7.556478508784291e-06, + "loss": 0.1049, + "step": 11295 + }, + { + "epoch": 4.73, + "learning_rate": 7.554834528704254e-06, + "loss": 0.1295, + "step": 11296 + }, + { + "epoch": 4.73, + "learning_rate": 7.553190618908327e-06, + "loss": 0.0941, + "step": 11297 + }, + { + "epoch": 4.73, + "learning_rate": 7.551546779443765e-06, + "loss": 0.0995, + "step": 11298 + }, + { + "epoch": 4.73, + "learning_rate": 7.54990301035782e-06, + "loss": 0.1157, + "step": 11299 + }, + { + "epoch": 4.73, + "learning_rate": 7.5482593116977355e-06, + "loss": 0.1149, + "step": 11300 + }, + { + "epoch": 4.73, + "learning_rate": 7.546615683510767e-06, + "loss": 0.1697, + "step": 11301 + }, + { + "epoch": 4.73, + "learning_rate": 7.544972125844149e-06, + "loss": 0.1343, + "step": 11302 + }, + { + "epoch": 4.73, + "learning_rate": 7.543328638745131e-06, + "loss": 0.099, + "step": 11303 + }, + { + "epoch": 4.73, + "learning_rate": 7.541685222260948e-06, + "loss": 0.1647, + "step": 11304 + }, + { + "epoch": 4.73, + "learning_rate": 7.540041876438842e-06, + "loss": 0.1305, + "step": 11305 + }, + { + "epoch": 4.73, + "learning_rate": 7.538398601326051e-06, + "loss": 0.1737, + "step": 11306 + }, + { + "epoch": 4.73, + "learning_rate": 7.5367553969698095e-06, + "loss": 0.1746, + "step": 11307 + }, + { + "epoch": 4.73, + "learning_rate": 7.535112263417346e-06, + "loss": 0.1647, + "step": 11308 + }, + { + "epoch": 4.73, + "learning_rate": 7.533469200715892e-06, + "loss": 0.1599, + "step": 11309 + }, + { + "epoch": 4.74, + "learning_rate": 7.53182620891268e-06, + "loss": 0.1466, + "step": 11310 + }, + { + "epoch": 4.74, + "learning_rate": 7.530183288054931e-06, + "loss": 0.1535, + "step": 11311 + }, + { + "epoch": 4.74, + "learning_rate": 7.528540438189872e-06, + "loss": 0.1489, + "step": 11312 + }, + { + "epoch": 4.74, + "learning_rate": 7.526897659364727e-06, + "loss": 0.1699, + "step": 11313 + }, + { + "epoch": 4.74, + "learning_rate": 7.525254951626712e-06, + "loss": 0.106, + "step": 11314 + }, + { + "epoch": 4.74, + "learning_rate": 7.523612315023048e-06, + "loss": 0.1492, + "step": 11315 + }, + { + "epoch": 4.74, + "learning_rate": 7.5219697496009525e-06, + "loss": 0.1654, + "step": 11316 + }, + { + "epoch": 4.74, + "learning_rate": 7.520327255407635e-06, + "loss": 0.1983, + "step": 11317 + }, + { + "epoch": 4.74, + "learning_rate": 7.518684832490311e-06, + "loss": 0.167, + "step": 11318 + }, + { + "epoch": 4.74, + "learning_rate": 7.51704248089619e-06, + "loss": 0.1525, + "step": 11319 + }, + { + "epoch": 4.74, + "learning_rate": 7.515400200672478e-06, + "loss": 0.1334, + "step": 11320 + }, + { + "epoch": 4.74, + "learning_rate": 7.513757991866383e-06, + "loss": 0.1492, + "step": 11321 + }, + { + "epoch": 4.74, + "learning_rate": 7.5121158545251085e-06, + "loss": 0.1698, + "step": 11322 + }, + { + "epoch": 4.74, + "learning_rate": 7.51047378869586e-06, + "loss": 0.1589, + "step": 11323 + }, + { + "epoch": 4.74, + "learning_rate": 7.5088317944258305e-06, + "loss": 0.1427, + "step": 11324 + }, + { + "epoch": 4.74, + "learning_rate": 7.50718987176222e-06, + "loss": 0.1504, + "step": 11325 + }, + { + "epoch": 4.74, + "learning_rate": 7.5055480207522225e-06, + "loss": 0.1366, + "step": 11326 + }, + { + "epoch": 4.74, + "learning_rate": 7.503906241443037e-06, + "loss": 0.1122, + "step": 11327 + }, + { + "epoch": 4.74, + "learning_rate": 7.502264533881849e-06, + "loss": 0.1707, + "step": 11328 + }, + { + "epoch": 4.74, + "learning_rate": 7.50062289811585e-06, + "loss": 0.1653, + "step": 11329 + }, + { + "epoch": 4.74, + "learning_rate": 7.49898133419223e-06, + "loss": 0.1242, + "step": 11330 + }, + { + "epoch": 4.74, + "learning_rate": 7.497339842158171e-06, + "loss": 0.1779, + "step": 11331 + }, + { + "epoch": 4.74, + "learning_rate": 7.495698422060857e-06, + "loss": 0.1762, + "step": 11332 + }, + { + "epoch": 4.74, + "learning_rate": 7.494057073947473e-06, + "loss": 0.1771, + "step": 11333 + }, + { + "epoch": 4.75, + "learning_rate": 7.492415797865191e-06, + "loss": 0.1462, + "step": 11334 + }, + { + "epoch": 4.75, + "learning_rate": 7.490774593861192e-06, + "loss": 0.1265, + "step": 11335 + }, + { + "epoch": 4.75, + "learning_rate": 7.489133461982651e-06, + "loss": 0.1422, + "step": 11336 + }, + { + "epoch": 4.75, + "learning_rate": 7.487492402276742e-06, + "loss": 0.1407, + "step": 11337 + }, + { + "epoch": 4.75, + "learning_rate": 7.485851414790632e-06, + "loss": 0.1442, + "step": 11338 + }, + { + "epoch": 4.75, + "learning_rate": 7.4842104995714935e-06, + "loss": 0.2042, + "step": 11339 + }, + { + "epoch": 4.75, + "learning_rate": 7.482569656666493e-06, + "loss": 0.1364, + "step": 11340 + }, + { + "epoch": 4.75, + "learning_rate": 7.480928886122792e-06, + "loss": 0.1347, + "step": 11341 + }, + { + "epoch": 4.75, + "learning_rate": 7.479288187987554e-06, + "loss": 0.1625, + "step": 11342 + }, + { + "epoch": 4.75, + "learning_rate": 7.477647562307943e-06, + "loss": 0.147, + "step": 11343 + }, + { + "epoch": 4.75, + "learning_rate": 7.476007009131112e-06, + "loss": 0.1096, + "step": 11344 + }, + { + "epoch": 4.75, + "learning_rate": 7.47436652850422e-06, + "loss": 0.1525, + "step": 11345 + }, + { + "epoch": 4.75, + "learning_rate": 7.472726120474425e-06, + "loss": 0.1605, + "step": 11346 + }, + { + "epoch": 4.75, + "learning_rate": 7.471085785088871e-06, + "loss": 0.1622, + "step": 11347 + }, + { + "epoch": 4.75, + "learning_rate": 7.469445522394714e-06, + "loss": 0.16, + "step": 11348 + }, + { + "epoch": 4.75, + "learning_rate": 7.467805332439098e-06, + "loss": 0.1244, + "step": 11349 + }, + { + "epoch": 4.75, + "learning_rate": 7.46616521526917e-06, + "loss": 0.1675, + "step": 11350 + }, + { + "epoch": 4.75, + "learning_rate": 7.4645251709320775e-06, + "loss": 0.1349, + "step": 11351 + }, + { + "epoch": 4.75, + "learning_rate": 7.462885199474956e-06, + "loss": 0.164, + "step": 11352 + }, + { + "epoch": 4.75, + "learning_rate": 7.4612453009449485e-06, + "loss": 0.1653, + "step": 11353 + }, + { + "epoch": 4.75, + "learning_rate": 7.459605475389193e-06, + "loss": 0.1166, + "step": 11354 + }, + { + "epoch": 4.75, + "learning_rate": 7.457965722854823e-06, + "loss": 0.1068, + "step": 11355 + }, + { + "epoch": 4.75, + "learning_rate": 7.4563260433889726e-06, + "loss": 0.1797, + "step": 11356 + }, + { + "epoch": 4.75, + "learning_rate": 7.454686437038775e-06, + "loss": 0.1814, + "step": 11357 + }, + { + "epoch": 4.76, + "learning_rate": 7.4530469038513545e-06, + "loss": 0.1696, + "step": 11358 + }, + { + "epoch": 4.76, + "learning_rate": 7.45140744387384e-06, + "loss": 0.1331, + "step": 11359 + }, + { + "epoch": 4.76, + "learning_rate": 7.44976805715336e-06, + "loss": 0.1058, + "step": 11360 + }, + { + "epoch": 4.76, + "learning_rate": 7.448128743737032e-06, + "loss": 0.1342, + "step": 11361 + }, + { + "epoch": 4.76, + "learning_rate": 7.446489503671979e-06, + "loss": 0.1599, + "step": 11362 + }, + { + "epoch": 4.76, + "learning_rate": 7.4448503370053204e-06, + "loss": 0.1291, + "step": 11363 + }, + { + "epoch": 4.76, + "learning_rate": 7.44321124378417e-06, + "loss": 0.1425, + "step": 11364 + }, + { + "epoch": 4.76, + "learning_rate": 7.441572224055644e-06, + "loss": 0.186, + "step": 11365 + }, + { + "epoch": 4.76, + "learning_rate": 7.4399332778668535e-06, + "loss": 0.1956, + "step": 11366 + }, + { + "epoch": 4.76, + "learning_rate": 7.438294405264912e-06, + "loss": 0.2012, + "step": 11367 + }, + { + "epoch": 4.76, + "learning_rate": 7.436655606296921e-06, + "loss": 0.1178, + "step": 11368 + }, + { + "epoch": 4.76, + "learning_rate": 7.4350168810099955e-06, + "loss": 0.1288, + "step": 11369 + }, + { + "epoch": 4.76, + "learning_rate": 7.433378229451229e-06, + "loss": 0.1542, + "step": 11370 + }, + { + "epoch": 4.76, + "learning_rate": 7.431739651667729e-06, + "loss": 0.1506, + "step": 11371 + }, + { + "epoch": 4.76, + "learning_rate": 7.430101147706592e-06, + "loss": 0.2029, + "step": 11372 + }, + { + "epoch": 4.76, + "learning_rate": 7.428462717614917e-06, + "loss": 0.1316, + "step": 11373 + }, + { + "epoch": 4.76, + "learning_rate": 7.426824361439803e-06, + "loss": 0.1284, + "step": 11374 + }, + { + "epoch": 4.76, + "learning_rate": 7.425186079228335e-06, + "loss": 0.1285, + "step": 11375 + }, + { + "epoch": 4.76, + "learning_rate": 7.42354787102761e-06, + "loss": 0.1689, + "step": 11376 + }, + { + "epoch": 4.76, + "learning_rate": 7.421909736884714e-06, + "loss": 0.1667, + "step": 11377 + }, + { + "epoch": 4.76, + "learning_rate": 7.420271676846737e-06, + "loss": 0.1947, + "step": 11378 + }, + { + "epoch": 4.76, + "learning_rate": 7.4186336909607594e-06, + "loss": 0.1408, + "step": 11379 + }, + { + "epoch": 4.76, + "learning_rate": 7.416995779273866e-06, + "loss": 0.1398, + "step": 11380 + }, + { + "epoch": 4.76, + "learning_rate": 7.415357941833139e-06, + "loss": 0.1497, + "step": 11381 + }, + { + "epoch": 4.77, + "learning_rate": 7.413720178685653e-06, + "loss": 0.1695, + "step": 11382 + }, + { + "epoch": 4.77, + "learning_rate": 7.412082489878484e-06, + "loss": 0.1638, + "step": 11383 + }, + { + "epoch": 4.77, + "learning_rate": 7.41044487545871e-06, + "loss": 0.1852, + "step": 11384 + }, + { + "epoch": 4.77, + "learning_rate": 7.408807335473398e-06, + "loss": 0.1481, + "step": 11385 + }, + { + "epoch": 4.77, + "learning_rate": 7.40716986996962e-06, + "loss": 0.1608, + "step": 11386 + }, + { + "epoch": 4.77, + "learning_rate": 7.4055324789944446e-06, + "loss": 0.1231, + "step": 11387 + }, + { + "epoch": 4.77, + "learning_rate": 7.403895162594934e-06, + "loss": 0.1494, + "step": 11388 + }, + { + "epoch": 4.77, + "learning_rate": 7.402257920818154e-06, + "loss": 0.1702, + "step": 11389 + }, + { + "epoch": 4.77, + "learning_rate": 7.400620753711167e-06, + "loss": 0.1696, + "step": 11390 + }, + { + "epoch": 4.77, + "learning_rate": 7.398983661321029e-06, + "loss": 0.1542, + "step": 11391 + }, + { + "epoch": 4.77, + "learning_rate": 7.3973466436947985e-06, + "loss": 0.1124, + "step": 11392 + }, + { + "epoch": 4.77, + "learning_rate": 7.395709700879525e-06, + "loss": 0.1556, + "step": 11393 + }, + { + "epoch": 4.77, + "learning_rate": 7.394072832922268e-06, + "loss": 0.1859, + "step": 11394 + }, + { + "epoch": 4.77, + "learning_rate": 7.3924360398700745e-06, + "loss": 0.1514, + "step": 11395 + }, + { + "epoch": 4.77, + "learning_rate": 7.390799321769993e-06, + "loss": 0.1577, + "step": 11396 + }, + { + "epoch": 4.77, + "learning_rate": 7.389162678669069e-06, + "loss": 0.1659, + "step": 11397 + }, + { + "epoch": 4.77, + "learning_rate": 7.387526110614347e-06, + "loss": 0.1558, + "step": 11398 + }, + { + "epoch": 4.77, + "learning_rate": 7.385889617652868e-06, + "loss": 0.1791, + "step": 11399 + }, + { + "epoch": 4.77, + "learning_rate": 7.384253199831674e-06, + "loss": 0.1584, + "step": 11400 + }, + { + "epoch": 4.77, + "learning_rate": 7.382616857197801e-06, + "loss": 0.1717, + "step": 11401 + }, + { + "epoch": 4.77, + "learning_rate": 7.380980589798281e-06, + "loss": 0.1607, + "step": 11402 + }, + { + "epoch": 4.77, + "learning_rate": 7.379344397680151e-06, + "loss": 0.1706, + "step": 11403 + }, + { + "epoch": 4.77, + "learning_rate": 7.3777082808904435e-06, + "loss": 0.1739, + "step": 11404 + }, + { + "epoch": 4.77, + "learning_rate": 7.376072239476182e-06, + "loss": 0.1403, + "step": 11405 + }, + { + "epoch": 4.78, + "learning_rate": 7.374436273484395e-06, + "loss": 0.1492, + "step": 11406 + }, + { + "epoch": 4.78, + "learning_rate": 7.372800382962107e-06, + "loss": 0.1787, + "step": 11407 + }, + { + "epoch": 4.78, + "learning_rate": 7.371164567956344e-06, + "loss": 0.155, + "step": 11408 + }, + { + "epoch": 4.78, + "learning_rate": 7.36952882851412e-06, + "loss": 0.1437, + "step": 11409 + }, + { + "epoch": 4.78, + "learning_rate": 7.367893164682456e-06, + "loss": 0.1398, + "step": 11410 + }, + { + "epoch": 4.78, + "learning_rate": 7.366257576508369e-06, + "loss": 0.1301, + "step": 11411 + }, + { + "epoch": 4.78, + "learning_rate": 7.364622064038868e-06, + "loss": 0.1722, + "step": 11412 + }, + { + "epoch": 4.78, + "learning_rate": 7.362986627320967e-06, + "loss": 0.1482, + "step": 11413 + }, + { + "epoch": 4.78, + "learning_rate": 7.361351266401682e-06, + "loss": 0.1344, + "step": 11414 + }, + { + "epoch": 4.78, + "learning_rate": 7.359715981328009e-06, + "loss": 0.1519, + "step": 11415 + }, + { + "epoch": 4.78, + "learning_rate": 7.358080772146956e-06, + "loss": 0.1482, + "step": 11416 + }, + { + "epoch": 4.78, + "learning_rate": 7.356445638905526e-06, + "loss": 0.1776, + "step": 11417 + }, + { + "epoch": 4.78, + "learning_rate": 7.354810581650721e-06, + "loss": 0.1388, + "step": 11418 + }, + { + "epoch": 4.78, + "learning_rate": 7.35317560042954e-06, + "loss": 0.1726, + "step": 11419 + }, + { + "epoch": 4.78, + "learning_rate": 7.351540695288977e-06, + "loss": 0.1506, + "step": 11420 + }, + { + "epoch": 4.78, + "learning_rate": 7.349905866276025e-06, + "loss": 0.1139, + "step": 11421 + }, + { + "epoch": 4.78, + "learning_rate": 7.348271113437679e-06, + "loss": 0.1037, + "step": 11422 + }, + { + "epoch": 4.78, + "learning_rate": 7.346636436820926e-06, + "loss": 0.1677, + "step": 11423 + }, + { + "epoch": 4.78, + "learning_rate": 7.345001836472754e-06, + "loss": 0.1564, + "step": 11424 + }, + { + "epoch": 4.78, + "learning_rate": 7.3433673124401495e-06, + "loss": 0.1574, + "step": 11425 + }, + { + "epoch": 4.78, + "learning_rate": 7.341732864770093e-06, + "loss": 0.1255, + "step": 11426 + }, + { + "epoch": 4.78, + "learning_rate": 7.340098493509565e-06, + "loss": 0.1382, + "step": 11427 + }, + { + "epoch": 4.78, + "learning_rate": 7.338464198705548e-06, + "loss": 0.1807, + "step": 11428 + }, + { + "epoch": 4.79, + "learning_rate": 7.336829980405014e-06, + "loss": 0.1531, + "step": 11429 + }, + { + "epoch": 4.79, + "learning_rate": 7.335195838654939e-06, + "loss": 0.1694, + "step": 11430 + }, + { + "epoch": 4.79, + "learning_rate": 7.333561773502297e-06, + "loss": 0.1758, + "step": 11431 + }, + { + "epoch": 4.79, + "learning_rate": 7.3319277849940525e-06, + "loss": 0.1775, + "step": 11432 + }, + { + "epoch": 4.79, + "learning_rate": 7.330293873177179e-06, + "loss": 0.1232, + "step": 11433 + }, + { + "epoch": 4.79, + "learning_rate": 7.328660038098636e-06, + "loss": 0.153, + "step": 11434 + }, + { + "epoch": 4.79, + "learning_rate": 7.327026279805394e-06, + "loss": 0.1797, + "step": 11435 + }, + { + "epoch": 4.79, + "learning_rate": 7.325392598344406e-06, + "loss": 0.1842, + "step": 11436 + }, + { + "epoch": 4.79, + "learning_rate": 7.3237589937626395e-06, + "loss": 0.1748, + "step": 11437 + }, + { + "epoch": 4.79, + "learning_rate": 7.322125466107042e-06, + "loss": 0.1729, + "step": 11438 + }, + { + "epoch": 4.79, + "learning_rate": 7.320492015424572e-06, + "loss": 0.1727, + "step": 11439 + }, + { + "epoch": 4.79, + "learning_rate": 7.318858641762181e-06, + "loss": 0.1761, + "step": 11440 + }, + { + "epoch": 4.79, + "learning_rate": 7.317225345166819e-06, + "loss": 0.1366, + "step": 11441 + }, + { + "epoch": 4.79, + "learning_rate": 7.315592125685434e-06, + "loss": 0.1397, + "step": 11442 + }, + { + "epoch": 4.79, + "learning_rate": 7.31395898336497e-06, + "loss": 0.107, + "step": 11443 + }, + { + "epoch": 4.79, + "learning_rate": 7.312325918252373e-06, + "loss": 0.1468, + "step": 11444 + }, + { + "epoch": 4.79, + "learning_rate": 7.310692930394583e-06, + "loss": 0.1194, + "step": 11445 + }, + { + "epoch": 4.79, + "learning_rate": 7.309060019838536e-06, + "loss": 0.1522, + "step": 11446 + }, + { + "epoch": 4.79, + "learning_rate": 7.307427186631173e-06, + "loss": 0.0958, + "step": 11447 + }, + { + "epoch": 4.79, + "learning_rate": 7.305794430819424e-06, + "loss": 0.1547, + "step": 11448 + }, + { + "epoch": 4.79, + "learning_rate": 7.304161752450227e-06, + "loss": 0.1457, + "step": 11449 + }, + { + "epoch": 4.79, + "learning_rate": 7.302529151570506e-06, + "loss": 0.1221, + "step": 11450 + }, + { + "epoch": 4.79, + "learning_rate": 7.3008966282271885e-06, + "loss": 0.1433, + "step": 11451 + }, + { + "epoch": 4.79, + "learning_rate": 7.299264182467206e-06, + "loss": 0.1421, + "step": 11452 + }, + { + "epoch": 4.8, + "learning_rate": 7.2976318143374755e-06, + "loss": 0.174, + "step": 11453 + }, + { + "epoch": 4.8, + "learning_rate": 7.295999523884921e-06, + "loss": 0.1891, + "step": 11454 + }, + { + "epoch": 4.8, + "learning_rate": 7.294367311156463e-06, + "loss": 0.1325, + "step": 11455 + }, + { + "epoch": 4.8, + "learning_rate": 7.292735176199013e-06, + "loss": 0.1817, + "step": 11456 + }, + { + "epoch": 4.8, + "learning_rate": 7.291103119059488e-06, + "loss": 0.1543, + "step": 11457 + }, + { + "epoch": 4.8, + "learning_rate": 7.289471139784803e-06, + "loss": 0.1808, + "step": 11458 + }, + { + "epoch": 4.8, + "learning_rate": 7.287839238421863e-06, + "loss": 0.1374, + "step": 11459 + }, + { + "epoch": 4.8, + "learning_rate": 7.286207415017579e-06, + "loss": 0.1189, + "step": 11460 + }, + { + "epoch": 4.8, + "learning_rate": 7.284575669618852e-06, + "loss": 0.1296, + "step": 11461 + }, + { + "epoch": 4.8, + "learning_rate": 7.282944002272587e-06, + "loss": 0.1559, + "step": 11462 + }, + { + "epoch": 4.8, + "learning_rate": 7.28131241302569e-06, + "loss": 0.1616, + "step": 11463 + }, + { + "epoch": 4.8, + "learning_rate": 7.279680901925051e-06, + "loss": 0.1827, + "step": 11464 + }, + { + "epoch": 4.8, + "learning_rate": 7.278049469017569e-06, + "loss": 0.1433, + "step": 11465 + }, + { + "epoch": 4.8, + "learning_rate": 7.276418114350145e-06, + "loss": 0.133, + "step": 11466 + }, + { + "epoch": 4.8, + "learning_rate": 7.274786837969663e-06, + "loss": 0.169, + "step": 11467 + }, + { + "epoch": 4.8, + "learning_rate": 7.273155639923013e-06, + "loss": 0.1406, + "step": 11468 + }, + { + "epoch": 4.8, + "learning_rate": 7.271524520257087e-06, + "loss": 0.1203, + "step": 11469 + }, + { + "epoch": 4.8, + "learning_rate": 7.269893479018766e-06, + "loss": 0.1757, + "step": 11470 + }, + { + "epoch": 4.8, + "learning_rate": 7.268262516254934e-06, + "loss": 0.1237, + "step": 11471 + }, + { + "epoch": 4.8, + "learning_rate": 7.2666316320124755e-06, + "loss": 0.1722, + "step": 11472 + }, + { + "epoch": 4.8, + "learning_rate": 7.265000826338261e-06, + "loss": 0.1364, + "step": 11473 + }, + { + "epoch": 4.8, + "learning_rate": 7.263370099279173e-06, + "loss": 0.1386, + "step": 11474 + }, + { + "epoch": 4.8, + "learning_rate": 7.2617394508820815e-06, + "loss": 0.1343, + "step": 11475 + }, + { + "epoch": 4.8, + "learning_rate": 7.26010888119386e-06, + "loss": 0.1714, + "step": 11476 + }, + { + "epoch": 4.81, + "learning_rate": 7.258478390261377e-06, + "loss": 0.156, + "step": 11477 + }, + { + "epoch": 4.81, + "learning_rate": 7.2568479781315e-06, + "loss": 0.1641, + "step": 11478 + }, + { + "epoch": 4.81, + "learning_rate": 7.255217644851096e-06, + "loss": 0.1244, + "step": 11479 + }, + { + "epoch": 4.81, + "learning_rate": 7.253587390467023e-06, + "loss": 0.1458, + "step": 11480 + }, + { + "epoch": 4.81, + "learning_rate": 7.251957215026142e-06, + "loss": 0.1585, + "step": 11481 + }, + { + "epoch": 4.81, + "learning_rate": 7.250327118575318e-06, + "loss": 0.1757, + "step": 11482 + }, + { + "epoch": 4.81, + "learning_rate": 7.2486971011614e-06, + "loss": 0.1324, + "step": 11483 + }, + { + "epoch": 4.81, + "learning_rate": 7.247067162831237e-06, + "loss": 0.1536, + "step": 11484 + }, + { + "epoch": 4.81, + "learning_rate": 7.245437303631687e-06, + "loss": 0.2021, + "step": 11485 + }, + { + "epoch": 4.81, + "learning_rate": 7.243807523609601e-06, + "loss": 0.1557, + "step": 11486 + }, + { + "epoch": 4.81, + "learning_rate": 7.242177822811818e-06, + "loss": 0.1699, + "step": 11487 + }, + { + "epoch": 4.81, + "learning_rate": 7.240548201285186e-06, + "loss": 0.1632, + "step": 11488 + }, + { + "epoch": 4.81, + "learning_rate": 7.238918659076548e-06, + "loss": 0.1633, + "step": 11489 + }, + { + "epoch": 4.81, + "learning_rate": 7.237289196232744e-06, + "loss": 0.1728, + "step": 11490 + }, + { + "epoch": 4.81, + "learning_rate": 7.235659812800611e-06, + "loss": 0.1672, + "step": 11491 + }, + { + "epoch": 4.81, + "learning_rate": 7.234030508826981e-06, + "loss": 0.1507, + "step": 11492 + }, + { + "epoch": 4.81, + "learning_rate": 7.232401284358692e-06, + "loss": 0.174, + "step": 11493 + }, + { + "epoch": 4.81, + "learning_rate": 7.230772139442572e-06, + "loss": 0.0923, + "step": 11494 + }, + { + "epoch": 4.81, + "learning_rate": 7.229143074125447e-06, + "loss": 0.1453, + "step": 11495 + }, + { + "epoch": 4.81, + "learning_rate": 7.2275140884541495e-06, + "loss": 0.1484, + "step": 11496 + }, + { + "epoch": 4.81, + "learning_rate": 7.225885182475496e-06, + "loss": 0.138, + "step": 11497 + }, + { + "epoch": 4.81, + "learning_rate": 7.224256356236312e-06, + "loss": 0.1439, + "step": 11498 + }, + { + "epoch": 4.81, + "learning_rate": 7.222627609783418e-06, + "loss": 0.1385, + "step": 11499 + }, + { + "epoch": 4.81, + "learning_rate": 7.2209989431636265e-06, + "loss": 0.1192, + "step": 11500 + }, + { + "epoch": 4.82, + "learning_rate": 7.219370356423754e-06, + "loss": 0.1667, + "step": 11501 + }, + { + "epoch": 4.82, + "learning_rate": 7.217741849610616e-06, + "loss": 0.1536, + "step": 11502 + }, + { + "epoch": 4.82, + "learning_rate": 7.216113422771016e-06, + "loss": 0.142, + "step": 11503 + }, + { + "epoch": 4.82, + "learning_rate": 7.214485075951771e-06, + "loss": 0.1855, + "step": 11504 + }, + { + "epoch": 4.82, + "learning_rate": 7.212856809199676e-06, + "loss": 0.1364, + "step": 11505 + }, + { + "epoch": 4.82, + "learning_rate": 7.211228622561538e-06, + "loss": 0.1593, + "step": 11506 + }, + { + "epoch": 4.82, + "learning_rate": 7.2096005160841605e-06, + "loss": 0.153, + "step": 11507 + }, + { + "epoch": 4.82, + "learning_rate": 7.207972489814339e-06, + "loss": 0.1313, + "step": 11508 + }, + { + "epoch": 4.82, + "learning_rate": 7.206344543798871e-06, + "loss": 0.1861, + "step": 11509 + }, + { + "epoch": 4.82, + "learning_rate": 7.204716678084551e-06, + "loss": 0.17, + "step": 11510 + }, + { + "epoch": 4.82, + "learning_rate": 7.203088892718168e-06, + "loss": 0.1709, + "step": 11511 + }, + { + "epoch": 4.82, + "learning_rate": 7.201461187746513e-06, + "loss": 0.0981, + "step": 11512 + }, + { + "epoch": 4.82, + "learning_rate": 7.199833563216375e-06, + "loss": 0.1039, + "step": 11513 + }, + { + "epoch": 4.82, + "learning_rate": 7.198206019174534e-06, + "loss": 0.176, + "step": 11514 + }, + { + "epoch": 4.82, + "learning_rate": 7.196578555667775e-06, + "loss": 0.1032, + "step": 11515 + }, + { + "epoch": 4.82, + "learning_rate": 7.194951172742879e-06, + "loss": 0.1818, + "step": 11516 + }, + { + "epoch": 4.82, + "learning_rate": 7.19332387044662e-06, + "loss": 0.1415, + "step": 11517 + }, + { + "epoch": 4.82, + "learning_rate": 7.191696648825778e-06, + "loss": 0.1941, + "step": 11518 + }, + { + "epoch": 4.82, + "learning_rate": 7.190069507927122e-06, + "loss": 0.1256, + "step": 11519 + }, + { + "epoch": 4.82, + "learning_rate": 7.1884424477974265e-06, + "loss": 0.1336, + "step": 11520 + }, + { + "epoch": 4.82, + "learning_rate": 7.186815468483457e-06, + "loss": 0.1809, + "step": 11521 + }, + { + "epoch": 4.82, + "learning_rate": 7.185188570031979e-06, + "loss": 0.1438, + "step": 11522 + }, + { + "epoch": 4.82, + "learning_rate": 7.183561752489761e-06, + "loss": 0.1717, + "step": 11523 + }, + { + "epoch": 4.82, + "learning_rate": 7.181935015903558e-06, + "loss": 0.1621, + "step": 11524 + }, + { + "epoch": 4.83, + "learning_rate": 7.180308360320134e-06, + "loss": 0.1906, + "step": 11525 + }, + { + "epoch": 4.83, + "learning_rate": 7.178681785786245e-06, + "loss": 0.1621, + "step": 11526 + }, + { + "epoch": 4.83, + "learning_rate": 7.177055292348646e-06, + "loss": 0.1375, + "step": 11527 + }, + { + "epoch": 4.83, + "learning_rate": 7.1754288800540836e-06, + "loss": 0.1286, + "step": 11528 + }, + { + "epoch": 4.83, + "learning_rate": 7.173802548949312e-06, + "loss": 0.1323, + "step": 11529 + }, + { + "epoch": 4.83, + "learning_rate": 7.172176299081078e-06, + "loss": 0.1918, + "step": 11530 + }, + { + "epoch": 4.83, + "learning_rate": 7.17055013049613e-06, + "loss": 0.1785, + "step": 11531 + }, + { + "epoch": 4.83, + "learning_rate": 7.168924043241205e-06, + "loss": 0.1526, + "step": 11532 + }, + { + "epoch": 4.83, + "learning_rate": 7.1672980373630455e-06, + "loss": 0.1177, + "step": 11533 + }, + { + "epoch": 4.83, + "learning_rate": 7.165672112908393e-06, + "loss": 0.0849, + "step": 11534 + }, + { + "epoch": 4.83, + "learning_rate": 7.164046269923978e-06, + "loss": 0.2089, + "step": 11535 + }, + { + "epoch": 4.83, + "learning_rate": 7.162420508456536e-06, + "loss": 0.1855, + "step": 11536 + }, + { + "epoch": 4.83, + "learning_rate": 7.160794828552802e-06, + "loss": 0.1502, + "step": 11537 + }, + { + "epoch": 4.83, + "learning_rate": 7.159169230259497e-06, + "loss": 0.1467, + "step": 11538 + }, + { + "epoch": 4.83, + "learning_rate": 7.157543713623353e-06, + "loss": 0.1252, + "step": 11539 + }, + { + "epoch": 4.83, + "learning_rate": 7.155918278691094e-06, + "loss": 0.1651, + "step": 11540 + }, + { + "epoch": 4.83, + "learning_rate": 7.1542929255094386e-06, + "loss": 0.1559, + "step": 11541 + }, + { + "epoch": 4.83, + "learning_rate": 7.1526676541251075e-06, + "loss": 0.1774, + "step": 11542 + }, + { + "epoch": 4.83, + "learning_rate": 7.151042464584821e-06, + "loss": 0.1794, + "step": 11543 + }, + { + "epoch": 4.83, + "learning_rate": 7.149417356935289e-06, + "loss": 0.1324, + "step": 11544 + }, + { + "epoch": 4.83, + "learning_rate": 7.1477923312232235e-06, + "loss": 0.1458, + "step": 11545 + }, + { + "epoch": 4.83, + "learning_rate": 7.14616738749534e-06, + "loss": 0.1399, + "step": 11546 + }, + { + "epoch": 4.83, + "learning_rate": 7.1445425257983395e-06, + "loss": 0.15, + "step": 11547 + }, + { + "epoch": 4.83, + "learning_rate": 7.14291774617893e-06, + "loss": 0.1499, + "step": 11548 + }, + { + "epoch": 4.84, + "learning_rate": 7.141293048683816e-06, + "loss": 0.1482, + "step": 11549 + }, + { + "epoch": 4.84, + "learning_rate": 7.1396684333597e-06, + "loss": 0.1563, + "step": 11550 + }, + { + "epoch": 4.84, + "learning_rate": 7.138043900253275e-06, + "loss": 0.1151, + "step": 11551 + }, + { + "epoch": 4.84, + "learning_rate": 7.136419449411236e-06, + "loss": 0.1529, + "step": 11552 + }, + { + "epoch": 4.84, + "learning_rate": 7.134795080880278e-06, + "loss": 0.1641, + "step": 11553 + }, + { + "epoch": 4.84, + "learning_rate": 7.133170794707096e-06, + "loss": 0.1526, + "step": 11554 + }, + { + "epoch": 4.84, + "learning_rate": 7.131546590938373e-06, + "loss": 0.1107, + "step": 11555 + }, + { + "epoch": 4.84, + "learning_rate": 7.129922469620798e-06, + "loss": 0.1722, + "step": 11556 + }, + { + "epoch": 4.84, + "learning_rate": 7.128298430801057e-06, + "loss": 0.1138, + "step": 11557 + }, + { + "epoch": 4.84, + "learning_rate": 7.1266744745258264e-06, + "loss": 0.1476, + "step": 11558 + }, + { + "epoch": 4.84, + "learning_rate": 7.12505060084179e-06, + "loss": 0.1776, + "step": 11559 + }, + { + "epoch": 4.84, + "learning_rate": 7.123426809795622e-06, + "loss": 0.1628, + "step": 11560 + }, + { + "epoch": 4.84, + "learning_rate": 7.121803101433998e-06, + "loss": 0.1727, + "step": 11561 + }, + { + "epoch": 4.84, + "learning_rate": 7.1201794758035905e-06, + "loss": 0.1536, + "step": 11562 + }, + { + "epoch": 4.84, + "learning_rate": 7.118555932951067e-06, + "loss": 0.1326, + "step": 11563 + }, + { + "epoch": 4.84, + "learning_rate": 7.116932472923098e-06, + "loss": 0.1376, + "step": 11564 + }, + { + "epoch": 4.84, + "learning_rate": 7.115309095766345e-06, + "loss": 0.1419, + "step": 11565 + }, + { + "epoch": 4.84, + "learning_rate": 7.113685801527473e-06, + "loss": 0.111, + "step": 11566 + }, + { + "epoch": 4.84, + "learning_rate": 7.112062590253143e-06, + "loss": 0.1201, + "step": 11567 + }, + { + "epoch": 4.84, + "learning_rate": 7.110439461990008e-06, + "loss": 0.1694, + "step": 11568 + }, + { + "epoch": 4.84, + "learning_rate": 7.108816416784728e-06, + "loss": 0.1499, + "step": 11569 + }, + { + "epoch": 4.84, + "learning_rate": 7.107193454683956e-06, + "loss": 0.1669, + "step": 11570 + }, + { + "epoch": 4.84, + "learning_rate": 7.105570575734339e-06, + "loss": 0.1315, + "step": 11571 + }, + { + "epoch": 4.84, + "learning_rate": 7.103947779982532e-06, + "loss": 0.1531, + "step": 11572 + }, + { + "epoch": 4.85, + "learning_rate": 7.102325067475171e-06, + "loss": 0.13, + "step": 11573 + }, + { + "epoch": 4.85, + "learning_rate": 7.100702438258905e-06, + "loss": 0.1609, + "step": 11574 + }, + { + "epoch": 4.85, + "learning_rate": 7.099079892380376e-06, + "loss": 0.1277, + "step": 11575 + }, + { + "epoch": 4.85, + "learning_rate": 7.097457429886221e-06, + "loss": 0.1312, + "step": 11576 + }, + { + "epoch": 4.85, + "learning_rate": 7.095835050823074e-06, + "loss": 0.1627, + "step": 11577 + }, + { + "epoch": 4.85, + "learning_rate": 7.094212755237576e-06, + "loss": 0.1633, + "step": 11578 + }, + { + "epoch": 4.85, + "learning_rate": 7.092590543176351e-06, + "loss": 0.156, + "step": 11579 + }, + { + "epoch": 4.85, + "learning_rate": 7.09096841468603e-06, + "loss": 0.1543, + "step": 11580 + }, + { + "epoch": 4.85, + "learning_rate": 7.089346369813244e-06, + "loss": 0.1305, + "step": 11581 + }, + { + "epoch": 4.85, + "learning_rate": 7.087724408604611e-06, + "loss": 0.15, + "step": 11582 + }, + { + "epoch": 4.85, + "learning_rate": 7.086102531106755e-06, + "loss": 0.1845, + "step": 11583 + }, + { + "epoch": 4.85, + "learning_rate": 7.084480737366297e-06, + "loss": 0.1297, + "step": 11584 + }, + { + "epoch": 4.85, + "learning_rate": 7.082859027429853e-06, + "loss": 0.1236, + "step": 11585 + }, + { + "epoch": 4.85, + "learning_rate": 7.0812374013440365e-06, + "loss": 0.1535, + "step": 11586 + }, + { + "epoch": 4.85, + "learning_rate": 7.079615859155464e-06, + "loss": 0.1349, + "step": 11587 + }, + { + "epoch": 4.85, + "learning_rate": 7.077994400910739e-06, + "loss": 0.1956, + "step": 11588 + }, + { + "epoch": 4.85, + "learning_rate": 7.0763730266564725e-06, + "loss": 0.1305, + "step": 11589 + }, + { + "epoch": 4.85, + "learning_rate": 7.0747517364392694e-06, + "loss": 0.1581, + "step": 11590 + }, + { + "epoch": 4.85, + "learning_rate": 7.073130530305733e-06, + "loss": 0.149, + "step": 11591 + }, + { + "epoch": 4.85, + "learning_rate": 7.0715094083024604e-06, + "loss": 0.1399, + "step": 11592 + }, + { + "epoch": 4.85, + "learning_rate": 7.069888370476051e-06, + "loss": 0.174, + "step": 11593 + }, + { + "epoch": 4.85, + "learning_rate": 7.068267416873102e-06, + "loss": 0.1423, + "step": 11594 + }, + { + "epoch": 4.85, + "learning_rate": 7.066646547540208e-06, + "loss": 0.148, + "step": 11595 + }, + { + "epoch": 4.85, + "learning_rate": 7.065025762523952e-06, + "loss": 0.174, + "step": 11596 + }, + { + "epoch": 4.86, + "learning_rate": 7.063405061870925e-06, + "loss": 0.1655, + "step": 11597 + }, + { + "epoch": 4.86, + "learning_rate": 7.061784445627715e-06, + "loss": 0.1518, + "step": 11598 + }, + { + "epoch": 4.86, + "learning_rate": 7.060163913840903e-06, + "loss": 0.1504, + "step": 11599 + }, + { + "epoch": 4.86, + "learning_rate": 7.058543466557069e-06, + "loss": 0.1705, + "step": 11600 + }, + { + "epoch": 4.86, + "learning_rate": 7.056923103822795e-06, + "loss": 0.1463, + "step": 11601 + }, + { + "epoch": 4.86, + "learning_rate": 7.0553028256846555e-06, + "loss": 0.1493, + "step": 11602 + }, + { + "epoch": 4.86, + "learning_rate": 7.053682632189221e-06, + "loss": 0.1341, + "step": 11603 + }, + { + "epoch": 4.86, + "learning_rate": 7.052062523383065e-06, + "loss": 0.1391, + "step": 11604 + }, + { + "epoch": 4.86, + "learning_rate": 7.050442499312757e-06, + "loss": 0.1184, + "step": 11605 + }, + { + "epoch": 4.86, + "learning_rate": 7.048822560024862e-06, + "loss": 0.1176, + "step": 11606 + }, + { + "epoch": 4.86, + "learning_rate": 7.047202705565942e-06, + "loss": 0.165, + "step": 11607 + }, + { + "epoch": 4.86, + "learning_rate": 7.045582935982562e-06, + "loss": 0.1549, + "step": 11608 + }, + { + "epoch": 4.86, + "learning_rate": 7.0439632513212775e-06, + "loss": 0.153, + "step": 11609 + }, + { + "epoch": 4.86, + "learning_rate": 7.042343651628646e-06, + "loss": 0.1586, + "step": 11610 + }, + { + "epoch": 4.86, + "learning_rate": 7.040724136951224e-06, + "loss": 0.1513, + "step": 11611 + }, + { + "epoch": 4.86, + "learning_rate": 7.0391047073355575e-06, + "loss": 0.1322, + "step": 11612 + }, + { + "epoch": 4.86, + "learning_rate": 7.0374853628282e-06, + "loss": 0.1278, + "step": 11613 + }, + { + "epoch": 4.86, + "learning_rate": 7.035866103475698e-06, + "loss": 0.1393, + "step": 11614 + }, + { + "epoch": 4.86, + "learning_rate": 7.034246929324593e-06, + "loss": 0.1838, + "step": 11615 + }, + { + "epoch": 4.86, + "learning_rate": 7.032627840421429e-06, + "loss": 0.1764, + "step": 11616 + }, + { + "epoch": 4.86, + "learning_rate": 7.031008836812743e-06, + "loss": 0.1596, + "step": 11617 + }, + { + "epoch": 4.86, + "learning_rate": 7.029389918545078e-06, + "loss": 0.1471, + "step": 11618 + }, + { + "epoch": 4.86, + "learning_rate": 7.0277710856649614e-06, + "loss": 0.1407, + "step": 11619 + }, + { + "epoch": 4.86, + "learning_rate": 7.026152338218925e-06, + "loss": 0.1156, + "step": 11620 + }, + { + "epoch": 4.87, + "learning_rate": 7.024533676253502e-06, + "loss": 0.1556, + "step": 11621 + }, + { + "epoch": 4.87, + "learning_rate": 7.022915099815219e-06, + "loss": 0.1736, + "step": 11622 + }, + { + "epoch": 4.87, + "learning_rate": 7.021296608950599e-06, + "loss": 0.1521, + "step": 11623 + }, + { + "epoch": 4.87, + "learning_rate": 7.019678203706164e-06, + "loss": 0.1597, + "step": 11624 + }, + { + "epoch": 4.87, + "learning_rate": 7.018059884128437e-06, + "loss": 0.1699, + "step": 11625 + }, + { + "epoch": 4.87, + "learning_rate": 7.01644165026393e-06, + "loss": 0.1904, + "step": 11626 + }, + { + "epoch": 4.87, + "learning_rate": 7.014823502159159e-06, + "loss": 0.1579, + "step": 11627 + }, + { + "epoch": 4.87, + "learning_rate": 7.0132054398606405e-06, + "loss": 0.1672, + "step": 11628 + }, + { + "epoch": 4.87, + "learning_rate": 7.011587463414879e-06, + "loss": 0.1292, + "step": 11629 + }, + { + "epoch": 4.87, + "learning_rate": 7.009969572868385e-06, + "loss": 0.1356, + "step": 11630 + }, + { + "epoch": 4.87, + "learning_rate": 7.008351768267661e-06, + "loss": 0.1077, + "step": 11631 + }, + { + "epoch": 4.87, + "learning_rate": 7.006734049659213e-06, + "loss": 0.1405, + "step": 11632 + }, + { + "epoch": 4.87, + "learning_rate": 7.005116417089537e-06, + "loss": 0.1182, + "step": 11633 + }, + { + "epoch": 4.87, + "learning_rate": 7.0034988706051325e-06, + "loss": 0.169, + "step": 11634 + }, + { + "epoch": 4.87, + "learning_rate": 7.001881410252494e-06, + "loss": 0.1308, + "step": 11635 + }, + { + "epoch": 4.87, + "learning_rate": 7.000264036078113e-06, + "loss": 0.1397, + "step": 11636 + }, + { + "epoch": 4.87, + "learning_rate": 6.9986467481284805e-06, + "loss": 0.1342, + "step": 11637 + }, + { + "epoch": 4.87, + "learning_rate": 6.997029546450085e-06, + "loss": 0.112, + "step": 11638 + }, + { + "epoch": 4.87, + "learning_rate": 6.99541243108941e-06, + "loss": 0.1643, + "step": 11639 + }, + { + "epoch": 4.87, + "learning_rate": 6.9937954020929414e-06, + "loss": 0.1515, + "step": 11640 + }, + { + "epoch": 4.87, + "learning_rate": 6.992178459507153e-06, + "loss": 0.1187, + "step": 11641 + }, + { + "epoch": 4.87, + "learning_rate": 6.990561603378524e-06, + "loss": 0.144, + "step": 11642 + }, + { + "epoch": 4.87, + "learning_rate": 6.9889448337535345e-06, + "loss": 0.1462, + "step": 11643 + }, + { + "epoch": 4.88, + "learning_rate": 6.987328150678652e-06, + "loss": 0.1524, + "step": 11644 + }, + { + "epoch": 4.88, + "learning_rate": 6.985711554200347e-06, + "loss": 0.1162, + "step": 11645 + }, + { + "epoch": 4.88, + "learning_rate": 6.984095044365092e-06, + "loss": 0.154, + "step": 11646 + }, + { + "epoch": 4.88, + "learning_rate": 6.982478621219346e-06, + "loss": 0.1642, + "step": 11647 + }, + { + "epoch": 4.88, + "learning_rate": 6.980862284809575e-06, + "loss": 0.1614, + "step": 11648 + }, + { + "epoch": 4.88, + "learning_rate": 6.97924603518224e-06, + "loss": 0.1626, + "step": 11649 + }, + { + "epoch": 4.88, + "learning_rate": 6.9776298723837956e-06, + "loss": 0.1456, + "step": 11650 + }, + { + "epoch": 4.88, + "learning_rate": 6.976013796460698e-06, + "loss": 0.1292, + "step": 11651 + }, + { + "epoch": 4.88, + "learning_rate": 6.974397807459402e-06, + "loss": 0.1422, + "step": 11652 + }, + { + "epoch": 4.88, + "learning_rate": 6.972781905426354e-06, + "loss": 0.133, + "step": 11653 + }, + { + "epoch": 4.88, + "learning_rate": 6.9711660904080055e-06, + "loss": 0.154, + "step": 11654 + }, + { + "epoch": 4.88, + "learning_rate": 6.9695503624508005e-06, + "loss": 0.1427, + "step": 11655 + }, + { + "epoch": 4.88, + "learning_rate": 6.96793472160118e-06, + "loss": 0.1513, + "step": 11656 + }, + { + "epoch": 4.88, + "learning_rate": 6.966319167905584e-06, + "loss": 0.1249, + "step": 11657 + }, + { + "epoch": 4.88, + "learning_rate": 6.964703701410455e-06, + "loss": 0.1561, + "step": 11658 + }, + { + "epoch": 4.88, + "learning_rate": 6.963088322162221e-06, + "loss": 0.139, + "step": 11659 + }, + { + "epoch": 4.88, + "learning_rate": 6.96147303020732e-06, + "loss": 0.194, + "step": 11660 + }, + { + "epoch": 4.88, + "learning_rate": 6.95985782559218e-06, + "loss": 0.1512, + "step": 11661 + }, + { + "epoch": 4.88, + "learning_rate": 6.958242708363231e-06, + "loss": 0.2009, + "step": 11662 + }, + { + "epoch": 4.88, + "learning_rate": 6.956627678566898e-06, + "loss": 0.1388, + "step": 11663 + }, + { + "epoch": 4.88, + "learning_rate": 6.955012736249598e-06, + "loss": 0.1439, + "step": 11664 + }, + { + "epoch": 4.88, + "learning_rate": 6.953397881457754e-06, + "loss": 0.1572, + "step": 11665 + }, + { + "epoch": 4.88, + "learning_rate": 6.951783114237786e-06, + "loss": 0.1202, + "step": 11666 + }, + { + "epoch": 4.88, + "learning_rate": 6.950168434636106e-06, + "loss": 0.1251, + "step": 11667 + }, + { + "epoch": 4.89, + "learning_rate": 6.94855384269913e-06, + "loss": 0.1462, + "step": 11668 + }, + { + "epoch": 4.89, + "learning_rate": 6.946939338473267e-06, + "loss": 0.1682, + "step": 11669 + }, + { + "epoch": 4.89, + "learning_rate": 6.945324922004919e-06, + "loss": 0.1335, + "step": 11670 + }, + { + "epoch": 4.89, + "learning_rate": 6.943710593340498e-06, + "loss": 0.1708, + "step": 11671 + }, + { + "epoch": 4.89, + "learning_rate": 6.9420963525264025e-06, + "loss": 0.1784, + "step": 11672 + }, + { + "epoch": 4.89, + "learning_rate": 6.940482199609036e-06, + "loss": 0.1363, + "step": 11673 + }, + { + "epoch": 4.89, + "learning_rate": 6.938868134634793e-06, + "loss": 0.1453, + "step": 11674 + }, + { + "epoch": 4.89, + "learning_rate": 6.937254157650068e-06, + "loss": 0.1673, + "step": 11675 + }, + { + "epoch": 4.89, + "learning_rate": 6.935640268701256e-06, + "loss": 0.1203, + "step": 11676 + }, + { + "epoch": 4.89, + "learning_rate": 6.934026467834745e-06, + "loss": 0.1493, + "step": 11677 + }, + { + "epoch": 4.89, + "learning_rate": 6.93241275509692e-06, + "loss": 0.1621, + "step": 11678 + }, + { + "epoch": 4.89, + "learning_rate": 6.930799130534172e-06, + "loss": 0.1438, + "step": 11679 + }, + { + "epoch": 4.89, + "learning_rate": 6.929185594192877e-06, + "loss": 0.1704, + "step": 11680 + }, + { + "epoch": 4.89, + "learning_rate": 6.927572146119416e-06, + "loss": 0.1914, + "step": 11681 + }, + { + "epoch": 4.89, + "learning_rate": 6.925958786360172e-06, + "loss": 0.1893, + "step": 11682 + }, + { + "epoch": 4.89, + "learning_rate": 6.92434551496151e-06, + "loss": 0.115, + "step": 11683 + }, + { + "epoch": 4.89, + "learning_rate": 6.922732331969808e-06, + "loss": 0.1342, + "step": 11684 + }, + { + "epoch": 4.89, + "learning_rate": 6.921119237431436e-06, + "loss": 0.1299, + "step": 11685 + }, + { + "epoch": 4.89, + "learning_rate": 6.9195062313927605e-06, + "loss": 0.1314, + "step": 11686 + }, + { + "epoch": 4.89, + "learning_rate": 6.917893313900144e-06, + "loss": 0.1386, + "step": 11687 + }, + { + "epoch": 4.89, + "learning_rate": 6.916280484999946e-06, + "loss": 0.152, + "step": 11688 + }, + { + "epoch": 4.89, + "learning_rate": 6.914667744738531e-06, + "loss": 0.1339, + "step": 11689 + }, + { + "epoch": 4.89, + "learning_rate": 6.9130550931622555e-06, + "loss": 0.1567, + "step": 11690 + }, + { + "epoch": 4.89, + "learning_rate": 6.911442530317469e-06, + "loss": 0.1512, + "step": 11691 + }, + { + "epoch": 4.9, + "learning_rate": 6.909830056250527e-06, + "loss": 0.1504, + "step": 11692 + }, + { + "epoch": 4.9, + "learning_rate": 6.90821767100778e-06, + "loss": 0.192, + "step": 11693 + }, + { + "epoch": 4.9, + "learning_rate": 6.90660537463557e-06, + "loss": 0.1369, + "step": 11694 + }, + { + "epoch": 4.9, + "learning_rate": 6.904993167180243e-06, + "loss": 0.1093, + "step": 11695 + }, + { + "epoch": 4.9, + "learning_rate": 6.9033810486881445e-06, + "loss": 0.159, + "step": 11696 + }, + { + "epoch": 4.9, + "learning_rate": 6.901769019205606e-06, + "loss": 0.1447, + "step": 11697 + }, + { + "epoch": 4.9, + "learning_rate": 6.900157078778969e-06, + "loss": 0.1699, + "step": 11698 + }, + { + "epoch": 4.9, + "learning_rate": 6.898545227454567e-06, + "loss": 0.1377, + "step": 11699 + }, + { + "epoch": 4.9, + "learning_rate": 6.89693346527873e-06, + "loss": 0.1256, + "step": 11700 + }, + { + "epoch": 4.9, + "learning_rate": 6.895321792297785e-06, + "loss": 0.1856, + "step": 11701 + }, + { + "epoch": 4.9, + "learning_rate": 6.893710208558061e-06, + "loss": 0.1304, + "step": 11702 + }, + { + "epoch": 4.9, + "learning_rate": 6.892098714105882e-06, + "loss": 0.1064, + "step": 11703 + }, + { + "epoch": 4.9, + "learning_rate": 6.890487308987567e-06, + "loss": 0.1524, + "step": 11704 + }, + { + "epoch": 4.9, + "learning_rate": 6.888875993249433e-06, + "loss": 0.1516, + "step": 11705 + }, + { + "epoch": 4.9, + "learning_rate": 6.887264766937801e-06, + "loss": 0.1015, + "step": 11706 + }, + { + "epoch": 4.9, + "learning_rate": 6.885653630098978e-06, + "loss": 0.16, + "step": 11707 + }, + { + "epoch": 4.9, + "learning_rate": 6.884042582779285e-06, + "loss": 0.1747, + "step": 11708 + }, + { + "epoch": 4.9, + "learning_rate": 6.882431625025016e-06, + "loss": 0.1781, + "step": 11709 + }, + { + "epoch": 4.9, + "learning_rate": 6.880820756882486e-06, + "loss": 0.1813, + "step": 11710 + }, + { + "epoch": 4.9, + "learning_rate": 6.879209978397993e-06, + "loss": 0.1389, + "step": 11711 + }, + { + "epoch": 4.9, + "learning_rate": 6.87759928961784e-06, + "loss": 0.1352, + "step": 11712 + }, + { + "epoch": 4.9, + "learning_rate": 6.875988690588325e-06, + "loss": 0.2103, + "step": 11713 + }, + { + "epoch": 4.9, + "learning_rate": 6.8743781813557455e-06, + "loss": 0.1271, + "step": 11714 + }, + { + "epoch": 4.9, + "learning_rate": 6.872767761966388e-06, + "loss": 0.1428, + "step": 11715 + }, + { + "epoch": 4.91, + "learning_rate": 6.8711574324665476e-06, + "loss": 0.1631, + "step": 11716 + }, + { + "epoch": 4.91, + "learning_rate": 6.869547192902511e-06, + "loss": 0.1268, + "step": 11717 + }, + { + "epoch": 4.91, + "learning_rate": 6.867937043320562e-06, + "loss": 0.1482, + "step": 11718 + }, + { + "epoch": 4.91, + "learning_rate": 6.866326983766982e-06, + "loss": 0.1294, + "step": 11719 + }, + { + "epoch": 4.91, + "learning_rate": 6.8647170142880545e-06, + "loss": 0.1384, + "step": 11720 + }, + { + "epoch": 4.91, + "learning_rate": 6.863107134930052e-06, + "loss": 0.13, + "step": 11721 + }, + { + "epoch": 4.91, + "learning_rate": 6.861497345739252e-06, + "loss": 0.111, + "step": 11722 + }, + { + "epoch": 4.91, + "learning_rate": 6.859887646761927e-06, + "loss": 0.152, + "step": 11723 + }, + { + "epoch": 4.91, + "learning_rate": 6.858278038044344e-06, + "loss": 0.1708, + "step": 11724 + }, + { + "epoch": 4.91, + "learning_rate": 6.85666851963277e-06, + "loss": 0.1705, + "step": 11725 + }, + { + "epoch": 4.91, + "learning_rate": 6.855059091573472e-06, + "loss": 0.1319, + "step": 11726 + }, + { + "epoch": 4.91, + "learning_rate": 6.853449753912708e-06, + "loss": 0.1653, + "step": 11727 + }, + { + "epoch": 4.91, + "learning_rate": 6.851840506696738e-06, + "loss": 0.1324, + "step": 11728 + }, + { + "epoch": 4.91, + "learning_rate": 6.8502313499718206e-06, + "loss": 0.1431, + "step": 11729 + }, + { + "epoch": 4.91, + "learning_rate": 6.848622283784209e-06, + "loss": 0.1088, + "step": 11730 + }, + { + "epoch": 4.91, + "learning_rate": 6.847013308180155e-06, + "loss": 0.1375, + "step": 11731 + }, + { + "epoch": 4.91, + "learning_rate": 6.845404423205903e-06, + "loss": 0.1367, + "step": 11732 + }, + { + "epoch": 4.91, + "learning_rate": 6.843795628907699e-06, + "loss": 0.1468, + "step": 11733 + }, + { + "epoch": 4.91, + "learning_rate": 6.842186925331793e-06, + "loss": 0.1308, + "step": 11734 + }, + { + "epoch": 4.91, + "learning_rate": 6.8405783125244196e-06, + "loss": 0.1238, + "step": 11735 + }, + { + "epoch": 4.91, + "learning_rate": 6.838969790531817e-06, + "loss": 0.1603, + "step": 11736 + }, + { + "epoch": 4.91, + "learning_rate": 6.837361359400227e-06, + "loss": 0.1417, + "step": 11737 + }, + { + "epoch": 4.91, + "learning_rate": 6.835753019175875e-06, + "loss": 0.1458, + "step": 11738 + }, + { + "epoch": 4.91, + "learning_rate": 6.834144769904996e-06, + "loss": 0.16, + "step": 11739 + }, + { + "epoch": 4.92, + "learning_rate": 6.832536611633816e-06, + "loss": 0.1393, + "step": 11740 + }, + { + "epoch": 4.92, + "learning_rate": 6.830928544408559e-06, + "loss": 0.1548, + "step": 11741 + }, + { + "epoch": 4.92, + "learning_rate": 6.82932056827545e-06, + "loss": 0.1411, + "step": 11742 + }, + { + "epoch": 4.92, + "learning_rate": 6.827712683280707e-06, + "loss": 0.1808, + "step": 11743 + }, + { + "epoch": 4.92, + "learning_rate": 6.8261048894705505e-06, + "loss": 0.181, + "step": 11744 + }, + { + "epoch": 4.92, + "learning_rate": 6.824497186891189e-06, + "loss": 0.1412, + "step": 11745 + }, + { + "epoch": 4.92, + "learning_rate": 6.822889575588838e-06, + "loss": 0.1402, + "step": 11746 + }, + { + "epoch": 4.92, + "learning_rate": 6.8212820556097105e-06, + "loss": 0.1246, + "step": 11747 + }, + { + "epoch": 4.92, + "learning_rate": 6.819674627000006e-06, + "loss": 0.1237, + "step": 11748 + }, + { + "epoch": 4.92, + "learning_rate": 6.818067289805934e-06, + "loss": 0.1781, + "step": 11749 + }, + { + "epoch": 4.92, + "learning_rate": 6.816460044073694e-06, + "loss": 0.144, + "step": 11750 + }, + { + "epoch": 4.92, + "learning_rate": 6.8148528898494845e-06, + "loss": 0.1312, + "step": 11751 + }, + { + "epoch": 4.92, + "learning_rate": 6.813245827179502e-06, + "loss": 0.1339, + "step": 11752 + }, + { + "epoch": 4.92, + "learning_rate": 6.811638856109941e-06, + "loss": 0.1753, + "step": 11753 + }, + { + "epoch": 4.92, + "learning_rate": 6.810031976686994e-06, + "loss": 0.1361, + "step": 11754 + }, + { + "epoch": 4.92, + "learning_rate": 6.808425188956847e-06, + "loss": 0.1599, + "step": 11755 + }, + { + "epoch": 4.92, + "learning_rate": 6.8068184929656815e-06, + "loss": 0.1234, + "step": 11756 + }, + { + "epoch": 4.92, + "learning_rate": 6.8052118887596865e-06, + "loss": 0.1611, + "step": 11757 + }, + { + "epoch": 4.92, + "learning_rate": 6.803605376385042e-06, + "loss": 0.1567, + "step": 11758 + }, + { + "epoch": 4.92, + "learning_rate": 6.801998955887923e-06, + "loss": 0.1952, + "step": 11759 + }, + { + "epoch": 4.92, + "learning_rate": 6.800392627314506e-06, + "loss": 0.1309, + "step": 11760 + }, + { + "epoch": 4.92, + "learning_rate": 6.7987863907109656e-06, + "loss": 0.1267, + "step": 11761 + }, + { + "epoch": 4.92, + "learning_rate": 6.797180246123467e-06, + "loss": 0.2057, + "step": 11762 + }, + { + "epoch": 4.92, + "learning_rate": 6.795574193598182e-06, + "loss": 0.182, + "step": 11763 + }, + { + "epoch": 4.93, + "learning_rate": 6.793968233181274e-06, + "loss": 0.1429, + "step": 11764 + }, + { + "epoch": 4.93, + "learning_rate": 6.792362364918902e-06, + "loss": 0.1341, + "step": 11765 + }, + { + "epoch": 4.93, + "learning_rate": 6.790756588857228e-06, + "loss": 0.1332, + "step": 11766 + }, + { + "epoch": 4.93, + "learning_rate": 6.7891509050424095e-06, + "loss": 0.146, + "step": 11767 + }, + { + "epoch": 4.93, + "learning_rate": 6.787545313520597e-06, + "loss": 0.1507, + "step": 11768 + }, + { + "epoch": 4.93, + "learning_rate": 6.785939814337944e-06, + "loss": 0.1471, + "step": 11769 + }, + { + "epoch": 4.93, + "learning_rate": 6.7843344075406005e-06, + "loss": 0.1545, + "step": 11770 + }, + { + "epoch": 4.93, + "learning_rate": 6.782729093174709e-06, + "loss": 0.1753, + "step": 11771 + }, + { + "epoch": 4.93, + "learning_rate": 6.781123871286416e-06, + "loss": 0.1701, + "step": 11772 + }, + { + "epoch": 4.93, + "learning_rate": 6.77951874192186e-06, + "loss": 0.1585, + "step": 11773 + }, + { + "epoch": 4.93, + "learning_rate": 6.777913705127181e-06, + "loss": 0.1698, + "step": 11774 + }, + { + "epoch": 4.93, + "learning_rate": 6.776308760948514e-06, + "loss": 0.1194, + "step": 11775 + }, + { + "epoch": 4.93, + "learning_rate": 6.774703909431992e-06, + "loss": 0.1379, + "step": 11776 + }, + { + "epoch": 4.93, + "learning_rate": 6.773099150623742e-06, + "loss": 0.1164, + "step": 11777 + }, + { + "epoch": 4.93, + "learning_rate": 6.771494484569895e-06, + "loss": 0.1853, + "step": 11778 + }, + { + "epoch": 4.93, + "learning_rate": 6.7698899113165715e-06, + "loss": 0.1434, + "step": 11779 + }, + { + "epoch": 4.93, + "learning_rate": 6.768285430909896e-06, + "loss": 0.1482, + "step": 11780 + }, + { + "epoch": 4.93, + "learning_rate": 6.76668104339599e-06, + "loss": 0.1466, + "step": 11781 + }, + { + "epoch": 4.93, + "learning_rate": 6.765076748820968e-06, + "loss": 0.1943, + "step": 11782 + }, + { + "epoch": 4.93, + "learning_rate": 6.763472547230943e-06, + "loss": 0.1793, + "step": 11783 + }, + { + "epoch": 4.93, + "learning_rate": 6.761868438672028e-06, + "loss": 0.1735, + "step": 11784 + }, + { + "epoch": 4.93, + "learning_rate": 6.760264423190332e-06, + "loss": 0.1779, + "step": 11785 + }, + { + "epoch": 4.93, + "learning_rate": 6.75866050083196e-06, + "loss": 0.1505, + "step": 11786 + }, + { + "epoch": 4.93, + "learning_rate": 6.757056671643014e-06, + "loss": 0.1513, + "step": 11787 + }, + { + "epoch": 4.94, + "learning_rate": 6.7554529356696e-06, + "loss": 0.1411, + "step": 11788 + }, + { + "epoch": 4.94, + "learning_rate": 6.753849292957808e-06, + "loss": 0.1627, + "step": 11789 + }, + { + "epoch": 4.94, + "learning_rate": 6.752245743553738e-06, + "loss": 0.1364, + "step": 11790 + }, + { + "epoch": 4.94, + "learning_rate": 6.750642287503486e-06, + "loss": 0.1319, + "step": 11791 + }, + { + "epoch": 4.94, + "learning_rate": 6.749038924853134e-06, + "loss": 0.1318, + "step": 11792 + }, + { + "epoch": 4.94, + "learning_rate": 6.747435655648774e-06, + "loss": 0.1795, + "step": 11793 + }, + { + "epoch": 4.94, + "learning_rate": 6.745832479936492e-06, + "loss": 0.1398, + "step": 11794 + }, + { + "epoch": 4.94, + "learning_rate": 6.744229397762365e-06, + "loss": 0.1839, + "step": 11795 + }, + { + "epoch": 4.94, + "learning_rate": 6.742626409172475e-06, + "loss": 0.1448, + "step": 11796 + }, + { + "epoch": 4.94, + "learning_rate": 6.741023514212901e-06, + "loss": 0.1557, + "step": 11797 + }, + { + "epoch": 4.94, + "learning_rate": 6.739420712929712e-06, + "loss": 0.139, + "step": 11798 + }, + { + "epoch": 4.94, + "learning_rate": 6.7378180053689854e-06, + "loss": 0.1092, + "step": 11799 + }, + { + "epoch": 4.94, + "learning_rate": 6.736215391576782e-06, + "loss": 0.1896, + "step": 11800 + }, + { + "epoch": 4.94, + "learning_rate": 6.734612871599169e-06, + "loss": 0.1533, + "step": 11801 + }, + { + "epoch": 4.94, + "learning_rate": 6.733010445482214e-06, + "loss": 0.1236, + "step": 11802 + }, + { + "epoch": 4.94, + "learning_rate": 6.7314081132719735e-06, + "loss": 0.1565, + "step": 11803 + }, + { + "epoch": 4.94, + "learning_rate": 6.729805875014506e-06, + "loss": 0.1567, + "step": 11804 + }, + { + "epoch": 4.94, + "learning_rate": 6.728203730755868e-06, + "loss": 0.1762, + "step": 11805 + }, + { + "epoch": 4.94, + "learning_rate": 6.726601680542109e-06, + "loss": 0.1186, + "step": 11806 + }, + { + "epoch": 4.94, + "learning_rate": 6.724999724419279e-06, + "loss": 0.1542, + "step": 11807 + }, + { + "epoch": 4.94, + "learning_rate": 6.723397862433429e-06, + "loss": 0.1656, + "step": 11808 + }, + { + "epoch": 4.94, + "learning_rate": 6.721796094630597e-06, + "loss": 0.1587, + "step": 11809 + }, + { + "epoch": 4.94, + "learning_rate": 6.720194421056828e-06, + "loss": 0.1193, + "step": 11810 + }, + { + "epoch": 4.94, + "learning_rate": 6.7185928417581615e-06, + "loss": 0.1541, + "step": 11811 + }, + { + "epoch": 4.95, + "learning_rate": 6.71699135678063e-06, + "loss": 0.1437, + "step": 11812 + }, + { + "epoch": 4.95, + "learning_rate": 6.715389966170269e-06, + "loss": 0.1602, + "step": 11813 + }, + { + "epoch": 4.95, + "learning_rate": 6.713788669973108e-06, + "loss": 0.1463, + "step": 11814 + }, + { + "epoch": 4.95, + "learning_rate": 6.7121874682351784e-06, + "loss": 0.1272, + "step": 11815 + }, + { + "epoch": 4.95, + "learning_rate": 6.710586361002499e-06, + "loss": 0.1211, + "step": 11816 + }, + { + "epoch": 4.95, + "learning_rate": 6.708985348321098e-06, + "loss": 0.1686, + "step": 11817 + }, + { + "epoch": 4.95, + "learning_rate": 6.707384430236991e-06, + "loss": 0.1725, + "step": 11818 + }, + { + "epoch": 4.95, + "learning_rate": 6.705783606796199e-06, + "loss": 0.1516, + "step": 11819 + }, + { + "epoch": 4.95, + "learning_rate": 6.704182878044732e-06, + "loss": 0.1333, + "step": 11820 + }, + { + "epoch": 4.95, + "learning_rate": 6.702582244028605e-06, + "loss": 0.1902, + "step": 11821 + }, + { + "epoch": 4.95, + "learning_rate": 6.700981704793828e-06, + "loss": 0.1359, + "step": 11822 + }, + { + "epoch": 4.95, + "learning_rate": 6.6993812603864e-06, + "loss": 0.1323, + "step": 11823 + }, + { + "epoch": 4.95, + "learning_rate": 6.6977809108523295e-06, + "loss": 0.1735, + "step": 11824 + }, + { + "epoch": 4.95, + "learning_rate": 6.696180656237616e-06, + "loss": 0.1605, + "step": 11825 + }, + { + "epoch": 4.95, + "learning_rate": 6.69458049658826e-06, + "loss": 0.1059, + "step": 11826 + }, + { + "epoch": 4.95, + "learning_rate": 6.6929804319502515e-06, + "loss": 0.1623, + "step": 11827 + }, + { + "epoch": 4.95, + "learning_rate": 6.691380462369588e-06, + "loss": 0.1609, + "step": 11828 + }, + { + "epoch": 4.95, + "learning_rate": 6.689780587892256e-06, + "loss": 0.1591, + "step": 11829 + }, + { + "epoch": 4.95, + "learning_rate": 6.688180808564244e-06, + "loss": 0.1501, + "step": 11830 + }, + { + "epoch": 4.95, + "learning_rate": 6.686581124431535e-06, + "loss": 0.1616, + "step": 11831 + }, + { + "epoch": 4.95, + "learning_rate": 6.684981535540114e-06, + "loss": 0.1992, + "step": 11832 + }, + { + "epoch": 4.95, + "learning_rate": 6.6833820419359526e-06, + "loss": 0.1278, + "step": 11833 + }, + { + "epoch": 4.95, + "learning_rate": 6.6817826436650335e-06, + "loss": 0.1619, + "step": 11834 + }, + { + "epoch": 4.95, + "learning_rate": 6.680183340773329e-06, + "loss": 0.1707, + "step": 11835 + }, + { + "epoch": 4.96, + "learning_rate": 6.678584133306808e-06, + "loss": 0.1649, + "step": 11836 + }, + { + "epoch": 4.96, + "learning_rate": 6.6769850213114375e-06, + "loss": 0.1744, + "step": 11837 + }, + { + "epoch": 4.96, + "learning_rate": 6.675386004833185e-06, + "loss": 0.1598, + "step": 11838 + }, + { + "epoch": 4.96, + "learning_rate": 6.673787083918011e-06, + "loss": 0.1677, + "step": 11839 + }, + { + "epoch": 4.96, + "learning_rate": 6.672188258611874e-06, + "loss": 0.1296, + "step": 11840 + }, + { + "epoch": 4.96, + "learning_rate": 6.670589528960734e-06, + "loss": 0.1734, + "step": 11841 + }, + { + "epoch": 4.96, + "learning_rate": 6.668990895010545e-06, + "loss": 0.1718, + "step": 11842 + }, + { + "epoch": 4.96, + "learning_rate": 6.667392356807255e-06, + "loss": 0.1761, + "step": 11843 + }, + { + "epoch": 4.96, + "learning_rate": 6.665793914396818e-06, + "loss": 0.1362, + "step": 11844 + }, + { + "epoch": 4.96, + "learning_rate": 6.664195567825174e-06, + "loss": 0.1253, + "step": 11845 + }, + { + "epoch": 4.96, + "learning_rate": 6.662597317138268e-06, + "loss": 0.1808, + "step": 11846 + }, + { + "epoch": 4.96, + "learning_rate": 6.660999162382038e-06, + "loss": 0.1541, + "step": 11847 + }, + { + "epoch": 4.96, + "learning_rate": 6.6594011036024255e-06, + "loss": 0.1515, + "step": 11848 + }, + { + "epoch": 4.96, + "learning_rate": 6.657803140845367e-06, + "loss": 0.1437, + "step": 11849 + }, + { + "epoch": 4.96, + "learning_rate": 6.656205274156788e-06, + "loss": 0.1805, + "step": 11850 + }, + { + "epoch": 4.96, + "learning_rate": 6.654607503582621e-06, + "loss": 0.1519, + "step": 11851 + }, + { + "epoch": 4.96, + "learning_rate": 6.653009829168795e-06, + "loss": 0.1387, + "step": 11852 + }, + { + "epoch": 4.96, + "learning_rate": 6.651412250961229e-06, + "loss": 0.1141, + "step": 11853 + }, + { + "epoch": 4.96, + "learning_rate": 6.649814769005847e-06, + "loss": 0.1398, + "step": 11854 + }, + { + "epoch": 4.96, + "learning_rate": 6.648217383348565e-06, + "loss": 0.1622, + "step": 11855 + }, + { + "epoch": 4.96, + "learning_rate": 6.646620094035305e-06, + "loss": 0.1627, + "step": 11856 + }, + { + "epoch": 4.96, + "learning_rate": 6.645022901111971e-06, + "loss": 0.1522, + "step": 11857 + }, + { + "epoch": 4.96, + "learning_rate": 6.643425804624476e-06, + "loss": 0.1588, + "step": 11858 + }, + { + "epoch": 4.97, + "learning_rate": 6.641828804618731e-06, + "loss": 0.1015, + "step": 11859 + }, + { + "epoch": 4.97, + "learning_rate": 6.640231901140633e-06, + "loss": 0.1449, + "step": 11860 + }, + { + "epoch": 4.97, + "learning_rate": 6.638635094236089e-06, + "loss": 0.1349, + "step": 11861 + }, + { + "epoch": 4.97, + "learning_rate": 6.637038383950998e-06, + "loss": 0.1586, + "step": 11862 + }, + { + "epoch": 4.97, + "learning_rate": 6.635441770331252e-06, + "loss": 0.1046, + "step": 11863 + }, + { + "epoch": 4.97, + "learning_rate": 6.633845253422748e-06, + "loss": 0.1767, + "step": 11864 + }, + { + "epoch": 4.97, + "learning_rate": 6.632248833271376e-06, + "loss": 0.1598, + "step": 11865 + }, + { + "epoch": 4.97, + "learning_rate": 6.6306525099230214e-06, + "loss": 0.1807, + "step": 11866 + }, + { + "epoch": 4.97, + "learning_rate": 6.629056283423574e-06, + "loss": 0.1622, + "step": 11867 + }, + { + "epoch": 4.97, + "learning_rate": 6.62746015381891e-06, + "loss": 0.2002, + "step": 11868 + }, + { + "epoch": 4.97, + "learning_rate": 6.625864121154908e-06, + "loss": 0.1498, + "step": 11869 + }, + { + "epoch": 4.97, + "learning_rate": 6.624268185477452e-06, + "loss": 0.1557, + "step": 11870 + }, + { + "epoch": 4.97, + "learning_rate": 6.622672346832409e-06, + "loss": 0.1275, + "step": 11871 + }, + { + "epoch": 4.97, + "learning_rate": 6.621076605265653e-06, + "loss": 0.1179, + "step": 11872 + }, + { + "epoch": 4.97, + "learning_rate": 6.619480960823053e-06, + "loss": 0.1549, + "step": 11873 + }, + { + "epoch": 4.97, + "learning_rate": 6.617885413550471e-06, + "loss": 0.1529, + "step": 11874 + }, + { + "epoch": 4.97, + "learning_rate": 6.616289963493771e-06, + "loss": 0.1529, + "step": 11875 + }, + { + "epoch": 4.97, + "learning_rate": 6.614694610698815e-06, + "loss": 0.1736, + "step": 11876 + }, + { + "epoch": 4.97, + "learning_rate": 6.613099355211457e-06, + "loss": 0.186, + "step": 11877 + }, + { + "epoch": 4.97, + "learning_rate": 6.611504197077554e-06, + "loss": 0.1384, + "step": 11878 + }, + { + "epoch": 4.97, + "learning_rate": 6.609909136342956e-06, + "loss": 0.1638, + "step": 11879 + }, + { + "epoch": 4.97, + "learning_rate": 6.608314173053509e-06, + "loss": 0.1224, + "step": 11880 + }, + { + "epoch": 4.97, + "learning_rate": 6.606719307255064e-06, + "loss": 0.1773, + "step": 11881 + }, + { + "epoch": 4.97, + "learning_rate": 6.605124538993462e-06, + "loss": 0.1682, + "step": 11882 + }, + { + "epoch": 4.98, + "learning_rate": 6.60352986831454e-06, + "loss": 0.1524, + "step": 11883 + }, + { + "epoch": 4.98, + "learning_rate": 6.601935295264139e-06, + "loss": 0.138, + "step": 11884 + }, + { + "epoch": 4.98, + "learning_rate": 6.600340819888091e-06, + "loss": 0.1323, + "step": 11885 + }, + { + "epoch": 4.98, + "learning_rate": 6.598746442232233e-06, + "loss": 0.1436, + "step": 11886 + }, + { + "epoch": 4.98, + "learning_rate": 6.597152162342388e-06, + "loss": 0.1648, + "step": 11887 + }, + { + "epoch": 4.98, + "learning_rate": 6.595557980264384e-06, + "loss": 0.1288, + "step": 11888 + }, + { + "epoch": 4.98, + "learning_rate": 6.593963896044047e-06, + "loss": 0.1178, + "step": 11889 + }, + { + "epoch": 4.98, + "learning_rate": 6.592369909727196e-06, + "loss": 0.1467, + "step": 11890 + }, + { + "epoch": 4.98, + "learning_rate": 6.590776021359646e-06, + "loss": 0.1334, + "step": 11891 + }, + { + "epoch": 4.98, + "learning_rate": 6.589182230987213e-06, + "loss": 0.1008, + "step": 11892 + }, + { + "epoch": 4.98, + "learning_rate": 6.587588538655714e-06, + "loss": 0.148, + "step": 11893 + }, + { + "epoch": 4.98, + "learning_rate": 6.5859949444109495e-06, + "loss": 0.1416, + "step": 11894 + }, + { + "epoch": 4.98, + "learning_rate": 6.584401448298732e-06, + "loss": 0.0951, + "step": 11895 + }, + { + "epoch": 4.98, + "learning_rate": 6.582808050364864e-06, + "loss": 0.137, + "step": 11896 + }, + { + "epoch": 4.98, + "learning_rate": 6.581214750655148e-06, + "loss": 0.1516, + "step": 11897 + }, + { + "epoch": 4.98, + "learning_rate": 6.579621549215378e-06, + "loss": 0.1675, + "step": 11898 + }, + { + "epoch": 4.98, + "learning_rate": 6.578028446091351e-06, + "loss": 0.0996, + "step": 11899 + }, + { + "epoch": 4.98, + "learning_rate": 6.5764354413288635e-06, + "loss": 0.1206, + "step": 11900 + }, + { + "epoch": 4.98, + "learning_rate": 6.574842534973698e-06, + "loss": 0.1579, + "step": 11901 + }, + { + "epoch": 4.98, + "learning_rate": 6.573249727071644e-06, + "loss": 0.1271, + "step": 11902 + }, + { + "epoch": 4.98, + "learning_rate": 6.571657017668489e-06, + "loss": 0.1465, + "step": 11903 + }, + { + "epoch": 4.98, + "learning_rate": 6.570064406810008e-06, + "loss": 0.1497, + "step": 11904 + }, + { + "epoch": 4.98, + "learning_rate": 6.568471894541982e-06, + "loss": 0.1262, + "step": 11905 + }, + { + "epoch": 4.98, + "learning_rate": 6.5668794809101884e-06, + "loss": 0.1509, + "step": 11906 + }, + { + "epoch": 4.99, + "learning_rate": 6.565287165960396e-06, + "loss": 0.1165, + "step": 11907 + }, + { + "epoch": 4.99, + "learning_rate": 6.563694949738375e-06, + "loss": 0.1142, + "step": 11908 + }, + { + "epoch": 4.99, + "learning_rate": 6.562102832289896e-06, + "loss": 0.1539, + "step": 11909 + }, + { + "epoch": 4.99, + "learning_rate": 6.560510813660719e-06, + "loss": 0.1892, + "step": 11910 + }, + { + "epoch": 4.99, + "learning_rate": 6.558918893896606e-06, + "loss": 0.1619, + "step": 11911 + }, + { + "epoch": 4.99, + "learning_rate": 6.55732707304332e-06, + "loss": 0.1679, + "step": 11912 + }, + { + "epoch": 4.99, + "learning_rate": 6.5557353511466085e-06, + "loss": 0.1434, + "step": 11913 + }, + { + "epoch": 4.99, + "learning_rate": 6.554143728252229e-06, + "loss": 0.1606, + "step": 11914 + }, + { + "epoch": 4.99, + "learning_rate": 6.552552204405928e-06, + "loss": 0.126, + "step": 11915 + }, + { + "epoch": 4.99, + "learning_rate": 6.550960779653456e-06, + "loss": 0.1764, + "step": 11916 + }, + { + "epoch": 4.99, + "learning_rate": 6.5493694540405575e-06, + "loss": 0.158, + "step": 11917 + }, + { + "epoch": 4.99, + "learning_rate": 6.547778227612969e-06, + "loss": 0.1304, + "step": 11918 + }, + { + "epoch": 4.99, + "learning_rate": 6.546187100416432e-06, + "loss": 0.1372, + "step": 11919 + }, + { + "epoch": 4.99, + "learning_rate": 6.544596072496685e-06, + "loss": 0.1006, + "step": 11920 + }, + { + "epoch": 4.99, + "learning_rate": 6.5430051438994545e-06, + "loss": 0.1562, + "step": 11921 + }, + { + "epoch": 4.99, + "learning_rate": 6.541414314670475e-06, + "loss": 0.1908, + "step": 11922 + }, + { + "epoch": 4.99, + "learning_rate": 6.539823584855472e-06, + "loss": 0.179, + "step": 11923 + }, + { + "epoch": 4.99, + "learning_rate": 6.53823295450017e-06, + "loss": 0.1202, + "step": 11924 + }, + { + "epoch": 4.99, + "learning_rate": 6.536642423650288e-06, + "loss": 0.1464, + "step": 11925 + }, + { + "epoch": 4.99, + "learning_rate": 6.5350519923515464e-06, + "loss": 0.1533, + "step": 11926 + }, + { + "epoch": 4.99, + "learning_rate": 6.533461660649663e-06, + "loss": 0.1619, + "step": 11927 + }, + { + "epoch": 4.99, + "learning_rate": 6.531871428590345e-06, + "loss": 0.1269, + "step": 11928 + }, + { + "epoch": 4.99, + "learning_rate": 6.530281296219305e-06, + "loss": 0.1226, + "step": 11929 + }, + { + "epoch": 4.99, + "learning_rate": 6.528691263582254e-06, + "loss": 0.1563, + "step": 11930 + }, + { + "epoch": 5.0, + "learning_rate": 6.52710133072489e-06, + "loss": 0.1348, + "step": 11931 + }, + { + "epoch": 5.0, + "learning_rate": 6.5255114976929154e-06, + "loss": 0.1795, + "step": 11932 + }, + { + "epoch": 5.0, + "learning_rate": 6.523921764532032e-06, + "loss": 0.1407, + "step": 11933 + }, + { + "epoch": 5.0, + "learning_rate": 6.522332131287931e-06, + "loss": 0.1487, + "step": 11934 + }, + { + "epoch": 5.0, + "learning_rate": 6.520742598006311e-06, + "loss": 0.1426, + "step": 11935 + }, + { + "epoch": 5.0, + "learning_rate": 6.519153164732852e-06, + "loss": 0.1117, + "step": 11936 + }, + { + "epoch": 5.0, + "learning_rate": 6.5175638315132484e-06, + "loss": 0.1438, + "step": 11937 + }, + { + "epoch": 5.0, + "learning_rate": 6.5159745983931846e-06, + "loss": 0.1838, + "step": 11938 + }, + { + "epoch": 5.0, + "learning_rate": 6.514385465418336e-06, + "loss": 0.1861, + "step": 11939 + }, + { + "epoch": 5.0, + "learning_rate": 6.512796432634383e-06, + "loss": 0.2126, + "step": 11940 + }, + { + "epoch": 5.0, + "learning_rate": 6.5112075000870065e-06, + "loss": 0.137, + "step": 11941 + }, + { + "epoch": 5.0, + "learning_rate": 6.5096186678218705e-06, + "loss": 0.1794, + "step": 11942 + }, + { + "epoch": 5.0, + "learning_rate": 6.50802993588465e-06, + "loss": 0.1587, + "step": 11943 + }, + { + "epoch": 5.0, + "learning_rate": 6.506441304321012e-06, + "loss": 0.1663, + "step": 11944 + }, + { + "epoch": 5.0, + "learning_rate": 6.5048527731766155e-06, + "loss": 0.1213, + "step": 11945 + }, + { + "epoch": 5.0, + "learning_rate": 6.503264342497127e-06, + "loss": 0.1663, + "step": 11946 + }, + { + "epoch": 5.0, + "learning_rate": 6.501676012328202e-06, + "loss": 0.1341, + "step": 11947 + }, + { + "epoch": 5.0, + "learning_rate": 6.500087782715494e-06, + "loss": 0.1334, + "step": 11948 + }, + { + "epoch": 5.0, + "learning_rate": 6.498499653704656e-06, + "loss": 0.096, + "step": 11949 + }, + { + "epoch": 5.0, + "learning_rate": 6.496911625341342e-06, + "loss": 0.1041, + "step": 11950 + }, + { + "epoch": 5.0, + "learning_rate": 6.495323697671192e-06, + "loss": 0.1372, + "step": 11951 + }, + { + "epoch": 5.0, + "learning_rate": 6.493735870739852e-06, + "loss": 0.1214, + "step": 11952 + }, + { + "epoch": 5.0, + "learning_rate": 6.492148144592962e-06, + "loss": 0.1367, + "step": 11953 + }, + { + "epoch": 5.0, + "learning_rate": 6.4905605192761656e-06, + "loss": 0.1148, + "step": 11954 + }, + { + "epoch": 5.01, + "learning_rate": 6.488972994835089e-06, + "loss": 0.144, + "step": 11955 + }, + { + "epoch": 5.01, + "learning_rate": 6.4873855713153675e-06, + "loss": 0.1576, + "step": 11956 + }, + { + "epoch": 5.01, + "learning_rate": 6.485798248762634e-06, + "loss": 0.1223, + "step": 11957 + }, + { + "epoch": 5.01, + "learning_rate": 6.484211027222513e-06, + "loss": 0.1417, + "step": 11958 + }, + { + "epoch": 5.01, + "learning_rate": 6.4826239067406216e-06, + "loss": 0.1227, + "step": 11959 + }, + { + "epoch": 5.01, + "learning_rate": 6.481036887362586e-06, + "loss": 0.1256, + "step": 11960 + }, + { + "epoch": 5.01, + "learning_rate": 6.479449969134024e-06, + "loss": 0.1219, + "step": 11961 + }, + { + "epoch": 5.01, + "learning_rate": 6.477863152100545e-06, + "loss": 0.128, + "step": 11962 + }, + { + "epoch": 5.01, + "learning_rate": 6.476276436307767e-06, + "loss": 0.1176, + "step": 11963 + }, + { + "epoch": 5.01, + "learning_rate": 6.474689821801295e-06, + "loss": 0.1425, + "step": 11964 + }, + { + "epoch": 5.01, + "learning_rate": 6.473103308626736e-06, + "loss": 0.0925, + "step": 11965 + }, + { + "epoch": 5.01, + "learning_rate": 6.471516896829692e-06, + "loss": 0.1364, + "step": 11966 + }, + { + "epoch": 5.01, + "learning_rate": 6.469930586455764e-06, + "loss": 0.1362, + "step": 11967 + }, + { + "epoch": 5.01, + "learning_rate": 6.468344377550549e-06, + "loss": 0.1267, + "step": 11968 + }, + { + "epoch": 5.01, + "learning_rate": 6.466758270159641e-06, + "loss": 0.156, + "step": 11969 + }, + { + "epoch": 5.01, + "learning_rate": 6.46517226432863e-06, + "loss": 0.1238, + "step": 11970 + }, + { + "epoch": 5.01, + "learning_rate": 6.4635863601031075e-06, + "loss": 0.1058, + "step": 11971 + }, + { + "epoch": 5.01, + "learning_rate": 6.462000557528656e-06, + "loss": 0.1324, + "step": 11972 + }, + { + "epoch": 5.01, + "learning_rate": 6.460414856650858e-06, + "loss": 0.0875, + "step": 11973 + }, + { + "epoch": 5.01, + "learning_rate": 6.4588292575152955e-06, + "loss": 0.1066, + "step": 11974 + }, + { + "epoch": 5.01, + "learning_rate": 6.4572437601675415e-06, + "loss": 0.1413, + "step": 11975 + }, + { + "epoch": 5.01, + "learning_rate": 6.455658364653174e-06, + "loss": 0.1057, + "step": 11976 + }, + { + "epoch": 5.01, + "learning_rate": 6.4540730710177614e-06, + "loss": 0.1148, + "step": 11977 + }, + { + "epoch": 5.01, + "learning_rate": 6.452487879306871e-06, + "loss": 0.1001, + "step": 11978 + }, + { + "epoch": 5.02, + "learning_rate": 6.450902789566068e-06, + "loss": 0.1373, + "step": 11979 + }, + { + "epoch": 5.02, + "learning_rate": 6.44931780184092e-06, + "loss": 0.107, + "step": 11980 + }, + { + "epoch": 5.02, + "learning_rate": 6.447732916176975e-06, + "loss": 0.1356, + "step": 11981 + }, + { + "epoch": 5.02, + "learning_rate": 6.446148132619798e-06, + "loss": 0.1585, + "step": 11982 + }, + { + "epoch": 5.02, + "learning_rate": 6.444563451214938e-06, + "loss": 0.1395, + "step": 11983 + }, + { + "epoch": 5.02, + "learning_rate": 6.442978872007946e-06, + "loss": 0.1106, + "step": 11984 + }, + { + "epoch": 5.02, + "learning_rate": 6.441394395044372e-06, + "loss": 0.1381, + "step": 11985 + }, + { + "epoch": 5.02, + "learning_rate": 6.439810020369755e-06, + "loss": 0.0928, + "step": 11986 + }, + { + "epoch": 5.02, + "learning_rate": 6.4382257480296405e-06, + "loss": 0.1299, + "step": 11987 + }, + { + "epoch": 5.02, + "learning_rate": 6.436641578069569e-06, + "loss": 0.1425, + "step": 11988 + }, + { + "epoch": 5.02, + "learning_rate": 6.435057510535071e-06, + "loss": 0.1285, + "step": 11989 + }, + { + "epoch": 5.02, + "learning_rate": 6.433473545471681e-06, + "loss": 0.1113, + "step": 11990 + }, + { + "epoch": 5.02, + "learning_rate": 6.431889682924932e-06, + "loss": 0.1415, + "step": 11991 + }, + { + "epoch": 5.02, + "learning_rate": 6.430305922940346e-06, + "loss": 0.1447, + "step": 11992 + }, + { + "epoch": 5.02, + "learning_rate": 6.428722265563446e-06, + "loss": 0.1038, + "step": 11993 + }, + { + "epoch": 5.02, + "learning_rate": 6.42713871083976e-06, + "loss": 0.1033, + "step": 11994 + }, + { + "epoch": 5.02, + "learning_rate": 6.4255552588147975e-06, + "loss": 0.1294, + "step": 11995 + }, + { + "epoch": 5.02, + "learning_rate": 6.423971909534077e-06, + "loss": 0.0996, + "step": 11996 + }, + { + "epoch": 5.02, + "learning_rate": 6.422388663043112e-06, + "loss": 0.1159, + "step": 11997 + }, + { + "epoch": 5.02, + "learning_rate": 6.420805519387412e-06, + "loss": 0.1138, + "step": 11998 + }, + { + "epoch": 5.02, + "learning_rate": 6.419222478612479e-06, + "loss": 0.0632, + "step": 11999 + }, + { + "epoch": 5.02, + "learning_rate": 6.417639540763818e-06, + "loss": 0.1196, + "step": 12000 + }, + { + "epoch": 5.02, + "learning_rate": 6.416056705886932e-06, + "loss": 0.1107, + "step": 12001 + }, + { + "epoch": 5.02, + "learning_rate": 6.414473974027313e-06, + "loss": 0.1428, + "step": 12002 + }, + { + "epoch": 5.03, + "learning_rate": 6.412891345230462e-06, + "loss": 0.1136, + "step": 12003 + }, + { + "epoch": 5.03, + "learning_rate": 6.411308819541863e-06, + "loss": 0.1047, + "step": 12004 + }, + { + "epoch": 5.03, + "learning_rate": 6.409726397007008e-06, + "loss": 0.1103, + "step": 12005 + }, + { + "epoch": 5.03, + "learning_rate": 6.408144077671381e-06, + "loss": 0.1787, + "step": 12006 + }, + { + "epoch": 5.03, + "learning_rate": 6.406561861580465e-06, + "loss": 0.0751, + "step": 12007 + }, + { + "epoch": 5.03, + "learning_rate": 6.404979748779738e-06, + "loss": 0.1224, + "step": 12008 + }, + { + "epoch": 5.03, + "learning_rate": 6.403397739314682e-06, + "loss": 0.1314, + "step": 12009 + }, + { + "epoch": 5.03, + "learning_rate": 6.4018158332307635e-06, + "loss": 0.1615, + "step": 12010 + }, + { + "epoch": 5.03, + "learning_rate": 6.400234030573456e-06, + "loss": 0.1406, + "step": 12011 + }, + { + "epoch": 5.03, + "learning_rate": 6.3986523313882285e-06, + "loss": 0.0791, + "step": 12012 + }, + { + "epoch": 5.03, + "learning_rate": 6.397070735720543e-06, + "loss": 0.1291, + "step": 12013 + }, + { + "epoch": 5.03, + "learning_rate": 6.395489243615861e-06, + "loss": 0.1109, + "step": 12014 + }, + { + "epoch": 5.03, + "learning_rate": 6.393907855119645e-06, + "loss": 0.1138, + "step": 12015 + }, + { + "epoch": 5.03, + "learning_rate": 6.392326570277345e-06, + "loss": 0.136, + "step": 12016 + }, + { + "epoch": 5.03, + "learning_rate": 6.390745389134416e-06, + "loss": 0.1366, + "step": 12017 + }, + { + "epoch": 5.03, + "learning_rate": 6.389164311736311e-06, + "loss": 0.1381, + "step": 12018 + }, + { + "epoch": 5.03, + "learning_rate": 6.387583338128471e-06, + "loss": 0.1289, + "step": 12019 + }, + { + "epoch": 5.03, + "learning_rate": 6.386002468356343e-06, + "loss": 0.1132, + "step": 12020 + }, + { + "epoch": 5.03, + "learning_rate": 6.384421702465369e-06, + "loss": 0.1303, + "step": 12021 + }, + { + "epoch": 5.03, + "learning_rate": 6.382841040500982e-06, + "loss": 0.1251, + "step": 12022 + }, + { + "epoch": 5.03, + "learning_rate": 6.38126048250862e-06, + "loss": 0.1139, + "step": 12023 + }, + { + "epoch": 5.03, + "learning_rate": 6.379680028533714e-06, + "loss": 0.0976, + "step": 12024 + }, + { + "epoch": 5.03, + "learning_rate": 6.378099678621697e-06, + "loss": 0.1316, + "step": 12025 + }, + { + "epoch": 5.03, + "learning_rate": 6.376519432817989e-06, + "loss": 0.1245, + "step": 12026 + }, + { + "epoch": 5.04, + "learning_rate": 6.374939291168012e-06, + "loss": 0.1231, + "step": 12027 + }, + { + "epoch": 5.04, + "learning_rate": 6.373359253717188e-06, + "loss": 0.1713, + "step": 12028 + }, + { + "epoch": 5.04, + "learning_rate": 6.3717793205109354e-06, + "loss": 0.1427, + "step": 12029 + }, + { + "epoch": 5.04, + "learning_rate": 6.370199491594667e-06, + "loss": 0.1467, + "step": 12030 + }, + { + "epoch": 5.04, + "learning_rate": 6.368619767013792e-06, + "loss": 0.1034, + "step": 12031 + }, + { + "epoch": 5.04, + "learning_rate": 6.367040146813721e-06, + "loss": 0.1073, + "step": 12032 + }, + { + "epoch": 5.04, + "learning_rate": 6.365460631039855e-06, + "loss": 0.1462, + "step": 12033 + }, + { + "epoch": 5.04, + "learning_rate": 6.363881219737599e-06, + "loss": 0.1324, + "step": 12034 + }, + { + "epoch": 5.04, + "learning_rate": 6.362301912952351e-06, + "loss": 0.1383, + "step": 12035 + }, + { + "epoch": 5.04, + "learning_rate": 6.360722710729506e-06, + "loss": 0.1589, + "step": 12036 + }, + { + "epoch": 5.04, + "learning_rate": 6.3591436131144565e-06, + "loss": 0.1377, + "step": 12037 + }, + { + "epoch": 5.04, + "learning_rate": 6.357564620152595e-06, + "loss": 0.1345, + "step": 12038 + }, + { + "epoch": 5.04, + "learning_rate": 6.355985731889305e-06, + "loss": 0.1272, + "step": 12039 + }, + { + "epoch": 5.04, + "learning_rate": 6.354406948369971e-06, + "loss": 0.1493, + "step": 12040 + }, + { + "epoch": 5.04, + "learning_rate": 6.352828269639974e-06, + "loss": 0.1311, + "step": 12041 + }, + { + "epoch": 5.04, + "learning_rate": 6.351249695744694e-06, + "loss": 0.1302, + "step": 12042 + }, + { + "epoch": 5.04, + "learning_rate": 6.349671226729502e-06, + "loss": 0.1074, + "step": 12043 + }, + { + "epoch": 5.04, + "learning_rate": 6.348092862639772e-06, + "loss": 0.1288, + "step": 12044 + }, + { + "epoch": 5.04, + "learning_rate": 6.346514603520873e-06, + "loss": 0.1185, + "step": 12045 + }, + { + "epoch": 5.04, + "learning_rate": 6.344936449418168e-06, + "loss": 0.1462, + "step": 12046 + }, + { + "epoch": 5.04, + "learning_rate": 6.343358400377021e-06, + "loss": 0.1116, + "step": 12047 + }, + { + "epoch": 5.04, + "learning_rate": 6.341780456442797e-06, + "loss": 0.1073, + "step": 12048 + }, + { + "epoch": 5.04, + "learning_rate": 6.340202617660842e-06, + "loss": 0.1362, + "step": 12049 + }, + { + "epoch": 5.05, + "learning_rate": 6.338624884076517e-06, + "loss": 0.0908, + "step": 12050 + }, + { + "epoch": 5.05, + "learning_rate": 6.337047255735169e-06, + "loss": 0.0934, + "step": 12051 + }, + { + "epoch": 5.05, + "learning_rate": 6.335469732682149e-06, + "loss": 0.1395, + "step": 12052 + }, + { + "epoch": 5.05, + "learning_rate": 6.3338923149628e-06, + "loss": 0.1196, + "step": 12053 + }, + { + "epoch": 5.05, + "learning_rate": 6.332315002622461e-06, + "loss": 0.1087, + "step": 12054 + }, + { + "epoch": 5.05, + "learning_rate": 6.330737795706474e-06, + "loss": 0.1016, + "step": 12055 + }, + { + "epoch": 5.05, + "learning_rate": 6.329160694260173e-06, + "loss": 0.1073, + "step": 12056 + }, + { + "epoch": 5.05, + "learning_rate": 6.3275836983288895e-06, + "loss": 0.1342, + "step": 12057 + }, + { + "epoch": 5.05, + "learning_rate": 6.326006807957954e-06, + "loss": 0.1287, + "step": 12058 + }, + { + "epoch": 5.05, + "learning_rate": 6.324430023192695e-06, + "loss": 0.1116, + "step": 12059 + }, + { + "epoch": 5.05, + "learning_rate": 6.32285334407843e-06, + "loss": 0.11, + "step": 12060 + }, + { + "epoch": 5.05, + "learning_rate": 6.3212767706604826e-06, + "loss": 0.1296, + "step": 12061 + }, + { + "epoch": 5.05, + "learning_rate": 6.319700302984173e-06, + "loss": 0.1233, + "step": 12062 + }, + { + "epoch": 5.05, + "learning_rate": 6.31812394109481e-06, + "loss": 0.0853, + "step": 12063 + }, + { + "epoch": 5.05, + "learning_rate": 6.316547685037707e-06, + "loss": 0.0856, + "step": 12064 + }, + { + "epoch": 5.05, + "learning_rate": 6.314971534858174e-06, + "loss": 0.1416, + "step": 12065 + }, + { + "epoch": 5.05, + "learning_rate": 6.313395490601513e-06, + "loss": 0.172, + "step": 12066 + }, + { + "epoch": 5.05, + "learning_rate": 6.311819552313026e-06, + "loss": 0.1376, + "step": 12067 + }, + { + "epoch": 5.05, + "learning_rate": 6.310243720038014e-06, + "loss": 0.1347, + "step": 12068 + }, + { + "epoch": 5.05, + "learning_rate": 6.308667993821773e-06, + "loss": 0.1733, + "step": 12069 + }, + { + "epoch": 5.05, + "learning_rate": 6.3070923737095946e-06, + "loss": 0.155, + "step": 12070 + }, + { + "epoch": 5.05, + "learning_rate": 6.305516859746772e-06, + "loss": 0.0885, + "step": 12071 + }, + { + "epoch": 5.05, + "learning_rate": 6.303941451978584e-06, + "loss": 0.1177, + "step": 12072 + }, + { + "epoch": 5.05, + "learning_rate": 6.302366150450323e-06, + "loss": 0.1401, + "step": 12073 + }, + { + "epoch": 5.06, + "learning_rate": 6.300790955207262e-06, + "loss": 0.1523, + "step": 12074 + }, + { + "epoch": 5.06, + "learning_rate": 6.299215866294683e-06, + "loss": 0.128, + "step": 12075 + }, + { + "epoch": 5.06, + "learning_rate": 6.297640883757863e-06, + "loss": 0.1039, + "step": 12076 + }, + { + "epoch": 5.06, + "learning_rate": 6.296066007642067e-06, + "loss": 0.1119, + "step": 12077 + }, + { + "epoch": 5.06, + "learning_rate": 6.294491237992566e-06, + "loss": 0.1343, + "step": 12078 + }, + { + "epoch": 5.06, + "learning_rate": 6.292916574854628e-06, + "loss": 0.1404, + "step": 12079 + }, + { + "epoch": 5.06, + "learning_rate": 6.291342018273514e-06, + "loss": 0.1355, + "step": 12080 + }, + { + "epoch": 5.06, + "learning_rate": 6.2897675682944824e-06, + "loss": 0.0829, + "step": 12081 + }, + { + "epoch": 5.06, + "learning_rate": 6.288193224962789e-06, + "loss": 0.0669, + "step": 12082 + }, + { + "epoch": 5.06, + "learning_rate": 6.2866189883236885e-06, + "loss": 0.1222, + "step": 12083 + }, + { + "epoch": 5.06, + "learning_rate": 6.285044858422429e-06, + "loss": 0.1228, + "step": 12084 + }, + { + "epoch": 5.06, + "learning_rate": 6.283470835304257e-06, + "loss": 0.116, + "step": 12085 + }, + { + "epoch": 5.06, + "learning_rate": 6.281896919014422e-06, + "loss": 0.12, + "step": 12086 + }, + { + "epoch": 5.06, + "learning_rate": 6.280323109598157e-06, + "loss": 0.138, + "step": 12087 + }, + { + "epoch": 5.06, + "learning_rate": 6.278749407100704e-06, + "loss": 0.1307, + "step": 12088 + }, + { + "epoch": 5.06, + "learning_rate": 6.2771758115673e-06, + "loss": 0.0881, + "step": 12089 + }, + { + "epoch": 5.06, + "learning_rate": 6.2756023230431705e-06, + "loss": 0.1232, + "step": 12090 + }, + { + "epoch": 5.06, + "learning_rate": 6.2740289415735474e-06, + "loss": 0.1266, + "step": 12091 + }, + { + "epoch": 5.06, + "learning_rate": 6.272455667203659e-06, + "loss": 0.1358, + "step": 12092 + }, + { + "epoch": 5.06, + "learning_rate": 6.270882499978723e-06, + "loss": 0.1208, + "step": 12093 + }, + { + "epoch": 5.06, + "learning_rate": 6.269309439943963e-06, + "loss": 0.1235, + "step": 12094 + }, + { + "epoch": 5.06, + "learning_rate": 6.2677364871445886e-06, + "loss": 0.1502, + "step": 12095 + }, + { + "epoch": 5.06, + "learning_rate": 6.2661636416258175e-06, + "loss": 0.1322, + "step": 12096 + }, + { + "epoch": 5.06, + "learning_rate": 6.264590903432861e-06, + "loss": 0.0914, + "step": 12097 + }, + { + "epoch": 5.07, + "learning_rate": 6.263018272610921e-06, + "loss": 0.123, + "step": 12098 + }, + { + "epoch": 5.07, + "learning_rate": 6.261445749205206e-06, + "loss": 0.1693, + "step": 12099 + }, + { + "epoch": 5.07, + "learning_rate": 6.259873333260917e-06, + "loss": 0.1072, + "step": 12100 + }, + { + "epoch": 5.07, + "learning_rate": 6.258301024823249e-06, + "loss": 0.0645, + "step": 12101 + }, + { + "epoch": 5.07, + "learning_rate": 6.256728823937397e-06, + "loss": 0.1359, + "step": 12102 + }, + { + "epoch": 5.07, + "learning_rate": 6.255156730648555e-06, + "loss": 0.1382, + "step": 12103 + }, + { + "epoch": 5.07, + "learning_rate": 6.253584745001908e-06, + "loss": 0.1039, + "step": 12104 + }, + { + "epoch": 5.07, + "learning_rate": 6.252012867042643e-06, + "loss": 0.1308, + "step": 12105 + }, + { + "epoch": 5.07, + "learning_rate": 6.250441096815945e-06, + "loss": 0.1262, + "step": 12106 + }, + { + "epoch": 5.07, + "learning_rate": 6.248869434366987e-06, + "loss": 0.1414, + "step": 12107 + }, + { + "epoch": 5.07, + "learning_rate": 6.247297879740948e-06, + "loss": 0.1307, + "step": 12108 + }, + { + "epoch": 5.07, + "learning_rate": 6.245726432983003e-06, + "loss": 0.1036, + "step": 12109 + }, + { + "epoch": 5.07, + "learning_rate": 6.244155094138322e-06, + "loss": 0.152, + "step": 12110 + }, + { + "epoch": 5.07, + "learning_rate": 6.2425838632520674e-06, + "loss": 0.1469, + "step": 12111 + }, + { + "epoch": 5.07, + "learning_rate": 6.2410127403694056e-06, + "loss": 0.1288, + "step": 12112 + }, + { + "epoch": 5.07, + "learning_rate": 6.239441725535498e-06, + "loss": 0.0963, + "step": 12113 + }, + { + "epoch": 5.07, + "learning_rate": 6.237870818795499e-06, + "loss": 0.0882, + "step": 12114 + }, + { + "epoch": 5.07, + "learning_rate": 6.2363000201945656e-06, + "loss": 0.1074, + "step": 12115 + }, + { + "epoch": 5.07, + "learning_rate": 6.234729329777853e-06, + "loss": 0.1028, + "step": 12116 + }, + { + "epoch": 5.07, + "learning_rate": 6.233158747590502e-06, + "loss": 0.1171, + "step": 12117 + }, + { + "epoch": 5.07, + "learning_rate": 6.231588273677658e-06, + "loss": 0.1217, + "step": 12118 + }, + { + "epoch": 5.07, + "learning_rate": 6.230017908084465e-06, + "loss": 0.1087, + "step": 12119 + }, + { + "epoch": 5.07, + "learning_rate": 6.228447650856062e-06, + "loss": 0.1036, + "step": 12120 + }, + { + "epoch": 5.07, + "learning_rate": 6.226877502037585e-06, + "loss": 0.0812, + "step": 12121 + }, + { + "epoch": 5.08, + "learning_rate": 6.225307461674166e-06, + "loss": 0.1108, + "step": 12122 + }, + { + "epoch": 5.08, + "learning_rate": 6.2237375298109336e-06, + "loss": 0.1215, + "step": 12123 + }, + { + "epoch": 5.08, + "learning_rate": 6.222167706493016e-06, + "loss": 0.1526, + "step": 12124 + }, + { + "epoch": 5.08, + "learning_rate": 6.220597991765536e-06, + "loss": 0.1058, + "step": 12125 + }, + { + "epoch": 5.08, + "learning_rate": 6.2190283856736115e-06, + "loss": 0.1068, + "step": 12126 + }, + { + "epoch": 5.08, + "learning_rate": 6.217458888262362e-06, + "loss": 0.1346, + "step": 12127 + }, + { + "epoch": 5.08, + "learning_rate": 6.215889499576898e-06, + "loss": 0.1367, + "step": 12128 + }, + { + "epoch": 5.08, + "learning_rate": 6.214320219662333e-06, + "loss": 0.1279, + "step": 12129 + }, + { + "epoch": 5.08, + "learning_rate": 6.212751048563776e-06, + "loss": 0.1141, + "step": 12130 + }, + { + "epoch": 5.08, + "learning_rate": 6.2111819863263275e-06, + "loss": 0.1321, + "step": 12131 + }, + { + "epoch": 5.08, + "learning_rate": 6.209613032995091e-06, + "loss": 0.1796, + "step": 12132 + }, + { + "epoch": 5.08, + "learning_rate": 6.208044188615166e-06, + "loss": 0.1399, + "step": 12133 + }, + { + "epoch": 5.08, + "learning_rate": 6.206475453231644e-06, + "loss": 0.1253, + "step": 12134 + }, + { + "epoch": 5.08, + "learning_rate": 6.204906826889618e-06, + "loss": 0.1111, + "step": 12135 + }, + { + "epoch": 5.08, + "learning_rate": 6.203338309634179e-06, + "loss": 0.1165, + "step": 12136 + }, + { + "epoch": 5.08, + "learning_rate": 6.201769901510412e-06, + "loss": 0.1071, + "step": 12137 + }, + { + "epoch": 5.08, + "learning_rate": 6.200201602563397e-06, + "loss": 0.1172, + "step": 12138 + }, + { + "epoch": 5.08, + "learning_rate": 6.1986334128382195e-06, + "loss": 0.1291, + "step": 12139 + }, + { + "epoch": 5.08, + "learning_rate": 6.197065332379947e-06, + "loss": 0.117, + "step": 12140 + }, + { + "epoch": 5.08, + "learning_rate": 6.195497361233658e-06, + "loss": 0.1428, + "step": 12141 + }, + { + "epoch": 5.08, + "learning_rate": 6.19392949944442e-06, + "loss": 0.1521, + "step": 12142 + }, + { + "epoch": 5.08, + "learning_rate": 6.192361747057301e-06, + "loss": 0.128, + "step": 12143 + }, + { + "epoch": 5.08, + "learning_rate": 6.190794104117368e-06, + "loss": 0.1004, + "step": 12144 + }, + { + "epoch": 5.08, + "learning_rate": 6.189226570669675e-06, + "loss": 0.1502, + "step": 12145 + }, + { + "epoch": 5.09, + "learning_rate": 6.187659146759282e-06, + "loss": 0.1702, + "step": 12146 + }, + { + "epoch": 5.09, + "learning_rate": 6.186091832431247e-06, + "loss": 0.1418, + "step": 12147 + }, + { + "epoch": 5.09, + "learning_rate": 6.1845246277306145e-06, + "loss": 0.1337, + "step": 12148 + }, + { + "epoch": 5.09, + "learning_rate": 6.182957532702437e-06, + "loss": 0.1371, + "step": 12149 + }, + { + "epoch": 5.09, + "learning_rate": 6.181390547391758e-06, + "loss": 0.1352, + "step": 12150 + }, + { + "epoch": 5.09, + "learning_rate": 6.179823671843621e-06, + "loss": 0.138, + "step": 12151 + }, + { + "epoch": 5.09, + "learning_rate": 6.178256906103061e-06, + "loss": 0.1226, + "step": 12152 + }, + { + "epoch": 5.09, + "learning_rate": 6.176690250215114e-06, + "loss": 0.1264, + "step": 12153 + }, + { + "epoch": 5.09, + "learning_rate": 6.175123704224814e-06, + "loss": 0.1309, + "step": 12154 + }, + { + "epoch": 5.09, + "learning_rate": 6.173557268177189e-06, + "loss": 0.135, + "step": 12155 + }, + { + "epoch": 5.09, + "learning_rate": 6.171990942117265e-06, + "loss": 0.1154, + "step": 12156 + }, + { + "epoch": 5.09, + "learning_rate": 6.170424726090065e-06, + "loss": 0.1521, + "step": 12157 + }, + { + "epoch": 5.09, + "learning_rate": 6.168858620140606e-06, + "loss": 0.1361, + "step": 12158 + }, + { + "epoch": 5.09, + "learning_rate": 6.167292624313908e-06, + "loss": 0.1125, + "step": 12159 + }, + { + "epoch": 5.09, + "learning_rate": 6.165726738654982e-06, + "loss": 0.0856, + "step": 12160 + }, + { + "epoch": 5.09, + "learning_rate": 6.16416096320884e-06, + "loss": 0.1175, + "step": 12161 + }, + { + "epoch": 5.09, + "learning_rate": 6.162595298020487e-06, + "loss": 0.0999, + "step": 12162 + }, + { + "epoch": 5.09, + "learning_rate": 6.1610297431349255e-06, + "loss": 0.1059, + "step": 12163 + }, + { + "epoch": 5.09, + "learning_rate": 6.1594642985971565e-06, + "loss": 0.1038, + "step": 12164 + }, + { + "epoch": 5.09, + "learning_rate": 6.15789896445218e-06, + "loss": 0.1068, + "step": 12165 + }, + { + "epoch": 5.09, + "learning_rate": 6.1563337407449865e-06, + "loss": 0.1114, + "step": 12166 + }, + { + "epoch": 5.09, + "learning_rate": 6.15476862752057e-06, + "loss": 0.1109, + "step": 12167 + }, + { + "epoch": 5.09, + "learning_rate": 6.153203624823918e-06, + "loss": 0.1219, + "step": 12168 + }, + { + "epoch": 5.09, + "learning_rate": 6.151638732700012e-06, + "loss": 0.1144, + "step": 12169 + }, + { + "epoch": 5.1, + "learning_rate": 6.150073951193836e-06, + "loss": 0.1246, + "step": 12170 + }, + { + "epoch": 5.1, + "learning_rate": 6.1485092803503695e-06, + "loss": 0.1424, + "step": 12171 + }, + { + "epoch": 5.1, + "learning_rate": 6.146944720214585e-06, + "loss": 0.1515, + "step": 12172 + }, + { + "epoch": 5.1, + "learning_rate": 6.145380270831455e-06, + "loss": 0.1316, + "step": 12173 + }, + { + "epoch": 5.1, + "learning_rate": 6.143815932245951e-06, + "loss": 0.1069, + "step": 12174 + }, + { + "epoch": 5.1, + "learning_rate": 6.142251704503033e-06, + "loss": 0.1006, + "step": 12175 + }, + { + "epoch": 5.1, + "learning_rate": 6.140687587647667e-06, + "loss": 0.0906, + "step": 12176 + }, + { + "epoch": 5.1, + "learning_rate": 6.139123581724813e-06, + "loss": 0.0751, + "step": 12177 + }, + { + "epoch": 5.1, + "learning_rate": 6.1375596867794245e-06, + "loss": 0.1319, + "step": 12178 + }, + { + "epoch": 5.1, + "learning_rate": 6.135995902856455e-06, + "loss": 0.0637, + "step": 12179 + }, + { + "epoch": 5.1, + "learning_rate": 6.134432230000854e-06, + "loss": 0.1416, + "step": 12180 + }, + { + "epoch": 5.1, + "learning_rate": 6.132868668257571e-06, + "loss": 0.1236, + "step": 12181 + }, + { + "epoch": 5.1, + "learning_rate": 6.131305217671544e-06, + "loss": 0.0791, + "step": 12182 + }, + { + "epoch": 5.1, + "learning_rate": 6.129741878287716e-06, + "loss": 0.106, + "step": 12183 + }, + { + "epoch": 5.1, + "learning_rate": 6.128178650151026e-06, + "loss": 0.1149, + "step": 12184 + }, + { + "epoch": 5.1, + "learning_rate": 6.126615533306405e-06, + "loss": 0.131, + "step": 12185 + }, + { + "epoch": 5.1, + "learning_rate": 6.12505252779878e-06, + "loss": 0.1268, + "step": 12186 + }, + { + "epoch": 5.1, + "learning_rate": 6.123489633673082e-06, + "loss": 0.0994, + "step": 12187 + }, + { + "epoch": 5.1, + "learning_rate": 6.121926850974236e-06, + "loss": 0.1267, + "step": 12188 + }, + { + "epoch": 5.1, + "learning_rate": 6.120364179747161e-06, + "loss": 0.1149, + "step": 12189 + }, + { + "epoch": 5.1, + "learning_rate": 6.1188016200367715e-06, + "loss": 0.1255, + "step": 12190 + }, + { + "epoch": 5.1, + "learning_rate": 6.117239171887987e-06, + "loss": 0.1498, + "step": 12191 + }, + { + "epoch": 5.1, + "learning_rate": 6.11567683534572e-06, + "loss": 0.106, + "step": 12192 + }, + { + "epoch": 5.1, + "learning_rate": 6.1141146104548735e-06, + "loss": 0.1239, + "step": 12193 + }, + { + "epoch": 5.11, + "learning_rate": 6.112552497260353e-06, + "loss": 0.0943, + "step": 12194 + }, + { + "epoch": 5.11, + "learning_rate": 6.110990495807064e-06, + "loss": 0.1374, + "step": 12195 + }, + { + "epoch": 5.11, + "learning_rate": 6.1094286061399e-06, + "loss": 0.0907, + "step": 12196 + }, + { + "epoch": 5.11, + "learning_rate": 6.107866828303759e-06, + "loss": 0.1415, + "step": 12197 + }, + { + "epoch": 5.11, + "learning_rate": 6.106305162343533e-06, + "loss": 0.0777, + "step": 12198 + }, + { + "epoch": 5.11, + "learning_rate": 6.104743608304109e-06, + "loss": 0.1269, + "step": 12199 + }, + { + "epoch": 5.11, + "learning_rate": 6.103182166230373e-06, + "loss": 0.1295, + "step": 12200 + }, + { + "epoch": 5.11, + "learning_rate": 6.101620836167208e-06, + "loss": 0.1536, + "step": 12201 + }, + { + "epoch": 5.11, + "learning_rate": 6.100059618159493e-06, + "loss": 0.1431, + "step": 12202 + }, + { + "epoch": 5.11, + "learning_rate": 6.098498512252102e-06, + "loss": 0.0914, + "step": 12203 + }, + { + "epoch": 5.11, + "learning_rate": 6.096937518489912e-06, + "loss": 0.1229, + "step": 12204 + }, + { + "epoch": 5.11, + "learning_rate": 6.095376636917787e-06, + "loss": 0.124, + "step": 12205 + }, + { + "epoch": 5.11, + "learning_rate": 6.093815867580597e-06, + "loss": 0.1116, + "step": 12206 + }, + { + "epoch": 5.11, + "learning_rate": 6.0922552105232055e-06, + "loss": 0.1509, + "step": 12207 + }, + { + "epoch": 5.11, + "learning_rate": 6.0906946657904685e-06, + "loss": 0.1288, + "step": 12208 + }, + { + "epoch": 5.11, + "learning_rate": 6.089134233427245e-06, + "loss": 0.1412, + "step": 12209 + }, + { + "epoch": 5.11, + "learning_rate": 6.087573913478386e-06, + "loss": 0.1191, + "step": 12210 + }, + { + "epoch": 5.11, + "learning_rate": 6.086013705988742e-06, + "loss": 0.1093, + "step": 12211 + }, + { + "epoch": 5.11, + "learning_rate": 6.084453611003164e-06, + "loss": 0.0971, + "step": 12212 + }, + { + "epoch": 5.11, + "learning_rate": 6.08289362856649e-06, + "loss": 0.0934, + "step": 12213 + }, + { + "epoch": 5.11, + "learning_rate": 6.081333758723562e-06, + "loss": 0.1281, + "step": 12214 + }, + { + "epoch": 5.11, + "learning_rate": 6.079774001519221e-06, + "loss": 0.1253, + "step": 12215 + }, + { + "epoch": 5.11, + "learning_rate": 6.078214356998294e-06, + "loss": 0.1323, + "step": 12216 + }, + { + "epoch": 5.11, + "learning_rate": 6.076654825205617e-06, + "loss": 0.154, + "step": 12217 + }, + { + "epoch": 5.12, + "learning_rate": 6.075095406186015e-06, + "loss": 0.0858, + "step": 12218 + }, + { + "epoch": 5.12, + "learning_rate": 6.073536099984314e-06, + "loss": 0.1258, + "step": 12219 + }, + { + "epoch": 5.12, + "learning_rate": 6.071976906645332e-06, + "loss": 0.1146, + "step": 12220 + }, + { + "epoch": 5.12, + "learning_rate": 6.070417826213888e-06, + "loss": 0.1333, + "step": 12221 + }, + { + "epoch": 5.12, + "learning_rate": 6.068858858734799e-06, + "loss": 0.096, + "step": 12222 + }, + { + "epoch": 5.12, + "learning_rate": 6.06730000425287e-06, + "loss": 0.1071, + "step": 12223 + }, + { + "epoch": 5.12, + "learning_rate": 6.065741262812914e-06, + "loss": 0.1269, + "step": 12224 + }, + { + "epoch": 5.12, + "learning_rate": 6.064182634459735e-06, + "loss": 0.1314, + "step": 12225 + }, + { + "epoch": 5.12, + "learning_rate": 6.062624119238133e-06, + "loss": 0.1292, + "step": 12226 + }, + { + "epoch": 5.12, + "learning_rate": 6.061065717192907e-06, + "loss": 0.087, + "step": 12227 + }, + { + "epoch": 5.12, + "learning_rate": 6.059507428368852e-06, + "loss": 0.1242, + "step": 12228 + }, + { + "epoch": 5.12, + "learning_rate": 6.057949252810762e-06, + "loss": 0.1421, + "step": 12229 + }, + { + "epoch": 5.12, + "learning_rate": 6.056391190563419e-06, + "loss": 0.1274, + "step": 12230 + }, + { + "epoch": 5.12, + "learning_rate": 6.054833241671611e-06, + "loss": 0.1439, + "step": 12231 + }, + { + "epoch": 5.12, + "learning_rate": 6.053275406180122e-06, + "loss": 0.1193, + "step": 12232 + }, + { + "epoch": 5.12, + "learning_rate": 6.051717684133729e-06, + "loss": 0.1221, + "step": 12233 + }, + { + "epoch": 5.12, + "learning_rate": 6.0501600755772074e-06, + "loss": 0.1642, + "step": 12234 + }, + { + "epoch": 5.12, + "learning_rate": 6.048602580555328e-06, + "loss": 0.1479, + "step": 12235 + }, + { + "epoch": 5.12, + "learning_rate": 6.047045199112865e-06, + "loss": 0.1157, + "step": 12236 + }, + { + "epoch": 5.12, + "learning_rate": 6.0454879312945755e-06, + "loss": 0.1253, + "step": 12237 + }, + { + "epoch": 5.12, + "learning_rate": 6.043930777145228e-06, + "loss": 0.1043, + "step": 12238 + }, + { + "epoch": 5.12, + "learning_rate": 6.042373736709581e-06, + "loss": 0.0997, + "step": 12239 + }, + { + "epoch": 5.12, + "learning_rate": 6.040816810032387e-06, + "loss": 0.1033, + "step": 12240 + }, + { + "epoch": 5.12, + "learning_rate": 6.039259997158401e-06, + "loss": 0.1221, + "step": 12241 + }, + { + "epoch": 5.13, + "learning_rate": 6.037703298132373e-06, + "loss": 0.1313, + "step": 12242 + }, + { + "epoch": 5.13, + "learning_rate": 6.036146712999047e-06, + "loss": 0.1127, + "step": 12243 + }, + { + "epoch": 5.13, + "learning_rate": 6.0345902418031645e-06, + "loss": 0.0897, + "step": 12244 + }, + { + "epoch": 5.13, + "learning_rate": 6.03303388458947e-06, + "loss": 0.1243, + "step": 12245 + }, + { + "epoch": 5.13, + "learning_rate": 6.031477641402693e-06, + "loss": 0.1178, + "step": 12246 + }, + { + "epoch": 5.13, + "learning_rate": 6.029921512287571e-06, + "loss": 0.0931, + "step": 12247 + }, + { + "epoch": 5.13, + "learning_rate": 6.028365497288831e-06, + "loss": 0.1232, + "step": 12248 + }, + { + "epoch": 5.13, + "learning_rate": 6.026809596451203e-06, + "loss": 0.099, + "step": 12249 + }, + { + "epoch": 5.13, + "learning_rate": 6.025253809819405e-06, + "loss": 0.0836, + "step": 12250 + }, + { + "epoch": 5.13, + "learning_rate": 6.023698137438159e-06, + "loss": 0.1112, + "step": 12251 + }, + { + "epoch": 5.13, + "learning_rate": 6.022142579352186e-06, + "loss": 0.0906, + "step": 12252 + }, + { + "epoch": 5.13, + "learning_rate": 6.020587135606193e-06, + "loss": 0.1082, + "step": 12253 + }, + { + "epoch": 5.13, + "learning_rate": 6.019031806244888e-06, + "loss": 0.0938, + "step": 12254 + }, + { + "epoch": 5.13, + "learning_rate": 6.017476591312983e-06, + "loss": 0.1267, + "step": 12255 + }, + { + "epoch": 5.13, + "learning_rate": 6.015921490855181e-06, + "loss": 0.1103, + "step": 12256 + }, + { + "epoch": 5.13, + "learning_rate": 6.014366504916178e-06, + "loss": 0.1016, + "step": 12257 + }, + { + "epoch": 5.13, + "learning_rate": 6.012811633540672e-06, + "loss": 0.1498, + "step": 12258 + }, + { + "epoch": 5.13, + "learning_rate": 6.011256876773362e-06, + "loss": 0.1537, + "step": 12259 + }, + { + "epoch": 5.13, + "learning_rate": 6.009702234658929e-06, + "loss": 0.1215, + "step": 12260 + }, + { + "epoch": 5.13, + "learning_rate": 6.008147707242065e-06, + "loss": 0.1021, + "step": 12261 + }, + { + "epoch": 5.13, + "learning_rate": 6.0065932945674535e-06, + "loss": 0.1175, + "step": 12262 + }, + { + "epoch": 5.13, + "learning_rate": 6.005038996679775e-06, + "loss": 0.1038, + "step": 12263 + }, + { + "epoch": 5.13, + "learning_rate": 6.0034848136237035e-06, + "loss": 0.118, + "step": 12264 + }, + { + "epoch": 5.14, + "learning_rate": 6.001930745443916e-06, + "loss": 0.1179, + "step": 12265 + }, + { + "epoch": 5.14, + "learning_rate": 6.000376792185082e-06, + "loss": 0.1159, + "step": 12266 + }, + { + "epoch": 5.14, + "learning_rate": 5.998822953891866e-06, + "loss": 0.1377, + "step": 12267 + }, + { + "epoch": 5.14, + "learning_rate": 5.997269230608935e-06, + "loss": 0.167, + "step": 12268 + }, + { + "epoch": 5.14, + "learning_rate": 5.995715622380949e-06, + "loss": 0.1558, + "step": 12269 + }, + { + "epoch": 5.14, + "learning_rate": 5.994162129252561e-06, + "loss": 0.1122, + "step": 12270 + }, + { + "epoch": 5.14, + "learning_rate": 5.992608751268429e-06, + "loss": 0.0987, + "step": 12271 + }, + { + "epoch": 5.14, + "learning_rate": 5.9910554884732056e-06, + "loss": 0.0981, + "step": 12272 + }, + { + "epoch": 5.14, + "learning_rate": 5.989502340911531e-06, + "loss": 0.1281, + "step": 12273 + }, + { + "epoch": 5.14, + "learning_rate": 5.987949308628054e-06, + "loss": 0.1062, + "step": 12274 + }, + { + "epoch": 5.14, + "learning_rate": 5.986396391667416e-06, + "loss": 0.0926, + "step": 12275 + }, + { + "epoch": 5.14, + "learning_rate": 5.98484359007425e-06, + "loss": 0.1085, + "step": 12276 + }, + { + "epoch": 5.14, + "learning_rate": 5.983290903893193e-06, + "loss": 0.1109, + "step": 12277 + }, + { + "epoch": 5.14, + "learning_rate": 5.981738333168873e-06, + "loss": 0.0862, + "step": 12278 + }, + { + "epoch": 5.14, + "learning_rate": 5.980185877945918e-06, + "loss": 0.1289, + "step": 12279 + }, + { + "epoch": 5.14, + "learning_rate": 5.978633538268957e-06, + "loss": 0.1209, + "step": 12280 + }, + { + "epoch": 5.14, + "learning_rate": 5.977081314182602e-06, + "loss": 0.1088, + "step": 12281 + }, + { + "epoch": 5.14, + "learning_rate": 5.975529205731476e-06, + "loss": 0.1414, + "step": 12282 + }, + { + "epoch": 5.14, + "learning_rate": 5.973977212960194e-06, + "loss": 0.0954, + "step": 12283 + }, + { + "epoch": 5.14, + "learning_rate": 5.9724253359133614e-06, + "loss": 0.1335, + "step": 12284 + }, + { + "epoch": 5.14, + "learning_rate": 5.97087357463559e-06, + "loss": 0.1011, + "step": 12285 + }, + { + "epoch": 5.14, + "learning_rate": 5.969321929171482e-06, + "loss": 0.104, + "step": 12286 + }, + { + "epoch": 5.14, + "learning_rate": 5.967770399565638e-06, + "loss": 0.1147, + "step": 12287 + }, + { + "epoch": 5.14, + "learning_rate": 5.966218985862655e-06, + "loss": 0.1452, + "step": 12288 + }, + { + "epoch": 5.15, + "learning_rate": 5.9646676881071305e-06, + "loss": 0.1098, + "step": 12289 + }, + { + "epoch": 5.15, + "learning_rate": 5.963116506343649e-06, + "loss": 0.1594, + "step": 12290 + }, + { + "epoch": 5.15, + "learning_rate": 5.961565440616803e-06, + "loss": 0.1281, + "step": 12291 + }, + { + "epoch": 5.15, + "learning_rate": 5.9600144909711745e-06, + "loss": 0.1219, + "step": 12292 + }, + { + "epoch": 5.15, + "learning_rate": 5.958463657451345e-06, + "loss": 0.1546, + "step": 12293 + }, + { + "epoch": 5.15, + "learning_rate": 5.956912940101891e-06, + "loss": 0.1529, + "step": 12294 + }, + { + "epoch": 5.15, + "learning_rate": 5.9553623389673855e-06, + "loss": 0.126, + "step": 12295 + }, + { + "epoch": 5.15, + "learning_rate": 5.953811854092404e-06, + "loss": 0.1451, + "step": 12296 + }, + { + "epoch": 5.15, + "learning_rate": 5.952261485521511e-06, + "loss": 0.15, + "step": 12297 + }, + { + "epoch": 5.15, + "learning_rate": 5.950711233299265e-06, + "loss": 0.1165, + "step": 12298 + }, + { + "epoch": 5.15, + "learning_rate": 5.949161097470233e-06, + "loss": 0.1291, + "step": 12299 + }, + { + "epoch": 5.15, + "learning_rate": 5.947611078078972e-06, + "loss": 0.1399, + "step": 12300 + }, + { + "epoch": 5.15, + "learning_rate": 5.9460611751700325e-06, + "loss": 0.1133, + "step": 12301 + }, + { + "epoch": 5.15, + "learning_rate": 5.944511388787967e-06, + "loss": 0.1605, + "step": 12302 + }, + { + "epoch": 5.15, + "learning_rate": 5.942961718977323e-06, + "loss": 0.089, + "step": 12303 + }, + { + "epoch": 5.15, + "learning_rate": 5.941412165782645e-06, + "loss": 0.1576, + "step": 12304 + }, + { + "epoch": 5.15, + "learning_rate": 5.939862729248472e-06, + "loss": 0.131, + "step": 12305 + }, + { + "epoch": 5.15, + "learning_rate": 5.9383134094193425e-06, + "loss": 0.0974, + "step": 12306 + }, + { + "epoch": 5.15, + "learning_rate": 5.936764206339791e-06, + "loss": 0.0774, + "step": 12307 + }, + { + "epoch": 5.15, + "learning_rate": 5.935215120054345e-06, + "loss": 0.1228, + "step": 12308 + }, + { + "epoch": 5.15, + "learning_rate": 5.933666150607533e-06, + "loss": 0.1299, + "step": 12309 + }, + { + "epoch": 5.15, + "learning_rate": 5.932117298043882e-06, + "loss": 0.1099, + "step": 12310 + }, + { + "epoch": 5.15, + "learning_rate": 5.930568562407905e-06, + "loss": 0.1544, + "step": 12311 + }, + { + "epoch": 5.15, + "learning_rate": 5.9290199437441256e-06, + "loss": 0.1006, + "step": 12312 + }, + { + "epoch": 5.16, + "learning_rate": 5.927471442097057e-06, + "loss": 0.1432, + "step": 12313 + }, + { + "epoch": 5.16, + "learning_rate": 5.925923057511207e-06, + "loss": 0.1034, + "step": 12314 + }, + { + "epoch": 5.16, + "learning_rate": 5.9243747900310825e-06, + "loss": 0.1427, + "step": 12315 + }, + { + "epoch": 5.16, + "learning_rate": 5.922826639701189e-06, + "loss": 0.1157, + "step": 12316 + }, + { + "epoch": 5.16, + "learning_rate": 5.921278606566026e-06, + "loss": 0.1101, + "step": 12317 + }, + { + "epoch": 5.16, + "learning_rate": 5.9197306906700865e-06, + "loss": 0.1233, + "step": 12318 + }, + { + "epoch": 5.16, + "learning_rate": 5.918182892057871e-06, + "loss": 0.1274, + "step": 12319 + }, + { + "epoch": 5.16, + "learning_rate": 5.916635210773869e-06, + "loss": 0.1298, + "step": 12320 + }, + { + "epoch": 5.16, + "learning_rate": 5.915087646862561e-06, + "loss": 0.0994, + "step": 12321 + }, + { + "epoch": 5.16, + "learning_rate": 5.913540200368434e-06, + "loss": 0.0831, + "step": 12322 + }, + { + "epoch": 5.16, + "learning_rate": 5.911992871335967e-06, + "loss": 0.1441, + "step": 12323 + }, + { + "epoch": 5.16, + "learning_rate": 5.910445659809638e-06, + "loss": 0.0953, + "step": 12324 + }, + { + "epoch": 5.16, + "learning_rate": 5.908898565833918e-06, + "loss": 0.1614, + "step": 12325 + }, + { + "epoch": 5.16, + "learning_rate": 5.907351589453279e-06, + "loss": 0.1289, + "step": 12326 + }, + { + "epoch": 5.16, + "learning_rate": 5.905804730712189e-06, + "loss": 0.1128, + "step": 12327 + }, + { + "epoch": 5.16, + "learning_rate": 5.904257989655107e-06, + "loss": 0.0952, + "step": 12328 + }, + { + "epoch": 5.16, + "learning_rate": 5.902711366326493e-06, + "loss": 0.1338, + "step": 12329 + }, + { + "epoch": 5.16, + "learning_rate": 5.901164860770808e-06, + "loss": 0.1145, + "step": 12330 + }, + { + "epoch": 5.16, + "learning_rate": 5.8996184730325005e-06, + "loss": 0.1594, + "step": 12331 + }, + { + "epoch": 5.16, + "learning_rate": 5.89807220315602e-06, + "loss": 0.1076, + "step": 12332 + }, + { + "epoch": 5.16, + "learning_rate": 5.896526051185814e-06, + "loss": 0.1145, + "step": 12333 + }, + { + "epoch": 5.16, + "learning_rate": 5.894980017166329e-06, + "loss": 0.1147, + "step": 12334 + }, + { + "epoch": 5.16, + "learning_rate": 5.893434101141997e-06, + "loss": 0.1126, + "step": 12335 + }, + { + "epoch": 5.16, + "learning_rate": 5.891888303157258e-06, + "loss": 0.0837, + "step": 12336 + }, + { + "epoch": 5.17, + "learning_rate": 5.890342623256546e-06, + "loss": 0.0819, + "step": 12337 + }, + { + "epoch": 5.17, + "learning_rate": 5.888797061484288e-06, + "loss": 0.1497, + "step": 12338 + }, + { + "epoch": 5.17, + "learning_rate": 5.887251617884909e-06, + "loss": 0.1503, + "step": 12339 + }, + { + "epoch": 5.17, + "learning_rate": 5.885706292502835e-06, + "loss": 0.1024, + "step": 12340 + }, + { + "epoch": 5.17, + "learning_rate": 5.884161085382481e-06, + "loss": 0.0938, + "step": 12341 + }, + { + "epoch": 5.17, + "learning_rate": 5.882615996568264e-06, + "loss": 0.1657, + "step": 12342 + }, + { + "epoch": 5.17, + "learning_rate": 5.8810710261046e-06, + "loss": 0.0848, + "step": 12343 + }, + { + "epoch": 5.17, + "learning_rate": 5.87952617403589e-06, + "loss": 0.1433, + "step": 12344 + }, + { + "epoch": 5.17, + "learning_rate": 5.877981440406547e-06, + "loss": 0.1287, + "step": 12345 + }, + { + "epoch": 5.17, + "learning_rate": 5.876436825260967e-06, + "loss": 0.1218, + "step": 12346 + }, + { + "epoch": 5.17, + "learning_rate": 5.8748923286435515e-06, + "loss": 0.1001, + "step": 12347 + }, + { + "epoch": 5.17, + "learning_rate": 5.873347950598697e-06, + "loss": 0.1375, + "step": 12348 + }, + { + "epoch": 5.17, + "learning_rate": 5.871803691170792e-06, + "loss": 0.1083, + "step": 12349 + }, + { + "epoch": 5.17, + "learning_rate": 5.870259550404227e-06, + "loss": 0.1457, + "step": 12350 + }, + { + "epoch": 5.17, + "learning_rate": 5.868715528343387e-06, + "loss": 0.1202, + "step": 12351 + }, + { + "epoch": 5.17, + "learning_rate": 5.8671716250326525e-06, + "loss": 0.0686, + "step": 12352 + }, + { + "epoch": 5.17, + "learning_rate": 5.865627840516402e-06, + "loss": 0.1238, + "step": 12353 + }, + { + "epoch": 5.17, + "learning_rate": 5.864084174839013e-06, + "loss": 0.0738, + "step": 12354 + }, + { + "epoch": 5.17, + "learning_rate": 5.86254062804485e-06, + "loss": 0.1097, + "step": 12355 + }, + { + "epoch": 5.17, + "learning_rate": 5.860997200178288e-06, + "loss": 0.1317, + "step": 12356 + }, + { + "epoch": 5.17, + "learning_rate": 5.859453891283688e-06, + "loss": 0.1103, + "step": 12357 + }, + { + "epoch": 5.17, + "learning_rate": 5.857910701405411e-06, + "loss": 0.1518, + "step": 12358 + }, + { + "epoch": 5.17, + "learning_rate": 5.8563676305878155e-06, + "loss": 0.1448, + "step": 12359 + }, + { + "epoch": 5.17, + "learning_rate": 5.854824678875256e-06, + "loss": 0.1091, + "step": 12360 + }, + { + "epoch": 5.18, + "learning_rate": 5.853281846312084e-06, + "loss": 0.1521, + "step": 12361 + }, + { + "epoch": 5.18, + "learning_rate": 5.851739132942645e-06, + "loss": 0.1361, + "step": 12362 + }, + { + "epoch": 5.18, + "learning_rate": 5.850196538811282e-06, + "loss": 0.1437, + "step": 12363 + }, + { + "epoch": 5.18, + "learning_rate": 5.848654063962338e-06, + "loss": 0.135, + "step": 12364 + }, + { + "epoch": 5.18, + "learning_rate": 5.847111708440152e-06, + "loss": 0.1703, + "step": 12365 + }, + { + "epoch": 5.18, + "learning_rate": 5.845569472289051e-06, + "loss": 0.1481, + "step": 12366 + }, + { + "epoch": 5.18, + "learning_rate": 5.844027355553369e-06, + "loss": 0.1317, + "step": 12367 + }, + { + "epoch": 5.18, + "learning_rate": 5.842485358277435e-06, + "loss": 0.1071, + "step": 12368 + }, + { + "epoch": 5.18, + "learning_rate": 5.840943480505567e-06, + "loss": 0.1361, + "step": 12369 + }, + { + "epoch": 5.18, + "learning_rate": 5.839401722282089e-06, + "loss": 0.113, + "step": 12370 + }, + { + "epoch": 5.18, + "learning_rate": 5.837860083651316e-06, + "loss": 0.0978, + "step": 12371 + }, + { + "epoch": 5.18, + "learning_rate": 5.836318564657561e-06, + "loss": 0.1384, + "step": 12372 + }, + { + "epoch": 5.18, + "learning_rate": 5.834777165345138e-06, + "loss": 0.0803, + "step": 12373 + }, + { + "epoch": 5.18, + "learning_rate": 5.833235885758346e-06, + "loss": 0.1489, + "step": 12374 + }, + { + "epoch": 5.18, + "learning_rate": 5.831694725941491e-06, + "loss": 0.1497, + "step": 12375 + }, + { + "epoch": 5.18, + "learning_rate": 5.830153685938873e-06, + "loss": 0.17, + "step": 12376 + }, + { + "epoch": 5.18, + "learning_rate": 5.8286127657947854e-06, + "loss": 0.1264, + "step": 12377 + }, + { + "epoch": 5.18, + "learning_rate": 5.827071965553524e-06, + "loss": 0.1041, + "step": 12378 + }, + { + "epoch": 5.18, + "learning_rate": 5.825531285259378e-06, + "loss": 0.1206, + "step": 12379 + }, + { + "epoch": 5.18, + "learning_rate": 5.8239907249566274e-06, + "loss": 0.1378, + "step": 12380 + }, + { + "epoch": 5.18, + "learning_rate": 5.822450284689558e-06, + "loss": 0.128, + "step": 12381 + }, + { + "epoch": 5.18, + "learning_rate": 5.820909964502448e-06, + "loss": 0.1407, + "step": 12382 + }, + { + "epoch": 5.18, + "learning_rate": 5.819369764439573e-06, + "loss": 0.1534, + "step": 12383 + }, + { + "epoch": 5.18, + "learning_rate": 5.817829684545204e-06, + "loss": 0.173, + "step": 12384 + }, + { + "epoch": 5.19, + "learning_rate": 5.816289724863609e-06, + "loss": 0.113, + "step": 12385 + }, + { + "epoch": 5.19, + "learning_rate": 5.814749885439057e-06, + "loss": 0.1181, + "step": 12386 + }, + { + "epoch": 5.19, + "learning_rate": 5.813210166315804e-06, + "loss": 0.1201, + "step": 12387 + }, + { + "epoch": 5.19, + "learning_rate": 5.81167056753811e-06, + "loss": 0.1776, + "step": 12388 + }, + { + "epoch": 5.19, + "learning_rate": 5.810131089150228e-06, + "loss": 0.146, + "step": 12389 + }, + { + "epoch": 5.19, + "learning_rate": 5.808591731196409e-06, + "loss": 0.1464, + "step": 12390 + }, + { + "epoch": 5.19, + "learning_rate": 5.8070524937209e-06, + "loss": 0.169, + "step": 12391 + }, + { + "epoch": 5.19, + "learning_rate": 5.805513376767947e-06, + "loss": 0.1243, + "step": 12392 + }, + { + "epoch": 5.19, + "learning_rate": 5.8039743803817904e-06, + "loss": 0.1317, + "step": 12393 + }, + { + "epoch": 5.19, + "learning_rate": 5.80243550460667e-06, + "loss": 0.1261, + "step": 12394 + }, + { + "epoch": 5.19, + "learning_rate": 5.800896749486813e-06, + "loss": 0.1377, + "step": 12395 + }, + { + "epoch": 5.19, + "learning_rate": 5.799358115066453e-06, + "loss": 0.1073, + "step": 12396 + }, + { + "epoch": 5.19, + "learning_rate": 5.797819601389815e-06, + "loss": 0.0979, + "step": 12397 + }, + { + "epoch": 5.19, + "learning_rate": 5.7962812085011245e-06, + "loss": 0.0694, + "step": 12398 + }, + { + "epoch": 5.19, + "learning_rate": 5.7947429364446e-06, + "loss": 0.1224, + "step": 12399 + }, + { + "epoch": 5.19, + "learning_rate": 5.793204785264461e-06, + "loss": 0.1102, + "step": 12400 + }, + { + "epoch": 5.19, + "learning_rate": 5.791666755004915e-06, + "loss": 0.1489, + "step": 12401 + }, + { + "epoch": 5.19, + "learning_rate": 5.790128845710175e-06, + "loss": 0.1471, + "step": 12402 + }, + { + "epoch": 5.19, + "learning_rate": 5.788591057424443e-06, + "loss": 0.123, + "step": 12403 + }, + { + "epoch": 5.19, + "learning_rate": 5.787053390191925e-06, + "loss": 0.0856, + "step": 12404 + }, + { + "epoch": 5.19, + "learning_rate": 5.785515844056818e-06, + "loss": 0.1065, + "step": 12405 + }, + { + "epoch": 5.19, + "learning_rate": 5.783978419063323e-06, + "loss": 0.1078, + "step": 12406 + }, + { + "epoch": 5.19, + "learning_rate": 5.782441115255624e-06, + "loss": 0.0962, + "step": 12407 + }, + { + "epoch": 5.19, + "learning_rate": 5.78090393267791e-06, + "loss": 0.13, + "step": 12408 + }, + { + "epoch": 5.2, + "learning_rate": 5.77936687137437e-06, + "loss": 0.134, + "step": 12409 + }, + { + "epoch": 5.2, + "learning_rate": 5.777829931389185e-06, + "loss": 0.125, + "step": 12410 + }, + { + "epoch": 5.2, + "learning_rate": 5.776293112766535e-06, + "loss": 0.1014, + "step": 12411 + }, + { + "epoch": 5.2, + "learning_rate": 5.774756415550587e-06, + "loss": 0.1036, + "step": 12412 + }, + { + "epoch": 5.2, + "learning_rate": 5.773219839785517e-06, + "loss": 0.1309, + "step": 12413 + }, + { + "epoch": 5.2, + "learning_rate": 5.771683385515495e-06, + "loss": 0.1354, + "step": 12414 + }, + { + "epoch": 5.2, + "learning_rate": 5.770147052784678e-06, + "loss": 0.1121, + "step": 12415 + }, + { + "epoch": 5.2, + "learning_rate": 5.768610841637232e-06, + "loss": 0.1305, + "step": 12416 + }, + { + "epoch": 5.2, + "learning_rate": 5.767074752117311e-06, + "loss": 0.1492, + "step": 12417 + }, + { + "epoch": 5.2, + "learning_rate": 5.765538784269071e-06, + "loss": 0.0997, + "step": 12418 + }, + { + "epoch": 5.2, + "learning_rate": 5.76400293813666e-06, + "loss": 0.0921, + "step": 12419 + }, + { + "epoch": 5.2, + "learning_rate": 5.762467213764231e-06, + "loss": 0.1206, + "step": 12420 + }, + { + "epoch": 5.2, + "learning_rate": 5.760931611195917e-06, + "loss": 0.1416, + "step": 12421 + }, + { + "epoch": 5.2, + "learning_rate": 5.75939613047586e-06, + "loss": 0.1282, + "step": 12422 + }, + { + "epoch": 5.2, + "learning_rate": 5.757860771648202e-06, + "loss": 0.1128, + "step": 12423 + }, + { + "epoch": 5.2, + "learning_rate": 5.75632553475707e-06, + "loss": 0.1247, + "step": 12424 + }, + { + "epoch": 5.2, + "learning_rate": 5.754790419846595e-06, + "loss": 0.1211, + "step": 12425 + }, + { + "epoch": 5.2, + "learning_rate": 5.753255426960901e-06, + "loss": 0.1361, + "step": 12426 + }, + { + "epoch": 5.2, + "learning_rate": 5.751720556144116e-06, + "loss": 0.1106, + "step": 12427 + }, + { + "epoch": 5.2, + "learning_rate": 5.750185807440349e-06, + "loss": 0.1182, + "step": 12428 + }, + { + "epoch": 5.2, + "learning_rate": 5.748651180893721e-06, + "loss": 0.1056, + "step": 12429 + }, + { + "epoch": 5.2, + "learning_rate": 5.747116676548342e-06, + "loss": 0.0918, + "step": 12430 + }, + { + "epoch": 5.2, + "learning_rate": 5.745582294448319e-06, + "loss": 0.1437, + "step": 12431 + }, + { + "epoch": 5.2, + "learning_rate": 5.744048034637759e-06, + "loss": 0.15, + "step": 12432 + }, + { + "epoch": 5.21, + "learning_rate": 5.742513897160766e-06, + "loss": 0.1019, + "step": 12433 + }, + { + "epoch": 5.21, + "learning_rate": 5.740979882061433e-06, + "loss": 0.0812, + "step": 12434 + }, + { + "epoch": 5.21, + "learning_rate": 5.7394459893838486e-06, + "loss": 0.0771, + "step": 12435 + }, + { + "epoch": 5.21, + "learning_rate": 5.737912219172108e-06, + "loss": 0.115, + "step": 12436 + }, + { + "epoch": 5.21, + "learning_rate": 5.7363785714703e-06, + "loss": 0.1307, + "step": 12437 + }, + { + "epoch": 5.21, + "learning_rate": 5.734845046322506e-06, + "loss": 0.1003, + "step": 12438 + }, + { + "epoch": 5.21, + "learning_rate": 5.733311643772807e-06, + "loss": 0.1297, + "step": 12439 + }, + { + "epoch": 5.21, + "learning_rate": 5.731778363865278e-06, + "loss": 0.1407, + "step": 12440 + }, + { + "epoch": 5.21, + "learning_rate": 5.730245206643996e-06, + "loss": 0.1088, + "step": 12441 + }, + { + "epoch": 5.21, + "learning_rate": 5.728712172153024e-06, + "loss": 0.1324, + "step": 12442 + }, + { + "epoch": 5.21, + "learning_rate": 5.727179260436428e-06, + "loss": 0.1405, + "step": 12443 + }, + { + "epoch": 5.21, + "learning_rate": 5.7256464715382754e-06, + "loss": 0.1619, + "step": 12444 + }, + { + "epoch": 5.21, + "learning_rate": 5.72411380550262e-06, + "loss": 0.088, + "step": 12445 + }, + { + "epoch": 5.21, + "learning_rate": 5.722581262373519e-06, + "loss": 0.1305, + "step": 12446 + }, + { + "epoch": 5.21, + "learning_rate": 5.721048842195028e-06, + "loss": 0.1233, + "step": 12447 + }, + { + "epoch": 5.21, + "learning_rate": 5.719516545011188e-06, + "loss": 0.127, + "step": 12448 + }, + { + "epoch": 5.21, + "learning_rate": 5.7179843708660454e-06, + "loss": 0.1287, + "step": 12449 + }, + { + "epoch": 5.21, + "learning_rate": 5.716452319803643e-06, + "loss": 0.1295, + "step": 12450 + }, + { + "epoch": 5.21, + "learning_rate": 5.7149203918680166e-06, + "loss": 0.1007, + "step": 12451 + }, + { + "epoch": 5.21, + "learning_rate": 5.713388587103201e-06, + "loss": 0.1025, + "step": 12452 + }, + { + "epoch": 5.21, + "learning_rate": 5.711856905553228e-06, + "loss": 0.1114, + "step": 12453 + }, + { + "epoch": 5.21, + "learning_rate": 5.710325347262125e-06, + "loss": 0.1141, + "step": 12454 + }, + { + "epoch": 5.21, + "learning_rate": 5.708793912273911e-06, + "loss": 0.1332, + "step": 12455 + }, + { + "epoch": 5.21, + "learning_rate": 5.707262600632612e-06, + "loss": 0.1265, + "step": 12456 + }, + { + "epoch": 5.22, + "learning_rate": 5.705731412382235e-06, + "loss": 0.1596, + "step": 12457 + }, + { + "epoch": 5.22, + "learning_rate": 5.704200347566798e-06, + "loss": 0.1179, + "step": 12458 + }, + { + "epoch": 5.22, + "learning_rate": 5.7026694062303115e-06, + "loss": 0.0919, + "step": 12459 + }, + { + "epoch": 5.22, + "learning_rate": 5.701138588416778e-06, + "loss": 0.1254, + "step": 12460 + }, + { + "epoch": 5.22, + "learning_rate": 5.699607894170206e-06, + "loss": 0.1102, + "step": 12461 + }, + { + "epoch": 5.22, + "learning_rate": 5.698077323534584e-06, + "loss": 0.1143, + "step": 12462 + }, + { + "epoch": 5.22, + "learning_rate": 5.696546876553912e-06, + "loss": 0.1093, + "step": 12463 + }, + { + "epoch": 5.22, + "learning_rate": 5.695016553272182e-06, + "loss": 0.1028, + "step": 12464 + }, + { + "epoch": 5.22, + "learning_rate": 5.693486353733382e-06, + "loss": 0.1587, + "step": 12465 + }, + { + "epoch": 5.22, + "learning_rate": 5.691956277981493e-06, + "loss": 0.1528, + "step": 12466 + }, + { + "epoch": 5.22, + "learning_rate": 5.690426326060498e-06, + "loss": 0.1301, + "step": 12467 + }, + { + "epoch": 5.22, + "learning_rate": 5.688896498014381e-06, + "loss": 0.1587, + "step": 12468 + }, + { + "epoch": 5.22, + "learning_rate": 5.687366793887101e-06, + "loss": 0.1473, + "step": 12469 + }, + { + "epoch": 5.22, + "learning_rate": 5.685837213722638e-06, + "loss": 0.1648, + "step": 12470 + }, + { + "epoch": 5.22, + "learning_rate": 5.684307757564954e-06, + "loss": 0.116, + "step": 12471 + }, + { + "epoch": 5.22, + "learning_rate": 5.682778425458015e-06, + "loss": 0.0865, + "step": 12472 + }, + { + "epoch": 5.22, + "learning_rate": 5.681249217445779e-06, + "loss": 0.1476, + "step": 12473 + }, + { + "epoch": 5.22, + "learning_rate": 5.6797201335722064e-06, + "loss": 0.142, + "step": 12474 + }, + { + "epoch": 5.22, + "learning_rate": 5.67819117388124e-06, + "loss": 0.1112, + "step": 12475 + }, + { + "epoch": 5.22, + "learning_rate": 5.676662338416833e-06, + "loss": 0.1331, + "step": 12476 + }, + { + "epoch": 5.22, + "learning_rate": 5.675133627222931e-06, + "loss": 0.1341, + "step": 12477 + }, + { + "epoch": 5.22, + "learning_rate": 5.673605040343475e-06, + "loss": 0.1504, + "step": 12478 + }, + { + "epoch": 5.22, + "learning_rate": 5.6720765778224075e-06, + "loss": 0.1181, + "step": 12479 + }, + { + "epoch": 5.23, + "learning_rate": 5.670548239703653e-06, + "loss": 0.1158, + "step": 12480 + }, + { + "epoch": 5.23, + "learning_rate": 5.669020026031148e-06, + "loss": 0.1225, + "step": 12481 + }, + { + "epoch": 5.23, + "learning_rate": 5.6674919368488235e-06, + "loss": 0.1307, + "step": 12482 + }, + { + "epoch": 5.23, + "learning_rate": 5.665963972200594e-06, + "loss": 0.121, + "step": 12483 + }, + { + "epoch": 5.23, + "learning_rate": 5.664436132130384e-06, + "loss": 0.1369, + "step": 12484 + }, + { + "epoch": 5.23, + "learning_rate": 5.662908416682111e-06, + "loss": 0.1592, + "step": 12485 + }, + { + "epoch": 5.23, + "learning_rate": 5.661380825899687e-06, + "loss": 0.1245, + "step": 12486 + }, + { + "epoch": 5.23, + "learning_rate": 5.6598533598270215e-06, + "loss": 0.0663, + "step": 12487 + }, + { + "epoch": 5.23, + "learning_rate": 5.658326018508022e-06, + "loss": 0.0951, + "step": 12488 + }, + { + "epoch": 5.23, + "learning_rate": 5.656798801986585e-06, + "loss": 0.1009, + "step": 12489 + }, + { + "epoch": 5.23, + "learning_rate": 5.655271710306613e-06, + "loss": 0.128, + "step": 12490 + }, + { + "epoch": 5.23, + "learning_rate": 5.653744743512e-06, + "loss": 0.1113, + "step": 12491 + }, + { + "epoch": 5.23, + "learning_rate": 5.652217901646637e-06, + "loss": 0.1389, + "step": 12492 + }, + { + "epoch": 5.23, + "learning_rate": 5.6506911847544134e-06, + "loss": 0.1356, + "step": 12493 + }, + { + "epoch": 5.23, + "learning_rate": 5.649164592879213e-06, + "loss": 0.1161, + "step": 12494 + }, + { + "epoch": 5.23, + "learning_rate": 5.647638126064917e-06, + "loss": 0.1361, + "step": 12495 + }, + { + "epoch": 5.23, + "learning_rate": 5.646111784355398e-06, + "loss": 0.112, + "step": 12496 + }, + { + "epoch": 5.23, + "learning_rate": 5.644585567794531e-06, + "loss": 0.1376, + "step": 12497 + }, + { + "epoch": 5.23, + "learning_rate": 5.643059476426189e-06, + "loss": 0.1104, + "step": 12498 + }, + { + "epoch": 5.23, + "learning_rate": 5.641533510294235e-06, + "loss": 0.115, + "step": 12499 + }, + { + "epoch": 5.23, + "learning_rate": 5.640007669442533e-06, + "loss": 0.0994, + "step": 12500 + }, + { + "epoch": 5.23, + "learning_rate": 5.638481953914945e-06, + "loss": 0.15, + "step": 12501 + }, + { + "epoch": 5.23, + "learning_rate": 5.636956363755324e-06, + "loss": 0.1126, + "step": 12502 + }, + { + "epoch": 5.23, + "learning_rate": 5.635430899007515e-06, + "loss": 0.1103, + "step": 12503 + }, + { + "epoch": 5.24, + "learning_rate": 5.6339055597153734e-06, + "loss": 0.1124, + "step": 12504 + }, + { + "epoch": 5.24, + "learning_rate": 5.632380345922741e-06, + "loss": 0.1223, + "step": 12505 + }, + { + "epoch": 5.24, + "learning_rate": 5.63085525767346e-06, + "loss": 0.1191, + "step": 12506 + }, + { + "epoch": 5.24, + "learning_rate": 5.629330295011367e-06, + "loss": 0.1154, + "step": 12507 + }, + { + "epoch": 5.24, + "learning_rate": 5.627805457980298e-06, + "loss": 0.1415, + "step": 12508 + }, + { + "epoch": 5.24, + "learning_rate": 5.626280746624083e-06, + "loss": 0.1086, + "step": 12509 + }, + { + "epoch": 5.24, + "learning_rate": 5.6247561609865445e-06, + "loss": 0.0688, + "step": 12510 + }, + { + "epoch": 5.24, + "learning_rate": 5.623231701111507e-06, + "loss": 0.1371, + "step": 12511 + }, + { + "epoch": 5.24, + "learning_rate": 5.6217073670427905e-06, + "loss": 0.1134, + "step": 12512 + }, + { + "epoch": 5.24, + "learning_rate": 5.62018315882421e-06, + "loss": 0.1037, + "step": 12513 + }, + { + "epoch": 5.24, + "learning_rate": 5.618659076499578e-06, + "loss": 0.126, + "step": 12514 + }, + { + "epoch": 5.24, + "learning_rate": 5.6171351201127065e-06, + "loss": 0.0937, + "step": 12515 + }, + { + "epoch": 5.24, + "learning_rate": 5.615611289707394e-06, + "loss": 0.1005, + "step": 12516 + }, + { + "epoch": 5.24, + "learning_rate": 5.614087585327444e-06, + "loss": 0.0986, + "step": 12517 + }, + { + "epoch": 5.24, + "learning_rate": 5.612564007016654e-06, + "loss": 0.1368, + "step": 12518 + }, + { + "epoch": 5.24, + "learning_rate": 5.611040554818819e-06, + "loss": 0.1007, + "step": 12519 + }, + { + "epoch": 5.24, + "learning_rate": 5.6095172287777275e-06, + "loss": 0.081, + "step": 12520 + }, + { + "epoch": 5.24, + "learning_rate": 5.607994028937171e-06, + "loss": 0.0976, + "step": 12521 + }, + { + "epoch": 5.24, + "learning_rate": 5.606470955340926e-06, + "loss": 0.1193, + "step": 12522 + }, + { + "epoch": 5.24, + "learning_rate": 5.604948008032774e-06, + "loss": 0.1216, + "step": 12523 + }, + { + "epoch": 5.24, + "learning_rate": 5.603425187056496e-06, + "loss": 0.0846, + "step": 12524 + }, + { + "epoch": 5.24, + "learning_rate": 5.601902492455855e-06, + "loss": 0.0941, + "step": 12525 + }, + { + "epoch": 5.24, + "learning_rate": 5.6003799242746235e-06, + "loss": 0.0941, + "step": 12526 + }, + { + "epoch": 5.24, + "learning_rate": 5.598857482556568e-06, + "loss": 0.107, + "step": 12527 + }, + { + "epoch": 5.25, + "learning_rate": 5.597335167345448e-06, + "loss": 0.1391, + "step": 12528 + }, + { + "epoch": 5.25, + "learning_rate": 5.595812978685027e-06, + "loss": 0.1228, + "step": 12529 + }, + { + "epoch": 5.25, + "learning_rate": 5.594290916619048e-06, + "loss": 0.1307, + "step": 12530 + }, + { + "epoch": 5.25, + "learning_rate": 5.592768981191268e-06, + "loss": 0.1124, + "step": 12531 + }, + { + "epoch": 5.25, + "learning_rate": 5.591247172445432e-06, + "loss": 0.1211, + "step": 12532 + }, + { + "epoch": 5.25, + "learning_rate": 5.589725490425284e-06, + "loss": 0.1426, + "step": 12533 + }, + { + "epoch": 5.25, + "learning_rate": 5.588203935174563e-06, + "loss": 0.1508, + "step": 12534 + }, + { + "epoch": 5.25, + "learning_rate": 5.586682506737007e-06, + "loss": 0.1433, + "step": 12535 + }, + { + "epoch": 5.25, + "learning_rate": 5.585161205156347e-06, + "loss": 0.0949, + "step": 12536 + }, + { + "epoch": 5.25, + "learning_rate": 5.583640030476309e-06, + "loss": 0.1226, + "step": 12537 + }, + { + "epoch": 5.25, + "learning_rate": 5.582118982740618e-06, + "loss": 0.1161, + "step": 12538 + }, + { + "epoch": 5.25, + "learning_rate": 5.580598061992997e-06, + "loss": 0.1382, + "step": 12539 + }, + { + "epoch": 5.25, + "learning_rate": 5.5790772682771625e-06, + "loss": 0.1094, + "step": 12540 + }, + { + "epoch": 5.25, + "learning_rate": 5.577556601636828e-06, + "loss": 0.1252, + "step": 12541 + }, + { + "epoch": 5.25, + "learning_rate": 5.576036062115709e-06, + "loss": 0.1765, + "step": 12542 + }, + { + "epoch": 5.25, + "learning_rate": 5.574515649757505e-06, + "loss": 0.1138, + "step": 12543 + }, + { + "epoch": 5.25, + "learning_rate": 5.572995364605921e-06, + "loss": 0.1526, + "step": 12544 + }, + { + "epoch": 5.25, + "learning_rate": 5.571475206704656e-06, + "loss": 0.1438, + "step": 12545 + }, + { + "epoch": 5.25, + "learning_rate": 5.569955176097408e-06, + "loss": 0.1175, + "step": 12546 + }, + { + "epoch": 5.25, + "learning_rate": 5.5684352728278705e-06, + "loss": 0.1257, + "step": 12547 + }, + { + "epoch": 5.25, + "learning_rate": 5.566915496939723e-06, + "loss": 0.1317, + "step": 12548 + }, + { + "epoch": 5.25, + "learning_rate": 5.565395848476658e-06, + "loss": 0.1251, + "step": 12549 + }, + { + "epoch": 5.25, + "learning_rate": 5.563876327482356e-06, + "loss": 0.1561, + "step": 12550 + }, + { + "epoch": 5.25, + "learning_rate": 5.56235693400049e-06, + "loss": 0.1404, + "step": 12551 + }, + { + "epoch": 5.26, + "learning_rate": 5.5608376680747354e-06, + "loss": 0.144, + "step": 12552 + }, + { + "epoch": 5.26, + "learning_rate": 5.559318529748763e-06, + "loss": 0.1175, + "step": 12553 + }, + { + "epoch": 5.26, + "learning_rate": 5.55779951906624e-06, + "loss": 0.1148, + "step": 12554 + }, + { + "epoch": 5.26, + "learning_rate": 5.556280636070826e-06, + "loss": 0.1124, + "step": 12555 + }, + { + "epoch": 5.26, + "learning_rate": 5.554761880806187e-06, + "loss": 0.102, + "step": 12556 + }, + { + "epoch": 5.26, + "learning_rate": 5.5532432533159676e-06, + "loss": 0.1266, + "step": 12557 + }, + { + "epoch": 5.26, + "learning_rate": 5.551724753643826e-06, + "loss": 0.1217, + "step": 12558 + }, + { + "epoch": 5.26, + "learning_rate": 5.550206381833409e-06, + "loss": 0.1142, + "step": 12559 + }, + { + "epoch": 5.26, + "learning_rate": 5.548688137928362e-06, + "loss": 0.0905, + "step": 12560 + }, + { + "epoch": 5.26, + "learning_rate": 5.547170021972323e-06, + "loss": 0.1191, + "step": 12561 + }, + { + "epoch": 5.26, + "learning_rate": 5.545652034008935e-06, + "loss": 0.1319, + "step": 12562 + }, + { + "epoch": 5.26, + "learning_rate": 5.544134174081821e-06, + "loss": 0.1422, + "step": 12563 + }, + { + "epoch": 5.26, + "learning_rate": 5.542616442234618e-06, + "loss": 0.0982, + "step": 12564 + }, + { + "epoch": 5.26, + "learning_rate": 5.54109883851095e-06, + "loss": 0.1629, + "step": 12565 + }, + { + "epoch": 5.26, + "learning_rate": 5.539581362954438e-06, + "loss": 0.1433, + "step": 12566 + }, + { + "epoch": 5.26, + "learning_rate": 5.5380640156087016e-06, + "loss": 0.0967, + "step": 12567 + }, + { + "epoch": 5.26, + "learning_rate": 5.536546796517357e-06, + "loss": 0.1558, + "step": 12568 + }, + { + "epoch": 5.26, + "learning_rate": 5.535029705724017e-06, + "loss": 0.0975, + "step": 12569 + }, + { + "epoch": 5.26, + "learning_rate": 5.533512743272285e-06, + "loss": 0.1051, + "step": 12570 + }, + { + "epoch": 5.26, + "learning_rate": 5.531995909205763e-06, + "loss": 0.1519, + "step": 12571 + }, + { + "epoch": 5.26, + "learning_rate": 5.530479203568053e-06, + "loss": 0.1319, + "step": 12572 + }, + { + "epoch": 5.26, + "learning_rate": 5.528962626402754e-06, + "loss": 0.1028, + "step": 12573 + }, + { + "epoch": 5.26, + "learning_rate": 5.527446177753455e-06, + "loss": 0.1478, + "step": 12574 + }, + { + "epoch": 5.26, + "learning_rate": 5.525929857663747e-06, + "loss": 0.1575, + "step": 12575 + }, + { + "epoch": 5.27, + "learning_rate": 5.524413666177216e-06, + "loss": 0.116, + "step": 12576 + }, + { + "epoch": 5.27, + "learning_rate": 5.522897603337445e-06, + "loss": 0.1317, + "step": 12577 + }, + { + "epoch": 5.27, + "learning_rate": 5.521381669188006e-06, + "loss": 0.1256, + "step": 12578 + }, + { + "epoch": 5.27, + "learning_rate": 5.519865863772477e-06, + "loss": 0.1148, + "step": 12579 + }, + { + "epoch": 5.27, + "learning_rate": 5.518350187134427e-06, + "loss": 0.1114, + "step": 12580 + }, + { + "epoch": 5.27, + "learning_rate": 5.516834639317426e-06, + "loss": 0.1411, + "step": 12581 + }, + { + "epoch": 5.27, + "learning_rate": 5.515319220365034e-06, + "loss": 0.1377, + "step": 12582 + }, + { + "epoch": 5.27, + "learning_rate": 5.513803930320815e-06, + "loss": 0.079, + "step": 12583 + }, + { + "epoch": 5.27, + "learning_rate": 5.512288769228318e-06, + "loss": 0.1564, + "step": 12584 + }, + { + "epoch": 5.27, + "learning_rate": 5.510773737131097e-06, + "loss": 0.1106, + "step": 12585 + }, + { + "epoch": 5.27, + "learning_rate": 5.509258834072702e-06, + "loss": 0.1236, + "step": 12586 + }, + { + "epoch": 5.27, + "learning_rate": 5.507744060096676e-06, + "loss": 0.1165, + "step": 12587 + }, + { + "epoch": 5.27, + "learning_rate": 5.50622941524656e-06, + "loss": 0.1437, + "step": 12588 + }, + { + "epoch": 5.27, + "learning_rate": 5.5047148995658976e-06, + "loss": 0.1063, + "step": 12589 + }, + { + "epoch": 5.27, + "learning_rate": 5.5032005130982115e-06, + "loss": 0.1413, + "step": 12590 + }, + { + "epoch": 5.27, + "learning_rate": 5.501686255887036e-06, + "loss": 0.1175, + "step": 12591 + }, + { + "epoch": 5.27, + "learning_rate": 5.500172127975903e-06, + "loss": 0.1253, + "step": 12592 + }, + { + "epoch": 5.27, + "learning_rate": 5.498658129408323e-06, + "loss": 0.1528, + "step": 12593 + }, + { + "epoch": 5.27, + "learning_rate": 5.49714426022782e-06, + "loss": 0.104, + "step": 12594 + }, + { + "epoch": 5.27, + "learning_rate": 5.495630520477912e-06, + "loss": 0.1079, + "step": 12595 + }, + { + "epoch": 5.27, + "learning_rate": 5.4941169102021065e-06, + "loss": 0.1383, + "step": 12596 + }, + { + "epoch": 5.27, + "learning_rate": 5.492603429443916e-06, + "loss": 0.1617, + "step": 12597 + }, + { + "epoch": 5.27, + "learning_rate": 5.491090078246836e-06, + "loss": 0.1331, + "step": 12598 + }, + { + "epoch": 5.27, + "learning_rate": 5.489576856654371e-06, + "loss": 0.161, + "step": 12599 + }, + { + "epoch": 5.28, + "learning_rate": 5.488063764710016e-06, + "loss": 0.1442, + "step": 12600 + }, + { + "epoch": 5.28, + "learning_rate": 5.486550802457264e-06, + "loss": 0.1255, + "step": 12601 + }, + { + "epoch": 5.28, + "learning_rate": 5.485037969939605e-06, + "loss": 0.1291, + "step": 12602 + }, + { + "epoch": 5.28, + "learning_rate": 5.483525267200527e-06, + "loss": 0.1409, + "step": 12603 + }, + { + "epoch": 5.28, + "learning_rate": 5.482012694283503e-06, + "loss": 0.1257, + "step": 12604 + }, + { + "epoch": 5.28, + "learning_rate": 5.480500251232015e-06, + "loss": 0.1324, + "step": 12605 + }, + { + "epoch": 5.28, + "learning_rate": 5.478987938089537e-06, + "loss": 0.1447, + "step": 12606 + }, + { + "epoch": 5.28, + "learning_rate": 5.477475754899538e-06, + "loss": 0.1482, + "step": 12607 + }, + { + "epoch": 5.28, + "learning_rate": 5.475963701705487e-06, + "loss": 0.1701, + "step": 12608 + }, + { + "epoch": 5.28, + "learning_rate": 5.4744517785508425e-06, + "loss": 0.1152, + "step": 12609 + }, + { + "epoch": 5.28, + "learning_rate": 5.472939985479071e-06, + "loss": 0.1722, + "step": 12610 + }, + { + "epoch": 5.28, + "learning_rate": 5.471428322533619e-06, + "loss": 0.124, + "step": 12611 + }, + { + "epoch": 5.28, + "learning_rate": 5.469916789757941e-06, + "loss": 0.1133, + "step": 12612 + }, + { + "epoch": 5.28, + "learning_rate": 5.4684053871954835e-06, + "loss": 0.0509, + "step": 12613 + }, + { + "epoch": 5.28, + "learning_rate": 5.466894114889697e-06, + "loss": 0.126, + "step": 12614 + }, + { + "epoch": 5.28, + "learning_rate": 5.4653829728840126e-06, + "loss": 0.117, + "step": 12615 + }, + { + "epoch": 5.28, + "learning_rate": 5.463871961221871e-06, + "loss": 0.115, + "step": 12616 + }, + { + "epoch": 5.28, + "learning_rate": 5.462361079946708e-06, + "loss": 0.1245, + "step": 12617 + }, + { + "epoch": 5.28, + "learning_rate": 5.460850329101944e-06, + "loss": 0.1079, + "step": 12618 + }, + { + "epoch": 5.28, + "learning_rate": 5.459339708731011e-06, + "loss": 0.0993, + "step": 12619 + }, + { + "epoch": 5.28, + "learning_rate": 5.457829218877328e-06, + "loss": 0.1243, + "step": 12620 + }, + { + "epoch": 5.28, + "learning_rate": 5.456318859584314e-06, + "loss": 0.1076, + "step": 12621 + }, + { + "epoch": 5.28, + "learning_rate": 5.454808630895383e-06, + "loss": 0.1082, + "step": 12622 + }, + { + "epoch": 5.28, + "learning_rate": 5.453298532853943e-06, + "loss": 0.1531, + "step": 12623 + }, + { + "epoch": 5.29, + "learning_rate": 5.451788565503406e-06, + "loss": 0.1146, + "step": 12624 + }, + { + "epoch": 5.29, + "learning_rate": 5.450278728887168e-06, + "loss": 0.1133, + "step": 12625 + }, + { + "epoch": 5.29, + "learning_rate": 5.44876902304863e-06, + "loss": 0.1498, + "step": 12626 + }, + { + "epoch": 5.29, + "learning_rate": 5.447259448031188e-06, + "loss": 0.1062, + "step": 12627 + }, + { + "epoch": 5.29, + "learning_rate": 5.445750003878233e-06, + "loss": 0.1089, + "step": 12628 + }, + { + "epoch": 5.29, + "learning_rate": 5.444240690633152e-06, + "loss": 0.1413, + "step": 12629 + }, + { + "epoch": 5.29, + "learning_rate": 5.442731508339334e-06, + "loss": 0.1212, + "step": 12630 + }, + { + "epoch": 5.29, + "learning_rate": 5.441222457040152e-06, + "loss": 0.1449, + "step": 12631 + }, + { + "epoch": 5.29, + "learning_rate": 5.439713536778983e-06, + "loss": 0.1023, + "step": 12632 + }, + { + "epoch": 5.29, + "learning_rate": 5.438204747599204e-06, + "loss": 0.135, + "step": 12633 + }, + { + "epoch": 5.29, + "learning_rate": 5.43669608954418e-06, + "loss": 0.1051, + "step": 12634 + }, + { + "epoch": 5.29, + "learning_rate": 5.435187562657278e-06, + "loss": 0.1177, + "step": 12635 + }, + { + "epoch": 5.29, + "learning_rate": 5.433679166981858e-06, + "loss": 0.1278, + "step": 12636 + }, + { + "epoch": 5.29, + "learning_rate": 5.432170902561282e-06, + "loss": 0.088, + "step": 12637 + }, + { + "epoch": 5.29, + "learning_rate": 5.4306627694389014e-06, + "loss": 0.1628, + "step": 12638 + }, + { + "epoch": 5.29, + "learning_rate": 5.42915476765806e-06, + "loss": 0.0765, + "step": 12639 + }, + { + "epoch": 5.29, + "learning_rate": 5.427646897262109e-06, + "loss": 0.1318, + "step": 12640 + }, + { + "epoch": 5.29, + "learning_rate": 5.42613915829439e-06, + "loss": 0.097, + "step": 12641 + }, + { + "epoch": 5.29, + "learning_rate": 5.424631550798243e-06, + "loss": 0.1231, + "step": 12642 + }, + { + "epoch": 5.29, + "learning_rate": 5.423124074817001e-06, + "loss": 0.1377, + "step": 12643 + }, + { + "epoch": 5.29, + "learning_rate": 5.421616730394e-06, + "loss": 0.1024, + "step": 12644 + }, + { + "epoch": 5.29, + "learning_rate": 5.42010951757256e-06, + "loss": 0.1169, + "step": 12645 + }, + { + "epoch": 5.29, + "learning_rate": 5.418602436396007e-06, + "loss": 0.1132, + "step": 12646 + }, + { + "epoch": 5.29, + "learning_rate": 5.417095486907663e-06, + "loss": 0.1527, + "step": 12647 + }, + { + "epoch": 5.3, + "learning_rate": 5.415588669150842e-06, + "loss": 0.1227, + "step": 12648 + }, + { + "epoch": 5.3, + "learning_rate": 5.414081983168855e-06, + "loss": 0.131, + "step": 12649 + }, + { + "epoch": 5.3, + "learning_rate": 5.412575429005014e-06, + "loss": 0.1329, + "step": 12650 + }, + { + "epoch": 5.3, + "learning_rate": 5.411069006702624e-06, + "loss": 0.1036, + "step": 12651 + }, + { + "epoch": 5.3, + "learning_rate": 5.409562716304979e-06, + "loss": 0.1419, + "step": 12652 + }, + { + "epoch": 5.3, + "learning_rate": 5.408056557855381e-06, + "loss": 0.125, + "step": 12653 + }, + { + "epoch": 5.3, + "learning_rate": 5.406550531397123e-06, + "loss": 0.1224, + "step": 12654 + }, + { + "epoch": 5.3, + "learning_rate": 5.405044636973493e-06, + "loss": 0.1051, + "step": 12655 + }, + { + "epoch": 5.3, + "learning_rate": 5.403538874627777e-06, + "loss": 0.131, + "step": 12656 + }, + { + "epoch": 5.3, + "learning_rate": 5.402033244403262e-06, + "loss": 0.0957, + "step": 12657 + }, + { + "epoch": 5.3, + "learning_rate": 5.400527746343217e-06, + "loss": 0.1377, + "step": 12658 + }, + { + "epoch": 5.3, + "learning_rate": 5.399022380490921e-06, + "loss": 0.086, + "step": 12659 + }, + { + "epoch": 5.3, + "learning_rate": 5.397517146889647e-06, + "loss": 0.0981, + "step": 12660 + }, + { + "epoch": 5.3, + "learning_rate": 5.3960120455826545e-06, + "loss": 0.1452, + "step": 12661 + }, + { + "epoch": 5.3, + "learning_rate": 5.394507076613212e-06, + "loss": 0.1332, + "step": 12662 + }, + { + "epoch": 5.3, + "learning_rate": 5.3930022400245745e-06, + "loss": 0.1038, + "step": 12663 + }, + { + "epoch": 5.3, + "learning_rate": 5.391497535859999e-06, + "loss": 0.1076, + "step": 12664 + }, + { + "epoch": 5.3, + "learning_rate": 5.3899929641627425e-06, + "loss": 0.1232, + "step": 12665 + }, + { + "epoch": 5.3, + "learning_rate": 5.388488524976043e-06, + "loss": 0.1222, + "step": 12666 + }, + { + "epoch": 5.3, + "learning_rate": 5.3869842183431495e-06, + "loss": 0.1345, + "step": 12667 + }, + { + "epoch": 5.3, + "learning_rate": 5.3854800443073006e-06, + "loss": 0.0914, + "step": 12668 + }, + { + "epoch": 5.3, + "learning_rate": 5.383976002911734e-06, + "loss": 0.152, + "step": 12669 + }, + { + "epoch": 5.3, + "learning_rate": 5.38247209419968e-06, + "loss": 0.1065, + "step": 12670 + }, + { + "epoch": 5.31, + "learning_rate": 5.3809683182143725e-06, + "loss": 0.1336, + "step": 12671 + }, + { + "epoch": 5.31, + "learning_rate": 5.3794646749990265e-06, + "loss": 0.1259, + "step": 12672 + }, + { + "epoch": 5.31, + "learning_rate": 5.3779611645968696e-06, + "loss": 0.1357, + "step": 12673 + }, + { + "epoch": 5.31, + "learning_rate": 5.376457787051117e-06, + "loss": 0.1132, + "step": 12674 + }, + { + "epoch": 5.31, + "learning_rate": 5.374954542404982e-06, + "loss": 0.1397, + "step": 12675 + }, + { + "epoch": 5.31, + "learning_rate": 5.373451430701676e-06, + "loss": 0.0993, + "step": 12676 + }, + { + "epoch": 5.31, + "learning_rate": 5.371948451984401e-06, + "loss": 0.0798, + "step": 12677 + }, + { + "epoch": 5.31, + "learning_rate": 5.3704456062963674e-06, + "loss": 0.1446, + "step": 12678 + }, + { + "epoch": 5.31, + "learning_rate": 5.368942893680761e-06, + "loss": 0.1033, + "step": 12679 + }, + { + "epoch": 5.31, + "learning_rate": 5.367440314180782e-06, + "loss": 0.125, + "step": 12680 + }, + { + "epoch": 5.31, + "learning_rate": 5.36593786783962e-06, + "loss": 0.1587, + "step": 12681 + }, + { + "epoch": 5.31, + "learning_rate": 5.3644355547004664e-06, + "loss": 0.1135, + "step": 12682 + }, + { + "epoch": 5.31, + "learning_rate": 5.362933374806494e-06, + "loss": 0.1085, + "step": 12683 + }, + { + "epoch": 5.31, + "learning_rate": 5.361431328200888e-06, + "loss": 0.1236, + "step": 12684 + }, + { + "epoch": 5.31, + "learning_rate": 5.3599294149268255e-06, + "loss": 0.1151, + "step": 12685 + }, + { + "epoch": 5.31, + "learning_rate": 5.3584276350274725e-06, + "loss": 0.1447, + "step": 12686 + }, + { + "epoch": 5.31, + "learning_rate": 5.356925988545996e-06, + "loss": 0.1179, + "step": 12687 + }, + { + "epoch": 5.31, + "learning_rate": 5.355424475525562e-06, + "loss": 0.1462, + "step": 12688 + }, + { + "epoch": 5.31, + "learning_rate": 5.3539230960093305e-06, + "loss": 0.1122, + "step": 12689 + }, + { + "epoch": 5.31, + "learning_rate": 5.352421850040457e-06, + "loss": 0.1016, + "step": 12690 + }, + { + "epoch": 5.31, + "learning_rate": 5.3509207376620935e-06, + "loss": 0.1547, + "step": 12691 + }, + { + "epoch": 5.31, + "learning_rate": 5.349419758917392e-06, + "loss": 0.1373, + "step": 12692 + }, + { + "epoch": 5.31, + "learning_rate": 5.347918913849487e-06, + "loss": 0.1006, + "step": 12693 + }, + { + "epoch": 5.31, + "learning_rate": 5.346418202501526e-06, + "loss": 0.1653, + "step": 12694 + }, + { + "epoch": 5.32, + "learning_rate": 5.344917624916644e-06, + "loss": 0.1047, + "step": 12695 + }, + { + "epoch": 5.32, + "learning_rate": 5.343417181137973e-06, + "loss": 0.1224, + "step": 12696 + }, + { + "epoch": 5.32, + "learning_rate": 5.341916871208644e-06, + "loss": 0.1222, + "step": 12697 + }, + { + "epoch": 5.32, + "learning_rate": 5.340416695171785e-06, + "loss": 0.1651, + "step": 12698 + }, + { + "epoch": 5.32, + "learning_rate": 5.3389166530705096e-06, + "loss": 0.1085, + "step": 12699 + }, + { + "epoch": 5.32, + "learning_rate": 5.337416744947939e-06, + "loss": 0.1256, + "step": 12700 + }, + { + "epoch": 5.32, + "learning_rate": 5.335916970847185e-06, + "loss": 0.0904, + "step": 12701 + }, + { + "epoch": 5.32, + "learning_rate": 5.3344173308113586e-06, + "loss": 0.1208, + "step": 12702 + }, + { + "epoch": 5.32, + "learning_rate": 5.332917824883568e-06, + "loss": 0.0958, + "step": 12703 + }, + { + "epoch": 5.32, + "learning_rate": 5.331418453106916e-06, + "loss": 0.1328, + "step": 12704 + }, + { + "epoch": 5.32, + "learning_rate": 5.329919215524494e-06, + "loss": 0.1195, + "step": 12705 + }, + { + "epoch": 5.32, + "learning_rate": 5.328420112179405e-06, + "loss": 0.094, + "step": 12706 + }, + { + "epoch": 5.32, + "learning_rate": 5.326921143114727e-06, + "loss": 0.1097, + "step": 12707 + }, + { + "epoch": 5.32, + "learning_rate": 5.325422308373557e-06, + "loss": 0.1508, + "step": 12708 + }, + { + "epoch": 5.32, + "learning_rate": 5.3239236079989735e-06, + "loss": 0.0881, + "step": 12709 + }, + { + "epoch": 5.32, + "learning_rate": 5.322425042034056e-06, + "loss": 0.1582, + "step": 12710 + }, + { + "epoch": 5.32, + "learning_rate": 5.3209266105218815e-06, + "loss": 0.1305, + "step": 12711 + }, + { + "epoch": 5.32, + "learning_rate": 5.319428313505523e-06, + "loss": 0.1466, + "step": 12712 + }, + { + "epoch": 5.32, + "learning_rate": 5.317930151028039e-06, + "loss": 0.1536, + "step": 12713 + }, + { + "epoch": 5.32, + "learning_rate": 5.316432123132498e-06, + "loss": 0.1155, + "step": 12714 + }, + { + "epoch": 5.32, + "learning_rate": 5.314934229861961e-06, + "loss": 0.1389, + "step": 12715 + }, + { + "epoch": 5.32, + "learning_rate": 5.3134364712594814e-06, + "loss": 0.1402, + "step": 12716 + }, + { + "epoch": 5.32, + "learning_rate": 5.311938847368111e-06, + "loss": 0.1609, + "step": 12717 + }, + { + "epoch": 5.32, + "learning_rate": 5.310441358230899e-06, + "loss": 0.1318, + "step": 12718 + }, + { + "epoch": 5.33, + "learning_rate": 5.308944003890892e-06, + "loss": 0.1152, + "step": 12719 + }, + { + "epoch": 5.33, + "learning_rate": 5.307446784391124e-06, + "loss": 0.1604, + "step": 12720 + }, + { + "epoch": 5.33, + "learning_rate": 5.305949699774633e-06, + "loss": 0.1279, + "step": 12721 + }, + { + "epoch": 5.33, + "learning_rate": 5.304452750084452e-06, + "loss": 0.1333, + "step": 12722 + }, + { + "epoch": 5.33, + "learning_rate": 5.3029559353636116e-06, + "loss": 0.1252, + "step": 12723 + }, + { + "epoch": 5.33, + "learning_rate": 5.301459255655134e-06, + "loss": 0.1178, + "step": 12724 + }, + { + "epoch": 5.33, + "learning_rate": 5.299962711002045e-06, + "loss": 0.1461, + "step": 12725 + }, + { + "epoch": 5.33, + "learning_rate": 5.298466301447353e-06, + "loss": 0.1093, + "step": 12726 + }, + { + "epoch": 5.33, + "learning_rate": 5.296970027034074e-06, + "loss": 0.1464, + "step": 12727 + }, + { + "epoch": 5.33, + "learning_rate": 5.295473887805223e-06, + "loss": 0.1338, + "step": 12728 + }, + { + "epoch": 5.33, + "learning_rate": 5.293977883803797e-06, + "loss": 0.1077, + "step": 12729 + }, + { + "epoch": 5.33, + "learning_rate": 5.292482015072799e-06, + "loss": 0.1061, + "step": 12730 + }, + { + "epoch": 5.33, + "learning_rate": 5.29098628165523e-06, + "loss": 0.1552, + "step": 12731 + }, + { + "epoch": 5.33, + "learning_rate": 5.289490683594081e-06, + "loss": 0.1318, + "step": 12732 + }, + { + "epoch": 5.33, + "learning_rate": 5.287995220932346e-06, + "loss": 0.1018, + "step": 12733 + }, + { + "epoch": 5.33, + "learning_rate": 5.286499893713003e-06, + "loss": 0.131, + "step": 12734 + }, + { + "epoch": 5.33, + "learning_rate": 5.285004701979037e-06, + "loss": 0.092, + "step": 12735 + }, + { + "epoch": 5.33, + "learning_rate": 5.283509645773429e-06, + "loss": 0.1369, + "step": 12736 + }, + { + "epoch": 5.33, + "learning_rate": 5.282014725139149e-06, + "loss": 0.1574, + "step": 12737 + }, + { + "epoch": 5.33, + "learning_rate": 5.280519940119171e-06, + "loss": 0.1436, + "step": 12738 + }, + { + "epoch": 5.33, + "learning_rate": 5.279025290756461e-06, + "loss": 0.0928, + "step": 12739 + }, + { + "epoch": 5.33, + "learning_rate": 5.2775307770939775e-06, + "loss": 0.1305, + "step": 12740 + }, + { + "epoch": 5.33, + "learning_rate": 5.27603639917468e-06, + "loss": 0.11, + "step": 12741 + }, + { + "epoch": 5.33, + "learning_rate": 5.2745421570415245e-06, + "loss": 0.1133, + "step": 12742 + }, + { + "epoch": 5.34, + "learning_rate": 5.273048050737461e-06, + "loss": 0.0938, + "step": 12743 + }, + { + "epoch": 5.34, + "learning_rate": 5.271554080305439e-06, + "loss": 0.1041, + "step": 12744 + }, + { + "epoch": 5.34, + "learning_rate": 5.270060245788401e-06, + "loss": 0.1204, + "step": 12745 + }, + { + "epoch": 5.34, + "learning_rate": 5.2685665472292805e-06, + "loss": 0.1563, + "step": 12746 + }, + { + "epoch": 5.34, + "learning_rate": 5.2670729846710155e-06, + "loss": 0.1355, + "step": 12747 + }, + { + "epoch": 5.34, + "learning_rate": 5.265579558156537e-06, + "loss": 0.1159, + "step": 12748 + }, + { + "epoch": 5.34, + "learning_rate": 5.264086267728774e-06, + "loss": 0.1063, + "step": 12749 + }, + { + "epoch": 5.34, + "learning_rate": 5.262593113430653e-06, + "loss": 0.1926, + "step": 12750 + }, + { + "epoch": 5.34, + "learning_rate": 5.261100095305085e-06, + "loss": 0.1129, + "step": 12751 + }, + { + "epoch": 5.34, + "learning_rate": 5.259607213394989e-06, + "loss": 0.1329, + "step": 12752 + }, + { + "epoch": 5.34, + "learning_rate": 5.2581144677432805e-06, + "loss": 0.1095, + "step": 12753 + }, + { + "epoch": 5.34, + "learning_rate": 5.256621858392861e-06, + "loss": 0.0983, + "step": 12754 + }, + { + "epoch": 5.34, + "learning_rate": 5.255129385386635e-06, + "loss": 0.1131, + "step": 12755 + }, + { + "epoch": 5.34, + "learning_rate": 5.2536370487675056e-06, + "loss": 0.1182, + "step": 12756 + }, + { + "epoch": 5.34, + "learning_rate": 5.252144848578368e-06, + "loss": 0.101, + "step": 12757 + }, + { + "epoch": 5.34, + "learning_rate": 5.250652784862114e-06, + "loss": 0.1241, + "step": 12758 + }, + { + "epoch": 5.34, + "learning_rate": 5.249160857661628e-06, + "loss": 0.1303, + "step": 12759 + }, + { + "epoch": 5.34, + "learning_rate": 5.247669067019805e-06, + "loss": 0.0913, + "step": 12760 + }, + { + "epoch": 5.34, + "learning_rate": 5.246177412979512e-06, + "loss": 0.0975, + "step": 12761 + }, + { + "epoch": 5.34, + "learning_rate": 5.2446858955836295e-06, + "loss": 0.0928, + "step": 12762 + }, + { + "epoch": 5.34, + "learning_rate": 5.243194514875033e-06, + "loss": 0.1463, + "step": 12763 + }, + { + "epoch": 5.34, + "learning_rate": 5.241703270896589e-06, + "loss": 0.1404, + "step": 12764 + }, + { + "epoch": 5.34, + "learning_rate": 5.240212163691161e-06, + "loss": 0.103, + "step": 12765 + }, + { + "epoch": 5.34, + "learning_rate": 5.2387211933016145e-06, + "loss": 0.1103, + "step": 12766 + }, + { + "epoch": 5.35, + "learning_rate": 5.237230359770799e-06, + "loss": 0.1328, + "step": 12767 + }, + { + "epoch": 5.35, + "learning_rate": 5.235739663141571e-06, + "loss": 0.1178, + "step": 12768 + }, + { + "epoch": 5.35, + "learning_rate": 5.2342491034567775e-06, + "loss": 0.124, + "step": 12769 + }, + { + "epoch": 5.35, + "learning_rate": 5.232758680759264e-06, + "loss": 0.0963, + "step": 12770 + }, + { + "epoch": 5.35, + "learning_rate": 5.231268395091874e-06, + "loss": 0.1234, + "step": 12771 + }, + { + "epoch": 5.35, + "learning_rate": 5.229778246497446e-06, + "loss": 0.117, + "step": 12772 + }, + { + "epoch": 5.35, + "learning_rate": 5.2282882350188035e-06, + "loss": 0.1322, + "step": 12773 + }, + { + "epoch": 5.35, + "learning_rate": 5.226798360698786e-06, + "loss": 0.1263, + "step": 12774 + }, + { + "epoch": 5.35, + "learning_rate": 5.225308623580209e-06, + "loss": 0.1048, + "step": 12775 + }, + { + "epoch": 5.35, + "learning_rate": 5.2238190237059e-06, + "loss": 0.1328, + "step": 12776 + }, + { + "epoch": 5.35, + "learning_rate": 5.222329561118675e-06, + "loss": 0.1141, + "step": 12777 + }, + { + "epoch": 5.35, + "learning_rate": 5.2208402358613465e-06, + "loss": 0.1137, + "step": 12778 + }, + { + "epoch": 5.35, + "learning_rate": 5.219351047976725e-06, + "loss": 0.1288, + "step": 12779 + }, + { + "epoch": 5.35, + "learning_rate": 5.217861997507618e-06, + "loss": 0.1239, + "step": 12780 + }, + { + "epoch": 5.35, + "learning_rate": 5.216373084496822e-06, + "loss": 0.1355, + "step": 12781 + }, + { + "epoch": 5.35, + "learning_rate": 5.214884308987136e-06, + "loss": 0.1093, + "step": 12782 + }, + { + "epoch": 5.35, + "learning_rate": 5.213395671021355e-06, + "loss": 0.1434, + "step": 12783 + }, + { + "epoch": 5.35, + "learning_rate": 5.211907170642267e-06, + "loss": 0.1132, + "step": 12784 + }, + { + "epoch": 5.35, + "learning_rate": 5.210418807892659e-06, + "loss": 0.1423, + "step": 12785 + }, + { + "epoch": 5.35, + "learning_rate": 5.208930582815316e-06, + "loss": 0.1334, + "step": 12786 + }, + { + "epoch": 5.35, + "learning_rate": 5.207442495453008e-06, + "loss": 0.1015, + "step": 12787 + }, + { + "epoch": 5.35, + "learning_rate": 5.205954545848513e-06, + "loss": 0.1216, + "step": 12788 + }, + { + "epoch": 5.35, + "learning_rate": 5.2044667340445985e-06, + "loss": 0.1396, + "step": 12789 + }, + { + "epoch": 5.35, + "learning_rate": 5.202979060084035e-06, + "loss": 0.1208, + "step": 12790 + }, + { + "epoch": 5.36, + "learning_rate": 5.2014915240095785e-06, + "loss": 0.1006, + "step": 12791 + }, + { + "epoch": 5.36, + "learning_rate": 5.200004125863993e-06, + "loss": 0.0823, + "step": 12792 + }, + { + "epoch": 5.36, + "learning_rate": 5.198516865690031e-06, + "loss": 0.1439, + "step": 12793 + }, + { + "epoch": 5.36, + "learning_rate": 5.1970297435304395e-06, + "loss": 0.0832, + "step": 12794 + }, + { + "epoch": 5.36, + "learning_rate": 5.195542759427964e-06, + "loss": 0.1474, + "step": 12795 + }, + { + "epoch": 5.36, + "learning_rate": 5.194055913425353e-06, + "loss": 0.1134, + "step": 12796 + }, + { + "epoch": 5.36, + "learning_rate": 5.192569205565336e-06, + "loss": 0.1257, + "step": 12797 + }, + { + "epoch": 5.36, + "learning_rate": 5.191082635890652e-06, + "loss": 0.0989, + "step": 12798 + }, + { + "epoch": 5.36, + "learning_rate": 5.1895962044440295e-06, + "loss": 0.0935, + "step": 12799 + }, + { + "epoch": 5.36, + "learning_rate": 5.188109911268196e-06, + "loss": 0.1132, + "step": 12800 + }, + { + "epoch": 5.36, + "learning_rate": 5.186623756405875e-06, + "loss": 0.1229, + "step": 12801 + }, + { + "epoch": 5.36, + "learning_rate": 5.185137739899781e-06, + "loss": 0.1153, + "step": 12802 + }, + { + "epoch": 5.36, + "learning_rate": 5.183651861792629e-06, + "loss": 0.125, + "step": 12803 + }, + { + "epoch": 5.36, + "learning_rate": 5.18216612212713e-06, + "loss": 0.1001, + "step": 12804 + }, + { + "epoch": 5.36, + "learning_rate": 5.180680520945992e-06, + "loss": 0.1196, + "step": 12805 + }, + { + "epoch": 5.36, + "learning_rate": 5.179195058291915e-06, + "loss": 0.0949, + "step": 12806 + }, + { + "epoch": 5.36, + "learning_rate": 5.177709734207603e-06, + "loss": 0.1265, + "step": 12807 + }, + { + "epoch": 5.36, + "learning_rate": 5.17622454873574e-06, + "loss": 0.1384, + "step": 12808 + }, + { + "epoch": 5.36, + "learning_rate": 5.174739501919023e-06, + "loss": 0.1639, + "step": 12809 + }, + { + "epoch": 5.36, + "learning_rate": 5.173254593800137e-06, + "loss": 0.1167, + "step": 12810 + }, + { + "epoch": 5.36, + "learning_rate": 5.171769824421764e-06, + "loss": 0.1537, + "step": 12811 + }, + { + "epoch": 5.36, + "learning_rate": 5.170285193826583e-06, + "loss": 0.1342, + "step": 12812 + }, + { + "epoch": 5.36, + "learning_rate": 5.168800702057272e-06, + "loss": 0.1059, + "step": 12813 + }, + { + "epoch": 5.36, + "learning_rate": 5.167316349156495e-06, + "loss": 0.0931, + "step": 12814 + }, + { + "epoch": 5.37, + "learning_rate": 5.16583213516692e-06, + "loss": 0.1204, + "step": 12815 + }, + { + "epoch": 5.37, + "learning_rate": 5.16434806013121e-06, + "loss": 0.1225, + "step": 12816 + }, + { + "epoch": 5.37, + "learning_rate": 5.162864124092024e-06, + "loss": 0.0935, + "step": 12817 + }, + { + "epoch": 5.37, + "learning_rate": 5.161380327092021e-06, + "loss": 0.1368, + "step": 12818 + }, + { + "epoch": 5.37, + "learning_rate": 5.1598966691738425e-06, + "loss": 0.0978, + "step": 12819 + }, + { + "epoch": 5.37, + "learning_rate": 5.158413150380139e-06, + "loss": 0.1169, + "step": 12820 + }, + { + "epoch": 5.37, + "learning_rate": 5.1569297707535585e-06, + "loss": 0.1467, + "step": 12821 + }, + { + "epoch": 5.37, + "learning_rate": 5.155446530336729e-06, + "loss": 0.132, + "step": 12822 + }, + { + "epoch": 5.37, + "learning_rate": 5.15396342917229e-06, + "loss": 0.128, + "step": 12823 + }, + { + "epoch": 5.37, + "learning_rate": 5.1524804673028725e-06, + "loss": 0.1187, + "step": 12824 + }, + { + "epoch": 5.37, + "learning_rate": 5.150997644771102e-06, + "loss": 0.1133, + "step": 12825 + }, + { + "epoch": 5.37, + "learning_rate": 5.149514961619602e-06, + "loss": 0.1066, + "step": 12826 + }, + { + "epoch": 5.37, + "learning_rate": 5.148032417890992e-06, + "loss": 0.1473, + "step": 12827 + }, + { + "epoch": 5.37, + "learning_rate": 5.1465500136278825e-06, + "loss": 0.1187, + "step": 12828 + }, + { + "epoch": 5.37, + "learning_rate": 5.145067748872885e-06, + "loss": 0.1354, + "step": 12829 + }, + { + "epoch": 5.37, + "learning_rate": 5.143585623668607e-06, + "loss": 0.129, + "step": 12830 + }, + { + "epoch": 5.37, + "learning_rate": 5.142103638057652e-06, + "loss": 0.14, + "step": 12831 + }, + { + "epoch": 5.37, + "learning_rate": 5.140621792082616e-06, + "loss": 0.1078, + "step": 12832 + }, + { + "epoch": 5.37, + "learning_rate": 5.139140085786095e-06, + "loss": 0.1107, + "step": 12833 + }, + { + "epoch": 5.37, + "learning_rate": 5.1376585192106824e-06, + "loss": 0.1486, + "step": 12834 + }, + { + "epoch": 5.37, + "learning_rate": 5.136177092398957e-06, + "loss": 0.1297, + "step": 12835 + }, + { + "epoch": 5.37, + "learning_rate": 5.1346958053935045e-06, + "loss": 0.1109, + "step": 12836 + }, + { + "epoch": 5.37, + "learning_rate": 5.133214658236905e-06, + "loss": 0.0966, + "step": 12837 + }, + { + "epoch": 5.37, + "learning_rate": 5.131733650971731e-06, + "loss": 0.1232, + "step": 12838 + }, + { + "epoch": 5.38, + "learning_rate": 5.130252783640553e-06, + "loss": 0.1504, + "step": 12839 + }, + { + "epoch": 5.38, + "learning_rate": 5.12877205628594e-06, + "loss": 0.0922, + "step": 12840 + }, + { + "epoch": 5.38, + "learning_rate": 5.127291468950453e-06, + "loss": 0.1228, + "step": 12841 + }, + { + "epoch": 5.38, + "learning_rate": 5.125811021676643e-06, + "loss": 0.1426, + "step": 12842 + }, + { + "epoch": 5.38, + "learning_rate": 5.124330714507073e-06, + "loss": 0.0989, + "step": 12843 + }, + { + "epoch": 5.38, + "learning_rate": 5.1228505474842885e-06, + "loss": 0.1034, + "step": 12844 + }, + { + "epoch": 5.38, + "learning_rate": 5.121370520650837e-06, + "loss": 0.1275, + "step": 12845 + }, + { + "epoch": 5.38, + "learning_rate": 5.1198906340492605e-06, + "loss": 0.1123, + "step": 12846 + }, + { + "epoch": 5.38, + "learning_rate": 5.118410887722096e-06, + "loss": 0.11, + "step": 12847 + }, + { + "epoch": 5.38, + "learning_rate": 5.116931281711886e-06, + "loss": 0.143, + "step": 12848 + }, + { + "epoch": 5.38, + "learning_rate": 5.115451816061146e-06, + "loss": 0.1328, + "step": 12849 + }, + { + "epoch": 5.38, + "learning_rate": 5.113972490812411e-06, + "loss": 0.1434, + "step": 12850 + }, + { + "epoch": 5.38, + "learning_rate": 5.112493306008199e-06, + "loss": 0.1443, + "step": 12851 + }, + { + "epoch": 5.38, + "learning_rate": 5.111014261691032e-06, + "loss": 0.088, + "step": 12852 + }, + { + "epoch": 5.38, + "learning_rate": 5.109535357903421e-06, + "loss": 0.1044, + "step": 12853 + }, + { + "epoch": 5.38, + "learning_rate": 5.10805659468788e-06, + "loss": 0.1146, + "step": 12854 + }, + { + "epoch": 5.38, + "learning_rate": 5.106577972086907e-06, + "loss": 0.1258, + "step": 12855 + }, + { + "epoch": 5.38, + "learning_rate": 5.105099490143008e-06, + "loss": 0.1459, + "step": 12856 + }, + { + "epoch": 5.38, + "learning_rate": 5.1036211488986806e-06, + "loss": 0.1133, + "step": 12857 + }, + { + "epoch": 5.38, + "learning_rate": 5.102142948396418e-06, + "loss": 0.1259, + "step": 12858 + }, + { + "epoch": 5.38, + "learning_rate": 5.10066488867871e-06, + "loss": 0.0901, + "step": 12859 + }, + { + "epoch": 5.38, + "learning_rate": 5.099186969788044e-06, + "loss": 0.12, + "step": 12860 + }, + { + "epoch": 5.38, + "learning_rate": 5.097709191766902e-06, + "loss": 0.1217, + "step": 12861 + }, + { + "epoch": 5.38, + "learning_rate": 5.096231554657755e-06, + "loss": 0.1196, + "step": 12862 + }, + { + "epoch": 5.39, + "learning_rate": 5.094754058503081e-06, + "loss": 0.1184, + "step": 12863 + }, + { + "epoch": 5.39, + "learning_rate": 5.093276703345351e-06, + "loss": 0.1329, + "step": 12864 + }, + { + "epoch": 5.39, + "learning_rate": 5.091799489227027e-06, + "loss": 0.112, + "step": 12865 + }, + { + "epoch": 5.39, + "learning_rate": 5.09032241619057e-06, + "loss": 0.13, + "step": 12866 + }, + { + "epoch": 5.39, + "learning_rate": 5.088845484278437e-06, + "loss": 0.1355, + "step": 12867 + }, + { + "epoch": 5.39, + "learning_rate": 5.087368693533088e-06, + "loss": 0.1024, + "step": 12868 + }, + { + "epoch": 5.39, + "learning_rate": 5.085892043996962e-06, + "loss": 0.1276, + "step": 12869 + }, + { + "epoch": 5.39, + "learning_rate": 5.084415535712508e-06, + "loss": 0.0943, + "step": 12870 + }, + { + "epoch": 5.39, + "learning_rate": 5.082939168722168e-06, + "loss": 0.1246, + "step": 12871 + }, + { + "epoch": 5.39, + "learning_rate": 5.081462943068378e-06, + "loss": 0.1012, + "step": 12872 + }, + { + "epoch": 5.39, + "learning_rate": 5.079986858793572e-06, + "loss": 0.1261, + "step": 12873 + }, + { + "epoch": 5.39, + "learning_rate": 5.078510915940178e-06, + "loss": 0.1254, + "step": 12874 + }, + { + "epoch": 5.39, + "learning_rate": 5.077035114550624e-06, + "loss": 0.1195, + "step": 12875 + }, + { + "epoch": 5.39, + "learning_rate": 5.075559454667321e-06, + "loss": 0.1738, + "step": 12876 + }, + { + "epoch": 5.39, + "learning_rate": 5.074083936332695e-06, + "loss": 0.132, + "step": 12877 + }, + { + "epoch": 5.39, + "learning_rate": 5.072608559589152e-06, + "loss": 0.1303, + "step": 12878 + }, + { + "epoch": 5.39, + "learning_rate": 5.071133324479105e-06, + "loss": 0.1372, + "step": 12879 + }, + { + "epoch": 5.39, + "learning_rate": 5.069658231044956e-06, + "loss": 0.1013, + "step": 12880 + }, + { + "epoch": 5.39, + "learning_rate": 5.0681832793291105e-06, + "loss": 0.0985, + "step": 12881 + }, + { + "epoch": 5.39, + "learning_rate": 5.066708469373958e-06, + "loss": 0.0946, + "step": 12882 + }, + { + "epoch": 5.39, + "learning_rate": 5.065233801221891e-06, + "loss": 0.1635, + "step": 12883 + }, + { + "epoch": 5.39, + "learning_rate": 5.063759274915299e-06, + "loss": 0.1048, + "step": 12884 + }, + { + "epoch": 5.39, + "learning_rate": 5.062284890496567e-06, + "loss": 0.1156, + "step": 12885 + }, + { + "epoch": 5.4, + "learning_rate": 5.060810648008078e-06, + "loss": 0.0879, + "step": 12886 + }, + { + "epoch": 5.4, + "learning_rate": 5.0593365474921995e-06, + "loss": 0.0964, + "step": 12887 + }, + { + "epoch": 5.4, + "learning_rate": 5.057862588991308e-06, + "loss": 0.1302, + "step": 12888 + }, + { + "epoch": 5.4, + "learning_rate": 5.056388772547775e-06, + "loss": 0.1282, + "step": 12889 + }, + { + "epoch": 5.4, + "learning_rate": 5.054915098203956e-06, + "loss": 0.086, + "step": 12890 + }, + { + "epoch": 5.4, + "learning_rate": 5.053441566002214e-06, + "loss": 0.1227, + "step": 12891 + }, + { + "epoch": 5.4, + "learning_rate": 5.051968175984905e-06, + "loss": 0.1448, + "step": 12892 + }, + { + "epoch": 5.4, + "learning_rate": 5.050494928194379e-06, + "loss": 0.1208, + "step": 12893 + }, + { + "epoch": 5.4, + "learning_rate": 5.0490218226729855e-06, + "loss": 0.1248, + "step": 12894 + }, + { + "epoch": 5.4, + "learning_rate": 5.047548859463069e-06, + "loss": 0.1448, + "step": 12895 + }, + { + "epoch": 5.4, + "learning_rate": 5.046076038606964e-06, + "loss": 0.1254, + "step": 12896 + }, + { + "epoch": 5.4, + "learning_rate": 5.0446033601470045e-06, + "loss": 0.1221, + "step": 12897 + }, + { + "epoch": 5.4, + "learning_rate": 5.043130824125526e-06, + "loss": 0.1134, + "step": 12898 + }, + { + "epoch": 5.4, + "learning_rate": 5.041658430584852e-06, + "loss": 0.1405, + "step": 12899 + }, + { + "epoch": 5.4, + "learning_rate": 5.040186179567308e-06, + "loss": 0.1081, + "step": 12900 + }, + { + "epoch": 5.4, + "learning_rate": 5.03871407111521e-06, + "loss": 0.137, + "step": 12901 + }, + { + "epoch": 5.4, + "learning_rate": 5.037242105270877e-06, + "loss": 0.151, + "step": 12902 + }, + { + "epoch": 5.4, + "learning_rate": 5.035770282076613e-06, + "loss": 0.1329, + "step": 12903 + }, + { + "epoch": 5.4, + "learning_rate": 5.034298601574726e-06, + "loss": 0.117, + "step": 12904 + }, + { + "epoch": 5.4, + "learning_rate": 5.03282706380752e-06, + "loss": 0.091, + "step": 12905 + }, + { + "epoch": 5.4, + "learning_rate": 5.031355668817292e-06, + "loss": 0.1177, + "step": 12906 + }, + { + "epoch": 5.4, + "learning_rate": 5.029884416646335e-06, + "loss": 0.1754, + "step": 12907 + }, + { + "epoch": 5.4, + "learning_rate": 5.0284133073369456e-06, + "loss": 0.122, + "step": 12908 + }, + { + "epoch": 5.4, + "learning_rate": 5.026942340931404e-06, + "loss": 0.1711, + "step": 12909 + }, + { + "epoch": 5.41, + "learning_rate": 5.025471517471987e-06, + "loss": 0.1217, + "step": 12910 + }, + { + "epoch": 5.41, + "learning_rate": 5.0240008370009775e-06, + "loss": 0.1192, + "step": 12911 + }, + { + "epoch": 5.41, + "learning_rate": 5.022530299560648e-06, + "loss": 0.1085, + "step": 12912 + }, + { + "epoch": 5.41, + "learning_rate": 5.021059905193268e-06, + "loss": 0.1179, + "step": 12913 + }, + { + "epoch": 5.41, + "learning_rate": 5.019589653941103e-06, + "loss": 0.1086, + "step": 12914 + }, + { + "epoch": 5.41, + "learning_rate": 5.018119545846414e-06, + "loss": 0.1005, + "step": 12915 + }, + { + "epoch": 5.41, + "learning_rate": 5.016649580951462e-06, + "loss": 0.0988, + "step": 12916 + }, + { + "epoch": 5.41, + "learning_rate": 5.015179759298492e-06, + "loss": 0.135, + "step": 12917 + }, + { + "epoch": 5.41, + "learning_rate": 5.013710080929754e-06, + "loss": 0.1036, + "step": 12918 + }, + { + "epoch": 5.41, + "learning_rate": 5.0122405458874975e-06, + "loss": 0.1485, + "step": 12919 + }, + { + "epoch": 5.41, + "learning_rate": 5.01077115421396e-06, + "loss": 0.0948, + "step": 12920 + }, + { + "epoch": 5.41, + "learning_rate": 5.009301905951378e-06, + "loss": 0.0954, + "step": 12921 + }, + { + "epoch": 5.41, + "learning_rate": 5.007832801141988e-06, + "loss": 0.1289, + "step": 12922 + }, + { + "epoch": 5.41, + "learning_rate": 5.00636383982801e-06, + "loss": 0.1359, + "step": 12923 + }, + { + "epoch": 5.41, + "learning_rate": 5.0048950220516725e-06, + "loss": 0.1199, + "step": 12924 + }, + { + "epoch": 5.41, + "learning_rate": 5.0034263478551935e-06, + "loss": 0.1447, + "step": 12925 + }, + { + "epoch": 5.41, + "learning_rate": 5.001957817280792e-06, + "loss": 0.1321, + "step": 12926 + }, + { + "epoch": 5.41, + "learning_rate": 5.0004894303706775e-06, + "loss": 0.116, + "step": 12927 + }, + { + "epoch": 5.41, + "learning_rate": 4.999021187167061e-06, + "loss": 0.085, + "step": 12928 + }, + { + "epoch": 5.41, + "learning_rate": 4.99755308771214e-06, + "loss": 0.1256, + "step": 12929 + }, + { + "epoch": 5.41, + "learning_rate": 4.996085132048116e-06, + "loss": 0.0945, + "step": 12930 + }, + { + "epoch": 5.41, + "learning_rate": 4.994617320217184e-06, + "loss": 0.1253, + "step": 12931 + }, + { + "epoch": 5.41, + "learning_rate": 4.99314965226154e-06, + "loss": 0.1201, + "step": 12932 + }, + { + "epoch": 5.41, + "learning_rate": 4.991682128223362e-06, + "loss": 0.1346, + "step": 12933 + }, + { + "epoch": 5.42, + "learning_rate": 4.990214748144837e-06, + "loss": 0.1149, + "step": 12934 + }, + { + "epoch": 5.42, + "learning_rate": 4.988747512068143e-06, + "loss": 0.1218, + "step": 12935 + }, + { + "epoch": 5.42, + "learning_rate": 4.9872804200354596e-06, + "loss": 0.0837, + "step": 12936 + }, + { + "epoch": 5.42, + "learning_rate": 4.985813472088947e-06, + "loss": 0.1447, + "step": 12937 + }, + { + "epoch": 5.42, + "learning_rate": 4.984346668270779e-06, + "loss": 0.0828, + "step": 12938 + }, + { + "epoch": 5.42, + "learning_rate": 4.9828800086231135e-06, + "loss": 0.1073, + "step": 12939 + }, + { + "epoch": 5.42, + "learning_rate": 4.98141349318811e-06, + "loss": 0.1124, + "step": 12940 + }, + { + "epoch": 5.42, + "learning_rate": 4.979947122007922e-06, + "loss": 0.115, + "step": 12941 + }, + { + "epoch": 5.42, + "learning_rate": 4.9784808951247e-06, + "loss": 0.0887, + "step": 12942 + }, + { + "epoch": 5.42, + "learning_rate": 4.977014812580593e-06, + "loss": 0.0724, + "step": 12943 + }, + { + "epoch": 5.42, + "learning_rate": 4.975548874417733e-06, + "loss": 0.126, + "step": 12944 + }, + { + "epoch": 5.42, + "learning_rate": 4.9740830806782625e-06, + "loss": 0.1083, + "step": 12945 + }, + { + "epoch": 5.42, + "learning_rate": 4.972617431404313e-06, + "loss": 0.0955, + "step": 12946 + }, + { + "epoch": 5.42, + "learning_rate": 4.971151926638016e-06, + "loss": 0.1092, + "step": 12947 + }, + { + "epoch": 5.42, + "learning_rate": 4.9696865664214936e-06, + "loss": 0.1395, + "step": 12948 + }, + { + "epoch": 5.42, + "learning_rate": 4.96822135079687e-06, + "loss": 0.1324, + "step": 12949 + }, + { + "epoch": 5.42, + "learning_rate": 4.966756279806255e-06, + "loss": 0.0848, + "step": 12950 + }, + { + "epoch": 5.42, + "learning_rate": 4.9652913534917645e-06, + "loss": 0.1021, + "step": 12951 + }, + { + "epoch": 5.42, + "learning_rate": 4.963826571895507e-06, + "loss": 0.11, + "step": 12952 + }, + { + "epoch": 5.42, + "learning_rate": 4.962361935059585e-06, + "loss": 0.1254, + "step": 12953 + }, + { + "epoch": 5.42, + "learning_rate": 4.9608974430261025e-06, + "loss": 0.1299, + "step": 12954 + }, + { + "epoch": 5.42, + "learning_rate": 4.959433095837149e-06, + "loss": 0.1142, + "step": 12955 + }, + { + "epoch": 5.42, + "learning_rate": 4.957968893534816e-06, + "loss": 0.1018, + "step": 12956 + }, + { + "epoch": 5.42, + "learning_rate": 4.956504836161198e-06, + "loss": 0.1073, + "step": 12957 + }, + { + "epoch": 5.43, + "learning_rate": 4.955040923758368e-06, + "loss": 0.1269, + "step": 12958 + }, + { + "epoch": 5.43, + "learning_rate": 4.953577156368412e-06, + "loss": 0.1611, + "step": 12959 + }, + { + "epoch": 5.43, + "learning_rate": 4.952113534033402e-06, + "loss": 0.1238, + "step": 12960 + }, + { + "epoch": 5.43, + "learning_rate": 4.950650056795407e-06, + "loss": 0.122, + "step": 12961 + }, + { + "epoch": 5.43, + "learning_rate": 4.949186724696498e-06, + "loss": 0.1164, + "step": 12962 + }, + { + "epoch": 5.43, + "learning_rate": 4.947723537778737e-06, + "loss": 0.1309, + "step": 12963 + }, + { + "epoch": 5.43, + "learning_rate": 4.946260496084175e-06, + "loss": 0.135, + "step": 12964 + }, + { + "epoch": 5.43, + "learning_rate": 4.9447975996548725e-06, + "loss": 0.1494, + "step": 12965 + }, + { + "epoch": 5.43, + "learning_rate": 4.943334848532876e-06, + "loss": 0.1116, + "step": 12966 + }, + { + "epoch": 5.43, + "learning_rate": 4.9418722427602315e-06, + "loss": 0.126, + "step": 12967 + }, + { + "epoch": 5.43, + "learning_rate": 4.940409782378982e-06, + "loss": 0.1028, + "step": 12968 + }, + { + "epoch": 5.43, + "learning_rate": 4.938947467431167e-06, + "loss": 0.1234, + "step": 12969 + }, + { + "epoch": 5.43, + "learning_rate": 4.937485297958812e-06, + "loss": 0.0987, + "step": 12970 + }, + { + "epoch": 5.43, + "learning_rate": 4.936023274003949e-06, + "loss": 0.1385, + "step": 12971 + }, + { + "epoch": 5.43, + "learning_rate": 4.934561395608603e-06, + "loss": 0.1048, + "step": 12972 + }, + { + "epoch": 5.43, + "learning_rate": 4.933099662814797e-06, + "loss": 0.1374, + "step": 12973 + }, + { + "epoch": 5.43, + "learning_rate": 4.931638075664544e-06, + "loss": 0.1106, + "step": 12974 + }, + { + "epoch": 5.43, + "learning_rate": 4.930176634199856e-06, + "loss": 0.166, + "step": 12975 + }, + { + "epoch": 5.43, + "learning_rate": 4.928715338462746e-06, + "loss": 0.1257, + "step": 12976 + }, + { + "epoch": 5.43, + "learning_rate": 4.927254188495213e-06, + "loss": 0.1514, + "step": 12977 + }, + { + "epoch": 5.43, + "learning_rate": 4.925793184339254e-06, + "loss": 0.1367, + "step": 12978 + }, + { + "epoch": 5.43, + "learning_rate": 4.924332326036867e-06, + "loss": 0.1215, + "step": 12979 + }, + { + "epoch": 5.43, + "learning_rate": 4.922871613630043e-06, + "loss": 0.0959, + "step": 12980 + }, + { + "epoch": 5.43, + "learning_rate": 4.921411047160768e-06, + "loss": 0.1108, + "step": 12981 + }, + { + "epoch": 5.44, + "learning_rate": 4.919950626671028e-06, + "loss": 0.0922, + "step": 12982 + }, + { + "epoch": 5.44, + "learning_rate": 4.918490352202797e-06, + "loss": 0.1245, + "step": 12983 + }, + { + "epoch": 5.44, + "learning_rate": 4.917030223798057e-06, + "loss": 0.1117, + "step": 12984 + }, + { + "epoch": 5.44, + "learning_rate": 4.9155702414987685e-06, + "loss": 0.098, + "step": 12985 + }, + { + "epoch": 5.44, + "learning_rate": 4.914110405346901e-06, + "loss": 0.0783, + "step": 12986 + }, + { + "epoch": 5.44, + "learning_rate": 4.912650715384417e-06, + "loss": 0.1356, + "step": 12987 + }, + { + "epoch": 5.44, + "learning_rate": 4.911191171653273e-06, + "loss": 0.0962, + "step": 12988 + }, + { + "epoch": 5.44, + "learning_rate": 4.909731774195425e-06, + "loss": 0.088, + "step": 12989 + }, + { + "epoch": 5.44, + "learning_rate": 4.908272523052823e-06, + "loss": 0.1381, + "step": 12990 + }, + { + "epoch": 5.44, + "learning_rate": 4.9068134182674044e-06, + "loss": 0.0844, + "step": 12991 + }, + { + "epoch": 5.44, + "learning_rate": 4.905354459881116e-06, + "loss": 0.1077, + "step": 12992 + }, + { + "epoch": 5.44, + "learning_rate": 4.9038956479358915e-06, + "loss": 0.1185, + "step": 12993 + }, + { + "epoch": 5.44, + "learning_rate": 4.9024369824736654e-06, + "loss": 0.1349, + "step": 12994 + }, + { + "epoch": 5.44, + "learning_rate": 4.900978463536363e-06, + "loss": 0.1354, + "step": 12995 + }, + { + "epoch": 5.44, + "learning_rate": 4.8995200911659155e-06, + "loss": 0.0975, + "step": 12996 + }, + { + "epoch": 5.44, + "learning_rate": 4.898061865404232e-06, + "loss": 0.1147, + "step": 12997 + }, + { + "epoch": 5.44, + "learning_rate": 4.896603786293233e-06, + "loss": 0.1106, + "step": 12998 + }, + { + "epoch": 5.44, + "learning_rate": 4.89514585387483e-06, + "loss": 0.0958, + "step": 12999 + }, + { + "epoch": 5.44, + "learning_rate": 4.893688068190933e-06, + "loss": 0.0883, + "step": 13000 + }, + { + "epoch": 5.44, + "learning_rate": 4.892230429283438e-06, + "loss": 0.0716, + "step": 13001 + }, + { + "epoch": 5.44, + "learning_rate": 4.890772937194246e-06, + "loss": 0.1229, + "step": 13002 + }, + { + "epoch": 5.44, + "learning_rate": 4.889315591965251e-06, + "loss": 0.1199, + "step": 13003 + }, + { + "epoch": 5.44, + "learning_rate": 4.88785839363835e-06, + "loss": 0.1117, + "step": 13004 + }, + { + "epoch": 5.44, + "learning_rate": 4.886401342255418e-06, + "loss": 0.1406, + "step": 13005 + }, + { + "epoch": 5.45, + "learning_rate": 4.884944437858342e-06, + "loss": 0.1029, + "step": 13006 + }, + { + "epoch": 5.45, + "learning_rate": 4.883487680489e-06, + "loss": 0.129, + "step": 13007 + }, + { + "epoch": 5.45, + "learning_rate": 4.882031070189262e-06, + "loss": 0.1123, + "step": 13008 + }, + { + "epoch": 5.45, + "learning_rate": 4.880574607001001e-06, + "loss": 0.1319, + "step": 13009 + }, + { + "epoch": 5.45, + "learning_rate": 4.879118290966085e-06, + "loss": 0.0958, + "step": 13010 + }, + { + "epoch": 5.45, + "learning_rate": 4.877662122126364e-06, + "loss": 0.1183, + "step": 13011 + }, + { + "epoch": 5.45, + "learning_rate": 4.876206100523701e-06, + "loss": 0.1054, + "step": 13012 + }, + { + "epoch": 5.45, + "learning_rate": 4.874750226199945e-06, + "loss": 0.0947, + "step": 13013 + }, + { + "epoch": 5.45, + "learning_rate": 4.873294499196946e-06, + "loss": 0.0976, + "step": 13014 + }, + { + "epoch": 5.45, + "learning_rate": 4.871838919556549e-06, + "loss": 0.1035, + "step": 13015 + }, + { + "epoch": 5.45, + "learning_rate": 4.87038348732059e-06, + "loss": 0.1318, + "step": 13016 + }, + { + "epoch": 5.45, + "learning_rate": 4.868928202530911e-06, + "loss": 0.1104, + "step": 13017 + }, + { + "epoch": 5.45, + "learning_rate": 4.867473065229332e-06, + "loss": 0.1334, + "step": 13018 + }, + { + "epoch": 5.45, + "learning_rate": 4.866018075457687e-06, + "loss": 0.0944, + "step": 13019 + }, + { + "epoch": 5.45, + "learning_rate": 4.864563233257795e-06, + "loss": 0.1089, + "step": 13020 + }, + { + "epoch": 5.45, + "learning_rate": 4.863108538671477e-06, + "loss": 0.1105, + "step": 13021 + }, + { + "epoch": 5.45, + "learning_rate": 4.861653991740549e-06, + "loss": 0.0989, + "step": 13022 + }, + { + "epoch": 5.45, + "learning_rate": 4.8601995925068136e-06, + "loss": 0.1159, + "step": 13023 + }, + { + "epoch": 5.45, + "learning_rate": 4.85874534101208e-06, + "loss": 0.1002, + "step": 13024 + }, + { + "epoch": 5.45, + "learning_rate": 4.8572912372981526e-06, + "loss": 0.1181, + "step": 13025 + }, + { + "epoch": 5.45, + "learning_rate": 4.855837281406821e-06, + "loss": 0.1238, + "step": 13026 + }, + { + "epoch": 5.45, + "learning_rate": 4.8543834733798826e-06, + "loss": 0.1285, + "step": 13027 + }, + { + "epoch": 5.45, + "learning_rate": 4.852929813259125e-06, + "loss": 0.1352, + "step": 13028 + }, + { + "epoch": 5.45, + "learning_rate": 4.851476301086333e-06, + "loss": 0.1426, + "step": 13029 + }, + { + "epoch": 5.46, + "learning_rate": 4.850022936903285e-06, + "loss": 0.1417, + "step": 13030 + }, + { + "epoch": 5.46, + "learning_rate": 4.848569720751762e-06, + "loss": 0.1086, + "step": 13031 + }, + { + "epoch": 5.46, + "learning_rate": 4.847116652673527e-06, + "loss": 0.128, + "step": 13032 + }, + { + "epoch": 5.46, + "learning_rate": 4.845663732710352e-06, + "loss": 0.1113, + "step": 13033 + }, + { + "epoch": 5.46, + "learning_rate": 4.844210960903998e-06, + "loss": 0.1092, + "step": 13034 + }, + { + "epoch": 5.46, + "learning_rate": 4.842758337296225e-06, + "loss": 0.0939, + "step": 13035 + }, + { + "epoch": 5.46, + "learning_rate": 4.841305861928787e-06, + "loss": 0.1224, + "step": 13036 + }, + { + "epoch": 5.46, + "learning_rate": 4.839853534843439e-06, + "loss": 0.131, + "step": 13037 + }, + { + "epoch": 5.46, + "learning_rate": 4.838401356081918e-06, + "loss": 0.1202, + "step": 13038 + }, + { + "epoch": 5.46, + "learning_rate": 4.83694932568597e-06, + "loss": 0.0958, + "step": 13039 + }, + { + "epoch": 5.46, + "learning_rate": 4.835497443697332e-06, + "loss": 0.1044, + "step": 13040 + }, + { + "epoch": 5.46, + "learning_rate": 4.834045710157735e-06, + "loss": 0.1014, + "step": 13041 + }, + { + "epoch": 5.46, + "learning_rate": 4.832594125108912e-06, + "loss": 0.1027, + "step": 13042 + }, + { + "epoch": 5.46, + "learning_rate": 4.831142688592585e-06, + "loss": 0.1258, + "step": 13043 + }, + { + "epoch": 5.46, + "learning_rate": 4.829691400650478e-06, + "loss": 0.1244, + "step": 13044 + }, + { + "epoch": 5.46, + "learning_rate": 4.828240261324304e-06, + "loss": 0.0685, + "step": 13045 + }, + { + "epoch": 5.46, + "learning_rate": 4.826789270655771e-06, + "loss": 0.1399, + "step": 13046 + }, + { + "epoch": 5.46, + "learning_rate": 4.825338428686589e-06, + "loss": 0.1374, + "step": 13047 + }, + { + "epoch": 5.46, + "learning_rate": 4.823887735458462e-06, + "loss": 0.1102, + "step": 13048 + }, + { + "epoch": 5.46, + "learning_rate": 4.82243719101309e-06, + "loss": 0.1116, + "step": 13049 + }, + { + "epoch": 5.46, + "learning_rate": 4.820986795392165e-06, + "loss": 0.1159, + "step": 13050 + }, + { + "epoch": 5.46, + "learning_rate": 4.819536548637383e-06, + "loss": 0.1404, + "step": 13051 + }, + { + "epoch": 5.46, + "learning_rate": 4.818086450790423e-06, + "loss": 0.1188, + "step": 13052 + }, + { + "epoch": 5.46, + "learning_rate": 4.816636501892968e-06, + "loss": 0.0973, + "step": 13053 + }, + { + "epoch": 5.47, + "learning_rate": 4.8151867019866985e-06, + "loss": 0.093, + "step": 13054 + }, + { + "epoch": 5.47, + "learning_rate": 4.813737051113286e-06, + "loss": 0.143, + "step": 13055 + }, + { + "epoch": 5.47, + "learning_rate": 4.812287549314399e-06, + "loss": 0.1674, + "step": 13056 + }, + { + "epoch": 5.47, + "learning_rate": 4.810838196631704e-06, + "loss": 0.1127, + "step": 13057 + }, + { + "epoch": 5.47, + "learning_rate": 4.809388993106864e-06, + "loss": 0.1456, + "step": 13058 + }, + { + "epoch": 5.47, + "learning_rate": 4.807939938781526e-06, + "loss": 0.1201, + "step": 13059 + }, + { + "epoch": 5.47, + "learning_rate": 4.806491033697348e-06, + "loss": 0.1025, + "step": 13060 + }, + { + "epoch": 5.47, + "learning_rate": 4.805042277895977e-06, + "loss": 0.1205, + "step": 13061 + }, + { + "epoch": 5.47, + "learning_rate": 4.803593671419056e-06, + "loss": 0.1222, + "step": 13062 + }, + { + "epoch": 5.47, + "learning_rate": 4.802145214308223e-06, + "loss": 0.1502, + "step": 13063 + }, + { + "epoch": 5.47, + "learning_rate": 4.800696906605117e-06, + "loss": 0.1467, + "step": 13064 + }, + { + "epoch": 5.47, + "learning_rate": 4.799248748351362e-06, + "loss": 0.1273, + "step": 13065 + }, + { + "epoch": 5.47, + "learning_rate": 4.797800739588585e-06, + "loss": 0.1242, + "step": 13066 + }, + { + "epoch": 5.47, + "learning_rate": 4.796352880358411e-06, + "loss": 0.1119, + "step": 13067 + }, + { + "epoch": 5.47, + "learning_rate": 4.794905170702458e-06, + "loss": 0.1283, + "step": 13068 + }, + { + "epoch": 5.47, + "learning_rate": 4.793457610662334e-06, + "loss": 0.1104, + "step": 13069 + }, + { + "epoch": 5.47, + "learning_rate": 4.792010200279651e-06, + "loss": 0.0905, + "step": 13070 + }, + { + "epoch": 5.47, + "learning_rate": 4.790562939596013e-06, + "loss": 0.0942, + "step": 13071 + }, + { + "epoch": 5.47, + "learning_rate": 4.789115828653025e-06, + "loss": 0.0938, + "step": 13072 + }, + { + "epoch": 5.47, + "learning_rate": 4.787668867492274e-06, + "loss": 0.112, + "step": 13073 + }, + { + "epoch": 5.47, + "learning_rate": 4.786222056155356e-06, + "loss": 0.1225, + "step": 13074 + }, + { + "epoch": 5.47, + "learning_rate": 4.784775394683859e-06, + "loss": 0.0973, + "step": 13075 + }, + { + "epoch": 5.47, + "learning_rate": 4.783328883119365e-06, + "loss": 0.1067, + "step": 13076 + }, + { + "epoch": 5.47, + "learning_rate": 4.781882521503452e-06, + "loss": 0.1221, + "step": 13077 + }, + { + "epoch": 5.48, + "learning_rate": 4.780436309877701e-06, + "loss": 0.0798, + "step": 13078 + }, + { + "epoch": 5.48, + "learning_rate": 4.778990248283671e-06, + "loss": 0.1351, + "step": 13079 + }, + { + "epoch": 5.48, + "learning_rate": 4.777544336762935e-06, + "loss": 0.1175, + "step": 13080 + }, + { + "epoch": 5.48, + "learning_rate": 4.776098575357052e-06, + "loss": 0.1549, + "step": 13081 + }, + { + "epoch": 5.48, + "learning_rate": 4.7746529641075785e-06, + "loss": 0.1547, + "step": 13082 + }, + { + "epoch": 5.48, + "learning_rate": 4.773207503056071e-06, + "loss": 0.0962, + "step": 13083 + }, + { + "epoch": 5.48, + "learning_rate": 4.771762192244075e-06, + "loss": 0.0888, + "step": 13084 + }, + { + "epoch": 5.48, + "learning_rate": 4.770317031713138e-06, + "loss": 0.1445, + "step": 13085 + }, + { + "epoch": 5.48, + "learning_rate": 4.768872021504795e-06, + "loss": 0.156, + "step": 13086 + }, + { + "epoch": 5.48, + "learning_rate": 4.7674271616605816e-06, + "loss": 0.1104, + "step": 13087 + }, + { + "epoch": 5.48, + "learning_rate": 4.765982452222032e-06, + "loss": 0.1365, + "step": 13088 + }, + { + "epoch": 5.48, + "learning_rate": 4.764537893230673e-06, + "loss": 0.1265, + "step": 13089 + }, + { + "epoch": 5.48, + "learning_rate": 4.76309348472803e-06, + "loss": 0.1443, + "step": 13090 + }, + { + "epoch": 5.48, + "learning_rate": 4.7616492267556134e-06, + "loss": 0.0837, + "step": 13091 + }, + { + "epoch": 5.48, + "learning_rate": 4.760205119354946e-06, + "loss": 0.1026, + "step": 13092 + }, + { + "epoch": 5.48, + "learning_rate": 4.758761162567528e-06, + "loss": 0.141, + "step": 13093 + }, + { + "epoch": 5.48, + "learning_rate": 4.757317356434869e-06, + "loss": 0.1503, + "step": 13094 + }, + { + "epoch": 5.48, + "learning_rate": 4.755873700998471e-06, + "loss": 0.1125, + "step": 13095 + }, + { + "epoch": 5.48, + "learning_rate": 4.75443019629983e-06, + "loss": 0.1134, + "step": 13096 + }, + { + "epoch": 5.48, + "learning_rate": 4.7529868423804385e-06, + "loss": 0.1306, + "step": 13097 + }, + { + "epoch": 5.48, + "learning_rate": 4.751543639281784e-06, + "loss": 0.1281, + "step": 13098 + }, + { + "epoch": 5.48, + "learning_rate": 4.750100587045353e-06, + "loss": 0.1129, + "step": 13099 + }, + { + "epoch": 5.48, + "learning_rate": 4.748657685712619e-06, + "loss": 0.11, + "step": 13100 + }, + { + "epoch": 5.49, + "learning_rate": 4.74721493532506e-06, + "loss": 0.1665, + "step": 13101 + }, + { + "epoch": 5.49, + "learning_rate": 4.745772335924146e-06, + "loss": 0.115, + "step": 13102 + }, + { + "epoch": 5.49, + "learning_rate": 4.744329887551343e-06, + "loss": 0.1227, + "step": 13103 + }, + { + "epoch": 5.49, + "learning_rate": 4.742887590248114e-06, + "loss": 0.1139, + "step": 13104 + }, + { + "epoch": 5.49, + "learning_rate": 4.741445444055921e-06, + "loss": 0.1154, + "step": 13105 + }, + { + "epoch": 5.49, + "learning_rate": 4.740003449016208e-06, + "loss": 0.0991, + "step": 13106 + }, + { + "epoch": 5.49, + "learning_rate": 4.7385616051704275e-06, + "loss": 0.1196, + "step": 13107 + }, + { + "epoch": 5.49, + "learning_rate": 4.737119912560025e-06, + "loss": 0.11, + "step": 13108 + }, + { + "epoch": 5.49, + "learning_rate": 4.7356783712264405e-06, + "loss": 0.117, + "step": 13109 + }, + { + "epoch": 5.49, + "learning_rate": 4.734236981211111e-06, + "loss": 0.0986, + "step": 13110 + }, + { + "epoch": 5.49, + "learning_rate": 4.73279574255547e-06, + "loss": 0.1196, + "step": 13111 + }, + { + "epoch": 5.49, + "learning_rate": 4.731354655300936e-06, + "loss": 0.0981, + "step": 13112 + }, + { + "epoch": 5.49, + "learning_rate": 4.729913719488942e-06, + "loss": 0.0829, + "step": 13113 + }, + { + "epoch": 5.49, + "learning_rate": 4.7284729351608985e-06, + "loss": 0.1425, + "step": 13114 + }, + { + "epoch": 5.49, + "learning_rate": 4.727032302358221e-06, + "loss": 0.1211, + "step": 13115 + }, + { + "epoch": 5.49, + "learning_rate": 4.725591821122322e-06, + "loss": 0.0995, + "step": 13116 + }, + { + "epoch": 5.49, + "learning_rate": 4.724151491494605e-06, + "loss": 0.1103, + "step": 13117 + }, + { + "epoch": 5.49, + "learning_rate": 4.722711313516473e-06, + "loss": 0.1047, + "step": 13118 + }, + { + "epoch": 5.49, + "learning_rate": 4.721271287229323e-06, + "loss": 0.1315, + "step": 13119 + }, + { + "epoch": 5.49, + "learning_rate": 4.7198314126745424e-06, + "loss": 0.1178, + "step": 13120 + }, + { + "epoch": 5.49, + "learning_rate": 4.7183916898935234e-06, + "loss": 0.1067, + "step": 13121 + }, + { + "epoch": 5.49, + "learning_rate": 4.716952118927648e-06, + "loss": 0.1286, + "step": 13122 + }, + { + "epoch": 5.49, + "learning_rate": 4.715512699818295e-06, + "loss": 0.103, + "step": 13123 + }, + { + "epoch": 5.49, + "learning_rate": 4.714073432606841e-06, + "loss": 0.1046, + "step": 13124 + }, + { + "epoch": 5.5, + "learning_rate": 4.712634317334655e-06, + "loss": 0.151, + "step": 13125 + }, + { + "epoch": 5.5, + "learning_rate": 4.711195354043109e-06, + "loss": 0.1192, + "step": 13126 + }, + { + "epoch": 5.5, + "learning_rate": 4.709756542773554e-06, + "loss": 0.125, + "step": 13127 + }, + { + "epoch": 5.5, + "learning_rate": 4.708317883567353e-06, + "loss": 0.1561, + "step": 13128 + }, + { + "epoch": 5.5, + "learning_rate": 4.706879376465858e-06, + "loss": 0.1347, + "step": 13129 + }, + { + "epoch": 5.5, + "learning_rate": 4.70544102151042e-06, + "loss": 0.1131, + "step": 13130 + }, + { + "epoch": 5.5, + "learning_rate": 4.7040028187423795e-06, + "loss": 0.0858, + "step": 13131 + }, + { + "epoch": 5.5, + "learning_rate": 4.702564768203083e-06, + "loss": 0.1196, + "step": 13132 + }, + { + "epoch": 5.5, + "learning_rate": 4.701126869933858e-06, + "loss": 0.1426, + "step": 13133 + }, + { + "epoch": 5.5, + "learning_rate": 4.699689123976039e-06, + "loss": 0.1101, + "step": 13134 + }, + { + "epoch": 5.5, + "learning_rate": 4.698251530370956e-06, + "loss": 0.1094, + "step": 13135 + }, + { + "epoch": 5.5, + "learning_rate": 4.696814089159924e-06, + "loss": 0.1387, + "step": 13136 + }, + { + "epoch": 5.5, + "learning_rate": 4.695376800384266e-06, + "loss": 0.1346, + "step": 13137 + }, + { + "epoch": 5.5, + "learning_rate": 4.6939396640852945e-06, + "loss": 0.1599, + "step": 13138 + }, + { + "epoch": 5.5, + "learning_rate": 4.6925026803043186e-06, + "loss": 0.1209, + "step": 13139 + }, + { + "epoch": 5.5, + "learning_rate": 4.691065849082647e-06, + "loss": 0.1277, + "step": 13140 + }, + { + "epoch": 5.5, + "learning_rate": 4.6896291704615725e-06, + "loss": 0.1685, + "step": 13141 + }, + { + "epoch": 5.5, + "learning_rate": 4.688192644482396e-06, + "loss": 0.1097, + "step": 13142 + }, + { + "epoch": 5.5, + "learning_rate": 4.686756271186407e-06, + "loss": 0.139, + "step": 13143 + }, + { + "epoch": 5.5, + "learning_rate": 4.6853200506148955e-06, + "loss": 0.1138, + "step": 13144 + }, + { + "epoch": 5.5, + "learning_rate": 4.683883982809142e-06, + "loss": 0.1214, + "step": 13145 + }, + { + "epoch": 5.5, + "learning_rate": 4.68244806781043e-06, + "loss": 0.1228, + "step": 13146 + }, + { + "epoch": 5.5, + "learning_rate": 4.681012305660026e-06, + "loss": 0.097, + "step": 13147 + }, + { + "epoch": 5.5, + "learning_rate": 4.679576696399203e-06, + "loss": 0.1354, + "step": 13148 + }, + { + "epoch": 5.51, + "learning_rate": 4.678141240069227e-06, + "loss": 0.1367, + "step": 13149 + }, + { + "epoch": 5.51, + "learning_rate": 4.676705936711358e-06, + "loss": 0.0956, + "step": 13150 + }, + { + "epoch": 5.51, + "learning_rate": 4.675270786366853e-06, + "loss": 0.1226, + "step": 13151 + }, + { + "epoch": 5.51, + "learning_rate": 4.673835789076969e-06, + "loss": 0.1573, + "step": 13152 + }, + { + "epoch": 5.51, + "learning_rate": 4.672400944882944e-06, + "loss": 0.1212, + "step": 13153 + }, + { + "epoch": 5.51, + "learning_rate": 4.670966253826027e-06, + "loss": 0.1723, + "step": 13154 + }, + { + "epoch": 5.51, + "learning_rate": 4.669531715947455e-06, + "loss": 0.1733, + "step": 13155 + }, + { + "epoch": 5.51, + "learning_rate": 4.668097331288465e-06, + "loss": 0.1056, + "step": 13156 + }, + { + "epoch": 5.51, + "learning_rate": 4.666663099890285e-06, + "loss": 0.088, + "step": 13157 + }, + { + "epoch": 5.51, + "learning_rate": 4.665229021794145e-06, + "loss": 0.1706, + "step": 13158 + }, + { + "epoch": 5.51, + "learning_rate": 4.66379509704126e-06, + "loss": 0.0982, + "step": 13159 + }, + { + "epoch": 5.51, + "learning_rate": 4.662361325672853e-06, + "loss": 0.1061, + "step": 13160 + }, + { + "epoch": 5.51, + "learning_rate": 4.660927707730129e-06, + "loss": 0.0863, + "step": 13161 + }, + { + "epoch": 5.51, + "learning_rate": 4.659494243254301e-06, + "loss": 0.1444, + "step": 13162 + }, + { + "epoch": 5.51, + "learning_rate": 4.658060932286571e-06, + "loss": 0.0896, + "step": 13163 + }, + { + "epoch": 5.51, + "learning_rate": 4.65662777486814e-06, + "loss": 0.1124, + "step": 13164 + }, + { + "epoch": 5.51, + "learning_rate": 4.6551947710402016e-06, + "loss": 0.13, + "step": 13165 + }, + { + "epoch": 5.51, + "learning_rate": 4.653761920843946e-06, + "loss": 0.1145, + "step": 13166 + }, + { + "epoch": 5.51, + "learning_rate": 4.652329224320564e-06, + "loss": 0.1004, + "step": 13167 + }, + { + "epoch": 5.51, + "learning_rate": 4.65089668151123e-06, + "loss": 0.0881, + "step": 13168 + }, + { + "epoch": 5.51, + "learning_rate": 4.649464292457123e-06, + "loss": 0.1377, + "step": 13169 + }, + { + "epoch": 5.51, + "learning_rate": 4.648032057199418e-06, + "loss": 0.107, + "step": 13170 + }, + { + "epoch": 5.51, + "learning_rate": 4.6465999757792815e-06, + "loss": 0.1474, + "step": 13171 + }, + { + "epoch": 5.51, + "learning_rate": 4.645168048237878e-06, + "loss": 0.145, + "step": 13172 + }, + { + "epoch": 5.52, + "learning_rate": 4.643736274616372e-06, + "loss": 0.1343, + "step": 13173 + }, + { + "epoch": 5.52, + "learning_rate": 4.6423046549559095e-06, + "loss": 0.1351, + "step": 13174 + }, + { + "epoch": 5.52, + "learning_rate": 4.640873189297645e-06, + "loss": 0.1167, + "step": 13175 + }, + { + "epoch": 5.52, + "learning_rate": 4.639441877682724e-06, + "loss": 0.1504, + "step": 13176 + }, + { + "epoch": 5.52, + "learning_rate": 4.638010720152292e-06, + "loss": 0.1185, + "step": 13177 + }, + { + "epoch": 5.52, + "learning_rate": 4.6365797167474815e-06, + "loss": 0.1556, + "step": 13178 + }, + { + "epoch": 5.52, + "learning_rate": 4.635148867509432e-06, + "loss": 0.1527, + "step": 13179 + }, + { + "epoch": 5.52, + "learning_rate": 4.6337181724792644e-06, + "loss": 0.1212, + "step": 13180 + }, + { + "epoch": 5.52, + "learning_rate": 4.6322876316981095e-06, + "loss": 0.1378, + "step": 13181 + }, + { + "epoch": 5.52, + "learning_rate": 4.630857245207079e-06, + "loss": 0.1086, + "step": 13182 + }, + { + "epoch": 5.52, + "learning_rate": 4.629427013047293e-06, + "loss": 0.1517, + "step": 13183 + }, + { + "epoch": 5.52, + "learning_rate": 4.627996935259862e-06, + "loss": 0.1087, + "step": 13184 + }, + { + "epoch": 5.52, + "learning_rate": 4.626567011885892e-06, + "loss": 0.1032, + "step": 13185 + }, + { + "epoch": 5.52, + "learning_rate": 4.625137242966484e-06, + "loss": 0.0961, + "step": 13186 + }, + { + "epoch": 5.52, + "learning_rate": 4.623707628542742e-06, + "loss": 0.1232, + "step": 13187 + }, + { + "epoch": 5.52, + "learning_rate": 4.6222781686557485e-06, + "loss": 0.1486, + "step": 13188 + }, + { + "epoch": 5.52, + "learning_rate": 4.620848863346596e-06, + "loss": 0.157, + "step": 13189 + }, + { + "epoch": 5.52, + "learning_rate": 4.6194197126563715e-06, + "loss": 0.1104, + "step": 13190 + }, + { + "epoch": 5.52, + "learning_rate": 4.617990716626151e-06, + "loss": 0.0985, + "step": 13191 + }, + { + "epoch": 5.52, + "learning_rate": 4.616561875297012e-06, + "loss": 0.0959, + "step": 13192 + }, + { + "epoch": 5.52, + "learning_rate": 4.615133188710029e-06, + "loss": 0.0781, + "step": 13193 + }, + { + "epoch": 5.52, + "learning_rate": 4.6137046569062595e-06, + "loss": 0.1041, + "step": 13194 + }, + { + "epoch": 5.52, + "learning_rate": 4.612276279926771e-06, + "loss": 0.1304, + "step": 13195 + }, + { + "epoch": 5.52, + "learning_rate": 4.610848057812619e-06, + "loss": 0.1292, + "step": 13196 + }, + { + "epoch": 5.53, + "learning_rate": 4.609419990604858e-06, + "loss": 0.1582, + "step": 13197 + }, + { + "epoch": 5.53, + "learning_rate": 4.607992078344534e-06, + "loss": 0.1267, + "step": 13198 + }, + { + "epoch": 5.53, + "learning_rate": 4.6065643210726955e-06, + "loss": 0.0792, + "step": 13199 + }, + { + "epoch": 5.53, + "learning_rate": 4.605136718830382e-06, + "loss": 0.1392, + "step": 13200 + }, + { + "epoch": 5.53, + "learning_rate": 4.603709271658621e-06, + "loss": 0.1323, + "step": 13201 + }, + { + "epoch": 5.53, + "learning_rate": 4.602281979598451e-06, + "loss": 0.1288, + "step": 13202 + }, + { + "epoch": 5.53, + "learning_rate": 4.600854842690898e-06, + "loss": 0.1464, + "step": 13203 + }, + { + "epoch": 5.53, + "learning_rate": 4.599427860976978e-06, + "loss": 0.092, + "step": 13204 + }, + { + "epoch": 5.53, + "learning_rate": 4.598001034497712e-06, + "loss": 0.1516, + "step": 13205 + }, + { + "epoch": 5.53, + "learning_rate": 4.596574363294113e-06, + "loss": 0.1023, + "step": 13206 + }, + { + "epoch": 5.53, + "learning_rate": 4.595147847407187e-06, + "loss": 0.173, + "step": 13207 + }, + { + "epoch": 5.53, + "learning_rate": 4.593721486877945e-06, + "loss": 0.1068, + "step": 13208 + }, + { + "epoch": 5.53, + "learning_rate": 4.592295281747379e-06, + "loss": 0.1424, + "step": 13209 + }, + { + "epoch": 5.53, + "learning_rate": 4.590869232056485e-06, + "loss": 0.1042, + "step": 13210 + }, + { + "epoch": 5.53, + "learning_rate": 4.589443337846256e-06, + "loss": 0.1413, + "step": 13211 + }, + { + "epoch": 5.53, + "learning_rate": 4.588017599157675e-06, + "loss": 0.1487, + "step": 13212 + }, + { + "epoch": 5.53, + "learning_rate": 4.5865920160317266e-06, + "loss": 0.1182, + "step": 13213 + }, + { + "epoch": 5.53, + "learning_rate": 4.585166588509391e-06, + "loss": 0.1448, + "step": 13214 + }, + { + "epoch": 5.53, + "learning_rate": 4.5837413166316335e-06, + "loss": 0.1197, + "step": 13215 + }, + { + "epoch": 5.53, + "learning_rate": 4.582316200439425e-06, + "loss": 0.1316, + "step": 13216 + }, + { + "epoch": 5.53, + "learning_rate": 4.580891239973729e-06, + "loss": 0.1344, + "step": 13217 + }, + { + "epoch": 5.53, + "learning_rate": 4.579466435275506e-06, + "loss": 0.1287, + "step": 13218 + }, + { + "epoch": 5.53, + "learning_rate": 4.578041786385711e-06, + "loss": 0.1488, + "step": 13219 + }, + { + "epoch": 5.53, + "learning_rate": 4.576617293345296e-06, + "loss": 0.1474, + "step": 13220 + }, + { + "epoch": 5.54, + "learning_rate": 4.575192956195201e-06, + "loss": 0.1238, + "step": 13221 + }, + { + "epoch": 5.54, + "learning_rate": 4.573768774976371e-06, + "loss": 0.1045, + "step": 13222 + }, + { + "epoch": 5.54, + "learning_rate": 4.5723447497297424e-06, + "loss": 0.1188, + "step": 13223 + }, + { + "epoch": 5.54, + "learning_rate": 4.570920880496247e-06, + "loss": 0.1278, + "step": 13224 + }, + { + "epoch": 5.54, + "learning_rate": 4.569497167316813e-06, + "loss": 0.1062, + "step": 13225 + }, + { + "epoch": 5.54, + "learning_rate": 4.568073610232368e-06, + "loss": 0.1161, + "step": 13226 + }, + { + "epoch": 5.54, + "learning_rate": 4.566650209283824e-06, + "loss": 0.1178, + "step": 13227 + }, + { + "epoch": 5.54, + "learning_rate": 4.565226964512102e-06, + "loss": 0.1045, + "step": 13228 + }, + { + "epoch": 5.54, + "learning_rate": 4.563803875958103e-06, + "loss": 0.112, + "step": 13229 + }, + { + "epoch": 5.54, + "learning_rate": 4.562380943662739e-06, + "loss": 0.1019, + "step": 13230 + }, + { + "epoch": 5.54, + "learning_rate": 4.560958167666908e-06, + "loss": 0.1122, + "step": 13231 + }, + { + "epoch": 5.54, + "learning_rate": 4.5595355480115086e-06, + "loss": 0.1071, + "step": 13232 + }, + { + "epoch": 5.54, + "learning_rate": 4.558113084737434e-06, + "loss": 0.1029, + "step": 13233 + }, + { + "epoch": 5.54, + "learning_rate": 4.556690777885571e-06, + "loss": 0.1232, + "step": 13234 + }, + { + "epoch": 5.54, + "learning_rate": 4.5552686274967984e-06, + "loss": 0.1612, + "step": 13235 + }, + { + "epoch": 5.54, + "learning_rate": 4.553846633611997e-06, + "loss": 0.1484, + "step": 13236 + }, + { + "epoch": 5.54, + "learning_rate": 4.552424796272041e-06, + "loss": 0.072, + "step": 13237 + }, + { + "epoch": 5.54, + "learning_rate": 4.551003115517801e-06, + "loss": 0.1069, + "step": 13238 + }, + { + "epoch": 5.54, + "learning_rate": 4.549581591390142e-06, + "loss": 0.1268, + "step": 13239 + }, + { + "epoch": 5.54, + "learning_rate": 4.548160223929923e-06, + "loss": 0.099, + "step": 13240 + }, + { + "epoch": 5.54, + "learning_rate": 4.546739013178003e-06, + "loss": 0.1366, + "step": 13241 + }, + { + "epoch": 5.54, + "learning_rate": 4.545317959175228e-06, + "loss": 0.1154, + "step": 13242 + }, + { + "epoch": 5.54, + "learning_rate": 4.543897061962448e-06, + "loss": 0.1007, + "step": 13243 + }, + { + "epoch": 5.54, + "learning_rate": 4.542476321580505e-06, + "loss": 0.1371, + "step": 13244 + }, + { + "epoch": 5.55, + "learning_rate": 4.541055738070237e-06, + "loss": 0.1588, + "step": 13245 + }, + { + "epoch": 5.55, + "learning_rate": 4.539635311472478e-06, + "loss": 0.154, + "step": 13246 + }, + { + "epoch": 5.55, + "learning_rate": 4.53821504182806e-06, + "loss": 0.0952, + "step": 13247 + }, + { + "epoch": 5.55, + "learning_rate": 4.5367949291778e-06, + "loss": 0.1177, + "step": 13248 + }, + { + "epoch": 5.55, + "learning_rate": 4.535374973562526e-06, + "loss": 0.1216, + "step": 13249 + }, + { + "epoch": 5.55, + "learning_rate": 4.533955175023046e-06, + "loss": 0.1066, + "step": 13250 + }, + { + "epoch": 5.55, + "learning_rate": 4.532535533600173e-06, + "loss": 0.0986, + "step": 13251 + }, + { + "epoch": 5.55, + "learning_rate": 4.531116049334715e-06, + "loss": 0.1444, + "step": 13252 + }, + { + "epoch": 5.55, + "learning_rate": 4.5296967222674724e-06, + "loss": 0.11, + "step": 13253 + }, + { + "epoch": 5.55, + "learning_rate": 4.528277552439244e-06, + "loss": 0.1187, + "step": 13254 + }, + { + "epoch": 5.55, + "learning_rate": 4.526858539890825e-06, + "loss": 0.1006, + "step": 13255 + }, + { + "epoch": 5.55, + "learning_rate": 4.5254396846629975e-06, + "loss": 0.086, + "step": 13256 + }, + { + "epoch": 5.55, + "learning_rate": 4.524020986796548e-06, + "loss": 0.1032, + "step": 13257 + }, + { + "epoch": 5.55, + "learning_rate": 4.522602446332256e-06, + "loss": 0.1266, + "step": 13258 + }, + { + "epoch": 5.55, + "learning_rate": 4.521184063310895e-06, + "loss": 0.1116, + "step": 13259 + }, + { + "epoch": 5.55, + "learning_rate": 4.519765837773238e-06, + "loss": 0.1249, + "step": 13260 + }, + { + "epoch": 5.55, + "learning_rate": 4.518347769760051e-06, + "loss": 0.0959, + "step": 13261 + }, + { + "epoch": 5.55, + "learning_rate": 4.51692985931209e-06, + "loss": 0.1262, + "step": 13262 + }, + { + "epoch": 5.55, + "learning_rate": 4.515512106470115e-06, + "loss": 0.1021, + "step": 13263 + }, + { + "epoch": 5.55, + "learning_rate": 4.514094511274877e-06, + "loss": 0.1451, + "step": 13264 + }, + { + "epoch": 5.55, + "learning_rate": 4.512677073767124e-06, + "loss": 0.1179, + "step": 13265 + }, + { + "epoch": 5.55, + "learning_rate": 4.5112597939876e-06, + "loss": 0.1279, + "step": 13266 + }, + { + "epoch": 5.55, + "learning_rate": 4.509842671977043e-06, + "loss": 0.0967, + "step": 13267 + }, + { + "epoch": 5.55, + "learning_rate": 4.50842570777619e-06, + "loss": 0.0605, + "step": 13268 + }, + { + "epoch": 5.56, + "learning_rate": 4.507008901425762e-06, + "loss": 0.0834, + "step": 13269 + }, + { + "epoch": 5.56, + "learning_rate": 4.50559225296649e-06, + "loss": 0.1333, + "step": 13270 + }, + { + "epoch": 5.56, + "learning_rate": 4.504175762439096e-06, + "loss": 0.1459, + "step": 13271 + }, + { + "epoch": 5.56, + "learning_rate": 4.5027594298842884e-06, + "loss": 0.1377, + "step": 13272 + }, + { + "epoch": 5.56, + "learning_rate": 4.5013432553427835e-06, + "loss": 0.0817, + "step": 13273 + }, + { + "epoch": 5.56, + "learning_rate": 4.499927238855287e-06, + "loss": 0.1335, + "step": 13274 + }, + { + "epoch": 5.56, + "learning_rate": 4.498511380462506e-06, + "loss": 0.1125, + "step": 13275 + }, + { + "epoch": 5.56, + "learning_rate": 4.4970956802051295e-06, + "loss": 0.1297, + "step": 13276 + }, + { + "epoch": 5.56, + "learning_rate": 4.495680138123855e-06, + "loss": 0.1283, + "step": 13277 + }, + { + "epoch": 5.56, + "learning_rate": 4.494264754259368e-06, + "loss": 0.1291, + "step": 13278 + }, + { + "epoch": 5.56, + "learning_rate": 4.492849528652357e-06, + "loss": 0.1326, + "step": 13279 + }, + { + "epoch": 5.56, + "learning_rate": 4.491434461343499e-06, + "loss": 0.109, + "step": 13280 + }, + { + "epoch": 5.56, + "learning_rate": 4.49001955237347e-06, + "loss": 0.1538, + "step": 13281 + }, + { + "epoch": 5.56, + "learning_rate": 4.488604801782942e-06, + "loss": 0.1198, + "step": 13282 + }, + { + "epoch": 5.56, + "learning_rate": 4.4871902096125755e-06, + "loss": 0.1009, + "step": 13283 + }, + { + "epoch": 5.56, + "learning_rate": 4.485775775903033e-06, + "loss": 0.1616, + "step": 13284 + }, + { + "epoch": 5.56, + "learning_rate": 4.484361500694974e-06, + "loss": 0.0977, + "step": 13285 + }, + { + "epoch": 5.56, + "learning_rate": 4.482947384029048e-06, + "loss": 0.1267, + "step": 13286 + }, + { + "epoch": 5.56, + "learning_rate": 4.481533425945904e-06, + "loss": 0.153, + "step": 13287 + }, + { + "epoch": 5.56, + "learning_rate": 4.480119626486189e-06, + "loss": 0.143, + "step": 13288 + }, + { + "epoch": 5.56, + "learning_rate": 4.4787059856905334e-06, + "loss": 0.0814, + "step": 13289 + }, + { + "epoch": 5.56, + "learning_rate": 4.477292503599574e-06, + "loss": 0.1323, + "step": 13290 + }, + { + "epoch": 5.56, + "learning_rate": 4.475879180253941e-06, + "loss": 0.1213, + "step": 13291 + }, + { + "epoch": 5.56, + "learning_rate": 4.47446601569426e-06, + "loss": 0.1197, + "step": 13292 + }, + { + "epoch": 5.57, + "learning_rate": 4.473053009961149e-06, + "loss": 0.1439, + "step": 13293 + }, + { + "epoch": 5.57, + "learning_rate": 4.471640163095229e-06, + "loss": 0.147, + "step": 13294 + }, + { + "epoch": 5.57, + "learning_rate": 4.470227475137102e-06, + "loss": 0.0922, + "step": 13295 + }, + { + "epoch": 5.57, + "learning_rate": 4.4688149461273845e-06, + "loss": 0.1109, + "step": 13296 + }, + { + "epoch": 5.57, + "learning_rate": 4.467402576106668e-06, + "loss": 0.1183, + "step": 13297 + }, + { + "epoch": 5.57, + "learning_rate": 4.465990365115555e-06, + "loss": 0.0781, + "step": 13298 + }, + { + "epoch": 5.57, + "learning_rate": 4.46457831319464e-06, + "loss": 0.1439, + "step": 13299 + }, + { + "epoch": 5.57, + "learning_rate": 4.463166420384507e-06, + "loss": 0.1276, + "step": 13300 + }, + { + "epoch": 5.57, + "learning_rate": 4.461754686725743e-06, + "loss": 0.1152, + "step": 13301 + }, + { + "epoch": 5.57, + "learning_rate": 4.4603431122589295e-06, + "loss": 0.1117, + "step": 13302 + }, + { + "epoch": 5.57, + "learning_rate": 4.458931697024633e-06, + "loss": 0.1098, + "step": 13303 + }, + { + "epoch": 5.57, + "learning_rate": 4.457520441063428e-06, + "loss": 0.1016, + "step": 13304 + }, + { + "epoch": 5.57, + "learning_rate": 4.456109344415881e-06, + "loss": 0.1145, + "step": 13305 + }, + { + "epoch": 5.57, + "learning_rate": 4.454698407122551e-06, + "loss": 0.1367, + "step": 13306 + }, + { + "epoch": 5.57, + "learning_rate": 4.453287629223994e-06, + "loss": 0.1371, + "step": 13307 + }, + { + "epoch": 5.57, + "learning_rate": 4.4518770107607626e-06, + "loss": 0.0942, + "step": 13308 + }, + { + "epoch": 5.57, + "learning_rate": 4.450466551773406e-06, + "loss": 0.114, + "step": 13309 + }, + { + "epoch": 5.57, + "learning_rate": 4.449056252302461e-06, + "loss": 0.1576, + "step": 13310 + }, + { + "epoch": 5.57, + "learning_rate": 4.447646112388467e-06, + "loss": 0.1172, + "step": 13311 + }, + { + "epoch": 5.57, + "learning_rate": 4.44623613207196e-06, + "loss": 0.1272, + "step": 13312 + }, + { + "epoch": 5.57, + "learning_rate": 4.444826311393465e-06, + "loss": 0.1114, + "step": 13313 + }, + { + "epoch": 5.57, + "learning_rate": 4.443416650393509e-06, + "loss": 0.112, + "step": 13314 + }, + { + "epoch": 5.57, + "learning_rate": 4.442007149112615e-06, + "loss": 0.1217, + "step": 13315 + }, + { + "epoch": 5.58, + "learning_rate": 4.44059780759129e-06, + "loss": 0.1172, + "step": 13316 + }, + { + "epoch": 5.58, + "learning_rate": 4.43918862587005e-06, + "loss": 0.1354, + "step": 13317 + }, + { + "epoch": 5.58, + "learning_rate": 4.437779603989395e-06, + "loss": 0.1031, + "step": 13318 + }, + { + "epoch": 5.58, + "learning_rate": 4.436370741989829e-06, + "loss": 0.154, + "step": 13319 + }, + { + "epoch": 5.58, + "learning_rate": 4.4349620399118495e-06, + "loss": 0.1524, + "step": 13320 + }, + { + "epoch": 5.58, + "learning_rate": 4.43355349779595e-06, + "loss": 0.0977, + "step": 13321 + }, + { + "epoch": 5.58, + "learning_rate": 4.432145115682613e-06, + "loss": 0.1257, + "step": 13322 + }, + { + "epoch": 5.58, + "learning_rate": 4.430736893612327e-06, + "loss": 0.1262, + "step": 13323 + }, + { + "epoch": 5.58, + "learning_rate": 4.429328831625565e-06, + "loss": 0.0848, + "step": 13324 + }, + { + "epoch": 5.58, + "learning_rate": 4.427920929762803e-06, + "loss": 0.1323, + "step": 13325 + }, + { + "epoch": 5.58, + "learning_rate": 4.426513188064507e-06, + "loss": 0.1331, + "step": 13326 + }, + { + "epoch": 5.58, + "learning_rate": 4.425105606571145e-06, + "loss": 0.1298, + "step": 13327 + }, + { + "epoch": 5.58, + "learning_rate": 4.423698185323175e-06, + "loss": 0.1379, + "step": 13328 + }, + { + "epoch": 5.58, + "learning_rate": 4.422290924361056e-06, + "loss": 0.1413, + "step": 13329 + }, + { + "epoch": 5.58, + "learning_rate": 4.42088382372523e-06, + "loss": 0.086, + "step": 13330 + }, + { + "epoch": 5.58, + "learning_rate": 4.41947688345615e-06, + "loss": 0.1054, + "step": 13331 + }, + { + "epoch": 5.58, + "learning_rate": 4.418070103594253e-06, + "loss": 0.1087, + "step": 13332 + }, + { + "epoch": 5.58, + "learning_rate": 4.416663484179978e-06, + "loss": 0.138, + "step": 13333 + }, + { + "epoch": 5.58, + "learning_rate": 4.415257025253758e-06, + "loss": 0.1474, + "step": 13334 + }, + { + "epoch": 5.58, + "learning_rate": 4.41385072685602e-06, + "loss": 0.1176, + "step": 13335 + }, + { + "epoch": 5.58, + "learning_rate": 4.412444589027185e-06, + "loss": 0.1425, + "step": 13336 + }, + { + "epoch": 5.58, + "learning_rate": 4.41103861180767e-06, + "loss": 0.1142, + "step": 13337 + }, + { + "epoch": 5.58, + "learning_rate": 4.4096327952378915e-06, + "loss": 0.0912, + "step": 13338 + }, + { + "epoch": 5.58, + "learning_rate": 4.408227139358261e-06, + "loss": 0.1089, + "step": 13339 + }, + { + "epoch": 5.59, + "learning_rate": 4.406821644209175e-06, + "loss": 0.1501, + "step": 13340 + }, + { + "epoch": 5.59, + "learning_rate": 4.4054163098310375e-06, + "loss": 0.1016, + "step": 13341 + }, + { + "epoch": 5.59, + "learning_rate": 4.404011136264245e-06, + "loss": 0.0911, + "step": 13342 + }, + { + "epoch": 5.59, + "learning_rate": 4.402606123549189e-06, + "loss": 0.128, + "step": 13343 + }, + { + "epoch": 5.59, + "learning_rate": 4.40120127172625e-06, + "loss": 0.1497, + "step": 13344 + }, + { + "epoch": 5.59, + "learning_rate": 4.399796580835811e-06, + "loss": 0.1469, + "step": 13345 + }, + { + "epoch": 5.59, + "learning_rate": 4.398392050918251e-06, + "loss": 0.1182, + "step": 13346 + }, + { + "epoch": 5.59, + "learning_rate": 4.39698768201394e-06, + "loss": 0.1152, + "step": 13347 + }, + { + "epoch": 5.59, + "learning_rate": 4.395583474163246e-06, + "loss": 0.1244, + "step": 13348 + }, + { + "epoch": 5.59, + "learning_rate": 4.394179427406533e-06, + "loss": 0.103, + "step": 13349 + }, + { + "epoch": 5.59, + "learning_rate": 4.392775541784161e-06, + "loss": 0.1087, + "step": 13350 + }, + { + "epoch": 5.59, + "learning_rate": 4.391371817336476e-06, + "loss": 0.1513, + "step": 13351 + }, + { + "epoch": 5.59, + "learning_rate": 4.3899682541038316e-06, + "loss": 0.1423, + "step": 13352 + }, + { + "epoch": 5.59, + "learning_rate": 4.388564852126571e-06, + "loss": 0.0978, + "step": 13353 + }, + { + "epoch": 5.59, + "learning_rate": 4.387161611445034e-06, + "loss": 0.1128, + "step": 13354 + }, + { + "epoch": 5.59, + "learning_rate": 4.385758532099556e-06, + "loss": 0.1407, + "step": 13355 + }, + { + "epoch": 5.59, + "learning_rate": 4.384355614130471e-06, + "loss": 0.1338, + "step": 13356 + }, + { + "epoch": 5.59, + "learning_rate": 4.382952857578098e-06, + "loss": 0.1307, + "step": 13357 + }, + { + "epoch": 5.59, + "learning_rate": 4.381550262482759e-06, + "loss": 0.1181, + "step": 13358 + }, + { + "epoch": 5.59, + "learning_rate": 4.380147828884772e-06, + "loss": 0.1042, + "step": 13359 + }, + { + "epoch": 5.59, + "learning_rate": 4.37874555682445e-06, + "loss": 0.1201, + "step": 13360 + }, + { + "epoch": 5.59, + "learning_rate": 4.377343446342097e-06, + "loss": 0.1473, + "step": 13361 + }, + { + "epoch": 5.59, + "learning_rate": 4.375941497478022e-06, + "loss": 0.1154, + "step": 13362 + }, + { + "epoch": 5.59, + "learning_rate": 4.374539710272514e-06, + "loss": 0.1125, + "step": 13363 + }, + { + "epoch": 5.6, + "learning_rate": 4.373138084765875e-06, + "loss": 0.132, + "step": 13364 + }, + { + "epoch": 5.6, + "learning_rate": 4.371736620998383e-06, + "loss": 0.1434, + "step": 13365 + }, + { + "epoch": 5.6, + "learning_rate": 4.370335319010328e-06, + "loss": 0.1287, + "step": 13366 + }, + { + "epoch": 5.6, + "learning_rate": 4.368934178841988e-06, + "loss": 0.1432, + "step": 13367 + }, + { + "epoch": 5.6, + "learning_rate": 4.367533200533639e-06, + "loss": 0.1219, + "step": 13368 + }, + { + "epoch": 5.6, + "learning_rate": 4.36613238412555e-06, + "loss": 0.1447, + "step": 13369 + }, + { + "epoch": 5.6, + "learning_rate": 4.36473172965799e-06, + "loss": 0.1143, + "step": 13370 + }, + { + "epoch": 5.6, + "learning_rate": 4.363331237171211e-06, + "loss": 0.111, + "step": 13371 + }, + { + "epoch": 5.6, + "learning_rate": 4.361930906705475e-06, + "loss": 0.115, + "step": 13372 + }, + { + "epoch": 5.6, + "learning_rate": 4.360530738301031e-06, + "loss": 0.1121, + "step": 13373 + }, + { + "epoch": 5.6, + "learning_rate": 4.359130731998127e-06, + "loss": 0.1524, + "step": 13374 + }, + { + "epoch": 5.6, + "learning_rate": 4.357730887837005e-06, + "loss": 0.1011, + "step": 13375 + }, + { + "epoch": 5.6, + "learning_rate": 4.356331205857904e-06, + "loss": 0.1064, + "step": 13376 + }, + { + "epoch": 5.6, + "learning_rate": 4.354931686101051e-06, + "loss": 0.1177, + "step": 13377 + }, + { + "epoch": 5.6, + "learning_rate": 4.353532328606679e-06, + "loss": 0.0824, + "step": 13378 + }, + { + "epoch": 5.6, + "learning_rate": 4.352133133415009e-06, + "loss": 0.12, + "step": 13379 + }, + { + "epoch": 5.6, + "learning_rate": 4.3507341005662595e-06, + "loss": 0.1201, + "step": 13380 + }, + { + "epoch": 5.6, + "learning_rate": 4.349335230100645e-06, + "loss": 0.0962, + "step": 13381 + }, + { + "epoch": 5.6, + "learning_rate": 4.3479365220583754e-06, + "loss": 0.1817, + "step": 13382 + }, + { + "epoch": 5.6, + "learning_rate": 4.346537976479659e-06, + "loss": 0.143, + "step": 13383 + }, + { + "epoch": 5.6, + "learning_rate": 4.345139593404688e-06, + "loss": 0.1115, + "step": 13384 + }, + { + "epoch": 5.6, + "learning_rate": 4.343741372873665e-06, + "loss": 0.104, + "step": 13385 + }, + { + "epoch": 5.6, + "learning_rate": 4.342343314926773e-06, + "loss": 0.1272, + "step": 13386 + }, + { + "epoch": 5.6, + "learning_rate": 4.340945419604201e-06, + "loss": 0.1128, + "step": 13387 + }, + { + "epoch": 5.61, + "learning_rate": 4.3395476869461325e-06, + "loss": 0.115, + "step": 13388 + }, + { + "epoch": 5.61, + "learning_rate": 4.338150116992743e-06, + "loss": 0.1597, + "step": 13389 + }, + { + "epoch": 5.61, + "learning_rate": 4.3367527097842034e-06, + "loss": 0.1409, + "step": 13390 + }, + { + "epoch": 5.61, + "learning_rate": 4.335355465360686e-06, + "loss": 0.1341, + "step": 13391 + }, + { + "epoch": 5.61, + "learning_rate": 4.333958383762345e-06, + "loss": 0.1257, + "step": 13392 + }, + { + "epoch": 5.61, + "learning_rate": 4.332561465029341e-06, + "loss": 0.1009, + "step": 13393 + }, + { + "epoch": 5.61, + "learning_rate": 4.331164709201829e-06, + "loss": 0.099, + "step": 13394 + }, + { + "epoch": 5.61, + "learning_rate": 4.329768116319958e-06, + "loss": 0.0919, + "step": 13395 + }, + { + "epoch": 5.61, + "learning_rate": 4.328371686423871e-06, + "loss": 0.085, + "step": 13396 + }, + { + "epoch": 5.61, + "learning_rate": 4.32697541955371e-06, + "loss": 0.1586, + "step": 13397 + }, + { + "epoch": 5.61, + "learning_rate": 4.325579315749603e-06, + "loss": 0.1115, + "step": 13398 + }, + { + "epoch": 5.61, + "learning_rate": 4.324183375051683e-06, + "loss": 0.1282, + "step": 13399 + }, + { + "epoch": 5.61, + "learning_rate": 4.322787597500074e-06, + "loss": 0.0924, + "step": 13400 + }, + { + "epoch": 5.61, + "learning_rate": 4.321391983134899e-06, + "loss": 0.1703, + "step": 13401 + }, + { + "epoch": 5.61, + "learning_rate": 4.319996531996272e-06, + "loss": 0.1231, + "step": 13402 + }, + { + "epoch": 5.61, + "learning_rate": 4.318601244124308e-06, + "loss": 0.1272, + "step": 13403 + }, + { + "epoch": 5.61, + "learning_rate": 4.317206119559108e-06, + "loss": 0.114, + "step": 13404 + }, + { + "epoch": 5.61, + "learning_rate": 4.315811158340773e-06, + "loss": 0.11, + "step": 13405 + }, + { + "epoch": 5.61, + "learning_rate": 4.314416360509403e-06, + "loss": 0.1258, + "step": 13406 + }, + { + "epoch": 5.61, + "learning_rate": 4.3130217261050925e-06, + "loss": 0.1358, + "step": 13407 + }, + { + "epoch": 5.61, + "learning_rate": 4.3116272551679215e-06, + "loss": 0.1249, + "step": 13408 + }, + { + "epoch": 5.61, + "learning_rate": 4.310232947737979e-06, + "loss": 0.1491, + "step": 13409 + }, + { + "epoch": 5.61, + "learning_rate": 4.308838803855341e-06, + "loss": 0.1461, + "step": 13410 + }, + { + "epoch": 5.61, + "learning_rate": 4.307444823560084e-06, + "loss": 0.1283, + "step": 13411 + }, + { + "epoch": 5.62, + "learning_rate": 4.306051006892271e-06, + "loss": 0.1214, + "step": 13412 + }, + { + "epoch": 5.62, + "learning_rate": 4.30465735389197e-06, + "loss": 0.1222, + "step": 13413 + }, + { + "epoch": 5.62, + "learning_rate": 4.303263864599239e-06, + "loss": 0.0874, + "step": 13414 + }, + { + "epoch": 5.62, + "learning_rate": 4.3018705390541316e-06, + "loss": 0.1077, + "step": 13415 + }, + { + "epoch": 5.62, + "learning_rate": 4.3004773772967e-06, + "loss": 0.1062, + "step": 13416 + }, + { + "epoch": 5.62, + "learning_rate": 4.2990843793669925e-06, + "loss": 0.0945, + "step": 13417 + }, + { + "epoch": 5.62, + "learning_rate": 4.297691545305042e-06, + "loss": 0.1373, + "step": 13418 + }, + { + "epoch": 5.62, + "learning_rate": 4.296298875150887e-06, + "loss": 0.1218, + "step": 13419 + }, + { + "epoch": 5.62, + "learning_rate": 4.29490636894456e-06, + "loss": 0.1227, + "step": 13420 + }, + { + "epoch": 5.62, + "learning_rate": 4.293514026726088e-06, + "loss": 0.1127, + "step": 13421 + }, + { + "epoch": 5.62, + "learning_rate": 4.29212184853549e-06, + "loss": 0.1294, + "step": 13422 + }, + { + "epoch": 5.62, + "learning_rate": 4.2907298344127844e-06, + "loss": 0.1136, + "step": 13423 + }, + { + "epoch": 5.62, + "learning_rate": 4.289337984397988e-06, + "loss": 0.0996, + "step": 13424 + }, + { + "epoch": 5.62, + "learning_rate": 4.287946298531099e-06, + "loss": 0.107, + "step": 13425 + }, + { + "epoch": 5.62, + "learning_rate": 4.286554776852125e-06, + "loss": 0.1048, + "step": 13426 + }, + { + "epoch": 5.62, + "learning_rate": 4.285163419401065e-06, + "loss": 0.1187, + "step": 13427 + }, + { + "epoch": 5.62, + "learning_rate": 4.2837722262179105e-06, + "loss": 0.1657, + "step": 13428 + }, + { + "epoch": 5.62, + "learning_rate": 4.282381197342651e-06, + "loss": 0.118, + "step": 13429 + }, + { + "epoch": 5.62, + "learning_rate": 4.280990332815274e-06, + "loss": 0.1607, + "step": 13430 + }, + { + "epoch": 5.62, + "learning_rate": 4.27959963267575e-06, + "loss": 0.151, + "step": 13431 + }, + { + "epoch": 5.62, + "learning_rate": 4.278209096964063e-06, + "loss": 0.1083, + "step": 13432 + }, + { + "epoch": 5.62, + "learning_rate": 4.276818725720176e-06, + "loss": 0.0842, + "step": 13433 + }, + { + "epoch": 5.62, + "learning_rate": 4.275428518984054e-06, + "loss": 0.1159, + "step": 13434 + }, + { + "epoch": 5.62, + "learning_rate": 4.27403847679566e-06, + "loss": 0.1504, + "step": 13435 + }, + { + "epoch": 5.63, + "learning_rate": 4.272648599194948e-06, + "loss": 0.1206, + "step": 13436 + }, + { + "epoch": 5.63, + "learning_rate": 4.27125888622187e-06, + "loss": 0.1466, + "step": 13437 + }, + { + "epoch": 5.63, + "learning_rate": 4.269869337916377e-06, + "loss": 0.1112, + "step": 13438 + }, + { + "epoch": 5.63, + "learning_rate": 4.2684799543183995e-06, + "loss": 0.0942, + "step": 13439 + }, + { + "epoch": 5.63, + "learning_rate": 4.26709073546788e-06, + "loss": 0.1187, + "step": 13440 + }, + { + "epoch": 5.63, + "learning_rate": 4.2657016814047515e-06, + "loss": 0.118, + "step": 13441 + }, + { + "epoch": 5.63, + "learning_rate": 4.264312792168938e-06, + "loss": 0.0935, + "step": 13442 + }, + { + "epoch": 5.63, + "learning_rate": 4.262924067800364e-06, + "loss": 0.1004, + "step": 13443 + }, + { + "epoch": 5.63, + "learning_rate": 4.261535508338951e-06, + "loss": 0.113, + "step": 13444 + }, + { + "epoch": 5.63, + "learning_rate": 4.260147113824603e-06, + "loss": 0.1241, + "step": 13445 + }, + { + "epoch": 5.63, + "learning_rate": 4.2587588842972325e-06, + "loss": 0.131, + "step": 13446 + }, + { + "epoch": 5.63, + "learning_rate": 4.257370819796745e-06, + "loss": 0.1177, + "step": 13447 + }, + { + "epoch": 5.63, + "learning_rate": 4.255982920363035e-06, + "loss": 0.1428, + "step": 13448 + }, + { + "epoch": 5.63, + "learning_rate": 4.254595186036001e-06, + "loss": 0.1097, + "step": 13449 + }, + { + "epoch": 5.63, + "learning_rate": 4.253207616855528e-06, + "loss": 0.1172, + "step": 13450 + }, + { + "epoch": 5.63, + "learning_rate": 4.251820212861507e-06, + "loss": 0.1338, + "step": 13451 + }, + { + "epoch": 5.63, + "learning_rate": 4.2504329740938084e-06, + "loss": 0.1281, + "step": 13452 + }, + { + "epoch": 5.63, + "learning_rate": 4.249045900592318e-06, + "loss": 0.1366, + "step": 13453 + }, + { + "epoch": 5.63, + "learning_rate": 4.2476589923968946e-06, + "loss": 0.0745, + "step": 13454 + }, + { + "epoch": 5.63, + "learning_rate": 4.2462722495474085e-06, + "loss": 0.1222, + "step": 13455 + }, + { + "epoch": 5.63, + "learning_rate": 4.244885672083722e-06, + "loss": 0.1208, + "step": 13456 + }, + { + "epoch": 5.63, + "learning_rate": 4.243499260045689e-06, + "loss": 0.1247, + "step": 13457 + }, + { + "epoch": 5.63, + "learning_rate": 4.242113013473166e-06, + "loss": 0.1226, + "step": 13458 + }, + { + "epoch": 5.63, + "learning_rate": 4.240726932405991e-06, + "loss": 0.1737, + "step": 13459 + }, + { + "epoch": 5.64, + "learning_rate": 4.239341016884008e-06, + "loss": 0.1072, + "step": 13460 + }, + { + "epoch": 5.64, + "learning_rate": 4.237955266947058e-06, + "loss": 0.1087, + "step": 13461 + }, + { + "epoch": 5.64, + "learning_rate": 4.23656968263497e-06, + "loss": 0.137, + "step": 13462 + }, + { + "epoch": 5.64, + "learning_rate": 4.235184263987572e-06, + "loss": 0.1212, + "step": 13463 + }, + { + "epoch": 5.64, + "learning_rate": 4.233799011044687e-06, + "loss": 0.1699, + "step": 13464 + }, + { + "epoch": 5.64, + "learning_rate": 4.232413923846135e-06, + "loss": 0.1369, + "step": 13465 + }, + { + "epoch": 5.64, + "learning_rate": 4.231029002431723e-06, + "loss": 0.0873, + "step": 13466 + }, + { + "epoch": 5.64, + "learning_rate": 4.2296442468412646e-06, + "loss": 0.1319, + "step": 13467 + }, + { + "epoch": 5.64, + "learning_rate": 4.2282596571145615e-06, + "loss": 0.1551, + "step": 13468 + }, + { + "epoch": 5.64, + "learning_rate": 4.226875233291412e-06, + "loss": 0.1186, + "step": 13469 + }, + { + "epoch": 5.64, + "learning_rate": 4.225490975411611e-06, + "loss": 0.1651, + "step": 13470 + }, + { + "epoch": 5.64, + "learning_rate": 4.224106883514952e-06, + "loss": 0.1228, + "step": 13471 + }, + { + "epoch": 5.64, + "learning_rate": 4.222722957641212e-06, + "loss": 0.1283, + "step": 13472 + }, + { + "epoch": 5.64, + "learning_rate": 4.221339197830173e-06, + "loss": 0.1121, + "step": 13473 + }, + { + "epoch": 5.64, + "learning_rate": 4.219955604121611e-06, + "loss": 0.1387, + "step": 13474 + }, + { + "epoch": 5.64, + "learning_rate": 4.2185721765553e-06, + "loss": 0.1105, + "step": 13475 + }, + { + "epoch": 5.64, + "learning_rate": 4.217188915170996e-06, + "loss": 0.12, + "step": 13476 + }, + { + "epoch": 5.64, + "learning_rate": 4.215805820008468e-06, + "loss": 0.1138, + "step": 13477 + }, + { + "epoch": 5.64, + "learning_rate": 4.214422891107467e-06, + "loss": 0.1054, + "step": 13478 + }, + { + "epoch": 5.64, + "learning_rate": 4.2130401285077505e-06, + "loss": 0.1142, + "step": 13479 + }, + { + "epoch": 5.64, + "learning_rate": 4.211657532249056e-06, + "loss": 0.1037, + "step": 13480 + }, + { + "epoch": 5.64, + "learning_rate": 4.21027510237113e-06, + "loss": 0.1261, + "step": 13481 + }, + { + "epoch": 5.64, + "learning_rate": 4.208892838913707e-06, + "loss": 0.1181, + "step": 13482 + }, + { + "epoch": 5.64, + "learning_rate": 4.207510741916522e-06, + "loss": 0.1007, + "step": 13483 + }, + { + "epoch": 5.65, + "learning_rate": 4.206128811419299e-06, + "loss": 0.1343, + "step": 13484 + }, + { + "epoch": 5.65, + "learning_rate": 4.204747047461767e-06, + "loss": 0.1377, + "step": 13485 + }, + { + "epoch": 5.65, + "learning_rate": 4.2033654500836344e-06, + "loss": 0.1334, + "step": 13486 + }, + { + "epoch": 5.65, + "learning_rate": 4.201984019324617e-06, + "loss": 0.1247, + "step": 13487 + }, + { + "epoch": 5.65, + "learning_rate": 4.200602755224426e-06, + "loss": 0.1618, + "step": 13488 + }, + { + "epoch": 5.65, + "learning_rate": 4.199221657822761e-06, + "loss": 0.1531, + "step": 13489 + }, + { + "epoch": 5.65, + "learning_rate": 4.1978407271593226e-06, + "loss": 0.1097, + "step": 13490 + }, + { + "epoch": 5.65, + "learning_rate": 4.1964599632738035e-06, + "loss": 0.1293, + "step": 13491 + }, + { + "epoch": 5.65, + "learning_rate": 4.1950793662058965e-06, + "loss": 0.1205, + "step": 13492 + }, + { + "epoch": 5.65, + "learning_rate": 4.1936989359952775e-06, + "loss": 0.1221, + "step": 13493 + }, + { + "epoch": 5.65, + "learning_rate": 4.192318672681631e-06, + "loss": 0.1379, + "step": 13494 + }, + { + "epoch": 5.65, + "learning_rate": 4.1909385763046305e-06, + "loss": 0.1072, + "step": 13495 + }, + { + "epoch": 5.65, + "learning_rate": 4.1895586469039466e-06, + "loss": 0.0995, + "step": 13496 + }, + { + "epoch": 5.65, + "learning_rate": 4.188178884519241e-06, + "loss": 0.1429, + "step": 13497 + }, + { + "epoch": 5.65, + "learning_rate": 4.186799289190182e-06, + "loss": 0.1256, + "step": 13498 + }, + { + "epoch": 5.65, + "learning_rate": 4.185419860956418e-06, + "loss": 0.1429, + "step": 13499 + }, + { + "epoch": 5.65, + "learning_rate": 4.184040599857596e-06, + "loss": 0.1564, + "step": 13500 + }, + { + "epoch": 5.65, + "learning_rate": 4.182661505933366e-06, + "loss": 0.1095, + "step": 13501 + }, + { + "epoch": 5.65, + "learning_rate": 4.18128257922337e-06, + "loss": 0.1098, + "step": 13502 + }, + { + "epoch": 5.65, + "learning_rate": 4.179903819767242e-06, + "loss": 0.1369, + "step": 13503 + }, + { + "epoch": 5.65, + "learning_rate": 4.178525227604615e-06, + "loss": 0.1407, + "step": 13504 + }, + { + "epoch": 5.65, + "learning_rate": 4.177146802775114e-06, + "loss": 0.1111, + "step": 13505 + }, + { + "epoch": 5.65, + "learning_rate": 4.175768545318364e-06, + "loss": 0.1186, + "step": 13506 + }, + { + "epoch": 5.66, + "learning_rate": 4.174390455273975e-06, + "loss": 0.1016, + "step": 13507 + }, + { + "epoch": 5.66, + "learning_rate": 4.173012532681564e-06, + "loss": 0.1468, + "step": 13508 + }, + { + "epoch": 5.66, + "learning_rate": 4.171634777580736e-06, + "loss": 0.1381, + "step": 13509 + }, + { + "epoch": 5.66, + "learning_rate": 4.170257190011096e-06, + "loss": 0.1473, + "step": 13510 + }, + { + "epoch": 5.66, + "learning_rate": 4.168879770012237e-06, + "loss": 0.0879, + "step": 13511 + }, + { + "epoch": 5.66, + "learning_rate": 4.16750251762376e-06, + "loss": 0.0742, + "step": 13512 + }, + { + "epoch": 5.66, + "learning_rate": 4.166125432885244e-06, + "loss": 0.1145, + "step": 13513 + }, + { + "epoch": 5.66, + "learning_rate": 4.164748515836276e-06, + "loss": 0.0923, + "step": 13514 + }, + { + "epoch": 5.66, + "learning_rate": 4.163371766516432e-06, + "loss": 0.1548, + "step": 13515 + }, + { + "epoch": 5.66, + "learning_rate": 4.1619951849652885e-06, + "loss": 0.1205, + "step": 13516 + }, + { + "epoch": 5.66, + "learning_rate": 4.160618771222412e-06, + "loss": 0.1102, + "step": 13517 + }, + { + "epoch": 5.66, + "learning_rate": 4.159242525327371e-06, + "loss": 0.112, + "step": 13518 + }, + { + "epoch": 5.66, + "learning_rate": 4.157866447319718e-06, + "loss": 0.1344, + "step": 13519 + }, + { + "epoch": 5.66, + "learning_rate": 4.15649053723901e-06, + "loss": 0.1181, + "step": 13520 + }, + { + "epoch": 5.66, + "learning_rate": 4.155114795124798e-06, + "loss": 0.1008, + "step": 13521 + }, + { + "epoch": 5.66, + "learning_rate": 4.153739221016622e-06, + "loss": 0.0864, + "step": 13522 + }, + { + "epoch": 5.66, + "learning_rate": 4.152363814954024e-06, + "loss": 0.1235, + "step": 13523 + }, + { + "epoch": 5.66, + "learning_rate": 4.150988576976539e-06, + "loss": 0.1005, + "step": 13524 + }, + { + "epoch": 5.66, + "learning_rate": 4.1496135071236965e-06, + "loss": 0.075, + "step": 13525 + }, + { + "epoch": 5.66, + "learning_rate": 4.1482386054350265e-06, + "loss": 0.1, + "step": 13526 + }, + { + "epoch": 5.66, + "learning_rate": 4.146863871950041e-06, + "loss": 0.0988, + "step": 13527 + }, + { + "epoch": 5.66, + "learning_rate": 4.1454893067082605e-06, + "loss": 0.1228, + "step": 13528 + }, + { + "epoch": 5.66, + "learning_rate": 4.144114909749193e-06, + "loss": 0.127, + "step": 13529 + }, + { + "epoch": 5.66, + "learning_rate": 4.142740681112347e-06, + "loss": 0.1128, + "step": 13530 + }, + { + "epoch": 5.67, + "learning_rate": 4.141366620837223e-06, + "loss": 0.1203, + "step": 13531 + }, + { + "epoch": 5.67, + "learning_rate": 4.139992728963315e-06, + "loss": 0.1196, + "step": 13532 + }, + { + "epoch": 5.67, + "learning_rate": 4.138619005530121e-06, + "loss": 0.1411, + "step": 13533 + }, + { + "epoch": 5.67, + "learning_rate": 4.1372454505771186e-06, + "loss": 0.1377, + "step": 13534 + }, + { + "epoch": 5.67, + "learning_rate": 4.135872064143792e-06, + "loss": 0.1149, + "step": 13535 + }, + { + "epoch": 5.67, + "learning_rate": 4.134498846269621e-06, + "loss": 0.1183, + "step": 13536 + }, + { + "epoch": 5.67, + "learning_rate": 4.133125796994074e-06, + "loss": 0.134, + "step": 13537 + }, + { + "epoch": 5.67, + "learning_rate": 4.131752916356622e-06, + "loss": 0.1001, + "step": 13538 + }, + { + "epoch": 5.67, + "learning_rate": 4.130380204396727e-06, + "loss": 0.1182, + "step": 13539 + }, + { + "epoch": 5.67, + "learning_rate": 4.129007661153841e-06, + "loss": 0.1349, + "step": 13540 + }, + { + "epoch": 5.67, + "learning_rate": 4.12763528666742e-06, + "loss": 0.1489, + "step": 13541 + }, + { + "epoch": 5.67, + "learning_rate": 4.126263080976911e-06, + "loss": 0.1143, + "step": 13542 + }, + { + "epoch": 5.67, + "learning_rate": 4.124891044121761e-06, + "loss": 0.1215, + "step": 13543 + }, + { + "epoch": 5.67, + "learning_rate": 4.1235191761414025e-06, + "loss": 0.0732, + "step": 13544 + }, + { + "epoch": 5.67, + "learning_rate": 4.12214747707527e-06, + "loss": 0.1264, + "step": 13545 + }, + { + "epoch": 5.67, + "learning_rate": 4.120775946962792e-06, + "loss": 0.1171, + "step": 13546 + }, + { + "epoch": 5.67, + "learning_rate": 4.1194045858433965e-06, + "loss": 0.1367, + "step": 13547 + }, + { + "epoch": 5.67, + "learning_rate": 4.118033393756493e-06, + "loss": 0.1333, + "step": 13548 + }, + { + "epoch": 5.67, + "learning_rate": 4.116662370741501e-06, + "loss": 0.0928, + "step": 13549 + }, + { + "epoch": 5.67, + "learning_rate": 4.115291516837827e-06, + "loss": 0.1202, + "step": 13550 + }, + { + "epoch": 5.67, + "learning_rate": 4.113920832084878e-06, + "loss": 0.1247, + "step": 13551 + }, + { + "epoch": 5.67, + "learning_rate": 4.11255031652205e-06, + "loss": 0.1122, + "step": 13552 + }, + { + "epoch": 5.67, + "learning_rate": 4.111179970188745e-06, + "loss": 0.1025, + "step": 13553 + }, + { + "epoch": 5.67, + "learning_rate": 4.10980979312434e-06, + "loss": 0.1193, + "step": 13554 + }, + { + "epoch": 5.68, + "learning_rate": 4.108439785368226e-06, + "loss": 0.1566, + "step": 13555 + }, + { + "epoch": 5.68, + "learning_rate": 4.107069946959783e-06, + "loss": 0.1232, + "step": 13556 + }, + { + "epoch": 5.68, + "learning_rate": 4.105700277938385e-06, + "loss": 0.1144, + "step": 13557 + }, + { + "epoch": 5.68, + "learning_rate": 4.1043307783434e-06, + "loss": 0.1173, + "step": 13558 + }, + { + "epoch": 5.68, + "learning_rate": 4.102961448214201e-06, + "loss": 0.1286, + "step": 13559 + }, + { + "epoch": 5.68, + "learning_rate": 4.101592287590138e-06, + "loss": 0.1327, + "step": 13560 + }, + { + "epoch": 5.68, + "learning_rate": 4.100223296510571e-06, + "loss": 0.1254, + "step": 13561 + }, + { + "epoch": 5.68, + "learning_rate": 4.098854475014849e-06, + "loss": 0.1013, + "step": 13562 + }, + { + "epoch": 5.68, + "learning_rate": 4.097485823142319e-06, + "loss": 0.1239, + "step": 13563 + }, + { + "epoch": 5.68, + "learning_rate": 4.096117340932322e-06, + "loss": 0.1021, + "step": 13564 + }, + { + "epoch": 5.68, + "learning_rate": 4.094749028424192e-06, + "loss": 0.1207, + "step": 13565 + }, + { + "epoch": 5.68, + "learning_rate": 4.093380885657264e-06, + "loss": 0.1143, + "step": 13566 + }, + { + "epoch": 5.68, + "learning_rate": 4.092012912670863e-06, + "loss": 0.1372, + "step": 13567 + }, + { + "epoch": 5.68, + "learning_rate": 4.090645109504303e-06, + "loss": 0.0726, + "step": 13568 + }, + { + "epoch": 5.68, + "learning_rate": 4.0892774761969055e-06, + "loss": 0.1092, + "step": 13569 + }, + { + "epoch": 5.68, + "learning_rate": 4.087910012787983e-06, + "loss": 0.1253, + "step": 13570 + }, + { + "epoch": 5.68, + "learning_rate": 4.08654271931684e-06, + "loss": 0.1181, + "step": 13571 + }, + { + "epoch": 5.68, + "learning_rate": 4.085175595822779e-06, + "loss": 0.1061, + "step": 13572 + }, + { + "epoch": 5.68, + "learning_rate": 4.083808642345098e-06, + "loss": 0.1261, + "step": 13573 + }, + { + "epoch": 5.68, + "learning_rate": 4.082441858923088e-06, + "loss": 0.0955, + "step": 13574 + }, + { + "epoch": 5.68, + "learning_rate": 4.0810752455960356e-06, + "loss": 0.1048, + "step": 13575 + }, + { + "epoch": 5.68, + "learning_rate": 4.07970880240322e-06, + "loss": 0.069, + "step": 13576 + }, + { + "epoch": 5.68, + "learning_rate": 4.0783425293839245e-06, + "loss": 0.1157, + "step": 13577 + }, + { + "epoch": 5.68, + "learning_rate": 4.0769764265774155e-06, + "loss": 0.111, + "step": 13578 + }, + { + "epoch": 5.69, + "learning_rate": 4.0756104940229645e-06, + "loss": 0.1169, + "step": 13579 + }, + { + "epoch": 5.69, + "learning_rate": 4.074244731759836e-06, + "loss": 0.1089, + "step": 13580 + }, + { + "epoch": 5.69, + "learning_rate": 4.07287913982728e-06, + "loss": 0.1629, + "step": 13581 + }, + { + "epoch": 5.69, + "learning_rate": 4.071513718264554e-06, + "loss": 0.1112, + "step": 13582 + }, + { + "epoch": 5.69, + "learning_rate": 4.070148467110906e-06, + "loss": 0.113, + "step": 13583 + }, + { + "epoch": 5.69, + "learning_rate": 4.068783386405577e-06, + "loss": 0.145, + "step": 13584 + }, + { + "epoch": 5.69, + "learning_rate": 4.067418476187807e-06, + "loss": 0.1243, + "step": 13585 + }, + { + "epoch": 5.69, + "learning_rate": 4.066053736496832e-06, + "loss": 0.0941, + "step": 13586 + }, + { + "epoch": 5.69, + "learning_rate": 4.0646891673718725e-06, + "loss": 0.1226, + "step": 13587 + }, + { + "epoch": 5.69, + "learning_rate": 4.063324768852157e-06, + "loss": 0.1108, + "step": 13588 + }, + { + "epoch": 5.69, + "learning_rate": 4.061960540976906e-06, + "loss": 0.1098, + "step": 13589 + }, + { + "epoch": 5.69, + "learning_rate": 4.060596483785326e-06, + "loss": 0.152, + "step": 13590 + }, + { + "epoch": 5.69, + "learning_rate": 4.059232597316629e-06, + "loss": 0.1005, + "step": 13591 + }, + { + "epoch": 5.69, + "learning_rate": 4.057868881610021e-06, + "loss": 0.1304, + "step": 13592 + }, + { + "epoch": 5.69, + "learning_rate": 4.056505336704698e-06, + "loss": 0.0902, + "step": 13593 + }, + { + "epoch": 5.69, + "learning_rate": 4.055141962639858e-06, + "loss": 0.1515, + "step": 13594 + }, + { + "epoch": 5.69, + "learning_rate": 4.053778759454684e-06, + "loss": 0.0776, + "step": 13595 + }, + { + "epoch": 5.69, + "learning_rate": 4.0524157271883635e-06, + "loss": 0.1422, + "step": 13596 + }, + { + "epoch": 5.69, + "learning_rate": 4.051052865880074e-06, + "loss": 0.124, + "step": 13597 + }, + { + "epoch": 5.69, + "learning_rate": 4.049690175568992e-06, + "loss": 0.093, + "step": 13598 + }, + { + "epoch": 5.69, + "learning_rate": 4.048327656294285e-06, + "loss": 0.09, + "step": 13599 + }, + { + "epoch": 5.69, + "learning_rate": 4.046965308095122e-06, + "loss": 0.1262, + "step": 13600 + }, + { + "epoch": 5.69, + "learning_rate": 4.045603131010655e-06, + "loss": 0.1213, + "step": 13601 + }, + { + "epoch": 5.69, + "learning_rate": 4.044241125080041e-06, + "loss": 0.1101, + "step": 13602 + }, + { + "epoch": 5.7, + "learning_rate": 4.042879290342432e-06, + "loss": 0.1136, + "step": 13603 + }, + { + "epoch": 5.7, + "learning_rate": 4.0415176268369714e-06, + "loss": 0.1203, + "step": 13604 + }, + { + "epoch": 5.7, + "learning_rate": 4.0401561346028e-06, + "loss": 0.084, + "step": 13605 + }, + { + "epoch": 5.7, + "learning_rate": 4.03879481367905e-06, + "loss": 0.1028, + "step": 13606 + }, + { + "epoch": 5.7, + "learning_rate": 4.037433664104857e-06, + "loss": 0.1255, + "step": 13607 + }, + { + "epoch": 5.7, + "learning_rate": 4.03607268591934e-06, + "loss": 0.1055, + "step": 13608 + }, + { + "epoch": 5.7, + "learning_rate": 4.03471187916162e-06, + "loss": 0.1279, + "step": 13609 + }, + { + "epoch": 5.7, + "learning_rate": 4.033351243870815e-06, + "loss": 0.1281, + "step": 13610 + }, + { + "epoch": 5.7, + "learning_rate": 4.031990780086037e-06, + "loss": 0.1073, + "step": 13611 + }, + { + "epoch": 5.7, + "learning_rate": 4.0306304878463834e-06, + "loss": 0.1445, + "step": 13612 + }, + { + "epoch": 5.7, + "learning_rate": 4.0292703671909605e-06, + "loss": 0.1192, + "step": 13613 + }, + { + "epoch": 5.7, + "learning_rate": 4.027910418158861e-06, + "loss": 0.1246, + "step": 13614 + }, + { + "epoch": 5.7, + "learning_rate": 4.026550640789183e-06, + "loss": 0.1184, + "step": 13615 + }, + { + "epoch": 5.7, + "learning_rate": 4.025191035121001e-06, + "loss": 0.1196, + "step": 13616 + }, + { + "epoch": 5.7, + "learning_rate": 4.023831601193401e-06, + "loss": 0.1371, + "step": 13617 + }, + { + "epoch": 5.7, + "learning_rate": 4.0224723390454575e-06, + "loss": 0.0976, + "step": 13618 + }, + { + "epoch": 5.7, + "learning_rate": 4.021113248716243e-06, + "loss": 0.0785, + "step": 13619 + }, + { + "epoch": 5.7, + "learning_rate": 4.019754330244822e-06, + "loss": 0.1165, + "step": 13620 + }, + { + "epoch": 5.7, + "learning_rate": 4.0183955836702615e-06, + "loss": 0.1068, + "step": 13621 + }, + { + "epoch": 5.7, + "learning_rate": 4.017037009031607e-06, + "loss": 0.125, + "step": 13622 + }, + { + "epoch": 5.7, + "learning_rate": 4.015678606367916e-06, + "loss": 0.1584, + "step": 13623 + }, + { + "epoch": 5.7, + "learning_rate": 4.01432037571823e-06, + "loss": 0.1086, + "step": 13624 + }, + { + "epoch": 5.7, + "learning_rate": 4.012962317121596e-06, + "loss": 0.1481, + "step": 13625 + }, + { + "epoch": 5.7, + "learning_rate": 4.011604430617046e-06, + "loss": 0.1074, + "step": 13626 + }, + { + "epoch": 5.71, + "learning_rate": 4.010246716243617e-06, + "loss": 0.1154, + "step": 13627 + }, + { + "epoch": 5.71, + "learning_rate": 4.008889174040327e-06, + "loss": 0.1414, + "step": 13628 + }, + { + "epoch": 5.71, + "learning_rate": 4.0075318040462005e-06, + "loss": 0.1165, + "step": 13629 + }, + { + "epoch": 5.71, + "learning_rate": 4.006174606300255e-06, + "loss": 0.0938, + "step": 13630 + }, + { + "epoch": 5.71, + "learning_rate": 4.004817580841503e-06, + "loss": 0.1314, + "step": 13631 + }, + { + "epoch": 5.71, + "learning_rate": 4.003460727708947e-06, + "loss": 0.1182, + "step": 13632 + }, + { + "epoch": 5.71, + "learning_rate": 4.002104046941594e-06, + "loss": 0.1373, + "step": 13633 + }, + { + "epoch": 5.71, + "learning_rate": 4.00074753857844e-06, + "loss": 0.0948, + "step": 13634 + }, + { + "epoch": 5.71, + "learning_rate": 3.9993912026584735e-06, + "loss": 0.1131, + "step": 13635 + }, + { + "epoch": 5.71, + "learning_rate": 3.99803503922068e-06, + "loss": 0.1216, + "step": 13636 + }, + { + "epoch": 5.71, + "learning_rate": 3.996679048304042e-06, + "loss": 0.0913, + "step": 13637 + }, + { + "epoch": 5.71, + "learning_rate": 3.9953232299475385e-06, + "loss": 0.1136, + "step": 13638 + }, + { + "epoch": 5.71, + "learning_rate": 3.993967584190141e-06, + "loss": 0.1164, + "step": 13639 + }, + { + "epoch": 5.71, + "learning_rate": 3.992612111070814e-06, + "loss": 0.0847, + "step": 13640 + }, + { + "epoch": 5.71, + "learning_rate": 3.991256810628525e-06, + "loss": 0.1079, + "step": 13641 + }, + { + "epoch": 5.71, + "learning_rate": 3.989901682902223e-06, + "loss": 0.1372, + "step": 13642 + }, + { + "epoch": 5.71, + "learning_rate": 3.9885467279308644e-06, + "loss": 0.1175, + "step": 13643 + }, + { + "epoch": 5.71, + "learning_rate": 3.987191945753396e-06, + "loss": 0.1287, + "step": 13644 + }, + { + "epoch": 5.71, + "learning_rate": 3.9858373364087586e-06, + "loss": 0.0921, + "step": 13645 + }, + { + "epoch": 5.71, + "learning_rate": 3.98448289993589e-06, + "loss": 0.1228, + "step": 13646 + }, + { + "epoch": 5.71, + "learning_rate": 3.983128636373723e-06, + "loss": 0.1311, + "step": 13647 + }, + { + "epoch": 5.71, + "learning_rate": 3.981774545761188e-06, + "loss": 0.1644, + "step": 13648 + }, + { + "epoch": 5.71, + "learning_rate": 3.980420628137199e-06, + "loss": 0.1409, + "step": 13649 + }, + { + "epoch": 5.71, + "learning_rate": 3.979066883540677e-06, + "loss": 0.115, + "step": 13650 + }, + { + "epoch": 5.72, + "learning_rate": 3.977713312010535e-06, + "loss": 0.1419, + "step": 13651 + }, + { + "epoch": 5.72, + "learning_rate": 3.97635991358568e-06, + "loss": 0.1248, + "step": 13652 + }, + { + "epoch": 5.72, + "learning_rate": 3.975006688305015e-06, + "loss": 0.0984, + "step": 13653 + }, + { + "epoch": 5.72, + "learning_rate": 3.973653636207437e-06, + "loss": 0.0916, + "step": 13654 + }, + { + "epoch": 5.72, + "learning_rate": 3.972300757331836e-06, + "loss": 0.1347, + "step": 13655 + }, + { + "epoch": 5.72, + "learning_rate": 3.970948051717104e-06, + "loss": 0.1198, + "step": 13656 + }, + { + "epoch": 5.72, + "learning_rate": 3.969595519402117e-06, + "loss": 0.1114, + "step": 13657 + }, + { + "epoch": 5.72, + "learning_rate": 3.9682431604257554e-06, + "loss": 0.111, + "step": 13658 + }, + { + "epoch": 5.72, + "learning_rate": 3.966890974826891e-06, + "loss": 0.0948, + "step": 13659 + }, + { + "epoch": 5.72, + "learning_rate": 3.965538962644392e-06, + "loss": 0.0953, + "step": 13660 + }, + { + "epoch": 5.72, + "learning_rate": 3.964187123917122e-06, + "loss": 0.1222, + "step": 13661 + }, + { + "epoch": 5.72, + "learning_rate": 3.962835458683939e-06, + "loss": 0.1081, + "step": 13662 + }, + { + "epoch": 5.72, + "learning_rate": 3.9614839669836905e-06, + "loss": 0.0928, + "step": 13663 + }, + { + "epoch": 5.72, + "learning_rate": 3.960132648855226e-06, + "loss": 0.0574, + "step": 13664 + }, + { + "epoch": 5.72, + "learning_rate": 3.95878150433739e-06, + "loss": 0.1443, + "step": 13665 + }, + { + "epoch": 5.72, + "learning_rate": 3.9574305334690175e-06, + "loss": 0.1078, + "step": 13666 + }, + { + "epoch": 5.72, + "learning_rate": 3.9560797362889436e-06, + "loss": 0.1208, + "step": 13667 + }, + { + "epoch": 5.72, + "learning_rate": 3.954729112835998e-06, + "loss": 0.1301, + "step": 13668 + }, + { + "epoch": 5.72, + "learning_rate": 3.953378663148994e-06, + "loss": 0.1397, + "step": 13669 + }, + { + "epoch": 5.72, + "learning_rate": 3.952028387266757e-06, + "loss": 0.1295, + "step": 13670 + }, + { + "epoch": 5.72, + "learning_rate": 3.950678285228097e-06, + "loss": 0.1382, + "step": 13671 + }, + { + "epoch": 5.72, + "learning_rate": 3.94932835707182e-06, + "loss": 0.1159, + "step": 13672 + }, + { + "epoch": 5.72, + "learning_rate": 3.947978602836731e-06, + "loss": 0.1644, + "step": 13673 + }, + { + "epoch": 5.72, + "learning_rate": 3.946629022561627e-06, + "loss": 0.1093, + "step": 13674 + }, + { + "epoch": 5.73, + "learning_rate": 3.945279616285303e-06, + "loss": 0.117, + "step": 13675 + }, + { + "epoch": 5.73, + "learning_rate": 3.943930384046541e-06, + "loss": 0.1518, + "step": 13676 + }, + { + "epoch": 5.73, + "learning_rate": 3.942581325884124e-06, + "loss": 0.0997, + "step": 13677 + }, + { + "epoch": 5.73, + "learning_rate": 3.941232441836833e-06, + "loss": 0.1424, + "step": 13678 + }, + { + "epoch": 5.73, + "learning_rate": 3.939883731943443e-06, + "loss": 0.1216, + "step": 13679 + }, + { + "epoch": 5.73, + "learning_rate": 3.938535196242713e-06, + "loss": 0.1252, + "step": 13680 + }, + { + "epoch": 5.73, + "learning_rate": 3.93718683477341e-06, + "loss": 0.1371, + "step": 13681 + }, + { + "epoch": 5.73, + "learning_rate": 3.935838647574295e-06, + "loss": 0.098, + "step": 13682 + }, + { + "epoch": 5.73, + "learning_rate": 3.934490634684114e-06, + "loss": 0.1125, + "step": 13683 + }, + { + "epoch": 5.73, + "learning_rate": 3.933142796141616e-06, + "loss": 0.0992, + "step": 13684 + }, + { + "epoch": 5.73, + "learning_rate": 3.931795131985545e-06, + "loss": 0.149, + "step": 13685 + }, + { + "epoch": 5.73, + "learning_rate": 3.930447642254639e-06, + "loss": 0.1305, + "step": 13686 + }, + { + "epoch": 5.73, + "learning_rate": 3.929100326987629e-06, + "loss": 0.1338, + "step": 13687 + }, + { + "epoch": 5.73, + "learning_rate": 3.927753186223243e-06, + "loss": 0.1654, + "step": 13688 + }, + { + "epoch": 5.73, + "learning_rate": 3.926406220000206e-06, + "loss": 0.1465, + "step": 13689 + }, + { + "epoch": 5.73, + "learning_rate": 3.925059428357231e-06, + "loss": 0.1099, + "step": 13690 + }, + { + "epoch": 5.73, + "learning_rate": 3.923712811333031e-06, + "loss": 0.1108, + "step": 13691 + }, + { + "epoch": 5.73, + "learning_rate": 3.922366368966313e-06, + "loss": 0.1221, + "step": 13692 + }, + { + "epoch": 5.73, + "learning_rate": 3.9210201012957816e-06, + "loss": 0.1164, + "step": 13693 + }, + { + "epoch": 5.73, + "learning_rate": 3.919674008360133e-06, + "loss": 0.1125, + "step": 13694 + }, + { + "epoch": 5.73, + "learning_rate": 3.918328090198063e-06, + "loss": 0.1316, + "step": 13695 + }, + { + "epoch": 5.73, + "learning_rate": 3.91698234684825e-06, + "loss": 0.0516, + "step": 13696 + }, + { + "epoch": 5.73, + "learning_rate": 3.915636778349385e-06, + "loss": 0.1248, + "step": 13697 + }, + { + "epoch": 5.73, + "learning_rate": 3.914291384740139e-06, + "loss": 0.1102, + "step": 13698 + }, + { + "epoch": 5.74, + "learning_rate": 3.9129461660591875e-06, + "loss": 0.1044, + "step": 13699 + }, + { + "epoch": 5.74, + "learning_rate": 3.911601122345197e-06, + "loss": 0.0924, + "step": 13700 + }, + { + "epoch": 5.74, + "learning_rate": 3.910256253636832e-06, + "loss": 0.1286, + "step": 13701 + }, + { + "epoch": 5.74, + "learning_rate": 3.9089115599727446e-06, + "loss": 0.135, + "step": 13702 + }, + { + "epoch": 5.74, + "learning_rate": 3.9075670413915925e-06, + "loss": 0.1467, + "step": 13703 + }, + { + "epoch": 5.74, + "learning_rate": 3.906222697932016e-06, + "loss": 0.0969, + "step": 13704 + }, + { + "epoch": 5.74, + "learning_rate": 3.90487852963266e-06, + "loss": 0.1121, + "step": 13705 + }, + { + "epoch": 5.74, + "learning_rate": 3.903534536532163e-06, + "loss": 0.1398, + "step": 13706 + }, + { + "epoch": 5.74, + "learning_rate": 3.902190718669155e-06, + "loss": 0.1083, + "step": 13707 + }, + { + "epoch": 5.74, + "learning_rate": 3.900847076082264e-06, + "loss": 0.1356, + "step": 13708 + }, + { + "epoch": 5.74, + "learning_rate": 3.899503608810114e-06, + "loss": 0.1313, + "step": 13709 + }, + { + "epoch": 5.74, + "learning_rate": 3.898160316891317e-06, + "loss": 0.0883, + "step": 13710 + }, + { + "epoch": 5.74, + "learning_rate": 3.896817200364488e-06, + "loss": 0.1363, + "step": 13711 + }, + { + "epoch": 5.74, + "learning_rate": 3.895474259268232e-06, + "loss": 0.0811, + "step": 13712 + }, + { + "epoch": 5.74, + "learning_rate": 3.89413149364115e-06, + "loss": 0.1213, + "step": 13713 + }, + { + "epoch": 5.74, + "learning_rate": 3.89278890352184e-06, + "loss": 0.1138, + "step": 13714 + }, + { + "epoch": 5.74, + "learning_rate": 3.891446488948894e-06, + "loss": 0.1276, + "step": 13715 + }, + { + "epoch": 5.74, + "learning_rate": 3.890104249960903e-06, + "loss": 0.0994, + "step": 13716 + }, + { + "epoch": 5.74, + "learning_rate": 3.888762186596439e-06, + "loss": 0.1112, + "step": 13717 + }, + { + "epoch": 5.74, + "learning_rate": 3.887420298894082e-06, + "loss": 0.1057, + "step": 13718 + }, + { + "epoch": 5.74, + "learning_rate": 3.886078586892406e-06, + "loss": 0.1357, + "step": 13719 + }, + { + "epoch": 5.74, + "learning_rate": 3.884737050629975e-06, + "loss": 0.136, + "step": 13720 + }, + { + "epoch": 5.74, + "learning_rate": 3.883395690145351e-06, + "loss": 0.1432, + "step": 13721 + }, + { + "epoch": 5.75, + "learning_rate": 3.882054505477092e-06, + "loss": 0.1476, + "step": 13722 + }, + { + "epoch": 5.75, + "learning_rate": 3.880713496663745e-06, + "loss": 0.1501, + "step": 13723 + }, + { + "epoch": 5.75, + "learning_rate": 3.879372663743862e-06, + "loss": 0.1247, + "step": 13724 + }, + { + "epoch": 5.75, + "learning_rate": 3.878032006755977e-06, + "loss": 0.1415, + "step": 13725 + }, + { + "epoch": 5.75, + "learning_rate": 3.87669152573863e-06, + "loss": 0.123, + "step": 13726 + }, + { + "epoch": 5.75, + "learning_rate": 3.87535122073035e-06, + "loss": 0.1068, + "step": 13727 + }, + { + "epoch": 5.75, + "learning_rate": 3.874011091769665e-06, + "loss": 0.1255, + "step": 13728 + }, + { + "epoch": 5.75, + "learning_rate": 3.872671138895095e-06, + "loss": 0.1134, + "step": 13729 + }, + { + "epoch": 5.75, + "learning_rate": 3.871331362145158e-06, + "loss": 0.0983, + "step": 13730 + }, + { + "epoch": 5.75, + "learning_rate": 3.869991761558362e-06, + "loss": 0.1482, + "step": 13731 + }, + { + "epoch": 5.75, + "learning_rate": 3.868652337173211e-06, + "loss": 0.132, + "step": 13732 + }, + { + "epoch": 5.75, + "learning_rate": 3.867313089028208e-06, + "loss": 0.1277, + "step": 13733 + }, + { + "epoch": 5.75, + "learning_rate": 3.8659740171618485e-06, + "loss": 0.1152, + "step": 13734 + }, + { + "epoch": 5.75, + "learning_rate": 3.864635121612622e-06, + "loss": 0.0775, + "step": 13735 + }, + { + "epoch": 5.75, + "learning_rate": 3.863296402419019e-06, + "loss": 0.1493, + "step": 13736 + }, + { + "epoch": 5.75, + "learning_rate": 3.861957859619511e-06, + "loss": 0.125, + "step": 13737 + }, + { + "epoch": 5.75, + "learning_rate": 3.860619493252578e-06, + "loss": 0.1011, + "step": 13738 + }, + { + "epoch": 5.75, + "learning_rate": 3.859281303356689e-06, + "loss": 0.163, + "step": 13739 + }, + { + "epoch": 5.75, + "learning_rate": 3.857943289970311e-06, + "loss": 0.1182, + "step": 13740 + }, + { + "epoch": 5.75, + "learning_rate": 3.856605453131901e-06, + "loss": 0.1114, + "step": 13741 + }, + { + "epoch": 5.75, + "learning_rate": 3.855267792879921e-06, + "loss": 0.1491, + "step": 13742 + }, + { + "epoch": 5.75, + "learning_rate": 3.85393030925281e-06, + "loss": 0.0939, + "step": 13743 + }, + { + "epoch": 5.75, + "learning_rate": 3.85259300228902e-06, + "loss": 0.0992, + "step": 13744 + }, + { + "epoch": 5.75, + "learning_rate": 3.851255872026989e-06, + "loss": 0.1063, + "step": 13745 + }, + { + "epoch": 5.76, + "learning_rate": 3.849918918505151e-06, + "loss": 0.0433, + "step": 13746 + }, + { + "epoch": 5.76, + "learning_rate": 3.848582141761939e-06, + "loss": 0.1548, + "step": 13747 + }, + { + "epoch": 5.76, + "learning_rate": 3.8472455418357735e-06, + "loss": 0.1037, + "step": 13748 + }, + { + "epoch": 5.76, + "learning_rate": 3.845909118765073e-06, + "loss": 0.0993, + "step": 13749 + }, + { + "epoch": 5.76, + "learning_rate": 3.844572872588258e-06, + "loss": 0.1247, + "step": 13750 + }, + { + "epoch": 5.76, + "learning_rate": 3.843236803343731e-06, + "loss": 0.1247, + "step": 13751 + }, + { + "epoch": 5.76, + "learning_rate": 3.841900911069897e-06, + "loss": 0.1141, + "step": 13752 + }, + { + "epoch": 5.76, + "learning_rate": 3.840565195805158e-06, + "loss": 0.0834, + "step": 13753 + }, + { + "epoch": 5.76, + "learning_rate": 3.839229657587906e-06, + "loss": 0.1405, + "step": 13754 + }, + { + "epoch": 5.76, + "learning_rate": 3.83789429645653e-06, + "loss": 0.1399, + "step": 13755 + }, + { + "epoch": 5.76, + "learning_rate": 3.836559112449415e-06, + "loss": 0.1061, + "step": 13756 + }, + { + "epoch": 5.76, + "learning_rate": 3.835224105604942e-06, + "loss": 0.0806, + "step": 13757 + }, + { + "epoch": 5.76, + "learning_rate": 3.833889275961479e-06, + "loss": 0.1568, + "step": 13758 + }, + { + "epoch": 5.76, + "learning_rate": 3.832554623557397e-06, + "loss": 0.1016, + "step": 13759 + }, + { + "epoch": 5.76, + "learning_rate": 3.831220148431058e-06, + "loss": 0.1125, + "step": 13760 + }, + { + "epoch": 5.76, + "learning_rate": 3.829885850620824e-06, + "loss": 0.0723, + "step": 13761 + }, + { + "epoch": 5.76, + "learning_rate": 3.8285517301650445e-06, + "loss": 0.1175, + "step": 13762 + }, + { + "epoch": 5.76, + "learning_rate": 3.827217787102072e-06, + "loss": 0.1329, + "step": 13763 + }, + { + "epoch": 5.76, + "learning_rate": 3.825884021470244e-06, + "loss": 0.1014, + "step": 13764 + }, + { + "epoch": 5.76, + "learning_rate": 3.8245504333079006e-06, + "loss": 0.0892, + "step": 13765 + }, + { + "epoch": 5.76, + "learning_rate": 3.823217022653376e-06, + "loss": 0.1108, + "step": 13766 + }, + { + "epoch": 5.76, + "learning_rate": 3.8218837895449956e-06, + "loss": 0.1202, + "step": 13767 + }, + { + "epoch": 5.76, + "learning_rate": 3.820550734021083e-06, + "loss": 0.1437, + "step": 13768 + }, + { + "epoch": 5.76, + "learning_rate": 3.819217856119961e-06, + "loss": 0.1, + "step": 13769 + }, + { + "epoch": 5.77, + "learning_rate": 3.817885155879933e-06, + "loss": 0.119, + "step": 13770 + }, + { + "epoch": 5.77, + "learning_rate": 3.816552633339313e-06, + "loss": 0.1255, + "step": 13771 + }, + { + "epoch": 5.77, + "learning_rate": 3.815220288536399e-06, + "loss": 0.1156, + "step": 13772 + }, + { + "epoch": 5.77, + "learning_rate": 3.813888121509487e-06, + "loss": 0.1489, + "step": 13773 + }, + { + "epoch": 5.77, + "learning_rate": 3.812556132296873e-06, + "loss": 0.1184, + "step": 13774 + }, + { + "epoch": 5.77, + "learning_rate": 3.8112243209368415e-06, + "loss": 0.1169, + "step": 13775 + }, + { + "epoch": 5.77, + "learning_rate": 3.8098926874676757e-06, + "loss": 0.1359, + "step": 13776 + }, + { + "epoch": 5.77, + "learning_rate": 3.8085612319276556e-06, + "loss": 0.1257, + "step": 13777 + }, + { + "epoch": 5.77, + "learning_rate": 3.8072299543550438e-06, + "loss": 0.1239, + "step": 13778 + }, + { + "epoch": 5.77, + "learning_rate": 3.8058988547881126e-06, + "loss": 0.0871, + "step": 13779 + }, + { + "epoch": 5.77, + "learning_rate": 3.804567933265122e-06, + "loss": 0.0929, + "step": 13780 + }, + { + "epoch": 5.77, + "learning_rate": 3.803237189824328e-06, + "loss": 0.1104, + "step": 13781 + }, + { + "epoch": 5.77, + "learning_rate": 3.8019066245039828e-06, + "loss": 0.0898, + "step": 13782 + }, + { + "epoch": 5.77, + "learning_rate": 3.800576237342335e-06, + "loss": 0.0989, + "step": 13783 + }, + { + "epoch": 5.77, + "learning_rate": 3.799246028377618e-06, + "loss": 0.1339, + "step": 13784 + }, + { + "epoch": 5.77, + "learning_rate": 3.7979159976480717e-06, + "loss": 0.1075, + "step": 13785 + }, + { + "epoch": 5.77, + "learning_rate": 3.796586145191926e-06, + "loss": 0.1306, + "step": 13786 + }, + { + "epoch": 5.77, + "learning_rate": 3.7952564710474072e-06, + "loss": 0.148, + "step": 13787 + }, + { + "epoch": 5.77, + "learning_rate": 3.793926975252735e-06, + "loss": 0.1047, + "step": 13788 + }, + { + "epoch": 5.77, + "learning_rate": 3.7925976578461255e-06, + "loss": 0.143, + "step": 13789 + }, + { + "epoch": 5.77, + "learning_rate": 3.7912685188657905e-06, + "loss": 0.1201, + "step": 13790 + }, + { + "epoch": 5.77, + "learning_rate": 3.789939558349929e-06, + "loss": 0.0978, + "step": 13791 + }, + { + "epoch": 5.77, + "learning_rate": 3.7886107763367474e-06, + "loss": 0.0969, + "step": 13792 + }, + { + "epoch": 5.77, + "learning_rate": 3.787282172864435e-06, + "loss": 0.1366, + "step": 13793 + }, + { + "epoch": 5.78, + "learning_rate": 3.785953747971183e-06, + "loss": 0.1331, + "step": 13794 + }, + { + "epoch": 5.78, + "learning_rate": 3.784625501695175e-06, + "loss": 0.1114, + "step": 13795 + }, + { + "epoch": 5.78, + "learning_rate": 3.783297434074592e-06, + "loss": 0.1316, + "step": 13796 + }, + { + "epoch": 5.78, + "learning_rate": 3.781969545147609e-06, + "loss": 0.1393, + "step": 13797 + }, + { + "epoch": 5.78, + "learning_rate": 3.7806418349523956e-06, + "loss": 0.1353, + "step": 13798 + }, + { + "epoch": 5.78, + "learning_rate": 3.7793143035271106e-06, + "loss": 0.1607, + "step": 13799 + }, + { + "epoch": 5.78, + "learning_rate": 3.7779869509099166e-06, + "loss": 0.0987, + "step": 13800 + }, + { + "epoch": 5.78, + "learning_rate": 3.776659777138966e-06, + "loss": 0.0866, + "step": 13801 + }, + { + "epoch": 5.78, + "learning_rate": 3.7753327822524067e-06, + "loss": 0.1508, + "step": 13802 + }, + { + "epoch": 5.78, + "learning_rate": 3.7740059662883843e-06, + "loss": 0.1343, + "step": 13803 + }, + { + "epoch": 5.78, + "learning_rate": 3.7726793292850382e-06, + "loss": 0.1392, + "step": 13804 + }, + { + "epoch": 5.78, + "learning_rate": 3.771352871280495e-06, + "loss": 0.1397, + "step": 13805 + }, + { + "epoch": 5.78, + "learning_rate": 3.7700265923128866e-06, + "loss": 0.087, + "step": 13806 + }, + { + "epoch": 5.78, + "learning_rate": 3.768700492420335e-06, + "loss": 0.1246, + "step": 13807 + }, + { + "epoch": 5.78, + "learning_rate": 3.7673745716409595e-06, + "loss": 0.1021, + "step": 13808 + }, + { + "epoch": 5.78, + "learning_rate": 3.7660488300128706e-06, + "loss": 0.1268, + "step": 13809 + }, + { + "epoch": 5.78, + "learning_rate": 3.7647232675741794e-06, + "loss": 0.0973, + "step": 13810 + }, + { + "epoch": 5.78, + "learning_rate": 3.7633978843629813e-06, + "loss": 0.0975, + "step": 13811 + }, + { + "epoch": 5.78, + "learning_rate": 3.7620726804173767e-06, + "loss": 0.1035, + "step": 13812 + }, + { + "epoch": 5.78, + "learning_rate": 3.7607476557754574e-06, + "loss": 0.1264, + "step": 13813 + }, + { + "epoch": 5.78, + "learning_rate": 3.7594228104753107e-06, + "loss": 0.1345, + "step": 13814 + }, + { + "epoch": 5.78, + "learning_rate": 3.75809814455502e-06, + "loss": 0.1187, + "step": 13815 + }, + { + "epoch": 5.78, + "learning_rate": 3.756773658052657e-06, + "loss": 0.1339, + "step": 13816 + }, + { + "epoch": 5.78, + "learning_rate": 3.755449351006294e-06, + "loss": 0.1081, + "step": 13817 + }, + { + "epoch": 5.79, + "learning_rate": 3.7541252234540028e-06, + "loss": 0.1106, + "step": 13818 + }, + { + "epoch": 5.79, + "learning_rate": 3.752801275433835e-06, + "loss": 0.1587, + "step": 13819 + }, + { + "epoch": 5.79, + "learning_rate": 3.7514775069838503e-06, + "loss": 0.1327, + "step": 13820 + }, + { + "epoch": 5.79, + "learning_rate": 3.750153918142101e-06, + "loss": 0.1187, + "step": 13821 + }, + { + "epoch": 5.79, + "learning_rate": 3.74883050894663e-06, + "loss": 0.1307, + "step": 13822 + }, + { + "epoch": 5.79, + "learning_rate": 3.7475072794354803e-06, + "loss": 0.0944, + "step": 13823 + }, + { + "epoch": 5.79, + "learning_rate": 3.746184229646689e-06, + "loss": 0.0824, + "step": 13824 + }, + { + "epoch": 5.79, + "learning_rate": 3.744861359618278e-06, + "loss": 0.1315, + "step": 13825 + }, + { + "epoch": 5.79, + "learning_rate": 3.7435386693882767e-06, + "loss": 0.1121, + "step": 13826 + }, + { + "epoch": 5.79, + "learning_rate": 3.742216158994705e-06, + "loss": 0.0785, + "step": 13827 + }, + { + "epoch": 5.79, + "learning_rate": 3.740893828475577e-06, + "loss": 0.1419, + "step": 13828 + }, + { + "epoch": 5.79, + "learning_rate": 3.7395716778689006e-06, + "loss": 0.1205, + "step": 13829 + }, + { + "epoch": 5.79, + "learning_rate": 3.738249707212681e-06, + "loss": 0.1453, + "step": 13830 + }, + { + "epoch": 5.79, + "learning_rate": 3.736927916544921e-06, + "loss": 0.129, + "step": 13831 + }, + { + "epoch": 5.79, + "learning_rate": 3.7356063059036073e-06, + "loss": 0.1384, + "step": 13832 + }, + { + "epoch": 5.79, + "learning_rate": 3.73428487532673e-06, + "loss": 0.1283, + "step": 13833 + }, + { + "epoch": 5.79, + "learning_rate": 3.732963624852275e-06, + "loss": 0.1252, + "step": 13834 + }, + { + "epoch": 5.79, + "learning_rate": 3.731642554518219e-06, + "loss": 0.1, + "step": 13835 + }, + { + "epoch": 5.79, + "learning_rate": 3.7303216643625364e-06, + "loss": 0.1462, + "step": 13836 + }, + { + "epoch": 5.79, + "learning_rate": 3.7290009544231955e-06, + "loss": 0.1298, + "step": 13837 + }, + { + "epoch": 5.79, + "learning_rate": 3.727680424738155e-06, + "loss": 0.1198, + "step": 13838 + }, + { + "epoch": 5.79, + "learning_rate": 3.726360075345378e-06, + "loss": 0.1026, + "step": 13839 + }, + { + "epoch": 5.79, + "learning_rate": 3.7250399062828104e-06, + "loss": 0.1226, + "step": 13840 + }, + { + "epoch": 5.79, + "learning_rate": 3.723719917588402e-06, + "loss": 0.1262, + "step": 13841 + }, + { + "epoch": 5.8, + "learning_rate": 3.722400109300095e-06, + "loss": 0.1533, + "step": 13842 + }, + { + "epoch": 5.8, + "learning_rate": 3.7210804814558253e-06, + "loss": 0.1316, + "step": 13843 + }, + { + "epoch": 5.8, + "learning_rate": 3.7197610340935263e-06, + "loss": 0.1205, + "step": 13844 + }, + { + "epoch": 5.8, + "learning_rate": 3.7184417672511273e-06, + "loss": 0.0987, + "step": 13845 + }, + { + "epoch": 5.8, + "learning_rate": 3.7171226809665417e-06, + "loss": 0.1437, + "step": 13846 + }, + { + "epoch": 5.8, + "learning_rate": 3.71580377527769e-06, + "loss": 0.1369, + "step": 13847 + }, + { + "epoch": 5.8, + "learning_rate": 3.7144850502224816e-06, + "loss": 0.135, + "step": 13848 + }, + { + "epoch": 5.8, + "learning_rate": 3.713166505838822e-06, + "loss": 0.102, + "step": 13849 + }, + { + "epoch": 5.8, + "learning_rate": 3.7118481421646134e-06, + "loss": 0.0825, + "step": 13850 + }, + { + "epoch": 5.8, + "learning_rate": 3.710529959237753e-06, + "loss": 0.0986, + "step": 13851 + }, + { + "epoch": 5.8, + "learning_rate": 3.7092119570961248e-06, + "loss": 0.1431, + "step": 13852 + }, + { + "epoch": 5.8, + "learning_rate": 3.7078941357776153e-06, + "loss": 0.0948, + "step": 13853 + }, + { + "epoch": 5.8, + "learning_rate": 3.7065764953201066e-06, + "loss": 0.1081, + "step": 13854 + }, + { + "epoch": 5.8, + "learning_rate": 3.7052590357614717e-06, + "loss": 0.0667, + "step": 13855 + }, + { + "epoch": 5.8, + "learning_rate": 3.7039417571395796e-06, + "loss": 0.1089, + "step": 13856 + }, + { + "epoch": 5.8, + "learning_rate": 3.7026246594922944e-06, + "loss": 0.1096, + "step": 13857 + }, + { + "epoch": 5.8, + "learning_rate": 3.7013077428574794e-06, + "loss": 0.1008, + "step": 13858 + }, + { + "epoch": 5.8, + "learning_rate": 3.6999910072729808e-06, + "loss": 0.1199, + "step": 13859 + }, + { + "epoch": 5.8, + "learning_rate": 3.698674452776653e-06, + "loss": 0.1097, + "step": 13860 + }, + { + "epoch": 5.8, + "learning_rate": 3.6973580794063323e-06, + "loss": 0.1472, + "step": 13861 + }, + { + "epoch": 5.8, + "learning_rate": 3.6960418871998615e-06, + "loss": 0.1396, + "step": 13862 + }, + { + "epoch": 5.8, + "learning_rate": 3.6947258761950723e-06, + "loss": 0.1408, + "step": 13863 + }, + { + "epoch": 5.8, + "learning_rate": 3.693410046429793e-06, + "loss": 0.0973, + "step": 13864 + }, + { + "epoch": 5.8, + "learning_rate": 3.6920943979418478e-06, + "loss": 0.1614, + "step": 13865 + }, + { + "epoch": 5.81, + "learning_rate": 3.690778930769049e-06, + "loss": 0.1382, + "step": 13866 + }, + { + "epoch": 5.81, + "learning_rate": 3.6894636449492116e-06, + "loss": 0.1073, + "step": 13867 + }, + { + "epoch": 5.81, + "learning_rate": 3.688148540520141e-06, + "loss": 0.1038, + "step": 13868 + }, + { + "epoch": 5.81, + "learning_rate": 3.6868336175196406e-06, + "loss": 0.1121, + "step": 13869 + }, + { + "epoch": 5.81, + "learning_rate": 3.6855188759855043e-06, + "loss": 0.125, + "step": 13870 + }, + { + "epoch": 5.81, + "learning_rate": 3.6842043159555263e-06, + "loss": 0.1256, + "step": 13871 + }, + { + "epoch": 5.81, + "learning_rate": 3.6828899374674933e-06, + "loss": 0.0791, + "step": 13872 + }, + { + "epoch": 5.81, + "learning_rate": 3.681575740559179e-06, + "loss": 0.1311, + "step": 13873 + }, + { + "epoch": 5.81, + "learning_rate": 3.680261725268364e-06, + "loss": 0.122, + "step": 13874 + }, + { + "epoch": 5.81, + "learning_rate": 3.678947891632818e-06, + "loss": 0.1136, + "step": 13875 + }, + { + "epoch": 5.81, + "learning_rate": 3.677634239690305e-06, + "loss": 0.0916, + "step": 13876 + }, + { + "epoch": 5.81, + "learning_rate": 3.676320769478585e-06, + "loss": 0.1307, + "step": 13877 + }, + { + "epoch": 5.81, + "learning_rate": 3.6750074810354173e-06, + "loss": 0.1332, + "step": 13878 + }, + { + "epoch": 5.81, + "learning_rate": 3.673694374398543e-06, + "loss": 0.1622, + "step": 13879 + }, + { + "epoch": 5.81, + "learning_rate": 3.672381449605711e-06, + "loss": 0.1185, + "step": 13880 + }, + { + "epoch": 5.81, + "learning_rate": 3.671068706694658e-06, + "loss": 0.1075, + "step": 13881 + }, + { + "epoch": 5.81, + "learning_rate": 3.6697561457031196e-06, + "loss": 0.0994, + "step": 13882 + }, + { + "epoch": 5.81, + "learning_rate": 3.6684437666688253e-06, + "loss": 0.1085, + "step": 13883 + }, + { + "epoch": 5.81, + "learning_rate": 3.667131569629495e-06, + "loss": 0.1145, + "step": 13884 + }, + { + "epoch": 5.81, + "learning_rate": 3.665819554622847e-06, + "loss": 0.1068, + "step": 13885 + }, + { + "epoch": 5.81, + "learning_rate": 3.6645077216865987e-06, + "loss": 0.0967, + "step": 13886 + }, + { + "epoch": 5.81, + "learning_rate": 3.6631960708584504e-06, + "loss": 0.1445, + "step": 13887 + }, + { + "epoch": 5.81, + "learning_rate": 3.6618846021761078e-06, + "loss": 0.1084, + "step": 13888 + }, + { + "epoch": 5.81, + "learning_rate": 3.660573315677268e-06, + "loss": 0.099, + "step": 13889 + }, + { + "epoch": 5.82, + "learning_rate": 3.6592622113996235e-06, + "loss": 0.1187, + "step": 13890 + }, + { + "epoch": 5.82, + "learning_rate": 3.657951289380859e-06, + "loss": 0.0873, + "step": 13891 + }, + { + "epoch": 5.82, + "learning_rate": 3.656640549658661e-06, + "loss": 0.1233, + "step": 13892 + }, + { + "epoch": 5.82, + "learning_rate": 3.6553299922706977e-06, + "loss": 0.0894, + "step": 13893 + }, + { + "epoch": 5.82, + "learning_rate": 3.654019617254644e-06, + "loss": 0.0926, + "step": 13894 + }, + { + "epoch": 5.82, + "learning_rate": 3.652709424648164e-06, + "loss": 0.0958, + "step": 13895 + }, + { + "epoch": 5.82, + "learning_rate": 3.65139941448892e-06, + "loss": 0.1193, + "step": 13896 + }, + { + "epoch": 5.82, + "learning_rate": 3.650089586814566e-06, + "loss": 0.1366, + "step": 13897 + }, + { + "epoch": 5.82, + "learning_rate": 3.648779941662752e-06, + "loss": 0.124, + "step": 13898 + }, + { + "epoch": 5.82, + "learning_rate": 3.6474704790711258e-06, + "loss": 0.1464, + "step": 13899 + }, + { + "epoch": 5.82, + "learning_rate": 3.6461611990773204e-06, + "loss": 0.1439, + "step": 13900 + }, + { + "epoch": 5.82, + "learning_rate": 3.6448521017189732e-06, + "loss": 0.1262, + "step": 13901 + }, + { + "epoch": 5.82, + "learning_rate": 3.6435431870337123e-06, + "loss": 0.1203, + "step": 13902 + }, + { + "epoch": 5.82, + "learning_rate": 3.6422344550591625e-06, + "loss": 0.1148, + "step": 13903 + }, + { + "epoch": 5.82, + "learning_rate": 3.6409259058329404e-06, + "loss": 0.082, + "step": 13904 + }, + { + "epoch": 5.82, + "learning_rate": 3.6396175393926646e-06, + "loss": 0.095, + "step": 13905 + }, + { + "epoch": 5.82, + "learning_rate": 3.6383093557759386e-06, + "loss": 0.1278, + "step": 13906 + }, + { + "epoch": 5.82, + "learning_rate": 3.637001355020361e-06, + "loss": 0.1405, + "step": 13907 + }, + { + "epoch": 5.82, + "learning_rate": 3.6356935371635336e-06, + "loss": 0.0856, + "step": 13908 + }, + { + "epoch": 5.82, + "learning_rate": 3.6343859022430474e-06, + "loss": 0.1269, + "step": 13909 + }, + { + "epoch": 5.82, + "learning_rate": 3.63307845029649e-06, + "loss": 0.1347, + "step": 13910 + }, + { + "epoch": 5.82, + "learning_rate": 3.631771181361442e-06, + "loss": 0.117, + "step": 13911 + }, + { + "epoch": 5.82, + "learning_rate": 3.630464095475481e-06, + "loss": 0.1103, + "step": 13912 + }, + { + "epoch": 5.82, + "learning_rate": 3.6291571926761805e-06, + "loss": 0.1385, + "step": 13913 + }, + { + "epoch": 5.83, + "learning_rate": 3.6278504730011e-06, + "loss": 0.1068, + "step": 13914 + }, + { + "epoch": 5.83, + "learning_rate": 3.626543936487803e-06, + "loss": 0.0955, + "step": 13915 + }, + { + "epoch": 5.83, + "learning_rate": 3.625237583173844e-06, + "loss": 0.0907, + "step": 13916 + }, + { + "epoch": 5.83, + "learning_rate": 3.6239314130967753e-06, + "loss": 0.1233, + "step": 13917 + }, + { + "epoch": 5.83, + "learning_rate": 3.6226254262941398e-06, + "loss": 0.1122, + "step": 13918 + }, + { + "epoch": 5.83, + "learning_rate": 3.6213196228034796e-06, + "loss": 0.1352, + "step": 13919 + }, + { + "epoch": 5.83, + "learning_rate": 3.6200140026623233e-06, + "loss": 0.1447, + "step": 13920 + }, + { + "epoch": 5.83, + "learning_rate": 3.6187085659082034e-06, + "loss": 0.1134, + "step": 13921 + }, + { + "epoch": 5.83, + "learning_rate": 3.6174033125786434e-06, + "loss": 0.1147, + "step": 13922 + }, + { + "epoch": 5.83, + "learning_rate": 3.6160982427111614e-06, + "loss": 0.0617, + "step": 13923 + }, + { + "epoch": 5.83, + "learning_rate": 3.614793356343269e-06, + "loss": 0.1265, + "step": 13924 + }, + { + "epoch": 5.83, + "learning_rate": 3.6134886535124804e-06, + "loss": 0.0609, + "step": 13925 + }, + { + "epoch": 5.83, + "learning_rate": 3.612184134256289e-06, + "loss": 0.1177, + "step": 13926 + }, + { + "epoch": 5.83, + "learning_rate": 3.6108797986121967e-06, + "loss": 0.1149, + "step": 13927 + }, + { + "epoch": 5.83, + "learning_rate": 3.609575646617699e-06, + "loss": 0.1173, + "step": 13928 + }, + { + "epoch": 5.83, + "learning_rate": 3.6082716783102735e-06, + "loss": 0.0993, + "step": 13929 + }, + { + "epoch": 5.83, + "learning_rate": 3.6069678937274076e-06, + "loss": 0.1415, + "step": 13930 + }, + { + "epoch": 5.83, + "learning_rate": 3.6056642929065765e-06, + "loss": 0.088, + "step": 13931 + }, + { + "epoch": 5.83, + "learning_rate": 3.604360875885251e-06, + "loss": 0.1072, + "step": 13932 + }, + { + "epoch": 5.83, + "learning_rate": 3.6030576427009e-06, + "loss": 0.131, + "step": 13933 + }, + { + "epoch": 5.83, + "learning_rate": 3.601754593390978e-06, + "loss": 0.1314, + "step": 13934 + }, + { + "epoch": 5.83, + "learning_rate": 3.600451727992942e-06, + "loss": 0.1126, + "step": 13935 + }, + { + "epoch": 5.83, + "learning_rate": 3.5991490465442413e-06, + "loss": 0.1463, + "step": 13936 + }, + { + "epoch": 5.84, + "learning_rate": 3.597846549082322e-06, + "loss": 0.1579, + "step": 13937 + }, + { + "epoch": 5.84, + "learning_rate": 3.5965442356446224e-06, + "loss": 0.0985, + "step": 13938 + }, + { + "epoch": 5.84, + "learning_rate": 3.595242106268577e-06, + "loss": 0.1163, + "step": 13939 + }, + { + "epoch": 5.84, + "learning_rate": 3.593940160991617e-06, + "loss": 0.1439, + "step": 13940 + }, + { + "epoch": 5.84, + "learning_rate": 3.592638399851158e-06, + "loss": 0.1152, + "step": 13941 + }, + { + "epoch": 5.84, + "learning_rate": 3.5913368228846236e-06, + "loss": 0.1404, + "step": 13942 + }, + { + "epoch": 5.84, + "learning_rate": 3.5900354301294236e-06, + "loss": 0.1272, + "step": 13943 + }, + { + "epoch": 5.84, + "learning_rate": 3.5887342216229694e-06, + "loss": 0.1131, + "step": 13944 + }, + { + "epoch": 5.84, + "learning_rate": 3.587433197402659e-06, + "loss": 0.1285, + "step": 13945 + }, + { + "epoch": 5.84, + "learning_rate": 3.586132357505895e-06, + "loss": 0.1054, + "step": 13946 + }, + { + "epoch": 5.84, + "learning_rate": 3.5848317019700608e-06, + "loss": 0.09, + "step": 13947 + }, + { + "epoch": 5.84, + "learning_rate": 3.5835312308325475e-06, + "loss": 0.1308, + "step": 13948 + }, + { + "epoch": 5.84, + "learning_rate": 3.5822309441307347e-06, + "loss": 0.1002, + "step": 13949 + }, + { + "epoch": 5.84, + "learning_rate": 3.5809308419019985e-06, + "loss": 0.0906, + "step": 13950 + }, + { + "epoch": 5.84, + "learning_rate": 3.5796309241837125e-06, + "loss": 0.1281, + "step": 13951 + }, + { + "epoch": 5.84, + "learning_rate": 3.5783311910132348e-06, + "loss": 0.1261, + "step": 13952 + }, + { + "epoch": 5.84, + "learning_rate": 3.5770316424279283e-06, + "loss": 0.1357, + "step": 13953 + }, + { + "epoch": 5.84, + "learning_rate": 3.5757322784651506e-06, + "loss": 0.1148, + "step": 13954 + }, + { + "epoch": 5.84, + "learning_rate": 3.5744330991622457e-06, + "loss": 0.0935, + "step": 13955 + }, + { + "epoch": 5.84, + "learning_rate": 3.573134104556558e-06, + "loss": 0.0908, + "step": 13956 + }, + { + "epoch": 5.84, + "learning_rate": 3.5718352946854286e-06, + "loss": 0.1236, + "step": 13957 + }, + { + "epoch": 5.84, + "learning_rate": 3.5705366695861886e-06, + "loss": 0.0759, + "step": 13958 + }, + { + "epoch": 5.84, + "learning_rate": 3.569238229296167e-06, + "loss": 0.1156, + "step": 13959 + }, + { + "epoch": 5.84, + "learning_rate": 3.567939973852689e-06, + "loss": 0.0935, + "step": 13960 + }, + { + "epoch": 5.85, + "learning_rate": 3.5666419032930653e-06, + "loss": 0.1497, + "step": 13961 + }, + { + "epoch": 5.85, + "learning_rate": 3.5653440176546115e-06, + "loss": 0.0916, + "step": 13962 + }, + { + "epoch": 5.85, + "learning_rate": 3.564046316974634e-06, + "loss": 0.0757, + "step": 13963 + }, + { + "epoch": 5.85, + "learning_rate": 3.5627488012904333e-06, + "loss": 0.0908, + "step": 13964 + }, + { + "epoch": 5.85, + "learning_rate": 3.5614514706393066e-06, + "loss": 0.1181, + "step": 13965 + }, + { + "epoch": 5.85, + "learning_rate": 3.560154325058547e-06, + "loss": 0.0886, + "step": 13966 + }, + { + "epoch": 5.85, + "learning_rate": 3.558857364585434e-06, + "loss": 0.1648, + "step": 13967 + }, + { + "epoch": 5.85, + "learning_rate": 3.5575605892572495e-06, + "loss": 0.0831, + "step": 13968 + }, + { + "epoch": 5.85, + "learning_rate": 3.556263999111269e-06, + "loss": 0.1066, + "step": 13969 + }, + { + "epoch": 5.85, + "learning_rate": 3.554967594184762e-06, + "loss": 0.1088, + "step": 13970 + }, + { + "epoch": 5.85, + "learning_rate": 3.553671374514992e-06, + "loss": 0.1059, + "step": 13971 + }, + { + "epoch": 5.85, + "learning_rate": 3.552375340139219e-06, + "loss": 0.1399, + "step": 13972 + }, + { + "epoch": 5.85, + "learning_rate": 3.551079491094698e-06, + "loss": 0.1216, + "step": 13973 + }, + { + "epoch": 5.85, + "learning_rate": 3.5497838274186737e-06, + "loss": 0.151, + "step": 13974 + }, + { + "epoch": 5.85, + "learning_rate": 3.5484883491483866e-06, + "loss": 0.1209, + "step": 13975 + }, + { + "epoch": 5.85, + "learning_rate": 3.5471930563210775e-06, + "loss": 0.14, + "step": 13976 + }, + { + "epoch": 5.85, + "learning_rate": 3.5458979489739778e-06, + "loss": 0.1103, + "step": 13977 + }, + { + "epoch": 5.85, + "learning_rate": 3.5446030271443132e-06, + "loss": 0.1041, + "step": 13978 + }, + { + "epoch": 5.85, + "learning_rate": 3.543308290869307e-06, + "loss": 0.0979, + "step": 13979 + }, + { + "epoch": 5.85, + "learning_rate": 3.5420137401861745e-06, + "loss": 0.1379, + "step": 13980 + }, + { + "epoch": 5.85, + "learning_rate": 3.540719375132129e-06, + "loss": 0.089, + "step": 13981 + }, + { + "epoch": 5.85, + "learning_rate": 3.5394251957443704e-06, + "loss": 0.1059, + "step": 13982 + }, + { + "epoch": 5.85, + "learning_rate": 3.538131202060102e-06, + "loss": 0.1206, + "step": 13983 + }, + { + "epoch": 5.85, + "learning_rate": 3.5368373941165167e-06, + "loss": 0.117, + "step": 13984 + }, + { + "epoch": 5.86, + "learning_rate": 3.5355437719508067e-06, + "loss": 0.0811, + "step": 13985 + }, + { + "epoch": 5.86, + "learning_rate": 3.5342503356001535e-06, + "loss": 0.1076, + "step": 13986 + }, + { + "epoch": 5.86, + "learning_rate": 3.5329570851017405e-06, + "loss": 0.112, + "step": 13987 + }, + { + "epoch": 5.86, + "learning_rate": 3.5316640204927342e-06, + "loss": 0.1049, + "step": 13988 + }, + { + "epoch": 5.86, + "learning_rate": 3.5303711418103047e-06, + "loss": 0.1239, + "step": 13989 + }, + { + "epoch": 5.86, + "learning_rate": 3.5290784490916175e-06, + "loss": 0.139, + "step": 13990 + }, + { + "epoch": 5.86, + "learning_rate": 3.527785942373827e-06, + "loss": 0.1611, + "step": 13991 + }, + { + "epoch": 5.86, + "learning_rate": 3.526493621694086e-06, + "loss": 0.1448, + "step": 13992 + }, + { + "epoch": 5.86, + "learning_rate": 3.5252014870895447e-06, + "loss": 0.1252, + "step": 13993 + }, + { + "epoch": 5.86, + "learning_rate": 3.5239095385973377e-06, + "loss": 0.1505, + "step": 13994 + }, + { + "epoch": 5.86, + "learning_rate": 3.522617776254604e-06, + "loss": 0.1341, + "step": 13995 + }, + { + "epoch": 5.86, + "learning_rate": 3.5213262000984793e-06, + "loss": 0.1043, + "step": 13996 + }, + { + "epoch": 5.86, + "learning_rate": 3.52003481016608e-06, + "loss": 0.1103, + "step": 13997 + }, + { + "epoch": 5.86, + "learning_rate": 3.518743606494529e-06, + "loss": 0.1107, + "step": 13998 + }, + { + "epoch": 5.86, + "learning_rate": 3.517452589120941e-06, + "loss": 0.0967, + "step": 13999 + }, + { + "epoch": 5.86, + "learning_rate": 3.516161758082427e-06, + "loss": 0.1111, + "step": 14000 + }, + { + "epoch": 5.86, + "learning_rate": 3.514871113416093e-06, + "loss": 0.0789, + "step": 14001 + }, + { + "epoch": 5.86, + "learning_rate": 3.51358065515903e-06, + "loss": 0.1114, + "step": 14002 + }, + { + "epoch": 5.86, + "learning_rate": 3.5122903833483357e-06, + "loss": 0.1079, + "step": 14003 + }, + { + "epoch": 5.86, + "learning_rate": 3.511000298021098e-06, + "loss": 0.118, + "step": 14004 + }, + { + "epoch": 5.86, + "learning_rate": 3.5097103992143978e-06, + "loss": 0.1236, + "step": 14005 + }, + { + "epoch": 5.86, + "learning_rate": 3.5084206869653126e-06, + "loss": 0.1465, + "step": 14006 + }, + { + "epoch": 5.86, + "learning_rate": 3.507131161310917e-06, + "loss": 0.1022, + "step": 14007 + }, + { + "epoch": 5.86, + "learning_rate": 3.505841822288272e-06, + "loss": 0.0969, + "step": 14008 + }, + { + "epoch": 5.87, + "learning_rate": 3.5045526699344413e-06, + "loss": 0.1046, + "step": 14009 + }, + { + "epoch": 5.87, + "learning_rate": 3.5032637042864793e-06, + "loss": 0.1009, + "step": 14010 + }, + { + "epoch": 5.87, + "learning_rate": 3.5019749253814372e-06, + "loss": 0.1093, + "step": 14011 + }, + { + "epoch": 5.87, + "learning_rate": 3.50068633325636e-06, + "loss": 0.1327, + "step": 14012 + }, + { + "epoch": 5.87, + "learning_rate": 3.499397927948286e-06, + "loss": 0.0909, + "step": 14013 + }, + { + "epoch": 5.87, + "learning_rate": 3.498109709494254e-06, + "loss": 0.1392, + "step": 14014 + }, + { + "epoch": 5.87, + "learning_rate": 3.496821677931286e-06, + "loss": 0.1085, + "step": 14015 + }, + { + "epoch": 5.87, + "learning_rate": 3.495533833296406e-06, + "loss": 0.1361, + "step": 14016 + }, + { + "epoch": 5.87, + "learning_rate": 3.4942461756266356e-06, + "loss": 0.1297, + "step": 14017 + }, + { + "epoch": 5.87, + "learning_rate": 3.4929587049589853e-06, + "loss": 0.1175, + "step": 14018 + }, + { + "epoch": 5.87, + "learning_rate": 3.491671421330467e-06, + "loss": 0.0825, + "step": 14019 + }, + { + "epoch": 5.87, + "learning_rate": 3.4903843247780743e-06, + "loss": 0.1452, + "step": 14020 + }, + { + "epoch": 5.87, + "learning_rate": 3.489097415338808e-06, + "loss": 0.1271, + "step": 14021 + }, + { + "epoch": 5.87, + "learning_rate": 3.4878106930496616e-06, + "loss": 0.1284, + "step": 14022 + }, + { + "epoch": 5.87, + "learning_rate": 3.486524157947615e-06, + "loss": 0.0917, + "step": 14023 + }, + { + "epoch": 5.87, + "learning_rate": 3.4852378100696505e-06, + "loss": 0.1334, + "step": 14024 + }, + { + "epoch": 5.87, + "learning_rate": 3.4839516494527457e-06, + "loss": 0.1086, + "step": 14025 + }, + { + "epoch": 5.87, + "learning_rate": 3.4826656761338673e-06, + "loss": 0.1463, + "step": 14026 + }, + { + "epoch": 5.87, + "learning_rate": 3.481379890149982e-06, + "loss": 0.1149, + "step": 14027 + }, + { + "epoch": 5.87, + "learning_rate": 3.4800942915380485e-06, + "loss": 0.127, + "step": 14028 + }, + { + "epoch": 5.87, + "learning_rate": 3.4788088803350173e-06, + "loss": 0.111, + "step": 14029 + }, + { + "epoch": 5.87, + "learning_rate": 3.477523656577837e-06, + "loss": 0.1165, + "step": 14030 + }, + { + "epoch": 5.87, + "learning_rate": 3.4762386203034527e-06, + "loss": 0.1217, + "step": 14031 + }, + { + "epoch": 5.87, + "learning_rate": 3.474953771548799e-06, + "loss": 0.1365, + "step": 14032 + }, + { + "epoch": 5.88, + "learning_rate": 3.473669110350809e-06, + "loss": 0.1499, + "step": 14033 + }, + { + "epoch": 5.88, + "learning_rate": 3.472384636746412e-06, + "loss": 0.1611, + "step": 14034 + }, + { + "epoch": 5.88, + "learning_rate": 3.471100350772524e-06, + "loss": 0.0953, + "step": 14035 + }, + { + "epoch": 5.88, + "learning_rate": 3.4698162524660615e-06, + "loss": 0.1255, + "step": 14036 + }, + { + "epoch": 5.88, + "learning_rate": 3.468532341863937e-06, + "loss": 0.1541, + "step": 14037 + }, + { + "epoch": 5.88, + "learning_rate": 3.4672486190030543e-06, + "loss": 0.1318, + "step": 14038 + }, + { + "epoch": 5.88, + "learning_rate": 3.4659650839203127e-06, + "loss": 0.1182, + "step": 14039 + }, + { + "epoch": 5.88, + "learning_rate": 3.4646817366526066e-06, + "loss": 0.1264, + "step": 14040 + }, + { + "epoch": 5.88, + "learning_rate": 3.4633985772368274e-06, + "loss": 0.1143, + "step": 14041 + }, + { + "epoch": 5.88, + "learning_rate": 3.4621156057098573e-06, + "loss": 0.1463, + "step": 14042 + }, + { + "epoch": 5.88, + "learning_rate": 3.4608328221085674e-06, + "loss": 0.1425, + "step": 14043 + }, + { + "epoch": 5.88, + "learning_rate": 3.4595502264698356e-06, + "loss": 0.1706, + "step": 14044 + }, + { + "epoch": 5.88, + "learning_rate": 3.4582678188305297e-06, + "loss": 0.1091, + "step": 14045 + }, + { + "epoch": 5.88, + "learning_rate": 3.4569855992275093e-06, + "loss": 0.1409, + "step": 14046 + }, + { + "epoch": 5.88, + "learning_rate": 3.4557035676976313e-06, + "loss": 0.0968, + "step": 14047 + }, + { + "epoch": 5.88, + "learning_rate": 3.454421724277751e-06, + "loss": 0.1727, + "step": 14048 + }, + { + "epoch": 5.88, + "learning_rate": 3.453140069004707e-06, + "loss": 0.1531, + "step": 14049 + }, + { + "epoch": 5.88, + "learning_rate": 3.451858601915341e-06, + "loss": 0.121, + "step": 14050 + }, + { + "epoch": 5.88, + "learning_rate": 3.45057732304649e-06, + "loss": 0.1237, + "step": 14051 + }, + { + "epoch": 5.88, + "learning_rate": 3.449296232434981e-06, + "loss": 0.0952, + "step": 14052 + }, + { + "epoch": 5.88, + "learning_rate": 3.448015330117639e-06, + "loss": 0.1005, + "step": 14053 + }, + { + "epoch": 5.88, + "learning_rate": 3.446734616131282e-06, + "loss": 0.1103, + "step": 14054 + }, + { + "epoch": 5.88, + "learning_rate": 3.4454540905127277e-06, + "loss": 0.1124, + "step": 14055 + }, + { + "epoch": 5.88, + "learning_rate": 3.444173753298775e-06, + "loss": 0.114, + "step": 14056 + }, + { + "epoch": 5.89, + "learning_rate": 3.4428936045262296e-06, + "loss": 0.0952, + "step": 14057 + }, + { + "epoch": 5.89, + "learning_rate": 3.4416136442318894e-06, + "loss": 0.1132, + "step": 14058 + }, + { + "epoch": 5.89, + "learning_rate": 3.440333872452546e-06, + "loss": 0.1363, + "step": 14059 + }, + { + "epoch": 5.89, + "learning_rate": 3.4390542892249833e-06, + "loss": 0.0637, + "step": 14060 + }, + { + "epoch": 5.89, + "learning_rate": 3.437774894585987e-06, + "loss": 0.1097, + "step": 14061 + }, + { + "epoch": 5.89, + "learning_rate": 3.4364956885723246e-06, + "loss": 0.1029, + "step": 14062 + }, + { + "epoch": 5.89, + "learning_rate": 3.435216671220769e-06, + "loss": 0.1439, + "step": 14063 + }, + { + "epoch": 5.89, + "learning_rate": 3.433937842568088e-06, + "loss": 0.1027, + "step": 14064 + }, + { + "epoch": 5.89, + "learning_rate": 3.432659202651034e-06, + "loss": 0.0999, + "step": 14065 + }, + { + "epoch": 5.89, + "learning_rate": 3.4313807515063635e-06, + "loss": 0.1, + "step": 14066 + }, + { + "epoch": 5.89, + "learning_rate": 3.4301024891708245e-06, + "loss": 0.1165, + "step": 14067 + }, + { + "epoch": 5.89, + "learning_rate": 3.42882441568116e-06, + "loss": 0.1132, + "step": 14068 + }, + { + "epoch": 5.89, + "learning_rate": 3.4275465310741083e-06, + "loss": 0.1187, + "step": 14069 + }, + { + "epoch": 5.89, + "learning_rate": 3.426268835386397e-06, + "loss": 0.0904, + "step": 14070 + }, + { + "epoch": 5.89, + "learning_rate": 3.4249913286547553e-06, + "loss": 0.1018, + "step": 14071 + }, + { + "epoch": 5.89, + "learning_rate": 3.4237140109159015e-06, + "loss": 0.1082, + "step": 14072 + }, + { + "epoch": 5.89, + "learning_rate": 3.4224368822065547e-06, + "loss": 0.1081, + "step": 14073 + }, + { + "epoch": 5.89, + "learning_rate": 3.421159942563421e-06, + "loss": 0.1229, + "step": 14074 + }, + { + "epoch": 5.89, + "learning_rate": 3.4198831920232113e-06, + "loss": 0.0917, + "step": 14075 + }, + { + "epoch": 5.89, + "learning_rate": 3.4186066306226174e-06, + "loss": 0.0752, + "step": 14076 + }, + { + "epoch": 5.89, + "learning_rate": 3.4173302583983335e-06, + "loss": 0.1235, + "step": 14077 + }, + { + "epoch": 5.89, + "learning_rate": 3.416054075387052e-06, + "loss": 0.0999, + "step": 14078 + }, + { + "epoch": 5.89, + "learning_rate": 3.414778081625453e-06, + "loss": 0.1152, + "step": 14079 + }, + { + "epoch": 5.89, + "learning_rate": 3.4135022771502146e-06, + "loss": 0.0914, + "step": 14080 + }, + { + "epoch": 5.9, + "learning_rate": 3.4122266619980083e-06, + "loss": 0.0757, + "step": 14081 + }, + { + "epoch": 5.9, + "learning_rate": 3.410951236205503e-06, + "loss": 0.0896, + "step": 14082 + }, + { + "epoch": 5.9, + "learning_rate": 3.4096759998093552e-06, + "loss": 0.1502, + "step": 14083 + }, + { + "epoch": 5.9, + "learning_rate": 3.4084009528462227e-06, + "loss": 0.1271, + "step": 14084 + }, + { + "epoch": 5.9, + "learning_rate": 3.407126095352755e-06, + "loss": 0.1263, + "step": 14085 + }, + { + "epoch": 5.9, + "learning_rate": 3.4058514273655984e-06, + "loss": 0.1307, + "step": 14086 + }, + { + "epoch": 5.9, + "learning_rate": 3.404576948921393e-06, + "loss": 0.1591, + "step": 14087 + }, + { + "epoch": 5.9, + "learning_rate": 3.4033026600567677e-06, + "loss": 0.1583, + "step": 14088 + }, + { + "epoch": 5.9, + "learning_rate": 3.402028560808357e-06, + "loss": 0.1016, + "step": 14089 + }, + { + "epoch": 5.9, + "learning_rate": 3.4007546512127764e-06, + "loss": 0.1096, + "step": 14090 + }, + { + "epoch": 5.9, + "learning_rate": 3.3994809313066477e-06, + "loss": 0.1148, + "step": 14091 + }, + { + "epoch": 5.9, + "learning_rate": 3.398207401126583e-06, + "loss": 0.1304, + "step": 14092 + }, + { + "epoch": 5.9, + "learning_rate": 3.3969340607091884e-06, + "loss": 0.0871, + "step": 14093 + }, + { + "epoch": 5.9, + "learning_rate": 3.3956609100910643e-06, + "loss": 0.1377, + "step": 14094 + }, + { + "epoch": 5.9, + "learning_rate": 3.394387949308806e-06, + "loss": 0.1212, + "step": 14095 + }, + { + "epoch": 5.9, + "learning_rate": 3.3931151783990092e-06, + "loss": 0.1206, + "step": 14096 + }, + { + "epoch": 5.9, + "learning_rate": 3.3918425973982496e-06, + "loss": 0.0933, + "step": 14097 + }, + { + "epoch": 5.9, + "learning_rate": 3.39057020634311e-06, + "loss": 0.0626, + "step": 14098 + }, + { + "epoch": 5.9, + "learning_rate": 3.389298005270165e-06, + "loss": 0.1276, + "step": 14099 + }, + { + "epoch": 5.9, + "learning_rate": 3.3880259942159833e-06, + "loss": 0.1148, + "step": 14100 + }, + { + "epoch": 5.9, + "learning_rate": 3.386754173217126e-06, + "loss": 0.1493, + "step": 14101 + }, + { + "epoch": 5.9, + "learning_rate": 3.3854825423101555e-06, + "loss": 0.1155, + "step": 14102 + }, + { + "epoch": 5.9, + "learning_rate": 3.384211101531617e-06, + "loss": 0.0956, + "step": 14103 + }, + { + "epoch": 5.9, + "learning_rate": 3.382939850918059e-06, + "loss": 0.0938, + "step": 14104 + }, + { + "epoch": 5.91, + "learning_rate": 3.3816687905060232e-06, + "loss": 0.1378, + "step": 14105 + }, + { + "epoch": 5.91, + "learning_rate": 3.380397920332045e-06, + "loss": 0.1013, + "step": 14106 + }, + { + "epoch": 5.91, + "learning_rate": 3.3791272404326535e-06, + "loss": 0.1205, + "step": 14107 + }, + { + "epoch": 5.91, + "learning_rate": 3.3778567508443782e-06, + "loss": 0.106, + "step": 14108 + }, + { + "epoch": 5.91, + "learning_rate": 3.3765864516037317e-06, + "loss": 0.1332, + "step": 14109 + }, + { + "epoch": 5.91, + "learning_rate": 3.3753163427472324e-06, + "loss": 0.1093, + "step": 14110 + }, + { + "epoch": 5.91, + "learning_rate": 3.3740464243113834e-06, + "loss": 0.1576, + "step": 14111 + }, + { + "epoch": 5.91, + "learning_rate": 3.3727766963326903e-06, + "loss": 0.1264, + "step": 14112 + }, + { + "epoch": 5.91, + "learning_rate": 3.37150715884765e-06, + "loss": 0.0976, + "step": 14113 + }, + { + "epoch": 5.91, + "learning_rate": 3.3702378118927537e-06, + "loss": 0.1251, + "step": 14114 + }, + { + "epoch": 5.91, + "learning_rate": 3.3689686555044888e-06, + "loss": 0.1625, + "step": 14115 + }, + { + "epoch": 5.91, + "learning_rate": 3.3676996897193393e-06, + "loss": 0.1327, + "step": 14116 + }, + { + "epoch": 5.91, + "learning_rate": 3.366430914573774e-06, + "loss": 0.1531, + "step": 14117 + }, + { + "epoch": 5.91, + "learning_rate": 3.3651623301042646e-06, + "loss": 0.1045, + "step": 14118 + }, + { + "epoch": 5.91, + "learning_rate": 3.363893936347277e-06, + "loss": 0.1105, + "step": 14119 + }, + { + "epoch": 5.91, + "learning_rate": 3.3626257333392687e-06, + "loss": 0.1239, + "step": 14120 + }, + { + "epoch": 5.91, + "learning_rate": 3.3613577211166936e-06, + "loss": 0.0995, + "step": 14121 + }, + { + "epoch": 5.91, + "learning_rate": 3.3600898997160005e-06, + "loss": 0.1189, + "step": 14122 + }, + { + "epoch": 5.91, + "learning_rate": 3.358822269173634e-06, + "loss": 0.1224, + "step": 14123 + }, + { + "epoch": 5.91, + "learning_rate": 3.357554829526025e-06, + "loss": 0.1318, + "step": 14124 + }, + { + "epoch": 5.91, + "learning_rate": 3.3562875808096083e-06, + "loss": 0.1149, + "step": 14125 + }, + { + "epoch": 5.91, + "learning_rate": 3.355020523060809e-06, + "loss": 0.1103, + "step": 14126 + }, + { + "epoch": 5.91, + "learning_rate": 3.3537536563160487e-06, + "loss": 0.1313, + "step": 14127 + }, + { + "epoch": 5.92, + "learning_rate": 3.352486980611741e-06, + "loss": 0.1419, + "step": 14128 + }, + { + "epoch": 5.92, + "learning_rate": 3.3512204959843e-06, + "loss": 0.1164, + "step": 14129 + }, + { + "epoch": 5.92, + "learning_rate": 3.3499542024701226e-06, + "loss": 0.1058, + "step": 14130 + }, + { + "epoch": 5.92, + "learning_rate": 3.3486881001056104e-06, + "loss": 0.1403, + "step": 14131 + }, + { + "epoch": 5.92, + "learning_rate": 3.3474221889271597e-06, + "loss": 0.0979, + "step": 14132 + }, + { + "epoch": 5.92, + "learning_rate": 3.346156468971151e-06, + "loss": 0.1352, + "step": 14133 + }, + { + "epoch": 5.92, + "learning_rate": 3.3448909402739716e-06, + "loss": 0.1098, + "step": 14134 + }, + { + "epoch": 5.92, + "learning_rate": 3.343625602871995e-06, + "loss": 0.1001, + "step": 14135 + }, + { + "epoch": 5.92, + "learning_rate": 3.3423604568015944e-06, + "loss": 0.139, + "step": 14136 + }, + { + "epoch": 5.92, + "learning_rate": 3.3410955020991377e-06, + "loss": 0.1335, + "step": 14137 + }, + { + "epoch": 5.92, + "learning_rate": 3.3398307388009786e-06, + "loss": 0.1197, + "step": 14138 + }, + { + "epoch": 5.92, + "learning_rate": 3.3385661669434744e-06, + "loss": 0.094, + "step": 14139 + }, + { + "epoch": 5.92, + "learning_rate": 3.3373017865629742e-06, + "loss": 0.1103, + "step": 14140 + }, + { + "epoch": 5.92, + "learning_rate": 3.3360375976958226e-06, + "loss": 0.1332, + "step": 14141 + }, + { + "epoch": 5.92, + "learning_rate": 3.334773600378357e-06, + "loss": 0.1125, + "step": 14142 + }, + { + "epoch": 5.92, + "learning_rate": 3.3335097946469118e-06, + "loss": 0.1256, + "step": 14143 + }, + { + "epoch": 5.92, + "learning_rate": 3.3322461805378094e-06, + "loss": 0.1456, + "step": 14144 + }, + { + "epoch": 5.92, + "learning_rate": 3.3309827580873734e-06, + "loss": 0.1428, + "step": 14145 + }, + { + "epoch": 5.92, + "learning_rate": 3.3297195273319195e-06, + "loss": 0.131, + "step": 14146 + }, + { + "epoch": 5.92, + "learning_rate": 3.32845648830776e-06, + "loss": 0.122, + "step": 14147 + }, + { + "epoch": 5.92, + "learning_rate": 3.3271936410511974e-06, + "loss": 0.1043, + "step": 14148 + }, + { + "epoch": 5.92, + "learning_rate": 3.3259309855985353e-06, + "loss": 0.1327, + "step": 14149 + }, + { + "epoch": 5.92, + "learning_rate": 3.324668521986062e-06, + "loss": 0.1611, + "step": 14150 + }, + { + "epoch": 5.92, + "learning_rate": 3.3234062502500685e-06, + "loss": 0.0954, + "step": 14151 + }, + { + "epoch": 5.93, + "learning_rate": 3.3221441704268375e-06, + "loss": 0.1467, + "step": 14152 + }, + { + "epoch": 5.93, + "learning_rate": 3.3208822825526464e-06, + "loss": 0.0981, + "step": 14153 + }, + { + "epoch": 5.93, + "learning_rate": 3.3196205866637667e-06, + "loss": 0.114, + "step": 14154 + }, + { + "epoch": 5.93, + "learning_rate": 3.3183590827964696e-06, + "loss": 0.1119, + "step": 14155 + }, + { + "epoch": 5.93, + "learning_rate": 3.3170977709870066e-06, + "loss": 0.12, + "step": 14156 + }, + { + "epoch": 5.93, + "learning_rate": 3.3158366512716424e-06, + "loss": 0.1203, + "step": 14157 + }, + { + "epoch": 5.93, + "learning_rate": 3.314575723686618e-06, + "loss": 0.1199, + "step": 14158 + }, + { + "epoch": 5.93, + "learning_rate": 3.3133149882681836e-06, + "loss": 0.1252, + "step": 14159 + }, + { + "epoch": 5.93, + "learning_rate": 3.3120544450525747e-06, + "loss": 0.095, + "step": 14160 + }, + { + "epoch": 5.93, + "learning_rate": 3.3107940940760264e-06, + "loss": 0.1227, + "step": 14161 + }, + { + "epoch": 5.93, + "learning_rate": 3.3095339353747657e-06, + "loss": 0.0897, + "step": 14162 + }, + { + "epoch": 5.93, + "learning_rate": 3.308273968985015e-06, + "loss": 0.1321, + "step": 14163 + }, + { + "epoch": 5.93, + "learning_rate": 3.3070141949429945e-06, + "loss": 0.0816, + "step": 14164 + }, + { + "epoch": 5.93, + "learning_rate": 3.3057546132849095e-06, + "loss": 0.1089, + "step": 14165 + }, + { + "epoch": 5.93, + "learning_rate": 3.304495224046966e-06, + "loss": 0.1081, + "step": 14166 + }, + { + "epoch": 5.93, + "learning_rate": 3.3032360272653664e-06, + "loss": 0.1332, + "step": 14167 + }, + { + "epoch": 5.93, + "learning_rate": 3.3019770229763048e-06, + "loss": 0.1229, + "step": 14168 + }, + { + "epoch": 5.93, + "learning_rate": 3.30071821121597e-06, + "loss": 0.1548, + "step": 14169 + }, + { + "epoch": 5.93, + "learning_rate": 3.2994595920205485e-06, + "loss": 0.1346, + "step": 14170 + }, + { + "epoch": 5.93, + "learning_rate": 3.2982011654262112e-06, + "loss": 0.1042, + "step": 14171 + }, + { + "epoch": 5.93, + "learning_rate": 3.296942931469136e-06, + "loss": 0.1263, + "step": 14172 + }, + { + "epoch": 5.93, + "learning_rate": 3.2956848901854863e-06, + "loss": 0.1211, + "step": 14173 + }, + { + "epoch": 5.93, + "learning_rate": 3.2944270416114256e-06, + "loss": 0.1317, + "step": 14174 + }, + { + "epoch": 5.93, + "learning_rate": 3.2931693857831094e-06, + "loss": 0.1582, + "step": 14175 + }, + { + "epoch": 5.94, + "learning_rate": 3.2919119227366913e-06, + "loss": 0.1371, + "step": 14176 + }, + { + "epoch": 5.94, + "learning_rate": 3.2906546525083082e-06, + "loss": 0.14, + "step": 14177 + }, + { + "epoch": 5.94, + "learning_rate": 3.2893975751341078e-06, + "loss": 0.1377, + "step": 14178 + }, + { + "epoch": 5.94, + "learning_rate": 3.2881406906502146e-06, + "loss": 0.134, + "step": 14179 + }, + { + "epoch": 5.94, + "learning_rate": 3.2868839990927627e-06, + "loss": 0.1461, + "step": 14180 + }, + { + "epoch": 5.94, + "learning_rate": 3.285627500497872e-06, + "loss": 0.1224, + "step": 14181 + }, + { + "epoch": 5.94, + "learning_rate": 3.284371194901661e-06, + "loss": 0.1106, + "step": 14182 + }, + { + "epoch": 5.94, + "learning_rate": 3.2831150823402403e-06, + "loss": 0.1034, + "step": 14183 + }, + { + "epoch": 5.94, + "learning_rate": 3.28185916284972e-06, + "loss": 0.1049, + "step": 14184 + }, + { + "epoch": 5.94, + "learning_rate": 3.280603436466192e-06, + "loss": 0.0994, + "step": 14185 + }, + { + "epoch": 5.94, + "learning_rate": 3.279347903225757e-06, + "loss": 0.1359, + "step": 14186 + }, + { + "epoch": 5.94, + "learning_rate": 3.278092563164502e-06, + "loss": 0.1302, + "step": 14187 + }, + { + "epoch": 5.94, + "learning_rate": 3.2768374163185114e-06, + "loss": 0.0997, + "step": 14188 + }, + { + "epoch": 5.94, + "learning_rate": 3.275582462723863e-06, + "loss": 0.1485, + "step": 14189 + }, + { + "epoch": 5.94, + "learning_rate": 3.274327702416633e-06, + "loss": 0.1328, + "step": 14190 + }, + { + "epoch": 5.94, + "learning_rate": 3.2730731354328827e-06, + "loss": 0.0914, + "step": 14191 + }, + { + "epoch": 5.94, + "learning_rate": 3.2718187618086737e-06, + "loss": 0.1373, + "step": 14192 + }, + { + "epoch": 5.94, + "learning_rate": 3.2705645815800658e-06, + "loss": 0.1274, + "step": 14193 + }, + { + "epoch": 5.94, + "learning_rate": 3.269310594783106e-06, + "loss": 0.0989, + "step": 14194 + }, + { + "epoch": 5.94, + "learning_rate": 3.2680568014538415e-06, + "loss": 0.1325, + "step": 14195 + }, + { + "epoch": 5.94, + "learning_rate": 3.2668032016283103e-06, + "loss": 0.1238, + "step": 14196 + }, + { + "epoch": 5.94, + "learning_rate": 3.2655497953425484e-06, + "loss": 0.1115, + "step": 14197 + }, + { + "epoch": 5.94, + "learning_rate": 3.26429658263258e-06, + "loss": 0.1297, + "step": 14198 + }, + { + "epoch": 5.94, + "learning_rate": 3.2630435635344283e-06, + "loss": 0.0975, + "step": 14199 + }, + { + "epoch": 5.95, + "learning_rate": 3.261790738084114e-06, + "loss": 0.1321, + "step": 14200 + }, + { + "epoch": 5.95, + "learning_rate": 3.2605381063176423e-06, + "loss": 0.0961, + "step": 14201 + }, + { + "epoch": 5.95, + "learning_rate": 3.259285668271023e-06, + "loss": 0.1038, + "step": 14202 + }, + { + "epoch": 5.95, + "learning_rate": 3.2580334239802545e-06, + "loss": 0.1285, + "step": 14203 + }, + { + "epoch": 5.95, + "learning_rate": 3.2567813734813336e-06, + "loss": 0.0986, + "step": 14204 + }, + { + "epoch": 5.95, + "learning_rate": 3.255529516810251e-06, + "loss": 0.128, + "step": 14205 + }, + { + "epoch": 5.95, + "learning_rate": 3.254277854002984e-06, + "loss": 0.0905, + "step": 14206 + }, + { + "epoch": 5.95, + "learning_rate": 3.2530263850955147e-06, + "loss": 0.1159, + "step": 14207 + }, + { + "epoch": 5.95, + "learning_rate": 3.251775110123814e-06, + "loss": 0.1275, + "step": 14208 + }, + { + "epoch": 5.95, + "learning_rate": 3.2505240291238505e-06, + "loss": 0.1211, + "step": 14209 + }, + { + "epoch": 5.95, + "learning_rate": 3.2492731421315838e-06, + "loss": 0.0921, + "step": 14210 + }, + { + "epoch": 5.95, + "learning_rate": 3.2480224491829728e-06, + "loss": 0.1079, + "step": 14211 + }, + { + "epoch": 5.95, + "learning_rate": 3.2467719503139626e-06, + "loss": 0.1004, + "step": 14212 + }, + { + "epoch": 5.95, + "learning_rate": 3.2455216455605e-06, + "loss": 0.1377, + "step": 14213 + }, + { + "epoch": 5.95, + "learning_rate": 3.244271534958523e-06, + "loss": 0.1096, + "step": 14214 + }, + { + "epoch": 5.95, + "learning_rate": 3.2430216185439667e-06, + "loss": 0.102, + "step": 14215 + }, + { + "epoch": 5.95, + "learning_rate": 3.2417718963527576e-06, + "loss": 0.1079, + "step": 14216 + }, + { + "epoch": 5.95, + "learning_rate": 3.2405223684208208e-06, + "loss": 0.12, + "step": 14217 + }, + { + "epoch": 5.95, + "learning_rate": 3.2392730347840686e-06, + "loss": 0.1327, + "step": 14218 + }, + { + "epoch": 5.95, + "learning_rate": 3.2380238954784115e-06, + "loss": 0.1363, + "step": 14219 + }, + { + "epoch": 5.95, + "learning_rate": 3.2367749505397582e-06, + "loss": 0.1274, + "step": 14220 + }, + { + "epoch": 5.95, + "learning_rate": 3.2355262000040077e-06, + "loss": 0.1217, + "step": 14221 + }, + { + "epoch": 5.95, + "learning_rate": 3.2342776439070545e-06, + "loss": 0.087, + "step": 14222 + }, + { + "epoch": 5.95, + "learning_rate": 3.2330292822847885e-06, + "loss": 0.1326, + "step": 14223 + }, + { + "epoch": 5.96, + "learning_rate": 3.231781115173088e-06, + "loss": 0.0972, + "step": 14224 + }, + { + "epoch": 5.96, + "learning_rate": 3.2305331426078367e-06, + "loss": 0.1557, + "step": 14225 + }, + { + "epoch": 5.96, + "learning_rate": 3.2292853646249002e-06, + "loss": 0.1304, + "step": 14226 + }, + { + "epoch": 5.96, + "learning_rate": 3.228037781260147e-06, + "loss": 0.1231, + "step": 14227 + }, + { + "epoch": 5.96, + "learning_rate": 3.2267903925494394e-06, + "loss": 0.1325, + "step": 14228 + }, + { + "epoch": 5.96, + "learning_rate": 3.22554319852863e-06, + "loss": 0.1139, + "step": 14229 + }, + { + "epoch": 5.96, + "learning_rate": 3.2242961992335707e-06, + "loss": 0.0909, + "step": 14230 + }, + { + "epoch": 5.96, + "learning_rate": 3.2230493947001084e-06, + "loss": 0.1008, + "step": 14231 + }, + { + "epoch": 5.96, + "learning_rate": 3.221802784964073e-06, + "loss": 0.1182, + "step": 14232 + }, + { + "epoch": 5.96, + "learning_rate": 3.2205563700613028e-06, + "loss": 0.123, + "step": 14233 + }, + { + "epoch": 5.96, + "learning_rate": 3.2193101500276226e-06, + "loss": 0.0822, + "step": 14234 + }, + { + "epoch": 5.96, + "learning_rate": 3.2180641248988566e-06, + "loss": 0.1333, + "step": 14235 + }, + { + "epoch": 5.96, + "learning_rate": 3.2168182947108173e-06, + "loss": 0.0668, + "step": 14236 + }, + { + "epoch": 5.96, + "learning_rate": 3.2155726594993177e-06, + "loss": 0.1355, + "step": 14237 + }, + { + "epoch": 5.96, + "learning_rate": 3.2143272193001652e-06, + "loss": 0.1102, + "step": 14238 + }, + { + "epoch": 5.96, + "learning_rate": 3.2130819741491513e-06, + "loss": 0.1288, + "step": 14239 + }, + { + "epoch": 5.96, + "learning_rate": 3.2118369240820745e-06, + "loss": 0.1122, + "step": 14240 + }, + { + "epoch": 5.96, + "learning_rate": 3.2105920691347216e-06, + "loss": 0.1103, + "step": 14241 + }, + { + "epoch": 5.96, + "learning_rate": 3.2093474093428733e-06, + "loss": 0.118, + "step": 14242 + }, + { + "epoch": 5.96, + "learning_rate": 3.2081029447423095e-06, + "loss": 0.1368, + "step": 14243 + }, + { + "epoch": 5.96, + "learning_rate": 3.2068586753688014e-06, + "loss": 0.1325, + "step": 14244 + }, + { + "epoch": 5.96, + "learning_rate": 3.20561460125811e-06, + "loss": 0.0986, + "step": 14245 + }, + { + "epoch": 5.96, + "learning_rate": 3.2043707224460006e-06, + "loss": 0.1024, + "step": 14246 + }, + { + "epoch": 5.96, + "learning_rate": 3.203127038968222e-06, + "loss": 0.1025, + "step": 14247 + }, + { + "epoch": 5.97, + "learning_rate": 3.201883550860525e-06, + "loss": 0.1007, + "step": 14248 + }, + { + "epoch": 5.97, + "learning_rate": 3.2006402581586527e-06, + "loss": 0.0998, + "step": 14249 + }, + { + "epoch": 5.97, + "learning_rate": 3.199397160898342e-06, + "loss": 0.1018, + "step": 14250 + }, + { + "epoch": 5.97, + "learning_rate": 3.1981542591153247e-06, + "loss": 0.1279, + "step": 14251 + }, + { + "epoch": 5.97, + "learning_rate": 3.196911552845332e-06, + "loss": 0.1581, + "step": 14252 + }, + { + "epoch": 5.97, + "learning_rate": 3.1956690421240754e-06, + "loss": 0.1524, + "step": 14253 + }, + { + "epoch": 5.97, + "learning_rate": 3.1944267269872754e-06, + "loss": 0.147, + "step": 14254 + }, + { + "epoch": 5.97, + "learning_rate": 3.1931846074706386e-06, + "loss": 0.1009, + "step": 14255 + }, + { + "epoch": 5.97, + "learning_rate": 3.191942683609871e-06, + "loss": 0.108, + "step": 14256 + }, + { + "epoch": 5.97, + "learning_rate": 3.190700955440669e-06, + "loss": 0.1121, + "step": 14257 + }, + { + "epoch": 5.97, + "learning_rate": 3.18945942299873e-06, + "loss": 0.1387, + "step": 14258 + }, + { + "epoch": 5.97, + "learning_rate": 3.1882180863197308e-06, + "loss": 0.121, + "step": 14259 + }, + { + "epoch": 5.97, + "learning_rate": 3.1869769454393596e-06, + "loss": 0.1329, + "step": 14260 + }, + { + "epoch": 5.97, + "learning_rate": 3.18573600039329e-06, + "loss": 0.1233, + "step": 14261 + }, + { + "epoch": 5.97, + "learning_rate": 3.1844952512171924e-06, + "loss": 0.094, + "step": 14262 + }, + { + "epoch": 5.97, + "learning_rate": 3.1832546979467304e-06, + "loss": 0.1106, + "step": 14263 + }, + { + "epoch": 5.97, + "learning_rate": 3.1820143406175642e-06, + "loss": 0.1171, + "step": 14264 + }, + { + "epoch": 5.97, + "learning_rate": 3.180774179265348e-06, + "loss": 0.1091, + "step": 14265 + }, + { + "epoch": 5.97, + "learning_rate": 3.1795342139257233e-06, + "loss": 0.1443, + "step": 14266 + }, + { + "epoch": 5.97, + "learning_rate": 3.1782944446343357e-06, + "loss": 0.1002, + "step": 14267 + }, + { + "epoch": 5.97, + "learning_rate": 3.1770548714268236e-06, + "loss": 0.0809, + "step": 14268 + }, + { + "epoch": 5.97, + "learning_rate": 3.175815494338811e-06, + "loss": 0.1243, + "step": 14269 + }, + { + "epoch": 5.97, + "learning_rate": 3.174576313405927e-06, + "loss": 0.1191, + "step": 14270 + }, + { + "epoch": 5.97, + "learning_rate": 3.17333732866379e-06, + "loss": 0.1145, + "step": 14271 + }, + { + "epoch": 5.98, + "learning_rate": 3.1720985401480176e-06, + "loss": 0.1007, + "step": 14272 + }, + { + "epoch": 5.98, + "learning_rate": 3.1708599478942094e-06, + "loss": 0.1246, + "step": 14273 + }, + { + "epoch": 5.98, + "learning_rate": 3.1696215519379726e-06, + "loss": 0.1157, + "step": 14274 + }, + { + "epoch": 5.98, + "learning_rate": 3.168383352314902e-06, + "loss": 0.1044, + "step": 14275 + }, + { + "epoch": 5.98, + "learning_rate": 3.1671453490605906e-06, + "loss": 0.1279, + "step": 14276 + }, + { + "epoch": 5.98, + "learning_rate": 3.1659075422106233e-06, + "loss": 0.1473, + "step": 14277 + }, + { + "epoch": 5.98, + "learning_rate": 3.164669931800578e-06, + "loss": 0.1236, + "step": 14278 + }, + { + "epoch": 5.98, + "learning_rate": 3.163432517866034e-06, + "loss": 0.1409, + "step": 14279 + }, + { + "epoch": 5.98, + "learning_rate": 3.162195300442552e-06, + "loss": 0.1223, + "step": 14280 + }, + { + "epoch": 5.98, + "learning_rate": 3.160958279565699e-06, + "loss": 0.1252, + "step": 14281 + }, + { + "epoch": 5.98, + "learning_rate": 3.15972145527103e-06, + "loss": 0.1619, + "step": 14282 + }, + { + "epoch": 5.98, + "learning_rate": 3.158484827594099e-06, + "loss": 0.0831, + "step": 14283 + }, + { + "epoch": 5.98, + "learning_rate": 3.1572483965704503e-06, + "loss": 0.1201, + "step": 14284 + }, + { + "epoch": 5.98, + "learning_rate": 3.156012162235628e-06, + "loss": 0.1578, + "step": 14285 + }, + { + "epoch": 5.98, + "learning_rate": 3.1547761246251587e-06, + "loss": 0.1137, + "step": 14286 + }, + { + "epoch": 5.98, + "learning_rate": 3.1535402837745753e-06, + "loss": 0.1081, + "step": 14287 + }, + { + "epoch": 5.98, + "learning_rate": 3.1523046397194015e-06, + "loss": 0.1271, + "step": 14288 + }, + { + "epoch": 5.98, + "learning_rate": 3.151069192495154e-06, + "loss": 0.1111, + "step": 14289 + }, + { + "epoch": 5.98, + "learning_rate": 3.149833942137346e-06, + "loss": 0.1529, + "step": 14290 + }, + { + "epoch": 5.98, + "learning_rate": 3.1485988886814846e-06, + "loss": 0.1303, + "step": 14291 + }, + { + "epoch": 5.98, + "learning_rate": 3.147364032163065e-06, + "loss": 0.1105, + "step": 14292 + }, + { + "epoch": 5.98, + "learning_rate": 3.146129372617589e-06, + "loss": 0.1075, + "step": 14293 + }, + { + "epoch": 5.98, + "learning_rate": 3.1448949100805394e-06, + "loss": 0.107, + "step": 14294 + }, + { + "epoch": 5.98, + "learning_rate": 3.1436606445874017e-06, + "loss": 0.1458, + "step": 14295 + }, + { + "epoch": 5.99, + "learning_rate": 3.142426576173655e-06, + "loss": 0.1445, + "step": 14296 + }, + { + "epoch": 5.99, + "learning_rate": 3.1411927048747715e-06, + "loss": 0.1194, + "step": 14297 + }, + { + "epoch": 5.99, + "learning_rate": 3.1399590307262175e-06, + "loss": 0.1093, + "step": 14298 + }, + { + "epoch": 5.99, + "learning_rate": 3.1387255537634553e-06, + "loss": 0.1034, + "step": 14299 + }, + { + "epoch": 5.99, + "learning_rate": 3.137492274021937e-06, + "loss": 0.0787, + "step": 14300 + }, + { + "epoch": 5.99, + "learning_rate": 3.1362591915371123e-06, + "loss": 0.1017, + "step": 14301 + }, + { + "epoch": 5.99, + "learning_rate": 3.1350263063444264e-06, + "loss": 0.1271, + "step": 14302 + }, + { + "epoch": 5.99, + "learning_rate": 3.1337936184793185e-06, + "loss": 0.1039, + "step": 14303 + }, + { + "epoch": 5.99, + "learning_rate": 3.1325611279772194e-06, + "loss": 0.0875, + "step": 14304 + }, + { + "epoch": 5.99, + "learning_rate": 3.131328834873557e-06, + "loss": 0.1423, + "step": 14305 + }, + { + "epoch": 5.99, + "learning_rate": 3.130096739203755e-06, + "loss": 0.1137, + "step": 14306 + }, + { + "epoch": 5.99, + "learning_rate": 3.128864841003223e-06, + "loss": 0.107, + "step": 14307 + }, + { + "epoch": 5.99, + "learning_rate": 3.1276331403073733e-06, + "loss": 0.1402, + "step": 14308 + }, + { + "epoch": 5.99, + "learning_rate": 3.1264016371516106e-06, + "loss": 0.1349, + "step": 14309 + }, + { + "epoch": 5.99, + "learning_rate": 3.1251703315713333e-06, + "loss": 0.1357, + "step": 14310 + }, + { + "epoch": 5.99, + "learning_rate": 3.123939223601935e-06, + "loss": 0.156, + "step": 14311 + }, + { + "epoch": 5.99, + "learning_rate": 3.122708313278805e-06, + "loss": 0.1171, + "step": 14312 + }, + { + "epoch": 5.99, + "learning_rate": 3.1214776006373205e-06, + "loss": 0.1377, + "step": 14313 + }, + { + "epoch": 5.99, + "learning_rate": 3.1202470857128564e-06, + "loss": 0.1145, + "step": 14314 + }, + { + "epoch": 5.99, + "learning_rate": 3.1190167685407848e-06, + "loss": 0.0946, + "step": 14315 + }, + { + "epoch": 5.99, + "learning_rate": 3.11778664915647e-06, + "loss": 0.1007, + "step": 14316 + }, + { + "epoch": 5.99, + "learning_rate": 3.1165567275952712e-06, + "loss": 0.1388, + "step": 14317 + }, + { + "epoch": 5.99, + "learning_rate": 3.1153270038925397e-06, + "loss": 0.1399, + "step": 14318 + }, + { + "epoch": 5.99, + "learning_rate": 3.1140974780836254e-06, + "loss": 0.1044, + "step": 14319 + }, + { + "epoch": 6.0, + "learning_rate": 3.1128681502038717e-06, + "loss": 0.1562, + "step": 14320 + }, + { + "epoch": 6.0, + "learning_rate": 3.1116390202886083e-06, + "loss": 0.1052, + "step": 14321 + }, + { + "epoch": 6.0, + "learning_rate": 3.110410088373168e-06, + "loss": 0.1, + "step": 14322 + }, + { + "epoch": 6.0, + "learning_rate": 3.1091813544928763e-06, + "loss": 0.1175, + "step": 14323 + }, + { + "epoch": 6.0, + "learning_rate": 3.107952818683052e-06, + "loss": 0.1078, + "step": 14324 + }, + { + "epoch": 6.0, + "learning_rate": 3.1067244809790076e-06, + "loss": 0.1231, + "step": 14325 + }, + { + "epoch": 6.0, + "learning_rate": 3.1054963414160545e-06, + "loss": 0.1265, + "step": 14326 + }, + { + "epoch": 6.0, + "learning_rate": 3.1042684000294876e-06, + "loss": 0.118, + "step": 14327 + }, + { + "epoch": 6.0, + "learning_rate": 3.1030406568546066e-06, + "loss": 0.1147, + "step": 14328 + }, + { + "epoch": 6.0, + "learning_rate": 3.1018131119267014e-06, + "loss": 0.0994, + "step": 14329 + }, + { + "epoch": 6.0, + "learning_rate": 3.100585765281057e-06, + "loss": 0.1185, + "step": 14330 + }, + { + "epoch": 6.0, + "learning_rate": 3.099358616952952e-06, + "loss": 0.1292, + "step": 14331 + }, + { + "epoch": 6.0, + "learning_rate": 3.098131666977664e-06, + "loss": 0.1093, + "step": 14332 + }, + { + "epoch": 6.0, + "learning_rate": 3.096904915390454e-06, + "loss": 0.0678, + "step": 14333 + }, + { + "epoch": 6.0, + "learning_rate": 3.095678362226585e-06, + "loss": 0.1091, + "step": 14334 + }, + { + "epoch": 6.0, + "learning_rate": 3.094452007521316e-06, + "loss": 0.1122, + "step": 14335 + }, + { + "epoch": 6.0, + "learning_rate": 3.0932258513098978e-06, + "loss": 0.0605, + "step": 14336 + }, + { + "epoch": 6.0, + "learning_rate": 3.091999893627572e-06, + "loss": 0.071, + "step": 14337 + }, + { + "epoch": 6.0, + "learning_rate": 3.090774134509579e-06, + "loss": 0.0975, + "step": 14338 + }, + { + "epoch": 6.0, + "learning_rate": 3.089548573991151e-06, + "loss": 0.0612, + "step": 14339 + }, + { + "epoch": 6.0, + "learning_rate": 3.088323212107521e-06, + "loss": 0.1026, + "step": 14340 + }, + { + "epoch": 6.0, + "learning_rate": 3.087098048893904e-06, + "loss": 0.0785, + "step": 14341 + }, + { + "epoch": 6.0, + "learning_rate": 3.0858730843855187e-06, + "loss": 0.0928, + "step": 14342 + }, + { + "epoch": 6.01, + "learning_rate": 3.084648318617577e-06, + "loss": 0.0686, + "step": 14343 + }, + { + "epoch": 6.01, + "learning_rate": 3.0834237516252817e-06, + "loss": 0.0737, + "step": 14344 + }, + { + "epoch": 6.01, + "learning_rate": 3.0821993834438337e-06, + "loss": 0.1265, + "step": 14345 + }, + { + "epoch": 6.01, + "learning_rate": 3.0809752141084248e-06, + "loss": 0.0865, + "step": 14346 + }, + { + "epoch": 6.01, + "learning_rate": 3.079751243654248e-06, + "loss": 0.0877, + "step": 14347 + }, + { + "epoch": 6.01, + "learning_rate": 3.078527472116477e-06, + "loss": 0.0758, + "step": 14348 + }, + { + "epoch": 6.01, + "learning_rate": 3.0773038995302918e-06, + "loss": 0.0773, + "step": 14349 + }, + { + "epoch": 6.01, + "learning_rate": 3.0760805259308626e-06, + "loss": 0.0597, + "step": 14350 + }, + { + "epoch": 6.01, + "learning_rate": 3.0748573513533554e-06, + "loss": 0.1065, + "step": 14351 + }, + { + "epoch": 6.01, + "learning_rate": 3.0736343758329277e-06, + "loss": 0.0947, + "step": 14352 + }, + { + "epoch": 6.01, + "learning_rate": 3.0724115994047367e-06, + "loss": 0.0626, + "step": 14353 + }, + { + "epoch": 6.01, + "learning_rate": 3.071189022103923e-06, + "loss": 0.0927, + "step": 14354 + }, + { + "epoch": 6.01, + "learning_rate": 3.069966643965634e-06, + "loss": 0.077, + "step": 14355 + }, + { + "epoch": 6.01, + "learning_rate": 3.068744465025003e-06, + "loss": 0.1038, + "step": 14356 + }, + { + "epoch": 6.01, + "learning_rate": 3.0675224853171613e-06, + "loss": 0.0589, + "step": 14357 + }, + { + "epoch": 6.01, + "learning_rate": 3.0663007048772354e-06, + "loss": 0.0928, + "step": 14358 + }, + { + "epoch": 6.01, + "learning_rate": 3.0650791237403456e-06, + "loss": 0.0821, + "step": 14359 + }, + { + "epoch": 6.01, + "learning_rate": 3.063857741941598e-06, + "loss": 0.066, + "step": 14360 + }, + { + "epoch": 6.01, + "learning_rate": 3.06263655951611e-06, + "loss": 0.0888, + "step": 14361 + }, + { + "epoch": 6.01, + "learning_rate": 3.061415576498973e-06, + "loss": 0.0679, + "step": 14362 + }, + { + "epoch": 6.01, + "learning_rate": 3.0601947929252883e-06, + "loss": 0.0956, + "step": 14363 + }, + { + "epoch": 6.01, + "learning_rate": 3.058974208830147e-06, + "loss": 0.0821, + "step": 14364 + }, + { + "epoch": 6.01, + "learning_rate": 3.0577538242486317e-06, + "loss": 0.0883, + "step": 14365 + }, + { + "epoch": 6.01, + "learning_rate": 3.0565336392158228e-06, + "loss": 0.0893, + "step": 14366 + }, + { + "epoch": 6.02, + "learning_rate": 3.055313653766796e-06, + "loss": 0.0802, + "step": 14367 + }, + { + "epoch": 6.02, + "learning_rate": 3.0540938679366126e-06, + "loss": 0.0919, + "step": 14368 + }, + { + "epoch": 6.02, + "learning_rate": 3.0528742817603364e-06, + "loss": 0.0956, + "step": 14369 + }, + { + "epoch": 6.02, + "learning_rate": 3.0516548952730253e-06, + "loss": 0.0819, + "step": 14370 + }, + { + "epoch": 6.02, + "learning_rate": 3.050435708509728e-06, + "loss": 0.0625, + "step": 14371 + }, + { + "epoch": 6.02, + "learning_rate": 3.049216721505489e-06, + "loss": 0.0944, + "step": 14372 + }, + { + "epoch": 6.02, + "learning_rate": 3.0479979342953504e-06, + "loss": 0.0734, + "step": 14373 + }, + { + "epoch": 6.02, + "learning_rate": 3.046779346914339e-06, + "loss": 0.0737, + "step": 14374 + }, + { + "epoch": 6.02, + "learning_rate": 3.0455609593974855e-06, + "loss": 0.0908, + "step": 14375 + }, + { + "epoch": 6.02, + "learning_rate": 3.044342771779811e-06, + "loss": 0.0815, + "step": 14376 + }, + { + "epoch": 6.02, + "learning_rate": 3.0431247840963316e-06, + "loss": 0.1037, + "step": 14377 + }, + { + "epoch": 6.02, + "learning_rate": 3.041906996382056e-06, + "loss": 0.0726, + "step": 14378 + }, + { + "epoch": 6.02, + "learning_rate": 3.0406894086719894e-06, + "loss": 0.1095, + "step": 14379 + }, + { + "epoch": 6.02, + "learning_rate": 3.039472021001134e-06, + "loss": 0.0771, + "step": 14380 + }, + { + "epoch": 6.02, + "learning_rate": 3.0382548334044783e-06, + "loss": 0.0466, + "step": 14381 + }, + { + "epoch": 6.02, + "learning_rate": 3.0370378459170058e-06, + "loss": 0.0773, + "step": 14382 + }, + { + "epoch": 6.02, + "learning_rate": 3.0358210585737024e-06, + "loss": 0.0787, + "step": 14383 + }, + { + "epoch": 6.02, + "learning_rate": 3.0346044714095425e-06, + "loss": 0.0997, + "step": 14384 + }, + { + "epoch": 6.02, + "learning_rate": 3.0333880844594963e-06, + "loss": 0.0779, + "step": 14385 + }, + { + "epoch": 6.02, + "learning_rate": 3.032171897758527e-06, + "loss": 0.0939, + "step": 14386 + }, + { + "epoch": 6.02, + "learning_rate": 3.030955911341594e-06, + "loss": 0.1008, + "step": 14387 + }, + { + "epoch": 6.02, + "learning_rate": 3.029740125243652e-06, + "loss": 0.0828, + "step": 14388 + }, + { + "epoch": 6.02, + "learning_rate": 3.028524539499642e-06, + "loss": 0.1049, + "step": 14389 + }, + { + "epoch": 6.02, + "learning_rate": 3.0273091541445065e-06, + "loss": 0.0757, + "step": 14390 + }, + { + "epoch": 6.03, + "learning_rate": 3.026093969213183e-06, + "loss": 0.0831, + "step": 14391 + }, + { + "epoch": 6.03, + "learning_rate": 3.0248789847406003e-06, + "loss": 0.0728, + "step": 14392 + }, + { + "epoch": 6.03, + "learning_rate": 3.02366420076168e-06, + "loss": 0.0453, + "step": 14393 + }, + { + "epoch": 6.03, + "learning_rate": 3.022449617311346e-06, + "loss": 0.089, + "step": 14394 + }, + { + "epoch": 6.03, + "learning_rate": 3.021235234424502e-06, + "loss": 0.0904, + "step": 14395 + }, + { + "epoch": 6.03, + "learning_rate": 3.0200210521360598e-06, + "loss": 0.0671, + "step": 14396 + }, + { + "epoch": 6.03, + "learning_rate": 3.0188070704809167e-06, + "loss": 0.1163, + "step": 14397 + }, + { + "epoch": 6.03, + "learning_rate": 3.017593289493971e-06, + "loss": 0.0829, + "step": 14398 + }, + { + "epoch": 6.03, + "learning_rate": 3.0163797092101087e-06, + "loss": 0.0714, + "step": 14399 + }, + { + "epoch": 6.03, + "learning_rate": 3.0151663296642188e-06, + "loss": 0.09, + "step": 14400 + }, + { + "epoch": 6.03, + "learning_rate": 3.01395315089117e-06, + "loss": 0.0753, + "step": 14401 + }, + { + "epoch": 6.03, + "learning_rate": 3.01274017292584e-06, + "loss": 0.0752, + "step": 14402 + }, + { + "epoch": 6.03, + "learning_rate": 3.011527395803092e-06, + "loss": 0.0562, + "step": 14403 + }, + { + "epoch": 6.03, + "learning_rate": 3.0103148195577914e-06, + "loss": 0.0999, + "step": 14404 + }, + { + "epoch": 6.03, + "learning_rate": 3.009102444224786e-06, + "loss": 0.0673, + "step": 14405 + }, + { + "epoch": 6.03, + "learning_rate": 3.0078902698389266e-06, + "loss": 0.0899, + "step": 14406 + }, + { + "epoch": 6.03, + "learning_rate": 3.0066782964350573e-06, + "loss": 0.1033, + "step": 14407 + }, + { + "epoch": 6.03, + "learning_rate": 3.0054665240480174e-06, + "loss": 0.0715, + "step": 14408 + }, + { + "epoch": 6.03, + "learning_rate": 3.004254952712632e-06, + "loss": 0.0692, + "step": 14409 + }, + { + "epoch": 6.03, + "learning_rate": 3.0030435824637304e-06, + "loss": 0.0857, + "step": 14410 + }, + { + "epoch": 6.03, + "learning_rate": 3.0018324133361323e-06, + "loss": 0.0605, + "step": 14411 + }, + { + "epoch": 6.03, + "learning_rate": 3.000621445364651e-06, + "loss": 0.0519, + "step": 14412 + }, + { + "epoch": 6.03, + "learning_rate": 2.999410678584096e-06, + "loss": 0.0537, + "step": 14413 + }, + { + "epoch": 6.03, + "learning_rate": 2.9982001130292703e-06, + "loss": 0.0751, + "step": 14414 + }, + { + "epoch": 6.04, + "learning_rate": 2.9969897487349675e-06, + "loss": 0.083, + "step": 14415 + }, + { + "epoch": 6.04, + "learning_rate": 2.9957795857359794e-06, + "loss": 0.0699, + "step": 14416 + }, + { + "epoch": 6.04, + "learning_rate": 2.9945696240670905e-06, + "loss": 0.1112, + "step": 14417 + }, + { + "epoch": 6.04, + "learning_rate": 2.9933598637630833e-06, + "loss": 0.0777, + "step": 14418 + }, + { + "epoch": 6.04, + "learning_rate": 2.9921503048587284e-06, + "loss": 0.0872, + "step": 14419 + }, + { + "epoch": 6.04, + "learning_rate": 2.990940947388794e-06, + "loss": 0.1089, + "step": 14420 + }, + { + "epoch": 6.04, + "learning_rate": 2.9897317913880454e-06, + "loss": 0.0951, + "step": 14421 + }, + { + "epoch": 6.04, + "learning_rate": 2.9885228368912333e-06, + "loss": 0.052, + "step": 14422 + }, + { + "epoch": 6.04, + "learning_rate": 2.9873140839331094e-06, + "loss": 0.0862, + "step": 14423 + }, + { + "epoch": 6.04, + "learning_rate": 2.98610553254842e-06, + "loss": 0.0954, + "step": 14424 + }, + { + "epoch": 6.04, + "learning_rate": 2.984897182771903e-06, + "loss": 0.061, + "step": 14425 + }, + { + "epoch": 6.04, + "learning_rate": 2.9836890346382907e-06, + "loss": 0.0772, + "step": 14426 + }, + { + "epoch": 6.04, + "learning_rate": 2.982481088182315e-06, + "loss": 0.0885, + "step": 14427 + }, + { + "epoch": 6.04, + "learning_rate": 2.98127334343869e-06, + "loss": 0.1007, + "step": 14428 + }, + { + "epoch": 6.04, + "learning_rate": 2.980065800442137e-06, + "loss": 0.046, + "step": 14429 + }, + { + "epoch": 6.04, + "learning_rate": 2.9788584592273595e-06, + "loss": 0.068, + "step": 14430 + }, + { + "epoch": 6.04, + "learning_rate": 2.9776513198290657e-06, + "loss": 0.0835, + "step": 14431 + }, + { + "epoch": 6.04, + "learning_rate": 2.9764443822819534e-06, + "loss": 0.0601, + "step": 14432 + }, + { + "epoch": 6.04, + "learning_rate": 2.9752376466207154e-06, + "loss": 0.0832, + "step": 14433 + }, + { + "epoch": 6.04, + "learning_rate": 2.974031112880037e-06, + "loss": 0.1016, + "step": 14434 + }, + { + "epoch": 6.04, + "learning_rate": 2.972824781094602e-06, + "loss": 0.0758, + "step": 14435 + }, + { + "epoch": 6.04, + "learning_rate": 2.9716186512990806e-06, + "loss": 0.062, + "step": 14436 + }, + { + "epoch": 6.04, + "learning_rate": 2.9704127235281433e-06, + "loss": 0.0975, + "step": 14437 + }, + { + "epoch": 6.04, + "learning_rate": 2.9692069978164552e-06, + "loss": 0.0696, + "step": 14438 + }, + { + "epoch": 6.05, + "learning_rate": 2.968001474198673e-06, + "loss": 0.0852, + "step": 14439 + }, + { + "epoch": 6.05, + "learning_rate": 2.966796152709447e-06, + "loss": 0.1034, + "step": 14440 + }, + { + "epoch": 6.05, + "learning_rate": 2.9655910333834294e-06, + "loss": 0.0885, + "step": 14441 + }, + { + "epoch": 6.05, + "learning_rate": 2.9643861162552512e-06, + "loss": 0.0735, + "step": 14442 + }, + { + "epoch": 6.05, + "learning_rate": 2.9631814013595507e-06, + "loss": 0.114, + "step": 14443 + }, + { + "epoch": 6.05, + "learning_rate": 2.961976888730956e-06, + "loss": 0.0802, + "step": 14444 + }, + { + "epoch": 6.05, + "learning_rate": 2.960772578404091e-06, + "loss": 0.0662, + "step": 14445 + }, + { + "epoch": 6.05, + "learning_rate": 2.959568470413572e-06, + "loss": 0.0662, + "step": 14446 + }, + { + "epoch": 6.05, + "learning_rate": 2.9583645647940097e-06, + "loss": 0.0978, + "step": 14447 + }, + { + "epoch": 6.05, + "learning_rate": 2.9571608615800118e-06, + "loss": 0.0529, + "step": 14448 + }, + { + "epoch": 6.05, + "learning_rate": 2.9559573608061755e-06, + "loss": 0.0754, + "step": 14449 + }, + { + "epoch": 6.05, + "learning_rate": 2.9547540625070903e-06, + "loss": 0.0878, + "step": 14450 + }, + { + "epoch": 6.05, + "learning_rate": 2.953550966717349e-06, + "loss": 0.0838, + "step": 14451 + }, + { + "epoch": 6.05, + "learning_rate": 2.9523480734715317e-06, + "loss": 0.0953, + "step": 14452 + }, + { + "epoch": 6.05, + "learning_rate": 2.9511453828042147e-06, + "loss": 0.1005, + "step": 14453 + }, + { + "epoch": 6.05, + "learning_rate": 2.9499428947499696e-06, + "loss": 0.0766, + "step": 14454 + }, + { + "epoch": 6.05, + "learning_rate": 2.9487406093433622e-06, + "loss": 0.0949, + "step": 14455 + }, + { + "epoch": 6.05, + "learning_rate": 2.947538526618946e-06, + "loss": 0.0926, + "step": 14456 + }, + { + "epoch": 6.05, + "learning_rate": 2.9463366466112764e-06, + "loss": 0.1011, + "step": 14457 + }, + { + "epoch": 6.05, + "learning_rate": 2.9451349693549e-06, + "loss": 0.0735, + "step": 14458 + }, + { + "epoch": 6.05, + "learning_rate": 2.9439334948843588e-06, + "loss": 0.1279, + "step": 14459 + }, + { + "epoch": 6.05, + "learning_rate": 2.9427322232341893e-06, + "loss": 0.067, + "step": 14460 + }, + { + "epoch": 6.05, + "learning_rate": 2.941531154438918e-06, + "loss": 0.064, + "step": 14461 + }, + { + "epoch": 6.05, + "learning_rate": 2.940330288533074e-06, + "loss": 0.0995, + "step": 14462 + }, + { + "epoch": 6.06, + "learning_rate": 2.9391296255511683e-06, + "loss": 0.1056, + "step": 14463 + }, + { + "epoch": 6.06, + "learning_rate": 2.9379291655277143e-06, + "loss": 0.0957, + "step": 14464 + }, + { + "epoch": 6.06, + "learning_rate": 2.9367289084972206e-06, + "loss": 0.0597, + "step": 14465 + }, + { + "epoch": 6.06, + "learning_rate": 2.9355288544941873e-06, + "loss": 0.0803, + "step": 14466 + }, + { + "epoch": 6.06, + "learning_rate": 2.934329003553107e-06, + "loss": 0.0615, + "step": 14467 + }, + { + "epoch": 6.06, + "learning_rate": 2.933129355708473e-06, + "loss": 0.0744, + "step": 14468 + }, + { + "epoch": 6.06, + "learning_rate": 2.931929910994762e-06, + "loss": 0.0773, + "step": 14469 + }, + { + "epoch": 6.06, + "learning_rate": 2.930730669446453e-06, + "loss": 0.0954, + "step": 14470 + }, + { + "epoch": 6.06, + "learning_rate": 2.9295316310980184e-06, + "loss": 0.0626, + "step": 14471 + }, + { + "epoch": 6.06, + "learning_rate": 2.928332795983926e-06, + "loss": 0.0806, + "step": 14472 + }, + { + "epoch": 6.06, + "learning_rate": 2.9271341641386285e-06, + "loss": 0.0626, + "step": 14473 + }, + { + "epoch": 6.06, + "learning_rate": 2.925935735596582e-06, + "loss": 0.0804, + "step": 14474 + }, + { + "epoch": 6.06, + "learning_rate": 2.9247375103922373e-06, + "loss": 0.0862, + "step": 14475 + }, + { + "epoch": 6.06, + "learning_rate": 2.9235394885600367e-06, + "loss": 0.0685, + "step": 14476 + }, + { + "epoch": 6.06, + "learning_rate": 2.922341670134411e-06, + "loss": 0.1191, + "step": 14477 + }, + { + "epoch": 6.06, + "learning_rate": 2.921144055149793e-06, + "loss": 0.0693, + "step": 14478 + }, + { + "epoch": 6.06, + "learning_rate": 2.9199466436406077e-06, + "loss": 0.0933, + "step": 14479 + }, + { + "epoch": 6.06, + "learning_rate": 2.918749435641274e-06, + "loss": 0.1152, + "step": 14480 + }, + { + "epoch": 6.06, + "learning_rate": 2.9175524311862034e-06, + "loss": 0.103, + "step": 14481 + }, + { + "epoch": 6.06, + "learning_rate": 2.916355630309806e-06, + "loss": 0.0681, + "step": 14482 + }, + { + "epoch": 6.06, + "learning_rate": 2.9151590330464775e-06, + "loss": 0.084, + "step": 14483 + }, + { + "epoch": 6.06, + "learning_rate": 2.913962639430615e-06, + "loss": 0.0755, + "step": 14484 + }, + { + "epoch": 6.06, + "learning_rate": 2.91276644949661e-06, + "loss": 0.0906, + "step": 14485 + }, + { + "epoch": 6.06, + "learning_rate": 2.911570463278843e-06, + "loss": 0.0835, + "step": 14486 + }, + { + "epoch": 6.07, + "learning_rate": 2.910374680811693e-06, + "loss": 0.0773, + "step": 14487 + }, + { + "epoch": 6.07, + "learning_rate": 2.909179102129531e-06, + "loss": 0.0837, + "step": 14488 + }, + { + "epoch": 6.07, + "learning_rate": 2.9079837272667276e-06, + "loss": 0.0662, + "step": 14489 + }, + { + "epoch": 6.07, + "learning_rate": 2.906788556257636e-06, + "loss": 0.077, + "step": 14490 + }, + { + "epoch": 6.07, + "learning_rate": 2.9055935891366117e-06, + "loss": 0.089, + "step": 14491 + }, + { + "epoch": 6.07, + "learning_rate": 2.904398825938005e-06, + "loss": 0.0903, + "step": 14492 + }, + { + "epoch": 6.07, + "learning_rate": 2.903204266696157e-06, + "loss": 0.0882, + "step": 14493 + }, + { + "epoch": 6.07, + "learning_rate": 2.9020099114454046e-06, + "loss": 0.0891, + "step": 14494 + }, + { + "epoch": 6.07, + "learning_rate": 2.9008157602200827e-06, + "loss": 0.0848, + "step": 14495 + }, + { + "epoch": 6.07, + "learning_rate": 2.8996218130545108e-06, + "loss": 0.1225, + "step": 14496 + }, + { + "epoch": 6.07, + "learning_rate": 2.898428069983007e-06, + "loss": 0.0876, + "step": 14497 + }, + { + "epoch": 6.07, + "learning_rate": 2.8972345310398863e-06, + "loss": 0.0819, + "step": 14498 + }, + { + "epoch": 6.07, + "learning_rate": 2.8960411962594558e-06, + "loss": 0.0752, + "step": 14499 + }, + { + "epoch": 6.07, + "learning_rate": 2.894848065676017e-06, + "loss": 0.0973, + "step": 14500 + }, + { + "epoch": 6.07, + "learning_rate": 2.8936551393238664e-06, + "loss": 0.0801, + "step": 14501 + }, + { + "epoch": 6.07, + "learning_rate": 2.892462417237292e-06, + "loss": 0.0798, + "step": 14502 + }, + { + "epoch": 6.07, + "learning_rate": 2.8912698994505816e-06, + "loss": 0.0809, + "step": 14503 + }, + { + "epoch": 6.07, + "learning_rate": 2.890077585998007e-06, + "loss": 0.0906, + "step": 14504 + }, + { + "epoch": 6.07, + "learning_rate": 2.888885476913842e-06, + "loss": 0.0601, + "step": 14505 + }, + { + "epoch": 6.07, + "learning_rate": 2.887693572232355e-06, + "loss": 0.0999, + "step": 14506 + }, + { + "epoch": 6.07, + "learning_rate": 2.8865018719878034e-06, + "loss": 0.0891, + "step": 14507 + }, + { + "epoch": 6.07, + "learning_rate": 2.8853103762144443e-06, + "loss": 0.0517, + "step": 14508 + }, + { + "epoch": 6.07, + "learning_rate": 2.8841190849465282e-06, + "loss": 0.0721, + "step": 14509 + }, + { + "epoch": 6.07, + "learning_rate": 2.882927998218291e-06, + "loss": 0.0974, + "step": 14510 + }, + { + "epoch": 6.08, + "learning_rate": 2.881737116063973e-06, + "loss": 0.0599, + "step": 14511 + }, + { + "epoch": 6.08, + "learning_rate": 2.880546438517804e-06, + "loss": 0.0839, + "step": 14512 + }, + { + "epoch": 6.08, + "learning_rate": 2.8793559656140114e-06, + "loss": 0.1268, + "step": 14513 + }, + { + "epoch": 6.08, + "learning_rate": 2.878165697386812e-06, + "loss": 0.0802, + "step": 14514 + }, + { + "epoch": 6.08, + "learning_rate": 2.8769756338704235e-06, + "loss": 0.0703, + "step": 14515 + }, + { + "epoch": 6.08, + "learning_rate": 2.875785775099046e-06, + "loss": 0.0828, + "step": 14516 + }, + { + "epoch": 6.08, + "learning_rate": 2.8745961211068873e-06, + "loss": 0.1026, + "step": 14517 + }, + { + "epoch": 6.08, + "learning_rate": 2.8734066719281372e-06, + "loss": 0.0841, + "step": 14518 + }, + { + "epoch": 6.08, + "learning_rate": 2.8722174275969884e-06, + "loss": 0.1171, + "step": 14519 + }, + { + "epoch": 6.08, + "learning_rate": 2.871028388147624e-06, + "loss": 0.1032, + "step": 14520 + }, + { + "epoch": 6.08, + "learning_rate": 2.8698395536142232e-06, + "loss": 0.0571, + "step": 14521 + }, + { + "epoch": 6.08, + "learning_rate": 2.868650924030958e-06, + "loss": 0.084, + "step": 14522 + }, + { + "epoch": 6.08, + "learning_rate": 2.8674624994319954e-06, + "loss": 0.0753, + "step": 14523 + }, + { + "epoch": 6.08, + "learning_rate": 2.866274279851491e-06, + "loss": 0.0792, + "step": 14524 + }, + { + "epoch": 6.08, + "learning_rate": 2.8650862653236034e-06, + "loss": 0.097, + "step": 14525 + }, + { + "epoch": 6.08, + "learning_rate": 2.8638984558824777e-06, + "loss": 0.0608, + "step": 14526 + }, + { + "epoch": 6.08, + "learning_rate": 2.8627108515622603e-06, + "loss": 0.1011, + "step": 14527 + }, + { + "epoch": 6.08, + "learning_rate": 2.861523452397085e-06, + "loss": 0.0821, + "step": 14528 + }, + { + "epoch": 6.08, + "learning_rate": 2.8603362584210838e-06, + "loss": 0.0893, + "step": 14529 + }, + { + "epoch": 6.08, + "learning_rate": 2.8591492696683833e-06, + "loss": 0.0995, + "step": 14530 + }, + { + "epoch": 6.08, + "learning_rate": 2.857962486173098e-06, + "loss": 0.0804, + "step": 14531 + }, + { + "epoch": 6.08, + "learning_rate": 2.8567759079693435e-06, + "loss": 0.101, + "step": 14532 + }, + { + "epoch": 6.08, + "learning_rate": 2.855589535091227e-06, + "loss": 0.0949, + "step": 14533 + }, + { + "epoch": 6.08, + "learning_rate": 2.854403367572849e-06, + "loss": 0.0694, + "step": 14534 + }, + { + "epoch": 6.09, + "learning_rate": 2.8532174054483053e-06, + "loss": 0.0773, + "step": 14535 + }, + { + "epoch": 6.09, + "learning_rate": 2.8520316487516884e-06, + "loss": 0.069, + "step": 14536 + }, + { + "epoch": 6.09, + "learning_rate": 2.850846097517076e-06, + "loss": 0.079, + "step": 14537 + }, + { + "epoch": 6.09, + "learning_rate": 2.849660751778548e-06, + "loss": 0.0623, + "step": 14538 + }, + { + "epoch": 6.09, + "learning_rate": 2.8484756115701773e-06, + "loss": 0.0909, + "step": 14539 + }, + { + "epoch": 6.09, + "learning_rate": 2.847290676926031e-06, + "loss": 0.0821, + "step": 14540 + }, + { + "epoch": 6.09, + "learning_rate": 2.8461059478801657e-06, + "loss": 0.1235, + "step": 14541 + }, + { + "epoch": 6.09, + "learning_rate": 2.8449214244666345e-06, + "loss": 0.0907, + "step": 14542 + }, + { + "epoch": 6.09, + "learning_rate": 2.84373710671949e-06, + "loss": 0.0718, + "step": 14543 + }, + { + "epoch": 6.09, + "learning_rate": 2.8425529946727725e-06, + "loss": 0.0996, + "step": 14544 + }, + { + "epoch": 6.09, + "learning_rate": 2.841369088360516e-06, + "loss": 0.0839, + "step": 14545 + }, + { + "epoch": 6.09, + "learning_rate": 2.8401853878167528e-06, + "loss": 0.1058, + "step": 14546 + }, + { + "epoch": 6.09, + "learning_rate": 2.839001893075507e-06, + "loss": 0.0762, + "step": 14547 + }, + { + "epoch": 6.09, + "learning_rate": 2.8378186041707977e-06, + "loss": 0.0689, + "step": 14548 + }, + { + "epoch": 6.09, + "learning_rate": 2.836635521136636e-06, + "loss": 0.0865, + "step": 14549 + }, + { + "epoch": 6.09, + "learning_rate": 2.835452644007033e-06, + "loss": 0.0619, + "step": 14550 + }, + { + "epoch": 6.09, + "learning_rate": 2.834269972815983e-06, + "loss": 0.0879, + "step": 14551 + }, + { + "epoch": 6.09, + "learning_rate": 2.833087507597484e-06, + "loss": 0.0751, + "step": 14552 + }, + { + "epoch": 6.09, + "learning_rate": 2.831905248385525e-06, + "loss": 0.0685, + "step": 14553 + }, + { + "epoch": 6.09, + "learning_rate": 2.830723195214088e-06, + "loss": 0.0894, + "step": 14554 + }, + { + "epoch": 6.09, + "learning_rate": 2.8295413481171517e-06, + "loss": 0.0732, + "step": 14555 + }, + { + "epoch": 6.09, + "learning_rate": 2.8283597071286897e-06, + "loss": 0.0846, + "step": 14556 + }, + { + "epoch": 6.09, + "learning_rate": 2.82717827228266e-06, + "loss": 0.0812, + "step": 14557 + }, + { + "epoch": 6.1, + "learning_rate": 2.825997043613026e-06, + "loss": 0.1015, + "step": 14558 + }, + { + "epoch": 6.1, + "learning_rate": 2.824816021153741e-06, + "loss": 0.0906, + "step": 14559 + }, + { + "epoch": 6.1, + "learning_rate": 2.8236352049387515e-06, + "loss": 0.0706, + "step": 14560 + }, + { + "epoch": 6.1, + "learning_rate": 2.822454595002001e-06, + "loss": 0.11, + "step": 14561 + }, + { + "epoch": 6.1, + "learning_rate": 2.821274191377423e-06, + "loss": 0.0869, + "step": 14562 + }, + { + "epoch": 6.1, + "learning_rate": 2.8200939940989524e-06, + "loss": 0.0942, + "step": 14563 + }, + { + "epoch": 6.1, + "learning_rate": 2.818914003200508e-06, + "loss": 0.0914, + "step": 14564 + }, + { + "epoch": 6.1, + "learning_rate": 2.817734218716005e-06, + "loss": 0.0794, + "step": 14565 + }, + { + "epoch": 6.1, + "learning_rate": 2.8165546406793586e-06, + "loss": 0.0576, + "step": 14566 + }, + { + "epoch": 6.1, + "learning_rate": 2.815375269124474e-06, + "loss": 0.0971, + "step": 14567 + }, + { + "epoch": 6.1, + "learning_rate": 2.814196104085253e-06, + "loss": 0.0714, + "step": 14568 + }, + { + "epoch": 6.1, + "learning_rate": 2.813017145595588e-06, + "loss": 0.0884, + "step": 14569 + }, + { + "epoch": 6.1, + "learning_rate": 2.811838393689367e-06, + "loss": 0.0818, + "step": 14570 + }, + { + "epoch": 6.1, + "learning_rate": 2.8106598484004756e-06, + "loss": 0.1187, + "step": 14571 + }, + { + "epoch": 6.1, + "learning_rate": 2.8094815097627846e-06, + "loss": 0.0948, + "step": 14572 + }, + { + "epoch": 6.1, + "learning_rate": 2.8083033778101664e-06, + "loss": 0.0907, + "step": 14573 + }, + { + "epoch": 6.1, + "learning_rate": 2.807125452576487e-06, + "loss": 0.1035, + "step": 14574 + }, + { + "epoch": 6.1, + "learning_rate": 2.805947734095602e-06, + "loss": 0.0851, + "step": 14575 + }, + { + "epoch": 6.1, + "learning_rate": 2.804770222401366e-06, + "loss": 0.1148, + "step": 14576 + }, + { + "epoch": 6.1, + "learning_rate": 2.803592917527629e-06, + "loss": 0.0831, + "step": 14577 + }, + { + "epoch": 6.1, + "learning_rate": 2.802415819508223e-06, + "loss": 0.0842, + "step": 14578 + }, + { + "epoch": 6.1, + "learning_rate": 2.8012389283769883e-06, + "loss": 0.0846, + "step": 14579 + }, + { + "epoch": 6.1, + "learning_rate": 2.8000622441677527e-06, + "loss": 0.0884, + "step": 14580 + }, + { + "epoch": 6.1, + "learning_rate": 2.798885766914339e-06, + "loss": 0.0936, + "step": 14581 + }, + { + "epoch": 6.11, + "learning_rate": 2.7977094966505624e-06, + "loss": 0.081, + "step": 14582 + }, + { + "epoch": 6.11, + "learning_rate": 2.796533433410239e-06, + "loss": 0.0873, + "step": 14583 + }, + { + "epoch": 6.11, + "learning_rate": 2.7953575772271667e-06, + "loss": 0.0961, + "step": 14584 + }, + { + "epoch": 6.11, + "learning_rate": 2.7941819281351513e-06, + "loss": 0.0902, + "step": 14585 + }, + { + "epoch": 6.11, + "learning_rate": 2.7930064861679783e-06, + "loss": 0.0795, + "step": 14586 + }, + { + "epoch": 6.11, + "learning_rate": 2.791831251359438e-06, + "loss": 0.0775, + "step": 14587 + }, + { + "epoch": 6.11, + "learning_rate": 2.790656223743312e-06, + "loss": 0.0974, + "step": 14588 + }, + { + "epoch": 6.11, + "learning_rate": 2.7894814033533768e-06, + "loss": 0.0759, + "step": 14589 + }, + { + "epoch": 6.11, + "learning_rate": 2.7883067902233985e-06, + "loss": 0.1007, + "step": 14590 + }, + { + "epoch": 6.11, + "learning_rate": 2.7871323843871467e-06, + "loss": 0.0767, + "step": 14591 + }, + { + "epoch": 6.11, + "learning_rate": 2.7859581858783693e-06, + "loss": 0.0675, + "step": 14592 + }, + { + "epoch": 6.11, + "learning_rate": 2.784784194730823e-06, + "loss": 0.1, + "step": 14593 + }, + { + "epoch": 6.11, + "learning_rate": 2.7836104109782525e-06, + "loss": 0.1084, + "step": 14594 + }, + { + "epoch": 6.11, + "learning_rate": 2.7824368346543963e-06, + "loss": 0.0986, + "step": 14595 + }, + { + "epoch": 6.11, + "learning_rate": 2.781263465792989e-06, + "loss": 0.078, + "step": 14596 + }, + { + "epoch": 6.11, + "learning_rate": 2.7800903044277604e-06, + "loss": 0.08, + "step": 14597 + }, + { + "epoch": 6.11, + "learning_rate": 2.7789173505924274e-06, + "loss": 0.0671, + "step": 14598 + }, + { + "epoch": 6.11, + "learning_rate": 2.777744604320706e-06, + "loss": 0.0647, + "step": 14599 + }, + { + "epoch": 6.11, + "learning_rate": 2.776572065646307e-06, + "loss": 0.0704, + "step": 14600 + }, + { + "epoch": 6.11, + "learning_rate": 2.775399734602934e-06, + "loss": 0.1009, + "step": 14601 + }, + { + "epoch": 6.11, + "learning_rate": 2.774227611224285e-06, + "loss": 0.1078, + "step": 14602 + }, + { + "epoch": 6.11, + "learning_rate": 2.7730556955440515e-06, + "loss": 0.0757, + "step": 14603 + }, + { + "epoch": 6.11, + "learning_rate": 2.7718839875959213e-06, + "loss": 0.0806, + "step": 14604 + }, + { + "epoch": 6.11, + "learning_rate": 2.7707124874135694e-06, + "loss": 0.0631, + "step": 14605 + }, + { + "epoch": 6.12, + "learning_rate": 2.7695411950306714e-06, + "loss": 0.1014, + "step": 14606 + }, + { + "epoch": 6.12, + "learning_rate": 2.7683701104808956e-06, + "loss": 0.0764, + "step": 14607 + }, + { + "epoch": 6.12, + "learning_rate": 2.767199233797907e-06, + "loss": 0.1016, + "step": 14608 + }, + { + "epoch": 6.12, + "learning_rate": 2.766028565015355e-06, + "loss": 0.0935, + "step": 14609 + }, + { + "epoch": 6.12, + "learning_rate": 2.7648581041668922e-06, + "loss": 0.0877, + "step": 14610 + }, + { + "epoch": 6.12, + "learning_rate": 2.7636878512861633e-06, + "loss": 0.0711, + "step": 14611 + }, + { + "epoch": 6.12, + "learning_rate": 2.7625178064068094e-06, + "loss": 0.0836, + "step": 14612 + }, + { + "epoch": 6.12, + "learning_rate": 2.7613479695624544e-06, + "loss": 0.0938, + "step": 14613 + }, + { + "epoch": 6.12, + "learning_rate": 2.760178340786729e-06, + "loss": 0.0893, + "step": 14614 + }, + { + "epoch": 6.12, + "learning_rate": 2.759008920113253e-06, + "loss": 0.096, + "step": 14615 + }, + { + "epoch": 6.12, + "learning_rate": 2.7578397075756404e-06, + "loss": 0.0646, + "step": 14616 + }, + { + "epoch": 6.12, + "learning_rate": 2.756670703207498e-06, + "loss": 0.0842, + "step": 14617 + }, + { + "epoch": 6.12, + "learning_rate": 2.7555019070424316e-06, + "loss": 0.0547, + "step": 14618 + }, + { + "epoch": 6.12, + "learning_rate": 2.754333319114032e-06, + "loss": 0.0824, + "step": 14619 + }, + { + "epoch": 6.12, + "learning_rate": 2.753164939455891e-06, + "loss": 0.0972, + "step": 14620 + }, + { + "epoch": 6.12, + "learning_rate": 2.7519967681015933e-06, + "loss": 0.0783, + "step": 14621 + }, + { + "epoch": 6.12, + "learning_rate": 2.7508288050847166e-06, + "loss": 0.1054, + "step": 14622 + }, + { + "epoch": 6.12, + "learning_rate": 2.7496610504388334e-06, + "loss": 0.0799, + "step": 14623 + }, + { + "epoch": 6.12, + "learning_rate": 2.7484935041975113e-06, + "loss": 0.0862, + "step": 14624 + }, + { + "epoch": 6.12, + "learning_rate": 2.7473261663943063e-06, + "loss": 0.1045, + "step": 14625 + }, + { + "epoch": 6.12, + "learning_rate": 2.7461590370627753e-06, + "loss": 0.0807, + "step": 14626 + }, + { + "epoch": 6.12, + "learning_rate": 2.7449921162364647e-06, + "loss": 0.0825, + "step": 14627 + }, + { + "epoch": 6.12, + "learning_rate": 2.7438254039489185e-06, + "loss": 0.0802, + "step": 14628 + }, + { + "epoch": 6.12, + "learning_rate": 2.7426589002336723e-06, + "loss": 0.0778, + "step": 14629 + }, + { + "epoch": 6.13, + "learning_rate": 2.741492605124255e-06, + "loss": 0.0931, + "step": 14630 + }, + { + "epoch": 6.13, + "learning_rate": 2.7403265186541963e-06, + "loss": 0.0809, + "step": 14631 + }, + { + "epoch": 6.13, + "learning_rate": 2.7391606408570084e-06, + "loss": 0.0576, + "step": 14632 + }, + { + "epoch": 6.13, + "learning_rate": 2.7379949717662014e-06, + "loss": 0.0951, + "step": 14633 + }, + { + "epoch": 6.13, + "learning_rate": 2.7368295114152852e-06, + "loss": 0.1012, + "step": 14634 + }, + { + "epoch": 6.13, + "learning_rate": 2.7356642598377604e-06, + "loss": 0.0901, + "step": 14635 + }, + { + "epoch": 6.13, + "learning_rate": 2.734499217067119e-06, + "loss": 0.0891, + "step": 14636 + }, + { + "epoch": 6.13, + "learning_rate": 2.733334383136851e-06, + "loss": 0.0668, + "step": 14637 + }, + { + "epoch": 6.13, + "learning_rate": 2.7321697580804407e-06, + "loss": 0.0831, + "step": 14638 + }, + { + "epoch": 6.13, + "learning_rate": 2.7310053419313576e-06, + "loss": 0.0706, + "step": 14639 + }, + { + "epoch": 6.13, + "learning_rate": 2.729841134723077e-06, + "loss": 0.0909, + "step": 14640 + }, + { + "epoch": 6.13, + "learning_rate": 2.7286771364890606e-06, + "loss": 0.0715, + "step": 14641 + }, + { + "epoch": 6.13, + "learning_rate": 2.727513347262767e-06, + "loss": 0.0648, + "step": 14642 + }, + { + "epoch": 6.13, + "learning_rate": 2.7263497670776495e-06, + "loss": 0.103, + "step": 14643 + }, + { + "epoch": 6.13, + "learning_rate": 2.7251863959671533e-06, + "loss": 0.0876, + "step": 14644 + }, + { + "epoch": 6.13, + "learning_rate": 2.724023233964721e-06, + "loss": 0.0984, + "step": 14645 + }, + { + "epoch": 6.13, + "learning_rate": 2.722860281103782e-06, + "loss": 0.0727, + "step": 14646 + }, + { + "epoch": 6.13, + "learning_rate": 2.721697537417767e-06, + "loss": 0.1052, + "step": 14647 + }, + { + "epoch": 6.13, + "learning_rate": 2.7205350029400966e-06, + "loss": 0.0902, + "step": 14648 + }, + { + "epoch": 6.13, + "learning_rate": 2.7193726777041885e-06, + "loss": 0.0884, + "step": 14649 + }, + { + "epoch": 6.13, + "learning_rate": 2.7182105617434516e-06, + "loss": 0.1211, + "step": 14650 + }, + { + "epoch": 6.13, + "learning_rate": 2.717048655091293e-06, + "loss": 0.0977, + "step": 14651 + }, + { + "epoch": 6.13, + "learning_rate": 2.7158869577811064e-06, + "loss": 0.0991, + "step": 14652 + }, + { + "epoch": 6.13, + "learning_rate": 2.7147254698462875e-06, + "loss": 0.0861, + "step": 14653 + }, + { + "epoch": 6.14, + "learning_rate": 2.713564191320217e-06, + "loss": 0.0808, + "step": 14654 + }, + { + "epoch": 6.14, + "learning_rate": 2.712403122236279e-06, + "loss": 0.0791, + "step": 14655 + }, + { + "epoch": 6.14, + "learning_rate": 2.7112422626278456e-06, + "loss": 0.0886, + "step": 14656 + }, + { + "epoch": 6.14, + "learning_rate": 2.7100816125282846e-06, + "loss": 0.0898, + "step": 14657 + }, + { + "epoch": 6.14, + "learning_rate": 2.7089211719709596e-06, + "loss": 0.0485, + "step": 14658 + }, + { + "epoch": 6.14, + "learning_rate": 2.707760940989229e-06, + "loss": 0.0637, + "step": 14659 + }, + { + "epoch": 6.14, + "learning_rate": 2.7066009196164355e-06, + "loss": 0.0994, + "step": 14660 + }, + { + "epoch": 6.14, + "learning_rate": 2.7054411078859267e-06, + "loss": 0.1317, + "step": 14661 + }, + { + "epoch": 6.14, + "learning_rate": 2.704281505831039e-06, + "loss": 0.0502, + "step": 14662 + }, + { + "epoch": 6.14, + "learning_rate": 2.7031221134851073e-06, + "loss": 0.0918, + "step": 14663 + }, + { + "epoch": 6.14, + "learning_rate": 2.7019629308814544e-06, + "loss": 0.081, + "step": 14664 + }, + { + "epoch": 6.14, + "learning_rate": 2.700803958053403e-06, + "loss": 0.1076, + "step": 14665 + }, + { + "epoch": 6.14, + "learning_rate": 2.699645195034263e-06, + "loss": 0.0651, + "step": 14666 + }, + { + "epoch": 6.14, + "learning_rate": 2.698486641857343e-06, + "loss": 0.0671, + "step": 14667 + }, + { + "epoch": 6.14, + "learning_rate": 2.6973282985559458e-06, + "loss": 0.1349, + "step": 14668 + }, + { + "epoch": 6.14, + "learning_rate": 2.6961701651633653e-06, + "loss": 0.0904, + "step": 14669 + }, + { + "epoch": 6.14, + "learning_rate": 2.6950122417128933e-06, + "loss": 0.0759, + "step": 14670 + }, + { + "epoch": 6.14, + "learning_rate": 2.6938545282378117e-06, + "loss": 0.0651, + "step": 14671 + }, + { + "epoch": 6.14, + "learning_rate": 2.6926970247714e-06, + "loss": 0.0691, + "step": 14672 + }, + { + "epoch": 6.14, + "learning_rate": 2.691539731346927e-06, + "loss": 0.1042, + "step": 14673 + }, + { + "epoch": 6.14, + "learning_rate": 2.6903826479976582e-06, + "loss": 0.0805, + "step": 14674 + }, + { + "epoch": 6.14, + "learning_rate": 2.689225774756853e-06, + "loss": 0.0479, + "step": 14675 + }, + { + "epoch": 6.14, + "learning_rate": 2.6880691116577694e-06, + "loss": 0.1131, + "step": 14676 + }, + { + "epoch": 6.14, + "learning_rate": 2.686912658733647e-06, + "loss": 0.1021, + "step": 14677 + }, + { + "epoch": 6.15, + "learning_rate": 2.685756416017732e-06, + "loss": 0.1051, + "step": 14678 + }, + { + "epoch": 6.15, + "learning_rate": 2.6846003835432597e-06, + "loss": 0.0603, + "step": 14679 + }, + { + "epoch": 6.15, + "learning_rate": 2.6834445613434555e-06, + "loss": 0.0844, + "step": 14680 + }, + { + "epoch": 6.15, + "learning_rate": 2.6822889494515445e-06, + "loss": 0.109, + "step": 14681 + }, + { + "epoch": 6.15, + "learning_rate": 2.681133547900745e-06, + "loss": 0.1006, + "step": 14682 + }, + { + "epoch": 6.15, + "learning_rate": 2.6799783567242653e-06, + "loss": 0.0806, + "step": 14683 + }, + { + "epoch": 6.15, + "learning_rate": 2.678823375955314e-06, + "loss": 0.0697, + "step": 14684 + }, + { + "epoch": 6.15, + "learning_rate": 2.677668605627086e-06, + "loss": 0.1013, + "step": 14685 + }, + { + "epoch": 6.15, + "learning_rate": 2.6765140457727813e-06, + "loss": 0.0662, + "step": 14686 + }, + { + "epoch": 6.15, + "learning_rate": 2.675359696425578e-06, + "loss": 0.0848, + "step": 14687 + }, + { + "epoch": 6.15, + "learning_rate": 2.67420555761866e-06, + "loss": 0.0681, + "step": 14688 + }, + { + "epoch": 6.15, + "learning_rate": 2.6730516293852014e-06, + "loss": 0.093, + "step": 14689 + }, + { + "epoch": 6.15, + "learning_rate": 2.671897911758373e-06, + "loss": 0.0732, + "step": 14690 + }, + { + "epoch": 6.15, + "learning_rate": 2.6707444047713373e-06, + "loss": 0.1054, + "step": 14691 + }, + { + "epoch": 6.15, + "learning_rate": 2.6695911084572514e-06, + "loss": 0.0781, + "step": 14692 + }, + { + "epoch": 6.15, + "learning_rate": 2.668438022849262e-06, + "loss": 0.0715, + "step": 14693 + }, + { + "epoch": 6.15, + "learning_rate": 2.667285147980515e-06, + "loss": 0.0876, + "step": 14694 + }, + { + "epoch": 6.15, + "learning_rate": 2.6661324838841495e-06, + "loss": 0.0982, + "step": 14695 + }, + { + "epoch": 6.15, + "learning_rate": 2.6649800305932984e-06, + "loss": 0.1052, + "step": 14696 + }, + { + "epoch": 6.15, + "learning_rate": 2.663827788141088e-06, + "loss": 0.0998, + "step": 14697 + }, + { + "epoch": 6.15, + "learning_rate": 2.6626757565606398e-06, + "loss": 0.0784, + "step": 14698 + }, + { + "epoch": 6.15, + "learning_rate": 2.661523935885063e-06, + "loss": 0.0928, + "step": 14699 + }, + { + "epoch": 6.15, + "learning_rate": 2.660372326147471e-06, + "loss": 0.0695, + "step": 14700 + }, + { + "epoch": 6.15, + "learning_rate": 2.659220927380961e-06, + "loss": 0.0439, + "step": 14701 + }, + { + "epoch": 6.16, + "learning_rate": 2.658069739618632e-06, + "loss": 0.0846, + "step": 14702 + }, + { + "epoch": 6.16, + "learning_rate": 2.6569187628935723e-06, + "loss": 0.072, + "step": 14703 + }, + { + "epoch": 6.16, + "learning_rate": 2.6557679972388674e-06, + "loss": 0.0804, + "step": 14704 + }, + { + "epoch": 6.16, + "learning_rate": 2.6546174426875937e-06, + "loss": 0.0963, + "step": 14705 + }, + { + "epoch": 6.16, + "learning_rate": 2.653467099272826e-06, + "loss": 0.0867, + "step": 14706 + }, + { + "epoch": 6.16, + "learning_rate": 2.652316967027625e-06, + "loss": 0.0503, + "step": 14707 + }, + { + "epoch": 6.16, + "learning_rate": 2.6511670459850526e-06, + "loss": 0.0855, + "step": 14708 + }, + { + "epoch": 6.16, + "learning_rate": 2.650017336178162e-06, + "loss": 0.0748, + "step": 14709 + }, + { + "epoch": 6.16, + "learning_rate": 2.648867837640001e-06, + "loss": 0.0822, + "step": 14710 + }, + { + "epoch": 6.16, + "learning_rate": 2.6477185504036106e-06, + "loss": 0.0679, + "step": 14711 + }, + { + "epoch": 6.16, + "learning_rate": 2.6465694745020265e-06, + "loss": 0.0653, + "step": 14712 + }, + { + "epoch": 6.16, + "learning_rate": 2.64542060996828e-06, + "loss": 0.0884, + "step": 14713 + }, + { + "epoch": 6.16, + "learning_rate": 2.6442719568353894e-06, + "loss": 0.0766, + "step": 14714 + }, + { + "epoch": 6.16, + "learning_rate": 2.643123515136373e-06, + "loss": 0.0957, + "step": 14715 + }, + { + "epoch": 6.16, + "learning_rate": 2.6419752849042437e-06, + "loss": 0.0947, + "step": 14716 + }, + { + "epoch": 6.16, + "learning_rate": 2.640827266172006e-06, + "loss": 0.1111, + "step": 14717 + }, + { + "epoch": 6.16, + "learning_rate": 2.639679458972657e-06, + "loss": 0.0927, + "step": 14718 + }, + { + "epoch": 6.16, + "learning_rate": 2.6385318633391934e-06, + "loss": 0.04, + "step": 14719 + }, + { + "epoch": 6.16, + "learning_rate": 2.6373844793045965e-06, + "loss": 0.0662, + "step": 14720 + }, + { + "epoch": 6.16, + "learning_rate": 2.636237306901852e-06, + "loss": 0.0734, + "step": 14721 + }, + { + "epoch": 6.16, + "learning_rate": 2.6350903461639287e-06, + "loss": 0.0683, + "step": 14722 + }, + { + "epoch": 6.16, + "learning_rate": 2.633943597123798e-06, + "loss": 0.0921, + "step": 14723 + }, + { + "epoch": 6.16, + "learning_rate": 2.6327970598144214e-06, + "loss": 0.0751, + "step": 14724 + }, + { + "epoch": 6.16, + "learning_rate": 2.631650734268757e-06, + "loss": 0.0901, + "step": 14725 + }, + { + "epoch": 6.17, + "learning_rate": 2.630504620519753e-06, + "loss": 0.095, + "step": 14726 + }, + { + "epoch": 6.17, + "learning_rate": 2.6293587186003568e-06, + "loss": 0.082, + "step": 14727 + }, + { + "epoch": 6.17, + "learning_rate": 2.6282130285435016e-06, + "loss": 0.1297, + "step": 14728 + }, + { + "epoch": 6.17, + "learning_rate": 2.6270675503821207e-06, + "loss": 0.0839, + "step": 14729 + }, + { + "epoch": 6.17, + "learning_rate": 2.6259222841491406e-06, + "loss": 0.0984, + "step": 14730 + }, + { + "epoch": 6.17, + "learning_rate": 2.6247772298774797e-06, + "loss": 0.0922, + "step": 14731 + }, + { + "epoch": 6.17, + "learning_rate": 2.6236323876000547e-06, + "loss": 0.0765, + "step": 14732 + }, + { + "epoch": 6.17, + "learning_rate": 2.622487757349772e-06, + "loss": 0.09, + "step": 14733 + }, + { + "epoch": 6.17, + "learning_rate": 2.62134333915953e-06, + "loss": 0.092, + "step": 14734 + }, + { + "epoch": 6.17, + "learning_rate": 2.620199133062227e-06, + "loss": 0.1081, + "step": 14735 + }, + { + "epoch": 6.17, + "learning_rate": 2.6190551390907495e-06, + "loss": 0.0826, + "step": 14736 + }, + { + "epoch": 6.17, + "learning_rate": 2.6179113572779835e-06, + "loss": 0.0785, + "step": 14737 + }, + { + "epoch": 6.17, + "learning_rate": 2.6167677876568044e-06, + "loss": 0.1054, + "step": 14738 + }, + { + "epoch": 6.17, + "learning_rate": 2.6156244302600864e-06, + "loss": 0.0771, + "step": 14739 + }, + { + "epoch": 6.17, + "learning_rate": 2.614481285120688e-06, + "loss": 0.0614, + "step": 14740 + }, + { + "epoch": 6.17, + "learning_rate": 2.6133383522714718e-06, + "loss": 0.0825, + "step": 14741 + }, + { + "epoch": 6.17, + "learning_rate": 2.612195631745289e-06, + "loss": 0.1003, + "step": 14742 + }, + { + "epoch": 6.17, + "learning_rate": 2.611053123574988e-06, + "loss": 0.0942, + "step": 14743 + }, + { + "epoch": 6.17, + "learning_rate": 2.6099108277934105e-06, + "loss": 0.1015, + "step": 14744 + }, + { + "epoch": 6.17, + "learning_rate": 2.6087687444333853e-06, + "loss": 0.0999, + "step": 14745 + }, + { + "epoch": 6.17, + "learning_rate": 2.6076268735277445e-06, + "loss": 0.068, + "step": 14746 + }, + { + "epoch": 6.17, + "learning_rate": 2.6064852151093124e-06, + "loss": 0.115, + "step": 14747 + }, + { + "epoch": 6.17, + "learning_rate": 2.6053437692108974e-06, + "loss": 0.0829, + "step": 14748 + }, + { + "epoch": 6.18, + "learning_rate": 2.6042025358653157e-06, + "loss": 0.0689, + "step": 14749 + }, + { + "epoch": 6.18, + "learning_rate": 2.6030615151053675e-06, + "loss": 0.0944, + "step": 14750 + }, + { + "epoch": 6.18, + "learning_rate": 2.601920706963853e-06, + "loss": 0.0837, + "step": 14751 + }, + { + "epoch": 6.18, + "learning_rate": 2.600780111473563e-06, + "loss": 0.0939, + "step": 14752 + }, + { + "epoch": 6.18, + "learning_rate": 2.5996397286672826e-06, + "loss": 0.1301, + "step": 14753 + }, + { + "epoch": 6.18, + "learning_rate": 2.598499558577794e-06, + "loss": 0.0731, + "step": 14754 + }, + { + "epoch": 6.18, + "learning_rate": 2.5973596012378644e-06, + "loss": 0.099, + "step": 14755 + }, + { + "epoch": 6.18, + "learning_rate": 2.5962198566802643e-06, + "loss": 0.0624, + "step": 14756 + }, + { + "epoch": 6.18, + "learning_rate": 2.5950803249377554e-06, + "loss": 0.0775, + "step": 14757 + }, + { + "epoch": 6.18, + "learning_rate": 2.5939410060430904e-06, + "loss": 0.0849, + "step": 14758 + }, + { + "epoch": 6.18, + "learning_rate": 2.5928019000290194e-06, + "loss": 0.0922, + "step": 14759 + }, + { + "epoch": 6.18, + "learning_rate": 2.591663006928288e-06, + "loss": 0.045, + "step": 14760 + }, + { + "epoch": 6.18, + "learning_rate": 2.5905243267736256e-06, + "loss": 0.1222, + "step": 14761 + }, + { + "epoch": 6.18, + "learning_rate": 2.589385859597767e-06, + "loss": 0.0858, + "step": 14762 + }, + { + "epoch": 6.18, + "learning_rate": 2.588247605433436e-06, + "loss": 0.0735, + "step": 14763 + }, + { + "epoch": 6.18, + "learning_rate": 2.5871095643133502e-06, + "loss": 0.0729, + "step": 14764 + }, + { + "epoch": 6.18, + "learning_rate": 2.585971736270223e-06, + "loss": 0.0955, + "step": 14765 + }, + { + "epoch": 6.18, + "learning_rate": 2.584834121336761e-06, + "loss": 0.0926, + "step": 14766 + }, + { + "epoch": 6.18, + "learning_rate": 2.5836967195456586e-06, + "loss": 0.0997, + "step": 14767 + }, + { + "epoch": 6.18, + "learning_rate": 2.582559530929617e-06, + "loss": 0.0763, + "step": 14768 + }, + { + "epoch": 6.18, + "learning_rate": 2.581422555521316e-06, + "loss": 0.0615, + "step": 14769 + }, + { + "epoch": 6.18, + "learning_rate": 2.5802857933534407e-06, + "loss": 0.0617, + "step": 14770 + }, + { + "epoch": 6.18, + "learning_rate": 2.579149244458666e-06, + "loss": 0.0791, + "step": 14771 + }, + { + "epoch": 6.18, + "learning_rate": 2.5780129088696614e-06, + "loss": 0.0943, + "step": 14772 + }, + { + "epoch": 6.19, + "learning_rate": 2.5768767866190893e-06, + "loss": 0.1239, + "step": 14773 + }, + { + "epoch": 6.19, + "learning_rate": 2.57574087773961e-06, + "loss": 0.101, + "step": 14774 + }, + { + "epoch": 6.19, + "learning_rate": 2.5746051822638675e-06, + "loss": 0.0708, + "step": 14775 + }, + { + "epoch": 6.19, + "learning_rate": 2.57346970022451e-06, + "loss": 0.0843, + "step": 14776 + }, + { + "epoch": 6.19, + "learning_rate": 2.5723344316541767e-06, + "loss": 0.0627, + "step": 14777 + }, + { + "epoch": 6.19, + "learning_rate": 2.5711993765854983e-06, + "loss": 0.0937, + "step": 14778 + }, + { + "epoch": 6.19, + "learning_rate": 2.570064535051101e-06, + "loss": 0.1021, + "step": 14779 + }, + { + "epoch": 6.19, + "learning_rate": 2.568929907083609e-06, + "loss": 0.0855, + "step": 14780 + }, + { + "epoch": 6.19, + "learning_rate": 2.56779549271563e-06, + "loss": 0.084, + "step": 14781 + }, + { + "epoch": 6.19, + "learning_rate": 2.566661291979774e-06, + "loss": 0.0663, + "step": 14782 + }, + { + "epoch": 6.19, + "learning_rate": 2.5655273049086425e-06, + "loss": 0.0879, + "step": 14783 + }, + { + "epoch": 6.19, + "learning_rate": 2.564393531534832e-06, + "loss": 0.0717, + "step": 14784 + }, + { + "epoch": 6.19, + "learning_rate": 2.563259971890931e-06, + "loss": 0.0717, + "step": 14785 + }, + { + "epoch": 6.19, + "learning_rate": 2.562126626009522e-06, + "loss": 0.1152, + "step": 14786 + }, + { + "epoch": 6.19, + "learning_rate": 2.5609934939231873e-06, + "loss": 0.0707, + "step": 14787 + }, + { + "epoch": 6.19, + "learning_rate": 2.5598605756644902e-06, + "loss": 0.0919, + "step": 14788 + }, + { + "epoch": 6.19, + "learning_rate": 2.558727871266001e-06, + "loss": 0.0781, + "step": 14789 + }, + { + "epoch": 6.19, + "learning_rate": 2.5575953807602727e-06, + "loss": 0.0687, + "step": 14790 + }, + { + "epoch": 6.19, + "learning_rate": 2.5564631041798615e-06, + "loss": 0.0992, + "step": 14791 + }, + { + "epoch": 6.19, + "learning_rate": 2.555331041557313e-06, + "loss": 0.1118, + "step": 14792 + }, + { + "epoch": 6.19, + "learning_rate": 2.5541991929251676e-06, + "loss": 0.0891, + "step": 14793 + }, + { + "epoch": 6.19, + "learning_rate": 2.553067558315958e-06, + "loss": 0.1247, + "step": 14794 + }, + { + "epoch": 6.19, + "learning_rate": 2.5519361377622176e-06, + "loss": 0.0741, + "step": 14795 + }, + { + "epoch": 6.19, + "learning_rate": 2.55080493129646e-06, + "loss": 0.0669, + "step": 14796 + }, + { + "epoch": 6.2, + "learning_rate": 2.5496739389512036e-06, + "loss": 0.0552, + "step": 14797 + }, + { + "epoch": 6.2, + "learning_rate": 2.5485431607589594e-06, + "loss": 0.0481, + "step": 14798 + }, + { + "epoch": 6.2, + "learning_rate": 2.54741259675223e-06, + "loss": 0.0748, + "step": 14799 + }, + { + "epoch": 6.2, + "learning_rate": 2.5462822469635117e-06, + "loss": 0.0843, + "step": 14800 + }, + { + "epoch": 6.2, + "learning_rate": 2.5451521114252985e-06, + "loss": 0.0868, + "step": 14801 + }, + { + "epoch": 6.2, + "learning_rate": 2.54402219017007e-06, + "loss": 0.0869, + "step": 14802 + }, + { + "epoch": 6.2, + "learning_rate": 2.542892483230308e-06, + "loss": 0.0633, + "step": 14803 + }, + { + "epoch": 6.2, + "learning_rate": 2.5417629906384834e-06, + "loss": 0.1067, + "step": 14804 + }, + { + "epoch": 6.2, + "learning_rate": 2.540633712427063e-06, + "loss": 0.0969, + "step": 14805 + }, + { + "epoch": 6.2, + "learning_rate": 2.539504648628508e-06, + "loss": 0.0993, + "step": 14806 + }, + { + "epoch": 6.2, + "learning_rate": 2.5383757992752744e-06, + "loss": 0.1001, + "step": 14807 + }, + { + "epoch": 6.2, + "learning_rate": 2.5372471643998042e-06, + "loss": 0.1042, + "step": 14808 + }, + { + "epoch": 6.2, + "learning_rate": 2.5361187440345416e-06, + "loss": 0.0951, + "step": 14809 + }, + { + "epoch": 6.2, + "learning_rate": 2.534990538211922e-06, + "loss": 0.0607, + "step": 14810 + }, + { + "epoch": 6.2, + "learning_rate": 2.533862546964376e-06, + "loss": 0.0593, + "step": 14811 + }, + { + "epoch": 6.2, + "learning_rate": 2.532734770324329e-06, + "loss": 0.1005, + "step": 14812 + }, + { + "epoch": 6.2, + "learning_rate": 2.5316072083241904e-06, + "loss": 0.0799, + "step": 14813 + }, + { + "epoch": 6.2, + "learning_rate": 2.5304798609963754e-06, + "loss": 0.0865, + "step": 14814 + }, + { + "epoch": 6.2, + "learning_rate": 2.529352728373292e-06, + "loss": 0.0834, + "step": 14815 + }, + { + "epoch": 6.2, + "learning_rate": 2.5282258104873324e-06, + "loss": 0.0644, + "step": 14816 + }, + { + "epoch": 6.2, + "learning_rate": 2.5270991073708905e-06, + "loss": 0.1018, + "step": 14817 + }, + { + "epoch": 6.2, + "learning_rate": 2.525972619056355e-06, + "loss": 0.0698, + "step": 14818 + }, + { + "epoch": 6.2, + "learning_rate": 2.524846345576103e-06, + "loss": 0.0751, + "step": 14819 + }, + { + "epoch": 6.2, + "learning_rate": 2.52372028696251e-06, + "loss": 0.062, + "step": 14820 + }, + { + "epoch": 6.21, + "learning_rate": 2.5225944432479452e-06, + "loss": 0.1017, + "step": 14821 + }, + { + "epoch": 6.21, + "learning_rate": 2.5214688144647658e-06, + "loss": 0.1301, + "step": 14822 + }, + { + "epoch": 6.21, + "learning_rate": 2.5203434006453287e-06, + "loss": 0.1092, + "step": 14823 + }, + { + "epoch": 6.21, + "learning_rate": 2.5192182018219826e-06, + "loss": 0.0819, + "step": 14824 + }, + { + "epoch": 6.21, + "learning_rate": 2.518093218027071e-06, + "loss": 0.0859, + "step": 14825 + }, + { + "epoch": 6.21, + "learning_rate": 2.5169684492929315e-06, + "loss": 0.0949, + "step": 14826 + }, + { + "epoch": 6.21, + "learning_rate": 2.515843895651893e-06, + "loss": 0.1155, + "step": 14827 + }, + { + "epoch": 6.21, + "learning_rate": 2.5147195571362837e-06, + "loss": 0.0719, + "step": 14828 + }, + { + "epoch": 6.21, + "learning_rate": 2.513595433778415e-06, + "loss": 0.0965, + "step": 14829 + }, + { + "epoch": 6.21, + "learning_rate": 2.5124715256106024e-06, + "loss": 0.0823, + "step": 14830 + }, + { + "epoch": 6.21, + "learning_rate": 2.511347832665152e-06, + "loss": 0.0948, + "step": 14831 + }, + { + "epoch": 6.21, + "learning_rate": 2.510224354974362e-06, + "loss": 0.1077, + "step": 14832 + }, + { + "epoch": 6.21, + "learning_rate": 2.509101092570527e-06, + "loss": 0.0804, + "step": 14833 + }, + { + "epoch": 6.21, + "learning_rate": 2.507978045485937e-06, + "loss": 0.0918, + "step": 14834 + }, + { + "epoch": 6.21, + "learning_rate": 2.5068552137528667e-06, + "loss": 0.0652, + "step": 14835 + }, + { + "epoch": 6.21, + "learning_rate": 2.5057325974035964e-06, + "loss": 0.0755, + "step": 14836 + }, + { + "epoch": 6.21, + "learning_rate": 2.5046101964703894e-06, + "loss": 0.0666, + "step": 14837 + }, + { + "epoch": 6.21, + "learning_rate": 2.503488010985512e-06, + "loss": 0.0726, + "step": 14838 + }, + { + "epoch": 6.21, + "learning_rate": 2.502366040981218e-06, + "loss": 0.0622, + "step": 14839 + }, + { + "epoch": 6.21, + "learning_rate": 2.50124428648976e-06, + "loss": 0.0765, + "step": 14840 + }, + { + "epoch": 6.21, + "learning_rate": 2.5001227475433797e-06, + "loss": 0.1044, + "step": 14841 + }, + { + "epoch": 6.21, + "learning_rate": 2.49900142417432e-06, + "loss": 0.0872, + "step": 14842 + }, + { + "epoch": 6.21, + "learning_rate": 2.4978803164148047e-06, + "loss": 0.1062, + "step": 14843 + }, + { + "epoch": 6.21, + "learning_rate": 2.496759424297062e-06, + "loss": 0.0653, + "step": 14844 + }, + { + "epoch": 6.22, + "learning_rate": 2.495638747853312e-06, + "loss": 0.0966, + "step": 14845 + }, + { + "epoch": 6.22, + "learning_rate": 2.4945182871157656e-06, + "loss": 0.1056, + "step": 14846 + }, + { + "epoch": 6.22, + "learning_rate": 2.4933980421166326e-06, + "loss": 0.1049, + "step": 14847 + }, + { + "epoch": 6.22, + "learning_rate": 2.492278012888113e-06, + "loss": 0.0643, + "step": 14848 + }, + { + "epoch": 6.22, + "learning_rate": 2.491158199462398e-06, + "loss": 0.0765, + "step": 14849 + }, + { + "epoch": 6.22, + "learning_rate": 2.490038601871676e-06, + "loss": 0.0877, + "step": 14850 + }, + { + "epoch": 6.22, + "learning_rate": 2.4889192201481315e-06, + "loss": 0.0766, + "step": 14851 + }, + { + "epoch": 6.22, + "learning_rate": 2.4878000543239377e-06, + "loss": 0.0746, + "step": 14852 + }, + { + "epoch": 6.22, + "learning_rate": 2.4866811044312667e-06, + "loss": 0.0629, + "step": 14853 + }, + { + "epoch": 6.22, + "learning_rate": 2.485562370502279e-06, + "loss": 0.0779, + "step": 14854 + }, + { + "epoch": 6.22, + "learning_rate": 2.484443852569135e-06, + "loss": 0.0831, + "step": 14855 + }, + { + "epoch": 6.22, + "learning_rate": 2.483325550663982e-06, + "loss": 0.0806, + "step": 14856 + }, + { + "epoch": 6.22, + "learning_rate": 2.4822074648189675e-06, + "loss": 0.0728, + "step": 14857 + }, + { + "epoch": 6.22, + "learning_rate": 2.481089595066226e-06, + "loss": 0.0995, + "step": 14858 + }, + { + "epoch": 6.22, + "learning_rate": 2.479971941437892e-06, + "loss": 0.0717, + "step": 14859 + }, + { + "epoch": 6.22, + "learning_rate": 2.4788545039660905e-06, + "loss": 0.082, + "step": 14860 + }, + { + "epoch": 6.22, + "learning_rate": 2.4777372826829427e-06, + "loss": 0.1138, + "step": 14861 + }, + { + "epoch": 6.22, + "learning_rate": 2.476620277620565e-06, + "loss": 0.0985, + "step": 14862 + }, + { + "epoch": 6.22, + "learning_rate": 2.4755034888110576e-06, + "loss": 0.1266, + "step": 14863 + }, + { + "epoch": 6.22, + "learning_rate": 2.474386916286526e-06, + "loss": 0.0751, + "step": 14864 + }, + { + "epoch": 6.22, + "learning_rate": 2.473270560079064e-06, + "loss": 0.0774, + "step": 14865 + }, + { + "epoch": 6.22, + "learning_rate": 2.47215442022076e-06, + "loss": 0.101, + "step": 14866 + }, + { + "epoch": 6.22, + "learning_rate": 2.4710384967436983e-06, + "loss": 0.0881, + "step": 14867 + }, + { + "epoch": 6.22, + "learning_rate": 2.469922789679954e-06, + "loss": 0.0825, + "step": 14868 + }, + { + "epoch": 6.23, + "learning_rate": 2.4688072990615987e-06, + "loss": 0.0701, + "step": 14869 + }, + { + "epoch": 6.23, + "learning_rate": 2.4676920249206926e-06, + "loss": 0.1069, + "step": 14870 + }, + { + "epoch": 6.23, + "learning_rate": 2.4665769672892946e-06, + "loss": 0.0739, + "step": 14871 + }, + { + "epoch": 6.23, + "learning_rate": 2.465462126199457e-06, + "loss": 0.0832, + "step": 14872 + }, + { + "epoch": 6.23, + "learning_rate": 2.464347501683224e-06, + "loss": 0.0882, + "step": 14873 + }, + { + "epoch": 6.23, + "learning_rate": 2.4632330937726356e-06, + "loss": 0.0955, + "step": 14874 + }, + { + "epoch": 6.23, + "learning_rate": 2.4621189024997273e-06, + "loss": 0.1045, + "step": 14875 + }, + { + "epoch": 6.23, + "learning_rate": 2.4610049278965175e-06, + "loss": 0.104, + "step": 14876 + }, + { + "epoch": 6.23, + "learning_rate": 2.4598911699950325e-06, + "loss": 0.0882, + "step": 14877 + }, + { + "epoch": 6.23, + "learning_rate": 2.4587776288272846e-06, + "loss": 0.0646, + "step": 14878 + }, + { + "epoch": 6.23, + "learning_rate": 2.4576643044252824e-06, + "loss": 0.0777, + "step": 14879 + }, + { + "epoch": 6.23, + "learning_rate": 2.4565511968210285e-06, + "loss": 0.0748, + "step": 14880 + }, + { + "epoch": 6.23, + "learning_rate": 2.455438306046515e-06, + "loss": 0.1052, + "step": 14881 + }, + { + "epoch": 6.23, + "learning_rate": 2.4543256321337306e-06, + "loss": 0.0863, + "step": 14882 + }, + { + "epoch": 6.23, + "learning_rate": 2.453213175114664e-06, + "loss": 0.0721, + "step": 14883 + }, + { + "epoch": 6.23, + "learning_rate": 2.4521009350212854e-06, + "loss": 0.0985, + "step": 14884 + }, + { + "epoch": 6.23, + "learning_rate": 2.4509889118855666e-06, + "loss": 0.0605, + "step": 14885 + }, + { + "epoch": 6.23, + "learning_rate": 2.449877105739473e-06, + "loss": 0.0552, + "step": 14886 + }, + { + "epoch": 6.23, + "learning_rate": 2.4487655166149625e-06, + "loss": 0.0864, + "step": 14887 + }, + { + "epoch": 6.23, + "learning_rate": 2.447654144543986e-06, + "loss": 0.0859, + "step": 14888 + }, + { + "epoch": 6.23, + "learning_rate": 2.4465429895584914e-06, + "loss": 0.0513, + "step": 14889 + }, + { + "epoch": 6.23, + "learning_rate": 2.445432051690414e-06, + "loss": 0.0702, + "step": 14890 + }, + { + "epoch": 6.23, + "learning_rate": 2.4443213309716884e-06, + "loss": 0.0878, + "step": 14891 + }, + { + "epoch": 6.23, + "learning_rate": 2.4432108274342405e-06, + "loss": 0.0621, + "step": 14892 + }, + { + "epoch": 6.24, + "learning_rate": 2.4421005411099928e-06, + "loss": 0.0659, + "step": 14893 + }, + { + "epoch": 6.24, + "learning_rate": 2.4409904720308576e-06, + "loss": 0.0877, + "step": 14894 + }, + { + "epoch": 6.24, + "learning_rate": 2.4398806202287427e-06, + "loss": 0.1058, + "step": 14895 + }, + { + "epoch": 6.24, + "learning_rate": 2.4387709857355545e-06, + "loss": 0.0755, + "step": 14896 + }, + { + "epoch": 6.24, + "learning_rate": 2.4376615685831817e-06, + "loss": 0.1033, + "step": 14897 + }, + { + "epoch": 6.24, + "learning_rate": 2.4365523688035163e-06, + "loss": 0.0995, + "step": 14898 + }, + { + "epoch": 6.24, + "learning_rate": 2.435443386428442e-06, + "loss": 0.0744, + "step": 14899 + }, + { + "epoch": 6.24, + "learning_rate": 2.4343346214898343e-06, + "loss": 0.0632, + "step": 14900 + }, + { + "epoch": 6.24, + "learning_rate": 2.4332260740195645e-06, + "loss": 0.0854, + "step": 14901 + }, + { + "epoch": 6.24, + "learning_rate": 2.4321177440494993e-06, + "loss": 0.0875, + "step": 14902 + }, + { + "epoch": 6.24, + "learning_rate": 2.431009631611493e-06, + "loss": 0.0932, + "step": 14903 + }, + { + "epoch": 6.24, + "learning_rate": 2.4299017367373966e-06, + "loss": 0.0912, + "step": 14904 + }, + { + "epoch": 6.24, + "learning_rate": 2.4287940594590564e-06, + "loss": 0.0737, + "step": 14905 + }, + { + "epoch": 6.24, + "learning_rate": 2.4276865998083132e-06, + "loss": 0.0779, + "step": 14906 + }, + { + "epoch": 6.24, + "learning_rate": 2.426579357816998e-06, + "loss": 0.0774, + "step": 14907 + }, + { + "epoch": 6.24, + "learning_rate": 2.4254723335169394e-06, + "loss": 0.0942, + "step": 14908 + }, + { + "epoch": 6.24, + "learning_rate": 2.424365526939958e-06, + "loss": 0.0811, + "step": 14909 + }, + { + "epoch": 6.24, + "learning_rate": 2.4232589381178685e-06, + "loss": 0.0895, + "step": 14910 + }, + { + "epoch": 6.24, + "learning_rate": 2.422152567082474e-06, + "loss": 0.1295, + "step": 14911 + }, + { + "epoch": 6.24, + "learning_rate": 2.4210464138655807e-06, + "loss": 0.1035, + "step": 14912 + }, + { + "epoch": 6.24, + "learning_rate": 2.4199404784989833e-06, + "loss": 0.0901, + "step": 14913 + }, + { + "epoch": 6.24, + "learning_rate": 2.4188347610144693e-06, + "loss": 0.0943, + "step": 14914 + }, + { + "epoch": 6.24, + "learning_rate": 2.417729261443823e-06, + "loss": 0.07, + "step": 14915 + }, + { + "epoch": 6.24, + "learning_rate": 2.416623979818823e-06, + "loss": 0.0666, + "step": 14916 + }, + { + "epoch": 6.25, + "learning_rate": 2.4155189161712358e-06, + "loss": 0.0687, + "step": 14917 + }, + { + "epoch": 6.25, + "learning_rate": 2.4144140705328258e-06, + "loss": 0.0905, + "step": 14918 + }, + { + "epoch": 6.25, + "learning_rate": 2.4133094429353522e-06, + "loss": 0.0994, + "step": 14919 + }, + { + "epoch": 6.25, + "learning_rate": 2.412205033410566e-06, + "loss": 0.0748, + "step": 14920 + }, + { + "epoch": 6.25, + "learning_rate": 2.4111008419902138e-06, + "loss": 0.1008, + "step": 14921 + }, + { + "epoch": 6.25, + "learning_rate": 2.409996868706036e-06, + "loss": 0.1057, + "step": 14922 + }, + { + "epoch": 6.25, + "learning_rate": 2.4088931135897596e-06, + "loss": 0.0828, + "step": 14923 + }, + { + "epoch": 6.25, + "learning_rate": 2.407789576673115e-06, + "loss": 0.0655, + "step": 14924 + }, + { + "epoch": 6.25, + "learning_rate": 2.406686257987825e-06, + "loss": 0.0944, + "step": 14925 + }, + { + "epoch": 6.25, + "learning_rate": 2.4055831575655975e-06, + "loss": 0.0802, + "step": 14926 + }, + { + "epoch": 6.25, + "learning_rate": 2.4044802754381423e-06, + "loss": 0.0785, + "step": 14927 + }, + { + "epoch": 6.25, + "learning_rate": 2.4033776116371623e-06, + "loss": 0.0997, + "step": 14928 + }, + { + "epoch": 6.25, + "learning_rate": 2.402275166194351e-06, + "loss": 0.084, + "step": 14929 + }, + { + "epoch": 6.25, + "learning_rate": 2.401172939141401e-06, + "loss": 0.0743, + "step": 14930 + }, + { + "epoch": 6.25, + "learning_rate": 2.400070930509989e-06, + "loss": 0.0961, + "step": 14931 + }, + { + "epoch": 6.25, + "learning_rate": 2.398969140331794e-06, + "loss": 0.0833, + "step": 14932 + }, + { + "epoch": 6.25, + "learning_rate": 2.3978675686384854e-06, + "loss": 0.0987, + "step": 14933 + }, + { + "epoch": 6.25, + "learning_rate": 2.396766215461728e-06, + "loss": 0.0845, + "step": 14934 + }, + { + "epoch": 6.25, + "learning_rate": 2.3956650808331795e-06, + "loss": 0.0589, + "step": 14935 + }, + { + "epoch": 6.25, + "learning_rate": 2.394564164784489e-06, + "loss": 0.1022, + "step": 14936 + }, + { + "epoch": 6.25, + "learning_rate": 2.3934634673473066e-06, + "loss": 0.0674, + "step": 14937 + }, + { + "epoch": 6.25, + "learning_rate": 2.392362988553263e-06, + "loss": 0.0875, + "step": 14938 + }, + { + "epoch": 6.25, + "learning_rate": 2.391262728433995e-06, + "loss": 0.084, + "step": 14939 + }, + { + "epoch": 6.25, + "learning_rate": 2.3901626870211282e-06, + "loss": 0.0814, + "step": 14940 + }, + { + "epoch": 6.26, + "learning_rate": 2.389062864346282e-06, + "loss": 0.0893, + "step": 14941 + }, + { + "epoch": 6.26, + "learning_rate": 2.3879632604410695e-06, + "loss": 0.101, + "step": 14942 + }, + { + "epoch": 6.26, + "learning_rate": 2.386863875337102e-06, + "loss": 0.0771, + "step": 14943 + }, + { + "epoch": 6.26, + "learning_rate": 2.385764709065973e-06, + "loss": 0.09, + "step": 14944 + }, + { + "epoch": 6.26, + "learning_rate": 2.3846657616592805e-06, + "loss": 0.0615, + "step": 14945 + }, + { + "epoch": 6.26, + "learning_rate": 2.3835670331486137e-06, + "loss": 0.094, + "step": 14946 + }, + { + "epoch": 6.26, + "learning_rate": 2.3824685235655532e-06, + "loss": 0.083, + "step": 14947 + }, + { + "epoch": 6.26, + "learning_rate": 2.381370232941679e-06, + "loss": 0.0625, + "step": 14948 + }, + { + "epoch": 6.26, + "learning_rate": 2.3802721613085534e-06, + "loss": 0.1044, + "step": 14949 + }, + { + "epoch": 6.26, + "learning_rate": 2.3791743086977425e-06, + "loss": 0.0795, + "step": 14950 + }, + { + "epoch": 6.26, + "learning_rate": 2.378076675140807e-06, + "loss": 0.0791, + "step": 14951 + }, + { + "epoch": 6.26, + "learning_rate": 2.376979260669291e-06, + "loss": 0.0932, + "step": 14952 + }, + { + "epoch": 6.26, + "learning_rate": 2.3758820653147405e-06, + "loss": 0.0702, + "step": 14953 + }, + { + "epoch": 6.26, + "learning_rate": 2.3747850891086956e-06, + "loss": 0.0805, + "step": 14954 + }, + { + "epoch": 6.26, + "learning_rate": 2.3736883320826864e-06, + "loss": 0.0756, + "step": 14955 + }, + { + "epoch": 6.26, + "learning_rate": 2.3725917942682397e-06, + "loss": 0.099, + "step": 14956 + }, + { + "epoch": 6.26, + "learning_rate": 2.3714954756968746e-06, + "loss": 0.0872, + "step": 14957 + }, + { + "epoch": 6.26, + "learning_rate": 2.370399376400101e-06, + "loss": 0.0862, + "step": 14958 + }, + { + "epoch": 6.26, + "learning_rate": 2.3693034964094264e-06, + "loss": 0.0518, + "step": 14959 + }, + { + "epoch": 6.26, + "learning_rate": 2.368207835756352e-06, + "loss": 0.0863, + "step": 14960 + }, + { + "epoch": 6.26, + "learning_rate": 2.367112394472371e-06, + "loss": 0.0785, + "step": 14961 + }, + { + "epoch": 6.26, + "learning_rate": 2.3660171725889703e-06, + "loss": 0.1134, + "step": 14962 + }, + { + "epoch": 6.26, + "learning_rate": 2.3649221701376346e-06, + "loss": 0.0889, + "step": 14963 + }, + { + "epoch": 6.27, + "learning_rate": 2.363827387149833e-06, + "loss": 0.0916, + "step": 14964 + }, + { + "epoch": 6.27, + "learning_rate": 2.3627328236570367e-06, + "loss": 0.087, + "step": 14965 + }, + { + "epoch": 6.27, + "learning_rate": 2.3616384796907087e-06, + "loss": 0.0704, + "step": 14966 + }, + { + "epoch": 6.27, + "learning_rate": 2.360544355282305e-06, + "loss": 0.0926, + "step": 14967 + }, + { + "epoch": 6.27, + "learning_rate": 2.359450450463274e-06, + "loss": 0.0874, + "step": 14968 + }, + { + "epoch": 6.27, + "learning_rate": 2.35835676526506e-06, + "loss": 0.1054, + "step": 14969 + }, + { + "epoch": 6.27, + "learning_rate": 2.3572632997191016e-06, + "loss": 0.0871, + "step": 14970 + }, + { + "epoch": 6.27, + "learning_rate": 2.356170053856829e-06, + "loss": 0.0815, + "step": 14971 + }, + { + "epoch": 6.27, + "learning_rate": 2.3550770277096615e-06, + "loss": 0.0962, + "step": 14972 + }, + { + "epoch": 6.27, + "learning_rate": 2.35398422130902e-06, + "loss": 0.0683, + "step": 14973 + }, + { + "epoch": 6.27, + "learning_rate": 2.3528916346863185e-06, + "loss": 0.0937, + "step": 14974 + }, + { + "epoch": 6.27, + "learning_rate": 2.3517992678729607e-06, + "loss": 0.0538, + "step": 14975 + }, + { + "epoch": 6.27, + "learning_rate": 2.3507071209003463e-06, + "loss": 0.0896, + "step": 14976 + }, + { + "epoch": 6.27, + "learning_rate": 2.349615193799868e-06, + "loss": 0.0762, + "step": 14977 + }, + { + "epoch": 6.27, + "learning_rate": 2.3485234866029137e-06, + "loss": 0.1219, + "step": 14978 + }, + { + "epoch": 6.27, + "learning_rate": 2.34743199934086e-06, + "loss": 0.1079, + "step": 14979 + }, + { + "epoch": 6.27, + "learning_rate": 2.3463407320450827e-06, + "loss": 0.0814, + "step": 14980 + }, + { + "epoch": 6.27, + "learning_rate": 2.34524968474695e-06, + "loss": 0.1027, + "step": 14981 + }, + { + "epoch": 6.27, + "learning_rate": 2.3441588574778207e-06, + "loss": 0.0672, + "step": 14982 + }, + { + "epoch": 6.27, + "learning_rate": 2.3430682502690527e-06, + "loss": 0.0835, + "step": 14983 + }, + { + "epoch": 6.27, + "learning_rate": 2.3419778631519963e-06, + "loss": 0.079, + "step": 14984 + }, + { + "epoch": 6.27, + "learning_rate": 2.340887696157986e-06, + "loss": 0.0962, + "step": 14985 + }, + { + "epoch": 6.27, + "learning_rate": 2.3397977493183643e-06, + "loss": 0.1165, + "step": 14986 + }, + { + "epoch": 6.27, + "learning_rate": 2.3387080226644578e-06, + "loss": 0.0685, + "step": 14987 + }, + { + "epoch": 6.28, + "learning_rate": 2.337618516227591e-06, + "loss": 0.058, + "step": 14988 + }, + { + "epoch": 6.28, + "learning_rate": 2.3365292300390795e-06, + "loss": 0.0789, + "step": 14989 + }, + { + "epoch": 6.28, + "learning_rate": 2.3354401641302395e-06, + "loss": 0.0779, + "step": 14990 + }, + { + "epoch": 6.28, + "learning_rate": 2.3343513185323664e-06, + "loss": 0.0699, + "step": 14991 + }, + { + "epoch": 6.28, + "learning_rate": 2.333262693276762e-06, + "loss": 0.0783, + "step": 14992 + }, + { + "epoch": 6.28, + "learning_rate": 2.332174288394722e-06, + "loss": 0.0981, + "step": 14993 + }, + { + "epoch": 6.28, + "learning_rate": 2.331086103917525e-06, + "loss": 0.0759, + "step": 14994 + }, + { + "epoch": 6.28, + "learning_rate": 2.3299981398764515e-06, + "loss": 0.0885, + "step": 14995 + }, + { + "epoch": 6.28, + "learning_rate": 2.328910396302776e-06, + "loss": 0.0812, + "step": 14996 + }, + { + "epoch": 6.28, + "learning_rate": 2.327822873227764e-06, + "loss": 0.0845, + "step": 14997 + }, + { + "epoch": 6.28, + "learning_rate": 2.326735570682678e-06, + "loss": 0.0802, + "step": 14998 + }, + { + "epoch": 6.28, + "learning_rate": 2.325648488698766e-06, + "loss": 0.0752, + "step": 14999 + }, + { + "epoch": 6.28, + "learning_rate": 2.324561627307278e-06, + "loss": 0.0878, + "step": 15000 + }, + { + "epoch": 6.28, + "learning_rate": 2.3234749865394545e-06, + "loss": 0.0812, + "step": 15001 + }, + { + "epoch": 6.28, + "learning_rate": 2.322388566426532e-06, + "loss": 0.055, + "step": 15002 + }, + { + "epoch": 6.28, + "learning_rate": 2.321302366999736e-06, + "loss": 0.0815, + "step": 15003 + }, + { + "epoch": 6.28, + "learning_rate": 2.320216388290293e-06, + "loss": 0.0931, + "step": 15004 + }, + { + "epoch": 6.28, + "learning_rate": 2.3191306303294114e-06, + "loss": 0.0878, + "step": 15005 + }, + { + "epoch": 6.28, + "learning_rate": 2.3180450931483045e-06, + "loss": 0.0696, + "step": 15006 + }, + { + "epoch": 6.28, + "learning_rate": 2.3169597767781736e-06, + "loss": 0.0859, + "step": 15007 + }, + { + "epoch": 6.28, + "learning_rate": 2.3158746812502174e-06, + "loss": 0.0825, + "step": 15008 + }, + { + "epoch": 6.28, + "learning_rate": 2.314789806595624e-06, + "loss": 0.1027, + "step": 15009 + }, + { + "epoch": 6.28, + "learning_rate": 2.3137051528455778e-06, + "loss": 0.1111, + "step": 15010 + }, + { + "epoch": 6.28, + "learning_rate": 2.31262072003126e-06, + "loss": 0.0816, + "step": 15011 + }, + { + "epoch": 6.29, + "learning_rate": 2.3115365081838346e-06, + "loss": 0.0815, + "step": 15012 + }, + { + "epoch": 6.29, + "learning_rate": 2.3104525173344693e-06, + "loss": 0.0544, + "step": 15013 + }, + { + "epoch": 6.29, + "learning_rate": 2.3093687475143236e-06, + "loss": 0.0848, + "step": 15014 + }, + { + "epoch": 6.29, + "learning_rate": 2.3082851987545474e-06, + "loss": 0.1051, + "step": 15015 + }, + { + "epoch": 6.29, + "learning_rate": 2.307201871086292e-06, + "loss": 0.0788, + "step": 15016 + }, + { + "epoch": 6.29, + "learning_rate": 2.306118764540688e-06, + "loss": 0.1185, + "step": 15017 + }, + { + "epoch": 6.29, + "learning_rate": 2.3050358791488734e-06, + "loss": 0.1118, + "step": 15018 + }, + { + "epoch": 6.29, + "learning_rate": 2.3039532149419775e-06, + "loss": 0.1037, + "step": 15019 + }, + { + "epoch": 6.29, + "learning_rate": 2.3028707719511134e-06, + "loss": 0.1002, + "step": 15020 + }, + { + "epoch": 6.29, + "learning_rate": 2.301788550207399e-06, + "loss": 0.0603, + "step": 15021 + }, + { + "epoch": 6.29, + "learning_rate": 2.3007065497419413e-06, + "loss": 0.1137, + "step": 15022 + }, + { + "epoch": 6.29, + "learning_rate": 2.299624770585842e-06, + "loss": 0.1029, + "step": 15023 + }, + { + "epoch": 6.29, + "learning_rate": 2.2985432127701945e-06, + "loss": 0.1103, + "step": 15024 + }, + { + "epoch": 6.29, + "learning_rate": 2.2974618763260927e-06, + "loss": 0.0929, + "step": 15025 + }, + { + "epoch": 6.29, + "learning_rate": 2.29638076128461e-06, + "loss": 0.0851, + "step": 15026 + }, + { + "epoch": 6.29, + "learning_rate": 2.2952998676768267e-06, + "loss": 0.0539, + "step": 15027 + }, + { + "epoch": 6.29, + "learning_rate": 2.294219195533811e-06, + "loss": 0.1102, + "step": 15028 + }, + { + "epoch": 6.29, + "learning_rate": 2.293138744886627e-06, + "loss": 0.0641, + "step": 15029 + }, + { + "epoch": 6.29, + "learning_rate": 2.292058515766331e-06, + "loss": 0.0838, + "step": 15030 + }, + { + "epoch": 6.29, + "learning_rate": 2.290978508203976e-06, + "loss": 0.0982, + "step": 15031 + }, + { + "epoch": 6.29, + "learning_rate": 2.2898987222306e-06, + "loss": 0.0812, + "step": 15032 + }, + { + "epoch": 6.29, + "learning_rate": 2.2888191578772436e-06, + "loss": 0.0796, + "step": 15033 + }, + { + "epoch": 6.29, + "learning_rate": 2.2877398151749375e-06, + "loss": 0.0761, + "step": 15034 + }, + { + "epoch": 6.29, + "learning_rate": 2.2866606941547063e-06, + "loss": 0.0802, + "step": 15035 + }, + { + "epoch": 6.3, + "learning_rate": 2.285581794847569e-06, + "loss": 0.0624, + "step": 15036 + }, + { + "epoch": 6.3, + "learning_rate": 2.284503117284538e-06, + "loss": 0.0575, + "step": 15037 + }, + { + "epoch": 6.3, + "learning_rate": 2.2834246614966192e-06, + "loss": 0.0717, + "step": 15038 + }, + { + "epoch": 6.3, + "learning_rate": 2.282346427514813e-06, + "loss": 0.0925, + "step": 15039 + }, + { + "epoch": 6.3, + "learning_rate": 2.281268415370107e-06, + "loss": 0.1024, + "step": 15040 + }, + { + "epoch": 6.3, + "learning_rate": 2.2801906250934904e-06, + "loss": 0.0787, + "step": 15041 + }, + { + "epoch": 6.3, + "learning_rate": 2.2791130567159434e-06, + "loss": 0.0818, + "step": 15042 + }, + { + "epoch": 6.3, + "learning_rate": 2.2780357102684415e-06, + "loss": 0.0743, + "step": 15043 + }, + { + "epoch": 6.3, + "learning_rate": 2.27695858578195e-06, + "loss": 0.0911, + "step": 15044 + }, + { + "epoch": 6.3, + "learning_rate": 2.2758816832874333e-06, + "loss": 0.0787, + "step": 15045 + }, + { + "epoch": 6.3, + "learning_rate": 2.274805002815841e-06, + "loss": 0.0562, + "step": 15046 + }, + { + "epoch": 6.3, + "learning_rate": 2.2737285443981226e-06, + "loss": 0.0959, + "step": 15047 + }, + { + "epoch": 6.3, + "learning_rate": 2.272652308065222e-06, + "loss": 0.0979, + "step": 15048 + }, + { + "epoch": 6.3, + "learning_rate": 2.2715762938480734e-06, + "loss": 0.0647, + "step": 15049 + }, + { + "epoch": 6.3, + "learning_rate": 2.2705005017776073e-06, + "loss": 0.0847, + "step": 15050 + }, + { + "epoch": 6.3, + "learning_rate": 2.269424931884744e-06, + "loss": 0.0836, + "step": 15051 + }, + { + "epoch": 6.3, + "learning_rate": 2.268349584200403e-06, + "loss": 0.0797, + "step": 15052 + }, + { + "epoch": 6.3, + "learning_rate": 2.2672744587554917e-06, + "loss": 0.0747, + "step": 15053 + }, + { + "epoch": 6.3, + "learning_rate": 2.2661995555809123e-06, + "loss": 0.0945, + "step": 15054 + }, + { + "epoch": 6.3, + "learning_rate": 2.265124874707565e-06, + "loss": 0.0886, + "step": 15055 + }, + { + "epoch": 6.3, + "learning_rate": 2.264050416166339e-06, + "loss": 0.1056, + "step": 15056 + }, + { + "epoch": 6.3, + "learning_rate": 2.262976179988119e-06, + "loss": 0.0627, + "step": 15057 + }, + { + "epoch": 6.3, + "learning_rate": 2.2619021662037855e-06, + "loss": 0.0828, + "step": 15058 + }, + { + "epoch": 6.3, + "learning_rate": 2.260828374844205e-06, + "loss": 0.0694, + "step": 15059 + }, + { + "epoch": 6.31, + "learning_rate": 2.2597548059402453e-06, + "loss": 0.0854, + "step": 15060 + }, + { + "epoch": 6.31, + "learning_rate": 2.258681459522768e-06, + "loss": 0.0818, + "step": 15061 + }, + { + "epoch": 6.31, + "learning_rate": 2.2576083356226197e-06, + "loss": 0.0873, + "step": 15062 + }, + { + "epoch": 6.31, + "learning_rate": 2.256535434270649e-06, + "loss": 0.0569, + "step": 15063 + }, + { + "epoch": 6.31, + "learning_rate": 2.255462755497696e-06, + "loss": 0.06, + "step": 15064 + }, + { + "epoch": 6.31, + "learning_rate": 2.2543902993345946e-06, + "loss": 0.0988, + "step": 15065 + }, + { + "epoch": 6.31, + "learning_rate": 2.2533180658121723e-06, + "loss": 0.0793, + "step": 15066 + }, + { + "epoch": 6.31, + "learning_rate": 2.252246054961246e-06, + "loss": 0.0841, + "step": 15067 + }, + { + "epoch": 6.31, + "learning_rate": 2.25117426681263e-06, + "loss": 0.0698, + "step": 15068 + }, + { + "epoch": 6.31, + "learning_rate": 2.2501027013971354e-06, + "loss": 0.1104, + "step": 15069 + }, + { + "epoch": 6.31, + "learning_rate": 2.2490313587455605e-06, + "loss": 0.0768, + "step": 15070 + }, + { + "epoch": 6.31, + "learning_rate": 2.2479602388887013e-06, + "loss": 0.0616, + "step": 15071 + }, + { + "epoch": 6.31, + "learning_rate": 2.2468893418573488e-06, + "loss": 0.0998, + "step": 15072 + }, + { + "epoch": 6.31, + "learning_rate": 2.24581866768228e-06, + "loss": 0.0846, + "step": 15073 + }, + { + "epoch": 6.31, + "learning_rate": 2.2447482163942725e-06, + "loss": 0.0625, + "step": 15074 + }, + { + "epoch": 6.31, + "learning_rate": 2.2436779880240943e-06, + "loss": 0.0746, + "step": 15075 + }, + { + "epoch": 6.31, + "learning_rate": 2.2426079826025118e-06, + "loss": 0.0756, + "step": 15076 + }, + { + "epoch": 6.31, + "learning_rate": 2.241538200160277e-06, + "loss": 0.0823, + "step": 15077 + }, + { + "epoch": 6.31, + "learning_rate": 2.2404686407281427e-06, + "loss": 0.0996, + "step": 15078 + }, + { + "epoch": 6.31, + "learning_rate": 2.2393993043368545e-06, + "loss": 0.0844, + "step": 15079 + }, + { + "epoch": 6.31, + "learning_rate": 2.238330191017144e-06, + "loss": 0.1074, + "step": 15080 + }, + { + "epoch": 6.31, + "learning_rate": 2.2372613007997435e-06, + "loss": 0.0777, + "step": 15081 + }, + { + "epoch": 6.31, + "learning_rate": 2.2361926337153796e-06, + "loss": 0.0737, + "step": 15082 + }, + { + "epoch": 6.31, + "learning_rate": 2.2351241897947684e-06, + "loss": 0.0865, + "step": 15083 + }, + { + "epoch": 6.32, + "learning_rate": 2.2340559690686247e-06, + "loss": 0.094, + "step": 15084 + }, + { + "epoch": 6.32, + "learning_rate": 2.232987971567648e-06, + "loss": 0.0923, + "step": 15085 + }, + { + "epoch": 6.32, + "learning_rate": 2.2319201973225434e-06, + "loss": 0.0568, + "step": 15086 + }, + { + "epoch": 6.32, + "learning_rate": 2.2308526463639955e-06, + "loss": 0.1045, + "step": 15087 + }, + { + "epoch": 6.32, + "learning_rate": 2.2297853187226946e-06, + "loss": 0.0798, + "step": 15088 + }, + { + "epoch": 6.32, + "learning_rate": 2.2287182144293196e-06, + "loss": 0.0816, + "step": 15089 + }, + { + "epoch": 6.32, + "learning_rate": 2.227651333514543e-06, + "loss": 0.0923, + "step": 15090 + }, + { + "epoch": 6.32, + "learning_rate": 2.2265846760090326e-06, + "loss": 0.0729, + "step": 15091 + }, + { + "epoch": 6.32, + "learning_rate": 2.225518241943446e-06, + "loss": 0.0955, + "step": 15092 + }, + { + "epoch": 6.32, + "learning_rate": 2.224452031348443e-06, + "loss": 0.0795, + "step": 15093 + }, + { + "epoch": 6.32, + "learning_rate": 2.223386044254663e-06, + "loss": 0.0862, + "step": 15094 + }, + { + "epoch": 6.32, + "learning_rate": 2.222320280692751e-06, + "loss": 0.0762, + "step": 15095 + }, + { + "epoch": 6.32, + "learning_rate": 2.2212547406933405e-06, + "loss": 0.0771, + "step": 15096 + }, + { + "epoch": 6.32, + "learning_rate": 2.220189424287059e-06, + "loss": 0.0832, + "step": 15097 + }, + { + "epoch": 6.32, + "learning_rate": 2.2191243315045296e-06, + "loss": 0.0759, + "step": 15098 + }, + { + "epoch": 6.32, + "learning_rate": 2.2180594623763686e-06, + "loss": 0.066, + "step": 15099 + }, + { + "epoch": 6.32, + "learning_rate": 2.2169948169331812e-06, + "loss": 0.0649, + "step": 15100 + }, + { + "epoch": 6.32, + "learning_rate": 2.21593039520557e-06, + "loss": 0.0836, + "step": 15101 + }, + { + "epoch": 6.32, + "learning_rate": 2.2148661972241324e-06, + "loss": 0.0737, + "step": 15102 + }, + { + "epoch": 6.32, + "learning_rate": 2.213802223019459e-06, + "loss": 0.0903, + "step": 15103 + }, + { + "epoch": 6.32, + "learning_rate": 2.2127384726221303e-06, + "loss": 0.0921, + "step": 15104 + }, + { + "epoch": 6.32, + "learning_rate": 2.211674946062726e-06, + "loss": 0.0945, + "step": 15105 + }, + { + "epoch": 6.32, + "learning_rate": 2.2106116433718115e-06, + "loss": 0.0758, + "step": 15106 + }, + { + "epoch": 6.32, + "learning_rate": 2.2095485645799564e-06, + "loss": 0.0526, + "step": 15107 + }, + { + "epoch": 6.33, + "learning_rate": 2.2084857097177125e-06, + "loss": 0.1023, + "step": 15108 + }, + { + "epoch": 6.33, + "learning_rate": 2.2074230788156325e-06, + "loss": 0.0932, + "step": 15109 + }, + { + "epoch": 6.33, + "learning_rate": 2.2063606719042605e-06, + "loss": 0.0918, + "step": 15110 + }, + { + "epoch": 6.33, + "learning_rate": 2.205298489014135e-06, + "loss": 0.0743, + "step": 15111 + }, + { + "epoch": 6.33, + "learning_rate": 2.2042365301757883e-06, + "loss": 0.0802, + "step": 15112 + }, + { + "epoch": 6.33, + "learning_rate": 2.203174795419747e-06, + "loss": 0.105, + "step": 15113 + }, + { + "epoch": 6.33, + "learning_rate": 2.2021132847765246e-06, + "loss": 0.1102, + "step": 15114 + }, + { + "epoch": 6.33, + "learning_rate": 2.2010519982766364e-06, + "loss": 0.0819, + "step": 15115 + }, + { + "epoch": 6.33, + "learning_rate": 2.1999909359505878e-06, + "loss": 0.0604, + "step": 15116 + }, + { + "epoch": 6.33, + "learning_rate": 2.1989300978288773e-06, + "loss": 0.0653, + "step": 15117 + }, + { + "epoch": 6.33, + "learning_rate": 2.197869483942e-06, + "loss": 0.0892, + "step": 15118 + }, + { + "epoch": 6.33, + "learning_rate": 2.1968090943204402e-06, + "loss": 0.0818, + "step": 15119 + }, + { + "epoch": 6.33, + "learning_rate": 2.195748928994681e-06, + "loss": 0.1011, + "step": 15120 + }, + { + "epoch": 6.33, + "learning_rate": 2.1946889879951903e-06, + "loss": 0.1059, + "step": 15121 + }, + { + "epoch": 6.33, + "learning_rate": 2.19362927135244e-06, + "loss": 0.0953, + "step": 15122 + }, + { + "epoch": 6.33, + "learning_rate": 2.192569779096888e-06, + "loss": 0.0665, + "step": 15123 + }, + { + "epoch": 6.33, + "learning_rate": 2.1915105112589894e-06, + "loss": 0.0824, + "step": 15124 + }, + { + "epoch": 6.33, + "learning_rate": 2.1904514678691924e-06, + "loss": 0.1006, + "step": 15125 + }, + { + "epoch": 6.33, + "learning_rate": 2.18939264895794e-06, + "loss": 0.0923, + "step": 15126 + }, + { + "epoch": 6.33, + "learning_rate": 2.188334054555663e-06, + "loss": 0.1008, + "step": 15127 + }, + { + "epoch": 6.33, + "learning_rate": 2.1872756846927913e-06, + "loss": 0.1174, + "step": 15128 + }, + { + "epoch": 6.33, + "learning_rate": 2.1862175393997497e-06, + "loss": 0.1017, + "step": 15129 + }, + { + "epoch": 6.33, + "learning_rate": 2.1851596187069478e-06, + "loss": 0.0738, + "step": 15130 + }, + { + "epoch": 6.33, + "learning_rate": 2.184101922644799e-06, + "loss": 0.086, + "step": 15131 + }, + { + "epoch": 6.34, + "learning_rate": 2.183044451243704e-06, + "loss": 0.0789, + "step": 15132 + }, + { + "epoch": 6.34, + "learning_rate": 2.181987204534061e-06, + "loss": 0.1091, + "step": 15133 + }, + { + "epoch": 6.34, + "learning_rate": 2.1809301825462593e-06, + "loss": 0.0802, + "step": 15134 + }, + { + "epoch": 6.34, + "learning_rate": 2.179873385310679e-06, + "loss": 0.1005, + "step": 15135 + }, + { + "epoch": 6.34, + "learning_rate": 2.178816812857698e-06, + "loss": 0.0425, + "step": 15136 + }, + { + "epoch": 6.34, + "learning_rate": 2.1777604652176885e-06, + "loss": 0.0622, + "step": 15137 + }, + { + "epoch": 6.34, + "learning_rate": 2.176704342421012e-06, + "loss": 0.0391, + "step": 15138 + }, + { + "epoch": 6.34, + "learning_rate": 2.1756484444980274e-06, + "loss": 0.0774, + "step": 15139 + }, + { + "epoch": 6.34, + "learning_rate": 2.1745927714790882e-06, + "loss": 0.0738, + "step": 15140 + }, + { + "epoch": 6.34, + "learning_rate": 2.1735373233945323e-06, + "loss": 0.079, + "step": 15141 + }, + { + "epoch": 6.34, + "learning_rate": 2.172482100274702e-06, + "loss": 0.0918, + "step": 15142 + }, + { + "epoch": 6.34, + "learning_rate": 2.171427102149927e-06, + "loss": 0.0805, + "step": 15143 + }, + { + "epoch": 6.34, + "learning_rate": 2.170372329050533e-06, + "loss": 0.057, + "step": 15144 + }, + { + "epoch": 6.34, + "learning_rate": 2.169317781006838e-06, + "loss": 0.0869, + "step": 15145 + }, + { + "epoch": 6.34, + "learning_rate": 2.168263458049158e-06, + "loss": 0.0805, + "step": 15146 + }, + { + "epoch": 6.34, + "learning_rate": 2.1672093602077914e-06, + "loss": 0.0829, + "step": 15147 + }, + { + "epoch": 6.34, + "learning_rate": 2.166155487513042e-06, + "loss": 0.0954, + "step": 15148 + }, + { + "epoch": 6.34, + "learning_rate": 2.1651018399952017e-06, + "loss": 0.0794, + "step": 15149 + }, + { + "epoch": 6.34, + "learning_rate": 2.1640484176845557e-06, + "loss": 0.1011, + "step": 15150 + }, + { + "epoch": 6.34, + "learning_rate": 2.1629952206113845e-06, + "loss": 0.0983, + "step": 15151 + }, + { + "epoch": 6.34, + "learning_rate": 2.161942248805965e-06, + "loss": 0.099, + "step": 15152 + }, + { + "epoch": 6.34, + "learning_rate": 2.1608895022985564e-06, + "loss": 0.0934, + "step": 15153 + }, + { + "epoch": 6.34, + "learning_rate": 2.159836981119425e-06, + "loss": 0.1135, + "step": 15154 + }, + { + "epoch": 6.34, + "learning_rate": 2.1587846852988203e-06, + "loss": 0.0888, + "step": 15155 + }, + { + "epoch": 6.35, + "learning_rate": 2.157732614866992e-06, + "loss": 0.0539, + "step": 15156 + }, + { + "epoch": 6.35, + "learning_rate": 2.1566807698541803e-06, + "loss": 0.0841, + "step": 15157 + }, + { + "epoch": 6.35, + "learning_rate": 2.15562915029062e-06, + "loss": 0.0645, + "step": 15158 + }, + { + "epoch": 6.35, + "learning_rate": 2.1545777562065383e-06, + "loss": 0.0846, + "step": 15159 + }, + { + "epoch": 6.35, + "learning_rate": 2.1535265876321574e-06, + "loss": 0.0775, + "step": 15160 + }, + { + "epoch": 6.35, + "learning_rate": 2.1524756445976947e-06, + "loss": 0.0767, + "step": 15161 + }, + { + "epoch": 6.35, + "learning_rate": 2.1514249271333532e-06, + "loss": 0.0852, + "step": 15162 + }, + { + "epoch": 6.35, + "learning_rate": 2.1503744352693377e-06, + "loss": 0.0796, + "step": 15163 + }, + { + "epoch": 6.35, + "learning_rate": 2.1493241690358437e-06, + "loss": 0.0639, + "step": 15164 + }, + { + "epoch": 6.35, + "learning_rate": 2.148274128463058e-06, + "loss": 0.0731, + "step": 15165 + }, + { + "epoch": 6.35, + "learning_rate": 2.1472243135811666e-06, + "loss": 0.0772, + "step": 15166 + }, + { + "epoch": 6.35, + "learning_rate": 2.146174724420347e-06, + "loss": 0.0946, + "step": 15167 + }, + { + "epoch": 6.35, + "learning_rate": 2.1451253610107616e-06, + "loss": 0.0783, + "step": 15168 + }, + { + "epoch": 6.35, + "learning_rate": 2.1440762233825774e-06, + "loss": 0.0823, + "step": 15169 + }, + { + "epoch": 6.35, + "learning_rate": 2.1430273115659516e-06, + "loss": 0.0751, + "step": 15170 + }, + { + "epoch": 6.35, + "learning_rate": 2.1419786255910336e-06, + "loss": 0.0774, + "step": 15171 + }, + { + "epoch": 6.35, + "learning_rate": 2.1409301654879665e-06, + "loss": 0.1268, + "step": 15172 + }, + { + "epoch": 6.35, + "learning_rate": 2.139881931286891e-06, + "loss": 0.1119, + "step": 15173 + }, + { + "epoch": 6.35, + "learning_rate": 2.138833923017931e-06, + "loss": 0.092, + "step": 15174 + }, + { + "epoch": 6.35, + "learning_rate": 2.1377861407112178e-06, + "loss": 0.09, + "step": 15175 + }, + { + "epoch": 6.35, + "learning_rate": 2.136738584396861e-06, + "loss": 0.0835, + "step": 15176 + }, + { + "epoch": 6.35, + "learning_rate": 2.135691254104977e-06, + "loss": 0.1081, + "step": 15177 + }, + { + "epoch": 6.35, + "learning_rate": 2.1346441498656688e-06, + "loss": 0.0438, + "step": 15178 + }, + { + "epoch": 6.36, + "learning_rate": 2.1335972717090358e-06, + "loss": 0.068, + "step": 15179 + }, + { + "epoch": 6.36, + "learning_rate": 2.132550619665168e-06, + "loss": 0.0942, + "step": 15180 + }, + { + "epoch": 6.36, + "learning_rate": 2.1315041937641534e-06, + "loss": 0.0728, + "step": 15181 + }, + { + "epoch": 6.36, + "learning_rate": 2.130457994036066e-06, + "loss": 0.1139, + "step": 15182 + }, + { + "epoch": 6.36, + "learning_rate": 2.12941202051098e-06, + "loss": 0.0853, + "step": 15183 + }, + { + "epoch": 6.36, + "learning_rate": 2.1283662732189624e-06, + "loss": 0.092, + "step": 15184 + }, + { + "epoch": 6.36, + "learning_rate": 2.12732075219007e-06, + "loss": 0.0503, + "step": 15185 + }, + { + "epoch": 6.36, + "learning_rate": 2.1262754574543566e-06, + "loss": 0.0582, + "step": 15186 + }, + { + "epoch": 6.36, + "learning_rate": 2.1252303890418714e-06, + "loss": 0.0838, + "step": 15187 + }, + { + "epoch": 6.36, + "learning_rate": 2.124185546982648e-06, + "loss": 0.0767, + "step": 15188 + }, + { + "epoch": 6.36, + "learning_rate": 2.123140931306722e-06, + "loss": 0.08, + "step": 15189 + }, + { + "epoch": 6.36, + "learning_rate": 2.1220965420441198e-06, + "loss": 0.0973, + "step": 15190 + }, + { + "epoch": 6.36, + "learning_rate": 2.1210523792248615e-06, + "loss": 0.0849, + "step": 15191 + }, + { + "epoch": 6.36, + "learning_rate": 2.1200084428789614e-06, + "loss": 0.0809, + "step": 15192 + }, + { + "epoch": 6.36, + "learning_rate": 2.118964733036426e-06, + "loss": 0.1049, + "step": 15193 + }, + { + "epoch": 6.36, + "learning_rate": 2.1179212497272582e-06, + "loss": 0.076, + "step": 15194 + }, + { + "epoch": 6.36, + "learning_rate": 2.1168779929814476e-06, + "loss": 0.0701, + "step": 15195 + }, + { + "epoch": 6.36, + "learning_rate": 2.1158349628289832e-06, + "loss": 0.0752, + "step": 15196 + }, + { + "epoch": 6.36, + "learning_rate": 2.1147921592998498e-06, + "loss": 0.0687, + "step": 15197 + }, + { + "epoch": 6.36, + "learning_rate": 2.1137495824240152e-06, + "loss": 0.0641, + "step": 15198 + }, + { + "epoch": 6.36, + "learning_rate": 2.1127072322314514e-06, + "loss": 0.0852, + "step": 15199 + }, + { + "epoch": 6.36, + "learning_rate": 2.1116651087521188e-06, + "loss": 0.0573, + "step": 15200 + }, + { + "epoch": 6.36, + "learning_rate": 2.110623212015972e-06, + "loss": 0.0811, + "step": 15201 + }, + { + "epoch": 6.36, + "learning_rate": 2.109581542052964e-06, + "loss": 0.0704, + "step": 15202 + }, + { + "epoch": 6.37, + "learning_rate": 2.108540098893029e-06, + "loss": 0.074, + "step": 15203 + }, + { + "epoch": 6.37, + "learning_rate": 2.107498882566107e-06, + "loss": 0.0836, + "step": 15204 + }, + { + "epoch": 6.37, + "learning_rate": 2.106457893102125e-06, + "loss": 0.0861, + "step": 15205 + }, + { + "epoch": 6.37, + "learning_rate": 2.1054171305310065e-06, + "loss": 0.0707, + "step": 15206 + }, + { + "epoch": 6.37, + "learning_rate": 2.1043765948826678e-06, + "loss": 0.0875, + "step": 15207 + }, + { + "epoch": 6.37, + "learning_rate": 2.1033362861870198e-06, + "loss": 0.083, + "step": 15208 + }, + { + "epoch": 6.37, + "learning_rate": 2.1022962044739602e-06, + "loss": 0.0617, + "step": 15209 + }, + { + "epoch": 6.37, + "learning_rate": 2.1012563497733873e-06, + "loss": 0.0751, + "step": 15210 + }, + { + "epoch": 6.37, + "learning_rate": 2.1002167221151915e-06, + "loss": 0.0753, + "step": 15211 + }, + { + "epoch": 6.37, + "learning_rate": 2.0991773215292566e-06, + "loss": 0.0736, + "step": 15212 + }, + { + "epoch": 6.37, + "learning_rate": 2.0981381480454575e-06, + "loss": 0.0571, + "step": 15213 + }, + { + "epoch": 6.37, + "learning_rate": 2.0970992016936685e-06, + "loss": 0.057, + "step": 15214 + }, + { + "epoch": 6.37, + "learning_rate": 2.096060482503748e-06, + "loss": 0.0662, + "step": 15215 + }, + { + "epoch": 6.37, + "learning_rate": 2.0950219905055548e-06, + "loss": 0.1205, + "step": 15216 + }, + { + "epoch": 6.37, + "learning_rate": 2.0939837257289395e-06, + "loss": 0.0759, + "step": 15217 + }, + { + "epoch": 6.37, + "learning_rate": 2.0929456882037458e-06, + "loss": 0.0748, + "step": 15218 + }, + { + "epoch": 6.37, + "learning_rate": 2.0919078779598123e-06, + "loss": 0.0819, + "step": 15219 + }, + { + "epoch": 6.37, + "learning_rate": 2.090870295026972e-06, + "loss": 0.0789, + "step": 15220 + }, + { + "epoch": 6.37, + "learning_rate": 2.0898329394350446e-06, + "loss": 0.0746, + "step": 15221 + }, + { + "epoch": 6.37, + "learning_rate": 2.0887958112138516e-06, + "loss": 0.0714, + "step": 15222 + }, + { + "epoch": 6.37, + "learning_rate": 2.0877589103932007e-06, + "loss": 0.0954, + "step": 15223 + }, + { + "epoch": 6.37, + "learning_rate": 2.0867222370028985e-06, + "loss": 0.1011, + "step": 15224 + }, + { + "epoch": 6.37, + "learning_rate": 2.0856857910727436e-06, + "loss": 0.0701, + "step": 15225 + }, + { + "epoch": 6.37, + "learning_rate": 2.0846495726325267e-06, + "loss": 0.1062, + "step": 15226 + }, + { + "epoch": 6.38, + "learning_rate": 2.0836135817120352e-06, + "loss": 0.0459, + "step": 15227 + }, + { + "epoch": 6.38, + "learning_rate": 2.0825778183410485e-06, + "loss": 0.0785, + "step": 15228 + }, + { + "epoch": 6.38, + "learning_rate": 2.0815422825493336e-06, + "loss": 0.1151, + "step": 15229 + }, + { + "epoch": 6.38, + "learning_rate": 2.0805069743666593e-06, + "loss": 0.0823, + "step": 15230 + }, + { + "epoch": 6.38, + "learning_rate": 2.0794718938227855e-06, + "loss": 0.0946, + "step": 15231 + }, + { + "epoch": 6.38, + "learning_rate": 2.078437040947462e-06, + "loss": 0.0926, + "step": 15232 + }, + { + "epoch": 6.38, + "learning_rate": 2.0774024157704364e-06, + "loss": 0.0794, + "step": 15233 + }, + { + "epoch": 6.38, + "learning_rate": 2.0763680183214485e-06, + "loss": 0.1051, + "step": 15234 + }, + { + "epoch": 6.38, + "learning_rate": 2.0753338486302334e-06, + "loss": 0.0725, + "step": 15235 + }, + { + "epoch": 6.38, + "learning_rate": 2.0742999067265125e-06, + "loss": 0.091, + "step": 15236 + }, + { + "epoch": 6.38, + "learning_rate": 2.0732661926400065e-06, + "loss": 0.0525, + "step": 15237 + }, + { + "epoch": 6.38, + "learning_rate": 2.07223270640043e-06, + "loss": 0.0917, + "step": 15238 + }, + { + "epoch": 6.38, + "learning_rate": 2.0711994480374895e-06, + "loss": 0.0621, + "step": 15239 + }, + { + "epoch": 6.38, + "learning_rate": 2.0701664175808843e-06, + "loss": 0.0761, + "step": 15240 + }, + { + "epoch": 6.38, + "learning_rate": 2.0691336150603124e-06, + "loss": 0.1114, + "step": 15241 + }, + { + "epoch": 6.38, + "learning_rate": 2.0681010405054535e-06, + "loss": 0.0802, + "step": 15242 + }, + { + "epoch": 6.38, + "learning_rate": 2.067068693945995e-06, + "loss": 0.1087, + "step": 15243 + }, + { + "epoch": 6.38, + "learning_rate": 2.0660365754116042e-06, + "loss": 0.0917, + "step": 15244 + }, + { + "epoch": 6.38, + "learning_rate": 2.0650046849319516e-06, + "loss": 0.0681, + "step": 15245 + }, + { + "epoch": 6.38, + "learning_rate": 2.063973022536698e-06, + "loss": 0.0731, + "step": 15246 + }, + { + "epoch": 6.38, + "learning_rate": 2.062941588255498e-06, + "loss": 0.081, + "step": 15247 + }, + { + "epoch": 6.38, + "learning_rate": 2.0619103821179975e-06, + "loss": 0.0566, + "step": 15248 + }, + { + "epoch": 6.38, + "learning_rate": 2.0608794041538426e-06, + "loss": 0.0903, + "step": 15249 + }, + { + "epoch": 6.38, + "learning_rate": 2.0598486543926623e-06, + "loss": 0.1197, + "step": 15250 + }, + { + "epoch": 6.39, + "learning_rate": 2.0588181328640856e-06, + "loss": 0.0987, + "step": 15251 + }, + { + "epoch": 6.39, + "learning_rate": 2.0577878395977347e-06, + "loss": 0.0553, + "step": 15252 + }, + { + "epoch": 6.39, + "learning_rate": 2.056757774623224e-06, + "loss": 0.077, + "step": 15253 + }, + { + "epoch": 6.39, + "learning_rate": 2.0557279379701632e-06, + "loss": 0.0711, + "step": 15254 + }, + { + "epoch": 6.39, + "learning_rate": 2.0546983296681554e-06, + "loss": 0.0822, + "step": 15255 + }, + { + "epoch": 6.39, + "learning_rate": 2.053668949746791e-06, + "loss": 0.1005, + "step": 15256 + }, + { + "epoch": 6.39, + "learning_rate": 2.0526397982356617e-06, + "loss": 0.0839, + "step": 15257 + }, + { + "epoch": 6.39, + "learning_rate": 2.0516108751643483e-06, + "loss": 0.1148, + "step": 15258 + }, + { + "epoch": 6.39, + "learning_rate": 2.050582180562427e-06, + "loss": 0.1279, + "step": 15259 + }, + { + "epoch": 6.39, + "learning_rate": 2.0495537144594678e-06, + "loss": 0.0718, + "step": 15260 + }, + { + "epoch": 6.39, + "learning_rate": 2.0485254768850317e-06, + "loss": 0.0816, + "step": 15261 + }, + { + "epoch": 6.39, + "learning_rate": 2.0474974678686764e-06, + "loss": 0.0784, + "step": 15262 + }, + { + "epoch": 6.39, + "learning_rate": 2.046469687439947e-06, + "loss": 0.075, + "step": 15263 + }, + { + "epoch": 6.39, + "learning_rate": 2.045442135628389e-06, + "loss": 0.0981, + "step": 15264 + }, + { + "epoch": 6.39, + "learning_rate": 2.044414812463541e-06, + "loss": 0.0727, + "step": 15265 + }, + { + "epoch": 6.39, + "learning_rate": 2.0433877179749273e-06, + "loss": 0.0525, + "step": 15266 + }, + { + "epoch": 6.39, + "learning_rate": 2.0423608521920723e-06, + "loss": 0.0993, + "step": 15267 + }, + { + "epoch": 6.39, + "learning_rate": 2.0413342151444928e-06, + "loss": 0.1002, + "step": 15268 + }, + { + "epoch": 6.39, + "learning_rate": 2.040307806861702e-06, + "loss": 0.065, + "step": 15269 + }, + { + "epoch": 6.39, + "learning_rate": 2.0392816273731973e-06, + "loss": 0.056, + "step": 15270 + }, + { + "epoch": 6.39, + "learning_rate": 2.0382556767084784e-06, + "loss": 0.0821, + "step": 15271 + }, + { + "epoch": 6.39, + "learning_rate": 2.0372299548970343e-06, + "loss": 0.0702, + "step": 15272 + }, + { + "epoch": 6.39, + "learning_rate": 2.0362044619683485e-06, + "loss": 0.1083, + "step": 15273 + }, + { + "epoch": 6.39, + "learning_rate": 2.0351791979518997e-06, + "loss": 0.0707, + "step": 15274 + }, + { + "epoch": 6.4, + "learning_rate": 2.0341541628771567e-06, + "loss": 0.0821, + "step": 15275 + }, + { + "epoch": 6.4, + "learning_rate": 2.0331293567735865e-06, + "loss": 0.055, + "step": 15276 + }, + { + "epoch": 6.4, + "learning_rate": 2.0321047796706393e-06, + "loss": 0.0862, + "step": 15277 + }, + { + "epoch": 6.4, + "learning_rate": 2.0310804315977717e-06, + "loss": 0.1026, + "step": 15278 + }, + { + "epoch": 6.4, + "learning_rate": 2.030056312584424e-06, + "loss": 0.0964, + "step": 15279 + }, + { + "epoch": 6.4, + "learning_rate": 2.0290324226600357e-06, + "loss": 0.104, + "step": 15280 + }, + { + "epoch": 6.4, + "learning_rate": 2.028008761854037e-06, + "loss": 0.101, + "step": 15281 + }, + { + "epoch": 6.4, + "learning_rate": 2.0269853301958554e-06, + "loss": 0.0903, + "step": 15282 + }, + { + "epoch": 6.4, + "learning_rate": 2.0259621277149023e-06, + "loss": 0.077, + "step": 15283 + }, + { + "epoch": 6.4, + "learning_rate": 2.0249391544405916e-06, + "loss": 0.1145, + "step": 15284 + }, + { + "epoch": 6.4, + "learning_rate": 2.0239164104023278e-06, + "loss": 0.0936, + "step": 15285 + }, + { + "epoch": 6.4, + "learning_rate": 2.0228938956295087e-06, + "loss": 0.0487, + "step": 15286 + }, + { + "epoch": 6.4, + "learning_rate": 2.0218716101515268e-06, + "loss": 0.1228, + "step": 15287 + }, + { + "epoch": 6.4, + "learning_rate": 2.020849553997768e-06, + "loss": 0.1073, + "step": 15288 + }, + { + "epoch": 6.4, + "learning_rate": 2.019827727197605e-06, + "loss": 0.0721, + "step": 15289 + }, + { + "epoch": 6.4, + "learning_rate": 2.018806129780416e-06, + "loss": 0.1021, + "step": 15290 + }, + { + "epoch": 6.4, + "learning_rate": 2.0177847617755595e-06, + "loss": 0.0533, + "step": 15291 + }, + { + "epoch": 6.4, + "learning_rate": 2.0167636232123967e-06, + "loss": 0.0427, + "step": 15292 + }, + { + "epoch": 6.4, + "learning_rate": 2.015742714120279e-06, + "loss": 0.082, + "step": 15293 + }, + { + "epoch": 6.4, + "learning_rate": 2.0147220345285524e-06, + "loss": 0.0755, + "step": 15294 + }, + { + "epoch": 6.4, + "learning_rate": 2.013701584466555e-06, + "loss": 0.1105, + "step": 15295 + }, + { + "epoch": 6.4, + "learning_rate": 2.012681363963621e-06, + "loss": 0.0404, + "step": 15296 + }, + { + "epoch": 6.4, + "learning_rate": 2.011661373049071e-06, + "loss": 0.1166, + "step": 15297 + }, + { + "epoch": 6.4, + "learning_rate": 2.0106416117522255e-06, + "loss": 0.0799, + "step": 15298 + }, + { + "epoch": 6.41, + "learning_rate": 2.009622080102398e-06, + "loss": 0.09, + "step": 15299 + }, + { + "epoch": 6.41, + "learning_rate": 2.008602778128892e-06, + "loss": 0.0517, + "step": 15300 + }, + { + "epoch": 6.41, + "learning_rate": 2.007583705861009e-06, + "loss": 0.1018, + "step": 15301 + }, + { + "epoch": 6.41, + "learning_rate": 2.00656486332804e-06, + "loss": 0.0705, + "step": 15302 + }, + { + "epoch": 6.41, + "learning_rate": 2.0055462505592725e-06, + "loss": 0.0839, + "step": 15303 + }, + { + "epoch": 6.41, + "learning_rate": 2.004527867583982e-06, + "loss": 0.0649, + "step": 15304 + }, + { + "epoch": 6.41, + "learning_rate": 2.003509714431442e-06, + "loss": 0.0991, + "step": 15305 + }, + { + "epoch": 6.41, + "learning_rate": 2.0024917911309207e-06, + "loss": 0.0896, + "step": 15306 + }, + { + "epoch": 6.41, + "learning_rate": 2.0014740977116743e-06, + "loss": 0.0785, + "step": 15307 + }, + { + "epoch": 6.41, + "learning_rate": 2.0004566342029586e-06, + "loss": 0.0783, + "step": 15308 + }, + { + "epoch": 6.41, + "learning_rate": 1.99943940063402e-06, + "loss": 0.0617, + "step": 15309 + }, + { + "epoch": 6.41, + "learning_rate": 1.9984223970340965e-06, + "loss": 0.1016, + "step": 15310 + }, + { + "epoch": 6.41, + "learning_rate": 1.997405623432419e-06, + "loss": 0.0764, + "step": 15311 + }, + { + "epoch": 6.41, + "learning_rate": 1.996389079858214e-06, + "loss": 0.0764, + "step": 15312 + }, + { + "epoch": 6.41, + "learning_rate": 1.9953727663407043e-06, + "loss": 0.0679, + "step": 15313 + }, + { + "epoch": 6.41, + "learning_rate": 1.9943566829091e-06, + "loss": 0.0698, + "step": 15314 + }, + { + "epoch": 6.41, + "learning_rate": 1.9933408295926095e-06, + "loss": 0.0677, + "step": 15315 + }, + { + "epoch": 6.41, + "learning_rate": 1.9923252064204313e-06, + "loss": 0.0525, + "step": 15316 + }, + { + "epoch": 6.41, + "learning_rate": 1.9913098134217612e-06, + "loss": 0.0881, + "step": 15317 + }, + { + "epoch": 6.41, + "learning_rate": 1.9902946506257825e-06, + "loss": 0.0696, + "step": 15318 + }, + { + "epoch": 6.41, + "learning_rate": 1.989279718061675e-06, + "loss": 0.095, + "step": 15319 + }, + { + "epoch": 6.41, + "learning_rate": 1.988265015758615e-06, + "loss": 0.0556, + "step": 15320 + }, + { + "epoch": 6.41, + "learning_rate": 1.9872505437457666e-06, + "loss": 0.1042, + "step": 15321 + }, + { + "epoch": 6.41, + "learning_rate": 1.986236302052291e-06, + "loss": 0.0944, + "step": 15322 + }, + { + "epoch": 6.42, + "learning_rate": 1.985222290707344e-06, + "loss": 0.1087, + "step": 15323 + }, + { + "epoch": 6.42, + "learning_rate": 1.9842085097400678e-06, + "loss": 0.0571, + "step": 15324 + }, + { + "epoch": 6.42, + "learning_rate": 1.9831949591796052e-06, + "loss": 0.1036, + "step": 15325 + }, + { + "epoch": 6.42, + "learning_rate": 1.9821816390550887e-06, + "loss": 0.0771, + "step": 15326 + }, + { + "epoch": 6.42, + "learning_rate": 1.981168549395647e-06, + "loss": 0.1008, + "step": 15327 + }, + { + "epoch": 6.42, + "learning_rate": 1.9801556902303986e-06, + "loss": 0.0746, + "step": 15328 + }, + { + "epoch": 6.42, + "learning_rate": 1.9791430615884622e-06, + "loss": 0.087, + "step": 15329 + }, + { + "epoch": 6.42, + "learning_rate": 1.9781306634989373e-06, + "loss": 0.0939, + "step": 15330 + }, + { + "epoch": 6.42, + "learning_rate": 1.9771184959909273e-06, + "loss": 0.083, + "step": 15331 + }, + { + "epoch": 6.42, + "learning_rate": 1.976106559093528e-06, + "loss": 0.0843, + "step": 15332 + }, + { + "epoch": 6.42, + "learning_rate": 1.9750948528358283e-06, + "loss": 0.0991, + "step": 15333 + }, + { + "epoch": 6.42, + "learning_rate": 1.9740833772469025e-06, + "loss": 0.0816, + "step": 15334 + }, + { + "epoch": 6.42, + "learning_rate": 1.973072132355828e-06, + "loss": 0.0678, + "step": 15335 + }, + { + "epoch": 6.42, + "learning_rate": 1.9720611181916716e-06, + "loss": 0.0686, + "step": 15336 + }, + { + "epoch": 6.42, + "learning_rate": 1.9710503347834975e-06, + "loss": 0.0766, + "step": 15337 + }, + { + "epoch": 6.42, + "learning_rate": 1.9700397821603546e-06, + "loss": 0.1118, + "step": 15338 + }, + { + "epoch": 6.42, + "learning_rate": 1.9690294603512915e-06, + "loss": 0.0656, + "step": 15339 + }, + { + "epoch": 6.42, + "learning_rate": 1.968019369385351e-06, + "loss": 0.0869, + "step": 15340 + }, + { + "epoch": 6.42, + "learning_rate": 1.9670095092915666e-06, + "loss": 0.0868, + "step": 15341 + }, + { + "epoch": 6.42, + "learning_rate": 1.9659998800989642e-06, + "loss": 0.0864, + "step": 15342 + }, + { + "epoch": 6.42, + "learning_rate": 1.9649904818365673e-06, + "loss": 0.096, + "step": 15343 + }, + { + "epoch": 6.42, + "learning_rate": 1.963981314533391e-06, + "loss": 0.0787, + "step": 15344 + }, + { + "epoch": 6.42, + "learning_rate": 1.962972378218437e-06, + "loss": 0.0652, + "step": 15345 + }, + { + "epoch": 6.42, + "learning_rate": 1.9619636729207116e-06, + "loss": 0.0911, + "step": 15346 + }, + { + "epoch": 6.43, + "learning_rate": 1.9609551986692067e-06, + "loss": 0.0688, + "step": 15347 + }, + { + "epoch": 6.43, + "learning_rate": 1.9599469554929116e-06, + "loss": 0.0916, + "step": 15348 + }, + { + "epoch": 6.43, + "learning_rate": 1.9589389434208063e-06, + "loss": 0.1026, + "step": 15349 + }, + { + "epoch": 6.43, + "learning_rate": 1.9579311624818674e-06, + "loss": 0.1018, + "step": 15350 + }, + { + "epoch": 6.43, + "learning_rate": 1.9569236127050596e-06, + "loss": 0.0661, + "step": 15351 + }, + { + "epoch": 6.43, + "learning_rate": 1.955916294119344e-06, + "loss": 0.0718, + "step": 15352 + }, + { + "epoch": 6.43, + "learning_rate": 1.954909206753678e-06, + "loss": 0.0781, + "step": 15353 + }, + { + "epoch": 6.43, + "learning_rate": 1.953902350637006e-06, + "loss": 0.0836, + "step": 15354 + }, + { + "epoch": 6.43, + "learning_rate": 1.9528957257982752e-06, + "loss": 0.0657, + "step": 15355 + }, + { + "epoch": 6.43, + "learning_rate": 1.9518893322664124e-06, + "loss": 0.0784, + "step": 15356 + }, + { + "epoch": 6.43, + "learning_rate": 1.9508831700703477e-06, + "loss": 0.0803, + "step": 15357 + }, + { + "epoch": 6.43, + "learning_rate": 1.9498772392390084e-06, + "loss": 0.0647, + "step": 15358 + }, + { + "epoch": 6.43, + "learning_rate": 1.9488715398013002e-06, + "loss": 0.0873, + "step": 15359 + }, + { + "epoch": 6.43, + "learning_rate": 1.9478660717861362e-06, + "loss": 0.0837, + "step": 15360 + }, + { + "epoch": 6.43, + "learning_rate": 1.946860835222416e-06, + "loss": 0.0908, + "step": 15361 + }, + { + "epoch": 6.43, + "learning_rate": 1.9458558301390353e-06, + "loss": 0.0487, + "step": 15362 + }, + { + "epoch": 6.43, + "learning_rate": 1.9448510565648813e-06, + "loss": 0.1196, + "step": 15363 + }, + { + "epoch": 6.43, + "learning_rate": 1.9438465145288377e-06, + "loss": 0.0769, + "step": 15364 + }, + { + "epoch": 6.43, + "learning_rate": 1.942842204059775e-06, + "loss": 0.0767, + "step": 15365 + }, + { + "epoch": 6.43, + "learning_rate": 1.9418381251865636e-06, + "loss": 0.0949, + "step": 15366 + }, + { + "epoch": 6.43, + "learning_rate": 1.940834277938065e-06, + "loss": 0.0808, + "step": 15367 + }, + { + "epoch": 6.43, + "learning_rate": 1.9398306623431327e-06, + "loss": 0.1119, + "step": 15368 + }, + { + "epoch": 6.43, + "learning_rate": 1.9388272784306163e-06, + "loss": 0.0912, + "step": 15369 + }, + { + "epoch": 6.44, + "learning_rate": 1.9378241262293583e-06, + "loss": 0.0776, + "step": 15370 + }, + { + "epoch": 6.44, + "learning_rate": 1.936821205768189e-06, + "loss": 0.062, + "step": 15371 + }, + { + "epoch": 6.44, + "learning_rate": 1.9358185170759394e-06, + "loss": 0.0768, + "step": 15372 + }, + { + "epoch": 6.44, + "learning_rate": 1.93481606018143e-06, + "loss": 0.095, + "step": 15373 + }, + { + "epoch": 6.44, + "learning_rate": 1.9338138351134773e-06, + "loss": 0.0929, + "step": 15374 + }, + { + "epoch": 6.44, + "learning_rate": 1.932811841900887e-06, + "loss": 0.0751, + "step": 15375 + }, + { + "epoch": 6.44, + "learning_rate": 1.9318100805724617e-06, + "loss": 0.0926, + "step": 15376 + }, + { + "epoch": 6.44, + "learning_rate": 1.9308085511569984e-06, + "loss": 0.073, + "step": 15377 + }, + { + "epoch": 6.44, + "learning_rate": 1.9298072536832834e-06, + "loss": 0.0421, + "step": 15378 + }, + { + "epoch": 6.44, + "learning_rate": 1.9288061881800945e-06, + "loss": 0.0576, + "step": 15379 + }, + { + "epoch": 6.44, + "learning_rate": 1.92780535467621e-06, + "loss": 0.106, + "step": 15380 + }, + { + "epoch": 6.44, + "learning_rate": 1.9268047532003962e-06, + "loss": 0.0936, + "step": 15381 + }, + { + "epoch": 6.44, + "learning_rate": 1.9258043837814164e-06, + "loss": 0.0728, + "step": 15382 + }, + { + "epoch": 6.44, + "learning_rate": 1.9248042464480255e-06, + "loss": 0.0798, + "step": 15383 + }, + { + "epoch": 6.44, + "learning_rate": 1.9238043412289696e-06, + "loss": 0.088, + "step": 15384 + }, + { + "epoch": 6.44, + "learning_rate": 1.9228046681529954e-06, + "loss": 0.1014, + "step": 15385 + }, + { + "epoch": 6.44, + "learning_rate": 1.92180522724883e-06, + "loss": 0.0671, + "step": 15386 + }, + { + "epoch": 6.44, + "learning_rate": 1.9208060185452037e-06, + "loss": 0.082, + "step": 15387 + }, + { + "epoch": 6.44, + "learning_rate": 1.9198070420708405e-06, + "loss": 0.0795, + "step": 15388 + }, + { + "epoch": 6.44, + "learning_rate": 1.918808297854453e-06, + "loss": 0.0988, + "step": 15389 + }, + { + "epoch": 6.44, + "learning_rate": 1.9178097859247502e-06, + "loss": 0.0596, + "step": 15390 + }, + { + "epoch": 6.44, + "learning_rate": 1.9168115063104355e-06, + "loss": 0.0862, + "step": 15391 + }, + { + "epoch": 6.44, + "learning_rate": 1.915813459040199e-06, + "loss": 0.089, + "step": 15392 + }, + { + "epoch": 6.44, + "learning_rate": 1.91481564414273e-06, + "loss": 0.113, + "step": 15393 + }, + { + "epoch": 6.45, + "learning_rate": 1.9138180616467105e-06, + "loss": 0.0811, + "step": 15394 + }, + { + "epoch": 6.45, + "learning_rate": 1.912820711580816e-06, + "loss": 0.0709, + "step": 15395 + }, + { + "epoch": 6.45, + "learning_rate": 1.911823593973714e-06, + "loss": 0.0679, + "step": 15396 + }, + { + "epoch": 6.45, + "learning_rate": 1.910826708854068e-06, + "loss": 0.0947, + "step": 15397 + }, + { + "epoch": 6.45, + "learning_rate": 1.9098300562505266e-06, + "loss": 0.1242, + "step": 15398 + }, + { + "epoch": 6.45, + "learning_rate": 1.9088336361917415e-06, + "loss": 0.0767, + "step": 15399 + }, + { + "epoch": 6.45, + "learning_rate": 1.9078374487063536e-06, + "loss": 0.0597, + "step": 15400 + }, + { + "epoch": 6.45, + "learning_rate": 1.9068414938229996e-06, + "loss": 0.0804, + "step": 15401 + }, + { + "epoch": 6.45, + "learning_rate": 1.9058457715703038e-06, + "loss": 0.0633, + "step": 15402 + }, + { + "epoch": 6.45, + "learning_rate": 1.9048502819768877e-06, + "loss": 0.0782, + "step": 15403 + }, + { + "epoch": 6.45, + "learning_rate": 1.903855025071366e-06, + "loss": 0.0845, + "step": 15404 + }, + { + "epoch": 6.45, + "learning_rate": 1.9028600008823517e-06, + "loss": 0.09, + "step": 15405 + }, + { + "epoch": 6.45, + "learning_rate": 1.901865209438437e-06, + "loss": 0.0917, + "step": 15406 + }, + { + "epoch": 6.45, + "learning_rate": 1.9008706507682217e-06, + "loss": 0.0871, + "step": 15407 + }, + { + "epoch": 6.45, + "learning_rate": 1.8998763249002928e-06, + "loss": 0.0657, + "step": 15408 + }, + { + "epoch": 6.45, + "learning_rate": 1.8988822318632305e-06, + "loss": 0.0579, + "step": 15409 + }, + { + "epoch": 6.45, + "learning_rate": 1.897888371685609e-06, + "loss": 0.0778, + "step": 15410 + }, + { + "epoch": 6.45, + "learning_rate": 1.8968947443959995e-06, + "loss": 0.0683, + "step": 15411 + }, + { + "epoch": 6.45, + "learning_rate": 1.8959013500229584e-06, + "loss": 0.0723, + "step": 15412 + }, + { + "epoch": 6.45, + "learning_rate": 1.8949081885950403e-06, + "loss": 0.0686, + "step": 15413 + }, + { + "epoch": 6.45, + "learning_rate": 1.8939152601407952e-06, + "loss": 0.0635, + "step": 15414 + }, + { + "epoch": 6.45, + "learning_rate": 1.8929225646887617e-06, + "loss": 0.0956, + "step": 15415 + }, + { + "epoch": 6.45, + "learning_rate": 1.8919301022674752e-06, + "loss": 0.1139, + "step": 15416 + }, + { + "epoch": 6.45, + "learning_rate": 1.890937872905463e-06, + "loss": 0.0914, + "step": 15417 + }, + { + "epoch": 6.46, + "learning_rate": 1.8899458766312484e-06, + "loss": 0.1168, + "step": 15418 + }, + { + "epoch": 6.46, + "learning_rate": 1.8889541134733401e-06, + "loss": 0.0744, + "step": 15419 + }, + { + "epoch": 6.46, + "learning_rate": 1.8879625834602488e-06, + "loss": 0.0642, + "step": 15420 + }, + { + "epoch": 6.46, + "learning_rate": 1.8869712866204748e-06, + "loss": 0.0822, + "step": 15421 + }, + { + "epoch": 6.46, + "learning_rate": 1.8859802229825109e-06, + "loss": 0.0684, + "step": 15422 + }, + { + "epoch": 6.46, + "learning_rate": 1.8849893925748474e-06, + "loss": 0.0957, + "step": 15423 + }, + { + "epoch": 6.46, + "learning_rate": 1.8839987954259609e-06, + "loss": 0.078, + "step": 15424 + }, + { + "epoch": 6.46, + "learning_rate": 1.883008431564327e-06, + "loss": 0.0764, + "step": 15425 + }, + { + "epoch": 6.46, + "learning_rate": 1.8820183010184157e-06, + "loss": 0.1057, + "step": 15426 + }, + { + "epoch": 6.46, + "learning_rate": 1.8810284038166815e-06, + "loss": 0.0593, + "step": 15427 + }, + { + "epoch": 6.46, + "learning_rate": 1.880038739987582e-06, + "loss": 0.0686, + "step": 15428 + }, + { + "epoch": 6.46, + "learning_rate": 1.8790493095595618e-06, + "loss": 0.0886, + "step": 15429 + }, + { + "epoch": 6.46, + "learning_rate": 1.8780601125610643e-06, + "loss": 0.0795, + "step": 15430 + }, + { + "epoch": 6.46, + "learning_rate": 1.8770711490205207e-06, + "loss": 0.0824, + "step": 15431 + }, + { + "epoch": 6.46, + "learning_rate": 1.8760824189663618e-06, + "loss": 0.0775, + "step": 15432 + }, + { + "epoch": 6.46, + "learning_rate": 1.8750939224270014e-06, + "loss": 0.0763, + "step": 15433 + }, + { + "epoch": 6.46, + "learning_rate": 1.8741056594308571e-06, + "loss": 0.0712, + "step": 15434 + }, + { + "epoch": 6.46, + "learning_rate": 1.8731176300063347e-06, + "loss": 0.0806, + "step": 15435 + }, + { + "epoch": 6.46, + "learning_rate": 1.8721298341818339e-06, + "loss": 0.0542, + "step": 15436 + }, + { + "epoch": 6.46, + "learning_rate": 1.8711422719857487e-06, + "loss": 0.07, + "step": 15437 + }, + { + "epoch": 6.46, + "learning_rate": 1.8701549434464684e-06, + "loss": 0.0852, + "step": 15438 + }, + { + "epoch": 6.46, + "learning_rate": 1.869167848592367e-06, + "loss": 0.1039, + "step": 15439 + }, + { + "epoch": 6.46, + "learning_rate": 1.8681809874518208e-06, + "loss": 0.0915, + "step": 15440 + }, + { + "epoch": 6.46, + "learning_rate": 1.8671943600531962e-06, + "loss": 0.0983, + "step": 15441 + }, + { + "epoch": 6.47, + "learning_rate": 1.8662079664248522e-06, + "loss": 0.047, + "step": 15442 + }, + { + "epoch": 6.47, + "learning_rate": 1.8652218065951432e-06, + "loss": 0.1026, + "step": 15443 + }, + { + "epoch": 6.47, + "learning_rate": 1.8642358805924144e-06, + "loss": 0.0711, + "step": 15444 + }, + { + "epoch": 6.47, + "learning_rate": 1.8632501884450082e-06, + "loss": 0.1029, + "step": 15445 + }, + { + "epoch": 6.47, + "learning_rate": 1.8622647301812546e-06, + "loss": 0.1079, + "step": 15446 + }, + { + "epoch": 6.47, + "learning_rate": 1.8612795058294775e-06, + "loss": 0.0836, + "step": 15447 + }, + { + "epoch": 6.47, + "learning_rate": 1.8602945154179997e-06, + "loss": 0.0754, + "step": 15448 + }, + { + "epoch": 6.47, + "learning_rate": 1.8593097589751318e-06, + "loss": 0.0884, + "step": 15449 + }, + { + "epoch": 6.47, + "learning_rate": 1.8583252365291814e-06, + "loss": 0.0766, + "step": 15450 + }, + { + "epoch": 6.47, + "learning_rate": 1.857340948108448e-06, + "loss": 0.0849, + "step": 15451 + }, + { + "epoch": 6.47, + "learning_rate": 1.8563568937412247e-06, + "loss": 0.0885, + "step": 15452 + }, + { + "epoch": 6.47, + "learning_rate": 1.855373073455794e-06, + "loss": 0.0845, + "step": 15453 + }, + { + "epoch": 6.47, + "learning_rate": 1.8543894872804368e-06, + "loss": 0.0834, + "step": 15454 + }, + { + "epoch": 6.47, + "learning_rate": 1.8534061352434263e-06, + "loss": 0.084, + "step": 15455 + }, + { + "epoch": 6.47, + "learning_rate": 1.8524230173730261e-06, + "loss": 0.0745, + "step": 15456 + }, + { + "epoch": 6.47, + "learning_rate": 1.8514401336974975e-06, + "loss": 0.1124, + "step": 15457 + }, + { + "epoch": 6.47, + "learning_rate": 1.8504574842450907e-06, + "loss": 0.0824, + "step": 15458 + }, + { + "epoch": 6.47, + "learning_rate": 1.8494750690440533e-06, + "loss": 0.0709, + "step": 15459 + }, + { + "epoch": 6.47, + "learning_rate": 1.8484928881226205e-06, + "loss": 0.0754, + "step": 15460 + }, + { + "epoch": 6.47, + "learning_rate": 1.8475109415090264e-06, + "loss": 0.0705, + "step": 15461 + }, + { + "epoch": 6.47, + "learning_rate": 1.8465292292314963e-06, + "loss": 0.0823, + "step": 15462 + }, + { + "epoch": 6.47, + "learning_rate": 1.8455477513182473e-06, + "loss": 0.089, + "step": 15463 + }, + { + "epoch": 6.47, + "learning_rate": 1.8445665077974928e-06, + "loss": 0.1006, + "step": 15464 + }, + { + "epoch": 6.47, + "learning_rate": 1.8435854986974399e-06, + "loss": 0.084, + "step": 15465 + }, + { + "epoch": 6.48, + "learning_rate": 1.8426047240462807e-06, + "loss": 0.1041, + "step": 15466 + }, + { + "epoch": 6.48, + "learning_rate": 1.8416241838722093e-06, + "loss": 0.0706, + "step": 15467 + }, + { + "epoch": 6.48, + "learning_rate": 1.840643878203412e-06, + "loss": 0.0696, + "step": 15468 + }, + { + "epoch": 6.48, + "learning_rate": 1.8396638070680685e-06, + "loss": 0.0779, + "step": 15469 + }, + { + "epoch": 6.48, + "learning_rate": 1.8386839704943449e-06, + "loss": 0.0984, + "step": 15470 + }, + { + "epoch": 6.48, + "learning_rate": 1.837704368510409e-06, + "loss": 0.0896, + "step": 15471 + }, + { + "epoch": 6.48, + "learning_rate": 1.836725001144417e-06, + "loss": 0.0775, + "step": 15472 + }, + { + "epoch": 6.48, + "learning_rate": 1.8357458684245234e-06, + "loss": 0.0882, + "step": 15473 + }, + { + "epoch": 6.48, + "learning_rate": 1.8347669703788683e-06, + "loss": 0.0753, + "step": 15474 + }, + { + "epoch": 6.48, + "learning_rate": 1.83378830703559e-06, + "loss": 0.0791, + "step": 15475 + }, + { + "epoch": 6.48, + "learning_rate": 1.832809878422821e-06, + "loss": 0.0847, + "step": 15476 + }, + { + "epoch": 6.48, + "learning_rate": 1.8318316845686834e-06, + "loss": 0.0809, + "step": 15477 + }, + { + "epoch": 6.48, + "learning_rate": 1.830853725501297e-06, + "loss": 0.0602, + "step": 15478 + }, + { + "epoch": 6.48, + "learning_rate": 1.8298760012487726e-06, + "loss": 0.0987, + "step": 15479 + }, + { + "epoch": 6.48, + "learning_rate": 1.828898511839211e-06, + "loss": 0.0938, + "step": 15480 + }, + { + "epoch": 6.48, + "learning_rate": 1.8279212573007099e-06, + "loss": 0.1222, + "step": 15481 + }, + { + "epoch": 6.48, + "learning_rate": 1.826944237661361e-06, + "loss": 0.0844, + "step": 15482 + }, + { + "epoch": 6.48, + "learning_rate": 1.8259674529492476e-06, + "loss": 0.0666, + "step": 15483 + }, + { + "epoch": 6.48, + "learning_rate": 1.824990903192445e-06, + "loss": 0.073, + "step": 15484 + }, + { + "epoch": 6.48, + "learning_rate": 1.8240145884190264e-06, + "loss": 0.0588, + "step": 15485 + }, + { + "epoch": 6.48, + "learning_rate": 1.8230385086570546e-06, + "loss": 0.0886, + "step": 15486 + }, + { + "epoch": 6.48, + "learning_rate": 1.8220626639345817e-06, + "loss": 0.0772, + "step": 15487 + }, + { + "epoch": 6.48, + "learning_rate": 1.8210870542796622e-06, + "loss": 0.1025, + "step": 15488 + }, + { + "epoch": 6.48, + "learning_rate": 1.8201116797203355e-06, + "loss": 0.0705, + "step": 15489 + }, + { + "epoch": 6.49, + "learning_rate": 1.8191365402846417e-06, + "loss": 0.0535, + "step": 15490 + }, + { + "epoch": 6.49, + "learning_rate": 1.8181616360006094e-06, + "loss": 0.0975, + "step": 15491 + }, + { + "epoch": 6.49, + "learning_rate": 1.8171869668962594e-06, + "loss": 0.0777, + "step": 15492 + }, + { + "epoch": 6.49, + "learning_rate": 1.8162125329996105e-06, + "loss": 0.0518, + "step": 15493 + }, + { + "epoch": 6.49, + "learning_rate": 1.815238334338667e-06, + "loss": 0.0665, + "step": 15494 + }, + { + "epoch": 6.49, + "learning_rate": 1.8142643709414355e-06, + "loss": 0.0883, + "step": 15495 + }, + { + "epoch": 6.49, + "learning_rate": 1.813290642835911e-06, + "loss": 0.0653, + "step": 15496 + }, + { + "epoch": 6.49, + "learning_rate": 1.8123171500500814e-06, + "loss": 0.0458, + "step": 15497 + }, + { + "epoch": 6.49, + "learning_rate": 1.8113438926119297e-06, + "loss": 0.0869, + "step": 15498 + }, + { + "epoch": 6.49, + "learning_rate": 1.8103708705494316e-06, + "loss": 0.092, + "step": 15499 + }, + { + "epoch": 6.49, + "learning_rate": 1.809398083890559e-06, + "loss": 0.0797, + "step": 15500 + }, + { + "epoch": 6.49, + "learning_rate": 1.808425532663266e-06, + "loss": 0.084, + "step": 15501 + }, + { + "epoch": 6.49, + "learning_rate": 1.8074532168955128e-06, + "loss": 0.0594, + "step": 15502 + }, + { + "epoch": 6.49, + "learning_rate": 1.8064811366152468e-06, + "loss": 0.0855, + "step": 15503 + }, + { + "epoch": 6.49, + "learning_rate": 1.8055092918504103e-06, + "loss": 0.0568, + "step": 15504 + }, + { + "epoch": 6.49, + "learning_rate": 1.8045376826289374e-06, + "loss": 0.0883, + "step": 15505 + }, + { + "epoch": 6.49, + "learning_rate": 1.8035663089787582e-06, + "loss": 0.1022, + "step": 15506 + }, + { + "epoch": 6.49, + "learning_rate": 1.80259517092779e-06, + "loss": 0.0616, + "step": 15507 + }, + { + "epoch": 6.49, + "learning_rate": 1.8016242685039486e-06, + "loss": 0.079, + "step": 15508 + }, + { + "epoch": 6.49, + "learning_rate": 1.8006536017351438e-06, + "loss": 0.084, + "step": 15509 + }, + { + "epoch": 6.49, + "learning_rate": 1.7996831706492745e-06, + "loss": 0.0961, + "step": 15510 + }, + { + "epoch": 6.49, + "learning_rate": 1.798712975274236e-06, + "loss": 0.0689, + "step": 15511 + }, + { + "epoch": 6.49, + "learning_rate": 1.7977430156379173e-06, + "loss": 0.0799, + "step": 15512 + }, + { + "epoch": 6.49, + "learning_rate": 1.7967732917681946e-06, + "loss": 0.0747, + "step": 15513 + }, + { + "epoch": 6.5, + "learning_rate": 1.795803803692947e-06, + "loss": 0.0868, + "step": 15514 + }, + { + "epoch": 6.5, + "learning_rate": 1.7948345514400356e-06, + "loss": 0.0968, + "step": 15515 + }, + { + "epoch": 6.5, + "learning_rate": 1.7938655350373246e-06, + "loss": 0.0793, + "step": 15516 + }, + { + "epoch": 6.5, + "learning_rate": 1.7928967545126674e-06, + "loss": 0.0978, + "step": 15517 + }, + { + "epoch": 6.5, + "learning_rate": 1.7919282098939095e-06, + "loss": 0.0584, + "step": 15518 + }, + { + "epoch": 6.5, + "learning_rate": 1.7909599012088908e-06, + "loss": 0.0498, + "step": 15519 + }, + { + "epoch": 6.5, + "learning_rate": 1.789991828485449e-06, + "loss": 0.0937, + "step": 15520 + }, + { + "epoch": 6.5, + "learning_rate": 1.789023991751403e-06, + "loss": 0.0835, + "step": 15521 + }, + { + "epoch": 6.5, + "learning_rate": 1.788056391034576e-06, + "loss": 0.0685, + "step": 15522 + }, + { + "epoch": 6.5, + "learning_rate": 1.7870890263627817e-06, + "loss": 0.0886, + "step": 15523 + }, + { + "epoch": 6.5, + "learning_rate": 1.7861218977638252e-06, + "loss": 0.0806, + "step": 15524 + }, + { + "epoch": 6.5, + "learning_rate": 1.7851550052655052e-06, + "loss": 0.0801, + "step": 15525 + }, + { + "epoch": 6.5, + "learning_rate": 1.7841883488956157e-06, + "loss": 0.0688, + "step": 15526 + }, + { + "epoch": 6.5, + "learning_rate": 1.7832219286819442e-06, + "loss": 0.1044, + "step": 15527 + }, + { + "epoch": 6.5, + "learning_rate": 1.7822557446522637e-06, + "loss": 0.0741, + "step": 15528 + }, + { + "epoch": 6.5, + "learning_rate": 1.781289796834349e-06, + "loss": 0.0708, + "step": 15529 + }, + { + "epoch": 6.5, + "learning_rate": 1.7803240852559667e-06, + "loss": 0.0867, + "step": 15530 + }, + { + "epoch": 6.5, + "learning_rate": 1.7793586099448745e-06, + "loss": 0.0651, + "step": 15531 + }, + { + "epoch": 6.5, + "learning_rate": 1.7783933709288249e-06, + "loss": 0.0988, + "step": 15532 + }, + { + "epoch": 6.5, + "learning_rate": 1.7774283682355642e-06, + "loss": 0.0839, + "step": 15533 + }, + { + "epoch": 6.5, + "learning_rate": 1.7764636018928249e-06, + "loss": 0.055, + "step": 15534 + }, + { + "epoch": 6.5, + "learning_rate": 1.7754990719283427e-06, + "loss": 0.083, + "step": 15535 + }, + { + "epoch": 6.5, + "learning_rate": 1.774534778369842e-06, + "loss": 0.0905, + "step": 15536 + }, + { + "epoch": 6.5, + "learning_rate": 1.7735707212450414e-06, + "loss": 0.0762, + "step": 15537 + }, + { + "epoch": 6.51, + "learning_rate": 1.7726069005816482e-06, + "loss": 0.0912, + "step": 15538 + }, + { + "epoch": 6.51, + "learning_rate": 1.7716433164073698e-06, + "loss": 0.0708, + "step": 15539 + }, + { + "epoch": 6.51, + "learning_rate": 1.770679968749901e-06, + "loss": 0.0814, + "step": 15540 + }, + { + "epoch": 6.51, + "learning_rate": 1.7697168576369373e-06, + "loss": 0.0873, + "step": 15541 + }, + { + "epoch": 6.51, + "learning_rate": 1.7687539830961564e-06, + "loss": 0.0748, + "step": 15542 + }, + { + "epoch": 6.51, + "learning_rate": 1.7677913451552386e-06, + "loss": 0.0954, + "step": 15543 + }, + { + "epoch": 6.51, + "learning_rate": 1.766828943841853e-06, + "loss": 0.0713, + "step": 15544 + }, + { + "epoch": 6.51, + "learning_rate": 1.7658667791836626e-06, + "loss": 0.0869, + "step": 15545 + }, + { + "epoch": 6.51, + "learning_rate": 1.7649048512083256e-06, + "loss": 0.0974, + "step": 15546 + }, + { + "epoch": 6.51, + "learning_rate": 1.763943159943493e-06, + "loss": 0.0788, + "step": 15547 + }, + { + "epoch": 6.51, + "learning_rate": 1.762981705416804e-06, + "loss": 0.0978, + "step": 15548 + }, + { + "epoch": 6.51, + "learning_rate": 1.7620204876558967e-06, + "loss": 0.0823, + "step": 15549 + }, + { + "epoch": 6.51, + "learning_rate": 1.7610595066883996e-06, + "loss": 0.0808, + "step": 15550 + }, + { + "epoch": 6.51, + "learning_rate": 1.7600987625419353e-06, + "loss": 0.0848, + "step": 15551 + }, + { + "epoch": 6.51, + "learning_rate": 1.7591382552441206e-06, + "loss": 0.1244, + "step": 15552 + }, + { + "epoch": 6.51, + "learning_rate": 1.7581779848225665e-06, + "loss": 0.0953, + "step": 15553 + }, + { + "epoch": 6.51, + "learning_rate": 1.7572179513048703e-06, + "loss": 0.0913, + "step": 15554 + }, + { + "epoch": 6.51, + "learning_rate": 1.7562581547186298e-06, + "loss": 0.1055, + "step": 15555 + }, + { + "epoch": 6.51, + "learning_rate": 1.7552985950914326e-06, + "loss": 0.0888, + "step": 15556 + }, + { + "epoch": 6.51, + "learning_rate": 1.7543392724508613e-06, + "loss": 0.0913, + "step": 15557 + }, + { + "epoch": 6.51, + "learning_rate": 1.7533801868244904e-06, + "loss": 0.0821, + "step": 15558 + }, + { + "epoch": 6.51, + "learning_rate": 1.7524213382398902e-06, + "loss": 0.1092, + "step": 15559 + }, + { + "epoch": 6.51, + "learning_rate": 1.7514627267246187e-06, + "loss": 0.0841, + "step": 15560 + }, + { + "epoch": 6.51, + "learning_rate": 1.7505043523062326e-06, + "loss": 0.0623, + "step": 15561 + }, + { + "epoch": 6.52, + "learning_rate": 1.7495462150122767e-06, + "loss": 0.0702, + "step": 15562 + }, + { + "epoch": 6.52, + "learning_rate": 1.7485883148702931e-06, + "loss": 0.0709, + "step": 15563 + }, + { + "epoch": 6.52, + "learning_rate": 1.7476306519078168e-06, + "loss": 0.1001, + "step": 15564 + }, + { + "epoch": 6.52, + "learning_rate": 1.7466732261523745e-06, + "loss": 0.079, + "step": 15565 + }, + { + "epoch": 6.52, + "learning_rate": 1.7457160376314852e-06, + "loss": 0.0705, + "step": 15566 + }, + { + "epoch": 6.52, + "learning_rate": 1.7447590863726649e-06, + "loss": 0.0983, + "step": 15567 + }, + { + "epoch": 6.52, + "learning_rate": 1.7438023724034215e-06, + "loss": 0.0806, + "step": 15568 + }, + { + "epoch": 6.52, + "learning_rate": 1.7428458957512496e-06, + "loss": 0.0615, + "step": 15569 + }, + { + "epoch": 6.52, + "learning_rate": 1.741889656443645e-06, + "loss": 0.0832, + "step": 15570 + }, + { + "epoch": 6.52, + "learning_rate": 1.7409336545080947e-06, + "loss": 0.0716, + "step": 15571 + }, + { + "epoch": 6.52, + "learning_rate": 1.7399778899720764e-06, + "loss": 0.1212, + "step": 15572 + }, + { + "epoch": 6.52, + "learning_rate": 1.7390223628630643e-06, + "loss": 0.1216, + "step": 15573 + }, + { + "epoch": 6.52, + "learning_rate": 1.738067073208527e-06, + "loss": 0.0805, + "step": 15574 + }, + { + "epoch": 6.52, + "learning_rate": 1.737112021035917e-06, + "loss": 0.0785, + "step": 15575 + }, + { + "epoch": 6.52, + "learning_rate": 1.7361572063726894e-06, + "loss": 0.1026, + "step": 15576 + }, + { + "epoch": 6.52, + "learning_rate": 1.735202629246291e-06, + "loss": 0.0766, + "step": 15577 + }, + { + "epoch": 6.52, + "learning_rate": 1.7342482896841572e-06, + "loss": 0.0695, + "step": 15578 + }, + { + "epoch": 6.52, + "learning_rate": 1.7332941877137233e-06, + "loss": 0.0673, + "step": 15579 + }, + { + "epoch": 6.52, + "learning_rate": 1.7323403233624136e-06, + "loss": 0.0663, + "step": 15580 + }, + { + "epoch": 6.52, + "learning_rate": 1.731386696657642e-06, + "loss": 0.0911, + "step": 15581 + }, + { + "epoch": 6.52, + "learning_rate": 1.7304333076268253e-06, + "loss": 0.0878, + "step": 15582 + }, + { + "epoch": 6.52, + "learning_rate": 1.7294801562973619e-06, + "loss": 0.0663, + "step": 15583 + }, + { + "epoch": 6.52, + "learning_rate": 1.7285272426966526e-06, + "loss": 0.0778, + "step": 15584 + }, + { + "epoch": 6.53, + "learning_rate": 1.7275745668520883e-06, + "loss": 0.0794, + "step": 15585 + }, + { + "epoch": 6.53, + "learning_rate": 1.7266221287910512e-06, + "loss": 0.0696, + "step": 15586 + }, + { + "epoch": 6.53, + "learning_rate": 1.7256699285409206e-06, + "loss": 0.0906, + "step": 15587 + }, + { + "epoch": 6.53, + "learning_rate": 1.724717966129067e-06, + "loss": 0.0507, + "step": 15588 + }, + { + "epoch": 6.53, + "learning_rate": 1.7237662415828493e-06, + "loss": 0.0617, + "step": 15589 + }, + { + "epoch": 6.53, + "learning_rate": 1.7228147549296271e-06, + "loss": 0.0886, + "step": 15590 + }, + { + "epoch": 6.53, + "learning_rate": 1.7218635061967505e-06, + "loss": 0.0585, + "step": 15591 + }, + { + "epoch": 6.53, + "learning_rate": 1.7209124954115608e-06, + "loss": 0.086, + "step": 15592 + }, + { + "epoch": 6.53, + "learning_rate": 1.7199617226013943e-06, + "loss": 0.0943, + "step": 15593 + }, + { + "epoch": 6.53, + "learning_rate": 1.7190111877935833e-06, + "loss": 0.0768, + "step": 15594 + }, + { + "epoch": 6.53, + "learning_rate": 1.718060891015445e-06, + "loss": 0.0918, + "step": 15595 + }, + { + "epoch": 6.53, + "learning_rate": 1.7171108322942975e-06, + "loss": 0.0945, + "step": 15596 + }, + { + "epoch": 6.53, + "learning_rate": 1.716161011657449e-06, + "loss": 0.0979, + "step": 15597 + }, + { + "epoch": 6.53, + "learning_rate": 1.7152114291322008e-06, + "loss": 0.1001, + "step": 15598 + }, + { + "epoch": 6.53, + "learning_rate": 1.714262084745848e-06, + "loss": 0.0842, + "step": 15599 + }, + { + "epoch": 6.53, + "learning_rate": 1.7133129785256797e-06, + "loss": 0.0716, + "step": 15600 + }, + { + "epoch": 6.53, + "learning_rate": 1.7123641104989786e-06, + "loss": 0.0983, + "step": 15601 + }, + { + "epoch": 6.53, + "learning_rate": 1.7114154806930138e-06, + "loss": 0.0482, + "step": 15602 + }, + { + "epoch": 6.53, + "learning_rate": 1.7104670891350572e-06, + "loss": 0.0885, + "step": 15603 + }, + { + "epoch": 6.53, + "learning_rate": 1.7095189358523668e-06, + "loss": 0.0932, + "step": 15604 + }, + { + "epoch": 6.53, + "learning_rate": 1.7085710208722006e-06, + "loss": 0.073, + "step": 15605 + }, + { + "epoch": 6.53, + "learning_rate": 1.7076233442218004e-06, + "loss": 0.0746, + "step": 15606 + }, + { + "epoch": 6.53, + "learning_rate": 1.7066759059284087e-06, + "loss": 0.0675, + "step": 15607 + }, + { + "epoch": 6.53, + "learning_rate": 1.7057287060192585e-06, + "loss": 0.0975, + "step": 15608 + }, + { + "epoch": 6.54, + "learning_rate": 1.7047817445215786e-06, + "loss": 0.0812, + "step": 15609 + }, + { + "epoch": 6.54, + "learning_rate": 1.7038350214625844e-06, + "loss": 0.1039, + "step": 15610 + }, + { + "epoch": 6.54, + "learning_rate": 1.7028885368694903e-06, + "loss": 0.0927, + "step": 15611 + }, + { + "epoch": 6.54, + "learning_rate": 1.7019422907695026e-06, + "loss": 0.0954, + "step": 15612 + }, + { + "epoch": 6.54, + "learning_rate": 1.7009962831898208e-06, + "loss": 0.0851, + "step": 15613 + }, + { + "epoch": 6.54, + "learning_rate": 1.700050514157635e-06, + "loss": 0.0976, + "step": 15614 + }, + { + "epoch": 6.54, + "learning_rate": 1.6991049837001362e-06, + "loss": 0.1027, + "step": 15615 + }, + { + "epoch": 6.54, + "learning_rate": 1.6981596918444953e-06, + "loss": 0.103, + "step": 15616 + }, + { + "epoch": 6.54, + "learning_rate": 1.6972146386178868e-06, + "loss": 0.0748, + "step": 15617 + }, + { + "epoch": 6.54, + "learning_rate": 1.6962698240474762e-06, + "loss": 0.092, + "step": 15618 + }, + { + "epoch": 6.54, + "learning_rate": 1.6953252481604198e-06, + "loss": 0.0901, + "step": 15619 + }, + { + "epoch": 6.54, + "learning_rate": 1.6943809109838716e-06, + "loss": 0.1037, + "step": 15620 + }, + { + "epoch": 6.54, + "learning_rate": 1.6934368125449752e-06, + "loss": 0.1248, + "step": 15621 + }, + { + "epoch": 6.54, + "learning_rate": 1.6924929528708633e-06, + "loss": 0.072, + "step": 15622 + }, + { + "epoch": 6.54, + "learning_rate": 1.6915493319886712e-06, + "loss": 0.0888, + "step": 15623 + }, + { + "epoch": 6.54, + "learning_rate": 1.690605949925519e-06, + "loss": 0.0678, + "step": 15624 + }, + { + "epoch": 6.54, + "learning_rate": 1.6896628067085264e-06, + "loss": 0.0959, + "step": 15625 + }, + { + "epoch": 6.54, + "learning_rate": 1.6887199023648016e-06, + "loss": 0.0664, + "step": 15626 + }, + { + "epoch": 6.54, + "learning_rate": 1.6877772369214507e-06, + "loss": 0.0693, + "step": 15627 + }, + { + "epoch": 6.54, + "learning_rate": 1.6868348104055642e-06, + "loss": 0.0777, + "step": 15628 + }, + { + "epoch": 6.54, + "learning_rate": 1.6858926228442362e-06, + "loss": 0.1035, + "step": 15629 + }, + { + "epoch": 6.54, + "learning_rate": 1.6849506742645449e-06, + "loss": 0.0704, + "step": 15630 + }, + { + "epoch": 6.54, + "learning_rate": 1.6840089646935686e-06, + "loss": 0.1098, + "step": 15631 + }, + { + "epoch": 6.54, + "learning_rate": 1.6830674941583736e-06, + "loss": 0.0978, + "step": 15632 + }, + { + "epoch": 6.55, + "learning_rate": 1.6821262626860246e-06, + "loss": 0.094, + "step": 15633 + }, + { + "epoch": 6.55, + "learning_rate": 1.6811852703035737e-06, + "loss": 0.0567, + "step": 15634 + }, + { + "epoch": 6.55, + "learning_rate": 1.6802445170380733e-06, + "loss": 0.098, + "step": 15635 + }, + { + "epoch": 6.55, + "learning_rate": 1.6793040029165596e-06, + "loss": 0.0864, + "step": 15636 + }, + { + "epoch": 6.55, + "learning_rate": 1.678363727966068e-06, + "loss": 0.0934, + "step": 15637 + }, + { + "epoch": 6.55, + "learning_rate": 1.6774236922136266e-06, + "loss": 0.0623, + "step": 15638 + }, + { + "epoch": 6.55, + "learning_rate": 1.676483895686255e-06, + "loss": 0.0902, + "step": 15639 + }, + { + "epoch": 6.55, + "learning_rate": 1.6755443384109694e-06, + "loss": 0.0686, + "step": 15640 + }, + { + "epoch": 6.55, + "learning_rate": 1.6746050204147734e-06, + "loss": 0.0762, + "step": 15641 + }, + { + "epoch": 6.55, + "learning_rate": 1.6736659417246714e-06, + "loss": 0.0639, + "step": 15642 + }, + { + "epoch": 6.55, + "learning_rate": 1.6727271023676506e-06, + "loss": 0.0844, + "step": 15643 + }, + { + "epoch": 6.55, + "learning_rate": 1.6717885023707004e-06, + "loss": 0.0582, + "step": 15644 + }, + { + "epoch": 6.55, + "learning_rate": 1.670850141760798e-06, + "loss": 0.1034, + "step": 15645 + }, + { + "epoch": 6.55, + "learning_rate": 1.6699120205649177e-06, + "loss": 0.0958, + "step": 15646 + }, + { + "epoch": 6.55, + "learning_rate": 1.6689741388100244e-06, + "loss": 0.086, + "step": 15647 + }, + { + "epoch": 6.55, + "learning_rate": 1.66803649652308e-06, + "loss": 0.08, + "step": 15648 + }, + { + "epoch": 6.55, + "learning_rate": 1.6670990937310293e-06, + "loss": 0.098, + "step": 15649 + }, + { + "epoch": 6.55, + "learning_rate": 1.6661619304608234e-06, + "loss": 0.0737, + "step": 15650 + }, + { + "epoch": 6.55, + "learning_rate": 1.665225006739396e-06, + "loss": 0.0956, + "step": 15651 + }, + { + "epoch": 6.55, + "learning_rate": 1.6642883225936779e-06, + "loss": 0.0917, + "step": 15652 + }, + { + "epoch": 6.55, + "learning_rate": 1.6633518780505964e-06, + "loss": 0.098, + "step": 15653 + }, + { + "epoch": 6.55, + "learning_rate": 1.6624156731370666e-06, + "loss": 0.0646, + "step": 15654 + }, + { + "epoch": 6.55, + "learning_rate": 1.6614797078799993e-06, + "loss": 0.0836, + "step": 15655 + }, + { + "epoch": 6.55, + "learning_rate": 1.6605439823063018e-06, + "loss": 0.0809, + "step": 15656 + }, + { + "epoch": 6.56, + "learning_rate": 1.659608496442865e-06, + "loss": 0.0779, + "step": 15657 + }, + { + "epoch": 6.56, + "learning_rate": 1.6586732503165793e-06, + "loss": 0.0667, + "step": 15658 + }, + { + "epoch": 6.56, + "learning_rate": 1.65773824395433e-06, + "loss": 0.1243, + "step": 15659 + }, + { + "epoch": 6.56, + "learning_rate": 1.6568034773829923e-06, + "loss": 0.0629, + "step": 15660 + }, + { + "epoch": 6.56, + "learning_rate": 1.6558689506294345e-06, + "loss": 0.1078, + "step": 15661 + }, + { + "epoch": 6.56, + "learning_rate": 1.654934663720522e-06, + "loss": 0.0776, + "step": 15662 + }, + { + "epoch": 6.56, + "learning_rate": 1.6540006166831046e-06, + "loss": 0.104, + "step": 15663 + }, + { + "epoch": 6.56, + "learning_rate": 1.6530668095440339e-06, + "loss": 0.0857, + "step": 15664 + }, + { + "epoch": 6.56, + "learning_rate": 1.6521332423301507e-06, + "loss": 0.091, + "step": 15665 + }, + { + "epoch": 6.56, + "learning_rate": 1.6511999150682888e-06, + "loss": 0.0714, + "step": 15666 + }, + { + "epoch": 6.56, + "learning_rate": 1.6502668277852774e-06, + "loss": 0.0698, + "step": 15667 + }, + { + "epoch": 6.56, + "learning_rate": 1.6493339805079367e-06, + "loss": 0.0865, + "step": 15668 + }, + { + "epoch": 6.56, + "learning_rate": 1.6484013732630833e-06, + "loss": 0.0685, + "step": 15669 + }, + { + "epoch": 6.56, + "learning_rate": 1.6474690060775178e-06, + "loss": 0.0712, + "step": 15670 + }, + { + "epoch": 6.56, + "learning_rate": 1.6465368789780446e-06, + "loss": 0.049, + "step": 15671 + }, + { + "epoch": 6.56, + "learning_rate": 1.6456049919914563e-06, + "loss": 0.0795, + "step": 15672 + }, + { + "epoch": 6.56, + "learning_rate": 1.6446733451445408e-06, + "loss": 0.0751, + "step": 15673 + }, + { + "epoch": 6.56, + "learning_rate": 1.643741938464073e-06, + "loss": 0.0589, + "step": 15674 + }, + { + "epoch": 6.56, + "learning_rate": 1.6428107719768293e-06, + "loss": 0.0733, + "step": 15675 + }, + { + "epoch": 6.56, + "learning_rate": 1.6418798457095753e-06, + "loss": 0.0658, + "step": 15676 + }, + { + "epoch": 6.56, + "learning_rate": 1.640949159689066e-06, + "loss": 0.0452, + "step": 15677 + }, + { + "epoch": 6.56, + "learning_rate": 1.6400187139420554e-06, + "loss": 0.0795, + "step": 15678 + }, + { + "epoch": 6.56, + "learning_rate": 1.6390885084952878e-06, + "loss": 0.1011, + "step": 15679 + }, + { + "epoch": 6.56, + "learning_rate": 1.638158543375502e-06, + "loss": 0.0588, + "step": 15680 + }, + { + "epoch": 6.57, + "learning_rate": 1.6372288186094287e-06, + "loss": 0.0611, + "step": 15681 + }, + { + "epoch": 6.57, + "learning_rate": 1.6362993342237931e-06, + "loss": 0.0871, + "step": 15682 + }, + { + "epoch": 6.57, + "learning_rate": 1.635370090245313e-06, + "loss": 0.1033, + "step": 15683 + }, + { + "epoch": 6.57, + "learning_rate": 1.6344410867006944e-06, + "loss": 0.0919, + "step": 15684 + }, + { + "epoch": 6.57, + "learning_rate": 1.6335123236166428e-06, + "loss": 0.0885, + "step": 15685 + }, + { + "epoch": 6.57, + "learning_rate": 1.6325838010198558e-06, + "loss": 0.0697, + "step": 15686 + }, + { + "epoch": 6.57, + "learning_rate": 1.631655518937022e-06, + "loss": 0.0664, + "step": 15687 + }, + { + "epoch": 6.57, + "learning_rate": 1.6307274773948244e-06, + "loss": 0.0457, + "step": 15688 + }, + { + "epoch": 6.57, + "learning_rate": 1.629799676419942e-06, + "loss": 0.0925, + "step": 15689 + }, + { + "epoch": 6.57, + "learning_rate": 1.6288721160390375e-06, + "loss": 0.0883, + "step": 15690 + }, + { + "epoch": 6.57, + "learning_rate": 1.6279447962787753e-06, + "loss": 0.0826, + "step": 15691 + }, + { + "epoch": 6.57, + "learning_rate": 1.6270177171658108e-06, + "loss": 0.0764, + "step": 15692 + }, + { + "epoch": 6.57, + "learning_rate": 1.6260908787267915e-06, + "loss": 0.0912, + "step": 15693 + }, + { + "epoch": 6.57, + "learning_rate": 1.6251642809883595e-06, + "loss": 0.0812, + "step": 15694 + }, + { + "epoch": 6.57, + "learning_rate": 1.624237923977151e-06, + "loss": 0.0999, + "step": 15695 + }, + { + "epoch": 6.57, + "learning_rate": 1.6233118077197884e-06, + "loss": 0.0605, + "step": 15696 + }, + { + "epoch": 6.57, + "learning_rate": 1.622385932242897e-06, + "loss": 0.0817, + "step": 15697 + }, + { + "epoch": 6.57, + "learning_rate": 1.6214602975730853e-06, + "loss": 0.0605, + "step": 15698 + }, + { + "epoch": 6.57, + "learning_rate": 1.6205349037369621e-06, + "loss": 0.101, + "step": 15699 + }, + { + "epoch": 6.57, + "learning_rate": 1.6196097507611274e-06, + "loss": 0.0774, + "step": 15700 + }, + { + "epoch": 6.57, + "learning_rate": 1.6186848386721743e-06, + "loss": 0.1025, + "step": 15701 + }, + { + "epoch": 6.57, + "learning_rate": 1.617760167496688e-06, + "loss": 0.0637, + "step": 15702 + }, + { + "epoch": 6.57, + "learning_rate": 1.6168357372612498e-06, + "loss": 0.046, + "step": 15703 + }, + { + "epoch": 6.57, + "learning_rate": 1.6159115479924259e-06, + "loss": 0.0925, + "step": 15704 + }, + { + "epoch": 6.58, + "learning_rate": 1.6149875997167851e-06, + "loss": 0.1007, + "step": 15705 + }, + { + "epoch": 6.58, + "learning_rate": 1.6140638924608853e-06, + "loss": 0.0874, + "step": 15706 + }, + { + "epoch": 6.58, + "learning_rate": 1.6131404262512774e-06, + "loss": 0.0554, + "step": 15707 + }, + { + "epoch": 6.58, + "learning_rate": 1.6122172011145053e-06, + "loss": 0.0789, + "step": 15708 + }, + { + "epoch": 6.58, + "learning_rate": 1.6112942170771073e-06, + "loss": 0.0864, + "step": 15709 + }, + { + "epoch": 6.58, + "learning_rate": 1.6103714741656152e-06, + "loss": 0.0571, + "step": 15710 + }, + { + "epoch": 6.58, + "learning_rate": 1.6094489724065476e-06, + "loss": 0.1119, + "step": 15711 + }, + { + "epoch": 6.58, + "learning_rate": 1.6085267118264235e-06, + "loss": 0.0649, + "step": 15712 + }, + { + "epoch": 6.58, + "learning_rate": 1.6076046924517518e-06, + "loss": 0.0511, + "step": 15713 + }, + { + "epoch": 6.58, + "learning_rate": 1.606682914309038e-06, + "loss": 0.0763, + "step": 15714 + }, + { + "epoch": 6.58, + "learning_rate": 1.6057613774247738e-06, + "loss": 0.0917, + "step": 15715 + }, + { + "epoch": 6.58, + "learning_rate": 1.6048400818254528e-06, + "loss": 0.0586, + "step": 15716 + }, + { + "epoch": 6.58, + "learning_rate": 1.6039190275375526e-06, + "loss": 0.0785, + "step": 15717 + }, + { + "epoch": 6.58, + "learning_rate": 1.60299821458755e-06, + "loss": 0.0894, + "step": 15718 + }, + { + "epoch": 6.58, + "learning_rate": 1.6020776430019113e-06, + "loss": 0.0793, + "step": 15719 + }, + { + "epoch": 6.58, + "learning_rate": 1.6011573128070978e-06, + "loss": 0.0858, + "step": 15720 + }, + { + "epoch": 6.58, + "learning_rate": 1.600237224029565e-06, + "loss": 0.0924, + "step": 15721 + }, + { + "epoch": 6.58, + "learning_rate": 1.5993173766957592e-06, + "loss": 0.0849, + "step": 15722 + }, + { + "epoch": 6.58, + "learning_rate": 1.5983977708321207e-06, + "loss": 0.0839, + "step": 15723 + }, + { + "epoch": 6.58, + "learning_rate": 1.597478406465085e-06, + "loss": 0.0739, + "step": 15724 + }, + { + "epoch": 6.58, + "learning_rate": 1.596559283621074e-06, + "loss": 0.0767, + "step": 15725 + }, + { + "epoch": 6.58, + "learning_rate": 1.5956404023265092e-06, + "loss": 0.0593, + "step": 15726 + }, + { + "epoch": 6.58, + "learning_rate": 1.5947217626078039e-06, + "loss": 0.0842, + "step": 15727 + }, + { + "epoch": 6.58, + "learning_rate": 1.593803364491362e-06, + "loss": 0.0605, + "step": 15728 + }, + { + "epoch": 6.59, + "learning_rate": 1.592885208003583e-06, + "loss": 0.0713, + "step": 15729 + }, + { + "epoch": 6.59, + "learning_rate": 1.5919672931708607e-06, + "loss": 0.1021, + "step": 15730 + }, + { + "epoch": 6.59, + "learning_rate": 1.5910496200195757e-06, + "loss": 0.0756, + "step": 15731 + }, + { + "epoch": 6.59, + "learning_rate": 1.5901321885761068e-06, + "loss": 0.0888, + "step": 15732 + }, + { + "epoch": 6.59, + "learning_rate": 1.5892149988668249e-06, + "loss": 0.0864, + "step": 15733 + }, + { + "epoch": 6.59, + "learning_rate": 1.5882980509180956e-06, + "loss": 0.083, + "step": 15734 + }, + { + "epoch": 6.59, + "learning_rate": 1.5873813447562735e-06, + "loss": 0.0715, + "step": 15735 + }, + { + "epoch": 6.59, + "learning_rate": 1.5864648804077122e-06, + "loss": 0.053, + "step": 15736 + }, + { + "epoch": 6.59, + "learning_rate": 1.5855486578987489e-06, + "loss": 0.0777, + "step": 15737 + }, + { + "epoch": 6.59, + "learning_rate": 1.584632677255723e-06, + "loss": 0.0909, + "step": 15738 + }, + { + "epoch": 6.59, + "learning_rate": 1.5837169385049623e-06, + "loss": 0.0655, + "step": 15739 + }, + { + "epoch": 6.59, + "learning_rate": 1.582801441672791e-06, + "loss": 0.0622, + "step": 15740 + }, + { + "epoch": 6.59, + "learning_rate": 1.5818861867855239e-06, + "loss": 0.0597, + "step": 15741 + }, + { + "epoch": 6.59, + "learning_rate": 1.5809711738694665e-06, + "loss": 0.0651, + "step": 15742 + }, + { + "epoch": 6.59, + "learning_rate": 1.5800564029509225e-06, + "loss": 0.0885, + "step": 15743 + }, + { + "epoch": 6.59, + "learning_rate": 1.579141874056187e-06, + "loss": 0.0718, + "step": 15744 + }, + { + "epoch": 6.59, + "learning_rate": 1.578227587211545e-06, + "loss": 0.0479, + "step": 15745 + }, + { + "epoch": 6.59, + "learning_rate": 1.5773135424432762e-06, + "loss": 0.0664, + "step": 15746 + }, + { + "epoch": 6.59, + "learning_rate": 1.5763997397776553e-06, + "loss": 0.0864, + "step": 15747 + }, + { + "epoch": 6.59, + "learning_rate": 1.5754861792409493e-06, + "loss": 0.081, + "step": 15748 + }, + { + "epoch": 6.59, + "learning_rate": 1.574572860859417e-06, + "loss": 0.0448, + "step": 15749 + }, + { + "epoch": 6.59, + "learning_rate": 1.5736597846593115e-06, + "loss": 0.0616, + "step": 15750 + }, + { + "epoch": 6.59, + "learning_rate": 1.5727469506668814e-06, + "loss": 0.0791, + "step": 15751 + }, + { + "epoch": 6.59, + "learning_rate": 1.5718343589083584e-06, + "loss": 0.0777, + "step": 15752 + }, + { + "epoch": 6.6, + "learning_rate": 1.5709220094099775e-06, + "loss": 0.087, + "step": 15753 + }, + { + "epoch": 6.6, + "learning_rate": 1.5700099021979642e-06, + "loss": 0.0565, + "step": 15754 + }, + { + "epoch": 6.6, + "learning_rate": 1.5690980372985354e-06, + "loss": 0.0675, + "step": 15755 + }, + { + "epoch": 6.6, + "learning_rate": 1.5681864147379022e-06, + "loss": 0.0759, + "step": 15756 + }, + { + "epoch": 6.6, + "learning_rate": 1.5672750345422716e-06, + "loss": 0.0632, + "step": 15757 + }, + { + "epoch": 6.6, + "learning_rate": 1.5663638967378337e-06, + "loss": 0.0824, + "step": 15758 + }, + { + "epoch": 6.6, + "learning_rate": 1.565453001350782e-06, + "loss": 0.0609, + "step": 15759 + }, + { + "epoch": 6.6, + "learning_rate": 1.5645423484072987e-06, + "loss": 0.0871, + "step": 15760 + }, + { + "epoch": 6.6, + "learning_rate": 1.5636319379335596e-06, + "loss": 0.103, + "step": 15761 + }, + { + "epoch": 6.6, + "learning_rate": 1.5627217699557352e-06, + "loss": 0.0919, + "step": 15762 + }, + { + "epoch": 6.6, + "learning_rate": 1.5618118444999885e-06, + "loss": 0.0721, + "step": 15763 + }, + { + "epoch": 6.6, + "learning_rate": 1.5609021615924702e-06, + "loss": 0.0606, + "step": 15764 + }, + { + "epoch": 6.6, + "learning_rate": 1.5599927212593325e-06, + "loss": 0.0896, + "step": 15765 + }, + { + "epoch": 6.6, + "learning_rate": 1.559083523526712e-06, + "loss": 0.1032, + "step": 15766 + }, + { + "epoch": 6.6, + "learning_rate": 1.558174568420746e-06, + "loss": 0.1055, + "step": 15767 + }, + { + "epoch": 6.6, + "learning_rate": 1.5572658559675602e-06, + "loss": 0.0868, + "step": 15768 + }, + { + "epoch": 6.6, + "learning_rate": 1.5563573861932756e-06, + "loss": 0.075, + "step": 15769 + }, + { + "epoch": 6.6, + "learning_rate": 1.555449159124005e-06, + "loss": 0.1048, + "step": 15770 + }, + { + "epoch": 6.6, + "learning_rate": 1.5545411747858574e-06, + "loss": 0.0805, + "step": 15771 + }, + { + "epoch": 6.6, + "learning_rate": 1.5536334332049274e-06, + "loss": 0.0666, + "step": 15772 + }, + { + "epoch": 6.6, + "learning_rate": 1.5527259344073087e-06, + "loss": 0.0646, + "step": 15773 + }, + { + "epoch": 6.6, + "learning_rate": 1.551818678419087e-06, + "loss": 0.1058, + "step": 15774 + }, + { + "epoch": 6.6, + "learning_rate": 1.5509116652663403e-06, + "loss": 0.0872, + "step": 15775 + }, + { + "epoch": 6.6, + "learning_rate": 1.5500048949751411e-06, + "loss": 0.0849, + "step": 15776 + }, + { + "epoch": 6.61, + "learning_rate": 1.5490983675715544e-06, + "loss": 0.0757, + "step": 15777 + }, + { + "epoch": 6.61, + "learning_rate": 1.5481920830816343e-06, + "loss": 0.107, + "step": 15778 + }, + { + "epoch": 6.61, + "learning_rate": 1.5472860415314317e-06, + "loss": 0.0731, + "step": 15779 + }, + { + "epoch": 6.61, + "learning_rate": 1.546380242946992e-06, + "loss": 0.1056, + "step": 15780 + }, + { + "epoch": 6.61, + "learning_rate": 1.5454746873543503e-06, + "loss": 0.078, + "step": 15781 + }, + { + "epoch": 6.61, + "learning_rate": 1.5445693747795366e-06, + "loss": 0.0975, + "step": 15782 + }, + { + "epoch": 6.61, + "learning_rate": 1.5436643052485723e-06, + "loss": 0.0566, + "step": 15783 + }, + { + "epoch": 6.61, + "learning_rate": 1.5427594787874766e-06, + "loss": 0.0795, + "step": 15784 + }, + { + "epoch": 6.61, + "learning_rate": 1.5418548954222512e-06, + "loss": 0.1206, + "step": 15785 + }, + { + "epoch": 6.61, + "learning_rate": 1.540950555178905e-06, + "loss": 0.0608, + "step": 15786 + }, + { + "epoch": 6.61, + "learning_rate": 1.5400464580834262e-06, + "loss": 0.0807, + "step": 15787 + }, + { + "epoch": 6.61, + "learning_rate": 1.539142604161804e-06, + "loss": 0.066, + "step": 15788 + }, + { + "epoch": 6.61, + "learning_rate": 1.5382389934400199e-06, + "loss": 0.0738, + "step": 15789 + }, + { + "epoch": 6.61, + "learning_rate": 1.5373356259440475e-06, + "loss": 0.0766, + "step": 15790 + }, + { + "epoch": 6.61, + "learning_rate": 1.5364325016998528e-06, + "loss": 0.0858, + "step": 15791 + }, + { + "epoch": 6.61, + "learning_rate": 1.535529620733397e-06, + "loss": 0.0799, + "step": 15792 + }, + { + "epoch": 6.61, + "learning_rate": 1.5346269830706295e-06, + "loss": 0.0814, + "step": 15793 + }, + { + "epoch": 6.61, + "learning_rate": 1.5337245887374962e-06, + "loss": 0.0911, + "step": 15794 + }, + { + "epoch": 6.61, + "learning_rate": 1.5328224377599376e-06, + "loss": 0.0756, + "step": 15795 + }, + { + "epoch": 6.61, + "learning_rate": 1.531920530163885e-06, + "loss": 0.1003, + "step": 15796 + }, + { + "epoch": 6.61, + "learning_rate": 1.5310188659752612e-06, + "loss": 0.0924, + "step": 15797 + }, + { + "epoch": 6.61, + "learning_rate": 1.5301174452199874e-06, + "loss": 0.0671, + "step": 15798 + }, + { + "epoch": 6.61, + "learning_rate": 1.5292162679239697e-06, + "loss": 0.0975, + "step": 15799 + }, + { + "epoch": 6.62, + "learning_rate": 1.528315334113114e-06, + "loss": 0.0744, + "step": 15800 + }, + { + "epoch": 6.62, + "learning_rate": 1.5274146438133154e-06, + "loss": 0.0805, + "step": 15801 + }, + { + "epoch": 6.62, + "learning_rate": 1.5265141970504648e-06, + "loss": 0.0738, + "step": 15802 + }, + { + "epoch": 6.62, + "learning_rate": 1.5256139938504445e-06, + "loss": 0.1022, + "step": 15803 + }, + { + "epoch": 6.62, + "learning_rate": 1.524714034239132e-06, + "loss": 0.0975, + "step": 15804 + }, + { + "epoch": 6.62, + "learning_rate": 1.5238143182423914e-06, + "loss": 0.0859, + "step": 15805 + }, + { + "epoch": 6.62, + "learning_rate": 1.5229148458860865e-06, + "loss": 0.0904, + "step": 15806 + }, + { + "epoch": 6.62, + "learning_rate": 1.522015617196072e-06, + "loss": 0.0979, + "step": 15807 + }, + { + "epoch": 6.62, + "learning_rate": 1.5211166321981963e-06, + "loss": 0.0959, + "step": 15808 + }, + { + "epoch": 6.62, + "learning_rate": 1.5202178909182997e-06, + "loss": 0.0619, + "step": 15809 + }, + { + "epoch": 6.62, + "learning_rate": 1.5193193933822136e-06, + "loss": 0.0899, + "step": 15810 + }, + { + "epoch": 6.62, + "learning_rate": 1.5184211396157655e-06, + "loss": 0.073, + "step": 15811 + }, + { + "epoch": 6.62, + "learning_rate": 1.5175231296447778e-06, + "loss": 0.0746, + "step": 15812 + }, + { + "epoch": 6.62, + "learning_rate": 1.516625363495059e-06, + "loss": 0.0975, + "step": 15813 + }, + { + "epoch": 6.62, + "learning_rate": 1.5157278411924148e-06, + "loss": 0.0644, + "step": 15814 + }, + { + "epoch": 6.62, + "learning_rate": 1.5148305627626458e-06, + "loss": 0.0867, + "step": 15815 + }, + { + "epoch": 6.62, + "learning_rate": 1.5139335282315427e-06, + "loss": 0.089, + "step": 15816 + }, + { + "epoch": 6.62, + "learning_rate": 1.51303673762489e-06, + "loss": 0.0824, + "step": 15817 + }, + { + "epoch": 6.62, + "learning_rate": 1.5121401909684663e-06, + "loss": 0.0783, + "step": 15818 + }, + { + "epoch": 6.62, + "learning_rate": 1.51124388828804e-06, + "loss": 0.0422, + "step": 15819 + }, + { + "epoch": 6.62, + "learning_rate": 1.5103478296093743e-06, + "loss": 0.0688, + "step": 15820 + }, + { + "epoch": 6.62, + "learning_rate": 1.509452014958227e-06, + "loss": 0.0618, + "step": 15821 + }, + { + "epoch": 6.62, + "learning_rate": 1.5085564443603473e-06, + "loss": 0.0895, + "step": 15822 + }, + { + "epoch": 6.62, + "learning_rate": 1.507661117841477e-06, + "loss": 0.0749, + "step": 15823 + }, + { + "epoch": 6.63, + "learning_rate": 1.5067660354273527e-06, + "loss": 0.0942, + "step": 15824 + }, + { + "epoch": 6.63, + "learning_rate": 1.5058711971437035e-06, + "loss": 0.0595, + "step": 15825 + }, + { + "epoch": 6.63, + "learning_rate": 1.504976603016247e-06, + "loss": 0.1109, + "step": 15826 + }, + { + "epoch": 6.63, + "learning_rate": 1.5040822530707e-06, + "loss": 0.0918, + "step": 15827 + }, + { + "epoch": 6.63, + "learning_rate": 1.5031881473327691e-06, + "loss": 0.0847, + "step": 15828 + }, + { + "epoch": 6.63, + "learning_rate": 1.5022942858281553e-06, + "loss": 0.0581, + "step": 15829 + }, + { + "epoch": 6.63, + "learning_rate": 1.5014006685825512e-06, + "loss": 0.0948, + "step": 15830 + }, + { + "epoch": 6.63, + "learning_rate": 1.500507295621646e-06, + "loss": 0.0823, + "step": 15831 + }, + { + "epoch": 6.63, + "learning_rate": 1.499614166971114e-06, + "loss": 0.0959, + "step": 15832 + }, + { + "epoch": 6.63, + "learning_rate": 1.4987212826566323e-06, + "loss": 0.0836, + "step": 15833 + }, + { + "epoch": 6.63, + "learning_rate": 1.4978286427038602e-06, + "loss": 0.0649, + "step": 15834 + }, + { + "epoch": 6.63, + "learning_rate": 1.4969362471384597e-06, + "loss": 0.1003, + "step": 15835 + }, + { + "epoch": 6.63, + "learning_rate": 1.4960440959860812e-06, + "loss": 0.0916, + "step": 15836 + }, + { + "epoch": 6.63, + "learning_rate": 1.4951521892723685e-06, + "loss": 0.0738, + "step": 15837 + }, + { + "epoch": 6.63, + "learning_rate": 1.4942605270229594e-06, + "loss": 0.0764, + "step": 15838 + }, + { + "epoch": 6.63, + "learning_rate": 1.4933691092634862e-06, + "loss": 0.0718, + "step": 15839 + }, + { + "epoch": 6.63, + "learning_rate": 1.4924779360195662e-06, + "loss": 0.0773, + "step": 15840 + }, + { + "epoch": 6.63, + "learning_rate": 1.4915870073168194e-06, + "loss": 0.0804, + "step": 15841 + }, + { + "epoch": 6.63, + "learning_rate": 1.4906963231808524e-06, + "loss": 0.0988, + "step": 15842 + }, + { + "epoch": 6.63, + "learning_rate": 1.4898058836372697e-06, + "loss": 0.0759, + "step": 15843 + }, + { + "epoch": 6.63, + "learning_rate": 1.4889156887116652e-06, + "loss": 0.0621, + "step": 15844 + }, + { + "epoch": 6.63, + "learning_rate": 1.4880257384296282e-06, + "loss": 0.0546, + "step": 15845 + }, + { + "epoch": 6.63, + "learning_rate": 1.487136032816736e-06, + "loss": 0.0796, + "step": 15846 + }, + { + "epoch": 6.63, + "learning_rate": 1.4862465718985653e-06, + "loss": 0.0723, + "step": 15847 + }, + { + "epoch": 6.64, + "learning_rate": 1.4853573557006807e-06, + "loss": 0.0789, + "step": 15848 + }, + { + "epoch": 6.64, + "learning_rate": 1.484468384248643e-06, + "loss": 0.0721, + "step": 15849 + }, + { + "epoch": 6.64, + "learning_rate": 1.4835796575680062e-06, + "loss": 0.0997, + "step": 15850 + }, + { + "epoch": 6.64, + "learning_rate": 1.4826911756843144e-06, + "loss": 0.095, + "step": 15851 + }, + { + "epoch": 6.64, + "learning_rate": 1.4818029386231082e-06, + "loss": 0.0576, + "step": 15852 + }, + { + "epoch": 6.64, + "learning_rate": 1.480914946409916e-06, + "loss": 0.0996, + "step": 15853 + }, + { + "epoch": 6.64, + "learning_rate": 1.4800271990702664e-06, + "loss": 0.0596, + "step": 15854 + }, + { + "epoch": 6.64, + "learning_rate": 1.4791396966296711e-06, + "loss": 0.0884, + "step": 15855 + }, + { + "epoch": 6.64, + "learning_rate": 1.4782524391136443e-06, + "loss": 0.0684, + "step": 15856 + }, + { + "epoch": 6.64, + "learning_rate": 1.47736542654769e-06, + "loss": 0.0636, + "step": 15857 + }, + { + "epoch": 6.64, + "learning_rate": 1.4764786589573033e-06, + "loss": 0.0788, + "step": 15858 + }, + { + "epoch": 6.64, + "learning_rate": 1.4755921363679758e-06, + "loss": 0.0973, + "step": 15859 + }, + { + "epoch": 6.64, + "learning_rate": 1.4747058588051854e-06, + "loss": 0.0763, + "step": 15860 + }, + { + "epoch": 6.64, + "learning_rate": 1.4738198262944103e-06, + "loss": 0.0712, + "step": 15861 + }, + { + "epoch": 6.64, + "learning_rate": 1.472934038861119e-06, + "loss": 0.0674, + "step": 15862 + }, + { + "epoch": 6.64, + "learning_rate": 1.4720484965307714e-06, + "loss": 0.1018, + "step": 15863 + }, + { + "epoch": 6.64, + "learning_rate": 1.4711631993288222e-06, + "loss": 0.1098, + "step": 15864 + }, + { + "epoch": 6.64, + "learning_rate": 1.4702781472807181e-06, + "loss": 0.0737, + "step": 15865 + }, + { + "epoch": 6.64, + "learning_rate": 1.469393340411902e-06, + "loss": 0.0964, + "step": 15866 + }, + { + "epoch": 6.64, + "learning_rate": 1.4685087787478025e-06, + "loss": 0.1032, + "step": 15867 + }, + { + "epoch": 6.64, + "learning_rate": 1.4676244623138468e-06, + "loss": 0.0856, + "step": 15868 + }, + { + "epoch": 6.64, + "learning_rate": 1.466740391135455e-06, + "loss": 0.1217, + "step": 15869 + }, + { + "epoch": 6.64, + "learning_rate": 1.4658565652380373e-06, + "loss": 0.0789, + "step": 15870 + }, + { + "epoch": 6.64, + "learning_rate": 1.4649729846470006e-06, + "loss": 0.073, + "step": 15871 + }, + { + "epoch": 6.65, + "learning_rate": 1.4640896493877443e-06, + "loss": 0.0972, + "step": 15872 + }, + { + "epoch": 6.65, + "learning_rate": 1.4632065594856527e-06, + "loss": 0.0651, + "step": 15873 + }, + { + "epoch": 6.65, + "learning_rate": 1.462323714966114e-06, + "loss": 0.1199, + "step": 15874 + }, + { + "epoch": 6.65, + "learning_rate": 1.4614411158545038e-06, + "loss": 0.1062, + "step": 15875 + }, + { + "epoch": 6.65, + "learning_rate": 1.460558762176194e-06, + "loss": 0.0922, + "step": 15876 + }, + { + "epoch": 6.65, + "learning_rate": 1.4596766539565432e-06, + "loss": 0.0987, + "step": 15877 + }, + { + "epoch": 6.65, + "learning_rate": 1.4587947912209076e-06, + "loss": 0.0748, + "step": 15878 + }, + { + "epoch": 6.65, + "learning_rate": 1.4579131739946361e-06, + "loss": 0.0895, + "step": 15879 + }, + { + "epoch": 6.65, + "learning_rate": 1.4570318023030738e-06, + "loss": 0.0804, + "step": 15880 + }, + { + "epoch": 6.65, + "learning_rate": 1.4561506761715482e-06, + "loss": 0.0968, + "step": 15881 + }, + { + "epoch": 6.65, + "learning_rate": 1.4552697956253903e-06, + "loss": 0.0778, + "step": 15882 + }, + { + "epoch": 6.65, + "learning_rate": 1.4543891606899196e-06, + "loss": 0.0658, + "step": 15883 + }, + { + "epoch": 6.65, + "learning_rate": 1.4535087713904484e-06, + "loss": 0.1075, + "step": 15884 + }, + { + "epoch": 6.65, + "learning_rate": 1.452628627752285e-06, + "loss": 0.0736, + "step": 15885 + }, + { + "epoch": 6.65, + "learning_rate": 1.4517487298007283e-06, + "loss": 0.086, + "step": 15886 + }, + { + "epoch": 6.65, + "learning_rate": 1.4508690775610656e-06, + "loss": 0.0852, + "step": 15887 + }, + { + "epoch": 6.65, + "learning_rate": 1.4499896710585859e-06, + "loss": 0.0803, + "step": 15888 + }, + { + "epoch": 6.65, + "learning_rate": 1.4491105103185666e-06, + "loss": 0.095, + "step": 15889 + }, + { + "epoch": 6.65, + "learning_rate": 1.4482315953662762e-06, + "loss": 0.1004, + "step": 15890 + }, + { + "epoch": 6.65, + "learning_rate": 1.4473529262269814e-06, + "loss": 0.0824, + "step": 15891 + }, + { + "epoch": 6.65, + "learning_rate": 1.4464745029259365e-06, + "loss": 0.0721, + "step": 15892 + }, + { + "epoch": 6.65, + "learning_rate": 1.4455963254883942e-06, + "loss": 0.0821, + "step": 15893 + }, + { + "epoch": 6.65, + "learning_rate": 1.4447183939395925e-06, + "loss": 0.0898, + "step": 15894 + }, + { + "epoch": 6.65, + "learning_rate": 1.4438407083047679e-06, + "loss": 0.092, + "step": 15895 + }, + { + "epoch": 6.66, + "learning_rate": 1.4429632686091498e-06, + "loss": 0.0992, + "step": 15896 + }, + { + "epoch": 6.66, + "learning_rate": 1.4420860748779596e-06, + "loss": 0.0756, + "step": 15897 + }, + { + "epoch": 6.66, + "learning_rate": 1.4412091271364105e-06, + "loss": 0.0674, + "step": 15898 + }, + { + "epoch": 6.66, + "learning_rate": 1.4403324254097128e-06, + "loss": 0.0817, + "step": 15899 + }, + { + "epoch": 6.66, + "learning_rate": 1.4394559697230637e-06, + "loss": 0.0912, + "step": 15900 + }, + { + "epoch": 6.66, + "learning_rate": 1.4385797601016538e-06, + "loss": 0.0836, + "step": 15901 + }, + { + "epoch": 6.66, + "learning_rate": 1.437703796570672e-06, + "loss": 0.0662, + "step": 15902 + }, + { + "epoch": 6.66, + "learning_rate": 1.4368280791552957e-06, + "loss": 0.1097, + "step": 15903 + }, + { + "epoch": 6.66, + "learning_rate": 1.4359526078806985e-06, + "loss": 0.0821, + "step": 15904 + }, + { + "epoch": 6.66, + "learning_rate": 1.4350773827720432e-06, + "loss": 0.0611, + "step": 15905 + }, + { + "epoch": 6.66, + "learning_rate": 1.4342024038544878e-06, + "loss": 0.1006, + "step": 15906 + }, + { + "epoch": 6.66, + "learning_rate": 1.4333276711531863e-06, + "loss": 0.0588, + "step": 15907 + }, + { + "epoch": 6.66, + "learning_rate": 1.4324531846932766e-06, + "loss": 0.0845, + "step": 15908 + }, + { + "epoch": 6.66, + "learning_rate": 1.4315789444998973e-06, + "loss": 0.0851, + "step": 15909 + }, + { + "epoch": 6.66, + "learning_rate": 1.4307049505981773e-06, + "loss": 0.0803, + "step": 15910 + }, + { + "epoch": 6.66, + "learning_rate": 1.4298312030132388e-06, + "loss": 0.0909, + "step": 15911 + }, + { + "epoch": 6.66, + "learning_rate": 1.4289577017701984e-06, + "loss": 0.1119, + "step": 15912 + }, + { + "epoch": 6.66, + "learning_rate": 1.4280844468941647e-06, + "loss": 0.0861, + "step": 15913 + }, + { + "epoch": 6.66, + "learning_rate": 1.4272114384102353e-06, + "loss": 0.1005, + "step": 15914 + }, + { + "epoch": 6.66, + "learning_rate": 1.426338676343505e-06, + "loss": 0.0662, + "step": 15915 + }, + { + "epoch": 6.66, + "learning_rate": 1.4254661607190624e-06, + "loss": 0.0832, + "step": 15916 + }, + { + "epoch": 6.66, + "learning_rate": 1.424593891561985e-06, + "loss": 0.0695, + "step": 15917 + }, + { + "epoch": 6.66, + "learning_rate": 1.4237218688973475e-06, + "loss": 0.0699, + "step": 15918 + }, + { + "epoch": 6.66, + "learning_rate": 1.4228500927502164e-06, + "loss": 0.0661, + "step": 15919 + }, + { + "epoch": 6.67, + "learning_rate": 1.4219785631456452e-06, + "loss": 0.0687, + "step": 15920 + }, + { + "epoch": 6.67, + "learning_rate": 1.421107280108689e-06, + "loss": 0.066, + "step": 15921 + }, + { + "epoch": 6.67, + "learning_rate": 1.4202362436643924e-06, + "loss": 0.1135, + "step": 15922 + }, + { + "epoch": 6.67, + "learning_rate": 1.41936545383779e-06, + "loss": 0.0978, + "step": 15923 + }, + { + "epoch": 6.67, + "learning_rate": 1.4184949106539137e-06, + "loss": 0.0523, + "step": 15924 + }, + { + "epoch": 6.67, + "learning_rate": 1.417624614137786e-06, + "loss": 0.105, + "step": 15925 + }, + { + "epoch": 6.67, + "learning_rate": 1.416754564314422e-06, + "loss": 0.085, + "step": 15926 + }, + { + "epoch": 6.67, + "learning_rate": 1.4158847612088344e-06, + "loss": 0.0774, + "step": 15927 + }, + { + "epoch": 6.67, + "learning_rate": 1.4150152048460198e-06, + "loss": 0.0513, + "step": 15928 + }, + { + "epoch": 6.67, + "learning_rate": 1.4141458952509745e-06, + "loss": 0.086, + "step": 15929 + }, + { + "epoch": 6.67, + "learning_rate": 1.4132768324486868e-06, + "loss": 0.0827, + "step": 15930 + }, + { + "epoch": 6.67, + "learning_rate": 1.4124080164641362e-06, + "loss": 0.0576, + "step": 15931 + }, + { + "epoch": 6.67, + "learning_rate": 1.4115394473222977e-06, + "loss": 0.052, + "step": 15932 + }, + { + "epoch": 6.67, + "learning_rate": 1.4106711250481353e-06, + "loss": 0.0713, + "step": 15933 + }, + { + "epoch": 6.67, + "learning_rate": 1.4098030496666126e-06, + "loss": 0.086, + "step": 15934 + }, + { + "epoch": 6.67, + "learning_rate": 1.4089352212026752e-06, + "loss": 0.0926, + "step": 15935 + }, + { + "epoch": 6.67, + "learning_rate": 1.408067639681272e-06, + "loss": 0.1034, + "step": 15936 + }, + { + "epoch": 6.67, + "learning_rate": 1.4072003051273408e-06, + "loss": 0.0879, + "step": 15937 + }, + { + "epoch": 6.67, + "learning_rate": 1.4063332175658107e-06, + "loss": 0.0597, + "step": 15938 + }, + { + "epoch": 6.67, + "learning_rate": 1.405466377021607e-06, + "loss": 0.0823, + "step": 15939 + }, + { + "epoch": 6.67, + "learning_rate": 1.404599783519648e-06, + "loss": 0.0752, + "step": 15940 + }, + { + "epoch": 6.67, + "learning_rate": 1.4037334370848376e-06, + "loss": 0.0663, + "step": 15941 + }, + { + "epoch": 6.67, + "learning_rate": 1.4028673377420821e-06, + "loss": 0.0863, + "step": 15942 + }, + { + "epoch": 6.67, + "learning_rate": 1.4020014855162755e-06, + "loss": 0.0547, + "step": 15943 + }, + { + "epoch": 6.68, + "learning_rate": 1.4011358804323095e-06, + "loss": 0.0713, + "step": 15944 + }, + { + "epoch": 6.68, + "learning_rate": 1.4002705225150593e-06, + "loss": 0.0967, + "step": 15945 + }, + { + "epoch": 6.68, + "learning_rate": 1.3994054117894018e-06, + "loss": 0.0731, + "step": 15946 + }, + { + "epoch": 6.68, + "learning_rate": 1.398540548280203e-06, + "loss": 0.0999, + "step": 15947 + }, + { + "epoch": 6.68, + "learning_rate": 1.3976759320123268e-06, + "loss": 0.0824, + "step": 15948 + }, + { + "epoch": 6.68, + "learning_rate": 1.3968115630106193e-06, + "loss": 0.0903, + "step": 15949 + }, + { + "epoch": 6.68, + "learning_rate": 1.3959474412999296e-06, + "loss": 0.0937, + "step": 15950 + }, + { + "epoch": 6.68, + "learning_rate": 1.3950835669050955e-06, + "loss": 0.0559, + "step": 15951 + }, + { + "epoch": 6.68, + "learning_rate": 1.394219939850948e-06, + "loss": 0.1039, + "step": 15952 + }, + { + "epoch": 6.68, + "learning_rate": 1.3933565601623123e-06, + "loss": 0.0779, + "step": 15953 + }, + { + "epoch": 6.68, + "learning_rate": 1.3924934278640068e-06, + "loss": 0.0767, + "step": 15954 + }, + { + "epoch": 6.68, + "learning_rate": 1.3916305429808375e-06, + "loss": 0.0572, + "step": 15955 + }, + { + "epoch": 6.68, + "learning_rate": 1.3907679055376089e-06, + "loss": 0.0967, + "step": 15956 + }, + { + "epoch": 6.68, + "learning_rate": 1.3899055155591167e-06, + "loss": 0.0821, + "step": 15957 + }, + { + "epoch": 6.68, + "learning_rate": 1.3890433730701513e-06, + "loss": 0.0777, + "step": 15958 + }, + { + "epoch": 6.68, + "learning_rate": 1.3881814780954916e-06, + "loss": 0.0568, + "step": 15959 + }, + { + "epoch": 6.68, + "learning_rate": 1.3873198306599168e-06, + "loss": 0.0594, + "step": 15960 + }, + { + "epoch": 6.68, + "learning_rate": 1.3864584307881878e-06, + "loss": 0.0879, + "step": 15961 + }, + { + "epoch": 6.68, + "learning_rate": 1.3855972785050675e-06, + "loss": 0.0866, + "step": 15962 + }, + { + "epoch": 6.68, + "learning_rate": 1.3847363738353091e-06, + "loss": 0.0754, + "step": 15963 + }, + { + "epoch": 6.68, + "learning_rate": 1.3838757168036598e-06, + "loss": 0.0802, + "step": 15964 + }, + { + "epoch": 6.68, + "learning_rate": 1.3830153074348562e-06, + "loss": 0.1218, + "step": 15965 + }, + { + "epoch": 6.68, + "learning_rate": 1.382155145753631e-06, + "loss": 0.0689, + "step": 15966 + }, + { + "epoch": 6.68, + "learning_rate": 1.3812952317847116e-06, + "loss": 0.0817, + "step": 15967 + }, + { + "epoch": 6.69, + "learning_rate": 1.3804355655528123e-06, + "loss": 0.0786, + "step": 15968 + }, + { + "epoch": 6.69, + "learning_rate": 1.3795761470826408e-06, + "loss": 0.0705, + "step": 15969 + }, + { + "epoch": 6.69, + "learning_rate": 1.3787169763989038e-06, + "loss": 0.0808, + "step": 15970 + }, + { + "epoch": 6.69, + "learning_rate": 1.3778580535262975e-06, + "loss": 0.0696, + "step": 15971 + }, + { + "epoch": 6.69, + "learning_rate": 1.3769993784895096e-06, + "loss": 0.0934, + "step": 15972 + }, + { + "epoch": 6.69, + "learning_rate": 1.376140951313224e-06, + "loss": 0.0787, + "step": 15973 + }, + { + "epoch": 6.69, + "learning_rate": 1.3752827720221128e-06, + "loss": 0.0771, + "step": 15974 + }, + { + "epoch": 6.69, + "learning_rate": 1.3744248406408467e-06, + "loss": 0.0788, + "step": 15975 + }, + { + "epoch": 6.69, + "learning_rate": 1.3735671571940835e-06, + "loss": 0.0711, + "step": 15976 + }, + { + "epoch": 6.69, + "learning_rate": 1.3727097217064767e-06, + "loss": 0.0902, + "step": 15977 + }, + { + "epoch": 6.69, + "learning_rate": 1.3718525342026733e-06, + "loss": 0.0761, + "step": 15978 + }, + { + "epoch": 6.69, + "learning_rate": 1.3709955947073117e-06, + "loss": 0.0823, + "step": 15979 + }, + { + "epoch": 6.69, + "learning_rate": 1.3701389032450252e-06, + "loss": 0.0721, + "step": 15980 + }, + { + "epoch": 6.69, + "learning_rate": 1.3692824598404398e-06, + "loss": 0.0955, + "step": 15981 + }, + { + "epoch": 6.69, + "learning_rate": 1.3684262645181702e-06, + "loss": 0.0907, + "step": 15982 + }, + { + "epoch": 6.69, + "learning_rate": 1.3675703173028266e-06, + "loss": 0.0557, + "step": 15983 + }, + { + "epoch": 6.69, + "learning_rate": 1.366714618219015e-06, + "loss": 0.0809, + "step": 15984 + }, + { + "epoch": 6.69, + "learning_rate": 1.3658591672913301e-06, + "loss": 0.0915, + "step": 15985 + }, + { + "epoch": 6.69, + "learning_rate": 1.3650039645443614e-06, + "loss": 0.0763, + "step": 15986 + }, + { + "epoch": 6.69, + "learning_rate": 1.3641490100026943e-06, + "loss": 0.0912, + "step": 15987 + }, + { + "epoch": 6.69, + "learning_rate": 1.3632943036908974e-06, + "loss": 0.0533, + "step": 15988 + }, + { + "epoch": 6.69, + "learning_rate": 1.362439845633542e-06, + "loss": 0.1142, + "step": 15989 + }, + { + "epoch": 6.69, + "learning_rate": 1.3615856358551904e-06, + "loss": 0.0713, + "step": 15990 + }, + { + "epoch": 6.69, + "learning_rate": 1.3607316743803922e-06, + "loss": 0.0435, + "step": 15991 + }, + { + "epoch": 6.7, + "learning_rate": 1.3598779612336965e-06, + "loss": 0.0534, + "step": 15992 + }, + { + "epoch": 6.7, + "learning_rate": 1.3590244964396403e-06, + "loss": 0.0946, + "step": 15993 + }, + { + "epoch": 6.7, + "learning_rate": 1.3581712800227576e-06, + "loss": 0.0693, + "step": 15994 + }, + { + "epoch": 6.7, + "learning_rate": 1.357318312007575e-06, + "loss": 0.0973, + "step": 15995 + }, + { + "epoch": 6.7, + "learning_rate": 1.3564655924186066e-06, + "loss": 0.0871, + "step": 15996 + }, + { + "epoch": 6.7, + "learning_rate": 1.3556131212803635e-06, + "loss": 0.0774, + "step": 15997 + }, + { + "epoch": 6.7, + "learning_rate": 1.354760898617351e-06, + "loss": 0.074, + "step": 15998 + }, + { + "epoch": 6.7, + "learning_rate": 1.3539089244540648e-06, + "loss": 0.0737, + "step": 15999 + }, + { + "epoch": 6.7, + "learning_rate": 1.3530571988149943e-06, + "loss": 0.0847, + "step": 16000 + }, + { + "epoch": 6.7, + "learning_rate": 1.3522057217246232e-06, + "loss": 0.0977, + "step": 16001 + }, + { + "epoch": 6.7, + "learning_rate": 1.351354493207423e-06, + "loss": 0.0814, + "step": 16002 + }, + { + "epoch": 6.7, + "learning_rate": 1.350503513287863e-06, + "loss": 0.0802, + "step": 16003 + }, + { + "epoch": 6.7, + "learning_rate": 1.3496527819904038e-06, + "loss": 0.0554, + "step": 16004 + }, + { + "epoch": 6.7, + "learning_rate": 1.348802299339499e-06, + "loss": 0.0923, + "step": 16005 + }, + { + "epoch": 6.7, + "learning_rate": 1.3479520653595957e-06, + "loss": 0.0737, + "step": 16006 + }, + { + "epoch": 6.7, + "learning_rate": 1.3471020800751322e-06, + "loss": 0.1104, + "step": 16007 + }, + { + "epoch": 6.7, + "learning_rate": 1.3462523435105436e-06, + "loss": 0.0921, + "step": 16008 + }, + { + "epoch": 6.7, + "learning_rate": 1.3454028556902498e-06, + "loss": 0.0635, + "step": 16009 + }, + { + "epoch": 6.7, + "learning_rate": 1.3445536166386708e-06, + "loss": 0.0707, + "step": 16010 + }, + { + "epoch": 6.7, + "learning_rate": 1.3437046263802178e-06, + "loss": 0.0919, + "step": 16011 + }, + { + "epoch": 6.7, + "learning_rate": 1.342855884939296e-06, + "loss": 0.0601, + "step": 16012 + }, + { + "epoch": 6.7, + "learning_rate": 1.3420073923402976e-06, + "loss": 0.0563, + "step": 16013 + }, + { + "epoch": 6.7, + "learning_rate": 1.3411591486076136e-06, + "loss": 0.0855, + "step": 16014 + }, + { + "epoch": 6.71, + "learning_rate": 1.3403111537656255e-06, + "loss": 0.0811, + "step": 16015 + }, + { + "epoch": 6.71, + "learning_rate": 1.3394634078387115e-06, + "loss": 0.0934, + "step": 16016 + }, + { + "epoch": 6.71, + "learning_rate": 1.3386159108512341e-06, + "loss": 0.0717, + "step": 16017 + }, + { + "epoch": 6.71, + "learning_rate": 1.3377686628275567e-06, + "loss": 0.0788, + "step": 16018 + }, + { + "epoch": 6.71, + "learning_rate": 1.3369216637920323e-06, + "loss": 0.0582, + "step": 16019 + }, + { + "epoch": 6.71, + "learning_rate": 1.3360749137690066e-06, + "loss": 0.0806, + "step": 16020 + }, + { + "epoch": 6.71, + "learning_rate": 1.3352284127828207e-06, + "loss": 0.0704, + "step": 16021 + }, + { + "epoch": 6.71, + "learning_rate": 1.3343821608578057e-06, + "loss": 0.0473, + "step": 16022 + }, + { + "epoch": 6.71, + "learning_rate": 1.3335361580182848e-06, + "loss": 0.0612, + "step": 16023 + }, + { + "epoch": 6.71, + "learning_rate": 1.3326904042885757e-06, + "loss": 0.0891, + "step": 16024 + }, + { + "epoch": 6.71, + "learning_rate": 1.33184489969299e-06, + "loss": 0.078, + "step": 16025 + }, + { + "epoch": 6.71, + "learning_rate": 1.3309996442558304e-06, + "loss": 0.0681, + "step": 16026 + }, + { + "epoch": 6.71, + "learning_rate": 1.3301546380013942e-06, + "loss": 0.1014, + "step": 16027 + }, + { + "epoch": 6.71, + "learning_rate": 1.3293098809539706e-06, + "loss": 0.0683, + "step": 16028 + }, + { + "epoch": 6.71, + "learning_rate": 1.3284653731378382e-06, + "loss": 0.0842, + "step": 16029 + }, + { + "epoch": 6.71, + "learning_rate": 1.3276211145772743e-06, + "loss": 0.0905, + "step": 16030 + }, + { + "epoch": 6.71, + "learning_rate": 1.326777105296546e-06, + "loss": 0.0827, + "step": 16031 + }, + { + "epoch": 6.71, + "learning_rate": 1.3259333453199119e-06, + "loss": 0.085, + "step": 16032 + }, + { + "epoch": 6.71, + "learning_rate": 1.3250898346716279e-06, + "loss": 0.1026, + "step": 16033 + }, + { + "epoch": 6.71, + "learning_rate": 1.324246573375938e-06, + "loss": 0.1129, + "step": 16034 + }, + { + "epoch": 6.71, + "learning_rate": 1.323403561457084e-06, + "loss": 0.092, + "step": 16035 + }, + { + "epoch": 6.71, + "learning_rate": 1.3225607989392953e-06, + "loss": 0.1044, + "step": 16036 + }, + { + "epoch": 6.71, + "learning_rate": 1.3217182858467936e-06, + "loss": 0.0621, + "step": 16037 + }, + { + "epoch": 6.71, + "learning_rate": 1.3208760222037986e-06, + "loss": 0.0796, + "step": 16038 + }, + { + "epoch": 6.72, + "learning_rate": 1.3200340080345208e-06, + "loss": 0.0831, + "step": 16039 + }, + { + "epoch": 6.72, + "learning_rate": 1.3191922433631632e-06, + "loss": 0.0672, + "step": 16040 + }, + { + "epoch": 6.72, + "learning_rate": 1.318350728213922e-06, + "loss": 0.0612, + "step": 16041 + }, + { + "epoch": 6.72, + "learning_rate": 1.3175094626109864e-06, + "loss": 0.0675, + "step": 16042 + }, + { + "epoch": 6.72, + "learning_rate": 1.316668446578535e-06, + "loss": 0.093, + "step": 16043 + }, + { + "epoch": 6.72, + "learning_rate": 1.3158276801407432e-06, + "loss": 0.0971, + "step": 16044 + }, + { + "epoch": 6.72, + "learning_rate": 1.314987163321778e-06, + "loss": 0.0573, + "step": 16045 + }, + { + "epoch": 6.72, + "learning_rate": 1.3141468961458003e-06, + "loss": 0.0859, + "step": 16046 + }, + { + "epoch": 6.72, + "learning_rate": 1.313306878636963e-06, + "loss": 0.0638, + "step": 16047 + }, + { + "epoch": 6.72, + "learning_rate": 1.3124671108194108e-06, + "loss": 0.0778, + "step": 16048 + }, + { + "epoch": 6.72, + "learning_rate": 1.3116275927172838e-06, + "loss": 0.0863, + "step": 16049 + }, + { + "epoch": 6.72, + "learning_rate": 1.3107883243547103e-06, + "loss": 0.0594, + "step": 16050 + }, + { + "epoch": 6.72, + "learning_rate": 1.3099493057558155e-06, + "loss": 0.0832, + "step": 16051 + }, + { + "epoch": 6.72, + "learning_rate": 1.3091105369447166e-06, + "loss": 0.0584, + "step": 16052 + }, + { + "epoch": 6.72, + "learning_rate": 1.3082720179455233e-06, + "loss": 0.1038, + "step": 16053 + }, + { + "epoch": 6.72, + "learning_rate": 1.3074337487823374e-06, + "loss": 0.0928, + "step": 16054 + }, + { + "epoch": 6.72, + "learning_rate": 1.3065957294792576e-06, + "loss": 0.0696, + "step": 16055 + }, + { + "epoch": 6.72, + "learning_rate": 1.3057579600603666e-06, + "loss": 0.0758, + "step": 16056 + }, + { + "epoch": 6.72, + "learning_rate": 1.3049204405497473e-06, + "loss": 0.0382, + "step": 16057 + }, + { + "epoch": 6.72, + "learning_rate": 1.3040831709714774e-06, + "loss": 0.0807, + "step": 16058 + }, + { + "epoch": 6.72, + "learning_rate": 1.3032461513496176e-06, + "loss": 0.0612, + "step": 16059 + }, + { + "epoch": 6.72, + "learning_rate": 1.3024093817082295e-06, + "loss": 0.0789, + "step": 16060 + }, + { + "epoch": 6.72, + "learning_rate": 1.3015728620713653e-06, + "loss": 0.0679, + "step": 16061 + }, + { + "epoch": 6.72, + "learning_rate": 1.300736592463071e-06, + "loss": 0.1048, + "step": 16062 + }, + { + "epoch": 6.73, + "learning_rate": 1.2999005729073855e-06, + "loss": 0.0882, + "step": 16063 + }, + { + "epoch": 6.73, + "learning_rate": 1.299064803428336e-06, + "loss": 0.0785, + "step": 16064 + }, + { + "epoch": 6.73, + "learning_rate": 1.2982292840499477e-06, + "loss": 0.0983, + "step": 16065 + }, + { + "epoch": 6.73, + "learning_rate": 1.2973940147962361e-06, + "loss": 0.0918, + "step": 16066 + }, + { + "epoch": 6.73, + "learning_rate": 1.2965589956912117e-06, + "loss": 0.0908, + "step": 16067 + }, + { + "epoch": 6.73, + "learning_rate": 1.2957242267588754e-06, + "loss": 0.069, + "step": 16068 + }, + { + "epoch": 6.73, + "learning_rate": 1.2948897080232248e-06, + "loss": 0.082, + "step": 16069 + }, + { + "epoch": 6.73, + "learning_rate": 1.2940554395082416e-06, + "loss": 0.0861, + "step": 16070 + }, + { + "epoch": 6.73, + "learning_rate": 1.2932214212379102e-06, + "loss": 0.094, + "step": 16071 + }, + { + "epoch": 6.73, + "learning_rate": 1.2923876532362022e-06, + "loss": 0.0741, + "step": 16072 + }, + { + "epoch": 6.73, + "learning_rate": 1.2915541355270845e-06, + "loss": 0.0595, + "step": 16073 + }, + { + "epoch": 6.73, + "learning_rate": 1.290720868134515e-06, + "loss": 0.0915, + "step": 16074 + }, + { + "epoch": 6.73, + "learning_rate": 1.2898878510824453e-06, + "loss": 0.0738, + "step": 16075 + }, + { + "epoch": 6.73, + "learning_rate": 1.2890550843948234e-06, + "loss": 0.0591, + "step": 16076 + }, + { + "epoch": 6.73, + "learning_rate": 1.2882225680955806e-06, + "loss": 0.0956, + "step": 16077 + }, + { + "epoch": 6.73, + "learning_rate": 1.2873903022086487e-06, + "loss": 0.0882, + "step": 16078 + }, + { + "epoch": 6.73, + "learning_rate": 1.2865582867579517e-06, + "loss": 0.0676, + "step": 16079 + }, + { + "epoch": 6.73, + "learning_rate": 1.2857265217674063e-06, + "loss": 0.0733, + "step": 16080 + }, + { + "epoch": 6.73, + "learning_rate": 1.2848950072609179e-06, + "loss": 0.0873, + "step": 16081 + }, + { + "epoch": 6.73, + "learning_rate": 1.2840637432623881e-06, + "loss": 0.0824, + "step": 16082 + }, + { + "epoch": 6.73, + "learning_rate": 1.2832327297957136e-06, + "loss": 0.057, + "step": 16083 + }, + { + "epoch": 6.73, + "learning_rate": 1.2824019668847775e-06, + "loss": 0.097, + "step": 16084 + }, + { + "epoch": 6.73, + "learning_rate": 1.2815714545534608e-06, + "loss": 0.0892, + "step": 16085 + }, + { + "epoch": 6.73, + "learning_rate": 1.2807411928256363e-06, + "loss": 0.0875, + "step": 16086 + }, + { + "epoch": 6.74, + "learning_rate": 1.2799111817251686e-06, + "loss": 0.0748, + "step": 16087 + }, + { + "epoch": 6.74, + "learning_rate": 1.2790814212759161e-06, + "loss": 0.0777, + "step": 16088 + }, + { + "epoch": 6.74, + "learning_rate": 1.27825191150173e-06, + "loss": 0.053, + "step": 16089 + }, + { + "epoch": 6.74, + "learning_rate": 1.2774226524264545e-06, + "loss": 0.0814, + "step": 16090 + }, + { + "epoch": 6.74, + "learning_rate": 1.2765936440739225e-06, + "loss": 0.0773, + "step": 16091 + }, + { + "epoch": 6.74, + "learning_rate": 1.275764886467965e-06, + "loss": 0.0506, + "step": 16092 + }, + { + "epoch": 6.74, + "learning_rate": 1.2749363796324043e-06, + "loss": 0.0743, + "step": 16093 + }, + { + "epoch": 6.74, + "learning_rate": 1.2741081235910547e-06, + "loss": 0.0992, + "step": 16094 + }, + { + "epoch": 6.74, + "learning_rate": 1.2732801183677246e-06, + "loss": 0.0694, + "step": 16095 + }, + { + "epoch": 6.74, + "learning_rate": 1.2724523639862153e-06, + "loss": 0.0733, + "step": 16096 + }, + { + "epoch": 6.74, + "learning_rate": 1.2716248604703153e-06, + "loss": 0.0818, + "step": 16097 + }, + { + "epoch": 6.74, + "learning_rate": 1.2707976078438132e-06, + "loss": 0.1016, + "step": 16098 + }, + { + "epoch": 6.74, + "learning_rate": 1.2699706061304884e-06, + "loss": 0.0834, + "step": 16099 + }, + { + "epoch": 6.74, + "learning_rate": 1.2691438553541112e-06, + "loss": 0.0431, + "step": 16100 + }, + { + "epoch": 6.74, + "learning_rate": 1.2683173555384476e-06, + "loss": 0.0985, + "step": 16101 + }, + { + "epoch": 6.74, + "learning_rate": 1.2674911067072514e-06, + "loss": 0.0744, + "step": 16102 + }, + { + "epoch": 6.74, + "learning_rate": 1.2666651088842773e-06, + "loss": 0.0862, + "step": 16103 + }, + { + "epoch": 6.74, + "learning_rate": 1.2658393620932652e-06, + "loss": 0.0848, + "step": 16104 + }, + { + "epoch": 6.74, + "learning_rate": 1.265013866357947e-06, + "loss": 0.0941, + "step": 16105 + }, + { + "epoch": 6.74, + "learning_rate": 1.264188621702055e-06, + "loss": 0.0776, + "step": 16106 + }, + { + "epoch": 6.74, + "learning_rate": 1.2633636281493078e-06, + "loss": 0.0564, + "step": 16107 + }, + { + "epoch": 6.74, + "learning_rate": 1.262538885723421e-06, + "loss": 0.0876, + "step": 16108 + }, + { + "epoch": 6.74, + "learning_rate": 1.2617143944481004e-06, + "loss": 0.093, + "step": 16109 + }, + { + "epoch": 6.74, + "learning_rate": 1.2608901543470475e-06, + "loss": 0.0993, + "step": 16110 + }, + { + "epoch": 6.75, + "learning_rate": 1.2600661654439505e-06, + "loss": 0.1058, + "step": 16111 + }, + { + "epoch": 6.75, + "learning_rate": 1.2592424277624948e-06, + "loss": 0.1112, + "step": 16112 + }, + { + "epoch": 6.75, + "learning_rate": 1.2584189413263604e-06, + "loss": 0.0813, + "step": 16113 + }, + { + "epoch": 6.75, + "learning_rate": 1.2575957061592159e-06, + "loss": 0.0847, + "step": 16114 + }, + { + "epoch": 6.75, + "learning_rate": 1.2567727222847237e-06, + "loss": 0.0952, + "step": 16115 + }, + { + "epoch": 6.75, + "learning_rate": 1.2559499897265426e-06, + "loss": 0.0755, + "step": 16116 + }, + { + "epoch": 6.75, + "learning_rate": 1.2551275085083215e-06, + "loss": 0.0503, + "step": 16117 + }, + { + "epoch": 6.75, + "learning_rate": 1.254305278653698e-06, + "loss": 0.0829, + "step": 16118 + }, + { + "epoch": 6.75, + "learning_rate": 1.2534833001863078e-06, + "loss": 0.0863, + "step": 16119 + }, + { + "epoch": 6.75, + "learning_rate": 1.2526615731297787e-06, + "loss": 0.0845, + "step": 16120 + }, + { + "epoch": 6.75, + "learning_rate": 1.2518400975077316e-06, + "loss": 0.0666, + "step": 16121 + }, + { + "epoch": 6.75, + "learning_rate": 1.2510188733437767e-06, + "loss": 0.0918, + "step": 16122 + }, + { + "epoch": 6.75, + "learning_rate": 1.250197900661523e-06, + "loss": 0.0965, + "step": 16123 + }, + { + "epoch": 6.75, + "learning_rate": 1.2493771794845644e-06, + "loss": 0.0821, + "step": 16124 + }, + { + "epoch": 6.75, + "learning_rate": 1.2485567098364937e-06, + "loss": 0.0706, + "step": 16125 + }, + { + "epoch": 6.75, + "learning_rate": 1.2477364917408964e-06, + "loss": 0.0577, + "step": 16126 + }, + { + "epoch": 6.75, + "learning_rate": 1.2469165252213445e-06, + "loss": 0.0698, + "step": 16127 + }, + { + "epoch": 6.75, + "learning_rate": 1.2460968103014104e-06, + "loss": 0.0853, + "step": 16128 + }, + { + "epoch": 6.75, + "learning_rate": 1.2452773470046543e-06, + "loss": 0.0955, + "step": 16129 + }, + { + "epoch": 6.75, + "learning_rate": 1.2444581353546326e-06, + "loss": 0.0921, + "step": 16130 + }, + { + "epoch": 6.75, + "learning_rate": 1.2436391753748955e-06, + "loss": 0.0521, + "step": 16131 + }, + { + "epoch": 6.75, + "learning_rate": 1.2428204670889766e-06, + "loss": 0.0614, + "step": 16132 + }, + { + "epoch": 6.75, + "learning_rate": 1.2420020105204122e-06, + "loss": 0.0705, + "step": 16133 + }, + { + "epoch": 6.75, + "learning_rate": 1.2411838056927295e-06, + "loss": 0.0681, + "step": 16134 + }, + { + "epoch": 6.76, + "learning_rate": 1.2403658526294448e-06, + "loss": 0.0632, + "step": 16135 + }, + { + "epoch": 6.76, + "learning_rate": 1.2395481513540719e-06, + "loss": 0.0803, + "step": 16136 + }, + { + "epoch": 6.76, + "learning_rate": 1.238730701890114e-06, + "loss": 0.0879, + "step": 16137 + }, + { + "epoch": 6.76, + "learning_rate": 1.2379135042610669e-06, + "loss": 0.0931, + "step": 16138 + }, + { + "epoch": 6.76, + "learning_rate": 1.2370965584904194e-06, + "loss": 0.078, + "step": 16139 + }, + { + "epoch": 6.76, + "learning_rate": 1.2362798646016561e-06, + "loss": 0.0907, + "step": 16140 + }, + { + "epoch": 6.76, + "learning_rate": 1.2354634226182506e-06, + "loss": 0.0921, + "step": 16141 + }, + { + "epoch": 6.76, + "learning_rate": 1.2346472325636717e-06, + "loss": 0.0967, + "step": 16142 + }, + { + "epoch": 6.76, + "learning_rate": 1.2338312944613828e-06, + "loss": 0.0937, + "step": 16143 + }, + { + "epoch": 6.76, + "learning_rate": 1.233015608334831e-06, + "loss": 0.0802, + "step": 16144 + }, + { + "epoch": 6.76, + "learning_rate": 1.2322001742074674e-06, + "loss": 0.0816, + "step": 16145 + }, + { + "epoch": 6.76, + "learning_rate": 1.2313849921027277e-06, + "loss": 0.0711, + "step": 16146 + }, + { + "epoch": 6.76, + "learning_rate": 1.2305700620440463e-06, + "loss": 0.1264, + "step": 16147 + }, + { + "epoch": 6.76, + "learning_rate": 1.2297553840548481e-06, + "loss": 0.0625, + "step": 16148 + }, + { + "epoch": 6.76, + "learning_rate": 1.2289409581585475e-06, + "loss": 0.1041, + "step": 16149 + }, + { + "epoch": 6.76, + "learning_rate": 1.228126784378555e-06, + "loss": 0.0655, + "step": 16150 + }, + { + "epoch": 6.76, + "learning_rate": 1.2273128627382768e-06, + "loss": 0.0905, + "step": 16151 + }, + { + "epoch": 6.76, + "learning_rate": 1.2264991932611036e-06, + "loss": 0.0756, + "step": 16152 + }, + { + "epoch": 6.76, + "learning_rate": 1.2256857759704255e-06, + "loss": 0.099, + "step": 16153 + }, + { + "epoch": 6.76, + "learning_rate": 1.2248726108896225e-06, + "loss": 0.075, + "step": 16154 + }, + { + "epoch": 6.76, + "learning_rate": 1.2240596980420704e-06, + "loss": 0.0853, + "step": 16155 + }, + { + "epoch": 6.76, + "learning_rate": 1.223247037451135e-06, + "loss": 0.0908, + "step": 16156 + }, + { + "epoch": 6.76, + "learning_rate": 1.2224346291401745e-06, + "loss": 0.0708, + "step": 16157 + }, + { + "epoch": 6.76, + "learning_rate": 1.2216224731325433e-06, + "loss": 0.0498, + "step": 16158 + }, + { + "epoch": 6.77, + "learning_rate": 1.2208105694515826e-06, + "loss": 0.0287, + "step": 16159 + }, + { + "epoch": 6.77, + "learning_rate": 1.2199989181206317e-06, + "loss": 0.1168, + "step": 16160 + }, + { + "epoch": 6.77, + "learning_rate": 1.2191875191630209e-06, + "loss": 0.098, + "step": 16161 + }, + { + "epoch": 6.77, + "learning_rate": 1.2183763726020726e-06, + "loss": 0.1242, + "step": 16162 + }, + { + "epoch": 6.77, + "learning_rate": 1.2175654784611024e-06, + "loss": 0.082, + "step": 16163 + }, + { + "epoch": 6.77, + "learning_rate": 1.216754836763422e-06, + "loss": 0.0883, + "step": 16164 + }, + { + "epoch": 6.77, + "learning_rate": 1.2159444475323268e-06, + "loss": 0.0557, + "step": 16165 + }, + { + "epoch": 6.77, + "learning_rate": 1.215134310791114e-06, + "loss": 0.0623, + "step": 16166 + }, + { + "epoch": 6.77, + "learning_rate": 1.2143244265630705e-06, + "loss": 0.0769, + "step": 16167 + }, + { + "epoch": 6.77, + "learning_rate": 1.2135147948714743e-06, + "loss": 0.0764, + "step": 16168 + }, + { + "epoch": 6.77, + "learning_rate": 1.212705415739599e-06, + "loss": 0.0581, + "step": 16169 + }, + { + "epoch": 6.77, + "learning_rate": 1.2118962891907105e-06, + "loss": 0.0961, + "step": 16170 + }, + { + "epoch": 6.77, + "learning_rate": 1.2110874152480622e-06, + "loss": 0.0848, + "step": 16171 + }, + { + "epoch": 6.77, + "learning_rate": 1.2102787939349091e-06, + "loss": 0.0636, + "step": 16172 + }, + { + "epoch": 6.77, + "learning_rate": 1.2094704252744916e-06, + "loss": 0.0711, + "step": 16173 + }, + { + "epoch": 6.77, + "learning_rate": 1.2086623092900441e-06, + "loss": 0.0806, + "step": 16174 + }, + { + "epoch": 6.77, + "learning_rate": 1.2078544460047991e-06, + "loss": 0.077, + "step": 16175 + }, + { + "epoch": 6.77, + "learning_rate": 1.207046835441975e-06, + "loss": 0.0656, + "step": 16176 + }, + { + "epoch": 6.77, + "learning_rate": 1.2062394776247865e-06, + "loss": 0.0556, + "step": 16177 + }, + { + "epoch": 6.77, + "learning_rate": 1.2054323725764439e-06, + "loss": 0.0943, + "step": 16178 + }, + { + "epoch": 6.77, + "learning_rate": 1.2046255203201407e-06, + "loss": 0.0666, + "step": 16179 + }, + { + "epoch": 6.77, + "learning_rate": 1.2038189208790718e-06, + "loss": 0.0834, + "step": 16180 + }, + { + "epoch": 6.77, + "learning_rate": 1.203012574276423e-06, + "loss": 0.0597, + "step": 16181 + }, + { + "epoch": 6.77, + "learning_rate": 1.2022064805353717e-06, + "loss": 0.0525, + "step": 16182 + }, + { + "epoch": 6.78, + "learning_rate": 1.2014006396790878e-06, + "loss": 0.0871, + "step": 16183 + }, + { + "epoch": 6.78, + "learning_rate": 1.2005950517307374e-06, + "loss": 0.077, + "step": 16184 + }, + { + "epoch": 6.78, + "learning_rate": 1.1997897167134708e-06, + "loss": 0.0911, + "step": 16185 + }, + { + "epoch": 6.78, + "learning_rate": 1.1989846346504396e-06, + "loss": 0.0771, + "step": 16186 + }, + { + "epoch": 6.78, + "learning_rate": 1.1981798055647865e-06, + "loss": 0.0922, + "step": 16187 + }, + { + "epoch": 6.78, + "learning_rate": 1.1973752294796426e-06, + "loss": 0.0934, + "step": 16188 + }, + { + "epoch": 6.78, + "learning_rate": 1.1965709064181374e-06, + "loss": 0.0966, + "step": 16189 + }, + { + "epoch": 6.78, + "learning_rate": 1.195766836403389e-06, + "loss": 0.0765, + "step": 16190 + }, + { + "epoch": 6.78, + "learning_rate": 1.1949630194585127e-06, + "loss": 0.0615, + "step": 16191 + }, + { + "epoch": 6.78, + "learning_rate": 1.1941594556066083e-06, + "loss": 0.0802, + "step": 16192 + }, + { + "epoch": 6.78, + "learning_rate": 1.1933561448707764e-06, + "loss": 0.0944, + "step": 16193 + }, + { + "epoch": 6.78, + "learning_rate": 1.1925530872741098e-06, + "loss": 0.0783, + "step": 16194 + }, + { + "epoch": 6.78, + "learning_rate": 1.1917502828396864e-06, + "loss": 0.0789, + "step": 16195 + }, + { + "epoch": 6.78, + "learning_rate": 1.190947731590586e-06, + "loss": 0.0661, + "step": 16196 + }, + { + "epoch": 6.78, + "learning_rate": 1.1901454335498752e-06, + "loss": 0.1021, + "step": 16197 + }, + { + "epoch": 6.78, + "learning_rate": 1.1893433887406158e-06, + "loss": 0.082, + "step": 16198 + }, + { + "epoch": 6.78, + "learning_rate": 1.188541597185865e-06, + "loss": 0.0746, + "step": 16199 + }, + { + "epoch": 6.78, + "learning_rate": 1.187740058908664e-06, + "loss": 0.1068, + "step": 16200 + }, + { + "epoch": 6.78, + "learning_rate": 1.186938773932056e-06, + "loss": 0.0991, + "step": 16201 + }, + { + "epoch": 6.78, + "learning_rate": 1.186137742279072e-06, + "loss": 0.09, + "step": 16202 + }, + { + "epoch": 6.78, + "learning_rate": 1.1853369639727376e-06, + "loss": 0.0617, + "step": 16203 + }, + { + "epoch": 6.78, + "learning_rate": 1.1845364390360692e-06, + "loss": 0.0599, + "step": 16204 + }, + { + "epoch": 6.78, + "learning_rate": 1.183736167492081e-06, + "loss": 0.0962, + "step": 16205 + }, + { + "epoch": 6.79, + "learning_rate": 1.1829361493637715e-06, + "loss": 0.0671, + "step": 16206 + }, + { + "epoch": 6.79, + "learning_rate": 1.1821363846741374e-06, + "loss": 0.0784, + "step": 16207 + }, + { + "epoch": 6.79, + "learning_rate": 1.1813368734461683e-06, + "loss": 0.0494, + "step": 16208 + }, + { + "epoch": 6.79, + "learning_rate": 1.1805376157028458e-06, + "loss": 0.0738, + "step": 16209 + }, + { + "epoch": 6.79, + "learning_rate": 1.1797386114671428e-06, + "loss": 0.0847, + "step": 16210 + }, + { + "epoch": 6.79, + "learning_rate": 1.1789398607620283e-06, + "loss": 0.0901, + "step": 16211 + }, + { + "epoch": 6.79, + "learning_rate": 1.1781413636104588e-06, + "loss": 0.0883, + "step": 16212 + }, + { + "epoch": 6.79, + "learning_rate": 1.1773431200353857e-06, + "loss": 0.0552, + "step": 16213 + }, + { + "epoch": 6.79, + "learning_rate": 1.1765451300597574e-06, + "loss": 0.0869, + "step": 16214 + }, + { + "epoch": 6.79, + "learning_rate": 1.175747393706509e-06, + "loss": 0.0581, + "step": 16215 + }, + { + "epoch": 6.79, + "learning_rate": 1.1749499109985728e-06, + "loss": 0.1278, + "step": 16216 + }, + { + "epoch": 6.79, + "learning_rate": 1.1741526819588679e-06, + "loss": 0.1016, + "step": 16217 + }, + { + "epoch": 6.79, + "learning_rate": 1.173355706610313e-06, + "loss": 0.0581, + "step": 16218 + }, + { + "epoch": 6.79, + "learning_rate": 1.1725589849758178e-06, + "loss": 0.0699, + "step": 16219 + }, + { + "epoch": 6.79, + "learning_rate": 1.1717625170782788e-06, + "loss": 0.0806, + "step": 16220 + }, + { + "epoch": 6.79, + "learning_rate": 1.1709663029405916e-06, + "loss": 0.0566, + "step": 16221 + }, + { + "epoch": 6.79, + "learning_rate": 1.1701703425856437e-06, + "loss": 0.0792, + "step": 16222 + }, + { + "epoch": 6.79, + "learning_rate": 1.1693746360363145e-06, + "loss": 0.0864, + "step": 16223 + }, + { + "epoch": 6.79, + "learning_rate": 1.1685791833154748e-06, + "loss": 0.0726, + "step": 16224 + }, + { + "epoch": 6.79, + "learning_rate": 1.1677839844459914e-06, + "loss": 0.0665, + "step": 16225 + }, + { + "epoch": 6.79, + "learning_rate": 1.1669890394507189e-06, + "loss": 0.08, + "step": 16226 + }, + { + "epoch": 6.79, + "learning_rate": 1.1661943483525074e-06, + "loss": 0.0594, + "step": 16227 + }, + { + "epoch": 6.79, + "learning_rate": 1.1653999111741998e-06, + "loss": 0.0887, + "step": 16228 + }, + { + "epoch": 6.79, + "learning_rate": 1.1646057279386325e-06, + "loss": 0.0723, + "step": 16229 + }, + { + "epoch": 6.8, + "learning_rate": 1.1638117986686336e-06, + "loss": 0.0681, + "step": 16230 + }, + { + "epoch": 6.8, + "learning_rate": 1.163018123387023e-06, + "loss": 0.0856, + "step": 16231 + }, + { + "epoch": 6.8, + "learning_rate": 1.1622247021166166e-06, + "loss": 0.0538, + "step": 16232 + }, + { + "epoch": 6.8, + "learning_rate": 1.1614315348802164e-06, + "loss": 0.0484, + "step": 16233 + }, + { + "epoch": 6.8, + "learning_rate": 1.160638621700624e-06, + "loss": 0.1014, + "step": 16234 + }, + { + "epoch": 6.8, + "learning_rate": 1.1598459626006308e-06, + "loss": 0.0939, + "step": 16235 + }, + { + "epoch": 6.8, + "learning_rate": 1.1590535576030204e-06, + "loss": 0.0629, + "step": 16236 + }, + { + "epoch": 6.8, + "learning_rate": 1.1582614067305697e-06, + "loss": 0.09, + "step": 16237 + }, + { + "epoch": 6.8, + "learning_rate": 1.1574695100060517e-06, + "loss": 0.0807, + "step": 16238 + }, + { + "epoch": 6.8, + "learning_rate": 1.1566778674522228e-06, + "loss": 0.0849, + "step": 16239 + }, + { + "epoch": 6.8, + "learning_rate": 1.1558864790918444e-06, + "loss": 0.0493, + "step": 16240 + }, + { + "epoch": 6.8, + "learning_rate": 1.1550953449476588e-06, + "loss": 0.1025, + "step": 16241 + }, + { + "epoch": 6.8, + "learning_rate": 1.1543044650424085e-06, + "loss": 0.0826, + "step": 16242 + }, + { + "epoch": 6.8, + "learning_rate": 1.1535138393988265e-06, + "loss": 0.0749, + "step": 16243 + }, + { + "epoch": 6.8, + "learning_rate": 1.1527234680396392e-06, + "loss": 0.0706, + "step": 16244 + }, + { + "epoch": 6.8, + "learning_rate": 1.1519333509875642e-06, + "loss": 0.079, + "step": 16245 + }, + { + "epoch": 6.8, + "learning_rate": 1.151143488265316e-06, + "loss": 0.0734, + "step": 16246 + }, + { + "epoch": 6.8, + "learning_rate": 1.1503538798955937e-06, + "loss": 0.1182, + "step": 16247 + }, + { + "epoch": 6.8, + "learning_rate": 1.1495645259010969e-06, + "loss": 0.083, + "step": 16248 + }, + { + "epoch": 6.8, + "learning_rate": 1.1487754263045138e-06, + "loss": 0.0921, + "step": 16249 + }, + { + "epoch": 6.8, + "learning_rate": 1.1479865811285262e-06, + "loss": 0.0853, + "step": 16250 + }, + { + "epoch": 6.8, + "learning_rate": 1.1471979903958087e-06, + "loss": 0.0741, + "step": 16251 + }, + { + "epoch": 6.8, + "learning_rate": 1.1464096541290326e-06, + "loss": 0.0754, + "step": 16252 + }, + { + "epoch": 6.8, + "learning_rate": 1.1456215723508512e-06, + "loss": 0.0952, + "step": 16253 + }, + { + "epoch": 6.81, + "learning_rate": 1.1448337450839208e-06, + "loss": 0.0692, + "step": 16254 + }, + { + "epoch": 6.81, + "learning_rate": 1.1440461723508867e-06, + "loss": 0.0656, + "step": 16255 + }, + { + "epoch": 6.81, + "learning_rate": 1.143258854174386e-06, + "loss": 0.0882, + "step": 16256 + }, + { + "epoch": 6.81, + "learning_rate": 1.142471790577051e-06, + "loss": 0.0911, + "step": 16257 + }, + { + "epoch": 6.81, + "learning_rate": 1.1416849815815045e-06, + "loss": 0.0977, + "step": 16258 + }, + { + "epoch": 6.81, + "learning_rate": 1.140898427210364e-06, + "loss": 0.0682, + "step": 16259 + }, + { + "epoch": 6.81, + "learning_rate": 1.1401121274862348e-06, + "loss": 0.0785, + "step": 16260 + }, + { + "epoch": 6.81, + "learning_rate": 1.1393260824317208e-06, + "loss": 0.0927, + "step": 16261 + }, + { + "epoch": 6.81, + "learning_rate": 1.1385402920694167e-06, + "loss": 0.0633, + "step": 16262 + }, + { + "epoch": 6.81, + "learning_rate": 1.137754756421907e-06, + "loss": 0.0938, + "step": 16263 + }, + { + "epoch": 6.81, + "learning_rate": 1.1369694755117733e-06, + "loss": 0.051, + "step": 16264 + }, + { + "epoch": 6.81, + "learning_rate": 1.136184449361586e-06, + "loss": 0.0968, + "step": 16265 + }, + { + "epoch": 6.81, + "learning_rate": 1.1353996779939135e-06, + "loss": 0.1161, + "step": 16266 + }, + { + "epoch": 6.81, + "learning_rate": 1.1346151614313095e-06, + "loss": 0.0944, + "step": 16267 + }, + { + "epoch": 6.81, + "learning_rate": 1.1338308996963255e-06, + "loss": 0.0622, + "step": 16268 + }, + { + "epoch": 6.81, + "learning_rate": 1.1330468928115036e-06, + "loss": 0.0881, + "step": 16269 + }, + { + "epoch": 6.81, + "learning_rate": 1.132263140799381e-06, + "loss": 0.1001, + "step": 16270 + }, + { + "epoch": 6.81, + "learning_rate": 1.131479643682486e-06, + "loss": 0.082, + "step": 16271 + }, + { + "epoch": 6.81, + "learning_rate": 1.1306964014833378e-06, + "loss": 0.0911, + "step": 16272 + }, + { + "epoch": 6.81, + "learning_rate": 1.129913414224454e-06, + "loss": 0.0798, + "step": 16273 + }, + { + "epoch": 6.81, + "learning_rate": 1.1291306819283353e-06, + "loss": 0.0896, + "step": 16274 + }, + { + "epoch": 6.81, + "learning_rate": 1.1283482046174832e-06, + "loss": 0.0488, + "step": 16275 + }, + { + "epoch": 6.81, + "learning_rate": 1.1275659823143903e-06, + "loss": 0.0652, + "step": 16276 + }, + { + "epoch": 6.81, + "learning_rate": 1.1267840150415388e-06, + "loss": 0.093, + "step": 16277 + }, + { + "epoch": 6.82, + "learning_rate": 1.1260023028214072e-06, + "loss": 0.0839, + "step": 16278 + }, + { + "epoch": 6.82, + "learning_rate": 1.125220845676467e-06, + "loss": 0.0774, + "step": 16279 + }, + { + "epoch": 6.82, + "learning_rate": 1.124439643629176e-06, + "loss": 0.0969, + "step": 16280 + }, + { + "epoch": 6.82, + "learning_rate": 1.1236586967019913e-06, + "loss": 0.0869, + "step": 16281 + }, + { + "epoch": 6.82, + "learning_rate": 1.1228780049173616e-06, + "loss": 0.0632, + "step": 16282 + }, + { + "epoch": 6.82, + "learning_rate": 1.1220975682977254e-06, + "loss": 0.0939, + "step": 16283 + }, + { + "epoch": 6.82, + "learning_rate": 1.121317386865518e-06, + "loss": 0.077, + "step": 16284 + }, + { + "epoch": 6.82, + "learning_rate": 1.120537460643163e-06, + "loss": 0.0857, + "step": 16285 + }, + { + "epoch": 6.82, + "learning_rate": 1.1197577896530776e-06, + "loss": 0.0703, + "step": 16286 + }, + { + "epoch": 6.82, + "learning_rate": 1.1189783739176774e-06, + "loss": 0.0719, + "step": 16287 + }, + { + "epoch": 6.82, + "learning_rate": 1.1181992134593612e-06, + "loss": 0.0721, + "step": 16288 + }, + { + "epoch": 6.82, + "learning_rate": 1.1174203083005263e-06, + "loss": 0.0799, + "step": 16289 + }, + { + "epoch": 6.82, + "learning_rate": 1.116641658463562e-06, + "loss": 0.0699, + "step": 16290 + }, + { + "epoch": 6.82, + "learning_rate": 1.1158632639708512e-06, + "loss": 0.0949, + "step": 16291 + }, + { + "epoch": 6.82, + "learning_rate": 1.1150851248447669e-06, + "loss": 0.0699, + "step": 16292 + }, + { + "epoch": 6.82, + "learning_rate": 1.1143072411076782e-06, + "loss": 0.0731, + "step": 16293 + }, + { + "epoch": 6.82, + "learning_rate": 1.1135296127819407e-06, + "loss": 0.0813, + "step": 16294 + }, + { + "epoch": 6.82, + "learning_rate": 1.1127522398899094e-06, + "loss": 0.081, + "step": 16295 + }, + { + "epoch": 6.82, + "learning_rate": 1.1119751224539278e-06, + "loss": 0.0699, + "step": 16296 + }, + { + "epoch": 6.82, + "learning_rate": 1.1111982604963345e-06, + "loss": 0.0821, + "step": 16297 + }, + { + "epoch": 6.82, + "learning_rate": 1.1104216540394596e-06, + "loss": 0.0959, + "step": 16298 + }, + { + "epoch": 6.82, + "learning_rate": 1.1096453031056265e-06, + "loss": 0.0907, + "step": 16299 + }, + { + "epoch": 6.82, + "learning_rate": 1.1088692077171503e-06, + "loss": 0.0513, + "step": 16300 + }, + { + "epoch": 6.82, + "learning_rate": 1.1080933678963379e-06, + "loss": 0.0845, + "step": 16301 + }, + { + "epoch": 6.83, + "learning_rate": 1.1073177836654913e-06, + "loss": 0.0945, + "step": 16302 + }, + { + "epoch": 6.83, + "learning_rate": 1.1065424550469028e-06, + "loss": 0.0662, + "step": 16303 + }, + { + "epoch": 6.83, + "learning_rate": 1.1057673820628612e-06, + "loss": 0.0858, + "step": 16304 + }, + { + "epoch": 6.83, + "learning_rate": 1.104992564735642e-06, + "loss": 0.0808, + "step": 16305 + }, + { + "epoch": 6.83, + "learning_rate": 1.1042180030875205e-06, + "loss": 0.0498, + "step": 16306 + }, + { + "epoch": 6.83, + "learning_rate": 1.1034436971407592e-06, + "loss": 0.0809, + "step": 16307 + }, + { + "epoch": 6.83, + "learning_rate": 1.1026696469176124e-06, + "loss": 0.094, + "step": 16308 + }, + { + "epoch": 6.83, + "learning_rate": 1.10189585244033e-06, + "loss": 0.1118, + "step": 16309 + }, + { + "epoch": 6.83, + "learning_rate": 1.1011223137311555e-06, + "loss": 0.0975, + "step": 16310 + }, + { + "epoch": 6.83, + "learning_rate": 1.1003490308123243e-06, + "loss": 0.0844, + "step": 16311 + }, + { + "epoch": 6.83, + "learning_rate": 1.099576003706062e-06, + "loss": 0.0935, + "step": 16312 + }, + { + "epoch": 6.83, + "learning_rate": 1.0988032324345898e-06, + "loss": 0.0727, + "step": 16313 + }, + { + "epoch": 6.83, + "learning_rate": 1.0980307170201222e-06, + "loss": 0.0765, + "step": 16314 + }, + { + "epoch": 6.83, + "learning_rate": 1.09725845748486e-06, + "loss": 0.074, + "step": 16315 + }, + { + "epoch": 6.83, + "learning_rate": 1.0964864538510022e-06, + "loss": 0.0684, + "step": 16316 + }, + { + "epoch": 6.83, + "learning_rate": 1.0957147061407414e-06, + "loss": 0.0592, + "step": 16317 + }, + { + "epoch": 6.83, + "learning_rate": 1.0949432143762596e-06, + "loss": 0.1047, + "step": 16318 + }, + { + "epoch": 6.83, + "learning_rate": 1.0941719785797333e-06, + "loss": 0.0862, + "step": 16319 + }, + { + "epoch": 6.83, + "learning_rate": 1.0934009987733318e-06, + "loss": 0.074, + "step": 16320 + }, + { + "epoch": 6.83, + "learning_rate": 1.092630274979214e-06, + "loss": 0.0656, + "step": 16321 + }, + { + "epoch": 6.83, + "learning_rate": 1.091859807219534e-06, + "loss": 0.1011, + "step": 16322 + }, + { + "epoch": 6.83, + "learning_rate": 1.091089595516439e-06, + "loss": 0.0674, + "step": 16323 + }, + { + "epoch": 6.83, + "learning_rate": 1.090319639892069e-06, + "loss": 0.0754, + "step": 16324 + }, + { + "epoch": 6.83, + "learning_rate": 1.0895499403685527e-06, + "loss": 0.0759, + "step": 16325 + }, + { + "epoch": 6.84, + "learning_rate": 1.0887804969680183e-06, + "loss": 0.0659, + "step": 16326 + }, + { + "epoch": 6.84, + "learning_rate": 1.0880113097125811e-06, + "loss": 0.0756, + "step": 16327 + }, + { + "epoch": 6.84, + "learning_rate": 1.0872423786243492e-06, + "loss": 0.0923, + "step": 16328 + }, + { + "epoch": 6.84, + "learning_rate": 1.0864737037254269e-06, + "loss": 0.0663, + "step": 16329 + }, + { + "epoch": 6.84, + "learning_rate": 1.085705285037909e-06, + "loss": 0.0682, + "step": 16330 + }, + { + "epoch": 6.84, + "learning_rate": 1.0849371225838812e-06, + "loss": 0.0596, + "step": 16331 + }, + { + "epoch": 6.84, + "learning_rate": 1.0841692163854244e-06, + "loss": 0.0642, + "step": 16332 + }, + { + "epoch": 6.84, + "learning_rate": 1.0834015664646114e-06, + "loss": 0.0716, + "step": 16333 + }, + { + "epoch": 6.84, + "learning_rate": 1.0826341728435096e-06, + "loss": 0.0658, + "step": 16334 + }, + { + "epoch": 6.84, + "learning_rate": 1.0818670355441741e-06, + "loss": 0.0611, + "step": 16335 + }, + { + "epoch": 6.84, + "learning_rate": 1.0811001545886568e-06, + "loss": 0.0533, + "step": 16336 + }, + { + "epoch": 6.84, + "learning_rate": 1.0803335299990015e-06, + "loss": 0.0934, + "step": 16337 + }, + { + "epoch": 6.84, + "learning_rate": 1.0795671617972424e-06, + "loss": 0.0777, + "step": 16338 + }, + { + "epoch": 6.84, + "learning_rate": 1.0788010500054102e-06, + "loss": 0.0946, + "step": 16339 + }, + { + "epoch": 6.84, + "learning_rate": 1.0780351946455258e-06, + "loss": 0.089, + "step": 16340 + }, + { + "epoch": 6.84, + "learning_rate": 1.077269595739604e-06, + "loss": 0.0611, + "step": 16341 + }, + { + "epoch": 6.84, + "learning_rate": 1.076504253309647e-06, + "loss": 0.096, + "step": 16342 + }, + { + "epoch": 6.84, + "learning_rate": 1.0757391673776573e-06, + "loss": 0.0851, + "step": 16343 + }, + { + "epoch": 6.84, + "learning_rate": 1.0749743379656263e-06, + "loss": 0.0889, + "step": 16344 + }, + { + "epoch": 6.84, + "learning_rate": 1.0742097650955375e-06, + "loss": 0.0625, + "step": 16345 + }, + { + "epoch": 6.84, + "learning_rate": 1.0734454487893687e-06, + "loss": 0.0934, + "step": 16346 + }, + { + "epoch": 6.84, + "learning_rate": 1.0726813890690913e-06, + "loss": 0.0325, + "step": 16347 + }, + { + "epoch": 6.84, + "learning_rate": 1.0719175859566634e-06, + "loss": 0.0627, + "step": 16348 + }, + { + "epoch": 6.84, + "learning_rate": 1.0711540394740416e-06, + "loss": 0.0725, + "step": 16349 + }, + { + "epoch": 6.85, + "learning_rate": 1.0703907496431743e-06, + "loss": 0.0745, + "step": 16350 + }, + { + "epoch": 6.85, + "learning_rate": 1.0696277164860002e-06, + "loss": 0.0632, + "step": 16351 + }, + { + "epoch": 6.85, + "learning_rate": 1.0688649400244544e-06, + "loss": 0.0666, + "step": 16352 + }, + { + "epoch": 6.85, + "learning_rate": 1.068102420280459e-06, + "loss": 0.1089, + "step": 16353 + }, + { + "epoch": 6.85, + "learning_rate": 1.0673401572759335e-06, + "loss": 0.0928, + "step": 16354 + }, + { + "epoch": 6.85, + "learning_rate": 1.0665781510327911e-06, + "loss": 0.0625, + "step": 16355 + }, + { + "epoch": 6.85, + "learning_rate": 1.06581640157293e-06, + "loss": 0.0707, + "step": 16356 + }, + { + "epoch": 6.85, + "learning_rate": 1.0650549089182482e-06, + "loss": 0.0622, + "step": 16357 + }, + { + "epoch": 6.85, + "learning_rate": 1.064293673090634e-06, + "loss": 0.0956, + "step": 16358 + }, + { + "epoch": 6.85, + "learning_rate": 1.0635326941119683e-06, + "loss": 0.0679, + "step": 16359 + }, + { + "epoch": 6.85, + "learning_rate": 1.0627719720041263e-06, + "loss": 0.073, + "step": 16360 + }, + { + "epoch": 6.85, + "learning_rate": 1.0620115067889747e-06, + "loss": 0.0908, + "step": 16361 + }, + { + "epoch": 6.85, + "learning_rate": 1.0612512984883694e-06, + "loss": 0.0869, + "step": 16362 + }, + { + "epoch": 6.85, + "learning_rate": 1.0604913471241629e-06, + "loss": 0.1014, + "step": 16363 + }, + { + "epoch": 6.85, + "learning_rate": 1.0597316527182e-06, + "loss": 0.0775, + "step": 16364 + }, + { + "epoch": 6.85, + "learning_rate": 1.0589722152923177e-06, + "loss": 0.0745, + "step": 16365 + }, + { + "epoch": 6.85, + "learning_rate": 1.058213034868345e-06, + "loss": 0.0947, + "step": 16366 + }, + { + "epoch": 6.85, + "learning_rate": 1.0574541114681046e-06, + "loss": 0.0809, + "step": 16367 + }, + { + "epoch": 6.85, + "learning_rate": 1.0566954451134093e-06, + "loss": 0.0749, + "step": 16368 + }, + { + "epoch": 6.85, + "learning_rate": 1.0559370358260667e-06, + "loss": 0.0803, + "step": 16369 + }, + { + "epoch": 6.85, + "learning_rate": 1.0551788836278776e-06, + "loss": 0.0862, + "step": 16370 + }, + { + "epoch": 6.85, + "learning_rate": 1.0544209885406342e-06, + "loss": 0.1013, + "step": 16371 + }, + { + "epoch": 6.85, + "learning_rate": 1.0536633505861216e-06, + "loss": 0.093, + "step": 16372 + }, + { + "epoch": 6.85, + "learning_rate": 1.0529059697861165e-06, + "loss": 0.0648, + "step": 16373 + }, + { + "epoch": 6.86, + "learning_rate": 1.0521488461623919e-06, + "loss": 0.0916, + "step": 16374 + }, + { + "epoch": 6.86, + "learning_rate": 1.0513919797367078e-06, + "loss": 0.0723, + "step": 16375 + }, + { + "epoch": 6.86, + "learning_rate": 1.0506353705308192e-06, + "loss": 0.0744, + "step": 16376 + }, + { + "epoch": 6.86, + "learning_rate": 1.0498790185664753e-06, + "loss": 0.0771, + "step": 16377 + }, + { + "epoch": 6.86, + "learning_rate": 1.0491229238654177e-06, + "loss": 0.0938, + "step": 16378 + }, + { + "epoch": 6.86, + "learning_rate": 1.0483670864493777e-06, + "loss": 0.0958, + "step": 16379 + }, + { + "epoch": 6.86, + "learning_rate": 1.0476115063400826e-06, + "loss": 0.0492, + "step": 16380 + }, + { + "epoch": 6.86, + "learning_rate": 1.0468561835592495e-06, + "loss": 0.0862, + "step": 16381 + }, + { + "epoch": 6.86, + "learning_rate": 1.046101118128594e-06, + "loss": 0.0814, + "step": 16382 + }, + { + "epoch": 6.86, + "learning_rate": 1.0453463100698135e-06, + "loss": 0.1073, + "step": 16383 + }, + { + "epoch": 6.86, + "learning_rate": 1.0445917594046073e-06, + "loss": 0.0732, + "step": 16384 + }, + { + "epoch": 6.86, + "learning_rate": 1.0438374661546646e-06, + "loss": 0.0696, + "step": 16385 + }, + { + "epoch": 6.86, + "learning_rate": 1.0430834303416658e-06, + "loss": 0.0517, + "step": 16386 + }, + { + "epoch": 6.86, + "learning_rate": 1.042329651987285e-06, + "loss": 0.0821, + "step": 16387 + }, + { + "epoch": 6.86, + "learning_rate": 1.0415761311131922e-06, + "loss": 0.0809, + "step": 16388 + }, + { + "epoch": 6.86, + "learning_rate": 1.0408228677410425e-06, + "loss": 0.086, + "step": 16389 + }, + { + "epoch": 6.86, + "learning_rate": 1.0400698618924888e-06, + "loss": 0.0722, + "step": 16390 + }, + { + "epoch": 6.86, + "learning_rate": 1.0393171135891754e-06, + "loss": 0.088, + "step": 16391 + }, + { + "epoch": 6.86, + "learning_rate": 1.0385646228527401e-06, + "loss": 0.0753, + "step": 16392 + }, + { + "epoch": 6.86, + "learning_rate": 1.0378123897048131e-06, + "loss": 0.0865, + "step": 16393 + }, + { + "epoch": 6.86, + "learning_rate": 1.037060414167017e-06, + "loss": 0.0897, + "step": 16394 + }, + { + "epoch": 6.86, + "learning_rate": 1.0363086962609648e-06, + "loss": 0.098, + "step": 16395 + }, + { + "epoch": 6.86, + "learning_rate": 1.0355572360082645e-06, + "loss": 0.0993, + "step": 16396 + }, + { + "epoch": 6.86, + "learning_rate": 1.0348060334305187e-06, + "loss": 0.0444, + "step": 16397 + }, + { + "epoch": 6.87, + "learning_rate": 1.0340550885493162e-06, + "loss": 0.099, + "step": 16398 + }, + { + "epoch": 6.87, + "learning_rate": 1.0333044013862437e-06, + "loss": 0.0992, + "step": 16399 + }, + { + "epoch": 6.87, + "learning_rate": 1.0325539719628785e-06, + "loss": 0.0673, + "step": 16400 + }, + { + "epoch": 6.87, + "learning_rate": 1.0318038003007924e-06, + "loss": 0.0799, + "step": 16401 + }, + { + "epoch": 6.87, + "learning_rate": 1.0310538864215503e-06, + "loss": 0.0739, + "step": 16402 + }, + { + "epoch": 6.87, + "learning_rate": 1.0303042303467027e-06, + "loss": 0.0847, + "step": 16403 + }, + { + "epoch": 6.87, + "learning_rate": 1.029554832097801e-06, + "loss": 0.0459, + "step": 16404 + }, + { + "epoch": 6.87, + "learning_rate": 1.0288056916963852e-06, + "loss": 0.0961, + "step": 16405 + }, + { + "epoch": 6.87, + "learning_rate": 1.0280568091639886e-06, + "loss": 0.0856, + "step": 16406 + }, + { + "epoch": 6.87, + "learning_rate": 1.0273081845221377e-06, + "loss": 0.0735, + "step": 16407 + }, + { + "epoch": 6.87, + "learning_rate": 1.026559817792352e-06, + "loss": 0.0828, + "step": 16408 + }, + { + "epoch": 6.87, + "learning_rate": 1.0258117089961405e-06, + "loss": 0.0677, + "step": 16409 + }, + { + "epoch": 6.87, + "learning_rate": 1.0250638581550076e-06, + "loss": 0.0815, + "step": 16410 + }, + { + "epoch": 6.87, + "learning_rate": 1.0243162652904492e-06, + "loss": 0.0893, + "step": 16411 + }, + { + "epoch": 6.87, + "learning_rate": 1.023568930423956e-06, + "loss": 0.0918, + "step": 16412 + }, + { + "epoch": 6.87, + "learning_rate": 1.0228218535770075e-06, + "loss": 0.0651, + "step": 16413 + }, + { + "epoch": 6.87, + "learning_rate": 1.0220750347710795e-06, + "loss": 0.0595, + "step": 16414 + }, + { + "epoch": 6.87, + "learning_rate": 1.0213284740276385e-06, + "loss": 0.089, + "step": 16415 + }, + { + "epoch": 6.87, + "learning_rate": 1.020582171368142e-06, + "loss": 0.0554, + "step": 16416 + }, + { + "epoch": 6.87, + "learning_rate": 1.0198361268140422e-06, + "loss": 0.0759, + "step": 16417 + }, + { + "epoch": 6.87, + "learning_rate": 1.0190903403867847e-06, + "loss": 0.0866, + "step": 16418 + }, + { + "epoch": 6.87, + "learning_rate": 1.0183448121078055e-06, + "loss": 0.0963, + "step": 16419 + }, + { + "epoch": 6.87, + "learning_rate": 1.0175995419985364e-06, + "loss": 0.0663, + "step": 16420 + }, + { + "epoch": 6.88, + "learning_rate": 1.0168545300803957e-06, + "loss": 0.0623, + "step": 16421 + }, + { + "epoch": 6.88, + "learning_rate": 1.0161097763748006e-06, + "loss": 0.0856, + "step": 16422 + }, + { + "epoch": 6.88, + "learning_rate": 1.015365280903159e-06, + "loss": 0.0765, + "step": 16423 + }, + { + "epoch": 6.88, + "learning_rate": 1.0146210436868675e-06, + "loss": 0.0778, + "step": 16424 + }, + { + "epoch": 6.88, + "learning_rate": 1.0138770647473206e-06, + "loss": 0.0706, + "step": 16425 + }, + { + "epoch": 6.88, + "learning_rate": 1.0131333441059045e-06, + "loss": 0.0771, + "step": 16426 + }, + { + "epoch": 6.88, + "learning_rate": 1.012389881783995e-06, + "loss": 0.0703, + "step": 16427 + }, + { + "epoch": 6.88, + "learning_rate": 1.011646677802962e-06, + "loss": 0.0617, + "step": 16428 + }, + { + "epoch": 6.88, + "learning_rate": 1.010903732184172e-06, + "loss": 0.097, + "step": 16429 + }, + { + "epoch": 6.88, + "learning_rate": 1.010161044948975e-06, + "loss": 0.0809, + "step": 16430 + }, + { + "epoch": 6.88, + "learning_rate": 1.0094186161187214e-06, + "loss": 0.0792, + "step": 16431 + }, + { + "epoch": 6.88, + "learning_rate": 1.0086764457147524e-06, + "loss": 0.0852, + "step": 16432 + }, + { + "epoch": 6.88, + "learning_rate": 1.007934533758399e-06, + "loss": 0.0726, + "step": 16433 + }, + { + "epoch": 6.88, + "learning_rate": 1.0071928802709886e-06, + "loss": 0.1071, + "step": 16434 + }, + { + "epoch": 6.88, + "learning_rate": 1.0064514852738406e-06, + "loss": 0.0572, + "step": 16435 + }, + { + "epoch": 6.88, + "learning_rate": 1.0057103487882624e-06, + "loss": 0.0718, + "step": 16436 + }, + { + "epoch": 6.88, + "learning_rate": 1.0049694708355583e-06, + "loss": 0.0643, + "step": 16437 + }, + { + "epoch": 6.88, + "learning_rate": 1.0042288514370258e-06, + "loss": 0.069, + "step": 16438 + }, + { + "epoch": 6.88, + "learning_rate": 1.0034884906139508e-06, + "loss": 0.079, + "step": 16439 + }, + { + "epoch": 6.88, + "learning_rate": 1.0027483883876166e-06, + "loss": 0.0699, + "step": 16440 + }, + { + "epoch": 6.88, + "learning_rate": 1.002008544779296e-06, + "loss": 0.1084, + "step": 16441 + }, + { + "epoch": 6.88, + "learning_rate": 1.0012689598102565e-06, + "loss": 0.096, + "step": 16442 + }, + { + "epoch": 6.88, + "learning_rate": 1.0005296335017557e-06, + "loss": 0.0603, + "step": 16443 + }, + { + "epoch": 6.88, + "learning_rate": 9.997905658750428e-07, + "loss": 0.0888, + "step": 16444 + }, + { + "epoch": 6.89, + "learning_rate": 9.990517569513624e-07, + "loss": 0.0721, + "step": 16445 + }, + { + "epoch": 6.89, + "learning_rate": 9.98313206751953e-07, + "loss": 0.0858, + "step": 16446 + }, + { + "epoch": 6.89, + "learning_rate": 9.975749152980429e-07, + "loss": 0.0755, + "step": 16447 + }, + { + "epoch": 6.89, + "learning_rate": 9.96836882610852e-07, + "loss": 0.0749, + "step": 16448 + }, + { + "epoch": 6.89, + "learning_rate": 9.96099108711599e-07, + "loss": 0.0614, + "step": 16449 + }, + { + "epoch": 6.89, + "learning_rate": 9.953615936214845e-07, + "loss": 0.1032, + "step": 16450 + }, + { + "epoch": 6.89, + "learning_rate": 9.946243373617104e-07, + "loss": 0.0466, + "step": 16451 + }, + { + "epoch": 6.89, + "learning_rate": 9.938873399534688e-07, + "loss": 0.0939, + "step": 16452 + }, + { + "epoch": 6.89, + "learning_rate": 9.931506014179427e-07, + "loss": 0.1175, + "step": 16453 + }, + { + "epoch": 6.89, + "learning_rate": 9.924141217763117e-07, + "loss": 0.0821, + "step": 16454 + }, + { + "epoch": 6.89, + "learning_rate": 9.916779010497424e-07, + "loss": 0.0936, + "step": 16455 + }, + { + "epoch": 6.89, + "learning_rate": 9.909419392594e-07, + "loss": 0.0785, + "step": 16456 + }, + { + "epoch": 6.89, + "learning_rate": 9.902062364264354e-07, + "loss": 0.0702, + "step": 16457 + }, + { + "epoch": 6.89, + "learning_rate": 9.89470792571997e-07, + "loss": 0.0996, + "step": 16458 + }, + { + "epoch": 6.89, + "learning_rate": 9.88735607717225e-07, + "loss": 0.0818, + "step": 16459 + }, + { + "epoch": 6.89, + "learning_rate": 9.880006818832522e-07, + "loss": 0.0734, + "step": 16460 + }, + { + "epoch": 6.89, + "learning_rate": 9.872660150912016e-07, + "loss": 0.1005, + "step": 16461 + }, + { + "epoch": 6.89, + "learning_rate": 9.865316073621945e-07, + "loss": 0.0835, + "step": 16462 + }, + { + "epoch": 6.89, + "learning_rate": 9.857974587173357e-07, + "loss": 0.1036, + "step": 16463 + }, + { + "epoch": 6.89, + "learning_rate": 9.8506356917773e-07, + "loss": 0.0803, + "step": 16464 + }, + { + "epoch": 6.89, + "learning_rate": 9.843299387644746e-07, + "loss": 0.068, + "step": 16465 + }, + { + "epoch": 6.89, + "learning_rate": 9.835965674986514e-07, + "loss": 0.0679, + "step": 16466 + }, + { + "epoch": 6.89, + "learning_rate": 9.82863455401345e-07, + "loss": 0.0814, + "step": 16467 + }, + { + "epoch": 6.89, + "learning_rate": 9.821306024936272e-07, + "loss": 0.0378, + "step": 16468 + }, + { + "epoch": 6.9, + "learning_rate": 9.813980087965625e-07, + "loss": 0.0549, + "step": 16469 + }, + { + "epoch": 6.9, + "learning_rate": 9.806656743312104e-07, + "loss": 0.0847, + "step": 16470 + }, + { + "epoch": 6.9, + "learning_rate": 9.799335991186186e-07, + "loss": 0.0792, + "step": 16471 + }, + { + "epoch": 6.9, + "learning_rate": 9.792017831798307e-07, + "loss": 0.0707, + "step": 16472 + }, + { + "epoch": 6.9, + "learning_rate": 9.784702265358826e-07, + "loss": 0.0938, + "step": 16473 + }, + { + "epoch": 6.9, + "learning_rate": 9.777389292078033e-07, + "loss": 0.0953, + "step": 16474 + }, + { + "epoch": 6.9, + "learning_rate": 9.770078912166114e-07, + "loss": 0.0846, + "step": 16475 + }, + { + "epoch": 6.9, + "learning_rate": 9.762771125833226e-07, + "loss": 0.0786, + "step": 16476 + }, + { + "epoch": 6.9, + "learning_rate": 9.7554659332894e-07, + "loss": 0.0914, + "step": 16477 + }, + { + "epoch": 6.9, + "learning_rate": 9.748163334744609e-07, + "loss": 0.0496, + "step": 16478 + }, + { + "epoch": 6.9, + "learning_rate": 9.740863330408778e-07, + "loss": 0.0769, + "step": 16479 + }, + { + "epoch": 6.9, + "learning_rate": 9.733565920491738e-07, + "loss": 0.1122, + "step": 16480 + }, + { + "epoch": 6.9, + "learning_rate": 9.726271105203255e-07, + "loss": 0.0569, + "step": 16481 + }, + { + "epoch": 6.9, + "learning_rate": 9.718978884752983e-07, + "loss": 0.0868, + "step": 16482 + }, + { + "epoch": 6.9, + "learning_rate": 9.711689259350587e-07, + "loss": 0.0809, + "step": 16483 + }, + { + "epoch": 6.9, + "learning_rate": 9.704402229205534e-07, + "loss": 0.0978, + "step": 16484 + }, + { + "epoch": 6.9, + "learning_rate": 9.697117794527311e-07, + "loss": 0.0694, + "step": 16485 + }, + { + "epoch": 6.9, + "learning_rate": 9.689835955525307e-07, + "loss": 0.0765, + "step": 16486 + }, + { + "epoch": 6.9, + "learning_rate": 9.68255671240882e-07, + "loss": 0.0612, + "step": 16487 + }, + { + "epoch": 6.9, + "learning_rate": 9.675280065387117e-07, + "loss": 0.071, + "step": 16488 + }, + { + "epoch": 6.9, + "learning_rate": 9.668006014669328e-07, + "loss": 0.0822, + "step": 16489 + }, + { + "epoch": 6.9, + "learning_rate": 9.660734560464557e-07, + "loss": 0.1086, + "step": 16490 + }, + { + "epoch": 6.9, + "learning_rate": 9.653465702981779e-07, + "loss": 0.0845, + "step": 16491 + }, + { + "epoch": 6.9, + "learning_rate": 9.64619944242997e-07, + "loss": 0.0677, + "step": 16492 + }, + { + "epoch": 6.91, + "learning_rate": 9.638935779017977e-07, + "loss": 0.0838, + "step": 16493 + }, + { + "epoch": 6.91, + "learning_rate": 9.631674712954586e-07, + "loss": 0.0735, + "step": 16494 + }, + { + "epoch": 6.91, + "learning_rate": 9.624416244448508e-07, + "loss": 0.0865, + "step": 16495 + }, + { + "epoch": 6.91, + "learning_rate": 9.6171603737084e-07, + "loss": 0.0667, + "step": 16496 + }, + { + "epoch": 6.91, + "learning_rate": 9.609907100942828e-07, + "loss": 0.0828, + "step": 16497 + }, + { + "epoch": 6.91, + "learning_rate": 9.602656426360246e-07, + "loss": 0.0794, + "step": 16498 + }, + { + "epoch": 6.91, + "learning_rate": 9.595408350169088e-07, + "loss": 0.0645, + "step": 16499 + }, + { + "epoch": 6.91, + "learning_rate": 9.588162872577688e-07, + "loss": 0.0823, + "step": 16500 + }, + { + "epoch": 6.91, + "learning_rate": 9.580919993794325e-07, + "loss": 0.1073, + "step": 16501 + }, + { + "epoch": 6.91, + "learning_rate": 9.573679714027173e-07, + "loss": 0.0848, + "step": 16502 + }, + { + "epoch": 6.91, + "learning_rate": 9.56644203348438e-07, + "loss": 0.0776, + "step": 16503 + }, + { + "epoch": 6.91, + "learning_rate": 9.559206952373933e-07, + "loss": 0.0622, + "step": 16504 + }, + { + "epoch": 6.91, + "learning_rate": 9.551974470903836e-07, + "loss": 0.098, + "step": 16505 + }, + { + "epoch": 6.91, + "learning_rate": 9.544744589281963e-07, + "loss": 0.0602, + "step": 16506 + }, + { + "epoch": 6.91, + "learning_rate": 9.53751730771615e-07, + "loss": 0.0916, + "step": 16507 + }, + { + "epoch": 6.91, + "learning_rate": 9.530292626414118e-07, + "loss": 0.0735, + "step": 16508 + }, + { + "epoch": 6.91, + "learning_rate": 9.523070545583535e-07, + "loss": 0.1054, + "step": 16509 + }, + { + "epoch": 6.91, + "learning_rate": 9.515851065432036e-07, + "loss": 0.0622, + "step": 16510 + }, + { + "epoch": 6.91, + "learning_rate": 9.508634186167098e-07, + "loss": 0.0743, + "step": 16511 + }, + { + "epoch": 6.91, + "learning_rate": 9.501419907996145e-07, + "loss": 0.0788, + "step": 16512 + }, + { + "epoch": 6.91, + "learning_rate": 9.494208231126578e-07, + "loss": 0.0596, + "step": 16513 + }, + { + "epoch": 6.91, + "learning_rate": 9.486999155765675e-07, + "loss": 0.0891, + "step": 16514 + }, + { + "epoch": 6.91, + "learning_rate": 9.47979268212067e-07, + "loss": 0.0679, + "step": 16515 + }, + { + "epoch": 6.91, + "learning_rate": 9.472588810398698e-07, + "loss": 0.0551, + "step": 16516 + }, + { + "epoch": 6.92, + "learning_rate": 9.465387540806836e-07, + "loss": 0.09, + "step": 16517 + }, + { + "epoch": 6.92, + "learning_rate": 9.458188873552054e-07, + "loss": 0.081, + "step": 16518 + }, + { + "epoch": 6.92, + "learning_rate": 9.450992808841297e-07, + "loss": 0.1218, + "step": 16519 + }, + { + "epoch": 6.92, + "learning_rate": 9.443799346881388e-07, + "loss": 0.0777, + "step": 16520 + }, + { + "epoch": 6.92, + "learning_rate": 9.436608487879118e-07, + "loss": 0.0809, + "step": 16521 + }, + { + "epoch": 6.92, + "learning_rate": 9.429420232041176e-07, + "loss": 0.089, + "step": 16522 + }, + { + "epoch": 6.92, + "learning_rate": 9.422234579574163e-07, + "loss": 0.0882, + "step": 16523 + }, + { + "epoch": 6.92, + "learning_rate": 9.415051530684671e-07, + "loss": 0.1084, + "step": 16524 + }, + { + "epoch": 6.92, + "learning_rate": 9.407871085579113e-07, + "loss": 0.0849, + "step": 16525 + }, + { + "epoch": 6.92, + "learning_rate": 9.400693244463921e-07, + "loss": 0.0915, + "step": 16526 + }, + { + "epoch": 6.92, + "learning_rate": 9.393518007545399e-07, + "loss": 0.0645, + "step": 16527 + }, + { + "epoch": 6.92, + "learning_rate": 9.386345375029793e-07, + "loss": 0.0911, + "step": 16528 + }, + { + "epoch": 6.92, + "learning_rate": 9.379175347123292e-07, + "loss": 0.0836, + "step": 16529 + }, + { + "epoch": 6.92, + "learning_rate": 9.372007924031989e-07, + "loss": 0.0984, + "step": 16530 + }, + { + "epoch": 6.92, + "learning_rate": 9.364843105961885e-07, + "loss": 0.0794, + "step": 16531 + }, + { + "epoch": 6.92, + "learning_rate": 9.357680893118937e-07, + "loss": 0.0728, + "step": 16532 + }, + { + "epoch": 6.92, + "learning_rate": 9.350521285709025e-07, + "loss": 0.0916, + "step": 16533 + }, + { + "epoch": 6.92, + "learning_rate": 9.343364283937928e-07, + "loss": 0.0867, + "step": 16534 + }, + { + "epoch": 6.92, + "learning_rate": 9.336209888011383e-07, + "loss": 0.0782, + "step": 16535 + }, + { + "epoch": 6.92, + "learning_rate": 9.329058098135024e-07, + "loss": 0.0911, + "step": 16536 + }, + { + "epoch": 6.92, + "learning_rate": 9.321908914514421e-07, + "loss": 0.0689, + "step": 16537 + }, + { + "epoch": 6.92, + "learning_rate": 9.31476233735511e-07, + "loss": 0.0549, + "step": 16538 + }, + { + "epoch": 6.92, + "learning_rate": 9.30761836686247e-07, + "loss": 0.0838, + "step": 16539 + }, + { + "epoch": 6.92, + "learning_rate": 9.300477003241848e-07, + "loss": 0.0678, + "step": 16540 + }, + { + "epoch": 6.93, + "learning_rate": 9.293338246698536e-07, + "loss": 0.0905, + "step": 16541 + }, + { + "epoch": 6.93, + "learning_rate": 9.286202097437725e-07, + "loss": 0.1143, + "step": 16542 + }, + { + "epoch": 6.93, + "learning_rate": 9.279068555664528e-07, + "loss": 0.0711, + "step": 16543 + }, + { + "epoch": 6.93, + "learning_rate": 9.271937621584026e-07, + "loss": 0.0915, + "step": 16544 + }, + { + "epoch": 6.93, + "learning_rate": 9.264809295401156e-07, + "loss": 0.0726, + "step": 16545 + }, + { + "epoch": 6.93, + "learning_rate": 9.25768357732082e-07, + "loss": 0.0738, + "step": 16546 + }, + { + "epoch": 6.93, + "learning_rate": 9.250560467547842e-07, + "loss": 0.101, + "step": 16547 + }, + { + "epoch": 6.93, + "learning_rate": 9.243439966286971e-07, + "loss": 0.0644, + "step": 16548 + }, + { + "epoch": 6.93, + "learning_rate": 9.236322073742887e-07, + "loss": 0.0633, + "step": 16549 + }, + { + "epoch": 6.93, + "learning_rate": 9.229206790120204e-07, + "loss": 0.0722, + "step": 16550 + }, + { + "epoch": 6.93, + "learning_rate": 9.222094115623404e-07, + "loss": 0.0666, + "step": 16551 + }, + { + "epoch": 6.93, + "learning_rate": 9.214984050456954e-07, + "loss": 0.0654, + "step": 16552 + }, + { + "epoch": 6.93, + "learning_rate": 9.207876594825216e-07, + "loss": 0.0772, + "step": 16553 + }, + { + "epoch": 6.93, + "learning_rate": 9.200771748932513e-07, + "loss": 0.0715, + "step": 16554 + }, + { + "epoch": 6.93, + "learning_rate": 9.193669512983038e-07, + "loss": 0.0896, + "step": 16555 + }, + { + "epoch": 6.93, + "learning_rate": 9.186569887180985e-07, + "loss": 0.0803, + "step": 16556 + }, + { + "epoch": 6.93, + "learning_rate": 9.179472871730366e-07, + "loss": 0.0934, + "step": 16557 + }, + { + "epoch": 6.93, + "learning_rate": 9.17237846683523e-07, + "loss": 0.1161, + "step": 16558 + }, + { + "epoch": 6.93, + "learning_rate": 9.165286672699459e-07, + "loss": 0.0756, + "step": 16559 + }, + { + "epoch": 6.93, + "learning_rate": 9.15819748952691e-07, + "loss": 0.0932, + "step": 16560 + }, + { + "epoch": 6.93, + "learning_rate": 9.151110917521366e-07, + "loss": 0.0653, + "step": 16561 + }, + { + "epoch": 6.93, + "learning_rate": 9.14402695688652e-07, + "loss": 0.08, + "step": 16562 + }, + { + "epoch": 6.93, + "learning_rate": 9.136945607825998e-07, + "loss": 0.076, + "step": 16563 + }, + { + "epoch": 6.93, + "learning_rate": 9.129866870543347e-07, + "loss": 0.0529, + "step": 16564 + }, + { + "epoch": 6.94, + "learning_rate": 9.122790745242049e-07, + "loss": 0.0748, + "step": 16565 + }, + { + "epoch": 6.94, + "learning_rate": 9.115717232125476e-07, + "loss": 0.065, + "step": 16566 + }, + { + "epoch": 6.94, + "learning_rate": 9.108646331396953e-07, + "loss": 0.0946, + "step": 16567 + }, + { + "epoch": 6.94, + "learning_rate": 9.10157804325974e-07, + "loss": 0.0623, + "step": 16568 + }, + { + "epoch": 6.94, + "learning_rate": 9.094512367916996e-07, + "loss": 0.0917, + "step": 16569 + }, + { + "epoch": 6.94, + "learning_rate": 9.087449305571838e-07, + "loss": 0.0906, + "step": 16570 + }, + { + "epoch": 6.94, + "learning_rate": 9.080388856427291e-07, + "loss": 0.0834, + "step": 16571 + }, + { + "epoch": 6.94, + "learning_rate": 9.07333102068626e-07, + "loss": 0.0968, + "step": 16572 + }, + { + "epoch": 6.94, + "learning_rate": 9.066275798551649e-07, + "loss": 0.0786, + "step": 16573 + }, + { + "epoch": 6.94, + "learning_rate": 9.059223190226251e-07, + "loss": 0.0709, + "step": 16574 + }, + { + "epoch": 6.94, + "learning_rate": 9.052173195912772e-07, + "loss": 0.0399, + "step": 16575 + }, + { + "epoch": 6.94, + "learning_rate": 9.045125815813883e-07, + "loss": 0.0838, + "step": 16576 + }, + { + "epoch": 6.94, + "learning_rate": 9.038081050132141e-07, + "loss": 0.0638, + "step": 16577 + }, + { + "epoch": 6.94, + "learning_rate": 9.031038899070033e-07, + "loss": 0.1018, + "step": 16578 + }, + { + "epoch": 6.94, + "learning_rate": 9.023999362830005e-07, + "loss": 0.0962, + "step": 16579 + }, + { + "epoch": 6.94, + "learning_rate": 9.016962441614363e-07, + "loss": 0.0945, + "step": 16580 + }, + { + "epoch": 6.94, + "learning_rate": 9.009928135625412e-07, + "loss": 0.0737, + "step": 16581 + }, + { + "epoch": 6.94, + "learning_rate": 9.002896445065312e-07, + "loss": 0.0624, + "step": 16582 + }, + { + "epoch": 6.94, + "learning_rate": 8.995867370136224e-07, + "loss": 0.0751, + "step": 16583 + }, + { + "epoch": 6.94, + "learning_rate": 8.988840911040165e-07, + "loss": 0.0876, + "step": 16584 + }, + { + "epoch": 6.94, + "learning_rate": 8.981817067979126e-07, + "loss": 0.087, + "step": 16585 + }, + { + "epoch": 6.94, + "learning_rate": 8.974795841154971e-07, + "loss": 0.1017, + "step": 16586 + }, + { + "epoch": 6.94, + "learning_rate": 8.967777230769537e-07, + "loss": 0.0746, + "step": 16587 + }, + { + "epoch": 6.94, + "learning_rate": 8.960761237024573e-07, + "loss": 0.0726, + "step": 16588 + }, + { + "epoch": 6.95, + "learning_rate": 8.953747860121731e-07, + "loss": 0.084, + "step": 16589 + }, + { + "epoch": 6.95, + "learning_rate": 8.946737100262614e-07, + "loss": 0.0726, + "step": 16590 + }, + { + "epoch": 6.95, + "learning_rate": 8.939728957648763e-07, + "loss": 0.0612, + "step": 16591 + }, + { + "epoch": 6.95, + "learning_rate": 8.932723432481572e-07, + "loss": 0.0952, + "step": 16592 + }, + { + "epoch": 6.95, + "learning_rate": 8.925720524962445e-07, + "loss": 0.0716, + "step": 16593 + }, + { + "epoch": 6.95, + "learning_rate": 8.918720235292655e-07, + "loss": 0.0679, + "step": 16594 + }, + { + "epoch": 6.95, + "learning_rate": 8.91172256367343e-07, + "loss": 0.0933, + "step": 16595 + }, + { + "epoch": 6.95, + "learning_rate": 8.90472751030591e-07, + "loss": 0.0544, + "step": 16596 + }, + { + "epoch": 6.95, + "learning_rate": 8.897735075391156e-07, + "loss": 0.0806, + "step": 16597 + }, + { + "epoch": 6.95, + "learning_rate": 8.890745259130196e-07, + "loss": 0.1142, + "step": 16598 + }, + { + "epoch": 6.95, + "learning_rate": 8.88375806172388e-07, + "loss": 0.0725, + "step": 16599 + }, + { + "epoch": 6.95, + "learning_rate": 8.876773483373091e-07, + "loss": 0.083, + "step": 16600 + }, + { + "epoch": 6.95, + "learning_rate": 8.869791524278604e-07, + "loss": 0.0751, + "step": 16601 + }, + { + "epoch": 6.95, + "learning_rate": 8.862812184641078e-07, + "loss": 0.0732, + "step": 16602 + }, + { + "epoch": 6.95, + "learning_rate": 8.855835464661145e-07, + "loss": 0.0782, + "step": 16603 + }, + { + "epoch": 6.95, + "learning_rate": 8.84886136453933e-07, + "loss": 0.0566, + "step": 16604 + }, + { + "epoch": 6.95, + "learning_rate": 8.841889884476118e-07, + "loss": 0.0722, + "step": 16605 + }, + { + "epoch": 6.95, + "learning_rate": 8.834921024671905e-07, + "loss": 0.0737, + "step": 16606 + }, + { + "epoch": 6.95, + "learning_rate": 8.827954785326975e-07, + "loss": 0.0725, + "step": 16607 + }, + { + "epoch": 6.95, + "learning_rate": 8.820991166641568e-07, + "loss": 0.0714, + "step": 16608 + }, + { + "epoch": 6.95, + "learning_rate": 8.814030168815868e-07, + "loss": 0.075, + "step": 16609 + }, + { + "epoch": 6.95, + "learning_rate": 8.807071792049948e-07, + "loss": 0.0669, + "step": 16610 + }, + { + "epoch": 6.95, + "learning_rate": 8.800116036543838e-07, + "loss": 0.1129, + "step": 16611 + }, + { + "epoch": 6.95, + "learning_rate": 8.793162902497465e-07, + "loss": 0.0735, + "step": 16612 + }, + { + "epoch": 6.96, + "learning_rate": 8.786212390110682e-07, + "loss": 0.0874, + "step": 16613 + }, + { + "epoch": 6.96, + "learning_rate": 8.779264499583284e-07, + "loss": 0.0807, + "step": 16614 + }, + { + "epoch": 6.96, + "learning_rate": 8.772319231114978e-07, + "loss": 0.0741, + "step": 16615 + }, + { + "epoch": 6.96, + "learning_rate": 8.765376584905395e-07, + "loss": 0.0974, + "step": 16616 + }, + { + "epoch": 6.96, + "learning_rate": 8.758436561154105e-07, + "loss": 0.0881, + "step": 16617 + }, + { + "epoch": 6.96, + "learning_rate": 8.751499160060605e-07, + "loss": 0.0807, + "step": 16618 + }, + { + "epoch": 6.96, + "learning_rate": 8.744564381824284e-07, + "loss": 0.1121, + "step": 16619 + }, + { + "epoch": 6.96, + "learning_rate": 8.737632226644465e-07, + "loss": 0.0894, + "step": 16620 + }, + { + "epoch": 6.96, + "learning_rate": 8.730702694720428e-07, + "loss": 0.0836, + "step": 16621 + }, + { + "epoch": 6.96, + "learning_rate": 8.723775786251354e-07, + "loss": 0.0837, + "step": 16622 + }, + { + "epoch": 6.96, + "learning_rate": 8.716851501436352e-07, + "loss": 0.0691, + "step": 16623 + }, + { + "epoch": 6.96, + "learning_rate": 8.709929840474463e-07, + "loss": 0.1011, + "step": 16624 + }, + { + "epoch": 6.96, + "learning_rate": 8.703010803564616e-07, + "loss": 0.0834, + "step": 16625 + }, + { + "epoch": 6.96, + "learning_rate": 8.696094390905718e-07, + "loss": 0.0848, + "step": 16626 + }, + { + "epoch": 6.96, + "learning_rate": 8.689180602696556e-07, + "loss": 0.1116, + "step": 16627 + }, + { + "epoch": 6.96, + "learning_rate": 8.682269439135871e-07, + "loss": 0.0851, + "step": 16628 + }, + { + "epoch": 6.96, + "learning_rate": 8.675360900422314e-07, + "loss": 0.081, + "step": 16629 + }, + { + "epoch": 6.96, + "learning_rate": 8.66845498675446e-07, + "loss": 0.0749, + "step": 16630 + }, + { + "epoch": 6.96, + "learning_rate": 8.661551698330828e-07, + "loss": 0.0705, + "step": 16631 + }, + { + "epoch": 6.96, + "learning_rate": 8.65465103534986e-07, + "loss": 0.0742, + "step": 16632 + }, + { + "epoch": 6.96, + "learning_rate": 8.647752998009873e-07, + "loss": 0.095, + "step": 16633 + }, + { + "epoch": 6.96, + "learning_rate": 8.640857586509155e-07, + "loss": 0.088, + "step": 16634 + }, + { + "epoch": 6.96, + "learning_rate": 8.633964801045913e-07, + "loss": 0.0972, + "step": 16635 + }, + { + "epoch": 6.97, + "learning_rate": 8.62707464181829e-07, + "loss": 0.1162, + "step": 16636 + }, + { + "epoch": 6.97, + "learning_rate": 8.620187109024314e-07, + "loss": 0.0883, + "step": 16637 + }, + { + "epoch": 6.97, + "learning_rate": 8.613302202861962e-07, + "loss": 0.092, + "step": 16638 + }, + { + "epoch": 6.97, + "learning_rate": 8.606419923529175e-07, + "loss": 0.0411, + "step": 16639 + }, + { + "epoch": 6.97, + "learning_rate": 8.599540271223728e-07, + "loss": 0.0708, + "step": 16640 + }, + { + "epoch": 6.97, + "learning_rate": 8.592663246143385e-07, + "loss": 0.0768, + "step": 16641 + }, + { + "epoch": 6.97, + "learning_rate": 8.58578884848582e-07, + "loss": 0.1036, + "step": 16642 + }, + { + "epoch": 6.97, + "learning_rate": 8.578917078448634e-07, + "loss": 0.09, + "step": 16643 + }, + { + "epoch": 6.97, + "learning_rate": 8.572047936229355e-07, + "loss": 0.1143, + "step": 16644 + }, + { + "epoch": 6.97, + "learning_rate": 8.565181422025448e-07, + "loss": 0.0634, + "step": 16645 + }, + { + "epoch": 6.97, + "learning_rate": 8.558317536034233e-07, + "loss": 0.0729, + "step": 16646 + }, + { + "epoch": 6.97, + "learning_rate": 8.551456278453063e-07, + "loss": 0.0873, + "step": 16647 + }, + { + "epoch": 6.97, + "learning_rate": 8.544597649479114e-07, + "loss": 0.0769, + "step": 16648 + }, + { + "epoch": 6.97, + "learning_rate": 8.53774164930955e-07, + "loss": 0.0891, + "step": 16649 + }, + { + "epoch": 6.97, + "learning_rate": 8.530888278141436e-07, + "loss": 0.0774, + "step": 16650 + }, + { + "epoch": 6.97, + "learning_rate": 8.524037536171781e-07, + "loss": 0.0801, + "step": 16651 + }, + { + "epoch": 6.97, + "learning_rate": 8.517189423597483e-07, + "loss": 0.0774, + "step": 16652 + }, + { + "epoch": 6.97, + "learning_rate": 8.510343940615417e-07, + "loss": 0.0767, + "step": 16653 + }, + { + "epoch": 6.97, + "learning_rate": 8.503501087422305e-07, + "loss": 0.0944, + "step": 16654 + }, + { + "epoch": 6.97, + "learning_rate": 8.496660864214868e-07, + "loss": 0.0833, + "step": 16655 + }, + { + "epoch": 6.97, + "learning_rate": 8.489823271189712e-07, + "loss": 0.0941, + "step": 16656 + }, + { + "epoch": 6.97, + "learning_rate": 8.482988308543383e-07, + "loss": 0.0652, + "step": 16657 + }, + { + "epoch": 6.97, + "learning_rate": 8.476155976472345e-07, + "loss": 0.0684, + "step": 16658 + }, + { + "epoch": 6.97, + "learning_rate": 8.469326275172995e-07, + "loss": 0.0905, + "step": 16659 + }, + { + "epoch": 6.98, + "learning_rate": 8.462499204841635e-07, + "loss": 0.0842, + "step": 16660 + }, + { + "epoch": 6.98, + "learning_rate": 8.455674765674493e-07, + "loss": 0.0685, + "step": 16661 + }, + { + "epoch": 6.98, + "learning_rate": 8.448852957867748e-07, + "loss": 0.0514, + "step": 16662 + }, + { + "epoch": 6.98, + "learning_rate": 8.442033781617486e-07, + "loss": 0.1013, + "step": 16663 + }, + { + "epoch": 6.98, + "learning_rate": 8.435217237119719e-07, + "loss": 0.0819, + "step": 16664 + }, + { + "epoch": 6.98, + "learning_rate": 8.428403324570378e-07, + "loss": 0.0854, + "step": 16665 + }, + { + "epoch": 6.98, + "learning_rate": 8.421592044165338e-07, + "loss": 0.0651, + "step": 16666 + }, + { + "epoch": 6.98, + "learning_rate": 8.414783396100357e-07, + "loss": 0.0824, + "step": 16667 + }, + { + "epoch": 6.98, + "learning_rate": 8.407977380571153e-07, + "loss": 0.0874, + "step": 16668 + }, + { + "epoch": 6.98, + "learning_rate": 8.401173997773381e-07, + "loss": 0.0745, + "step": 16669 + }, + { + "epoch": 6.98, + "learning_rate": 8.394373247902565e-07, + "loss": 0.0691, + "step": 16670 + }, + { + "epoch": 6.98, + "learning_rate": 8.387575131154202e-07, + "loss": 0.0817, + "step": 16671 + }, + { + "epoch": 6.98, + "learning_rate": 8.380779647723692e-07, + "loss": 0.0862, + "step": 16672 + }, + { + "epoch": 6.98, + "learning_rate": 8.373986797806377e-07, + "loss": 0.0664, + "step": 16673 + }, + { + "epoch": 6.98, + "learning_rate": 8.367196581597503e-07, + "loss": 0.0629, + "step": 16674 + }, + { + "epoch": 6.98, + "learning_rate": 8.360408999292236e-07, + "loss": 0.0735, + "step": 16675 + }, + { + "epoch": 6.98, + "learning_rate": 8.353624051085696e-07, + "loss": 0.0621, + "step": 16676 + }, + { + "epoch": 6.98, + "learning_rate": 8.346841737172917e-07, + "loss": 0.0812, + "step": 16677 + }, + { + "epoch": 6.98, + "learning_rate": 8.340062057748832e-07, + "loss": 0.0491, + "step": 16678 + }, + { + "epoch": 6.98, + "learning_rate": 8.333285013008329e-07, + "loss": 0.1196, + "step": 16679 + }, + { + "epoch": 6.98, + "learning_rate": 8.32651060314622e-07, + "loss": 0.0909, + "step": 16680 + }, + { + "epoch": 6.98, + "learning_rate": 8.319738828357193e-07, + "loss": 0.0886, + "step": 16681 + }, + { + "epoch": 6.98, + "learning_rate": 8.312969688835926e-07, + "loss": 0.0663, + "step": 16682 + }, + { + "epoch": 6.98, + "learning_rate": 8.306203184776985e-07, + "loss": 0.0835, + "step": 16683 + }, + { + "epoch": 6.99, + "learning_rate": 8.299439316374858e-07, + "loss": 0.1121, + "step": 16684 + }, + { + "epoch": 6.99, + "learning_rate": 8.29267808382398e-07, + "loss": 0.085, + "step": 16685 + }, + { + "epoch": 6.99, + "learning_rate": 8.285919487318694e-07, + "loss": 0.0473, + "step": 16686 + }, + { + "epoch": 6.99, + "learning_rate": 8.279163527053268e-07, + "loss": 0.0749, + "step": 16687 + }, + { + "epoch": 6.99, + "learning_rate": 8.272410203221881e-07, + "loss": 0.0831, + "step": 16688 + }, + { + "epoch": 6.99, + "learning_rate": 8.265659516018676e-07, + "loss": 0.0757, + "step": 16689 + }, + { + "epoch": 6.99, + "learning_rate": 8.258911465637675e-07, + "loss": 0.1019, + "step": 16690 + }, + { + "epoch": 6.99, + "learning_rate": 8.252166052272859e-07, + "loss": 0.0633, + "step": 16691 + }, + { + "epoch": 6.99, + "learning_rate": 8.245423276118136e-07, + "loss": 0.0781, + "step": 16692 + }, + { + "epoch": 6.99, + "learning_rate": 8.238683137367276e-07, + "loss": 0.0457, + "step": 16693 + }, + { + "epoch": 6.99, + "learning_rate": 8.231945636214068e-07, + "loss": 0.065, + "step": 16694 + }, + { + "epoch": 6.99, + "learning_rate": 8.225210772852122e-07, + "loss": 0.0584, + "step": 16695 + }, + { + "epoch": 6.99, + "learning_rate": 8.218478547475051e-07, + "loss": 0.0754, + "step": 16696 + }, + { + "epoch": 6.99, + "learning_rate": 8.211748960276378e-07, + "loss": 0.094, + "step": 16697 + }, + { + "epoch": 6.99, + "learning_rate": 8.205022011449526e-07, + "loss": 0.0867, + "step": 16698 + }, + { + "epoch": 6.99, + "learning_rate": 8.198297701187852e-07, + "loss": 0.0904, + "step": 16699 + }, + { + "epoch": 6.99, + "learning_rate": 8.191576029684667e-07, + "loss": 0.1052, + "step": 16700 + }, + { + "epoch": 6.99, + "learning_rate": 8.18485699713315e-07, + "loss": 0.0751, + "step": 16701 + }, + { + "epoch": 6.99, + "learning_rate": 8.178140603726437e-07, + "loss": 0.0655, + "step": 16702 + }, + { + "epoch": 6.99, + "learning_rate": 8.171426849657594e-07, + "loss": 0.0832, + "step": 16703 + }, + { + "epoch": 6.99, + "learning_rate": 8.164715735119599e-07, + "loss": 0.0788, + "step": 16704 + }, + { + "epoch": 6.99, + "learning_rate": 8.158007260305356e-07, + "loss": 0.0843, + "step": 16705 + }, + { + "epoch": 6.99, + "learning_rate": 8.151301425407699e-07, + "loss": 0.1059, + "step": 16706 + }, + { + "epoch": 6.99, + "learning_rate": 8.144598230619383e-07, + "loss": 0.0749, + "step": 16707 + }, + { + "epoch": 7.0, + "learning_rate": 8.137897676133077e-07, + "loss": 0.0613, + "step": 16708 + }, + { + "epoch": 7.0, + "learning_rate": 8.131199762141373e-07, + "loss": 0.0587, + "step": 16709 + }, + { + "epoch": 7.0, + "learning_rate": 8.124504488836815e-07, + "loss": 0.0732, + "step": 16710 + }, + { + "epoch": 7.0, + "learning_rate": 8.11781185641185e-07, + "loss": 0.0801, + "step": 16711 + }, + { + "epoch": 7.0, + "learning_rate": 8.111121865058846e-07, + "loss": 0.0597, + "step": 16712 + }, + { + "epoch": 7.0, + "learning_rate": 8.104434514970117e-07, + "loss": 0.0654, + "step": 16713 + }, + { + "epoch": 7.0, + "learning_rate": 8.097749806337851e-07, + "loss": 0.0664, + "step": 16714 + }, + { + "epoch": 7.0, + "learning_rate": 8.091067739354241e-07, + "loss": 0.0875, + "step": 16715 + }, + { + "epoch": 7.0, + "learning_rate": 8.084388314211311e-07, + "loss": 0.0907, + "step": 16716 + }, + { + "epoch": 7.0, + "learning_rate": 8.077711531101074e-07, + "loss": 0.0948, + "step": 16717 + }, + { + "epoch": 7.0, + "learning_rate": 8.071037390215442e-07, + "loss": 0.1112, + "step": 16718 + }, + { + "epoch": 7.0, + "learning_rate": 8.064365891746273e-07, + "loss": 0.0912, + "step": 16719 + }, + { + "epoch": 7.0, + "learning_rate": 8.057697035885315e-07, + "loss": 0.0744, + "step": 16720 + }, + { + "epoch": 7.0, + "learning_rate": 8.05103082282429e-07, + "loss": 0.0586, + "step": 16721 + }, + { + "epoch": 7.0, + "learning_rate": 8.04436725275477e-07, + "loss": 0.0533, + "step": 16722 + }, + { + "epoch": 7.0, + "learning_rate": 8.037706325868322e-07, + "loss": 0.0551, + "step": 16723 + }, + { + "epoch": 7.0, + "learning_rate": 8.031048042356393e-07, + "loss": 0.0567, + "step": 16724 + }, + { + "epoch": 7.0, + "learning_rate": 8.024392402410375e-07, + "loss": 0.059, + "step": 16725 + }, + { + "epoch": 7.0, + "learning_rate": 8.017739406221581e-07, + "loss": 0.039, + "step": 16726 + }, + { + "epoch": 7.0, + "learning_rate": 8.011089053981258e-07, + "loss": 0.0584, + "step": 16727 + }, + { + "epoch": 7.0, + "learning_rate": 8.004441345880542e-07, + "loss": 0.0674, + "step": 16728 + }, + { + "epoch": 7.0, + "learning_rate": 7.997796282110514e-07, + "loss": 0.0421, + "step": 16729 + }, + { + "epoch": 7.0, + "learning_rate": 7.9911538628622e-07, + "loss": 0.0638, + "step": 16730 + }, + { + "epoch": 7.0, + "learning_rate": 7.984514088326512e-07, + "loss": 0.0644, + "step": 16731 + }, + { + "epoch": 7.01, + "learning_rate": 7.977876958694319e-07, + "loss": 0.0453, + "step": 16732 + }, + { + "epoch": 7.01, + "learning_rate": 7.971242474156393e-07, + "loss": 0.0523, + "step": 16733 + }, + { + "epoch": 7.01, + "learning_rate": 7.964610634903458e-07, + "loss": 0.0488, + "step": 16734 + }, + { + "epoch": 7.01, + "learning_rate": 7.957981441126094e-07, + "loss": 0.0442, + "step": 16735 + }, + { + "epoch": 7.01, + "learning_rate": 7.951354893014885e-07, + "loss": 0.0343, + "step": 16736 + }, + { + "epoch": 7.01, + "learning_rate": 7.944730990760307e-07, + "loss": 0.0534, + "step": 16737 + }, + { + "epoch": 7.01, + "learning_rate": 7.938109734552735e-07, + "loss": 0.0261, + "step": 16738 + }, + { + "epoch": 7.01, + "learning_rate": 7.931491124582503e-07, + "loss": 0.0428, + "step": 16739 + }, + { + "epoch": 7.01, + "learning_rate": 7.924875161039869e-07, + "loss": 0.0633, + "step": 16740 + }, + { + "epoch": 7.01, + "learning_rate": 7.918261844114994e-07, + "loss": 0.0419, + "step": 16741 + }, + { + "epoch": 7.01, + "learning_rate": 7.911651173997959e-07, + "loss": 0.0469, + "step": 16742 + }, + { + "epoch": 7.01, + "learning_rate": 7.9050431508788e-07, + "loss": 0.0343, + "step": 16743 + }, + { + "epoch": 7.01, + "learning_rate": 7.898437774947442e-07, + "loss": 0.0416, + "step": 16744 + }, + { + "epoch": 7.01, + "learning_rate": 7.89183504639377e-07, + "loss": 0.0553, + "step": 16745 + }, + { + "epoch": 7.01, + "learning_rate": 7.885234965407551e-07, + "loss": 0.0456, + "step": 16746 + }, + { + "epoch": 7.01, + "learning_rate": 7.878637532178523e-07, + "loss": 0.0493, + "step": 16747 + }, + { + "epoch": 7.01, + "learning_rate": 7.872042746896324e-07, + "loss": 0.0584, + "step": 16748 + }, + { + "epoch": 7.01, + "learning_rate": 7.865450609750492e-07, + "loss": 0.0441, + "step": 16749 + }, + { + "epoch": 7.01, + "learning_rate": 7.858861120930516e-07, + "loss": 0.069, + "step": 16750 + }, + { + "epoch": 7.01, + "learning_rate": 7.852274280625815e-07, + "loss": 0.0656, + "step": 16751 + }, + { + "epoch": 7.01, + "learning_rate": 7.845690089025726e-07, + "loss": 0.0386, + "step": 16752 + }, + { + "epoch": 7.01, + "learning_rate": 7.839108546319485e-07, + "loss": 0.0722, + "step": 16753 + }, + { + "epoch": 7.01, + "learning_rate": 7.832529652696319e-07, + "loss": 0.0487, + "step": 16754 + }, + { + "epoch": 7.01, + "learning_rate": 7.825953408345277e-07, + "loss": 0.049, + "step": 16755 + }, + { + "epoch": 7.02, + "learning_rate": 7.819379813455407e-07, + "loss": 0.0345, + "step": 16756 + }, + { + "epoch": 7.02, + "learning_rate": 7.812808868215671e-07, + "loss": 0.0533, + "step": 16757 + }, + { + "epoch": 7.02, + "learning_rate": 7.806240572814927e-07, + "loss": 0.0435, + "step": 16758 + }, + { + "epoch": 7.02, + "learning_rate": 7.799674927442003e-07, + "loss": 0.0526, + "step": 16759 + }, + { + "epoch": 7.02, + "learning_rate": 7.793111932285613e-07, + "loss": 0.032, + "step": 16760 + }, + { + "epoch": 7.02, + "learning_rate": 7.786551587534386e-07, + "loss": 0.0518, + "step": 16761 + }, + { + "epoch": 7.02, + "learning_rate": 7.779993893376914e-07, + "loss": 0.0535, + "step": 16762 + }, + { + "epoch": 7.02, + "learning_rate": 7.773438850001669e-07, + "loss": 0.0357, + "step": 16763 + }, + { + "epoch": 7.02, + "learning_rate": 7.766886457597089e-07, + "loss": 0.0406, + "step": 16764 + }, + { + "epoch": 7.02, + "learning_rate": 7.760336716351502e-07, + "loss": 0.0399, + "step": 16765 + }, + { + "epoch": 7.02, + "learning_rate": 7.75378962645319e-07, + "loss": 0.0497, + "step": 16766 + }, + { + "epoch": 7.02, + "learning_rate": 7.747245188090335e-07, + "loss": 0.0407, + "step": 16767 + }, + { + "epoch": 7.02, + "learning_rate": 7.740703401451078e-07, + "loss": 0.0417, + "step": 16768 + }, + { + "epoch": 7.02, + "learning_rate": 7.734164266723409e-07, + "loss": 0.0706, + "step": 16769 + }, + { + "epoch": 7.02, + "learning_rate": 7.727627784095315e-07, + "loss": 0.0491, + "step": 16770 + }, + { + "epoch": 7.02, + "learning_rate": 7.721093953754677e-07, + "loss": 0.0575, + "step": 16771 + }, + { + "epoch": 7.02, + "learning_rate": 7.714562775889301e-07, + "loss": 0.0509, + "step": 16772 + }, + { + "epoch": 7.02, + "learning_rate": 7.708034250686936e-07, + "loss": 0.0662, + "step": 16773 + }, + { + "epoch": 7.02, + "learning_rate": 7.701508378335243e-07, + "loss": 0.0647, + "step": 16774 + }, + { + "epoch": 7.02, + "learning_rate": 7.694985159021762e-07, + "loss": 0.0538, + "step": 16775 + }, + { + "epoch": 7.02, + "learning_rate": 7.68846459293403e-07, + "loss": 0.0805, + "step": 16776 + }, + { + "epoch": 7.02, + "learning_rate": 7.681946680259467e-07, + "loss": 0.043, + "step": 16777 + }, + { + "epoch": 7.02, + "learning_rate": 7.675431421185431e-07, + "loss": 0.0544, + "step": 16778 + }, + { + "epoch": 7.02, + "learning_rate": 7.668918815899184e-07, + "loss": 0.0521, + "step": 16779 + }, + { + "epoch": 7.03, + "learning_rate": 7.662408864587933e-07, + "loss": 0.0247, + "step": 16780 + }, + { + "epoch": 7.03, + "learning_rate": 7.655901567438817e-07, + "loss": 0.0483, + "step": 16781 + }, + { + "epoch": 7.03, + "learning_rate": 7.649396924638841e-07, + "loss": 0.0487, + "step": 16782 + }, + { + "epoch": 7.03, + "learning_rate": 7.642894936375034e-07, + "loss": 0.0609, + "step": 16783 + }, + { + "epoch": 7.03, + "learning_rate": 7.636395602834223e-07, + "loss": 0.0555, + "step": 16784 + }, + { + "epoch": 7.03, + "learning_rate": 7.629898924203272e-07, + "loss": 0.04, + "step": 16785 + }, + { + "epoch": 7.03, + "learning_rate": 7.623404900668896e-07, + "loss": 0.0494, + "step": 16786 + }, + { + "epoch": 7.03, + "learning_rate": 7.616913532417781e-07, + "loss": 0.039, + "step": 16787 + }, + { + "epoch": 7.03, + "learning_rate": 7.610424819636508e-07, + "loss": 0.0592, + "step": 16788 + }, + { + "epoch": 7.03, + "learning_rate": 7.603938762511597e-07, + "loss": 0.0331, + "step": 16789 + }, + { + "epoch": 7.03, + "learning_rate": 7.597455361229466e-07, + "loss": 0.0446, + "step": 16790 + }, + { + "epoch": 7.03, + "learning_rate": 7.590974615976477e-07, + "loss": 0.0436, + "step": 16791 + }, + { + "epoch": 7.03, + "learning_rate": 7.584496526938933e-07, + "loss": 0.0654, + "step": 16792 + }, + { + "epoch": 7.03, + "learning_rate": 7.578021094303012e-07, + "loss": 0.0466, + "step": 16793 + }, + { + "epoch": 7.03, + "learning_rate": 7.571548318254874e-07, + "loss": 0.0442, + "step": 16794 + }, + { + "epoch": 7.03, + "learning_rate": 7.565078198980569e-07, + "loss": 0.0503, + "step": 16795 + }, + { + "epoch": 7.03, + "learning_rate": 7.558610736666061e-07, + "loss": 0.0542, + "step": 16796 + }, + { + "epoch": 7.03, + "learning_rate": 7.552145931497246e-07, + "loss": 0.0584, + "step": 16797 + }, + { + "epoch": 7.03, + "learning_rate": 7.545683783659963e-07, + "loss": 0.0533, + "step": 16798 + }, + { + "epoch": 7.03, + "learning_rate": 7.539224293339953e-07, + "loss": 0.0321, + "step": 16799 + }, + { + "epoch": 7.03, + "learning_rate": 7.53276746072289e-07, + "loss": 0.058, + "step": 16800 + }, + { + "epoch": 7.03, + "learning_rate": 7.526313285994391e-07, + "loss": 0.0487, + "step": 16801 + }, + { + "epoch": 7.03, + "learning_rate": 7.519861769339942e-07, + "loss": 0.0787, + "step": 16802 + }, + { + "epoch": 7.03, + "learning_rate": 7.513412910945006e-07, + "loss": 0.0295, + "step": 16803 + }, + { + "epoch": 7.04, + "learning_rate": 7.506966710994945e-07, + "loss": 0.0386, + "step": 16804 + }, + { + "epoch": 7.04, + "learning_rate": 7.500523169675056e-07, + "loss": 0.0329, + "step": 16805 + }, + { + "epoch": 7.04, + "learning_rate": 7.494082287170535e-07, + "loss": 0.044, + "step": 16806 + }, + { + "epoch": 7.04, + "learning_rate": 7.487644063666532e-07, + "loss": 0.0363, + "step": 16807 + }, + { + "epoch": 7.04, + "learning_rate": 7.481208499348102e-07, + "loss": 0.0662, + "step": 16808 + }, + { + "epoch": 7.04, + "learning_rate": 7.474775594400252e-07, + "loss": 0.0397, + "step": 16809 + }, + { + "epoch": 7.04, + "learning_rate": 7.468345349007855e-07, + "loss": 0.0364, + "step": 16810 + }, + { + "epoch": 7.04, + "learning_rate": 7.461917763355764e-07, + "loss": 0.0466, + "step": 16811 + }, + { + "epoch": 7.04, + "learning_rate": 7.455492837628731e-07, + "loss": 0.0587, + "step": 16812 + }, + { + "epoch": 7.04, + "learning_rate": 7.449070572011429e-07, + "loss": 0.0536, + "step": 16813 + }, + { + "epoch": 7.04, + "learning_rate": 7.442650966688469e-07, + "loss": 0.0586, + "step": 16814 + }, + { + "epoch": 7.04, + "learning_rate": 7.43623402184438e-07, + "loss": 0.0411, + "step": 16815 + }, + { + "epoch": 7.04, + "learning_rate": 7.429819737663602e-07, + "loss": 0.0405, + "step": 16816 + }, + { + "epoch": 7.04, + "learning_rate": 7.423408114330499e-07, + "loss": 0.0509, + "step": 16817 + }, + { + "epoch": 7.04, + "learning_rate": 7.41699915202938e-07, + "loss": 0.0519, + "step": 16818 + }, + { + "epoch": 7.04, + "learning_rate": 7.410592850944476e-07, + "loss": 0.0512, + "step": 16819 + }, + { + "epoch": 7.04, + "learning_rate": 7.404189211259905e-07, + "loss": 0.0485, + "step": 16820 + }, + { + "epoch": 7.04, + "learning_rate": 7.397788233159753e-07, + "loss": 0.0434, + "step": 16821 + }, + { + "epoch": 7.04, + "learning_rate": 7.391389916828018e-07, + "loss": 0.0478, + "step": 16822 + }, + { + "epoch": 7.04, + "learning_rate": 7.384994262448586e-07, + "loss": 0.0408, + "step": 16823 + }, + { + "epoch": 7.04, + "learning_rate": 7.378601270205299e-07, + "loss": 0.0521, + "step": 16824 + }, + { + "epoch": 7.04, + "learning_rate": 7.372210940281943e-07, + "loss": 0.0469, + "step": 16825 + }, + { + "epoch": 7.04, + "learning_rate": 7.365823272862183e-07, + "loss": 0.056, + "step": 16826 + }, + { + "epoch": 7.05, + "learning_rate": 7.359438268129626e-07, + "loss": 0.0539, + "step": 16827 + }, + { + "epoch": 7.05, + "learning_rate": 7.353055926267827e-07, + "loss": 0.0601, + "step": 16828 + }, + { + "epoch": 7.05, + "learning_rate": 7.346676247460205e-07, + "loss": 0.0615, + "step": 16829 + }, + { + "epoch": 7.05, + "learning_rate": 7.340299231890169e-07, + "loss": 0.0734, + "step": 16830 + }, + { + "epoch": 7.05, + "learning_rate": 7.333924879740994e-07, + "loss": 0.0506, + "step": 16831 + }, + { + "epoch": 7.05, + "learning_rate": 7.327553191195913e-07, + "loss": 0.0775, + "step": 16832 + }, + { + "epoch": 7.05, + "learning_rate": 7.321184166438089e-07, + "loss": 0.0427, + "step": 16833 + }, + { + "epoch": 7.05, + "learning_rate": 7.314817805650576e-07, + "loss": 0.0436, + "step": 16834 + }, + { + "epoch": 7.05, + "learning_rate": 7.308454109016383e-07, + "loss": 0.063, + "step": 16835 + }, + { + "epoch": 7.05, + "learning_rate": 7.302093076718442e-07, + "loss": 0.0398, + "step": 16836 + }, + { + "epoch": 7.05, + "learning_rate": 7.295734708939551e-07, + "loss": 0.062, + "step": 16837 + }, + { + "epoch": 7.05, + "learning_rate": 7.289379005862507e-07, + "loss": 0.0487, + "step": 16838 + }, + { + "epoch": 7.05, + "learning_rate": 7.283025967669998e-07, + "loss": 0.0399, + "step": 16839 + }, + { + "epoch": 7.05, + "learning_rate": 7.276675594544624e-07, + "loss": 0.0421, + "step": 16840 + }, + { + "epoch": 7.05, + "learning_rate": 7.270327886668937e-07, + "loss": 0.0248, + "step": 16841 + }, + { + "epoch": 7.05, + "learning_rate": 7.263982844225393e-07, + "loss": 0.0509, + "step": 16842 + }, + { + "epoch": 7.05, + "learning_rate": 7.257640467396365e-07, + "loss": 0.0543, + "step": 16843 + }, + { + "epoch": 7.05, + "learning_rate": 7.251300756364154e-07, + "loss": 0.0344, + "step": 16844 + }, + { + "epoch": 7.05, + "learning_rate": 7.244963711311003e-07, + "loss": 0.0437, + "step": 16845 + }, + { + "epoch": 7.05, + "learning_rate": 7.238629332419068e-07, + "loss": 0.0437, + "step": 16846 + }, + { + "epoch": 7.05, + "learning_rate": 7.23229761987041e-07, + "loss": 0.059, + "step": 16847 + }, + { + "epoch": 7.05, + "learning_rate": 7.225968573847031e-07, + "loss": 0.0455, + "step": 16848 + }, + { + "epoch": 7.05, + "learning_rate": 7.219642194530885e-07, + "loss": 0.0446, + "step": 16849 + }, + { + "epoch": 7.05, + "learning_rate": 7.21331848210377e-07, + "loss": 0.0713, + "step": 16850 + }, + { + "epoch": 7.06, + "learning_rate": 7.206997436747498e-07, + "loss": 0.0498, + "step": 16851 + }, + { + "epoch": 7.06, + "learning_rate": 7.200679058643733e-07, + "loss": 0.0443, + "step": 16852 + }, + { + "epoch": 7.06, + "learning_rate": 7.194363347974087e-07, + "loss": 0.0665, + "step": 16853 + }, + { + "epoch": 7.06, + "learning_rate": 7.188050304920113e-07, + "loss": 0.0543, + "step": 16854 + }, + { + "epoch": 7.06, + "learning_rate": 7.181739929663279e-07, + "loss": 0.0494, + "step": 16855 + }, + { + "epoch": 7.06, + "learning_rate": 7.175432222384981e-07, + "loss": 0.0478, + "step": 16856 + }, + { + "epoch": 7.06, + "learning_rate": 7.169127183266489e-07, + "loss": 0.0669, + "step": 16857 + }, + { + "epoch": 7.06, + "learning_rate": 7.162824812489067e-07, + "loss": 0.0413, + "step": 16858 + }, + { + "epoch": 7.06, + "learning_rate": 7.15652511023387e-07, + "loss": 0.0468, + "step": 16859 + }, + { + "epoch": 7.06, + "learning_rate": 7.150228076681954e-07, + "loss": 0.0416, + "step": 16860 + }, + { + "epoch": 7.06, + "learning_rate": 7.14393371201435e-07, + "loss": 0.0476, + "step": 16861 + }, + { + "epoch": 7.06, + "learning_rate": 7.137642016411972e-07, + "loss": 0.0451, + "step": 16862 + }, + { + "epoch": 7.06, + "learning_rate": 7.131352990055684e-07, + "loss": 0.0461, + "step": 16863 + }, + { + "epoch": 7.06, + "learning_rate": 7.125066633126221e-07, + "loss": 0.0596, + "step": 16864 + }, + { + "epoch": 7.06, + "learning_rate": 7.118782945804315e-07, + "loss": 0.0413, + "step": 16865 + }, + { + "epoch": 7.06, + "learning_rate": 7.112501928270565e-07, + "loss": 0.0538, + "step": 16866 + }, + { + "epoch": 7.06, + "learning_rate": 7.106223580705518e-07, + "loss": 0.0509, + "step": 16867 + }, + { + "epoch": 7.06, + "learning_rate": 7.099947903289639e-07, + "loss": 0.0664, + "step": 16868 + }, + { + "epoch": 7.06, + "learning_rate": 7.09367489620334e-07, + "loss": 0.0486, + "step": 16869 + }, + { + "epoch": 7.06, + "learning_rate": 7.087404559626898e-07, + "loss": 0.0577, + "step": 16870 + }, + { + "epoch": 7.06, + "learning_rate": 7.08113689374057e-07, + "loss": 0.0703, + "step": 16871 + }, + { + "epoch": 7.06, + "learning_rate": 7.074871898724489e-07, + "loss": 0.0454, + "step": 16872 + }, + { + "epoch": 7.06, + "learning_rate": 7.068609574758789e-07, + "loss": 0.045, + "step": 16873 + }, + { + "epoch": 7.06, + "learning_rate": 7.062349922023414e-07, + "loss": 0.0393, + "step": 16874 + }, + { + "epoch": 7.07, + "learning_rate": 7.056092940698323e-07, + "loss": 0.0451, + "step": 16875 + }, + { + "epoch": 7.07, + "learning_rate": 7.049838630963357e-07, + "loss": 0.0395, + "step": 16876 + }, + { + "epoch": 7.07, + "learning_rate": 7.04358699299832e-07, + "loss": 0.047, + "step": 16877 + }, + { + "epoch": 7.07, + "learning_rate": 7.037338026982877e-07, + "loss": 0.0573, + "step": 16878 + }, + { + "epoch": 7.07, + "learning_rate": 7.03109173309664e-07, + "loss": 0.0497, + "step": 16879 + }, + { + "epoch": 7.07, + "learning_rate": 7.024848111519189e-07, + "loss": 0.0619, + "step": 16880 + }, + { + "epoch": 7.07, + "learning_rate": 7.018607162429969e-07, + "loss": 0.0582, + "step": 16881 + }, + { + "epoch": 7.07, + "learning_rate": 7.012368886008369e-07, + "loss": 0.0479, + "step": 16882 + }, + { + "epoch": 7.07, + "learning_rate": 7.006133282433736e-07, + "loss": 0.0595, + "step": 16883 + }, + { + "epoch": 7.07, + "learning_rate": 6.999900351885258e-07, + "loss": 0.0317, + "step": 16884 + }, + { + "epoch": 7.07, + "learning_rate": 6.993670094542126e-07, + "loss": 0.0367, + "step": 16885 + }, + { + "epoch": 7.07, + "learning_rate": 6.987442510583409e-07, + "loss": 0.0383, + "step": 16886 + }, + { + "epoch": 7.07, + "learning_rate": 6.981217600188117e-07, + "loss": 0.0412, + "step": 16887 + }, + { + "epoch": 7.07, + "learning_rate": 6.974995363535186e-07, + "loss": 0.0623, + "step": 16888 + }, + { + "epoch": 7.07, + "learning_rate": 6.968775800803462e-07, + "loss": 0.0444, + "step": 16889 + }, + { + "epoch": 7.07, + "learning_rate": 6.962558912171735e-07, + "loss": 0.0342, + "step": 16890 + }, + { + "epoch": 7.07, + "learning_rate": 6.956344697818684e-07, + "loss": 0.0453, + "step": 16891 + }, + { + "epoch": 7.07, + "learning_rate": 6.950133157922934e-07, + "loss": 0.0417, + "step": 16892 + }, + { + "epoch": 7.07, + "learning_rate": 6.94392429266304e-07, + "loss": 0.0654, + "step": 16893 + }, + { + "epoch": 7.07, + "learning_rate": 6.937718102217461e-07, + "loss": 0.0427, + "step": 16894 + }, + { + "epoch": 7.07, + "learning_rate": 6.931514586764587e-07, + "loss": 0.0379, + "step": 16895 + }, + { + "epoch": 7.07, + "learning_rate": 6.925313746482765e-07, + "loss": 0.0404, + "step": 16896 + }, + { + "epoch": 7.07, + "learning_rate": 6.919115581550207e-07, + "loss": 0.0481, + "step": 16897 + }, + { + "epoch": 7.07, + "learning_rate": 6.912920092145048e-07, + "loss": 0.0624, + "step": 16898 + }, + { + "epoch": 7.08, + "learning_rate": 6.906727278445402e-07, + "loss": 0.0395, + "step": 16899 + }, + { + "epoch": 7.08, + "learning_rate": 6.900537140629282e-07, + "loss": 0.0761, + "step": 16900 + }, + { + "epoch": 7.08, + "learning_rate": 6.89434967887459e-07, + "loss": 0.0683, + "step": 16901 + }, + { + "epoch": 7.08, + "learning_rate": 6.888164893359206e-07, + "loss": 0.0487, + "step": 16902 + }, + { + "epoch": 7.08, + "learning_rate": 6.881982784260887e-07, + "loss": 0.0503, + "step": 16903 + }, + { + "epoch": 7.08, + "learning_rate": 6.875803351757371e-07, + "loss": 0.0449, + "step": 16904 + }, + { + "epoch": 7.08, + "learning_rate": 6.869626596026225e-07, + "loss": 0.052, + "step": 16905 + }, + { + "epoch": 7.08, + "learning_rate": 6.863452517245017e-07, + "loss": 0.0368, + "step": 16906 + }, + { + "epoch": 7.08, + "learning_rate": 6.85728111559123e-07, + "loss": 0.0517, + "step": 16907 + }, + { + "epoch": 7.08, + "learning_rate": 6.851112391242231e-07, + "loss": 0.049, + "step": 16908 + }, + { + "epoch": 7.08, + "learning_rate": 6.844946344375358e-07, + "loss": 0.0619, + "step": 16909 + }, + { + "epoch": 7.08, + "learning_rate": 6.838782975167846e-07, + "loss": 0.0458, + "step": 16910 + }, + { + "epoch": 7.08, + "learning_rate": 6.832622283796842e-07, + "loss": 0.0333, + "step": 16911 + }, + { + "epoch": 7.08, + "learning_rate": 6.826464270439426e-07, + "loss": 0.0416, + "step": 16912 + }, + { + "epoch": 7.08, + "learning_rate": 6.820308935272612e-07, + "loss": 0.046, + "step": 16913 + }, + { + "epoch": 7.08, + "learning_rate": 6.814156278473338e-07, + "loss": 0.0552, + "step": 16914 + }, + { + "epoch": 7.08, + "learning_rate": 6.80800630021845e-07, + "loss": 0.059, + "step": 16915 + }, + { + "epoch": 7.08, + "learning_rate": 6.801859000684719e-07, + "loss": 0.05, + "step": 16916 + }, + { + "epoch": 7.08, + "learning_rate": 6.795714380048868e-07, + "loss": 0.0466, + "step": 16917 + }, + { + "epoch": 7.08, + "learning_rate": 6.78957243848748e-07, + "loss": 0.0438, + "step": 16918 + }, + { + "epoch": 7.08, + "learning_rate": 6.783433176177134e-07, + "loss": 0.0654, + "step": 16919 + }, + { + "epoch": 7.08, + "learning_rate": 6.777296593294269e-07, + "loss": 0.0482, + "step": 16920 + }, + { + "epoch": 7.08, + "learning_rate": 6.771162690015298e-07, + "loss": 0.0327, + "step": 16921 + }, + { + "epoch": 7.08, + "learning_rate": 6.765031466516514e-07, + "loss": 0.0551, + "step": 16922 + }, + { + "epoch": 7.09, + "learning_rate": 6.758902922974165e-07, + "loss": 0.0455, + "step": 16923 + }, + { + "epoch": 7.09, + "learning_rate": 6.752777059564431e-07, + "loss": 0.0405, + "step": 16924 + }, + { + "epoch": 7.09, + "learning_rate": 6.74665387646336e-07, + "loss": 0.0494, + "step": 16925 + }, + { + "epoch": 7.09, + "learning_rate": 6.740533373846969e-07, + "loss": 0.0416, + "step": 16926 + }, + { + "epoch": 7.09, + "learning_rate": 6.734415551891193e-07, + "loss": 0.0524, + "step": 16927 + }, + { + "epoch": 7.09, + "learning_rate": 6.728300410771871e-07, + "loss": 0.0335, + "step": 16928 + }, + { + "epoch": 7.09, + "learning_rate": 6.722187950664794e-07, + "loss": 0.065, + "step": 16929 + }, + { + "epoch": 7.09, + "learning_rate": 6.716078171745643e-07, + "loss": 0.0502, + "step": 16930 + }, + { + "epoch": 7.09, + "learning_rate": 6.709971074190069e-07, + "loss": 0.0467, + "step": 16931 + }, + { + "epoch": 7.09, + "learning_rate": 6.703866658173574e-07, + "loss": 0.0301, + "step": 16932 + }, + { + "epoch": 7.09, + "learning_rate": 6.697764923871652e-07, + "loss": 0.0571, + "step": 16933 + }, + { + "epoch": 7.09, + "learning_rate": 6.691665871459674e-07, + "loss": 0.0473, + "step": 16934 + }, + { + "epoch": 7.09, + "learning_rate": 6.685569501112954e-07, + "loss": 0.0477, + "step": 16935 + }, + { + "epoch": 7.09, + "learning_rate": 6.679475813006742e-07, + "loss": 0.0532, + "step": 16936 + }, + { + "epoch": 7.09, + "learning_rate": 6.673384807316197e-07, + "loss": 0.0344, + "step": 16937 + }, + { + "epoch": 7.09, + "learning_rate": 6.66729648421638e-07, + "loss": 0.0632, + "step": 16938 + }, + { + "epoch": 7.09, + "learning_rate": 6.661210843882293e-07, + "loss": 0.0556, + "step": 16939 + }, + { + "epoch": 7.09, + "learning_rate": 6.655127886488877e-07, + "loss": 0.0644, + "step": 16940 + }, + { + "epoch": 7.09, + "learning_rate": 6.649047612211002e-07, + "loss": 0.0481, + "step": 16941 + }, + { + "epoch": 7.09, + "learning_rate": 6.642970021223383e-07, + "loss": 0.0634, + "step": 16942 + }, + { + "epoch": 7.09, + "learning_rate": 6.636895113700759e-07, + "loss": 0.0725, + "step": 16943 + }, + { + "epoch": 7.09, + "learning_rate": 6.630822889817734e-07, + "loss": 0.053, + "step": 16944 + }, + { + "epoch": 7.09, + "learning_rate": 6.62475334974887e-07, + "loss": 0.0539, + "step": 16945 + }, + { + "epoch": 7.09, + "learning_rate": 6.618686493668591e-07, + "loss": 0.0409, + "step": 16946 + }, + { + "epoch": 7.1, + "learning_rate": 6.612622321751306e-07, + "loss": 0.0357, + "step": 16947 + }, + { + "epoch": 7.1, + "learning_rate": 6.606560834171316e-07, + "loss": 0.0606, + "step": 16948 + }, + { + "epoch": 7.1, + "learning_rate": 6.600502031102863e-07, + "loss": 0.0367, + "step": 16949 + }, + { + "epoch": 7.1, + "learning_rate": 6.594445912720093e-07, + "loss": 0.0457, + "step": 16950 + }, + { + "epoch": 7.1, + "learning_rate": 6.588392479197115e-07, + "loss": 0.0349, + "step": 16951 + }, + { + "epoch": 7.1, + "learning_rate": 6.582341730707875e-07, + "loss": 0.0528, + "step": 16952 + }, + { + "epoch": 7.1, + "learning_rate": 6.576293667426337e-07, + "loss": 0.0508, + "step": 16953 + }, + { + "epoch": 7.1, + "learning_rate": 6.570248289526326e-07, + "loss": 0.0495, + "step": 16954 + }, + { + "epoch": 7.1, + "learning_rate": 6.564205597181627e-07, + "loss": 0.0423, + "step": 16955 + }, + { + "epoch": 7.1, + "learning_rate": 6.558165590565924e-07, + "loss": 0.059, + "step": 16956 + }, + { + "epoch": 7.1, + "learning_rate": 6.552128269852842e-07, + "loss": 0.0425, + "step": 16957 + }, + { + "epoch": 7.1, + "learning_rate": 6.546093635215888e-07, + "loss": 0.0472, + "step": 16958 + }, + { + "epoch": 7.1, + "learning_rate": 6.540061686828559e-07, + "loss": 0.0475, + "step": 16959 + }, + { + "epoch": 7.1, + "learning_rate": 6.534032424864211e-07, + "loss": 0.0521, + "step": 16960 + }, + { + "epoch": 7.1, + "learning_rate": 6.528005849496166e-07, + "loss": 0.0355, + "step": 16961 + }, + { + "epoch": 7.1, + "learning_rate": 6.521981960897639e-07, + "loss": 0.0614, + "step": 16962 + }, + { + "epoch": 7.1, + "learning_rate": 6.515960759241791e-07, + "loss": 0.0398, + "step": 16963 + }, + { + "epoch": 7.1, + "learning_rate": 6.509942244701717e-07, + "loss": 0.0477, + "step": 16964 + }, + { + "epoch": 7.1, + "learning_rate": 6.50392641745039e-07, + "loss": 0.0532, + "step": 16965 + }, + { + "epoch": 7.1, + "learning_rate": 6.497913277660716e-07, + "loss": 0.0637, + "step": 16966 + }, + { + "epoch": 7.1, + "learning_rate": 6.491902825505548e-07, + "loss": 0.0484, + "step": 16967 + }, + { + "epoch": 7.1, + "learning_rate": 6.485895061157665e-07, + "loss": 0.0621, + "step": 16968 + }, + { + "epoch": 7.1, + "learning_rate": 6.479889984789733e-07, + "loss": 0.0641, + "step": 16969 + }, + { + "epoch": 7.1, + "learning_rate": 6.47388759657438e-07, + "loss": 0.0537, + "step": 16970 + }, + { + "epoch": 7.11, + "learning_rate": 6.467887896684144e-07, + "loss": 0.0448, + "step": 16971 + }, + { + "epoch": 7.11, + "learning_rate": 6.461890885291477e-07, + "loss": 0.0457, + "step": 16972 + }, + { + "epoch": 7.11, + "learning_rate": 6.455896562568731e-07, + "loss": 0.0567, + "step": 16973 + }, + { + "epoch": 7.11, + "learning_rate": 6.449904928688244e-07, + "loss": 0.0442, + "step": 16974 + }, + { + "epoch": 7.11, + "learning_rate": 6.443915983822213e-07, + "loss": 0.056, + "step": 16975 + }, + { + "epoch": 7.11, + "learning_rate": 6.4379297281428e-07, + "loss": 0.0709, + "step": 16976 + }, + { + "epoch": 7.11, + "learning_rate": 6.431946161822078e-07, + "loss": 0.0519, + "step": 16977 + }, + { + "epoch": 7.11, + "learning_rate": 6.425965285032043e-07, + "loss": 0.0594, + "step": 16978 + }, + { + "epoch": 7.11, + "learning_rate": 6.41998709794458e-07, + "loss": 0.0358, + "step": 16979 + }, + { + "epoch": 7.11, + "learning_rate": 6.414011600731552e-07, + "loss": 0.0299, + "step": 16980 + }, + { + "epoch": 7.11, + "learning_rate": 6.408038793564708e-07, + "loss": 0.0414, + "step": 16981 + }, + { + "epoch": 7.11, + "learning_rate": 6.402068676615747e-07, + "loss": 0.0359, + "step": 16982 + }, + { + "epoch": 7.11, + "learning_rate": 6.396101250056252e-07, + "loss": 0.0433, + "step": 16983 + }, + { + "epoch": 7.11, + "learning_rate": 6.390136514057788e-07, + "loss": 0.0505, + "step": 16984 + }, + { + "epoch": 7.11, + "learning_rate": 6.384174468791771e-07, + "loss": 0.053, + "step": 16985 + }, + { + "epoch": 7.11, + "learning_rate": 6.378215114429587e-07, + "loss": 0.0392, + "step": 16986 + }, + { + "epoch": 7.11, + "learning_rate": 6.372258451142533e-07, + "loss": 0.0524, + "step": 16987 + }, + { + "epoch": 7.11, + "learning_rate": 6.366304479101815e-07, + "loss": 0.0684, + "step": 16988 + }, + { + "epoch": 7.11, + "learning_rate": 6.360353198478586e-07, + "loss": 0.0296, + "step": 16989 + }, + { + "epoch": 7.11, + "learning_rate": 6.354404609443909e-07, + "loss": 0.042, + "step": 16990 + }, + { + "epoch": 7.11, + "learning_rate": 6.348458712168771e-07, + "loss": 0.0531, + "step": 16991 + }, + { + "epoch": 7.11, + "learning_rate": 6.342515506824099e-07, + "loss": 0.0372, + "step": 16992 + }, + { + "epoch": 7.11, + "learning_rate": 6.336574993580702e-07, + "loss": 0.0533, + "step": 16993 + }, + { + "epoch": 7.11, + "learning_rate": 6.330637172609333e-07, + "loss": 0.0447, + "step": 16994 + }, + { + "epoch": 7.12, + "learning_rate": 6.324702044080678e-07, + "loss": 0.052, + "step": 16995 + }, + { + "epoch": 7.12, + "learning_rate": 6.318769608165332e-07, + "loss": 0.0692, + "step": 16996 + }, + { + "epoch": 7.12, + "learning_rate": 6.312839865033826e-07, + "loss": 0.0409, + "step": 16997 + }, + { + "epoch": 7.12, + "learning_rate": 6.306912814856625e-07, + "loss": 0.0613, + "step": 16998 + }, + { + "epoch": 7.12, + "learning_rate": 6.300988457804047e-07, + "loss": 0.0461, + "step": 16999 + }, + { + "epoch": 7.12, + "learning_rate": 6.295066794046423e-07, + "loss": 0.05, + "step": 17000 + }, + { + "epoch": 7.12, + "learning_rate": 6.289147823753939e-07, + "loss": 0.0657, + "step": 17001 + }, + { + "epoch": 7.12, + "learning_rate": 6.283231547096746e-07, + "loss": 0.0571, + "step": 17002 + }, + { + "epoch": 7.12, + "learning_rate": 6.2773179642449e-07, + "loss": 0.0354, + "step": 17003 + }, + { + "epoch": 7.12, + "learning_rate": 6.271407075368385e-07, + "loss": 0.0361, + "step": 17004 + }, + { + "epoch": 7.12, + "learning_rate": 6.26549888063711e-07, + "loss": 0.0458, + "step": 17005 + }, + { + "epoch": 7.12, + "learning_rate": 6.259593380220875e-07, + "loss": 0.0539, + "step": 17006 + }, + { + "epoch": 7.12, + "learning_rate": 6.253690574289451e-07, + "loss": 0.0529, + "step": 17007 + }, + { + "epoch": 7.12, + "learning_rate": 6.247790463012493e-07, + "loss": 0.0425, + "step": 17008 + }, + { + "epoch": 7.12, + "learning_rate": 6.241893046559633e-07, + "loss": 0.0618, + "step": 17009 + }, + { + "epoch": 7.12, + "learning_rate": 6.235998325100334e-07, + "loss": 0.0516, + "step": 17010 + }, + { + "epoch": 7.12, + "learning_rate": 6.23010629880405e-07, + "loss": 0.0726, + "step": 17011 + }, + { + "epoch": 7.12, + "learning_rate": 6.224216967840169e-07, + "loss": 0.0473, + "step": 17012 + }, + { + "epoch": 7.12, + "learning_rate": 6.218330332377953e-07, + "loss": 0.0543, + "step": 17013 + }, + { + "epoch": 7.12, + "learning_rate": 6.212446392586613e-07, + "loss": 0.0578, + "step": 17014 + }, + { + "epoch": 7.12, + "learning_rate": 6.206565148635269e-07, + "loss": 0.0423, + "step": 17015 + }, + { + "epoch": 7.12, + "learning_rate": 6.200686600692974e-07, + "loss": 0.0524, + "step": 17016 + }, + { + "epoch": 7.12, + "learning_rate": 6.194810748928714e-07, + "loss": 0.041, + "step": 17017 + }, + { + "epoch": 7.12, + "learning_rate": 6.188937593511368e-07, + "loss": 0.0781, + "step": 17018 + }, + { + "epoch": 7.13, + "learning_rate": 6.183067134609788e-07, + "loss": 0.0519, + "step": 17019 + }, + { + "epoch": 7.13, + "learning_rate": 6.177199372392673e-07, + "loss": 0.0491, + "step": 17020 + }, + { + "epoch": 7.13, + "learning_rate": 6.171334307028709e-07, + "loss": 0.0446, + "step": 17021 + }, + { + "epoch": 7.13, + "learning_rate": 6.165471938686474e-07, + "loss": 0.0611, + "step": 17022 + }, + { + "epoch": 7.13, + "learning_rate": 6.159612267534487e-07, + "loss": 0.0382, + "step": 17023 + }, + { + "epoch": 7.13, + "learning_rate": 6.15375529374116e-07, + "loss": 0.0452, + "step": 17024 + }, + { + "epoch": 7.13, + "learning_rate": 6.147901017474889e-07, + "loss": 0.0458, + "step": 17025 + }, + { + "epoch": 7.13, + "learning_rate": 6.142049438903896e-07, + "loss": 0.0655, + "step": 17026 + }, + { + "epoch": 7.13, + "learning_rate": 6.136200558196404e-07, + "loss": 0.0524, + "step": 17027 + }, + { + "epoch": 7.13, + "learning_rate": 6.130354375520542e-07, + "loss": 0.064, + "step": 17028 + }, + { + "epoch": 7.13, + "learning_rate": 6.124510891044333e-07, + "loss": 0.0449, + "step": 17029 + }, + { + "epoch": 7.13, + "learning_rate": 6.118670104935765e-07, + "loss": 0.0404, + "step": 17030 + }, + { + "epoch": 7.13, + "learning_rate": 6.112832017362713e-07, + "loss": 0.0543, + "step": 17031 + }, + { + "epoch": 7.13, + "learning_rate": 6.106996628493e-07, + "loss": 0.0434, + "step": 17032 + }, + { + "epoch": 7.13, + "learning_rate": 6.101163938494359e-07, + "loss": 0.0572, + "step": 17033 + }, + { + "epoch": 7.13, + "learning_rate": 6.095333947534421e-07, + "loss": 0.0562, + "step": 17034 + }, + { + "epoch": 7.13, + "learning_rate": 6.089506655780774e-07, + "loss": 0.0447, + "step": 17035 + }, + { + "epoch": 7.13, + "learning_rate": 6.083682063400919e-07, + "loss": 0.0567, + "step": 17036 + }, + { + "epoch": 7.13, + "learning_rate": 6.077860170562288e-07, + "loss": 0.0344, + "step": 17037 + }, + { + "epoch": 7.13, + "learning_rate": 6.072040977432225e-07, + "loss": 0.0537, + "step": 17038 + }, + { + "epoch": 7.13, + "learning_rate": 6.066224484178007e-07, + "loss": 0.0579, + "step": 17039 + }, + { + "epoch": 7.13, + "learning_rate": 6.0604106909668e-07, + "loss": 0.0518, + "step": 17040 + }, + { + "epoch": 7.13, + "learning_rate": 6.054599597965716e-07, + "loss": 0.0407, + "step": 17041 + }, + { + "epoch": 7.14, + "learning_rate": 6.048791205341809e-07, + "loss": 0.0791, + "step": 17042 + }, + { + "epoch": 7.14, + "learning_rate": 6.042985513262023e-07, + "loss": 0.0451, + "step": 17043 + }, + { + "epoch": 7.14, + "learning_rate": 6.037182521893247e-07, + "loss": 0.0292, + "step": 17044 + }, + { + "epoch": 7.14, + "learning_rate": 6.031382231402283e-07, + "loss": 0.0701, + "step": 17045 + }, + { + "epoch": 7.14, + "learning_rate": 6.025584641955862e-07, + "loss": 0.0321, + "step": 17046 + }, + { + "epoch": 7.14, + "learning_rate": 6.019789753720595e-07, + "loss": 0.0325, + "step": 17047 + }, + { + "epoch": 7.14, + "learning_rate": 6.013997566863084e-07, + "loss": 0.0744, + "step": 17048 + }, + { + "epoch": 7.14, + "learning_rate": 6.008208081549805e-07, + "loss": 0.0507, + "step": 17049 + }, + { + "epoch": 7.14, + "learning_rate": 6.002421297947181e-07, + "loss": 0.048, + "step": 17050 + }, + { + "epoch": 7.14, + "learning_rate": 5.996637216221545e-07, + "loss": 0.0519, + "step": 17051 + }, + { + "epoch": 7.14, + "learning_rate": 5.990855836539178e-07, + "loss": 0.0617, + "step": 17052 + }, + { + "epoch": 7.14, + "learning_rate": 5.985077159066211e-07, + "loss": 0.047, + "step": 17053 + }, + { + "epoch": 7.14, + "learning_rate": 5.979301183968789e-07, + "loss": 0.0532, + "step": 17054 + }, + { + "epoch": 7.14, + "learning_rate": 5.973527911412913e-07, + "loss": 0.0493, + "step": 17055 + }, + { + "epoch": 7.14, + "learning_rate": 5.96775734156454e-07, + "loss": 0.0379, + "step": 17056 + }, + { + "epoch": 7.14, + "learning_rate": 5.961989474589536e-07, + "loss": 0.0469, + "step": 17057 + }, + { + "epoch": 7.14, + "learning_rate": 5.956224310653702e-07, + "loss": 0.0569, + "step": 17058 + }, + { + "epoch": 7.14, + "learning_rate": 5.95046184992274e-07, + "loss": 0.0599, + "step": 17059 + }, + { + "epoch": 7.14, + "learning_rate": 5.944702092562315e-07, + "loss": 0.0446, + "step": 17060 + }, + { + "epoch": 7.14, + "learning_rate": 5.938945038737942e-07, + "loss": 0.043, + "step": 17061 + }, + { + "epoch": 7.14, + "learning_rate": 5.933190688615131e-07, + "loss": 0.035, + "step": 17062 + }, + { + "epoch": 7.14, + "learning_rate": 5.927439042359284e-07, + "loss": 0.0371, + "step": 17063 + }, + { + "epoch": 7.14, + "learning_rate": 5.921690100135713e-07, + "loss": 0.0493, + "step": 17064 + }, + { + "epoch": 7.14, + "learning_rate": 5.915943862109685e-07, + "loss": 0.0422, + "step": 17065 + }, + { + "epoch": 7.15, + "learning_rate": 5.91020032844638e-07, + "loss": 0.0691, + "step": 17066 + }, + { + "epoch": 7.15, + "learning_rate": 5.904459499310844e-07, + "loss": 0.0396, + "step": 17067 + }, + { + "epoch": 7.15, + "learning_rate": 5.898721374868132e-07, + "loss": 0.0283, + "step": 17068 + }, + { + "epoch": 7.15, + "learning_rate": 5.89298595528317e-07, + "loss": 0.0624, + "step": 17069 + }, + { + "epoch": 7.15, + "learning_rate": 5.887253240720814e-07, + "loss": 0.0461, + "step": 17070 + }, + { + "epoch": 7.15, + "learning_rate": 5.881523231345854e-07, + "loss": 0.0477, + "step": 17071 + }, + { + "epoch": 7.15, + "learning_rate": 5.875795927322981e-07, + "loss": 0.0541, + "step": 17072 + }, + { + "epoch": 7.15, + "learning_rate": 5.87007132881685e-07, + "loss": 0.052, + "step": 17073 + }, + { + "epoch": 7.15, + "learning_rate": 5.864349435991978e-07, + "loss": 0.0411, + "step": 17074 + }, + { + "epoch": 7.15, + "learning_rate": 5.858630249012842e-07, + "loss": 0.0537, + "step": 17075 + }, + { + "epoch": 7.15, + "learning_rate": 5.852913768043833e-07, + "loss": 0.057, + "step": 17076 + }, + { + "epoch": 7.15, + "learning_rate": 5.847199993249297e-07, + "loss": 0.0607, + "step": 17077 + }, + { + "epoch": 7.15, + "learning_rate": 5.841488924793437e-07, + "loss": 0.0495, + "step": 17078 + }, + { + "epoch": 7.15, + "learning_rate": 5.83578056284041e-07, + "loss": 0.0314, + "step": 17079 + }, + { + "epoch": 7.15, + "learning_rate": 5.83007490755434e-07, + "loss": 0.051, + "step": 17080 + }, + { + "epoch": 7.15, + "learning_rate": 5.824371959099174e-07, + "loss": 0.0399, + "step": 17081 + }, + { + "epoch": 7.15, + "learning_rate": 5.81867171763888e-07, + "loss": 0.0408, + "step": 17082 + }, + { + "epoch": 7.15, + "learning_rate": 5.812974183337283e-07, + "loss": 0.0347, + "step": 17083 + }, + { + "epoch": 7.15, + "learning_rate": 5.807279356358164e-07, + "loss": 0.0461, + "step": 17084 + }, + { + "epoch": 7.15, + "learning_rate": 5.801587236865214e-07, + "loss": 0.0652, + "step": 17085 + }, + { + "epoch": 7.15, + "learning_rate": 5.795897825022046e-07, + "loss": 0.0383, + "step": 17086 + }, + { + "epoch": 7.15, + "learning_rate": 5.790211120992218e-07, + "loss": 0.0328, + "step": 17087 + }, + { + "epoch": 7.15, + "learning_rate": 5.784527124939154e-07, + "loss": 0.0486, + "step": 17088 + }, + { + "epoch": 7.15, + "learning_rate": 5.778845837026259e-07, + "loss": 0.0435, + "step": 17089 + }, + { + "epoch": 7.16, + "learning_rate": 5.773167257416823e-07, + "loss": 0.0477, + "step": 17090 + }, + { + "epoch": 7.16, + "learning_rate": 5.767491386274082e-07, + "loss": 0.0498, + "step": 17091 + }, + { + "epoch": 7.16, + "learning_rate": 5.761818223761184e-07, + "loss": 0.0428, + "step": 17092 + }, + { + "epoch": 7.16, + "learning_rate": 5.756147770041209e-07, + "loss": 0.0436, + "step": 17093 + }, + { + "epoch": 7.16, + "learning_rate": 5.750480025277116e-07, + "loss": 0.0549, + "step": 17094 + }, + { + "epoch": 7.16, + "learning_rate": 5.744814989631842e-07, + "loss": 0.033, + "step": 17095 + }, + { + "epoch": 7.16, + "learning_rate": 5.739152663268222e-07, + "loss": 0.0581, + "step": 17096 + }, + { + "epoch": 7.16, + "learning_rate": 5.733493046349015e-07, + "loss": 0.0454, + "step": 17097 + }, + { + "epoch": 7.16, + "learning_rate": 5.727836139036902e-07, + "loss": 0.0521, + "step": 17098 + }, + { + "epoch": 7.16, + "learning_rate": 5.722181941494476e-07, + "loss": 0.0507, + "step": 17099 + }, + { + "epoch": 7.16, + "learning_rate": 5.716530453884283e-07, + "loss": 0.0441, + "step": 17100 + }, + { + "epoch": 7.16, + "learning_rate": 5.710881676368762e-07, + "loss": 0.0264, + "step": 17101 + }, + { + "epoch": 7.16, + "learning_rate": 5.70523560911026e-07, + "loss": 0.0546, + "step": 17102 + }, + { + "epoch": 7.16, + "learning_rate": 5.69959225227108e-07, + "loss": 0.0417, + "step": 17103 + }, + { + "epoch": 7.16, + "learning_rate": 5.693951606013448e-07, + "loss": 0.052, + "step": 17104 + }, + { + "epoch": 7.16, + "learning_rate": 5.68831367049949e-07, + "loss": 0.0548, + "step": 17105 + }, + { + "epoch": 7.16, + "learning_rate": 5.682678445891265e-07, + "loss": 0.058, + "step": 17106 + }, + { + "epoch": 7.16, + "learning_rate": 5.677045932350766e-07, + "loss": 0.064, + "step": 17107 + }, + { + "epoch": 7.16, + "learning_rate": 5.671416130039864e-07, + "loss": 0.0672, + "step": 17108 + }, + { + "epoch": 7.16, + "learning_rate": 5.665789039120395e-07, + "loss": 0.0542, + "step": 17109 + }, + { + "epoch": 7.16, + "learning_rate": 5.66016465975412e-07, + "loss": 0.0412, + "step": 17110 + }, + { + "epoch": 7.16, + "learning_rate": 5.654542992102685e-07, + "loss": 0.0584, + "step": 17111 + }, + { + "epoch": 7.16, + "learning_rate": 5.648924036327697e-07, + "loss": 0.0335, + "step": 17112 + }, + { + "epoch": 7.16, + "learning_rate": 5.643307792590669e-07, + "loss": 0.0471, + "step": 17113 + }, + { + "epoch": 7.17, + "learning_rate": 5.637694261053029e-07, + "loss": 0.0373, + "step": 17114 + }, + { + "epoch": 7.17, + "learning_rate": 5.632083441876123e-07, + "loss": 0.0523, + "step": 17115 + }, + { + "epoch": 7.17, + "learning_rate": 5.626475335221248e-07, + "loss": 0.0573, + "step": 17116 + }, + { + "epoch": 7.17, + "learning_rate": 5.620869941249585e-07, + "loss": 0.0413, + "step": 17117 + }, + { + "epoch": 7.17, + "learning_rate": 5.615267260122259e-07, + "loss": 0.0619, + "step": 17118 + }, + { + "epoch": 7.17, + "learning_rate": 5.609667292000331e-07, + "loss": 0.0465, + "step": 17119 + }, + { + "epoch": 7.17, + "learning_rate": 5.604070037044762e-07, + "loss": 0.0475, + "step": 17120 + }, + { + "epoch": 7.17, + "learning_rate": 5.598475495416422e-07, + "loss": 0.0465, + "step": 17121 + }, + { + "epoch": 7.17, + "learning_rate": 5.59288366727615e-07, + "loss": 0.0392, + "step": 17122 + }, + { + "epoch": 7.17, + "learning_rate": 5.58729455278465e-07, + "loss": 0.0409, + "step": 17123 + }, + { + "epoch": 7.17, + "learning_rate": 5.581708152102583e-07, + "loss": 0.0346, + "step": 17124 + }, + { + "epoch": 7.17, + "learning_rate": 5.576124465390532e-07, + "loss": 0.0457, + "step": 17125 + }, + { + "epoch": 7.17, + "learning_rate": 5.570543492808988e-07, + "loss": 0.0469, + "step": 17126 + }, + { + "epoch": 7.17, + "learning_rate": 5.564965234518383e-07, + "loss": 0.06, + "step": 17127 + }, + { + "epoch": 7.17, + "learning_rate": 5.559389690679063e-07, + "loss": 0.068, + "step": 17128 + }, + { + "epoch": 7.17, + "learning_rate": 5.553816861451267e-07, + "loss": 0.0508, + "step": 17129 + }, + { + "epoch": 7.17, + "learning_rate": 5.5482467469952e-07, + "loss": 0.0634, + "step": 17130 + }, + { + "epoch": 7.17, + "learning_rate": 5.542679347470958e-07, + "loss": 0.0611, + "step": 17131 + }, + { + "epoch": 7.17, + "learning_rate": 5.537114663038579e-07, + "loss": 0.0391, + "step": 17132 + }, + { + "epoch": 7.17, + "learning_rate": 5.531552693858022e-07, + "loss": 0.0572, + "step": 17133 + }, + { + "epoch": 7.17, + "learning_rate": 5.525993440089161e-07, + "loss": 0.0517, + "step": 17134 + }, + { + "epoch": 7.17, + "learning_rate": 5.520436901891768e-07, + "loss": 0.0466, + "step": 17135 + }, + { + "epoch": 7.17, + "learning_rate": 5.514883079425581e-07, + "loss": 0.0506, + "step": 17136 + }, + { + "epoch": 7.17, + "learning_rate": 5.509331972850229e-07, + "loss": 0.0642, + "step": 17137 + }, + { + "epoch": 7.18, + "learning_rate": 5.503783582325283e-07, + "loss": 0.0459, + "step": 17138 + }, + { + "epoch": 7.18, + "learning_rate": 5.498237908010229e-07, + "loss": 0.0536, + "step": 17139 + }, + { + "epoch": 7.18, + "learning_rate": 5.492694950064459e-07, + "loss": 0.0532, + "step": 17140 + }, + { + "epoch": 7.18, + "learning_rate": 5.487154708647324e-07, + "loss": 0.0545, + "step": 17141 + }, + { + "epoch": 7.18, + "learning_rate": 5.481617183918053e-07, + "loss": 0.049, + "step": 17142 + }, + { + "epoch": 7.18, + "learning_rate": 5.476082376035807e-07, + "loss": 0.0629, + "step": 17143 + }, + { + "epoch": 7.18, + "learning_rate": 5.470550285159704e-07, + "loss": 0.0462, + "step": 17144 + }, + { + "epoch": 7.18, + "learning_rate": 5.465020911448759e-07, + "loss": 0.047, + "step": 17145 + }, + { + "epoch": 7.18, + "learning_rate": 5.459494255061892e-07, + "loss": 0.0412, + "step": 17146 + }, + { + "epoch": 7.18, + "learning_rate": 5.453970316157964e-07, + "loss": 0.0722, + "step": 17147 + }, + { + "epoch": 7.18, + "learning_rate": 5.448449094895769e-07, + "loss": 0.0416, + "step": 17148 + }, + { + "epoch": 7.18, + "learning_rate": 5.442930591433992e-07, + "loss": 0.0392, + "step": 17149 + }, + { + "epoch": 7.18, + "learning_rate": 5.437414805931274e-07, + "loss": 0.0586, + "step": 17150 + }, + { + "epoch": 7.18, + "learning_rate": 5.431901738546142e-07, + "loss": 0.0482, + "step": 17151 + }, + { + "epoch": 7.18, + "learning_rate": 5.42639138943708e-07, + "loss": 0.0548, + "step": 17152 + }, + { + "epoch": 7.18, + "learning_rate": 5.420883758762474e-07, + "loss": 0.0484, + "step": 17153 + }, + { + "epoch": 7.18, + "learning_rate": 5.415378846680641e-07, + "loss": 0.0542, + "step": 17154 + }, + { + "epoch": 7.18, + "learning_rate": 5.409876653349822e-07, + "loss": 0.0526, + "step": 17155 + }, + { + "epoch": 7.18, + "learning_rate": 5.404377178928144e-07, + "loss": 0.047, + "step": 17156 + }, + { + "epoch": 7.18, + "learning_rate": 5.398880423573704e-07, + "loss": 0.0596, + "step": 17157 + }, + { + "epoch": 7.18, + "learning_rate": 5.393386387444499e-07, + "loss": 0.0485, + "step": 17158 + }, + { + "epoch": 7.18, + "learning_rate": 5.387895070698457e-07, + "loss": 0.0361, + "step": 17159 + }, + { + "epoch": 7.18, + "learning_rate": 5.382406473493395e-07, + "loss": 0.0459, + "step": 17160 + }, + { + "epoch": 7.18, + "learning_rate": 5.376920595987123e-07, + "loss": 0.0495, + "step": 17161 + }, + { + "epoch": 7.19, + "learning_rate": 5.37143743833729e-07, + "loss": 0.0529, + "step": 17162 + }, + { + "epoch": 7.19, + "learning_rate": 5.365957000701516e-07, + "loss": 0.0338, + "step": 17163 + }, + { + "epoch": 7.19, + "learning_rate": 5.360479283237319e-07, + "loss": 0.0757, + "step": 17164 + }, + { + "epoch": 7.19, + "learning_rate": 5.355004286102172e-07, + "loss": 0.0415, + "step": 17165 + }, + { + "epoch": 7.19, + "learning_rate": 5.34953200945344e-07, + "loss": 0.0604, + "step": 17166 + }, + { + "epoch": 7.19, + "learning_rate": 5.34406245344844e-07, + "loss": 0.0416, + "step": 17167 + }, + { + "epoch": 7.19, + "learning_rate": 5.338595618244347e-07, + "loss": 0.049, + "step": 17168 + }, + { + "epoch": 7.19, + "learning_rate": 5.333131503998334e-07, + "loss": 0.059, + "step": 17169 + }, + { + "epoch": 7.19, + "learning_rate": 5.327670110867444e-07, + "loss": 0.0566, + "step": 17170 + }, + { + "epoch": 7.19, + "learning_rate": 5.322211439008651e-07, + "loss": 0.0499, + "step": 17171 + }, + { + "epoch": 7.19, + "learning_rate": 5.316755488578884e-07, + "loss": 0.0497, + "step": 17172 + }, + { + "epoch": 7.19, + "learning_rate": 5.311302259734963e-07, + "loss": 0.0619, + "step": 17173 + }, + { + "epoch": 7.19, + "learning_rate": 5.305851752633628e-07, + "loss": 0.0443, + "step": 17174 + }, + { + "epoch": 7.19, + "learning_rate": 5.300403967431578e-07, + "loss": 0.0595, + "step": 17175 + }, + { + "epoch": 7.19, + "learning_rate": 5.294958904285363e-07, + "loss": 0.0332, + "step": 17176 + }, + { + "epoch": 7.19, + "learning_rate": 5.289516563351504e-07, + "loss": 0.0602, + "step": 17177 + }, + { + "epoch": 7.19, + "learning_rate": 5.284076944786454e-07, + "loss": 0.0471, + "step": 17178 + }, + { + "epoch": 7.19, + "learning_rate": 5.278640048746564e-07, + "loss": 0.0685, + "step": 17179 + }, + { + "epoch": 7.19, + "learning_rate": 5.273205875388109e-07, + "loss": 0.0361, + "step": 17180 + }, + { + "epoch": 7.19, + "learning_rate": 5.267774424867312e-07, + "loss": 0.0628, + "step": 17181 + }, + { + "epoch": 7.19, + "learning_rate": 5.262345697340254e-07, + "loss": 0.036, + "step": 17182 + }, + { + "epoch": 7.19, + "learning_rate": 5.256919692963003e-07, + "loss": 0.0546, + "step": 17183 + }, + { + "epoch": 7.19, + "learning_rate": 5.251496411891521e-07, + "loss": 0.0416, + "step": 17184 + }, + { + "epoch": 7.19, + "learning_rate": 5.246075854281696e-07, + "loss": 0.0642, + "step": 17185 + }, + { + "epoch": 7.2, + "learning_rate": 5.240658020289347e-07, + "loss": 0.0384, + "step": 17186 + }, + { + "epoch": 7.2, + "learning_rate": 5.235242910070182e-07, + "loss": 0.0674, + "step": 17187 + }, + { + "epoch": 7.2, + "learning_rate": 5.229830523779888e-07, + "loss": 0.0377, + "step": 17188 + }, + { + "epoch": 7.2, + "learning_rate": 5.224420861573997e-07, + "loss": 0.0552, + "step": 17189 + }, + { + "epoch": 7.2, + "learning_rate": 5.219013923608051e-07, + "loss": 0.0486, + "step": 17190 + }, + { + "epoch": 7.2, + "learning_rate": 5.213609710037415e-07, + "loss": 0.0612, + "step": 17191 + }, + { + "epoch": 7.2, + "learning_rate": 5.208208221017463e-07, + "loss": 0.0494, + "step": 17192 + }, + { + "epoch": 7.2, + "learning_rate": 5.202809456703451e-07, + "loss": 0.0492, + "step": 17193 + }, + { + "epoch": 7.2, + "learning_rate": 5.197413417250563e-07, + "loss": 0.0579, + "step": 17194 + }, + { + "epoch": 7.2, + "learning_rate": 5.192020102813888e-07, + "loss": 0.0604, + "step": 17195 + }, + { + "epoch": 7.2, + "learning_rate": 5.186629513548491e-07, + "loss": 0.0378, + "step": 17196 + }, + { + "epoch": 7.2, + "learning_rate": 5.18124164960927e-07, + "loss": 0.039, + "step": 17197 + }, + { + "epoch": 7.2, + "learning_rate": 5.175856511151123e-07, + "loss": 0.0578, + "step": 17198 + }, + { + "epoch": 7.2, + "learning_rate": 5.170474098328826e-07, + "loss": 0.0604, + "step": 17199 + }, + { + "epoch": 7.2, + "learning_rate": 5.165094411297111e-07, + "loss": 0.0532, + "step": 17200 + }, + { + "epoch": 7.2, + "learning_rate": 5.1597174502106e-07, + "loss": 0.0469, + "step": 17201 + }, + { + "epoch": 7.2, + "learning_rate": 5.154343215223856e-07, + "loss": 0.0451, + "step": 17202 + }, + { + "epoch": 7.2, + "learning_rate": 5.148971706491345e-07, + "loss": 0.0635, + "step": 17203 + }, + { + "epoch": 7.2, + "learning_rate": 5.143602924167479e-07, + "loss": 0.0566, + "step": 17204 + }, + { + "epoch": 7.2, + "learning_rate": 5.138236868406565e-07, + "loss": 0.0526, + "step": 17205 + }, + { + "epoch": 7.2, + "learning_rate": 5.13287353936286e-07, + "loss": 0.0509, + "step": 17206 + }, + { + "epoch": 7.2, + "learning_rate": 5.127512937190515e-07, + "loss": 0.0469, + "step": 17207 + }, + { + "epoch": 7.2, + "learning_rate": 5.122155062043632e-07, + "loss": 0.058, + "step": 17208 + }, + { + "epoch": 7.2, + "learning_rate": 5.116799914076198e-07, + "loss": 0.0455, + "step": 17209 + }, + { + "epoch": 7.21, + "learning_rate": 5.111447493442156e-07, + "loss": 0.072, + "step": 17210 + }, + { + "epoch": 7.21, + "learning_rate": 5.106097800295351e-07, + "loss": 0.0577, + "step": 17211 + }, + { + "epoch": 7.21, + "learning_rate": 5.100750834789559e-07, + "loss": 0.0573, + "step": 17212 + }, + { + "epoch": 7.21, + "learning_rate": 5.09540659707849e-07, + "loss": 0.047, + "step": 17213 + }, + { + "epoch": 7.21, + "learning_rate": 5.090065087315721e-07, + "loss": 0.0399, + "step": 17214 + }, + { + "epoch": 7.21, + "learning_rate": 5.084726305654819e-07, + "loss": 0.0516, + "step": 17215 + }, + { + "epoch": 7.21, + "learning_rate": 5.079390252249239e-07, + "loss": 0.0476, + "step": 17216 + }, + { + "epoch": 7.21, + "learning_rate": 5.074056927252335e-07, + "loss": 0.0331, + "step": 17217 + }, + { + "epoch": 7.21, + "learning_rate": 5.068726330817442e-07, + "loss": 0.0627, + "step": 17218 + }, + { + "epoch": 7.21, + "learning_rate": 5.063398463097758e-07, + "loss": 0.049, + "step": 17219 + }, + { + "epoch": 7.21, + "learning_rate": 5.058073324246448e-07, + "loss": 0.0441, + "step": 17220 + }, + { + "epoch": 7.21, + "learning_rate": 5.052750914416571e-07, + "loss": 0.0365, + "step": 17221 + }, + { + "epoch": 7.21, + "learning_rate": 5.047431233761124e-07, + "loss": 0.0519, + "step": 17222 + }, + { + "epoch": 7.21, + "learning_rate": 5.042114282432998e-07, + "loss": 0.0562, + "step": 17223 + }, + { + "epoch": 7.21, + "learning_rate": 5.036800060585023e-07, + "loss": 0.0381, + "step": 17224 + }, + { + "epoch": 7.21, + "learning_rate": 5.031488568369969e-07, + "loss": 0.0396, + "step": 17225 + }, + { + "epoch": 7.21, + "learning_rate": 5.026179805940501e-07, + "loss": 0.0411, + "step": 17226 + }, + { + "epoch": 7.21, + "learning_rate": 5.02087377344922e-07, + "loss": 0.0682, + "step": 17227 + }, + { + "epoch": 7.21, + "learning_rate": 5.015570471048637e-07, + "loss": 0.0456, + "step": 17228 + }, + { + "epoch": 7.21, + "learning_rate": 5.010269898891196e-07, + "loss": 0.0765, + "step": 17229 + }, + { + "epoch": 7.21, + "learning_rate": 5.004972057129253e-07, + "loss": 0.064, + "step": 17230 + }, + { + "epoch": 7.21, + "learning_rate": 4.999676945915088e-07, + "loss": 0.048, + "step": 17231 + }, + { + "epoch": 7.21, + "learning_rate": 4.994384565400912e-07, + "loss": 0.0618, + "step": 17232 + }, + { + "epoch": 7.21, + "learning_rate": 4.989094915738835e-07, + "loss": 0.0445, + "step": 17233 + }, + { + "epoch": 7.22, + "learning_rate": 4.983807997080925e-07, + "loss": 0.0666, + "step": 17234 + }, + { + "epoch": 7.22, + "learning_rate": 4.978523809579139e-07, + "loss": 0.06, + "step": 17235 + }, + { + "epoch": 7.22, + "learning_rate": 4.973242353385365e-07, + "loss": 0.052, + "step": 17236 + }, + { + "epoch": 7.22, + "learning_rate": 4.967963628651418e-07, + "loss": 0.0713, + "step": 17237 + }, + { + "epoch": 7.22, + "learning_rate": 4.962687635529018e-07, + "loss": 0.0548, + "step": 17238 + }, + { + "epoch": 7.22, + "learning_rate": 4.957414374169834e-07, + "loss": 0.0518, + "step": 17239 + }, + { + "epoch": 7.22, + "learning_rate": 4.952143844725421e-07, + "loss": 0.0585, + "step": 17240 + }, + { + "epoch": 7.22, + "learning_rate": 4.946876047347293e-07, + "loss": 0.0657, + "step": 17241 + }, + { + "epoch": 7.22, + "learning_rate": 4.941610982186873e-07, + "loss": 0.0384, + "step": 17242 + }, + { + "epoch": 7.22, + "learning_rate": 4.936348649395495e-07, + "loss": 0.0612, + "step": 17243 + }, + { + "epoch": 7.22, + "learning_rate": 4.931089049124405e-07, + "loss": 0.047, + "step": 17244 + }, + { + "epoch": 7.22, + "learning_rate": 4.925832181524803e-07, + "loss": 0.0511, + "step": 17245 + }, + { + "epoch": 7.22, + "learning_rate": 4.920578046747782e-07, + "loss": 0.0712, + "step": 17246 + }, + { + "epoch": 7.22, + "learning_rate": 4.915326644944374e-07, + "loss": 0.0619, + "step": 17247 + }, + { + "epoch": 7.22, + "learning_rate": 4.910077976265526e-07, + "loss": 0.0476, + "step": 17248 + }, + { + "epoch": 7.22, + "learning_rate": 4.904832040862129e-07, + "loss": 0.0398, + "step": 17249 + }, + { + "epoch": 7.22, + "learning_rate": 4.899588838884917e-07, + "loss": 0.0612, + "step": 17250 + }, + { + "epoch": 7.22, + "learning_rate": 4.894348370484648e-07, + "loss": 0.0294, + "step": 17251 + }, + { + "epoch": 7.22, + "learning_rate": 4.889110635811933e-07, + "loss": 0.0357, + "step": 17252 + }, + { + "epoch": 7.22, + "learning_rate": 4.883875635017332e-07, + "loss": 0.0531, + "step": 17253 + }, + { + "epoch": 7.22, + "learning_rate": 4.878643368251323e-07, + "loss": 0.0412, + "step": 17254 + }, + { + "epoch": 7.22, + "learning_rate": 4.873413835664298e-07, + "loss": 0.035, + "step": 17255 + }, + { + "epoch": 7.22, + "learning_rate": 4.868187037406602e-07, + "loss": 0.0401, + "step": 17256 + }, + { + "epoch": 7.23, + "learning_rate": 4.862962973628427e-07, + "loss": 0.0583, + "step": 17257 + }, + { + "epoch": 7.23, + "learning_rate": 4.857741644479974e-07, + "loss": 0.0446, + "step": 17258 + }, + { + "epoch": 7.23, + "learning_rate": 4.852523050111301e-07, + "loss": 0.0301, + "step": 17259 + }, + { + "epoch": 7.23, + "learning_rate": 4.847307190672424e-07, + "loss": 0.0308, + "step": 17260 + }, + { + "epoch": 7.23, + "learning_rate": 4.842094066313252e-07, + "loss": 0.0493, + "step": 17261 + }, + { + "epoch": 7.23, + "learning_rate": 4.836883677183656e-07, + "loss": 0.0656, + "step": 17262 + }, + { + "epoch": 7.23, + "learning_rate": 4.831676023433396e-07, + "loss": 0.0511, + "step": 17263 + }, + { + "epoch": 7.23, + "learning_rate": 4.826471105212149e-07, + "loss": 0.0567, + "step": 17264 + }, + { + "epoch": 7.23, + "learning_rate": 4.821268922669542e-07, + "loss": 0.047, + "step": 17265 + }, + { + "epoch": 7.23, + "learning_rate": 4.816069475955087e-07, + "loss": 0.0684, + "step": 17266 + }, + { + "epoch": 7.23, + "learning_rate": 4.810872765218255e-07, + "loss": 0.0433, + "step": 17267 + }, + { + "epoch": 7.23, + "learning_rate": 4.805678790608415e-07, + "loss": 0.0373, + "step": 17268 + }, + { + "epoch": 7.23, + "learning_rate": 4.800487552274868e-07, + "loss": 0.0461, + "step": 17269 + }, + { + "epoch": 7.23, + "learning_rate": 4.795299050366841e-07, + "loss": 0.056, + "step": 17270 + }, + { + "epoch": 7.23, + "learning_rate": 4.790113285033437e-07, + "loss": 0.0434, + "step": 17271 + }, + { + "epoch": 7.23, + "learning_rate": 4.784930256423748e-07, + "loss": 0.0303, + "step": 17272 + }, + { + "epoch": 7.23, + "learning_rate": 4.779749964686752e-07, + "loss": 0.0588, + "step": 17273 + }, + { + "epoch": 7.23, + "learning_rate": 4.774572409971334e-07, + "loss": 0.0565, + "step": 17274 + }, + { + "epoch": 7.23, + "learning_rate": 4.769397592426339e-07, + "loss": 0.0705, + "step": 17275 + }, + { + "epoch": 7.23, + "learning_rate": 4.7642255122005155e-07, + "loss": 0.0575, + "step": 17276 + }, + { + "epoch": 7.23, + "learning_rate": 4.7590561694424995e-07, + "loss": 0.058, + "step": 17277 + }, + { + "epoch": 7.23, + "learning_rate": 4.7538895643009066e-07, + "loss": 0.0532, + "step": 17278 + }, + { + "epoch": 7.23, + "learning_rate": 4.7487256969242277e-07, + "loss": 0.0531, + "step": 17279 + }, + { + "epoch": 7.23, + "learning_rate": 4.743564567460912e-07, + "loss": 0.0315, + "step": 17280 + }, + { + "epoch": 7.24, + "learning_rate": 4.738406176059318e-07, + "loss": 0.0385, + "step": 17281 + }, + { + "epoch": 7.24, + "learning_rate": 4.7332505228676937e-07, + "loss": 0.0513, + "step": 17282 + }, + { + "epoch": 7.24, + "learning_rate": 4.7280976080342434e-07, + "loss": 0.0437, + "step": 17283 + }, + { + "epoch": 7.24, + "learning_rate": 4.722947431707092e-07, + "loss": 0.0482, + "step": 17284 + }, + { + "epoch": 7.24, + "learning_rate": 4.7177999940342554e-07, + "loss": 0.0481, + "step": 17285 + }, + { + "epoch": 7.24, + "learning_rate": 4.712655295163715e-07, + "loss": 0.0416, + "step": 17286 + }, + { + "epoch": 7.24, + "learning_rate": 4.70751333524333e-07, + "loss": 0.0451, + "step": 17287 + }, + { + "epoch": 7.24, + "learning_rate": 4.702374114420927e-07, + "loss": 0.0611, + "step": 17288 + }, + { + "epoch": 7.24, + "learning_rate": 4.6972376328442095e-07, + "loss": 0.0486, + "step": 17289 + }, + { + "epoch": 7.24, + "learning_rate": 4.6921038906608375e-07, + "loss": 0.0418, + "step": 17290 + }, + { + "epoch": 7.24, + "learning_rate": 4.6869728880183486e-07, + "loss": 0.0622, + "step": 17291 + }, + { + "epoch": 7.24, + "learning_rate": 4.681844625064247e-07, + "loss": 0.0411, + "step": 17292 + }, + { + "epoch": 7.24, + "learning_rate": 4.676719101945937e-07, + "loss": 0.0574, + "step": 17293 + }, + { + "epoch": 7.24, + "learning_rate": 4.671596318810745e-07, + "loss": 0.0523, + "step": 17294 + }, + { + "epoch": 7.24, + "learning_rate": 4.666476275805931e-07, + "loss": 0.0486, + "step": 17295 + }, + { + "epoch": 7.24, + "learning_rate": 4.6613589730786557e-07, + "loss": 0.0604, + "step": 17296 + }, + { + "epoch": 7.24, + "learning_rate": 4.6562444107760227e-07, + "loss": 0.0412, + "step": 17297 + }, + { + "epoch": 7.24, + "learning_rate": 4.6511325890450374e-07, + "loss": 0.0475, + "step": 17298 + }, + { + "epoch": 7.24, + "learning_rate": 4.6460235080326264e-07, + "loss": 0.042, + "step": 17299 + }, + { + "epoch": 7.24, + "learning_rate": 4.64091716788565e-07, + "loss": 0.0475, + "step": 17300 + }, + { + "epoch": 7.24, + "learning_rate": 4.635813568750891e-07, + "loss": 0.0411, + "step": 17301 + }, + { + "epoch": 7.24, + "learning_rate": 4.630712710775054e-07, + "loss": 0.061, + "step": 17302 + }, + { + "epoch": 7.24, + "learning_rate": 4.625614594104755e-07, + "loss": 0.0551, + "step": 17303 + }, + { + "epoch": 7.24, + "learning_rate": 4.620519218886543e-07, + "loss": 0.0708, + "step": 17304 + }, + { + "epoch": 7.25, + "learning_rate": 4.6154265852668466e-07, + "loss": 0.0362, + "step": 17305 + }, + { + "epoch": 7.25, + "learning_rate": 4.61033669339207e-07, + "loss": 0.0756, + "step": 17306 + }, + { + "epoch": 7.25, + "learning_rate": 4.6052495434085185e-07, + "loss": 0.0703, + "step": 17307 + }, + { + "epoch": 7.25, + "learning_rate": 4.6001651354624087e-07, + "loss": 0.0462, + "step": 17308 + }, + { + "epoch": 7.25, + "learning_rate": 4.5950834696999127e-07, + "loss": 0.0519, + "step": 17309 + }, + { + "epoch": 7.25, + "learning_rate": 4.5900045462670684e-07, + "loss": 0.0645, + "step": 17310 + }, + { + "epoch": 7.25, + "learning_rate": 4.584928365309893e-07, + "loss": 0.05, + "step": 17311 + }, + { + "epoch": 7.25, + "learning_rate": 4.579854926974281e-07, + "loss": 0.061, + "step": 17312 + }, + { + "epoch": 7.25, + "learning_rate": 4.5747842314060486e-07, + "loss": 0.0557, + "step": 17313 + }, + { + "epoch": 7.25, + "learning_rate": 4.5697162787509686e-07, + "loss": 0.0476, + "step": 17314 + }, + { + "epoch": 7.25, + "learning_rate": 4.564651069154713e-07, + "loss": 0.053, + "step": 17315 + }, + { + "epoch": 7.25, + "learning_rate": 4.5595886027628765e-07, + "loss": 0.0552, + "step": 17316 + }, + { + "epoch": 7.25, + "learning_rate": 4.5545288797209875e-07, + "loss": 0.0505, + "step": 17317 + }, + { + "epoch": 7.25, + "learning_rate": 4.549471900174451e-07, + "loss": 0.0499, + "step": 17318 + }, + { + "epoch": 7.25, + "learning_rate": 4.544417664268652e-07, + "loss": 0.0369, + "step": 17319 + }, + { + "epoch": 7.25, + "learning_rate": 4.539366172148851e-07, + "loss": 0.0433, + "step": 17320 + }, + { + "epoch": 7.25, + "learning_rate": 4.534317423960266e-07, + "loss": 0.0449, + "step": 17321 + }, + { + "epoch": 7.25, + "learning_rate": 4.5292714198480135e-07, + "loss": 0.0608, + "step": 17322 + }, + { + "epoch": 7.25, + "learning_rate": 4.5242281599571223e-07, + "loss": 0.0682, + "step": 17323 + }, + { + "epoch": 7.25, + "learning_rate": 4.5191876444325986e-07, + "loss": 0.0345, + "step": 17324 + }, + { + "epoch": 7.25, + "learning_rate": 4.514149873419271e-07, + "loss": 0.0469, + "step": 17325 + }, + { + "epoch": 7.25, + "learning_rate": 4.5091148470619904e-07, + "loss": 0.0465, + "step": 17326 + }, + { + "epoch": 7.25, + "learning_rate": 4.504082565505452e-07, + "loss": 0.0575, + "step": 17327 + }, + { + "epoch": 7.25, + "learning_rate": 4.499053028894329e-07, + "loss": 0.055, + "step": 17328 + }, + { + "epoch": 7.26, + "learning_rate": 4.494026237373172e-07, + "loss": 0.0386, + "step": 17329 + }, + { + "epoch": 7.26, + "learning_rate": 4.489002191086478e-07, + "loss": 0.0713, + "step": 17330 + }, + { + "epoch": 7.26, + "learning_rate": 4.483980890178674e-07, + "loss": 0.0319, + "step": 17331 + }, + { + "epoch": 7.26, + "learning_rate": 4.478962334794068e-07, + "loss": 0.0569, + "step": 17332 + }, + { + "epoch": 7.26, + "learning_rate": 4.473946525076933e-07, + "loss": 0.0645, + "step": 17333 + }, + { + "epoch": 7.26, + "learning_rate": 4.468933461171432e-07, + "loss": 0.0429, + "step": 17334 + }, + { + "epoch": 7.26, + "learning_rate": 4.46392314322166e-07, + "loss": 0.0457, + "step": 17335 + }, + { + "epoch": 7.26, + "learning_rate": 4.458915571371647e-07, + "loss": 0.0455, + "step": 17336 + }, + { + "epoch": 7.26, + "learning_rate": 4.453910745765322e-07, + "loss": 0.0348, + "step": 17337 + }, + { + "epoch": 7.26, + "learning_rate": 4.4489086665465697e-07, + "loss": 0.0363, + "step": 17338 + }, + { + "epoch": 7.26, + "learning_rate": 4.44390933385912e-07, + "loss": 0.0432, + "step": 17339 + }, + { + "epoch": 7.26, + "learning_rate": 4.438912747846702e-07, + "loss": 0.0539, + "step": 17340 + }, + { + "epoch": 7.26, + "learning_rate": 4.4339189086529453e-07, + "loss": 0.058, + "step": 17341 + }, + { + "epoch": 7.26, + "learning_rate": 4.42892781642138e-07, + "loss": 0.0456, + "step": 17342 + }, + { + "epoch": 7.26, + "learning_rate": 4.42393947129548e-07, + "loss": 0.0764, + "step": 17343 + }, + { + "epoch": 7.26, + "learning_rate": 4.418953873418641e-07, + "loss": 0.0617, + "step": 17344 + }, + { + "epoch": 7.26, + "learning_rate": 4.413971022934138e-07, + "loss": 0.0425, + "step": 17345 + }, + { + "epoch": 7.26, + "learning_rate": 4.408990919985223e-07, + "loss": 0.0586, + "step": 17346 + }, + { + "epoch": 7.26, + "learning_rate": 4.4040135647150374e-07, + "loss": 0.0401, + "step": 17347 + }, + { + "epoch": 7.26, + "learning_rate": 4.399038957266644e-07, + "loss": 0.0491, + "step": 17348 + }, + { + "epoch": 7.26, + "learning_rate": 4.3940670977830614e-07, + "loss": 0.0355, + "step": 17349 + }, + { + "epoch": 7.26, + "learning_rate": 4.389097986407165e-07, + "loss": 0.069, + "step": 17350 + }, + { + "epoch": 7.26, + "learning_rate": 4.3841316232818064e-07, + "loss": 0.0486, + "step": 17351 + }, + { + "epoch": 7.26, + "learning_rate": 4.379168008549739e-07, + "loss": 0.0434, + "step": 17352 + }, + { + "epoch": 7.27, + "learning_rate": 4.3742071423536256e-07, + "loss": 0.045, + "step": 17353 + }, + { + "epoch": 7.27, + "learning_rate": 4.3692490248360755e-07, + "loss": 0.0497, + "step": 17354 + }, + { + "epoch": 7.27, + "learning_rate": 4.3642936561396063e-07, + "loss": 0.029, + "step": 17355 + }, + { + "epoch": 7.27, + "learning_rate": 4.359341036406639e-07, + "loss": 0.0545, + "step": 17356 + }, + { + "epoch": 7.27, + "learning_rate": 4.354391165779548e-07, + "loss": 0.0284, + "step": 17357 + }, + { + "epoch": 7.27, + "learning_rate": 4.3494440444006304e-07, + "loss": 0.0697, + "step": 17358 + }, + { + "epoch": 7.27, + "learning_rate": 4.344499672412039e-07, + "loss": 0.03, + "step": 17359 + }, + { + "epoch": 7.27, + "learning_rate": 4.3395580499559276e-07, + "loss": 0.06, + "step": 17360 + }, + { + "epoch": 7.27, + "learning_rate": 4.334619177174326e-07, + "loss": 0.0472, + "step": 17361 + }, + { + "epoch": 7.27, + "learning_rate": 4.3296830542092104e-07, + "loss": 0.0438, + "step": 17362 + }, + { + "epoch": 7.27, + "learning_rate": 4.3247496812024557e-07, + "loss": 0.0427, + "step": 17363 + }, + { + "epoch": 7.27, + "learning_rate": 4.319819058295893e-07, + "loss": 0.0448, + "step": 17364 + }, + { + "epoch": 7.27, + "learning_rate": 4.314891185631198e-07, + "loss": 0.05, + "step": 17365 + }, + { + "epoch": 7.27, + "learning_rate": 4.309966063350057e-07, + "loss": 0.0309, + "step": 17366 + }, + { + "epoch": 7.27, + "learning_rate": 4.305043691594024e-07, + "loss": 0.0388, + "step": 17367 + }, + { + "epoch": 7.27, + "learning_rate": 4.300124070504597e-07, + "loss": 0.0507, + "step": 17368 + }, + { + "epoch": 7.27, + "learning_rate": 4.2952072002231727e-07, + "loss": 0.045, + "step": 17369 + }, + { + "epoch": 7.27, + "learning_rate": 4.2902930808910946e-07, + "loss": 0.047, + "step": 17370 + }, + { + "epoch": 7.27, + "learning_rate": 4.285381712649628e-07, + "loss": 0.0591, + "step": 17371 + }, + { + "epoch": 7.27, + "learning_rate": 4.2804730956399255e-07, + "loss": 0.0318, + "step": 17372 + }, + { + "epoch": 7.27, + "learning_rate": 4.275567230003064e-07, + "loss": 0.0318, + "step": 17373 + }, + { + "epoch": 7.27, + "learning_rate": 4.2706641158800855e-07, + "loss": 0.0538, + "step": 17374 + }, + { + "epoch": 7.27, + "learning_rate": 4.265763753411922e-07, + "loss": 0.0387, + "step": 17375 + }, + { + "epoch": 7.27, + "learning_rate": 4.260866142739417e-07, + "loss": 0.0493, + "step": 17376 + }, + { + "epoch": 7.28, + "learning_rate": 4.2559712840033573e-07, + "loss": 0.0295, + "step": 17377 + }, + { + "epoch": 7.28, + "learning_rate": 4.2510791773444526e-07, + "loss": 0.0197, + "step": 17378 + }, + { + "epoch": 7.28, + "learning_rate": 4.2461898229033125e-07, + "loss": 0.033, + "step": 17379 + }, + { + "epoch": 7.28, + "learning_rate": 4.24130322082047e-07, + "loss": 0.0443, + "step": 17380 + }, + { + "epoch": 7.28, + "learning_rate": 4.236419371236389e-07, + "loss": 0.0478, + "step": 17381 + }, + { + "epoch": 7.28, + "learning_rate": 4.231538274291447e-07, + "loss": 0.0407, + "step": 17382 + }, + { + "epoch": 7.28, + "learning_rate": 4.226659930125965e-07, + "loss": 0.0503, + "step": 17383 + }, + { + "epoch": 7.28, + "learning_rate": 4.221784338880153e-07, + "loss": 0.0384, + "step": 17384 + }, + { + "epoch": 7.28, + "learning_rate": 4.216911500694165e-07, + "loss": 0.0419, + "step": 17385 + }, + { + "epoch": 7.28, + "learning_rate": 4.2120414157080457e-07, + "loss": 0.0671, + "step": 17386 + }, + { + "epoch": 7.28, + "learning_rate": 4.2071740840618045e-07, + "loss": 0.0418, + "step": 17387 + }, + { + "epoch": 7.28, + "learning_rate": 4.2023095058953303e-07, + "loss": 0.0592, + "step": 17388 + }, + { + "epoch": 7.28, + "learning_rate": 4.1974476813484654e-07, + "loss": 0.046, + "step": 17389 + }, + { + "epoch": 7.28, + "learning_rate": 4.1925886105609545e-07, + "loss": 0.0326, + "step": 17390 + }, + { + "epoch": 7.28, + "learning_rate": 4.187732293672475e-07, + "loss": 0.0493, + "step": 17391 + }, + { + "epoch": 7.28, + "learning_rate": 4.1828787308225927e-07, + "loss": 0.0503, + "step": 17392 + }, + { + "epoch": 7.28, + "learning_rate": 4.17802792215084e-07, + "loss": 0.0494, + "step": 17393 + }, + { + "epoch": 7.28, + "learning_rate": 4.173179867796662e-07, + "loss": 0.0586, + "step": 17394 + }, + { + "epoch": 7.28, + "learning_rate": 4.168334567899368e-07, + "loss": 0.051, + "step": 17395 + }, + { + "epoch": 7.28, + "learning_rate": 4.16349202259827e-07, + "loss": 0.0591, + "step": 17396 + }, + { + "epoch": 7.28, + "learning_rate": 4.158652232032545e-07, + "loss": 0.0404, + "step": 17397 + }, + { + "epoch": 7.28, + "learning_rate": 4.1538151963413043e-07, + "loss": 0.0403, + "step": 17398 + }, + { + "epoch": 7.28, + "learning_rate": 4.148980915663614e-07, + "loss": 0.0516, + "step": 17399 + }, + { + "epoch": 7.28, + "learning_rate": 4.1441493901383966e-07, + "loss": 0.063, + "step": 17400 + }, + { + "epoch": 7.29, + "learning_rate": 4.139320619904552e-07, + "loss": 0.0619, + "step": 17401 + }, + { + "epoch": 7.29, + "learning_rate": 4.134494605100858e-07, + "loss": 0.032, + "step": 17402 + }, + { + "epoch": 7.29, + "learning_rate": 4.12967134586606e-07, + "loss": 0.051, + "step": 17403 + }, + { + "epoch": 7.29, + "learning_rate": 4.124850842338779e-07, + "loss": 0.0308, + "step": 17404 + }, + { + "epoch": 7.29, + "learning_rate": 4.120033094657594e-07, + "loss": 0.0452, + "step": 17405 + }, + { + "epoch": 7.29, + "learning_rate": 4.1152181029609606e-07, + "loss": 0.0388, + "step": 17406 + }, + { + "epoch": 7.29, + "learning_rate": 4.1104058673873016e-07, + "loss": 0.0414, + "step": 17407 + }, + { + "epoch": 7.29, + "learning_rate": 4.105596388074928e-07, + "loss": 0.0416, + "step": 17408 + }, + { + "epoch": 7.29, + "learning_rate": 4.100789665162097e-07, + "loss": 0.0766, + "step": 17409 + }, + { + "epoch": 7.29, + "learning_rate": 4.095985698786964e-07, + "loss": 0.0513, + "step": 17410 + }, + { + "epoch": 7.29, + "learning_rate": 4.0911844890876183e-07, + "loss": 0.0473, + "step": 17411 + }, + { + "epoch": 7.29, + "learning_rate": 4.086386036202084e-07, + "loss": 0.0582, + "step": 17412 + }, + { + "epoch": 7.29, + "learning_rate": 4.0815903402682597e-07, + "loss": 0.0538, + "step": 17413 + }, + { + "epoch": 7.29, + "learning_rate": 4.076797401423993e-07, + "loss": 0.0553, + "step": 17414 + }, + { + "epoch": 7.29, + "learning_rate": 4.0720072198070725e-07, + "loss": 0.0634, + "step": 17415 + }, + { + "epoch": 7.29, + "learning_rate": 4.0672197955551773e-07, + "loss": 0.048, + "step": 17416 + }, + { + "epoch": 7.29, + "learning_rate": 4.0624351288059414e-07, + "loss": 0.0345, + "step": 17417 + }, + { + "epoch": 7.29, + "learning_rate": 4.057653219696844e-07, + "loss": 0.0466, + "step": 17418 + }, + { + "epoch": 7.29, + "learning_rate": 4.0528740683653864e-07, + "loss": 0.0434, + "step": 17419 + }, + { + "epoch": 7.29, + "learning_rate": 4.0480976749489254e-07, + "loss": 0.061, + "step": 17420 + }, + { + "epoch": 7.29, + "learning_rate": 4.0433240395847284e-07, + "loss": 0.0559, + "step": 17421 + }, + { + "epoch": 7.29, + "learning_rate": 4.0385531624100413e-07, + "loss": 0.0368, + "step": 17422 + }, + { + "epoch": 7.29, + "learning_rate": 4.033785043561989e-07, + "loss": 0.0419, + "step": 17423 + }, + { + "epoch": 7.29, + "learning_rate": 4.0290196831776264e-07, + "loss": 0.038, + "step": 17424 + }, + { + "epoch": 7.3, + "learning_rate": 4.024257081393934e-07, + "loss": 0.0545, + "step": 17425 + }, + { + "epoch": 7.3, + "learning_rate": 4.019497238347814e-07, + "loss": 0.0368, + "step": 17426 + }, + { + "epoch": 7.3, + "learning_rate": 4.014740154176056e-07, + "loss": 0.0479, + "step": 17427 + }, + { + "epoch": 7.3, + "learning_rate": 4.009985829015417e-07, + "loss": 0.0473, + "step": 17428 + }, + { + "epoch": 7.3, + "learning_rate": 4.0052342630025664e-07, + "loss": 0.0411, + "step": 17429 + }, + { + "epoch": 7.3, + "learning_rate": 4.0004854562740615e-07, + "loss": 0.0445, + "step": 17430 + }, + { + "epoch": 7.3, + "learning_rate": 3.9957394089664147e-07, + "loss": 0.0387, + "step": 17431 + }, + { + "epoch": 7.3, + "learning_rate": 3.990996121216062e-07, + "loss": 0.0619, + "step": 17432 + }, + { + "epoch": 7.3, + "learning_rate": 3.986255593159316e-07, + "loss": 0.0792, + "step": 17433 + }, + { + "epoch": 7.3, + "learning_rate": 3.981517824932446e-07, + "loss": 0.0331, + "step": 17434 + }, + { + "epoch": 7.3, + "learning_rate": 3.9767828166716535e-07, + "loss": 0.063, + "step": 17435 + }, + { + "epoch": 7.3, + "learning_rate": 3.9720505685130195e-07, + "loss": 0.0544, + "step": 17436 + }, + { + "epoch": 7.3, + "learning_rate": 3.967321080592579e-07, + "loss": 0.0414, + "step": 17437 + }, + { + "epoch": 7.3, + "learning_rate": 3.9625943530462787e-07, + "loss": 0.0573, + "step": 17438 + }, + { + "epoch": 7.3, + "learning_rate": 3.9578703860099987e-07, + "loss": 0.0431, + "step": 17439 + }, + { + "epoch": 7.3, + "learning_rate": 3.9531491796194975e-07, + "loss": 0.0329, + "step": 17440 + }, + { + "epoch": 7.3, + "learning_rate": 3.9484307340104886e-07, + "loss": 0.0559, + "step": 17441 + }, + { + "epoch": 7.3, + "learning_rate": 3.943715049318597e-07, + "loss": 0.0369, + "step": 17442 + }, + { + "epoch": 7.3, + "learning_rate": 3.939002125679381e-07, + "loss": 0.0628, + "step": 17443 + }, + { + "epoch": 7.3, + "learning_rate": 3.9342919632283093e-07, + "loss": 0.0454, + "step": 17444 + }, + { + "epoch": 7.3, + "learning_rate": 3.9295845621007635e-07, + "loss": 0.038, + "step": 17445 + }, + { + "epoch": 7.3, + "learning_rate": 3.924879922432068e-07, + "loss": 0.0705, + "step": 17446 + }, + { + "epoch": 7.3, + "learning_rate": 3.9201780443574367e-07, + "loss": 0.0651, + "step": 17447 + }, + { + "epoch": 7.31, + "learning_rate": 3.9154789280120287e-07, + "loss": 0.054, + "step": 17448 + }, + { + "epoch": 7.31, + "learning_rate": 3.9107825735309136e-07, + "loss": 0.0405, + "step": 17449 + }, + { + "epoch": 7.31, + "learning_rate": 3.906088981049083e-07, + "loss": 0.033, + "step": 17450 + }, + { + "epoch": 7.31, + "learning_rate": 3.901398150701452e-07, + "loss": 0.047, + "step": 17451 + }, + { + "epoch": 7.31, + "learning_rate": 3.8967100826228563e-07, + "loss": 0.0572, + "step": 17452 + }, + { + "epoch": 7.31, + "learning_rate": 3.8920247769480557e-07, + "loss": 0.0324, + "step": 17453 + }, + { + "epoch": 7.31, + "learning_rate": 3.8873422338117084e-07, + "loss": 0.0422, + "step": 17454 + }, + { + "epoch": 7.31, + "learning_rate": 3.8826624533484293e-07, + "loss": 0.0374, + "step": 17455 + }, + { + "epoch": 7.31, + "learning_rate": 3.8779854356927107e-07, + "loss": 0.0715, + "step": 17456 + }, + { + "epoch": 7.31, + "learning_rate": 3.873311180979011e-07, + "loss": 0.0487, + "step": 17457 + }, + { + "epoch": 7.31, + "learning_rate": 3.8686396893416797e-07, + "loss": 0.0439, + "step": 17458 + }, + { + "epoch": 7.31, + "learning_rate": 3.863970960915009e-07, + "loss": 0.0619, + "step": 17459 + }, + { + "epoch": 7.31, + "learning_rate": 3.8593049958331684e-07, + "loss": 0.0463, + "step": 17460 + }, + { + "epoch": 7.31, + "learning_rate": 3.854641794230296e-07, + "loss": 0.0467, + "step": 17461 + }, + { + "epoch": 7.31, + "learning_rate": 3.84998135624044e-07, + "loss": 0.0415, + "step": 17462 + }, + { + "epoch": 7.31, + "learning_rate": 3.845323681997526e-07, + "loss": 0.0458, + "step": 17463 + }, + { + "epoch": 7.31, + "learning_rate": 3.8406687716354695e-07, + "loss": 0.037, + "step": 17464 + }, + { + "epoch": 7.31, + "learning_rate": 3.836016625288053e-07, + "loss": 0.0711, + "step": 17465 + }, + { + "epoch": 7.31, + "learning_rate": 3.8313672430890016e-07, + "loss": 0.0652, + "step": 17466 + }, + { + "epoch": 7.31, + "learning_rate": 3.8267206251719756e-07, + "loss": 0.0621, + "step": 17467 + }, + { + "epoch": 7.31, + "learning_rate": 3.8220767716705133e-07, + "loss": 0.0326, + "step": 17468 + }, + { + "epoch": 7.31, + "learning_rate": 3.817435682718096e-07, + "loss": 0.0543, + "step": 17469 + }, + { + "epoch": 7.31, + "learning_rate": 3.81279735844815e-07, + "loss": 0.0264, + "step": 17470 + }, + { + "epoch": 7.31, + "learning_rate": 3.808161798993981e-07, + "loss": 0.0525, + "step": 17471 + }, + { + "epoch": 7.32, + "learning_rate": 3.803529004488848e-07, + "loss": 0.0604, + "step": 17472 + }, + { + "epoch": 7.32, + "learning_rate": 3.7988989750659235e-07, + "loss": 0.0645, + "step": 17473 + }, + { + "epoch": 7.32, + "learning_rate": 3.7942717108582663e-07, + "loss": 0.0479, + "step": 17474 + }, + { + "epoch": 7.32, + "learning_rate": 3.7896472119988927e-07, + "loss": 0.0339, + "step": 17475 + }, + { + "epoch": 7.32, + "learning_rate": 3.7850254786207406e-07, + "loss": 0.065, + "step": 17476 + }, + { + "epoch": 7.32, + "learning_rate": 3.780406510856649e-07, + "loss": 0.0603, + "step": 17477 + }, + { + "epoch": 7.32, + "learning_rate": 3.7757903088393886e-07, + "loss": 0.0492, + "step": 17478 + }, + { + "epoch": 7.32, + "learning_rate": 3.771176872701654e-07, + "loss": 0.0541, + "step": 17479 + }, + { + "epoch": 7.32, + "learning_rate": 3.7665662025760496e-07, + "loss": 0.0519, + "step": 17480 + }, + { + "epoch": 7.32, + "learning_rate": 3.7619582985951033e-07, + "loss": 0.0461, + "step": 17481 + }, + { + "epoch": 7.32, + "learning_rate": 3.757353160891253e-07, + "loss": 0.0587, + "step": 17482 + }, + { + "epoch": 7.32, + "learning_rate": 3.752750789596893e-07, + "loss": 0.0386, + "step": 17483 + }, + { + "epoch": 7.32, + "learning_rate": 3.748151184844295e-07, + "loss": 0.067, + "step": 17484 + }, + { + "epoch": 7.32, + "learning_rate": 3.743554346765699e-07, + "loss": 0.0416, + "step": 17485 + }, + { + "epoch": 7.32, + "learning_rate": 3.7389602754931973e-07, + "loss": 0.041, + "step": 17486 + }, + { + "epoch": 7.32, + "learning_rate": 3.734368971158875e-07, + "loss": 0.0373, + "step": 17487 + }, + { + "epoch": 7.32, + "learning_rate": 3.7297804338946807e-07, + "loss": 0.05, + "step": 17488 + }, + { + "epoch": 7.32, + "learning_rate": 3.725194663832521e-07, + "loss": 0.0446, + "step": 17489 + }, + { + "epoch": 7.32, + "learning_rate": 3.720611661104212e-07, + "loss": 0.0261, + "step": 17490 + }, + { + "epoch": 7.32, + "learning_rate": 3.716031425841482e-07, + "loss": 0.045, + "step": 17491 + }, + { + "epoch": 7.32, + "learning_rate": 3.7114539581759814e-07, + "loss": 0.0463, + "step": 17492 + }, + { + "epoch": 7.32, + "learning_rate": 3.706879258239293e-07, + "loss": 0.0356, + "step": 17493 + }, + { + "epoch": 7.32, + "learning_rate": 3.702307326162924e-07, + "loss": 0.0485, + "step": 17494 + }, + { + "epoch": 7.32, + "learning_rate": 3.697738162078268e-07, + "loss": 0.0504, + "step": 17495 + }, + { + "epoch": 7.33, + "learning_rate": 3.6931717661166764e-07, + "loss": 0.0378, + "step": 17496 + }, + { + "epoch": 7.33, + "learning_rate": 3.688608138409389e-07, + "loss": 0.0618, + "step": 17497 + }, + { + "epoch": 7.33, + "learning_rate": 3.6840472790876104e-07, + "loss": 0.0616, + "step": 17498 + }, + { + "epoch": 7.33, + "learning_rate": 3.6794891882824147e-07, + "loss": 0.0413, + "step": 17499 + }, + { + "epoch": 7.33, + "learning_rate": 3.6749338661248414e-07, + "loss": 0.0602, + "step": 17500 + }, + { + "epoch": 7.33, + "learning_rate": 3.6703813127458074e-07, + "loss": 0.064, + "step": 17501 + }, + { + "epoch": 7.33, + "learning_rate": 3.6658315282761757e-07, + "loss": 0.0716, + "step": 17502 + }, + { + "epoch": 7.33, + "learning_rate": 3.66128451284673e-07, + "loss": 0.0456, + "step": 17503 + }, + { + "epoch": 7.33, + "learning_rate": 3.656740266588177e-07, + "loss": 0.0486, + "step": 17504 + }, + { + "epoch": 7.33, + "learning_rate": 3.652198789631134e-07, + "loss": 0.0662, + "step": 17505 + }, + { + "epoch": 7.33, + "learning_rate": 3.64766008210613e-07, + "loss": 0.0621, + "step": 17506 + }, + { + "epoch": 7.33, + "learning_rate": 3.6431241441436617e-07, + "loss": 0.0472, + "step": 17507 + }, + { + "epoch": 7.33, + "learning_rate": 3.638590975874068e-07, + "loss": 0.0423, + "step": 17508 + }, + { + "epoch": 7.33, + "learning_rate": 3.634060577427667e-07, + "loss": 0.0613, + "step": 17509 + }, + { + "epoch": 7.33, + "learning_rate": 3.6295329489346776e-07, + "loss": 0.0548, + "step": 17510 + }, + { + "epoch": 7.33, + "learning_rate": 3.625008090525239e-07, + "loss": 0.0712, + "step": 17511 + }, + { + "epoch": 7.33, + "learning_rate": 3.6204860023294263e-07, + "loss": 0.0473, + "step": 17512 + }, + { + "epoch": 7.33, + "learning_rate": 3.6159666844772234e-07, + "loss": 0.0541, + "step": 17513 + }, + { + "epoch": 7.33, + "learning_rate": 3.611450137098538e-07, + "loss": 0.037, + "step": 17514 + }, + { + "epoch": 7.33, + "learning_rate": 3.606936360323166e-07, + "loss": 0.054, + "step": 17515 + }, + { + "epoch": 7.33, + "learning_rate": 3.6024253542808806e-07, + "loss": 0.0396, + "step": 17516 + }, + { + "epoch": 7.33, + "learning_rate": 3.597917119101335e-07, + "loss": 0.0425, + "step": 17517 + }, + { + "epoch": 7.33, + "learning_rate": 3.5934116549141027e-07, + "loss": 0.0495, + "step": 17518 + }, + { + "epoch": 7.33, + "learning_rate": 3.588908961848714e-07, + "loss": 0.0519, + "step": 17519 + }, + { + "epoch": 7.34, + "learning_rate": 3.5844090400345867e-07, + "loss": 0.0897, + "step": 17520 + }, + { + "epoch": 7.34, + "learning_rate": 3.579911889601073e-07, + "loss": 0.0397, + "step": 17521 + }, + { + "epoch": 7.34, + "learning_rate": 3.5754175106774147e-07, + "loss": 0.0531, + "step": 17522 + }, + { + "epoch": 7.34, + "learning_rate": 3.570925903392819e-07, + "loss": 0.0579, + "step": 17523 + }, + { + "epoch": 7.34, + "learning_rate": 3.5664370678763825e-07, + "loss": 0.0585, + "step": 17524 + }, + { + "epoch": 7.34, + "learning_rate": 3.561951004257147e-07, + "loss": 0.055, + "step": 17525 + }, + { + "epoch": 7.34, + "learning_rate": 3.5574677126640423e-07, + "loss": 0.0424, + "step": 17526 + }, + { + "epoch": 7.34, + "learning_rate": 3.5529871932259654e-07, + "loss": 0.0539, + "step": 17527 + }, + { + "epoch": 7.34, + "learning_rate": 3.5485094460716684e-07, + "loss": 0.0451, + "step": 17528 + }, + { + "epoch": 7.34, + "learning_rate": 3.5440344713298826e-07, + "loss": 0.0494, + "step": 17529 + }, + { + "epoch": 7.34, + "learning_rate": 3.5395622691292485e-07, + "loss": 0.0639, + "step": 17530 + }, + { + "epoch": 7.34, + "learning_rate": 3.5350928395982755e-07, + "loss": 0.0447, + "step": 17531 + }, + { + "epoch": 7.34, + "learning_rate": 3.530626182865471e-07, + "loss": 0.0295, + "step": 17532 + }, + { + "epoch": 7.34, + "learning_rate": 3.526162299059199e-07, + "loss": 0.0497, + "step": 17533 + }, + { + "epoch": 7.34, + "learning_rate": 3.5217011883077914e-07, + "loss": 0.0466, + "step": 17534 + }, + { + "epoch": 7.34, + "learning_rate": 3.5172428507394773e-07, + "loss": 0.0546, + "step": 17535 + }, + { + "epoch": 7.34, + "learning_rate": 3.5127872864823887e-07, + "loss": 0.0519, + "step": 17536 + }, + { + "epoch": 7.34, + "learning_rate": 3.5083344956646116e-07, + "loss": 0.0519, + "step": 17537 + }, + { + "epoch": 7.34, + "learning_rate": 3.5038844784141324e-07, + "loss": 0.0443, + "step": 17538 + }, + { + "epoch": 7.34, + "learning_rate": 3.4994372348588714e-07, + "loss": 0.0528, + "step": 17539 + }, + { + "epoch": 7.34, + "learning_rate": 3.494992765126637e-07, + "loss": 0.0274, + "step": 17540 + }, + { + "epoch": 7.34, + "learning_rate": 3.4905510693452273e-07, + "loss": 0.0516, + "step": 17541 + }, + { + "epoch": 7.34, + "learning_rate": 3.486112147642262e-07, + "loss": 0.0367, + "step": 17542 + }, + { + "epoch": 7.34, + "learning_rate": 3.481676000145362e-07, + "loss": 0.0601, + "step": 17543 + }, + { + "epoch": 7.35, + "learning_rate": 3.477242626982047e-07, + "loss": 0.0567, + "step": 17544 + }, + { + "epoch": 7.35, + "learning_rate": 3.4728120282797263e-07, + "loss": 0.0485, + "step": 17545 + }, + { + "epoch": 7.35, + "learning_rate": 3.4683842041657757e-07, + "loss": 0.0624, + "step": 17546 + }, + { + "epoch": 7.35, + "learning_rate": 3.463959154767449e-07, + "loss": 0.0641, + "step": 17547 + }, + { + "epoch": 7.35, + "learning_rate": 3.459536880211978e-07, + "loss": 0.035, + "step": 17548 + }, + { + "epoch": 7.35, + "learning_rate": 3.455117380626427e-07, + "loss": 0.0531, + "step": 17549 + }, + { + "epoch": 7.35, + "learning_rate": 3.450700656137862e-07, + "loss": 0.0611, + "step": 17550 + }, + { + "epoch": 7.35, + "learning_rate": 3.4462867068732253e-07, + "loss": 0.0546, + "step": 17551 + }, + { + "epoch": 7.35, + "learning_rate": 3.441875532959404e-07, + "loss": 0.0589, + "step": 17552 + }, + { + "epoch": 7.35, + "learning_rate": 3.437467134523198e-07, + "loss": 0.022, + "step": 17553 + }, + { + "epoch": 7.35, + "learning_rate": 3.4330615116912933e-07, + "loss": 0.0305, + "step": 17554 + }, + { + "epoch": 7.35, + "learning_rate": 3.428658664590356e-07, + "loss": 0.0283, + "step": 17555 + }, + { + "epoch": 7.35, + "learning_rate": 3.424258593346919e-07, + "loss": 0.0343, + "step": 17556 + }, + { + "epoch": 7.35, + "learning_rate": 3.419861298087468e-07, + "loss": 0.0567, + "step": 17557 + }, + { + "epoch": 7.35, + "learning_rate": 3.415466778938403e-07, + "loss": 0.052, + "step": 17558 + }, + { + "epoch": 7.35, + "learning_rate": 3.411075036026035e-07, + "loss": 0.0637, + "step": 17559 + }, + { + "epoch": 7.35, + "learning_rate": 3.406686069476606e-07, + "loss": 0.0489, + "step": 17560 + }, + { + "epoch": 7.35, + "learning_rate": 3.402299879416271e-07, + "loss": 0.0592, + "step": 17561 + }, + { + "epoch": 7.35, + "learning_rate": 3.3979164659711184e-07, + "loss": 0.0508, + "step": 17562 + }, + { + "epoch": 7.35, + "learning_rate": 3.393535829267114e-07, + "loss": 0.0409, + "step": 17563 + }, + { + "epoch": 7.35, + "learning_rate": 3.389157969430201e-07, + "loss": 0.0555, + "step": 17564 + }, + { + "epoch": 7.35, + "learning_rate": 3.384782886586213e-07, + "loss": 0.0502, + "step": 17565 + }, + { + "epoch": 7.35, + "learning_rate": 3.3804105808609047e-07, + "loss": 0.0659, + "step": 17566 + }, + { + "epoch": 7.35, + "learning_rate": 3.3760410523799526e-07, + "loss": 0.058, + "step": 17567 + }, + { + "epoch": 7.36, + "learning_rate": 3.371674301268979e-07, + "loss": 0.0324, + "step": 17568 + }, + { + "epoch": 7.36, + "learning_rate": 3.3673103276534616e-07, + "loss": 0.0626, + "step": 17569 + }, + { + "epoch": 7.36, + "learning_rate": 3.3629491316588546e-07, + "loss": 0.0406, + "step": 17570 + }, + { + "epoch": 7.36, + "learning_rate": 3.3585907134105253e-07, + "loss": 0.0371, + "step": 17571 + }, + { + "epoch": 7.36, + "learning_rate": 3.3542350730337516e-07, + "loss": 0.0589, + "step": 17572 + }, + { + "epoch": 7.36, + "learning_rate": 3.349882210653732e-07, + "loss": 0.0353, + "step": 17573 + }, + { + "epoch": 7.36, + "learning_rate": 3.345532126395579e-07, + "loss": 0.0669, + "step": 17574 + }, + { + "epoch": 7.36, + "learning_rate": 3.3411848203843357e-07, + "loss": 0.0595, + "step": 17575 + }, + { + "epoch": 7.36, + "learning_rate": 3.336840292744969e-07, + "loss": 0.0619, + "step": 17576 + }, + { + "epoch": 7.36, + "learning_rate": 3.332498543602347e-07, + "loss": 0.0613, + "step": 17577 + }, + { + "epoch": 7.36, + "learning_rate": 3.328159573081258e-07, + "loss": 0.0544, + "step": 17578 + }, + { + "epoch": 7.36, + "learning_rate": 3.323823381306446e-07, + "loss": 0.0532, + "step": 17579 + }, + { + "epoch": 7.36, + "learning_rate": 3.319489968402545e-07, + "loss": 0.0328, + "step": 17580 + }, + { + "epoch": 7.36, + "learning_rate": 3.3151593344941113e-07, + "loss": 0.0449, + "step": 17581 + }, + { + "epoch": 7.36, + "learning_rate": 3.310831479705634e-07, + "loss": 0.0474, + "step": 17582 + }, + { + "epoch": 7.36, + "learning_rate": 3.3065064041614914e-07, + "loss": 0.057, + "step": 17583 + }, + { + "epoch": 7.36, + "learning_rate": 3.302184107986017e-07, + "loss": 0.0557, + "step": 17584 + }, + { + "epoch": 7.36, + "learning_rate": 3.297864591303457e-07, + "loss": 0.0552, + "step": 17585 + }, + { + "epoch": 7.36, + "learning_rate": 3.2935478542379664e-07, + "loss": 0.0535, + "step": 17586 + }, + { + "epoch": 7.36, + "learning_rate": 3.289233896913624e-07, + "loss": 0.0573, + "step": 17587 + }, + { + "epoch": 7.36, + "learning_rate": 3.284922719454453e-07, + "loss": 0.0596, + "step": 17588 + }, + { + "epoch": 7.36, + "learning_rate": 3.2806143219843324e-07, + "loss": 0.0332, + "step": 17589 + }, + { + "epoch": 7.36, + "learning_rate": 3.2763087046271404e-07, + "loss": 0.0382, + "step": 17590 + }, + { + "epoch": 7.36, + "learning_rate": 3.272005867506611e-07, + "loss": 0.047, + "step": 17591 + }, + { + "epoch": 7.37, + "learning_rate": 3.2677058107464467e-07, + "loss": 0.0582, + "step": 17592 + }, + { + "epoch": 7.37, + "learning_rate": 3.263408534470236e-07, + "loss": 0.0572, + "step": 17593 + }, + { + "epoch": 7.37, + "learning_rate": 3.259114038801503e-07, + "loss": 0.0606, + "step": 17594 + }, + { + "epoch": 7.37, + "learning_rate": 3.2548223238637045e-07, + "loss": 0.0543, + "step": 17595 + }, + { + "epoch": 7.37, + "learning_rate": 3.250533389780175e-07, + "loss": 0.0382, + "step": 17596 + }, + { + "epoch": 7.37, + "learning_rate": 3.246247236674205e-07, + "loss": 0.0457, + "step": 17597 + }, + { + "epoch": 7.37, + "learning_rate": 3.241963864669018e-07, + "loss": 0.0439, + "step": 17598 + }, + { + "epoch": 7.37, + "learning_rate": 3.237683273887704e-07, + "loss": 0.0578, + "step": 17599 + }, + { + "epoch": 7.37, + "learning_rate": 3.233405464453321e-07, + "loss": 0.0534, + "step": 17600 + }, + { + "epoch": 7.37, + "learning_rate": 3.2291304364888256e-07, + "loss": 0.0599, + "step": 17601 + }, + { + "epoch": 7.37, + "learning_rate": 3.224858190117108e-07, + "loss": 0.0686, + "step": 17602 + }, + { + "epoch": 7.37, + "learning_rate": 3.220588725460971e-07, + "loss": 0.0503, + "step": 17603 + }, + { + "epoch": 7.37, + "learning_rate": 3.216322042643127e-07, + "loss": 0.0489, + "step": 17604 + }, + { + "epoch": 7.37, + "learning_rate": 3.2120581417862115e-07, + "loss": 0.0296, + "step": 17605 + }, + { + "epoch": 7.37, + "learning_rate": 3.2077970230128043e-07, + "loss": 0.0557, + "step": 17606 + }, + { + "epoch": 7.37, + "learning_rate": 3.203538686445373e-07, + "loss": 0.0553, + "step": 17607 + }, + { + "epoch": 7.37, + "learning_rate": 3.1992831322063324e-07, + "loss": 0.042, + "step": 17608 + }, + { + "epoch": 7.37, + "learning_rate": 3.1950303604180057e-07, + "loss": 0.046, + "step": 17609 + }, + { + "epoch": 7.37, + "learning_rate": 3.1907803712026173e-07, + "loss": 0.0723, + "step": 17610 + }, + { + "epoch": 7.37, + "learning_rate": 3.186533164682348e-07, + "loss": 0.0562, + "step": 17611 + }, + { + "epoch": 7.37, + "learning_rate": 3.182288740979267e-07, + "loss": 0.0347, + "step": 17612 + }, + { + "epoch": 7.37, + "learning_rate": 3.1780471002153755e-07, + "loss": 0.0586, + "step": 17613 + }, + { + "epoch": 7.37, + "learning_rate": 3.17380824251261e-07, + "loss": 0.0364, + "step": 17614 + }, + { + "epoch": 7.37, + "learning_rate": 3.169572167992818e-07, + "loss": 0.0607, + "step": 17615 + }, + { + "epoch": 7.38, + "learning_rate": 3.1653388767777237e-07, + "loss": 0.0533, + "step": 17616 + }, + { + "epoch": 7.38, + "learning_rate": 3.161108368989052e-07, + "loss": 0.0389, + "step": 17617 + }, + { + "epoch": 7.38, + "learning_rate": 3.1568806447483726e-07, + "loss": 0.0412, + "step": 17618 + }, + { + "epoch": 7.38, + "learning_rate": 3.152655704177221e-07, + "loss": 0.0473, + "step": 17619 + }, + { + "epoch": 7.38, + "learning_rate": 3.1484335473970564e-07, + "loss": 0.0459, + "step": 17620 + }, + { + "epoch": 7.38, + "learning_rate": 3.144214174529214e-07, + "loss": 0.0393, + "step": 17621 + }, + { + "epoch": 7.38, + "learning_rate": 3.139997585694998e-07, + "loss": 0.0354, + "step": 17622 + }, + { + "epoch": 7.38, + "learning_rate": 3.1357837810156e-07, + "loss": 0.0451, + "step": 17623 + }, + { + "epoch": 7.38, + "learning_rate": 3.1315727606121226e-07, + "loss": 0.0397, + "step": 17624 + }, + { + "epoch": 7.38, + "learning_rate": 3.127364524605636e-07, + "loss": 0.0487, + "step": 17625 + }, + { + "epoch": 7.38, + "learning_rate": 3.123159073117088e-07, + "loss": 0.0584, + "step": 17626 + }, + { + "epoch": 7.38, + "learning_rate": 3.1189564062673704e-07, + "loss": 0.0406, + "step": 17627 + }, + { + "epoch": 7.38, + "learning_rate": 3.1147565241772757e-07, + "loss": 0.0539, + "step": 17628 + }, + { + "epoch": 7.38, + "learning_rate": 3.110559426967552e-07, + "loss": 0.0452, + "step": 17629 + }, + { + "epoch": 7.38, + "learning_rate": 3.1063651147587913e-07, + "loss": 0.0421, + "step": 17630 + }, + { + "epoch": 7.38, + "learning_rate": 3.1021735876715973e-07, + "loss": 0.0597, + "step": 17631 + }, + { + "epoch": 7.38, + "learning_rate": 3.0979848458264293e-07, + "loss": 0.0551, + "step": 17632 + }, + { + "epoch": 7.38, + "learning_rate": 3.093798889343702e-07, + "loss": 0.0538, + "step": 17633 + }, + { + "epoch": 7.38, + "learning_rate": 3.0896157183437305e-07, + "loss": 0.0426, + "step": 17634 + }, + { + "epoch": 7.38, + "learning_rate": 3.0854353329467624e-07, + "loss": 0.0381, + "step": 17635 + }, + { + "epoch": 7.38, + "learning_rate": 3.081257733272969e-07, + "loss": 0.0315, + "step": 17636 + }, + { + "epoch": 7.38, + "learning_rate": 3.077082919442398e-07, + "loss": 0.0684, + "step": 17637 + }, + { + "epoch": 7.38, + "learning_rate": 3.072910891575076e-07, + "loss": 0.0511, + "step": 17638 + }, + { + "epoch": 7.38, + "learning_rate": 3.0687416497909185e-07, + "loss": 0.0481, + "step": 17639 + }, + { + "epoch": 7.39, + "learning_rate": 3.0645751942097623e-07, + "loss": 0.0658, + "step": 17640 + }, + { + "epoch": 7.39, + "learning_rate": 3.060411524951379e-07, + "loss": 0.0427, + "step": 17641 + }, + { + "epoch": 7.39, + "learning_rate": 3.05625064213545e-07, + "loss": 0.0536, + "step": 17642 + }, + { + "epoch": 7.39, + "learning_rate": 3.0520925458815576e-07, + "loss": 0.0519, + "step": 17643 + }, + { + "epoch": 7.39, + "learning_rate": 3.0479372363092506e-07, + "loss": 0.0414, + "step": 17644 + }, + { + "epoch": 7.39, + "learning_rate": 3.043784713537945e-07, + "loss": 0.0615, + "step": 17645 + }, + { + "epoch": 7.39, + "learning_rate": 3.0396349776870004e-07, + "loss": 0.0434, + "step": 17646 + }, + { + "epoch": 7.39, + "learning_rate": 3.0354880288757214e-07, + "loss": 0.0634, + "step": 17647 + }, + { + "epoch": 7.39, + "learning_rate": 3.0313438672232796e-07, + "loss": 0.0856, + "step": 17648 + }, + { + "epoch": 7.39, + "learning_rate": 3.0272024928488244e-07, + "loss": 0.0379, + "step": 17649 + }, + { + "epoch": 7.39, + "learning_rate": 3.023063905871382e-07, + "loss": 0.0397, + "step": 17650 + }, + { + "epoch": 7.39, + "learning_rate": 3.018928106409902e-07, + "loss": 0.0358, + "step": 17651 + }, + { + "epoch": 7.39, + "learning_rate": 3.014795094583278e-07, + "loss": 0.039, + "step": 17652 + }, + { + "epoch": 7.39, + "learning_rate": 3.0106648705103045e-07, + "loss": 0.0523, + "step": 17653 + }, + { + "epoch": 7.39, + "learning_rate": 3.0065374343096974e-07, + "loss": 0.0526, + "step": 17654 + }, + { + "epoch": 7.39, + "learning_rate": 3.002412786100106e-07, + "loss": 0.0791, + "step": 17655 + }, + { + "epoch": 7.39, + "learning_rate": 2.9982909260001023e-07, + "loss": 0.0536, + "step": 17656 + }, + { + "epoch": 7.39, + "learning_rate": 2.994171854128125e-07, + "loss": 0.0419, + "step": 17657 + }, + { + "epoch": 7.39, + "learning_rate": 2.9900555706026015e-07, + "loss": 0.036, + "step": 17658 + }, + { + "epoch": 7.39, + "learning_rate": 2.985942075541848e-07, + "loss": 0.0623, + "step": 17659 + }, + { + "epoch": 7.39, + "learning_rate": 2.981831369064103e-07, + "loss": 0.051, + "step": 17660 + }, + { + "epoch": 7.39, + "learning_rate": 2.9777234512875174e-07, + "loss": 0.0405, + "step": 17661 + }, + { + "epoch": 7.39, + "learning_rate": 2.973618322330174e-07, + "loss": 0.0401, + "step": 17662 + }, + { + "epoch": 7.4, + "learning_rate": 2.9695159823100896e-07, + "loss": 0.0441, + "step": 17663 + }, + { + "epoch": 7.4, + "learning_rate": 2.965416431345147e-07, + "loss": 0.0691, + "step": 17664 + }, + { + "epoch": 7.4, + "learning_rate": 2.9613196695532085e-07, + "loss": 0.0394, + "step": 17665 + }, + { + "epoch": 7.4, + "learning_rate": 2.9572256970520353e-07, + "loss": 0.0536, + "step": 17666 + }, + { + "epoch": 7.4, + "learning_rate": 2.953134513959277e-07, + "loss": 0.0639, + "step": 17667 + }, + { + "epoch": 7.4, + "learning_rate": 2.9490461203925513e-07, + "loss": 0.0285, + "step": 17668 + }, + { + "epoch": 7.4, + "learning_rate": 2.944960516469375e-07, + "loss": 0.0541, + "step": 17669 + }, + { + "epoch": 7.4, + "learning_rate": 2.940877702307199e-07, + "loss": 0.0451, + "step": 17670 + }, + { + "epoch": 7.4, + "learning_rate": 2.936797678023351e-07, + "loss": 0.0423, + "step": 17671 + }, + { + "epoch": 7.4, + "learning_rate": 2.9327204437351266e-07, + "loss": 0.0591, + "step": 17672 + }, + { + "epoch": 7.4, + "learning_rate": 2.9286459995597096e-07, + "loss": 0.0501, + "step": 17673 + }, + { + "epoch": 7.4, + "learning_rate": 2.9245743456142283e-07, + "loss": 0.0577, + "step": 17674 + }, + { + "epoch": 7.4, + "learning_rate": 2.9205054820157007e-07, + "loss": 0.0485, + "step": 17675 + }, + { + "epoch": 7.4, + "learning_rate": 2.916439408881111e-07, + "loss": 0.0518, + "step": 17676 + }, + { + "epoch": 7.4, + "learning_rate": 2.912376126327332e-07, + "loss": 0.0455, + "step": 17677 + }, + { + "epoch": 7.4, + "learning_rate": 2.908315634471126e-07, + "loss": 0.0631, + "step": 17678 + }, + { + "epoch": 7.4, + "learning_rate": 2.9042579334292333e-07, + "loss": 0.0396, + "step": 17679 + }, + { + "epoch": 7.4, + "learning_rate": 2.900203023318293e-07, + "loss": 0.0586, + "step": 17680 + }, + { + "epoch": 7.4, + "learning_rate": 2.8961509042548466e-07, + "loss": 0.0546, + "step": 17681 + }, + { + "epoch": 7.4, + "learning_rate": 2.892101576355366e-07, + "loss": 0.0392, + "step": 17682 + }, + { + "epoch": 7.4, + "learning_rate": 2.8880550397362705e-07, + "loss": 0.0524, + "step": 17683 + }, + { + "epoch": 7.4, + "learning_rate": 2.8840112945138553e-07, + "loss": 0.0387, + "step": 17684 + }, + { + "epoch": 7.4, + "learning_rate": 2.879970340804339e-07, + "loss": 0.0469, + "step": 17685 + }, + { + "epoch": 7.4, + "learning_rate": 2.875932178723906e-07, + "loss": 0.0548, + "step": 17686 + }, + { + "epoch": 7.41, + "learning_rate": 2.871896808388608e-07, + "loss": 0.0616, + "step": 17687 + }, + { + "epoch": 7.41, + "learning_rate": 2.867864229914452e-07, + "loss": 0.0609, + "step": 17688 + }, + { + "epoch": 7.41, + "learning_rate": 2.863834443417357e-07, + "loss": 0.0426, + "step": 17689 + }, + { + "epoch": 7.41, + "learning_rate": 2.85980744901313e-07, + "loss": 0.0461, + "step": 17690 + }, + { + "epoch": 7.41, + "learning_rate": 2.855783246817545e-07, + "loss": 0.0395, + "step": 17691 + }, + { + "epoch": 7.41, + "learning_rate": 2.851761836946265e-07, + "loss": 0.0499, + "step": 17692 + }, + { + "epoch": 7.41, + "learning_rate": 2.847743219514876e-07, + "loss": 0.0421, + "step": 17693 + }, + { + "epoch": 7.41, + "learning_rate": 2.8437273946388954e-07, + "loss": 0.0492, + "step": 17694 + }, + { + "epoch": 7.41, + "learning_rate": 2.839714362433754e-07, + "loss": 0.0464, + "step": 17695 + }, + { + "epoch": 7.41, + "learning_rate": 2.8357041230148153e-07, + "loss": 0.0623, + "step": 17696 + }, + { + "epoch": 7.41, + "learning_rate": 2.8316966764973417e-07, + "loss": 0.0539, + "step": 17697 + }, + { + "epoch": 7.41, + "learning_rate": 2.827692022996509e-07, + "loss": 0.0522, + "step": 17698 + }, + { + "epoch": 7.41, + "learning_rate": 2.823690162627446e-07, + "loss": 0.028, + "step": 17699 + }, + { + "epoch": 7.41, + "learning_rate": 2.819691095505173e-07, + "loss": 0.0448, + "step": 17700 + }, + { + "epoch": 7.41, + "learning_rate": 2.815694821744652e-07, + "loss": 0.0507, + "step": 17701 + }, + { + "epoch": 7.41, + "learning_rate": 2.811701341460737e-07, + "loss": 0.0512, + "step": 17702 + }, + { + "epoch": 7.41, + "learning_rate": 2.807710654768214e-07, + "loss": 0.0467, + "step": 17703 + }, + { + "epoch": 7.41, + "learning_rate": 2.803722761781824e-07, + "loss": 0.0349, + "step": 17704 + }, + { + "epoch": 7.41, + "learning_rate": 2.7997376626161644e-07, + "loss": 0.0453, + "step": 17705 + }, + { + "epoch": 7.41, + "learning_rate": 2.7957553573857876e-07, + "loss": 0.0674, + "step": 17706 + }, + { + "epoch": 7.41, + "learning_rate": 2.791775846205158e-07, + "loss": 0.0486, + "step": 17707 + }, + { + "epoch": 7.41, + "learning_rate": 2.787799129188673e-07, + "loss": 0.0415, + "step": 17708 + }, + { + "epoch": 7.41, + "learning_rate": 2.783825206450641e-07, + "loss": 0.0466, + "step": 17709 + }, + { + "epoch": 7.41, + "learning_rate": 2.779854078105304e-07, + "loss": 0.0504, + "step": 17710 + }, + { + "epoch": 7.42, + "learning_rate": 2.775885744266782e-07, + "loss": 0.0748, + "step": 17711 + }, + { + "epoch": 7.42, + "learning_rate": 2.771920205049139e-07, + "loss": 0.0579, + "step": 17712 + }, + { + "epoch": 7.42, + "learning_rate": 2.767957460566373e-07, + "loss": 0.05, + "step": 17713 + }, + { + "epoch": 7.42, + "learning_rate": 2.763997510932381e-07, + "loss": 0.0371, + "step": 17714 + }, + { + "epoch": 7.42, + "learning_rate": 2.7600403562609955e-07, + "loss": 0.0392, + "step": 17715 + }, + { + "epoch": 7.42, + "learning_rate": 2.756085996665969e-07, + "loss": 0.0545, + "step": 17716 + }, + { + "epoch": 7.42, + "learning_rate": 2.752134432260956e-07, + "loss": 0.0601, + "step": 17717 + }, + { + "epoch": 7.42, + "learning_rate": 2.7481856631595415e-07, + "loss": 0.0515, + "step": 17718 + }, + { + "epoch": 7.42, + "learning_rate": 2.7442396894752365e-07, + "loss": 0.0655, + "step": 17719 + }, + { + "epoch": 7.42, + "learning_rate": 2.7402965113214495e-07, + "loss": 0.0449, + "step": 17720 + }, + { + "epoch": 7.42, + "learning_rate": 2.7363561288115347e-07, + "loss": 0.039, + "step": 17721 + }, + { + "epoch": 7.42, + "learning_rate": 2.7324185420587457e-07, + "loss": 0.0347, + "step": 17722 + }, + { + "epoch": 7.42, + "learning_rate": 2.728483751176281e-07, + "loss": 0.0471, + "step": 17723 + }, + { + "epoch": 7.42, + "learning_rate": 2.7245517562772385e-07, + "loss": 0.0554, + "step": 17724 + }, + { + "epoch": 7.42, + "learning_rate": 2.720622557474628e-07, + "loss": 0.0563, + "step": 17725 + }, + { + "epoch": 7.42, + "learning_rate": 2.716696154881393e-07, + "loss": 0.0543, + "step": 17726 + }, + { + "epoch": 7.42, + "learning_rate": 2.7127725486104094e-07, + "loss": 0.067, + "step": 17727 + }, + { + "epoch": 7.42, + "learning_rate": 2.7088517387744426e-07, + "loss": 0.0382, + "step": 17728 + }, + { + "epoch": 7.42, + "learning_rate": 2.704933725486192e-07, + "loss": 0.0575, + "step": 17729 + }, + { + "epoch": 7.42, + "learning_rate": 2.7010185088582886e-07, + "loss": 0.0454, + "step": 17730 + }, + { + "epoch": 7.42, + "learning_rate": 2.697106089003276e-07, + "loss": 0.0555, + "step": 17731 + }, + { + "epoch": 7.42, + "learning_rate": 2.6931964660335985e-07, + "loss": 0.0613, + "step": 17732 + }, + { + "epoch": 7.42, + "learning_rate": 2.689289640061632e-07, + "loss": 0.0592, + "step": 17733 + }, + { + "epoch": 7.42, + "learning_rate": 2.685385611199698e-07, + "loss": 0.0351, + "step": 17734 + }, + { + "epoch": 7.43, + "learning_rate": 2.681484379559984e-07, + "loss": 0.0521, + "step": 17735 + }, + { + "epoch": 7.43, + "learning_rate": 2.677585945254646e-07, + "loss": 0.065, + "step": 17736 + }, + { + "epoch": 7.43, + "learning_rate": 2.673690308395738e-07, + "loss": 0.045, + "step": 17737 + }, + { + "epoch": 7.43, + "learning_rate": 2.6697974690952477e-07, + "loss": 0.03, + "step": 17738 + }, + { + "epoch": 7.43, + "learning_rate": 2.6659074274650533e-07, + "loss": 0.0582, + "step": 17739 + }, + { + "epoch": 7.43, + "learning_rate": 2.662020183616976e-07, + "loss": 0.0468, + "step": 17740 + }, + { + "epoch": 7.43, + "learning_rate": 2.6581357376627483e-07, + "loss": 0.0598, + "step": 17741 + }, + { + "epoch": 7.43, + "learning_rate": 2.6542540897140256e-07, + "loss": 0.0732, + "step": 17742 + }, + { + "epoch": 7.43, + "learning_rate": 2.650375239882386e-07, + "loss": 0.0396, + "step": 17743 + }, + { + "epoch": 7.43, + "learning_rate": 2.646499188279328e-07, + "loss": 0.0468, + "step": 17744 + }, + { + "epoch": 7.43, + "learning_rate": 2.642625935016274e-07, + "loss": 0.0367, + "step": 17745 + }, + { + "epoch": 7.43, + "learning_rate": 2.6387554802045246e-07, + "loss": 0.0477, + "step": 17746 + }, + { + "epoch": 7.43, + "learning_rate": 2.6348878239553677e-07, + "loss": 0.0628, + "step": 17747 + }, + { + "epoch": 7.43, + "learning_rate": 2.631022966379948e-07, + "loss": 0.0582, + "step": 17748 + }, + { + "epoch": 7.43, + "learning_rate": 2.6271609075893656e-07, + "loss": 0.0567, + "step": 17749 + }, + { + "epoch": 7.43, + "learning_rate": 2.6233016476946427e-07, + "loss": 0.0632, + "step": 17750 + }, + { + "epoch": 7.43, + "learning_rate": 2.619445186806713e-07, + "loss": 0.0567, + "step": 17751 + }, + { + "epoch": 7.43, + "learning_rate": 2.6155915250364094e-07, + "loss": 0.0392, + "step": 17752 + }, + { + "epoch": 7.43, + "learning_rate": 2.611740662494511e-07, + "loss": 0.0517, + "step": 17753 + }, + { + "epoch": 7.43, + "learning_rate": 2.607892599291695e-07, + "loss": 0.0436, + "step": 17754 + }, + { + "epoch": 7.43, + "learning_rate": 2.6040473355385954e-07, + "loss": 0.0579, + "step": 17755 + }, + { + "epoch": 7.43, + "learning_rate": 2.600204871345724e-07, + "loss": 0.0332, + "step": 17756 + }, + { + "epoch": 7.43, + "learning_rate": 2.596365206823537e-07, + "loss": 0.0377, + "step": 17757 + }, + { + "epoch": 7.43, + "learning_rate": 2.59252834208239e-07, + "loss": 0.0295, + "step": 17758 + }, + { + "epoch": 7.44, + "learning_rate": 2.588694277232595e-07, + "loss": 0.0364, + "step": 17759 + }, + { + "epoch": 7.44, + "learning_rate": 2.58486301238432e-07, + "loss": 0.0565, + "step": 17760 + }, + { + "epoch": 7.44, + "learning_rate": 2.5810345476477207e-07, + "loss": 0.0415, + "step": 17761 + }, + { + "epoch": 7.44, + "learning_rate": 2.5772088831328313e-07, + "loss": 0.0432, + "step": 17762 + }, + { + "epoch": 7.44, + "learning_rate": 2.5733860189496306e-07, + "loss": 0.0583, + "step": 17763 + }, + { + "epoch": 7.44, + "learning_rate": 2.5695659552079866e-07, + "loss": 0.0525, + "step": 17764 + }, + { + "epoch": 7.44, + "learning_rate": 2.565748692017722e-07, + "loss": 0.0465, + "step": 17765 + }, + { + "epoch": 7.44, + "learning_rate": 2.5619342294885387e-07, + "loss": 0.0576, + "step": 17766 + }, + { + "epoch": 7.44, + "learning_rate": 2.5581225677300925e-07, + "loss": 0.0588, + "step": 17767 + }, + { + "epoch": 7.44, + "learning_rate": 2.5543137068519407e-07, + "loss": 0.0535, + "step": 17768 + }, + { + "epoch": 7.44, + "learning_rate": 2.5505076469635734e-07, + "loss": 0.0495, + "step": 17769 + }, + { + "epoch": 7.44, + "learning_rate": 2.5467043881743814e-07, + "loss": 0.0792, + "step": 17770 + }, + { + "epoch": 7.44, + "learning_rate": 2.5429039305937095e-07, + "loss": 0.0626, + "step": 17771 + }, + { + "epoch": 7.44, + "learning_rate": 2.5391062743307717e-07, + "loss": 0.0569, + "step": 17772 + }, + { + "epoch": 7.44, + "learning_rate": 2.5353114194947457e-07, + "loss": 0.0197, + "step": 17773 + }, + { + "epoch": 7.44, + "learning_rate": 2.5315193661947013e-07, + "loss": 0.0386, + "step": 17774 + }, + { + "epoch": 7.44, + "learning_rate": 2.5277301145396283e-07, + "loss": 0.0483, + "step": 17775 + }, + { + "epoch": 7.44, + "learning_rate": 2.523943664638473e-07, + "loss": 0.0573, + "step": 17776 + }, + { + "epoch": 7.44, + "learning_rate": 2.5201600166000485e-07, + "loss": 0.0374, + "step": 17777 + }, + { + "epoch": 7.44, + "learning_rate": 2.5163791705331343e-07, + "loss": 0.0404, + "step": 17778 + }, + { + "epoch": 7.44, + "learning_rate": 2.512601126546399e-07, + "loss": 0.0402, + "step": 17779 + }, + { + "epoch": 7.44, + "learning_rate": 2.508825884748423e-07, + "loss": 0.0507, + "step": 17780 + }, + { + "epoch": 7.44, + "learning_rate": 2.5050534452477406e-07, + "loss": 0.0486, + "step": 17781 + }, + { + "epoch": 7.44, + "learning_rate": 2.5012838081527766e-07, + "loss": 0.0545, + "step": 17782 + }, + { + "epoch": 7.45, + "learning_rate": 2.4975169735718896e-07, + "loss": 0.0653, + "step": 17783 + }, + { + "epoch": 7.45, + "learning_rate": 2.493752941613348e-07, + "loss": 0.0716, + "step": 17784 + }, + { + "epoch": 7.45, + "learning_rate": 2.4899917123853647e-07, + "loss": 0.0518, + "step": 17785 + }, + { + "epoch": 7.45, + "learning_rate": 2.4862332859960426e-07, + "loss": 0.0181, + "step": 17786 + }, + { + "epoch": 7.45, + "learning_rate": 2.4824776625534064e-07, + "loss": 0.0417, + "step": 17787 + }, + { + "epoch": 7.45, + "learning_rate": 2.478724842165414e-07, + "loss": 0.0314, + "step": 17788 + }, + { + "epoch": 7.45, + "learning_rate": 2.4749748249399463e-07, + "loss": 0.034, + "step": 17789 + }, + { + "epoch": 7.45, + "learning_rate": 2.4712276109847723e-07, + "loss": 0.0728, + "step": 17790 + }, + { + "epoch": 7.45, + "learning_rate": 2.4674832004076166e-07, + "loss": 0.0472, + "step": 17791 + }, + { + "epoch": 7.45, + "learning_rate": 2.4637415933161155e-07, + "loss": 0.0512, + "step": 17792 + }, + { + "epoch": 7.45, + "learning_rate": 2.460002789817806e-07, + "loss": 0.0484, + "step": 17793 + }, + { + "epoch": 7.45, + "learning_rate": 2.4562667900201677e-07, + "loss": 0.0547, + "step": 17794 + }, + { + "epoch": 7.45, + "learning_rate": 2.4525335940305704e-07, + "loss": 0.0613, + "step": 17795 + }, + { + "epoch": 7.45, + "learning_rate": 2.44880320195634e-07, + "loss": 0.0636, + "step": 17796 + }, + { + "epoch": 7.45, + "learning_rate": 2.4450756139046906e-07, + "loss": 0.051, + "step": 17797 + }, + { + "epoch": 7.45, + "learning_rate": 2.4413508299827805e-07, + "loss": 0.0516, + "step": 17798 + }, + { + "epoch": 7.45, + "learning_rate": 2.4376288502976685e-07, + "loss": 0.0318, + "step": 17799 + }, + { + "epoch": 7.45, + "learning_rate": 2.433909674956336e-07, + "loss": 0.0468, + "step": 17800 + }, + { + "epoch": 7.45, + "learning_rate": 2.430193304065687e-07, + "loss": 0.0524, + "step": 17801 + }, + { + "epoch": 7.45, + "learning_rate": 2.4264797377325566e-07, + "loss": 0.0567, + "step": 17802 + }, + { + "epoch": 7.45, + "learning_rate": 2.422768976063683e-07, + "loss": 0.0401, + "step": 17803 + }, + { + "epoch": 7.45, + "learning_rate": 2.4190610191657246e-07, + "loss": 0.0437, + "step": 17804 + }, + { + "epoch": 7.45, + "learning_rate": 2.4153558671452524e-07, + "loss": 0.032, + "step": 17805 + }, + { + "epoch": 7.45, + "learning_rate": 2.411653520108803e-07, + "loss": 0.0429, + "step": 17806 + }, + { + "epoch": 7.46, + "learning_rate": 2.4079539781627693e-07, + "loss": 0.0478, + "step": 17807 + }, + { + "epoch": 7.46, + "learning_rate": 2.4042572414134877e-07, + "loss": 0.041, + "step": 17808 + }, + { + "epoch": 7.46, + "learning_rate": 2.4005633099672296e-07, + "loss": 0.045, + "step": 17809 + }, + { + "epoch": 7.46, + "learning_rate": 2.396872183930166e-07, + "loss": 0.0517, + "step": 17810 + }, + { + "epoch": 7.46, + "learning_rate": 2.3931838634083994e-07, + "loss": 0.0358, + "step": 17811 + }, + { + "epoch": 7.46, + "learning_rate": 2.3894983485079683e-07, + "loss": 0.0513, + "step": 17812 + }, + { + "epoch": 7.46, + "learning_rate": 2.3858156393347654e-07, + "loss": 0.0579, + "step": 17813 + }, + { + "epoch": 7.46, + "learning_rate": 2.3821357359946838e-07, + "loss": 0.0584, + "step": 17814 + }, + { + "epoch": 7.46, + "learning_rate": 2.3784586385934839e-07, + "loss": 0.0381, + "step": 17815 + }, + { + "epoch": 7.46, + "learning_rate": 2.374784347236858e-07, + "loss": 0.052, + "step": 17816 + }, + { + "epoch": 7.46, + "learning_rate": 2.3711128620304226e-07, + "loss": 0.046, + "step": 17817 + }, + { + "epoch": 7.46, + "learning_rate": 2.3674441830797147e-07, + "loss": 0.0637, + "step": 17818 + }, + { + "epoch": 7.46, + "learning_rate": 2.3637783104902057e-07, + "loss": 0.053, + "step": 17819 + }, + { + "epoch": 7.46, + "learning_rate": 2.3601152443672336e-07, + "loss": 0.0594, + "step": 17820 + }, + { + "epoch": 7.46, + "learning_rate": 2.3564549848161033e-07, + "loss": 0.0661, + "step": 17821 + }, + { + "epoch": 7.46, + "learning_rate": 2.3527975319420192e-07, + "loss": 0.0536, + "step": 17822 + }, + { + "epoch": 7.46, + "learning_rate": 2.349142885850131e-07, + "loss": 0.0418, + "step": 17823 + }, + { + "epoch": 7.46, + "learning_rate": 2.345491046645465e-07, + "loss": 0.05, + "step": 17824 + }, + { + "epoch": 7.46, + "learning_rate": 2.3418420144330268e-07, + "loss": 0.0583, + "step": 17825 + }, + { + "epoch": 7.46, + "learning_rate": 2.3381957893176543e-07, + "loss": 0.0828, + "step": 17826 + }, + { + "epoch": 7.46, + "learning_rate": 2.3345523714041973e-07, + "loss": 0.043, + "step": 17827 + }, + { + "epoch": 7.46, + "learning_rate": 2.330911760797361e-07, + "loss": 0.0615, + "step": 17828 + }, + { + "epoch": 7.46, + "learning_rate": 2.3272739576017945e-07, + "loss": 0.0439, + "step": 17829 + }, + { + "epoch": 7.46, + "learning_rate": 2.323638961922059e-07, + "loss": 0.0408, + "step": 17830 + }, + { + "epoch": 7.47, + "learning_rate": 2.3200067738626487e-07, + "loss": 0.0589, + "step": 17831 + }, + { + "epoch": 7.47, + "learning_rate": 2.3163773935279577e-07, + "loss": 0.0638, + "step": 17832 + }, + { + "epoch": 7.47, + "learning_rate": 2.3127508210223359e-07, + "loss": 0.0529, + "step": 17833 + }, + { + "epoch": 7.47, + "learning_rate": 2.309127056449989e-07, + "loss": 0.0459, + "step": 17834 + }, + { + "epoch": 7.47, + "learning_rate": 2.3055060999151003e-07, + "loss": 0.0486, + "step": 17835 + }, + { + "epoch": 7.47, + "learning_rate": 2.3018879515217418e-07, + "loss": 0.0327, + "step": 17836 + }, + { + "epoch": 7.47, + "learning_rate": 2.2982726113739195e-07, + "loss": 0.0491, + "step": 17837 + }, + { + "epoch": 7.47, + "learning_rate": 2.2946600795755502e-07, + "loss": 0.0493, + "step": 17838 + }, + { + "epoch": 7.47, + "learning_rate": 2.2910503562304843e-07, + "loss": 0.0534, + "step": 17839 + }, + { + "epoch": 7.47, + "learning_rate": 2.2874434414424606e-07, + "loss": 0.0509, + "step": 17840 + }, + { + "epoch": 7.47, + "learning_rate": 2.2838393353151745e-07, + "loss": 0.0737, + "step": 17841 + }, + { + "epoch": 7.47, + "learning_rate": 2.2802380379522093e-07, + "loss": 0.0368, + "step": 17842 + }, + { + "epoch": 7.47, + "learning_rate": 2.2766395494570825e-07, + "loss": 0.0512, + "step": 17843 + }, + { + "epoch": 7.47, + "learning_rate": 2.2730438699332446e-07, + "loss": 0.0422, + "step": 17844 + }, + { + "epoch": 7.47, + "learning_rate": 2.2694509994840353e-07, + "loss": 0.0436, + "step": 17845 + }, + { + "epoch": 7.47, + "learning_rate": 2.2658609382127384e-07, + "loss": 0.0599, + "step": 17846 + }, + { + "epoch": 7.47, + "learning_rate": 2.2622736862225382e-07, + "loss": 0.0471, + "step": 17847 + }, + { + "epoch": 7.47, + "learning_rate": 2.258689243616541e-07, + "loss": 0.0448, + "step": 17848 + }, + { + "epoch": 7.47, + "learning_rate": 2.2551076104977975e-07, + "loss": 0.0355, + "step": 17849 + }, + { + "epoch": 7.47, + "learning_rate": 2.251528786969237e-07, + "loss": 0.0359, + "step": 17850 + }, + { + "epoch": 7.47, + "learning_rate": 2.2479527731337437e-07, + "loss": 0.0344, + "step": 17851 + }, + { + "epoch": 7.47, + "learning_rate": 2.2443795690941127e-07, + "loss": 0.0577, + "step": 17852 + }, + { + "epoch": 7.47, + "learning_rate": 2.2408091749530402e-07, + "loss": 0.059, + "step": 17853 + }, + { + "epoch": 7.47, + "learning_rate": 2.237241590813155e-07, + "loss": 0.0708, + "step": 17854 + }, + { + "epoch": 7.48, + "learning_rate": 2.2336768167770084e-07, + "loss": 0.0472, + "step": 17855 + }, + { + "epoch": 7.48, + "learning_rate": 2.230114852947074e-07, + "loss": 0.0545, + "step": 17856 + }, + { + "epoch": 7.48, + "learning_rate": 2.2265556994257144e-07, + "loss": 0.0575, + "step": 17857 + }, + { + "epoch": 7.48, + "learning_rate": 2.2229993563152585e-07, + "loss": 0.0539, + "step": 17858 + }, + { + "epoch": 7.48, + "learning_rate": 2.2194458237179139e-07, + "loss": 0.0631, + "step": 17859 + }, + { + "epoch": 7.48, + "learning_rate": 2.2158951017358432e-07, + "loss": 0.0624, + "step": 17860 + }, + { + "epoch": 7.48, + "learning_rate": 2.2123471904710757e-07, + "loss": 0.0565, + "step": 17861 + }, + { + "epoch": 7.48, + "learning_rate": 2.20880209002563e-07, + "loss": 0.0399, + "step": 17862 + }, + { + "epoch": 7.48, + "learning_rate": 2.2052598005013803e-07, + "loss": 0.047, + "step": 17863 + }, + { + "epoch": 7.48, + "learning_rate": 2.2017203220001559e-07, + "loss": 0.0494, + "step": 17864 + }, + { + "epoch": 7.48, + "learning_rate": 2.1981836546236979e-07, + "loss": 0.0559, + "step": 17865 + }, + { + "epoch": 7.48, + "learning_rate": 2.1946497984736693e-07, + "loss": 0.0383, + "step": 17866 + }, + { + "epoch": 7.48, + "learning_rate": 2.1911187536516332e-07, + "loss": 0.0442, + "step": 17867 + }, + { + "epoch": 7.48, + "learning_rate": 2.1875905202590974e-07, + "loss": 0.0464, + "step": 17868 + }, + { + "epoch": 7.48, + "learning_rate": 2.1840650983974698e-07, + "loss": 0.0628, + "step": 17869 + }, + { + "epoch": 7.48, + "learning_rate": 2.1805424881681137e-07, + "loss": 0.036, + "step": 17870 + }, + { + "epoch": 7.48, + "learning_rate": 2.1770226896722369e-07, + "loss": 0.0673, + "step": 17871 + }, + { + "epoch": 7.48, + "learning_rate": 2.1735057030110362e-07, + "loss": 0.0502, + "step": 17872 + }, + { + "epoch": 7.48, + "learning_rate": 2.1699915282856087e-07, + "loss": 0.0514, + "step": 17873 + }, + { + "epoch": 7.48, + "learning_rate": 2.1664801655969735e-07, + "loss": 0.0612, + "step": 17874 + }, + { + "epoch": 7.48, + "learning_rate": 2.162971615046039e-07, + "loss": 0.048, + "step": 17875 + }, + { + "epoch": 7.48, + "learning_rate": 2.1594658767336685e-07, + "loss": 0.0522, + "step": 17876 + }, + { + "epoch": 7.48, + "learning_rate": 2.155962950760626e-07, + "loss": 0.0434, + "step": 17877 + }, + { + "epoch": 7.49, + "learning_rate": 2.152462837227598e-07, + "loss": 0.0527, + "step": 17878 + }, + { + "epoch": 7.49, + "learning_rate": 2.1489655362352034e-07, + "loss": 0.048, + "step": 17879 + }, + { + "epoch": 7.49, + "learning_rate": 2.1454710478839736e-07, + "loss": 0.0406, + "step": 17880 + }, + { + "epoch": 7.49, + "learning_rate": 2.1419793722743388e-07, + "loss": 0.044, + "step": 17881 + }, + { + "epoch": 7.49, + "learning_rate": 2.1384905095066634e-07, + "loss": 0.0478, + "step": 17882 + }, + { + "epoch": 7.49, + "learning_rate": 2.1350044596812335e-07, + "loss": 0.0491, + "step": 17883 + }, + { + "epoch": 7.49, + "learning_rate": 2.1315212228982584e-07, + "loss": 0.0352, + "step": 17884 + }, + { + "epoch": 7.49, + "learning_rate": 2.1280407992578578e-07, + "loss": 0.0437, + "step": 17885 + }, + { + "epoch": 7.49, + "learning_rate": 2.1245631888600848e-07, + "loss": 0.0512, + "step": 17886 + }, + { + "epoch": 7.49, + "learning_rate": 2.121088391804893e-07, + "loss": 0.0427, + "step": 17887 + }, + { + "epoch": 7.49, + "learning_rate": 2.117616408192147e-07, + "loss": 0.0554, + "step": 17888 + }, + { + "epoch": 7.49, + "learning_rate": 2.1141472381216554e-07, + "loss": 0.037, + "step": 17889 + }, + { + "epoch": 7.49, + "learning_rate": 2.11068088169315e-07, + "loss": 0.0567, + "step": 17890 + }, + { + "epoch": 7.49, + "learning_rate": 2.107217339006251e-07, + "loss": 0.0412, + "step": 17891 + }, + { + "epoch": 7.49, + "learning_rate": 2.103756610160512e-07, + "loss": 0.0677, + "step": 17892 + }, + { + "epoch": 7.49, + "learning_rate": 2.1002986952554428e-07, + "loss": 0.0494, + "step": 17893 + }, + { + "epoch": 7.49, + "learning_rate": 2.0968435943904075e-07, + "loss": 0.0517, + "step": 17894 + }, + { + "epoch": 7.49, + "learning_rate": 2.093391307664716e-07, + "loss": 0.0252, + "step": 17895 + }, + { + "epoch": 7.49, + "learning_rate": 2.0899418351776e-07, + "loss": 0.0519, + "step": 17896 + }, + { + "epoch": 7.49, + "learning_rate": 2.086495177028236e-07, + "loss": 0.0646, + "step": 17897 + }, + { + "epoch": 7.49, + "learning_rate": 2.0830513333156776e-07, + "loss": 0.0583, + "step": 17898 + }, + { + "epoch": 7.49, + "learning_rate": 2.0796103041389126e-07, + "loss": 0.0538, + "step": 17899 + }, + { + "epoch": 7.49, + "learning_rate": 2.0761720895968617e-07, + "loss": 0.0497, + "step": 17900 + }, + { + "epoch": 7.49, + "learning_rate": 2.0727366897883571e-07, + "loss": 0.0612, + "step": 17901 + }, + { + "epoch": 7.5, + "learning_rate": 2.069304104812131e-07, + "loss": 0.0552, + "step": 17902 + }, + { + "epoch": 7.5, + "learning_rate": 2.0658743347668487e-07, + "loss": 0.0488, + "step": 17903 + }, + { + "epoch": 7.5, + "learning_rate": 2.062447379751109e-07, + "loss": 0.0491, + "step": 17904 + }, + { + "epoch": 7.5, + "learning_rate": 2.0590232398634114e-07, + "loss": 0.0561, + "step": 17905 + }, + { + "epoch": 7.5, + "learning_rate": 2.0556019152021767e-07, + "loss": 0.0515, + "step": 17906 + }, + { + "epoch": 7.5, + "learning_rate": 2.0521834058657597e-07, + "loss": 0.0684, + "step": 17907 + }, + { + "epoch": 7.5, + "learning_rate": 2.0487677119524152e-07, + "loss": 0.043, + "step": 17908 + }, + { + "epoch": 7.5, + "learning_rate": 2.045354833560309e-07, + "loss": 0.0524, + "step": 17909 + }, + { + "epoch": 7.5, + "learning_rate": 2.0419447707875518e-07, + "loss": 0.053, + "step": 17910 + }, + { + "epoch": 7.5, + "learning_rate": 2.0385375237321759e-07, + "loss": 0.0464, + "step": 17911 + }, + { + "epoch": 7.5, + "learning_rate": 2.0351330924921031e-07, + "loss": 0.0389, + "step": 17912 + }, + { + "epoch": 7.5, + "learning_rate": 2.0317314771651996e-07, + "loss": 0.0343, + "step": 17913 + }, + { + "epoch": 7.5, + "learning_rate": 2.028332677849254e-07, + "loss": 0.0581, + "step": 17914 + }, + { + "epoch": 7.5, + "learning_rate": 2.0249366946419436e-07, + "loss": 0.0411, + "step": 17915 + }, + { + "epoch": 7.5, + "learning_rate": 2.021543527640879e-07, + "loss": 0.0523, + "step": 17916 + }, + { + "epoch": 7.5, + "learning_rate": 2.0181531769435936e-07, + "loss": 0.0686, + "step": 17917 + }, + { + "epoch": 7.5, + "learning_rate": 2.0147656426475537e-07, + "loss": 0.0555, + "step": 17918 + }, + { + "epoch": 7.5, + "learning_rate": 2.011380924850115e-07, + "loss": 0.0443, + "step": 17919 + }, + { + "epoch": 7.5, + "learning_rate": 2.007999023648588e-07, + "loss": 0.0613, + "step": 17920 + }, + { + "epoch": 7.5, + "learning_rate": 2.0046199391401844e-07, + "loss": 0.039, + "step": 17921 + }, + { + "epoch": 7.5, + "learning_rate": 2.001243671422004e-07, + "loss": 0.0493, + "step": 17922 + }, + { + "epoch": 7.5, + "learning_rate": 1.9978702205911137e-07, + "loss": 0.0779, + "step": 17923 + }, + { + "epoch": 7.5, + "learning_rate": 1.9944995867444806e-07, + "loss": 0.0601, + "step": 17924 + }, + { + "epoch": 7.5, + "learning_rate": 1.9911317699789822e-07, + "loss": 0.0622, + "step": 17925 + }, + { + "epoch": 7.51, + "learning_rate": 1.9877667703914417e-07, + "loss": 0.0615, + "step": 17926 + }, + { + "epoch": 7.51, + "learning_rate": 1.984404588078559e-07, + "loss": 0.0469, + "step": 17927 + }, + { + "epoch": 7.51, + "learning_rate": 1.9810452231370013e-07, + "loss": 0.034, + "step": 17928 + }, + { + "epoch": 7.51, + "learning_rate": 1.977688675663314e-07, + "loss": 0.0664, + "step": 17929 + }, + { + "epoch": 7.51, + "learning_rate": 1.974334945753975e-07, + "loss": 0.0598, + "step": 17930 + }, + { + "epoch": 7.51, + "learning_rate": 1.9709840335053965e-07, + "loss": 0.0547, + "step": 17931 + }, + { + "epoch": 7.51, + "learning_rate": 1.96763593901389e-07, + "loss": 0.0456, + "step": 17932 + }, + { + "epoch": 7.51, + "learning_rate": 1.9642906623757008e-07, + "loss": 0.0449, + "step": 17933 + }, + { + "epoch": 7.51, + "learning_rate": 1.9609482036869853e-07, + "loss": 0.0373, + "step": 17934 + }, + { + "epoch": 7.51, + "learning_rate": 1.9576085630438114e-07, + "loss": 0.0501, + "step": 17935 + }, + { + "epoch": 7.51, + "learning_rate": 1.9542717405421796e-07, + "loss": 0.0594, + "step": 17936 + }, + { + "epoch": 7.51, + "learning_rate": 1.9509377362779913e-07, + "loss": 0.0462, + "step": 17937 + }, + { + "epoch": 7.51, + "learning_rate": 1.9476065503471032e-07, + "loss": 0.0492, + "step": 17938 + }, + { + "epoch": 7.51, + "learning_rate": 1.9442781828452496e-07, + "loss": 0.0553, + "step": 17939 + }, + { + "epoch": 7.51, + "learning_rate": 1.9409526338681096e-07, + "loss": 0.0564, + "step": 17940 + }, + { + "epoch": 7.51, + "learning_rate": 1.9376299035112623e-07, + "loss": 0.069, + "step": 17941 + }, + { + "epoch": 7.51, + "learning_rate": 1.9343099918702424e-07, + "loss": 0.048, + "step": 17942 + }, + { + "epoch": 7.51, + "learning_rate": 1.9309928990404403e-07, + "loss": 0.0558, + "step": 17943 + }, + { + "epoch": 7.51, + "learning_rate": 1.927678625117224e-07, + "loss": 0.0586, + "step": 17944 + }, + { + "epoch": 7.51, + "learning_rate": 1.924367170195862e-07, + "loss": 0.0553, + "step": 17945 + }, + { + "epoch": 7.51, + "learning_rate": 1.9210585343715337e-07, + "loss": 0.0448, + "step": 17946 + }, + { + "epoch": 7.51, + "learning_rate": 1.9177527177393518e-07, + "loss": 0.0435, + "step": 17947 + }, + { + "epoch": 7.51, + "learning_rate": 1.914449720394329e-07, + "loss": 0.0372, + "step": 17948 + }, + { + "epoch": 7.51, + "learning_rate": 1.9111495424314009e-07, + "loss": 0.0541, + "step": 17949 + }, + { + "epoch": 7.52, + "learning_rate": 1.9078521839454466e-07, + "loss": 0.058, + "step": 17950 + }, + { + "epoch": 7.52, + "learning_rate": 1.9045576450312242e-07, + "loss": 0.039, + "step": 17951 + }, + { + "epoch": 7.52, + "learning_rate": 1.9012659257834464e-07, + "loss": 0.0564, + "step": 17952 + }, + { + "epoch": 7.52, + "learning_rate": 1.8979770262967378e-07, + "loss": 0.0444, + "step": 17953 + }, + { + "epoch": 7.52, + "learning_rate": 1.8946909466656115e-07, + "loss": 0.0515, + "step": 17954 + }, + { + "epoch": 7.52, + "learning_rate": 1.8914076869845587e-07, + "loss": 0.0674, + "step": 17955 + }, + { + "epoch": 7.52, + "learning_rate": 1.8881272473479152e-07, + "loss": 0.0523, + "step": 17956 + }, + { + "epoch": 7.52, + "learning_rate": 1.8848496278499938e-07, + "loss": 0.056, + "step": 17957 + }, + { + "epoch": 7.52, + "learning_rate": 1.8815748285849977e-07, + "loss": 0.0546, + "step": 17958 + }, + { + "epoch": 7.52, + "learning_rate": 1.8783028496470735e-07, + "loss": 0.0405, + "step": 17959 + }, + { + "epoch": 7.52, + "learning_rate": 1.875033691130257e-07, + "loss": 0.0504, + "step": 17960 + }, + { + "epoch": 7.52, + "learning_rate": 1.8717673531285398e-07, + "loss": 0.0564, + "step": 17961 + }, + { + "epoch": 7.52, + "learning_rate": 1.86850383573578e-07, + "loss": 0.0531, + "step": 17962 + }, + { + "epoch": 7.52, + "learning_rate": 1.8652431390457916e-07, + "loss": 0.0326, + "step": 17963 + }, + { + "epoch": 7.52, + "learning_rate": 1.8619852631523105e-07, + "loss": 0.0476, + "step": 17964 + }, + { + "epoch": 7.52, + "learning_rate": 1.8587302081489734e-07, + "loss": 0.0564, + "step": 17965 + }, + { + "epoch": 7.52, + "learning_rate": 1.8554779741293493e-07, + "loss": 0.0391, + "step": 17966 + }, + { + "epoch": 7.52, + "learning_rate": 1.8522285611869194e-07, + "loss": 0.0564, + "step": 17967 + }, + { + "epoch": 7.52, + "learning_rate": 1.848981969415087e-07, + "loss": 0.0552, + "step": 17968 + }, + { + "epoch": 7.52, + "learning_rate": 1.8457381989071767e-07, + "loss": 0.0359, + "step": 17969 + }, + { + "epoch": 7.52, + "learning_rate": 1.8424972497564143e-07, + "loss": 0.0379, + "step": 17970 + }, + { + "epoch": 7.52, + "learning_rate": 1.8392591220559586e-07, + "loss": 0.0391, + "step": 17971 + }, + { + "epoch": 7.52, + "learning_rate": 1.836023815898902e-07, + "loss": 0.0349, + "step": 17972 + }, + { + "epoch": 7.52, + "learning_rate": 1.8327913313782252e-07, + "loss": 0.0289, + "step": 17973 + }, + { + "epoch": 7.53, + "learning_rate": 1.8295616685868434e-07, + "loss": 0.0499, + "step": 17974 + }, + { + "epoch": 7.53, + "learning_rate": 1.8263348276176152e-07, + "loss": 0.063, + "step": 17975 + }, + { + "epoch": 7.53, + "learning_rate": 1.8231108085632664e-07, + "loss": 0.0428, + "step": 17976 + }, + { + "epoch": 7.53, + "learning_rate": 1.8198896115164676e-07, + "loss": 0.0729, + "step": 17977 + }, + { + "epoch": 7.53, + "learning_rate": 1.8166712365698335e-07, + "loss": 0.0325, + "step": 17978 + }, + { + "epoch": 7.53, + "learning_rate": 1.8134556838158457e-07, + "loss": 0.038, + "step": 17979 + }, + { + "epoch": 7.53, + "learning_rate": 1.8102429533469523e-07, + "loss": 0.0474, + "step": 17980 + }, + { + "epoch": 7.53, + "learning_rate": 1.8070330452554907e-07, + "loss": 0.0502, + "step": 17981 + }, + { + "epoch": 7.53, + "learning_rate": 1.8038259596337316e-07, + "loss": 0.0523, + "step": 17982 + }, + { + "epoch": 7.53, + "learning_rate": 1.8006216965738676e-07, + "loss": 0.0457, + "step": 17983 + }, + { + "epoch": 7.53, + "learning_rate": 1.797420256167981e-07, + "loss": 0.0496, + "step": 17984 + }, + { + "epoch": 7.53, + "learning_rate": 1.794221638508098e-07, + "loss": 0.051, + "step": 17985 + }, + { + "epoch": 7.53, + "learning_rate": 1.7910258436861782e-07, + "loss": 0.0541, + "step": 17986 + }, + { + "epoch": 7.53, + "learning_rate": 1.7878328717940597e-07, + "loss": 0.0444, + "step": 17987 + }, + { + "epoch": 7.53, + "learning_rate": 1.7846427229235462e-07, + "loss": 0.0563, + "step": 17988 + }, + { + "epoch": 7.53, + "learning_rate": 1.7814553971663207e-07, + "loss": 0.0434, + "step": 17989 + }, + { + "epoch": 7.53, + "learning_rate": 1.778270894613998e-07, + "loss": 0.0503, + "step": 17990 + }, + { + "epoch": 7.53, + "learning_rate": 1.7750892153581278e-07, + "loss": 0.0337, + "step": 17991 + }, + { + "epoch": 7.53, + "learning_rate": 1.7719103594901366e-07, + "loss": 0.0494, + "step": 17992 + }, + { + "epoch": 7.53, + "learning_rate": 1.768734327101429e-07, + "loss": 0.0563, + "step": 17993 + }, + { + "epoch": 7.53, + "learning_rate": 1.7655611182832767e-07, + "loss": 0.0568, + "step": 17994 + }, + { + "epoch": 7.53, + "learning_rate": 1.7623907331269174e-07, + "loss": 0.0511, + "step": 17995 + }, + { + "epoch": 7.53, + "learning_rate": 1.7592231717234454e-07, + "loss": 0.0627, + "step": 17996 + }, + { + "epoch": 7.53, + "learning_rate": 1.7560584341639208e-07, + "loss": 0.0462, + "step": 17997 + }, + { + "epoch": 7.54, + "learning_rate": 1.7528965205393266e-07, + "loss": 0.0687, + "step": 17998 + }, + { + "epoch": 7.54, + "learning_rate": 1.7497374309405346e-07, + "loss": 0.0531, + "step": 17999 + }, + { + "epoch": 7.54, + "learning_rate": 1.7465811654583608e-07, + "loss": 0.0545, + "step": 18000 + } + ], + "max_steps": 19104, + "num_train_epochs": 8, + "total_flos": 3.2964440252334735e+18, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-18000/training_args.bin b/checkpoint-18000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..8b1b50208b5713a1848bb03b9cf39bd33c748f72 --- /dev/null +++ b/checkpoint-18000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:457d7e76c0fce8d03d8cfe3fdedf1b89cbf25e0f3a605b6976f387e928f69251 +size 3899 diff --git a/config.json b/config.json new file mode 100644 index 0000000000000000000000000000000000000000..066a0107385020bf48a41a53e63dd916e2f96670 --- /dev/null +++ b/config.json @@ -0,0 +1,30 @@ +{ + "_name_or_path": "/scratch/wzhu59/LLaVA-Med/weights/llava-med-7b-pretrain", + "architectures": [ + "LlavaLlamaForCausalLM" + ], + "bos_token_id": 0, + "eos_token_id": 1, + "freeze_mm_mlp_adapter": false, + "hidden_act": "silu", + "hidden_size": 4096, + "initializer_range": 0.02, + "intermediate_size": 11008, + "max_sequence_length": 2048, + "mm_hidden_size": 1024, + "mm_use_im_start_end": true, + "mm_vision_select_layer": -2, + "mm_vision_tower": "openai/clip-vit-large-patch14", + "model_type": "llava", + "num_attention_heads": 32, + "num_hidden_layers": 32, + "pad_token_id": -1, + "rms_norm_eps": 1e-06, + "tie_word_embeddings": false, + "torch_dtype": "float32", + "transformers_version": "4.28.0.dev0", + "tune_mm_mlp_adapter": false, + "use_cache": false, + "use_mm_proj": true, + "vocab_size": 32004 +} diff --git a/generation_config.json b/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..128b48946713166edcc3e3c62ae5fe285df776be --- /dev/null +++ b/generation_config.json @@ -0,0 +1,7 @@ +{ + "_from_model_config": true, + "bos_token_id": 0, + "eos_token_id": 1, + "pad_token_id": 0, + "transformers_version": "4.28.0.dev0" +} diff --git a/pytorch_model-00001-of-00003.bin b/pytorch_model-00001-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..70bfcda89803e223011f4c279280bdc66862e738 --- /dev/null +++ b/pytorch_model-00001-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f019af333c131952cc474d40e1dbd2db2d4afeaea03b1e3b70242c57c73787cb +size 9878055122 diff --git a/pytorch_model-00002-of-00003.bin b/pytorch_model-00002-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..1a37a50c7af295b6fbe229d793f18562e12afa25 --- /dev/null +++ b/pytorch_model-00002-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b144830160a142e4a2d47e494043f863899c9940c984c7607074bad525e6aaea +size 9894801014 diff --git a/pytorch_model-00003-of-00003.bin b/pytorch_model-00003-of-00003.bin new file mode 100644 index 0000000000000000000000000000000000000000..43617e3823108c72c6d643a52805608eb3bedcac --- /dev/null +++ b/pytorch_model-00003-of-00003.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97dbae40d85779dcce1916d0905fad5b498f4cd07b18e62fdcfba07607361aa3 +size 7197850485 diff --git a/pytorch_model.bin.index.json b/pytorch_model.bin.index.json new file mode 100644 index 0000000000000000000000000000000000000000..50d9a764a2282762889d821cfaa49b284963c070 --- /dev/null +++ b/pytorch_model.bin.index.json @@ -0,0 +1,332 @@ +{ + "metadata": { + "total_size": 26970595328 + }, + "weight_map": { + "lm_head.weight": "pytorch_model-00003-of-00003.bin", + "model.embed_tokens.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.12.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.20.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.input_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.mlp.up_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00003.bin", + "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00003.bin", + "model.layers.24.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.30.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.input_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.down_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.mlp.up_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00003.bin", + "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00003-of-00003.bin", + "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00003.bin", + "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00003.bin", + "model.mm_projector.bias": "pytorch_model-00003-of-00003.bin", + "model.mm_projector.weight": "pytorch_model-00003-of-00003.bin", + "model.norm.weight": "pytorch_model-00003-of-00003.bin" + } +} diff --git a/special_tokens_map.json b/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..5ce7dbc9759ecf5596c5122abbdf0b431355efa4 --- /dev/null +++ b/special_tokens_map.json @@ -0,0 +1,6 @@ +{ + "bos_token": "", + "eos_token": "", + "pad_token": "[PAD]", + "unk_token": "" +} diff --git a/tokenizer.model b/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/tokenizer_config.json b/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..22bcdfb351739d00df7b9b12f99757c6e7e0b540 --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,32 @@ +{ + "bos_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + }, + "clean_up_tokenization_spaces": false, + "eos_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + }, + "model_max_length": 2048, + "pad_token": null, + "padding_side": "right", + "special_tokens_map_file": "/nobackup/haotian/.cache/huggingface/hub/models--lmsys--vicuna-7b-delta-v0/snapshots/6d2e645f62e734821be8c392c08c8b2098b35434/special_tokens_map.json", + "tokenizer_class": "LlamaTokenizer", + "unk_token": { + "__type": "AddedToken", + "content": "", + "lstrip": false, + "normalized": true, + "rstrip": false, + "single_word": false + } +} diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..ce59b71da7fcb0151072e84745214989e7e06ffd --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,114649 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 7.998325308771196, + "global_step": 19104, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 3.484320557491289e-08, + "loss": 1.487, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 6.968641114982578e-08, + "loss": 1.5418, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 1.045296167247387e-07, + "loss": 1.2413, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 1.3937282229965157e-07, + "loss": 1.1777, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 1.7421602787456448e-07, + "loss": 0.9729, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 2.090592334494774e-07, + "loss": 0.9887, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 2.439024390243903e-07, + "loss": 0.9151, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 2.7874564459930313e-07, + "loss": 0.9341, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 3.1358885017421604e-07, + "loss": 0.9024, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 3.4843205574912896e-07, + "loss": 0.96, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 3.832752613240418e-07, + "loss": 0.9397, + "step": 11 + }, + { + "epoch": 0.01, + "learning_rate": 4.181184668989548e-07, + "loss": 1.0473, + "step": 12 + }, + { + "epoch": 0.01, + "learning_rate": 4.5296167247386764e-07, + "loss": 0.8521, + "step": 13 + }, + { + "epoch": 0.01, + "learning_rate": 4.878048780487805e-07, + "loss": 0.9829, + "step": 14 + }, + { + "epoch": 0.01, + "learning_rate": 5.226480836236935e-07, + "loss": 1.1014, + "step": 15 + }, + { + "epoch": 0.01, + "learning_rate": 5.574912891986063e-07, + "loss": 0.9186, + "step": 16 + }, + { + "epoch": 0.01, + "learning_rate": 5.923344947735192e-07, + "loss": 0.8867, + "step": 17 + }, + { + "epoch": 0.01, + "learning_rate": 6.271777003484321e-07, + "loss": 0.8071, + "step": 18 + }, + { + "epoch": 0.01, + "learning_rate": 6.62020905923345e-07, + "loss": 0.8481, + "step": 19 + }, + { + "epoch": 0.01, + "learning_rate": 6.968641114982579e-07, + "loss": 0.8675, + "step": 20 + }, + { + "epoch": 0.01, + "learning_rate": 7.317073170731707e-07, + "loss": 0.9635, + "step": 21 + }, + { + "epoch": 0.01, + "learning_rate": 7.665505226480836e-07, + "loss": 0.8868, + "step": 22 + }, + { + "epoch": 0.01, + "learning_rate": 8.013937282229965e-07, + "loss": 0.9029, + "step": 23 + }, + { + "epoch": 0.01, + "learning_rate": 8.362369337979096e-07, + "loss": 0.9118, + "step": 24 + }, + { + "epoch": 0.01, + "learning_rate": 8.710801393728225e-07, + "loss": 0.9768, + "step": 25 + }, + { + "epoch": 0.01, + "learning_rate": 9.059233449477353e-07, + "loss": 0.8292, + "step": 26 + }, + { + "epoch": 0.01, + "learning_rate": 9.407665505226482e-07, + "loss": 0.8865, + "step": 27 + }, + { + "epoch": 0.01, + "learning_rate": 9.75609756097561e-07, + "loss": 0.9918, + "step": 28 + }, + { + "epoch": 0.01, + "learning_rate": 1.010452961672474e-06, + "loss": 0.7146, + "step": 29 + }, + { + "epoch": 0.01, + "learning_rate": 1.045296167247387e-06, + "loss": 0.907, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 1.0801393728222997e-06, + "loss": 0.8015, + "step": 31 + }, + { + "epoch": 0.01, + "learning_rate": 1.1149825783972125e-06, + "loss": 0.7597, + "step": 32 + }, + { + "epoch": 0.01, + "learning_rate": 1.1498257839721255e-06, + "loss": 0.8095, + "step": 33 + }, + { + "epoch": 0.01, + "learning_rate": 1.1846689895470384e-06, + "loss": 0.9036, + "step": 34 + }, + { + "epoch": 0.01, + "learning_rate": 1.2195121951219514e-06, + "loss": 0.8162, + "step": 35 + }, + { + "epoch": 0.02, + "learning_rate": 1.2543554006968642e-06, + "loss": 0.9883, + "step": 36 + }, + { + "epoch": 0.02, + "learning_rate": 1.289198606271777e-06, + "loss": 0.8634, + "step": 37 + }, + { + "epoch": 0.02, + "learning_rate": 1.32404181184669e-06, + "loss": 0.7832, + "step": 38 + }, + { + "epoch": 0.02, + "learning_rate": 1.3588850174216028e-06, + "loss": 0.7378, + "step": 39 + }, + { + "epoch": 0.02, + "learning_rate": 1.3937282229965158e-06, + "loss": 0.8348, + "step": 40 + }, + { + "epoch": 0.02, + "learning_rate": 1.4285714285714286e-06, + "loss": 0.6925, + "step": 41 + }, + { + "epoch": 0.02, + "learning_rate": 1.4634146341463414e-06, + "loss": 0.8969, + "step": 42 + }, + { + "epoch": 0.02, + "learning_rate": 1.4982578397212545e-06, + "loss": 0.7099, + "step": 43 + }, + { + "epoch": 0.02, + "learning_rate": 1.5331010452961673e-06, + "loss": 0.8763, + "step": 44 + }, + { + "epoch": 0.02, + "learning_rate": 1.56794425087108e-06, + "loss": 0.7967, + "step": 45 + }, + { + "epoch": 0.02, + "learning_rate": 1.602787456445993e-06, + "loss": 0.7528, + "step": 46 + }, + { + "epoch": 0.02, + "learning_rate": 1.6376306620209059e-06, + "loss": 0.8367, + "step": 47 + }, + { + "epoch": 0.02, + "learning_rate": 1.6724738675958191e-06, + "loss": 0.8172, + "step": 48 + }, + { + "epoch": 0.02, + "learning_rate": 1.707317073170732e-06, + "loss": 0.7745, + "step": 49 + }, + { + "epoch": 0.02, + "learning_rate": 1.742160278745645e-06, + "loss": 0.8643, + "step": 50 + }, + { + "epoch": 0.02, + "learning_rate": 1.7770034843205577e-06, + "loss": 0.7374, + "step": 51 + }, + { + "epoch": 0.02, + "learning_rate": 1.8118466898954705e-06, + "loss": 0.8095, + "step": 52 + }, + { + "epoch": 0.02, + "learning_rate": 1.8466898954703836e-06, + "loss": 0.7755, + "step": 53 + }, + { + "epoch": 0.02, + "learning_rate": 1.8815331010452964e-06, + "loss": 0.8084, + "step": 54 + }, + { + "epoch": 0.02, + "learning_rate": 1.916376306620209e-06, + "loss": 0.7396, + "step": 55 + }, + { + "epoch": 0.02, + "learning_rate": 1.951219512195122e-06, + "loss": 0.7355, + "step": 56 + }, + { + "epoch": 0.02, + "learning_rate": 1.986062717770035e-06, + "loss": 0.7007, + "step": 57 + }, + { + "epoch": 0.02, + "learning_rate": 2.020905923344948e-06, + "loss": 0.8282, + "step": 58 + }, + { + "epoch": 0.02, + "learning_rate": 2.055749128919861e-06, + "loss": 0.7561, + "step": 59 + }, + { + "epoch": 0.03, + "learning_rate": 2.090592334494774e-06, + "loss": 0.8712, + "step": 60 + }, + { + "epoch": 0.03, + "learning_rate": 2.1254355400696864e-06, + "loss": 0.8559, + "step": 61 + }, + { + "epoch": 0.03, + "learning_rate": 2.1602787456445995e-06, + "loss": 0.7767, + "step": 62 + }, + { + "epoch": 0.03, + "learning_rate": 2.1951219512195125e-06, + "loss": 0.81, + "step": 63 + }, + { + "epoch": 0.03, + "learning_rate": 2.229965156794425e-06, + "loss": 0.805, + "step": 64 + }, + { + "epoch": 0.03, + "learning_rate": 2.264808362369338e-06, + "loss": 0.6648, + "step": 65 + }, + { + "epoch": 0.03, + "learning_rate": 2.299651567944251e-06, + "loss": 0.8027, + "step": 66 + }, + { + "epoch": 0.03, + "learning_rate": 2.334494773519164e-06, + "loss": 0.7623, + "step": 67 + }, + { + "epoch": 0.03, + "learning_rate": 2.3693379790940767e-06, + "loss": 0.7903, + "step": 68 + }, + { + "epoch": 0.03, + "learning_rate": 2.4041811846689897e-06, + "loss": 0.6917, + "step": 69 + }, + { + "epoch": 0.03, + "learning_rate": 2.4390243902439027e-06, + "loss": 0.8177, + "step": 70 + }, + { + "epoch": 0.03, + "learning_rate": 2.4738675958188153e-06, + "loss": 0.7348, + "step": 71 + }, + { + "epoch": 0.03, + "learning_rate": 2.5087108013937284e-06, + "loss": 0.6916, + "step": 72 + }, + { + "epoch": 0.03, + "learning_rate": 2.5435540069686414e-06, + "loss": 0.6663, + "step": 73 + }, + { + "epoch": 0.03, + "learning_rate": 2.578397212543554e-06, + "loss": 0.6881, + "step": 74 + }, + { + "epoch": 0.03, + "learning_rate": 2.6132404181184674e-06, + "loss": 0.6372, + "step": 75 + }, + { + "epoch": 0.03, + "learning_rate": 2.64808362369338e-06, + "loss": 0.7268, + "step": 76 + }, + { + "epoch": 0.03, + "learning_rate": 2.682926829268293e-06, + "loss": 0.7138, + "step": 77 + }, + { + "epoch": 0.03, + "learning_rate": 2.7177700348432056e-06, + "loss": 0.7341, + "step": 78 + }, + { + "epoch": 0.03, + "learning_rate": 2.7526132404181186e-06, + "loss": 0.7041, + "step": 79 + }, + { + "epoch": 0.03, + "learning_rate": 2.7874564459930316e-06, + "loss": 0.7239, + "step": 80 + }, + { + "epoch": 0.03, + "learning_rate": 2.8222996515679447e-06, + "loss": 0.6809, + "step": 81 + }, + { + "epoch": 0.03, + "learning_rate": 2.8571428571428573e-06, + "loss": 0.7256, + "step": 82 + }, + { + "epoch": 0.03, + "learning_rate": 2.8919860627177703e-06, + "loss": 0.7205, + "step": 83 + }, + { + "epoch": 0.04, + "learning_rate": 2.926829268292683e-06, + "loss": 0.6596, + "step": 84 + }, + { + "epoch": 0.04, + "learning_rate": 2.9616724738675963e-06, + "loss": 0.657, + "step": 85 + }, + { + "epoch": 0.04, + "learning_rate": 2.996515679442509e-06, + "loss": 0.6502, + "step": 86 + }, + { + "epoch": 0.04, + "learning_rate": 3.031358885017422e-06, + "loss": 0.7095, + "step": 87 + }, + { + "epoch": 0.04, + "learning_rate": 3.0662020905923345e-06, + "loss": 0.6643, + "step": 88 + }, + { + "epoch": 0.04, + "learning_rate": 3.1010452961672475e-06, + "loss": 0.6621, + "step": 89 + }, + { + "epoch": 0.04, + "learning_rate": 3.13588850174216e-06, + "loss": 0.7324, + "step": 90 + }, + { + "epoch": 0.04, + "learning_rate": 3.1707317073170736e-06, + "loss": 0.73, + "step": 91 + }, + { + "epoch": 0.04, + "learning_rate": 3.205574912891986e-06, + "loss": 0.7109, + "step": 92 + }, + { + "epoch": 0.04, + "learning_rate": 3.240418118466899e-06, + "loss": 0.695, + "step": 93 + }, + { + "epoch": 0.04, + "learning_rate": 3.2752613240418118e-06, + "loss": 0.5556, + "step": 94 + }, + { + "epoch": 0.04, + "learning_rate": 3.310104529616725e-06, + "loss": 0.6237, + "step": 95 + }, + { + "epoch": 0.04, + "learning_rate": 3.3449477351916382e-06, + "loss": 0.6923, + "step": 96 + }, + { + "epoch": 0.04, + "learning_rate": 3.379790940766551e-06, + "loss": 0.7276, + "step": 97 + }, + { + "epoch": 0.04, + "learning_rate": 3.414634146341464e-06, + "loss": 0.6814, + "step": 98 + }, + { + "epoch": 0.04, + "learning_rate": 3.4494773519163764e-06, + "loss": 0.7419, + "step": 99 + }, + { + "epoch": 0.04, + "learning_rate": 3.48432055749129e-06, + "loss": 0.6808, + "step": 100 + }, + { + "epoch": 0.04, + "learning_rate": 3.5191637630662025e-06, + "loss": 0.6811, + "step": 101 + }, + { + "epoch": 0.04, + "learning_rate": 3.5540069686411155e-06, + "loss": 0.6577, + "step": 102 + }, + { + "epoch": 0.04, + "learning_rate": 3.588850174216028e-06, + "loss": 0.6782, + "step": 103 + }, + { + "epoch": 0.04, + "learning_rate": 3.623693379790941e-06, + "loss": 0.7048, + "step": 104 + }, + { + "epoch": 0.04, + "learning_rate": 3.6585365853658537e-06, + "loss": 0.6673, + "step": 105 + }, + { + "epoch": 0.04, + "learning_rate": 3.693379790940767e-06, + "loss": 0.6239, + "step": 106 + }, + { + "epoch": 0.04, + "learning_rate": 3.7282229965156797e-06, + "loss": 0.6954, + "step": 107 + }, + { + "epoch": 0.05, + "learning_rate": 3.7630662020905927e-06, + "loss": 0.6656, + "step": 108 + }, + { + "epoch": 0.05, + "learning_rate": 3.7979094076655053e-06, + "loss": 0.6732, + "step": 109 + }, + { + "epoch": 0.05, + "learning_rate": 3.832752613240418e-06, + "loss": 0.7624, + "step": 110 + }, + { + "epoch": 0.05, + "learning_rate": 3.867595818815331e-06, + "loss": 0.7086, + "step": 111 + }, + { + "epoch": 0.05, + "learning_rate": 3.902439024390244e-06, + "loss": 0.6284, + "step": 112 + }, + { + "epoch": 0.05, + "learning_rate": 3.937282229965157e-06, + "loss": 0.6643, + "step": 113 + }, + { + "epoch": 0.05, + "learning_rate": 3.97212543554007e-06, + "loss": 0.6443, + "step": 114 + }, + { + "epoch": 0.05, + "learning_rate": 4.006968641114983e-06, + "loss": 0.6025, + "step": 115 + }, + { + "epoch": 0.05, + "learning_rate": 4.041811846689896e-06, + "loss": 0.6651, + "step": 116 + }, + { + "epoch": 0.05, + "learning_rate": 4.076655052264808e-06, + "loss": 0.6758, + "step": 117 + }, + { + "epoch": 0.05, + "learning_rate": 4.111498257839722e-06, + "loss": 0.6962, + "step": 118 + }, + { + "epoch": 0.05, + "learning_rate": 4.146341463414634e-06, + "loss": 0.6613, + "step": 119 + }, + { + "epoch": 0.05, + "learning_rate": 4.181184668989548e-06, + "loss": 0.6735, + "step": 120 + }, + { + "epoch": 0.05, + "learning_rate": 4.21602787456446e-06, + "loss": 0.5424, + "step": 121 + }, + { + "epoch": 0.05, + "learning_rate": 4.250871080139373e-06, + "loss": 0.7245, + "step": 122 + }, + { + "epoch": 0.05, + "learning_rate": 4.2857142857142855e-06, + "loss": 0.5873, + "step": 123 + }, + { + "epoch": 0.05, + "learning_rate": 4.320557491289199e-06, + "loss": 0.7426, + "step": 124 + }, + { + "epoch": 0.05, + "learning_rate": 4.3554006968641115e-06, + "loss": 0.7477, + "step": 125 + }, + { + "epoch": 0.05, + "learning_rate": 4.390243902439025e-06, + "loss": 0.6925, + "step": 126 + }, + { + "epoch": 0.05, + "learning_rate": 4.4250871080139375e-06, + "loss": 0.7745, + "step": 127 + }, + { + "epoch": 0.05, + "learning_rate": 4.45993031358885e-06, + "loss": 0.6976, + "step": 128 + }, + { + "epoch": 0.05, + "learning_rate": 4.4947735191637636e-06, + "loss": 0.6296, + "step": 129 + }, + { + "epoch": 0.05, + "learning_rate": 4.529616724738676e-06, + "loss": 0.6011, + "step": 130 + }, + { + "epoch": 0.05, + "learning_rate": 4.56445993031359e-06, + "loss": 0.6661, + "step": 131 + }, + { + "epoch": 0.06, + "learning_rate": 4.599303135888502e-06, + "loss": 0.6491, + "step": 132 + }, + { + "epoch": 0.06, + "learning_rate": 4.634146341463416e-06, + "loss": 0.6139, + "step": 133 + }, + { + "epoch": 0.06, + "learning_rate": 4.668989547038328e-06, + "loss": 0.6212, + "step": 134 + }, + { + "epoch": 0.06, + "learning_rate": 4.703832752613241e-06, + "loss": 0.7824, + "step": 135 + }, + { + "epoch": 0.06, + "learning_rate": 4.738675958188153e-06, + "loss": 0.6765, + "step": 136 + }, + { + "epoch": 0.06, + "learning_rate": 4.773519163763067e-06, + "loss": 0.6975, + "step": 137 + }, + { + "epoch": 0.06, + "learning_rate": 4.8083623693379794e-06, + "loss": 0.7251, + "step": 138 + }, + { + "epoch": 0.06, + "learning_rate": 4.843205574912893e-06, + "loss": 0.7232, + "step": 139 + }, + { + "epoch": 0.06, + "learning_rate": 4.8780487804878055e-06, + "loss": 0.6243, + "step": 140 + }, + { + "epoch": 0.06, + "learning_rate": 4.912891986062718e-06, + "loss": 0.6315, + "step": 141 + }, + { + "epoch": 0.06, + "learning_rate": 4.947735191637631e-06, + "loss": 0.632, + "step": 142 + }, + { + "epoch": 0.06, + "learning_rate": 4.982578397212544e-06, + "loss": 0.7962, + "step": 143 + }, + { + "epoch": 0.06, + "learning_rate": 5.017421602787457e-06, + "loss": 0.5554, + "step": 144 + }, + { + "epoch": 0.06, + "learning_rate": 5.052264808362369e-06, + "loss": 0.6004, + "step": 145 + }, + { + "epoch": 0.06, + "learning_rate": 5.087108013937283e-06, + "loss": 0.5713, + "step": 146 + }, + { + "epoch": 0.06, + "learning_rate": 5.121951219512195e-06, + "loss": 0.6607, + "step": 147 + }, + { + "epoch": 0.06, + "learning_rate": 5.156794425087108e-06, + "loss": 0.627, + "step": 148 + }, + { + "epoch": 0.06, + "learning_rate": 5.1916376306620205e-06, + "loss": 0.6944, + "step": 149 + }, + { + "epoch": 0.06, + "learning_rate": 5.226480836236935e-06, + "loss": 0.6617, + "step": 150 + }, + { + "epoch": 0.06, + "learning_rate": 5.261324041811847e-06, + "loss": 0.6516, + "step": 151 + }, + { + "epoch": 0.06, + "learning_rate": 5.29616724738676e-06, + "loss": 0.6271, + "step": 152 + }, + { + "epoch": 0.06, + "learning_rate": 5.331010452961673e-06, + "loss": 0.5682, + "step": 153 + }, + { + "epoch": 0.06, + "learning_rate": 5.365853658536586e-06, + "loss": 0.6439, + "step": 154 + }, + { + "epoch": 0.06, + "learning_rate": 5.400696864111499e-06, + "loss": 0.7387, + "step": 155 + }, + { + "epoch": 0.07, + "learning_rate": 5.435540069686411e-06, + "loss": 0.5682, + "step": 156 + }, + { + "epoch": 0.07, + "learning_rate": 5.470383275261324e-06, + "loss": 0.5933, + "step": 157 + }, + { + "epoch": 0.07, + "learning_rate": 5.505226480836237e-06, + "loss": 0.6817, + "step": 158 + }, + { + "epoch": 0.07, + "learning_rate": 5.540069686411151e-06, + "loss": 0.6364, + "step": 159 + }, + { + "epoch": 0.07, + "learning_rate": 5.574912891986063e-06, + "loss": 0.5878, + "step": 160 + }, + { + "epoch": 0.07, + "learning_rate": 5.609756097560977e-06, + "loss": 0.6615, + "step": 161 + }, + { + "epoch": 0.07, + "learning_rate": 5.644599303135889e-06, + "loss": 0.6586, + "step": 162 + }, + { + "epoch": 0.07, + "learning_rate": 5.679442508710802e-06, + "loss": 0.6446, + "step": 163 + }, + { + "epoch": 0.07, + "learning_rate": 5.7142857142857145e-06, + "loss": 0.733, + "step": 164 + }, + { + "epoch": 0.07, + "learning_rate": 5.749128919860628e-06, + "loss": 0.6642, + "step": 165 + }, + { + "epoch": 0.07, + "learning_rate": 5.7839721254355405e-06, + "loss": 0.6861, + "step": 166 + }, + { + "epoch": 0.07, + "learning_rate": 5.818815331010453e-06, + "loss": 0.623, + "step": 167 + }, + { + "epoch": 0.07, + "learning_rate": 5.853658536585366e-06, + "loss": 0.5951, + "step": 168 + }, + { + "epoch": 0.07, + "learning_rate": 5.88850174216028e-06, + "loss": 0.6887, + "step": 169 + }, + { + "epoch": 0.07, + "learning_rate": 5.923344947735193e-06, + "loss": 0.5783, + "step": 170 + }, + { + "epoch": 0.07, + "learning_rate": 5.958188153310105e-06, + "loss": 0.6036, + "step": 171 + }, + { + "epoch": 0.07, + "learning_rate": 5.993031358885018e-06, + "loss": 0.6031, + "step": 172 + }, + { + "epoch": 0.07, + "learning_rate": 6.027874564459931e-06, + "loss": 0.659, + "step": 173 + }, + { + "epoch": 0.07, + "learning_rate": 6.062717770034844e-06, + "loss": 0.6425, + "step": 174 + }, + { + "epoch": 0.07, + "learning_rate": 6.0975609756097564e-06, + "loss": 0.5897, + "step": 175 + }, + { + "epoch": 0.07, + "learning_rate": 6.132404181184669e-06, + "loss": 0.6508, + "step": 176 + }, + { + "epoch": 0.07, + "learning_rate": 6.1672473867595825e-06, + "loss": 0.6807, + "step": 177 + }, + { + "epoch": 0.07, + "learning_rate": 6.202090592334495e-06, + "loss": 0.5439, + "step": 178 + }, + { + "epoch": 0.07, + "learning_rate": 6.236933797909408e-06, + "loss": 0.5265, + "step": 179 + }, + { + "epoch": 0.08, + "learning_rate": 6.27177700348432e-06, + "loss": 0.6912, + "step": 180 + }, + { + "epoch": 0.08, + "learning_rate": 6.3066202090592345e-06, + "loss": 0.6348, + "step": 181 + }, + { + "epoch": 0.08, + "learning_rate": 6.341463414634147e-06, + "loss": 0.6374, + "step": 182 + }, + { + "epoch": 0.08, + "learning_rate": 6.37630662020906e-06, + "loss": 0.626, + "step": 183 + }, + { + "epoch": 0.08, + "learning_rate": 6.411149825783972e-06, + "loss": 0.6239, + "step": 184 + }, + { + "epoch": 0.08, + "learning_rate": 6.445993031358886e-06, + "loss": 0.6928, + "step": 185 + }, + { + "epoch": 0.08, + "learning_rate": 6.480836236933798e-06, + "loss": 0.6335, + "step": 186 + }, + { + "epoch": 0.08, + "learning_rate": 6.515679442508711e-06, + "loss": 0.5033, + "step": 187 + }, + { + "epoch": 0.08, + "learning_rate": 6.5505226480836235e-06, + "loss": 0.6619, + "step": 188 + }, + { + "epoch": 0.08, + "learning_rate": 6.585365853658538e-06, + "loss": 0.6189, + "step": 189 + }, + { + "epoch": 0.08, + "learning_rate": 6.62020905923345e-06, + "loss": 0.6277, + "step": 190 + }, + { + "epoch": 0.08, + "learning_rate": 6.655052264808363e-06, + "loss": 0.6073, + "step": 191 + }, + { + "epoch": 0.08, + "learning_rate": 6.6898954703832765e-06, + "loss": 0.6097, + "step": 192 + }, + { + "epoch": 0.08, + "learning_rate": 6.724738675958189e-06, + "loss": 0.5943, + "step": 193 + }, + { + "epoch": 0.08, + "learning_rate": 6.759581881533102e-06, + "loss": 0.5979, + "step": 194 + }, + { + "epoch": 0.08, + "learning_rate": 6.794425087108014e-06, + "loss": 0.598, + "step": 195 + }, + { + "epoch": 0.08, + "learning_rate": 6.829268292682928e-06, + "loss": 0.6136, + "step": 196 + }, + { + "epoch": 0.08, + "learning_rate": 6.86411149825784e-06, + "loss": 0.6153, + "step": 197 + }, + { + "epoch": 0.08, + "learning_rate": 6.898954703832753e-06, + "loss": 0.6817, + "step": 198 + }, + { + "epoch": 0.08, + "learning_rate": 6.9337979094076655e-06, + "loss": 0.5501, + "step": 199 + }, + { + "epoch": 0.08, + "learning_rate": 6.96864111498258e-06, + "loss": 0.7031, + "step": 200 + }, + { + "epoch": 0.08, + "learning_rate": 7.003484320557492e-06, + "loss": 0.5989, + "step": 201 + }, + { + "epoch": 0.08, + "learning_rate": 7.038327526132405e-06, + "loss": 0.6308, + "step": 202 + }, + { + "epoch": 0.08, + "learning_rate": 7.0731707317073175e-06, + "loss": 0.5342, + "step": 203 + }, + { + "epoch": 0.09, + "learning_rate": 7.108013937282231e-06, + "loss": 0.6175, + "step": 204 + }, + { + "epoch": 0.09, + "learning_rate": 7.1428571428571436e-06, + "loss": 0.5792, + "step": 205 + }, + { + "epoch": 0.09, + "learning_rate": 7.177700348432056e-06, + "loss": 0.6591, + "step": 206 + }, + { + "epoch": 0.09, + "learning_rate": 7.212543554006969e-06, + "loss": 0.6793, + "step": 207 + }, + { + "epoch": 0.09, + "learning_rate": 7.247386759581882e-06, + "loss": 0.6891, + "step": 208 + }, + { + "epoch": 0.09, + "learning_rate": 7.282229965156795e-06, + "loss": 0.5279, + "step": 209 + }, + { + "epoch": 0.09, + "learning_rate": 7.317073170731707e-06, + "loss": 0.6229, + "step": 210 + }, + { + "epoch": 0.09, + "learning_rate": 7.35191637630662e-06, + "loss": 0.6168, + "step": 211 + }, + { + "epoch": 0.09, + "learning_rate": 7.386759581881534e-06, + "loss": 0.4843, + "step": 212 + }, + { + "epoch": 0.09, + "learning_rate": 7.421602787456447e-06, + "loss": 0.5995, + "step": 213 + }, + { + "epoch": 0.09, + "learning_rate": 7.4564459930313594e-06, + "loss": 0.5968, + "step": 214 + }, + { + "epoch": 0.09, + "learning_rate": 7.491289198606272e-06, + "loss": 0.5597, + "step": 215 + }, + { + "epoch": 0.09, + "learning_rate": 7.5261324041811855e-06, + "loss": 0.6491, + "step": 216 + }, + { + "epoch": 0.09, + "learning_rate": 7.560975609756098e-06, + "loss": 0.643, + "step": 217 + }, + { + "epoch": 0.09, + "learning_rate": 7.595818815331011e-06, + "loss": 0.6927, + "step": 218 + }, + { + "epoch": 0.09, + "learning_rate": 7.630662020905924e-06, + "loss": 0.5916, + "step": 219 + }, + { + "epoch": 0.09, + "learning_rate": 7.665505226480837e-06, + "loss": 0.5445, + "step": 220 + }, + { + "epoch": 0.09, + "learning_rate": 7.70034843205575e-06, + "loss": 0.6517, + "step": 221 + }, + { + "epoch": 0.09, + "learning_rate": 7.735191637630662e-06, + "loss": 0.6, + "step": 222 + }, + { + "epoch": 0.09, + "learning_rate": 7.770034843205574e-06, + "loss": 0.6441, + "step": 223 + }, + { + "epoch": 0.09, + "learning_rate": 7.804878048780489e-06, + "loss": 0.6266, + "step": 224 + }, + { + "epoch": 0.09, + "learning_rate": 7.839721254355401e-06, + "loss": 0.5312, + "step": 225 + }, + { + "epoch": 0.09, + "learning_rate": 7.874564459930314e-06, + "loss": 0.614, + "step": 226 + }, + { + "epoch": 0.1, + "learning_rate": 7.909407665505228e-06, + "loss": 0.6742, + "step": 227 + }, + { + "epoch": 0.1, + "learning_rate": 7.94425087108014e-06, + "loss": 0.588, + "step": 228 + }, + { + "epoch": 0.1, + "learning_rate": 7.979094076655053e-06, + "loss": 0.5782, + "step": 229 + }, + { + "epoch": 0.1, + "learning_rate": 8.013937282229966e-06, + "loss": 0.6051, + "step": 230 + }, + { + "epoch": 0.1, + "learning_rate": 8.048780487804879e-06, + "loss": 0.671, + "step": 231 + }, + { + "epoch": 0.1, + "learning_rate": 8.083623693379791e-06, + "loss": 0.6003, + "step": 232 + }, + { + "epoch": 0.1, + "learning_rate": 8.118466898954704e-06, + "loss": 0.6023, + "step": 233 + }, + { + "epoch": 0.1, + "learning_rate": 8.153310104529616e-06, + "loss": 0.6305, + "step": 234 + }, + { + "epoch": 0.1, + "learning_rate": 8.18815331010453e-06, + "loss": 0.5449, + "step": 235 + }, + { + "epoch": 0.1, + "learning_rate": 8.222996515679443e-06, + "loss": 0.5957, + "step": 236 + }, + { + "epoch": 0.1, + "learning_rate": 8.257839721254356e-06, + "loss": 0.6335, + "step": 237 + }, + { + "epoch": 0.1, + "learning_rate": 8.292682926829268e-06, + "loss": 0.6214, + "step": 238 + }, + { + "epoch": 0.1, + "learning_rate": 8.327526132404183e-06, + "loss": 0.6096, + "step": 239 + }, + { + "epoch": 0.1, + "learning_rate": 8.362369337979095e-06, + "loss": 0.55, + "step": 240 + }, + { + "epoch": 0.1, + "learning_rate": 8.397212543554008e-06, + "loss": 0.641, + "step": 241 + }, + { + "epoch": 0.1, + "learning_rate": 8.43205574912892e-06, + "loss": 0.6294, + "step": 242 + }, + { + "epoch": 0.1, + "learning_rate": 8.466898954703833e-06, + "loss": 0.6216, + "step": 243 + }, + { + "epoch": 0.1, + "learning_rate": 8.501742160278746e-06, + "loss": 0.5949, + "step": 244 + }, + { + "epoch": 0.1, + "learning_rate": 8.536585365853658e-06, + "loss": 0.568, + "step": 245 + }, + { + "epoch": 0.1, + "learning_rate": 8.571428571428571e-06, + "loss": 0.6762, + "step": 246 + }, + { + "epoch": 0.1, + "learning_rate": 8.606271777003485e-06, + "loss": 0.6068, + "step": 247 + }, + { + "epoch": 0.1, + "learning_rate": 8.641114982578398e-06, + "loss": 0.584, + "step": 248 + }, + { + "epoch": 0.1, + "learning_rate": 8.67595818815331e-06, + "loss": 0.5967, + "step": 249 + }, + { + "epoch": 0.1, + "learning_rate": 8.710801393728223e-06, + "loss": 0.5988, + "step": 250 + }, + { + "epoch": 0.11, + "learning_rate": 8.745644599303137e-06, + "loss": 0.611, + "step": 251 + }, + { + "epoch": 0.11, + "learning_rate": 8.78048780487805e-06, + "loss": 0.6672, + "step": 252 + }, + { + "epoch": 0.11, + "learning_rate": 8.815331010452962e-06, + "loss": 0.5487, + "step": 253 + }, + { + "epoch": 0.11, + "learning_rate": 8.850174216027875e-06, + "loss": 0.5397, + "step": 254 + }, + { + "epoch": 0.11, + "learning_rate": 8.885017421602788e-06, + "loss": 0.7101, + "step": 255 + }, + { + "epoch": 0.11, + "learning_rate": 8.9198606271777e-06, + "loss": 0.6165, + "step": 256 + }, + { + "epoch": 0.11, + "learning_rate": 8.954703832752613e-06, + "loss": 0.5816, + "step": 257 + }, + { + "epoch": 0.11, + "learning_rate": 8.989547038327527e-06, + "loss": 0.6, + "step": 258 + }, + { + "epoch": 0.11, + "learning_rate": 9.02439024390244e-06, + "loss": 0.5065, + "step": 259 + }, + { + "epoch": 0.11, + "learning_rate": 9.059233449477352e-06, + "loss": 0.7077, + "step": 260 + }, + { + "epoch": 0.11, + "learning_rate": 9.094076655052265e-06, + "loss": 0.7149, + "step": 261 + }, + { + "epoch": 0.11, + "learning_rate": 9.12891986062718e-06, + "loss": 0.5832, + "step": 262 + }, + { + "epoch": 0.11, + "learning_rate": 9.163763066202092e-06, + "loss": 0.5587, + "step": 263 + }, + { + "epoch": 0.11, + "learning_rate": 9.198606271777004e-06, + "loss": 0.5781, + "step": 264 + }, + { + "epoch": 0.11, + "learning_rate": 9.233449477351917e-06, + "loss": 0.5852, + "step": 265 + }, + { + "epoch": 0.11, + "learning_rate": 9.268292682926831e-06, + "loss": 0.6214, + "step": 266 + }, + { + "epoch": 0.11, + "learning_rate": 9.303135888501744e-06, + "loss": 0.4923, + "step": 267 + }, + { + "epoch": 0.11, + "learning_rate": 9.337979094076656e-06, + "loss": 0.5257, + "step": 268 + }, + { + "epoch": 0.11, + "learning_rate": 9.372822299651569e-06, + "loss": 0.6273, + "step": 269 + }, + { + "epoch": 0.11, + "learning_rate": 9.407665505226482e-06, + "loss": 0.638, + "step": 270 + }, + { + "epoch": 0.11, + "learning_rate": 9.442508710801394e-06, + "loss": 0.5369, + "step": 271 + }, + { + "epoch": 0.11, + "learning_rate": 9.477351916376307e-06, + "loss": 0.6128, + "step": 272 + }, + { + "epoch": 0.11, + "learning_rate": 9.51219512195122e-06, + "loss": 0.61, + "step": 273 + }, + { + "epoch": 0.11, + "learning_rate": 9.547038327526134e-06, + "loss": 0.6117, + "step": 274 + }, + { + "epoch": 0.12, + "learning_rate": 9.581881533101046e-06, + "loss": 0.5191, + "step": 275 + }, + { + "epoch": 0.12, + "learning_rate": 9.616724738675959e-06, + "loss": 0.5381, + "step": 276 + }, + { + "epoch": 0.12, + "learning_rate": 9.651567944250871e-06, + "loss": 0.6686, + "step": 277 + }, + { + "epoch": 0.12, + "learning_rate": 9.686411149825786e-06, + "loss": 0.6415, + "step": 278 + }, + { + "epoch": 0.12, + "learning_rate": 9.721254355400698e-06, + "loss": 0.5333, + "step": 279 + }, + { + "epoch": 0.12, + "learning_rate": 9.756097560975611e-06, + "loss": 0.6254, + "step": 280 + }, + { + "epoch": 0.12, + "learning_rate": 9.790940766550524e-06, + "loss": 0.5899, + "step": 281 + }, + { + "epoch": 0.12, + "learning_rate": 9.825783972125436e-06, + "loss": 0.5635, + "step": 282 + }, + { + "epoch": 0.12, + "learning_rate": 9.860627177700349e-06, + "loss": 0.5338, + "step": 283 + }, + { + "epoch": 0.12, + "learning_rate": 9.895470383275261e-06, + "loss": 0.6047, + "step": 284 + }, + { + "epoch": 0.12, + "learning_rate": 9.930313588850174e-06, + "loss": 0.6124, + "step": 285 + }, + { + "epoch": 0.12, + "learning_rate": 9.965156794425088e-06, + "loss": 0.5827, + "step": 286 + }, + { + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.52, + "step": 287 + }, + { + "epoch": 0.12, + "learning_rate": 1.0034843205574913e-05, + "loss": 0.6033, + "step": 288 + }, + { + "epoch": 0.12, + "learning_rate": 1.0069686411149826e-05, + "loss": 0.5744, + "step": 289 + }, + { + "epoch": 0.12, + "learning_rate": 1.0104529616724739e-05, + "loss": 0.6823, + "step": 290 + }, + { + "epoch": 0.12, + "learning_rate": 1.0139372822299653e-05, + "loss": 0.6204, + "step": 291 + }, + { + "epoch": 0.12, + "learning_rate": 1.0174216027874565e-05, + "loss": 0.5409, + "step": 292 + }, + { + "epoch": 0.12, + "learning_rate": 1.0209059233449478e-05, + "loss": 0.5721, + "step": 293 + }, + { + "epoch": 0.12, + "learning_rate": 1.024390243902439e-05, + "loss": 0.6182, + "step": 294 + }, + { + "epoch": 0.12, + "learning_rate": 1.0278745644599303e-05, + "loss": 0.579, + "step": 295 + }, + { + "epoch": 0.12, + "learning_rate": 1.0313588850174216e-05, + "loss": 0.6389, + "step": 296 + }, + { + "epoch": 0.12, + "learning_rate": 1.0348432055749128e-05, + "loss": 0.6308, + "step": 297 + }, + { + "epoch": 0.12, + "learning_rate": 1.0383275261324041e-05, + "loss": 0.5235, + "step": 298 + }, + { + "epoch": 0.13, + "learning_rate": 1.0418118466898957e-05, + "loss": 0.5905, + "step": 299 + }, + { + "epoch": 0.13, + "learning_rate": 1.045296167247387e-05, + "loss": 0.6244, + "step": 300 + }, + { + "epoch": 0.13, + "learning_rate": 1.0487804878048782e-05, + "loss": 0.692, + "step": 301 + }, + { + "epoch": 0.13, + "learning_rate": 1.0522648083623695e-05, + "loss": 0.5482, + "step": 302 + }, + { + "epoch": 0.13, + "learning_rate": 1.0557491289198607e-05, + "loss": 0.6385, + "step": 303 + }, + { + "epoch": 0.13, + "learning_rate": 1.059233449477352e-05, + "loss": 0.643, + "step": 304 + }, + { + "epoch": 0.13, + "learning_rate": 1.0627177700348433e-05, + "loss": 0.5917, + "step": 305 + }, + { + "epoch": 0.13, + "learning_rate": 1.0662020905923345e-05, + "loss": 0.6183, + "step": 306 + }, + { + "epoch": 0.13, + "learning_rate": 1.069686411149826e-05, + "loss": 0.557, + "step": 307 + }, + { + "epoch": 0.13, + "learning_rate": 1.0731707317073172e-05, + "loss": 0.6811, + "step": 308 + }, + { + "epoch": 0.13, + "learning_rate": 1.0766550522648085e-05, + "loss": 0.6266, + "step": 309 + }, + { + "epoch": 0.13, + "learning_rate": 1.0801393728222997e-05, + "loss": 0.5146, + "step": 310 + }, + { + "epoch": 0.13, + "learning_rate": 1.083623693379791e-05, + "loss": 0.6395, + "step": 311 + }, + { + "epoch": 0.13, + "learning_rate": 1.0871080139372822e-05, + "loss": 0.6394, + "step": 312 + }, + { + "epoch": 0.13, + "learning_rate": 1.0905923344947735e-05, + "loss": 0.5787, + "step": 313 + }, + { + "epoch": 0.13, + "learning_rate": 1.0940766550522648e-05, + "loss": 0.5919, + "step": 314 + }, + { + "epoch": 0.13, + "learning_rate": 1.0975609756097562e-05, + "loss": 0.5283, + "step": 315 + }, + { + "epoch": 0.13, + "learning_rate": 1.1010452961672475e-05, + "loss": 0.5572, + "step": 316 + }, + { + "epoch": 0.13, + "learning_rate": 1.1045296167247389e-05, + "loss": 0.5774, + "step": 317 + }, + { + "epoch": 0.13, + "learning_rate": 1.1080139372822301e-05, + "loss": 0.5864, + "step": 318 + }, + { + "epoch": 0.13, + "learning_rate": 1.1114982578397214e-05, + "loss": 0.5731, + "step": 319 + }, + { + "epoch": 0.13, + "learning_rate": 1.1149825783972127e-05, + "loss": 0.5627, + "step": 320 + }, + { + "epoch": 0.13, + "learning_rate": 1.118466898954704e-05, + "loss": 0.5659, + "step": 321 + }, + { + "epoch": 0.13, + "learning_rate": 1.1219512195121953e-05, + "loss": 0.6295, + "step": 322 + }, + { + "epoch": 0.14, + "learning_rate": 1.1254355400696866e-05, + "loss": 0.5662, + "step": 323 + }, + { + "epoch": 0.14, + "learning_rate": 1.1289198606271779e-05, + "loss": 0.5659, + "step": 324 + }, + { + "epoch": 0.14, + "learning_rate": 1.1324041811846691e-05, + "loss": 0.74, + "step": 325 + }, + { + "epoch": 0.14, + "learning_rate": 1.1358885017421604e-05, + "loss": 0.604, + "step": 326 + }, + { + "epoch": 0.14, + "learning_rate": 1.1393728222996516e-05, + "loss": 0.5228, + "step": 327 + }, + { + "epoch": 0.14, + "learning_rate": 1.1428571428571429e-05, + "loss": 0.5745, + "step": 328 + }, + { + "epoch": 0.14, + "learning_rate": 1.1463414634146342e-05, + "loss": 0.5464, + "step": 329 + }, + { + "epoch": 0.14, + "learning_rate": 1.1498257839721256e-05, + "loss": 0.5523, + "step": 330 + }, + { + "epoch": 0.14, + "learning_rate": 1.1533101045296168e-05, + "loss": 0.5893, + "step": 331 + }, + { + "epoch": 0.14, + "learning_rate": 1.1567944250871081e-05, + "loss": 0.6265, + "step": 332 + }, + { + "epoch": 0.14, + "learning_rate": 1.1602787456445994e-05, + "loss": 0.6626, + "step": 333 + }, + { + "epoch": 0.14, + "learning_rate": 1.1637630662020906e-05, + "loss": 0.6379, + "step": 334 + }, + { + "epoch": 0.14, + "learning_rate": 1.1672473867595819e-05, + "loss": 0.6768, + "step": 335 + }, + { + "epoch": 0.14, + "learning_rate": 1.1707317073170731e-05, + "loss": 0.5926, + "step": 336 + }, + { + "epoch": 0.14, + "learning_rate": 1.1742160278745644e-05, + "loss": 0.5531, + "step": 337 + }, + { + "epoch": 0.14, + "learning_rate": 1.177700348432056e-05, + "loss": 0.5478, + "step": 338 + }, + { + "epoch": 0.14, + "learning_rate": 1.1811846689895473e-05, + "loss": 0.5894, + "step": 339 + }, + { + "epoch": 0.14, + "learning_rate": 1.1846689895470385e-05, + "loss": 0.5819, + "step": 340 + }, + { + "epoch": 0.14, + "learning_rate": 1.1881533101045298e-05, + "loss": 0.4793, + "step": 341 + }, + { + "epoch": 0.14, + "learning_rate": 1.191637630662021e-05, + "loss": 0.6046, + "step": 342 + }, + { + "epoch": 0.14, + "learning_rate": 1.1951219512195123e-05, + "loss": 0.5306, + "step": 343 + }, + { + "epoch": 0.14, + "learning_rate": 1.1986062717770036e-05, + "loss": 0.6153, + "step": 344 + }, + { + "epoch": 0.14, + "learning_rate": 1.2020905923344948e-05, + "loss": 0.6093, + "step": 345 + }, + { + "epoch": 0.14, + "learning_rate": 1.2055749128919862e-05, + "loss": 0.6022, + "step": 346 + }, + { + "epoch": 0.15, + "learning_rate": 1.2090592334494775e-05, + "loss": 0.515, + "step": 347 + }, + { + "epoch": 0.15, + "learning_rate": 1.2125435540069688e-05, + "loss": 0.6889, + "step": 348 + }, + { + "epoch": 0.15, + "learning_rate": 1.21602787456446e-05, + "loss": 0.576, + "step": 349 + }, + { + "epoch": 0.15, + "learning_rate": 1.2195121951219513e-05, + "loss": 0.5454, + "step": 350 + }, + { + "epoch": 0.15, + "learning_rate": 1.2229965156794425e-05, + "loss": 0.4593, + "step": 351 + }, + { + "epoch": 0.15, + "learning_rate": 1.2264808362369338e-05, + "loss": 0.5302, + "step": 352 + }, + { + "epoch": 0.15, + "learning_rate": 1.2299651567944252e-05, + "loss": 0.5591, + "step": 353 + }, + { + "epoch": 0.15, + "learning_rate": 1.2334494773519165e-05, + "loss": 0.6179, + "step": 354 + }, + { + "epoch": 0.15, + "learning_rate": 1.2369337979094078e-05, + "loss": 0.6015, + "step": 355 + }, + { + "epoch": 0.15, + "learning_rate": 1.240418118466899e-05, + "loss": 0.5615, + "step": 356 + }, + { + "epoch": 0.15, + "learning_rate": 1.2439024390243903e-05, + "loss": 0.6798, + "step": 357 + }, + { + "epoch": 0.15, + "learning_rate": 1.2473867595818815e-05, + "loss": 0.578, + "step": 358 + }, + { + "epoch": 0.15, + "learning_rate": 1.2508710801393728e-05, + "loss": 0.629, + "step": 359 + }, + { + "epoch": 0.15, + "learning_rate": 1.254355400696864e-05, + "loss": 0.6236, + "step": 360 + }, + { + "epoch": 0.15, + "learning_rate": 1.2578397212543556e-05, + "loss": 0.577, + "step": 361 + }, + { + "epoch": 0.15, + "learning_rate": 1.2613240418118469e-05, + "loss": 0.5588, + "step": 362 + }, + { + "epoch": 0.15, + "learning_rate": 1.2648083623693382e-05, + "loss": 0.6655, + "step": 363 + }, + { + "epoch": 0.15, + "learning_rate": 1.2682926829268294e-05, + "loss": 0.6124, + "step": 364 + }, + { + "epoch": 0.15, + "learning_rate": 1.2717770034843207e-05, + "loss": 0.4786, + "step": 365 + }, + { + "epoch": 0.15, + "learning_rate": 1.275261324041812e-05, + "loss": 0.566, + "step": 366 + }, + { + "epoch": 0.15, + "learning_rate": 1.2787456445993032e-05, + "loss": 0.5676, + "step": 367 + }, + { + "epoch": 0.15, + "learning_rate": 1.2822299651567945e-05, + "loss": 0.6284, + "step": 368 + }, + { + "epoch": 0.15, + "learning_rate": 1.2857142857142859e-05, + "loss": 0.5828, + "step": 369 + }, + { + "epoch": 0.15, + "learning_rate": 1.2891986062717772e-05, + "loss": 0.6464, + "step": 370 + }, + { + "epoch": 0.16, + "learning_rate": 1.2926829268292684e-05, + "loss": 0.5833, + "step": 371 + }, + { + "epoch": 0.16, + "learning_rate": 1.2961672473867597e-05, + "loss": 0.578, + "step": 372 + }, + { + "epoch": 0.16, + "learning_rate": 1.299651567944251e-05, + "loss": 0.5904, + "step": 373 + }, + { + "epoch": 0.16, + "learning_rate": 1.3031358885017422e-05, + "loss": 0.5454, + "step": 374 + }, + { + "epoch": 0.16, + "learning_rate": 1.3066202090592334e-05, + "loss": 0.5759, + "step": 375 + }, + { + "epoch": 0.16, + "learning_rate": 1.3101045296167247e-05, + "loss": 0.5288, + "step": 376 + }, + { + "epoch": 0.16, + "learning_rate": 1.3135888501742163e-05, + "loss": 0.6116, + "step": 377 + }, + { + "epoch": 0.16, + "learning_rate": 1.3170731707317076e-05, + "loss": 0.5446, + "step": 378 + }, + { + "epoch": 0.16, + "learning_rate": 1.3205574912891988e-05, + "loss": 0.5642, + "step": 379 + }, + { + "epoch": 0.16, + "learning_rate": 1.32404181184669e-05, + "loss": 0.5638, + "step": 380 + }, + { + "epoch": 0.16, + "learning_rate": 1.3275261324041813e-05, + "loss": 0.779, + "step": 381 + }, + { + "epoch": 0.16, + "learning_rate": 1.3310104529616726e-05, + "loss": 0.5757, + "step": 382 + }, + { + "epoch": 0.16, + "learning_rate": 1.3344947735191639e-05, + "loss": 0.6283, + "step": 383 + }, + { + "epoch": 0.16, + "learning_rate": 1.3379790940766553e-05, + "loss": 0.6289, + "step": 384 + }, + { + "epoch": 0.16, + "learning_rate": 1.3414634146341466e-05, + "loss": 0.5548, + "step": 385 + }, + { + "epoch": 0.16, + "learning_rate": 1.3449477351916378e-05, + "loss": 0.5725, + "step": 386 + }, + { + "epoch": 0.16, + "learning_rate": 1.348432055749129e-05, + "loss": 0.6109, + "step": 387 + }, + { + "epoch": 0.16, + "learning_rate": 1.3519163763066203e-05, + "loss": 0.5975, + "step": 388 + }, + { + "epoch": 0.16, + "learning_rate": 1.3554006968641116e-05, + "loss": 0.5972, + "step": 389 + }, + { + "epoch": 0.16, + "learning_rate": 1.3588850174216028e-05, + "loss": 0.6673, + "step": 390 + }, + { + "epoch": 0.16, + "learning_rate": 1.3623693379790941e-05, + "loss": 0.6687, + "step": 391 + }, + { + "epoch": 0.16, + "learning_rate": 1.3658536585365855e-05, + "loss": 0.504, + "step": 392 + }, + { + "epoch": 0.16, + "learning_rate": 1.3693379790940768e-05, + "loss": 0.6478, + "step": 393 + }, + { + "epoch": 0.16, + "learning_rate": 1.372822299651568e-05, + "loss": 0.6042, + "step": 394 + }, + { + "epoch": 0.17, + "learning_rate": 1.3763066202090593e-05, + "loss": 0.6066, + "step": 395 + }, + { + "epoch": 0.17, + "learning_rate": 1.3797909407665506e-05, + "loss": 0.5724, + "step": 396 + }, + { + "epoch": 0.17, + "learning_rate": 1.3832752613240418e-05, + "loss": 0.6821, + "step": 397 + }, + { + "epoch": 0.17, + "learning_rate": 1.3867595818815331e-05, + "loss": 0.544, + "step": 398 + }, + { + "epoch": 0.17, + "learning_rate": 1.3902439024390244e-05, + "loss": 0.5192, + "step": 399 + }, + { + "epoch": 0.17, + "learning_rate": 1.393728222996516e-05, + "loss": 0.6457, + "step": 400 + }, + { + "epoch": 0.17, + "learning_rate": 1.3972125435540072e-05, + "loss": 0.5996, + "step": 401 + }, + { + "epoch": 0.17, + "learning_rate": 1.4006968641114985e-05, + "loss": 0.5737, + "step": 402 + }, + { + "epoch": 0.17, + "learning_rate": 1.4041811846689897e-05, + "loss": 0.6506, + "step": 403 + }, + { + "epoch": 0.17, + "learning_rate": 1.407665505226481e-05, + "loss": 0.5541, + "step": 404 + }, + { + "epoch": 0.17, + "learning_rate": 1.4111498257839722e-05, + "loss": 0.5502, + "step": 405 + }, + { + "epoch": 0.17, + "learning_rate": 1.4146341463414635e-05, + "loss": 0.5909, + "step": 406 + }, + { + "epoch": 0.17, + "learning_rate": 1.4181184668989548e-05, + "loss": 0.5953, + "step": 407 + }, + { + "epoch": 0.17, + "learning_rate": 1.4216027874564462e-05, + "loss": 0.5376, + "step": 408 + }, + { + "epoch": 0.17, + "learning_rate": 1.4250871080139375e-05, + "loss": 0.5377, + "step": 409 + }, + { + "epoch": 0.17, + "learning_rate": 1.4285714285714287e-05, + "loss": 0.6134, + "step": 410 + }, + { + "epoch": 0.17, + "learning_rate": 1.43205574912892e-05, + "loss": 0.6472, + "step": 411 + }, + { + "epoch": 0.17, + "learning_rate": 1.4355400696864112e-05, + "loss": 0.4815, + "step": 412 + }, + { + "epoch": 0.17, + "learning_rate": 1.4390243902439025e-05, + "loss": 0.4832, + "step": 413 + }, + { + "epoch": 0.17, + "learning_rate": 1.4425087108013938e-05, + "loss": 0.5455, + "step": 414 + }, + { + "epoch": 0.17, + "learning_rate": 1.4459930313588852e-05, + "loss": 0.5445, + "step": 415 + }, + { + "epoch": 0.17, + "learning_rate": 1.4494773519163764e-05, + "loss": 0.5713, + "step": 416 + }, + { + "epoch": 0.17, + "learning_rate": 1.4529616724738677e-05, + "loss": 0.5713, + "step": 417 + }, + { + "epoch": 0.18, + "learning_rate": 1.456445993031359e-05, + "loss": 0.6725, + "step": 418 + }, + { + "epoch": 0.18, + "learning_rate": 1.4599303135888502e-05, + "loss": 0.6199, + "step": 419 + }, + { + "epoch": 0.18, + "learning_rate": 1.4634146341463415e-05, + "loss": 0.5049, + "step": 420 + }, + { + "epoch": 0.18, + "learning_rate": 1.4668989547038327e-05, + "loss": 0.5712, + "step": 421 + }, + { + "epoch": 0.18, + "learning_rate": 1.470383275261324e-05, + "loss": 0.6426, + "step": 422 + }, + { + "epoch": 0.18, + "learning_rate": 1.4738675958188156e-05, + "loss": 0.5879, + "step": 423 + }, + { + "epoch": 0.18, + "learning_rate": 1.4773519163763069e-05, + "loss": 0.5529, + "step": 424 + }, + { + "epoch": 0.18, + "learning_rate": 1.4808362369337981e-05, + "loss": 0.6029, + "step": 425 + }, + { + "epoch": 0.18, + "learning_rate": 1.4843205574912894e-05, + "loss": 0.6285, + "step": 426 + }, + { + "epoch": 0.18, + "learning_rate": 1.4878048780487806e-05, + "loss": 0.5832, + "step": 427 + }, + { + "epoch": 0.18, + "learning_rate": 1.4912891986062719e-05, + "loss": 0.506, + "step": 428 + }, + { + "epoch": 0.18, + "learning_rate": 1.4947735191637631e-05, + "loss": 0.6044, + "step": 429 + }, + { + "epoch": 0.18, + "learning_rate": 1.4982578397212544e-05, + "loss": 0.5128, + "step": 430 + }, + { + "epoch": 0.18, + "learning_rate": 1.5017421602787458e-05, + "loss": 0.5998, + "step": 431 + }, + { + "epoch": 0.18, + "learning_rate": 1.5052264808362371e-05, + "loss": 0.6959, + "step": 432 + }, + { + "epoch": 0.18, + "learning_rate": 1.5087108013937284e-05, + "loss": 0.6206, + "step": 433 + }, + { + "epoch": 0.18, + "learning_rate": 1.5121951219512196e-05, + "loss": 0.6337, + "step": 434 + }, + { + "epoch": 0.18, + "learning_rate": 1.5156794425087109e-05, + "loss": 0.5991, + "step": 435 + }, + { + "epoch": 0.18, + "learning_rate": 1.5191637630662021e-05, + "loss": 0.6201, + "step": 436 + }, + { + "epoch": 0.18, + "learning_rate": 1.5226480836236934e-05, + "loss": 0.4951, + "step": 437 + }, + { + "epoch": 0.18, + "learning_rate": 1.5261324041811848e-05, + "loss": 0.5876, + "step": 438 + }, + { + "epoch": 0.18, + "learning_rate": 1.529616724738676e-05, + "loss": 0.5744, + "step": 439 + }, + { + "epoch": 0.18, + "learning_rate": 1.5331010452961673e-05, + "loss": 0.5939, + "step": 440 + }, + { + "epoch": 0.18, + "learning_rate": 1.5365853658536586e-05, + "loss": 0.6046, + "step": 441 + }, + { + "epoch": 0.19, + "learning_rate": 1.54006968641115e-05, + "loss": 0.656, + "step": 442 + }, + { + "epoch": 0.19, + "learning_rate": 1.543554006968641e-05, + "loss": 0.6, + "step": 443 + }, + { + "epoch": 0.19, + "learning_rate": 1.5470383275261324e-05, + "loss": 0.5162, + "step": 444 + }, + { + "epoch": 0.19, + "learning_rate": 1.5505226480836236e-05, + "loss": 0.5459, + "step": 445 + }, + { + "epoch": 0.19, + "learning_rate": 1.554006968641115e-05, + "loss": 0.486, + "step": 446 + }, + { + "epoch": 0.19, + "learning_rate": 1.5574912891986065e-05, + "loss": 0.6013, + "step": 447 + }, + { + "epoch": 0.19, + "learning_rate": 1.5609756097560978e-05, + "loss": 0.6249, + "step": 448 + }, + { + "epoch": 0.19, + "learning_rate": 1.564459930313589e-05, + "loss": 0.545, + "step": 449 + }, + { + "epoch": 0.19, + "learning_rate": 1.5679442508710803e-05, + "loss": 0.6635, + "step": 450 + }, + { + "epoch": 0.19, + "learning_rate": 1.5714285714285715e-05, + "loss": 0.6335, + "step": 451 + }, + { + "epoch": 0.19, + "learning_rate": 1.5749128919860628e-05, + "loss": 0.5401, + "step": 452 + }, + { + "epoch": 0.19, + "learning_rate": 1.578397212543554e-05, + "loss": 0.5021, + "step": 453 + }, + { + "epoch": 0.19, + "learning_rate": 1.5818815331010456e-05, + "loss": 0.5869, + "step": 454 + }, + { + "epoch": 0.19, + "learning_rate": 1.585365853658537e-05, + "loss": 0.6381, + "step": 455 + }, + { + "epoch": 0.19, + "learning_rate": 1.588850174216028e-05, + "loss": 0.4976, + "step": 456 + }, + { + "epoch": 0.19, + "learning_rate": 1.5923344947735194e-05, + "loss": 0.5761, + "step": 457 + }, + { + "epoch": 0.19, + "learning_rate": 1.5958188153310107e-05, + "loss": 0.6614, + "step": 458 + }, + { + "epoch": 0.19, + "learning_rate": 1.599303135888502e-05, + "loss": 0.6228, + "step": 459 + }, + { + "epoch": 0.19, + "learning_rate": 1.6027874564459932e-05, + "loss": 0.5755, + "step": 460 + }, + { + "epoch": 0.19, + "learning_rate": 1.6062717770034845e-05, + "loss": 0.5943, + "step": 461 + }, + { + "epoch": 0.19, + "learning_rate": 1.6097560975609757e-05, + "loss": 0.5631, + "step": 462 + }, + { + "epoch": 0.19, + "learning_rate": 1.613240418118467e-05, + "loss": 0.6802, + "step": 463 + }, + { + "epoch": 0.19, + "learning_rate": 1.6167247386759582e-05, + "loss": 0.5835, + "step": 464 + }, + { + "epoch": 0.19, + "learning_rate": 1.6202090592334495e-05, + "loss": 0.5725, + "step": 465 + }, + { + "epoch": 0.2, + "learning_rate": 1.6236933797909408e-05, + "loss": 0.5487, + "step": 466 + }, + { + "epoch": 0.2, + "learning_rate": 1.627177700348432e-05, + "loss": 0.5889, + "step": 467 + }, + { + "epoch": 0.2, + "learning_rate": 1.6306620209059233e-05, + "loss": 0.5585, + "step": 468 + }, + { + "epoch": 0.2, + "learning_rate": 1.6341463414634145e-05, + "loss": 0.5731, + "step": 469 + }, + { + "epoch": 0.2, + "learning_rate": 1.637630662020906e-05, + "loss": 0.6129, + "step": 470 + }, + { + "epoch": 0.2, + "learning_rate": 1.6411149825783974e-05, + "loss": 0.586, + "step": 471 + }, + { + "epoch": 0.2, + "learning_rate": 1.6445993031358887e-05, + "loss": 0.6418, + "step": 472 + }, + { + "epoch": 0.2, + "learning_rate": 1.64808362369338e-05, + "loss": 0.6442, + "step": 473 + }, + { + "epoch": 0.2, + "learning_rate": 1.6515679442508712e-05, + "loss": 0.5419, + "step": 474 + }, + { + "epoch": 0.2, + "learning_rate": 1.6550522648083624e-05, + "loss": 0.5282, + "step": 475 + }, + { + "epoch": 0.2, + "learning_rate": 1.6585365853658537e-05, + "loss": 0.5588, + "step": 476 + }, + { + "epoch": 0.2, + "learning_rate": 1.662020905923345e-05, + "loss": 0.5536, + "step": 477 + }, + { + "epoch": 0.2, + "learning_rate": 1.6655052264808366e-05, + "loss": 0.5514, + "step": 478 + }, + { + "epoch": 0.2, + "learning_rate": 1.6689895470383278e-05, + "loss": 0.5554, + "step": 479 + }, + { + "epoch": 0.2, + "learning_rate": 1.672473867595819e-05, + "loss": 0.555, + "step": 480 + }, + { + "epoch": 0.2, + "learning_rate": 1.6759581881533103e-05, + "loss": 0.5606, + "step": 481 + }, + { + "epoch": 0.2, + "learning_rate": 1.6794425087108016e-05, + "loss": 0.5154, + "step": 482 + }, + { + "epoch": 0.2, + "learning_rate": 1.682926829268293e-05, + "loss": 0.5407, + "step": 483 + }, + { + "epoch": 0.2, + "learning_rate": 1.686411149825784e-05, + "loss": 0.6047, + "step": 484 + }, + { + "epoch": 0.2, + "learning_rate": 1.6898954703832754e-05, + "loss": 0.5323, + "step": 485 + }, + { + "epoch": 0.2, + "learning_rate": 1.6933797909407666e-05, + "loss": 0.5395, + "step": 486 + }, + { + "epoch": 0.2, + "learning_rate": 1.696864111498258e-05, + "loss": 0.6181, + "step": 487 + }, + { + "epoch": 0.2, + "learning_rate": 1.700348432055749e-05, + "loss": 0.5764, + "step": 488 + }, + { + "epoch": 0.2, + "learning_rate": 1.7038327526132404e-05, + "loss": 0.6155, + "step": 489 + }, + { + "epoch": 0.21, + "learning_rate": 1.7073170731707317e-05, + "loss": 0.5058, + "step": 490 + }, + { + "epoch": 0.21, + "learning_rate": 1.710801393728223e-05, + "loss": 0.6552, + "step": 491 + }, + { + "epoch": 0.21, + "learning_rate": 1.7142857142857142e-05, + "loss": 0.5362, + "step": 492 + }, + { + "epoch": 0.21, + "learning_rate": 1.7177700348432058e-05, + "loss": 0.6479, + "step": 493 + }, + { + "epoch": 0.21, + "learning_rate": 1.721254355400697e-05, + "loss": 0.4939, + "step": 494 + }, + { + "epoch": 0.21, + "learning_rate": 1.7247386759581883e-05, + "loss": 0.6442, + "step": 495 + }, + { + "epoch": 0.21, + "learning_rate": 1.7282229965156796e-05, + "loss": 0.5753, + "step": 496 + }, + { + "epoch": 0.21, + "learning_rate": 1.7317073170731708e-05, + "loss": 0.701, + "step": 497 + }, + { + "epoch": 0.21, + "learning_rate": 1.735191637630662e-05, + "loss": 0.5235, + "step": 498 + }, + { + "epoch": 0.21, + "learning_rate": 1.7386759581881533e-05, + "loss": 0.4857, + "step": 499 + }, + { + "epoch": 0.21, + "learning_rate": 1.7421602787456446e-05, + "loss": 0.5072, + "step": 500 + }, + { + "epoch": 0.21, + "learning_rate": 1.7456445993031362e-05, + "loss": 0.6019, + "step": 501 + }, + { + "epoch": 0.21, + "learning_rate": 1.7491289198606275e-05, + "loss": 0.5263, + "step": 502 + }, + { + "epoch": 0.21, + "learning_rate": 1.7526132404181187e-05, + "loss": 0.5893, + "step": 503 + }, + { + "epoch": 0.21, + "learning_rate": 1.75609756097561e-05, + "loss": 0.6769, + "step": 504 + }, + { + "epoch": 0.21, + "learning_rate": 1.7595818815331012e-05, + "loss": 0.5069, + "step": 505 + }, + { + "epoch": 0.21, + "learning_rate": 1.7630662020905925e-05, + "loss": 0.6241, + "step": 506 + }, + { + "epoch": 0.21, + "learning_rate": 1.7665505226480838e-05, + "loss": 0.5585, + "step": 507 + }, + { + "epoch": 0.21, + "learning_rate": 1.770034843205575e-05, + "loss": 0.5546, + "step": 508 + }, + { + "epoch": 0.21, + "learning_rate": 1.7735191637630663e-05, + "loss": 0.5824, + "step": 509 + }, + { + "epoch": 0.21, + "learning_rate": 1.7770034843205575e-05, + "loss": 0.5741, + "step": 510 + }, + { + "epoch": 0.21, + "learning_rate": 1.7804878048780488e-05, + "loss": 0.5867, + "step": 511 + }, + { + "epoch": 0.21, + "learning_rate": 1.78397212543554e-05, + "loss": 0.5562, + "step": 512 + }, + { + "epoch": 0.21, + "learning_rate": 1.7874564459930313e-05, + "loss": 0.5487, + "step": 513 + }, + { + "epoch": 0.22, + "learning_rate": 1.7909407665505226e-05, + "loss": 0.6476, + "step": 514 + }, + { + "epoch": 0.22, + "learning_rate": 1.7944250871080138e-05, + "loss": 0.5481, + "step": 515 + }, + { + "epoch": 0.22, + "learning_rate": 1.7979094076655054e-05, + "loss": 0.5381, + "step": 516 + }, + { + "epoch": 0.22, + "learning_rate": 1.8013937282229967e-05, + "loss": 0.5591, + "step": 517 + }, + { + "epoch": 0.22, + "learning_rate": 1.804878048780488e-05, + "loss": 0.591, + "step": 518 + }, + { + "epoch": 0.22, + "learning_rate": 1.8083623693379792e-05, + "loss": 0.465, + "step": 519 + }, + { + "epoch": 0.22, + "learning_rate": 1.8118466898954705e-05, + "loss": 0.6022, + "step": 520 + }, + { + "epoch": 0.22, + "learning_rate": 1.8153310104529617e-05, + "loss": 0.4784, + "step": 521 + }, + { + "epoch": 0.22, + "learning_rate": 1.818815331010453e-05, + "loss": 0.5983, + "step": 522 + }, + { + "epoch": 0.22, + "learning_rate": 1.8222996515679442e-05, + "loss": 0.5676, + "step": 523 + }, + { + "epoch": 0.22, + "learning_rate": 1.825783972125436e-05, + "loss": 0.5733, + "step": 524 + }, + { + "epoch": 0.22, + "learning_rate": 1.829268292682927e-05, + "loss": 0.5837, + "step": 525 + }, + { + "epoch": 0.22, + "learning_rate": 1.8327526132404184e-05, + "loss": 0.5893, + "step": 526 + }, + { + "epoch": 0.22, + "learning_rate": 1.8362369337979096e-05, + "loss": 0.6748, + "step": 527 + }, + { + "epoch": 0.22, + "learning_rate": 1.839721254355401e-05, + "loss": 0.5764, + "step": 528 + }, + { + "epoch": 0.22, + "learning_rate": 1.843205574912892e-05, + "loss": 0.6803, + "step": 529 + }, + { + "epoch": 0.22, + "learning_rate": 1.8466898954703834e-05, + "loss": 0.5717, + "step": 530 + }, + { + "epoch": 0.22, + "learning_rate": 1.8501742160278747e-05, + "loss": 0.5755, + "step": 531 + }, + { + "epoch": 0.22, + "learning_rate": 1.8536585365853663e-05, + "loss": 0.5581, + "step": 532 + }, + { + "epoch": 0.22, + "learning_rate": 1.8571428571428575e-05, + "loss": 0.5884, + "step": 533 + }, + { + "epoch": 0.22, + "learning_rate": 1.8606271777003488e-05, + "loss": 0.5363, + "step": 534 + }, + { + "epoch": 0.22, + "learning_rate": 1.86411149825784e-05, + "loss": 0.5798, + "step": 535 + }, + { + "epoch": 0.22, + "learning_rate": 1.8675958188153313e-05, + "loss": 0.5236, + "step": 536 + }, + { + "epoch": 0.22, + "learning_rate": 1.8710801393728226e-05, + "loss": 0.5025, + "step": 537 + }, + { + "epoch": 0.23, + "learning_rate": 1.8745644599303138e-05, + "loss": 0.6242, + "step": 538 + }, + { + "epoch": 0.23, + "learning_rate": 1.878048780487805e-05, + "loss": 0.6023, + "step": 539 + }, + { + "epoch": 0.23, + "learning_rate": 1.8815331010452963e-05, + "loss": 0.6211, + "step": 540 + }, + { + "epoch": 0.23, + "learning_rate": 1.8850174216027876e-05, + "loss": 0.5077, + "step": 541 + }, + { + "epoch": 0.23, + "learning_rate": 1.888501742160279e-05, + "loss": 0.5018, + "step": 542 + }, + { + "epoch": 0.23, + "learning_rate": 1.89198606271777e-05, + "loss": 0.6053, + "step": 543 + }, + { + "epoch": 0.23, + "learning_rate": 1.8954703832752614e-05, + "loss": 0.6225, + "step": 544 + }, + { + "epoch": 0.23, + "learning_rate": 1.8989547038327526e-05, + "loss": 0.6134, + "step": 545 + }, + { + "epoch": 0.23, + "learning_rate": 1.902439024390244e-05, + "loss": 0.5871, + "step": 546 + }, + { + "epoch": 0.23, + "learning_rate": 1.9059233449477355e-05, + "loss": 0.7059, + "step": 547 + }, + { + "epoch": 0.23, + "learning_rate": 1.9094076655052267e-05, + "loss": 0.5859, + "step": 548 + }, + { + "epoch": 0.23, + "learning_rate": 1.912891986062718e-05, + "loss": 0.5957, + "step": 549 + }, + { + "epoch": 0.23, + "learning_rate": 1.9163763066202093e-05, + "loss": 0.5299, + "step": 550 + }, + { + "epoch": 0.23, + "learning_rate": 1.9198606271777005e-05, + "loss": 0.6158, + "step": 551 + }, + { + "epoch": 0.23, + "learning_rate": 1.9233449477351918e-05, + "loss": 0.5808, + "step": 552 + }, + { + "epoch": 0.23, + "learning_rate": 1.926829268292683e-05, + "loss": 0.573, + "step": 553 + }, + { + "epoch": 0.23, + "learning_rate": 1.9303135888501743e-05, + "loss": 0.6362, + "step": 554 + }, + { + "epoch": 0.23, + "learning_rate": 1.933797909407666e-05, + "loss": 0.5059, + "step": 555 + }, + { + "epoch": 0.23, + "learning_rate": 1.937282229965157e-05, + "loss": 0.5784, + "step": 556 + }, + { + "epoch": 0.23, + "learning_rate": 1.9407665505226484e-05, + "loss": 0.649, + "step": 557 + }, + { + "epoch": 0.23, + "learning_rate": 1.9442508710801397e-05, + "loss": 0.6128, + "step": 558 + }, + { + "epoch": 0.23, + "learning_rate": 1.947735191637631e-05, + "loss": 0.595, + "step": 559 + }, + { + "epoch": 0.23, + "learning_rate": 1.9512195121951222e-05, + "loss": 0.5921, + "step": 560 + }, + { + "epoch": 0.23, + "learning_rate": 1.9547038327526135e-05, + "loss": 0.5053, + "step": 561 + }, + { + "epoch": 0.24, + "learning_rate": 1.9581881533101047e-05, + "loss": 0.4499, + "step": 562 + }, + { + "epoch": 0.24, + "learning_rate": 1.961672473867596e-05, + "loss": 0.5918, + "step": 563 + }, + { + "epoch": 0.24, + "learning_rate": 1.9651567944250872e-05, + "loss": 0.5511, + "step": 564 + }, + { + "epoch": 0.24, + "learning_rate": 1.9686411149825785e-05, + "loss": 0.6467, + "step": 565 + }, + { + "epoch": 0.24, + "learning_rate": 1.9721254355400697e-05, + "loss": 0.5508, + "step": 566 + }, + { + "epoch": 0.24, + "learning_rate": 1.975609756097561e-05, + "loss": 0.5592, + "step": 567 + }, + { + "epoch": 0.24, + "learning_rate": 1.9790940766550523e-05, + "loss": 0.5966, + "step": 568 + }, + { + "epoch": 0.24, + "learning_rate": 1.9825783972125435e-05, + "loss": 0.5011, + "step": 569 + }, + { + "epoch": 0.24, + "learning_rate": 1.9860627177700348e-05, + "loss": 0.593, + "step": 570 + }, + { + "epoch": 0.24, + "learning_rate": 1.9895470383275264e-05, + "loss": 0.5659, + "step": 571 + }, + { + "epoch": 0.24, + "learning_rate": 1.9930313588850176e-05, + "loss": 0.5767, + "step": 572 + }, + { + "epoch": 0.24, + "learning_rate": 1.996515679442509e-05, + "loss": 0.5624, + "step": 573 + }, + { + "epoch": 0.24, + "learning_rate": 2e-05, + "loss": 0.4904, + "step": 574 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999985627944e-05, + "loss": 0.5734, + "step": 575 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999999425117754e-05, + "loss": 0.5985, + "step": 576 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999998706514958e-05, + "loss": 0.5729, + "step": 577 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999997700471074e-05, + "loss": 0.558, + "step": 578 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999640698613e-05, + "loss": 0.5446, + "step": 579 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999994826060166e-05, + "loss": 0.5934, + "step": 580 + }, + { + "epoch": 0.24, + "learning_rate": 1.999999295769322e-05, + "loss": 0.6511, + "step": 581 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999990801885353e-05, + "loss": 0.5262, + "step": 582 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999988358636622e-05, + "loss": 0.5777, + "step": 583 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999985627947104e-05, + "loss": 0.4609, + "step": 584 + }, + { + "epoch": 0.24, + "learning_rate": 1.9999982609816867e-05, + "loss": 0.6021, + "step": 585 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999979304246006e-05, + "loss": 0.6274, + "step": 586 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999975711234616e-05, + "loss": 0.517, + "step": 587 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999971830782795e-05, + "loss": 0.562, + "step": 588 + }, + { + "epoch": 0.25, + "learning_rate": 1.999996766289066e-05, + "loss": 0.5583, + "step": 589 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999963207558326e-05, + "loss": 0.5975, + "step": 590 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999958464785925e-05, + "loss": 0.6094, + "step": 591 + }, + { + "epoch": 0.25, + "learning_rate": 1.999995343457359e-05, + "loss": 0.5579, + "step": 592 + }, + { + "epoch": 0.25, + "learning_rate": 1.999994811692147e-05, + "loss": 0.551, + "step": 593 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999942511829714e-05, + "loss": 0.5576, + "step": 594 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999936619298483e-05, + "loss": 0.5486, + "step": 595 + }, + { + "epoch": 0.25, + "learning_rate": 1.999993043932795e-05, + "loss": 0.588, + "step": 596 + }, + { + "epoch": 0.25, + "learning_rate": 1.999992397191829e-05, + "loss": 0.5231, + "step": 597 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999917217069687e-05, + "loss": 0.5773, + "step": 598 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999910174782337e-05, + "loss": 0.5609, + "step": 599 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999902845056444e-05, + "loss": 0.6032, + "step": 600 + }, + { + "epoch": 0.25, + "learning_rate": 1.999989522789222e-05, + "loss": 0.5846, + "step": 601 + }, + { + "epoch": 0.25, + "learning_rate": 1.999988732328988e-05, + "loss": 0.6105, + "step": 602 + }, + { + "epoch": 0.25, + "learning_rate": 1.999987913124965e-05, + "loss": 0.5158, + "step": 603 + }, + { + "epoch": 0.25, + "learning_rate": 1.999987065177177e-05, + "loss": 0.6727, + "step": 604 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999861884856483e-05, + "loss": 0.5327, + "step": 605 + }, + { + "epoch": 0.25, + "learning_rate": 1.9999852830504038e-05, + "loss": 0.5825, + "step": 606 + }, + { + "epoch": 0.25, + "learning_rate": 1.99998434887147e-05, + "loss": 0.5495, + "step": 607 + }, + { + "epoch": 0.25, + "learning_rate": 1.999983385948873e-05, + "loss": 0.542, + "step": 608 + }, + { + "epoch": 0.25, + "learning_rate": 1.999982394282641e-05, + "loss": 0.6037, + "step": 609 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999813738728028e-05, + "loss": 0.6024, + "step": 610 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999803247193873e-05, + "loss": 0.6103, + "step": 611 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999792468224247e-05, + "loss": 0.6136, + "step": 612 + }, + { + "epoch": 0.26, + "learning_rate": 1.999978140181946e-05, + "loss": 0.5756, + "step": 613 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999770047979828e-05, + "loss": 0.6016, + "step": 614 + }, + { + "epoch": 0.26, + "learning_rate": 1.999975840670568e-05, + "loss": 0.6508, + "step": 615 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999746477997352e-05, + "loss": 0.5523, + "step": 616 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999734261855188e-05, + "loss": 0.6448, + "step": 617 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999721758279532e-05, + "loss": 0.5188, + "step": 618 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999708967270747e-05, + "loss": 0.5533, + "step": 619 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999695888829206e-05, + "loss": 0.4754, + "step": 620 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999682522955277e-05, + "loss": 0.5404, + "step": 621 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999668869649347e-05, + "loss": 0.5205, + "step": 622 + }, + { + "epoch": 0.26, + "learning_rate": 1.999965492891181e-05, + "loss": 0.5183, + "step": 623 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999640700743063e-05, + "loss": 0.5141, + "step": 624 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999626185143518e-05, + "loss": 0.5411, + "step": 625 + }, + { + "epoch": 0.26, + "learning_rate": 1.99996113821136e-05, + "loss": 0.5665, + "step": 626 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999596291653715e-05, + "loss": 0.5936, + "step": 627 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999580913764313e-05, + "loss": 0.5419, + "step": 628 + }, + { + "epoch": 0.26, + "learning_rate": 1.999956524844583e-05, + "loss": 0.5021, + "step": 629 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999549295698722e-05, + "loss": 0.5776, + "step": 630 + }, + { + "epoch": 0.26, + "learning_rate": 1.9999533055523437e-05, + "loss": 0.4954, + "step": 631 + }, + { + "epoch": 0.26, + "learning_rate": 1.999951652792045e-05, + "loss": 0.5628, + "step": 632 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999499712890235e-05, + "loss": 0.6356, + "step": 633 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999482610433275e-05, + "loss": 0.5385, + "step": 634 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999465220550057e-05, + "loss": 0.5615, + "step": 635 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999447543241086e-05, + "loss": 0.5947, + "step": 636 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999429578506867e-05, + "loss": 0.6466, + "step": 637 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999411326347924e-05, + "loss": 0.5652, + "step": 638 + }, + { + "epoch": 0.27, + "learning_rate": 1.999939278676477e-05, + "loss": 0.571, + "step": 639 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999373959757945e-05, + "loss": 0.6267, + "step": 640 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999354845327993e-05, + "loss": 0.5371, + "step": 641 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999335443475452e-05, + "loss": 0.5337, + "step": 642 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999315754200892e-05, + "loss": 0.542, + "step": 643 + }, + { + "epoch": 0.27, + "learning_rate": 1.999929577750487e-05, + "loss": 0.5757, + "step": 644 + }, + { + "epoch": 0.27, + "learning_rate": 1.999927551338797e-05, + "loss": 0.5567, + "step": 645 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999254961850765e-05, + "loss": 0.5668, + "step": 646 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999234122893848e-05, + "loss": 0.5143, + "step": 647 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999212996517823e-05, + "loss": 0.5506, + "step": 648 + }, + { + "epoch": 0.27, + "learning_rate": 1.999919158272329e-05, + "loss": 0.6022, + "step": 649 + }, + { + "epoch": 0.27, + "learning_rate": 1.999916988151087e-05, + "loss": 0.5667, + "step": 650 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999147892881182e-05, + "loss": 0.5127, + "step": 651 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999125616834866e-05, + "loss": 0.6013, + "step": 652 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999103053372554e-05, + "loss": 0.4694, + "step": 653 + }, + { + "epoch": 0.27, + "learning_rate": 1.99990802024949e-05, + "loss": 0.5073, + "step": 654 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999057064202555e-05, + "loss": 0.5867, + "step": 655 + }, + { + "epoch": 0.27, + "learning_rate": 1.9999033638496187e-05, + "loss": 0.5578, + "step": 656 + }, + { + "epoch": 0.28, + "learning_rate": 1.9999009925376475e-05, + "loss": 0.588, + "step": 657 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998985924844093e-05, + "loss": 0.5796, + "step": 658 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998961636899736e-05, + "loss": 0.559, + "step": 659 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998937061544096e-05, + "loss": 0.4719, + "step": 660 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998912198777886e-05, + "loss": 0.5707, + "step": 661 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998887048601812e-05, + "loss": 0.5999, + "step": 662 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998861611016608e-05, + "loss": 0.49, + "step": 663 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998835886022997e-05, + "loss": 0.6035, + "step": 664 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998809873621722e-05, + "loss": 0.5609, + "step": 665 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998783573813535e-05, + "loss": 0.5078, + "step": 666 + }, + { + "epoch": 0.28, + "learning_rate": 1.999875698659918e-05, + "loss": 0.5402, + "step": 667 + }, + { + "epoch": 0.28, + "learning_rate": 1.999873011197943e-05, + "loss": 0.5853, + "step": 668 + }, + { + "epoch": 0.28, + "learning_rate": 1.999870294995506e-05, + "loss": 0.4822, + "step": 669 + }, + { + "epoch": 0.28, + "learning_rate": 1.999867550052684e-05, + "loss": 0.6014, + "step": 670 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998647763695568e-05, + "loss": 0.5914, + "step": 671 + }, + { + "epoch": 0.28, + "learning_rate": 1.999861973946204e-05, + "loss": 0.5698, + "step": 672 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998591427827062e-05, + "loss": 0.5685, + "step": 673 + }, + { + "epoch": 0.28, + "learning_rate": 1.999856282879144e-05, + "loss": 0.6114, + "step": 674 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998533942356008e-05, + "loss": 0.5575, + "step": 675 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998504768521588e-05, + "loss": 0.6381, + "step": 676 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998475307289018e-05, + "loss": 0.5428, + "step": 677 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998445558659152e-05, + "loss": 0.5533, + "step": 678 + }, + { + "epoch": 0.28, + "learning_rate": 1.999841552263284e-05, + "loss": 0.4876, + "step": 679 + }, + { + "epoch": 0.28, + "learning_rate": 1.9998385199210945e-05, + "loss": 0.5705, + "step": 680 + }, + { + "epoch": 0.29, + "learning_rate": 1.999835458839434e-05, + "loss": 0.6025, + "step": 681 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998323690183907e-05, + "loss": 0.6213, + "step": 682 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998292504580528e-05, + "loss": 0.5695, + "step": 683 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998261031585104e-05, + "loss": 0.5282, + "step": 684 + }, + { + "epoch": 0.29, + "learning_rate": 1.999822927119854e-05, + "loss": 0.5748, + "step": 685 + }, + { + "epoch": 0.29, + "learning_rate": 1.999819722342175e-05, + "loss": 0.5202, + "step": 686 + }, + { + "epoch": 0.29, + "learning_rate": 1.999816488825565e-05, + "loss": 0.5667, + "step": 687 + }, + { + "epoch": 0.29, + "learning_rate": 1.9998132265701176e-05, + "loss": 0.5365, + "step": 688 + }, + { + "epoch": 0.29, + "learning_rate": 1.999809935575926e-05, + "loss": 0.6048, + "step": 689 + }, + { + "epoch": 0.29, + "learning_rate": 1.999806615843085e-05, + "loss": 0.6033, + "step": 690 + }, + { + "epoch": 0.29, + "learning_rate": 1.99980326737169e-05, + "loss": 0.5729, + "step": 691 + }, + { + "epoch": 0.29, + "learning_rate": 1.999799890161837e-05, + "loss": 0.492, + "step": 692 + }, + { + "epoch": 0.29, + "learning_rate": 1.999796484213624e-05, + "loss": 0.5711, + "step": 693 + }, + { + "epoch": 0.29, + "learning_rate": 1.999793049527148e-05, + "loss": 0.5011, + "step": 694 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997895861025078e-05, + "loss": 0.5375, + "step": 695 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997860939398035e-05, + "loss": 0.5716, + "step": 696 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997825730391345e-05, + "loss": 0.5735, + "step": 697 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997790234006034e-05, + "loss": 0.529, + "step": 698 + }, + { + "epoch": 0.29, + "learning_rate": 1.999775445024311e-05, + "loss": 0.5457, + "step": 699 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997718379103608e-05, + "loss": 0.5196, + "step": 700 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997682020588563e-05, + "loss": 0.5419, + "step": 701 + }, + { + "epoch": 0.29, + "learning_rate": 1.999764537469902e-05, + "loss": 0.4548, + "step": 702 + }, + { + "epoch": 0.29, + "learning_rate": 1.9997608441436033e-05, + "loss": 0.4813, + "step": 703 + }, + { + "epoch": 0.29, + "learning_rate": 1.999757122080066e-05, + "loss": 0.5566, + "step": 704 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997533712793976e-05, + "loss": 0.5593, + "step": 705 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997495917417054e-05, + "loss": 0.5683, + "step": 706 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997457834670988e-05, + "loss": 0.5919, + "step": 707 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997419464556868e-05, + "loss": 0.6129, + "step": 708 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997380807075796e-05, + "loss": 0.4535, + "step": 709 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997341862228886e-05, + "loss": 0.6517, + "step": 710 + }, + { + "epoch": 0.3, + "learning_rate": 1.999730263001725e-05, + "loss": 0.5925, + "step": 711 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997263110442025e-05, + "loss": 0.4815, + "step": 712 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997223303504345e-05, + "loss": 0.5777, + "step": 713 + }, + { + "epoch": 0.3, + "learning_rate": 1.999718320920535e-05, + "loss": 0.4826, + "step": 714 + }, + { + "epoch": 0.3, + "learning_rate": 1.9997142827546196e-05, + "loss": 0.6005, + "step": 715 + }, + { + "epoch": 0.3, + "learning_rate": 1.999710215852804e-05, + "loss": 0.6185, + "step": 716 + }, + { + "epoch": 0.3, + "learning_rate": 1.999706120215206e-05, + "loss": 0.6308, + "step": 717 + }, + { + "epoch": 0.3, + "learning_rate": 1.999701995841942e-05, + "loss": 0.6157, + "step": 718 + }, + { + "epoch": 0.3, + "learning_rate": 1.999697842733132e-05, + "loss": 0.517, + "step": 719 + }, + { + "epoch": 0.3, + "learning_rate": 1.999693660888894e-05, + "loss": 0.5469, + "step": 720 + }, + { + "epoch": 0.3, + "learning_rate": 1.999689450309349e-05, + "loss": 0.5627, + "step": 721 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996852109946178e-05, + "loss": 0.5886, + "step": 722 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996809429448224e-05, + "loss": 0.5238, + "step": 723 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996766461600855e-05, + "loss": 0.6008, + "step": 724 + }, + { + "epoch": 0.3, + "learning_rate": 1.99967232064053e-05, + "loss": 0.4986, + "step": 725 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996679663862815e-05, + "loss": 0.5396, + "step": 726 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996635833974637e-05, + "loss": 0.497, + "step": 727 + }, + { + "epoch": 0.3, + "learning_rate": 1.9996591716742037e-05, + "loss": 0.5982, + "step": 728 + }, + { + "epoch": 0.31, + "learning_rate": 1.999654731216628e-05, + "loss": 0.5701, + "step": 729 + }, + { + "epoch": 0.31, + "learning_rate": 1.999650262024864e-05, + "loss": 0.5235, + "step": 730 + }, + { + "epoch": 0.31, + "learning_rate": 1.99964576409904e-05, + "loss": 0.5917, + "step": 731 + }, + { + "epoch": 0.31, + "learning_rate": 1.999641237439286e-05, + "loss": 0.5691, + "step": 732 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996366820457313e-05, + "loss": 0.6141, + "step": 733 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996320979185074e-05, + "loss": 0.5376, + "step": 734 + }, + { + "epoch": 0.31, + "learning_rate": 1.999627485057746e-05, + "loss": 0.4333, + "step": 735 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996228434635797e-05, + "loss": 0.5657, + "step": 736 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996181731361415e-05, + "loss": 0.5721, + "step": 737 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996134740755662e-05, + "loss": 0.6029, + "step": 738 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996087462819884e-05, + "loss": 0.5711, + "step": 739 + }, + { + "epoch": 0.31, + "learning_rate": 1.9996039897555443e-05, + "loss": 0.5509, + "step": 740 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995992044963705e-05, + "loss": 0.555, + "step": 741 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995943905046048e-05, + "loss": 0.4628, + "step": 742 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995895477803854e-05, + "loss": 0.5218, + "step": 743 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995846763238514e-05, + "loss": 0.4748, + "step": 744 + }, + { + "epoch": 0.31, + "learning_rate": 1.999579776135143e-05, + "loss": 0.6094, + "step": 745 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995748472144005e-05, + "loss": 0.5397, + "step": 746 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995698895617663e-05, + "loss": 0.6416, + "step": 747 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995649031773828e-05, + "loss": 0.5666, + "step": 748 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995598880613928e-05, + "loss": 0.5718, + "step": 749 + }, + { + "epoch": 0.31, + "learning_rate": 1.999554844213941e-05, + "loss": 0.5107, + "step": 750 + }, + { + "epoch": 0.31, + "learning_rate": 1.9995497716351725e-05, + "loss": 0.5182, + "step": 751 + }, + { + "epoch": 0.31, + "learning_rate": 1.999544670325232e-05, + "loss": 0.645, + "step": 752 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995395402842678e-05, + "loss": 0.6027, + "step": 753 + }, + { + "epoch": 0.32, + "learning_rate": 1.999534381512426e-05, + "loss": 0.506, + "step": 754 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995291940098555e-05, + "loss": 0.4984, + "step": 755 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995239777767053e-05, + "loss": 0.4942, + "step": 756 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995187328131252e-05, + "loss": 0.5343, + "step": 757 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995134591192664e-05, + "loss": 0.4752, + "step": 758 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995081566952795e-05, + "loss": 0.4636, + "step": 759 + }, + { + "epoch": 0.32, + "learning_rate": 1.9995028255413184e-05, + "loss": 0.5338, + "step": 760 + }, + { + "epoch": 0.32, + "learning_rate": 1.999497465657535e-05, + "loss": 0.5684, + "step": 761 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994920770440842e-05, + "loss": 0.4523, + "step": 762 + }, + { + "epoch": 0.32, + "learning_rate": 1.99948665970112e-05, + "loss": 0.6039, + "step": 763 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994812136287994e-05, + "loss": 0.4691, + "step": 764 + }, + { + "epoch": 0.32, + "learning_rate": 1.999475738827278e-05, + "loss": 0.5422, + "step": 765 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994702352967132e-05, + "loss": 0.5557, + "step": 766 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994647030372638e-05, + "loss": 0.545, + "step": 767 + }, + { + "epoch": 0.32, + "learning_rate": 1.999459142049088e-05, + "loss": 0.5399, + "step": 768 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994535523323462e-05, + "loss": 0.5359, + "step": 769 + }, + { + "epoch": 0.32, + "learning_rate": 1.999447933887199e-05, + "loss": 0.5282, + "step": 770 + }, + { + "epoch": 0.32, + "learning_rate": 1.999442286713808e-05, + "loss": 0.5645, + "step": 771 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994366108123352e-05, + "loss": 0.5232, + "step": 772 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994309061829438e-05, + "loss": 0.537, + "step": 773 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994251728257978e-05, + "loss": 0.6248, + "step": 774 + }, + { + "epoch": 0.32, + "learning_rate": 1.999419410741062e-05, + "loss": 0.5334, + "step": 775 + }, + { + "epoch": 0.32, + "learning_rate": 1.9994136199289028e-05, + "loss": 0.5189, + "step": 776 + }, + { + "epoch": 0.33, + "learning_rate": 1.9994078003894856e-05, + "loss": 0.64, + "step": 777 + }, + { + "epoch": 0.33, + "learning_rate": 1.999401952122978e-05, + "loss": 0.6003, + "step": 778 + }, + { + "epoch": 0.33, + "learning_rate": 1.999396075129548e-05, + "loss": 0.5584, + "step": 779 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993901694093645e-05, + "loss": 0.4774, + "step": 780 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993842349625978e-05, + "loss": 0.5653, + "step": 781 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993782717894178e-05, + "loss": 0.5828, + "step": 782 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993722798899963e-05, + "loss": 0.6803, + "step": 783 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993662592645057e-05, + "loss": 0.487, + "step": 784 + }, + { + "epoch": 0.33, + "learning_rate": 1.999360209913119e-05, + "loss": 0.5226, + "step": 785 + }, + { + "epoch": 0.33, + "learning_rate": 1.999354131836009e-05, + "loss": 0.5717, + "step": 786 + }, + { + "epoch": 0.33, + "learning_rate": 1.999348025033352e-05, + "loss": 0.5765, + "step": 787 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993418895053227e-05, + "loss": 0.4651, + "step": 788 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993357252520975e-05, + "loss": 0.5336, + "step": 789 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993295322738537e-05, + "loss": 0.5058, + "step": 790 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993233105707693e-05, + "loss": 0.5516, + "step": 791 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993170601430233e-05, + "loss": 0.57, + "step": 792 + }, + { + "epoch": 0.33, + "learning_rate": 1.9993107809907946e-05, + "loss": 0.481, + "step": 793 + }, + { + "epoch": 0.33, + "learning_rate": 1.999304473114265e-05, + "loss": 0.6331, + "step": 794 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992981365136145e-05, + "loss": 0.5095, + "step": 795 + }, + { + "epoch": 0.33, + "learning_rate": 1.999291771189026e-05, + "loss": 0.5001, + "step": 796 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992853771406825e-05, + "loss": 0.5308, + "step": 797 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992789543687673e-05, + "loss": 0.5813, + "step": 798 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992725028734655e-05, + "loss": 0.6555, + "step": 799 + }, + { + "epoch": 0.33, + "learning_rate": 1.9992660226549624e-05, + "loss": 0.4987, + "step": 800 + }, + { + "epoch": 0.34, + "learning_rate": 1.999259513713444e-05, + "loss": 0.5393, + "step": 801 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992529760490976e-05, + "loss": 0.554, + "step": 802 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992464096621113e-05, + "loss": 0.484, + "step": 803 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992398145526738e-05, + "loss": 0.5911, + "step": 804 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992331907209742e-05, + "loss": 0.5786, + "step": 805 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992265381672036e-05, + "loss": 0.5447, + "step": 806 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992198568915526e-05, + "loss": 0.5934, + "step": 807 + }, + { + "epoch": 0.34, + "learning_rate": 1.9992131468942135e-05, + "loss": 0.5529, + "step": 808 + }, + { + "epoch": 0.34, + "learning_rate": 1.999206408175379e-05, + "loss": 0.539, + "step": 809 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991996407352432e-05, + "loss": 0.5555, + "step": 810 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991928445740007e-05, + "loss": 0.5249, + "step": 811 + }, + { + "epoch": 0.34, + "learning_rate": 1.999186019691846e-05, + "loss": 0.5433, + "step": 812 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991791660889762e-05, + "loss": 0.6182, + "step": 813 + }, + { + "epoch": 0.34, + "learning_rate": 1.999172283765588e-05, + "loss": 0.5141, + "step": 814 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991653727218787e-05, + "loss": 0.5161, + "step": 815 + }, + { + "epoch": 0.34, + "learning_rate": 1.999158432958048e-05, + "loss": 0.5049, + "step": 816 + }, + { + "epoch": 0.34, + "learning_rate": 1.999151464474294e-05, + "loss": 0.4883, + "step": 817 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991444672708183e-05, + "loss": 0.5744, + "step": 818 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991374413478214e-05, + "loss": 0.5944, + "step": 819 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991303867055056e-05, + "loss": 0.5067, + "step": 820 + }, + { + "epoch": 0.34, + "learning_rate": 1.999123303344073e-05, + "loss": 0.5539, + "step": 821 + }, + { + "epoch": 0.34, + "learning_rate": 1.999116191263728e-05, + "loss": 0.5636, + "step": 822 + }, + { + "epoch": 0.34, + "learning_rate": 1.9991090504646748e-05, + "loss": 0.63, + "step": 823 + }, + { + "epoch": 0.34, + "learning_rate": 1.999101880947118e-05, + "loss": 0.5853, + "step": 824 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990946827112648e-05, + "loss": 0.5308, + "step": 825 + }, + { + "epoch": 0.35, + "learning_rate": 1.999087455757321e-05, + "loss": 0.5656, + "step": 826 + }, + { + "epoch": 0.35, + "learning_rate": 1.999080200085495e-05, + "loss": 0.5307, + "step": 827 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990729156959956e-05, + "loss": 0.4947, + "step": 828 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990656025890315e-05, + "loss": 0.5177, + "step": 829 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990582607648133e-05, + "loss": 0.5555, + "step": 830 + }, + { + "epoch": 0.35, + "learning_rate": 1.999050890223552e-05, + "loss": 0.4916, + "step": 831 + }, + { + "epoch": 0.35, + "learning_rate": 1.999043490965459e-05, + "loss": 0.5135, + "step": 832 + }, + { + "epoch": 0.35, + "learning_rate": 1.999036062990748e-05, + "loss": 0.63, + "step": 833 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990286062996316e-05, + "loss": 0.5116, + "step": 834 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990211208923242e-05, + "loss": 0.5488, + "step": 835 + }, + { + "epoch": 0.35, + "learning_rate": 1.9990136067690416e-05, + "loss": 0.6538, + "step": 836 + }, + { + "epoch": 0.35, + "learning_rate": 1.999006063929999e-05, + "loss": 0.5465, + "step": 837 + }, + { + "epoch": 0.35, + "learning_rate": 1.998998492375414e-05, + "loss": 0.5929, + "step": 838 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989908921055036e-05, + "loss": 0.5986, + "step": 839 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989832631204863e-05, + "loss": 0.613, + "step": 840 + }, + { + "epoch": 0.35, + "learning_rate": 1.998975605420582e-05, + "loss": 0.5219, + "step": 841 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989679190060104e-05, + "loss": 0.5513, + "step": 842 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989602038769922e-05, + "loss": 0.4865, + "step": 843 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989524600337493e-05, + "loss": 0.6124, + "step": 844 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989446874765048e-05, + "loss": 0.5157, + "step": 845 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989368862054814e-05, + "loss": 0.5596, + "step": 846 + }, + { + "epoch": 0.35, + "learning_rate": 1.9989290562209036e-05, + "loss": 0.5604, + "step": 847 + }, + { + "epoch": 0.36, + "learning_rate": 1.9989211975229966e-05, + "loss": 0.5181, + "step": 848 + }, + { + "epoch": 0.36, + "learning_rate": 1.998913310111986e-05, + "loss": 0.5237, + "step": 849 + }, + { + "epoch": 0.36, + "learning_rate": 1.998905393988099e-05, + "loss": 0.5698, + "step": 850 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988974491515628e-05, + "loss": 0.5563, + "step": 851 + }, + { + "epoch": 0.36, + "learning_rate": 1.998889475602606e-05, + "loss": 0.5638, + "step": 852 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988814733414573e-05, + "loss": 0.5097, + "step": 853 + }, + { + "epoch": 0.36, + "learning_rate": 1.998873442368347e-05, + "loss": 0.4871, + "step": 854 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988653826835062e-05, + "loss": 0.4984, + "step": 855 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988572942871666e-05, + "loss": 0.5244, + "step": 856 + }, + { + "epoch": 0.36, + "learning_rate": 1.99884917717956e-05, + "loss": 0.6335, + "step": 857 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988410313609203e-05, + "loss": 0.5901, + "step": 858 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988328568314818e-05, + "loss": 0.54, + "step": 859 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988246535914787e-05, + "loss": 0.5668, + "step": 860 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988164216411474e-05, + "loss": 0.5125, + "step": 861 + }, + { + "epoch": 0.36, + "learning_rate": 1.9988081609807243e-05, + "loss": 0.5225, + "step": 862 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987998716104473e-05, + "loss": 0.5736, + "step": 863 + }, + { + "epoch": 0.36, + "learning_rate": 1.998791553530554e-05, + "loss": 0.4624, + "step": 864 + }, + { + "epoch": 0.36, + "learning_rate": 1.998783206741284e-05, + "loss": 0.5752, + "step": 865 + }, + { + "epoch": 0.36, + "learning_rate": 1.998774831242877e-05, + "loss": 0.5861, + "step": 866 + }, + { + "epoch": 0.36, + "learning_rate": 1.998766427035574e-05, + "loss": 0.5887, + "step": 867 + }, + { + "epoch": 0.36, + "learning_rate": 1.998757994119616e-05, + "loss": 0.5148, + "step": 868 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987495324952458e-05, + "loss": 0.5131, + "step": 869 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987410421627063e-05, + "loss": 0.5495, + "step": 870 + }, + { + "epoch": 0.36, + "learning_rate": 1.9987325231222423e-05, + "loss": 0.5244, + "step": 871 + }, + { + "epoch": 0.37, + "learning_rate": 1.998723975374098e-05, + "loss": 0.5098, + "step": 872 + }, + { + "epoch": 0.37, + "learning_rate": 1.998715398918519e-05, + "loss": 0.5323, + "step": 873 + }, + { + "epoch": 0.37, + "learning_rate": 1.9987067937557527e-05, + "loss": 0.6008, + "step": 874 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986981598860454e-05, + "loss": 0.613, + "step": 875 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986894973096458e-05, + "loss": 0.5766, + "step": 876 + }, + { + "epoch": 0.37, + "learning_rate": 1.998680806026803e-05, + "loss": 0.5188, + "step": 877 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986720860377664e-05, + "loss": 0.5137, + "step": 878 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986633373427868e-05, + "loss": 0.6744, + "step": 879 + }, + { + "epoch": 0.37, + "learning_rate": 1.998654559942116e-05, + "loss": 0.5557, + "step": 880 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986457538360062e-05, + "loss": 0.5192, + "step": 881 + }, + { + "epoch": 0.37, + "learning_rate": 1.99863691902471e-05, + "loss": 0.5091, + "step": 882 + }, + { + "epoch": 0.37, + "learning_rate": 1.998628055508482e-05, + "loss": 0.568, + "step": 883 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986191632875768e-05, + "loss": 0.5641, + "step": 884 + }, + { + "epoch": 0.37, + "learning_rate": 1.99861024236225e-05, + "loss": 0.5126, + "step": 885 + }, + { + "epoch": 0.37, + "learning_rate": 1.9986012927327575e-05, + "loss": 0.5267, + "step": 886 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985923143993572e-05, + "loss": 0.4799, + "step": 887 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985833073623068e-05, + "loss": 0.5367, + "step": 888 + }, + { + "epoch": 0.37, + "learning_rate": 1.998574271621866e-05, + "loss": 0.5242, + "step": 889 + }, + { + "epoch": 0.37, + "learning_rate": 1.998565207178293e-05, + "loss": 0.5858, + "step": 890 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985561140318496e-05, + "loss": 0.5344, + "step": 891 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985469921827968e-05, + "loss": 0.5244, + "step": 892 + }, + { + "epoch": 0.37, + "learning_rate": 1.998537841631397e-05, + "loss": 0.5289, + "step": 893 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985286623779126e-05, + "loss": 0.5555, + "step": 894 + }, + { + "epoch": 0.37, + "learning_rate": 1.9985194544226082e-05, + "loss": 0.5863, + "step": 895 + }, + { + "epoch": 0.38, + "learning_rate": 1.998510217765748e-05, + "loss": 0.4871, + "step": 896 + }, + { + "epoch": 0.38, + "learning_rate": 1.998500952407598e-05, + "loss": 0.5624, + "step": 897 + }, + { + "epoch": 0.38, + "learning_rate": 1.998491658348424e-05, + "loss": 0.5368, + "step": 898 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984823355884926e-05, + "loss": 0.4979, + "step": 899 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984729841280734e-05, + "loss": 0.6757, + "step": 900 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984636039674342e-05, + "loss": 0.5738, + "step": 901 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984541951068445e-05, + "loss": 0.5586, + "step": 902 + }, + { + "epoch": 0.38, + "learning_rate": 1.998444757546575e-05, + "loss": 0.5576, + "step": 903 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984352912868973e-05, + "loss": 0.5864, + "step": 904 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984257963280823e-05, + "loss": 0.5936, + "step": 905 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984162726704047e-05, + "loss": 0.6112, + "step": 906 + }, + { + "epoch": 0.38, + "learning_rate": 1.9984067203141368e-05, + "loss": 0.5352, + "step": 907 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983971392595537e-05, + "loss": 0.6031, + "step": 908 + }, + { + "epoch": 0.38, + "learning_rate": 1.998387529506931e-05, + "loss": 0.5561, + "step": 909 + }, + { + "epoch": 0.38, + "learning_rate": 1.998377891056545e-05, + "loss": 0.5093, + "step": 910 + }, + { + "epoch": 0.38, + "learning_rate": 1.998368223908672e-05, + "loss": 0.6654, + "step": 911 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983585280635902e-05, + "loss": 0.4926, + "step": 912 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983488035215788e-05, + "loss": 0.5659, + "step": 913 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983390502829168e-05, + "loss": 0.5259, + "step": 914 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983292683478844e-05, + "loss": 0.4566, + "step": 915 + }, + { + "epoch": 0.38, + "learning_rate": 1.9983194577167636e-05, + "loss": 0.524, + "step": 916 + }, + { + "epoch": 0.38, + "learning_rate": 1.998309618389836e-05, + "loss": 0.4984, + "step": 917 + }, + { + "epoch": 0.38, + "learning_rate": 1.9982997503673837e-05, + "loss": 0.5231, + "step": 918 + }, + { + "epoch": 0.38, + "learning_rate": 1.9982898536496912e-05, + "loss": 0.5507, + "step": 919 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982799282370432e-05, + "loss": 0.5801, + "step": 920 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982699741297236e-05, + "loss": 0.5397, + "step": 921 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982599913280202e-05, + "loss": 0.5434, + "step": 922 + }, + { + "epoch": 0.39, + "learning_rate": 1.998249979832219e-05, + "loss": 0.507, + "step": 923 + }, + { + "epoch": 0.39, + "learning_rate": 1.998239939642608e-05, + "loss": 0.5271, + "step": 924 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982298707594757e-05, + "loss": 0.5165, + "step": 925 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982197731831115e-05, + "loss": 0.6061, + "step": 926 + }, + { + "epoch": 0.39, + "learning_rate": 1.9982096469138056e-05, + "loss": 0.6236, + "step": 927 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981994919518496e-05, + "loss": 0.5399, + "step": 928 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981893082975348e-05, + "loss": 0.5696, + "step": 929 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981790959511538e-05, + "loss": 0.5813, + "step": 930 + }, + { + "epoch": 0.39, + "learning_rate": 1.998168854913001e-05, + "loss": 0.5144, + "step": 931 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981585851833697e-05, + "loss": 0.4815, + "step": 932 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981482867625557e-05, + "loss": 0.5817, + "step": 933 + }, + { + "epoch": 0.39, + "learning_rate": 1.998137959650855e-05, + "loss": 0.5457, + "step": 934 + }, + { + "epoch": 0.39, + "learning_rate": 1.998127603848565e-05, + "loss": 0.4847, + "step": 935 + }, + { + "epoch": 0.39, + "learning_rate": 1.998117219355982e-05, + "loss": 0.4814, + "step": 936 + }, + { + "epoch": 0.39, + "learning_rate": 1.9981068061734055e-05, + "loss": 0.5047, + "step": 937 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980963643011346e-05, + "loss": 0.5589, + "step": 938 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980858937394695e-05, + "loss": 0.5293, + "step": 939 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980753944887108e-05, + "loss": 0.5762, + "step": 940 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980648665491606e-05, + "loss": 0.557, + "step": 941 + }, + { + "epoch": 0.39, + "learning_rate": 1.9980543099211215e-05, + "loss": 0.577, + "step": 942 + }, + { + "epoch": 0.39, + "learning_rate": 1.998043724604897e-05, + "loss": 0.5131, + "step": 943 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980331106007912e-05, + "loss": 0.5274, + "step": 944 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980224679091092e-05, + "loss": 0.5165, + "step": 945 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980117965301574e-05, + "loss": 0.5571, + "step": 946 + }, + { + "epoch": 0.4, + "learning_rate": 1.9980010964642417e-05, + "loss": 0.5509, + "step": 947 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979903677116705e-05, + "loss": 0.4886, + "step": 948 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979796102727513e-05, + "loss": 0.5009, + "step": 949 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979688241477938e-05, + "loss": 0.5245, + "step": 950 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979580093371084e-05, + "loss": 0.5203, + "step": 951 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979471658410055e-05, + "loss": 0.5339, + "step": 952 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979362936597967e-05, + "loss": 0.6407, + "step": 953 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979253927937948e-05, + "loss": 0.5562, + "step": 954 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979144632433132e-05, + "loss": 0.5537, + "step": 955 + }, + { + "epoch": 0.4, + "learning_rate": 1.9979035050086656e-05, + "loss": 0.537, + "step": 956 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978925180901675e-05, + "loss": 0.5218, + "step": 957 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978815024881342e-05, + "loss": 0.619, + "step": 958 + }, + { + "epoch": 0.4, + "learning_rate": 1.997870458202883e-05, + "loss": 0.5215, + "step": 959 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978593852347306e-05, + "loss": 0.5597, + "step": 960 + }, + { + "epoch": 0.4, + "learning_rate": 1.997848283583996e-05, + "loss": 0.4835, + "step": 961 + }, + { + "epoch": 0.4, + "learning_rate": 1.997837153250998e-05, + "loss": 0.5569, + "step": 962 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978259942360562e-05, + "loss": 0.5173, + "step": 963 + }, + { + "epoch": 0.4, + "learning_rate": 1.9978148065394917e-05, + "loss": 0.5005, + "step": 964 + }, + { + "epoch": 0.4, + "learning_rate": 1.997803590161626e-05, + "loss": 0.6021, + "step": 965 + }, + { + "epoch": 0.4, + "learning_rate": 1.9977923451027816e-05, + "loss": 0.5151, + "step": 966 + }, + { + "epoch": 0.4, + "learning_rate": 1.9977810713632815e-05, + "loss": 0.5957, + "step": 967 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977697689434504e-05, + "loss": 0.5009, + "step": 968 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977584378436123e-05, + "loss": 0.5259, + "step": 969 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977470780640934e-05, + "loss": 0.4887, + "step": 970 + }, + { + "epoch": 0.41, + "learning_rate": 1.99773568960522e-05, + "loss": 0.5379, + "step": 971 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977242724673198e-05, + "loss": 0.6189, + "step": 972 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977128266507203e-05, + "loss": 0.5615, + "step": 973 + }, + { + "epoch": 0.41, + "learning_rate": 1.9977013521557516e-05, + "loss": 0.5451, + "step": 974 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976898489827423e-05, + "loss": 0.5693, + "step": 975 + }, + { + "epoch": 0.41, + "learning_rate": 1.997678317132024e-05, + "loss": 0.5098, + "step": 976 + }, + { + "epoch": 0.41, + "learning_rate": 1.997666756603928e-05, + "loss": 0.5261, + "step": 977 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976551673987857e-05, + "loss": 0.525, + "step": 978 + }, + { + "epoch": 0.41, + "learning_rate": 1.997643549516931e-05, + "loss": 0.6046, + "step": 979 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976319029586982e-05, + "loss": 0.572, + "step": 980 + }, + { + "epoch": 0.41, + "learning_rate": 1.9976202277244216e-05, + "loss": 0.5908, + "step": 981 + }, + { + "epoch": 0.41, + "learning_rate": 1.997608523814437e-05, + "loss": 0.5843, + "step": 982 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975967912290805e-05, + "loss": 0.5956, + "step": 983 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975850299686894e-05, + "loss": 0.5536, + "step": 984 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975732400336018e-05, + "loss": 0.5441, + "step": 985 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975614214241562e-05, + "loss": 0.5195, + "step": 986 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975495741406934e-05, + "loss": 0.5481, + "step": 987 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975376981835533e-05, + "loss": 0.5622, + "step": 988 + }, + { + "epoch": 0.41, + "learning_rate": 1.997525793553077e-05, + "loss": 0.4714, + "step": 989 + }, + { + "epoch": 0.41, + "learning_rate": 1.997513860249607e-05, + "loss": 0.5424, + "step": 990 + }, + { + "epoch": 0.41, + "learning_rate": 1.9975018982734863e-05, + "loss": 0.5466, + "step": 991 + }, + { + "epoch": 0.42, + "learning_rate": 1.997489907625059e-05, + "loss": 0.5853, + "step": 992 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974778883046688e-05, + "loss": 0.4975, + "step": 993 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974658403126624e-05, + "loss": 0.4859, + "step": 994 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974537636493853e-05, + "loss": 0.4968, + "step": 995 + }, + { + "epoch": 0.42, + "learning_rate": 1.997441658315185e-05, + "loss": 0.5131, + "step": 996 + }, + { + "epoch": 0.42, + "learning_rate": 1.9974295243104093e-05, + "loss": 0.5391, + "step": 997 + }, + { + "epoch": 0.42, + "learning_rate": 1.997417361635407e-05, + "loss": 0.4989, + "step": 998 + }, + { + "epoch": 0.42, + "learning_rate": 1.997405170290528e-05, + "loss": 0.5016, + "step": 999 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973929502761222e-05, + "loss": 0.5196, + "step": 1000 + }, + { + "epoch": 0.42, + "learning_rate": 1.997380701592541e-05, + "loss": 0.5563, + "step": 1001 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973684242401366e-05, + "loss": 0.5229, + "step": 1002 + }, + { + "epoch": 0.42, + "learning_rate": 1.997356118219262e-05, + "loss": 0.4715, + "step": 1003 + }, + { + "epoch": 0.42, + "learning_rate": 1.997343783530271e-05, + "loss": 0.5847, + "step": 1004 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973314201735176e-05, + "loss": 0.6229, + "step": 1005 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973190281493578e-05, + "loss": 0.4641, + "step": 1006 + }, + { + "epoch": 0.42, + "learning_rate": 1.9973066074581475e-05, + "loss": 0.4784, + "step": 1007 + }, + { + "epoch": 0.42, + "learning_rate": 1.997294158100244e-05, + "loss": 0.4818, + "step": 1008 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972816800760046e-05, + "loss": 0.4652, + "step": 1009 + }, + { + "epoch": 0.42, + "learning_rate": 1.997269173385788e-05, + "loss": 0.5223, + "step": 1010 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972566380299544e-05, + "loss": 0.5002, + "step": 1011 + }, + { + "epoch": 0.42, + "learning_rate": 1.997244074008864e-05, + "loss": 0.487, + "step": 1012 + }, + { + "epoch": 0.42, + "learning_rate": 1.997231481322877e-05, + "loss": 0.5619, + "step": 1013 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972188599723564e-05, + "loss": 0.5361, + "step": 1014 + }, + { + "epoch": 0.42, + "learning_rate": 1.9972062099576642e-05, + "loss": 0.524, + "step": 1015 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971935312791646e-05, + "loss": 0.4828, + "step": 1016 + }, + { + "epoch": 0.43, + "learning_rate": 1.997180823937222e-05, + "loss": 0.5571, + "step": 1017 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971680879322012e-05, + "loss": 0.528, + "step": 1018 + }, + { + "epoch": 0.43, + "learning_rate": 1.997155323264469e-05, + "loss": 0.5309, + "step": 1019 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971425299343915e-05, + "loss": 0.4413, + "step": 1020 + }, + { + "epoch": 0.43, + "learning_rate": 1.997129707942337e-05, + "loss": 0.4623, + "step": 1021 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971168572886738e-05, + "loss": 0.4915, + "step": 1022 + }, + { + "epoch": 0.43, + "learning_rate": 1.9971039779737717e-05, + "loss": 0.5633, + "step": 1023 + }, + { + "epoch": 0.43, + "learning_rate": 1.997091069998e-05, + "loss": 0.577, + "step": 1024 + }, + { + "epoch": 0.43, + "learning_rate": 1.9970781333617303e-05, + "loss": 0.5143, + "step": 1025 + }, + { + "epoch": 0.43, + "learning_rate": 1.997065168065335e-05, + "loss": 0.5277, + "step": 1026 + }, + { + "epoch": 0.43, + "learning_rate": 1.997052174109186e-05, + "loss": 0.5045, + "step": 1027 + }, + { + "epoch": 0.43, + "learning_rate": 1.997039151493657e-05, + "loss": 0.5114, + "step": 1028 + }, + { + "epoch": 0.43, + "learning_rate": 1.997026100219122e-05, + "loss": 0.4649, + "step": 1029 + }, + { + "epoch": 0.43, + "learning_rate": 1.997013020285957e-05, + "loss": 0.4951, + "step": 1030 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969999116945374e-05, + "loss": 0.5396, + "step": 1031 + }, + { + "epoch": 0.43, + "learning_rate": 1.99698677444524e-05, + "loss": 0.5606, + "step": 1032 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969736085384422e-05, + "loss": 0.5713, + "step": 1033 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969604139745227e-05, + "loss": 0.5012, + "step": 1034 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969471907538615e-05, + "loss": 0.6019, + "step": 1035 + }, + { + "epoch": 0.43, + "learning_rate": 1.9969339388768372e-05, + "loss": 0.5543, + "step": 1036 + }, + { + "epoch": 0.43, + "learning_rate": 1.996920658343832e-05, + "loss": 0.5305, + "step": 1037 + }, + { + "epoch": 0.43, + "learning_rate": 1.996907349155227e-05, + "loss": 0.5374, + "step": 1038 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968940113114048e-05, + "loss": 0.5252, + "step": 1039 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968806448127488e-05, + "loss": 0.5686, + "step": 1040 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968672496596434e-05, + "loss": 0.5571, + "step": 1041 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968538258524736e-05, + "loss": 0.5308, + "step": 1042 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968403733916246e-05, + "loss": 0.5229, + "step": 1043 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968268922774842e-05, + "loss": 0.5767, + "step": 1044 + }, + { + "epoch": 0.44, + "learning_rate": 1.9968133825104396e-05, + "loss": 0.3965, + "step": 1045 + }, + { + "epoch": 0.44, + "learning_rate": 1.996799844090878e-05, + "loss": 0.6016, + "step": 1046 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967862770191902e-05, + "loss": 0.5493, + "step": 1047 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967726812957652e-05, + "loss": 0.5211, + "step": 1048 + }, + { + "epoch": 0.44, + "learning_rate": 1.996759056920994e-05, + "loss": 0.6184, + "step": 1049 + }, + { + "epoch": 0.44, + "learning_rate": 1.996745403895268e-05, + "loss": 0.5019, + "step": 1050 + }, + { + "epoch": 0.44, + "learning_rate": 1.99673172221898e-05, + "loss": 0.549, + "step": 1051 + }, + { + "epoch": 0.44, + "learning_rate": 1.996718011892523e-05, + "loss": 0.484, + "step": 1052 + }, + { + "epoch": 0.44, + "learning_rate": 1.9967042729162914e-05, + "loss": 0.5575, + "step": 1053 + }, + { + "epoch": 0.44, + "learning_rate": 1.99669050529068e-05, + "loss": 0.5138, + "step": 1054 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966767090160843e-05, + "loss": 0.4945, + "step": 1055 + }, + { + "epoch": 0.44, + "learning_rate": 1.996662884092901e-05, + "loss": 0.4617, + "step": 1056 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966490305215277e-05, + "loss": 0.5356, + "step": 1057 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966351483023627e-05, + "loss": 0.4732, + "step": 1058 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966212374358044e-05, + "loss": 0.4211, + "step": 1059 + }, + { + "epoch": 0.44, + "learning_rate": 1.9966072979222527e-05, + "loss": 0.5779, + "step": 1060 + }, + { + "epoch": 0.44, + "learning_rate": 1.996593329762109e-05, + "loss": 0.5561, + "step": 1061 + }, + { + "epoch": 0.44, + "learning_rate": 1.9965793329557743e-05, + "loss": 0.5145, + "step": 1062 + }, + { + "epoch": 0.45, + "learning_rate": 1.996565307503651e-05, + "loss": 0.5379, + "step": 1063 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965512534061426e-05, + "loss": 0.446, + "step": 1064 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965371706636522e-05, + "loss": 0.5324, + "step": 1065 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965230592765856e-05, + "loss": 0.4506, + "step": 1066 + }, + { + "epoch": 0.45, + "learning_rate": 1.9965089192453475e-05, + "loss": 0.5063, + "step": 1067 + }, + { + "epoch": 0.45, + "learning_rate": 1.996494750570345e-05, + "loss": 0.5347, + "step": 1068 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964805532519852e-05, + "loss": 0.5617, + "step": 1069 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964663272906763e-05, + "loss": 0.4943, + "step": 1070 + }, + { + "epoch": 0.45, + "learning_rate": 1.996452072686827e-05, + "loss": 0.5119, + "step": 1071 + }, + { + "epoch": 0.45, + "learning_rate": 1.996437789440847e-05, + "loss": 0.5418, + "step": 1072 + }, + { + "epoch": 0.45, + "learning_rate": 1.996423477553147e-05, + "loss": 0.5345, + "step": 1073 + }, + { + "epoch": 0.45, + "learning_rate": 1.9964091370241384e-05, + "loss": 0.4945, + "step": 1074 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963947678542335e-05, + "loss": 0.5395, + "step": 1075 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963803700438448e-05, + "loss": 0.5527, + "step": 1076 + }, + { + "epoch": 0.45, + "learning_rate": 1.996365943593387e-05, + "loss": 0.4826, + "step": 1077 + }, + { + "epoch": 0.45, + "learning_rate": 1.996351488503274e-05, + "loss": 0.5785, + "step": 1078 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963370047739218e-05, + "loss": 0.5382, + "step": 1079 + }, + { + "epoch": 0.45, + "learning_rate": 1.9963224924057463e-05, + "loss": 0.5902, + "step": 1080 + }, + { + "epoch": 0.45, + "learning_rate": 1.996307951399165e-05, + "loss": 0.5723, + "step": 1081 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962933817545957e-05, + "loss": 0.5868, + "step": 1082 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962787834724572e-05, + "loss": 0.6129, + "step": 1083 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962641565531694e-05, + "loss": 0.501, + "step": 1084 + }, + { + "epoch": 0.45, + "learning_rate": 1.996249500997152e-05, + "loss": 0.5659, + "step": 1085 + }, + { + "epoch": 0.45, + "learning_rate": 1.9962348168048272e-05, + "loss": 0.556, + "step": 1086 + }, + { + "epoch": 0.46, + "learning_rate": 1.9962201039766163e-05, + "loss": 0.556, + "step": 1087 + }, + { + "epoch": 0.46, + "learning_rate": 1.996205362512943e-05, + "loss": 0.5758, + "step": 1088 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961905924142303e-05, + "loss": 0.4728, + "step": 1089 + }, + { + "epoch": 0.46, + "learning_rate": 1.996175793680903e-05, + "loss": 0.5484, + "step": 1090 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961609663133865e-05, + "loss": 0.5067, + "step": 1091 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961461103121073e-05, + "loss": 0.5013, + "step": 1092 + }, + { + "epoch": 0.46, + "learning_rate": 1.9961312256774915e-05, + "loss": 0.5285, + "step": 1093 + }, + { + "epoch": 0.46, + "learning_rate": 1.996116312409968e-05, + "loss": 0.5439, + "step": 1094 + }, + { + "epoch": 0.46, + "learning_rate": 1.996101370509965e-05, + "loss": 0.5521, + "step": 1095 + }, + { + "epoch": 0.46, + "learning_rate": 1.996086399977912e-05, + "loss": 0.5894, + "step": 1096 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960714008142394e-05, + "loss": 0.4348, + "step": 1097 + }, + { + "epoch": 0.46, + "learning_rate": 1.996056373019378e-05, + "loss": 0.5884, + "step": 1098 + }, + { + "epoch": 0.46, + "learning_rate": 1.99604131659376e-05, + "loss": 0.4819, + "step": 1099 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960262315378188e-05, + "loss": 0.4674, + "step": 1100 + }, + { + "epoch": 0.46, + "learning_rate": 1.9960111178519873e-05, + "loss": 0.5385, + "step": 1101 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959959755366997e-05, + "loss": 0.5105, + "step": 1102 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959808045923917e-05, + "loss": 0.4483, + "step": 1103 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959656050194994e-05, + "loss": 0.5014, + "step": 1104 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959503768184597e-05, + "loss": 0.5339, + "step": 1105 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959351199897102e-05, + "loss": 0.4732, + "step": 1106 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959198345336893e-05, + "loss": 0.5171, + "step": 1107 + }, + { + "epoch": 0.46, + "learning_rate": 1.9959045204508366e-05, + "loss": 0.5483, + "step": 1108 + }, + { + "epoch": 0.46, + "learning_rate": 1.9958891777415917e-05, + "loss": 0.5923, + "step": 1109 + }, + { + "epoch": 0.46, + "learning_rate": 1.9958738064063968e-05, + "loss": 0.5747, + "step": 1110 + }, + { + "epoch": 0.47, + "learning_rate": 1.995858406445693e-05, + "loss": 0.5271, + "step": 1111 + }, + { + "epoch": 0.47, + "learning_rate": 1.9958429778599225e-05, + "loss": 0.5466, + "step": 1112 + }, + { + "epoch": 0.47, + "learning_rate": 1.99582752064953e-05, + "loss": 0.4861, + "step": 1113 + }, + { + "epoch": 0.47, + "learning_rate": 1.9958120348149585e-05, + "loss": 0.5899, + "step": 1114 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957965203566542e-05, + "loss": 0.4815, + "step": 1115 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957809772750623e-05, + "loss": 0.5185, + "step": 1116 + }, + { + "epoch": 0.47, + "learning_rate": 1.99576540557063e-05, + "loss": 0.4802, + "step": 1117 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957498052438046e-05, + "loss": 0.5159, + "step": 1118 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957341762950346e-05, + "loss": 0.6096, + "step": 1119 + }, + { + "epoch": 0.47, + "learning_rate": 1.9957185187247697e-05, + "loss": 0.5007, + "step": 1120 + }, + { + "epoch": 0.47, + "learning_rate": 1.995702832533459e-05, + "loss": 0.5597, + "step": 1121 + }, + { + "epoch": 0.47, + "learning_rate": 1.995687117721554e-05, + "loss": 0.4951, + "step": 1122 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956713742895068e-05, + "loss": 0.4691, + "step": 1123 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956556022377693e-05, + "loss": 0.5219, + "step": 1124 + }, + { + "epoch": 0.47, + "learning_rate": 1.995639801566795e-05, + "loss": 0.545, + "step": 1125 + }, + { + "epoch": 0.47, + "learning_rate": 1.9956239722770382e-05, + "loss": 0.4985, + "step": 1126 + }, + { + "epoch": 0.47, + "learning_rate": 1.995608114368954e-05, + "loss": 0.5149, + "step": 1127 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955922278429975e-05, + "loss": 0.5308, + "step": 1128 + }, + { + "epoch": 0.47, + "learning_rate": 1.995576312699626e-05, + "loss": 0.5482, + "step": 1129 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955603689392974e-05, + "loss": 0.5766, + "step": 1130 + }, + { + "epoch": 0.47, + "learning_rate": 1.995544396562469e-05, + "loss": 0.4821, + "step": 1131 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955283955696004e-05, + "loss": 0.4794, + "step": 1132 + }, + { + "epoch": 0.47, + "learning_rate": 1.9955123659611516e-05, + "loss": 0.5762, + "step": 1133 + }, + { + "epoch": 0.47, + "learning_rate": 1.995496307737583e-05, + "loss": 0.5338, + "step": 1134 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954802208993565e-05, + "loss": 0.574, + "step": 1135 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954641054469347e-05, + "loss": 0.6228, + "step": 1136 + }, + { + "epoch": 0.48, + "learning_rate": 1.99544796138078e-05, + "loss": 0.5616, + "step": 1137 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954317887013576e-05, + "loss": 0.5199, + "step": 1138 + }, + { + "epoch": 0.48, + "learning_rate": 1.9954155874091313e-05, + "loss": 0.5464, + "step": 1139 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953993575045674e-05, + "loss": 0.5484, + "step": 1140 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953830989881324e-05, + "loss": 0.5235, + "step": 1141 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953668118602937e-05, + "loss": 0.5152, + "step": 1142 + }, + { + "epoch": 0.48, + "learning_rate": 1.995350496121519e-05, + "loss": 0.5461, + "step": 1143 + }, + { + "epoch": 0.48, + "learning_rate": 1.9953341517722773e-05, + "loss": 0.4547, + "step": 1144 + }, + { + "epoch": 0.48, + "learning_rate": 1.995317778813039e-05, + "loss": 0.5252, + "step": 1145 + }, + { + "epoch": 0.48, + "learning_rate": 1.995301377244274e-05, + "loss": 0.5046, + "step": 1146 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952849470664547e-05, + "loss": 0.4885, + "step": 1147 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952684882800525e-05, + "loss": 0.5105, + "step": 1148 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952520008855405e-05, + "loss": 0.55, + "step": 1149 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952354848833932e-05, + "loss": 0.5413, + "step": 1150 + }, + { + "epoch": 0.48, + "learning_rate": 1.995218940274085e-05, + "loss": 0.552, + "step": 1151 + }, + { + "epoch": 0.48, + "learning_rate": 1.9952023670580915e-05, + "loss": 0.6011, + "step": 1152 + }, + { + "epoch": 0.48, + "learning_rate": 1.995185765235889e-05, + "loss": 0.5404, + "step": 1153 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951691348079553e-05, + "loss": 0.4995, + "step": 1154 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951524757747675e-05, + "loss": 0.5654, + "step": 1155 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951357881368047e-05, + "loss": 0.5018, + "step": 1156 + }, + { + "epoch": 0.48, + "learning_rate": 1.995119071894547e-05, + "loss": 0.5189, + "step": 1157 + }, + { + "epoch": 0.48, + "learning_rate": 1.9951023270484748e-05, + "loss": 0.5147, + "step": 1158 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950855535990688e-05, + "loss": 0.577, + "step": 1159 + }, + { + "epoch": 0.49, + "learning_rate": 1.995068751546812e-05, + "loss": 0.4954, + "step": 1160 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950519208921866e-05, + "loss": 0.5634, + "step": 1161 + }, + { + "epoch": 0.49, + "learning_rate": 1.995035061635677e-05, + "loss": 0.4313, + "step": 1162 + }, + { + "epoch": 0.49, + "learning_rate": 1.9950181737777673e-05, + "loss": 0.4947, + "step": 1163 + }, + { + "epoch": 0.49, + "learning_rate": 1.995001257318943e-05, + "loss": 0.467, + "step": 1164 + }, + { + "epoch": 0.49, + "learning_rate": 1.994984312259691e-05, + "loss": 0.4954, + "step": 1165 + }, + { + "epoch": 0.49, + "learning_rate": 1.9949673386004975e-05, + "loss": 0.5324, + "step": 1166 + }, + { + "epoch": 0.49, + "learning_rate": 1.994950336341851e-05, + "loss": 0.4987, + "step": 1167 + }, + { + "epoch": 0.49, + "learning_rate": 1.9949333054842398e-05, + "loss": 0.5435, + "step": 1168 + }, + { + "epoch": 0.49, + "learning_rate": 1.994916246028154e-05, + "loss": 0.4749, + "step": 1169 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948991579740832e-05, + "loss": 0.4948, + "step": 1170 + }, + { + "epoch": 0.49, + "learning_rate": 1.994882041322519e-05, + "loss": 0.5437, + "step": 1171 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948648960739536e-05, + "loss": 0.5135, + "step": 1172 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948477222288792e-05, + "loss": 0.5212, + "step": 1173 + }, + { + "epoch": 0.49, + "learning_rate": 1.99483051978779e-05, + "loss": 0.4861, + "step": 1174 + }, + { + "epoch": 0.49, + "learning_rate": 1.9948132887511805e-05, + "loss": 0.5396, + "step": 1175 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947960291195457e-05, + "loss": 0.5566, + "step": 1176 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947787408933816e-05, + "loss": 0.653, + "step": 1177 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947614240731856e-05, + "loss": 0.5085, + "step": 1178 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947440786594552e-05, + "loss": 0.4604, + "step": 1179 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947267046526887e-05, + "loss": 0.538, + "step": 1180 + }, + { + "epoch": 0.49, + "learning_rate": 1.9947093020533863e-05, + "loss": 0.479, + "step": 1181 + }, + { + "epoch": 0.49, + "learning_rate": 1.9946918708620472e-05, + "loss": 0.5215, + "step": 1182 + }, + { + "epoch": 0.5, + "learning_rate": 1.994674411079173e-05, + "loss": 0.49, + "step": 1183 + }, + { + "epoch": 0.5, + "learning_rate": 1.994656922705266e-05, + "loss": 0.5296, + "step": 1184 + }, + { + "epoch": 0.5, + "learning_rate": 1.9946394057408278e-05, + "loss": 0.4532, + "step": 1185 + }, + { + "epoch": 0.5, + "learning_rate": 1.9946218601863626e-05, + "loss": 0.4952, + "step": 1186 + }, + { + "epoch": 0.5, + "learning_rate": 1.994604286042375e-05, + "loss": 0.5172, + "step": 1187 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945866833093694e-05, + "loss": 0.4908, + "step": 1188 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945690519878523e-05, + "loss": 0.6495, + "step": 1189 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945513920783303e-05, + "loss": 0.5633, + "step": 1190 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945337035813114e-05, + "loss": 0.4542, + "step": 1191 + }, + { + "epoch": 0.5, + "learning_rate": 1.9945159864973035e-05, + "loss": 0.5391, + "step": 1192 + }, + { + "epoch": 0.5, + "learning_rate": 1.994498240826816e-05, + "loss": 0.5666, + "step": 1193 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944804665703593e-05, + "loss": 0.5144, + "step": 1194 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944626637284438e-05, + "loss": 0.4959, + "step": 1195 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944448323015818e-05, + "loss": 0.5488, + "step": 1196 + }, + { + "epoch": 0.5, + "learning_rate": 1.9944269722902854e-05, + "loss": 0.4746, + "step": 1197 + }, + { + "epoch": 0.5, + "learning_rate": 1.994409083695068e-05, + "loss": 0.5414, + "step": 1198 + }, + { + "epoch": 0.5, + "learning_rate": 1.994391166516444e-05, + "loss": 0.5357, + "step": 1199 + }, + { + "epoch": 0.5, + "learning_rate": 1.9943732207549287e-05, + "loss": 0.5277, + "step": 1200 + }, + { + "epoch": 0.5, + "learning_rate": 1.994355246411037e-05, + "loss": 0.5926, + "step": 1201 + }, + { + "epoch": 0.5, + "learning_rate": 1.9943372434852864e-05, + "loss": 0.5177, + "step": 1202 + }, + { + "epoch": 0.5, + "learning_rate": 1.994319211978194e-05, + "loss": 0.5034, + "step": 1203 + }, + { + "epoch": 0.5, + "learning_rate": 1.994301151890278e-05, + "loss": 0.5085, + "step": 1204 + }, + { + "epoch": 0.5, + "learning_rate": 1.994283063222058e-05, + "loss": 0.6082, + "step": 1205 + }, + { + "epoch": 0.5, + "learning_rate": 1.9942649459740537e-05, + "loss": 0.4875, + "step": 1206 + }, + { + "epoch": 0.51, + "learning_rate": 1.9942468001467857e-05, + "loss": 0.5117, + "step": 1207 + }, + { + "epoch": 0.51, + "learning_rate": 1.994228625740776e-05, + "loss": 0.453, + "step": 1208 + }, + { + "epoch": 0.51, + "learning_rate": 1.994210422756546e-05, + "loss": 0.509, + "step": 1209 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941921911946202e-05, + "loss": 0.5297, + "step": 1210 + }, + { + "epoch": 0.51, + "learning_rate": 1.994173931055522e-05, + "loss": 0.5244, + "step": 1211 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941556423397762e-05, + "loss": 0.4848, + "step": 1212 + }, + { + "epoch": 0.51, + "learning_rate": 1.994137325047909e-05, + "loss": 0.5528, + "step": 1213 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941189791804466e-05, + "loss": 0.5117, + "step": 1214 + }, + { + "epoch": 0.51, + "learning_rate": 1.9941006047379158e-05, + "loss": 0.4961, + "step": 1215 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940822017208458e-05, + "loss": 0.5069, + "step": 1216 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940637701297647e-05, + "loss": 0.4534, + "step": 1217 + }, + { + "epoch": 0.51, + "learning_rate": 1.994045309965203e-05, + "loss": 0.5828, + "step": 1218 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940268212276904e-05, + "loss": 0.4312, + "step": 1219 + }, + { + "epoch": 0.51, + "learning_rate": 1.9940083039177594e-05, + "loss": 0.4555, + "step": 1220 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939897580359417e-05, + "loss": 0.5016, + "step": 1221 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939711835827703e-05, + "loss": 0.5193, + "step": 1222 + }, + { + "epoch": 0.51, + "learning_rate": 1.993952580558779e-05, + "loss": 0.4985, + "step": 1223 + }, + { + "epoch": 0.51, + "learning_rate": 1.9939339489645034e-05, + "loss": 0.4723, + "step": 1224 + }, + { + "epoch": 0.51, + "learning_rate": 1.993915288800478e-05, + "loss": 0.4698, + "step": 1225 + }, + { + "epoch": 0.51, + "learning_rate": 1.99389660006724e-05, + "loss": 0.4951, + "step": 1226 + }, + { + "epoch": 0.51, + "learning_rate": 1.9938778827653256e-05, + "loss": 0.4973, + "step": 1227 + }, + { + "epoch": 0.51, + "learning_rate": 1.993859136895274e-05, + "loss": 0.5482, + "step": 1228 + }, + { + "epoch": 0.51, + "learning_rate": 1.993840362457623e-05, + "loss": 0.5661, + "step": 1229 + }, + { + "epoch": 0.51, + "learning_rate": 1.993821559452913e-05, + "loss": 0.5084, + "step": 1230 + }, + { + "epoch": 0.52, + "learning_rate": 1.993802727881684e-05, + "loss": 0.6094, + "step": 1231 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937838677444776e-05, + "loss": 0.4842, + "step": 1232 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937649790418357e-05, + "loss": 0.4732, + "step": 1233 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937460617743013e-05, + "loss": 0.5847, + "step": 1234 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937271159424185e-05, + "loss": 0.5418, + "step": 1235 + }, + { + "epoch": 0.52, + "learning_rate": 1.9937081415467313e-05, + "loss": 0.4498, + "step": 1236 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936891385877854e-05, + "loss": 0.5086, + "step": 1237 + }, + { + "epoch": 0.52, + "learning_rate": 1.993670107066127e-05, + "loss": 0.4876, + "step": 1238 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936510469823028e-05, + "loss": 0.5137, + "step": 1239 + }, + { + "epoch": 0.52, + "learning_rate": 1.9936319583368616e-05, + "loss": 0.4546, + "step": 1240 + }, + { + "epoch": 0.52, + "learning_rate": 1.993612841130351e-05, + "loss": 0.5644, + "step": 1241 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935936953633214e-05, + "loss": 0.5763, + "step": 1242 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935745210363224e-05, + "loss": 0.5405, + "step": 1243 + }, + { + "epoch": 0.52, + "learning_rate": 1.993555318149906e-05, + "loss": 0.4923, + "step": 1244 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935360867046228e-05, + "loss": 0.5439, + "step": 1245 + }, + { + "epoch": 0.52, + "learning_rate": 1.9935168267010268e-05, + "loss": 0.4501, + "step": 1246 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934975381396716e-05, + "loss": 0.5177, + "step": 1247 + }, + { + "epoch": 0.52, + "learning_rate": 1.993478221021111e-05, + "loss": 0.5299, + "step": 1248 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934588753459006e-05, + "loss": 0.5357, + "step": 1249 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934395011145967e-05, + "loss": 0.4541, + "step": 1250 + }, + { + "epoch": 0.52, + "learning_rate": 1.9934200983277553e-05, + "loss": 0.464, + "step": 1251 + }, + { + "epoch": 0.52, + "learning_rate": 1.993400666985935e-05, + "loss": 0.5593, + "step": 1252 + }, + { + "epoch": 0.52, + "learning_rate": 1.993381207089694e-05, + "loss": 0.516, + "step": 1253 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933617186395917e-05, + "loss": 0.6846, + "step": 1254 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933422016361886e-05, + "loss": 0.5269, + "step": 1255 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933226560800452e-05, + "loss": 0.5366, + "step": 1256 + }, + { + "epoch": 0.53, + "learning_rate": 1.9933030819717235e-05, + "loss": 0.4933, + "step": 1257 + }, + { + "epoch": 0.53, + "learning_rate": 1.993283479311786e-05, + "loss": 0.5686, + "step": 1258 + }, + { + "epoch": 0.53, + "learning_rate": 1.993263848100796e-05, + "loss": 0.4922, + "step": 1259 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932441883393186e-05, + "loss": 0.546, + "step": 1260 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932245000279184e-05, + "loss": 0.5403, + "step": 1261 + }, + { + "epoch": 0.53, + "learning_rate": 1.9932047831671615e-05, + "loss": 0.4769, + "step": 1262 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931850377576142e-05, + "loss": 0.5271, + "step": 1263 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931652637998443e-05, + "loss": 0.5356, + "step": 1264 + }, + { + "epoch": 0.53, + "learning_rate": 1.9931454612944203e-05, + "loss": 0.5345, + "step": 1265 + }, + { + "epoch": 0.53, + "learning_rate": 1.993125630241911e-05, + "loss": 0.5233, + "step": 1266 + }, + { + "epoch": 0.53, + "learning_rate": 1.993105770642887e-05, + "loss": 0.4715, + "step": 1267 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930858824979192e-05, + "loss": 0.5269, + "step": 1268 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930659658075784e-05, + "loss": 0.5421, + "step": 1269 + }, + { + "epoch": 0.53, + "learning_rate": 1.993046020572438e-05, + "loss": 0.5315, + "step": 1270 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930260467930706e-05, + "loss": 0.6047, + "step": 1271 + }, + { + "epoch": 0.53, + "learning_rate": 1.9930060444700508e-05, + "loss": 0.4419, + "step": 1272 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929860136039537e-05, + "loss": 0.6234, + "step": 1273 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929659541953548e-05, + "loss": 0.4614, + "step": 1274 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929458662448305e-05, + "loss": 0.5827, + "step": 1275 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929257497529584e-05, + "loss": 0.4641, + "step": 1276 + }, + { + "epoch": 0.53, + "learning_rate": 1.9929056047203166e-05, + "loss": 0.4859, + "step": 1277 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928854311474844e-05, + "loss": 0.4513, + "step": 1278 + }, + { + "epoch": 0.54, + "learning_rate": 1.992865229035042e-05, + "loss": 0.6051, + "step": 1279 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928449983835688e-05, + "loss": 0.5138, + "step": 1280 + }, + { + "epoch": 0.54, + "learning_rate": 1.992824739193648e-05, + "loss": 0.5685, + "step": 1281 + }, + { + "epoch": 0.54, + "learning_rate": 1.9928044514658604e-05, + "loss": 0.5332, + "step": 1282 + }, + { + "epoch": 0.54, + "learning_rate": 1.99278413520079e-05, + "loss": 0.5414, + "step": 1283 + }, + { + "epoch": 0.54, + "learning_rate": 1.992763790399021e-05, + "loss": 0.6006, + "step": 1284 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927434170611373e-05, + "loss": 0.519, + "step": 1285 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927230151877253e-05, + "loss": 0.5043, + "step": 1286 + }, + { + "epoch": 0.54, + "learning_rate": 1.9927025847793714e-05, + "loss": 0.5028, + "step": 1287 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926821258366622e-05, + "loss": 0.5702, + "step": 1288 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926616383601862e-05, + "loss": 0.5905, + "step": 1289 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926411223505323e-05, + "loss": 0.5257, + "step": 1290 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926205778082903e-05, + "loss": 0.4561, + "step": 1291 + }, + { + "epoch": 0.54, + "learning_rate": 1.9926000047340507e-05, + "loss": 0.5467, + "step": 1292 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925794031284046e-05, + "loss": 0.5829, + "step": 1293 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925587729919447e-05, + "loss": 0.5134, + "step": 1294 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925381143252633e-05, + "loss": 0.5486, + "step": 1295 + }, + { + "epoch": 0.54, + "learning_rate": 1.9925174271289548e-05, + "loss": 0.5339, + "step": 1296 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924967114036133e-05, + "loss": 0.5039, + "step": 1297 + }, + { + "epoch": 0.54, + "learning_rate": 1.992475967149835e-05, + "loss": 0.5455, + "step": 1298 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924551943682153e-05, + "loss": 0.5704, + "step": 1299 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924343930593516e-05, + "loss": 0.5534, + "step": 1300 + }, + { + "epoch": 0.54, + "learning_rate": 1.9924135632238423e-05, + "loss": 0.4944, + "step": 1301 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923927048622857e-05, + "loss": 0.5032, + "step": 1302 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923718179752815e-05, + "loss": 0.557, + "step": 1303 + }, + { + "epoch": 0.55, + "learning_rate": 1.99235090256343e-05, + "loss": 0.4871, + "step": 1304 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923299586273323e-05, + "loss": 0.5246, + "step": 1305 + }, + { + "epoch": 0.55, + "learning_rate": 1.9923089861675906e-05, + "loss": 0.5811, + "step": 1306 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922879851848077e-05, + "loss": 0.5093, + "step": 1307 + }, + { + "epoch": 0.55, + "learning_rate": 1.992266955679587e-05, + "loss": 0.5393, + "step": 1308 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922458976525335e-05, + "loss": 0.533, + "step": 1309 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922248111042518e-05, + "loss": 0.4694, + "step": 1310 + }, + { + "epoch": 0.55, + "learning_rate": 1.9922036960353486e-05, + "loss": 0.5957, + "step": 1311 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921825524464306e-05, + "loss": 0.5424, + "step": 1312 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921613803381055e-05, + "loss": 0.5007, + "step": 1313 + }, + { + "epoch": 0.55, + "learning_rate": 1.992140179710982e-05, + "loss": 0.4663, + "step": 1314 + }, + { + "epoch": 0.55, + "learning_rate": 1.9921189505656696e-05, + "loss": 0.5117, + "step": 1315 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920976929027782e-05, + "loss": 0.4508, + "step": 1316 + }, + { + "epoch": 0.55, + "learning_rate": 1.992076406722919e-05, + "loss": 0.5698, + "step": 1317 + }, + { + "epoch": 0.55, + "learning_rate": 1.992055092026704e-05, + "loss": 0.5356, + "step": 1318 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920337488147453e-05, + "loss": 0.5155, + "step": 1319 + }, + { + "epoch": 0.55, + "learning_rate": 1.9920123770876573e-05, + "loss": 0.6008, + "step": 1320 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919909768460535e-05, + "loss": 0.5054, + "step": 1321 + }, + { + "epoch": 0.55, + "learning_rate": 1.991969548090549e-05, + "loss": 0.4738, + "step": 1322 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919480908217606e-05, + "loss": 0.4882, + "step": 1323 + }, + { + "epoch": 0.55, + "learning_rate": 1.991926605040305e-05, + "loss": 0.5006, + "step": 1324 + }, + { + "epoch": 0.55, + "learning_rate": 1.9919050907467986e-05, + "loss": 0.5821, + "step": 1325 + }, + { + "epoch": 0.56, + "learning_rate": 1.9918835479418607e-05, + "loss": 0.5611, + "step": 1326 + }, + { + "epoch": 0.56, + "learning_rate": 1.9918619766261106e-05, + "loss": 0.5482, + "step": 1327 + }, + { + "epoch": 0.56, + "learning_rate": 1.991840376800168e-05, + "loss": 0.6395, + "step": 1328 + }, + { + "epoch": 0.56, + "learning_rate": 1.991818748464654e-05, + "loss": 0.5458, + "step": 1329 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917970916201904e-05, + "loss": 0.5189, + "step": 1330 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917754062673992e-05, + "loss": 0.5461, + "step": 1331 + }, + { + "epoch": 0.56, + "learning_rate": 1.991753692406904e-05, + "loss": 0.4995, + "step": 1332 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917319500393294e-05, + "loss": 0.609, + "step": 1333 + }, + { + "epoch": 0.56, + "learning_rate": 1.9917101791652997e-05, + "loss": 0.5375, + "step": 1334 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916883797854407e-05, + "loss": 0.4678, + "step": 1335 + }, + { + "epoch": 0.56, + "learning_rate": 1.991666551900379e-05, + "loss": 0.5375, + "step": 1336 + }, + { + "epoch": 0.56, + "learning_rate": 1.991644695510743e-05, + "loss": 0.441, + "step": 1337 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916228106171596e-05, + "loss": 0.5054, + "step": 1338 + }, + { + "epoch": 0.56, + "learning_rate": 1.9916008972202586e-05, + "loss": 0.5001, + "step": 1339 + }, + { + "epoch": 0.56, + "learning_rate": 1.99157895532067e-05, + "loss": 0.574, + "step": 1340 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915569849190243e-05, + "loss": 0.5413, + "step": 1341 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915349860159528e-05, + "loss": 0.5348, + "step": 1342 + }, + { + "epoch": 0.56, + "learning_rate": 1.9915129586120878e-05, + "loss": 0.4805, + "step": 1343 + }, + { + "epoch": 0.56, + "learning_rate": 1.991490902708063e-05, + "loss": 0.3818, + "step": 1344 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914688183045116e-05, + "loss": 0.5991, + "step": 1345 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914467054020692e-05, + "loss": 0.5398, + "step": 1346 + }, + { + "epoch": 0.56, + "learning_rate": 1.991424564001371e-05, + "loss": 0.5381, + "step": 1347 + }, + { + "epoch": 0.56, + "learning_rate": 1.9914023941030534e-05, + "loss": 0.5163, + "step": 1348 + }, + { + "epoch": 0.56, + "learning_rate": 1.9913801957077535e-05, + "loss": 0.5226, + "step": 1349 + }, + { + "epoch": 0.57, + "learning_rate": 1.99135796881611e-05, + "loss": 0.5332, + "step": 1350 + }, + { + "epoch": 0.57, + "learning_rate": 1.991335713428761e-05, + "loss": 0.5834, + "step": 1351 + }, + { + "epoch": 0.57, + "learning_rate": 1.991313429546347e-05, + "loss": 0.5745, + "step": 1352 + }, + { + "epoch": 0.57, + "learning_rate": 1.991291117169508e-05, + "loss": 0.5629, + "step": 1353 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912687762988855e-05, + "loss": 0.4963, + "step": 1354 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912464069351217e-05, + "loss": 0.4239, + "step": 1355 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912240090788595e-05, + "loss": 0.4935, + "step": 1356 + }, + { + "epoch": 0.57, + "learning_rate": 1.9912015827307426e-05, + "loss": 0.53, + "step": 1357 + }, + { + "epoch": 0.57, + "learning_rate": 1.991179127891416e-05, + "loss": 0.469, + "step": 1358 + }, + { + "epoch": 0.57, + "learning_rate": 1.9911566445615253e-05, + "loss": 0.537, + "step": 1359 + }, + { + "epoch": 0.57, + "learning_rate": 1.9911341327417156e-05, + "loss": 0.5467, + "step": 1360 + }, + { + "epoch": 0.57, + "learning_rate": 1.991111592432635e-05, + "loss": 0.4584, + "step": 1361 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910890236349316e-05, + "loss": 0.6264, + "step": 1362 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910664263492533e-05, + "loss": 0.5039, + "step": 1363 + }, + { + "epoch": 0.57, + "learning_rate": 1.9910438005762503e-05, + "loss": 0.4549, + "step": 1364 + }, + { + "epoch": 0.57, + "learning_rate": 1.991021146316572e-05, + "loss": 0.5509, + "step": 1365 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909984635708714e-05, + "loss": 0.4958, + "step": 1366 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909757523397985e-05, + "loss": 0.5492, + "step": 1367 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909530126240074e-05, + "loss": 0.5841, + "step": 1368 + }, + { + "epoch": 0.57, + "learning_rate": 1.9909302444241508e-05, + "loss": 0.545, + "step": 1369 + }, + { + "epoch": 0.57, + "learning_rate": 1.990907447740884e-05, + "loss": 0.4326, + "step": 1370 + }, + { + "epoch": 0.57, + "learning_rate": 1.9908846225748623e-05, + "loss": 0.5381, + "step": 1371 + }, + { + "epoch": 0.57, + "learning_rate": 1.990861768926741e-05, + "loss": 0.4573, + "step": 1372 + }, + { + "epoch": 0.57, + "learning_rate": 1.9908388867971772e-05, + "loss": 0.4962, + "step": 1373 + }, + { + "epoch": 0.58, + "learning_rate": 1.9908159761868292e-05, + "loss": 0.4974, + "step": 1374 + }, + { + "epoch": 0.58, + "learning_rate": 1.990793037096355e-05, + "loss": 0.4819, + "step": 1375 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907700695264144e-05, + "loss": 0.532, + "step": 1376 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907470734776674e-05, + "loss": 0.4813, + "step": 1377 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907240489507748e-05, + "loss": 0.483, + "step": 1378 + }, + { + "epoch": 0.58, + "learning_rate": 1.9907009959463983e-05, + "loss": 0.6402, + "step": 1379 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906779144652014e-05, + "loss": 0.5279, + "step": 1380 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906548045078464e-05, + "loss": 0.5868, + "step": 1381 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906316660749985e-05, + "loss": 0.4664, + "step": 1382 + }, + { + "epoch": 0.58, + "learning_rate": 1.9906084991673224e-05, + "loss": 0.571, + "step": 1383 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905853037854834e-05, + "loss": 0.5611, + "step": 1384 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905620799301495e-05, + "loss": 0.4523, + "step": 1385 + }, + { + "epoch": 0.58, + "learning_rate": 1.9905388276019873e-05, + "loss": 0.6086, + "step": 1386 + }, + { + "epoch": 0.58, + "learning_rate": 1.990515546801666e-05, + "loss": 0.483, + "step": 1387 + }, + { + "epoch": 0.58, + "learning_rate": 1.990492237529854e-05, + "loss": 0.6169, + "step": 1388 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904688997872215e-05, + "loss": 0.4782, + "step": 1389 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904455335744395e-05, + "loss": 0.5676, + "step": 1390 + }, + { + "epoch": 0.58, + "learning_rate": 1.9904221388921794e-05, + "loss": 0.6075, + "step": 1391 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903987157411138e-05, + "loss": 0.4489, + "step": 1392 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903752641219163e-05, + "loss": 0.5618, + "step": 1393 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903517840352605e-05, + "loss": 0.5132, + "step": 1394 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903282754818216e-05, + "loss": 0.4228, + "step": 1395 + }, + { + "epoch": 0.58, + "learning_rate": 1.9903047384622748e-05, + "loss": 0.513, + "step": 1396 + }, + { + "epoch": 0.58, + "learning_rate": 1.9902811729772974e-05, + "loss": 0.5782, + "step": 1397 + }, + { + "epoch": 0.59, + "learning_rate": 1.990257579027566e-05, + "loss": 0.5138, + "step": 1398 + }, + { + "epoch": 0.59, + "learning_rate": 1.99023395661376e-05, + "loss": 0.5087, + "step": 1399 + }, + { + "epoch": 0.59, + "learning_rate": 1.9902103057365573e-05, + "loss": 0.6183, + "step": 1400 + }, + { + "epoch": 0.59, + "learning_rate": 1.990186626396638e-05, + "loss": 0.5741, + "step": 1401 + }, + { + "epoch": 0.59, + "learning_rate": 1.9901629185946827e-05, + "loss": 0.4439, + "step": 1402 + }, + { + "epoch": 0.59, + "learning_rate": 1.990139182331373e-05, + "loss": 0.6, + "step": 1403 + }, + { + "epoch": 0.59, + "learning_rate": 1.9901154176073914e-05, + "loss": 0.5193, + "step": 1404 + }, + { + "epoch": 0.59, + "learning_rate": 1.9900916244234203e-05, + "loss": 0.6317, + "step": 1405 + }, + { + "epoch": 0.59, + "learning_rate": 1.9900678027801445e-05, + "loss": 0.5093, + "step": 1406 + }, + { + "epoch": 0.59, + "learning_rate": 1.990043952678248e-05, + "loss": 0.4804, + "step": 1407 + }, + { + "epoch": 0.59, + "learning_rate": 1.990020074118417e-05, + "loss": 0.5628, + "step": 1408 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899961671013367e-05, + "loss": 0.5073, + "step": 1409 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899722316276958e-05, + "loss": 0.5252, + "step": 1410 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899482676981812e-05, + "loss": 0.5157, + "step": 1411 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899242753134824e-05, + "loss": 0.5309, + "step": 1412 + }, + { + "epoch": 0.59, + "learning_rate": 1.9899002544742886e-05, + "loss": 0.4717, + "step": 1413 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898762051812906e-05, + "loss": 0.4811, + "step": 1414 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898521274351793e-05, + "loss": 0.496, + "step": 1415 + }, + { + "epoch": 0.59, + "learning_rate": 1.989828021236647e-05, + "loss": 0.4783, + "step": 1416 + }, + { + "epoch": 0.59, + "learning_rate": 1.9898038865863864e-05, + "loss": 0.4625, + "step": 1417 + }, + { + "epoch": 0.59, + "learning_rate": 1.9897797234850916e-05, + "loss": 0.4816, + "step": 1418 + }, + { + "epoch": 0.59, + "learning_rate": 1.989755531933457e-05, + "loss": 0.5341, + "step": 1419 + }, + { + "epoch": 0.59, + "learning_rate": 1.989731311932178e-05, + "loss": 0.5596, + "step": 1420 + }, + { + "epoch": 0.59, + "learning_rate": 1.9897070634819506e-05, + "loss": 0.4642, + "step": 1421 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896827865834717e-05, + "loss": 0.4795, + "step": 1422 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896584812374397e-05, + "loss": 0.4566, + "step": 1423 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896341474445526e-05, + "loss": 0.5784, + "step": 1424 + }, + { + "epoch": 0.6, + "learning_rate": 1.9896097852055103e-05, + "loss": 0.5538, + "step": 1425 + }, + { + "epoch": 0.6, + "learning_rate": 1.989585394521013e-05, + "loss": 0.5432, + "step": 1426 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895609753917614e-05, + "loss": 0.4524, + "step": 1427 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895365278184577e-05, + "loss": 0.4955, + "step": 1428 + }, + { + "epoch": 0.6, + "learning_rate": 1.9895120518018046e-05, + "loss": 0.443, + "step": 1429 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894875473425057e-05, + "loss": 0.5583, + "step": 1430 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894630144412656e-05, + "loss": 0.5227, + "step": 1431 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894384530987885e-05, + "loss": 0.5223, + "step": 1432 + }, + { + "epoch": 0.6, + "learning_rate": 1.9894138633157816e-05, + "loss": 0.4718, + "step": 1433 + }, + { + "epoch": 0.6, + "learning_rate": 1.989389245092951e-05, + "loss": 0.4776, + "step": 1434 + }, + { + "epoch": 0.6, + "learning_rate": 1.9893645984310047e-05, + "loss": 0.4808, + "step": 1435 + }, + { + "epoch": 0.6, + "learning_rate": 1.989339923330651e-05, + "loss": 0.5571, + "step": 1436 + }, + { + "epoch": 0.6, + "learning_rate": 1.989315219792599e-05, + "loss": 0.4956, + "step": 1437 + }, + { + "epoch": 0.6, + "learning_rate": 1.989290487817559e-05, + "loss": 0.5159, + "step": 1438 + }, + { + "epoch": 0.6, + "learning_rate": 1.9892657274062415e-05, + "loss": 0.5385, + "step": 1439 + }, + { + "epoch": 0.6, + "learning_rate": 1.9892409385593586e-05, + "loss": 0.6072, + "step": 1440 + }, + { + "epoch": 0.6, + "learning_rate": 1.989216121277623e-05, + "loss": 0.4643, + "step": 1441 + }, + { + "epoch": 0.6, + "learning_rate": 1.9891912755617477e-05, + "loss": 0.4857, + "step": 1442 + }, + { + "epoch": 0.6, + "learning_rate": 1.989166401412447e-05, + "loss": 0.4994, + "step": 1443 + }, + { + "epoch": 0.6, + "learning_rate": 1.9891414988304357e-05, + "loss": 0.4851, + "step": 1444 + }, + { + "epoch": 0.6, + "learning_rate": 1.98911656781643e-05, + "loss": 0.5248, + "step": 1445 + }, + { + "epoch": 0.61, + "learning_rate": 1.9890916083711463e-05, + "loss": 0.5135, + "step": 1446 + }, + { + "epoch": 0.61, + "learning_rate": 1.9890666204953018e-05, + "loss": 0.5424, + "step": 1447 + }, + { + "epoch": 0.61, + "learning_rate": 1.989041604189615e-05, + "loss": 0.4628, + "step": 1448 + }, + { + "epoch": 0.61, + "learning_rate": 1.989016559454805e-05, + "loss": 0.5106, + "step": 1449 + }, + { + "epoch": 0.61, + "learning_rate": 1.988991486291592e-05, + "loss": 0.5695, + "step": 1450 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889663847006956e-05, + "loss": 0.5715, + "step": 1451 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889412546828386e-05, + "loss": 0.4649, + "step": 1452 + }, + { + "epoch": 0.61, + "learning_rate": 1.9889160962387427e-05, + "loss": 0.5909, + "step": 1453 + }, + { + "epoch": 0.61, + "learning_rate": 1.988890909369131e-05, + "loss": 0.5654, + "step": 1454 + }, + { + "epoch": 0.61, + "learning_rate": 1.988865694074728e-05, + "loss": 0.5696, + "step": 1455 + }, + { + "epoch": 0.61, + "learning_rate": 1.988840450356258e-05, + "loss": 0.486, + "step": 1456 + }, + { + "epoch": 0.61, + "learning_rate": 1.9888151782144463e-05, + "loss": 0.4263, + "step": 1457 + }, + { + "epoch": 0.61, + "learning_rate": 1.9887898776500203e-05, + "loss": 0.4926, + "step": 1458 + }, + { + "epoch": 0.61, + "learning_rate": 1.9887645486637063e-05, + "loss": 0.5234, + "step": 1459 + }, + { + "epoch": 0.61, + "learning_rate": 1.988739191256233e-05, + "loss": 0.4934, + "step": 1460 + }, + { + "epoch": 0.61, + "learning_rate": 1.988713805428329e-05, + "loss": 0.5068, + "step": 1461 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886883911807242e-05, + "loss": 0.5427, + "step": 1462 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886629485141488e-05, + "loss": 0.524, + "step": 1463 + }, + { + "epoch": 0.61, + "learning_rate": 1.988637477429334e-05, + "loss": 0.4719, + "step": 1464 + }, + { + "epoch": 0.61, + "learning_rate": 1.9886119779270126e-05, + "loss": 0.5084, + "step": 1465 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885864500079172e-05, + "loss": 0.4963, + "step": 1466 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885608936727814e-05, + "loss": 0.5211, + "step": 1467 + }, + { + "epoch": 0.61, + "learning_rate": 1.9885353089223397e-05, + "loss": 0.4711, + "step": 1468 + }, + { + "epoch": 0.62, + "learning_rate": 1.988509695757328e-05, + "loss": 0.5843, + "step": 1469 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884840541784824e-05, + "loss": 0.5093, + "step": 1470 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884583841865395e-05, + "loss": 0.4973, + "step": 1471 + }, + { + "epoch": 0.62, + "learning_rate": 1.9884326857822378e-05, + "loss": 0.5495, + "step": 1472 + }, + { + "epoch": 0.62, + "learning_rate": 1.988406958966316e-05, + "loss": 0.6042, + "step": 1473 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883812037395126e-05, + "loss": 0.4894, + "step": 1474 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883554201025687e-05, + "loss": 0.5177, + "step": 1475 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883296080562255e-05, + "loss": 0.4454, + "step": 1476 + }, + { + "epoch": 0.62, + "learning_rate": 1.9883037676012245e-05, + "loss": 0.534, + "step": 1477 + }, + { + "epoch": 0.62, + "learning_rate": 1.988277898738309e-05, + "loss": 0.4713, + "step": 1478 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882520014682218e-05, + "loss": 0.5647, + "step": 1479 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882260757917083e-05, + "loss": 0.5868, + "step": 1480 + }, + { + "epoch": 0.62, + "learning_rate": 1.9882001217095127e-05, + "loss": 0.5463, + "step": 1481 + }, + { + "epoch": 0.62, + "learning_rate": 1.9881741392223822e-05, + "loss": 0.5546, + "step": 1482 + }, + { + "epoch": 0.62, + "learning_rate": 1.9881481283310624e-05, + "loss": 0.5102, + "step": 1483 + }, + { + "epoch": 0.62, + "learning_rate": 1.988122089036302e-05, + "loss": 0.4513, + "step": 1484 + }, + { + "epoch": 0.62, + "learning_rate": 1.9880960213388483e-05, + "loss": 0.5327, + "step": 1485 + }, + { + "epoch": 0.62, + "learning_rate": 1.988069925239452e-05, + "loss": 0.5329, + "step": 1486 + }, + { + "epoch": 0.62, + "learning_rate": 1.988043800738862e-05, + "loss": 0.522, + "step": 1487 + }, + { + "epoch": 0.62, + "learning_rate": 1.98801764783783e-05, + "loss": 0.5105, + "step": 1488 + }, + { + "epoch": 0.62, + "learning_rate": 1.987991466537107e-05, + "loss": 0.5621, + "step": 1489 + }, + { + "epoch": 0.62, + "learning_rate": 1.9879652568374467e-05, + "loss": 0.5259, + "step": 1490 + }, + { + "epoch": 0.62, + "learning_rate": 1.987939018739601e-05, + "loss": 0.464, + "step": 1491 + }, + { + "epoch": 0.62, + "learning_rate": 1.9879127522443255e-05, + "loss": 0.5808, + "step": 1492 + }, + { + "epoch": 0.63, + "learning_rate": 1.9878864573523744e-05, + "loss": 0.5933, + "step": 1493 + }, + { + "epoch": 0.63, + "learning_rate": 1.987860134064504e-05, + "loss": 0.4712, + "step": 1494 + }, + { + "epoch": 0.63, + "learning_rate": 1.98783378238147e-05, + "loss": 0.5199, + "step": 1495 + }, + { + "epoch": 0.63, + "learning_rate": 1.9878074023040313e-05, + "loss": 0.5006, + "step": 1496 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877809938329447e-05, + "loss": 0.5374, + "step": 1497 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877545569689705e-05, + "loss": 0.4978, + "step": 1498 + }, + { + "epoch": 0.63, + "learning_rate": 1.987728091712868e-05, + "loss": 0.4904, + "step": 1499 + }, + { + "epoch": 0.63, + "learning_rate": 1.9877015980653978e-05, + "loss": 0.5521, + "step": 1500 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876750760273218e-05, + "loss": 0.6263, + "step": 1501 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876485255994022e-05, + "loss": 0.5446, + "step": 1502 + }, + { + "epoch": 0.63, + "learning_rate": 1.9876219467824018e-05, + "loss": 0.4413, + "step": 1503 + }, + { + "epoch": 0.63, + "learning_rate": 1.987595339577085e-05, + "loss": 0.4664, + "step": 1504 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875687039842168e-05, + "loss": 0.5025, + "step": 1505 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875420400045625e-05, + "loss": 0.5362, + "step": 1506 + }, + { + "epoch": 0.63, + "learning_rate": 1.9875153476388887e-05, + "loss": 0.5398, + "step": 1507 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874886268879625e-05, + "loss": 0.4848, + "step": 1508 + }, + { + "epoch": 0.63, + "learning_rate": 1.987461877752552e-05, + "loss": 0.4712, + "step": 1509 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874351002334257e-05, + "loss": 0.5298, + "step": 1510 + }, + { + "epoch": 0.63, + "learning_rate": 1.9874082943313543e-05, + "loss": 0.5308, + "step": 1511 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873814600471073e-05, + "loss": 0.5389, + "step": 1512 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873545973814564e-05, + "loss": 0.4738, + "step": 1513 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873277063351737e-05, + "loss": 0.5454, + "step": 1514 + }, + { + "epoch": 0.63, + "learning_rate": 1.9873007869090326e-05, + "loss": 0.5225, + "step": 1515 + }, + { + "epoch": 0.63, + "learning_rate": 1.9872738391038063e-05, + "loss": 0.4892, + "step": 1516 + }, + { + "epoch": 0.64, + "learning_rate": 1.9872468629202696e-05, + "loss": 0.5003, + "step": 1517 + }, + { + "epoch": 0.64, + "learning_rate": 1.9872198583591976e-05, + "loss": 0.5106, + "step": 1518 + }, + { + "epoch": 0.64, + "learning_rate": 1.987192825421367e-05, + "loss": 0.6494, + "step": 1519 + }, + { + "epoch": 0.64, + "learning_rate": 1.987165764107555e-05, + "loss": 0.5218, + "step": 1520 + }, + { + "epoch": 0.64, + "learning_rate": 1.9871386744185386e-05, + "loss": 0.5341, + "step": 1521 + }, + { + "epoch": 0.64, + "learning_rate": 1.9871115563550974e-05, + "loss": 0.4962, + "step": 1522 + }, + { + "epoch": 0.64, + "learning_rate": 1.98708440991801e-05, + "loss": 0.5314, + "step": 1523 + }, + { + "epoch": 0.64, + "learning_rate": 1.9870572351080573e-05, + "loss": 0.5039, + "step": 1524 + }, + { + "epoch": 0.64, + "learning_rate": 1.9870300319260204e-05, + "loss": 0.4734, + "step": 1525 + }, + { + "epoch": 0.64, + "learning_rate": 1.987002800372681e-05, + "loss": 0.565, + "step": 1526 + }, + { + "epoch": 0.64, + "learning_rate": 1.986975540448822e-05, + "loss": 0.5463, + "step": 1527 + }, + { + "epoch": 0.64, + "learning_rate": 1.986948252155227e-05, + "loss": 0.504, + "step": 1528 + }, + { + "epoch": 0.64, + "learning_rate": 1.9869209354926802e-05, + "loss": 0.4311, + "step": 1529 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868935904619673e-05, + "loss": 0.4794, + "step": 1530 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868662170638732e-05, + "loss": 0.5152, + "step": 1531 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868388152991858e-05, + "loss": 0.5112, + "step": 1532 + }, + { + "epoch": 0.64, + "learning_rate": 1.9868113851686925e-05, + "loss": 0.5007, + "step": 1533 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867839266731813e-05, + "loss": 0.5184, + "step": 1534 + }, + { + "epoch": 0.64, + "learning_rate": 1.986756439813442e-05, + "loss": 0.4357, + "step": 1535 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867289245902645e-05, + "loss": 0.532, + "step": 1536 + }, + { + "epoch": 0.64, + "learning_rate": 1.9867013810044394e-05, + "loss": 0.5021, + "step": 1537 + }, + { + "epoch": 0.64, + "learning_rate": 1.9866738090567586e-05, + "loss": 0.4668, + "step": 1538 + }, + { + "epoch": 0.64, + "learning_rate": 1.9866462087480153e-05, + "loss": 0.4493, + "step": 1539 + }, + { + "epoch": 0.64, + "learning_rate": 1.986618580079002e-05, + "loss": 0.503, + "step": 1540 + }, + { + "epoch": 0.65, + "learning_rate": 1.986590923050513e-05, + "loss": 0.5003, + "step": 1541 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865632376633433e-05, + "loss": 0.5519, + "step": 1542 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865355239182892e-05, + "loss": 0.4793, + "step": 1543 + }, + { + "epoch": 0.65, + "learning_rate": 1.9865077818161464e-05, + "loss": 0.4704, + "step": 1544 + }, + { + "epoch": 0.65, + "learning_rate": 1.9864800113577133e-05, + "loss": 0.5161, + "step": 1545 + }, + { + "epoch": 0.65, + "learning_rate": 1.9864522125437874e-05, + "loss": 0.4979, + "step": 1546 + }, + { + "epoch": 0.65, + "learning_rate": 1.986424385375168e-05, + "loss": 0.5197, + "step": 1547 + }, + { + "epoch": 0.65, + "learning_rate": 1.986396529852655e-05, + "loss": 0.4897, + "step": 1548 + }, + { + "epoch": 0.65, + "learning_rate": 1.986368645977049e-05, + "loss": 0.4824, + "step": 1549 + }, + { + "epoch": 0.65, + "learning_rate": 1.9863407337491516e-05, + "loss": 0.565, + "step": 1550 + }, + { + "epoch": 0.65, + "learning_rate": 1.986312793169765e-05, + "loss": 0.5372, + "step": 1551 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862848242396927e-05, + "loss": 0.485, + "step": 1552 + }, + { + "epoch": 0.65, + "learning_rate": 1.986256826959738e-05, + "loss": 0.5229, + "step": 1553 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862288013307062e-05, + "loss": 0.583, + "step": 1554 + }, + { + "epoch": 0.65, + "learning_rate": 1.9862007473534026e-05, + "loss": 0.5444, + "step": 1555 + }, + { + "epoch": 0.65, + "learning_rate": 1.9861726650286338e-05, + "loss": 0.5535, + "step": 1556 + }, + { + "epoch": 0.65, + "learning_rate": 1.986144554357207e-05, + "loss": 0.4921, + "step": 1557 + }, + { + "epoch": 0.65, + "learning_rate": 1.9861164153399297e-05, + "loss": 0.5669, + "step": 1558 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860882479776113e-05, + "loss": 0.5139, + "step": 1559 + }, + { + "epoch": 0.65, + "learning_rate": 1.986060052271061e-05, + "loss": 0.5647, + "step": 1560 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860318282210896e-05, + "loss": 0.6061, + "step": 1561 + }, + { + "epoch": 0.65, + "learning_rate": 1.9860035758285082e-05, + "loss": 0.5984, + "step": 1562 + }, + { + "epoch": 0.65, + "learning_rate": 1.9859752950941292e-05, + "loss": 0.4112, + "step": 1563 + }, + { + "epoch": 0.65, + "learning_rate": 1.9859469860187654e-05, + "loss": 0.4325, + "step": 1564 + }, + { + "epoch": 0.66, + "learning_rate": 1.9859186486032302e-05, + "loss": 0.5618, + "step": 1565 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858902828483382e-05, + "loss": 0.5109, + "step": 1566 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858618887549053e-05, + "loss": 0.5181, + "step": 1567 + }, + { + "epoch": 0.66, + "learning_rate": 1.985833466323747e-05, + "loss": 0.5925, + "step": 1568 + }, + { + "epoch": 0.66, + "learning_rate": 1.9858050155556807e-05, + "loss": 0.5414, + "step": 1569 + }, + { + "epoch": 0.66, + "learning_rate": 1.9857765364515234e-05, + "loss": 0.5568, + "step": 1570 + }, + { + "epoch": 0.66, + "learning_rate": 1.9857480290120947e-05, + "loss": 0.5295, + "step": 1571 + }, + { + "epoch": 0.66, + "learning_rate": 1.985719493238214e-05, + "loss": 0.5144, + "step": 1572 + }, + { + "epoch": 0.66, + "learning_rate": 1.9856909291307008e-05, + "loss": 0.4398, + "step": 1573 + }, + { + "epoch": 0.66, + "learning_rate": 1.9856623366903765e-05, + "loss": 0.5049, + "step": 1574 + }, + { + "epoch": 0.66, + "learning_rate": 1.985633715918063e-05, + "loss": 0.539, + "step": 1575 + }, + { + "epoch": 0.66, + "learning_rate": 1.985605066814583e-05, + "loss": 0.4789, + "step": 1576 + }, + { + "epoch": 0.66, + "learning_rate": 1.9855763893807603e-05, + "loss": 0.4971, + "step": 1577 + }, + { + "epoch": 0.66, + "learning_rate": 1.9855476836174182e-05, + "loss": 0.4828, + "step": 1578 + }, + { + "epoch": 0.66, + "learning_rate": 1.985518949525383e-05, + "loss": 0.4806, + "step": 1579 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854901871054803e-05, + "loss": 0.5103, + "step": 1580 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854613963585362e-05, + "loss": 0.4901, + "step": 1581 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854325772853792e-05, + "loss": 0.5219, + "step": 1582 + }, + { + "epoch": 0.66, + "learning_rate": 1.9854037298868367e-05, + "loss": 0.5073, + "step": 1583 + }, + { + "epoch": 0.66, + "learning_rate": 1.9853748541637387e-05, + "loss": 0.5223, + "step": 1584 + }, + { + "epoch": 0.66, + "learning_rate": 1.985345950116915e-05, + "loss": 0.525, + "step": 1585 + }, + { + "epoch": 0.66, + "learning_rate": 1.9853170177471962e-05, + "loss": 0.4888, + "step": 1586 + }, + { + "epoch": 0.66, + "learning_rate": 1.9852880570554142e-05, + "loss": 0.5425, + "step": 1587 + }, + { + "epoch": 0.66, + "learning_rate": 1.9852590680424012e-05, + "loss": 0.5606, + "step": 1588 + }, + { + "epoch": 0.67, + "learning_rate": 1.9852300507089907e-05, + "loss": 0.5552, + "step": 1589 + }, + { + "epoch": 0.67, + "learning_rate": 1.9852010050560165e-05, + "loss": 0.4529, + "step": 1590 + }, + { + "epoch": 0.67, + "learning_rate": 1.985171931084314e-05, + "loss": 0.3869, + "step": 1591 + }, + { + "epoch": 0.67, + "learning_rate": 1.9851428287947184e-05, + "loss": 0.4744, + "step": 1592 + }, + { + "epoch": 0.67, + "learning_rate": 1.9851136981880662e-05, + "loss": 0.5176, + "step": 1593 + }, + { + "epoch": 0.67, + "learning_rate": 1.985084539265195e-05, + "loss": 0.4943, + "step": 1594 + }, + { + "epoch": 0.67, + "learning_rate": 1.985055352026943e-05, + "loss": 0.4605, + "step": 1595 + }, + { + "epoch": 0.67, + "learning_rate": 1.985026136474149e-05, + "loss": 0.564, + "step": 1596 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849968926076528e-05, + "loss": 0.5434, + "step": 1597 + }, + { + "epoch": 0.67, + "learning_rate": 1.984967620428295e-05, + "loss": 0.5173, + "step": 1598 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849383199369167e-05, + "loss": 0.5304, + "step": 1599 + }, + { + "epoch": 0.67, + "learning_rate": 1.9849089911343608e-05, + "loss": 0.5106, + "step": 1600 + }, + { + "epoch": 0.67, + "learning_rate": 1.98487963402147e-05, + "loss": 0.4964, + "step": 1601 + }, + { + "epoch": 0.67, + "learning_rate": 1.9848502485990876e-05, + "loss": 0.4417, + "step": 1602 + }, + { + "epoch": 0.67, + "learning_rate": 1.9848208348680594e-05, + "loss": 0.4463, + "step": 1603 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847913928292297e-05, + "loss": 0.5096, + "step": 1604 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847619224834455e-05, + "loss": 0.5129, + "step": 1605 + }, + { + "epoch": 0.67, + "learning_rate": 1.9847324238315536e-05, + "loss": 0.4326, + "step": 1606 + }, + { + "epoch": 0.67, + "learning_rate": 1.984702896874402e-05, + "loss": 0.5151, + "step": 1607 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846733416128393e-05, + "loss": 0.6225, + "step": 1608 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846437580477155e-05, + "loss": 0.5483, + "step": 1609 + }, + { + "epoch": 0.67, + "learning_rate": 1.9846141461798804e-05, + "loss": 0.5227, + "step": 1610 + }, + { + "epoch": 0.67, + "learning_rate": 1.984584506010186e-05, + "loss": 0.4964, + "step": 1611 + }, + { + "epoch": 0.67, + "learning_rate": 1.984554837539483e-05, + "loss": 0.5653, + "step": 1612 + }, + { + "epoch": 0.68, + "learning_rate": 1.984525140768625e-05, + "loss": 0.5379, + "step": 1613 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844954156984656e-05, + "loss": 0.5595, + "step": 1614 + }, + { + "epoch": 0.68, + "learning_rate": 1.984465662329859e-05, + "loss": 0.5941, + "step": 1615 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844358806636605e-05, + "loss": 0.4752, + "step": 1616 + }, + { + "epoch": 0.68, + "learning_rate": 1.9844060707007263e-05, + "loss": 0.5019, + "step": 1617 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843762324419132e-05, + "loss": 0.5074, + "step": 1618 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843463658880786e-05, + "loss": 0.628, + "step": 1619 + }, + { + "epoch": 0.68, + "learning_rate": 1.9843164710400815e-05, + "loss": 0.5083, + "step": 1620 + }, + { + "epoch": 0.68, + "learning_rate": 1.9842865478987808e-05, + "loss": 0.4711, + "step": 1621 + }, + { + "epoch": 0.68, + "learning_rate": 1.9842565964650363e-05, + "loss": 0.5057, + "step": 1622 + }, + { + "epoch": 0.68, + "learning_rate": 1.98422661673971e-05, + "loss": 0.5801, + "step": 1623 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841966087236626e-05, + "loss": 0.4771, + "step": 1624 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841665724177572e-05, + "loss": 0.5187, + "step": 1625 + }, + { + "epoch": 0.68, + "learning_rate": 1.984136507822857e-05, + "loss": 0.4719, + "step": 1626 + }, + { + "epoch": 0.68, + "learning_rate": 1.9841064149398262e-05, + "loss": 0.4859, + "step": 1627 + }, + { + "epoch": 0.68, + "learning_rate": 1.9840762937695296e-05, + "loss": 0.4938, + "step": 1628 + }, + { + "epoch": 0.68, + "learning_rate": 1.9840461443128333e-05, + "loss": 0.5628, + "step": 1629 + }, + { + "epoch": 0.68, + "learning_rate": 1.984015966570604e-05, + "loss": 0.5753, + "step": 1630 + }, + { + "epoch": 0.68, + "learning_rate": 1.9839857605437086e-05, + "loss": 0.4204, + "step": 1631 + }, + { + "epoch": 0.68, + "learning_rate": 1.983955526233016e-05, + "loss": 0.4693, + "step": 1632 + }, + { + "epoch": 0.68, + "learning_rate": 1.9839252636393946e-05, + "loss": 0.5352, + "step": 1633 + }, + { + "epoch": 0.68, + "learning_rate": 1.983894972763715e-05, + "loss": 0.5136, + "step": 1634 + }, + { + "epoch": 0.68, + "learning_rate": 1.9838646536068473e-05, + "loss": 0.5446, + "step": 1635 + }, + { + "epoch": 0.68, + "learning_rate": 1.9838343061696633e-05, + "loss": 0.5841, + "step": 1636 + }, + { + "epoch": 0.69, + "learning_rate": 1.983803930453035e-05, + "loss": 0.5229, + "step": 1637 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837735264578358e-05, + "loss": 0.5148, + "step": 1638 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837430941849395e-05, + "loss": 0.4822, + "step": 1639 + }, + { + "epoch": 0.69, + "learning_rate": 1.9837126336352214e-05, + "loss": 0.5169, + "step": 1640 + }, + { + "epoch": 0.69, + "learning_rate": 1.983682144809556e-05, + "loss": 0.4126, + "step": 1641 + }, + { + "epoch": 0.69, + "learning_rate": 1.9836516277088203e-05, + "loss": 0.4774, + "step": 1642 + }, + { + "epoch": 0.69, + "learning_rate": 1.9836210823338917e-05, + "loss": 0.5266, + "step": 1643 + }, + { + "epoch": 0.69, + "learning_rate": 1.983590508685648e-05, + "loss": 0.4671, + "step": 1644 + }, + { + "epoch": 0.69, + "learning_rate": 1.9835599067649676e-05, + "loss": 0.5408, + "step": 1645 + }, + { + "epoch": 0.69, + "learning_rate": 1.983529276572731e-05, + "loss": 0.5083, + "step": 1646 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834986181098176e-05, + "loss": 0.5176, + "step": 1647 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834679313771095e-05, + "loss": 0.5439, + "step": 1648 + }, + { + "epoch": 0.69, + "learning_rate": 1.9834372163754882e-05, + "loss": 0.5107, + "step": 1649 + }, + { + "epoch": 0.69, + "learning_rate": 1.983406473105837e-05, + "loss": 0.5676, + "step": 1650 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833757015690393e-05, + "loss": 0.4696, + "step": 1651 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833449017659796e-05, + "loss": 0.5984, + "step": 1652 + }, + { + "epoch": 0.69, + "learning_rate": 1.9833140736975433e-05, + "loss": 0.5636, + "step": 1653 + }, + { + "epoch": 0.69, + "learning_rate": 1.9832832173646164e-05, + "loss": 0.5231, + "step": 1654 + }, + { + "epoch": 0.69, + "learning_rate": 1.9832523327680865e-05, + "loss": 0.5034, + "step": 1655 + }, + { + "epoch": 0.69, + "learning_rate": 1.98322141990884e-05, + "loss": 0.5617, + "step": 1656 + }, + { + "epoch": 0.69, + "learning_rate": 1.983190478787767e-05, + "loss": 0.4909, + "step": 1657 + }, + { + "epoch": 0.69, + "learning_rate": 1.983159509405756e-05, + "loss": 0.4736, + "step": 1658 + }, + { + "epoch": 0.69, + "learning_rate": 1.9831285117636972e-05, + "loss": 0.592, + "step": 1659 + }, + { + "epoch": 0.69, + "learning_rate": 1.983097485862482e-05, + "loss": 0.5005, + "step": 1660 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830664317030014e-05, + "loss": 0.5292, + "step": 1661 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830353492861493e-05, + "loss": 0.5129, + "step": 1662 + }, + { + "epoch": 0.7, + "learning_rate": 1.9830042386128178e-05, + "loss": 0.575, + "step": 1663 + }, + { + "epoch": 0.7, + "learning_rate": 1.982973099683902e-05, + "loss": 0.4696, + "step": 1664 + }, + { + "epoch": 0.7, + "learning_rate": 1.9829419325002966e-05, + "loss": 0.4746, + "step": 1665 + }, + { + "epoch": 0.7, + "learning_rate": 1.982910737062898e-05, + "loss": 0.4449, + "step": 1666 + }, + { + "epoch": 0.7, + "learning_rate": 1.982879513372602e-05, + "loss": 0.4704, + "step": 1667 + }, + { + "epoch": 0.7, + "learning_rate": 1.9828482614303068e-05, + "loss": 0.5179, + "step": 1668 + }, + { + "epoch": 0.7, + "learning_rate": 1.9828169812369106e-05, + "loss": 0.4781, + "step": 1669 + }, + { + "epoch": 0.7, + "learning_rate": 1.9827856727933122e-05, + "loss": 0.5388, + "step": 1670 + }, + { + "epoch": 0.7, + "learning_rate": 1.9827543361004118e-05, + "loss": 0.5813, + "step": 1671 + }, + { + "epoch": 0.7, + "learning_rate": 1.98272297115911e-05, + "loss": 0.4643, + "step": 1672 + }, + { + "epoch": 0.7, + "learning_rate": 1.9826915779703087e-05, + "loss": 0.4306, + "step": 1673 + }, + { + "epoch": 0.7, + "learning_rate": 1.98266015653491e-05, + "loss": 0.5216, + "step": 1674 + }, + { + "epoch": 0.7, + "learning_rate": 1.9826287068538167e-05, + "loss": 0.5218, + "step": 1675 + }, + { + "epoch": 0.7, + "learning_rate": 1.9825972289279336e-05, + "loss": 0.519, + "step": 1676 + }, + { + "epoch": 0.7, + "learning_rate": 1.982565722758165e-05, + "loss": 0.4881, + "step": 1677 + }, + { + "epoch": 0.7, + "learning_rate": 1.9825341883454165e-05, + "loss": 0.5389, + "step": 1678 + }, + { + "epoch": 0.7, + "learning_rate": 1.982502625690595e-05, + "loss": 0.5541, + "step": 1679 + }, + { + "epoch": 0.7, + "learning_rate": 1.982471034794607e-05, + "loss": 0.4128, + "step": 1680 + }, + { + "epoch": 0.7, + "learning_rate": 1.982439415658361e-05, + "loss": 0.4948, + "step": 1681 + }, + { + "epoch": 0.7, + "learning_rate": 1.9824077682827658e-05, + "loss": 0.4972, + "step": 1682 + }, + { + "epoch": 0.7, + "learning_rate": 1.982376092668731e-05, + "loss": 0.5727, + "step": 1683 + }, + { + "epoch": 0.71, + "learning_rate": 1.9823443888171676e-05, + "loss": 0.5451, + "step": 1684 + }, + { + "epoch": 0.71, + "learning_rate": 1.982312656728986e-05, + "loss": 0.4172, + "step": 1685 + }, + { + "epoch": 0.71, + "learning_rate": 1.9822808964050987e-05, + "loss": 0.5163, + "step": 1686 + }, + { + "epoch": 0.71, + "learning_rate": 1.982249107846419e-05, + "loss": 0.5484, + "step": 1687 + }, + { + "epoch": 0.71, + "learning_rate": 1.98221729105386e-05, + "loss": 0.5193, + "step": 1688 + }, + { + "epoch": 0.71, + "learning_rate": 1.982185446028337e-05, + "loss": 0.483, + "step": 1689 + }, + { + "epoch": 0.71, + "learning_rate": 1.9821535727707647e-05, + "loss": 0.519, + "step": 1690 + }, + { + "epoch": 0.71, + "learning_rate": 1.9821216712820597e-05, + "loss": 0.5161, + "step": 1691 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820897415631383e-05, + "loss": 0.5653, + "step": 1692 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820577836149192e-05, + "loss": 0.4888, + "step": 1693 + }, + { + "epoch": 0.71, + "learning_rate": 1.9820257974383205e-05, + "loss": 0.4977, + "step": 1694 + }, + { + "epoch": 0.71, + "learning_rate": 1.9819937830342618e-05, + "loss": 0.4693, + "step": 1695 + }, + { + "epoch": 0.71, + "learning_rate": 1.981961740403663e-05, + "loss": 0.493, + "step": 1696 + }, + { + "epoch": 0.71, + "learning_rate": 1.9819296695474454e-05, + "loss": 0.5228, + "step": 1697 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818975704665306e-05, + "loss": 0.4938, + "step": 1698 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818654431618416e-05, + "loss": 0.4252, + "step": 1699 + }, + { + "epoch": 0.71, + "learning_rate": 1.981833287634302e-05, + "loss": 0.4584, + "step": 1700 + }, + { + "epoch": 0.71, + "learning_rate": 1.9818011038848355e-05, + "loss": 0.5541, + "step": 1701 + }, + { + "epoch": 0.71, + "learning_rate": 1.9817688919143676e-05, + "loss": 0.412, + "step": 1702 + }, + { + "epoch": 0.71, + "learning_rate": 1.981736651723824e-05, + "loss": 0.5411, + "step": 1703 + }, + { + "epoch": 0.71, + "learning_rate": 1.9817043833141317e-05, + "loss": 0.4207, + "step": 1704 + }, + { + "epoch": 0.71, + "learning_rate": 1.981672086686218e-05, + "loss": 0.502, + "step": 1705 + }, + { + "epoch": 0.71, + "learning_rate": 1.9816397618410112e-05, + "loss": 0.5266, + "step": 1706 + }, + { + "epoch": 0.71, + "learning_rate": 1.9816074087794406e-05, + "loss": 0.5749, + "step": 1707 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815750275024362e-05, + "loss": 0.5609, + "step": 1708 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815426180109286e-05, + "loss": 0.5089, + "step": 1709 + }, + { + "epoch": 0.72, + "learning_rate": 1.9815101803058492e-05, + "loss": 0.5283, + "step": 1710 + }, + { + "epoch": 0.72, + "learning_rate": 1.981477714388131e-05, + "loss": 0.4316, + "step": 1711 + }, + { + "epoch": 0.72, + "learning_rate": 1.9814452202587066e-05, + "loss": 0.4375, + "step": 1712 + }, + { + "epoch": 0.72, + "learning_rate": 1.9814126979185105e-05, + "loss": 0.5121, + "step": 1713 + }, + { + "epoch": 0.72, + "learning_rate": 1.981380147368477e-05, + "loss": 0.4719, + "step": 1714 + }, + { + "epoch": 0.72, + "learning_rate": 1.981347568609542e-05, + "loss": 0.4747, + "step": 1715 + }, + { + "epoch": 0.72, + "learning_rate": 1.9813149616426424e-05, + "loss": 0.5267, + "step": 1716 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812823264687148e-05, + "loss": 0.4687, + "step": 1717 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812496630886974e-05, + "loss": 0.5243, + "step": 1718 + }, + { + "epoch": 0.72, + "learning_rate": 1.9812169715035295e-05, + "loss": 0.5228, + "step": 1719 + }, + { + "epoch": 0.72, + "learning_rate": 1.98118425171415e-05, + "loss": 0.5324, + "step": 1720 + }, + { + "epoch": 0.72, + "learning_rate": 1.9811515037215002e-05, + "loss": 0.5004, + "step": 1721 + }, + { + "epoch": 0.72, + "learning_rate": 1.981118727526521e-05, + "loss": 0.4853, + "step": 1722 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810859231301545e-05, + "loss": 0.551, + "step": 1723 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810530905333437e-05, + "loss": 0.4566, + "step": 1724 + }, + { + "epoch": 0.72, + "learning_rate": 1.9810202297370328e-05, + "loss": 0.4481, + "step": 1725 + }, + { + "epoch": 0.72, + "learning_rate": 1.9809873407421656e-05, + "loss": 0.5405, + "step": 1726 + }, + { + "epoch": 0.72, + "learning_rate": 1.980954423549688e-05, + "loss": 0.5531, + "step": 1727 + }, + { + "epoch": 0.72, + "learning_rate": 1.9809214781605458e-05, + "loss": 0.5072, + "step": 1728 + }, + { + "epoch": 0.72, + "learning_rate": 1.980888504575686e-05, + "loss": 0.6357, + "step": 1729 + }, + { + "epoch": 0.72, + "learning_rate": 1.980855502796057e-05, + "loss": 0.523, + "step": 1730 + }, + { + "epoch": 0.72, + "learning_rate": 1.9808224728226067e-05, + "loss": 0.4832, + "step": 1731 + }, + { + "epoch": 0.73, + "learning_rate": 1.9807894146562847e-05, + "loss": 0.4624, + "step": 1732 + }, + { + "epoch": 0.73, + "learning_rate": 1.9807563282980415e-05, + "loss": 0.444, + "step": 1733 + }, + { + "epoch": 0.73, + "learning_rate": 1.980723213748828e-05, + "loss": 0.485, + "step": 1734 + }, + { + "epoch": 0.73, + "learning_rate": 1.9806900710095958e-05, + "loss": 0.6331, + "step": 1735 + }, + { + "epoch": 0.73, + "learning_rate": 1.980656900081298e-05, + "loss": 0.551, + "step": 1736 + }, + { + "epoch": 0.73, + "learning_rate": 1.9806237009648876e-05, + "loss": 0.4904, + "step": 1737 + }, + { + "epoch": 0.73, + "learning_rate": 1.980590473661319e-05, + "loss": 0.5215, + "step": 1738 + }, + { + "epoch": 0.73, + "learning_rate": 1.9805572181715478e-05, + "loss": 0.5491, + "step": 1739 + }, + { + "epoch": 0.73, + "learning_rate": 1.980523934496529e-05, + "loss": 0.5062, + "step": 1740 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804906226372202e-05, + "loss": 0.5369, + "step": 1741 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804572825945784e-05, + "loss": 0.4987, + "step": 1742 + }, + { + "epoch": 0.73, + "learning_rate": 1.9804239143695622e-05, + "loss": 0.4956, + "step": 1743 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803905179631303e-05, + "loss": 0.4876, + "step": 1744 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803570933762432e-05, + "loss": 0.5259, + "step": 1745 + }, + { + "epoch": 0.73, + "learning_rate": 1.9803236406098612e-05, + "loss": 0.4896, + "step": 1746 + }, + { + "epoch": 0.73, + "learning_rate": 1.9802901596649462e-05, + "loss": 0.4446, + "step": 1747 + }, + { + "epoch": 0.73, + "learning_rate": 1.9802566505424603e-05, + "loss": 0.5677, + "step": 1748 + }, + { + "epoch": 0.73, + "learning_rate": 1.980223113243367e-05, + "loss": 0.5355, + "step": 1749 + }, + { + "epoch": 0.73, + "learning_rate": 1.98018954776863e-05, + "loss": 0.4811, + "step": 1750 + }, + { + "epoch": 0.73, + "learning_rate": 1.9801559541192147e-05, + "loss": 0.4993, + "step": 1751 + }, + { + "epoch": 0.73, + "learning_rate": 1.9801223322960858e-05, + "loss": 0.5684, + "step": 1752 + }, + { + "epoch": 0.73, + "learning_rate": 1.9800886823002105e-05, + "loss": 0.5324, + "step": 1753 + }, + { + "epoch": 0.73, + "learning_rate": 1.9800550041325553e-05, + "loss": 0.5637, + "step": 1754 + }, + { + "epoch": 0.73, + "learning_rate": 1.980021297794089e-05, + "loss": 0.5459, + "step": 1755 + }, + { + "epoch": 0.74, + "learning_rate": 1.9799875632857802e-05, + "loss": 0.5847, + "step": 1756 + }, + { + "epoch": 0.74, + "learning_rate": 1.9799538006085982e-05, + "loss": 0.5407, + "step": 1757 + }, + { + "epoch": 0.74, + "learning_rate": 1.979920009763514e-05, + "loss": 0.4979, + "step": 1758 + }, + { + "epoch": 0.74, + "learning_rate": 1.979886190751499e-05, + "loss": 0.438, + "step": 1759 + }, + { + "epoch": 0.74, + "learning_rate": 1.9798523435735246e-05, + "loss": 0.5051, + "step": 1760 + }, + { + "epoch": 0.74, + "learning_rate": 1.9798184682305645e-05, + "loss": 0.4852, + "step": 1761 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797845647235915e-05, + "loss": 0.5432, + "step": 1762 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797506330535808e-05, + "loss": 0.4505, + "step": 1763 + }, + { + "epoch": 0.74, + "learning_rate": 1.9797166732215078e-05, + "loss": 0.6174, + "step": 1764 + }, + { + "epoch": 0.74, + "learning_rate": 1.979682685228348e-05, + "loss": 0.4877, + "step": 1765 + }, + { + "epoch": 0.74, + "learning_rate": 1.9796486690750792e-05, + "loss": 0.515, + "step": 1766 + }, + { + "epoch": 0.74, + "learning_rate": 1.9796146247626783e-05, + "loss": 0.4858, + "step": 1767 + }, + { + "epoch": 0.74, + "learning_rate": 1.9795805522921246e-05, + "loss": 0.5304, + "step": 1768 + }, + { + "epoch": 0.74, + "learning_rate": 1.979546451664397e-05, + "loss": 0.5342, + "step": 1769 + }, + { + "epoch": 0.74, + "learning_rate": 1.979512322880476e-05, + "loss": 0.5117, + "step": 1770 + }, + { + "epoch": 0.74, + "learning_rate": 1.9794781659413425e-05, + "loss": 0.5361, + "step": 1771 + }, + { + "epoch": 0.74, + "learning_rate": 1.9794439808479783e-05, + "loss": 0.4229, + "step": 1772 + }, + { + "epoch": 0.74, + "learning_rate": 1.979409767601366e-05, + "loss": 0.413, + "step": 1773 + }, + { + "epoch": 0.74, + "learning_rate": 1.979375526202489e-05, + "loss": 0.464, + "step": 1774 + }, + { + "epoch": 0.74, + "learning_rate": 1.9793412566523318e-05, + "loss": 0.5194, + "step": 1775 + }, + { + "epoch": 0.74, + "learning_rate": 1.9793069589518787e-05, + "loss": 0.5144, + "step": 1776 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792726331021166e-05, + "loss": 0.451, + "step": 1777 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792382791040317e-05, + "loss": 0.5394, + "step": 1778 + }, + { + "epoch": 0.74, + "learning_rate": 1.9792038969586113e-05, + "loss": 0.5391, + "step": 1779 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791694866668434e-05, + "loss": 0.5101, + "step": 1780 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791350482297177e-05, + "loss": 0.4901, + "step": 1781 + }, + { + "epoch": 0.75, + "learning_rate": 1.9791005816482242e-05, + "loss": 0.4868, + "step": 1782 + }, + { + "epoch": 0.75, + "learning_rate": 1.9790660869233532e-05, + "loss": 0.4358, + "step": 1783 + }, + { + "epoch": 0.75, + "learning_rate": 1.9790315640560962e-05, + "loss": 0.5186, + "step": 1784 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789970130474455e-05, + "loss": 0.5941, + "step": 1785 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789624338983948e-05, + "loss": 0.4653, + "step": 1786 + }, + { + "epoch": 0.75, + "learning_rate": 1.9789278266099375e-05, + "loss": 0.4051, + "step": 1787 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788931911830688e-05, + "loss": 0.4873, + "step": 1788 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788585276187836e-05, + "loss": 0.5416, + "step": 1789 + }, + { + "epoch": 0.75, + "learning_rate": 1.9788238359180786e-05, + "loss": 0.6702, + "step": 1790 + }, + { + "epoch": 0.75, + "learning_rate": 1.978789116081951e-05, + "loss": 0.5041, + "step": 1791 + }, + { + "epoch": 0.75, + "learning_rate": 1.9787543681113993e-05, + "loss": 0.4556, + "step": 1792 + }, + { + "epoch": 0.75, + "learning_rate": 1.9787195920074213e-05, + "loss": 0.5527, + "step": 1793 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786847877710177e-05, + "loss": 0.5213, + "step": 1794 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786499554031876e-05, + "loss": 0.5291, + "step": 1795 + }, + { + "epoch": 0.75, + "learning_rate": 1.9786150949049335e-05, + "loss": 0.4473, + "step": 1796 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785802062772566e-05, + "loss": 0.5462, + "step": 1797 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785452895211606e-05, + "loss": 0.5014, + "step": 1798 + }, + { + "epoch": 0.75, + "learning_rate": 1.9785103446376483e-05, + "loss": 0.5301, + "step": 1799 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784753716277242e-05, + "loss": 0.5775, + "step": 1800 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784403704923937e-05, + "loss": 0.4802, + "step": 1801 + }, + { + "epoch": 0.75, + "learning_rate": 1.9784053412326637e-05, + "loss": 0.4719, + "step": 1802 + }, + { + "epoch": 0.75, + "learning_rate": 1.97837028384954e-05, + "loss": 0.5638, + "step": 1803 + }, + { + "epoch": 0.76, + "learning_rate": 1.9783351983440306e-05, + "loss": 0.5138, + "step": 1804 + }, + { + "epoch": 0.76, + "learning_rate": 1.978300084717144e-05, + "loss": 0.4597, + "step": 1805 + }, + { + "epoch": 0.76, + "learning_rate": 1.9782649429698896e-05, + "loss": 0.4601, + "step": 1806 + }, + { + "epoch": 0.76, + "learning_rate": 1.9782297731032778e-05, + "loss": 0.5567, + "step": 1807 + }, + { + "epoch": 0.76, + "learning_rate": 1.978194575118319e-05, + "loss": 0.5288, + "step": 1808 + }, + { + "epoch": 0.76, + "learning_rate": 1.9781593490160255e-05, + "loss": 0.5387, + "step": 1809 + }, + { + "epoch": 0.76, + "learning_rate": 1.9781240947974093e-05, + "loss": 0.4789, + "step": 1810 + }, + { + "epoch": 0.76, + "learning_rate": 1.978088812463484e-05, + "loss": 0.4638, + "step": 1811 + }, + { + "epoch": 0.76, + "learning_rate": 1.9780535020152632e-05, + "loss": 0.5389, + "step": 1812 + }, + { + "epoch": 0.76, + "learning_rate": 1.978018163453763e-05, + "loss": 0.4942, + "step": 1813 + }, + { + "epoch": 0.76, + "learning_rate": 1.9779827967799985e-05, + "loss": 0.4926, + "step": 1814 + }, + { + "epoch": 0.76, + "learning_rate": 1.9779474019949864e-05, + "loss": 0.5134, + "step": 1815 + }, + { + "epoch": 0.76, + "learning_rate": 1.977911979099744e-05, + "loss": 0.5047, + "step": 1816 + }, + { + "epoch": 0.76, + "learning_rate": 1.9778765280952896e-05, + "loss": 0.492, + "step": 1817 + }, + { + "epoch": 0.76, + "learning_rate": 1.977841048982642e-05, + "loss": 0.5855, + "step": 1818 + }, + { + "epoch": 0.76, + "learning_rate": 1.977805541762821e-05, + "loss": 0.487, + "step": 1819 + }, + { + "epoch": 0.76, + "learning_rate": 1.9777700064368475e-05, + "loss": 0.562, + "step": 1820 + }, + { + "epoch": 0.76, + "learning_rate": 1.977734443005743e-05, + "loss": 0.4709, + "step": 1821 + }, + { + "epoch": 0.76, + "learning_rate": 1.9776988514705294e-05, + "loss": 0.5106, + "step": 1822 + }, + { + "epoch": 0.76, + "learning_rate": 1.97766323183223e-05, + "loss": 0.5815, + "step": 1823 + }, + { + "epoch": 0.76, + "learning_rate": 1.9776275840918685e-05, + "loss": 0.4957, + "step": 1824 + }, + { + "epoch": 0.76, + "learning_rate": 1.97759190825047e-05, + "loss": 0.4592, + "step": 1825 + }, + { + "epoch": 0.76, + "learning_rate": 1.977556204309059e-05, + "loss": 0.5652, + "step": 1826 + }, + { + "epoch": 0.76, + "learning_rate": 1.9775204722686628e-05, + "loss": 0.4936, + "step": 1827 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774847121303078e-05, + "loss": 0.5499, + "step": 1828 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774489238950222e-05, + "loss": 0.5595, + "step": 1829 + }, + { + "epoch": 0.77, + "learning_rate": 1.9774131075638347e-05, + "loss": 0.5254, + "step": 1830 + }, + { + "epoch": 0.77, + "learning_rate": 1.977377263137775e-05, + "loss": 0.5087, + "step": 1831 + }, + { + "epoch": 0.77, + "learning_rate": 1.977341390617873e-05, + "loss": 0.4864, + "step": 1832 + }, + { + "epoch": 0.77, + "learning_rate": 1.97730549000516e-05, + "loss": 0.5294, + "step": 1833 + }, + { + "epoch": 0.77, + "learning_rate": 1.9772695613006676e-05, + "loss": 0.4775, + "step": 1834 + }, + { + "epoch": 0.77, + "learning_rate": 1.977233604505429e-05, + "loss": 0.5719, + "step": 1835 + }, + { + "epoch": 0.77, + "learning_rate": 1.977197619620478e-05, + "loss": 0.5053, + "step": 1836 + }, + { + "epoch": 0.77, + "learning_rate": 1.9771616066468484e-05, + "loss": 0.4699, + "step": 1837 + }, + { + "epoch": 0.77, + "learning_rate": 1.9771255655855753e-05, + "loss": 0.5384, + "step": 1838 + }, + { + "epoch": 0.77, + "learning_rate": 1.9770894964376956e-05, + "loss": 0.5049, + "step": 1839 + }, + { + "epoch": 0.77, + "learning_rate": 1.9770533992042447e-05, + "loss": 0.5714, + "step": 1840 + }, + { + "epoch": 0.77, + "learning_rate": 1.977017273886261e-05, + "loss": 0.5266, + "step": 1841 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769811204847825e-05, + "loss": 0.5479, + "step": 1842 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769449390008492e-05, + "loss": 0.5449, + "step": 1843 + }, + { + "epoch": 0.77, + "learning_rate": 1.9769087294355006e-05, + "loss": 0.4606, + "step": 1844 + }, + { + "epoch": 0.77, + "learning_rate": 1.976872491789777e-05, + "loss": 0.5291, + "step": 1845 + }, + { + "epoch": 0.77, + "learning_rate": 1.9768362260647205e-05, + "loss": 0.5553, + "step": 1846 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767999322613737e-05, + "loss": 0.5144, + "step": 1847 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767636103807796e-05, + "loss": 0.5458, + "step": 1848 + }, + { + "epoch": 0.77, + "learning_rate": 1.9767272604239823e-05, + "loss": 0.559, + "step": 1849 + }, + { + "epoch": 0.77, + "learning_rate": 1.9766908823920268e-05, + "loss": 0.4688, + "step": 1850 + }, + { + "epoch": 0.77, + "learning_rate": 1.976654476285958e-05, + "loss": 0.558, + "step": 1851 + }, + { + "epoch": 0.78, + "learning_rate": 1.9766180421068236e-05, + "loss": 0.5534, + "step": 1852 + }, + { + "epoch": 0.78, + "learning_rate": 1.97658157985567e-05, + "loss": 0.469, + "step": 1853 + }, + { + "epoch": 0.78, + "learning_rate": 1.9765450895335454e-05, + "loss": 0.4969, + "step": 1854 + }, + { + "epoch": 0.78, + "learning_rate": 1.9765085711414987e-05, + "loss": 0.5451, + "step": 1855 + }, + { + "epoch": 0.78, + "learning_rate": 1.97647202468058e-05, + "loss": 0.5496, + "step": 1856 + }, + { + "epoch": 0.78, + "learning_rate": 1.9764354501518394e-05, + "loss": 0.4641, + "step": 1857 + }, + { + "epoch": 0.78, + "learning_rate": 1.976398847556328e-05, + "loss": 0.513, + "step": 1858 + }, + { + "epoch": 0.78, + "learning_rate": 1.9763622168950982e-05, + "loss": 0.6003, + "step": 1859 + }, + { + "epoch": 0.78, + "learning_rate": 1.9763255581692027e-05, + "loss": 0.5382, + "step": 1860 + }, + { + "epoch": 0.78, + "learning_rate": 1.976288871379696e-05, + "loss": 0.5244, + "step": 1861 + }, + { + "epoch": 0.78, + "learning_rate": 1.9762521565276314e-05, + "loss": 0.5568, + "step": 1862 + }, + { + "epoch": 0.78, + "learning_rate": 1.976215413614065e-05, + "loss": 0.509, + "step": 1863 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761786426400532e-05, + "loss": 0.5482, + "step": 1864 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761418436066523e-05, + "loss": 0.5047, + "step": 1865 + }, + { + "epoch": 0.78, + "learning_rate": 1.9761050165149208e-05, + "loss": 0.4403, + "step": 1866 + }, + { + "epoch": 0.78, + "learning_rate": 1.976068161365916e-05, + "loss": 0.509, + "step": 1867 + }, + { + "epoch": 0.78, + "learning_rate": 1.9760312781606985e-05, + "loss": 0.5031, + "step": 1868 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759943669003282e-05, + "loss": 0.5258, + "step": 1869 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759574275858653e-05, + "loss": 0.5036, + "step": 1870 + }, + { + "epoch": 0.78, + "learning_rate": 1.9759204602183725e-05, + "loss": 0.4541, + "step": 1871 + }, + { + "epoch": 0.78, + "learning_rate": 1.9758834647989123e-05, + "loss": 0.4063, + "step": 1872 + }, + { + "epoch": 0.78, + "learning_rate": 1.9758464413285477e-05, + "loss": 0.5494, + "step": 1873 + }, + { + "epoch": 0.78, + "learning_rate": 1.975809389808343e-05, + "loss": 0.5205, + "step": 1874 + }, + { + "epoch": 0.79, + "learning_rate": 1.9757723102393634e-05, + "loss": 0.4654, + "step": 1875 + }, + { + "epoch": 0.79, + "learning_rate": 1.9757352026226746e-05, + "loss": 0.4682, + "step": 1876 + }, + { + "epoch": 0.79, + "learning_rate": 1.9756980669593432e-05, + "loss": 0.5617, + "step": 1877 + }, + { + "epoch": 0.79, + "learning_rate": 1.975660903250437e-05, + "loss": 0.4518, + "step": 1878 + }, + { + "epoch": 0.79, + "learning_rate": 1.9756237114970235e-05, + "loss": 0.4759, + "step": 1879 + }, + { + "epoch": 0.79, + "learning_rate": 1.9755864917001724e-05, + "loss": 0.6428, + "step": 1880 + }, + { + "epoch": 0.79, + "learning_rate": 1.9755492438609532e-05, + "loss": 0.5162, + "step": 1881 + }, + { + "epoch": 0.79, + "learning_rate": 1.975511967980437e-05, + "loss": 0.4909, + "step": 1882 + }, + { + "epoch": 0.79, + "learning_rate": 1.9754746640596945e-05, + "loss": 0.4749, + "step": 1883 + }, + { + "epoch": 0.79, + "learning_rate": 1.9754373320997987e-05, + "loss": 0.4545, + "step": 1884 + }, + { + "epoch": 0.79, + "learning_rate": 1.975399972101822e-05, + "loss": 0.5069, + "step": 1885 + }, + { + "epoch": 0.79, + "learning_rate": 1.975362584066839e-05, + "loss": 0.4973, + "step": 1886 + }, + { + "epoch": 0.79, + "learning_rate": 1.975325167995924e-05, + "loss": 0.4698, + "step": 1887 + }, + { + "epoch": 0.79, + "learning_rate": 1.9752877238901525e-05, + "loss": 0.4487, + "step": 1888 + }, + { + "epoch": 0.79, + "learning_rate": 1.975250251750601e-05, + "loss": 0.5072, + "step": 1889 + }, + { + "epoch": 0.79, + "learning_rate": 1.9752127515783458e-05, + "loss": 0.5311, + "step": 1890 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751752233744662e-05, + "loss": 0.4764, + "step": 1891 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751376671400398e-05, + "loss": 0.4509, + "step": 1892 + }, + { + "epoch": 0.79, + "learning_rate": 1.9751000828761465e-05, + "loss": 0.5794, + "step": 1893 + }, + { + "epoch": 0.79, + "learning_rate": 1.9750624705838666e-05, + "loss": 0.4593, + "step": 1894 + }, + { + "epoch": 0.79, + "learning_rate": 1.9750248302642813e-05, + "loss": 0.472, + "step": 1895 + }, + { + "epoch": 0.79, + "learning_rate": 1.9749871619184726e-05, + "loss": 0.4706, + "step": 1896 + }, + { + "epoch": 0.79, + "learning_rate": 1.974949465547523e-05, + "loss": 0.5228, + "step": 1897 + }, + { + "epoch": 0.79, + "learning_rate": 1.974911741152516e-05, + "loss": 0.5407, + "step": 1898 + }, + { + "epoch": 0.8, + "learning_rate": 1.9748739887345362e-05, + "loss": 0.6125, + "step": 1899 + }, + { + "epoch": 0.8, + "learning_rate": 1.974836208294669e-05, + "loss": 0.4526, + "step": 1900 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747983998339996e-05, + "loss": 0.4489, + "step": 1901 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747605633536152e-05, + "loss": 0.4619, + "step": 1902 + }, + { + "epoch": 0.8, + "learning_rate": 1.9747226988546038e-05, + "loss": 0.5118, + "step": 1903 + }, + { + "epoch": 0.8, + "learning_rate": 1.974684806338053e-05, + "loss": 0.5502, + "step": 1904 + }, + { + "epoch": 0.8, + "learning_rate": 1.9746468858050526e-05, + "loss": 0.5278, + "step": 1905 + }, + { + "epoch": 0.8, + "learning_rate": 1.9746089372566924e-05, + "loss": 0.4562, + "step": 1906 + }, + { + "epoch": 0.8, + "learning_rate": 1.974570960694063e-05, + "loss": 0.5074, + "step": 1907 + }, + { + "epoch": 0.8, + "learning_rate": 1.974532956118256e-05, + "loss": 0.4635, + "step": 1908 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744949235303644e-05, + "loss": 0.5443, + "step": 1909 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744568629314807e-05, + "loss": 0.4998, + "step": 1910 + }, + { + "epoch": 0.8, + "learning_rate": 1.9744187743226994e-05, + "loss": 0.4901, + "step": 1911 + }, + { + "epoch": 0.8, + "learning_rate": 1.974380657705115e-05, + "loss": 0.5269, + "step": 1912 + }, + { + "epoch": 0.8, + "learning_rate": 1.974342513079823e-05, + "loss": 0.4695, + "step": 1913 + }, + { + "epoch": 0.8, + "learning_rate": 1.9743043404479203e-05, + "loss": 0.4963, + "step": 1914 + }, + { + "epoch": 0.8, + "learning_rate": 1.9742661398105038e-05, + "loss": 0.5258, + "step": 1915 + }, + { + "epoch": 0.8, + "learning_rate": 1.9742279111686717e-05, + "loss": 0.4798, + "step": 1916 + }, + { + "epoch": 0.8, + "learning_rate": 1.974189654523523e-05, + "loss": 0.5146, + "step": 1917 + }, + { + "epoch": 0.8, + "learning_rate": 1.974151369876157e-05, + "loss": 0.4735, + "step": 1918 + }, + { + "epoch": 0.8, + "learning_rate": 1.9741130572276742e-05, + "loss": 0.5074, + "step": 1919 + }, + { + "epoch": 0.8, + "learning_rate": 1.9740747165791762e-05, + "loss": 0.5401, + "step": 1920 + }, + { + "epoch": 0.8, + "learning_rate": 1.974036347931765e-05, + "loss": 0.5378, + "step": 1921 + }, + { + "epoch": 0.8, + "learning_rate": 1.973997951286543e-05, + "loss": 0.5635, + "step": 1922 + }, + { + "epoch": 0.81, + "learning_rate": 1.9739595266446143e-05, + "loss": 0.6037, + "step": 1923 + }, + { + "epoch": 0.81, + "learning_rate": 1.9739210740070833e-05, + "loss": 0.4684, + "step": 1924 + }, + { + "epoch": 0.81, + "learning_rate": 1.973882593375055e-05, + "loss": 0.48, + "step": 1925 + }, + { + "epoch": 0.81, + "learning_rate": 1.973844084749636e-05, + "loss": 0.5285, + "step": 1926 + }, + { + "epoch": 0.81, + "learning_rate": 1.973805548131933e-05, + "loss": 0.4726, + "step": 1927 + }, + { + "epoch": 0.81, + "learning_rate": 1.9737669835230537e-05, + "loss": 0.5762, + "step": 1928 + }, + { + "epoch": 0.81, + "learning_rate": 1.9737283909241065e-05, + "loss": 0.5675, + "step": 1929 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736897703362006e-05, + "loss": 0.5074, + "step": 1930 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736511217604465e-05, + "loss": 0.4967, + "step": 1931 + }, + { + "epoch": 0.81, + "learning_rate": 1.9736124451979547e-05, + "loss": 0.5085, + "step": 1932 + }, + { + "epoch": 0.81, + "learning_rate": 1.9735737406498375e-05, + "loss": 0.5721, + "step": 1933 + }, + { + "epoch": 0.81, + "learning_rate": 1.973535008117207e-05, + "loss": 0.5391, + "step": 1934 + }, + { + "epoch": 0.81, + "learning_rate": 1.9734962476011763e-05, + "loss": 0.4891, + "step": 1935 + }, + { + "epoch": 0.81, + "learning_rate": 1.97345745910286e-05, + "loss": 0.4809, + "step": 1936 + }, + { + "epoch": 0.81, + "learning_rate": 1.9734186426233728e-05, + "loss": 0.4708, + "step": 1937 + }, + { + "epoch": 0.81, + "learning_rate": 1.9733797981638306e-05, + "loss": 0.4815, + "step": 1938 + }, + { + "epoch": 0.81, + "learning_rate": 1.97334092572535e-05, + "loss": 0.5476, + "step": 1939 + }, + { + "epoch": 0.81, + "learning_rate": 1.9733020253090477e-05, + "loss": 0.5123, + "step": 1940 + }, + { + "epoch": 0.81, + "learning_rate": 1.9732630969160428e-05, + "loss": 0.5376, + "step": 1941 + }, + { + "epoch": 0.81, + "learning_rate": 1.9732241405474536e-05, + "loss": 0.5071, + "step": 1942 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731851562044002e-05, + "loss": 0.4348, + "step": 1943 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731461438880034e-05, + "loss": 0.4937, + "step": 1944 + }, + { + "epoch": 0.81, + "learning_rate": 1.9731071035993837e-05, + "loss": 0.5239, + "step": 1945 + }, + { + "epoch": 0.81, + "learning_rate": 1.9730680353396643e-05, + "loss": 0.497, + "step": 1946 + }, + { + "epoch": 0.82, + "learning_rate": 1.9730289391099675e-05, + "loss": 0.563, + "step": 1947 + }, + { + "epoch": 0.82, + "learning_rate": 1.9729898149114174e-05, + "loss": 0.6107, + "step": 1948 + }, + { + "epoch": 0.82, + "learning_rate": 1.9729506627451383e-05, + "loss": 0.5381, + "step": 1949 + }, + { + "epoch": 0.82, + "learning_rate": 1.972911482612256e-05, + "loss": 0.5316, + "step": 1950 + }, + { + "epoch": 0.82, + "learning_rate": 1.972872274513896e-05, + "loss": 0.5669, + "step": 1951 + }, + { + "epoch": 0.82, + "learning_rate": 1.972833038451186e-05, + "loss": 0.5279, + "step": 1952 + }, + { + "epoch": 0.82, + "learning_rate": 1.972793774425254e-05, + "loss": 0.485, + "step": 1953 + }, + { + "epoch": 0.82, + "learning_rate": 1.972754482437228e-05, + "loss": 0.4867, + "step": 1954 + }, + { + "epoch": 0.82, + "learning_rate": 1.972715162488237e-05, + "loss": 0.4395, + "step": 1955 + }, + { + "epoch": 0.82, + "learning_rate": 1.9726758145794127e-05, + "loss": 0.5581, + "step": 1956 + }, + { + "epoch": 0.82, + "learning_rate": 1.9726364387118847e-05, + "loss": 0.4973, + "step": 1957 + }, + { + "epoch": 0.82, + "learning_rate": 1.9725970348867858e-05, + "loss": 0.5836, + "step": 1958 + }, + { + "epoch": 0.82, + "learning_rate": 1.972557603105248e-05, + "loss": 0.461, + "step": 1959 + }, + { + "epoch": 0.82, + "learning_rate": 1.9725181433684047e-05, + "loss": 0.5794, + "step": 1960 + }, + { + "epoch": 0.82, + "learning_rate": 1.9724786556773906e-05, + "loss": 0.4638, + "step": 1961 + }, + { + "epoch": 0.82, + "learning_rate": 1.9724391400333405e-05, + "loss": 0.4794, + "step": 1962 + }, + { + "epoch": 0.82, + "learning_rate": 1.97239959643739e-05, + "loss": 0.4787, + "step": 1963 + }, + { + "epoch": 0.82, + "learning_rate": 1.9723600248906766e-05, + "loss": 0.4904, + "step": 1964 + }, + { + "epoch": 0.82, + "learning_rate": 1.9723204253943366e-05, + "loss": 0.482, + "step": 1965 + }, + { + "epoch": 0.82, + "learning_rate": 1.972280797949509e-05, + "loss": 0.5006, + "step": 1966 + }, + { + "epoch": 0.82, + "learning_rate": 1.9722411425573323e-05, + "loss": 0.506, + "step": 1967 + }, + { + "epoch": 0.82, + "learning_rate": 1.9722014592189472e-05, + "loss": 0.5763, + "step": 1968 + }, + { + "epoch": 0.82, + "learning_rate": 1.9721617479354936e-05, + "loss": 0.5273, + "step": 1969 + }, + { + "epoch": 0.82, + "learning_rate": 1.972122008708113e-05, + "loss": 0.4863, + "step": 1970 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720822415379485e-05, + "loss": 0.4972, + "step": 1971 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720424464261426e-05, + "loss": 0.5359, + "step": 1972 + }, + { + "epoch": 0.83, + "learning_rate": 1.9720026233738388e-05, + "loss": 0.4415, + "step": 1973 + }, + { + "epoch": 0.83, + "learning_rate": 1.9719627723821818e-05, + "loss": 0.5304, + "step": 1974 + }, + { + "epoch": 0.83, + "learning_rate": 1.971922893452318e-05, + "loss": 0.489, + "step": 1975 + }, + { + "epoch": 0.83, + "learning_rate": 1.971882986585393e-05, + "loss": 0.49, + "step": 1976 + }, + { + "epoch": 0.83, + "learning_rate": 1.9718430517825538e-05, + "loss": 0.5563, + "step": 1977 + }, + { + "epoch": 0.83, + "learning_rate": 1.9718030890449483e-05, + "loss": 0.5101, + "step": 1978 + }, + { + "epoch": 0.83, + "learning_rate": 1.9717630983737256e-05, + "loss": 0.5219, + "step": 1979 + }, + { + "epoch": 0.83, + "learning_rate": 1.971723079770035e-05, + "loss": 0.4326, + "step": 1980 + }, + { + "epoch": 0.83, + "learning_rate": 1.971683033235027e-05, + "loss": 0.4955, + "step": 1981 + }, + { + "epoch": 0.83, + "learning_rate": 1.971642958769852e-05, + "loss": 0.5168, + "step": 1982 + }, + { + "epoch": 0.83, + "learning_rate": 1.9716028563756628e-05, + "loss": 0.608, + "step": 1983 + }, + { + "epoch": 0.83, + "learning_rate": 1.971562726053611e-05, + "loss": 0.4624, + "step": 1984 + }, + { + "epoch": 0.83, + "learning_rate": 1.9715225678048515e-05, + "loss": 0.4922, + "step": 1985 + }, + { + "epoch": 0.83, + "learning_rate": 1.9714823816305376e-05, + "loss": 0.6225, + "step": 1986 + }, + { + "epoch": 0.83, + "learning_rate": 1.9714421675318247e-05, + "loss": 0.4787, + "step": 1987 + }, + { + "epoch": 0.83, + "learning_rate": 1.971401925509869e-05, + "loss": 0.4666, + "step": 1988 + }, + { + "epoch": 0.83, + "learning_rate": 1.9713616555658266e-05, + "loss": 0.5071, + "step": 1989 + }, + { + "epoch": 0.83, + "learning_rate": 1.9713213577008556e-05, + "loss": 0.4458, + "step": 1990 + }, + { + "epoch": 0.83, + "learning_rate": 1.971281031916114e-05, + "loss": 0.4941, + "step": 1991 + }, + { + "epoch": 0.83, + "learning_rate": 1.9712406782127612e-05, + "loss": 0.4653, + "step": 1992 + }, + { + "epoch": 0.83, + "learning_rate": 1.9712002965919567e-05, + "loss": 0.5734, + "step": 1993 + }, + { + "epoch": 0.83, + "learning_rate": 1.971159887054862e-05, + "loss": 0.535, + "step": 1994 + }, + { + "epoch": 0.84, + "learning_rate": 1.9711194496026372e-05, + "loss": 0.5193, + "step": 1995 + }, + { + "epoch": 0.84, + "learning_rate": 1.9710789842364465e-05, + "loss": 0.5112, + "step": 1996 + }, + { + "epoch": 0.84, + "learning_rate": 1.971038490957452e-05, + "loss": 0.4812, + "step": 1997 + }, + { + "epoch": 0.84, + "learning_rate": 1.9709979697668172e-05, + "loss": 0.4981, + "step": 1998 + }, + { + "epoch": 0.84, + "learning_rate": 1.970957420665708e-05, + "loss": 0.4752, + "step": 1999 + }, + { + "epoch": 0.84, + "learning_rate": 1.9709168436552887e-05, + "loss": 0.4864, + "step": 2000 + }, + { + "epoch": 0.84, + "learning_rate": 1.9708762387367272e-05, + "loss": 0.4548, + "step": 2001 + }, + { + "epoch": 0.84, + "learning_rate": 1.970835605911189e-05, + "loss": 0.511, + "step": 2002 + }, + { + "epoch": 0.84, + "learning_rate": 1.970794945179843e-05, + "loss": 0.5372, + "step": 2003 + }, + { + "epoch": 0.84, + "learning_rate": 1.970754256543858e-05, + "loss": 0.4788, + "step": 2004 + }, + { + "epoch": 0.84, + "learning_rate": 1.970713540004403e-05, + "loss": 0.5105, + "step": 2005 + }, + { + "epoch": 0.84, + "learning_rate": 1.970672795562649e-05, + "loss": 0.4853, + "step": 2006 + }, + { + "epoch": 0.84, + "learning_rate": 1.970632023219767e-05, + "loss": 0.5353, + "step": 2007 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705912229769285e-05, + "loss": 0.5399, + "step": 2008 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705503948353064e-05, + "loss": 0.4605, + "step": 2009 + }, + { + "epoch": 0.84, + "learning_rate": 1.9705095387960746e-05, + "loss": 0.4799, + "step": 2010 + }, + { + "epoch": 0.84, + "learning_rate": 1.9704686548604076e-05, + "loss": 0.4942, + "step": 2011 + }, + { + "epoch": 0.84, + "learning_rate": 1.9704277430294798e-05, + "loss": 0.5017, + "step": 2012 + }, + { + "epoch": 0.84, + "learning_rate": 1.970386803304468e-05, + "loss": 0.5192, + "step": 2013 + }, + { + "epoch": 0.84, + "learning_rate": 1.9703458356865484e-05, + "loss": 0.4614, + "step": 2014 + }, + { + "epoch": 0.84, + "learning_rate": 1.9703048401768993e-05, + "loss": 0.4899, + "step": 2015 + }, + { + "epoch": 0.84, + "learning_rate": 1.970263816776698e-05, + "loss": 0.4431, + "step": 2016 + }, + { + "epoch": 0.84, + "learning_rate": 1.970222765487125e-05, + "loss": 0.5404, + "step": 2017 + }, + { + "epoch": 0.84, + "learning_rate": 1.970181686309359e-05, + "loss": 0.4927, + "step": 2018 + }, + { + "epoch": 0.85, + "learning_rate": 1.9701405792445815e-05, + "loss": 0.4271, + "step": 2019 + }, + { + "epoch": 0.85, + "learning_rate": 1.970099444293974e-05, + "loss": 0.4929, + "step": 2020 + }, + { + "epoch": 0.85, + "learning_rate": 1.970058281458719e-05, + "loss": 0.4559, + "step": 2021 + }, + { + "epoch": 0.85, + "learning_rate": 1.9700170907399992e-05, + "loss": 0.5275, + "step": 2022 + }, + { + "epoch": 0.85, + "learning_rate": 1.969975872138999e-05, + "loss": 0.5423, + "step": 2023 + }, + { + "epoch": 0.85, + "learning_rate": 1.9699346256569033e-05, + "loss": 0.4918, + "step": 2024 + }, + { + "epoch": 0.85, + "learning_rate": 1.9698933512948973e-05, + "loss": 0.4484, + "step": 2025 + }, + { + "epoch": 0.85, + "learning_rate": 1.9698520490541675e-05, + "loss": 0.4909, + "step": 2026 + }, + { + "epoch": 0.85, + "learning_rate": 1.969810718935901e-05, + "loss": 0.4184, + "step": 2027 + }, + { + "epoch": 0.85, + "learning_rate": 1.9697693609412864e-05, + "loss": 0.516, + "step": 2028 + }, + { + "epoch": 0.85, + "learning_rate": 1.9697279750715118e-05, + "loss": 0.6038, + "step": 2029 + }, + { + "epoch": 0.85, + "learning_rate": 1.9696865613277673e-05, + "loss": 0.5819, + "step": 2030 + }, + { + "epoch": 0.85, + "learning_rate": 1.969645119711243e-05, + "loss": 0.512, + "step": 2031 + }, + { + "epoch": 0.85, + "learning_rate": 1.96960365022313e-05, + "loss": 0.4182, + "step": 2032 + }, + { + "epoch": 0.85, + "learning_rate": 1.969562152864621e-05, + "loss": 0.5777, + "step": 2033 + }, + { + "epoch": 0.85, + "learning_rate": 1.9695206276369077e-05, + "loss": 0.481, + "step": 2034 + }, + { + "epoch": 0.85, + "learning_rate": 1.9694790745411846e-05, + "loss": 0.4749, + "step": 2035 + }, + { + "epoch": 0.85, + "learning_rate": 1.9694374935786457e-05, + "loss": 0.4691, + "step": 2036 + }, + { + "epoch": 0.85, + "learning_rate": 1.9693958847504863e-05, + "loss": 0.5651, + "step": 2037 + }, + { + "epoch": 0.85, + "learning_rate": 1.9693542480579025e-05, + "loss": 0.4447, + "step": 2038 + }, + { + "epoch": 0.85, + "learning_rate": 1.969312583502091e-05, + "loss": 0.4686, + "step": 2039 + }, + { + "epoch": 0.85, + "learning_rate": 1.9692708910842494e-05, + "loss": 0.5077, + "step": 2040 + }, + { + "epoch": 0.85, + "learning_rate": 1.9692291708055762e-05, + "loss": 0.5205, + "step": 2041 + }, + { + "epoch": 0.85, + "learning_rate": 1.9691874226672704e-05, + "loss": 0.4826, + "step": 2042 + }, + { + "epoch": 0.86, + "learning_rate": 1.9691456466705324e-05, + "loss": 0.5561, + "step": 2043 + }, + { + "epoch": 0.86, + "learning_rate": 1.969103842816563e-05, + "loss": 0.4831, + "step": 2044 + }, + { + "epoch": 0.86, + "learning_rate": 1.969062011106563e-05, + "loss": 0.546, + "step": 2045 + }, + { + "epoch": 0.86, + "learning_rate": 1.969020151541736e-05, + "loss": 0.5363, + "step": 2046 + }, + { + "epoch": 0.86, + "learning_rate": 1.968978264123284e-05, + "loss": 0.4719, + "step": 2047 + }, + { + "epoch": 0.86, + "learning_rate": 1.968936348852412e-05, + "loss": 0.5133, + "step": 2048 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688944057303247e-05, + "loss": 0.465, + "step": 2049 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688524347582272e-05, + "loss": 0.5911, + "step": 2050 + }, + { + "epoch": 0.86, + "learning_rate": 1.9688104359373264e-05, + "loss": 0.486, + "step": 2051 + }, + { + "epoch": 0.86, + "learning_rate": 1.9687684092688292e-05, + "loss": 0.4789, + "step": 2052 + }, + { + "epoch": 0.86, + "learning_rate": 1.968726354753944e-05, + "loss": 0.4695, + "step": 2053 + }, + { + "epoch": 0.86, + "learning_rate": 1.968684272393879e-05, + "loss": 0.6171, + "step": 2054 + }, + { + "epoch": 0.86, + "learning_rate": 1.9686421621898443e-05, + "loss": 0.4325, + "step": 2055 + }, + { + "epoch": 0.86, + "learning_rate": 1.9686000241430503e-05, + "loss": 0.513, + "step": 2056 + }, + { + "epoch": 0.86, + "learning_rate": 1.968557858254708e-05, + "loss": 0.4462, + "step": 2057 + }, + { + "epoch": 0.86, + "learning_rate": 1.9685156645260296e-05, + "loss": 0.5465, + "step": 2058 + }, + { + "epoch": 0.86, + "learning_rate": 1.968473442958228e-05, + "loss": 0.4689, + "step": 2059 + }, + { + "epoch": 0.86, + "learning_rate": 1.9684311935525163e-05, + "loss": 0.5581, + "step": 2060 + }, + { + "epoch": 0.86, + "learning_rate": 1.9683889163101096e-05, + "loss": 0.4729, + "step": 2061 + }, + { + "epoch": 0.86, + "learning_rate": 1.968346611232223e-05, + "loss": 0.5558, + "step": 2062 + }, + { + "epoch": 0.86, + "learning_rate": 1.968304278320072e-05, + "loss": 0.5112, + "step": 2063 + }, + { + "epoch": 0.86, + "learning_rate": 1.968261917574874e-05, + "loss": 0.5189, + "step": 2064 + }, + { + "epoch": 0.86, + "learning_rate": 1.9682195289978466e-05, + "loss": 0.5116, + "step": 2065 + }, + { + "epoch": 0.86, + "learning_rate": 1.9681771125902075e-05, + "loss": 0.524, + "step": 2066 + }, + { + "epoch": 0.87, + "learning_rate": 1.9681346683531766e-05, + "loss": 0.5492, + "step": 2067 + }, + { + "epoch": 0.87, + "learning_rate": 1.968092196287974e-05, + "loss": 0.513, + "step": 2068 + }, + { + "epoch": 0.87, + "learning_rate": 1.96804969639582e-05, + "loss": 0.561, + "step": 2069 + }, + { + "epoch": 0.87, + "learning_rate": 1.9680071686779368e-05, + "loss": 0.5189, + "step": 2070 + }, + { + "epoch": 0.87, + "learning_rate": 1.9679646131355463e-05, + "loss": 0.5128, + "step": 2071 + }, + { + "epoch": 0.87, + "learning_rate": 1.9679220297698722e-05, + "loss": 0.45, + "step": 2072 + }, + { + "epoch": 0.87, + "learning_rate": 1.967879418582138e-05, + "loss": 0.5464, + "step": 2073 + }, + { + "epoch": 0.87, + "learning_rate": 1.967836779573569e-05, + "loss": 0.4759, + "step": 2074 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677941127453907e-05, + "loss": 0.5496, + "step": 2075 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677514180988292e-05, + "loss": 0.5275, + "step": 2076 + }, + { + "epoch": 0.87, + "learning_rate": 1.9677086956351118e-05, + "loss": 0.5698, + "step": 2077 + }, + { + "epoch": 0.87, + "learning_rate": 1.967665945355467e-05, + "loss": 0.5322, + "step": 2078 + }, + { + "epoch": 0.87, + "learning_rate": 1.9676231672611233e-05, + "loss": 0.4737, + "step": 2079 + }, + { + "epoch": 0.87, + "learning_rate": 1.96758036135331e-05, + "loss": 0.51, + "step": 2080 + }, + { + "epoch": 0.87, + "learning_rate": 1.9675375276332583e-05, + "loss": 0.4943, + "step": 2081 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674946661021984e-05, + "loss": 0.4394, + "step": 2082 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674517767613633e-05, + "loss": 0.5582, + "step": 2083 + }, + { + "epoch": 0.87, + "learning_rate": 1.9674088596119853e-05, + "loss": 0.4703, + "step": 2084 + }, + { + "epoch": 0.87, + "learning_rate": 1.967365914655298e-05, + "loss": 0.5181, + "step": 2085 + }, + { + "epoch": 0.87, + "learning_rate": 1.9673229418925358e-05, + "loss": 0.4687, + "step": 2086 + }, + { + "epoch": 0.87, + "learning_rate": 1.9672799413249343e-05, + "loss": 0.5428, + "step": 2087 + }, + { + "epoch": 0.87, + "learning_rate": 1.967236912953729e-05, + "loss": 0.5192, + "step": 2088 + }, + { + "epoch": 0.87, + "learning_rate": 1.9671938567801568e-05, + "loss": 0.5889, + "step": 2089 + }, + { + "epoch": 0.88, + "learning_rate": 1.9671507728054558e-05, + "loss": 0.4703, + "step": 2090 + }, + { + "epoch": 0.88, + "learning_rate": 1.9671076610308638e-05, + "loss": 0.5094, + "step": 2091 + }, + { + "epoch": 0.88, + "learning_rate": 1.9670645214576204e-05, + "loss": 0.5278, + "step": 2092 + }, + { + "epoch": 0.88, + "learning_rate": 1.9670213540869655e-05, + "loss": 0.6058, + "step": 2093 + }, + { + "epoch": 0.88, + "learning_rate": 1.96697815892014e-05, + "loss": 0.5246, + "step": 2094 + }, + { + "epoch": 0.88, + "learning_rate": 1.9669349359583852e-05, + "loss": 0.5534, + "step": 2095 + }, + { + "epoch": 0.88, + "learning_rate": 1.966891685202944e-05, + "loss": 0.4617, + "step": 2096 + }, + { + "epoch": 0.88, + "learning_rate": 1.9668484066550592e-05, + "loss": 0.519, + "step": 2097 + }, + { + "epoch": 0.88, + "learning_rate": 1.9668051003159746e-05, + "loss": 0.548, + "step": 2098 + }, + { + "epoch": 0.88, + "learning_rate": 1.9667617661869357e-05, + "loss": 0.4658, + "step": 2099 + }, + { + "epoch": 0.88, + "learning_rate": 1.9667184042691877e-05, + "loss": 0.4929, + "step": 2100 + }, + { + "epoch": 0.88, + "learning_rate": 1.966675014563977e-05, + "loss": 0.4961, + "step": 2101 + }, + { + "epoch": 0.88, + "learning_rate": 1.9666315970725507e-05, + "loss": 0.493, + "step": 2102 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665881517961568e-05, + "loss": 0.4827, + "step": 2103 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665446787360444e-05, + "loss": 0.4241, + "step": 2104 + }, + { + "epoch": 0.88, + "learning_rate": 1.9665011778934628e-05, + "loss": 0.4034, + "step": 2105 + }, + { + "epoch": 0.88, + "learning_rate": 1.9664576492696628e-05, + "loss": 0.463, + "step": 2106 + }, + { + "epoch": 0.88, + "learning_rate": 1.9664140928658947e-05, + "loss": 0.5554, + "step": 2107 + }, + { + "epoch": 0.88, + "learning_rate": 1.9663705086834114e-05, + "loss": 0.481, + "step": 2108 + }, + { + "epoch": 0.88, + "learning_rate": 1.9663268967234655e-05, + "loss": 0.4523, + "step": 2109 + }, + { + "epoch": 0.88, + "learning_rate": 1.9662832569873106e-05, + "loss": 0.5055, + "step": 2110 + }, + { + "epoch": 0.88, + "learning_rate": 1.9662395894762007e-05, + "loss": 0.506, + "step": 2111 + }, + { + "epoch": 0.88, + "learning_rate": 1.9661958941913913e-05, + "loss": 0.4366, + "step": 2112 + }, + { + "epoch": 0.88, + "learning_rate": 1.966152171134138e-05, + "loss": 0.4667, + "step": 2113 + }, + { + "epoch": 0.89, + "learning_rate": 1.9661084203056983e-05, + "loss": 0.4995, + "step": 2114 + }, + { + "epoch": 0.89, + "learning_rate": 1.966064641707329e-05, + "loss": 0.5362, + "step": 2115 + }, + { + "epoch": 0.89, + "learning_rate": 1.9660208353402888e-05, + "loss": 0.5667, + "step": 2116 + }, + { + "epoch": 0.89, + "learning_rate": 1.9659770012058374e-05, + "loss": 0.5167, + "step": 2117 + }, + { + "epoch": 0.89, + "learning_rate": 1.9659331393052342e-05, + "loss": 0.5972, + "step": 2118 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658892496397398e-05, + "loss": 0.4921, + "step": 2119 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658453322106165e-05, + "loss": 0.5383, + "step": 2120 + }, + { + "epoch": 0.89, + "learning_rate": 1.9658013870191258e-05, + "loss": 0.55, + "step": 2121 + }, + { + "epoch": 0.89, + "learning_rate": 1.965757414066531e-05, + "loss": 0.4841, + "step": 2122 + }, + { + "epoch": 0.89, + "learning_rate": 1.9657134133540966e-05, + "loss": 0.4447, + "step": 2123 + }, + { + "epoch": 0.89, + "learning_rate": 1.965669384883087e-05, + "loss": 0.5766, + "step": 2124 + }, + { + "epoch": 0.89, + "learning_rate": 1.9656253286547683e-05, + "loss": 0.53, + "step": 2125 + }, + { + "epoch": 0.89, + "learning_rate": 1.965581244670406e-05, + "loss": 0.5233, + "step": 2126 + }, + { + "epoch": 0.89, + "learning_rate": 1.9655371329312677e-05, + "loss": 0.4755, + "step": 2127 + }, + { + "epoch": 0.89, + "learning_rate": 1.9654929934386215e-05, + "loss": 0.4842, + "step": 2128 + }, + { + "epoch": 0.89, + "learning_rate": 1.965448826193736e-05, + "loss": 0.4566, + "step": 2129 + }, + { + "epoch": 0.89, + "learning_rate": 1.9654046311978803e-05, + "loss": 0.4718, + "step": 2130 + }, + { + "epoch": 0.89, + "learning_rate": 1.9653604084523255e-05, + "loss": 0.4792, + "step": 2131 + }, + { + "epoch": 0.89, + "learning_rate": 1.9653161579583423e-05, + "loss": 0.5052, + "step": 2132 + }, + { + "epoch": 0.89, + "learning_rate": 1.965271879717203e-05, + "loss": 0.5531, + "step": 2133 + }, + { + "epoch": 0.89, + "learning_rate": 1.9652275737301797e-05, + "loss": 0.5135, + "step": 2134 + }, + { + "epoch": 0.89, + "learning_rate": 1.9651832399985466e-05, + "loss": 0.5318, + "step": 2135 + }, + { + "epoch": 0.89, + "learning_rate": 1.9651388785235775e-05, + "loss": 0.548, + "step": 2136 + }, + { + "epoch": 0.89, + "learning_rate": 1.965094489306548e-05, + "loss": 0.4697, + "step": 2137 + }, + { + "epoch": 0.9, + "learning_rate": 1.9650500723487335e-05, + "loss": 0.5317, + "step": 2138 + }, + { + "epoch": 0.9, + "learning_rate": 1.9650056276514115e-05, + "loss": 0.4823, + "step": 2139 + }, + { + "epoch": 0.9, + "learning_rate": 1.964961155215859e-05, + "loss": 0.5023, + "step": 2140 + }, + { + "epoch": 0.9, + "learning_rate": 1.964916655043354e-05, + "loss": 0.4548, + "step": 2141 + }, + { + "epoch": 0.9, + "learning_rate": 1.9648721271351763e-05, + "loss": 0.48, + "step": 2142 + }, + { + "epoch": 0.9, + "learning_rate": 1.9648275714926055e-05, + "loss": 0.5174, + "step": 2143 + }, + { + "epoch": 0.9, + "learning_rate": 1.9647829881169223e-05, + "loss": 0.5626, + "step": 2144 + }, + { + "epoch": 0.9, + "learning_rate": 1.9647383770094083e-05, + "loss": 0.4561, + "step": 2145 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646937381713454e-05, + "loss": 0.4395, + "step": 2146 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646490716040174e-05, + "loss": 0.4942, + "step": 2147 + }, + { + "epoch": 0.9, + "learning_rate": 1.9646043773087078e-05, + "loss": 0.4702, + "step": 2148 + }, + { + "epoch": 0.9, + "learning_rate": 1.9645596552867013e-05, + "loss": 0.4292, + "step": 2149 + }, + { + "epoch": 0.9, + "learning_rate": 1.9645149055392834e-05, + "loss": 0.5833, + "step": 2150 + }, + { + "epoch": 0.9, + "learning_rate": 1.9644701280677406e-05, + "loss": 0.4801, + "step": 2151 + }, + { + "epoch": 0.9, + "learning_rate": 1.9644253228733596e-05, + "loss": 0.4909, + "step": 2152 + }, + { + "epoch": 0.9, + "learning_rate": 1.9643804899574287e-05, + "loss": 0.5129, + "step": 2153 + }, + { + "epoch": 0.9, + "learning_rate": 1.9643356293212363e-05, + "loss": 0.5068, + "step": 2154 + }, + { + "epoch": 0.9, + "learning_rate": 1.964290740966072e-05, + "loss": 0.4858, + "step": 2155 + }, + { + "epoch": 0.9, + "learning_rate": 1.9642458248932262e-05, + "loss": 0.5096, + "step": 2156 + }, + { + "epoch": 0.9, + "learning_rate": 1.9642008811039893e-05, + "loss": 0.5049, + "step": 2157 + }, + { + "epoch": 0.9, + "learning_rate": 1.9641559095996544e-05, + "loss": 0.4885, + "step": 2158 + }, + { + "epoch": 0.9, + "learning_rate": 1.964110910381513e-05, + "loss": 0.4504, + "step": 2159 + }, + { + "epoch": 0.9, + "learning_rate": 1.964065883450859e-05, + "loss": 0.5344, + "step": 2160 + }, + { + "epoch": 0.9, + "learning_rate": 1.9640208288089872e-05, + "loss": 0.4163, + "step": 2161 + }, + { + "epoch": 0.91, + "learning_rate": 1.9639757464571914e-05, + "loss": 0.4949, + "step": 2162 + }, + { + "epoch": 0.91, + "learning_rate": 1.9639306363967686e-05, + "loss": 0.4949, + "step": 2163 + }, + { + "epoch": 0.91, + "learning_rate": 1.963885498629015e-05, + "loss": 0.5139, + "step": 2164 + }, + { + "epoch": 0.91, + "learning_rate": 1.9638403331552277e-05, + "loss": 0.4826, + "step": 2165 + }, + { + "epoch": 0.91, + "learning_rate": 1.963795139976706e-05, + "loss": 0.5396, + "step": 2166 + }, + { + "epoch": 0.91, + "learning_rate": 1.9637499190947478e-05, + "loss": 0.507, + "step": 2167 + }, + { + "epoch": 0.91, + "learning_rate": 1.9637046705106534e-05, + "loss": 0.5509, + "step": 2168 + }, + { + "epoch": 0.91, + "learning_rate": 1.9636593942257237e-05, + "loss": 0.4484, + "step": 2169 + }, + { + "epoch": 0.91, + "learning_rate": 1.9636140902412598e-05, + "loss": 0.4527, + "step": 2170 + }, + { + "epoch": 0.91, + "learning_rate": 1.9635687585585637e-05, + "loss": 0.524, + "step": 2171 + }, + { + "epoch": 0.91, + "learning_rate": 1.963523399178939e-05, + "loss": 0.4281, + "step": 2172 + }, + { + "epoch": 0.91, + "learning_rate": 1.9634780121036887e-05, + "loss": 0.3896, + "step": 2173 + }, + { + "epoch": 0.91, + "learning_rate": 1.9634325973341183e-05, + "loss": 0.5746, + "step": 2174 + }, + { + "epoch": 0.91, + "learning_rate": 1.963387154871533e-05, + "loss": 0.5507, + "step": 2175 + }, + { + "epoch": 0.91, + "learning_rate": 1.9633416847172384e-05, + "loss": 0.4694, + "step": 2176 + }, + { + "epoch": 0.91, + "learning_rate": 1.9632961868725424e-05, + "loss": 0.4982, + "step": 2177 + }, + { + "epoch": 0.91, + "learning_rate": 1.963250661338752e-05, + "loss": 0.4752, + "step": 2178 + }, + { + "epoch": 0.91, + "learning_rate": 1.963205108117176e-05, + "loss": 0.6507, + "step": 2179 + }, + { + "epoch": 0.91, + "learning_rate": 1.963159527209124e-05, + "loss": 0.5155, + "step": 2180 + }, + { + "epoch": 0.91, + "learning_rate": 1.9631139186159064e-05, + "loss": 0.6029, + "step": 2181 + }, + { + "epoch": 0.91, + "learning_rate": 1.9630682823388337e-05, + "loss": 0.5069, + "step": 2182 + }, + { + "epoch": 0.91, + "learning_rate": 1.9630226183792175e-05, + "loss": 0.5005, + "step": 2183 + }, + { + "epoch": 0.91, + "learning_rate": 1.962976926738371e-05, + "loss": 0.5772, + "step": 2184 + }, + { + "epoch": 0.91, + "learning_rate": 1.9629312074176073e-05, + "loss": 0.4924, + "step": 2185 + }, + { + "epoch": 0.92, + "learning_rate": 1.9628854604182403e-05, + "loss": 0.5, + "step": 2186 + }, + { + "epoch": 0.92, + "learning_rate": 1.9628396857415856e-05, + "loss": 0.4859, + "step": 2187 + }, + { + "epoch": 0.92, + "learning_rate": 1.962793883388958e-05, + "loss": 0.4494, + "step": 2188 + }, + { + "epoch": 0.92, + "learning_rate": 1.962748053361675e-05, + "loss": 0.4656, + "step": 2189 + }, + { + "epoch": 0.92, + "learning_rate": 1.9627021956610534e-05, + "loss": 0.475, + "step": 2190 + }, + { + "epoch": 0.92, + "learning_rate": 1.9626563102884117e-05, + "loss": 0.4323, + "step": 2191 + }, + { + "epoch": 0.92, + "learning_rate": 1.9626103972450683e-05, + "loss": 0.578, + "step": 2192 + }, + { + "epoch": 0.92, + "learning_rate": 1.962564456532343e-05, + "loss": 0.4718, + "step": 2193 + }, + { + "epoch": 0.92, + "learning_rate": 1.962518488151557e-05, + "loss": 0.4104, + "step": 2194 + }, + { + "epoch": 0.92, + "learning_rate": 1.962472492104031e-05, + "loss": 0.548, + "step": 2195 + }, + { + "epoch": 0.92, + "learning_rate": 1.9624264683910877e-05, + "loss": 0.555, + "step": 2196 + }, + { + "epoch": 0.92, + "learning_rate": 1.9623804170140494e-05, + "loss": 0.4559, + "step": 2197 + }, + { + "epoch": 0.92, + "learning_rate": 1.9623343379742397e-05, + "loss": 0.5425, + "step": 2198 + }, + { + "epoch": 0.92, + "learning_rate": 1.9622882312729835e-05, + "loss": 0.5496, + "step": 2199 + }, + { + "epoch": 0.92, + "learning_rate": 1.9622420969116062e-05, + "loss": 0.5709, + "step": 2200 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621959348914335e-05, + "loss": 0.5725, + "step": 2201 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621497452137927e-05, + "loss": 0.4704, + "step": 2202 + }, + { + "epoch": 0.92, + "learning_rate": 1.9621035278800115e-05, + "loss": 0.5526, + "step": 2203 + }, + { + "epoch": 0.92, + "learning_rate": 1.9620572828914177e-05, + "loss": 0.5194, + "step": 2204 + }, + { + "epoch": 0.92, + "learning_rate": 1.9620110102493412e-05, + "loss": 0.551, + "step": 2205 + }, + { + "epoch": 0.92, + "learning_rate": 1.9619647099551118e-05, + "loss": 0.53, + "step": 2206 + }, + { + "epoch": 0.92, + "learning_rate": 1.9619183820100603e-05, + "loss": 0.4799, + "step": 2207 + }, + { + "epoch": 0.92, + "learning_rate": 1.9618720264155187e-05, + "loss": 0.4713, + "step": 2208 + }, + { + "epoch": 0.92, + "learning_rate": 1.961825643172819e-05, + "loss": 0.4345, + "step": 2209 + }, + { + "epoch": 0.93, + "learning_rate": 1.9617792322832952e-05, + "loss": 0.5059, + "step": 2210 + }, + { + "epoch": 0.93, + "learning_rate": 1.9617327937482805e-05, + "loss": 0.5279, + "step": 2211 + }, + { + "epoch": 0.93, + "learning_rate": 1.9616863275691102e-05, + "loss": 0.5578, + "step": 2212 + }, + { + "epoch": 0.93, + "learning_rate": 1.9616398337471196e-05, + "loss": 0.5448, + "step": 2213 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615933122836454e-05, + "loss": 0.5377, + "step": 2214 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615467631800248e-05, + "loss": 0.4938, + "step": 2215 + }, + { + "epoch": 0.93, + "learning_rate": 1.9615001864375957e-05, + "loss": 0.4626, + "step": 2216 + }, + { + "epoch": 0.93, + "learning_rate": 1.9614535820576973e-05, + "loss": 0.5208, + "step": 2217 + }, + { + "epoch": 0.93, + "learning_rate": 1.9614069500416685e-05, + "loss": 0.494, + "step": 2218 + }, + { + "epoch": 0.93, + "learning_rate": 1.9613602903908503e-05, + "loss": 0.5347, + "step": 2219 + }, + { + "epoch": 0.93, + "learning_rate": 1.961313603106583e-05, + "loss": 0.6241, + "step": 2220 + }, + { + "epoch": 0.93, + "learning_rate": 1.96126688819021e-05, + "loss": 0.5039, + "step": 2221 + }, + { + "epoch": 0.93, + "learning_rate": 1.9612201456430732e-05, + "loss": 0.4839, + "step": 2222 + }, + { + "epoch": 0.93, + "learning_rate": 1.9611733754665157e-05, + "loss": 0.5787, + "step": 2223 + }, + { + "epoch": 0.93, + "learning_rate": 1.961126577661883e-05, + "loss": 0.5715, + "step": 2224 + }, + { + "epoch": 0.93, + "learning_rate": 1.9610797522305195e-05, + "loss": 0.4819, + "step": 2225 + }, + { + "epoch": 0.93, + "learning_rate": 1.9610328991737716e-05, + "loss": 0.4741, + "step": 2226 + }, + { + "epoch": 0.93, + "learning_rate": 1.9609860184929858e-05, + "loss": 0.4819, + "step": 2227 + }, + { + "epoch": 0.93, + "learning_rate": 1.9609391101895093e-05, + "loss": 0.5646, + "step": 2228 + }, + { + "epoch": 0.93, + "learning_rate": 1.9608921742646913e-05, + "loss": 0.4428, + "step": 2229 + }, + { + "epoch": 0.93, + "learning_rate": 1.9608452107198798e-05, + "loss": 0.4098, + "step": 2230 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607982195564258e-05, + "loss": 0.4311, + "step": 2231 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607512007756793e-05, + "loss": 0.503, + "step": 2232 + }, + { + "epoch": 0.93, + "learning_rate": 1.9607041543789927e-05, + "loss": 0.5069, + "step": 2233 + }, + { + "epoch": 0.94, + "learning_rate": 1.960657080367717e-05, + "loss": 0.4942, + "step": 2234 + }, + { + "epoch": 0.94, + "learning_rate": 1.9606099787432066e-05, + "loss": 0.5272, + "step": 2235 + }, + { + "epoch": 0.94, + "learning_rate": 1.9605628495068143e-05, + "loss": 0.5224, + "step": 2236 + }, + { + "epoch": 0.94, + "learning_rate": 1.9605156926598953e-05, + "loss": 0.6102, + "step": 2237 + }, + { + "epoch": 0.94, + "learning_rate": 1.9604685082038053e-05, + "loss": 0.5718, + "step": 2238 + }, + { + "epoch": 0.94, + "learning_rate": 1.9604212961399004e-05, + "loss": 0.4199, + "step": 2239 + }, + { + "epoch": 0.94, + "learning_rate": 1.960374056469537e-05, + "loss": 0.5303, + "step": 2240 + }, + { + "epoch": 0.94, + "learning_rate": 1.9603267891940742e-05, + "loss": 0.4223, + "step": 2241 + }, + { + "epoch": 0.94, + "learning_rate": 1.9602794943148698e-05, + "loss": 0.6121, + "step": 2242 + }, + { + "epoch": 0.94, + "learning_rate": 1.9602321718332838e-05, + "loss": 0.5694, + "step": 2243 + }, + { + "epoch": 0.94, + "learning_rate": 1.9601848217506755e-05, + "loss": 0.5002, + "step": 2244 + }, + { + "epoch": 0.94, + "learning_rate": 1.960137444068407e-05, + "loss": 0.5088, + "step": 2245 + }, + { + "epoch": 0.94, + "learning_rate": 1.9600900387878397e-05, + "loss": 0.555, + "step": 2246 + }, + { + "epoch": 0.94, + "learning_rate": 1.9600426059103358e-05, + "loss": 0.5601, + "step": 2247 + }, + { + "epoch": 0.94, + "learning_rate": 1.9599951454372594e-05, + "loss": 0.4603, + "step": 2248 + }, + { + "epoch": 0.94, + "learning_rate": 1.9599476573699747e-05, + "loss": 0.5543, + "step": 2249 + }, + { + "epoch": 0.94, + "learning_rate": 1.959900141709846e-05, + "loss": 0.5286, + "step": 2250 + }, + { + "epoch": 0.94, + "learning_rate": 1.9598525984582398e-05, + "loss": 0.494, + "step": 2251 + }, + { + "epoch": 0.94, + "learning_rate": 1.959805027616522e-05, + "loss": 0.4904, + "step": 2252 + }, + { + "epoch": 0.94, + "learning_rate": 1.9597574291860607e-05, + "loss": 0.5076, + "step": 2253 + }, + { + "epoch": 0.94, + "learning_rate": 1.9597098031682238e-05, + "loss": 0.5447, + "step": 2254 + }, + { + "epoch": 0.94, + "learning_rate": 1.9596621495643805e-05, + "loss": 0.4445, + "step": 2255 + }, + { + "epoch": 0.94, + "learning_rate": 1.9596144683759e-05, + "loss": 0.5893, + "step": 2256 + }, + { + "epoch": 0.94, + "learning_rate": 1.959566759604153e-05, + "loss": 0.479, + "step": 2257 + }, + { + "epoch": 0.95, + "learning_rate": 1.959519023250511e-05, + "loss": 0.5299, + "step": 2258 + }, + { + "epoch": 0.95, + "learning_rate": 1.959471259316346e-05, + "loss": 0.4844, + "step": 2259 + }, + { + "epoch": 0.95, + "learning_rate": 1.9594234678030317e-05, + "loss": 0.4575, + "step": 2260 + }, + { + "epoch": 0.95, + "learning_rate": 1.959375648711941e-05, + "loss": 0.538, + "step": 2261 + }, + { + "epoch": 0.95, + "learning_rate": 1.9593278020444483e-05, + "loss": 0.5077, + "step": 2262 + }, + { + "epoch": 0.95, + "learning_rate": 1.9592799278019294e-05, + "loss": 0.5273, + "step": 2263 + }, + { + "epoch": 0.95, + "learning_rate": 1.9592320259857603e-05, + "loss": 0.4944, + "step": 2264 + }, + { + "epoch": 0.95, + "learning_rate": 1.9591840965973175e-05, + "loss": 0.525, + "step": 2265 + }, + { + "epoch": 0.95, + "learning_rate": 1.9591361396379795e-05, + "loss": 0.5158, + "step": 2266 + }, + { + "epoch": 0.95, + "learning_rate": 1.959088155109124e-05, + "loss": 0.5111, + "step": 2267 + }, + { + "epoch": 0.95, + "learning_rate": 1.9590401430121305e-05, + "loss": 0.4892, + "step": 2268 + }, + { + "epoch": 0.95, + "learning_rate": 1.9589921033483793e-05, + "loss": 0.5192, + "step": 2269 + }, + { + "epoch": 0.95, + "learning_rate": 1.958944036119251e-05, + "loss": 0.4518, + "step": 2270 + }, + { + "epoch": 0.95, + "learning_rate": 1.9588959413261272e-05, + "loss": 0.4856, + "step": 2271 + }, + { + "epoch": 0.95, + "learning_rate": 1.9588478189703908e-05, + "loss": 0.4939, + "step": 2272 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587996690534244e-05, + "loss": 0.5192, + "step": 2273 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587514915766124e-05, + "loss": 0.5328, + "step": 2274 + }, + { + "epoch": 0.95, + "learning_rate": 1.9587032865413394e-05, + "loss": 0.4875, + "step": 2275 + }, + { + "epoch": 0.95, + "learning_rate": 1.9586550539489914e-05, + "loss": 0.4893, + "step": 2276 + }, + { + "epoch": 0.95, + "learning_rate": 1.9586067938009546e-05, + "loss": 0.4676, + "step": 2277 + }, + { + "epoch": 0.95, + "learning_rate": 1.9585585060986162e-05, + "loss": 0.4876, + "step": 2278 + }, + { + "epoch": 0.95, + "learning_rate": 1.958510190843364e-05, + "loss": 0.4743, + "step": 2279 + }, + { + "epoch": 0.95, + "learning_rate": 1.958461848036587e-05, + "loss": 0.5614, + "step": 2280 + }, + { + "epoch": 0.95, + "learning_rate": 1.958413477679675e-05, + "loss": 0.4286, + "step": 2281 + }, + { + "epoch": 0.96, + "learning_rate": 1.9583650797740173e-05, + "loss": 0.4811, + "step": 2282 + }, + { + "epoch": 0.96, + "learning_rate": 1.9583166543210065e-05, + "loss": 0.5306, + "step": 2283 + }, + { + "epoch": 0.96, + "learning_rate": 1.9582682013220337e-05, + "loss": 0.4897, + "step": 2284 + }, + { + "epoch": 0.96, + "learning_rate": 1.958219720778492e-05, + "loss": 0.4841, + "step": 2285 + }, + { + "epoch": 0.96, + "learning_rate": 1.9581712126917742e-05, + "loss": 0.5035, + "step": 2286 + }, + { + "epoch": 0.96, + "learning_rate": 1.9581226770632753e-05, + "loss": 0.4354, + "step": 2287 + }, + { + "epoch": 0.96, + "learning_rate": 1.9580741138943906e-05, + "loss": 0.461, + "step": 2288 + }, + { + "epoch": 0.96, + "learning_rate": 1.9580255231865156e-05, + "loss": 0.4559, + "step": 2289 + }, + { + "epoch": 0.96, + "learning_rate": 1.9579769049410468e-05, + "loss": 0.5464, + "step": 2290 + }, + { + "epoch": 0.96, + "learning_rate": 1.957928259159382e-05, + "loss": 0.4297, + "step": 2291 + }, + { + "epoch": 0.96, + "learning_rate": 1.95787958584292e-05, + "loss": 0.5206, + "step": 2292 + }, + { + "epoch": 0.96, + "learning_rate": 1.9578308849930586e-05, + "loss": 0.5657, + "step": 2293 + }, + { + "epoch": 0.96, + "learning_rate": 1.957782156611199e-05, + "loss": 0.4817, + "step": 2294 + }, + { + "epoch": 0.96, + "learning_rate": 1.9577334006987405e-05, + "loss": 0.5052, + "step": 2295 + }, + { + "epoch": 0.96, + "learning_rate": 1.9576846172570855e-05, + "loss": 0.5689, + "step": 2296 + }, + { + "epoch": 0.96, + "learning_rate": 1.9576358062876363e-05, + "loss": 0.5839, + "step": 2297 + }, + { + "epoch": 0.96, + "learning_rate": 1.9575869677917956e-05, + "loss": 0.4988, + "step": 2298 + }, + { + "epoch": 0.96, + "learning_rate": 1.957538101770967e-05, + "loss": 0.4993, + "step": 2299 + }, + { + "epoch": 0.96, + "learning_rate": 1.9574892082265557e-05, + "loss": 0.4793, + "step": 2300 + }, + { + "epoch": 0.96, + "learning_rate": 1.9574402871599666e-05, + "loss": 0.5144, + "step": 2301 + }, + { + "epoch": 0.96, + "learning_rate": 1.957391338572606e-05, + "loss": 0.5921, + "step": 2302 + }, + { + "epoch": 0.96, + "learning_rate": 1.957342362465881e-05, + "loss": 0.5145, + "step": 2303 + }, + { + "epoch": 0.96, + "learning_rate": 1.9572933588411997e-05, + "loss": 0.4551, + "step": 2304 + }, + { + "epoch": 0.97, + "learning_rate": 1.9572443276999696e-05, + "loss": 0.5018, + "step": 2305 + }, + { + "epoch": 0.97, + "learning_rate": 1.957195269043601e-05, + "loss": 0.5091, + "step": 2306 + }, + { + "epoch": 0.97, + "learning_rate": 1.957146182873504e-05, + "loss": 0.5403, + "step": 2307 + }, + { + "epoch": 0.97, + "learning_rate": 1.957097069191089e-05, + "loss": 0.5032, + "step": 2308 + }, + { + "epoch": 0.97, + "learning_rate": 1.9570479279977682e-05, + "loss": 0.5441, + "step": 2309 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569987592949543e-05, + "loss": 0.5361, + "step": 2310 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569495630840597e-05, + "loss": 0.4916, + "step": 2311 + }, + { + "epoch": 0.97, + "learning_rate": 1.9569003393664996e-05, + "loss": 0.4256, + "step": 2312 + }, + { + "epoch": 0.97, + "learning_rate": 1.9568510881436883e-05, + "loss": 0.484, + "step": 2313 + }, + { + "epoch": 0.97, + "learning_rate": 1.9568018094170415e-05, + "loss": 0.4544, + "step": 2314 + }, + { + "epoch": 0.97, + "learning_rate": 1.9567525031879756e-05, + "loss": 0.4739, + "step": 2315 + }, + { + "epoch": 0.97, + "learning_rate": 1.956703169457908e-05, + "loss": 0.4568, + "step": 2316 + }, + { + "epoch": 0.97, + "learning_rate": 1.956653808228257e-05, + "loss": 0.5526, + "step": 2317 + }, + { + "epoch": 0.97, + "learning_rate": 1.956604419500441e-05, + "loss": 0.5471, + "step": 2318 + }, + { + "epoch": 0.97, + "learning_rate": 1.95655500327588e-05, + "loss": 0.4968, + "step": 2319 + }, + { + "epoch": 0.97, + "learning_rate": 1.956505559555994e-05, + "loss": 0.5308, + "step": 2320 + }, + { + "epoch": 0.97, + "learning_rate": 1.9564560883422045e-05, + "loss": 0.6437, + "step": 2321 + }, + { + "epoch": 0.97, + "learning_rate": 1.9564065896359337e-05, + "loss": 0.4779, + "step": 2322 + }, + { + "epoch": 0.97, + "learning_rate": 1.9563570634386042e-05, + "loss": 0.4739, + "step": 2323 + }, + { + "epoch": 0.97, + "learning_rate": 1.9563075097516397e-05, + "loss": 0.4708, + "step": 2324 + }, + { + "epoch": 0.97, + "learning_rate": 1.9562579285764638e-05, + "loss": 0.4772, + "step": 2325 + }, + { + "epoch": 0.97, + "learning_rate": 1.956208319914503e-05, + "loss": 0.5105, + "step": 2326 + }, + { + "epoch": 0.97, + "learning_rate": 1.956158683767182e-05, + "loss": 0.4218, + "step": 2327 + }, + { + "epoch": 0.97, + "learning_rate": 1.9561090201359285e-05, + "loss": 0.5393, + "step": 2328 + }, + { + "epoch": 0.98, + "learning_rate": 1.9560593290221695e-05, + "loss": 0.439, + "step": 2329 + }, + { + "epoch": 0.98, + "learning_rate": 1.9560096104273335e-05, + "loss": 0.4925, + "step": 2330 + }, + { + "epoch": 0.98, + "learning_rate": 1.95595986435285e-05, + "loss": 0.5582, + "step": 2331 + }, + { + "epoch": 0.98, + "learning_rate": 1.9559100908001478e-05, + "loss": 0.4094, + "step": 2332 + }, + { + "epoch": 0.98, + "learning_rate": 1.9558602897706586e-05, + "loss": 0.5477, + "step": 2333 + }, + { + "epoch": 0.98, + "learning_rate": 1.9558104612658137e-05, + "loss": 0.5561, + "step": 2334 + }, + { + "epoch": 0.98, + "learning_rate": 1.9557606052870452e-05, + "loss": 0.4986, + "step": 2335 + }, + { + "epoch": 0.98, + "learning_rate": 1.9557107218357862e-05, + "loss": 0.4391, + "step": 2336 + }, + { + "epoch": 0.98, + "learning_rate": 1.9556608109134707e-05, + "loss": 0.432, + "step": 2337 + }, + { + "epoch": 0.98, + "learning_rate": 1.955610872521533e-05, + "loss": 0.4751, + "step": 2338 + }, + { + "epoch": 0.98, + "learning_rate": 1.955560906661409e-05, + "loss": 0.5742, + "step": 2339 + }, + { + "epoch": 0.98, + "learning_rate": 1.9555109133345347e-05, + "loss": 0.5143, + "step": 2340 + }, + { + "epoch": 0.98, + "learning_rate": 1.9554608925423467e-05, + "loss": 0.4496, + "step": 2341 + }, + { + "epoch": 0.98, + "learning_rate": 1.9554108442862836e-05, + "loss": 0.4794, + "step": 2342 + }, + { + "epoch": 0.98, + "learning_rate": 1.9553607685677834e-05, + "loss": 0.4721, + "step": 2343 + }, + { + "epoch": 0.98, + "learning_rate": 1.955310665388286e-05, + "loss": 0.5335, + "step": 2344 + }, + { + "epoch": 0.98, + "learning_rate": 1.955260534749231e-05, + "loss": 0.4888, + "step": 2345 + }, + { + "epoch": 0.98, + "learning_rate": 1.9552103766520596e-05, + "loss": 0.452, + "step": 2346 + }, + { + "epoch": 0.98, + "learning_rate": 1.9551601910982136e-05, + "loss": 0.5632, + "step": 2347 + }, + { + "epoch": 0.98, + "learning_rate": 1.9551099780891353e-05, + "loss": 0.4978, + "step": 2348 + }, + { + "epoch": 0.98, + "learning_rate": 1.9550597376262685e-05, + "loss": 0.5115, + "step": 2349 + }, + { + "epoch": 0.98, + "learning_rate": 1.955009469711057e-05, + "loss": 0.5622, + "step": 2350 + }, + { + "epoch": 0.98, + "learning_rate": 1.954959174344946e-05, + "loss": 0.5278, + "step": 2351 + }, + { + "epoch": 0.98, + "learning_rate": 1.9549088515293804e-05, + "loss": 0.4838, + "step": 2352 + }, + { + "epoch": 0.99, + "learning_rate": 1.9548585012658074e-05, + "loss": 0.5064, + "step": 2353 + }, + { + "epoch": 0.99, + "learning_rate": 1.954808123555674e-05, + "loss": 0.5339, + "step": 2354 + }, + { + "epoch": 0.99, + "learning_rate": 1.954757718400429e-05, + "loss": 0.4907, + "step": 2355 + }, + { + "epoch": 0.99, + "learning_rate": 1.95470728580152e-05, + "loss": 0.485, + "step": 2356 + }, + { + "epoch": 0.99, + "learning_rate": 1.9546568257603976e-05, + "loss": 0.4973, + "step": 2357 + }, + { + "epoch": 0.99, + "learning_rate": 1.9546063382785117e-05, + "loss": 0.5277, + "step": 2358 + }, + { + "epoch": 0.99, + "learning_rate": 1.9545558233573136e-05, + "loss": 0.4577, + "step": 2359 + }, + { + "epoch": 0.99, + "learning_rate": 1.9545052809982556e-05, + "loss": 0.4347, + "step": 2360 + }, + { + "epoch": 0.99, + "learning_rate": 1.9544547112027904e-05, + "loss": 0.4502, + "step": 2361 + }, + { + "epoch": 0.99, + "learning_rate": 1.9544041139723712e-05, + "loss": 0.446, + "step": 2362 + }, + { + "epoch": 0.99, + "learning_rate": 1.954353489308453e-05, + "loss": 0.618, + "step": 2363 + }, + { + "epoch": 0.99, + "learning_rate": 1.9543028372124903e-05, + "loss": 0.4332, + "step": 2364 + }, + { + "epoch": 0.99, + "learning_rate": 1.9542521576859397e-05, + "loss": 0.4302, + "step": 2365 + }, + { + "epoch": 0.99, + "learning_rate": 1.9542014507302576e-05, + "loss": 0.5073, + "step": 2366 + }, + { + "epoch": 0.99, + "learning_rate": 1.9541507163469012e-05, + "loss": 0.4632, + "step": 2367 + }, + { + "epoch": 0.99, + "learning_rate": 1.9540999545373295e-05, + "loss": 0.5273, + "step": 2368 + }, + { + "epoch": 0.99, + "learning_rate": 1.954049165303001e-05, + "loss": 0.4725, + "step": 2369 + }, + { + "epoch": 0.99, + "learning_rate": 1.9539983486453762e-05, + "loss": 0.4652, + "step": 2370 + }, + { + "epoch": 0.99, + "learning_rate": 1.953947504565915e-05, + "loss": 0.4843, + "step": 2371 + }, + { + "epoch": 0.99, + "learning_rate": 1.9538966330660796e-05, + "loss": 0.511, + "step": 2372 + }, + { + "epoch": 0.99, + "learning_rate": 1.9538457341473318e-05, + "loss": 0.5827, + "step": 2373 + }, + { + "epoch": 0.99, + "learning_rate": 1.9537948078111348e-05, + "loss": 0.4844, + "step": 2374 + }, + { + "epoch": 0.99, + "learning_rate": 1.9537438540589526e-05, + "loss": 0.5425, + "step": 2375 + }, + { + "epoch": 0.99, + "learning_rate": 1.9536928728922496e-05, + "loss": 0.4983, + "step": 2376 + }, + { + "epoch": 1.0, + "learning_rate": 1.953641864312491e-05, + "loss": 0.5085, + "step": 2377 + }, + { + "epoch": 1.0, + "learning_rate": 1.9535908283211435e-05, + "loss": 0.5282, + "step": 2378 + }, + { + "epoch": 1.0, + "learning_rate": 1.9535397649196738e-05, + "loss": 0.4554, + "step": 2379 + }, + { + "epoch": 1.0, + "learning_rate": 1.95348867410955e-05, + "loss": 0.4774, + "step": 2380 + }, + { + "epoch": 1.0, + "learning_rate": 1.95343755589224e-05, + "loss": 0.6091, + "step": 2381 + }, + { + "epoch": 1.0, + "learning_rate": 1.9533864102692134e-05, + "loss": 0.6011, + "step": 2382 + }, + { + "epoch": 1.0, + "learning_rate": 1.9533352372419408e-05, + "loss": 0.4555, + "step": 2383 + }, + { + "epoch": 1.0, + "learning_rate": 1.953284036811893e-05, + "loss": 0.435, + "step": 2384 + }, + { + "epoch": 1.0, + "learning_rate": 1.9532328089805407e-05, + "loss": 0.5451, + "step": 2385 + }, + { + "epoch": 1.0, + "learning_rate": 1.953181553749358e-05, + "loss": 0.4929, + "step": 2386 + }, + { + "epoch": 1.0, + "learning_rate": 1.953130271119817e-05, + "loss": 0.5192, + "step": 2387 + }, + { + "epoch": 1.0, + "learning_rate": 1.953078961093392e-05, + "loss": 0.5065, + "step": 2388 + }, + { + "epoch": 1.0, + "learning_rate": 1.953027623671558e-05, + "loss": 0.4521, + "step": 2389 + }, + { + "epoch": 1.0, + "learning_rate": 1.952976258855791e-05, + "loss": 0.3904, + "step": 2390 + }, + { + "epoch": 1.0, + "learning_rate": 1.9529248666475668e-05, + "loss": 0.3474, + "step": 2391 + }, + { + "epoch": 1.0, + "learning_rate": 1.9528734470483632e-05, + "loss": 0.3506, + "step": 2392 + }, + { + "epoch": 1.0, + "learning_rate": 1.9528220000596577e-05, + "loss": 0.399, + "step": 2393 + }, + { + "epoch": 1.0, + "learning_rate": 1.9527705256829293e-05, + "loss": 0.3954, + "step": 2394 + }, + { + "epoch": 1.0, + "learning_rate": 1.952719023919658e-05, + "loss": 0.3772, + "step": 2395 + }, + { + "epoch": 1.0, + "learning_rate": 1.9526674947713234e-05, + "loss": 0.4016, + "step": 2396 + }, + { + "epoch": 1.0, + "learning_rate": 1.9526159382394068e-05, + "loss": 0.3883, + "step": 2397 + }, + { + "epoch": 1.0, + "learning_rate": 1.952564354325391e-05, + "loss": 0.3803, + "step": 2398 + }, + { + "epoch": 1.0, + "learning_rate": 1.952512743030758e-05, + "loss": 0.3478, + "step": 2399 + }, + { + "epoch": 1.0, + "learning_rate": 1.952461104356991e-05, + "loss": 0.4181, + "step": 2400 + }, + { + "epoch": 1.01, + "learning_rate": 1.952409438305575e-05, + "loss": 0.3764, + "step": 2401 + }, + { + "epoch": 1.01, + "learning_rate": 1.952357744877995e-05, + "loss": 0.412, + "step": 2402 + }, + { + "epoch": 1.01, + "learning_rate": 1.952306024075737e-05, + "loss": 0.3514, + "step": 2403 + }, + { + "epoch": 1.01, + "learning_rate": 1.9522542759002866e-05, + "loss": 0.3896, + "step": 2404 + }, + { + "epoch": 1.01, + "learning_rate": 1.9522025003531327e-05, + "loss": 0.3816, + "step": 2405 + }, + { + "epoch": 1.01, + "learning_rate": 1.9521506974357628e-05, + "loss": 0.4014, + "step": 2406 + }, + { + "epoch": 1.01, + "learning_rate": 1.952098867149666e-05, + "loss": 0.3944, + "step": 2407 + }, + { + "epoch": 1.01, + "learning_rate": 1.952047009496332e-05, + "loss": 0.376, + "step": 2408 + }, + { + "epoch": 1.01, + "learning_rate": 1.9519951244772514e-05, + "loss": 0.363, + "step": 2409 + }, + { + "epoch": 1.01, + "learning_rate": 1.951943212093916e-05, + "loss": 0.4244, + "step": 2410 + }, + { + "epoch": 1.01, + "learning_rate": 1.9518912723478176e-05, + "loss": 0.4064, + "step": 2411 + }, + { + "epoch": 1.01, + "learning_rate": 1.9518393052404495e-05, + "loss": 0.3723, + "step": 2412 + }, + { + "epoch": 1.01, + "learning_rate": 1.951787310773305e-05, + "loss": 0.3509, + "step": 2413 + }, + { + "epoch": 1.01, + "learning_rate": 1.9517352889478787e-05, + "loss": 0.3573, + "step": 2414 + }, + { + "epoch": 1.01, + "learning_rate": 1.9516832397656663e-05, + "loss": 0.3915, + "step": 2415 + }, + { + "epoch": 1.01, + "learning_rate": 1.9516311632281636e-05, + "loss": 0.3676, + "step": 2416 + }, + { + "epoch": 1.01, + "learning_rate": 1.9515790593368677e-05, + "loss": 0.4132, + "step": 2417 + }, + { + "epoch": 1.01, + "learning_rate": 1.951526928093276e-05, + "loss": 0.4165, + "step": 2418 + }, + { + "epoch": 1.01, + "learning_rate": 1.9514747694988873e-05, + "loss": 0.3968, + "step": 2419 + }, + { + "epoch": 1.01, + "learning_rate": 1.9514225835552004e-05, + "loss": 0.3525, + "step": 2420 + }, + { + "epoch": 1.01, + "learning_rate": 1.9513703702637158e-05, + "loss": 0.3924, + "step": 2421 + }, + { + "epoch": 1.01, + "learning_rate": 1.951318129625934e-05, + "loss": 0.3927, + "step": 2422 + }, + { + "epoch": 1.01, + "learning_rate": 1.951265861643357e-05, + "loss": 0.4062, + "step": 2423 + }, + { + "epoch": 1.01, + "learning_rate": 1.951213566317487e-05, + "loss": 0.3723, + "step": 2424 + }, + { + "epoch": 1.02, + "learning_rate": 1.9511612436498267e-05, + "loss": 0.3841, + "step": 2425 + }, + { + "epoch": 1.02, + "learning_rate": 1.9511088936418807e-05, + "loss": 0.3915, + "step": 2426 + }, + { + "epoch": 1.02, + "learning_rate": 1.9510565162951538e-05, + "loss": 0.4367, + "step": 2427 + }, + { + "epoch": 1.02, + "learning_rate": 1.951004111611151e-05, + "loss": 0.3407, + "step": 2428 + }, + { + "epoch": 1.02, + "learning_rate": 1.950951679591379e-05, + "loss": 0.3957, + "step": 2429 + }, + { + "epoch": 1.02, + "learning_rate": 1.9508992202373448e-05, + "loss": 0.3546, + "step": 2430 + }, + { + "epoch": 1.02, + "learning_rate": 1.9508467335505564e-05, + "loss": 0.3823, + "step": 2431 + }, + { + "epoch": 1.02, + "learning_rate": 1.9507942195325223e-05, + "loss": 0.3666, + "step": 2432 + }, + { + "epoch": 1.02, + "learning_rate": 1.9507416781847524e-05, + "loss": 0.4182, + "step": 2433 + }, + { + "epoch": 1.02, + "learning_rate": 1.9506891095087563e-05, + "loss": 0.4023, + "step": 2434 + }, + { + "epoch": 1.02, + "learning_rate": 1.950636513506045e-05, + "loss": 0.4253, + "step": 2435 + }, + { + "epoch": 1.02, + "learning_rate": 1.9505838901781314e-05, + "loss": 0.3536, + "step": 2436 + }, + { + "epoch": 1.02, + "learning_rate": 1.9505312395265273e-05, + "loss": 0.3591, + "step": 2437 + }, + { + "epoch": 1.02, + "learning_rate": 1.950478561552746e-05, + "loss": 0.3788, + "step": 2438 + }, + { + "epoch": 1.02, + "learning_rate": 1.9504258562583018e-05, + "loss": 0.3627, + "step": 2439 + }, + { + "epoch": 1.02, + "learning_rate": 1.95037312364471e-05, + "loss": 0.3955, + "step": 2440 + }, + { + "epoch": 1.02, + "learning_rate": 1.950320363713486e-05, + "loss": 0.3654, + "step": 2441 + }, + { + "epoch": 1.02, + "learning_rate": 1.9502675764661464e-05, + "loss": 0.3787, + "step": 2442 + }, + { + "epoch": 1.02, + "learning_rate": 1.9502147619042086e-05, + "loss": 0.3312, + "step": 2443 + }, + { + "epoch": 1.02, + "learning_rate": 1.950161920029191e-05, + "loss": 0.3557, + "step": 2444 + }, + { + "epoch": 1.02, + "learning_rate": 1.9501090508426117e-05, + "loss": 0.3739, + "step": 2445 + }, + { + "epoch": 1.02, + "learning_rate": 1.950056154345991e-05, + "loss": 0.3747, + "step": 2446 + }, + { + "epoch": 1.02, + "learning_rate": 1.9500032305408494e-05, + "loss": 0.3263, + "step": 2447 + }, + { + "epoch": 1.02, + "learning_rate": 1.9499502794287078e-05, + "loss": 0.3685, + "step": 2448 + }, + { + "epoch": 1.03, + "learning_rate": 1.9498973010110884e-05, + "loss": 0.3698, + "step": 2449 + }, + { + "epoch": 1.03, + "learning_rate": 1.949844295289514e-05, + "loss": 0.3479, + "step": 2450 + }, + { + "epoch": 1.03, + "learning_rate": 1.949791262265508e-05, + "loss": 0.3987, + "step": 2451 + }, + { + "epoch": 1.03, + "learning_rate": 1.949738201940595e-05, + "loss": 0.3598, + "step": 2452 + }, + { + "epoch": 1.03, + "learning_rate": 1.9496851143163004e-05, + "loss": 0.3291, + "step": 2453 + }, + { + "epoch": 1.03, + "learning_rate": 1.9496319993941498e-05, + "loss": 0.4026, + "step": 2454 + }, + { + "epoch": 1.03, + "learning_rate": 1.9495788571756704e-05, + "loss": 0.3992, + "step": 2455 + }, + { + "epoch": 1.03, + "learning_rate": 1.949525687662389e-05, + "loss": 0.3673, + "step": 2456 + }, + { + "epoch": 1.03, + "learning_rate": 1.9494724908558345e-05, + "loss": 0.391, + "step": 2457 + }, + { + "epoch": 1.03, + "learning_rate": 1.9494192667575355e-05, + "loss": 0.3577, + "step": 2458 + }, + { + "epoch": 1.03, + "learning_rate": 1.9493660153690225e-05, + "loss": 0.3866, + "step": 2459 + }, + { + "epoch": 1.03, + "learning_rate": 1.949312736691826e-05, + "loss": 0.3643, + "step": 2460 + }, + { + "epoch": 1.03, + "learning_rate": 1.9492594307274772e-05, + "loss": 0.3734, + "step": 2461 + }, + { + "epoch": 1.03, + "learning_rate": 1.949206097477508e-05, + "loss": 0.4079, + "step": 2462 + }, + { + "epoch": 1.03, + "learning_rate": 1.949152736943452e-05, + "loss": 0.3489, + "step": 2463 + }, + { + "epoch": 1.03, + "learning_rate": 1.949099349126843e-05, + "loss": 0.3724, + "step": 2464 + }, + { + "epoch": 1.03, + "learning_rate": 1.9490459340292153e-05, + "loss": 0.3742, + "step": 2465 + }, + { + "epoch": 1.03, + "learning_rate": 1.9489924916521047e-05, + "loss": 0.3711, + "step": 2466 + }, + { + "epoch": 1.03, + "learning_rate": 1.9489390219970468e-05, + "loss": 0.3826, + "step": 2467 + }, + { + "epoch": 1.03, + "learning_rate": 1.9488855250655785e-05, + "loss": 0.3584, + "step": 2468 + }, + { + "epoch": 1.03, + "learning_rate": 1.948832000859238e-05, + "loss": 0.3569, + "step": 2469 + }, + { + "epoch": 1.03, + "learning_rate": 1.9487784493795637e-05, + "loss": 0.3772, + "step": 2470 + }, + { + "epoch": 1.03, + "learning_rate": 1.948724870628095e-05, + "loss": 0.3452, + "step": 2471 + }, + { + "epoch": 1.03, + "learning_rate": 1.9486712646063715e-05, + "loss": 0.4041, + "step": 2472 + }, + { + "epoch": 1.04, + "learning_rate": 1.9486176313159344e-05, + "loss": 0.3882, + "step": 2473 + }, + { + "epoch": 1.04, + "learning_rate": 1.9485639707583255e-05, + "loss": 0.4075, + "step": 2474 + }, + { + "epoch": 1.04, + "learning_rate": 1.948510282935087e-05, + "loss": 0.355, + "step": 2475 + }, + { + "epoch": 1.04, + "learning_rate": 1.9484565678477616e-05, + "loss": 0.3827, + "step": 2476 + }, + { + "epoch": 1.04, + "learning_rate": 1.9484028254978942e-05, + "loss": 0.3761, + "step": 2477 + }, + { + "epoch": 1.04, + "learning_rate": 1.948349055887029e-05, + "loss": 0.4096, + "step": 2478 + }, + { + "epoch": 1.04, + "learning_rate": 1.9482952590167118e-05, + "loss": 0.3582, + "step": 2479 + }, + { + "epoch": 1.04, + "learning_rate": 1.9482414348884892e-05, + "loss": 0.3936, + "step": 2480 + }, + { + "epoch": 1.04, + "learning_rate": 1.9481875835039075e-05, + "loss": 0.3534, + "step": 2481 + }, + { + "epoch": 1.04, + "learning_rate": 1.9481337048645154e-05, + "loss": 0.3723, + "step": 2482 + }, + { + "epoch": 1.04, + "learning_rate": 1.9480797989718612e-05, + "loss": 0.4351, + "step": 2483 + }, + { + "epoch": 1.04, + "learning_rate": 1.9480258658274947e-05, + "loss": 0.3678, + "step": 2484 + }, + { + "epoch": 1.04, + "learning_rate": 1.9479719054329657e-05, + "loss": 0.3977, + "step": 2485 + }, + { + "epoch": 1.04, + "learning_rate": 1.9479179177898255e-05, + "loss": 0.3488, + "step": 2486 + }, + { + "epoch": 1.04, + "learning_rate": 1.9478639028996262e-05, + "loss": 0.3655, + "step": 2487 + }, + { + "epoch": 1.04, + "learning_rate": 1.9478098607639197e-05, + "loss": 0.4027, + "step": 2488 + }, + { + "epoch": 1.04, + "learning_rate": 1.94775579138426e-05, + "loss": 0.3937, + "step": 2489 + }, + { + "epoch": 1.04, + "learning_rate": 1.9477016947622013e-05, + "loss": 0.354, + "step": 2490 + }, + { + "epoch": 1.04, + "learning_rate": 1.947647570899298e-05, + "loss": 0.4085, + "step": 2491 + }, + { + "epoch": 1.04, + "learning_rate": 1.947593419797107e-05, + "loss": 0.3629, + "step": 2492 + }, + { + "epoch": 1.04, + "learning_rate": 1.947539241457183e-05, + "loss": 0.3365, + "step": 2493 + }, + { + "epoch": 1.04, + "learning_rate": 1.947485035881085e-05, + "loss": 0.3338, + "step": 2494 + }, + { + "epoch": 1.04, + "learning_rate": 1.94743080307037e-05, + "loss": 0.4101, + "step": 2495 + }, + { + "epoch": 1.05, + "learning_rate": 1.947376543026598e-05, + "loss": 0.3708, + "step": 2496 + }, + { + "epoch": 1.05, + "learning_rate": 1.9473222557513273e-05, + "loss": 0.3908, + "step": 2497 + }, + { + "epoch": 1.05, + "learning_rate": 1.947267941246119e-05, + "loss": 0.3852, + "step": 2498 + }, + { + "epoch": 1.05, + "learning_rate": 1.9472135995125344e-05, + "loss": 0.3899, + "step": 2499 + }, + { + "epoch": 1.05, + "learning_rate": 1.9471592305521355e-05, + "loss": 0.4517, + "step": 2500 + }, + { + "epoch": 1.05, + "learning_rate": 1.9471048343664853e-05, + "loss": 0.3975, + "step": 2501 + }, + { + "epoch": 1.05, + "learning_rate": 1.9470504109571467e-05, + "loss": 0.4205, + "step": 2502 + }, + { + "epoch": 1.05, + "learning_rate": 1.9469959603256844e-05, + "loss": 0.3591, + "step": 2503 + }, + { + "epoch": 1.05, + "learning_rate": 1.946941482473664e-05, + "loss": 0.3632, + "step": 2504 + }, + { + "epoch": 1.05, + "learning_rate": 1.9468869774026508e-05, + "loss": 0.4138, + "step": 2505 + }, + { + "epoch": 1.05, + "learning_rate": 1.9468324451142115e-05, + "loss": 0.3705, + "step": 2506 + }, + { + "epoch": 1.05, + "learning_rate": 1.9467778856099134e-05, + "loss": 0.3899, + "step": 2507 + }, + { + "epoch": 1.05, + "learning_rate": 1.946723298891326e-05, + "loss": 0.342, + "step": 2508 + }, + { + "epoch": 1.05, + "learning_rate": 1.9466686849600168e-05, + "loss": 0.4147, + "step": 2509 + }, + { + "epoch": 1.05, + "learning_rate": 1.9466140438175567e-05, + "loss": 0.391, + "step": 2510 + }, + { + "epoch": 1.05, + "learning_rate": 1.9465593754655158e-05, + "loss": 0.3563, + "step": 2511 + }, + { + "epoch": 1.05, + "learning_rate": 1.9465046799054657e-05, + "loss": 0.4002, + "step": 2512 + }, + { + "epoch": 1.05, + "learning_rate": 1.946449957138978e-05, + "loss": 0.4012, + "step": 2513 + }, + { + "epoch": 1.05, + "learning_rate": 1.9463952071676268e-05, + "loss": 0.3771, + "step": 2514 + }, + { + "epoch": 1.05, + "learning_rate": 1.9463404299929852e-05, + "loss": 0.3488, + "step": 2515 + }, + { + "epoch": 1.05, + "learning_rate": 1.9462856256166273e-05, + "loss": 0.379, + "step": 2516 + }, + { + "epoch": 1.05, + "learning_rate": 1.9462307940401288e-05, + "loss": 0.4167, + "step": 2517 + }, + { + "epoch": 1.05, + "learning_rate": 1.9461759352650663e-05, + "loss": 0.3738, + "step": 2518 + }, + { + "epoch": 1.05, + "learning_rate": 1.9461210492930156e-05, + "loss": 0.3786, + "step": 2519 + }, + { + "epoch": 1.06, + "learning_rate": 1.9460661361255552e-05, + "loss": 0.422, + "step": 2520 + }, + { + "epoch": 1.06, + "learning_rate": 1.946011195764263e-05, + "loss": 0.3372, + "step": 2521 + }, + { + "epoch": 1.06, + "learning_rate": 1.9459562282107188e-05, + "loss": 0.3704, + "step": 2522 + }, + { + "epoch": 1.06, + "learning_rate": 1.945901233466502e-05, + "loss": 0.3586, + "step": 2523 + }, + { + "epoch": 1.06, + "learning_rate": 1.9458462115331938e-05, + "loss": 0.3876, + "step": 2524 + }, + { + "epoch": 1.06, + "learning_rate": 1.9457911624123754e-05, + "loss": 0.3911, + "step": 2525 + }, + { + "epoch": 1.06, + "learning_rate": 1.9457360861056294e-05, + "loss": 0.3998, + "step": 2526 + }, + { + "epoch": 1.06, + "learning_rate": 1.9456809826145387e-05, + "loss": 0.3516, + "step": 2527 + }, + { + "epoch": 1.06, + "learning_rate": 1.9456258519406875e-05, + "loss": 0.3591, + "step": 2528 + }, + { + "epoch": 1.06, + "learning_rate": 1.9455706940856602e-05, + "loss": 0.4332, + "step": 2529 + }, + { + "epoch": 1.06, + "learning_rate": 1.9455155090510427e-05, + "loss": 0.4036, + "step": 2530 + }, + { + "epoch": 1.06, + "learning_rate": 1.9454602968384204e-05, + "loss": 0.3703, + "step": 2531 + }, + { + "epoch": 1.06, + "learning_rate": 1.9454050574493815e-05, + "loss": 0.3515, + "step": 2532 + }, + { + "epoch": 1.06, + "learning_rate": 1.9453497908855125e-05, + "loss": 0.4121, + "step": 2533 + }, + { + "epoch": 1.06, + "learning_rate": 1.945294497148403e-05, + "loss": 0.3881, + "step": 2534 + }, + { + "epoch": 1.06, + "learning_rate": 1.945239176239642e-05, + "loss": 0.3798, + "step": 2535 + }, + { + "epoch": 1.06, + "learning_rate": 1.94518382816082e-05, + "loss": 0.3696, + "step": 2536 + }, + { + "epoch": 1.06, + "learning_rate": 1.945128452913527e-05, + "loss": 0.3494, + "step": 2537 + }, + { + "epoch": 1.06, + "learning_rate": 1.9450730504993554e-05, + "loss": 0.4292, + "step": 2538 + }, + { + "epoch": 1.06, + "learning_rate": 1.9450176209198977e-05, + "loss": 0.3643, + "step": 2539 + }, + { + "epoch": 1.06, + "learning_rate": 1.9449621641767474e-05, + "loss": 0.4396, + "step": 2540 + }, + { + "epoch": 1.06, + "learning_rate": 1.944906680271498e-05, + "loss": 0.3905, + "step": 2541 + }, + { + "epoch": 1.06, + "learning_rate": 1.9448511692057446e-05, + "loss": 0.377, + "step": 2542 + }, + { + "epoch": 1.06, + "learning_rate": 1.9447956309810825e-05, + "loss": 0.3605, + "step": 2543 + }, + { + "epoch": 1.07, + "learning_rate": 1.9447400655991086e-05, + "loss": 0.4316, + "step": 2544 + }, + { + "epoch": 1.07, + "learning_rate": 1.94468447306142e-05, + "loss": 0.3723, + "step": 2545 + }, + { + "epoch": 1.07, + "learning_rate": 1.9446288533696145e-05, + "loss": 0.4058, + "step": 2546 + }, + { + "epoch": 1.07, + "learning_rate": 1.9445732065252904e-05, + "loss": 0.379, + "step": 2547 + }, + { + "epoch": 1.07, + "learning_rate": 1.9445175325300482e-05, + "loss": 0.4104, + "step": 2548 + }, + { + "epoch": 1.07, + "learning_rate": 1.9444618313854878e-05, + "loss": 0.4294, + "step": 2549 + }, + { + "epoch": 1.07, + "learning_rate": 1.9444061030932094e-05, + "loss": 0.3712, + "step": 2550 + }, + { + "epoch": 1.07, + "learning_rate": 1.9443503476548162e-05, + "loss": 0.3953, + "step": 2551 + }, + { + "epoch": 1.07, + "learning_rate": 1.9442945650719102e-05, + "loss": 0.3694, + "step": 2552 + }, + { + "epoch": 1.07, + "learning_rate": 1.944238755346095e-05, + "loss": 0.4309, + "step": 2553 + }, + { + "epoch": 1.07, + "learning_rate": 1.9441829184789745e-05, + "loss": 0.3836, + "step": 2554 + }, + { + "epoch": 1.07, + "learning_rate": 1.9441270544721536e-05, + "loss": 0.4283, + "step": 2555 + }, + { + "epoch": 1.07, + "learning_rate": 1.9440711633272387e-05, + "loss": 0.3818, + "step": 2556 + }, + { + "epoch": 1.07, + "learning_rate": 1.944015245045836e-05, + "loss": 0.3676, + "step": 2557 + }, + { + "epoch": 1.07, + "learning_rate": 1.9439592996295525e-05, + "loss": 0.3511, + "step": 2558 + }, + { + "epoch": 1.07, + "learning_rate": 1.943903327079997e-05, + "loss": 0.3633, + "step": 2559 + }, + { + "epoch": 1.07, + "learning_rate": 1.9438473273987775e-05, + "loss": 0.3664, + "step": 2560 + }, + { + "epoch": 1.07, + "learning_rate": 1.9437913005875046e-05, + "loss": 0.4262, + "step": 2561 + }, + { + "epoch": 1.07, + "learning_rate": 1.943735246647788e-05, + "loss": 0.4054, + "step": 2562 + }, + { + "epoch": 1.07, + "learning_rate": 1.9436791655812392e-05, + "loss": 0.4121, + "step": 2563 + }, + { + "epoch": 1.07, + "learning_rate": 1.94362305738947e-05, + "loss": 0.3511, + "step": 2564 + }, + { + "epoch": 1.07, + "learning_rate": 1.9435669220740933e-05, + "loss": 0.3689, + "step": 2565 + }, + { + "epoch": 1.07, + "learning_rate": 1.943510759636723e-05, + "loss": 0.3972, + "step": 2566 + }, + { + "epoch": 1.07, + "learning_rate": 1.9434545700789732e-05, + "loss": 0.404, + "step": 2567 + }, + { + "epoch": 1.08, + "learning_rate": 1.943398353402459e-05, + "loss": 0.3609, + "step": 2568 + }, + { + "epoch": 1.08, + "learning_rate": 1.9433421096087963e-05, + "loss": 0.3381, + "step": 2569 + }, + { + "epoch": 1.08, + "learning_rate": 1.9432858386996018e-05, + "loss": 0.3895, + "step": 2570 + }, + { + "epoch": 1.08, + "learning_rate": 1.9432295406764927e-05, + "loss": 0.3767, + "step": 2571 + }, + { + "epoch": 1.08, + "learning_rate": 1.9431732155410878e-05, + "loss": 0.3764, + "step": 2572 + }, + { + "epoch": 1.08, + "learning_rate": 1.9431168632950053e-05, + "loss": 0.3678, + "step": 2573 + }, + { + "epoch": 1.08, + "learning_rate": 1.9430604839398657e-05, + "loss": 0.3845, + "step": 2574 + }, + { + "epoch": 1.08, + "learning_rate": 1.9430040774772893e-05, + "loss": 0.42, + "step": 2575 + }, + { + "epoch": 1.08, + "learning_rate": 1.9429476439088976e-05, + "loss": 0.385, + "step": 2576 + }, + { + "epoch": 1.08, + "learning_rate": 1.9428911832363126e-05, + "loss": 0.4227, + "step": 2577 + }, + { + "epoch": 1.08, + "learning_rate": 1.9428346954611574e-05, + "loss": 0.3842, + "step": 2578 + }, + { + "epoch": 1.08, + "learning_rate": 1.9427781805850553e-05, + "loss": 0.3735, + "step": 2579 + }, + { + "epoch": 1.08, + "learning_rate": 1.9427216386096313e-05, + "loss": 0.3866, + "step": 2580 + }, + { + "epoch": 1.08, + "learning_rate": 1.94266506953651e-05, + "loss": 0.3834, + "step": 2581 + }, + { + "epoch": 1.08, + "learning_rate": 1.942608473367318e-05, + "loss": 0.4074, + "step": 2582 + }, + { + "epoch": 1.08, + "learning_rate": 1.942551850103682e-05, + "loss": 0.4092, + "step": 2583 + }, + { + "epoch": 1.08, + "learning_rate": 1.942495199747229e-05, + "loss": 0.3959, + "step": 2584 + }, + { + "epoch": 1.08, + "learning_rate": 1.9424385222995883e-05, + "loss": 0.3953, + "step": 2585 + }, + { + "epoch": 1.08, + "learning_rate": 1.9423818177623884e-05, + "loss": 0.3939, + "step": 2586 + }, + { + "epoch": 1.08, + "learning_rate": 1.9423250861372592e-05, + "loss": 0.4049, + "step": 2587 + }, + { + "epoch": 1.08, + "learning_rate": 1.9422683274258318e-05, + "loss": 0.3754, + "step": 2588 + }, + { + "epoch": 1.08, + "learning_rate": 1.9422115416297377e-05, + "loss": 0.4024, + "step": 2589 + }, + { + "epoch": 1.08, + "learning_rate": 1.9421547287506085e-05, + "loss": 0.358, + "step": 2590 + }, + { + "epoch": 1.08, + "learning_rate": 1.942097888790078e-05, + "loss": 0.3966, + "step": 2591 + }, + { + "epoch": 1.09, + "learning_rate": 1.9420410217497795e-05, + "loss": 0.3881, + "step": 2592 + }, + { + "epoch": 1.09, + "learning_rate": 1.941984127631348e-05, + "loss": 0.3901, + "step": 2593 + }, + { + "epoch": 1.09, + "learning_rate": 1.9419272064364186e-05, + "loss": 0.4057, + "step": 2594 + }, + { + "epoch": 1.09, + "learning_rate": 1.9418702581666276e-05, + "loss": 0.351, + "step": 2595 + }, + { + "epoch": 1.09, + "learning_rate": 1.9418132828236114e-05, + "loss": 0.3222, + "step": 2596 + }, + { + "epoch": 1.09, + "learning_rate": 1.9417562804090086e-05, + "loss": 0.3725, + "step": 2597 + }, + { + "epoch": 1.09, + "learning_rate": 1.941699250924457e-05, + "loss": 0.4544, + "step": 2598 + }, + { + "epoch": 1.09, + "learning_rate": 1.941642194371596e-05, + "loss": 0.408, + "step": 2599 + }, + { + "epoch": 1.09, + "learning_rate": 1.941585110752066e-05, + "loss": 0.3905, + "step": 2600 + }, + { + "epoch": 1.09, + "learning_rate": 1.9415280000675074e-05, + "loss": 0.39, + "step": 2601 + }, + { + "epoch": 1.09, + "learning_rate": 1.9414708623195618e-05, + "loss": 0.363, + "step": 2602 + }, + { + "epoch": 1.09, + "learning_rate": 1.9414136975098718e-05, + "loss": 0.3951, + "step": 2603 + }, + { + "epoch": 1.09, + "learning_rate": 1.9413565056400804e-05, + "loss": 0.3983, + "step": 2604 + }, + { + "epoch": 1.09, + "learning_rate": 1.9412992867118318e-05, + "loss": 0.3811, + "step": 2605 + }, + { + "epoch": 1.09, + "learning_rate": 1.9412420407267704e-05, + "loss": 0.3948, + "step": 2606 + }, + { + "epoch": 1.09, + "learning_rate": 1.9411847676865415e-05, + "loss": 0.3897, + "step": 2607 + }, + { + "epoch": 1.09, + "learning_rate": 1.9411274675927922e-05, + "loss": 0.3905, + "step": 2608 + }, + { + "epoch": 1.09, + "learning_rate": 1.9410701404471686e-05, + "loss": 0.4154, + "step": 2609 + }, + { + "epoch": 1.09, + "learning_rate": 1.941012786251319e-05, + "loss": 0.3532, + "step": 2610 + }, + { + "epoch": 1.09, + "learning_rate": 1.9409554050068916e-05, + "loss": 0.3639, + "step": 2611 + }, + { + "epoch": 1.09, + "learning_rate": 1.9408979967155364e-05, + "loss": 0.3617, + "step": 2612 + }, + { + "epoch": 1.09, + "learning_rate": 1.940840561378903e-05, + "loss": 0.3866, + "step": 2613 + }, + { + "epoch": 1.09, + "learning_rate": 1.940783098998643e-05, + "loss": 0.3641, + "step": 2614 + }, + { + "epoch": 1.09, + "learning_rate": 1.9407256095764074e-05, + "loss": 0.3998, + "step": 2615 + }, + { + "epoch": 1.1, + "learning_rate": 1.9406680931138488e-05, + "loss": 0.3803, + "step": 2616 + }, + { + "epoch": 1.1, + "learning_rate": 1.940610549612621e-05, + "loss": 0.3922, + "step": 2617 + }, + { + "epoch": 1.1, + "learning_rate": 1.940552979074377e-05, + "loss": 0.372, + "step": 2618 + }, + { + "epoch": 1.1, + "learning_rate": 1.940495381500773e-05, + "loss": 0.3683, + "step": 2619 + }, + { + "epoch": 1.1, + "learning_rate": 1.9404377568934633e-05, + "loss": 0.4032, + "step": 2620 + }, + { + "epoch": 1.1, + "learning_rate": 1.940380105254105e-05, + "loss": 0.3778, + "step": 2621 + }, + { + "epoch": 1.1, + "learning_rate": 1.9403224265843547e-05, + "loss": 0.3555, + "step": 2622 + }, + { + "epoch": 1.1, + "learning_rate": 1.9402647208858712e-05, + "loss": 0.3768, + "step": 2623 + }, + { + "epoch": 1.1, + "learning_rate": 1.9402069881603123e-05, + "loss": 0.3493, + "step": 2624 + }, + { + "epoch": 1.1, + "learning_rate": 1.9401492284093382e-05, + "loss": 0.4207, + "step": 2625 + }, + { + "epoch": 1.1, + "learning_rate": 1.9400914416346083e-05, + "loss": 0.3887, + "step": 2626 + }, + { + "epoch": 1.1, + "learning_rate": 1.9400336278377845e-05, + "loss": 0.3926, + "step": 2627 + }, + { + "epoch": 1.1, + "learning_rate": 1.939975787020528e-05, + "loss": 0.3902, + "step": 2628 + }, + { + "epoch": 1.1, + "learning_rate": 1.9399179191845022e-05, + "loss": 0.3577, + "step": 2629 + }, + { + "epoch": 1.1, + "learning_rate": 1.9398600243313695e-05, + "loss": 0.3764, + "step": 2630 + }, + { + "epoch": 1.1, + "learning_rate": 1.939802102462794e-05, + "loss": 0.3581, + "step": 2631 + }, + { + "epoch": 1.1, + "learning_rate": 1.9397441535804417e-05, + "loss": 0.3631, + "step": 2632 + }, + { + "epoch": 1.1, + "learning_rate": 1.939686177685977e-05, + "loss": 0.3528, + "step": 2633 + }, + { + "epoch": 1.1, + "learning_rate": 1.939628174781068e-05, + "loss": 0.3927, + "step": 2634 + }, + { + "epoch": 1.1, + "learning_rate": 1.9395701448673797e-05, + "loss": 0.3593, + "step": 2635 + }, + { + "epoch": 1.1, + "learning_rate": 1.939512087946582e-05, + "loss": 0.3985, + "step": 2636 + }, + { + "epoch": 1.1, + "learning_rate": 1.9394540040203428e-05, + "loss": 0.3532, + "step": 2637 + }, + { + "epoch": 1.1, + "learning_rate": 1.9393958930903323e-05, + "loss": 0.3842, + "step": 2638 + }, + { + "epoch": 1.1, + "learning_rate": 1.93933775515822e-05, + "loss": 0.3824, + "step": 2639 + }, + { + "epoch": 1.11, + "learning_rate": 1.939279590225678e-05, + "loss": 0.3778, + "step": 2640 + }, + { + "epoch": 1.11, + "learning_rate": 1.9392213982943774e-05, + "loss": 0.3495, + "step": 2641 + }, + { + "epoch": 1.11, + "learning_rate": 1.9391631793659908e-05, + "loss": 0.3768, + "step": 2642 + }, + { + "epoch": 1.11, + "learning_rate": 1.9391049334421925e-05, + "loss": 0.3629, + "step": 2643 + }, + { + "epoch": 1.11, + "learning_rate": 1.939046660524656e-05, + "loss": 0.375, + "step": 2644 + }, + { + "epoch": 1.11, + "learning_rate": 1.938988360615057e-05, + "loss": 0.3699, + "step": 2645 + }, + { + "epoch": 1.11, + "learning_rate": 1.93893003371507e-05, + "loss": 0.3716, + "step": 2646 + }, + { + "epoch": 1.11, + "learning_rate": 1.9388716798263728e-05, + "loss": 0.4273, + "step": 2647 + }, + { + "epoch": 1.11, + "learning_rate": 1.9388132989506422e-05, + "loss": 0.3377, + "step": 2648 + }, + { + "epoch": 1.11, + "learning_rate": 1.9387548910895567e-05, + "loss": 0.3908, + "step": 2649 + }, + { + "epoch": 1.11, + "learning_rate": 1.9386964562447947e-05, + "loss": 0.3764, + "step": 2650 + }, + { + "epoch": 1.11, + "learning_rate": 1.938637994418036e-05, + "loss": 0.4319, + "step": 2651 + }, + { + "epoch": 1.11, + "learning_rate": 1.9385795056109614e-05, + "loss": 0.3947, + "step": 2652 + }, + { + "epoch": 1.11, + "learning_rate": 1.938520989825251e-05, + "loss": 0.3839, + "step": 2653 + }, + { + "epoch": 1.11, + "learning_rate": 1.9384624470625884e-05, + "loss": 0.4071, + "step": 2654 + }, + { + "epoch": 1.11, + "learning_rate": 1.9384038773246555e-05, + "loss": 0.3845, + "step": 2655 + }, + { + "epoch": 1.11, + "learning_rate": 1.9383452806131356e-05, + "loss": 0.4514, + "step": 2656 + }, + { + "epoch": 1.11, + "learning_rate": 1.9382866569297133e-05, + "loss": 0.34, + "step": 2657 + }, + { + "epoch": 1.11, + "learning_rate": 1.9382280062760738e-05, + "loss": 0.3823, + "step": 2658 + }, + { + "epoch": 1.11, + "learning_rate": 1.9381693286539024e-05, + "loss": 0.381, + "step": 2659 + }, + { + "epoch": 1.11, + "learning_rate": 1.9381106240648867e-05, + "loss": 0.3852, + "step": 2660 + }, + { + "epoch": 1.11, + "learning_rate": 1.9380518925107133e-05, + "loss": 0.4091, + "step": 2661 + }, + { + "epoch": 1.11, + "learning_rate": 1.9379931339930705e-05, + "loss": 0.3806, + "step": 2662 + }, + { + "epoch": 1.11, + "learning_rate": 1.9379343485136478e-05, + "loss": 0.3209, + "step": 2663 + }, + { + "epoch": 1.12, + "learning_rate": 1.9378755360741342e-05, + "loss": 0.3436, + "step": 2664 + }, + { + "epoch": 1.12, + "learning_rate": 1.9378166966762207e-05, + "loss": 0.392, + "step": 2665 + }, + { + "epoch": 1.12, + "learning_rate": 1.9377578303215984e-05, + "loss": 0.3796, + "step": 2666 + }, + { + "epoch": 1.12, + "learning_rate": 1.9376989370119595e-05, + "loss": 0.3759, + "step": 2667 + }, + { + "epoch": 1.12, + "learning_rate": 1.937640016748997e-05, + "loss": 0.4186, + "step": 2668 + }, + { + "epoch": 1.12, + "learning_rate": 1.937581069534404e-05, + "loss": 0.3675, + "step": 2669 + }, + { + "epoch": 1.12, + "learning_rate": 1.937522095369875e-05, + "loss": 0.3504, + "step": 2670 + }, + { + "epoch": 1.12, + "learning_rate": 1.9374630942571055e-05, + "loss": 0.3958, + "step": 2671 + }, + { + "epoch": 1.12, + "learning_rate": 1.9374040661977913e-05, + "loss": 0.3544, + "step": 2672 + }, + { + "epoch": 1.12, + "learning_rate": 1.937345011193629e-05, + "loss": 0.376, + "step": 2673 + }, + { + "epoch": 1.12, + "learning_rate": 1.9372859292463163e-05, + "loss": 0.3849, + "step": 2674 + }, + { + "epoch": 1.12, + "learning_rate": 1.9372268203575512e-05, + "loss": 0.3558, + "step": 2675 + }, + { + "epoch": 1.12, + "learning_rate": 1.9371676845290327e-05, + "loss": 0.3714, + "step": 2676 + }, + { + "epoch": 1.12, + "learning_rate": 1.9371085217624608e-05, + "loss": 0.4111, + "step": 2677 + }, + { + "epoch": 1.12, + "learning_rate": 1.937049332059536e-05, + "loss": 0.3657, + "step": 2678 + }, + { + "epoch": 1.12, + "learning_rate": 1.9369901154219595e-05, + "loss": 0.3718, + "step": 2679 + }, + { + "epoch": 1.12, + "learning_rate": 1.9369308718514342e-05, + "loss": 0.3688, + "step": 2680 + }, + { + "epoch": 1.12, + "learning_rate": 1.9368716013496616e-05, + "loss": 0.4108, + "step": 2681 + }, + { + "epoch": 1.12, + "learning_rate": 1.9368123039183468e-05, + "loss": 0.3761, + "step": 2682 + }, + { + "epoch": 1.12, + "learning_rate": 1.9367529795591935e-05, + "loss": 0.3954, + "step": 2683 + }, + { + "epoch": 1.12, + "learning_rate": 1.9366936282739067e-05, + "loss": 0.3762, + "step": 2684 + }, + { + "epoch": 1.12, + "learning_rate": 1.9366342500641933e-05, + "loss": 0.3663, + "step": 2685 + }, + { + "epoch": 1.12, + "learning_rate": 1.9365748449317593e-05, + "loss": 0.3908, + "step": 2686 + }, + { + "epoch": 1.12, + "learning_rate": 1.9365154128783126e-05, + "loss": 0.3524, + "step": 2687 + }, + { + "epoch": 1.13, + "learning_rate": 1.9364559539055612e-05, + "loss": 0.364, + "step": 2688 + }, + { + "epoch": 1.13, + "learning_rate": 1.9363964680152147e-05, + "loss": 0.3783, + "step": 2689 + }, + { + "epoch": 1.13, + "learning_rate": 1.936336955208982e-05, + "loss": 0.3973, + "step": 2690 + }, + { + "epoch": 1.13, + "learning_rate": 1.936277415488575e-05, + "loss": 0.3912, + "step": 2691 + }, + { + "epoch": 1.13, + "learning_rate": 1.9362178488557043e-05, + "loss": 0.4189, + "step": 2692 + }, + { + "epoch": 1.13, + "learning_rate": 1.9361582553120824e-05, + "loss": 0.3805, + "step": 2693 + }, + { + "epoch": 1.13, + "learning_rate": 1.9360986348594226e-05, + "loss": 0.3507, + "step": 2694 + }, + { + "epoch": 1.13, + "learning_rate": 1.9360389874994374e-05, + "loss": 0.4058, + "step": 2695 + }, + { + "epoch": 1.13, + "learning_rate": 1.9359793132338427e-05, + "loss": 0.3807, + "step": 2696 + }, + { + "epoch": 1.13, + "learning_rate": 1.935919612064353e-05, + "loss": 0.4055, + "step": 2697 + }, + { + "epoch": 1.13, + "learning_rate": 1.9358598839926848e-05, + "loss": 0.3797, + "step": 2698 + }, + { + "epoch": 1.13, + "learning_rate": 1.9358001290205542e-05, + "loss": 0.3917, + "step": 2699 + }, + { + "epoch": 1.13, + "learning_rate": 1.9357403471496797e-05, + "loss": 0.3894, + "step": 2700 + }, + { + "epoch": 1.13, + "learning_rate": 1.9356805383817792e-05, + "loss": 0.3666, + "step": 2701 + }, + { + "epoch": 1.13, + "learning_rate": 1.935620702718572e-05, + "loss": 0.4063, + "step": 2702 + }, + { + "epoch": 1.13, + "learning_rate": 1.9355608401617782e-05, + "loss": 0.3768, + "step": 2703 + }, + { + "epoch": 1.13, + "learning_rate": 1.935500950713118e-05, + "loss": 0.3813, + "step": 2704 + }, + { + "epoch": 1.13, + "learning_rate": 1.9354410343743127e-05, + "loss": 0.3715, + "step": 2705 + }, + { + "epoch": 1.13, + "learning_rate": 1.9353810911470855e-05, + "loss": 0.3931, + "step": 2706 + }, + { + "epoch": 1.13, + "learning_rate": 1.935321121033159e-05, + "loss": 0.41, + "step": 2707 + }, + { + "epoch": 1.13, + "learning_rate": 1.9352611240342565e-05, + "loss": 0.4062, + "step": 2708 + }, + { + "epoch": 1.13, + "learning_rate": 1.935201100152103e-05, + "loss": 0.3632, + "step": 2709 + }, + { + "epoch": 1.13, + "learning_rate": 1.9351410493884234e-05, + "loss": 0.3832, + "step": 2710 + }, + { + "epoch": 1.14, + "learning_rate": 1.9350809717449448e-05, + "loss": 0.3753, + "step": 2711 + }, + { + "epoch": 1.14, + "learning_rate": 1.935020867223393e-05, + "loss": 0.346, + "step": 2712 + }, + { + "epoch": 1.14, + "learning_rate": 1.9349607358254965e-05, + "loss": 0.3275, + "step": 2713 + }, + { + "epoch": 1.14, + "learning_rate": 1.934900577552983e-05, + "loss": 0.4153, + "step": 2714 + }, + { + "epoch": 1.14, + "learning_rate": 1.934840392407582e-05, + "loss": 0.3351, + "step": 2715 + }, + { + "epoch": 1.14, + "learning_rate": 1.9347801803910236e-05, + "loss": 0.4013, + "step": 2716 + }, + { + "epoch": 1.14, + "learning_rate": 1.9347199415050383e-05, + "loss": 0.37, + "step": 2717 + }, + { + "epoch": 1.14, + "learning_rate": 1.934659675751358e-05, + "loss": 0.3452, + "step": 2718 + }, + { + "epoch": 1.14, + "learning_rate": 1.9345993831317147e-05, + "loss": 0.307, + "step": 2719 + }, + { + "epoch": 1.14, + "learning_rate": 1.9345390636478414e-05, + "loss": 0.3732, + "step": 2720 + }, + { + "epoch": 1.14, + "learning_rate": 1.934478717301472e-05, + "loss": 0.4037, + "step": 2721 + }, + { + "epoch": 1.14, + "learning_rate": 1.9344183440943408e-05, + "loss": 0.4228, + "step": 2722 + }, + { + "epoch": 1.14, + "learning_rate": 1.934357944028184e-05, + "loss": 0.3822, + "step": 2723 + }, + { + "epoch": 1.14, + "learning_rate": 1.934297517104737e-05, + "loss": 0.3925, + "step": 2724 + }, + { + "epoch": 1.14, + "learning_rate": 1.9342370633257367e-05, + "loss": 0.3673, + "step": 2725 + }, + { + "epoch": 1.14, + "learning_rate": 1.9341765826929216e-05, + "loss": 0.4204, + "step": 2726 + }, + { + "epoch": 1.14, + "learning_rate": 1.9341160752080294e-05, + "loss": 0.429, + "step": 2727 + }, + { + "epoch": 1.14, + "learning_rate": 1.9340555408727993e-05, + "loss": 0.3993, + "step": 2728 + }, + { + "epoch": 1.14, + "learning_rate": 1.9339949796889717e-05, + "loss": 0.4509, + "step": 2729 + }, + { + "epoch": 1.14, + "learning_rate": 1.933934391658287e-05, + "loss": 0.3768, + "step": 2730 + }, + { + "epoch": 1.14, + "learning_rate": 1.933873776782487e-05, + "loss": 0.3504, + "step": 2731 + }, + { + "epoch": 1.14, + "learning_rate": 1.9338131350633143e-05, + "loss": 0.4084, + "step": 2732 + }, + { + "epoch": 1.14, + "learning_rate": 1.9337524665025114e-05, + "loss": 0.3781, + "step": 2733 + }, + { + "epoch": 1.14, + "learning_rate": 1.9336917711018227e-05, + "loss": 0.4215, + "step": 2734 + }, + { + "epoch": 1.15, + "learning_rate": 1.9336310488629926e-05, + "loss": 0.4066, + "step": 2735 + }, + { + "epoch": 1.15, + "learning_rate": 1.9335702997877664e-05, + "loss": 0.4147, + "step": 2736 + }, + { + "epoch": 1.15, + "learning_rate": 1.9335095238778903e-05, + "loss": 0.3655, + "step": 2737 + }, + { + "epoch": 1.15, + "learning_rate": 1.9334487211351113e-05, + "loss": 0.3868, + "step": 2738 + }, + { + "epoch": 1.15, + "learning_rate": 1.933387891561177e-05, + "loss": 0.3515, + "step": 2739 + }, + { + "epoch": 1.15, + "learning_rate": 1.9333270351578367e-05, + "loss": 0.3973, + "step": 2740 + }, + { + "epoch": 1.15, + "learning_rate": 1.9332661519268384e-05, + "loss": 0.4615, + "step": 2741 + }, + { + "epoch": 1.15, + "learning_rate": 1.933205241869933e-05, + "loss": 0.4009, + "step": 2742 + }, + { + "epoch": 1.15, + "learning_rate": 1.9331443049888706e-05, + "loss": 0.3969, + "step": 2743 + }, + { + "epoch": 1.15, + "learning_rate": 1.9330833412854035e-05, + "loss": 0.3819, + "step": 2744 + }, + { + "epoch": 1.15, + "learning_rate": 1.933022350761284e-05, + "loss": 0.3491, + "step": 2745 + }, + { + "epoch": 1.15, + "learning_rate": 1.9329613334182644e-05, + "loss": 0.3933, + "step": 2746 + }, + { + "epoch": 1.15, + "learning_rate": 1.9329002892580995e-05, + "loss": 0.3607, + "step": 2747 + }, + { + "epoch": 1.15, + "learning_rate": 1.932839218282544e-05, + "loss": 0.3915, + "step": 2748 + }, + { + "epoch": 1.15, + "learning_rate": 1.9327781204933522e-05, + "loss": 0.355, + "step": 2749 + }, + { + "epoch": 1.15, + "learning_rate": 1.9327169958922813e-05, + "loss": 0.4151, + "step": 2750 + }, + { + "epoch": 1.15, + "learning_rate": 1.9326558444810883e-05, + "loss": 0.3827, + "step": 2751 + }, + { + "epoch": 1.15, + "learning_rate": 1.9325946662615304e-05, + "loss": 0.4245, + "step": 2752 + }, + { + "epoch": 1.15, + "learning_rate": 1.932533461235367e-05, + "loss": 0.3551, + "step": 2753 + }, + { + "epoch": 1.15, + "learning_rate": 1.932472229404356e-05, + "loss": 0.3895, + "step": 2754 + }, + { + "epoch": 1.15, + "learning_rate": 1.9324109707702585e-05, + "loss": 0.3821, + "step": 2755 + }, + { + "epoch": 1.15, + "learning_rate": 1.932349685334835e-05, + "loss": 0.3928, + "step": 2756 + }, + { + "epoch": 1.15, + "learning_rate": 1.9322883730998474e-05, + "loss": 0.3673, + "step": 2757 + }, + { + "epoch": 1.15, + "learning_rate": 1.9322270340670574e-05, + "loss": 0.3496, + "step": 2758 + }, + { + "epoch": 1.16, + "learning_rate": 1.9321656682382287e-05, + "loss": 0.3826, + "step": 2759 + }, + { + "epoch": 1.16, + "learning_rate": 1.9321042756151254e-05, + "loss": 0.3732, + "step": 2760 + }, + { + "epoch": 1.16, + "learning_rate": 1.9320428561995114e-05, + "loss": 0.3547, + "step": 2761 + }, + { + "epoch": 1.16, + "learning_rate": 1.931981409993153e-05, + "loss": 0.337, + "step": 2762 + }, + { + "epoch": 1.16, + "learning_rate": 1.9319199369978157e-05, + "loss": 0.3734, + "step": 2763 + }, + { + "epoch": 1.16, + "learning_rate": 1.9318584372152668e-05, + "loss": 0.3379, + "step": 2764 + }, + { + "epoch": 1.16, + "learning_rate": 1.9317969106472742e-05, + "loss": 0.3891, + "step": 2765 + }, + { + "epoch": 1.16, + "learning_rate": 1.9317353572956058e-05, + "loss": 0.4175, + "step": 2766 + }, + { + "epoch": 1.16, + "learning_rate": 1.931673777162032e-05, + "loss": 0.3745, + "step": 2767 + }, + { + "epoch": 1.16, + "learning_rate": 1.9316121702483218e-05, + "loss": 0.421, + "step": 2768 + }, + { + "epoch": 1.16, + "learning_rate": 1.9315505365562464e-05, + "loss": 0.3868, + "step": 2769 + }, + { + "epoch": 1.16, + "learning_rate": 1.931488876087578e-05, + "loss": 0.3765, + "step": 2770 + }, + { + "epoch": 1.16, + "learning_rate": 1.9314271888440878e-05, + "loss": 0.3419, + "step": 2771 + }, + { + "epoch": 1.16, + "learning_rate": 1.9313654748275497e-05, + "loss": 0.4157, + "step": 2772 + }, + { + "epoch": 1.16, + "learning_rate": 1.931303734039738e-05, + "loss": 0.4043, + "step": 2773 + }, + { + "epoch": 1.16, + "learning_rate": 1.9312419664824268e-05, + "loss": 0.4398, + "step": 2774 + }, + { + "epoch": 1.16, + "learning_rate": 1.9311801721573913e-05, + "loss": 0.39, + "step": 2775 + }, + { + "epoch": 1.16, + "learning_rate": 1.931118351066408e-05, + "loss": 0.3564, + "step": 2776 + }, + { + "epoch": 1.16, + "learning_rate": 1.9310565032112542e-05, + "loss": 0.3823, + "step": 2777 + }, + { + "epoch": 1.16, + "learning_rate": 1.9309946285937075e-05, + "loss": 0.3924, + "step": 2778 + }, + { + "epoch": 1.16, + "learning_rate": 1.9309327272155462e-05, + "loss": 0.3895, + "step": 2779 + }, + { + "epoch": 1.16, + "learning_rate": 1.9308707990785497e-05, + "loss": 0.3785, + "step": 2780 + }, + { + "epoch": 1.16, + "learning_rate": 1.9308088441844982e-05, + "loss": 0.3857, + "step": 2781 + }, + { + "epoch": 1.16, + "learning_rate": 1.9307468625351723e-05, + "loss": 0.4335, + "step": 2782 + }, + { + "epoch": 1.17, + "learning_rate": 1.930684854132354e-05, + "loss": 0.3893, + "step": 2783 + }, + { + "epoch": 1.17, + "learning_rate": 1.9306228189778255e-05, + "loss": 0.4093, + "step": 2784 + }, + { + "epoch": 1.17, + "learning_rate": 1.9305607570733696e-05, + "loss": 0.4062, + "step": 2785 + }, + { + "epoch": 1.17, + "learning_rate": 1.930498668420771e-05, + "loss": 0.3321, + "step": 2786 + }, + { + "epoch": 1.17, + "learning_rate": 1.9304365530218134e-05, + "loss": 0.4056, + "step": 2787 + }, + { + "epoch": 1.17, + "learning_rate": 1.930374410878283e-05, + "loss": 0.4217, + "step": 2788 + }, + { + "epoch": 1.17, + "learning_rate": 1.9303122419919655e-05, + "loss": 0.3904, + "step": 2789 + }, + { + "epoch": 1.17, + "learning_rate": 1.9302500463646483e-05, + "loss": 0.3738, + "step": 2790 + }, + { + "epoch": 1.17, + "learning_rate": 1.930187823998119e-05, + "loss": 0.3978, + "step": 2791 + }, + { + "epoch": 1.17, + "learning_rate": 1.930125574894166e-05, + "loss": 0.3531, + "step": 2792 + }, + { + "epoch": 1.17, + "learning_rate": 1.930063299054579e-05, + "loss": 0.3876, + "step": 2793 + }, + { + "epoch": 1.17, + "learning_rate": 1.9300009964811475e-05, + "loss": 0.4144, + "step": 2794 + }, + { + "epoch": 1.17, + "learning_rate": 1.9299386671756628e-05, + "loss": 0.4041, + "step": 2795 + }, + { + "epoch": 1.17, + "learning_rate": 1.9298763111399162e-05, + "loss": 0.4336, + "step": 2796 + }, + { + "epoch": 1.17, + "learning_rate": 1.9298139283757004e-05, + "loss": 0.3942, + "step": 2797 + }, + { + "epoch": 1.17, + "learning_rate": 1.9297515188848083e-05, + "loss": 0.3601, + "step": 2798 + }, + { + "epoch": 1.17, + "learning_rate": 1.9296890826690337e-05, + "loss": 0.3929, + "step": 2799 + }, + { + "epoch": 1.17, + "learning_rate": 1.9296266197301716e-05, + "loss": 0.4062, + "step": 2800 + }, + { + "epoch": 1.17, + "learning_rate": 1.9295641300700173e-05, + "loss": 0.354, + "step": 2801 + }, + { + "epoch": 1.17, + "learning_rate": 1.9295016136903666e-05, + "loss": 0.4031, + "step": 2802 + }, + { + "epoch": 1.17, + "learning_rate": 1.929439070593017e-05, + "loss": 0.4086, + "step": 2803 + }, + { + "epoch": 1.17, + "learning_rate": 1.9293765007797662e-05, + "loss": 0.4178, + "step": 2804 + }, + { + "epoch": 1.17, + "learning_rate": 1.9293139042524124e-05, + "loss": 0.3496, + "step": 2805 + }, + { + "epoch": 1.17, + "learning_rate": 1.9292512810127554e-05, + "loss": 0.397, + "step": 2806 + }, + { + "epoch": 1.18, + "learning_rate": 1.9291886310625948e-05, + "loss": 0.3759, + "step": 2807 + }, + { + "epoch": 1.18, + "learning_rate": 1.9291259544037313e-05, + "loss": 0.3958, + "step": 2808 + }, + { + "epoch": 1.18, + "learning_rate": 1.9290632510379667e-05, + "loss": 0.404, + "step": 2809 + }, + { + "epoch": 1.18, + "learning_rate": 1.9290005209671036e-05, + "loss": 0.394, + "step": 2810 + }, + { + "epoch": 1.18, + "learning_rate": 1.928937764192945e-05, + "loss": 0.4051, + "step": 2811 + }, + { + "epoch": 1.18, + "learning_rate": 1.9288749807172947e-05, + "loss": 0.4203, + "step": 2812 + }, + { + "epoch": 1.18, + "learning_rate": 1.9288121705419572e-05, + "loss": 0.3902, + "step": 2813 + }, + { + "epoch": 1.18, + "learning_rate": 1.928749333668738e-05, + "loss": 0.3657, + "step": 2814 + }, + { + "epoch": 1.18, + "learning_rate": 1.9286864700994433e-05, + "loss": 0.3203, + "step": 2815 + }, + { + "epoch": 1.18, + "learning_rate": 1.9286235798358803e-05, + "loss": 0.3667, + "step": 2816 + }, + { + "epoch": 1.18, + "learning_rate": 1.9285606628798564e-05, + "loss": 0.4075, + "step": 2817 + }, + { + "epoch": 1.18, + "learning_rate": 1.9284977192331807e-05, + "loss": 0.4056, + "step": 2818 + }, + { + "epoch": 1.18, + "learning_rate": 1.9284347488976616e-05, + "loss": 0.4313, + "step": 2819 + }, + { + "epoch": 1.18, + "learning_rate": 1.9283717518751095e-05, + "loss": 0.3834, + "step": 2820 + }, + { + "epoch": 1.18, + "learning_rate": 1.9283087281673352e-05, + "loss": 0.3633, + "step": 2821 + }, + { + "epoch": 1.18, + "learning_rate": 1.9282456777761505e-05, + "loss": 0.3872, + "step": 2822 + }, + { + "epoch": 1.18, + "learning_rate": 1.9281826007033674e-05, + "loss": 0.4124, + "step": 2823 + }, + { + "epoch": 1.18, + "learning_rate": 1.928119496950799e-05, + "loss": 0.4013, + "step": 2824 + }, + { + "epoch": 1.18, + "learning_rate": 1.9280563665202595e-05, + "loss": 0.4354, + "step": 2825 + }, + { + "epoch": 1.18, + "learning_rate": 1.927993209413563e-05, + "loss": 0.3694, + "step": 2826 + }, + { + "epoch": 1.18, + "learning_rate": 1.927930025632525e-05, + "loss": 0.3888, + "step": 2827 + }, + { + "epoch": 1.18, + "learning_rate": 1.9278668151789627e-05, + "loss": 0.4104, + "step": 2828 + }, + { + "epoch": 1.18, + "learning_rate": 1.9278035780546912e-05, + "loss": 0.4169, + "step": 2829 + }, + { + "epoch": 1.18, + "learning_rate": 1.9277403142615297e-05, + "loss": 0.385, + "step": 2830 + }, + { + "epoch": 1.19, + "learning_rate": 1.927677023801296e-05, + "loss": 0.4508, + "step": 2831 + }, + { + "epoch": 1.19, + "learning_rate": 1.9276137066758097e-05, + "loss": 0.4015, + "step": 2832 + }, + { + "epoch": 1.19, + "learning_rate": 1.9275503628868902e-05, + "loss": 0.4012, + "step": 2833 + }, + { + "epoch": 1.19, + "learning_rate": 1.9274869924363585e-05, + "loss": 0.3566, + "step": 2834 + }, + { + "epoch": 1.19, + "learning_rate": 1.9274235953260367e-05, + "loss": 0.3656, + "step": 2835 + }, + { + "epoch": 1.19, + "learning_rate": 1.9273601715577463e-05, + "loss": 0.3389, + "step": 2836 + }, + { + "epoch": 1.19, + "learning_rate": 1.9272967211333107e-05, + "loss": 0.3845, + "step": 2837 + }, + { + "epoch": 1.19, + "learning_rate": 1.927233244054554e-05, + "loss": 0.372, + "step": 2838 + }, + { + "epoch": 1.19, + "learning_rate": 1.9271697403233e-05, + "loss": 0.4459, + "step": 2839 + }, + { + "epoch": 1.19, + "learning_rate": 1.9271062099413752e-05, + "loss": 0.3977, + "step": 2840 + }, + { + "epoch": 1.19, + "learning_rate": 1.9270426529106048e-05, + "loss": 0.3697, + "step": 2841 + }, + { + "epoch": 1.19, + "learning_rate": 1.926979069232816e-05, + "loss": 0.3596, + "step": 2842 + }, + { + "epoch": 1.19, + "learning_rate": 1.9269154589098363e-05, + "loss": 0.388, + "step": 2843 + }, + { + "epoch": 1.19, + "learning_rate": 1.9268518219434943e-05, + "loss": 0.4069, + "step": 2844 + }, + { + "epoch": 1.19, + "learning_rate": 1.9267881583356192e-05, + "loss": 0.367, + "step": 2845 + }, + { + "epoch": 1.19, + "learning_rate": 1.926724468088041e-05, + "loss": 0.3711, + "step": 2846 + }, + { + "epoch": 1.19, + "learning_rate": 1.9266607512025903e-05, + "loss": 0.3994, + "step": 2847 + }, + { + "epoch": 1.19, + "learning_rate": 1.9265970076810987e-05, + "loss": 0.4422, + "step": 2848 + }, + { + "epoch": 1.19, + "learning_rate": 1.9265332375253984e-05, + "loss": 0.3868, + "step": 2849 + }, + { + "epoch": 1.19, + "learning_rate": 1.926469440737322e-05, + "loss": 0.4381, + "step": 2850 + }, + { + "epoch": 1.19, + "learning_rate": 1.9264056173187037e-05, + "loss": 0.4296, + "step": 2851 + }, + { + "epoch": 1.19, + "learning_rate": 1.9263417672713786e-05, + "loss": 0.3709, + "step": 2852 + }, + { + "epoch": 1.19, + "learning_rate": 1.9262778905971808e-05, + "loss": 0.4049, + "step": 2853 + }, + { + "epoch": 1.19, + "learning_rate": 1.9262139872979474e-05, + "loss": 0.3886, + "step": 2854 + }, + { + "epoch": 1.2, + "learning_rate": 1.9261500573755146e-05, + "loss": 0.3822, + "step": 2855 + }, + { + "epoch": 1.2, + "learning_rate": 1.9260861008317203e-05, + "loss": 0.3676, + "step": 2856 + }, + { + "epoch": 1.2, + "learning_rate": 1.9260221176684025e-05, + "loss": 0.386, + "step": 2857 + }, + { + "epoch": 1.2, + "learning_rate": 1.925958107887401e-05, + "loss": 0.3864, + "step": 2858 + }, + { + "epoch": 1.2, + "learning_rate": 1.9258940714905554e-05, + "loss": 0.3853, + "step": 2859 + }, + { + "epoch": 1.2, + "learning_rate": 1.9258300084797063e-05, + "loss": 0.4195, + "step": 2860 + }, + { + "epoch": 1.2, + "learning_rate": 1.925765918856695e-05, + "loss": 0.3893, + "step": 2861 + }, + { + "epoch": 1.2, + "learning_rate": 1.9257018026233642e-05, + "loss": 0.4311, + "step": 2862 + }, + { + "epoch": 1.2, + "learning_rate": 1.9256376597815565e-05, + "loss": 0.3936, + "step": 2863 + }, + { + "epoch": 1.2, + "learning_rate": 1.9255734903331157e-05, + "loss": 0.4135, + "step": 2864 + }, + { + "epoch": 1.2, + "learning_rate": 1.9255092942798858e-05, + "loss": 0.3841, + "step": 2865 + }, + { + "epoch": 1.2, + "learning_rate": 1.9254450716237128e-05, + "loss": 0.394, + "step": 2866 + }, + { + "epoch": 1.2, + "learning_rate": 1.9253808223664427e-05, + "loss": 0.4184, + "step": 2867 + }, + { + "epoch": 1.2, + "learning_rate": 1.9253165465099217e-05, + "loss": 0.3678, + "step": 2868 + }, + { + "epoch": 1.2, + "learning_rate": 1.925252244055998e-05, + "loss": 0.4486, + "step": 2869 + }, + { + "epoch": 1.2, + "learning_rate": 1.9251879150065192e-05, + "loss": 0.3573, + "step": 2870 + }, + { + "epoch": 1.2, + "learning_rate": 1.925123559363335e-05, + "loss": 0.3508, + "step": 2871 + }, + { + "epoch": 1.2, + "learning_rate": 1.925059177128295e-05, + "loss": 0.3864, + "step": 2872 + }, + { + "epoch": 1.2, + "learning_rate": 1.9249947683032497e-05, + "loss": 0.3628, + "step": 2873 + }, + { + "epoch": 1.2, + "learning_rate": 1.9249303328900505e-05, + "loss": 0.3605, + "step": 2874 + }, + { + "epoch": 1.2, + "learning_rate": 1.92486587089055e-05, + "loss": 0.391, + "step": 2875 + }, + { + "epoch": 1.2, + "learning_rate": 1.9248013823066006e-05, + "loss": 0.4173, + "step": 2876 + }, + { + "epoch": 1.2, + "learning_rate": 1.924736867140056e-05, + "loss": 0.413, + "step": 2877 + }, + { + "epoch": 1.2, + "learning_rate": 1.9246723253927714e-05, + "loss": 0.4276, + "step": 2878 + }, + { + "epoch": 1.21, + "learning_rate": 1.9246077570666008e-05, + "loss": 0.4009, + "step": 2879 + }, + { + "epoch": 1.21, + "learning_rate": 1.9245431621634006e-05, + "loss": 0.3918, + "step": 2880 + }, + { + "epoch": 1.21, + "learning_rate": 1.924478540685028e-05, + "loss": 0.3812, + "step": 2881 + }, + { + "epoch": 1.21, + "learning_rate": 1.9244138926333396e-05, + "loss": 0.4158, + "step": 2882 + }, + { + "epoch": 1.21, + "learning_rate": 1.9243492180101947e-05, + "loss": 0.4269, + "step": 2883 + }, + { + "epoch": 1.21, + "learning_rate": 1.9242845168174514e-05, + "loss": 0.3786, + "step": 2884 + }, + { + "epoch": 1.21, + "learning_rate": 1.92421978905697e-05, + "loss": 0.3487, + "step": 2885 + }, + { + "epoch": 1.21, + "learning_rate": 1.924155034730611e-05, + "loss": 0.3736, + "step": 2886 + }, + { + "epoch": 1.21, + "learning_rate": 1.9240902538402354e-05, + "loss": 0.3477, + "step": 2887 + }, + { + "epoch": 1.21, + "learning_rate": 1.9240254463877055e-05, + "loss": 0.3401, + "step": 2888 + }, + { + "epoch": 1.21, + "learning_rate": 1.9239606123748845e-05, + "loss": 0.3752, + "step": 2889 + }, + { + "epoch": 1.21, + "learning_rate": 1.923895751803635e-05, + "loss": 0.37, + "step": 2890 + }, + { + "epoch": 1.21, + "learning_rate": 1.9238308646758224e-05, + "loss": 0.3883, + "step": 2891 + }, + { + "epoch": 1.21, + "learning_rate": 1.9237659509933114e-05, + "loss": 0.3907, + "step": 2892 + }, + { + "epoch": 1.21, + "learning_rate": 1.9237010107579676e-05, + "loss": 0.3955, + "step": 2893 + }, + { + "epoch": 1.21, + "learning_rate": 1.923636043971658e-05, + "loss": 0.3549, + "step": 2894 + }, + { + "epoch": 1.21, + "learning_rate": 1.92357105063625e-05, + "loss": 0.3955, + "step": 2895 + }, + { + "epoch": 1.21, + "learning_rate": 1.9235060307536117e-05, + "loss": 0.3917, + "step": 2896 + }, + { + "epoch": 1.21, + "learning_rate": 1.923440984325612e-05, + "loss": 0.3484, + "step": 2897 + }, + { + "epoch": 1.21, + "learning_rate": 1.923375911354121e-05, + "loss": 0.4145, + "step": 2898 + }, + { + "epoch": 1.21, + "learning_rate": 1.9233108118410082e-05, + "loss": 0.3835, + "step": 2899 + }, + { + "epoch": 1.21, + "learning_rate": 1.923245685788146e-05, + "loss": 0.4056, + "step": 2900 + }, + { + "epoch": 1.21, + "learning_rate": 1.9231805331974053e-05, + "loss": 0.3901, + "step": 2901 + }, + { + "epoch": 1.21, + "learning_rate": 1.9231153540706597e-05, + "loss": 0.4369, + "step": 2902 + }, + { + "epoch": 1.22, + "learning_rate": 1.9230501484097826e-05, + "loss": 0.379, + "step": 2903 + }, + { + "epoch": 1.22, + "learning_rate": 1.9229849162166478e-05, + "loss": 0.3816, + "step": 2904 + }, + { + "epoch": 1.22, + "learning_rate": 1.922919657493131e-05, + "loss": 0.4186, + "step": 2905 + }, + { + "epoch": 1.22, + "learning_rate": 1.922854372241107e-05, + "loss": 0.4063, + "step": 2906 + }, + { + "epoch": 1.22, + "learning_rate": 1.9227890604624534e-05, + "loss": 0.4478, + "step": 2907 + }, + { + "epoch": 1.22, + "learning_rate": 1.9227237221590472e-05, + "loss": 0.4357, + "step": 2908 + }, + { + "epoch": 1.22, + "learning_rate": 1.922658357332766e-05, + "loss": 0.3672, + "step": 2909 + }, + { + "epoch": 1.22, + "learning_rate": 1.9225929659854893e-05, + "loss": 0.3548, + "step": 2910 + }, + { + "epoch": 1.22, + "learning_rate": 1.922527548119097e-05, + "loss": 0.3264, + "step": 2911 + }, + { + "epoch": 1.22, + "learning_rate": 1.922462103735468e-05, + "loss": 0.3426, + "step": 2912 + }, + { + "epoch": 1.22, + "learning_rate": 1.9223966328364853e-05, + "loss": 0.3965, + "step": 2913 + }, + { + "epoch": 1.22, + "learning_rate": 1.922331135424029e-05, + "loss": 0.4179, + "step": 2914 + }, + { + "epoch": 1.22, + "learning_rate": 1.9222656114999835e-05, + "loss": 0.372, + "step": 2915 + }, + { + "epoch": 1.22, + "learning_rate": 1.922200061066231e-05, + "loss": 0.4026, + "step": 2916 + }, + { + "epoch": 1.22, + "learning_rate": 1.9221344841246562e-05, + "loss": 0.356, + "step": 2917 + }, + { + "epoch": 1.22, + "learning_rate": 1.9220688806771443e-05, + "loss": 0.3802, + "step": 2918 + }, + { + "epoch": 1.22, + "learning_rate": 1.92200325072558e-05, + "loss": 0.4339, + "step": 2919 + }, + { + "epoch": 1.22, + "learning_rate": 1.9219375942718508e-05, + "loss": 0.3936, + "step": 2920 + }, + { + "epoch": 1.22, + "learning_rate": 1.9218719113178433e-05, + "loss": 0.3785, + "step": 2921 + }, + { + "epoch": 1.22, + "learning_rate": 1.9218062018654458e-05, + "loss": 0.3854, + "step": 2922 + }, + { + "epoch": 1.22, + "learning_rate": 1.9217404659165476e-05, + "loss": 0.3722, + "step": 2923 + }, + { + "epoch": 1.22, + "learning_rate": 1.921674703473037e-05, + "loss": 0.3839, + "step": 2924 + }, + { + "epoch": 1.22, + "learning_rate": 1.9216089145368052e-05, + "loss": 0.4487, + "step": 2925 + }, + { + "epoch": 1.23, + "learning_rate": 1.921543099109743e-05, + "loss": 0.3922, + "step": 2926 + }, + { + "epoch": 1.23, + "learning_rate": 1.921477257193742e-05, + "loss": 0.393, + "step": 2927 + }, + { + "epoch": 1.23, + "learning_rate": 1.9214113887906947e-05, + "loss": 0.3938, + "step": 2928 + }, + { + "epoch": 1.23, + "learning_rate": 1.9213454939024953e-05, + "loss": 0.3843, + "step": 2929 + }, + { + "epoch": 1.23, + "learning_rate": 1.9212795725310372e-05, + "loss": 0.3286, + "step": 2930 + }, + { + "epoch": 1.23, + "learning_rate": 1.921213624678215e-05, + "loss": 0.3765, + "step": 2931 + }, + { + "epoch": 1.23, + "learning_rate": 1.921147650345925e-05, + "loss": 0.3984, + "step": 2932 + }, + { + "epoch": 1.23, + "learning_rate": 1.9210816495360625e-05, + "loss": 0.4522, + "step": 2933 + }, + { + "epoch": 1.23, + "learning_rate": 1.921015622250526e-05, + "loss": 0.3989, + "step": 2934 + }, + { + "epoch": 1.23, + "learning_rate": 1.920949568491212e-05, + "loss": 0.4184, + "step": 2935 + }, + { + "epoch": 1.23, + "learning_rate": 1.9208834882600207e-05, + "loss": 0.3787, + "step": 2936 + }, + { + "epoch": 1.23, + "learning_rate": 1.9208173815588506e-05, + "loss": 0.4151, + "step": 2937 + }, + { + "epoch": 1.23, + "learning_rate": 1.9207512483896016e-05, + "loss": 0.4006, + "step": 2938 + }, + { + "epoch": 1.23, + "learning_rate": 1.920685088754175e-05, + "loss": 0.3855, + "step": 2939 + }, + { + "epoch": 1.23, + "learning_rate": 1.9206189026544728e-05, + "loss": 0.417, + "step": 2940 + }, + { + "epoch": 1.23, + "learning_rate": 1.920552690092397e-05, + "loss": 0.3611, + "step": 2941 + }, + { + "epoch": 1.23, + "learning_rate": 1.9204864510698513e-05, + "loss": 0.3916, + "step": 2942 + }, + { + "epoch": 1.23, + "learning_rate": 1.9204201855887394e-05, + "loss": 0.3599, + "step": 2943 + }, + { + "epoch": 1.23, + "learning_rate": 1.920353893650966e-05, + "loss": 0.3919, + "step": 2944 + }, + { + "epoch": 1.23, + "learning_rate": 1.9202875752584363e-05, + "loss": 0.3776, + "step": 2945 + }, + { + "epoch": 1.23, + "learning_rate": 1.920221230413057e-05, + "loss": 0.41, + "step": 2946 + }, + { + "epoch": 1.23, + "learning_rate": 1.920154859116735e-05, + "loss": 0.4023, + "step": 2947 + }, + { + "epoch": 1.23, + "learning_rate": 1.9200884613713782e-05, + "loss": 0.3815, + "step": 2948 + }, + { + "epoch": 1.23, + "learning_rate": 1.9200220371788952e-05, + "loss": 0.3629, + "step": 2949 + }, + { + "epoch": 1.24, + "learning_rate": 1.9199555865411948e-05, + "loss": 0.4138, + "step": 2950 + }, + { + "epoch": 1.24, + "learning_rate": 1.9198891094601876e-05, + "loss": 0.3576, + "step": 2951 + }, + { + "epoch": 1.24, + "learning_rate": 1.9198226059377846e-05, + "loss": 0.3736, + "step": 2952 + }, + { + "epoch": 1.24, + "learning_rate": 1.9197560759758963e-05, + "loss": 0.4181, + "step": 2953 + }, + { + "epoch": 1.24, + "learning_rate": 1.9196895195764363e-05, + "loss": 0.3947, + "step": 2954 + }, + { + "epoch": 1.24, + "learning_rate": 1.9196229367413172e-05, + "loss": 0.3668, + "step": 2955 + }, + { + "epoch": 1.24, + "learning_rate": 1.9195563274724523e-05, + "loss": 0.3783, + "step": 2956 + }, + { + "epoch": 1.24, + "learning_rate": 1.9194896917717573e-05, + "loss": 0.3779, + "step": 2957 + }, + { + "epoch": 1.24, + "learning_rate": 1.919423029641147e-05, + "loss": 0.3688, + "step": 2958 + }, + { + "epoch": 1.24, + "learning_rate": 1.9193563410825376e-05, + "loss": 0.3632, + "step": 2959 + }, + { + "epoch": 1.24, + "learning_rate": 1.9192896260978457e-05, + "loss": 0.4211, + "step": 2960 + }, + { + "epoch": 1.24, + "learning_rate": 1.9192228846889896e-05, + "loss": 0.4101, + "step": 2961 + }, + { + "epoch": 1.24, + "learning_rate": 1.9191561168578873e-05, + "loss": 0.3919, + "step": 2962 + }, + { + "epoch": 1.24, + "learning_rate": 1.919089322606458e-05, + "loss": 0.3991, + "step": 2963 + }, + { + "epoch": 1.24, + "learning_rate": 1.919022501936622e-05, + "loss": 0.3567, + "step": 2964 + }, + { + "epoch": 1.24, + "learning_rate": 1.918955654850299e-05, + "loss": 0.3857, + "step": 2965 + }, + { + "epoch": 1.24, + "learning_rate": 1.9188887813494118e-05, + "loss": 0.4223, + "step": 2966 + }, + { + "epoch": 1.24, + "learning_rate": 1.9188218814358817e-05, + "loss": 0.4015, + "step": 2967 + }, + { + "epoch": 1.24, + "learning_rate": 1.9187549551116322e-05, + "loss": 0.4114, + "step": 2968 + }, + { + "epoch": 1.24, + "learning_rate": 1.9186880023785864e-05, + "loss": 0.4167, + "step": 2969 + }, + { + "epoch": 1.24, + "learning_rate": 1.9186210232386694e-05, + "loss": 0.3801, + "step": 2970 + }, + { + "epoch": 1.24, + "learning_rate": 1.9185540176938065e-05, + "loss": 0.3913, + "step": 2971 + }, + { + "epoch": 1.24, + "learning_rate": 1.9184869857459233e-05, + "loss": 0.3867, + "step": 2972 + }, + { + "epoch": 1.24, + "learning_rate": 1.9184199273969465e-05, + "loss": 0.3883, + "step": 2973 + }, + { + "epoch": 1.25, + "learning_rate": 1.918352842648804e-05, + "loss": 0.4711, + "step": 2974 + }, + { + "epoch": 1.25, + "learning_rate": 1.918285731503424e-05, + "loss": 0.4078, + "step": 2975 + }, + { + "epoch": 1.25, + "learning_rate": 1.9182185939627356e-05, + "loss": 0.3993, + "step": 2976 + }, + { + "epoch": 1.25, + "learning_rate": 1.918151430028669e-05, + "loss": 0.3231, + "step": 2977 + }, + { + "epoch": 1.25, + "learning_rate": 1.9180842397031535e-05, + "loss": 0.4097, + "step": 2978 + }, + { + "epoch": 1.25, + "learning_rate": 1.9180170229881216e-05, + "loss": 0.3751, + "step": 2979 + }, + { + "epoch": 1.25, + "learning_rate": 1.9179497798855048e-05, + "loss": 0.3605, + "step": 2980 + }, + { + "epoch": 1.25, + "learning_rate": 1.9178825103972364e-05, + "loss": 0.3395, + "step": 2981 + }, + { + "epoch": 1.25, + "learning_rate": 1.9178152145252497e-05, + "loss": 0.3948, + "step": 2982 + }, + { + "epoch": 1.25, + "learning_rate": 1.917747892271479e-05, + "loss": 0.3671, + "step": 2983 + }, + { + "epoch": 1.25, + "learning_rate": 1.9176805436378597e-05, + "loss": 0.3953, + "step": 2984 + }, + { + "epoch": 1.25, + "learning_rate": 1.9176131686263274e-05, + "loss": 0.4075, + "step": 2985 + }, + { + "epoch": 1.25, + "learning_rate": 1.917545767238819e-05, + "loss": 0.3921, + "step": 2986 + }, + { + "epoch": 1.25, + "learning_rate": 1.9174783394772714e-05, + "loss": 0.4225, + "step": 2987 + }, + { + "epoch": 1.25, + "learning_rate": 1.9174108853436234e-05, + "loss": 0.3768, + "step": 2988 + }, + { + "epoch": 1.25, + "learning_rate": 1.9173434048398133e-05, + "loss": 0.3695, + "step": 2989 + }, + { + "epoch": 1.25, + "learning_rate": 1.917275897967781e-05, + "loss": 0.4141, + "step": 2990 + }, + { + "epoch": 1.25, + "learning_rate": 1.9172083647294676e-05, + "loss": 0.4165, + "step": 2991 + }, + { + "epoch": 1.25, + "learning_rate": 1.9171408051268132e-05, + "loss": 0.4475, + "step": 2992 + }, + { + "epoch": 1.25, + "learning_rate": 1.9170732191617603e-05, + "loss": 0.3717, + "step": 2993 + }, + { + "epoch": 1.25, + "learning_rate": 1.9170056068362516e-05, + "loss": 0.4643, + "step": 2994 + }, + { + "epoch": 1.25, + "learning_rate": 1.9169379681522305e-05, + "loss": 0.3985, + "step": 2995 + }, + { + "epoch": 1.25, + "learning_rate": 1.9168703031116412e-05, + "loss": 0.3602, + "step": 2996 + }, + { + "epoch": 1.25, + "learning_rate": 1.9168026117164285e-05, + "loss": 0.3634, + "step": 2997 + }, + { + "epoch": 1.26, + "learning_rate": 1.916734893968538e-05, + "loss": 0.4244, + "step": 2998 + }, + { + "epoch": 1.26, + "learning_rate": 1.916667149869917e-05, + "loss": 0.3957, + "step": 2999 + }, + { + "epoch": 1.26, + "learning_rate": 1.916599379422512e-05, + "loss": 0.3868, + "step": 3000 + }, + { + "epoch": 1.26, + "learning_rate": 1.9165315826282712e-05, + "loss": 0.3846, + "step": 3001 + }, + { + "epoch": 1.26, + "learning_rate": 1.916463759489143e-05, + "loss": 0.4195, + "step": 3002 + }, + { + "epoch": 1.26, + "learning_rate": 1.916395910007078e-05, + "loss": 0.4048, + "step": 3003 + }, + { + "epoch": 1.26, + "learning_rate": 1.9163280341840253e-05, + "loss": 0.3728, + "step": 3004 + }, + { + "epoch": 1.26, + "learning_rate": 1.9162601320219364e-05, + "loss": 0.419, + "step": 3005 + }, + { + "epoch": 1.26, + "learning_rate": 1.9161922035227633e-05, + "loss": 0.3842, + "step": 3006 + }, + { + "epoch": 1.26, + "learning_rate": 1.9161242486884584e-05, + "loss": 0.4437, + "step": 3007 + }, + { + "epoch": 1.26, + "learning_rate": 1.9160562675209746e-05, + "loss": 0.4448, + "step": 3008 + }, + { + "epoch": 1.26, + "learning_rate": 1.9159882600222664e-05, + "loss": 0.4838, + "step": 3009 + }, + { + "epoch": 1.26, + "learning_rate": 1.9159202261942886e-05, + "loss": 0.3661, + "step": 3010 + }, + { + "epoch": 1.26, + "learning_rate": 1.9158521660389966e-05, + "loss": 0.3588, + "step": 3011 + }, + { + "epoch": 1.26, + "learning_rate": 1.9157840795583466e-05, + "loss": 0.452, + "step": 3012 + }, + { + "epoch": 1.26, + "learning_rate": 1.9157159667542964e-05, + "loss": 0.3669, + "step": 3013 + }, + { + "epoch": 1.26, + "learning_rate": 1.9156478276288027e-05, + "loss": 0.4242, + "step": 3014 + }, + { + "epoch": 1.26, + "learning_rate": 1.9155796621838253e-05, + "loss": 0.3988, + "step": 3015 + }, + { + "epoch": 1.26, + "learning_rate": 1.915511470421323e-05, + "loss": 0.3667, + "step": 3016 + }, + { + "epoch": 1.26, + "learning_rate": 1.9154432523432553e-05, + "loss": 0.4029, + "step": 3017 + }, + { + "epoch": 1.26, + "learning_rate": 1.915375007951584e-05, + "loss": 0.3726, + "step": 3018 + }, + { + "epoch": 1.26, + "learning_rate": 1.9153067372482702e-05, + "loss": 0.4151, + "step": 3019 + }, + { + "epoch": 1.26, + "learning_rate": 1.915238440235277e-05, + "loss": 0.3597, + "step": 3020 + }, + { + "epoch": 1.26, + "learning_rate": 1.9151701169145664e-05, + "loss": 0.3604, + "step": 3021 + }, + { + "epoch": 1.27, + "learning_rate": 1.9151017672881032e-05, + "loss": 0.4356, + "step": 3022 + }, + { + "epoch": 1.27, + "learning_rate": 1.9150333913578518e-05, + "loss": 0.4045, + "step": 3023 + }, + { + "epoch": 1.27, + "learning_rate": 1.9149649891257772e-05, + "loss": 0.379, + "step": 3024 + }, + { + "epoch": 1.27, + "learning_rate": 1.9148965605938463e-05, + "loss": 0.3753, + "step": 3025 + }, + { + "epoch": 1.27, + "learning_rate": 1.9148281057640252e-05, + "loss": 0.4016, + "step": 3026 + }, + { + "epoch": 1.27, + "learning_rate": 1.914759624638282e-05, + "loss": 0.3454, + "step": 3027 + }, + { + "epoch": 1.27, + "learning_rate": 1.914691117218586e-05, + "loss": 0.3758, + "step": 3028 + }, + { + "epoch": 1.27, + "learning_rate": 1.914622583506905e-05, + "loss": 0.449, + "step": 3029 + }, + { + "epoch": 1.27, + "learning_rate": 1.914554023505209e-05, + "loss": 0.3803, + "step": 3030 + }, + { + "epoch": 1.27, + "learning_rate": 1.9144854372154697e-05, + "loss": 0.3536, + "step": 3031 + }, + { + "epoch": 1.27, + "learning_rate": 1.9144168246396576e-05, + "loss": 0.4062, + "step": 3032 + }, + { + "epoch": 1.27, + "learning_rate": 1.914348185779746e-05, + "loss": 0.3956, + "step": 3033 + }, + { + "epoch": 1.27, + "learning_rate": 1.9142795206377066e-05, + "loss": 0.4077, + "step": 3034 + }, + { + "epoch": 1.27, + "learning_rate": 1.914210829215514e-05, + "loss": 0.4146, + "step": 3035 + }, + { + "epoch": 1.27, + "learning_rate": 1.914142111515142e-05, + "loss": 0.4158, + "step": 3036 + }, + { + "epoch": 1.27, + "learning_rate": 1.9140733675385666e-05, + "loss": 0.4014, + "step": 3037 + }, + { + "epoch": 1.27, + "learning_rate": 1.914004597287763e-05, + "loss": 0.3813, + "step": 3038 + }, + { + "epoch": 1.27, + "learning_rate": 1.9139358007647085e-05, + "loss": 0.3681, + "step": 3039 + }, + { + "epoch": 1.27, + "learning_rate": 1.9138669779713807e-05, + "loss": 0.409, + "step": 3040 + }, + { + "epoch": 1.27, + "learning_rate": 1.9137981289097573e-05, + "loss": 0.4381, + "step": 3041 + }, + { + "epoch": 1.27, + "learning_rate": 1.9137292535818174e-05, + "loss": 0.3638, + "step": 3042 + }, + { + "epoch": 1.27, + "learning_rate": 1.913660351989541e-05, + "loss": 0.4291, + "step": 3043 + }, + { + "epoch": 1.27, + "learning_rate": 1.9135914241349088e-05, + "loss": 0.4002, + "step": 3044 + }, + { + "epoch": 1.27, + "learning_rate": 1.9135224700199015e-05, + "loss": 0.3697, + "step": 3045 + }, + { + "epoch": 1.28, + "learning_rate": 1.913453489646502e-05, + "loss": 0.4188, + "step": 3046 + }, + { + "epoch": 1.28, + "learning_rate": 1.913384483016692e-05, + "loss": 0.4102, + "step": 3047 + }, + { + "epoch": 1.28, + "learning_rate": 1.9133154501324557e-05, + "loss": 0.3559, + "step": 3048 + }, + { + "epoch": 1.28, + "learning_rate": 1.9132463909957774e-05, + "loss": 0.4174, + "step": 3049 + }, + { + "epoch": 1.28, + "learning_rate": 1.9131773056086417e-05, + "loss": 0.4116, + "step": 3050 + }, + { + "epoch": 1.28, + "learning_rate": 1.9131081939730348e-05, + "loss": 0.4361, + "step": 3051 + }, + { + "epoch": 1.28, + "learning_rate": 1.913039056090943e-05, + "loss": 0.3952, + "step": 3052 + }, + { + "epoch": 1.28, + "learning_rate": 1.912969891964354e-05, + "loss": 0.3878, + "step": 3053 + }, + { + "epoch": 1.28, + "learning_rate": 1.9129007015952556e-05, + "loss": 0.36, + "step": 3054 + }, + { + "epoch": 1.28, + "learning_rate": 1.9128314849856365e-05, + "loss": 0.3476, + "step": 3055 + }, + { + "epoch": 1.28, + "learning_rate": 1.9127622421374866e-05, + "loss": 0.3948, + "step": 3056 + }, + { + "epoch": 1.28, + "learning_rate": 1.9126929730527957e-05, + "loss": 0.3804, + "step": 3057 + }, + { + "epoch": 1.28, + "learning_rate": 1.9126236777335554e-05, + "loss": 0.4198, + "step": 3058 + }, + { + "epoch": 1.28, + "learning_rate": 1.9125543561817573e-05, + "loss": 0.3692, + "step": 3059 + }, + { + "epoch": 1.28, + "learning_rate": 1.9124850083993942e-05, + "loss": 0.3862, + "step": 3060 + }, + { + "epoch": 1.28, + "learning_rate": 1.912415634388459e-05, + "loss": 0.4109, + "step": 3061 + }, + { + "epoch": 1.28, + "learning_rate": 1.912346234150946e-05, + "loss": 0.3838, + "step": 3062 + }, + { + "epoch": 1.28, + "learning_rate": 1.9122768076888506e-05, + "loss": 0.3874, + "step": 3063 + }, + { + "epoch": 1.28, + "learning_rate": 1.9122073550041675e-05, + "loss": 0.4108, + "step": 3064 + }, + { + "epoch": 1.28, + "learning_rate": 1.9121378760988934e-05, + "loss": 0.3832, + "step": 3065 + }, + { + "epoch": 1.28, + "learning_rate": 1.9120683709750257e-05, + "loss": 0.3902, + "step": 3066 + }, + { + "epoch": 1.28, + "learning_rate": 1.9119988396345615e-05, + "loss": 0.3786, + "step": 3067 + }, + { + "epoch": 1.28, + "learning_rate": 1.9119292820795007e-05, + "loss": 0.4007, + "step": 3068 + }, + { + "epoch": 1.28, + "learning_rate": 1.9118596983118415e-05, + "loss": 0.3548, + "step": 3069 + }, + { + "epoch": 1.29, + "learning_rate": 1.9117900883335846e-05, + "loss": 0.3979, + "step": 3070 + }, + { + "epoch": 1.29, + "learning_rate": 1.9117204521467308e-05, + "loss": 0.3693, + "step": 3071 + }, + { + "epoch": 1.29, + "learning_rate": 1.9116507897532813e-05, + "loss": 0.3875, + "step": 3072 + }, + { + "epoch": 1.29, + "learning_rate": 1.911581101155239e-05, + "loss": 0.3935, + "step": 3073 + }, + { + "epoch": 1.29, + "learning_rate": 1.911511386354607e-05, + "loss": 0.3525, + "step": 3074 + }, + { + "epoch": 1.29, + "learning_rate": 1.911441645353389e-05, + "loss": 0.3911, + "step": 3075 + }, + { + "epoch": 1.29, + "learning_rate": 1.9113718781535896e-05, + "loss": 0.3531, + "step": 3076 + }, + { + "epoch": 1.29, + "learning_rate": 1.9113020847572144e-05, + "loss": 0.4023, + "step": 3077 + }, + { + "epoch": 1.29, + "learning_rate": 1.911232265166269e-05, + "loss": 0.3949, + "step": 3078 + }, + { + "epoch": 1.29, + "learning_rate": 1.9111624193827614e-05, + "loss": 0.3762, + "step": 3079 + }, + { + "epoch": 1.29, + "learning_rate": 1.9110925474086982e-05, + "loss": 0.3759, + "step": 3080 + }, + { + "epoch": 1.29, + "learning_rate": 1.9110226492460886e-05, + "loss": 0.4094, + "step": 3081 + }, + { + "epoch": 1.29, + "learning_rate": 1.9109527248969412e-05, + "loss": 0.4151, + "step": 3082 + }, + { + "epoch": 1.29, + "learning_rate": 1.910882774363266e-05, + "loss": 0.3605, + "step": 3083 + }, + { + "epoch": 1.29, + "learning_rate": 1.9108127976470737e-05, + "loss": 0.3758, + "step": 3084 + }, + { + "epoch": 1.29, + "learning_rate": 1.9107427947503757e-05, + "loss": 0.3833, + "step": 3085 + }, + { + "epoch": 1.29, + "learning_rate": 1.9106727656751846e-05, + "loss": 0.3986, + "step": 3086 + }, + { + "epoch": 1.29, + "learning_rate": 1.9106027104235126e-05, + "loss": 0.3984, + "step": 3087 + }, + { + "epoch": 1.29, + "learning_rate": 1.910532628997374e-05, + "loss": 0.4009, + "step": 3088 + }, + { + "epoch": 1.29, + "learning_rate": 1.910462521398783e-05, + "loss": 0.3672, + "step": 3089 + }, + { + "epoch": 1.29, + "learning_rate": 1.9103923876297544e-05, + "loss": 0.3765, + "step": 3090 + }, + { + "epoch": 1.29, + "learning_rate": 1.9103222276923047e-05, + "loss": 0.3638, + "step": 3091 + }, + { + "epoch": 1.29, + "learning_rate": 1.9102520415884504e-05, + "loss": 0.3996, + "step": 3092 + }, + { + "epoch": 1.29, + "learning_rate": 1.910181829320209e-05, + "loss": 0.3812, + "step": 3093 + }, + { + "epoch": 1.3, + "learning_rate": 1.9101115908895985e-05, + "loss": 0.3676, + "step": 3094 + }, + { + "epoch": 1.3, + "learning_rate": 1.9100413262986383e-05, + "loss": 0.3927, + "step": 3095 + }, + { + "epoch": 1.3, + "learning_rate": 1.909971035549347e-05, + "loss": 0.4282, + "step": 3096 + }, + { + "epoch": 1.3, + "learning_rate": 1.9099007186437465e-05, + "loss": 0.4489, + "step": 3097 + }, + { + "epoch": 1.3, + "learning_rate": 1.9098303755838566e-05, + "loss": 0.3486, + "step": 3098 + }, + { + "epoch": 1.3, + "learning_rate": 1.9097600063717005e-05, + "loss": 0.3687, + "step": 3099 + }, + { + "epoch": 1.3, + "learning_rate": 1.9096896110092998e-05, + "loss": 0.4529, + "step": 3100 + }, + { + "epoch": 1.3, + "learning_rate": 1.909619189498679e-05, + "loss": 0.3496, + "step": 3101 + }, + { + "epoch": 1.3, + "learning_rate": 1.9095487418418614e-05, + "loss": 0.4179, + "step": 3102 + }, + { + "epoch": 1.3, + "learning_rate": 1.9094782680408722e-05, + "loss": 0.4081, + "step": 3103 + }, + { + "epoch": 1.3, + "learning_rate": 1.9094077680977377e-05, + "loss": 0.3881, + "step": 3104 + }, + { + "epoch": 1.3, + "learning_rate": 1.9093372420144835e-05, + "loss": 0.4002, + "step": 3105 + }, + { + "epoch": 1.3, + "learning_rate": 1.9092666897931375e-05, + "loss": 0.3316, + "step": 3106 + }, + { + "epoch": 1.3, + "learning_rate": 1.9091961114357274e-05, + "loss": 0.4296, + "step": 3107 + }, + { + "epoch": 1.3, + "learning_rate": 1.909125506944282e-05, + "loss": 0.3749, + "step": 3108 + }, + { + "epoch": 1.3, + "learning_rate": 1.90905487632083e-05, + "loss": 0.427, + "step": 3109 + }, + { + "epoch": 1.3, + "learning_rate": 1.908984219567403e-05, + "loss": 0.386, + "step": 3110 + }, + { + "epoch": 1.3, + "learning_rate": 1.908913536686031e-05, + "loss": 0.4013, + "step": 3111 + }, + { + "epoch": 1.3, + "learning_rate": 1.9088428276787455e-05, + "loss": 0.3712, + "step": 3112 + }, + { + "epoch": 1.3, + "learning_rate": 1.90877209254758e-05, + "loss": 0.4584, + "step": 3113 + }, + { + "epoch": 1.3, + "learning_rate": 1.9087013312945668e-05, + "loss": 0.41, + "step": 3114 + }, + { + "epoch": 1.3, + "learning_rate": 1.9086305439217402e-05, + "loss": 0.3809, + "step": 3115 + }, + { + "epoch": 1.3, + "learning_rate": 1.908559730431135e-05, + "loss": 0.3908, + "step": 3116 + }, + { + "epoch": 1.31, + "learning_rate": 1.9084888908247864e-05, + "loss": 0.4066, + "step": 3117 + }, + { + "epoch": 1.31, + "learning_rate": 1.908418025104731e-05, + "loss": 0.4241, + "step": 3118 + }, + { + "epoch": 1.31, + "learning_rate": 1.9083471332730058e-05, + "loss": 0.3628, + "step": 3119 + }, + { + "epoch": 1.31, + "learning_rate": 1.908276215331648e-05, + "loss": 0.469, + "step": 3120 + }, + { + "epoch": 1.31, + "learning_rate": 1.9082052712826967e-05, + "loss": 0.3736, + "step": 3121 + }, + { + "epoch": 1.31, + "learning_rate": 1.9081343011281907e-05, + "loss": 0.3628, + "step": 3122 + }, + { + "epoch": 1.31, + "learning_rate": 1.9080633048701696e-05, + "loss": 0.436, + "step": 3123 + }, + { + "epoch": 1.31, + "learning_rate": 1.907992282510675e-05, + "loss": 0.4212, + "step": 3124 + }, + { + "epoch": 1.31, + "learning_rate": 1.907921234051748e-05, + "loss": 0.3986, + "step": 3125 + }, + { + "epoch": 1.31, + "learning_rate": 1.9078501594954305e-05, + "loss": 0.4126, + "step": 3126 + }, + { + "epoch": 1.31, + "learning_rate": 1.9077790588437663e-05, + "loss": 0.3649, + "step": 3127 + }, + { + "epoch": 1.31, + "learning_rate": 1.907707932098798e-05, + "loss": 0.3977, + "step": 3128 + }, + { + "epoch": 1.31, + "learning_rate": 1.9076367792625713e-05, + "loss": 0.368, + "step": 3129 + }, + { + "epoch": 1.31, + "learning_rate": 1.9075656003371304e-05, + "loss": 0.3954, + "step": 3130 + }, + { + "epoch": 1.31, + "learning_rate": 1.907494395324522e-05, + "loss": 0.3712, + "step": 3131 + }, + { + "epoch": 1.31, + "learning_rate": 1.907423164226792e-05, + "loss": 0.3879, + "step": 3132 + }, + { + "epoch": 1.31, + "learning_rate": 1.907351907045989e-05, + "loss": 0.375, + "step": 3133 + }, + { + "epoch": 1.31, + "learning_rate": 1.90728062378416e-05, + "loss": 0.4481, + "step": 3134 + }, + { + "epoch": 1.31, + "learning_rate": 1.907209314443355e-05, + "loss": 0.4373, + "step": 3135 + }, + { + "epoch": 1.31, + "learning_rate": 1.907137979025623e-05, + "loss": 0.4099, + "step": 3136 + }, + { + "epoch": 1.31, + "learning_rate": 1.907066617533015e-05, + "loss": 0.39, + "step": 3137 + }, + { + "epoch": 1.31, + "learning_rate": 1.906995229967582e-05, + "loss": 0.3978, + "step": 3138 + }, + { + "epoch": 1.31, + "learning_rate": 1.9069238163313757e-05, + "loss": 0.4477, + "step": 3139 + }, + { + "epoch": 1.31, + "learning_rate": 1.906852376626449e-05, + "loss": 0.3745, + "step": 3140 + }, + { + "epoch": 1.32, + "learning_rate": 1.906780910854856e-05, + "loss": 0.3677, + "step": 3141 + }, + { + "epoch": 1.32, + "learning_rate": 1.90670941901865e-05, + "loss": 0.359, + "step": 3142 + }, + { + "epoch": 1.32, + "learning_rate": 1.9066379011198863e-05, + "loss": 0.3517, + "step": 3143 + }, + { + "epoch": 1.32, + "learning_rate": 1.906566357160621e-05, + "loss": 0.3351, + "step": 3144 + }, + { + "epoch": 1.32, + "learning_rate": 1.9064947871429102e-05, + "loss": 0.3869, + "step": 3145 + }, + { + "epoch": 1.32, + "learning_rate": 1.9064231910688108e-05, + "loss": 0.4992, + "step": 3146 + }, + { + "epoch": 1.32, + "learning_rate": 1.9063515689403815e-05, + "loss": 0.3779, + "step": 3147 + }, + { + "epoch": 1.32, + "learning_rate": 1.90627992075968e-05, + "loss": 0.3634, + "step": 3148 + }, + { + "epoch": 1.32, + "learning_rate": 1.9062082465287672e-05, + "loss": 0.38, + "step": 3149 + }, + { + "epoch": 1.32, + "learning_rate": 1.9061365462497022e-05, + "loss": 0.3647, + "step": 3150 + }, + { + "epoch": 1.32, + "learning_rate": 1.9060648199245463e-05, + "loss": 0.4137, + "step": 3151 + }, + { + "epoch": 1.32, + "learning_rate": 1.905993067555361e-05, + "loss": 0.392, + "step": 3152 + }, + { + "epoch": 1.32, + "learning_rate": 1.905921289144209e-05, + "loss": 0.3798, + "step": 3153 + }, + { + "epoch": 1.32, + "learning_rate": 1.9058494846931538e-05, + "loss": 0.3981, + "step": 3154 + }, + { + "epoch": 1.32, + "learning_rate": 1.9057776542042587e-05, + "loss": 0.4565, + "step": 3155 + }, + { + "epoch": 1.32, + "learning_rate": 1.9057057976795884e-05, + "loss": 0.3855, + "step": 3156 + }, + { + "epoch": 1.32, + "learning_rate": 1.905633915121209e-05, + "loss": 0.3821, + "step": 3157 + }, + { + "epoch": 1.32, + "learning_rate": 1.905562006531186e-05, + "loss": 0.3874, + "step": 3158 + }, + { + "epoch": 1.32, + "learning_rate": 1.9054900719115874e-05, + "loss": 0.3713, + "step": 3159 + }, + { + "epoch": 1.32, + "learning_rate": 1.9054181112644797e-05, + "loss": 0.3942, + "step": 3160 + }, + { + "epoch": 1.32, + "learning_rate": 1.905346124591932e-05, + "loss": 0.4181, + "step": 3161 + }, + { + "epoch": 1.32, + "learning_rate": 1.9052741118960133e-05, + "loss": 0.405, + "step": 3162 + }, + { + "epoch": 1.32, + "learning_rate": 1.9052020731787936e-05, + "loss": 0.3995, + "step": 3163 + }, + { + "epoch": 1.32, + "learning_rate": 1.9051300084423433e-05, + "loss": 0.4052, + "step": 3164 + }, + { + "epoch": 1.33, + "learning_rate": 1.9050579176887343e-05, + "loss": 0.4243, + "step": 3165 + }, + { + "epoch": 1.33, + "learning_rate": 1.9049858009200388e-05, + "loss": 0.373, + "step": 3166 + }, + { + "epoch": 1.33, + "learning_rate": 1.9049136581383295e-05, + "loss": 0.4055, + "step": 3167 + }, + { + "epoch": 1.33, + "learning_rate": 1.90484148934568e-05, + "loss": 0.3674, + "step": 3168 + }, + { + "epoch": 1.33, + "learning_rate": 1.9047692945441647e-05, + "loss": 0.4224, + "step": 3169 + }, + { + "epoch": 1.33, + "learning_rate": 1.904697073735859e-05, + "loss": 0.4768, + "step": 3170 + }, + { + "epoch": 1.33, + "learning_rate": 1.9046248269228388e-05, + "loss": 0.3627, + "step": 3171 + }, + { + "epoch": 1.33, + "learning_rate": 1.9045525541071807e-05, + "loss": 0.3667, + "step": 3172 + }, + { + "epoch": 1.33, + "learning_rate": 1.904480255290962e-05, + "loss": 0.3859, + "step": 3173 + }, + { + "epoch": 1.33, + "learning_rate": 1.904407930476261e-05, + "loss": 0.3877, + "step": 3174 + }, + { + "epoch": 1.33, + "learning_rate": 1.9043355796651565e-05, + "loss": 0.4244, + "step": 3175 + }, + { + "epoch": 1.33, + "learning_rate": 1.9042632028597286e-05, + "loss": 0.4116, + "step": 3176 + }, + { + "epoch": 1.33, + "learning_rate": 1.904190800062057e-05, + "loss": 0.4461, + "step": 3177 + }, + { + "epoch": 1.33, + "learning_rate": 1.9041183712742233e-05, + "loss": 0.3593, + "step": 3178 + }, + { + "epoch": 1.33, + "learning_rate": 1.9040459164983095e-05, + "loss": 0.3896, + "step": 3179 + }, + { + "epoch": 1.33, + "learning_rate": 1.903973435736398e-05, + "loss": 0.4172, + "step": 3180 + }, + { + "epoch": 1.33, + "learning_rate": 1.903900928990572e-05, + "loss": 0.3866, + "step": 3181 + }, + { + "epoch": 1.33, + "learning_rate": 1.9038283962629164e-05, + "loss": 0.3818, + "step": 3182 + }, + { + "epoch": 1.33, + "learning_rate": 1.903755837555515e-05, + "loss": 0.3322, + "step": 3183 + }, + { + "epoch": 1.33, + "learning_rate": 1.9036832528704547e-05, + "loss": 0.4535, + "step": 3184 + }, + { + "epoch": 1.33, + "learning_rate": 1.9036106422098207e-05, + "loss": 0.4, + "step": 3185 + }, + { + "epoch": 1.33, + "learning_rate": 1.9035380055757007e-05, + "loss": 0.3732, + "step": 3186 + }, + { + "epoch": 1.33, + "learning_rate": 1.9034653429701824e-05, + "loss": 0.3552, + "step": 3187 + }, + { + "epoch": 1.33, + "learning_rate": 1.9033926543953546e-05, + "loss": 0.4101, + "step": 3188 + }, + { + "epoch": 1.34, + "learning_rate": 1.903319939853307e-05, + "loss": 0.409, + "step": 3189 + }, + { + "epoch": 1.34, + "learning_rate": 1.903247199346129e-05, + "loss": 0.3997, + "step": 3190 + }, + { + "epoch": 1.34, + "learning_rate": 1.9031744328759118e-05, + "loss": 0.4128, + "step": 3191 + }, + { + "epoch": 1.34, + "learning_rate": 1.903101640444747e-05, + "loss": 0.4079, + "step": 3192 + }, + { + "epoch": 1.34, + "learning_rate": 1.903028822054727e-05, + "loss": 0.3743, + "step": 3193 + }, + { + "epoch": 1.34, + "learning_rate": 1.902955977707945e-05, + "loss": 0.353, + "step": 3194 + }, + { + "epoch": 1.34, + "learning_rate": 1.9028831074064945e-05, + "loss": 0.3681, + "step": 3195 + }, + { + "epoch": 1.34, + "learning_rate": 1.9028102111524702e-05, + "loss": 0.4156, + "step": 3196 + }, + { + "epoch": 1.34, + "learning_rate": 1.9027372889479676e-05, + "loss": 0.4007, + "step": 3197 + }, + { + "epoch": 1.34, + "learning_rate": 1.9026643407950826e-05, + "loss": 0.3695, + "step": 3198 + }, + { + "epoch": 1.34, + "learning_rate": 1.9025913666959122e-05, + "loss": 0.3651, + "step": 3199 + }, + { + "epoch": 1.34, + "learning_rate": 1.9025183666525543e-05, + "loss": 0.3975, + "step": 3200 + }, + { + "epoch": 1.34, + "learning_rate": 1.9024453406671064e-05, + "loss": 0.4004, + "step": 3201 + }, + { + "epoch": 1.34, + "learning_rate": 1.902372288741668e-05, + "loss": 0.3727, + "step": 3202 + }, + { + "epoch": 1.34, + "learning_rate": 1.902299210878339e-05, + "loss": 0.4255, + "step": 3203 + }, + { + "epoch": 1.34, + "learning_rate": 1.9022261070792197e-05, + "loss": 0.3589, + "step": 3204 + }, + { + "epoch": 1.34, + "learning_rate": 1.902152977346412e-05, + "loss": 0.3909, + "step": 3205 + }, + { + "epoch": 1.34, + "learning_rate": 1.9020798216820174e-05, + "loss": 0.3787, + "step": 3206 + }, + { + "epoch": 1.34, + "learning_rate": 1.9020066400881386e-05, + "loss": 0.3855, + "step": 3207 + }, + { + "epoch": 1.34, + "learning_rate": 1.9019334325668795e-05, + "loss": 0.3937, + "step": 3208 + }, + { + "epoch": 1.34, + "learning_rate": 1.901860199120344e-05, + "loss": 0.448, + "step": 3209 + }, + { + "epoch": 1.34, + "learning_rate": 1.9017869397506374e-05, + "loss": 0.4216, + "step": 3210 + }, + { + "epoch": 1.34, + "learning_rate": 1.9017136544598657e-05, + "loss": 0.3929, + "step": 3211 + }, + { + "epoch": 1.34, + "learning_rate": 1.901640343250135e-05, + "loss": 0.4304, + "step": 3212 + }, + { + "epoch": 1.35, + "learning_rate": 1.9015670061235528e-05, + "loss": 0.3553, + "step": 3213 + }, + { + "epoch": 1.35, + "learning_rate": 1.901493643082227e-05, + "loss": 0.3978, + "step": 3214 + }, + { + "epoch": 1.35, + "learning_rate": 1.9014202541282663e-05, + "loss": 0.425, + "step": 3215 + }, + { + "epoch": 1.35, + "learning_rate": 1.901346839263781e-05, + "loss": 0.4167, + "step": 3216 + }, + { + "epoch": 1.35, + "learning_rate": 1.9012733984908798e-05, + "loss": 0.3343, + "step": 3217 + }, + { + "epoch": 1.35, + "learning_rate": 1.901199931811675e-05, + "loss": 0.3673, + "step": 3218 + }, + { + "epoch": 1.35, + "learning_rate": 1.9011264392282774e-05, + "loss": 0.4347, + "step": 3219 + }, + { + "epoch": 1.35, + "learning_rate": 1.9010529207428004e-05, + "loss": 0.337, + "step": 3220 + }, + { + "epoch": 1.35, + "learning_rate": 1.9009793763573564e-05, + "loss": 0.4128, + "step": 3221 + }, + { + "epoch": 1.35, + "learning_rate": 1.9009058060740603e-05, + "loss": 0.3688, + "step": 3222 + }, + { + "epoch": 1.35, + "learning_rate": 1.900832209895026e-05, + "loss": 0.4163, + "step": 3223 + }, + { + "epoch": 1.35, + "learning_rate": 1.900758587822369e-05, + "loss": 0.417, + "step": 3224 + }, + { + "epoch": 1.35, + "learning_rate": 1.900684939858206e-05, + "loss": 0.3658, + "step": 3225 + }, + { + "epoch": 1.35, + "learning_rate": 1.9006112660046535e-05, + "loss": 0.3795, + "step": 3226 + }, + { + "epoch": 1.35, + "learning_rate": 1.9005375662638294e-05, + "loss": 0.3625, + "step": 3227 + }, + { + "epoch": 1.35, + "learning_rate": 1.900463840637852e-05, + "loss": 0.3675, + "step": 3228 + }, + { + "epoch": 1.35, + "learning_rate": 1.9003900891288406e-05, + "loss": 0.3909, + "step": 3229 + }, + { + "epoch": 1.35, + "learning_rate": 1.900316311738915e-05, + "loss": 0.4061, + "step": 3230 + }, + { + "epoch": 1.35, + "learning_rate": 1.9002425084701958e-05, + "loss": 0.3623, + "step": 3231 + }, + { + "epoch": 1.35, + "learning_rate": 1.900168679324805e-05, + "loss": 0.4009, + "step": 3232 + }, + { + "epoch": 1.35, + "learning_rate": 1.900094824304864e-05, + "loss": 0.373, + "step": 3233 + }, + { + "epoch": 1.35, + "learning_rate": 1.900020943412496e-05, + "loss": 0.3667, + "step": 3234 + }, + { + "epoch": 1.35, + "learning_rate": 1.899947036649825e-05, + "loss": 0.3848, + "step": 3235 + }, + { + "epoch": 1.35, + "learning_rate": 1.8998731040189748e-05, + "loss": 0.4338, + "step": 3236 + }, + { + "epoch": 1.36, + "learning_rate": 1.8997991455220702e-05, + "loss": 0.4409, + "step": 3237 + }, + { + "epoch": 1.36, + "learning_rate": 1.8997251611612385e-05, + "loss": 0.4362, + "step": 3238 + }, + { + "epoch": 1.36, + "learning_rate": 1.899651150938605e-05, + "loss": 0.3413, + "step": 3239 + }, + { + "epoch": 1.36, + "learning_rate": 1.8995771148562978e-05, + "loss": 0.4013, + "step": 3240 + }, + { + "epoch": 1.36, + "learning_rate": 1.8995030529164446e-05, + "loss": 0.3302, + "step": 3241 + }, + { + "epoch": 1.36, + "learning_rate": 1.8994289651211742e-05, + "loss": 0.3261, + "step": 3242 + }, + { + "epoch": 1.36, + "learning_rate": 1.8993548514726162e-05, + "loss": 0.4185, + "step": 3243 + }, + { + "epoch": 1.36, + "learning_rate": 1.8992807119729013e-05, + "loss": 0.4119, + "step": 3244 + }, + { + "epoch": 1.36, + "learning_rate": 1.8992065466241603e-05, + "loss": 0.3402, + "step": 3245 + }, + { + "epoch": 1.36, + "learning_rate": 1.899132355428525e-05, + "loss": 0.3759, + "step": 3246 + }, + { + "epoch": 1.36, + "learning_rate": 1.899058138388128e-05, + "loss": 0.4098, + "step": 3247 + }, + { + "epoch": 1.36, + "learning_rate": 1.8989838955051027e-05, + "loss": 0.4127, + "step": 3248 + }, + { + "epoch": 1.36, + "learning_rate": 1.898909626781583e-05, + "loss": 0.3761, + "step": 3249 + }, + { + "epoch": 1.36, + "learning_rate": 1.898835332219704e-05, + "loss": 0.3541, + "step": 3250 + }, + { + "epoch": 1.36, + "learning_rate": 1.8987610118216006e-05, + "loss": 0.4133, + "step": 3251 + }, + { + "epoch": 1.36, + "learning_rate": 1.8986866655894098e-05, + "loss": 0.4228, + "step": 3252 + }, + { + "epoch": 1.36, + "learning_rate": 1.8986122935252684e-05, + "loss": 0.4267, + "step": 3253 + }, + { + "epoch": 1.36, + "learning_rate": 1.8985378956313133e-05, + "loss": 0.3861, + "step": 3254 + }, + { + "epoch": 1.36, + "learning_rate": 1.8984634719096844e-05, + "loss": 0.3509, + "step": 3255 + }, + { + "epoch": 1.36, + "learning_rate": 1.8983890223625205e-05, + "loss": 0.3746, + "step": 3256 + }, + { + "epoch": 1.36, + "learning_rate": 1.8983145469919606e-05, + "loss": 0.4175, + "step": 3257 + }, + { + "epoch": 1.36, + "learning_rate": 1.8982400458001465e-05, + "loss": 0.4045, + "step": 3258 + }, + { + "epoch": 1.36, + "learning_rate": 1.8981655187892194e-05, + "loss": 0.3661, + "step": 3259 + }, + { + "epoch": 1.36, + "learning_rate": 1.8980909659613217e-05, + "loss": 0.4329, + "step": 3260 + }, + { + "epoch": 1.37, + "learning_rate": 1.898016387318596e-05, + "loss": 0.3702, + "step": 3261 + }, + { + "epoch": 1.37, + "learning_rate": 1.8979417828631858e-05, + "loss": 0.4432, + "step": 3262 + }, + { + "epoch": 1.37, + "learning_rate": 1.8978671525972365e-05, + "loss": 0.3797, + "step": 3263 + }, + { + "epoch": 1.37, + "learning_rate": 1.8977924965228923e-05, + "loss": 0.3628, + "step": 3264 + }, + { + "epoch": 1.37, + "learning_rate": 1.8977178146422992e-05, + "loss": 0.3518, + "step": 3265 + }, + { + "epoch": 1.37, + "learning_rate": 1.8976431069576047e-05, + "loss": 0.3573, + "step": 3266 + }, + { + "epoch": 1.37, + "learning_rate": 1.8975683734709554e-05, + "loss": 0.3757, + "step": 3267 + }, + { + "epoch": 1.37, + "learning_rate": 1.8974936141844997e-05, + "loss": 0.37, + "step": 3268 + }, + { + "epoch": 1.37, + "learning_rate": 1.8974188291003864e-05, + "loss": 0.3833, + "step": 3269 + }, + { + "epoch": 1.37, + "learning_rate": 1.897344018220765e-05, + "loss": 0.3984, + "step": 3270 + }, + { + "epoch": 1.37, + "learning_rate": 1.8972691815477864e-05, + "loss": 0.4031, + "step": 3271 + }, + { + "epoch": 1.37, + "learning_rate": 1.8971943190836013e-05, + "loss": 0.3859, + "step": 3272 + }, + { + "epoch": 1.37, + "learning_rate": 1.8971194308303617e-05, + "loss": 0.3791, + "step": 3273 + }, + { + "epoch": 1.37, + "learning_rate": 1.89704451679022e-05, + "loss": 0.4298, + "step": 3274 + }, + { + "epoch": 1.37, + "learning_rate": 1.89696957696533e-05, + "loss": 0.3776, + "step": 3275 + }, + { + "epoch": 1.37, + "learning_rate": 1.8968946113578455e-05, + "loss": 0.3676, + "step": 3276 + }, + { + "epoch": 1.37, + "learning_rate": 1.896819619969921e-05, + "loss": 0.393, + "step": 3277 + }, + { + "epoch": 1.37, + "learning_rate": 1.8967446028037126e-05, + "loss": 0.372, + "step": 3278 + }, + { + "epoch": 1.37, + "learning_rate": 1.8966695598613758e-05, + "loss": 0.3877, + "step": 3279 + }, + { + "epoch": 1.37, + "learning_rate": 1.8965944911450685e-05, + "loss": 0.3731, + "step": 3280 + }, + { + "epoch": 1.37, + "learning_rate": 1.8965193966569483e-05, + "loss": 0.417, + "step": 3281 + }, + { + "epoch": 1.37, + "learning_rate": 1.8964442763991737e-05, + "loss": 0.4348, + "step": 3282 + }, + { + "epoch": 1.37, + "learning_rate": 1.8963691303739037e-05, + "loss": 0.4124, + "step": 3283 + }, + { + "epoch": 1.37, + "learning_rate": 1.8962939585832985e-05, + "loss": 0.3836, + "step": 3284 + }, + { + "epoch": 1.38, + "learning_rate": 1.8962187610295188e-05, + "loss": 0.4002, + "step": 3285 + }, + { + "epoch": 1.38, + "learning_rate": 1.896143537714726e-05, + "loss": 0.3544, + "step": 3286 + }, + { + "epoch": 1.38, + "learning_rate": 1.8960682886410825e-05, + "loss": 0.3982, + "step": 3287 + }, + { + "epoch": 1.38, + "learning_rate": 1.8959930138107512e-05, + "loss": 0.3865, + "step": 3288 + }, + { + "epoch": 1.38, + "learning_rate": 1.895917713225896e-05, + "loss": 0.3951, + "step": 3289 + }, + { + "epoch": 1.38, + "learning_rate": 1.895842386888681e-05, + "loss": 0.3978, + "step": 3290 + }, + { + "epoch": 1.38, + "learning_rate": 1.8957670348012717e-05, + "loss": 0.3807, + "step": 3291 + }, + { + "epoch": 1.38, + "learning_rate": 1.8956916569658336e-05, + "loss": 0.396, + "step": 3292 + }, + { + "epoch": 1.38, + "learning_rate": 1.895616253384534e-05, + "loss": 0.3715, + "step": 3293 + }, + { + "epoch": 1.38, + "learning_rate": 1.8955408240595396e-05, + "loss": 0.4017, + "step": 3294 + }, + { + "epoch": 1.38, + "learning_rate": 1.895465368993019e-05, + "loss": 0.4133, + "step": 3295 + }, + { + "epoch": 1.38, + "learning_rate": 1.8953898881871412e-05, + "loss": 0.3663, + "step": 3296 + }, + { + "epoch": 1.38, + "learning_rate": 1.895314381644075e-05, + "loss": 0.4083, + "step": 3297 + }, + { + "epoch": 1.38, + "learning_rate": 1.8952388493659918e-05, + "loss": 0.3797, + "step": 3298 + }, + { + "epoch": 1.38, + "learning_rate": 1.8951632913550625e-05, + "loss": 0.3964, + "step": 3299 + }, + { + "epoch": 1.38, + "learning_rate": 1.8950877076134583e-05, + "loss": 0.4279, + "step": 3300 + }, + { + "epoch": 1.38, + "learning_rate": 1.8950120981433527e-05, + "loss": 0.3264, + "step": 3301 + }, + { + "epoch": 1.38, + "learning_rate": 1.8949364629469184e-05, + "loss": 0.3627, + "step": 3302 + }, + { + "epoch": 1.38, + "learning_rate": 1.8948608020263293e-05, + "loss": 0.3707, + "step": 3303 + }, + { + "epoch": 1.38, + "learning_rate": 1.894785115383761e-05, + "loss": 0.3831, + "step": 3304 + }, + { + "epoch": 1.38, + "learning_rate": 1.8947094030213883e-05, + "loss": 0.3846, + "step": 3305 + }, + { + "epoch": 1.38, + "learning_rate": 1.8946336649413883e-05, + "loss": 0.3935, + "step": 3306 + }, + { + "epoch": 1.38, + "learning_rate": 1.8945579011459366e-05, + "loss": 0.4725, + "step": 3307 + }, + { + "epoch": 1.38, + "learning_rate": 1.8944821116372125e-05, + "loss": 0.401, + "step": 3308 + }, + { + "epoch": 1.39, + "learning_rate": 1.8944062964173934e-05, + "loss": 0.446, + "step": 3309 + }, + { + "epoch": 1.39, + "learning_rate": 1.8943304554886594e-05, + "loss": 0.3866, + "step": 3310 + }, + { + "epoch": 1.39, + "learning_rate": 1.8942545888531897e-05, + "loss": 0.4263, + "step": 3311 + }, + { + "epoch": 1.39, + "learning_rate": 1.8941786965131658e-05, + "loss": 0.4343, + "step": 3312 + }, + { + "epoch": 1.39, + "learning_rate": 1.8941027784707685e-05, + "loss": 0.3569, + "step": 3313 + }, + { + "epoch": 1.39, + "learning_rate": 1.8940268347281802e-05, + "loss": 0.3838, + "step": 3314 + }, + { + "epoch": 1.39, + "learning_rate": 1.8939508652875837e-05, + "loss": 0.3983, + "step": 3315 + }, + { + "epoch": 1.39, + "learning_rate": 1.8938748701511633e-05, + "loss": 0.3926, + "step": 3316 + }, + { + "epoch": 1.39, + "learning_rate": 1.8937988493211027e-05, + "loss": 0.3967, + "step": 3317 + }, + { + "epoch": 1.39, + "learning_rate": 1.8937228027995874e-05, + "loss": 0.3469, + "step": 3318 + }, + { + "epoch": 1.39, + "learning_rate": 1.8936467305888033e-05, + "loss": 0.3752, + "step": 3319 + }, + { + "epoch": 1.39, + "learning_rate": 1.893570632690937e-05, + "loss": 0.3537, + "step": 3320 + }, + { + "epoch": 1.39, + "learning_rate": 1.8934945091081755e-05, + "loss": 0.3916, + "step": 3321 + }, + { + "epoch": 1.39, + "learning_rate": 1.8934183598427072e-05, + "loss": 0.3867, + "step": 3322 + }, + { + "epoch": 1.39, + "learning_rate": 1.8933421848967213e-05, + "loss": 0.3915, + "step": 3323 + }, + { + "epoch": 1.39, + "learning_rate": 1.8932659842724067e-05, + "loss": 0.3388, + "step": 3324 + }, + { + "epoch": 1.39, + "learning_rate": 1.8931897579719543e-05, + "loss": 0.4215, + "step": 3325 + }, + { + "epoch": 1.39, + "learning_rate": 1.8931135059975548e-05, + "loss": 0.519, + "step": 3326 + }, + { + "epoch": 1.39, + "learning_rate": 1.8930372283514004e-05, + "loss": 0.3756, + "step": 3327 + }, + { + "epoch": 1.39, + "learning_rate": 1.8929609250356827e-05, + "loss": 0.354, + "step": 3328 + }, + { + "epoch": 1.39, + "learning_rate": 1.8928845960525957e-05, + "loss": 0.3802, + "step": 3329 + }, + { + "epoch": 1.39, + "learning_rate": 1.8928082414043342e-05, + "loss": 0.3898, + "step": 3330 + }, + { + "epoch": 1.39, + "learning_rate": 1.8927318610930912e-05, + "loss": 0.414, + "step": 3331 + }, + { + "epoch": 1.4, + "learning_rate": 1.892655455121063e-05, + "loss": 0.4356, + "step": 3332 + }, + { + "epoch": 1.4, + "learning_rate": 1.8925790234904463e-05, + "loss": 0.3701, + "step": 3333 + }, + { + "epoch": 1.4, + "learning_rate": 1.8925025662034376e-05, + "loss": 0.4053, + "step": 3334 + }, + { + "epoch": 1.4, + "learning_rate": 1.8924260832622347e-05, + "loss": 0.3682, + "step": 3335 + }, + { + "epoch": 1.4, + "learning_rate": 1.8923495746690354e-05, + "loss": 0.3782, + "step": 3336 + }, + { + "epoch": 1.4, + "learning_rate": 1.89227304042604e-05, + "loss": 0.4035, + "step": 3337 + }, + { + "epoch": 1.4, + "learning_rate": 1.8921964805354475e-05, + "loss": 0.3436, + "step": 3338 + }, + { + "epoch": 1.4, + "learning_rate": 1.892119894999459e-05, + "loss": 0.3769, + "step": 3339 + }, + { + "epoch": 1.4, + "learning_rate": 1.892043283820276e-05, + "loss": 0.4044, + "step": 3340 + }, + { + "epoch": 1.4, + "learning_rate": 1.8919666470001003e-05, + "loss": 0.4397, + "step": 3341 + }, + { + "epoch": 1.4, + "learning_rate": 1.8918899845411343e-05, + "loss": 0.3762, + "step": 3342 + }, + { + "epoch": 1.4, + "learning_rate": 1.891813296445583e-05, + "loss": 0.3667, + "step": 3343 + }, + { + "epoch": 1.4, + "learning_rate": 1.8917365827156493e-05, + "loss": 0.4077, + "step": 3344 + }, + { + "epoch": 1.4, + "learning_rate": 1.891659843353539e-05, + "loss": 0.4158, + "step": 3345 + }, + { + "epoch": 1.4, + "learning_rate": 1.8915830783614576e-05, + "loss": 0.3733, + "step": 3346 + }, + { + "epoch": 1.4, + "learning_rate": 1.891506287741612e-05, + "loss": 0.3935, + "step": 3347 + }, + { + "epoch": 1.4, + "learning_rate": 1.8914294714962093e-05, + "loss": 0.3812, + "step": 3348 + }, + { + "epoch": 1.4, + "learning_rate": 1.8913526296274573e-05, + "loss": 0.3571, + "step": 3349 + }, + { + "epoch": 1.4, + "learning_rate": 1.8912757621375654e-05, + "loss": 0.3853, + "step": 3350 + }, + { + "epoch": 1.4, + "learning_rate": 1.8911988690287422e-05, + "loss": 0.412, + "step": 3351 + }, + { + "epoch": 1.4, + "learning_rate": 1.891121950303198e-05, + "loss": 0.4038, + "step": 3352 + }, + { + "epoch": 1.4, + "learning_rate": 1.891045005963145e-05, + "loss": 0.3953, + "step": 3353 + }, + { + "epoch": 1.4, + "learning_rate": 1.890968036010793e-05, + "loss": 0.4022, + "step": 3354 + }, + { + "epoch": 1.4, + "learning_rate": 1.8908910404483564e-05, + "loss": 0.3732, + "step": 3355 + }, + { + "epoch": 1.41, + "learning_rate": 1.8908140192780466e-05, + "loss": 0.3713, + "step": 3356 + }, + { + "epoch": 1.41, + "learning_rate": 1.890736972502079e-05, + "loss": 0.4261, + "step": 3357 + }, + { + "epoch": 1.41, + "learning_rate": 1.890659900122667e-05, + "loss": 0.3903, + "step": 3358 + }, + { + "epoch": 1.41, + "learning_rate": 1.890582802142027e-05, + "loss": 0.4046, + "step": 3359 + }, + { + "epoch": 1.41, + "learning_rate": 1.8905056785623742e-05, + "loss": 0.3788, + "step": 3360 + }, + { + "epoch": 1.41, + "learning_rate": 1.890428529385926e-05, + "loss": 0.3524, + "step": 3361 + }, + { + "epoch": 1.41, + "learning_rate": 1.8903513546149e-05, + "loss": 0.3807, + "step": 3362 + }, + { + "epoch": 1.41, + "learning_rate": 1.8902741542515143e-05, + "loss": 0.33, + "step": 3363 + }, + { + "epoch": 1.41, + "learning_rate": 1.8901969282979882e-05, + "loss": 0.3963, + "step": 3364 + }, + { + "epoch": 1.41, + "learning_rate": 1.890119676756541e-05, + "loss": 0.3798, + "step": 3365 + }, + { + "epoch": 1.41, + "learning_rate": 1.8900423996293938e-05, + "loss": 0.4029, + "step": 3366 + }, + { + "epoch": 1.41, + "learning_rate": 1.889965096918768e-05, + "loss": 0.3793, + "step": 3367 + }, + { + "epoch": 1.41, + "learning_rate": 1.889887768626885e-05, + "loss": 0.3876, + "step": 3368 + }, + { + "epoch": 1.41, + "learning_rate": 1.8898104147559674e-05, + "loss": 0.3914, + "step": 3369 + }, + { + "epoch": 1.41, + "learning_rate": 1.8897330353082392e-05, + "loss": 0.3742, + "step": 3370 + }, + { + "epoch": 1.41, + "learning_rate": 1.8896556302859243e-05, + "loss": 0.4183, + "step": 3371 + }, + { + "epoch": 1.41, + "learning_rate": 1.889578199691248e-05, + "loss": 0.4014, + "step": 3372 + }, + { + "epoch": 1.41, + "learning_rate": 1.889500743526436e-05, + "loss": 0.3522, + "step": 3373 + }, + { + "epoch": 1.41, + "learning_rate": 1.889423261793714e-05, + "loss": 0.4073, + "step": 3374 + }, + { + "epoch": 1.41, + "learning_rate": 1.88934575449531e-05, + "loss": 0.3872, + "step": 3375 + }, + { + "epoch": 1.41, + "learning_rate": 1.889268221633451e-05, + "loss": 0.3361, + "step": 3376 + }, + { + "epoch": 1.41, + "learning_rate": 1.8891906632103664e-05, + "loss": 0.4251, + "step": 3377 + }, + { + "epoch": 1.41, + "learning_rate": 1.889113079228285e-05, + "loss": 0.3913, + "step": 3378 + }, + { + "epoch": 1.41, + "learning_rate": 1.8890354696894374e-05, + "loss": 0.3846, + "step": 3379 + }, + { + "epoch": 1.42, + "learning_rate": 1.888957834596054e-05, + "loss": 0.3444, + "step": 3380 + }, + { + "epoch": 1.42, + "learning_rate": 1.8888801739503667e-05, + "loss": 0.4703, + "step": 3381 + }, + { + "epoch": 1.42, + "learning_rate": 1.8888024877546074e-05, + "loss": 0.3833, + "step": 3382 + }, + { + "epoch": 1.42, + "learning_rate": 1.8887247760110094e-05, + "loss": 0.3652, + "step": 3383 + }, + { + "epoch": 1.42, + "learning_rate": 1.888647038721806e-05, + "loss": 0.4107, + "step": 3384 + }, + { + "epoch": 1.42, + "learning_rate": 1.8885692758892324e-05, + "loss": 0.3597, + "step": 3385 + }, + { + "epoch": 1.42, + "learning_rate": 1.8884914875155235e-05, + "loss": 0.3862, + "step": 3386 + }, + { + "epoch": 1.42, + "learning_rate": 1.8884136736029154e-05, + "loss": 0.4022, + "step": 3387 + }, + { + "epoch": 1.42, + "learning_rate": 1.8883358341536442e-05, + "loss": 0.4315, + "step": 3388 + }, + { + "epoch": 1.42, + "learning_rate": 1.8882579691699475e-05, + "loss": 0.4305, + "step": 3389 + }, + { + "epoch": 1.42, + "learning_rate": 1.8881800786540642e-05, + "loss": 0.3692, + "step": 3390 + }, + { + "epoch": 1.42, + "learning_rate": 1.8881021626082328e-05, + "loss": 0.3415, + "step": 3391 + }, + { + "epoch": 1.42, + "learning_rate": 1.8880242210346925e-05, + "loss": 0.4347, + "step": 3392 + }, + { + "epoch": 1.42, + "learning_rate": 1.8879462539356842e-05, + "loss": 0.3729, + "step": 3393 + }, + { + "epoch": 1.42, + "learning_rate": 1.8878682613134486e-05, + "loss": 0.3826, + "step": 3394 + }, + { + "epoch": 1.42, + "learning_rate": 1.8877902431702274e-05, + "loss": 0.4082, + "step": 3395 + }, + { + "epoch": 1.42, + "learning_rate": 1.8877121995082638e-05, + "loss": 0.3841, + "step": 3396 + }, + { + "epoch": 1.42, + "learning_rate": 1.8876341303298012e-05, + "loss": 0.3887, + "step": 3397 + }, + { + "epoch": 1.42, + "learning_rate": 1.8875560356370827e-05, + "loss": 0.3646, + "step": 3398 + }, + { + "epoch": 1.42, + "learning_rate": 1.8874779154323538e-05, + "loss": 0.3887, + "step": 3399 + }, + { + "epoch": 1.42, + "learning_rate": 1.8873997697178593e-05, + "loss": 0.3638, + "step": 3400 + }, + { + "epoch": 1.42, + "learning_rate": 1.8873215984958465e-05, + "loss": 0.4295, + "step": 3401 + }, + { + "epoch": 1.42, + "learning_rate": 1.8872434017685613e-05, + "loss": 0.3792, + "step": 3402 + }, + { + "epoch": 1.42, + "learning_rate": 1.8871651795382518e-05, + "loss": 0.3996, + "step": 3403 + }, + { + "epoch": 1.43, + "learning_rate": 1.8870869318071667e-05, + "loss": 0.4269, + "step": 3404 + }, + { + "epoch": 1.43, + "learning_rate": 1.8870086585775548e-05, + "loss": 0.3915, + "step": 3405 + }, + { + "epoch": 1.43, + "learning_rate": 1.8869303598516663e-05, + "loss": 0.3593, + "step": 3406 + }, + { + "epoch": 1.43, + "learning_rate": 1.8868520356317517e-05, + "loss": 0.3277, + "step": 3407 + }, + { + "epoch": 1.43, + "learning_rate": 1.886773685920062e-05, + "loss": 0.3957, + "step": 3408 + }, + { + "epoch": 1.43, + "learning_rate": 1.8866953107188498e-05, + "loss": 0.4089, + "step": 3409 + }, + { + "epoch": 1.43, + "learning_rate": 1.8866169100303677e-05, + "loss": 0.368, + "step": 3410 + }, + { + "epoch": 1.43, + "learning_rate": 1.8865384838568693e-05, + "loss": 0.3971, + "step": 3411 + }, + { + "epoch": 1.43, + "learning_rate": 1.886460032200609e-05, + "loss": 0.4052, + "step": 3412 + }, + { + "epoch": 1.43, + "learning_rate": 1.8863815550638414e-05, + "loss": 0.3769, + "step": 3413 + }, + { + "epoch": 1.43, + "learning_rate": 1.886303052448823e-05, + "loss": 0.3886, + "step": 3414 + }, + { + "epoch": 1.43, + "learning_rate": 1.8862245243578097e-05, + "loss": 0.4253, + "step": 3415 + }, + { + "epoch": 1.43, + "learning_rate": 1.8861459707930585e-05, + "loss": 0.3974, + "step": 3416 + }, + { + "epoch": 1.43, + "learning_rate": 1.886067391756828e-05, + "loss": 0.4251, + "step": 3417 + }, + { + "epoch": 1.43, + "learning_rate": 1.885988787251377e-05, + "loss": 0.3961, + "step": 3418 + }, + { + "epoch": 1.43, + "learning_rate": 1.885910157278964e-05, + "loss": 0.3985, + "step": 3419 + }, + { + "epoch": 1.43, + "learning_rate": 1.8858315018418497e-05, + "loss": 0.4277, + "step": 3420 + }, + { + "epoch": 1.43, + "learning_rate": 1.885752820942295e-05, + "loss": 0.4149, + "step": 3421 + }, + { + "epoch": 1.43, + "learning_rate": 1.8856741145825616e-05, + "loss": 0.3769, + "step": 3422 + }, + { + "epoch": 1.43, + "learning_rate": 1.8855953827649115e-05, + "loss": 0.3979, + "step": 3423 + }, + { + "epoch": 1.43, + "learning_rate": 1.8855166254916082e-05, + "loss": 0.398, + "step": 3424 + }, + { + "epoch": 1.43, + "learning_rate": 1.885437842764915e-05, + "loss": 0.3734, + "step": 3425 + }, + { + "epoch": 1.43, + "learning_rate": 1.885359034587097e-05, + "loss": 0.3775, + "step": 3426 + }, + { + "epoch": 1.43, + "learning_rate": 1.885280200960419e-05, + "loss": 0.3443, + "step": 3427 + }, + { + "epoch": 1.44, + "learning_rate": 1.8852013418871474e-05, + "loss": 0.4162, + "step": 3428 + }, + { + "epoch": 1.44, + "learning_rate": 1.885122457369549e-05, + "loss": 0.3529, + "step": 3429 + }, + { + "epoch": 1.44, + "learning_rate": 1.8850435474098903e-05, + "loss": 0.4343, + "step": 3430 + }, + { + "epoch": 1.44, + "learning_rate": 1.884964612010441e-05, + "loss": 0.3688, + "step": 3431 + }, + { + "epoch": 1.44, + "learning_rate": 1.884885651173469e-05, + "loss": 0.4096, + "step": 3432 + }, + { + "epoch": 1.44, + "learning_rate": 1.8848066649012437e-05, + "loss": 0.3725, + "step": 3433 + }, + { + "epoch": 1.44, + "learning_rate": 1.8847276531960363e-05, + "loss": 0.3621, + "step": 3434 + }, + { + "epoch": 1.44, + "learning_rate": 1.8846486160601175e-05, + "loss": 0.3706, + "step": 3435 + }, + { + "epoch": 1.44, + "learning_rate": 1.8845695534957594e-05, + "loss": 0.4113, + "step": 3436 + }, + { + "epoch": 1.44, + "learning_rate": 1.8844904655052344e-05, + "loss": 0.3906, + "step": 3437 + }, + { + "epoch": 1.44, + "learning_rate": 1.884411352090816e-05, + "loss": 0.3552, + "step": 3438 + }, + { + "epoch": 1.44, + "learning_rate": 1.8843322132547778e-05, + "loss": 0.4262, + "step": 3439 + }, + { + "epoch": 1.44, + "learning_rate": 1.884253048999395e-05, + "loss": 0.4011, + "step": 3440 + }, + { + "epoch": 1.44, + "learning_rate": 1.8841738593269432e-05, + "loss": 0.3708, + "step": 3441 + }, + { + "epoch": 1.44, + "learning_rate": 1.884094644239698e-05, + "loss": 0.3951, + "step": 3442 + }, + { + "epoch": 1.44, + "learning_rate": 1.8840154037399368e-05, + "loss": 0.3575, + "step": 3443 + }, + { + "epoch": 1.44, + "learning_rate": 1.8839361378299376e-05, + "loss": 0.4257, + "step": 3444 + }, + { + "epoch": 1.44, + "learning_rate": 1.8838568465119786e-05, + "loss": 0.3638, + "step": 3445 + }, + { + "epoch": 1.44, + "learning_rate": 1.8837775297883384e-05, + "loss": 0.3661, + "step": 3446 + }, + { + "epoch": 1.44, + "learning_rate": 1.883698187661298e-05, + "loss": 0.4143, + "step": 3447 + }, + { + "epoch": 1.44, + "learning_rate": 1.8836188201331368e-05, + "loss": 0.3825, + "step": 3448 + }, + { + "epoch": 1.44, + "learning_rate": 1.8835394272061368e-05, + "loss": 0.3355, + "step": 3449 + }, + { + "epoch": 1.44, + "learning_rate": 1.88346000888258e-05, + "loss": 0.4143, + "step": 3450 + }, + { + "epoch": 1.44, + "learning_rate": 1.8833805651647498e-05, + "loss": 0.428, + "step": 3451 + }, + { + "epoch": 1.45, + "learning_rate": 1.8833010960549286e-05, + "loss": 0.3992, + "step": 3452 + }, + { + "epoch": 1.45, + "learning_rate": 1.883221601555401e-05, + "loss": 0.3523, + "step": 3453 + }, + { + "epoch": 1.45, + "learning_rate": 1.8831420816684528e-05, + "loss": 0.3957, + "step": 3454 + }, + { + "epoch": 1.45, + "learning_rate": 1.8830625363963685e-05, + "loss": 0.366, + "step": 3455 + }, + { + "epoch": 1.45, + "learning_rate": 1.882982965741436e-05, + "loss": 0.3754, + "step": 3456 + }, + { + "epoch": 1.45, + "learning_rate": 1.882903369705941e-05, + "loss": 0.4156, + "step": 3457 + }, + { + "epoch": 1.45, + "learning_rate": 1.8828237482921725e-05, + "loss": 0.4268, + "step": 3458 + }, + { + "epoch": 1.45, + "learning_rate": 1.8827441015024187e-05, + "loss": 0.3932, + "step": 3459 + }, + { + "epoch": 1.45, + "learning_rate": 1.882664429338969e-05, + "loss": 0.4338, + "step": 3460 + }, + { + "epoch": 1.45, + "learning_rate": 1.8825847318041136e-05, + "loss": 0.4078, + "step": 3461 + }, + { + "epoch": 1.45, + "learning_rate": 1.882505008900143e-05, + "loss": 0.4001, + "step": 3462 + }, + { + "epoch": 1.45, + "learning_rate": 1.8824252606293492e-05, + "loss": 0.3735, + "step": 3463 + }, + { + "epoch": 1.45, + "learning_rate": 1.8823454869940243e-05, + "loss": 0.3941, + "step": 3464 + }, + { + "epoch": 1.45, + "learning_rate": 1.8822656879964617e-05, + "loss": 0.3981, + "step": 3465 + }, + { + "epoch": 1.45, + "learning_rate": 1.8821858636389547e-05, + "loss": 0.4284, + "step": 3466 + }, + { + "epoch": 1.45, + "learning_rate": 1.8821060139237975e-05, + "loss": 0.3802, + "step": 3467 + }, + { + "epoch": 1.45, + "learning_rate": 1.8820261388532858e-05, + "loss": 0.3674, + "step": 3468 + }, + { + "epoch": 1.45, + "learning_rate": 1.8819462384297155e-05, + "loss": 0.3818, + "step": 3469 + }, + { + "epoch": 1.45, + "learning_rate": 1.8818663126553833e-05, + "loss": 0.4197, + "step": 3470 + }, + { + "epoch": 1.45, + "learning_rate": 1.8817863615325865e-05, + "loss": 0.425, + "step": 3471 + }, + { + "epoch": 1.45, + "learning_rate": 1.881706385063623e-05, + "loss": 0.3757, + "step": 3472 + }, + { + "epoch": 1.45, + "learning_rate": 1.881626383250792e-05, + "loss": 0.3716, + "step": 3473 + }, + { + "epoch": 1.45, + "learning_rate": 1.8815463560963932e-05, + "loss": 0.392, + "step": 3474 + }, + { + "epoch": 1.45, + "learning_rate": 1.8814663036027262e-05, + "loss": 0.3917, + "step": 3475 + }, + { + "epoch": 1.46, + "learning_rate": 1.881386225772093e-05, + "loss": 0.356, + "step": 3476 + }, + { + "epoch": 1.46, + "learning_rate": 1.8813061226067945e-05, + "loss": 0.3455, + "step": 3477 + }, + { + "epoch": 1.46, + "learning_rate": 1.881225994109134e-05, + "loss": 0.4187, + "step": 3478 + }, + { + "epoch": 1.46, + "learning_rate": 1.881145840281414e-05, + "loss": 0.3953, + "step": 3479 + }, + { + "epoch": 1.46, + "learning_rate": 1.8810656611259387e-05, + "loss": 0.3728, + "step": 3480 + }, + { + "epoch": 1.46, + "learning_rate": 1.8809854566450128e-05, + "loss": 0.3603, + "step": 3481 + }, + { + "epoch": 1.46, + "learning_rate": 1.8809052268409414e-05, + "loss": 0.3356, + "step": 3482 + }, + { + "epoch": 1.46, + "learning_rate": 1.8808249717160318e-05, + "loss": 0.4093, + "step": 3483 + }, + { + "epoch": 1.46, + "learning_rate": 1.8807446912725893e-05, + "loss": 0.4172, + "step": 3484 + }, + { + "epoch": 1.46, + "learning_rate": 1.8806643855129225e-05, + "loss": 0.3647, + "step": 3485 + }, + { + "epoch": 1.46, + "learning_rate": 1.8805840544393393e-05, + "loss": 0.4281, + "step": 3486 + }, + { + "epoch": 1.46, + "learning_rate": 1.880503698054149e-05, + "loss": 0.3747, + "step": 3487 + }, + { + "epoch": 1.46, + "learning_rate": 1.8804233163596613e-05, + "loss": 0.3777, + "step": 3488 + }, + { + "epoch": 1.46, + "learning_rate": 1.8803429093581863e-05, + "loss": 0.3581, + "step": 3489 + }, + { + "epoch": 1.46, + "learning_rate": 1.880262477052036e-05, + "loss": 0.3936, + "step": 3490 + }, + { + "epoch": 1.46, + "learning_rate": 1.8801820194435215e-05, + "loss": 0.3912, + "step": 3491 + }, + { + "epoch": 1.46, + "learning_rate": 1.8801015365349562e-05, + "loss": 0.3921, + "step": 3492 + }, + { + "epoch": 1.46, + "learning_rate": 1.880021028328653e-05, + "loss": 0.358, + "step": 3493 + }, + { + "epoch": 1.46, + "learning_rate": 1.8799404948269267e-05, + "loss": 0.3555, + "step": 3494 + }, + { + "epoch": 1.46, + "learning_rate": 1.879859936032091e-05, + "loss": 0.3767, + "step": 3495 + }, + { + "epoch": 1.46, + "learning_rate": 1.8797793519464628e-05, + "loss": 0.3615, + "step": 3496 + }, + { + "epoch": 1.46, + "learning_rate": 1.8796987425723578e-05, + "loss": 0.3714, + "step": 3497 + }, + { + "epoch": 1.46, + "learning_rate": 1.879618107912093e-05, + "loss": 0.3775, + "step": 3498 + }, + { + "epoch": 1.46, + "learning_rate": 1.8795374479679862e-05, + "loss": 0.4144, + "step": 3499 + }, + { + "epoch": 1.47, + "learning_rate": 1.879456762742356e-05, + "loss": 0.3544, + "step": 3500 + }, + { + "epoch": 1.47, + "learning_rate": 1.8793760522375214e-05, + "loss": 0.3959, + "step": 3501 + }, + { + "epoch": 1.47, + "learning_rate": 1.8792953164558026e-05, + "loss": 0.4117, + "step": 3502 + }, + { + "epoch": 1.47, + "learning_rate": 1.8792145553995203e-05, + "loss": 0.3799, + "step": 3503 + }, + { + "epoch": 1.47, + "learning_rate": 1.879133769070996e-05, + "loss": 0.4163, + "step": 3504 + }, + { + "epoch": 1.47, + "learning_rate": 1.8790529574725514e-05, + "loss": 0.4168, + "step": 3505 + }, + { + "epoch": 1.47, + "learning_rate": 1.8789721206065096e-05, + "loss": 0.3888, + "step": 3506 + }, + { + "epoch": 1.47, + "learning_rate": 1.8788912584751938e-05, + "loss": 0.3918, + "step": 3507 + }, + { + "epoch": 1.47, + "learning_rate": 1.8788103710809292e-05, + "loss": 0.3519, + "step": 3508 + }, + { + "epoch": 1.47, + "learning_rate": 1.8787294584260402e-05, + "loss": 0.377, + "step": 3509 + }, + { + "epoch": 1.47, + "learning_rate": 1.8786485205128526e-05, + "loss": 0.3672, + "step": 3510 + }, + { + "epoch": 1.47, + "learning_rate": 1.878567557343693e-05, + "loss": 0.4139, + "step": 3511 + }, + { + "epoch": 1.47, + "learning_rate": 1.8784865689208888e-05, + "loss": 0.3682, + "step": 3512 + }, + { + "epoch": 1.47, + "learning_rate": 1.8784055552467674e-05, + "loss": 0.4095, + "step": 3513 + }, + { + "epoch": 1.47, + "learning_rate": 1.878324516323658e-05, + "loss": 0.4295, + "step": 3514 + }, + { + "epoch": 1.47, + "learning_rate": 1.87824345215389e-05, + "loss": 0.4504, + "step": 3515 + }, + { + "epoch": 1.47, + "learning_rate": 1.878162362739793e-05, + "loss": 0.3981, + "step": 3516 + }, + { + "epoch": 1.47, + "learning_rate": 1.878081248083698e-05, + "loss": 0.4034, + "step": 3517 + }, + { + "epoch": 1.47, + "learning_rate": 1.878000108187937e-05, + "loss": 0.3617, + "step": 3518 + }, + { + "epoch": 1.47, + "learning_rate": 1.877918943054842e-05, + "loss": 0.3635, + "step": 3519 + }, + { + "epoch": 1.47, + "learning_rate": 1.877837752686746e-05, + "loss": 0.4073, + "step": 3520 + }, + { + "epoch": 1.47, + "learning_rate": 1.8777565370859826e-05, + "loss": 0.3933, + "step": 3521 + }, + { + "epoch": 1.47, + "learning_rate": 1.8776752962548866e-05, + "loss": 0.3756, + "step": 3522 + }, + { + "epoch": 1.47, + "learning_rate": 1.8775940301957932e-05, + "loss": 0.3968, + "step": 3523 + }, + { + "epoch": 1.48, + "learning_rate": 1.877512738911038e-05, + "loss": 0.3968, + "step": 3524 + }, + { + "epoch": 1.48, + "learning_rate": 1.8774314224029578e-05, + "loss": 0.4166, + "step": 3525 + }, + { + "epoch": 1.48, + "learning_rate": 1.87735008067389e-05, + "loss": 0.3903, + "step": 3526 + }, + { + "epoch": 1.48, + "learning_rate": 1.8772687137261728e-05, + "loss": 0.3604, + "step": 3527 + }, + { + "epoch": 1.48, + "learning_rate": 1.8771873215621447e-05, + "loss": 0.3919, + "step": 3528 + }, + { + "epoch": 1.48, + "learning_rate": 1.8771059041841454e-05, + "loss": 0.4071, + "step": 3529 + }, + { + "epoch": 1.48, + "learning_rate": 1.8770244615945153e-05, + "loss": 0.4072, + "step": 3530 + }, + { + "epoch": 1.48, + "learning_rate": 1.8769429937955954e-05, + "loss": 0.3793, + "step": 3531 + }, + { + "epoch": 1.48, + "learning_rate": 1.8768615007897274e-05, + "loss": 0.3788, + "step": 3532 + }, + { + "epoch": 1.48, + "learning_rate": 1.8767799825792533e-05, + "loss": 0.3738, + "step": 3533 + }, + { + "epoch": 1.48, + "learning_rate": 1.8766984391665173e-05, + "loss": 0.3879, + "step": 3534 + }, + { + "epoch": 1.48, + "learning_rate": 1.876616870553862e-05, + "loss": 0.3786, + "step": 3535 + }, + { + "epoch": 1.48, + "learning_rate": 1.876535276743633e-05, + "loss": 0.4138, + "step": 3536 + }, + { + "epoch": 1.48, + "learning_rate": 1.8764536577381748e-05, + "loss": 0.4008, + "step": 3537 + }, + { + "epoch": 1.48, + "learning_rate": 1.8763720135398348e-05, + "loss": 0.3867, + "step": 3538 + }, + { + "epoch": 1.48, + "learning_rate": 1.8762903441509582e-05, + "loss": 0.3631, + "step": 3539 + }, + { + "epoch": 1.48, + "learning_rate": 1.8762086495738937e-05, + "loss": 0.35, + "step": 3540 + }, + { + "epoch": 1.48, + "learning_rate": 1.876126929810989e-05, + "loss": 0.3891, + "step": 3541 + }, + { + "epoch": 1.48, + "learning_rate": 1.8760451848645932e-05, + "loss": 0.3692, + "step": 3542 + }, + { + "epoch": 1.48, + "learning_rate": 1.8759634147370555e-05, + "loss": 0.4024, + "step": 3543 + }, + { + "epoch": 1.48, + "learning_rate": 1.8758816194307272e-05, + "loss": 0.3913, + "step": 3544 + }, + { + "epoch": 1.48, + "learning_rate": 1.875799798947959e-05, + "loss": 0.3868, + "step": 3545 + }, + { + "epoch": 1.48, + "learning_rate": 1.8757179532911026e-05, + "loss": 0.3665, + "step": 3546 + }, + { + "epoch": 1.49, + "learning_rate": 1.8756360824625107e-05, + "loss": 0.4118, + "step": 3547 + }, + { + "epoch": 1.49, + "learning_rate": 1.875554186464537e-05, + "loss": 0.3749, + "step": 3548 + }, + { + "epoch": 1.49, + "learning_rate": 1.8754722652995346e-05, + "loss": 0.4166, + "step": 3549 + }, + { + "epoch": 1.49, + "learning_rate": 1.8753903189698594e-05, + "loss": 0.3962, + "step": 3550 + }, + { + "epoch": 1.49, + "learning_rate": 1.875308347477866e-05, + "loss": 0.4516, + "step": 3551 + }, + { + "epoch": 1.49, + "learning_rate": 1.8752263508259107e-05, + "loss": 0.4046, + "step": 3552 + }, + { + "epoch": 1.49, + "learning_rate": 1.875144329016351e-05, + "loss": 0.4097, + "step": 3553 + }, + { + "epoch": 1.49, + "learning_rate": 1.875062282051544e-05, + "loss": 0.4241, + "step": 3554 + }, + { + "epoch": 1.49, + "learning_rate": 1.874980209933848e-05, + "loss": 0.38, + "step": 3555 + }, + { + "epoch": 1.49, + "learning_rate": 1.8748981126656224e-05, + "loss": 0.3898, + "step": 3556 + }, + { + "epoch": 1.49, + "learning_rate": 1.8748159902492273e-05, + "loss": 0.4338, + "step": 3557 + }, + { + "epoch": 1.49, + "learning_rate": 1.8747338426870222e-05, + "loss": 0.4228, + "step": 3558 + }, + { + "epoch": 1.49, + "learning_rate": 1.8746516699813693e-05, + "loss": 0.35, + "step": 3559 + }, + { + "epoch": 1.49, + "learning_rate": 1.8745694721346305e-05, + "loss": 0.3797, + "step": 3560 + }, + { + "epoch": 1.49, + "learning_rate": 1.874487249149168e-05, + "loss": 0.4204, + "step": 3561 + }, + { + "epoch": 1.49, + "learning_rate": 1.8744050010273458e-05, + "loss": 0.4093, + "step": 3562 + }, + { + "epoch": 1.49, + "learning_rate": 1.874322727771528e-05, + "loss": 0.4614, + "step": 3563 + }, + { + "epoch": 1.49, + "learning_rate": 1.8742404293840784e-05, + "loss": 0.3874, + "step": 3564 + }, + { + "epoch": 1.49, + "learning_rate": 1.8741581058673642e-05, + "loss": 0.3658, + "step": 3565 + }, + { + "epoch": 1.49, + "learning_rate": 1.874075757223751e-05, + "loss": 0.4201, + "step": 3566 + }, + { + "epoch": 1.49, + "learning_rate": 1.8739933834556054e-05, + "loss": 0.3875, + "step": 3567 + }, + { + "epoch": 1.49, + "learning_rate": 1.8739109845652957e-05, + "loss": 0.4209, + "step": 3568 + }, + { + "epoch": 1.49, + "learning_rate": 1.87382856055519e-05, + "loss": 0.347, + "step": 3569 + }, + { + "epoch": 1.49, + "learning_rate": 1.873746111427658e-05, + "loss": 0.4023, + "step": 3570 + }, + { + "epoch": 1.5, + "learning_rate": 1.8736636371850694e-05, + "loss": 0.3956, + "step": 3571 + }, + { + "epoch": 1.5, + "learning_rate": 1.8735811378297946e-05, + "loss": 0.406, + "step": 3572 + }, + { + "epoch": 1.5, + "learning_rate": 1.8734986133642058e-05, + "loss": 0.3692, + "step": 3573 + }, + { + "epoch": 1.5, + "learning_rate": 1.8734160637906738e-05, + "loss": 0.3746, + "step": 3574 + }, + { + "epoch": 1.5, + "learning_rate": 1.8733334891115725e-05, + "loss": 0.3654, + "step": 3575 + }, + { + "epoch": 1.5, + "learning_rate": 1.873250889329275e-05, + "loss": 0.3501, + "step": 3576 + }, + { + "epoch": 1.5, + "learning_rate": 1.8731682644461555e-05, + "loss": 0.4146, + "step": 3577 + }, + { + "epoch": 1.5, + "learning_rate": 1.873085614464589e-05, + "loss": 0.3821, + "step": 3578 + }, + { + "epoch": 1.5, + "learning_rate": 1.8730029393869512e-05, + "loss": 0.3517, + "step": 3579 + }, + { + "epoch": 1.5, + "learning_rate": 1.8729202392156186e-05, + "loss": 0.4147, + "step": 3580 + }, + { + "epoch": 1.5, + "learning_rate": 1.8728375139529688e-05, + "loss": 0.4, + "step": 3581 + }, + { + "epoch": 1.5, + "learning_rate": 1.872754763601379e-05, + "loss": 0.3598, + "step": 3582 + }, + { + "epoch": 1.5, + "learning_rate": 1.8726719881632278e-05, + "loss": 0.405, + "step": 3583 + }, + { + "epoch": 1.5, + "learning_rate": 1.8725891876408946e-05, + "loss": 0.4087, + "step": 3584 + }, + { + "epoch": 1.5, + "learning_rate": 1.8725063620367598e-05, + "loss": 0.4424, + "step": 3585 + }, + { + "epoch": 1.5, + "learning_rate": 1.872423511353204e-05, + "loss": 0.3516, + "step": 3586 + }, + { + "epoch": 1.5, + "learning_rate": 1.8723406355926078e-05, + "loss": 0.4001, + "step": 3587 + }, + { + "epoch": 1.5, + "learning_rate": 1.872257734757355e-05, + "loss": 0.3657, + "step": 3588 + }, + { + "epoch": 1.5, + "learning_rate": 1.8721748088498272e-05, + "loss": 0.3784, + "step": 3589 + }, + { + "epoch": 1.5, + "learning_rate": 1.8720918578724085e-05, + "loss": 0.4014, + "step": 3590 + }, + { + "epoch": 1.5, + "learning_rate": 1.8720088818274835e-05, + "loss": 0.4215, + "step": 3591 + }, + { + "epoch": 1.5, + "learning_rate": 1.8719258807174366e-05, + "loss": 0.3759, + "step": 3592 + }, + { + "epoch": 1.5, + "learning_rate": 1.8718428545446545e-05, + "loss": 0.3367, + "step": 3593 + }, + { + "epoch": 1.5, + "learning_rate": 1.8717598033115224e-05, + "loss": 0.3609, + "step": 3594 + }, + { + "epoch": 1.51, + "learning_rate": 1.871676727020429e-05, + "loss": 0.3814, + "step": 3595 + }, + { + "epoch": 1.51, + "learning_rate": 1.871593625673761e-05, + "loss": 0.4125, + "step": 3596 + }, + { + "epoch": 1.51, + "learning_rate": 1.8715104992739085e-05, + "loss": 0.3841, + "step": 3597 + }, + { + "epoch": 1.51, + "learning_rate": 1.8714273478232597e-05, + "loss": 0.3855, + "step": 3598 + }, + { + "epoch": 1.51, + "learning_rate": 1.871344171324205e-05, + "loss": 0.4007, + "step": 3599 + }, + { + "epoch": 1.51, + "learning_rate": 1.871260969779135e-05, + "loss": 0.3679, + "step": 3600 + }, + { + "epoch": 1.51, + "learning_rate": 1.871177743190442e-05, + "loss": 0.3722, + "step": 3601 + }, + { + "epoch": 1.51, + "learning_rate": 1.871094491560518e-05, + "loss": 0.3753, + "step": 3602 + }, + { + "epoch": 1.51, + "learning_rate": 1.8710112148917555e-05, + "loss": 0.4449, + "step": 3603 + }, + { + "epoch": 1.51, + "learning_rate": 1.8709279131865486e-05, + "loss": 0.4068, + "step": 3604 + }, + { + "epoch": 1.51, + "learning_rate": 1.8708445864472917e-05, + "loss": 0.3895, + "step": 3605 + }, + { + "epoch": 1.51, + "learning_rate": 1.87076123467638e-05, + "loss": 0.3729, + "step": 3606 + }, + { + "epoch": 1.51, + "learning_rate": 1.8706778578762094e-05, + "loss": 0.3501, + "step": 3607 + }, + { + "epoch": 1.51, + "learning_rate": 1.870594456049176e-05, + "loss": 0.337, + "step": 3608 + }, + { + "epoch": 1.51, + "learning_rate": 1.870511029197678e-05, + "loss": 0.3914, + "step": 3609 + }, + { + "epoch": 1.51, + "learning_rate": 1.8704275773241126e-05, + "loss": 0.3952, + "step": 3610 + }, + { + "epoch": 1.51, + "learning_rate": 1.8703441004308792e-05, + "loss": 0.3786, + "step": 3611 + }, + { + "epoch": 1.51, + "learning_rate": 1.8702605985203767e-05, + "loss": 0.38, + "step": 3612 + }, + { + "epoch": 1.51, + "learning_rate": 1.8701770715950058e-05, + "loss": 0.4337, + "step": 3613 + }, + { + "epoch": 1.51, + "learning_rate": 1.8700935196571666e-05, + "loss": 0.3772, + "step": 3614 + }, + { + "epoch": 1.51, + "learning_rate": 1.8700099427092618e-05, + "loss": 0.3557, + "step": 3615 + }, + { + "epoch": 1.51, + "learning_rate": 1.869926340753693e-05, + "loss": 0.363, + "step": 3616 + }, + { + "epoch": 1.51, + "learning_rate": 1.8698427137928637e-05, + "loss": 0.4203, + "step": 3617 + }, + { + "epoch": 1.51, + "learning_rate": 1.8697590618291772e-05, + "loss": 0.4204, + "step": 3618 + }, + { + "epoch": 1.52, + "learning_rate": 1.8696753848650386e-05, + "loss": 0.3818, + "step": 3619 + }, + { + "epoch": 1.52, + "learning_rate": 1.8695916829028525e-05, + "loss": 0.4434, + "step": 3620 + }, + { + "epoch": 1.52, + "learning_rate": 1.8695079559450255e-05, + "loss": 0.4064, + "step": 3621 + }, + { + "epoch": 1.52, + "learning_rate": 1.8694242039939636e-05, + "loss": 0.3619, + "step": 3622 + }, + { + "epoch": 1.52, + "learning_rate": 1.8693404270520745e-05, + "loss": 0.3922, + "step": 3623 + }, + { + "epoch": 1.52, + "learning_rate": 1.8692566251217665e-05, + "loss": 0.3681, + "step": 3624 + }, + { + "epoch": 1.52, + "learning_rate": 1.8691727982054478e-05, + "loss": 0.3653, + "step": 3625 + }, + { + "epoch": 1.52, + "learning_rate": 1.8690889463055285e-05, + "loss": 0.3871, + "step": 3626 + }, + { + "epoch": 1.52, + "learning_rate": 1.8690050694244188e-05, + "loss": 0.3941, + "step": 3627 + }, + { + "epoch": 1.52, + "learning_rate": 1.868921167564529e-05, + "loss": 0.3549, + "step": 3628 + }, + { + "epoch": 1.52, + "learning_rate": 1.8688372407282717e-05, + "loss": 0.3583, + "step": 3629 + }, + { + "epoch": 1.52, + "learning_rate": 1.868753288918059e-05, + "loss": 0.3913, + "step": 3630 + }, + { + "epoch": 1.52, + "learning_rate": 1.868669312136304e-05, + "loss": 0.3775, + "step": 3631 + }, + { + "epoch": 1.52, + "learning_rate": 1.86858531038542e-05, + "loss": 0.3765, + "step": 3632 + }, + { + "epoch": 1.52, + "learning_rate": 1.8685012836678225e-05, + "loss": 0.3596, + "step": 3633 + }, + { + "epoch": 1.52, + "learning_rate": 1.8684172319859258e-05, + "loss": 0.3955, + "step": 3634 + }, + { + "epoch": 1.52, + "learning_rate": 1.8683331553421467e-05, + "loss": 0.468, + "step": 3635 + }, + { + "epoch": 1.52, + "learning_rate": 1.8682490537389017e-05, + "loss": 0.4153, + "step": 3636 + }, + { + "epoch": 1.52, + "learning_rate": 1.868164927178608e-05, + "loss": 0.3604, + "step": 3637 + }, + { + "epoch": 1.52, + "learning_rate": 1.8680807756636837e-05, + "loss": 0.3752, + "step": 3638 + }, + { + "epoch": 1.52, + "learning_rate": 1.8679965991965484e-05, + "loss": 0.3992, + "step": 3639 + }, + { + "epoch": 1.52, + "learning_rate": 1.8679123977796205e-05, + "loss": 0.3463, + "step": 3640 + }, + { + "epoch": 1.52, + "learning_rate": 1.867828171415321e-05, + "loss": 0.412, + "step": 3641 + }, + { + "epoch": 1.52, + "learning_rate": 1.867743920106071e-05, + "loss": 0.3994, + "step": 3642 + }, + { + "epoch": 1.53, + "learning_rate": 1.867659643854292e-05, + "loss": 0.4042, + "step": 3643 + }, + { + "epoch": 1.53, + "learning_rate": 1.867575342662406e-05, + "loss": 0.4215, + "step": 3644 + }, + { + "epoch": 1.53, + "learning_rate": 1.8674910165328374e-05, + "loss": 0.423, + "step": 3645 + }, + { + "epoch": 1.53, + "learning_rate": 1.8674066654680088e-05, + "loss": 0.3863, + "step": 3646 + }, + { + "epoch": 1.53, + "learning_rate": 1.8673222894703456e-05, + "loss": 0.419, + "step": 3647 + }, + { + "epoch": 1.53, + "learning_rate": 1.8672378885422728e-05, + "loss": 0.4253, + "step": 3648 + }, + { + "epoch": 1.53, + "learning_rate": 1.867153462686216e-05, + "loss": 0.3741, + "step": 3649 + }, + { + "epoch": 1.53, + "learning_rate": 1.8670690119046032e-05, + "loss": 0.3937, + "step": 3650 + }, + { + "epoch": 1.53, + "learning_rate": 1.8669845361998607e-05, + "loss": 0.3863, + "step": 3651 + }, + { + "epoch": 1.53, + "learning_rate": 1.8669000355744172e-05, + "loss": 0.4042, + "step": 3652 + }, + { + "epoch": 1.53, + "learning_rate": 1.8668155100307013e-05, + "loss": 0.4318, + "step": 3653 + }, + { + "epoch": 1.53, + "learning_rate": 1.8667309595711426e-05, + "loss": 0.4049, + "step": 3654 + }, + { + "epoch": 1.53, + "learning_rate": 1.8666463841981717e-05, + "loss": 0.4273, + "step": 3655 + }, + { + "epoch": 1.53, + "learning_rate": 1.8665617839142198e-05, + "loss": 0.3994, + "step": 3656 + }, + { + "epoch": 1.53, + "learning_rate": 1.866477158721718e-05, + "loss": 0.3565, + "step": 3657 + }, + { + "epoch": 1.53, + "learning_rate": 1.8663925086230998e-05, + "loss": 0.3925, + "step": 3658 + }, + { + "epoch": 1.53, + "learning_rate": 1.866307833620797e-05, + "loss": 0.446, + "step": 3659 + }, + { + "epoch": 1.53, + "learning_rate": 1.8662231337172446e-05, + "loss": 0.388, + "step": 3660 + }, + { + "epoch": 1.53, + "learning_rate": 1.866138408914877e-05, + "loss": 0.3558, + "step": 3661 + }, + { + "epoch": 1.53, + "learning_rate": 1.8660536592161293e-05, + "loss": 0.3339, + "step": 3662 + }, + { + "epoch": 1.53, + "learning_rate": 1.8659688846234376e-05, + "loss": 0.357, + "step": 3663 + }, + { + "epoch": 1.53, + "learning_rate": 1.8658840851392386e-05, + "loss": 0.4077, + "step": 3664 + }, + { + "epoch": 1.53, + "learning_rate": 1.8657992607659706e-05, + "loss": 0.3655, + "step": 3665 + }, + { + "epoch": 1.53, + "learning_rate": 1.8657144115060707e-05, + "loss": 0.3917, + "step": 3666 + }, + { + "epoch": 1.54, + "learning_rate": 1.8656295373619785e-05, + "loss": 0.4059, + "step": 3667 + }, + { + "epoch": 1.54, + "learning_rate": 1.8655446383361332e-05, + "loss": 0.4232, + "step": 3668 + }, + { + "epoch": 1.54, + "learning_rate": 1.8654597144309753e-05, + "loss": 0.3869, + "step": 3669 + }, + { + "epoch": 1.54, + "learning_rate": 1.865374765648946e-05, + "loss": 0.3615, + "step": 3670 + }, + { + "epoch": 1.54, + "learning_rate": 1.865289791992487e-05, + "loss": 0.368, + "step": 3671 + }, + { + "epoch": 1.54, + "learning_rate": 1.8652047934640405e-05, + "loss": 0.3497, + "step": 3672 + }, + { + "epoch": 1.54, + "learning_rate": 1.86511977006605e-05, + "loss": 0.4531, + "step": 3673 + }, + { + "epoch": 1.54, + "learning_rate": 1.8650347218009596e-05, + "loss": 0.3943, + "step": 3674 + }, + { + "epoch": 1.54, + "learning_rate": 1.864949648671214e-05, + "loss": 0.3597, + "step": 3675 + }, + { + "epoch": 1.54, + "learning_rate": 1.8648645506792578e-05, + "loss": 0.396, + "step": 3676 + }, + { + "epoch": 1.54, + "learning_rate": 1.864779427827538e-05, + "loss": 0.3928, + "step": 3677 + }, + { + "epoch": 1.54, + "learning_rate": 1.8646942801185006e-05, + "loss": 0.3414, + "step": 3678 + }, + { + "epoch": 1.54, + "learning_rate": 1.8646091075545936e-05, + "loss": 0.3588, + "step": 3679 + }, + { + "epoch": 1.54, + "learning_rate": 1.864523910138265e-05, + "loss": 0.4464, + "step": 3680 + }, + { + "epoch": 1.54, + "learning_rate": 1.864438687871964e-05, + "loss": 0.3666, + "step": 3681 + }, + { + "epoch": 1.54, + "learning_rate": 1.8643534407581397e-05, + "loss": 0.4229, + "step": 3682 + }, + { + "epoch": 1.54, + "learning_rate": 1.8642681687992427e-05, + "loss": 0.3988, + "step": 3683 + }, + { + "epoch": 1.54, + "learning_rate": 1.8641828719977244e-05, + "loss": 0.3685, + "step": 3684 + }, + { + "epoch": 1.54, + "learning_rate": 1.864097550356036e-05, + "loss": 0.3398, + "step": 3685 + }, + { + "epoch": 1.54, + "learning_rate": 1.8640122038766307e-05, + "loss": 0.4044, + "step": 3686 + }, + { + "epoch": 1.54, + "learning_rate": 1.863926832561961e-05, + "loss": 0.3999, + "step": 3687 + }, + { + "epoch": 1.54, + "learning_rate": 1.8638414364144813e-05, + "loss": 0.3707, + "step": 3688 + }, + { + "epoch": 1.54, + "learning_rate": 1.863756015436646e-05, + "loss": 0.3947, + "step": 3689 + }, + { + "epoch": 1.54, + "learning_rate": 1.8636705696309104e-05, + "loss": 0.4347, + "step": 3690 + }, + { + "epoch": 1.55, + "learning_rate": 1.8635850989997308e-05, + "loss": 0.3899, + "step": 3691 + }, + { + "epoch": 1.55, + "learning_rate": 1.863499603545564e-05, + "loss": 0.362, + "step": 3692 + }, + { + "epoch": 1.55, + "learning_rate": 1.8634140832708673e-05, + "loss": 0.4191, + "step": 3693 + }, + { + "epoch": 1.55, + "learning_rate": 1.8633285381780987e-05, + "loss": 0.4036, + "step": 3694 + }, + { + "epoch": 1.55, + "learning_rate": 1.8632429682697175e-05, + "loss": 0.4099, + "step": 3695 + }, + { + "epoch": 1.55, + "learning_rate": 1.8631573735481834e-05, + "loss": 0.43, + "step": 3696 + }, + { + "epoch": 1.55, + "learning_rate": 1.8630717540159562e-05, + "loss": 0.3987, + "step": 3697 + }, + { + "epoch": 1.55, + "learning_rate": 1.8629861096754977e-05, + "loss": 0.3406, + "step": 3698 + }, + { + "epoch": 1.55, + "learning_rate": 1.862900440529269e-05, + "loss": 0.3903, + "step": 3699 + }, + { + "epoch": 1.55, + "learning_rate": 1.862814746579733e-05, + "loss": 0.3315, + "step": 3700 + }, + { + "epoch": 1.55, + "learning_rate": 1.8627290278293525e-05, + "loss": 0.3851, + "step": 3701 + }, + { + "epoch": 1.55, + "learning_rate": 1.862643284280592e-05, + "loss": 0.3995, + "step": 3702 + }, + { + "epoch": 1.55, + "learning_rate": 1.8625575159359157e-05, + "loss": 0.393, + "step": 3703 + }, + { + "epoch": 1.55, + "learning_rate": 1.862471722797789e-05, + "loss": 0.405, + "step": 3704 + }, + { + "epoch": 1.55, + "learning_rate": 1.8623859048686777e-05, + "loss": 0.3704, + "step": 3705 + }, + { + "epoch": 1.55, + "learning_rate": 1.862300062151049e-05, + "loss": 0.3837, + "step": 3706 + }, + { + "epoch": 1.55, + "learning_rate": 1.8622141946473705e-05, + "loss": 0.3839, + "step": 3707 + }, + { + "epoch": 1.55, + "learning_rate": 1.8621283023601097e-05, + "loss": 0.4162, + "step": 3708 + }, + { + "epoch": 1.55, + "learning_rate": 1.862042385291736e-05, + "loss": 0.4516, + "step": 3709 + }, + { + "epoch": 1.55, + "learning_rate": 1.8619564434447192e-05, + "loss": 0.3776, + "step": 3710 + }, + { + "epoch": 1.55, + "learning_rate": 1.861870476821529e-05, + "loss": 0.4294, + "step": 3711 + }, + { + "epoch": 1.55, + "learning_rate": 1.8617844854246368e-05, + "loss": 0.3658, + "step": 3712 + }, + { + "epoch": 1.55, + "learning_rate": 1.8616984692565144e-05, + "loss": 0.3766, + "step": 3713 + }, + { + "epoch": 1.55, + "learning_rate": 1.861612428319634e-05, + "loss": 0.3958, + "step": 3714 + }, + { + "epoch": 1.56, + "learning_rate": 1.861526362616469e-05, + "loss": 0.4002, + "step": 3715 + }, + { + "epoch": 1.56, + "learning_rate": 1.8614402721494934e-05, + "loss": 0.3409, + "step": 3716 + }, + { + "epoch": 1.56, + "learning_rate": 1.8613541569211815e-05, + "loss": 0.3693, + "step": 3717 + }, + { + "epoch": 1.56, + "learning_rate": 1.8612680169340085e-05, + "loss": 0.3764, + "step": 3718 + }, + { + "epoch": 1.56, + "learning_rate": 1.861181852190451e-05, + "loss": 0.3525, + "step": 3719 + }, + { + "epoch": 1.56, + "learning_rate": 1.8610956626929852e-05, + "loss": 0.3591, + "step": 3720 + }, + { + "epoch": 1.56, + "learning_rate": 1.8610094484440884e-05, + "loss": 0.3761, + "step": 3721 + }, + { + "epoch": 1.56, + "learning_rate": 1.8609232094462394e-05, + "loss": 0.3549, + "step": 3722 + }, + { + "epoch": 1.56, + "learning_rate": 1.8608369457019166e-05, + "loss": 0.3992, + "step": 3723 + }, + { + "epoch": 1.56, + "learning_rate": 1.8607506572135994e-05, + "loss": 0.3731, + "step": 3724 + }, + { + "epoch": 1.56, + "learning_rate": 1.860664343983769e-05, + "loss": 0.3797, + "step": 3725 + }, + { + "epoch": 1.56, + "learning_rate": 1.8605780060149056e-05, + "loss": 0.3526, + "step": 3726 + }, + { + "epoch": 1.56, + "learning_rate": 1.8604916433094908e-05, + "loss": 0.3517, + "step": 3727 + }, + { + "epoch": 1.56, + "learning_rate": 1.860405255870007e-05, + "loss": 0.3643, + "step": 3728 + }, + { + "epoch": 1.56, + "learning_rate": 1.8603188436989382e-05, + "loss": 0.3997, + "step": 3729 + }, + { + "epoch": 1.56, + "learning_rate": 1.8602324067987678e-05, + "loss": 0.3371, + "step": 3730 + }, + { + "epoch": 1.56, + "learning_rate": 1.86014594517198e-05, + "loss": 0.4331, + "step": 3731 + }, + { + "epoch": 1.56, + "learning_rate": 1.8600594588210602e-05, + "loss": 0.4084, + "step": 3732 + }, + { + "epoch": 1.56, + "learning_rate": 1.8599729477484944e-05, + "loss": 0.3512, + "step": 3733 + }, + { + "epoch": 1.56, + "learning_rate": 1.8598864119567693e-05, + "loss": 0.4022, + "step": 3734 + }, + { + "epoch": 1.56, + "learning_rate": 1.8597998514483724e-05, + "loss": 0.3846, + "step": 3735 + }, + { + "epoch": 1.56, + "learning_rate": 1.859713266225792e-05, + "loss": 0.3828, + "step": 3736 + }, + { + "epoch": 1.56, + "learning_rate": 1.8596266562915163e-05, + "loss": 0.4152, + "step": 3737 + }, + { + "epoch": 1.56, + "learning_rate": 1.8595400216480356e-05, + "loss": 0.347, + "step": 3738 + }, + { + "epoch": 1.57, + "learning_rate": 1.8594533622978394e-05, + "loss": 0.4058, + "step": 3739 + }, + { + "epoch": 1.57, + "learning_rate": 1.859366678243419e-05, + "loss": 0.4159, + "step": 3740 + }, + { + "epoch": 1.57, + "learning_rate": 1.859279969487266e-05, + "loss": 0.3883, + "step": 3741 + }, + { + "epoch": 1.57, + "learning_rate": 1.859193236031873e-05, + "loss": 0.3622, + "step": 3742 + }, + { + "epoch": 1.57, + "learning_rate": 1.859106477879733e-05, + "loss": 0.3714, + "step": 3743 + }, + { + "epoch": 1.57, + "learning_rate": 1.859019695033339e-05, + "loss": 0.3634, + "step": 3744 + }, + { + "epoch": 1.57, + "learning_rate": 1.8589328874951866e-05, + "loss": 0.442, + "step": 3745 + }, + { + "epoch": 1.57, + "learning_rate": 1.8588460552677703e-05, + "loss": 0.426, + "step": 3746 + }, + { + "epoch": 1.57, + "learning_rate": 1.8587591983535868e-05, + "loss": 0.3693, + "step": 3747 + }, + { + "epoch": 1.57, + "learning_rate": 1.8586723167551315e-05, + "loss": 0.4352, + "step": 3748 + }, + { + "epoch": 1.57, + "learning_rate": 1.858585410474903e-05, + "loss": 0.3763, + "step": 3749 + }, + { + "epoch": 1.57, + "learning_rate": 1.8584984795153984e-05, + "loss": 0.4033, + "step": 3750 + }, + { + "epoch": 1.57, + "learning_rate": 1.858411523879117e-05, + "loss": 0.4202, + "step": 3751 + }, + { + "epoch": 1.57, + "learning_rate": 1.858324543568558e-05, + "loss": 0.4004, + "step": 3752 + }, + { + "epoch": 1.57, + "learning_rate": 1.8582375385862215e-05, + "loss": 0.4286, + "step": 3753 + }, + { + "epoch": 1.57, + "learning_rate": 1.858150508934609e-05, + "loss": 0.3823, + "step": 3754 + }, + { + "epoch": 1.57, + "learning_rate": 1.8580634546162212e-05, + "loss": 0.3807, + "step": 3755 + }, + { + "epoch": 1.57, + "learning_rate": 1.857976375633561e-05, + "loss": 0.3462, + "step": 3756 + }, + { + "epoch": 1.57, + "learning_rate": 1.8578892719891314e-05, + "loss": 0.379, + "step": 3757 + }, + { + "epoch": 1.57, + "learning_rate": 1.8578021436854357e-05, + "loss": 0.3501, + "step": 3758 + }, + { + "epoch": 1.57, + "learning_rate": 1.8577149907249786e-05, + "loss": 0.384, + "step": 3759 + }, + { + "epoch": 1.57, + "learning_rate": 1.8576278131102655e-05, + "loss": 0.3863, + "step": 3760 + }, + { + "epoch": 1.57, + "learning_rate": 1.8575406108438017e-05, + "loss": 0.3424, + "step": 3761 + }, + { + "epoch": 1.58, + "learning_rate": 1.857453383928094e-05, + "loss": 0.395, + "step": 3762 + }, + { + "epoch": 1.58, + "learning_rate": 1.8573661323656495e-05, + "loss": 0.3597, + "step": 3763 + }, + { + "epoch": 1.58, + "learning_rate": 1.8572788561589766e-05, + "loss": 0.391, + "step": 3764 + }, + { + "epoch": 1.58, + "learning_rate": 1.857191555310584e-05, + "loss": 0.3818, + "step": 3765 + }, + { + "epoch": 1.58, + "learning_rate": 1.85710422982298e-05, + "loss": 0.4242, + "step": 3766 + }, + { + "epoch": 1.58, + "learning_rate": 1.8570168796986764e-05, + "loss": 0.42, + "step": 3767 + }, + { + "epoch": 1.58, + "learning_rate": 1.8569295049401826e-05, + "loss": 0.3615, + "step": 3768 + }, + { + "epoch": 1.58, + "learning_rate": 1.8568421055500107e-05, + "loss": 0.3709, + "step": 3769 + }, + { + "epoch": 1.58, + "learning_rate": 1.8567546815306726e-05, + "loss": 0.3988, + "step": 3770 + }, + { + "epoch": 1.58, + "learning_rate": 1.8566672328846816e-05, + "loss": 0.3838, + "step": 3771 + }, + { + "epoch": 1.58, + "learning_rate": 1.8565797596145515e-05, + "loss": 0.3679, + "step": 3772 + }, + { + "epoch": 1.58, + "learning_rate": 1.8564922617227958e-05, + "loss": 0.3765, + "step": 3773 + }, + { + "epoch": 1.58, + "learning_rate": 1.8564047392119306e-05, + "loss": 0.4416, + "step": 3774 + }, + { + "epoch": 1.58, + "learning_rate": 1.8563171920844706e-05, + "loss": 0.3616, + "step": 3775 + }, + { + "epoch": 1.58, + "learning_rate": 1.856229620342933e-05, + "loss": 0.3814, + "step": 3776 + }, + { + "epoch": 1.58, + "learning_rate": 1.8561420239898348e-05, + "loss": 0.3511, + "step": 3777 + }, + { + "epoch": 1.58, + "learning_rate": 1.856054403027694e-05, + "loss": 0.3562, + "step": 3778 + }, + { + "epoch": 1.58, + "learning_rate": 1.855966757459029e-05, + "loss": 0.3642, + "step": 3779 + }, + { + "epoch": 1.58, + "learning_rate": 1.855879087286359e-05, + "loss": 0.4004, + "step": 3780 + }, + { + "epoch": 1.58, + "learning_rate": 1.8557913925122042e-05, + "loss": 0.3681, + "step": 3781 + }, + { + "epoch": 1.58, + "learning_rate": 1.8557036731390853e-05, + "loss": 0.3826, + "step": 3782 + }, + { + "epoch": 1.58, + "learning_rate": 1.8556159291695235e-05, + "loss": 0.3754, + "step": 3783 + }, + { + "epoch": 1.58, + "learning_rate": 1.855528160606041e-05, + "loss": 0.3771, + "step": 3784 + }, + { + "epoch": 1.58, + "learning_rate": 1.855440367451161e-05, + "loss": 0.4514, + "step": 3785 + }, + { + "epoch": 1.59, + "learning_rate": 1.8553525497074065e-05, + "loss": 0.5203, + "step": 3786 + }, + { + "epoch": 1.59, + "learning_rate": 1.855264707377302e-05, + "loss": 0.4171, + "step": 3787 + }, + { + "epoch": 1.59, + "learning_rate": 1.8551768404633724e-05, + "loss": 0.3869, + "step": 3788 + }, + { + "epoch": 1.59, + "learning_rate": 1.8550889489681435e-05, + "loss": 0.4159, + "step": 3789 + }, + { + "epoch": 1.59, + "learning_rate": 1.8550010328941417e-05, + "loss": 0.3714, + "step": 3790 + }, + { + "epoch": 1.59, + "learning_rate": 1.8549130922438938e-05, + "loss": 0.375, + "step": 3791 + }, + { + "epoch": 1.59, + "learning_rate": 1.8548251270199274e-05, + "loss": 0.4059, + "step": 3792 + }, + { + "epoch": 1.59, + "learning_rate": 1.854737137224772e-05, + "loss": 0.4218, + "step": 3793 + }, + { + "epoch": 1.59, + "learning_rate": 1.8546491228609552e-05, + "loss": 0.3925, + "step": 3794 + }, + { + "epoch": 1.59, + "learning_rate": 1.8545610839310085e-05, + "loss": 0.3834, + "step": 3795 + }, + { + "epoch": 1.59, + "learning_rate": 1.854473020437461e-05, + "loss": 0.3778, + "step": 3796 + }, + { + "epoch": 1.59, + "learning_rate": 1.8543849323828452e-05, + "loss": 0.427, + "step": 3797 + }, + { + "epoch": 1.59, + "learning_rate": 1.854296819769693e-05, + "loss": 0.3654, + "step": 3798 + }, + { + "epoch": 1.59, + "learning_rate": 1.8542086826005366e-05, + "loss": 0.3874, + "step": 3799 + }, + { + "epoch": 1.59, + "learning_rate": 1.8541205208779095e-05, + "loss": 0.432, + "step": 3800 + }, + { + "epoch": 1.59, + "learning_rate": 1.8540323346043458e-05, + "loss": 0.3539, + "step": 3801 + }, + { + "epoch": 1.59, + "learning_rate": 1.853944123782381e-05, + "loss": 0.421, + "step": 3802 + }, + { + "epoch": 1.59, + "learning_rate": 1.8538558884145497e-05, + "loss": 0.3837, + "step": 3803 + }, + { + "epoch": 1.59, + "learning_rate": 1.8537676285033886e-05, + "loss": 0.3663, + "step": 3804 + }, + { + "epoch": 1.59, + "learning_rate": 1.853679344051435e-05, + "loss": 0.3943, + "step": 3805 + }, + { + "epoch": 1.59, + "learning_rate": 1.8535910350612257e-05, + "loss": 0.3913, + "step": 3806 + }, + { + "epoch": 1.59, + "learning_rate": 1.8535027015353e-05, + "loss": 0.3979, + "step": 3807 + }, + { + "epoch": 1.59, + "learning_rate": 1.8534143434761965e-05, + "loss": 0.4269, + "step": 3808 + }, + { + "epoch": 1.59, + "learning_rate": 1.8533259608864547e-05, + "loss": 0.3765, + "step": 3809 + }, + { + "epoch": 1.6, + "learning_rate": 1.8532375537686155e-05, + "loss": 0.3696, + "step": 3810 + }, + { + "epoch": 1.6, + "learning_rate": 1.85314912212522e-05, + "loss": 0.4306, + "step": 3811 + }, + { + "epoch": 1.6, + "learning_rate": 1.85306066595881e-05, + "loss": 0.3987, + "step": 3812 + }, + { + "epoch": 1.6, + "learning_rate": 1.8529721852719283e-05, + "loss": 0.4022, + "step": 3813 + }, + { + "epoch": 1.6, + "learning_rate": 1.852883680067118e-05, + "loss": 0.4001, + "step": 3814 + }, + { + "epoch": 1.6, + "learning_rate": 1.852795150346923e-05, + "loss": 0.3886, + "step": 3815 + }, + { + "epoch": 1.6, + "learning_rate": 1.8527065961138885e-05, + "loss": 0.339, + "step": 3816 + }, + { + "epoch": 1.6, + "learning_rate": 1.852618017370559e-05, + "loss": 0.3473, + "step": 3817 + }, + { + "epoch": 1.6, + "learning_rate": 1.8525294141194816e-05, + "loss": 0.3534, + "step": 3818 + }, + { + "epoch": 1.6, + "learning_rate": 1.852440786363203e-05, + "loss": 0.3533, + "step": 3819 + }, + { + "epoch": 1.6, + "learning_rate": 1.85235213410427e-05, + "loss": 0.4324, + "step": 3820 + }, + { + "epoch": 1.6, + "learning_rate": 1.8522634573452314e-05, + "loss": 0.3952, + "step": 3821 + }, + { + "epoch": 1.6, + "learning_rate": 1.8521747560886357e-05, + "loss": 0.4197, + "step": 3822 + }, + { + "epoch": 1.6, + "learning_rate": 1.8520860303370333e-05, + "loss": 0.3743, + "step": 3823 + }, + { + "epoch": 1.6, + "learning_rate": 1.851997280092974e-05, + "loss": 0.3709, + "step": 3824 + }, + { + "epoch": 1.6, + "learning_rate": 1.8519085053590083e-05, + "loss": 0.4111, + "step": 3825 + }, + { + "epoch": 1.6, + "learning_rate": 1.8518197061376895e-05, + "loss": 0.4298, + "step": 3826 + }, + { + "epoch": 1.6, + "learning_rate": 1.8517308824315688e-05, + "loss": 0.3054, + "step": 3827 + }, + { + "epoch": 1.6, + "learning_rate": 1.8516420342431994e-05, + "loss": 0.3889, + "step": 3828 + }, + { + "epoch": 1.6, + "learning_rate": 1.8515531615751357e-05, + "loss": 0.387, + "step": 3829 + }, + { + "epoch": 1.6, + "learning_rate": 1.851464264429932e-05, + "loss": 0.3876, + "step": 3830 + }, + { + "epoch": 1.6, + "learning_rate": 1.8513753428101436e-05, + "loss": 0.3743, + "step": 3831 + }, + { + "epoch": 1.6, + "learning_rate": 1.8512863967183265e-05, + "loss": 0.4047, + "step": 3832 + }, + { + "epoch": 1.6, + "learning_rate": 1.8511974261570375e-05, + "loss": 0.3511, + "step": 3833 + }, + { + "epoch": 1.61, + "learning_rate": 1.8511084311288338e-05, + "loss": 0.3873, + "step": 3834 + }, + { + "epoch": 1.61, + "learning_rate": 1.851019411636273e-05, + "loss": 0.3827, + "step": 3835 + }, + { + "epoch": 1.61, + "learning_rate": 1.850930367681915e-05, + "loss": 0.4201, + "step": 3836 + }, + { + "epoch": 1.61, + "learning_rate": 1.8508412992683183e-05, + "loss": 0.3714, + "step": 3837 + }, + { + "epoch": 1.61, + "learning_rate": 1.8507522063980436e-05, + "loss": 0.4159, + "step": 3838 + }, + { + "epoch": 1.61, + "learning_rate": 1.8506630890736515e-05, + "loss": 0.4006, + "step": 3839 + }, + { + "epoch": 1.61, + "learning_rate": 1.8505739472977043e-05, + "loss": 0.3933, + "step": 3840 + }, + { + "epoch": 1.61, + "learning_rate": 1.8504847810727632e-05, + "loss": 0.3986, + "step": 3841 + }, + { + "epoch": 1.61, + "learning_rate": 1.8503955904013922e-05, + "loss": 0.3899, + "step": 3842 + }, + { + "epoch": 1.61, + "learning_rate": 1.8503063752861543e-05, + "loss": 0.4075, + "step": 3843 + }, + { + "epoch": 1.61, + "learning_rate": 1.8502171357296144e-05, + "loss": 0.3603, + "step": 3844 + }, + { + "epoch": 1.61, + "learning_rate": 1.8501278717343372e-05, + "loss": 0.4044, + "step": 3845 + }, + { + "epoch": 1.61, + "learning_rate": 1.850038583302889e-05, + "loss": 0.375, + "step": 3846 + }, + { + "epoch": 1.61, + "learning_rate": 1.8499492704378355e-05, + "loss": 0.3622, + "step": 3847 + }, + { + "epoch": 1.61, + "learning_rate": 1.849859933141745e-05, + "loss": 0.4129, + "step": 3848 + }, + { + "epoch": 1.61, + "learning_rate": 1.8497705714171846e-05, + "loss": 0.4007, + "step": 3849 + }, + { + "epoch": 1.61, + "learning_rate": 1.849681185266723e-05, + "loss": 0.3879, + "step": 3850 + }, + { + "epoch": 1.61, + "learning_rate": 1.8495917746929303e-05, + "loss": 0.4252, + "step": 3851 + }, + { + "epoch": 1.61, + "learning_rate": 1.8495023396983754e-05, + "loss": 0.3669, + "step": 3852 + }, + { + "epoch": 1.61, + "learning_rate": 1.84941288028563e-05, + "loss": 0.3351, + "step": 3853 + }, + { + "epoch": 1.61, + "learning_rate": 1.8493233964572648e-05, + "loss": 0.3886, + "step": 3854 + }, + { + "epoch": 1.61, + "learning_rate": 1.8492338882158526e-05, + "loss": 0.415, + "step": 3855 + }, + { + "epoch": 1.61, + "learning_rate": 1.8491443555639656e-05, + "loss": 0.4072, + "step": 3856 + }, + { + "epoch": 1.61, + "learning_rate": 1.8490547985041774e-05, + "loss": 0.349, + "step": 3857 + }, + { + "epoch": 1.62, + "learning_rate": 1.8489652170390627e-05, + "loss": 0.3835, + "step": 3858 + }, + { + "epoch": 1.62, + "learning_rate": 1.848875611171196e-05, + "loss": 0.4041, + "step": 3859 + }, + { + "epoch": 1.62, + "learning_rate": 1.8487859809031537e-05, + "loss": 0.4144, + "step": 3860 + }, + { + "epoch": 1.62, + "learning_rate": 1.8486963262375113e-05, + "loss": 0.4173, + "step": 3861 + }, + { + "epoch": 1.62, + "learning_rate": 1.8486066471768457e-05, + "loss": 0.329, + "step": 3862 + }, + { + "epoch": 1.62, + "learning_rate": 1.848516943723736e-05, + "loss": 0.4009, + "step": 3863 + }, + { + "epoch": 1.62, + "learning_rate": 1.8484272158807587e-05, + "loss": 0.3595, + "step": 3864 + }, + { + "epoch": 1.62, + "learning_rate": 1.8483374636504943e-05, + "loss": 0.4056, + "step": 3865 + }, + { + "epoch": 1.62, + "learning_rate": 1.8482476870355226e-05, + "loss": 0.3897, + "step": 3866 + }, + { + "epoch": 1.62, + "learning_rate": 1.848157886038424e-05, + "loss": 0.3094, + "step": 3867 + }, + { + "epoch": 1.62, + "learning_rate": 1.848068060661779e-05, + "loss": 0.3604, + "step": 3868 + }, + { + "epoch": 1.62, + "learning_rate": 1.8479782109081702e-05, + "loss": 0.3688, + "step": 3869 + }, + { + "epoch": 1.62, + "learning_rate": 1.8478883367801804e-05, + "loss": 0.3693, + "step": 3870 + }, + { + "epoch": 1.62, + "learning_rate": 1.8477984382803928e-05, + "loss": 0.4919, + "step": 3871 + }, + { + "epoch": 1.62, + "learning_rate": 1.8477085154113913e-05, + "loss": 0.3725, + "step": 3872 + }, + { + "epoch": 1.62, + "learning_rate": 1.847618568175761e-05, + "loss": 0.3605, + "step": 3873 + }, + { + "epoch": 1.62, + "learning_rate": 1.847528596576087e-05, + "loss": 0.4308, + "step": 3874 + }, + { + "epoch": 1.62, + "learning_rate": 1.8474386006149556e-05, + "loss": 0.403, + "step": 3875 + }, + { + "epoch": 1.62, + "learning_rate": 1.847348580294954e-05, + "loss": 0.4061, + "step": 3876 + }, + { + "epoch": 1.62, + "learning_rate": 1.8472585356186688e-05, + "loss": 0.4084, + "step": 3877 + }, + { + "epoch": 1.62, + "learning_rate": 1.847168466588689e-05, + "loss": 0.4078, + "step": 3878 + }, + { + "epoch": 1.62, + "learning_rate": 1.8470783732076034e-05, + "loss": 0.3814, + "step": 3879 + }, + { + "epoch": 1.62, + "learning_rate": 1.8469882554780016e-05, + "loss": 0.4037, + "step": 3880 + }, + { + "epoch": 1.62, + "learning_rate": 1.846898113402474e-05, + "loss": 0.3961, + "step": 3881 + }, + { + "epoch": 1.63, + "learning_rate": 1.8468079469836117e-05, + "loss": 0.3874, + "step": 3882 + }, + { + "epoch": 1.63, + "learning_rate": 1.8467177562240062e-05, + "loss": 0.3401, + "step": 3883 + }, + { + "epoch": 1.63, + "learning_rate": 1.8466275411262504e-05, + "loss": 0.3788, + "step": 3884 + }, + { + "epoch": 1.63, + "learning_rate": 1.8465373016929377e-05, + "loss": 0.3697, + "step": 3885 + }, + { + "epoch": 1.63, + "learning_rate": 1.8464470379266606e-05, + "loss": 0.3385, + "step": 3886 + }, + { + "epoch": 1.63, + "learning_rate": 1.846356749830015e-05, + "loss": 0.4286, + "step": 3887 + }, + { + "epoch": 1.63, + "learning_rate": 1.8462664374055953e-05, + "loss": 0.3876, + "step": 3888 + }, + { + "epoch": 1.63, + "learning_rate": 1.8461761006559982e-05, + "loss": 0.3762, + "step": 3889 + }, + { + "epoch": 1.63, + "learning_rate": 1.84608573958382e-05, + "loss": 0.3909, + "step": 3890 + }, + { + "epoch": 1.63, + "learning_rate": 1.8459953541916576e-05, + "loss": 0.3997, + "step": 3891 + }, + { + "epoch": 1.63, + "learning_rate": 1.84590494448211e-05, + "loss": 0.3785, + "step": 3892 + }, + { + "epoch": 1.63, + "learning_rate": 1.8458145104577752e-05, + "loss": 0.4019, + "step": 3893 + }, + { + "epoch": 1.63, + "learning_rate": 1.8457240521212525e-05, + "loss": 0.3651, + "step": 3894 + }, + { + "epoch": 1.63, + "learning_rate": 1.845633569475143e-05, + "loss": 0.3841, + "step": 3895 + }, + { + "epoch": 1.63, + "learning_rate": 1.8455430625220465e-05, + "loss": 0.4117, + "step": 3896 + }, + { + "epoch": 1.63, + "learning_rate": 1.8454525312645654e-05, + "loss": 0.4056, + "step": 3897 + }, + { + "epoch": 1.63, + "learning_rate": 1.845361975705301e-05, + "loss": 0.379, + "step": 3898 + }, + { + "epoch": 1.63, + "learning_rate": 1.845271395846857e-05, + "loss": 0.4022, + "step": 3899 + }, + { + "epoch": 1.63, + "learning_rate": 1.8451807916918367e-05, + "loss": 0.4109, + "step": 3900 + }, + { + "epoch": 1.63, + "learning_rate": 1.845090163242845e-05, + "loss": 0.3745, + "step": 3901 + }, + { + "epoch": 1.63, + "learning_rate": 1.844999510502486e-05, + "loss": 0.4296, + "step": 3902 + }, + { + "epoch": 1.63, + "learning_rate": 1.844908833473366e-05, + "loss": 0.4323, + "step": 3903 + }, + { + "epoch": 1.63, + "learning_rate": 1.8448181321580916e-05, + "loss": 0.392, + "step": 3904 + }, + { + "epoch": 1.63, + "learning_rate": 1.8447274065592694e-05, + "loss": 0.4409, + "step": 3905 + }, + { + "epoch": 1.64, + "learning_rate": 1.8446366566795075e-05, + "loss": 0.4219, + "step": 3906 + }, + { + "epoch": 1.64, + "learning_rate": 1.8445458825214148e-05, + "loss": 0.4145, + "step": 3907 + }, + { + "epoch": 1.64, + "learning_rate": 1.8444550840875996e-05, + "loss": 0.3887, + "step": 3908 + }, + { + "epoch": 1.64, + "learning_rate": 1.8443642613806727e-05, + "loss": 0.3598, + "step": 3909 + }, + { + "epoch": 1.64, + "learning_rate": 1.8442734144032442e-05, + "loss": 0.3373, + "step": 3910 + }, + { + "epoch": 1.64, + "learning_rate": 1.844182543157926e-05, + "loss": 0.4054, + "step": 3911 + }, + { + "epoch": 1.64, + "learning_rate": 1.844091647647329e-05, + "loss": 0.3754, + "step": 3912 + }, + { + "epoch": 1.64, + "learning_rate": 1.8440007278740673e-05, + "loss": 0.4236, + "step": 3913 + }, + { + "epoch": 1.64, + "learning_rate": 1.8439097838407532e-05, + "loss": 0.3872, + "step": 3914 + }, + { + "epoch": 1.64, + "learning_rate": 1.8438188155500016e-05, + "loss": 0.3718, + "step": 3915 + }, + { + "epoch": 1.64, + "learning_rate": 1.8437278230044266e-05, + "loss": 0.4069, + "step": 3916 + }, + { + "epoch": 1.64, + "learning_rate": 1.8436368062066442e-05, + "loss": 0.4603, + "step": 3917 + }, + { + "epoch": 1.64, + "learning_rate": 1.8435457651592704e-05, + "loss": 0.382, + "step": 3918 + }, + { + "epoch": 1.64, + "learning_rate": 1.843454699864922e-05, + "loss": 0.3786, + "step": 3919 + }, + { + "epoch": 1.64, + "learning_rate": 1.843363610326217e-05, + "loss": 0.3877, + "step": 3920 + }, + { + "epoch": 1.64, + "learning_rate": 1.843272496545773e-05, + "loss": 0.3959, + "step": 3921 + }, + { + "epoch": 1.64, + "learning_rate": 1.8431813585262098e-05, + "loss": 0.3778, + "step": 3922 + }, + { + "epoch": 1.64, + "learning_rate": 1.8430901962701465e-05, + "loss": 0.4074, + "step": 3923 + }, + { + "epoch": 1.64, + "learning_rate": 1.8429990097802037e-05, + "loss": 0.403, + "step": 3924 + }, + { + "epoch": 1.64, + "learning_rate": 1.8429077990590026e-05, + "loss": 0.4485, + "step": 3925 + }, + { + "epoch": 1.64, + "learning_rate": 1.8428165641091643e-05, + "loss": 0.3935, + "step": 3926 + }, + { + "epoch": 1.64, + "learning_rate": 1.8427253049333124e-05, + "loss": 0.3865, + "step": 3927 + }, + { + "epoch": 1.64, + "learning_rate": 1.8426340215340688e-05, + "loss": 0.3605, + "step": 3928 + }, + { + "epoch": 1.64, + "learning_rate": 1.8425427139140586e-05, + "loss": 0.38, + "step": 3929 + }, + { + "epoch": 1.65, + "learning_rate": 1.8424513820759054e-05, + "loss": 0.4266, + "step": 3930 + }, + { + "epoch": 1.65, + "learning_rate": 1.8423600260222346e-05, + "loss": 0.4254, + "step": 3931 + }, + { + "epoch": 1.65, + "learning_rate": 1.8422686457556726e-05, + "loss": 0.398, + "step": 3932 + }, + { + "epoch": 1.65, + "learning_rate": 1.842177241278846e-05, + "loss": 0.3574, + "step": 3933 + }, + { + "epoch": 1.65, + "learning_rate": 1.8420858125943818e-05, + "loss": 0.3633, + "step": 3934 + }, + { + "epoch": 1.65, + "learning_rate": 1.841994359704908e-05, + "loss": 0.4486, + "step": 3935 + }, + { + "epoch": 1.65, + "learning_rate": 1.8419028826130537e-05, + "loss": 0.3329, + "step": 3936 + }, + { + "epoch": 1.65, + "learning_rate": 1.8418113813214477e-05, + "loss": 0.5135, + "step": 3937 + }, + { + "epoch": 1.65, + "learning_rate": 1.841719855832721e-05, + "loss": 0.4119, + "step": 3938 + }, + { + "epoch": 1.65, + "learning_rate": 1.8416283061495038e-05, + "loss": 0.3705, + "step": 3939 + }, + { + "epoch": 1.65, + "learning_rate": 1.841536732274428e-05, + "loss": 0.3832, + "step": 3940 + }, + { + "epoch": 1.65, + "learning_rate": 1.8414451342101253e-05, + "loss": 0.3343, + "step": 3941 + }, + { + "epoch": 1.65, + "learning_rate": 1.8413535119592292e-05, + "loss": 0.364, + "step": 3942 + }, + { + "epoch": 1.65, + "learning_rate": 1.841261865524373e-05, + "loss": 0.3662, + "step": 3943 + }, + { + "epoch": 1.65, + "learning_rate": 1.8411701949081906e-05, + "loss": 0.4828, + "step": 3944 + }, + { + "epoch": 1.65, + "learning_rate": 1.8410785001133177e-05, + "loss": 0.3881, + "step": 3945 + }, + { + "epoch": 1.65, + "learning_rate": 1.8409867811423897e-05, + "loss": 0.4021, + "step": 3946 + }, + { + "epoch": 1.65, + "learning_rate": 1.840895037998043e-05, + "loss": 0.4208, + "step": 3947 + }, + { + "epoch": 1.65, + "learning_rate": 1.840803270682914e-05, + "loss": 0.3583, + "step": 3948 + }, + { + "epoch": 1.65, + "learning_rate": 1.8407114791996416e-05, + "loss": 0.3859, + "step": 3949 + }, + { + "epoch": 1.65, + "learning_rate": 1.840619663550864e-05, + "loss": 0.373, + "step": 3950 + }, + { + "epoch": 1.65, + "learning_rate": 1.84052782373922e-05, + "loss": 0.3767, + "step": 3951 + }, + { + "epoch": 1.65, + "learning_rate": 1.8404359597673492e-05, + "loss": 0.3835, + "step": 3952 + }, + { + "epoch": 1.66, + "learning_rate": 1.840344071637893e-05, + "loss": 0.3904, + "step": 3953 + }, + { + "epoch": 1.66, + "learning_rate": 1.840252159353492e-05, + "loss": 0.3753, + "step": 3954 + }, + { + "epoch": 1.66, + "learning_rate": 1.840160222916788e-05, + "loss": 0.3811, + "step": 3955 + }, + { + "epoch": 1.66, + "learning_rate": 1.8400682623304243e-05, + "loss": 0.362, + "step": 3956 + }, + { + "epoch": 1.66, + "learning_rate": 1.839976277597044e-05, + "loss": 0.4012, + "step": 3957 + }, + { + "epoch": 1.66, + "learning_rate": 1.8398842687192906e-05, + "loss": 0.408, + "step": 3958 + }, + { + "epoch": 1.66, + "learning_rate": 1.839792235699809e-05, + "loss": 0.3913, + "step": 3959 + }, + { + "epoch": 1.66, + "learning_rate": 1.8397001785412455e-05, + "loss": 0.3546, + "step": 3960 + }, + { + "epoch": 1.66, + "learning_rate": 1.839608097246245e-05, + "loss": 0.3846, + "step": 3961 + }, + { + "epoch": 1.66, + "learning_rate": 1.839515991817455e-05, + "loss": 0.3486, + "step": 3962 + }, + { + "epoch": 1.66, + "learning_rate": 1.8394238622575228e-05, + "loss": 0.4618, + "step": 3963 + }, + { + "epoch": 1.66, + "learning_rate": 1.8393317085690964e-05, + "loss": 0.3483, + "step": 3964 + }, + { + "epoch": 1.66, + "learning_rate": 1.839239530754825e-05, + "loss": 0.4279, + "step": 3965 + }, + { + "epoch": 1.66, + "learning_rate": 1.839147328817358e-05, + "loss": 0.4616, + "step": 3966 + }, + { + "epoch": 1.66, + "learning_rate": 1.8390551027593456e-05, + "loss": 0.4158, + "step": 3967 + }, + { + "epoch": 1.66, + "learning_rate": 1.8389628525834385e-05, + "loss": 0.3667, + "step": 3968 + }, + { + "epoch": 1.66, + "learning_rate": 1.8388705782922893e-05, + "loss": 0.4405, + "step": 3969 + }, + { + "epoch": 1.66, + "learning_rate": 1.8387782798885495e-05, + "loss": 0.4028, + "step": 3970 + }, + { + "epoch": 1.66, + "learning_rate": 1.8386859573748724e-05, + "loss": 0.3949, + "step": 3971 + }, + { + "epoch": 1.66, + "learning_rate": 1.8385936107539115e-05, + "loss": 0.3959, + "step": 3972 + }, + { + "epoch": 1.66, + "learning_rate": 1.8385012400283218e-05, + "loss": 0.4073, + "step": 3973 + }, + { + "epoch": 1.66, + "learning_rate": 1.838408845200758e-05, + "loss": 0.3725, + "step": 3974 + }, + { + "epoch": 1.66, + "learning_rate": 1.8383164262738756e-05, + "loss": 0.3761, + "step": 3975 + }, + { + "epoch": 1.66, + "learning_rate": 1.838223983250331e-05, + "loss": 0.3953, + "step": 3976 + }, + { + "epoch": 1.67, + "learning_rate": 1.8381315161327827e-05, + "loss": 0.3925, + "step": 3977 + }, + { + "epoch": 1.67, + "learning_rate": 1.8380390249238876e-05, + "loss": 0.3745, + "step": 3978 + }, + { + "epoch": 1.67, + "learning_rate": 1.837946509626304e-05, + "loss": 0.3934, + "step": 3979 + }, + { + "epoch": 1.67, + "learning_rate": 1.8378539702426917e-05, + "loss": 0.348, + "step": 3980 + }, + { + "epoch": 1.67, + "learning_rate": 1.8377614067757106e-05, + "loss": 0.3714, + "step": 3981 + }, + { + "epoch": 1.67, + "learning_rate": 1.8376688192280213e-05, + "loss": 0.384, + "step": 3982 + }, + { + "epoch": 1.67, + "learning_rate": 1.837576207602285e-05, + "loss": 0.3941, + "step": 3983 + }, + { + "epoch": 1.67, + "learning_rate": 1.837483571901164e-05, + "loss": 0.3607, + "step": 3984 + }, + { + "epoch": 1.67, + "learning_rate": 1.837390912127321e-05, + "loss": 0.3494, + "step": 3985 + }, + { + "epoch": 1.67, + "learning_rate": 1.837298228283419e-05, + "loss": 0.3867, + "step": 3986 + }, + { + "epoch": 1.67, + "learning_rate": 1.837205520372123e-05, + "loss": 0.369, + "step": 3987 + }, + { + "epoch": 1.67, + "learning_rate": 1.8371127883960963e-05, + "loss": 0.3949, + "step": 3988 + }, + { + "epoch": 1.67, + "learning_rate": 1.837020032358006e-05, + "loss": 0.3878, + "step": 3989 + }, + { + "epoch": 1.67, + "learning_rate": 1.8369272522605177e-05, + "loss": 0.4191, + "step": 3990 + }, + { + "epoch": 1.67, + "learning_rate": 1.836834448106298e-05, + "loss": 0.3945, + "step": 3991 + }, + { + "epoch": 1.67, + "learning_rate": 1.836741619898015e-05, + "loss": 0.4298, + "step": 3992 + }, + { + "epoch": 1.67, + "learning_rate": 1.836648767638336e-05, + "loss": 0.3937, + "step": 3993 + }, + { + "epoch": 1.67, + "learning_rate": 1.8365558913299307e-05, + "loss": 0.4084, + "step": 3994 + }, + { + "epoch": 1.67, + "learning_rate": 1.836462990975469e-05, + "loss": 0.3711, + "step": 3995 + }, + { + "epoch": 1.67, + "learning_rate": 1.836370066577621e-05, + "loss": 0.3859, + "step": 3996 + }, + { + "epoch": 1.67, + "learning_rate": 1.836277118139057e-05, + "loss": 0.3787, + "step": 3997 + }, + { + "epoch": 1.67, + "learning_rate": 1.83618414566245e-05, + "loss": 0.387, + "step": 3998 + }, + { + "epoch": 1.67, + "learning_rate": 1.8360911491504714e-05, + "loss": 0.3551, + "step": 3999 + }, + { + "epoch": 1.67, + "learning_rate": 1.835998128605795e-05, + "loss": 0.409, + "step": 4000 + }, + { + "epoch": 1.68, + "learning_rate": 1.8359050840310936e-05, + "loss": 0.3925, + "step": 4001 + }, + { + "epoch": 1.68, + "learning_rate": 1.8358120154290427e-05, + "loss": 0.4395, + "step": 4002 + }, + { + "epoch": 1.68, + "learning_rate": 1.8357189228023174e-05, + "loss": 0.425, + "step": 4003 + }, + { + "epoch": 1.68, + "learning_rate": 1.835625806153593e-05, + "loss": 0.3778, + "step": 4004 + }, + { + "epoch": 1.68, + "learning_rate": 1.835532665485546e-05, + "loss": 0.3671, + "step": 4005 + }, + { + "epoch": 1.68, + "learning_rate": 1.8354395008008545e-05, + "loss": 0.4004, + "step": 4006 + }, + { + "epoch": 1.68, + "learning_rate": 1.835346312102196e-05, + "loss": 0.3547, + "step": 4007 + }, + { + "epoch": 1.68, + "learning_rate": 1.8352530993922483e-05, + "loss": 0.4045, + "step": 4008 + }, + { + "epoch": 1.68, + "learning_rate": 1.835159862673692e-05, + "loss": 0.3707, + "step": 4009 + }, + { + "epoch": 1.68, + "learning_rate": 1.8350666019492065e-05, + "loss": 0.3821, + "step": 4010 + }, + { + "epoch": 1.68, + "learning_rate": 1.8349733172214722e-05, + "loss": 0.4071, + "step": 4011 + }, + { + "epoch": 1.68, + "learning_rate": 1.8348800084931712e-05, + "loss": 0.4002, + "step": 4012 + }, + { + "epoch": 1.68, + "learning_rate": 1.834786675766985e-05, + "loss": 0.3747, + "step": 4013 + }, + { + "epoch": 1.68, + "learning_rate": 1.834693319045597e-05, + "loss": 0.4455, + "step": 4014 + }, + { + "epoch": 1.68, + "learning_rate": 1.8345999383316897e-05, + "loss": 0.3962, + "step": 4015 + }, + { + "epoch": 1.68, + "learning_rate": 1.8345065336279482e-05, + "loss": 0.3184, + "step": 4016 + }, + { + "epoch": 1.68, + "learning_rate": 1.8344131049370566e-05, + "loss": 0.3856, + "step": 4017 + }, + { + "epoch": 1.68, + "learning_rate": 1.834319652261701e-05, + "loss": 0.4466, + "step": 4018 + }, + { + "epoch": 1.68, + "learning_rate": 1.8342261756045673e-05, + "loss": 0.4317, + "step": 4019 + }, + { + "epoch": 1.68, + "learning_rate": 1.834132674968342e-05, + "loss": 0.3718, + "step": 4020 + }, + { + "epoch": 1.68, + "learning_rate": 1.834039150355714e-05, + "loss": 0.3773, + "step": 4021 + }, + { + "epoch": 1.68, + "learning_rate": 1.83394560176937e-05, + "loss": 0.3929, + "step": 4022 + }, + { + "epoch": 1.68, + "learning_rate": 1.8338520292120002e-05, + "loss": 0.4016, + "step": 4023 + }, + { + "epoch": 1.68, + "learning_rate": 1.8337584326862935e-05, + "loss": 0.4165, + "step": 4024 + }, + { + "epoch": 1.69, + "learning_rate": 1.8336648121949405e-05, + "loss": 0.3676, + "step": 4025 + }, + { + "epoch": 1.69, + "learning_rate": 1.8335711677406324e-05, + "loss": 0.3762, + "step": 4026 + }, + { + "epoch": 1.69, + "learning_rate": 1.833477499326061e-05, + "loss": 0.4556, + "step": 4027 + }, + { + "epoch": 1.69, + "learning_rate": 1.833383806953918e-05, + "loss": 0.3767, + "step": 4028 + }, + { + "epoch": 1.69, + "learning_rate": 1.8332900906268972e-05, + "loss": 0.3779, + "step": 4029 + }, + { + "epoch": 1.69, + "learning_rate": 1.8331963503476925e-05, + "loss": 0.3795, + "step": 4030 + }, + { + "epoch": 1.69, + "learning_rate": 1.8331025861189974e-05, + "loss": 0.4227, + "step": 4031 + }, + { + "epoch": 1.69, + "learning_rate": 1.8330087979435083e-05, + "loss": 0.3875, + "step": 4032 + }, + { + "epoch": 1.69, + "learning_rate": 1.8329149858239204e-05, + "loss": 0.4157, + "step": 4033 + }, + { + "epoch": 1.69, + "learning_rate": 1.8328211497629304e-05, + "loss": 0.4011, + "step": 4034 + }, + { + "epoch": 1.69, + "learning_rate": 1.832727289763235e-05, + "loss": 0.3993, + "step": 4035 + }, + { + "epoch": 1.69, + "learning_rate": 1.832633405827533e-05, + "loss": 0.3993, + "step": 4036 + }, + { + "epoch": 1.69, + "learning_rate": 1.8325394979585227e-05, + "loss": 0.4001, + "step": 4037 + }, + { + "epoch": 1.69, + "learning_rate": 1.8324455661589033e-05, + "loss": 0.3964, + "step": 4038 + }, + { + "epoch": 1.69, + "learning_rate": 1.8323516104313746e-05, + "loss": 0.424, + "step": 4039 + }, + { + "epoch": 1.69, + "learning_rate": 1.8322576307786376e-05, + "loss": 0.4006, + "step": 4040 + }, + { + "epoch": 1.69, + "learning_rate": 1.8321636272033934e-05, + "loss": 0.3782, + "step": 4041 + }, + { + "epoch": 1.69, + "learning_rate": 1.8320695997083443e-05, + "loss": 0.3963, + "step": 4042 + }, + { + "epoch": 1.69, + "learning_rate": 1.831975548296193e-05, + "loss": 0.3684, + "step": 4043 + }, + { + "epoch": 1.69, + "learning_rate": 1.8318814729696427e-05, + "loss": 0.4097, + "step": 4044 + }, + { + "epoch": 1.69, + "learning_rate": 1.8317873737313977e-05, + "loss": 0.3876, + "step": 4045 + }, + { + "epoch": 1.69, + "learning_rate": 1.831693250584163e-05, + "loss": 0.4022, + "step": 4046 + }, + { + "epoch": 1.69, + "learning_rate": 1.8315991035306435e-05, + "loss": 0.3946, + "step": 4047 + }, + { + "epoch": 1.69, + "learning_rate": 1.831504932573546e-05, + "loss": 0.4278, + "step": 4048 + }, + { + "epoch": 1.7, + "learning_rate": 1.831410737715577e-05, + "loss": 0.391, + "step": 4049 + }, + { + "epoch": 1.7, + "learning_rate": 1.8313165189594438e-05, + "loss": 0.3522, + "step": 4050 + }, + { + "epoch": 1.7, + "learning_rate": 1.8312222763078552e-05, + "loss": 0.3493, + "step": 4051 + }, + { + "epoch": 1.7, + "learning_rate": 1.83112800976352e-05, + "loss": 0.3808, + "step": 4052 + }, + { + "epoch": 1.7, + "learning_rate": 1.8310337193291473e-05, + "loss": 0.376, + "step": 4053 + }, + { + "epoch": 1.7, + "learning_rate": 1.8309394050074482e-05, + "loss": 0.3485, + "step": 4054 + }, + { + "epoch": 1.7, + "learning_rate": 1.8308450668011334e-05, + "loss": 0.3439, + "step": 4055 + }, + { + "epoch": 1.7, + "learning_rate": 1.830750704712914e-05, + "loss": 0.4055, + "step": 4056 + }, + { + "epoch": 1.7, + "learning_rate": 1.830656318745503e-05, + "loss": 0.4463, + "step": 4057 + }, + { + "epoch": 1.7, + "learning_rate": 1.830561908901613e-05, + "loss": 0.3713, + "step": 4058 + }, + { + "epoch": 1.7, + "learning_rate": 1.8304674751839583e-05, + "loss": 0.3569, + "step": 4059 + }, + { + "epoch": 1.7, + "learning_rate": 1.830373017595253e-05, + "loss": 0.3846, + "step": 4060 + }, + { + "epoch": 1.7, + "learning_rate": 1.8302785361382117e-05, + "loss": 0.4214, + "step": 4061 + }, + { + "epoch": 1.7, + "learning_rate": 1.8301840308155507e-05, + "loss": 0.3752, + "step": 4062 + }, + { + "epoch": 1.7, + "learning_rate": 1.8300895016299868e-05, + "loss": 0.4175, + "step": 4063 + }, + { + "epoch": 1.7, + "learning_rate": 1.8299949485842366e-05, + "loss": 0.4112, + "step": 4064 + }, + { + "epoch": 1.7, + "learning_rate": 1.829900371681018e-05, + "loss": 0.436, + "step": 4065 + }, + { + "epoch": 1.7, + "learning_rate": 1.8298057709230498e-05, + "loss": 0.4509, + "step": 4066 + }, + { + "epoch": 1.7, + "learning_rate": 1.8297111463130512e-05, + "loss": 0.4404, + "step": 4067 + }, + { + "epoch": 1.7, + "learning_rate": 1.8296164978537422e-05, + "loss": 0.3855, + "step": 4068 + }, + { + "epoch": 1.7, + "learning_rate": 1.8295218255478425e-05, + "loss": 0.3585, + "step": 4069 + }, + { + "epoch": 1.7, + "learning_rate": 1.8294271293980744e-05, + "loss": 0.4391, + "step": 4070 + }, + { + "epoch": 1.7, + "learning_rate": 1.8293324094071593e-05, + "loss": 0.34, + "step": 4071 + }, + { + "epoch": 1.7, + "learning_rate": 1.8292376655778202e-05, + "loss": 0.3956, + "step": 4072 + }, + { + "epoch": 1.71, + "learning_rate": 1.8291428979127802e-05, + "loss": 0.3659, + "step": 4073 + }, + { + "epoch": 1.71, + "learning_rate": 1.8290481064147637e-05, + "loss": 0.3756, + "step": 4074 + }, + { + "epoch": 1.71, + "learning_rate": 1.8289532910864943e-05, + "loss": 0.357, + "step": 4075 + }, + { + "epoch": 1.71, + "learning_rate": 1.8288584519306985e-05, + "loss": 0.3498, + "step": 4076 + }, + { + "epoch": 1.71, + "learning_rate": 1.8287635889501024e-05, + "loss": 0.3925, + "step": 4077 + }, + { + "epoch": 1.71, + "learning_rate": 1.828668702147432e-05, + "loss": 0.3581, + "step": 4078 + }, + { + "epoch": 1.71, + "learning_rate": 1.828573791525415e-05, + "loss": 0.372, + "step": 4079 + }, + { + "epoch": 1.71, + "learning_rate": 1.82847885708678e-05, + "loss": 0.388, + "step": 4080 + }, + { + "epoch": 1.71, + "learning_rate": 1.8283838988342553e-05, + "loss": 0.3954, + "step": 4081 + }, + { + "epoch": 1.71, + "learning_rate": 1.8282889167705704e-05, + "loss": 0.3438, + "step": 4082 + }, + { + "epoch": 1.71, + "learning_rate": 1.8281939108984557e-05, + "loss": 0.3651, + "step": 4083 + }, + { + "epoch": 1.71, + "learning_rate": 1.828098881220642e-05, + "loss": 0.3769, + "step": 4084 + }, + { + "epoch": 1.71, + "learning_rate": 1.8280038277398606e-05, + "loss": 0.39, + "step": 4085 + }, + { + "epoch": 1.71, + "learning_rate": 1.827908750458844e-05, + "loss": 0.3994, + "step": 4086 + }, + { + "epoch": 1.71, + "learning_rate": 1.8278136493803253e-05, + "loss": 0.3713, + "step": 4087 + }, + { + "epoch": 1.71, + "learning_rate": 1.8277185245070373e-05, + "loss": 0.4595, + "step": 4088 + }, + { + "epoch": 1.71, + "learning_rate": 1.8276233758417153e-05, + "loss": 0.4077, + "step": 4089 + }, + { + "epoch": 1.71, + "learning_rate": 1.827528203387094e-05, + "loss": 0.4482, + "step": 4090 + }, + { + "epoch": 1.71, + "learning_rate": 1.8274330071459082e-05, + "loss": 0.3704, + "step": 4091 + }, + { + "epoch": 1.71, + "learning_rate": 1.8273377871208952e-05, + "loss": 0.392, + "step": 4092 + }, + { + "epoch": 1.71, + "learning_rate": 1.8272425433147916e-05, + "loss": 0.3883, + "step": 4093 + }, + { + "epoch": 1.71, + "learning_rate": 1.8271472757303352e-05, + "loss": 0.3909, + "step": 4094 + }, + { + "epoch": 1.71, + "learning_rate": 1.827051984370264e-05, + "loss": 0.4504, + "step": 4095 + }, + { + "epoch": 1.71, + "learning_rate": 1.826956669237318e-05, + "loss": 0.3842, + "step": 4096 + }, + { + "epoch": 1.72, + "learning_rate": 1.8268613303342362e-05, + "loss": 0.3769, + "step": 4097 + }, + { + "epoch": 1.72, + "learning_rate": 1.826765967663759e-05, + "loss": 0.3996, + "step": 4098 + }, + { + "epoch": 1.72, + "learning_rate": 1.8266705812286276e-05, + "loss": 0.3751, + "step": 4099 + }, + { + "epoch": 1.72, + "learning_rate": 1.8265751710315843e-05, + "loss": 0.3925, + "step": 4100 + }, + { + "epoch": 1.72, + "learning_rate": 1.826479737075371e-05, + "loss": 0.3749, + "step": 4101 + }, + { + "epoch": 1.72, + "learning_rate": 1.8263842793627314e-05, + "loss": 0.3599, + "step": 4102 + }, + { + "epoch": 1.72, + "learning_rate": 1.8262887978964087e-05, + "loss": 0.4009, + "step": 4103 + }, + { + "epoch": 1.72, + "learning_rate": 1.8261932926791477e-05, + "loss": 0.3794, + "step": 4104 + }, + { + "epoch": 1.72, + "learning_rate": 1.8260977637136937e-05, + "loss": 0.3634, + "step": 4105 + }, + { + "epoch": 1.72, + "learning_rate": 1.8260022110027925e-05, + "loss": 0.3815, + "step": 4106 + }, + { + "epoch": 1.72, + "learning_rate": 1.8259066345491908e-05, + "loss": 0.447, + "step": 4107 + }, + { + "epoch": 1.72, + "learning_rate": 1.8258110343556358e-05, + "loss": 0.4141, + "step": 4108 + }, + { + "epoch": 1.72, + "learning_rate": 1.8257154104248752e-05, + "loss": 0.407, + "step": 4109 + }, + { + "epoch": 1.72, + "learning_rate": 1.8256197627596582e-05, + "loss": 0.395, + "step": 4110 + }, + { + "epoch": 1.72, + "learning_rate": 1.8255240913627335e-05, + "loss": 0.4496, + "step": 4111 + }, + { + "epoch": 1.72, + "learning_rate": 1.8254283962368517e-05, + "loss": 0.366, + "step": 4112 + }, + { + "epoch": 1.72, + "learning_rate": 1.8253326773847628e-05, + "loss": 0.3714, + "step": 4113 + }, + { + "epoch": 1.72, + "learning_rate": 1.8252369348092184e-05, + "loss": 0.3777, + "step": 4114 + }, + { + "epoch": 1.72, + "learning_rate": 1.825141168512971e-05, + "loss": 0.4085, + "step": 4115 + }, + { + "epoch": 1.72, + "learning_rate": 1.825045378498773e-05, + "loss": 0.4169, + "step": 4116 + }, + { + "epoch": 1.72, + "learning_rate": 1.824949564769377e-05, + "loss": 0.4316, + "step": 4117 + }, + { + "epoch": 1.72, + "learning_rate": 1.8248537273275385e-05, + "loss": 0.3995, + "step": 4118 + }, + { + "epoch": 1.72, + "learning_rate": 1.8247578661760114e-05, + "loss": 0.4057, + "step": 4119 + }, + { + "epoch": 1.72, + "learning_rate": 1.824661981317551e-05, + "loss": 0.389, + "step": 4120 + }, + { + "epoch": 1.73, + "learning_rate": 1.8245660727549142e-05, + "loss": 0.4068, + "step": 4121 + }, + { + "epoch": 1.73, + "learning_rate": 1.824470140490857e-05, + "loss": 0.4061, + "step": 4122 + }, + { + "epoch": 1.73, + "learning_rate": 1.8243741845281373e-05, + "loss": 0.4385, + "step": 4123 + }, + { + "epoch": 1.73, + "learning_rate": 1.824278204869513e-05, + "loss": 0.4081, + "step": 4124 + }, + { + "epoch": 1.73, + "learning_rate": 1.8241822015177433e-05, + "loss": 0.4129, + "step": 4125 + }, + { + "epoch": 1.73, + "learning_rate": 1.824086174475588e-05, + "loss": 0.3608, + "step": 4126 + }, + { + "epoch": 1.73, + "learning_rate": 1.8239901237458064e-05, + "loss": 0.4721, + "step": 4127 + }, + { + "epoch": 1.73, + "learning_rate": 1.8238940493311604e-05, + "loss": 0.3642, + "step": 4128 + }, + { + "epoch": 1.73, + "learning_rate": 1.8237979512344107e-05, + "loss": 0.3647, + "step": 4129 + }, + { + "epoch": 1.73, + "learning_rate": 1.8237018294583196e-05, + "loss": 0.4653, + "step": 4130 + }, + { + "epoch": 1.73, + "learning_rate": 1.8236056840056512e-05, + "loss": 0.4066, + "step": 4131 + }, + { + "epoch": 1.73, + "learning_rate": 1.8235095148791675e-05, + "loss": 0.3798, + "step": 4132 + }, + { + "epoch": 1.73, + "learning_rate": 1.823413322081634e-05, + "loss": 0.388, + "step": 4133 + }, + { + "epoch": 1.73, + "learning_rate": 1.823317105615815e-05, + "loss": 0.4043, + "step": 4134 + }, + { + "epoch": 1.73, + "learning_rate": 1.8232208654844766e-05, + "loss": 0.3316, + "step": 4135 + }, + { + "epoch": 1.73, + "learning_rate": 1.823124601690385e-05, + "loss": 0.3562, + "step": 4136 + }, + { + "epoch": 1.73, + "learning_rate": 1.8230283142363068e-05, + "loss": 0.4149, + "step": 4137 + }, + { + "epoch": 1.73, + "learning_rate": 1.8229320031250098e-05, + "loss": 0.3378, + "step": 4138 + }, + { + "epoch": 1.73, + "learning_rate": 1.822835668359263e-05, + "loss": 0.4156, + "step": 4139 + }, + { + "epoch": 1.73, + "learning_rate": 1.8227393099418353e-05, + "loss": 0.3673, + "step": 4140 + }, + { + "epoch": 1.73, + "learning_rate": 1.8226429278754963e-05, + "loss": 0.4593, + "step": 4141 + }, + { + "epoch": 1.73, + "learning_rate": 1.822546522163016e-05, + "loss": 0.3879, + "step": 4142 + }, + { + "epoch": 1.73, + "learning_rate": 1.822450092807166e-05, + "loss": 0.3846, + "step": 4143 + }, + { + "epoch": 1.73, + "learning_rate": 1.8223536398107177e-05, + "loss": 0.3752, + "step": 4144 + }, + { + "epoch": 1.74, + "learning_rate": 1.822257163176444e-05, + "loss": 0.4094, + "step": 4145 + }, + { + "epoch": 1.74, + "learning_rate": 1.8221606629071175e-05, + "loss": 0.3852, + "step": 4146 + }, + { + "epoch": 1.74, + "learning_rate": 1.8220641390055127e-05, + "loss": 0.3741, + "step": 4147 + }, + { + "epoch": 1.74, + "learning_rate": 1.8219675914744035e-05, + "loss": 0.4307, + "step": 4148 + }, + { + "epoch": 1.74, + "learning_rate": 1.8218710203165653e-05, + "loss": 0.3892, + "step": 4149 + }, + { + "epoch": 1.74, + "learning_rate": 1.821774425534774e-05, + "loss": 0.3891, + "step": 4150 + }, + { + "epoch": 1.74, + "learning_rate": 1.821677807131806e-05, + "loss": 0.389, + "step": 4151 + }, + { + "epoch": 1.74, + "learning_rate": 1.8215811651104383e-05, + "loss": 0.3828, + "step": 4152 + }, + { + "epoch": 1.74, + "learning_rate": 1.8214844994734498e-05, + "loss": 0.3996, + "step": 4153 + }, + { + "epoch": 1.74, + "learning_rate": 1.8213878102236177e-05, + "loss": 0.4122, + "step": 4154 + }, + { + "epoch": 1.74, + "learning_rate": 1.821291097363722e-05, + "loss": 0.3949, + "step": 4155 + }, + { + "epoch": 1.74, + "learning_rate": 1.821194360896543e-05, + "loss": 0.3584, + "step": 4156 + }, + { + "epoch": 1.74, + "learning_rate": 1.8210976008248603e-05, + "loss": 0.3853, + "step": 4157 + }, + { + "epoch": 1.74, + "learning_rate": 1.821000817151456e-05, + "loss": 0.4165, + "step": 4158 + }, + { + "epoch": 1.74, + "learning_rate": 1.820904009879111e-05, + "loss": 0.4278, + "step": 4159 + }, + { + "epoch": 1.74, + "learning_rate": 1.8208071790106096e-05, + "loss": 0.3944, + "step": 4160 + }, + { + "epoch": 1.74, + "learning_rate": 1.8207103245487335e-05, + "loss": 0.3955, + "step": 4161 + }, + { + "epoch": 1.74, + "learning_rate": 1.8206134464962678e-05, + "loss": 0.3999, + "step": 4162 + }, + { + "epoch": 1.74, + "learning_rate": 1.820516544855997e-05, + "loss": 0.4308, + "step": 4163 + }, + { + "epoch": 1.74, + "learning_rate": 1.8204196196307058e-05, + "loss": 0.3431, + "step": 4164 + }, + { + "epoch": 1.74, + "learning_rate": 1.8203226708231806e-05, + "loss": 0.3637, + "step": 4165 + }, + { + "epoch": 1.74, + "learning_rate": 1.8202256984362087e-05, + "loss": 0.3876, + "step": 4166 + }, + { + "epoch": 1.74, + "learning_rate": 1.8201287024725765e-05, + "loss": 0.3851, + "step": 4167 + }, + { + "epoch": 1.75, + "learning_rate": 1.8200316829350728e-05, + "loss": 0.3599, + "step": 4168 + }, + { + "epoch": 1.75, + "learning_rate": 1.8199346398264856e-05, + "loss": 0.4277, + "step": 4169 + }, + { + "epoch": 1.75, + "learning_rate": 1.8198375731496054e-05, + "loss": 0.3442, + "step": 4170 + }, + { + "epoch": 1.75, + "learning_rate": 1.8197404829072214e-05, + "loss": 0.4514, + "step": 4171 + }, + { + "epoch": 1.75, + "learning_rate": 1.8196433691021243e-05, + "loss": 0.4127, + "step": 4172 + }, + { + "epoch": 1.75, + "learning_rate": 1.8195462317371063e-05, + "loss": 0.4018, + "step": 4173 + }, + { + "epoch": 1.75, + "learning_rate": 1.819449070814959e-05, + "loss": 0.4155, + "step": 4174 + }, + { + "epoch": 1.75, + "learning_rate": 1.8193518863384753e-05, + "loss": 0.3667, + "step": 4175 + }, + { + "epoch": 1.75, + "learning_rate": 1.819254678310449e-05, + "loss": 0.3539, + "step": 4176 + }, + { + "epoch": 1.75, + "learning_rate": 1.8191574467336737e-05, + "loss": 0.4054, + "step": 4177 + }, + { + "epoch": 1.75, + "learning_rate": 1.8190601916109444e-05, + "loss": 0.4086, + "step": 4178 + }, + { + "epoch": 1.75, + "learning_rate": 1.818962912945057e-05, + "loss": 0.3798, + "step": 4179 + }, + { + "epoch": 1.75, + "learning_rate": 1.818865610738807e-05, + "loss": 0.4052, + "step": 4180 + }, + { + "epoch": 1.75, + "learning_rate": 1.818768284994992e-05, + "loss": 0.3307, + "step": 4181 + }, + { + "epoch": 1.75, + "learning_rate": 1.818670935716409e-05, + "loss": 0.3998, + "step": 4182 + }, + { + "epoch": 1.75, + "learning_rate": 1.8185735629058566e-05, + "loss": 0.3693, + "step": 4183 + }, + { + "epoch": 1.75, + "learning_rate": 1.8184761665661335e-05, + "loss": 0.3871, + "step": 4184 + }, + { + "epoch": 1.75, + "learning_rate": 1.8183787467000392e-05, + "loss": 0.3721, + "step": 4185 + }, + { + "epoch": 1.75, + "learning_rate": 1.8182813033103742e-05, + "loss": 0.3355, + "step": 4186 + }, + { + "epoch": 1.75, + "learning_rate": 1.8181838363999392e-05, + "loss": 0.4374, + "step": 4187 + }, + { + "epoch": 1.75, + "learning_rate": 1.8180863459715358e-05, + "loss": 0.3902, + "step": 4188 + }, + { + "epoch": 1.75, + "learning_rate": 1.8179888320279663e-05, + "loss": 0.4099, + "step": 4189 + }, + { + "epoch": 1.75, + "learning_rate": 1.817891294572034e-05, + "loss": 0.3684, + "step": 4190 + }, + { + "epoch": 1.75, + "learning_rate": 1.817793733606542e-05, + "loss": 0.4386, + "step": 4191 + }, + { + "epoch": 1.76, + "learning_rate": 1.8176961491342947e-05, + "loss": 0.3864, + "step": 4192 + }, + { + "epoch": 1.76, + "learning_rate": 1.8175985411580975e-05, + "loss": 0.4155, + "step": 4193 + }, + { + "epoch": 1.76, + "learning_rate": 1.8175009096807557e-05, + "loss": 0.4204, + "step": 4194 + }, + { + "epoch": 1.76, + "learning_rate": 1.8174032547050754e-05, + "loss": 0.382, + "step": 4195 + }, + { + "epoch": 1.76, + "learning_rate": 1.8173055762338642e-05, + "loss": 0.4276, + "step": 4196 + }, + { + "epoch": 1.76, + "learning_rate": 1.8172078742699293e-05, + "loss": 0.3971, + "step": 4197 + }, + { + "epoch": 1.76, + "learning_rate": 1.817110148816079e-05, + "loss": 0.3628, + "step": 4198 + }, + { + "epoch": 1.76, + "learning_rate": 1.817012399875123e-05, + "loss": 0.4342, + "step": 4199 + }, + { + "epoch": 1.76, + "learning_rate": 1.8169146274498707e-05, + "loss": 0.3692, + "step": 4200 + }, + { + "epoch": 1.76, + "learning_rate": 1.8168168315431318e-05, + "loss": 0.3981, + "step": 4201 + }, + { + "epoch": 1.76, + "learning_rate": 1.8167190121577184e-05, + "loss": 0.4388, + "step": 4202 + }, + { + "epoch": 1.76, + "learning_rate": 1.8166211692964413e-05, + "loss": 0.3119, + "step": 4203 + }, + { + "epoch": 1.76, + "learning_rate": 1.8165233029621136e-05, + "loss": 0.3513, + "step": 4204 + }, + { + "epoch": 1.76, + "learning_rate": 1.816425413157548e-05, + "loss": 0.4071, + "step": 4205 + }, + { + "epoch": 1.76, + "learning_rate": 1.8163274998855585e-05, + "loss": 0.4342, + "step": 4206 + }, + { + "epoch": 1.76, + "learning_rate": 1.8162295631489595e-05, + "loss": 0.418, + "step": 4207 + }, + { + "epoch": 1.76, + "learning_rate": 1.8161316029505656e-05, + "loss": 0.3763, + "step": 4208 + }, + { + "epoch": 1.76, + "learning_rate": 1.8160336192931935e-05, + "loss": 0.4215, + "step": 4209 + }, + { + "epoch": 1.76, + "learning_rate": 1.8159356121796586e-05, + "loss": 0.3893, + "step": 4210 + }, + { + "epoch": 1.76, + "learning_rate": 1.8158375816127792e-05, + "loss": 0.4154, + "step": 4211 + }, + { + "epoch": 1.76, + "learning_rate": 1.8157395275953722e-05, + "loss": 0.375, + "step": 4212 + }, + { + "epoch": 1.76, + "learning_rate": 1.8156414501302565e-05, + "loss": 0.3681, + "step": 4213 + }, + { + "epoch": 1.76, + "learning_rate": 1.8155433492202507e-05, + "loss": 0.3854, + "step": 4214 + }, + { + "epoch": 1.76, + "learning_rate": 1.8154452248681754e-05, + "loss": 0.3947, + "step": 4215 + }, + { + "epoch": 1.77, + "learning_rate": 1.8153470770768506e-05, + "loss": 0.3914, + "step": 4216 + }, + { + "epoch": 1.77, + "learning_rate": 1.8152489058490975e-05, + "loss": 0.3556, + "step": 4217 + }, + { + "epoch": 1.77, + "learning_rate": 1.8151507111877382e-05, + "loss": 0.3644, + "step": 4218 + }, + { + "epoch": 1.77, + "learning_rate": 1.815052493095595e-05, + "loss": 0.3726, + "step": 4219 + }, + { + "epoch": 1.77, + "learning_rate": 1.814954251575491e-05, + "loss": 0.4142, + "step": 4220 + }, + { + "epoch": 1.77, + "learning_rate": 1.8148559866302505e-05, + "loss": 0.3856, + "step": 4221 + }, + { + "epoch": 1.77, + "learning_rate": 1.8147576982626974e-05, + "loss": 0.3581, + "step": 4222 + }, + { + "epoch": 1.77, + "learning_rate": 1.8146593864756577e-05, + "loss": 0.4268, + "step": 4223 + }, + { + "epoch": 1.77, + "learning_rate": 1.8145610512719565e-05, + "loss": 0.3792, + "step": 4224 + }, + { + "epoch": 1.77, + "learning_rate": 1.8144626926544208e-05, + "loss": 0.3814, + "step": 4225 + }, + { + "epoch": 1.77, + "learning_rate": 1.814364310625878e-05, + "loss": 0.3749, + "step": 4226 + }, + { + "epoch": 1.77, + "learning_rate": 1.8142659051891552e-05, + "loss": 0.3667, + "step": 4227 + }, + { + "epoch": 1.77, + "learning_rate": 1.814167476347082e-05, + "loss": 0.3915, + "step": 4228 + }, + { + "epoch": 1.77, + "learning_rate": 1.8140690241024872e-05, + "loss": 0.4039, + "step": 4229 + }, + { + "epoch": 1.77, + "learning_rate": 1.8139705484582005e-05, + "loss": 0.3712, + "step": 4230 + }, + { + "epoch": 1.77, + "learning_rate": 1.813872049417053e-05, + "loss": 0.3634, + "step": 4231 + }, + { + "epoch": 1.77, + "learning_rate": 1.813773526981875e-05, + "loss": 0.4128, + "step": 4232 + }, + { + "epoch": 1.77, + "learning_rate": 1.8136749811554995e-05, + "loss": 0.3902, + "step": 4233 + }, + { + "epoch": 1.77, + "learning_rate": 1.8135764119407585e-05, + "loss": 0.3534, + "step": 4234 + }, + { + "epoch": 1.77, + "learning_rate": 1.8134778193404857e-05, + "loss": 0.3815, + "step": 4235 + }, + { + "epoch": 1.77, + "learning_rate": 1.813379203357515e-05, + "loss": 0.3744, + "step": 4236 + }, + { + "epoch": 1.77, + "learning_rate": 1.8132805639946806e-05, + "loss": 0.3964, + "step": 4237 + }, + { + "epoch": 1.77, + "learning_rate": 1.8131819012548183e-05, + "loss": 0.3992, + "step": 4238 + }, + { + "epoch": 1.77, + "learning_rate": 1.8130832151407634e-05, + "loss": 0.3965, + "step": 4239 + }, + { + "epoch": 1.78, + "learning_rate": 1.8129845056553534e-05, + "loss": 0.4237, + "step": 4240 + }, + { + "epoch": 1.78, + "learning_rate": 1.812885772801425e-05, + "loss": 0.3852, + "step": 4241 + }, + { + "epoch": 1.78, + "learning_rate": 1.8127870165818166e-05, + "loss": 0.4264, + "step": 4242 + }, + { + "epoch": 1.78, + "learning_rate": 1.812688236999367e-05, + "loss": 0.4305, + "step": 4243 + }, + { + "epoch": 1.78, + "learning_rate": 1.8125894340569145e-05, + "loss": 0.3854, + "step": 4244 + }, + { + "epoch": 1.78, + "learning_rate": 1.8124906077573e-05, + "loss": 0.3823, + "step": 4245 + }, + { + "epoch": 1.78, + "learning_rate": 1.8123917581033642e-05, + "loss": 0.3893, + "step": 4246 + }, + { + "epoch": 1.78, + "learning_rate": 1.8122928850979483e-05, + "loss": 0.4558, + "step": 4247 + }, + { + "epoch": 1.78, + "learning_rate": 1.8121939887438936e-05, + "loss": 0.3923, + "step": 4248 + }, + { + "epoch": 1.78, + "learning_rate": 1.812095069044044e-05, + "loss": 0.37, + "step": 4249 + }, + { + "epoch": 1.78, + "learning_rate": 1.8119961260012422e-05, + "loss": 0.423, + "step": 4250 + }, + { + "epoch": 1.78, + "learning_rate": 1.811897159618332e-05, + "loss": 0.3789, + "step": 4251 + }, + { + "epoch": 1.78, + "learning_rate": 1.811798169898159e-05, + "loss": 0.3585, + "step": 4252 + }, + { + "epoch": 1.78, + "learning_rate": 1.8116991568435674e-05, + "loss": 0.3785, + "step": 4253 + }, + { + "epoch": 1.78, + "learning_rate": 1.8116001204574042e-05, + "loss": 0.4106, + "step": 4254 + }, + { + "epoch": 1.78, + "learning_rate": 1.8115010607425154e-05, + "loss": 0.4085, + "step": 4255 + }, + { + "epoch": 1.78, + "learning_rate": 1.811401977701749e-05, + "loss": 0.45, + "step": 4256 + }, + { + "epoch": 1.78, + "learning_rate": 1.8113028713379524e-05, + "loss": 0.4488, + "step": 4257 + }, + { + "epoch": 1.78, + "learning_rate": 1.811203741653975e-05, + "loss": 0.409, + "step": 4258 + }, + { + "epoch": 1.78, + "learning_rate": 1.811104588652666e-05, + "loss": 0.3769, + "step": 4259 + }, + { + "epoch": 1.78, + "learning_rate": 1.8110054123368755e-05, + "loss": 0.3857, + "step": 4260 + }, + { + "epoch": 1.78, + "learning_rate": 1.8109062127094538e-05, + "loss": 0.4303, + "step": 4261 + }, + { + "epoch": 1.78, + "learning_rate": 1.8108069897732527e-05, + "loss": 0.4154, + "step": 4262 + }, + { + "epoch": 1.78, + "learning_rate": 1.810707743531124e-05, + "loss": 0.3739, + "step": 4263 + }, + { + "epoch": 1.79, + "learning_rate": 1.8106084739859206e-05, + "loss": 0.3749, + "step": 4264 + }, + { + "epoch": 1.79, + "learning_rate": 1.8105091811404962e-05, + "loss": 0.4192, + "step": 4265 + }, + { + "epoch": 1.79, + "learning_rate": 1.8104098649977044e-05, + "loss": 0.4326, + "step": 4266 + }, + { + "epoch": 1.79, + "learning_rate": 1.8103105255604003e-05, + "loss": 0.4022, + "step": 4267 + }, + { + "epoch": 1.79, + "learning_rate": 1.8102111628314394e-05, + "loss": 0.4064, + "step": 4268 + }, + { + "epoch": 1.79, + "learning_rate": 1.810111776813677e-05, + "loss": 0.3597, + "step": 4269 + }, + { + "epoch": 1.79, + "learning_rate": 1.810012367509971e-05, + "loss": 0.397, + "step": 4270 + }, + { + "epoch": 1.79, + "learning_rate": 1.8099129349231783e-05, + "loss": 0.4079, + "step": 4271 + }, + { + "epoch": 1.79, + "learning_rate": 1.8098134790561567e-05, + "loss": 0.3914, + "step": 4272 + }, + { + "epoch": 1.79, + "learning_rate": 1.8097139999117655e-05, + "loss": 0.3927, + "step": 4273 + }, + { + "epoch": 1.79, + "learning_rate": 1.8096144974928635e-05, + "loss": 0.4111, + "step": 4274 + }, + { + "epoch": 1.79, + "learning_rate": 1.8095149718023114e-05, + "loss": 0.4026, + "step": 4275 + }, + { + "epoch": 1.79, + "learning_rate": 1.80941542284297e-05, + "loss": 0.3648, + "step": 4276 + }, + { + "epoch": 1.79, + "learning_rate": 1.8093158506177003e-05, + "loss": 0.3546, + "step": 4277 + }, + { + "epoch": 1.79, + "learning_rate": 1.8092162551293646e-05, + "loss": 0.3978, + "step": 4278 + }, + { + "epoch": 1.79, + "learning_rate": 1.8091166363808258e-05, + "loss": 0.3781, + "step": 4279 + }, + { + "epoch": 1.79, + "learning_rate": 1.8090169943749477e-05, + "loss": 0.3416, + "step": 4280 + }, + { + "epoch": 1.79, + "learning_rate": 1.8089173291145935e-05, + "loss": 0.3681, + "step": 4281 + }, + { + "epoch": 1.79, + "learning_rate": 1.808817640602629e-05, + "loss": 0.402, + "step": 4282 + }, + { + "epoch": 1.79, + "learning_rate": 1.8087179288419182e-05, + "loss": 0.4024, + "step": 4283 + }, + { + "epoch": 1.79, + "learning_rate": 1.808618193835329e-05, + "loss": 0.3581, + "step": 4284 + }, + { + "epoch": 1.79, + "learning_rate": 1.808518435585727e-05, + "loss": 0.4094, + "step": 4285 + }, + { + "epoch": 1.79, + "learning_rate": 1.80841865409598e-05, + "loss": 0.3766, + "step": 4286 + }, + { + "epoch": 1.79, + "learning_rate": 1.808318849368957e-05, + "loss": 0.4021, + "step": 4287 + }, + { + "epoch": 1.8, + "learning_rate": 1.8082190214075252e-05, + "loss": 0.3986, + "step": 4288 + }, + { + "epoch": 1.8, + "learning_rate": 1.8081191702145547e-05, + "loss": 0.397, + "step": 4289 + }, + { + "epoch": 1.8, + "learning_rate": 1.8080192957929162e-05, + "loss": 0.3872, + "step": 4290 + }, + { + "epoch": 1.8, + "learning_rate": 1.8079193981454798e-05, + "loss": 0.3809, + "step": 4291 + }, + { + "epoch": 1.8, + "learning_rate": 1.8078194772751173e-05, + "loss": 0.4337, + "step": 4292 + }, + { + "epoch": 1.8, + "learning_rate": 1.8077195331847007e-05, + "loss": 0.4027, + "step": 4293 + }, + { + "epoch": 1.8, + "learning_rate": 1.8076195658771033e-05, + "loss": 0.3748, + "step": 4294 + }, + { + "epoch": 1.8, + "learning_rate": 1.8075195753551974e-05, + "loss": 0.3974, + "step": 4295 + }, + { + "epoch": 1.8, + "learning_rate": 1.8074195616218583e-05, + "loss": 0.3615, + "step": 4296 + }, + { + "epoch": 1.8, + "learning_rate": 1.8073195246799605e-05, + "loss": 0.375, + "step": 4297 + }, + { + "epoch": 1.8, + "learning_rate": 1.8072194645323794e-05, + "loss": 0.449, + "step": 4298 + }, + { + "epoch": 1.8, + "learning_rate": 1.8071193811819912e-05, + "loss": 0.3788, + "step": 4299 + }, + { + "epoch": 1.8, + "learning_rate": 1.807019274631672e-05, + "loss": 0.3921, + "step": 4300 + }, + { + "epoch": 1.8, + "learning_rate": 1.8069191448843005e-05, + "loss": 0.3985, + "step": 4301 + }, + { + "epoch": 1.8, + "learning_rate": 1.806818991942754e-05, + "loss": 0.3922, + "step": 4302 + }, + { + "epoch": 1.8, + "learning_rate": 1.8067188158099112e-05, + "loss": 0.367, + "step": 4303 + }, + { + "epoch": 1.8, + "learning_rate": 1.8066186164886525e-05, + "loss": 0.4072, + "step": 4304 + }, + { + "epoch": 1.8, + "learning_rate": 1.806518393981857e-05, + "loss": 0.3879, + "step": 4305 + }, + { + "epoch": 1.8, + "learning_rate": 1.8064181482924065e-05, + "loss": 0.377, + "step": 4306 + }, + { + "epoch": 1.8, + "learning_rate": 1.8063178794231814e-05, + "loss": 0.4025, + "step": 4307 + }, + { + "epoch": 1.8, + "learning_rate": 1.8062175873770643e-05, + "loss": 0.3922, + "step": 4308 + }, + { + "epoch": 1.8, + "learning_rate": 1.8061172721569383e-05, + "loss": 0.4284, + "step": 4309 + }, + { + "epoch": 1.8, + "learning_rate": 1.8060169337656866e-05, + "loss": 0.4754, + "step": 4310 + }, + { + "epoch": 1.8, + "learning_rate": 1.805916572206194e-05, + "loss": 0.3792, + "step": 4311 + }, + { + "epoch": 1.81, + "learning_rate": 1.805816187481344e-05, + "loss": 0.374, + "step": 4312 + }, + { + "epoch": 1.81, + "learning_rate": 1.8057157795940228e-05, + "loss": 0.4356, + "step": 4313 + }, + { + "epoch": 1.81, + "learning_rate": 1.8056153485471167e-05, + "loss": 0.3891, + "step": 4314 + }, + { + "epoch": 1.81, + "learning_rate": 1.805514894343512e-05, + "loss": 0.3978, + "step": 4315 + }, + { + "epoch": 1.81, + "learning_rate": 1.805414416986097e-05, + "loss": 0.411, + "step": 4316 + }, + { + "epoch": 1.81, + "learning_rate": 1.8053139164777587e-05, + "loss": 0.3439, + "step": 4317 + }, + { + "epoch": 1.81, + "learning_rate": 1.805213392821387e-05, + "loss": 0.42, + "step": 4318 + }, + { + "epoch": 1.81, + "learning_rate": 1.8051128460198703e-05, + "loss": 0.3687, + "step": 4319 + }, + { + "epoch": 1.81, + "learning_rate": 1.8050122760760994e-05, + "loss": 0.3926, + "step": 4320 + }, + { + "epoch": 1.81, + "learning_rate": 1.8049116829929655e-05, + "loss": 0.3792, + "step": 4321 + }, + { + "epoch": 1.81, + "learning_rate": 1.804811066773359e-05, + "loss": 0.3436, + "step": 4322 + }, + { + "epoch": 1.81, + "learning_rate": 1.804710427420173e-05, + "loss": 0.386, + "step": 4323 + }, + { + "epoch": 1.81, + "learning_rate": 1.8046097649362994e-05, + "loss": 0.3816, + "step": 4324 + }, + { + "epoch": 1.81, + "learning_rate": 1.804509079324632e-05, + "loss": 0.3676, + "step": 4325 + }, + { + "epoch": 1.81, + "learning_rate": 1.804408370588066e-05, + "loss": 0.3695, + "step": 4326 + }, + { + "epoch": 1.81, + "learning_rate": 1.8043076387294944e-05, + "loss": 0.4537, + "step": 4327 + }, + { + "epoch": 1.81, + "learning_rate": 1.8042068837518135e-05, + "loss": 0.3724, + "step": 4328 + }, + { + "epoch": 1.81, + "learning_rate": 1.8041061056579194e-05, + "loss": 0.3988, + "step": 4329 + }, + { + "epoch": 1.81, + "learning_rate": 1.8040053044507087e-05, + "loss": 0.3737, + "step": 4330 + }, + { + "epoch": 1.81, + "learning_rate": 1.8039044801330795e-05, + "loss": 0.4418, + "step": 4331 + }, + { + "epoch": 1.81, + "learning_rate": 1.803803632707929e-05, + "loss": 0.3556, + "step": 4332 + }, + { + "epoch": 1.81, + "learning_rate": 1.8037027621781568e-05, + "loss": 0.3917, + "step": 4333 + }, + { + "epoch": 1.81, + "learning_rate": 1.8036018685466614e-05, + "loss": 0.4307, + "step": 4334 + }, + { + "epoch": 1.81, + "learning_rate": 1.8035009518163434e-05, + "loss": 0.4482, + "step": 4335 + }, + { + "epoch": 1.82, + "learning_rate": 1.8034000119901038e-05, + "loss": 0.4407, + "step": 4336 + }, + { + "epoch": 1.82, + "learning_rate": 1.8032990490708434e-05, + "loss": 0.3863, + "step": 4337 + }, + { + "epoch": 1.82, + "learning_rate": 1.8031980630614652e-05, + "loss": 0.3908, + "step": 4338 + }, + { + "epoch": 1.82, + "learning_rate": 1.803097053964871e-05, + "loss": 0.3528, + "step": 4339 + }, + { + "epoch": 1.82, + "learning_rate": 1.8029960217839648e-05, + "loss": 0.3768, + "step": 4340 + }, + { + "epoch": 1.82, + "learning_rate": 1.8028949665216505e-05, + "loss": 0.4015, + "step": 4341 + }, + { + "epoch": 1.82, + "learning_rate": 1.802793888180833e-05, + "loss": 0.4411, + "step": 4342 + }, + { + "epoch": 1.82, + "learning_rate": 1.8026927867644175e-05, + "loss": 0.4535, + "step": 4343 + }, + { + "epoch": 1.82, + "learning_rate": 1.80259166227531e-05, + "loss": 0.3384, + "step": 4344 + }, + { + "epoch": 1.82, + "learning_rate": 1.8024905147164176e-05, + "loss": 0.3852, + "step": 4345 + }, + { + "epoch": 1.82, + "learning_rate": 1.8023893440906476e-05, + "loss": 0.3559, + "step": 4346 + }, + { + "epoch": 1.82, + "learning_rate": 1.8022881504009074e-05, + "loss": 0.4396, + "step": 4347 + }, + { + "epoch": 1.82, + "learning_rate": 1.8021869336501066e-05, + "loss": 0.3677, + "step": 4348 + }, + { + "epoch": 1.82, + "learning_rate": 1.802085693841154e-05, + "loss": 0.3632, + "step": 4349 + }, + { + "epoch": 1.82, + "learning_rate": 1.8019844309769602e-05, + "loss": 0.4229, + "step": 4350 + }, + { + "epoch": 1.82, + "learning_rate": 1.8018831450604356e-05, + "loss": 0.4812, + "step": 4351 + }, + { + "epoch": 1.82, + "learning_rate": 1.801781836094491e-05, + "loss": 0.387, + "step": 4352 + }, + { + "epoch": 1.82, + "learning_rate": 1.80168050408204e-05, + "loss": 0.3763, + "step": 4353 + }, + { + "epoch": 1.82, + "learning_rate": 1.8015791490259932e-05, + "loss": 0.3933, + "step": 4354 + }, + { + "epoch": 1.82, + "learning_rate": 1.801477770929266e-05, + "loss": 0.4103, + "step": 4355 + }, + { + "epoch": 1.82, + "learning_rate": 1.801376369794771e-05, + "loss": 0.4138, + "step": 4356 + }, + { + "epoch": 1.82, + "learning_rate": 1.8012749456254238e-05, + "loss": 0.4028, + "step": 4357 + }, + { + "epoch": 1.82, + "learning_rate": 1.801173498424139e-05, + "loss": 0.3912, + "step": 4358 + }, + { + "epoch": 1.82, + "learning_rate": 1.8010720281938327e-05, + "loss": 0.429, + "step": 4359 + }, + { + "epoch": 1.83, + "learning_rate": 1.800970534937422e-05, + "loss": 0.3638, + "step": 4360 + }, + { + "epoch": 1.83, + "learning_rate": 1.8008690186578243e-05, + "loss": 0.4019, + "step": 4361 + }, + { + "epoch": 1.83, + "learning_rate": 1.800767479357957e-05, + "loss": 0.3711, + "step": 4362 + }, + { + "epoch": 1.83, + "learning_rate": 1.8006659170407396e-05, + "loss": 0.3859, + "step": 4363 + }, + { + "epoch": 1.83, + "learning_rate": 1.80056433170909e-05, + "loss": 0.441, + "step": 4364 + }, + { + "epoch": 1.83, + "learning_rate": 1.80046272336593e-05, + "loss": 0.375, + "step": 4365 + }, + { + "epoch": 1.83, + "learning_rate": 1.800361092014179e-05, + "loss": 0.4191, + "step": 4366 + }, + { + "epoch": 1.83, + "learning_rate": 1.8002594376567585e-05, + "loss": 0.3702, + "step": 4367 + }, + { + "epoch": 1.83, + "learning_rate": 1.8001577602965907e-05, + "loss": 0.4165, + "step": 4368 + }, + { + "epoch": 1.83, + "learning_rate": 1.800056059936598e-05, + "loss": 0.4246, + "step": 4369 + }, + { + "epoch": 1.83, + "learning_rate": 1.7999543365797043e-05, + "loss": 0.4047, + "step": 4370 + }, + { + "epoch": 1.83, + "learning_rate": 1.7998525902288327e-05, + "loss": 0.3736, + "step": 4371 + }, + { + "epoch": 1.83, + "learning_rate": 1.7997508208869082e-05, + "loss": 0.4108, + "step": 4372 + }, + { + "epoch": 1.83, + "learning_rate": 1.799649028556856e-05, + "loss": 0.3926, + "step": 4373 + }, + { + "epoch": 1.83, + "learning_rate": 1.799547213241602e-05, + "loss": 0.403, + "step": 4374 + }, + { + "epoch": 1.83, + "learning_rate": 1.799445374944073e-05, + "loss": 0.3477, + "step": 4375 + }, + { + "epoch": 1.83, + "learning_rate": 1.799343513667196e-05, + "loss": 0.4046, + "step": 4376 + }, + { + "epoch": 1.83, + "learning_rate": 1.7992416294138992e-05, + "loss": 0.3881, + "step": 4377 + }, + { + "epoch": 1.83, + "learning_rate": 1.799139722187111e-05, + "loss": 0.4091, + "step": 4378 + }, + { + "epoch": 1.83, + "learning_rate": 1.7990377919897605e-05, + "loss": 0.399, + "step": 4379 + }, + { + "epoch": 1.83, + "learning_rate": 1.7989358388247776e-05, + "loss": 0.4123, + "step": 4380 + }, + { + "epoch": 1.83, + "learning_rate": 1.798833862695093e-05, + "loss": 0.4001, + "step": 4381 + }, + { + "epoch": 1.83, + "learning_rate": 1.7987318636036382e-05, + "loss": 0.3674, + "step": 4382 + }, + { + "epoch": 1.84, + "learning_rate": 1.7986298415533444e-05, + "loss": 0.3603, + "step": 4383 + }, + { + "epoch": 1.84, + "learning_rate": 1.798527796547145e-05, + "loss": 0.4306, + "step": 4384 + }, + { + "epoch": 1.84, + "learning_rate": 1.7984257285879724e-05, + "loss": 0.3837, + "step": 4385 + }, + { + "epoch": 1.84, + "learning_rate": 1.7983236376787607e-05, + "loss": 0.4104, + "step": 4386 + }, + { + "epoch": 1.84, + "learning_rate": 1.7982215238224445e-05, + "loss": 0.368, + "step": 4387 + }, + { + "epoch": 1.84, + "learning_rate": 1.7981193870219587e-05, + "loss": 0.3786, + "step": 4388 + }, + { + "epoch": 1.84, + "learning_rate": 1.7980172272802398e-05, + "loss": 0.4624, + "step": 4389 + }, + { + "epoch": 1.84, + "learning_rate": 1.7979150446002233e-05, + "loss": 0.4058, + "step": 4390 + }, + { + "epoch": 1.84, + "learning_rate": 1.7978128389848476e-05, + "loss": 0.3636, + "step": 4391 + }, + { + "epoch": 1.84, + "learning_rate": 1.7977106104370493e-05, + "loss": 0.3949, + "step": 4392 + }, + { + "epoch": 1.84, + "learning_rate": 1.7976083589597672e-05, + "loss": 0.3636, + "step": 4393 + }, + { + "epoch": 1.84, + "learning_rate": 1.7975060845559412e-05, + "loss": 0.3717, + "step": 4394 + }, + { + "epoch": 1.84, + "learning_rate": 1.7974037872285102e-05, + "loss": 0.4207, + "step": 4395 + }, + { + "epoch": 1.84, + "learning_rate": 1.797301466980415e-05, + "loss": 0.3758, + "step": 4396 + }, + { + "epoch": 1.84, + "learning_rate": 1.7971991238145965e-05, + "loss": 0.3979, + "step": 4397 + }, + { + "epoch": 1.84, + "learning_rate": 1.7970967577339965e-05, + "loss": 0.4325, + "step": 4398 + }, + { + "epoch": 1.84, + "learning_rate": 1.7969943687415575e-05, + "loss": 0.4123, + "step": 4399 + }, + { + "epoch": 1.84, + "learning_rate": 1.7968919568402234e-05, + "loss": 0.3432, + "step": 4400 + }, + { + "epoch": 1.84, + "learning_rate": 1.7967895220329364e-05, + "loss": 0.3518, + "step": 4401 + }, + { + "epoch": 1.84, + "learning_rate": 1.796687064322642e-05, + "loss": 0.4224, + "step": 4402 + }, + { + "epoch": 1.84, + "learning_rate": 1.7965845837122844e-05, + "loss": 0.3905, + "step": 4403 + }, + { + "epoch": 1.84, + "learning_rate": 1.79648208020481e-05, + "loss": 0.3709, + "step": 4404 + }, + { + "epoch": 1.84, + "learning_rate": 1.796379553803165e-05, + "loss": 0.4077, + "step": 4405 + }, + { + "epoch": 1.84, + "learning_rate": 1.7962770045102968e-05, + "loss": 0.3806, + "step": 4406 + }, + { + "epoch": 1.85, + "learning_rate": 1.7961744323291525e-05, + "loss": 0.3653, + "step": 4407 + }, + { + "epoch": 1.85, + "learning_rate": 1.7960718372626807e-05, + "loss": 0.435, + "step": 4408 + }, + { + "epoch": 1.85, + "learning_rate": 1.7959692193138303e-05, + "loss": 0.3866, + "step": 4409 + }, + { + "epoch": 1.85, + "learning_rate": 1.795866578485551e-05, + "loss": 0.4393, + "step": 4410 + }, + { + "epoch": 1.85, + "learning_rate": 1.795763914780793e-05, + "loss": 0.4025, + "step": 4411 + }, + { + "epoch": 1.85, + "learning_rate": 1.7956612282025078e-05, + "loss": 0.3961, + "step": 4412 + }, + { + "epoch": 1.85, + "learning_rate": 1.7955585187536463e-05, + "loss": 0.4066, + "step": 4413 + }, + { + "epoch": 1.85, + "learning_rate": 1.7954557864371615e-05, + "loss": 0.3567, + "step": 4414 + }, + { + "epoch": 1.85, + "learning_rate": 1.7953530312560056e-05, + "loss": 0.4295, + "step": 4415 + }, + { + "epoch": 1.85, + "learning_rate": 1.7952502532131326e-05, + "loss": 0.3641, + "step": 4416 + }, + { + "epoch": 1.85, + "learning_rate": 1.795147452311497e-05, + "loss": 0.4159, + "step": 4417 + }, + { + "epoch": 1.85, + "learning_rate": 1.7950446285540532e-05, + "loss": 0.3475, + "step": 4418 + }, + { + "epoch": 1.85, + "learning_rate": 1.7949417819437577e-05, + "loss": 0.3599, + "step": 4419 + }, + { + "epoch": 1.85, + "learning_rate": 1.7948389124835654e-05, + "loss": 0.3755, + "step": 4420 + }, + { + "epoch": 1.85, + "learning_rate": 1.794736020176434e-05, + "loss": 0.4222, + "step": 4421 + }, + { + "epoch": 1.85, + "learning_rate": 1.794633105025321e-05, + "loss": 0.3815, + "step": 4422 + }, + { + "epoch": 1.85, + "learning_rate": 1.7945301670331847e-05, + "loss": 0.4236, + "step": 4423 + }, + { + "epoch": 1.85, + "learning_rate": 1.794427206202984e-05, + "loss": 0.3942, + "step": 4424 + }, + { + "epoch": 1.85, + "learning_rate": 1.7943242225376778e-05, + "loss": 0.3716, + "step": 4425 + }, + { + "epoch": 1.85, + "learning_rate": 1.794221216040227e-05, + "loss": 0.3625, + "step": 4426 + }, + { + "epoch": 1.85, + "learning_rate": 1.7941181867135917e-05, + "loss": 0.4135, + "step": 4427 + }, + { + "epoch": 1.85, + "learning_rate": 1.7940151345607343e-05, + "loss": 0.3409, + "step": 4428 + }, + { + "epoch": 1.85, + "learning_rate": 1.793912059584616e-05, + "loss": 0.3957, + "step": 4429 + }, + { + "epoch": 1.85, + "learning_rate": 1.7938089617882004e-05, + "loss": 0.3973, + "step": 4430 + }, + { + "epoch": 1.86, + "learning_rate": 1.7937058411744504e-05, + "loss": 0.4159, + "step": 4431 + }, + { + "epoch": 1.86, + "learning_rate": 1.7936026977463306e-05, + "loss": 0.3604, + "step": 4432 + }, + { + "epoch": 1.86, + "learning_rate": 1.793499531506805e-05, + "loss": 0.3864, + "step": 4433 + }, + { + "epoch": 1.86, + "learning_rate": 1.79339634245884e-05, + "loss": 0.4238, + "step": 4434 + }, + { + "epoch": 1.86, + "learning_rate": 1.7932931306054007e-05, + "loss": 0.3839, + "step": 4435 + }, + { + "epoch": 1.86, + "learning_rate": 1.7931898959494546e-05, + "loss": 0.3724, + "step": 4436 + }, + { + "epoch": 1.86, + "learning_rate": 1.793086638493969e-05, + "loss": 0.383, + "step": 4437 + }, + { + "epoch": 1.86, + "learning_rate": 1.7929833582419115e-05, + "loss": 0.4443, + "step": 4438 + }, + { + "epoch": 1.86, + "learning_rate": 1.792880055196251e-05, + "loss": 0.4123, + "step": 4439 + }, + { + "epoch": 1.86, + "learning_rate": 1.792776729359957e-05, + "loss": 0.3807, + "step": 4440 + }, + { + "epoch": 1.86, + "learning_rate": 1.7926733807359993e-05, + "loss": 0.4245, + "step": 4441 + }, + { + "epoch": 1.86, + "learning_rate": 1.792570009327349e-05, + "loss": 0.3401, + "step": 4442 + }, + { + "epoch": 1.86, + "learning_rate": 1.792466615136977e-05, + "loss": 0.4053, + "step": 4443 + }, + { + "epoch": 1.86, + "learning_rate": 1.7923631981678552e-05, + "loss": 0.4292, + "step": 4444 + }, + { + "epoch": 1.86, + "learning_rate": 1.7922597584229566e-05, + "loss": 0.3729, + "step": 4445 + }, + { + "epoch": 1.86, + "learning_rate": 1.792156295905254e-05, + "loss": 0.4087, + "step": 4446 + }, + { + "epoch": 1.86, + "learning_rate": 1.792052810617722e-05, + "loss": 0.3385, + "step": 4447 + }, + { + "epoch": 1.86, + "learning_rate": 1.7919493025633342e-05, + "loss": 0.4045, + "step": 4448 + }, + { + "epoch": 1.86, + "learning_rate": 1.791845771745067e-05, + "loss": 0.3663, + "step": 4449 + }, + { + "epoch": 1.86, + "learning_rate": 1.7917422181658954e-05, + "loss": 0.4042, + "step": 4450 + }, + { + "epoch": 1.86, + "learning_rate": 1.7916386418287968e-05, + "loss": 0.3809, + "step": 4451 + }, + { + "epoch": 1.86, + "learning_rate": 1.7915350427367476e-05, + "loss": 0.4189, + "step": 4452 + }, + { + "epoch": 1.86, + "learning_rate": 1.791431420892726e-05, + "loss": 0.3534, + "step": 4453 + }, + { + "epoch": 1.86, + "learning_rate": 1.7913277762997105e-05, + "loss": 0.3893, + "step": 4454 + }, + { + "epoch": 1.87, + "learning_rate": 1.7912241089606802e-05, + "loss": 0.3716, + "step": 4455 + }, + { + "epoch": 1.87, + "learning_rate": 1.7911204188786154e-05, + "loss": 0.4082, + "step": 4456 + }, + { + "epoch": 1.87, + "learning_rate": 1.791016706056496e-05, + "loss": 0.3839, + "step": 4457 + }, + { + "epoch": 1.87, + "learning_rate": 1.790912970497303e-05, + "loss": 0.4122, + "step": 4458 + }, + { + "epoch": 1.87, + "learning_rate": 1.790809212204019e-05, + "loss": 0.3856, + "step": 4459 + }, + { + "epoch": 1.87, + "learning_rate": 1.7907054311796254e-05, + "loss": 0.4127, + "step": 4460 + }, + { + "epoch": 1.87, + "learning_rate": 1.7906016274271062e-05, + "loss": 0.3855, + "step": 4461 + }, + { + "epoch": 1.87, + "learning_rate": 1.790497800949445e-05, + "loss": 0.4138, + "step": 4462 + }, + { + "epoch": 1.87, + "learning_rate": 1.7903939517496253e-05, + "loss": 0.4138, + "step": 4463 + }, + { + "epoch": 1.87, + "learning_rate": 1.7902900798306334e-05, + "loss": 0.3514, + "step": 4464 + }, + { + "epoch": 1.87, + "learning_rate": 1.7901861851954543e-05, + "loss": 0.3967, + "step": 4465 + }, + { + "epoch": 1.87, + "learning_rate": 1.7900822678470745e-05, + "loss": 0.4038, + "step": 4466 + }, + { + "epoch": 1.87, + "learning_rate": 1.789978327788481e-05, + "loss": 0.3741, + "step": 4467 + }, + { + "epoch": 1.87, + "learning_rate": 1.7898743650226617e-05, + "loss": 0.4083, + "step": 4468 + }, + { + "epoch": 1.87, + "learning_rate": 1.7897703795526044e-05, + "loss": 0.3672, + "step": 4469 + }, + { + "epoch": 1.87, + "learning_rate": 1.7896663713812984e-05, + "loss": 0.4927, + "step": 4470 + }, + { + "epoch": 1.87, + "learning_rate": 1.7895623405117334e-05, + "loss": 0.4271, + "step": 4471 + }, + { + "epoch": 1.87, + "learning_rate": 1.7894582869468994e-05, + "loss": 0.3796, + "step": 4472 + }, + { + "epoch": 1.87, + "learning_rate": 1.7893542106897878e-05, + "loss": 0.4172, + "step": 4473 + }, + { + "epoch": 1.87, + "learning_rate": 1.7892501117433896e-05, + "loss": 0.3696, + "step": 4474 + }, + { + "epoch": 1.87, + "learning_rate": 1.7891459901106976e-05, + "loss": 0.3844, + "step": 4475 + }, + { + "epoch": 1.87, + "learning_rate": 1.7890418457947038e-05, + "loss": 0.3763, + "step": 4476 + }, + { + "epoch": 1.87, + "learning_rate": 1.788937678798403e-05, + "loss": 0.3572, + "step": 4477 + }, + { + "epoch": 1.87, + "learning_rate": 1.7888334891247883e-05, + "loss": 0.3744, + "step": 4478 + }, + { + "epoch": 1.88, + "learning_rate": 1.788729276776855e-05, + "loss": 0.3865, + "step": 4479 + }, + { + "epoch": 1.88, + "learning_rate": 1.7886250417575986e-05, + "loss": 0.4056, + "step": 4480 + }, + { + "epoch": 1.88, + "learning_rate": 1.7885207840700155e-05, + "loss": 0.4122, + "step": 4481 + }, + { + "epoch": 1.88, + "learning_rate": 1.7884165037171016e-05, + "loss": 0.3764, + "step": 4482 + }, + { + "epoch": 1.88, + "learning_rate": 1.7883122007018554e-05, + "loss": 0.3989, + "step": 4483 + }, + { + "epoch": 1.88, + "learning_rate": 1.788207875027274e-05, + "loss": 0.4024, + "step": 4484 + }, + { + "epoch": 1.88, + "learning_rate": 1.7881035266963575e-05, + "loss": 0.3932, + "step": 4485 + }, + { + "epoch": 1.88, + "learning_rate": 1.787999155712104e-05, + "loss": 0.3933, + "step": 4486 + }, + { + "epoch": 1.88, + "learning_rate": 1.787894762077514e-05, + "loss": 0.3786, + "step": 4487 + }, + { + "epoch": 1.88, + "learning_rate": 1.787790345795588e-05, + "loss": 0.4072, + "step": 4488 + }, + { + "epoch": 1.88, + "learning_rate": 1.787685906869328e-05, + "loss": 0.4105, + "step": 4489 + }, + { + "epoch": 1.88, + "learning_rate": 1.7875814453017355e-05, + "loss": 0.3618, + "step": 4490 + }, + { + "epoch": 1.88, + "learning_rate": 1.787476961095813e-05, + "loss": 0.4214, + "step": 4491 + }, + { + "epoch": 1.88, + "learning_rate": 1.7873724542545647e-05, + "loss": 0.3606, + "step": 4492 + }, + { + "epoch": 1.88, + "learning_rate": 1.7872679247809933e-05, + "loss": 0.397, + "step": 4493 + }, + { + "epoch": 1.88, + "learning_rate": 1.7871633726781042e-05, + "loss": 0.4022, + "step": 4494 + }, + { + "epoch": 1.88, + "learning_rate": 1.787058797948902e-05, + "loss": 0.3629, + "step": 4495 + }, + { + "epoch": 1.88, + "learning_rate": 1.786954200596394e-05, + "loss": 0.427, + "step": 4496 + }, + { + "epoch": 1.88, + "learning_rate": 1.786849580623585e-05, + "loss": 0.3878, + "step": 4497 + }, + { + "epoch": 1.88, + "learning_rate": 1.7867449380334834e-05, + "loss": 0.4738, + "step": 4498 + }, + { + "epoch": 1.88, + "learning_rate": 1.7866402728290967e-05, + "loss": 0.3804, + "step": 4499 + }, + { + "epoch": 1.88, + "learning_rate": 1.786535585013433e-05, + "loss": 0.4011, + "step": 4500 + }, + { + "epoch": 1.88, + "learning_rate": 1.7864308745895024e-05, + "loss": 0.4079, + "step": 4501 + }, + { + "epoch": 1.88, + "learning_rate": 1.7863261415603142e-05, + "loss": 0.3485, + "step": 4502 + }, + { + "epoch": 1.89, + "learning_rate": 1.7862213859288788e-05, + "loss": 0.4219, + "step": 4503 + }, + { + "epoch": 1.89, + "learning_rate": 1.786116607698207e-05, + "loss": 0.3978, + "step": 4504 + }, + { + "epoch": 1.89, + "learning_rate": 1.786011806871311e-05, + "loss": 0.3497, + "step": 4505 + }, + { + "epoch": 1.89, + "learning_rate": 1.7859069834512033e-05, + "loss": 0.354, + "step": 4506 + }, + { + "epoch": 1.89, + "learning_rate": 1.7858021374408968e-05, + "loss": 0.3959, + "step": 4507 + }, + { + "epoch": 1.89, + "learning_rate": 1.785697268843405e-05, + "loss": 0.375, + "step": 4508 + }, + { + "epoch": 1.89, + "learning_rate": 1.7855923776617424e-05, + "loss": 0.3681, + "step": 4509 + }, + { + "epoch": 1.89, + "learning_rate": 1.785487463898924e-05, + "loss": 0.4029, + "step": 4510 + }, + { + "epoch": 1.89, + "learning_rate": 1.7853825275579656e-05, + "loss": 0.4052, + "step": 4511 + }, + { + "epoch": 1.89, + "learning_rate": 1.7852775686418832e-05, + "loss": 0.4151, + "step": 4512 + }, + { + "epoch": 1.89, + "learning_rate": 1.7851725871536945e-05, + "loss": 0.4123, + "step": 4513 + }, + { + "epoch": 1.89, + "learning_rate": 1.785067583096416e-05, + "loss": 0.3862, + "step": 4514 + }, + { + "epoch": 1.89, + "learning_rate": 1.7849625564730667e-05, + "loss": 0.3823, + "step": 4515 + }, + { + "epoch": 1.89, + "learning_rate": 1.784857507286665e-05, + "loss": 0.3856, + "step": 4516 + }, + { + "epoch": 1.89, + "learning_rate": 1.784752435540231e-05, + "loss": 0.3882, + "step": 4517 + }, + { + "epoch": 1.89, + "learning_rate": 1.7846473412367845e-05, + "loss": 0.4147, + "step": 4518 + }, + { + "epoch": 1.89, + "learning_rate": 1.7845422243793464e-05, + "loss": 0.4482, + "step": 4519 + }, + { + "epoch": 1.89, + "learning_rate": 1.784437084970938e-05, + "loss": 0.3736, + "step": 4520 + }, + { + "epoch": 1.89, + "learning_rate": 1.7843319230145822e-05, + "loss": 0.3869, + "step": 4521 + }, + { + "epoch": 1.89, + "learning_rate": 1.7842267385133013e-05, + "loss": 0.4484, + "step": 4522 + }, + { + "epoch": 1.89, + "learning_rate": 1.7841215314701182e-05, + "loss": 0.4245, + "step": 4523 + }, + { + "epoch": 1.89, + "learning_rate": 1.784016301888058e-05, + "loss": 0.3619, + "step": 4524 + }, + { + "epoch": 1.89, + "learning_rate": 1.7839110497701446e-05, + "loss": 0.4076, + "step": 4525 + }, + { + "epoch": 1.89, + "learning_rate": 1.783805775119404e-05, + "loss": 0.4029, + "step": 4526 + }, + { + "epoch": 1.9, + "learning_rate": 1.7837004779388615e-05, + "loss": 0.4297, + "step": 4527 + }, + { + "epoch": 1.9, + "learning_rate": 1.7835951582315445e-05, + "loss": 0.3741, + "step": 4528 + }, + { + "epoch": 1.9, + "learning_rate": 1.78348981600048e-05, + "loss": 0.4124, + "step": 4529 + }, + { + "epoch": 1.9, + "learning_rate": 1.783384451248696e-05, + "loss": 0.3873, + "step": 4530 + }, + { + "epoch": 1.9, + "learning_rate": 1.783279063979221e-05, + "loss": 0.3949, + "step": 4531 + }, + { + "epoch": 1.9, + "learning_rate": 1.7831736541950845e-05, + "loss": 0.399, + "step": 4532 + }, + { + "epoch": 1.9, + "learning_rate": 1.783068221899316e-05, + "loss": 0.4076, + "step": 4533 + }, + { + "epoch": 1.9, + "learning_rate": 1.7829627670949468e-05, + "loss": 0.4031, + "step": 4534 + }, + { + "epoch": 1.9, + "learning_rate": 1.7828572897850076e-05, + "loss": 0.3656, + "step": 4535 + }, + { + "epoch": 1.9, + "learning_rate": 1.78275178997253e-05, + "loss": 0.3846, + "step": 4536 + }, + { + "epoch": 1.9, + "learning_rate": 1.782646267660547e-05, + "loss": 0.409, + "step": 4537 + }, + { + "epoch": 1.9, + "learning_rate": 1.7825407228520917e-05, + "loss": 0.3568, + "step": 4538 + }, + { + "epoch": 1.9, + "learning_rate": 1.7824351555501972e-05, + "loss": 0.4116, + "step": 4539 + }, + { + "epoch": 1.9, + "learning_rate": 1.782329565757899e-05, + "loss": 0.4143, + "step": 4540 + }, + { + "epoch": 1.9, + "learning_rate": 1.7822239534782312e-05, + "loss": 0.3885, + "step": 4541 + }, + { + "epoch": 1.9, + "learning_rate": 1.7821183187142302e-05, + "loss": 0.4098, + "step": 4542 + }, + { + "epoch": 1.9, + "learning_rate": 1.7820126614689326e-05, + "loss": 0.4534, + "step": 4543 + }, + { + "epoch": 1.9, + "learning_rate": 1.7819069817453746e-05, + "loss": 0.3467, + "step": 4544 + }, + { + "epoch": 1.9, + "learning_rate": 1.781801279546594e-05, + "loss": 0.408, + "step": 4545 + }, + { + "epoch": 1.9, + "learning_rate": 1.7816955548756298e-05, + "loss": 0.3673, + "step": 4546 + }, + { + "epoch": 1.9, + "learning_rate": 1.7815898077355204e-05, + "loss": 0.4112, + "step": 4547 + }, + { + "epoch": 1.9, + "learning_rate": 1.7814840381293054e-05, + "loss": 0.3596, + "step": 4548 + }, + { + "epoch": 1.9, + "learning_rate": 1.7813782460600257e-05, + "loss": 0.3993, + "step": 4549 + }, + { + "epoch": 1.9, + "learning_rate": 1.781272431530721e-05, + "loss": 0.3849, + "step": 4550 + }, + { + "epoch": 1.91, + "learning_rate": 1.7811665945444337e-05, + "loss": 0.3657, + "step": 4551 + }, + { + "epoch": 1.91, + "learning_rate": 1.7810607351042062e-05, + "loss": 0.3464, + "step": 4552 + }, + { + "epoch": 1.91, + "learning_rate": 1.7809548532130808e-05, + "loss": 0.3813, + "step": 4553 + }, + { + "epoch": 1.91, + "learning_rate": 1.7808489488741013e-05, + "loss": 0.4107, + "step": 4554 + }, + { + "epoch": 1.91, + "learning_rate": 1.7807430220903113e-05, + "loss": 0.319, + "step": 4555 + }, + { + "epoch": 1.91, + "learning_rate": 1.780637072864756e-05, + "loss": 0.3817, + "step": 4556 + }, + { + "epoch": 1.91, + "learning_rate": 1.7805311012004812e-05, + "loss": 0.36, + "step": 4557 + }, + { + "epoch": 1.91, + "learning_rate": 1.7804251071005323e-05, + "loss": 0.4005, + "step": 4558 + }, + { + "epoch": 1.91, + "learning_rate": 1.7803190905679562e-05, + "loss": 0.3722, + "step": 4559 + }, + { + "epoch": 1.91, + "learning_rate": 1.7802130516058003e-05, + "loss": 0.4129, + "step": 4560 + }, + { + "epoch": 1.91, + "learning_rate": 1.7801069902171123e-05, + "loss": 0.3759, + "step": 4561 + }, + { + "epoch": 1.91, + "learning_rate": 1.7800009064049416e-05, + "loss": 0.414, + "step": 4562 + }, + { + "epoch": 1.91, + "learning_rate": 1.7798948001723368e-05, + "loss": 0.4868, + "step": 4563 + }, + { + "epoch": 1.91, + "learning_rate": 1.779788671522348e-05, + "loss": 0.3889, + "step": 4564 + }, + { + "epoch": 1.91, + "learning_rate": 1.779682520458026e-05, + "loss": 0.3491, + "step": 4565 + }, + { + "epoch": 1.91, + "learning_rate": 1.779576346982421e-05, + "loss": 0.4432, + "step": 4566 + }, + { + "epoch": 1.91, + "learning_rate": 1.7794701510985867e-05, + "loss": 0.3424, + "step": 4567 + }, + { + "epoch": 1.91, + "learning_rate": 1.779363932809574e-05, + "loss": 0.3825, + "step": 4568 + }, + { + "epoch": 1.91, + "learning_rate": 1.7792576921184374e-05, + "loss": 0.3551, + "step": 4569 + }, + { + "epoch": 1.91, + "learning_rate": 1.7791514290282293e-05, + "loss": 0.3705, + "step": 4570 + }, + { + "epoch": 1.91, + "learning_rate": 1.7790451435420046e-05, + "loss": 0.3369, + "step": 4571 + }, + { + "epoch": 1.91, + "learning_rate": 1.778938835662819e-05, + "loss": 0.402, + "step": 4572 + }, + { + "epoch": 1.91, + "learning_rate": 1.7788325053937278e-05, + "loss": 0.3878, + "step": 4573 + }, + { + "epoch": 1.92, + "learning_rate": 1.778726152737787e-05, + "loss": 0.4031, + "step": 4574 + }, + { + "epoch": 1.92, + "learning_rate": 1.7786197776980542e-05, + "loss": 0.384, + "step": 4575 + }, + { + "epoch": 1.92, + "learning_rate": 1.778513380277587e-05, + "loss": 0.3849, + "step": 4576 + }, + { + "epoch": 1.92, + "learning_rate": 1.7784069604794433e-05, + "loss": 0.3563, + "step": 4577 + }, + { + "epoch": 1.92, + "learning_rate": 1.778300518306682e-05, + "loss": 0.4222, + "step": 4578 + }, + { + "epoch": 1.92, + "learning_rate": 1.7781940537623636e-05, + "loss": 0.3769, + "step": 4579 + }, + { + "epoch": 1.92, + "learning_rate": 1.778087566849547e-05, + "loss": 0.3815, + "step": 4580 + }, + { + "epoch": 1.92, + "learning_rate": 1.7779810575712942e-05, + "loss": 0.3749, + "step": 4581 + }, + { + "epoch": 1.92, + "learning_rate": 1.777874525930666e-05, + "loss": 0.3943, + "step": 4582 + }, + { + "epoch": 1.92, + "learning_rate": 1.7777679719307253e-05, + "loss": 0.3639, + "step": 4583 + }, + { + "epoch": 1.92, + "learning_rate": 1.7776613955745337e-05, + "loss": 0.3701, + "step": 4584 + }, + { + "epoch": 1.92, + "learning_rate": 1.7775547968651562e-05, + "loss": 0.3833, + "step": 4585 + }, + { + "epoch": 1.92, + "learning_rate": 1.7774481758056553e-05, + "loss": 0.3607, + "step": 4586 + }, + { + "epoch": 1.92, + "learning_rate": 1.777341532399097e-05, + "loss": 0.3518, + "step": 4587 + }, + { + "epoch": 1.92, + "learning_rate": 1.777234866648546e-05, + "loss": 0.4178, + "step": 4588 + }, + { + "epoch": 1.92, + "learning_rate": 1.7771281785570683e-05, + "loss": 0.4101, + "step": 4589 + }, + { + "epoch": 1.92, + "learning_rate": 1.777021468127731e-05, + "loss": 0.3854, + "step": 4590 + }, + { + "epoch": 1.92, + "learning_rate": 1.776914735363601e-05, + "loss": 0.3975, + "step": 4591 + }, + { + "epoch": 1.92, + "learning_rate": 1.776807980267746e-05, + "loss": 0.409, + "step": 4592 + }, + { + "epoch": 1.92, + "learning_rate": 1.7767012028432352e-05, + "loss": 0.4173, + "step": 4593 + }, + { + "epoch": 1.92, + "learning_rate": 1.776594403093138e-05, + "loss": 0.3405, + "step": 4594 + }, + { + "epoch": 1.92, + "learning_rate": 1.7764875810205234e-05, + "loss": 0.3914, + "step": 4595 + }, + { + "epoch": 1.92, + "learning_rate": 1.776380736628462e-05, + "loss": 0.3719, + "step": 4596 + }, + { + "epoch": 1.92, + "learning_rate": 1.7762738699200256e-05, + "loss": 0.4169, + "step": 4597 + }, + { + "epoch": 1.93, + "learning_rate": 1.7761669808982857e-05, + "loss": 0.4051, + "step": 4598 + }, + { + "epoch": 1.93, + "learning_rate": 1.776060069566315e-05, + "loss": 0.3898, + "step": 4599 + }, + { + "epoch": 1.93, + "learning_rate": 1.7759531359271857e-05, + "loss": 0.3679, + "step": 4600 + }, + { + "epoch": 1.93, + "learning_rate": 1.7758461799839723e-05, + "loss": 0.3402, + "step": 4601 + }, + { + "epoch": 1.93, + "learning_rate": 1.7757392017397493e-05, + "loss": 0.391, + "step": 4602 + }, + { + "epoch": 1.93, + "learning_rate": 1.7756322011975906e-05, + "loss": 0.3498, + "step": 4603 + }, + { + "epoch": 1.93, + "learning_rate": 1.7755251783605733e-05, + "loss": 0.3862, + "step": 4604 + }, + { + "epoch": 1.93, + "learning_rate": 1.7754181332317725e-05, + "loss": 0.4389, + "step": 4605 + }, + { + "epoch": 1.93, + "learning_rate": 1.7753110658142657e-05, + "loss": 0.3772, + "step": 4606 + }, + { + "epoch": 1.93, + "learning_rate": 1.77520397611113e-05, + "loss": 0.4095, + "step": 4607 + }, + { + "epoch": 1.93, + "learning_rate": 1.7750968641254444e-05, + "loss": 0.4164, + "step": 4608 + }, + { + "epoch": 1.93, + "learning_rate": 1.7749897298602867e-05, + "loss": 0.4049, + "step": 4609 + }, + { + "epoch": 1.93, + "learning_rate": 1.7748825733187372e-05, + "loss": 0.3771, + "step": 4610 + }, + { + "epoch": 1.93, + "learning_rate": 1.774775394503876e-05, + "loss": 0.3696, + "step": 4611 + }, + { + "epoch": 1.93, + "learning_rate": 1.7746681934187832e-05, + "loss": 0.3992, + "step": 4612 + }, + { + "epoch": 1.93, + "learning_rate": 1.7745609700665406e-05, + "loss": 0.3612, + "step": 4613 + }, + { + "epoch": 1.93, + "learning_rate": 1.774453724450231e-05, + "loss": 0.3978, + "step": 4614 + }, + { + "epoch": 1.93, + "learning_rate": 1.7743464565729354e-05, + "loss": 0.4378, + "step": 4615 + }, + { + "epoch": 1.93, + "learning_rate": 1.7742391664377384e-05, + "loss": 0.3375, + "step": 4616 + }, + { + "epoch": 1.93, + "learning_rate": 1.7741318540477234e-05, + "loss": 0.4127, + "step": 4617 + }, + { + "epoch": 1.93, + "learning_rate": 1.7740245194059754e-05, + "loss": 0.3915, + "step": 4618 + }, + { + "epoch": 1.93, + "learning_rate": 1.77391716251558e-05, + "loss": 0.3586, + "step": 4619 + }, + { + "epoch": 1.93, + "learning_rate": 1.7738097833796218e-05, + "loss": 0.3378, + "step": 4620 + }, + { + "epoch": 1.93, + "learning_rate": 1.7737023820011884e-05, + "loss": 0.4145, + "step": 4621 + }, + { + "epoch": 1.94, + "learning_rate": 1.7735949583833662e-05, + "loss": 0.4131, + "step": 4622 + }, + { + "epoch": 1.94, + "learning_rate": 1.7734875125292436e-05, + "loss": 0.3654, + "step": 4623 + }, + { + "epoch": 1.94, + "learning_rate": 1.7733800444419088e-05, + "loss": 0.3907, + "step": 4624 + }, + { + "epoch": 1.94, + "learning_rate": 1.773272554124451e-05, + "loss": 0.4214, + "step": 4625 + }, + { + "epoch": 1.94, + "learning_rate": 1.77316504157996e-05, + "loss": 0.3629, + "step": 4626 + }, + { + "epoch": 1.94, + "learning_rate": 1.773057506811526e-05, + "loss": 0.3772, + "step": 4627 + }, + { + "epoch": 1.94, + "learning_rate": 1.7729499498222392e-05, + "loss": 0.4109, + "step": 4628 + }, + { + "epoch": 1.94, + "learning_rate": 1.7728423706151928e-05, + "loss": 0.36, + "step": 4629 + }, + { + "epoch": 1.94, + "learning_rate": 1.772734769193478e-05, + "loss": 0.436, + "step": 4630 + }, + { + "epoch": 1.94, + "learning_rate": 1.772627145560188e-05, + "loss": 0.3648, + "step": 4631 + }, + { + "epoch": 1.94, + "learning_rate": 1.772519499718416e-05, + "loss": 0.4065, + "step": 4632 + }, + { + "epoch": 1.94, + "learning_rate": 1.772411831671257e-05, + "loss": 0.387, + "step": 4633 + }, + { + "epoch": 1.94, + "learning_rate": 1.7723041414218053e-05, + "loss": 0.3432, + "step": 4634 + }, + { + "epoch": 1.94, + "learning_rate": 1.7721964289731562e-05, + "loss": 0.4029, + "step": 4635 + }, + { + "epoch": 1.94, + "learning_rate": 1.772088694328406e-05, + "loss": 0.3698, + "step": 4636 + }, + { + "epoch": 1.94, + "learning_rate": 1.771980937490651e-05, + "loss": 0.4266, + "step": 4637 + }, + { + "epoch": 1.94, + "learning_rate": 1.7718731584629897e-05, + "loss": 0.3723, + "step": 4638 + }, + { + "epoch": 1.94, + "learning_rate": 1.771765357248519e-05, + "loss": 0.3856, + "step": 4639 + }, + { + "epoch": 1.94, + "learning_rate": 1.7716575338503382e-05, + "loss": 0.3812, + "step": 4640 + }, + { + "epoch": 1.94, + "learning_rate": 1.7715496882715463e-05, + "loss": 0.353, + "step": 4641 + }, + { + "epoch": 1.94, + "learning_rate": 1.7714418205152434e-05, + "loss": 0.4068, + "step": 4642 + }, + { + "epoch": 1.94, + "learning_rate": 1.7713339305845293e-05, + "loss": 0.4075, + "step": 4643 + }, + { + "epoch": 1.94, + "learning_rate": 1.7712260184825063e-05, + "loss": 0.3755, + "step": 4644 + }, + { + "epoch": 1.94, + "learning_rate": 1.771118084212276e-05, + "loss": 0.4403, + "step": 4645 + }, + { + "epoch": 1.95, + "learning_rate": 1.7710101277769403e-05, + "loss": 0.4173, + "step": 4646 + }, + { + "epoch": 1.95, + "learning_rate": 1.7709021491796027e-05, + "loss": 0.3637, + "step": 4647 + }, + { + "epoch": 1.95, + "learning_rate": 1.770794148423367e-05, + "loss": 0.4326, + "step": 4648 + }, + { + "epoch": 1.95, + "learning_rate": 1.7706861255113376e-05, + "loss": 0.3747, + "step": 4649 + }, + { + "epoch": 1.95, + "learning_rate": 1.7705780804466192e-05, + "loss": 0.3416, + "step": 4650 + }, + { + "epoch": 1.95, + "learning_rate": 1.7704700132323175e-05, + "loss": 0.3899, + "step": 4651 + }, + { + "epoch": 1.95, + "learning_rate": 1.7703619238715392e-05, + "loss": 0.4293, + "step": 4652 + }, + { + "epoch": 1.95, + "learning_rate": 1.770253812367391e-05, + "loss": 0.4505, + "step": 4653 + }, + { + "epoch": 1.95, + "learning_rate": 1.7701456787229805e-05, + "loss": 0.3799, + "step": 4654 + }, + { + "epoch": 1.95, + "learning_rate": 1.770037522941416e-05, + "loss": 0.3909, + "step": 4655 + }, + { + "epoch": 1.95, + "learning_rate": 1.7699293450258062e-05, + "loss": 0.4465, + "step": 4656 + }, + { + "epoch": 1.95, + "learning_rate": 1.7698211449792604e-05, + "loss": 0.3893, + "step": 4657 + }, + { + "epoch": 1.95, + "learning_rate": 1.769712922804889e-05, + "loss": 0.3736, + "step": 4658 + }, + { + "epoch": 1.95, + "learning_rate": 1.7696046785058024e-05, + "loss": 0.3389, + "step": 4659 + }, + { + "epoch": 1.95, + "learning_rate": 1.7694964120851128e-05, + "loss": 0.4159, + "step": 4660 + }, + { + "epoch": 1.95, + "learning_rate": 1.7693881235459315e-05, + "loss": 0.391, + "step": 4661 + }, + { + "epoch": 1.95, + "learning_rate": 1.7692798128913714e-05, + "loss": 0.3976, + "step": 4662 + }, + { + "epoch": 1.95, + "learning_rate": 1.7691714801245453e-05, + "loss": 0.371, + "step": 4663 + }, + { + "epoch": 1.95, + "learning_rate": 1.7690631252485676e-05, + "loss": 0.3931, + "step": 4664 + }, + { + "epoch": 1.95, + "learning_rate": 1.7689547482665532e-05, + "loss": 0.3735, + "step": 4665 + }, + { + "epoch": 1.95, + "learning_rate": 1.768846349181617e-05, + "loss": 0.4045, + "step": 4666 + }, + { + "epoch": 1.95, + "learning_rate": 1.7687379279968745e-05, + "loss": 0.4001, + "step": 4667 + }, + { + "epoch": 1.95, + "learning_rate": 1.7686294847154423e-05, + "loss": 0.3443, + "step": 4668 + }, + { + "epoch": 1.95, + "learning_rate": 1.7685210193404377e-05, + "loss": 0.4312, + "step": 4669 + }, + { + "epoch": 1.96, + "learning_rate": 1.7684125318749785e-05, + "loss": 0.3976, + "step": 4670 + }, + { + "epoch": 1.96, + "learning_rate": 1.7683040223221828e-05, + "loss": 0.4214, + "step": 4671 + }, + { + "epoch": 1.96, + "learning_rate": 1.76819549068517e-05, + "loss": 0.3772, + "step": 4672 + }, + { + "epoch": 1.96, + "learning_rate": 1.768086936967059e-05, + "loss": 0.3983, + "step": 4673 + }, + { + "epoch": 1.96, + "learning_rate": 1.767978361170971e-05, + "loss": 0.4451, + "step": 4674 + }, + { + "epoch": 1.96, + "learning_rate": 1.7678697633000267e-05, + "loss": 0.3974, + "step": 4675 + }, + { + "epoch": 1.96, + "learning_rate": 1.767761143357347e-05, + "loss": 0.3598, + "step": 4676 + }, + { + "epoch": 1.96, + "learning_rate": 1.7676525013460547e-05, + "loss": 0.3909, + "step": 4677 + }, + { + "epoch": 1.96, + "learning_rate": 1.7675438372692726e-05, + "loss": 0.4153, + "step": 4678 + }, + { + "epoch": 1.96, + "learning_rate": 1.7674351511301238e-05, + "loss": 0.3957, + "step": 4679 + }, + { + "epoch": 1.96, + "learning_rate": 1.767326442931733e-05, + "loss": 0.3571, + "step": 4680 + }, + { + "epoch": 1.96, + "learning_rate": 1.7672177126772237e-05, + "loss": 0.3678, + "step": 4681 + }, + { + "epoch": 1.96, + "learning_rate": 1.7671089603697228e-05, + "loss": 0.4258, + "step": 4682 + }, + { + "epoch": 1.96, + "learning_rate": 1.767000186012355e-05, + "loss": 0.4408, + "step": 4683 + }, + { + "epoch": 1.96, + "learning_rate": 1.766891389608248e-05, + "loss": 0.409, + "step": 4684 + }, + { + "epoch": 1.96, + "learning_rate": 1.7667825711605283e-05, + "loss": 0.4514, + "step": 4685 + }, + { + "epoch": 1.96, + "learning_rate": 1.766673730672324e-05, + "loss": 0.3947, + "step": 4686 + }, + { + "epoch": 1.96, + "learning_rate": 1.766564868146764e-05, + "loss": 0.349, + "step": 4687 + }, + { + "epoch": 1.96, + "learning_rate": 1.7664559835869763e-05, + "loss": 0.4218, + "step": 4688 + }, + { + "epoch": 1.96, + "learning_rate": 1.766347076996092e-05, + "loss": 0.4156, + "step": 4689 + }, + { + "epoch": 1.96, + "learning_rate": 1.7662381483772412e-05, + "loss": 0.3596, + "step": 4690 + }, + { + "epoch": 1.96, + "learning_rate": 1.7661291977335544e-05, + "loss": 0.4368, + "step": 4691 + }, + { + "epoch": 1.96, + "learning_rate": 1.766020225068164e-05, + "loss": 0.3567, + "step": 4692 + }, + { + "epoch": 1.96, + "learning_rate": 1.7659112303842015e-05, + "loss": 0.3661, + "step": 4693 + }, + { + "epoch": 1.97, + "learning_rate": 1.7658022136848008e-05, + "loss": 0.3629, + "step": 4694 + }, + { + "epoch": 1.97, + "learning_rate": 1.7656931749730947e-05, + "loss": 0.3877, + "step": 4695 + }, + { + "epoch": 1.97, + "learning_rate": 1.765584114252218e-05, + "loss": 0.3517, + "step": 4696 + }, + { + "epoch": 1.97, + "learning_rate": 1.7654750315253055e-05, + "loss": 0.3632, + "step": 4697 + }, + { + "epoch": 1.97, + "learning_rate": 1.765365926795492e-05, + "loss": 0.3851, + "step": 4698 + }, + { + "epoch": 1.97, + "learning_rate": 1.7652568000659144e-05, + "loss": 0.4156, + "step": 4699 + }, + { + "epoch": 1.97, + "learning_rate": 1.765147651339709e-05, + "loss": 0.3623, + "step": 4700 + }, + { + "epoch": 1.97, + "learning_rate": 1.7650384806200134e-05, + "loss": 0.4243, + "step": 4701 + }, + { + "epoch": 1.97, + "learning_rate": 1.7649292879099655e-05, + "loss": 0.3677, + "step": 4702 + }, + { + "epoch": 1.97, + "learning_rate": 1.764820073212704e-05, + "loss": 0.4255, + "step": 4703 + }, + { + "epoch": 1.97, + "learning_rate": 1.7647108365313683e-05, + "loss": 0.4099, + "step": 4704 + }, + { + "epoch": 1.97, + "learning_rate": 1.7646015778690983e-05, + "loss": 0.4019, + "step": 4705 + }, + { + "epoch": 1.97, + "learning_rate": 1.7644922972290344e-05, + "loss": 0.4247, + "step": 4706 + }, + { + "epoch": 1.97, + "learning_rate": 1.7643829946143176e-05, + "loss": 0.4158, + "step": 4707 + }, + { + "epoch": 1.97, + "learning_rate": 1.76427367002809e-05, + "loss": 0.354, + "step": 4708 + }, + { + "epoch": 1.97, + "learning_rate": 1.764164323473494e-05, + "loss": 0.4356, + "step": 4709 + }, + { + "epoch": 1.97, + "learning_rate": 1.7640549549536727e-05, + "loss": 0.38, + "step": 4710 + }, + { + "epoch": 1.97, + "learning_rate": 1.7639455644717694e-05, + "loss": 0.3709, + "step": 4711 + }, + { + "epoch": 1.97, + "learning_rate": 1.7638361520309294e-05, + "loss": 0.4102, + "step": 4712 + }, + { + "epoch": 1.97, + "learning_rate": 1.7637267176342964e-05, + "loss": 0.3607, + "step": 4713 + }, + { + "epoch": 1.97, + "learning_rate": 1.763617261285017e-05, + "loss": 0.417, + "step": 4714 + }, + { + "epoch": 1.97, + "learning_rate": 1.763507782986237e-05, + "loss": 0.3558, + "step": 4715 + }, + { + "epoch": 1.97, + "learning_rate": 1.763398282741103e-05, + "loss": 0.381, + "step": 4716 + }, + { + "epoch": 1.97, + "learning_rate": 1.763288760552763e-05, + "loss": 0.3835, + "step": 4717 + }, + { + "epoch": 1.98, + "learning_rate": 1.763179216424365e-05, + "loss": 0.4098, + "step": 4718 + }, + { + "epoch": 1.98, + "learning_rate": 1.7630696503590577e-05, + "loss": 0.3972, + "step": 4719 + }, + { + "epoch": 1.98, + "learning_rate": 1.7629600623599904e-05, + "loss": 0.4298, + "step": 4720 + }, + { + "epoch": 1.98, + "learning_rate": 1.7628504524303128e-05, + "loss": 0.382, + "step": 4721 + }, + { + "epoch": 1.98, + "learning_rate": 1.7627408205731762e-05, + "loss": 0.43, + "step": 4722 + }, + { + "epoch": 1.98, + "learning_rate": 1.7626311667917315e-05, + "loss": 0.3753, + "step": 4723 + }, + { + "epoch": 1.98, + "learning_rate": 1.7625214910891307e-05, + "loss": 0.4242, + "step": 4724 + }, + { + "epoch": 1.98, + "learning_rate": 1.7624117934685264e-05, + "loss": 0.4101, + "step": 4725 + }, + { + "epoch": 1.98, + "learning_rate": 1.7623020739330714e-05, + "loss": 0.3907, + "step": 4726 + }, + { + "epoch": 1.98, + "learning_rate": 1.7621923324859196e-05, + "loss": 0.3705, + "step": 4727 + }, + { + "epoch": 1.98, + "learning_rate": 1.762082569130226e-05, + "loss": 0.3592, + "step": 4728 + }, + { + "epoch": 1.98, + "learning_rate": 1.761972783869145e-05, + "loss": 0.401, + "step": 4729 + }, + { + "epoch": 1.98, + "learning_rate": 1.7618629767058326e-05, + "loss": 0.4288, + "step": 4730 + }, + { + "epoch": 1.98, + "learning_rate": 1.7617531476434447e-05, + "loss": 0.3446, + "step": 4731 + }, + { + "epoch": 1.98, + "learning_rate": 1.7616432966851386e-05, + "loss": 0.3834, + "step": 4732 + }, + { + "epoch": 1.98, + "learning_rate": 1.761533423834072e-05, + "loss": 0.4065, + "step": 4733 + }, + { + "epoch": 1.98, + "learning_rate": 1.761423529093403e-05, + "loss": 0.4022, + "step": 4734 + }, + { + "epoch": 1.98, + "learning_rate": 1.76131361246629e-05, + "loss": 0.3807, + "step": 4735 + }, + { + "epoch": 1.98, + "learning_rate": 1.7612036739558932e-05, + "loss": 0.3898, + "step": 4736 + }, + { + "epoch": 1.98, + "learning_rate": 1.761093713565372e-05, + "loss": 0.3873, + "step": 4737 + }, + { + "epoch": 1.98, + "learning_rate": 1.7609837312978874e-05, + "loss": 0.4046, + "step": 4738 + }, + { + "epoch": 1.98, + "learning_rate": 1.7608737271566004e-05, + "loss": 0.4069, + "step": 4739 + }, + { + "epoch": 1.98, + "learning_rate": 1.7607637011446737e-05, + "loss": 0.3928, + "step": 4740 + }, + { + "epoch": 1.98, + "learning_rate": 1.7606536532652695e-05, + "loss": 0.3691, + "step": 4741 + }, + { + "epoch": 1.99, + "learning_rate": 1.760543583521551e-05, + "loss": 0.4312, + "step": 4742 + }, + { + "epoch": 1.99, + "learning_rate": 1.7604334919166822e-05, + "loss": 0.3853, + "step": 4743 + }, + { + "epoch": 1.99, + "learning_rate": 1.7603233784538275e-05, + "loss": 0.4136, + "step": 4744 + }, + { + "epoch": 1.99, + "learning_rate": 1.7602132431361516e-05, + "loss": 0.4116, + "step": 4745 + }, + { + "epoch": 1.99, + "learning_rate": 1.7601030859668207e-05, + "loss": 0.3834, + "step": 4746 + }, + { + "epoch": 1.99, + "learning_rate": 1.7599929069490014e-05, + "loss": 0.4323, + "step": 4747 + }, + { + "epoch": 1.99, + "learning_rate": 1.7598827060858605e-05, + "loss": 0.4095, + "step": 4748 + }, + { + "epoch": 1.99, + "learning_rate": 1.759772483380565e-05, + "loss": 0.4095, + "step": 4749 + }, + { + "epoch": 1.99, + "learning_rate": 1.759662238836284e-05, + "loss": 0.4272, + "step": 4750 + }, + { + "epoch": 1.99, + "learning_rate": 1.759551972456186e-05, + "loss": 0.405, + "step": 4751 + }, + { + "epoch": 1.99, + "learning_rate": 1.7594416842434405e-05, + "loss": 0.3806, + "step": 4752 + }, + { + "epoch": 1.99, + "learning_rate": 1.759331374201218e-05, + "loss": 0.3747, + "step": 4753 + }, + { + "epoch": 1.99, + "learning_rate": 1.759221042332689e-05, + "loss": 0.3804, + "step": 4754 + }, + { + "epoch": 1.99, + "learning_rate": 1.759110688641024e-05, + "loss": 0.4165, + "step": 4755 + }, + { + "epoch": 1.99, + "learning_rate": 1.7590003131293967e-05, + "loss": 0.3855, + "step": 4756 + }, + { + "epoch": 1.99, + "learning_rate": 1.7588899158009786e-05, + "loss": 0.373, + "step": 4757 + }, + { + "epoch": 1.99, + "learning_rate": 1.7587794966589437e-05, + "loss": 0.3976, + "step": 4758 + }, + { + "epoch": 1.99, + "learning_rate": 1.758669055706465e-05, + "loss": 0.3288, + "step": 4759 + }, + { + "epoch": 1.99, + "learning_rate": 1.758558592946718e-05, + "loss": 0.3985, + "step": 4760 + }, + { + "epoch": 1.99, + "learning_rate": 1.7584481083828766e-05, + "loss": 0.4002, + "step": 4761 + }, + { + "epoch": 1.99, + "learning_rate": 1.7583376020181183e-05, + "loss": 0.3729, + "step": 4762 + }, + { + "epoch": 1.99, + "learning_rate": 1.758227073855618e-05, + "loss": 0.4247, + "step": 4763 + }, + { + "epoch": 1.99, + "learning_rate": 1.7581165238985533e-05, + "loss": 0.4128, + "step": 4764 + }, + { + "epoch": 1.99, + "learning_rate": 1.7580059521501017e-05, + "loss": 0.3545, + "step": 4765 + }, + { + "epoch": 2.0, + "learning_rate": 1.7578953586134423e-05, + "loss": 0.3787, + "step": 4766 + }, + { + "epoch": 2.0, + "learning_rate": 1.7577847432917528e-05, + "loss": 0.3957, + "step": 4767 + }, + { + "epoch": 2.0, + "learning_rate": 1.7576741061882136e-05, + "loss": 0.3975, + "step": 4768 + }, + { + "epoch": 2.0, + "learning_rate": 1.7575634473060045e-05, + "loss": 0.3689, + "step": 4769 + }, + { + "epoch": 2.0, + "learning_rate": 1.757452766648306e-05, + "loss": 0.4133, + "step": 4770 + }, + { + "epoch": 2.0, + "learning_rate": 1.7573420642183004e-05, + "loss": 0.3487, + "step": 4771 + }, + { + "epoch": 2.0, + "learning_rate": 1.757231340019169e-05, + "loss": 0.3746, + "step": 4772 + }, + { + "epoch": 2.0, + "learning_rate": 1.7571205940540945e-05, + "loss": 0.4144, + "step": 4773 + }, + { + "epoch": 2.0, + "learning_rate": 1.7570098263262608e-05, + "loss": 0.3845, + "step": 4774 + }, + { + "epoch": 2.0, + "learning_rate": 1.756899036838851e-05, + "loss": 0.4326, + "step": 4775 + }, + { + "epoch": 2.0, + "learning_rate": 1.7567882255950503e-05, + "loss": 0.3347, + "step": 4776 + }, + { + "epoch": 2.0, + "learning_rate": 1.7566773925980435e-05, + "loss": 0.3743, + "step": 4777 + }, + { + "epoch": 2.0, + "learning_rate": 1.7565665378510165e-05, + "loss": 0.2799, + "step": 4778 + }, + { + "epoch": 2.0, + "learning_rate": 1.756455661357156e-05, + "loss": 0.315, + "step": 4779 + }, + { + "epoch": 2.0, + "learning_rate": 1.7563447631196484e-05, + "loss": 0.2846, + "step": 4780 + }, + { + "epoch": 2.0, + "learning_rate": 1.7562338431416818e-05, + "loss": 0.262, + "step": 4781 + }, + { + "epoch": 2.0, + "learning_rate": 1.7561229014264448e-05, + "loss": 0.2669, + "step": 4782 + }, + { + "epoch": 2.0, + "learning_rate": 1.7560119379771257e-05, + "loss": 0.2785, + "step": 4783 + }, + { + "epoch": 2.0, + "learning_rate": 1.7559009527969145e-05, + "loss": 0.27, + "step": 4784 + }, + { + "epoch": 2.0, + "learning_rate": 1.755789945889001e-05, + "loss": 0.2426, + "step": 4785 + }, + { + "epoch": 2.0, + "learning_rate": 1.7556789172565762e-05, + "loss": 0.266, + "step": 4786 + }, + { + "epoch": 2.0, + "learning_rate": 1.7555678669028315e-05, + "loss": 0.242, + "step": 4787 + }, + { + "epoch": 2.0, + "learning_rate": 1.755456794830959e-05, + "loss": 0.246, + "step": 4788 + }, + { + "epoch": 2.01, + "learning_rate": 1.7553457010441514e-05, + "loss": 0.2788, + "step": 4789 + }, + { + "epoch": 2.01, + "learning_rate": 1.7552345855456017e-05, + "loss": 0.2595, + "step": 4790 + }, + { + "epoch": 2.01, + "learning_rate": 1.755123448338504e-05, + "loss": 0.2886, + "step": 4791 + }, + { + "epoch": 2.01, + "learning_rate": 1.755012289426053e-05, + "loss": 0.2787, + "step": 4792 + }, + { + "epoch": 2.01, + "learning_rate": 1.754901108811444e-05, + "loss": 0.2442, + "step": 4793 + }, + { + "epoch": 2.01, + "learning_rate": 1.754789906497872e-05, + "loss": 0.2656, + "step": 4794 + }, + { + "epoch": 2.01, + "learning_rate": 1.754678682488534e-05, + "loss": 0.2701, + "step": 4795 + }, + { + "epoch": 2.01, + "learning_rate": 1.7545674367866273e-05, + "loss": 0.2597, + "step": 4796 + }, + { + "epoch": 2.01, + "learning_rate": 1.754456169395349e-05, + "loss": 0.2762, + "step": 4797 + }, + { + "epoch": 2.01, + "learning_rate": 1.7543448803178975e-05, + "loss": 0.2363, + "step": 4798 + }, + { + "epoch": 2.01, + "learning_rate": 1.754233569557472e-05, + "loss": 0.2452, + "step": 4799 + }, + { + "epoch": 2.01, + "learning_rate": 1.7541222371172715e-05, + "loss": 0.2096, + "step": 4800 + }, + { + "epoch": 2.01, + "learning_rate": 1.754010883000497e-05, + "loss": 0.264, + "step": 4801 + }, + { + "epoch": 2.01, + "learning_rate": 1.7538995072103485e-05, + "loss": 0.2442, + "step": 4802 + }, + { + "epoch": 2.01, + "learning_rate": 1.7537881097500277e-05, + "loss": 0.2588, + "step": 4803 + }, + { + "epoch": 2.01, + "learning_rate": 1.7536766906227365e-05, + "loss": 0.2637, + "step": 4804 + }, + { + "epoch": 2.01, + "learning_rate": 1.753565249831678e-05, + "loss": 0.255, + "step": 4805 + }, + { + "epoch": 2.01, + "learning_rate": 1.7534537873800544e-05, + "loss": 0.2544, + "step": 4806 + }, + { + "epoch": 2.01, + "learning_rate": 1.7533423032710705e-05, + "loss": 0.2545, + "step": 4807 + }, + { + "epoch": 2.01, + "learning_rate": 1.753230797507931e-05, + "loss": 0.2724, + "step": 4808 + }, + { + "epoch": 2.01, + "learning_rate": 1.7531192700938406e-05, + "loss": 0.2539, + "step": 4809 + }, + { + "epoch": 2.01, + "learning_rate": 1.7530077210320047e-05, + "loss": 0.2681, + "step": 4810 + }, + { + "epoch": 2.01, + "learning_rate": 1.7528961503256302e-05, + "loss": 0.2636, + "step": 4811 + }, + { + "epoch": 2.01, + "learning_rate": 1.752784557977924e-05, + "loss": 0.2572, + "step": 4812 + }, + { + "epoch": 2.02, + "learning_rate": 1.752672943992094e-05, + "loss": 0.27, + "step": 4813 + }, + { + "epoch": 2.02, + "learning_rate": 1.752561308371348e-05, + "loss": 0.2464, + "step": 4814 + }, + { + "epoch": 2.02, + "learning_rate": 1.7524496511188948e-05, + "loss": 0.2508, + "step": 4815 + }, + { + "epoch": 2.02, + "learning_rate": 1.752337972237944e-05, + "loss": 0.2722, + "step": 4816 + }, + { + "epoch": 2.02, + "learning_rate": 1.752226271731706e-05, + "loss": 0.2406, + "step": 4817 + }, + { + "epoch": 2.02, + "learning_rate": 1.7521145496033912e-05, + "loss": 0.258, + "step": 4818 + }, + { + "epoch": 2.02, + "learning_rate": 1.752002805856211e-05, + "loss": 0.2664, + "step": 4819 + }, + { + "epoch": 2.02, + "learning_rate": 1.7518910404933777e-05, + "loss": 0.2426, + "step": 4820 + }, + { + "epoch": 2.02, + "learning_rate": 1.751779253518104e-05, + "loss": 0.2263, + "step": 4821 + }, + { + "epoch": 2.02, + "learning_rate": 1.751667444933602e-05, + "loss": 0.2425, + "step": 4822 + }, + { + "epoch": 2.02, + "learning_rate": 1.7515556147430867e-05, + "loss": 0.2282, + "step": 4823 + }, + { + "epoch": 2.02, + "learning_rate": 1.751443762949772e-05, + "loss": 0.2368, + "step": 4824 + }, + { + "epoch": 2.02, + "learning_rate": 1.7513318895568734e-05, + "loss": 0.2329, + "step": 4825 + }, + { + "epoch": 2.02, + "learning_rate": 1.7512199945676065e-05, + "loss": 0.2456, + "step": 4826 + }, + { + "epoch": 2.02, + "learning_rate": 1.751108077985187e-05, + "loss": 0.2523, + "step": 4827 + }, + { + "epoch": 2.02, + "learning_rate": 1.7509961398128325e-05, + "loss": 0.2661, + "step": 4828 + }, + { + "epoch": 2.02, + "learning_rate": 1.7508841800537608e-05, + "loss": 0.2412, + "step": 4829 + }, + { + "epoch": 2.02, + "learning_rate": 1.750772198711189e-05, + "loss": 0.242, + "step": 4830 + }, + { + "epoch": 2.02, + "learning_rate": 1.750660195788337e-05, + "loss": 0.2648, + "step": 4831 + }, + { + "epoch": 2.02, + "learning_rate": 1.7505481712884235e-05, + "loss": 0.2752, + "step": 4832 + }, + { + "epoch": 2.02, + "learning_rate": 1.750436125214669e-05, + "loss": 0.2557, + "step": 4833 + }, + { + "epoch": 2.02, + "learning_rate": 1.750324057570294e-05, + "loss": 0.2343, + "step": 4834 + }, + { + "epoch": 2.02, + "learning_rate": 1.75021196835852e-05, + "loss": 0.2432, + "step": 4835 + }, + { + "epoch": 2.02, + "learning_rate": 1.750099857582568e-05, + "loss": 0.2425, + "step": 4836 + }, + { + "epoch": 2.03, + "learning_rate": 1.7499877252456624e-05, + "loss": 0.234, + "step": 4837 + }, + { + "epoch": 2.03, + "learning_rate": 1.7498755713510244e-05, + "loss": 0.2568, + "step": 4838 + }, + { + "epoch": 2.03, + "learning_rate": 1.7497633959018784e-05, + "loss": 0.2783, + "step": 4839 + }, + { + "epoch": 2.03, + "learning_rate": 1.749651198901449e-05, + "loss": 0.2693, + "step": 4840 + }, + { + "epoch": 2.03, + "learning_rate": 1.7495389803529615e-05, + "loss": 0.2427, + "step": 4841 + }, + { + "epoch": 2.03, + "learning_rate": 1.749426740259641e-05, + "loss": 0.2645, + "step": 4842 + }, + { + "epoch": 2.03, + "learning_rate": 1.7493144786247136e-05, + "loss": 0.2245, + "step": 4843 + }, + { + "epoch": 2.03, + "learning_rate": 1.7492021954514068e-05, + "loss": 0.2344, + "step": 4844 + }, + { + "epoch": 2.03, + "learning_rate": 1.7490898907429473e-05, + "loss": 0.2408, + "step": 4845 + }, + { + "epoch": 2.03, + "learning_rate": 1.748977564502564e-05, + "loss": 0.2739, + "step": 4846 + }, + { + "epoch": 2.03, + "learning_rate": 1.748865216733485e-05, + "loss": 0.2663, + "step": 4847 + }, + { + "epoch": 2.03, + "learning_rate": 1.7487528474389403e-05, + "loss": 0.2342, + "step": 4848 + }, + { + "epoch": 2.03, + "learning_rate": 1.748640456622159e-05, + "loss": 0.2576, + "step": 4849 + }, + { + "epoch": 2.03, + "learning_rate": 1.7485280442863718e-05, + "loss": 0.2432, + "step": 4850 + }, + { + "epoch": 2.03, + "learning_rate": 1.7484156104348107e-05, + "loss": 0.2371, + "step": 4851 + }, + { + "epoch": 2.03, + "learning_rate": 1.748303155070707e-05, + "loss": 0.2528, + "step": 4852 + }, + { + "epoch": 2.03, + "learning_rate": 1.7481906781972932e-05, + "loss": 0.2573, + "step": 4853 + }, + { + "epoch": 2.03, + "learning_rate": 1.748078179817802e-05, + "loss": 0.2821, + "step": 4854 + }, + { + "epoch": 2.03, + "learning_rate": 1.7479656599354677e-05, + "loss": 0.2348, + "step": 4855 + }, + { + "epoch": 2.03, + "learning_rate": 1.7478531185535235e-05, + "loss": 0.2659, + "step": 4856 + }, + { + "epoch": 2.03, + "learning_rate": 1.747740555675206e-05, + "loss": 0.2252, + "step": 4857 + }, + { + "epoch": 2.03, + "learning_rate": 1.747627971303749e-05, + "loss": 0.2632, + "step": 4858 + }, + { + "epoch": 2.03, + "learning_rate": 1.74751536544239e-05, + "loss": 0.2315, + "step": 4859 + }, + { + "epoch": 2.03, + "learning_rate": 1.747402738094365e-05, + "loss": 0.2415, + "step": 4860 + }, + { + "epoch": 2.04, + "learning_rate": 1.7472900892629114e-05, + "loss": 0.234, + "step": 4861 + }, + { + "epoch": 2.04, + "learning_rate": 1.747177418951267e-05, + "loss": 0.2695, + "step": 4862 + }, + { + "epoch": 2.04, + "learning_rate": 1.7470647271626712e-05, + "loss": 0.2861, + "step": 4863 + }, + { + "epoch": 2.04, + "learning_rate": 1.7469520139003627e-05, + "loss": 0.2673, + "step": 4864 + }, + { + "epoch": 2.04, + "learning_rate": 1.7468392791675812e-05, + "loss": 0.254, + "step": 4865 + }, + { + "epoch": 2.04, + "learning_rate": 1.7467265229675673e-05, + "loss": 0.2536, + "step": 4866 + }, + { + "epoch": 2.04, + "learning_rate": 1.7466137453035626e-05, + "loss": 0.2572, + "step": 4867 + }, + { + "epoch": 2.04, + "learning_rate": 1.746500946178808e-05, + "loss": 0.2533, + "step": 4868 + }, + { + "epoch": 2.04, + "learning_rate": 1.746388125596546e-05, + "loss": 0.2277, + "step": 4869 + }, + { + "epoch": 2.04, + "learning_rate": 1.7462752835600198e-05, + "loss": 0.226, + "step": 4870 + }, + { + "epoch": 2.04, + "learning_rate": 1.746162420072473e-05, + "loss": 0.2779, + "step": 4871 + }, + { + "epoch": 2.04, + "learning_rate": 1.7460495351371492e-05, + "loss": 0.2378, + "step": 4872 + }, + { + "epoch": 2.04, + "learning_rate": 1.745936628757294e-05, + "loss": 0.2827, + "step": 4873 + }, + { + "epoch": 2.04, + "learning_rate": 1.745823700936152e-05, + "loss": 0.2935, + "step": 4874 + }, + { + "epoch": 2.04, + "learning_rate": 1.7457107516769697e-05, + "loss": 0.2524, + "step": 4875 + }, + { + "epoch": 2.04, + "learning_rate": 1.7455977809829933e-05, + "loss": 0.2445, + "step": 4876 + }, + { + "epoch": 2.04, + "learning_rate": 1.7454847888574705e-05, + "loss": 0.2442, + "step": 4877 + }, + { + "epoch": 2.04, + "learning_rate": 1.745371775303649e-05, + "loss": 0.255, + "step": 4878 + }, + { + "epoch": 2.04, + "learning_rate": 1.7452587403247772e-05, + "loss": 0.2344, + "step": 4879 + }, + { + "epoch": 2.04, + "learning_rate": 1.7451456839241044e-05, + "loss": 0.2695, + "step": 4880 + }, + { + "epoch": 2.04, + "learning_rate": 1.7450326061048798e-05, + "loss": 0.2985, + "step": 4881 + }, + { + "epoch": 2.04, + "learning_rate": 1.7449195068703542e-05, + "loss": 0.2883, + "step": 4882 + }, + { + "epoch": 2.04, + "learning_rate": 1.7448063862237784e-05, + "loss": 0.2591, + "step": 4883 + }, + { + "epoch": 2.04, + "learning_rate": 1.7446932441684044e-05, + "loss": 0.2699, + "step": 4884 + }, + { + "epoch": 2.05, + "learning_rate": 1.7445800807074834e-05, + "loss": 0.2448, + "step": 4885 + }, + { + "epoch": 2.05, + "learning_rate": 1.744466895844269e-05, + "loss": 0.2717, + "step": 4886 + }, + { + "epoch": 2.05, + "learning_rate": 1.7443536895820143e-05, + "loss": 0.2549, + "step": 4887 + }, + { + "epoch": 2.05, + "learning_rate": 1.744240461923973e-05, + "loss": 0.261, + "step": 4888 + }, + { + "epoch": 2.05, + "learning_rate": 1.7441272128734004e-05, + "loss": 0.2292, + "step": 4889 + }, + { + "epoch": 2.05, + "learning_rate": 1.7440139424335513e-05, + "loss": 0.2494, + "step": 4890 + }, + { + "epoch": 2.05, + "learning_rate": 1.7439006506076816e-05, + "loss": 0.2344, + "step": 4891 + }, + { + "epoch": 2.05, + "learning_rate": 1.7437873373990478e-05, + "loss": 0.2476, + "step": 4892 + }, + { + "epoch": 2.05, + "learning_rate": 1.743674002810907e-05, + "loss": 0.2553, + "step": 4893 + }, + { + "epoch": 2.05, + "learning_rate": 1.743560646846517e-05, + "loss": 0.2397, + "step": 4894 + }, + { + "epoch": 2.05, + "learning_rate": 1.7434472695091358e-05, + "loss": 0.2433, + "step": 4895 + }, + { + "epoch": 2.05, + "learning_rate": 1.743333870802023e-05, + "loss": 0.2665, + "step": 4896 + }, + { + "epoch": 2.05, + "learning_rate": 1.7432204507284373e-05, + "loss": 0.2498, + "step": 4897 + }, + { + "epoch": 2.05, + "learning_rate": 1.7431070092916395e-05, + "loss": 0.2239, + "step": 4898 + }, + { + "epoch": 2.05, + "learning_rate": 1.74299354649489e-05, + "loss": 0.2373, + "step": 4899 + }, + { + "epoch": 2.05, + "learning_rate": 1.7428800623414502e-05, + "loss": 0.246, + "step": 4900 + }, + { + "epoch": 2.05, + "learning_rate": 1.7427665568345826e-05, + "loss": 0.2617, + "step": 4901 + }, + { + "epoch": 2.05, + "learning_rate": 1.7426530299775492e-05, + "loss": 0.247, + "step": 4902 + }, + { + "epoch": 2.05, + "learning_rate": 1.7425394817736135e-05, + "loss": 0.2407, + "step": 4903 + }, + { + "epoch": 2.05, + "learning_rate": 1.7424259122260396e-05, + "loss": 0.2377, + "step": 4904 + }, + { + "epoch": 2.05, + "learning_rate": 1.7423123213380914e-05, + "loss": 0.2585, + "step": 4905 + }, + { + "epoch": 2.05, + "learning_rate": 1.742198709113034e-05, + "loss": 0.2342, + "step": 4906 + }, + { + "epoch": 2.05, + "learning_rate": 1.7420850755541335e-05, + "loss": 0.2528, + "step": 4907 + }, + { + "epoch": 2.05, + "learning_rate": 1.7419714206646563e-05, + "loss": 0.2572, + "step": 4908 + }, + { + "epoch": 2.06, + "learning_rate": 1.741857744447869e-05, + "loss": 0.2499, + "step": 4909 + }, + { + "epoch": 2.06, + "learning_rate": 1.7417440469070386e-05, + "loss": 0.2417, + "step": 4910 + }, + { + "epoch": 2.06, + "learning_rate": 1.7416303280454343e-05, + "loss": 0.2383, + "step": 4911 + }, + { + "epoch": 2.06, + "learning_rate": 1.741516587866324e-05, + "loss": 0.2331, + "step": 4912 + }, + { + "epoch": 2.06, + "learning_rate": 1.7414028263729776e-05, + "loss": 0.2658, + "step": 4913 + }, + { + "epoch": 2.06, + "learning_rate": 1.741289043568665e-05, + "loss": 0.2369, + "step": 4914 + }, + { + "epoch": 2.06, + "learning_rate": 1.7411752394566565e-05, + "loss": 0.279, + "step": 4915 + }, + { + "epoch": 2.06, + "learning_rate": 1.7410614140402234e-05, + "loss": 0.2687, + "step": 4916 + }, + { + "epoch": 2.06, + "learning_rate": 1.7409475673226376e-05, + "loss": 0.2613, + "step": 4917 + }, + { + "epoch": 2.06, + "learning_rate": 1.7408336993071716e-05, + "loss": 0.2657, + "step": 4918 + }, + { + "epoch": 2.06, + "learning_rate": 1.7407198099970982e-05, + "loss": 0.2621, + "step": 4919 + }, + { + "epoch": 2.06, + "learning_rate": 1.7406058993956908e-05, + "loss": 0.2789, + "step": 4920 + }, + { + "epoch": 2.06, + "learning_rate": 1.7404919675062245e-05, + "loss": 0.2448, + "step": 4921 + }, + { + "epoch": 2.06, + "learning_rate": 1.7403780143319736e-05, + "loss": 0.2405, + "step": 4922 + }, + { + "epoch": 2.06, + "learning_rate": 1.7402640398762137e-05, + "loss": 0.2727, + "step": 4923 + }, + { + "epoch": 2.06, + "learning_rate": 1.740150044142221e-05, + "loss": 0.2417, + "step": 4924 + }, + { + "epoch": 2.06, + "learning_rate": 1.7400360271332718e-05, + "loss": 0.2379, + "step": 4925 + }, + { + "epoch": 2.06, + "learning_rate": 1.7399219888526438e-05, + "loss": 0.2408, + "step": 4926 + }, + { + "epoch": 2.06, + "learning_rate": 1.739807929303615e-05, + "loss": 0.242, + "step": 4927 + }, + { + "epoch": 2.06, + "learning_rate": 1.7396938484894636e-05, + "loss": 0.251, + "step": 4928 + }, + { + "epoch": 2.06, + "learning_rate": 1.739579746413469e-05, + "loss": 0.2409, + "step": 4929 + }, + { + "epoch": 2.06, + "learning_rate": 1.7394656230789106e-05, + "loss": 0.2426, + "step": 4930 + }, + { + "epoch": 2.06, + "learning_rate": 1.7393514784890694e-05, + "loss": 0.2722, + "step": 4931 + }, + { + "epoch": 2.06, + "learning_rate": 1.7392373126472257e-05, + "loss": 0.2441, + "step": 4932 + }, + { + "epoch": 2.07, + "learning_rate": 1.739123125556662e-05, + "loss": 0.2293, + "step": 4933 + }, + { + "epoch": 2.07, + "learning_rate": 1.7390089172206594e-05, + "loss": 0.2598, + "step": 4934 + }, + { + "epoch": 2.07, + "learning_rate": 1.7388946876425016e-05, + "loss": 0.2543, + "step": 4935 + }, + { + "epoch": 2.07, + "learning_rate": 1.7387804368254713e-05, + "loss": 0.2322, + "step": 4936 + }, + { + "epoch": 2.07, + "learning_rate": 1.738666164772853e-05, + "loss": 0.2598, + "step": 4937 + }, + { + "epoch": 2.07, + "learning_rate": 1.7385518714879315e-05, + "loss": 0.2482, + "step": 4938 + }, + { + "epoch": 2.07, + "learning_rate": 1.7384375569739918e-05, + "loss": 0.2783, + "step": 4939 + }, + { + "epoch": 2.07, + "learning_rate": 1.73832322123432e-05, + "loss": 0.2674, + "step": 4940 + }, + { + "epoch": 2.07, + "learning_rate": 1.738208864272202e-05, + "loss": 0.2536, + "step": 4941 + }, + { + "epoch": 2.07, + "learning_rate": 1.7380944860909253e-05, + "loss": 0.2471, + "step": 4942 + }, + { + "epoch": 2.07, + "learning_rate": 1.7379800866937773e-05, + "loss": 0.2384, + "step": 4943 + }, + { + "epoch": 2.07, + "learning_rate": 1.7378656660840475e-05, + "loss": 0.238, + "step": 4944 + }, + { + "epoch": 2.07, + "learning_rate": 1.7377512242650234e-05, + "loss": 0.2664, + "step": 4945 + }, + { + "epoch": 2.07, + "learning_rate": 1.7376367612399947e-05, + "loss": 0.2523, + "step": 4946 + }, + { + "epoch": 2.07, + "learning_rate": 1.737522277012252e-05, + "loss": 0.2584, + "step": 4947 + }, + { + "epoch": 2.07, + "learning_rate": 1.7374077715850865e-05, + "loss": 0.2505, + "step": 4948 + }, + { + "epoch": 2.07, + "learning_rate": 1.7372932449617884e-05, + "loss": 0.2087, + "step": 4949 + }, + { + "epoch": 2.07, + "learning_rate": 1.7371786971456504e-05, + "loss": 0.2631, + "step": 4950 + }, + { + "epoch": 2.07, + "learning_rate": 1.7370641281399647e-05, + "loss": 0.2492, + "step": 4951 + }, + { + "epoch": 2.07, + "learning_rate": 1.736949537948025e-05, + "loss": 0.2934, + "step": 4952 + }, + { + "epoch": 2.07, + "learning_rate": 1.7368349265731246e-05, + "loss": 0.2197, + "step": 4953 + }, + { + "epoch": 2.07, + "learning_rate": 1.7367202940185583e-05, + "loss": 0.2569, + "step": 4954 + }, + { + "epoch": 2.07, + "learning_rate": 1.7366056402876204e-05, + "loss": 0.2391, + "step": 4955 + }, + { + "epoch": 2.07, + "learning_rate": 1.7364909653836074e-05, + "loss": 0.2742, + "step": 4956 + }, + { + "epoch": 2.08, + "learning_rate": 1.7363762693098154e-05, + "loss": 0.2206, + "step": 4957 + }, + { + "epoch": 2.08, + "learning_rate": 1.7362615520695405e-05, + "loss": 0.2756, + "step": 4958 + }, + { + "epoch": 2.08, + "learning_rate": 1.736146813666081e-05, + "loss": 0.2251, + "step": 4959 + }, + { + "epoch": 2.08, + "learning_rate": 1.7360320541027342e-05, + "loss": 0.2713, + "step": 4960 + }, + { + "epoch": 2.08, + "learning_rate": 1.7359172733827995e-05, + "loss": 0.2465, + "step": 4961 + }, + { + "epoch": 2.08, + "learning_rate": 1.735802471509576e-05, + "loss": 0.2497, + "step": 4962 + }, + { + "epoch": 2.08, + "learning_rate": 1.735687648486363e-05, + "loss": 0.2337, + "step": 4963 + }, + { + "epoch": 2.08, + "learning_rate": 1.7355728043164613e-05, + "loss": 0.2509, + "step": 4964 + }, + { + "epoch": 2.08, + "learning_rate": 1.7354579390031724e-05, + "loss": 0.2376, + "step": 4965 + }, + { + "epoch": 2.08, + "learning_rate": 1.7353430525497974e-05, + "loss": 0.2525, + "step": 4966 + }, + { + "epoch": 2.08, + "learning_rate": 1.7352281449596392e-05, + "loss": 0.2847, + "step": 4967 + }, + { + "epoch": 2.08, + "learning_rate": 1.7351132162360002e-05, + "loss": 0.255, + "step": 4968 + }, + { + "epoch": 2.08, + "learning_rate": 1.734998266382184e-05, + "loss": 0.253, + "step": 4969 + }, + { + "epoch": 2.08, + "learning_rate": 1.734883295401495e-05, + "loss": 0.2528, + "step": 4970 + }, + { + "epoch": 2.08, + "learning_rate": 1.7347683032972377e-05, + "loss": 0.2893, + "step": 4971 + }, + { + "epoch": 2.08, + "learning_rate": 1.7346532900727177e-05, + "loss": 0.2915, + "step": 4972 + }, + { + "epoch": 2.08, + "learning_rate": 1.7345382557312408e-05, + "loss": 0.2729, + "step": 4973 + }, + { + "epoch": 2.08, + "learning_rate": 1.7344232002761138e-05, + "loss": 0.2677, + "step": 4974 + }, + { + "epoch": 2.08, + "learning_rate": 1.734308123710643e-05, + "loss": 0.2816, + "step": 4975 + }, + { + "epoch": 2.08, + "learning_rate": 1.734193026038137e-05, + "loss": 0.2447, + "step": 4976 + }, + { + "epoch": 2.08, + "learning_rate": 1.7340779072619043e-05, + "loss": 0.2602, + "step": 4977 + }, + { + "epoch": 2.08, + "learning_rate": 1.7339627673852534e-05, + "loss": 0.2614, + "step": 4978 + }, + { + "epoch": 2.08, + "learning_rate": 1.733847606411494e-05, + "loss": 0.2689, + "step": 4979 + }, + { + "epoch": 2.08, + "learning_rate": 1.7337324243439363e-05, + "loss": 0.2257, + "step": 4980 + }, + { + "epoch": 2.09, + "learning_rate": 1.7336172211858914e-05, + "loss": 0.2506, + "step": 4981 + }, + { + "epoch": 2.09, + "learning_rate": 1.7335019969406702e-05, + "loss": 0.2852, + "step": 4982 + }, + { + "epoch": 2.09, + "learning_rate": 1.7333867516115853e-05, + "loss": 0.226, + "step": 4983 + }, + { + "epoch": 2.09, + "learning_rate": 1.7332714852019487e-05, + "loss": 0.2242, + "step": 4984 + }, + { + "epoch": 2.09, + "learning_rate": 1.733156197715074e-05, + "loss": 0.2465, + "step": 4985 + }, + { + "epoch": 2.09, + "learning_rate": 1.733040889154275e-05, + "loss": 0.2993, + "step": 4986 + }, + { + "epoch": 2.09, + "learning_rate": 1.7329255595228664e-05, + "loss": 0.2579, + "step": 4987 + }, + { + "epoch": 2.09, + "learning_rate": 1.732810208824163e-05, + "loss": 0.2166, + "step": 4988 + }, + { + "epoch": 2.09, + "learning_rate": 1.73269483706148e-05, + "loss": 0.2538, + "step": 4989 + }, + { + "epoch": 2.09, + "learning_rate": 1.7325794442381344e-05, + "loss": 0.2515, + "step": 4990 + }, + { + "epoch": 2.09, + "learning_rate": 1.7324640303574426e-05, + "loss": 0.2579, + "step": 4991 + }, + { + "epoch": 2.09, + "learning_rate": 1.7323485954227223e-05, + "loss": 0.2537, + "step": 4992 + }, + { + "epoch": 2.09, + "learning_rate": 1.7322331394372916e-05, + "loss": 0.2616, + "step": 4993 + }, + { + "epoch": 2.09, + "learning_rate": 1.732117662404469e-05, + "loss": 0.2548, + "step": 4994 + }, + { + "epoch": 2.09, + "learning_rate": 1.7320021643275734e-05, + "loss": 0.2309, + "step": 4995 + }, + { + "epoch": 2.09, + "learning_rate": 1.731886645209926e-05, + "loss": 0.2589, + "step": 4996 + }, + { + "epoch": 2.09, + "learning_rate": 1.731771105054846e-05, + "loss": 0.2616, + "step": 4997 + }, + { + "epoch": 2.09, + "learning_rate": 1.7316555438656547e-05, + "loss": 0.261, + "step": 4998 + }, + { + "epoch": 2.09, + "learning_rate": 1.7315399616456746e-05, + "loss": 0.2548, + "step": 4999 + }, + { + "epoch": 2.09, + "learning_rate": 1.731424358398227e-05, + "loss": 0.2641, + "step": 5000 + }, + { + "epoch": 2.09, + "learning_rate": 1.7313087341266355e-05, + "loss": 0.232, + "step": 5001 + }, + { + "epoch": 2.09, + "learning_rate": 1.7311930888342234e-05, + "loss": 0.2712, + "step": 5002 + }, + { + "epoch": 2.09, + "learning_rate": 1.7310774225243146e-05, + "loss": 0.2451, + "step": 5003 + }, + { + "epoch": 2.1, + "learning_rate": 1.7309617352002343e-05, + "loss": 0.2562, + "step": 5004 + }, + { + "epoch": 2.1, + "learning_rate": 1.7308460268653077e-05, + "loss": 0.2448, + "step": 5005 + }, + { + "epoch": 2.1, + "learning_rate": 1.73073029752286e-05, + "loss": 0.2452, + "step": 5006 + }, + { + "epoch": 2.1, + "learning_rate": 1.730614547176219e-05, + "loss": 0.2327, + "step": 5007 + }, + { + "epoch": 2.1, + "learning_rate": 1.730498775828711e-05, + "loss": 0.2429, + "step": 5008 + }, + { + "epoch": 2.1, + "learning_rate": 1.7303829834836636e-05, + "loss": 0.2622, + "step": 5009 + }, + { + "epoch": 2.1, + "learning_rate": 1.7302671701444057e-05, + "loss": 0.2703, + "step": 5010 + }, + { + "epoch": 2.1, + "learning_rate": 1.730151335814266e-05, + "loss": 0.2536, + "step": 5011 + }, + { + "epoch": 2.1, + "learning_rate": 1.730035480496574e-05, + "loss": 0.2419, + "step": 5012 + }, + { + "epoch": 2.1, + "learning_rate": 1.72991960419466e-05, + "loss": 0.2594, + "step": 5013 + }, + { + "epoch": 2.1, + "learning_rate": 1.7298037069118546e-05, + "loss": 0.2761, + "step": 5014 + }, + { + "epoch": 2.1, + "learning_rate": 1.7296877886514892e-05, + "loss": 0.2803, + "step": 5015 + }, + { + "epoch": 2.1, + "learning_rate": 1.7295718494168963e-05, + "loss": 0.2587, + "step": 5016 + }, + { + "epoch": 2.1, + "learning_rate": 1.7294558892114077e-05, + "loss": 0.2206, + "step": 5017 + }, + { + "epoch": 2.1, + "learning_rate": 1.729339908038357e-05, + "loss": 0.2495, + "step": 5018 + }, + { + "epoch": 2.1, + "learning_rate": 1.7292239059010777e-05, + "loss": 0.2289, + "step": 5019 + }, + { + "epoch": 2.1, + "learning_rate": 1.7291078828029042e-05, + "loss": 0.2807, + "step": 5020 + }, + { + "epoch": 2.1, + "learning_rate": 1.728991838747172e-05, + "loss": 0.242, + "step": 5021 + }, + { + "epoch": 2.1, + "learning_rate": 1.7288757737372157e-05, + "loss": 0.2491, + "step": 5022 + }, + { + "epoch": 2.1, + "learning_rate": 1.7287596877763725e-05, + "loss": 0.2257, + "step": 5023 + }, + { + "epoch": 2.1, + "learning_rate": 1.7286435808679787e-05, + "loss": 0.2492, + "step": 5024 + }, + { + "epoch": 2.1, + "learning_rate": 1.7285274530153718e-05, + "loss": 0.2578, + "step": 5025 + }, + { + "epoch": 2.1, + "learning_rate": 1.7284113042218895e-05, + "loss": 0.26, + "step": 5026 + }, + { + "epoch": 2.1, + "learning_rate": 1.728295134490871e-05, + "loss": 0.2653, + "step": 5027 + }, + { + "epoch": 2.11, + "learning_rate": 1.728178943825655e-05, + "loss": 0.279, + "step": 5028 + }, + { + "epoch": 2.11, + "learning_rate": 1.7280627322295812e-05, + "loss": 0.2164, + "step": 5029 + }, + { + "epoch": 2.11, + "learning_rate": 1.7279464997059905e-05, + "loss": 0.2582, + "step": 5030 + }, + { + "epoch": 2.11, + "learning_rate": 1.727830246258224e-05, + "loss": 0.2465, + "step": 5031 + }, + { + "epoch": 2.11, + "learning_rate": 1.727713971889622e-05, + "loss": 0.2576, + "step": 5032 + }, + { + "epoch": 2.11, + "learning_rate": 1.7275976766035282e-05, + "loss": 0.2423, + "step": 5033 + }, + { + "epoch": 2.11, + "learning_rate": 1.7274813604032847e-05, + "loss": 0.2154, + "step": 5034 + }, + { + "epoch": 2.11, + "learning_rate": 1.727365023292235e-05, + "loss": 0.2747, + "step": 5035 + }, + { + "epoch": 2.11, + "learning_rate": 1.7272486652737236e-05, + "loss": 0.2407, + "step": 5036 + }, + { + "epoch": 2.11, + "learning_rate": 1.7271322863510942e-05, + "loss": 0.2562, + "step": 5037 + }, + { + "epoch": 2.11, + "learning_rate": 1.7270158865276927e-05, + "loss": 0.2562, + "step": 5038 + }, + { + "epoch": 2.11, + "learning_rate": 1.7268994658068645e-05, + "loss": 0.2782, + "step": 5039 + }, + { + "epoch": 2.11, + "learning_rate": 1.7267830241919562e-05, + "loss": 0.2556, + "step": 5040 + }, + { + "epoch": 2.11, + "learning_rate": 1.726666561686315e-05, + "loss": 0.2463, + "step": 5041 + }, + { + "epoch": 2.11, + "learning_rate": 1.7265500782932885e-05, + "loss": 0.2729, + "step": 5042 + }, + { + "epoch": 2.11, + "learning_rate": 1.7264335740162244e-05, + "loss": 0.223, + "step": 5043 + }, + { + "epoch": 2.11, + "learning_rate": 1.7263170488584717e-05, + "loss": 0.2527, + "step": 5044 + }, + { + "epoch": 2.11, + "learning_rate": 1.7262005028233802e-05, + "loss": 0.2513, + "step": 5045 + }, + { + "epoch": 2.11, + "learning_rate": 1.7260839359142995e-05, + "loss": 0.2431, + "step": 5046 + }, + { + "epoch": 2.11, + "learning_rate": 1.7259673481345808e-05, + "loss": 0.2269, + "step": 5047 + }, + { + "epoch": 2.11, + "learning_rate": 1.7258507394875742e-05, + "loss": 0.249, + "step": 5048 + }, + { + "epoch": 2.11, + "learning_rate": 1.725734109976633e-05, + "loss": 0.2325, + "step": 5049 + }, + { + "epoch": 2.11, + "learning_rate": 1.725617459605108e-05, + "loss": 0.2591, + "step": 5050 + }, + { + "epoch": 2.11, + "learning_rate": 1.7255007883763537e-05, + "loss": 0.2405, + "step": 5051 + }, + { + "epoch": 2.12, + "learning_rate": 1.7253840962937226e-05, + "loss": 0.2675, + "step": 5052 + }, + { + "epoch": 2.12, + "learning_rate": 1.7252673833605697e-05, + "loss": 0.2355, + "step": 5053 + }, + { + "epoch": 2.12, + "learning_rate": 1.725150649580249e-05, + "loss": 0.2465, + "step": 5054 + }, + { + "epoch": 2.12, + "learning_rate": 1.7250338949561167e-05, + "loss": 0.2481, + "step": 5055 + }, + { + "epoch": 2.12, + "learning_rate": 1.7249171194915283e-05, + "loss": 0.2396, + "step": 5056 + }, + { + "epoch": 2.12, + "learning_rate": 1.724800323189841e-05, + "loss": 0.2557, + "step": 5057 + }, + { + "epoch": 2.12, + "learning_rate": 1.724683506054411e-05, + "loss": 0.2235, + "step": 5058 + }, + { + "epoch": 2.12, + "learning_rate": 1.7245666680885975e-05, + "loss": 0.2663, + "step": 5059 + }, + { + "epoch": 2.12, + "learning_rate": 1.724449809295757e-05, + "loss": 0.2585, + "step": 5060 + }, + { + "epoch": 2.12, + "learning_rate": 1.7243329296792502e-05, + "loss": 0.2593, + "step": 5061 + }, + { + "epoch": 2.12, + "learning_rate": 1.7242160292424362e-05, + "loss": 0.2311, + "step": 5062 + }, + { + "epoch": 2.12, + "learning_rate": 1.724099107988675e-05, + "loss": 0.2416, + "step": 5063 + }, + { + "epoch": 2.12, + "learning_rate": 1.7239821659213272e-05, + "loss": 0.2705, + "step": 5064 + }, + { + "epoch": 2.12, + "learning_rate": 1.723865203043755e-05, + "loss": 0.2453, + "step": 5065 + }, + { + "epoch": 2.12, + "learning_rate": 1.7237482193593195e-05, + "loss": 0.2731, + "step": 5066 + }, + { + "epoch": 2.12, + "learning_rate": 1.7236312148713838e-05, + "loss": 0.2653, + "step": 5067 + }, + { + "epoch": 2.12, + "learning_rate": 1.723514189583311e-05, + "loss": 0.2623, + "step": 5068 + }, + { + "epoch": 2.12, + "learning_rate": 1.7233971434984648e-05, + "loss": 0.2766, + "step": 5069 + }, + { + "epoch": 2.12, + "learning_rate": 1.72328007662021e-05, + "loss": 0.2531, + "step": 5070 + }, + { + "epoch": 2.12, + "learning_rate": 1.7231629889519106e-05, + "loss": 0.256, + "step": 5071 + }, + { + "epoch": 2.12, + "learning_rate": 1.723045880496933e-05, + "loss": 0.2576, + "step": 5072 + }, + { + "epoch": 2.12, + "learning_rate": 1.7229287512586432e-05, + "loss": 0.2503, + "step": 5073 + }, + { + "epoch": 2.12, + "learning_rate": 1.7228116012404085e-05, + "loss": 0.2751, + "step": 5074 + }, + { + "epoch": 2.12, + "learning_rate": 1.722694430445595e-05, + "loss": 0.258, + "step": 5075 + }, + { + "epoch": 2.13, + "learning_rate": 1.7225772388775716e-05, + "loss": 0.2496, + "step": 5076 + }, + { + "epoch": 2.13, + "learning_rate": 1.722460026539707e-05, + "loss": 0.2667, + "step": 5077 + }, + { + "epoch": 2.13, + "learning_rate": 1.7223427934353695e-05, + "loss": 0.2329, + "step": 5078 + }, + { + "epoch": 2.13, + "learning_rate": 1.7222255395679298e-05, + "loss": 0.2524, + "step": 5079 + }, + { + "epoch": 2.13, + "learning_rate": 1.7221082649407578e-05, + "loss": 0.25, + "step": 5080 + }, + { + "epoch": 2.13, + "learning_rate": 1.7219909695572244e-05, + "loss": 0.2948, + "step": 5081 + }, + { + "epoch": 2.13, + "learning_rate": 1.7218736534207014e-05, + "loss": 0.2579, + "step": 5082 + }, + { + "epoch": 2.13, + "learning_rate": 1.7217563165345607e-05, + "loss": 0.2617, + "step": 5083 + }, + { + "epoch": 2.13, + "learning_rate": 1.721638958902175e-05, + "loss": 0.2608, + "step": 5084 + }, + { + "epoch": 2.13, + "learning_rate": 1.721521580526918e-05, + "loss": 0.2515, + "step": 5085 + }, + { + "epoch": 2.13, + "learning_rate": 1.7214041814121635e-05, + "loss": 0.2472, + "step": 5086 + }, + { + "epoch": 2.13, + "learning_rate": 1.721286761561286e-05, + "loss": 0.2589, + "step": 5087 + }, + { + "epoch": 2.13, + "learning_rate": 1.7211693209776604e-05, + "loss": 0.2718, + "step": 5088 + }, + { + "epoch": 2.13, + "learning_rate": 1.7210518596646627e-05, + "loss": 0.2498, + "step": 5089 + }, + { + "epoch": 2.13, + "learning_rate": 1.720934377625669e-05, + "loss": 0.2506, + "step": 5090 + }, + { + "epoch": 2.13, + "learning_rate": 1.7208168748640565e-05, + "loss": 0.2388, + "step": 5091 + }, + { + "epoch": 2.13, + "learning_rate": 1.7206993513832026e-05, + "loss": 0.2686, + "step": 5092 + }, + { + "epoch": 2.13, + "learning_rate": 1.7205818071864853e-05, + "loss": 0.2474, + "step": 5093 + }, + { + "epoch": 2.13, + "learning_rate": 1.7204642422772836e-05, + "loss": 0.2179, + "step": 5094 + }, + { + "epoch": 2.13, + "learning_rate": 1.7203466566589763e-05, + "loss": 0.257, + "step": 5095 + }, + { + "epoch": 2.13, + "learning_rate": 1.7202290503349436e-05, + "loss": 0.2763, + "step": 5096 + }, + { + "epoch": 2.13, + "learning_rate": 1.7201114233085664e-05, + "loss": 0.2821, + "step": 5097 + }, + { + "epoch": 2.13, + "learning_rate": 1.719993775583225e-05, + "loss": 0.2528, + "step": 5098 + }, + { + "epoch": 2.13, + "learning_rate": 1.7198761071623014e-05, + "loss": 0.2878, + "step": 5099 + }, + { + "epoch": 2.14, + "learning_rate": 1.7197584180491776e-05, + "loss": 0.2428, + "step": 5100 + }, + { + "epoch": 2.14, + "learning_rate": 1.7196407082472375e-05, + "loss": 0.2827, + "step": 5101 + }, + { + "epoch": 2.14, + "learning_rate": 1.7195229777598635e-05, + "loss": 0.2712, + "step": 5102 + }, + { + "epoch": 2.14, + "learning_rate": 1.71940522659044e-05, + "loss": 0.271, + "step": 5103 + }, + { + "epoch": 2.14, + "learning_rate": 1.7192874547423514e-05, + "loss": 0.223, + "step": 5104 + }, + { + "epoch": 2.14, + "learning_rate": 1.7191696622189836e-05, + "loss": 0.2587, + "step": 5105 + }, + { + "epoch": 2.14, + "learning_rate": 1.719051849023722e-05, + "loss": 0.2812, + "step": 5106 + }, + { + "epoch": 2.14, + "learning_rate": 1.718934015159953e-05, + "loss": 0.258, + "step": 5107 + }, + { + "epoch": 2.14, + "learning_rate": 1.7188161606310635e-05, + "loss": 0.2294, + "step": 5108 + }, + { + "epoch": 2.14, + "learning_rate": 1.7186982854404415e-05, + "loss": 0.256, + "step": 5109 + }, + { + "epoch": 2.14, + "learning_rate": 1.718580389591475e-05, + "loss": 0.252, + "step": 5110 + }, + { + "epoch": 2.14, + "learning_rate": 1.7184624730875528e-05, + "loss": 0.2629, + "step": 5111 + }, + { + "epoch": 2.14, + "learning_rate": 1.7183445359320645e-05, + "loss": 0.2399, + "step": 5112 + }, + { + "epoch": 2.14, + "learning_rate": 1.7182265781284e-05, + "loss": 0.257, + "step": 5113 + }, + { + "epoch": 2.14, + "learning_rate": 1.71810859967995e-05, + "loss": 0.2499, + "step": 5114 + }, + { + "epoch": 2.14, + "learning_rate": 1.717990600590105e-05, + "loss": 0.24, + "step": 5115 + }, + { + "epoch": 2.14, + "learning_rate": 1.7178725808622575e-05, + "loss": 0.2659, + "step": 5116 + }, + { + "epoch": 2.14, + "learning_rate": 1.7177545404997998e-05, + "loss": 0.2335, + "step": 5117 + }, + { + "epoch": 2.14, + "learning_rate": 1.717636479506125e-05, + "loss": 0.2443, + "step": 5118 + }, + { + "epoch": 2.14, + "learning_rate": 1.717518397884626e-05, + "loss": 0.2775, + "step": 5119 + }, + { + "epoch": 2.14, + "learning_rate": 1.717400295638698e-05, + "loss": 0.235, + "step": 5120 + }, + { + "epoch": 2.14, + "learning_rate": 1.7172821727717345e-05, + "loss": 0.2412, + "step": 5121 + }, + { + "epoch": 2.14, + "learning_rate": 1.7171640292871314e-05, + "loss": 0.2423, + "step": 5122 + }, + { + "epoch": 2.14, + "learning_rate": 1.7170458651882847e-05, + "loss": 0.2764, + "step": 5123 + }, + { + "epoch": 2.15, + "learning_rate": 1.716927680478591e-05, + "loss": 0.2641, + "step": 5124 + }, + { + "epoch": 2.15, + "learning_rate": 1.716809475161448e-05, + "loss": 0.2793, + "step": 5125 + }, + { + "epoch": 2.15, + "learning_rate": 1.716691249240252e-05, + "loss": 0.2427, + "step": 5126 + }, + { + "epoch": 2.15, + "learning_rate": 1.716573002718402e-05, + "loss": 0.2391, + "step": 5127 + }, + { + "epoch": 2.15, + "learning_rate": 1.7164547355992972e-05, + "loss": 0.2451, + "step": 5128 + }, + { + "epoch": 2.15, + "learning_rate": 1.7163364478863365e-05, + "loss": 0.2525, + "step": 5129 + }, + { + "epoch": 2.15, + "learning_rate": 1.7162181395829204e-05, + "loss": 0.2202, + "step": 5130 + }, + { + "epoch": 2.15, + "learning_rate": 1.7160998106924494e-05, + "loss": 0.2548, + "step": 5131 + }, + { + "epoch": 2.15, + "learning_rate": 1.715981461218325e-05, + "loss": 0.2345, + "step": 5132 + }, + { + "epoch": 2.15, + "learning_rate": 1.7158630911639488e-05, + "loss": 0.2593, + "step": 5133 + }, + { + "epoch": 2.15, + "learning_rate": 1.715744700532723e-05, + "loss": 0.2576, + "step": 5134 + }, + { + "epoch": 2.15, + "learning_rate": 1.7156262893280512e-05, + "loss": 0.2574, + "step": 5135 + }, + { + "epoch": 2.15, + "learning_rate": 1.7155078575533366e-05, + "loss": 0.2137, + "step": 5136 + }, + { + "epoch": 2.15, + "learning_rate": 1.7153894052119837e-05, + "loss": 0.2537, + "step": 5137 + }, + { + "epoch": 2.15, + "learning_rate": 1.7152709323073972e-05, + "loss": 0.2346, + "step": 5138 + }, + { + "epoch": 2.15, + "learning_rate": 1.7151524388429823e-05, + "loss": 0.2037, + "step": 5139 + }, + { + "epoch": 2.15, + "learning_rate": 1.7150339248221454e-05, + "loss": 0.2163, + "step": 5140 + }, + { + "epoch": 2.15, + "learning_rate": 1.7149153902482926e-05, + "loss": 0.2761, + "step": 5141 + }, + { + "epoch": 2.15, + "learning_rate": 1.7147968351248315e-05, + "loss": 0.2492, + "step": 5142 + }, + { + "epoch": 2.15, + "learning_rate": 1.7146782594551698e-05, + "loss": 0.2327, + "step": 5143 + }, + { + "epoch": 2.15, + "learning_rate": 1.714559663242715e-05, + "loss": 0.2507, + "step": 5144 + }, + { + "epoch": 2.15, + "learning_rate": 1.7144410464908774e-05, + "loss": 0.2678, + "step": 5145 + }, + { + "epoch": 2.15, + "learning_rate": 1.714322409203066e-05, + "loss": 0.2523, + "step": 5146 + }, + { + "epoch": 2.15, + "learning_rate": 1.7142037513826904e-05, + "loss": 0.2563, + "step": 5147 + }, + { + "epoch": 2.16, + "learning_rate": 1.714085073033162e-05, + "loss": 0.2264, + "step": 5148 + }, + { + "epoch": 2.16, + "learning_rate": 1.7139663741578917e-05, + "loss": 0.2479, + "step": 5149 + }, + { + "epoch": 2.16, + "learning_rate": 1.7138476547602917e-05, + "loss": 0.2299, + "step": 5150 + }, + { + "epoch": 2.16, + "learning_rate": 1.7137289148437743e-05, + "loss": 0.2625, + "step": 5151 + }, + { + "epoch": 2.16, + "learning_rate": 1.7136101544117526e-05, + "loss": 0.2654, + "step": 5152 + }, + { + "epoch": 2.16, + "learning_rate": 1.7134913734676398e-05, + "loss": 0.2451, + "step": 5153 + }, + { + "epoch": 2.16, + "learning_rate": 1.7133725720148513e-05, + "loss": 0.2287, + "step": 5154 + }, + { + "epoch": 2.16, + "learning_rate": 1.713253750056801e-05, + "loss": 0.2551, + "step": 5155 + }, + { + "epoch": 2.16, + "learning_rate": 1.7131349075969044e-05, + "loss": 0.2604, + "step": 5156 + }, + { + "epoch": 2.16, + "learning_rate": 1.713016044638578e-05, + "loss": 0.2559, + "step": 5157 + }, + { + "epoch": 2.16, + "learning_rate": 1.7128971611852378e-05, + "loss": 0.2613, + "step": 5158 + }, + { + "epoch": 2.16, + "learning_rate": 1.7127782572403015e-05, + "loss": 0.2773, + "step": 5159 + }, + { + "epoch": 2.16, + "learning_rate": 1.7126593328071867e-05, + "loss": 0.2502, + "step": 5160 + }, + { + "epoch": 2.16, + "learning_rate": 1.7125403878893118e-05, + "loss": 0.2821, + "step": 5161 + }, + { + "epoch": 2.16, + "learning_rate": 1.7124214224900955e-05, + "loss": 0.2593, + "step": 5162 + }, + { + "epoch": 2.16, + "learning_rate": 1.712302436612958e-05, + "loss": 0.2377, + "step": 5163 + }, + { + "epoch": 2.16, + "learning_rate": 1.712183430261319e-05, + "loss": 0.2523, + "step": 5164 + }, + { + "epoch": 2.16, + "learning_rate": 1.712064403438599e-05, + "loss": 0.2733, + "step": 5165 + }, + { + "epoch": 2.16, + "learning_rate": 1.71194535614822e-05, + "loss": 0.2376, + "step": 5166 + }, + { + "epoch": 2.16, + "learning_rate": 1.711826288393603e-05, + "loss": 0.2457, + "step": 5167 + }, + { + "epoch": 2.16, + "learning_rate": 1.7117072001781713e-05, + "loss": 0.2349, + "step": 5168 + }, + { + "epoch": 2.16, + "learning_rate": 1.7115880915053476e-05, + "loss": 0.2544, + "step": 5169 + }, + { + "epoch": 2.16, + "learning_rate": 1.7114689623785556e-05, + "loss": 0.2306, + "step": 5170 + }, + { + "epoch": 2.16, + "learning_rate": 1.7113498128012197e-05, + "loss": 0.2666, + "step": 5171 + }, + { + "epoch": 2.17, + "learning_rate": 1.711230642776765e-05, + "loss": 0.2648, + "step": 5172 + }, + { + "epoch": 2.17, + "learning_rate": 1.711111452308616e-05, + "loss": 0.2542, + "step": 5173 + }, + { + "epoch": 2.17, + "learning_rate": 1.7109922414001997e-05, + "loss": 0.2421, + "step": 5174 + }, + { + "epoch": 2.17, + "learning_rate": 1.710873010054942e-05, + "loss": 0.2632, + "step": 5175 + }, + { + "epoch": 2.17, + "learning_rate": 1.710753758276271e-05, + "loss": 0.2471, + "step": 5176 + }, + { + "epoch": 2.17, + "learning_rate": 1.7106344860676137e-05, + "loss": 0.2528, + "step": 5177 + }, + { + "epoch": 2.17, + "learning_rate": 1.7105151934323988e-05, + "loss": 0.2461, + "step": 5178 + }, + { + "epoch": 2.17, + "learning_rate": 1.7103958803740547e-05, + "loss": 0.2155, + "step": 5179 + }, + { + "epoch": 2.17, + "learning_rate": 1.7102765468960116e-05, + "loss": 0.2406, + "step": 5180 + }, + { + "epoch": 2.17, + "learning_rate": 1.7101571930017e-05, + "loss": 0.2621, + "step": 5181 + }, + { + "epoch": 2.17, + "learning_rate": 1.7100378186945494e-05, + "loss": 0.2702, + "step": 5182 + }, + { + "epoch": 2.17, + "learning_rate": 1.709918423977992e-05, + "loss": 0.256, + "step": 5183 + }, + { + "epoch": 2.17, + "learning_rate": 1.7097990088554595e-05, + "loss": 0.2888, + "step": 5184 + }, + { + "epoch": 2.17, + "learning_rate": 1.7096795733303844e-05, + "loss": 0.2608, + "step": 5185 + }, + { + "epoch": 2.17, + "learning_rate": 1.7095601174061997e-05, + "loss": 0.2615, + "step": 5186 + }, + { + "epoch": 2.17, + "learning_rate": 1.709440641086339e-05, + "loss": 0.2309, + "step": 5187 + }, + { + "epoch": 2.17, + "learning_rate": 1.7093211443742368e-05, + "loss": 0.2574, + "step": 5188 + }, + { + "epoch": 2.17, + "learning_rate": 1.7092016272733274e-05, + "loss": 0.2316, + "step": 5189 + }, + { + "epoch": 2.17, + "learning_rate": 1.709082089787047e-05, + "loss": 0.2482, + "step": 5190 + }, + { + "epoch": 2.17, + "learning_rate": 1.7089625319188308e-05, + "loss": 0.264, + "step": 5191 + }, + { + "epoch": 2.17, + "learning_rate": 1.708842953672116e-05, + "loss": 0.2641, + "step": 5192 + }, + { + "epoch": 2.17, + "learning_rate": 1.7087233550503394e-05, + "loss": 0.2737, + "step": 5193 + }, + { + "epoch": 2.17, + "learning_rate": 1.7086037360569386e-05, + "loss": 0.2495, + "step": 5194 + }, + { + "epoch": 2.18, + "learning_rate": 1.7084840966953526e-05, + "loss": 0.2728, + "step": 5195 + }, + { + "epoch": 2.18, + "learning_rate": 1.7083644369690197e-05, + "loss": 0.2599, + "step": 5196 + }, + { + "epoch": 2.18, + "learning_rate": 1.7082447568813798e-05, + "loss": 0.2787, + "step": 5197 + }, + { + "epoch": 2.18, + "learning_rate": 1.708125056435873e-05, + "loss": 0.2641, + "step": 5198 + }, + { + "epoch": 2.18, + "learning_rate": 1.7080053356359396e-05, + "loss": 0.2461, + "step": 5199 + }, + { + "epoch": 2.18, + "learning_rate": 1.707885594485021e-05, + "loss": 0.2538, + "step": 5200 + }, + { + "epoch": 2.18, + "learning_rate": 1.7077658329865593e-05, + "loss": 0.2623, + "step": 5201 + }, + { + "epoch": 2.18, + "learning_rate": 1.7076460511439966e-05, + "loss": 0.2551, + "step": 5202 + }, + { + "epoch": 2.18, + "learning_rate": 1.7075262489607766e-05, + "loss": 0.2842, + "step": 5203 + }, + { + "epoch": 2.18, + "learning_rate": 1.707406426440342e-05, + "loss": 0.2662, + "step": 5204 + }, + { + "epoch": 2.18, + "learning_rate": 1.7072865835861377e-05, + "loss": 0.2783, + "step": 5205 + }, + { + "epoch": 2.18, + "learning_rate": 1.707166720401608e-05, + "loss": 0.2565, + "step": 5206 + }, + { + "epoch": 2.18, + "learning_rate": 1.7070468368901982e-05, + "loss": 0.2397, + "step": 5207 + }, + { + "epoch": 2.18, + "learning_rate": 1.706926933055355e-05, + "loss": 0.2461, + "step": 5208 + }, + { + "epoch": 2.18, + "learning_rate": 1.706807008900524e-05, + "loss": 0.2689, + "step": 5209 + }, + { + "epoch": 2.18, + "learning_rate": 1.7066870644291527e-05, + "loss": 0.2655, + "step": 5210 + }, + { + "epoch": 2.18, + "learning_rate": 1.7065670996446892e-05, + "loss": 0.2626, + "step": 5211 + }, + { + "epoch": 2.18, + "learning_rate": 1.7064471145505813e-05, + "loss": 0.2378, + "step": 5212 + }, + { + "epoch": 2.18, + "learning_rate": 1.706327109150278e-05, + "loss": 0.2428, + "step": 5213 + }, + { + "epoch": 2.18, + "learning_rate": 1.7062070834472287e-05, + "loss": 0.2442, + "step": 5214 + }, + { + "epoch": 2.18, + "learning_rate": 1.7060870374448833e-05, + "loss": 0.2648, + "step": 5215 + }, + { + "epoch": 2.18, + "learning_rate": 1.705966971146693e-05, + "loss": 0.2259, + "step": 5216 + }, + { + "epoch": 2.18, + "learning_rate": 1.7058468845561084e-05, + "loss": 0.2766, + "step": 5217 + }, + { + "epoch": 2.18, + "learning_rate": 1.7057267776765814e-05, + "loss": 0.2512, + "step": 5218 + }, + { + "epoch": 2.19, + "learning_rate": 1.705606650511564e-05, + "loss": 0.2322, + "step": 5219 + }, + { + "epoch": 2.19, + "learning_rate": 1.7054865030645102e-05, + "loss": 0.2896, + "step": 5220 + }, + { + "epoch": 2.19, + "learning_rate": 1.7053663353388727e-05, + "loss": 0.2412, + "step": 5221 + }, + { + "epoch": 2.19, + "learning_rate": 1.7052461473381056e-05, + "loss": 0.2456, + "step": 5222 + }, + { + "epoch": 2.19, + "learning_rate": 1.7051259390656642e-05, + "loss": 0.2572, + "step": 5223 + }, + { + "epoch": 2.19, + "learning_rate": 1.7050057105250034e-05, + "loss": 0.2393, + "step": 5224 + }, + { + "epoch": 2.19, + "learning_rate": 1.7048854617195788e-05, + "loss": 0.2851, + "step": 5225 + }, + { + "epoch": 2.19, + "learning_rate": 1.7047651926528472e-05, + "loss": 0.2603, + "step": 5226 + }, + { + "epoch": 2.19, + "learning_rate": 1.7046449033282656e-05, + "loss": 0.2448, + "step": 5227 + }, + { + "epoch": 2.19, + "learning_rate": 1.7045245937492912e-05, + "loss": 0.2544, + "step": 5228 + }, + { + "epoch": 2.19, + "learning_rate": 1.704404263919383e-05, + "loss": 0.2433, + "step": 5229 + }, + { + "epoch": 2.19, + "learning_rate": 1.7042839138419992e-05, + "loss": 0.2589, + "step": 5230 + }, + { + "epoch": 2.19, + "learning_rate": 1.704163543520599e-05, + "loss": 0.2382, + "step": 5231 + }, + { + "epoch": 2.19, + "learning_rate": 1.7040431529586427e-05, + "loss": 0.2312, + "step": 5232 + }, + { + "epoch": 2.19, + "learning_rate": 1.703922742159591e-05, + "loss": 0.2521, + "step": 5233 + }, + { + "epoch": 2.19, + "learning_rate": 1.7038023111269045e-05, + "loss": 0.258, + "step": 5234 + }, + { + "epoch": 2.19, + "learning_rate": 1.7036818598640456e-05, + "loss": 0.2897, + "step": 5235 + }, + { + "epoch": 2.19, + "learning_rate": 1.7035613883744753e-05, + "loss": 0.2337, + "step": 5236 + }, + { + "epoch": 2.19, + "learning_rate": 1.7034408966616577e-05, + "loss": 0.2502, + "step": 5237 + }, + { + "epoch": 2.19, + "learning_rate": 1.7033203847290553e-05, + "loss": 0.2469, + "step": 5238 + }, + { + "epoch": 2.19, + "learning_rate": 1.703199852580133e-05, + "loss": 0.2517, + "step": 5239 + }, + { + "epoch": 2.19, + "learning_rate": 1.7030793002183546e-05, + "loss": 0.2594, + "step": 5240 + }, + { + "epoch": 2.19, + "learning_rate": 1.702958727647186e-05, + "loss": 0.2306, + "step": 5241 + }, + { + "epoch": 2.19, + "learning_rate": 1.7028381348700926e-05, + "loss": 0.2449, + "step": 5242 + }, + { + "epoch": 2.2, + "learning_rate": 1.70271752189054e-05, + "loss": 0.2469, + "step": 5243 + }, + { + "epoch": 2.2, + "learning_rate": 1.7025968887119962e-05, + "loss": 0.3028, + "step": 5244 + }, + { + "epoch": 2.2, + "learning_rate": 1.702476235337929e-05, + "loss": 0.2763, + "step": 5245 + }, + { + "epoch": 2.2, + "learning_rate": 1.702355561771805e-05, + "loss": 0.2196, + "step": 5246 + }, + { + "epoch": 2.2, + "learning_rate": 1.7022348680170936e-05, + "loss": 0.2704, + "step": 5247 + }, + { + "epoch": 2.2, + "learning_rate": 1.7021141540772645e-05, + "loss": 0.2302, + "step": 5248 + }, + { + "epoch": 2.2, + "learning_rate": 1.7019934199557868e-05, + "loss": 0.2421, + "step": 5249 + }, + { + "epoch": 2.2, + "learning_rate": 1.7018726656561314e-05, + "loss": 0.254, + "step": 5250 + }, + { + "epoch": 2.2, + "learning_rate": 1.7017518911817686e-05, + "loss": 0.251, + "step": 5251 + }, + { + "epoch": 2.2, + "learning_rate": 1.7016310965361708e-05, + "loss": 0.2674, + "step": 5252 + }, + { + "epoch": 2.2, + "learning_rate": 1.70151028172281e-05, + "loss": 0.268, + "step": 5253 + }, + { + "epoch": 2.2, + "learning_rate": 1.701389446745158e-05, + "loss": 0.2271, + "step": 5254 + }, + { + "epoch": 2.2, + "learning_rate": 1.7012685916066894e-05, + "loss": 0.2591, + "step": 5255 + }, + { + "epoch": 2.2, + "learning_rate": 1.7011477163108766e-05, + "loss": 0.2526, + "step": 5256 + }, + { + "epoch": 2.2, + "learning_rate": 1.7010268208611958e-05, + "loss": 0.271, + "step": 5257 + }, + { + "epoch": 2.2, + "learning_rate": 1.7009059052611206e-05, + "loss": 0.238, + "step": 5258 + }, + { + "epoch": 2.2, + "learning_rate": 1.7007849695141275e-05, + "loss": 0.2589, + "step": 5259 + }, + { + "epoch": 2.2, + "learning_rate": 1.700664013623692e-05, + "loss": 0.2659, + "step": 5260 + }, + { + "epoch": 2.2, + "learning_rate": 1.700543037593291e-05, + "loss": 0.2963, + "step": 5261 + }, + { + "epoch": 2.2, + "learning_rate": 1.7004220414264025e-05, + "loss": 0.2727, + "step": 5262 + }, + { + "epoch": 2.2, + "learning_rate": 1.7003010251265036e-05, + "loss": 0.2739, + "step": 5263 + }, + { + "epoch": 2.2, + "learning_rate": 1.7001799886970735e-05, + "loss": 0.2541, + "step": 5264 + }, + { + "epoch": 2.2, + "learning_rate": 1.7000589321415906e-05, + "loss": 0.2445, + "step": 5265 + }, + { + "epoch": 2.2, + "learning_rate": 1.699937855463535e-05, + "loss": 0.2615, + "step": 5266 + }, + { + "epoch": 2.21, + "learning_rate": 1.6998167586663868e-05, + "loss": 0.2552, + "step": 5267 + }, + { + "epoch": 2.21, + "learning_rate": 1.699695641753627e-05, + "loss": 0.2752, + "step": 5268 + }, + { + "epoch": 2.21, + "learning_rate": 1.6995745047287373e-05, + "loss": 0.2623, + "step": 5269 + }, + { + "epoch": 2.21, + "learning_rate": 1.6994533475951985e-05, + "loss": 0.2388, + "step": 5270 + }, + { + "epoch": 2.21, + "learning_rate": 1.6993321703564946e-05, + "loss": 0.2854, + "step": 5271 + }, + { + "epoch": 2.21, + "learning_rate": 1.6992109730161077e-05, + "loss": 0.264, + "step": 5272 + }, + { + "epoch": 2.21, + "learning_rate": 1.699089755577522e-05, + "loss": 0.2667, + "step": 5273 + }, + { + "epoch": 2.21, + "learning_rate": 1.6989685180442212e-05, + "loss": 0.2484, + "step": 5274 + }, + { + "epoch": 2.21, + "learning_rate": 1.6988472604196907e-05, + "loss": 0.2643, + "step": 5275 + }, + { + "epoch": 2.21, + "learning_rate": 1.698725982707416e-05, + "loss": 0.2616, + "step": 5276 + }, + { + "epoch": 2.21, + "learning_rate": 1.6986046849108834e-05, + "loss": 0.2566, + "step": 5277 + }, + { + "epoch": 2.21, + "learning_rate": 1.6984833670335783e-05, + "loss": 0.26, + "step": 5278 + }, + { + "epoch": 2.21, + "learning_rate": 1.698362029078989e-05, + "loss": 0.2434, + "step": 5279 + }, + { + "epoch": 2.21, + "learning_rate": 1.698240671050603e-05, + "loss": 0.2577, + "step": 5280 + }, + { + "epoch": 2.21, + "learning_rate": 1.6981192929519087e-05, + "loss": 0.2326, + "step": 5281 + }, + { + "epoch": 2.21, + "learning_rate": 1.6979978947863946e-05, + "loss": 0.2519, + "step": 5282 + }, + { + "epoch": 2.21, + "learning_rate": 1.69787647655755e-05, + "loss": 0.2774, + "step": 5283 + }, + { + "epoch": 2.21, + "learning_rate": 1.697755038268866e-05, + "loss": 0.2594, + "step": 5284 + }, + { + "epoch": 2.21, + "learning_rate": 1.6976335799238318e-05, + "loss": 0.2421, + "step": 5285 + }, + { + "epoch": 2.21, + "learning_rate": 1.69751210152594e-05, + "loss": 0.2378, + "step": 5286 + }, + { + "epoch": 2.21, + "learning_rate": 1.697390603078682e-05, + "loss": 0.2452, + "step": 5287 + }, + { + "epoch": 2.21, + "learning_rate": 1.6972690845855494e-05, + "loss": 0.2753, + "step": 5288 + }, + { + "epoch": 2.21, + "learning_rate": 1.697147546050036e-05, + "loss": 0.2547, + "step": 5289 + }, + { + "epoch": 2.21, + "learning_rate": 1.6970259874756353e-05, + "loss": 0.2338, + "step": 5290 + }, + { + "epoch": 2.22, + "learning_rate": 1.696904408865841e-05, + "loss": 0.2653, + "step": 5291 + }, + { + "epoch": 2.22, + "learning_rate": 1.6967828102241475e-05, + "loss": 0.2689, + "step": 5292 + }, + { + "epoch": 2.22, + "learning_rate": 1.6966611915540506e-05, + "loss": 0.2678, + "step": 5293 + }, + { + "epoch": 2.22, + "learning_rate": 1.696539552859046e-05, + "loss": 0.2517, + "step": 5294 + }, + { + "epoch": 2.22, + "learning_rate": 1.69641789414263e-05, + "loss": 0.2531, + "step": 5295 + }, + { + "epoch": 2.22, + "learning_rate": 1.6962962154083e-05, + "loss": 0.263, + "step": 5296 + }, + { + "epoch": 2.22, + "learning_rate": 1.6961745166595528e-05, + "loss": 0.26, + "step": 5297 + }, + { + "epoch": 2.22, + "learning_rate": 1.696052797899887e-05, + "loss": 0.2513, + "step": 5298 + }, + { + "epoch": 2.22, + "learning_rate": 1.695931059132801e-05, + "loss": 0.2297, + "step": 5299 + }, + { + "epoch": 2.22, + "learning_rate": 1.6958093003617942e-05, + "loss": 0.2827, + "step": 5300 + }, + { + "epoch": 2.22, + "learning_rate": 1.695687521590367e-05, + "loss": 0.2749, + "step": 5301 + }, + { + "epoch": 2.22, + "learning_rate": 1.695565722822019e-05, + "loss": 0.2556, + "step": 5302 + }, + { + "epoch": 2.22, + "learning_rate": 1.6954439040602516e-05, + "loss": 0.2822, + "step": 5303 + }, + { + "epoch": 2.22, + "learning_rate": 1.6953220653085662e-05, + "loss": 0.2487, + "step": 5304 + }, + { + "epoch": 2.22, + "learning_rate": 1.6952002065704652e-05, + "loss": 0.2533, + "step": 5305 + }, + { + "epoch": 2.22, + "learning_rate": 1.6950783278494513e-05, + "loss": 0.241, + "step": 5306 + }, + { + "epoch": 2.22, + "learning_rate": 1.6949564291490274e-05, + "loss": 0.2709, + "step": 5307 + }, + { + "epoch": 2.22, + "learning_rate": 1.694834510472698e-05, + "loss": 0.2417, + "step": 5308 + }, + { + "epoch": 2.22, + "learning_rate": 1.6947125718239664e-05, + "loss": 0.2468, + "step": 5309 + }, + { + "epoch": 2.22, + "learning_rate": 1.694590613206339e-05, + "loss": 0.2712, + "step": 5310 + }, + { + "epoch": 2.22, + "learning_rate": 1.6944686346233207e-05, + "loss": 0.2649, + "step": 5311 + }, + { + "epoch": 2.22, + "learning_rate": 1.694346636078418e-05, + "loss": 0.25, + "step": 5312 + }, + { + "epoch": 2.22, + "learning_rate": 1.6942246175751372e-05, + "loss": 0.2659, + "step": 5313 + }, + { + "epoch": 2.22, + "learning_rate": 1.6941025791169855e-05, + "loss": 0.2657, + "step": 5314 + }, + { + "epoch": 2.23, + "learning_rate": 1.6939805207074714e-05, + "loss": 0.253, + "step": 5315 + }, + { + "epoch": 2.23, + "learning_rate": 1.693858442350103e-05, + "loss": 0.2573, + "step": 5316 + }, + { + "epoch": 2.23, + "learning_rate": 1.6937363440483892e-05, + "loss": 0.2307, + "step": 5317 + }, + { + "epoch": 2.23, + "learning_rate": 1.6936142258058403e-05, + "loss": 0.2418, + "step": 5318 + }, + { + "epoch": 2.23, + "learning_rate": 1.6934920876259657e-05, + "loss": 0.2507, + "step": 5319 + }, + { + "epoch": 2.23, + "learning_rate": 1.6933699295122766e-05, + "loss": 0.2765, + "step": 5320 + }, + { + "epoch": 2.23, + "learning_rate": 1.6932477514682838e-05, + "loss": 0.2427, + "step": 5321 + }, + { + "epoch": 2.23, + "learning_rate": 1.6931255534974997e-05, + "loss": 0.2684, + "step": 5322 + }, + { + "epoch": 2.23, + "learning_rate": 1.6930033356034368e-05, + "loss": 0.2533, + "step": 5323 + }, + { + "epoch": 2.23, + "learning_rate": 1.692881097789608e-05, + "loss": 0.2579, + "step": 5324 + }, + { + "epoch": 2.23, + "learning_rate": 1.6927588400595265e-05, + "loss": 0.2453, + "step": 5325 + }, + { + "epoch": 2.23, + "learning_rate": 1.6926365624167075e-05, + "loss": 0.2447, + "step": 5326 + }, + { + "epoch": 2.23, + "learning_rate": 1.6925142648646647e-05, + "loss": 0.2478, + "step": 5327 + }, + { + "epoch": 2.23, + "learning_rate": 1.692391947406914e-05, + "loss": 0.2556, + "step": 5328 + }, + { + "epoch": 2.23, + "learning_rate": 1.692269610046971e-05, + "loss": 0.2543, + "step": 5329 + }, + { + "epoch": 2.23, + "learning_rate": 1.6921472527883527e-05, + "loss": 0.2752, + "step": 5330 + }, + { + "epoch": 2.23, + "learning_rate": 1.6920248756345757e-05, + "loss": 0.2648, + "step": 5331 + }, + { + "epoch": 2.23, + "learning_rate": 1.6919024785891576e-05, + "loss": 0.2655, + "step": 5332 + }, + { + "epoch": 2.23, + "learning_rate": 1.691780061655617e-05, + "loss": 0.2572, + "step": 5333 + }, + { + "epoch": 2.23, + "learning_rate": 1.691657624837472e-05, + "loss": 0.2709, + "step": 5334 + }, + { + "epoch": 2.23, + "learning_rate": 1.6915351681382427e-05, + "loss": 0.2539, + "step": 5335 + }, + { + "epoch": 2.23, + "learning_rate": 1.6914126915614483e-05, + "loss": 0.2223, + "step": 5336 + }, + { + "epoch": 2.23, + "learning_rate": 1.69129019511061e-05, + "loss": 0.263, + "step": 5337 + }, + { + "epoch": 2.23, + "learning_rate": 1.6911676787892484e-05, + "loss": 0.2504, + "step": 5338 + }, + { + "epoch": 2.24, + "learning_rate": 1.691045142600885e-05, + "loss": 0.2343, + "step": 5339 + }, + { + "epoch": 2.24, + "learning_rate": 1.6909225865490424e-05, + "loss": 0.2212, + "step": 5340 + }, + { + "epoch": 2.24, + "learning_rate": 1.6908000106372432e-05, + "loss": 0.2872, + "step": 5341 + }, + { + "epoch": 2.24, + "learning_rate": 1.6906774148690106e-05, + "loss": 0.2537, + "step": 5342 + }, + { + "epoch": 2.24, + "learning_rate": 1.6905547992478687e-05, + "loss": 0.2897, + "step": 5343 + }, + { + "epoch": 2.24, + "learning_rate": 1.6904321637773414e-05, + "loss": 0.2736, + "step": 5344 + }, + { + "epoch": 2.24, + "learning_rate": 1.690309508460955e-05, + "loss": 0.2745, + "step": 5345 + }, + { + "epoch": 2.24, + "learning_rate": 1.6901868333022337e-05, + "loss": 0.2602, + "step": 5346 + }, + { + "epoch": 2.24, + "learning_rate": 1.6900641383047046e-05, + "loss": 0.2841, + "step": 5347 + }, + { + "epoch": 2.24, + "learning_rate": 1.6899414234718943e-05, + "loss": 0.2467, + "step": 5348 + }, + { + "epoch": 2.24, + "learning_rate": 1.68981868880733e-05, + "loss": 0.2693, + "step": 5349 + }, + { + "epoch": 2.24, + "learning_rate": 1.6896959343145396e-05, + "loss": 0.267, + "step": 5350 + }, + { + "epoch": 2.24, + "learning_rate": 1.6895731599970516e-05, + "loss": 0.2706, + "step": 5351 + }, + { + "epoch": 2.24, + "learning_rate": 1.689450365858395e-05, + "loss": 0.2446, + "step": 5352 + }, + { + "epoch": 2.24, + "learning_rate": 1.689327551902099e-05, + "loss": 0.2822, + "step": 5353 + }, + { + "epoch": 2.24, + "learning_rate": 1.6892047181316952e-05, + "loss": 0.2618, + "step": 5354 + }, + { + "epoch": 2.24, + "learning_rate": 1.6890818645507126e-05, + "loss": 0.2618, + "step": 5355 + }, + { + "epoch": 2.24, + "learning_rate": 1.6889589911626834e-05, + "loss": 0.2688, + "step": 5356 + }, + { + "epoch": 2.24, + "learning_rate": 1.6888360979711395e-05, + "loss": 0.2756, + "step": 5357 + }, + { + "epoch": 2.24, + "learning_rate": 1.6887131849796132e-05, + "loss": 0.2537, + "step": 5358 + }, + { + "epoch": 2.24, + "learning_rate": 1.6885902521916375e-05, + "loss": 0.2688, + "step": 5359 + }, + { + "epoch": 2.24, + "learning_rate": 1.6884672996107464e-05, + "loss": 0.2811, + "step": 5360 + }, + { + "epoch": 2.24, + "learning_rate": 1.6883443272404735e-05, + "loss": 0.2519, + "step": 5361 + }, + { + "epoch": 2.24, + "learning_rate": 1.688221335084353e-05, + "loss": 0.2722, + "step": 5362 + }, + { + "epoch": 2.25, + "learning_rate": 1.6880983231459217e-05, + "loss": 0.2626, + "step": 5363 + }, + { + "epoch": 2.25, + "learning_rate": 1.687975291428715e-05, + "loss": 0.2585, + "step": 5364 + }, + { + "epoch": 2.25, + "learning_rate": 1.6878522399362683e-05, + "loss": 0.2621, + "step": 5365 + }, + { + "epoch": 2.25, + "learning_rate": 1.68772916867212e-05, + "loss": 0.218, + "step": 5366 + }, + { + "epoch": 2.25, + "learning_rate": 1.6876060776398063e-05, + "loss": 0.2472, + "step": 5367 + }, + { + "epoch": 2.25, + "learning_rate": 1.6874829668428667e-05, + "loss": 0.2508, + "step": 5368 + }, + { + "epoch": 2.25, + "learning_rate": 1.687359836284839e-05, + "loss": 0.2585, + "step": 5369 + }, + { + "epoch": 2.25, + "learning_rate": 1.687236685969263e-05, + "loss": 0.2585, + "step": 5370 + }, + { + "epoch": 2.25, + "learning_rate": 1.6871135158996782e-05, + "loss": 0.2454, + "step": 5371 + }, + { + "epoch": 2.25, + "learning_rate": 1.686990326079625e-05, + "loss": 0.2705, + "step": 5372 + }, + { + "epoch": 2.25, + "learning_rate": 1.6868671165126445e-05, + "loss": 0.299, + "step": 5373 + }, + { + "epoch": 2.25, + "learning_rate": 1.6867438872022782e-05, + "loss": 0.2597, + "step": 5374 + }, + { + "epoch": 2.25, + "learning_rate": 1.6866206381520683e-05, + "loss": 0.2684, + "step": 5375 + }, + { + "epoch": 2.25, + "learning_rate": 1.6864973693655577e-05, + "loss": 0.243, + "step": 5376 + }, + { + "epoch": 2.25, + "learning_rate": 1.686374080846289e-05, + "loss": 0.2693, + "step": 5377 + }, + { + "epoch": 2.25, + "learning_rate": 1.6862507725978067e-05, + "loss": 0.2694, + "step": 5378 + }, + { + "epoch": 2.25, + "learning_rate": 1.686127444623655e-05, + "loss": 0.2304, + "step": 5379 + }, + { + "epoch": 2.25, + "learning_rate": 1.686004096927378e-05, + "loss": 0.2439, + "step": 5380 + }, + { + "epoch": 2.25, + "learning_rate": 1.685880729512523e-05, + "loss": 0.2477, + "step": 5381 + }, + { + "epoch": 2.25, + "learning_rate": 1.6857573423826344e-05, + "loss": 0.264, + "step": 5382 + }, + { + "epoch": 2.25, + "learning_rate": 1.6856339355412602e-05, + "loss": 0.233, + "step": 5383 + }, + { + "epoch": 2.25, + "learning_rate": 1.6855105089919467e-05, + "loss": 0.2768, + "step": 5384 + }, + { + "epoch": 2.25, + "learning_rate": 1.6853870627382416e-05, + "loss": 0.2539, + "step": 5385 + }, + { + "epoch": 2.25, + "learning_rate": 1.6852635967836936e-05, + "loss": 0.2581, + "step": 5386 + }, + { + "epoch": 2.26, + "learning_rate": 1.685140111131852e-05, + "loss": 0.2704, + "step": 5387 + }, + { + "epoch": 2.26, + "learning_rate": 1.6850166057862654e-05, + "loss": 0.2789, + "step": 5388 + }, + { + "epoch": 2.26, + "learning_rate": 1.6848930807504846e-05, + "loss": 0.2843, + "step": 5389 + }, + { + "epoch": 2.26, + "learning_rate": 1.68476953602806e-05, + "loss": 0.2948, + "step": 5390 + }, + { + "epoch": 2.26, + "learning_rate": 1.684645971622543e-05, + "loss": 0.255, + "step": 5391 + }, + { + "epoch": 2.26, + "learning_rate": 1.6845223875374845e-05, + "loss": 0.2635, + "step": 5392 + }, + { + "epoch": 2.26, + "learning_rate": 1.6843987837764375e-05, + "loss": 0.2963, + "step": 5393 + }, + { + "epoch": 2.26, + "learning_rate": 1.684275160342955e-05, + "loss": 0.2592, + "step": 5394 + }, + { + "epoch": 2.26, + "learning_rate": 1.68415151724059e-05, + "loss": 0.2682, + "step": 5395 + }, + { + "epoch": 2.26, + "learning_rate": 1.684027854472897e-05, + "loss": 0.2681, + "step": 5396 + }, + { + "epoch": 2.26, + "learning_rate": 1.6839041720434303e-05, + "loss": 0.2474, + "step": 5397 + }, + { + "epoch": 2.26, + "learning_rate": 1.683780469955745e-05, + "loss": 0.2892, + "step": 5398 + }, + { + "epoch": 2.26, + "learning_rate": 1.683656748213397e-05, + "loss": 0.2526, + "step": 5399 + }, + { + "epoch": 2.26, + "learning_rate": 1.6835330068199424e-05, + "loss": 0.2477, + "step": 5400 + }, + { + "epoch": 2.26, + "learning_rate": 1.683409245778938e-05, + "loss": 0.2851, + "step": 5401 + }, + { + "epoch": 2.26, + "learning_rate": 1.683285465093941e-05, + "loss": 0.2454, + "step": 5402 + }, + { + "epoch": 2.26, + "learning_rate": 1.6831616647685102e-05, + "loss": 0.2666, + "step": 5403 + }, + { + "epoch": 2.26, + "learning_rate": 1.683037844806203e-05, + "loss": 0.2602, + "step": 5404 + }, + { + "epoch": 2.26, + "learning_rate": 1.6829140052105795e-05, + "loss": 0.2423, + "step": 5405 + }, + { + "epoch": 2.26, + "learning_rate": 1.6827901459851986e-05, + "loss": 0.2727, + "step": 5406 + }, + { + "epoch": 2.26, + "learning_rate": 1.6826662671336212e-05, + "loss": 0.2452, + "step": 5407 + }, + { + "epoch": 2.26, + "learning_rate": 1.6825423686594075e-05, + "loss": 0.2498, + "step": 5408 + }, + { + "epoch": 2.26, + "learning_rate": 1.682418450566119e-05, + "loss": 0.2509, + "step": 5409 + }, + { + "epoch": 2.27, + "learning_rate": 1.6822945128573182e-05, + "loss": 0.2815, + "step": 5410 + }, + { + "epoch": 2.27, + "learning_rate": 1.6821705555365665e-05, + "loss": 0.2747, + "step": 5411 + }, + { + "epoch": 2.27, + "learning_rate": 1.6820465786074277e-05, + "loss": 0.2632, + "step": 5412 + }, + { + "epoch": 2.27, + "learning_rate": 1.6819225820734656e-05, + "loss": 0.2268, + "step": 5413 + }, + { + "epoch": 2.27, + "learning_rate": 1.6817985659382437e-05, + "loss": 0.2475, + "step": 5414 + }, + { + "epoch": 2.27, + "learning_rate": 1.681674530205327e-05, + "loss": 0.2406, + "step": 5415 + }, + { + "epoch": 2.27, + "learning_rate": 1.681550474878281e-05, + "loss": 0.2551, + "step": 5416 + }, + { + "epoch": 2.27, + "learning_rate": 1.6814263999606713e-05, + "loss": 0.2745, + "step": 5417 + }, + { + "epoch": 2.27, + "learning_rate": 1.681302305456064e-05, + "loss": 0.2644, + "step": 5418 + }, + { + "epoch": 2.27, + "learning_rate": 1.6811781913680273e-05, + "loss": 0.2814, + "step": 5419 + }, + { + "epoch": 2.27, + "learning_rate": 1.6810540577001277e-05, + "loss": 0.2529, + "step": 5420 + }, + { + "epoch": 2.27, + "learning_rate": 1.6809299044559333e-05, + "loss": 0.2791, + "step": 5421 + }, + { + "epoch": 2.27, + "learning_rate": 1.6808057316390128e-05, + "loss": 0.255, + "step": 5422 + }, + { + "epoch": 2.27, + "learning_rate": 1.6806815392529363e-05, + "loss": 0.26, + "step": 5423 + }, + { + "epoch": 2.27, + "learning_rate": 1.680557327301273e-05, + "loss": 0.256, + "step": 5424 + }, + { + "epoch": 2.27, + "learning_rate": 1.6804330957875926e-05, + "loss": 0.2486, + "step": 5425 + }, + { + "epoch": 2.27, + "learning_rate": 1.6803088447154672e-05, + "loss": 0.2407, + "step": 5426 + }, + { + "epoch": 2.27, + "learning_rate": 1.6801845740884678e-05, + "loss": 0.2346, + "step": 5427 + }, + { + "epoch": 2.27, + "learning_rate": 1.680060283910166e-05, + "loss": 0.2524, + "step": 5428 + }, + { + "epoch": 2.27, + "learning_rate": 1.6799359741841354e-05, + "loss": 0.2708, + "step": 5429 + }, + { + "epoch": 2.27, + "learning_rate": 1.6798116449139483e-05, + "loss": 0.2632, + "step": 5430 + }, + { + "epoch": 2.27, + "learning_rate": 1.6796872961031783e-05, + "loss": 0.2237, + "step": 5431 + }, + { + "epoch": 2.27, + "learning_rate": 1.6795629277554003e-05, + "loss": 0.2382, + "step": 5432 + }, + { + "epoch": 2.27, + "learning_rate": 1.6794385398741894e-05, + "loss": 0.2691, + "step": 5433 + }, + { + "epoch": 2.28, + "learning_rate": 1.67931413246312e-05, + "loss": 0.2636, + "step": 5434 + }, + { + "epoch": 2.28, + "learning_rate": 1.679189705525769e-05, + "loss": 0.2594, + "step": 5435 + }, + { + "epoch": 2.28, + "learning_rate": 1.6790652590657125e-05, + "loss": 0.2543, + "step": 5436 + }, + { + "epoch": 2.28, + "learning_rate": 1.678940793086528e-05, + "loss": 0.2591, + "step": 5437 + }, + { + "epoch": 2.28, + "learning_rate": 1.678816307591793e-05, + "loss": 0.2541, + "step": 5438 + }, + { + "epoch": 2.28, + "learning_rate": 1.6786918025850852e-05, + "loss": 0.2863, + "step": 5439 + }, + { + "epoch": 2.28, + "learning_rate": 1.6785672780699836e-05, + "loss": 0.2611, + "step": 5440 + }, + { + "epoch": 2.28, + "learning_rate": 1.6784427340500684e-05, + "loss": 0.286, + "step": 5441 + }, + { + "epoch": 2.28, + "learning_rate": 1.6783181705289184e-05, + "loss": 0.244, + "step": 5442 + }, + { + "epoch": 2.28, + "learning_rate": 1.6781935875101148e-05, + "loss": 0.2727, + "step": 5443 + }, + { + "epoch": 2.28, + "learning_rate": 1.678068984997238e-05, + "loss": 0.2597, + "step": 5444 + }, + { + "epoch": 2.28, + "learning_rate": 1.67794436299387e-05, + "loss": 0.2972, + "step": 5445 + }, + { + "epoch": 2.28, + "learning_rate": 1.677819721503593e-05, + "loss": 0.2636, + "step": 5446 + }, + { + "epoch": 2.28, + "learning_rate": 1.6776950605299897e-05, + "loss": 0.2327, + "step": 5447 + }, + { + "epoch": 2.28, + "learning_rate": 1.677570380076643e-05, + "loss": 0.275, + "step": 5448 + }, + { + "epoch": 2.28, + "learning_rate": 1.677445680147137e-05, + "loss": 0.2507, + "step": 5449 + }, + { + "epoch": 2.28, + "learning_rate": 1.6773209607450564e-05, + "loss": 0.2583, + "step": 5450 + }, + { + "epoch": 2.28, + "learning_rate": 1.6771962218739856e-05, + "loss": 0.2828, + "step": 5451 + }, + { + "epoch": 2.28, + "learning_rate": 1.6770714635375104e-05, + "loss": 0.2481, + "step": 5452 + }, + { + "epoch": 2.28, + "learning_rate": 1.6769466857392167e-05, + "loss": 0.2325, + "step": 5453 + }, + { + "epoch": 2.28, + "learning_rate": 1.6768218884826916e-05, + "loss": 0.2542, + "step": 5454 + }, + { + "epoch": 2.28, + "learning_rate": 1.6766970717715216e-05, + "loss": 0.2508, + "step": 5455 + }, + { + "epoch": 2.28, + "learning_rate": 1.6765722356092947e-05, + "loss": 0.256, + "step": 5456 + }, + { + "epoch": 2.28, + "learning_rate": 1.6764473799995993e-05, + "loss": 0.261, + "step": 5457 + }, + { + "epoch": 2.29, + "learning_rate": 1.676322504946024e-05, + "loss": 0.2655, + "step": 5458 + }, + { + "epoch": 2.29, + "learning_rate": 1.676197610452159e-05, + "loss": 0.2763, + "step": 5459 + }, + { + "epoch": 2.29, + "learning_rate": 1.6760726965215936e-05, + "loss": 0.2443, + "step": 5460 + }, + { + "epoch": 2.29, + "learning_rate": 1.6759477631579182e-05, + "loss": 0.2783, + "step": 5461 + }, + { + "epoch": 2.29, + "learning_rate": 1.6758228103647246e-05, + "loss": 0.2794, + "step": 5462 + }, + { + "epoch": 2.29, + "learning_rate": 1.6756978381456035e-05, + "loss": 0.2522, + "step": 5463 + }, + { + "epoch": 2.29, + "learning_rate": 1.675572846504148e-05, + "loss": 0.2734, + "step": 5464 + }, + { + "epoch": 2.29, + "learning_rate": 1.6754478354439504e-05, + "loss": 0.2389, + "step": 5465 + }, + { + "epoch": 2.29, + "learning_rate": 1.6753228049686042e-05, + "loss": 0.2517, + "step": 5466 + }, + { + "epoch": 2.29, + "learning_rate": 1.675197755081703e-05, + "loss": 0.2726, + "step": 5467 + }, + { + "epoch": 2.29, + "learning_rate": 1.675072685786842e-05, + "loss": 0.2493, + "step": 5468 + }, + { + "epoch": 2.29, + "learning_rate": 1.674947597087615e-05, + "loss": 0.2413, + "step": 5469 + }, + { + "epoch": 2.29, + "learning_rate": 1.6748224889876188e-05, + "loss": 0.2812, + "step": 5470 + }, + { + "epoch": 2.29, + "learning_rate": 1.674697361490449e-05, + "loss": 0.2636, + "step": 5471 + }, + { + "epoch": 2.29, + "learning_rate": 1.674572214599702e-05, + "loss": 0.2142, + "step": 5472 + }, + { + "epoch": 2.29, + "learning_rate": 1.6744470483189755e-05, + "loss": 0.2527, + "step": 5473 + }, + { + "epoch": 2.29, + "learning_rate": 1.674321862651867e-05, + "loss": 0.2542, + "step": 5474 + }, + { + "epoch": 2.29, + "learning_rate": 1.674196657601975e-05, + "loss": 0.2612, + "step": 5475 + }, + { + "epoch": 2.29, + "learning_rate": 1.6740714331728984e-05, + "loss": 0.2351, + "step": 5476 + }, + { + "epoch": 2.29, + "learning_rate": 1.673946189368236e-05, + "loss": 0.2701, + "step": 5477 + }, + { + "epoch": 2.29, + "learning_rate": 1.673820926191589e-05, + "loss": 0.2584, + "step": 5478 + }, + { + "epoch": 2.29, + "learning_rate": 1.6736956436465573e-05, + "loss": 0.2623, + "step": 5479 + }, + { + "epoch": 2.29, + "learning_rate": 1.6735703417367422e-05, + "loss": 0.2528, + "step": 5480 + }, + { + "epoch": 2.29, + "learning_rate": 1.673445020465745e-05, + "loss": 0.2376, + "step": 5481 + }, + { + "epoch": 2.3, + "learning_rate": 1.673319679837169e-05, + "loss": 0.2837, + "step": 5482 + }, + { + "epoch": 2.3, + "learning_rate": 1.673194319854616e-05, + "loss": 0.261, + "step": 5483 + }, + { + "epoch": 2.3, + "learning_rate": 1.6730689405216895e-05, + "loss": 0.2461, + "step": 5484 + }, + { + "epoch": 2.3, + "learning_rate": 1.6729435418419936e-05, + "loss": 0.2953, + "step": 5485 + }, + { + "epoch": 2.3, + "learning_rate": 1.6728181238191328e-05, + "loss": 0.2853, + "step": 5486 + }, + { + "epoch": 2.3, + "learning_rate": 1.672692686456712e-05, + "loss": 0.2504, + "step": 5487 + }, + { + "epoch": 2.3, + "learning_rate": 1.672567229758337e-05, + "loss": 0.2589, + "step": 5488 + }, + { + "epoch": 2.3, + "learning_rate": 1.6724417537276138e-05, + "loss": 0.2655, + "step": 5489 + }, + { + "epoch": 2.3, + "learning_rate": 1.6723162583681492e-05, + "loss": 0.249, + "step": 5490 + }, + { + "epoch": 2.3, + "learning_rate": 1.67219074368355e-05, + "loss": 0.2526, + "step": 5491 + }, + { + "epoch": 2.3, + "learning_rate": 1.6720652096774245e-05, + "loss": 0.2313, + "step": 5492 + }, + { + "epoch": 2.3, + "learning_rate": 1.6719396563533812e-05, + "loss": 0.2452, + "step": 5493 + }, + { + "epoch": 2.3, + "learning_rate": 1.6718140837150286e-05, + "loss": 0.2643, + "step": 5494 + }, + { + "epoch": 2.3, + "learning_rate": 1.6716884917659763e-05, + "loss": 0.2852, + "step": 5495 + }, + { + "epoch": 2.3, + "learning_rate": 1.671562880509834e-05, + "loss": 0.2577, + "step": 5496 + }, + { + "epoch": 2.3, + "learning_rate": 1.6714372499502133e-05, + "loss": 0.2387, + "step": 5497 + }, + { + "epoch": 2.3, + "learning_rate": 1.671311600090724e-05, + "loss": 0.2763, + "step": 5498 + }, + { + "epoch": 2.3, + "learning_rate": 1.6711859309349788e-05, + "loss": 0.2555, + "step": 5499 + }, + { + "epoch": 2.3, + "learning_rate": 1.67106024248659e-05, + "loss": 0.2557, + "step": 5500 + }, + { + "epoch": 2.3, + "learning_rate": 1.6709345347491695e-05, + "loss": 0.2474, + "step": 5501 + }, + { + "epoch": 2.3, + "learning_rate": 1.670808807726331e-05, + "loss": 0.261, + "step": 5502 + }, + { + "epoch": 2.3, + "learning_rate": 1.670683061421689e-05, + "loss": 0.247, + "step": 5503 + }, + { + "epoch": 2.3, + "learning_rate": 1.6705572958388576e-05, + "loss": 0.26, + "step": 5504 + }, + { + "epoch": 2.3, + "learning_rate": 1.6704315109814514e-05, + "loss": 0.2832, + "step": 5505 + }, + { + "epoch": 2.31, + "learning_rate": 1.670305706853087e-05, + "loss": 0.2386, + "step": 5506 + }, + { + "epoch": 2.31, + "learning_rate": 1.670179883457379e-05, + "loss": 0.286, + "step": 5507 + }, + { + "epoch": 2.31, + "learning_rate": 1.6700540407979457e-05, + "loss": 0.2268, + "step": 5508 + }, + { + "epoch": 2.31, + "learning_rate": 1.669928178878403e-05, + "loss": 0.2397, + "step": 5509 + }, + { + "epoch": 2.31, + "learning_rate": 1.6698022977023697e-05, + "loss": 0.25, + "step": 5510 + }, + { + "epoch": 2.31, + "learning_rate": 1.6696763972734637e-05, + "loss": 0.2595, + "step": 5511 + }, + { + "epoch": 2.31, + "learning_rate": 1.6695504775953036e-05, + "loss": 0.244, + "step": 5512 + }, + { + "epoch": 2.31, + "learning_rate": 1.6694245386715094e-05, + "loss": 0.2475, + "step": 5513 + }, + { + "epoch": 2.31, + "learning_rate": 1.669298580505701e-05, + "loss": 0.253, + "step": 5514 + }, + { + "epoch": 2.31, + "learning_rate": 1.6691726031014985e-05, + "loss": 0.2752, + "step": 5515 + }, + { + "epoch": 2.31, + "learning_rate": 1.669046606462524e-05, + "loss": 0.2726, + "step": 5516 + }, + { + "epoch": 2.31, + "learning_rate": 1.6689205905923977e-05, + "loss": 0.238, + "step": 5517 + }, + { + "epoch": 2.31, + "learning_rate": 1.6687945554947428e-05, + "loss": 0.2627, + "step": 5518 + }, + { + "epoch": 2.31, + "learning_rate": 1.668668501173182e-05, + "loss": 0.2485, + "step": 5519 + }, + { + "epoch": 2.31, + "learning_rate": 1.6685424276313386e-05, + "loss": 0.2596, + "step": 5520 + }, + { + "epoch": 2.31, + "learning_rate": 1.6684163348728364e-05, + "loss": 0.2893, + "step": 5521 + }, + { + "epoch": 2.31, + "learning_rate": 1.6682902229012997e-05, + "loss": 0.2622, + "step": 5522 + }, + { + "epoch": 2.31, + "learning_rate": 1.6681640917203534e-05, + "loss": 0.2604, + "step": 5523 + }, + { + "epoch": 2.31, + "learning_rate": 1.668037941333623e-05, + "loss": 0.2601, + "step": 5524 + }, + { + "epoch": 2.31, + "learning_rate": 1.6679117717447356e-05, + "loss": 0.2353, + "step": 5525 + }, + { + "epoch": 2.31, + "learning_rate": 1.6677855829573164e-05, + "loss": 0.242, + "step": 5526 + }, + { + "epoch": 2.31, + "learning_rate": 1.6676593749749936e-05, + "loss": 0.2716, + "step": 5527 + }, + { + "epoch": 2.31, + "learning_rate": 1.667533147801394e-05, + "loss": 0.258, + "step": 5528 + }, + { + "epoch": 2.31, + "learning_rate": 1.667406901440147e-05, + "loss": 0.273, + "step": 5529 + }, + { + "epoch": 2.32, + "learning_rate": 1.6672806358948803e-05, + "loss": 0.2658, + "step": 5530 + }, + { + "epoch": 2.32, + "learning_rate": 1.667154351169224e-05, + "loss": 0.2419, + "step": 5531 + }, + { + "epoch": 2.32, + "learning_rate": 1.6670280472668082e-05, + "loss": 0.2512, + "step": 5532 + }, + { + "epoch": 2.32, + "learning_rate": 1.666901724191263e-05, + "loss": 0.2694, + "step": 5533 + }, + { + "epoch": 2.32, + "learning_rate": 1.6667753819462196e-05, + "loss": 0.2649, + "step": 5534 + }, + { + "epoch": 2.32, + "learning_rate": 1.6666490205353092e-05, + "loss": 0.2606, + "step": 5535 + }, + { + "epoch": 2.32, + "learning_rate": 1.6665226399621642e-05, + "loss": 0.2565, + "step": 5536 + }, + { + "epoch": 2.32, + "learning_rate": 1.6663962402304178e-05, + "loss": 0.239, + "step": 5537 + }, + { + "epoch": 2.32, + "learning_rate": 1.666269821343703e-05, + "loss": 0.2428, + "step": 5538 + }, + { + "epoch": 2.32, + "learning_rate": 1.6661433833056527e-05, + "loss": 0.2325, + "step": 5539 + }, + { + "epoch": 2.32, + "learning_rate": 1.6660169261199027e-05, + "loss": 0.2501, + "step": 5540 + }, + { + "epoch": 2.32, + "learning_rate": 1.665890449790087e-05, + "loss": 0.2478, + "step": 5541 + }, + { + "epoch": 2.32, + "learning_rate": 1.665763954319841e-05, + "loss": 0.23, + "step": 5542 + }, + { + "epoch": 2.32, + "learning_rate": 1.6656374397128007e-05, + "loss": 0.2581, + "step": 5543 + }, + { + "epoch": 2.32, + "learning_rate": 1.665510905972603e-05, + "loss": 0.2491, + "step": 5544 + }, + { + "epoch": 2.32, + "learning_rate": 1.6653843531028852e-05, + "loss": 0.255, + "step": 5545 + }, + { + "epoch": 2.32, + "learning_rate": 1.6652577811072842e-05, + "loss": 0.2237, + "step": 5546 + }, + { + "epoch": 2.32, + "learning_rate": 1.665131189989439e-05, + "loss": 0.2275, + "step": 5547 + }, + { + "epoch": 2.32, + "learning_rate": 1.665004579752988e-05, + "loss": 0.2532, + "step": 5548 + }, + { + "epoch": 2.32, + "learning_rate": 1.66487795040157e-05, + "loss": 0.2732, + "step": 5549 + }, + { + "epoch": 2.32, + "learning_rate": 1.664751301938826e-05, + "loss": 0.2887, + "step": 5550 + }, + { + "epoch": 2.32, + "learning_rate": 1.6646246343683953e-05, + "loss": 0.2389, + "step": 5551 + }, + { + "epoch": 2.32, + "learning_rate": 1.6644979476939194e-05, + "loss": 0.2584, + "step": 5552 + }, + { + "epoch": 2.32, + "learning_rate": 1.6643712419190396e-05, + "loss": 0.2517, + "step": 5553 + }, + { + "epoch": 2.33, + "learning_rate": 1.664244517047398e-05, + "loss": 0.2563, + "step": 5554 + }, + { + "epoch": 2.33, + "learning_rate": 1.664117773082637e-05, + "loss": 0.2428, + "step": 5555 + }, + { + "epoch": 2.33, + "learning_rate": 1.6639910100284e-05, + "loss": 0.2671, + "step": 5556 + }, + { + "epoch": 2.33, + "learning_rate": 1.663864227888331e-05, + "loss": 0.2196, + "step": 5557 + }, + { + "epoch": 2.33, + "learning_rate": 1.6637374266660734e-05, + "loss": 0.2565, + "step": 5558 + }, + { + "epoch": 2.33, + "learning_rate": 1.6636106063652725e-05, + "loss": 0.2578, + "step": 5559 + }, + { + "epoch": 2.33, + "learning_rate": 1.663483766989574e-05, + "loss": 0.2725, + "step": 5560 + }, + { + "epoch": 2.33, + "learning_rate": 1.663356908542623e-05, + "loss": 0.2808, + "step": 5561 + }, + { + "epoch": 2.33, + "learning_rate": 1.6632300310280664e-05, + "loss": 0.2417, + "step": 5562 + }, + { + "epoch": 2.33, + "learning_rate": 1.663103134449551e-05, + "loss": 0.2653, + "step": 5563 + }, + { + "epoch": 2.33, + "learning_rate": 1.6629762188107248e-05, + "loss": 0.2614, + "step": 5564 + }, + { + "epoch": 2.33, + "learning_rate": 1.6628492841152357e-05, + "loss": 0.2483, + "step": 5565 + }, + { + "epoch": 2.33, + "learning_rate": 1.6627223303667314e-05, + "loss": 0.2439, + "step": 5566 + }, + { + "epoch": 2.33, + "learning_rate": 1.6625953575688624e-05, + "loss": 0.2314, + "step": 5567 + }, + { + "epoch": 2.33, + "learning_rate": 1.6624683657252773e-05, + "loss": 0.216, + "step": 5568 + }, + { + "epoch": 2.33, + "learning_rate": 1.6623413548396272e-05, + "loss": 0.253, + "step": 5569 + }, + { + "epoch": 2.33, + "learning_rate": 1.6622143249155625e-05, + "loss": 0.2446, + "step": 5570 + }, + { + "epoch": 2.33, + "learning_rate": 1.6620872759567345e-05, + "loss": 0.256, + "step": 5571 + }, + { + "epoch": 2.33, + "learning_rate": 1.6619602079667956e-05, + "loss": 0.2411, + "step": 5572 + }, + { + "epoch": 2.33, + "learning_rate": 1.661833120949398e-05, + "loss": 0.2755, + "step": 5573 + }, + { + "epoch": 2.33, + "learning_rate": 1.6617060149081944e-05, + "loss": 0.2411, + "step": 5574 + }, + { + "epoch": 2.33, + "learning_rate": 1.6615788898468384e-05, + "loss": 0.2439, + "step": 5575 + }, + { + "epoch": 2.33, + "learning_rate": 1.661451745768985e-05, + "loss": 0.2606, + "step": 5576 + }, + { + "epoch": 2.33, + "learning_rate": 1.6613245826782875e-05, + "loss": 0.288, + "step": 5577 + }, + { + "epoch": 2.34, + "learning_rate": 1.6611974005784017e-05, + "loss": 0.2454, + "step": 5578 + }, + { + "epoch": 2.34, + "learning_rate": 1.6610701994729838e-05, + "loss": 0.2253, + "step": 5579 + }, + { + "epoch": 2.34, + "learning_rate": 1.660942979365689e-05, + "loss": 0.2392, + "step": 5580 + }, + { + "epoch": 2.34, + "learning_rate": 1.6608157402601753e-05, + "loss": 0.2397, + "step": 5581 + }, + { + "epoch": 2.34, + "learning_rate": 1.6606884821600995e-05, + "loss": 0.2303, + "step": 5582 + }, + { + "epoch": 2.34, + "learning_rate": 1.6605612050691197e-05, + "loss": 0.2431, + "step": 5583 + }, + { + "epoch": 2.34, + "learning_rate": 1.6604339089908935e-05, + "loss": 0.2579, + "step": 5584 + }, + { + "epoch": 2.34, + "learning_rate": 1.6603065939290815e-05, + "loss": 0.2297, + "step": 5585 + }, + { + "epoch": 2.34, + "learning_rate": 1.660179259887342e-05, + "loss": 0.2695, + "step": 5586 + }, + { + "epoch": 2.34, + "learning_rate": 1.6600519068693356e-05, + "loss": 0.2433, + "step": 5587 + }, + { + "epoch": 2.34, + "learning_rate": 1.659924534878723e-05, + "loss": 0.2382, + "step": 5588 + }, + { + "epoch": 2.34, + "learning_rate": 1.659797143919165e-05, + "loss": 0.2639, + "step": 5589 + }, + { + "epoch": 2.34, + "learning_rate": 1.6596697339943234e-05, + "loss": 0.2494, + "step": 5590 + }, + { + "epoch": 2.34, + "learning_rate": 1.659542305107861e-05, + "loss": 0.283, + "step": 5591 + }, + { + "epoch": 2.34, + "learning_rate": 1.6594148572634404e-05, + "loss": 0.2544, + "step": 5592 + }, + { + "epoch": 2.34, + "learning_rate": 1.6592873904647245e-05, + "loss": 0.2498, + "step": 5593 + }, + { + "epoch": 2.34, + "learning_rate": 1.6591599047153778e-05, + "loss": 0.2804, + "step": 5594 + }, + { + "epoch": 2.34, + "learning_rate": 1.6590324000190645e-05, + "loss": 0.2287, + "step": 5595 + }, + { + "epoch": 2.34, + "learning_rate": 1.65890487637945e-05, + "loss": 0.2674, + "step": 5596 + }, + { + "epoch": 2.34, + "learning_rate": 1.6587773338001993e-05, + "loss": 0.2598, + "step": 5597 + }, + { + "epoch": 2.34, + "learning_rate": 1.658649772284979e-05, + "loss": 0.2656, + "step": 5598 + }, + { + "epoch": 2.34, + "learning_rate": 1.658522191837455e-05, + "loss": 0.2451, + "step": 5599 + }, + { + "epoch": 2.34, + "learning_rate": 1.658394592461295e-05, + "loss": 0.2575, + "step": 5600 + }, + { + "epoch": 2.34, + "learning_rate": 1.658266974160167e-05, + "loss": 0.259, + "step": 5601 + }, + { + "epoch": 2.35, + "learning_rate": 1.6581393369377385e-05, + "loss": 0.2611, + "step": 5602 + }, + { + "epoch": 2.35, + "learning_rate": 1.6580116807976795e-05, + "loss": 0.2447, + "step": 5603 + }, + { + "epoch": 2.35, + "learning_rate": 1.657884005743658e-05, + "loss": 0.2716, + "step": 5604 + }, + { + "epoch": 2.35, + "learning_rate": 1.6577563117793447e-05, + "loss": 0.2492, + "step": 5605 + }, + { + "epoch": 2.35, + "learning_rate": 1.65762859890841e-05, + "loss": 0.2713, + "step": 5606 + }, + { + "epoch": 2.35, + "learning_rate": 1.6575008671345248e-05, + "loss": 0.2381, + "step": 5607 + }, + { + "epoch": 2.35, + "learning_rate": 1.6573731164613604e-05, + "loss": 0.2521, + "step": 5608 + }, + { + "epoch": 2.35, + "learning_rate": 1.6572453468925895e-05, + "loss": 0.2688, + "step": 5609 + }, + { + "epoch": 2.35, + "learning_rate": 1.6571175584318845e-05, + "loss": 0.2438, + "step": 5610 + }, + { + "epoch": 2.35, + "learning_rate": 1.656989751082918e-05, + "loss": 0.2655, + "step": 5611 + }, + { + "epoch": 2.35, + "learning_rate": 1.656861924849364e-05, + "loss": 0.2333, + "step": 5612 + }, + { + "epoch": 2.35, + "learning_rate": 1.656734079734897e-05, + "loss": 0.277, + "step": 5613 + }, + { + "epoch": 2.35, + "learning_rate": 1.6566062157431915e-05, + "loss": 0.2572, + "step": 5614 + }, + { + "epoch": 2.35, + "learning_rate": 1.656478332877923e-05, + "loss": 0.2439, + "step": 5615 + }, + { + "epoch": 2.35, + "learning_rate": 1.6563504311427678e-05, + "loss": 0.2518, + "step": 5616 + }, + { + "epoch": 2.35, + "learning_rate": 1.6562225105414016e-05, + "loss": 0.2696, + "step": 5617 + }, + { + "epoch": 2.35, + "learning_rate": 1.656094571077502e-05, + "loss": 0.2628, + "step": 5618 + }, + { + "epoch": 2.35, + "learning_rate": 1.6559666127547455e-05, + "loss": 0.2394, + "step": 5619 + }, + { + "epoch": 2.35, + "learning_rate": 1.655838635576811e-05, + "loss": 0.2871, + "step": 5620 + }, + { + "epoch": 2.35, + "learning_rate": 1.6557106395473775e-05, + "loss": 0.2729, + "step": 5621 + }, + { + "epoch": 2.35, + "learning_rate": 1.655582624670123e-05, + "loss": 0.2643, + "step": 5622 + }, + { + "epoch": 2.35, + "learning_rate": 1.6554545909487277e-05, + "loss": 0.2522, + "step": 5623 + }, + { + "epoch": 2.35, + "learning_rate": 1.655326538386872e-05, + "loss": 0.2616, + "step": 5624 + }, + { + "epoch": 2.36, + "learning_rate": 1.6551984669882362e-05, + "loss": 0.2816, + "step": 5625 + }, + { + "epoch": 2.36, + "learning_rate": 1.655070376756502e-05, + "loss": 0.2558, + "step": 5626 + }, + { + "epoch": 2.36, + "learning_rate": 1.6549422676953513e-05, + "loss": 0.2902, + "step": 5627 + }, + { + "epoch": 2.36, + "learning_rate": 1.6548141398084662e-05, + "loss": 0.2461, + "step": 5628 + }, + { + "epoch": 2.36, + "learning_rate": 1.6546859930995295e-05, + "loss": 0.2601, + "step": 5629 + }, + { + "epoch": 2.36, + "learning_rate": 1.654557827572225e-05, + "loss": 0.2647, + "step": 5630 + }, + { + "epoch": 2.36, + "learning_rate": 1.6544296432302373e-05, + "loss": 0.2559, + "step": 5631 + }, + { + "epoch": 2.36, + "learning_rate": 1.6543014400772496e-05, + "loss": 0.2638, + "step": 5632 + }, + { + "epoch": 2.36, + "learning_rate": 1.6541732181169474e-05, + "loss": 0.2409, + "step": 5633 + }, + { + "epoch": 2.36, + "learning_rate": 1.6540449773530168e-05, + "loss": 0.2524, + "step": 5634 + }, + { + "epoch": 2.36, + "learning_rate": 1.6539167177891436e-05, + "loss": 0.2701, + "step": 5635 + }, + { + "epoch": 2.36, + "learning_rate": 1.653788439429015e-05, + "loss": 0.2518, + "step": 5636 + }, + { + "epoch": 2.36, + "learning_rate": 1.6536601422763173e-05, + "loss": 0.2835, + "step": 5637 + }, + { + "epoch": 2.36, + "learning_rate": 1.6535318263347395e-05, + "loss": 0.2598, + "step": 5638 + }, + { + "epoch": 2.36, + "learning_rate": 1.6534034916079686e-05, + "loss": 0.2588, + "step": 5639 + }, + { + "epoch": 2.36, + "learning_rate": 1.653275138099695e-05, + "loss": 0.222, + "step": 5640 + }, + { + "epoch": 2.36, + "learning_rate": 1.6531467658136064e-05, + "loss": 0.2774, + "step": 5641 + }, + { + "epoch": 2.36, + "learning_rate": 1.6530183747533942e-05, + "loss": 0.2674, + "step": 5642 + }, + { + "epoch": 2.36, + "learning_rate": 1.6528899649227482e-05, + "loss": 0.26, + "step": 5643 + }, + { + "epoch": 2.36, + "learning_rate": 1.652761536325359e-05, + "loss": 0.2622, + "step": 5644 + }, + { + "epoch": 2.36, + "learning_rate": 1.6526330889649194e-05, + "loss": 0.2339, + "step": 5645 + }, + { + "epoch": 2.36, + "learning_rate": 1.6525046228451204e-05, + "loss": 0.279, + "step": 5646 + }, + { + "epoch": 2.36, + "learning_rate": 1.652376137969655e-05, + "loss": 0.2597, + "step": 5647 + }, + { + "epoch": 2.36, + "learning_rate": 1.6522476343422164e-05, + "loss": 0.2777, + "step": 5648 + }, + { + "epoch": 2.37, + "learning_rate": 1.652119111966499e-05, + "loss": 0.2719, + "step": 5649 + }, + { + "epoch": 2.37, + "learning_rate": 1.6519905708461953e-05, + "loss": 0.2532, + "step": 5650 + }, + { + "epoch": 2.37, + "learning_rate": 1.651862010985002e-05, + "loss": 0.2652, + "step": 5651 + }, + { + "epoch": 2.37, + "learning_rate": 1.6517334323866136e-05, + "loss": 0.2313, + "step": 5652 + }, + { + "epoch": 2.37, + "learning_rate": 1.651604835054726e-05, + "loss": 0.253, + "step": 5653 + }, + { + "epoch": 2.37, + "learning_rate": 1.6514762189930353e-05, + "loss": 0.2641, + "step": 5654 + }, + { + "epoch": 2.37, + "learning_rate": 1.651347584205239e-05, + "loss": 0.2449, + "step": 5655 + }, + { + "epoch": 2.37, + "learning_rate": 1.6512189306950345e-05, + "loss": 0.2649, + "step": 5656 + }, + { + "epoch": 2.37, + "learning_rate": 1.6510902584661195e-05, + "loss": 0.2837, + "step": 5657 + }, + { + "epoch": 2.37, + "learning_rate": 1.650961567522193e-05, + "loss": 0.2499, + "step": 5658 + }, + { + "epoch": 2.37, + "learning_rate": 1.6508328578669537e-05, + "loss": 0.2475, + "step": 5659 + }, + { + "epoch": 2.37, + "learning_rate": 1.6507041295041015e-05, + "loss": 0.246, + "step": 5660 + }, + { + "epoch": 2.37, + "learning_rate": 1.6505753824373364e-05, + "loss": 0.2631, + "step": 5661 + }, + { + "epoch": 2.37, + "learning_rate": 1.6504466166703593e-05, + "loss": 0.2594, + "step": 5662 + }, + { + "epoch": 2.37, + "learning_rate": 1.6503178322068718e-05, + "loss": 0.2271, + "step": 5663 + }, + { + "epoch": 2.37, + "learning_rate": 1.650189029050575e-05, + "loss": 0.2303, + "step": 5664 + }, + { + "epoch": 2.37, + "learning_rate": 1.6500602072051714e-05, + "loss": 0.2522, + "step": 5665 + }, + { + "epoch": 2.37, + "learning_rate": 1.6499313666743643e-05, + "loss": 0.2741, + "step": 5666 + }, + { + "epoch": 2.37, + "learning_rate": 1.6498025074618563e-05, + "loss": 0.2291, + "step": 5667 + }, + { + "epoch": 2.37, + "learning_rate": 1.6496736295713526e-05, + "loss": 0.2841, + "step": 5668 + }, + { + "epoch": 2.37, + "learning_rate": 1.649544733006556e-05, + "loss": 0.246, + "step": 5669 + }, + { + "epoch": 2.37, + "learning_rate": 1.649415817771173e-05, + "loss": 0.2773, + "step": 5670 + }, + { + "epoch": 2.37, + "learning_rate": 1.649286883868909e-05, + "loss": 0.2308, + "step": 5671 + }, + { + "epoch": 2.37, + "learning_rate": 1.649157931303469e-05, + "loss": 0.244, + "step": 5672 + }, + { + "epoch": 2.38, + "learning_rate": 1.6490289600785603e-05, + "loss": 0.2649, + "step": 5673 + }, + { + "epoch": 2.38, + "learning_rate": 1.6488999701978905e-05, + "loss": 0.2718, + "step": 5674 + }, + { + "epoch": 2.38, + "learning_rate": 1.6487709616651667e-05, + "loss": 0.2615, + "step": 5675 + }, + { + "epoch": 2.38, + "learning_rate": 1.6486419344840974e-05, + "loss": 0.2741, + "step": 5676 + }, + { + "epoch": 2.38, + "learning_rate": 1.6485128886583913e-05, + "loss": 0.2812, + "step": 5677 + }, + { + "epoch": 2.38, + "learning_rate": 1.6483838241917577e-05, + "loss": 0.2596, + "step": 5678 + }, + { + "epoch": 2.38, + "learning_rate": 1.6482547410879063e-05, + "loss": 0.2359, + "step": 5679 + }, + { + "epoch": 2.38, + "learning_rate": 1.6481256393505477e-05, + "loss": 0.2907, + "step": 5680 + }, + { + "epoch": 2.38, + "learning_rate": 1.6479965189833923e-05, + "loss": 0.2376, + "step": 5681 + }, + { + "epoch": 2.38, + "learning_rate": 1.6478673799901525e-05, + "loss": 0.2418, + "step": 5682 + }, + { + "epoch": 2.38, + "learning_rate": 1.6477382223745396e-05, + "loss": 0.2486, + "step": 5683 + }, + { + "epoch": 2.38, + "learning_rate": 1.6476090461402664e-05, + "loss": 0.2549, + "step": 5684 + }, + { + "epoch": 2.38, + "learning_rate": 1.647479851291046e-05, + "loss": 0.2417, + "step": 5685 + }, + { + "epoch": 2.38, + "learning_rate": 1.6473506378305914e-05, + "loss": 0.2823, + "step": 5686 + }, + { + "epoch": 2.38, + "learning_rate": 1.6472214057626175e-05, + "loss": 0.2542, + "step": 5687 + }, + { + "epoch": 2.38, + "learning_rate": 1.6470921550908385e-05, + "loss": 0.2549, + "step": 5688 + }, + { + "epoch": 2.38, + "learning_rate": 1.6469628858189698e-05, + "loss": 0.2689, + "step": 5689 + }, + { + "epoch": 2.38, + "learning_rate": 1.646833597950727e-05, + "loss": 0.2706, + "step": 5690 + }, + { + "epoch": 2.38, + "learning_rate": 1.6467042914898262e-05, + "loss": 0.2514, + "step": 5691 + }, + { + "epoch": 2.38, + "learning_rate": 1.6465749664399848e-05, + "loss": 0.2491, + "step": 5692 + }, + { + "epoch": 2.38, + "learning_rate": 1.6464456228049195e-05, + "loss": 0.2418, + "step": 5693 + }, + { + "epoch": 2.38, + "learning_rate": 1.6463162605883485e-05, + "loss": 0.2389, + "step": 5694 + }, + { + "epoch": 2.38, + "learning_rate": 1.64618687979399e-05, + "loss": 0.2294, + "step": 5695 + }, + { + "epoch": 2.38, + "learning_rate": 1.6460574804255634e-05, + "loss": 0.2445, + "step": 5696 + }, + { + "epoch": 2.39, + "learning_rate": 1.6459280624867876e-05, + "loss": 0.2614, + "step": 5697 + }, + { + "epoch": 2.39, + "learning_rate": 1.6457986259813827e-05, + "loss": 0.2873, + "step": 5698 + }, + { + "epoch": 2.39, + "learning_rate": 1.6456691709130694e-05, + "loss": 0.251, + "step": 5699 + }, + { + "epoch": 2.39, + "learning_rate": 1.6455396972855688e-05, + "loss": 0.2761, + "step": 5700 + }, + { + "epoch": 2.39, + "learning_rate": 1.6454102051026025e-05, + "loss": 0.2521, + "step": 5701 + }, + { + "epoch": 2.39, + "learning_rate": 1.6452806943678928e-05, + "loss": 0.2566, + "step": 5702 + }, + { + "epoch": 2.39, + "learning_rate": 1.645151165085162e-05, + "loss": 0.2531, + "step": 5703 + }, + { + "epoch": 2.39, + "learning_rate": 1.645021617258133e-05, + "loss": 0.2512, + "step": 5704 + }, + { + "epoch": 2.39, + "learning_rate": 1.6448920508905305e-05, + "loss": 0.239, + "step": 5705 + }, + { + "epoch": 2.39, + "learning_rate": 1.644762465986078e-05, + "loss": 0.2325, + "step": 5706 + }, + { + "epoch": 2.39, + "learning_rate": 1.6446328625485007e-05, + "loss": 0.2521, + "step": 5707 + }, + { + "epoch": 2.39, + "learning_rate": 1.644503240581524e-05, + "loss": 0.2613, + "step": 5708 + }, + { + "epoch": 2.39, + "learning_rate": 1.6443736000888732e-05, + "loss": 0.2535, + "step": 5709 + }, + { + "epoch": 2.39, + "learning_rate": 1.6442439410742753e-05, + "loss": 0.2512, + "step": 5710 + }, + { + "epoch": 2.39, + "learning_rate": 1.6441142635414568e-05, + "loss": 0.2478, + "step": 5711 + }, + { + "epoch": 2.39, + "learning_rate": 1.6439845674941458e-05, + "loss": 0.2688, + "step": 5712 + }, + { + "epoch": 2.39, + "learning_rate": 1.6438548529360698e-05, + "loss": 0.2423, + "step": 5713 + }, + { + "epoch": 2.39, + "learning_rate": 1.6437251198709567e-05, + "loss": 0.2904, + "step": 5714 + }, + { + "epoch": 2.39, + "learning_rate": 1.643595368302537e-05, + "loss": 0.2544, + "step": 5715 + }, + { + "epoch": 2.39, + "learning_rate": 1.643465598234539e-05, + "loss": 0.2706, + "step": 5716 + }, + { + "epoch": 2.39, + "learning_rate": 1.6433358096706937e-05, + "loss": 0.2799, + "step": 5717 + }, + { + "epoch": 2.39, + "learning_rate": 1.6432060026147313e-05, + "loss": 0.2739, + "step": 5718 + }, + { + "epoch": 2.39, + "learning_rate": 1.6430761770703836e-05, + "loss": 0.2693, + "step": 5719 + }, + { + "epoch": 2.39, + "learning_rate": 1.6429463330413812e-05, + "loss": 0.2574, + "step": 5720 + }, + { + "epoch": 2.4, + "learning_rate": 1.6428164705314572e-05, + "loss": 0.2597, + "step": 5721 + }, + { + "epoch": 2.4, + "learning_rate": 1.6426865895443446e-05, + "loss": 0.2549, + "step": 5722 + }, + { + "epoch": 2.4, + "learning_rate": 1.642556690083776e-05, + "loss": 0.2729, + "step": 5723 + }, + { + "epoch": 2.4, + "learning_rate": 1.6424267721534856e-05, + "loss": 0.2518, + "step": 5724 + }, + { + "epoch": 2.4, + "learning_rate": 1.6422968357572076e-05, + "loss": 0.2618, + "step": 5725 + }, + { + "epoch": 2.4, + "learning_rate": 1.642166880898677e-05, + "loss": 0.2654, + "step": 5726 + }, + { + "epoch": 2.4, + "learning_rate": 1.642036907581629e-05, + "loss": 0.2463, + "step": 5727 + }, + { + "epoch": 2.4, + "learning_rate": 1.6419069158098004e-05, + "loss": 0.2646, + "step": 5728 + }, + { + "epoch": 2.4, + "learning_rate": 1.6417769055869266e-05, + "loss": 0.2662, + "step": 5729 + }, + { + "epoch": 2.4, + "learning_rate": 1.641646876916745e-05, + "loss": 0.2697, + "step": 5730 + }, + { + "epoch": 2.4, + "learning_rate": 1.6415168298029943e-05, + "loss": 0.241, + "step": 5731 + }, + { + "epoch": 2.4, + "learning_rate": 1.6413867642494106e-05, + "loss": 0.2836, + "step": 5732 + }, + { + "epoch": 2.4, + "learning_rate": 1.641256680259734e-05, + "loss": 0.2531, + "step": 5733 + }, + { + "epoch": 2.4, + "learning_rate": 1.641126577837703e-05, + "loss": 0.2368, + "step": 5734 + }, + { + "epoch": 2.4, + "learning_rate": 1.640996456987058e-05, + "loss": 0.2644, + "step": 5735 + }, + { + "epoch": 2.4, + "learning_rate": 1.6408663177115382e-05, + "loss": 0.2916, + "step": 5736 + }, + { + "epoch": 2.4, + "learning_rate": 1.6407361600148845e-05, + "loss": 0.2774, + "step": 5737 + }, + { + "epoch": 2.4, + "learning_rate": 1.6406059839008387e-05, + "loss": 0.2308, + "step": 5738 + }, + { + "epoch": 2.4, + "learning_rate": 1.6404757893731424e-05, + "loss": 0.2639, + "step": 5739 + }, + { + "epoch": 2.4, + "learning_rate": 1.640345576435538e-05, + "loss": 0.2969, + "step": 5740 + }, + { + "epoch": 2.4, + "learning_rate": 1.640215345091768e-05, + "loss": 0.2689, + "step": 5741 + }, + { + "epoch": 2.4, + "learning_rate": 1.640085095345576e-05, + "loss": 0.2603, + "step": 5742 + }, + { + "epoch": 2.4, + "learning_rate": 1.6399548272007063e-05, + "loss": 0.2703, + "step": 5743 + }, + { + "epoch": 2.4, + "learning_rate": 1.6398245406609027e-05, + "loss": 0.2651, + "step": 5744 + }, + { + "epoch": 2.41, + "learning_rate": 1.6396942357299104e-05, + "loss": 0.2588, + "step": 5745 + }, + { + "epoch": 2.41, + "learning_rate": 1.639563912411475e-05, + "loss": 0.2697, + "step": 5746 + }, + { + "epoch": 2.41, + "learning_rate": 1.6394335707093423e-05, + "loss": 0.2693, + "step": 5747 + }, + { + "epoch": 2.41, + "learning_rate": 1.6393032106272594e-05, + "loss": 0.2228, + "step": 5748 + }, + { + "epoch": 2.41, + "learning_rate": 1.639172832168973e-05, + "loss": 0.2704, + "step": 5749 + }, + { + "epoch": 2.41, + "learning_rate": 1.639042435338231e-05, + "loss": 0.2572, + "step": 5750 + }, + { + "epoch": 2.41, + "learning_rate": 1.6389120201387806e-05, + "loss": 0.2515, + "step": 5751 + }, + { + "epoch": 2.41, + "learning_rate": 1.6387815865743712e-05, + "loss": 0.2549, + "step": 5752 + }, + { + "epoch": 2.41, + "learning_rate": 1.638651134648752e-05, + "loss": 0.244, + "step": 5753 + }, + { + "epoch": 2.41, + "learning_rate": 1.638520664365673e-05, + "loss": 0.272, + "step": 5754 + }, + { + "epoch": 2.41, + "learning_rate": 1.638390175728884e-05, + "loss": 0.2539, + "step": 5755 + }, + { + "epoch": 2.41, + "learning_rate": 1.638259668742136e-05, + "loss": 0.251, + "step": 5756 + }, + { + "epoch": 2.41, + "learning_rate": 1.63812914340918e-05, + "loss": 0.2558, + "step": 5757 + }, + { + "epoch": 2.41, + "learning_rate": 1.6379985997337678e-05, + "loss": 0.2609, + "step": 5758 + }, + { + "epoch": 2.41, + "learning_rate": 1.6378680377196526e-05, + "loss": 0.2463, + "step": 5759 + }, + { + "epoch": 2.41, + "learning_rate": 1.6377374573705863e-05, + "loss": 0.2639, + "step": 5760 + }, + { + "epoch": 2.41, + "learning_rate": 1.6376068586903227e-05, + "loss": 0.272, + "step": 5761 + }, + { + "epoch": 2.41, + "learning_rate": 1.6374762416826157e-05, + "loss": 0.2646, + "step": 5762 + }, + { + "epoch": 2.41, + "learning_rate": 1.63734560635122e-05, + "loss": 0.251, + "step": 5763 + }, + { + "epoch": 2.41, + "learning_rate": 1.6372149526998905e-05, + "loss": 0.253, + "step": 5764 + }, + { + "epoch": 2.41, + "learning_rate": 1.6370842807323823e-05, + "loss": 0.243, + "step": 5765 + }, + { + "epoch": 2.41, + "learning_rate": 1.636953590452452e-05, + "loss": 0.2471, + "step": 5766 + }, + { + "epoch": 2.41, + "learning_rate": 1.636822881863856e-05, + "loss": 0.2734, + "step": 5767 + }, + { + "epoch": 2.41, + "learning_rate": 1.6366921549703513e-05, + "loss": 0.2282, + "step": 5768 + }, + { + "epoch": 2.42, + "learning_rate": 1.6365614097756956e-05, + "loss": 0.257, + "step": 5769 + }, + { + "epoch": 2.42, + "learning_rate": 1.636430646283647e-05, + "loss": 0.2875, + "step": 5770 + }, + { + "epoch": 2.42, + "learning_rate": 1.6362998644979645e-05, + "loss": 0.2659, + "step": 5771 + }, + { + "epoch": 2.42, + "learning_rate": 1.6361690644224066e-05, + "loss": 0.2742, + "step": 5772 + }, + { + "epoch": 2.42, + "learning_rate": 1.636038246060734e-05, + "loss": 0.2699, + "step": 5773 + }, + { + "epoch": 2.42, + "learning_rate": 1.635907409416706e-05, + "loss": 0.2658, + "step": 5774 + }, + { + "epoch": 2.42, + "learning_rate": 1.6357765544940837e-05, + "loss": 0.2769, + "step": 5775 + }, + { + "epoch": 2.42, + "learning_rate": 1.635645681296629e-05, + "loss": 0.2656, + "step": 5776 + }, + { + "epoch": 2.42, + "learning_rate": 1.6355147898281026e-05, + "loss": 0.2624, + "step": 5777 + }, + { + "epoch": 2.42, + "learning_rate": 1.6353838800922682e-05, + "loss": 0.2353, + "step": 5778 + }, + { + "epoch": 2.42, + "learning_rate": 1.6352529520928877e-05, + "loss": 0.2584, + "step": 5779 + }, + { + "epoch": 2.42, + "learning_rate": 1.6351220058337248e-05, + "loss": 0.2717, + "step": 5780 + }, + { + "epoch": 2.42, + "learning_rate": 1.6349910413185438e-05, + "loss": 0.2829, + "step": 5781 + }, + { + "epoch": 2.42, + "learning_rate": 1.6348600585511082e-05, + "loss": 0.2519, + "step": 5782 + }, + { + "epoch": 2.42, + "learning_rate": 1.634729057535184e-05, + "loss": 0.2674, + "step": 5783 + }, + { + "epoch": 2.42, + "learning_rate": 1.634598038274536e-05, + "loss": 0.2715, + "step": 5784 + }, + { + "epoch": 2.42, + "learning_rate": 1.6344670007729305e-05, + "loss": 0.2771, + "step": 5785 + }, + { + "epoch": 2.42, + "learning_rate": 1.6343359450341344e-05, + "loss": 0.2167, + "step": 5786 + }, + { + "epoch": 2.42, + "learning_rate": 1.6342048710619143e-05, + "loss": 0.2417, + "step": 5787 + }, + { + "epoch": 2.42, + "learning_rate": 1.6340737788600378e-05, + "loss": 0.2649, + "step": 5788 + }, + { + "epoch": 2.42, + "learning_rate": 1.6339426684322735e-05, + "loss": 0.2508, + "step": 5789 + }, + { + "epoch": 2.42, + "learning_rate": 1.6338115397823897e-05, + "loss": 0.2406, + "step": 5790 + }, + { + "epoch": 2.42, + "learning_rate": 1.6336803929141557e-05, + "loss": 0.2676, + "step": 5791 + }, + { + "epoch": 2.42, + "learning_rate": 1.6335492278313405e-05, + "loss": 0.2332, + "step": 5792 + }, + { + "epoch": 2.43, + "learning_rate": 1.6334180445377156e-05, + "loss": 0.251, + "step": 5793 + }, + { + "epoch": 2.43, + "learning_rate": 1.6332868430370508e-05, + "loss": 0.2471, + "step": 5794 + }, + { + "epoch": 2.43, + "learning_rate": 1.6331556233331178e-05, + "loss": 0.2658, + "step": 5795 + }, + { + "epoch": 2.43, + "learning_rate": 1.633024385429688e-05, + "loss": 0.2872, + "step": 5796 + }, + { + "epoch": 2.43, + "learning_rate": 1.6328931293305345e-05, + "loss": 0.2564, + "step": 5797 + }, + { + "epoch": 2.43, + "learning_rate": 1.632761855039429e-05, + "loss": 0.2626, + "step": 5798 + }, + { + "epoch": 2.43, + "learning_rate": 1.632630562560146e-05, + "loss": 0.2719, + "step": 5799 + }, + { + "epoch": 2.43, + "learning_rate": 1.6324992518964586e-05, + "loss": 0.248, + "step": 5800 + }, + { + "epoch": 2.43, + "learning_rate": 1.6323679230521413e-05, + "loss": 0.2723, + "step": 5801 + }, + { + "epoch": 2.43, + "learning_rate": 1.6322365760309698e-05, + "loss": 0.2574, + "step": 5802 + }, + { + "epoch": 2.43, + "learning_rate": 1.6321052108367187e-05, + "loss": 0.2379, + "step": 5803 + }, + { + "epoch": 2.43, + "learning_rate": 1.631973827473164e-05, + "loss": 0.2748, + "step": 5804 + }, + { + "epoch": 2.43, + "learning_rate": 1.6318424259440827e-05, + "loss": 0.2571, + "step": 5805 + }, + { + "epoch": 2.43, + "learning_rate": 1.631711006253251e-05, + "loss": 0.266, + "step": 5806 + }, + { + "epoch": 2.43, + "learning_rate": 1.6315795684044476e-05, + "loss": 0.2561, + "step": 5807 + }, + { + "epoch": 2.43, + "learning_rate": 1.63144811240145e-05, + "loss": 0.2786, + "step": 5808 + }, + { + "epoch": 2.43, + "learning_rate": 1.6313166382480362e-05, + "loss": 0.2648, + "step": 5809 + }, + { + "epoch": 2.43, + "learning_rate": 1.631185145947986e-05, + "loss": 0.2931, + "step": 5810 + }, + { + "epoch": 2.43, + "learning_rate": 1.631053635505079e-05, + "loss": 0.2583, + "step": 5811 + }, + { + "epoch": 2.43, + "learning_rate": 1.6309221069230954e-05, + "loss": 0.2413, + "step": 5812 + }, + { + "epoch": 2.43, + "learning_rate": 1.6307905602058158e-05, + "loss": 0.2587, + "step": 5813 + }, + { + "epoch": 2.43, + "learning_rate": 1.6306589953570207e-05, + "loss": 0.2623, + "step": 5814 + }, + { + "epoch": 2.43, + "learning_rate": 1.6305274123804934e-05, + "loss": 0.2557, + "step": 5815 + }, + { + "epoch": 2.44, + "learning_rate": 1.6303958112800144e-05, + "loss": 0.2457, + "step": 5816 + }, + { + "epoch": 2.44, + "learning_rate": 1.630264192059367e-05, + "loss": 0.2665, + "step": 5817 + }, + { + "epoch": 2.44, + "learning_rate": 1.6301325547223353e-05, + "loss": 0.2545, + "step": 5818 + }, + { + "epoch": 2.44, + "learning_rate": 1.630000899272702e-05, + "loss": 0.2427, + "step": 5819 + }, + { + "epoch": 2.44, + "learning_rate": 1.6298692257142525e-05, + "loss": 0.2687, + "step": 5820 + }, + { + "epoch": 2.44, + "learning_rate": 1.6297375340507705e-05, + "loss": 0.2529, + "step": 5821 + }, + { + "epoch": 2.44, + "learning_rate": 1.629605824286042e-05, + "loss": 0.2595, + "step": 5822 + }, + { + "epoch": 2.44, + "learning_rate": 1.629474096423853e-05, + "loss": 0.2597, + "step": 5823 + }, + { + "epoch": 2.44, + "learning_rate": 1.6293423504679898e-05, + "loss": 0.2469, + "step": 5824 + }, + { + "epoch": 2.44, + "learning_rate": 1.6292105864222386e-05, + "loss": 0.2507, + "step": 5825 + }, + { + "epoch": 2.44, + "learning_rate": 1.629078804290388e-05, + "loss": 0.2638, + "step": 5826 + }, + { + "epoch": 2.44, + "learning_rate": 1.628947004076225e-05, + "loss": 0.2452, + "step": 5827 + }, + { + "epoch": 2.44, + "learning_rate": 1.628815185783539e-05, + "loss": 0.2736, + "step": 5828 + }, + { + "epoch": 2.44, + "learning_rate": 1.628683349416118e-05, + "loss": 0.2595, + "step": 5829 + }, + { + "epoch": 2.44, + "learning_rate": 1.6285514949777522e-05, + "loss": 0.2602, + "step": 5830 + }, + { + "epoch": 2.44, + "learning_rate": 1.628419622472231e-05, + "loss": 0.2559, + "step": 5831 + }, + { + "epoch": 2.44, + "learning_rate": 1.6282877319033462e-05, + "loss": 0.2772, + "step": 5832 + }, + { + "epoch": 2.44, + "learning_rate": 1.6281558232748874e-05, + "loss": 0.2583, + "step": 5833 + }, + { + "epoch": 2.44, + "learning_rate": 1.6280238965906474e-05, + "loss": 0.2928, + "step": 5834 + }, + { + "epoch": 2.44, + "learning_rate": 1.6278919518544175e-05, + "loss": 0.2403, + "step": 5835 + }, + { + "epoch": 2.44, + "learning_rate": 1.627759989069991e-05, + "loss": 0.2579, + "step": 5836 + }, + { + "epoch": 2.44, + "learning_rate": 1.6276280082411604e-05, + "loss": 0.2968, + "step": 5837 + }, + { + "epoch": 2.44, + "learning_rate": 1.6274960093717192e-05, + "loss": 0.2501, + "step": 5838 + }, + { + "epoch": 2.44, + "learning_rate": 1.6273639924654626e-05, + "loss": 0.2598, + "step": 5839 + }, + { + "epoch": 2.45, + "learning_rate": 1.627231957526185e-05, + "loss": 0.2334, + "step": 5840 + }, + { + "epoch": 2.45, + "learning_rate": 1.6270999045576808e-05, + "loss": 0.2504, + "step": 5841 + }, + { + "epoch": 2.45, + "learning_rate": 1.6269678335637464e-05, + "loss": 0.2664, + "step": 5842 + }, + { + "epoch": 2.45, + "learning_rate": 1.626835744548178e-05, + "loss": 0.2308, + "step": 5843 + }, + { + "epoch": 2.45, + "learning_rate": 1.6267036375147728e-05, + "loss": 0.2584, + "step": 5844 + }, + { + "epoch": 2.45, + "learning_rate": 1.6265715124673272e-05, + "loss": 0.2493, + "step": 5845 + }, + { + "epoch": 2.45, + "learning_rate": 1.6264393694096396e-05, + "loss": 0.2382, + "step": 5846 + }, + { + "epoch": 2.45, + "learning_rate": 1.6263072083455084e-05, + "loss": 0.2394, + "step": 5847 + }, + { + "epoch": 2.45, + "learning_rate": 1.626175029278732e-05, + "loss": 0.2658, + "step": 5848 + }, + { + "epoch": 2.45, + "learning_rate": 1.62604283221311e-05, + "loss": 0.2416, + "step": 5849 + }, + { + "epoch": 2.45, + "learning_rate": 1.6259106171524424e-05, + "loss": 0.2642, + "step": 5850 + }, + { + "epoch": 2.45, + "learning_rate": 1.6257783841005297e-05, + "loss": 0.2426, + "step": 5851 + }, + { + "epoch": 2.45, + "learning_rate": 1.6256461330611727e-05, + "loss": 0.2554, + "step": 5852 + }, + { + "epoch": 2.45, + "learning_rate": 1.6255138640381723e-05, + "loss": 0.2516, + "step": 5853 + }, + { + "epoch": 2.45, + "learning_rate": 1.6253815770353316e-05, + "loss": 0.2651, + "step": 5854 + }, + { + "epoch": 2.45, + "learning_rate": 1.625249272056452e-05, + "loss": 0.2797, + "step": 5855 + }, + { + "epoch": 2.45, + "learning_rate": 1.625116949105337e-05, + "loss": 0.2711, + "step": 5856 + }, + { + "epoch": 2.45, + "learning_rate": 1.62498460818579e-05, + "loss": 0.2555, + "step": 5857 + }, + { + "epoch": 2.45, + "learning_rate": 1.6248522493016153e-05, + "loss": 0.2479, + "step": 5858 + }, + { + "epoch": 2.45, + "learning_rate": 1.624719872456617e-05, + "loss": 0.2662, + "step": 5859 + }, + { + "epoch": 2.45, + "learning_rate": 1.6245874776546005e-05, + "loss": 0.2627, + "step": 5860 + }, + { + "epoch": 2.45, + "learning_rate": 1.624455064899371e-05, + "loss": 0.247, + "step": 5861 + }, + { + "epoch": 2.45, + "learning_rate": 1.6243226341947348e-05, + "loss": 0.266, + "step": 5862 + }, + { + "epoch": 2.45, + "learning_rate": 1.624190185544498e-05, + "loss": 0.25, + "step": 5863 + }, + { + "epoch": 2.46, + "learning_rate": 1.6240577189524688e-05, + "loss": 0.2389, + "step": 5864 + }, + { + "epoch": 2.46, + "learning_rate": 1.6239252344224545e-05, + "loss": 0.2448, + "step": 5865 + }, + { + "epoch": 2.46, + "learning_rate": 1.6237927319582624e-05, + "loss": 0.2697, + "step": 5866 + }, + { + "epoch": 2.46, + "learning_rate": 1.623660211563702e-05, + "loss": 0.2607, + "step": 5867 + }, + { + "epoch": 2.46, + "learning_rate": 1.6235276732425823e-05, + "loss": 0.2711, + "step": 5868 + }, + { + "epoch": 2.46, + "learning_rate": 1.623395116998713e-05, + "loss": 0.2854, + "step": 5869 + }, + { + "epoch": 2.46, + "learning_rate": 1.6232625428359042e-05, + "loss": 0.2689, + "step": 5870 + }, + { + "epoch": 2.46, + "learning_rate": 1.6231299507579668e-05, + "loss": 0.2591, + "step": 5871 + }, + { + "epoch": 2.46, + "learning_rate": 1.6229973407687117e-05, + "loss": 0.2603, + "step": 5872 + }, + { + "epoch": 2.46, + "learning_rate": 1.622864712871951e-05, + "loss": 0.2572, + "step": 5873 + }, + { + "epoch": 2.46, + "learning_rate": 1.6227320670714966e-05, + "loss": 0.2879, + "step": 5874 + }, + { + "epoch": 2.46, + "learning_rate": 1.6225994033711617e-05, + "loss": 0.2058, + "step": 5875 + }, + { + "epoch": 2.46, + "learning_rate": 1.6224667217747594e-05, + "loss": 0.26, + "step": 5876 + }, + { + "epoch": 2.46, + "learning_rate": 1.6223340222861037e-05, + "loss": 0.2241, + "step": 5877 + }, + { + "epoch": 2.46, + "learning_rate": 1.6222013049090086e-05, + "loss": 0.2499, + "step": 5878 + }, + { + "epoch": 2.46, + "learning_rate": 1.622068569647289e-05, + "loss": 0.2711, + "step": 5879 + }, + { + "epoch": 2.46, + "learning_rate": 1.6219358165047606e-05, + "loss": 0.2468, + "step": 5880 + }, + { + "epoch": 2.46, + "learning_rate": 1.6218030454852395e-05, + "loss": 0.2711, + "step": 5881 + }, + { + "epoch": 2.46, + "learning_rate": 1.621670256592541e-05, + "loss": 0.2464, + "step": 5882 + }, + { + "epoch": 2.46, + "learning_rate": 1.6215374498304827e-05, + "loss": 0.2739, + "step": 5883 + }, + { + "epoch": 2.46, + "learning_rate": 1.6214046252028823e-05, + "loss": 0.2529, + "step": 5884 + }, + { + "epoch": 2.46, + "learning_rate": 1.621271782713557e-05, + "loss": 0.2508, + "step": 5885 + }, + { + "epoch": 2.46, + "learning_rate": 1.621138922366326e-05, + "loss": 0.245, + "step": 5886 + }, + { + "epoch": 2.46, + "learning_rate": 1.6210060441650073e-05, + "loss": 0.2693, + "step": 5887 + }, + { + "epoch": 2.47, + "learning_rate": 1.6208731481134214e-05, + "loss": 0.2377, + "step": 5888 + }, + { + "epoch": 2.47, + "learning_rate": 1.6207402342153876e-05, + "loss": 0.2442, + "step": 5889 + }, + { + "epoch": 2.47, + "learning_rate": 1.6206073024747264e-05, + "loss": 0.2673, + "step": 5890 + }, + { + "epoch": 2.47, + "learning_rate": 1.6204743528952592e-05, + "loss": 0.2758, + "step": 5891 + }, + { + "epoch": 2.47, + "learning_rate": 1.6203413854808076e-05, + "loss": 0.2612, + "step": 5892 + }, + { + "epoch": 2.47, + "learning_rate": 1.620208400235193e-05, + "loss": 0.2591, + "step": 5893 + }, + { + "epoch": 2.47, + "learning_rate": 1.6200753971622387e-05, + "loss": 0.2512, + "step": 5894 + }, + { + "epoch": 2.47, + "learning_rate": 1.619942376265767e-05, + "loss": 0.2587, + "step": 5895 + }, + { + "epoch": 2.47, + "learning_rate": 1.619809337549602e-05, + "loss": 0.2153, + "step": 5896 + }, + { + "epoch": 2.47, + "learning_rate": 1.6196762810175674e-05, + "loss": 0.2638, + "step": 5897 + }, + { + "epoch": 2.47, + "learning_rate": 1.6195432066734878e-05, + "loss": 0.2553, + "step": 5898 + }, + { + "epoch": 2.47, + "learning_rate": 1.619410114521189e-05, + "loss": 0.2529, + "step": 5899 + }, + { + "epoch": 2.47, + "learning_rate": 1.619277004564496e-05, + "loss": 0.2724, + "step": 5900 + }, + { + "epoch": 2.47, + "learning_rate": 1.619143876807235e-05, + "loss": 0.2458, + "step": 5901 + }, + { + "epoch": 2.47, + "learning_rate": 1.6190107312532327e-05, + "loss": 0.2857, + "step": 5902 + }, + { + "epoch": 2.47, + "learning_rate": 1.618877567906316e-05, + "loss": 0.2431, + "step": 5903 + }, + { + "epoch": 2.47, + "learning_rate": 1.618744386770313e-05, + "loss": 0.2515, + "step": 5904 + }, + { + "epoch": 2.47, + "learning_rate": 1.6186111878490515e-05, + "loss": 0.2522, + "step": 5905 + }, + { + "epoch": 2.47, + "learning_rate": 1.6184779711463607e-05, + "loss": 0.2592, + "step": 5906 + }, + { + "epoch": 2.47, + "learning_rate": 1.618344736666069e-05, + "loss": 0.2663, + "step": 5907 + }, + { + "epoch": 2.47, + "learning_rate": 1.618211484412007e-05, + "loss": 0.2633, + "step": 5908 + }, + { + "epoch": 2.47, + "learning_rate": 1.6180782143880042e-05, + "loss": 0.2422, + "step": 5909 + }, + { + "epoch": 2.47, + "learning_rate": 1.6179449265978913e-05, + "loss": 0.2889, + "step": 5910 + }, + { + "epoch": 2.47, + "learning_rate": 1.6178116210455006e-05, + "loss": 0.3035, + "step": 5911 + }, + { + "epoch": 2.48, + "learning_rate": 1.6176782977346626e-05, + "loss": 0.2714, + "step": 5912 + }, + { + "epoch": 2.48, + "learning_rate": 1.61754495666921e-05, + "loss": 0.254, + "step": 5913 + }, + { + "epoch": 2.48, + "learning_rate": 1.617411597852976e-05, + "loss": 0.2454, + "step": 5914 + }, + { + "epoch": 2.48, + "learning_rate": 1.617278221289793e-05, + "loss": 0.2544, + "step": 5915 + }, + { + "epoch": 2.48, + "learning_rate": 1.6171448269834956e-05, + "loss": 0.2736, + "step": 5916 + }, + { + "epoch": 2.48, + "learning_rate": 1.6170114149379178e-05, + "loss": 0.2428, + "step": 5917 + }, + { + "epoch": 2.48, + "learning_rate": 1.6168779851568947e-05, + "loss": 0.2423, + "step": 5918 + }, + { + "epoch": 2.48, + "learning_rate": 1.6167445376442604e-05, + "loss": 0.2222, + "step": 5919 + }, + { + "epoch": 2.48, + "learning_rate": 1.6166110724038524e-05, + "loss": 0.2606, + "step": 5920 + }, + { + "epoch": 2.48, + "learning_rate": 1.616477589439506e-05, + "loss": 0.2681, + "step": 5921 + }, + { + "epoch": 2.48, + "learning_rate": 1.6163440887550587e-05, + "loss": 0.2595, + "step": 5922 + }, + { + "epoch": 2.48, + "learning_rate": 1.616210570354347e-05, + "loss": 0.2684, + "step": 5923 + }, + { + "epoch": 2.48, + "learning_rate": 1.6160770342412097e-05, + "loss": 0.2561, + "step": 5924 + }, + { + "epoch": 2.48, + "learning_rate": 1.6159434804194845e-05, + "loss": 0.2489, + "step": 5925 + }, + { + "epoch": 2.48, + "learning_rate": 1.615809908893011e-05, + "loss": 0.2594, + "step": 5926 + }, + { + "epoch": 2.48, + "learning_rate": 1.6156763196656276e-05, + "loss": 0.2583, + "step": 5927 + }, + { + "epoch": 2.48, + "learning_rate": 1.6155427127411748e-05, + "loss": 0.2552, + "step": 5928 + }, + { + "epoch": 2.48, + "learning_rate": 1.615409088123493e-05, + "loss": 0.2666, + "step": 5929 + }, + { + "epoch": 2.48, + "learning_rate": 1.615275445816423e-05, + "loss": 0.2609, + "step": 5930 + }, + { + "epoch": 2.48, + "learning_rate": 1.6151417858238064e-05, + "loss": 0.2475, + "step": 5931 + }, + { + "epoch": 2.48, + "learning_rate": 1.615008108149485e-05, + "loss": 0.2644, + "step": 5932 + }, + { + "epoch": 2.48, + "learning_rate": 1.6148744127973014e-05, + "loss": 0.2486, + "step": 5933 + }, + { + "epoch": 2.48, + "learning_rate": 1.614740699771098e-05, + "loss": 0.2463, + "step": 5934 + }, + { + "epoch": 2.48, + "learning_rate": 1.614606969074719e-05, + "loss": 0.2454, + "step": 5935 + }, + { + "epoch": 2.49, + "learning_rate": 1.6144732207120083e-05, + "loss": 0.2456, + "step": 5936 + }, + { + "epoch": 2.49, + "learning_rate": 1.61433945468681e-05, + "loss": 0.2615, + "step": 5937 + }, + { + "epoch": 2.49, + "learning_rate": 1.614205671002969e-05, + "loss": 0.2378, + "step": 5938 + }, + { + "epoch": 2.49, + "learning_rate": 1.6140718696643313e-05, + "loss": 0.2868, + "step": 5939 + }, + { + "epoch": 2.49, + "learning_rate": 1.6139380506747424e-05, + "loss": 0.2496, + "step": 5940 + }, + { + "epoch": 2.49, + "learning_rate": 1.6138042140380493e-05, + "loss": 0.2343, + "step": 5941 + }, + { + "epoch": 2.49, + "learning_rate": 1.6136703597580984e-05, + "loss": 0.2396, + "step": 5942 + }, + { + "epoch": 2.49, + "learning_rate": 1.613536487838738e-05, + "loss": 0.2665, + "step": 5943 + }, + { + "epoch": 2.49, + "learning_rate": 1.6134025982838153e-05, + "loss": 0.2772, + "step": 5944 + }, + { + "epoch": 2.49, + "learning_rate": 1.6132686910971792e-05, + "loss": 0.2544, + "step": 5945 + }, + { + "epoch": 2.49, + "learning_rate": 1.6131347662826793e-05, + "loss": 0.2489, + "step": 5946 + }, + { + "epoch": 2.49, + "learning_rate": 1.613000823844164e-05, + "loss": 0.2798, + "step": 5947 + }, + { + "epoch": 2.49, + "learning_rate": 1.6128668637854847e-05, + "loss": 0.2367, + "step": 5948 + }, + { + "epoch": 2.49, + "learning_rate": 1.6127328861104906e-05, + "loss": 0.2688, + "step": 5949 + }, + { + "epoch": 2.49, + "learning_rate": 1.6125988908230337e-05, + "loss": 0.2514, + "step": 5950 + }, + { + "epoch": 2.49, + "learning_rate": 1.6124648779269654e-05, + "loss": 0.2594, + "step": 5951 + }, + { + "epoch": 2.49, + "learning_rate": 1.6123308474261374e-05, + "loss": 0.2542, + "step": 5952 + }, + { + "epoch": 2.49, + "learning_rate": 1.6121967993244026e-05, + "loss": 0.2548, + "step": 5953 + }, + { + "epoch": 2.49, + "learning_rate": 1.6120627336256142e-05, + "loss": 0.2928, + "step": 5954 + }, + { + "epoch": 2.49, + "learning_rate": 1.6119286503336256e-05, + "loss": 0.2584, + "step": 5955 + }, + { + "epoch": 2.49, + "learning_rate": 1.611794549452291e-05, + "loss": 0.2732, + "step": 5956 + }, + { + "epoch": 2.49, + "learning_rate": 1.611660430985465e-05, + "loss": 0.2218, + "step": 5957 + }, + { + "epoch": 2.49, + "learning_rate": 1.6115262949370025e-05, + "loss": 0.2893, + "step": 5958 + }, + { + "epoch": 2.49, + "learning_rate": 1.6113921413107597e-05, + "loss": 0.2684, + "step": 5959 + }, + { + "epoch": 2.5, + "learning_rate": 1.6112579701105923e-05, + "loss": 0.2812, + "step": 5960 + }, + { + "epoch": 2.5, + "learning_rate": 1.6111237813403568e-05, + "loss": 0.2759, + "step": 5961 + }, + { + "epoch": 2.5, + "learning_rate": 1.61098957500391e-05, + "loss": 0.285, + "step": 5962 + }, + { + "epoch": 2.5, + "learning_rate": 1.6108553511051106e-05, + "loss": 0.2418, + "step": 5963 + }, + { + "epoch": 2.5, + "learning_rate": 1.610721109647816e-05, + "loss": 0.2605, + "step": 5964 + }, + { + "epoch": 2.5, + "learning_rate": 1.610586850635885e-05, + "loss": 0.263, + "step": 5965 + }, + { + "epoch": 2.5, + "learning_rate": 1.6104525740731772e-05, + "loss": 0.28, + "step": 5966 + }, + { + "epoch": 2.5, + "learning_rate": 1.6103182799635514e-05, + "loss": 0.246, + "step": 5967 + }, + { + "epoch": 2.5, + "learning_rate": 1.6101839683108686e-05, + "loss": 0.2725, + "step": 5968 + }, + { + "epoch": 2.5, + "learning_rate": 1.610049639118989e-05, + "loss": 0.2355, + "step": 5969 + }, + { + "epoch": 2.5, + "learning_rate": 1.6099152923917737e-05, + "loss": 0.2961, + "step": 5970 + }, + { + "epoch": 2.5, + "learning_rate": 1.6097809281330848e-05, + "loss": 0.2723, + "step": 5971 + }, + { + "epoch": 2.5, + "learning_rate": 1.609646546346784e-05, + "loss": 0.2695, + "step": 5972 + }, + { + "epoch": 2.5, + "learning_rate": 1.6095121470367345e-05, + "loss": 0.287, + "step": 5973 + }, + { + "epoch": 2.5, + "learning_rate": 1.609377730206799e-05, + "loss": 0.2501, + "step": 5974 + }, + { + "epoch": 2.5, + "learning_rate": 1.6092432958608413e-05, + "loss": 0.2542, + "step": 5975 + }, + { + "epoch": 2.5, + "learning_rate": 1.609108844002726e-05, + "loss": 0.259, + "step": 5976 + }, + { + "epoch": 2.5, + "learning_rate": 1.608974374636317e-05, + "loss": 0.2713, + "step": 5977 + }, + { + "epoch": 2.5, + "learning_rate": 1.6088398877654806e-05, + "loss": 0.2765, + "step": 5978 + }, + { + "epoch": 2.5, + "learning_rate": 1.6087053833940815e-05, + "loss": 0.2254, + "step": 5979 + }, + { + "epoch": 2.5, + "learning_rate": 1.6085708615259863e-05, + "loss": 0.2434, + "step": 5980 + }, + { + "epoch": 2.5, + "learning_rate": 1.608436322165062e-05, + "loss": 0.2638, + "step": 5981 + }, + { + "epoch": 2.5, + "learning_rate": 1.6083017653151752e-05, + "loss": 0.2749, + "step": 5982 + }, + { + "epoch": 2.5, + "learning_rate": 1.6081671909801944e-05, + "loss": 0.2938, + "step": 5983 + }, + { + "epoch": 2.51, + "learning_rate": 1.6080325991639866e-05, + "loss": 0.2419, + "step": 5984 + }, + { + "epoch": 2.51, + "learning_rate": 1.6078979898704223e-05, + "loss": 0.2736, + "step": 5985 + }, + { + "epoch": 2.51, + "learning_rate": 1.607763363103369e-05, + "loss": 0.2498, + "step": 5986 + }, + { + "epoch": 2.51, + "learning_rate": 1.6076287188666972e-05, + "loss": 0.2468, + "step": 5987 + }, + { + "epoch": 2.51, + "learning_rate": 1.6074940571642773e-05, + "loss": 0.2543, + "step": 5988 + }, + { + "epoch": 2.51, + "learning_rate": 1.6073593779999795e-05, + "loss": 0.2549, + "step": 5989 + }, + { + "epoch": 2.51, + "learning_rate": 1.6072246813776758e-05, + "loss": 0.2918, + "step": 5990 + }, + { + "epoch": 2.51, + "learning_rate": 1.6070899673012372e-05, + "loss": 0.2755, + "step": 5991 + }, + { + "epoch": 2.51, + "learning_rate": 1.6069552357745363e-05, + "loss": 0.2557, + "step": 5992 + }, + { + "epoch": 2.51, + "learning_rate": 1.6068204868014455e-05, + "loss": 0.2732, + "step": 5993 + }, + { + "epoch": 2.51, + "learning_rate": 1.6066857203858387e-05, + "loss": 0.2549, + "step": 5994 + }, + { + "epoch": 2.51, + "learning_rate": 1.6065509365315893e-05, + "loss": 0.2497, + "step": 5995 + }, + { + "epoch": 2.51, + "learning_rate": 1.6064161352425708e-05, + "loss": 0.2364, + "step": 5996 + }, + { + "epoch": 2.51, + "learning_rate": 1.6062813165226594e-05, + "loss": 0.2453, + "step": 5997 + }, + { + "epoch": 2.51, + "learning_rate": 1.606146480375729e-05, + "loss": 0.2279, + "step": 5998 + }, + { + "epoch": 2.51, + "learning_rate": 1.606011626805656e-05, + "loss": 0.2577, + "step": 5999 + }, + { + "epoch": 2.51, + "learning_rate": 1.6058767558163168e-05, + "loss": 0.2731, + "step": 6000 + }, + { + "epoch": 2.51, + "learning_rate": 1.6057418674115876e-05, + "loss": 0.2782, + "step": 6001 + }, + { + "epoch": 2.51, + "learning_rate": 1.605606961595346e-05, + "loss": 0.2451, + "step": 6002 + }, + { + "epoch": 2.51, + "learning_rate": 1.60547203837147e-05, + "loss": 0.2513, + "step": 6003 + }, + { + "epoch": 2.51, + "learning_rate": 1.605337097743837e-05, + "loss": 0.2531, + "step": 6004 + }, + { + "epoch": 2.51, + "learning_rate": 1.605202139716327e-05, + "loss": 0.2754, + "step": 6005 + }, + { + "epoch": 2.51, + "learning_rate": 1.605067164292818e-05, + "loss": 0.2522, + "step": 6006 + }, + { + "epoch": 2.51, + "learning_rate": 1.6049321714771907e-05, + "loss": 0.2427, + "step": 6007 + }, + { + "epoch": 2.52, + "learning_rate": 1.6047971612733246e-05, + "loss": 0.2295, + "step": 6008 + }, + { + "epoch": 2.52, + "learning_rate": 1.6046621336851007e-05, + "loss": 0.272, + "step": 6009 + }, + { + "epoch": 2.52, + "learning_rate": 1.6045270887164005e-05, + "loss": 0.2655, + "step": 6010 + }, + { + "epoch": 2.52, + "learning_rate": 1.6043920263711056e-05, + "loss": 0.2788, + "step": 6011 + }, + { + "epoch": 2.52, + "learning_rate": 1.6042569466530985e-05, + "loss": 0.2805, + "step": 6012 + }, + { + "epoch": 2.52, + "learning_rate": 1.604121849566261e-05, + "loss": 0.2553, + "step": 6013 + }, + { + "epoch": 2.52, + "learning_rate": 1.6039867351144778e-05, + "loss": 0.2407, + "step": 6014 + }, + { + "epoch": 2.52, + "learning_rate": 1.6038516033016314e-05, + "loss": 0.2728, + "step": 6015 + }, + { + "epoch": 2.52, + "learning_rate": 1.6037164541316067e-05, + "loss": 0.2442, + "step": 6016 + }, + { + "epoch": 2.52, + "learning_rate": 1.603581287608288e-05, + "loss": 0.2559, + "step": 6017 + }, + { + "epoch": 2.52, + "learning_rate": 1.603446103735561e-05, + "loss": 0.2662, + "step": 6018 + }, + { + "epoch": 2.52, + "learning_rate": 1.6033109025173113e-05, + "loss": 0.2541, + "step": 6019 + }, + { + "epoch": 2.52, + "learning_rate": 1.6031756839574247e-05, + "loss": 0.2909, + "step": 6020 + }, + { + "epoch": 2.52, + "learning_rate": 1.6030404480597888e-05, + "loss": 0.2366, + "step": 6021 + }, + { + "epoch": 2.52, + "learning_rate": 1.60290519482829e-05, + "loss": 0.2578, + "step": 6022 + }, + { + "epoch": 2.52, + "learning_rate": 1.6027699242668166e-05, + "loss": 0.2686, + "step": 6023 + }, + { + "epoch": 2.52, + "learning_rate": 1.6026346363792565e-05, + "loss": 0.2634, + "step": 6024 + }, + { + "epoch": 2.52, + "learning_rate": 1.6024993311694986e-05, + "loss": 0.267, + "step": 6025 + }, + { + "epoch": 2.52, + "learning_rate": 1.6023640086414318e-05, + "loss": 0.2677, + "step": 6026 + }, + { + "epoch": 2.52, + "learning_rate": 1.6022286687989466e-05, + "loss": 0.2588, + "step": 6027 + }, + { + "epoch": 2.52, + "learning_rate": 1.6020933116459324e-05, + "loss": 0.254, + "step": 6028 + }, + { + "epoch": 2.52, + "learning_rate": 1.6019579371862803e-05, + "loss": 0.2767, + "step": 6029 + }, + { + "epoch": 2.52, + "learning_rate": 1.6018225454238816e-05, + "loss": 0.248, + "step": 6030 + }, + { + "epoch": 2.53, + "learning_rate": 1.6016871363626277e-05, + "loss": 0.2448, + "step": 6031 + }, + { + "epoch": 2.53, + "learning_rate": 1.6015517100064114e-05, + "loss": 0.2472, + "step": 6032 + }, + { + "epoch": 2.53, + "learning_rate": 1.601416266359124e-05, + "loss": 0.2665, + "step": 6033 + }, + { + "epoch": 2.53, + "learning_rate": 1.6012808054246608e-05, + "loss": 0.2622, + "step": 6034 + }, + { + "epoch": 2.53, + "learning_rate": 1.601145327206914e-05, + "loss": 0.2483, + "step": 6035 + }, + { + "epoch": 2.53, + "learning_rate": 1.601009831709778e-05, + "loss": 0.2636, + "step": 6036 + }, + { + "epoch": 2.53, + "learning_rate": 1.600874318937148e-05, + "loss": 0.2706, + "step": 6037 + }, + { + "epoch": 2.53, + "learning_rate": 1.600738788892919e-05, + "loss": 0.2803, + "step": 6038 + }, + { + "epoch": 2.53, + "learning_rate": 1.6006032415809866e-05, + "loss": 0.2737, + "step": 6039 + }, + { + "epoch": 2.53, + "learning_rate": 1.6004676770052466e-05, + "loss": 0.2622, + "step": 6040 + }, + { + "epoch": 2.53, + "learning_rate": 1.600332095169596e-05, + "loss": 0.2615, + "step": 6041 + }, + { + "epoch": 2.53, + "learning_rate": 1.6001964960779326e-05, + "loss": 0.2653, + "step": 6042 + }, + { + "epoch": 2.53, + "learning_rate": 1.600060879734153e-05, + "loss": 0.2388, + "step": 6043 + }, + { + "epoch": 2.53, + "learning_rate": 1.5999252461421564e-05, + "loss": 0.2292, + "step": 6044 + }, + { + "epoch": 2.53, + "learning_rate": 1.5997895953058407e-05, + "loss": 0.2427, + "step": 6045 + }, + { + "epoch": 2.53, + "learning_rate": 1.5996539272291052e-05, + "loss": 0.2385, + "step": 6046 + }, + { + "epoch": 2.53, + "learning_rate": 1.59951824191585e-05, + "loss": 0.2729, + "step": 6047 + }, + { + "epoch": 2.53, + "learning_rate": 1.5993825393699746e-05, + "loss": 0.2514, + "step": 6048 + }, + { + "epoch": 2.53, + "learning_rate": 1.59924681959538e-05, + "loss": 0.2577, + "step": 6049 + }, + { + "epoch": 2.53, + "learning_rate": 1.5991110825959676e-05, + "loss": 0.245, + "step": 6050 + }, + { + "epoch": 2.53, + "learning_rate": 1.598975328375639e-05, + "loss": 0.2652, + "step": 6051 + }, + { + "epoch": 2.53, + "learning_rate": 1.5988395569382954e-05, + "loss": 0.2477, + "step": 6052 + }, + { + "epoch": 2.53, + "learning_rate": 1.5987037682878405e-05, + "loss": 0.2296, + "step": 6053 + }, + { + "epoch": 2.53, + "learning_rate": 1.5985679624281772e-05, + "loss": 0.244, + "step": 6054 + }, + { + "epoch": 2.54, + "learning_rate": 1.5984321393632087e-05, + "loss": 0.2329, + "step": 6055 + }, + { + "epoch": 2.54, + "learning_rate": 1.5982962990968397e-05, + "loss": 0.2332, + "step": 6056 + }, + { + "epoch": 2.54, + "learning_rate": 1.5981604416329742e-05, + "loss": 0.243, + "step": 6057 + }, + { + "epoch": 2.54, + "learning_rate": 1.598024566975518e-05, + "loss": 0.2464, + "step": 6058 + }, + { + "epoch": 2.54, + "learning_rate": 1.597888675128376e-05, + "loss": 0.2263, + "step": 6059 + }, + { + "epoch": 2.54, + "learning_rate": 1.5977527660954545e-05, + "loss": 0.2715, + "step": 6060 + }, + { + "epoch": 2.54, + "learning_rate": 1.5976168398806602e-05, + "loss": 0.2625, + "step": 6061 + }, + { + "epoch": 2.54, + "learning_rate": 1.5974808964879005e-05, + "loss": 0.2729, + "step": 6062 + }, + { + "epoch": 2.54, + "learning_rate": 1.5973449359210823e-05, + "loss": 0.2417, + "step": 6063 + }, + { + "epoch": 2.54, + "learning_rate": 1.597208958184114e-05, + "loss": 0.2355, + "step": 6064 + }, + { + "epoch": 2.54, + "learning_rate": 1.5970729632809044e-05, + "loss": 0.2282, + "step": 6065 + }, + { + "epoch": 2.54, + "learning_rate": 1.596936951215362e-05, + "loss": 0.2426, + "step": 6066 + }, + { + "epoch": 2.54, + "learning_rate": 1.5968009219913966e-05, + "loss": 0.2719, + "step": 6067 + }, + { + "epoch": 2.54, + "learning_rate": 1.5966648756129187e-05, + "loss": 0.2581, + "step": 6068 + }, + { + "epoch": 2.54, + "learning_rate": 1.5965288120838383e-05, + "loss": 0.2915, + "step": 6069 + }, + { + "epoch": 2.54, + "learning_rate": 1.596392731408066e-05, + "loss": 0.2598, + "step": 6070 + }, + { + "epoch": 2.54, + "learning_rate": 1.5962566335895146e-05, + "loss": 0.2513, + "step": 6071 + }, + { + "epoch": 2.54, + "learning_rate": 1.5961205186320952e-05, + "loss": 0.2531, + "step": 6072 + }, + { + "epoch": 2.54, + "learning_rate": 1.59598438653972e-05, + "loss": 0.2473, + "step": 6073 + }, + { + "epoch": 2.54, + "learning_rate": 1.595848237316303e-05, + "loss": 0.2286, + "step": 6074 + }, + { + "epoch": 2.54, + "learning_rate": 1.595712070965757e-05, + "loss": 0.2766, + "step": 6075 + }, + { + "epoch": 2.54, + "learning_rate": 1.595575887491996e-05, + "loss": 0.2845, + "step": 6076 + }, + { + "epoch": 2.54, + "learning_rate": 1.595439686898935e-05, + "loss": 0.2496, + "step": 6077 + }, + { + "epoch": 2.54, + "learning_rate": 1.595303469190488e-05, + "loss": 0.2742, + "step": 6078 + }, + { + "epoch": 2.55, + "learning_rate": 1.5951672343705714e-05, + "loss": 0.2465, + "step": 6079 + }, + { + "epoch": 2.55, + "learning_rate": 1.5950309824431008e-05, + "loss": 0.2647, + "step": 6080 + }, + { + "epoch": 2.55, + "learning_rate": 1.5948947134119926e-05, + "loss": 0.2563, + "step": 6081 + }, + { + "epoch": 2.55, + "learning_rate": 1.594758427281164e-05, + "loss": 0.2532, + "step": 6082 + }, + { + "epoch": 2.55, + "learning_rate": 1.594622124054532e-05, + "loss": 0.2513, + "step": 6083 + }, + { + "epoch": 2.55, + "learning_rate": 1.5944858037360145e-05, + "loss": 0.2514, + "step": 6084 + }, + { + "epoch": 2.55, + "learning_rate": 1.5943494663295303e-05, + "loss": 0.2461, + "step": 6085 + }, + { + "epoch": 2.55, + "learning_rate": 1.5942131118389984e-05, + "loss": 0.2538, + "step": 6086 + }, + { + "epoch": 2.55, + "learning_rate": 1.594076740268337e-05, + "loss": 0.2439, + "step": 6087 + }, + { + "epoch": 2.55, + "learning_rate": 1.593940351621468e-05, + "loss": 0.2267, + "step": 6088 + }, + { + "epoch": 2.55, + "learning_rate": 1.5938039459023098e-05, + "loss": 0.2541, + "step": 6089 + }, + { + "epoch": 2.55, + "learning_rate": 1.5936675231147845e-05, + "loss": 0.2191, + "step": 6090 + }, + { + "epoch": 2.55, + "learning_rate": 1.593531083262813e-05, + "loss": 0.268, + "step": 6091 + }, + { + "epoch": 2.55, + "learning_rate": 1.593394626350317e-05, + "loss": 0.2505, + "step": 6092 + }, + { + "epoch": 2.55, + "learning_rate": 1.5932581523812193e-05, + "loss": 0.2636, + "step": 6093 + }, + { + "epoch": 2.55, + "learning_rate": 1.5931216613594426e-05, + "loss": 0.2326, + "step": 6094 + }, + { + "epoch": 2.55, + "learning_rate": 1.5929851532889095e-05, + "loss": 0.2709, + "step": 6095 + }, + { + "epoch": 2.55, + "learning_rate": 1.592848628173545e-05, + "loss": 0.2567, + "step": 6096 + }, + { + "epoch": 2.55, + "learning_rate": 1.5927120860172722e-05, + "loss": 0.2475, + "step": 6097 + }, + { + "epoch": 2.55, + "learning_rate": 1.592575526824017e-05, + "loss": 0.2789, + "step": 6098 + }, + { + "epoch": 2.55, + "learning_rate": 1.5924389505977038e-05, + "loss": 0.2341, + "step": 6099 + }, + { + "epoch": 2.55, + "learning_rate": 1.5923023573422585e-05, + "loss": 0.2605, + "step": 6100 + }, + { + "epoch": 2.55, + "learning_rate": 1.592165747061608e-05, + "loss": 0.2788, + "step": 6101 + }, + { + "epoch": 2.55, + "learning_rate": 1.5920291197596784e-05, + "loss": 0.2772, + "step": 6102 + }, + { + "epoch": 2.56, + "learning_rate": 1.591892475440397e-05, + "loss": 0.232, + "step": 6103 + }, + { + "epoch": 2.56, + "learning_rate": 1.5917558141076915e-05, + "loss": 0.2478, + "step": 6104 + }, + { + "epoch": 2.56, + "learning_rate": 1.5916191357654906e-05, + "loss": 0.2352, + "step": 6105 + }, + { + "epoch": 2.56, + "learning_rate": 1.5914824404177222e-05, + "loss": 0.2803, + "step": 6106 + }, + { + "epoch": 2.56, + "learning_rate": 1.5913457280683163e-05, + "loss": 0.2376, + "step": 6107 + }, + { + "epoch": 2.56, + "learning_rate": 1.591208998721202e-05, + "loss": 0.264, + "step": 6108 + }, + { + "epoch": 2.56, + "learning_rate": 1.5910722523803098e-05, + "loss": 0.2767, + "step": 6109 + }, + { + "epoch": 2.56, + "learning_rate": 1.5909354890495703e-05, + "loss": 0.2921, + "step": 6110 + }, + { + "epoch": 2.56, + "learning_rate": 1.5907987087329143e-05, + "loss": 0.2388, + "step": 6111 + }, + { + "epoch": 2.56, + "learning_rate": 1.590661911434274e-05, + "loss": 0.2408, + "step": 6112 + }, + { + "epoch": 2.56, + "learning_rate": 1.5905250971575808e-05, + "loss": 0.2688, + "step": 6113 + }, + { + "epoch": 2.56, + "learning_rate": 1.590388265906768e-05, + "loss": 0.2435, + "step": 6114 + }, + { + "epoch": 2.56, + "learning_rate": 1.590251417685768e-05, + "loss": 0.2595, + "step": 6115 + }, + { + "epoch": 2.56, + "learning_rate": 1.5901145524985155e-05, + "loss": 0.26, + "step": 6116 + }, + { + "epoch": 2.56, + "learning_rate": 1.5899776703489432e-05, + "loss": 0.304, + "step": 6117 + }, + { + "epoch": 2.56, + "learning_rate": 1.5898407712409863e-05, + "loss": 0.2597, + "step": 6118 + }, + { + "epoch": 2.56, + "learning_rate": 1.58970385517858e-05, + "loss": 0.2655, + "step": 6119 + }, + { + "epoch": 2.56, + "learning_rate": 1.58956692216566e-05, + "loss": 0.2699, + "step": 6120 + }, + { + "epoch": 2.56, + "learning_rate": 1.5894299722061617e-05, + "loss": 0.273, + "step": 6121 + }, + { + "epoch": 2.56, + "learning_rate": 1.5892930053040218e-05, + "loss": 0.2488, + "step": 6122 + }, + { + "epoch": 2.56, + "learning_rate": 1.5891560214631775e-05, + "loss": 0.2928, + "step": 6123 + }, + { + "epoch": 2.56, + "learning_rate": 1.5890190206875663e-05, + "loss": 0.2671, + "step": 6124 + }, + { + "epoch": 2.56, + "learning_rate": 1.588882002981126e-05, + "loss": 0.2488, + "step": 6125 + }, + { + "epoch": 2.56, + "learning_rate": 1.588744968347795e-05, + "loss": 0.2903, + "step": 6126 + }, + { + "epoch": 2.57, + "learning_rate": 1.5886079167915123e-05, + "loss": 0.2526, + "step": 6127 + }, + { + "epoch": 2.57, + "learning_rate": 1.5884708483162175e-05, + "loss": 0.2429, + "step": 6128 + }, + { + "epoch": 2.57, + "learning_rate": 1.5883337629258502e-05, + "loss": 0.2909, + "step": 6129 + }, + { + "epoch": 2.57, + "learning_rate": 1.5881966606243512e-05, + "loss": 0.2554, + "step": 6130 + }, + { + "epoch": 2.57, + "learning_rate": 1.588059541415661e-05, + "loss": 0.2652, + "step": 6131 + }, + { + "epoch": 2.57, + "learning_rate": 1.5879224053037213e-05, + "loss": 0.2657, + "step": 6132 + }, + { + "epoch": 2.57, + "learning_rate": 1.5877852522924733e-05, + "loss": 0.2627, + "step": 6133 + }, + { + "epoch": 2.57, + "learning_rate": 1.58764808238586e-05, + "loss": 0.2544, + "step": 6134 + }, + { + "epoch": 2.57, + "learning_rate": 1.587510895587824e-05, + "loss": 0.2426, + "step": 6135 + }, + { + "epoch": 2.57, + "learning_rate": 1.587373691902309e-05, + "loss": 0.2401, + "step": 6136 + }, + { + "epoch": 2.57, + "learning_rate": 1.5872364713332583e-05, + "loss": 0.2544, + "step": 6137 + }, + { + "epoch": 2.57, + "learning_rate": 1.587099233884616e-05, + "loss": 0.2522, + "step": 6138 + }, + { + "epoch": 2.57, + "learning_rate": 1.586961979560328e-05, + "loss": 0.2804, + "step": 6139 + }, + { + "epoch": 2.57, + "learning_rate": 1.586824708364338e-05, + "loss": 0.2559, + "step": 6140 + }, + { + "epoch": 2.57, + "learning_rate": 1.5866874203005928e-05, + "loss": 0.2641, + "step": 6141 + }, + { + "epoch": 2.57, + "learning_rate": 1.5865501153730383e-05, + "loss": 0.2238, + "step": 6142 + }, + { + "epoch": 2.57, + "learning_rate": 1.586412793585621e-05, + "loss": 0.2681, + "step": 6143 + }, + { + "epoch": 2.57, + "learning_rate": 1.5862754549422884e-05, + "loss": 0.272, + "step": 6144 + }, + { + "epoch": 2.57, + "learning_rate": 1.5861380994469883e-05, + "loss": 0.2693, + "step": 6145 + }, + { + "epoch": 2.57, + "learning_rate": 1.5860007271036682e-05, + "loss": 0.2674, + "step": 6146 + }, + { + "epoch": 2.57, + "learning_rate": 1.585863337916278e-05, + "loss": 0.2741, + "step": 6147 + }, + { + "epoch": 2.57, + "learning_rate": 1.5857259318887653e-05, + "loss": 0.2629, + "step": 6148 + }, + { + "epoch": 2.57, + "learning_rate": 1.585588509025081e-05, + "loss": 0.202, + "step": 6149 + }, + { + "epoch": 2.57, + "learning_rate": 1.5854510693291744e-05, + "loss": 0.26, + "step": 6150 + }, + { + "epoch": 2.58, + "learning_rate": 1.5853136128049963e-05, + "loss": 0.2583, + "step": 6151 + }, + { + "epoch": 2.58, + "learning_rate": 1.585176139456498e-05, + "loss": 0.243, + "step": 6152 + }, + { + "epoch": 2.58, + "learning_rate": 1.5850386492876307e-05, + "loss": 0.2652, + "step": 6153 + }, + { + "epoch": 2.58, + "learning_rate": 1.5849011423023463e-05, + "loss": 0.2457, + "step": 6154 + }, + { + "epoch": 2.58, + "learning_rate": 1.584763618504598e-05, + "loss": 0.2511, + "step": 6155 + }, + { + "epoch": 2.58, + "learning_rate": 1.5846260778983382e-05, + "loss": 0.2497, + "step": 6156 + }, + { + "epoch": 2.58, + "learning_rate": 1.5844885204875207e-05, + "loss": 0.2725, + "step": 6157 + }, + { + "epoch": 2.58, + "learning_rate": 1.584350946276099e-05, + "loss": 0.292, + "step": 6158 + }, + { + "epoch": 2.58, + "learning_rate": 1.5842133552680286e-05, + "loss": 0.2717, + "step": 6159 + }, + { + "epoch": 2.58, + "learning_rate": 1.5840757474672634e-05, + "loss": 0.2678, + "step": 6160 + }, + { + "epoch": 2.58, + "learning_rate": 1.5839381228777586e-05, + "loss": 0.2544, + "step": 6161 + }, + { + "epoch": 2.58, + "learning_rate": 1.583800481503471e-05, + "loss": 0.2599, + "step": 6162 + }, + { + "epoch": 2.58, + "learning_rate": 1.5836628233483568e-05, + "loss": 0.2436, + "step": 6163 + }, + { + "epoch": 2.58, + "learning_rate": 1.5835251484163728e-05, + "loss": 0.2356, + "step": 6164 + }, + { + "epoch": 2.58, + "learning_rate": 1.583387456711476e-05, + "loss": 0.2818, + "step": 6165 + }, + { + "epoch": 2.58, + "learning_rate": 1.5832497482376244e-05, + "loss": 0.2922, + "step": 6166 + }, + { + "epoch": 2.58, + "learning_rate": 1.583112022998776e-05, + "loss": 0.2658, + "step": 6167 + }, + { + "epoch": 2.58, + "learning_rate": 1.5829742809988907e-05, + "loss": 0.2644, + "step": 6168 + }, + { + "epoch": 2.58, + "learning_rate": 1.5828365222419268e-05, + "loss": 0.2485, + "step": 6169 + }, + { + "epoch": 2.58, + "learning_rate": 1.582698746731844e-05, + "loss": 0.2583, + "step": 6170 + }, + { + "epoch": 2.58, + "learning_rate": 1.582560954472603e-05, + "loss": 0.2407, + "step": 6171 + }, + { + "epoch": 2.58, + "learning_rate": 1.582423145468164e-05, + "loss": 0.2561, + "step": 6172 + }, + { + "epoch": 2.58, + "learning_rate": 1.5822853197224888e-05, + "loss": 0.2346, + "step": 6173 + }, + { + "epoch": 2.58, + "learning_rate": 1.5821474772395385e-05, + "loss": 0.2575, + "step": 6174 + }, + { + "epoch": 2.59, + "learning_rate": 1.5820096180232762e-05, + "loss": 0.2469, + "step": 6175 + }, + { + "epoch": 2.59, + "learning_rate": 1.5818717420776635e-05, + "loss": 0.2751, + "step": 6176 + }, + { + "epoch": 2.59, + "learning_rate": 1.5817338494066636e-05, + "loss": 0.251, + "step": 6177 + }, + { + "epoch": 2.59, + "learning_rate": 1.581595940014241e-05, + "loss": 0.2643, + "step": 6178 + }, + { + "epoch": 2.59, + "learning_rate": 1.5814580139043587e-05, + "loss": 0.2536, + "step": 6179 + }, + { + "epoch": 2.59, + "learning_rate": 1.581320071080982e-05, + "loss": 0.2741, + "step": 6180 + }, + { + "epoch": 2.59, + "learning_rate": 1.581182111548076e-05, + "loss": 0.2593, + "step": 6181 + }, + { + "epoch": 2.59, + "learning_rate": 1.5810441353096054e-05, + "loss": 0.2871, + "step": 6182 + }, + { + "epoch": 2.59, + "learning_rate": 1.5809061423695368e-05, + "loss": 0.2528, + "step": 6183 + }, + { + "epoch": 2.59, + "learning_rate": 1.5807681327318372e-05, + "loss": 0.2764, + "step": 6184 + }, + { + "epoch": 2.59, + "learning_rate": 1.5806301064004724e-05, + "loss": 0.2689, + "step": 6185 + }, + { + "epoch": 2.59, + "learning_rate": 1.5804920633794107e-05, + "loss": 0.2624, + "step": 6186 + }, + { + "epoch": 2.59, + "learning_rate": 1.5803540036726198e-05, + "loss": 0.2553, + "step": 6187 + }, + { + "epoch": 2.59, + "learning_rate": 1.5802159272840677e-05, + "loss": 0.24, + "step": 6188 + }, + { + "epoch": 2.59, + "learning_rate": 1.580077834217724e-05, + "loss": 0.2369, + "step": 6189 + }, + { + "epoch": 2.59, + "learning_rate": 1.5799397244775576e-05, + "loss": 0.2515, + "step": 6190 + }, + { + "epoch": 2.59, + "learning_rate": 1.5798015980675384e-05, + "loss": 0.2577, + "step": 6191 + }, + { + "epoch": 2.59, + "learning_rate": 1.579663454991637e-05, + "loss": 0.2721, + "step": 6192 + }, + { + "epoch": 2.59, + "learning_rate": 1.579525295253824e-05, + "loss": 0.2448, + "step": 6193 + }, + { + "epoch": 2.59, + "learning_rate": 1.5793871188580704e-05, + "loss": 0.2729, + "step": 6194 + }, + { + "epoch": 2.59, + "learning_rate": 1.579248925808348e-05, + "loss": 0.2618, + "step": 6195 + }, + { + "epoch": 2.59, + "learning_rate": 1.5791107161086294e-05, + "loss": 0.2722, + "step": 6196 + }, + { + "epoch": 2.59, + "learning_rate": 1.5789724897628873e-05, + "loss": 0.261, + "step": 6197 + }, + { + "epoch": 2.59, + "learning_rate": 1.578834246775095e-05, + "loss": 0.2592, + "step": 6198 + }, + { + "epoch": 2.6, + "learning_rate": 1.5786959871492254e-05, + "loss": 0.2786, + "step": 6199 + }, + { + "epoch": 2.6, + "learning_rate": 1.5785577108892537e-05, + "loss": 0.2406, + "step": 6200 + }, + { + "epoch": 2.6, + "learning_rate": 1.5784194179991534e-05, + "loss": 0.2347, + "step": 6201 + }, + { + "epoch": 2.6, + "learning_rate": 1.5782811084829005e-05, + "loss": 0.2531, + "step": 6202 + }, + { + "epoch": 2.6, + "learning_rate": 1.5781427823444704e-05, + "loss": 0.2519, + "step": 6203 + }, + { + "epoch": 2.6, + "learning_rate": 1.578004439587839e-05, + "loss": 0.2603, + "step": 6204 + }, + { + "epoch": 2.6, + "learning_rate": 1.5778660802169826e-05, + "loss": 0.2648, + "step": 6205 + }, + { + "epoch": 2.6, + "learning_rate": 1.577727704235879e-05, + "loss": 0.2316, + "step": 6206 + }, + { + "epoch": 2.6, + "learning_rate": 1.5775893116485052e-05, + "loss": 0.2395, + "step": 6207 + }, + { + "epoch": 2.6, + "learning_rate": 1.5774509024588388e-05, + "loss": 0.2766, + "step": 6208 + }, + { + "epoch": 2.6, + "learning_rate": 1.5773124766708588e-05, + "loss": 0.2556, + "step": 6209 + }, + { + "epoch": 2.6, + "learning_rate": 1.577174034288544e-05, + "loss": 0.2498, + "step": 6210 + }, + { + "epoch": 2.6, + "learning_rate": 1.5770355753158736e-05, + "loss": 0.2674, + "step": 6211 + }, + { + "epoch": 2.6, + "learning_rate": 1.5768970997568282e-05, + "loss": 0.2553, + "step": 6212 + }, + { + "epoch": 2.6, + "learning_rate": 1.576758607615387e-05, + "loss": 0.2363, + "step": 6213 + }, + { + "epoch": 2.6, + "learning_rate": 1.5766200988955315e-05, + "loss": 0.2762, + "step": 6214 + }, + { + "epoch": 2.6, + "learning_rate": 1.576481573601243e-05, + "loss": 0.2845, + "step": 6215 + }, + { + "epoch": 2.6, + "learning_rate": 1.5763430317365032e-05, + "loss": 0.2653, + "step": 6216 + }, + { + "epoch": 2.6, + "learning_rate": 1.5762044733052944e-05, + "loss": 0.2545, + "step": 6217 + }, + { + "epoch": 2.6, + "learning_rate": 1.576065898311599e-05, + "loss": 0.2495, + "step": 6218 + }, + { + "epoch": 2.6, + "learning_rate": 1.575927306759401e-05, + "loss": 0.2674, + "step": 6219 + }, + { + "epoch": 2.6, + "learning_rate": 1.575788698652684e-05, + "loss": 0.2691, + "step": 6220 + }, + { + "epoch": 2.6, + "learning_rate": 1.5756500739954312e-05, + "loss": 0.2548, + "step": 6221 + }, + { + "epoch": 2.6, + "learning_rate": 1.5755114327916282e-05, + "loss": 0.2469, + "step": 6222 + }, + { + "epoch": 2.61, + "learning_rate": 1.5753727750452596e-05, + "loss": 0.2675, + "step": 6223 + }, + { + "epoch": 2.61, + "learning_rate": 1.575234100760311e-05, + "loss": 0.2719, + "step": 6224 + }, + { + "epoch": 2.61, + "learning_rate": 1.5750954099407688e-05, + "loss": 0.2764, + "step": 6225 + }, + { + "epoch": 2.61, + "learning_rate": 1.5749567025906193e-05, + "loss": 0.2575, + "step": 6226 + }, + { + "epoch": 2.61, + "learning_rate": 1.5748179787138494e-05, + "loss": 0.2601, + "step": 6227 + }, + { + "epoch": 2.61, + "learning_rate": 1.5746792383144472e-05, + "loss": 0.2749, + "step": 6228 + }, + { + "epoch": 2.61, + "learning_rate": 1.5745404813964e-05, + "loss": 0.2885, + "step": 6229 + }, + { + "epoch": 2.61, + "learning_rate": 1.5744017079636967e-05, + "loss": 0.2574, + "step": 6230 + }, + { + "epoch": 2.61, + "learning_rate": 1.5742629180203258e-05, + "loss": 0.2448, + "step": 6231 + }, + { + "epoch": 2.61, + "learning_rate": 1.574124111570277e-05, + "loss": 0.2523, + "step": 6232 + }, + { + "epoch": 2.61, + "learning_rate": 1.57398528861754e-05, + "loss": 0.277, + "step": 6233 + }, + { + "epoch": 2.61, + "learning_rate": 1.5738464491661054e-05, + "loss": 0.2612, + "step": 6234 + }, + { + "epoch": 2.61, + "learning_rate": 1.5737075932199637e-05, + "loss": 0.2514, + "step": 6235 + }, + { + "epoch": 2.61, + "learning_rate": 1.5735687207831064e-05, + "loss": 0.2762, + "step": 6236 + }, + { + "epoch": 2.61, + "learning_rate": 1.5734298318595252e-05, + "loss": 0.266, + "step": 6237 + }, + { + "epoch": 2.61, + "learning_rate": 1.573290926453212e-05, + "loss": 0.2705, + "step": 6238 + }, + { + "epoch": 2.61, + "learning_rate": 1.5731520045681606e-05, + "loss": 0.2766, + "step": 6239 + }, + { + "epoch": 2.61, + "learning_rate": 1.5730130662083628e-05, + "loss": 0.2422, + "step": 6240 + }, + { + "epoch": 2.61, + "learning_rate": 1.572874111377813e-05, + "loss": 0.2674, + "step": 6241 + }, + { + "epoch": 2.61, + "learning_rate": 1.5727351400805054e-05, + "loss": 0.268, + "step": 6242 + }, + { + "epoch": 2.61, + "learning_rate": 1.5725961523204345e-05, + "loss": 0.2462, + "step": 6243 + }, + { + "epoch": 2.61, + "learning_rate": 1.5724571481015953e-05, + "loss": 0.2585, + "step": 6244 + }, + { + "epoch": 2.61, + "learning_rate": 1.572318127427983e-05, + "loss": 0.308, + "step": 6245 + }, + { + "epoch": 2.62, + "learning_rate": 1.5721790903035942e-05, + "loss": 0.2525, + "step": 6246 + }, + { + "epoch": 2.62, + "learning_rate": 1.572040036732425e-05, + "loss": 0.2731, + "step": 6247 + }, + { + "epoch": 2.62, + "learning_rate": 1.571900966718473e-05, + "loss": 0.246, + "step": 6248 + }, + { + "epoch": 2.62, + "learning_rate": 1.571761880265735e-05, + "loss": 0.2731, + "step": 6249 + }, + { + "epoch": 2.62, + "learning_rate": 1.5716227773782088e-05, + "loss": 0.2903, + "step": 6250 + }, + { + "epoch": 2.62, + "learning_rate": 1.5714836580598936e-05, + "loss": 0.2622, + "step": 6251 + }, + { + "epoch": 2.62, + "learning_rate": 1.5713445223147876e-05, + "loss": 0.2476, + "step": 6252 + }, + { + "epoch": 2.62, + "learning_rate": 1.5712053701468902e-05, + "loss": 0.259, + "step": 6253 + }, + { + "epoch": 2.62, + "learning_rate": 1.5710662015602016e-05, + "loss": 0.2912, + "step": 6254 + }, + { + "epoch": 2.62, + "learning_rate": 1.5709270165587216e-05, + "loss": 0.2395, + "step": 6255 + }, + { + "epoch": 2.62, + "learning_rate": 1.570787815146451e-05, + "loss": 0.2619, + "step": 6256 + }, + { + "epoch": 2.62, + "learning_rate": 1.5706485973273916e-05, + "loss": 0.2478, + "step": 6257 + }, + { + "epoch": 2.62, + "learning_rate": 1.570509363105544e-05, + "loss": 0.2812, + "step": 6258 + }, + { + "epoch": 2.62, + "learning_rate": 1.5703701124849115e-05, + "loss": 0.2282, + "step": 6259 + }, + { + "epoch": 2.62, + "learning_rate": 1.5702308454694963e-05, + "loss": 0.2505, + "step": 6260 + }, + { + "epoch": 2.62, + "learning_rate": 1.570091562063301e-05, + "loss": 0.2522, + "step": 6261 + }, + { + "epoch": 2.62, + "learning_rate": 1.5699522622703303e-05, + "loss": 0.2635, + "step": 6262 + }, + { + "epoch": 2.62, + "learning_rate": 1.569812946094587e-05, + "loss": 0.2555, + "step": 6263 + }, + { + "epoch": 2.62, + "learning_rate": 1.5696736135400765e-05, + "loss": 0.2671, + "step": 6264 + }, + { + "epoch": 2.62, + "learning_rate": 1.569534264610803e-05, + "loss": 0.2352, + "step": 6265 + }, + { + "epoch": 2.62, + "learning_rate": 1.5693948993107733e-05, + "loss": 0.2657, + "step": 6266 + }, + { + "epoch": 2.62, + "learning_rate": 1.569255517643992e-05, + "loss": 0.2495, + "step": 6267 + }, + { + "epoch": 2.62, + "learning_rate": 1.569116119614466e-05, + "loss": 0.2657, + "step": 6268 + }, + { + "epoch": 2.62, + "learning_rate": 1.5689767052262028e-05, + "loss": 0.2685, + "step": 6269 + }, + { + "epoch": 2.63, + "learning_rate": 1.568837274483208e-05, + "loss": 0.2572, + "step": 6270 + }, + { + "epoch": 2.63, + "learning_rate": 1.5686978273894914e-05, + "loss": 0.2648, + "step": 6271 + }, + { + "epoch": 2.63, + "learning_rate": 1.5685583639490598e-05, + "loss": 0.2516, + "step": 6272 + }, + { + "epoch": 2.63, + "learning_rate": 1.568418884165923e-05, + "loss": 0.2292, + "step": 6273 + }, + { + "epoch": 2.63, + "learning_rate": 1.5682793880440895e-05, + "loss": 0.2718, + "step": 6274 + }, + { + "epoch": 2.63, + "learning_rate": 1.5681398755875695e-05, + "loss": 0.2613, + "step": 6275 + }, + { + "epoch": 2.63, + "learning_rate": 1.5680003468003725e-05, + "loss": 0.2443, + "step": 6276 + }, + { + "epoch": 2.63, + "learning_rate": 1.56786080168651e-05, + "loss": 0.2538, + "step": 6277 + }, + { + "epoch": 2.63, + "learning_rate": 1.567721240249993e-05, + "loss": 0.2497, + "step": 6278 + }, + { + "epoch": 2.63, + "learning_rate": 1.5675816624948322e-05, + "loss": 0.2681, + "step": 6279 + }, + { + "epoch": 2.63, + "learning_rate": 1.56744206842504e-05, + "loss": 0.2525, + "step": 6280 + }, + { + "epoch": 2.63, + "learning_rate": 1.5673024580446294e-05, + "loss": 0.2496, + "step": 6281 + }, + { + "epoch": 2.63, + "learning_rate": 1.567162831357613e-05, + "loss": 0.2768, + "step": 6282 + }, + { + "epoch": 2.63, + "learning_rate": 1.5670231883680044e-05, + "loss": 0.2721, + "step": 6283 + }, + { + "epoch": 2.63, + "learning_rate": 1.566883529079817e-05, + "loss": 0.274, + "step": 6284 + }, + { + "epoch": 2.63, + "learning_rate": 1.5667438534970664e-05, + "loss": 0.283, + "step": 6285 + }, + { + "epoch": 2.63, + "learning_rate": 1.566604161623766e-05, + "loss": 0.284, + "step": 6286 + }, + { + "epoch": 2.63, + "learning_rate": 1.566464453463932e-05, + "loss": 0.2795, + "step": 6287 + }, + { + "epoch": 2.63, + "learning_rate": 1.5663247290215797e-05, + "loss": 0.2495, + "step": 6288 + }, + { + "epoch": 2.63, + "learning_rate": 1.566184988300726e-05, + "loss": 0.261, + "step": 6289 + }, + { + "epoch": 2.63, + "learning_rate": 1.5660452313053873e-05, + "loss": 0.2663, + "step": 6290 + }, + { + "epoch": 2.63, + "learning_rate": 1.56590545803958e-05, + "loss": 0.2632, + "step": 6291 + }, + { + "epoch": 2.63, + "learning_rate": 1.565765668507323e-05, + "loss": 0.2589, + "step": 6292 + }, + { + "epoch": 2.63, + "learning_rate": 1.565625862712634e-05, + "loss": 0.26, + "step": 6293 + }, + { + "epoch": 2.64, + "learning_rate": 1.5654860406595312e-05, + "loss": 0.2677, + "step": 6294 + }, + { + "epoch": 2.64, + "learning_rate": 1.5653462023520344e-05, + "loss": 0.2423, + "step": 6295 + }, + { + "epoch": 2.64, + "learning_rate": 1.5652063477941625e-05, + "loss": 0.2646, + "step": 6296 + }, + { + "epoch": 2.64, + "learning_rate": 1.5650664769899356e-05, + "loss": 0.2359, + "step": 6297 + }, + { + "epoch": 2.64, + "learning_rate": 1.564926589943374e-05, + "loss": 0.2378, + "step": 6298 + }, + { + "epoch": 2.64, + "learning_rate": 1.5647866866584992e-05, + "loss": 0.2465, + "step": 6299 + }, + { + "epoch": 2.64, + "learning_rate": 1.5646467671393323e-05, + "loss": 0.2394, + "step": 6300 + }, + { + "epoch": 2.64, + "learning_rate": 1.564506831389895e-05, + "loss": 0.2737, + "step": 6301 + }, + { + "epoch": 2.64, + "learning_rate": 1.5643668794142102e-05, + "loss": 0.2833, + "step": 6302 + }, + { + "epoch": 2.64, + "learning_rate": 1.5642269112162998e-05, + "loss": 0.2895, + "step": 6303 + }, + { + "epoch": 2.64, + "learning_rate": 1.5640869268001875e-05, + "loss": 0.2499, + "step": 6304 + }, + { + "epoch": 2.64, + "learning_rate": 1.563946926169897e-05, + "loss": 0.2881, + "step": 6305 + }, + { + "epoch": 2.64, + "learning_rate": 1.5638069093294527e-05, + "loss": 0.271, + "step": 6306 + }, + { + "epoch": 2.64, + "learning_rate": 1.5636668762828793e-05, + "loss": 0.2463, + "step": 6307 + }, + { + "epoch": 2.64, + "learning_rate": 1.5635268270342015e-05, + "loss": 0.2836, + "step": 6308 + }, + { + "epoch": 2.64, + "learning_rate": 1.5633867615874453e-05, + "loss": 0.2646, + "step": 6309 + }, + { + "epoch": 2.64, + "learning_rate": 1.5632466799466362e-05, + "loss": 0.2348, + "step": 6310 + }, + { + "epoch": 2.64, + "learning_rate": 1.5631065821158016e-05, + "loss": 0.2564, + "step": 6311 + }, + { + "epoch": 2.64, + "learning_rate": 1.5629664680989677e-05, + "loss": 0.2469, + "step": 6312 + }, + { + "epoch": 2.64, + "learning_rate": 1.562826337900162e-05, + "loss": 0.2531, + "step": 6313 + }, + { + "epoch": 2.64, + "learning_rate": 1.562686191523413e-05, + "loss": 0.2948, + "step": 6314 + }, + { + "epoch": 2.64, + "learning_rate": 1.562546028972749e-05, + "loss": 0.2857, + "step": 6315 + }, + { + "epoch": 2.64, + "learning_rate": 1.562405850252198e-05, + "loss": 0.2549, + "step": 6316 + }, + { + "epoch": 2.64, + "learning_rate": 1.5622656553657904e-05, + "loss": 0.2772, + "step": 6317 + }, + { + "epoch": 2.65, + "learning_rate": 1.5621254443175553e-05, + "loss": 0.28, + "step": 6318 + }, + { + "epoch": 2.65, + "learning_rate": 1.561985217111523e-05, + "loss": 0.2754, + "step": 6319 + }, + { + "epoch": 2.65, + "learning_rate": 1.5618449737517242e-05, + "loss": 0.2899, + "step": 6320 + }, + { + "epoch": 2.65, + "learning_rate": 1.5617047142421904e-05, + "loss": 0.2448, + "step": 6321 + }, + { + "epoch": 2.65, + "learning_rate": 1.5615644385869532e-05, + "loss": 0.2475, + "step": 6322 + }, + { + "epoch": 2.65, + "learning_rate": 1.5614241467900444e-05, + "loss": 0.276, + "step": 6323 + }, + { + "epoch": 2.65, + "learning_rate": 1.5612838388554966e-05, + "loss": 0.2666, + "step": 6324 + }, + { + "epoch": 2.65, + "learning_rate": 1.561143514787343e-05, + "loss": 0.2741, + "step": 6325 + }, + { + "epoch": 2.65, + "learning_rate": 1.561003174589617e-05, + "loss": 0.2765, + "step": 6326 + }, + { + "epoch": 2.65, + "learning_rate": 1.5608628182663527e-05, + "loss": 0.2711, + "step": 6327 + }, + { + "epoch": 2.65, + "learning_rate": 1.5607224458215843e-05, + "loss": 0.2404, + "step": 6328 + }, + { + "epoch": 2.65, + "learning_rate": 1.560582057259347e-05, + "loss": 0.2638, + "step": 6329 + }, + { + "epoch": 2.65, + "learning_rate": 1.5604416525836756e-05, + "loss": 0.2678, + "step": 6330 + }, + { + "epoch": 2.65, + "learning_rate": 1.560301231798606e-05, + "loss": 0.2781, + "step": 6331 + }, + { + "epoch": 2.65, + "learning_rate": 1.560160794908175e-05, + "loss": 0.2909, + "step": 6332 + }, + { + "epoch": 2.65, + "learning_rate": 1.560020341916419e-05, + "loss": 0.2679, + "step": 6333 + }, + { + "epoch": 2.65, + "learning_rate": 1.5598798728273753e-05, + "loss": 0.251, + "step": 6334 + }, + { + "epoch": 2.65, + "learning_rate": 1.5597393876450813e-05, + "loss": 0.2647, + "step": 6335 + }, + { + "epoch": 2.65, + "learning_rate": 1.5595988863735758e-05, + "loss": 0.2343, + "step": 6336 + }, + { + "epoch": 2.65, + "learning_rate": 1.5594583690168966e-05, + "loss": 0.2831, + "step": 6337 + }, + { + "epoch": 2.65, + "learning_rate": 1.559317835579083e-05, + "loss": 0.2257, + "step": 6338 + }, + { + "epoch": 2.65, + "learning_rate": 1.5591772860641743e-05, + "loss": 0.2514, + "step": 6339 + }, + { + "epoch": 2.65, + "learning_rate": 1.5590367204762113e-05, + "loss": 0.2614, + "step": 6340 + }, + { + "epoch": 2.65, + "learning_rate": 1.5588961388192333e-05, + "loss": 0.2447, + "step": 6341 + }, + { + "epoch": 2.66, + "learning_rate": 1.558755541097282e-05, + "loss": 0.2514, + "step": 6342 + }, + { + "epoch": 2.66, + "learning_rate": 1.5586149273143982e-05, + "loss": 0.2586, + "step": 6343 + }, + { + "epoch": 2.66, + "learning_rate": 1.5584742974746242e-05, + "loss": 0.2454, + "step": 6344 + }, + { + "epoch": 2.66, + "learning_rate": 1.5583336515820024e-05, + "loss": 0.2618, + "step": 6345 + }, + { + "epoch": 2.66, + "learning_rate": 1.558192989640575e-05, + "loss": 0.2594, + "step": 6346 + }, + { + "epoch": 2.66, + "learning_rate": 1.5580523116543853e-05, + "loss": 0.2391, + "step": 6347 + }, + { + "epoch": 2.66, + "learning_rate": 1.557911617627477e-05, + "loss": 0.2645, + "step": 6348 + }, + { + "epoch": 2.66, + "learning_rate": 1.557770907563895e-05, + "loss": 0.2572, + "step": 6349 + }, + { + "epoch": 2.66, + "learning_rate": 1.5576301814676825e-05, + "loss": 0.2587, + "step": 6350 + }, + { + "epoch": 2.66, + "learning_rate": 1.5574894393428856e-05, + "loss": 0.2847, + "step": 6351 + }, + { + "epoch": 2.66, + "learning_rate": 1.5573486811935494e-05, + "loss": 0.2696, + "step": 6352 + }, + { + "epoch": 2.66, + "learning_rate": 1.55720790702372e-05, + "loss": 0.2752, + "step": 6353 + }, + { + "epoch": 2.66, + "learning_rate": 1.557067116837444e-05, + "loss": 0.233, + "step": 6354 + }, + { + "epoch": 2.66, + "learning_rate": 1.5569263106387676e-05, + "loss": 0.2646, + "step": 6355 + }, + { + "epoch": 2.66, + "learning_rate": 1.5567854884317387e-05, + "loss": 0.273, + "step": 6356 + }, + { + "epoch": 2.66, + "learning_rate": 1.5566446502204052e-05, + "loss": 0.2578, + "step": 6357 + }, + { + "epoch": 2.66, + "learning_rate": 1.556503796008815e-05, + "loss": 0.2444, + "step": 6358 + }, + { + "epoch": 2.66, + "learning_rate": 1.5563629258010173e-05, + "loss": 0.2624, + "step": 6359 + }, + { + "epoch": 2.66, + "learning_rate": 1.556222039601061e-05, + "loss": 0.2674, + "step": 6360 + }, + { + "epoch": 2.66, + "learning_rate": 1.5560811374129955e-05, + "loss": 0.2442, + "step": 6361 + }, + { + "epoch": 2.66, + "learning_rate": 1.5559402192408712e-05, + "loss": 0.283, + "step": 6362 + }, + { + "epoch": 2.66, + "learning_rate": 1.5557992850887388e-05, + "loss": 0.2245, + "step": 6363 + }, + { + "epoch": 2.66, + "learning_rate": 1.5556583349606492e-05, + "loss": 0.2622, + "step": 6364 + }, + { + "epoch": 2.66, + "learning_rate": 1.5555173688606537e-05, + "loss": 0.2813, + "step": 6365 + }, + { + "epoch": 2.67, + "learning_rate": 1.5553763867928043e-05, + "loss": 0.2728, + "step": 6366 + }, + { + "epoch": 2.67, + "learning_rate": 1.5552353887611536e-05, + "loss": 0.2805, + "step": 6367 + }, + { + "epoch": 2.67, + "learning_rate": 1.5550943747697546e-05, + "loss": 0.2557, + "step": 6368 + }, + { + "epoch": 2.67, + "learning_rate": 1.5549533448226598e-05, + "loss": 0.2733, + "step": 6369 + }, + { + "epoch": 2.67, + "learning_rate": 1.554812298923924e-05, + "loss": 0.2685, + "step": 6370 + }, + { + "epoch": 2.67, + "learning_rate": 1.5546712370776005e-05, + "loss": 0.2678, + "step": 6371 + }, + { + "epoch": 2.67, + "learning_rate": 1.554530159287745e-05, + "loss": 0.2202, + "step": 6372 + }, + { + "epoch": 2.67, + "learning_rate": 1.5543890655584122e-05, + "loss": 0.2725, + "step": 6373 + }, + { + "epoch": 2.67, + "learning_rate": 1.5542479558936572e-05, + "loss": 0.2627, + "step": 6374 + }, + { + "epoch": 2.67, + "learning_rate": 1.554106830297537e-05, + "loss": 0.2835, + "step": 6375 + }, + { + "epoch": 2.67, + "learning_rate": 1.5539656887741076e-05, + "loss": 0.2782, + "step": 6376 + }, + { + "epoch": 2.67, + "learning_rate": 1.5538245313274258e-05, + "loss": 0.278, + "step": 6377 + }, + { + "epoch": 2.67, + "learning_rate": 1.5536833579615495e-05, + "loss": 0.2575, + "step": 6378 + }, + { + "epoch": 2.67, + "learning_rate": 1.5535421686805367e-05, + "loss": 0.242, + "step": 6379 + }, + { + "epoch": 2.67, + "learning_rate": 1.553400963488445e-05, + "loss": 0.2558, + "step": 6380 + }, + { + "epoch": 2.67, + "learning_rate": 1.553259742389334e-05, + "loss": 0.2637, + "step": 6381 + }, + { + "epoch": 2.67, + "learning_rate": 1.553118505387262e-05, + "loss": 0.2492, + "step": 6382 + }, + { + "epoch": 2.67, + "learning_rate": 1.55297725248629e-05, + "loss": 0.2669, + "step": 6383 + }, + { + "epoch": 2.67, + "learning_rate": 1.5528359836904774e-05, + "loss": 0.2425, + "step": 6384 + }, + { + "epoch": 2.67, + "learning_rate": 1.552694699003885e-05, + "loss": 0.2539, + "step": 6385 + }, + { + "epoch": 2.67, + "learning_rate": 1.5525533984305742e-05, + "loss": 0.2474, + "step": 6386 + }, + { + "epoch": 2.67, + "learning_rate": 1.552412081974606e-05, + "loss": 0.254, + "step": 6387 + }, + { + "epoch": 2.67, + "learning_rate": 1.5522707496400425e-05, + "loss": 0.2662, + "step": 6388 + }, + { + "epoch": 2.67, + "learning_rate": 1.552129401430947e-05, + "loss": 0.2675, + "step": 6389 + }, + { + "epoch": 2.68, + "learning_rate": 1.5519880373513814e-05, + "loss": 0.2582, + "step": 6390 + }, + { + "epoch": 2.68, + "learning_rate": 1.5518466574054097e-05, + "loss": 0.2468, + "step": 6391 + }, + { + "epoch": 2.68, + "learning_rate": 1.5517052615970955e-05, + "loss": 0.2305, + "step": 6392 + }, + { + "epoch": 2.68, + "learning_rate": 1.551563849930503e-05, + "loss": 0.3024, + "step": 6393 + }, + { + "epoch": 2.68, + "learning_rate": 1.551422422409697e-05, + "loss": 0.2718, + "step": 6394 + }, + { + "epoch": 2.68, + "learning_rate": 1.551280979038743e-05, + "loss": 0.2705, + "step": 6395 + }, + { + "epoch": 2.68, + "learning_rate": 1.551139519821706e-05, + "loss": 0.2716, + "step": 6396 + }, + { + "epoch": 2.68, + "learning_rate": 1.5509980447626532e-05, + "loss": 0.2879, + "step": 6397 + }, + { + "epoch": 2.68, + "learning_rate": 1.5508565538656505e-05, + "loss": 0.2648, + "step": 6398 + }, + { + "epoch": 2.68, + "learning_rate": 1.5507150471347644e-05, + "loss": 0.2476, + "step": 6399 + }, + { + "epoch": 2.68, + "learning_rate": 1.5505735245740635e-05, + "loss": 0.2499, + "step": 6400 + }, + { + "epoch": 2.68, + "learning_rate": 1.5504319861876148e-05, + "loss": 0.2537, + "step": 6401 + }, + { + "epoch": 2.68, + "learning_rate": 1.5502904319794875e-05, + "loss": 0.2677, + "step": 6402 + }, + { + "epoch": 2.68, + "learning_rate": 1.55014886195375e-05, + "loss": 0.2583, + "step": 6403 + }, + { + "epoch": 2.68, + "learning_rate": 1.5500072761144712e-05, + "loss": 0.2513, + "step": 6404 + }, + { + "epoch": 2.68, + "learning_rate": 1.5498656744657222e-05, + "loss": 0.2343, + "step": 6405 + }, + { + "epoch": 2.68, + "learning_rate": 1.5497240570115716e-05, + "loss": 0.2468, + "step": 6406 + }, + { + "epoch": 2.68, + "learning_rate": 1.5495824237560907e-05, + "loss": 0.2654, + "step": 6407 + }, + { + "epoch": 2.68, + "learning_rate": 1.5494407747033513e-05, + "loss": 0.2677, + "step": 6408 + }, + { + "epoch": 2.68, + "learning_rate": 1.5492991098574243e-05, + "loss": 0.2488, + "step": 6409 + }, + { + "epoch": 2.68, + "learning_rate": 1.5491574292223814e-05, + "loss": 0.2519, + "step": 6410 + }, + { + "epoch": 2.68, + "learning_rate": 1.5490157328022957e-05, + "loss": 0.2988, + "step": 6411 + }, + { + "epoch": 2.68, + "learning_rate": 1.54887402060124e-05, + "loss": 0.281, + "step": 6412 + }, + { + "epoch": 2.68, + "learning_rate": 1.5487322926232876e-05, + "loss": 0.262, + "step": 6413 + }, + { + "epoch": 2.69, + "learning_rate": 1.5485905488725126e-05, + "loss": 0.2738, + "step": 6414 + }, + { + "epoch": 2.69, + "learning_rate": 1.548448789352989e-05, + "loss": 0.2776, + "step": 6415 + }, + { + "epoch": 2.69, + "learning_rate": 1.5483070140687916e-05, + "loss": 0.2465, + "step": 6416 + }, + { + "epoch": 2.69, + "learning_rate": 1.5481652230239954e-05, + "loss": 0.2877, + "step": 6417 + }, + { + "epoch": 2.69, + "learning_rate": 1.5480234162226763e-05, + "loss": 0.2721, + "step": 6418 + }, + { + "epoch": 2.69, + "learning_rate": 1.5478815936689107e-05, + "loss": 0.2494, + "step": 6419 + }, + { + "epoch": 2.69, + "learning_rate": 1.5477397553667745e-05, + "loss": 0.2774, + "step": 6420 + }, + { + "epoch": 2.69, + "learning_rate": 1.5475979013203456e-05, + "loss": 0.2249, + "step": 6421 + }, + { + "epoch": 2.69, + "learning_rate": 1.5474560315337007e-05, + "loss": 0.2835, + "step": 6422 + }, + { + "epoch": 2.69, + "learning_rate": 1.547314146010918e-05, + "loss": 0.2793, + "step": 6423 + }, + { + "epoch": 2.69, + "learning_rate": 1.5471722447560757e-05, + "loss": 0.2306, + "step": 6424 + }, + { + "epoch": 2.69, + "learning_rate": 1.547030327773253e-05, + "loss": 0.2505, + "step": 6425 + }, + { + "epoch": 2.69, + "learning_rate": 1.546888395066529e-05, + "loss": 0.2627, + "step": 6426 + }, + { + "epoch": 2.69, + "learning_rate": 1.546746446639983e-05, + "loss": 0.2706, + "step": 6427 + }, + { + "epoch": 2.69, + "learning_rate": 1.546604482497696e-05, + "loss": 0.2982, + "step": 6428 + }, + { + "epoch": 2.69, + "learning_rate": 1.546462502643748e-05, + "loss": 0.2484, + "step": 6429 + }, + { + "epoch": 2.69, + "learning_rate": 1.5463205070822202e-05, + "loss": 0.2251, + "step": 6430 + }, + { + "epoch": 2.69, + "learning_rate": 1.5461784958171945e-05, + "loss": 0.2504, + "step": 6431 + }, + { + "epoch": 2.69, + "learning_rate": 1.5460364688527523e-05, + "loss": 0.2677, + "step": 6432 + }, + { + "epoch": 2.69, + "learning_rate": 1.5458944261929763e-05, + "loss": 0.243, + "step": 6433 + }, + { + "epoch": 2.69, + "learning_rate": 1.5457523678419498e-05, + "loss": 0.2672, + "step": 6434 + }, + { + "epoch": 2.69, + "learning_rate": 1.5456102938037555e-05, + "loss": 0.2635, + "step": 6435 + }, + { + "epoch": 2.69, + "learning_rate": 1.5454682040824776e-05, + "loss": 0.2272, + "step": 6436 + }, + { + "epoch": 2.69, + "learning_rate": 1.5453260986822e-05, + "loss": 0.2772, + "step": 6437 + }, + { + "epoch": 2.7, + "learning_rate": 1.545183977607008e-05, + "loss": 0.2579, + "step": 6438 + }, + { + "epoch": 2.7, + "learning_rate": 1.545041840860986e-05, + "loss": 0.2872, + "step": 6439 + }, + { + "epoch": 2.7, + "learning_rate": 1.5448996884482198e-05, + "loss": 0.2657, + "step": 6440 + }, + { + "epoch": 2.7, + "learning_rate": 1.5447575203727964e-05, + "loss": 0.2656, + "step": 6441 + }, + { + "epoch": 2.7, + "learning_rate": 1.5446153366388007e-05, + "loss": 0.2635, + "step": 6442 + }, + { + "epoch": 2.7, + "learning_rate": 1.5444731372503206e-05, + "loss": 0.235, + "step": 6443 + }, + { + "epoch": 2.7, + "learning_rate": 1.5443309222114434e-05, + "loss": 0.2488, + "step": 6444 + }, + { + "epoch": 2.7, + "learning_rate": 1.544188691526257e-05, + "loss": 0.2692, + "step": 6445 + }, + { + "epoch": 2.7, + "learning_rate": 1.5440464451988494e-05, + "loss": 0.2598, + "step": 6446 + }, + { + "epoch": 2.7, + "learning_rate": 1.5439041832333095e-05, + "loss": 0.2663, + "step": 6447 + }, + { + "epoch": 2.7, + "learning_rate": 1.5437619056337267e-05, + "loss": 0.2528, + "step": 6448 + }, + { + "epoch": 2.7, + "learning_rate": 1.5436196124041902e-05, + "loss": 0.2391, + "step": 6449 + }, + { + "epoch": 2.7, + "learning_rate": 1.5434773035487905e-05, + "loss": 0.2803, + "step": 6450 + }, + { + "epoch": 2.7, + "learning_rate": 1.543334979071618e-05, + "loss": 0.2625, + "step": 6451 + }, + { + "epoch": 2.7, + "learning_rate": 1.5431926389767637e-05, + "loss": 0.2697, + "step": 6452 + }, + { + "epoch": 2.7, + "learning_rate": 1.543050283268319e-05, + "loss": 0.2431, + "step": 6453 + }, + { + "epoch": 2.7, + "learning_rate": 1.5429079119503752e-05, + "loss": 0.2628, + "step": 6454 + }, + { + "epoch": 2.7, + "learning_rate": 1.542765525027026e-05, + "loss": 0.266, + "step": 6455 + }, + { + "epoch": 2.7, + "learning_rate": 1.542623122502363e-05, + "loss": 0.2715, + "step": 6456 + }, + { + "epoch": 2.7, + "learning_rate": 1.54248070438048e-05, + "loss": 0.2704, + "step": 6457 + }, + { + "epoch": 2.7, + "learning_rate": 1.5423382706654705e-05, + "loss": 0.2164, + "step": 6458 + }, + { + "epoch": 2.7, + "learning_rate": 1.542195821361429e-05, + "loss": 0.2669, + "step": 6459 + }, + { + "epoch": 2.7, + "learning_rate": 1.5420533564724495e-05, + "loss": 0.2869, + "step": 6460 + }, + { + "epoch": 2.71, + "learning_rate": 1.5419108760026273e-05, + "loss": 0.2427, + "step": 6461 + }, + { + "epoch": 2.71, + "learning_rate": 1.5417683799560576e-05, + "loss": 0.248, + "step": 6462 + }, + { + "epoch": 2.71, + "learning_rate": 1.541625868336837e-05, + "loss": 0.2846, + "step": 6463 + }, + { + "epoch": 2.71, + "learning_rate": 1.5414833411490612e-05, + "loss": 0.2726, + "step": 6464 + }, + { + "epoch": 2.71, + "learning_rate": 1.5413407983968274e-05, + "loss": 0.2504, + "step": 6465 + }, + { + "epoch": 2.71, + "learning_rate": 1.5411982400842328e-05, + "loss": 0.239, + "step": 6466 + }, + { + "epoch": 2.71, + "learning_rate": 1.5410556662153747e-05, + "loss": 0.2547, + "step": 6467 + }, + { + "epoch": 2.71, + "learning_rate": 1.5409130767943516e-05, + "loss": 0.284, + "step": 6468 + }, + { + "epoch": 2.71, + "learning_rate": 1.5407704718252625e-05, + "loss": 0.2341, + "step": 6469 + }, + { + "epoch": 2.71, + "learning_rate": 1.5406278513122058e-05, + "loss": 0.2427, + "step": 6470 + }, + { + "epoch": 2.71, + "learning_rate": 1.5404852152592814e-05, + "loss": 0.2431, + "step": 6471 + }, + { + "epoch": 2.71, + "learning_rate": 1.540342563670589e-05, + "loss": 0.2642, + "step": 6472 + }, + { + "epoch": 2.71, + "learning_rate": 1.5401998965502293e-05, + "loss": 0.2666, + "step": 6473 + }, + { + "epoch": 2.71, + "learning_rate": 1.5400572139023026e-05, + "loss": 0.2834, + "step": 6474 + }, + { + "epoch": 2.71, + "learning_rate": 1.5399145157309105e-05, + "loss": 0.281, + "step": 6475 + }, + { + "epoch": 2.71, + "learning_rate": 1.539771802040155e-05, + "loss": 0.2498, + "step": 6476 + }, + { + "epoch": 2.71, + "learning_rate": 1.539629072834138e-05, + "loss": 0.2607, + "step": 6477 + }, + { + "epoch": 2.71, + "learning_rate": 1.539486328116962e-05, + "loss": 0.2705, + "step": 6478 + }, + { + "epoch": 2.71, + "learning_rate": 1.5393435678927305e-05, + "loss": 0.2686, + "step": 6479 + }, + { + "epoch": 2.71, + "learning_rate": 1.5392007921655466e-05, + "loss": 0.2472, + "step": 6480 + }, + { + "epoch": 2.71, + "learning_rate": 1.5390580009395147e-05, + "loss": 0.2202, + "step": 6481 + }, + { + "epoch": 2.71, + "learning_rate": 1.5389151942187383e-05, + "loss": 0.2606, + "step": 6482 + }, + { + "epoch": 2.71, + "learning_rate": 1.538772372007323e-05, + "loss": 0.2618, + "step": 6483 + }, + { + "epoch": 2.71, + "learning_rate": 1.5386295343093743e-05, + "loss": 0.2498, + "step": 6484 + }, + { + "epoch": 2.72, + "learning_rate": 1.5384866811289976e-05, + "loss": 0.263, + "step": 6485 + }, + { + "epoch": 2.72, + "learning_rate": 1.538343812470299e-05, + "loss": 0.2487, + "step": 6486 + }, + { + "epoch": 2.72, + "learning_rate": 1.538200928337385e-05, + "loss": 0.273, + "step": 6487 + }, + { + "epoch": 2.72, + "learning_rate": 1.538058028734363e-05, + "loss": 0.2388, + "step": 6488 + }, + { + "epoch": 2.72, + "learning_rate": 1.5379151136653407e-05, + "loss": 0.2524, + "step": 6489 + }, + { + "epoch": 2.72, + "learning_rate": 1.5377721831344258e-05, + "loss": 0.244, + "step": 6490 + }, + { + "epoch": 2.72, + "learning_rate": 1.537629237145726e-05, + "loss": 0.2628, + "step": 6491 + }, + { + "epoch": 2.72, + "learning_rate": 1.5374862757033518e-05, + "loss": 0.2961, + "step": 6492 + }, + { + "epoch": 2.72, + "learning_rate": 1.5373432988114113e-05, + "loss": 0.2753, + "step": 6493 + }, + { + "epoch": 2.72, + "learning_rate": 1.537200306474014e-05, + "loss": 0.2629, + "step": 6494 + }, + { + "epoch": 2.72, + "learning_rate": 1.5370572986952712e-05, + "loss": 0.2631, + "step": 6495 + }, + { + "epoch": 2.72, + "learning_rate": 1.5369142754792926e-05, + "loss": 0.2519, + "step": 6496 + }, + { + "epoch": 2.72, + "learning_rate": 1.5367712368301897e-05, + "loss": 0.2696, + "step": 6497 + }, + { + "epoch": 2.72, + "learning_rate": 1.536628182752074e-05, + "loss": 0.2714, + "step": 6498 + }, + { + "epoch": 2.72, + "learning_rate": 1.536485113249057e-05, + "loss": 0.2673, + "step": 6499 + }, + { + "epoch": 2.72, + "learning_rate": 1.5363420283252518e-05, + "loss": 0.2689, + "step": 6500 + }, + { + "epoch": 2.72, + "learning_rate": 1.536198927984771e-05, + "loss": 0.2581, + "step": 6501 + }, + { + "epoch": 2.72, + "learning_rate": 1.5360558122317275e-05, + "loss": 0.2747, + "step": 6502 + }, + { + "epoch": 2.72, + "learning_rate": 1.5359126810702357e-05, + "loss": 0.2599, + "step": 6503 + }, + { + "epoch": 2.72, + "learning_rate": 1.5357695345044096e-05, + "loss": 0.2317, + "step": 6504 + }, + { + "epoch": 2.72, + "learning_rate": 1.5356263725383633e-05, + "loss": 0.264, + "step": 6505 + }, + { + "epoch": 2.72, + "learning_rate": 1.5354831951762122e-05, + "loss": 0.2814, + "step": 6506 + }, + { + "epoch": 2.72, + "learning_rate": 1.535340002422072e-05, + "loss": 0.2611, + "step": 6507 + }, + { + "epoch": 2.72, + "learning_rate": 1.5351967942800584e-05, + "loss": 0.2676, + "step": 6508 + }, + { + "epoch": 2.73, + "learning_rate": 1.535053570754288e-05, + "loss": 0.2563, + "step": 6509 + }, + { + "epoch": 2.73, + "learning_rate": 1.5349103318488772e-05, + "loss": 0.2836, + "step": 6510 + }, + { + "epoch": 2.73, + "learning_rate": 1.534767077567944e-05, + "loss": 0.257, + "step": 6511 + }, + { + "epoch": 2.73, + "learning_rate": 1.5346238079156055e-05, + "loss": 0.2543, + "step": 6512 + }, + { + "epoch": 2.73, + "learning_rate": 1.53448052289598e-05, + "loss": 0.2448, + "step": 6513 + }, + { + "epoch": 2.73, + "learning_rate": 1.5343372225131863e-05, + "loss": 0.2426, + "step": 6514 + }, + { + "epoch": 2.73, + "learning_rate": 1.5341939067713433e-05, + "loss": 0.2579, + "step": 6515 + }, + { + "epoch": 2.73, + "learning_rate": 1.5340505756745703e-05, + "loss": 0.2629, + "step": 6516 + }, + { + "epoch": 2.73, + "learning_rate": 1.5339072292269874e-05, + "loss": 0.2614, + "step": 6517 + }, + { + "epoch": 2.73, + "learning_rate": 1.5337638674327152e-05, + "loss": 0.2702, + "step": 6518 + }, + { + "epoch": 2.73, + "learning_rate": 1.5336204902958743e-05, + "loss": 0.249, + "step": 6519 + }, + { + "epoch": 2.73, + "learning_rate": 1.5334770978205857e-05, + "loss": 0.2522, + "step": 6520 + }, + { + "epoch": 2.73, + "learning_rate": 1.5333336900109715e-05, + "loss": 0.2293, + "step": 6521 + }, + { + "epoch": 2.73, + "learning_rate": 1.5331902668711536e-05, + "loss": 0.2922, + "step": 6522 + }, + { + "epoch": 2.73, + "learning_rate": 1.5330468284052544e-05, + "loss": 0.2786, + "step": 6523 + }, + { + "epoch": 2.73, + "learning_rate": 1.5329033746173975e-05, + "loss": 0.2407, + "step": 6524 + }, + { + "epoch": 2.73, + "learning_rate": 1.5327599055117057e-05, + "loss": 0.2681, + "step": 6525 + }, + { + "epoch": 2.73, + "learning_rate": 1.5326164210923032e-05, + "loss": 0.2567, + "step": 6526 + }, + { + "epoch": 2.73, + "learning_rate": 1.5324729213633146e-05, + "loss": 0.268, + "step": 6527 + }, + { + "epoch": 2.73, + "learning_rate": 1.532329406328864e-05, + "loss": 0.2753, + "step": 6528 + }, + { + "epoch": 2.73, + "learning_rate": 1.5321858759930776e-05, + "loss": 0.256, + "step": 6529 + }, + { + "epoch": 2.73, + "learning_rate": 1.53204233036008e-05, + "loss": 0.2613, + "step": 6530 + }, + { + "epoch": 2.73, + "learning_rate": 1.531898769433998e-05, + "loss": 0.2602, + "step": 6531 + }, + { + "epoch": 2.73, + "learning_rate": 1.5317551932189574e-05, + "loss": 0.2707, + "step": 6532 + }, + { + "epoch": 2.74, + "learning_rate": 1.531611601719086e-05, + "loss": 0.2616, + "step": 6533 + }, + { + "epoch": 2.74, + "learning_rate": 1.5314679949385105e-05, + "loss": 0.2707, + "step": 6534 + }, + { + "epoch": 2.74, + "learning_rate": 1.5313243728813598e-05, + "loss": 0.2467, + "step": 6535 + }, + { + "epoch": 2.74, + "learning_rate": 1.5311807355517608e-05, + "loss": 0.257, + "step": 6536 + }, + { + "epoch": 2.74, + "learning_rate": 1.531037082953843e-05, + "loss": 0.2538, + "step": 6537 + }, + { + "epoch": 2.74, + "learning_rate": 1.530893415091736e-05, + "loss": 0.2585, + "step": 6538 + }, + { + "epoch": 2.74, + "learning_rate": 1.5307497319695683e-05, + "loss": 0.2764, + "step": 6539 + }, + { + "epoch": 2.74, + "learning_rate": 1.5306060335914706e-05, + "loss": 0.2576, + "step": 6540 + }, + { + "epoch": 2.74, + "learning_rate": 1.5304623199615737e-05, + "loss": 0.2602, + "step": 6541 + }, + { + "epoch": 2.74, + "learning_rate": 1.530318591084008e-05, + "loss": 0.2584, + "step": 6542 + }, + { + "epoch": 2.74, + "learning_rate": 1.530174846962905e-05, + "loss": 0.2904, + "step": 6543 + }, + { + "epoch": 2.74, + "learning_rate": 1.530031087602396e-05, + "loss": 0.2618, + "step": 6544 + }, + { + "epoch": 2.74, + "learning_rate": 1.5298873130066144e-05, + "loss": 0.272, + "step": 6545 + }, + { + "epoch": 2.74, + "learning_rate": 1.529743523179692e-05, + "loss": 0.2769, + "step": 6546 + }, + { + "epoch": 2.74, + "learning_rate": 1.529599718125762e-05, + "loss": 0.228, + "step": 6547 + }, + { + "epoch": 2.74, + "learning_rate": 1.5294558978489583e-05, + "loss": 0.2458, + "step": 6548 + }, + { + "epoch": 2.74, + "learning_rate": 1.5293120623534145e-05, + "loss": 0.2771, + "step": 6549 + }, + { + "epoch": 2.74, + "learning_rate": 1.529168211643265e-05, + "loss": 0.244, + "step": 6550 + }, + { + "epoch": 2.74, + "learning_rate": 1.529024345722645e-05, + "loss": 0.2994, + "step": 6551 + }, + { + "epoch": 2.74, + "learning_rate": 1.5288804645956896e-05, + "loss": 0.2486, + "step": 6552 + }, + { + "epoch": 2.74, + "learning_rate": 1.5287365682665346e-05, + "loss": 0.2665, + "step": 6553 + }, + { + "epoch": 2.74, + "learning_rate": 1.5285926567393158e-05, + "loss": 0.2559, + "step": 6554 + }, + { + "epoch": 2.74, + "learning_rate": 1.5284487300181707e-05, + "loss": 0.2747, + "step": 6555 + }, + { + "epoch": 2.74, + "learning_rate": 1.5283047881072357e-05, + "loss": 0.2544, + "step": 6556 + }, + { + "epoch": 2.75, + "learning_rate": 1.5281608310106478e-05, + "loss": 0.2588, + "step": 6557 + }, + { + "epoch": 2.75, + "learning_rate": 1.5280168587325462e-05, + "loss": 0.2612, + "step": 6558 + }, + { + "epoch": 2.75, + "learning_rate": 1.5278728712770683e-05, + "loss": 0.2553, + "step": 6559 + }, + { + "epoch": 2.75, + "learning_rate": 1.5277288686483528e-05, + "loss": 0.2601, + "step": 6560 + }, + { + "epoch": 2.75, + "learning_rate": 1.5275848508505396e-05, + "loss": 0.2463, + "step": 6561 + }, + { + "epoch": 2.75, + "learning_rate": 1.527440817887768e-05, + "loss": 0.2555, + "step": 6562 + }, + { + "epoch": 2.75, + "learning_rate": 1.5272967697641784e-05, + "loss": 0.2678, + "step": 6563 + }, + { + "epoch": 2.75, + "learning_rate": 1.5271527064839107e-05, + "loss": 0.265, + "step": 6564 + }, + { + "epoch": 2.75, + "learning_rate": 1.5270086280511064e-05, + "loss": 0.2693, + "step": 6565 + }, + { + "epoch": 2.75, + "learning_rate": 1.5268645344699066e-05, + "loss": 0.2654, + "step": 6566 + }, + { + "epoch": 2.75, + "learning_rate": 1.5267204257444534e-05, + "loss": 0.2538, + "step": 6567 + }, + { + "epoch": 2.75, + "learning_rate": 1.526576301878889e-05, + "loss": 0.2554, + "step": 6568 + }, + { + "epoch": 2.75, + "learning_rate": 1.526432162877356e-05, + "loss": 0.2549, + "step": 6569 + }, + { + "epoch": 2.75, + "learning_rate": 1.5262880087439975e-05, + "loss": 0.2483, + "step": 6570 + }, + { + "epoch": 2.75, + "learning_rate": 1.526143839482957e-05, + "loss": 0.2568, + "step": 6571 + }, + { + "epoch": 2.75, + "learning_rate": 1.5259996550983797e-05, + "loss": 0.259, + "step": 6572 + }, + { + "epoch": 2.75, + "learning_rate": 1.5258554555944085e-05, + "loss": 0.2885, + "step": 6573 + }, + { + "epoch": 2.75, + "learning_rate": 1.5257112409751886e-05, + "loss": 0.2732, + "step": 6574 + }, + { + "epoch": 2.75, + "learning_rate": 1.5255670112448657e-05, + "loss": 0.2344, + "step": 6575 + }, + { + "epoch": 2.75, + "learning_rate": 1.5254227664075858e-05, + "loss": 0.2616, + "step": 6576 + }, + { + "epoch": 2.75, + "learning_rate": 1.5252785064674945e-05, + "loss": 0.27, + "step": 6577 + }, + { + "epoch": 2.75, + "learning_rate": 1.5251342314287385e-05, + "loss": 0.2484, + "step": 6578 + }, + { + "epoch": 2.75, + "learning_rate": 1.5249899412954652e-05, + "loss": 0.2413, + "step": 6579 + }, + { + "epoch": 2.75, + "learning_rate": 1.5248456360718218e-05, + "loss": 0.2837, + "step": 6580 + }, + { + "epoch": 2.76, + "learning_rate": 1.524701315761956e-05, + "loss": 0.2716, + "step": 6581 + }, + { + "epoch": 2.76, + "learning_rate": 1.5245569803700174e-05, + "loss": 0.2518, + "step": 6582 + }, + { + "epoch": 2.76, + "learning_rate": 1.5244126299001534e-05, + "loss": 0.249, + "step": 6583 + }, + { + "epoch": 2.76, + "learning_rate": 1.5242682643565135e-05, + "loss": 0.2523, + "step": 6584 + }, + { + "epoch": 2.76, + "learning_rate": 1.5241238837432477e-05, + "loss": 0.2322, + "step": 6585 + }, + { + "epoch": 2.76, + "learning_rate": 1.523979488064506e-05, + "loss": 0.2686, + "step": 6586 + }, + { + "epoch": 2.76, + "learning_rate": 1.523835077324439e-05, + "loss": 0.2603, + "step": 6587 + }, + { + "epoch": 2.76, + "learning_rate": 1.5236906515271974e-05, + "loss": 0.2674, + "step": 6588 + }, + { + "epoch": 2.76, + "learning_rate": 1.5235462106769327e-05, + "loss": 0.2468, + "step": 6589 + }, + { + "epoch": 2.76, + "learning_rate": 1.5234017547777969e-05, + "loss": 0.2629, + "step": 6590 + }, + { + "epoch": 2.76, + "learning_rate": 1.5232572838339418e-05, + "loss": 0.2581, + "step": 6591 + }, + { + "epoch": 2.76, + "learning_rate": 1.5231127978495208e-05, + "loss": 0.2561, + "step": 6592 + }, + { + "epoch": 2.76, + "learning_rate": 1.5229682968286862e-05, + "loss": 0.2629, + "step": 6593 + }, + { + "epoch": 2.76, + "learning_rate": 1.5228237807755925e-05, + "loss": 0.254, + "step": 6594 + }, + { + "epoch": 2.76, + "learning_rate": 1.522679249694393e-05, + "loss": 0.2554, + "step": 6595 + }, + { + "epoch": 2.76, + "learning_rate": 1.5225347035892423e-05, + "loss": 0.2422, + "step": 6596 + }, + { + "epoch": 2.76, + "learning_rate": 1.522390142464295e-05, + "loss": 0.2471, + "step": 6597 + }, + { + "epoch": 2.76, + "learning_rate": 1.5222455663237068e-05, + "loss": 0.2608, + "step": 6598 + }, + { + "epoch": 2.76, + "learning_rate": 1.5221009751716333e-05, + "loss": 0.2835, + "step": 6599 + }, + { + "epoch": 2.76, + "learning_rate": 1.5219563690122306e-05, + "loss": 0.2757, + "step": 6600 + }, + { + "epoch": 2.76, + "learning_rate": 1.5218117478496548e-05, + "loss": 0.2538, + "step": 6601 + }, + { + "epoch": 2.76, + "learning_rate": 1.5216671116880637e-05, + "loss": 0.274, + "step": 6602 + }, + { + "epoch": 2.76, + "learning_rate": 1.5215224605316143e-05, + "loss": 0.2488, + "step": 6603 + }, + { + "epoch": 2.76, + "learning_rate": 1.5213777943844648e-05, + "loss": 0.2378, + "step": 6604 + }, + { + "epoch": 2.77, + "learning_rate": 1.521233113250773e-05, + "loss": 0.2754, + "step": 6605 + }, + { + "epoch": 2.77, + "learning_rate": 1.5210884171346983e-05, + "loss": 0.263, + "step": 6606 + }, + { + "epoch": 2.77, + "learning_rate": 1.5209437060403988e-05, + "loss": 0.2472, + "step": 6607 + }, + { + "epoch": 2.77, + "learning_rate": 1.5207989799720352e-05, + "loss": 0.2798, + "step": 6608 + }, + { + "epoch": 2.77, + "learning_rate": 1.520654238933767e-05, + "loss": 0.2764, + "step": 6609 + }, + { + "epoch": 2.77, + "learning_rate": 1.5205094829297546e-05, + "loss": 0.2551, + "step": 6610 + }, + { + "epoch": 2.77, + "learning_rate": 1.5203647119641591e-05, + "loss": 0.2434, + "step": 6611 + }, + { + "epoch": 2.77, + "learning_rate": 1.5202199260411417e-05, + "loss": 0.2527, + "step": 6612 + }, + { + "epoch": 2.77, + "learning_rate": 1.5200751251648641e-05, + "loss": 0.2886, + "step": 6613 + }, + { + "epoch": 2.77, + "learning_rate": 1.5199303093394883e-05, + "loss": 0.2617, + "step": 6614 + }, + { + "epoch": 2.77, + "learning_rate": 1.5197854785691775e-05, + "loss": 0.2527, + "step": 6615 + }, + { + "epoch": 2.77, + "learning_rate": 1.5196406328580943e-05, + "loss": 0.273, + "step": 6616 + }, + { + "epoch": 2.77, + "learning_rate": 1.5194957722104024e-05, + "loss": 0.2436, + "step": 6617 + }, + { + "epoch": 2.77, + "learning_rate": 1.5193508966302655e-05, + "loss": 0.2875, + "step": 6618 + }, + { + "epoch": 2.77, + "learning_rate": 1.5192060061218477e-05, + "loss": 0.2511, + "step": 6619 + }, + { + "epoch": 2.77, + "learning_rate": 1.519061100689314e-05, + "loss": 0.2614, + "step": 6620 + }, + { + "epoch": 2.77, + "learning_rate": 1.5189161803368297e-05, + "loss": 0.2674, + "step": 6621 + }, + { + "epoch": 2.77, + "learning_rate": 1.51877124506856e-05, + "loss": 0.2546, + "step": 6622 + }, + { + "epoch": 2.77, + "learning_rate": 1.5186262948886716e-05, + "loss": 0.2376, + "step": 6623 + }, + { + "epoch": 2.77, + "learning_rate": 1.5184813298013304e-05, + "loss": 0.243, + "step": 6624 + }, + { + "epoch": 2.77, + "learning_rate": 1.5183363498107033e-05, + "loss": 0.2854, + "step": 6625 + }, + { + "epoch": 2.77, + "learning_rate": 1.5181913549209582e-05, + "loss": 0.2663, + "step": 6626 + }, + { + "epoch": 2.77, + "learning_rate": 1.5180463451362621e-05, + "loss": 0.2811, + "step": 6627 + }, + { + "epoch": 2.77, + "learning_rate": 1.5179013204607838e-05, + "loss": 0.2553, + "step": 6628 + }, + { + "epoch": 2.78, + "learning_rate": 1.5177562808986914e-05, + "loss": 0.2549, + "step": 6629 + }, + { + "epoch": 2.78, + "learning_rate": 1.5176112264541542e-05, + "loss": 0.249, + "step": 6630 + }, + { + "epoch": 2.78, + "learning_rate": 1.5174661571313417e-05, + "loss": 0.2608, + "step": 6631 + }, + { + "epoch": 2.78, + "learning_rate": 1.5173210729344234e-05, + "loss": 0.2462, + "step": 6632 + }, + { + "epoch": 2.78, + "learning_rate": 1.5171759738675702e-05, + "loss": 0.2513, + "step": 6633 + }, + { + "epoch": 2.78, + "learning_rate": 1.5170308599349524e-05, + "loss": 0.2768, + "step": 6634 + }, + { + "epoch": 2.78, + "learning_rate": 1.5168857311407411e-05, + "loss": 0.2487, + "step": 6635 + }, + { + "epoch": 2.78, + "learning_rate": 1.516740587489109e-05, + "loss": 0.2677, + "step": 6636 + }, + { + "epoch": 2.78, + "learning_rate": 1.5165954289842267e-05, + "loss": 0.2875, + "step": 6637 + }, + { + "epoch": 2.78, + "learning_rate": 1.516450255630267e-05, + "loss": 0.2793, + "step": 6638 + }, + { + "epoch": 2.78, + "learning_rate": 1.5163050674314033e-05, + "loss": 0.2797, + "step": 6639 + }, + { + "epoch": 2.78, + "learning_rate": 1.5161598643918086e-05, + "loss": 0.2556, + "step": 6640 + }, + { + "epoch": 2.78, + "learning_rate": 1.5160146465156564e-05, + "loss": 0.2648, + "step": 6641 + }, + { + "epoch": 2.78, + "learning_rate": 1.5158694138071213e-05, + "loss": 0.2395, + "step": 6642 + }, + { + "epoch": 2.78, + "learning_rate": 1.5157241662703777e-05, + "loss": 0.2652, + "step": 6643 + }, + { + "epoch": 2.78, + "learning_rate": 1.5155789039096005e-05, + "loss": 0.2498, + "step": 6644 + }, + { + "epoch": 2.78, + "learning_rate": 1.515433626728965e-05, + "loss": 0.2564, + "step": 6645 + }, + { + "epoch": 2.78, + "learning_rate": 1.5152883347326475e-05, + "loss": 0.2747, + "step": 6646 + }, + { + "epoch": 2.78, + "learning_rate": 1.5151430279248242e-05, + "loss": 0.2571, + "step": 6647 + }, + { + "epoch": 2.78, + "learning_rate": 1.5149977063096715e-05, + "loss": 0.2651, + "step": 6648 + }, + { + "epoch": 2.78, + "learning_rate": 1.5148523698913668e-05, + "loss": 0.2875, + "step": 6649 + }, + { + "epoch": 2.78, + "learning_rate": 1.5147070186740877e-05, + "loss": 0.2477, + "step": 6650 + }, + { + "epoch": 2.78, + "learning_rate": 1.514561652662012e-05, + "loss": 0.2882, + "step": 6651 + }, + { + "epoch": 2.79, + "learning_rate": 1.5144162718593183e-05, + "loss": 0.2645, + "step": 6652 + }, + { + "epoch": 2.79, + "learning_rate": 1.5142708762701852e-05, + "loss": 0.279, + "step": 6653 + }, + { + "epoch": 2.79, + "learning_rate": 1.5141254658987922e-05, + "loss": 0.2428, + "step": 6654 + }, + { + "epoch": 2.79, + "learning_rate": 1.5139800407493189e-05, + "loss": 0.2613, + "step": 6655 + }, + { + "epoch": 2.79, + "learning_rate": 1.5138346008259452e-05, + "loss": 0.2361, + "step": 6656 + }, + { + "epoch": 2.79, + "learning_rate": 1.5136891461328525e-05, + "loss": 0.2499, + "step": 6657 + }, + { + "epoch": 2.79, + "learning_rate": 1.5135436766742207e-05, + "loss": 0.2513, + "step": 6658 + }, + { + "epoch": 2.79, + "learning_rate": 1.5133981924542314e-05, + "loss": 0.2727, + "step": 6659 + }, + { + "epoch": 2.79, + "learning_rate": 1.513252693477067e-05, + "loss": 0.2188, + "step": 6660 + }, + { + "epoch": 2.79, + "learning_rate": 1.5131071797469093e-05, + "loss": 0.2486, + "step": 6661 + }, + { + "epoch": 2.79, + "learning_rate": 1.512961651267941e-05, + "loss": 0.2672, + "step": 6662 + }, + { + "epoch": 2.79, + "learning_rate": 1.5128161080443451e-05, + "loss": 0.2455, + "step": 6663 + }, + { + "epoch": 2.79, + "learning_rate": 1.5126705500803055e-05, + "loss": 0.2667, + "step": 6664 + }, + { + "epoch": 2.79, + "learning_rate": 1.5125249773800057e-05, + "loss": 0.2744, + "step": 6665 + }, + { + "epoch": 2.79, + "learning_rate": 1.5123793899476303e-05, + "loss": 0.2415, + "step": 6666 + }, + { + "epoch": 2.79, + "learning_rate": 1.5122337877873641e-05, + "loss": 0.2695, + "step": 6667 + }, + { + "epoch": 2.79, + "learning_rate": 1.5120881709033919e-05, + "loss": 0.2449, + "step": 6668 + }, + { + "epoch": 2.79, + "learning_rate": 1.5119425392999e-05, + "loss": 0.2488, + "step": 6669 + }, + { + "epoch": 2.79, + "learning_rate": 1.511796892981074e-05, + "loss": 0.2597, + "step": 6670 + }, + { + "epoch": 2.79, + "learning_rate": 1.5116512319511003e-05, + "loss": 0.253, + "step": 6671 + }, + { + "epoch": 2.79, + "learning_rate": 1.5115055562141661e-05, + "loss": 0.2759, + "step": 6672 + }, + { + "epoch": 2.79, + "learning_rate": 1.5113598657744585e-05, + "loss": 0.3121, + "step": 6673 + }, + { + "epoch": 2.79, + "learning_rate": 1.5112141606361657e-05, + "loss": 0.2661, + "step": 6674 + }, + { + "epoch": 2.79, + "learning_rate": 1.511068440803475e-05, + "loss": 0.2716, + "step": 6675 + }, + { + "epoch": 2.8, + "learning_rate": 1.5109227062805758e-05, + "loss": 0.2753, + "step": 6676 + }, + { + "epoch": 2.8, + "learning_rate": 1.5107769570716566e-05, + "loss": 0.2742, + "step": 6677 + }, + { + "epoch": 2.8, + "learning_rate": 1.510631193180907e-05, + "loss": 0.2392, + "step": 6678 + }, + { + "epoch": 2.8, + "learning_rate": 1.510485414612517e-05, + "loss": 0.2766, + "step": 6679 + }, + { + "epoch": 2.8, + "learning_rate": 1.510339621370677e-05, + "loss": 0.2514, + "step": 6680 + }, + { + "epoch": 2.8, + "learning_rate": 1.510193813459577e-05, + "loss": 0.2646, + "step": 6681 + }, + { + "epoch": 2.8, + "learning_rate": 1.5100479908834087e-05, + "loss": 0.2824, + "step": 6682 + }, + { + "epoch": 2.8, + "learning_rate": 1.5099021536463636e-05, + "loss": 0.2641, + "step": 6683 + }, + { + "epoch": 2.8, + "learning_rate": 1.5097563017526336e-05, + "loss": 0.2753, + "step": 6684 + }, + { + "epoch": 2.8, + "learning_rate": 1.5096104352064112e-05, + "loss": 0.2636, + "step": 6685 + }, + { + "epoch": 2.8, + "learning_rate": 1.5094645540118888e-05, + "loss": 0.2799, + "step": 6686 + }, + { + "epoch": 2.8, + "learning_rate": 1.5093186581732599e-05, + "loss": 0.2483, + "step": 6687 + }, + { + "epoch": 2.8, + "learning_rate": 1.5091727476947182e-05, + "loss": 0.2439, + "step": 6688 + }, + { + "epoch": 2.8, + "learning_rate": 1.5090268225804577e-05, + "loss": 0.25, + "step": 6689 + }, + { + "epoch": 2.8, + "learning_rate": 1.5088808828346727e-05, + "loss": 0.2747, + "step": 6690 + }, + { + "epoch": 2.8, + "learning_rate": 1.5087349284615584e-05, + "loss": 0.255, + "step": 6691 + }, + { + "epoch": 2.8, + "learning_rate": 1.50858895946531e-05, + "loss": 0.2375, + "step": 6692 + }, + { + "epoch": 2.8, + "learning_rate": 1.5084429758501235e-05, + "loss": 0.2505, + "step": 6693 + }, + { + "epoch": 2.8, + "learning_rate": 1.5082969776201948e-05, + "loss": 0.2722, + "step": 6694 + }, + { + "epoch": 2.8, + "learning_rate": 1.5081509647797204e-05, + "loss": 0.2526, + "step": 6695 + }, + { + "epoch": 2.8, + "learning_rate": 1.5080049373328976e-05, + "loss": 0.263, + "step": 6696 + }, + { + "epoch": 2.8, + "learning_rate": 1.5078588952839235e-05, + "loss": 0.2623, + "step": 6697 + }, + { + "epoch": 2.8, + "learning_rate": 1.507712838636996e-05, + "loss": 0.245, + "step": 6698 + }, + { + "epoch": 2.8, + "learning_rate": 1.5075667673963139e-05, + "loss": 0.2681, + "step": 6699 + }, + { + "epoch": 2.81, + "learning_rate": 1.507420681566075e-05, + "loss": 0.2744, + "step": 6700 + }, + { + "epoch": 2.81, + "learning_rate": 1.5072745811504793e-05, + "loss": 0.2359, + "step": 6701 + }, + { + "epoch": 2.81, + "learning_rate": 1.5071284661537257e-05, + "loss": 0.2707, + "step": 6702 + }, + { + "epoch": 2.81, + "learning_rate": 1.5069823365800146e-05, + "loss": 0.2543, + "step": 6703 + }, + { + "epoch": 2.81, + "learning_rate": 1.5068361924335457e-05, + "loss": 0.2385, + "step": 6704 + }, + { + "epoch": 2.81, + "learning_rate": 1.5066900337185203e-05, + "loss": 0.2427, + "step": 6705 + }, + { + "epoch": 2.81, + "learning_rate": 1.5065438604391398e-05, + "loss": 0.2722, + "step": 6706 + }, + { + "epoch": 2.81, + "learning_rate": 1.5063976725996052e-05, + "loss": 0.2621, + "step": 6707 + }, + { + "epoch": 2.81, + "learning_rate": 1.5062514702041193e-05, + "loss": 0.253, + "step": 6708 + }, + { + "epoch": 2.81, + "learning_rate": 1.5061052532568838e-05, + "loss": 0.2611, + "step": 6709 + }, + { + "epoch": 2.81, + "learning_rate": 1.5059590217621018e-05, + "loss": 0.2644, + "step": 6710 + }, + { + "epoch": 2.81, + "learning_rate": 1.5058127757239772e-05, + "loss": 0.2422, + "step": 6711 + }, + { + "epoch": 2.81, + "learning_rate": 1.5056665151467126e-05, + "loss": 0.2587, + "step": 6712 + }, + { + "epoch": 2.81, + "learning_rate": 1.505520240034513e-05, + "loss": 0.2658, + "step": 6713 + }, + { + "epoch": 2.81, + "learning_rate": 1.505373950391583e-05, + "loss": 0.248, + "step": 6714 + }, + { + "epoch": 2.81, + "learning_rate": 1.5052276462221267e-05, + "loss": 0.3038, + "step": 6715 + }, + { + "epoch": 2.81, + "learning_rate": 1.5050813275303501e-05, + "loss": 0.2474, + "step": 6716 + }, + { + "epoch": 2.81, + "learning_rate": 1.5049349943204595e-05, + "loss": 0.2528, + "step": 6717 + }, + { + "epoch": 2.81, + "learning_rate": 1.50478864659666e-05, + "loss": 0.2764, + "step": 6718 + }, + { + "epoch": 2.81, + "learning_rate": 1.5046422843631594e-05, + "loss": 0.2536, + "step": 6719 + }, + { + "epoch": 2.81, + "learning_rate": 1.5044959076241634e-05, + "loss": 0.246, + "step": 6720 + }, + { + "epoch": 2.81, + "learning_rate": 1.5043495163838809e-05, + "loss": 0.2581, + "step": 6721 + }, + { + "epoch": 2.81, + "learning_rate": 1.5042031106465186e-05, + "loss": 0.2556, + "step": 6722 + }, + { + "epoch": 2.81, + "learning_rate": 1.5040566904162855e-05, + "loss": 0.2449, + "step": 6723 + }, + { + "epoch": 2.82, + "learning_rate": 1.5039102556973899e-05, + "loss": 0.2626, + "step": 6724 + }, + { + "epoch": 2.82, + "learning_rate": 1.5037638064940416e-05, + "loss": 0.2653, + "step": 6725 + }, + { + "epoch": 2.82, + "learning_rate": 1.5036173428104493e-05, + "loss": 0.256, + "step": 6726 + }, + { + "epoch": 2.82, + "learning_rate": 1.5034708646508236e-05, + "loss": 0.2708, + "step": 6727 + }, + { + "epoch": 2.82, + "learning_rate": 1.5033243720193746e-05, + "loss": 0.2576, + "step": 6728 + }, + { + "epoch": 2.82, + "learning_rate": 1.5031778649203133e-05, + "loss": 0.2685, + "step": 6729 + }, + { + "epoch": 2.82, + "learning_rate": 1.5030313433578508e-05, + "loss": 0.2787, + "step": 6730 + }, + { + "epoch": 2.82, + "learning_rate": 1.5028848073361987e-05, + "loss": 0.2648, + "step": 6731 + }, + { + "epoch": 2.82, + "learning_rate": 1.5027382568595688e-05, + "loss": 0.2672, + "step": 6732 + }, + { + "epoch": 2.82, + "learning_rate": 1.5025916919321741e-05, + "loss": 0.2528, + "step": 6733 + }, + { + "epoch": 2.82, + "learning_rate": 1.5024451125582271e-05, + "loss": 0.2618, + "step": 6734 + }, + { + "epoch": 2.82, + "learning_rate": 1.5022985187419411e-05, + "loss": 0.2609, + "step": 6735 + }, + { + "epoch": 2.82, + "learning_rate": 1.50215191048753e-05, + "loss": 0.254, + "step": 6736 + }, + { + "epoch": 2.82, + "learning_rate": 1.5020052877992077e-05, + "loss": 0.2502, + "step": 6737 + }, + { + "epoch": 2.82, + "learning_rate": 1.5018586506811893e-05, + "loss": 0.24, + "step": 6738 + }, + { + "epoch": 2.82, + "learning_rate": 1.501711999137689e-05, + "loss": 0.2637, + "step": 6739 + }, + { + "epoch": 2.82, + "learning_rate": 1.5015653331729225e-05, + "loss": 0.2965, + "step": 6740 + }, + { + "epoch": 2.82, + "learning_rate": 1.5014186527911056e-05, + "loss": 0.2677, + "step": 6741 + }, + { + "epoch": 2.82, + "learning_rate": 1.5012719579964548e-05, + "loss": 0.2725, + "step": 6742 + }, + { + "epoch": 2.82, + "learning_rate": 1.501125248793186e-05, + "loss": 0.2443, + "step": 6743 + }, + { + "epoch": 2.82, + "learning_rate": 1.5009785251855164e-05, + "loss": 0.2601, + "step": 6744 + }, + { + "epoch": 2.82, + "learning_rate": 1.5008317871776643e-05, + "loss": 0.241, + "step": 6745 + }, + { + "epoch": 2.82, + "learning_rate": 1.5006850347738463e-05, + "loss": 0.2925, + "step": 6746 + }, + { + "epoch": 2.82, + "learning_rate": 1.5005382679782817e-05, + "loss": 0.26, + "step": 6747 + }, + { + "epoch": 2.83, + "learning_rate": 1.5003914867951888e-05, + "loss": 0.2512, + "step": 6748 + }, + { + "epoch": 2.83, + "learning_rate": 1.5002446912287862e-05, + "loss": 0.2741, + "step": 6749 + }, + { + "epoch": 2.83, + "learning_rate": 1.5000978812832942e-05, + "loss": 0.2485, + "step": 6750 + }, + { + "epoch": 2.83, + "learning_rate": 1.4999510569629323e-05, + "loss": 0.2567, + "step": 6751 + }, + { + "epoch": 2.83, + "learning_rate": 1.4998042182719207e-05, + "loss": 0.2542, + "step": 6752 + }, + { + "epoch": 2.83, + "learning_rate": 1.4996573652144807e-05, + "loss": 0.2555, + "step": 6753 + }, + { + "epoch": 2.83, + "learning_rate": 1.4995104977948331e-05, + "loss": 0.2351, + "step": 6754 + }, + { + "epoch": 2.83, + "learning_rate": 1.4993636160171994e-05, + "loss": 0.2558, + "step": 6755 + }, + { + "epoch": 2.83, + "learning_rate": 1.4992167198858016e-05, + "loss": 0.2562, + "step": 6756 + }, + { + "epoch": 2.83, + "learning_rate": 1.4990698094048624e-05, + "loss": 0.2462, + "step": 6757 + }, + { + "epoch": 2.83, + "learning_rate": 1.498922884578604e-05, + "loss": 0.2662, + "step": 6758 + }, + { + "epoch": 2.83, + "learning_rate": 1.4987759454112505e-05, + "loss": 0.2444, + "step": 6759 + }, + { + "epoch": 2.83, + "learning_rate": 1.4986289919070247e-05, + "loss": 0.2565, + "step": 6760 + }, + { + "epoch": 2.83, + "learning_rate": 1.4984820240701513e-05, + "loss": 0.2805, + "step": 6761 + }, + { + "epoch": 2.83, + "learning_rate": 1.4983350419048544e-05, + "loss": 0.254, + "step": 6762 + }, + { + "epoch": 2.83, + "learning_rate": 1.4981880454153586e-05, + "loss": 0.2438, + "step": 6763 + }, + { + "epoch": 2.83, + "learning_rate": 1.49804103460589e-05, + "loss": 0.2595, + "step": 6764 + }, + { + "epoch": 2.83, + "learning_rate": 1.4978940094806733e-05, + "loss": 0.254, + "step": 6765 + }, + { + "epoch": 2.83, + "learning_rate": 1.4977469700439358e-05, + "loss": 0.2385, + "step": 6766 + }, + { + "epoch": 2.83, + "learning_rate": 1.4975999162999027e-05, + "loss": 0.2545, + "step": 6767 + }, + { + "epoch": 2.83, + "learning_rate": 1.4974528482528018e-05, + "loss": 0.2445, + "step": 6768 + }, + { + "epoch": 2.83, + "learning_rate": 1.4973057659068601e-05, + "loss": 0.2655, + "step": 6769 + }, + { + "epoch": 2.83, + "learning_rate": 1.4971586692663055e-05, + "loss": 0.2913, + "step": 6770 + }, + { + "epoch": 2.83, + "learning_rate": 1.4970115583353661e-05, + "loss": 0.2701, + "step": 6771 + }, + { + "epoch": 2.84, + "learning_rate": 1.496864433118271e-05, + "loss": 0.263, + "step": 6772 + }, + { + "epoch": 2.84, + "learning_rate": 1.496717293619248e-05, + "loss": 0.2663, + "step": 6773 + }, + { + "epoch": 2.84, + "learning_rate": 1.4965701398425275e-05, + "loss": 0.2575, + "step": 6774 + }, + { + "epoch": 2.84, + "learning_rate": 1.496422971792339e-05, + "loss": 0.2518, + "step": 6775 + }, + { + "epoch": 2.84, + "learning_rate": 1.4962757894729128e-05, + "loss": 0.2548, + "step": 6776 + }, + { + "epoch": 2.84, + "learning_rate": 1.496128592888479e-05, + "loss": 0.2347, + "step": 6777 + }, + { + "epoch": 2.84, + "learning_rate": 1.4959813820432693e-05, + "loss": 0.2599, + "step": 6778 + }, + { + "epoch": 2.84, + "learning_rate": 1.4958341569415149e-05, + "loss": 0.2783, + "step": 6779 + }, + { + "epoch": 2.84, + "learning_rate": 1.4956869175874475e-05, + "loss": 0.2505, + "step": 6780 + }, + { + "epoch": 2.84, + "learning_rate": 1.4955396639852996e-05, + "loss": 0.2879, + "step": 6781 + }, + { + "epoch": 2.84, + "learning_rate": 1.4953923961393043e-05, + "loss": 0.251, + "step": 6782 + }, + { + "epoch": 2.84, + "learning_rate": 1.4952451140536934e-05, + "loss": 0.2481, + "step": 6783 + }, + { + "epoch": 2.84, + "learning_rate": 1.4950978177327016e-05, + "loss": 0.2704, + "step": 6784 + }, + { + "epoch": 2.84, + "learning_rate": 1.4949505071805624e-05, + "loss": 0.2769, + "step": 6785 + }, + { + "epoch": 2.84, + "learning_rate": 1.4948031824015098e-05, + "loss": 0.2836, + "step": 6786 + }, + { + "epoch": 2.84, + "learning_rate": 1.4946558433997792e-05, + "loss": 0.2574, + "step": 6787 + }, + { + "epoch": 2.84, + "learning_rate": 1.4945084901796048e-05, + "loss": 0.2546, + "step": 6788 + }, + { + "epoch": 2.84, + "learning_rate": 1.494361122745223e-05, + "loss": 0.2575, + "step": 6789 + }, + { + "epoch": 2.84, + "learning_rate": 1.4942137411008696e-05, + "loss": 0.2564, + "step": 6790 + }, + { + "epoch": 2.84, + "learning_rate": 1.4940663452507802e-05, + "loss": 0.2227, + "step": 6791 + }, + { + "epoch": 2.84, + "learning_rate": 1.4939189351991928e-05, + "loss": 0.2171, + "step": 6792 + }, + { + "epoch": 2.84, + "learning_rate": 1.4937715109503434e-05, + "loss": 0.2628, + "step": 6793 + }, + { + "epoch": 2.84, + "learning_rate": 1.4936240725084701e-05, + "loss": 0.256, + "step": 6794 + }, + { + "epoch": 2.84, + "learning_rate": 1.4934766198778111e-05, + "loss": 0.2534, + "step": 6795 + }, + { + "epoch": 2.85, + "learning_rate": 1.4933291530626047e-05, + "loss": 0.279, + "step": 6796 + }, + { + "epoch": 2.85, + "learning_rate": 1.493181672067089e-05, + "loss": 0.2554, + "step": 6797 + }, + { + "epoch": 2.85, + "learning_rate": 1.4930341768955043e-05, + "loss": 0.233, + "step": 6798 + }, + { + "epoch": 2.85, + "learning_rate": 1.4928866675520897e-05, + "loss": 0.2738, + "step": 6799 + }, + { + "epoch": 2.85, + "learning_rate": 1.492739144041085e-05, + "loss": 0.2688, + "step": 6800 + }, + { + "epoch": 2.85, + "learning_rate": 1.4925916063667309e-05, + "loss": 0.2855, + "step": 6801 + }, + { + "epoch": 2.85, + "learning_rate": 1.4924440545332681e-05, + "loss": 0.2513, + "step": 6802 + }, + { + "epoch": 2.85, + "learning_rate": 1.4922964885449382e-05, + "loss": 0.2448, + "step": 6803 + }, + { + "epoch": 2.85, + "learning_rate": 1.4921489084059823e-05, + "loss": 0.2518, + "step": 6804 + }, + { + "epoch": 2.85, + "learning_rate": 1.4920013141206428e-05, + "loss": 0.2839, + "step": 6805 + }, + { + "epoch": 2.85, + "learning_rate": 1.4918537056931623e-05, + "loss": 0.2599, + "step": 6806 + }, + { + "epoch": 2.85, + "learning_rate": 1.4917060831277835e-05, + "loss": 0.2528, + "step": 6807 + }, + { + "epoch": 2.85, + "learning_rate": 1.4915584464287494e-05, + "loss": 0.2628, + "step": 6808 + }, + { + "epoch": 2.85, + "learning_rate": 1.4914107956003044e-05, + "loss": 0.2673, + "step": 6809 + }, + { + "epoch": 2.85, + "learning_rate": 1.4912631306466918e-05, + "loss": 0.243, + "step": 6810 + }, + { + "epoch": 2.85, + "learning_rate": 1.4911154515721564e-05, + "loss": 0.2547, + "step": 6811 + }, + { + "epoch": 2.85, + "learning_rate": 1.4909677583809435e-05, + "loss": 0.2695, + "step": 6812 + }, + { + "epoch": 2.85, + "learning_rate": 1.4908200510772979e-05, + "loss": 0.2457, + "step": 6813 + }, + { + "epoch": 2.85, + "learning_rate": 1.4906723296654652e-05, + "loss": 0.2445, + "step": 6814 + }, + { + "epoch": 2.85, + "learning_rate": 1.490524594149692e-05, + "loss": 0.2652, + "step": 6815 + }, + { + "epoch": 2.85, + "learning_rate": 1.4903768445342248e-05, + "loss": 0.2757, + "step": 6816 + }, + { + "epoch": 2.85, + "learning_rate": 1.4902290808233103e-05, + "loss": 0.2322, + "step": 6817 + }, + { + "epoch": 2.85, + "learning_rate": 1.4900813030211956e-05, + "loss": 0.2429, + "step": 6818 + }, + { + "epoch": 2.85, + "learning_rate": 1.4899335111321289e-05, + "loss": 0.2251, + "step": 6819 + }, + { + "epoch": 2.86, + "learning_rate": 1.489785705160358e-05, + "loss": 0.2629, + "step": 6820 + }, + { + "epoch": 2.86, + "learning_rate": 1.4896378851101321e-05, + "loss": 0.2627, + "step": 6821 + }, + { + "epoch": 2.86, + "learning_rate": 1.4894900509856995e-05, + "loss": 0.2647, + "step": 6822 + }, + { + "epoch": 2.86, + "learning_rate": 1.4893422027913099e-05, + "loss": 0.2637, + "step": 6823 + }, + { + "epoch": 2.86, + "learning_rate": 1.4891943405312126e-05, + "loss": 0.25, + "step": 6824 + }, + { + "epoch": 2.86, + "learning_rate": 1.489046464209658e-05, + "loss": 0.2694, + "step": 6825 + }, + { + "epoch": 2.86, + "learning_rate": 1.488898573830897e-05, + "loss": 0.2741, + "step": 6826 + }, + { + "epoch": 2.86, + "learning_rate": 1.4887506693991803e-05, + "loss": 0.269, + "step": 6827 + }, + { + "epoch": 2.86, + "learning_rate": 1.4886027509187593e-05, + "loss": 0.2546, + "step": 6828 + }, + { + "epoch": 2.86, + "learning_rate": 1.4884548183938857e-05, + "loss": 0.2611, + "step": 6829 + }, + { + "epoch": 2.86, + "learning_rate": 1.4883068718288121e-05, + "loss": 0.2603, + "step": 6830 + }, + { + "epoch": 2.86, + "learning_rate": 1.4881589112277903e-05, + "loss": 0.2771, + "step": 6831 + }, + { + "epoch": 2.86, + "learning_rate": 1.4880109365950744e-05, + "loss": 0.2664, + "step": 6832 + }, + { + "epoch": 2.86, + "learning_rate": 1.4878629479349167e-05, + "loss": 0.2743, + "step": 6833 + }, + { + "epoch": 2.86, + "learning_rate": 1.4877149452515718e-05, + "loss": 0.2375, + "step": 6834 + }, + { + "epoch": 2.86, + "learning_rate": 1.4875669285492933e-05, + "loss": 0.2422, + "step": 6835 + }, + { + "epoch": 2.86, + "learning_rate": 1.487418897832336e-05, + "loss": 0.2433, + "step": 6836 + }, + { + "epoch": 2.86, + "learning_rate": 1.4872708531049554e-05, + "loss": 0.2557, + "step": 6837 + }, + { + "epoch": 2.86, + "learning_rate": 1.4871227943714063e-05, + "loss": 0.2563, + "step": 6838 + }, + { + "epoch": 2.86, + "learning_rate": 1.4869747216359446e-05, + "loss": 0.2676, + "step": 6839 + }, + { + "epoch": 2.86, + "learning_rate": 1.486826634902827e-05, + "loss": 0.2543, + "step": 6840 + }, + { + "epoch": 2.86, + "learning_rate": 1.4866785341763097e-05, + "loss": 0.2639, + "step": 6841 + }, + { + "epoch": 2.86, + "learning_rate": 1.4865304194606496e-05, + "loss": 0.2607, + "step": 6842 + }, + { + "epoch": 2.86, + "learning_rate": 1.4863822907601046e-05, + "loss": 0.2511, + "step": 6843 + }, + { + "epoch": 2.87, + "learning_rate": 1.4862341480789322e-05, + "loss": 0.2529, + "step": 6844 + }, + { + "epoch": 2.87, + "learning_rate": 1.4860859914213905e-05, + "loss": 0.2363, + "step": 6845 + }, + { + "epoch": 2.87, + "learning_rate": 1.4859378207917385e-05, + "loss": 0.2423, + "step": 6846 + }, + { + "epoch": 2.87, + "learning_rate": 1.485789636194235e-05, + "loss": 0.2616, + "step": 6847 + }, + { + "epoch": 2.87, + "learning_rate": 1.4856414376331394e-05, + "loss": 0.264, + "step": 6848 + }, + { + "epoch": 2.87, + "learning_rate": 1.4854932251127117e-05, + "loss": 0.2839, + "step": 6849 + }, + { + "epoch": 2.87, + "learning_rate": 1.4853449986372122e-05, + "loss": 0.253, + "step": 6850 + }, + { + "epoch": 2.87, + "learning_rate": 1.4851967582109013e-05, + "loss": 0.288, + "step": 6851 + }, + { + "epoch": 2.87, + "learning_rate": 1.48504850383804e-05, + "loss": 0.2654, + "step": 6852 + }, + { + "epoch": 2.87, + "learning_rate": 1.4849002355228901e-05, + "loss": 0.2839, + "step": 6853 + }, + { + "epoch": 2.87, + "learning_rate": 1.484751953269713e-05, + "loss": 0.2706, + "step": 6854 + }, + { + "epoch": 2.87, + "learning_rate": 1.4846036570827714e-05, + "loss": 0.2648, + "step": 6855 + }, + { + "epoch": 2.87, + "learning_rate": 1.4844553469663274e-05, + "loss": 0.2387, + "step": 6856 + }, + { + "epoch": 2.87, + "learning_rate": 1.4843070229246448e-05, + "loss": 0.2518, + "step": 6857 + }, + { + "epoch": 2.87, + "learning_rate": 1.4841586849619862e-05, + "loss": 0.2505, + "step": 6858 + }, + { + "epoch": 2.87, + "learning_rate": 1.4840103330826158e-05, + "loss": 0.2374, + "step": 6859 + }, + { + "epoch": 2.87, + "learning_rate": 1.4838619672907984e-05, + "loss": 0.2631, + "step": 6860 + }, + { + "epoch": 2.87, + "learning_rate": 1.4837135875907978e-05, + "loss": 0.2686, + "step": 6861 + }, + { + "epoch": 2.87, + "learning_rate": 1.4835651939868792e-05, + "loss": 0.2546, + "step": 6862 + }, + { + "epoch": 2.87, + "learning_rate": 1.4834167864833085e-05, + "loss": 0.2985, + "step": 6863 + }, + { + "epoch": 2.87, + "learning_rate": 1.483268365084351e-05, + "loss": 0.2512, + "step": 6864 + }, + { + "epoch": 2.87, + "learning_rate": 1.4831199297942731e-05, + "loss": 0.2642, + "step": 6865 + }, + { + "epoch": 2.87, + "learning_rate": 1.482971480617342e-05, + "loss": 0.2639, + "step": 6866 + }, + { + "epoch": 2.88, + "learning_rate": 1.482823017557824e-05, + "loss": 0.2675, + "step": 6867 + }, + { + "epoch": 2.88, + "learning_rate": 1.4826745406199865e-05, + "loss": 0.2623, + "step": 6868 + }, + { + "epoch": 2.88, + "learning_rate": 1.482526049808098e-05, + "loss": 0.2545, + "step": 6869 + }, + { + "epoch": 2.88, + "learning_rate": 1.4823775451264262e-05, + "loss": 0.2215, + "step": 6870 + }, + { + "epoch": 2.88, + "learning_rate": 1.4822290265792403e-05, + "loss": 0.229, + "step": 6871 + }, + { + "epoch": 2.88, + "learning_rate": 1.4820804941708086e-05, + "loss": 0.2672, + "step": 6872 + }, + { + "epoch": 2.88, + "learning_rate": 1.4819319479054009e-05, + "loss": 0.2455, + "step": 6873 + }, + { + "epoch": 2.88, + "learning_rate": 1.481783387787287e-05, + "loss": 0.2786, + "step": 6874 + }, + { + "epoch": 2.88, + "learning_rate": 1.4816348138207373e-05, + "loss": 0.2453, + "step": 6875 + }, + { + "epoch": 2.88, + "learning_rate": 1.4814862260100222e-05, + "loss": 0.2477, + "step": 6876 + }, + { + "epoch": 2.88, + "learning_rate": 1.4813376243594128e-05, + "loss": 0.275, + "step": 6877 + }, + { + "epoch": 2.88, + "learning_rate": 1.4811890088731808e-05, + "loss": 0.2415, + "step": 6878 + }, + { + "epoch": 2.88, + "learning_rate": 1.4810403795555975e-05, + "loss": 0.2582, + "step": 6879 + }, + { + "epoch": 2.88, + "learning_rate": 1.4808917364109352e-05, + "loss": 0.2525, + "step": 6880 + }, + { + "epoch": 2.88, + "learning_rate": 1.480743079443467e-05, + "loss": 0.2591, + "step": 6881 + }, + { + "epoch": 2.88, + "learning_rate": 1.4805944086574653e-05, + "loss": 0.2261, + "step": 6882 + }, + { + "epoch": 2.88, + "learning_rate": 1.4804457240572037e-05, + "loss": 0.2787, + "step": 6883 + }, + { + "epoch": 2.88, + "learning_rate": 1.4802970256469564e-05, + "loss": 0.2833, + "step": 6884 + }, + { + "epoch": 2.88, + "learning_rate": 1.4801483134309971e-05, + "loss": 0.2346, + "step": 6885 + }, + { + "epoch": 2.88, + "learning_rate": 1.4799995874136006e-05, + "loss": 0.2583, + "step": 6886 + }, + { + "epoch": 2.88, + "learning_rate": 1.4798508475990421e-05, + "loss": 0.2704, + "step": 6887 + }, + { + "epoch": 2.88, + "learning_rate": 1.4797020939915967e-05, + "loss": 0.2709, + "step": 6888 + }, + { + "epoch": 2.88, + "learning_rate": 1.4795533265955404e-05, + "loss": 0.267, + "step": 6889 + }, + { + "epoch": 2.88, + "learning_rate": 1.4794045454151491e-05, + "loss": 0.2466, + "step": 6890 + }, + { + "epoch": 2.89, + "learning_rate": 1.4792557504546997e-05, + "loss": 0.2482, + "step": 6891 + }, + { + "epoch": 2.89, + "learning_rate": 1.4791069417184689e-05, + "loss": 0.2537, + "step": 6892 + }, + { + "epoch": 2.89, + "learning_rate": 1.4789581192107343e-05, + "loss": 0.2675, + "step": 6893 + }, + { + "epoch": 2.89, + "learning_rate": 1.4788092829357736e-05, + "loss": 0.2571, + "step": 6894 + }, + { + "epoch": 2.89, + "learning_rate": 1.4786604328978648e-05, + "loss": 0.267, + "step": 6895 + }, + { + "epoch": 2.89, + "learning_rate": 1.4785115691012866e-05, + "loss": 0.2748, + "step": 6896 + }, + { + "epoch": 2.89, + "learning_rate": 1.4783626915503182e-05, + "loss": 0.2911, + "step": 6897 + }, + { + "epoch": 2.89, + "learning_rate": 1.4782138002492385e-05, + "loss": 0.2753, + "step": 6898 + }, + { + "epoch": 2.89, + "learning_rate": 1.4780648952023276e-05, + "loss": 0.245, + "step": 6899 + }, + { + "epoch": 2.89, + "learning_rate": 1.4779159764138656e-05, + "loss": 0.2457, + "step": 6900 + }, + { + "epoch": 2.89, + "learning_rate": 1.4777670438881328e-05, + "loss": 0.2733, + "step": 6901 + }, + { + "epoch": 2.89, + "learning_rate": 1.4776180976294105e-05, + "loss": 0.2523, + "step": 6902 + }, + { + "epoch": 2.89, + "learning_rate": 1.4774691376419795e-05, + "loss": 0.2498, + "step": 6903 + }, + { + "epoch": 2.89, + "learning_rate": 1.4773201639301223e-05, + "loss": 0.2705, + "step": 6904 + }, + { + "epoch": 2.89, + "learning_rate": 1.47717117649812e-05, + "loss": 0.2358, + "step": 6905 + }, + { + "epoch": 2.89, + "learning_rate": 1.4770221753502558e-05, + "loss": 0.2867, + "step": 6906 + }, + { + "epoch": 2.89, + "learning_rate": 1.4768731604908125e-05, + "loss": 0.2619, + "step": 6907 + }, + { + "epoch": 2.89, + "learning_rate": 1.4767241319240735e-05, + "loss": 0.2755, + "step": 6908 + }, + { + "epoch": 2.89, + "learning_rate": 1.4765750896543222e-05, + "loss": 0.2694, + "step": 6909 + }, + { + "epoch": 2.89, + "learning_rate": 1.476426033685843e-05, + "loss": 0.294, + "step": 6910 + }, + { + "epoch": 2.89, + "learning_rate": 1.4762769640229202e-05, + "loss": 0.235, + "step": 6911 + }, + { + "epoch": 2.89, + "learning_rate": 1.4761278806698389e-05, + "loss": 0.2529, + "step": 6912 + }, + { + "epoch": 2.89, + "learning_rate": 1.475978783630884e-05, + "loss": 0.2626, + "step": 6913 + }, + { + "epoch": 2.89, + "learning_rate": 1.4758296729103413e-05, + "loss": 0.2572, + "step": 6914 + }, + { + "epoch": 2.9, + "learning_rate": 1.475680548512497e-05, + "loss": 0.2899, + "step": 6915 + }, + { + "epoch": 2.9, + "learning_rate": 1.4755314104416371e-05, + "loss": 0.2459, + "step": 6916 + }, + { + "epoch": 2.9, + "learning_rate": 1.4753822587020491e-05, + "loss": 0.255, + "step": 6917 + }, + { + "epoch": 2.9, + "learning_rate": 1.47523309329802e-05, + "loss": 0.2422, + "step": 6918 + }, + { + "epoch": 2.9, + "learning_rate": 1.4750839142338372e-05, + "loss": 0.2726, + "step": 6919 + }, + { + "epoch": 2.9, + "learning_rate": 1.4749347215137888e-05, + "loss": 0.2683, + "step": 6920 + }, + { + "epoch": 2.9, + "learning_rate": 1.4747855151421636e-05, + "loss": 0.2625, + "step": 6921 + }, + { + "epoch": 2.9, + "learning_rate": 1.4746362951232499e-05, + "loss": 0.253, + "step": 6922 + }, + { + "epoch": 2.9, + "learning_rate": 1.4744870614613369e-05, + "loss": 0.2732, + "step": 6923 + }, + { + "epoch": 2.9, + "learning_rate": 1.4743378141607146e-05, + "loss": 0.2718, + "step": 6924 + }, + { + "epoch": 2.9, + "learning_rate": 1.4741885532256726e-05, + "loss": 0.2532, + "step": 6925 + }, + { + "epoch": 2.9, + "learning_rate": 1.4740392786605014e-05, + "loss": 0.2521, + "step": 6926 + }, + { + "epoch": 2.9, + "learning_rate": 1.473889990469492e-05, + "loss": 0.2389, + "step": 6927 + }, + { + "epoch": 2.9, + "learning_rate": 1.4737406886569352e-05, + "loss": 0.2739, + "step": 6928 + }, + { + "epoch": 2.9, + "learning_rate": 1.4735913732271226e-05, + "loss": 0.2551, + "step": 6929 + }, + { + "epoch": 2.9, + "learning_rate": 1.4734420441843462e-05, + "loss": 0.2653, + "step": 6930 + }, + { + "epoch": 2.9, + "learning_rate": 1.4732927015328989e-05, + "loss": 0.2595, + "step": 6931 + }, + { + "epoch": 2.9, + "learning_rate": 1.4731433452770723e-05, + "loss": 0.2651, + "step": 6932 + }, + { + "epoch": 2.9, + "learning_rate": 1.4729939754211602e-05, + "loss": 0.2988, + "step": 6933 + }, + { + "epoch": 2.9, + "learning_rate": 1.4728445919694563e-05, + "loss": 0.2706, + "step": 6934 + }, + { + "epoch": 2.9, + "learning_rate": 1.472695194926254e-05, + "loss": 0.2431, + "step": 6935 + }, + { + "epoch": 2.9, + "learning_rate": 1.4725457842958477e-05, + "loss": 0.2552, + "step": 6936 + }, + { + "epoch": 2.9, + "learning_rate": 1.4723963600825322e-05, + "loss": 0.2553, + "step": 6937 + }, + { + "epoch": 2.9, + "learning_rate": 1.472246922290603e-05, + "loss": 0.2611, + "step": 6938 + }, + { + "epoch": 2.91, + "learning_rate": 1.4720974709243542e-05, + "loss": 0.2267, + "step": 6939 + }, + { + "epoch": 2.91, + "learning_rate": 1.471948005988083e-05, + "loss": 0.2719, + "step": 6940 + }, + { + "epoch": 2.91, + "learning_rate": 1.471798527486085e-05, + "loss": 0.2622, + "step": 6941 + }, + { + "epoch": 2.91, + "learning_rate": 1.4716490354226573e-05, + "loss": 0.2586, + "step": 6942 + }, + { + "epoch": 2.91, + "learning_rate": 1.4714995298020963e-05, + "loss": 0.2677, + "step": 6943 + }, + { + "epoch": 2.91, + "learning_rate": 1.4713500106287e-05, + "loss": 0.2544, + "step": 6944 + }, + { + "epoch": 2.91, + "learning_rate": 1.4712004779067659e-05, + "loss": 0.2792, + "step": 6945 + }, + { + "epoch": 2.91, + "learning_rate": 1.471050931640592e-05, + "loss": 0.243, + "step": 6946 + }, + { + "epoch": 2.91, + "learning_rate": 1.4709013718344773e-05, + "loss": 0.2542, + "step": 6947 + }, + { + "epoch": 2.91, + "learning_rate": 1.4707517984927201e-05, + "loss": 0.2478, + "step": 6948 + }, + { + "epoch": 2.91, + "learning_rate": 1.4706022116196208e-05, + "loss": 0.2705, + "step": 6949 + }, + { + "epoch": 2.91, + "learning_rate": 1.4704526112194781e-05, + "loss": 0.2619, + "step": 6950 + }, + { + "epoch": 2.91, + "learning_rate": 1.4703029972965927e-05, + "loss": 0.2677, + "step": 6951 + }, + { + "epoch": 2.91, + "learning_rate": 1.470153369855265e-05, + "loss": 0.2332, + "step": 6952 + }, + { + "epoch": 2.91, + "learning_rate": 1.470003728899796e-05, + "loss": 0.2818, + "step": 6953 + }, + { + "epoch": 2.91, + "learning_rate": 1.4698540744344864e-05, + "loss": 0.2309, + "step": 6954 + }, + { + "epoch": 2.91, + "learning_rate": 1.469704406463639e-05, + "loss": 0.2381, + "step": 6955 + }, + { + "epoch": 2.91, + "learning_rate": 1.469554724991555e-05, + "loss": 0.2561, + "step": 6956 + }, + { + "epoch": 2.91, + "learning_rate": 1.469405030022537e-05, + "loss": 0.2776, + "step": 6957 + }, + { + "epoch": 2.91, + "learning_rate": 1.4692553215608878e-05, + "loss": 0.2784, + "step": 6958 + }, + { + "epoch": 2.91, + "learning_rate": 1.4691055996109112e-05, + "loss": 0.265, + "step": 6959 + }, + { + "epoch": 2.91, + "learning_rate": 1.4689558641769101e-05, + "loss": 0.2791, + "step": 6960 + }, + { + "epoch": 2.91, + "learning_rate": 1.4688061152631889e-05, + "loss": 0.2317, + "step": 6961 + }, + { + "epoch": 2.91, + "learning_rate": 1.468656352874052e-05, + "loss": 0.2672, + "step": 6962 + }, + { + "epoch": 2.92, + "learning_rate": 1.4685065770138043e-05, + "loss": 0.2632, + "step": 6963 + }, + { + "epoch": 2.92, + "learning_rate": 1.4683567876867503e-05, + "loss": 0.272, + "step": 6964 + }, + { + "epoch": 2.92, + "learning_rate": 1.4682069848971966e-05, + "loss": 0.2292, + "step": 6965 + }, + { + "epoch": 2.92, + "learning_rate": 1.4680571686494483e-05, + "loss": 0.2913, + "step": 6966 + }, + { + "epoch": 2.92, + "learning_rate": 1.467907338947812e-05, + "loss": 0.2553, + "step": 6967 + }, + { + "epoch": 2.92, + "learning_rate": 1.4677574957965946e-05, + "loss": 0.2323, + "step": 6968 + }, + { + "epoch": 2.92, + "learning_rate": 1.4676076392001032e-05, + "loss": 0.2707, + "step": 6969 + }, + { + "epoch": 2.92, + "learning_rate": 1.4674577691626448e-05, + "loss": 0.2586, + "step": 6970 + }, + { + "epoch": 2.92, + "learning_rate": 1.4673078856885277e-05, + "loss": 0.2988, + "step": 6971 + }, + { + "epoch": 2.92, + "learning_rate": 1.4671579887820602e-05, + "loss": 0.2444, + "step": 6972 + }, + { + "epoch": 2.92, + "learning_rate": 1.467008078447551e-05, + "loss": 0.2504, + "step": 6973 + }, + { + "epoch": 2.92, + "learning_rate": 1.4668581546893086e-05, + "loss": 0.2456, + "step": 6974 + }, + { + "epoch": 2.92, + "learning_rate": 1.4667082175116431e-05, + "loss": 0.2501, + "step": 6975 + }, + { + "epoch": 2.92, + "learning_rate": 1.466558266918864e-05, + "loss": 0.2867, + "step": 6976 + }, + { + "epoch": 2.92, + "learning_rate": 1.4664083029152815e-05, + "loss": 0.2356, + "step": 6977 + }, + { + "epoch": 2.92, + "learning_rate": 1.4662583255052064e-05, + "loss": 0.2852, + "step": 6978 + }, + { + "epoch": 2.92, + "learning_rate": 1.4661083346929492e-05, + "loss": 0.2831, + "step": 6979 + }, + { + "epoch": 2.92, + "learning_rate": 1.465958330482822e-05, + "loss": 0.2642, + "step": 6980 + }, + { + "epoch": 2.92, + "learning_rate": 1.4658083128791356e-05, + "loss": 0.251, + "step": 6981 + }, + { + "epoch": 2.92, + "learning_rate": 1.4656582818862028e-05, + "loss": 0.3004, + "step": 6982 + }, + { + "epoch": 2.92, + "learning_rate": 1.465508237508336e-05, + "loss": 0.2536, + "step": 6983 + }, + { + "epoch": 2.92, + "learning_rate": 1.4653581797498477e-05, + "loss": 0.2684, + "step": 6984 + }, + { + "epoch": 2.92, + "learning_rate": 1.4652081086150516e-05, + "loss": 0.2621, + "step": 6985 + }, + { + "epoch": 2.92, + "learning_rate": 1.4650580241082613e-05, + "loss": 0.2872, + "step": 6986 + }, + { + "epoch": 2.93, + "learning_rate": 1.4649079262337909e-05, + "loss": 0.2488, + "step": 6987 + }, + { + "epoch": 2.93, + "learning_rate": 1.4647578149959545e-05, + "loss": 0.2557, + "step": 6988 + }, + { + "epoch": 2.93, + "learning_rate": 1.4646076903990671e-05, + "loss": 0.2869, + "step": 6989 + }, + { + "epoch": 2.93, + "learning_rate": 1.4644575524474441e-05, + "loss": 0.2733, + "step": 6990 + }, + { + "epoch": 2.93, + "learning_rate": 1.4643074011454007e-05, + "loss": 0.2676, + "step": 6991 + }, + { + "epoch": 2.93, + "learning_rate": 1.4641572364972532e-05, + "loss": 0.2571, + "step": 6992 + }, + { + "epoch": 2.93, + "learning_rate": 1.464007058507318e-05, + "loss": 0.2744, + "step": 6993 + }, + { + "epoch": 2.93, + "learning_rate": 1.4638568671799113e-05, + "loss": 0.2485, + "step": 6994 + }, + { + "epoch": 2.93, + "learning_rate": 1.4637066625193508e-05, + "loss": 0.2437, + "step": 6995 + }, + { + "epoch": 2.93, + "learning_rate": 1.463556444529954e-05, + "loss": 0.2428, + "step": 6996 + }, + { + "epoch": 2.93, + "learning_rate": 1.4634062132160381e-05, + "loss": 0.2526, + "step": 6997 + }, + { + "epoch": 2.93, + "learning_rate": 1.4632559685819219e-05, + "loss": 0.2696, + "step": 6998 + }, + { + "epoch": 2.93, + "learning_rate": 1.4631057106319242e-05, + "loss": 0.2625, + "step": 6999 + }, + { + "epoch": 2.93, + "learning_rate": 1.4629554393703635e-05, + "loss": 0.259, + "step": 7000 + }, + { + "epoch": 2.93, + "learning_rate": 1.4628051548015596e-05, + "loss": 0.2856, + "step": 7001 + }, + { + "epoch": 2.93, + "learning_rate": 1.4626548569298324e-05, + "loss": 0.2357, + "step": 7002 + }, + { + "epoch": 2.93, + "learning_rate": 1.462504545759502e-05, + "loss": 0.2855, + "step": 7003 + }, + { + "epoch": 2.93, + "learning_rate": 1.4623542212948887e-05, + "loss": 0.2517, + "step": 7004 + }, + { + "epoch": 2.93, + "learning_rate": 1.4622038835403135e-05, + "loss": 0.2457, + "step": 7005 + }, + { + "epoch": 2.93, + "learning_rate": 1.4620535325000978e-05, + "loss": 0.2696, + "step": 7006 + }, + { + "epoch": 2.93, + "learning_rate": 1.4619031681785633e-05, + "loss": 0.2551, + "step": 7007 + }, + { + "epoch": 2.93, + "learning_rate": 1.4617527905800321e-05, + "loss": 0.2464, + "step": 7008 + }, + { + "epoch": 2.93, + "learning_rate": 1.461602399708827e-05, + "loss": 0.2891, + "step": 7009 + }, + { + "epoch": 2.93, + "learning_rate": 1.4614519955692701e-05, + "loss": 0.2391, + "step": 7010 + }, + { + "epoch": 2.94, + "learning_rate": 1.4613015781656854e-05, + "loss": 0.2525, + "step": 7011 + }, + { + "epoch": 2.94, + "learning_rate": 1.4611511475023958e-05, + "loss": 0.2371, + "step": 7012 + }, + { + "epoch": 2.94, + "learning_rate": 1.461000703583726e-05, + "loss": 0.2491, + "step": 7013 + }, + { + "epoch": 2.94, + "learning_rate": 1.4608502464140002e-05, + "loss": 0.2961, + "step": 7014 + }, + { + "epoch": 2.94, + "learning_rate": 1.4606997759975427e-05, + "loss": 0.2514, + "step": 7015 + }, + { + "epoch": 2.94, + "learning_rate": 1.4605492923386795e-05, + "loss": 0.2443, + "step": 7016 + }, + { + "epoch": 2.94, + "learning_rate": 1.4603987954417349e-05, + "loss": 0.2637, + "step": 7017 + }, + { + "epoch": 2.94, + "learning_rate": 1.4602482853110358e-05, + "loss": 0.2675, + "step": 7018 + }, + { + "epoch": 2.94, + "learning_rate": 1.460097761950908e-05, + "loss": 0.2693, + "step": 7019 + }, + { + "epoch": 2.94, + "learning_rate": 1.4599472253656786e-05, + "loss": 0.2458, + "step": 7020 + }, + { + "epoch": 2.94, + "learning_rate": 1.459796675559674e-05, + "loss": 0.2845, + "step": 7021 + }, + { + "epoch": 2.94, + "learning_rate": 1.4596461125372221e-05, + "loss": 0.2674, + "step": 7022 + }, + { + "epoch": 2.94, + "learning_rate": 1.4594955363026509e-05, + "loss": 0.2461, + "step": 7023 + }, + { + "epoch": 2.94, + "learning_rate": 1.4593449468602878e-05, + "loss": 0.2608, + "step": 7024 + }, + { + "epoch": 2.94, + "learning_rate": 1.459194344214462e-05, + "loss": 0.25, + "step": 7025 + }, + { + "epoch": 2.94, + "learning_rate": 1.4590437283695022e-05, + "loss": 0.2608, + "step": 7026 + }, + { + "epoch": 2.94, + "learning_rate": 1.4588930993297381e-05, + "loss": 0.2597, + "step": 7027 + }, + { + "epoch": 2.94, + "learning_rate": 1.4587424570994989e-05, + "loss": 0.2485, + "step": 7028 + }, + { + "epoch": 2.94, + "learning_rate": 1.4585918016831144e-05, + "loss": 0.3006, + "step": 7029 + }, + { + "epoch": 2.94, + "learning_rate": 1.4584411330849162e-05, + "loss": 0.2837, + "step": 7030 + }, + { + "epoch": 2.94, + "learning_rate": 1.458290451309234e-05, + "loss": 0.243, + "step": 7031 + }, + { + "epoch": 2.94, + "learning_rate": 1.4581397563603994e-05, + "loss": 0.2504, + "step": 7032 + }, + { + "epoch": 2.94, + "learning_rate": 1.4579890482427446e-05, + "loss": 0.27, + "step": 7033 + }, + { + "epoch": 2.94, + "learning_rate": 1.4578383269606004e-05, + "loss": 0.228, + "step": 7034 + }, + { + "epoch": 2.95, + "learning_rate": 1.4576875925182998e-05, + "loss": 0.2386, + "step": 7035 + }, + { + "epoch": 2.95, + "learning_rate": 1.4575368449201759e-05, + "loss": 0.2873, + "step": 7036 + }, + { + "epoch": 2.95, + "learning_rate": 1.4573860841705613e-05, + "loss": 0.2577, + "step": 7037 + }, + { + "epoch": 2.95, + "learning_rate": 1.4572353102737894e-05, + "loss": 0.2532, + "step": 7038 + }, + { + "epoch": 2.95, + "learning_rate": 1.4570845232341943e-05, + "loss": 0.249, + "step": 7039 + }, + { + "epoch": 2.95, + "learning_rate": 1.4569337230561105e-05, + "loss": 0.2475, + "step": 7040 + }, + { + "epoch": 2.95, + "learning_rate": 1.4567829097438719e-05, + "loss": 0.2634, + "step": 7041 + }, + { + "epoch": 2.95, + "learning_rate": 1.456632083301814e-05, + "loss": 0.2866, + "step": 7042 + }, + { + "epoch": 2.95, + "learning_rate": 1.4564812437342724e-05, + "loss": 0.2312, + "step": 7043 + }, + { + "epoch": 2.95, + "learning_rate": 1.4563303910455821e-05, + "loss": 0.2826, + "step": 7044 + }, + { + "epoch": 2.95, + "learning_rate": 1.4561795252400797e-05, + "loss": 0.2687, + "step": 7045 + }, + { + "epoch": 2.95, + "learning_rate": 1.4560286463221018e-05, + "loss": 0.2409, + "step": 7046 + }, + { + "epoch": 2.95, + "learning_rate": 1.455877754295985e-05, + "loss": 0.2738, + "step": 7047 + }, + { + "epoch": 2.95, + "learning_rate": 1.455726849166067e-05, + "loss": 0.2755, + "step": 7048 + }, + { + "epoch": 2.95, + "learning_rate": 1.4555759309366848e-05, + "loss": 0.2729, + "step": 7049 + }, + { + "epoch": 2.95, + "learning_rate": 1.455424999612177e-05, + "loss": 0.2863, + "step": 7050 + }, + { + "epoch": 2.95, + "learning_rate": 1.4552740551968817e-05, + "loss": 0.2536, + "step": 7051 + }, + { + "epoch": 2.95, + "learning_rate": 1.4551230976951373e-05, + "loss": 0.2727, + "step": 7052 + }, + { + "epoch": 2.95, + "learning_rate": 1.4549721271112835e-05, + "loss": 0.2709, + "step": 7053 + }, + { + "epoch": 2.95, + "learning_rate": 1.4548211434496598e-05, + "loss": 0.2484, + "step": 7054 + }, + { + "epoch": 2.95, + "learning_rate": 1.4546701467146058e-05, + "loss": 0.2855, + "step": 7055 + }, + { + "epoch": 2.95, + "learning_rate": 1.4545191369104618e-05, + "loss": 0.265, + "step": 7056 + }, + { + "epoch": 2.95, + "learning_rate": 1.4543681140415688e-05, + "loss": 0.2499, + "step": 7057 + }, + { + "epoch": 2.95, + "learning_rate": 1.4542170781122673e-05, + "loss": 0.26, + "step": 7058 + }, + { + "epoch": 2.96, + "learning_rate": 1.4540660291268993e-05, + "loss": 0.2607, + "step": 7059 + }, + { + "epoch": 2.96, + "learning_rate": 1.453914967089806e-05, + "loss": 0.2486, + "step": 7060 + }, + { + "epoch": 2.96, + "learning_rate": 1.4537638920053298e-05, + "loss": 0.3044, + "step": 7061 + }, + { + "epoch": 2.96, + "learning_rate": 1.4536128038778132e-05, + "loss": 0.2689, + "step": 7062 + }, + { + "epoch": 2.96, + "learning_rate": 1.453461702711599e-05, + "loss": 0.2797, + "step": 7063 + }, + { + "epoch": 2.96, + "learning_rate": 1.4533105885110307e-05, + "loss": 0.2443, + "step": 7064 + }, + { + "epoch": 2.96, + "learning_rate": 1.4531594612804516e-05, + "loss": 0.2563, + "step": 7065 + }, + { + "epoch": 2.96, + "learning_rate": 1.4530083210242062e-05, + "loss": 0.2787, + "step": 7066 + }, + { + "epoch": 2.96, + "learning_rate": 1.4528571677466383e-05, + "loss": 0.2638, + "step": 7067 + }, + { + "epoch": 2.96, + "learning_rate": 1.4527060014520932e-05, + "loss": 0.2708, + "step": 7068 + }, + { + "epoch": 2.96, + "learning_rate": 1.4525548221449156e-05, + "loss": 0.235, + "step": 7069 + }, + { + "epoch": 2.96, + "learning_rate": 1.4524036298294515e-05, + "loss": 0.2447, + "step": 7070 + }, + { + "epoch": 2.96, + "learning_rate": 1.4522524245100465e-05, + "loss": 0.2411, + "step": 7071 + }, + { + "epoch": 2.96, + "learning_rate": 1.4521012061910466e-05, + "loss": 0.26, + "step": 7072 + }, + { + "epoch": 2.96, + "learning_rate": 1.4519499748767989e-05, + "loss": 0.2866, + "step": 7073 + }, + { + "epoch": 2.96, + "learning_rate": 1.4517987305716502e-05, + "loss": 0.2783, + "step": 7074 + }, + { + "epoch": 2.96, + "learning_rate": 1.4516474732799476e-05, + "loss": 0.2697, + "step": 7075 + }, + { + "epoch": 2.96, + "learning_rate": 1.4514962030060395e-05, + "loss": 0.2484, + "step": 7076 + }, + { + "epoch": 2.96, + "learning_rate": 1.451344919754274e-05, + "loss": 0.2751, + "step": 7077 + }, + { + "epoch": 2.96, + "learning_rate": 1.4511936235289987e-05, + "loss": 0.2818, + "step": 7078 + }, + { + "epoch": 2.96, + "learning_rate": 1.451042314334563e-05, + "loss": 0.2503, + "step": 7079 + }, + { + "epoch": 2.96, + "learning_rate": 1.450890992175317e-05, + "loss": 0.2401, + "step": 7080 + }, + { + "epoch": 2.96, + "learning_rate": 1.450739657055609e-05, + "loss": 0.2721, + "step": 7081 + }, + { + "epoch": 2.97, + "learning_rate": 1.4505883089797896e-05, + "loss": 0.2605, + "step": 7082 + }, + { + "epoch": 2.97, + "learning_rate": 1.4504369479522092e-05, + "loss": 0.2714, + "step": 7083 + }, + { + "epoch": 2.97, + "learning_rate": 1.4502855739772184e-05, + "loss": 0.2754, + "step": 7084 + }, + { + "epoch": 2.97, + "learning_rate": 1.4501341870591683e-05, + "loss": 0.2411, + "step": 7085 + }, + { + "epoch": 2.97, + "learning_rate": 1.4499827872024104e-05, + "loss": 0.2623, + "step": 7086 + }, + { + "epoch": 2.97, + "learning_rate": 1.4498313744112966e-05, + "loss": 0.2632, + "step": 7087 + }, + { + "epoch": 2.97, + "learning_rate": 1.4496799486901791e-05, + "loss": 0.2663, + "step": 7088 + }, + { + "epoch": 2.97, + "learning_rate": 1.4495285100434107e-05, + "loss": 0.2793, + "step": 7089 + }, + { + "epoch": 2.97, + "learning_rate": 1.449377058475344e-05, + "loss": 0.2695, + "step": 7090 + }, + { + "epoch": 2.97, + "learning_rate": 1.4492255939903326e-05, + "loss": 0.2576, + "step": 7091 + }, + { + "epoch": 2.97, + "learning_rate": 1.44907411659273e-05, + "loss": 0.2428, + "step": 7092 + }, + { + "epoch": 2.97, + "learning_rate": 1.4489226262868906e-05, + "loss": 0.2305, + "step": 7093 + }, + { + "epoch": 2.97, + "learning_rate": 1.4487711230771686e-05, + "loss": 0.2953, + "step": 7094 + }, + { + "epoch": 2.97, + "learning_rate": 1.4486196069679191e-05, + "loss": 0.2662, + "step": 7095 + }, + { + "epoch": 2.97, + "learning_rate": 1.4484680779634968e-05, + "loss": 0.2287, + "step": 7096 + }, + { + "epoch": 2.97, + "learning_rate": 1.4483165360682577e-05, + "loss": 0.2736, + "step": 7097 + }, + { + "epoch": 2.97, + "learning_rate": 1.4481649812865574e-05, + "loss": 0.2746, + "step": 7098 + }, + { + "epoch": 2.97, + "learning_rate": 1.4480134136227526e-05, + "loss": 0.2411, + "step": 7099 + }, + { + "epoch": 2.97, + "learning_rate": 1.4478618330811997e-05, + "loss": 0.2587, + "step": 7100 + }, + { + "epoch": 2.97, + "learning_rate": 1.447710239666256e-05, + "loss": 0.2382, + "step": 7101 + }, + { + "epoch": 2.97, + "learning_rate": 1.4475586333822787e-05, + "loss": 0.2577, + "step": 7102 + }, + { + "epoch": 2.97, + "learning_rate": 1.4474070142336254e-05, + "loss": 0.2714, + "step": 7103 + }, + { + "epoch": 2.97, + "learning_rate": 1.4472553822246548e-05, + "loss": 0.2527, + "step": 7104 + }, + { + "epoch": 2.97, + "learning_rate": 1.4471037373597251e-05, + "loss": 0.292, + "step": 7105 + }, + { + "epoch": 2.98, + "learning_rate": 1.446952079643195e-05, + "loss": 0.2405, + "step": 7106 + }, + { + "epoch": 2.98, + "learning_rate": 1.4468004090794241e-05, + "loss": 0.2724, + "step": 7107 + }, + { + "epoch": 2.98, + "learning_rate": 1.4466487256727721e-05, + "loss": 0.2636, + "step": 7108 + }, + { + "epoch": 2.98, + "learning_rate": 1.4464970294275986e-05, + "loss": 0.2888, + "step": 7109 + }, + { + "epoch": 2.98, + "learning_rate": 1.4463453203482643e-05, + "loss": 0.2672, + "step": 7110 + }, + { + "epoch": 2.98, + "learning_rate": 1.44619359843913e-05, + "loss": 0.2623, + "step": 7111 + }, + { + "epoch": 2.98, + "learning_rate": 1.4460418637045564e-05, + "loss": 0.2394, + "step": 7112 + }, + { + "epoch": 2.98, + "learning_rate": 1.4458901161489051e-05, + "loss": 0.2543, + "step": 7113 + }, + { + "epoch": 2.98, + "learning_rate": 1.4457383557765385e-05, + "loss": 0.2451, + "step": 7114 + }, + { + "epoch": 2.98, + "learning_rate": 1.4455865825918182e-05, + "loss": 0.2542, + "step": 7115 + }, + { + "epoch": 2.98, + "learning_rate": 1.445434796599107e-05, + "loss": 0.2598, + "step": 7116 + }, + { + "epoch": 2.98, + "learning_rate": 1.4452829978027678e-05, + "loss": 0.2718, + "step": 7117 + }, + { + "epoch": 2.98, + "learning_rate": 1.4451311862071641e-05, + "loss": 0.2405, + "step": 7118 + }, + { + "epoch": 2.98, + "learning_rate": 1.4449793618166594e-05, + "loss": 0.2497, + "step": 7119 + }, + { + "epoch": 2.98, + "learning_rate": 1.4448275246356175e-05, + "loss": 0.2675, + "step": 7120 + }, + { + "epoch": 2.98, + "learning_rate": 1.4446756746684036e-05, + "loss": 0.2779, + "step": 7121 + }, + { + "epoch": 2.98, + "learning_rate": 1.4445238119193818e-05, + "loss": 0.2645, + "step": 7122 + }, + { + "epoch": 2.98, + "learning_rate": 1.4443719363929173e-05, + "loss": 0.2663, + "step": 7123 + }, + { + "epoch": 2.98, + "learning_rate": 1.4442200480933761e-05, + "loss": 0.2581, + "step": 7124 + }, + { + "epoch": 2.98, + "learning_rate": 1.4440681470251239e-05, + "loss": 0.2753, + "step": 7125 + }, + { + "epoch": 2.98, + "learning_rate": 1.4439162331925268e-05, + "loss": 0.2406, + "step": 7126 + }, + { + "epoch": 2.98, + "learning_rate": 1.4437643065999516e-05, + "loss": 0.2575, + "step": 7127 + }, + { + "epoch": 2.98, + "learning_rate": 1.443612367251765e-05, + "loss": 0.2494, + "step": 7128 + }, + { + "epoch": 2.98, + "learning_rate": 1.4434604151523345e-05, + "loss": 0.2552, + "step": 7129 + }, + { + "epoch": 2.99, + "learning_rate": 1.443308450306028e-05, + "loss": 0.2535, + "step": 7130 + }, + { + "epoch": 2.99, + "learning_rate": 1.4431564727172135e-05, + "loss": 0.2476, + "step": 7131 + }, + { + "epoch": 2.99, + "learning_rate": 1.4430044823902594e-05, + "loss": 0.2381, + "step": 7132 + }, + { + "epoch": 2.99, + "learning_rate": 1.4428524793295344e-05, + "loss": 0.2544, + "step": 7133 + }, + { + "epoch": 2.99, + "learning_rate": 1.442700463539408e-05, + "loss": 0.269, + "step": 7134 + }, + { + "epoch": 2.99, + "learning_rate": 1.4425484350242498e-05, + "loss": 0.24, + "step": 7135 + }, + { + "epoch": 2.99, + "learning_rate": 1.4423963937884293e-05, + "loss": 0.2364, + "step": 7136 + }, + { + "epoch": 2.99, + "learning_rate": 1.442244339836317e-05, + "loss": 0.2322, + "step": 7137 + }, + { + "epoch": 2.99, + "learning_rate": 1.442092273172284e-05, + "loss": 0.2413, + "step": 7138 + }, + { + "epoch": 2.99, + "learning_rate": 1.4419401938007007e-05, + "loss": 0.2707, + "step": 7139 + }, + { + "epoch": 2.99, + "learning_rate": 1.4417881017259385e-05, + "loss": 0.3087, + "step": 7140 + }, + { + "epoch": 2.99, + "learning_rate": 1.4416359969523694e-05, + "loss": 0.265, + "step": 7141 + }, + { + "epoch": 2.99, + "learning_rate": 1.4414838794843658e-05, + "loss": 0.2298, + "step": 7142 + }, + { + "epoch": 2.99, + "learning_rate": 1.4413317493262996e-05, + "loss": 0.2708, + "step": 7143 + }, + { + "epoch": 2.99, + "learning_rate": 1.4411796064825436e-05, + "loss": 0.2349, + "step": 7144 + }, + { + "epoch": 2.99, + "learning_rate": 1.4410274509574718e-05, + "loss": 0.2678, + "step": 7145 + }, + { + "epoch": 2.99, + "learning_rate": 1.4408752827554571e-05, + "loss": 0.2931, + "step": 7146 + }, + { + "epoch": 2.99, + "learning_rate": 1.4407231018808734e-05, + "loss": 0.2673, + "step": 7147 + }, + { + "epoch": 2.99, + "learning_rate": 1.4405709083380957e-05, + "loss": 0.2571, + "step": 7148 + }, + { + "epoch": 2.99, + "learning_rate": 1.4404187021314982e-05, + "loss": 0.2492, + "step": 7149 + }, + { + "epoch": 2.99, + "learning_rate": 1.4402664832654555e-05, + "loss": 0.2863, + "step": 7150 + }, + { + "epoch": 2.99, + "learning_rate": 1.4401142517443434e-05, + "loss": 0.2501, + "step": 7151 + }, + { + "epoch": 2.99, + "learning_rate": 1.439962007572538e-05, + "loss": 0.2647, + "step": 7152 + }, + { + "epoch": 2.99, + "learning_rate": 1.4398097507544152e-05, + "loss": 0.2374, + "step": 7153 + }, + { + "epoch": 3.0, + "learning_rate": 1.439657481294351e-05, + "loss": 0.2462, + "step": 7154 + }, + { + "epoch": 3.0, + "learning_rate": 1.4395051991967228e-05, + "loss": 0.2512, + "step": 7155 + }, + { + "epoch": 3.0, + "learning_rate": 1.4393529044659077e-05, + "loss": 0.2685, + "step": 7156 + }, + { + "epoch": 3.0, + "learning_rate": 1.439200597106283e-05, + "loss": 0.2732, + "step": 7157 + }, + { + "epoch": 3.0, + "learning_rate": 1.4390482771222273e-05, + "loss": 0.2834, + "step": 7158 + }, + { + "epoch": 3.0, + "learning_rate": 1.4388959445181181e-05, + "loss": 0.2718, + "step": 7159 + }, + { + "epoch": 3.0, + "learning_rate": 1.4387435992983347e-05, + "loss": 0.2553, + "step": 7160 + }, + { + "epoch": 3.0, + "learning_rate": 1.4385912414672559e-05, + "loss": 0.23, + "step": 7161 + }, + { + "epoch": 3.0, + "learning_rate": 1.438438871029261e-05, + "loss": 0.2289, + "step": 7162 + }, + { + "epoch": 3.0, + "learning_rate": 1.4382864879887298e-05, + "loss": 0.2558, + "step": 7163 + }, + { + "epoch": 3.0, + "learning_rate": 1.4381340923500422e-05, + "loss": 0.2601, + "step": 7164 + }, + { + "epoch": 3.0, + "learning_rate": 1.437981684117579e-05, + "loss": 0.2692, + "step": 7165 + }, + { + "epoch": 3.0, + "learning_rate": 1.4378292632957214e-05, + "loss": 0.2586, + "step": 7166 + }, + { + "epoch": 3.0, + "learning_rate": 1.4376768298888498e-05, + "loss": 0.1887, + "step": 7167 + }, + { + "epoch": 3.0, + "learning_rate": 1.4375243839013458e-05, + "loss": 0.2059, + "step": 7168 + }, + { + "epoch": 3.0, + "learning_rate": 1.4373719253375921e-05, + "loss": 0.231, + "step": 7169 + }, + { + "epoch": 3.0, + "learning_rate": 1.4372194542019705e-05, + "loss": 0.1802, + "step": 7170 + }, + { + "epoch": 3.0, + "learning_rate": 1.4370669704988634e-05, + "loss": 0.1511, + "step": 7171 + }, + { + "epoch": 3.0, + "learning_rate": 1.4369144742326544e-05, + "loss": 0.1502, + "step": 7172 + }, + { + "epoch": 3.0, + "learning_rate": 1.4367619654077264e-05, + "loss": 0.1817, + "step": 7173 + }, + { + "epoch": 3.0, + "learning_rate": 1.4366094440284632e-05, + "loss": 0.1603, + "step": 7174 + }, + { + "epoch": 3.0, + "learning_rate": 1.4364569100992489e-05, + "loss": 0.2057, + "step": 7175 + }, + { + "epoch": 3.0, + "learning_rate": 1.4363043636244682e-05, + "loss": 0.1826, + "step": 7176 + }, + { + "epoch": 3.0, + "learning_rate": 1.4361518046085057e-05, + "loss": 0.1632, + "step": 7177 + }, + { + "epoch": 3.01, + "learning_rate": 1.4359992330557467e-05, + "loss": 0.1868, + "step": 7178 + }, + { + "epoch": 3.01, + "learning_rate": 1.4358466489705766e-05, + "loss": 0.1837, + "step": 7179 + }, + { + "epoch": 3.01, + "learning_rate": 1.4356940523573812e-05, + "loss": 0.2101, + "step": 7180 + }, + { + "epoch": 3.01, + "learning_rate": 1.4355414432205468e-05, + "loss": 0.1869, + "step": 7181 + }, + { + "epoch": 3.01, + "learning_rate": 1.4353888215644604e-05, + "loss": 0.2015, + "step": 7182 + }, + { + "epoch": 3.01, + "learning_rate": 1.4352361873935088e-05, + "loss": 0.1817, + "step": 7183 + }, + { + "epoch": 3.01, + "learning_rate": 1.435083540712079e-05, + "loss": 0.2035, + "step": 7184 + }, + { + "epoch": 3.01, + "learning_rate": 1.4349308815245586e-05, + "loss": 0.1981, + "step": 7185 + }, + { + "epoch": 3.01, + "learning_rate": 1.4347782098353364e-05, + "loss": 0.1769, + "step": 7186 + }, + { + "epoch": 3.01, + "learning_rate": 1.4346255256488002e-05, + "loss": 0.1934, + "step": 7187 + }, + { + "epoch": 3.01, + "learning_rate": 1.4344728289693391e-05, + "loss": 0.1702, + "step": 7188 + }, + { + "epoch": 3.01, + "learning_rate": 1.434320119801342e-05, + "loss": 0.177, + "step": 7189 + }, + { + "epoch": 3.01, + "learning_rate": 1.4341673981491982e-05, + "loss": 0.1806, + "step": 7190 + }, + { + "epoch": 3.01, + "learning_rate": 1.4340146640172981e-05, + "loss": 0.2014, + "step": 7191 + }, + { + "epoch": 3.01, + "learning_rate": 1.4338619174100315e-05, + "loss": 0.1654, + "step": 7192 + }, + { + "epoch": 3.01, + "learning_rate": 1.433709158331789e-05, + "loss": 0.1999, + "step": 7193 + }, + { + "epoch": 3.01, + "learning_rate": 1.4335563867869618e-05, + "loss": 0.1922, + "step": 7194 + }, + { + "epoch": 3.01, + "learning_rate": 1.433403602779941e-05, + "loss": 0.1618, + "step": 7195 + }, + { + "epoch": 3.01, + "learning_rate": 1.4332508063151184e-05, + "loss": 0.1569, + "step": 7196 + }, + { + "epoch": 3.01, + "learning_rate": 1.4330979973968856e-05, + "loss": 0.2257, + "step": 7197 + }, + { + "epoch": 3.01, + "learning_rate": 1.4329451760296352e-05, + "loss": 0.1733, + "step": 7198 + }, + { + "epoch": 3.01, + "learning_rate": 1.43279234221776e-05, + "loss": 0.2231, + "step": 7199 + }, + { + "epoch": 3.01, + "learning_rate": 1.4326394959656527e-05, + "loss": 0.1967, + "step": 7200 + }, + { + "epoch": 3.01, + "learning_rate": 1.4324866372777072e-05, + "loss": 0.2143, + "step": 7201 + }, + { + "epoch": 3.02, + "learning_rate": 1.4323337661583168e-05, + "loss": 0.1895, + "step": 7202 + }, + { + "epoch": 3.02, + "learning_rate": 1.4321808826118764e-05, + "loss": 0.2105, + "step": 7203 + }, + { + "epoch": 3.02, + "learning_rate": 1.4320279866427798e-05, + "loss": 0.1458, + "step": 7204 + }, + { + "epoch": 3.02, + "learning_rate": 1.431875078255422e-05, + "loss": 0.1854, + "step": 7205 + }, + { + "epoch": 3.02, + "learning_rate": 1.4317221574541985e-05, + "loss": 0.1838, + "step": 7206 + }, + { + "epoch": 3.02, + "learning_rate": 1.4315692242435048e-05, + "loss": 0.1533, + "step": 7207 + }, + { + "epoch": 3.02, + "learning_rate": 1.4314162786277366e-05, + "loss": 0.1994, + "step": 7208 + }, + { + "epoch": 3.02, + "learning_rate": 1.43126332061129e-05, + "loss": 0.2194, + "step": 7209 + }, + { + "epoch": 3.02, + "learning_rate": 1.4311103501985625e-05, + "loss": 0.169, + "step": 7210 + }, + { + "epoch": 3.02, + "learning_rate": 1.4309573673939503e-05, + "loss": 0.2104, + "step": 7211 + }, + { + "epoch": 3.02, + "learning_rate": 1.4308043722018508e-05, + "loss": 0.2042, + "step": 7212 + }, + { + "epoch": 3.02, + "learning_rate": 1.4306513646266621e-05, + "loss": 0.1725, + "step": 7213 + }, + { + "epoch": 3.02, + "learning_rate": 1.430498344672782e-05, + "loss": 0.1701, + "step": 7214 + }, + { + "epoch": 3.02, + "learning_rate": 1.430345312344609e-05, + "loss": 0.1718, + "step": 7215 + }, + { + "epoch": 3.02, + "learning_rate": 1.430192267646542e-05, + "loss": 0.1763, + "step": 7216 + }, + { + "epoch": 3.02, + "learning_rate": 1.4300392105829802e-05, + "loss": 0.17, + "step": 7217 + }, + { + "epoch": 3.02, + "learning_rate": 1.4298861411583224e-05, + "loss": 0.1941, + "step": 7218 + }, + { + "epoch": 3.02, + "learning_rate": 1.4297330593769691e-05, + "loss": 0.2126, + "step": 7219 + }, + { + "epoch": 3.02, + "learning_rate": 1.4295799652433206e-05, + "loss": 0.1735, + "step": 7220 + }, + { + "epoch": 3.02, + "learning_rate": 1.429426858761777e-05, + "loss": 0.1709, + "step": 7221 + }, + { + "epoch": 3.02, + "learning_rate": 1.4292737399367394e-05, + "loss": 0.1972, + "step": 7222 + }, + { + "epoch": 3.02, + "learning_rate": 1.429120608772609e-05, + "loss": 0.19, + "step": 7223 + }, + { + "epoch": 3.02, + "learning_rate": 1.4289674652737877e-05, + "loss": 0.1853, + "step": 7224 + }, + { + "epoch": 3.02, + "learning_rate": 1.4288143094446771e-05, + "loss": 0.1501, + "step": 7225 + }, + { + "epoch": 3.03, + "learning_rate": 1.42866114128968e-05, + "loss": 0.1467, + "step": 7226 + }, + { + "epoch": 3.03, + "learning_rate": 1.4285079608131984e-05, + "loss": 0.148, + "step": 7227 + }, + { + "epoch": 3.03, + "learning_rate": 1.4283547680196359e-05, + "loss": 0.2309, + "step": 7228 + }, + { + "epoch": 3.03, + "learning_rate": 1.4282015629133958e-05, + "loss": 0.2142, + "step": 7229 + }, + { + "epoch": 3.03, + "learning_rate": 1.4280483454988816e-05, + "loss": 0.2085, + "step": 7230 + }, + { + "epoch": 3.03, + "learning_rate": 1.4278951157804977e-05, + "loss": 0.1577, + "step": 7231 + }, + { + "epoch": 3.03, + "learning_rate": 1.4277418737626483e-05, + "loss": 0.1826, + "step": 7232 + }, + { + "epoch": 3.03, + "learning_rate": 1.4275886194497384e-05, + "loss": 0.2294, + "step": 7233 + }, + { + "epoch": 3.03, + "learning_rate": 1.4274353528461728e-05, + "loss": 0.1659, + "step": 7234 + }, + { + "epoch": 3.03, + "learning_rate": 1.4272820739563575e-05, + "loss": 0.226, + "step": 7235 + }, + { + "epoch": 3.03, + "learning_rate": 1.427128782784698e-05, + "loss": 0.1884, + "step": 7236 + }, + { + "epoch": 3.03, + "learning_rate": 1.4269754793356008e-05, + "loss": 0.1751, + "step": 7237 + }, + { + "epoch": 3.03, + "learning_rate": 1.4268221636134722e-05, + "loss": 0.2024, + "step": 7238 + }, + { + "epoch": 3.03, + "learning_rate": 1.4266688356227194e-05, + "loss": 0.1908, + "step": 7239 + }, + { + "epoch": 3.03, + "learning_rate": 1.4265154953677498e-05, + "loss": 0.1581, + "step": 7240 + }, + { + "epoch": 3.03, + "learning_rate": 1.4263621428529703e-05, + "loss": 0.1458, + "step": 7241 + }, + { + "epoch": 3.03, + "learning_rate": 1.4262087780827896e-05, + "loss": 0.18, + "step": 7242 + }, + { + "epoch": 3.03, + "learning_rate": 1.4260554010616156e-05, + "loss": 0.1969, + "step": 7243 + }, + { + "epoch": 3.03, + "learning_rate": 1.4259020117938574e-05, + "loss": 0.1862, + "step": 7244 + }, + { + "epoch": 3.03, + "learning_rate": 1.4257486102839236e-05, + "loss": 0.2063, + "step": 7245 + }, + { + "epoch": 3.03, + "learning_rate": 1.4255951965362238e-05, + "loss": 0.1843, + "step": 7246 + }, + { + "epoch": 3.03, + "learning_rate": 1.4254417705551679e-05, + "loss": 0.1448, + "step": 7247 + }, + { + "epoch": 3.03, + "learning_rate": 1.4252883323451658e-05, + "loss": 0.2114, + "step": 7248 + }, + { + "epoch": 3.03, + "learning_rate": 1.425134881910628e-05, + "loss": 0.2203, + "step": 7249 + }, + { + "epoch": 3.04, + "learning_rate": 1.4249814192559653e-05, + "loss": 0.1884, + "step": 7250 + }, + { + "epoch": 3.04, + "learning_rate": 1.4248279443855891e-05, + "loss": 0.1889, + "step": 7251 + }, + { + "epoch": 3.04, + "learning_rate": 1.42467445730391e-05, + "loss": 0.1804, + "step": 7252 + }, + { + "epoch": 3.04, + "learning_rate": 1.4245209580153408e-05, + "loss": 0.174, + "step": 7253 + }, + { + "epoch": 3.04, + "learning_rate": 1.4243674465242934e-05, + "loss": 0.1691, + "step": 7254 + }, + { + "epoch": 3.04, + "learning_rate": 1.42421392283518e-05, + "loss": 0.1507, + "step": 7255 + }, + { + "epoch": 3.04, + "learning_rate": 1.424060386952414e-05, + "loss": 0.1765, + "step": 7256 + }, + { + "epoch": 3.04, + "learning_rate": 1.4239068388804089e-05, + "loss": 0.1813, + "step": 7257 + }, + { + "epoch": 3.04, + "learning_rate": 1.4237532786235775e-05, + "loss": 0.1951, + "step": 7258 + }, + { + "epoch": 3.04, + "learning_rate": 1.423599706186334e-05, + "loss": 0.2148, + "step": 7259 + }, + { + "epoch": 3.04, + "learning_rate": 1.4234461215730932e-05, + "loss": 0.1641, + "step": 7260 + }, + { + "epoch": 3.04, + "learning_rate": 1.4232925247882692e-05, + "loss": 0.1737, + "step": 7261 + }, + { + "epoch": 3.04, + "learning_rate": 1.4231389158362771e-05, + "loss": 0.1844, + "step": 7262 + }, + { + "epoch": 3.04, + "learning_rate": 1.4229852947215326e-05, + "loss": 0.179, + "step": 7263 + }, + { + "epoch": 3.04, + "learning_rate": 1.4228316614484511e-05, + "loss": 0.1891, + "step": 7264 + }, + { + "epoch": 3.04, + "learning_rate": 1.4226780160214486e-05, + "loss": 0.166, + "step": 7265 + }, + { + "epoch": 3.04, + "learning_rate": 1.4225243584449415e-05, + "loss": 0.1638, + "step": 7266 + }, + { + "epoch": 3.04, + "learning_rate": 1.422370688723347e-05, + "loss": 0.2241, + "step": 7267 + }, + { + "epoch": 3.04, + "learning_rate": 1.4222170068610816e-05, + "loss": 0.2143, + "step": 7268 + }, + { + "epoch": 3.04, + "learning_rate": 1.4220633128625632e-05, + "loss": 0.2018, + "step": 7269 + }, + { + "epoch": 3.04, + "learning_rate": 1.4219096067322094e-05, + "loss": 0.1457, + "step": 7270 + }, + { + "epoch": 3.04, + "learning_rate": 1.421755888474438e-05, + "loss": 0.1974, + "step": 7271 + }, + { + "epoch": 3.04, + "learning_rate": 1.4216021580936681e-05, + "loss": 0.1919, + "step": 7272 + }, + { + "epoch": 3.05, + "learning_rate": 1.421448415594318e-05, + "loss": 0.2016, + "step": 7273 + }, + { + "epoch": 3.05, + "learning_rate": 1.4212946609808079e-05, + "loss": 0.1676, + "step": 7274 + }, + { + "epoch": 3.05, + "learning_rate": 1.421140894257556e-05, + "loss": 0.2124, + "step": 7275 + }, + { + "epoch": 3.05, + "learning_rate": 1.420987115428983e-05, + "loss": 0.2214, + "step": 7276 + }, + { + "epoch": 3.05, + "learning_rate": 1.4208333244995087e-05, + "loss": 0.1987, + "step": 7277 + }, + { + "epoch": 3.05, + "learning_rate": 1.4206795214735543e-05, + "loss": 0.1791, + "step": 7278 + }, + { + "epoch": 3.05, + "learning_rate": 1.42052570635554e-05, + "loss": 0.2003, + "step": 7279 + }, + { + "epoch": 3.05, + "learning_rate": 1.4203718791498877e-05, + "loss": 0.1513, + "step": 7280 + }, + { + "epoch": 3.05, + "learning_rate": 1.4202180398610188e-05, + "loss": 0.1887, + "step": 7281 + }, + { + "epoch": 3.05, + "learning_rate": 1.420064188493355e-05, + "loss": 0.2105, + "step": 7282 + }, + { + "epoch": 3.05, + "learning_rate": 1.4199103250513192e-05, + "loss": 0.1794, + "step": 7283 + }, + { + "epoch": 3.05, + "learning_rate": 1.4197564495393334e-05, + "loss": 0.1558, + "step": 7284 + }, + { + "epoch": 3.05, + "learning_rate": 1.4196025619618212e-05, + "loss": 0.1765, + "step": 7285 + }, + { + "epoch": 3.05, + "learning_rate": 1.4194486623232056e-05, + "loss": 0.1902, + "step": 7286 + }, + { + "epoch": 3.05, + "learning_rate": 1.4192947506279102e-05, + "loss": 0.1742, + "step": 7287 + }, + { + "epoch": 3.05, + "learning_rate": 1.4191408268803596e-05, + "loss": 0.1754, + "step": 7288 + }, + { + "epoch": 3.05, + "learning_rate": 1.4189868910849779e-05, + "loss": 0.1759, + "step": 7289 + }, + { + "epoch": 3.05, + "learning_rate": 1.4188329432461894e-05, + "loss": 0.1869, + "step": 7290 + }, + { + "epoch": 3.05, + "learning_rate": 1.41867898336842e-05, + "loss": 0.1341, + "step": 7291 + }, + { + "epoch": 3.05, + "learning_rate": 1.4185250114560945e-05, + "loss": 0.1637, + "step": 7292 + }, + { + "epoch": 3.05, + "learning_rate": 1.4183710275136391e-05, + "loss": 0.1879, + "step": 7293 + }, + { + "epoch": 3.05, + "learning_rate": 1.4182170315454798e-05, + "loss": 0.1744, + "step": 7294 + }, + { + "epoch": 3.05, + "learning_rate": 1.4180630235560428e-05, + "loss": 0.1949, + "step": 7295 + }, + { + "epoch": 3.05, + "learning_rate": 1.4179090035497554e-05, + "loss": 0.1792, + "step": 7296 + }, + { + "epoch": 3.06, + "learning_rate": 1.4177549715310445e-05, + "loss": 0.169, + "step": 7297 + }, + { + "epoch": 3.06, + "learning_rate": 1.4176009275043376e-05, + "loss": 0.1775, + "step": 7298 + }, + { + "epoch": 3.06, + "learning_rate": 1.4174468714740628e-05, + "loss": 0.1806, + "step": 7299 + }, + { + "epoch": 3.06, + "learning_rate": 1.4172928034446478e-05, + "loss": 0.217, + "step": 7300 + }, + { + "epoch": 3.06, + "learning_rate": 1.4171387234205219e-05, + "loss": 0.1439, + "step": 7301 + }, + { + "epoch": 3.06, + "learning_rate": 1.4169846314061132e-05, + "loss": 0.1562, + "step": 7302 + }, + { + "epoch": 3.06, + "learning_rate": 1.4168305274058511e-05, + "loss": 0.1626, + "step": 7303 + }, + { + "epoch": 3.06, + "learning_rate": 1.4166764114241657e-05, + "loss": 0.1897, + "step": 7304 + }, + { + "epoch": 3.06, + "learning_rate": 1.4165222834654867e-05, + "loss": 0.2121, + "step": 7305 + }, + { + "epoch": 3.06, + "learning_rate": 1.416368143534244e-05, + "loss": 0.1895, + "step": 7306 + }, + { + "epoch": 3.06, + "learning_rate": 1.4162139916348685e-05, + "loss": 0.2021, + "step": 7307 + }, + { + "epoch": 3.06, + "learning_rate": 1.4160598277717916e-05, + "loss": 0.1783, + "step": 7308 + }, + { + "epoch": 3.06, + "learning_rate": 1.4159056519494439e-05, + "loss": 0.2108, + "step": 7309 + }, + { + "epoch": 3.06, + "learning_rate": 1.4157514641722569e-05, + "loss": 0.2187, + "step": 7310 + }, + { + "epoch": 3.06, + "learning_rate": 1.4155972644446635e-05, + "loss": 0.1467, + "step": 7311 + }, + { + "epoch": 3.06, + "learning_rate": 1.4154430527710954e-05, + "loss": 0.1728, + "step": 7312 + }, + { + "epoch": 3.06, + "learning_rate": 1.4152888291559852e-05, + "loss": 0.1972, + "step": 7313 + }, + { + "epoch": 3.06, + "learning_rate": 1.4151345936037664e-05, + "loss": 0.2267, + "step": 7314 + }, + { + "epoch": 3.06, + "learning_rate": 1.414980346118872e-05, + "loss": 0.1653, + "step": 7315 + }, + { + "epoch": 3.06, + "learning_rate": 1.4148260867057357e-05, + "loss": 0.1404, + "step": 7316 + }, + { + "epoch": 3.06, + "learning_rate": 1.4146718153687917e-05, + "loss": 0.1793, + "step": 7317 + }, + { + "epoch": 3.06, + "learning_rate": 1.4145175321124744e-05, + "loss": 0.1372, + "step": 7318 + }, + { + "epoch": 3.06, + "learning_rate": 1.4143632369412185e-05, + "loss": 0.1705, + "step": 7319 + }, + { + "epoch": 3.06, + "learning_rate": 1.414208929859459e-05, + "loss": 0.2271, + "step": 7320 + }, + { + "epoch": 3.07, + "learning_rate": 1.4140546108716313e-05, + "loss": 0.1695, + "step": 7321 + }, + { + "epoch": 3.07, + "learning_rate": 1.4139002799821714e-05, + "loss": 0.1739, + "step": 7322 + }, + { + "epoch": 3.07, + "learning_rate": 1.4137459371955151e-05, + "loss": 0.2014, + "step": 7323 + }, + { + "epoch": 3.07, + "learning_rate": 1.413591582516099e-05, + "loss": 0.1908, + "step": 7324 + }, + { + "epoch": 3.07, + "learning_rate": 1.41343721594836e-05, + "loss": 0.1706, + "step": 7325 + }, + { + "epoch": 3.07, + "learning_rate": 1.4132828374967348e-05, + "loss": 0.1861, + "step": 7326 + }, + { + "epoch": 3.07, + "learning_rate": 1.4131284471656615e-05, + "loss": 0.1634, + "step": 7327 + }, + { + "epoch": 3.07, + "learning_rate": 1.4129740449595777e-05, + "loss": 0.1741, + "step": 7328 + }, + { + "epoch": 3.07, + "learning_rate": 1.4128196308829209e-05, + "loss": 0.1747, + "step": 7329 + }, + { + "epoch": 3.07, + "learning_rate": 1.4126652049401306e-05, + "loss": 0.1682, + "step": 7330 + }, + { + "epoch": 3.07, + "learning_rate": 1.4125107671356451e-05, + "loss": 0.19, + "step": 7331 + }, + { + "epoch": 3.07, + "learning_rate": 1.4123563174739036e-05, + "loss": 0.1916, + "step": 7332 + }, + { + "epoch": 3.07, + "learning_rate": 1.4122018559593457e-05, + "loss": 0.1645, + "step": 7333 + }, + { + "epoch": 3.07, + "learning_rate": 1.4120473825964112e-05, + "loss": 0.1863, + "step": 7334 + }, + { + "epoch": 3.07, + "learning_rate": 1.4118928973895406e-05, + "loss": 0.1731, + "step": 7335 + }, + { + "epoch": 3.07, + "learning_rate": 1.4117384003431738e-05, + "loss": 0.1718, + "step": 7336 + }, + { + "epoch": 3.07, + "learning_rate": 1.4115838914617518e-05, + "loss": 0.2038, + "step": 7337 + }, + { + "epoch": 3.07, + "learning_rate": 1.4114293707497168e-05, + "loss": 0.1613, + "step": 7338 + }, + { + "epoch": 3.07, + "learning_rate": 1.4112748382115091e-05, + "loss": 0.2105, + "step": 7339 + }, + { + "epoch": 3.07, + "learning_rate": 1.4111202938515711e-05, + "loss": 0.1805, + "step": 7340 + }, + { + "epoch": 3.07, + "learning_rate": 1.4109657376743455e-05, + "loss": 0.157, + "step": 7341 + }, + { + "epoch": 3.07, + "learning_rate": 1.4108111696842743e-05, + "loss": 0.2109, + "step": 7342 + }, + { + "epoch": 3.07, + "learning_rate": 1.4106565898858007e-05, + "loss": 0.1733, + "step": 7343 + }, + { + "epoch": 3.07, + "learning_rate": 1.4105019982833674e-05, + "loss": 0.1645, + "step": 7344 + }, + { + "epoch": 3.08, + "learning_rate": 1.4103473948814189e-05, + "loss": 0.1983, + "step": 7345 + }, + { + "epoch": 3.08, + "learning_rate": 1.4101927796843984e-05, + "loss": 0.1839, + "step": 7346 + }, + { + "epoch": 3.08, + "learning_rate": 1.4100381526967502e-05, + "loss": 0.2312, + "step": 7347 + }, + { + "epoch": 3.08, + "learning_rate": 1.4098835139229195e-05, + "loss": 0.1856, + "step": 7348 + }, + { + "epoch": 3.08, + "learning_rate": 1.409728863367351e-05, + "loss": 0.1912, + "step": 7349 + }, + { + "epoch": 3.08, + "learning_rate": 1.4095742010344898e-05, + "loss": 0.1981, + "step": 7350 + }, + { + "epoch": 3.08, + "learning_rate": 1.4094195269287814e-05, + "loss": 0.197, + "step": 7351 + }, + { + "epoch": 3.08, + "learning_rate": 1.4092648410546723e-05, + "loss": 0.1855, + "step": 7352 + }, + { + "epoch": 3.08, + "learning_rate": 1.4091101434166086e-05, + "loss": 0.1751, + "step": 7353 + }, + { + "epoch": 3.08, + "learning_rate": 1.4089554340190365e-05, + "loss": 0.1941, + "step": 7354 + }, + { + "epoch": 3.08, + "learning_rate": 1.4088007128664036e-05, + "loss": 0.1755, + "step": 7355 + }, + { + "epoch": 3.08, + "learning_rate": 1.4086459799631572e-05, + "loss": 0.1973, + "step": 7356 + }, + { + "epoch": 3.08, + "learning_rate": 1.4084912353137444e-05, + "loss": 0.1854, + "step": 7357 + }, + { + "epoch": 3.08, + "learning_rate": 1.4083364789226135e-05, + "loss": 0.2054, + "step": 7358 + }, + { + "epoch": 3.08, + "learning_rate": 1.4081817107942131e-05, + "loss": 0.1948, + "step": 7359 + }, + { + "epoch": 3.08, + "learning_rate": 1.4080269309329912e-05, + "loss": 0.157, + "step": 7360 + }, + { + "epoch": 3.08, + "learning_rate": 1.4078721393433977e-05, + "loss": 0.2108, + "step": 7361 + }, + { + "epoch": 3.08, + "learning_rate": 1.4077173360298812e-05, + "loss": 0.1796, + "step": 7362 + }, + { + "epoch": 3.08, + "learning_rate": 1.407562520996892e-05, + "loss": 0.2147, + "step": 7363 + }, + { + "epoch": 3.08, + "learning_rate": 1.4074076942488794e-05, + "loss": 0.2133, + "step": 7364 + }, + { + "epoch": 3.08, + "learning_rate": 1.4072528557902944e-05, + "loss": 0.1498, + "step": 7365 + }, + { + "epoch": 3.08, + "learning_rate": 1.4070980056255876e-05, + "loss": 0.2077, + "step": 7366 + }, + { + "epoch": 3.08, + "learning_rate": 1.4069431437592095e-05, + "loss": 0.2182, + "step": 7367 + }, + { + "epoch": 3.08, + "learning_rate": 1.4067882701956121e-05, + "loss": 0.2054, + "step": 7368 + }, + { + "epoch": 3.09, + "learning_rate": 1.406633384939247e-05, + "loss": 0.1672, + "step": 7369 + }, + { + "epoch": 3.09, + "learning_rate": 1.4064784879945658e-05, + "loss": 0.1746, + "step": 7370 + }, + { + "epoch": 3.09, + "learning_rate": 1.4063235793660212e-05, + "loss": 0.1901, + "step": 7371 + }, + { + "epoch": 3.09, + "learning_rate": 1.406168659058066e-05, + "loss": 0.1435, + "step": 7372 + }, + { + "epoch": 3.09, + "learning_rate": 1.4060137270751531e-05, + "loss": 0.2062, + "step": 7373 + }, + { + "epoch": 3.09, + "learning_rate": 1.4058587834217356e-05, + "loss": 0.1671, + "step": 7374 + }, + { + "epoch": 3.09, + "learning_rate": 1.405703828102268e-05, + "loss": 0.1781, + "step": 7375 + }, + { + "epoch": 3.09, + "learning_rate": 1.4055488611212038e-05, + "loss": 0.1978, + "step": 7376 + }, + { + "epoch": 3.09, + "learning_rate": 1.405393882482997e-05, + "loss": 0.1653, + "step": 7377 + }, + { + "epoch": 3.09, + "learning_rate": 1.4052388921921033e-05, + "loss": 0.1671, + "step": 7378 + }, + { + "epoch": 3.09, + "learning_rate": 1.405083890252977e-05, + "loss": 0.1926, + "step": 7379 + }, + { + "epoch": 3.09, + "learning_rate": 1.4049288766700739e-05, + "loss": 0.1676, + "step": 7380 + }, + { + "epoch": 3.09, + "learning_rate": 1.4047738514478493e-05, + "loss": 0.1606, + "step": 7381 + }, + { + "epoch": 3.09, + "learning_rate": 1.40461881459076e-05, + "loss": 0.1998, + "step": 7382 + }, + { + "epoch": 3.09, + "learning_rate": 1.4044637661032614e-05, + "loss": 0.2041, + "step": 7383 + }, + { + "epoch": 3.09, + "learning_rate": 1.404308705989811e-05, + "loss": 0.1909, + "step": 7384 + }, + { + "epoch": 3.09, + "learning_rate": 1.4041536342548656e-05, + "loss": 0.1531, + "step": 7385 + }, + { + "epoch": 3.09, + "learning_rate": 1.4039985509028825e-05, + "loss": 0.1778, + "step": 7386 + }, + { + "epoch": 3.09, + "learning_rate": 1.4038434559383198e-05, + "loss": 0.1942, + "step": 7387 + }, + { + "epoch": 3.09, + "learning_rate": 1.4036883493656353e-05, + "loss": 0.2008, + "step": 7388 + }, + { + "epoch": 3.09, + "learning_rate": 1.4035332311892873e-05, + "loss": 0.1685, + "step": 7389 + }, + { + "epoch": 3.09, + "learning_rate": 1.4033781014137348e-05, + "loss": 0.1667, + "step": 7390 + }, + { + "epoch": 3.09, + "learning_rate": 1.4032229600434363e-05, + "loss": 0.1642, + "step": 7391 + }, + { + "epoch": 3.09, + "learning_rate": 1.403067807082852e-05, + "loss": 0.169, + "step": 7392 + }, + { + "epoch": 3.1, + "learning_rate": 1.4029126425364414e-05, + "loss": 0.183, + "step": 7393 + }, + { + "epoch": 3.1, + "learning_rate": 1.4027574664086641e-05, + "loss": 0.1868, + "step": 7394 + }, + { + "epoch": 3.1, + "learning_rate": 1.402602278703981e-05, + "loss": 0.2027, + "step": 7395 + }, + { + "epoch": 3.1, + "learning_rate": 1.4024470794268528e-05, + "loss": 0.2004, + "step": 7396 + }, + { + "epoch": 3.1, + "learning_rate": 1.4022918685817399e-05, + "loss": 0.1987, + "step": 7397 + }, + { + "epoch": 3.1, + "learning_rate": 1.4021366461731047e-05, + "loss": 0.2168, + "step": 7398 + }, + { + "epoch": 3.1, + "learning_rate": 1.4019814122054082e-05, + "loss": 0.1792, + "step": 7399 + }, + { + "epoch": 3.1, + "learning_rate": 1.4018261666831132e-05, + "loss": 0.1826, + "step": 7400 + }, + { + "epoch": 3.1, + "learning_rate": 1.4016709096106813e-05, + "loss": 0.1979, + "step": 7401 + }, + { + "epoch": 3.1, + "learning_rate": 1.4015156409925754e-05, + "loss": 0.1863, + "step": 7402 + }, + { + "epoch": 3.1, + "learning_rate": 1.4013603608332589e-05, + "loss": 0.1953, + "step": 7403 + }, + { + "epoch": 3.1, + "learning_rate": 1.4012050691371948e-05, + "loss": 0.1579, + "step": 7404 + }, + { + "epoch": 3.1, + "learning_rate": 1.4010497659088468e-05, + "loss": 0.2044, + "step": 7405 + }, + { + "epoch": 3.1, + "learning_rate": 1.4008944511526798e-05, + "loss": 0.1969, + "step": 7406 + }, + { + "epoch": 3.1, + "learning_rate": 1.400739124873157e-05, + "loss": 0.1812, + "step": 7407 + }, + { + "epoch": 3.1, + "learning_rate": 1.4005837870747439e-05, + "loss": 0.2105, + "step": 7408 + }, + { + "epoch": 3.1, + "learning_rate": 1.4004284377619053e-05, + "loss": 0.2011, + "step": 7409 + }, + { + "epoch": 3.1, + "learning_rate": 1.4002730769391069e-05, + "loss": 0.1904, + "step": 7410 + }, + { + "epoch": 3.1, + "learning_rate": 1.4001177046108136e-05, + "loss": 0.1919, + "step": 7411 + }, + { + "epoch": 3.1, + "learning_rate": 1.3999623207814921e-05, + "loss": 0.1803, + "step": 7412 + }, + { + "epoch": 3.1, + "learning_rate": 1.3998069254556088e-05, + "loss": 0.1908, + "step": 7413 + }, + { + "epoch": 3.1, + "learning_rate": 1.3996515186376298e-05, + "loss": 0.22, + "step": 7414 + }, + { + "epoch": 3.1, + "learning_rate": 1.399496100332023e-05, + "loss": 0.174, + "step": 7415 + }, + { + "epoch": 3.1, + "learning_rate": 1.399340670543255e-05, + "loss": 0.2192, + "step": 7416 + }, + { + "epoch": 3.11, + "learning_rate": 1.3991852292757937e-05, + "loss": 0.1978, + "step": 7417 + }, + { + "epoch": 3.11, + "learning_rate": 1.3990297765341074e-05, + "loss": 0.1474, + "step": 7418 + }, + { + "epoch": 3.11, + "learning_rate": 1.3988743123226643e-05, + "loss": 0.1784, + "step": 7419 + }, + { + "epoch": 3.11, + "learning_rate": 1.398718836645933e-05, + "loss": 0.177, + "step": 7420 + }, + { + "epoch": 3.11, + "learning_rate": 1.3985633495083828e-05, + "loss": 0.2083, + "step": 7421 + }, + { + "epoch": 3.11, + "learning_rate": 1.3984078509144824e-05, + "loss": 0.2083, + "step": 7422 + }, + { + "epoch": 3.11, + "learning_rate": 1.398252340868702e-05, + "loss": 0.1653, + "step": 7423 + }, + { + "epoch": 3.11, + "learning_rate": 1.3980968193755115e-05, + "loss": 0.1788, + "step": 7424 + }, + { + "epoch": 3.11, + "learning_rate": 1.397941286439381e-05, + "loss": 0.17, + "step": 7425 + }, + { + "epoch": 3.11, + "learning_rate": 1.3977857420647816e-05, + "loss": 0.1635, + "step": 7426 + }, + { + "epoch": 3.11, + "learning_rate": 1.3976301862561841e-05, + "loss": 0.1704, + "step": 7427 + }, + { + "epoch": 3.11, + "learning_rate": 1.3974746190180595e-05, + "loss": 0.1724, + "step": 7428 + }, + { + "epoch": 3.11, + "learning_rate": 1.3973190403548798e-05, + "loss": 0.1812, + "step": 7429 + }, + { + "epoch": 3.11, + "learning_rate": 1.3971634502711169e-05, + "loss": 0.1525, + "step": 7430 + }, + { + "epoch": 3.11, + "learning_rate": 1.397007848771243e-05, + "loss": 0.1913, + "step": 7431 + }, + { + "epoch": 3.11, + "learning_rate": 1.3968522358597307e-05, + "loss": 0.1774, + "step": 7432 + }, + { + "epoch": 3.11, + "learning_rate": 1.3966966115410533e-05, + "loss": 0.1482, + "step": 7433 + }, + { + "epoch": 3.11, + "learning_rate": 1.3965409758196838e-05, + "loss": 0.2149, + "step": 7434 + }, + { + "epoch": 3.11, + "learning_rate": 1.3963853287000956e-05, + "loss": 0.1504, + "step": 7435 + }, + { + "epoch": 3.11, + "learning_rate": 1.396229670186763e-05, + "loss": 0.2362, + "step": 7436 + }, + { + "epoch": 3.11, + "learning_rate": 1.3960740002841602e-05, + "loss": 0.1958, + "step": 7437 + }, + { + "epoch": 3.11, + "learning_rate": 1.3959183189967615e-05, + "loss": 0.1642, + "step": 7438 + }, + { + "epoch": 3.11, + "learning_rate": 1.395762626329042e-05, + "loss": 0.1807, + "step": 7439 + }, + { + "epoch": 3.11, + "learning_rate": 1.3956069222854774e-05, + "loss": 0.1854, + "step": 7440 + }, + { + "epoch": 3.12, + "learning_rate": 1.3954512068705425e-05, + "loss": 0.2109, + "step": 7441 + }, + { + "epoch": 3.12, + "learning_rate": 1.3952954800887137e-05, + "loss": 0.1881, + "step": 7442 + }, + { + "epoch": 3.12, + "learning_rate": 1.3951397419444675e-05, + "loss": 0.1886, + "step": 7443 + }, + { + "epoch": 3.12, + "learning_rate": 1.3949839924422798e-05, + "loss": 0.1795, + "step": 7444 + }, + { + "epoch": 3.12, + "learning_rate": 1.3948282315866274e-05, + "loss": 0.2422, + "step": 7445 + }, + { + "epoch": 3.12, + "learning_rate": 1.394672459381988e-05, + "loss": 0.2021, + "step": 7446 + }, + { + "epoch": 3.12, + "learning_rate": 1.3945166758328394e-05, + "loss": 0.1772, + "step": 7447 + }, + { + "epoch": 3.12, + "learning_rate": 1.3943608809436587e-05, + "loss": 0.2052, + "step": 7448 + }, + { + "epoch": 3.12, + "learning_rate": 1.3942050747189242e-05, + "loss": 0.1767, + "step": 7449 + }, + { + "epoch": 3.12, + "learning_rate": 1.394049257163115e-05, + "loss": 0.1822, + "step": 7450 + }, + { + "epoch": 3.12, + "learning_rate": 1.3938934282807095e-05, + "loss": 0.1777, + "step": 7451 + }, + { + "epoch": 3.12, + "learning_rate": 1.3937375880761868e-05, + "loss": 0.1736, + "step": 7452 + }, + { + "epoch": 3.12, + "learning_rate": 1.3935817365540268e-05, + "loss": 0.1833, + "step": 7453 + }, + { + "epoch": 3.12, + "learning_rate": 1.3934258737187087e-05, + "loss": 0.2147, + "step": 7454 + }, + { + "epoch": 3.12, + "learning_rate": 1.3932699995747134e-05, + "loss": 0.1847, + "step": 7455 + }, + { + "epoch": 3.12, + "learning_rate": 1.3931141141265206e-05, + "loss": 0.1931, + "step": 7456 + }, + { + "epoch": 3.12, + "learning_rate": 1.3929582173786116e-05, + "loss": 0.15, + "step": 7457 + }, + { + "epoch": 3.12, + "learning_rate": 1.3928023093354671e-05, + "loss": 0.2478, + "step": 7458 + }, + { + "epoch": 3.12, + "learning_rate": 1.392646390001569e-05, + "loss": 0.158, + "step": 7459 + }, + { + "epoch": 3.12, + "learning_rate": 1.3924904593813986e-05, + "loss": 0.1903, + "step": 7460 + }, + { + "epoch": 3.12, + "learning_rate": 1.3923345174794386e-05, + "loss": 0.2119, + "step": 7461 + }, + { + "epoch": 3.12, + "learning_rate": 1.3921785643001708e-05, + "loss": 0.165, + "step": 7462 + }, + { + "epoch": 3.12, + "learning_rate": 1.3920225998480782e-05, + "loss": 0.1985, + "step": 7463 + }, + { + "epoch": 3.12, + "learning_rate": 1.3918666241276442e-05, + "loss": 0.1734, + "step": 7464 + }, + { + "epoch": 3.13, + "learning_rate": 1.3917106371433513e-05, + "loss": 0.202, + "step": 7465 + }, + { + "epoch": 3.13, + "learning_rate": 1.3915546388996838e-05, + "loss": 0.1719, + "step": 7466 + }, + { + "epoch": 3.13, + "learning_rate": 1.3913986294011261e-05, + "loss": 0.1945, + "step": 7467 + }, + { + "epoch": 3.13, + "learning_rate": 1.391242608652162e-05, + "loss": 0.1596, + "step": 7468 + }, + { + "epoch": 3.13, + "learning_rate": 1.391086576657276e-05, + "loss": 0.2132, + "step": 7469 + }, + { + "epoch": 3.13, + "learning_rate": 1.3909305334209536e-05, + "loss": 0.1928, + "step": 7470 + }, + { + "epoch": 3.13, + "learning_rate": 1.3907744789476799e-05, + "loss": 0.1837, + "step": 7471 + }, + { + "epoch": 3.13, + "learning_rate": 1.3906184132419405e-05, + "loss": 0.1827, + "step": 7472 + }, + { + "epoch": 3.13, + "learning_rate": 1.3904623363082213e-05, + "loss": 0.192, + "step": 7473 + }, + { + "epoch": 3.13, + "learning_rate": 1.390306248151009e-05, + "loss": 0.1669, + "step": 7474 + }, + { + "epoch": 3.13, + "learning_rate": 1.3901501487747897e-05, + "loss": 0.163, + "step": 7475 + }, + { + "epoch": 3.13, + "learning_rate": 1.389994038184051e-05, + "loss": 0.1953, + "step": 7476 + }, + { + "epoch": 3.13, + "learning_rate": 1.3898379163832794e-05, + "loss": 0.1866, + "step": 7477 + }, + { + "epoch": 3.13, + "learning_rate": 1.3896817833769631e-05, + "loss": 0.1751, + "step": 7478 + }, + { + "epoch": 3.13, + "learning_rate": 1.3895256391695896e-05, + "loss": 0.1909, + "step": 7479 + }, + { + "epoch": 3.13, + "learning_rate": 1.3893694837656471e-05, + "loss": 0.2215, + "step": 7480 + }, + { + "epoch": 3.13, + "learning_rate": 1.3892133171696246e-05, + "loss": 0.1652, + "step": 7481 + }, + { + "epoch": 3.13, + "learning_rate": 1.3890571393860101e-05, + "loss": 0.1398, + "step": 7482 + }, + { + "epoch": 3.13, + "learning_rate": 1.3889009504192938e-05, + "loss": 0.1765, + "step": 7483 + }, + { + "epoch": 3.13, + "learning_rate": 1.3887447502739649e-05, + "loss": 0.1829, + "step": 7484 + }, + { + "epoch": 3.13, + "learning_rate": 1.388588538954513e-05, + "loss": 0.2005, + "step": 7485 + }, + { + "epoch": 3.13, + "learning_rate": 1.3884323164654284e-05, + "loss": 0.1762, + "step": 7486 + }, + { + "epoch": 3.13, + "learning_rate": 1.3882760828112014e-05, + "loss": 0.2078, + "step": 7487 + }, + { + "epoch": 3.14, + "learning_rate": 1.388119837996323e-05, + "loss": 0.1795, + "step": 7488 + }, + { + "epoch": 3.14, + "learning_rate": 1.3879635820252846e-05, + "loss": 0.1307, + "step": 7489 + }, + { + "epoch": 3.14, + "learning_rate": 1.3878073149025768e-05, + "loss": 0.176, + "step": 7490 + }, + { + "epoch": 3.14, + "learning_rate": 1.3876510366326922e-05, + "loss": 0.207, + "step": 7491 + }, + { + "epoch": 3.14, + "learning_rate": 1.3874947472201226e-05, + "loss": 0.1517, + "step": 7492 + }, + { + "epoch": 3.14, + "learning_rate": 1.38733844666936e-05, + "loss": 0.1875, + "step": 7493 + }, + { + "epoch": 3.14, + "learning_rate": 1.3871821349848977e-05, + "loss": 0.1754, + "step": 7494 + }, + { + "epoch": 3.14, + "learning_rate": 1.3870258121712285e-05, + "loss": 0.1777, + "step": 7495 + }, + { + "epoch": 3.14, + "learning_rate": 1.3868694782328456e-05, + "loss": 0.2025, + "step": 7496 + }, + { + "epoch": 3.14, + "learning_rate": 1.3867131331742429e-05, + "loss": 0.1829, + "step": 7497 + }, + { + "epoch": 3.14, + "learning_rate": 1.3865567769999146e-05, + "loss": 0.176, + "step": 7498 + }, + { + "epoch": 3.14, + "learning_rate": 1.3864004097143546e-05, + "loss": 0.1477, + "step": 7499 + }, + { + "epoch": 3.14, + "learning_rate": 1.3862440313220575e-05, + "loss": 0.1657, + "step": 7500 + }, + { + "epoch": 3.14, + "learning_rate": 1.3860876418275189e-05, + "loss": 0.1488, + "step": 7501 + }, + { + "epoch": 3.14, + "learning_rate": 1.3859312412352335e-05, + "loss": 0.193, + "step": 7502 + }, + { + "epoch": 3.14, + "learning_rate": 1.3857748295496972e-05, + "loss": 0.1975, + "step": 7503 + }, + { + "epoch": 3.14, + "learning_rate": 1.3856184067754053e-05, + "loss": 0.1794, + "step": 7504 + }, + { + "epoch": 3.14, + "learning_rate": 1.3854619729168548e-05, + "loss": 0.161, + "step": 7505 + }, + { + "epoch": 3.14, + "learning_rate": 1.3853055279785416e-05, + "loss": 0.1941, + "step": 7506 + }, + { + "epoch": 3.14, + "learning_rate": 1.3851490719649631e-05, + "loss": 0.2179, + "step": 7507 + }, + { + "epoch": 3.14, + "learning_rate": 1.3849926048806166e-05, + "loss": 0.1785, + "step": 7508 + }, + { + "epoch": 3.14, + "learning_rate": 1.384836126729999e-05, + "loss": 0.2004, + "step": 7509 + }, + { + "epoch": 3.14, + "learning_rate": 1.3846796375176083e-05, + "loss": 0.2078, + "step": 7510 + }, + { + "epoch": 3.14, + "learning_rate": 1.3845231372479432e-05, + "loss": 0.1929, + "step": 7511 + }, + { + "epoch": 3.15, + "learning_rate": 1.3843666259255018e-05, + "loss": 0.1956, + "step": 7512 + }, + { + "epoch": 3.15, + "learning_rate": 1.3842101035547827e-05, + "loss": 0.2089, + "step": 7513 + }, + { + "epoch": 3.15, + "learning_rate": 1.3840535701402847e-05, + "loss": 0.184, + "step": 7514 + }, + { + "epoch": 3.15, + "learning_rate": 1.3838970256865078e-05, + "loss": 0.1952, + "step": 7515 + }, + { + "epoch": 3.15, + "learning_rate": 1.3837404701979517e-05, + "loss": 0.1959, + "step": 7516 + }, + { + "epoch": 3.15, + "learning_rate": 1.3835839036791163e-05, + "loss": 0.2267, + "step": 7517 + }, + { + "epoch": 3.15, + "learning_rate": 1.383427326134502e-05, + "loss": 0.2016, + "step": 7518 + }, + { + "epoch": 3.15, + "learning_rate": 1.3832707375686095e-05, + "loss": 0.1755, + "step": 7519 + }, + { + "epoch": 3.15, + "learning_rate": 1.3831141379859394e-05, + "loss": 0.2223, + "step": 7520 + }, + { + "epoch": 3.15, + "learning_rate": 1.3829575273909939e-05, + "loss": 0.1427, + "step": 7521 + }, + { + "epoch": 3.15, + "learning_rate": 1.3828009057882737e-05, + "loss": 0.1875, + "step": 7522 + }, + { + "epoch": 3.15, + "learning_rate": 1.3826442731822814e-05, + "loss": 0.1887, + "step": 7523 + }, + { + "epoch": 3.15, + "learning_rate": 1.3824876295775187e-05, + "loss": 0.1875, + "step": 7524 + }, + { + "epoch": 3.15, + "learning_rate": 1.3823309749784888e-05, + "loss": 0.1904, + "step": 7525 + }, + { + "epoch": 3.15, + "learning_rate": 1.3821743093896942e-05, + "loss": 0.1796, + "step": 7526 + }, + { + "epoch": 3.15, + "learning_rate": 1.3820176328156381e-05, + "loss": 0.1889, + "step": 7527 + }, + { + "epoch": 3.15, + "learning_rate": 1.3818609452608245e-05, + "loss": 0.1661, + "step": 7528 + }, + { + "epoch": 3.15, + "learning_rate": 1.3817042467297565e-05, + "loss": 0.173, + "step": 7529 + }, + { + "epoch": 3.15, + "learning_rate": 1.3815475372269386e-05, + "loss": 0.1865, + "step": 7530 + }, + { + "epoch": 3.15, + "learning_rate": 1.3813908167568757e-05, + "loss": 0.1942, + "step": 7531 + }, + { + "epoch": 3.15, + "learning_rate": 1.381234085324072e-05, + "loss": 0.2108, + "step": 7532 + }, + { + "epoch": 3.15, + "learning_rate": 1.381077342933033e-05, + "loss": 0.1843, + "step": 7533 + }, + { + "epoch": 3.15, + "learning_rate": 1.3809205895882635e-05, + "loss": 0.2048, + "step": 7534 + }, + { + "epoch": 3.15, + "learning_rate": 1.3807638252942704e-05, + "loss": 0.1397, + "step": 7535 + }, + { + "epoch": 3.16, + "learning_rate": 1.3806070500555583e-05, + "loss": 0.2272, + "step": 7536 + }, + { + "epoch": 3.16, + "learning_rate": 1.3804502638766346e-05, + "loss": 0.1862, + "step": 7537 + }, + { + "epoch": 3.16, + "learning_rate": 1.3802934667620055e-05, + "loss": 0.1913, + "step": 7538 + }, + { + "epoch": 3.16, + "learning_rate": 1.3801366587161786e-05, + "loss": 0.1688, + "step": 7539 + }, + { + "epoch": 3.16, + "learning_rate": 1.3799798397436603e-05, + "loss": 0.2065, + "step": 7540 + }, + { + "epoch": 3.16, + "learning_rate": 1.379823009848959e-05, + "loss": 0.1557, + "step": 7541 + }, + { + "epoch": 3.16, + "learning_rate": 1.3796661690365823e-05, + "loss": 0.1426, + "step": 7542 + }, + { + "epoch": 3.16, + "learning_rate": 1.379509317311038e-05, + "loss": 0.2024, + "step": 7543 + }, + { + "epoch": 3.16, + "learning_rate": 1.3793524546768358e-05, + "loss": 0.2058, + "step": 7544 + }, + { + "epoch": 3.16, + "learning_rate": 1.3791955811384837e-05, + "loss": 0.1913, + "step": 7545 + }, + { + "epoch": 3.16, + "learning_rate": 1.3790386967004911e-05, + "loss": 0.1702, + "step": 7546 + }, + { + "epoch": 3.16, + "learning_rate": 1.3788818013673675e-05, + "loss": 0.1924, + "step": 7547 + }, + { + "epoch": 3.16, + "learning_rate": 1.3787248951436225e-05, + "loss": 0.1676, + "step": 7548 + }, + { + "epoch": 3.16, + "learning_rate": 1.3785679780337668e-05, + "loss": 0.1956, + "step": 7549 + }, + { + "epoch": 3.16, + "learning_rate": 1.3784110500423104e-05, + "loss": 0.2423, + "step": 7550 + }, + { + "epoch": 3.16, + "learning_rate": 1.3782541111737642e-05, + "loss": 0.2112, + "step": 7551 + }, + { + "epoch": 3.16, + "learning_rate": 1.3780971614326393e-05, + "loss": 0.1864, + "step": 7552 + }, + { + "epoch": 3.16, + "learning_rate": 1.3779402008234467e-05, + "loss": 0.2137, + "step": 7553 + }, + { + "epoch": 3.16, + "learning_rate": 1.3777832293506985e-05, + "loss": 0.211, + "step": 7554 + }, + { + "epoch": 3.16, + "learning_rate": 1.377626247018907e-05, + "loss": 0.1677, + "step": 7555 + }, + { + "epoch": 3.16, + "learning_rate": 1.3774692538325835e-05, + "loss": 0.2024, + "step": 7556 + }, + { + "epoch": 3.16, + "learning_rate": 1.3773122497962418e-05, + "loss": 0.1875, + "step": 7557 + }, + { + "epoch": 3.16, + "learning_rate": 1.3771552349143942e-05, + "loss": 0.2106, + "step": 7558 + }, + { + "epoch": 3.16, + "learning_rate": 1.3769982091915539e-05, + "loss": 0.1844, + "step": 7559 + }, + { + "epoch": 3.17, + "learning_rate": 1.3768411726322347e-05, + "loss": 0.241, + "step": 7560 + }, + { + "epoch": 3.17, + "learning_rate": 1.3766841252409502e-05, + "loss": 0.1707, + "step": 7561 + }, + { + "epoch": 3.17, + "learning_rate": 1.3765270670222153e-05, + "loss": 0.1793, + "step": 7562 + }, + { + "epoch": 3.17, + "learning_rate": 1.3763699979805434e-05, + "loss": 0.1644, + "step": 7563 + }, + { + "epoch": 3.17, + "learning_rate": 1.3762129181204502e-05, + "loss": 0.2117, + "step": 7564 + }, + { + "epoch": 3.17, + "learning_rate": 1.3760558274464506e-05, + "loss": 0.1701, + "step": 7565 + }, + { + "epoch": 3.17, + "learning_rate": 1.3758987259630597e-05, + "loss": 0.185, + "step": 7566 + }, + { + "epoch": 3.17, + "learning_rate": 1.3757416136747935e-05, + "loss": 0.1732, + "step": 7567 + }, + { + "epoch": 3.17, + "learning_rate": 1.375584490586168e-05, + "loss": 0.218, + "step": 7568 + }, + { + "epoch": 3.17, + "learning_rate": 1.3754273567017e-05, + "loss": 0.1982, + "step": 7569 + }, + { + "epoch": 3.17, + "learning_rate": 1.3752702120259055e-05, + "loss": 0.2061, + "step": 7570 + }, + { + "epoch": 3.17, + "learning_rate": 1.3751130565633015e-05, + "loss": 0.1819, + "step": 7571 + }, + { + "epoch": 3.17, + "learning_rate": 1.3749558903184059e-05, + "loss": 0.1826, + "step": 7572 + }, + { + "epoch": 3.17, + "learning_rate": 1.3747987132957361e-05, + "loss": 0.1503, + "step": 7573 + }, + { + "epoch": 3.17, + "learning_rate": 1.3746415254998095e-05, + "loss": 0.1801, + "step": 7574 + }, + { + "epoch": 3.17, + "learning_rate": 1.3744843269351446e-05, + "loss": 0.1277, + "step": 7575 + }, + { + "epoch": 3.17, + "learning_rate": 1.3743271176062605e-05, + "loss": 0.1731, + "step": 7576 + }, + { + "epoch": 3.17, + "learning_rate": 1.3741698975176754e-05, + "loss": 0.2044, + "step": 7577 + }, + { + "epoch": 3.17, + "learning_rate": 1.3740126666739086e-05, + "loss": 0.1722, + "step": 7578 + }, + { + "epoch": 3.17, + "learning_rate": 1.3738554250794795e-05, + "loss": 0.1917, + "step": 7579 + }, + { + "epoch": 3.17, + "learning_rate": 1.3736981727389083e-05, + "loss": 0.2002, + "step": 7580 + }, + { + "epoch": 3.17, + "learning_rate": 1.3735409096567145e-05, + "loss": 0.1809, + "step": 7581 + }, + { + "epoch": 3.17, + "learning_rate": 1.3733836358374186e-05, + "loss": 0.1463, + "step": 7582 + }, + { + "epoch": 3.17, + "learning_rate": 1.3732263512855416e-05, + "loss": 0.1965, + "step": 7583 + }, + { + "epoch": 3.18, + "learning_rate": 1.3730690560056043e-05, + "loss": 0.1524, + "step": 7584 + }, + { + "epoch": 3.18, + "learning_rate": 1.3729117500021279e-05, + "loss": 0.1706, + "step": 7585 + }, + { + "epoch": 3.18, + "learning_rate": 1.3727544332796344e-05, + "loss": 0.1952, + "step": 7586 + }, + { + "epoch": 3.18, + "learning_rate": 1.3725971058426453e-05, + "loss": 0.1621, + "step": 7587 + }, + { + "epoch": 3.18, + "learning_rate": 1.3724397676956831e-05, + "loss": 0.2063, + "step": 7588 + }, + { + "epoch": 3.18, + "learning_rate": 1.3722824188432702e-05, + "loss": 0.2065, + "step": 7589 + }, + { + "epoch": 3.18, + "learning_rate": 1.3721250592899295e-05, + "loss": 0.2085, + "step": 7590 + }, + { + "epoch": 3.18, + "learning_rate": 1.3719676890401845e-05, + "loss": 0.1861, + "step": 7591 + }, + { + "epoch": 3.18, + "learning_rate": 1.371810308098558e-05, + "loss": 0.1507, + "step": 7592 + }, + { + "epoch": 3.18, + "learning_rate": 1.3716529164695744e-05, + "loss": 0.1772, + "step": 7593 + }, + { + "epoch": 3.18, + "learning_rate": 1.3714955141577573e-05, + "loss": 0.2075, + "step": 7594 + }, + { + "epoch": 3.18, + "learning_rate": 1.3713381011676315e-05, + "loss": 0.2258, + "step": 7595 + }, + { + "epoch": 3.18, + "learning_rate": 1.3711806775037215e-05, + "loss": 0.2036, + "step": 7596 + }, + { + "epoch": 3.18, + "learning_rate": 1.3710232431705521e-05, + "loss": 0.1845, + "step": 7597 + }, + { + "epoch": 3.18, + "learning_rate": 1.3708657981726487e-05, + "loss": 0.1761, + "step": 7598 + }, + { + "epoch": 3.18, + "learning_rate": 1.3707083425145375e-05, + "loss": 0.1685, + "step": 7599 + }, + { + "epoch": 3.18, + "learning_rate": 1.3705508762007436e-05, + "loss": 0.1957, + "step": 7600 + }, + { + "epoch": 3.18, + "learning_rate": 1.3703933992357936e-05, + "loss": 0.1601, + "step": 7601 + }, + { + "epoch": 3.18, + "learning_rate": 1.370235911624214e-05, + "loss": 0.1924, + "step": 7602 + }, + { + "epoch": 3.18, + "learning_rate": 1.370078413370532e-05, + "loss": 0.1813, + "step": 7603 + }, + { + "epoch": 3.18, + "learning_rate": 1.3699209044792743e-05, + "loss": 0.2021, + "step": 7604 + }, + { + "epoch": 3.18, + "learning_rate": 1.3697633849549683e-05, + "loss": 0.1347, + "step": 7605 + }, + { + "epoch": 3.18, + "learning_rate": 1.369605854802142e-05, + "loss": 0.1988, + "step": 7606 + }, + { + "epoch": 3.18, + "learning_rate": 1.3694483140253234e-05, + "loss": 0.1635, + "step": 7607 + }, + { + "epoch": 3.19, + "learning_rate": 1.3692907626290406e-05, + "loss": 0.1745, + "step": 7608 + }, + { + "epoch": 3.19, + "learning_rate": 1.3691332006178226e-05, + "loss": 0.1576, + "step": 7609 + }, + { + "epoch": 3.19, + "learning_rate": 1.3689756279961986e-05, + "loss": 0.1684, + "step": 7610 + }, + { + "epoch": 3.19, + "learning_rate": 1.3688180447686975e-05, + "loss": 0.1723, + "step": 7611 + }, + { + "epoch": 3.19, + "learning_rate": 1.3686604509398489e-05, + "loss": 0.1553, + "step": 7612 + }, + { + "epoch": 3.19, + "learning_rate": 1.3685028465141828e-05, + "loss": 0.1632, + "step": 7613 + }, + { + "epoch": 3.19, + "learning_rate": 1.3683452314962295e-05, + "loss": 0.1662, + "step": 7614 + }, + { + "epoch": 3.19, + "learning_rate": 1.3681876058905193e-05, + "loss": 0.2028, + "step": 7615 + }, + { + "epoch": 3.19, + "learning_rate": 1.368029969701583e-05, + "loss": 0.2133, + "step": 7616 + }, + { + "epoch": 3.19, + "learning_rate": 1.3678723229339518e-05, + "loss": 0.1861, + "step": 7617 + }, + { + "epoch": 3.19, + "learning_rate": 1.3677146655921573e-05, + "loss": 0.2252, + "step": 7618 + }, + { + "epoch": 3.19, + "learning_rate": 1.3675569976807307e-05, + "loss": 0.1522, + "step": 7619 + }, + { + "epoch": 3.19, + "learning_rate": 1.3673993192042048e-05, + "loss": 0.2132, + "step": 7620 + }, + { + "epoch": 3.19, + "learning_rate": 1.3672416301671111e-05, + "loss": 0.1779, + "step": 7621 + }, + { + "epoch": 3.19, + "learning_rate": 1.3670839305739828e-05, + "loss": 0.1832, + "step": 7622 + }, + { + "epoch": 3.19, + "learning_rate": 1.366926220429353e-05, + "loss": 0.1864, + "step": 7623 + }, + { + "epoch": 3.19, + "learning_rate": 1.3667684997377542e-05, + "loss": 0.2032, + "step": 7624 + }, + { + "epoch": 3.19, + "learning_rate": 1.3666107685037204e-05, + "loss": 0.1911, + "step": 7625 + }, + { + "epoch": 3.19, + "learning_rate": 1.3664530267317855e-05, + "loss": 0.1678, + "step": 7626 + }, + { + "epoch": 3.19, + "learning_rate": 1.3662952744264835e-05, + "loss": 0.1906, + "step": 7627 + }, + { + "epoch": 3.19, + "learning_rate": 1.3661375115923487e-05, + "loss": 0.2059, + "step": 7628 + }, + { + "epoch": 3.19, + "learning_rate": 1.3659797382339162e-05, + "loss": 0.1447, + "step": 7629 + }, + { + "epoch": 3.19, + "learning_rate": 1.3658219543557208e-05, + "loss": 0.1612, + "step": 7630 + }, + { + "epoch": 3.19, + "learning_rate": 1.365664159962298e-05, + "loss": 0.1815, + "step": 7631 + }, + { + "epoch": 3.2, + "learning_rate": 1.3655063550581834e-05, + "loss": 0.1874, + "step": 7632 + }, + { + "epoch": 3.2, + "learning_rate": 1.3653485396479129e-05, + "loss": 0.1905, + "step": 7633 + }, + { + "epoch": 3.2, + "learning_rate": 1.365190713736023e-05, + "loss": 0.1837, + "step": 7634 + }, + { + "epoch": 3.2, + "learning_rate": 1.3650328773270498e-05, + "loss": 0.2112, + "step": 7635 + }, + { + "epoch": 3.2, + "learning_rate": 1.3648750304255307e-05, + "loss": 0.1932, + "step": 7636 + }, + { + "epoch": 3.2, + "learning_rate": 1.3647171730360026e-05, + "loss": 0.134, + "step": 7637 + }, + { + "epoch": 3.2, + "learning_rate": 1.3645593051630032e-05, + "loss": 0.18, + "step": 7638 + }, + { + "epoch": 3.2, + "learning_rate": 1.3644014268110697e-05, + "loss": 0.1485, + "step": 7639 + }, + { + "epoch": 3.2, + "learning_rate": 1.364243537984741e-05, + "loss": 0.1891, + "step": 7640 + }, + { + "epoch": 3.2, + "learning_rate": 1.3640856386885546e-05, + "loss": 0.202, + "step": 7641 + }, + { + "epoch": 3.2, + "learning_rate": 1.3639277289270497e-05, + "loss": 0.1828, + "step": 7642 + }, + { + "epoch": 3.2, + "learning_rate": 1.3637698087047651e-05, + "loss": 0.149, + "step": 7643 + }, + { + "epoch": 3.2, + "learning_rate": 1.3636118780262405e-05, + "loss": 0.1346, + "step": 7644 + }, + { + "epoch": 3.2, + "learning_rate": 1.3634539368960147e-05, + "loss": 0.1852, + "step": 7645 + }, + { + "epoch": 3.2, + "learning_rate": 1.363295985318628e-05, + "loss": 0.1923, + "step": 7646 + }, + { + "epoch": 3.2, + "learning_rate": 1.363138023298621e-05, + "loss": 0.182, + "step": 7647 + }, + { + "epoch": 3.2, + "learning_rate": 1.3629800508405338e-05, + "loss": 0.1883, + "step": 7648 + }, + { + "epoch": 3.2, + "learning_rate": 1.3628220679489069e-05, + "loss": 0.2061, + "step": 7649 + }, + { + "epoch": 3.2, + "learning_rate": 1.3626640746282815e-05, + "loss": 0.2, + "step": 7650 + }, + { + "epoch": 3.2, + "learning_rate": 1.3625060708831992e-05, + "loss": 0.1932, + "step": 7651 + }, + { + "epoch": 3.2, + "learning_rate": 1.3623480567182018e-05, + "loss": 0.154, + "step": 7652 + }, + { + "epoch": 3.2, + "learning_rate": 1.3621900321378308e-05, + "loss": 0.1829, + "step": 7653 + }, + { + "epoch": 3.2, + "learning_rate": 1.3620319971466288e-05, + "loss": 0.2042, + "step": 7654 + }, + { + "epoch": 3.2, + "learning_rate": 1.3618739517491381e-05, + "loss": 0.2031, + "step": 7655 + }, + { + "epoch": 3.21, + "learning_rate": 1.361715895949902e-05, + "loss": 0.1759, + "step": 7656 + }, + { + "epoch": 3.21, + "learning_rate": 1.3615578297534635e-05, + "loss": 0.2407, + "step": 7657 + }, + { + "epoch": 3.21, + "learning_rate": 1.3613997531643657e-05, + "loss": 0.1647, + "step": 7658 + }, + { + "epoch": 3.21, + "learning_rate": 1.3612416661871532e-05, + "loss": 0.2102, + "step": 7659 + }, + { + "epoch": 3.21, + "learning_rate": 1.361083568826369e-05, + "loss": 0.1727, + "step": 7660 + }, + { + "epoch": 3.21, + "learning_rate": 1.3609254610865587e-05, + "loss": 0.2066, + "step": 7661 + }, + { + "epoch": 3.21, + "learning_rate": 1.3607673429722658e-05, + "loss": 0.1782, + "step": 7662 + }, + { + "epoch": 3.21, + "learning_rate": 1.3606092144880359e-05, + "loss": 0.1969, + "step": 7663 + }, + { + "epoch": 3.21, + "learning_rate": 1.3604510756384141e-05, + "loss": 0.1575, + "step": 7664 + }, + { + "epoch": 3.21, + "learning_rate": 1.360292926427946e-05, + "loss": 0.1964, + "step": 7665 + }, + { + "epoch": 3.21, + "learning_rate": 1.3601347668611772e-05, + "loss": 0.1888, + "step": 7666 + }, + { + "epoch": 3.21, + "learning_rate": 1.3599765969426546e-05, + "loss": 0.1649, + "step": 7667 + }, + { + "epoch": 3.21, + "learning_rate": 1.359818416676924e-05, + "loss": 0.1653, + "step": 7668 + }, + { + "epoch": 3.21, + "learning_rate": 1.3596602260685321e-05, + "loss": 0.1853, + "step": 7669 + }, + { + "epoch": 3.21, + "learning_rate": 1.3595020251220263e-05, + "loss": 0.1915, + "step": 7670 + }, + { + "epoch": 3.21, + "learning_rate": 1.3593438138419539e-05, + "loss": 0.1966, + "step": 7671 + }, + { + "epoch": 3.21, + "learning_rate": 1.3591855922328624e-05, + "loss": 0.2002, + "step": 7672 + }, + { + "epoch": 3.21, + "learning_rate": 1.3590273602992996e-05, + "loss": 0.1996, + "step": 7673 + }, + { + "epoch": 3.21, + "learning_rate": 1.3588691180458143e-05, + "loss": 0.1883, + "step": 7674 + }, + { + "epoch": 3.21, + "learning_rate": 1.3587108654769542e-05, + "loss": 0.1751, + "step": 7675 + }, + { + "epoch": 3.21, + "learning_rate": 1.3585526025972688e-05, + "loss": 0.1636, + "step": 7676 + }, + { + "epoch": 3.21, + "learning_rate": 1.3583943294113071e-05, + "loss": 0.2001, + "step": 7677 + }, + { + "epoch": 3.21, + "learning_rate": 1.3582360459236183e-05, + "loss": 0.1908, + "step": 7678 + }, + { + "epoch": 3.21, + "learning_rate": 1.3580777521387523e-05, + "loss": 0.1748, + "step": 7679 + }, + { + "epoch": 3.22, + "learning_rate": 1.357919448061259e-05, + "loss": 0.1909, + "step": 7680 + }, + { + "epoch": 3.22, + "learning_rate": 1.3577611336956888e-05, + "loss": 0.1817, + "step": 7681 + }, + { + "epoch": 3.22, + "learning_rate": 1.3576028090465924e-05, + "loss": 0.1583, + "step": 7682 + }, + { + "epoch": 3.22, + "learning_rate": 1.3574444741185207e-05, + "loss": 0.1666, + "step": 7683 + }, + { + "epoch": 3.22, + "learning_rate": 1.3572861289160245e-05, + "loss": 0.2047, + "step": 7684 + }, + { + "epoch": 3.22, + "learning_rate": 1.3571277734436556e-05, + "loss": 0.1697, + "step": 7685 + }, + { + "epoch": 3.22, + "learning_rate": 1.356969407705966e-05, + "loss": 0.1998, + "step": 7686 + }, + { + "epoch": 3.22, + "learning_rate": 1.3568110317075071e-05, + "loss": 0.1905, + "step": 7687 + }, + { + "epoch": 3.22, + "learning_rate": 1.356652645452832e-05, + "loss": 0.1819, + "step": 7688 + }, + { + "epoch": 3.22, + "learning_rate": 1.3564942489464932e-05, + "loss": 0.2108, + "step": 7689 + }, + { + "epoch": 3.22, + "learning_rate": 1.3563358421930433e-05, + "loss": 0.2004, + "step": 7690 + }, + { + "epoch": 3.22, + "learning_rate": 1.356177425197036e-05, + "loss": 0.178, + "step": 7691 + }, + { + "epoch": 3.22, + "learning_rate": 1.3560189979630247e-05, + "loss": 0.2014, + "step": 7692 + }, + { + "epoch": 3.22, + "learning_rate": 1.3558605604955633e-05, + "loss": 0.1777, + "step": 7693 + }, + { + "epoch": 3.22, + "learning_rate": 1.3557021127992055e-05, + "loss": 0.172, + "step": 7694 + }, + { + "epoch": 3.22, + "learning_rate": 1.3555436548785066e-05, + "loss": 0.1939, + "step": 7695 + }, + { + "epoch": 3.22, + "learning_rate": 1.3553851867380205e-05, + "loss": 0.1617, + "step": 7696 + }, + { + "epoch": 3.22, + "learning_rate": 1.3552267083823028e-05, + "loss": 0.1244, + "step": 7697 + }, + { + "epoch": 3.22, + "learning_rate": 1.3550682198159084e-05, + "loss": 0.1537, + "step": 7698 + }, + { + "epoch": 3.22, + "learning_rate": 1.3549097210433932e-05, + "loss": 0.1861, + "step": 7699 + }, + { + "epoch": 3.22, + "learning_rate": 1.354751212069313e-05, + "loss": 0.2075, + "step": 7700 + }, + { + "epoch": 3.22, + "learning_rate": 1.3545926928982241e-05, + "loss": 0.1893, + "step": 7701 + }, + { + "epoch": 3.22, + "learning_rate": 1.3544341635346827e-05, + "loss": 0.215, + "step": 7702 + }, + { + "epoch": 3.23, + "learning_rate": 1.3542756239832458e-05, + "loss": 0.1982, + "step": 7703 + }, + { + "epoch": 3.23, + "learning_rate": 1.3541170742484705e-05, + "loss": 0.2129, + "step": 7704 + }, + { + "epoch": 3.23, + "learning_rate": 1.3539585143349146e-05, + "loss": 0.1981, + "step": 7705 + }, + { + "epoch": 3.23, + "learning_rate": 1.3537999442471349e-05, + "loss": 0.1974, + "step": 7706 + }, + { + "epoch": 3.23, + "learning_rate": 1.3536413639896895e-05, + "loss": 0.176, + "step": 7707 + }, + { + "epoch": 3.23, + "learning_rate": 1.3534827735671373e-05, + "loss": 0.2061, + "step": 7708 + }, + { + "epoch": 3.23, + "learning_rate": 1.3533241729840362e-05, + "loss": 0.1642, + "step": 7709 + }, + { + "epoch": 3.23, + "learning_rate": 1.3531655622449454e-05, + "loss": 0.1684, + "step": 7710 + }, + { + "epoch": 3.23, + "learning_rate": 1.3530069413544241e-05, + "loss": 0.1948, + "step": 7711 + }, + { + "epoch": 3.23, + "learning_rate": 1.3528483103170312e-05, + "loss": 0.1634, + "step": 7712 + }, + { + "epoch": 3.23, + "learning_rate": 1.3526896691373269e-05, + "loss": 0.1526, + "step": 7713 + }, + { + "epoch": 3.23, + "learning_rate": 1.3525310178198707e-05, + "loss": 0.2019, + "step": 7714 + }, + { + "epoch": 3.23, + "learning_rate": 1.3523723563692237e-05, + "loss": 0.2137, + "step": 7715 + }, + { + "epoch": 3.23, + "learning_rate": 1.352213684789946e-05, + "loss": 0.1549, + "step": 7716 + }, + { + "epoch": 3.23, + "learning_rate": 1.3520550030865983e-05, + "loss": 0.1934, + "step": 7717 + }, + { + "epoch": 3.23, + "learning_rate": 1.3518963112637419e-05, + "loss": 0.1872, + "step": 7718 + }, + { + "epoch": 3.23, + "learning_rate": 1.3517376093259383e-05, + "loss": 0.2245, + "step": 7719 + }, + { + "epoch": 3.23, + "learning_rate": 1.3515788972777491e-05, + "loss": 0.1961, + "step": 7720 + }, + { + "epoch": 3.23, + "learning_rate": 1.3514201751237366e-05, + "loss": 0.1684, + "step": 7721 + }, + { + "epoch": 3.23, + "learning_rate": 1.3512614428684633e-05, + "loss": 0.1901, + "step": 7722 + }, + { + "epoch": 3.23, + "learning_rate": 1.3511027005164911e-05, + "loss": 0.1897, + "step": 7723 + }, + { + "epoch": 3.23, + "learning_rate": 1.3509439480723835e-05, + "loss": 0.1878, + "step": 7724 + }, + { + "epoch": 3.23, + "learning_rate": 1.3507851855407038e-05, + "loss": 0.1834, + "step": 7725 + }, + { + "epoch": 3.23, + "learning_rate": 1.350626412926015e-05, + "loss": 0.1637, + "step": 7726 + }, + { + "epoch": 3.24, + "learning_rate": 1.3504676302328813e-05, + "loss": 0.1433, + "step": 7727 + }, + { + "epoch": 3.24, + "learning_rate": 1.3503088374658661e-05, + "loss": 0.2066, + "step": 7728 + }, + { + "epoch": 3.24, + "learning_rate": 1.3501500346295347e-05, + "loss": 0.1597, + "step": 7729 + }, + { + "epoch": 3.24, + "learning_rate": 1.349991221728451e-05, + "loss": 0.201, + "step": 7730 + }, + { + "epoch": 3.24, + "learning_rate": 1.3498323987671801e-05, + "loss": 0.1925, + "step": 7731 + }, + { + "epoch": 3.24, + "learning_rate": 1.3496735657502877e-05, + "loss": 0.1942, + "step": 7732 + }, + { + "epoch": 3.24, + "learning_rate": 1.3495147226823386e-05, + "loss": 0.2095, + "step": 7733 + }, + { + "epoch": 3.24, + "learning_rate": 1.3493558695678992e-05, + "loss": 0.1603, + "step": 7734 + }, + { + "epoch": 3.24, + "learning_rate": 1.3491970064115351e-05, + "loss": 0.17, + "step": 7735 + }, + { + "epoch": 3.24, + "learning_rate": 1.3490381332178131e-05, + "loss": 0.1829, + "step": 7736 + }, + { + "epoch": 3.24, + "learning_rate": 1.3488792499912998e-05, + "loss": 0.2311, + "step": 7737 + }, + { + "epoch": 3.24, + "learning_rate": 1.348720356736562e-05, + "loss": 0.2073, + "step": 7738 + }, + { + "epoch": 3.24, + "learning_rate": 1.348561453458167e-05, + "loss": 0.1692, + "step": 7739 + }, + { + "epoch": 3.24, + "learning_rate": 1.3484025401606821e-05, + "loss": 0.1917, + "step": 7740 + }, + { + "epoch": 3.24, + "learning_rate": 1.3482436168486754e-05, + "loss": 0.1508, + "step": 7741 + }, + { + "epoch": 3.24, + "learning_rate": 1.3480846835267152e-05, + "loss": 0.1405, + "step": 7742 + }, + { + "epoch": 3.24, + "learning_rate": 1.3479257401993696e-05, + "loss": 0.2095, + "step": 7743 + }, + { + "epoch": 3.24, + "learning_rate": 1.347766786871207e-05, + "loss": 0.2075, + "step": 7744 + }, + { + "epoch": 3.24, + "learning_rate": 1.347607823546797e-05, + "loss": 0.1521, + "step": 7745 + }, + { + "epoch": 3.24, + "learning_rate": 1.3474488502307085e-05, + "loss": 0.1693, + "step": 7746 + }, + { + "epoch": 3.24, + "learning_rate": 1.3472898669275112e-05, + "loss": 0.1621, + "step": 7747 + }, + { + "epoch": 3.24, + "learning_rate": 1.347130873641775e-05, + "loss": 0.2106, + "step": 7748 + }, + { + "epoch": 3.24, + "learning_rate": 1.3469718703780694e-05, + "loss": 0.2237, + "step": 7749 + }, + { + "epoch": 3.24, + "learning_rate": 1.3468128571409658e-05, + "loss": 0.1928, + "step": 7750 + }, + { + "epoch": 3.25, + "learning_rate": 1.346653833935034e-05, + "loss": 0.2011, + "step": 7751 + }, + { + "epoch": 3.25, + "learning_rate": 1.3464948007648457e-05, + "loss": 0.1587, + "step": 7752 + }, + { + "epoch": 3.25, + "learning_rate": 1.3463357576349716e-05, + "loss": 0.1522, + "step": 7753 + }, + { + "epoch": 3.25, + "learning_rate": 1.3461767045499834e-05, + "loss": 0.2028, + "step": 7754 + }, + { + "epoch": 3.25, + "learning_rate": 1.346017641514453e-05, + "loss": 0.1807, + "step": 7755 + }, + { + "epoch": 3.25, + "learning_rate": 1.3458585685329529e-05, + "loss": 0.1705, + "step": 7756 + }, + { + "epoch": 3.25, + "learning_rate": 1.3456994856100547e-05, + "loss": 0.2111, + "step": 7757 + }, + { + "epoch": 3.25, + "learning_rate": 1.3455403927503317e-05, + "loss": 0.1834, + "step": 7758 + }, + { + "epoch": 3.25, + "learning_rate": 1.3453812899583572e-05, + "loss": 0.1738, + "step": 7759 + }, + { + "epoch": 3.25, + "learning_rate": 1.3452221772387036e-05, + "loss": 0.1727, + "step": 7760 + }, + { + "epoch": 3.25, + "learning_rate": 1.3450630545959448e-05, + "loss": 0.1906, + "step": 7761 + }, + { + "epoch": 3.25, + "learning_rate": 1.3449039220346547e-05, + "loss": 0.1901, + "step": 7762 + }, + { + "epoch": 3.25, + "learning_rate": 1.3447447795594078e-05, + "loss": 0.1671, + "step": 7763 + }, + { + "epoch": 3.25, + "learning_rate": 1.3445856271747777e-05, + "loss": 0.1833, + "step": 7764 + }, + { + "epoch": 3.25, + "learning_rate": 1.3444264648853395e-05, + "loss": 0.2022, + "step": 7765 + }, + { + "epoch": 3.25, + "learning_rate": 1.3442672926956685e-05, + "loss": 0.1815, + "step": 7766 + }, + { + "epoch": 3.25, + "learning_rate": 1.3441081106103395e-05, + "loss": 0.164, + "step": 7767 + }, + { + "epoch": 3.25, + "learning_rate": 1.3439489186339283e-05, + "loss": 0.208, + "step": 7768 + }, + { + "epoch": 3.25, + "learning_rate": 1.3437897167710105e-05, + "loss": 0.2039, + "step": 7769 + }, + { + "epoch": 3.25, + "learning_rate": 1.3436305050261625e-05, + "loss": 0.1856, + "step": 7770 + }, + { + "epoch": 3.25, + "learning_rate": 1.3434712834039606e-05, + "loss": 0.1896, + "step": 7771 + }, + { + "epoch": 3.25, + "learning_rate": 1.3433120519089813e-05, + "loss": 0.1631, + "step": 7772 + }, + { + "epoch": 3.25, + "learning_rate": 1.3431528105458021e-05, + "loss": 0.2288, + "step": 7773 + }, + { + "epoch": 3.25, + "learning_rate": 1.3429935593189996e-05, + "loss": 0.1845, + "step": 7774 + }, + { + "epoch": 3.26, + "learning_rate": 1.3428342982331515e-05, + "loss": 0.2092, + "step": 7775 + }, + { + "epoch": 3.26, + "learning_rate": 1.3426750272928359e-05, + "loss": 0.1466, + "step": 7776 + }, + { + "epoch": 3.26, + "learning_rate": 1.3425157465026305e-05, + "loss": 0.2149, + "step": 7777 + }, + { + "epoch": 3.26, + "learning_rate": 1.342356455867114e-05, + "loss": 0.2079, + "step": 7778 + }, + { + "epoch": 3.26, + "learning_rate": 1.342197155390865e-05, + "loss": 0.1664, + "step": 7779 + }, + { + "epoch": 3.26, + "learning_rate": 1.3420378450784623e-05, + "loss": 0.1936, + "step": 7780 + }, + { + "epoch": 3.26, + "learning_rate": 1.3418785249344855e-05, + "loss": 0.1982, + "step": 7781 + }, + { + "epoch": 3.26, + "learning_rate": 1.3417191949635137e-05, + "loss": 0.1644, + "step": 7782 + }, + { + "epoch": 3.26, + "learning_rate": 1.341559855170127e-05, + "loss": 0.1735, + "step": 7783 + }, + { + "epoch": 3.26, + "learning_rate": 1.3414005055589057e-05, + "loss": 0.1726, + "step": 7784 + }, + { + "epoch": 3.26, + "learning_rate": 1.3412411461344293e-05, + "loss": 0.1926, + "step": 7785 + }, + { + "epoch": 3.26, + "learning_rate": 1.3410817769012792e-05, + "loss": 0.1661, + "step": 7786 + }, + { + "epoch": 3.26, + "learning_rate": 1.340922397864036e-05, + "loss": 0.2143, + "step": 7787 + }, + { + "epoch": 3.26, + "learning_rate": 1.3407630090272808e-05, + "loss": 0.1829, + "step": 7788 + }, + { + "epoch": 3.26, + "learning_rate": 1.3406036103955957e-05, + "loss": 0.1304, + "step": 7789 + }, + { + "epoch": 3.26, + "learning_rate": 1.3404442019735617e-05, + "loss": 0.1993, + "step": 7790 + }, + { + "epoch": 3.26, + "learning_rate": 1.3402847837657614e-05, + "loss": 0.1636, + "step": 7791 + }, + { + "epoch": 3.26, + "learning_rate": 1.3401253557767768e-05, + "loss": 0.1769, + "step": 7792 + }, + { + "epoch": 3.26, + "learning_rate": 1.339965918011191e-05, + "loss": 0.1884, + "step": 7793 + }, + { + "epoch": 3.26, + "learning_rate": 1.3398064704735862e-05, + "loss": 0.1943, + "step": 7794 + }, + { + "epoch": 3.26, + "learning_rate": 1.3396470131685462e-05, + "loss": 0.2005, + "step": 7795 + }, + { + "epoch": 3.26, + "learning_rate": 1.3394875461006543e-05, + "loss": 0.2422, + "step": 7796 + }, + { + "epoch": 3.26, + "learning_rate": 1.3393280692744938e-05, + "loss": 0.1647, + "step": 7797 + }, + { + "epoch": 3.26, + "learning_rate": 1.3391685826946492e-05, + "loss": 0.2119, + "step": 7798 + }, + { + "epoch": 3.27, + "learning_rate": 1.3390090863657048e-05, + "loss": 0.1554, + "step": 7799 + }, + { + "epoch": 3.27, + "learning_rate": 1.338849580292245e-05, + "loss": 0.1796, + "step": 7800 + }, + { + "epoch": 3.27, + "learning_rate": 1.3386900644788546e-05, + "loss": 0.1523, + "step": 7801 + }, + { + "epoch": 3.27, + "learning_rate": 1.3385305389301187e-05, + "loss": 0.1896, + "step": 7802 + }, + { + "epoch": 3.27, + "learning_rate": 1.3383710036506232e-05, + "loss": 0.2046, + "step": 7803 + }, + { + "epoch": 3.27, + "learning_rate": 1.3382114586449533e-05, + "loss": 0.1643, + "step": 7804 + }, + { + "epoch": 3.27, + "learning_rate": 1.3380519039176951e-05, + "loss": 0.1594, + "step": 7805 + }, + { + "epoch": 3.27, + "learning_rate": 1.337892339473435e-05, + "loss": 0.1781, + "step": 7806 + }, + { + "epoch": 3.27, + "learning_rate": 1.3377327653167596e-05, + "loss": 0.1904, + "step": 7807 + }, + { + "epoch": 3.27, + "learning_rate": 1.3375731814522552e-05, + "loss": 0.1833, + "step": 7808 + }, + { + "epoch": 3.27, + "learning_rate": 1.3374135878845094e-05, + "loss": 0.1631, + "step": 7809 + }, + { + "epoch": 3.27, + "learning_rate": 1.3372539846181096e-05, + "loss": 0.2096, + "step": 7810 + }, + { + "epoch": 3.27, + "learning_rate": 1.337094371657643e-05, + "loss": 0.1732, + "step": 7811 + }, + { + "epoch": 3.27, + "learning_rate": 1.3369347490076981e-05, + "loss": 0.2059, + "step": 7812 + }, + { + "epoch": 3.27, + "learning_rate": 1.3367751166728626e-05, + "loss": 0.1556, + "step": 7813 + }, + { + "epoch": 3.27, + "learning_rate": 1.3366154746577254e-05, + "loss": 0.152, + "step": 7814 + }, + { + "epoch": 3.27, + "learning_rate": 1.3364558229668748e-05, + "loss": 0.1588, + "step": 7815 + }, + { + "epoch": 3.27, + "learning_rate": 1.3362961616049006e-05, + "loss": 0.1508, + "step": 7816 + }, + { + "epoch": 3.27, + "learning_rate": 1.3361364905763912e-05, + "loss": 0.2059, + "step": 7817 + }, + { + "epoch": 3.27, + "learning_rate": 1.335976809885937e-05, + "loss": 0.1803, + "step": 7818 + }, + { + "epoch": 3.27, + "learning_rate": 1.3358171195381272e-05, + "loss": 0.2117, + "step": 7819 + }, + { + "epoch": 3.27, + "learning_rate": 1.3356574195375526e-05, + "loss": 0.118, + "step": 7820 + }, + { + "epoch": 3.27, + "learning_rate": 1.335497709888803e-05, + "loss": 0.1968, + "step": 7821 + }, + { + "epoch": 3.27, + "learning_rate": 1.3353379905964698e-05, + "loss": 0.1957, + "step": 7822 + }, + { + "epoch": 3.28, + "learning_rate": 1.3351782616651436e-05, + "loss": 0.1985, + "step": 7823 + }, + { + "epoch": 3.28, + "learning_rate": 1.3350185230994157e-05, + "loss": 0.227, + "step": 7824 + }, + { + "epoch": 3.28, + "learning_rate": 1.3348587749038773e-05, + "loss": 0.2088, + "step": 7825 + }, + { + "epoch": 3.28, + "learning_rate": 1.334699017083121e-05, + "loss": 0.1898, + "step": 7826 + }, + { + "epoch": 3.28, + "learning_rate": 1.3345392496417381e-05, + "loss": 0.1721, + "step": 7827 + }, + { + "epoch": 3.28, + "learning_rate": 1.3343794725843215e-05, + "loss": 0.1599, + "step": 7828 + }, + { + "epoch": 3.28, + "learning_rate": 1.3342196859154639e-05, + "loss": 0.2064, + "step": 7829 + }, + { + "epoch": 3.28, + "learning_rate": 1.3340598896397579e-05, + "loss": 0.1717, + "step": 7830 + }, + { + "epoch": 3.28, + "learning_rate": 1.3339000837617967e-05, + "loss": 0.161, + "step": 7831 + }, + { + "epoch": 3.28, + "learning_rate": 1.3337402682861737e-05, + "loss": 0.1659, + "step": 7832 + }, + { + "epoch": 3.28, + "learning_rate": 1.333580443217483e-05, + "loss": 0.1689, + "step": 7833 + }, + { + "epoch": 3.28, + "learning_rate": 1.3334206085603186e-05, + "loss": 0.2025, + "step": 7834 + }, + { + "epoch": 3.28, + "learning_rate": 1.3332607643192745e-05, + "loss": 0.1741, + "step": 7835 + }, + { + "epoch": 3.28, + "learning_rate": 1.3331009104989456e-05, + "loss": 0.1395, + "step": 7836 + }, + { + "epoch": 3.28, + "learning_rate": 1.3329410471039266e-05, + "loss": 0.2303, + "step": 7837 + }, + { + "epoch": 3.28, + "learning_rate": 1.3327811741388126e-05, + "loss": 0.1997, + "step": 7838 + }, + { + "epoch": 3.28, + "learning_rate": 1.3326212916081991e-05, + "loss": 0.1643, + "step": 7839 + }, + { + "epoch": 3.28, + "learning_rate": 1.3324613995166818e-05, + "loss": 0.1636, + "step": 7840 + }, + { + "epoch": 3.28, + "learning_rate": 1.3323014978688567e-05, + "loss": 0.1608, + "step": 7841 + }, + { + "epoch": 3.28, + "learning_rate": 1.3321415866693195e-05, + "loss": 0.2052, + "step": 7842 + }, + { + "epoch": 3.28, + "learning_rate": 1.3319816659226672e-05, + "loss": 0.1693, + "step": 7843 + }, + { + "epoch": 3.28, + "learning_rate": 1.3318217356334967e-05, + "loss": 0.1523, + "step": 7844 + }, + { + "epoch": 3.28, + "learning_rate": 1.3316617958064049e-05, + "loss": 0.1823, + "step": 7845 + }, + { + "epoch": 3.28, + "learning_rate": 1.331501846445989e-05, + "loss": 0.2051, + "step": 7846 + }, + { + "epoch": 3.29, + "learning_rate": 1.3313418875568469e-05, + "loss": 0.197, + "step": 7847 + }, + { + "epoch": 3.29, + "learning_rate": 1.331181919143576e-05, + "loss": 0.1993, + "step": 7848 + }, + { + "epoch": 3.29, + "learning_rate": 1.3310219412107745e-05, + "loss": 0.1948, + "step": 7849 + }, + { + "epoch": 3.29, + "learning_rate": 1.3308619537630416e-05, + "loss": 0.1773, + "step": 7850 + }, + { + "epoch": 3.29, + "learning_rate": 1.3307019568049751e-05, + "loss": 0.1666, + "step": 7851 + }, + { + "epoch": 3.29, + "learning_rate": 1.3305419503411745e-05, + "loss": 0.1913, + "step": 7852 + }, + { + "epoch": 3.29, + "learning_rate": 1.3303819343762387e-05, + "loss": 0.2112, + "step": 7853 + }, + { + "epoch": 3.29, + "learning_rate": 1.3302219089147675e-05, + "loss": 0.1746, + "step": 7854 + }, + { + "epoch": 3.29, + "learning_rate": 1.3300618739613605e-05, + "loss": 0.1547, + "step": 7855 + }, + { + "epoch": 3.29, + "learning_rate": 1.3299018295206176e-05, + "loss": 0.1833, + "step": 7856 + }, + { + "epoch": 3.29, + "learning_rate": 1.3297417755971399e-05, + "loss": 0.1845, + "step": 7857 + }, + { + "epoch": 3.29, + "learning_rate": 1.329581712195527e-05, + "loss": 0.1692, + "step": 7858 + }, + { + "epoch": 3.29, + "learning_rate": 1.3294216393203803e-05, + "loss": 0.1781, + "step": 7859 + }, + { + "epoch": 3.29, + "learning_rate": 1.329261556976301e-05, + "loss": 0.1617, + "step": 7860 + }, + { + "epoch": 3.29, + "learning_rate": 1.3291014651678905e-05, + "loss": 0.215, + "step": 7861 + }, + { + "epoch": 3.29, + "learning_rate": 1.3289413638997502e-05, + "loss": 0.1892, + "step": 7862 + }, + { + "epoch": 3.29, + "learning_rate": 1.3287812531764826e-05, + "loss": 0.1876, + "step": 7863 + }, + { + "epoch": 3.29, + "learning_rate": 1.3286211330026894e-05, + "loss": 0.1248, + "step": 7864 + }, + { + "epoch": 3.29, + "learning_rate": 1.3284610033829734e-05, + "loss": 0.2129, + "step": 7865 + }, + { + "epoch": 3.29, + "learning_rate": 1.3283008643219373e-05, + "loss": 0.2075, + "step": 7866 + }, + { + "epoch": 3.29, + "learning_rate": 1.3281407158241842e-05, + "loss": 0.1871, + "step": 7867 + }, + { + "epoch": 3.29, + "learning_rate": 1.3279805578943174e-05, + "loss": 0.1847, + "step": 7868 + }, + { + "epoch": 3.29, + "learning_rate": 1.3278203905369407e-05, + "loss": 0.1944, + "step": 7869 + }, + { + "epoch": 3.29, + "learning_rate": 1.3276602137566575e-05, + "loss": 0.2002, + "step": 7870 + }, + { + "epoch": 3.3, + "learning_rate": 1.3275000275580723e-05, + "loss": 0.1795, + "step": 7871 + }, + { + "epoch": 3.3, + "learning_rate": 1.3273398319457895e-05, + "loss": 0.1858, + "step": 7872 + }, + { + "epoch": 3.3, + "learning_rate": 1.3271796269244135e-05, + "loss": 0.1892, + "step": 7873 + }, + { + "epoch": 3.3, + "learning_rate": 1.3270194124985499e-05, + "loss": 0.1832, + "step": 7874 + }, + { + "epoch": 3.3, + "learning_rate": 1.326859188672803e-05, + "loss": 0.1631, + "step": 7875 + }, + { + "epoch": 3.3, + "learning_rate": 1.3266989554517791e-05, + "loss": 0.2076, + "step": 7876 + }, + { + "epoch": 3.3, + "learning_rate": 1.3265387128400833e-05, + "loss": 0.2149, + "step": 7877 + }, + { + "epoch": 3.3, + "learning_rate": 1.3263784608423225e-05, + "loss": 0.186, + "step": 7878 + }, + { + "epoch": 3.3, + "learning_rate": 1.326218199463102e-05, + "loss": 0.1097, + "step": 7879 + }, + { + "epoch": 3.3, + "learning_rate": 1.3260579287070288e-05, + "loss": 0.1991, + "step": 7880 + }, + { + "epoch": 3.3, + "learning_rate": 1.3258976485787101e-05, + "loss": 0.1648, + "step": 7881 + }, + { + "epoch": 3.3, + "learning_rate": 1.3257373590827525e-05, + "loss": 0.1858, + "step": 7882 + }, + { + "epoch": 3.3, + "learning_rate": 1.3255770602237634e-05, + "loss": 0.2108, + "step": 7883 + }, + { + "epoch": 3.3, + "learning_rate": 1.325416752006351e-05, + "loss": 0.1963, + "step": 7884 + }, + { + "epoch": 3.3, + "learning_rate": 1.3252564344351227e-05, + "loss": 0.2124, + "step": 7885 + }, + { + "epoch": 3.3, + "learning_rate": 1.3250961075146868e-05, + "loss": 0.1754, + "step": 7886 + }, + { + "epoch": 3.3, + "learning_rate": 1.3249357712496518e-05, + "loss": 0.1954, + "step": 7887 + }, + { + "epoch": 3.3, + "learning_rate": 1.3247754256446263e-05, + "loss": 0.1774, + "step": 7888 + }, + { + "epoch": 3.3, + "learning_rate": 1.3246150707042196e-05, + "loss": 0.1806, + "step": 7889 + }, + { + "epoch": 3.3, + "learning_rate": 1.3244547064330404e-05, + "loss": 0.1706, + "step": 7890 + }, + { + "epoch": 3.3, + "learning_rate": 1.3242943328356988e-05, + "loss": 0.1864, + "step": 7891 + }, + { + "epoch": 3.3, + "learning_rate": 1.3241339499168042e-05, + "loss": 0.1964, + "step": 7892 + }, + { + "epoch": 3.3, + "learning_rate": 1.323973557680967e-05, + "loss": 0.1794, + "step": 7893 + }, + { + "epoch": 3.31, + "learning_rate": 1.3238131561327976e-05, + "loss": 0.162, + "step": 7894 + }, + { + "epoch": 3.31, + "learning_rate": 1.3236527452769061e-05, + "loss": 0.1988, + "step": 7895 + }, + { + "epoch": 3.31, + "learning_rate": 1.3234923251179036e-05, + "loss": 0.1611, + "step": 7896 + }, + { + "epoch": 3.31, + "learning_rate": 1.3233318956604013e-05, + "loss": 0.1859, + "step": 7897 + }, + { + "epoch": 3.31, + "learning_rate": 1.3231714569090108e-05, + "loss": 0.1789, + "step": 7898 + }, + { + "epoch": 3.31, + "learning_rate": 1.3230110088683433e-05, + "loss": 0.1955, + "step": 7899 + }, + { + "epoch": 3.31, + "learning_rate": 1.322850551543011e-05, + "loss": 0.2029, + "step": 7900 + }, + { + "epoch": 3.31, + "learning_rate": 1.3226900849376263e-05, + "loss": 0.1774, + "step": 7901 + }, + { + "epoch": 3.31, + "learning_rate": 1.3225296090568011e-05, + "loss": 0.204, + "step": 7902 + }, + { + "epoch": 3.31, + "learning_rate": 1.3223691239051489e-05, + "loss": 0.1998, + "step": 7903 + }, + { + "epoch": 3.31, + "learning_rate": 1.3222086294872818e-05, + "loss": 0.1742, + "step": 7904 + }, + { + "epoch": 3.31, + "learning_rate": 1.3220481258078142e-05, + "loss": 0.2086, + "step": 7905 + }, + { + "epoch": 3.31, + "learning_rate": 1.3218876128713584e-05, + "loss": 0.2133, + "step": 7906 + }, + { + "epoch": 3.31, + "learning_rate": 1.321727090682529e-05, + "loss": 0.2085, + "step": 7907 + }, + { + "epoch": 3.31, + "learning_rate": 1.3215665592459401e-05, + "loss": 0.1646, + "step": 7908 + }, + { + "epoch": 3.31, + "learning_rate": 1.3214060185662058e-05, + "loss": 0.1632, + "step": 7909 + }, + { + "epoch": 3.31, + "learning_rate": 1.3212454686479406e-05, + "loss": 0.1936, + "step": 7910 + }, + { + "epoch": 3.31, + "learning_rate": 1.3210849094957594e-05, + "loss": 0.1792, + "step": 7911 + }, + { + "epoch": 3.31, + "learning_rate": 1.3209243411142776e-05, + "loss": 0.1455, + "step": 7912 + }, + { + "epoch": 3.31, + "learning_rate": 1.32076376350811e-05, + "loss": 0.1829, + "step": 7913 + }, + { + "epoch": 3.31, + "learning_rate": 1.320603176681873e-05, + "loss": 0.2218, + "step": 7914 + }, + { + "epoch": 3.31, + "learning_rate": 1.3204425806401822e-05, + "loss": 0.1586, + "step": 7915 + }, + { + "epoch": 3.31, + "learning_rate": 1.3202819753876535e-05, + "loss": 0.204, + "step": 7916 + }, + { + "epoch": 3.31, + "learning_rate": 1.3201213609289036e-05, + "loss": 0.168, + "step": 7917 + }, + { + "epoch": 3.32, + "learning_rate": 1.3199607372685497e-05, + "loss": 0.1784, + "step": 7918 + }, + { + "epoch": 3.32, + "learning_rate": 1.319800104411208e-05, + "loss": 0.2257, + "step": 7919 + }, + { + "epoch": 3.32, + "learning_rate": 1.3196394623614963e-05, + "loss": 0.1622, + "step": 7920 + }, + { + "epoch": 3.32, + "learning_rate": 1.3194788111240317e-05, + "loss": 0.1386, + "step": 7921 + }, + { + "epoch": 3.32, + "learning_rate": 1.3193181507034323e-05, + "loss": 0.1977, + "step": 7922 + }, + { + "epoch": 3.32, + "learning_rate": 1.3191574811043159e-05, + "loss": 0.2258, + "step": 7923 + }, + { + "epoch": 3.32, + "learning_rate": 1.318996802331301e-05, + "loss": 0.2303, + "step": 7924 + }, + { + "epoch": 3.32, + "learning_rate": 1.318836114389006e-05, + "loss": 0.1982, + "step": 7925 + }, + { + "epoch": 3.32, + "learning_rate": 1.31867541728205e-05, + "loss": 0.1936, + "step": 7926 + }, + { + "epoch": 3.32, + "learning_rate": 1.3185147110150516e-05, + "loss": 0.1944, + "step": 7927 + }, + { + "epoch": 3.32, + "learning_rate": 1.3183539955926307e-05, + "loss": 0.191, + "step": 7928 + }, + { + "epoch": 3.32, + "learning_rate": 1.3181932710194067e-05, + "loss": 0.2047, + "step": 7929 + }, + { + "epoch": 3.32, + "learning_rate": 1.3180325372999994e-05, + "loss": 0.167, + "step": 7930 + }, + { + "epoch": 3.32, + "learning_rate": 1.3178717944390292e-05, + "loss": 0.2035, + "step": 7931 + }, + { + "epoch": 3.32, + "learning_rate": 1.3177110424411164e-05, + "loss": 0.2037, + "step": 7932 + }, + { + "epoch": 3.32, + "learning_rate": 1.3175502813108813e-05, + "loss": 0.2023, + "step": 7933 + }, + { + "epoch": 3.32, + "learning_rate": 1.3173895110529454e-05, + "loss": 0.1541, + "step": 7934 + }, + { + "epoch": 3.32, + "learning_rate": 1.3172287316719297e-05, + "loss": 0.1985, + "step": 7935 + }, + { + "epoch": 3.32, + "learning_rate": 1.3170679431724553e-05, + "loss": 0.1705, + "step": 7936 + }, + { + "epoch": 3.32, + "learning_rate": 1.3169071455591443e-05, + "loss": 0.2345, + "step": 7937 + }, + { + "epoch": 3.32, + "learning_rate": 1.3167463388366186e-05, + "loss": 0.2211, + "step": 7938 + }, + { + "epoch": 3.32, + "learning_rate": 1.3165855230095008e-05, + "loss": 0.2096, + "step": 7939 + }, + { + "epoch": 3.32, + "learning_rate": 1.3164246980824126e-05, + "loss": 0.1247, + "step": 7940 + }, + { + "epoch": 3.32, + "learning_rate": 1.3162638640599777e-05, + "loss": 0.1589, + "step": 7941 + }, + { + "epoch": 3.33, + "learning_rate": 1.3161030209468187e-05, + "loss": 0.1761, + "step": 7942 + }, + { + "epoch": 3.33, + "learning_rate": 1.3159421687475586e-05, + "loss": 0.1564, + "step": 7943 + }, + { + "epoch": 3.33, + "learning_rate": 1.3157813074668213e-05, + "loss": 0.1976, + "step": 7944 + }, + { + "epoch": 3.33, + "learning_rate": 1.3156204371092302e-05, + "loss": 0.1818, + "step": 7945 + }, + { + "epoch": 3.33, + "learning_rate": 1.3154595576794103e-05, + "loss": 0.2022, + "step": 7946 + }, + { + "epoch": 3.33, + "learning_rate": 1.315298669181985e-05, + "loss": 0.1456, + "step": 7947 + }, + { + "epoch": 3.33, + "learning_rate": 1.3151377716215791e-05, + "loss": 0.1878, + "step": 7948 + }, + { + "epoch": 3.33, + "learning_rate": 1.3149768650028181e-05, + "loss": 0.1428, + "step": 7949 + }, + { + "epoch": 3.33, + "learning_rate": 1.3148159493303262e-05, + "loss": 0.1644, + "step": 7950 + }, + { + "epoch": 3.33, + "learning_rate": 1.3146550246087293e-05, + "loss": 0.1733, + "step": 7951 + }, + { + "epoch": 3.33, + "learning_rate": 1.3144940908426532e-05, + "loss": 0.1824, + "step": 7952 + }, + { + "epoch": 3.33, + "learning_rate": 1.314333148036723e-05, + "loss": 0.1695, + "step": 7953 + }, + { + "epoch": 3.33, + "learning_rate": 1.3141721961955658e-05, + "loss": 0.1797, + "step": 7954 + }, + { + "epoch": 3.33, + "learning_rate": 1.3140112353238075e-05, + "loss": 0.1447, + "step": 7955 + }, + { + "epoch": 3.33, + "learning_rate": 1.3138502654260749e-05, + "loss": 0.1697, + "step": 7956 + }, + { + "epoch": 3.33, + "learning_rate": 1.313689286506995e-05, + "loss": 0.1789, + "step": 7957 + }, + { + "epoch": 3.33, + "learning_rate": 1.3135282985711948e-05, + "loss": 0.1959, + "step": 7958 + }, + { + "epoch": 3.33, + "learning_rate": 1.313367301623302e-05, + "loss": 0.1866, + "step": 7959 + }, + { + "epoch": 3.33, + "learning_rate": 1.3132062956679441e-05, + "loss": 0.1782, + "step": 7960 + }, + { + "epoch": 3.33, + "learning_rate": 1.3130452807097491e-05, + "loss": 0.195, + "step": 7961 + }, + { + "epoch": 3.33, + "learning_rate": 1.3128842567533456e-05, + "loss": 0.2005, + "step": 7962 + }, + { + "epoch": 3.33, + "learning_rate": 1.3127232238033613e-05, + "loss": 0.1895, + "step": 7963 + }, + { + "epoch": 3.33, + "learning_rate": 1.3125621818644258e-05, + "loss": 0.2153, + "step": 7964 + }, + { + "epoch": 3.33, + "learning_rate": 1.3124011309411678e-05, + "loss": 0.2004, + "step": 7965 + }, + { + "epoch": 3.34, + "learning_rate": 1.3122400710382164e-05, + "loss": 0.211, + "step": 7966 + }, + { + "epoch": 3.34, + "learning_rate": 1.3120790021602012e-05, + "loss": 0.1833, + "step": 7967 + }, + { + "epoch": 3.34, + "learning_rate": 1.311917924311752e-05, + "loss": 0.2227, + "step": 7968 + }, + { + "epoch": 3.34, + "learning_rate": 1.311756837497499e-05, + "loss": 0.2323, + "step": 7969 + }, + { + "epoch": 3.34, + "learning_rate": 1.3115957417220722e-05, + "loss": 0.1494, + "step": 7970 + }, + { + "epoch": 3.34, + "learning_rate": 1.3114346369901021e-05, + "loss": 0.1479, + "step": 7971 + }, + { + "epoch": 3.34, + "learning_rate": 1.3112735233062201e-05, + "loss": 0.1661, + "step": 7972 + }, + { + "epoch": 3.34, + "learning_rate": 1.311112400675057e-05, + "loss": 0.1722, + "step": 7973 + }, + { + "epoch": 3.34, + "learning_rate": 1.3109512691012434e-05, + "loss": 0.1949, + "step": 7974 + }, + { + "epoch": 3.34, + "learning_rate": 1.3107901285894118e-05, + "loss": 0.2101, + "step": 7975 + }, + { + "epoch": 3.34, + "learning_rate": 1.3106289791441941e-05, + "loss": 0.1923, + "step": 7976 + }, + { + "epoch": 3.34, + "learning_rate": 1.3104678207702218e-05, + "loss": 0.1802, + "step": 7977 + }, + { + "epoch": 3.34, + "learning_rate": 1.3103066534721273e-05, + "loss": 0.2054, + "step": 7978 + }, + { + "epoch": 3.34, + "learning_rate": 1.3101454772545435e-05, + "loss": 0.2031, + "step": 7979 + }, + { + "epoch": 3.34, + "learning_rate": 1.3099842921221034e-05, + "loss": 0.2368, + "step": 7980 + }, + { + "epoch": 3.34, + "learning_rate": 1.3098230980794394e-05, + "loss": 0.1559, + "step": 7981 + }, + { + "epoch": 3.34, + "learning_rate": 1.309661895131186e-05, + "loss": 0.1767, + "step": 7982 + }, + { + "epoch": 3.34, + "learning_rate": 1.3095006832819759e-05, + "loss": 0.1372, + "step": 7983 + }, + { + "epoch": 3.34, + "learning_rate": 1.3093394625364433e-05, + "loss": 0.1748, + "step": 7984 + }, + { + "epoch": 3.34, + "learning_rate": 1.3091782328992225e-05, + "loss": 0.1735, + "step": 7985 + }, + { + "epoch": 3.34, + "learning_rate": 1.3090169943749475e-05, + "loss": 0.1624, + "step": 7986 + }, + { + "epoch": 3.34, + "learning_rate": 1.3088557469682535e-05, + "loss": 0.1809, + "step": 7987 + }, + { + "epoch": 3.34, + "learning_rate": 1.3086944906837751e-05, + "loss": 0.2157, + "step": 7988 + }, + { + "epoch": 3.34, + "learning_rate": 1.3085332255261472e-05, + "loss": 0.1759, + "step": 7989 + }, + { + "epoch": 3.35, + "learning_rate": 1.3083719515000057e-05, + "loss": 0.1399, + "step": 7990 + }, + { + "epoch": 3.35, + "learning_rate": 1.308210668609986e-05, + "loss": 0.1942, + "step": 7991 + }, + { + "epoch": 3.35, + "learning_rate": 1.3080493768607243e-05, + "loss": 0.1867, + "step": 7992 + }, + { + "epoch": 3.35, + "learning_rate": 1.3078880762568566e-05, + "loss": 0.2199, + "step": 7993 + }, + { + "epoch": 3.35, + "learning_rate": 1.3077267668030194e-05, + "loss": 0.2082, + "step": 7994 + }, + { + "epoch": 3.35, + "learning_rate": 1.3075654485038491e-05, + "loss": 0.166, + "step": 7995 + }, + { + "epoch": 3.35, + "learning_rate": 1.3074041213639832e-05, + "loss": 0.1959, + "step": 7996 + }, + { + "epoch": 3.35, + "learning_rate": 1.3072427853880583e-05, + "loss": 0.1902, + "step": 7997 + }, + { + "epoch": 3.35, + "learning_rate": 1.3070814405807124e-05, + "loss": 0.1851, + "step": 7998 + }, + { + "epoch": 3.35, + "learning_rate": 1.3069200869465832e-05, + "loss": 0.1786, + "step": 7999 + }, + { + "epoch": 3.35, + "learning_rate": 1.3067587244903081e-05, + "loss": 0.1853, + "step": 8000 + }, + { + "epoch": 3.35, + "learning_rate": 1.3065973532165259e-05, + "loss": 0.1811, + "step": 8001 + }, + { + "epoch": 3.35, + "learning_rate": 1.3064359731298746e-05, + "loss": 0.2069, + "step": 8002 + }, + { + "epoch": 3.35, + "learning_rate": 1.3062745842349935e-05, + "loss": 0.2145, + "step": 8003 + }, + { + "epoch": 3.35, + "learning_rate": 1.3061131865365211e-05, + "loss": 0.2082, + "step": 8004 + }, + { + "epoch": 3.35, + "learning_rate": 1.3059517800390967e-05, + "loss": 0.1653, + "step": 8005 + }, + { + "epoch": 3.35, + "learning_rate": 1.30579036474736e-05, + "loss": 0.1863, + "step": 8006 + }, + { + "epoch": 3.35, + "learning_rate": 1.3056289406659506e-05, + "loss": 0.172, + "step": 8007 + }, + { + "epoch": 3.35, + "learning_rate": 1.3054675077995082e-05, + "loss": 0.1559, + "step": 8008 + }, + { + "epoch": 3.35, + "learning_rate": 1.3053060661526738e-05, + "loss": 0.202, + "step": 8009 + }, + { + "epoch": 3.35, + "learning_rate": 1.3051446157300875e-05, + "loss": 0.1904, + "step": 8010 + }, + { + "epoch": 3.35, + "learning_rate": 1.3049831565363897e-05, + "loss": 0.2119, + "step": 8011 + }, + { + "epoch": 3.35, + "learning_rate": 1.3048216885762218e-05, + "loss": 0.2065, + "step": 8012 + }, + { + "epoch": 3.35, + "learning_rate": 1.3046602118542251e-05, + "loss": 0.2191, + "step": 8013 + }, + { + "epoch": 3.36, + "learning_rate": 1.3044987263750409e-05, + "loss": 0.1768, + "step": 8014 + }, + { + "epoch": 3.36, + "learning_rate": 1.3043372321433108e-05, + "loss": 0.1888, + "step": 8015 + }, + { + "epoch": 3.36, + "learning_rate": 1.304175729163677e-05, + "loss": 0.1931, + "step": 8016 + }, + { + "epoch": 3.36, + "learning_rate": 1.304014217440782e-05, + "loss": 0.1912, + "step": 8017 + }, + { + "epoch": 3.36, + "learning_rate": 1.303852696979268e-05, + "loss": 0.1944, + "step": 8018 + }, + { + "epoch": 3.36, + "learning_rate": 1.3036911677837778e-05, + "loss": 0.1651, + "step": 8019 + }, + { + "epoch": 3.36, + "learning_rate": 1.3035296298589549e-05, + "loss": 0.2085, + "step": 8020 + }, + { + "epoch": 3.36, + "learning_rate": 1.3033680832094416e-05, + "loss": 0.2003, + "step": 8021 + }, + { + "epoch": 3.36, + "learning_rate": 1.3032065278398823e-05, + "loss": 0.1749, + "step": 8022 + }, + { + "epoch": 3.36, + "learning_rate": 1.3030449637549203e-05, + "loss": 0.1729, + "step": 8023 + }, + { + "epoch": 3.36, + "learning_rate": 1.3028833909591999e-05, + "loss": 0.2078, + "step": 8024 + }, + { + "epoch": 3.36, + "learning_rate": 1.3027218094573648e-05, + "loss": 0.1727, + "step": 8025 + }, + { + "epoch": 3.36, + "learning_rate": 1.30256021925406e-05, + "loss": 0.1386, + "step": 8026 + }, + { + "epoch": 3.36, + "learning_rate": 1.3023986203539304e-05, + "loss": 0.1952, + "step": 8027 + }, + { + "epoch": 3.36, + "learning_rate": 1.3022370127616209e-05, + "loss": 0.1911, + "step": 8028 + }, + { + "epoch": 3.36, + "learning_rate": 1.3020753964817763e-05, + "loss": 0.1608, + "step": 8029 + }, + { + "epoch": 3.36, + "learning_rate": 1.301913771519043e-05, + "loss": 0.225, + "step": 8030 + }, + { + "epoch": 3.36, + "learning_rate": 1.3017521378780655e-05, + "loss": 0.2167, + "step": 8031 + }, + { + "epoch": 3.36, + "learning_rate": 1.301590495563491e-05, + "loss": 0.1651, + "step": 8032 + }, + { + "epoch": 3.36, + "learning_rate": 1.3014288445799656e-05, + "loss": 0.1953, + "step": 8033 + }, + { + "epoch": 3.36, + "learning_rate": 1.3012671849321351e-05, + "loss": 0.2044, + "step": 8034 + }, + { + "epoch": 3.36, + "learning_rate": 1.3011055166246469e-05, + "loss": 0.2039, + "step": 8035 + }, + { + "epoch": 3.36, + "learning_rate": 1.3009438396621479e-05, + "loss": 0.1804, + "step": 8036 + }, + { + "epoch": 3.36, + "learning_rate": 1.3007821540492854e-05, + "loss": 0.1906, + "step": 8037 + }, + { + "epoch": 3.37, + "learning_rate": 1.3006204597907064e-05, + "loss": 0.2217, + "step": 8038 + }, + { + "epoch": 3.37, + "learning_rate": 1.3004587568910592e-05, + "loss": 0.2237, + "step": 8039 + }, + { + "epoch": 3.37, + "learning_rate": 1.3002970453549918e-05, + "loss": 0.1782, + "step": 8040 + }, + { + "epoch": 3.37, + "learning_rate": 1.300135325187152e-05, + "loss": 0.211, + "step": 8041 + }, + { + "epoch": 3.37, + "learning_rate": 1.2999735963921887e-05, + "loss": 0.184, + "step": 8042 + }, + { + "epoch": 3.37, + "learning_rate": 1.2998118589747509e-05, + "loss": 0.1948, + "step": 8043 + }, + { + "epoch": 3.37, + "learning_rate": 1.2996501129394872e-05, + "loss": 0.1965, + "step": 8044 + }, + { + "epoch": 3.37, + "learning_rate": 1.2994883582910465e-05, + "loss": 0.1551, + "step": 8045 + }, + { + "epoch": 3.37, + "learning_rate": 1.299326595034079e-05, + "loss": 0.1572, + "step": 8046 + }, + { + "epoch": 3.37, + "learning_rate": 1.2991648231732342e-05, + "loss": 0.1887, + "step": 8047 + }, + { + "epoch": 3.37, + "learning_rate": 1.2990030427131618e-05, + "loss": 0.1751, + "step": 8048 + }, + { + "epoch": 3.37, + "learning_rate": 1.2988412536585125e-05, + "loss": 0.2004, + "step": 8049 + }, + { + "epoch": 3.37, + "learning_rate": 1.2986794560139362e-05, + "loss": 0.2151, + "step": 8050 + }, + { + "epoch": 3.37, + "learning_rate": 1.2985176497840843e-05, + "loss": 0.2198, + "step": 8051 + }, + { + "epoch": 3.37, + "learning_rate": 1.2983558349736074e-05, + "loss": 0.1641, + "step": 8052 + }, + { + "epoch": 3.37, + "learning_rate": 1.2981940115871566e-05, + "loss": 0.176, + "step": 8053 + }, + { + "epoch": 3.37, + "learning_rate": 1.2980321796293838e-05, + "loss": 0.1961, + "step": 8054 + }, + { + "epoch": 3.37, + "learning_rate": 1.2978703391049402e-05, + "loss": 0.1903, + "step": 8055 + }, + { + "epoch": 3.37, + "learning_rate": 1.2977084900184784e-05, + "loss": 0.1761, + "step": 8056 + }, + { + "epoch": 3.37, + "learning_rate": 1.2975466323746499e-05, + "loss": 0.1296, + "step": 8057 + }, + { + "epoch": 3.37, + "learning_rate": 1.2973847661781078e-05, + "loss": 0.1581, + "step": 8058 + }, + { + "epoch": 3.37, + "learning_rate": 1.2972228914335044e-05, + "loss": 0.1424, + "step": 8059 + }, + { + "epoch": 3.37, + "learning_rate": 1.2970610081454925e-05, + "loss": 0.1769, + "step": 8060 + }, + { + "epoch": 3.37, + "learning_rate": 1.2968991163187256e-05, + "loss": 0.187, + "step": 8061 + }, + { + "epoch": 3.38, + "learning_rate": 1.2967372159578574e-05, + "loss": 0.1795, + "step": 8062 + }, + { + "epoch": 3.38, + "learning_rate": 1.2965753070675406e-05, + "loss": 0.1714, + "step": 8063 + }, + { + "epoch": 3.38, + "learning_rate": 1.2964133896524305e-05, + "loss": 0.1511, + "step": 8064 + }, + { + "epoch": 3.38, + "learning_rate": 1.2962514637171802e-05, + "loss": 0.1504, + "step": 8065 + }, + { + "epoch": 3.38, + "learning_rate": 1.2960895292664443e-05, + "loss": 0.1826, + "step": 8066 + }, + { + "epoch": 3.38, + "learning_rate": 1.295927586304878e-05, + "loss": 0.1782, + "step": 8067 + }, + { + "epoch": 3.38, + "learning_rate": 1.2957656348371357e-05, + "loss": 0.2208, + "step": 8068 + }, + { + "epoch": 3.38, + "learning_rate": 1.2956036748678727e-05, + "loss": 0.1828, + "step": 8069 + }, + { + "epoch": 3.38, + "learning_rate": 1.2954417064017441e-05, + "loss": 0.1838, + "step": 8070 + }, + { + "epoch": 3.38, + "learning_rate": 1.295279729443406e-05, + "loss": 0.1932, + "step": 8071 + }, + { + "epoch": 3.38, + "learning_rate": 1.2951177439975141e-05, + "loss": 0.2389, + "step": 8072 + }, + { + "epoch": 3.38, + "learning_rate": 1.2949557500687244e-05, + "loss": 0.1934, + "step": 8073 + }, + { + "epoch": 3.38, + "learning_rate": 1.2947937476616938e-05, + "loss": 0.1962, + "step": 8074 + }, + { + "epoch": 3.38, + "learning_rate": 1.2946317367810784e-05, + "loss": 0.2111, + "step": 8075 + }, + { + "epoch": 3.38, + "learning_rate": 1.2944697174315349e-05, + "loss": 0.1932, + "step": 8076 + }, + { + "epoch": 3.38, + "learning_rate": 1.2943076896177208e-05, + "loss": 0.1867, + "step": 8077 + }, + { + "epoch": 3.38, + "learning_rate": 1.2941456533442935e-05, + "loss": 0.1605, + "step": 8078 + }, + { + "epoch": 3.38, + "learning_rate": 1.2939836086159102e-05, + "loss": 0.2104, + "step": 8079 + }, + { + "epoch": 3.38, + "learning_rate": 1.2938215554372292e-05, + "loss": 0.1872, + "step": 8080 + }, + { + "epoch": 3.38, + "learning_rate": 1.2936594938129082e-05, + "loss": 0.1978, + "step": 8081 + }, + { + "epoch": 3.38, + "learning_rate": 1.2934974237476054e-05, + "loss": 0.2306, + "step": 8082 + }, + { + "epoch": 3.38, + "learning_rate": 1.2933353452459796e-05, + "loss": 0.1974, + "step": 8083 + }, + { + "epoch": 3.38, + "learning_rate": 1.29317325831269e-05, + "loss": 0.1785, + "step": 8084 + }, + { + "epoch": 3.38, + "learning_rate": 1.293011162952395e-05, + "loss": 0.1864, + "step": 8085 + }, + { + "epoch": 3.39, + "learning_rate": 1.292849059169754e-05, + "loss": 0.2104, + "step": 8086 + }, + { + "epoch": 3.39, + "learning_rate": 1.2926869469694267e-05, + "loss": 0.2014, + "step": 8087 + }, + { + "epoch": 3.39, + "learning_rate": 1.2925248263560733e-05, + "loss": 0.2069, + "step": 8088 + }, + { + "epoch": 3.39, + "learning_rate": 1.2923626973343527e-05, + "loss": 0.1916, + "step": 8089 + }, + { + "epoch": 3.39, + "learning_rate": 1.2922005599089263e-05, + "loss": 0.1521, + "step": 8090 + }, + { + "epoch": 3.39, + "learning_rate": 1.2920384140844539e-05, + "loss": 0.2251, + "step": 8091 + }, + { + "epoch": 3.39, + "learning_rate": 1.2918762598655965e-05, + "loss": 0.2019, + "step": 8092 + }, + { + "epoch": 3.39, + "learning_rate": 1.291714097257015e-05, + "loss": 0.1826, + "step": 8093 + }, + { + "epoch": 3.39, + "learning_rate": 1.2915519262633704e-05, + "loss": 0.2038, + "step": 8094 + }, + { + "epoch": 3.39, + "learning_rate": 1.2913897468893249e-05, + "loss": 0.187, + "step": 8095 + }, + { + "epoch": 3.39, + "learning_rate": 1.2912275591395393e-05, + "loss": 0.2268, + "step": 8096 + }, + { + "epoch": 3.39, + "learning_rate": 1.291065363018676e-05, + "loss": 0.2099, + "step": 8097 + }, + { + "epoch": 3.39, + "learning_rate": 1.290903158531397e-05, + "loss": 0.1623, + "step": 8098 + }, + { + "epoch": 3.39, + "learning_rate": 1.2907409456823651e-05, + "loss": 0.2001, + "step": 8099 + }, + { + "epoch": 3.39, + "learning_rate": 1.2905787244762427e-05, + "loss": 0.1717, + "step": 8100 + }, + { + "epoch": 3.39, + "learning_rate": 1.2904164949176927e-05, + "loss": 0.1823, + "step": 8101 + }, + { + "epoch": 3.39, + "learning_rate": 1.2902542570113786e-05, + "loss": 0.2004, + "step": 8102 + }, + { + "epoch": 3.39, + "learning_rate": 1.2900920107619627e-05, + "loss": 0.1961, + "step": 8103 + }, + { + "epoch": 3.39, + "learning_rate": 1.2899297561741097e-05, + "loss": 0.2076, + "step": 8104 + }, + { + "epoch": 3.39, + "learning_rate": 1.2897674932524832e-05, + "loss": 0.1968, + "step": 8105 + }, + { + "epoch": 3.39, + "learning_rate": 1.2896052220017474e-05, + "loss": 0.1579, + "step": 8106 + }, + { + "epoch": 3.39, + "learning_rate": 1.2894429424265661e-05, + "loss": 0.1819, + "step": 8107 + }, + { + "epoch": 3.39, + "learning_rate": 1.2892806545316048e-05, + "loss": 0.191, + "step": 8108 + }, + { + "epoch": 3.4, + "learning_rate": 1.2891183583215272e-05, + "loss": 0.1599, + "step": 8109 + }, + { + "epoch": 3.4, + "learning_rate": 1.2889560538009993e-05, + "loss": 0.164, + "step": 8110 + }, + { + "epoch": 3.4, + "learning_rate": 1.288793740974686e-05, + "loss": 0.1802, + "step": 8111 + }, + { + "epoch": 3.4, + "learning_rate": 1.288631419847253e-05, + "loss": 0.174, + "step": 8112 + }, + { + "epoch": 3.4, + "learning_rate": 1.2884690904233658e-05, + "loss": 0.2084, + "step": 8113 + }, + { + "epoch": 3.4, + "learning_rate": 1.2883067527076904e-05, + "loss": 0.1475, + "step": 8114 + }, + { + "epoch": 3.4, + "learning_rate": 1.2881444067048936e-05, + "loss": 0.1961, + "step": 8115 + }, + { + "epoch": 3.4, + "learning_rate": 1.2879820524196414e-05, + "loss": 0.1655, + "step": 8116 + }, + { + "epoch": 3.4, + "learning_rate": 1.2878196898566004e-05, + "loss": 0.1942, + "step": 8117 + }, + { + "epoch": 3.4, + "learning_rate": 1.2876573190204381e-05, + "loss": 0.1498, + "step": 8118 + }, + { + "epoch": 3.4, + "learning_rate": 1.2874949399158213e-05, + "loss": 0.2078, + "step": 8119 + }, + { + "epoch": 3.4, + "learning_rate": 1.2873325525474176e-05, + "loss": 0.2005, + "step": 8120 + }, + { + "epoch": 3.4, + "learning_rate": 1.2871701569198946e-05, + "loss": 0.1994, + "step": 8121 + }, + { + "epoch": 3.4, + "learning_rate": 1.2870077530379205e-05, + "loss": 0.1859, + "step": 8122 + }, + { + "epoch": 3.4, + "learning_rate": 1.286845340906163e-05, + "loss": 0.1879, + "step": 8123 + }, + { + "epoch": 3.4, + "learning_rate": 1.286682920529291e-05, + "loss": 0.1767, + "step": 8124 + }, + { + "epoch": 3.4, + "learning_rate": 1.2865204919119726e-05, + "loss": 0.203, + "step": 8125 + }, + { + "epoch": 3.4, + "learning_rate": 1.286358055058877e-05, + "loss": 0.2012, + "step": 8126 + }, + { + "epoch": 3.4, + "learning_rate": 1.2861956099746731e-05, + "loss": 0.1452, + "step": 8127 + }, + { + "epoch": 3.4, + "learning_rate": 1.2860331566640303e-05, + "loss": 0.1434, + "step": 8128 + }, + { + "epoch": 3.4, + "learning_rate": 1.2858706951316184e-05, + "loss": 0.1894, + "step": 8129 + }, + { + "epoch": 3.4, + "learning_rate": 1.285708225382107e-05, + "loss": 0.2196, + "step": 8130 + }, + { + "epoch": 3.4, + "learning_rate": 1.2855457474201663e-05, + "loss": 0.1776, + "step": 8131 + }, + { + "epoch": 3.4, + "learning_rate": 1.2853832612504662e-05, + "loss": 0.1807, + "step": 8132 + }, + { + "epoch": 3.41, + "learning_rate": 1.2852207668776777e-05, + "loss": 0.1557, + "step": 8133 + }, + { + "epoch": 3.41, + "learning_rate": 1.2850582643064713e-05, + "loss": 0.1438, + "step": 8134 + }, + { + "epoch": 3.41, + "learning_rate": 1.2848957535415183e-05, + "loss": 0.179, + "step": 8135 + }, + { + "epoch": 3.41, + "learning_rate": 1.2847332345874895e-05, + "loss": 0.1775, + "step": 8136 + }, + { + "epoch": 3.41, + "learning_rate": 1.2845707074490563e-05, + "loss": 0.1993, + "step": 8137 + }, + { + "epoch": 3.41, + "learning_rate": 1.284408172130891e-05, + "loss": 0.173, + "step": 8138 + }, + { + "epoch": 3.41, + "learning_rate": 1.284245628637665e-05, + "loss": 0.1968, + "step": 8139 + }, + { + "epoch": 3.41, + "learning_rate": 1.2840830769740506e-05, + "loss": 0.1977, + "step": 8140 + }, + { + "epoch": 3.41, + "learning_rate": 1.2839205171447203e-05, + "loss": 0.1816, + "step": 8141 + }, + { + "epoch": 3.41, + "learning_rate": 1.2837579491543468e-05, + "loss": 0.186, + "step": 8142 + }, + { + "epoch": 3.41, + "learning_rate": 1.2835953730076026e-05, + "loss": 0.2, + "step": 8143 + }, + { + "epoch": 3.41, + "learning_rate": 1.2834327887091611e-05, + "loss": 0.2119, + "step": 8144 + }, + { + "epoch": 3.41, + "learning_rate": 1.2832701962636957e-05, + "loss": 0.1799, + "step": 8145 + }, + { + "epoch": 3.41, + "learning_rate": 1.2831075956758801e-05, + "loss": 0.1658, + "step": 8146 + }, + { + "epoch": 3.41, + "learning_rate": 1.2829449869503876e-05, + "loss": 0.1929, + "step": 8147 + }, + { + "epoch": 3.41, + "learning_rate": 1.2827823700918924e-05, + "loss": 0.203, + "step": 8148 + }, + { + "epoch": 3.41, + "learning_rate": 1.2826197451050692e-05, + "loss": 0.1814, + "step": 8149 + }, + { + "epoch": 3.41, + "learning_rate": 1.2824571119945921e-05, + "loss": 0.1706, + "step": 8150 + }, + { + "epoch": 3.41, + "learning_rate": 1.282294470765136e-05, + "loss": 0.1477, + "step": 8151 + }, + { + "epoch": 3.41, + "learning_rate": 1.2821318214213758e-05, + "loss": 0.2028, + "step": 8152 + }, + { + "epoch": 3.41, + "learning_rate": 1.2819691639679867e-05, + "loss": 0.1899, + "step": 8153 + }, + { + "epoch": 3.41, + "learning_rate": 1.2818064984096443e-05, + "loss": 0.2138, + "step": 8154 + }, + { + "epoch": 3.41, + "learning_rate": 1.2816438247510243e-05, + "loss": 0.1841, + "step": 8155 + }, + { + "epoch": 3.41, + "learning_rate": 1.2814811429968022e-05, + "loss": 0.1999, + "step": 8156 + }, + { + "epoch": 3.42, + "learning_rate": 1.2813184531516545e-05, + "loss": 0.1649, + "step": 8157 + }, + { + "epoch": 3.42, + "learning_rate": 1.2811557552202576e-05, + "loss": 0.185, + "step": 8158 + }, + { + "epoch": 3.42, + "learning_rate": 1.2809930492072882e-05, + "loss": 0.1613, + "step": 8159 + }, + { + "epoch": 3.42, + "learning_rate": 1.2808303351174224e-05, + "loss": 0.189, + "step": 8160 + }, + { + "epoch": 3.42, + "learning_rate": 1.2806676129553381e-05, + "loss": 0.2033, + "step": 8161 + }, + { + "epoch": 3.42, + "learning_rate": 1.2805048827257123e-05, + "loss": 0.2193, + "step": 8162 + }, + { + "epoch": 3.42, + "learning_rate": 1.280342144433223e-05, + "loss": 0.2024, + "step": 8163 + }, + { + "epoch": 3.42, + "learning_rate": 1.280179398082547e-05, + "loss": 0.1882, + "step": 8164 + }, + { + "epoch": 3.42, + "learning_rate": 1.2800166436783627e-05, + "loss": 0.1955, + "step": 8165 + }, + { + "epoch": 3.42, + "learning_rate": 1.279853881225349e-05, + "loss": 0.1965, + "step": 8166 + }, + { + "epoch": 3.42, + "learning_rate": 1.2796911107281833e-05, + "loss": 0.1778, + "step": 8167 + }, + { + "epoch": 3.42, + "learning_rate": 1.279528332191545e-05, + "loss": 0.2018, + "step": 8168 + }, + { + "epoch": 3.42, + "learning_rate": 1.279365545620113e-05, + "loss": 0.1709, + "step": 8169 + }, + { + "epoch": 3.42, + "learning_rate": 1.2792027510185665e-05, + "loss": 0.186, + "step": 8170 + }, + { + "epoch": 3.42, + "learning_rate": 1.2790399483915843e-05, + "loss": 0.1931, + "step": 8171 + }, + { + "epoch": 3.42, + "learning_rate": 1.2788771377438465e-05, + "loss": 0.17, + "step": 8172 + }, + { + "epoch": 3.42, + "learning_rate": 1.278714319080033e-05, + "loss": 0.1962, + "step": 8173 + }, + { + "epoch": 3.42, + "learning_rate": 1.2785514924048235e-05, + "loss": 0.1863, + "step": 8174 + }, + { + "epoch": 3.42, + "learning_rate": 1.2783886577228986e-05, + "loss": 0.1965, + "step": 8175 + }, + { + "epoch": 3.42, + "learning_rate": 1.2782258150389387e-05, + "loss": 0.1853, + "step": 8176 + }, + { + "epoch": 3.42, + "learning_rate": 1.2780629643576249e-05, + "loss": 0.2044, + "step": 8177 + }, + { + "epoch": 3.42, + "learning_rate": 1.2779001056836373e-05, + "loss": 0.1349, + "step": 8178 + }, + { + "epoch": 3.42, + "learning_rate": 1.2777372390216585e-05, + "loss": 0.1896, + "step": 8179 + }, + { + "epoch": 3.42, + "learning_rate": 1.2775743643763691e-05, + "loss": 0.1824, + "step": 8180 + }, + { + "epoch": 3.43, + "learning_rate": 1.2774114817524505e-05, + "loss": 0.2179, + "step": 8181 + }, + { + "epoch": 3.43, + "learning_rate": 1.2772485911545854e-05, + "loss": 0.1746, + "step": 8182 + }, + { + "epoch": 3.43, + "learning_rate": 1.2770856925874555e-05, + "loss": 0.1649, + "step": 8183 + }, + { + "epoch": 3.43, + "learning_rate": 1.2769227860557431e-05, + "loss": 0.1964, + "step": 8184 + }, + { + "epoch": 3.43, + "learning_rate": 1.2767598715641309e-05, + "loss": 0.1795, + "step": 8185 + }, + { + "epoch": 3.43, + "learning_rate": 1.2765969491173023e-05, + "loss": 0.1827, + "step": 8186 + }, + { + "epoch": 3.43, + "learning_rate": 1.2764340187199393e-05, + "loss": 0.1434, + "step": 8187 + }, + { + "epoch": 3.43, + "learning_rate": 1.2762710803767258e-05, + "loss": 0.2034, + "step": 8188 + }, + { + "epoch": 3.43, + "learning_rate": 1.2761081340923455e-05, + "loss": 0.1645, + "step": 8189 + }, + { + "epoch": 3.43, + "learning_rate": 1.2759451798714816e-05, + "loss": 0.1894, + "step": 8190 + }, + { + "epoch": 3.43, + "learning_rate": 1.2757822177188187e-05, + "loss": 0.1838, + "step": 8191 + }, + { + "epoch": 3.43, + "learning_rate": 1.2756192476390406e-05, + "loss": 0.2094, + "step": 8192 + }, + { + "epoch": 3.43, + "learning_rate": 1.2754562696368318e-05, + "loss": 0.2115, + "step": 8193 + }, + { + "epoch": 3.43, + "learning_rate": 1.2752932837168768e-05, + "loss": 0.1835, + "step": 8194 + }, + { + "epoch": 3.43, + "learning_rate": 1.2751302898838607e-05, + "loss": 0.1596, + "step": 8195 + }, + { + "epoch": 3.43, + "learning_rate": 1.2749672881424687e-05, + "loss": 0.1958, + "step": 8196 + }, + { + "epoch": 3.43, + "learning_rate": 1.2748042784973857e-05, + "loss": 0.1959, + "step": 8197 + }, + { + "epoch": 3.43, + "learning_rate": 1.274641260953298e-05, + "loss": 0.2248, + "step": 8198 + }, + { + "epoch": 3.43, + "learning_rate": 1.2744782355148905e-05, + "loss": 0.2259, + "step": 8199 + }, + { + "epoch": 3.43, + "learning_rate": 1.27431520218685e-05, + "loss": 0.2132, + "step": 8200 + }, + { + "epoch": 3.43, + "learning_rate": 1.2741521609738623e-05, + "loss": 0.195, + "step": 8201 + }, + { + "epoch": 3.43, + "learning_rate": 1.273989111880614e-05, + "loss": 0.1968, + "step": 8202 + }, + { + "epoch": 3.43, + "learning_rate": 1.273826054911792e-05, + "loss": 0.2022, + "step": 8203 + }, + { + "epoch": 3.43, + "learning_rate": 1.2736629900720832e-05, + "loss": 0.2075, + "step": 8204 + }, + { + "epoch": 3.44, + "learning_rate": 1.2734999173661742e-05, + "loss": 0.173, + "step": 8205 + }, + { + "epoch": 3.44, + "learning_rate": 1.2733368367987528e-05, + "loss": 0.1718, + "step": 8206 + }, + { + "epoch": 3.44, + "learning_rate": 1.2731737483745068e-05, + "loss": 0.1859, + "step": 8207 + }, + { + "epoch": 3.44, + "learning_rate": 1.2730106520981236e-05, + "loss": 0.207, + "step": 8208 + }, + { + "epoch": 3.44, + "learning_rate": 1.2728475479742914e-05, + "loss": 0.1998, + "step": 8209 + }, + { + "epoch": 3.44, + "learning_rate": 1.272684436007699e-05, + "loss": 0.1945, + "step": 8210 + }, + { + "epoch": 3.44, + "learning_rate": 1.272521316203034e-05, + "loss": 0.1999, + "step": 8211 + }, + { + "epoch": 3.44, + "learning_rate": 1.2723581885649858e-05, + "loss": 0.1899, + "step": 8212 + }, + { + "epoch": 3.44, + "learning_rate": 1.2721950530982431e-05, + "loss": 0.1733, + "step": 8213 + }, + { + "epoch": 3.44, + "learning_rate": 1.2720319098074954e-05, + "loss": 0.2122, + "step": 8214 + }, + { + "epoch": 3.44, + "learning_rate": 1.2718687586974317e-05, + "loss": 0.1633, + "step": 8215 + }, + { + "epoch": 3.44, + "learning_rate": 1.2717055997727416e-05, + "loss": 0.1858, + "step": 8216 + }, + { + "epoch": 3.44, + "learning_rate": 1.2715424330381155e-05, + "loss": 0.1657, + "step": 8217 + }, + { + "epoch": 3.44, + "learning_rate": 1.2713792584982426e-05, + "loss": 0.2026, + "step": 8218 + }, + { + "epoch": 3.44, + "learning_rate": 1.2712160761578138e-05, + "loss": 0.1809, + "step": 8219 + }, + { + "epoch": 3.44, + "learning_rate": 1.27105288602152e-05, + "loss": 0.1778, + "step": 8220 + }, + { + "epoch": 3.44, + "learning_rate": 1.2708896880940513e-05, + "loss": 0.2055, + "step": 8221 + }, + { + "epoch": 3.44, + "learning_rate": 1.270726482380099e-05, + "loss": 0.211, + "step": 8222 + }, + { + "epoch": 3.44, + "learning_rate": 1.270563268884354e-05, + "loss": 0.1524, + "step": 8223 + }, + { + "epoch": 3.44, + "learning_rate": 1.2704000476115079e-05, + "loss": 0.1888, + "step": 8224 + }, + { + "epoch": 3.44, + "learning_rate": 1.2702368185662529e-05, + "loss": 0.1963, + "step": 8225 + }, + { + "epoch": 3.44, + "learning_rate": 1.2700735817532796e-05, + "loss": 0.1832, + "step": 8226 + }, + { + "epoch": 3.44, + "learning_rate": 1.2699103371772815e-05, + "loss": 0.1948, + "step": 8227 + }, + { + "epoch": 3.44, + "learning_rate": 1.26974708484295e-05, + "loss": 0.2054, + "step": 8228 + }, + { + "epoch": 3.45, + "learning_rate": 1.2695838247549777e-05, + "loss": 0.1572, + "step": 8229 + }, + { + "epoch": 3.45, + "learning_rate": 1.2694205569180577e-05, + "loss": 0.1622, + "step": 8230 + }, + { + "epoch": 3.45, + "learning_rate": 1.269257281336883e-05, + "loss": 0.1943, + "step": 8231 + }, + { + "epoch": 3.45, + "learning_rate": 1.2690939980161465e-05, + "loss": 0.1662, + "step": 8232 + }, + { + "epoch": 3.45, + "learning_rate": 1.2689307069605418e-05, + "loss": 0.1833, + "step": 8233 + }, + { + "epoch": 3.45, + "learning_rate": 1.268767408174763e-05, + "loss": 0.2034, + "step": 8234 + }, + { + "epoch": 3.45, + "learning_rate": 1.2686041016635031e-05, + "loss": 0.1898, + "step": 8235 + }, + { + "epoch": 3.45, + "learning_rate": 1.2684407874314568e-05, + "loss": 0.1917, + "step": 8236 + }, + { + "epoch": 3.45, + "learning_rate": 1.2682774654833188e-05, + "loss": 0.1471, + "step": 8237 + }, + { + "epoch": 3.45, + "learning_rate": 1.2681141358237823e-05, + "loss": 0.1666, + "step": 8238 + }, + { + "epoch": 3.45, + "learning_rate": 1.2679507984575431e-05, + "loss": 0.162, + "step": 8239 + }, + { + "epoch": 3.45, + "learning_rate": 1.2677874533892963e-05, + "loss": 0.18, + "step": 8240 + }, + { + "epoch": 3.45, + "learning_rate": 1.2676241006237367e-05, + "loss": 0.1862, + "step": 8241 + }, + { + "epoch": 3.45, + "learning_rate": 1.2674607401655595e-05, + "loss": 0.1574, + "step": 8242 + }, + { + "epoch": 3.45, + "learning_rate": 1.2672973720194608e-05, + "loss": 0.1919, + "step": 8243 + }, + { + "epoch": 3.45, + "learning_rate": 1.2671339961901366e-05, + "loss": 0.2196, + "step": 8244 + }, + { + "epoch": 3.45, + "learning_rate": 1.2669706126822823e-05, + "loss": 0.1901, + "step": 8245 + }, + { + "epoch": 3.45, + "learning_rate": 1.2668072215005946e-05, + "loss": 0.1999, + "step": 8246 + }, + { + "epoch": 3.45, + "learning_rate": 1.2666438226497706e-05, + "loss": 0.2016, + "step": 8247 + }, + { + "epoch": 3.45, + "learning_rate": 1.2664804161345064e-05, + "loss": 0.2066, + "step": 8248 + }, + { + "epoch": 3.45, + "learning_rate": 1.2663170019594988e-05, + "loss": 0.1709, + "step": 8249 + }, + { + "epoch": 3.45, + "learning_rate": 1.2661535801294455e-05, + "loss": 0.1545, + "step": 8250 + }, + { + "epoch": 3.45, + "learning_rate": 1.2659901506490437e-05, + "loss": 0.1295, + "step": 8251 + }, + { + "epoch": 3.45, + "learning_rate": 1.2658267135229911e-05, + "loss": 0.1584, + "step": 8252 + }, + { + "epoch": 3.46, + "learning_rate": 1.2656632687559853e-05, + "loss": 0.1862, + "step": 8253 + }, + { + "epoch": 3.46, + "learning_rate": 1.265499816352725e-05, + "loss": 0.1749, + "step": 8254 + }, + { + "epoch": 3.46, + "learning_rate": 1.2653363563179078e-05, + "loss": 0.1752, + "step": 8255 + }, + { + "epoch": 3.46, + "learning_rate": 1.2651728886562324e-05, + "loss": 0.178, + "step": 8256 + }, + { + "epoch": 3.46, + "learning_rate": 1.2650094133723978e-05, + "loss": 0.1874, + "step": 8257 + }, + { + "epoch": 3.46, + "learning_rate": 1.2648459304711026e-05, + "loss": 0.1988, + "step": 8258 + }, + { + "epoch": 3.46, + "learning_rate": 1.2646824399570466e-05, + "loss": 0.1609, + "step": 8259 + }, + { + "epoch": 3.46, + "learning_rate": 1.2645189418349282e-05, + "loss": 0.2107, + "step": 8260 + }, + { + "epoch": 3.46, + "learning_rate": 1.2643554361094478e-05, + "loss": 0.2183, + "step": 8261 + }, + { + "epoch": 3.46, + "learning_rate": 1.264191922785305e-05, + "loss": 0.201, + "step": 8262 + }, + { + "epoch": 3.46, + "learning_rate": 1.2640284018671995e-05, + "loss": 0.1965, + "step": 8263 + }, + { + "epoch": 3.46, + "learning_rate": 1.2638648733598325e-05, + "loss": 0.1527, + "step": 8264 + }, + { + "epoch": 3.46, + "learning_rate": 1.2637013372679032e-05, + "loss": 0.188, + "step": 8265 + }, + { + "epoch": 3.46, + "learning_rate": 1.2635377935961134e-05, + "loss": 0.1662, + "step": 8266 + }, + { + "epoch": 3.46, + "learning_rate": 1.2633742423491636e-05, + "loss": 0.1792, + "step": 8267 + }, + { + "epoch": 3.46, + "learning_rate": 1.2632106835317545e-05, + "loss": 0.1997, + "step": 8268 + }, + { + "epoch": 3.46, + "learning_rate": 1.263047117148588e-05, + "loss": 0.1857, + "step": 8269 + }, + { + "epoch": 3.46, + "learning_rate": 1.2628835432043655e-05, + "loss": 0.2021, + "step": 8270 + }, + { + "epoch": 3.46, + "learning_rate": 1.2627199617037895e-05, + "loss": 0.1827, + "step": 8271 + }, + { + "epoch": 3.46, + "learning_rate": 1.2625563726515609e-05, + "loss": 0.2168, + "step": 8272 + }, + { + "epoch": 3.46, + "learning_rate": 1.2623927760523822e-05, + "loss": 0.2037, + "step": 8273 + }, + { + "epoch": 3.46, + "learning_rate": 1.2622291719109559e-05, + "loss": 0.1773, + "step": 8274 + }, + { + "epoch": 3.46, + "learning_rate": 1.2620655602319851e-05, + "loss": 0.1975, + "step": 8275 + }, + { + "epoch": 3.46, + "learning_rate": 1.261901941020172e-05, + "loss": 0.1958, + "step": 8276 + }, + { + "epoch": 3.47, + "learning_rate": 1.2617383142802202e-05, + "loss": 0.2037, + "step": 8277 + }, + { + "epoch": 3.47, + "learning_rate": 1.261574680016833e-05, + "loss": 0.166, + "step": 8278 + }, + { + "epoch": 3.47, + "learning_rate": 1.2614110382347133e-05, + "loss": 0.2285, + "step": 8279 + }, + { + "epoch": 3.47, + "learning_rate": 1.2612473889385655e-05, + "loss": 0.1713, + "step": 8280 + }, + { + "epoch": 3.47, + "learning_rate": 1.2610837321330935e-05, + "loss": 0.1785, + "step": 8281 + }, + { + "epoch": 3.47, + "learning_rate": 1.2609200678230014e-05, + "loss": 0.1969, + "step": 8282 + }, + { + "epoch": 3.47, + "learning_rate": 1.260756396012993e-05, + "loss": 0.2023, + "step": 8283 + }, + { + "epoch": 3.47, + "learning_rate": 1.2605927167077736e-05, + "loss": 0.2079, + "step": 8284 + }, + { + "epoch": 3.47, + "learning_rate": 1.2604290299120478e-05, + "loss": 0.1478, + "step": 8285 + }, + { + "epoch": 3.47, + "learning_rate": 1.2602653356305207e-05, + "loss": 0.1633, + "step": 8286 + }, + { + "epoch": 3.47, + "learning_rate": 1.2601016338678976e-05, + "loss": 0.2344, + "step": 8287 + }, + { + "epoch": 3.47, + "learning_rate": 1.2599379246288837e-05, + "loss": 0.179, + "step": 8288 + }, + { + "epoch": 3.47, + "learning_rate": 1.2597742079181847e-05, + "loss": 0.2131, + "step": 8289 + }, + { + "epoch": 3.47, + "learning_rate": 1.2596104837405065e-05, + "loss": 0.1779, + "step": 8290 + }, + { + "epoch": 3.47, + "learning_rate": 1.2594467521005557e-05, + "loss": 0.203, + "step": 8291 + }, + { + "epoch": 3.47, + "learning_rate": 1.259283013003038e-05, + "loss": 0.1678, + "step": 8292 + }, + { + "epoch": 3.47, + "learning_rate": 1.2591192664526606e-05, + "loss": 0.1857, + "step": 8293 + }, + { + "epoch": 3.47, + "learning_rate": 1.2589555124541293e-05, + "loss": 0.1401, + "step": 8294 + }, + { + "epoch": 3.47, + "learning_rate": 1.2587917510121518e-05, + "loss": 0.1979, + "step": 8295 + }, + { + "epoch": 3.47, + "learning_rate": 1.2586279821314351e-05, + "loss": 0.1971, + "step": 8296 + }, + { + "epoch": 3.47, + "learning_rate": 1.2584642058166863e-05, + "loss": 0.1511, + "step": 8297 + }, + { + "epoch": 3.47, + "learning_rate": 1.2583004220726136e-05, + "loss": 0.1771, + "step": 8298 + }, + { + "epoch": 3.47, + "learning_rate": 1.2581366309039242e-05, + "loss": 0.1956, + "step": 8299 + }, + { + "epoch": 3.47, + "learning_rate": 1.2579728323153267e-05, + "loss": 0.1676, + "step": 8300 + }, + { + "epoch": 3.48, + "learning_rate": 1.257809026311529e-05, + "loss": 0.1678, + "step": 8301 + }, + { + "epoch": 3.48, + "learning_rate": 1.2576452128972394e-05, + "loss": 0.2065, + "step": 8302 + }, + { + "epoch": 3.48, + "learning_rate": 1.2574813920771667e-05, + "loss": 0.1917, + "step": 8303 + }, + { + "epoch": 3.48, + "learning_rate": 1.25731756385602e-05, + "loss": 0.1817, + "step": 8304 + }, + { + "epoch": 3.48, + "learning_rate": 1.2571537282385087e-05, + "loss": 0.2061, + "step": 8305 + }, + { + "epoch": 3.48, + "learning_rate": 1.2569898852293412e-05, + "loss": 0.1832, + "step": 8306 + }, + { + "epoch": 3.48, + "learning_rate": 1.2568260348332276e-05, + "loss": 0.1655, + "step": 8307 + }, + { + "epoch": 3.48, + "learning_rate": 1.2566621770548777e-05, + "loss": 0.1663, + "step": 8308 + }, + { + "epoch": 3.48, + "learning_rate": 1.256498311899001e-05, + "loss": 0.1934, + "step": 8309 + }, + { + "epoch": 3.48, + "learning_rate": 1.2563344393703081e-05, + "loss": 0.1964, + "step": 8310 + }, + { + "epoch": 3.48, + "learning_rate": 1.256170559473509e-05, + "loss": 0.1858, + "step": 8311 + }, + { + "epoch": 3.48, + "learning_rate": 1.2560066722133146e-05, + "loss": 0.1925, + "step": 8312 + }, + { + "epoch": 3.48, + "learning_rate": 1.2558427775944357e-05, + "loss": 0.2114, + "step": 8313 + }, + { + "epoch": 3.48, + "learning_rate": 1.255678875621583e-05, + "loss": 0.1507, + "step": 8314 + }, + { + "epoch": 3.48, + "learning_rate": 1.2555149662994684e-05, + "loss": 0.1966, + "step": 8315 + }, + { + "epoch": 3.48, + "learning_rate": 1.2553510496328025e-05, + "loss": 0.1924, + "step": 8316 + }, + { + "epoch": 3.48, + "learning_rate": 1.255187125626297e-05, + "loss": 0.1676, + "step": 8317 + }, + { + "epoch": 3.48, + "learning_rate": 1.2550231942846643e-05, + "loss": 0.1972, + "step": 8318 + }, + { + "epoch": 3.48, + "learning_rate": 1.2548592556126163e-05, + "loss": 0.1806, + "step": 8319 + }, + { + "epoch": 3.48, + "learning_rate": 1.2546953096148648e-05, + "loss": 0.18, + "step": 8320 + }, + { + "epoch": 3.48, + "learning_rate": 1.2545313562961228e-05, + "loss": 0.201, + "step": 8321 + }, + { + "epoch": 3.48, + "learning_rate": 1.2543673956611029e-05, + "loss": 0.1876, + "step": 8322 + }, + { + "epoch": 3.48, + "learning_rate": 1.2542034277145178e-05, + "loss": 0.1419, + "step": 8323 + }, + { + "epoch": 3.49, + "learning_rate": 1.2540394524610808e-05, + "loss": 0.1911, + "step": 8324 + }, + { + "epoch": 3.49, + "learning_rate": 1.2538754699055057e-05, + "loss": 0.1678, + "step": 8325 + }, + { + "epoch": 3.49, + "learning_rate": 1.2537114800525047e-05, + "loss": 0.2095, + "step": 8326 + }, + { + "epoch": 3.49, + "learning_rate": 1.2535474829067928e-05, + "loss": 0.2091, + "step": 8327 + }, + { + "epoch": 3.49, + "learning_rate": 1.2533834784730835e-05, + "loss": 0.191, + "step": 8328 + }, + { + "epoch": 3.49, + "learning_rate": 1.2532194667560908e-05, + "loss": 0.2376, + "step": 8329 + }, + { + "epoch": 3.49, + "learning_rate": 1.253055447760529e-05, + "loss": 0.1688, + "step": 8330 + }, + { + "epoch": 3.49, + "learning_rate": 1.2528914214911132e-05, + "loss": 0.2203, + "step": 8331 + }, + { + "epoch": 3.49, + "learning_rate": 1.2527273879525579e-05, + "loss": 0.1921, + "step": 8332 + }, + { + "epoch": 3.49, + "learning_rate": 1.2525633471495782e-05, + "loss": 0.1735, + "step": 8333 + }, + { + "epoch": 3.49, + "learning_rate": 1.252399299086889e-05, + "loss": 0.1715, + "step": 8334 + }, + { + "epoch": 3.49, + "learning_rate": 1.252235243769206e-05, + "loss": 0.2148, + "step": 8335 + }, + { + "epoch": 3.49, + "learning_rate": 1.2520711812012447e-05, + "loss": 0.1942, + "step": 8336 + }, + { + "epoch": 3.49, + "learning_rate": 1.2519071113877211e-05, + "loss": 0.2036, + "step": 8337 + }, + { + "epoch": 3.49, + "learning_rate": 1.2517430343333508e-05, + "loss": 0.1769, + "step": 8338 + }, + { + "epoch": 3.49, + "learning_rate": 1.251578950042851e-05, + "loss": 0.1718, + "step": 8339 + }, + { + "epoch": 3.49, + "learning_rate": 1.2514148585209372e-05, + "loss": 0.1998, + "step": 8340 + }, + { + "epoch": 3.49, + "learning_rate": 1.251250759772326e-05, + "loss": 0.1851, + "step": 8341 + }, + { + "epoch": 3.49, + "learning_rate": 1.2510866538017352e-05, + "loss": 0.2033, + "step": 8342 + }, + { + "epoch": 3.49, + "learning_rate": 1.2509225406138809e-05, + "loss": 0.1818, + "step": 8343 + }, + { + "epoch": 3.49, + "learning_rate": 1.2507584202134812e-05, + "loss": 0.2065, + "step": 8344 + }, + { + "epoch": 3.49, + "learning_rate": 1.250594292605253e-05, + "loss": 0.1722, + "step": 8345 + }, + { + "epoch": 3.49, + "learning_rate": 1.2504301577939146e-05, + "loss": 0.2298, + "step": 8346 + }, + { + "epoch": 3.49, + "learning_rate": 1.250266015784183e-05, + "loss": 0.2453, + "step": 8347 + }, + { + "epoch": 3.5, + "learning_rate": 1.2501018665807772e-05, + "loss": 0.1601, + "step": 8348 + }, + { + "epoch": 3.5, + "learning_rate": 1.2499377101884153e-05, + "loss": 0.1856, + "step": 8349 + }, + { + "epoch": 3.5, + "learning_rate": 1.2497735466118156e-05, + "loss": 0.224, + "step": 8350 + }, + { + "epoch": 3.5, + "learning_rate": 1.249609375855697e-05, + "loss": 0.2147, + "step": 8351 + }, + { + "epoch": 3.5, + "learning_rate": 1.2494451979247782e-05, + "loss": 0.2016, + "step": 8352 + }, + { + "epoch": 3.5, + "learning_rate": 1.2492810128237787e-05, + "loss": 0.1871, + "step": 8353 + }, + { + "epoch": 3.5, + "learning_rate": 1.2491168205574175e-05, + "loss": 0.2007, + "step": 8354 + }, + { + "epoch": 3.5, + "learning_rate": 1.2489526211304143e-05, + "loss": 0.1742, + "step": 8355 + }, + { + "epoch": 3.5, + "learning_rate": 1.2487884145474894e-05, + "loss": 0.1818, + "step": 8356 + }, + { + "epoch": 3.5, + "learning_rate": 1.2486242008133618e-05, + "loss": 0.1848, + "step": 8357 + }, + { + "epoch": 3.5, + "learning_rate": 1.2484599799327524e-05, + "loss": 0.1861, + "step": 8358 + }, + { + "epoch": 3.5, + "learning_rate": 1.2482957519103813e-05, + "loss": 0.1962, + "step": 8359 + }, + { + "epoch": 3.5, + "learning_rate": 1.2481315167509691e-05, + "loss": 0.163, + "step": 8360 + }, + { + "epoch": 3.5, + "learning_rate": 1.2479672744592368e-05, + "loss": 0.1704, + "step": 8361 + }, + { + "epoch": 3.5, + "learning_rate": 1.2478030250399052e-05, + "loss": 0.1996, + "step": 8362 + }, + { + "epoch": 3.5, + "learning_rate": 1.2476387684976956e-05, + "loss": 0.1665, + "step": 8363 + }, + { + "epoch": 3.5, + "learning_rate": 1.2474745048373292e-05, + "loss": 0.1742, + "step": 8364 + }, + { + "epoch": 3.5, + "learning_rate": 1.2473102340635275e-05, + "loss": 0.1708, + "step": 8365 + }, + { + "epoch": 3.5, + "learning_rate": 1.247145956181013e-05, + "loss": 0.2358, + "step": 8366 + }, + { + "epoch": 3.5, + "learning_rate": 1.246981671194507e-05, + "loss": 0.1767, + "step": 8367 + }, + { + "epoch": 3.5, + "learning_rate": 1.2468173791087324e-05, + "loss": 0.1535, + "step": 8368 + }, + { + "epoch": 3.5, + "learning_rate": 1.2466530799284111e-05, + "loss": 0.1793, + "step": 8369 + }, + { + "epoch": 3.5, + "learning_rate": 1.2464887736582659e-05, + "loss": 0.1774, + "step": 8370 + }, + { + "epoch": 3.5, + "learning_rate": 1.2463244603030194e-05, + "loss": 0.1839, + "step": 8371 + }, + { + "epoch": 3.51, + "learning_rate": 1.2461601398673953e-05, + "loss": 0.2015, + "step": 8372 + }, + { + "epoch": 3.51, + "learning_rate": 1.2459958123561163e-05, + "loss": 0.1921, + "step": 8373 + }, + { + "epoch": 3.51, + "learning_rate": 1.2458314777739056e-05, + "loss": 0.1948, + "step": 8374 + }, + { + "epoch": 3.51, + "learning_rate": 1.2456671361254876e-05, + "loss": 0.2294, + "step": 8375 + }, + { + "epoch": 3.51, + "learning_rate": 1.2455027874155856e-05, + "loss": 0.193, + "step": 8376 + }, + { + "epoch": 3.51, + "learning_rate": 1.2453384316489239e-05, + "loss": 0.1784, + "step": 8377 + }, + { + "epoch": 3.51, + "learning_rate": 1.2451740688302265e-05, + "loss": 0.1754, + "step": 8378 + }, + { + "epoch": 3.51, + "learning_rate": 1.2450096989642184e-05, + "loss": 0.1777, + "step": 8379 + }, + { + "epoch": 3.51, + "learning_rate": 1.2448453220556235e-05, + "loss": 0.1634, + "step": 8380 + }, + { + "epoch": 3.51, + "learning_rate": 1.2446809381091674e-05, + "loss": 0.1739, + "step": 8381 + }, + { + "epoch": 3.51, + "learning_rate": 1.2445165471295745e-05, + "loss": 0.1825, + "step": 8382 + }, + { + "epoch": 3.51, + "learning_rate": 1.244352149121571e-05, + "loss": 0.1394, + "step": 8383 + }, + { + "epoch": 3.51, + "learning_rate": 1.2441877440898817e-05, + "loss": 0.1721, + "step": 8384 + }, + { + "epoch": 3.51, + "learning_rate": 1.2440233320392322e-05, + "loss": 0.1992, + "step": 8385 + }, + { + "epoch": 3.51, + "learning_rate": 1.2438589129743484e-05, + "loss": 0.1969, + "step": 8386 + }, + { + "epoch": 3.51, + "learning_rate": 1.243694486899957e-05, + "loss": 0.214, + "step": 8387 + }, + { + "epoch": 3.51, + "learning_rate": 1.2435300538207832e-05, + "loss": 0.1699, + "step": 8388 + }, + { + "epoch": 3.51, + "learning_rate": 1.2433656137415545e-05, + "loss": 0.1888, + "step": 8389 + }, + { + "epoch": 3.51, + "learning_rate": 1.2432011666669972e-05, + "loss": 0.1743, + "step": 8390 + }, + { + "epoch": 3.51, + "learning_rate": 1.243036712601838e-05, + "loss": 0.1822, + "step": 8391 + }, + { + "epoch": 3.51, + "learning_rate": 1.2428722515508043e-05, + "loss": 0.1835, + "step": 8392 + }, + { + "epoch": 3.51, + "learning_rate": 1.2427077835186235e-05, + "loss": 0.2189, + "step": 8393 + }, + { + "epoch": 3.51, + "learning_rate": 1.2425433085100224e-05, + "loss": 0.1872, + "step": 8394 + }, + { + "epoch": 3.51, + "learning_rate": 1.2423788265297294e-05, + "loss": 0.2121, + "step": 8395 + }, + { + "epoch": 3.52, + "learning_rate": 1.242214337582472e-05, + "loss": 0.1972, + "step": 8396 + }, + { + "epoch": 3.52, + "learning_rate": 1.2420498416729785e-05, + "loss": 0.2004, + "step": 8397 + }, + { + "epoch": 3.52, + "learning_rate": 1.2418853388059771e-05, + "loss": 0.1933, + "step": 8398 + }, + { + "epoch": 3.52, + "learning_rate": 1.241720828986196e-05, + "loss": 0.18, + "step": 8399 + }, + { + "epoch": 3.52, + "learning_rate": 1.2415563122183644e-05, + "loss": 0.2336, + "step": 8400 + }, + { + "epoch": 3.52, + "learning_rate": 1.2413917885072109e-05, + "loss": 0.1966, + "step": 8401 + }, + { + "epoch": 3.52, + "learning_rate": 1.2412272578574647e-05, + "loss": 0.2009, + "step": 8402 + }, + { + "epoch": 3.52, + "learning_rate": 1.241062720273855e-05, + "loss": 0.1812, + "step": 8403 + }, + { + "epoch": 3.52, + "learning_rate": 1.2408981757611112e-05, + "loss": 0.2079, + "step": 8404 + }, + { + "epoch": 3.52, + "learning_rate": 1.2407336243239632e-05, + "loss": 0.1831, + "step": 8405 + }, + { + "epoch": 3.52, + "learning_rate": 1.240569065967141e-05, + "loss": 0.2116, + "step": 8406 + }, + { + "epoch": 3.52, + "learning_rate": 1.2404045006953744e-05, + "loss": 0.1673, + "step": 8407 + }, + { + "epoch": 3.52, + "learning_rate": 1.2402399285133935e-05, + "loss": 0.2214, + "step": 8408 + }, + { + "epoch": 3.52, + "learning_rate": 1.2400753494259293e-05, + "loss": 0.1845, + "step": 8409 + }, + { + "epoch": 3.52, + "learning_rate": 1.2399107634377122e-05, + "loss": 0.1506, + "step": 8410 + }, + { + "epoch": 3.52, + "learning_rate": 1.2397461705534727e-05, + "loss": 0.1963, + "step": 8411 + }, + { + "epoch": 3.52, + "learning_rate": 1.2395815707779426e-05, + "loss": 0.2047, + "step": 8412 + }, + { + "epoch": 3.52, + "learning_rate": 1.239416964115853e-05, + "loss": 0.2205, + "step": 8413 + }, + { + "epoch": 3.52, + "learning_rate": 1.2392523505719349e-05, + "loss": 0.1948, + "step": 8414 + }, + { + "epoch": 3.52, + "learning_rate": 1.2390877301509205e-05, + "loss": 0.2205, + "step": 8415 + }, + { + "epoch": 3.52, + "learning_rate": 1.2389231028575412e-05, + "loss": 0.1596, + "step": 8416 + }, + { + "epoch": 3.52, + "learning_rate": 1.2387584686965297e-05, + "loss": 0.2314, + "step": 8417 + }, + { + "epoch": 3.52, + "learning_rate": 1.2385938276726179e-05, + "loss": 0.1764, + "step": 8418 + }, + { + "epoch": 3.52, + "learning_rate": 1.2384291797905382e-05, + "loss": 0.1856, + "step": 8419 + }, + { + "epoch": 3.53, + "learning_rate": 1.2382645250550232e-05, + "loss": 0.2062, + "step": 8420 + }, + { + "epoch": 3.53, + "learning_rate": 1.2380998634708057e-05, + "loss": 0.2007, + "step": 8421 + }, + { + "epoch": 3.53, + "learning_rate": 1.2379351950426188e-05, + "loss": 0.2188, + "step": 8422 + }, + { + "epoch": 3.53, + "learning_rate": 1.2377705197751961e-05, + "loss": 0.212, + "step": 8423 + }, + { + "epoch": 3.53, + "learning_rate": 1.237605837673271e-05, + "loss": 0.2021, + "step": 8424 + }, + { + "epoch": 3.53, + "learning_rate": 1.2374411487415766e-05, + "loss": 0.1813, + "step": 8425 + }, + { + "epoch": 3.53, + "learning_rate": 1.2372764529848471e-05, + "loss": 0.1811, + "step": 8426 + }, + { + "epoch": 3.53, + "learning_rate": 1.237111750407817e-05, + "loss": 0.1925, + "step": 8427 + }, + { + "epoch": 3.53, + "learning_rate": 1.2369470410152195e-05, + "loss": 0.1922, + "step": 8428 + }, + { + "epoch": 3.53, + "learning_rate": 1.2367823248117899e-05, + "loss": 0.2002, + "step": 8429 + }, + { + "epoch": 3.53, + "learning_rate": 1.2366176018022621e-05, + "loss": 0.2103, + "step": 8430 + }, + { + "epoch": 3.53, + "learning_rate": 1.2364528719913717e-05, + "loss": 0.1977, + "step": 8431 + }, + { + "epoch": 3.53, + "learning_rate": 1.2362881353838532e-05, + "loss": 0.1977, + "step": 8432 + }, + { + "epoch": 3.53, + "learning_rate": 1.236123391984442e-05, + "loss": 0.1646, + "step": 8433 + }, + { + "epoch": 3.53, + "learning_rate": 1.2359586417978733e-05, + "loss": 0.1623, + "step": 8434 + }, + { + "epoch": 3.53, + "learning_rate": 1.2357938848288829e-05, + "loss": 0.2135, + "step": 8435 + }, + { + "epoch": 3.53, + "learning_rate": 1.2356291210822063e-05, + "loss": 0.2054, + "step": 8436 + }, + { + "epoch": 3.53, + "learning_rate": 1.23546435056258e-05, + "loss": 0.1851, + "step": 8437 + }, + { + "epoch": 3.53, + "learning_rate": 1.2352995732747398e-05, + "loss": 0.2118, + "step": 8438 + }, + { + "epoch": 3.53, + "learning_rate": 1.235134789223422e-05, + "loss": 0.2034, + "step": 8439 + }, + { + "epoch": 3.53, + "learning_rate": 1.2349699984133636e-05, + "loss": 0.1801, + "step": 8440 + }, + { + "epoch": 3.53, + "learning_rate": 1.234805200849301e-05, + "loss": 0.1968, + "step": 8441 + }, + { + "epoch": 3.53, + "learning_rate": 1.2346403965359712e-05, + "loss": 0.1928, + "step": 8442 + }, + { + "epoch": 3.53, + "learning_rate": 1.2344755854781114e-05, + "loss": 0.1793, + "step": 8443 + }, + { + "epoch": 3.54, + "learning_rate": 1.2343107676804593e-05, + "loss": 0.1609, + "step": 8444 + }, + { + "epoch": 3.54, + "learning_rate": 1.2341459431477516e-05, + "loss": 0.1669, + "step": 8445 + }, + { + "epoch": 3.54, + "learning_rate": 1.2339811118847269e-05, + "loss": 0.1921, + "step": 8446 + }, + { + "epoch": 3.54, + "learning_rate": 1.2338162738961227e-05, + "loss": 0.1854, + "step": 8447 + }, + { + "epoch": 3.54, + "learning_rate": 1.233651429186677e-05, + "loss": 0.1917, + "step": 8448 + }, + { + "epoch": 3.54, + "learning_rate": 1.2334865777611284e-05, + "loss": 0.1943, + "step": 8449 + }, + { + "epoch": 3.54, + "learning_rate": 1.2333217196242152e-05, + "loss": 0.1806, + "step": 8450 + }, + { + "epoch": 3.54, + "learning_rate": 1.2331568547806768e-05, + "loss": 0.1831, + "step": 8451 + }, + { + "epoch": 3.54, + "learning_rate": 1.232991983235251e-05, + "loss": 0.2109, + "step": 8452 + }, + { + "epoch": 3.54, + "learning_rate": 1.2328271049926774e-05, + "loss": 0.213, + "step": 8453 + }, + { + "epoch": 3.54, + "learning_rate": 1.2326622200576956e-05, + "loss": 0.1713, + "step": 8454 + }, + { + "epoch": 3.54, + "learning_rate": 1.2324973284350444e-05, + "loss": 0.2101, + "step": 8455 + }, + { + "epoch": 3.54, + "learning_rate": 1.2323324301294637e-05, + "loss": 0.1893, + "step": 8456 + }, + { + "epoch": 3.54, + "learning_rate": 1.2321675251456937e-05, + "loss": 0.1919, + "step": 8457 + }, + { + "epoch": 3.54, + "learning_rate": 1.2320026134884744e-05, + "loss": 0.1883, + "step": 8458 + }, + { + "epoch": 3.54, + "learning_rate": 1.2318376951625455e-05, + "loss": 0.1351, + "step": 8459 + }, + { + "epoch": 3.54, + "learning_rate": 1.2316727701726479e-05, + "loss": 0.2197, + "step": 8460 + }, + { + "epoch": 3.54, + "learning_rate": 1.2315078385235223e-05, + "loss": 0.1724, + "step": 8461 + }, + { + "epoch": 3.54, + "learning_rate": 1.2313429002199088e-05, + "loss": 0.1657, + "step": 8462 + }, + { + "epoch": 3.54, + "learning_rate": 1.2311779552665495e-05, + "loss": 0.1867, + "step": 8463 + }, + { + "epoch": 3.54, + "learning_rate": 1.2310130036681847e-05, + "loss": 0.1593, + "step": 8464 + }, + { + "epoch": 3.54, + "learning_rate": 1.2308480454295563e-05, + "loss": 0.1857, + "step": 8465 + }, + { + "epoch": 3.54, + "learning_rate": 1.2306830805554055e-05, + "loss": 0.1832, + "step": 8466 + }, + { + "epoch": 3.54, + "learning_rate": 1.2305181090504744e-05, + "loss": 0.1751, + "step": 8467 + }, + { + "epoch": 3.55, + "learning_rate": 1.2303531309195047e-05, + "loss": 0.178, + "step": 8468 + }, + { + "epoch": 3.55, + "learning_rate": 1.2301881461672387e-05, + "loss": 0.1923, + "step": 8469 + }, + { + "epoch": 3.55, + "learning_rate": 1.2300231547984187e-05, + "loss": 0.1603, + "step": 8470 + }, + { + "epoch": 3.55, + "learning_rate": 1.2298581568177874e-05, + "loss": 0.2233, + "step": 8471 + }, + { + "epoch": 3.55, + "learning_rate": 1.229693152230087e-05, + "loss": 0.1936, + "step": 8472 + }, + { + "epoch": 3.55, + "learning_rate": 1.2295281410400611e-05, + "loss": 0.198, + "step": 8473 + }, + { + "epoch": 3.55, + "learning_rate": 1.2293631232524524e-05, + "loss": 0.1754, + "step": 8474 + }, + { + "epoch": 3.55, + "learning_rate": 1.2291980988720042e-05, + "loss": 0.156, + "step": 8475 + }, + { + "epoch": 3.55, + "learning_rate": 1.22903306790346e-05, + "loss": 0.192, + "step": 8476 + }, + { + "epoch": 3.55, + "learning_rate": 1.2288680303515634e-05, + "loss": 0.1959, + "step": 8477 + }, + { + "epoch": 3.55, + "learning_rate": 1.2287029862210587e-05, + "loss": 0.2003, + "step": 8478 + }, + { + "epoch": 3.55, + "learning_rate": 1.2285379355166893e-05, + "loss": 0.1684, + "step": 8479 + }, + { + "epoch": 3.55, + "learning_rate": 1.2283728782431996e-05, + "loss": 0.2383, + "step": 8480 + }, + { + "epoch": 3.55, + "learning_rate": 1.2282078144053347e-05, + "loss": 0.212, + "step": 8481 + }, + { + "epoch": 3.55, + "learning_rate": 1.2280427440078383e-05, + "loss": 0.1748, + "step": 8482 + }, + { + "epoch": 3.55, + "learning_rate": 1.2278776670554555e-05, + "loss": 0.1972, + "step": 8483 + }, + { + "epoch": 3.55, + "learning_rate": 1.2277125835529318e-05, + "loss": 0.1675, + "step": 8484 + }, + { + "epoch": 3.55, + "learning_rate": 1.2275474935050116e-05, + "loss": 0.184, + "step": 8485 + }, + { + "epoch": 3.55, + "learning_rate": 1.227382396916441e-05, + "loss": 0.1557, + "step": 8486 + }, + { + "epoch": 3.55, + "learning_rate": 1.2272172937919646e-05, + "loss": 0.2133, + "step": 8487 + }, + { + "epoch": 3.55, + "learning_rate": 1.2270521841363293e-05, + "loss": 0.2386, + "step": 8488 + }, + { + "epoch": 3.55, + "learning_rate": 1.2268870679542799e-05, + "loss": 0.1974, + "step": 8489 + }, + { + "epoch": 3.55, + "learning_rate": 1.2267219452505635e-05, + "loss": 0.1825, + "step": 8490 + }, + { + "epoch": 3.55, + "learning_rate": 1.226556816029926e-05, + "loss": 0.2028, + "step": 8491 + }, + { + "epoch": 3.56, + "learning_rate": 1.2263916802971134e-05, + "loss": 0.1763, + "step": 8492 + }, + { + "epoch": 3.56, + "learning_rate": 1.2262265380568729e-05, + "loss": 0.1288, + "step": 8493 + }, + { + "epoch": 3.56, + "learning_rate": 1.2260613893139517e-05, + "loss": 0.1676, + "step": 8494 + }, + { + "epoch": 3.56, + "learning_rate": 1.2258962340730965e-05, + "loss": 0.173, + "step": 8495 + }, + { + "epoch": 3.56, + "learning_rate": 1.2257310723390541e-05, + "loss": 0.1756, + "step": 8496 + }, + { + "epoch": 3.56, + "learning_rate": 1.2255659041165726e-05, + "loss": 0.1826, + "step": 8497 + }, + { + "epoch": 3.56, + "learning_rate": 1.2254007294103994e-05, + "loss": 0.2019, + "step": 8498 + }, + { + "epoch": 3.56, + "learning_rate": 1.2252355482252823e-05, + "loss": 0.2117, + "step": 8499 + }, + { + "epoch": 3.56, + "learning_rate": 1.2250703605659688e-05, + "loss": 0.1707, + "step": 8500 + }, + { + "epoch": 3.56, + "learning_rate": 1.2249051664372078e-05, + "loss": 0.1901, + "step": 8501 + }, + { + "epoch": 3.56, + "learning_rate": 1.2247399658437475e-05, + "loss": 0.2087, + "step": 8502 + }, + { + "epoch": 3.56, + "learning_rate": 1.224574758790336e-05, + "loss": 0.2222, + "step": 8503 + }, + { + "epoch": 3.56, + "learning_rate": 1.2244095452817227e-05, + "loss": 0.208, + "step": 8504 + }, + { + "epoch": 3.56, + "learning_rate": 1.2242443253226558e-05, + "loss": 0.2085, + "step": 8505 + }, + { + "epoch": 3.56, + "learning_rate": 1.224079098917885e-05, + "loss": 0.2235, + "step": 8506 + }, + { + "epoch": 3.56, + "learning_rate": 1.2239138660721593e-05, + "loss": 0.1747, + "step": 8507 + }, + { + "epoch": 3.56, + "learning_rate": 1.2237486267902284e-05, + "loss": 0.1916, + "step": 8508 + }, + { + "epoch": 3.56, + "learning_rate": 1.2235833810768417e-05, + "loss": 0.1648, + "step": 8509 + }, + { + "epoch": 3.56, + "learning_rate": 1.2234181289367489e-05, + "loss": 0.2035, + "step": 8510 + }, + { + "epoch": 3.56, + "learning_rate": 1.2232528703747002e-05, + "loss": 0.2268, + "step": 8511 + }, + { + "epoch": 3.56, + "learning_rate": 1.2230876053954462e-05, + "loss": 0.1868, + "step": 8512 + }, + { + "epoch": 3.56, + "learning_rate": 1.2229223340037364e-05, + "loss": 0.1953, + "step": 8513 + }, + { + "epoch": 3.56, + "learning_rate": 1.2227570562043223e-05, + "loss": 0.1992, + "step": 8514 + }, + { + "epoch": 3.56, + "learning_rate": 1.2225917720019544e-05, + "loss": 0.1824, + "step": 8515 + }, + { + "epoch": 3.57, + "learning_rate": 1.2224264814013833e-05, + "loss": 0.1768, + "step": 8516 + }, + { + "epoch": 3.57, + "learning_rate": 1.2222611844073604e-05, + "loss": 0.185, + "step": 8517 + }, + { + "epoch": 3.57, + "learning_rate": 1.2220958810246371e-05, + "loss": 0.1673, + "step": 8518 + }, + { + "epoch": 3.57, + "learning_rate": 1.2219305712579647e-05, + "loss": 0.1732, + "step": 8519 + }, + { + "epoch": 3.57, + "learning_rate": 1.221765255112095e-05, + "loss": 0.2109, + "step": 8520 + }, + { + "epoch": 3.57, + "learning_rate": 1.2215999325917797e-05, + "loss": 0.2072, + "step": 8521 + }, + { + "epoch": 3.57, + "learning_rate": 1.2214346037017715e-05, + "loss": 0.2434, + "step": 8522 + }, + { + "epoch": 3.57, + "learning_rate": 1.2212692684468215e-05, + "loss": 0.1588, + "step": 8523 + }, + { + "epoch": 3.57, + "learning_rate": 1.221103926831683e-05, + "loss": 0.2165, + "step": 8524 + }, + { + "epoch": 3.57, + "learning_rate": 1.2209385788611084e-05, + "loss": 0.2039, + "step": 8525 + }, + { + "epoch": 3.57, + "learning_rate": 1.2207732245398502e-05, + "loss": 0.1687, + "step": 8526 + }, + { + "epoch": 3.57, + "learning_rate": 1.2206078638726615e-05, + "loss": 0.1956, + "step": 8527 + }, + { + "epoch": 3.57, + "learning_rate": 1.2204424968642957e-05, + "loss": 0.2377, + "step": 8528 + }, + { + "epoch": 3.57, + "learning_rate": 1.220277123519506e-05, + "loss": 0.2237, + "step": 8529 + }, + { + "epoch": 3.57, + "learning_rate": 1.2201117438430456e-05, + "loss": 0.1893, + "step": 8530 + }, + { + "epoch": 3.57, + "learning_rate": 1.2199463578396688e-05, + "loss": 0.1917, + "step": 8531 + }, + { + "epoch": 3.57, + "learning_rate": 1.219780965514129e-05, + "loss": 0.2059, + "step": 8532 + }, + { + "epoch": 3.57, + "learning_rate": 1.2196155668711804e-05, + "loss": 0.1924, + "step": 8533 + }, + { + "epoch": 3.57, + "learning_rate": 1.219450161915577e-05, + "loss": 0.1509, + "step": 8534 + }, + { + "epoch": 3.57, + "learning_rate": 1.2192847506520735e-05, + "loss": 0.1846, + "step": 8535 + }, + { + "epoch": 3.57, + "learning_rate": 1.2191193330854246e-05, + "loss": 0.1921, + "step": 8536 + }, + { + "epoch": 3.57, + "learning_rate": 1.2189539092203849e-05, + "loss": 0.1988, + "step": 8537 + }, + { + "epoch": 3.57, + "learning_rate": 1.218788479061709e-05, + "loss": 0.1842, + "step": 8538 + }, + { + "epoch": 3.58, + "learning_rate": 1.2186230426141528e-05, + "loss": 0.1653, + "step": 8539 + }, + { + "epoch": 3.58, + "learning_rate": 1.2184575998824713e-05, + "loss": 0.1869, + "step": 8540 + }, + { + "epoch": 3.58, + "learning_rate": 1.2182921508714196e-05, + "loss": 0.1924, + "step": 8541 + }, + { + "epoch": 3.58, + "learning_rate": 1.218126695585754e-05, + "loss": 0.1826, + "step": 8542 + }, + { + "epoch": 3.58, + "learning_rate": 1.2179612340302302e-05, + "loss": 0.1999, + "step": 8543 + }, + { + "epoch": 3.58, + "learning_rate": 1.2177957662096038e-05, + "loss": 0.217, + "step": 8544 + }, + { + "epoch": 3.58, + "learning_rate": 1.2176302921286316e-05, + "loss": 0.2095, + "step": 8545 + }, + { + "epoch": 3.58, + "learning_rate": 1.2174648117920698e-05, + "loss": 0.202, + "step": 8546 + }, + { + "epoch": 3.58, + "learning_rate": 1.2172993252046747e-05, + "loss": 0.1597, + "step": 8547 + }, + { + "epoch": 3.58, + "learning_rate": 1.2171338323712034e-05, + "loss": 0.1915, + "step": 8548 + }, + { + "epoch": 3.58, + "learning_rate": 1.216968333296413e-05, + "loss": 0.1843, + "step": 8549 + }, + { + "epoch": 3.58, + "learning_rate": 1.2168028279850603e-05, + "loss": 0.2172, + "step": 8550 + }, + { + "epoch": 3.58, + "learning_rate": 1.2166373164419026e-05, + "loss": 0.1981, + "step": 8551 + }, + { + "epoch": 3.58, + "learning_rate": 1.2164717986716977e-05, + "loss": 0.1738, + "step": 8552 + }, + { + "epoch": 3.58, + "learning_rate": 1.2163062746792027e-05, + "loss": 0.1915, + "step": 8553 + }, + { + "epoch": 3.58, + "learning_rate": 1.2161407444691761e-05, + "loss": 0.1887, + "step": 8554 + }, + { + "epoch": 3.58, + "learning_rate": 1.2159752080463753e-05, + "loss": 0.1902, + "step": 8555 + }, + { + "epoch": 3.58, + "learning_rate": 1.2158096654155592e-05, + "loss": 0.1652, + "step": 8556 + }, + { + "epoch": 3.58, + "learning_rate": 1.2156441165814855e-05, + "loss": 0.187, + "step": 8557 + }, + { + "epoch": 3.58, + "learning_rate": 1.2154785615489129e-05, + "loss": 0.1937, + "step": 8558 + }, + { + "epoch": 3.58, + "learning_rate": 1.2153130003226006e-05, + "loss": 0.1814, + "step": 8559 + }, + { + "epoch": 3.58, + "learning_rate": 1.215147432907307e-05, + "loss": 0.1693, + "step": 8560 + }, + { + "epoch": 3.58, + "learning_rate": 1.2149818593077912e-05, + "loss": 0.1944, + "step": 8561 + }, + { + "epoch": 3.58, + "learning_rate": 1.2148162795288127e-05, + "loss": 0.1962, + "step": 8562 + }, + { + "epoch": 3.59, + "learning_rate": 1.214650693575131e-05, + "loss": 0.1754, + "step": 8563 + }, + { + "epoch": 3.59, + "learning_rate": 1.2144851014515055e-05, + "loss": 0.1715, + "step": 8564 + }, + { + "epoch": 3.59, + "learning_rate": 1.2143195031626962e-05, + "loss": 0.1796, + "step": 8565 + }, + { + "epoch": 3.59, + "learning_rate": 1.214153898713463e-05, + "loss": 0.1753, + "step": 8566 + }, + { + "epoch": 3.59, + "learning_rate": 1.2139882881085658e-05, + "loss": 0.1809, + "step": 8567 + }, + { + "epoch": 3.59, + "learning_rate": 1.2138226713527652e-05, + "loss": 0.1781, + "step": 8568 + }, + { + "epoch": 3.59, + "learning_rate": 1.2136570484508217e-05, + "loss": 0.2059, + "step": 8569 + }, + { + "epoch": 3.59, + "learning_rate": 1.213491419407496e-05, + "loss": 0.1986, + "step": 8570 + }, + { + "epoch": 3.59, + "learning_rate": 1.2133257842275487e-05, + "loss": 0.193, + "step": 8571 + }, + { + "epoch": 3.59, + "learning_rate": 1.2131601429157413e-05, + "loss": 0.2063, + "step": 8572 + }, + { + "epoch": 3.59, + "learning_rate": 1.2129944954768346e-05, + "loss": 0.1816, + "step": 8573 + }, + { + "epoch": 3.59, + "learning_rate": 1.2128288419155903e-05, + "loss": 0.2111, + "step": 8574 + }, + { + "epoch": 3.59, + "learning_rate": 1.2126631822367693e-05, + "loss": 0.1456, + "step": 8575 + }, + { + "epoch": 3.59, + "learning_rate": 1.2124975164451345e-05, + "loss": 0.1482, + "step": 8576 + }, + { + "epoch": 3.59, + "learning_rate": 1.212331844545447e-05, + "loss": 0.2008, + "step": 8577 + }, + { + "epoch": 3.59, + "learning_rate": 1.2121661665424688e-05, + "loss": 0.176, + "step": 8578 + }, + { + "epoch": 3.59, + "learning_rate": 1.2120004824409626e-05, + "loss": 0.1779, + "step": 8579 + }, + { + "epoch": 3.59, + "learning_rate": 1.2118347922456905e-05, + "loss": 0.162, + "step": 8580 + }, + { + "epoch": 3.59, + "learning_rate": 1.2116690959614154e-05, + "loss": 0.1954, + "step": 8581 + }, + { + "epoch": 3.59, + "learning_rate": 1.2115033935929e-05, + "loss": 0.183, + "step": 8582 + }, + { + "epoch": 3.59, + "learning_rate": 1.211337685144907e-05, + "loss": 0.1416, + "step": 8583 + }, + { + "epoch": 3.59, + "learning_rate": 1.2111719706222e-05, + "loss": 0.2038, + "step": 8584 + }, + { + "epoch": 3.59, + "learning_rate": 1.2110062500295418e-05, + "loss": 0.1634, + "step": 8585 + }, + { + "epoch": 3.59, + "learning_rate": 1.2108405233716967e-05, + "loss": 0.1703, + "step": 8586 + }, + { + "epoch": 3.6, + "learning_rate": 1.2106747906534275e-05, + "loss": 0.193, + "step": 8587 + }, + { + "epoch": 3.6, + "learning_rate": 1.2105090518794986e-05, + "loss": 0.1664, + "step": 8588 + }, + { + "epoch": 3.6, + "learning_rate": 1.2103433070546737e-05, + "loss": 0.1917, + "step": 8589 + }, + { + "epoch": 3.6, + "learning_rate": 1.2101775561837172e-05, + "loss": 0.1935, + "step": 8590 + }, + { + "epoch": 3.6, + "learning_rate": 1.2100117992713931e-05, + "loss": 0.1685, + "step": 8591 + }, + { + "epoch": 3.6, + "learning_rate": 1.2098460363224662e-05, + "loss": 0.1723, + "step": 8592 + }, + { + "epoch": 3.6, + "learning_rate": 1.2096802673417014e-05, + "loss": 0.1534, + "step": 8593 + }, + { + "epoch": 3.6, + "learning_rate": 1.2095144923338633e-05, + "loss": 0.1846, + "step": 8594 + }, + { + "epoch": 3.6, + "learning_rate": 1.209348711303717e-05, + "loss": 0.217, + "step": 8595 + }, + { + "epoch": 3.6, + "learning_rate": 1.2091829242560278e-05, + "loss": 0.2127, + "step": 8596 + }, + { + "epoch": 3.6, + "learning_rate": 1.209017131195561e-05, + "loss": 0.182, + "step": 8597 + }, + { + "epoch": 3.6, + "learning_rate": 1.2088513321270823e-05, + "loss": 0.2203, + "step": 8598 + }, + { + "epoch": 3.6, + "learning_rate": 1.2086855270553575e-05, + "loss": 0.2172, + "step": 8599 + }, + { + "epoch": 3.6, + "learning_rate": 1.2085197159851525e-05, + "loss": 0.2137, + "step": 8600 + }, + { + "epoch": 3.6, + "learning_rate": 1.208353898921233e-05, + "loss": 0.1924, + "step": 8601 + }, + { + "epoch": 3.6, + "learning_rate": 1.2081880758683656e-05, + "loss": 0.22, + "step": 8602 + }, + { + "epoch": 3.6, + "learning_rate": 1.208022246831317e-05, + "loss": 0.2001, + "step": 8603 + }, + { + "epoch": 3.6, + "learning_rate": 1.2078564118148531e-05, + "loss": 0.2161, + "step": 8604 + }, + { + "epoch": 3.6, + "learning_rate": 1.2076905708237412e-05, + "loss": 0.19, + "step": 8605 + }, + { + "epoch": 3.6, + "learning_rate": 1.2075247238627482e-05, + "loss": 0.1804, + "step": 8606 + }, + { + "epoch": 3.6, + "learning_rate": 1.2073588709366413e-05, + "loss": 0.2071, + "step": 8607 + }, + { + "epoch": 3.6, + "learning_rate": 1.2071930120501874e-05, + "loss": 0.2102, + "step": 8608 + }, + { + "epoch": 3.6, + "learning_rate": 1.2070271472081542e-05, + "loss": 0.1457, + "step": 8609 + }, + { + "epoch": 3.6, + "learning_rate": 1.2068612764153099e-05, + "loss": 0.2246, + "step": 8610 + }, + { + "epoch": 3.61, + "learning_rate": 1.2066953996764215e-05, + "loss": 0.1485, + "step": 8611 + }, + { + "epoch": 3.61, + "learning_rate": 1.2065295169962573e-05, + "loss": 0.1568, + "step": 8612 + }, + { + "epoch": 3.61, + "learning_rate": 1.206363628379585e-05, + "loss": 0.1886, + "step": 8613 + }, + { + "epoch": 3.61, + "learning_rate": 1.2061977338311737e-05, + "loss": 0.1763, + "step": 8614 + }, + { + "epoch": 3.61, + "learning_rate": 1.2060318333557915e-05, + "loss": 0.166, + "step": 8615 + }, + { + "epoch": 3.61, + "learning_rate": 1.2058659269582067e-05, + "loss": 0.1715, + "step": 8616 + }, + { + "epoch": 3.61, + "learning_rate": 1.205700014643189e-05, + "loss": 0.2064, + "step": 8617 + }, + { + "epoch": 3.61, + "learning_rate": 1.2055340964155067e-05, + "loss": 0.2055, + "step": 8618 + }, + { + "epoch": 3.61, + "learning_rate": 1.2053681722799291e-05, + "loss": 0.2265, + "step": 8619 + }, + { + "epoch": 3.61, + "learning_rate": 1.2052022422412259e-05, + "loss": 0.2, + "step": 8620 + }, + { + "epoch": 3.61, + "learning_rate": 1.2050363063041662e-05, + "loss": 0.1541, + "step": 8621 + }, + { + "epoch": 3.61, + "learning_rate": 1.2048703644735198e-05, + "loss": 0.1835, + "step": 8622 + }, + { + "epoch": 3.61, + "learning_rate": 1.2047044167540564e-05, + "loss": 0.1863, + "step": 8623 + }, + { + "epoch": 3.61, + "learning_rate": 1.2045384631505466e-05, + "loss": 0.1417, + "step": 8624 + }, + { + "epoch": 3.61, + "learning_rate": 1.2043725036677599e-05, + "loss": 0.1641, + "step": 8625 + }, + { + "epoch": 3.61, + "learning_rate": 1.204206538310467e-05, + "loss": 0.1823, + "step": 8626 + }, + { + "epoch": 3.61, + "learning_rate": 1.2040405670834385e-05, + "loss": 0.2046, + "step": 8627 + }, + { + "epoch": 3.61, + "learning_rate": 1.2038745899914446e-05, + "loss": 0.1768, + "step": 8628 + }, + { + "epoch": 3.61, + "learning_rate": 1.2037086070392567e-05, + "loss": 0.1889, + "step": 8629 + }, + { + "epoch": 3.61, + "learning_rate": 1.2035426182316459e-05, + "loss": 0.2021, + "step": 8630 + }, + { + "epoch": 3.61, + "learning_rate": 1.2033766235733828e-05, + "loss": 0.2051, + "step": 8631 + }, + { + "epoch": 3.61, + "learning_rate": 1.2032106230692394e-05, + "loss": 0.2008, + "step": 8632 + }, + { + "epoch": 3.61, + "learning_rate": 1.2030446167239869e-05, + "loss": 0.171, + "step": 8633 + }, + { + "epoch": 3.61, + "learning_rate": 1.202878604542397e-05, + "loss": 0.2133, + "step": 8634 + }, + { + "epoch": 3.62, + "learning_rate": 1.2027125865292417e-05, + "loss": 0.1647, + "step": 8635 + }, + { + "epoch": 3.62, + "learning_rate": 1.202546562689293e-05, + "loss": 0.1962, + "step": 8636 + }, + { + "epoch": 3.62, + "learning_rate": 1.202380533027323e-05, + "loss": 0.2258, + "step": 8637 + }, + { + "epoch": 3.62, + "learning_rate": 1.2022144975481043e-05, + "loss": 0.1678, + "step": 8638 + }, + { + "epoch": 3.62, + "learning_rate": 1.202048456256409e-05, + "loss": 0.1997, + "step": 8639 + }, + { + "epoch": 3.62, + "learning_rate": 1.2018824091570103e-05, + "loss": 0.1733, + "step": 8640 + }, + { + "epoch": 3.62, + "learning_rate": 1.201716356254681e-05, + "loss": 0.1818, + "step": 8641 + }, + { + "epoch": 3.62, + "learning_rate": 1.201550297554194e-05, + "loss": 0.1849, + "step": 8642 + }, + { + "epoch": 3.62, + "learning_rate": 1.2013842330603226e-05, + "loss": 0.1728, + "step": 8643 + }, + { + "epoch": 3.62, + "learning_rate": 1.20121816277784e-05, + "loss": 0.1764, + "step": 8644 + }, + { + "epoch": 3.62, + "learning_rate": 1.20105208671152e-05, + "loss": 0.2053, + "step": 8645 + }, + { + "epoch": 3.62, + "learning_rate": 1.2008860048661362e-05, + "loss": 0.1708, + "step": 8646 + }, + { + "epoch": 3.62, + "learning_rate": 1.200719917246462e-05, + "loss": 0.1796, + "step": 8647 + }, + { + "epoch": 3.62, + "learning_rate": 1.2005538238572724e-05, + "loss": 0.1694, + "step": 8648 + }, + { + "epoch": 3.62, + "learning_rate": 1.2003877247033411e-05, + "loss": 0.1919, + "step": 8649 + }, + { + "epoch": 3.62, + "learning_rate": 1.200221619789442e-05, + "loss": 0.242, + "step": 8650 + }, + { + "epoch": 3.62, + "learning_rate": 1.2000555091203507e-05, + "loss": 0.1495, + "step": 8651 + }, + { + "epoch": 3.62, + "learning_rate": 1.1998893927008411e-05, + "loss": 0.1781, + "step": 8652 + }, + { + "epoch": 3.62, + "learning_rate": 1.199723270535688e-05, + "loss": 0.1978, + "step": 8653 + }, + { + "epoch": 3.62, + "learning_rate": 1.1995571426296671e-05, + "loss": 0.2193, + "step": 8654 + }, + { + "epoch": 3.62, + "learning_rate": 1.1993910089875531e-05, + "loss": 0.1623, + "step": 8655 + }, + { + "epoch": 3.62, + "learning_rate": 1.1992248696141215e-05, + "loss": 0.2132, + "step": 8656 + }, + { + "epoch": 3.62, + "learning_rate": 1.1990587245141477e-05, + "loss": 0.1865, + "step": 8657 + }, + { + "epoch": 3.62, + "learning_rate": 1.1988925736924078e-05, + "loss": 0.1778, + "step": 8658 + }, + { + "epoch": 3.63, + "learning_rate": 1.198726417153677e-05, + "loss": 0.1642, + "step": 8659 + }, + { + "epoch": 3.63, + "learning_rate": 1.1985602549027318e-05, + "loss": 0.1997, + "step": 8660 + }, + { + "epoch": 3.63, + "learning_rate": 1.1983940869443484e-05, + "loss": 0.2021, + "step": 8661 + }, + { + "epoch": 3.63, + "learning_rate": 1.1982279132833028e-05, + "loss": 0.1802, + "step": 8662 + }, + { + "epoch": 3.63, + "learning_rate": 1.1980617339243719e-05, + "loss": 0.215, + "step": 8663 + }, + { + "epoch": 3.63, + "learning_rate": 1.1978955488723323e-05, + "loss": 0.1522, + "step": 8664 + }, + { + "epoch": 3.63, + "learning_rate": 1.1977293581319605e-05, + "loss": 0.1754, + "step": 8665 + }, + { + "epoch": 3.63, + "learning_rate": 1.1975631617080339e-05, + "loss": 0.1967, + "step": 8666 + }, + { + "epoch": 3.63, + "learning_rate": 1.1973969596053295e-05, + "loss": 0.1964, + "step": 8667 + }, + { + "epoch": 3.63, + "learning_rate": 1.1972307518286248e-05, + "loss": 0.163, + "step": 8668 + }, + { + "epoch": 3.63, + "learning_rate": 1.1970645383826971e-05, + "loss": 0.2226, + "step": 8669 + }, + { + "epoch": 3.63, + "learning_rate": 1.1968983192723238e-05, + "loss": 0.1526, + "step": 8670 + }, + { + "epoch": 3.63, + "learning_rate": 1.1967320945022836e-05, + "loss": 0.1703, + "step": 8671 + }, + { + "epoch": 3.63, + "learning_rate": 1.1965658640773535e-05, + "loss": 0.1807, + "step": 8672 + }, + { + "epoch": 3.63, + "learning_rate": 1.196399628002312e-05, + "loss": 0.1969, + "step": 8673 + }, + { + "epoch": 3.63, + "learning_rate": 1.1962333862819379e-05, + "loss": 0.1891, + "step": 8674 + }, + { + "epoch": 3.63, + "learning_rate": 1.1960671389210087e-05, + "loss": 0.1824, + "step": 8675 + }, + { + "epoch": 3.63, + "learning_rate": 1.195900885924304e-05, + "loss": 0.1831, + "step": 8676 + }, + { + "epoch": 3.63, + "learning_rate": 1.1957346272966019e-05, + "loss": 0.1629, + "step": 8677 + }, + { + "epoch": 3.63, + "learning_rate": 1.195568363042682e-05, + "loss": 0.187, + "step": 8678 + }, + { + "epoch": 3.63, + "learning_rate": 1.1954020931673229e-05, + "loss": 0.1698, + "step": 8679 + }, + { + "epoch": 3.63, + "learning_rate": 1.1952358176753038e-05, + "loss": 0.1691, + "step": 8680 + }, + { + "epoch": 3.63, + "learning_rate": 1.1950695365714046e-05, + "loss": 0.1688, + "step": 8681 + }, + { + "epoch": 3.63, + "learning_rate": 1.1949032498604047e-05, + "loss": 0.232, + "step": 8682 + }, + { + "epoch": 3.64, + "learning_rate": 1.1947369575470837e-05, + "loss": 0.1872, + "step": 8683 + }, + { + "epoch": 3.64, + "learning_rate": 1.1945706596362216e-05, + "loss": 0.1597, + "step": 8684 + }, + { + "epoch": 3.64, + "learning_rate": 1.1944043561325989e-05, + "loss": 0.2131, + "step": 8685 + }, + { + "epoch": 3.64, + "learning_rate": 1.194238047040995e-05, + "loss": 0.1646, + "step": 8686 + }, + { + "epoch": 3.64, + "learning_rate": 1.194071732366191e-05, + "loss": 0.1981, + "step": 8687 + }, + { + "epoch": 3.64, + "learning_rate": 1.1939054121129677e-05, + "loss": 0.1612, + "step": 8688 + }, + { + "epoch": 3.64, + "learning_rate": 1.1937390862861047e-05, + "loss": 0.1734, + "step": 8689 + }, + { + "epoch": 3.64, + "learning_rate": 1.1935727548903842e-05, + "loss": 0.1866, + "step": 8690 + }, + { + "epoch": 3.64, + "learning_rate": 1.1934064179305861e-05, + "loss": 0.1663, + "step": 8691 + }, + { + "epoch": 3.64, + "learning_rate": 1.1932400754114925e-05, + "loss": 0.1951, + "step": 8692 + }, + { + "epoch": 3.64, + "learning_rate": 1.1930737273378842e-05, + "loss": 0.1756, + "step": 8693 + }, + { + "epoch": 3.64, + "learning_rate": 1.1929073737145432e-05, + "loss": 0.1713, + "step": 8694 + }, + { + "epoch": 3.64, + "learning_rate": 1.1927410145462507e-05, + "loss": 0.2078, + "step": 8695 + }, + { + "epoch": 3.64, + "learning_rate": 1.1925746498377888e-05, + "loss": 0.1786, + "step": 8696 + }, + { + "epoch": 3.64, + "learning_rate": 1.1924082795939394e-05, + "loss": 0.1771, + "step": 8697 + }, + { + "epoch": 3.64, + "learning_rate": 1.192241903819485e-05, + "loss": 0.1918, + "step": 8698 + }, + { + "epoch": 3.64, + "learning_rate": 1.1920755225192074e-05, + "loss": 0.167, + "step": 8699 + }, + { + "epoch": 3.64, + "learning_rate": 1.1919091356978894e-05, + "loss": 0.1738, + "step": 8700 + }, + { + "epoch": 3.64, + "learning_rate": 1.1917427433603138e-05, + "loss": 0.203, + "step": 8701 + }, + { + "epoch": 3.64, + "learning_rate": 1.1915763455112633e-05, + "loss": 0.2271, + "step": 8702 + }, + { + "epoch": 3.64, + "learning_rate": 1.1914099421555203e-05, + "loss": 0.1783, + "step": 8703 + }, + { + "epoch": 3.64, + "learning_rate": 1.1912435332978684e-05, + "loss": 0.1941, + "step": 8704 + }, + { + "epoch": 3.64, + "learning_rate": 1.1910771189430913e-05, + "loss": 0.2174, + "step": 8705 + }, + { + "epoch": 3.64, + "learning_rate": 1.1909106990959716e-05, + "loss": 0.1742, + "step": 8706 + }, + { + "epoch": 3.65, + "learning_rate": 1.1907442737612932e-05, + "loss": 0.1669, + "step": 8707 + }, + { + "epoch": 3.65, + "learning_rate": 1.1905778429438401e-05, + "loss": 0.2056, + "step": 8708 + }, + { + "epoch": 3.65, + "learning_rate": 1.190411406648396e-05, + "loss": 0.2232, + "step": 8709 + }, + { + "epoch": 3.65, + "learning_rate": 1.190244964879745e-05, + "loss": 0.1932, + "step": 8710 + }, + { + "epoch": 3.65, + "learning_rate": 1.190078517642671e-05, + "loss": 0.1979, + "step": 8711 + }, + { + "epoch": 3.65, + "learning_rate": 1.189912064941959e-05, + "loss": 0.1841, + "step": 8712 + }, + { + "epoch": 3.65, + "learning_rate": 1.1897456067823934e-05, + "loss": 0.1931, + "step": 8713 + }, + { + "epoch": 3.65, + "learning_rate": 1.1895791431687585e-05, + "loss": 0.1672, + "step": 8714 + }, + { + "epoch": 3.65, + "learning_rate": 1.1894126741058393e-05, + "loss": 0.1963, + "step": 8715 + }, + { + "epoch": 3.65, + "learning_rate": 1.189246199598421e-05, + "loss": 0.2083, + "step": 8716 + }, + { + "epoch": 3.65, + "learning_rate": 1.1890797196512884e-05, + "loss": 0.218, + "step": 8717 + }, + { + "epoch": 3.65, + "learning_rate": 1.1889132342692272e-05, + "loss": 0.1758, + "step": 8718 + }, + { + "epoch": 3.65, + "learning_rate": 1.1887467434570227e-05, + "loss": 0.1998, + "step": 8719 + }, + { + "epoch": 3.65, + "learning_rate": 1.1885802472194606e-05, + "loss": 0.1669, + "step": 8720 + }, + { + "epoch": 3.65, + "learning_rate": 1.1884137455613266e-05, + "loss": 0.1861, + "step": 8721 + }, + { + "epoch": 3.65, + "learning_rate": 1.1882472384874067e-05, + "loss": 0.2202, + "step": 8722 + }, + { + "epoch": 3.65, + "learning_rate": 1.1880807260024867e-05, + "loss": 0.1837, + "step": 8723 + }, + { + "epoch": 3.65, + "learning_rate": 1.1879142081113535e-05, + "loss": 0.1654, + "step": 8724 + }, + { + "epoch": 3.65, + "learning_rate": 1.187747684818793e-05, + "loss": 0.1711, + "step": 8725 + }, + { + "epoch": 3.65, + "learning_rate": 1.1875811561295921e-05, + "loss": 0.1968, + "step": 8726 + }, + { + "epoch": 3.65, + "learning_rate": 1.187414622048537e-05, + "loss": 0.1816, + "step": 8727 + }, + { + "epoch": 3.65, + "learning_rate": 1.1872480825804149e-05, + "loss": 0.1843, + "step": 8728 + }, + { + "epoch": 3.65, + "learning_rate": 1.1870815377300131e-05, + "loss": 0.171, + "step": 8729 + }, + { + "epoch": 3.66, + "learning_rate": 1.1869149875021184e-05, + "loss": 0.1602, + "step": 8730 + }, + { + "epoch": 3.66, + "learning_rate": 1.186748431901518e-05, + "loss": 0.1338, + "step": 8731 + }, + { + "epoch": 3.66, + "learning_rate": 1.1865818709330001e-05, + "loss": 0.211, + "step": 8732 + }, + { + "epoch": 3.66, + "learning_rate": 1.1864153046013515e-05, + "loss": 0.2019, + "step": 8733 + }, + { + "epoch": 3.66, + "learning_rate": 1.1862487329113606e-05, + "loss": 0.2191, + "step": 8734 + }, + { + "epoch": 3.66, + "learning_rate": 1.1860821558678152e-05, + "loss": 0.1587, + "step": 8735 + }, + { + "epoch": 3.66, + "learning_rate": 1.1859155734755034e-05, + "loss": 0.22, + "step": 8736 + }, + { + "epoch": 3.66, + "learning_rate": 1.1857489857392133e-05, + "loss": 0.1619, + "step": 8737 + }, + { + "epoch": 3.66, + "learning_rate": 1.1855823926637333e-05, + "loss": 0.1612, + "step": 8738 + }, + { + "epoch": 3.66, + "learning_rate": 1.1854157942538522e-05, + "loss": 0.1793, + "step": 8739 + }, + { + "epoch": 3.66, + "learning_rate": 1.1852491905143586e-05, + "loss": 0.1781, + "step": 8740 + }, + { + "epoch": 3.66, + "learning_rate": 1.1850825814500413e-05, + "loss": 0.1956, + "step": 8741 + }, + { + "epoch": 3.66, + "learning_rate": 1.1849159670656897e-05, + "loss": 0.1916, + "step": 8742 + }, + { + "epoch": 3.66, + "learning_rate": 1.1847493473660925e-05, + "loss": 0.1949, + "step": 8743 + }, + { + "epoch": 3.66, + "learning_rate": 1.1845827223560394e-05, + "loss": 0.1847, + "step": 8744 + }, + { + "epoch": 3.66, + "learning_rate": 1.1844160920403197e-05, + "loss": 0.1929, + "step": 8745 + }, + { + "epoch": 3.66, + "learning_rate": 1.1842494564237231e-05, + "loss": 0.131, + "step": 8746 + }, + { + "epoch": 3.66, + "learning_rate": 1.1840828155110395e-05, + "loss": 0.1856, + "step": 8747 + }, + { + "epoch": 3.66, + "learning_rate": 1.1839161693070584e-05, + "loss": 0.1538, + "step": 8748 + }, + { + "epoch": 3.66, + "learning_rate": 1.1837495178165706e-05, + "loss": 0.1963, + "step": 8749 + }, + { + "epoch": 3.66, + "learning_rate": 1.1835828610443656e-05, + "loss": 0.1969, + "step": 8750 + }, + { + "epoch": 3.66, + "learning_rate": 1.1834161989952345e-05, + "loss": 0.1978, + "step": 8751 + }, + { + "epoch": 3.66, + "learning_rate": 1.1832495316739672e-05, + "loss": 0.2095, + "step": 8752 + }, + { + "epoch": 3.66, + "learning_rate": 1.1830828590853548e-05, + "loss": 0.1731, + "step": 8753 + }, + { + "epoch": 3.67, + "learning_rate": 1.1829161812341883e-05, + "loss": 0.1978, + "step": 8754 + }, + { + "epoch": 3.67, + "learning_rate": 1.182749498125258e-05, + "loss": 0.1767, + "step": 8755 + }, + { + "epoch": 3.67, + "learning_rate": 1.1825828097633562e-05, + "loss": 0.1752, + "step": 8756 + }, + { + "epoch": 3.67, + "learning_rate": 1.1824161161532731e-05, + "loss": 0.1937, + "step": 8757 + }, + { + "epoch": 3.67, + "learning_rate": 1.1822494172998008e-05, + "loss": 0.1824, + "step": 8758 + }, + { + "epoch": 3.67, + "learning_rate": 1.1820827132077306e-05, + "loss": 0.1865, + "step": 8759 + }, + { + "epoch": 3.67, + "learning_rate": 1.1819160038818547e-05, + "loss": 0.1787, + "step": 8760 + }, + { + "epoch": 3.67, + "learning_rate": 1.1817492893269645e-05, + "loss": 0.1637, + "step": 8761 + }, + { + "epoch": 3.67, + "learning_rate": 1.181582569547852e-05, + "loss": 0.2163, + "step": 8762 + }, + { + "epoch": 3.67, + "learning_rate": 1.18141584454931e-05, + "loss": 0.1649, + "step": 8763 + }, + { + "epoch": 3.67, + "learning_rate": 1.1812491143361305e-05, + "loss": 0.2028, + "step": 8764 + }, + { + "epoch": 3.67, + "learning_rate": 1.1810823789131059e-05, + "loss": 0.1657, + "step": 8765 + }, + { + "epoch": 3.67, + "learning_rate": 1.1809156382850291e-05, + "loss": 0.1668, + "step": 8766 + }, + { + "epoch": 3.67, + "learning_rate": 1.1807488924566929e-05, + "loss": 0.1856, + "step": 8767 + }, + { + "epoch": 3.67, + "learning_rate": 1.1805821414328897e-05, + "loss": 0.1946, + "step": 8768 + }, + { + "epoch": 3.67, + "learning_rate": 1.1804153852184137e-05, + "loss": 0.2073, + "step": 8769 + }, + { + "epoch": 3.67, + "learning_rate": 1.1802486238180575e-05, + "loss": 0.1951, + "step": 8770 + }, + { + "epoch": 3.67, + "learning_rate": 1.1800818572366144e-05, + "loss": 0.1153, + "step": 8771 + }, + { + "epoch": 3.67, + "learning_rate": 1.179915085478878e-05, + "loss": 0.1781, + "step": 8772 + }, + { + "epoch": 3.67, + "learning_rate": 1.1797483085496424e-05, + "loss": 0.2328, + "step": 8773 + }, + { + "epoch": 3.67, + "learning_rate": 1.1795815264537008e-05, + "loss": 0.2347, + "step": 8774 + }, + { + "epoch": 3.67, + "learning_rate": 1.1794147391958477e-05, + "loss": 0.2036, + "step": 8775 + }, + { + "epoch": 3.67, + "learning_rate": 1.1792479467808774e-05, + "loss": 0.205, + "step": 8776 + }, + { + "epoch": 3.67, + "learning_rate": 1.1790811492135837e-05, + "loss": 0.1933, + "step": 8777 + }, + { + "epoch": 3.68, + "learning_rate": 1.1789143464987611e-05, + "loss": 0.1803, + "step": 8778 + }, + { + "epoch": 3.68, + "learning_rate": 1.1787475386412048e-05, + "loss": 0.2011, + "step": 8779 + }, + { + "epoch": 3.68, + "learning_rate": 1.178580725645709e-05, + "loss": 0.1484, + "step": 8780 + }, + { + "epoch": 3.68, + "learning_rate": 1.1784139075170686e-05, + "loss": 0.1619, + "step": 8781 + }, + { + "epoch": 3.68, + "learning_rate": 1.1782470842600785e-05, + "loss": 0.1636, + "step": 8782 + }, + { + "epoch": 3.68, + "learning_rate": 1.1780802558795346e-05, + "loss": 0.166, + "step": 8783 + }, + { + "epoch": 3.68, + "learning_rate": 1.1779134223802315e-05, + "loss": 0.1686, + "step": 8784 + }, + { + "epoch": 3.68, + "learning_rate": 1.177746583766965e-05, + "loss": 0.224, + "step": 8785 + }, + { + "epoch": 3.68, + "learning_rate": 1.1775797400445307e-05, + "loss": 0.1735, + "step": 8786 + }, + { + "epoch": 3.68, + "learning_rate": 1.1774128912177245e-05, + "loss": 0.1761, + "step": 8787 + }, + { + "epoch": 3.68, + "learning_rate": 1.177246037291342e-05, + "loss": 0.2169, + "step": 8788 + }, + { + "epoch": 3.68, + "learning_rate": 1.1770791782701794e-05, + "loss": 0.1702, + "step": 8789 + }, + { + "epoch": 3.68, + "learning_rate": 1.176912314159033e-05, + "loss": 0.1961, + "step": 8790 + }, + { + "epoch": 3.68, + "learning_rate": 1.1767454449626992e-05, + "loss": 0.2112, + "step": 8791 + }, + { + "epoch": 3.68, + "learning_rate": 1.1765785706859744e-05, + "loss": 0.2216, + "step": 8792 + }, + { + "epoch": 3.68, + "learning_rate": 1.1764116913336552e-05, + "loss": 0.2142, + "step": 8793 + }, + { + "epoch": 3.68, + "learning_rate": 1.1762448069105387e-05, + "loss": 0.2356, + "step": 8794 + }, + { + "epoch": 3.68, + "learning_rate": 1.1760779174214215e-05, + "loss": 0.1855, + "step": 8795 + }, + { + "epoch": 3.68, + "learning_rate": 1.1759110228711006e-05, + "loss": 0.2006, + "step": 8796 + }, + { + "epoch": 3.68, + "learning_rate": 1.1757441232643737e-05, + "loss": 0.2181, + "step": 8797 + }, + { + "epoch": 3.68, + "learning_rate": 1.1755772186060379e-05, + "loss": 0.1845, + "step": 8798 + }, + { + "epoch": 3.68, + "learning_rate": 1.1754103089008905e-05, + "loss": 0.1632, + "step": 8799 + }, + { + "epoch": 3.68, + "learning_rate": 1.1752433941537297e-05, + "loss": 0.1701, + "step": 8800 + }, + { + "epoch": 3.68, + "learning_rate": 1.175076474369353e-05, + "loss": 0.1762, + "step": 8801 + }, + { + "epoch": 3.69, + "learning_rate": 1.1749095495525584e-05, + "loss": 0.2207, + "step": 8802 + }, + { + "epoch": 3.69, + "learning_rate": 1.1747426197081443e-05, + "loss": 0.2055, + "step": 8803 + }, + { + "epoch": 3.69, + "learning_rate": 1.1745756848409083e-05, + "loss": 0.1968, + "step": 8804 + }, + { + "epoch": 3.69, + "learning_rate": 1.1744087449556492e-05, + "loss": 0.164, + "step": 8805 + }, + { + "epoch": 3.69, + "learning_rate": 1.1742418000571653e-05, + "loss": 0.2076, + "step": 8806 + }, + { + "epoch": 3.69, + "learning_rate": 1.1740748501502559e-05, + "loss": 0.1933, + "step": 8807 + }, + { + "epoch": 3.69, + "learning_rate": 1.1739078952397189e-05, + "loss": 0.1662, + "step": 8808 + }, + { + "epoch": 3.69, + "learning_rate": 1.1737409353303541e-05, + "loss": 0.1819, + "step": 8809 + }, + { + "epoch": 3.69, + "learning_rate": 1.1735739704269605e-05, + "loss": 0.1808, + "step": 8810 + }, + { + "epoch": 3.69, + "learning_rate": 1.1734070005343368e-05, + "loss": 0.184, + "step": 8811 + }, + { + "epoch": 3.69, + "learning_rate": 1.1732400256572827e-05, + "loss": 0.1854, + "step": 8812 + }, + { + "epoch": 3.69, + "learning_rate": 1.1730730458005981e-05, + "loss": 0.1776, + "step": 8813 + }, + { + "epoch": 3.69, + "learning_rate": 1.1729060609690824e-05, + "loss": 0.188, + "step": 8814 + }, + { + "epoch": 3.69, + "learning_rate": 1.172739071167535e-05, + "loss": 0.1515, + "step": 8815 + }, + { + "epoch": 3.69, + "learning_rate": 1.1725720764007564e-05, + "loss": 0.2115, + "step": 8816 + }, + { + "epoch": 3.69, + "learning_rate": 1.1724050766735469e-05, + "loss": 0.1856, + "step": 8817 + }, + { + "epoch": 3.69, + "learning_rate": 1.1722380719907063e-05, + "loss": 0.1764, + "step": 8818 + }, + { + "epoch": 3.69, + "learning_rate": 1.172071062357035e-05, + "loss": 0.1929, + "step": 8819 + }, + { + "epoch": 3.69, + "learning_rate": 1.171904047777334e-05, + "loss": 0.1946, + "step": 8820 + }, + { + "epoch": 3.69, + "learning_rate": 1.1717370282564034e-05, + "loss": 0.1833, + "step": 8821 + }, + { + "epoch": 3.69, + "learning_rate": 1.1715700037990443e-05, + "loss": 0.1939, + "step": 8822 + }, + { + "epoch": 3.69, + "learning_rate": 1.171402974410058e-05, + "loss": 0.167, + "step": 8823 + }, + { + "epoch": 3.69, + "learning_rate": 1.171235940094245e-05, + "loss": 0.223, + "step": 8824 + }, + { + "epoch": 3.69, + "learning_rate": 1.1710689008564068e-05, + "loss": 0.1941, + "step": 8825 + }, + { + "epoch": 3.7, + "learning_rate": 1.1709018567013455e-05, + "loss": 0.1697, + "step": 8826 + }, + { + "epoch": 3.7, + "learning_rate": 1.1707348076338616e-05, + "loss": 0.1743, + "step": 8827 + }, + { + "epoch": 3.7, + "learning_rate": 1.170567753658757e-05, + "loss": 0.2077, + "step": 8828 + }, + { + "epoch": 3.7, + "learning_rate": 1.1704006947808338e-05, + "loss": 0.1682, + "step": 8829 + }, + { + "epoch": 3.7, + "learning_rate": 1.1702336310048936e-05, + "loss": 0.1975, + "step": 8830 + }, + { + "epoch": 3.7, + "learning_rate": 1.170066562335739e-05, + "loss": 0.1564, + "step": 8831 + }, + { + "epoch": 3.7, + "learning_rate": 1.1698994887781718e-05, + "loss": 0.1795, + "step": 8832 + }, + { + "epoch": 3.7, + "learning_rate": 1.1697324103369947e-05, + "loss": 0.1962, + "step": 8833 + }, + { + "epoch": 3.7, + "learning_rate": 1.1695653270170102e-05, + "loss": 0.2126, + "step": 8834 + }, + { + "epoch": 3.7, + "learning_rate": 1.1693982388230207e-05, + "loss": 0.2115, + "step": 8835 + }, + { + "epoch": 3.7, + "learning_rate": 1.1692311457598291e-05, + "loss": 0.2035, + "step": 8836 + }, + { + "epoch": 3.7, + "learning_rate": 1.1690640478322386e-05, + "loss": 0.1825, + "step": 8837 + }, + { + "epoch": 3.7, + "learning_rate": 1.168896945045052e-05, + "loss": 0.1644, + "step": 8838 + }, + { + "epoch": 3.7, + "learning_rate": 1.1687298374030724e-05, + "loss": 0.1886, + "step": 8839 + }, + { + "epoch": 3.7, + "learning_rate": 1.1685627249111034e-05, + "loss": 0.1961, + "step": 8840 + }, + { + "epoch": 3.7, + "learning_rate": 1.1683956075739487e-05, + "loss": 0.2219, + "step": 8841 + }, + { + "epoch": 3.7, + "learning_rate": 1.1682284853964114e-05, + "loss": 0.1937, + "step": 8842 + }, + { + "epoch": 3.7, + "learning_rate": 1.1680613583832956e-05, + "loss": 0.1764, + "step": 8843 + }, + { + "epoch": 3.7, + "learning_rate": 1.1678942265394055e-05, + "loss": 0.1845, + "step": 8844 + }, + { + "epoch": 3.7, + "learning_rate": 1.1677270898695445e-05, + "loss": 0.1797, + "step": 8845 + }, + { + "epoch": 3.7, + "learning_rate": 1.1675599483785171e-05, + "loss": 0.1832, + "step": 8846 + }, + { + "epoch": 3.7, + "learning_rate": 1.167392802071128e-05, + "loss": 0.2014, + "step": 8847 + }, + { + "epoch": 3.7, + "learning_rate": 1.1672256509521814e-05, + "loss": 0.2086, + "step": 8848 + }, + { + "epoch": 3.7, + "learning_rate": 1.1670584950264816e-05, + "loss": 0.1592, + "step": 8849 + }, + { + "epoch": 3.71, + "learning_rate": 1.1668913342988335e-05, + "loss": 0.2036, + "step": 8850 + }, + { + "epoch": 3.71, + "learning_rate": 1.1667241687740422e-05, + "loss": 0.1597, + "step": 8851 + }, + { + "epoch": 3.71, + "learning_rate": 1.1665569984569127e-05, + "loss": 0.2258, + "step": 8852 + }, + { + "epoch": 3.71, + "learning_rate": 1.1663898233522499e-05, + "loss": 0.1592, + "step": 8853 + }, + { + "epoch": 3.71, + "learning_rate": 1.1662226434648595e-05, + "loss": 0.1749, + "step": 8854 + }, + { + "epoch": 3.71, + "learning_rate": 1.1660554587995465e-05, + "loss": 0.1716, + "step": 8855 + }, + { + "epoch": 3.71, + "learning_rate": 1.1658882693611169e-05, + "loss": 0.1531, + "step": 8856 + }, + { + "epoch": 3.71, + "learning_rate": 1.165721075154376e-05, + "loss": 0.1848, + "step": 8857 + }, + { + "epoch": 3.71, + "learning_rate": 1.16555387618413e-05, + "loss": 0.1777, + "step": 8858 + }, + { + "epoch": 3.71, + "learning_rate": 1.1653866724551848e-05, + "loss": 0.2096, + "step": 8859 + }, + { + "epoch": 3.71, + "learning_rate": 1.1652194639723465e-05, + "loss": 0.1777, + "step": 8860 + }, + { + "epoch": 3.71, + "learning_rate": 1.1650522507404211e-05, + "loss": 0.2106, + "step": 8861 + }, + { + "epoch": 3.71, + "learning_rate": 1.1648850327642154e-05, + "loss": 0.1917, + "step": 8862 + }, + { + "epoch": 3.71, + "learning_rate": 1.1647178100485354e-05, + "loss": 0.1864, + "step": 8863 + }, + { + "epoch": 3.71, + "learning_rate": 1.1645505825981884e-05, + "loss": 0.1869, + "step": 8864 + }, + { + "epoch": 3.71, + "learning_rate": 1.1643833504179811e-05, + "loss": 0.1971, + "step": 8865 + }, + { + "epoch": 3.71, + "learning_rate": 1.1642161135127198e-05, + "loss": 0.1678, + "step": 8866 + }, + { + "epoch": 3.71, + "learning_rate": 1.1640488718872124e-05, + "loss": 0.1875, + "step": 8867 + }, + { + "epoch": 3.71, + "learning_rate": 1.1638816255462657e-05, + "loss": 0.1905, + "step": 8868 + }, + { + "epoch": 3.71, + "learning_rate": 1.1637143744946871e-05, + "loss": 0.2219, + "step": 8869 + }, + { + "epoch": 3.71, + "learning_rate": 1.163547118737284e-05, + "loss": 0.1915, + "step": 8870 + }, + { + "epoch": 3.71, + "learning_rate": 1.1633798582788642e-05, + "loss": 0.1979, + "step": 8871 + }, + { + "epoch": 3.71, + "learning_rate": 1.1632125931242356e-05, + "loss": 0.2147, + "step": 8872 + }, + { + "epoch": 3.71, + "learning_rate": 1.1630453232782053e-05, + "loss": 0.2346, + "step": 8873 + }, + { + "epoch": 3.72, + "learning_rate": 1.162878048745582e-05, + "loss": 0.1713, + "step": 8874 + }, + { + "epoch": 3.72, + "learning_rate": 1.1627107695311738e-05, + "loss": 0.2076, + "step": 8875 + }, + { + "epoch": 3.72, + "learning_rate": 1.1625434856397891e-05, + "loss": 0.1693, + "step": 8876 + }, + { + "epoch": 3.72, + "learning_rate": 1.1623761970762359e-05, + "loss": 0.1629, + "step": 8877 + }, + { + "epoch": 3.72, + "learning_rate": 1.1622089038453232e-05, + "loss": 0.1809, + "step": 8878 + }, + { + "epoch": 3.72, + "learning_rate": 1.1620416059518593e-05, + "loss": 0.1831, + "step": 8879 + }, + { + "epoch": 3.72, + "learning_rate": 1.1618743034006532e-05, + "loss": 0.1921, + "step": 8880 + }, + { + "epoch": 3.72, + "learning_rate": 1.1617069961965141e-05, + "loss": 0.1877, + "step": 8881 + }, + { + "epoch": 3.72, + "learning_rate": 1.161539684344251e-05, + "loss": 0.2086, + "step": 8882 + }, + { + "epoch": 3.72, + "learning_rate": 1.1613723678486728e-05, + "loss": 0.1888, + "step": 8883 + }, + { + "epoch": 3.72, + "learning_rate": 1.161205046714589e-05, + "loss": 0.2187, + "step": 8884 + }, + { + "epoch": 3.72, + "learning_rate": 1.1610377209468095e-05, + "loss": 0.1716, + "step": 8885 + }, + { + "epoch": 3.72, + "learning_rate": 1.1608703905501432e-05, + "loss": 0.2116, + "step": 8886 + }, + { + "epoch": 3.72, + "learning_rate": 1.1607030555294005e-05, + "loss": 0.2013, + "step": 8887 + }, + { + "epoch": 3.72, + "learning_rate": 1.1605357158893913e-05, + "loss": 0.1809, + "step": 8888 + }, + { + "epoch": 3.72, + "learning_rate": 1.160368371634925e-05, + "loss": 0.196, + "step": 8889 + }, + { + "epoch": 3.72, + "learning_rate": 1.1602010227708122e-05, + "loss": 0.2002, + "step": 8890 + }, + { + "epoch": 3.72, + "learning_rate": 1.1600336693018635e-05, + "loss": 0.1671, + "step": 8891 + }, + { + "epoch": 3.72, + "learning_rate": 1.1598663112328888e-05, + "loss": 0.1837, + "step": 8892 + }, + { + "epoch": 3.72, + "learning_rate": 1.1596989485686987e-05, + "loss": 0.202, + "step": 8893 + }, + { + "epoch": 3.72, + "learning_rate": 1.1595315813141041e-05, + "loss": 0.1697, + "step": 8894 + }, + { + "epoch": 3.72, + "learning_rate": 1.1593642094739161e-05, + "loss": 0.1923, + "step": 8895 + }, + { + "epoch": 3.72, + "learning_rate": 1.159196833052945e-05, + "loss": 0.166, + "step": 8896 + }, + { + "epoch": 3.72, + "learning_rate": 1.1590294520560022e-05, + "loss": 0.2166, + "step": 8897 + }, + { + "epoch": 3.73, + "learning_rate": 1.1588620664878989e-05, + "loss": 0.1934, + "step": 8898 + }, + { + "epoch": 3.73, + "learning_rate": 1.1586946763534465e-05, + "loss": 0.1996, + "step": 8899 + }, + { + "epoch": 3.73, + "learning_rate": 1.1585272816574565e-05, + "loss": 0.1903, + "step": 8900 + }, + { + "epoch": 3.73, + "learning_rate": 1.1583598824047404e-05, + "loss": 0.1923, + "step": 8901 + }, + { + "epoch": 3.73, + "learning_rate": 1.1581924786001102e-05, + "loss": 0.1939, + "step": 8902 + }, + { + "epoch": 3.73, + "learning_rate": 1.1580250702483773e-05, + "loss": 0.1808, + "step": 8903 + }, + { + "epoch": 3.73, + "learning_rate": 1.1578576573543541e-05, + "loss": 0.1719, + "step": 8904 + }, + { + "epoch": 3.73, + "learning_rate": 1.1576902399228527e-05, + "loss": 0.1946, + "step": 8905 + }, + { + "epoch": 3.73, + "learning_rate": 1.1575228179586855e-05, + "loss": 0.1857, + "step": 8906 + }, + { + "epoch": 3.73, + "learning_rate": 1.1573553914666643e-05, + "loss": 0.1814, + "step": 8907 + }, + { + "epoch": 3.73, + "learning_rate": 1.157187960451602e-05, + "loss": 0.1515, + "step": 8908 + }, + { + "epoch": 3.73, + "learning_rate": 1.1570205249183115e-05, + "loss": 0.1651, + "step": 8909 + }, + { + "epoch": 3.73, + "learning_rate": 1.156853084871605e-05, + "loss": 0.1432, + "step": 8910 + }, + { + "epoch": 3.73, + "learning_rate": 1.156685640316296e-05, + "loss": 0.1824, + "step": 8911 + }, + { + "epoch": 3.73, + "learning_rate": 1.1565181912571975e-05, + "loss": 0.1531, + "step": 8912 + }, + { + "epoch": 3.73, + "learning_rate": 1.1563507376991222e-05, + "loss": 0.2062, + "step": 8913 + }, + { + "epoch": 3.73, + "learning_rate": 1.1561832796468837e-05, + "loss": 0.209, + "step": 8914 + }, + { + "epoch": 3.73, + "learning_rate": 1.1560158171052956e-05, + "loss": 0.1206, + "step": 8915 + }, + { + "epoch": 3.73, + "learning_rate": 1.1558483500791713e-05, + "loss": 0.1754, + "step": 8916 + }, + { + "epoch": 3.73, + "learning_rate": 1.1556808785733242e-05, + "loss": 0.2223, + "step": 8917 + }, + { + "epoch": 3.73, + "learning_rate": 1.1555134025925685e-05, + "loss": 0.1631, + "step": 8918 + }, + { + "epoch": 3.73, + "learning_rate": 1.1553459221417182e-05, + "loss": 0.1875, + "step": 8919 + }, + { + "epoch": 3.73, + "learning_rate": 1.155178437225587e-05, + "loss": 0.2032, + "step": 8920 + }, + { + "epoch": 3.73, + "learning_rate": 1.1550109478489896e-05, + "loss": 0.1785, + "step": 8921 + }, + { + "epoch": 3.74, + "learning_rate": 1.1548434540167399e-05, + "loss": 0.1759, + "step": 8922 + }, + { + "epoch": 3.74, + "learning_rate": 1.1546759557336524e-05, + "loss": 0.2091, + "step": 8923 + }, + { + "epoch": 3.74, + "learning_rate": 1.1545084530045419e-05, + "loss": 0.211, + "step": 8924 + }, + { + "epoch": 3.74, + "learning_rate": 1.154340945834223e-05, + "loss": 0.2067, + "step": 8925 + }, + { + "epoch": 3.74, + "learning_rate": 1.1541734342275106e-05, + "loss": 0.2118, + "step": 8926 + }, + { + "epoch": 3.74, + "learning_rate": 1.1540059181892197e-05, + "loss": 0.1759, + "step": 8927 + }, + { + "epoch": 3.74, + "learning_rate": 1.1538383977241654e-05, + "loss": 0.179, + "step": 8928 + }, + { + "epoch": 3.74, + "learning_rate": 1.1536708728371628e-05, + "loss": 0.1841, + "step": 8929 + }, + { + "epoch": 3.74, + "learning_rate": 1.1535033435330272e-05, + "loss": 0.1893, + "step": 8930 + }, + { + "epoch": 3.74, + "learning_rate": 1.1533358098165744e-05, + "loss": 0.2098, + "step": 8931 + }, + { + "epoch": 3.74, + "learning_rate": 1.1531682716926199e-05, + "loss": 0.1792, + "step": 8932 + }, + { + "epoch": 3.74, + "learning_rate": 1.153000729165979e-05, + "loss": 0.1708, + "step": 8933 + }, + { + "epoch": 3.74, + "learning_rate": 1.1528331822414681e-05, + "loss": 0.1958, + "step": 8934 + }, + { + "epoch": 3.74, + "learning_rate": 1.1526656309239028e-05, + "loss": 0.1662, + "step": 8935 + }, + { + "epoch": 3.74, + "learning_rate": 1.1524980752180996e-05, + "loss": 0.208, + "step": 8936 + }, + { + "epoch": 3.74, + "learning_rate": 1.1523305151288747e-05, + "loss": 0.173, + "step": 8937 + }, + { + "epoch": 3.74, + "learning_rate": 1.1521629506610439e-05, + "loss": 0.216, + "step": 8938 + }, + { + "epoch": 3.74, + "learning_rate": 1.1519953818194246e-05, + "loss": 0.1837, + "step": 8939 + }, + { + "epoch": 3.74, + "learning_rate": 1.1518278086088328e-05, + "loss": 0.1763, + "step": 8940 + }, + { + "epoch": 3.74, + "learning_rate": 1.1516602310340852e-05, + "loss": 0.1977, + "step": 8941 + }, + { + "epoch": 3.74, + "learning_rate": 1.151492649099999e-05, + "loss": 0.2517, + "step": 8942 + }, + { + "epoch": 3.74, + "learning_rate": 1.151325062811391e-05, + "loss": 0.234, + "step": 8943 + }, + { + "epoch": 3.74, + "learning_rate": 1.1511574721730781e-05, + "loss": 0.1756, + "step": 8944 + }, + { + "epoch": 3.75, + "learning_rate": 1.150989877189878e-05, + "loss": 0.1934, + "step": 8945 + }, + { + "epoch": 3.75, + "learning_rate": 1.1508222778666078e-05, + "loss": 0.1277, + "step": 8946 + }, + { + "epoch": 3.75, + "learning_rate": 1.1506546742080854e-05, + "loss": 0.1748, + "step": 8947 + }, + { + "epoch": 3.75, + "learning_rate": 1.1504870662191275e-05, + "loss": 0.169, + "step": 8948 + }, + { + "epoch": 3.75, + "learning_rate": 1.1503194539045527e-05, + "loss": 0.1794, + "step": 8949 + }, + { + "epoch": 3.75, + "learning_rate": 1.1501518372691789e-05, + "loss": 0.134, + "step": 8950 + }, + { + "epoch": 3.75, + "learning_rate": 1.1499842163178234e-05, + "loss": 0.1981, + "step": 8951 + }, + { + "epoch": 3.75, + "learning_rate": 1.1498165910553048e-05, + "loss": 0.1996, + "step": 8952 + }, + { + "epoch": 3.75, + "learning_rate": 1.1496489614864412e-05, + "loss": 0.1395, + "step": 8953 + }, + { + "epoch": 3.75, + "learning_rate": 1.1494813276160511e-05, + "loss": 0.1912, + "step": 8954 + }, + { + "epoch": 3.75, + "learning_rate": 1.1493136894489527e-05, + "loss": 0.1953, + "step": 8955 + }, + { + "epoch": 3.75, + "learning_rate": 1.149146046989965e-05, + "loss": 0.1645, + "step": 8956 + }, + { + "epoch": 3.75, + "learning_rate": 1.1489784002439064e-05, + "loss": 0.2151, + "step": 8957 + }, + { + "epoch": 3.75, + "learning_rate": 1.1488107492155958e-05, + "loss": 0.171, + "step": 8958 + }, + { + "epoch": 3.75, + "learning_rate": 1.1486430939098525e-05, + "loss": 0.1875, + "step": 8959 + }, + { + "epoch": 3.75, + "learning_rate": 1.1484754343314952e-05, + "loss": 0.2094, + "step": 8960 + }, + { + "epoch": 3.75, + "learning_rate": 1.1483077704853431e-05, + "loss": 0.1987, + "step": 8961 + }, + { + "epoch": 3.75, + "learning_rate": 1.1481401023762162e-05, + "loss": 0.2137, + "step": 8962 + }, + { + "epoch": 3.75, + "learning_rate": 1.1479724300089335e-05, + "loss": 0.1704, + "step": 8963 + }, + { + "epoch": 3.75, + "learning_rate": 1.1478047533883143e-05, + "loss": 0.1562, + "step": 8964 + }, + { + "epoch": 3.75, + "learning_rate": 1.1476370725191788e-05, + "loss": 0.1986, + "step": 8965 + }, + { + "epoch": 3.75, + "learning_rate": 1.1474693874063467e-05, + "loss": 0.1907, + "step": 8966 + }, + { + "epoch": 3.75, + "learning_rate": 1.1473016980546377e-05, + "loss": 0.1945, + "step": 8967 + }, + { + "epoch": 3.75, + "learning_rate": 1.147134004468872e-05, + "loss": 0.2249, + "step": 8968 + }, + { + "epoch": 3.76, + "learning_rate": 1.1469663066538704e-05, + "loss": 0.1739, + "step": 8969 + }, + { + "epoch": 3.76, + "learning_rate": 1.1467986046144523e-05, + "loss": 0.1978, + "step": 8970 + }, + { + "epoch": 3.76, + "learning_rate": 1.1466308983554386e-05, + "loss": 0.1379, + "step": 8971 + }, + { + "epoch": 3.76, + "learning_rate": 1.1464631878816502e-05, + "loss": 0.1583, + "step": 8972 + }, + { + "epoch": 3.76, + "learning_rate": 1.1462954731979074e-05, + "loss": 0.1796, + "step": 8973 + }, + { + "epoch": 3.76, + "learning_rate": 1.1461277543090308e-05, + "loss": 0.1897, + "step": 8974 + }, + { + "epoch": 3.76, + "learning_rate": 1.1459600312198416e-05, + "loss": 0.2038, + "step": 8975 + }, + { + "epoch": 3.76, + "learning_rate": 1.1457923039351606e-05, + "loss": 0.1697, + "step": 8976 + }, + { + "epoch": 3.76, + "learning_rate": 1.1456245724598096e-05, + "loss": 0.1876, + "step": 8977 + }, + { + "epoch": 3.76, + "learning_rate": 1.1454568367986094e-05, + "loss": 0.19, + "step": 8978 + }, + { + "epoch": 3.76, + "learning_rate": 1.1452890969563813e-05, + "loss": 0.2207, + "step": 8979 + }, + { + "epoch": 3.76, + "learning_rate": 1.1451213529379472e-05, + "loss": 0.1888, + "step": 8980 + }, + { + "epoch": 3.76, + "learning_rate": 1.1449536047481287e-05, + "loss": 0.1905, + "step": 8981 + }, + { + "epoch": 3.76, + "learning_rate": 1.1447858523917472e-05, + "loss": 0.2155, + "step": 8982 + }, + { + "epoch": 3.76, + "learning_rate": 1.144618095873625e-05, + "loss": 0.2026, + "step": 8983 + }, + { + "epoch": 3.76, + "learning_rate": 1.144450335198584e-05, + "loss": 0.1414, + "step": 8984 + }, + { + "epoch": 3.76, + "learning_rate": 1.1442825703714464e-05, + "loss": 0.1805, + "step": 8985 + }, + { + "epoch": 3.76, + "learning_rate": 1.144114801397034e-05, + "loss": 0.1571, + "step": 8986 + }, + { + "epoch": 3.76, + "learning_rate": 1.14394702828017e-05, + "loss": 0.2133, + "step": 8987 + }, + { + "epoch": 3.76, + "learning_rate": 1.143779251025676e-05, + "loss": 0.1764, + "step": 8988 + }, + { + "epoch": 3.76, + "learning_rate": 1.1436114696383749e-05, + "loss": 0.2045, + "step": 8989 + }, + { + "epoch": 3.76, + "learning_rate": 1.14344368412309e-05, + "loss": 0.1676, + "step": 8990 + }, + { + "epoch": 3.76, + "learning_rate": 1.1432758944846433e-05, + "loss": 0.1914, + "step": 8991 + }, + { + "epoch": 3.76, + "learning_rate": 1.1431081007278584e-05, + "loss": 0.1964, + "step": 8992 + }, + { + "epoch": 3.77, + "learning_rate": 1.142940302857558e-05, + "loss": 0.1722, + "step": 8993 + }, + { + "epoch": 3.77, + "learning_rate": 1.1427725008785655e-05, + "loss": 0.173, + "step": 8994 + }, + { + "epoch": 3.77, + "learning_rate": 1.1426046947957041e-05, + "loss": 0.2282, + "step": 8995 + }, + { + "epoch": 3.77, + "learning_rate": 1.1424368846137975e-05, + "loss": 0.1957, + "step": 8996 + }, + { + "epoch": 3.77, + "learning_rate": 1.142269070337669e-05, + "loss": 0.1834, + "step": 8997 + }, + { + "epoch": 3.77, + "learning_rate": 1.1421012519721421e-05, + "loss": 0.1803, + "step": 8998 + }, + { + "epoch": 3.77, + "learning_rate": 1.141933429522041e-05, + "loss": 0.2323, + "step": 8999 + }, + { + "epoch": 3.77, + "learning_rate": 1.1417656029921895e-05, + "loss": 0.2027, + "step": 9000 + }, + { + "epoch": 3.77, + "learning_rate": 1.1415977723874112e-05, + "loss": 0.1768, + "step": 9001 + }, + { + "epoch": 3.77, + "learning_rate": 1.1414299377125308e-05, + "loss": 0.1486, + "step": 9002 + }, + { + "epoch": 3.77, + "learning_rate": 1.1412620989723724e-05, + "loss": 0.1703, + "step": 9003 + }, + { + "epoch": 3.77, + "learning_rate": 1.1410942561717601e-05, + "loss": 0.1961, + "step": 9004 + }, + { + "epoch": 3.77, + "learning_rate": 1.1409264093155189e-05, + "loss": 0.2018, + "step": 9005 + }, + { + "epoch": 3.77, + "learning_rate": 1.1407585584084726e-05, + "loss": 0.1878, + "step": 9006 + }, + { + "epoch": 3.77, + "learning_rate": 1.1405907034554473e-05, + "loss": 0.2096, + "step": 9007 + }, + { + "epoch": 3.77, + "learning_rate": 1.1404228444612664e-05, + "loss": 0.1901, + "step": 9008 + }, + { + "epoch": 3.77, + "learning_rate": 1.1402549814307555e-05, + "loss": 0.1547, + "step": 9009 + }, + { + "epoch": 3.77, + "learning_rate": 1.1400871143687398e-05, + "loss": 0.193, + "step": 9010 + }, + { + "epoch": 3.77, + "learning_rate": 1.1399192432800442e-05, + "loss": 0.1671, + "step": 9011 + }, + { + "epoch": 3.77, + "learning_rate": 1.1397513681694938e-05, + "loss": 0.1926, + "step": 9012 + }, + { + "epoch": 3.77, + "learning_rate": 1.1395834890419147e-05, + "loss": 0.1855, + "step": 9013 + }, + { + "epoch": 3.77, + "learning_rate": 1.139415605902132e-05, + "loss": 0.1667, + "step": 9014 + }, + { + "epoch": 3.77, + "learning_rate": 1.1392477187549714e-05, + "loss": 0.2235, + "step": 9015 + }, + { + "epoch": 3.77, + "learning_rate": 1.1390798276052587e-05, + "loss": 0.206, + "step": 9016 + }, + { + "epoch": 3.78, + "learning_rate": 1.1389119324578197e-05, + "loss": 0.1925, + "step": 9017 + }, + { + "epoch": 3.78, + "learning_rate": 1.1387440333174809e-05, + "loss": 0.1357, + "step": 9018 + }, + { + "epoch": 3.78, + "learning_rate": 1.1385761301890674e-05, + "loss": 0.1902, + "step": 9019 + }, + { + "epoch": 3.78, + "learning_rate": 1.1384082230774063e-05, + "loss": 0.2036, + "step": 9020 + }, + { + "epoch": 3.78, + "learning_rate": 1.1382403119873236e-05, + "loss": 0.1801, + "step": 9021 + }, + { + "epoch": 3.78, + "learning_rate": 1.1380723969236458e-05, + "loss": 0.204, + "step": 9022 + }, + { + "epoch": 3.78, + "learning_rate": 1.1379044778911994e-05, + "loss": 0.2092, + "step": 9023 + }, + { + "epoch": 3.78, + "learning_rate": 1.1377365548948114e-05, + "loss": 0.164, + "step": 9024 + }, + { + "epoch": 3.78, + "learning_rate": 1.137568627939308e-05, + "loss": 0.1681, + "step": 9025 + }, + { + "epoch": 3.78, + "learning_rate": 1.1374006970295167e-05, + "loss": 0.1726, + "step": 9026 + }, + { + "epoch": 3.78, + "learning_rate": 1.137232762170264e-05, + "loss": 0.1698, + "step": 9027 + }, + { + "epoch": 3.78, + "learning_rate": 1.1370648233663776e-05, + "loss": 0.1895, + "step": 9028 + }, + { + "epoch": 3.78, + "learning_rate": 1.1368968806226844e-05, + "loss": 0.195, + "step": 9029 + }, + { + "epoch": 3.78, + "learning_rate": 1.1367289339440123e-05, + "loss": 0.2167, + "step": 9030 + }, + { + "epoch": 3.78, + "learning_rate": 1.1365609833351877e-05, + "loss": 0.232, + "step": 9031 + }, + { + "epoch": 3.78, + "learning_rate": 1.136393028801039e-05, + "loss": 0.1908, + "step": 9032 + }, + { + "epoch": 3.78, + "learning_rate": 1.1362250703463937e-05, + "loss": 0.1637, + "step": 9033 + }, + { + "epoch": 3.78, + "learning_rate": 1.1360571079760798e-05, + "loss": 0.1488, + "step": 9034 + }, + { + "epoch": 3.78, + "learning_rate": 1.1358891416949249e-05, + "loss": 0.2147, + "step": 9035 + }, + { + "epoch": 3.78, + "learning_rate": 1.1357211715077571e-05, + "loss": 0.2032, + "step": 9036 + }, + { + "epoch": 3.78, + "learning_rate": 1.1355531974194051e-05, + "loss": 0.1986, + "step": 9037 + }, + { + "epoch": 3.78, + "learning_rate": 1.1353852194346963e-05, + "loss": 0.1634, + "step": 9038 + }, + { + "epoch": 3.78, + "learning_rate": 1.1352172375584596e-05, + "loss": 0.2022, + "step": 9039 + }, + { + "epoch": 3.78, + "learning_rate": 1.1350492517955234e-05, + "loss": 0.1578, + "step": 9040 + }, + { + "epoch": 3.79, + "learning_rate": 1.1348812621507167e-05, + "loss": 0.1488, + "step": 9041 + }, + { + "epoch": 3.79, + "learning_rate": 1.1347132686288673e-05, + "loss": 0.1921, + "step": 9042 + }, + { + "epoch": 3.79, + "learning_rate": 1.1345452712348048e-05, + "loss": 0.1615, + "step": 9043 + }, + { + "epoch": 3.79, + "learning_rate": 1.1343772699733579e-05, + "loss": 0.2077, + "step": 9044 + }, + { + "epoch": 3.79, + "learning_rate": 1.1342092648493555e-05, + "loss": 0.1969, + "step": 9045 + }, + { + "epoch": 3.79, + "learning_rate": 1.1340412558676266e-05, + "loss": 0.2035, + "step": 9046 + }, + { + "epoch": 3.79, + "learning_rate": 1.1338732430330011e-05, + "loss": 0.1633, + "step": 9047 + }, + { + "epoch": 3.79, + "learning_rate": 1.1337052263503079e-05, + "loss": 0.1723, + "step": 9048 + }, + { + "epoch": 3.79, + "learning_rate": 1.1335372058243768e-05, + "loss": 0.2138, + "step": 9049 + }, + { + "epoch": 3.79, + "learning_rate": 1.1333691814600369e-05, + "loss": 0.1746, + "step": 9050 + }, + { + "epoch": 3.79, + "learning_rate": 1.1332011532621186e-05, + "loss": 0.2055, + "step": 9051 + }, + { + "epoch": 3.79, + "learning_rate": 1.1330331212354513e-05, + "loss": 0.1866, + "step": 9052 + }, + { + "epoch": 3.79, + "learning_rate": 1.1328650853848646e-05, + "loss": 0.1995, + "step": 9053 + }, + { + "epoch": 3.79, + "learning_rate": 1.132697045715189e-05, + "loss": 0.1582, + "step": 9054 + }, + { + "epoch": 3.79, + "learning_rate": 1.132529002231255e-05, + "loss": 0.2029, + "step": 9055 + }, + { + "epoch": 3.79, + "learning_rate": 1.132360954937892e-05, + "loss": 0.1756, + "step": 9056 + }, + { + "epoch": 3.79, + "learning_rate": 1.1321929038399309e-05, + "loss": 0.2151, + "step": 9057 + }, + { + "epoch": 3.79, + "learning_rate": 1.1320248489422024e-05, + "loss": 0.211, + "step": 9058 + }, + { + "epoch": 3.79, + "learning_rate": 1.1318567902495364e-05, + "loss": 0.2204, + "step": 9059 + }, + { + "epoch": 3.79, + "learning_rate": 1.131688727766764e-05, + "loss": 0.1669, + "step": 9060 + }, + { + "epoch": 3.79, + "learning_rate": 1.1315206614987161e-05, + "loss": 0.1861, + "step": 9061 + }, + { + "epoch": 3.79, + "learning_rate": 1.1313525914502235e-05, + "loss": 0.1728, + "step": 9062 + }, + { + "epoch": 3.79, + "learning_rate": 1.1311845176261172e-05, + "loss": 0.1886, + "step": 9063 + }, + { + "epoch": 3.79, + "learning_rate": 1.1310164400312284e-05, + "loss": 0.1917, + "step": 9064 + }, + { + "epoch": 3.8, + "learning_rate": 1.1308483586703885e-05, + "loss": 0.1719, + "step": 9065 + }, + { + "epoch": 3.8, + "learning_rate": 1.1306802735484282e-05, + "loss": 0.1885, + "step": 9066 + }, + { + "epoch": 3.8, + "learning_rate": 1.1305121846701797e-05, + "loss": 0.2081, + "step": 9067 + }, + { + "epoch": 3.8, + "learning_rate": 1.130344092040474e-05, + "loss": 0.2003, + "step": 9068 + }, + { + "epoch": 3.8, + "learning_rate": 1.1301759956641432e-05, + "loss": 0.199, + "step": 9069 + }, + { + "epoch": 3.8, + "learning_rate": 1.1300078955460187e-05, + "loss": 0.2198, + "step": 9070 + }, + { + "epoch": 3.8, + "learning_rate": 1.1298397916909332e-05, + "loss": 0.187, + "step": 9071 + }, + { + "epoch": 3.8, + "learning_rate": 1.1296716841037176e-05, + "loss": 0.2268, + "step": 9072 + }, + { + "epoch": 3.8, + "learning_rate": 1.1295035727892046e-05, + "loss": 0.2108, + "step": 9073 + }, + { + "epoch": 3.8, + "learning_rate": 1.1293354577522264e-05, + "loss": 0.2043, + "step": 9074 + }, + { + "epoch": 3.8, + "learning_rate": 1.1291673389976153e-05, + "loss": 0.1764, + "step": 9075 + }, + { + "epoch": 3.8, + "learning_rate": 1.1289992165302036e-05, + "loss": 0.2057, + "step": 9076 + }, + { + "epoch": 3.8, + "learning_rate": 1.1288310903548236e-05, + "loss": 0.1947, + "step": 9077 + }, + { + "epoch": 3.8, + "learning_rate": 1.1286629604763087e-05, + "loss": 0.2054, + "step": 9078 + }, + { + "epoch": 3.8, + "learning_rate": 1.1284948268994908e-05, + "loss": 0.2173, + "step": 9079 + }, + { + "epoch": 3.8, + "learning_rate": 1.1283266896292033e-05, + "loss": 0.1699, + "step": 9080 + }, + { + "epoch": 3.8, + "learning_rate": 1.1281585486702791e-05, + "loss": 0.1738, + "step": 9081 + }, + { + "epoch": 3.8, + "learning_rate": 1.1279904040275508e-05, + "loss": 0.1923, + "step": 9082 + }, + { + "epoch": 3.8, + "learning_rate": 1.1278222557058521e-05, + "loss": 0.191, + "step": 9083 + }, + { + "epoch": 3.8, + "learning_rate": 1.127654103710016e-05, + "loss": 0.2003, + "step": 9084 + }, + { + "epoch": 3.8, + "learning_rate": 1.127485948044876e-05, + "loss": 0.1608, + "step": 9085 + }, + { + "epoch": 3.8, + "learning_rate": 1.127317788715266e-05, + "loss": 0.1259, + "step": 9086 + }, + { + "epoch": 3.8, + "learning_rate": 1.1271496257260186e-05, + "loss": 0.1972, + "step": 9087 + }, + { + "epoch": 3.8, + "learning_rate": 1.126981459081968e-05, + "loss": 0.1557, + "step": 9088 + }, + { + "epoch": 3.81, + "learning_rate": 1.1268132887879485e-05, + "loss": 0.2127, + "step": 9089 + }, + { + "epoch": 3.81, + "learning_rate": 1.1266451148487934e-05, + "loss": 0.1941, + "step": 9090 + }, + { + "epoch": 3.81, + "learning_rate": 1.1264769372693365e-05, + "loss": 0.1913, + "step": 9091 + }, + { + "epoch": 3.81, + "learning_rate": 1.1263087560544128e-05, + "loss": 0.1689, + "step": 9092 + }, + { + "epoch": 3.81, + "learning_rate": 1.1261405712088556e-05, + "loss": 0.1506, + "step": 9093 + }, + { + "epoch": 3.81, + "learning_rate": 1.1259723827374996e-05, + "loss": 0.2145, + "step": 9094 + }, + { + "epoch": 3.81, + "learning_rate": 1.1258041906451795e-05, + "loss": 0.1616, + "step": 9095 + }, + { + "epoch": 3.81, + "learning_rate": 1.1256359949367293e-05, + "loss": 0.2278, + "step": 9096 + }, + { + "epoch": 3.81, + "learning_rate": 1.1254677956169839e-05, + "loss": 0.1865, + "step": 9097 + }, + { + "epoch": 3.81, + "learning_rate": 1.1252995926907783e-05, + "loss": 0.1616, + "step": 9098 + }, + { + "epoch": 3.81, + "learning_rate": 1.1251313861629474e-05, + "loss": 0.1893, + "step": 9099 + }, + { + "epoch": 3.81, + "learning_rate": 1.1249631760383253e-05, + "loss": 0.2181, + "step": 9100 + }, + { + "epoch": 3.81, + "learning_rate": 1.1247949623217476e-05, + "loss": 0.1697, + "step": 9101 + }, + { + "epoch": 3.81, + "learning_rate": 1.1246267450180497e-05, + "loss": 0.1728, + "step": 9102 + }, + { + "epoch": 3.81, + "learning_rate": 1.1244585241320664e-05, + "loss": 0.2064, + "step": 9103 + }, + { + "epoch": 3.81, + "learning_rate": 1.1242902996686333e-05, + "loss": 0.186, + "step": 9104 + }, + { + "epoch": 3.81, + "learning_rate": 1.1241220716325863e-05, + "loss": 0.1907, + "step": 9105 + }, + { + "epoch": 3.81, + "learning_rate": 1.12395384002876e-05, + "loss": 0.1635, + "step": 9106 + }, + { + "epoch": 3.81, + "learning_rate": 1.1237856048619904e-05, + "loss": 0.1679, + "step": 9107 + }, + { + "epoch": 3.81, + "learning_rate": 1.1236173661371139e-05, + "loss": 0.1374, + "step": 9108 + }, + { + "epoch": 3.81, + "learning_rate": 1.1234491238589658e-05, + "loss": 0.1645, + "step": 9109 + }, + { + "epoch": 3.81, + "learning_rate": 1.1232808780323821e-05, + "loss": 0.2045, + "step": 9110 + }, + { + "epoch": 3.81, + "learning_rate": 1.1231126286621988e-05, + "loss": 0.1994, + "step": 9111 + }, + { + "epoch": 3.81, + "learning_rate": 1.1229443757532526e-05, + "loss": 0.1896, + "step": 9112 + }, + { + "epoch": 3.82, + "learning_rate": 1.122776119310379e-05, + "loss": 0.1959, + "step": 9113 + }, + { + "epoch": 3.82, + "learning_rate": 1.122607859338415e-05, + "loss": 0.2131, + "step": 9114 + }, + { + "epoch": 3.82, + "learning_rate": 1.122439595842197e-05, + "loss": 0.1619, + "step": 9115 + }, + { + "epoch": 3.82, + "learning_rate": 1.1222713288265613e-05, + "loss": 0.1915, + "step": 9116 + }, + { + "epoch": 3.82, + "learning_rate": 1.1221030582963447e-05, + "loss": 0.1749, + "step": 9117 + }, + { + "epoch": 3.82, + "learning_rate": 1.1219347842563838e-05, + "loss": 0.1934, + "step": 9118 + }, + { + "epoch": 3.82, + "learning_rate": 1.1217665067115164e-05, + "loss": 0.1642, + "step": 9119 + }, + { + "epoch": 3.82, + "learning_rate": 1.1215982256665786e-05, + "loss": 0.2118, + "step": 9120 + }, + { + "epoch": 3.82, + "learning_rate": 1.1214299411264075e-05, + "loss": 0.1747, + "step": 9121 + }, + { + "epoch": 3.82, + "learning_rate": 1.1212616530958407e-05, + "loss": 0.1927, + "step": 9122 + }, + { + "epoch": 3.82, + "learning_rate": 1.121093361579715e-05, + "loss": 0.1762, + "step": 9123 + }, + { + "epoch": 3.82, + "learning_rate": 1.1209250665828684e-05, + "loss": 0.1817, + "step": 9124 + }, + { + "epoch": 3.82, + "learning_rate": 1.1207567681101377e-05, + "loss": 0.2285, + "step": 9125 + }, + { + "epoch": 3.82, + "learning_rate": 1.1205884661663614e-05, + "loss": 0.1896, + "step": 9126 + }, + { + "epoch": 3.82, + "learning_rate": 1.1204201607563761e-05, + "loss": 0.2009, + "step": 9127 + }, + { + "epoch": 3.82, + "learning_rate": 1.1202518518850202e-05, + "loss": 0.1918, + "step": 9128 + }, + { + "epoch": 3.82, + "learning_rate": 1.1200835395571319e-05, + "loss": 0.1798, + "step": 9129 + }, + { + "epoch": 3.82, + "learning_rate": 1.1199152237775485e-05, + "loss": 0.1773, + "step": 9130 + }, + { + "epoch": 3.82, + "learning_rate": 1.1197469045511085e-05, + "loss": 0.1865, + "step": 9131 + }, + { + "epoch": 3.82, + "learning_rate": 1.11957858188265e-05, + "loss": 0.1732, + "step": 9132 + }, + { + "epoch": 3.82, + "learning_rate": 1.1194102557770117e-05, + "loss": 0.2204, + "step": 9133 + }, + { + "epoch": 3.82, + "learning_rate": 1.119241926239031e-05, + "loss": 0.1778, + "step": 9134 + }, + { + "epoch": 3.82, + "learning_rate": 1.119073593273547e-05, + "loss": 0.1671, + "step": 9135 + }, + { + "epoch": 3.82, + "learning_rate": 1.1189052568853985e-05, + "loss": 0.2029, + "step": 9136 + }, + { + "epoch": 3.83, + "learning_rate": 1.1187369170794235e-05, + "loss": 0.2035, + "step": 9137 + }, + { + "epoch": 3.83, + "learning_rate": 1.1185685738604615e-05, + "loss": 0.1841, + "step": 9138 + }, + { + "epoch": 3.83, + "learning_rate": 1.1184002272333509e-05, + "loss": 0.1578, + "step": 9139 + }, + { + "epoch": 3.83, + "learning_rate": 1.1182318772029309e-05, + "loss": 0.1937, + "step": 9140 + }, + { + "epoch": 3.83, + "learning_rate": 1.1180635237740405e-05, + "loss": 0.1827, + "step": 9141 + }, + { + "epoch": 3.83, + "learning_rate": 1.117895166951519e-05, + "loss": 0.2154, + "step": 9142 + }, + { + "epoch": 3.83, + "learning_rate": 1.1177268067402057e-05, + "loss": 0.1988, + "step": 9143 + }, + { + "epoch": 3.83, + "learning_rate": 1.1175584431449393e-05, + "loss": 0.1743, + "step": 9144 + }, + { + "epoch": 3.83, + "learning_rate": 1.11739007617056e-05, + "loss": 0.2047, + "step": 9145 + }, + { + "epoch": 3.83, + "learning_rate": 1.1172217058219073e-05, + "loss": 0.1919, + "step": 9146 + }, + { + "epoch": 3.83, + "learning_rate": 1.1170533321038204e-05, + "loss": 0.186, + "step": 9147 + }, + { + "epoch": 3.83, + "learning_rate": 1.1168849550211394e-05, + "loss": 0.1803, + "step": 9148 + }, + { + "epoch": 3.83, + "learning_rate": 1.1167165745787041e-05, + "loss": 0.2222, + "step": 9149 + }, + { + "epoch": 3.83, + "learning_rate": 1.1165481907813544e-05, + "loss": 0.1488, + "step": 9150 + }, + { + "epoch": 3.83, + "learning_rate": 1.1163798036339303e-05, + "loss": 0.1842, + "step": 9151 + }, + { + "epoch": 3.83, + "learning_rate": 1.1162114131412723e-05, + "loss": 0.1747, + "step": 9152 + }, + { + "epoch": 3.83, + "learning_rate": 1.1160430193082203e-05, + "loss": 0.1926, + "step": 9153 + }, + { + "epoch": 3.83, + "learning_rate": 1.1158746221396148e-05, + "loss": 0.1933, + "step": 9154 + }, + { + "epoch": 3.83, + "learning_rate": 1.1157062216402959e-05, + "loss": 0.1534, + "step": 9155 + }, + { + "epoch": 3.83, + "learning_rate": 1.1155378178151044e-05, + "loss": 0.1914, + "step": 9156 + }, + { + "epoch": 3.83, + "learning_rate": 1.1153694106688809e-05, + "loss": 0.1767, + "step": 9157 + }, + { + "epoch": 3.83, + "learning_rate": 1.115201000206466e-05, + "loss": 0.1747, + "step": 9158 + }, + { + "epoch": 3.83, + "learning_rate": 1.1150325864327003e-05, + "loss": 0.193, + "step": 9159 + }, + { + "epoch": 3.84, + "learning_rate": 1.1148641693524254e-05, + "loss": 0.1653, + "step": 9160 + }, + { + "epoch": 3.84, + "learning_rate": 1.1146957489704817e-05, + "loss": 0.1859, + "step": 9161 + }, + { + "epoch": 3.84, + "learning_rate": 1.1145273252917104e-05, + "loss": 0.1975, + "step": 9162 + }, + { + "epoch": 3.84, + "learning_rate": 1.114358898320953e-05, + "loss": 0.1936, + "step": 9163 + }, + { + "epoch": 3.84, + "learning_rate": 1.1141904680630504e-05, + "loss": 0.2159, + "step": 9164 + }, + { + "epoch": 3.84, + "learning_rate": 1.114022034522844e-05, + "loss": 0.1748, + "step": 9165 + }, + { + "epoch": 3.84, + "learning_rate": 1.1138535977051757e-05, + "loss": 0.2104, + "step": 9166 + }, + { + "epoch": 3.84, + "learning_rate": 1.1136851576148867e-05, + "loss": 0.1722, + "step": 9167 + }, + { + "epoch": 3.84, + "learning_rate": 1.1135167142568186e-05, + "loss": 0.2082, + "step": 9168 + }, + { + "epoch": 3.84, + "learning_rate": 1.1133482676358131e-05, + "loss": 0.1616, + "step": 9169 + }, + { + "epoch": 3.84, + "learning_rate": 1.1131798177567129e-05, + "loss": 0.1815, + "step": 9170 + }, + { + "epoch": 3.84, + "learning_rate": 1.1130113646243585e-05, + "loss": 0.2332, + "step": 9171 + }, + { + "epoch": 3.84, + "learning_rate": 1.112842908243593e-05, + "loss": 0.1734, + "step": 9172 + }, + { + "epoch": 3.84, + "learning_rate": 1.1126744486192585e-05, + "loss": 0.1712, + "step": 9173 + }, + { + "epoch": 3.84, + "learning_rate": 1.1125059857561966e-05, + "loss": 0.2049, + "step": 9174 + }, + { + "epoch": 3.84, + "learning_rate": 1.11233751965925e-05, + "loss": 0.2336, + "step": 9175 + }, + { + "epoch": 3.84, + "learning_rate": 1.1121690503332613e-05, + "loss": 0.1608, + "step": 9176 + }, + { + "epoch": 3.84, + "learning_rate": 1.1120005777830729e-05, + "loss": 0.2013, + "step": 9177 + }, + { + "epoch": 3.84, + "learning_rate": 1.111832102013527e-05, + "loss": 0.1843, + "step": 9178 + }, + { + "epoch": 3.84, + "learning_rate": 1.1116636230294663e-05, + "loss": 0.1676, + "step": 9179 + }, + { + "epoch": 3.84, + "learning_rate": 1.1114951408357344e-05, + "loss": 0.1899, + "step": 9180 + }, + { + "epoch": 3.84, + "learning_rate": 1.1113266554371734e-05, + "loss": 0.2414, + "step": 9181 + }, + { + "epoch": 3.84, + "learning_rate": 1.1111581668386262e-05, + "loss": 0.1606, + "step": 9182 + }, + { + "epoch": 3.84, + "learning_rate": 1.1109896750449367e-05, + "loss": 0.1816, + "step": 9183 + }, + { + "epoch": 3.85, + "learning_rate": 1.110821180060947e-05, + "loss": 0.207, + "step": 9184 + }, + { + "epoch": 3.85, + "learning_rate": 1.1106526818915008e-05, + "loss": 0.1648, + "step": 9185 + }, + { + "epoch": 3.85, + "learning_rate": 1.1104841805414418e-05, + "loss": 0.2001, + "step": 9186 + }, + { + "epoch": 3.85, + "learning_rate": 1.1103156760156129e-05, + "loss": 0.154, + "step": 9187 + }, + { + "epoch": 3.85, + "learning_rate": 1.110147168318858e-05, + "loss": 0.183, + "step": 9188 + }, + { + "epoch": 3.85, + "learning_rate": 1.10997865745602e-05, + "loss": 0.2053, + "step": 9189 + }, + { + "epoch": 3.85, + "learning_rate": 1.1098101434319438e-05, + "loss": 0.2275, + "step": 9190 + }, + { + "epoch": 3.85, + "learning_rate": 1.1096416262514717e-05, + "loss": 0.197, + "step": 9191 + }, + { + "epoch": 3.85, + "learning_rate": 1.1094731059194488e-05, + "loss": 0.2249, + "step": 9192 + }, + { + "epoch": 3.85, + "learning_rate": 1.1093045824407186e-05, + "loss": 0.1649, + "step": 9193 + }, + { + "epoch": 3.85, + "learning_rate": 1.1091360558201249e-05, + "loss": 0.205, + "step": 9194 + }, + { + "epoch": 3.85, + "learning_rate": 1.1089675260625124e-05, + "loss": 0.1797, + "step": 9195 + }, + { + "epoch": 3.85, + "learning_rate": 1.1087989931727248e-05, + "loss": 0.1717, + "step": 9196 + }, + { + "epoch": 3.85, + "learning_rate": 1.1086304571556071e-05, + "loss": 0.1843, + "step": 9197 + }, + { + "epoch": 3.85, + "learning_rate": 1.1084619180160028e-05, + "loss": 0.2124, + "step": 9198 + }, + { + "epoch": 3.85, + "learning_rate": 1.1082933757587571e-05, + "loss": 0.1779, + "step": 9199 + }, + { + "epoch": 3.85, + "learning_rate": 1.1081248303887147e-05, + "loss": 0.2332, + "step": 9200 + }, + { + "epoch": 3.85, + "learning_rate": 1.1079562819107198e-05, + "loss": 0.2471, + "step": 9201 + }, + { + "epoch": 3.85, + "learning_rate": 1.1077877303296174e-05, + "loss": 0.1762, + "step": 9202 + }, + { + "epoch": 3.85, + "learning_rate": 1.1076191756502521e-05, + "loss": 0.2314, + "step": 9203 + }, + { + "epoch": 3.85, + "learning_rate": 1.1074506178774696e-05, + "loss": 0.173, + "step": 9204 + }, + { + "epoch": 3.85, + "learning_rate": 1.107282057016114e-05, + "loss": 0.2055, + "step": 9205 + }, + { + "epoch": 3.85, + "learning_rate": 1.1071134930710308e-05, + "loss": 0.1714, + "step": 9206 + }, + { + "epoch": 3.85, + "learning_rate": 1.1069449260470658e-05, + "loss": 0.1473, + "step": 9207 + }, + { + "epoch": 3.86, + "learning_rate": 1.1067763559490632e-05, + "loss": 0.149, + "step": 9208 + }, + { + "epoch": 3.86, + "learning_rate": 1.1066077827818694e-05, + "loss": 0.2025, + "step": 9209 + }, + { + "epoch": 3.86, + "learning_rate": 1.1064392065503294e-05, + "loss": 0.171, + "step": 9210 + }, + { + "epoch": 3.86, + "learning_rate": 1.106270627259289e-05, + "loss": 0.1824, + "step": 9211 + }, + { + "epoch": 3.86, + "learning_rate": 1.1061020449135935e-05, + "loss": 0.186, + "step": 9212 + }, + { + "epoch": 3.86, + "learning_rate": 1.1059334595180888e-05, + "loss": 0.1983, + "step": 9213 + }, + { + "epoch": 3.86, + "learning_rate": 1.1057648710776212e-05, + "loss": 0.1745, + "step": 9214 + }, + { + "epoch": 3.86, + "learning_rate": 1.1055962795970359e-05, + "loss": 0.2321, + "step": 9215 + }, + { + "epoch": 3.86, + "learning_rate": 1.1054276850811791e-05, + "loss": 0.2003, + "step": 9216 + }, + { + "epoch": 3.86, + "learning_rate": 1.1052590875348975e-05, + "loss": 0.2135, + "step": 9217 + }, + { + "epoch": 3.86, + "learning_rate": 1.1050904869630365e-05, + "loss": 0.2082, + "step": 9218 + }, + { + "epoch": 3.86, + "learning_rate": 1.1049218833704427e-05, + "loss": 0.2232, + "step": 9219 + }, + { + "epoch": 3.86, + "learning_rate": 1.1047532767619629e-05, + "loss": 0.2073, + "step": 9220 + }, + { + "epoch": 3.86, + "learning_rate": 1.1045846671424424e-05, + "loss": 0.1981, + "step": 9221 + }, + { + "epoch": 3.86, + "learning_rate": 1.1044160545167294e-05, + "loss": 0.1434, + "step": 9222 + }, + { + "epoch": 3.86, + "learning_rate": 1.1042474388896687e-05, + "loss": 0.1945, + "step": 9223 + }, + { + "epoch": 3.86, + "learning_rate": 1.1040788202661084e-05, + "loss": 0.1881, + "step": 9224 + }, + { + "epoch": 3.86, + "learning_rate": 1.1039101986508945e-05, + "loss": 0.1577, + "step": 9225 + }, + { + "epoch": 3.86, + "learning_rate": 1.1037415740488742e-05, + "loss": 0.1527, + "step": 9226 + }, + { + "epoch": 3.86, + "learning_rate": 1.1035729464648945e-05, + "loss": 0.2051, + "step": 9227 + }, + { + "epoch": 3.86, + "learning_rate": 1.1034043159038021e-05, + "loss": 0.1761, + "step": 9228 + }, + { + "epoch": 3.86, + "learning_rate": 1.1032356823704446e-05, + "loss": 0.2139, + "step": 9229 + }, + { + "epoch": 3.86, + "learning_rate": 1.1030670458696688e-05, + "loss": 0.2011, + "step": 9230 + }, + { + "epoch": 3.86, + "learning_rate": 1.1028984064063226e-05, + "loss": 0.1833, + "step": 9231 + }, + { + "epoch": 3.87, + "learning_rate": 1.1027297639852528e-05, + "loss": 0.1715, + "step": 9232 + }, + { + "epoch": 3.87, + "learning_rate": 1.1025611186113073e-05, + "loss": 0.1462, + "step": 9233 + }, + { + "epoch": 3.87, + "learning_rate": 1.1023924702893334e-05, + "loss": 0.1673, + "step": 9234 + }, + { + "epoch": 3.87, + "learning_rate": 1.1022238190241787e-05, + "loss": 0.1945, + "step": 9235 + }, + { + "epoch": 3.87, + "learning_rate": 1.102055164820691e-05, + "loss": 0.1562, + "step": 9236 + }, + { + "epoch": 3.87, + "learning_rate": 1.1018865076837182e-05, + "loss": 0.1894, + "step": 9237 + }, + { + "epoch": 3.87, + "learning_rate": 1.101717847618108e-05, + "loss": 0.1953, + "step": 9238 + }, + { + "epoch": 3.87, + "learning_rate": 1.1015491846287088e-05, + "loss": 0.2008, + "step": 9239 + }, + { + "epoch": 3.87, + "learning_rate": 1.1013805187203682e-05, + "loss": 0.2021, + "step": 9240 + }, + { + "epoch": 3.87, + "learning_rate": 1.1012118498979347e-05, + "loss": 0.199, + "step": 9241 + }, + { + "epoch": 3.87, + "learning_rate": 1.1010431781662563e-05, + "loss": 0.1611, + "step": 9242 + }, + { + "epoch": 3.87, + "learning_rate": 1.1008745035301814e-05, + "loss": 0.208, + "step": 9243 + }, + { + "epoch": 3.87, + "learning_rate": 1.1007058259945584e-05, + "loss": 0.1939, + "step": 9244 + }, + { + "epoch": 3.87, + "learning_rate": 1.100537145564236e-05, + "loss": 0.1942, + "step": 9245 + }, + { + "epoch": 3.87, + "learning_rate": 1.1003684622440622e-05, + "loss": 0.1912, + "step": 9246 + }, + { + "epoch": 3.87, + "learning_rate": 1.1001997760388862e-05, + "loss": 0.2018, + "step": 9247 + }, + { + "epoch": 3.87, + "learning_rate": 1.100031086953557e-05, + "loss": 0.1898, + "step": 9248 + }, + { + "epoch": 3.87, + "learning_rate": 1.0998623949929223e-05, + "loss": 0.1977, + "step": 9249 + }, + { + "epoch": 3.87, + "learning_rate": 1.0996937001618317e-05, + "loss": 0.1582, + "step": 9250 + }, + { + "epoch": 3.87, + "learning_rate": 1.0995250024651345e-05, + "loss": 0.1945, + "step": 9251 + }, + { + "epoch": 3.87, + "learning_rate": 1.0993563019076794e-05, + "loss": 0.1894, + "step": 9252 + }, + { + "epoch": 3.87, + "learning_rate": 1.0991875984943152e-05, + "loss": 0.1623, + "step": 9253 + }, + { + "epoch": 3.87, + "learning_rate": 1.0990188922298918e-05, + "loss": 0.1477, + "step": 9254 + }, + { + "epoch": 3.87, + "learning_rate": 1.0988501831192582e-05, + "loss": 0.1554, + "step": 9255 + }, + { + "epoch": 3.88, + "learning_rate": 1.098681471167264e-05, + "loss": 0.2056, + "step": 9256 + }, + { + "epoch": 3.88, + "learning_rate": 1.098512756378758e-05, + "loss": 0.2109, + "step": 9257 + }, + { + "epoch": 3.88, + "learning_rate": 1.0983440387585909e-05, + "loss": 0.15, + "step": 9258 + }, + { + "epoch": 3.88, + "learning_rate": 1.0981753183116111e-05, + "loss": 0.2154, + "step": 9259 + }, + { + "epoch": 3.88, + "learning_rate": 1.0980065950426687e-05, + "loss": 0.2, + "step": 9260 + }, + { + "epoch": 3.88, + "learning_rate": 1.0978378689566144e-05, + "loss": 0.217, + "step": 9261 + }, + { + "epoch": 3.88, + "learning_rate": 1.097669140058297e-05, + "loss": 0.2133, + "step": 9262 + }, + { + "epoch": 3.88, + "learning_rate": 1.0975004083525667e-05, + "loss": 0.198, + "step": 9263 + }, + { + "epoch": 3.88, + "learning_rate": 1.0973316738442738e-05, + "loss": 0.2014, + "step": 9264 + }, + { + "epoch": 3.88, + "learning_rate": 1.0971629365382686e-05, + "loss": 0.1903, + "step": 9265 + }, + { + "epoch": 3.88, + "learning_rate": 1.0969941964394006e-05, + "loss": 0.1916, + "step": 9266 + }, + { + "epoch": 3.88, + "learning_rate": 1.0968254535525204e-05, + "loss": 0.2014, + "step": 9267 + }, + { + "epoch": 3.88, + "learning_rate": 1.0966567078824792e-05, + "loss": 0.1985, + "step": 9268 + }, + { + "epoch": 3.88, + "learning_rate": 1.0964879594341262e-05, + "loss": 0.1646, + "step": 9269 + }, + { + "epoch": 3.88, + "learning_rate": 1.0963192082123124e-05, + "loss": 0.1842, + "step": 9270 + }, + { + "epoch": 3.88, + "learning_rate": 1.0961504542218884e-05, + "loss": 0.2025, + "step": 9271 + }, + { + "epoch": 3.88, + "learning_rate": 1.0959816974677052e-05, + "loss": 0.1725, + "step": 9272 + }, + { + "epoch": 3.88, + "learning_rate": 1.095812937954613e-05, + "loss": 0.2005, + "step": 9273 + }, + { + "epoch": 3.88, + "learning_rate": 1.0956441756874628e-05, + "loss": 0.1933, + "step": 9274 + }, + { + "epoch": 3.88, + "learning_rate": 1.095475410671106e-05, + "loss": 0.1843, + "step": 9275 + }, + { + "epoch": 3.88, + "learning_rate": 1.0953066429103929e-05, + "loss": 0.2184, + "step": 9276 + }, + { + "epoch": 3.88, + "learning_rate": 1.0951378724101751e-05, + "loss": 0.2005, + "step": 9277 + }, + { + "epoch": 3.88, + "learning_rate": 1.0949690991753036e-05, + "loss": 0.1757, + "step": 9278 + }, + { + "epoch": 3.88, + "learning_rate": 1.0948003232106298e-05, + "loss": 0.1942, + "step": 9279 + }, + { + "epoch": 3.89, + "learning_rate": 1.0946315445210048e-05, + "loss": 0.1583, + "step": 9280 + }, + { + "epoch": 3.89, + "learning_rate": 1.0944627631112796e-05, + "loss": 0.1767, + "step": 9281 + }, + { + "epoch": 3.89, + "learning_rate": 1.0942939789863065e-05, + "loss": 0.2008, + "step": 9282 + }, + { + "epoch": 3.89, + "learning_rate": 1.0941251921509367e-05, + "loss": 0.1965, + "step": 9283 + }, + { + "epoch": 3.89, + "learning_rate": 1.0939564026100213e-05, + "loss": 0.2036, + "step": 9284 + }, + { + "epoch": 3.89, + "learning_rate": 1.093787610368413e-05, + "loss": 0.1682, + "step": 9285 + }, + { + "epoch": 3.89, + "learning_rate": 1.0936188154309626e-05, + "loss": 0.1771, + "step": 9286 + }, + { + "epoch": 3.89, + "learning_rate": 1.0934500178025228e-05, + "loss": 0.1683, + "step": 9287 + }, + { + "epoch": 3.89, + "learning_rate": 1.093281217487945e-05, + "loss": 0.1783, + "step": 9288 + }, + { + "epoch": 3.89, + "learning_rate": 1.0931124144920813e-05, + "loss": 0.2245, + "step": 9289 + }, + { + "epoch": 3.89, + "learning_rate": 1.0929436088197842e-05, + "loss": 0.2076, + "step": 9290 + }, + { + "epoch": 3.89, + "learning_rate": 1.0927748004759053e-05, + "loss": 0.1446, + "step": 9291 + }, + { + "epoch": 3.89, + "learning_rate": 1.0926059894652972e-05, + "loss": 0.1343, + "step": 9292 + }, + { + "epoch": 3.89, + "learning_rate": 1.0924371757928121e-05, + "loss": 0.1693, + "step": 9293 + }, + { + "epoch": 3.89, + "learning_rate": 1.092268359463302e-05, + "loss": 0.2063, + "step": 9294 + }, + { + "epoch": 3.89, + "learning_rate": 1.0920995404816204e-05, + "loss": 0.1792, + "step": 9295 + }, + { + "epoch": 3.89, + "learning_rate": 1.091930718852619e-05, + "loss": 0.1761, + "step": 9296 + }, + { + "epoch": 3.89, + "learning_rate": 1.0917618945811503e-05, + "loss": 0.1832, + "step": 9297 + }, + { + "epoch": 3.89, + "learning_rate": 1.0915930676720678e-05, + "loss": 0.1779, + "step": 9298 + }, + { + "epoch": 3.89, + "learning_rate": 1.0914242381302238e-05, + "loss": 0.1716, + "step": 9299 + }, + { + "epoch": 3.89, + "learning_rate": 1.091255405960471e-05, + "loss": 0.173, + "step": 9300 + }, + { + "epoch": 3.89, + "learning_rate": 1.0910865711676628e-05, + "loss": 0.1813, + "step": 9301 + }, + { + "epoch": 3.89, + "learning_rate": 1.0909177337566518e-05, + "loss": 0.1746, + "step": 9302 + }, + { + "epoch": 3.89, + "learning_rate": 1.0907488937322915e-05, + "loss": 0.1858, + "step": 9303 + }, + { + "epoch": 3.9, + "learning_rate": 1.0905800510994341e-05, + "loss": 0.2029, + "step": 9304 + }, + { + "epoch": 3.9, + "learning_rate": 1.0904112058629341e-05, + "loss": 0.1727, + "step": 9305 + }, + { + "epoch": 3.9, + "learning_rate": 1.090242358027644e-05, + "loss": 0.1983, + "step": 9306 + }, + { + "epoch": 3.9, + "learning_rate": 1.0900735075984174e-05, + "loss": 0.1724, + "step": 9307 + }, + { + "epoch": 3.9, + "learning_rate": 1.0899046545801078e-05, + "loss": 0.1676, + "step": 9308 + }, + { + "epoch": 3.9, + "learning_rate": 1.0897357989775687e-05, + "loss": 0.1845, + "step": 9309 + }, + { + "epoch": 3.9, + "learning_rate": 1.0895669407956538e-05, + "loss": 0.1733, + "step": 9310 + }, + { + "epoch": 3.9, + "learning_rate": 1.0893980800392164e-05, + "loss": 0.209, + "step": 9311 + }, + { + "epoch": 3.9, + "learning_rate": 1.0892292167131107e-05, + "loss": 0.2143, + "step": 9312 + }, + { + "epoch": 3.9, + "learning_rate": 1.0890603508221905e-05, + "loss": 0.1837, + "step": 9313 + }, + { + "epoch": 3.9, + "learning_rate": 1.0888914823713092e-05, + "loss": 0.2079, + "step": 9314 + }, + { + "epoch": 3.9, + "learning_rate": 1.0887226113653211e-05, + "loss": 0.1714, + "step": 9315 + }, + { + "epoch": 3.9, + "learning_rate": 1.0885537378090807e-05, + "loss": 0.1785, + "step": 9316 + }, + { + "epoch": 3.9, + "learning_rate": 1.0883848617074412e-05, + "loss": 0.1692, + "step": 9317 + }, + { + "epoch": 3.9, + "learning_rate": 1.0882159830652574e-05, + "loss": 0.2079, + "step": 9318 + }, + { + "epoch": 3.9, + "learning_rate": 1.0880471018873836e-05, + "loss": 0.2076, + "step": 9319 + }, + { + "epoch": 3.9, + "learning_rate": 1.087878218178674e-05, + "loss": 0.2011, + "step": 9320 + }, + { + "epoch": 3.9, + "learning_rate": 1.0877093319439828e-05, + "loss": 0.1759, + "step": 9321 + }, + { + "epoch": 3.9, + "learning_rate": 1.0875404431881651e-05, + "loss": 0.2012, + "step": 9322 + }, + { + "epoch": 3.9, + "learning_rate": 1.0873715519160746e-05, + "loss": 0.1937, + "step": 9323 + }, + { + "epoch": 3.9, + "learning_rate": 1.0872026581325667e-05, + "loss": 0.1947, + "step": 9324 + }, + { + "epoch": 3.9, + "learning_rate": 1.0870337618424956e-05, + "loss": 0.1981, + "step": 9325 + }, + { + "epoch": 3.9, + "learning_rate": 1.0868648630507164e-05, + "loss": 0.2003, + "step": 9326 + }, + { + "epoch": 3.9, + "learning_rate": 1.0866959617620835e-05, + "loss": 0.2131, + "step": 9327 + }, + { + "epoch": 3.91, + "learning_rate": 1.0865270579814522e-05, + "loss": 0.1698, + "step": 9328 + }, + { + "epoch": 3.91, + "learning_rate": 1.0863581517136776e-05, + "loss": 0.1846, + "step": 9329 + }, + { + "epoch": 3.91, + "learning_rate": 1.0861892429636143e-05, + "loss": 0.1581, + "step": 9330 + }, + { + "epoch": 3.91, + "learning_rate": 1.0860203317361178e-05, + "loss": 0.1566, + "step": 9331 + }, + { + "epoch": 3.91, + "learning_rate": 1.0858514180360432e-05, + "loss": 0.2144, + "step": 9332 + }, + { + "epoch": 3.91, + "learning_rate": 1.085682501868246e-05, + "loss": 0.1654, + "step": 9333 + }, + { + "epoch": 3.91, + "learning_rate": 1.0855135832375809e-05, + "loss": 0.2098, + "step": 9334 + }, + { + "epoch": 3.91, + "learning_rate": 1.0853446621489042e-05, + "loss": 0.2019, + "step": 9335 + }, + { + "epoch": 3.91, + "learning_rate": 1.0851757386070706e-05, + "loss": 0.1809, + "step": 9336 + }, + { + "epoch": 3.91, + "learning_rate": 1.085006812616936e-05, + "loss": 0.1719, + "step": 9337 + }, + { + "epoch": 3.91, + "learning_rate": 1.084837884183356e-05, + "loss": 0.2099, + "step": 9338 + }, + { + "epoch": 3.91, + "learning_rate": 1.0846689533111865e-05, + "loss": 0.1807, + "step": 9339 + }, + { + "epoch": 3.91, + "learning_rate": 1.0845000200052828e-05, + "loss": 0.198, + "step": 9340 + }, + { + "epoch": 3.91, + "learning_rate": 1.0843310842705011e-05, + "loss": 0.1822, + "step": 9341 + }, + { + "epoch": 3.91, + "learning_rate": 1.0841621461116971e-05, + "loss": 0.2146, + "step": 9342 + }, + { + "epoch": 3.91, + "learning_rate": 1.0839932055337273e-05, + "loss": 0.1954, + "step": 9343 + }, + { + "epoch": 3.91, + "learning_rate": 1.083824262541447e-05, + "loss": 0.1589, + "step": 9344 + }, + { + "epoch": 3.91, + "learning_rate": 1.0836553171397126e-05, + "loss": 0.2323, + "step": 9345 + }, + { + "epoch": 3.91, + "learning_rate": 1.0834863693333805e-05, + "loss": 0.1982, + "step": 9346 + }, + { + "epoch": 3.91, + "learning_rate": 1.083317419127307e-05, + "loss": 0.1637, + "step": 9347 + }, + { + "epoch": 3.91, + "learning_rate": 1.0831484665263479e-05, + "loss": 0.1847, + "step": 9348 + }, + { + "epoch": 3.91, + "learning_rate": 1.0829795115353601e-05, + "loss": 0.1476, + "step": 9349 + }, + { + "epoch": 3.91, + "learning_rate": 1.0828105541591998e-05, + "loss": 0.1757, + "step": 9350 + }, + { + "epoch": 3.92, + "learning_rate": 1.0826415944027235e-05, + "loss": 0.214, + "step": 9351 + }, + { + "epoch": 3.92, + "learning_rate": 1.0824726322707878e-05, + "loss": 0.217, + "step": 9352 + }, + { + "epoch": 3.92, + "learning_rate": 1.0823036677682499e-05, + "loss": 0.1895, + "step": 9353 + }, + { + "epoch": 3.92, + "learning_rate": 1.0821347008999658e-05, + "loss": 0.1615, + "step": 9354 + }, + { + "epoch": 3.92, + "learning_rate": 1.0819657316707926e-05, + "loss": 0.1782, + "step": 9355 + }, + { + "epoch": 3.92, + "learning_rate": 1.0817967600855874e-05, + "loss": 0.1674, + "step": 9356 + }, + { + "epoch": 3.92, + "learning_rate": 1.0816277861492068e-05, + "loss": 0.1784, + "step": 9357 + }, + { + "epoch": 3.92, + "learning_rate": 1.081458809866508e-05, + "loss": 0.1907, + "step": 9358 + }, + { + "epoch": 3.92, + "learning_rate": 1.0812898312423478e-05, + "loss": 0.1914, + "step": 9359 + }, + { + "epoch": 3.92, + "learning_rate": 1.0811208502815838e-05, + "loss": 0.21, + "step": 9360 + }, + { + "epoch": 3.92, + "learning_rate": 1.0809518669890726e-05, + "loss": 0.1671, + "step": 9361 + }, + { + "epoch": 3.92, + "learning_rate": 1.0807828813696719e-05, + "loss": 0.1829, + "step": 9362 + }, + { + "epoch": 3.92, + "learning_rate": 1.0806138934282393e-05, + "loss": 0.2003, + "step": 9363 + }, + { + "epoch": 3.92, + "learning_rate": 1.0804449031696315e-05, + "loss": 0.1604, + "step": 9364 + }, + { + "epoch": 3.92, + "learning_rate": 1.0802759105987065e-05, + "loss": 0.1791, + "step": 9365 + }, + { + "epoch": 3.92, + "learning_rate": 1.0801069157203217e-05, + "loss": 0.1715, + "step": 9366 + }, + { + "epoch": 3.92, + "learning_rate": 1.0799379185393347e-05, + "loss": 0.1648, + "step": 9367 + }, + { + "epoch": 3.92, + "learning_rate": 1.0797689190606028e-05, + "loss": 0.1898, + "step": 9368 + }, + { + "epoch": 3.92, + "learning_rate": 1.0795999172889847e-05, + "loss": 0.1862, + "step": 9369 + }, + { + "epoch": 3.92, + "learning_rate": 1.0794309132293374e-05, + "loss": 0.1888, + "step": 9370 + }, + { + "epoch": 3.92, + "learning_rate": 1.0792619068865189e-05, + "loss": 0.1692, + "step": 9371 + }, + { + "epoch": 3.92, + "learning_rate": 1.0790928982653873e-05, + "loss": 0.2446, + "step": 9372 + }, + { + "epoch": 3.92, + "learning_rate": 1.0789238873708004e-05, + "loss": 0.2284, + "step": 9373 + }, + { + "epoch": 3.92, + "learning_rate": 1.0787548742076164e-05, + "loss": 0.1938, + "step": 9374 + }, + { + "epoch": 3.93, + "learning_rate": 1.0785858587806935e-05, + "loss": 0.1542, + "step": 9375 + }, + { + "epoch": 3.93, + "learning_rate": 1.0784168410948897e-05, + "loss": 0.164, + "step": 9376 + }, + { + "epoch": 3.93, + "learning_rate": 1.0782478211550636e-05, + "loss": 0.1857, + "step": 9377 + }, + { + "epoch": 3.93, + "learning_rate": 1.078078798966073e-05, + "loss": 0.189, + "step": 9378 + }, + { + "epoch": 3.93, + "learning_rate": 1.0779097745327768e-05, + "loss": 0.2207, + "step": 9379 + }, + { + "epoch": 3.93, + "learning_rate": 1.0777407478600334e-05, + "loss": 0.1534, + "step": 9380 + }, + { + "epoch": 3.93, + "learning_rate": 1.0775717189527012e-05, + "loss": 0.2344, + "step": 9381 + }, + { + "epoch": 3.93, + "learning_rate": 1.0774026878156387e-05, + "loss": 0.218, + "step": 9382 + }, + { + "epoch": 3.93, + "learning_rate": 1.0772336544537045e-05, + "loss": 0.2319, + "step": 9383 + }, + { + "epoch": 3.93, + "learning_rate": 1.0770646188717578e-05, + "loss": 0.177, + "step": 9384 + }, + { + "epoch": 3.93, + "learning_rate": 1.0768955810746564e-05, + "loss": 0.1836, + "step": 9385 + }, + { + "epoch": 3.93, + "learning_rate": 1.07672654106726e-05, + "loss": 0.1955, + "step": 9386 + }, + { + "epoch": 3.93, + "learning_rate": 1.0765574988544276e-05, + "loss": 0.1846, + "step": 9387 + }, + { + "epoch": 3.93, + "learning_rate": 1.0763884544410174e-05, + "loss": 0.1924, + "step": 9388 + }, + { + "epoch": 3.93, + "learning_rate": 1.0762194078318891e-05, + "loss": 0.1258, + "step": 9389 + }, + { + "epoch": 3.93, + "learning_rate": 1.0760503590319016e-05, + "loss": 0.1639, + "step": 9390 + }, + { + "epoch": 3.93, + "learning_rate": 1.0758813080459138e-05, + "loss": 0.1734, + "step": 9391 + }, + { + "epoch": 3.93, + "learning_rate": 1.0757122548787855e-05, + "loss": 0.1761, + "step": 9392 + }, + { + "epoch": 3.93, + "learning_rate": 1.0755431995353753e-05, + "loss": 0.199, + "step": 9393 + }, + { + "epoch": 3.93, + "learning_rate": 1.0753741420205431e-05, + "loss": 0.1722, + "step": 9394 + }, + { + "epoch": 3.93, + "learning_rate": 1.0752050823391478e-05, + "loss": 0.1927, + "step": 9395 + }, + { + "epoch": 3.93, + "learning_rate": 1.0750360204960493e-05, + "loss": 0.1614, + "step": 9396 + }, + { + "epoch": 3.93, + "learning_rate": 1.074866956496107e-05, + "loss": 0.2037, + "step": 9397 + }, + { + "epoch": 3.93, + "learning_rate": 1.0746978903441803e-05, + "loss": 0.1731, + "step": 9398 + }, + { + "epoch": 3.94, + "learning_rate": 1.0745288220451291e-05, + "loss": 0.1679, + "step": 9399 + }, + { + "epoch": 3.94, + "learning_rate": 1.0743597516038132e-05, + "loss": 0.1661, + "step": 9400 + }, + { + "epoch": 3.94, + "learning_rate": 1.074190679025092e-05, + "loss": 0.2065, + "step": 9401 + }, + { + "epoch": 3.94, + "learning_rate": 1.0740216043138256e-05, + "loss": 0.1841, + "step": 9402 + }, + { + "epoch": 3.94, + "learning_rate": 1.073852527474874e-05, + "loss": 0.1781, + "step": 9403 + }, + { + "epoch": 3.94, + "learning_rate": 1.0736834485130972e-05, + "loss": 0.1686, + "step": 9404 + }, + { + "epoch": 3.94, + "learning_rate": 1.0735143674333549e-05, + "loss": 0.188, + "step": 9405 + }, + { + "epoch": 3.94, + "learning_rate": 1.073345284240507e-05, + "loss": 0.1495, + "step": 9406 + }, + { + "epoch": 3.94, + "learning_rate": 1.0731761989394145e-05, + "loss": 0.2022, + "step": 9407 + }, + { + "epoch": 3.94, + "learning_rate": 1.0730071115349367e-05, + "loss": 0.2069, + "step": 9408 + }, + { + "epoch": 3.94, + "learning_rate": 1.0728380220319345e-05, + "loss": 0.2076, + "step": 9409 + }, + { + "epoch": 3.94, + "learning_rate": 1.072668930435268e-05, + "loss": 0.2082, + "step": 9410 + }, + { + "epoch": 3.94, + "learning_rate": 1.0724998367497974e-05, + "loss": 0.2061, + "step": 9411 + }, + { + "epoch": 3.94, + "learning_rate": 1.0723307409803834e-05, + "loss": 0.2191, + "step": 9412 + }, + { + "epoch": 3.94, + "learning_rate": 1.0721616431318865e-05, + "loss": 0.1913, + "step": 9413 + }, + { + "epoch": 3.94, + "learning_rate": 1.0719925432091671e-05, + "loss": 0.2349, + "step": 9414 + }, + { + "epoch": 3.94, + "learning_rate": 1.0718234412170862e-05, + "loss": 0.1887, + "step": 9415 + }, + { + "epoch": 3.94, + "learning_rate": 1.0716543371605039e-05, + "loss": 0.1745, + "step": 9416 + }, + { + "epoch": 3.94, + "learning_rate": 1.0714852310442817e-05, + "loss": 0.1596, + "step": 9417 + }, + { + "epoch": 3.94, + "learning_rate": 1.0713161228732795e-05, + "loss": 0.2257, + "step": 9418 + }, + { + "epoch": 3.94, + "learning_rate": 1.0711470126523587e-05, + "loss": 0.2298, + "step": 9419 + }, + { + "epoch": 3.94, + "learning_rate": 1.0709779003863801e-05, + "loss": 0.2116, + "step": 9420 + }, + { + "epoch": 3.94, + "learning_rate": 1.070808786080205e-05, + "loss": 0.1933, + "step": 9421 + }, + { + "epoch": 3.94, + "learning_rate": 1.070639669738694e-05, + "loss": 0.1895, + "step": 9422 + }, + { + "epoch": 3.95, + "learning_rate": 1.0704705513667084e-05, + "loss": 0.2109, + "step": 9423 + }, + { + "epoch": 3.95, + "learning_rate": 1.0703014309691096e-05, + "loss": 0.1643, + "step": 9424 + }, + { + "epoch": 3.95, + "learning_rate": 1.0701323085507583e-05, + "loss": 0.1672, + "step": 9425 + }, + { + "epoch": 3.95, + "learning_rate": 1.0699631841165162e-05, + "loss": 0.1769, + "step": 9426 + }, + { + "epoch": 3.95, + "learning_rate": 1.069794057671244e-05, + "loss": 0.2088, + "step": 9427 + }, + { + "epoch": 3.95, + "learning_rate": 1.069624929219804e-05, + "loss": 0.1518, + "step": 9428 + }, + { + "epoch": 3.95, + "learning_rate": 1.069455798767057e-05, + "loss": 0.1869, + "step": 9429 + }, + { + "epoch": 3.95, + "learning_rate": 1.0692866663178646e-05, + "loss": 0.1764, + "step": 9430 + }, + { + "epoch": 3.95, + "learning_rate": 1.0691175318770887e-05, + "loss": 0.1703, + "step": 9431 + }, + { + "epoch": 3.95, + "learning_rate": 1.0689483954495906e-05, + "loss": 0.1435, + "step": 9432 + }, + { + "epoch": 3.95, + "learning_rate": 1.068779257040232e-05, + "loss": 0.1748, + "step": 9433 + }, + { + "epoch": 3.95, + "learning_rate": 1.068610116653875e-05, + "loss": 0.1518, + "step": 9434 + }, + { + "epoch": 3.95, + "learning_rate": 1.0684409742953805e-05, + "loss": 0.1742, + "step": 9435 + }, + { + "epoch": 3.95, + "learning_rate": 1.0682718299696113e-05, + "loss": 0.1625, + "step": 9436 + }, + { + "epoch": 3.95, + "learning_rate": 1.068102683681429e-05, + "loss": 0.1579, + "step": 9437 + }, + { + "epoch": 3.95, + "learning_rate": 1.0679335354356954e-05, + "loss": 0.2154, + "step": 9438 + }, + { + "epoch": 3.95, + "learning_rate": 1.0677643852372726e-05, + "loss": 0.1967, + "step": 9439 + }, + { + "epoch": 3.95, + "learning_rate": 1.0675952330910224e-05, + "loss": 0.1679, + "step": 9440 + }, + { + "epoch": 3.95, + "learning_rate": 1.0674260790018076e-05, + "loss": 0.1793, + "step": 9441 + }, + { + "epoch": 3.95, + "learning_rate": 1.0672569229744899e-05, + "loss": 0.2155, + "step": 9442 + }, + { + "epoch": 3.95, + "learning_rate": 1.0670877650139316e-05, + "loss": 0.1693, + "step": 9443 + }, + { + "epoch": 3.95, + "learning_rate": 1.0669186051249948e-05, + "loss": 0.1585, + "step": 9444 + }, + { + "epoch": 3.95, + "learning_rate": 1.0667494433125423e-05, + "loss": 0.1796, + "step": 9445 + }, + { + "epoch": 3.95, + "learning_rate": 1.0665802795814362e-05, + "loss": 0.1399, + "step": 9446 + }, + { + "epoch": 3.96, + "learning_rate": 1.0664111139365392e-05, + "loss": 0.1708, + "step": 9447 + }, + { + "epoch": 3.96, + "learning_rate": 1.0662419463827136e-05, + "loss": 0.1761, + "step": 9448 + }, + { + "epoch": 3.96, + "learning_rate": 1.0660727769248221e-05, + "loss": 0.1746, + "step": 9449 + }, + { + "epoch": 3.96, + "learning_rate": 1.0659036055677272e-05, + "loss": 0.1718, + "step": 9450 + }, + { + "epoch": 3.96, + "learning_rate": 1.0657344323162916e-05, + "loss": 0.2022, + "step": 9451 + }, + { + "epoch": 3.96, + "learning_rate": 1.0655652571753781e-05, + "loss": 0.1998, + "step": 9452 + }, + { + "epoch": 3.96, + "learning_rate": 1.0653960801498493e-05, + "loss": 0.1449, + "step": 9453 + }, + { + "epoch": 3.96, + "learning_rate": 1.0652269012445684e-05, + "loss": 0.2073, + "step": 9454 + }, + { + "epoch": 3.96, + "learning_rate": 1.0650577204643981e-05, + "loss": 0.2105, + "step": 9455 + }, + { + "epoch": 3.96, + "learning_rate": 1.0648885378142013e-05, + "loss": 0.159, + "step": 9456 + }, + { + "epoch": 3.96, + "learning_rate": 1.0647193532988409e-05, + "loss": 0.2047, + "step": 9457 + }, + { + "epoch": 3.96, + "learning_rate": 1.0645501669231803e-05, + "loss": 0.1614, + "step": 9458 + }, + { + "epoch": 3.96, + "learning_rate": 1.0643809786920824e-05, + "loss": 0.2312, + "step": 9459 + }, + { + "epoch": 3.96, + "learning_rate": 1.0642117886104106e-05, + "loss": 0.195, + "step": 9460 + }, + { + "epoch": 3.96, + "learning_rate": 1.0640425966830276e-05, + "loss": 0.1706, + "step": 9461 + }, + { + "epoch": 3.96, + "learning_rate": 1.0638734029147974e-05, + "loss": 0.1486, + "step": 9462 + }, + { + "epoch": 3.96, + "learning_rate": 1.0637042073105826e-05, + "loss": 0.1422, + "step": 9463 + }, + { + "epoch": 3.96, + "learning_rate": 1.0635350098752468e-05, + "loss": 0.1756, + "step": 9464 + }, + { + "epoch": 3.96, + "learning_rate": 1.0633658106136538e-05, + "loss": 0.2265, + "step": 9465 + }, + { + "epoch": 3.96, + "learning_rate": 1.0631966095306668e-05, + "loss": 0.2164, + "step": 9466 + }, + { + "epoch": 3.96, + "learning_rate": 1.063027406631149e-05, + "loss": 0.204, + "step": 9467 + }, + { + "epoch": 3.96, + "learning_rate": 1.0628582019199648e-05, + "loss": 0.2023, + "step": 9468 + }, + { + "epoch": 3.96, + "learning_rate": 1.0626889954019768e-05, + "loss": 0.1844, + "step": 9469 + }, + { + "epoch": 3.96, + "learning_rate": 1.0625197870820493e-05, + "loss": 0.1986, + "step": 9470 + }, + { + "epoch": 3.97, + "learning_rate": 1.0623505769650466e-05, + "loss": 0.1838, + "step": 9471 + }, + { + "epoch": 3.97, + "learning_rate": 1.0621813650558317e-05, + "loss": 0.1745, + "step": 9472 + }, + { + "epoch": 3.97, + "learning_rate": 1.0620121513592682e-05, + "loss": 0.1978, + "step": 9473 + }, + { + "epoch": 3.97, + "learning_rate": 1.0618429358802209e-05, + "loss": 0.2103, + "step": 9474 + }, + { + "epoch": 3.97, + "learning_rate": 1.061673718623553e-05, + "loss": 0.1763, + "step": 9475 + }, + { + "epoch": 3.97, + "learning_rate": 1.061504499594129e-05, + "loss": 0.1444, + "step": 9476 + }, + { + "epoch": 3.97, + "learning_rate": 1.0613352787968125e-05, + "loss": 0.1943, + "step": 9477 + }, + { + "epoch": 3.97, + "learning_rate": 1.061166056236468e-05, + "loss": 0.1833, + "step": 9478 + }, + { + "epoch": 3.97, + "learning_rate": 1.0609968319179595e-05, + "loss": 0.1546, + "step": 9479 + }, + { + "epoch": 3.97, + "learning_rate": 1.060827605846151e-05, + "loss": 0.2261, + "step": 9480 + }, + { + "epoch": 3.97, + "learning_rate": 1.0606583780259073e-05, + "loss": 0.1785, + "step": 9481 + }, + { + "epoch": 3.97, + "learning_rate": 1.060489148462092e-05, + "loss": 0.1689, + "step": 9482 + }, + { + "epoch": 3.97, + "learning_rate": 1.0603199171595701e-05, + "loss": 0.1889, + "step": 9483 + }, + { + "epoch": 3.97, + "learning_rate": 1.0601506841232057e-05, + "loss": 0.1917, + "step": 9484 + }, + { + "epoch": 3.97, + "learning_rate": 1.059981449357863e-05, + "loss": 0.2113, + "step": 9485 + }, + { + "epoch": 3.97, + "learning_rate": 1.0598122128684069e-05, + "loss": 0.1792, + "step": 9486 + }, + { + "epoch": 3.97, + "learning_rate": 1.0596429746597017e-05, + "loss": 0.2004, + "step": 9487 + }, + { + "epoch": 3.97, + "learning_rate": 1.0594737347366125e-05, + "loss": 0.166, + "step": 9488 + }, + { + "epoch": 3.97, + "learning_rate": 1.059304493104003e-05, + "loss": 0.1494, + "step": 9489 + }, + { + "epoch": 3.97, + "learning_rate": 1.0591352497667388e-05, + "loss": 0.179, + "step": 9490 + }, + { + "epoch": 3.97, + "learning_rate": 1.0589660047296844e-05, + "loss": 0.1768, + "step": 9491 + }, + { + "epoch": 3.97, + "learning_rate": 1.0587967579977043e-05, + "loss": 0.2044, + "step": 9492 + }, + { + "epoch": 3.97, + "learning_rate": 1.0586275095756638e-05, + "loss": 0.1723, + "step": 9493 + }, + { + "epoch": 3.97, + "learning_rate": 1.0584582594684274e-05, + "loss": 0.1643, + "step": 9494 + }, + { + "epoch": 3.98, + "learning_rate": 1.0582890076808603e-05, + "loss": 0.1814, + "step": 9495 + }, + { + "epoch": 3.98, + "learning_rate": 1.0581197542178274e-05, + "loss": 0.1636, + "step": 9496 + }, + { + "epoch": 3.98, + "learning_rate": 1.0579504990841935e-05, + "loss": 0.2118, + "step": 9497 + }, + { + "epoch": 3.98, + "learning_rate": 1.0577812422848241e-05, + "loss": 0.1787, + "step": 9498 + }, + { + "epoch": 3.98, + "learning_rate": 1.0576119838245843e-05, + "loss": 0.211, + "step": 9499 + }, + { + "epoch": 3.98, + "learning_rate": 1.057442723708339e-05, + "loss": 0.1991, + "step": 9500 + }, + { + "epoch": 3.98, + "learning_rate": 1.0572734619409535e-05, + "loss": 0.1995, + "step": 9501 + }, + { + "epoch": 3.98, + "learning_rate": 1.0571041985272934e-05, + "loss": 0.1729, + "step": 9502 + }, + { + "epoch": 3.98, + "learning_rate": 1.0569349334722236e-05, + "loss": 0.1999, + "step": 9503 + }, + { + "epoch": 3.98, + "learning_rate": 1.0567656667806097e-05, + "loss": 0.192, + "step": 9504 + }, + { + "epoch": 3.98, + "learning_rate": 1.0565963984573174e-05, + "loss": 0.1887, + "step": 9505 + }, + { + "epoch": 3.98, + "learning_rate": 1.0564271285072117e-05, + "loss": 0.1744, + "step": 9506 + }, + { + "epoch": 3.98, + "learning_rate": 1.0562578569351579e-05, + "loss": 0.1877, + "step": 9507 + }, + { + "epoch": 3.98, + "learning_rate": 1.056088583746022e-05, + "loss": 0.1796, + "step": 9508 + }, + { + "epoch": 3.98, + "learning_rate": 1.0559193089446698e-05, + "loss": 0.2141, + "step": 9509 + }, + { + "epoch": 3.98, + "learning_rate": 1.0557500325359665e-05, + "loss": 0.1828, + "step": 9510 + }, + { + "epoch": 3.98, + "learning_rate": 1.055580754524778e-05, + "loss": 0.1784, + "step": 9511 + }, + { + "epoch": 3.98, + "learning_rate": 1.05541147491597e-05, + "loss": 0.1886, + "step": 9512 + }, + { + "epoch": 3.98, + "learning_rate": 1.0552421937144083e-05, + "loss": 0.1875, + "step": 9513 + }, + { + "epoch": 3.98, + "learning_rate": 1.055072910924959e-05, + "loss": 0.1703, + "step": 9514 + }, + { + "epoch": 3.98, + "learning_rate": 1.0549036265524877e-05, + "loss": 0.1968, + "step": 9515 + }, + { + "epoch": 3.98, + "learning_rate": 1.0547343406018602e-05, + "loss": 0.1663, + "step": 9516 + }, + { + "epoch": 3.98, + "learning_rate": 1.0545650530779428e-05, + "loss": 0.1886, + "step": 9517 + }, + { + "epoch": 3.98, + "learning_rate": 1.0543957639856014e-05, + "loss": 0.1585, + "step": 9518 + }, + { + "epoch": 3.99, + "learning_rate": 1.0542264733297021e-05, + "loss": 0.1914, + "step": 9519 + }, + { + "epoch": 3.99, + "learning_rate": 1.0540571811151106e-05, + "loss": 0.163, + "step": 9520 + }, + { + "epoch": 3.99, + "learning_rate": 1.0538878873466936e-05, + "loss": 0.2158, + "step": 9521 + }, + { + "epoch": 3.99, + "learning_rate": 1.0537185920293171e-05, + "loss": 0.1514, + "step": 9522 + }, + { + "epoch": 3.99, + "learning_rate": 1.0535492951678471e-05, + "loss": 0.1868, + "step": 9523 + }, + { + "epoch": 3.99, + "learning_rate": 1.0533799967671503e-05, + "loss": 0.1927, + "step": 9524 + }, + { + "epoch": 3.99, + "learning_rate": 1.053210696832093e-05, + "loss": 0.1469, + "step": 9525 + }, + { + "epoch": 3.99, + "learning_rate": 1.0530413953675414e-05, + "loss": 0.1823, + "step": 9526 + }, + { + "epoch": 3.99, + "learning_rate": 1.0528720923783617e-05, + "loss": 0.2002, + "step": 9527 + }, + { + "epoch": 3.99, + "learning_rate": 1.0527027878694212e-05, + "loss": 0.2211, + "step": 9528 + }, + { + "epoch": 3.99, + "learning_rate": 1.0525334818455855e-05, + "loss": 0.1998, + "step": 9529 + }, + { + "epoch": 3.99, + "learning_rate": 1.0523641743117215e-05, + "loss": 0.1809, + "step": 9530 + }, + { + "epoch": 3.99, + "learning_rate": 1.0521948652726955e-05, + "loss": 0.1798, + "step": 9531 + }, + { + "epoch": 3.99, + "learning_rate": 1.0520255547333744e-05, + "loss": 0.1796, + "step": 9532 + }, + { + "epoch": 3.99, + "learning_rate": 1.0518562426986253e-05, + "loss": 0.202, + "step": 9533 + }, + { + "epoch": 3.99, + "learning_rate": 1.0516869291733142e-05, + "loss": 0.1884, + "step": 9534 + }, + { + "epoch": 3.99, + "learning_rate": 1.0515176141623085e-05, + "loss": 0.1454, + "step": 9535 + }, + { + "epoch": 3.99, + "learning_rate": 1.0513482976704746e-05, + "loss": 0.1957, + "step": 9536 + }, + { + "epoch": 3.99, + "learning_rate": 1.0511789797026792e-05, + "loss": 0.2049, + "step": 9537 + }, + { + "epoch": 3.99, + "learning_rate": 1.0510096602637896e-05, + "loss": 0.179, + "step": 9538 + }, + { + "epoch": 3.99, + "learning_rate": 1.0508403393586728e-05, + "loss": 0.1709, + "step": 9539 + }, + { + "epoch": 3.99, + "learning_rate": 1.0506710169921954e-05, + "loss": 0.1893, + "step": 9540 + }, + { + "epoch": 3.99, + "learning_rate": 1.0505016931692244e-05, + "loss": 0.1522, + "step": 9541 + }, + { + "epoch": 3.99, + "learning_rate": 1.0503323678946272e-05, + "loss": 0.1592, + "step": 9542 + }, + { + "epoch": 4.0, + "learning_rate": 1.0501630411732709e-05, + "loss": 0.1703, + "step": 9543 + }, + { + "epoch": 4.0, + "learning_rate": 1.0499937130100222e-05, + "loss": 0.1883, + "step": 9544 + }, + { + "epoch": 4.0, + "learning_rate": 1.0498243834097485e-05, + "loss": 0.1816, + "step": 9545 + }, + { + "epoch": 4.0, + "learning_rate": 1.0496550523773177e-05, + "loss": 0.1701, + "step": 9546 + }, + { + "epoch": 4.0, + "learning_rate": 1.0494857199175961e-05, + "loss": 0.174, + "step": 9547 + }, + { + "epoch": 4.0, + "learning_rate": 1.0493163860354512e-05, + "loss": 0.1885, + "step": 9548 + }, + { + "epoch": 4.0, + "learning_rate": 1.0491470507357508e-05, + "loss": 0.2248, + "step": 9549 + }, + { + "epoch": 4.0, + "learning_rate": 1.0489777140233619e-05, + "loss": 0.1783, + "step": 9550 + }, + { + "epoch": 4.0, + "learning_rate": 1.0488083759031524e-05, + "loss": 0.1625, + "step": 9551 + }, + { + "epoch": 4.0, + "learning_rate": 1.0486390363799891e-05, + "loss": 0.2177, + "step": 9552 + }, + { + "epoch": 4.0, + "learning_rate": 1.04846969545874e-05, + "loss": 0.1887, + "step": 9553 + }, + { + "epoch": 4.0, + "learning_rate": 1.0483003531442726e-05, + "loss": 0.1784, + "step": 9554 + }, + { + "epoch": 4.0, + "learning_rate": 1.048131009441454e-05, + "loss": 0.1487, + "step": 9555 + }, + { + "epoch": 4.0, + "learning_rate": 1.047961664355153e-05, + "loss": 0.1338, + "step": 9556 + }, + { + "epoch": 4.0, + "learning_rate": 1.0477923178902359e-05, + "loss": 0.1734, + "step": 9557 + }, + { + "epoch": 4.0, + "learning_rate": 1.047622970051571e-05, + "loss": 0.133, + "step": 9558 + }, + { + "epoch": 4.0, + "learning_rate": 1.0474536208440263e-05, + "loss": 0.1673, + "step": 9559 + }, + { + "epoch": 4.0, + "learning_rate": 1.0472842702724695e-05, + "loss": 0.1659, + "step": 9560 + }, + { + "epoch": 4.0, + "learning_rate": 1.047114918341768e-05, + "loss": 0.1201, + "step": 9561 + }, + { + "epoch": 4.0, + "learning_rate": 1.0469455650567903e-05, + "loss": 0.1705, + "step": 9562 + }, + { + "epoch": 4.0, + "learning_rate": 1.0467762104224039e-05, + "loss": 0.164, + "step": 9563 + }, + { + "epoch": 4.0, + "learning_rate": 1.0466068544434766e-05, + "loss": 0.1443, + "step": 9564 + }, + { + "epoch": 4.0, + "learning_rate": 1.046437497124877e-05, + "loss": 0.1925, + "step": 9565 + }, + { + "epoch": 4.01, + "learning_rate": 1.0462681384714723e-05, + "loss": 0.1494, + "step": 9566 + }, + { + "epoch": 4.01, + "learning_rate": 1.0460987784881313e-05, + "loss": 0.1724, + "step": 9567 + }, + { + "epoch": 4.01, + "learning_rate": 1.0459294171797218e-05, + "loss": 0.1702, + "step": 9568 + }, + { + "epoch": 4.01, + "learning_rate": 1.045760054551112e-05, + "loss": 0.1659, + "step": 9569 + }, + { + "epoch": 4.01, + "learning_rate": 1.04559069060717e-05, + "loss": 0.1584, + "step": 9570 + }, + { + "epoch": 4.01, + "learning_rate": 1.045421325352764e-05, + "loss": 0.1523, + "step": 9571 + }, + { + "epoch": 4.01, + "learning_rate": 1.0452519587927623e-05, + "loss": 0.1254, + "step": 9572 + }, + { + "epoch": 4.01, + "learning_rate": 1.0450825909320335e-05, + "loss": 0.136, + "step": 9573 + }, + { + "epoch": 4.01, + "learning_rate": 1.0449132217754455e-05, + "loss": 0.1503, + "step": 9574 + }, + { + "epoch": 4.01, + "learning_rate": 1.0447438513278664e-05, + "loss": 0.1935, + "step": 9575 + }, + { + "epoch": 4.01, + "learning_rate": 1.0445744795941652e-05, + "loss": 0.146, + "step": 9576 + }, + { + "epoch": 4.01, + "learning_rate": 1.0444051065792105e-05, + "loss": 0.192, + "step": 9577 + }, + { + "epoch": 4.01, + "learning_rate": 1.04423573228787e-05, + "loss": 0.0897, + "step": 9578 + }, + { + "epoch": 4.01, + "learning_rate": 1.0440663567250127e-05, + "loss": 0.1551, + "step": 9579 + }, + { + "epoch": 4.01, + "learning_rate": 1.0438969798955072e-05, + "loss": 0.1595, + "step": 9580 + }, + { + "epoch": 4.01, + "learning_rate": 1.0437276018042219e-05, + "loss": 0.1983, + "step": 9581 + }, + { + "epoch": 4.01, + "learning_rate": 1.0435582224560255e-05, + "loss": 0.1687, + "step": 9582 + }, + { + "epoch": 4.01, + "learning_rate": 1.0433888418557867e-05, + "loss": 0.1532, + "step": 9583 + }, + { + "epoch": 4.01, + "learning_rate": 1.043219460008374e-05, + "loss": 0.1476, + "step": 9584 + }, + { + "epoch": 4.01, + "learning_rate": 1.0430500769186565e-05, + "loss": 0.1043, + "step": 9585 + }, + { + "epoch": 4.01, + "learning_rate": 1.0428806925915026e-05, + "loss": 0.1767, + "step": 9586 + }, + { + "epoch": 4.01, + "learning_rate": 1.0427113070317813e-05, + "loss": 0.1923, + "step": 9587 + }, + { + "epoch": 4.01, + "learning_rate": 1.0425419202443614e-05, + "loss": 0.1675, + "step": 9588 + }, + { + "epoch": 4.01, + "learning_rate": 1.0423725322341115e-05, + "loss": 0.1778, + "step": 9589 + }, + { + "epoch": 4.02, + "learning_rate": 1.0422031430059011e-05, + "loss": 0.1971, + "step": 9590 + }, + { + "epoch": 4.02, + "learning_rate": 1.0420337525645984e-05, + "loss": 0.1652, + "step": 9591 + }, + { + "epoch": 4.02, + "learning_rate": 1.0418643609150732e-05, + "loss": 0.1324, + "step": 9592 + }, + { + "epoch": 4.02, + "learning_rate": 1.041694968062194e-05, + "loss": 0.1142, + "step": 9593 + }, + { + "epoch": 4.02, + "learning_rate": 1.0415255740108299e-05, + "loss": 0.1709, + "step": 9594 + }, + { + "epoch": 4.02, + "learning_rate": 1.0413561787658497e-05, + "loss": 0.1278, + "step": 9595 + }, + { + "epoch": 4.02, + "learning_rate": 1.0411867823321233e-05, + "loss": 0.1172, + "step": 9596 + }, + { + "epoch": 4.02, + "learning_rate": 1.0410173847145195e-05, + "loss": 0.1617, + "step": 9597 + }, + { + "epoch": 4.02, + "learning_rate": 1.040847985917907e-05, + "loss": 0.1151, + "step": 9598 + }, + { + "epoch": 4.02, + "learning_rate": 1.0406785859471554e-05, + "loss": 0.1156, + "step": 9599 + }, + { + "epoch": 4.02, + "learning_rate": 1.0405091848071342e-05, + "loss": 0.1262, + "step": 9600 + }, + { + "epoch": 4.02, + "learning_rate": 1.0403397825027122e-05, + "loss": 0.157, + "step": 9601 + }, + { + "epoch": 4.02, + "learning_rate": 1.0401703790387589e-05, + "loss": 0.175, + "step": 9602 + }, + { + "epoch": 4.02, + "learning_rate": 1.0400009744201438e-05, + "loss": 0.1851, + "step": 9603 + }, + { + "epoch": 4.02, + "learning_rate": 1.0398315686517362e-05, + "loss": 0.1766, + "step": 9604 + }, + { + "epoch": 4.02, + "learning_rate": 1.0396621617384055e-05, + "loss": 0.1459, + "step": 9605 + }, + { + "epoch": 4.02, + "learning_rate": 1.039492753685021e-05, + "loss": 0.1324, + "step": 9606 + }, + { + "epoch": 4.02, + "learning_rate": 1.0393233444964528e-05, + "loss": 0.1526, + "step": 9607 + }, + { + "epoch": 4.02, + "learning_rate": 1.0391539341775696e-05, + "loss": 0.1222, + "step": 9608 + }, + { + "epoch": 4.02, + "learning_rate": 1.0389845227332413e-05, + "loss": 0.1661, + "step": 9609 + }, + { + "epoch": 4.02, + "learning_rate": 1.0388151101683375e-05, + "loss": 0.1224, + "step": 9610 + }, + { + "epoch": 4.02, + "learning_rate": 1.0386456964877281e-05, + "loss": 0.154, + "step": 9611 + }, + { + "epoch": 4.02, + "learning_rate": 1.038476281696282e-05, + "loss": 0.1473, + "step": 9612 + }, + { + "epoch": 4.02, + "learning_rate": 1.0383068657988695e-05, + "loss": 0.0998, + "step": 9613 + }, + { + "epoch": 4.03, + "learning_rate": 1.0381374488003603e-05, + "loss": 0.1953, + "step": 9614 + }, + { + "epoch": 4.03, + "learning_rate": 1.0379680307056238e-05, + "loss": 0.1348, + "step": 9615 + }, + { + "epoch": 4.03, + "learning_rate": 1.03779861151953e-05, + "loss": 0.1649, + "step": 9616 + }, + { + "epoch": 4.03, + "learning_rate": 1.0376291912469488e-05, + "loss": 0.1258, + "step": 9617 + }, + { + "epoch": 4.03, + "learning_rate": 1.0374597698927496e-05, + "loss": 0.1482, + "step": 9618 + }, + { + "epoch": 4.03, + "learning_rate": 1.0372903474618027e-05, + "loss": 0.1671, + "step": 9619 + }, + { + "epoch": 4.03, + "learning_rate": 1.0371209239589779e-05, + "loss": 0.1486, + "step": 9620 + }, + { + "epoch": 4.03, + "learning_rate": 1.0369514993891451e-05, + "loss": 0.1308, + "step": 9621 + }, + { + "epoch": 4.03, + "learning_rate": 1.036782073757174e-05, + "loss": 0.1495, + "step": 9622 + }, + { + "epoch": 4.03, + "learning_rate": 1.036612647067935e-05, + "loss": 0.0969, + "step": 9623 + }, + { + "epoch": 4.03, + "learning_rate": 1.036443219326298e-05, + "loss": 0.184, + "step": 9624 + }, + { + "epoch": 4.03, + "learning_rate": 1.036273790537133e-05, + "loss": 0.127, + "step": 9625 + }, + { + "epoch": 4.03, + "learning_rate": 1.0361043607053101e-05, + "loss": 0.1336, + "step": 9626 + }, + { + "epoch": 4.03, + "learning_rate": 1.0359349298356993e-05, + "loss": 0.1505, + "step": 9627 + }, + { + "epoch": 4.03, + "learning_rate": 1.0357654979331708e-05, + "loss": 0.1709, + "step": 9628 + }, + { + "epoch": 4.03, + "learning_rate": 1.0355960650025946e-05, + "loss": 0.1573, + "step": 9629 + }, + { + "epoch": 4.03, + "learning_rate": 1.0354266310488416e-05, + "loss": 0.2003, + "step": 9630 + }, + { + "epoch": 4.03, + "learning_rate": 1.0352571960767814e-05, + "loss": 0.1725, + "step": 9631 + }, + { + "epoch": 4.03, + "learning_rate": 1.0350877600912844e-05, + "loss": 0.1442, + "step": 9632 + }, + { + "epoch": 4.03, + "learning_rate": 1.0349183230972204e-05, + "loss": 0.1851, + "step": 9633 + }, + { + "epoch": 4.03, + "learning_rate": 1.0347488850994608e-05, + "loss": 0.1321, + "step": 9634 + }, + { + "epoch": 4.03, + "learning_rate": 1.034579446102875e-05, + "loss": 0.1416, + "step": 9635 + }, + { + "epoch": 4.03, + "learning_rate": 1.0344100061123337e-05, + "loss": 0.1843, + "step": 9636 + }, + { + "epoch": 4.03, + "learning_rate": 1.0342405651327072e-05, + "loss": 0.1499, + "step": 9637 + }, + { + "epoch": 4.04, + "learning_rate": 1.0340711231688662e-05, + "loss": 0.1437, + "step": 9638 + }, + { + "epoch": 4.04, + "learning_rate": 1.033901680225681e-05, + "loss": 0.1823, + "step": 9639 + }, + { + "epoch": 4.04, + "learning_rate": 1.0337322363080219e-05, + "loss": 0.1039, + "step": 9640 + }, + { + "epoch": 4.04, + "learning_rate": 1.0335627914207602e-05, + "loss": 0.1494, + "step": 9641 + }, + { + "epoch": 4.04, + "learning_rate": 1.0333933455687655e-05, + "loss": 0.166, + "step": 9642 + }, + { + "epoch": 4.04, + "learning_rate": 1.0332238987569086e-05, + "loss": 0.1793, + "step": 9643 + }, + { + "epoch": 4.04, + "learning_rate": 1.0330544509900603e-05, + "loss": 0.1315, + "step": 9644 + }, + { + "epoch": 4.04, + "learning_rate": 1.032885002273091e-05, + "loss": 0.1267, + "step": 9645 + }, + { + "epoch": 4.04, + "learning_rate": 1.0327155526108716e-05, + "loss": 0.1621, + "step": 9646 + }, + { + "epoch": 4.04, + "learning_rate": 1.0325461020082726e-05, + "loss": 0.1321, + "step": 9647 + }, + { + "epoch": 4.04, + "learning_rate": 1.0323766504701649e-05, + "loss": 0.1195, + "step": 9648 + }, + { + "epoch": 4.04, + "learning_rate": 1.032207198001419e-05, + "loss": 0.1324, + "step": 9649 + }, + { + "epoch": 4.04, + "learning_rate": 1.0320377446069059e-05, + "loss": 0.1025, + "step": 9650 + }, + { + "epoch": 4.04, + "learning_rate": 1.0318682902914962e-05, + "loss": 0.1751, + "step": 9651 + }, + { + "epoch": 4.04, + "learning_rate": 1.0316988350600608e-05, + "loss": 0.1372, + "step": 9652 + }, + { + "epoch": 4.04, + "learning_rate": 1.0315293789174705e-05, + "loss": 0.2112, + "step": 9653 + }, + { + "epoch": 4.04, + "learning_rate": 1.0313599218685961e-05, + "loss": 0.1442, + "step": 9654 + }, + { + "epoch": 4.04, + "learning_rate": 1.0311904639183088e-05, + "loss": 0.0953, + "step": 9655 + }, + { + "epoch": 4.04, + "learning_rate": 1.031021005071479e-05, + "loss": 0.1931, + "step": 9656 + }, + { + "epoch": 4.04, + "learning_rate": 1.030851545332978e-05, + "loss": 0.1386, + "step": 9657 + }, + { + "epoch": 4.04, + "learning_rate": 1.0306820847076768e-05, + "loss": 0.1469, + "step": 9658 + }, + { + "epoch": 4.04, + "learning_rate": 1.0305126232004464e-05, + "loss": 0.1834, + "step": 9659 + }, + { + "epoch": 4.04, + "learning_rate": 1.0303431608161574e-05, + "loss": 0.1344, + "step": 9660 + }, + { + "epoch": 4.04, + "learning_rate": 1.0301736975596816e-05, + "loss": 0.1186, + "step": 9661 + }, + { + "epoch": 4.05, + "learning_rate": 1.0300042334358893e-05, + "loss": 0.1086, + "step": 9662 + }, + { + "epoch": 4.05, + "learning_rate": 1.029834768449652e-05, + "loss": 0.1411, + "step": 9663 + }, + { + "epoch": 4.05, + "learning_rate": 1.029665302605841e-05, + "loss": 0.1346, + "step": 9664 + }, + { + "epoch": 4.05, + "learning_rate": 1.029495835909327e-05, + "loss": 0.1431, + "step": 9665 + }, + { + "epoch": 4.05, + "learning_rate": 1.0293263683649813e-05, + "loss": 0.1703, + "step": 9666 + }, + { + "epoch": 4.05, + "learning_rate": 1.0291568999776751e-05, + "loss": 0.164, + "step": 9667 + }, + { + "epoch": 4.05, + "learning_rate": 1.0289874307522801e-05, + "loss": 0.131, + "step": 9668 + }, + { + "epoch": 4.05, + "learning_rate": 1.0288179606936666e-05, + "loss": 0.1263, + "step": 9669 + }, + { + "epoch": 4.05, + "learning_rate": 1.0286484898067067e-05, + "loss": 0.1508, + "step": 9670 + }, + { + "epoch": 4.05, + "learning_rate": 1.0284790180962712e-05, + "loss": 0.1621, + "step": 9671 + }, + { + "epoch": 4.05, + "learning_rate": 1.0283095455672319e-05, + "loss": 0.0984, + "step": 9672 + }, + { + "epoch": 4.05, + "learning_rate": 1.0281400722244594e-05, + "loss": 0.1535, + "step": 9673 + }, + { + "epoch": 4.05, + "learning_rate": 1.0279705980728258e-05, + "loss": 0.1426, + "step": 9674 + }, + { + "epoch": 4.05, + "learning_rate": 1.0278011231172024e-05, + "loss": 0.1577, + "step": 9675 + }, + { + "epoch": 4.05, + "learning_rate": 1.02763164736246e-05, + "loss": 0.1846, + "step": 9676 + }, + { + "epoch": 4.05, + "learning_rate": 1.0274621708134705e-05, + "loss": 0.1882, + "step": 9677 + }, + { + "epoch": 4.05, + "learning_rate": 1.0272926934751053e-05, + "loss": 0.0994, + "step": 9678 + }, + { + "epoch": 4.05, + "learning_rate": 1.0271232153522359e-05, + "loss": 0.193, + "step": 9679 + }, + { + "epoch": 4.05, + "learning_rate": 1.0269537364497337e-05, + "loss": 0.1684, + "step": 9680 + }, + { + "epoch": 4.05, + "learning_rate": 1.0267842567724702e-05, + "loss": 0.1234, + "step": 9681 + }, + { + "epoch": 4.05, + "learning_rate": 1.0266147763253171e-05, + "loss": 0.1486, + "step": 9682 + }, + { + "epoch": 4.05, + "learning_rate": 1.0264452951131459e-05, + "loss": 0.1551, + "step": 9683 + }, + { + "epoch": 4.05, + "learning_rate": 1.026275813140828e-05, + "loss": 0.1232, + "step": 9684 + }, + { + "epoch": 4.05, + "learning_rate": 1.0261063304132353e-05, + "loss": 0.1866, + "step": 9685 + }, + { + "epoch": 4.06, + "learning_rate": 1.025936846935239e-05, + "loss": 0.1786, + "step": 9686 + }, + { + "epoch": 4.06, + "learning_rate": 1.0257673627117115e-05, + "loss": 0.1304, + "step": 9687 + }, + { + "epoch": 4.06, + "learning_rate": 1.0255978777475238e-05, + "loss": 0.1339, + "step": 9688 + }, + { + "epoch": 4.06, + "learning_rate": 1.025428392047548e-05, + "loss": 0.1277, + "step": 9689 + }, + { + "epoch": 4.06, + "learning_rate": 1.0252589056166554e-05, + "loss": 0.1696, + "step": 9690 + }, + { + "epoch": 4.06, + "learning_rate": 1.025089418459718e-05, + "loss": 0.1633, + "step": 9691 + }, + { + "epoch": 4.06, + "learning_rate": 1.0249199305816078e-05, + "loss": 0.179, + "step": 9692 + }, + { + "epoch": 4.06, + "learning_rate": 1.024750441987196e-05, + "loss": 0.1495, + "step": 9693 + }, + { + "epoch": 4.06, + "learning_rate": 1.0245809526813547e-05, + "loss": 0.1376, + "step": 9694 + }, + { + "epoch": 4.06, + "learning_rate": 1.0244114626689558e-05, + "loss": 0.1755, + "step": 9695 + }, + { + "epoch": 4.06, + "learning_rate": 1.0242419719548711e-05, + "loss": 0.0943, + "step": 9696 + }, + { + "epoch": 4.06, + "learning_rate": 1.0240724805439724e-05, + "loss": 0.1556, + "step": 9697 + }, + { + "epoch": 4.06, + "learning_rate": 1.0239029884411318e-05, + "loss": 0.1504, + "step": 9698 + }, + { + "epoch": 4.06, + "learning_rate": 1.0237334956512209e-05, + "loss": 0.1953, + "step": 9699 + }, + { + "epoch": 4.06, + "learning_rate": 1.0235640021791115e-05, + "loss": 0.1541, + "step": 9700 + }, + { + "epoch": 4.06, + "learning_rate": 1.0233945080296758e-05, + "loss": 0.1627, + "step": 9701 + }, + { + "epoch": 4.06, + "learning_rate": 1.0232250132077861e-05, + "loss": 0.1356, + "step": 9702 + }, + { + "epoch": 4.06, + "learning_rate": 1.0230555177183136e-05, + "loss": 0.1696, + "step": 9703 + }, + { + "epoch": 4.06, + "learning_rate": 1.0228860215661309e-05, + "loss": 0.1229, + "step": 9704 + }, + { + "epoch": 4.06, + "learning_rate": 1.02271652475611e-05, + "loss": 0.1677, + "step": 9705 + }, + { + "epoch": 4.06, + "learning_rate": 1.0225470272931224e-05, + "loss": 0.0818, + "step": 9706 + }, + { + "epoch": 4.06, + "learning_rate": 1.0223775291820406e-05, + "loss": 0.163, + "step": 9707 + }, + { + "epoch": 4.06, + "learning_rate": 1.0222080304277366e-05, + "loss": 0.1689, + "step": 9708 + }, + { + "epoch": 4.06, + "learning_rate": 1.0220385310350829e-05, + "loss": 0.1454, + "step": 9709 + }, + { + "epoch": 4.07, + "learning_rate": 1.0218690310089508e-05, + "loss": 0.1118, + "step": 9710 + }, + { + "epoch": 4.07, + "learning_rate": 1.0216995303542126e-05, + "loss": 0.1471, + "step": 9711 + }, + { + "epoch": 4.07, + "learning_rate": 1.021530029075741e-05, + "loss": 0.1493, + "step": 9712 + }, + { + "epoch": 4.07, + "learning_rate": 1.0213605271784076e-05, + "loss": 0.156, + "step": 9713 + }, + { + "epoch": 4.07, + "learning_rate": 1.0211910246670848e-05, + "loss": 0.1622, + "step": 9714 + }, + { + "epoch": 4.07, + "learning_rate": 1.0210215215466447e-05, + "loss": 0.1617, + "step": 9715 + }, + { + "epoch": 4.07, + "learning_rate": 1.0208520178219602e-05, + "loss": 0.1603, + "step": 9716 + }, + { + "epoch": 4.07, + "learning_rate": 1.0206825134979024e-05, + "loss": 0.1312, + "step": 9717 + }, + { + "epoch": 4.07, + "learning_rate": 1.020513008579344e-05, + "loss": 0.1757, + "step": 9718 + }, + { + "epoch": 4.07, + "learning_rate": 1.0203435030711577e-05, + "loss": 0.1677, + "step": 9719 + }, + { + "epoch": 4.07, + "learning_rate": 1.0201739969782154e-05, + "loss": 0.1854, + "step": 9720 + }, + { + "epoch": 4.07, + "learning_rate": 1.0200044903053896e-05, + "loss": 0.1221, + "step": 9721 + }, + { + "epoch": 4.07, + "learning_rate": 1.0198349830575522e-05, + "loss": 0.1636, + "step": 9722 + }, + { + "epoch": 4.07, + "learning_rate": 1.0196654752395761e-05, + "loss": 0.1262, + "step": 9723 + }, + { + "epoch": 4.07, + "learning_rate": 1.0194959668563332e-05, + "loss": 0.1673, + "step": 9724 + }, + { + "epoch": 4.07, + "learning_rate": 1.0193264579126957e-05, + "loss": 0.1389, + "step": 9725 + }, + { + "epoch": 4.07, + "learning_rate": 1.0191569484135368e-05, + "loss": 0.1372, + "step": 9726 + }, + { + "epoch": 4.07, + "learning_rate": 1.0189874383637284e-05, + "loss": 0.149, + "step": 9727 + }, + { + "epoch": 4.07, + "learning_rate": 1.0188179277681426e-05, + "loss": 0.1391, + "step": 9728 + }, + { + "epoch": 4.07, + "learning_rate": 1.0186484166316526e-05, + "loss": 0.1838, + "step": 9729 + }, + { + "epoch": 4.07, + "learning_rate": 1.01847890495913e-05, + "loss": 0.1652, + "step": 9730 + }, + { + "epoch": 4.07, + "learning_rate": 1.0183093927554478e-05, + "loss": 0.1445, + "step": 9731 + }, + { + "epoch": 4.07, + "learning_rate": 1.0181398800254785e-05, + "loss": 0.1057, + "step": 9732 + }, + { + "epoch": 4.07, + "learning_rate": 1.0179703667740946e-05, + "loss": 0.1414, + "step": 9733 + }, + { + "epoch": 4.08, + "learning_rate": 1.017800853006168e-05, + "loss": 0.1351, + "step": 9734 + }, + { + "epoch": 4.08, + "learning_rate": 1.0176313387265717e-05, + "loss": 0.1709, + "step": 9735 + }, + { + "epoch": 4.08, + "learning_rate": 1.0174618239401786e-05, + "loss": 0.1435, + "step": 9736 + }, + { + "epoch": 4.08, + "learning_rate": 1.0172923086518607e-05, + "loss": 0.1688, + "step": 9737 + }, + { + "epoch": 4.08, + "learning_rate": 1.0171227928664905e-05, + "loss": 0.1496, + "step": 9738 + }, + { + "epoch": 4.08, + "learning_rate": 1.016953276588941e-05, + "loss": 0.1329, + "step": 9739 + }, + { + "epoch": 4.08, + "learning_rate": 1.0167837598240847e-05, + "loss": 0.1073, + "step": 9740 + }, + { + "epoch": 4.08, + "learning_rate": 1.016614242576794e-05, + "loss": 0.106, + "step": 9741 + }, + { + "epoch": 4.08, + "learning_rate": 1.0164447248519415e-05, + "loss": 0.1252, + "step": 9742 + }, + { + "epoch": 4.08, + "learning_rate": 1.0162752066544007e-05, + "loss": 0.1417, + "step": 9743 + }, + { + "epoch": 4.08, + "learning_rate": 1.0161056879890427e-05, + "loss": 0.1441, + "step": 9744 + }, + { + "epoch": 4.08, + "learning_rate": 1.0159361688607411e-05, + "loss": 0.1247, + "step": 9745 + }, + { + "epoch": 4.08, + "learning_rate": 1.0157666492743687e-05, + "loss": 0.2127, + "step": 9746 + }, + { + "epoch": 4.08, + "learning_rate": 1.0155971292347978e-05, + "loss": 0.1592, + "step": 9747 + }, + { + "epoch": 4.08, + "learning_rate": 1.0154276087469015e-05, + "loss": 0.0748, + "step": 9748 + }, + { + "epoch": 4.08, + "learning_rate": 1.015258087815552e-05, + "loss": 0.1466, + "step": 9749 + }, + { + "epoch": 4.08, + "learning_rate": 1.0150885664456224e-05, + "loss": 0.1454, + "step": 9750 + }, + { + "epoch": 4.08, + "learning_rate": 1.0149190446419854e-05, + "loss": 0.1237, + "step": 9751 + }, + { + "epoch": 4.08, + "learning_rate": 1.0147495224095136e-05, + "loss": 0.1437, + "step": 9752 + }, + { + "epoch": 4.08, + "learning_rate": 1.0145799997530801e-05, + "loss": 0.1919, + "step": 9753 + }, + { + "epoch": 4.08, + "learning_rate": 1.0144104766775574e-05, + "loss": 0.1484, + "step": 9754 + }, + { + "epoch": 4.08, + "learning_rate": 1.014240953187818e-05, + "loss": 0.0954, + "step": 9755 + }, + { + "epoch": 4.08, + "learning_rate": 1.0140714292887355e-05, + "loss": 0.1143, + "step": 9756 + }, + { + "epoch": 4.08, + "learning_rate": 1.0139019049851818e-05, + "loss": 0.1436, + "step": 9757 + }, + { + "epoch": 4.09, + "learning_rate": 1.0137323802820304e-05, + "loss": 0.1065, + "step": 9758 + }, + { + "epoch": 4.09, + "learning_rate": 1.0135628551841538e-05, + "loss": 0.1562, + "step": 9759 + }, + { + "epoch": 4.09, + "learning_rate": 1.0133933296964254e-05, + "loss": 0.1523, + "step": 9760 + }, + { + "epoch": 4.09, + "learning_rate": 1.0132238038237173e-05, + "loss": 0.1694, + "step": 9761 + }, + { + "epoch": 4.09, + "learning_rate": 1.0130542775709025e-05, + "loss": 0.1982, + "step": 9762 + }, + { + "epoch": 4.09, + "learning_rate": 1.0128847509428543e-05, + "loss": 0.1591, + "step": 9763 + }, + { + "epoch": 4.09, + "learning_rate": 1.0127152239444453e-05, + "loss": 0.1483, + "step": 9764 + }, + { + "epoch": 4.09, + "learning_rate": 1.0125456965805485e-05, + "loss": 0.1438, + "step": 9765 + }, + { + "epoch": 4.09, + "learning_rate": 1.012376168856037e-05, + "loss": 0.1265, + "step": 9766 + }, + { + "epoch": 4.09, + "learning_rate": 1.0122066407757833e-05, + "loss": 0.1673, + "step": 9767 + }, + { + "epoch": 4.09, + "learning_rate": 1.0120371123446603e-05, + "loss": 0.1653, + "step": 9768 + }, + { + "epoch": 4.09, + "learning_rate": 1.0118675835675416e-05, + "loss": 0.1294, + "step": 9769 + }, + { + "epoch": 4.09, + "learning_rate": 1.0116980544492996e-05, + "loss": 0.1593, + "step": 9770 + }, + { + "epoch": 4.09, + "learning_rate": 1.0115285249948073e-05, + "loss": 0.1596, + "step": 9771 + }, + { + "epoch": 4.09, + "learning_rate": 1.0113589952089377e-05, + "loss": 0.1486, + "step": 9772 + }, + { + "epoch": 4.09, + "learning_rate": 1.0111894650965644e-05, + "loss": 0.2042, + "step": 9773 + }, + { + "epoch": 4.09, + "learning_rate": 1.0110199346625593e-05, + "loss": 0.1378, + "step": 9774 + }, + { + "epoch": 4.09, + "learning_rate": 1.010850403911796e-05, + "loss": 0.1603, + "step": 9775 + }, + { + "epoch": 4.09, + "learning_rate": 1.0106808728491477e-05, + "loss": 0.1564, + "step": 9776 + }, + { + "epoch": 4.09, + "learning_rate": 1.0105113414794872e-05, + "loss": 0.1183, + "step": 9777 + }, + { + "epoch": 4.09, + "learning_rate": 1.0103418098076874e-05, + "loss": 0.1432, + "step": 9778 + }, + { + "epoch": 4.09, + "learning_rate": 1.0101722778386214e-05, + "loss": 0.1461, + "step": 9779 + }, + { + "epoch": 4.09, + "learning_rate": 1.0100027455771624e-05, + "loss": 0.1482, + "step": 9780 + }, + { + "epoch": 4.1, + "learning_rate": 1.0098332130281831e-05, + "loss": 0.156, + "step": 9781 + }, + { + "epoch": 4.1, + "learning_rate": 1.0096636801965571e-05, + "loss": 0.1696, + "step": 9782 + }, + { + "epoch": 4.1, + "learning_rate": 1.0094941470871571e-05, + "loss": 0.1552, + "step": 9783 + }, + { + "epoch": 4.1, + "learning_rate": 1.0093246137048563e-05, + "loss": 0.1347, + "step": 9784 + }, + { + "epoch": 4.1, + "learning_rate": 1.0091550800545279e-05, + "loss": 0.1917, + "step": 9785 + }, + { + "epoch": 4.1, + "learning_rate": 1.0089855461410446e-05, + "loss": 0.135, + "step": 9786 + }, + { + "epoch": 4.1, + "learning_rate": 1.0088160119692798e-05, + "loss": 0.1596, + "step": 9787 + }, + { + "epoch": 4.1, + "learning_rate": 1.0086464775441064e-05, + "loss": 0.126, + "step": 9788 + }, + { + "epoch": 4.1, + "learning_rate": 1.008476942870398e-05, + "loss": 0.1321, + "step": 9789 + }, + { + "epoch": 4.1, + "learning_rate": 1.0083074079530272e-05, + "loss": 0.1722, + "step": 9790 + }, + { + "epoch": 4.1, + "learning_rate": 1.0081378727968676e-05, + "loss": 0.1506, + "step": 9791 + }, + { + "epoch": 4.1, + "learning_rate": 1.007968337406792e-05, + "loss": 0.1526, + "step": 9792 + }, + { + "epoch": 4.1, + "learning_rate": 1.0077988017876732e-05, + "loss": 0.147, + "step": 9793 + }, + { + "epoch": 4.1, + "learning_rate": 1.0076292659443852e-05, + "loss": 0.1657, + "step": 9794 + }, + { + "epoch": 4.1, + "learning_rate": 1.0074597298818005e-05, + "loss": 0.1669, + "step": 9795 + }, + { + "epoch": 4.1, + "learning_rate": 1.0072901936047923e-05, + "loss": 0.1401, + "step": 9796 + }, + { + "epoch": 4.1, + "learning_rate": 1.0071206571182344e-05, + "loss": 0.1403, + "step": 9797 + }, + { + "epoch": 4.1, + "learning_rate": 1.0069511204269992e-05, + "loss": 0.1263, + "step": 9798 + }, + { + "epoch": 4.1, + "learning_rate": 1.0067815835359605e-05, + "loss": 0.1474, + "step": 9799 + }, + { + "epoch": 4.1, + "learning_rate": 1.0066120464499912e-05, + "loss": 0.1316, + "step": 9800 + }, + { + "epoch": 4.1, + "learning_rate": 1.0064425091739643e-05, + "loss": 0.156, + "step": 9801 + }, + { + "epoch": 4.1, + "learning_rate": 1.0062729717127531e-05, + "loss": 0.1534, + "step": 9802 + }, + { + "epoch": 4.1, + "learning_rate": 1.006103434071231e-05, + "loss": 0.1773, + "step": 9803 + }, + { + "epoch": 4.1, + "learning_rate": 1.0059338962542713e-05, + "loss": 0.1738, + "step": 9804 + }, + { + "epoch": 4.11, + "learning_rate": 1.0057643582667467e-05, + "loss": 0.1704, + "step": 9805 + }, + { + "epoch": 4.11, + "learning_rate": 1.0055948201135308e-05, + "loss": 0.1433, + "step": 9806 + }, + { + "epoch": 4.11, + "learning_rate": 1.0054252817994969e-05, + "loss": 0.1176, + "step": 9807 + }, + { + "epoch": 4.11, + "learning_rate": 1.0052557433295181e-05, + "loss": 0.1655, + "step": 9808 + }, + { + "epoch": 4.11, + "learning_rate": 1.0050862047084673e-05, + "loss": 0.1327, + "step": 9809 + }, + { + "epoch": 4.11, + "learning_rate": 1.0049166659412185e-05, + "loss": 0.1708, + "step": 9810 + }, + { + "epoch": 4.11, + "learning_rate": 1.0047471270326446e-05, + "loss": 0.1619, + "step": 9811 + }, + { + "epoch": 4.11, + "learning_rate": 1.0045775879876183e-05, + "loss": 0.1612, + "step": 9812 + }, + { + "epoch": 4.11, + "learning_rate": 1.0044080488110133e-05, + "loss": 0.1654, + "step": 9813 + }, + { + "epoch": 4.11, + "learning_rate": 1.0042385095077032e-05, + "loss": 0.1458, + "step": 9814 + }, + { + "epoch": 4.11, + "learning_rate": 1.0040689700825606e-05, + "loss": 0.1526, + "step": 9815 + }, + { + "epoch": 4.11, + "learning_rate": 1.003899430540459e-05, + "loss": 0.18, + "step": 9816 + }, + { + "epoch": 4.11, + "learning_rate": 1.0037298908862721e-05, + "loss": 0.1752, + "step": 9817 + }, + { + "epoch": 4.11, + "learning_rate": 1.0035603511248723e-05, + "loss": 0.1416, + "step": 9818 + }, + { + "epoch": 4.11, + "learning_rate": 1.0033908112611337e-05, + "loss": 0.141, + "step": 9819 + }, + { + "epoch": 4.11, + "learning_rate": 1.0032212712999292e-05, + "loss": 0.1333, + "step": 9820 + }, + { + "epoch": 4.11, + "learning_rate": 1.0030517312461323e-05, + "loss": 0.1397, + "step": 9821 + }, + { + "epoch": 4.11, + "learning_rate": 1.0028821911046158e-05, + "loss": 0.1191, + "step": 9822 + }, + { + "epoch": 4.11, + "learning_rate": 1.0027126508802534e-05, + "loss": 0.1196, + "step": 9823 + }, + { + "epoch": 4.11, + "learning_rate": 1.0025431105779184e-05, + "loss": 0.1096, + "step": 9824 + }, + { + "epoch": 4.11, + "learning_rate": 1.0023735702024838e-05, + "loss": 0.1517, + "step": 9825 + }, + { + "epoch": 4.11, + "learning_rate": 1.0022040297588229e-05, + "loss": 0.147, + "step": 9826 + }, + { + "epoch": 4.11, + "learning_rate": 1.0020344892518093e-05, + "loss": 0.1235, + "step": 9827 + }, + { + "epoch": 4.11, + "learning_rate": 1.001864948686316e-05, + "loss": 0.1635, + "step": 9828 + }, + { + "epoch": 4.12, + "learning_rate": 1.0016954080672163e-05, + "loss": 0.1218, + "step": 9829 + }, + { + "epoch": 4.12, + "learning_rate": 1.0015258673993837e-05, + "loss": 0.1867, + "step": 9830 + }, + { + "epoch": 4.12, + "learning_rate": 1.0013563266876916e-05, + "loss": 0.1583, + "step": 9831 + }, + { + "epoch": 4.12, + "learning_rate": 1.001186785937013e-05, + "loss": 0.1475, + "step": 9832 + }, + { + "epoch": 4.12, + "learning_rate": 1.0010172451522211e-05, + "loss": 0.1283, + "step": 9833 + }, + { + "epoch": 4.12, + "learning_rate": 1.0008477043381896e-05, + "loss": 0.1581, + "step": 9834 + }, + { + "epoch": 4.12, + "learning_rate": 1.0006781634997917e-05, + "loss": 0.1534, + "step": 9835 + }, + { + "epoch": 4.12, + "learning_rate": 1.0005086226419003e-05, + "loss": 0.127, + "step": 9836 + }, + { + "epoch": 4.12, + "learning_rate": 1.000339081769389e-05, + "loss": 0.1332, + "step": 9837 + }, + { + "epoch": 4.12, + "learning_rate": 1.0001695408871312e-05, + "loss": 0.142, + "step": 9838 + }, + { + "epoch": 4.12, + "learning_rate": 1e-05, + "loss": 0.1187, + "step": 9839 + }, + { + "epoch": 4.12, + "learning_rate": 9.99830459112869e-06, + "loss": 0.1665, + "step": 9840 + }, + { + "epoch": 4.12, + "learning_rate": 9.996609182306113e-06, + "loss": 0.1124, + "step": 9841 + }, + { + "epoch": 4.12, + "learning_rate": 9.994913773581e-06, + "loss": 0.1864, + "step": 9842 + }, + { + "epoch": 4.12, + "learning_rate": 9.993218365002086e-06, + "loss": 0.153, + "step": 9843 + }, + { + "epoch": 4.12, + "learning_rate": 9.991522956618107e-06, + "loss": 0.1748, + "step": 9844 + }, + { + "epoch": 4.12, + "learning_rate": 9.98982754847779e-06, + "loss": 0.1491, + "step": 9845 + }, + { + "epoch": 4.12, + "learning_rate": 9.988132140629871e-06, + "loss": 0.1691, + "step": 9846 + }, + { + "epoch": 4.12, + "learning_rate": 9.986436733123086e-06, + "loss": 0.1863, + "step": 9847 + }, + { + "epoch": 4.12, + "learning_rate": 9.984741326006163e-06, + "loss": 0.1493, + "step": 9848 + }, + { + "epoch": 4.12, + "learning_rate": 9.983045919327837e-06, + "loss": 0.1843, + "step": 9849 + }, + { + "epoch": 4.12, + "learning_rate": 9.981350513136841e-06, + "loss": 0.1709, + "step": 9850 + }, + { + "epoch": 4.12, + "learning_rate": 9.979655107481912e-06, + "loss": 0.1936, + "step": 9851 + }, + { + "epoch": 4.12, + "learning_rate": 9.977959702411776e-06, + "loss": 0.1036, + "step": 9852 + }, + { + "epoch": 4.13, + "learning_rate": 9.976264297975167e-06, + "loss": 0.1825, + "step": 9853 + }, + { + "epoch": 4.13, + "learning_rate": 9.974568894220821e-06, + "loss": 0.1521, + "step": 9854 + }, + { + "epoch": 4.13, + "learning_rate": 9.97287349119747e-06, + "loss": 0.1488, + "step": 9855 + }, + { + "epoch": 4.13, + "learning_rate": 9.971178088953845e-06, + "loss": 0.1464, + "step": 9856 + }, + { + "epoch": 4.13, + "learning_rate": 9.96948268753868e-06, + "loss": 0.1594, + "step": 9857 + }, + { + "epoch": 4.13, + "learning_rate": 9.967787287000711e-06, + "loss": 0.1306, + "step": 9858 + }, + { + "epoch": 4.13, + "learning_rate": 9.966091887388665e-06, + "loss": 0.1487, + "step": 9859 + }, + { + "epoch": 4.13, + "learning_rate": 9.964396488751278e-06, + "loss": 0.1637, + "step": 9860 + }, + { + "epoch": 4.13, + "learning_rate": 9.962701091137282e-06, + "loss": 0.1266, + "step": 9861 + }, + { + "epoch": 4.13, + "learning_rate": 9.961005694595413e-06, + "loss": 0.1867, + "step": 9862 + }, + { + "epoch": 4.13, + "learning_rate": 9.959310299174397e-06, + "loss": 0.1562, + "step": 9863 + }, + { + "epoch": 4.13, + "learning_rate": 9.957614904922971e-06, + "loss": 0.1554, + "step": 9864 + }, + { + "epoch": 4.13, + "learning_rate": 9.95591951188987e-06, + "loss": 0.1215, + "step": 9865 + }, + { + "epoch": 4.13, + "learning_rate": 9.954224120123822e-06, + "loss": 0.1551, + "step": 9866 + }, + { + "epoch": 4.13, + "learning_rate": 9.95252872967356e-06, + "loss": 0.1345, + "step": 9867 + }, + { + "epoch": 4.13, + "learning_rate": 9.950833340587818e-06, + "loss": 0.2001, + "step": 9868 + }, + { + "epoch": 4.13, + "learning_rate": 9.949137952915327e-06, + "loss": 0.1969, + "step": 9869 + }, + { + "epoch": 4.13, + "learning_rate": 9.94744256670482e-06, + "loss": 0.1802, + "step": 9870 + }, + { + "epoch": 4.13, + "learning_rate": 9.945747182005033e-06, + "loss": 0.1551, + "step": 9871 + }, + { + "epoch": 4.13, + "learning_rate": 9.944051798864692e-06, + "loss": 0.1237, + "step": 9872 + }, + { + "epoch": 4.13, + "learning_rate": 9.942356417332538e-06, + "loss": 0.1763, + "step": 9873 + }, + { + "epoch": 4.13, + "learning_rate": 9.940661037457294e-06, + "loss": 0.1686, + "step": 9874 + }, + { + "epoch": 4.13, + "learning_rate": 9.938965659287693e-06, + "loss": 0.1286, + "step": 9875 + }, + { + "epoch": 4.13, + "learning_rate": 9.937270282872474e-06, + "loss": 0.1735, + "step": 9876 + }, + { + "epoch": 4.14, + "learning_rate": 9.935574908260362e-06, + "loss": 0.128, + "step": 9877 + }, + { + "epoch": 4.14, + "learning_rate": 9.933879535500093e-06, + "loss": 0.1503, + "step": 9878 + }, + { + "epoch": 4.14, + "learning_rate": 9.932184164640399e-06, + "loss": 0.1645, + "step": 9879 + }, + { + "epoch": 4.14, + "learning_rate": 9.93048879573001e-06, + "loss": 0.1779, + "step": 9880 + }, + { + "epoch": 4.14, + "learning_rate": 9.928793428817658e-06, + "loss": 0.1424, + "step": 9881 + }, + { + "epoch": 4.14, + "learning_rate": 9.927098063952078e-06, + "loss": 0.1156, + "step": 9882 + }, + { + "epoch": 4.14, + "learning_rate": 9.925402701181999e-06, + "loss": 0.1213, + "step": 9883 + }, + { + "epoch": 4.14, + "learning_rate": 9.923707340556151e-06, + "loss": 0.1282, + "step": 9884 + }, + { + "epoch": 4.14, + "learning_rate": 9.92201198212327e-06, + "loss": 0.1577, + "step": 9885 + }, + { + "epoch": 4.14, + "learning_rate": 9.920316625932085e-06, + "loss": 0.1339, + "step": 9886 + }, + { + "epoch": 4.14, + "learning_rate": 9.918621272031327e-06, + "loss": 0.1632, + "step": 9887 + }, + { + "epoch": 4.14, + "learning_rate": 9.91692592046973e-06, + "loss": 0.1693, + "step": 9888 + }, + { + "epoch": 4.14, + "learning_rate": 9.915230571296021e-06, + "loss": 0.1508, + "step": 9889 + }, + { + "epoch": 4.14, + "learning_rate": 9.913535224558936e-06, + "loss": 0.1149, + "step": 9890 + }, + { + "epoch": 4.14, + "learning_rate": 9.911839880307202e-06, + "loss": 0.1464, + "step": 9891 + }, + { + "epoch": 4.14, + "learning_rate": 9.910144538589556e-06, + "loss": 0.1719, + "step": 9892 + }, + { + "epoch": 4.14, + "learning_rate": 9.908449199454723e-06, + "loss": 0.1044, + "step": 9893 + }, + { + "epoch": 4.14, + "learning_rate": 9.906753862951437e-06, + "loss": 0.1697, + "step": 9894 + }, + { + "epoch": 4.14, + "learning_rate": 9.90505852912843e-06, + "loss": 0.173, + "step": 9895 + }, + { + "epoch": 4.14, + "learning_rate": 9.903363198034434e-06, + "loss": 0.158, + "step": 9896 + }, + { + "epoch": 4.14, + "learning_rate": 9.901667869718172e-06, + "loss": 0.1206, + "step": 9897 + }, + { + "epoch": 4.14, + "learning_rate": 9.89997254422838e-06, + "loss": 0.1652, + "step": 9898 + }, + { + "epoch": 4.14, + "learning_rate": 9.89827722161379e-06, + "loss": 0.1206, + "step": 9899 + }, + { + "epoch": 4.14, + "learning_rate": 9.896581901923131e-06, + "loss": 0.1813, + "step": 9900 + }, + { + "epoch": 4.15, + "learning_rate": 9.894886585205132e-06, + "loss": 0.1526, + "step": 9901 + }, + { + "epoch": 4.15, + "learning_rate": 9.893191271508526e-06, + "loss": 0.1389, + "step": 9902 + }, + { + "epoch": 4.15, + "learning_rate": 9.891495960882042e-06, + "loss": 0.1889, + "step": 9903 + }, + { + "epoch": 4.15, + "learning_rate": 9.88980065337441e-06, + "loss": 0.1551, + "step": 9904 + }, + { + "epoch": 4.15, + "learning_rate": 9.88810534903436e-06, + "loss": 0.1207, + "step": 9905 + }, + { + "epoch": 4.15, + "learning_rate": 9.886410047910624e-06, + "loss": 0.1304, + "step": 9906 + }, + { + "epoch": 4.15, + "learning_rate": 9.884714750051929e-06, + "loss": 0.1537, + "step": 9907 + }, + { + "epoch": 4.15, + "learning_rate": 9.883019455507008e-06, + "loss": 0.1206, + "step": 9908 + }, + { + "epoch": 4.15, + "learning_rate": 9.881324164324587e-06, + "loss": 0.14, + "step": 9909 + }, + { + "epoch": 4.15, + "learning_rate": 9.879628876553398e-06, + "loss": 0.1357, + "step": 9910 + }, + { + "epoch": 4.15, + "learning_rate": 9.87793359224217e-06, + "loss": 0.1855, + "step": 9911 + }, + { + "epoch": 4.15, + "learning_rate": 9.876238311439635e-06, + "loss": 0.1721, + "step": 9912 + }, + { + "epoch": 4.15, + "learning_rate": 9.874543034194516e-06, + "loss": 0.172, + "step": 9913 + }, + { + "epoch": 4.15, + "learning_rate": 9.872847760555547e-06, + "loss": 0.1699, + "step": 9914 + }, + { + "epoch": 4.15, + "learning_rate": 9.871152490571459e-06, + "loss": 0.1347, + "step": 9915 + }, + { + "epoch": 4.15, + "learning_rate": 9.869457224290975e-06, + "loss": 0.1038, + "step": 9916 + }, + { + "epoch": 4.15, + "learning_rate": 9.867761961762829e-06, + "loss": 0.1226, + "step": 9917 + }, + { + "epoch": 4.15, + "learning_rate": 9.86606670303575e-06, + "loss": 0.1654, + "step": 9918 + }, + { + "epoch": 4.15, + "learning_rate": 9.864371448158463e-06, + "loss": 0.1471, + "step": 9919 + }, + { + "epoch": 4.15, + "learning_rate": 9.8626761971797e-06, + "loss": 0.1598, + "step": 9920 + }, + { + "epoch": 4.15, + "learning_rate": 9.860980950148185e-06, + "loss": 0.185, + "step": 9921 + }, + { + "epoch": 4.15, + "learning_rate": 9.859285707112648e-06, + "loss": 0.1481, + "step": 9922 + }, + { + "epoch": 4.15, + "learning_rate": 9.857590468121823e-06, + "loss": 0.1243, + "step": 9923 + }, + { + "epoch": 4.15, + "learning_rate": 9.855895233224431e-06, + "loss": 0.1544, + "step": 9924 + }, + { + "epoch": 4.16, + "learning_rate": 9.854200002469202e-06, + "loss": 0.094, + "step": 9925 + }, + { + "epoch": 4.16, + "learning_rate": 9.852504775904867e-06, + "loss": 0.1216, + "step": 9926 + }, + { + "epoch": 4.16, + "learning_rate": 9.850809553580148e-06, + "loss": 0.1627, + "step": 9927 + }, + { + "epoch": 4.16, + "learning_rate": 9.849114335543778e-06, + "loss": 0.1341, + "step": 9928 + }, + { + "epoch": 4.16, + "learning_rate": 9.847419121844484e-06, + "loss": 0.1723, + "step": 9929 + }, + { + "epoch": 4.16, + "learning_rate": 9.845723912530989e-06, + "loss": 0.1964, + "step": 9930 + }, + { + "epoch": 4.16, + "learning_rate": 9.844028707652023e-06, + "loss": 0.1807, + "step": 9931 + }, + { + "epoch": 4.16, + "learning_rate": 9.842333507256315e-06, + "loss": 0.19, + "step": 9932 + }, + { + "epoch": 4.16, + "learning_rate": 9.84063831139259e-06, + "loss": 0.1478, + "step": 9933 + }, + { + "epoch": 4.16, + "learning_rate": 9.838943120109576e-06, + "loss": 0.1816, + "step": 9934 + }, + { + "epoch": 4.16, + "learning_rate": 9.837247933455998e-06, + "loss": 0.1445, + "step": 9935 + }, + { + "epoch": 4.16, + "learning_rate": 9.835552751480587e-06, + "loss": 0.1875, + "step": 9936 + }, + { + "epoch": 4.16, + "learning_rate": 9.833857574232061e-06, + "loss": 0.1343, + "step": 9937 + }, + { + "epoch": 4.16, + "learning_rate": 9.832162401759155e-06, + "loss": 0.1191, + "step": 9938 + }, + { + "epoch": 4.16, + "learning_rate": 9.83046723411059e-06, + "loss": 0.1396, + "step": 9939 + }, + { + "epoch": 4.16, + "learning_rate": 9.828772071335095e-06, + "loss": 0.148, + "step": 9940 + }, + { + "epoch": 4.16, + "learning_rate": 9.8270769134814e-06, + "loss": 0.1541, + "step": 9941 + }, + { + "epoch": 4.16, + "learning_rate": 9.82538176059822e-06, + "loss": 0.1429, + "step": 9942 + }, + { + "epoch": 4.16, + "learning_rate": 9.823686612734285e-06, + "loss": 0.1397, + "step": 9943 + }, + { + "epoch": 4.16, + "learning_rate": 9.821991469938325e-06, + "loss": 0.1751, + "step": 9944 + }, + { + "epoch": 4.16, + "learning_rate": 9.82029633225906e-06, + "loss": 0.1865, + "step": 9945 + }, + { + "epoch": 4.16, + "learning_rate": 9.818601199745218e-06, + "loss": 0.101, + "step": 9946 + }, + { + "epoch": 4.16, + "learning_rate": 9.816906072445526e-06, + "loss": 0.1478, + "step": 9947 + }, + { + "epoch": 4.16, + "learning_rate": 9.815210950408703e-06, + "loss": 0.1287, + "step": 9948 + }, + { + "epoch": 4.17, + "learning_rate": 9.813515833683477e-06, + "loss": 0.168, + "step": 9949 + }, + { + "epoch": 4.17, + "learning_rate": 9.811820722318577e-06, + "loss": 0.1253, + "step": 9950 + }, + { + "epoch": 4.17, + "learning_rate": 9.81012561636272e-06, + "loss": 0.1594, + "step": 9951 + }, + { + "epoch": 4.17, + "learning_rate": 9.808430515864633e-06, + "loss": 0.1339, + "step": 9952 + }, + { + "epoch": 4.17, + "learning_rate": 9.806735420873044e-06, + "loss": 0.1805, + "step": 9953 + }, + { + "epoch": 4.17, + "learning_rate": 9.805040331436672e-06, + "loss": 0.1188, + "step": 9954 + }, + { + "epoch": 4.17, + "learning_rate": 9.803345247604242e-06, + "loss": 0.1549, + "step": 9955 + }, + { + "epoch": 4.17, + "learning_rate": 9.80165016942448e-06, + "loss": 0.1802, + "step": 9956 + }, + { + "epoch": 4.17, + "learning_rate": 9.799955096946107e-06, + "loss": 0.1507, + "step": 9957 + }, + { + "epoch": 4.17, + "learning_rate": 9.798260030217846e-06, + "loss": 0.112, + "step": 9958 + }, + { + "epoch": 4.17, + "learning_rate": 9.796564969288425e-06, + "loss": 0.1529, + "step": 9959 + }, + { + "epoch": 4.17, + "learning_rate": 9.79486991420656e-06, + "loss": 0.1983, + "step": 9960 + }, + { + "epoch": 4.17, + "learning_rate": 9.793174865020978e-06, + "loss": 0.124, + "step": 9961 + }, + { + "epoch": 4.17, + "learning_rate": 9.7914798217804e-06, + "loss": 0.1237, + "step": 9962 + }, + { + "epoch": 4.17, + "learning_rate": 9.789784784533553e-06, + "loss": 0.1132, + "step": 9963 + }, + { + "epoch": 4.17, + "learning_rate": 9.788089753329157e-06, + "loss": 0.0795, + "step": 9964 + }, + { + "epoch": 4.17, + "learning_rate": 9.78639472821593e-06, + "loss": 0.1575, + "step": 9965 + }, + { + "epoch": 4.17, + "learning_rate": 9.784699709242595e-06, + "loss": 0.1362, + "step": 9966 + }, + { + "epoch": 4.17, + "learning_rate": 9.78300469645788e-06, + "loss": 0.1231, + "step": 9967 + }, + { + "epoch": 4.17, + "learning_rate": 9.7813096899105e-06, + "loss": 0.1493, + "step": 9968 + }, + { + "epoch": 4.17, + "learning_rate": 9.779614689649176e-06, + "loss": 0.1198, + "step": 9969 + }, + { + "epoch": 4.17, + "learning_rate": 9.777919695722638e-06, + "loss": 0.1976, + "step": 9970 + }, + { + "epoch": 4.17, + "learning_rate": 9.776224708179597e-06, + "loss": 0.1473, + "step": 9971 + }, + { + "epoch": 4.18, + "learning_rate": 9.774529727068778e-06, + "loss": 0.1737, + "step": 9972 + }, + { + "epoch": 4.18, + "learning_rate": 9.772834752438903e-06, + "loss": 0.1593, + "step": 9973 + }, + { + "epoch": 4.18, + "learning_rate": 9.771139784338693e-06, + "loss": 0.1489, + "step": 9974 + }, + { + "epoch": 4.18, + "learning_rate": 9.769444822816865e-06, + "loss": 0.174, + "step": 9975 + }, + { + "epoch": 4.18, + "learning_rate": 9.76774986792214e-06, + "loss": 0.1863, + "step": 9976 + }, + { + "epoch": 4.18, + "learning_rate": 9.766054919703243e-06, + "loss": 0.1683, + "step": 9977 + }, + { + "epoch": 4.18, + "learning_rate": 9.764359978208886e-06, + "loss": 0.1814, + "step": 9978 + }, + { + "epoch": 4.18, + "learning_rate": 9.762665043487793e-06, + "loss": 0.1657, + "step": 9979 + }, + { + "epoch": 4.18, + "learning_rate": 9.760970115588686e-06, + "loss": 0.1244, + "step": 9980 + }, + { + "epoch": 4.18, + "learning_rate": 9.759275194560278e-06, + "loss": 0.1577, + "step": 9981 + }, + { + "epoch": 4.18, + "learning_rate": 9.757580280451289e-06, + "loss": 0.1613, + "step": 9982 + }, + { + "epoch": 4.18, + "learning_rate": 9.755885373310444e-06, + "loss": 0.1368, + "step": 9983 + }, + { + "epoch": 4.18, + "learning_rate": 9.754190473186455e-06, + "loss": 0.1572, + "step": 9984 + }, + { + "epoch": 4.18, + "learning_rate": 9.752495580128042e-06, + "loss": 0.1111, + "step": 9985 + }, + { + "epoch": 4.18, + "learning_rate": 9.750800694183926e-06, + "loss": 0.1778, + "step": 9986 + }, + { + "epoch": 4.18, + "learning_rate": 9.749105815402823e-06, + "loss": 0.1719, + "step": 9987 + }, + { + "epoch": 4.18, + "learning_rate": 9.747410943833451e-06, + "loss": 0.1589, + "step": 9988 + }, + { + "epoch": 4.18, + "learning_rate": 9.745716079524526e-06, + "loss": 0.1702, + "step": 9989 + }, + { + "epoch": 4.18, + "learning_rate": 9.744021222524764e-06, + "loss": 0.1895, + "step": 9990 + }, + { + "epoch": 4.18, + "learning_rate": 9.74232637288289e-06, + "loss": 0.143, + "step": 9991 + }, + { + "epoch": 4.18, + "learning_rate": 9.740631530647611e-06, + "loss": 0.0912, + "step": 9992 + }, + { + "epoch": 4.18, + "learning_rate": 9.73893669586765e-06, + "loss": 0.1575, + "step": 9993 + }, + { + "epoch": 4.18, + "learning_rate": 9.737241868591724e-06, + "loss": 0.1241, + "step": 9994 + }, + { + "epoch": 4.18, + "learning_rate": 9.735547048868545e-06, + "loss": 0.1368, + "step": 9995 + }, + { + "epoch": 4.19, + "learning_rate": 9.733852236746832e-06, + "loss": 0.1415, + "step": 9996 + }, + { + "epoch": 4.19, + "learning_rate": 9.732157432275301e-06, + "loss": 0.1183, + "step": 9997 + }, + { + "epoch": 4.19, + "learning_rate": 9.730462635502666e-06, + "loss": 0.1642, + "step": 9998 + }, + { + "epoch": 4.19, + "learning_rate": 9.728767846477643e-06, + "loss": 0.153, + "step": 9999 + }, + { + "epoch": 4.19, + "learning_rate": 9.72707306524895e-06, + "loss": 0.1753, + "step": 10000 + }, + { + "epoch": 4.19, + "learning_rate": 9.725378291865297e-06, + "loss": 0.1548, + "step": 10001 + }, + { + "epoch": 4.19, + "learning_rate": 9.723683526375403e-06, + "loss": 0.1527, + "step": 10002 + }, + { + "epoch": 4.19, + "learning_rate": 9.721988768827977e-06, + "loss": 0.1509, + "step": 10003 + }, + { + "epoch": 4.19, + "learning_rate": 9.720294019271744e-06, + "loss": 0.1601, + "step": 10004 + }, + { + "epoch": 4.19, + "learning_rate": 9.718599277755406e-06, + "loss": 0.1228, + "step": 10005 + }, + { + "epoch": 4.19, + "learning_rate": 9.716904544327683e-06, + "loss": 0.1562, + "step": 10006 + }, + { + "epoch": 4.19, + "learning_rate": 9.71520981903729e-06, + "loss": 0.1742, + "step": 10007 + }, + { + "epoch": 4.19, + "learning_rate": 9.713515101932935e-06, + "loss": 0.1533, + "step": 10008 + }, + { + "epoch": 4.19, + "learning_rate": 9.711820393063337e-06, + "loss": 0.1423, + "step": 10009 + }, + { + "epoch": 4.19, + "learning_rate": 9.710125692477204e-06, + "loss": 0.1037, + "step": 10010 + }, + { + "epoch": 4.19, + "learning_rate": 9.708431000223252e-06, + "loss": 0.18, + "step": 10011 + }, + { + "epoch": 4.19, + "learning_rate": 9.706736316350192e-06, + "loss": 0.1598, + "step": 10012 + }, + { + "epoch": 4.19, + "learning_rate": 9.705041640906736e-06, + "loss": 0.1305, + "step": 10013 + }, + { + "epoch": 4.19, + "learning_rate": 9.703346973941595e-06, + "loss": 0.1635, + "step": 10014 + }, + { + "epoch": 4.19, + "learning_rate": 9.701652315503484e-06, + "loss": 0.1586, + "step": 10015 + }, + { + "epoch": 4.19, + "learning_rate": 9.69995766564111e-06, + "loss": 0.1495, + "step": 10016 + }, + { + "epoch": 4.19, + "learning_rate": 9.698263024403187e-06, + "loss": 0.1283, + "step": 10017 + }, + { + "epoch": 4.19, + "learning_rate": 9.696568391838428e-06, + "loss": 0.1545, + "step": 10018 + }, + { + "epoch": 4.19, + "learning_rate": 9.69487376799554e-06, + "loss": 0.1423, + "step": 10019 + }, + { + "epoch": 4.2, + "learning_rate": 9.693179152923233e-06, + "loss": 0.1542, + "step": 10020 + }, + { + "epoch": 4.2, + "learning_rate": 9.691484546670222e-06, + "loss": 0.1238, + "step": 10021 + }, + { + "epoch": 4.2, + "learning_rate": 9.689789949285213e-06, + "loss": 0.1384, + "step": 10022 + }, + { + "epoch": 4.2, + "learning_rate": 9.688095360816916e-06, + "loss": 0.1603, + "step": 10023 + }, + { + "epoch": 4.2, + "learning_rate": 9.686400781314042e-06, + "loss": 0.1407, + "step": 10024 + }, + { + "epoch": 4.2, + "learning_rate": 9.684706210825298e-06, + "loss": 0.156, + "step": 10025 + }, + { + "epoch": 4.2, + "learning_rate": 9.683011649399393e-06, + "loss": 0.1402, + "step": 10026 + }, + { + "epoch": 4.2, + "learning_rate": 9.681317097085041e-06, + "loss": 0.1731, + "step": 10027 + }, + { + "epoch": 4.2, + "learning_rate": 9.679622553930943e-06, + "loss": 0.1694, + "step": 10028 + }, + { + "epoch": 4.2, + "learning_rate": 9.677928019985811e-06, + "loss": 0.179, + "step": 10029 + }, + { + "epoch": 4.2, + "learning_rate": 9.676233495298353e-06, + "loss": 0.1417, + "step": 10030 + }, + { + "epoch": 4.2, + "learning_rate": 9.674538979917274e-06, + "loss": 0.1471, + "step": 10031 + }, + { + "epoch": 4.2, + "learning_rate": 9.67284447389129e-06, + "loss": 0.127, + "step": 10032 + }, + { + "epoch": 4.2, + "learning_rate": 9.671149977269095e-06, + "loss": 0.1081, + "step": 10033 + }, + { + "epoch": 4.2, + "learning_rate": 9.669455490099402e-06, + "loss": 0.1672, + "step": 10034 + }, + { + "epoch": 4.2, + "learning_rate": 9.667761012430919e-06, + "loss": 0.2019, + "step": 10035 + }, + { + "epoch": 4.2, + "learning_rate": 9.66606654431235e-06, + "loss": 0.144, + "step": 10036 + }, + { + "epoch": 4.2, + "learning_rate": 9.664372085792401e-06, + "loss": 0.1272, + "step": 10037 + }, + { + "epoch": 4.2, + "learning_rate": 9.662677636919783e-06, + "loss": 0.1226, + "step": 10038 + }, + { + "epoch": 4.2, + "learning_rate": 9.660983197743192e-06, + "loss": 0.1758, + "step": 10039 + }, + { + "epoch": 4.2, + "learning_rate": 9.65928876831134e-06, + "loss": 0.1523, + "step": 10040 + }, + { + "epoch": 4.2, + "learning_rate": 9.657594348672931e-06, + "loss": 0.1713, + "step": 10041 + }, + { + "epoch": 4.2, + "learning_rate": 9.655899938876668e-06, + "loss": 0.1666, + "step": 10042 + }, + { + "epoch": 4.2, + "learning_rate": 9.654205538971254e-06, + "loss": 0.1977, + "step": 10043 + }, + { + "epoch": 4.21, + "learning_rate": 9.652511149005396e-06, + "loss": 0.1409, + "step": 10044 + }, + { + "epoch": 4.21, + "learning_rate": 9.650816769027797e-06, + "loss": 0.1365, + "step": 10045 + }, + { + "epoch": 4.21, + "learning_rate": 9.649122399087162e-06, + "loss": 0.1512, + "step": 10046 + }, + { + "epoch": 4.21, + "learning_rate": 9.647428039232188e-06, + "loss": 0.1452, + "step": 10047 + }, + { + "epoch": 4.21, + "learning_rate": 9.645733689511585e-06, + "loss": 0.148, + "step": 10048 + }, + { + "epoch": 4.21, + "learning_rate": 9.644039349974054e-06, + "loss": 0.1768, + "step": 10049 + }, + { + "epoch": 4.21, + "learning_rate": 9.642345020668294e-06, + "loss": 0.1461, + "step": 10050 + }, + { + "epoch": 4.21, + "learning_rate": 9.64065070164301e-06, + "loss": 0.1729, + "step": 10051 + }, + { + "epoch": 4.21, + "learning_rate": 9.6389563929469e-06, + "loss": 0.1695, + "step": 10052 + }, + { + "epoch": 4.21, + "learning_rate": 9.63726209462867e-06, + "loss": 0.1668, + "step": 10053 + }, + { + "epoch": 4.21, + "learning_rate": 9.635567806737021e-06, + "loss": 0.1861, + "step": 10054 + }, + { + "epoch": 4.21, + "learning_rate": 9.633873529320652e-06, + "loss": 0.1445, + "step": 10055 + }, + { + "epoch": 4.21, + "learning_rate": 9.632179262428264e-06, + "loss": 0.1459, + "step": 10056 + }, + { + "epoch": 4.21, + "learning_rate": 9.630485006108554e-06, + "loss": 0.2078, + "step": 10057 + }, + { + "epoch": 4.21, + "learning_rate": 9.628790760410224e-06, + "loss": 0.2002, + "step": 10058 + }, + { + "epoch": 4.21, + "learning_rate": 9.627096525381978e-06, + "loss": 0.1857, + "step": 10059 + }, + { + "epoch": 4.21, + "learning_rate": 9.625402301072508e-06, + "loss": 0.1554, + "step": 10060 + }, + { + "epoch": 4.21, + "learning_rate": 9.623708087530515e-06, + "loss": 0.1844, + "step": 10061 + }, + { + "epoch": 4.21, + "learning_rate": 9.622013884804704e-06, + "loss": 0.1169, + "step": 10062 + }, + { + "epoch": 4.21, + "learning_rate": 9.620319692943765e-06, + "loss": 0.1747, + "step": 10063 + }, + { + "epoch": 4.21, + "learning_rate": 9.618625511996398e-06, + "loss": 0.1651, + "step": 10064 + }, + { + "epoch": 4.21, + "learning_rate": 9.616931342011307e-06, + "loss": 0.1641, + "step": 10065 + }, + { + "epoch": 4.21, + "learning_rate": 9.615237183037181e-06, + "loss": 0.1532, + "step": 10066 + }, + { + "epoch": 4.21, + "learning_rate": 9.613543035122722e-06, + "loss": 0.1131, + "step": 10067 + }, + { + "epoch": 4.22, + "learning_rate": 9.611848898316626e-06, + "loss": 0.1774, + "step": 10068 + }, + { + "epoch": 4.22, + "learning_rate": 9.610154772667589e-06, + "loss": 0.1478, + "step": 10069 + }, + { + "epoch": 4.22, + "learning_rate": 9.608460658224306e-06, + "loss": 0.1161, + "step": 10070 + }, + { + "epoch": 4.22, + "learning_rate": 9.606766555035475e-06, + "loss": 0.1271, + "step": 10071 + }, + { + "epoch": 4.22, + "learning_rate": 9.60507246314979e-06, + "loss": 0.1846, + "step": 10072 + }, + { + "epoch": 4.22, + "learning_rate": 9.603378382615946e-06, + "loss": 0.0966, + "step": 10073 + }, + { + "epoch": 4.22, + "learning_rate": 9.601684313482638e-06, + "loss": 0.1844, + "step": 10074 + }, + { + "epoch": 4.22, + "learning_rate": 9.599990255798564e-06, + "loss": 0.1447, + "step": 10075 + }, + { + "epoch": 4.22, + "learning_rate": 9.598296209612413e-06, + "loss": 0.1451, + "step": 10076 + }, + { + "epoch": 4.22, + "learning_rate": 9.596602174972883e-06, + "loss": 0.1175, + "step": 10077 + }, + { + "epoch": 4.22, + "learning_rate": 9.594908151928663e-06, + "loss": 0.1086, + "step": 10078 + }, + { + "epoch": 4.22, + "learning_rate": 9.593214140528451e-06, + "loss": 0.1838, + "step": 10079 + }, + { + "epoch": 4.22, + "learning_rate": 9.591520140820936e-06, + "loss": 0.1618, + "step": 10080 + }, + { + "epoch": 4.22, + "learning_rate": 9.58982615285481e-06, + "loss": 0.1676, + "step": 10081 + }, + { + "epoch": 4.22, + "learning_rate": 9.588132176678772e-06, + "loss": 0.1582, + "step": 10082 + }, + { + "epoch": 4.22, + "learning_rate": 9.586438212341505e-06, + "loss": 0.1614, + "step": 10083 + }, + { + "epoch": 4.22, + "learning_rate": 9.584744259891706e-06, + "loss": 0.1732, + "step": 10084 + }, + { + "epoch": 4.22, + "learning_rate": 9.583050319378063e-06, + "loss": 0.1284, + "step": 10085 + }, + { + "epoch": 4.22, + "learning_rate": 9.581356390849272e-06, + "loss": 0.1317, + "step": 10086 + }, + { + "epoch": 4.22, + "learning_rate": 9.579662474354017e-06, + "loss": 0.1301, + "step": 10087 + }, + { + "epoch": 4.22, + "learning_rate": 9.577968569940992e-06, + "loss": 0.1623, + "step": 10088 + }, + { + "epoch": 4.22, + "learning_rate": 9.576274677658888e-06, + "loss": 0.1329, + "step": 10089 + }, + { + "epoch": 4.22, + "learning_rate": 9.574580797556389e-06, + "loss": 0.162, + "step": 10090 + }, + { + "epoch": 4.22, + "learning_rate": 9.57288692968219e-06, + "loss": 0.1418, + "step": 10091 + }, + { + "epoch": 4.23, + "learning_rate": 9.571193074084976e-06, + "loss": 0.1598, + "step": 10092 + }, + { + "epoch": 4.23, + "learning_rate": 9.569499230813437e-06, + "loss": 0.14, + "step": 10093 + }, + { + "epoch": 4.23, + "learning_rate": 9.56780539991626e-06, + "loss": 0.1781, + "step": 10094 + }, + { + "epoch": 4.23, + "learning_rate": 9.566111581442136e-06, + "loss": 0.0888, + "step": 10095 + }, + { + "epoch": 4.23, + "learning_rate": 9.564417775439747e-06, + "loss": 0.1739, + "step": 10096 + }, + { + "epoch": 4.23, + "learning_rate": 9.562723981957781e-06, + "loss": 0.1608, + "step": 10097 + }, + { + "epoch": 4.23, + "learning_rate": 9.56103020104493e-06, + "loss": 0.1476, + "step": 10098 + }, + { + "epoch": 4.23, + "learning_rate": 9.559336432749873e-06, + "loss": 0.1451, + "step": 10099 + }, + { + "epoch": 4.23, + "learning_rate": 9.557642677121306e-06, + "loss": 0.1617, + "step": 10100 + }, + { + "epoch": 4.23, + "learning_rate": 9.5559489342079e-06, + "loss": 0.2092, + "step": 10101 + }, + { + "epoch": 4.23, + "learning_rate": 9.55425520405835e-06, + "loss": 0.1794, + "step": 10102 + }, + { + "epoch": 4.23, + "learning_rate": 9.55256148672134e-06, + "loss": 0.1469, + "step": 10103 + }, + { + "epoch": 4.23, + "learning_rate": 9.55086778224555e-06, + "loss": 0.1448, + "step": 10104 + }, + { + "epoch": 4.23, + "learning_rate": 9.54917409067967e-06, + "loss": 0.1527, + "step": 10105 + }, + { + "epoch": 4.23, + "learning_rate": 9.547480412072379e-06, + "loss": 0.1258, + "step": 10106 + }, + { + "epoch": 4.23, + "learning_rate": 9.545786746472363e-06, + "loss": 0.1697, + "step": 10107 + }, + { + "epoch": 4.23, + "learning_rate": 9.544093093928302e-06, + "loss": 0.1573, + "step": 10108 + }, + { + "epoch": 4.23, + "learning_rate": 9.542399454488883e-06, + "loss": 0.161, + "step": 10109 + }, + { + "epoch": 4.23, + "learning_rate": 9.540705828202783e-06, + "loss": 0.1356, + "step": 10110 + }, + { + "epoch": 4.23, + "learning_rate": 9.539012215118688e-06, + "loss": 0.1737, + "step": 10111 + }, + { + "epoch": 4.23, + "learning_rate": 9.537318615285279e-06, + "loss": 0.1357, + "step": 10112 + }, + { + "epoch": 4.23, + "learning_rate": 9.535625028751235e-06, + "loss": 0.1329, + "step": 10113 + }, + { + "epoch": 4.23, + "learning_rate": 9.533931455565236e-06, + "loss": 0.1777, + "step": 10114 + }, + { + "epoch": 4.23, + "learning_rate": 9.532237895775963e-06, + "loss": 0.1467, + "step": 10115 + }, + { + "epoch": 4.24, + "learning_rate": 9.5305443494321e-06, + "loss": 0.1216, + "step": 10116 + }, + { + "epoch": 4.24, + "learning_rate": 9.52885081658232e-06, + "loss": 0.1411, + "step": 10117 + }, + { + "epoch": 4.24, + "learning_rate": 9.527157297275306e-06, + "loss": 0.1181, + "step": 10118 + }, + { + "epoch": 4.24, + "learning_rate": 9.525463791559737e-06, + "loss": 0.1709, + "step": 10119 + }, + { + "epoch": 4.24, + "learning_rate": 9.52377029948429e-06, + "loss": 0.1502, + "step": 10120 + }, + { + "epoch": 4.24, + "learning_rate": 9.522076821097643e-06, + "loss": 0.1715, + "step": 10121 + }, + { + "epoch": 4.24, + "learning_rate": 9.520383356448476e-06, + "loss": 0.1614, + "step": 10122 + }, + { + "epoch": 4.24, + "learning_rate": 9.518689905585463e-06, + "loss": 0.1279, + "step": 10123 + }, + { + "epoch": 4.24, + "learning_rate": 9.51699646855728e-06, + "loss": 0.1557, + "step": 10124 + }, + { + "epoch": 4.24, + "learning_rate": 9.515303045412604e-06, + "loss": 0.1346, + "step": 10125 + }, + { + "epoch": 4.24, + "learning_rate": 9.513609636200112e-06, + "loss": 0.1711, + "step": 10126 + }, + { + "epoch": 4.24, + "learning_rate": 9.511916240968481e-06, + "loss": 0.1474, + "step": 10127 + }, + { + "epoch": 4.24, + "learning_rate": 9.510222859766383e-06, + "loss": 0.1593, + "step": 10128 + }, + { + "epoch": 4.24, + "learning_rate": 9.508529492642493e-06, + "loss": 0.1478, + "step": 10129 + }, + { + "epoch": 4.24, + "learning_rate": 9.506836139645491e-06, + "loss": 0.175, + "step": 10130 + }, + { + "epoch": 4.24, + "learning_rate": 9.505142800824044e-06, + "loss": 0.1369, + "step": 10131 + }, + { + "epoch": 4.24, + "learning_rate": 9.503449476226826e-06, + "loss": 0.1794, + "step": 10132 + }, + { + "epoch": 4.24, + "learning_rate": 9.501756165902517e-06, + "loss": 0.1383, + "step": 10133 + }, + { + "epoch": 4.24, + "learning_rate": 9.50006286989978e-06, + "loss": 0.1675, + "step": 10134 + }, + { + "epoch": 4.24, + "learning_rate": 9.498369588267295e-06, + "loss": 0.1657, + "step": 10135 + }, + { + "epoch": 4.24, + "learning_rate": 9.49667632105373e-06, + "loss": 0.1763, + "step": 10136 + }, + { + "epoch": 4.24, + "learning_rate": 9.494983068307759e-06, + "loss": 0.1615, + "step": 10137 + }, + { + "epoch": 4.24, + "learning_rate": 9.493289830078049e-06, + "loss": 0.1786, + "step": 10138 + }, + { + "epoch": 4.24, + "learning_rate": 9.491596606413275e-06, + "loss": 0.1827, + "step": 10139 + }, + { + "epoch": 4.25, + "learning_rate": 9.489903397362104e-06, + "loss": 0.146, + "step": 10140 + }, + { + "epoch": 4.25, + "learning_rate": 9.488210202973208e-06, + "loss": 0.1449, + "step": 10141 + }, + { + "epoch": 4.25, + "learning_rate": 9.486517023295257e-06, + "loss": 0.1465, + "step": 10142 + }, + { + "epoch": 4.25, + "learning_rate": 9.484823858376916e-06, + "loss": 0.1521, + "step": 10143 + }, + { + "epoch": 4.25, + "learning_rate": 9.483130708266856e-06, + "loss": 0.1594, + "step": 10144 + }, + { + "epoch": 4.25, + "learning_rate": 9.48143757301375e-06, + "loss": 0.1279, + "step": 10145 + }, + { + "epoch": 4.25, + "learning_rate": 9.479744452666257e-06, + "loss": 0.2213, + "step": 10146 + }, + { + "epoch": 4.25, + "learning_rate": 9.47805134727305e-06, + "loss": 0.133, + "step": 10147 + }, + { + "epoch": 4.25, + "learning_rate": 9.476358256882792e-06, + "loss": 0.1305, + "step": 10148 + }, + { + "epoch": 4.25, + "learning_rate": 9.47466518154415e-06, + "loss": 0.1897, + "step": 10149 + }, + { + "epoch": 4.25, + "learning_rate": 9.472972121305795e-06, + "loss": 0.1382, + "step": 10150 + }, + { + "epoch": 4.25, + "learning_rate": 9.471279076216385e-06, + "loss": 0.1311, + "step": 10151 + }, + { + "epoch": 4.25, + "learning_rate": 9.46958604632459e-06, + "loss": 0.1497, + "step": 10152 + }, + { + "epoch": 4.25, + "learning_rate": 9.467893031679073e-06, + "loss": 0.1205, + "step": 10153 + }, + { + "epoch": 4.25, + "learning_rate": 9.466200032328499e-06, + "loss": 0.162, + "step": 10154 + }, + { + "epoch": 4.25, + "learning_rate": 9.46450704832153e-06, + "loss": 0.1396, + "step": 10155 + }, + { + "epoch": 4.25, + "learning_rate": 9.462814079706832e-06, + "loss": 0.1599, + "step": 10156 + }, + { + "epoch": 4.25, + "learning_rate": 9.461121126533069e-06, + "loss": 0.1311, + "step": 10157 + }, + { + "epoch": 4.25, + "learning_rate": 9.459428188848897e-06, + "loss": 0.1662, + "step": 10158 + }, + { + "epoch": 4.25, + "learning_rate": 9.457735266702984e-06, + "loss": 0.1174, + "step": 10159 + }, + { + "epoch": 4.25, + "learning_rate": 9.45604236014399e-06, + "loss": 0.1673, + "step": 10160 + }, + { + "epoch": 4.25, + "learning_rate": 9.454349469220573e-06, + "loss": 0.1501, + "step": 10161 + }, + { + "epoch": 4.25, + "learning_rate": 9.452656593981398e-06, + "loss": 0.1671, + "step": 10162 + }, + { + "epoch": 4.25, + "learning_rate": 9.450963734475125e-06, + "loss": 0.1354, + "step": 10163 + }, + { + "epoch": 4.26, + "learning_rate": 9.44927089075041e-06, + "loss": 0.2105, + "step": 10164 + }, + { + "epoch": 4.26, + "learning_rate": 9.447578062855915e-06, + "loss": 0.1904, + "step": 10165 + }, + { + "epoch": 4.26, + "learning_rate": 9.445885250840301e-06, + "loss": 0.1371, + "step": 10166 + }, + { + "epoch": 4.26, + "learning_rate": 9.444192454752221e-06, + "loss": 0.1447, + "step": 10167 + }, + { + "epoch": 4.26, + "learning_rate": 9.44249967464034e-06, + "loss": 0.1443, + "step": 10168 + }, + { + "epoch": 4.26, + "learning_rate": 9.440806910553308e-06, + "loss": 0.1216, + "step": 10169 + }, + { + "epoch": 4.26, + "learning_rate": 9.439114162539784e-06, + "loss": 0.138, + "step": 10170 + }, + { + "epoch": 4.26, + "learning_rate": 9.437421430648426e-06, + "loss": 0.1481, + "step": 10171 + }, + { + "epoch": 4.26, + "learning_rate": 9.43572871492789e-06, + "loss": 0.152, + "step": 10172 + }, + { + "epoch": 4.26, + "learning_rate": 9.434036015426831e-06, + "loss": 0.1472, + "step": 10173 + }, + { + "epoch": 4.26, + "learning_rate": 9.432343332193907e-06, + "loss": 0.1496, + "step": 10174 + }, + { + "epoch": 4.26, + "learning_rate": 9.430650665277767e-06, + "loss": 0.1479, + "step": 10175 + }, + { + "epoch": 4.26, + "learning_rate": 9.428958014727068e-06, + "loss": 0.1076, + "step": 10176 + }, + { + "epoch": 4.26, + "learning_rate": 9.427265380590466e-06, + "loss": 0.1564, + "step": 10177 + }, + { + "epoch": 4.26, + "learning_rate": 9.425572762916613e-06, + "loss": 0.1509, + "step": 10178 + }, + { + "epoch": 4.26, + "learning_rate": 9.423880161754158e-06, + "loss": 0.1453, + "step": 10179 + }, + { + "epoch": 4.26, + "learning_rate": 9.42218757715176e-06, + "loss": 0.1317, + "step": 10180 + }, + { + "epoch": 4.26, + "learning_rate": 9.420495009158066e-06, + "loss": 0.1405, + "step": 10181 + }, + { + "epoch": 4.26, + "learning_rate": 9.418802457821729e-06, + "loss": 0.1627, + "step": 10182 + }, + { + "epoch": 4.26, + "learning_rate": 9.4171099231914e-06, + "loss": 0.1302, + "step": 10183 + }, + { + "epoch": 4.26, + "learning_rate": 9.415417405315727e-06, + "loss": 0.1721, + "step": 10184 + }, + { + "epoch": 4.26, + "learning_rate": 9.413724904243363e-06, + "loss": 0.164, + "step": 10185 + }, + { + "epoch": 4.26, + "learning_rate": 9.412032420022956e-06, + "loss": 0.1351, + "step": 10186 + }, + { + "epoch": 4.27, + "learning_rate": 9.410339952703157e-06, + "loss": 0.116, + "step": 10187 + }, + { + "epoch": 4.27, + "learning_rate": 9.408647502332612e-06, + "loss": 0.1532, + "step": 10188 + }, + { + "epoch": 4.27, + "learning_rate": 9.40695506895997e-06, + "loss": 0.1497, + "step": 10189 + }, + { + "epoch": 4.27, + "learning_rate": 9.405262652633879e-06, + "loss": 0.1906, + "step": 10190 + }, + { + "epoch": 4.27, + "learning_rate": 9.403570253402987e-06, + "loss": 0.1771, + "step": 10191 + }, + { + "epoch": 4.27, + "learning_rate": 9.401877871315936e-06, + "loss": 0.1526, + "step": 10192 + }, + { + "epoch": 4.27, + "learning_rate": 9.400185506421373e-06, + "loss": 0.1485, + "step": 10193 + }, + { + "epoch": 4.27, + "learning_rate": 9.39849315876795e-06, + "loss": 0.0871, + "step": 10194 + }, + { + "epoch": 4.27, + "learning_rate": 9.396800828404304e-06, + "loss": 0.1601, + "step": 10195 + }, + { + "epoch": 4.27, + "learning_rate": 9.395108515379082e-06, + "loss": 0.1222, + "step": 10196 + }, + { + "epoch": 4.27, + "learning_rate": 9.393416219740929e-06, + "loss": 0.1923, + "step": 10197 + }, + { + "epoch": 4.27, + "learning_rate": 9.391723941538492e-06, + "loss": 0.1477, + "step": 10198 + }, + { + "epoch": 4.27, + "learning_rate": 9.390031680820409e-06, + "loss": 0.1737, + "step": 10199 + }, + { + "epoch": 4.27, + "learning_rate": 9.388339437635321e-06, + "loss": 0.1635, + "step": 10200 + }, + { + "epoch": 4.27, + "learning_rate": 9.386647212031877e-06, + "loss": 0.1653, + "step": 10201 + }, + { + "epoch": 4.27, + "learning_rate": 9.384955004058714e-06, + "loss": 0.1863, + "step": 10202 + }, + { + "epoch": 4.27, + "learning_rate": 9.383262813764471e-06, + "loss": 0.1611, + "step": 10203 + }, + { + "epoch": 4.27, + "learning_rate": 9.381570641197795e-06, + "loss": 0.1518, + "step": 10204 + }, + { + "epoch": 4.27, + "learning_rate": 9.37987848640732e-06, + "loss": 0.183, + "step": 10205 + }, + { + "epoch": 4.27, + "learning_rate": 9.378186349441685e-06, + "loss": 0.1258, + "step": 10206 + }, + { + "epoch": 4.27, + "learning_rate": 9.376494230349537e-06, + "loss": 0.1622, + "step": 10207 + }, + { + "epoch": 4.27, + "learning_rate": 9.374802129179505e-06, + "loss": 0.1771, + "step": 10208 + }, + { + "epoch": 4.27, + "learning_rate": 9.373110045980232e-06, + "loss": 0.1735, + "step": 10209 + }, + { + "epoch": 4.27, + "learning_rate": 9.371417980800357e-06, + "loss": 0.1723, + "step": 10210 + }, + { + "epoch": 4.28, + "learning_rate": 9.369725933688512e-06, + "loss": 0.1516, + "step": 10211 + }, + { + "epoch": 4.28, + "learning_rate": 9.368033904693334e-06, + "loss": 0.1744, + "step": 10212 + }, + { + "epoch": 4.28, + "learning_rate": 9.366341893863467e-06, + "loss": 0.1628, + "step": 10213 + }, + { + "epoch": 4.28, + "learning_rate": 9.364649901247534e-06, + "loss": 0.1779, + "step": 10214 + }, + { + "epoch": 4.28, + "learning_rate": 9.362957926894179e-06, + "loss": 0.1468, + "step": 10215 + }, + { + "epoch": 4.28, + "learning_rate": 9.361265970852031e-06, + "loss": 0.1671, + "step": 10216 + }, + { + "epoch": 4.28, + "learning_rate": 9.359574033169725e-06, + "loss": 0.1538, + "step": 10217 + }, + { + "epoch": 4.28, + "learning_rate": 9.357882113895899e-06, + "loss": 0.1673, + "step": 10218 + }, + { + "epoch": 4.28, + "learning_rate": 9.356190213079178e-06, + "loss": 0.1574, + "step": 10219 + }, + { + "epoch": 4.28, + "learning_rate": 9.354498330768198e-06, + "loss": 0.15, + "step": 10220 + }, + { + "epoch": 4.28, + "learning_rate": 9.352806467011595e-06, + "loss": 0.133, + "step": 10221 + }, + { + "epoch": 4.28, + "learning_rate": 9.35111462185799e-06, + "loss": 0.158, + "step": 10222 + }, + { + "epoch": 4.28, + "learning_rate": 9.349422795356022e-06, + "loss": 0.1445, + "step": 10223 + }, + { + "epoch": 4.28, + "learning_rate": 9.34773098755432e-06, + "loss": 0.2036, + "step": 10224 + }, + { + "epoch": 4.28, + "learning_rate": 9.34603919850151e-06, + "loss": 0.1615, + "step": 10225 + }, + { + "epoch": 4.28, + "learning_rate": 9.344347428246222e-06, + "loss": 0.1625, + "step": 10226 + }, + { + "epoch": 4.28, + "learning_rate": 9.342655676837085e-06, + "loss": 0.1402, + "step": 10227 + }, + { + "epoch": 4.28, + "learning_rate": 9.340963944322731e-06, + "loss": 0.185, + "step": 10228 + }, + { + "epoch": 4.28, + "learning_rate": 9.33927223075178e-06, + "loss": 0.1517, + "step": 10229 + }, + { + "epoch": 4.28, + "learning_rate": 9.337580536172864e-06, + "loss": 0.1605, + "step": 10230 + }, + { + "epoch": 4.28, + "learning_rate": 9.33588886063461e-06, + "loss": 0.1153, + "step": 10231 + }, + { + "epoch": 4.28, + "learning_rate": 9.334197204185638e-06, + "loss": 0.1657, + "step": 10232 + }, + { + "epoch": 4.28, + "learning_rate": 9.332505566874577e-06, + "loss": 0.1653, + "step": 10233 + }, + { + "epoch": 4.28, + "learning_rate": 9.330813948750053e-06, + "loss": 0.1501, + "step": 10234 + }, + { + "epoch": 4.29, + "learning_rate": 9.329122349860686e-06, + "loss": 0.1349, + "step": 10235 + }, + { + "epoch": 4.29, + "learning_rate": 9.327430770255106e-06, + "loss": 0.1312, + "step": 10236 + }, + { + "epoch": 4.29, + "learning_rate": 9.325739209981929e-06, + "loss": 0.155, + "step": 10237 + }, + { + "epoch": 4.29, + "learning_rate": 9.324047669089778e-06, + "loss": 0.1361, + "step": 10238 + }, + { + "epoch": 4.29, + "learning_rate": 9.32235614762728e-06, + "loss": 0.1375, + "step": 10239 + }, + { + "epoch": 4.29, + "learning_rate": 9.320664645643051e-06, + "loss": 0.1078, + "step": 10240 + }, + { + "epoch": 4.29, + "learning_rate": 9.318973163185713e-06, + "loss": 0.1587, + "step": 10241 + }, + { + "epoch": 4.29, + "learning_rate": 9.31728170030389e-06, + "loss": 0.172, + "step": 10242 + }, + { + "epoch": 4.29, + "learning_rate": 9.315590257046197e-06, + "loss": 0.1524, + "step": 10243 + }, + { + "epoch": 4.29, + "learning_rate": 9.313898833461255e-06, + "loss": 0.1059, + "step": 10244 + }, + { + "epoch": 4.29, + "learning_rate": 9.312207429597684e-06, + "loss": 0.1627, + "step": 10245 + }, + { + "epoch": 4.29, + "learning_rate": 9.310516045504096e-06, + "loss": 0.1688, + "step": 10246 + }, + { + "epoch": 4.29, + "learning_rate": 9.308824681229115e-06, + "loss": 0.1488, + "step": 10247 + }, + { + "epoch": 4.29, + "learning_rate": 9.307133336821357e-06, + "loss": 0.1414, + "step": 10248 + }, + { + "epoch": 4.29, + "learning_rate": 9.305442012329433e-06, + "loss": 0.1393, + "step": 10249 + }, + { + "epoch": 4.29, + "learning_rate": 9.303750707801963e-06, + "loss": 0.1709, + "step": 10250 + }, + { + "epoch": 4.29, + "learning_rate": 9.302059423287563e-06, + "loss": 0.1623, + "step": 10251 + }, + { + "epoch": 4.29, + "learning_rate": 9.300368158834843e-06, + "loss": 0.1603, + "step": 10252 + }, + { + "epoch": 4.29, + "learning_rate": 9.298676914492419e-06, + "loss": 0.174, + "step": 10253 + }, + { + "epoch": 4.29, + "learning_rate": 9.29698569030891e-06, + "loss": 0.1618, + "step": 10254 + }, + { + "epoch": 4.29, + "learning_rate": 9.295294486332916e-06, + "loss": 0.1749, + "step": 10255 + }, + { + "epoch": 4.29, + "learning_rate": 9.29360330261306e-06, + "loss": 0.175, + "step": 10256 + }, + { + "epoch": 4.29, + "learning_rate": 9.29191213919795e-06, + "loss": 0.1675, + "step": 10257 + }, + { + "epoch": 4.29, + "learning_rate": 9.290220996136199e-06, + "loss": 0.1704, + "step": 10258 + }, + { + "epoch": 4.3, + "learning_rate": 9.288529873476418e-06, + "loss": 0.1415, + "step": 10259 + }, + { + "epoch": 4.3, + "learning_rate": 9.28683877126721e-06, + "loss": 0.1458, + "step": 10260 + }, + { + "epoch": 4.3, + "learning_rate": 9.285147689557188e-06, + "loss": 0.1277, + "step": 10261 + }, + { + "epoch": 4.3, + "learning_rate": 9.283456628394966e-06, + "loss": 0.1414, + "step": 10262 + }, + { + "epoch": 4.3, + "learning_rate": 9.281765587829143e-06, + "loss": 0.1546, + "step": 10263 + }, + { + "epoch": 4.3, + "learning_rate": 9.28007456790833e-06, + "loss": 0.1535, + "step": 10264 + }, + { + "epoch": 4.3, + "learning_rate": 9.278383568681139e-06, + "loss": 0.161, + "step": 10265 + }, + { + "epoch": 4.3, + "learning_rate": 9.27669259019617e-06, + "loss": 0.1447, + "step": 10266 + }, + { + "epoch": 4.3, + "learning_rate": 9.275001632502029e-06, + "loss": 0.1506, + "step": 10267 + }, + { + "epoch": 4.3, + "learning_rate": 9.273310695647323e-06, + "loss": 0.1517, + "step": 10268 + }, + { + "epoch": 4.3, + "learning_rate": 9.271619779680658e-06, + "loss": 0.1645, + "step": 10269 + }, + { + "epoch": 4.3, + "learning_rate": 9.269928884650635e-06, + "loss": 0.1579, + "step": 10270 + }, + { + "epoch": 4.3, + "learning_rate": 9.268238010605858e-06, + "loss": 0.1617, + "step": 10271 + }, + { + "epoch": 4.3, + "learning_rate": 9.266547157594932e-06, + "loss": 0.1459, + "step": 10272 + }, + { + "epoch": 4.3, + "learning_rate": 9.264856325666455e-06, + "loss": 0.1122, + "step": 10273 + }, + { + "epoch": 4.3, + "learning_rate": 9.26316551486903e-06, + "loss": 0.1507, + "step": 10274 + }, + { + "epoch": 4.3, + "learning_rate": 9.261474725251261e-06, + "loss": 0.1768, + "step": 10275 + }, + { + "epoch": 4.3, + "learning_rate": 9.259783956861744e-06, + "loss": 0.1558, + "step": 10276 + }, + { + "epoch": 4.3, + "learning_rate": 9.258093209749081e-06, + "loss": 0.1558, + "step": 10277 + }, + { + "epoch": 4.3, + "learning_rate": 9.25640248396187e-06, + "loss": 0.1407, + "step": 10278 + }, + { + "epoch": 4.3, + "learning_rate": 9.254711779548709e-06, + "loss": 0.1367, + "step": 10279 + }, + { + "epoch": 4.3, + "learning_rate": 9.253021096558197e-06, + "loss": 0.1778, + "step": 10280 + }, + { + "epoch": 4.3, + "learning_rate": 9.251330435038935e-06, + "loss": 0.1635, + "step": 10281 + }, + { + "epoch": 4.3, + "learning_rate": 9.24963979503951e-06, + "loss": 0.111, + "step": 10282 + }, + { + "epoch": 4.31, + "learning_rate": 9.247949176608527e-06, + "loss": 0.1337, + "step": 10283 + }, + { + "epoch": 4.31, + "learning_rate": 9.246258579794574e-06, + "loss": 0.1602, + "step": 10284 + }, + { + "epoch": 4.31, + "learning_rate": 9.244568004646249e-06, + "loss": 0.1061, + "step": 10285 + }, + { + "epoch": 4.31, + "learning_rate": 9.24287745121215e-06, + "loss": 0.1671, + "step": 10286 + }, + { + "epoch": 4.31, + "learning_rate": 9.241186919540864e-06, + "loss": 0.202, + "step": 10287 + }, + { + "epoch": 4.31, + "learning_rate": 9.239496409680986e-06, + "loss": 0.1655, + "step": 10288 + }, + { + "epoch": 4.31, + "learning_rate": 9.237805921681112e-06, + "loss": 0.1347, + "step": 10289 + }, + { + "epoch": 4.31, + "learning_rate": 9.236115455589827e-06, + "loss": 0.1724, + "step": 10290 + }, + { + "epoch": 4.31, + "learning_rate": 9.234425011455727e-06, + "loss": 0.1192, + "step": 10291 + }, + { + "epoch": 4.31, + "learning_rate": 9.232734589327402e-06, + "loss": 0.1302, + "step": 10292 + }, + { + "epoch": 4.31, + "learning_rate": 9.231044189253438e-06, + "loss": 0.144, + "step": 10293 + }, + { + "epoch": 4.31, + "learning_rate": 9.229353811282427e-06, + "loss": 0.1895, + "step": 10294 + }, + { + "epoch": 4.31, + "learning_rate": 9.227663455462957e-06, + "loss": 0.1672, + "step": 10295 + }, + { + "epoch": 4.31, + "learning_rate": 9.225973121843615e-06, + "loss": 0.1519, + "step": 10296 + }, + { + "epoch": 4.31, + "learning_rate": 9.22428281047299e-06, + "loss": 0.1131, + "step": 10297 + }, + { + "epoch": 4.31, + "learning_rate": 9.222592521399666e-06, + "loss": 0.1864, + "step": 10298 + }, + { + "epoch": 4.31, + "learning_rate": 9.220902254672232e-06, + "loss": 0.1448, + "step": 10299 + }, + { + "epoch": 4.31, + "learning_rate": 9.21921201033927e-06, + "loss": 0.1724, + "step": 10300 + }, + { + "epoch": 4.31, + "learning_rate": 9.217521788449365e-06, + "loss": 0.1048, + "step": 10301 + }, + { + "epoch": 4.31, + "learning_rate": 9.215831589051103e-06, + "loss": 0.141, + "step": 10302 + }, + { + "epoch": 4.31, + "learning_rate": 9.214141412193065e-06, + "loss": 0.1361, + "step": 10303 + }, + { + "epoch": 4.31, + "learning_rate": 9.212451257923841e-06, + "loss": 0.1285, + "step": 10304 + }, + { + "epoch": 4.31, + "learning_rate": 9.210761126292e-06, + "loss": 0.1157, + "step": 10305 + }, + { + "epoch": 4.31, + "learning_rate": 9.209071017346134e-06, + "loss": 0.1308, + "step": 10306 + }, + { + "epoch": 4.32, + "learning_rate": 9.207380931134816e-06, + "loss": 0.1388, + "step": 10307 + }, + { + "epoch": 4.32, + "learning_rate": 9.205690867706631e-06, + "loss": 0.1371, + "step": 10308 + }, + { + "epoch": 4.32, + "learning_rate": 9.204000827110157e-06, + "loss": 0.1566, + "step": 10309 + }, + { + "epoch": 4.32, + "learning_rate": 9.202310809393973e-06, + "loss": 0.1379, + "step": 10310 + }, + { + "epoch": 4.32, + "learning_rate": 9.200620814606659e-06, + "loss": 0.1757, + "step": 10311 + }, + { + "epoch": 4.32, + "learning_rate": 9.198930842796786e-06, + "loss": 0.1404, + "step": 10312 + }, + { + "epoch": 4.32, + "learning_rate": 9.197240894012939e-06, + "loss": 0.1312, + "step": 10313 + }, + { + "epoch": 4.32, + "learning_rate": 9.195550968303688e-06, + "loss": 0.1814, + "step": 10314 + }, + { + "epoch": 4.32, + "learning_rate": 9.19386106571761e-06, + "loss": 0.1651, + "step": 10315 + }, + { + "epoch": 4.32, + "learning_rate": 9.192171186303283e-06, + "loss": 0.1172, + "step": 10316 + }, + { + "epoch": 4.32, + "learning_rate": 9.190481330109276e-06, + "loss": 0.1769, + "step": 10317 + }, + { + "epoch": 4.32, + "learning_rate": 9.188791497184165e-06, + "loss": 0.1593, + "step": 10318 + }, + { + "epoch": 4.32, + "learning_rate": 9.187101687576525e-06, + "loss": 0.1895, + "step": 10319 + }, + { + "epoch": 4.32, + "learning_rate": 9.185411901334924e-06, + "loss": 0.1344, + "step": 10320 + }, + { + "epoch": 4.32, + "learning_rate": 9.183722138507934e-06, + "loss": 0.1776, + "step": 10321 + }, + { + "epoch": 4.32, + "learning_rate": 9.18203239914413e-06, + "loss": 0.2088, + "step": 10322 + }, + { + "epoch": 4.32, + "learning_rate": 9.180342683292074e-06, + "loss": 0.1396, + "step": 10323 + }, + { + "epoch": 4.32, + "learning_rate": 9.178652991000342e-06, + "loss": 0.1484, + "step": 10324 + }, + { + "epoch": 4.32, + "learning_rate": 9.176963322317503e-06, + "loss": 0.1141, + "step": 10325 + }, + { + "epoch": 4.32, + "learning_rate": 9.175273677292122e-06, + "loss": 0.1434, + "step": 10326 + }, + { + "epoch": 4.32, + "learning_rate": 9.17358405597277e-06, + "loss": 0.1797, + "step": 10327 + }, + { + "epoch": 4.32, + "learning_rate": 9.171894458408007e-06, + "loss": 0.1074, + "step": 10328 + }, + { + "epoch": 4.32, + "learning_rate": 9.170204884646402e-06, + "loss": 0.1534, + "step": 10329 + }, + { + "epoch": 4.32, + "learning_rate": 9.168515334736526e-06, + "loss": 0.1112, + "step": 10330 + }, + { + "epoch": 4.33, + "learning_rate": 9.166825808726934e-06, + "loss": 0.1436, + "step": 10331 + }, + { + "epoch": 4.33, + "learning_rate": 9.165136306666196e-06, + "loss": 0.1286, + "step": 10332 + }, + { + "epoch": 4.33, + "learning_rate": 9.163446828602876e-06, + "loss": 0.1368, + "step": 10333 + }, + { + "epoch": 4.33, + "learning_rate": 9.161757374585533e-06, + "loss": 0.1154, + "step": 10334 + }, + { + "epoch": 4.33, + "learning_rate": 9.16006794466273e-06, + "loss": 0.1594, + "step": 10335 + }, + { + "epoch": 4.33, + "learning_rate": 9.158378538883032e-06, + "loss": 0.1529, + "step": 10336 + }, + { + "epoch": 4.33, + "learning_rate": 9.156689157294992e-06, + "loss": 0.1316, + "step": 10337 + }, + { + "epoch": 4.33, + "learning_rate": 9.154999799947174e-06, + "loss": 0.1473, + "step": 10338 + }, + { + "epoch": 4.33, + "learning_rate": 9.153310466888138e-06, + "loss": 0.1211, + "step": 10339 + }, + { + "epoch": 4.33, + "learning_rate": 9.151621158166443e-06, + "loss": 0.1527, + "step": 10340 + }, + { + "epoch": 4.33, + "learning_rate": 9.149931873830642e-06, + "loss": 0.1493, + "step": 10341 + }, + { + "epoch": 4.33, + "learning_rate": 9.148242613929295e-06, + "loss": 0.1741, + "step": 10342 + }, + { + "epoch": 4.33, + "learning_rate": 9.146553378510961e-06, + "loss": 0.1669, + "step": 10343 + }, + { + "epoch": 4.33, + "learning_rate": 9.144864167624191e-06, + "loss": 0.1436, + "step": 10344 + }, + { + "epoch": 4.33, + "learning_rate": 9.143174981317542e-06, + "loss": 0.1587, + "step": 10345 + }, + { + "epoch": 4.33, + "learning_rate": 9.141485819639569e-06, + "loss": 0.2177, + "step": 10346 + }, + { + "epoch": 4.33, + "learning_rate": 9.139796682638824e-06, + "loss": 0.1276, + "step": 10347 + }, + { + "epoch": 4.33, + "learning_rate": 9.138107570363857e-06, + "loss": 0.1801, + "step": 10348 + }, + { + "epoch": 4.33, + "learning_rate": 9.13641848286323e-06, + "loss": 0.1598, + "step": 10349 + }, + { + "epoch": 4.33, + "learning_rate": 9.134729420185481e-06, + "loss": 0.1672, + "step": 10350 + }, + { + "epoch": 4.33, + "learning_rate": 9.13304038237917e-06, + "loss": 0.1669, + "step": 10351 + }, + { + "epoch": 4.33, + "learning_rate": 9.13135136949284e-06, + "loss": 0.1917, + "step": 10352 + }, + { + "epoch": 4.33, + "learning_rate": 9.129662381575047e-06, + "loss": 0.1498, + "step": 10353 + }, + { + "epoch": 4.33, + "learning_rate": 9.127973418674338e-06, + "loss": 0.1808, + "step": 10354 + }, + { + "epoch": 4.34, + "learning_rate": 9.126284480839256e-06, + "loss": 0.1474, + "step": 10355 + }, + { + "epoch": 4.34, + "learning_rate": 9.124595568118352e-06, + "loss": 0.1349, + "step": 10356 + }, + { + "epoch": 4.34, + "learning_rate": 9.122906680560173e-06, + "loss": 0.1752, + "step": 10357 + }, + { + "epoch": 4.34, + "learning_rate": 9.121217818213263e-06, + "loss": 0.126, + "step": 10358 + }, + { + "epoch": 4.34, + "learning_rate": 9.119528981126165e-06, + "loss": 0.1602, + "step": 10359 + }, + { + "epoch": 4.34, + "learning_rate": 9.117840169347428e-06, + "loss": 0.1327, + "step": 10360 + }, + { + "epoch": 4.34, + "learning_rate": 9.11615138292559e-06, + "loss": 0.1333, + "step": 10361 + }, + { + "epoch": 4.34, + "learning_rate": 9.114462621909196e-06, + "loss": 0.1549, + "step": 10362 + }, + { + "epoch": 4.34, + "learning_rate": 9.11277388634679e-06, + "loss": 0.2197, + "step": 10363 + }, + { + "epoch": 4.34, + "learning_rate": 9.111085176286912e-06, + "loss": 0.1698, + "step": 10364 + }, + { + "epoch": 4.34, + "learning_rate": 9.109396491778099e-06, + "loss": 0.1588, + "step": 10365 + }, + { + "epoch": 4.34, + "learning_rate": 9.107707832868896e-06, + "loss": 0.1549, + "step": 10366 + }, + { + "epoch": 4.34, + "learning_rate": 9.106019199607838e-06, + "loss": 0.1317, + "step": 10367 + }, + { + "epoch": 4.34, + "learning_rate": 9.104330592043464e-06, + "loss": 0.1306, + "step": 10368 + }, + { + "epoch": 4.34, + "learning_rate": 9.102642010224313e-06, + "loss": 0.1434, + "step": 10369 + }, + { + "epoch": 4.34, + "learning_rate": 9.100953454198924e-06, + "loss": 0.1234, + "step": 10370 + }, + { + "epoch": 4.34, + "learning_rate": 9.099264924015826e-06, + "loss": 0.1644, + "step": 10371 + }, + { + "epoch": 4.34, + "learning_rate": 9.097576419723564e-06, + "loss": 0.1702, + "step": 10372 + }, + { + "epoch": 4.34, + "learning_rate": 9.095887941370662e-06, + "loss": 0.1733, + "step": 10373 + }, + { + "epoch": 4.34, + "learning_rate": 9.094199489005664e-06, + "loss": 0.0816, + "step": 10374 + }, + { + "epoch": 4.34, + "learning_rate": 9.092511062677092e-06, + "loss": 0.1213, + "step": 10375 + }, + { + "epoch": 4.34, + "learning_rate": 9.090822662433485e-06, + "loss": 0.1588, + "step": 10376 + }, + { + "epoch": 4.34, + "learning_rate": 9.089134288323377e-06, + "loss": 0.1595, + "step": 10377 + }, + { + "epoch": 4.34, + "learning_rate": 9.087445940395292e-06, + "loss": 0.1875, + "step": 10378 + }, + { + "epoch": 4.35, + "learning_rate": 9.085757618697765e-06, + "loss": 0.1474, + "step": 10379 + }, + { + "epoch": 4.35, + "learning_rate": 9.084069323279325e-06, + "loss": 0.1392, + "step": 10380 + }, + { + "epoch": 4.35, + "learning_rate": 9.0823810541885e-06, + "loss": 0.1543, + "step": 10381 + }, + { + "epoch": 4.35, + "learning_rate": 9.080692811473815e-06, + "loss": 0.1155, + "step": 10382 + }, + { + "epoch": 4.35, + "learning_rate": 9.079004595183799e-06, + "loss": 0.1661, + "step": 10383 + }, + { + "epoch": 4.35, + "learning_rate": 9.07731640536698e-06, + "loss": 0.1738, + "step": 10384 + }, + { + "epoch": 4.35, + "learning_rate": 9.075628242071884e-06, + "loss": 0.1154, + "step": 10385 + }, + { + "epoch": 4.35, + "learning_rate": 9.073940105347032e-06, + "loss": 0.1177, + "step": 10386 + }, + { + "epoch": 4.35, + "learning_rate": 9.07225199524095e-06, + "loss": 0.1713, + "step": 10387 + }, + { + "epoch": 4.35, + "learning_rate": 9.07056391180216e-06, + "loss": 0.1511, + "step": 10388 + }, + { + "epoch": 4.35, + "learning_rate": 9.068875855079186e-06, + "loss": 0.1077, + "step": 10389 + }, + { + "epoch": 4.35, + "learning_rate": 9.067187825120552e-06, + "loss": 0.1106, + "step": 10390 + }, + { + "epoch": 4.35, + "learning_rate": 9.065499821974774e-06, + "loss": 0.1994, + "step": 10391 + }, + { + "epoch": 4.35, + "learning_rate": 9.063811845690374e-06, + "loss": 0.1749, + "step": 10392 + }, + { + "epoch": 4.35, + "learning_rate": 9.062123896315872e-06, + "loss": 0.1925, + "step": 10393 + }, + { + "epoch": 4.35, + "learning_rate": 9.06043597389979e-06, + "loss": 0.1688, + "step": 10394 + }, + { + "epoch": 4.35, + "learning_rate": 9.05874807849064e-06, + "loss": 0.1216, + "step": 10395 + }, + { + "epoch": 4.35, + "learning_rate": 9.05706021013694e-06, + "loss": 0.1676, + "step": 10396 + }, + { + "epoch": 4.35, + "learning_rate": 9.055372368887207e-06, + "loss": 0.1076, + "step": 10397 + }, + { + "epoch": 4.35, + "learning_rate": 9.053684554789959e-06, + "loss": 0.1369, + "step": 10398 + }, + { + "epoch": 4.35, + "learning_rate": 9.051996767893707e-06, + "loss": 0.1862, + "step": 10399 + }, + { + "epoch": 4.35, + "learning_rate": 9.050309008246965e-06, + "loss": 0.1327, + "step": 10400 + }, + { + "epoch": 4.35, + "learning_rate": 9.048621275898252e-06, + "loss": 0.1802, + "step": 10401 + }, + { + "epoch": 4.36, + "learning_rate": 9.046933570896073e-06, + "loss": 0.1715, + "step": 10402 + }, + { + "epoch": 4.36, + "learning_rate": 9.045245893288943e-06, + "loss": 0.1338, + "step": 10403 + }, + { + "epoch": 4.36, + "learning_rate": 9.043558243125374e-06, + "loss": 0.145, + "step": 10404 + }, + { + "epoch": 4.36, + "learning_rate": 9.041870620453874e-06, + "loss": 0.1623, + "step": 10405 + }, + { + "epoch": 4.36, + "learning_rate": 9.040183025322952e-06, + "loss": 0.199, + "step": 10406 + }, + { + "epoch": 4.36, + "learning_rate": 9.03849545778112e-06, + "loss": 0.149, + "step": 10407 + }, + { + "epoch": 4.36, + "learning_rate": 9.03680791787688e-06, + "loss": 0.1117, + "step": 10408 + }, + { + "epoch": 4.36, + "learning_rate": 9.035120405658741e-06, + "loss": 0.1546, + "step": 10409 + }, + { + "epoch": 4.36, + "learning_rate": 9.03343292117521e-06, + "loss": 0.1345, + "step": 10410 + }, + { + "epoch": 4.36, + "learning_rate": 9.031745464474794e-06, + "loss": 0.131, + "step": 10411 + }, + { + "epoch": 4.36, + "learning_rate": 9.030058035605996e-06, + "loss": 0.1347, + "step": 10412 + }, + { + "epoch": 4.36, + "learning_rate": 9.028370634617316e-06, + "loss": 0.1506, + "step": 10413 + }, + { + "epoch": 4.36, + "learning_rate": 9.026683261557263e-06, + "loss": 0.1459, + "step": 10414 + }, + { + "epoch": 4.36, + "learning_rate": 9.024995916474333e-06, + "loss": 0.1284, + "step": 10415 + }, + { + "epoch": 4.36, + "learning_rate": 9.023308599417032e-06, + "loss": 0.1399, + "step": 10416 + }, + { + "epoch": 4.36, + "learning_rate": 9.021621310433861e-06, + "loss": 0.164, + "step": 10417 + }, + { + "epoch": 4.36, + "learning_rate": 9.019934049573316e-06, + "loss": 0.1458, + "step": 10418 + }, + { + "epoch": 4.36, + "learning_rate": 9.018246816883894e-06, + "loss": 0.1446, + "step": 10419 + }, + { + "epoch": 4.36, + "learning_rate": 9.016559612414098e-06, + "loss": 0.151, + "step": 10420 + }, + { + "epoch": 4.36, + "learning_rate": 9.014872436212423e-06, + "loss": 0.1738, + "step": 10421 + }, + { + "epoch": 4.36, + "learning_rate": 9.013185288327365e-06, + "loss": 0.1546, + "step": 10422 + }, + { + "epoch": 4.36, + "learning_rate": 9.01149816880742e-06, + "loss": 0.182, + "step": 10423 + }, + { + "epoch": 4.36, + "learning_rate": 9.009811077701083e-06, + "loss": 0.1314, + "step": 10424 + }, + { + "epoch": 4.36, + "learning_rate": 9.008124015056851e-06, + "loss": 0.1467, + "step": 10425 + }, + { + "epoch": 4.37, + "learning_rate": 9.00643698092321e-06, + "loss": 0.1292, + "step": 10426 + }, + { + "epoch": 4.37, + "learning_rate": 9.004749975348657e-06, + "loss": 0.0717, + "step": 10427 + }, + { + "epoch": 4.37, + "learning_rate": 9.003062998381685e-06, + "loss": 0.1891, + "step": 10428 + }, + { + "epoch": 4.37, + "learning_rate": 9.00137605007078e-06, + "loss": 0.1756, + "step": 10429 + }, + { + "epoch": 4.37, + "learning_rate": 8.999689130464436e-06, + "loss": 0.1271, + "step": 10430 + }, + { + "epoch": 4.37, + "learning_rate": 8.99800223961114e-06, + "loss": 0.156, + "step": 10431 + }, + { + "epoch": 4.37, + "learning_rate": 8.99631537755938e-06, + "loss": 0.1119, + "step": 10432 + }, + { + "epoch": 4.37, + "learning_rate": 8.994628544357641e-06, + "loss": 0.1482, + "step": 10433 + }, + { + "epoch": 4.37, + "learning_rate": 8.992941740054418e-06, + "loss": 0.1023, + "step": 10434 + }, + { + "epoch": 4.37, + "learning_rate": 8.991254964698188e-06, + "loss": 0.1573, + "step": 10435 + }, + { + "epoch": 4.37, + "learning_rate": 8.989568218337438e-06, + "loss": 0.1465, + "step": 10436 + }, + { + "epoch": 4.37, + "learning_rate": 8.987881501020656e-06, + "loss": 0.155, + "step": 10437 + }, + { + "epoch": 4.37, + "learning_rate": 8.98619481279632e-06, + "loss": 0.1424, + "step": 10438 + }, + { + "epoch": 4.37, + "learning_rate": 8.984508153712912e-06, + "loss": 0.0995, + "step": 10439 + }, + { + "epoch": 4.37, + "learning_rate": 8.982821523818923e-06, + "loss": 0.1584, + "step": 10440 + }, + { + "epoch": 4.37, + "learning_rate": 8.981134923162823e-06, + "loss": 0.1327, + "step": 10441 + }, + { + "epoch": 4.37, + "learning_rate": 8.979448351793095e-06, + "loss": 0.1538, + "step": 10442 + }, + { + "epoch": 4.37, + "learning_rate": 8.977761809758218e-06, + "loss": 0.1266, + "step": 10443 + }, + { + "epoch": 4.37, + "learning_rate": 8.97607529710667e-06, + "loss": 0.1584, + "step": 10444 + }, + { + "epoch": 4.37, + "learning_rate": 8.974388813886932e-06, + "loss": 0.1403, + "step": 10445 + }, + { + "epoch": 4.37, + "learning_rate": 8.972702360147474e-06, + "loss": 0.1293, + "step": 10446 + }, + { + "epoch": 4.37, + "learning_rate": 8.971015935936777e-06, + "loss": 0.142, + "step": 10447 + }, + { + "epoch": 4.37, + "learning_rate": 8.969329541303315e-06, + "loss": 0.1632, + "step": 10448 + }, + { + "epoch": 4.37, + "learning_rate": 8.967643176295557e-06, + "loss": 0.1862, + "step": 10449 + }, + { + "epoch": 4.38, + "learning_rate": 8.96595684096198e-06, + "loss": 0.1808, + "step": 10450 + }, + { + "epoch": 4.38, + "learning_rate": 8.964270535351058e-06, + "loss": 0.1826, + "step": 10451 + }, + { + "epoch": 4.38, + "learning_rate": 8.962584259511262e-06, + "loss": 0.1768, + "step": 10452 + }, + { + "epoch": 4.38, + "learning_rate": 8.960898013491058e-06, + "loss": 0.1202, + "step": 10453 + }, + { + "epoch": 4.38, + "learning_rate": 8.959211797338919e-06, + "loss": 0.1497, + "step": 10454 + }, + { + "epoch": 4.38, + "learning_rate": 8.957525611103316e-06, + "loss": 0.1282, + "step": 10455 + }, + { + "epoch": 4.38, + "learning_rate": 8.955839454832711e-06, + "loss": 0.148, + "step": 10456 + }, + { + "epoch": 4.38, + "learning_rate": 8.954153328575574e-06, + "loss": 0.1241, + "step": 10457 + }, + { + "epoch": 4.38, + "learning_rate": 8.952467232380375e-06, + "loss": 0.1631, + "step": 10458 + }, + { + "epoch": 4.38, + "learning_rate": 8.950781166295573e-06, + "loss": 0.1293, + "step": 10459 + }, + { + "epoch": 4.38, + "learning_rate": 8.949095130369636e-06, + "loss": 0.1547, + "step": 10460 + }, + { + "epoch": 4.38, + "learning_rate": 8.947409124651027e-06, + "loss": 0.1538, + "step": 10461 + }, + { + "epoch": 4.38, + "learning_rate": 8.94572314918821e-06, + "loss": 0.1362, + "step": 10462 + }, + { + "epoch": 4.38, + "learning_rate": 8.944037204029646e-06, + "loss": 0.1879, + "step": 10463 + }, + { + "epoch": 4.38, + "learning_rate": 8.942351289223795e-06, + "loss": 0.1804, + "step": 10464 + }, + { + "epoch": 4.38, + "learning_rate": 8.940665404819114e-06, + "loss": 0.1195, + "step": 10465 + }, + { + "epoch": 4.38, + "learning_rate": 8.93897955086407e-06, + "loss": 0.1211, + "step": 10466 + }, + { + "epoch": 4.38, + "learning_rate": 8.937293727407115e-06, + "loss": 0.1555, + "step": 10467 + }, + { + "epoch": 4.38, + "learning_rate": 8.935607934496708e-06, + "loss": 0.1263, + "step": 10468 + }, + { + "epoch": 4.38, + "learning_rate": 8.93392217218131e-06, + "loss": 0.184, + "step": 10469 + }, + { + "epoch": 4.38, + "learning_rate": 8.93223644050937e-06, + "loss": 0.1298, + "step": 10470 + }, + { + "epoch": 4.38, + "learning_rate": 8.930550739529345e-06, + "loss": 0.13, + "step": 10471 + }, + { + "epoch": 4.38, + "learning_rate": 8.928865069289693e-06, + "loss": 0.134, + "step": 10472 + }, + { + "epoch": 4.38, + "learning_rate": 8.927179429838864e-06, + "loss": 0.145, + "step": 10473 + }, + { + "epoch": 4.39, + "learning_rate": 8.925493821225309e-06, + "loss": 0.1614, + "step": 10474 + }, + { + "epoch": 4.39, + "learning_rate": 8.923808243497482e-06, + "loss": 0.1361, + "step": 10475 + }, + { + "epoch": 4.39, + "learning_rate": 8.92212269670383e-06, + "loss": 0.1655, + "step": 10476 + }, + { + "epoch": 4.39, + "learning_rate": 8.920437180892803e-06, + "loss": 0.1206, + "step": 10477 + }, + { + "epoch": 4.39, + "learning_rate": 8.918751696112856e-06, + "loss": 0.1356, + "step": 10478 + }, + { + "epoch": 4.39, + "learning_rate": 8.917066242412429e-06, + "loss": 0.1348, + "step": 10479 + }, + { + "epoch": 4.39, + "learning_rate": 8.915380819839972e-06, + "loss": 0.1483, + "step": 10480 + }, + { + "epoch": 4.39, + "learning_rate": 8.91369542844393e-06, + "loss": 0.1389, + "step": 10481 + }, + { + "epoch": 4.39, + "learning_rate": 8.912010068272752e-06, + "loss": 0.1384, + "step": 10482 + }, + { + "epoch": 4.39, + "learning_rate": 8.910324739374877e-06, + "loss": 0.1575, + "step": 10483 + }, + { + "epoch": 4.39, + "learning_rate": 8.908639441798751e-06, + "loss": 0.1206, + "step": 10484 + }, + { + "epoch": 4.39, + "learning_rate": 8.906954175592819e-06, + "loss": 0.1597, + "step": 10485 + }, + { + "epoch": 4.39, + "learning_rate": 8.905268940805519e-06, + "loss": 0.1651, + "step": 10486 + }, + { + "epoch": 4.39, + "learning_rate": 8.903583737485286e-06, + "loss": 0.1341, + "step": 10487 + }, + { + "epoch": 4.39, + "learning_rate": 8.901898565680567e-06, + "loss": 0.1445, + "step": 10488 + }, + { + "epoch": 4.39, + "learning_rate": 8.900213425439803e-06, + "loss": 0.1532, + "step": 10489 + }, + { + "epoch": 4.39, + "learning_rate": 8.898528316811425e-06, + "loss": 0.1891, + "step": 10490 + }, + { + "epoch": 4.39, + "learning_rate": 8.896843239843873e-06, + "loss": 0.1499, + "step": 10491 + }, + { + "epoch": 4.39, + "learning_rate": 8.895158194585583e-06, + "loss": 0.1366, + "step": 10492 + }, + { + "epoch": 4.39, + "learning_rate": 8.893473181084993e-06, + "loss": 0.1487, + "step": 10493 + }, + { + "epoch": 4.39, + "learning_rate": 8.891788199390533e-06, + "loss": 0.1514, + "step": 10494 + }, + { + "epoch": 4.39, + "learning_rate": 8.890103249550638e-06, + "loss": 0.1496, + "step": 10495 + }, + { + "epoch": 4.39, + "learning_rate": 8.88841833161374e-06, + "loss": 0.1665, + "step": 10496 + }, + { + "epoch": 4.39, + "learning_rate": 8.88673344562827e-06, + "loss": 0.1539, + "step": 10497 + }, + { + "epoch": 4.4, + "learning_rate": 8.88504859164266e-06, + "loss": 0.1255, + "step": 10498 + }, + { + "epoch": 4.4, + "learning_rate": 8.883363769705338e-06, + "loss": 0.1813, + "step": 10499 + }, + { + "epoch": 4.4, + "learning_rate": 8.881678979864734e-06, + "loss": 0.1768, + "step": 10500 + }, + { + "epoch": 4.4, + "learning_rate": 8.879994222169275e-06, + "loss": 0.1486, + "step": 10501 + }, + { + "epoch": 4.4, + "learning_rate": 8.87830949666739e-06, + "loss": 0.1229, + "step": 10502 + }, + { + "epoch": 4.4, + "learning_rate": 8.876624803407501e-06, + "loss": 0.1755, + "step": 10503 + }, + { + "epoch": 4.4, + "learning_rate": 8.874940142438036e-06, + "loss": 0.1787, + "step": 10504 + }, + { + "epoch": 4.4, + "learning_rate": 8.873255513807418e-06, + "loss": 0.1251, + "step": 10505 + }, + { + "epoch": 4.4, + "learning_rate": 8.87157091756407e-06, + "loss": 0.1068, + "step": 10506 + }, + { + "epoch": 4.4, + "learning_rate": 8.869886353756415e-06, + "loss": 0.1293, + "step": 10507 + }, + { + "epoch": 4.4, + "learning_rate": 8.868201822432878e-06, + "loss": 0.1115, + "step": 10508 + }, + { + "epoch": 4.4, + "learning_rate": 8.86651732364187e-06, + "loss": 0.1715, + "step": 10509 + }, + { + "epoch": 4.4, + "learning_rate": 8.86483285743182e-06, + "loss": 0.1266, + "step": 10510 + }, + { + "epoch": 4.4, + "learning_rate": 8.863148423851138e-06, + "loss": 0.1839, + "step": 10511 + }, + { + "epoch": 4.4, + "learning_rate": 8.861464022948247e-06, + "loss": 0.1993, + "step": 10512 + }, + { + "epoch": 4.4, + "learning_rate": 8.859779654771563e-06, + "loss": 0.163, + "step": 10513 + }, + { + "epoch": 4.4, + "learning_rate": 8.858095319369499e-06, + "loss": 0.1587, + "step": 10514 + }, + { + "epoch": 4.4, + "learning_rate": 8.856411016790472e-06, + "loss": 0.1455, + "step": 10515 + }, + { + "epoch": 4.4, + "learning_rate": 8.8547267470829e-06, + "loss": 0.1502, + "step": 10516 + }, + { + "epoch": 4.4, + "learning_rate": 8.853042510295187e-06, + "loss": 0.1293, + "step": 10517 + }, + { + "epoch": 4.4, + "learning_rate": 8.851358306475748e-06, + "loss": 0.1647, + "step": 10518 + }, + { + "epoch": 4.4, + "learning_rate": 8.849674135672999e-06, + "loss": 0.1932, + "step": 10519 + }, + { + "epoch": 4.4, + "learning_rate": 8.847989997935345e-06, + "loss": 0.1486, + "step": 10520 + }, + { + "epoch": 4.4, + "learning_rate": 8.846305893311194e-06, + "loss": 0.1692, + "step": 10521 + }, + { + "epoch": 4.41, + "learning_rate": 8.844621821848958e-06, + "loss": 0.1121, + "step": 10522 + }, + { + "epoch": 4.41, + "learning_rate": 8.842937783597045e-06, + "loss": 0.1358, + "step": 10523 + }, + { + "epoch": 4.41, + "learning_rate": 8.841253778603855e-06, + "loss": 0.1654, + "step": 10524 + }, + { + "epoch": 4.41, + "learning_rate": 8.839569806917797e-06, + "loss": 0.1473, + "step": 10525 + }, + { + "epoch": 4.41, + "learning_rate": 8.837885868587279e-06, + "loss": 0.1257, + "step": 10526 + }, + { + "epoch": 4.41, + "learning_rate": 8.836201963660695e-06, + "loss": 0.1375, + "step": 10527 + }, + { + "epoch": 4.41, + "learning_rate": 8.834518092186456e-06, + "loss": 0.1417, + "step": 10528 + }, + { + "epoch": 4.41, + "learning_rate": 8.83283425421296e-06, + "loss": 0.139, + "step": 10529 + }, + { + "epoch": 4.41, + "learning_rate": 8.831150449788613e-06, + "loss": 0.1568, + "step": 10530 + }, + { + "epoch": 4.41, + "learning_rate": 8.829466678961803e-06, + "loss": 0.189, + "step": 10531 + }, + { + "epoch": 4.41, + "learning_rate": 8.827782941780934e-06, + "loss": 0.1277, + "step": 10532 + }, + { + "epoch": 4.41, + "learning_rate": 8.826099238294404e-06, + "loss": 0.1296, + "step": 10533 + }, + { + "epoch": 4.41, + "learning_rate": 8.824415568550612e-06, + "loss": 0.1962, + "step": 10534 + }, + { + "epoch": 4.41, + "learning_rate": 8.82273193259795e-06, + "loss": 0.1299, + "step": 10535 + }, + { + "epoch": 4.41, + "learning_rate": 8.821048330484814e-06, + "loss": 0.0992, + "step": 10536 + }, + { + "epoch": 4.41, + "learning_rate": 8.819364762259598e-06, + "loss": 0.1378, + "step": 10537 + }, + { + "epoch": 4.41, + "learning_rate": 8.817681227970693e-06, + "loss": 0.1181, + "step": 10538 + }, + { + "epoch": 4.41, + "learning_rate": 8.815997727666493e-06, + "loss": 0.1476, + "step": 10539 + }, + { + "epoch": 4.41, + "learning_rate": 8.814314261395389e-06, + "loss": 0.1612, + "step": 10540 + }, + { + "epoch": 4.41, + "learning_rate": 8.812630829205766e-06, + "loss": 0.1735, + "step": 10541 + }, + { + "epoch": 4.41, + "learning_rate": 8.810947431146019e-06, + "loss": 0.1622, + "step": 10542 + }, + { + "epoch": 4.41, + "learning_rate": 8.809264067264533e-06, + "loss": 0.1423, + "step": 10543 + }, + { + "epoch": 4.41, + "learning_rate": 8.807580737609693e-06, + "loss": 0.151, + "step": 10544 + }, + { + "epoch": 4.41, + "learning_rate": 8.805897442229888e-06, + "loss": 0.1505, + "step": 10545 + }, + { + "epoch": 4.42, + "learning_rate": 8.804214181173501e-06, + "loss": 0.1616, + "step": 10546 + }, + { + "epoch": 4.42, + "learning_rate": 8.802530954488915e-06, + "loss": 0.1944, + "step": 10547 + }, + { + "epoch": 4.42, + "learning_rate": 8.800847762224514e-06, + "loss": 0.1288, + "step": 10548 + }, + { + "epoch": 4.42, + "learning_rate": 8.799164604428683e-06, + "loss": 0.123, + "step": 10549 + }, + { + "epoch": 4.42, + "learning_rate": 8.797481481149798e-06, + "loss": 0.1643, + "step": 10550 + }, + { + "epoch": 4.42, + "learning_rate": 8.795798392436239e-06, + "loss": 0.1086, + "step": 10551 + }, + { + "epoch": 4.42, + "learning_rate": 8.794115338336388e-06, + "loss": 0.1179, + "step": 10552 + }, + { + "epoch": 4.42, + "learning_rate": 8.792432318898624e-06, + "loss": 0.1645, + "step": 10553 + }, + { + "epoch": 4.42, + "learning_rate": 8.790749334171323e-06, + "loss": 0.1417, + "step": 10554 + }, + { + "epoch": 4.42, + "learning_rate": 8.789066384202853e-06, + "loss": 0.1391, + "step": 10555 + }, + { + "epoch": 4.42, + "learning_rate": 8.787383469041598e-06, + "loss": 0.1525, + "step": 10556 + }, + { + "epoch": 4.42, + "learning_rate": 8.78570058873593e-06, + "loss": 0.1494, + "step": 10557 + }, + { + "epoch": 4.42, + "learning_rate": 8.78401774333422e-06, + "loss": 0.1371, + "step": 10558 + }, + { + "epoch": 4.42, + "learning_rate": 8.78233493288484e-06, + "loss": 0.1782, + "step": 10559 + }, + { + "epoch": 4.42, + "learning_rate": 8.780652157436163e-06, + "loss": 0.1544, + "step": 10560 + }, + { + "epoch": 4.42, + "learning_rate": 8.778969417036557e-06, + "loss": 0.1573, + "step": 10561 + }, + { + "epoch": 4.42, + "learning_rate": 8.77728671173439e-06, + "loss": 0.1482, + "step": 10562 + }, + { + "epoch": 4.42, + "learning_rate": 8.775604041578034e-06, + "loss": 0.1629, + "step": 10563 + }, + { + "epoch": 4.42, + "learning_rate": 8.773921406615854e-06, + "loss": 0.1126, + "step": 10564 + }, + { + "epoch": 4.42, + "learning_rate": 8.772238806896213e-06, + "loss": 0.1316, + "step": 10565 + }, + { + "epoch": 4.42, + "learning_rate": 8.770556242467479e-06, + "loss": 0.1752, + "step": 10566 + }, + { + "epoch": 4.42, + "learning_rate": 8.768873713378015e-06, + "loss": 0.1287, + "step": 10567 + }, + { + "epoch": 4.42, + "learning_rate": 8.767191219676184e-06, + "loss": 0.1922, + "step": 10568 + }, + { + "epoch": 4.42, + "learning_rate": 8.765508761410345e-06, + "loss": 0.1686, + "step": 10569 + }, + { + "epoch": 4.43, + "learning_rate": 8.763826338628865e-06, + "loss": 0.1586, + "step": 10570 + }, + { + "epoch": 4.43, + "learning_rate": 8.762143951380096e-06, + "loss": 0.168, + "step": 10571 + }, + { + "epoch": 4.43, + "learning_rate": 8.760461599712402e-06, + "loss": 0.1332, + "step": 10572 + }, + { + "epoch": 4.43, + "learning_rate": 8.758779283674142e-06, + "loss": 0.1216, + "step": 10573 + }, + { + "epoch": 4.43, + "learning_rate": 8.757097003313665e-06, + "loss": 0.1546, + "step": 10574 + }, + { + "epoch": 4.43, + "learning_rate": 8.755414758679336e-06, + "loss": 0.1556, + "step": 10575 + }, + { + "epoch": 4.43, + "learning_rate": 8.753732549819508e-06, + "loss": 0.1918, + "step": 10576 + }, + { + "epoch": 4.43, + "learning_rate": 8.752050376782525e-06, + "loss": 0.141, + "step": 10577 + }, + { + "epoch": 4.43, + "learning_rate": 8.750368239616752e-06, + "loss": 0.1161, + "step": 10578 + }, + { + "epoch": 4.43, + "learning_rate": 8.748686138370533e-06, + "loss": 0.1649, + "step": 10579 + }, + { + "epoch": 4.43, + "learning_rate": 8.747004073092218e-06, + "loss": 0.1614, + "step": 10580 + }, + { + "epoch": 4.43, + "learning_rate": 8.745322043830163e-06, + "loss": 0.1292, + "step": 10581 + }, + { + "epoch": 4.43, + "learning_rate": 8.743640050632709e-06, + "loss": 0.1331, + "step": 10582 + }, + { + "epoch": 4.43, + "learning_rate": 8.741958093548208e-06, + "loss": 0.166, + "step": 10583 + }, + { + "epoch": 4.43, + "learning_rate": 8.740276172625007e-06, + "loss": 0.1588, + "step": 10584 + }, + { + "epoch": 4.43, + "learning_rate": 8.738594287911447e-06, + "loss": 0.1624, + "step": 10585 + }, + { + "epoch": 4.43, + "learning_rate": 8.736912439455876e-06, + "loss": 0.1911, + "step": 10586 + }, + { + "epoch": 4.43, + "learning_rate": 8.735230627306637e-06, + "loss": 0.1964, + "step": 10587 + }, + { + "epoch": 4.43, + "learning_rate": 8.733548851512071e-06, + "loss": 0.1723, + "step": 10588 + }, + { + "epoch": 4.43, + "learning_rate": 8.731867112120516e-06, + "loss": 0.1644, + "step": 10589 + }, + { + "epoch": 4.43, + "learning_rate": 8.730185409180321e-06, + "loss": 0.1857, + "step": 10590 + }, + { + "epoch": 4.43, + "learning_rate": 8.728503742739816e-06, + "loss": 0.1473, + "step": 10591 + }, + { + "epoch": 4.43, + "learning_rate": 8.726822112847343e-06, + "loss": 0.1511, + "step": 10592 + }, + { + "epoch": 4.44, + "learning_rate": 8.725140519551238e-06, + "loss": 0.1551, + "step": 10593 + }, + { + "epoch": 4.44, + "learning_rate": 8.72345896289984e-06, + "loss": 0.1422, + "step": 10594 + }, + { + "epoch": 4.44, + "learning_rate": 8.721777442941479e-06, + "loss": 0.145, + "step": 10595 + }, + { + "epoch": 4.44, + "learning_rate": 8.720095959724492e-06, + "loss": 0.1595, + "step": 10596 + }, + { + "epoch": 4.44, + "learning_rate": 8.718414513297214e-06, + "loss": 0.1792, + "step": 10597 + }, + { + "epoch": 4.44, + "learning_rate": 8.716733103707972e-06, + "loss": 0.1171, + "step": 10598 + }, + { + "epoch": 4.44, + "learning_rate": 8.715051731005097e-06, + "loss": 0.1347, + "step": 10599 + }, + { + "epoch": 4.44, + "learning_rate": 8.713370395236918e-06, + "loss": 0.1383, + "step": 10600 + }, + { + "epoch": 4.44, + "learning_rate": 8.711689096451769e-06, + "loss": 0.1394, + "step": 10601 + }, + { + "epoch": 4.44, + "learning_rate": 8.71000783469797e-06, + "loss": 0.1451, + "step": 10602 + }, + { + "epoch": 4.44, + "learning_rate": 8.708326610023852e-06, + "loss": 0.1374, + "step": 10603 + }, + { + "epoch": 4.44, + "learning_rate": 8.706645422477739e-06, + "loss": 0.1479, + "step": 10604 + }, + { + "epoch": 4.44, + "learning_rate": 8.704964272107957e-06, + "loss": 0.1753, + "step": 10605 + }, + { + "epoch": 4.44, + "learning_rate": 8.703283158962828e-06, + "loss": 0.1383, + "step": 10606 + }, + { + "epoch": 4.44, + "learning_rate": 8.701602083090673e-06, + "loss": 0.1204, + "step": 10607 + }, + { + "epoch": 4.44, + "learning_rate": 8.699921044539815e-06, + "loss": 0.1408, + "step": 10608 + }, + { + "epoch": 4.44, + "learning_rate": 8.69824004335857e-06, + "loss": 0.1552, + "step": 10609 + }, + { + "epoch": 4.44, + "learning_rate": 8.696559079595261e-06, + "loss": 0.1682, + "step": 10610 + }, + { + "epoch": 4.44, + "learning_rate": 8.694878153298207e-06, + "loss": 0.1135, + "step": 10611 + }, + { + "epoch": 4.44, + "learning_rate": 8.69319726451572e-06, + "loss": 0.1273, + "step": 10612 + }, + { + "epoch": 4.44, + "learning_rate": 8.691516413296118e-06, + "loss": 0.0887, + "step": 10613 + }, + { + "epoch": 4.44, + "learning_rate": 8.689835599687717e-06, + "loss": 0.136, + "step": 10614 + }, + { + "epoch": 4.44, + "learning_rate": 8.68815482373883e-06, + "loss": 0.1311, + "step": 10615 + }, + { + "epoch": 4.44, + "learning_rate": 8.686474085497765e-06, + "loss": 0.1339, + "step": 10616 + }, + { + "epoch": 4.45, + "learning_rate": 8.68479338501284e-06, + "loss": 0.1429, + "step": 10617 + }, + { + "epoch": 4.45, + "learning_rate": 8.683112722332361e-06, + "loss": 0.1564, + "step": 10618 + }, + { + "epoch": 4.45, + "learning_rate": 8.681432097504636e-06, + "loss": 0.1621, + "step": 10619 + }, + { + "epoch": 4.45, + "learning_rate": 8.67975151057798e-06, + "loss": 0.1507, + "step": 10620 + }, + { + "epoch": 4.45, + "learning_rate": 8.678070961600693e-06, + "loss": 0.1407, + "step": 10621 + }, + { + "epoch": 4.45, + "learning_rate": 8.676390450621084e-06, + "loss": 0.1741, + "step": 10622 + }, + { + "epoch": 4.45, + "learning_rate": 8.674709977687456e-06, + "loss": 0.1286, + "step": 10623 + }, + { + "epoch": 4.45, + "learning_rate": 8.67302954284811e-06, + "loss": 0.1865, + "step": 10624 + }, + { + "epoch": 4.45, + "learning_rate": 8.67134914615136e-06, + "loss": 0.1679, + "step": 10625 + }, + { + "epoch": 4.45, + "learning_rate": 8.669668787645494e-06, + "loss": 0.1714, + "step": 10626 + }, + { + "epoch": 4.45, + "learning_rate": 8.667988467378819e-06, + "loss": 0.1482, + "step": 10627 + }, + { + "epoch": 4.45, + "learning_rate": 8.666308185399634e-06, + "loss": 0.2011, + "step": 10628 + }, + { + "epoch": 4.45, + "learning_rate": 8.664627941756236e-06, + "loss": 0.1634, + "step": 10629 + }, + { + "epoch": 4.45, + "learning_rate": 8.662947736496923e-06, + "loss": 0.1275, + "step": 10630 + }, + { + "epoch": 4.45, + "learning_rate": 8.661267569669994e-06, + "loss": 0.1586, + "step": 10631 + }, + { + "epoch": 4.45, + "learning_rate": 8.659587441323736e-06, + "loss": 0.16, + "step": 10632 + }, + { + "epoch": 4.45, + "learning_rate": 8.65790735150645e-06, + "loss": 0.1889, + "step": 10633 + }, + { + "epoch": 4.45, + "learning_rate": 8.656227300266425e-06, + "loss": 0.2088, + "step": 10634 + }, + { + "epoch": 4.45, + "learning_rate": 8.654547287651955e-06, + "loss": 0.1768, + "step": 10635 + }, + { + "epoch": 4.45, + "learning_rate": 8.652867313711328e-06, + "loss": 0.141, + "step": 10636 + }, + { + "epoch": 4.45, + "learning_rate": 8.651187378492836e-06, + "loss": 0.1424, + "step": 10637 + }, + { + "epoch": 4.45, + "learning_rate": 8.649507482044766e-06, + "loss": 0.1509, + "step": 10638 + }, + { + "epoch": 4.45, + "learning_rate": 8.647827624415403e-06, + "loss": 0.176, + "step": 10639 + }, + { + "epoch": 4.45, + "learning_rate": 8.646147805653037e-06, + "loss": 0.1154, + "step": 10640 + }, + { + "epoch": 4.46, + "learning_rate": 8.644468025805952e-06, + "loss": 0.1167, + "step": 10641 + }, + { + "epoch": 4.46, + "learning_rate": 8.642788284922427e-06, + "loss": 0.149, + "step": 10642 + }, + { + "epoch": 4.46, + "learning_rate": 8.641108583050751e-06, + "loss": 0.19, + "step": 10643 + }, + { + "epoch": 4.46, + "learning_rate": 8.639428920239205e-06, + "loss": 0.1419, + "step": 10644 + }, + { + "epoch": 4.46, + "learning_rate": 8.637749296536065e-06, + "loss": 0.2231, + "step": 10645 + }, + { + "epoch": 4.46, + "learning_rate": 8.636069711989614e-06, + "loss": 0.1218, + "step": 10646 + }, + { + "epoch": 4.46, + "learning_rate": 8.634390166648126e-06, + "loss": 0.1739, + "step": 10647 + }, + { + "epoch": 4.46, + "learning_rate": 8.632710660559883e-06, + "loss": 0.1368, + "step": 10648 + }, + { + "epoch": 4.46, + "learning_rate": 8.631031193773158e-06, + "loss": 0.1488, + "step": 10649 + }, + { + "epoch": 4.46, + "learning_rate": 8.629351766336225e-06, + "loss": 0.1991, + "step": 10650 + }, + { + "epoch": 4.46, + "learning_rate": 8.627672378297361e-06, + "loss": 0.154, + "step": 10651 + }, + { + "epoch": 4.46, + "learning_rate": 8.625993029704838e-06, + "loss": 0.1503, + "step": 10652 + }, + { + "epoch": 4.46, + "learning_rate": 8.624313720606922e-06, + "loss": 0.1851, + "step": 10653 + }, + { + "epoch": 4.46, + "learning_rate": 8.62263445105189e-06, + "loss": 0.1845, + "step": 10654 + }, + { + "epoch": 4.46, + "learning_rate": 8.62095522108801e-06, + "loss": 0.1492, + "step": 10655 + }, + { + "epoch": 4.46, + "learning_rate": 8.619276030763545e-06, + "loss": 0.218, + "step": 10656 + }, + { + "epoch": 4.46, + "learning_rate": 8.617596880126766e-06, + "loss": 0.1528, + "step": 10657 + }, + { + "epoch": 4.46, + "learning_rate": 8.61591776922594e-06, + "loss": 0.1055, + "step": 10658 + }, + { + "epoch": 4.46, + "learning_rate": 8.614238698109329e-06, + "loss": 0.1446, + "step": 10659 + }, + { + "epoch": 4.46, + "learning_rate": 8.612559666825196e-06, + "loss": 0.1775, + "step": 10660 + }, + { + "epoch": 4.46, + "learning_rate": 8.610880675421805e-06, + "loss": 0.1524, + "step": 10661 + }, + { + "epoch": 4.46, + "learning_rate": 8.609201723947415e-06, + "loss": 0.1336, + "step": 10662 + }, + { + "epoch": 4.46, + "learning_rate": 8.607522812450288e-06, + "loss": 0.1588, + "step": 10663 + }, + { + "epoch": 4.46, + "learning_rate": 8.60584394097868e-06, + "loss": 0.1851, + "step": 10664 + }, + { + "epoch": 4.47, + "learning_rate": 8.604165109580855e-06, + "loss": 0.1002, + "step": 10665 + }, + { + "epoch": 4.47, + "learning_rate": 8.602486318305067e-06, + "loss": 0.172, + "step": 10666 + }, + { + "epoch": 4.47, + "learning_rate": 8.600807567199565e-06, + "loss": 0.1638, + "step": 10667 + }, + { + "epoch": 4.47, + "learning_rate": 8.599128856312607e-06, + "loss": 0.1558, + "step": 10668 + }, + { + "epoch": 4.47, + "learning_rate": 8.597450185692451e-06, + "loss": 0.1578, + "step": 10669 + }, + { + "epoch": 4.47, + "learning_rate": 8.59577155538734e-06, + "loss": 0.1578, + "step": 10670 + }, + { + "epoch": 4.47, + "learning_rate": 8.594092965445534e-06, + "loss": 0.1351, + "step": 10671 + }, + { + "epoch": 4.47, + "learning_rate": 8.592414415915275e-06, + "loss": 0.1154, + "step": 10672 + }, + { + "epoch": 4.47, + "learning_rate": 8.590735906844816e-06, + "loss": 0.1396, + "step": 10673 + }, + { + "epoch": 4.47, + "learning_rate": 8.5890574382824e-06, + "loss": 0.1288, + "step": 10674 + }, + { + "epoch": 4.47, + "learning_rate": 8.58737901027628e-06, + "loss": 0.154, + "step": 10675 + }, + { + "epoch": 4.47, + "learning_rate": 8.585700622874695e-06, + "loss": 0.1194, + "step": 10676 + }, + { + "epoch": 4.47, + "learning_rate": 8.584022276125891e-06, + "loss": 0.1551, + "step": 10677 + }, + { + "epoch": 4.47, + "learning_rate": 8.582343970078108e-06, + "loss": 0.1489, + "step": 10678 + }, + { + "epoch": 4.47, + "learning_rate": 8.580665704779593e-06, + "loss": 0.1997, + "step": 10679 + }, + { + "epoch": 4.47, + "learning_rate": 8.57898748027858e-06, + "loss": 0.1953, + "step": 10680 + }, + { + "epoch": 4.47, + "learning_rate": 8.577309296623311e-06, + "loss": 0.1389, + "step": 10681 + }, + { + "epoch": 4.47, + "learning_rate": 8.575631153862027e-06, + "loss": 0.1535, + "step": 10682 + }, + { + "epoch": 4.47, + "learning_rate": 8.573953052042959e-06, + "loss": 0.147, + "step": 10683 + }, + { + "epoch": 4.47, + "learning_rate": 8.572274991214345e-06, + "loss": 0.1321, + "step": 10684 + }, + { + "epoch": 4.47, + "learning_rate": 8.570596971424421e-06, + "loss": 0.1436, + "step": 10685 + }, + { + "epoch": 4.47, + "learning_rate": 8.568918992721418e-06, + "loss": 0.121, + "step": 10686 + }, + { + "epoch": 4.47, + "learning_rate": 8.567241055153567e-06, + "loss": 0.137, + "step": 10687 + }, + { + "epoch": 4.47, + "learning_rate": 8.565563158769103e-06, + "loss": 0.1576, + "step": 10688 + }, + { + "epoch": 4.48, + "learning_rate": 8.563885303616253e-06, + "loss": 0.1183, + "step": 10689 + }, + { + "epoch": 4.48, + "learning_rate": 8.562207489743245e-06, + "loss": 0.1324, + "step": 10690 + }, + { + "epoch": 4.48, + "learning_rate": 8.560529717198305e-06, + "loss": 0.1612, + "step": 10691 + }, + { + "epoch": 4.48, + "learning_rate": 8.558851986029663e-06, + "loss": 0.1455, + "step": 10692 + }, + { + "epoch": 4.48, + "learning_rate": 8.557174296285542e-06, + "loss": 0.1424, + "step": 10693 + }, + { + "epoch": 4.48, + "learning_rate": 8.555496648014163e-06, + "loss": 0.1679, + "step": 10694 + }, + { + "epoch": 4.48, + "learning_rate": 8.553819041263752e-06, + "loss": 0.15, + "step": 10695 + }, + { + "epoch": 4.48, + "learning_rate": 8.552141476082531e-06, + "loss": 0.1507, + "step": 10696 + }, + { + "epoch": 4.48, + "learning_rate": 8.550463952518716e-06, + "loss": 0.1299, + "step": 10697 + }, + { + "epoch": 4.48, + "learning_rate": 8.54878647062053e-06, + "loss": 0.1896, + "step": 10698 + }, + { + "epoch": 4.48, + "learning_rate": 8.547109030436189e-06, + "loss": 0.1801, + "step": 10699 + }, + { + "epoch": 4.48, + "learning_rate": 8.545431632013909e-06, + "loss": 0.1252, + "step": 10700 + }, + { + "epoch": 4.48, + "learning_rate": 8.543754275401906e-06, + "loss": 0.1409, + "step": 10701 + }, + { + "epoch": 4.48, + "learning_rate": 8.542076960648395e-06, + "loss": 0.1873, + "step": 10702 + }, + { + "epoch": 4.48, + "learning_rate": 8.540399687801588e-06, + "loss": 0.1732, + "step": 10703 + }, + { + "epoch": 4.48, + "learning_rate": 8.538722456909694e-06, + "loss": 0.1139, + "step": 10704 + }, + { + "epoch": 4.48, + "learning_rate": 8.53704526802093e-06, + "loss": 0.157, + "step": 10705 + }, + { + "epoch": 4.48, + "learning_rate": 8.5353681211835e-06, + "loss": 0.1486, + "step": 10706 + }, + { + "epoch": 4.48, + "learning_rate": 8.533691016445614e-06, + "loss": 0.1798, + "step": 10707 + }, + { + "epoch": 4.48, + "learning_rate": 8.532013953855477e-06, + "loss": 0.1528, + "step": 10708 + }, + { + "epoch": 4.48, + "learning_rate": 8.530336933461298e-06, + "loss": 0.1536, + "step": 10709 + }, + { + "epoch": 4.48, + "learning_rate": 8.52865995531128e-06, + "loss": 0.1856, + "step": 10710 + }, + { + "epoch": 4.48, + "learning_rate": 8.526983019453624e-06, + "loss": 0.1538, + "step": 10711 + }, + { + "epoch": 4.48, + "learning_rate": 8.52530612593654e-06, + "loss": 0.1494, + "step": 10712 + }, + { + "epoch": 4.49, + "learning_rate": 8.523629274808219e-06, + "loss": 0.1811, + "step": 10713 + }, + { + "epoch": 4.49, + "learning_rate": 8.521952466116862e-06, + "loss": 0.2089, + "step": 10714 + }, + { + "epoch": 4.49, + "learning_rate": 8.52027569991067e-06, + "loss": 0.1166, + "step": 10715 + }, + { + "epoch": 4.49, + "learning_rate": 8.51859897623784e-06, + "loss": 0.1852, + "step": 10716 + }, + { + "epoch": 4.49, + "learning_rate": 8.51692229514657e-06, + "loss": 0.1098, + "step": 10717 + }, + { + "epoch": 4.49, + "learning_rate": 8.515245656685051e-06, + "loss": 0.1062, + "step": 10718 + }, + { + "epoch": 4.49, + "learning_rate": 8.513569060901477e-06, + "loss": 0.1651, + "step": 10719 + }, + { + "epoch": 4.49, + "learning_rate": 8.511892507844045e-06, + "loss": 0.1643, + "step": 10720 + }, + { + "epoch": 4.49, + "learning_rate": 8.510215997560939e-06, + "loss": 0.1554, + "step": 10721 + }, + { + "epoch": 4.49, + "learning_rate": 8.508539530100352e-06, + "loss": 0.1887, + "step": 10722 + }, + { + "epoch": 4.49, + "learning_rate": 8.506863105510476e-06, + "loss": 0.1091, + "step": 10723 + }, + { + "epoch": 4.49, + "learning_rate": 8.505186723839492e-06, + "loss": 0.1308, + "step": 10724 + }, + { + "epoch": 4.49, + "learning_rate": 8.50351038513559e-06, + "loss": 0.1375, + "step": 10725 + }, + { + "epoch": 4.49, + "learning_rate": 8.501834089446956e-06, + "loss": 0.1762, + "step": 10726 + }, + { + "epoch": 4.49, + "learning_rate": 8.500157836821768e-06, + "loss": 0.1493, + "step": 10727 + }, + { + "epoch": 4.49, + "learning_rate": 8.498481627308215e-06, + "loss": 0.1416, + "step": 10728 + }, + { + "epoch": 4.49, + "learning_rate": 8.496805460954474e-06, + "loss": 0.1904, + "step": 10729 + }, + { + "epoch": 4.49, + "learning_rate": 8.495129337808725e-06, + "loss": 0.1275, + "step": 10730 + }, + { + "epoch": 4.49, + "learning_rate": 8.493453257919148e-06, + "loss": 0.1303, + "step": 10731 + }, + { + "epoch": 4.49, + "learning_rate": 8.491777221333923e-06, + "loss": 0.1428, + "step": 10732 + }, + { + "epoch": 4.49, + "learning_rate": 8.490101228101222e-06, + "loss": 0.1747, + "step": 10733 + }, + { + "epoch": 4.49, + "learning_rate": 8.488425278269224e-06, + "loss": 0.1431, + "step": 10734 + }, + { + "epoch": 4.49, + "learning_rate": 8.486749371886096e-06, + "loss": 0.1246, + "step": 10735 + }, + { + "epoch": 4.49, + "learning_rate": 8.485073509000015e-06, + "loss": 0.136, + "step": 10736 + }, + { + "epoch": 4.5, + "learning_rate": 8.483397689659153e-06, + "loss": 0.1523, + "step": 10737 + }, + { + "epoch": 4.5, + "learning_rate": 8.481721913911677e-06, + "loss": 0.1385, + "step": 10738 + }, + { + "epoch": 4.5, + "learning_rate": 8.480046181805757e-06, + "loss": 0.1603, + "step": 10739 + }, + { + "epoch": 4.5, + "learning_rate": 8.478370493389563e-06, + "loss": 0.1634, + "step": 10740 + }, + { + "epoch": 4.5, + "learning_rate": 8.476694848711258e-06, + "loss": 0.1531, + "step": 10741 + }, + { + "epoch": 4.5, + "learning_rate": 8.475019247819006e-06, + "loss": 0.1607, + "step": 10742 + }, + { + "epoch": 4.5, + "learning_rate": 8.473343690760976e-06, + "loss": 0.1643, + "step": 10743 + }, + { + "epoch": 4.5, + "learning_rate": 8.471668177585324e-06, + "loss": 0.1783, + "step": 10744 + }, + { + "epoch": 4.5, + "learning_rate": 8.469992708340213e-06, + "loss": 0.1064, + "step": 10745 + }, + { + "epoch": 4.5, + "learning_rate": 8.468317283073805e-06, + "loss": 0.1407, + "step": 10746 + }, + { + "epoch": 4.5, + "learning_rate": 8.46664190183426e-06, + "loss": 0.1761, + "step": 10747 + }, + { + "epoch": 4.5, + "learning_rate": 8.46496656466973e-06, + "loss": 0.143, + "step": 10748 + }, + { + "epoch": 4.5, + "learning_rate": 8.463291271628375e-06, + "loss": 0.1405, + "step": 10749 + }, + { + "epoch": 4.5, + "learning_rate": 8.46161602275835e-06, + "loss": 0.1192, + "step": 10750 + }, + { + "epoch": 4.5, + "learning_rate": 8.459940818107805e-06, + "loss": 0.1546, + "step": 10751 + }, + { + "epoch": 4.5, + "learning_rate": 8.458265657724894e-06, + "loss": 0.1785, + "step": 10752 + }, + { + "epoch": 4.5, + "learning_rate": 8.456590541657772e-06, + "loss": 0.1608, + "step": 10753 + }, + { + "epoch": 4.5, + "learning_rate": 8.454915469954583e-06, + "loss": 0.1609, + "step": 10754 + }, + { + "epoch": 4.5, + "learning_rate": 8.453240442663477e-06, + "loss": 0.1414, + "step": 10755 + }, + { + "epoch": 4.5, + "learning_rate": 8.451565459832604e-06, + "loss": 0.1398, + "step": 10756 + }, + { + "epoch": 4.5, + "learning_rate": 8.44989052151011e-06, + "loss": 0.1418, + "step": 10757 + }, + { + "epoch": 4.5, + "learning_rate": 8.448215627744134e-06, + "loss": 0.1435, + "step": 10758 + }, + { + "epoch": 4.5, + "learning_rate": 8.446540778582823e-06, + "loss": 0.1296, + "step": 10759 + }, + { + "epoch": 4.5, + "learning_rate": 8.444865974074319e-06, + "loss": 0.2264, + "step": 10760 + }, + { + "epoch": 4.51, + "learning_rate": 8.443191214266763e-06, + "loss": 0.1684, + "step": 10761 + }, + { + "epoch": 4.51, + "learning_rate": 8.441516499208292e-06, + "loss": 0.1324, + "step": 10762 + }, + { + "epoch": 4.51, + "learning_rate": 8.439841828947048e-06, + "loss": 0.1965, + "step": 10763 + }, + { + "epoch": 4.51, + "learning_rate": 8.438167203531166e-06, + "loss": 0.1786, + "step": 10764 + }, + { + "epoch": 4.51, + "learning_rate": 8.436492623008782e-06, + "loss": 0.1435, + "step": 10765 + }, + { + "epoch": 4.51, + "learning_rate": 8.434818087428029e-06, + "loss": 0.1287, + "step": 10766 + }, + { + "epoch": 4.51, + "learning_rate": 8.433143596837043e-06, + "loss": 0.1772, + "step": 10767 + }, + { + "epoch": 4.51, + "learning_rate": 8.431469151283953e-06, + "loss": 0.1788, + "step": 10768 + }, + { + "epoch": 4.51, + "learning_rate": 8.429794750816888e-06, + "loss": 0.1539, + "step": 10769 + }, + { + "epoch": 4.51, + "learning_rate": 8.428120395483982e-06, + "loss": 0.14, + "step": 10770 + }, + { + "epoch": 4.51, + "learning_rate": 8.42644608533336e-06, + "loss": 0.1312, + "step": 10771 + }, + { + "epoch": 4.51, + "learning_rate": 8.42477182041315e-06, + "loss": 0.1656, + "step": 10772 + }, + { + "epoch": 4.51, + "learning_rate": 8.423097600771476e-06, + "loss": 0.1824, + "step": 10773 + }, + { + "epoch": 4.51, + "learning_rate": 8.42142342645646e-06, + "loss": 0.1685, + "step": 10774 + }, + { + "epoch": 4.51, + "learning_rate": 8.419749297516227e-06, + "loss": 0.1578, + "step": 10775 + }, + { + "epoch": 4.51, + "learning_rate": 8.4180752139989e-06, + "loss": 0.1387, + "step": 10776 + }, + { + "epoch": 4.51, + "learning_rate": 8.416401175952598e-06, + "loss": 0.1427, + "step": 10777 + }, + { + "epoch": 4.51, + "learning_rate": 8.414727183425437e-06, + "loss": 0.1592, + "step": 10778 + }, + { + "epoch": 4.51, + "learning_rate": 8.413053236465535e-06, + "loss": 0.1332, + "step": 10779 + }, + { + "epoch": 4.51, + "learning_rate": 8.411379335121014e-06, + "loss": 0.1034, + "step": 10780 + }, + { + "epoch": 4.51, + "learning_rate": 8.409705479439985e-06, + "loss": 0.185, + "step": 10781 + }, + { + "epoch": 4.51, + "learning_rate": 8.408031669470555e-06, + "loss": 0.1347, + "step": 10782 + }, + { + "epoch": 4.51, + "learning_rate": 8.406357905260845e-06, + "loss": 0.1821, + "step": 10783 + }, + { + "epoch": 4.51, + "learning_rate": 8.404684186858962e-06, + "loss": 0.1364, + "step": 10784 + }, + { + "epoch": 4.52, + "learning_rate": 8.403010514313016e-06, + "loss": 0.1522, + "step": 10785 + }, + { + "epoch": 4.52, + "learning_rate": 8.401336887671117e-06, + "loss": 0.1601, + "step": 10786 + }, + { + "epoch": 4.52, + "learning_rate": 8.399663306981368e-06, + "loss": 0.1565, + "step": 10787 + }, + { + "epoch": 4.52, + "learning_rate": 8.39798977229188e-06, + "loss": 0.1307, + "step": 10788 + }, + { + "epoch": 4.52, + "learning_rate": 8.396316283650753e-06, + "loss": 0.1278, + "step": 10789 + }, + { + "epoch": 4.52, + "learning_rate": 8.39464284110609e-06, + "loss": 0.1808, + "step": 10790 + }, + { + "epoch": 4.52, + "learning_rate": 8.392969444705998e-06, + "loss": 0.1575, + "step": 10791 + }, + { + "epoch": 4.52, + "learning_rate": 8.39129609449857e-06, + "loss": 0.163, + "step": 10792 + }, + { + "epoch": 4.52, + "learning_rate": 8.389622790531908e-06, + "loss": 0.1786, + "step": 10793 + }, + { + "epoch": 4.52, + "learning_rate": 8.387949532854113e-06, + "loss": 0.1419, + "step": 10794 + }, + { + "epoch": 4.52, + "learning_rate": 8.386276321513275e-06, + "loss": 0.1721, + "step": 10795 + }, + { + "epoch": 4.52, + "learning_rate": 8.384603156557492e-06, + "loss": 0.145, + "step": 10796 + }, + { + "epoch": 4.52, + "learning_rate": 8.38293003803486e-06, + "loss": 0.1321, + "step": 10797 + }, + { + "epoch": 4.52, + "learning_rate": 8.381256965993468e-06, + "loss": 0.1727, + "step": 10798 + }, + { + "epoch": 4.52, + "learning_rate": 8.379583940481407e-06, + "loss": 0.1064, + "step": 10799 + }, + { + "epoch": 4.52, + "learning_rate": 8.377910961546771e-06, + "loss": 0.1787, + "step": 10800 + }, + { + "epoch": 4.52, + "learning_rate": 8.376238029237643e-06, + "loss": 0.1235, + "step": 10801 + }, + { + "epoch": 4.52, + "learning_rate": 8.374565143602114e-06, + "loss": 0.1486, + "step": 10802 + }, + { + "epoch": 4.52, + "learning_rate": 8.372892304688265e-06, + "loss": 0.1193, + "step": 10803 + }, + { + "epoch": 4.52, + "learning_rate": 8.371219512544182e-06, + "loss": 0.1624, + "step": 10804 + }, + { + "epoch": 4.52, + "learning_rate": 8.369546767217952e-06, + "loss": 0.1455, + "step": 10805 + }, + { + "epoch": 4.52, + "learning_rate": 8.367874068757651e-06, + "loss": 0.1264, + "step": 10806 + }, + { + "epoch": 4.52, + "learning_rate": 8.366201417211361e-06, + "loss": 0.171, + "step": 10807 + }, + { + "epoch": 4.53, + "learning_rate": 8.364528812627164e-06, + "loss": 0.1191, + "step": 10808 + }, + { + "epoch": 4.53, + "learning_rate": 8.362856255053132e-06, + "loss": 0.1946, + "step": 10809 + }, + { + "epoch": 4.53, + "learning_rate": 8.361183744537345e-06, + "loss": 0.1596, + "step": 10810 + }, + { + "epoch": 4.53, + "learning_rate": 8.359511281127879e-06, + "loss": 0.2023, + "step": 10811 + }, + { + "epoch": 4.53, + "learning_rate": 8.357838864872803e-06, + "loss": 0.1452, + "step": 10812 + }, + { + "epoch": 4.53, + "learning_rate": 8.356166495820192e-06, + "loss": 0.1487, + "step": 10813 + }, + { + "epoch": 4.53, + "learning_rate": 8.354494174018118e-06, + "loss": 0.1538, + "step": 10814 + }, + { + "epoch": 4.53, + "learning_rate": 8.352821899514648e-06, + "loss": 0.1443, + "step": 10815 + }, + { + "epoch": 4.53, + "learning_rate": 8.35114967235785e-06, + "loss": 0.1427, + "step": 10816 + }, + { + "epoch": 4.53, + "learning_rate": 8.34947749259579e-06, + "loss": 0.1366, + "step": 10817 + }, + { + "epoch": 4.53, + "learning_rate": 8.347805360276538e-06, + "loss": 0.157, + "step": 10818 + }, + { + "epoch": 4.53, + "learning_rate": 8.346133275448153e-06, + "loss": 0.1599, + "step": 10819 + }, + { + "epoch": 4.53, + "learning_rate": 8.3444612381587e-06, + "loss": 0.1406, + "step": 10820 + }, + { + "epoch": 4.53, + "learning_rate": 8.342789248456242e-06, + "loss": 0.1796, + "step": 10821 + }, + { + "epoch": 4.53, + "learning_rate": 8.341117306388833e-06, + "loss": 0.15, + "step": 10822 + }, + { + "epoch": 4.53, + "learning_rate": 8.339445412004533e-06, + "loss": 0.1105, + "step": 10823 + }, + { + "epoch": 4.53, + "learning_rate": 8.337773565351407e-06, + "loss": 0.153, + "step": 10824 + }, + { + "epoch": 4.53, + "learning_rate": 8.336101766477506e-06, + "loss": 0.1288, + "step": 10825 + }, + { + "epoch": 4.53, + "learning_rate": 8.334430015430878e-06, + "loss": 0.1934, + "step": 10826 + }, + { + "epoch": 4.53, + "learning_rate": 8.332758312259583e-06, + "loss": 0.1263, + "step": 10827 + }, + { + "epoch": 4.53, + "learning_rate": 8.331086657011669e-06, + "loss": 0.1337, + "step": 10828 + }, + { + "epoch": 4.53, + "learning_rate": 8.32941504973519e-06, + "loss": 0.1528, + "step": 10829 + }, + { + "epoch": 4.53, + "learning_rate": 8.327743490478192e-06, + "loss": 0.1801, + "step": 10830 + }, + { + "epoch": 4.53, + "learning_rate": 8.326071979288723e-06, + "loss": 0.1432, + "step": 10831 + }, + { + "epoch": 4.54, + "learning_rate": 8.32440051621483e-06, + "loss": 0.1296, + "step": 10832 + }, + { + "epoch": 4.54, + "learning_rate": 8.322729101304558e-06, + "loss": 0.1795, + "step": 10833 + }, + { + "epoch": 4.54, + "learning_rate": 8.321057734605949e-06, + "loss": 0.1599, + "step": 10834 + }, + { + "epoch": 4.54, + "learning_rate": 8.319386416167047e-06, + "loss": 0.1965, + "step": 10835 + }, + { + "epoch": 4.54, + "learning_rate": 8.31771514603589e-06, + "loss": 0.1628, + "step": 10836 + }, + { + "epoch": 4.54, + "learning_rate": 8.316043924260514e-06, + "loss": 0.1537, + "step": 10837 + }, + { + "epoch": 4.54, + "learning_rate": 8.314372750888968e-06, + "loss": 0.1814, + "step": 10838 + }, + { + "epoch": 4.54, + "learning_rate": 8.312701625969278e-06, + "loss": 0.1303, + "step": 10839 + }, + { + "epoch": 4.54, + "learning_rate": 8.311030549549483e-06, + "loss": 0.1532, + "step": 10840 + }, + { + "epoch": 4.54, + "learning_rate": 8.309359521677617e-06, + "loss": 0.1477, + "step": 10841 + }, + { + "epoch": 4.54, + "learning_rate": 8.307688542401709e-06, + "loss": 0.1307, + "step": 10842 + }, + { + "epoch": 4.54, + "learning_rate": 8.306017611769795e-06, + "loss": 0.122, + "step": 10843 + }, + { + "epoch": 4.54, + "learning_rate": 8.3043467298299e-06, + "loss": 0.1751, + "step": 10844 + }, + { + "epoch": 4.54, + "learning_rate": 8.302675896630054e-06, + "loss": 0.1655, + "step": 10845 + }, + { + "epoch": 4.54, + "learning_rate": 8.301005112218282e-06, + "loss": 0.1444, + "step": 10846 + }, + { + "epoch": 4.54, + "learning_rate": 8.299334376642609e-06, + "loss": 0.1179, + "step": 10847 + }, + { + "epoch": 4.54, + "learning_rate": 8.297663689951067e-06, + "loss": 0.1252, + "step": 10848 + }, + { + "epoch": 4.54, + "learning_rate": 8.295993052191669e-06, + "loss": 0.1143, + "step": 10849 + }, + { + "epoch": 4.54, + "learning_rate": 8.294322463412435e-06, + "loss": 0.132, + "step": 10850 + }, + { + "epoch": 4.54, + "learning_rate": 8.292651923661391e-06, + "loss": 0.1182, + "step": 10851 + }, + { + "epoch": 4.54, + "learning_rate": 8.290981432986551e-06, + "loss": 0.1625, + "step": 10852 + }, + { + "epoch": 4.54, + "learning_rate": 8.289310991435933e-06, + "loss": 0.1747, + "step": 10853 + }, + { + "epoch": 4.54, + "learning_rate": 8.287640599057551e-06, + "loss": 0.162, + "step": 10854 + }, + { + "epoch": 4.54, + "learning_rate": 8.285970255899425e-06, + "loss": 0.1723, + "step": 10855 + }, + { + "epoch": 4.55, + "learning_rate": 8.284299962009559e-06, + "loss": 0.1552, + "step": 10856 + }, + { + "epoch": 4.55, + "learning_rate": 8.28262971743597e-06, + "loss": 0.1625, + "step": 10857 + }, + { + "epoch": 4.55, + "learning_rate": 8.280959522226663e-06, + "loss": 0.1036, + "step": 10858 + }, + { + "epoch": 4.55, + "learning_rate": 8.279289376429653e-06, + "loss": 0.1547, + "step": 10859 + }, + { + "epoch": 4.55, + "learning_rate": 8.27761928009294e-06, + "loss": 0.1245, + "step": 10860 + }, + { + "epoch": 4.55, + "learning_rate": 8.275949233264534e-06, + "loss": 0.1447, + "step": 10861 + }, + { + "epoch": 4.55, + "learning_rate": 8.274279235992438e-06, + "loss": 0.1469, + "step": 10862 + }, + { + "epoch": 4.55, + "learning_rate": 8.272609288324653e-06, + "loss": 0.1636, + "step": 10863 + }, + { + "epoch": 4.55, + "learning_rate": 8.270939390309181e-06, + "loss": 0.2064, + "step": 10864 + }, + { + "epoch": 4.55, + "learning_rate": 8.269269541994022e-06, + "loss": 0.1293, + "step": 10865 + }, + { + "epoch": 4.55, + "learning_rate": 8.267599743427173e-06, + "loss": 0.126, + "step": 10866 + }, + { + "epoch": 4.55, + "learning_rate": 8.265929994656634e-06, + "loss": 0.1713, + "step": 10867 + }, + { + "epoch": 4.55, + "learning_rate": 8.2642602957304e-06, + "loss": 0.1381, + "step": 10868 + }, + { + "epoch": 4.55, + "learning_rate": 8.26259064669646e-06, + "loss": 0.0982, + "step": 10869 + }, + { + "epoch": 4.55, + "learning_rate": 8.260921047602814e-06, + "loss": 0.1539, + "step": 10870 + }, + { + "epoch": 4.55, + "learning_rate": 8.259251498497446e-06, + "loss": 0.1549, + "step": 10871 + }, + { + "epoch": 4.55, + "learning_rate": 8.25758199942835e-06, + "loss": 0.1301, + "step": 10872 + }, + { + "epoch": 4.55, + "learning_rate": 8.255912550443515e-06, + "loss": 0.133, + "step": 10873 + }, + { + "epoch": 4.55, + "learning_rate": 8.254243151590922e-06, + "loss": 0.195, + "step": 10874 + }, + { + "epoch": 4.55, + "learning_rate": 8.252573802918562e-06, + "loss": 0.1498, + "step": 10875 + }, + { + "epoch": 4.55, + "learning_rate": 8.25090450447442e-06, + "loss": 0.173, + "step": 10876 + }, + { + "epoch": 4.55, + "learning_rate": 8.249235256306472e-06, + "loss": 0.1879, + "step": 10877 + }, + { + "epoch": 4.55, + "learning_rate": 8.247566058462704e-06, + "loss": 0.1443, + "step": 10878 + }, + { + "epoch": 4.55, + "learning_rate": 8.245896910991096e-06, + "loss": 0.1435, + "step": 10879 + }, + { + "epoch": 4.56, + "learning_rate": 8.244227813939624e-06, + "loss": 0.1807, + "step": 10880 + }, + { + "epoch": 4.56, + "learning_rate": 8.242558767356266e-06, + "loss": 0.1714, + "step": 10881 + }, + { + "epoch": 4.56, + "learning_rate": 8.240889771288997e-06, + "loss": 0.1307, + "step": 10882 + }, + { + "epoch": 4.56, + "learning_rate": 8.239220825785789e-06, + "loss": 0.1448, + "step": 10883 + }, + { + "epoch": 4.56, + "learning_rate": 8.237551930894616e-06, + "loss": 0.1597, + "step": 10884 + }, + { + "epoch": 4.56, + "learning_rate": 8.23588308666345e-06, + "loss": 0.1435, + "step": 10885 + }, + { + "epoch": 4.56, + "learning_rate": 8.234214293140258e-06, + "loss": 0.1298, + "step": 10886 + }, + { + "epoch": 4.56, + "learning_rate": 8.23254555037301e-06, + "loss": 0.1396, + "step": 10887 + }, + { + "epoch": 4.56, + "learning_rate": 8.230876858409669e-06, + "loss": 0.1867, + "step": 10888 + }, + { + "epoch": 4.56, + "learning_rate": 8.229208217298208e-06, + "loss": 0.149, + "step": 10889 + }, + { + "epoch": 4.56, + "learning_rate": 8.227539627086582e-06, + "loss": 0.1266, + "step": 10890 + }, + { + "epoch": 4.56, + "learning_rate": 8.225871087822757e-06, + "loss": 0.1307, + "step": 10891 + }, + { + "epoch": 4.56, + "learning_rate": 8.224202599554695e-06, + "loss": 0.1753, + "step": 10892 + }, + { + "epoch": 4.56, + "learning_rate": 8.222534162330355e-06, + "loss": 0.1343, + "step": 10893 + }, + { + "epoch": 4.56, + "learning_rate": 8.22086577619769e-06, + "loss": 0.145, + "step": 10894 + }, + { + "epoch": 4.56, + "learning_rate": 8.219197441204659e-06, + "loss": 0.1279, + "step": 10895 + }, + { + "epoch": 4.56, + "learning_rate": 8.21752915739922e-06, + "loss": 0.1412, + "step": 10896 + }, + { + "epoch": 4.56, + "learning_rate": 8.21586092482932e-06, + "loss": 0.1737, + "step": 10897 + }, + { + "epoch": 4.56, + "learning_rate": 8.214192743542914e-06, + "loss": 0.1382, + "step": 10898 + }, + { + "epoch": 4.56, + "learning_rate": 8.212524613587955e-06, + "loss": 0.14, + "step": 10899 + }, + { + "epoch": 4.56, + "learning_rate": 8.210856535012392e-06, + "loss": 0.1652, + "step": 10900 + }, + { + "epoch": 4.56, + "learning_rate": 8.209188507864167e-06, + "loss": 0.1024, + "step": 10901 + }, + { + "epoch": 4.56, + "learning_rate": 8.20752053219123e-06, + "loss": 0.2096, + "step": 10902 + }, + { + "epoch": 4.56, + "learning_rate": 8.205852608041526e-06, + "loss": 0.1255, + "step": 10903 + }, + { + "epoch": 4.57, + "learning_rate": 8.204184735462994e-06, + "loss": 0.1135, + "step": 10904 + }, + { + "epoch": 4.57, + "learning_rate": 8.20251691450358e-06, + "loss": 0.1712, + "step": 10905 + }, + { + "epoch": 4.57, + "learning_rate": 8.200849145211224e-06, + "loss": 0.1354, + "step": 10906 + }, + { + "epoch": 4.57, + "learning_rate": 8.19918142763386e-06, + "loss": 0.14, + "step": 10907 + }, + { + "epoch": 4.57, + "learning_rate": 8.197513761819428e-06, + "loss": 0.1466, + "step": 10908 + }, + { + "epoch": 4.57, + "learning_rate": 8.195846147815865e-06, + "loss": 0.1407, + "step": 10909 + }, + { + "epoch": 4.57, + "learning_rate": 8.194178585671102e-06, + "loss": 0.1501, + "step": 10910 + }, + { + "epoch": 4.57, + "learning_rate": 8.192511075433073e-06, + "loss": 0.1264, + "step": 10911 + }, + { + "epoch": 4.57, + "learning_rate": 8.190843617149712e-06, + "loss": 0.181, + "step": 10912 + }, + { + "epoch": 4.57, + "learning_rate": 8.189176210868942e-06, + "loss": 0.1677, + "step": 10913 + }, + { + "epoch": 4.57, + "learning_rate": 8.187508856638697e-06, + "loss": 0.174, + "step": 10914 + }, + { + "epoch": 4.57, + "learning_rate": 8.185841554506901e-06, + "loss": 0.1589, + "step": 10915 + }, + { + "epoch": 4.57, + "learning_rate": 8.184174304521483e-06, + "loss": 0.1424, + "step": 10916 + }, + { + "epoch": 4.57, + "learning_rate": 8.18250710673036e-06, + "loss": 0.1473, + "step": 10917 + }, + { + "epoch": 4.57, + "learning_rate": 8.180839961181458e-06, + "loss": 0.1487, + "step": 10918 + }, + { + "epoch": 4.57, + "learning_rate": 8.179172867922695e-06, + "loss": 0.1064, + "step": 10919 + }, + { + "epoch": 4.57, + "learning_rate": 8.177505827001997e-06, + "loss": 0.1629, + "step": 10920 + }, + { + "epoch": 4.57, + "learning_rate": 8.175838838467272e-06, + "loss": 0.168, + "step": 10921 + }, + { + "epoch": 4.57, + "learning_rate": 8.174171902366441e-06, + "loss": 0.1765, + "step": 10922 + }, + { + "epoch": 4.57, + "learning_rate": 8.172505018747421e-06, + "loss": 0.1488, + "step": 10923 + }, + { + "epoch": 4.57, + "learning_rate": 8.17083818765812e-06, + "loss": 0.1875, + "step": 10924 + }, + { + "epoch": 4.57, + "learning_rate": 8.169171409146453e-06, + "loss": 0.1606, + "step": 10925 + }, + { + "epoch": 4.57, + "learning_rate": 8.167504683260332e-06, + "loss": 0.1725, + "step": 10926 + }, + { + "epoch": 4.57, + "learning_rate": 8.16583801004766e-06, + "loss": 0.129, + "step": 10927 + }, + { + "epoch": 4.58, + "learning_rate": 8.164171389556346e-06, + "loss": 0.1683, + "step": 10928 + }, + { + "epoch": 4.58, + "learning_rate": 8.162504821834296e-06, + "loss": 0.1483, + "step": 10929 + }, + { + "epoch": 4.58, + "learning_rate": 8.160838306929418e-06, + "loss": 0.1661, + "step": 10930 + }, + { + "epoch": 4.58, + "learning_rate": 8.159171844889609e-06, + "loss": 0.149, + "step": 10931 + }, + { + "epoch": 4.58, + "learning_rate": 8.157505435762769e-06, + "loss": 0.161, + "step": 10932 + }, + { + "epoch": 4.58, + "learning_rate": 8.155839079596805e-06, + "loss": 0.1306, + "step": 10933 + }, + { + "epoch": 4.58, + "learning_rate": 8.154172776439608e-06, + "loss": 0.1119, + "step": 10934 + }, + { + "epoch": 4.58, + "learning_rate": 8.152506526339075e-06, + "loss": 0.1548, + "step": 10935 + }, + { + "epoch": 4.58, + "learning_rate": 8.150840329343104e-06, + "loss": 0.1768, + "step": 10936 + }, + { + "epoch": 4.58, + "learning_rate": 8.149174185499587e-06, + "loss": 0.1677, + "step": 10937 + }, + { + "epoch": 4.58, + "learning_rate": 8.14750809485642e-06, + "loss": 0.127, + "step": 10938 + }, + { + "epoch": 4.58, + "learning_rate": 8.145842057461483e-06, + "loss": 0.1572, + "step": 10939 + }, + { + "epoch": 4.58, + "learning_rate": 8.14417607336267e-06, + "loss": 0.1449, + "step": 10940 + }, + { + "epoch": 4.58, + "learning_rate": 8.142510142607874e-06, + "loss": 0.1592, + "step": 10941 + }, + { + "epoch": 4.58, + "learning_rate": 8.140844265244971e-06, + "loss": 0.1251, + "step": 10942 + }, + { + "epoch": 4.58, + "learning_rate": 8.139178441321851e-06, + "loss": 0.1163, + "step": 10943 + }, + { + "epoch": 4.58, + "learning_rate": 8.137512670886397e-06, + "loss": 0.1183, + "step": 10944 + }, + { + "epoch": 4.58, + "learning_rate": 8.135846953986486e-06, + "loss": 0.1977, + "step": 10945 + }, + { + "epoch": 4.58, + "learning_rate": 8.134181290670002e-06, + "loss": 0.1227, + "step": 10946 + }, + { + "epoch": 4.58, + "learning_rate": 8.132515680984821e-06, + "loss": 0.1084, + "step": 10947 + }, + { + "epoch": 4.58, + "learning_rate": 8.130850124978819e-06, + "loss": 0.163, + "step": 10948 + }, + { + "epoch": 4.58, + "learning_rate": 8.12918462269987e-06, + "loss": 0.1608, + "step": 10949 + }, + { + "epoch": 4.58, + "learning_rate": 8.127519174195853e-06, + "loss": 0.1399, + "step": 10950 + }, + { + "epoch": 4.58, + "learning_rate": 8.125853779514632e-06, + "loss": 0.1631, + "step": 10951 + }, + { + "epoch": 4.59, + "learning_rate": 8.124188438704082e-06, + "loss": 0.1099, + "step": 10952 + }, + { + "epoch": 4.59, + "learning_rate": 8.122523151812072e-06, + "loss": 0.1351, + "step": 10953 + }, + { + "epoch": 4.59, + "learning_rate": 8.120857918886467e-06, + "loss": 0.1401, + "step": 10954 + }, + { + "epoch": 4.59, + "learning_rate": 8.119192739975133e-06, + "loss": 0.1397, + "step": 10955 + }, + { + "epoch": 4.59, + "learning_rate": 8.117527615125938e-06, + "loss": 0.1302, + "step": 10956 + }, + { + "epoch": 4.59, + "learning_rate": 8.115862544386736e-06, + "loss": 0.1344, + "step": 10957 + }, + { + "epoch": 4.59, + "learning_rate": 8.114197527805396e-06, + "loss": 0.1636, + "step": 10958 + }, + { + "epoch": 4.59, + "learning_rate": 8.112532565429774e-06, + "loss": 0.1649, + "step": 10959 + }, + { + "epoch": 4.59, + "learning_rate": 8.11086765730773e-06, + "loss": 0.1666, + "step": 10960 + }, + { + "epoch": 4.59, + "learning_rate": 8.109202803487121e-06, + "loss": 0.1521, + "step": 10961 + }, + { + "epoch": 4.59, + "learning_rate": 8.107538004015796e-06, + "loss": 0.1471, + "step": 10962 + }, + { + "epoch": 4.59, + "learning_rate": 8.105873258941612e-06, + "loss": 0.1451, + "step": 10963 + }, + { + "epoch": 4.59, + "learning_rate": 8.10420856831242e-06, + "loss": 0.1674, + "step": 10964 + }, + { + "epoch": 4.59, + "learning_rate": 8.10254393217607e-06, + "loss": 0.1801, + "step": 10965 + }, + { + "epoch": 4.59, + "learning_rate": 8.100879350580411e-06, + "loss": 0.1484, + "step": 10966 + }, + { + "epoch": 4.59, + "learning_rate": 8.099214823573293e-06, + "loss": 0.1657, + "step": 10967 + }, + { + "epoch": 4.59, + "learning_rate": 8.097550351202554e-06, + "loss": 0.179, + "step": 10968 + }, + { + "epoch": 4.59, + "learning_rate": 8.095885933516043e-06, + "loss": 0.1169, + "step": 10969 + }, + { + "epoch": 4.59, + "learning_rate": 8.094221570561602e-06, + "loss": 0.1453, + "step": 10970 + }, + { + "epoch": 4.59, + "learning_rate": 8.09255726238707e-06, + "loss": 0.1403, + "step": 10971 + }, + { + "epoch": 4.59, + "learning_rate": 8.090893009040288e-06, + "loss": 0.1554, + "step": 10972 + }, + { + "epoch": 4.59, + "learning_rate": 8.08922881056909e-06, + "loss": 0.1699, + "step": 10973 + }, + { + "epoch": 4.59, + "learning_rate": 8.087564667021317e-06, + "loss": 0.1783, + "step": 10974 + }, + { + "epoch": 4.59, + "learning_rate": 8.0859005784448e-06, + "loss": 0.1717, + "step": 10975 + }, + { + "epoch": 4.6, + "learning_rate": 8.084236544887372e-06, + "loss": 0.1571, + "step": 10976 + }, + { + "epoch": 4.6, + "learning_rate": 8.082572566396864e-06, + "loss": 0.1587, + "step": 10977 + }, + { + "epoch": 4.6, + "learning_rate": 8.080908643021107e-06, + "loss": 0.1206, + "step": 10978 + }, + { + "epoch": 4.6, + "learning_rate": 8.079244774807926e-06, + "loss": 0.1667, + "step": 10979 + }, + { + "epoch": 4.6, + "learning_rate": 8.077580961805153e-06, + "loss": 0.1552, + "step": 10980 + }, + { + "epoch": 4.6, + "learning_rate": 8.075917204060606e-06, + "loss": 0.1245, + "step": 10981 + }, + { + "epoch": 4.6, + "learning_rate": 8.074253501622113e-06, + "loss": 0.1779, + "step": 10982 + }, + { + "epoch": 4.6, + "learning_rate": 8.072589854537498e-06, + "loss": 0.1806, + "step": 10983 + }, + { + "epoch": 4.6, + "learning_rate": 8.070926262854573e-06, + "loss": 0.1765, + "step": 10984 + }, + { + "epoch": 4.6, + "learning_rate": 8.069262726621163e-06, + "loss": 0.1469, + "step": 10985 + }, + { + "epoch": 4.6, + "learning_rate": 8.06759924588508e-06, + "loss": 0.1461, + "step": 10986 + }, + { + "epoch": 4.6, + "learning_rate": 8.065935820694142e-06, + "loss": 0.1394, + "step": 10987 + }, + { + "epoch": 4.6, + "learning_rate": 8.064272451096165e-06, + "loss": 0.1073, + "step": 10988 + }, + { + "epoch": 4.6, + "learning_rate": 8.062609137138955e-06, + "loss": 0.126, + "step": 10989 + }, + { + "epoch": 4.6, + "learning_rate": 8.060945878870328e-06, + "loss": 0.1782, + "step": 10990 + }, + { + "epoch": 4.6, + "learning_rate": 8.059282676338092e-06, + "loss": 0.1596, + "step": 10991 + }, + { + "epoch": 4.6, + "learning_rate": 8.057619529590051e-06, + "loss": 0.138, + "step": 10992 + }, + { + "epoch": 4.6, + "learning_rate": 8.055956438674015e-06, + "loss": 0.1796, + "step": 10993 + }, + { + "epoch": 4.6, + "learning_rate": 8.054293403637787e-06, + "loss": 0.153, + "step": 10994 + }, + { + "epoch": 4.6, + "learning_rate": 8.052630424529166e-06, + "loss": 0.1554, + "step": 10995 + }, + { + "epoch": 4.6, + "learning_rate": 8.050967501395955e-06, + "loss": 0.1505, + "step": 10996 + }, + { + "epoch": 4.6, + "learning_rate": 8.049304634285957e-06, + "loss": 0.1279, + "step": 10997 + }, + { + "epoch": 4.6, + "learning_rate": 8.047641823246964e-06, + "loss": 0.1643, + "step": 10998 + }, + { + "epoch": 4.6, + "learning_rate": 8.045979068326773e-06, + "loss": 0.1534, + "step": 10999 + }, + { + "epoch": 4.61, + "learning_rate": 8.044316369573181e-06, + "loss": 0.1411, + "step": 11000 + }, + { + "epoch": 4.61, + "learning_rate": 8.042653727033981e-06, + "loss": 0.1839, + "step": 11001 + }, + { + "epoch": 4.61, + "learning_rate": 8.04099114075696e-06, + "loss": 0.158, + "step": 11002 + }, + { + "epoch": 4.61, + "learning_rate": 8.039328610789911e-06, + "loss": 0.1694, + "step": 11003 + }, + { + "epoch": 4.61, + "learning_rate": 8.037666137180624e-06, + "loss": 0.1096, + "step": 11004 + }, + { + "epoch": 4.61, + "learning_rate": 8.03600371997688e-06, + "loss": 0.1871, + "step": 11005 + }, + { + "epoch": 4.61, + "learning_rate": 8.03434135922647e-06, + "loss": 0.12, + "step": 11006 + }, + { + "epoch": 4.61, + "learning_rate": 8.032679054977171e-06, + "loss": 0.1471, + "step": 11007 + }, + { + "epoch": 4.61, + "learning_rate": 8.031016807276765e-06, + "loss": 0.1441, + "step": 11008 + }, + { + "epoch": 4.61, + "learning_rate": 8.029354616173036e-06, + "loss": 0.1719, + "step": 11009 + }, + { + "epoch": 4.61, + "learning_rate": 8.027692481713757e-06, + "loss": 0.1434, + "step": 11010 + }, + { + "epoch": 4.61, + "learning_rate": 8.026030403946707e-06, + "loss": 0.1538, + "step": 11011 + }, + { + "epoch": 4.61, + "learning_rate": 8.024368382919665e-06, + "loss": 0.1316, + "step": 11012 + }, + { + "epoch": 4.61, + "learning_rate": 8.022706418680398e-06, + "loss": 0.1592, + "step": 11013 + }, + { + "epoch": 4.61, + "learning_rate": 8.021044511276682e-06, + "loss": 0.1634, + "step": 11014 + }, + { + "epoch": 4.61, + "learning_rate": 8.019382660756284e-06, + "loss": 0.1346, + "step": 11015 + }, + { + "epoch": 4.61, + "learning_rate": 8.017720867166976e-06, + "loss": 0.142, + "step": 11016 + }, + { + "epoch": 4.61, + "learning_rate": 8.01605913055652e-06, + "loss": 0.1682, + "step": 11017 + }, + { + "epoch": 4.61, + "learning_rate": 8.014397450972685e-06, + "loss": 0.156, + "step": 11018 + }, + { + "epoch": 4.61, + "learning_rate": 8.012735828463233e-06, + "loss": 0.1653, + "step": 11019 + }, + { + "epoch": 4.61, + "learning_rate": 8.011074263075925e-06, + "loss": 0.1269, + "step": 11020 + }, + { + "epoch": 4.61, + "learning_rate": 8.009412754858525e-06, + "loss": 0.1328, + "step": 11021 + }, + { + "epoch": 4.61, + "learning_rate": 8.007751303858788e-06, + "loss": 0.1473, + "step": 11022 + }, + { + "epoch": 4.62, + "learning_rate": 8.00608991012447e-06, + "loss": 0.108, + "step": 11023 + }, + { + "epoch": 4.62, + "learning_rate": 8.00442857370333e-06, + "loss": 0.1977, + "step": 11024 + }, + { + "epoch": 4.62, + "learning_rate": 8.00276729464312e-06, + "loss": 0.1484, + "step": 11025 + }, + { + "epoch": 4.62, + "learning_rate": 8.00110607299159e-06, + "loss": 0.1721, + "step": 11026 + }, + { + "epoch": 4.62, + "learning_rate": 7.999444908796493e-06, + "loss": 0.1437, + "step": 11027 + }, + { + "epoch": 4.62, + "learning_rate": 7.99778380210558e-06, + "loss": 0.132, + "step": 11028 + }, + { + "epoch": 4.62, + "learning_rate": 7.996122752966596e-06, + "loss": 0.148, + "step": 11029 + }, + { + "epoch": 4.62, + "learning_rate": 7.99446176142728e-06, + "loss": 0.1464, + "step": 11030 + }, + { + "epoch": 4.62, + "learning_rate": 7.992800827535381e-06, + "loss": 0.1883, + "step": 11031 + }, + { + "epoch": 4.62, + "learning_rate": 7.991139951338645e-06, + "loss": 0.1545, + "step": 11032 + }, + { + "epoch": 4.62, + "learning_rate": 7.989479132884803e-06, + "loss": 0.1339, + "step": 11033 + }, + { + "epoch": 4.62, + "learning_rate": 7.987818372221603e-06, + "loss": 0.1387, + "step": 11034 + }, + { + "epoch": 4.62, + "learning_rate": 7.98615766939678e-06, + "loss": 0.1584, + "step": 11035 + }, + { + "epoch": 4.62, + "learning_rate": 7.984497024458062e-06, + "loss": 0.128, + "step": 11036 + }, + { + "epoch": 4.62, + "learning_rate": 7.982836437453192e-06, + "loss": 0.1636, + "step": 11037 + }, + { + "epoch": 4.62, + "learning_rate": 7.9811759084299e-06, + "loss": 0.1627, + "step": 11038 + }, + { + "epoch": 4.62, + "learning_rate": 7.979515437435913e-06, + "loss": 0.1573, + "step": 11039 + }, + { + "epoch": 4.62, + "learning_rate": 7.977855024518962e-06, + "loss": 0.1787, + "step": 11040 + }, + { + "epoch": 4.62, + "learning_rate": 7.976194669726773e-06, + "loss": 0.1583, + "step": 11041 + }, + { + "epoch": 4.62, + "learning_rate": 7.974534373107074e-06, + "loss": 0.1613, + "step": 11042 + }, + { + "epoch": 4.62, + "learning_rate": 7.972874134707585e-06, + "loss": 0.1165, + "step": 11043 + }, + { + "epoch": 4.62, + "learning_rate": 7.971213954576031e-06, + "loss": 0.1546, + "step": 11044 + }, + { + "epoch": 4.62, + "learning_rate": 7.969553832760134e-06, + "loss": 0.1246, + "step": 11045 + }, + { + "epoch": 4.62, + "learning_rate": 7.967893769307608e-06, + "loss": 0.1284, + "step": 11046 + }, + { + "epoch": 4.63, + "learning_rate": 7.966233764266172e-06, + "loss": 0.1343, + "step": 11047 + }, + { + "epoch": 4.63, + "learning_rate": 7.964573817683545e-06, + "loss": 0.1521, + "step": 11048 + }, + { + "epoch": 4.63, + "learning_rate": 7.962913929607433e-06, + "loss": 0.1585, + "step": 11049 + }, + { + "epoch": 4.63, + "learning_rate": 7.961254100085554e-06, + "loss": 0.132, + "step": 11050 + }, + { + "epoch": 4.63, + "learning_rate": 7.95959432916562e-06, + "loss": 0.1382, + "step": 11051 + }, + { + "epoch": 4.63, + "learning_rate": 7.957934616895332e-06, + "loss": 0.1273, + "step": 11052 + }, + { + "epoch": 4.63, + "learning_rate": 7.956274963322406e-06, + "loss": 0.1409, + "step": 11053 + }, + { + "epoch": 4.63, + "learning_rate": 7.954615368494538e-06, + "loss": 0.1615, + "step": 11054 + }, + { + "epoch": 4.63, + "learning_rate": 7.952955832459437e-06, + "loss": 0.1566, + "step": 11055 + }, + { + "epoch": 4.63, + "learning_rate": 7.951296355264805e-06, + "loss": 0.1366, + "step": 11056 + }, + { + "epoch": 4.63, + "learning_rate": 7.949636936958342e-06, + "loss": 0.1468, + "step": 11057 + }, + { + "epoch": 4.63, + "learning_rate": 7.947977577587745e-06, + "loss": 0.1938, + "step": 11058 + }, + { + "epoch": 4.63, + "learning_rate": 7.946318277200712e-06, + "loss": 0.1244, + "step": 11059 + }, + { + "epoch": 4.63, + "learning_rate": 7.944659035844937e-06, + "loss": 0.1601, + "step": 11060 + }, + { + "epoch": 4.63, + "learning_rate": 7.942999853568112e-06, + "loss": 0.1841, + "step": 11061 + }, + { + "epoch": 4.63, + "learning_rate": 7.941340730417934e-06, + "loss": 0.1518, + "step": 11062 + }, + { + "epoch": 4.63, + "learning_rate": 7.93968166644209e-06, + "loss": 0.1456, + "step": 11063 + }, + { + "epoch": 4.63, + "learning_rate": 7.938022661688265e-06, + "loss": 0.137, + "step": 11064 + }, + { + "epoch": 4.63, + "learning_rate": 7.936363716204153e-06, + "loss": 0.1501, + "step": 11065 + }, + { + "epoch": 4.63, + "learning_rate": 7.934704830037432e-06, + "loss": 0.1474, + "step": 11066 + }, + { + "epoch": 4.63, + "learning_rate": 7.933046003235789e-06, + "loss": 0.1521, + "step": 11067 + }, + { + "epoch": 4.63, + "learning_rate": 7.931387235846905e-06, + "loss": 0.1458, + "step": 11068 + }, + { + "epoch": 4.63, + "learning_rate": 7.929728527918456e-06, + "loss": 0.1276, + "step": 11069 + }, + { + "epoch": 4.63, + "learning_rate": 7.928069879498126e-06, + "loss": 0.1618, + "step": 11070 + }, + { + "epoch": 4.64, + "learning_rate": 7.926411290633588e-06, + "loss": 0.1723, + "step": 11071 + }, + { + "epoch": 4.64, + "learning_rate": 7.92475276137252e-06, + "loss": 0.151, + "step": 11072 + }, + { + "epoch": 4.64, + "learning_rate": 7.923094291762588e-06, + "loss": 0.1688, + "step": 11073 + }, + { + "epoch": 4.64, + "learning_rate": 7.921435881851474e-06, + "loss": 0.1447, + "step": 11074 + }, + { + "epoch": 4.64, + "learning_rate": 7.919777531686837e-06, + "loss": 0.1317, + "step": 11075 + }, + { + "epoch": 4.64, + "learning_rate": 7.91811924131635e-06, + "loss": 0.1612, + "step": 11076 + }, + { + "epoch": 4.64, + "learning_rate": 7.916461010787676e-06, + "loss": 0.1556, + "step": 11077 + }, + { + "epoch": 4.64, + "learning_rate": 7.914802840148482e-06, + "loss": 0.1761, + "step": 11078 + }, + { + "epoch": 4.64, + "learning_rate": 7.91314472944643e-06, + "loss": 0.1232, + "step": 11079 + }, + { + "epoch": 4.64, + "learning_rate": 7.91148667872918e-06, + "loss": 0.1865, + "step": 11080 + }, + { + "epoch": 4.64, + "learning_rate": 7.909828688044392e-06, + "loss": 0.1404, + "step": 11081 + }, + { + "epoch": 4.64, + "learning_rate": 7.908170757439724e-06, + "loss": 0.148, + "step": 11082 + }, + { + "epoch": 4.64, + "learning_rate": 7.906512886962833e-06, + "loss": 0.1539, + "step": 11083 + }, + { + "epoch": 4.64, + "learning_rate": 7.90485507666137e-06, + "loss": 0.1607, + "step": 11084 + }, + { + "epoch": 4.64, + "learning_rate": 7.903197326582988e-06, + "loss": 0.1351, + "step": 11085 + }, + { + "epoch": 4.64, + "learning_rate": 7.90153963677534e-06, + "loss": 0.1095, + "step": 11086 + }, + { + "epoch": 4.64, + "learning_rate": 7.899882007286074e-06, + "loss": 0.1682, + "step": 11087 + }, + { + "epoch": 4.64, + "learning_rate": 7.898224438162831e-06, + "loss": 0.1204, + "step": 11088 + }, + { + "epoch": 4.64, + "learning_rate": 7.896566929453267e-06, + "loss": 0.166, + "step": 11089 + }, + { + "epoch": 4.64, + "learning_rate": 7.894909481205017e-06, + "loss": 0.118, + "step": 11090 + }, + { + "epoch": 4.64, + "learning_rate": 7.893252093465725e-06, + "loss": 0.133, + "step": 11091 + }, + { + "epoch": 4.64, + "learning_rate": 7.891594766283034e-06, + "loss": 0.1369, + "step": 11092 + }, + { + "epoch": 4.64, + "learning_rate": 7.88993749970458e-06, + "loss": 0.1728, + "step": 11093 + }, + { + "epoch": 4.64, + "learning_rate": 7.888280293778001e-06, + "loss": 0.1538, + "step": 11094 + }, + { + "epoch": 4.65, + "learning_rate": 7.886623148550932e-06, + "loss": 0.1335, + "step": 11095 + }, + { + "epoch": 4.65, + "learning_rate": 7.884966064071002e-06, + "loss": 0.1191, + "step": 11096 + }, + { + "epoch": 4.65, + "learning_rate": 7.88330904038585e-06, + "loss": 0.1459, + "step": 11097 + }, + { + "epoch": 4.65, + "learning_rate": 7.881652077543098e-06, + "loss": 0.1629, + "step": 11098 + }, + { + "epoch": 4.65, + "learning_rate": 7.879995175590378e-06, + "loss": 0.1501, + "step": 11099 + }, + { + "epoch": 4.65, + "learning_rate": 7.878338334575317e-06, + "loss": 0.189, + "step": 11100 + }, + { + "epoch": 4.65, + "learning_rate": 7.876681554545535e-06, + "loss": 0.1861, + "step": 11101 + }, + { + "epoch": 4.65, + "learning_rate": 7.875024835548658e-06, + "loss": 0.1758, + "step": 11102 + }, + { + "epoch": 4.65, + "learning_rate": 7.873368177632308e-06, + "loss": 0.1761, + "step": 11103 + }, + { + "epoch": 4.65, + "learning_rate": 7.871711580844102e-06, + "loss": 0.159, + "step": 11104 + }, + { + "epoch": 4.65, + "learning_rate": 7.870055045231656e-06, + "loss": 0.1334, + "step": 11105 + }, + { + "epoch": 4.65, + "learning_rate": 7.86839857084259e-06, + "loss": 0.1364, + "step": 11106 + }, + { + "epoch": 4.65, + "learning_rate": 7.866742157724515e-06, + "loss": 0.1134, + "step": 11107 + }, + { + "epoch": 4.65, + "learning_rate": 7.865085805925042e-06, + "loss": 0.1967, + "step": 11108 + }, + { + "epoch": 4.65, + "learning_rate": 7.863429515491786e-06, + "loss": 0.1544, + "step": 11109 + }, + { + "epoch": 4.65, + "learning_rate": 7.861773286472351e-06, + "loss": 0.1231, + "step": 11110 + }, + { + "epoch": 4.65, + "learning_rate": 7.860117118914345e-06, + "loss": 0.1251, + "step": 11111 + }, + { + "epoch": 4.65, + "learning_rate": 7.858461012865373e-06, + "loss": 0.1512, + "step": 11112 + }, + { + "epoch": 4.65, + "learning_rate": 7.856804968373041e-06, + "loss": 0.2215, + "step": 11113 + }, + { + "epoch": 4.65, + "learning_rate": 7.855148985484946e-06, + "loss": 0.135, + "step": 11114 + }, + { + "epoch": 4.65, + "learning_rate": 7.85349306424869e-06, + "loss": 0.1436, + "step": 11115 + }, + { + "epoch": 4.65, + "learning_rate": 7.851837204711874e-06, + "loss": 0.2166, + "step": 11116 + }, + { + "epoch": 4.65, + "learning_rate": 7.85018140692209e-06, + "loss": 0.1807, + "step": 11117 + }, + { + "epoch": 4.65, + "learning_rate": 7.848525670926932e-06, + "loss": 0.1591, + "step": 11118 + }, + { + "epoch": 4.66, + "learning_rate": 7.846869996774e-06, + "loss": 0.1464, + "step": 11119 + }, + { + "epoch": 4.66, + "learning_rate": 7.845214384510875e-06, + "loss": 0.1911, + "step": 11120 + }, + { + "epoch": 4.66, + "learning_rate": 7.84355883418515e-06, + "loss": 0.1695, + "step": 11121 + }, + { + "epoch": 4.66, + "learning_rate": 7.841903345844413e-06, + "loss": 0.1471, + "step": 11122 + }, + { + "epoch": 4.66, + "learning_rate": 7.840247919536249e-06, + "loss": 0.1358, + "step": 11123 + }, + { + "epoch": 4.66, + "learning_rate": 7.838592555308244e-06, + "loss": 0.1547, + "step": 11124 + }, + { + "epoch": 4.66, + "learning_rate": 7.836937253207976e-06, + "loss": 0.1534, + "step": 11125 + }, + { + "epoch": 4.66, + "learning_rate": 7.835282013283027e-06, + "loss": 0.1508, + "step": 11126 + }, + { + "epoch": 4.66, + "learning_rate": 7.833626835580977e-06, + "loss": 0.1305, + "step": 11127 + }, + { + "epoch": 4.66, + "learning_rate": 7.8319717201494e-06, + "loss": 0.1499, + "step": 11128 + }, + { + "epoch": 4.66, + "learning_rate": 7.830316667035872e-06, + "loss": 0.1538, + "step": 11129 + }, + { + "epoch": 4.66, + "learning_rate": 7.828661676287968e-06, + "loss": 0.1356, + "step": 11130 + }, + { + "epoch": 4.66, + "learning_rate": 7.827006747953255e-06, + "loss": 0.1237, + "step": 11131 + }, + { + "epoch": 4.66, + "learning_rate": 7.825351882079305e-06, + "loss": 0.1834, + "step": 11132 + }, + { + "epoch": 4.66, + "learning_rate": 7.823697078713689e-06, + "loss": 0.1415, + "step": 11133 + }, + { + "epoch": 4.66, + "learning_rate": 7.822042337903964e-06, + "loss": 0.1404, + "step": 11134 + }, + { + "epoch": 4.66, + "learning_rate": 7.820387659697701e-06, + "loss": 0.1441, + "step": 11135 + }, + { + "epoch": 4.66, + "learning_rate": 7.818733044142462e-06, + "loss": 0.1424, + "step": 11136 + }, + { + "epoch": 4.66, + "learning_rate": 7.817078491285806e-06, + "loss": 0.1405, + "step": 11137 + }, + { + "epoch": 4.66, + "learning_rate": 7.815424001175289e-06, + "loss": 0.1463, + "step": 11138 + }, + { + "epoch": 4.66, + "learning_rate": 7.813769573858474e-06, + "loss": 0.1954, + "step": 11139 + }, + { + "epoch": 4.66, + "learning_rate": 7.81211520938291e-06, + "loss": 0.1358, + "step": 11140 + }, + { + "epoch": 4.66, + "learning_rate": 7.810460907796153e-06, + "loss": 0.1541, + "step": 11141 + }, + { + "epoch": 4.66, + "learning_rate": 7.80880666914576e-06, + "loss": 0.1566, + "step": 11142 + }, + { + "epoch": 4.67, + "learning_rate": 7.807152493479268e-06, + "loss": 0.1288, + "step": 11143 + }, + { + "epoch": 4.67, + "learning_rate": 7.805498380844236e-06, + "loss": 0.1411, + "step": 11144 + }, + { + "epoch": 4.67, + "learning_rate": 7.803844331288201e-06, + "loss": 0.1424, + "step": 11145 + }, + { + "epoch": 4.67, + "learning_rate": 7.802190344858715e-06, + "loss": 0.1793, + "step": 11146 + }, + { + "epoch": 4.67, + "learning_rate": 7.800536421603317e-06, + "loss": 0.1316, + "step": 11147 + }, + { + "epoch": 4.67, + "learning_rate": 7.798882561569546e-06, + "loss": 0.1403, + "step": 11148 + }, + { + "epoch": 4.67, + "learning_rate": 7.797228764804942e-06, + "loss": 0.1246, + "step": 11149 + }, + { + "epoch": 4.67, + "learning_rate": 7.795575031357047e-06, + "loss": 0.1353, + "step": 11150 + }, + { + "epoch": 4.67, + "learning_rate": 7.793921361273388e-06, + "loss": 0.183, + "step": 11151 + }, + { + "epoch": 4.67, + "learning_rate": 7.792267754601501e-06, + "loss": 0.1039, + "step": 11152 + }, + { + "epoch": 4.67, + "learning_rate": 7.79061421138892e-06, + "loss": 0.1485, + "step": 11153 + }, + { + "epoch": 4.67, + "learning_rate": 7.788960731683174e-06, + "loss": 0.1523, + "step": 11154 + }, + { + "epoch": 4.67, + "learning_rate": 7.787307315531787e-06, + "loss": 0.1523, + "step": 11155 + }, + { + "epoch": 4.67, + "learning_rate": 7.785653962982288e-06, + "loss": 0.1329, + "step": 11156 + }, + { + "epoch": 4.67, + "learning_rate": 7.784000674082205e-06, + "loss": 0.1695, + "step": 11157 + }, + { + "epoch": 4.67, + "learning_rate": 7.782347448879051e-06, + "loss": 0.1421, + "step": 11158 + }, + { + "epoch": 4.67, + "learning_rate": 7.780694287420355e-06, + "loss": 0.1418, + "step": 11159 + }, + { + "epoch": 4.67, + "learning_rate": 7.77904118975363e-06, + "loss": 0.1341, + "step": 11160 + }, + { + "epoch": 4.67, + "learning_rate": 7.777388155926397e-06, + "loss": 0.16, + "step": 11161 + }, + { + "epoch": 4.67, + "learning_rate": 7.775735185986168e-06, + "loss": 0.1736, + "step": 11162 + }, + { + "epoch": 4.67, + "learning_rate": 7.774082279980459e-06, + "loss": 0.1503, + "step": 11163 + }, + { + "epoch": 4.67, + "learning_rate": 7.772429437956777e-06, + "loss": 0.1389, + "step": 11164 + }, + { + "epoch": 4.67, + "learning_rate": 7.77077665996264e-06, + "loss": 0.1582, + "step": 11165 + }, + { + "epoch": 4.67, + "learning_rate": 7.769123946045545e-06, + "loss": 0.1867, + "step": 11166 + }, + { + "epoch": 4.68, + "learning_rate": 7.767471296253003e-06, + "loss": 0.1555, + "step": 11167 + }, + { + "epoch": 4.68, + "learning_rate": 7.765818710632516e-06, + "loss": 0.1818, + "step": 11168 + }, + { + "epoch": 4.68, + "learning_rate": 7.764166189231588e-06, + "loss": 0.1169, + "step": 11169 + }, + { + "epoch": 4.68, + "learning_rate": 7.762513732097721e-06, + "loss": 0.1792, + "step": 11170 + }, + { + "epoch": 4.68, + "learning_rate": 7.76086133927841e-06, + "loss": 0.1554, + "step": 11171 + }, + { + "epoch": 4.68, + "learning_rate": 7.759209010821152e-06, + "loss": 0.1676, + "step": 11172 + }, + { + "epoch": 4.68, + "learning_rate": 7.757556746773443e-06, + "loss": 0.1391, + "step": 11173 + }, + { + "epoch": 4.68, + "learning_rate": 7.755904547182778e-06, + "loss": 0.1281, + "step": 11174 + }, + { + "epoch": 4.68, + "learning_rate": 7.754252412096642e-06, + "loss": 0.1502, + "step": 11175 + }, + { + "epoch": 4.68, + "learning_rate": 7.752600341562527e-06, + "loss": 0.152, + "step": 11176 + }, + { + "epoch": 4.68, + "learning_rate": 7.750948335627925e-06, + "loss": 0.1148, + "step": 11177 + }, + { + "epoch": 4.68, + "learning_rate": 7.749296394340313e-06, + "loss": 0.1783, + "step": 11178 + }, + { + "epoch": 4.68, + "learning_rate": 7.747644517747182e-06, + "loss": 0.1686, + "step": 11179 + }, + { + "epoch": 4.68, + "learning_rate": 7.74599270589601e-06, + "loss": 0.1007, + "step": 11180 + }, + { + "epoch": 4.68, + "learning_rate": 7.744340958834275e-06, + "loss": 0.1236, + "step": 11181 + }, + { + "epoch": 4.68, + "learning_rate": 7.742689276609459e-06, + "loss": 0.1305, + "step": 11182 + }, + { + "epoch": 4.68, + "learning_rate": 7.741037659269037e-06, + "loss": 0.1455, + "step": 11183 + }, + { + "epoch": 4.68, + "learning_rate": 7.739386106860485e-06, + "loss": 0.1433, + "step": 11184 + }, + { + "epoch": 4.68, + "learning_rate": 7.73773461943127e-06, + "loss": 0.1277, + "step": 11185 + }, + { + "epoch": 4.68, + "learning_rate": 7.736083197028866e-06, + "loss": 0.2145, + "step": 11186 + }, + { + "epoch": 4.68, + "learning_rate": 7.734431839700747e-06, + "loss": 0.1557, + "step": 11187 + }, + { + "epoch": 4.68, + "learning_rate": 7.732780547494372e-06, + "loss": 0.1628, + "step": 11188 + }, + { + "epoch": 4.68, + "learning_rate": 7.731129320457205e-06, + "loss": 0.1719, + "step": 11189 + }, + { + "epoch": 4.68, + "learning_rate": 7.729478158636712e-06, + "loss": 0.1289, + "step": 11190 + }, + { + "epoch": 4.69, + "learning_rate": 7.727827062080359e-06, + "loss": 0.1182, + "step": 11191 + }, + { + "epoch": 4.69, + "learning_rate": 7.726176030835597e-06, + "loss": 0.1314, + "step": 11192 + }, + { + "epoch": 4.69, + "learning_rate": 7.724525064949887e-06, + "loss": 0.1119, + "step": 11193 + }, + { + "epoch": 4.69, + "learning_rate": 7.722874164470685e-06, + "loss": 0.1527, + "step": 11194 + }, + { + "epoch": 4.69, + "learning_rate": 7.721223329445448e-06, + "loss": 0.1525, + "step": 11195 + }, + { + "epoch": 4.69, + "learning_rate": 7.719572559921619e-06, + "loss": 0.177, + "step": 11196 + }, + { + "epoch": 4.69, + "learning_rate": 7.717921855946656e-06, + "loss": 0.1636, + "step": 11197 + }, + { + "epoch": 4.69, + "learning_rate": 7.716271217568005e-06, + "loss": 0.1738, + "step": 11198 + }, + { + "epoch": 4.69, + "learning_rate": 7.71462064483311e-06, + "loss": 0.1189, + "step": 11199 + }, + { + "epoch": 4.69, + "learning_rate": 7.712970137789416e-06, + "loss": 0.1319, + "step": 11200 + }, + { + "epoch": 4.69, + "learning_rate": 7.711319696484368e-06, + "loss": 0.1306, + "step": 11201 + }, + { + "epoch": 4.69, + "learning_rate": 7.709669320965402e-06, + "loss": 0.1446, + "step": 11202 + }, + { + "epoch": 4.69, + "learning_rate": 7.70801901127996e-06, + "loss": 0.1587, + "step": 11203 + }, + { + "epoch": 4.69, + "learning_rate": 7.70636876747548e-06, + "loss": 0.1619, + "step": 11204 + }, + { + "epoch": 4.69, + "learning_rate": 7.70471858959939e-06, + "loss": 0.1482, + "step": 11205 + }, + { + "epoch": 4.69, + "learning_rate": 7.70306847769913e-06, + "loss": 0.1366, + "step": 11206 + }, + { + "epoch": 4.69, + "learning_rate": 7.70141843182213e-06, + "loss": 0.1257, + "step": 11207 + }, + { + "epoch": 4.69, + "learning_rate": 7.699768452015815e-06, + "loss": 0.151, + "step": 11208 + }, + { + "epoch": 4.69, + "learning_rate": 7.698118538327613e-06, + "loss": 0.1567, + "step": 11209 + }, + { + "epoch": 4.69, + "learning_rate": 7.696468690804956e-06, + "loss": 0.1596, + "step": 11210 + }, + { + "epoch": 4.69, + "learning_rate": 7.69481890949526e-06, + "loss": 0.1584, + "step": 11211 + }, + { + "epoch": 4.69, + "learning_rate": 7.69316919444595e-06, + "loss": 0.1776, + "step": 11212 + }, + { + "epoch": 4.69, + "learning_rate": 7.691519545704442e-06, + "loss": 0.1568, + "step": 11213 + }, + { + "epoch": 4.69, + "learning_rate": 7.689869963318154e-06, + "loss": 0.1598, + "step": 11214 + }, + { + "epoch": 4.7, + "learning_rate": 7.688220447334508e-06, + "loss": 0.1739, + "step": 11215 + }, + { + "epoch": 4.7, + "learning_rate": 7.686570997800914e-06, + "loss": 0.1726, + "step": 11216 + }, + { + "epoch": 4.7, + "learning_rate": 7.68492161476478e-06, + "loss": 0.1606, + "step": 11217 + }, + { + "epoch": 4.7, + "learning_rate": 7.683272298273525e-06, + "loss": 0.1754, + "step": 11218 + }, + { + "epoch": 4.7, + "learning_rate": 7.681623048374549e-06, + "loss": 0.1424, + "step": 11219 + }, + { + "epoch": 4.7, + "learning_rate": 7.67997386511526e-06, + "loss": 0.2035, + "step": 11220 + }, + { + "epoch": 4.7, + "learning_rate": 7.678324748543066e-06, + "loss": 0.2215, + "step": 11221 + }, + { + "epoch": 4.7, + "learning_rate": 7.676675698705364e-06, + "loss": 0.1417, + "step": 11222 + }, + { + "epoch": 4.7, + "learning_rate": 7.67502671564956e-06, + "loss": 0.1331, + "step": 11223 + }, + { + "epoch": 4.7, + "learning_rate": 7.673377799423047e-06, + "loss": 0.1634, + "step": 11224 + }, + { + "epoch": 4.7, + "learning_rate": 7.671728950073228e-06, + "loss": 0.1274, + "step": 11225 + }, + { + "epoch": 4.7, + "learning_rate": 7.670080167647492e-06, + "loss": 0.1679, + "step": 11226 + }, + { + "epoch": 4.7, + "learning_rate": 7.668431452193234e-06, + "loss": 0.1328, + "step": 11227 + }, + { + "epoch": 4.7, + "learning_rate": 7.666782803757848e-06, + "loss": 0.1429, + "step": 11228 + }, + { + "epoch": 4.7, + "learning_rate": 7.665134222388717e-06, + "loss": 0.1673, + "step": 11229 + }, + { + "epoch": 4.7, + "learning_rate": 7.66348570813323e-06, + "loss": 0.1549, + "step": 11230 + }, + { + "epoch": 4.7, + "learning_rate": 7.661837261038776e-06, + "loss": 0.1568, + "step": 11231 + }, + { + "epoch": 4.7, + "learning_rate": 7.660188881152732e-06, + "loss": 0.1671, + "step": 11232 + }, + { + "epoch": 4.7, + "learning_rate": 7.658540568522487e-06, + "loss": 0.1755, + "step": 11233 + }, + { + "epoch": 4.7, + "learning_rate": 7.656892323195414e-06, + "loss": 0.1713, + "step": 11234 + }, + { + "epoch": 4.7, + "learning_rate": 7.655244145218887e-06, + "loss": 0.1667, + "step": 11235 + }, + { + "epoch": 4.7, + "learning_rate": 7.653596034640291e-06, + "loss": 0.166, + "step": 11236 + }, + { + "epoch": 4.7, + "learning_rate": 7.651947991506994e-06, + "loss": 0.1592, + "step": 11237 + }, + { + "epoch": 4.71, + "learning_rate": 7.650300015866367e-06, + "loss": 0.1462, + "step": 11238 + }, + { + "epoch": 4.71, + "learning_rate": 7.648652107765783e-06, + "loss": 0.1501, + "step": 11239 + }, + { + "epoch": 4.71, + "learning_rate": 7.647004267252606e-06, + "loss": 0.1555, + "step": 11240 + }, + { + "epoch": 4.71, + "learning_rate": 7.645356494374203e-06, + "loss": 0.1119, + "step": 11241 + }, + { + "epoch": 4.71, + "learning_rate": 7.643708789177939e-06, + "loss": 0.1433, + "step": 11242 + }, + { + "epoch": 4.71, + "learning_rate": 7.642061151711176e-06, + "loss": 0.1535, + "step": 11243 + }, + { + "epoch": 4.71, + "learning_rate": 7.64041358202127e-06, + "loss": 0.1613, + "step": 11244 + }, + { + "epoch": 4.71, + "learning_rate": 7.638766080155584e-06, + "loss": 0.1381, + "step": 11245 + }, + { + "epoch": 4.71, + "learning_rate": 7.63711864616147e-06, + "loss": 0.1383, + "step": 11246 + }, + { + "epoch": 4.71, + "learning_rate": 7.635471280086285e-06, + "loss": 0.1832, + "step": 11247 + }, + { + "epoch": 4.71, + "learning_rate": 7.63382398197738e-06, + "loss": 0.1316, + "step": 11248 + }, + { + "epoch": 4.71, + "learning_rate": 7.632176751882103e-06, + "loss": 0.1394, + "step": 11249 + }, + { + "epoch": 4.71, + "learning_rate": 7.630529589847807e-06, + "loss": 0.1282, + "step": 11250 + }, + { + "epoch": 4.71, + "learning_rate": 7.628882495921834e-06, + "loss": 0.1677, + "step": 11251 + }, + { + "epoch": 4.71, + "learning_rate": 7.627235470151528e-06, + "loss": 0.2038, + "step": 11252 + }, + { + "epoch": 4.71, + "learning_rate": 7.625588512584234e-06, + "loss": 0.1632, + "step": 11253 + }, + { + "epoch": 4.71, + "learning_rate": 7.6239416232672904e-06, + "loss": 0.1553, + "step": 11254 + }, + { + "epoch": 4.71, + "learning_rate": 7.622294802248041e-06, + "loss": 0.1566, + "step": 11255 + }, + { + "epoch": 4.71, + "learning_rate": 7.620648049573815e-06, + "loss": 0.1448, + "step": 11256 + }, + { + "epoch": 4.71, + "learning_rate": 7.619001365291948e-06, + "loss": 0.141, + "step": 11257 + }, + { + "epoch": 4.71, + "learning_rate": 7.617354749449774e-06, + "loss": 0.1419, + "step": 11258 + }, + { + "epoch": 4.71, + "learning_rate": 7.615708202094625e-06, + "loss": 0.1363, + "step": 11259 + }, + { + "epoch": 4.71, + "learning_rate": 7.614061723273825e-06, + "loss": 0.1554, + "step": 11260 + }, + { + "epoch": 4.71, + "learning_rate": 7.612415313034706e-06, + "loss": 0.1467, + "step": 11261 + }, + { + "epoch": 4.72, + "learning_rate": 7.61076897142459e-06, + "loss": 0.1473, + "step": 11262 + }, + { + "epoch": 4.72, + "learning_rate": 7.6091226984907985e-06, + "loss": 0.1502, + "step": 11263 + }, + { + "epoch": 4.72, + "learning_rate": 7.607476494280653e-06, + "loss": 0.1061, + "step": 11264 + }, + { + "epoch": 4.72, + "learning_rate": 7.6058303588414725e-06, + "loss": 0.2, + "step": 11265 + }, + { + "epoch": 4.72, + "learning_rate": 7.604184292220577e-06, + "loss": 0.1315, + "step": 11266 + }, + { + "epoch": 4.72, + "learning_rate": 7.602538294465273e-06, + "loss": 0.1253, + "step": 11267 + }, + { + "epoch": 4.72, + "learning_rate": 7.6008923656228814e-06, + "loss": 0.1226, + "step": 11268 + }, + { + "epoch": 4.72, + "learning_rate": 7.599246505740711e-06, + "loss": 0.1397, + "step": 11269 + }, + { + "epoch": 4.72, + "learning_rate": 7.597600714866067e-06, + "loss": 0.1627, + "step": 11270 + }, + { + "epoch": 4.72, + "learning_rate": 7.595954993046259e-06, + "loss": 0.1419, + "step": 11271 + }, + { + "epoch": 4.72, + "learning_rate": 7.594309340328594e-06, + "loss": 0.1498, + "step": 11272 + }, + { + "epoch": 4.72, + "learning_rate": 7.592663756760369e-06, + "loss": 0.1064, + "step": 11273 + }, + { + "epoch": 4.72, + "learning_rate": 7.591018242388888e-06, + "loss": 0.1151, + "step": 11274 + }, + { + "epoch": 4.72, + "learning_rate": 7.589372797261453e-06, + "loss": 0.1374, + "step": 11275 + }, + { + "epoch": 4.72, + "learning_rate": 7.587727421425355e-06, + "loss": 0.1407, + "step": 11276 + }, + { + "epoch": 4.72, + "learning_rate": 7.586082114927892e-06, + "loss": 0.1609, + "step": 11277 + }, + { + "epoch": 4.72, + "learning_rate": 7.58443687781636e-06, + "loss": 0.1706, + "step": 11278 + }, + { + "epoch": 4.72, + "learning_rate": 7.582791710138043e-06, + "loss": 0.1377, + "step": 11279 + }, + { + "epoch": 4.72, + "learning_rate": 7.581146611940235e-06, + "loss": 0.1213, + "step": 11280 + }, + { + "epoch": 4.72, + "learning_rate": 7.57950158327022e-06, + "loss": 0.1416, + "step": 11281 + }, + { + "epoch": 4.72, + "learning_rate": 7.577856624175283e-06, + "loss": 0.1573, + "step": 11282 + }, + { + "epoch": 4.72, + "learning_rate": 7.576211734702711e-06, + "loss": 0.155, + "step": 11283 + }, + { + "epoch": 4.72, + "learning_rate": 7.574566914899779e-06, + "loss": 0.0967, + "step": 11284 + }, + { + "epoch": 4.72, + "learning_rate": 7.5729221648137686e-06, + "loss": 0.1409, + "step": 11285 + }, + { + "epoch": 4.73, + "learning_rate": 7.571277484491959e-06, + "loss": 0.1336, + "step": 11286 + }, + { + "epoch": 4.73, + "learning_rate": 7.569632873981622e-06, + "loss": 0.1512, + "step": 11287 + }, + { + "epoch": 4.73, + "learning_rate": 7.56798833333003e-06, + "loss": 0.1534, + "step": 11288 + }, + { + "epoch": 4.73, + "learning_rate": 7.566343862584458e-06, + "loss": 0.1724, + "step": 11289 + }, + { + "epoch": 4.73, + "learning_rate": 7.56469946179217e-06, + "loss": 0.129, + "step": 11290 + }, + { + "epoch": 4.73, + "learning_rate": 7.563055131000434e-06, + "loss": 0.1654, + "step": 11291 + }, + { + "epoch": 4.73, + "learning_rate": 7.561410870256518e-06, + "loss": 0.1414, + "step": 11292 + }, + { + "epoch": 4.73, + "learning_rate": 7.559766679607682e-06, + "loss": 0.1662, + "step": 11293 + }, + { + "epoch": 4.73, + "learning_rate": 7.5581225591011864e-06, + "loss": 0.1258, + "step": 11294 + }, + { + "epoch": 4.73, + "learning_rate": 7.556478508784291e-06, + "loss": 0.1049, + "step": 11295 + }, + { + "epoch": 4.73, + "learning_rate": 7.554834528704254e-06, + "loss": 0.1295, + "step": 11296 + }, + { + "epoch": 4.73, + "learning_rate": 7.553190618908327e-06, + "loss": 0.0941, + "step": 11297 + }, + { + "epoch": 4.73, + "learning_rate": 7.551546779443765e-06, + "loss": 0.0995, + "step": 11298 + }, + { + "epoch": 4.73, + "learning_rate": 7.54990301035782e-06, + "loss": 0.1157, + "step": 11299 + }, + { + "epoch": 4.73, + "learning_rate": 7.5482593116977355e-06, + "loss": 0.1149, + "step": 11300 + }, + { + "epoch": 4.73, + "learning_rate": 7.546615683510767e-06, + "loss": 0.1697, + "step": 11301 + }, + { + "epoch": 4.73, + "learning_rate": 7.544972125844149e-06, + "loss": 0.1343, + "step": 11302 + }, + { + "epoch": 4.73, + "learning_rate": 7.543328638745131e-06, + "loss": 0.099, + "step": 11303 + }, + { + "epoch": 4.73, + "learning_rate": 7.541685222260948e-06, + "loss": 0.1647, + "step": 11304 + }, + { + "epoch": 4.73, + "learning_rate": 7.540041876438842e-06, + "loss": 0.1305, + "step": 11305 + }, + { + "epoch": 4.73, + "learning_rate": 7.538398601326051e-06, + "loss": 0.1737, + "step": 11306 + }, + { + "epoch": 4.73, + "learning_rate": 7.5367553969698095e-06, + "loss": 0.1746, + "step": 11307 + }, + { + "epoch": 4.73, + "learning_rate": 7.535112263417346e-06, + "loss": 0.1647, + "step": 11308 + }, + { + "epoch": 4.73, + "learning_rate": 7.533469200715892e-06, + "loss": 0.1599, + "step": 11309 + }, + { + "epoch": 4.74, + "learning_rate": 7.53182620891268e-06, + "loss": 0.1466, + "step": 11310 + }, + { + "epoch": 4.74, + "learning_rate": 7.530183288054931e-06, + "loss": 0.1535, + "step": 11311 + }, + { + "epoch": 4.74, + "learning_rate": 7.528540438189872e-06, + "loss": 0.1489, + "step": 11312 + }, + { + "epoch": 4.74, + "learning_rate": 7.526897659364727e-06, + "loss": 0.1699, + "step": 11313 + }, + { + "epoch": 4.74, + "learning_rate": 7.525254951626712e-06, + "loss": 0.106, + "step": 11314 + }, + { + "epoch": 4.74, + "learning_rate": 7.523612315023048e-06, + "loss": 0.1492, + "step": 11315 + }, + { + "epoch": 4.74, + "learning_rate": 7.5219697496009525e-06, + "loss": 0.1654, + "step": 11316 + }, + { + "epoch": 4.74, + "learning_rate": 7.520327255407635e-06, + "loss": 0.1983, + "step": 11317 + }, + { + "epoch": 4.74, + "learning_rate": 7.518684832490311e-06, + "loss": 0.167, + "step": 11318 + }, + { + "epoch": 4.74, + "learning_rate": 7.51704248089619e-06, + "loss": 0.1525, + "step": 11319 + }, + { + "epoch": 4.74, + "learning_rate": 7.515400200672478e-06, + "loss": 0.1334, + "step": 11320 + }, + { + "epoch": 4.74, + "learning_rate": 7.513757991866383e-06, + "loss": 0.1492, + "step": 11321 + }, + { + "epoch": 4.74, + "learning_rate": 7.5121158545251085e-06, + "loss": 0.1698, + "step": 11322 + }, + { + "epoch": 4.74, + "learning_rate": 7.51047378869586e-06, + "loss": 0.1589, + "step": 11323 + }, + { + "epoch": 4.74, + "learning_rate": 7.5088317944258305e-06, + "loss": 0.1427, + "step": 11324 + }, + { + "epoch": 4.74, + "learning_rate": 7.50718987176222e-06, + "loss": 0.1504, + "step": 11325 + }, + { + "epoch": 4.74, + "learning_rate": 7.5055480207522225e-06, + "loss": 0.1366, + "step": 11326 + }, + { + "epoch": 4.74, + "learning_rate": 7.503906241443037e-06, + "loss": 0.1122, + "step": 11327 + }, + { + "epoch": 4.74, + "learning_rate": 7.502264533881849e-06, + "loss": 0.1707, + "step": 11328 + }, + { + "epoch": 4.74, + "learning_rate": 7.50062289811585e-06, + "loss": 0.1653, + "step": 11329 + }, + { + "epoch": 4.74, + "learning_rate": 7.49898133419223e-06, + "loss": 0.1242, + "step": 11330 + }, + { + "epoch": 4.74, + "learning_rate": 7.497339842158171e-06, + "loss": 0.1779, + "step": 11331 + }, + { + "epoch": 4.74, + "learning_rate": 7.495698422060857e-06, + "loss": 0.1762, + "step": 11332 + }, + { + "epoch": 4.74, + "learning_rate": 7.494057073947473e-06, + "loss": 0.1771, + "step": 11333 + }, + { + "epoch": 4.75, + "learning_rate": 7.492415797865191e-06, + "loss": 0.1462, + "step": 11334 + }, + { + "epoch": 4.75, + "learning_rate": 7.490774593861192e-06, + "loss": 0.1265, + "step": 11335 + }, + { + "epoch": 4.75, + "learning_rate": 7.489133461982651e-06, + "loss": 0.1422, + "step": 11336 + }, + { + "epoch": 4.75, + "learning_rate": 7.487492402276742e-06, + "loss": 0.1407, + "step": 11337 + }, + { + "epoch": 4.75, + "learning_rate": 7.485851414790632e-06, + "loss": 0.1442, + "step": 11338 + }, + { + "epoch": 4.75, + "learning_rate": 7.4842104995714935e-06, + "loss": 0.2042, + "step": 11339 + }, + { + "epoch": 4.75, + "learning_rate": 7.482569656666493e-06, + "loss": 0.1364, + "step": 11340 + }, + { + "epoch": 4.75, + "learning_rate": 7.480928886122792e-06, + "loss": 0.1347, + "step": 11341 + }, + { + "epoch": 4.75, + "learning_rate": 7.479288187987554e-06, + "loss": 0.1625, + "step": 11342 + }, + { + "epoch": 4.75, + "learning_rate": 7.477647562307943e-06, + "loss": 0.147, + "step": 11343 + }, + { + "epoch": 4.75, + "learning_rate": 7.476007009131112e-06, + "loss": 0.1096, + "step": 11344 + }, + { + "epoch": 4.75, + "learning_rate": 7.47436652850422e-06, + "loss": 0.1525, + "step": 11345 + }, + { + "epoch": 4.75, + "learning_rate": 7.472726120474425e-06, + "loss": 0.1605, + "step": 11346 + }, + { + "epoch": 4.75, + "learning_rate": 7.471085785088871e-06, + "loss": 0.1622, + "step": 11347 + }, + { + "epoch": 4.75, + "learning_rate": 7.469445522394714e-06, + "loss": 0.16, + "step": 11348 + }, + { + "epoch": 4.75, + "learning_rate": 7.467805332439098e-06, + "loss": 0.1244, + "step": 11349 + }, + { + "epoch": 4.75, + "learning_rate": 7.46616521526917e-06, + "loss": 0.1675, + "step": 11350 + }, + { + "epoch": 4.75, + "learning_rate": 7.4645251709320775e-06, + "loss": 0.1349, + "step": 11351 + }, + { + "epoch": 4.75, + "learning_rate": 7.462885199474956e-06, + "loss": 0.164, + "step": 11352 + }, + { + "epoch": 4.75, + "learning_rate": 7.4612453009449485e-06, + "loss": 0.1653, + "step": 11353 + }, + { + "epoch": 4.75, + "learning_rate": 7.459605475389193e-06, + "loss": 0.1166, + "step": 11354 + }, + { + "epoch": 4.75, + "learning_rate": 7.457965722854823e-06, + "loss": 0.1068, + "step": 11355 + }, + { + "epoch": 4.75, + "learning_rate": 7.4563260433889726e-06, + "loss": 0.1797, + "step": 11356 + }, + { + "epoch": 4.75, + "learning_rate": 7.454686437038775e-06, + "loss": 0.1814, + "step": 11357 + }, + { + "epoch": 4.76, + "learning_rate": 7.4530469038513545e-06, + "loss": 0.1696, + "step": 11358 + }, + { + "epoch": 4.76, + "learning_rate": 7.45140744387384e-06, + "loss": 0.1331, + "step": 11359 + }, + { + "epoch": 4.76, + "learning_rate": 7.44976805715336e-06, + "loss": 0.1058, + "step": 11360 + }, + { + "epoch": 4.76, + "learning_rate": 7.448128743737032e-06, + "loss": 0.1342, + "step": 11361 + }, + { + "epoch": 4.76, + "learning_rate": 7.446489503671979e-06, + "loss": 0.1599, + "step": 11362 + }, + { + "epoch": 4.76, + "learning_rate": 7.4448503370053204e-06, + "loss": 0.1291, + "step": 11363 + }, + { + "epoch": 4.76, + "learning_rate": 7.44321124378417e-06, + "loss": 0.1425, + "step": 11364 + }, + { + "epoch": 4.76, + "learning_rate": 7.441572224055644e-06, + "loss": 0.186, + "step": 11365 + }, + { + "epoch": 4.76, + "learning_rate": 7.4399332778668535e-06, + "loss": 0.1956, + "step": 11366 + }, + { + "epoch": 4.76, + "learning_rate": 7.438294405264912e-06, + "loss": 0.2012, + "step": 11367 + }, + { + "epoch": 4.76, + "learning_rate": 7.436655606296921e-06, + "loss": 0.1178, + "step": 11368 + }, + { + "epoch": 4.76, + "learning_rate": 7.4350168810099955e-06, + "loss": 0.1288, + "step": 11369 + }, + { + "epoch": 4.76, + "learning_rate": 7.433378229451229e-06, + "loss": 0.1542, + "step": 11370 + }, + { + "epoch": 4.76, + "learning_rate": 7.431739651667729e-06, + "loss": 0.1506, + "step": 11371 + }, + { + "epoch": 4.76, + "learning_rate": 7.430101147706592e-06, + "loss": 0.2029, + "step": 11372 + }, + { + "epoch": 4.76, + "learning_rate": 7.428462717614917e-06, + "loss": 0.1316, + "step": 11373 + }, + { + "epoch": 4.76, + "learning_rate": 7.426824361439803e-06, + "loss": 0.1284, + "step": 11374 + }, + { + "epoch": 4.76, + "learning_rate": 7.425186079228335e-06, + "loss": 0.1285, + "step": 11375 + }, + { + "epoch": 4.76, + "learning_rate": 7.42354787102761e-06, + "loss": 0.1689, + "step": 11376 + }, + { + "epoch": 4.76, + "learning_rate": 7.421909736884714e-06, + "loss": 0.1667, + "step": 11377 + }, + { + "epoch": 4.76, + "learning_rate": 7.420271676846737e-06, + "loss": 0.1947, + "step": 11378 + }, + { + "epoch": 4.76, + "learning_rate": 7.4186336909607594e-06, + "loss": 0.1408, + "step": 11379 + }, + { + "epoch": 4.76, + "learning_rate": 7.416995779273866e-06, + "loss": 0.1398, + "step": 11380 + }, + { + "epoch": 4.76, + "learning_rate": 7.415357941833139e-06, + "loss": 0.1497, + "step": 11381 + }, + { + "epoch": 4.77, + "learning_rate": 7.413720178685653e-06, + "loss": 0.1695, + "step": 11382 + }, + { + "epoch": 4.77, + "learning_rate": 7.412082489878484e-06, + "loss": 0.1638, + "step": 11383 + }, + { + "epoch": 4.77, + "learning_rate": 7.41044487545871e-06, + "loss": 0.1852, + "step": 11384 + }, + { + "epoch": 4.77, + "learning_rate": 7.408807335473398e-06, + "loss": 0.1481, + "step": 11385 + }, + { + "epoch": 4.77, + "learning_rate": 7.40716986996962e-06, + "loss": 0.1608, + "step": 11386 + }, + { + "epoch": 4.77, + "learning_rate": 7.4055324789944446e-06, + "loss": 0.1231, + "step": 11387 + }, + { + "epoch": 4.77, + "learning_rate": 7.403895162594934e-06, + "loss": 0.1494, + "step": 11388 + }, + { + "epoch": 4.77, + "learning_rate": 7.402257920818154e-06, + "loss": 0.1702, + "step": 11389 + }, + { + "epoch": 4.77, + "learning_rate": 7.400620753711167e-06, + "loss": 0.1696, + "step": 11390 + }, + { + "epoch": 4.77, + "learning_rate": 7.398983661321029e-06, + "loss": 0.1542, + "step": 11391 + }, + { + "epoch": 4.77, + "learning_rate": 7.3973466436947985e-06, + "loss": 0.1124, + "step": 11392 + }, + { + "epoch": 4.77, + "learning_rate": 7.395709700879525e-06, + "loss": 0.1556, + "step": 11393 + }, + { + "epoch": 4.77, + "learning_rate": 7.394072832922268e-06, + "loss": 0.1859, + "step": 11394 + }, + { + "epoch": 4.77, + "learning_rate": 7.3924360398700745e-06, + "loss": 0.1514, + "step": 11395 + }, + { + "epoch": 4.77, + "learning_rate": 7.390799321769993e-06, + "loss": 0.1577, + "step": 11396 + }, + { + "epoch": 4.77, + "learning_rate": 7.389162678669069e-06, + "loss": 0.1659, + "step": 11397 + }, + { + "epoch": 4.77, + "learning_rate": 7.387526110614347e-06, + "loss": 0.1558, + "step": 11398 + }, + { + "epoch": 4.77, + "learning_rate": 7.385889617652868e-06, + "loss": 0.1791, + "step": 11399 + }, + { + "epoch": 4.77, + "learning_rate": 7.384253199831674e-06, + "loss": 0.1584, + "step": 11400 + }, + { + "epoch": 4.77, + "learning_rate": 7.382616857197801e-06, + "loss": 0.1717, + "step": 11401 + }, + { + "epoch": 4.77, + "learning_rate": 7.380980589798281e-06, + "loss": 0.1607, + "step": 11402 + }, + { + "epoch": 4.77, + "learning_rate": 7.379344397680151e-06, + "loss": 0.1706, + "step": 11403 + }, + { + "epoch": 4.77, + "learning_rate": 7.3777082808904435e-06, + "loss": 0.1739, + "step": 11404 + }, + { + "epoch": 4.77, + "learning_rate": 7.376072239476182e-06, + "loss": 0.1403, + "step": 11405 + }, + { + "epoch": 4.78, + "learning_rate": 7.374436273484395e-06, + "loss": 0.1492, + "step": 11406 + }, + { + "epoch": 4.78, + "learning_rate": 7.372800382962107e-06, + "loss": 0.1787, + "step": 11407 + }, + { + "epoch": 4.78, + "learning_rate": 7.371164567956344e-06, + "loss": 0.155, + "step": 11408 + }, + { + "epoch": 4.78, + "learning_rate": 7.36952882851412e-06, + "loss": 0.1437, + "step": 11409 + }, + { + "epoch": 4.78, + "learning_rate": 7.367893164682456e-06, + "loss": 0.1398, + "step": 11410 + }, + { + "epoch": 4.78, + "learning_rate": 7.366257576508369e-06, + "loss": 0.1301, + "step": 11411 + }, + { + "epoch": 4.78, + "learning_rate": 7.364622064038868e-06, + "loss": 0.1722, + "step": 11412 + }, + { + "epoch": 4.78, + "learning_rate": 7.362986627320967e-06, + "loss": 0.1482, + "step": 11413 + }, + { + "epoch": 4.78, + "learning_rate": 7.361351266401682e-06, + "loss": 0.1344, + "step": 11414 + }, + { + "epoch": 4.78, + "learning_rate": 7.359715981328009e-06, + "loss": 0.1519, + "step": 11415 + }, + { + "epoch": 4.78, + "learning_rate": 7.358080772146956e-06, + "loss": 0.1482, + "step": 11416 + }, + { + "epoch": 4.78, + "learning_rate": 7.356445638905526e-06, + "loss": 0.1776, + "step": 11417 + }, + { + "epoch": 4.78, + "learning_rate": 7.354810581650721e-06, + "loss": 0.1388, + "step": 11418 + }, + { + "epoch": 4.78, + "learning_rate": 7.35317560042954e-06, + "loss": 0.1726, + "step": 11419 + }, + { + "epoch": 4.78, + "learning_rate": 7.351540695288977e-06, + "loss": 0.1506, + "step": 11420 + }, + { + "epoch": 4.78, + "learning_rate": 7.349905866276025e-06, + "loss": 0.1139, + "step": 11421 + }, + { + "epoch": 4.78, + "learning_rate": 7.348271113437679e-06, + "loss": 0.1037, + "step": 11422 + }, + { + "epoch": 4.78, + "learning_rate": 7.346636436820926e-06, + "loss": 0.1677, + "step": 11423 + }, + { + "epoch": 4.78, + "learning_rate": 7.345001836472754e-06, + "loss": 0.1564, + "step": 11424 + }, + { + "epoch": 4.78, + "learning_rate": 7.3433673124401495e-06, + "loss": 0.1574, + "step": 11425 + }, + { + "epoch": 4.78, + "learning_rate": 7.341732864770093e-06, + "loss": 0.1255, + "step": 11426 + }, + { + "epoch": 4.78, + "learning_rate": 7.340098493509565e-06, + "loss": 0.1382, + "step": 11427 + }, + { + "epoch": 4.78, + "learning_rate": 7.338464198705548e-06, + "loss": 0.1807, + "step": 11428 + }, + { + "epoch": 4.79, + "learning_rate": 7.336829980405014e-06, + "loss": 0.1531, + "step": 11429 + }, + { + "epoch": 4.79, + "learning_rate": 7.335195838654939e-06, + "loss": 0.1694, + "step": 11430 + }, + { + "epoch": 4.79, + "learning_rate": 7.333561773502297e-06, + "loss": 0.1758, + "step": 11431 + }, + { + "epoch": 4.79, + "learning_rate": 7.3319277849940525e-06, + "loss": 0.1775, + "step": 11432 + }, + { + "epoch": 4.79, + "learning_rate": 7.330293873177179e-06, + "loss": 0.1232, + "step": 11433 + }, + { + "epoch": 4.79, + "learning_rate": 7.328660038098636e-06, + "loss": 0.153, + "step": 11434 + }, + { + "epoch": 4.79, + "learning_rate": 7.327026279805394e-06, + "loss": 0.1797, + "step": 11435 + }, + { + "epoch": 4.79, + "learning_rate": 7.325392598344406e-06, + "loss": 0.1842, + "step": 11436 + }, + { + "epoch": 4.79, + "learning_rate": 7.3237589937626395e-06, + "loss": 0.1748, + "step": 11437 + }, + { + "epoch": 4.79, + "learning_rate": 7.322125466107042e-06, + "loss": 0.1729, + "step": 11438 + }, + { + "epoch": 4.79, + "learning_rate": 7.320492015424572e-06, + "loss": 0.1727, + "step": 11439 + }, + { + "epoch": 4.79, + "learning_rate": 7.318858641762181e-06, + "loss": 0.1761, + "step": 11440 + }, + { + "epoch": 4.79, + "learning_rate": 7.317225345166819e-06, + "loss": 0.1366, + "step": 11441 + }, + { + "epoch": 4.79, + "learning_rate": 7.315592125685434e-06, + "loss": 0.1397, + "step": 11442 + }, + { + "epoch": 4.79, + "learning_rate": 7.31395898336497e-06, + "loss": 0.107, + "step": 11443 + }, + { + "epoch": 4.79, + "learning_rate": 7.312325918252373e-06, + "loss": 0.1468, + "step": 11444 + }, + { + "epoch": 4.79, + "learning_rate": 7.310692930394583e-06, + "loss": 0.1194, + "step": 11445 + }, + { + "epoch": 4.79, + "learning_rate": 7.309060019838536e-06, + "loss": 0.1522, + "step": 11446 + }, + { + "epoch": 4.79, + "learning_rate": 7.307427186631173e-06, + "loss": 0.0958, + "step": 11447 + }, + { + "epoch": 4.79, + "learning_rate": 7.305794430819424e-06, + "loss": 0.1547, + "step": 11448 + }, + { + "epoch": 4.79, + "learning_rate": 7.304161752450227e-06, + "loss": 0.1457, + "step": 11449 + }, + { + "epoch": 4.79, + "learning_rate": 7.302529151570506e-06, + "loss": 0.1221, + "step": 11450 + }, + { + "epoch": 4.79, + "learning_rate": 7.3008966282271885e-06, + "loss": 0.1433, + "step": 11451 + }, + { + "epoch": 4.79, + "learning_rate": 7.299264182467206e-06, + "loss": 0.1421, + "step": 11452 + }, + { + "epoch": 4.8, + "learning_rate": 7.2976318143374755e-06, + "loss": 0.174, + "step": 11453 + }, + { + "epoch": 4.8, + "learning_rate": 7.295999523884921e-06, + "loss": 0.1891, + "step": 11454 + }, + { + "epoch": 4.8, + "learning_rate": 7.294367311156463e-06, + "loss": 0.1325, + "step": 11455 + }, + { + "epoch": 4.8, + "learning_rate": 7.292735176199013e-06, + "loss": 0.1817, + "step": 11456 + }, + { + "epoch": 4.8, + "learning_rate": 7.291103119059488e-06, + "loss": 0.1543, + "step": 11457 + }, + { + "epoch": 4.8, + "learning_rate": 7.289471139784803e-06, + "loss": 0.1808, + "step": 11458 + }, + { + "epoch": 4.8, + "learning_rate": 7.287839238421863e-06, + "loss": 0.1374, + "step": 11459 + }, + { + "epoch": 4.8, + "learning_rate": 7.286207415017579e-06, + "loss": 0.1189, + "step": 11460 + }, + { + "epoch": 4.8, + "learning_rate": 7.284575669618852e-06, + "loss": 0.1296, + "step": 11461 + }, + { + "epoch": 4.8, + "learning_rate": 7.282944002272587e-06, + "loss": 0.1559, + "step": 11462 + }, + { + "epoch": 4.8, + "learning_rate": 7.28131241302569e-06, + "loss": 0.1616, + "step": 11463 + }, + { + "epoch": 4.8, + "learning_rate": 7.279680901925051e-06, + "loss": 0.1827, + "step": 11464 + }, + { + "epoch": 4.8, + "learning_rate": 7.278049469017569e-06, + "loss": 0.1433, + "step": 11465 + }, + { + "epoch": 4.8, + "learning_rate": 7.276418114350145e-06, + "loss": 0.133, + "step": 11466 + }, + { + "epoch": 4.8, + "learning_rate": 7.274786837969663e-06, + "loss": 0.169, + "step": 11467 + }, + { + "epoch": 4.8, + "learning_rate": 7.273155639923013e-06, + "loss": 0.1406, + "step": 11468 + }, + { + "epoch": 4.8, + "learning_rate": 7.271524520257087e-06, + "loss": 0.1203, + "step": 11469 + }, + { + "epoch": 4.8, + "learning_rate": 7.269893479018766e-06, + "loss": 0.1757, + "step": 11470 + }, + { + "epoch": 4.8, + "learning_rate": 7.268262516254934e-06, + "loss": 0.1237, + "step": 11471 + }, + { + "epoch": 4.8, + "learning_rate": 7.2666316320124755e-06, + "loss": 0.1722, + "step": 11472 + }, + { + "epoch": 4.8, + "learning_rate": 7.265000826338261e-06, + "loss": 0.1364, + "step": 11473 + }, + { + "epoch": 4.8, + "learning_rate": 7.263370099279173e-06, + "loss": 0.1386, + "step": 11474 + }, + { + "epoch": 4.8, + "learning_rate": 7.2617394508820815e-06, + "loss": 0.1343, + "step": 11475 + }, + { + "epoch": 4.8, + "learning_rate": 7.26010888119386e-06, + "loss": 0.1714, + "step": 11476 + }, + { + "epoch": 4.81, + "learning_rate": 7.258478390261377e-06, + "loss": 0.156, + "step": 11477 + }, + { + "epoch": 4.81, + "learning_rate": 7.2568479781315e-06, + "loss": 0.1641, + "step": 11478 + }, + { + "epoch": 4.81, + "learning_rate": 7.255217644851096e-06, + "loss": 0.1244, + "step": 11479 + }, + { + "epoch": 4.81, + "learning_rate": 7.253587390467023e-06, + "loss": 0.1458, + "step": 11480 + }, + { + "epoch": 4.81, + "learning_rate": 7.251957215026142e-06, + "loss": 0.1585, + "step": 11481 + }, + { + "epoch": 4.81, + "learning_rate": 7.250327118575318e-06, + "loss": 0.1757, + "step": 11482 + }, + { + "epoch": 4.81, + "learning_rate": 7.2486971011614e-06, + "loss": 0.1324, + "step": 11483 + }, + { + "epoch": 4.81, + "learning_rate": 7.247067162831237e-06, + "loss": 0.1536, + "step": 11484 + }, + { + "epoch": 4.81, + "learning_rate": 7.245437303631687e-06, + "loss": 0.2021, + "step": 11485 + }, + { + "epoch": 4.81, + "learning_rate": 7.243807523609601e-06, + "loss": 0.1557, + "step": 11486 + }, + { + "epoch": 4.81, + "learning_rate": 7.242177822811818e-06, + "loss": 0.1699, + "step": 11487 + }, + { + "epoch": 4.81, + "learning_rate": 7.240548201285186e-06, + "loss": 0.1632, + "step": 11488 + }, + { + "epoch": 4.81, + "learning_rate": 7.238918659076548e-06, + "loss": 0.1633, + "step": 11489 + }, + { + "epoch": 4.81, + "learning_rate": 7.237289196232744e-06, + "loss": 0.1728, + "step": 11490 + }, + { + "epoch": 4.81, + "learning_rate": 7.235659812800611e-06, + "loss": 0.1672, + "step": 11491 + }, + { + "epoch": 4.81, + "learning_rate": 7.234030508826981e-06, + "loss": 0.1507, + "step": 11492 + }, + { + "epoch": 4.81, + "learning_rate": 7.232401284358692e-06, + "loss": 0.174, + "step": 11493 + }, + { + "epoch": 4.81, + "learning_rate": 7.230772139442572e-06, + "loss": 0.0923, + "step": 11494 + }, + { + "epoch": 4.81, + "learning_rate": 7.229143074125447e-06, + "loss": 0.1453, + "step": 11495 + }, + { + "epoch": 4.81, + "learning_rate": 7.2275140884541495e-06, + "loss": 0.1484, + "step": 11496 + }, + { + "epoch": 4.81, + "learning_rate": 7.225885182475496e-06, + "loss": 0.138, + "step": 11497 + }, + { + "epoch": 4.81, + "learning_rate": 7.224256356236312e-06, + "loss": 0.1439, + "step": 11498 + }, + { + "epoch": 4.81, + "learning_rate": 7.222627609783418e-06, + "loss": 0.1385, + "step": 11499 + }, + { + "epoch": 4.81, + "learning_rate": 7.2209989431636265e-06, + "loss": 0.1192, + "step": 11500 + }, + { + "epoch": 4.82, + "learning_rate": 7.219370356423754e-06, + "loss": 0.1667, + "step": 11501 + }, + { + "epoch": 4.82, + "learning_rate": 7.217741849610616e-06, + "loss": 0.1536, + "step": 11502 + }, + { + "epoch": 4.82, + "learning_rate": 7.216113422771016e-06, + "loss": 0.142, + "step": 11503 + }, + { + "epoch": 4.82, + "learning_rate": 7.214485075951771e-06, + "loss": 0.1855, + "step": 11504 + }, + { + "epoch": 4.82, + "learning_rate": 7.212856809199676e-06, + "loss": 0.1364, + "step": 11505 + }, + { + "epoch": 4.82, + "learning_rate": 7.211228622561538e-06, + "loss": 0.1593, + "step": 11506 + }, + { + "epoch": 4.82, + "learning_rate": 7.2096005160841605e-06, + "loss": 0.153, + "step": 11507 + }, + { + "epoch": 4.82, + "learning_rate": 7.207972489814339e-06, + "loss": 0.1313, + "step": 11508 + }, + { + "epoch": 4.82, + "learning_rate": 7.206344543798871e-06, + "loss": 0.1861, + "step": 11509 + }, + { + "epoch": 4.82, + "learning_rate": 7.204716678084551e-06, + "loss": 0.17, + "step": 11510 + }, + { + "epoch": 4.82, + "learning_rate": 7.203088892718168e-06, + "loss": 0.1709, + "step": 11511 + }, + { + "epoch": 4.82, + "learning_rate": 7.201461187746513e-06, + "loss": 0.0981, + "step": 11512 + }, + { + "epoch": 4.82, + "learning_rate": 7.199833563216375e-06, + "loss": 0.1039, + "step": 11513 + }, + { + "epoch": 4.82, + "learning_rate": 7.198206019174534e-06, + "loss": 0.176, + "step": 11514 + }, + { + "epoch": 4.82, + "learning_rate": 7.196578555667775e-06, + "loss": 0.1032, + "step": 11515 + }, + { + "epoch": 4.82, + "learning_rate": 7.194951172742879e-06, + "loss": 0.1818, + "step": 11516 + }, + { + "epoch": 4.82, + "learning_rate": 7.19332387044662e-06, + "loss": 0.1415, + "step": 11517 + }, + { + "epoch": 4.82, + "learning_rate": 7.191696648825778e-06, + "loss": 0.1941, + "step": 11518 + }, + { + "epoch": 4.82, + "learning_rate": 7.190069507927122e-06, + "loss": 0.1256, + "step": 11519 + }, + { + "epoch": 4.82, + "learning_rate": 7.1884424477974265e-06, + "loss": 0.1336, + "step": 11520 + }, + { + "epoch": 4.82, + "learning_rate": 7.186815468483457e-06, + "loss": 0.1809, + "step": 11521 + }, + { + "epoch": 4.82, + "learning_rate": 7.185188570031979e-06, + "loss": 0.1438, + "step": 11522 + }, + { + "epoch": 4.82, + "learning_rate": 7.183561752489761e-06, + "loss": 0.1717, + "step": 11523 + }, + { + "epoch": 4.82, + "learning_rate": 7.181935015903558e-06, + "loss": 0.1621, + "step": 11524 + }, + { + "epoch": 4.83, + "learning_rate": 7.180308360320134e-06, + "loss": 0.1906, + "step": 11525 + }, + { + "epoch": 4.83, + "learning_rate": 7.178681785786245e-06, + "loss": 0.1621, + "step": 11526 + }, + { + "epoch": 4.83, + "learning_rate": 7.177055292348646e-06, + "loss": 0.1375, + "step": 11527 + }, + { + "epoch": 4.83, + "learning_rate": 7.1754288800540836e-06, + "loss": 0.1286, + "step": 11528 + }, + { + "epoch": 4.83, + "learning_rate": 7.173802548949312e-06, + "loss": 0.1323, + "step": 11529 + }, + { + "epoch": 4.83, + "learning_rate": 7.172176299081078e-06, + "loss": 0.1918, + "step": 11530 + }, + { + "epoch": 4.83, + "learning_rate": 7.17055013049613e-06, + "loss": 0.1785, + "step": 11531 + }, + { + "epoch": 4.83, + "learning_rate": 7.168924043241205e-06, + "loss": 0.1526, + "step": 11532 + }, + { + "epoch": 4.83, + "learning_rate": 7.1672980373630455e-06, + "loss": 0.1177, + "step": 11533 + }, + { + "epoch": 4.83, + "learning_rate": 7.165672112908393e-06, + "loss": 0.0849, + "step": 11534 + }, + { + "epoch": 4.83, + "learning_rate": 7.164046269923978e-06, + "loss": 0.2089, + "step": 11535 + }, + { + "epoch": 4.83, + "learning_rate": 7.162420508456536e-06, + "loss": 0.1855, + "step": 11536 + }, + { + "epoch": 4.83, + "learning_rate": 7.160794828552802e-06, + "loss": 0.1502, + "step": 11537 + }, + { + "epoch": 4.83, + "learning_rate": 7.159169230259497e-06, + "loss": 0.1467, + "step": 11538 + }, + { + "epoch": 4.83, + "learning_rate": 7.157543713623353e-06, + "loss": 0.1252, + "step": 11539 + }, + { + "epoch": 4.83, + "learning_rate": 7.155918278691094e-06, + "loss": 0.1651, + "step": 11540 + }, + { + "epoch": 4.83, + "learning_rate": 7.1542929255094386e-06, + "loss": 0.1559, + "step": 11541 + }, + { + "epoch": 4.83, + "learning_rate": 7.1526676541251075e-06, + "loss": 0.1774, + "step": 11542 + }, + { + "epoch": 4.83, + "learning_rate": 7.151042464584821e-06, + "loss": 0.1794, + "step": 11543 + }, + { + "epoch": 4.83, + "learning_rate": 7.149417356935289e-06, + "loss": 0.1324, + "step": 11544 + }, + { + "epoch": 4.83, + "learning_rate": 7.1477923312232235e-06, + "loss": 0.1458, + "step": 11545 + }, + { + "epoch": 4.83, + "learning_rate": 7.14616738749534e-06, + "loss": 0.1399, + "step": 11546 + }, + { + "epoch": 4.83, + "learning_rate": 7.1445425257983395e-06, + "loss": 0.15, + "step": 11547 + }, + { + "epoch": 4.83, + "learning_rate": 7.14291774617893e-06, + "loss": 0.1499, + "step": 11548 + }, + { + "epoch": 4.84, + "learning_rate": 7.141293048683816e-06, + "loss": 0.1482, + "step": 11549 + }, + { + "epoch": 4.84, + "learning_rate": 7.1396684333597e-06, + "loss": 0.1563, + "step": 11550 + }, + { + "epoch": 4.84, + "learning_rate": 7.138043900253275e-06, + "loss": 0.1151, + "step": 11551 + }, + { + "epoch": 4.84, + "learning_rate": 7.136419449411236e-06, + "loss": 0.1529, + "step": 11552 + }, + { + "epoch": 4.84, + "learning_rate": 7.134795080880278e-06, + "loss": 0.1641, + "step": 11553 + }, + { + "epoch": 4.84, + "learning_rate": 7.133170794707096e-06, + "loss": 0.1526, + "step": 11554 + }, + { + "epoch": 4.84, + "learning_rate": 7.131546590938373e-06, + "loss": 0.1107, + "step": 11555 + }, + { + "epoch": 4.84, + "learning_rate": 7.129922469620798e-06, + "loss": 0.1722, + "step": 11556 + }, + { + "epoch": 4.84, + "learning_rate": 7.128298430801057e-06, + "loss": 0.1138, + "step": 11557 + }, + { + "epoch": 4.84, + "learning_rate": 7.1266744745258264e-06, + "loss": 0.1476, + "step": 11558 + }, + { + "epoch": 4.84, + "learning_rate": 7.12505060084179e-06, + "loss": 0.1776, + "step": 11559 + }, + { + "epoch": 4.84, + "learning_rate": 7.123426809795622e-06, + "loss": 0.1628, + "step": 11560 + }, + { + "epoch": 4.84, + "learning_rate": 7.121803101433998e-06, + "loss": 0.1727, + "step": 11561 + }, + { + "epoch": 4.84, + "learning_rate": 7.1201794758035905e-06, + "loss": 0.1536, + "step": 11562 + }, + { + "epoch": 4.84, + "learning_rate": 7.118555932951067e-06, + "loss": 0.1326, + "step": 11563 + }, + { + "epoch": 4.84, + "learning_rate": 7.116932472923098e-06, + "loss": 0.1376, + "step": 11564 + }, + { + "epoch": 4.84, + "learning_rate": 7.115309095766345e-06, + "loss": 0.1419, + "step": 11565 + }, + { + "epoch": 4.84, + "learning_rate": 7.113685801527473e-06, + "loss": 0.111, + "step": 11566 + }, + { + "epoch": 4.84, + "learning_rate": 7.112062590253143e-06, + "loss": 0.1201, + "step": 11567 + }, + { + "epoch": 4.84, + "learning_rate": 7.110439461990008e-06, + "loss": 0.1694, + "step": 11568 + }, + { + "epoch": 4.84, + "learning_rate": 7.108816416784728e-06, + "loss": 0.1499, + "step": 11569 + }, + { + "epoch": 4.84, + "learning_rate": 7.107193454683956e-06, + "loss": 0.1669, + "step": 11570 + }, + { + "epoch": 4.84, + "learning_rate": 7.105570575734339e-06, + "loss": 0.1315, + "step": 11571 + }, + { + "epoch": 4.84, + "learning_rate": 7.103947779982532e-06, + "loss": 0.1531, + "step": 11572 + }, + { + "epoch": 4.85, + "learning_rate": 7.102325067475171e-06, + "loss": 0.13, + "step": 11573 + }, + { + "epoch": 4.85, + "learning_rate": 7.100702438258905e-06, + "loss": 0.1609, + "step": 11574 + }, + { + "epoch": 4.85, + "learning_rate": 7.099079892380376e-06, + "loss": 0.1277, + "step": 11575 + }, + { + "epoch": 4.85, + "learning_rate": 7.097457429886221e-06, + "loss": 0.1312, + "step": 11576 + }, + { + "epoch": 4.85, + "learning_rate": 7.095835050823074e-06, + "loss": 0.1627, + "step": 11577 + }, + { + "epoch": 4.85, + "learning_rate": 7.094212755237576e-06, + "loss": 0.1633, + "step": 11578 + }, + { + "epoch": 4.85, + "learning_rate": 7.092590543176351e-06, + "loss": 0.156, + "step": 11579 + }, + { + "epoch": 4.85, + "learning_rate": 7.09096841468603e-06, + "loss": 0.1543, + "step": 11580 + }, + { + "epoch": 4.85, + "learning_rate": 7.089346369813244e-06, + "loss": 0.1305, + "step": 11581 + }, + { + "epoch": 4.85, + "learning_rate": 7.087724408604611e-06, + "loss": 0.15, + "step": 11582 + }, + { + "epoch": 4.85, + "learning_rate": 7.086102531106755e-06, + "loss": 0.1845, + "step": 11583 + }, + { + "epoch": 4.85, + "learning_rate": 7.084480737366297e-06, + "loss": 0.1297, + "step": 11584 + }, + { + "epoch": 4.85, + "learning_rate": 7.082859027429853e-06, + "loss": 0.1236, + "step": 11585 + }, + { + "epoch": 4.85, + "learning_rate": 7.0812374013440365e-06, + "loss": 0.1535, + "step": 11586 + }, + { + "epoch": 4.85, + "learning_rate": 7.079615859155464e-06, + "loss": 0.1349, + "step": 11587 + }, + { + "epoch": 4.85, + "learning_rate": 7.077994400910739e-06, + "loss": 0.1956, + "step": 11588 + }, + { + "epoch": 4.85, + "learning_rate": 7.0763730266564725e-06, + "loss": 0.1305, + "step": 11589 + }, + { + "epoch": 4.85, + "learning_rate": 7.0747517364392694e-06, + "loss": 0.1581, + "step": 11590 + }, + { + "epoch": 4.85, + "learning_rate": 7.073130530305733e-06, + "loss": 0.149, + "step": 11591 + }, + { + "epoch": 4.85, + "learning_rate": 7.0715094083024604e-06, + "loss": 0.1399, + "step": 11592 + }, + { + "epoch": 4.85, + "learning_rate": 7.069888370476051e-06, + "loss": 0.174, + "step": 11593 + }, + { + "epoch": 4.85, + "learning_rate": 7.068267416873102e-06, + "loss": 0.1423, + "step": 11594 + }, + { + "epoch": 4.85, + "learning_rate": 7.066646547540208e-06, + "loss": 0.148, + "step": 11595 + }, + { + "epoch": 4.85, + "learning_rate": 7.065025762523952e-06, + "loss": 0.174, + "step": 11596 + }, + { + "epoch": 4.86, + "learning_rate": 7.063405061870925e-06, + "loss": 0.1655, + "step": 11597 + }, + { + "epoch": 4.86, + "learning_rate": 7.061784445627715e-06, + "loss": 0.1518, + "step": 11598 + }, + { + "epoch": 4.86, + "learning_rate": 7.060163913840903e-06, + "loss": 0.1504, + "step": 11599 + }, + { + "epoch": 4.86, + "learning_rate": 7.058543466557069e-06, + "loss": 0.1705, + "step": 11600 + }, + { + "epoch": 4.86, + "learning_rate": 7.056923103822795e-06, + "loss": 0.1463, + "step": 11601 + }, + { + "epoch": 4.86, + "learning_rate": 7.0553028256846555e-06, + "loss": 0.1493, + "step": 11602 + }, + { + "epoch": 4.86, + "learning_rate": 7.053682632189221e-06, + "loss": 0.1341, + "step": 11603 + }, + { + "epoch": 4.86, + "learning_rate": 7.052062523383065e-06, + "loss": 0.1391, + "step": 11604 + }, + { + "epoch": 4.86, + "learning_rate": 7.050442499312757e-06, + "loss": 0.1184, + "step": 11605 + }, + { + "epoch": 4.86, + "learning_rate": 7.048822560024862e-06, + "loss": 0.1176, + "step": 11606 + }, + { + "epoch": 4.86, + "learning_rate": 7.047202705565942e-06, + "loss": 0.165, + "step": 11607 + }, + { + "epoch": 4.86, + "learning_rate": 7.045582935982562e-06, + "loss": 0.1549, + "step": 11608 + }, + { + "epoch": 4.86, + "learning_rate": 7.0439632513212775e-06, + "loss": 0.153, + "step": 11609 + }, + { + "epoch": 4.86, + "learning_rate": 7.042343651628646e-06, + "loss": 0.1586, + "step": 11610 + }, + { + "epoch": 4.86, + "learning_rate": 7.040724136951224e-06, + "loss": 0.1513, + "step": 11611 + }, + { + "epoch": 4.86, + "learning_rate": 7.0391047073355575e-06, + "loss": 0.1322, + "step": 11612 + }, + { + "epoch": 4.86, + "learning_rate": 7.0374853628282e-06, + "loss": 0.1278, + "step": 11613 + }, + { + "epoch": 4.86, + "learning_rate": 7.035866103475698e-06, + "loss": 0.1393, + "step": 11614 + }, + { + "epoch": 4.86, + "learning_rate": 7.034246929324593e-06, + "loss": 0.1838, + "step": 11615 + }, + { + "epoch": 4.86, + "learning_rate": 7.032627840421429e-06, + "loss": 0.1764, + "step": 11616 + }, + { + "epoch": 4.86, + "learning_rate": 7.031008836812743e-06, + "loss": 0.1596, + "step": 11617 + }, + { + "epoch": 4.86, + "learning_rate": 7.029389918545078e-06, + "loss": 0.1471, + "step": 11618 + }, + { + "epoch": 4.86, + "learning_rate": 7.0277710856649614e-06, + "loss": 0.1407, + "step": 11619 + }, + { + "epoch": 4.86, + "learning_rate": 7.026152338218925e-06, + "loss": 0.1156, + "step": 11620 + }, + { + "epoch": 4.87, + "learning_rate": 7.024533676253502e-06, + "loss": 0.1556, + "step": 11621 + }, + { + "epoch": 4.87, + "learning_rate": 7.022915099815219e-06, + "loss": 0.1736, + "step": 11622 + }, + { + "epoch": 4.87, + "learning_rate": 7.021296608950599e-06, + "loss": 0.1521, + "step": 11623 + }, + { + "epoch": 4.87, + "learning_rate": 7.019678203706164e-06, + "loss": 0.1597, + "step": 11624 + }, + { + "epoch": 4.87, + "learning_rate": 7.018059884128437e-06, + "loss": 0.1699, + "step": 11625 + }, + { + "epoch": 4.87, + "learning_rate": 7.01644165026393e-06, + "loss": 0.1904, + "step": 11626 + }, + { + "epoch": 4.87, + "learning_rate": 7.014823502159159e-06, + "loss": 0.1579, + "step": 11627 + }, + { + "epoch": 4.87, + "learning_rate": 7.0132054398606405e-06, + "loss": 0.1672, + "step": 11628 + }, + { + "epoch": 4.87, + "learning_rate": 7.011587463414879e-06, + "loss": 0.1292, + "step": 11629 + }, + { + "epoch": 4.87, + "learning_rate": 7.009969572868385e-06, + "loss": 0.1356, + "step": 11630 + }, + { + "epoch": 4.87, + "learning_rate": 7.008351768267661e-06, + "loss": 0.1077, + "step": 11631 + }, + { + "epoch": 4.87, + "learning_rate": 7.006734049659213e-06, + "loss": 0.1405, + "step": 11632 + }, + { + "epoch": 4.87, + "learning_rate": 7.005116417089537e-06, + "loss": 0.1182, + "step": 11633 + }, + { + "epoch": 4.87, + "learning_rate": 7.0034988706051325e-06, + "loss": 0.169, + "step": 11634 + }, + { + "epoch": 4.87, + "learning_rate": 7.001881410252494e-06, + "loss": 0.1308, + "step": 11635 + }, + { + "epoch": 4.87, + "learning_rate": 7.000264036078113e-06, + "loss": 0.1397, + "step": 11636 + }, + { + "epoch": 4.87, + "learning_rate": 6.9986467481284805e-06, + "loss": 0.1342, + "step": 11637 + }, + { + "epoch": 4.87, + "learning_rate": 6.997029546450085e-06, + "loss": 0.112, + "step": 11638 + }, + { + "epoch": 4.87, + "learning_rate": 6.99541243108941e-06, + "loss": 0.1643, + "step": 11639 + }, + { + "epoch": 4.87, + "learning_rate": 6.9937954020929414e-06, + "loss": 0.1515, + "step": 11640 + }, + { + "epoch": 4.87, + "learning_rate": 6.992178459507153e-06, + "loss": 0.1187, + "step": 11641 + }, + { + "epoch": 4.87, + "learning_rate": 6.990561603378524e-06, + "loss": 0.144, + "step": 11642 + }, + { + "epoch": 4.87, + "learning_rate": 6.9889448337535345e-06, + "loss": 0.1462, + "step": 11643 + }, + { + "epoch": 4.88, + "learning_rate": 6.987328150678652e-06, + "loss": 0.1524, + "step": 11644 + }, + { + "epoch": 4.88, + "learning_rate": 6.985711554200347e-06, + "loss": 0.1162, + "step": 11645 + }, + { + "epoch": 4.88, + "learning_rate": 6.984095044365092e-06, + "loss": 0.154, + "step": 11646 + }, + { + "epoch": 4.88, + "learning_rate": 6.982478621219346e-06, + "loss": 0.1642, + "step": 11647 + }, + { + "epoch": 4.88, + "learning_rate": 6.980862284809575e-06, + "loss": 0.1614, + "step": 11648 + }, + { + "epoch": 4.88, + "learning_rate": 6.97924603518224e-06, + "loss": 0.1626, + "step": 11649 + }, + { + "epoch": 4.88, + "learning_rate": 6.9776298723837956e-06, + "loss": 0.1456, + "step": 11650 + }, + { + "epoch": 4.88, + "learning_rate": 6.976013796460698e-06, + "loss": 0.1292, + "step": 11651 + }, + { + "epoch": 4.88, + "learning_rate": 6.974397807459402e-06, + "loss": 0.1422, + "step": 11652 + }, + { + "epoch": 4.88, + "learning_rate": 6.972781905426354e-06, + "loss": 0.133, + "step": 11653 + }, + { + "epoch": 4.88, + "learning_rate": 6.9711660904080055e-06, + "loss": 0.154, + "step": 11654 + }, + { + "epoch": 4.88, + "learning_rate": 6.9695503624508005e-06, + "loss": 0.1427, + "step": 11655 + }, + { + "epoch": 4.88, + "learning_rate": 6.96793472160118e-06, + "loss": 0.1513, + "step": 11656 + }, + { + "epoch": 4.88, + "learning_rate": 6.966319167905584e-06, + "loss": 0.1249, + "step": 11657 + }, + { + "epoch": 4.88, + "learning_rate": 6.964703701410455e-06, + "loss": 0.1561, + "step": 11658 + }, + { + "epoch": 4.88, + "learning_rate": 6.963088322162221e-06, + "loss": 0.139, + "step": 11659 + }, + { + "epoch": 4.88, + "learning_rate": 6.96147303020732e-06, + "loss": 0.194, + "step": 11660 + }, + { + "epoch": 4.88, + "learning_rate": 6.95985782559218e-06, + "loss": 0.1512, + "step": 11661 + }, + { + "epoch": 4.88, + "learning_rate": 6.958242708363231e-06, + "loss": 0.2009, + "step": 11662 + }, + { + "epoch": 4.88, + "learning_rate": 6.956627678566898e-06, + "loss": 0.1388, + "step": 11663 + }, + { + "epoch": 4.88, + "learning_rate": 6.955012736249598e-06, + "loss": 0.1439, + "step": 11664 + }, + { + "epoch": 4.88, + "learning_rate": 6.953397881457754e-06, + "loss": 0.1572, + "step": 11665 + }, + { + "epoch": 4.88, + "learning_rate": 6.951783114237786e-06, + "loss": 0.1202, + "step": 11666 + }, + { + "epoch": 4.88, + "learning_rate": 6.950168434636106e-06, + "loss": 0.1251, + "step": 11667 + }, + { + "epoch": 4.89, + "learning_rate": 6.94855384269913e-06, + "loss": 0.1462, + "step": 11668 + }, + { + "epoch": 4.89, + "learning_rate": 6.946939338473267e-06, + "loss": 0.1682, + "step": 11669 + }, + { + "epoch": 4.89, + "learning_rate": 6.945324922004919e-06, + "loss": 0.1335, + "step": 11670 + }, + { + "epoch": 4.89, + "learning_rate": 6.943710593340498e-06, + "loss": 0.1708, + "step": 11671 + }, + { + "epoch": 4.89, + "learning_rate": 6.9420963525264025e-06, + "loss": 0.1784, + "step": 11672 + }, + { + "epoch": 4.89, + "learning_rate": 6.940482199609036e-06, + "loss": 0.1363, + "step": 11673 + }, + { + "epoch": 4.89, + "learning_rate": 6.938868134634793e-06, + "loss": 0.1453, + "step": 11674 + }, + { + "epoch": 4.89, + "learning_rate": 6.937254157650068e-06, + "loss": 0.1673, + "step": 11675 + }, + { + "epoch": 4.89, + "learning_rate": 6.935640268701256e-06, + "loss": 0.1203, + "step": 11676 + }, + { + "epoch": 4.89, + "learning_rate": 6.934026467834745e-06, + "loss": 0.1493, + "step": 11677 + }, + { + "epoch": 4.89, + "learning_rate": 6.93241275509692e-06, + "loss": 0.1621, + "step": 11678 + }, + { + "epoch": 4.89, + "learning_rate": 6.930799130534172e-06, + "loss": 0.1438, + "step": 11679 + }, + { + "epoch": 4.89, + "learning_rate": 6.929185594192877e-06, + "loss": 0.1704, + "step": 11680 + }, + { + "epoch": 4.89, + "learning_rate": 6.927572146119416e-06, + "loss": 0.1914, + "step": 11681 + }, + { + "epoch": 4.89, + "learning_rate": 6.925958786360172e-06, + "loss": 0.1893, + "step": 11682 + }, + { + "epoch": 4.89, + "learning_rate": 6.92434551496151e-06, + "loss": 0.115, + "step": 11683 + }, + { + "epoch": 4.89, + "learning_rate": 6.922732331969808e-06, + "loss": 0.1342, + "step": 11684 + }, + { + "epoch": 4.89, + "learning_rate": 6.921119237431436e-06, + "loss": 0.1299, + "step": 11685 + }, + { + "epoch": 4.89, + "learning_rate": 6.9195062313927605e-06, + "loss": 0.1314, + "step": 11686 + }, + { + "epoch": 4.89, + "learning_rate": 6.917893313900144e-06, + "loss": 0.1386, + "step": 11687 + }, + { + "epoch": 4.89, + "learning_rate": 6.916280484999946e-06, + "loss": 0.152, + "step": 11688 + }, + { + "epoch": 4.89, + "learning_rate": 6.914667744738531e-06, + "loss": 0.1339, + "step": 11689 + }, + { + "epoch": 4.89, + "learning_rate": 6.9130550931622555e-06, + "loss": 0.1567, + "step": 11690 + }, + { + "epoch": 4.89, + "learning_rate": 6.911442530317469e-06, + "loss": 0.1512, + "step": 11691 + }, + { + "epoch": 4.9, + "learning_rate": 6.909830056250527e-06, + "loss": 0.1504, + "step": 11692 + }, + { + "epoch": 4.9, + "learning_rate": 6.90821767100778e-06, + "loss": 0.192, + "step": 11693 + }, + { + "epoch": 4.9, + "learning_rate": 6.90660537463557e-06, + "loss": 0.1369, + "step": 11694 + }, + { + "epoch": 4.9, + "learning_rate": 6.904993167180243e-06, + "loss": 0.1093, + "step": 11695 + }, + { + "epoch": 4.9, + "learning_rate": 6.9033810486881445e-06, + "loss": 0.159, + "step": 11696 + }, + { + "epoch": 4.9, + "learning_rate": 6.901769019205606e-06, + "loss": 0.1447, + "step": 11697 + }, + { + "epoch": 4.9, + "learning_rate": 6.900157078778969e-06, + "loss": 0.1699, + "step": 11698 + }, + { + "epoch": 4.9, + "learning_rate": 6.898545227454567e-06, + "loss": 0.1377, + "step": 11699 + }, + { + "epoch": 4.9, + "learning_rate": 6.89693346527873e-06, + "loss": 0.1256, + "step": 11700 + }, + { + "epoch": 4.9, + "learning_rate": 6.895321792297785e-06, + "loss": 0.1856, + "step": 11701 + }, + { + "epoch": 4.9, + "learning_rate": 6.893710208558061e-06, + "loss": 0.1304, + "step": 11702 + }, + { + "epoch": 4.9, + "learning_rate": 6.892098714105882e-06, + "loss": 0.1064, + "step": 11703 + }, + { + "epoch": 4.9, + "learning_rate": 6.890487308987567e-06, + "loss": 0.1524, + "step": 11704 + }, + { + "epoch": 4.9, + "learning_rate": 6.888875993249433e-06, + "loss": 0.1516, + "step": 11705 + }, + { + "epoch": 4.9, + "learning_rate": 6.887264766937801e-06, + "loss": 0.1015, + "step": 11706 + }, + { + "epoch": 4.9, + "learning_rate": 6.885653630098978e-06, + "loss": 0.16, + "step": 11707 + }, + { + "epoch": 4.9, + "learning_rate": 6.884042582779285e-06, + "loss": 0.1747, + "step": 11708 + }, + { + "epoch": 4.9, + "learning_rate": 6.882431625025016e-06, + "loss": 0.1781, + "step": 11709 + }, + { + "epoch": 4.9, + "learning_rate": 6.880820756882486e-06, + "loss": 0.1813, + "step": 11710 + }, + { + "epoch": 4.9, + "learning_rate": 6.879209978397993e-06, + "loss": 0.1389, + "step": 11711 + }, + { + "epoch": 4.9, + "learning_rate": 6.87759928961784e-06, + "loss": 0.1352, + "step": 11712 + }, + { + "epoch": 4.9, + "learning_rate": 6.875988690588325e-06, + "loss": 0.2103, + "step": 11713 + }, + { + "epoch": 4.9, + "learning_rate": 6.8743781813557455e-06, + "loss": 0.1271, + "step": 11714 + }, + { + "epoch": 4.9, + "learning_rate": 6.872767761966388e-06, + "loss": 0.1428, + "step": 11715 + }, + { + "epoch": 4.91, + "learning_rate": 6.8711574324665476e-06, + "loss": 0.1631, + "step": 11716 + }, + { + "epoch": 4.91, + "learning_rate": 6.869547192902511e-06, + "loss": 0.1268, + "step": 11717 + }, + { + "epoch": 4.91, + "learning_rate": 6.867937043320562e-06, + "loss": 0.1482, + "step": 11718 + }, + { + "epoch": 4.91, + "learning_rate": 6.866326983766982e-06, + "loss": 0.1294, + "step": 11719 + }, + { + "epoch": 4.91, + "learning_rate": 6.8647170142880545e-06, + "loss": 0.1384, + "step": 11720 + }, + { + "epoch": 4.91, + "learning_rate": 6.863107134930052e-06, + "loss": 0.13, + "step": 11721 + }, + { + "epoch": 4.91, + "learning_rate": 6.861497345739252e-06, + "loss": 0.111, + "step": 11722 + }, + { + "epoch": 4.91, + "learning_rate": 6.859887646761927e-06, + "loss": 0.152, + "step": 11723 + }, + { + "epoch": 4.91, + "learning_rate": 6.858278038044344e-06, + "loss": 0.1708, + "step": 11724 + }, + { + "epoch": 4.91, + "learning_rate": 6.85666851963277e-06, + "loss": 0.1705, + "step": 11725 + }, + { + "epoch": 4.91, + "learning_rate": 6.855059091573472e-06, + "loss": 0.1319, + "step": 11726 + }, + { + "epoch": 4.91, + "learning_rate": 6.853449753912708e-06, + "loss": 0.1653, + "step": 11727 + }, + { + "epoch": 4.91, + "learning_rate": 6.851840506696738e-06, + "loss": 0.1324, + "step": 11728 + }, + { + "epoch": 4.91, + "learning_rate": 6.8502313499718206e-06, + "loss": 0.1431, + "step": 11729 + }, + { + "epoch": 4.91, + "learning_rate": 6.848622283784209e-06, + "loss": 0.1088, + "step": 11730 + }, + { + "epoch": 4.91, + "learning_rate": 6.847013308180155e-06, + "loss": 0.1375, + "step": 11731 + }, + { + "epoch": 4.91, + "learning_rate": 6.845404423205903e-06, + "loss": 0.1367, + "step": 11732 + }, + { + "epoch": 4.91, + "learning_rate": 6.843795628907699e-06, + "loss": 0.1468, + "step": 11733 + }, + { + "epoch": 4.91, + "learning_rate": 6.842186925331793e-06, + "loss": 0.1308, + "step": 11734 + }, + { + "epoch": 4.91, + "learning_rate": 6.8405783125244196e-06, + "loss": 0.1238, + "step": 11735 + }, + { + "epoch": 4.91, + "learning_rate": 6.838969790531817e-06, + "loss": 0.1603, + "step": 11736 + }, + { + "epoch": 4.91, + "learning_rate": 6.837361359400227e-06, + "loss": 0.1417, + "step": 11737 + }, + { + "epoch": 4.91, + "learning_rate": 6.835753019175875e-06, + "loss": 0.1458, + "step": 11738 + }, + { + "epoch": 4.91, + "learning_rate": 6.834144769904996e-06, + "loss": 0.16, + "step": 11739 + }, + { + "epoch": 4.92, + "learning_rate": 6.832536611633816e-06, + "loss": 0.1393, + "step": 11740 + }, + { + "epoch": 4.92, + "learning_rate": 6.830928544408559e-06, + "loss": 0.1548, + "step": 11741 + }, + { + "epoch": 4.92, + "learning_rate": 6.82932056827545e-06, + "loss": 0.1411, + "step": 11742 + }, + { + "epoch": 4.92, + "learning_rate": 6.827712683280707e-06, + "loss": 0.1808, + "step": 11743 + }, + { + "epoch": 4.92, + "learning_rate": 6.8261048894705505e-06, + "loss": 0.181, + "step": 11744 + }, + { + "epoch": 4.92, + "learning_rate": 6.824497186891189e-06, + "loss": 0.1412, + "step": 11745 + }, + { + "epoch": 4.92, + "learning_rate": 6.822889575588838e-06, + "loss": 0.1402, + "step": 11746 + }, + { + "epoch": 4.92, + "learning_rate": 6.8212820556097105e-06, + "loss": 0.1246, + "step": 11747 + }, + { + "epoch": 4.92, + "learning_rate": 6.819674627000006e-06, + "loss": 0.1237, + "step": 11748 + }, + { + "epoch": 4.92, + "learning_rate": 6.818067289805934e-06, + "loss": 0.1781, + "step": 11749 + }, + { + "epoch": 4.92, + "learning_rate": 6.816460044073694e-06, + "loss": 0.144, + "step": 11750 + }, + { + "epoch": 4.92, + "learning_rate": 6.8148528898494845e-06, + "loss": 0.1312, + "step": 11751 + }, + { + "epoch": 4.92, + "learning_rate": 6.813245827179502e-06, + "loss": 0.1339, + "step": 11752 + }, + { + "epoch": 4.92, + "learning_rate": 6.811638856109941e-06, + "loss": 0.1753, + "step": 11753 + }, + { + "epoch": 4.92, + "learning_rate": 6.810031976686994e-06, + "loss": 0.1361, + "step": 11754 + }, + { + "epoch": 4.92, + "learning_rate": 6.808425188956847e-06, + "loss": 0.1599, + "step": 11755 + }, + { + "epoch": 4.92, + "learning_rate": 6.8068184929656815e-06, + "loss": 0.1234, + "step": 11756 + }, + { + "epoch": 4.92, + "learning_rate": 6.8052118887596865e-06, + "loss": 0.1611, + "step": 11757 + }, + { + "epoch": 4.92, + "learning_rate": 6.803605376385042e-06, + "loss": 0.1567, + "step": 11758 + }, + { + "epoch": 4.92, + "learning_rate": 6.801998955887923e-06, + "loss": 0.1952, + "step": 11759 + }, + { + "epoch": 4.92, + "learning_rate": 6.800392627314506e-06, + "loss": 0.1309, + "step": 11760 + }, + { + "epoch": 4.92, + "learning_rate": 6.7987863907109656e-06, + "loss": 0.1267, + "step": 11761 + }, + { + "epoch": 4.92, + "learning_rate": 6.797180246123467e-06, + "loss": 0.2057, + "step": 11762 + }, + { + "epoch": 4.92, + "learning_rate": 6.795574193598182e-06, + "loss": 0.182, + "step": 11763 + }, + { + "epoch": 4.93, + "learning_rate": 6.793968233181274e-06, + "loss": 0.1429, + "step": 11764 + }, + { + "epoch": 4.93, + "learning_rate": 6.792362364918902e-06, + "loss": 0.1341, + "step": 11765 + }, + { + "epoch": 4.93, + "learning_rate": 6.790756588857228e-06, + "loss": 0.1332, + "step": 11766 + }, + { + "epoch": 4.93, + "learning_rate": 6.7891509050424095e-06, + "loss": 0.146, + "step": 11767 + }, + { + "epoch": 4.93, + "learning_rate": 6.787545313520597e-06, + "loss": 0.1507, + "step": 11768 + }, + { + "epoch": 4.93, + "learning_rate": 6.785939814337944e-06, + "loss": 0.1471, + "step": 11769 + }, + { + "epoch": 4.93, + "learning_rate": 6.7843344075406005e-06, + "loss": 0.1545, + "step": 11770 + }, + { + "epoch": 4.93, + "learning_rate": 6.782729093174709e-06, + "loss": 0.1753, + "step": 11771 + }, + { + "epoch": 4.93, + "learning_rate": 6.781123871286416e-06, + "loss": 0.1701, + "step": 11772 + }, + { + "epoch": 4.93, + "learning_rate": 6.77951874192186e-06, + "loss": 0.1585, + "step": 11773 + }, + { + "epoch": 4.93, + "learning_rate": 6.777913705127181e-06, + "loss": 0.1698, + "step": 11774 + }, + { + "epoch": 4.93, + "learning_rate": 6.776308760948514e-06, + "loss": 0.1194, + "step": 11775 + }, + { + "epoch": 4.93, + "learning_rate": 6.774703909431992e-06, + "loss": 0.1379, + "step": 11776 + }, + { + "epoch": 4.93, + "learning_rate": 6.773099150623742e-06, + "loss": 0.1164, + "step": 11777 + }, + { + "epoch": 4.93, + "learning_rate": 6.771494484569895e-06, + "loss": 0.1853, + "step": 11778 + }, + { + "epoch": 4.93, + "learning_rate": 6.7698899113165715e-06, + "loss": 0.1434, + "step": 11779 + }, + { + "epoch": 4.93, + "learning_rate": 6.768285430909896e-06, + "loss": 0.1482, + "step": 11780 + }, + { + "epoch": 4.93, + "learning_rate": 6.76668104339599e-06, + "loss": 0.1466, + "step": 11781 + }, + { + "epoch": 4.93, + "learning_rate": 6.765076748820968e-06, + "loss": 0.1943, + "step": 11782 + }, + { + "epoch": 4.93, + "learning_rate": 6.763472547230943e-06, + "loss": 0.1793, + "step": 11783 + }, + { + "epoch": 4.93, + "learning_rate": 6.761868438672028e-06, + "loss": 0.1735, + "step": 11784 + }, + { + "epoch": 4.93, + "learning_rate": 6.760264423190332e-06, + "loss": 0.1779, + "step": 11785 + }, + { + "epoch": 4.93, + "learning_rate": 6.75866050083196e-06, + "loss": 0.1505, + "step": 11786 + }, + { + "epoch": 4.93, + "learning_rate": 6.757056671643014e-06, + "loss": 0.1513, + "step": 11787 + }, + { + "epoch": 4.94, + "learning_rate": 6.7554529356696e-06, + "loss": 0.1411, + "step": 11788 + }, + { + "epoch": 4.94, + "learning_rate": 6.753849292957808e-06, + "loss": 0.1627, + "step": 11789 + }, + { + "epoch": 4.94, + "learning_rate": 6.752245743553738e-06, + "loss": 0.1364, + "step": 11790 + }, + { + "epoch": 4.94, + "learning_rate": 6.750642287503486e-06, + "loss": 0.1319, + "step": 11791 + }, + { + "epoch": 4.94, + "learning_rate": 6.749038924853134e-06, + "loss": 0.1318, + "step": 11792 + }, + { + "epoch": 4.94, + "learning_rate": 6.747435655648774e-06, + "loss": 0.1795, + "step": 11793 + }, + { + "epoch": 4.94, + "learning_rate": 6.745832479936492e-06, + "loss": 0.1398, + "step": 11794 + }, + { + "epoch": 4.94, + "learning_rate": 6.744229397762365e-06, + "loss": 0.1839, + "step": 11795 + }, + { + "epoch": 4.94, + "learning_rate": 6.742626409172475e-06, + "loss": 0.1448, + "step": 11796 + }, + { + "epoch": 4.94, + "learning_rate": 6.741023514212901e-06, + "loss": 0.1557, + "step": 11797 + }, + { + "epoch": 4.94, + "learning_rate": 6.739420712929712e-06, + "loss": 0.139, + "step": 11798 + }, + { + "epoch": 4.94, + "learning_rate": 6.7378180053689854e-06, + "loss": 0.1092, + "step": 11799 + }, + { + "epoch": 4.94, + "learning_rate": 6.736215391576782e-06, + "loss": 0.1896, + "step": 11800 + }, + { + "epoch": 4.94, + "learning_rate": 6.734612871599169e-06, + "loss": 0.1533, + "step": 11801 + }, + { + "epoch": 4.94, + "learning_rate": 6.733010445482214e-06, + "loss": 0.1236, + "step": 11802 + }, + { + "epoch": 4.94, + "learning_rate": 6.7314081132719735e-06, + "loss": 0.1565, + "step": 11803 + }, + { + "epoch": 4.94, + "learning_rate": 6.729805875014506e-06, + "loss": 0.1567, + "step": 11804 + }, + { + "epoch": 4.94, + "learning_rate": 6.728203730755868e-06, + "loss": 0.1762, + "step": 11805 + }, + { + "epoch": 4.94, + "learning_rate": 6.726601680542109e-06, + "loss": 0.1186, + "step": 11806 + }, + { + "epoch": 4.94, + "learning_rate": 6.724999724419279e-06, + "loss": 0.1542, + "step": 11807 + }, + { + "epoch": 4.94, + "learning_rate": 6.723397862433429e-06, + "loss": 0.1656, + "step": 11808 + }, + { + "epoch": 4.94, + "learning_rate": 6.721796094630597e-06, + "loss": 0.1587, + "step": 11809 + }, + { + "epoch": 4.94, + "learning_rate": 6.720194421056828e-06, + "loss": 0.1193, + "step": 11810 + }, + { + "epoch": 4.94, + "learning_rate": 6.7185928417581615e-06, + "loss": 0.1541, + "step": 11811 + }, + { + "epoch": 4.95, + "learning_rate": 6.71699135678063e-06, + "loss": 0.1437, + "step": 11812 + }, + { + "epoch": 4.95, + "learning_rate": 6.715389966170269e-06, + "loss": 0.1602, + "step": 11813 + }, + { + "epoch": 4.95, + "learning_rate": 6.713788669973108e-06, + "loss": 0.1463, + "step": 11814 + }, + { + "epoch": 4.95, + "learning_rate": 6.7121874682351784e-06, + "loss": 0.1272, + "step": 11815 + }, + { + "epoch": 4.95, + "learning_rate": 6.710586361002499e-06, + "loss": 0.1211, + "step": 11816 + }, + { + "epoch": 4.95, + "learning_rate": 6.708985348321098e-06, + "loss": 0.1686, + "step": 11817 + }, + { + "epoch": 4.95, + "learning_rate": 6.707384430236991e-06, + "loss": 0.1725, + "step": 11818 + }, + { + "epoch": 4.95, + "learning_rate": 6.705783606796199e-06, + "loss": 0.1516, + "step": 11819 + }, + { + "epoch": 4.95, + "learning_rate": 6.704182878044732e-06, + "loss": 0.1333, + "step": 11820 + }, + { + "epoch": 4.95, + "learning_rate": 6.702582244028605e-06, + "loss": 0.1902, + "step": 11821 + }, + { + "epoch": 4.95, + "learning_rate": 6.700981704793828e-06, + "loss": 0.1359, + "step": 11822 + }, + { + "epoch": 4.95, + "learning_rate": 6.6993812603864e-06, + "loss": 0.1323, + "step": 11823 + }, + { + "epoch": 4.95, + "learning_rate": 6.6977809108523295e-06, + "loss": 0.1735, + "step": 11824 + }, + { + "epoch": 4.95, + "learning_rate": 6.696180656237616e-06, + "loss": 0.1605, + "step": 11825 + }, + { + "epoch": 4.95, + "learning_rate": 6.69458049658826e-06, + "loss": 0.1059, + "step": 11826 + }, + { + "epoch": 4.95, + "learning_rate": 6.6929804319502515e-06, + "loss": 0.1623, + "step": 11827 + }, + { + "epoch": 4.95, + "learning_rate": 6.691380462369588e-06, + "loss": 0.1609, + "step": 11828 + }, + { + "epoch": 4.95, + "learning_rate": 6.689780587892256e-06, + "loss": 0.1591, + "step": 11829 + }, + { + "epoch": 4.95, + "learning_rate": 6.688180808564244e-06, + "loss": 0.1501, + "step": 11830 + }, + { + "epoch": 4.95, + "learning_rate": 6.686581124431535e-06, + "loss": 0.1616, + "step": 11831 + }, + { + "epoch": 4.95, + "learning_rate": 6.684981535540114e-06, + "loss": 0.1992, + "step": 11832 + }, + { + "epoch": 4.95, + "learning_rate": 6.6833820419359526e-06, + "loss": 0.1278, + "step": 11833 + }, + { + "epoch": 4.95, + "learning_rate": 6.6817826436650335e-06, + "loss": 0.1619, + "step": 11834 + }, + { + "epoch": 4.95, + "learning_rate": 6.680183340773329e-06, + "loss": 0.1707, + "step": 11835 + }, + { + "epoch": 4.96, + "learning_rate": 6.678584133306808e-06, + "loss": 0.1649, + "step": 11836 + }, + { + "epoch": 4.96, + "learning_rate": 6.6769850213114375e-06, + "loss": 0.1744, + "step": 11837 + }, + { + "epoch": 4.96, + "learning_rate": 6.675386004833185e-06, + "loss": 0.1598, + "step": 11838 + }, + { + "epoch": 4.96, + "learning_rate": 6.673787083918011e-06, + "loss": 0.1677, + "step": 11839 + }, + { + "epoch": 4.96, + "learning_rate": 6.672188258611874e-06, + "loss": 0.1296, + "step": 11840 + }, + { + "epoch": 4.96, + "learning_rate": 6.670589528960734e-06, + "loss": 0.1734, + "step": 11841 + }, + { + "epoch": 4.96, + "learning_rate": 6.668990895010545e-06, + "loss": 0.1718, + "step": 11842 + }, + { + "epoch": 4.96, + "learning_rate": 6.667392356807255e-06, + "loss": 0.1761, + "step": 11843 + }, + { + "epoch": 4.96, + "learning_rate": 6.665793914396818e-06, + "loss": 0.1362, + "step": 11844 + }, + { + "epoch": 4.96, + "learning_rate": 6.664195567825174e-06, + "loss": 0.1253, + "step": 11845 + }, + { + "epoch": 4.96, + "learning_rate": 6.662597317138268e-06, + "loss": 0.1808, + "step": 11846 + }, + { + "epoch": 4.96, + "learning_rate": 6.660999162382038e-06, + "loss": 0.1541, + "step": 11847 + }, + { + "epoch": 4.96, + "learning_rate": 6.6594011036024255e-06, + "loss": 0.1515, + "step": 11848 + }, + { + "epoch": 4.96, + "learning_rate": 6.657803140845367e-06, + "loss": 0.1437, + "step": 11849 + }, + { + "epoch": 4.96, + "learning_rate": 6.656205274156788e-06, + "loss": 0.1805, + "step": 11850 + }, + { + "epoch": 4.96, + "learning_rate": 6.654607503582621e-06, + "loss": 0.1519, + "step": 11851 + }, + { + "epoch": 4.96, + "learning_rate": 6.653009829168795e-06, + "loss": 0.1387, + "step": 11852 + }, + { + "epoch": 4.96, + "learning_rate": 6.651412250961229e-06, + "loss": 0.1141, + "step": 11853 + }, + { + "epoch": 4.96, + "learning_rate": 6.649814769005847e-06, + "loss": 0.1398, + "step": 11854 + }, + { + "epoch": 4.96, + "learning_rate": 6.648217383348565e-06, + "loss": 0.1622, + "step": 11855 + }, + { + "epoch": 4.96, + "learning_rate": 6.646620094035305e-06, + "loss": 0.1627, + "step": 11856 + }, + { + "epoch": 4.96, + "learning_rate": 6.645022901111971e-06, + "loss": 0.1522, + "step": 11857 + }, + { + "epoch": 4.96, + "learning_rate": 6.643425804624476e-06, + "loss": 0.1588, + "step": 11858 + }, + { + "epoch": 4.97, + "learning_rate": 6.641828804618731e-06, + "loss": 0.1015, + "step": 11859 + }, + { + "epoch": 4.97, + "learning_rate": 6.640231901140633e-06, + "loss": 0.1449, + "step": 11860 + }, + { + "epoch": 4.97, + "learning_rate": 6.638635094236089e-06, + "loss": 0.1349, + "step": 11861 + }, + { + "epoch": 4.97, + "learning_rate": 6.637038383950998e-06, + "loss": 0.1586, + "step": 11862 + }, + { + "epoch": 4.97, + "learning_rate": 6.635441770331252e-06, + "loss": 0.1046, + "step": 11863 + }, + { + "epoch": 4.97, + "learning_rate": 6.633845253422748e-06, + "loss": 0.1767, + "step": 11864 + }, + { + "epoch": 4.97, + "learning_rate": 6.632248833271376e-06, + "loss": 0.1598, + "step": 11865 + }, + { + "epoch": 4.97, + "learning_rate": 6.6306525099230214e-06, + "loss": 0.1807, + "step": 11866 + }, + { + "epoch": 4.97, + "learning_rate": 6.629056283423574e-06, + "loss": 0.1622, + "step": 11867 + }, + { + "epoch": 4.97, + "learning_rate": 6.62746015381891e-06, + "loss": 0.2002, + "step": 11868 + }, + { + "epoch": 4.97, + "learning_rate": 6.625864121154908e-06, + "loss": 0.1498, + "step": 11869 + }, + { + "epoch": 4.97, + "learning_rate": 6.624268185477452e-06, + "loss": 0.1557, + "step": 11870 + }, + { + "epoch": 4.97, + "learning_rate": 6.622672346832409e-06, + "loss": 0.1275, + "step": 11871 + }, + { + "epoch": 4.97, + "learning_rate": 6.621076605265653e-06, + "loss": 0.1179, + "step": 11872 + }, + { + "epoch": 4.97, + "learning_rate": 6.619480960823053e-06, + "loss": 0.1549, + "step": 11873 + }, + { + "epoch": 4.97, + "learning_rate": 6.617885413550471e-06, + "loss": 0.1529, + "step": 11874 + }, + { + "epoch": 4.97, + "learning_rate": 6.616289963493771e-06, + "loss": 0.1529, + "step": 11875 + }, + { + "epoch": 4.97, + "learning_rate": 6.614694610698815e-06, + "loss": 0.1736, + "step": 11876 + }, + { + "epoch": 4.97, + "learning_rate": 6.613099355211457e-06, + "loss": 0.186, + "step": 11877 + }, + { + "epoch": 4.97, + "learning_rate": 6.611504197077554e-06, + "loss": 0.1384, + "step": 11878 + }, + { + "epoch": 4.97, + "learning_rate": 6.609909136342956e-06, + "loss": 0.1638, + "step": 11879 + }, + { + "epoch": 4.97, + "learning_rate": 6.608314173053509e-06, + "loss": 0.1224, + "step": 11880 + }, + { + "epoch": 4.97, + "learning_rate": 6.606719307255064e-06, + "loss": 0.1773, + "step": 11881 + }, + { + "epoch": 4.97, + "learning_rate": 6.605124538993462e-06, + "loss": 0.1682, + "step": 11882 + }, + { + "epoch": 4.98, + "learning_rate": 6.60352986831454e-06, + "loss": 0.1524, + "step": 11883 + }, + { + "epoch": 4.98, + "learning_rate": 6.601935295264139e-06, + "loss": 0.138, + "step": 11884 + }, + { + "epoch": 4.98, + "learning_rate": 6.600340819888091e-06, + "loss": 0.1323, + "step": 11885 + }, + { + "epoch": 4.98, + "learning_rate": 6.598746442232233e-06, + "loss": 0.1436, + "step": 11886 + }, + { + "epoch": 4.98, + "learning_rate": 6.597152162342388e-06, + "loss": 0.1648, + "step": 11887 + }, + { + "epoch": 4.98, + "learning_rate": 6.595557980264384e-06, + "loss": 0.1288, + "step": 11888 + }, + { + "epoch": 4.98, + "learning_rate": 6.593963896044047e-06, + "loss": 0.1178, + "step": 11889 + }, + { + "epoch": 4.98, + "learning_rate": 6.592369909727196e-06, + "loss": 0.1467, + "step": 11890 + }, + { + "epoch": 4.98, + "learning_rate": 6.590776021359646e-06, + "loss": 0.1334, + "step": 11891 + }, + { + "epoch": 4.98, + "learning_rate": 6.589182230987213e-06, + "loss": 0.1008, + "step": 11892 + }, + { + "epoch": 4.98, + "learning_rate": 6.587588538655714e-06, + "loss": 0.148, + "step": 11893 + }, + { + "epoch": 4.98, + "learning_rate": 6.5859949444109495e-06, + "loss": 0.1416, + "step": 11894 + }, + { + "epoch": 4.98, + "learning_rate": 6.584401448298732e-06, + "loss": 0.0951, + "step": 11895 + }, + { + "epoch": 4.98, + "learning_rate": 6.582808050364864e-06, + "loss": 0.137, + "step": 11896 + }, + { + "epoch": 4.98, + "learning_rate": 6.581214750655148e-06, + "loss": 0.1516, + "step": 11897 + }, + { + "epoch": 4.98, + "learning_rate": 6.579621549215378e-06, + "loss": 0.1675, + "step": 11898 + }, + { + "epoch": 4.98, + "learning_rate": 6.578028446091351e-06, + "loss": 0.0996, + "step": 11899 + }, + { + "epoch": 4.98, + "learning_rate": 6.5764354413288635e-06, + "loss": 0.1206, + "step": 11900 + }, + { + "epoch": 4.98, + "learning_rate": 6.574842534973698e-06, + "loss": 0.1579, + "step": 11901 + }, + { + "epoch": 4.98, + "learning_rate": 6.573249727071644e-06, + "loss": 0.1271, + "step": 11902 + }, + { + "epoch": 4.98, + "learning_rate": 6.571657017668489e-06, + "loss": 0.1465, + "step": 11903 + }, + { + "epoch": 4.98, + "learning_rate": 6.570064406810008e-06, + "loss": 0.1497, + "step": 11904 + }, + { + "epoch": 4.98, + "learning_rate": 6.568471894541982e-06, + "loss": 0.1262, + "step": 11905 + }, + { + "epoch": 4.98, + "learning_rate": 6.5668794809101884e-06, + "loss": 0.1509, + "step": 11906 + }, + { + "epoch": 4.99, + "learning_rate": 6.565287165960396e-06, + "loss": 0.1165, + "step": 11907 + }, + { + "epoch": 4.99, + "learning_rate": 6.563694949738375e-06, + "loss": 0.1142, + "step": 11908 + }, + { + "epoch": 4.99, + "learning_rate": 6.562102832289896e-06, + "loss": 0.1539, + "step": 11909 + }, + { + "epoch": 4.99, + "learning_rate": 6.560510813660719e-06, + "loss": 0.1892, + "step": 11910 + }, + { + "epoch": 4.99, + "learning_rate": 6.558918893896606e-06, + "loss": 0.1619, + "step": 11911 + }, + { + "epoch": 4.99, + "learning_rate": 6.55732707304332e-06, + "loss": 0.1679, + "step": 11912 + }, + { + "epoch": 4.99, + "learning_rate": 6.5557353511466085e-06, + "loss": 0.1434, + "step": 11913 + }, + { + "epoch": 4.99, + "learning_rate": 6.554143728252229e-06, + "loss": 0.1606, + "step": 11914 + }, + { + "epoch": 4.99, + "learning_rate": 6.552552204405928e-06, + "loss": 0.126, + "step": 11915 + }, + { + "epoch": 4.99, + "learning_rate": 6.550960779653456e-06, + "loss": 0.1764, + "step": 11916 + }, + { + "epoch": 4.99, + "learning_rate": 6.5493694540405575e-06, + "loss": 0.158, + "step": 11917 + }, + { + "epoch": 4.99, + "learning_rate": 6.547778227612969e-06, + "loss": 0.1304, + "step": 11918 + }, + { + "epoch": 4.99, + "learning_rate": 6.546187100416432e-06, + "loss": 0.1372, + "step": 11919 + }, + { + "epoch": 4.99, + "learning_rate": 6.544596072496685e-06, + "loss": 0.1006, + "step": 11920 + }, + { + "epoch": 4.99, + "learning_rate": 6.5430051438994545e-06, + "loss": 0.1562, + "step": 11921 + }, + { + "epoch": 4.99, + "learning_rate": 6.541414314670475e-06, + "loss": 0.1908, + "step": 11922 + }, + { + "epoch": 4.99, + "learning_rate": 6.539823584855472e-06, + "loss": 0.179, + "step": 11923 + }, + { + "epoch": 4.99, + "learning_rate": 6.53823295450017e-06, + "loss": 0.1202, + "step": 11924 + }, + { + "epoch": 4.99, + "learning_rate": 6.536642423650288e-06, + "loss": 0.1464, + "step": 11925 + }, + { + "epoch": 4.99, + "learning_rate": 6.5350519923515464e-06, + "loss": 0.1533, + "step": 11926 + }, + { + "epoch": 4.99, + "learning_rate": 6.533461660649663e-06, + "loss": 0.1619, + "step": 11927 + }, + { + "epoch": 4.99, + "learning_rate": 6.531871428590345e-06, + "loss": 0.1269, + "step": 11928 + }, + { + "epoch": 4.99, + "learning_rate": 6.530281296219305e-06, + "loss": 0.1226, + "step": 11929 + }, + { + "epoch": 4.99, + "learning_rate": 6.528691263582254e-06, + "loss": 0.1563, + "step": 11930 + }, + { + "epoch": 5.0, + "learning_rate": 6.52710133072489e-06, + "loss": 0.1348, + "step": 11931 + }, + { + "epoch": 5.0, + "learning_rate": 6.5255114976929154e-06, + "loss": 0.1795, + "step": 11932 + }, + { + "epoch": 5.0, + "learning_rate": 6.523921764532032e-06, + "loss": 0.1407, + "step": 11933 + }, + { + "epoch": 5.0, + "learning_rate": 6.522332131287931e-06, + "loss": 0.1487, + "step": 11934 + }, + { + "epoch": 5.0, + "learning_rate": 6.520742598006311e-06, + "loss": 0.1426, + "step": 11935 + }, + { + "epoch": 5.0, + "learning_rate": 6.519153164732852e-06, + "loss": 0.1117, + "step": 11936 + }, + { + "epoch": 5.0, + "learning_rate": 6.5175638315132484e-06, + "loss": 0.1438, + "step": 11937 + }, + { + "epoch": 5.0, + "learning_rate": 6.5159745983931846e-06, + "loss": 0.1838, + "step": 11938 + }, + { + "epoch": 5.0, + "learning_rate": 6.514385465418336e-06, + "loss": 0.1861, + "step": 11939 + }, + { + "epoch": 5.0, + "learning_rate": 6.512796432634383e-06, + "loss": 0.2126, + "step": 11940 + }, + { + "epoch": 5.0, + "learning_rate": 6.5112075000870065e-06, + "loss": 0.137, + "step": 11941 + }, + { + "epoch": 5.0, + "learning_rate": 6.5096186678218705e-06, + "loss": 0.1794, + "step": 11942 + }, + { + "epoch": 5.0, + "learning_rate": 6.50802993588465e-06, + "loss": 0.1587, + "step": 11943 + }, + { + "epoch": 5.0, + "learning_rate": 6.506441304321012e-06, + "loss": 0.1663, + "step": 11944 + }, + { + "epoch": 5.0, + "learning_rate": 6.5048527731766155e-06, + "loss": 0.1213, + "step": 11945 + }, + { + "epoch": 5.0, + "learning_rate": 6.503264342497127e-06, + "loss": 0.1663, + "step": 11946 + }, + { + "epoch": 5.0, + "learning_rate": 6.501676012328202e-06, + "loss": 0.1341, + "step": 11947 + }, + { + "epoch": 5.0, + "learning_rate": 6.500087782715494e-06, + "loss": 0.1334, + "step": 11948 + }, + { + "epoch": 5.0, + "learning_rate": 6.498499653704656e-06, + "loss": 0.096, + "step": 11949 + }, + { + "epoch": 5.0, + "learning_rate": 6.496911625341342e-06, + "loss": 0.1041, + "step": 11950 + }, + { + "epoch": 5.0, + "learning_rate": 6.495323697671192e-06, + "loss": 0.1372, + "step": 11951 + }, + { + "epoch": 5.0, + "learning_rate": 6.493735870739852e-06, + "loss": 0.1214, + "step": 11952 + }, + { + "epoch": 5.0, + "learning_rate": 6.492148144592962e-06, + "loss": 0.1367, + "step": 11953 + }, + { + "epoch": 5.0, + "learning_rate": 6.4905605192761656e-06, + "loss": 0.1148, + "step": 11954 + }, + { + "epoch": 5.01, + "learning_rate": 6.488972994835089e-06, + "loss": 0.144, + "step": 11955 + }, + { + "epoch": 5.01, + "learning_rate": 6.4873855713153675e-06, + "loss": 0.1576, + "step": 11956 + }, + { + "epoch": 5.01, + "learning_rate": 6.485798248762634e-06, + "loss": 0.1223, + "step": 11957 + }, + { + "epoch": 5.01, + "learning_rate": 6.484211027222513e-06, + "loss": 0.1417, + "step": 11958 + }, + { + "epoch": 5.01, + "learning_rate": 6.4826239067406216e-06, + "loss": 0.1227, + "step": 11959 + }, + { + "epoch": 5.01, + "learning_rate": 6.481036887362586e-06, + "loss": 0.1256, + "step": 11960 + }, + { + "epoch": 5.01, + "learning_rate": 6.479449969134024e-06, + "loss": 0.1219, + "step": 11961 + }, + { + "epoch": 5.01, + "learning_rate": 6.477863152100545e-06, + "loss": 0.128, + "step": 11962 + }, + { + "epoch": 5.01, + "learning_rate": 6.476276436307767e-06, + "loss": 0.1176, + "step": 11963 + }, + { + "epoch": 5.01, + "learning_rate": 6.474689821801295e-06, + "loss": 0.1425, + "step": 11964 + }, + { + "epoch": 5.01, + "learning_rate": 6.473103308626736e-06, + "loss": 0.0925, + "step": 11965 + }, + { + "epoch": 5.01, + "learning_rate": 6.471516896829692e-06, + "loss": 0.1364, + "step": 11966 + }, + { + "epoch": 5.01, + "learning_rate": 6.469930586455764e-06, + "loss": 0.1362, + "step": 11967 + }, + { + "epoch": 5.01, + "learning_rate": 6.468344377550549e-06, + "loss": 0.1267, + "step": 11968 + }, + { + "epoch": 5.01, + "learning_rate": 6.466758270159641e-06, + "loss": 0.156, + "step": 11969 + }, + { + "epoch": 5.01, + "learning_rate": 6.46517226432863e-06, + "loss": 0.1238, + "step": 11970 + }, + { + "epoch": 5.01, + "learning_rate": 6.4635863601031075e-06, + "loss": 0.1058, + "step": 11971 + }, + { + "epoch": 5.01, + "learning_rate": 6.462000557528656e-06, + "loss": 0.1324, + "step": 11972 + }, + { + "epoch": 5.01, + "learning_rate": 6.460414856650858e-06, + "loss": 0.0875, + "step": 11973 + }, + { + "epoch": 5.01, + "learning_rate": 6.4588292575152955e-06, + "loss": 0.1066, + "step": 11974 + }, + { + "epoch": 5.01, + "learning_rate": 6.4572437601675415e-06, + "loss": 0.1413, + "step": 11975 + }, + { + "epoch": 5.01, + "learning_rate": 6.455658364653174e-06, + "loss": 0.1057, + "step": 11976 + }, + { + "epoch": 5.01, + "learning_rate": 6.4540730710177614e-06, + "loss": 0.1148, + "step": 11977 + }, + { + "epoch": 5.01, + "learning_rate": 6.452487879306871e-06, + "loss": 0.1001, + "step": 11978 + }, + { + "epoch": 5.02, + "learning_rate": 6.450902789566068e-06, + "loss": 0.1373, + "step": 11979 + }, + { + "epoch": 5.02, + "learning_rate": 6.44931780184092e-06, + "loss": 0.107, + "step": 11980 + }, + { + "epoch": 5.02, + "learning_rate": 6.447732916176975e-06, + "loss": 0.1356, + "step": 11981 + }, + { + "epoch": 5.02, + "learning_rate": 6.446148132619798e-06, + "loss": 0.1585, + "step": 11982 + }, + { + "epoch": 5.02, + "learning_rate": 6.444563451214938e-06, + "loss": 0.1395, + "step": 11983 + }, + { + "epoch": 5.02, + "learning_rate": 6.442978872007946e-06, + "loss": 0.1106, + "step": 11984 + }, + { + "epoch": 5.02, + "learning_rate": 6.441394395044372e-06, + "loss": 0.1381, + "step": 11985 + }, + { + "epoch": 5.02, + "learning_rate": 6.439810020369755e-06, + "loss": 0.0928, + "step": 11986 + }, + { + "epoch": 5.02, + "learning_rate": 6.4382257480296405e-06, + "loss": 0.1299, + "step": 11987 + }, + { + "epoch": 5.02, + "learning_rate": 6.436641578069569e-06, + "loss": 0.1425, + "step": 11988 + }, + { + "epoch": 5.02, + "learning_rate": 6.435057510535071e-06, + "loss": 0.1285, + "step": 11989 + }, + { + "epoch": 5.02, + "learning_rate": 6.433473545471681e-06, + "loss": 0.1113, + "step": 11990 + }, + { + "epoch": 5.02, + "learning_rate": 6.431889682924932e-06, + "loss": 0.1415, + "step": 11991 + }, + { + "epoch": 5.02, + "learning_rate": 6.430305922940346e-06, + "loss": 0.1447, + "step": 11992 + }, + { + "epoch": 5.02, + "learning_rate": 6.428722265563446e-06, + "loss": 0.1038, + "step": 11993 + }, + { + "epoch": 5.02, + "learning_rate": 6.42713871083976e-06, + "loss": 0.1033, + "step": 11994 + }, + { + "epoch": 5.02, + "learning_rate": 6.4255552588147975e-06, + "loss": 0.1294, + "step": 11995 + }, + { + "epoch": 5.02, + "learning_rate": 6.423971909534077e-06, + "loss": 0.0996, + "step": 11996 + }, + { + "epoch": 5.02, + "learning_rate": 6.422388663043112e-06, + "loss": 0.1159, + "step": 11997 + }, + { + "epoch": 5.02, + "learning_rate": 6.420805519387412e-06, + "loss": 0.1138, + "step": 11998 + }, + { + "epoch": 5.02, + "learning_rate": 6.419222478612479e-06, + "loss": 0.0632, + "step": 11999 + }, + { + "epoch": 5.02, + "learning_rate": 6.417639540763818e-06, + "loss": 0.1196, + "step": 12000 + }, + { + "epoch": 5.02, + "learning_rate": 6.416056705886932e-06, + "loss": 0.1107, + "step": 12001 + }, + { + "epoch": 5.02, + "learning_rate": 6.414473974027313e-06, + "loss": 0.1428, + "step": 12002 + }, + { + "epoch": 5.03, + "learning_rate": 6.412891345230462e-06, + "loss": 0.1136, + "step": 12003 + }, + { + "epoch": 5.03, + "learning_rate": 6.411308819541863e-06, + "loss": 0.1047, + "step": 12004 + }, + { + "epoch": 5.03, + "learning_rate": 6.409726397007008e-06, + "loss": 0.1103, + "step": 12005 + }, + { + "epoch": 5.03, + "learning_rate": 6.408144077671381e-06, + "loss": 0.1787, + "step": 12006 + }, + { + "epoch": 5.03, + "learning_rate": 6.406561861580465e-06, + "loss": 0.0751, + "step": 12007 + }, + { + "epoch": 5.03, + "learning_rate": 6.404979748779738e-06, + "loss": 0.1224, + "step": 12008 + }, + { + "epoch": 5.03, + "learning_rate": 6.403397739314682e-06, + "loss": 0.1314, + "step": 12009 + }, + { + "epoch": 5.03, + "learning_rate": 6.4018158332307635e-06, + "loss": 0.1615, + "step": 12010 + }, + { + "epoch": 5.03, + "learning_rate": 6.400234030573456e-06, + "loss": 0.1406, + "step": 12011 + }, + { + "epoch": 5.03, + "learning_rate": 6.3986523313882285e-06, + "loss": 0.0791, + "step": 12012 + }, + { + "epoch": 5.03, + "learning_rate": 6.397070735720543e-06, + "loss": 0.1291, + "step": 12013 + }, + { + "epoch": 5.03, + "learning_rate": 6.395489243615861e-06, + "loss": 0.1109, + "step": 12014 + }, + { + "epoch": 5.03, + "learning_rate": 6.393907855119645e-06, + "loss": 0.1138, + "step": 12015 + }, + { + "epoch": 5.03, + "learning_rate": 6.392326570277345e-06, + "loss": 0.136, + "step": 12016 + }, + { + "epoch": 5.03, + "learning_rate": 6.390745389134416e-06, + "loss": 0.1366, + "step": 12017 + }, + { + "epoch": 5.03, + "learning_rate": 6.389164311736311e-06, + "loss": 0.1381, + "step": 12018 + }, + { + "epoch": 5.03, + "learning_rate": 6.387583338128471e-06, + "loss": 0.1289, + "step": 12019 + }, + { + "epoch": 5.03, + "learning_rate": 6.386002468356343e-06, + "loss": 0.1132, + "step": 12020 + }, + { + "epoch": 5.03, + "learning_rate": 6.384421702465369e-06, + "loss": 0.1303, + "step": 12021 + }, + { + "epoch": 5.03, + "learning_rate": 6.382841040500982e-06, + "loss": 0.1251, + "step": 12022 + }, + { + "epoch": 5.03, + "learning_rate": 6.38126048250862e-06, + "loss": 0.1139, + "step": 12023 + }, + { + "epoch": 5.03, + "learning_rate": 6.379680028533714e-06, + "loss": 0.0976, + "step": 12024 + }, + { + "epoch": 5.03, + "learning_rate": 6.378099678621697e-06, + "loss": 0.1316, + "step": 12025 + }, + { + "epoch": 5.03, + "learning_rate": 6.376519432817989e-06, + "loss": 0.1245, + "step": 12026 + }, + { + "epoch": 5.04, + "learning_rate": 6.374939291168012e-06, + "loss": 0.1231, + "step": 12027 + }, + { + "epoch": 5.04, + "learning_rate": 6.373359253717188e-06, + "loss": 0.1713, + "step": 12028 + }, + { + "epoch": 5.04, + "learning_rate": 6.3717793205109354e-06, + "loss": 0.1427, + "step": 12029 + }, + { + "epoch": 5.04, + "learning_rate": 6.370199491594667e-06, + "loss": 0.1467, + "step": 12030 + }, + { + "epoch": 5.04, + "learning_rate": 6.368619767013792e-06, + "loss": 0.1034, + "step": 12031 + }, + { + "epoch": 5.04, + "learning_rate": 6.367040146813721e-06, + "loss": 0.1073, + "step": 12032 + }, + { + "epoch": 5.04, + "learning_rate": 6.365460631039855e-06, + "loss": 0.1462, + "step": 12033 + }, + { + "epoch": 5.04, + "learning_rate": 6.363881219737599e-06, + "loss": 0.1324, + "step": 12034 + }, + { + "epoch": 5.04, + "learning_rate": 6.362301912952351e-06, + "loss": 0.1383, + "step": 12035 + }, + { + "epoch": 5.04, + "learning_rate": 6.360722710729506e-06, + "loss": 0.1589, + "step": 12036 + }, + { + "epoch": 5.04, + "learning_rate": 6.3591436131144565e-06, + "loss": 0.1377, + "step": 12037 + }, + { + "epoch": 5.04, + "learning_rate": 6.357564620152595e-06, + "loss": 0.1345, + "step": 12038 + }, + { + "epoch": 5.04, + "learning_rate": 6.355985731889305e-06, + "loss": 0.1272, + "step": 12039 + }, + { + "epoch": 5.04, + "learning_rate": 6.354406948369971e-06, + "loss": 0.1493, + "step": 12040 + }, + { + "epoch": 5.04, + "learning_rate": 6.352828269639974e-06, + "loss": 0.1311, + "step": 12041 + }, + { + "epoch": 5.04, + "learning_rate": 6.351249695744694e-06, + "loss": 0.1302, + "step": 12042 + }, + { + "epoch": 5.04, + "learning_rate": 6.349671226729502e-06, + "loss": 0.1074, + "step": 12043 + }, + { + "epoch": 5.04, + "learning_rate": 6.348092862639772e-06, + "loss": 0.1288, + "step": 12044 + }, + { + "epoch": 5.04, + "learning_rate": 6.346514603520873e-06, + "loss": 0.1185, + "step": 12045 + }, + { + "epoch": 5.04, + "learning_rate": 6.344936449418168e-06, + "loss": 0.1462, + "step": 12046 + }, + { + "epoch": 5.04, + "learning_rate": 6.343358400377021e-06, + "loss": 0.1116, + "step": 12047 + }, + { + "epoch": 5.04, + "learning_rate": 6.341780456442797e-06, + "loss": 0.1073, + "step": 12048 + }, + { + "epoch": 5.04, + "learning_rate": 6.340202617660842e-06, + "loss": 0.1362, + "step": 12049 + }, + { + "epoch": 5.05, + "learning_rate": 6.338624884076517e-06, + "loss": 0.0908, + "step": 12050 + }, + { + "epoch": 5.05, + "learning_rate": 6.337047255735169e-06, + "loss": 0.0934, + "step": 12051 + }, + { + "epoch": 5.05, + "learning_rate": 6.335469732682149e-06, + "loss": 0.1395, + "step": 12052 + }, + { + "epoch": 5.05, + "learning_rate": 6.3338923149628e-06, + "loss": 0.1196, + "step": 12053 + }, + { + "epoch": 5.05, + "learning_rate": 6.332315002622461e-06, + "loss": 0.1087, + "step": 12054 + }, + { + "epoch": 5.05, + "learning_rate": 6.330737795706474e-06, + "loss": 0.1016, + "step": 12055 + }, + { + "epoch": 5.05, + "learning_rate": 6.329160694260173e-06, + "loss": 0.1073, + "step": 12056 + }, + { + "epoch": 5.05, + "learning_rate": 6.3275836983288895e-06, + "loss": 0.1342, + "step": 12057 + }, + { + "epoch": 5.05, + "learning_rate": 6.326006807957954e-06, + "loss": 0.1287, + "step": 12058 + }, + { + "epoch": 5.05, + "learning_rate": 6.324430023192695e-06, + "loss": 0.1116, + "step": 12059 + }, + { + "epoch": 5.05, + "learning_rate": 6.32285334407843e-06, + "loss": 0.11, + "step": 12060 + }, + { + "epoch": 5.05, + "learning_rate": 6.3212767706604826e-06, + "loss": 0.1296, + "step": 12061 + }, + { + "epoch": 5.05, + "learning_rate": 6.319700302984173e-06, + "loss": 0.1233, + "step": 12062 + }, + { + "epoch": 5.05, + "learning_rate": 6.31812394109481e-06, + "loss": 0.0853, + "step": 12063 + }, + { + "epoch": 5.05, + "learning_rate": 6.316547685037707e-06, + "loss": 0.0856, + "step": 12064 + }, + { + "epoch": 5.05, + "learning_rate": 6.314971534858174e-06, + "loss": 0.1416, + "step": 12065 + }, + { + "epoch": 5.05, + "learning_rate": 6.313395490601513e-06, + "loss": 0.172, + "step": 12066 + }, + { + "epoch": 5.05, + "learning_rate": 6.311819552313026e-06, + "loss": 0.1376, + "step": 12067 + }, + { + "epoch": 5.05, + "learning_rate": 6.310243720038014e-06, + "loss": 0.1347, + "step": 12068 + }, + { + "epoch": 5.05, + "learning_rate": 6.308667993821773e-06, + "loss": 0.1733, + "step": 12069 + }, + { + "epoch": 5.05, + "learning_rate": 6.3070923737095946e-06, + "loss": 0.155, + "step": 12070 + }, + { + "epoch": 5.05, + "learning_rate": 6.305516859746772e-06, + "loss": 0.0885, + "step": 12071 + }, + { + "epoch": 5.05, + "learning_rate": 6.303941451978584e-06, + "loss": 0.1177, + "step": 12072 + }, + { + "epoch": 5.05, + "learning_rate": 6.302366150450323e-06, + "loss": 0.1401, + "step": 12073 + }, + { + "epoch": 5.06, + "learning_rate": 6.300790955207262e-06, + "loss": 0.1523, + "step": 12074 + }, + { + "epoch": 5.06, + "learning_rate": 6.299215866294683e-06, + "loss": 0.128, + "step": 12075 + }, + { + "epoch": 5.06, + "learning_rate": 6.297640883757863e-06, + "loss": 0.1039, + "step": 12076 + }, + { + "epoch": 5.06, + "learning_rate": 6.296066007642067e-06, + "loss": 0.1119, + "step": 12077 + }, + { + "epoch": 5.06, + "learning_rate": 6.294491237992566e-06, + "loss": 0.1343, + "step": 12078 + }, + { + "epoch": 5.06, + "learning_rate": 6.292916574854628e-06, + "loss": 0.1404, + "step": 12079 + }, + { + "epoch": 5.06, + "learning_rate": 6.291342018273514e-06, + "loss": 0.1355, + "step": 12080 + }, + { + "epoch": 5.06, + "learning_rate": 6.2897675682944824e-06, + "loss": 0.0829, + "step": 12081 + }, + { + "epoch": 5.06, + "learning_rate": 6.288193224962789e-06, + "loss": 0.0669, + "step": 12082 + }, + { + "epoch": 5.06, + "learning_rate": 6.2866189883236885e-06, + "loss": 0.1222, + "step": 12083 + }, + { + "epoch": 5.06, + "learning_rate": 6.285044858422429e-06, + "loss": 0.1228, + "step": 12084 + }, + { + "epoch": 5.06, + "learning_rate": 6.283470835304257e-06, + "loss": 0.116, + "step": 12085 + }, + { + "epoch": 5.06, + "learning_rate": 6.281896919014422e-06, + "loss": 0.12, + "step": 12086 + }, + { + "epoch": 5.06, + "learning_rate": 6.280323109598157e-06, + "loss": 0.138, + "step": 12087 + }, + { + "epoch": 5.06, + "learning_rate": 6.278749407100704e-06, + "loss": 0.1307, + "step": 12088 + }, + { + "epoch": 5.06, + "learning_rate": 6.2771758115673e-06, + "loss": 0.0881, + "step": 12089 + }, + { + "epoch": 5.06, + "learning_rate": 6.2756023230431705e-06, + "loss": 0.1232, + "step": 12090 + }, + { + "epoch": 5.06, + "learning_rate": 6.2740289415735474e-06, + "loss": 0.1266, + "step": 12091 + }, + { + "epoch": 5.06, + "learning_rate": 6.272455667203659e-06, + "loss": 0.1358, + "step": 12092 + }, + { + "epoch": 5.06, + "learning_rate": 6.270882499978723e-06, + "loss": 0.1208, + "step": 12093 + }, + { + "epoch": 5.06, + "learning_rate": 6.269309439943963e-06, + "loss": 0.1235, + "step": 12094 + }, + { + "epoch": 5.06, + "learning_rate": 6.2677364871445886e-06, + "loss": 0.1502, + "step": 12095 + }, + { + "epoch": 5.06, + "learning_rate": 6.2661636416258175e-06, + "loss": 0.1322, + "step": 12096 + }, + { + "epoch": 5.06, + "learning_rate": 6.264590903432861e-06, + "loss": 0.0914, + "step": 12097 + }, + { + "epoch": 5.07, + "learning_rate": 6.263018272610921e-06, + "loss": 0.123, + "step": 12098 + }, + { + "epoch": 5.07, + "learning_rate": 6.261445749205206e-06, + "loss": 0.1693, + "step": 12099 + }, + { + "epoch": 5.07, + "learning_rate": 6.259873333260917e-06, + "loss": 0.1072, + "step": 12100 + }, + { + "epoch": 5.07, + "learning_rate": 6.258301024823249e-06, + "loss": 0.0645, + "step": 12101 + }, + { + "epoch": 5.07, + "learning_rate": 6.256728823937397e-06, + "loss": 0.1359, + "step": 12102 + }, + { + "epoch": 5.07, + "learning_rate": 6.255156730648555e-06, + "loss": 0.1382, + "step": 12103 + }, + { + "epoch": 5.07, + "learning_rate": 6.253584745001908e-06, + "loss": 0.1039, + "step": 12104 + }, + { + "epoch": 5.07, + "learning_rate": 6.252012867042643e-06, + "loss": 0.1308, + "step": 12105 + }, + { + "epoch": 5.07, + "learning_rate": 6.250441096815945e-06, + "loss": 0.1262, + "step": 12106 + }, + { + "epoch": 5.07, + "learning_rate": 6.248869434366987e-06, + "loss": 0.1414, + "step": 12107 + }, + { + "epoch": 5.07, + "learning_rate": 6.247297879740948e-06, + "loss": 0.1307, + "step": 12108 + }, + { + "epoch": 5.07, + "learning_rate": 6.245726432983003e-06, + "loss": 0.1036, + "step": 12109 + }, + { + "epoch": 5.07, + "learning_rate": 6.244155094138322e-06, + "loss": 0.152, + "step": 12110 + }, + { + "epoch": 5.07, + "learning_rate": 6.2425838632520674e-06, + "loss": 0.1469, + "step": 12111 + }, + { + "epoch": 5.07, + "learning_rate": 6.2410127403694056e-06, + "loss": 0.1288, + "step": 12112 + }, + { + "epoch": 5.07, + "learning_rate": 6.239441725535498e-06, + "loss": 0.0963, + "step": 12113 + }, + { + "epoch": 5.07, + "learning_rate": 6.237870818795499e-06, + "loss": 0.0882, + "step": 12114 + }, + { + "epoch": 5.07, + "learning_rate": 6.2363000201945656e-06, + "loss": 0.1074, + "step": 12115 + }, + { + "epoch": 5.07, + "learning_rate": 6.234729329777853e-06, + "loss": 0.1028, + "step": 12116 + }, + { + "epoch": 5.07, + "learning_rate": 6.233158747590502e-06, + "loss": 0.1171, + "step": 12117 + }, + { + "epoch": 5.07, + "learning_rate": 6.231588273677658e-06, + "loss": 0.1217, + "step": 12118 + }, + { + "epoch": 5.07, + "learning_rate": 6.230017908084465e-06, + "loss": 0.1087, + "step": 12119 + }, + { + "epoch": 5.07, + "learning_rate": 6.228447650856062e-06, + "loss": 0.1036, + "step": 12120 + }, + { + "epoch": 5.07, + "learning_rate": 6.226877502037585e-06, + "loss": 0.0812, + "step": 12121 + }, + { + "epoch": 5.08, + "learning_rate": 6.225307461674166e-06, + "loss": 0.1108, + "step": 12122 + }, + { + "epoch": 5.08, + "learning_rate": 6.2237375298109336e-06, + "loss": 0.1215, + "step": 12123 + }, + { + "epoch": 5.08, + "learning_rate": 6.222167706493016e-06, + "loss": 0.1526, + "step": 12124 + }, + { + "epoch": 5.08, + "learning_rate": 6.220597991765536e-06, + "loss": 0.1058, + "step": 12125 + }, + { + "epoch": 5.08, + "learning_rate": 6.2190283856736115e-06, + "loss": 0.1068, + "step": 12126 + }, + { + "epoch": 5.08, + "learning_rate": 6.217458888262362e-06, + "loss": 0.1346, + "step": 12127 + }, + { + "epoch": 5.08, + "learning_rate": 6.215889499576898e-06, + "loss": 0.1367, + "step": 12128 + }, + { + "epoch": 5.08, + "learning_rate": 6.214320219662333e-06, + "loss": 0.1279, + "step": 12129 + }, + { + "epoch": 5.08, + "learning_rate": 6.212751048563776e-06, + "loss": 0.1141, + "step": 12130 + }, + { + "epoch": 5.08, + "learning_rate": 6.2111819863263275e-06, + "loss": 0.1321, + "step": 12131 + }, + { + "epoch": 5.08, + "learning_rate": 6.209613032995091e-06, + "loss": 0.1796, + "step": 12132 + }, + { + "epoch": 5.08, + "learning_rate": 6.208044188615166e-06, + "loss": 0.1399, + "step": 12133 + }, + { + "epoch": 5.08, + "learning_rate": 6.206475453231644e-06, + "loss": 0.1253, + "step": 12134 + }, + { + "epoch": 5.08, + "learning_rate": 6.204906826889618e-06, + "loss": 0.1111, + "step": 12135 + }, + { + "epoch": 5.08, + "learning_rate": 6.203338309634179e-06, + "loss": 0.1165, + "step": 12136 + }, + { + "epoch": 5.08, + "learning_rate": 6.201769901510412e-06, + "loss": 0.1071, + "step": 12137 + }, + { + "epoch": 5.08, + "learning_rate": 6.200201602563397e-06, + "loss": 0.1172, + "step": 12138 + }, + { + "epoch": 5.08, + "learning_rate": 6.1986334128382195e-06, + "loss": 0.1291, + "step": 12139 + }, + { + "epoch": 5.08, + "learning_rate": 6.197065332379947e-06, + "loss": 0.117, + "step": 12140 + }, + { + "epoch": 5.08, + "learning_rate": 6.195497361233658e-06, + "loss": 0.1428, + "step": 12141 + }, + { + "epoch": 5.08, + "learning_rate": 6.19392949944442e-06, + "loss": 0.1521, + "step": 12142 + }, + { + "epoch": 5.08, + "learning_rate": 6.192361747057301e-06, + "loss": 0.128, + "step": 12143 + }, + { + "epoch": 5.08, + "learning_rate": 6.190794104117368e-06, + "loss": 0.1004, + "step": 12144 + }, + { + "epoch": 5.08, + "learning_rate": 6.189226570669675e-06, + "loss": 0.1502, + "step": 12145 + }, + { + "epoch": 5.09, + "learning_rate": 6.187659146759282e-06, + "loss": 0.1702, + "step": 12146 + }, + { + "epoch": 5.09, + "learning_rate": 6.186091832431247e-06, + "loss": 0.1418, + "step": 12147 + }, + { + "epoch": 5.09, + "learning_rate": 6.1845246277306145e-06, + "loss": 0.1337, + "step": 12148 + }, + { + "epoch": 5.09, + "learning_rate": 6.182957532702437e-06, + "loss": 0.1371, + "step": 12149 + }, + { + "epoch": 5.09, + "learning_rate": 6.181390547391758e-06, + "loss": 0.1352, + "step": 12150 + }, + { + "epoch": 5.09, + "learning_rate": 6.179823671843621e-06, + "loss": 0.138, + "step": 12151 + }, + { + "epoch": 5.09, + "learning_rate": 6.178256906103061e-06, + "loss": 0.1226, + "step": 12152 + }, + { + "epoch": 5.09, + "learning_rate": 6.176690250215114e-06, + "loss": 0.1264, + "step": 12153 + }, + { + "epoch": 5.09, + "learning_rate": 6.175123704224814e-06, + "loss": 0.1309, + "step": 12154 + }, + { + "epoch": 5.09, + "learning_rate": 6.173557268177189e-06, + "loss": 0.135, + "step": 12155 + }, + { + "epoch": 5.09, + "learning_rate": 6.171990942117265e-06, + "loss": 0.1154, + "step": 12156 + }, + { + "epoch": 5.09, + "learning_rate": 6.170424726090065e-06, + "loss": 0.1521, + "step": 12157 + }, + { + "epoch": 5.09, + "learning_rate": 6.168858620140606e-06, + "loss": 0.1361, + "step": 12158 + }, + { + "epoch": 5.09, + "learning_rate": 6.167292624313908e-06, + "loss": 0.1125, + "step": 12159 + }, + { + "epoch": 5.09, + "learning_rate": 6.165726738654982e-06, + "loss": 0.0856, + "step": 12160 + }, + { + "epoch": 5.09, + "learning_rate": 6.16416096320884e-06, + "loss": 0.1175, + "step": 12161 + }, + { + "epoch": 5.09, + "learning_rate": 6.162595298020487e-06, + "loss": 0.0999, + "step": 12162 + }, + { + "epoch": 5.09, + "learning_rate": 6.1610297431349255e-06, + "loss": 0.1059, + "step": 12163 + }, + { + "epoch": 5.09, + "learning_rate": 6.1594642985971565e-06, + "loss": 0.1038, + "step": 12164 + }, + { + "epoch": 5.09, + "learning_rate": 6.15789896445218e-06, + "loss": 0.1068, + "step": 12165 + }, + { + "epoch": 5.09, + "learning_rate": 6.1563337407449865e-06, + "loss": 0.1114, + "step": 12166 + }, + { + "epoch": 5.09, + "learning_rate": 6.15476862752057e-06, + "loss": 0.1109, + "step": 12167 + }, + { + "epoch": 5.09, + "learning_rate": 6.153203624823918e-06, + "loss": 0.1219, + "step": 12168 + }, + { + "epoch": 5.09, + "learning_rate": 6.151638732700012e-06, + "loss": 0.1144, + "step": 12169 + }, + { + "epoch": 5.1, + "learning_rate": 6.150073951193836e-06, + "loss": 0.1246, + "step": 12170 + }, + { + "epoch": 5.1, + "learning_rate": 6.1485092803503695e-06, + "loss": 0.1424, + "step": 12171 + }, + { + "epoch": 5.1, + "learning_rate": 6.146944720214585e-06, + "loss": 0.1515, + "step": 12172 + }, + { + "epoch": 5.1, + "learning_rate": 6.145380270831455e-06, + "loss": 0.1316, + "step": 12173 + }, + { + "epoch": 5.1, + "learning_rate": 6.143815932245951e-06, + "loss": 0.1069, + "step": 12174 + }, + { + "epoch": 5.1, + "learning_rate": 6.142251704503033e-06, + "loss": 0.1006, + "step": 12175 + }, + { + "epoch": 5.1, + "learning_rate": 6.140687587647667e-06, + "loss": 0.0906, + "step": 12176 + }, + { + "epoch": 5.1, + "learning_rate": 6.139123581724813e-06, + "loss": 0.0751, + "step": 12177 + }, + { + "epoch": 5.1, + "learning_rate": 6.1375596867794245e-06, + "loss": 0.1319, + "step": 12178 + }, + { + "epoch": 5.1, + "learning_rate": 6.135995902856455e-06, + "loss": 0.0637, + "step": 12179 + }, + { + "epoch": 5.1, + "learning_rate": 6.134432230000854e-06, + "loss": 0.1416, + "step": 12180 + }, + { + "epoch": 5.1, + "learning_rate": 6.132868668257571e-06, + "loss": 0.1236, + "step": 12181 + }, + { + "epoch": 5.1, + "learning_rate": 6.131305217671544e-06, + "loss": 0.0791, + "step": 12182 + }, + { + "epoch": 5.1, + "learning_rate": 6.129741878287716e-06, + "loss": 0.106, + "step": 12183 + }, + { + "epoch": 5.1, + "learning_rate": 6.128178650151026e-06, + "loss": 0.1149, + "step": 12184 + }, + { + "epoch": 5.1, + "learning_rate": 6.126615533306405e-06, + "loss": 0.131, + "step": 12185 + }, + { + "epoch": 5.1, + "learning_rate": 6.12505252779878e-06, + "loss": 0.1268, + "step": 12186 + }, + { + "epoch": 5.1, + "learning_rate": 6.123489633673082e-06, + "loss": 0.0994, + "step": 12187 + }, + { + "epoch": 5.1, + "learning_rate": 6.121926850974236e-06, + "loss": 0.1267, + "step": 12188 + }, + { + "epoch": 5.1, + "learning_rate": 6.120364179747161e-06, + "loss": 0.1149, + "step": 12189 + }, + { + "epoch": 5.1, + "learning_rate": 6.1188016200367715e-06, + "loss": 0.1255, + "step": 12190 + }, + { + "epoch": 5.1, + "learning_rate": 6.117239171887987e-06, + "loss": 0.1498, + "step": 12191 + }, + { + "epoch": 5.1, + "learning_rate": 6.11567683534572e-06, + "loss": 0.106, + "step": 12192 + }, + { + "epoch": 5.1, + "learning_rate": 6.1141146104548735e-06, + "loss": 0.1239, + "step": 12193 + }, + { + "epoch": 5.11, + "learning_rate": 6.112552497260353e-06, + "loss": 0.0943, + "step": 12194 + }, + { + "epoch": 5.11, + "learning_rate": 6.110990495807064e-06, + "loss": 0.1374, + "step": 12195 + }, + { + "epoch": 5.11, + "learning_rate": 6.1094286061399e-06, + "loss": 0.0907, + "step": 12196 + }, + { + "epoch": 5.11, + "learning_rate": 6.107866828303759e-06, + "loss": 0.1415, + "step": 12197 + }, + { + "epoch": 5.11, + "learning_rate": 6.106305162343533e-06, + "loss": 0.0777, + "step": 12198 + }, + { + "epoch": 5.11, + "learning_rate": 6.104743608304109e-06, + "loss": 0.1269, + "step": 12199 + }, + { + "epoch": 5.11, + "learning_rate": 6.103182166230373e-06, + "loss": 0.1295, + "step": 12200 + }, + { + "epoch": 5.11, + "learning_rate": 6.101620836167208e-06, + "loss": 0.1536, + "step": 12201 + }, + { + "epoch": 5.11, + "learning_rate": 6.100059618159493e-06, + "loss": 0.1431, + "step": 12202 + }, + { + "epoch": 5.11, + "learning_rate": 6.098498512252102e-06, + "loss": 0.0914, + "step": 12203 + }, + { + "epoch": 5.11, + "learning_rate": 6.096937518489912e-06, + "loss": 0.1229, + "step": 12204 + }, + { + "epoch": 5.11, + "learning_rate": 6.095376636917787e-06, + "loss": 0.124, + "step": 12205 + }, + { + "epoch": 5.11, + "learning_rate": 6.093815867580597e-06, + "loss": 0.1116, + "step": 12206 + }, + { + "epoch": 5.11, + "learning_rate": 6.0922552105232055e-06, + "loss": 0.1509, + "step": 12207 + }, + { + "epoch": 5.11, + "learning_rate": 6.0906946657904685e-06, + "loss": 0.1288, + "step": 12208 + }, + { + "epoch": 5.11, + "learning_rate": 6.089134233427245e-06, + "loss": 0.1412, + "step": 12209 + }, + { + "epoch": 5.11, + "learning_rate": 6.087573913478386e-06, + "loss": 0.1191, + "step": 12210 + }, + { + "epoch": 5.11, + "learning_rate": 6.086013705988742e-06, + "loss": 0.1093, + "step": 12211 + }, + { + "epoch": 5.11, + "learning_rate": 6.084453611003164e-06, + "loss": 0.0971, + "step": 12212 + }, + { + "epoch": 5.11, + "learning_rate": 6.08289362856649e-06, + "loss": 0.0934, + "step": 12213 + }, + { + "epoch": 5.11, + "learning_rate": 6.081333758723562e-06, + "loss": 0.1281, + "step": 12214 + }, + { + "epoch": 5.11, + "learning_rate": 6.079774001519221e-06, + "loss": 0.1253, + "step": 12215 + }, + { + "epoch": 5.11, + "learning_rate": 6.078214356998294e-06, + "loss": 0.1323, + "step": 12216 + }, + { + "epoch": 5.11, + "learning_rate": 6.076654825205617e-06, + "loss": 0.154, + "step": 12217 + }, + { + "epoch": 5.12, + "learning_rate": 6.075095406186015e-06, + "loss": 0.0858, + "step": 12218 + }, + { + "epoch": 5.12, + "learning_rate": 6.073536099984314e-06, + "loss": 0.1258, + "step": 12219 + }, + { + "epoch": 5.12, + "learning_rate": 6.071976906645332e-06, + "loss": 0.1146, + "step": 12220 + }, + { + "epoch": 5.12, + "learning_rate": 6.070417826213888e-06, + "loss": 0.1333, + "step": 12221 + }, + { + "epoch": 5.12, + "learning_rate": 6.068858858734799e-06, + "loss": 0.096, + "step": 12222 + }, + { + "epoch": 5.12, + "learning_rate": 6.06730000425287e-06, + "loss": 0.1071, + "step": 12223 + }, + { + "epoch": 5.12, + "learning_rate": 6.065741262812914e-06, + "loss": 0.1269, + "step": 12224 + }, + { + "epoch": 5.12, + "learning_rate": 6.064182634459735e-06, + "loss": 0.1314, + "step": 12225 + }, + { + "epoch": 5.12, + "learning_rate": 6.062624119238133e-06, + "loss": 0.1292, + "step": 12226 + }, + { + "epoch": 5.12, + "learning_rate": 6.061065717192907e-06, + "loss": 0.087, + "step": 12227 + }, + { + "epoch": 5.12, + "learning_rate": 6.059507428368852e-06, + "loss": 0.1242, + "step": 12228 + }, + { + "epoch": 5.12, + "learning_rate": 6.057949252810762e-06, + "loss": 0.1421, + "step": 12229 + }, + { + "epoch": 5.12, + "learning_rate": 6.056391190563419e-06, + "loss": 0.1274, + "step": 12230 + }, + { + "epoch": 5.12, + "learning_rate": 6.054833241671611e-06, + "loss": 0.1439, + "step": 12231 + }, + { + "epoch": 5.12, + "learning_rate": 6.053275406180122e-06, + "loss": 0.1193, + "step": 12232 + }, + { + "epoch": 5.12, + "learning_rate": 6.051717684133729e-06, + "loss": 0.1221, + "step": 12233 + }, + { + "epoch": 5.12, + "learning_rate": 6.0501600755772074e-06, + "loss": 0.1642, + "step": 12234 + }, + { + "epoch": 5.12, + "learning_rate": 6.048602580555328e-06, + "loss": 0.1479, + "step": 12235 + }, + { + "epoch": 5.12, + "learning_rate": 6.047045199112865e-06, + "loss": 0.1157, + "step": 12236 + }, + { + "epoch": 5.12, + "learning_rate": 6.0454879312945755e-06, + "loss": 0.1253, + "step": 12237 + }, + { + "epoch": 5.12, + "learning_rate": 6.043930777145228e-06, + "loss": 0.1043, + "step": 12238 + }, + { + "epoch": 5.12, + "learning_rate": 6.042373736709581e-06, + "loss": 0.0997, + "step": 12239 + }, + { + "epoch": 5.12, + "learning_rate": 6.040816810032387e-06, + "loss": 0.1033, + "step": 12240 + }, + { + "epoch": 5.12, + "learning_rate": 6.039259997158401e-06, + "loss": 0.1221, + "step": 12241 + }, + { + "epoch": 5.13, + "learning_rate": 6.037703298132373e-06, + "loss": 0.1313, + "step": 12242 + }, + { + "epoch": 5.13, + "learning_rate": 6.036146712999047e-06, + "loss": 0.1127, + "step": 12243 + }, + { + "epoch": 5.13, + "learning_rate": 6.0345902418031645e-06, + "loss": 0.0897, + "step": 12244 + }, + { + "epoch": 5.13, + "learning_rate": 6.03303388458947e-06, + "loss": 0.1243, + "step": 12245 + }, + { + "epoch": 5.13, + "learning_rate": 6.031477641402693e-06, + "loss": 0.1178, + "step": 12246 + }, + { + "epoch": 5.13, + "learning_rate": 6.029921512287571e-06, + "loss": 0.0931, + "step": 12247 + }, + { + "epoch": 5.13, + "learning_rate": 6.028365497288831e-06, + "loss": 0.1232, + "step": 12248 + }, + { + "epoch": 5.13, + "learning_rate": 6.026809596451203e-06, + "loss": 0.099, + "step": 12249 + }, + { + "epoch": 5.13, + "learning_rate": 6.025253809819405e-06, + "loss": 0.0836, + "step": 12250 + }, + { + "epoch": 5.13, + "learning_rate": 6.023698137438159e-06, + "loss": 0.1112, + "step": 12251 + }, + { + "epoch": 5.13, + "learning_rate": 6.022142579352186e-06, + "loss": 0.0906, + "step": 12252 + }, + { + "epoch": 5.13, + "learning_rate": 6.020587135606193e-06, + "loss": 0.1082, + "step": 12253 + }, + { + "epoch": 5.13, + "learning_rate": 6.019031806244888e-06, + "loss": 0.0938, + "step": 12254 + }, + { + "epoch": 5.13, + "learning_rate": 6.017476591312983e-06, + "loss": 0.1267, + "step": 12255 + }, + { + "epoch": 5.13, + "learning_rate": 6.015921490855181e-06, + "loss": 0.1103, + "step": 12256 + }, + { + "epoch": 5.13, + "learning_rate": 6.014366504916178e-06, + "loss": 0.1016, + "step": 12257 + }, + { + "epoch": 5.13, + "learning_rate": 6.012811633540672e-06, + "loss": 0.1498, + "step": 12258 + }, + { + "epoch": 5.13, + "learning_rate": 6.011256876773362e-06, + "loss": 0.1537, + "step": 12259 + }, + { + "epoch": 5.13, + "learning_rate": 6.009702234658929e-06, + "loss": 0.1215, + "step": 12260 + }, + { + "epoch": 5.13, + "learning_rate": 6.008147707242065e-06, + "loss": 0.1021, + "step": 12261 + }, + { + "epoch": 5.13, + "learning_rate": 6.0065932945674535e-06, + "loss": 0.1175, + "step": 12262 + }, + { + "epoch": 5.13, + "learning_rate": 6.005038996679775e-06, + "loss": 0.1038, + "step": 12263 + }, + { + "epoch": 5.13, + "learning_rate": 6.0034848136237035e-06, + "loss": 0.118, + "step": 12264 + }, + { + "epoch": 5.14, + "learning_rate": 6.001930745443916e-06, + "loss": 0.1179, + "step": 12265 + }, + { + "epoch": 5.14, + "learning_rate": 6.000376792185082e-06, + "loss": 0.1159, + "step": 12266 + }, + { + "epoch": 5.14, + "learning_rate": 5.998822953891866e-06, + "loss": 0.1377, + "step": 12267 + }, + { + "epoch": 5.14, + "learning_rate": 5.997269230608935e-06, + "loss": 0.167, + "step": 12268 + }, + { + "epoch": 5.14, + "learning_rate": 5.995715622380949e-06, + "loss": 0.1558, + "step": 12269 + }, + { + "epoch": 5.14, + "learning_rate": 5.994162129252561e-06, + "loss": 0.1122, + "step": 12270 + }, + { + "epoch": 5.14, + "learning_rate": 5.992608751268429e-06, + "loss": 0.0987, + "step": 12271 + }, + { + "epoch": 5.14, + "learning_rate": 5.9910554884732056e-06, + "loss": 0.0981, + "step": 12272 + }, + { + "epoch": 5.14, + "learning_rate": 5.989502340911531e-06, + "loss": 0.1281, + "step": 12273 + }, + { + "epoch": 5.14, + "learning_rate": 5.987949308628054e-06, + "loss": 0.1062, + "step": 12274 + }, + { + "epoch": 5.14, + "learning_rate": 5.986396391667416e-06, + "loss": 0.0926, + "step": 12275 + }, + { + "epoch": 5.14, + "learning_rate": 5.98484359007425e-06, + "loss": 0.1085, + "step": 12276 + }, + { + "epoch": 5.14, + "learning_rate": 5.983290903893193e-06, + "loss": 0.1109, + "step": 12277 + }, + { + "epoch": 5.14, + "learning_rate": 5.981738333168873e-06, + "loss": 0.0862, + "step": 12278 + }, + { + "epoch": 5.14, + "learning_rate": 5.980185877945918e-06, + "loss": 0.1289, + "step": 12279 + }, + { + "epoch": 5.14, + "learning_rate": 5.978633538268957e-06, + "loss": 0.1209, + "step": 12280 + }, + { + "epoch": 5.14, + "learning_rate": 5.977081314182602e-06, + "loss": 0.1088, + "step": 12281 + }, + { + "epoch": 5.14, + "learning_rate": 5.975529205731476e-06, + "loss": 0.1414, + "step": 12282 + }, + { + "epoch": 5.14, + "learning_rate": 5.973977212960194e-06, + "loss": 0.0954, + "step": 12283 + }, + { + "epoch": 5.14, + "learning_rate": 5.9724253359133614e-06, + "loss": 0.1335, + "step": 12284 + }, + { + "epoch": 5.14, + "learning_rate": 5.97087357463559e-06, + "loss": 0.1011, + "step": 12285 + }, + { + "epoch": 5.14, + "learning_rate": 5.969321929171482e-06, + "loss": 0.104, + "step": 12286 + }, + { + "epoch": 5.14, + "learning_rate": 5.967770399565638e-06, + "loss": 0.1147, + "step": 12287 + }, + { + "epoch": 5.14, + "learning_rate": 5.966218985862655e-06, + "loss": 0.1452, + "step": 12288 + }, + { + "epoch": 5.15, + "learning_rate": 5.9646676881071305e-06, + "loss": 0.1098, + "step": 12289 + }, + { + "epoch": 5.15, + "learning_rate": 5.963116506343649e-06, + "loss": 0.1594, + "step": 12290 + }, + { + "epoch": 5.15, + "learning_rate": 5.961565440616803e-06, + "loss": 0.1281, + "step": 12291 + }, + { + "epoch": 5.15, + "learning_rate": 5.9600144909711745e-06, + "loss": 0.1219, + "step": 12292 + }, + { + "epoch": 5.15, + "learning_rate": 5.958463657451345e-06, + "loss": 0.1546, + "step": 12293 + }, + { + "epoch": 5.15, + "learning_rate": 5.956912940101891e-06, + "loss": 0.1529, + "step": 12294 + }, + { + "epoch": 5.15, + "learning_rate": 5.9553623389673855e-06, + "loss": 0.126, + "step": 12295 + }, + { + "epoch": 5.15, + "learning_rate": 5.953811854092404e-06, + "loss": 0.1451, + "step": 12296 + }, + { + "epoch": 5.15, + "learning_rate": 5.952261485521511e-06, + "loss": 0.15, + "step": 12297 + }, + { + "epoch": 5.15, + "learning_rate": 5.950711233299265e-06, + "loss": 0.1165, + "step": 12298 + }, + { + "epoch": 5.15, + "learning_rate": 5.949161097470233e-06, + "loss": 0.1291, + "step": 12299 + }, + { + "epoch": 5.15, + "learning_rate": 5.947611078078972e-06, + "loss": 0.1399, + "step": 12300 + }, + { + "epoch": 5.15, + "learning_rate": 5.9460611751700325e-06, + "loss": 0.1133, + "step": 12301 + }, + { + "epoch": 5.15, + "learning_rate": 5.944511388787967e-06, + "loss": 0.1605, + "step": 12302 + }, + { + "epoch": 5.15, + "learning_rate": 5.942961718977323e-06, + "loss": 0.089, + "step": 12303 + }, + { + "epoch": 5.15, + "learning_rate": 5.941412165782645e-06, + "loss": 0.1576, + "step": 12304 + }, + { + "epoch": 5.15, + "learning_rate": 5.939862729248472e-06, + "loss": 0.131, + "step": 12305 + }, + { + "epoch": 5.15, + "learning_rate": 5.9383134094193425e-06, + "loss": 0.0974, + "step": 12306 + }, + { + "epoch": 5.15, + "learning_rate": 5.936764206339791e-06, + "loss": 0.0774, + "step": 12307 + }, + { + "epoch": 5.15, + "learning_rate": 5.935215120054345e-06, + "loss": 0.1228, + "step": 12308 + }, + { + "epoch": 5.15, + "learning_rate": 5.933666150607533e-06, + "loss": 0.1299, + "step": 12309 + }, + { + "epoch": 5.15, + "learning_rate": 5.932117298043882e-06, + "loss": 0.1099, + "step": 12310 + }, + { + "epoch": 5.15, + "learning_rate": 5.930568562407905e-06, + "loss": 0.1544, + "step": 12311 + }, + { + "epoch": 5.15, + "learning_rate": 5.9290199437441256e-06, + "loss": 0.1006, + "step": 12312 + }, + { + "epoch": 5.16, + "learning_rate": 5.927471442097057e-06, + "loss": 0.1432, + "step": 12313 + }, + { + "epoch": 5.16, + "learning_rate": 5.925923057511207e-06, + "loss": 0.1034, + "step": 12314 + }, + { + "epoch": 5.16, + "learning_rate": 5.9243747900310825e-06, + "loss": 0.1427, + "step": 12315 + }, + { + "epoch": 5.16, + "learning_rate": 5.922826639701189e-06, + "loss": 0.1157, + "step": 12316 + }, + { + "epoch": 5.16, + "learning_rate": 5.921278606566026e-06, + "loss": 0.1101, + "step": 12317 + }, + { + "epoch": 5.16, + "learning_rate": 5.9197306906700865e-06, + "loss": 0.1233, + "step": 12318 + }, + { + "epoch": 5.16, + "learning_rate": 5.918182892057871e-06, + "loss": 0.1274, + "step": 12319 + }, + { + "epoch": 5.16, + "learning_rate": 5.916635210773869e-06, + "loss": 0.1298, + "step": 12320 + }, + { + "epoch": 5.16, + "learning_rate": 5.915087646862561e-06, + "loss": 0.0994, + "step": 12321 + }, + { + "epoch": 5.16, + "learning_rate": 5.913540200368434e-06, + "loss": 0.0831, + "step": 12322 + }, + { + "epoch": 5.16, + "learning_rate": 5.911992871335967e-06, + "loss": 0.1441, + "step": 12323 + }, + { + "epoch": 5.16, + "learning_rate": 5.910445659809638e-06, + "loss": 0.0953, + "step": 12324 + }, + { + "epoch": 5.16, + "learning_rate": 5.908898565833918e-06, + "loss": 0.1614, + "step": 12325 + }, + { + "epoch": 5.16, + "learning_rate": 5.907351589453279e-06, + "loss": 0.1289, + "step": 12326 + }, + { + "epoch": 5.16, + "learning_rate": 5.905804730712189e-06, + "loss": 0.1128, + "step": 12327 + }, + { + "epoch": 5.16, + "learning_rate": 5.904257989655107e-06, + "loss": 0.0952, + "step": 12328 + }, + { + "epoch": 5.16, + "learning_rate": 5.902711366326493e-06, + "loss": 0.1338, + "step": 12329 + }, + { + "epoch": 5.16, + "learning_rate": 5.901164860770808e-06, + "loss": 0.1145, + "step": 12330 + }, + { + "epoch": 5.16, + "learning_rate": 5.8996184730325005e-06, + "loss": 0.1594, + "step": 12331 + }, + { + "epoch": 5.16, + "learning_rate": 5.89807220315602e-06, + "loss": 0.1076, + "step": 12332 + }, + { + "epoch": 5.16, + "learning_rate": 5.896526051185814e-06, + "loss": 0.1145, + "step": 12333 + }, + { + "epoch": 5.16, + "learning_rate": 5.894980017166329e-06, + "loss": 0.1147, + "step": 12334 + }, + { + "epoch": 5.16, + "learning_rate": 5.893434101141997e-06, + "loss": 0.1126, + "step": 12335 + }, + { + "epoch": 5.16, + "learning_rate": 5.891888303157258e-06, + "loss": 0.0837, + "step": 12336 + }, + { + "epoch": 5.17, + "learning_rate": 5.890342623256546e-06, + "loss": 0.0819, + "step": 12337 + }, + { + "epoch": 5.17, + "learning_rate": 5.888797061484288e-06, + "loss": 0.1497, + "step": 12338 + }, + { + "epoch": 5.17, + "learning_rate": 5.887251617884909e-06, + "loss": 0.1503, + "step": 12339 + }, + { + "epoch": 5.17, + "learning_rate": 5.885706292502835e-06, + "loss": 0.1024, + "step": 12340 + }, + { + "epoch": 5.17, + "learning_rate": 5.884161085382481e-06, + "loss": 0.0938, + "step": 12341 + }, + { + "epoch": 5.17, + "learning_rate": 5.882615996568264e-06, + "loss": 0.1657, + "step": 12342 + }, + { + "epoch": 5.17, + "learning_rate": 5.8810710261046e-06, + "loss": 0.0848, + "step": 12343 + }, + { + "epoch": 5.17, + "learning_rate": 5.87952617403589e-06, + "loss": 0.1433, + "step": 12344 + }, + { + "epoch": 5.17, + "learning_rate": 5.877981440406547e-06, + "loss": 0.1287, + "step": 12345 + }, + { + "epoch": 5.17, + "learning_rate": 5.876436825260967e-06, + "loss": 0.1218, + "step": 12346 + }, + { + "epoch": 5.17, + "learning_rate": 5.8748923286435515e-06, + "loss": 0.1001, + "step": 12347 + }, + { + "epoch": 5.17, + "learning_rate": 5.873347950598697e-06, + "loss": 0.1375, + "step": 12348 + }, + { + "epoch": 5.17, + "learning_rate": 5.871803691170792e-06, + "loss": 0.1083, + "step": 12349 + }, + { + "epoch": 5.17, + "learning_rate": 5.870259550404227e-06, + "loss": 0.1457, + "step": 12350 + }, + { + "epoch": 5.17, + "learning_rate": 5.868715528343387e-06, + "loss": 0.1202, + "step": 12351 + }, + { + "epoch": 5.17, + "learning_rate": 5.8671716250326525e-06, + "loss": 0.0686, + "step": 12352 + }, + { + "epoch": 5.17, + "learning_rate": 5.865627840516402e-06, + "loss": 0.1238, + "step": 12353 + }, + { + "epoch": 5.17, + "learning_rate": 5.864084174839013e-06, + "loss": 0.0738, + "step": 12354 + }, + { + "epoch": 5.17, + "learning_rate": 5.86254062804485e-06, + "loss": 0.1097, + "step": 12355 + }, + { + "epoch": 5.17, + "learning_rate": 5.860997200178288e-06, + "loss": 0.1317, + "step": 12356 + }, + { + "epoch": 5.17, + "learning_rate": 5.859453891283688e-06, + "loss": 0.1103, + "step": 12357 + }, + { + "epoch": 5.17, + "learning_rate": 5.857910701405411e-06, + "loss": 0.1518, + "step": 12358 + }, + { + "epoch": 5.17, + "learning_rate": 5.8563676305878155e-06, + "loss": 0.1448, + "step": 12359 + }, + { + "epoch": 5.17, + "learning_rate": 5.854824678875256e-06, + "loss": 0.1091, + "step": 12360 + }, + { + "epoch": 5.18, + "learning_rate": 5.853281846312084e-06, + "loss": 0.1521, + "step": 12361 + }, + { + "epoch": 5.18, + "learning_rate": 5.851739132942645e-06, + "loss": 0.1361, + "step": 12362 + }, + { + "epoch": 5.18, + "learning_rate": 5.850196538811282e-06, + "loss": 0.1437, + "step": 12363 + }, + { + "epoch": 5.18, + "learning_rate": 5.848654063962338e-06, + "loss": 0.135, + "step": 12364 + }, + { + "epoch": 5.18, + "learning_rate": 5.847111708440152e-06, + "loss": 0.1703, + "step": 12365 + }, + { + "epoch": 5.18, + "learning_rate": 5.845569472289051e-06, + "loss": 0.1481, + "step": 12366 + }, + { + "epoch": 5.18, + "learning_rate": 5.844027355553369e-06, + "loss": 0.1317, + "step": 12367 + }, + { + "epoch": 5.18, + "learning_rate": 5.842485358277435e-06, + "loss": 0.1071, + "step": 12368 + }, + { + "epoch": 5.18, + "learning_rate": 5.840943480505567e-06, + "loss": 0.1361, + "step": 12369 + }, + { + "epoch": 5.18, + "learning_rate": 5.839401722282089e-06, + "loss": 0.113, + "step": 12370 + }, + { + "epoch": 5.18, + "learning_rate": 5.837860083651316e-06, + "loss": 0.0978, + "step": 12371 + }, + { + "epoch": 5.18, + "learning_rate": 5.836318564657561e-06, + "loss": 0.1384, + "step": 12372 + }, + { + "epoch": 5.18, + "learning_rate": 5.834777165345138e-06, + "loss": 0.0803, + "step": 12373 + }, + { + "epoch": 5.18, + "learning_rate": 5.833235885758346e-06, + "loss": 0.1489, + "step": 12374 + }, + { + "epoch": 5.18, + "learning_rate": 5.831694725941491e-06, + "loss": 0.1497, + "step": 12375 + }, + { + "epoch": 5.18, + "learning_rate": 5.830153685938873e-06, + "loss": 0.17, + "step": 12376 + }, + { + "epoch": 5.18, + "learning_rate": 5.8286127657947854e-06, + "loss": 0.1264, + "step": 12377 + }, + { + "epoch": 5.18, + "learning_rate": 5.827071965553524e-06, + "loss": 0.1041, + "step": 12378 + }, + { + "epoch": 5.18, + "learning_rate": 5.825531285259378e-06, + "loss": 0.1206, + "step": 12379 + }, + { + "epoch": 5.18, + "learning_rate": 5.8239907249566274e-06, + "loss": 0.1378, + "step": 12380 + }, + { + "epoch": 5.18, + "learning_rate": 5.822450284689558e-06, + "loss": 0.128, + "step": 12381 + }, + { + "epoch": 5.18, + "learning_rate": 5.820909964502448e-06, + "loss": 0.1407, + "step": 12382 + }, + { + "epoch": 5.18, + "learning_rate": 5.819369764439573e-06, + "loss": 0.1534, + "step": 12383 + }, + { + "epoch": 5.18, + "learning_rate": 5.817829684545204e-06, + "loss": 0.173, + "step": 12384 + }, + { + "epoch": 5.19, + "learning_rate": 5.816289724863609e-06, + "loss": 0.113, + "step": 12385 + }, + { + "epoch": 5.19, + "learning_rate": 5.814749885439057e-06, + "loss": 0.1181, + "step": 12386 + }, + { + "epoch": 5.19, + "learning_rate": 5.813210166315804e-06, + "loss": 0.1201, + "step": 12387 + }, + { + "epoch": 5.19, + "learning_rate": 5.81167056753811e-06, + "loss": 0.1776, + "step": 12388 + }, + { + "epoch": 5.19, + "learning_rate": 5.810131089150228e-06, + "loss": 0.146, + "step": 12389 + }, + { + "epoch": 5.19, + "learning_rate": 5.808591731196409e-06, + "loss": 0.1464, + "step": 12390 + }, + { + "epoch": 5.19, + "learning_rate": 5.8070524937209e-06, + "loss": 0.169, + "step": 12391 + }, + { + "epoch": 5.19, + "learning_rate": 5.805513376767947e-06, + "loss": 0.1243, + "step": 12392 + }, + { + "epoch": 5.19, + "learning_rate": 5.8039743803817904e-06, + "loss": 0.1317, + "step": 12393 + }, + { + "epoch": 5.19, + "learning_rate": 5.80243550460667e-06, + "loss": 0.1261, + "step": 12394 + }, + { + "epoch": 5.19, + "learning_rate": 5.800896749486813e-06, + "loss": 0.1377, + "step": 12395 + }, + { + "epoch": 5.19, + "learning_rate": 5.799358115066453e-06, + "loss": 0.1073, + "step": 12396 + }, + { + "epoch": 5.19, + "learning_rate": 5.797819601389815e-06, + "loss": 0.0979, + "step": 12397 + }, + { + "epoch": 5.19, + "learning_rate": 5.7962812085011245e-06, + "loss": 0.0694, + "step": 12398 + }, + { + "epoch": 5.19, + "learning_rate": 5.7947429364446e-06, + "loss": 0.1224, + "step": 12399 + }, + { + "epoch": 5.19, + "learning_rate": 5.793204785264461e-06, + "loss": 0.1102, + "step": 12400 + }, + { + "epoch": 5.19, + "learning_rate": 5.791666755004915e-06, + "loss": 0.1489, + "step": 12401 + }, + { + "epoch": 5.19, + "learning_rate": 5.790128845710175e-06, + "loss": 0.1471, + "step": 12402 + }, + { + "epoch": 5.19, + "learning_rate": 5.788591057424443e-06, + "loss": 0.123, + "step": 12403 + }, + { + "epoch": 5.19, + "learning_rate": 5.787053390191925e-06, + "loss": 0.0856, + "step": 12404 + }, + { + "epoch": 5.19, + "learning_rate": 5.785515844056818e-06, + "loss": 0.1065, + "step": 12405 + }, + { + "epoch": 5.19, + "learning_rate": 5.783978419063323e-06, + "loss": 0.1078, + "step": 12406 + }, + { + "epoch": 5.19, + "learning_rate": 5.782441115255624e-06, + "loss": 0.0962, + "step": 12407 + }, + { + "epoch": 5.19, + "learning_rate": 5.78090393267791e-06, + "loss": 0.13, + "step": 12408 + }, + { + "epoch": 5.2, + "learning_rate": 5.77936687137437e-06, + "loss": 0.134, + "step": 12409 + }, + { + "epoch": 5.2, + "learning_rate": 5.777829931389185e-06, + "loss": 0.125, + "step": 12410 + }, + { + "epoch": 5.2, + "learning_rate": 5.776293112766535e-06, + "loss": 0.1014, + "step": 12411 + }, + { + "epoch": 5.2, + "learning_rate": 5.774756415550587e-06, + "loss": 0.1036, + "step": 12412 + }, + { + "epoch": 5.2, + "learning_rate": 5.773219839785517e-06, + "loss": 0.1309, + "step": 12413 + }, + { + "epoch": 5.2, + "learning_rate": 5.771683385515495e-06, + "loss": 0.1354, + "step": 12414 + }, + { + "epoch": 5.2, + "learning_rate": 5.770147052784678e-06, + "loss": 0.1121, + "step": 12415 + }, + { + "epoch": 5.2, + "learning_rate": 5.768610841637232e-06, + "loss": 0.1305, + "step": 12416 + }, + { + "epoch": 5.2, + "learning_rate": 5.767074752117311e-06, + "loss": 0.1492, + "step": 12417 + }, + { + "epoch": 5.2, + "learning_rate": 5.765538784269071e-06, + "loss": 0.0997, + "step": 12418 + }, + { + "epoch": 5.2, + "learning_rate": 5.76400293813666e-06, + "loss": 0.0921, + "step": 12419 + }, + { + "epoch": 5.2, + "learning_rate": 5.762467213764231e-06, + "loss": 0.1206, + "step": 12420 + }, + { + "epoch": 5.2, + "learning_rate": 5.760931611195917e-06, + "loss": 0.1416, + "step": 12421 + }, + { + "epoch": 5.2, + "learning_rate": 5.75939613047586e-06, + "loss": 0.1282, + "step": 12422 + }, + { + "epoch": 5.2, + "learning_rate": 5.757860771648202e-06, + "loss": 0.1128, + "step": 12423 + }, + { + "epoch": 5.2, + "learning_rate": 5.75632553475707e-06, + "loss": 0.1247, + "step": 12424 + }, + { + "epoch": 5.2, + "learning_rate": 5.754790419846595e-06, + "loss": 0.1211, + "step": 12425 + }, + { + "epoch": 5.2, + "learning_rate": 5.753255426960901e-06, + "loss": 0.1361, + "step": 12426 + }, + { + "epoch": 5.2, + "learning_rate": 5.751720556144116e-06, + "loss": 0.1106, + "step": 12427 + }, + { + "epoch": 5.2, + "learning_rate": 5.750185807440349e-06, + "loss": 0.1182, + "step": 12428 + }, + { + "epoch": 5.2, + "learning_rate": 5.748651180893721e-06, + "loss": 0.1056, + "step": 12429 + }, + { + "epoch": 5.2, + "learning_rate": 5.747116676548342e-06, + "loss": 0.0918, + "step": 12430 + }, + { + "epoch": 5.2, + "learning_rate": 5.745582294448319e-06, + "loss": 0.1437, + "step": 12431 + }, + { + "epoch": 5.2, + "learning_rate": 5.744048034637759e-06, + "loss": 0.15, + "step": 12432 + }, + { + "epoch": 5.21, + "learning_rate": 5.742513897160766e-06, + "loss": 0.1019, + "step": 12433 + }, + { + "epoch": 5.21, + "learning_rate": 5.740979882061433e-06, + "loss": 0.0812, + "step": 12434 + }, + { + "epoch": 5.21, + "learning_rate": 5.7394459893838486e-06, + "loss": 0.0771, + "step": 12435 + }, + { + "epoch": 5.21, + "learning_rate": 5.737912219172108e-06, + "loss": 0.115, + "step": 12436 + }, + { + "epoch": 5.21, + "learning_rate": 5.7363785714703e-06, + "loss": 0.1307, + "step": 12437 + }, + { + "epoch": 5.21, + "learning_rate": 5.734845046322506e-06, + "loss": 0.1003, + "step": 12438 + }, + { + "epoch": 5.21, + "learning_rate": 5.733311643772807e-06, + "loss": 0.1297, + "step": 12439 + }, + { + "epoch": 5.21, + "learning_rate": 5.731778363865278e-06, + "loss": 0.1407, + "step": 12440 + }, + { + "epoch": 5.21, + "learning_rate": 5.730245206643996e-06, + "loss": 0.1088, + "step": 12441 + }, + { + "epoch": 5.21, + "learning_rate": 5.728712172153024e-06, + "loss": 0.1324, + "step": 12442 + }, + { + "epoch": 5.21, + "learning_rate": 5.727179260436428e-06, + "loss": 0.1405, + "step": 12443 + }, + { + "epoch": 5.21, + "learning_rate": 5.7256464715382754e-06, + "loss": 0.1619, + "step": 12444 + }, + { + "epoch": 5.21, + "learning_rate": 5.72411380550262e-06, + "loss": 0.088, + "step": 12445 + }, + { + "epoch": 5.21, + "learning_rate": 5.722581262373519e-06, + "loss": 0.1305, + "step": 12446 + }, + { + "epoch": 5.21, + "learning_rate": 5.721048842195028e-06, + "loss": 0.1233, + "step": 12447 + }, + { + "epoch": 5.21, + "learning_rate": 5.719516545011188e-06, + "loss": 0.127, + "step": 12448 + }, + { + "epoch": 5.21, + "learning_rate": 5.7179843708660454e-06, + "loss": 0.1287, + "step": 12449 + }, + { + "epoch": 5.21, + "learning_rate": 5.716452319803643e-06, + "loss": 0.1295, + "step": 12450 + }, + { + "epoch": 5.21, + "learning_rate": 5.7149203918680166e-06, + "loss": 0.1007, + "step": 12451 + }, + { + "epoch": 5.21, + "learning_rate": 5.713388587103201e-06, + "loss": 0.1025, + "step": 12452 + }, + { + "epoch": 5.21, + "learning_rate": 5.711856905553228e-06, + "loss": 0.1114, + "step": 12453 + }, + { + "epoch": 5.21, + "learning_rate": 5.710325347262125e-06, + "loss": 0.1141, + "step": 12454 + }, + { + "epoch": 5.21, + "learning_rate": 5.708793912273911e-06, + "loss": 0.1332, + "step": 12455 + }, + { + "epoch": 5.21, + "learning_rate": 5.707262600632612e-06, + "loss": 0.1265, + "step": 12456 + }, + { + "epoch": 5.22, + "learning_rate": 5.705731412382235e-06, + "loss": 0.1596, + "step": 12457 + }, + { + "epoch": 5.22, + "learning_rate": 5.704200347566798e-06, + "loss": 0.1179, + "step": 12458 + }, + { + "epoch": 5.22, + "learning_rate": 5.7026694062303115e-06, + "loss": 0.0919, + "step": 12459 + }, + { + "epoch": 5.22, + "learning_rate": 5.701138588416778e-06, + "loss": 0.1254, + "step": 12460 + }, + { + "epoch": 5.22, + "learning_rate": 5.699607894170206e-06, + "loss": 0.1102, + "step": 12461 + }, + { + "epoch": 5.22, + "learning_rate": 5.698077323534584e-06, + "loss": 0.1143, + "step": 12462 + }, + { + "epoch": 5.22, + "learning_rate": 5.696546876553912e-06, + "loss": 0.1093, + "step": 12463 + }, + { + "epoch": 5.22, + "learning_rate": 5.695016553272182e-06, + "loss": 0.1028, + "step": 12464 + }, + { + "epoch": 5.22, + "learning_rate": 5.693486353733382e-06, + "loss": 0.1587, + "step": 12465 + }, + { + "epoch": 5.22, + "learning_rate": 5.691956277981493e-06, + "loss": 0.1528, + "step": 12466 + }, + { + "epoch": 5.22, + "learning_rate": 5.690426326060498e-06, + "loss": 0.1301, + "step": 12467 + }, + { + "epoch": 5.22, + "learning_rate": 5.688896498014381e-06, + "loss": 0.1587, + "step": 12468 + }, + { + "epoch": 5.22, + "learning_rate": 5.687366793887101e-06, + "loss": 0.1473, + "step": 12469 + }, + { + "epoch": 5.22, + "learning_rate": 5.685837213722638e-06, + "loss": 0.1648, + "step": 12470 + }, + { + "epoch": 5.22, + "learning_rate": 5.684307757564954e-06, + "loss": 0.116, + "step": 12471 + }, + { + "epoch": 5.22, + "learning_rate": 5.682778425458015e-06, + "loss": 0.0865, + "step": 12472 + }, + { + "epoch": 5.22, + "learning_rate": 5.681249217445779e-06, + "loss": 0.1476, + "step": 12473 + }, + { + "epoch": 5.22, + "learning_rate": 5.6797201335722064e-06, + "loss": 0.142, + "step": 12474 + }, + { + "epoch": 5.22, + "learning_rate": 5.67819117388124e-06, + "loss": 0.1112, + "step": 12475 + }, + { + "epoch": 5.22, + "learning_rate": 5.676662338416833e-06, + "loss": 0.1331, + "step": 12476 + }, + { + "epoch": 5.22, + "learning_rate": 5.675133627222931e-06, + "loss": 0.1341, + "step": 12477 + }, + { + "epoch": 5.22, + "learning_rate": 5.673605040343475e-06, + "loss": 0.1504, + "step": 12478 + }, + { + "epoch": 5.22, + "learning_rate": 5.6720765778224075e-06, + "loss": 0.1181, + "step": 12479 + }, + { + "epoch": 5.23, + "learning_rate": 5.670548239703653e-06, + "loss": 0.1158, + "step": 12480 + }, + { + "epoch": 5.23, + "learning_rate": 5.669020026031148e-06, + "loss": 0.1225, + "step": 12481 + }, + { + "epoch": 5.23, + "learning_rate": 5.6674919368488235e-06, + "loss": 0.1307, + "step": 12482 + }, + { + "epoch": 5.23, + "learning_rate": 5.665963972200594e-06, + "loss": 0.121, + "step": 12483 + }, + { + "epoch": 5.23, + "learning_rate": 5.664436132130384e-06, + "loss": 0.1369, + "step": 12484 + }, + { + "epoch": 5.23, + "learning_rate": 5.662908416682111e-06, + "loss": 0.1592, + "step": 12485 + }, + { + "epoch": 5.23, + "learning_rate": 5.661380825899687e-06, + "loss": 0.1245, + "step": 12486 + }, + { + "epoch": 5.23, + "learning_rate": 5.6598533598270215e-06, + "loss": 0.0663, + "step": 12487 + }, + { + "epoch": 5.23, + "learning_rate": 5.658326018508022e-06, + "loss": 0.0951, + "step": 12488 + }, + { + "epoch": 5.23, + "learning_rate": 5.656798801986585e-06, + "loss": 0.1009, + "step": 12489 + }, + { + "epoch": 5.23, + "learning_rate": 5.655271710306613e-06, + "loss": 0.128, + "step": 12490 + }, + { + "epoch": 5.23, + "learning_rate": 5.653744743512e-06, + "loss": 0.1113, + "step": 12491 + }, + { + "epoch": 5.23, + "learning_rate": 5.652217901646637e-06, + "loss": 0.1389, + "step": 12492 + }, + { + "epoch": 5.23, + "learning_rate": 5.6506911847544134e-06, + "loss": 0.1356, + "step": 12493 + }, + { + "epoch": 5.23, + "learning_rate": 5.649164592879213e-06, + "loss": 0.1161, + "step": 12494 + }, + { + "epoch": 5.23, + "learning_rate": 5.647638126064917e-06, + "loss": 0.1361, + "step": 12495 + }, + { + "epoch": 5.23, + "learning_rate": 5.646111784355398e-06, + "loss": 0.112, + "step": 12496 + }, + { + "epoch": 5.23, + "learning_rate": 5.644585567794531e-06, + "loss": 0.1376, + "step": 12497 + }, + { + "epoch": 5.23, + "learning_rate": 5.643059476426189e-06, + "loss": 0.1104, + "step": 12498 + }, + { + "epoch": 5.23, + "learning_rate": 5.641533510294235e-06, + "loss": 0.115, + "step": 12499 + }, + { + "epoch": 5.23, + "learning_rate": 5.640007669442533e-06, + "loss": 0.0994, + "step": 12500 + }, + { + "epoch": 5.23, + "learning_rate": 5.638481953914945e-06, + "loss": 0.15, + "step": 12501 + }, + { + "epoch": 5.23, + "learning_rate": 5.636956363755324e-06, + "loss": 0.1126, + "step": 12502 + }, + { + "epoch": 5.23, + "learning_rate": 5.635430899007515e-06, + "loss": 0.1103, + "step": 12503 + }, + { + "epoch": 5.24, + "learning_rate": 5.6339055597153734e-06, + "loss": 0.1124, + "step": 12504 + }, + { + "epoch": 5.24, + "learning_rate": 5.632380345922741e-06, + "loss": 0.1223, + "step": 12505 + }, + { + "epoch": 5.24, + "learning_rate": 5.63085525767346e-06, + "loss": 0.1191, + "step": 12506 + }, + { + "epoch": 5.24, + "learning_rate": 5.629330295011367e-06, + "loss": 0.1154, + "step": 12507 + }, + { + "epoch": 5.24, + "learning_rate": 5.627805457980298e-06, + "loss": 0.1415, + "step": 12508 + }, + { + "epoch": 5.24, + "learning_rate": 5.626280746624083e-06, + "loss": 0.1086, + "step": 12509 + }, + { + "epoch": 5.24, + "learning_rate": 5.6247561609865445e-06, + "loss": 0.0688, + "step": 12510 + }, + { + "epoch": 5.24, + "learning_rate": 5.623231701111507e-06, + "loss": 0.1371, + "step": 12511 + }, + { + "epoch": 5.24, + "learning_rate": 5.6217073670427905e-06, + "loss": 0.1134, + "step": 12512 + }, + { + "epoch": 5.24, + "learning_rate": 5.62018315882421e-06, + "loss": 0.1037, + "step": 12513 + }, + { + "epoch": 5.24, + "learning_rate": 5.618659076499578e-06, + "loss": 0.126, + "step": 12514 + }, + { + "epoch": 5.24, + "learning_rate": 5.6171351201127065e-06, + "loss": 0.0937, + "step": 12515 + }, + { + "epoch": 5.24, + "learning_rate": 5.615611289707394e-06, + "loss": 0.1005, + "step": 12516 + }, + { + "epoch": 5.24, + "learning_rate": 5.614087585327444e-06, + "loss": 0.0986, + "step": 12517 + }, + { + "epoch": 5.24, + "learning_rate": 5.612564007016654e-06, + "loss": 0.1368, + "step": 12518 + }, + { + "epoch": 5.24, + "learning_rate": 5.611040554818819e-06, + "loss": 0.1007, + "step": 12519 + }, + { + "epoch": 5.24, + "learning_rate": 5.6095172287777275e-06, + "loss": 0.081, + "step": 12520 + }, + { + "epoch": 5.24, + "learning_rate": 5.607994028937171e-06, + "loss": 0.0976, + "step": 12521 + }, + { + "epoch": 5.24, + "learning_rate": 5.606470955340926e-06, + "loss": 0.1193, + "step": 12522 + }, + { + "epoch": 5.24, + "learning_rate": 5.604948008032774e-06, + "loss": 0.1216, + "step": 12523 + }, + { + "epoch": 5.24, + "learning_rate": 5.603425187056496e-06, + "loss": 0.0846, + "step": 12524 + }, + { + "epoch": 5.24, + "learning_rate": 5.601902492455855e-06, + "loss": 0.0941, + "step": 12525 + }, + { + "epoch": 5.24, + "learning_rate": 5.6003799242746235e-06, + "loss": 0.0941, + "step": 12526 + }, + { + "epoch": 5.24, + "learning_rate": 5.598857482556568e-06, + "loss": 0.107, + "step": 12527 + }, + { + "epoch": 5.25, + "learning_rate": 5.597335167345448e-06, + "loss": 0.1391, + "step": 12528 + }, + { + "epoch": 5.25, + "learning_rate": 5.595812978685027e-06, + "loss": 0.1228, + "step": 12529 + }, + { + "epoch": 5.25, + "learning_rate": 5.594290916619048e-06, + "loss": 0.1307, + "step": 12530 + }, + { + "epoch": 5.25, + "learning_rate": 5.592768981191268e-06, + "loss": 0.1124, + "step": 12531 + }, + { + "epoch": 5.25, + "learning_rate": 5.591247172445432e-06, + "loss": 0.1211, + "step": 12532 + }, + { + "epoch": 5.25, + "learning_rate": 5.589725490425284e-06, + "loss": 0.1426, + "step": 12533 + }, + { + "epoch": 5.25, + "learning_rate": 5.588203935174563e-06, + "loss": 0.1508, + "step": 12534 + }, + { + "epoch": 5.25, + "learning_rate": 5.586682506737007e-06, + "loss": 0.1433, + "step": 12535 + }, + { + "epoch": 5.25, + "learning_rate": 5.585161205156347e-06, + "loss": 0.0949, + "step": 12536 + }, + { + "epoch": 5.25, + "learning_rate": 5.583640030476309e-06, + "loss": 0.1226, + "step": 12537 + }, + { + "epoch": 5.25, + "learning_rate": 5.582118982740618e-06, + "loss": 0.1161, + "step": 12538 + }, + { + "epoch": 5.25, + "learning_rate": 5.580598061992997e-06, + "loss": 0.1382, + "step": 12539 + }, + { + "epoch": 5.25, + "learning_rate": 5.5790772682771625e-06, + "loss": 0.1094, + "step": 12540 + }, + { + "epoch": 5.25, + "learning_rate": 5.577556601636828e-06, + "loss": 0.1252, + "step": 12541 + }, + { + "epoch": 5.25, + "learning_rate": 5.576036062115709e-06, + "loss": 0.1765, + "step": 12542 + }, + { + "epoch": 5.25, + "learning_rate": 5.574515649757505e-06, + "loss": 0.1138, + "step": 12543 + }, + { + "epoch": 5.25, + "learning_rate": 5.572995364605921e-06, + "loss": 0.1526, + "step": 12544 + }, + { + "epoch": 5.25, + "learning_rate": 5.571475206704656e-06, + "loss": 0.1438, + "step": 12545 + }, + { + "epoch": 5.25, + "learning_rate": 5.569955176097408e-06, + "loss": 0.1175, + "step": 12546 + }, + { + "epoch": 5.25, + "learning_rate": 5.5684352728278705e-06, + "loss": 0.1257, + "step": 12547 + }, + { + "epoch": 5.25, + "learning_rate": 5.566915496939723e-06, + "loss": 0.1317, + "step": 12548 + }, + { + "epoch": 5.25, + "learning_rate": 5.565395848476658e-06, + "loss": 0.1251, + "step": 12549 + }, + { + "epoch": 5.25, + "learning_rate": 5.563876327482356e-06, + "loss": 0.1561, + "step": 12550 + }, + { + "epoch": 5.25, + "learning_rate": 5.56235693400049e-06, + "loss": 0.1404, + "step": 12551 + }, + { + "epoch": 5.26, + "learning_rate": 5.5608376680747354e-06, + "loss": 0.144, + "step": 12552 + }, + { + "epoch": 5.26, + "learning_rate": 5.559318529748763e-06, + "loss": 0.1175, + "step": 12553 + }, + { + "epoch": 5.26, + "learning_rate": 5.55779951906624e-06, + "loss": 0.1148, + "step": 12554 + }, + { + "epoch": 5.26, + "learning_rate": 5.556280636070826e-06, + "loss": 0.1124, + "step": 12555 + }, + { + "epoch": 5.26, + "learning_rate": 5.554761880806187e-06, + "loss": 0.102, + "step": 12556 + }, + { + "epoch": 5.26, + "learning_rate": 5.5532432533159676e-06, + "loss": 0.1266, + "step": 12557 + }, + { + "epoch": 5.26, + "learning_rate": 5.551724753643826e-06, + "loss": 0.1217, + "step": 12558 + }, + { + "epoch": 5.26, + "learning_rate": 5.550206381833409e-06, + "loss": 0.1142, + "step": 12559 + }, + { + "epoch": 5.26, + "learning_rate": 5.548688137928362e-06, + "loss": 0.0905, + "step": 12560 + }, + { + "epoch": 5.26, + "learning_rate": 5.547170021972323e-06, + "loss": 0.1191, + "step": 12561 + }, + { + "epoch": 5.26, + "learning_rate": 5.545652034008935e-06, + "loss": 0.1319, + "step": 12562 + }, + { + "epoch": 5.26, + "learning_rate": 5.544134174081821e-06, + "loss": 0.1422, + "step": 12563 + }, + { + "epoch": 5.26, + "learning_rate": 5.542616442234618e-06, + "loss": 0.0982, + "step": 12564 + }, + { + "epoch": 5.26, + "learning_rate": 5.54109883851095e-06, + "loss": 0.1629, + "step": 12565 + }, + { + "epoch": 5.26, + "learning_rate": 5.539581362954438e-06, + "loss": 0.1433, + "step": 12566 + }, + { + "epoch": 5.26, + "learning_rate": 5.5380640156087016e-06, + "loss": 0.0967, + "step": 12567 + }, + { + "epoch": 5.26, + "learning_rate": 5.536546796517357e-06, + "loss": 0.1558, + "step": 12568 + }, + { + "epoch": 5.26, + "learning_rate": 5.535029705724017e-06, + "loss": 0.0975, + "step": 12569 + }, + { + "epoch": 5.26, + "learning_rate": 5.533512743272285e-06, + "loss": 0.1051, + "step": 12570 + }, + { + "epoch": 5.26, + "learning_rate": 5.531995909205763e-06, + "loss": 0.1519, + "step": 12571 + }, + { + "epoch": 5.26, + "learning_rate": 5.530479203568053e-06, + "loss": 0.1319, + "step": 12572 + }, + { + "epoch": 5.26, + "learning_rate": 5.528962626402754e-06, + "loss": 0.1028, + "step": 12573 + }, + { + "epoch": 5.26, + "learning_rate": 5.527446177753455e-06, + "loss": 0.1478, + "step": 12574 + }, + { + "epoch": 5.26, + "learning_rate": 5.525929857663747e-06, + "loss": 0.1575, + "step": 12575 + }, + { + "epoch": 5.27, + "learning_rate": 5.524413666177216e-06, + "loss": 0.116, + "step": 12576 + }, + { + "epoch": 5.27, + "learning_rate": 5.522897603337445e-06, + "loss": 0.1317, + "step": 12577 + }, + { + "epoch": 5.27, + "learning_rate": 5.521381669188006e-06, + "loss": 0.1256, + "step": 12578 + }, + { + "epoch": 5.27, + "learning_rate": 5.519865863772477e-06, + "loss": 0.1148, + "step": 12579 + }, + { + "epoch": 5.27, + "learning_rate": 5.518350187134427e-06, + "loss": 0.1114, + "step": 12580 + }, + { + "epoch": 5.27, + "learning_rate": 5.516834639317426e-06, + "loss": 0.1411, + "step": 12581 + }, + { + "epoch": 5.27, + "learning_rate": 5.515319220365034e-06, + "loss": 0.1377, + "step": 12582 + }, + { + "epoch": 5.27, + "learning_rate": 5.513803930320815e-06, + "loss": 0.079, + "step": 12583 + }, + { + "epoch": 5.27, + "learning_rate": 5.512288769228318e-06, + "loss": 0.1564, + "step": 12584 + }, + { + "epoch": 5.27, + "learning_rate": 5.510773737131097e-06, + "loss": 0.1106, + "step": 12585 + }, + { + "epoch": 5.27, + "learning_rate": 5.509258834072702e-06, + "loss": 0.1236, + "step": 12586 + }, + { + "epoch": 5.27, + "learning_rate": 5.507744060096676e-06, + "loss": 0.1165, + "step": 12587 + }, + { + "epoch": 5.27, + "learning_rate": 5.50622941524656e-06, + "loss": 0.1437, + "step": 12588 + }, + { + "epoch": 5.27, + "learning_rate": 5.5047148995658976e-06, + "loss": 0.1063, + "step": 12589 + }, + { + "epoch": 5.27, + "learning_rate": 5.5032005130982115e-06, + "loss": 0.1413, + "step": 12590 + }, + { + "epoch": 5.27, + "learning_rate": 5.501686255887036e-06, + "loss": 0.1175, + "step": 12591 + }, + { + "epoch": 5.27, + "learning_rate": 5.500172127975903e-06, + "loss": 0.1253, + "step": 12592 + }, + { + "epoch": 5.27, + "learning_rate": 5.498658129408323e-06, + "loss": 0.1528, + "step": 12593 + }, + { + "epoch": 5.27, + "learning_rate": 5.49714426022782e-06, + "loss": 0.104, + "step": 12594 + }, + { + "epoch": 5.27, + "learning_rate": 5.495630520477912e-06, + "loss": 0.1079, + "step": 12595 + }, + { + "epoch": 5.27, + "learning_rate": 5.4941169102021065e-06, + "loss": 0.1383, + "step": 12596 + }, + { + "epoch": 5.27, + "learning_rate": 5.492603429443916e-06, + "loss": 0.1617, + "step": 12597 + }, + { + "epoch": 5.27, + "learning_rate": 5.491090078246836e-06, + "loss": 0.1331, + "step": 12598 + }, + { + "epoch": 5.27, + "learning_rate": 5.489576856654371e-06, + "loss": 0.161, + "step": 12599 + }, + { + "epoch": 5.28, + "learning_rate": 5.488063764710016e-06, + "loss": 0.1442, + "step": 12600 + }, + { + "epoch": 5.28, + "learning_rate": 5.486550802457264e-06, + "loss": 0.1255, + "step": 12601 + }, + { + "epoch": 5.28, + "learning_rate": 5.485037969939605e-06, + "loss": 0.1291, + "step": 12602 + }, + { + "epoch": 5.28, + "learning_rate": 5.483525267200527e-06, + "loss": 0.1409, + "step": 12603 + }, + { + "epoch": 5.28, + "learning_rate": 5.482012694283503e-06, + "loss": 0.1257, + "step": 12604 + }, + { + "epoch": 5.28, + "learning_rate": 5.480500251232015e-06, + "loss": 0.1324, + "step": 12605 + }, + { + "epoch": 5.28, + "learning_rate": 5.478987938089537e-06, + "loss": 0.1447, + "step": 12606 + }, + { + "epoch": 5.28, + "learning_rate": 5.477475754899538e-06, + "loss": 0.1482, + "step": 12607 + }, + { + "epoch": 5.28, + "learning_rate": 5.475963701705487e-06, + "loss": 0.1701, + "step": 12608 + }, + { + "epoch": 5.28, + "learning_rate": 5.4744517785508425e-06, + "loss": 0.1152, + "step": 12609 + }, + { + "epoch": 5.28, + "learning_rate": 5.472939985479071e-06, + "loss": 0.1722, + "step": 12610 + }, + { + "epoch": 5.28, + "learning_rate": 5.471428322533619e-06, + "loss": 0.124, + "step": 12611 + }, + { + "epoch": 5.28, + "learning_rate": 5.469916789757941e-06, + "loss": 0.1133, + "step": 12612 + }, + { + "epoch": 5.28, + "learning_rate": 5.4684053871954835e-06, + "loss": 0.0509, + "step": 12613 + }, + { + "epoch": 5.28, + "learning_rate": 5.466894114889697e-06, + "loss": 0.126, + "step": 12614 + }, + { + "epoch": 5.28, + "learning_rate": 5.4653829728840126e-06, + "loss": 0.117, + "step": 12615 + }, + { + "epoch": 5.28, + "learning_rate": 5.463871961221871e-06, + "loss": 0.115, + "step": 12616 + }, + { + "epoch": 5.28, + "learning_rate": 5.462361079946708e-06, + "loss": 0.1245, + "step": 12617 + }, + { + "epoch": 5.28, + "learning_rate": 5.460850329101944e-06, + "loss": 0.1079, + "step": 12618 + }, + { + "epoch": 5.28, + "learning_rate": 5.459339708731011e-06, + "loss": 0.0993, + "step": 12619 + }, + { + "epoch": 5.28, + "learning_rate": 5.457829218877328e-06, + "loss": 0.1243, + "step": 12620 + }, + { + "epoch": 5.28, + "learning_rate": 5.456318859584314e-06, + "loss": 0.1076, + "step": 12621 + }, + { + "epoch": 5.28, + "learning_rate": 5.454808630895383e-06, + "loss": 0.1082, + "step": 12622 + }, + { + "epoch": 5.28, + "learning_rate": 5.453298532853943e-06, + "loss": 0.1531, + "step": 12623 + }, + { + "epoch": 5.29, + "learning_rate": 5.451788565503406e-06, + "loss": 0.1146, + "step": 12624 + }, + { + "epoch": 5.29, + "learning_rate": 5.450278728887168e-06, + "loss": 0.1133, + "step": 12625 + }, + { + "epoch": 5.29, + "learning_rate": 5.44876902304863e-06, + "loss": 0.1498, + "step": 12626 + }, + { + "epoch": 5.29, + "learning_rate": 5.447259448031188e-06, + "loss": 0.1062, + "step": 12627 + }, + { + "epoch": 5.29, + "learning_rate": 5.445750003878233e-06, + "loss": 0.1089, + "step": 12628 + }, + { + "epoch": 5.29, + "learning_rate": 5.444240690633152e-06, + "loss": 0.1413, + "step": 12629 + }, + { + "epoch": 5.29, + "learning_rate": 5.442731508339334e-06, + "loss": 0.1212, + "step": 12630 + }, + { + "epoch": 5.29, + "learning_rate": 5.441222457040152e-06, + "loss": 0.1449, + "step": 12631 + }, + { + "epoch": 5.29, + "learning_rate": 5.439713536778983e-06, + "loss": 0.1023, + "step": 12632 + }, + { + "epoch": 5.29, + "learning_rate": 5.438204747599204e-06, + "loss": 0.135, + "step": 12633 + }, + { + "epoch": 5.29, + "learning_rate": 5.43669608954418e-06, + "loss": 0.1051, + "step": 12634 + }, + { + "epoch": 5.29, + "learning_rate": 5.435187562657278e-06, + "loss": 0.1177, + "step": 12635 + }, + { + "epoch": 5.29, + "learning_rate": 5.433679166981858e-06, + "loss": 0.1278, + "step": 12636 + }, + { + "epoch": 5.29, + "learning_rate": 5.432170902561282e-06, + "loss": 0.088, + "step": 12637 + }, + { + "epoch": 5.29, + "learning_rate": 5.4306627694389014e-06, + "loss": 0.1628, + "step": 12638 + }, + { + "epoch": 5.29, + "learning_rate": 5.42915476765806e-06, + "loss": 0.0765, + "step": 12639 + }, + { + "epoch": 5.29, + "learning_rate": 5.427646897262109e-06, + "loss": 0.1318, + "step": 12640 + }, + { + "epoch": 5.29, + "learning_rate": 5.42613915829439e-06, + "loss": 0.097, + "step": 12641 + }, + { + "epoch": 5.29, + "learning_rate": 5.424631550798243e-06, + "loss": 0.1231, + "step": 12642 + }, + { + "epoch": 5.29, + "learning_rate": 5.423124074817001e-06, + "loss": 0.1377, + "step": 12643 + }, + { + "epoch": 5.29, + "learning_rate": 5.421616730394e-06, + "loss": 0.1024, + "step": 12644 + }, + { + "epoch": 5.29, + "learning_rate": 5.42010951757256e-06, + "loss": 0.1169, + "step": 12645 + }, + { + "epoch": 5.29, + "learning_rate": 5.418602436396007e-06, + "loss": 0.1132, + "step": 12646 + }, + { + "epoch": 5.29, + "learning_rate": 5.417095486907663e-06, + "loss": 0.1527, + "step": 12647 + }, + { + "epoch": 5.3, + "learning_rate": 5.415588669150842e-06, + "loss": 0.1227, + "step": 12648 + }, + { + "epoch": 5.3, + "learning_rate": 5.414081983168855e-06, + "loss": 0.131, + "step": 12649 + }, + { + "epoch": 5.3, + "learning_rate": 5.412575429005014e-06, + "loss": 0.1329, + "step": 12650 + }, + { + "epoch": 5.3, + "learning_rate": 5.411069006702624e-06, + "loss": 0.1036, + "step": 12651 + }, + { + "epoch": 5.3, + "learning_rate": 5.409562716304979e-06, + "loss": 0.1419, + "step": 12652 + }, + { + "epoch": 5.3, + "learning_rate": 5.408056557855381e-06, + "loss": 0.125, + "step": 12653 + }, + { + "epoch": 5.3, + "learning_rate": 5.406550531397123e-06, + "loss": 0.1224, + "step": 12654 + }, + { + "epoch": 5.3, + "learning_rate": 5.405044636973493e-06, + "loss": 0.1051, + "step": 12655 + }, + { + "epoch": 5.3, + "learning_rate": 5.403538874627777e-06, + "loss": 0.131, + "step": 12656 + }, + { + "epoch": 5.3, + "learning_rate": 5.402033244403262e-06, + "loss": 0.0957, + "step": 12657 + }, + { + "epoch": 5.3, + "learning_rate": 5.400527746343217e-06, + "loss": 0.1377, + "step": 12658 + }, + { + "epoch": 5.3, + "learning_rate": 5.399022380490921e-06, + "loss": 0.086, + "step": 12659 + }, + { + "epoch": 5.3, + "learning_rate": 5.397517146889647e-06, + "loss": 0.0981, + "step": 12660 + }, + { + "epoch": 5.3, + "learning_rate": 5.3960120455826545e-06, + "loss": 0.1452, + "step": 12661 + }, + { + "epoch": 5.3, + "learning_rate": 5.394507076613212e-06, + "loss": 0.1332, + "step": 12662 + }, + { + "epoch": 5.3, + "learning_rate": 5.3930022400245745e-06, + "loss": 0.1038, + "step": 12663 + }, + { + "epoch": 5.3, + "learning_rate": 5.391497535859999e-06, + "loss": 0.1076, + "step": 12664 + }, + { + "epoch": 5.3, + "learning_rate": 5.3899929641627425e-06, + "loss": 0.1232, + "step": 12665 + }, + { + "epoch": 5.3, + "learning_rate": 5.388488524976043e-06, + "loss": 0.1222, + "step": 12666 + }, + { + "epoch": 5.3, + "learning_rate": 5.3869842183431495e-06, + "loss": 0.1345, + "step": 12667 + }, + { + "epoch": 5.3, + "learning_rate": 5.3854800443073006e-06, + "loss": 0.0914, + "step": 12668 + }, + { + "epoch": 5.3, + "learning_rate": 5.383976002911734e-06, + "loss": 0.152, + "step": 12669 + }, + { + "epoch": 5.3, + "learning_rate": 5.38247209419968e-06, + "loss": 0.1065, + "step": 12670 + }, + { + "epoch": 5.31, + "learning_rate": 5.3809683182143725e-06, + "loss": 0.1336, + "step": 12671 + }, + { + "epoch": 5.31, + "learning_rate": 5.3794646749990265e-06, + "loss": 0.1259, + "step": 12672 + }, + { + "epoch": 5.31, + "learning_rate": 5.3779611645968696e-06, + "loss": 0.1357, + "step": 12673 + }, + { + "epoch": 5.31, + "learning_rate": 5.376457787051117e-06, + "loss": 0.1132, + "step": 12674 + }, + { + "epoch": 5.31, + "learning_rate": 5.374954542404982e-06, + "loss": 0.1397, + "step": 12675 + }, + { + "epoch": 5.31, + "learning_rate": 5.373451430701676e-06, + "loss": 0.0993, + "step": 12676 + }, + { + "epoch": 5.31, + "learning_rate": 5.371948451984401e-06, + "loss": 0.0798, + "step": 12677 + }, + { + "epoch": 5.31, + "learning_rate": 5.3704456062963674e-06, + "loss": 0.1446, + "step": 12678 + }, + { + "epoch": 5.31, + "learning_rate": 5.368942893680761e-06, + "loss": 0.1033, + "step": 12679 + }, + { + "epoch": 5.31, + "learning_rate": 5.367440314180782e-06, + "loss": 0.125, + "step": 12680 + }, + { + "epoch": 5.31, + "learning_rate": 5.36593786783962e-06, + "loss": 0.1587, + "step": 12681 + }, + { + "epoch": 5.31, + "learning_rate": 5.3644355547004664e-06, + "loss": 0.1135, + "step": 12682 + }, + { + "epoch": 5.31, + "learning_rate": 5.362933374806494e-06, + "loss": 0.1085, + "step": 12683 + }, + { + "epoch": 5.31, + "learning_rate": 5.361431328200888e-06, + "loss": 0.1236, + "step": 12684 + }, + { + "epoch": 5.31, + "learning_rate": 5.3599294149268255e-06, + "loss": 0.1151, + "step": 12685 + }, + { + "epoch": 5.31, + "learning_rate": 5.3584276350274725e-06, + "loss": 0.1447, + "step": 12686 + }, + { + "epoch": 5.31, + "learning_rate": 5.356925988545996e-06, + "loss": 0.1179, + "step": 12687 + }, + { + "epoch": 5.31, + "learning_rate": 5.355424475525562e-06, + "loss": 0.1462, + "step": 12688 + }, + { + "epoch": 5.31, + "learning_rate": 5.3539230960093305e-06, + "loss": 0.1122, + "step": 12689 + }, + { + "epoch": 5.31, + "learning_rate": 5.352421850040457e-06, + "loss": 0.1016, + "step": 12690 + }, + { + "epoch": 5.31, + "learning_rate": 5.3509207376620935e-06, + "loss": 0.1547, + "step": 12691 + }, + { + "epoch": 5.31, + "learning_rate": 5.349419758917392e-06, + "loss": 0.1373, + "step": 12692 + }, + { + "epoch": 5.31, + "learning_rate": 5.347918913849487e-06, + "loss": 0.1006, + "step": 12693 + }, + { + "epoch": 5.31, + "learning_rate": 5.346418202501526e-06, + "loss": 0.1653, + "step": 12694 + }, + { + "epoch": 5.32, + "learning_rate": 5.344917624916644e-06, + "loss": 0.1047, + "step": 12695 + }, + { + "epoch": 5.32, + "learning_rate": 5.343417181137973e-06, + "loss": 0.1224, + "step": 12696 + }, + { + "epoch": 5.32, + "learning_rate": 5.341916871208644e-06, + "loss": 0.1222, + "step": 12697 + }, + { + "epoch": 5.32, + "learning_rate": 5.340416695171785e-06, + "loss": 0.1651, + "step": 12698 + }, + { + "epoch": 5.32, + "learning_rate": 5.3389166530705096e-06, + "loss": 0.1085, + "step": 12699 + }, + { + "epoch": 5.32, + "learning_rate": 5.337416744947939e-06, + "loss": 0.1256, + "step": 12700 + }, + { + "epoch": 5.32, + "learning_rate": 5.335916970847185e-06, + "loss": 0.0904, + "step": 12701 + }, + { + "epoch": 5.32, + "learning_rate": 5.3344173308113586e-06, + "loss": 0.1208, + "step": 12702 + }, + { + "epoch": 5.32, + "learning_rate": 5.332917824883568e-06, + "loss": 0.0958, + "step": 12703 + }, + { + "epoch": 5.32, + "learning_rate": 5.331418453106916e-06, + "loss": 0.1328, + "step": 12704 + }, + { + "epoch": 5.32, + "learning_rate": 5.329919215524494e-06, + "loss": 0.1195, + "step": 12705 + }, + { + "epoch": 5.32, + "learning_rate": 5.328420112179405e-06, + "loss": 0.094, + "step": 12706 + }, + { + "epoch": 5.32, + "learning_rate": 5.326921143114727e-06, + "loss": 0.1097, + "step": 12707 + }, + { + "epoch": 5.32, + "learning_rate": 5.325422308373557e-06, + "loss": 0.1508, + "step": 12708 + }, + { + "epoch": 5.32, + "learning_rate": 5.3239236079989735e-06, + "loss": 0.0881, + "step": 12709 + }, + { + "epoch": 5.32, + "learning_rate": 5.322425042034056e-06, + "loss": 0.1582, + "step": 12710 + }, + { + "epoch": 5.32, + "learning_rate": 5.3209266105218815e-06, + "loss": 0.1305, + "step": 12711 + }, + { + "epoch": 5.32, + "learning_rate": 5.319428313505523e-06, + "loss": 0.1466, + "step": 12712 + }, + { + "epoch": 5.32, + "learning_rate": 5.317930151028039e-06, + "loss": 0.1536, + "step": 12713 + }, + { + "epoch": 5.32, + "learning_rate": 5.316432123132498e-06, + "loss": 0.1155, + "step": 12714 + }, + { + "epoch": 5.32, + "learning_rate": 5.314934229861961e-06, + "loss": 0.1389, + "step": 12715 + }, + { + "epoch": 5.32, + "learning_rate": 5.3134364712594814e-06, + "loss": 0.1402, + "step": 12716 + }, + { + "epoch": 5.32, + "learning_rate": 5.311938847368111e-06, + "loss": 0.1609, + "step": 12717 + }, + { + "epoch": 5.32, + "learning_rate": 5.310441358230899e-06, + "loss": 0.1318, + "step": 12718 + }, + { + "epoch": 5.33, + "learning_rate": 5.308944003890892e-06, + "loss": 0.1152, + "step": 12719 + }, + { + "epoch": 5.33, + "learning_rate": 5.307446784391124e-06, + "loss": 0.1604, + "step": 12720 + }, + { + "epoch": 5.33, + "learning_rate": 5.305949699774633e-06, + "loss": 0.1279, + "step": 12721 + }, + { + "epoch": 5.33, + "learning_rate": 5.304452750084452e-06, + "loss": 0.1333, + "step": 12722 + }, + { + "epoch": 5.33, + "learning_rate": 5.3029559353636116e-06, + "loss": 0.1252, + "step": 12723 + }, + { + "epoch": 5.33, + "learning_rate": 5.301459255655134e-06, + "loss": 0.1178, + "step": 12724 + }, + { + "epoch": 5.33, + "learning_rate": 5.299962711002045e-06, + "loss": 0.1461, + "step": 12725 + }, + { + "epoch": 5.33, + "learning_rate": 5.298466301447353e-06, + "loss": 0.1093, + "step": 12726 + }, + { + "epoch": 5.33, + "learning_rate": 5.296970027034074e-06, + "loss": 0.1464, + "step": 12727 + }, + { + "epoch": 5.33, + "learning_rate": 5.295473887805223e-06, + "loss": 0.1338, + "step": 12728 + }, + { + "epoch": 5.33, + "learning_rate": 5.293977883803797e-06, + "loss": 0.1077, + "step": 12729 + }, + { + "epoch": 5.33, + "learning_rate": 5.292482015072799e-06, + "loss": 0.1061, + "step": 12730 + }, + { + "epoch": 5.33, + "learning_rate": 5.29098628165523e-06, + "loss": 0.1552, + "step": 12731 + }, + { + "epoch": 5.33, + "learning_rate": 5.289490683594081e-06, + "loss": 0.1318, + "step": 12732 + }, + { + "epoch": 5.33, + "learning_rate": 5.287995220932346e-06, + "loss": 0.1018, + "step": 12733 + }, + { + "epoch": 5.33, + "learning_rate": 5.286499893713003e-06, + "loss": 0.131, + "step": 12734 + }, + { + "epoch": 5.33, + "learning_rate": 5.285004701979037e-06, + "loss": 0.092, + "step": 12735 + }, + { + "epoch": 5.33, + "learning_rate": 5.283509645773429e-06, + "loss": 0.1369, + "step": 12736 + }, + { + "epoch": 5.33, + "learning_rate": 5.282014725139149e-06, + "loss": 0.1574, + "step": 12737 + }, + { + "epoch": 5.33, + "learning_rate": 5.280519940119171e-06, + "loss": 0.1436, + "step": 12738 + }, + { + "epoch": 5.33, + "learning_rate": 5.279025290756461e-06, + "loss": 0.0928, + "step": 12739 + }, + { + "epoch": 5.33, + "learning_rate": 5.2775307770939775e-06, + "loss": 0.1305, + "step": 12740 + }, + { + "epoch": 5.33, + "learning_rate": 5.27603639917468e-06, + "loss": 0.11, + "step": 12741 + }, + { + "epoch": 5.33, + "learning_rate": 5.2745421570415245e-06, + "loss": 0.1133, + "step": 12742 + }, + { + "epoch": 5.34, + "learning_rate": 5.273048050737461e-06, + "loss": 0.0938, + "step": 12743 + }, + { + "epoch": 5.34, + "learning_rate": 5.271554080305439e-06, + "loss": 0.1041, + "step": 12744 + }, + { + "epoch": 5.34, + "learning_rate": 5.270060245788401e-06, + "loss": 0.1204, + "step": 12745 + }, + { + "epoch": 5.34, + "learning_rate": 5.2685665472292805e-06, + "loss": 0.1563, + "step": 12746 + }, + { + "epoch": 5.34, + "learning_rate": 5.2670729846710155e-06, + "loss": 0.1355, + "step": 12747 + }, + { + "epoch": 5.34, + "learning_rate": 5.265579558156537e-06, + "loss": 0.1159, + "step": 12748 + }, + { + "epoch": 5.34, + "learning_rate": 5.264086267728774e-06, + "loss": 0.1063, + "step": 12749 + }, + { + "epoch": 5.34, + "learning_rate": 5.262593113430653e-06, + "loss": 0.1926, + "step": 12750 + }, + { + "epoch": 5.34, + "learning_rate": 5.261100095305085e-06, + "loss": 0.1129, + "step": 12751 + }, + { + "epoch": 5.34, + "learning_rate": 5.259607213394989e-06, + "loss": 0.1329, + "step": 12752 + }, + { + "epoch": 5.34, + "learning_rate": 5.2581144677432805e-06, + "loss": 0.1095, + "step": 12753 + }, + { + "epoch": 5.34, + "learning_rate": 5.256621858392861e-06, + "loss": 0.0983, + "step": 12754 + }, + { + "epoch": 5.34, + "learning_rate": 5.255129385386635e-06, + "loss": 0.1131, + "step": 12755 + }, + { + "epoch": 5.34, + "learning_rate": 5.2536370487675056e-06, + "loss": 0.1182, + "step": 12756 + }, + { + "epoch": 5.34, + "learning_rate": 5.252144848578368e-06, + "loss": 0.101, + "step": 12757 + }, + { + "epoch": 5.34, + "learning_rate": 5.250652784862114e-06, + "loss": 0.1241, + "step": 12758 + }, + { + "epoch": 5.34, + "learning_rate": 5.249160857661628e-06, + "loss": 0.1303, + "step": 12759 + }, + { + "epoch": 5.34, + "learning_rate": 5.247669067019805e-06, + "loss": 0.0913, + "step": 12760 + }, + { + "epoch": 5.34, + "learning_rate": 5.246177412979512e-06, + "loss": 0.0975, + "step": 12761 + }, + { + "epoch": 5.34, + "learning_rate": 5.2446858955836295e-06, + "loss": 0.0928, + "step": 12762 + }, + { + "epoch": 5.34, + "learning_rate": 5.243194514875033e-06, + "loss": 0.1463, + "step": 12763 + }, + { + "epoch": 5.34, + "learning_rate": 5.241703270896589e-06, + "loss": 0.1404, + "step": 12764 + }, + { + "epoch": 5.34, + "learning_rate": 5.240212163691161e-06, + "loss": 0.103, + "step": 12765 + }, + { + "epoch": 5.34, + "learning_rate": 5.2387211933016145e-06, + "loss": 0.1103, + "step": 12766 + }, + { + "epoch": 5.35, + "learning_rate": 5.237230359770799e-06, + "loss": 0.1328, + "step": 12767 + }, + { + "epoch": 5.35, + "learning_rate": 5.235739663141571e-06, + "loss": 0.1178, + "step": 12768 + }, + { + "epoch": 5.35, + "learning_rate": 5.2342491034567775e-06, + "loss": 0.124, + "step": 12769 + }, + { + "epoch": 5.35, + "learning_rate": 5.232758680759264e-06, + "loss": 0.0963, + "step": 12770 + }, + { + "epoch": 5.35, + "learning_rate": 5.231268395091874e-06, + "loss": 0.1234, + "step": 12771 + }, + { + "epoch": 5.35, + "learning_rate": 5.229778246497446e-06, + "loss": 0.117, + "step": 12772 + }, + { + "epoch": 5.35, + "learning_rate": 5.2282882350188035e-06, + "loss": 0.1322, + "step": 12773 + }, + { + "epoch": 5.35, + "learning_rate": 5.226798360698786e-06, + "loss": 0.1263, + "step": 12774 + }, + { + "epoch": 5.35, + "learning_rate": 5.225308623580209e-06, + "loss": 0.1048, + "step": 12775 + }, + { + "epoch": 5.35, + "learning_rate": 5.2238190237059e-06, + "loss": 0.1328, + "step": 12776 + }, + { + "epoch": 5.35, + "learning_rate": 5.222329561118675e-06, + "loss": 0.1141, + "step": 12777 + }, + { + "epoch": 5.35, + "learning_rate": 5.2208402358613465e-06, + "loss": 0.1137, + "step": 12778 + }, + { + "epoch": 5.35, + "learning_rate": 5.219351047976725e-06, + "loss": 0.1288, + "step": 12779 + }, + { + "epoch": 5.35, + "learning_rate": 5.217861997507618e-06, + "loss": 0.1239, + "step": 12780 + }, + { + "epoch": 5.35, + "learning_rate": 5.216373084496822e-06, + "loss": 0.1355, + "step": 12781 + }, + { + "epoch": 5.35, + "learning_rate": 5.214884308987136e-06, + "loss": 0.1093, + "step": 12782 + }, + { + "epoch": 5.35, + "learning_rate": 5.213395671021355e-06, + "loss": 0.1434, + "step": 12783 + }, + { + "epoch": 5.35, + "learning_rate": 5.211907170642267e-06, + "loss": 0.1132, + "step": 12784 + }, + { + "epoch": 5.35, + "learning_rate": 5.210418807892659e-06, + "loss": 0.1423, + "step": 12785 + }, + { + "epoch": 5.35, + "learning_rate": 5.208930582815316e-06, + "loss": 0.1334, + "step": 12786 + }, + { + "epoch": 5.35, + "learning_rate": 5.207442495453008e-06, + "loss": 0.1015, + "step": 12787 + }, + { + "epoch": 5.35, + "learning_rate": 5.205954545848513e-06, + "loss": 0.1216, + "step": 12788 + }, + { + "epoch": 5.35, + "learning_rate": 5.2044667340445985e-06, + "loss": 0.1396, + "step": 12789 + }, + { + "epoch": 5.35, + "learning_rate": 5.202979060084035e-06, + "loss": 0.1208, + "step": 12790 + }, + { + "epoch": 5.36, + "learning_rate": 5.2014915240095785e-06, + "loss": 0.1006, + "step": 12791 + }, + { + "epoch": 5.36, + "learning_rate": 5.200004125863993e-06, + "loss": 0.0823, + "step": 12792 + }, + { + "epoch": 5.36, + "learning_rate": 5.198516865690031e-06, + "loss": 0.1439, + "step": 12793 + }, + { + "epoch": 5.36, + "learning_rate": 5.1970297435304395e-06, + "loss": 0.0832, + "step": 12794 + }, + { + "epoch": 5.36, + "learning_rate": 5.195542759427964e-06, + "loss": 0.1474, + "step": 12795 + }, + { + "epoch": 5.36, + "learning_rate": 5.194055913425353e-06, + "loss": 0.1134, + "step": 12796 + }, + { + "epoch": 5.36, + "learning_rate": 5.192569205565336e-06, + "loss": 0.1257, + "step": 12797 + }, + { + "epoch": 5.36, + "learning_rate": 5.191082635890652e-06, + "loss": 0.0989, + "step": 12798 + }, + { + "epoch": 5.36, + "learning_rate": 5.1895962044440295e-06, + "loss": 0.0935, + "step": 12799 + }, + { + "epoch": 5.36, + "learning_rate": 5.188109911268196e-06, + "loss": 0.1132, + "step": 12800 + }, + { + "epoch": 5.36, + "learning_rate": 5.186623756405875e-06, + "loss": 0.1229, + "step": 12801 + }, + { + "epoch": 5.36, + "learning_rate": 5.185137739899781e-06, + "loss": 0.1153, + "step": 12802 + }, + { + "epoch": 5.36, + "learning_rate": 5.183651861792629e-06, + "loss": 0.125, + "step": 12803 + }, + { + "epoch": 5.36, + "learning_rate": 5.18216612212713e-06, + "loss": 0.1001, + "step": 12804 + }, + { + "epoch": 5.36, + "learning_rate": 5.180680520945992e-06, + "loss": 0.1196, + "step": 12805 + }, + { + "epoch": 5.36, + "learning_rate": 5.179195058291915e-06, + "loss": 0.0949, + "step": 12806 + }, + { + "epoch": 5.36, + "learning_rate": 5.177709734207603e-06, + "loss": 0.1265, + "step": 12807 + }, + { + "epoch": 5.36, + "learning_rate": 5.17622454873574e-06, + "loss": 0.1384, + "step": 12808 + }, + { + "epoch": 5.36, + "learning_rate": 5.174739501919023e-06, + "loss": 0.1639, + "step": 12809 + }, + { + "epoch": 5.36, + "learning_rate": 5.173254593800137e-06, + "loss": 0.1167, + "step": 12810 + }, + { + "epoch": 5.36, + "learning_rate": 5.171769824421764e-06, + "loss": 0.1537, + "step": 12811 + }, + { + "epoch": 5.36, + "learning_rate": 5.170285193826583e-06, + "loss": 0.1342, + "step": 12812 + }, + { + "epoch": 5.36, + "learning_rate": 5.168800702057272e-06, + "loss": 0.1059, + "step": 12813 + }, + { + "epoch": 5.36, + "learning_rate": 5.167316349156495e-06, + "loss": 0.0931, + "step": 12814 + }, + { + "epoch": 5.37, + "learning_rate": 5.16583213516692e-06, + "loss": 0.1204, + "step": 12815 + }, + { + "epoch": 5.37, + "learning_rate": 5.16434806013121e-06, + "loss": 0.1225, + "step": 12816 + }, + { + "epoch": 5.37, + "learning_rate": 5.162864124092024e-06, + "loss": 0.0935, + "step": 12817 + }, + { + "epoch": 5.37, + "learning_rate": 5.161380327092021e-06, + "loss": 0.1368, + "step": 12818 + }, + { + "epoch": 5.37, + "learning_rate": 5.1598966691738425e-06, + "loss": 0.0978, + "step": 12819 + }, + { + "epoch": 5.37, + "learning_rate": 5.158413150380139e-06, + "loss": 0.1169, + "step": 12820 + }, + { + "epoch": 5.37, + "learning_rate": 5.1569297707535585e-06, + "loss": 0.1467, + "step": 12821 + }, + { + "epoch": 5.37, + "learning_rate": 5.155446530336729e-06, + "loss": 0.132, + "step": 12822 + }, + { + "epoch": 5.37, + "learning_rate": 5.15396342917229e-06, + "loss": 0.128, + "step": 12823 + }, + { + "epoch": 5.37, + "learning_rate": 5.1524804673028725e-06, + "loss": 0.1187, + "step": 12824 + }, + { + "epoch": 5.37, + "learning_rate": 5.150997644771102e-06, + "loss": 0.1133, + "step": 12825 + }, + { + "epoch": 5.37, + "learning_rate": 5.149514961619602e-06, + "loss": 0.1066, + "step": 12826 + }, + { + "epoch": 5.37, + "learning_rate": 5.148032417890992e-06, + "loss": 0.1473, + "step": 12827 + }, + { + "epoch": 5.37, + "learning_rate": 5.1465500136278825e-06, + "loss": 0.1187, + "step": 12828 + }, + { + "epoch": 5.37, + "learning_rate": 5.145067748872885e-06, + "loss": 0.1354, + "step": 12829 + }, + { + "epoch": 5.37, + "learning_rate": 5.143585623668607e-06, + "loss": 0.129, + "step": 12830 + }, + { + "epoch": 5.37, + "learning_rate": 5.142103638057652e-06, + "loss": 0.14, + "step": 12831 + }, + { + "epoch": 5.37, + "learning_rate": 5.140621792082616e-06, + "loss": 0.1078, + "step": 12832 + }, + { + "epoch": 5.37, + "learning_rate": 5.139140085786095e-06, + "loss": 0.1107, + "step": 12833 + }, + { + "epoch": 5.37, + "learning_rate": 5.1376585192106824e-06, + "loss": 0.1486, + "step": 12834 + }, + { + "epoch": 5.37, + "learning_rate": 5.136177092398957e-06, + "loss": 0.1297, + "step": 12835 + }, + { + "epoch": 5.37, + "learning_rate": 5.1346958053935045e-06, + "loss": 0.1109, + "step": 12836 + }, + { + "epoch": 5.37, + "learning_rate": 5.133214658236905e-06, + "loss": 0.0966, + "step": 12837 + }, + { + "epoch": 5.37, + "learning_rate": 5.131733650971731e-06, + "loss": 0.1232, + "step": 12838 + }, + { + "epoch": 5.38, + "learning_rate": 5.130252783640553e-06, + "loss": 0.1504, + "step": 12839 + }, + { + "epoch": 5.38, + "learning_rate": 5.12877205628594e-06, + "loss": 0.0922, + "step": 12840 + }, + { + "epoch": 5.38, + "learning_rate": 5.127291468950453e-06, + "loss": 0.1228, + "step": 12841 + }, + { + "epoch": 5.38, + "learning_rate": 5.125811021676643e-06, + "loss": 0.1426, + "step": 12842 + }, + { + "epoch": 5.38, + "learning_rate": 5.124330714507073e-06, + "loss": 0.0989, + "step": 12843 + }, + { + "epoch": 5.38, + "learning_rate": 5.1228505474842885e-06, + "loss": 0.1034, + "step": 12844 + }, + { + "epoch": 5.38, + "learning_rate": 5.121370520650837e-06, + "loss": 0.1275, + "step": 12845 + }, + { + "epoch": 5.38, + "learning_rate": 5.1198906340492605e-06, + "loss": 0.1123, + "step": 12846 + }, + { + "epoch": 5.38, + "learning_rate": 5.118410887722096e-06, + "loss": 0.11, + "step": 12847 + }, + { + "epoch": 5.38, + "learning_rate": 5.116931281711886e-06, + "loss": 0.143, + "step": 12848 + }, + { + "epoch": 5.38, + "learning_rate": 5.115451816061146e-06, + "loss": 0.1328, + "step": 12849 + }, + { + "epoch": 5.38, + "learning_rate": 5.113972490812411e-06, + "loss": 0.1434, + "step": 12850 + }, + { + "epoch": 5.38, + "learning_rate": 5.112493306008199e-06, + "loss": 0.1443, + "step": 12851 + }, + { + "epoch": 5.38, + "learning_rate": 5.111014261691032e-06, + "loss": 0.088, + "step": 12852 + }, + { + "epoch": 5.38, + "learning_rate": 5.109535357903421e-06, + "loss": 0.1044, + "step": 12853 + }, + { + "epoch": 5.38, + "learning_rate": 5.10805659468788e-06, + "loss": 0.1146, + "step": 12854 + }, + { + "epoch": 5.38, + "learning_rate": 5.106577972086907e-06, + "loss": 0.1258, + "step": 12855 + }, + { + "epoch": 5.38, + "learning_rate": 5.105099490143008e-06, + "loss": 0.1459, + "step": 12856 + }, + { + "epoch": 5.38, + "learning_rate": 5.1036211488986806e-06, + "loss": 0.1133, + "step": 12857 + }, + { + "epoch": 5.38, + "learning_rate": 5.102142948396418e-06, + "loss": 0.1259, + "step": 12858 + }, + { + "epoch": 5.38, + "learning_rate": 5.10066488867871e-06, + "loss": 0.0901, + "step": 12859 + }, + { + "epoch": 5.38, + "learning_rate": 5.099186969788044e-06, + "loss": 0.12, + "step": 12860 + }, + { + "epoch": 5.38, + "learning_rate": 5.097709191766902e-06, + "loss": 0.1217, + "step": 12861 + }, + { + "epoch": 5.38, + "learning_rate": 5.096231554657755e-06, + "loss": 0.1196, + "step": 12862 + }, + { + "epoch": 5.39, + "learning_rate": 5.094754058503081e-06, + "loss": 0.1184, + "step": 12863 + }, + { + "epoch": 5.39, + "learning_rate": 5.093276703345351e-06, + "loss": 0.1329, + "step": 12864 + }, + { + "epoch": 5.39, + "learning_rate": 5.091799489227027e-06, + "loss": 0.112, + "step": 12865 + }, + { + "epoch": 5.39, + "learning_rate": 5.09032241619057e-06, + "loss": 0.13, + "step": 12866 + }, + { + "epoch": 5.39, + "learning_rate": 5.088845484278437e-06, + "loss": 0.1355, + "step": 12867 + }, + { + "epoch": 5.39, + "learning_rate": 5.087368693533088e-06, + "loss": 0.1024, + "step": 12868 + }, + { + "epoch": 5.39, + "learning_rate": 5.085892043996962e-06, + "loss": 0.1276, + "step": 12869 + }, + { + "epoch": 5.39, + "learning_rate": 5.084415535712508e-06, + "loss": 0.0943, + "step": 12870 + }, + { + "epoch": 5.39, + "learning_rate": 5.082939168722168e-06, + "loss": 0.1246, + "step": 12871 + }, + { + "epoch": 5.39, + "learning_rate": 5.081462943068378e-06, + "loss": 0.1012, + "step": 12872 + }, + { + "epoch": 5.39, + "learning_rate": 5.079986858793572e-06, + "loss": 0.1261, + "step": 12873 + }, + { + "epoch": 5.39, + "learning_rate": 5.078510915940178e-06, + "loss": 0.1254, + "step": 12874 + }, + { + "epoch": 5.39, + "learning_rate": 5.077035114550624e-06, + "loss": 0.1195, + "step": 12875 + }, + { + "epoch": 5.39, + "learning_rate": 5.075559454667321e-06, + "loss": 0.1738, + "step": 12876 + }, + { + "epoch": 5.39, + "learning_rate": 5.074083936332695e-06, + "loss": 0.132, + "step": 12877 + }, + { + "epoch": 5.39, + "learning_rate": 5.072608559589152e-06, + "loss": 0.1303, + "step": 12878 + }, + { + "epoch": 5.39, + "learning_rate": 5.071133324479105e-06, + "loss": 0.1372, + "step": 12879 + }, + { + "epoch": 5.39, + "learning_rate": 5.069658231044956e-06, + "loss": 0.1013, + "step": 12880 + }, + { + "epoch": 5.39, + "learning_rate": 5.0681832793291105e-06, + "loss": 0.0985, + "step": 12881 + }, + { + "epoch": 5.39, + "learning_rate": 5.066708469373958e-06, + "loss": 0.0946, + "step": 12882 + }, + { + "epoch": 5.39, + "learning_rate": 5.065233801221891e-06, + "loss": 0.1635, + "step": 12883 + }, + { + "epoch": 5.39, + "learning_rate": 5.063759274915299e-06, + "loss": 0.1048, + "step": 12884 + }, + { + "epoch": 5.39, + "learning_rate": 5.062284890496567e-06, + "loss": 0.1156, + "step": 12885 + }, + { + "epoch": 5.4, + "learning_rate": 5.060810648008078e-06, + "loss": 0.0879, + "step": 12886 + }, + { + "epoch": 5.4, + "learning_rate": 5.0593365474921995e-06, + "loss": 0.0964, + "step": 12887 + }, + { + "epoch": 5.4, + "learning_rate": 5.057862588991308e-06, + "loss": 0.1302, + "step": 12888 + }, + { + "epoch": 5.4, + "learning_rate": 5.056388772547775e-06, + "loss": 0.1282, + "step": 12889 + }, + { + "epoch": 5.4, + "learning_rate": 5.054915098203956e-06, + "loss": 0.086, + "step": 12890 + }, + { + "epoch": 5.4, + "learning_rate": 5.053441566002214e-06, + "loss": 0.1227, + "step": 12891 + }, + { + "epoch": 5.4, + "learning_rate": 5.051968175984905e-06, + "loss": 0.1448, + "step": 12892 + }, + { + "epoch": 5.4, + "learning_rate": 5.050494928194379e-06, + "loss": 0.1208, + "step": 12893 + }, + { + "epoch": 5.4, + "learning_rate": 5.0490218226729855e-06, + "loss": 0.1248, + "step": 12894 + }, + { + "epoch": 5.4, + "learning_rate": 5.047548859463069e-06, + "loss": 0.1448, + "step": 12895 + }, + { + "epoch": 5.4, + "learning_rate": 5.046076038606964e-06, + "loss": 0.1254, + "step": 12896 + }, + { + "epoch": 5.4, + "learning_rate": 5.0446033601470045e-06, + "loss": 0.1221, + "step": 12897 + }, + { + "epoch": 5.4, + "learning_rate": 5.043130824125526e-06, + "loss": 0.1134, + "step": 12898 + }, + { + "epoch": 5.4, + "learning_rate": 5.041658430584852e-06, + "loss": 0.1405, + "step": 12899 + }, + { + "epoch": 5.4, + "learning_rate": 5.040186179567308e-06, + "loss": 0.1081, + "step": 12900 + }, + { + "epoch": 5.4, + "learning_rate": 5.03871407111521e-06, + "loss": 0.137, + "step": 12901 + }, + { + "epoch": 5.4, + "learning_rate": 5.037242105270877e-06, + "loss": 0.151, + "step": 12902 + }, + { + "epoch": 5.4, + "learning_rate": 5.035770282076613e-06, + "loss": 0.1329, + "step": 12903 + }, + { + "epoch": 5.4, + "learning_rate": 5.034298601574726e-06, + "loss": 0.117, + "step": 12904 + }, + { + "epoch": 5.4, + "learning_rate": 5.03282706380752e-06, + "loss": 0.091, + "step": 12905 + }, + { + "epoch": 5.4, + "learning_rate": 5.031355668817292e-06, + "loss": 0.1177, + "step": 12906 + }, + { + "epoch": 5.4, + "learning_rate": 5.029884416646335e-06, + "loss": 0.1754, + "step": 12907 + }, + { + "epoch": 5.4, + "learning_rate": 5.0284133073369456e-06, + "loss": 0.122, + "step": 12908 + }, + { + "epoch": 5.4, + "learning_rate": 5.026942340931404e-06, + "loss": 0.1711, + "step": 12909 + }, + { + "epoch": 5.41, + "learning_rate": 5.025471517471987e-06, + "loss": 0.1217, + "step": 12910 + }, + { + "epoch": 5.41, + "learning_rate": 5.0240008370009775e-06, + "loss": 0.1192, + "step": 12911 + }, + { + "epoch": 5.41, + "learning_rate": 5.022530299560648e-06, + "loss": 0.1085, + "step": 12912 + }, + { + "epoch": 5.41, + "learning_rate": 5.021059905193268e-06, + "loss": 0.1179, + "step": 12913 + }, + { + "epoch": 5.41, + "learning_rate": 5.019589653941103e-06, + "loss": 0.1086, + "step": 12914 + }, + { + "epoch": 5.41, + "learning_rate": 5.018119545846414e-06, + "loss": 0.1005, + "step": 12915 + }, + { + "epoch": 5.41, + "learning_rate": 5.016649580951462e-06, + "loss": 0.0988, + "step": 12916 + }, + { + "epoch": 5.41, + "learning_rate": 5.015179759298492e-06, + "loss": 0.135, + "step": 12917 + }, + { + "epoch": 5.41, + "learning_rate": 5.013710080929754e-06, + "loss": 0.1036, + "step": 12918 + }, + { + "epoch": 5.41, + "learning_rate": 5.0122405458874975e-06, + "loss": 0.1485, + "step": 12919 + }, + { + "epoch": 5.41, + "learning_rate": 5.01077115421396e-06, + "loss": 0.0948, + "step": 12920 + }, + { + "epoch": 5.41, + "learning_rate": 5.009301905951378e-06, + "loss": 0.0954, + "step": 12921 + }, + { + "epoch": 5.41, + "learning_rate": 5.007832801141988e-06, + "loss": 0.1289, + "step": 12922 + }, + { + "epoch": 5.41, + "learning_rate": 5.00636383982801e-06, + "loss": 0.1359, + "step": 12923 + }, + { + "epoch": 5.41, + "learning_rate": 5.0048950220516725e-06, + "loss": 0.1199, + "step": 12924 + }, + { + "epoch": 5.41, + "learning_rate": 5.0034263478551935e-06, + "loss": 0.1447, + "step": 12925 + }, + { + "epoch": 5.41, + "learning_rate": 5.001957817280792e-06, + "loss": 0.1321, + "step": 12926 + }, + { + "epoch": 5.41, + "learning_rate": 5.0004894303706775e-06, + "loss": 0.116, + "step": 12927 + }, + { + "epoch": 5.41, + "learning_rate": 4.999021187167061e-06, + "loss": 0.085, + "step": 12928 + }, + { + "epoch": 5.41, + "learning_rate": 4.99755308771214e-06, + "loss": 0.1256, + "step": 12929 + }, + { + "epoch": 5.41, + "learning_rate": 4.996085132048116e-06, + "loss": 0.0945, + "step": 12930 + }, + { + "epoch": 5.41, + "learning_rate": 4.994617320217184e-06, + "loss": 0.1253, + "step": 12931 + }, + { + "epoch": 5.41, + "learning_rate": 4.99314965226154e-06, + "loss": 0.1201, + "step": 12932 + }, + { + "epoch": 5.41, + "learning_rate": 4.991682128223362e-06, + "loss": 0.1346, + "step": 12933 + }, + { + "epoch": 5.42, + "learning_rate": 4.990214748144837e-06, + "loss": 0.1149, + "step": 12934 + }, + { + "epoch": 5.42, + "learning_rate": 4.988747512068143e-06, + "loss": 0.1218, + "step": 12935 + }, + { + "epoch": 5.42, + "learning_rate": 4.9872804200354596e-06, + "loss": 0.0837, + "step": 12936 + }, + { + "epoch": 5.42, + "learning_rate": 4.985813472088947e-06, + "loss": 0.1447, + "step": 12937 + }, + { + "epoch": 5.42, + "learning_rate": 4.984346668270779e-06, + "loss": 0.0828, + "step": 12938 + }, + { + "epoch": 5.42, + "learning_rate": 4.9828800086231135e-06, + "loss": 0.1073, + "step": 12939 + }, + { + "epoch": 5.42, + "learning_rate": 4.98141349318811e-06, + "loss": 0.1124, + "step": 12940 + }, + { + "epoch": 5.42, + "learning_rate": 4.979947122007922e-06, + "loss": 0.115, + "step": 12941 + }, + { + "epoch": 5.42, + "learning_rate": 4.9784808951247e-06, + "loss": 0.0887, + "step": 12942 + }, + { + "epoch": 5.42, + "learning_rate": 4.977014812580593e-06, + "loss": 0.0724, + "step": 12943 + }, + { + "epoch": 5.42, + "learning_rate": 4.975548874417733e-06, + "loss": 0.126, + "step": 12944 + }, + { + "epoch": 5.42, + "learning_rate": 4.9740830806782625e-06, + "loss": 0.1083, + "step": 12945 + }, + { + "epoch": 5.42, + "learning_rate": 4.972617431404313e-06, + "loss": 0.0955, + "step": 12946 + }, + { + "epoch": 5.42, + "learning_rate": 4.971151926638016e-06, + "loss": 0.1092, + "step": 12947 + }, + { + "epoch": 5.42, + "learning_rate": 4.9696865664214936e-06, + "loss": 0.1395, + "step": 12948 + }, + { + "epoch": 5.42, + "learning_rate": 4.96822135079687e-06, + "loss": 0.1324, + "step": 12949 + }, + { + "epoch": 5.42, + "learning_rate": 4.966756279806255e-06, + "loss": 0.0848, + "step": 12950 + }, + { + "epoch": 5.42, + "learning_rate": 4.9652913534917645e-06, + "loss": 0.1021, + "step": 12951 + }, + { + "epoch": 5.42, + "learning_rate": 4.963826571895507e-06, + "loss": 0.11, + "step": 12952 + }, + { + "epoch": 5.42, + "learning_rate": 4.962361935059585e-06, + "loss": 0.1254, + "step": 12953 + }, + { + "epoch": 5.42, + "learning_rate": 4.9608974430261025e-06, + "loss": 0.1299, + "step": 12954 + }, + { + "epoch": 5.42, + "learning_rate": 4.959433095837149e-06, + "loss": 0.1142, + "step": 12955 + }, + { + "epoch": 5.42, + "learning_rate": 4.957968893534816e-06, + "loss": 0.1018, + "step": 12956 + }, + { + "epoch": 5.42, + "learning_rate": 4.956504836161198e-06, + "loss": 0.1073, + "step": 12957 + }, + { + "epoch": 5.43, + "learning_rate": 4.955040923758368e-06, + "loss": 0.1269, + "step": 12958 + }, + { + "epoch": 5.43, + "learning_rate": 4.953577156368412e-06, + "loss": 0.1611, + "step": 12959 + }, + { + "epoch": 5.43, + "learning_rate": 4.952113534033402e-06, + "loss": 0.1238, + "step": 12960 + }, + { + "epoch": 5.43, + "learning_rate": 4.950650056795407e-06, + "loss": 0.122, + "step": 12961 + }, + { + "epoch": 5.43, + "learning_rate": 4.949186724696498e-06, + "loss": 0.1164, + "step": 12962 + }, + { + "epoch": 5.43, + "learning_rate": 4.947723537778737e-06, + "loss": 0.1309, + "step": 12963 + }, + { + "epoch": 5.43, + "learning_rate": 4.946260496084175e-06, + "loss": 0.135, + "step": 12964 + }, + { + "epoch": 5.43, + "learning_rate": 4.9447975996548725e-06, + "loss": 0.1494, + "step": 12965 + }, + { + "epoch": 5.43, + "learning_rate": 4.943334848532876e-06, + "loss": 0.1116, + "step": 12966 + }, + { + "epoch": 5.43, + "learning_rate": 4.9418722427602315e-06, + "loss": 0.126, + "step": 12967 + }, + { + "epoch": 5.43, + "learning_rate": 4.940409782378982e-06, + "loss": 0.1028, + "step": 12968 + }, + { + "epoch": 5.43, + "learning_rate": 4.938947467431167e-06, + "loss": 0.1234, + "step": 12969 + }, + { + "epoch": 5.43, + "learning_rate": 4.937485297958812e-06, + "loss": 0.0987, + "step": 12970 + }, + { + "epoch": 5.43, + "learning_rate": 4.936023274003949e-06, + "loss": 0.1385, + "step": 12971 + }, + { + "epoch": 5.43, + "learning_rate": 4.934561395608603e-06, + "loss": 0.1048, + "step": 12972 + }, + { + "epoch": 5.43, + "learning_rate": 4.933099662814797e-06, + "loss": 0.1374, + "step": 12973 + }, + { + "epoch": 5.43, + "learning_rate": 4.931638075664544e-06, + "loss": 0.1106, + "step": 12974 + }, + { + "epoch": 5.43, + "learning_rate": 4.930176634199856e-06, + "loss": 0.166, + "step": 12975 + }, + { + "epoch": 5.43, + "learning_rate": 4.928715338462746e-06, + "loss": 0.1257, + "step": 12976 + }, + { + "epoch": 5.43, + "learning_rate": 4.927254188495213e-06, + "loss": 0.1514, + "step": 12977 + }, + { + "epoch": 5.43, + "learning_rate": 4.925793184339254e-06, + "loss": 0.1367, + "step": 12978 + }, + { + "epoch": 5.43, + "learning_rate": 4.924332326036867e-06, + "loss": 0.1215, + "step": 12979 + }, + { + "epoch": 5.43, + "learning_rate": 4.922871613630043e-06, + "loss": 0.0959, + "step": 12980 + }, + { + "epoch": 5.43, + "learning_rate": 4.921411047160768e-06, + "loss": 0.1108, + "step": 12981 + }, + { + "epoch": 5.44, + "learning_rate": 4.919950626671028e-06, + "loss": 0.0922, + "step": 12982 + }, + { + "epoch": 5.44, + "learning_rate": 4.918490352202797e-06, + "loss": 0.1245, + "step": 12983 + }, + { + "epoch": 5.44, + "learning_rate": 4.917030223798057e-06, + "loss": 0.1117, + "step": 12984 + }, + { + "epoch": 5.44, + "learning_rate": 4.9155702414987685e-06, + "loss": 0.098, + "step": 12985 + }, + { + "epoch": 5.44, + "learning_rate": 4.914110405346901e-06, + "loss": 0.0783, + "step": 12986 + }, + { + "epoch": 5.44, + "learning_rate": 4.912650715384417e-06, + "loss": 0.1356, + "step": 12987 + }, + { + "epoch": 5.44, + "learning_rate": 4.911191171653273e-06, + "loss": 0.0962, + "step": 12988 + }, + { + "epoch": 5.44, + "learning_rate": 4.909731774195425e-06, + "loss": 0.088, + "step": 12989 + }, + { + "epoch": 5.44, + "learning_rate": 4.908272523052823e-06, + "loss": 0.1381, + "step": 12990 + }, + { + "epoch": 5.44, + "learning_rate": 4.9068134182674044e-06, + "loss": 0.0844, + "step": 12991 + }, + { + "epoch": 5.44, + "learning_rate": 4.905354459881116e-06, + "loss": 0.1077, + "step": 12992 + }, + { + "epoch": 5.44, + "learning_rate": 4.9038956479358915e-06, + "loss": 0.1185, + "step": 12993 + }, + { + "epoch": 5.44, + "learning_rate": 4.9024369824736654e-06, + "loss": 0.1349, + "step": 12994 + }, + { + "epoch": 5.44, + "learning_rate": 4.900978463536363e-06, + "loss": 0.1354, + "step": 12995 + }, + { + "epoch": 5.44, + "learning_rate": 4.8995200911659155e-06, + "loss": 0.0975, + "step": 12996 + }, + { + "epoch": 5.44, + "learning_rate": 4.898061865404232e-06, + "loss": 0.1147, + "step": 12997 + }, + { + "epoch": 5.44, + "learning_rate": 4.896603786293233e-06, + "loss": 0.1106, + "step": 12998 + }, + { + "epoch": 5.44, + "learning_rate": 4.89514585387483e-06, + "loss": 0.0958, + "step": 12999 + }, + { + "epoch": 5.44, + "learning_rate": 4.893688068190933e-06, + "loss": 0.0883, + "step": 13000 + }, + { + "epoch": 5.44, + "learning_rate": 4.892230429283438e-06, + "loss": 0.0716, + "step": 13001 + }, + { + "epoch": 5.44, + "learning_rate": 4.890772937194246e-06, + "loss": 0.1229, + "step": 13002 + }, + { + "epoch": 5.44, + "learning_rate": 4.889315591965251e-06, + "loss": 0.1199, + "step": 13003 + }, + { + "epoch": 5.44, + "learning_rate": 4.88785839363835e-06, + "loss": 0.1117, + "step": 13004 + }, + { + "epoch": 5.44, + "learning_rate": 4.886401342255418e-06, + "loss": 0.1406, + "step": 13005 + }, + { + "epoch": 5.45, + "learning_rate": 4.884944437858342e-06, + "loss": 0.1029, + "step": 13006 + }, + { + "epoch": 5.45, + "learning_rate": 4.883487680489e-06, + "loss": 0.129, + "step": 13007 + }, + { + "epoch": 5.45, + "learning_rate": 4.882031070189262e-06, + "loss": 0.1123, + "step": 13008 + }, + { + "epoch": 5.45, + "learning_rate": 4.880574607001001e-06, + "loss": 0.1319, + "step": 13009 + }, + { + "epoch": 5.45, + "learning_rate": 4.879118290966085e-06, + "loss": 0.0958, + "step": 13010 + }, + { + "epoch": 5.45, + "learning_rate": 4.877662122126364e-06, + "loss": 0.1183, + "step": 13011 + }, + { + "epoch": 5.45, + "learning_rate": 4.876206100523701e-06, + "loss": 0.1054, + "step": 13012 + }, + { + "epoch": 5.45, + "learning_rate": 4.874750226199945e-06, + "loss": 0.0947, + "step": 13013 + }, + { + "epoch": 5.45, + "learning_rate": 4.873294499196946e-06, + "loss": 0.0976, + "step": 13014 + }, + { + "epoch": 5.45, + "learning_rate": 4.871838919556549e-06, + "loss": 0.1035, + "step": 13015 + }, + { + "epoch": 5.45, + "learning_rate": 4.87038348732059e-06, + "loss": 0.1318, + "step": 13016 + }, + { + "epoch": 5.45, + "learning_rate": 4.868928202530911e-06, + "loss": 0.1104, + "step": 13017 + }, + { + "epoch": 5.45, + "learning_rate": 4.867473065229332e-06, + "loss": 0.1334, + "step": 13018 + }, + { + "epoch": 5.45, + "learning_rate": 4.866018075457687e-06, + "loss": 0.0944, + "step": 13019 + }, + { + "epoch": 5.45, + "learning_rate": 4.864563233257795e-06, + "loss": 0.1089, + "step": 13020 + }, + { + "epoch": 5.45, + "learning_rate": 4.863108538671477e-06, + "loss": 0.1105, + "step": 13021 + }, + { + "epoch": 5.45, + "learning_rate": 4.861653991740549e-06, + "loss": 0.0989, + "step": 13022 + }, + { + "epoch": 5.45, + "learning_rate": 4.8601995925068136e-06, + "loss": 0.1159, + "step": 13023 + }, + { + "epoch": 5.45, + "learning_rate": 4.85874534101208e-06, + "loss": 0.1002, + "step": 13024 + }, + { + "epoch": 5.45, + "learning_rate": 4.8572912372981526e-06, + "loss": 0.1181, + "step": 13025 + }, + { + "epoch": 5.45, + "learning_rate": 4.855837281406821e-06, + "loss": 0.1238, + "step": 13026 + }, + { + "epoch": 5.45, + "learning_rate": 4.8543834733798826e-06, + "loss": 0.1285, + "step": 13027 + }, + { + "epoch": 5.45, + "learning_rate": 4.852929813259125e-06, + "loss": 0.1352, + "step": 13028 + }, + { + "epoch": 5.45, + "learning_rate": 4.851476301086333e-06, + "loss": 0.1426, + "step": 13029 + }, + { + "epoch": 5.46, + "learning_rate": 4.850022936903285e-06, + "loss": 0.1417, + "step": 13030 + }, + { + "epoch": 5.46, + "learning_rate": 4.848569720751762e-06, + "loss": 0.1086, + "step": 13031 + }, + { + "epoch": 5.46, + "learning_rate": 4.847116652673527e-06, + "loss": 0.128, + "step": 13032 + }, + { + "epoch": 5.46, + "learning_rate": 4.845663732710352e-06, + "loss": 0.1113, + "step": 13033 + }, + { + "epoch": 5.46, + "learning_rate": 4.844210960903998e-06, + "loss": 0.1092, + "step": 13034 + }, + { + "epoch": 5.46, + "learning_rate": 4.842758337296225e-06, + "loss": 0.0939, + "step": 13035 + }, + { + "epoch": 5.46, + "learning_rate": 4.841305861928787e-06, + "loss": 0.1224, + "step": 13036 + }, + { + "epoch": 5.46, + "learning_rate": 4.839853534843439e-06, + "loss": 0.131, + "step": 13037 + }, + { + "epoch": 5.46, + "learning_rate": 4.838401356081918e-06, + "loss": 0.1202, + "step": 13038 + }, + { + "epoch": 5.46, + "learning_rate": 4.83694932568597e-06, + "loss": 0.0958, + "step": 13039 + }, + { + "epoch": 5.46, + "learning_rate": 4.835497443697332e-06, + "loss": 0.1044, + "step": 13040 + }, + { + "epoch": 5.46, + "learning_rate": 4.834045710157735e-06, + "loss": 0.1014, + "step": 13041 + }, + { + "epoch": 5.46, + "learning_rate": 4.832594125108912e-06, + "loss": 0.1027, + "step": 13042 + }, + { + "epoch": 5.46, + "learning_rate": 4.831142688592585e-06, + "loss": 0.1258, + "step": 13043 + }, + { + "epoch": 5.46, + "learning_rate": 4.829691400650478e-06, + "loss": 0.1244, + "step": 13044 + }, + { + "epoch": 5.46, + "learning_rate": 4.828240261324304e-06, + "loss": 0.0685, + "step": 13045 + }, + { + "epoch": 5.46, + "learning_rate": 4.826789270655771e-06, + "loss": 0.1399, + "step": 13046 + }, + { + "epoch": 5.46, + "learning_rate": 4.825338428686589e-06, + "loss": 0.1374, + "step": 13047 + }, + { + "epoch": 5.46, + "learning_rate": 4.823887735458462e-06, + "loss": 0.1102, + "step": 13048 + }, + { + "epoch": 5.46, + "learning_rate": 4.82243719101309e-06, + "loss": 0.1116, + "step": 13049 + }, + { + "epoch": 5.46, + "learning_rate": 4.820986795392165e-06, + "loss": 0.1159, + "step": 13050 + }, + { + "epoch": 5.46, + "learning_rate": 4.819536548637383e-06, + "loss": 0.1404, + "step": 13051 + }, + { + "epoch": 5.46, + "learning_rate": 4.818086450790423e-06, + "loss": 0.1188, + "step": 13052 + }, + { + "epoch": 5.46, + "learning_rate": 4.816636501892968e-06, + "loss": 0.0973, + "step": 13053 + }, + { + "epoch": 5.47, + "learning_rate": 4.8151867019866985e-06, + "loss": 0.093, + "step": 13054 + }, + { + "epoch": 5.47, + "learning_rate": 4.813737051113286e-06, + "loss": 0.143, + "step": 13055 + }, + { + "epoch": 5.47, + "learning_rate": 4.812287549314399e-06, + "loss": 0.1674, + "step": 13056 + }, + { + "epoch": 5.47, + "learning_rate": 4.810838196631704e-06, + "loss": 0.1127, + "step": 13057 + }, + { + "epoch": 5.47, + "learning_rate": 4.809388993106864e-06, + "loss": 0.1456, + "step": 13058 + }, + { + "epoch": 5.47, + "learning_rate": 4.807939938781526e-06, + "loss": 0.1201, + "step": 13059 + }, + { + "epoch": 5.47, + "learning_rate": 4.806491033697348e-06, + "loss": 0.1025, + "step": 13060 + }, + { + "epoch": 5.47, + "learning_rate": 4.805042277895977e-06, + "loss": 0.1205, + "step": 13061 + }, + { + "epoch": 5.47, + "learning_rate": 4.803593671419056e-06, + "loss": 0.1222, + "step": 13062 + }, + { + "epoch": 5.47, + "learning_rate": 4.802145214308223e-06, + "loss": 0.1502, + "step": 13063 + }, + { + "epoch": 5.47, + "learning_rate": 4.800696906605117e-06, + "loss": 0.1467, + "step": 13064 + }, + { + "epoch": 5.47, + "learning_rate": 4.799248748351362e-06, + "loss": 0.1273, + "step": 13065 + }, + { + "epoch": 5.47, + "learning_rate": 4.797800739588585e-06, + "loss": 0.1242, + "step": 13066 + }, + { + "epoch": 5.47, + "learning_rate": 4.796352880358411e-06, + "loss": 0.1119, + "step": 13067 + }, + { + "epoch": 5.47, + "learning_rate": 4.794905170702458e-06, + "loss": 0.1283, + "step": 13068 + }, + { + "epoch": 5.47, + "learning_rate": 4.793457610662334e-06, + "loss": 0.1104, + "step": 13069 + }, + { + "epoch": 5.47, + "learning_rate": 4.792010200279651e-06, + "loss": 0.0905, + "step": 13070 + }, + { + "epoch": 5.47, + "learning_rate": 4.790562939596013e-06, + "loss": 0.0942, + "step": 13071 + }, + { + "epoch": 5.47, + "learning_rate": 4.789115828653025e-06, + "loss": 0.0938, + "step": 13072 + }, + { + "epoch": 5.47, + "learning_rate": 4.787668867492274e-06, + "loss": 0.112, + "step": 13073 + }, + { + "epoch": 5.47, + "learning_rate": 4.786222056155356e-06, + "loss": 0.1225, + "step": 13074 + }, + { + "epoch": 5.47, + "learning_rate": 4.784775394683859e-06, + "loss": 0.0973, + "step": 13075 + }, + { + "epoch": 5.47, + "learning_rate": 4.783328883119365e-06, + "loss": 0.1067, + "step": 13076 + }, + { + "epoch": 5.47, + "learning_rate": 4.781882521503452e-06, + "loss": 0.1221, + "step": 13077 + }, + { + "epoch": 5.48, + "learning_rate": 4.780436309877701e-06, + "loss": 0.0798, + "step": 13078 + }, + { + "epoch": 5.48, + "learning_rate": 4.778990248283671e-06, + "loss": 0.1351, + "step": 13079 + }, + { + "epoch": 5.48, + "learning_rate": 4.777544336762935e-06, + "loss": 0.1175, + "step": 13080 + }, + { + "epoch": 5.48, + "learning_rate": 4.776098575357052e-06, + "loss": 0.1549, + "step": 13081 + }, + { + "epoch": 5.48, + "learning_rate": 4.7746529641075785e-06, + "loss": 0.1547, + "step": 13082 + }, + { + "epoch": 5.48, + "learning_rate": 4.773207503056071e-06, + "loss": 0.0962, + "step": 13083 + }, + { + "epoch": 5.48, + "learning_rate": 4.771762192244075e-06, + "loss": 0.0888, + "step": 13084 + }, + { + "epoch": 5.48, + "learning_rate": 4.770317031713138e-06, + "loss": 0.1445, + "step": 13085 + }, + { + "epoch": 5.48, + "learning_rate": 4.768872021504795e-06, + "loss": 0.156, + "step": 13086 + }, + { + "epoch": 5.48, + "learning_rate": 4.7674271616605816e-06, + "loss": 0.1104, + "step": 13087 + }, + { + "epoch": 5.48, + "learning_rate": 4.765982452222032e-06, + "loss": 0.1365, + "step": 13088 + }, + { + "epoch": 5.48, + "learning_rate": 4.764537893230673e-06, + "loss": 0.1265, + "step": 13089 + }, + { + "epoch": 5.48, + "learning_rate": 4.76309348472803e-06, + "loss": 0.1443, + "step": 13090 + }, + { + "epoch": 5.48, + "learning_rate": 4.7616492267556134e-06, + "loss": 0.0837, + "step": 13091 + }, + { + "epoch": 5.48, + "learning_rate": 4.760205119354946e-06, + "loss": 0.1026, + "step": 13092 + }, + { + "epoch": 5.48, + "learning_rate": 4.758761162567528e-06, + "loss": 0.141, + "step": 13093 + }, + { + "epoch": 5.48, + "learning_rate": 4.757317356434869e-06, + "loss": 0.1503, + "step": 13094 + }, + { + "epoch": 5.48, + "learning_rate": 4.755873700998471e-06, + "loss": 0.1125, + "step": 13095 + }, + { + "epoch": 5.48, + "learning_rate": 4.75443019629983e-06, + "loss": 0.1134, + "step": 13096 + }, + { + "epoch": 5.48, + "learning_rate": 4.7529868423804385e-06, + "loss": 0.1306, + "step": 13097 + }, + { + "epoch": 5.48, + "learning_rate": 4.751543639281784e-06, + "loss": 0.1281, + "step": 13098 + }, + { + "epoch": 5.48, + "learning_rate": 4.750100587045353e-06, + "loss": 0.1129, + "step": 13099 + }, + { + "epoch": 5.48, + "learning_rate": 4.748657685712619e-06, + "loss": 0.11, + "step": 13100 + }, + { + "epoch": 5.49, + "learning_rate": 4.74721493532506e-06, + "loss": 0.1665, + "step": 13101 + }, + { + "epoch": 5.49, + "learning_rate": 4.745772335924146e-06, + "loss": 0.115, + "step": 13102 + }, + { + "epoch": 5.49, + "learning_rate": 4.744329887551343e-06, + "loss": 0.1227, + "step": 13103 + }, + { + "epoch": 5.49, + "learning_rate": 4.742887590248114e-06, + "loss": 0.1139, + "step": 13104 + }, + { + "epoch": 5.49, + "learning_rate": 4.741445444055921e-06, + "loss": 0.1154, + "step": 13105 + }, + { + "epoch": 5.49, + "learning_rate": 4.740003449016208e-06, + "loss": 0.0991, + "step": 13106 + }, + { + "epoch": 5.49, + "learning_rate": 4.7385616051704275e-06, + "loss": 0.1196, + "step": 13107 + }, + { + "epoch": 5.49, + "learning_rate": 4.737119912560025e-06, + "loss": 0.11, + "step": 13108 + }, + { + "epoch": 5.49, + "learning_rate": 4.7356783712264405e-06, + "loss": 0.117, + "step": 13109 + }, + { + "epoch": 5.49, + "learning_rate": 4.734236981211111e-06, + "loss": 0.0986, + "step": 13110 + }, + { + "epoch": 5.49, + "learning_rate": 4.73279574255547e-06, + "loss": 0.1196, + "step": 13111 + }, + { + "epoch": 5.49, + "learning_rate": 4.731354655300936e-06, + "loss": 0.0981, + "step": 13112 + }, + { + "epoch": 5.49, + "learning_rate": 4.729913719488942e-06, + "loss": 0.0829, + "step": 13113 + }, + { + "epoch": 5.49, + "learning_rate": 4.7284729351608985e-06, + "loss": 0.1425, + "step": 13114 + }, + { + "epoch": 5.49, + "learning_rate": 4.727032302358221e-06, + "loss": 0.1211, + "step": 13115 + }, + { + "epoch": 5.49, + "learning_rate": 4.725591821122322e-06, + "loss": 0.0995, + "step": 13116 + }, + { + "epoch": 5.49, + "learning_rate": 4.724151491494605e-06, + "loss": 0.1103, + "step": 13117 + }, + { + "epoch": 5.49, + "learning_rate": 4.722711313516473e-06, + "loss": 0.1047, + "step": 13118 + }, + { + "epoch": 5.49, + "learning_rate": 4.721271287229323e-06, + "loss": 0.1315, + "step": 13119 + }, + { + "epoch": 5.49, + "learning_rate": 4.7198314126745424e-06, + "loss": 0.1178, + "step": 13120 + }, + { + "epoch": 5.49, + "learning_rate": 4.7183916898935234e-06, + "loss": 0.1067, + "step": 13121 + }, + { + "epoch": 5.49, + "learning_rate": 4.716952118927648e-06, + "loss": 0.1286, + "step": 13122 + }, + { + "epoch": 5.49, + "learning_rate": 4.715512699818295e-06, + "loss": 0.103, + "step": 13123 + }, + { + "epoch": 5.49, + "learning_rate": 4.714073432606841e-06, + "loss": 0.1046, + "step": 13124 + }, + { + "epoch": 5.5, + "learning_rate": 4.712634317334655e-06, + "loss": 0.151, + "step": 13125 + }, + { + "epoch": 5.5, + "learning_rate": 4.711195354043109e-06, + "loss": 0.1192, + "step": 13126 + }, + { + "epoch": 5.5, + "learning_rate": 4.709756542773554e-06, + "loss": 0.125, + "step": 13127 + }, + { + "epoch": 5.5, + "learning_rate": 4.708317883567353e-06, + "loss": 0.1561, + "step": 13128 + }, + { + "epoch": 5.5, + "learning_rate": 4.706879376465858e-06, + "loss": 0.1347, + "step": 13129 + }, + { + "epoch": 5.5, + "learning_rate": 4.70544102151042e-06, + "loss": 0.1131, + "step": 13130 + }, + { + "epoch": 5.5, + "learning_rate": 4.7040028187423795e-06, + "loss": 0.0858, + "step": 13131 + }, + { + "epoch": 5.5, + "learning_rate": 4.702564768203083e-06, + "loss": 0.1196, + "step": 13132 + }, + { + "epoch": 5.5, + "learning_rate": 4.701126869933858e-06, + "loss": 0.1426, + "step": 13133 + }, + { + "epoch": 5.5, + "learning_rate": 4.699689123976039e-06, + "loss": 0.1101, + "step": 13134 + }, + { + "epoch": 5.5, + "learning_rate": 4.698251530370956e-06, + "loss": 0.1094, + "step": 13135 + }, + { + "epoch": 5.5, + "learning_rate": 4.696814089159924e-06, + "loss": 0.1387, + "step": 13136 + }, + { + "epoch": 5.5, + "learning_rate": 4.695376800384266e-06, + "loss": 0.1346, + "step": 13137 + }, + { + "epoch": 5.5, + "learning_rate": 4.6939396640852945e-06, + "loss": 0.1599, + "step": 13138 + }, + { + "epoch": 5.5, + "learning_rate": 4.6925026803043186e-06, + "loss": 0.1209, + "step": 13139 + }, + { + "epoch": 5.5, + "learning_rate": 4.691065849082647e-06, + "loss": 0.1277, + "step": 13140 + }, + { + "epoch": 5.5, + "learning_rate": 4.6896291704615725e-06, + "loss": 0.1685, + "step": 13141 + }, + { + "epoch": 5.5, + "learning_rate": 4.688192644482396e-06, + "loss": 0.1097, + "step": 13142 + }, + { + "epoch": 5.5, + "learning_rate": 4.686756271186407e-06, + "loss": 0.139, + "step": 13143 + }, + { + "epoch": 5.5, + "learning_rate": 4.6853200506148955e-06, + "loss": 0.1138, + "step": 13144 + }, + { + "epoch": 5.5, + "learning_rate": 4.683883982809142e-06, + "loss": 0.1214, + "step": 13145 + }, + { + "epoch": 5.5, + "learning_rate": 4.68244806781043e-06, + "loss": 0.1228, + "step": 13146 + }, + { + "epoch": 5.5, + "learning_rate": 4.681012305660026e-06, + "loss": 0.097, + "step": 13147 + }, + { + "epoch": 5.5, + "learning_rate": 4.679576696399203e-06, + "loss": 0.1354, + "step": 13148 + }, + { + "epoch": 5.51, + "learning_rate": 4.678141240069227e-06, + "loss": 0.1367, + "step": 13149 + }, + { + "epoch": 5.51, + "learning_rate": 4.676705936711358e-06, + "loss": 0.0956, + "step": 13150 + }, + { + "epoch": 5.51, + "learning_rate": 4.675270786366853e-06, + "loss": 0.1226, + "step": 13151 + }, + { + "epoch": 5.51, + "learning_rate": 4.673835789076969e-06, + "loss": 0.1573, + "step": 13152 + }, + { + "epoch": 5.51, + "learning_rate": 4.672400944882944e-06, + "loss": 0.1212, + "step": 13153 + }, + { + "epoch": 5.51, + "learning_rate": 4.670966253826027e-06, + "loss": 0.1723, + "step": 13154 + }, + { + "epoch": 5.51, + "learning_rate": 4.669531715947455e-06, + "loss": 0.1733, + "step": 13155 + }, + { + "epoch": 5.51, + "learning_rate": 4.668097331288465e-06, + "loss": 0.1056, + "step": 13156 + }, + { + "epoch": 5.51, + "learning_rate": 4.666663099890285e-06, + "loss": 0.088, + "step": 13157 + }, + { + "epoch": 5.51, + "learning_rate": 4.665229021794145e-06, + "loss": 0.1706, + "step": 13158 + }, + { + "epoch": 5.51, + "learning_rate": 4.66379509704126e-06, + "loss": 0.0982, + "step": 13159 + }, + { + "epoch": 5.51, + "learning_rate": 4.662361325672853e-06, + "loss": 0.1061, + "step": 13160 + }, + { + "epoch": 5.51, + "learning_rate": 4.660927707730129e-06, + "loss": 0.0863, + "step": 13161 + }, + { + "epoch": 5.51, + "learning_rate": 4.659494243254301e-06, + "loss": 0.1444, + "step": 13162 + }, + { + "epoch": 5.51, + "learning_rate": 4.658060932286571e-06, + "loss": 0.0896, + "step": 13163 + }, + { + "epoch": 5.51, + "learning_rate": 4.65662777486814e-06, + "loss": 0.1124, + "step": 13164 + }, + { + "epoch": 5.51, + "learning_rate": 4.6551947710402016e-06, + "loss": 0.13, + "step": 13165 + }, + { + "epoch": 5.51, + "learning_rate": 4.653761920843946e-06, + "loss": 0.1145, + "step": 13166 + }, + { + "epoch": 5.51, + "learning_rate": 4.652329224320564e-06, + "loss": 0.1004, + "step": 13167 + }, + { + "epoch": 5.51, + "learning_rate": 4.65089668151123e-06, + "loss": 0.0881, + "step": 13168 + }, + { + "epoch": 5.51, + "learning_rate": 4.649464292457123e-06, + "loss": 0.1377, + "step": 13169 + }, + { + "epoch": 5.51, + "learning_rate": 4.648032057199418e-06, + "loss": 0.107, + "step": 13170 + }, + { + "epoch": 5.51, + "learning_rate": 4.6465999757792815e-06, + "loss": 0.1474, + "step": 13171 + }, + { + "epoch": 5.51, + "learning_rate": 4.645168048237878e-06, + "loss": 0.145, + "step": 13172 + }, + { + "epoch": 5.52, + "learning_rate": 4.643736274616372e-06, + "loss": 0.1343, + "step": 13173 + }, + { + "epoch": 5.52, + "learning_rate": 4.6423046549559095e-06, + "loss": 0.1351, + "step": 13174 + }, + { + "epoch": 5.52, + "learning_rate": 4.640873189297645e-06, + "loss": 0.1167, + "step": 13175 + }, + { + "epoch": 5.52, + "learning_rate": 4.639441877682724e-06, + "loss": 0.1504, + "step": 13176 + }, + { + "epoch": 5.52, + "learning_rate": 4.638010720152292e-06, + "loss": 0.1185, + "step": 13177 + }, + { + "epoch": 5.52, + "learning_rate": 4.6365797167474815e-06, + "loss": 0.1556, + "step": 13178 + }, + { + "epoch": 5.52, + "learning_rate": 4.635148867509432e-06, + "loss": 0.1527, + "step": 13179 + }, + { + "epoch": 5.52, + "learning_rate": 4.6337181724792644e-06, + "loss": 0.1212, + "step": 13180 + }, + { + "epoch": 5.52, + "learning_rate": 4.6322876316981095e-06, + "loss": 0.1378, + "step": 13181 + }, + { + "epoch": 5.52, + "learning_rate": 4.630857245207079e-06, + "loss": 0.1086, + "step": 13182 + }, + { + "epoch": 5.52, + "learning_rate": 4.629427013047293e-06, + "loss": 0.1517, + "step": 13183 + }, + { + "epoch": 5.52, + "learning_rate": 4.627996935259862e-06, + "loss": 0.1087, + "step": 13184 + }, + { + "epoch": 5.52, + "learning_rate": 4.626567011885892e-06, + "loss": 0.1032, + "step": 13185 + }, + { + "epoch": 5.52, + "learning_rate": 4.625137242966484e-06, + "loss": 0.0961, + "step": 13186 + }, + { + "epoch": 5.52, + "learning_rate": 4.623707628542742e-06, + "loss": 0.1232, + "step": 13187 + }, + { + "epoch": 5.52, + "learning_rate": 4.6222781686557485e-06, + "loss": 0.1486, + "step": 13188 + }, + { + "epoch": 5.52, + "learning_rate": 4.620848863346596e-06, + "loss": 0.157, + "step": 13189 + }, + { + "epoch": 5.52, + "learning_rate": 4.6194197126563715e-06, + "loss": 0.1104, + "step": 13190 + }, + { + "epoch": 5.52, + "learning_rate": 4.617990716626151e-06, + "loss": 0.0985, + "step": 13191 + }, + { + "epoch": 5.52, + "learning_rate": 4.616561875297012e-06, + "loss": 0.0959, + "step": 13192 + }, + { + "epoch": 5.52, + "learning_rate": 4.615133188710029e-06, + "loss": 0.0781, + "step": 13193 + }, + { + "epoch": 5.52, + "learning_rate": 4.6137046569062595e-06, + "loss": 0.1041, + "step": 13194 + }, + { + "epoch": 5.52, + "learning_rate": 4.612276279926771e-06, + "loss": 0.1304, + "step": 13195 + }, + { + "epoch": 5.52, + "learning_rate": 4.610848057812619e-06, + "loss": 0.1292, + "step": 13196 + }, + { + "epoch": 5.53, + "learning_rate": 4.609419990604858e-06, + "loss": 0.1582, + "step": 13197 + }, + { + "epoch": 5.53, + "learning_rate": 4.607992078344534e-06, + "loss": 0.1267, + "step": 13198 + }, + { + "epoch": 5.53, + "learning_rate": 4.6065643210726955e-06, + "loss": 0.0792, + "step": 13199 + }, + { + "epoch": 5.53, + "learning_rate": 4.605136718830382e-06, + "loss": 0.1392, + "step": 13200 + }, + { + "epoch": 5.53, + "learning_rate": 4.603709271658621e-06, + "loss": 0.1323, + "step": 13201 + }, + { + "epoch": 5.53, + "learning_rate": 4.602281979598451e-06, + "loss": 0.1288, + "step": 13202 + }, + { + "epoch": 5.53, + "learning_rate": 4.600854842690898e-06, + "loss": 0.1464, + "step": 13203 + }, + { + "epoch": 5.53, + "learning_rate": 4.599427860976978e-06, + "loss": 0.092, + "step": 13204 + }, + { + "epoch": 5.53, + "learning_rate": 4.598001034497712e-06, + "loss": 0.1516, + "step": 13205 + }, + { + "epoch": 5.53, + "learning_rate": 4.596574363294113e-06, + "loss": 0.1023, + "step": 13206 + }, + { + "epoch": 5.53, + "learning_rate": 4.595147847407187e-06, + "loss": 0.173, + "step": 13207 + }, + { + "epoch": 5.53, + "learning_rate": 4.593721486877945e-06, + "loss": 0.1068, + "step": 13208 + }, + { + "epoch": 5.53, + "learning_rate": 4.592295281747379e-06, + "loss": 0.1424, + "step": 13209 + }, + { + "epoch": 5.53, + "learning_rate": 4.590869232056485e-06, + "loss": 0.1042, + "step": 13210 + }, + { + "epoch": 5.53, + "learning_rate": 4.589443337846256e-06, + "loss": 0.1413, + "step": 13211 + }, + { + "epoch": 5.53, + "learning_rate": 4.588017599157675e-06, + "loss": 0.1487, + "step": 13212 + }, + { + "epoch": 5.53, + "learning_rate": 4.5865920160317266e-06, + "loss": 0.1182, + "step": 13213 + }, + { + "epoch": 5.53, + "learning_rate": 4.585166588509391e-06, + "loss": 0.1448, + "step": 13214 + }, + { + "epoch": 5.53, + "learning_rate": 4.5837413166316335e-06, + "loss": 0.1197, + "step": 13215 + }, + { + "epoch": 5.53, + "learning_rate": 4.582316200439425e-06, + "loss": 0.1316, + "step": 13216 + }, + { + "epoch": 5.53, + "learning_rate": 4.580891239973729e-06, + "loss": 0.1344, + "step": 13217 + }, + { + "epoch": 5.53, + "learning_rate": 4.579466435275506e-06, + "loss": 0.1287, + "step": 13218 + }, + { + "epoch": 5.53, + "learning_rate": 4.578041786385711e-06, + "loss": 0.1488, + "step": 13219 + }, + { + "epoch": 5.53, + "learning_rate": 4.576617293345296e-06, + "loss": 0.1474, + "step": 13220 + }, + { + "epoch": 5.54, + "learning_rate": 4.575192956195201e-06, + "loss": 0.1238, + "step": 13221 + }, + { + "epoch": 5.54, + "learning_rate": 4.573768774976371e-06, + "loss": 0.1045, + "step": 13222 + }, + { + "epoch": 5.54, + "learning_rate": 4.5723447497297424e-06, + "loss": 0.1188, + "step": 13223 + }, + { + "epoch": 5.54, + "learning_rate": 4.570920880496247e-06, + "loss": 0.1278, + "step": 13224 + }, + { + "epoch": 5.54, + "learning_rate": 4.569497167316813e-06, + "loss": 0.1062, + "step": 13225 + }, + { + "epoch": 5.54, + "learning_rate": 4.568073610232368e-06, + "loss": 0.1161, + "step": 13226 + }, + { + "epoch": 5.54, + "learning_rate": 4.566650209283824e-06, + "loss": 0.1178, + "step": 13227 + }, + { + "epoch": 5.54, + "learning_rate": 4.565226964512102e-06, + "loss": 0.1045, + "step": 13228 + }, + { + "epoch": 5.54, + "learning_rate": 4.563803875958103e-06, + "loss": 0.112, + "step": 13229 + }, + { + "epoch": 5.54, + "learning_rate": 4.562380943662739e-06, + "loss": 0.1019, + "step": 13230 + }, + { + "epoch": 5.54, + "learning_rate": 4.560958167666908e-06, + "loss": 0.1122, + "step": 13231 + }, + { + "epoch": 5.54, + "learning_rate": 4.5595355480115086e-06, + "loss": 0.1071, + "step": 13232 + }, + { + "epoch": 5.54, + "learning_rate": 4.558113084737434e-06, + "loss": 0.1029, + "step": 13233 + }, + { + "epoch": 5.54, + "learning_rate": 4.556690777885571e-06, + "loss": 0.1232, + "step": 13234 + }, + { + "epoch": 5.54, + "learning_rate": 4.5552686274967984e-06, + "loss": 0.1612, + "step": 13235 + }, + { + "epoch": 5.54, + "learning_rate": 4.553846633611997e-06, + "loss": 0.1484, + "step": 13236 + }, + { + "epoch": 5.54, + "learning_rate": 4.552424796272041e-06, + "loss": 0.072, + "step": 13237 + }, + { + "epoch": 5.54, + "learning_rate": 4.551003115517801e-06, + "loss": 0.1069, + "step": 13238 + }, + { + "epoch": 5.54, + "learning_rate": 4.549581591390142e-06, + "loss": 0.1268, + "step": 13239 + }, + { + "epoch": 5.54, + "learning_rate": 4.548160223929923e-06, + "loss": 0.099, + "step": 13240 + }, + { + "epoch": 5.54, + "learning_rate": 4.546739013178003e-06, + "loss": 0.1366, + "step": 13241 + }, + { + "epoch": 5.54, + "learning_rate": 4.545317959175228e-06, + "loss": 0.1154, + "step": 13242 + }, + { + "epoch": 5.54, + "learning_rate": 4.543897061962448e-06, + "loss": 0.1007, + "step": 13243 + }, + { + "epoch": 5.54, + "learning_rate": 4.542476321580505e-06, + "loss": 0.1371, + "step": 13244 + }, + { + "epoch": 5.55, + "learning_rate": 4.541055738070237e-06, + "loss": 0.1588, + "step": 13245 + }, + { + "epoch": 5.55, + "learning_rate": 4.539635311472478e-06, + "loss": 0.154, + "step": 13246 + }, + { + "epoch": 5.55, + "learning_rate": 4.53821504182806e-06, + "loss": 0.0952, + "step": 13247 + }, + { + "epoch": 5.55, + "learning_rate": 4.5367949291778e-06, + "loss": 0.1177, + "step": 13248 + }, + { + "epoch": 5.55, + "learning_rate": 4.535374973562526e-06, + "loss": 0.1216, + "step": 13249 + }, + { + "epoch": 5.55, + "learning_rate": 4.533955175023046e-06, + "loss": 0.1066, + "step": 13250 + }, + { + "epoch": 5.55, + "learning_rate": 4.532535533600173e-06, + "loss": 0.0986, + "step": 13251 + }, + { + "epoch": 5.55, + "learning_rate": 4.531116049334715e-06, + "loss": 0.1444, + "step": 13252 + }, + { + "epoch": 5.55, + "learning_rate": 4.5296967222674724e-06, + "loss": 0.11, + "step": 13253 + }, + { + "epoch": 5.55, + "learning_rate": 4.528277552439244e-06, + "loss": 0.1187, + "step": 13254 + }, + { + "epoch": 5.55, + "learning_rate": 4.526858539890825e-06, + "loss": 0.1006, + "step": 13255 + }, + { + "epoch": 5.55, + "learning_rate": 4.5254396846629975e-06, + "loss": 0.086, + "step": 13256 + }, + { + "epoch": 5.55, + "learning_rate": 4.524020986796548e-06, + "loss": 0.1032, + "step": 13257 + }, + { + "epoch": 5.55, + "learning_rate": 4.522602446332256e-06, + "loss": 0.1266, + "step": 13258 + }, + { + "epoch": 5.55, + "learning_rate": 4.521184063310895e-06, + "loss": 0.1116, + "step": 13259 + }, + { + "epoch": 5.55, + "learning_rate": 4.519765837773238e-06, + "loss": 0.1249, + "step": 13260 + }, + { + "epoch": 5.55, + "learning_rate": 4.518347769760051e-06, + "loss": 0.0959, + "step": 13261 + }, + { + "epoch": 5.55, + "learning_rate": 4.51692985931209e-06, + "loss": 0.1262, + "step": 13262 + }, + { + "epoch": 5.55, + "learning_rate": 4.515512106470115e-06, + "loss": 0.1021, + "step": 13263 + }, + { + "epoch": 5.55, + "learning_rate": 4.514094511274877e-06, + "loss": 0.1451, + "step": 13264 + }, + { + "epoch": 5.55, + "learning_rate": 4.512677073767124e-06, + "loss": 0.1179, + "step": 13265 + }, + { + "epoch": 5.55, + "learning_rate": 4.5112597939876e-06, + "loss": 0.1279, + "step": 13266 + }, + { + "epoch": 5.55, + "learning_rate": 4.509842671977043e-06, + "loss": 0.0967, + "step": 13267 + }, + { + "epoch": 5.55, + "learning_rate": 4.50842570777619e-06, + "loss": 0.0605, + "step": 13268 + }, + { + "epoch": 5.56, + "learning_rate": 4.507008901425762e-06, + "loss": 0.0834, + "step": 13269 + }, + { + "epoch": 5.56, + "learning_rate": 4.50559225296649e-06, + "loss": 0.1333, + "step": 13270 + }, + { + "epoch": 5.56, + "learning_rate": 4.504175762439096e-06, + "loss": 0.1459, + "step": 13271 + }, + { + "epoch": 5.56, + "learning_rate": 4.5027594298842884e-06, + "loss": 0.1377, + "step": 13272 + }, + { + "epoch": 5.56, + "learning_rate": 4.5013432553427835e-06, + "loss": 0.0817, + "step": 13273 + }, + { + "epoch": 5.56, + "learning_rate": 4.499927238855287e-06, + "loss": 0.1335, + "step": 13274 + }, + { + "epoch": 5.56, + "learning_rate": 4.498511380462506e-06, + "loss": 0.1125, + "step": 13275 + }, + { + "epoch": 5.56, + "learning_rate": 4.4970956802051295e-06, + "loss": 0.1297, + "step": 13276 + }, + { + "epoch": 5.56, + "learning_rate": 4.495680138123855e-06, + "loss": 0.1283, + "step": 13277 + }, + { + "epoch": 5.56, + "learning_rate": 4.494264754259368e-06, + "loss": 0.1291, + "step": 13278 + }, + { + "epoch": 5.56, + "learning_rate": 4.492849528652357e-06, + "loss": 0.1326, + "step": 13279 + }, + { + "epoch": 5.56, + "learning_rate": 4.491434461343499e-06, + "loss": 0.109, + "step": 13280 + }, + { + "epoch": 5.56, + "learning_rate": 4.49001955237347e-06, + "loss": 0.1538, + "step": 13281 + }, + { + "epoch": 5.56, + "learning_rate": 4.488604801782942e-06, + "loss": 0.1198, + "step": 13282 + }, + { + "epoch": 5.56, + "learning_rate": 4.4871902096125755e-06, + "loss": 0.1009, + "step": 13283 + }, + { + "epoch": 5.56, + "learning_rate": 4.485775775903033e-06, + "loss": 0.1616, + "step": 13284 + }, + { + "epoch": 5.56, + "learning_rate": 4.484361500694974e-06, + "loss": 0.0977, + "step": 13285 + }, + { + "epoch": 5.56, + "learning_rate": 4.482947384029048e-06, + "loss": 0.1267, + "step": 13286 + }, + { + "epoch": 5.56, + "learning_rate": 4.481533425945904e-06, + "loss": 0.153, + "step": 13287 + }, + { + "epoch": 5.56, + "learning_rate": 4.480119626486189e-06, + "loss": 0.143, + "step": 13288 + }, + { + "epoch": 5.56, + "learning_rate": 4.4787059856905334e-06, + "loss": 0.0814, + "step": 13289 + }, + { + "epoch": 5.56, + "learning_rate": 4.477292503599574e-06, + "loss": 0.1323, + "step": 13290 + }, + { + "epoch": 5.56, + "learning_rate": 4.475879180253941e-06, + "loss": 0.1213, + "step": 13291 + }, + { + "epoch": 5.56, + "learning_rate": 4.47446601569426e-06, + "loss": 0.1197, + "step": 13292 + }, + { + "epoch": 5.57, + "learning_rate": 4.473053009961149e-06, + "loss": 0.1439, + "step": 13293 + }, + { + "epoch": 5.57, + "learning_rate": 4.471640163095229e-06, + "loss": 0.147, + "step": 13294 + }, + { + "epoch": 5.57, + "learning_rate": 4.470227475137102e-06, + "loss": 0.0922, + "step": 13295 + }, + { + "epoch": 5.57, + "learning_rate": 4.4688149461273845e-06, + "loss": 0.1109, + "step": 13296 + }, + { + "epoch": 5.57, + "learning_rate": 4.467402576106668e-06, + "loss": 0.1183, + "step": 13297 + }, + { + "epoch": 5.57, + "learning_rate": 4.465990365115555e-06, + "loss": 0.0781, + "step": 13298 + }, + { + "epoch": 5.57, + "learning_rate": 4.46457831319464e-06, + "loss": 0.1439, + "step": 13299 + }, + { + "epoch": 5.57, + "learning_rate": 4.463166420384507e-06, + "loss": 0.1276, + "step": 13300 + }, + { + "epoch": 5.57, + "learning_rate": 4.461754686725743e-06, + "loss": 0.1152, + "step": 13301 + }, + { + "epoch": 5.57, + "learning_rate": 4.4603431122589295e-06, + "loss": 0.1117, + "step": 13302 + }, + { + "epoch": 5.57, + "learning_rate": 4.458931697024633e-06, + "loss": 0.1098, + "step": 13303 + }, + { + "epoch": 5.57, + "learning_rate": 4.457520441063428e-06, + "loss": 0.1016, + "step": 13304 + }, + { + "epoch": 5.57, + "learning_rate": 4.456109344415881e-06, + "loss": 0.1145, + "step": 13305 + }, + { + "epoch": 5.57, + "learning_rate": 4.454698407122551e-06, + "loss": 0.1367, + "step": 13306 + }, + { + "epoch": 5.57, + "learning_rate": 4.453287629223994e-06, + "loss": 0.1371, + "step": 13307 + }, + { + "epoch": 5.57, + "learning_rate": 4.4518770107607626e-06, + "loss": 0.0942, + "step": 13308 + }, + { + "epoch": 5.57, + "learning_rate": 4.450466551773406e-06, + "loss": 0.114, + "step": 13309 + }, + { + "epoch": 5.57, + "learning_rate": 4.449056252302461e-06, + "loss": 0.1576, + "step": 13310 + }, + { + "epoch": 5.57, + "learning_rate": 4.447646112388467e-06, + "loss": 0.1172, + "step": 13311 + }, + { + "epoch": 5.57, + "learning_rate": 4.44623613207196e-06, + "loss": 0.1272, + "step": 13312 + }, + { + "epoch": 5.57, + "learning_rate": 4.444826311393465e-06, + "loss": 0.1114, + "step": 13313 + }, + { + "epoch": 5.57, + "learning_rate": 4.443416650393509e-06, + "loss": 0.112, + "step": 13314 + }, + { + "epoch": 5.57, + "learning_rate": 4.442007149112615e-06, + "loss": 0.1217, + "step": 13315 + }, + { + "epoch": 5.58, + "learning_rate": 4.44059780759129e-06, + "loss": 0.1172, + "step": 13316 + }, + { + "epoch": 5.58, + "learning_rate": 4.43918862587005e-06, + "loss": 0.1354, + "step": 13317 + }, + { + "epoch": 5.58, + "learning_rate": 4.437779603989395e-06, + "loss": 0.1031, + "step": 13318 + }, + { + "epoch": 5.58, + "learning_rate": 4.436370741989829e-06, + "loss": 0.154, + "step": 13319 + }, + { + "epoch": 5.58, + "learning_rate": 4.4349620399118495e-06, + "loss": 0.1524, + "step": 13320 + }, + { + "epoch": 5.58, + "learning_rate": 4.43355349779595e-06, + "loss": 0.0977, + "step": 13321 + }, + { + "epoch": 5.58, + "learning_rate": 4.432145115682613e-06, + "loss": 0.1257, + "step": 13322 + }, + { + "epoch": 5.58, + "learning_rate": 4.430736893612327e-06, + "loss": 0.1262, + "step": 13323 + }, + { + "epoch": 5.58, + "learning_rate": 4.429328831625565e-06, + "loss": 0.0848, + "step": 13324 + }, + { + "epoch": 5.58, + "learning_rate": 4.427920929762803e-06, + "loss": 0.1323, + "step": 13325 + }, + { + "epoch": 5.58, + "learning_rate": 4.426513188064507e-06, + "loss": 0.1331, + "step": 13326 + }, + { + "epoch": 5.58, + "learning_rate": 4.425105606571145e-06, + "loss": 0.1298, + "step": 13327 + }, + { + "epoch": 5.58, + "learning_rate": 4.423698185323175e-06, + "loss": 0.1379, + "step": 13328 + }, + { + "epoch": 5.58, + "learning_rate": 4.422290924361056e-06, + "loss": 0.1413, + "step": 13329 + }, + { + "epoch": 5.58, + "learning_rate": 4.42088382372523e-06, + "loss": 0.086, + "step": 13330 + }, + { + "epoch": 5.58, + "learning_rate": 4.41947688345615e-06, + "loss": 0.1054, + "step": 13331 + }, + { + "epoch": 5.58, + "learning_rate": 4.418070103594253e-06, + "loss": 0.1087, + "step": 13332 + }, + { + "epoch": 5.58, + "learning_rate": 4.416663484179978e-06, + "loss": 0.138, + "step": 13333 + }, + { + "epoch": 5.58, + "learning_rate": 4.415257025253758e-06, + "loss": 0.1474, + "step": 13334 + }, + { + "epoch": 5.58, + "learning_rate": 4.41385072685602e-06, + "loss": 0.1176, + "step": 13335 + }, + { + "epoch": 5.58, + "learning_rate": 4.412444589027185e-06, + "loss": 0.1425, + "step": 13336 + }, + { + "epoch": 5.58, + "learning_rate": 4.41103861180767e-06, + "loss": 0.1142, + "step": 13337 + }, + { + "epoch": 5.58, + "learning_rate": 4.4096327952378915e-06, + "loss": 0.0912, + "step": 13338 + }, + { + "epoch": 5.58, + "learning_rate": 4.408227139358261e-06, + "loss": 0.1089, + "step": 13339 + }, + { + "epoch": 5.59, + "learning_rate": 4.406821644209175e-06, + "loss": 0.1501, + "step": 13340 + }, + { + "epoch": 5.59, + "learning_rate": 4.4054163098310375e-06, + "loss": 0.1016, + "step": 13341 + }, + { + "epoch": 5.59, + "learning_rate": 4.404011136264245e-06, + "loss": 0.0911, + "step": 13342 + }, + { + "epoch": 5.59, + "learning_rate": 4.402606123549189e-06, + "loss": 0.128, + "step": 13343 + }, + { + "epoch": 5.59, + "learning_rate": 4.40120127172625e-06, + "loss": 0.1497, + "step": 13344 + }, + { + "epoch": 5.59, + "learning_rate": 4.399796580835811e-06, + "loss": 0.1469, + "step": 13345 + }, + { + "epoch": 5.59, + "learning_rate": 4.398392050918251e-06, + "loss": 0.1182, + "step": 13346 + }, + { + "epoch": 5.59, + "learning_rate": 4.39698768201394e-06, + "loss": 0.1152, + "step": 13347 + }, + { + "epoch": 5.59, + "learning_rate": 4.395583474163246e-06, + "loss": 0.1244, + "step": 13348 + }, + { + "epoch": 5.59, + "learning_rate": 4.394179427406533e-06, + "loss": 0.103, + "step": 13349 + }, + { + "epoch": 5.59, + "learning_rate": 4.392775541784161e-06, + "loss": 0.1087, + "step": 13350 + }, + { + "epoch": 5.59, + "learning_rate": 4.391371817336476e-06, + "loss": 0.1513, + "step": 13351 + }, + { + "epoch": 5.59, + "learning_rate": 4.3899682541038316e-06, + "loss": 0.1423, + "step": 13352 + }, + { + "epoch": 5.59, + "learning_rate": 4.388564852126571e-06, + "loss": 0.0978, + "step": 13353 + }, + { + "epoch": 5.59, + "learning_rate": 4.387161611445034e-06, + "loss": 0.1128, + "step": 13354 + }, + { + "epoch": 5.59, + "learning_rate": 4.385758532099556e-06, + "loss": 0.1407, + "step": 13355 + }, + { + "epoch": 5.59, + "learning_rate": 4.384355614130471e-06, + "loss": 0.1338, + "step": 13356 + }, + { + "epoch": 5.59, + "learning_rate": 4.382952857578098e-06, + "loss": 0.1307, + "step": 13357 + }, + { + "epoch": 5.59, + "learning_rate": 4.381550262482759e-06, + "loss": 0.1181, + "step": 13358 + }, + { + "epoch": 5.59, + "learning_rate": 4.380147828884772e-06, + "loss": 0.1042, + "step": 13359 + }, + { + "epoch": 5.59, + "learning_rate": 4.37874555682445e-06, + "loss": 0.1201, + "step": 13360 + }, + { + "epoch": 5.59, + "learning_rate": 4.377343446342097e-06, + "loss": 0.1473, + "step": 13361 + }, + { + "epoch": 5.59, + "learning_rate": 4.375941497478022e-06, + "loss": 0.1154, + "step": 13362 + }, + { + "epoch": 5.59, + "learning_rate": 4.374539710272514e-06, + "loss": 0.1125, + "step": 13363 + }, + { + "epoch": 5.6, + "learning_rate": 4.373138084765875e-06, + "loss": 0.132, + "step": 13364 + }, + { + "epoch": 5.6, + "learning_rate": 4.371736620998383e-06, + "loss": 0.1434, + "step": 13365 + }, + { + "epoch": 5.6, + "learning_rate": 4.370335319010328e-06, + "loss": 0.1287, + "step": 13366 + }, + { + "epoch": 5.6, + "learning_rate": 4.368934178841988e-06, + "loss": 0.1432, + "step": 13367 + }, + { + "epoch": 5.6, + "learning_rate": 4.367533200533639e-06, + "loss": 0.1219, + "step": 13368 + }, + { + "epoch": 5.6, + "learning_rate": 4.36613238412555e-06, + "loss": 0.1447, + "step": 13369 + }, + { + "epoch": 5.6, + "learning_rate": 4.36473172965799e-06, + "loss": 0.1143, + "step": 13370 + }, + { + "epoch": 5.6, + "learning_rate": 4.363331237171211e-06, + "loss": 0.111, + "step": 13371 + }, + { + "epoch": 5.6, + "learning_rate": 4.361930906705475e-06, + "loss": 0.115, + "step": 13372 + }, + { + "epoch": 5.6, + "learning_rate": 4.360530738301031e-06, + "loss": 0.1121, + "step": 13373 + }, + { + "epoch": 5.6, + "learning_rate": 4.359130731998127e-06, + "loss": 0.1524, + "step": 13374 + }, + { + "epoch": 5.6, + "learning_rate": 4.357730887837005e-06, + "loss": 0.1011, + "step": 13375 + }, + { + "epoch": 5.6, + "learning_rate": 4.356331205857904e-06, + "loss": 0.1064, + "step": 13376 + }, + { + "epoch": 5.6, + "learning_rate": 4.354931686101051e-06, + "loss": 0.1177, + "step": 13377 + }, + { + "epoch": 5.6, + "learning_rate": 4.353532328606679e-06, + "loss": 0.0824, + "step": 13378 + }, + { + "epoch": 5.6, + "learning_rate": 4.352133133415009e-06, + "loss": 0.12, + "step": 13379 + }, + { + "epoch": 5.6, + "learning_rate": 4.3507341005662595e-06, + "loss": 0.1201, + "step": 13380 + }, + { + "epoch": 5.6, + "learning_rate": 4.349335230100645e-06, + "loss": 0.0962, + "step": 13381 + }, + { + "epoch": 5.6, + "learning_rate": 4.3479365220583754e-06, + "loss": 0.1817, + "step": 13382 + }, + { + "epoch": 5.6, + "learning_rate": 4.346537976479659e-06, + "loss": 0.143, + "step": 13383 + }, + { + "epoch": 5.6, + "learning_rate": 4.345139593404688e-06, + "loss": 0.1115, + "step": 13384 + }, + { + "epoch": 5.6, + "learning_rate": 4.343741372873665e-06, + "loss": 0.104, + "step": 13385 + }, + { + "epoch": 5.6, + "learning_rate": 4.342343314926773e-06, + "loss": 0.1272, + "step": 13386 + }, + { + "epoch": 5.6, + "learning_rate": 4.340945419604201e-06, + "loss": 0.1128, + "step": 13387 + }, + { + "epoch": 5.61, + "learning_rate": 4.3395476869461325e-06, + "loss": 0.115, + "step": 13388 + }, + { + "epoch": 5.61, + "learning_rate": 4.338150116992743e-06, + "loss": 0.1597, + "step": 13389 + }, + { + "epoch": 5.61, + "learning_rate": 4.3367527097842034e-06, + "loss": 0.1409, + "step": 13390 + }, + { + "epoch": 5.61, + "learning_rate": 4.335355465360686e-06, + "loss": 0.1341, + "step": 13391 + }, + { + "epoch": 5.61, + "learning_rate": 4.333958383762345e-06, + "loss": 0.1257, + "step": 13392 + }, + { + "epoch": 5.61, + "learning_rate": 4.332561465029341e-06, + "loss": 0.1009, + "step": 13393 + }, + { + "epoch": 5.61, + "learning_rate": 4.331164709201829e-06, + "loss": 0.099, + "step": 13394 + }, + { + "epoch": 5.61, + "learning_rate": 4.329768116319958e-06, + "loss": 0.0919, + "step": 13395 + }, + { + "epoch": 5.61, + "learning_rate": 4.328371686423871e-06, + "loss": 0.085, + "step": 13396 + }, + { + "epoch": 5.61, + "learning_rate": 4.32697541955371e-06, + "loss": 0.1586, + "step": 13397 + }, + { + "epoch": 5.61, + "learning_rate": 4.325579315749603e-06, + "loss": 0.1115, + "step": 13398 + }, + { + "epoch": 5.61, + "learning_rate": 4.324183375051683e-06, + "loss": 0.1282, + "step": 13399 + }, + { + "epoch": 5.61, + "learning_rate": 4.322787597500074e-06, + "loss": 0.0924, + "step": 13400 + }, + { + "epoch": 5.61, + "learning_rate": 4.321391983134899e-06, + "loss": 0.1703, + "step": 13401 + }, + { + "epoch": 5.61, + "learning_rate": 4.319996531996272e-06, + "loss": 0.1231, + "step": 13402 + }, + { + "epoch": 5.61, + "learning_rate": 4.318601244124308e-06, + "loss": 0.1272, + "step": 13403 + }, + { + "epoch": 5.61, + "learning_rate": 4.317206119559108e-06, + "loss": 0.114, + "step": 13404 + }, + { + "epoch": 5.61, + "learning_rate": 4.315811158340773e-06, + "loss": 0.11, + "step": 13405 + }, + { + "epoch": 5.61, + "learning_rate": 4.314416360509403e-06, + "loss": 0.1258, + "step": 13406 + }, + { + "epoch": 5.61, + "learning_rate": 4.3130217261050925e-06, + "loss": 0.1358, + "step": 13407 + }, + { + "epoch": 5.61, + "learning_rate": 4.3116272551679215e-06, + "loss": 0.1249, + "step": 13408 + }, + { + "epoch": 5.61, + "learning_rate": 4.310232947737979e-06, + "loss": 0.1491, + "step": 13409 + }, + { + "epoch": 5.61, + "learning_rate": 4.308838803855341e-06, + "loss": 0.1461, + "step": 13410 + }, + { + "epoch": 5.61, + "learning_rate": 4.307444823560084e-06, + "loss": 0.1283, + "step": 13411 + }, + { + "epoch": 5.62, + "learning_rate": 4.306051006892271e-06, + "loss": 0.1214, + "step": 13412 + }, + { + "epoch": 5.62, + "learning_rate": 4.30465735389197e-06, + "loss": 0.1222, + "step": 13413 + }, + { + "epoch": 5.62, + "learning_rate": 4.303263864599239e-06, + "loss": 0.0874, + "step": 13414 + }, + { + "epoch": 5.62, + "learning_rate": 4.3018705390541316e-06, + "loss": 0.1077, + "step": 13415 + }, + { + "epoch": 5.62, + "learning_rate": 4.3004773772967e-06, + "loss": 0.1062, + "step": 13416 + }, + { + "epoch": 5.62, + "learning_rate": 4.2990843793669925e-06, + "loss": 0.0945, + "step": 13417 + }, + { + "epoch": 5.62, + "learning_rate": 4.297691545305042e-06, + "loss": 0.1373, + "step": 13418 + }, + { + "epoch": 5.62, + "learning_rate": 4.296298875150887e-06, + "loss": 0.1218, + "step": 13419 + }, + { + "epoch": 5.62, + "learning_rate": 4.29490636894456e-06, + "loss": 0.1227, + "step": 13420 + }, + { + "epoch": 5.62, + "learning_rate": 4.293514026726088e-06, + "loss": 0.1127, + "step": 13421 + }, + { + "epoch": 5.62, + "learning_rate": 4.29212184853549e-06, + "loss": 0.1294, + "step": 13422 + }, + { + "epoch": 5.62, + "learning_rate": 4.2907298344127844e-06, + "loss": 0.1136, + "step": 13423 + }, + { + "epoch": 5.62, + "learning_rate": 4.289337984397988e-06, + "loss": 0.0996, + "step": 13424 + }, + { + "epoch": 5.62, + "learning_rate": 4.287946298531099e-06, + "loss": 0.107, + "step": 13425 + }, + { + "epoch": 5.62, + "learning_rate": 4.286554776852125e-06, + "loss": 0.1048, + "step": 13426 + }, + { + "epoch": 5.62, + "learning_rate": 4.285163419401065e-06, + "loss": 0.1187, + "step": 13427 + }, + { + "epoch": 5.62, + "learning_rate": 4.2837722262179105e-06, + "loss": 0.1657, + "step": 13428 + }, + { + "epoch": 5.62, + "learning_rate": 4.282381197342651e-06, + "loss": 0.118, + "step": 13429 + }, + { + "epoch": 5.62, + "learning_rate": 4.280990332815274e-06, + "loss": 0.1607, + "step": 13430 + }, + { + "epoch": 5.62, + "learning_rate": 4.27959963267575e-06, + "loss": 0.151, + "step": 13431 + }, + { + "epoch": 5.62, + "learning_rate": 4.278209096964063e-06, + "loss": 0.1083, + "step": 13432 + }, + { + "epoch": 5.62, + "learning_rate": 4.276818725720176e-06, + "loss": 0.0842, + "step": 13433 + }, + { + "epoch": 5.62, + "learning_rate": 4.275428518984054e-06, + "loss": 0.1159, + "step": 13434 + }, + { + "epoch": 5.62, + "learning_rate": 4.27403847679566e-06, + "loss": 0.1504, + "step": 13435 + }, + { + "epoch": 5.63, + "learning_rate": 4.272648599194948e-06, + "loss": 0.1206, + "step": 13436 + }, + { + "epoch": 5.63, + "learning_rate": 4.27125888622187e-06, + "loss": 0.1466, + "step": 13437 + }, + { + "epoch": 5.63, + "learning_rate": 4.269869337916377e-06, + "loss": 0.1112, + "step": 13438 + }, + { + "epoch": 5.63, + "learning_rate": 4.2684799543183995e-06, + "loss": 0.0942, + "step": 13439 + }, + { + "epoch": 5.63, + "learning_rate": 4.26709073546788e-06, + "loss": 0.1187, + "step": 13440 + }, + { + "epoch": 5.63, + "learning_rate": 4.2657016814047515e-06, + "loss": 0.118, + "step": 13441 + }, + { + "epoch": 5.63, + "learning_rate": 4.264312792168938e-06, + "loss": 0.0935, + "step": 13442 + }, + { + "epoch": 5.63, + "learning_rate": 4.262924067800364e-06, + "loss": 0.1004, + "step": 13443 + }, + { + "epoch": 5.63, + "learning_rate": 4.261535508338951e-06, + "loss": 0.113, + "step": 13444 + }, + { + "epoch": 5.63, + "learning_rate": 4.260147113824603e-06, + "loss": 0.1241, + "step": 13445 + }, + { + "epoch": 5.63, + "learning_rate": 4.2587588842972325e-06, + "loss": 0.131, + "step": 13446 + }, + { + "epoch": 5.63, + "learning_rate": 4.257370819796745e-06, + "loss": 0.1177, + "step": 13447 + }, + { + "epoch": 5.63, + "learning_rate": 4.255982920363035e-06, + "loss": 0.1428, + "step": 13448 + }, + { + "epoch": 5.63, + "learning_rate": 4.254595186036001e-06, + "loss": 0.1097, + "step": 13449 + }, + { + "epoch": 5.63, + "learning_rate": 4.253207616855528e-06, + "loss": 0.1172, + "step": 13450 + }, + { + "epoch": 5.63, + "learning_rate": 4.251820212861507e-06, + "loss": 0.1338, + "step": 13451 + }, + { + "epoch": 5.63, + "learning_rate": 4.2504329740938084e-06, + "loss": 0.1281, + "step": 13452 + }, + { + "epoch": 5.63, + "learning_rate": 4.249045900592318e-06, + "loss": 0.1366, + "step": 13453 + }, + { + "epoch": 5.63, + "learning_rate": 4.2476589923968946e-06, + "loss": 0.0745, + "step": 13454 + }, + { + "epoch": 5.63, + "learning_rate": 4.2462722495474085e-06, + "loss": 0.1222, + "step": 13455 + }, + { + "epoch": 5.63, + "learning_rate": 4.244885672083722e-06, + "loss": 0.1208, + "step": 13456 + }, + { + "epoch": 5.63, + "learning_rate": 4.243499260045689e-06, + "loss": 0.1247, + "step": 13457 + }, + { + "epoch": 5.63, + "learning_rate": 4.242113013473166e-06, + "loss": 0.1226, + "step": 13458 + }, + { + "epoch": 5.63, + "learning_rate": 4.240726932405991e-06, + "loss": 0.1737, + "step": 13459 + }, + { + "epoch": 5.64, + "learning_rate": 4.239341016884008e-06, + "loss": 0.1072, + "step": 13460 + }, + { + "epoch": 5.64, + "learning_rate": 4.237955266947058e-06, + "loss": 0.1087, + "step": 13461 + }, + { + "epoch": 5.64, + "learning_rate": 4.23656968263497e-06, + "loss": 0.137, + "step": 13462 + }, + { + "epoch": 5.64, + "learning_rate": 4.235184263987572e-06, + "loss": 0.1212, + "step": 13463 + }, + { + "epoch": 5.64, + "learning_rate": 4.233799011044687e-06, + "loss": 0.1699, + "step": 13464 + }, + { + "epoch": 5.64, + "learning_rate": 4.232413923846135e-06, + "loss": 0.1369, + "step": 13465 + }, + { + "epoch": 5.64, + "learning_rate": 4.231029002431723e-06, + "loss": 0.0873, + "step": 13466 + }, + { + "epoch": 5.64, + "learning_rate": 4.2296442468412646e-06, + "loss": 0.1319, + "step": 13467 + }, + { + "epoch": 5.64, + "learning_rate": 4.2282596571145615e-06, + "loss": 0.1551, + "step": 13468 + }, + { + "epoch": 5.64, + "learning_rate": 4.226875233291412e-06, + "loss": 0.1186, + "step": 13469 + }, + { + "epoch": 5.64, + "learning_rate": 4.225490975411611e-06, + "loss": 0.1651, + "step": 13470 + }, + { + "epoch": 5.64, + "learning_rate": 4.224106883514952e-06, + "loss": 0.1228, + "step": 13471 + }, + { + "epoch": 5.64, + "learning_rate": 4.222722957641212e-06, + "loss": 0.1283, + "step": 13472 + }, + { + "epoch": 5.64, + "learning_rate": 4.221339197830173e-06, + "loss": 0.1121, + "step": 13473 + }, + { + "epoch": 5.64, + "learning_rate": 4.219955604121611e-06, + "loss": 0.1387, + "step": 13474 + }, + { + "epoch": 5.64, + "learning_rate": 4.2185721765553e-06, + "loss": 0.1105, + "step": 13475 + }, + { + "epoch": 5.64, + "learning_rate": 4.217188915170996e-06, + "loss": 0.12, + "step": 13476 + }, + { + "epoch": 5.64, + "learning_rate": 4.215805820008468e-06, + "loss": 0.1138, + "step": 13477 + }, + { + "epoch": 5.64, + "learning_rate": 4.214422891107467e-06, + "loss": 0.1054, + "step": 13478 + }, + { + "epoch": 5.64, + "learning_rate": 4.2130401285077505e-06, + "loss": 0.1142, + "step": 13479 + }, + { + "epoch": 5.64, + "learning_rate": 4.211657532249056e-06, + "loss": 0.1037, + "step": 13480 + }, + { + "epoch": 5.64, + "learning_rate": 4.21027510237113e-06, + "loss": 0.1261, + "step": 13481 + }, + { + "epoch": 5.64, + "learning_rate": 4.208892838913707e-06, + "loss": 0.1181, + "step": 13482 + }, + { + "epoch": 5.64, + "learning_rate": 4.207510741916522e-06, + "loss": 0.1007, + "step": 13483 + }, + { + "epoch": 5.65, + "learning_rate": 4.206128811419299e-06, + "loss": 0.1343, + "step": 13484 + }, + { + "epoch": 5.65, + "learning_rate": 4.204747047461767e-06, + "loss": 0.1377, + "step": 13485 + }, + { + "epoch": 5.65, + "learning_rate": 4.2033654500836344e-06, + "loss": 0.1334, + "step": 13486 + }, + { + "epoch": 5.65, + "learning_rate": 4.201984019324617e-06, + "loss": 0.1247, + "step": 13487 + }, + { + "epoch": 5.65, + "learning_rate": 4.200602755224426e-06, + "loss": 0.1618, + "step": 13488 + }, + { + "epoch": 5.65, + "learning_rate": 4.199221657822761e-06, + "loss": 0.1531, + "step": 13489 + }, + { + "epoch": 5.65, + "learning_rate": 4.1978407271593226e-06, + "loss": 0.1097, + "step": 13490 + }, + { + "epoch": 5.65, + "learning_rate": 4.1964599632738035e-06, + "loss": 0.1293, + "step": 13491 + }, + { + "epoch": 5.65, + "learning_rate": 4.1950793662058965e-06, + "loss": 0.1205, + "step": 13492 + }, + { + "epoch": 5.65, + "learning_rate": 4.1936989359952775e-06, + "loss": 0.1221, + "step": 13493 + }, + { + "epoch": 5.65, + "learning_rate": 4.192318672681631e-06, + "loss": 0.1379, + "step": 13494 + }, + { + "epoch": 5.65, + "learning_rate": 4.1909385763046305e-06, + "loss": 0.1072, + "step": 13495 + }, + { + "epoch": 5.65, + "learning_rate": 4.1895586469039466e-06, + "loss": 0.0995, + "step": 13496 + }, + { + "epoch": 5.65, + "learning_rate": 4.188178884519241e-06, + "loss": 0.1429, + "step": 13497 + }, + { + "epoch": 5.65, + "learning_rate": 4.186799289190182e-06, + "loss": 0.1256, + "step": 13498 + }, + { + "epoch": 5.65, + "learning_rate": 4.185419860956418e-06, + "loss": 0.1429, + "step": 13499 + }, + { + "epoch": 5.65, + "learning_rate": 4.184040599857596e-06, + "loss": 0.1564, + "step": 13500 + }, + { + "epoch": 5.65, + "learning_rate": 4.182661505933366e-06, + "loss": 0.1095, + "step": 13501 + }, + { + "epoch": 5.65, + "learning_rate": 4.18128257922337e-06, + "loss": 0.1098, + "step": 13502 + }, + { + "epoch": 5.65, + "learning_rate": 4.179903819767242e-06, + "loss": 0.1369, + "step": 13503 + }, + { + "epoch": 5.65, + "learning_rate": 4.178525227604615e-06, + "loss": 0.1407, + "step": 13504 + }, + { + "epoch": 5.65, + "learning_rate": 4.177146802775114e-06, + "loss": 0.1111, + "step": 13505 + }, + { + "epoch": 5.65, + "learning_rate": 4.175768545318364e-06, + "loss": 0.1186, + "step": 13506 + }, + { + "epoch": 5.66, + "learning_rate": 4.174390455273975e-06, + "loss": 0.1016, + "step": 13507 + }, + { + "epoch": 5.66, + "learning_rate": 4.173012532681564e-06, + "loss": 0.1468, + "step": 13508 + }, + { + "epoch": 5.66, + "learning_rate": 4.171634777580736e-06, + "loss": 0.1381, + "step": 13509 + }, + { + "epoch": 5.66, + "learning_rate": 4.170257190011096e-06, + "loss": 0.1473, + "step": 13510 + }, + { + "epoch": 5.66, + "learning_rate": 4.168879770012237e-06, + "loss": 0.0879, + "step": 13511 + }, + { + "epoch": 5.66, + "learning_rate": 4.16750251762376e-06, + "loss": 0.0742, + "step": 13512 + }, + { + "epoch": 5.66, + "learning_rate": 4.166125432885244e-06, + "loss": 0.1145, + "step": 13513 + }, + { + "epoch": 5.66, + "learning_rate": 4.164748515836276e-06, + "loss": 0.0923, + "step": 13514 + }, + { + "epoch": 5.66, + "learning_rate": 4.163371766516432e-06, + "loss": 0.1548, + "step": 13515 + }, + { + "epoch": 5.66, + "learning_rate": 4.1619951849652885e-06, + "loss": 0.1205, + "step": 13516 + }, + { + "epoch": 5.66, + "learning_rate": 4.160618771222412e-06, + "loss": 0.1102, + "step": 13517 + }, + { + "epoch": 5.66, + "learning_rate": 4.159242525327371e-06, + "loss": 0.112, + "step": 13518 + }, + { + "epoch": 5.66, + "learning_rate": 4.157866447319718e-06, + "loss": 0.1344, + "step": 13519 + }, + { + "epoch": 5.66, + "learning_rate": 4.15649053723901e-06, + "loss": 0.1181, + "step": 13520 + }, + { + "epoch": 5.66, + "learning_rate": 4.155114795124798e-06, + "loss": 0.1008, + "step": 13521 + }, + { + "epoch": 5.66, + "learning_rate": 4.153739221016622e-06, + "loss": 0.0864, + "step": 13522 + }, + { + "epoch": 5.66, + "learning_rate": 4.152363814954024e-06, + "loss": 0.1235, + "step": 13523 + }, + { + "epoch": 5.66, + "learning_rate": 4.150988576976539e-06, + "loss": 0.1005, + "step": 13524 + }, + { + "epoch": 5.66, + "learning_rate": 4.1496135071236965e-06, + "loss": 0.075, + "step": 13525 + }, + { + "epoch": 5.66, + "learning_rate": 4.1482386054350265e-06, + "loss": 0.1, + "step": 13526 + }, + { + "epoch": 5.66, + "learning_rate": 4.146863871950041e-06, + "loss": 0.0988, + "step": 13527 + }, + { + "epoch": 5.66, + "learning_rate": 4.1454893067082605e-06, + "loss": 0.1228, + "step": 13528 + }, + { + "epoch": 5.66, + "learning_rate": 4.144114909749193e-06, + "loss": 0.127, + "step": 13529 + }, + { + "epoch": 5.66, + "learning_rate": 4.142740681112347e-06, + "loss": 0.1128, + "step": 13530 + }, + { + "epoch": 5.67, + "learning_rate": 4.141366620837223e-06, + "loss": 0.1203, + "step": 13531 + }, + { + "epoch": 5.67, + "learning_rate": 4.139992728963315e-06, + "loss": 0.1196, + "step": 13532 + }, + { + "epoch": 5.67, + "learning_rate": 4.138619005530121e-06, + "loss": 0.1411, + "step": 13533 + }, + { + "epoch": 5.67, + "learning_rate": 4.1372454505771186e-06, + "loss": 0.1377, + "step": 13534 + }, + { + "epoch": 5.67, + "learning_rate": 4.135872064143792e-06, + "loss": 0.1149, + "step": 13535 + }, + { + "epoch": 5.67, + "learning_rate": 4.134498846269621e-06, + "loss": 0.1183, + "step": 13536 + }, + { + "epoch": 5.67, + "learning_rate": 4.133125796994074e-06, + "loss": 0.134, + "step": 13537 + }, + { + "epoch": 5.67, + "learning_rate": 4.131752916356622e-06, + "loss": 0.1001, + "step": 13538 + }, + { + "epoch": 5.67, + "learning_rate": 4.130380204396727e-06, + "loss": 0.1182, + "step": 13539 + }, + { + "epoch": 5.67, + "learning_rate": 4.129007661153841e-06, + "loss": 0.1349, + "step": 13540 + }, + { + "epoch": 5.67, + "learning_rate": 4.12763528666742e-06, + "loss": 0.1489, + "step": 13541 + }, + { + "epoch": 5.67, + "learning_rate": 4.126263080976911e-06, + "loss": 0.1143, + "step": 13542 + }, + { + "epoch": 5.67, + "learning_rate": 4.124891044121761e-06, + "loss": 0.1215, + "step": 13543 + }, + { + "epoch": 5.67, + "learning_rate": 4.1235191761414025e-06, + "loss": 0.0732, + "step": 13544 + }, + { + "epoch": 5.67, + "learning_rate": 4.12214747707527e-06, + "loss": 0.1264, + "step": 13545 + }, + { + "epoch": 5.67, + "learning_rate": 4.120775946962792e-06, + "loss": 0.1171, + "step": 13546 + }, + { + "epoch": 5.67, + "learning_rate": 4.1194045858433965e-06, + "loss": 0.1367, + "step": 13547 + }, + { + "epoch": 5.67, + "learning_rate": 4.118033393756493e-06, + "loss": 0.1333, + "step": 13548 + }, + { + "epoch": 5.67, + "learning_rate": 4.116662370741501e-06, + "loss": 0.0928, + "step": 13549 + }, + { + "epoch": 5.67, + "learning_rate": 4.115291516837827e-06, + "loss": 0.1202, + "step": 13550 + }, + { + "epoch": 5.67, + "learning_rate": 4.113920832084878e-06, + "loss": 0.1247, + "step": 13551 + }, + { + "epoch": 5.67, + "learning_rate": 4.11255031652205e-06, + "loss": 0.1122, + "step": 13552 + }, + { + "epoch": 5.67, + "learning_rate": 4.111179970188745e-06, + "loss": 0.1025, + "step": 13553 + }, + { + "epoch": 5.67, + "learning_rate": 4.10980979312434e-06, + "loss": 0.1193, + "step": 13554 + }, + { + "epoch": 5.68, + "learning_rate": 4.108439785368226e-06, + "loss": 0.1566, + "step": 13555 + }, + { + "epoch": 5.68, + "learning_rate": 4.107069946959783e-06, + "loss": 0.1232, + "step": 13556 + }, + { + "epoch": 5.68, + "learning_rate": 4.105700277938385e-06, + "loss": 0.1144, + "step": 13557 + }, + { + "epoch": 5.68, + "learning_rate": 4.1043307783434e-06, + "loss": 0.1173, + "step": 13558 + }, + { + "epoch": 5.68, + "learning_rate": 4.102961448214201e-06, + "loss": 0.1286, + "step": 13559 + }, + { + "epoch": 5.68, + "learning_rate": 4.101592287590138e-06, + "loss": 0.1327, + "step": 13560 + }, + { + "epoch": 5.68, + "learning_rate": 4.100223296510571e-06, + "loss": 0.1254, + "step": 13561 + }, + { + "epoch": 5.68, + "learning_rate": 4.098854475014849e-06, + "loss": 0.1013, + "step": 13562 + }, + { + "epoch": 5.68, + "learning_rate": 4.097485823142319e-06, + "loss": 0.1239, + "step": 13563 + }, + { + "epoch": 5.68, + "learning_rate": 4.096117340932322e-06, + "loss": 0.1021, + "step": 13564 + }, + { + "epoch": 5.68, + "learning_rate": 4.094749028424192e-06, + "loss": 0.1207, + "step": 13565 + }, + { + "epoch": 5.68, + "learning_rate": 4.093380885657264e-06, + "loss": 0.1143, + "step": 13566 + }, + { + "epoch": 5.68, + "learning_rate": 4.092012912670863e-06, + "loss": 0.1372, + "step": 13567 + }, + { + "epoch": 5.68, + "learning_rate": 4.090645109504303e-06, + "loss": 0.0726, + "step": 13568 + }, + { + "epoch": 5.68, + "learning_rate": 4.0892774761969055e-06, + "loss": 0.1092, + "step": 13569 + }, + { + "epoch": 5.68, + "learning_rate": 4.087910012787983e-06, + "loss": 0.1253, + "step": 13570 + }, + { + "epoch": 5.68, + "learning_rate": 4.08654271931684e-06, + "loss": 0.1181, + "step": 13571 + }, + { + "epoch": 5.68, + "learning_rate": 4.085175595822779e-06, + "loss": 0.1061, + "step": 13572 + }, + { + "epoch": 5.68, + "learning_rate": 4.083808642345098e-06, + "loss": 0.1261, + "step": 13573 + }, + { + "epoch": 5.68, + "learning_rate": 4.082441858923088e-06, + "loss": 0.0955, + "step": 13574 + }, + { + "epoch": 5.68, + "learning_rate": 4.0810752455960356e-06, + "loss": 0.1048, + "step": 13575 + }, + { + "epoch": 5.68, + "learning_rate": 4.07970880240322e-06, + "loss": 0.069, + "step": 13576 + }, + { + "epoch": 5.68, + "learning_rate": 4.0783425293839245e-06, + "loss": 0.1157, + "step": 13577 + }, + { + "epoch": 5.68, + "learning_rate": 4.0769764265774155e-06, + "loss": 0.111, + "step": 13578 + }, + { + "epoch": 5.69, + "learning_rate": 4.0756104940229645e-06, + "loss": 0.1169, + "step": 13579 + }, + { + "epoch": 5.69, + "learning_rate": 4.074244731759836e-06, + "loss": 0.1089, + "step": 13580 + }, + { + "epoch": 5.69, + "learning_rate": 4.07287913982728e-06, + "loss": 0.1629, + "step": 13581 + }, + { + "epoch": 5.69, + "learning_rate": 4.071513718264554e-06, + "loss": 0.1112, + "step": 13582 + }, + { + "epoch": 5.69, + "learning_rate": 4.070148467110906e-06, + "loss": 0.113, + "step": 13583 + }, + { + "epoch": 5.69, + "learning_rate": 4.068783386405577e-06, + "loss": 0.145, + "step": 13584 + }, + { + "epoch": 5.69, + "learning_rate": 4.067418476187807e-06, + "loss": 0.1243, + "step": 13585 + }, + { + "epoch": 5.69, + "learning_rate": 4.066053736496832e-06, + "loss": 0.0941, + "step": 13586 + }, + { + "epoch": 5.69, + "learning_rate": 4.0646891673718725e-06, + "loss": 0.1226, + "step": 13587 + }, + { + "epoch": 5.69, + "learning_rate": 4.063324768852157e-06, + "loss": 0.1108, + "step": 13588 + }, + { + "epoch": 5.69, + "learning_rate": 4.061960540976906e-06, + "loss": 0.1098, + "step": 13589 + }, + { + "epoch": 5.69, + "learning_rate": 4.060596483785326e-06, + "loss": 0.152, + "step": 13590 + }, + { + "epoch": 5.69, + "learning_rate": 4.059232597316629e-06, + "loss": 0.1005, + "step": 13591 + }, + { + "epoch": 5.69, + "learning_rate": 4.057868881610021e-06, + "loss": 0.1304, + "step": 13592 + }, + { + "epoch": 5.69, + "learning_rate": 4.056505336704698e-06, + "loss": 0.0902, + "step": 13593 + }, + { + "epoch": 5.69, + "learning_rate": 4.055141962639858e-06, + "loss": 0.1515, + "step": 13594 + }, + { + "epoch": 5.69, + "learning_rate": 4.053778759454684e-06, + "loss": 0.0776, + "step": 13595 + }, + { + "epoch": 5.69, + "learning_rate": 4.0524157271883635e-06, + "loss": 0.1422, + "step": 13596 + }, + { + "epoch": 5.69, + "learning_rate": 4.051052865880074e-06, + "loss": 0.124, + "step": 13597 + }, + { + "epoch": 5.69, + "learning_rate": 4.049690175568992e-06, + "loss": 0.093, + "step": 13598 + }, + { + "epoch": 5.69, + "learning_rate": 4.048327656294285e-06, + "loss": 0.09, + "step": 13599 + }, + { + "epoch": 5.69, + "learning_rate": 4.046965308095122e-06, + "loss": 0.1262, + "step": 13600 + }, + { + "epoch": 5.69, + "learning_rate": 4.045603131010655e-06, + "loss": 0.1213, + "step": 13601 + }, + { + "epoch": 5.69, + "learning_rate": 4.044241125080041e-06, + "loss": 0.1101, + "step": 13602 + }, + { + "epoch": 5.7, + "learning_rate": 4.042879290342432e-06, + "loss": 0.1136, + "step": 13603 + }, + { + "epoch": 5.7, + "learning_rate": 4.0415176268369714e-06, + "loss": 0.1203, + "step": 13604 + }, + { + "epoch": 5.7, + "learning_rate": 4.0401561346028e-06, + "loss": 0.084, + "step": 13605 + }, + { + "epoch": 5.7, + "learning_rate": 4.03879481367905e-06, + "loss": 0.1028, + "step": 13606 + }, + { + "epoch": 5.7, + "learning_rate": 4.037433664104857e-06, + "loss": 0.1255, + "step": 13607 + }, + { + "epoch": 5.7, + "learning_rate": 4.03607268591934e-06, + "loss": 0.1055, + "step": 13608 + }, + { + "epoch": 5.7, + "learning_rate": 4.03471187916162e-06, + "loss": 0.1279, + "step": 13609 + }, + { + "epoch": 5.7, + "learning_rate": 4.033351243870815e-06, + "loss": 0.1281, + "step": 13610 + }, + { + "epoch": 5.7, + "learning_rate": 4.031990780086037e-06, + "loss": 0.1073, + "step": 13611 + }, + { + "epoch": 5.7, + "learning_rate": 4.0306304878463834e-06, + "loss": 0.1445, + "step": 13612 + }, + { + "epoch": 5.7, + "learning_rate": 4.0292703671909605e-06, + "loss": 0.1192, + "step": 13613 + }, + { + "epoch": 5.7, + "learning_rate": 4.027910418158861e-06, + "loss": 0.1246, + "step": 13614 + }, + { + "epoch": 5.7, + "learning_rate": 4.026550640789183e-06, + "loss": 0.1184, + "step": 13615 + }, + { + "epoch": 5.7, + "learning_rate": 4.025191035121001e-06, + "loss": 0.1196, + "step": 13616 + }, + { + "epoch": 5.7, + "learning_rate": 4.023831601193401e-06, + "loss": 0.1371, + "step": 13617 + }, + { + "epoch": 5.7, + "learning_rate": 4.0224723390454575e-06, + "loss": 0.0976, + "step": 13618 + }, + { + "epoch": 5.7, + "learning_rate": 4.021113248716243e-06, + "loss": 0.0785, + "step": 13619 + }, + { + "epoch": 5.7, + "learning_rate": 4.019754330244822e-06, + "loss": 0.1165, + "step": 13620 + }, + { + "epoch": 5.7, + "learning_rate": 4.0183955836702615e-06, + "loss": 0.1068, + "step": 13621 + }, + { + "epoch": 5.7, + "learning_rate": 4.017037009031607e-06, + "loss": 0.125, + "step": 13622 + }, + { + "epoch": 5.7, + "learning_rate": 4.015678606367916e-06, + "loss": 0.1584, + "step": 13623 + }, + { + "epoch": 5.7, + "learning_rate": 4.01432037571823e-06, + "loss": 0.1086, + "step": 13624 + }, + { + "epoch": 5.7, + "learning_rate": 4.012962317121596e-06, + "loss": 0.1481, + "step": 13625 + }, + { + "epoch": 5.7, + "learning_rate": 4.011604430617046e-06, + "loss": 0.1074, + "step": 13626 + }, + { + "epoch": 5.71, + "learning_rate": 4.010246716243617e-06, + "loss": 0.1154, + "step": 13627 + }, + { + "epoch": 5.71, + "learning_rate": 4.008889174040327e-06, + "loss": 0.1414, + "step": 13628 + }, + { + "epoch": 5.71, + "learning_rate": 4.0075318040462005e-06, + "loss": 0.1165, + "step": 13629 + }, + { + "epoch": 5.71, + "learning_rate": 4.006174606300255e-06, + "loss": 0.0938, + "step": 13630 + }, + { + "epoch": 5.71, + "learning_rate": 4.004817580841503e-06, + "loss": 0.1314, + "step": 13631 + }, + { + "epoch": 5.71, + "learning_rate": 4.003460727708947e-06, + "loss": 0.1182, + "step": 13632 + }, + { + "epoch": 5.71, + "learning_rate": 4.002104046941594e-06, + "loss": 0.1373, + "step": 13633 + }, + { + "epoch": 5.71, + "learning_rate": 4.00074753857844e-06, + "loss": 0.0948, + "step": 13634 + }, + { + "epoch": 5.71, + "learning_rate": 3.9993912026584735e-06, + "loss": 0.1131, + "step": 13635 + }, + { + "epoch": 5.71, + "learning_rate": 3.99803503922068e-06, + "loss": 0.1216, + "step": 13636 + }, + { + "epoch": 5.71, + "learning_rate": 3.996679048304042e-06, + "loss": 0.0913, + "step": 13637 + }, + { + "epoch": 5.71, + "learning_rate": 3.9953232299475385e-06, + "loss": 0.1136, + "step": 13638 + }, + { + "epoch": 5.71, + "learning_rate": 3.993967584190141e-06, + "loss": 0.1164, + "step": 13639 + }, + { + "epoch": 5.71, + "learning_rate": 3.992612111070814e-06, + "loss": 0.0847, + "step": 13640 + }, + { + "epoch": 5.71, + "learning_rate": 3.991256810628525e-06, + "loss": 0.1079, + "step": 13641 + }, + { + "epoch": 5.71, + "learning_rate": 3.989901682902223e-06, + "loss": 0.1372, + "step": 13642 + }, + { + "epoch": 5.71, + "learning_rate": 3.9885467279308644e-06, + "loss": 0.1175, + "step": 13643 + }, + { + "epoch": 5.71, + "learning_rate": 3.987191945753396e-06, + "loss": 0.1287, + "step": 13644 + }, + { + "epoch": 5.71, + "learning_rate": 3.9858373364087586e-06, + "loss": 0.0921, + "step": 13645 + }, + { + "epoch": 5.71, + "learning_rate": 3.98448289993589e-06, + "loss": 0.1228, + "step": 13646 + }, + { + "epoch": 5.71, + "learning_rate": 3.983128636373723e-06, + "loss": 0.1311, + "step": 13647 + }, + { + "epoch": 5.71, + "learning_rate": 3.981774545761188e-06, + "loss": 0.1644, + "step": 13648 + }, + { + "epoch": 5.71, + "learning_rate": 3.980420628137199e-06, + "loss": 0.1409, + "step": 13649 + }, + { + "epoch": 5.71, + "learning_rate": 3.979066883540677e-06, + "loss": 0.115, + "step": 13650 + }, + { + "epoch": 5.72, + "learning_rate": 3.977713312010535e-06, + "loss": 0.1419, + "step": 13651 + }, + { + "epoch": 5.72, + "learning_rate": 3.97635991358568e-06, + "loss": 0.1248, + "step": 13652 + }, + { + "epoch": 5.72, + "learning_rate": 3.975006688305015e-06, + "loss": 0.0984, + "step": 13653 + }, + { + "epoch": 5.72, + "learning_rate": 3.973653636207437e-06, + "loss": 0.0916, + "step": 13654 + }, + { + "epoch": 5.72, + "learning_rate": 3.972300757331836e-06, + "loss": 0.1347, + "step": 13655 + }, + { + "epoch": 5.72, + "learning_rate": 3.970948051717104e-06, + "loss": 0.1198, + "step": 13656 + }, + { + "epoch": 5.72, + "learning_rate": 3.969595519402117e-06, + "loss": 0.1114, + "step": 13657 + }, + { + "epoch": 5.72, + "learning_rate": 3.9682431604257554e-06, + "loss": 0.111, + "step": 13658 + }, + { + "epoch": 5.72, + "learning_rate": 3.966890974826891e-06, + "loss": 0.0948, + "step": 13659 + }, + { + "epoch": 5.72, + "learning_rate": 3.965538962644392e-06, + "loss": 0.0953, + "step": 13660 + }, + { + "epoch": 5.72, + "learning_rate": 3.964187123917122e-06, + "loss": 0.1222, + "step": 13661 + }, + { + "epoch": 5.72, + "learning_rate": 3.962835458683939e-06, + "loss": 0.1081, + "step": 13662 + }, + { + "epoch": 5.72, + "learning_rate": 3.9614839669836905e-06, + "loss": 0.0928, + "step": 13663 + }, + { + "epoch": 5.72, + "learning_rate": 3.960132648855226e-06, + "loss": 0.0574, + "step": 13664 + }, + { + "epoch": 5.72, + "learning_rate": 3.95878150433739e-06, + "loss": 0.1443, + "step": 13665 + }, + { + "epoch": 5.72, + "learning_rate": 3.9574305334690175e-06, + "loss": 0.1078, + "step": 13666 + }, + { + "epoch": 5.72, + "learning_rate": 3.9560797362889436e-06, + "loss": 0.1208, + "step": 13667 + }, + { + "epoch": 5.72, + "learning_rate": 3.954729112835998e-06, + "loss": 0.1301, + "step": 13668 + }, + { + "epoch": 5.72, + "learning_rate": 3.953378663148994e-06, + "loss": 0.1397, + "step": 13669 + }, + { + "epoch": 5.72, + "learning_rate": 3.952028387266757e-06, + "loss": 0.1295, + "step": 13670 + }, + { + "epoch": 5.72, + "learning_rate": 3.950678285228097e-06, + "loss": 0.1382, + "step": 13671 + }, + { + "epoch": 5.72, + "learning_rate": 3.94932835707182e-06, + "loss": 0.1159, + "step": 13672 + }, + { + "epoch": 5.72, + "learning_rate": 3.947978602836731e-06, + "loss": 0.1644, + "step": 13673 + }, + { + "epoch": 5.72, + "learning_rate": 3.946629022561627e-06, + "loss": 0.1093, + "step": 13674 + }, + { + "epoch": 5.73, + "learning_rate": 3.945279616285303e-06, + "loss": 0.117, + "step": 13675 + }, + { + "epoch": 5.73, + "learning_rate": 3.943930384046541e-06, + "loss": 0.1518, + "step": 13676 + }, + { + "epoch": 5.73, + "learning_rate": 3.942581325884124e-06, + "loss": 0.0997, + "step": 13677 + }, + { + "epoch": 5.73, + "learning_rate": 3.941232441836833e-06, + "loss": 0.1424, + "step": 13678 + }, + { + "epoch": 5.73, + "learning_rate": 3.939883731943443e-06, + "loss": 0.1216, + "step": 13679 + }, + { + "epoch": 5.73, + "learning_rate": 3.938535196242713e-06, + "loss": 0.1252, + "step": 13680 + }, + { + "epoch": 5.73, + "learning_rate": 3.93718683477341e-06, + "loss": 0.1371, + "step": 13681 + }, + { + "epoch": 5.73, + "learning_rate": 3.935838647574295e-06, + "loss": 0.098, + "step": 13682 + }, + { + "epoch": 5.73, + "learning_rate": 3.934490634684114e-06, + "loss": 0.1125, + "step": 13683 + }, + { + "epoch": 5.73, + "learning_rate": 3.933142796141616e-06, + "loss": 0.0992, + "step": 13684 + }, + { + "epoch": 5.73, + "learning_rate": 3.931795131985545e-06, + "loss": 0.149, + "step": 13685 + }, + { + "epoch": 5.73, + "learning_rate": 3.930447642254639e-06, + "loss": 0.1305, + "step": 13686 + }, + { + "epoch": 5.73, + "learning_rate": 3.929100326987629e-06, + "loss": 0.1338, + "step": 13687 + }, + { + "epoch": 5.73, + "learning_rate": 3.927753186223243e-06, + "loss": 0.1654, + "step": 13688 + }, + { + "epoch": 5.73, + "learning_rate": 3.926406220000206e-06, + "loss": 0.1465, + "step": 13689 + }, + { + "epoch": 5.73, + "learning_rate": 3.925059428357231e-06, + "loss": 0.1099, + "step": 13690 + }, + { + "epoch": 5.73, + "learning_rate": 3.923712811333031e-06, + "loss": 0.1108, + "step": 13691 + }, + { + "epoch": 5.73, + "learning_rate": 3.922366368966313e-06, + "loss": 0.1221, + "step": 13692 + }, + { + "epoch": 5.73, + "learning_rate": 3.9210201012957816e-06, + "loss": 0.1164, + "step": 13693 + }, + { + "epoch": 5.73, + "learning_rate": 3.919674008360133e-06, + "loss": 0.1125, + "step": 13694 + }, + { + "epoch": 5.73, + "learning_rate": 3.918328090198063e-06, + "loss": 0.1316, + "step": 13695 + }, + { + "epoch": 5.73, + "learning_rate": 3.91698234684825e-06, + "loss": 0.0516, + "step": 13696 + }, + { + "epoch": 5.73, + "learning_rate": 3.915636778349385e-06, + "loss": 0.1248, + "step": 13697 + }, + { + "epoch": 5.73, + "learning_rate": 3.914291384740139e-06, + "loss": 0.1102, + "step": 13698 + }, + { + "epoch": 5.74, + "learning_rate": 3.9129461660591875e-06, + "loss": 0.1044, + "step": 13699 + }, + { + "epoch": 5.74, + "learning_rate": 3.911601122345197e-06, + "loss": 0.0924, + "step": 13700 + }, + { + "epoch": 5.74, + "learning_rate": 3.910256253636832e-06, + "loss": 0.1286, + "step": 13701 + }, + { + "epoch": 5.74, + "learning_rate": 3.9089115599727446e-06, + "loss": 0.135, + "step": 13702 + }, + { + "epoch": 5.74, + "learning_rate": 3.9075670413915925e-06, + "loss": 0.1467, + "step": 13703 + }, + { + "epoch": 5.74, + "learning_rate": 3.906222697932016e-06, + "loss": 0.0969, + "step": 13704 + }, + { + "epoch": 5.74, + "learning_rate": 3.90487852963266e-06, + "loss": 0.1121, + "step": 13705 + }, + { + "epoch": 5.74, + "learning_rate": 3.903534536532163e-06, + "loss": 0.1398, + "step": 13706 + }, + { + "epoch": 5.74, + "learning_rate": 3.902190718669155e-06, + "loss": 0.1083, + "step": 13707 + }, + { + "epoch": 5.74, + "learning_rate": 3.900847076082264e-06, + "loss": 0.1356, + "step": 13708 + }, + { + "epoch": 5.74, + "learning_rate": 3.899503608810114e-06, + "loss": 0.1313, + "step": 13709 + }, + { + "epoch": 5.74, + "learning_rate": 3.898160316891317e-06, + "loss": 0.0883, + "step": 13710 + }, + { + "epoch": 5.74, + "learning_rate": 3.896817200364488e-06, + "loss": 0.1363, + "step": 13711 + }, + { + "epoch": 5.74, + "learning_rate": 3.895474259268232e-06, + "loss": 0.0811, + "step": 13712 + }, + { + "epoch": 5.74, + "learning_rate": 3.89413149364115e-06, + "loss": 0.1213, + "step": 13713 + }, + { + "epoch": 5.74, + "learning_rate": 3.89278890352184e-06, + "loss": 0.1138, + "step": 13714 + }, + { + "epoch": 5.74, + "learning_rate": 3.891446488948894e-06, + "loss": 0.1276, + "step": 13715 + }, + { + "epoch": 5.74, + "learning_rate": 3.890104249960903e-06, + "loss": 0.0994, + "step": 13716 + }, + { + "epoch": 5.74, + "learning_rate": 3.888762186596439e-06, + "loss": 0.1112, + "step": 13717 + }, + { + "epoch": 5.74, + "learning_rate": 3.887420298894082e-06, + "loss": 0.1057, + "step": 13718 + }, + { + "epoch": 5.74, + "learning_rate": 3.886078586892406e-06, + "loss": 0.1357, + "step": 13719 + }, + { + "epoch": 5.74, + "learning_rate": 3.884737050629975e-06, + "loss": 0.136, + "step": 13720 + }, + { + "epoch": 5.74, + "learning_rate": 3.883395690145351e-06, + "loss": 0.1432, + "step": 13721 + }, + { + "epoch": 5.75, + "learning_rate": 3.882054505477092e-06, + "loss": 0.1476, + "step": 13722 + }, + { + "epoch": 5.75, + "learning_rate": 3.880713496663745e-06, + "loss": 0.1501, + "step": 13723 + }, + { + "epoch": 5.75, + "learning_rate": 3.879372663743862e-06, + "loss": 0.1247, + "step": 13724 + }, + { + "epoch": 5.75, + "learning_rate": 3.878032006755977e-06, + "loss": 0.1415, + "step": 13725 + }, + { + "epoch": 5.75, + "learning_rate": 3.87669152573863e-06, + "loss": 0.123, + "step": 13726 + }, + { + "epoch": 5.75, + "learning_rate": 3.87535122073035e-06, + "loss": 0.1068, + "step": 13727 + }, + { + "epoch": 5.75, + "learning_rate": 3.874011091769665e-06, + "loss": 0.1255, + "step": 13728 + }, + { + "epoch": 5.75, + "learning_rate": 3.872671138895095e-06, + "loss": 0.1134, + "step": 13729 + }, + { + "epoch": 5.75, + "learning_rate": 3.871331362145158e-06, + "loss": 0.0983, + "step": 13730 + }, + { + "epoch": 5.75, + "learning_rate": 3.869991761558362e-06, + "loss": 0.1482, + "step": 13731 + }, + { + "epoch": 5.75, + "learning_rate": 3.868652337173211e-06, + "loss": 0.132, + "step": 13732 + }, + { + "epoch": 5.75, + "learning_rate": 3.867313089028208e-06, + "loss": 0.1277, + "step": 13733 + }, + { + "epoch": 5.75, + "learning_rate": 3.8659740171618485e-06, + "loss": 0.1152, + "step": 13734 + }, + { + "epoch": 5.75, + "learning_rate": 3.864635121612622e-06, + "loss": 0.0775, + "step": 13735 + }, + { + "epoch": 5.75, + "learning_rate": 3.863296402419019e-06, + "loss": 0.1493, + "step": 13736 + }, + { + "epoch": 5.75, + "learning_rate": 3.861957859619511e-06, + "loss": 0.125, + "step": 13737 + }, + { + "epoch": 5.75, + "learning_rate": 3.860619493252578e-06, + "loss": 0.1011, + "step": 13738 + }, + { + "epoch": 5.75, + "learning_rate": 3.859281303356689e-06, + "loss": 0.163, + "step": 13739 + }, + { + "epoch": 5.75, + "learning_rate": 3.857943289970311e-06, + "loss": 0.1182, + "step": 13740 + }, + { + "epoch": 5.75, + "learning_rate": 3.856605453131901e-06, + "loss": 0.1114, + "step": 13741 + }, + { + "epoch": 5.75, + "learning_rate": 3.855267792879921e-06, + "loss": 0.1491, + "step": 13742 + }, + { + "epoch": 5.75, + "learning_rate": 3.85393030925281e-06, + "loss": 0.0939, + "step": 13743 + }, + { + "epoch": 5.75, + "learning_rate": 3.85259300228902e-06, + "loss": 0.0992, + "step": 13744 + }, + { + "epoch": 5.75, + "learning_rate": 3.851255872026989e-06, + "loss": 0.1063, + "step": 13745 + }, + { + "epoch": 5.76, + "learning_rate": 3.849918918505151e-06, + "loss": 0.0433, + "step": 13746 + }, + { + "epoch": 5.76, + "learning_rate": 3.848582141761939e-06, + "loss": 0.1548, + "step": 13747 + }, + { + "epoch": 5.76, + "learning_rate": 3.8472455418357735e-06, + "loss": 0.1037, + "step": 13748 + }, + { + "epoch": 5.76, + "learning_rate": 3.845909118765073e-06, + "loss": 0.0993, + "step": 13749 + }, + { + "epoch": 5.76, + "learning_rate": 3.844572872588258e-06, + "loss": 0.1247, + "step": 13750 + }, + { + "epoch": 5.76, + "learning_rate": 3.843236803343731e-06, + "loss": 0.1247, + "step": 13751 + }, + { + "epoch": 5.76, + "learning_rate": 3.841900911069897e-06, + "loss": 0.1141, + "step": 13752 + }, + { + "epoch": 5.76, + "learning_rate": 3.840565195805158e-06, + "loss": 0.0834, + "step": 13753 + }, + { + "epoch": 5.76, + "learning_rate": 3.839229657587906e-06, + "loss": 0.1405, + "step": 13754 + }, + { + "epoch": 5.76, + "learning_rate": 3.83789429645653e-06, + "loss": 0.1399, + "step": 13755 + }, + { + "epoch": 5.76, + "learning_rate": 3.836559112449415e-06, + "loss": 0.1061, + "step": 13756 + }, + { + "epoch": 5.76, + "learning_rate": 3.835224105604942e-06, + "loss": 0.0806, + "step": 13757 + }, + { + "epoch": 5.76, + "learning_rate": 3.833889275961479e-06, + "loss": 0.1568, + "step": 13758 + }, + { + "epoch": 5.76, + "learning_rate": 3.832554623557397e-06, + "loss": 0.1016, + "step": 13759 + }, + { + "epoch": 5.76, + "learning_rate": 3.831220148431058e-06, + "loss": 0.1125, + "step": 13760 + }, + { + "epoch": 5.76, + "learning_rate": 3.829885850620824e-06, + "loss": 0.0723, + "step": 13761 + }, + { + "epoch": 5.76, + "learning_rate": 3.8285517301650445e-06, + "loss": 0.1175, + "step": 13762 + }, + { + "epoch": 5.76, + "learning_rate": 3.827217787102072e-06, + "loss": 0.1329, + "step": 13763 + }, + { + "epoch": 5.76, + "learning_rate": 3.825884021470244e-06, + "loss": 0.1014, + "step": 13764 + }, + { + "epoch": 5.76, + "learning_rate": 3.8245504333079006e-06, + "loss": 0.0892, + "step": 13765 + }, + { + "epoch": 5.76, + "learning_rate": 3.823217022653376e-06, + "loss": 0.1108, + "step": 13766 + }, + { + "epoch": 5.76, + "learning_rate": 3.8218837895449956e-06, + "loss": 0.1202, + "step": 13767 + }, + { + "epoch": 5.76, + "learning_rate": 3.820550734021083e-06, + "loss": 0.1437, + "step": 13768 + }, + { + "epoch": 5.76, + "learning_rate": 3.819217856119961e-06, + "loss": 0.1, + "step": 13769 + }, + { + "epoch": 5.77, + "learning_rate": 3.817885155879933e-06, + "loss": 0.119, + "step": 13770 + }, + { + "epoch": 5.77, + "learning_rate": 3.816552633339313e-06, + "loss": 0.1255, + "step": 13771 + }, + { + "epoch": 5.77, + "learning_rate": 3.815220288536399e-06, + "loss": 0.1156, + "step": 13772 + }, + { + "epoch": 5.77, + "learning_rate": 3.813888121509487e-06, + "loss": 0.1489, + "step": 13773 + }, + { + "epoch": 5.77, + "learning_rate": 3.812556132296873e-06, + "loss": 0.1184, + "step": 13774 + }, + { + "epoch": 5.77, + "learning_rate": 3.8112243209368415e-06, + "loss": 0.1169, + "step": 13775 + }, + { + "epoch": 5.77, + "learning_rate": 3.8098926874676757e-06, + "loss": 0.1359, + "step": 13776 + }, + { + "epoch": 5.77, + "learning_rate": 3.8085612319276556e-06, + "loss": 0.1257, + "step": 13777 + }, + { + "epoch": 5.77, + "learning_rate": 3.8072299543550438e-06, + "loss": 0.1239, + "step": 13778 + }, + { + "epoch": 5.77, + "learning_rate": 3.8058988547881126e-06, + "loss": 0.0871, + "step": 13779 + }, + { + "epoch": 5.77, + "learning_rate": 3.804567933265122e-06, + "loss": 0.0929, + "step": 13780 + }, + { + "epoch": 5.77, + "learning_rate": 3.803237189824328e-06, + "loss": 0.1104, + "step": 13781 + }, + { + "epoch": 5.77, + "learning_rate": 3.8019066245039828e-06, + "loss": 0.0898, + "step": 13782 + }, + { + "epoch": 5.77, + "learning_rate": 3.800576237342335e-06, + "loss": 0.0989, + "step": 13783 + }, + { + "epoch": 5.77, + "learning_rate": 3.799246028377618e-06, + "loss": 0.1339, + "step": 13784 + }, + { + "epoch": 5.77, + "learning_rate": 3.7979159976480717e-06, + "loss": 0.1075, + "step": 13785 + }, + { + "epoch": 5.77, + "learning_rate": 3.796586145191926e-06, + "loss": 0.1306, + "step": 13786 + }, + { + "epoch": 5.77, + "learning_rate": 3.7952564710474072e-06, + "loss": 0.148, + "step": 13787 + }, + { + "epoch": 5.77, + "learning_rate": 3.793926975252735e-06, + "loss": 0.1047, + "step": 13788 + }, + { + "epoch": 5.77, + "learning_rate": 3.7925976578461255e-06, + "loss": 0.143, + "step": 13789 + }, + { + "epoch": 5.77, + "learning_rate": 3.7912685188657905e-06, + "loss": 0.1201, + "step": 13790 + }, + { + "epoch": 5.77, + "learning_rate": 3.789939558349929e-06, + "loss": 0.0978, + "step": 13791 + }, + { + "epoch": 5.77, + "learning_rate": 3.7886107763367474e-06, + "loss": 0.0969, + "step": 13792 + }, + { + "epoch": 5.77, + "learning_rate": 3.787282172864435e-06, + "loss": 0.1366, + "step": 13793 + }, + { + "epoch": 5.78, + "learning_rate": 3.785953747971183e-06, + "loss": 0.1331, + "step": 13794 + }, + { + "epoch": 5.78, + "learning_rate": 3.784625501695175e-06, + "loss": 0.1114, + "step": 13795 + }, + { + "epoch": 5.78, + "learning_rate": 3.783297434074592e-06, + "loss": 0.1316, + "step": 13796 + }, + { + "epoch": 5.78, + "learning_rate": 3.781969545147609e-06, + "loss": 0.1393, + "step": 13797 + }, + { + "epoch": 5.78, + "learning_rate": 3.7806418349523956e-06, + "loss": 0.1353, + "step": 13798 + }, + { + "epoch": 5.78, + "learning_rate": 3.7793143035271106e-06, + "loss": 0.1607, + "step": 13799 + }, + { + "epoch": 5.78, + "learning_rate": 3.7779869509099166e-06, + "loss": 0.0987, + "step": 13800 + }, + { + "epoch": 5.78, + "learning_rate": 3.776659777138966e-06, + "loss": 0.0866, + "step": 13801 + }, + { + "epoch": 5.78, + "learning_rate": 3.7753327822524067e-06, + "loss": 0.1508, + "step": 13802 + }, + { + "epoch": 5.78, + "learning_rate": 3.7740059662883843e-06, + "loss": 0.1343, + "step": 13803 + }, + { + "epoch": 5.78, + "learning_rate": 3.7726793292850382e-06, + "loss": 0.1392, + "step": 13804 + }, + { + "epoch": 5.78, + "learning_rate": 3.771352871280495e-06, + "loss": 0.1397, + "step": 13805 + }, + { + "epoch": 5.78, + "learning_rate": 3.7700265923128866e-06, + "loss": 0.087, + "step": 13806 + }, + { + "epoch": 5.78, + "learning_rate": 3.768700492420335e-06, + "loss": 0.1246, + "step": 13807 + }, + { + "epoch": 5.78, + "learning_rate": 3.7673745716409595e-06, + "loss": 0.1021, + "step": 13808 + }, + { + "epoch": 5.78, + "learning_rate": 3.7660488300128706e-06, + "loss": 0.1268, + "step": 13809 + }, + { + "epoch": 5.78, + "learning_rate": 3.7647232675741794e-06, + "loss": 0.0973, + "step": 13810 + }, + { + "epoch": 5.78, + "learning_rate": 3.7633978843629813e-06, + "loss": 0.0975, + "step": 13811 + }, + { + "epoch": 5.78, + "learning_rate": 3.7620726804173767e-06, + "loss": 0.1035, + "step": 13812 + }, + { + "epoch": 5.78, + "learning_rate": 3.7607476557754574e-06, + "loss": 0.1264, + "step": 13813 + }, + { + "epoch": 5.78, + "learning_rate": 3.7594228104753107e-06, + "loss": 0.1345, + "step": 13814 + }, + { + "epoch": 5.78, + "learning_rate": 3.75809814455502e-06, + "loss": 0.1187, + "step": 13815 + }, + { + "epoch": 5.78, + "learning_rate": 3.756773658052657e-06, + "loss": 0.1339, + "step": 13816 + }, + { + "epoch": 5.78, + "learning_rate": 3.755449351006294e-06, + "loss": 0.1081, + "step": 13817 + }, + { + "epoch": 5.79, + "learning_rate": 3.7541252234540028e-06, + "loss": 0.1106, + "step": 13818 + }, + { + "epoch": 5.79, + "learning_rate": 3.752801275433835e-06, + "loss": 0.1587, + "step": 13819 + }, + { + "epoch": 5.79, + "learning_rate": 3.7514775069838503e-06, + "loss": 0.1327, + "step": 13820 + }, + { + "epoch": 5.79, + "learning_rate": 3.750153918142101e-06, + "loss": 0.1187, + "step": 13821 + }, + { + "epoch": 5.79, + "learning_rate": 3.74883050894663e-06, + "loss": 0.1307, + "step": 13822 + }, + { + "epoch": 5.79, + "learning_rate": 3.7475072794354803e-06, + "loss": 0.0944, + "step": 13823 + }, + { + "epoch": 5.79, + "learning_rate": 3.746184229646689e-06, + "loss": 0.0824, + "step": 13824 + }, + { + "epoch": 5.79, + "learning_rate": 3.744861359618278e-06, + "loss": 0.1315, + "step": 13825 + }, + { + "epoch": 5.79, + "learning_rate": 3.7435386693882767e-06, + "loss": 0.1121, + "step": 13826 + }, + { + "epoch": 5.79, + "learning_rate": 3.742216158994705e-06, + "loss": 0.0785, + "step": 13827 + }, + { + "epoch": 5.79, + "learning_rate": 3.740893828475577e-06, + "loss": 0.1419, + "step": 13828 + }, + { + "epoch": 5.79, + "learning_rate": 3.7395716778689006e-06, + "loss": 0.1205, + "step": 13829 + }, + { + "epoch": 5.79, + "learning_rate": 3.738249707212681e-06, + "loss": 0.1453, + "step": 13830 + }, + { + "epoch": 5.79, + "learning_rate": 3.736927916544921e-06, + "loss": 0.129, + "step": 13831 + }, + { + "epoch": 5.79, + "learning_rate": 3.7356063059036073e-06, + "loss": 0.1384, + "step": 13832 + }, + { + "epoch": 5.79, + "learning_rate": 3.73428487532673e-06, + "loss": 0.1283, + "step": 13833 + }, + { + "epoch": 5.79, + "learning_rate": 3.732963624852275e-06, + "loss": 0.1252, + "step": 13834 + }, + { + "epoch": 5.79, + "learning_rate": 3.731642554518219e-06, + "loss": 0.1, + "step": 13835 + }, + { + "epoch": 5.79, + "learning_rate": 3.7303216643625364e-06, + "loss": 0.1462, + "step": 13836 + }, + { + "epoch": 5.79, + "learning_rate": 3.7290009544231955e-06, + "loss": 0.1298, + "step": 13837 + }, + { + "epoch": 5.79, + "learning_rate": 3.727680424738155e-06, + "loss": 0.1198, + "step": 13838 + }, + { + "epoch": 5.79, + "learning_rate": 3.726360075345378e-06, + "loss": 0.1026, + "step": 13839 + }, + { + "epoch": 5.79, + "learning_rate": 3.7250399062828104e-06, + "loss": 0.1226, + "step": 13840 + }, + { + "epoch": 5.79, + "learning_rate": 3.723719917588402e-06, + "loss": 0.1262, + "step": 13841 + }, + { + "epoch": 5.8, + "learning_rate": 3.722400109300095e-06, + "loss": 0.1533, + "step": 13842 + }, + { + "epoch": 5.8, + "learning_rate": 3.7210804814558253e-06, + "loss": 0.1316, + "step": 13843 + }, + { + "epoch": 5.8, + "learning_rate": 3.7197610340935263e-06, + "loss": 0.1205, + "step": 13844 + }, + { + "epoch": 5.8, + "learning_rate": 3.7184417672511273e-06, + "loss": 0.0987, + "step": 13845 + }, + { + "epoch": 5.8, + "learning_rate": 3.7171226809665417e-06, + "loss": 0.1437, + "step": 13846 + }, + { + "epoch": 5.8, + "learning_rate": 3.71580377527769e-06, + "loss": 0.1369, + "step": 13847 + }, + { + "epoch": 5.8, + "learning_rate": 3.7144850502224816e-06, + "loss": 0.135, + "step": 13848 + }, + { + "epoch": 5.8, + "learning_rate": 3.713166505838822e-06, + "loss": 0.102, + "step": 13849 + }, + { + "epoch": 5.8, + "learning_rate": 3.7118481421646134e-06, + "loss": 0.0825, + "step": 13850 + }, + { + "epoch": 5.8, + "learning_rate": 3.710529959237753e-06, + "loss": 0.0986, + "step": 13851 + }, + { + "epoch": 5.8, + "learning_rate": 3.7092119570961248e-06, + "loss": 0.1431, + "step": 13852 + }, + { + "epoch": 5.8, + "learning_rate": 3.7078941357776153e-06, + "loss": 0.0948, + "step": 13853 + }, + { + "epoch": 5.8, + "learning_rate": 3.7065764953201066e-06, + "loss": 0.1081, + "step": 13854 + }, + { + "epoch": 5.8, + "learning_rate": 3.7052590357614717e-06, + "loss": 0.0667, + "step": 13855 + }, + { + "epoch": 5.8, + "learning_rate": 3.7039417571395796e-06, + "loss": 0.1089, + "step": 13856 + }, + { + "epoch": 5.8, + "learning_rate": 3.7026246594922944e-06, + "loss": 0.1096, + "step": 13857 + }, + { + "epoch": 5.8, + "learning_rate": 3.7013077428574794e-06, + "loss": 0.1008, + "step": 13858 + }, + { + "epoch": 5.8, + "learning_rate": 3.6999910072729808e-06, + "loss": 0.1199, + "step": 13859 + }, + { + "epoch": 5.8, + "learning_rate": 3.698674452776653e-06, + "loss": 0.1097, + "step": 13860 + }, + { + "epoch": 5.8, + "learning_rate": 3.6973580794063323e-06, + "loss": 0.1472, + "step": 13861 + }, + { + "epoch": 5.8, + "learning_rate": 3.6960418871998615e-06, + "loss": 0.1396, + "step": 13862 + }, + { + "epoch": 5.8, + "learning_rate": 3.6947258761950723e-06, + "loss": 0.1408, + "step": 13863 + }, + { + "epoch": 5.8, + "learning_rate": 3.693410046429793e-06, + "loss": 0.0973, + "step": 13864 + }, + { + "epoch": 5.8, + "learning_rate": 3.6920943979418478e-06, + "loss": 0.1614, + "step": 13865 + }, + { + "epoch": 5.81, + "learning_rate": 3.690778930769049e-06, + "loss": 0.1382, + "step": 13866 + }, + { + "epoch": 5.81, + "learning_rate": 3.6894636449492116e-06, + "loss": 0.1073, + "step": 13867 + }, + { + "epoch": 5.81, + "learning_rate": 3.688148540520141e-06, + "loss": 0.1038, + "step": 13868 + }, + { + "epoch": 5.81, + "learning_rate": 3.6868336175196406e-06, + "loss": 0.1121, + "step": 13869 + }, + { + "epoch": 5.81, + "learning_rate": 3.6855188759855043e-06, + "loss": 0.125, + "step": 13870 + }, + { + "epoch": 5.81, + "learning_rate": 3.6842043159555263e-06, + "loss": 0.1256, + "step": 13871 + }, + { + "epoch": 5.81, + "learning_rate": 3.6828899374674933e-06, + "loss": 0.0791, + "step": 13872 + }, + { + "epoch": 5.81, + "learning_rate": 3.681575740559179e-06, + "loss": 0.1311, + "step": 13873 + }, + { + "epoch": 5.81, + "learning_rate": 3.680261725268364e-06, + "loss": 0.122, + "step": 13874 + }, + { + "epoch": 5.81, + "learning_rate": 3.678947891632818e-06, + "loss": 0.1136, + "step": 13875 + }, + { + "epoch": 5.81, + "learning_rate": 3.677634239690305e-06, + "loss": 0.0916, + "step": 13876 + }, + { + "epoch": 5.81, + "learning_rate": 3.676320769478585e-06, + "loss": 0.1307, + "step": 13877 + }, + { + "epoch": 5.81, + "learning_rate": 3.6750074810354173e-06, + "loss": 0.1332, + "step": 13878 + }, + { + "epoch": 5.81, + "learning_rate": 3.673694374398543e-06, + "loss": 0.1622, + "step": 13879 + }, + { + "epoch": 5.81, + "learning_rate": 3.672381449605711e-06, + "loss": 0.1185, + "step": 13880 + }, + { + "epoch": 5.81, + "learning_rate": 3.671068706694658e-06, + "loss": 0.1075, + "step": 13881 + }, + { + "epoch": 5.81, + "learning_rate": 3.6697561457031196e-06, + "loss": 0.0994, + "step": 13882 + }, + { + "epoch": 5.81, + "learning_rate": 3.6684437666688253e-06, + "loss": 0.1085, + "step": 13883 + }, + { + "epoch": 5.81, + "learning_rate": 3.667131569629495e-06, + "loss": 0.1145, + "step": 13884 + }, + { + "epoch": 5.81, + "learning_rate": 3.665819554622847e-06, + "loss": 0.1068, + "step": 13885 + }, + { + "epoch": 5.81, + "learning_rate": 3.6645077216865987e-06, + "loss": 0.0967, + "step": 13886 + }, + { + "epoch": 5.81, + "learning_rate": 3.6631960708584504e-06, + "loss": 0.1445, + "step": 13887 + }, + { + "epoch": 5.81, + "learning_rate": 3.6618846021761078e-06, + "loss": 0.1084, + "step": 13888 + }, + { + "epoch": 5.81, + "learning_rate": 3.660573315677268e-06, + "loss": 0.099, + "step": 13889 + }, + { + "epoch": 5.82, + "learning_rate": 3.6592622113996235e-06, + "loss": 0.1187, + "step": 13890 + }, + { + "epoch": 5.82, + "learning_rate": 3.657951289380859e-06, + "loss": 0.0873, + "step": 13891 + }, + { + "epoch": 5.82, + "learning_rate": 3.656640549658661e-06, + "loss": 0.1233, + "step": 13892 + }, + { + "epoch": 5.82, + "learning_rate": 3.6553299922706977e-06, + "loss": 0.0894, + "step": 13893 + }, + { + "epoch": 5.82, + "learning_rate": 3.654019617254644e-06, + "loss": 0.0926, + "step": 13894 + }, + { + "epoch": 5.82, + "learning_rate": 3.652709424648164e-06, + "loss": 0.0958, + "step": 13895 + }, + { + "epoch": 5.82, + "learning_rate": 3.65139941448892e-06, + "loss": 0.1193, + "step": 13896 + }, + { + "epoch": 5.82, + "learning_rate": 3.650089586814566e-06, + "loss": 0.1366, + "step": 13897 + }, + { + "epoch": 5.82, + "learning_rate": 3.648779941662752e-06, + "loss": 0.124, + "step": 13898 + }, + { + "epoch": 5.82, + "learning_rate": 3.6474704790711258e-06, + "loss": 0.1464, + "step": 13899 + }, + { + "epoch": 5.82, + "learning_rate": 3.6461611990773204e-06, + "loss": 0.1439, + "step": 13900 + }, + { + "epoch": 5.82, + "learning_rate": 3.6448521017189732e-06, + "loss": 0.1262, + "step": 13901 + }, + { + "epoch": 5.82, + "learning_rate": 3.6435431870337123e-06, + "loss": 0.1203, + "step": 13902 + }, + { + "epoch": 5.82, + "learning_rate": 3.6422344550591625e-06, + "loss": 0.1148, + "step": 13903 + }, + { + "epoch": 5.82, + "learning_rate": 3.6409259058329404e-06, + "loss": 0.082, + "step": 13904 + }, + { + "epoch": 5.82, + "learning_rate": 3.6396175393926646e-06, + "loss": 0.095, + "step": 13905 + }, + { + "epoch": 5.82, + "learning_rate": 3.6383093557759386e-06, + "loss": 0.1278, + "step": 13906 + }, + { + "epoch": 5.82, + "learning_rate": 3.637001355020361e-06, + "loss": 0.1405, + "step": 13907 + }, + { + "epoch": 5.82, + "learning_rate": 3.6356935371635336e-06, + "loss": 0.0856, + "step": 13908 + }, + { + "epoch": 5.82, + "learning_rate": 3.6343859022430474e-06, + "loss": 0.1269, + "step": 13909 + }, + { + "epoch": 5.82, + "learning_rate": 3.63307845029649e-06, + "loss": 0.1347, + "step": 13910 + }, + { + "epoch": 5.82, + "learning_rate": 3.631771181361442e-06, + "loss": 0.117, + "step": 13911 + }, + { + "epoch": 5.82, + "learning_rate": 3.630464095475481e-06, + "loss": 0.1103, + "step": 13912 + }, + { + "epoch": 5.82, + "learning_rate": 3.6291571926761805e-06, + "loss": 0.1385, + "step": 13913 + }, + { + "epoch": 5.83, + "learning_rate": 3.6278504730011e-06, + "loss": 0.1068, + "step": 13914 + }, + { + "epoch": 5.83, + "learning_rate": 3.626543936487803e-06, + "loss": 0.0955, + "step": 13915 + }, + { + "epoch": 5.83, + "learning_rate": 3.625237583173844e-06, + "loss": 0.0907, + "step": 13916 + }, + { + "epoch": 5.83, + "learning_rate": 3.6239314130967753e-06, + "loss": 0.1233, + "step": 13917 + }, + { + "epoch": 5.83, + "learning_rate": 3.6226254262941398e-06, + "loss": 0.1122, + "step": 13918 + }, + { + "epoch": 5.83, + "learning_rate": 3.6213196228034796e-06, + "loss": 0.1352, + "step": 13919 + }, + { + "epoch": 5.83, + "learning_rate": 3.6200140026623233e-06, + "loss": 0.1447, + "step": 13920 + }, + { + "epoch": 5.83, + "learning_rate": 3.6187085659082034e-06, + "loss": 0.1134, + "step": 13921 + }, + { + "epoch": 5.83, + "learning_rate": 3.6174033125786434e-06, + "loss": 0.1147, + "step": 13922 + }, + { + "epoch": 5.83, + "learning_rate": 3.6160982427111614e-06, + "loss": 0.0617, + "step": 13923 + }, + { + "epoch": 5.83, + "learning_rate": 3.614793356343269e-06, + "loss": 0.1265, + "step": 13924 + }, + { + "epoch": 5.83, + "learning_rate": 3.6134886535124804e-06, + "loss": 0.0609, + "step": 13925 + }, + { + "epoch": 5.83, + "learning_rate": 3.612184134256289e-06, + "loss": 0.1177, + "step": 13926 + }, + { + "epoch": 5.83, + "learning_rate": 3.6108797986121967e-06, + "loss": 0.1149, + "step": 13927 + }, + { + "epoch": 5.83, + "learning_rate": 3.609575646617699e-06, + "loss": 0.1173, + "step": 13928 + }, + { + "epoch": 5.83, + "learning_rate": 3.6082716783102735e-06, + "loss": 0.0993, + "step": 13929 + }, + { + "epoch": 5.83, + "learning_rate": 3.6069678937274076e-06, + "loss": 0.1415, + "step": 13930 + }, + { + "epoch": 5.83, + "learning_rate": 3.6056642929065765e-06, + "loss": 0.088, + "step": 13931 + }, + { + "epoch": 5.83, + "learning_rate": 3.604360875885251e-06, + "loss": 0.1072, + "step": 13932 + }, + { + "epoch": 5.83, + "learning_rate": 3.6030576427009e-06, + "loss": 0.131, + "step": 13933 + }, + { + "epoch": 5.83, + "learning_rate": 3.601754593390978e-06, + "loss": 0.1314, + "step": 13934 + }, + { + "epoch": 5.83, + "learning_rate": 3.600451727992942e-06, + "loss": 0.1126, + "step": 13935 + }, + { + "epoch": 5.83, + "learning_rate": 3.5991490465442413e-06, + "loss": 0.1463, + "step": 13936 + }, + { + "epoch": 5.84, + "learning_rate": 3.597846549082322e-06, + "loss": 0.1579, + "step": 13937 + }, + { + "epoch": 5.84, + "learning_rate": 3.5965442356446224e-06, + "loss": 0.0985, + "step": 13938 + }, + { + "epoch": 5.84, + "learning_rate": 3.595242106268577e-06, + "loss": 0.1163, + "step": 13939 + }, + { + "epoch": 5.84, + "learning_rate": 3.593940160991617e-06, + "loss": 0.1439, + "step": 13940 + }, + { + "epoch": 5.84, + "learning_rate": 3.592638399851158e-06, + "loss": 0.1152, + "step": 13941 + }, + { + "epoch": 5.84, + "learning_rate": 3.5913368228846236e-06, + "loss": 0.1404, + "step": 13942 + }, + { + "epoch": 5.84, + "learning_rate": 3.5900354301294236e-06, + "loss": 0.1272, + "step": 13943 + }, + { + "epoch": 5.84, + "learning_rate": 3.5887342216229694e-06, + "loss": 0.1131, + "step": 13944 + }, + { + "epoch": 5.84, + "learning_rate": 3.587433197402659e-06, + "loss": 0.1285, + "step": 13945 + }, + { + "epoch": 5.84, + "learning_rate": 3.586132357505895e-06, + "loss": 0.1054, + "step": 13946 + }, + { + "epoch": 5.84, + "learning_rate": 3.5848317019700608e-06, + "loss": 0.09, + "step": 13947 + }, + { + "epoch": 5.84, + "learning_rate": 3.5835312308325475e-06, + "loss": 0.1308, + "step": 13948 + }, + { + "epoch": 5.84, + "learning_rate": 3.5822309441307347e-06, + "loss": 0.1002, + "step": 13949 + }, + { + "epoch": 5.84, + "learning_rate": 3.5809308419019985e-06, + "loss": 0.0906, + "step": 13950 + }, + { + "epoch": 5.84, + "learning_rate": 3.5796309241837125e-06, + "loss": 0.1281, + "step": 13951 + }, + { + "epoch": 5.84, + "learning_rate": 3.5783311910132348e-06, + "loss": 0.1261, + "step": 13952 + }, + { + "epoch": 5.84, + "learning_rate": 3.5770316424279283e-06, + "loss": 0.1357, + "step": 13953 + }, + { + "epoch": 5.84, + "learning_rate": 3.5757322784651506e-06, + "loss": 0.1148, + "step": 13954 + }, + { + "epoch": 5.84, + "learning_rate": 3.5744330991622457e-06, + "loss": 0.0935, + "step": 13955 + }, + { + "epoch": 5.84, + "learning_rate": 3.573134104556558e-06, + "loss": 0.0908, + "step": 13956 + }, + { + "epoch": 5.84, + "learning_rate": 3.5718352946854286e-06, + "loss": 0.1236, + "step": 13957 + }, + { + "epoch": 5.84, + "learning_rate": 3.5705366695861886e-06, + "loss": 0.0759, + "step": 13958 + }, + { + "epoch": 5.84, + "learning_rate": 3.569238229296167e-06, + "loss": 0.1156, + "step": 13959 + }, + { + "epoch": 5.84, + "learning_rate": 3.567939973852689e-06, + "loss": 0.0935, + "step": 13960 + }, + { + "epoch": 5.85, + "learning_rate": 3.5666419032930653e-06, + "loss": 0.1497, + "step": 13961 + }, + { + "epoch": 5.85, + "learning_rate": 3.5653440176546115e-06, + "loss": 0.0916, + "step": 13962 + }, + { + "epoch": 5.85, + "learning_rate": 3.564046316974634e-06, + "loss": 0.0757, + "step": 13963 + }, + { + "epoch": 5.85, + "learning_rate": 3.5627488012904333e-06, + "loss": 0.0908, + "step": 13964 + }, + { + "epoch": 5.85, + "learning_rate": 3.5614514706393066e-06, + "loss": 0.1181, + "step": 13965 + }, + { + "epoch": 5.85, + "learning_rate": 3.560154325058547e-06, + "loss": 0.0886, + "step": 13966 + }, + { + "epoch": 5.85, + "learning_rate": 3.558857364585434e-06, + "loss": 0.1648, + "step": 13967 + }, + { + "epoch": 5.85, + "learning_rate": 3.5575605892572495e-06, + "loss": 0.0831, + "step": 13968 + }, + { + "epoch": 5.85, + "learning_rate": 3.556263999111269e-06, + "loss": 0.1066, + "step": 13969 + }, + { + "epoch": 5.85, + "learning_rate": 3.554967594184762e-06, + "loss": 0.1088, + "step": 13970 + }, + { + "epoch": 5.85, + "learning_rate": 3.553671374514992e-06, + "loss": 0.1059, + "step": 13971 + }, + { + "epoch": 5.85, + "learning_rate": 3.552375340139219e-06, + "loss": 0.1399, + "step": 13972 + }, + { + "epoch": 5.85, + "learning_rate": 3.551079491094698e-06, + "loss": 0.1216, + "step": 13973 + }, + { + "epoch": 5.85, + "learning_rate": 3.5497838274186737e-06, + "loss": 0.151, + "step": 13974 + }, + { + "epoch": 5.85, + "learning_rate": 3.5484883491483866e-06, + "loss": 0.1209, + "step": 13975 + }, + { + "epoch": 5.85, + "learning_rate": 3.5471930563210775e-06, + "loss": 0.14, + "step": 13976 + }, + { + "epoch": 5.85, + "learning_rate": 3.5458979489739778e-06, + "loss": 0.1103, + "step": 13977 + }, + { + "epoch": 5.85, + "learning_rate": 3.5446030271443132e-06, + "loss": 0.1041, + "step": 13978 + }, + { + "epoch": 5.85, + "learning_rate": 3.543308290869307e-06, + "loss": 0.0979, + "step": 13979 + }, + { + "epoch": 5.85, + "learning_rate": 3.5420137401861745e-06, + "loss": 0.1379, + "step": 13980 + }, + { + "epoch": 5.85, + "learning_rate": 3.540719375132129e-06, + "loss": 0.089, + "step": 13981 + }, + { + "epoch": 5.85, + "learning_rate": 3.5394251957443704e-06, + "loss": 0.1059, + "step": 13982 + }, + { + "epoch": 5.85, + "learning_rate": 3.538131202060102e-06, + "loss": 0.1206, + "step": 13983 + }, + { + "epoch": 5.85, + "learning_rate": 3.5368373941165167e-06, + "loss": 0.117, + "step": 13984 + }, + { + "epoch": 5.86, + "learning_rate": 3.5355437719508067e-06, + "loss": 0.0811, + "step": 13985 + }, + { + "epoch": 5.86, + "learning_rate": 3.5342503356001535e-06, + "loss": 0.1076, + "step": 13986 + }, + { + "epoch": 5.86, + "learning_rate": 3.5329570851017405e-06, + "loss": 0.112, + "step": 13987 + }, + { + "epoch": 5.86, + "learning_rate": 3.5316640204927342e-06, + "loss": 0.1049, + "step": 13988 + }, + { + "epoch": 5.86, + "learning_rate": 3.5303711418103047e-06, + "loss": 0.1239, + "step": 13989 + }, + { + "epoch": 5.86, + "learning_rate": 3.5290784490916175e-06, + "loss": 0.139, + "step": 13990 + }, + { + "epoch": 5.86, + "learning_rate": 3.527785942373827e-06, + "loss": 0.1611, + "step": 13991 + }, + { + "epoch": 5.86, + "learning_rate": 3.526493621694086e-06, + "loss": 0.1448, + "step": 13992 + }, + { + "epoch": 5.86, + "learning_rate": 3.5252014870895447e-06, + "loss": 0.1252, + "step": 13993 + }, + { + "epoch": 5.86, + "learning_rate": 3.5239095385973377e-06, + "loss": 0.1505, + "step": 13994 + }, + { + "epoch": 5.86, + "learning_rate": 3.522617776254604e-06, + "loss": 0.1341, + "step": 13995 + }, + { + "epoch": 5.86, + "learning_rate": 3.5213262000984793e-06, + "loss": 0.1043, + "step": 13996 + }, + { + "epoch": 5.86, + "learning_rate": 3.52003481016608e-06, + "loss": 0.1103, + "step": 13997 + }, + { + "epoch": 5.86, + "learning_rate": 3.518743606494529e-06, + "loss": 0.1107, + "step": 13998 + }, + { + "epoch": 5.86, + "learning_rate": 3.517452589120941e-06, + "loss": 0.0967, + "step": 13999 + }, + { + "epoch": 5.86, + "learning_rate": 3.516161758082427e-06, + "loss": 0.1111, + "step": 14000 + }, + { + "epoch": 5.86, + "learning_rate": 3.514871113416093e-06, + "loss": 0.0789, + "step": 14001 + }, + { + "epoch": 5.86, + "learning_rate": 3.51358065515903e-06, + "loss": 0.1114, + "step": 14002 + }, + { + "epoch": 5.86, + "learning_rate": 3.5122903833483357e-06, + "loss": 0.1079, + "step": 14003 + }, + { + "epoch": 5.86, + "learning_rate": 3.511000298021098e-06, + "loss": 0.118, + "step": 14004 + }, + { + "epoch": 5.86, + "learning_rate": 3.5097103992143978e-06, + "loss": 0.1236, + "step": 14005 + }, + { + "epoch": 5.86, + "learning_rate": 3.5084206869653126e-06, + "loss": 0.1465, + "step": 14006 + }, + { + "epoch": 5.86, + "learning_rate": 3.507131161310917e-06, + "loss": 0.1022, + "step": 14007 + }, + { + "epoch": 5.86, + "learning_rate": 3.505841822288272e-06, + "loss": 0.0969, + "step": 14008 + }, + { + "epoch": 5.87, + "learning_rate": 3.5045526699344413e-06, + "loss": 0.1046, + "step": 14009 + }, + { + "epoch": 5.87, + "learning_rate": 3.5032637042864793e-06, + "loss": 0.1009, + "step": 14010 + }, + { + "epoch": 5.87, + "learning_rate": 3.5019749253814372e-06, + "loss": 0.1093, + "step": 14011 + }, + { + "epoch": 5.87, + "learning_rate": 3.50068633325636e-06, + "loss": 0.1327, + "step": 14012 + }, + { + "epoch": 5.87, + "learning_rate": 3.499397927948286e-06, + "loss": 0.0909, + "step": 14013 + }, + { + "epoch": 5.87, + "learning_rate": 3.498109709494254e-06, + "loss": 0.1392, + "step": 14014 + }, + { + "epoch": 5.87, + "learning_rate": 3.496821677931286e-06, + "loss": 0.1085, + "step": 14015 + }, + { + "epoch": 5.87, + "learning_rate": 3.495533833296406e-06, + "loss": 0.1361, + "step": 14016 + }, + { + "epoch": 5.87, + "learning_rate": 3.4942461756266356e-06, + "loss": 0.1297, + "step": 14017 + }, + { + "epoch": 5.87, + "learning_rate": 3.4929587049589853e-06, + "loss": 0.1175, + "step": 14018 + }, + { + "epoch": 5.87, + "learning_rate": 3.491671421330467e-06, + "loss": 0.0825, + "step": 14019 + }, + { + "epoch": 5.87, + "learning_rate": 3.4903843247780743e-06, + "loss": 0.1452, + "step": 14020 + }, + { + "epoch": 5.87, + "learning_rate": 3.489097415338808e-06, + "loss": 0.1271, + "step": 14021 + }, + { + "epoch": 5.87, + "learning_rate": 3.4878106930496616e-06, + "loss": 0.1284, + "step": 14022 + }, + { + "epoch": 5.87, + "learning_rate": 3.486524157947615e-06, + "loss": 0.0917, + "step": 14023 + }, + { + "epoch": 5.87, + "learning_rate": 3.4852378100696505e-06, + "loss": 0.1334, + "step": 14024 + }, + { + "epoch": 5.87, + "learning_rate": 3.4839516494527457e-06, + "loss": 0.1086, + "step": 14025 + }, + { + "epoch": 5.87, + "learning_rate": 3.4826656761338673e-06, + "loss": 0.1463, + "step": 14026 + }, + { + "epoch": 5.87, + "learning_rate": 3.481379890149982e-06, + "loss": 0.1149, + "step": 14027 + }, + { + "epoch": 5.87, + "learning_rate": 3.4800942915380485e-06, + "loss": 0.127, + "step": 14028 + }, + { + "epoch": 5.87, + "learning_rate": 3.4788088803350173e-06, + "loss": 0.111, + "step": 14029 + }, + { + "epoch": 5.87, + "learning_rate": 3.477523656577837e-06, + "loss": 0.1165, + "step": 14030 + }, + { + "epoch": 5.87, + "learning_rate": 3.4762386203034527e-06, + "loss": 0.1217, + "step": 14031 + }, + { + "epoch": 5.87, + "learning_rate": 3.474953771548799e-06, + "loss": 0.1365, + "step": 14032 + }, + { + "epoch": 5.88, + "learning_rate": 3.473669110350809e-06, + "loss": 0.1499, + "step": 14033 + }, + { + "epoch": 5.88, + "learning_rate": 3.472384636746412e-06, + "loss": 0.1611, + "step": 14034 + }, + { + "epoch": 5.88, + "learning_rate": 3.471100350772524e-06, + "loss": 0.0953, + "step": 14035 + }, + { + "epoch": 5.88, + "learning_rate": 3.4698162524660615e-06, + "loss": 0.1255, + "step": 14036 + }, + { + "epoch": 5.88, + "learning_rate": 3.468532341863937e-06, + "loss": 0.1541, + "step": 14037 + }, + { + "epoch": 5.88, + "learning_rate": 3.4672486190030543e-06, + "loss": 0.1318, + "step": 14038 + }, + { + "epoch": 5.88, + "learning_rate": 3.4659650839203127e-06, + "loss": 0.1182, + "step": 14039 + }, + { + "epoch": 5.88, + "learning_rate": 3.4646817366526066e-06, + "loss": 0.1264, + "step": 14040 + }, + { + "epoch": 5.88, + "learning_rate": 3.4633985772368274e-06, + "loss": 0.1143, + "step": 14041 + }, + { + "epoch": 5.88, + "learning_rate": 3.4621156057098573e-06, + "loss": 0.1463, + "step": 14042 + }, + { + "epoch": 5.88, + "learning_rate": 3.4608328221085674e-06, + "loss": 0.1425, + "step": 14043 + }, + { + "epoch": 5.88, + "learning_rate": 3.4595502264698356e-06, + "loss": 0.1706, + "step": 14044 + }, + { + "epoch": 5.88, + "learning_rate": 3.4582678188305297e-06, + "loss": 0.1091, + "step": 14045 + }, + { + "epoch": 5.88, + "learning_rate": 3.4569855992275093e-06, + "loss": 0.1409, + "step": 14046 + }, + { + "epoch": 5.88, + "learning_rate": 3.4557035676976313e-06, + "loss": 0.0968, + "step": 14047 + }, + { + "epoch": 5.88, + "learning_rate": 3.454421724277751e-06, + "loss": 0.1727, + "step": 14048 + }, + { + "epoch": 5.88, + "learning_rate": 3.453140069004707e-06, + "loss": 0.1531, + "step": 14049 + }, + { + "epoch": 5.88, + "learning_rate": 3.451858601915341e-06, + "loss": 0.121, + "step": 14050 + }, + { + "epoch": 5.88, + "learning_rate": 3.45057732304649e-06, + "loss": 0.1237, + "step": 14051 + }, + { + "epoch": 5.88, + "learning_rate": 3.449296232434981e-06, + "loss": 0.0952, + "step": 14052 + }, + { + "epoch": 5.88, + "learning_rate": 3.448015330117639e-06, + "loss": 0.1005, + "step": 14053 + }, + { + "epoch": 5.88, + "learning_rate": 3.446734616131282e-06, + "loss": 0.1103, + "step": 14054 + }, + { + "epoch": 5.88, + "learning_rate": 3.4454540905127277e-06, + "loss": 0.1124, + "step": 14055 + }, + { + "epoch": 5.88, + "learning_rate": 3.444173753298775e-06, + "loss": 0.114, + "step": 14056 + }, + { + "epoch": 5.89, + "learning_rate": 3.4428936045262296e-06, + "loss": 0.0952, + "step": 14057 + }, + { + "epoch": 5.89, + "learning_rate": 3.4416136442318894e-06, + "loss": 0.1132, + "step": 14058 + }, + { + "epoch": 5.89, + "learning_rate": 3.440333872452546e-06, + "loss": 0.1363, + "step": 14059 + }, + { + "epoch": 5.89, + "learning_rate": 3.4390542892249833e-06, + "loss": 0.0637, + "step": 14060 + }, + { + "epoch": 5.89, + "learning_rate": 3.437774894585987e-06, + "loss": 0.1097, + "step": 14061 + }, + { + "epoch": 5.89, + "learning_rate": 3.4364956885723246e-06, + "loss": 0.1029, + "step": 14062 + }, + { + "epoch": 5.89, + "learning_rate": 3.435216671220769e-06, + "loss": 0.1439, + "step": 14063 + }, + { + "epoch": 5.89, + "learning_rate": 3.433937842568088e-06, + "loss": 0.1027, + "step": 14064 + }, + { + "epoch": 5.89, + "learning_rate": 3.432659202651034e-06, + "loss": 0.0999, + "step": 14065 + }, + { + "epoch": 5.89, + "learning_rate": 3.4313807515063635e-06, + "loss": 0.1, + "step": 14066 + }, + { + "epoch": 5.89, + "learning_rate": 3.4301024891708245e-06, + "loss": 0.1165, + "step": 14067 + }, + { + "epoch": 5.89, + "learning_rate": 3.42882441568116e-06, + "loss": 0.1132, + "step": 14068 + }, + { + "epoch": 5.89, + "learning_rate": 3.4275465310741083e-06, + "loss": 0.1187, + "step": 14069 + }, + { + "epoch": 5.89, + "learning_rate": 3.426268835386397e-06, + "loss": 0.0904, + "step": 14070 + }, + { + "epoch": 5.89, + "learning_rate": 3.4249913286547553e-06, + "loss": 0.1018, + "step": 14071 + }, + { + "epoch": 5.89, + "learning_rate": 3.4237140109159015e-06, + "loss": 0.1082, + "step": 14072 + }, + { + "epoch": 5.89, + "learning_rate": 3.4224368822065547e-06, + "loss": 0.1081, + "step": 14073 + }, + { + "epoch": 5.89, + "learning_rate": 3.421159942563421e-06, + "loss": 0.1229, + "step": 14074 + }, + { + "epoch": 5.89, + "learning_rate": 3.4198831920232113e-06, + "loss": 0.0917, + "step": 14075 + }, + { + "epoch": 5.89, + "learning_rate": 3.4186066306226174e-06, + "loss": 0.0752, + "step": 14076 + }, + { + "epoch": 5.89, + "learning_rate": 3.4173302583983335e-06, + "loss": 0.1235, + "step": 14077 + }, + { + "epoch": 5.89, + "learning_rate": 3.416054075387052e-06, + "loss": 0.0999, + "step": 14078 + }, + { + "epoch": 5.89, + "learning_rate": 3.414778081625453e-06, + "loss": 0.1152, + "step": 14079 + }, + { + "epoch": 5.89, + "learning_rate": 3.4135022771502146e-06, + "loss": 0.0914, + "step": 14080 + }, + { + "epoch": 5.9, + "learning_rate": 3.4122266619980083e-06, + "loss": 0.0757, + "step": 14081 + }, + { + "epoch": 5.9, + "learning_rate": 3.410951236205503e-06, + "loss": 0.0896, + "step": 14082 + }, + { + "epoch": 5.9, + "learning_rate": 3.4096759998093552e-06, + "loss": 0.1502, + "step": 14083 + }, + { + "epoch": 5.9, + "learning_rate": 3.4084009528462227e-06, + "loss": 0.1271, + "step": 14084 + }, + { + "epoch": 5.9, + "learning_rate": 3.407126095352755e-06, + "loss": 0.1263, + "step": 14085 + }, + { + "epoch": 5.9, + "learning_rate": 3.4058514273655984e-06, + "loss": 0.1307, + "step": 14086 + }, + { + "epoch": 5.9, + "learning_rate": 3.404576948921393e-06, + "loss": 0.1591, + "step": 14087 + }, + { + "epoch": 5.9, + "learning_rate": 3.4033026600567677e-06, + "loss": 0.1583, + "step": 14088 + }, + { + "epoch": 5.9, + "learning_rate": 3.402028560808357e-06, + "loss": 0.1016, + "step": 14089 + }, + { + "epoch": 5.9, + "learning_rate": 3.4007546512127764e-06, + "loss": 0.1096, + "step": 14090 + }, + { + "epoch": 5.9, + "learning_rate": 3.3994809313066477e-06, + "loss": 0.1148, + "step": 14091 + }, + { + "epoch": 5.9, + "learning_rate": 3.398207401126583e-06, + "loss": 0.1304, + "step": 14092 + }, + { + "epoch": 5.9, + "learning_rate": 3.3969340607091884e-06, + "loss": 0.0871, + "step": 14093 + }, + { + "epoch": 5.9, + "learning_rate": 3.3956609100910643e-06, + "loss": 0.1377, + "step": 14094 + }, + { + "epoch": 5.9, + "learning_rate": 3.394387949308806e-06, + "loss": 0.1212, + "step": 14095 + }, + { + "epoch": 5.9, + "learning_rate": 3.3931151783990092e-06, + "loss": 0.1206, + "step": 14096 + }, + { + "epoch": 5.9, + "learning_rate": 3.3918425973982496e-06, + "loss": 0.0933, + "step": 14097 + }, + { + "epoch": 5.9, + "learning_rate": 3.39057020634311e-06, + "loss": 0.0626, + "step": 14098 + }, + { + "epoch": 5.9, + "learning_rate": 3.389298005270165e-06, + "loss": 0.1276, + "step": 14099 + }, + { + "epoch": 5.9, + "learning_rate": 3.3880259942159833e-06, + "loss": 0.1148, + "step": 14100 + }, + { + "epoch": 5.9, + "learning_rate": 3.386754173217126e-06, + "loss": 0.1493, + "step": 14101 + }, + { + "epoch": 5.9, + "learning_rate": 3.3854825423101555e-06, + "loss": 0.1155, + "step": 14102 + }, + { + "epoch": 5.9, + "learning_rate": 3.384211101531617e-06, + "loss": 0.0956, + "step": 14103 + }, + { + "epoch": 5.9, + "learning_rate": 3.382939850918059e-06, + "loss": 0.0938, + "step": 14104 + }, + { + "epoch": 5.91, + "learning_rate": 3.3816687905060232e-06, + "loss": 0.1378, + "step": 14105 + }, + { + "epoch": 5.91, + "learning_rate": 3.380397920332045e-06, + "loss": 0.1013, + "step": 14106 + }, + { + "epoch": 5.91, + "learning_rate": 3.3791272404326535e-06, + "loss": 0.1205, + "step": 14107 + }, + { + "epoch": 5.91, + "learning_rate": 3.3778567508443782e-06, + "loss": 0.106, + "step": 14108 + }, + { + "epoch": 5.91, + "learning_rate": 3.3765864516037317e-06, + "loss": 0.1332, + "step": 14109 + }, + { + "epoch": 5.91, + "learning_rate": 3.3753163427472324e-06, + "loss": 0.1093, + "step": 14110 + }, + { + "epoch": 5.91, + "learning_rate": 3.3740464243113834e-06, + "loss": 0.1576, + "step": 14111 + }, + { + "epoch": 5.91, + "learning_rate": 3.3727766963326903e-06, + "loss": 0.1264, + "step": 14112 + }, + { + "epoch": 5.91, + "learning_rate": 3.37150715884765e-06, + "loss": 0.0976, + "step": 14113 + }, + { + "epoch": 5.91, + "learning_rate": 3.3702378118927537e-06, + "loss": 0.1251, + "step": 14114 + }, + { + "epoch": 5.91, + "learning_rate": 3.3689686555044888e-06, + "loss": 0.1625, + "step": 14115 + }, + { + "epoch": 5.91, + "learning_rate": 3.3676996897193393e-06, + "loss": 0.1327, + "step": 14116 + }, + { + "epoch": 5.91, + "learning_rate": 3.366430914573774e-06, + "loss": 0.1531, + "step": 14117 + }, + { + "epoch": 5.91, + "learning_rate": 3.3651623301042646e-06, + "loss": 0.1045, + "step": 14118 + }, + { + "epoch": 5.91, + "learning_rate": 3.363893936347277e-06, + "loss": 0.1105, + "step": 14119 + }, + { + "epoch": 5.91, + "learning_rate": 3.3626257333392687e-06, + "loss": 0.1239, + "step": 14120 + }, + { + "epoch": 5.91, + "learning_rate": 3.3613577211166936e-06, + "loss": 0.0995, + "step": 14121 + }, + { + "epoch": 5.91, + "learning_rate": 3.3600898997160005e-06, + "loss": 0.1189, + "step": 14122 + }, + { + "epoch": 5.91, + "learning_rate": 3.358822269173634e-06, + "loss": 0.1224, + "step": 14123 + }, + { + "epoch": 5.91, + "learning_rate": 3.357554829526025e-06, + "loss": 0.1318, + "step": 14124 + }, + { + "epoch": 5.91, + "learning_rate": 3.3562875808096083e-06, + "loss": 0.1149, + "step": 14125 + }, + { + "epoch": 5.91, + "learning_rate": 3.355020523060809e-06, + "loss": 0.1103, + "step": 14126 + }, + { + "epoch": 5.91, + "learning_rate": 3.3537536563160487e-06, + "loss": 0.1313, + "step": 14127 + }, + { + "epoch": 5.92, + "learning_rate": 3.352486980611741e-06, + "loss": 0.1419, + "step": 14128 + }, + { + "epoch": 5.92, + "learning_rate": 3.3512204959843e-06, + "loss": 0.1164, + "step": 14129 + }, + { + "epoch": 5.92, + "learning_rate": 3.3499542024701226e-06, + "loss": 0.1058, + "step": 14130 + }, + { + "epoch": 5.92, + "learning_rate": 3.3486881001056104e-06, + "loss": 0.1403, + "step": 14131 + }, + { + "epoch": 5.92, + "learning_rate": 3.3474221889271597e-06, + "loss": 0.0979, + "step": 14132 + }, + { + "epoch": 5.92, + "learning_rate": 3.346156468971151e-06, + "loss": 0.1352, + "step": 14133 + }, + { + "epoch": 5.92, + "learning_rate": 3.3448909402739716e-06, + "loss": 0.1098, + "step": 14134 + }, + { + "epoch": 5.92, + "learning_rate": 3.343625602871995e-06, + "loss": 0.1001, + "step": 14135 + }, + { + "epoch": 5.92, + "learning_rate": 3.3423604568015944e-06, + "loss": 0.139, + "step": 14136 + }, + { + "epoch": 5.92, + "learning_rate": 3.3410955020991377e-06, + "loss": 0.1335, + "step": 14137 + }, + { + "epoch": 5.92, + "learning_rate": 3.3398307388009786e-06, + "loss": 0.1197, + "step": 14138 + }, + { + "epoch": 5.92, + "learning_rate": 3.3385661669434744e-06, + "loss": 0.094, + "step": 14139 + }, + { + "epoch": 5.92, + "learning_rate": 3.3373017865629742e-06, + "loss": 0.1103, + "step": 14140 + }, + { + "epoch": 5.92, + "learning_rate": 3.3360375976958226e-06, + "loss": 0.1332, + "step": 14141 + }, + { + "epoch": 5.92, + "learning_rate": 3.334773600378357e-06, + "loss": 0.1125, + "step": 14142 + }, + { + "epoch": 5.92, + "learning_rate": 3.3335097946469118e-06, + "loss": 0.1256, + "step": 14143 + }, + { + "epoch": 5.92, + "learning_rate": 3.3322461805378094e-06, + "loss": 0.1456, + "step": 14144 + }, + { + "epoch": 5.92, + "learning_rate": 3.3309827580873734e-06, + "loss": 0.1428, + "step": 14145 + }, + { + "epoch": 5.92, + "learning_rate": 3.3297195273319195e-06, + "loss": 0.131, + "step": 14146 + }, + { + "epoch": 5.92, + "learning_rate": 3.32845648830776e-06, + "loss": 0.122, + "step": 14147 + }, + { + "epoch": 5.92, + "learning_rate": 3.3271936410511974e-06, + "loss": 0.1043, + "step": 14148 + }, + { + "epoch": 5.92, + "learning_rate": 3.3259309855985353e-06, + "loss": 0.1327, + "step": 14149 + }, + { + "epoch": 5.92, + "learning_rate": 3.324668521986062e-06, + "loss": 0.1611, + "step": 14150 + }, + { + "epoch": 5.92, + "learning_rate": 3.3234062502500685e-06, + "loss": 0.0954, + "step": 14151 + }, + { + "epoch": 5.93, + "learning_rate": 3.3221441704268375e-06, + "loss": 0.1467, + "step": 14152 + }, + { + "epoch": 5.93, + "learning_rate": 3.3208822825526464e-06, + "loss": 0.0981, + "step": 14153 + }, + { + "epoch": 5.93, + "learning_rate": 3.3196205866637667e-06, + "loss": 0.114, + "step": 14154 + }, + { + "epoch": 5.93, + "learning_rate": 3.3183590827964696e-06, + "loss": 0.1119, + "step": 14155 + }, + { + "epoch": 5.93, + "learning_rate": 3.3170977709870066e-06, + "loss": 0.12, + "step": 14156 + }, + { + "epoch": 5.93, + "learning_rate": 3.3158366512716424e-06, + "loss": 0.1203, + "step": 14157 + }, + { + "epoch": 5.93, + "learning_rate": 3.314575723686618e-06, + "loss": 0.1199, + "step": 14158 + }, + { + "epoch": 5.93, + "learning_rate": 3.3133149882681836e-06, + "loss": 0.1252, + "step": 14159 + }, + { + "epoch": 5.93, + "learning_rate": 3.3120544450525747e-06, + "loss": 0.095, + "step": 14160 + }, + { + "epoch": 5.93, + "learning_rate": 3.3107940940760264e-06, + "loss": 0.1227, + "step": 14161 + }, + { + "epoch": 5.93, + "learning_rate": 3.3095339353747657e-06, + "loss": 0.0897, + "step": 14162 + }, + { + "epoch": 5.93, + "learning_rate": 3.308273968985015e-06, + "loss": 0.1321, + "step": 14163 + }, + { + "epoch": 5.93, + "learning_rate": 3.3070141949429945e-06, + "loss": 0.0816, + "step": 14164 + }, + { + "epoch": 5.93, + "learning_rate": 3.3057546132849095e-06, + "loss": 0.1089, + "step": 14165 + }, + { + "epoch": 5.93, + "learning_rate": 3.304495224046966e-06, + "loss": 0.1081, + "step": 14166 + }, + { + "epoch": 5.93, + "learning_rate": 3.3032360272653664e-06, + "loss": 0.1332, + "step": 14167 + }, + { + "epoch": 5.93, + "learning_rate": 3.3019770229763048e-06, + "loss": 0.1229, + "step": 14168 + }, + { + "epoch": 5.93, + "learning_rate": 3.30071821121597e-06, + "loss": 0.1548, + "step": 14169 + }, + { + "epoch": 5.93, + "learning_rate": 3.2994595920205485e-06, + "loss": 0.1346, + "step": 14170 + }, + { + "epoch": 5.93, + "learning_rate": 3.2982011654262112e-06, + "loss": 0.1042, + "step": 14171 + }, + { + "epoch": 5.93, + "learning_rate": 3.296942931469136e-06, + "loss": 0.1263, + "step": 14172 + }, + { + "epoch": 5.93, + "learning_rate": 3.2956848901854863e-06, + "loss": 0.1211, + "step": 14173 + }, + { + "epoch": 5.93, + "learning_rate": 3.2944270416114256e-06, + "loss": 0.1317, + "step": 14174 + }, + { + "epoch": 5.93, + "learning_rate": 3.2931693857831094e-06, + "loss": 0.1582, + "step": 14175 + }, + { + "epoch": 5.94, + "learning_rate": 3.2919119227366913e-06, + "loss": 0.1371, + "step": 14176 + }, + { + "epoch": 5.94, + "learning_rate": 3.2906546525083082e-06, + "loss": 0.14, + "step": 14177 + }, + { + "epoch": 5.94, + "learning_rate": 3.2893975751341078e-06, + "loss": 0.1377, + "step": 14178 + }, + { + "epoch": 5.94, + "learning_rate": 3.2881406906502146e-06, + "loss": 0.134, + "step": 14179 + }, + { + "epoch": 5.94, + "learning_rate": 3.2868839990927627e-06, + "loss": 0.1461, + "step": 14180 + }, + { + "epoch": 5.94, + "learning_rate": 3.285627500497872e-06, + "loss": 0.1224, + "step": 14181 + }, + { + "epoch": 5.94, + "learning_rate": 3.284371194901661e-06, + "loss": 0.1106, + "step": 14182 + }, + { + "epoch": 5.94, + "learning_rate": 3.2831150823402403e-06, + "loss": 0.1034, + "step": 14183 + }, + { + "epoch": 5.94, + "learning_rate": 3.28185916284972e-06, + "loss": 0.1049, + "step": 14184 + }, + { + "epoch": 5.94, + "learning_rate": 3.280603436466192e-06, + "loss": 0.0994, + "step": 14185 + }, + { + "epoch": 5.94, + "learning_rate": 3.279347903225757e-06, + "loss": 0.1359, + "step": 14186 + }, + { + "epoch": 5.94, + "learning_rate": 3.278092563164502e-06, + "loss": 0.1302, + "step": 14187 + }, + { + "epoch": 5.94, + "learning_rate": 3.2768374163185114e-06, + "loss": 0.0997, + "step": 14188 + }, + { + "epoch": 5.94, + "learning_rate": 3.275582462723863e-06, + "loss": 0.1485, + "step": 14189 + }, + { + "epoch": 5.94, + "learning_rate": 3.274327702416633e-06, + "loss": 0.1328, + "step": 14190 + }, + { + "epoch": 5.94, + "learning_rate": 3.2730731354328827e-06, + "loss": 0.0914, + "step": 14191 + }, + { + "epoch": 5.94, + "learning_rate": 3.2718187618086737e-06, + "loss": 0.1373, + "step": 14192 + }, + { + "epoch": 5.94, + "learning_rate": 3.2705645815800658e-06, + "loss": 0.1274, + "step": 14193 + }, + { + "epoch": 5.94, + "learning_rate": 3.269310594783106e-06, + "loss": 0.0989, + "step": 14194 + }, + { + "epoch": 5.94, + "learning_rate": 3.2680568014538415e-06, + "loss": 0.1325, + "step": 14195 + }, + { + "epoch": 5.94, + "learning_rate": 3.2668032016283103e-06, + "loss": 0.1238, + "step": 14196 + }, + { + "epoch": 5.94, + "learning_rate": 3.2655497953425484e-06, + "loss": 0.1115, + "step": 14197 + }, + { + "epoch": 5.94, + "learning_rate": 3.26429658263258e-06, + "loss": 0.1297, + "step": 14198 + }, + { + "epoch": 5.94, + "learning_rate": 3.2630435635344283e-06, + "loss": 0.0975, + "step": 14199 + }, + { + "epoch": 5.95, + "learning_rate": 3.261790738084114e-06, + "loss": 0.1321, + "step": 14200 + }, + { + "epoch": 5.95, + "learning_rate": 3.2605381063176423e-06, + "loss": 0.0961, + "step": 14201 + }, + { + "epoch": 5.95, + "learning_rate": 3.259285668271023e-06, + "loss": 0.1038, + "step": 14202 + }, + { + "epoch": 5.95, + "learning_rate": 3.2580334239802545e-06, + "loss": 0.1285, + "step": 14203 + }, + { + "epoch": 5.95, + "learning_rate": 3.2567813734813336e-06, + "loss": 0.0986, + "step": 14204 + }, + { + "epoch": 5.95, + "learning_rate": 3.255529516810251e-06, + "loss": 0.128, + "step": 14205 + }, + { + "epoch": 5.95, + "learning_rate": 3.254277854002984e-06, + "loss": 0.0905, + "step": 14206 + }, + { + "epoch": 5.95, + "learning_rate": 3.2530263850955147e-06, + "loss": 0.1159, + "step": 14207 + }, + { + "epoch": 5.95, + "learning_rate": 3.251775110123814e-06, + "loss": 0.1275, + "step": 14208 + }, + { + "epoch": 5.95, + "learning_rate": 3.2505240291238505e-06, + "loss": 0.1211, + "step": 14209 + }, + { + "epoch": 5.95, + "learning_rate": 3.2492731421315838e-06, + "loss": 0.0921, + "step": 14210 + }, + { + "epoch": 5.95, + "learning_rate": 3.2480224491829728e-06, + "loss": 0.1079, + "step": 14211 + }, + { + "epoch": 5.95, + "learning_rate": 3.2467719503139626e-06, + "loss": 0.1004, + "step": 14212 + }, + { + "epoch": 5.95, + "learning_rate": 3.2455216455605e-06, + "loss": 0.1377, + "step": 14213 + }, + { + "epoch": 5.95, + "learning_rate": 3.244271534958523e-06, + "loss": 0.1096, + "step": 14214 + }, + { + "epoch": 5.95, + "learning_rate": 3.2430216185439667e-06, + "loss": 0.102, + "step": 14215 + }, + { + "epoch": 5.95, + "learning_rate": 3.2417718963527576e-06, + "loss": 0.1079, + "step": 14216 + }, + { + "epoch": 5.95, + "learning_rate": 3.2405223684208208e-06, + "loss": 0.12, + "step": 14217 + }, + { + "epoch": 5.95, + "learning_rate": 3.2392730347840686e-06, + "loss": 0.1327, + "step": 14218 + }, + { + "epoch": 5.95, + "learning_rate": 3.2380238954784115e-06, + "loss": 0.1363, + "step": 14219 + }, + { + "epoch": 5.95, + "learning_rate": 3.2367749505397582e-06, + "loss": 0.1274, + "step": 14220 + }, + { + "epoch": 5.95, + "learning_rate": 3.2355262000040077e-06, + "loss": 0.1217, + "step": 14221 + }, + { + "epoch": 5.95, + "learning_rate": 3.2342776439070545e-06, + "loss": 0.087, + "step": 14222 + }, + { + "epoch": 5.95, + "learning_rate": 3.2330292822847885e-06, + "loss": 0.1326, + "step": 14223 + }, + { + "epoch": 5.96, + "learning_rate": 3.231781115173088e-06, + "loss": 0.0972, + "step": 14224 + }, + { + "epoch": 5.96, + "learning_rate": 3.2305331426078367e-06, + "loss": 0.1557, + "step": 14225 + }, + { + "epoch": 5.96, + "learning_rate": 3.2292853646249002e-06, + "loss": 0.1304, + "step": 14226 + }, + { + "epoch": 5.96, + "learning_rate": 3.228037781260147e-06, + "loss": 0.1231, + "step": 14227 + }, + { + "epoch": 5.96, + "learning_rate": 3.2267903925494394e-06, + "loss": 0.1325, + "step": 14228 + }, + { + "epoch": 5.96, + "learning_rate": 3.22554319852863e-06, + "loss": 0.1139, + "step": 14229 + }, + { + "epoch": 5.96, + "learning_rate": 3.2242961992335707e-06, + "loss": 0.0909, + "step": 14230 + }, + { + "epoch": 5.96, + "learning_rate": 3.2230493947001084e-06, + "loss": 0.1008, + "step": 14231 + }, + { + "epoch": 5.96, + "learning_rate": 3.221802784964073e-06, + "loss": 0.1182, + "step": 14232 + }, + { + "epoch": 5.96, + "learning_rate": 3.2205563700613028e-06, + "loss": 0.123, + "step": 14233 + }, + { + "epoch": 5.96, + "learning_rate": 3.2193101500276226e-06, + "loss": 0.0822, + "step": 14234 + }, + { + "epoch": 5.96, + "learning_rate": 3.2180641248988566e-06, + "loss": 0.1333, + "step": 14235 + }, + { + "epoch": 5.96, + "learning_rate": 3.2168182947108173e-06, + "loss": 0.0668, + "step": 14236 + }, + { + "epoch": 5.96, + "learning_rate": 3.2155726594993177e-06, + "loss": 0.1355, + "step": 14237 + }, + { + "epoch": 5.96, + "learning_rate": 3.2143272193001652e-06, + "loss": 0.1102, + "step": 14238 + }, + { + "epoch": 5.96, + "learning_rate": 3.2130819741491513e-06, + "loss": 0.1288, + "step": 14239 + }, + { + "epoch": 5.96, + "learning_rate": 3.2118369240820745e-06, + "loss": 0.1122, + "step": 14240 + }, + { + "epoch": 5.96, + "learning_rate": 3.2105920691347216e-06, + "loss": 0.1103, + "step": 14241 + }, + { + "epoch": 5.96, + "learning_rate": 3.2093474093428733e-06, + "loss": 0.118, + "step": 14242 + }, + { + "epoch": 5.96, + "learning_rate": 3.2081029447423095e-06, + "loss": 0.1368, + "step": 14243 + }, + { + "epoch": 5.96, + "learning_rate": 3.2068586753688014e-06, + "loss": 0.1325, + "step": 14244 + }, + { + "epoch": 5.96, + "learning_rate": 3.20561460125811e-06, + "loss": 0.0986, + "step": 14245 + }, + { + "epoch": 5.96, + "learning_rate": 3.2043707224460006e-06, + "loss": 0.1024, + "step": 14246 + }, + { + "epoch": 5.96, + "learning_rate": 3.203127038968222e-06, + "loss": 0.1025, + "step": 14247 + }, + { + "epoch": 5.97, + "learning_rate": 3.201883550860525e-06, + "loss": 0.1007, + "step": 14248 + }, + { + "epoch": 5.97, + "learning_rate": 3.2006402581586527e-06, + "loss": 0.0998, + "step": 14249 + }, + { + "epoch": 5.97, + "learning_rate": 3.199397160898342e-06, + "loss": 0.1018, + "step": 14250 + }, + { + "epoch": 5.97, + "learning_rate": 3.1981542591153247e-06, + "loss": 0.1279, + "step": 14251 + }, + { + "epoch": 5.97, + "learning_rate": 3.196911552845332e-06, + "loss": 0.1581, + "step": 14252 + }, + { + "epoch": 5.97, + "learning_rate": 3.1956690421240754e-06, + "loss": 0.1524, + "step": 14253 + }, + { + "epoch": 5.97, + "learning_rate": 3.1944267269872754e-06, + "loss": 0.147, + "step": 14254 + }, + { + "epoch": 5.97, + "learning_rate": 3.1931846074706386e-06, + "loss": 0.1009, + "step": 14255 + }, + { + "epoch": 5.97, + "learning_rate": 3.191942683609871e-06, + "loss": 0.108, + "step": 14256 + }, + { + "epoch": 5.97, + "learning_rate": 3.190700955440669e-06, + "loss": 0.1121, + "step": 14257 + }, + { + "epoch": 5.97, + "learning_rate": 3.18945942299873e-06, + "loss": 0.1387, + "step": 14258 + }, + { + "epoch": 5.97, + "learning_rate": 3.1882180863197308e-06, + "loss": 0.121, + "step": 14259 + }, + { + "epoch": 5.97, + "learning_rate": 3.1869769454393596e-06, + "loss": 0.1329, + "step": 14260 + }, + { + "epoch": 5.97, + "learning_rate": 3.18573600039329e-06, + "loss": 0.1233, + "step": 14261 + }, + { + "epoch": 5.97, + "learning_rate": 3.1844952512171924e-06, + "loss": 0.094, + "step": 14262 + }, + { + "epoch": 5.97, + "learning_rate": 3.1832546979467304e-06, + "loss": 0.1106, + "step": 14263 + }, + { + "epoch": 5.97, + "learning_rate": 3.1820143406175642e-06, + "loss": 0.1171, + "step": 14264 + }, + { + "epoch": 5.97, + "learning_rate": 3.180774179265348e-06, + "loss": 0.1091, + "step": 14265 + }, + { + "epoch": 5.97, + "learning_rate": 3.1795342139257233e-06, + "loss": 0.1443, + "step": 14266 + }, + { + "epoch": 5.97, + "learning_rate": 3.1782944446343357e-06, + "loss": 0.1002, + "step": 14267 + }, + { + "epoch": 5.97, + "learning_rate": 3.1770548714268236e-06, + "loss": 0.0809, + "step": 14268 + }, + { + "epoch": 5.97, + "learning_rate": 3.175815494338811e-06, + "loss": 0.1243, + "step": 14269 + }, + { + "epoch": 5.97, + "learning_rate": 3.174576313405927e-06, + "loss": 0.1191, + "step": 14270 + }, + { + "epoch": 5.97, + "learning_rate": 3.17333732866379e-06, + "loss": 0.1145, + "step": 14271 + }, + { + "epoch": 5.98, + "learning_rate": 3.1720985401480176e-06, + "loss": 0.1007, + "step": 14272 + }, + { + "epoch": 5.98, + "learning_rate": 3.1708599478942094e-06, + "loss": 0.1246, + "step": 14273 + }, + { + "epoch": 5.98, + "learning_rate": 3.1696215519379726e-06, + "loss": 0.1157, + "step": 14274 + }, + { + "epoch": 5.98, + "learning_rate": 3.168383352314902e-06, + "loss": 0.1044, + "step": 14275 + }, + { + "epoch": 5.98, + "learning_rate": 3.1671453490605906e-06, + "loss": 0.1279, + "step": 14276 + }, + { + "epoch": 5.98, + "learning_rate": 3.1659075422106233e-06, + "loss": 0.1473, + "step": 14277 + }, + { + "epoch": 5.98, + "learning_rate": 3.164669931800578e-06, + "loss": 0.1236, + "step": 14278 + }, + { + "epoch": 5.98, + "learning_rate": 3.163432517866034e-06, + "loss": 0.1409, + "step": 14279 + }, + { + "epoch": 5.98, + "learning_rate": 3.162195300442552e-06, + "loss": 0.1223, + "step": 14280 + }, + { + "epoch": 5.98, + "learning_rate": 3.160958279565699e-06, + "loss": 0.1252, + "step": 14281 + }, + { + "epoch": 5.98, + "learning_rate": 3.15972145527103e-06, + "loss": 0.1619, + "step": 14282 + }, + { + "epoch": 5.98, + "learning_rate": 3.158484827594099e-06, + "loss": 0.0831, + "step": 14283 + }, + { + "epoch": 5.98, + "learning_rate": 3.1572483965704503e-06, + "loss": 0.1201, + "step": 14284 + }, + { + "epoch": 5.98, + "learning_rate": 3.156012162235628e-06, + "loss": 0.1578, + "step": 14285 + }, + { + "epoch": 5.98, + "learning_rate": 3.1547761246251587e-06, + "loss": 0.1137, + "step": 14286 + }, + { + "epoch": 5.98, + "learning_rate": 3.1535402837745753e-06, + "loss": 0.1081, + "step": 14287 + }, + { + "epoch": 5.98, + "learning_rate": 3.1523046397194015e-06, + "loss": 0.1271, + "step": 14288 + }, + { + "epoch": 5.98, + "learning_rate": 3.151069192495154e-06, + "loss": 0.1111, + "step": 14289 + }, + { + "epoch": 5.98, + "learning_rate": 3.149833942137346e-06, + "loss": 0.1529, + "step": 14290 + }, + { + "epoch": 5.98, + "learning_rate": 3.1485988886814846e-06, + "loss": 0.1303, + "step": 14291 + }, + { + "epoch": 5.98, + "learning_rate": 3.147364032163065e-06, + "loss": 0.1105, + "step": 14292 + }, + { + "epoch": 5.98, + "learning_rate": 3.146129372617589e-06, + "loss": 0.1075, + "step": 14293 + }, + { + "epoch": 5.98, + "learning_rate": 3.1448949100805394e-06, + "loss": 0.107, + "step": 14294 + }, + { + "epoch": 5.98, + "learning_rate": 3.1436606445874017e-06, + "loss": 0.1458, + "step": 14295 + }, + { + "epoch": 5.99, + "learning_rate": 3.142426576173655e-06, + "loss": 0.1445, + "step": 14296 + }, + { + "epoch": 5.99, + "learning_rate": 3.1411927048747715e-06, + "loss": 0.1194, + "step": 14297 + }, + { + "epoch": 5.99, + "learning_rate": 3.1399590307262175e-06, + "loss": 0.1093, + "step": 14298 + }, + { + "epoch": 5.99, + "learning_rate": 3.1387255537634553e-06, + "loss": 0.1034, + "step": 14299 + }, + { + "epoch": 5.99, + "learning_rate": 3.137492274021937e-06, + "loss": 0.0787, + "step": 14300 + }, + { + "epoch": 5.99, + "learning_rate": 3.1362591915371123e-06, + "loss": 0.1017, + "step": 14301 + }, + { + "epoch": 5.99, + "learning_rate": 3.1350263063444264e-06, + "loss": 0.1271, + "step": 14302 + }, + { + "epoch": 5.99, + "learning_rate": 3.1337936184793185e-06, + "loss": 0.1039, + "step": 14303 + }, + { + "epoch": 5.99, + "learning_rate": 3.1325611279772194e-06, + "loss": 0.0875, + "step": 14304 + }, + { + "epoch": 5.99, + "learning_rate": 3.131328834873557e-06, + "loss": 0.1423, + "step": 14305 + }, + { + "epoch": 5.99, + "learning_rate": 3.130096739203755e-06, + "loss": 0.1137, + "step": 14306 + }, + { + "epoch": 5.99, + "learning_rate": 3.128864841003223e-06, + "loss": 0.107, + "step": 14307 + }, + { + "epoch": 5.99, + "learning_rate": 3.1276331403073733e-06, + "loss": 0.1402, + "step": 14308 + }, + { + "epoch": 5.99, + "learning_rate": 3.1264016371516106e-06, + "loss": 0.1349, + "step": 14309 + }, + { + "epoch": 5.99, + "learning_rate": 3.1251703315713333e-06, + "loss": 0.1357, + "step": 14310 + }, + { + "epoch": 5.99, + "learning_rate": 3.123939223601935e-06, + "loss": 0.156, + "step": 14311 + }, + { + "epoch": 5.99, + "learning_rate": 3.122708313278805e-06, + "loss": 0.1171, + "step": 14312 + }, + { + "epoch": 5.99, + "learning_rate": 3.1214776006373205e-06, + "loss": 0.1377, + "step": 14313 + }, + { + "epoch": 5.99, + "learning_rate": 3.1202470857128564e-06, + "loss": 0.1145, + "step": 14314 + }, + { + "epoch": 5.99, + "learning_rate": 3.1190167685407848e-06, + "loss": 0.0946, + "step": 14315 + }, + { + "epoch": 5.99, + "learning_rate": 3.11778664915647e-06, + "loss": 0.1007, + "step": 14316 + }, + { + "epoch": 5.99, + "learning_rate": 3.1165567275952712e-06, + "loss": 0.1388, + "step": 14317 + }, + { + "epoch": 5.99, + "learning_rate": 3.1153270038925397e-06, + "loss": 0.1399, + "step": 14318 + }, + { + "epoch": 5.99, + "learning_rate": 3.1140974780836254e-06, + "loss": 0.1044, + "step": 14319 + }, + { + "epoch": 6.0, + "learning_rate": 3.1128681502038717e-06, + "loss": 0.1562, + "step": 14320 + }, + { + "epoch": 6.0, + "learning_rate": 3.1116390202886083e-06, + "loss": 0.1052, + "step": 14321 + }, + { + "epoch": 6.0, + "learning_rate": 3.110410088373168e-06, + "loss": 0.1, + "step": 14322 + }, + { + "epoch": 6.0, + "learning_rate": 3.1091813544928763e-06, + "loss": 0.1175, + "step": 14323 + }, + { + "epoch": 6.0, + "learning_rate": 3.107952818683052e-06, + "loss": 0.1078, + "step": 14324 + }, + { + "epoch": 6.0, + "learning_rate": 3.1067244809790076e-06, + "loss": 0.1231, + "step": 14325 + }, + { + "epoch": 6.0, + "learning_rate": 3.1054963414160545e-06, + "loss": 0.1265, + "step": 14326 + }, + { + "epoch": 6.0, + "learning_rate": 3.1042684000294876e-06, + "loss": 0.118, + "step": 14327 + }, + { + "epoch": 6.0, + "learning_rate": 3.1030406568546066e-06, + "loss": 0.1147, + "step": 14328 + }, + { + "epoch": 6.0, + "learning_rate": 3.1018131119267014e-06, + "loss": 0.0994, + "step": 14329 + }, + { + "epoch": 6.0, + "learning_rate": 3.100585765281057e-06, + "loss": 0.1185, + "step": 14330 + }, + { + "epoch": 6.0, + "learning_rate": 3.099358616952952e-06, + "loss": 0.1292, + "step": 14331 + }, + { + "epoch": 6.0, + "learning_rate": 3.098131666977664e-06, + "loss": 0.1093, + "step": 14332 + }, + { + "epoch": 6.0, + "learning_rate": 3.096904915390454e-06, + "loss": 0.0678, + "step": 14333 + }, + { + "epoch": 6.0, + "learning_rate": 3.095678362226585e-06, + "loss": 0.1091, + "step": 14334 + }, + { + "epoch": 6.0, + "learning_rate": 3.094452007521316e-06, + "loss": 0.1122, + "step": 14335 + }, + { + "epoch": 6.0, + "learning_rate": 3.0932258513098978e-06, + "loss": 0.0605, + "step": 14336 + }, + { + "epoch": 6.0, + "learning_rate": 3.091999893627572e-06, + "loss": 0.071, + "step": 14337 + }, + { + "epoch": 6.0, + "learning_rate": 3.090774134509579e-06, + "loss": 0.0975, + "step": 14338 + }, + { + "epoch": 6.0, + "learning_rate": 3.089548573991151e-06, + "loss": 0.0612, + "step": 14339 + }, + { + "epoch": 6.0, + "learning_rate": 3.088323212107521e-06, + "loss": 0.1026, + "step": 14340 + }, + { + "epoch": 6.0, + "learning_rate": 3.087098048893904e-06, + "loss": 0.0785, + "step": 14341 + }, + { + "epoch": 6.0, + "learning_rate": 3.0858730843855187e-06, + "loss": 0.0928, + "step": 14342 + }, + { + "epoch": 6.01, + "learning_rate": 3.084648318617577e-06, + "loss": 0.0686, + "step": 14343 + }, + { + "epoch": 6.01, + "learning_rate": 3.0834237516252817e-06, + "loss": 0.0737, + "step": 14344 + }, + { + "epoch": 6.01, + "learning_rate": 3.0821993834438337e-06, + "loss": 0.1265, + "step": 14345 + }, + { + "epoch": 6.01, + "learning_rate": 3.0809752141084248e-06, + "loss": 0.0865, + "step": 14346 + }, + { + "epoch": 6.01, + "learning_rate": 3.079751243654248e-06, + "loss": 0.0877, + "step": 14347 + }, + { + "epoch": 6.01, + "learning_rate": 3.078527472116477e-06, + "loss": 0.0758, + "step": 14348 + }, + { + "epoch": 6.01, + "learning_rate": 3.0773038995302918e-06, + "loss": 0.0773, + "step": 14349 + }, + { + "epoch": 6.01, + "learning_rate": 3.0760805259308626e-06, + "loss": 0.0597, + "step": 14350 + }, + { + "epoch": 6.01, + "learning_rate": 3.0748573513533554e-06, + "loss": 0.1065, + "step": 14351 + }, + { + "epoch": 6.01, + "learning_rate": 3.0736343758329277e-06, + "loss": 0.0947, + "step": 14352 + }, + { + "epoch": 6.01, + "learning_rate": 3.0724115994047367e-06, + "loss": 0.0626, + "step": 14353 + }, + { + "epoch": 6.01, + "learning_rate": 3.071189022103923e-06, + "loss": 0.0927, + "step": 14354 + }, + { + "epoch": 6.01, + "learning_rate": 3.069966643965634e-06, + "loss": 0.077, + "step": 14355 + }, + { + "epoch": 6.01, + "learning_rate": 3.068744465025003e-06, + "loss": 0.1038, + "step": 14356 + }, + { + "epoch": 6.01, + "learning_rate": 3.0675224853171613e-06, + "loss": 0.0589, + "step": 14357 + }, + { + "epoch": 6.01, + "learning_rate": 3.0663007048772354e-06, + "loss": 0.0928, + "step": 14358 + }, + { + "epoch": 6.01, + "learning_rate": 3.0650791237403456e-06, + "loss": 0.0821, + "step": 14359 + }, + { + "epoch": 6.01, + "learning_rate": 3.063857741941598e-06, + "loss": 0.066, + "step": 14360 + }, + { + "epoch": 6.01, + "learning_rate": 3.06263655951611e-06, + "loss": 0.0888, + "step": 14361 + }, + { + "epoch": 6.01, + "learning_rate": 3.061415576498973e-06, + "loss": 0.0679, + "step": 14362 + }, + { + "epoch": 6.01, + "learning_rate": 3.0601947929252883e-06, + "loss": 0.0956, + "step": 14363 + }, + { + "epoch": 6.01, + "learning_rate": 3.058974208830147e-06, + "loss": 0.0821, + "step": 14364 + }, + { + "epoch": 6.01, + "learning_rate": 3.0577538242486317e-06, + "loss": 0.0883, + "step": 14365 + }, + { + "epoch": 6.01, + "learning_rate": 3.0565336392158228e-06, + "loss": 0.0893, + "step": 14366 + }, + { + "epoch": 6.02, + "learning_rate": 3.055313653766796e-06, + "loss": 0.0802, + "step": 14367 + }, + { + "epoch": 6.02, + "learning_rate": 3.0540938679366126e-06, + "loss": 0.0919, + "step": 14368 + }, + { + "epoch": 6.02, + "learning_rate": 3.0528742817603364e-06, + "loss": 0.0956, + "step": 14369 + }, + { + "epoch": 6.02, + "learning_rate": 3.0516548952730253e-06, + "loss": 0.0819, + "step": 14370 + }, + { + "epoch": 6.02, + "learning_rate": 3.050435708509728e-06, + "loss": 0.0625, + "step": 14371 + }, + { + "epoch": 6.02, + "learning_rate": 3.049216721505489e-06, + "loss": 0.0944, + "step": 14372 + }, + { + "epoch": 6.02, + "learning_rate": 3.0479979342953504e-06, + "loss": 0.0734, + "step": 14373 + }, + { + "epoch": 6.02, + "learning_rate": 3.046779346914339e-06, + "loss": 0.0737, + "step": 14374 + }, + { + "epoch": 6.02, + "learning_rate": 3.0455609593974855e-06, + "loss": 0.0908, + "step": 14375 + }, + { + "epoch": 6.02, + "learning_rate": 3.044342771779811e-06, + "loss": 0.0815, + "step": 14376 + }, + { + "epoch": 6.02, + "learning_rate": 3.0431247840963316e-06, + "loss": 0.1037, + "step": 14377 + }, + { + "epoch": 6.02, + "learning_rate": 3.041906996382056e-06, + "loss": 0.0726, + "step": 14378 + }, + { + "epoch": 6.02, + "learning_rate": 3.0406894086719894e-06, + "loss": 0.1095, + "step": 14379 + }, + { + "epoch": 6.02, + "learning_rate": 3.039472021001134e-06, + "loss": 0.0771, + "step": 14380 + }, + { + "epoch": 6.02, + "learning_rate": 3.0382548334044783e-06, + "loss": 0.0466, + "step": 14381 + }, + { + "epoch": 6.02, + "learning_rate": 3.0370378459170058e-06, + "loss": 0.0773, + "step": 14382 + }, + { + "epoch": 6.02, + "learning_rate": 3.0358210585737024e-06, + "loss": 0.0787, + "step": 14383 + }, + { + "epoch": 6.02, + "learning_rate": 3.0346044714095425e-06, + "loss": 0.0997, + "step": 14384 + }, + { + "epoch": 6.02, + "learning_rate": 3.0333880844594963e-06, + "loss": 0.0779, + "step": 14385 + }, + { + "epoch": 6.02, + "learning_rate": 3.032171897758527e-06, + "loss": 0.0939, + "step": 14386 + }, + { + "epoch": 6.02, + "learning_rate": 3.030955911341594e-06, + "loss": 0.1008, + "step": 14387 + }, + { + "epoch": 6.02, + "learning_rate": 3.029740125243652e-06, + "loss": 0.0828, + "step": 14388 + }, + { + "epoch": 6.02, + "learning_rate": 3.028524539499642e-06, + "loss": 0.1049, + "step": 14389 + }, + { + "epoch": 6.02, + "learning_rate": 3.0273091541445065e-06, + "loss": 0.0757, + "step": 14390 + }, + { + "epoch": 6.03, + "learning_rate": 3.026093969213183e-06, + "loss": 0.0831, + "step": 14391 + }, + { + "epoch": 6.03, + "learning_rate": 3.0248789847406003e-06, + "loss": 0.0728, + "step": 14392 + }, + { + "epoch": 6.03, + "learning_rate": 3.02366420076168e-06, + "loss": 0.0453, + "step": 14393 + }, + { + "epoch": 6.03, + "learning_rate": 3.022449617311346e-06, + "loss": 0.089, + "step": 14394 + }, + { + "epoch": 6.03, + "learning_rate": 3.021235234424502e-06, + "loss": 0.0904, + "step": 14395 + }, + { + "epoch": 6.03, + "learning_rate": 3.0200210521360598e-06, + "loss": 0.0671, + "step": 14396 + }, + { + "epoch": 6.03, + "learning_rate": 3.0188070704809167e-06, + "loss": 0.1163, + "step": 14397 + }, + { + "epoch": 6.03, + "learning_rate": 3.017593289493971e-06, + "loss": 0.0829, + "step": 14398 + }, + { + "epoch": 6.03, + "learning_rate": 3.0163797092101087e-06, + "loss": 0.0714, + "step": 14399 + }, + { + "epoch": 6.03, + "learning_rate": 3.0151663296642188e-06, + "loss": 0.09, + "step": 14400 + }, + { + "epoch": 6.03, + "learning_rate": 3.01395315089117e-06, + "loss": 0.0753, + "step": 14401 + }, + { + "epoch": 6.03, + "learning_rate": 3.01274017292584e-06, + "loss": 0.0752, + "step": 14402 + }, + { + "epoch": 6.03, + "learning_rate": 3.011527395803092e-06, + "loss": 0.0562, + "step": 14403 + }, + { + "epoch": 6.03, + "learning_rate": 3.0103148195577914e-06, + "loss": 0.0999, + "step": 14404 + }, + { + "epoch": 6.03, + "learning_rate": 3.009102444224786e-06, + "loss": 0.0673, + "step": 14405 + }, + { + "epoch": 6.03, + "learning_rate": 3.0078902698389266e-06, + "loss": 0.0899, + "step": 14406 + }, + { + "epoch": 6.03, + "learning_rate": 3.0066782964350573e-06, + "loss": 0.1033, + "step": 14407 + }, + { + "epoch": 6.03, + "learning_rate": 3.0054665240480174e-06, + "loss": 0.0715, + "step": 14408 + }, + { + "epoch": 6.03, + "learning_rate": 3.004254952712632e-06, + "loss": 0.0692, + "step": 14409 + }, + { + "epoch": 6.03, + "learning_rate": 3.0030435824637304e-06, + "loss": 0.0857, + "step": 14410 + }, + { + "epoch": 6.03, + "learning_rate": 3.0018324133361323e-06, + "loss": 0.0605, + "step": 14411 + }, + { + "epoch": 6.03, + "learning_rate": 3.000621445364651e-06, + "loss": 0.0519, + "step": 14412 + }, + { + "epoch": 6.03, + "learning_rate": 2.999410678584096e-06, + "loss": 0.0537, + "step": 14413 + }, + { + "epoch": 6.03, + "learning_rate": 2.9982001130292703e-06, + "loss": 0.0751, + "step": 14414 + }, + { + "epoch": 6.04, + "learning_rate": 2.9969897487349675e-06, + "loss": 0.083, + "step": 14415 + }, + { + "epoch": 6.04, + "learning_rate": 2.9957795857359794e-06, + "loss": 0.0699, + "step": 14416 + }, + { + "epoch": 6.04, + "learning_rate": 2.9945696240670905e-06, + "loss": 0.1112, + "step": 14417 + }, + { + "epoch": 6.04, + "learning_rate": 2.9933598637630833e-06, + "loss": 0.0777, + "step": 14418 + }, + { + "epoch": 6.04, + "learning_rate": 2.9921503048587284e-06, + "loss": 0.0872, + "step": 14419 + }, + { + "epoch": 6.04, + "learning_rate": 2.990940947388794e-06, + "loss": 0.1089, + "step": 14420 + }, + { + "epoch": 6.04, + "learning_rate": 2.9897317913880454e-06, + "loss": 0.0951, + "step": 14421 + }, + { + "epoch": 6.04, + "learning_rate": 2.9885228368912333e-06, + "loss": 0.052, + "step": 14422 + }, + { + "epoch": 6.04, + "learning_rate": 2.9873140839331094e-06, + "loss": 0.0862, + "step": 14423 + }, + { + "epoch": 6.04, + "learning_rate": 2.98610553254842e-06, + "loss": 0.0954, + "step": 14424 + }, + { + "epoch": 6.04, + "learning_rate": 2.984897182771903e-06, + "loss": 0.061, + "step": 14425 + }, + { + "epoch": 6.04, + "learning_rate": 2.9836890346382907e-06, + "loss": 0.0772, + "step": 14426 + }, + { + "epoch": 6.04, + "learning_rate": 2.982481088182315e-06, + "loss": 0.0885, + "step": 14427 + }, + { + "epoch": 6.04, + "learning_rate": 2.98127334343869e-06, + "loss": 0.1007, + "step": 14428 + }, + { + "epoch": 6.04, + "learning_rate": 2.980065800442137e-06, + "loss": 0.046, + "step": 14429 + }, + { + "epoch": 6.04, + "learning_rate": 2.9788584592273595e-06, + "loss": 0.068, + "step": 14430 + }, + { + "epoch": 6.04, + "learning_rate": 2.9776513198290657e-06, + "loss": 0.0835, + "step": 14431 + }, + { + "epoch": 6.04, + "learning_rate": 2.9764443822819534e-06, + "loss": 0.0601, + "step": 14432 + }, + { + "epoch": 6.04, + "learning_rate": 2.9752376466207154e-06, + "loss": 0.0832, + "step": 14433 + }, + { + "epoch": 6.04, + "learning_rate": 2.974031112880037e-06, + "loss": 0.1016, + "step": 14434 + }, + { + "epoch": 6.04, + "learning_rate": 2.972824781094602e-06, + "loss": 0.0758, + "step": 14435 + }, + { + "epoch": 6.04, + "learning_rate": 2.9716186512990806e-06, + "loss": 0.062, + "step": 14436 + }, + { + "epoch": 6.04, + "learning_rate": 2.9704127235281433e-06, + "loss": 0.0975, + "step": 14437 + }, + { + "epoch": 6.04, + "learning_rate": 2.9692069978164552e-06, + "loss": 0.0696, + "step": 14438 + }, + { + "epoch": 6.05, + "learning_rate": 2.968001474198673e-06, + "loss": 0.0852, + "step": 14439 + }, + { + "epoch": 6.05, + "learning_rate": 2.966796152709447e-06, + "loss": 0.1034, + "step": 14440 + }, + { + "epoch": 6.05, + "learning_rate": 2.9655910333834294e-06, + "loss": 0.0885, + "step": 14441 + }, + { + "epoch": 6.05, + "learning_rate": 2.9643861162552512e-06, + "loss": 0.0735, + "step": 14442 + }, + { + "epoch": 6.05, + "learning_rate": 2.9631814013595507e-06, + "loss": 0.114, + "step": 14443 + }, + { + "epoch": 6.05, + "learning_rate": 2.961976888730956e-06, + "loss": 0.0802, + "step": 14444 + }, + { + "epoch": 6.05, + "learning_rate": 2.960772578404091e-06, + "loss": 0.0662, + "step": 14445 + }, + { + "epoch": 6.05, + "learning_rate": 2.959568470413572e-06, + "loss": 0.0662, + "step": 14446 + }, + { + "epoch": 6.05, + "learning_rate": 2.9583645647940097e-06, + "loss": 0.0978, + "step": 14447 + }, + { + "epoch": 6.05, + "learning_rate": 2.9571608615800118e-06, + "loss": 0.0529, + "step": 14448 + }, + { + "epoch": 6.05, + "learning_rate": 2.9559573608061755e-06, + "loss": 0.0754, + "step": 14449 + }, + { + "epoch": 6.05, + "learning_rate": 2.9547540625070903e-06, + "loss": 0.0878, + "step": 14450 + }, + { + "epoch": 6.05, + "learning_rate": 2.953550966717349e-06, + "loss": 0.0838, + "step": 14451 + }, + { + "epoch": 6.05, + "learning_rate": 2.9523480734715317e-06, + "loss": 0.0953, + "step": 14452 + }, + { + "epoch": 6.05, + "learning_rate": 2.9511453828042147e-06, + "loss": 0.1005, + "step": 14453 + }, + { + "epoch": 6.05, + "learning_rate": 2.9499428947499696e-06, + "loss": 0.0766, + "step": 14454 + }, + { + "epoch": 6.05, + "learning_rate": 2.9487406093433622e-06, + "loss": 0.0949, + "step": 14455 + }, + { + "epoch": 6.05, + "learning_rate": 2.947538526618946e-06, + "loss": 0.0926, + "step": 14456 + }, + { + "epoch": 6.05, + "learning_rate": 2.9463366466112764e-06, + "loss": 0.1011, + "step": 14457 + }, + { + "epoch": 6.05, + "learning_rate": 2.9451349693549e-06, + "loss": 0.0735, + "step": 14458 + }, + { + "epoch": 6.05, + "learning_rate": 2.9439334948843588e-06, + "loss": 0.1279, + "step": 14459 + }, + { + "epoch": 6.05, + "learning_rate": 2.9427322232341893e-06, + "loss": 0.067, + "step": 14460 + }, + { + "epoch": 6.05, + "learning_rate": 2.941531154438918e-06, + "loss": 0.064, + "step": 14461 + }, + { + "epoch": 6.05, + "learning_rate": 2.940330288533074e-06, + "loss": 0.0995, + "step": 14462 + }, + { + "epoch": 6.06, + "learning_rate": 2.9391296255511683e-06, + "loss": 0.1056, + "step": 14463 + }, + { + "epoch": 6.06, + "learning_rate": 2.9379291655277143e-06, + "loss": 0.0957, + "step": 14464 + }, + { + "epoch": 6.06, + "learning_rate": 2.9367289084972206e-06, + "loss": 0.0597, + "step": 14465 + }, + { + "epoch": 6.06, + "learning_rate": 2.9355288544941873e-06, + "loss": 0.0803, + "step": 14466 + }, + { + "epoch": 6.06, + "learning_rate": 2.934329003553107e-06, + "loss": 0.0615, + "step": 14467 + }, + { + "epoch": 6.06, + "learning_rate": 2.933129355708473e-06, + "loss": 0.0744, + "step": 14468 + }, + { + "epoch": 6.06, + "learning_rate": 2.931929910994762e-06, + "loss": 0.0773, + "step": 14469 + }, + { + "epoch": 6.06, + "learning_rate": 2.930730669446453e-06, + "loss": 0.0954, + "step": 14470 + }, + { + "epoch": 6.06, + "learning_rate": 2.9295316310980184e-06, + "loss": 0.0626, + "step": 14471 + }, + { + "epoch": 6.06, + "learning_rate": 2.928332795983926e-06, + "loss": 0.0806, + "step": 14472 + }, + { + "epoch": 6.06, + "learning_rate": 2.9271341641386285e-06, + "loss": 0.0626, + "step": 14473 + }, + { + "epoch": 6.06, + "learning_rate": 2.925935735596582e-06, + "loss": 0.0804, + "step": 14474 + }, + { + "epoch": 6.06, + "learning_rate": 2.9247375103922373e-06, + "loss": 0.0862, + "step": 14475 + }, + { + "epoch": 6.06, + "learning_rate": 2.9235394885600367e-06, + "loss": 0.0685, + "step": 14476 + }, + { + "epoch": 6.06, + "learning_rate": 2.922341670134411e-06, + "loss": 0.1191, + "step": 14477 + }, + { + "epoch": 6.06, + "learning_rate": 2.921144055149793e-06, + "loss": 0.0693, + "step": 14478 + }, + { + "epoch": 6.06, + "learning_rate": 2.9199466436406077e-06, + "loss": 0.0933, + "step": 14479 + }, + { + "epoch": 6.06, + "learning_rate": 2.918749435641274e-06, + "loss": 0.1152, + "step": 14480 + }, + { + "epoch": 6.06, + "learning_rate": 2.9175524311862034e-06, + "loss": 0.103, + "step": 14481 + }, + { + "epoch": 6.06, + "learning_rate": 2.916355630309806e-06, + "loss": 0.0681, + "step": 14482 + }, + { + "epoch": 6.06, + "learning_rate": 2.9151590330464775e-06, + "loss": 0.084, + "step": 14483 + }, + { + "epoch": 6.06, + "learning_rate": 2.913962639430615e-06, + "loss": 0.0755, + "step": 14484 + }, + { + "epoch": 6.06, + "learning_rate": 2.91276644949661e-06, + "loss": 0.0906, + "step": 14485 + }, + { + "epoch": 6.06, + "learning_rate": 2.911570463278843e-06, + "loss": 0.0835, + "step": 14486 + }, + { + "epoch": 6.07, + "learning_rate": 2.910374680811693e-06, + "loss": 0.0773, + "step": 14487 + }, + { + "epoch": 6.07, + "learning_rate": 2.909179102129531e-06, + "loss": 0.0837, + "step": 14488 + }, + { + "epoch": 6.07, + "learning_rate": 2.9079837272667276e-06, + "loss": 0.0662, + "step": 14489 + }, + { + "epoch": 6.07, + "learning_rate": 2.906788556257636e-06, + "loss": 0.077, + "step": 14490 + }, + { + "epoch": 6.07, + "learning_rate": 2.9055935891366117e-06, + "loss": 0.089, + "step": 14491 + }, + { + "epoch": 6.07, + "learning_rate": 2.904398825938005e-06, + "loss": 0.0903, + "step": 14492 + }, + { + "epoch": 6.07, + "learning_rate": 2.903204266696157e-06, + "loss": 0.0882, + "step": 14493 + }, + { + "epoch": 6.07, + "learning_rate": 2.9020099114454046e-06, + "loss": 0.0891, + "step": 14494 + }, + { + "epoch": 6.07, + "learning_rate": 2.9008157602200827e-06, + "loss": 0.0848, + "step": 14495 + }, + { + "epoch": 6.07, + "learning_rate": 2.8996218130545108e-06, + "loss": 0.1225, + "step": 14496 + }, + { + "epoch": 6.07, + "learning_rate": 2.898428069983007e-06, + "loss": 0.0876, + "step": 14497 + }, + { + "epoch": 6.07, + "learning_rate": 2.8972345310398863e-06, + "loss": 0.0819, + "step": 14498 + }, + { + "epoch": 6.07, + "learning_rate": 2.8960411962594558e-06, + "loss": 0.0752, + "step": 14499 + }, + { + "epoch": 6.07, + "learning_rate": 2.894848065676017e-06, + "loss": 0.0973, + "step": 14500 + }, + { + "epoch": 6.07, + "learning_rate": 2.8936551393238664e-06, + "loss": 0.0801, + "step": 14501 + }, + { + "epoch": 6.07, + "learning_rate": 2.892462417237292e-06, + "loss": 0.0798, + "step": 14502 + }, + { + "epoch": 6.07, + "learning_rate": 2.8912698994505816e-06, + "loss": 0.0809, + "step": 14503 + }, + { + "epoch": 6.07, + "learning_rate": 2.890077585998007e-06, + "loss": 0.0906, + "step": 14504 + }, + { + "epoch": 6.07, + "learning_rate": 2.888885476913842e-06, + "loss": 0.0601, + "step": 14505 + }, + { + "epoch": 6.07, + "learning_rate": 2.887693572232355e-06, + "loss": 0.0999, + "step": 14506 + }, + { + "epoch": 6.07, + "learning_rate": 2.8865018719878034e-06, + "loss": 0.0891, + "step": 14507 + }, + { + "epoch": 6.07, + "learning_rate": 2.8853103762144443e-06, + "loss": 0.0517, + "step": 14508 + }, + { + "epoch": 6.07, + "learning_rate": 2.8841190849465282e-06, + "loss": 0.0721, + "step": 14509 + }, + { + "epoch": 6.07, + "learning_rate": 2.882927998218291e-06, + "loss": 0.0974, + "step": 14510 + }, + { + "epoch": 6.08, + "learning_rate": 2.881737116063973e-06, + "loss": 0.0599, + "step": 14511 + }, + { + "epoch": 6.08, + "learning_rate": 2.880546438517804e-06, + "loss": 0.0839, + "step": 14512 + }, + { + "epoch": 6.08, + "learning_rate": 2.8793559656140114e-06, + "loss": 0.1268, + "step": 14513 + }, + { + "epoch": 6.08, + "learning_rate": 2.878165697386812e-06, + "loss": 0.0802, + "step": 14514 + }, + { + "epoch": 6.08, + "learning_rate": 2.8769756338704235e-06, + "loss": 0.0703, + "step": 14515 + }, + { + "epoch": 6.08, + "learning_rate": 2.875785775099046e-06, + "loss": 0.0828, + "step": 14516 + }, + { + "epoch": 6.08, + "learning_rate": 2.8745961211068873e-06, + "loss": 0.1026, + "step": 14517 + }, + { + "epoch": 6.08, + "learning_rate": 2.8734066719281372e-06, + "loss": 0.0841, + "step": 14518 + }, + { + "epoch": 6.08, + "learning_rate": 2.8722174275969884e-06, + "loss": 0.1171, + "step": 14519 + }, + { + "epoch": 6.08, + "learning_rate": 2.871028388147624e-06, + "loss": 0.1032, + "step": 14520 + }, + { + "epoch": 6.08, + "learning_rate": 2.8698395536142232e-06, + "loss": 0.0571, + "step": 14521 + }, + { + "epoch": 6.08, + "learning_rate": 2.868650924030958e-06, + "loss": 0.084, + "step": 14522 + }, + { + "epoch": 6.08, + "learning_rate": 2.8674624994319954e-06, + "loss": 0.0753, + "step": 14523 + }, + { + "epoch": 6.08, + "learning_rate": 2.866274279851491e-06, + "loss": 0.0792, + "step": 14524 + }, + { + "epoch": 6.08, + "learning_rate": 2.8650862653236034e-06, + "loss": 0.097, + "step": 14525 + }, + { + "epoch": 6.08, + "learning_rate": 2.8638984558824777e-06, + "loss": 0.0608, + "step": 14526 + }, + { + "epoch": 6.08, + "learning_rate": 2.8627108515622603e-06, + "loss": 0.1011, + "step": 14527 + }, + { + "epoch": 6.08, + "learning_rate": 2.861523452397085e-06, + "loss": 0.0821, + "step": 14528 + }, + { + "epoch": 6.08, + "learning_rate": 2.8603362584210838e-06, + "loss": 0.0893, + "step": 14529 + }, + { + "epoch": 6.08, + "learning_rate": 2.8591492696683833e-06, + "loss": 0.0995, + "step": 14530 + }, + { + "epoch": 6.08, + "learning_rate": 2.857962486173098e-06, + "loss": 0.0804, + "step": 14531 + }, + { + "epoch": 6.08, + "learning_rate": 2.8567759079693435e-06, + "loss": 0.101, + "step": 14532 + }, + { + "epoch": 6.08, + "learning_rate": 2.855589535091227e-06, + "loss": 0.0949, + "step": 14533 + }, + { + "epoch": 6.08, + "learning_rate": 2.854403367572849e-06, + "loss": 0.0694, + "step": 14534 + }, + { + "epoch": 6.09, + "learning_rate": 2.8532174054483053e-06, + "loss": 0.0773, + "step": 14535 + }, + { + "epoch": 6.09, + "learning_rate": 2.8520316487516884e-06, + "loss": 0.069, + "step": 14536 + }, + { + "epoch": 6.09, + "learning_rate": 2.850846097517076e-06, + "loss": 0.079, + "step": 14537 + }, + { + "epoch": 6.09, + "learning_rate": 2.849660751778548e-06, + "loss": 0.0623, + "step": 14538 + }, + { + "epoch": 6.09, + "learning_rate": 2.8484756115701773e-06, + "loss": 0.0909, + "step": 14539 + }, + { + "epoch": 6.09, + "learning_rate": 2.847290676926031e-06, + "loss": 0.0821, + "step": 14540 + }, + { + "epoch": 6.09, + "learning_rate": 2.8461059478801657e-06, + "loss": 0.1235, + "step": 14541 + }, + { + "epoch": 6.09, + "learning_rate": 2.8449214244666345e-06, + "loss": 0.0907, + "step": 14542 + }, + { + "epoch": 6.09, + "learning_rate": 2.84373710671949e-06, + "loss": 0.0718, + "step": 14543 + }, + { + "epoch": 6.09, + "learning_rate": 2.8425529946727725e-06, + "loss": 0.0996, + "step": 14544 + }, + { + "epoch": 6.09, + "learning_rate": 2.841369088360516e-06, + "loss": 0.0839, + "step": 14545 + }, + { + "epoch": 6.09, + "learning_rate": 2.8401853878167528e-06, + "loss": 0.1058, + "step": 14546 + }, + { + "epoch": 6.09, + "learning_rate": 2.839001893075507e-06, + "loss": 0.0762, + "step": 14547 + }, + { + "epoch": 6.09, + "learning_rate": 2.8378186041707977e-06, + "loss": 0.0689, + "step": 14548 + }, + { + "epoch": 6.09, + "learning_rate": 2.836635521136636e-06, + "loss": 0.0865, + "step": 14549 + }, + { + "epoch": 6.09, + "learning_rate": 2.835452644007033e-06, + "loss": 0.0619, + "step": 14550 + }, + { + "epoch": 6.09, + "learning_rate": 2.834269972815983e-06, + "loss": 0.0879, + "step": 14551 + }, + { + "epoch": 6.09, + "learning_rate": 2.833087507597484e-06, + "loss": 0.0751, + "step": 14552 + }, + { + "epoch": 6.09, + "learning_rate": 2.831905248385525e-06, + "loss": 0.0685, + "step": 14553 + }, + { + "epoch": 6.09, + "learning_rate": 2.830723195214088e-06, + "loss": 0.0894, + "step": 14554 + }, + { + "epoch": 6.09, + "learning_rate": 2.8295413481171517e-06, + "loss": 0.0732, + "step": 14555 + }, + { + "epoch": 6.09, + "learning_rate": 2.8283597071286897e-06, + "loss": 0.0846, + "step": 14556 + }, + { + "epoch": 6.09, + "learning_rate": 2.82717827228266e-06, + "loss": 0.0812, + "step": 14557 + }, + { + "epoch": 6.1, + "learning_rate": 2.825997043613026e-06, + "loss": 0.1015, + "step": 14558 + }, + { + "epoch": 6.1, + "learning_rate": 2.824816021153741e-06, + "loss": 0.0906, + "step": 14559 + }, + { + "epoch": 6.1, + "learning_rate": 2.8236352049387515e-06, + "loss": 0.0706, + "step": 14560 + }, + { + "epoch": 6.1, + "learning_rate": 2.822454595002001e-06, + "loss": 0.11, + "step": 14561 + }, + { + "epoch": 6.1, + "learning_rate": 2.821274191377423e-06, + "loss": 0.0869, + "step": 14562 + }, + { + "epoch": 6.1, + "learning_rate": 2.8200939940989524e-06, + "loss": 0.0942, + "step": 14563 + }, + { + "epoch": 6.1, + "learning_rate": 2.818914003200508e-06, + "loss": 0.0914, + "step": 14564 + }, + { + "epoch": 6.1, + "learning_rate": 2.817734218716005e-06, + "loss": 0.0794, + "step": 14565 + }, + { + "epoch": 6.1, + "learning_rate": 2.8165546406793586e-06, + "loss": 0.0576, + "step": 14566 + }, + { + "epoch": 6.1, + "learning_rate": 2.815375269124474e-06, + "loss": 0.0971, + "step": 14567 + }, + { + "epoch": 6.1, + "learning_rate": 2.814196104085253e-06, + "loss": 0.0714, + "step": 14568 + }, + { + "epoch": 6.1, + "learning_rate": 2.813017145595588e-06, + "loss": 0.0884, + "step": 14569 + }, + { + "epoch": 6.1, + "learning_rate": 2.811838393689367e-06, + "loss": 0.0818, + "step": 14570 + }, + { + "epoch": 6.1, + "learning_rate": 2.8106598484004756e-06, + "loss": 0.1187, + "step": 14571 + }, + { + "epoch": 6.1, + "learning_rate": 2.8094815097627846e-06, + "loss": 0.0948, + "step": 14572 + }, + { + "epoch": 6.1, + "learning_rate": 2.8083033778101664e-06, + "loss": 0.0907, + "step": 14573 + }, + { + "epoch": 6.1, + "learning_rate": 2.807125452576487e-06, + "loss": 0.1035, + "step": 14574 + }, + { + "epoch": 6.1, + "learning_rate": 2.805947734095602e-06, + "loss": 0.0851, + "step": 14575 + }, + { + "epoch": 6.1, + "learning_rate": 2.804770222401366e-06, + "loss": 0.1148, + "step": 14576 + }, + { + "epoch": 6.1, + "learning_rate": 2.803592917527629e-06, + "loss": 0.0831, + "step": 14577 + }, + { + "epoch": 6.1, + "learning_rate": 2.802415819508223e-06, + "loss": 0.0842, + "step": 14578 + }, + { + "epoch": 6.1, + "learning_rate": 2.8012389283769883e-06, + "loss": 0.0846, + "step": 14579 + }, + { + "epoch": 6.1, + "learning_rate": 2.8000622441677527e-06, + "loss": 0.0884, + "step": 14580 + }, + { + "epoch": 6.1, + "learning_rate": 2.798885766914339e-06, + "loss": 0.0936, + "step": 14581 + }, + { + "epoch": 6.11, + "learning_rate": 2.7977094966505624e-06, + "loss": 0.081, + "step": 14582 + }, + { + "epoch": 6.11, + "learning_rate": 2.796533433410239e-06, + "loss": 0.0873, + "step": 14583 + }, + { + "epoch": 6.11, + "learning_rate": 2.7953575772271667e-06, + "loss": 0.0961, + "step": 14584 + }, + { + "epoch": 6.11, + "learning_rate": 2.7941819281351513e-06, + "loss": 0.0902, + "step": 14585 + }, + { + "epoch": 6.11, + "learning_rate": 2.7930064861679783e-06, + "loss": 0.0795, + "step": 14586 + }, + { + "epoch": 6.11, + "learning_rate": 2.791831251359438e-06, + "loss": 0.0775, + "step": 14587 + }, + { + "epoch": 6.11, + "learning_rate": 2.790656223743312e-06, + "loss": 0.0974, + "step": 14588 + }, + { + "epoch": 6.11, + "learning_rate": 2.7894814033533768e-06, + "loss": 0.0759, + "step": 14589 + }, + { + "epoch": 6.11, + "learning_rate": 2.7883067902233985e-06, + "loss": 0.1007, + "step": 14590 + }, + { + "epoch": 6.11, + "learning_rate": 2.7871323843871467e-06, + "loss": 0.0767, + "step": 14591 + }, + { + "epoch": 6.11, + "learning_rate": 2.7859581858783693e-06, + "loss": 0.0675, + "step": 14592 + }, + { + "epoch": 6.11, + "learning_rate": 2.784784194730823e-06, + "loss": 0.1, + "step": 14593 + }, + { + "epoch": 6.11, + "learning_rate": 2.7836104109782525e-06, + "loss": 0.1084, + "step": 14594 + }, + { + "epoch": 6.11, + "learning_rate": 2.7824368346543963e-06, + "loss": 0.0986, + "step": 14595 + }, + { + "epoch": 6.11, + "learning_rate": 2.781263465792989e-06, + "loss": 0.078, + "step": 14596 + }, + { + "epoch": 6.11, + "learning_rate": 2.7800903044277604e-06, + "loss": 0.08, + "step": 14597 + }, + { + "epoch": 6.11, + "learning_rate": 2.7789173505924274e-06, + "loss": 0.0671, + "step": 14598 + }, + { + "epoch": 6.11, + "learning_rate": 2.777744604320706e-06, + "loss": 0.0647, + "step": 14599 + }, + { + "epoch": 6.11, + "learning_rate": 2.776572065646307e-06, + "loss": 0.0704, + "step": 14600 + }, + { + "epoch": 6.11, + "learning_rate": 2.775399734602934e-06, + "loss": 0.1009, + "step": 14601 + }, + { + "epoch": 6.11, + "learning_rate": 2.774227611224285e-06, + "loss": 0.1078, + "step": 14602 + }, + { + "epoch": 6.11, + "learning_rate": 2.7730556955440515e-06, + "loss": 0.0757, + "step": 14603 + }, + { + "epoch": 6.11, + "learning_rate": 2.7718839875959213e-06, + "loss": 0.0806, + "step": 14604 + }, + { + "epoch": 6.11, + "learning_rate": 2.7707124874135694e-06, + "loss": 0.0631, + "step": 14605 + }, + { + "epoch": 6.12, + "learning_rate": 2.7695411950306714e-06, + "loss": 0.1014, + "step": 14606 + }, + { + "epoch": 6.12, + "learning_rate": 2.7683701104808956e-06, + "loss": 0.0764, + "step": 14607 + }, + { + "epoch": 6.12, + "learning_rate": 2.767199233797907e-06, + "loss": 0.1016, + "step": 14608 + }, + { + "epoch": 6.12, + "learning_rate": 2.766028565015355e-06, + "loss": 0.0935, + "step": 14609 + }, + { + "epoch": 6.12, + "learning_rate": 2.7648581041668922e-06, + "loss": 0.0877, + "step": 14610 + }, + { + "epoch": 6.12, + "learning_rate": 2.7636878512861633e-06, + "loss": 0.0711, + "step": 14611 + }, + { + "epoch": 6.12, + "learning_rate": 2.7625178064068094e-06, + "loss": 0.0836, + "step": 14612 + }, + { + "epoch": 6.12, + "learning_rate": 2.7613479695624544e-06, + "loss": 0.0938, + "step": 14613 + }, + { + "epoch": 6.12, + "learning_rate": 2.760178340786729e-06, + "loss": 0.0893, + "step": 14614 + }, + { + "epoch": 6.12, + "learning_rate": 2.759008920113253e-06, + "loss": 0.096, + "step": 14615 + }, + { + "epoch": 6.12, + "learning_rate": 2.7578397075756404e-06, + "loss": 0.0646, + "step": 14616 + }, + { + "epoch": 6.12, + "learning_rate": 2.756670703207498e-06, + "loss": 0.0842, + "step": 14617 + }, + { + "epoch": 6.12, + "learning_rate": 2.7555019070424316e-06, + "loss": 0.0547, + "step": 14618 + }, + { + "epoch": 6.12, + "learning_rate": 2.754333319114032e-06, + "loss": 0.0824, + "step": 14619 + }, + { + "epoch": 6.12, + "learning_rate": 2.753164939455891e-06, + "loss": 0.0972, + "step": 14620 + }, + { + "epoch": 6.12, + "learning_rate": 2.7519967681015933e-06, + "loss": 0.0783, + "step": 14621 + }, + { + "epoch": 6.12, + "learning_rate": 2.7508288050847166e-06, + "loss": 0.1054, + "step": 14622 + }, + { + "epoch": 6.12, + "learning_rate": 2.7496610504388334e-06, + "loss": 0.0799, + "step": 14623 + }, + { + "epoch": 6.12, + "learning_rate": 2.7484935041975113e-06, + "loss": 0.0862, + "step": 14624 + }, + { + "epoch": 6.12, + "learning_rate": 2.7473261663943063e-06, + "loss": 0.1045, + "step": 14625 + }, + { + "epoch": 6.12, + "learning_rate": 2.7461590370627753e-06, + "loss": 0.0807, + "step": 14626 + }, + { + "epoch": 6.12, + "learning_rate": 2.7449921162364647e-06, + "loss": 0.0825, + "step": 14627 + }, + { + "epoch": 6.12, + "learning_rate": 2.7438254039489185e-06, + "loss": 0.0802, + "step": 14628 + }, + { + "epoch": 6.12, + "learning_rate": 2.7426589002336723e-06, + "loss": 0.0778, + "step": 14629 + }, + { + "epoch": 6.13, + "learning_rate": 2.741492605124255e-06, + "loss": 0.0931, + "step": 14630 + }, + { + "epoch": 6.13, + "learning_rate": 2.7403265186541963e-06, + "loss": 0.0809, + "step": 14631 + }, + { + "epoch": 6.13, + "learning_rate": 2.7391606408570084e-06, + "loss": 0.0576, + "step": 14632 + }, + { + "epoch": 6.13, + "learning_rate": 2.7379949717662014e-06, + "loss": 0.0951, + "step": 14633 + }, + { + "epoch": 6.13, + "learning_rate": 2.7368295114152852e-06, + "loss": 0.1012, + "step": 14634 + }, + { + "epoch": 6.13, + "learning_rate": 2.7356642598377604e-06, + "loss": 0.0901, + "step": 14635 + }, + { + "epoch": 6.13, + "learning_rate": 2.734499217067119e-06, + "loss": 0.0891, + "step": 14636 + }, + { + "epoch": 6.13, + "learning_rate": 2.733334383136851e-06, + "loss": 0.0668, + "step": 14637 + }, + { + "epoch": 6.13, + "learning_rate": 2.7321697580804407e-06, + "loss": 0.0831, + "step": 14638 + }, + { + "epoch": 6.13, + "learning_rate": 2.7310053419313576e-06, + "loss": 0.0706, + "step": 14639 + }, + { + "epoch": 6.13, + "learning_rate": 2.729841134723077e-06, + "loss": 0.0909, + "step": 14640 + }, + { + "epoch": 6.13, + "learning_rate": 2.7286771364890606e-06, + "loss": 0.0715, + "step": 14641 + }, + { + "epoch": 6.13, + "learning_rate": 2.727513347262767e-06, + "loss": 0.0648, + "step": 14642 + }, + { + "epoch": 6.13, + "learning_rate": 2.7263497670776495e-06, + "loss": 0.103, + "step": 14643 + }, + { + "epoch": 6.13, + "learning_rate": 2.7251863959671533e-06, + "loss": 0.0876, + "step": 14644 + }, + { + "epoch": 6.13, + "learning_rate": 2.724023233964721e-06, + "loss": 0.0984, + "step": 14645 + }, + { + "epoch": 6.13, + "learning_rate": 2.722860281103782e-06, + "loss": 0.0727, + "step": 14646 + }, + { + "epoch": 6.13, + "learning_rate": 2.721697537417767e-06, + "loss": 0.1052, + "step": 14647 + }, + { + "epoch": 6.13, + "learning_rate": 2.7205350029400966e-06, + "loss": 0.0902, + "step": 14648 + }, + { + "epoch": 6.13, + "learning_rate": 2.7193726777041885e-06, + "loss": 0.0884, + "step": 14649 + }, + { + "epoch": 6.13, + "learning_rate": 2.7182105617434516e-06, + "loss": 0.1211, + "step": 14650 + }, + { + "epoch": 6.13, + "learning_rate": 2.717048655091293e-06, + "loss": 0.0977, + "step": 14651 + }, + { + "epoch": 6.13, + "learning_rate": 2.7158869577811064e-06, + "loss": 0.0991, + "step": 14652 + }, + { + "epoch": 6.13, + "learning_rate": 2.7147254698462875e-06, + "loss": 0.0861, + "step": 14653 + }, + { + "epoch": 6.14, + "learning_rate": 2.713564191320217e-06, + "loss": 0.0808, + "step": 14654 + }, + { + "epoch": 6.14, + "learning_rate": 2.712403122236279e-06, + "loss": 0.0791, + "step": 14655 + }, + { + "epoch": 6.14, + "learning_rate": 2.7112422626278456e-06, + "loss": 0.0886, + "step": 14656 + }, + { + "epoch": 6.14, + "learning_rate": 2.7100816125282846e-06, + "loss": 0.0898, + "step": 14657 + }, + { + "epoch": 6.14, + "learning_rate": 2.7089211719709596e-06, + "loss": 0.0485, + "step": 14658 + }, + { + "epoch": 6.14, + "learning_rate": 2.707760940989229e-06, + "loss": 0.0637, + "step": 14659 + }, + { + "epoch": 6.14, + "learning_rate": 2.7066009196164355e-06, + "loss": 0.0994, + "step": 14660 + }, + { + "epoch": 6.14, + "learning_rate": 2.7054411078859267e-06, + "loss": 0.1317, + "step": 14661 + }, + { + "epoch": 6.14, + "learning_rate": 2.704281505831039e-06, + "loss": 0.0502, + "step": 14662 + }, + { + "epoch": 6.14, + "learning_rate": 2.7031221134851073e-06, + "loss": 0.0918, + "step": 14663 + }, + { + "epoch": 6.14, + "learning_rate": 2.7019629308814544e-06, + "loss": 0.081, + "step": 14664 + }, + { + "epoch": 6.14, + "learning_rate": 2.700803958053403e-06, + "loss": 0.1076, + "step": 14665 + }, + { + "epoch": 6.14, + "learning_rate": 2.699645195034263e-06, + "loss": 0.0651, + "step": 14666 + }, + { + "epoch": 6.14, + "learning_rate": 2.698486641857343e-06, + "loss": 0.0671, + "step": 14667 + }, + { + "epoch": 6.14, + "learning_rate": 2.6973282985559458e-06, + "loss": 0.1349, + "step": 14668 + }, + { + "epoch": 6.14, + "learning_rate": 2.6961701651633653e-06, + "loss": 0.0904, + "step": 14669 + }, + { + "epoch": 6.14, + "learning_rate": 2.6950122417128933e-06, + "loss": 0.0759, + "step": 14670 + }, + { + "epoch": 6.14, + "learning_rate": 2.6938545282378117e-06, + "loss": 0.0651, + "step": 14671 + }, + { + "epoch": 6.14, + "learning_rate": 2.6926970247714e-06, + "loss": 0.0691, + "step": 14672 + }, + { + "epoch": 6.14, + "learning_rate": 2.691539731346927e-06, + "loss": 0.1042, + "step": 14673 + }, + { + "epoch": 6.14, + "learning_rate": 2.6903826479976582e-06, + "loss": 0.0805, + "step": 14674 + }, + { + "epoch": 6.14, + "learning_rate": 2.689225774756853e-06, + "loss": 0.0479, + "step": 14675 + }, + { + "epoch": 6.14, + "learning_rate": 2.6880691116577694e-06, + "loss": 0.1131, + "step": 14676 + }, + { + "epoch": 6.14, + "learning_rate": 2.686912658733647e-06, + "loss": 0.1021, + "step": 14677 + }, + { + "epoch": 6.15, + "learning_rate": 2.685756416017732e-06, + "loss": 0.1051, + "step": 14678 + }, + { + "epoch": 6.15, + "learning_rate": 2.6846003835432597e-06, + "loss": 0.0603, + "step": 14679 + }, + { + "epoch": 6.15, + "learning_rate": 2.6834445613434555e-06, + "loss": 0.0844, + "step": 14680 + }, + { + "epoch": 6.15, + "learning_rate": 2.6822889494515445e-06, + "loss": 0.109, + "step": 14681 + }, + { + "epoch": 6.15, + "learning_rate": 2.681133547900745e-06, + "loss": 0.1006, + "step": 14682 + }, + { + "epoch": 6.15, + "learning_rate": 2.6799783567242653e-06, + "loss": 0.0806, + "step": 14683 + }, + { + "epoch": 6.15, + "learning_rate": 2.678823375955314e-06, + "loss": 0.0697, + "step": 14684 + }, + { + "epoch": 6.15, + "learning_rate": 2.677668605627086e-06, + "loss": 0.1013, + "step": 14685 + }, + { + "epoch": 6.15, + "learning_rate": 2.6765140457727813e-06, + "loss": 0.0662, + "step": 14686 + }, + { + "epoch": 6.15, + "learning_rate": 2.675359696425578e-06, + "loss": 0.0848, + "step": 14687 + }, + { + "epoch": 6.15, + "learning_rate": 2.67420555761866e-06, + "loss": 0.0681, + "step": 14688 + }, + { + "epoch": 6.15, + "learning_rate": 2.6730516293852014e-06, + "loss": 0.093, + "step": 14689 + }, + { + "epoch": 6.15, + "learning_rate": 2.671897911758373e-06, + "loss": 0.0732, + "step": 14690 + }, + { + "epoch": 6.15, + "learning_rate": 2.6707444047713373e-06, + "loss": 0.1054, + "step": 14691 + }, + { + "epoch": 6.15, + "learning_rate": 2.6695911084572514e-06, + "loss": 0.0781, + "step": 14692 + }, + { + "epoch": 6.15, + "learning_rate": 2.668438022849262e-06, + "loss": 0.0715, + "step": 14693 + }, + { + "epoch": 6.15, + "learning_rate": 2.667285147980515e-06, + "loss": 0.0876, + "step": 14694 + }, + { + "epoch": 6.15, + "learning_rate": 2.6661324838841495e-06, + "loss": 0.0982, + "step": 14695 + }, + { + "epoch": 6.15, + "learning_rate": 2.6649800305932984e-06, + "loss": 0.1052, + "step": 14696 + }, + { + "epoch": 6.15, + "learning_rate": 2.663827788141088e-06, + "loss": 0.0998, + "step": 14697 + }, + { + "epoch": 6.15, + "learning_rate": 2.6626757565606398e-06, + "loss": 0.0784, + "step": 14698 + }, + { + "epoch": 6.15, + "learning_rate": 2.661523935885063e-06, + "loss": 0.0928, + "step": 14699 + }, + { + "epoch": 6.15, + "learning_rate": 2.660372326147471e-06, + "loss": 0.0695, + "step": 14700 + }, + { + "epoch": 6.15, + "learning_rate": 2.659220927380961e-06, + "loss": 0.0439, + "step": 14701 + }, + { + "epoch": 6.16, + "learning_rate": 2.658069739618632e-06, + "loss": 0.0846, + "step": 14702 + }, + { + "epoch": 6.16, + "learning_rate": 2.6569187628935723e-06, + "loss": 0.072, + "step": 14703 + }, + { + "epoch": 6.16, + "learning_rate": 2.6557679972388674e-06, + "loss": 0.0804, + "step": 14704 + }, + { + "epoch": 6.16, + "learning_rate": 2.6546174426875937e-06, + "loss": 0.0963, + "step": 14705 + }, + { + "epoch": 6.16, + "learning_rate": 2.653467099272826e-06, + "loss": 0.0867, + "step": 14706 + }, + { + "epoch": 6.16, + "learning_rate": 2.652316967027625e-06, + "loss": 0.0503, + "step": 14707 + }, + { + "epoch": 6.16, + "learning_rate": 2.6511670459850526e-06, + "loss": 0.0855, + "step": 14708 + }, + { + "epoch": 6.16, + "learning_rate": 2.650017336178162e-06, + "loss": 0.0748, + "step": 14709 + }, + { + "epoch": 6.16, + "learning_rate": 2.648867837640001e-06, + "loss": 0.0822, + "step": 14710 + }, + { + "epoch": 6.16, + "learning_rate": 2.6477185504036106e-06, + "loss": 0.0679, + "step": 14711 + }, + { + "epoch": 6.16, + "learning_rate": 2.6465694745020265e-06, + "loss": 0.0653, + "step": 14712 + }, + { + "epoch": 6.16, + "learning_rate": 2.64542060996828e-06, + "loss": 0.0884, + "step": 14713 + }, + { + "epoch": 6.16, + "learning_rate": 2.6442719568353894e-06, + "loss": 0.0766, + "step": 14714 + }, + { + "epoch": 6.16, + "learning_rate": 2.643123515136373e-06, + "loss": 0.0957, + "step": 14715 + }, + { + "epoch": 6.16, + "learning_rate": 2.6419752849042437e-06, + "loss": 0.0947, + "step": 14716 + }, + { + "epoch": 6.16, + "learning_rate": 2.640827266172006e-06, + "loss": 0.1111, + "step": 14717 + }, + { + "epoch": 6.16, + "learning_rate": 2.639679458972657e-06, + "loss": 0.0927, + "step": 14718 + }, + { + "epoch": 6.16, + "learning_rate": 2.6385318633391934e-06, + "loss": 0.04, + "step": 14719 + }, + { + "epoch": 6.16, + "learning_rate": 2.6373844793045965e-06, + "loss": 0.0662, + "step": 14720 + }, + { + "epoch": 6.16, + "learning_rate": 2.636237306901852e-06, + "loss": 0.0734, + "step": 14721 + }, + { + "epoch": 6.16, + "learning_rate": 2.6350903461639287e-06, + "loss": 0.0683, + "step": 14722 + }, + { + "epoch": 6.16, + "learning_rate": 2.633943597123798e-06, + "loss": 0.0921, + "step": 14723 + }, + { + "epoch": 6.16, + "learning_rate": 2.6327970598144214e-06, + "loss": 0.0751, + "step": 14724 + }, + { + "epoch": 6.16, + "learning_rate": 2.631650734268757e-06, + "loss": 0.0901, + "step": 14725 + }, + { + "epoch": 6.17, + "learning_rate": 2.630504620519753e-06, + "loss": 0.095, + "step": 14726 + }, + { + "epoch": 6.17, + "learning_rate": 2.6293587186003568e-06, + "loss": 0.082, + "step": 14727 + }, + { + "epoch": 6.17, + "learning_rate": 2.6282130285435016e-06, + "loss": 0.1297, + "step": 14728 + }, + { + "epoch": 6.17, + "learning_rate": 2.6270675503821207e-06, + "loss": 0.0839, + "step": 14729 + }, + { + "epoch": 6.17, + "learning_rate": 2.6259222841491406e-06, + "loss": 0.0984, + "step": 14730 + }, + { + "epoch": 6.17, + "learning_rate": 2.6247772298774797e-06, + "loss": 0.0922, + "step": 14731 + }, + { + "epoch": 6.17, + "learning_rate": 2.6236323876000547e-06, + "loss": 0.0765, + "step": 14732 + }, + { + "epoch": 6.17, + "learning_rate": 2.622487757349772e-06, + "loss": 0.09, + "step": 14733 + }, + { + "epoch": 6.17, + "learning_rate": 2.62134333915953e-06, + "loss": 0.092, + "step": 14734 + }, + { + "epoch": 6.17, + "learning_rate": 2.620199133062227e-06, + "loss": 0.1081, + "step": 14735 + }, + { + "epoch": 6.17, + "learning_rate": 2.6190551390907495e-06, + "loss": 0.0826, + "step": 14736 + }, + { + "epoch": 6.17, + "learning_rate": 2.6179113572779835e-06, + "loss": 0.0785, + "step": 14737 + }, + { + "epoch": 6.17, + "learning_rate": 2.6167677876568044e-06, + "loss": 0.1054, + "step": 14738 + }, + { + "epoch": 6.17, + "learning_rate": 2.6156244302600864e-06, + "loss": 0.0771, + "step": 14739 + }, + { + "epoch": 6.17, + "learning_rate": 2.614481285120688e-06, + "loss": 0.0614, + "step": 14740 + }, + { + "epoch": 6.17, + "learning_rate": 2.6133383522714718e-06, + "loss": 0.0825, + "step": 14741 + }, + { + "epoch": 6.17, + "learning_rate": 2.612195631745289e-06, + "loss": 0.1003, + "step": 14742 + }, + { + "epoch": 6.17, + "learning_rate": 2.611053123574988e-06, + "loss": 0.0942, + "step": 14743 + }, + { + "epoch": 6.17, + "learning_rate": 2.6099108277934105e-06, + "loss": 0.1015, + "step": 14744 + }, + { + "epoch": 6.17, + "learning_rate": 2.6087687444333853e-06, + "loss": 0.0999, + "step": 14745 + }, + { + "epoch": 6.17, + "learning_rate": 2.6076268735277445e-06, + "loss": 0.068, + "step": 14746 + }, + { + "epoch": 6.17, + "learning_rate": 2.6064852151093124e-06, + "loss": 0.115, + "step": 14747 + }, + { + "epoch": 6.17, + "learning_rate": 2.6053437692108974e-06, + "loss": 0.0829, + "step": 14748 + }, + { + "epoch": 6.18, + "learning_rate": 2.6042025358653157e-06, + "loss": 0.0689, + "step": 14749 + }, + { + "epoch": 6.18, + "learning_rate": 2.6030615151053675e-06, + "loss": 0.0944, + "step": 14750 + }, + { + "epoch": 6.18, + "learning_rate": 2.601920706963853e-06, + "loss": 0.0837, + "step": 14751 + }, + { + "epoch": 6.18, + "learning_rate": 2.600780111473563e-06, + "loss": 0.0939, + "step": 14752 + }, + { + "epoch": 6.18, + "learning_rate": 2.5996397286672826e-06, + "loss": 0.1301, + "step": 14753 + }, + { + "epoch": 6.18, + "learning_rate": 2.598499558577794e-06, + "loss": 0.0731, + "step": 14754 + }, + { + "epoch": 6.18, + "learning_rate": 2.5973596012378644e-06, + "loss": 0.099, + "step": 14755 + }, + { + "epoch": 6.18, + "learning_rate": 2.5962198566802643e-06, + "loss": 0.0624, + "step": 14756 + }, + { + "epoch": 6.18, + "learning_rate": 2.5950803249377554e-06, + "loss": 0.0775, + "step": 14757 + }, + { + "epoch": 6.18, + "learning_rate": 2.5939410060430904e-06, + "loss": 0.0849, + "step": 14758 + }, + { + "epoch": 6.18, + "learning_rate": 2.5928019000290194e-06, + "loss": 0.0922, + "step": 14759 + }, + { + "epoch": 6.18, + "learning_rate": 2.591663006928288e-06, + "loss": 0.045, + "step": 14760 + }, + { + "epoch": 6.18, + "learning_rate": 2.5905243267736256e-06, + "loss": 0.1222, + "step": 14761 + }, + { + "epoch": 6.18, + "learning_rate": 2.589385859597767e-06, + "loss": 0.0858, + "step": 14762 + }, + { + "epoch": 6.18, + "learning_rate": 2.588247605433436e-06, + "loss": 0.0735, + "step": 14763 + }, + { + "epoch": 6.18, + "learning_rate": 2.5871095643133502e-06, + "loss": 0.0729, + "step": 14764 + }, + { + "epoch": 6.18, + "learning_rate": 2.585971736270223e-06, + "loss": 0.0955, + "step": 14765 + }, + { + "epoch": 6.18, + "learning_rate": 2.584834121336761e-06, + "loss": 0.0926, + "step": 14766 + }, + { + "epoch": 6.18, + "learning_rate": 2.5836967195456586e-06, + "loss": 0.0997, + "step": 14767 + }, + { + "epoch": 6.18, + "learning_rate": 2.582559530929617e-06, + "loss": 0.0763, + "step": 14768 + }, + { + "epoch": 6.18, + "learning_rate": 2.581422555521316e-06, + "loss": 0.0615, + "step": 14769 + }, + { + "epoch": 6.18, + "learning_rate": 2.5802857933534407e-06, + "loss": 0.0617, + "step": 14770 + }, + { + "epoch": 6.18, + "learning_rate": 2.579149244458666e-06, + "loss": 0.0791, + "step": 14771 + }, + { + "epoch": 6.18, + "learning_rate": 2.5780129088696614e-06, + "loss": 0.0943, + "step": 14772 + }, + { + "epoch": 6.19, + "learning_rate": 2.5768767866190893e-06, + "loss": 0.1239, + "step": 14773 + }, + { + "epoch": 6.19, + "learning_rate": 2.57574087773961e-06, + "loss": 0.101, + "step": 14774 + }, + { + "epoch": 6.19, + "learning_rate": 2.5746051822638675e-06, + "loss": 0.0708, + "step": 14775 + }, + { + "epoch": 6.19, + "learning_rate": 2.57346970022451e-06, + "loss": 0.0843, + "step": 14776 + }, + { + "epoch": 6.19, + "learning_rate": 2.5723344316541767e-06, + "loss": 0.0627, + "step": 14777 + }, + { + "epoch": 6.19, + "learning_rate": 2.5711993765854983e-06, + "loss": 0.0937, + "step": 14778 + }, + { + "epoch": 6.19, + "learning_rate": 2.570064535051101e-06, + "loss": 0.1021, + "step": 14779 + }, + { + "epoch": 6.19, + "learning_rate": 2.568929907083609e-06, + "loss": 0.0855, + "step": 14780 + }, + { + "epoch": 6.19, + "learning_rate": 2.56779549271563e-06, + "loss": 0.084, + "step": 14781 + }, + { + "epoch": 6.19, + "learning_rate": 2.566661291979774e-06, + "loss": 0.0663, + "step": 14782 + }, + { + "epoch": 6.19, + "learning_rate": 2.5655273049086425e-06, + "loss": 0.0879, + "step": 14783 + }, + { + "epoch": 6.19, + "learning_rate": 2.564393531534832e-06, + "loss": 0.0717, + "step": 14784 + }, + { + "epoch": 6.19, + "learning_rate": 2.563259971890931e-06, + "loss": 0.0717, + "step": 14785 + }, + { + "epoch": 6.19, + "learning_rate": 2.562126626009522e-06, + "loss": 0.1152, + "step": 14786 + }, + { + "epoch": 6.19, + "learning_rate": 2.5609934939231873e-06, + "loss": 0.0707, + "step": 14787 + }, + { + "epoch": 6.19, + "learning_rate": 2.5598605756644902e-06, + "loss": 0.0919, + "step": 14788 + }, + { + "epoch": 6.19, + "learning_rate": 2.558727871266001e-06, + "loss": 0.0781, + "step": 14789 + }, + { + "epoch": 6.19, + "learning_rate": 2.5575953807602727e-06, + "loss": 0.0687, + "step": 14790 + }, + { + "epoch": 6.19, + "learning_rate": 2.5564631041798615e-06, + "loss": 0.0992, + "step": 14791 + }, + { + "epoch": 6.19, + "learning_rate": 2.555331041557313e-06, + "loss": 0.1118, + "step": 14792 + }, + { + "epoch": 6.19, + "learning_rate": 2.5541991929251676e-06, + "loss": 0.0891, + "step": 14793 + }, + { + "epoch": 6.19, + "learning_rate": 2.553067558315958e-06, + "loss": 0.1247, + "step": 14794 + }, + { + "epoch": 6.19, + "learning_rate": 2.5519361377622176e-06, + "loss": 0.0741, + "step": 14795 + }, + { + "epoch": 6.19, + "learning_rate": 2.55080493129646e-06, + "loss": 0.0669, + "step": 14796 + }, + { + "epoch": 6.2, + "learning_rate": 2.5496739389512036e-06, + "loss": 0.0552, + "step": 14797 + }, + { + "epoch": 6.2, + "learning_rate": 2.5485431607589594e-06, + "loss": 0.0481, + "step": 14798 + }, + { + "epoch": 6.2, + "learning_rate": 2.54741259675223e-06, + "loss": 0.0748, + "step": 14799 + }, + { + "epoch": 6.2, + "learning_rate": 2.5462822469635117e-06, + "loss": 0.0843, + "step": 14800 + }, + { + "epoch": 6.2, + "learning_rate": 2.5451521114252985e-06, + "loss": 0.0868, + "step": 14801 + }, + { + "epoch": 6.2, + "learning_rate": 2.54402219017007e-06, + "loss": 0.0869, + "step": 14802 + }, + { + "epoch": 6.2, + "learning_rate": 2.542892483230308e-06, + "loss": 0.0633, + "step": 14803 + }, + { + "epoch": 6.2, + "learning_rate": 2.5417629906384834e-06, + "loss": 0.1067, + "step": 14804 + }, + { + "epoch": 6.2, + "learning_rate": 2.540633712427063e-06, + "loss": 0.0969, + "step": 14805 + }, + { + "epoch": 6.2, + "learning_rate": 2.539504648628508e-06, + "loss": 0.0993, + "step": 14806 + }, + { + "epoch": 6.2, + "learning_rate": 2.5383757992752744e-06, + "loss": 0.1001, + "step": 14807 + }, + { + "epoch": 6.2, + "learning_rate": 2.5372471643998042e-06, + "loss": 0.1042, + "step": 14808 + }, + { + "epoch": 6.2, + "learning_rate": 2.5361187440345416e-06, + "loss": 0.0951, + "step": 14809 + }, + { + "epoch": 6.2, + "learning_rate": 2.534990538211922e-06, + "loss": 0.0607, + "step": 14810 + }, + { + "epoch": 6.2, + "learning_rate": 2.533862546964376e-06, + "loss": 0.0593, + "step": 14811 + }, + { + "epoch": 6.2, + "learning_rate": 2.532734770324329e-06, + "loss": 0.1005, + "step": 14812 + }, + { + "epoch": 6.2, + "learning_rate": 2.5316072083241904e-06, + "loss": 0.0799, + "step": 14813 + }, + { + "epoch": 6.2, + "learning_rate": 2.5304798609963754e-06, + "loss": 0.0865, + "step": 14814 + }, + { + "epoch": 6.2, + "learning_rate": 2.529352728373292e-06, + "loss": 0.0834, + "step": 14815 + }, + { + "epoch": 6.2, + "learning_rate": 2.5282258104873324e-06, + "loss": 0.0644, + "step": 14816 + }, + { + "epoch": 6.2, + "learning_rate": 2.5270991073708905e-06, + "loss": 0.1018, + "step": 14817 + }, + { + "epoch": 6.2, + "learning_rate": 2.525972619056355e-06, + "loss": 0.0698, + "step": 14818 + }, + { + "epoch": 6.2, + "learning_rate": 2.524846345576103e-06, + "loss": 0.0751, + "step": 14819 + }, + { + "epoch": 6.2, + "learning_rate": 2.52372028696251e-06, + "loss": 0.062, + "step": 14820 + }, + { + "epoch": 6.21, + "learning_rate": 2.5225944432479452e-06, + "loss": 0.1017, + "step": 14821 + }, + { + "epoch": 6.21, + "learning_rate": 2.5214688144647658e-06, + "loss": 0.1301, + "step": 14822 + }, + { + "epoch": 6.21, + "learning_rate": 2.5203434006453287e-06, + "loss": 0.1092, + "step": 14823 + }, + { + "epoch": 6.21, + "learning_rate": 2.5192182018219826e-06, + "loss": 0.0819, + "step": 14824 + }, + { + "epoch": 6.21, + "learning_rate": 2.518093218027071e-06, + "loss": 0.0859, + "step": 14825 + }, + { + "epoch": 6.21, + "learning_rate": 2.5169684492929315e-06, + "loss": 0.0949, + "step": 14826 + }, + { + "epoch": 6.21, + "learning_rate": 2.515843895651893e-06, + "loss": 0.1155, + "step": 14827 + }, + { + "epoch": 6.21, + "learning_rate": 2.5147195571362837e-06, + "loss": 0.0719, + "step": 14828 + }, + { + "epoch": 6.21, + "learning_rate": 2.513595433778415e-06, + "loss": 0.0965, + "step": 14829 + }, + { + "epoch": 6.21, + "learning_rate": 2.5124715256106024e-06, + "loss": 0.0823, + "step": 14830 + }, + { + "epoch": 6.21, + "learning_rate": 2.511347832665152e-06, + "loss": 0.0948, + "step": 14831 + }, + { + "epoch": 6.21, + "learning_rate": 2.510224354974362e-06, + "loss": 0.1077, + "step": 14832 + }, + { + "epoch": 6.21, + "learning_rate": 2.509101092570527e-06, + "loss": 0.0804, + "step": 14833 + }, + { + "epoch": 6.21, + "learning_rate": 2.507978045485937e-06, + "loss": 0.0918, + "step": 14834 + }, + { + "epoch": 6.21, + "learning_rate": 2.5068552137528667e-06, + "loss": 0.0652, + "step": 14835 + }, + { + "epoch": 6.21, + "learning_rate": 2.5057325974035964e-06, + "loss": 0.0755, + "step": 14836 + }, + { + "epoch": 6.21, + "learning_rate": 2.5046101964703894e-06, + "loss": 0.0666, + "step": 14837 + }, + { + "epoch": 6.21, + "learning_rate": 2.503488010985512e-06, + "loss": 0.0726, + "step": 14838 + }, + { + "epoch": 6.21, + "learning_rate": 2.502366040981218e-06, + "loss": 0.0622, + "step": 14839 + }, + { + "epoch": 6.21, + "learning_rate": 2.50124428648976e-06, + "loss": 0.0765, + "step": 14840 + }, + { + "epoch": 6.21, + "learning_rate": 2.5001227475433797e-06, + "loss": 0.1044, + "step": 14841 + }, + { + "epoch": 6.21, + "learning_rate": 2.49900142417432e-06, + "loss": 0.0872, + "step": 14842 + }, + { + "epoch": 6.21, + "learning_rate": 2.4978803164148047e-06, + "loss": 0.1062, + "step": 14843 + }, + { + "epoch": 6.21, + "learning_rate": 2.496759424297062e-06, + "loss": 0.0653, + "step": 14844 + }, + { + "epoch": 6.22, + "learning_rate": 2.495638747853312e-06, + "loss": 0.0966, + "step": 14845 + }, + { + "epoch": 6.22, + "learning_rate": 2.4945182871157656e-06, + "loss": 0.1056, + "step": 14846 + }, + { + "epoch": 6.22, + "learning_rate": 2.4933980421166326e-06, + "loss": 0.1049, + "step": 14847 + }, + { + "epoch": 6.22, + "learning_rate": 2.492278012888113e-06, + "loss": 0.0643, + "step": 14848 + }, + { + "epoch": 6.22, + "learning_rate": 2.491158199462398e-06, + "loss": 0.0765, + "step": 14849 + }, + { + "epoch": 6.22, + "learning_rate": 2.490038601871676e-06, + "loss": 0.0877, + "step": 14850 + }, + { + "epoch": 6.22, + "learning_rate": 2.4889192201481315e-06, + "loss": 0.0766, + "step": 14851 + }, + { + "epoch": 6.22, + "learning_rate": 2.4878000543239377e-06, + "loss": 0.0746, + "step": 14852 + }, + { + "epoch": 6.22, + "learning_rate": 2.4866811044312667e-06, + "loss": 0.0629, + "step": 14853 + }, + { + "epoch": 6.22, + "learning_rate": 2.485562370502279e-06, + "loss": 0.0779, + "step": 14854 + }, + { + "epoch": 6.22, + "learning_rate": 2.484443852569135e-06, + "loss": 0.0831, + "step": 14855 + }, + { + "epoch": 6.22, + "learning_rate": 2.483325550663982e-06, + "loss": 0.0806, + "step": 14856 + }, + { + "epoch": 6.22, + "learning_rate": 2.4822074648189675e-06, + "loss": 0.0728, + "step": 14857 + }, + { + "epoch": 6.22, + "learning_rate": 2.481089595066226e-06, + "loss": 0.0995, + "step": 14858 + }, + { + "epoch": 6.22, + "learning_rate": 2.479971941437892e-06, + "loss": 0.0717, + "step": 14859 + }, + { + "epoch": 6.22, + "learning_rate": 2.4788545039660905e-06, + "loss": 0.082, + "step": 14860 + }, + { + "epoch": 6.22, + "learning_rate": 2.4777372826829427e-06, + "loss": 0.1138, + "step": 14861 + }, + { + "epoch": 6.22, + "learning_rate": 2.476620277620565e-06, + "loss": 0.0985, + "step": 14862 + }, + { + "epoch": 6.22, + "learning_rate": 2.4755034888110576e-06, + "loss": 0.1266, + "step": 14863 + }, + { + "epoch": 6.22, + "learning_rate": 2.474386916286526e-06, + "loss": 0.0751, + "step": 14864 + }, + { + "epoch": 6.22, + "learning_rate": 2.473270560079064e-06, + "loss": 0.0774, + "step": 14865 + }, + { + "epoch": 6.22, + "learning_rate": 2.47215442022076e-06, + "loss": 0.101, + "step": 14866 + }, + { + "epoch": 6.22, + "learning_rate": 2.4710384967436983e-06, + "loss": 0.0881, + "step": 14867 + }, + { + "epoch": 6.22, + "learning_rate": 2.469922789679954e-06, + "loss": 0.0825, + "step": 14868 + }, + { + "epoch": 6.23, + "learning_rate": 2.4688072990615987e-06, + "loss": 0.0701, + "step": 14869 + }, + { + "epoch": 6.23, + "learning_rate": 2.4676920249206926e-06, + "loss": 0.1069, + "step": 14870 + }, + { + "epoch": 6.23, + "learning_rate": 2.4665769672892946e-06, + "loss": 0.0739, + "step": 14871 + }, + { + "epoch": 6.23, + "learning_rate": 2.465462126199457e-06, + "loss": 0.0832, + "step": 14872 + }, + { + "epoch": 6.23, + "learning_rate": 2.464347501683224e-06, + "loss": 0.0882, + "step": 14873 + }, + { + "epoch": 6.23, + "learning_rate": 2.4632330937726356e-06, + "loss": 0.0955, + "step": 14874 + }, + { + "epoch": 6.23, + "learning_rate": 2.4621189024997273e-06, + "loss": 0.1045, + "step": 14875 + }, + { + "epoch": 6.23, + "learning_rate": 2.4610049278965175e-06, + "loss": 0.104, + "step": 14876 + }, + { + "epoch": 6.23, + "learning_rate": 2.4598911699950325e-06, + "loss": 0.0882, + "step": 14877 + }, + { + "epoch": 6.23, + "learning_rate": 2.4587776288272846e-06, + "loss": 0.0646, + "step": 14878 + }, + { + "epoch": 6.23, + "learning_rate": 2.4576643044252824e-06, + "loss": 0.0777, + "step": 14879 + }, + { + "epoch": 6.23, + "learning_rate": 2.4565511968210285e-06, + "loss": 0.0748, + "step": 14880 + }, + { + "epoch": 6.23, + "learning_rate": 2.455438306046515e-06, + "loss": 0.1052, + "step": 14881 + }, + { + "epoch": 6.23, + "learning_rate": 2.4543256321337306e-06, + "loss": 0.0863, + "step": 14882 + }, + { + "epoch": 6.23, + "learning_rate": 2.453213175114664e-06, + "loss": 0.0721, + "step": 14883 + }, + { + "epoch": 6.23, + "learning_rate": 2.4521009350212854e-06, + "loss": 0.0985, + "step": 14884 + }, + { + "epoch": 6.23, + "learning_rate": 2.4509889118855666e-06, + "loss": 0.0605, + "step": 14885 + }, + { + "epoch": 6.23, + "learning_rate": 2.449877105739473e-06, + "loss": 0.0552, + "step": 14886 + }, + { + "epoch": 6.23, + "learning_rate": 2.4487655166149625e-06, + "loss": 0.0864, + "step": 14887 + }, + { + "epoch": 6.23, + "learning_rate": 2.447654144543986e-06, + "loss": 0.0859, + "step": 14888 + }, + { + "epoch": 6.23, + "learning_rate": 2.4465429895584914e-06, + "loss": 0.0513, + "step": 14889 + }, + { + "epoch": 6.23, + "learning_rate": 2.445432051690414e-06, + "loss": 0.0702, + "step": 14890 + }, + { + "epoch": 6.23, + "learning_rate": 2.4443213309716884e-06, + "loss": 0.0878, + "step": 14891 + }, + { + "epoch": 6.23, + "learning_rate": 2.4432108274342405e-06, + "loss": 0.0621, + "step": 14892 + }, + { + "epoch": 6.24, + "learning_rate": 2.4421005411099928e-06, + "loss": 0.0659, + "step": 14893 + }, + { + "epoch": 6.24, + "learning_rate": 2.4409904720308576e-06, + "loss": 0.0877, + "step": 14894 + }, + { + "epoch": 6.24, + "learning_rate": 2.4398806202287427e-06, + "loss": 0.1058, + "step": 14895 + }, + { + "epoch": 6.24, + "learning_rate": 2.4387709857355545e-06, + "loss": 0.0755, + "step": 14896 + }, + { + "epoch": 6.24, + "learning_rate": 2.4376615685831817e-06, + "loss": 0.1033, + "step": 14897 + }, + { + "epoch": 6.24, + "learning_rate": 2.4365523688035163e-06, + "loss": 0.0995, + "step": 14898 + }, + { + "epoch": 6.24, + "learning_rate": 2.435443386428442e-06, + "loss": 0.0744, + "step": 14899 + }, + { + "epoch": 6.24, + "learning_rate": 2.4343346214898343e-06, + "loss": 0.0632, + "step": 14900 + }, + { + "epoch": 6.24, + "learning_rate": 2.4332260740195645e-06, + "loss": 0.0854, + "step": 14901 + }, + { + "epoch": 6.24, + "learning_rate": 2.4321177440494993e-06, + "loss": 0.0875, + "step": 14902 + }, + { + "epoch": 6.24, + "learning_rate": 2.431009631611493e-06, + "loss": 0.0932, + "step": 14903 + }, + { + "epoch": 6.24, + "learning_rate": 2.4299017367373966e-06, + "loss": 0.0912, + "step": 14904 + }, + { + "epoch": 6.24, + "learning_rate": 2.4287940594590564e-06, + "loss": 0.0737, + "step": 14905 + }, + { + "epoch": 6.24, + "learning_rate": 2.4276865998083132e-06, + "loss": 0.0779, + "step": 14906 + }, + { + "epoch": 6.24, + "learning_rate": 2.426579357816998e-06, + "loss": 0.0774, + "step": 14907 + }, + { + "epoch": 6.24, + "learning_rate": 2.4254723335169394e-06, + "loss": 0.0942, + "step": 14908 + }, + { + "epoch": 6.24, + "learning_rate": 2.424365526939958e-06, + "loss": 0.0811, + "step": 14909 + }, + { + "epoch": 6.24, + "learning_rate": 2.4232589381178685e-06, + "loss": 0.0895, + "step": 14910 + }, + { + "epoch": 6.24, + "learning_rate": 2.422152567082474e-06, + "loss": 0.1295, + "step": 14911 + }, + { + "epoch": 6.24, + "learning_rate": 2.4210464138655807e-06, + "loss": 0.1035, + "step": 14912 + }, + { + "epoch": 6.24, + "learning_rate": 2.4199404784989833e-06, + "loss": 0.0901, + "step": 14913 + }, + { + "epoch": 6.24, + "learning_rate": 2.4188347610144693e-06, + "loss": 0.0943, + "step": 14914 + }, + { + "epoch": 6.24, + "learning_rate": 2.417729261443823e-06, + "loss": 0.07, + "step": 14915 + }, + { + "epoch": 6.24, + "learning_rate": 2.416623979818823e-06, + "loss": 0.0666, + "step": 14916 + }, + { + "epoch": 6.25, + "learning_rate": 2.4155189161712358e-06, + "loss": 0.0687, + "step": 14917 + }, + { + "epoch": 6.25, + "learning_rate": 2.4144140705328258e-06, + "loss": 0.0905, + "step": 14918 + }, + { + "epoch": 6.25, + "learning_rate": 2.4133094429353522e-06, + "loss": 0.0994, + "step": 14919 + }, + { + "epoch": 6.25, + "learning_rate": 2.412205033410566e-06, + "loss": 0.0748, + "step": 14920 + }, + { + "epoch": 6.25, + "learning_rate": 2.4111008419902138e-06, + "loss": 0.1008, + "step": 14921 + }, + { + "epoch": 6.25, + "learning_rate": 2.409996868706036e-06, + "loss": 0.1057, + "step": 14922 + }, + { + "epoch": 6.25, + "learning_rate": 2.4088931135897596e-06, + "loss": 0.0828, + "step": 14923 + }, + { + "epoch": 6.25, + "learning_rate": 2.407789576673115e-06, + "loss": 0.0655, + "step": 14924 + }, + { + "epoch": 6.25, + "learning_rate": 2.406686257987825e-06, + "loss": 0.0944, + "step": 14925 + }, + { + "epoch": 6.25, + "learning_rate": 2.4055831575655975e-06, + "loss": 0.0802, + "step": 14926 + }, + { + "epoch": 6.25, + "learning_rate": 2.4044802754381423e-06, + "loss": 0.0785, + "step": 14927 + }, + { + "epoch": 6.25, + "learning_rate": 2.4033776116371623e-06, + "loss": 0.0997, + "step": 14928 + }, + { + "epoch": 6.25, + "learning_rate": 2.402275166194351e-06, + "loss": 0.084, + "step": 14929 + }, + { + "epoch": 6.25, + "learning_rate": 2.401172939141401e-06, + "loss": 0.0743, + "step": 14930 + }, + { + "epoch": 6.25, + "learning_rate": 2.400070930509989e-06, + "loss": 0.0961, + "step": 14931 + }, + { + "epoch": 6.25, + "learning_rate": 2.398969140331794e-06, + "loss": 0.0833, + "step": 14932 + }, + { + "epoch": 6.25, + "learning_rate": 2.3978675686384854e-06, + "loss": 0.0987, + "step": 14933 + }, + { + "epoch": 6.25, + "learning_rate": 2.396766215461728e-06, + "loss": 0.0845, + "step": 14934 + }, + { + "epoch": 6.25, + "learning_rate": 2.3956650808331795e-06, + "loss": 0.0589, + "step": 14935 + }, + { + "epoch": 6.25, + "learning_rate": 2.394564164784489e-06, + "loss": 0.1022, + "step": 14936 + }, + { + "epoch": 6.25, + "learning_rate": 2.3934634673473066e-06, + "loss": 0.0674, + "step": 14937 + }, + { + "epoch": 6.25, + "learning_rate": 2.392362988553263e-06, + "loss": 0.0875, + "step": 14938 + }, + { + "epoch": 6.25, + "learning_rate": 2.391262728433995e-06, + "loss": 0.084, + "step": 14939 + }, + { + "epoch": 6.25, + "learning_rate": 2.3901626870211282e-06, + "loss": 0.0814, + "step": 14940 + }, + { + "epoch": 6.26, + "learning_rate": 2.389062864346282e-06, + "loss": 0.0893, + "step": 14941 + }, + { + "epoch": 6.26, + "learning_rate": 2.3879632604410695e-06, + "loss": 0.101, + "step": 14942 + }, + { + "epoch": 6.26, + "learning_rate": 2.386863875337102e-06, + "loss": 0.0771, + "step": 14943 + }, + { + "epoch": 6.26, + "learning_rate": 2.385764709065973e-06, + "loss": 0.09, + "step": 14944 + }, + { + "epoch": 6.26, + "learning_rate": 2.3846657616592805e-06, + "loss": 0.0615, + "step": 14945 + }, + { + "epoch": 6.26, + "learning_rate": 2.3835670331486137e-06, + "loss": 0.094, + "step": 14946 + }, + { + "epoch": 6.26, + "learning_rate": 2.3824685235655532e-06, + "loss": 0.083, + "step": 14947 + }, + { + "epoch": 6.26, + "learning_rate": 2.381370232941679e-06, + "loss": 0.0625, + "step": 14948 + }, + { + "epoch": 6.26, + "learning_rate": 2.3802721613085534e-06, + "loss": 0.1044, + "step": 14949 + }, + { + "epoch": 6.26, + "learning_rate": 2.3791743086977425e-06, + "loss": 0.0795, + "step": 14950 + }, + { + "epoch": 6.26, + "learning_rate": 2.378076675140807e-06, + "loss": 0.0791, + "step": 14951 + }, + { + "epoch": 6.26, + "learning_rate": 2.376979260669291e-06, + "loss": 0.0932, + "step": 14952 + }, + { + "epoch": 6.26, + "learning_rate": 2.3758820653147405e-06, + "loss": 0.0702, + "step": 14953 + }, + { + "epoch": 6.26, + "learning_rate": 2.3747850891086956e-06, + "loss": 0.0805, + "step": 14954 + }, + { + "epoch": 6.26, + "learning_rate": 2.3736883320826864e-06, + "loss": 0.0756, + "step": 14955 + }, + { + "epoch": 6.26, + "learning_rate": 2.3725917942682397e-06, + "loss": 0.099, + "step": 14956 + }, + { + "epoch": 6.26, + "learning_rate": 2.3714954756968746e-06, + "loss": 0.0872, + "step": 14957 + }, + { + "epoch": 6.26, + "learning_rate": 2.370399376400101e-06, + "loss": 0.0862, + "step": 14958 + }, + { + "epoch": 6.26, + "learning_rate": 2.3693034964094264e-06, + "loss": 0.0518, + "step": 14959 + }, + { + "epoch": 6.26, + "learning_rate": 2.368207835756352e-06, + "loss": 0.0863, + "step": 14960 + }, + { + "epoch": 6.26, + "learning_rate": 2.367112394472371e-06, + "loss": 0.0785, + "step": 14961 + }, + { + "epoch": 6.26, + "learning_rate": 2.3660171725889703e-06, + "loss": 0.1134, + "step": 14962 + }, + { + "epoch": 6.26, + "learning_rate": 2.3649221701376346e-06, + "loss": 0.0889, + "step": 14963 + }, + { + "epoch": 6.27, + "learning_rate": 2.363827387149833e-06, + "loss": 0.0916, + "step": 14964 + }, + { + "epoch": 6.27, + "learning_rate": 2.3627328236570367e-06, + "loss": 0.087, + "step": 14965 + }, + { + "epoch": 6.27, + "learning_rate": 2.3616384796907087e-06, + "loss": 0.0704, + "step": 14966 + }, + { + "epoch": 6.27, + "learning_rate": 2.360544355282305e-06, + "loss": 0.0926, + "step": 14967 + }, + { + "epoch": 6.27, + "learning_rate": 2.359450450463274e-06, + "loss": 0.0874, + "step": 14968 + }, + { + "epoch": 6.27, + "learning_rate": 2.35835676526506e-06, + "loss": 0.1054, + "step": 14969 + }, + { + "epoch": 6.27, + "learning_rate": 2.3572632997191016e-06, + "loss": 0.0871, + "step": 14970 + }, + { + "epoch": 6.27, + "learning_rate": 2.356170053856829e-06, + "loss": 0.0815, + "step": 14971 + }, + { + "epoch": 6.27, + "learning_rate": 2.3550770277096615e-06, + "loss": 0.0962, + "step": 14972 + }, + { + "epoch": 6.27, + "learning_rate": 2.35398422130902e-06, + "loss": 0.0683, + "step": 14973 + }, + { + "epoch": 6.27, + "learning_rate": 2.3528916346863185e-06, + "loss": 0.0937, + "step": 14974 + }, + { + "epoch": 6.27, + "learning_rate": 2.3517992678729607e-06, + "loss": 0.0538, + "step": 14975 + }, + { + "epoch": 6.27, + "learning_rate": 2.3507071209003463e-06, + "loss": 0.0896, + "step": 14976 + }, + { + "epoch": 6.27, + "learning_rate": 2.349615193799868e-06, + "loss": 0.0762, + "step": 14977 + }, + { + "epoch": 6.27, + "learning_rate": 2.3485234866029137e-06, + "loss": 0.1219, + "step": 14978 + }, + { + "epoch": 6.27, + "learning_rate": 2.34743199934086e-06, + "loss": 0.1079, + "step": 14979 + }, + { + "epoch": 6.27, + "learning_rate": 2.3463407320450827e-06, + "loss": 0.0814, + "step": 14980 + }, + { + "epoch": 6.27, + "learning_rate": 2.34524968474695e-06, + "loss": 0.1027, + "step": 14981 + }, + { + "epoch": 6.27, + "learning_rate": 2.3441588574778207e-06, + "loss": 0.0672, + "step": 14982 + }, + { + "epoch": 6.27, + "learning_rate": 2.3430682502690527e-06, + "loss": 0.0835, + "step": 14983 + }, + { + "epoch": 6.27, + "learning_rate": 2.3419778631519963e-06, + "loss": 0.079, + "step": 14984 + }, + { + "epoch": 6.27, + "learning_rate": 2.340887696157986e-06, + "loss": 0.0962, + "step": 14985 + }, + { + "epoch": 6.27, + "learning_rate": 2.3397977493183643e-06, + "loss": 0.1165, + "step": 14986 + }, + { + "epoch": 6.27, + "learning_rate": 2.3387080226644578e-06, + "loss": 0.0685, + "step": 14987 + }, + { + "epoch": 6.28, + "learning_rate": 2.337618516227591e-06, + "loss": 0.058, + "step": 14988 + }, + { + "epoch": 6.28, + "learning_rate": 2.3365292300390795e-06, + "loss": 0.0789, + "step": 14989 + }, + { + "epoch": 6.28, + "learning_rate": 2.3354401641302395e-06, + "loss": 0.0779, + "step": 14990 + }, + { + "epoch": 6.28, + "learning_rate": 2.3343513185323664e-06, + "loss": 0.0699, + "step": 14991 + }, + { + "epoch": 6.28, + "learning_rate": 2.333262693276762e-06, + "loss": 0.0783, + "step": 14992 + }, + { + "epoch": 6.28, + "learning_rate": 2.332174288394722e-06, + "loss": 0.0981, + "step": 14993 + }, + { + "epoch": 6.28, + "learning_rate": 2.331086103917525e-06, + "loss": 0.0759, + "step": 14994 + }, + { + "epoch": 6.28, + "learning_rate": 2.3299981398764515e-06, + "loss": 0.0885, + "step": 14995 + }, + { + "epoch": 6.28, + "learning_rate": 2.328910396302776e-06, + "loss": 0.0812, + "step": 14996 + }, + { + "epoch": 6.28, + "learning_rate": 2.327822873227764e-06, + "loss": 0.0845, + "step": 14997 + }, + { + "epoch": 6.28, + "learning_rate": 2.326735570682678e-06, + "loss": 0.0802, + "step": 14998 + }, + { + "epoch": 6.28, + "learning_rate": 2.325648488698766e-06, + "loss": 0.0752, + "step": 14999 + }, + { + "epoch": 6.28, + "learning_rate": 2.324561627307278e-06, + "loss": 0.0878, + "step": 15000 + }, + { + "epoch": 6.28, + "learning_rate": 2.3234749865394545e-06, + "loss": 0.0812, + "step": 15001 + }, + { + "epoch": 6.28, + "learning_rate": 2.322388566426532e-06, + "loss": 0.055, + "step": 15002 + }, + { + "epoch": 6.28, + "learning_rate": 2.321302366999736e-06, + "loss": 0.0815, + "step": 15003 + }, + { + "epoch": 6.28, + "learning_rate": 2.320216388290293e-06, + "loss": 0.0931, + "step": 15004 + }, + { + "epoch": 6.28, + "learning_rate": 2.3191306303294114e-06, + "loss": 0.0878, + "step": 15005 + }, + { + "epoch": 6.28, + "learning_rate": 2.3180450931483045e-06, + "loss": 0.0696, + "step": 15006 + }, + { + "epoch": 6.28, + "learning_rate": 2.3169597767781736e-06, + "loss": 0.0859, + "step": 15007 + }, + { + "epoch": 6.28, + "learning_rate": 2.3158746812502174e-06, + "loss": 0.0825, + "step": 15008 + }, + { + "epoch": 6.28, + "learning_rate": 2.314789806595624e-06, + "loss": 0.1027, + "step": 15009 + }, + { + "epoch": 6.28, + "learning_rate": 2.3137051528455778e-06, + "loss": 0.1111, + "step": 15010 + }, + { + "epoch": 6.28, + "learning_rate": 2.31262072003126e-06, + "loss": 0.0816, + "step": 15011 + }, + { + "epoch": 6.29, + "learning_rate": 2.3115365081838346e-06, + "loss": 0.0815, + "step": 15012 + }, + { + "epoch": 6.29, + "learning_rate": 2.3104525173344693e-06, + "loss": 0.0544, + "step": 15013 + }, + { + "epoch": 6.29, + "learning_rate": 2.3093687475143236e-06, + "loss": 0.0848, + "step": 15014 + }, + { + "epoch": 6.29, + "learning_rate": 2.3082851987545474e-06, + "loss": 0.1051, + "step": 15015 + }, + { + "epoch": 6.29, + "learning_rate": 2.307201871086292e-06, + "loss": 0.0788, + "step": 15016 + }, + { + "epoch": 6.29, + "learning_rate": 2.306118764540688e-06, + "loss": 0.1185, + "step": 15017 + }, + { + "epoch": 6.29, + "learning_rate": 2.3050358791488734e-06, + "loss": 0.1118, + "step": 15018 + }, + { + "epoch": 6.29, + "learning_rate": 2.3039532149419775e-06, + "loss": 0.1037, + "step": 15019 + }, + { + "epoch": 6.29, + "learning_rate": 2.3028707719511134e-06, + "loss": 0.1002, + "step": 15020 + }, + { + "epoch": 6.29, + "learning_rate": 2.301788550207399e-06, + "loss": 0.0603, + "step": 15021 + }, + { + "epoch": 6.29, + "learning_rate": 2.3007065497419413e-06, + "loss": 0.1137, + "step": 15022 + }, + { + "epoch": 6.29, + "learning_rate": 2.299624770585842e-06, + "loss": 0.1029, + "step": 15023 + }, + { + "epoch": 6.29, + "learning_rate": 2.2985432127701945e-06, + "loss": 0.1103, + "step": 15024 + }, + { + "epoch": 6.29, + "learning_rate": 2.2974618763260927e-06, + "loss": 0.0929, + "step": 15025 + }, + { + "epoch": 6.29, + "learning_rate": 2.29638076128461e-06, + "loss": 0.0851, + "step": 15026 + }, + { + "epoch": 6.29, + "learning_rate": 2.2952998676768267e-06, + "loss": 0.0539, + "step": 15027 + }, + { + "epoch": 6.29, + "learning_rate": 2.294219195533811e-06, + "loss": 0.1102, + "step": 15028 + }, + { + "epoch": 6.29, + "learning_rate": 2.293138744886627e-06, + "loss": 0.0641, + "step": 15029 + }, + { + "epoch": 6.29, + "learning_rate": 2.292058515766331e-06, + "loss": 0.0838, + "step": 15030 + }, + { + "epoch": 6.29, + "learning_rate": 2.290978508203976e-06, + "loss": 0.0982, + "step": 15031 + }, + { + "epoch": 6.29, + "learning_rate": 2.2898987222306e-06, + "loss": 0.0812, + "step": 15032 + }, + { + "epoch": 6.29, + "learning_rate": 2.2888191578772436e-06, + "loss": 0.0796, + "step": 15033 + }, + { + "epoch": 6.29, + "learning_rate": 2.2877398151749375e-06, + "loss": 0.0761, + "step": 15034 + }, + { + "epoch": 6.29, + "learning_rate": 2.2866606941547063e-06, + "loss": 0.0802, + "step": 15035 + }, + { + "epoch": 6.3, + "learning_rate": 2.285581794847569e-06, + "loss": 0.0624, + "step": 15036 + }, + { + "epoch": 6.3, + "learning_rate": 2.284503117284538e-06, + "loss": 0.0575, + "step": 15037 + }, + { + "epoch": 6.3, + "learning_rate": 2.2834246614966192e-06, + "loss": 0.0717, + "step": 15038 + }, + { + "epoch": 6.3, + "learning_rate": 2.282346427514813e-06, + "loss": 0.0925, + "step": 15039 + }, + { + "epoch": 6.3, + "learning_rate": 2.281268415370107e-06, + "loss": 0.1024, + "step": 15040 + }, + { + "epoch": 6.3, + "learning_rate": 2.2801906250934904e-06, + "loss": 0.0787, + "step": 15041 + }, + { + "epoch": 6.3, + "learning_rate": 2.2791130567159434e-06, + "loss": 0.0818, + "step": 15042 + }, + { + "epoch": 6.3, + "learning_rate": 2.2780357102684415e-06, + "loss": 0.0743, + "step": 15043 + }, + { + "epoch": 6.3, + "learning_rate": 2.27695858578195e-06, + "loss": 0.0911, + "step": 15044 + }, + { + "epoch": 6.3, + "learning_rate": 2.2758816832874333e-06, + "loss": 0.0787, + "step": 15045 + }, + { + "epoch": 6.3, + "learning_rate": 2.274805002815841e-06, + "loss": 0.0562, + "step": 15046 + }, + { + "epoch": 6.3, + "learning_rate": 2.2737285443981226e-06, + "loss": 0.0959, + "step": 15047 + }, + { + "epoch": 6.3, + "learning_rate": 2.272652308065222e-06, + "loss": 0.0979, + "step": 15048 + }, + { + "epoch": 6.3, + "learning_rate": 2.2715762938480734e-06, + "loss": 0.0647, + "step": 15049 + }, + { + "epoch": 6.3, + "learning_rate": 2.2705005017776073e-06, + "loss": 0.0847, + "step": 15050 + }, + { + "epoch": 6.3, + "learning_rate": 2.269424931884744e-06, + "loss": 0.0836, + "step": 15051 + }, + { + "epoch": 6.3, + "learning_rate": 2.268349584200403e-06, + "loss": 0.0797, + "step": 15052 + }, + { + "epoch": 6.3, + "learning_rate": 2.2672744587554917e-06, + "loss": 0.0747, + "step": 15053 + }, + { + "epoch": 6.3, + "learning_rate": 2.2661995555809123e-06, + "loss": 0.0945, + "step": 15054 + }, + { + "epoch": 6.3, + "learning_rate": 2.265124874707565e-06, + "loss": 0.0886, + "step": 15055 + }, + { + "epoch": 6.3, + "learning_rate": 2.264050416166339e-06, + "loss": 0.1056, + "step": 15056 + }, + { + "epoch": 6.3, + "learning_rate": 2.262976179988119e-06, + "loss": 0.0627, + "step": 15057 + }, + { + "epoch": 6.3, + "learning_rate": 2.2619021662037855e-06, + "loss": 0.0828, + "step": 15058 + }, + { + "epoch": 6.3, + "learning_rate": 2.260828374844205e-06, + "loss": 0.0694, + "step": 15059 + }, + { + "epoch": 6.31, + "learning_rate": 2.2597548059402453e-06, + "loss": 0.0854, + "step": 15060 + }, + { + "epoch": 6.31, + "learning_rate": 2.258681459522768e-06, + "loss": 0.0818, + "step": 15061 + }, + { + "epoch": 6.31, + "learning_rate": 2.2576083356226197e-06, + "loss": 0.0873, + "step": 15062 + }, + { + "epoch": 6.31, + "learning_rate": 2.256535434270649e-06, + "loss": 0.0569, + "step": 15063 + }, + { + "epoch": 6.31, + "learning_rate": 2.255462755497696e-06, + "loss": 0.06, + "step": 15064 + }, + { + "epoch": 6.31, + "learning_rate": 2.2543902993345946e-06, + "loss": 0.0988, + "step": 15065 + }, + { + "epoch": 6.31, + "learning_rate": 2.2533180658121723e-06, + "loss": 0.0793, + "step": 15066 + }, + { + "epoch": 6.31, + "learning_rate": 2.252246054961246e-06, + "loss": 0.0841, + "step": 15067 + }, + { + "epoch": 6.31, + "learning_rate": 2.25117426681263e-06, + "loss": 0.0698, + "step": 15068 + }, + { + "epoch": 6.31, + "learning_rate": 2.2501027013971354e-06, + "loss": 0.1104, + "step": 15069 + }, + { + "epoch": 6.31, + "learning_rate": 2.2490313587455605e-06, + "loss": 0.0768, + "step": 15070 + }, + { + "epoch": 6.31, + "learning_rate": 2.2479602388887013e-06, + "loss": 0.0616, + "step": 15071 + }, + { + "epoch": 6.31, + "learning_rate": 2.2468893418573488e-06, + "loss": 0.0998, + "step": 15072 + }, + { + "epoch": 6.31, + "learning_rate": 2.24581866768228e-06, + "loss": 0.0846, + "step": 15073 + }, + { + "epoch": 6.31, + "learning_rate": 2.2447482163942725e-06, + "loss": 0.0625, + "step": 15074 + }, + { + "epoch": 6.31, + "learning_rate": 2.2436779880240943e-06, + "loss": 0.0746, + "step": 15075 + }, + { + "epoch": 6.31, + "learning_rate": 2.2426079826025118e-06, + "loss": 0.0756, + "step": 15076 + }, + { + "epoch": 6.31, + "learning_rate": 2.241538200160277e-06, + "loss": 0.0823, + "step": 15077 + }, + { + "epoch": 6.31, + "learning_rate": 2.2404686407281427e-06, + "loss": 0.0996, + "step": 15078 + }, + { + "epoch": 6.31, + "learning_rate": 2.2393993043368545e-06, + "loss": 0.0844, + "step": 15079 + }, + { + "epoch": 6.31, + "learning_rate": 2.238330191017144e-06, + "loss": 0.1074, + "step": 15080 + }, + { + "epoch": 6.31, + "learning_rate": 2.2372613007997435e-06, + "loss": 0.0777, + "step": 15081 + }, + { + "epoch": 6.31, + "learning_rate": 2.2361926337153796e-06, + "loss": 0.0737, + "step": 15082 + }, + { + "epoch": 6.31, + "learning_rate": 2.2351241897947684e-06, + "loss": 0.0865, + "step": 15083 + }, + { + "epoch": 6.32, + "learning_rate": 2.2340559690686247e-06, + "loss": 0.094, + "step": 15084 + }, + { + "epoch": 6.32, + "learning_rate": 2.232987971567648e-06, + "loss": 0.0923, + "step": 15085 + }, + { + "epoch": 6.32, + "learning_rate": 2.2319201973225434e-06, + "loss": 0.0568, + "step": 15086 + }, + { + "epoch": 6.32, + "learning_rate": 2.2308526463639955e-06, + "loss": 0.1045, + "step": 15087 + }, + { + "epoch": 6.32, + "learning_rate": 2.2297853187226946e-06, + "loss": 0.0798, + "step": 15088 + }, + { + "epoch": 6.32, + "learning_rate": 2.2287182144293196e-06, + "loss": 0.0816, + "step": 15089 + }, + { + "epoch": 6.32, + "learning_rate": 2.227651333514543e-06, + "loss": 0.0923, + "step": 15090 + }, + { + "epoch": 6.32, + "learning_rate": 2.2265846760090326e-06, + "loss": 0.0729, + "step": 15091 + }, + { + "epoch": 6.32, + "learning_rate": 2.225518241943446e-06, + "loss": 0.0955, + "step": 15092 + }, + { + "epoch": 6.32, + "learning_rate": 2.224452031348443e-06, + "loss": 0.0795, + "step": 15093 + }, + { + "epoch": 6.32, + "learning_rate": 2.223386044254663e-06, + "loss": 0.0862, + "step": 15094 + }, + { + "epoch": 6.32, + "learning_rate": 2.222320280692751e-06, + "loss": 0.0762, + "step": 15095 + }, + { + "epoch": 6.32, + "learning_rate": 2.2212547406933405e-06, + "loss": 0.0771, + "step": 15096 + }, + { + "epoch": 6.32, + "learning_rate": 2.220189424287059e-06, + "loss": 0.0832, + "step": 15097 + }, + { + "epoch": 6.32, + "learning_rate": 2.2191243315045296e-06, + "loss": 0.0759, + "step": 15098 + }, + { + "epoch": 6.32, + "learning_rate": 2.2180594623763686e-06, + "loss": 0.066, + "step": 15099 + }, + { + "epoch": 6.32, + "learning_rate": 2.2169948169331812e-06, + "loss": 0.0649, + "step": 15100 + }, + { + "epoch": 6.32, + "learning_rate": 2.21593039520557e-06, + "loss": 0.0836, + "step": 15101 + }, + { + "epoch": 6.32, + "learning_rate": 2.2148661972241324e-06, + "loss": 0.0737, + "step": 15102 + }, + { + "epoch": 6.32, + "learning_rate": 2.213802223019459e-06, + "loss": 0.0903, + "step": 15103 + }, + { + "epoch": 6.32, + "learning_rate": 2.2127384726221303e-06, + "loss": 0.0921, + "step": 15104 + }, + { + "epoch": 6.32, + "learning_rate": 2.211674946062726e-06, + "loss": 0.0945, + "step": 15105 + }, + { + "epoch": 6.32, + "learning_rate": 2.2106116433718115e-06, + "loss": 0.0758, + "step": 15106 + }, + { + "epoch": 6.32, + "learning_rate": 2.2095485645799564e-06, + "loss": 0.0526, + "step": 15107 + }, + { + "epoch": 6.33, + "learning_rate": 2.2084857097177125e-06, + "loss": 0.1023, + "step": 15108 + }, + { + "epoch": 6.33, + "learning_rate": 2.2074230788156325e-06, + "loss": 0.0932, + "step": 15109 + }, + { + "epoch": 6.33, + "learning_rate": 2.2063606719042605e-06, + "loss": 0.0918, + "step": 15110 + }, + { + "epoch": 6.33, + "learning_rate": 2.205298489014135e-06, + "loss": 0.0743, + "step": 15111 + }, + { + "epoch": 6.33, + "learning_rate": 2.2042365301757883e-06, + "loss": 0.0802, + "step": 15112 + }, + { + "epoch": 6.33, + "learning_rate": 2.203174795419747e-06, + "loss": 0.105, + "step": 15113 + }, + { + "epoch": 6.33, + "learning_rate": 2.2021132847765246e-06, + "loss": 0.1102, + "step": 15114 + }, + { + "epoch": 6.33, + "learning_rate": 2.2010519982766364e-06, + "loss": 0.0819, + "step": 15115 + }, + { + "epoch": 6.33, + "learning_rate": 2.1999909359505878e-06, + "loss": 0.0604, + "step": 15116 + }, + { + "epoch": 6.33, + "learning_rate": 2.1989300978288773e-06, + "loss": 0.0653, + "step": 15117 + }, + { + "epoch": 6.33, + "learning_rate": 2.197869483942e-06, + "loss": 0.0892, + "step": 15118 + }, + { + "epoch": 6.33, + "learning_rate": 2.1968090943204402e-06, + "loss": 0.0818, + "step": 15119 + }, + { + "epoch": 6.33, + "learning_rate": 2.195748928994681e-06, + "loss": 0.1011, + "step": 15120 + }, + { + "epoch": 6.33, + "learning_rate": 2.1946889879951903e-06, + "loss": 0.1059, + "step": 15121 + }, + { + "epoch": 6.33, + "learning_rate": 2.19362927135244e-06, + "loss": 0.0953, + "step": 15122 + }, + { + "epoch": 6.33, + "learning_rate": 2.192569779096888e-06, + "loss": 0.0665, + "step": 15123 + }, + { + "epoch": 6.33, + "learning_rate": 2.1915105112589894e-06, + "loss": 0.0824, + "step": 15124 + }, + { + "epoch": 6.33, + "learning_rate": 2.1904514678691924e-06, + "loss": 0.1006, + "step": 15125 + }, + { + "epoch": 6.33, + "learning_rate": 2.18939264895794e-06, + "loss": 0.0923, + "step": 15126 + }, + { + "epoch": 6.33, + "learning_rate": 2.188334054555663e-06, + "loss": 0.1008, + "step": 15127 + }, + { + "epoch": 6.33, + "learning_rate": 2.1872756846927913e-06, + "loss": 0.1174, + "step": 15128 + }, + { + "epoch": 6.33, + "learning_rate": 2.1862175393997497e-06, + "loss": 0.1017, + "step": 15129 + }, + { + "epoch": 6.33, + "learning_rate": 2.1851596187069478e-06, + "loss": 0.0738, + "step": 15130 + }, + { + "epoch": 6.33, + "learning_rate": 2.184101922644799e-06, + "loss": 0.086, + "step": 15131 + }, + { + "epoch": 6.34, + "learning_rate": 2.183044451243704e-06, + "loss": 0.0789, + "step": 15132 + }, + { + "epoch": 6.34, + "learning_rate": 2.181987204534061e-06, + "loss": 0.1091, + "step": 15133 + }, + { + "epoch": 6.34, + "learning_rate": 2.1809301825462593e-06, + "loss": 0.0802, + "step": 15134 + }, + { + "epoch": 6.34, + "learning_rate": 2.179873385310679e-06, + "loss": 0.1005, + "step": 15135 + }, + { + "epoch": 6.34, + "learning_rate": 2.178816812857698e-06, + "loss": 0.0425, + "step": 15136 + }, + { + "epoch": 6.34, + "learning_rate": 2.1777604652176885e-06, + "loss": 0.0622, + "step": 15137 + }, + { + "epoch": 6.34, + "learning_rate": 2.176704342421012e-06, + "loss": 0.0391, + "step": 15138 + }, + { + "epoch": 6.34, + "learning_rate": 2.1756484444980274e-06, + "loss": 0.0774, + "step": 15139 + }, + { + "epoch": 6.34, + "learning_rate": 2.1745927714790882e-06, + "loss": 0.0738, + "step": 15140 + }, + { + "epoch": 6.34, + "learning_rate": 2.1735373233945323e-06, + "loss": 0.079, + "step": 15141 + }, + { + "epoch": 6.34, + "learning_rate": 2.172482100274702e-06, + "loss": 0.0918, + "step": 15142 + }, + { + "epoch": 6.34, + "learning_rate": 2.171427102149927e-06, + "loss": 0.0805, + "step": 15143 + }, + { + "epoch": 6.34, + "learning_rate": 2.170372329050533e-06, + "loss": 0.057, + "step": 15144 + }, + { + "epoch": 6.34, + "learning_rate": 2.169317781006838e-06, + "loss": 0.0869, + "step": 15145 + }, + { + "epoch": 6.34, + "learning_rate": 2.168263458049158e-06, + "loss": 0.0805, + "step": 15146 + }, + { + "epoch": 6.34, + "learning_rate": 2.1672093602077914e-06, + "loss": 0.0829, + "step": 15147 + }, + { + "epoch": 6.34, + "learning_rate": 2.166155487513042e-06, + "loss": 0.0954, + "step": 15148 + }, + { + "epoch": 6.34, + "learning_rate": 2.1651018399952017e-06, + "loss": 0.0794, + "step": 15149 + }, + { + "epoch": 6.34, + "learning_rate": 2.1640484176845557e-06, + "loss": 0.1011, + "step": 15150 + }, + { + "epoch": 6.34, + "learning_rate": 2.1629952206113845e-06, + "loss": 0.0983, + "step": 15151 + }, + { + "epoch": 6.34, + "learning_rate": 2.161942248805965e-06, + "loss": 0.099, + "step": 15152 + }, + { + "epoch": 6.34, + "learning_rate": 2.1608895022985564e-06, + "loss": 0.0934, + "step": 15153 + }, + { + "epoch": 6.34, + "learning_rate": 2.159836981119425e-06, + "loss": 0.1135, + "step": 15154 + }, + { + "epoch": 6.34, + "learning_rate": 2.1587846852988203e-06, + "loss": 0.0888, + "step": 15155 + }, + { + "epoch": 6.35, + "learning_rate": 2.157732614866992e-06, + "loss": 0.0539, + "step": 15156 + }, + { + "epoch": 6.35, + "learning_rate": 2.1566807698541803e-06, + "loss": 0.0841, + "step": 15157 + }, + { + "epoch": 6.35, + "learning_rate": 2.15562915029062e-06, + "loss": 0.0645, + "step": 15158 + }, + { + "epoch": 6.35, + "learning_rate": 2.1545777562065383e-06, + "loss": 0.0846, + "step": 15159 + }, + { + "epoch": 6.35, + "learning_rate": 2.1535265876321574e-06, + "loss": 0.0775, + "step": 15160 + }, + { + "epoch": 6.35, + "learning_rate": 2.1524756445976947e-06, + "loss": 0.0767, + "step": 15161 + }, + { + "epoch": 6.35, + "learning_rate": 2.1514249271333532e-06, + "loss": 0.0852, + "step": 15162 + }, + { + "epoch": 6.35, + "learning_rate": 2.1503744352693377e-06, + "loss": 0.0796, + "step": 15163 + }, + { + "epoch": 6.35, + "learning_rate": 2.1493241690358437e-06, + "loss": 0.0639, + "step": 15164 + }, + { + "epoch": 6.35, + "learning_rate": 2.148274128463058e-06, + "loss": 0.0731, + "step": 15165 + }, + { + "epoch": 6.35, + "learning_rate": 2.1472243135811666e-06, + "loss": 0.0772, + "step": 15166 + }, + { + "epoch": 6.35, + "learning_rate": 2.146174724420347e-06, + "loss": 0.0946, + "step": 15167 + }, + { + "epoch": 6.35, + "learning_rate": 2.1451253610107616e-06, + "loss": 0.0783, + "step": 15168 + }, + { + "epoch": 6.35, + "learning_rate": 2.1440762233825774e-06, + "loss": 0.0823, + "step": 15169 + }, + { + "epoch": 6.35, + "learning_rate": 2.1430273115659516e-06, + "loss": 0.0751, + "step": 15170 + }, + { + "epoch": 6.35, + "learning_rate": 2.1419786255910336e-06, + "loss": 0.0774, + "step": 15171 + }, + { + "epoch": 6.35, + "learning_rate": 2.1409301654879665e-06, + "loss": 0.1268, + "step": 15172 + }, + { + "epoch": 6.35, + "learning_rate": 2.139881931286891e-06, + "loss": 0.1119, + "step": 15173 + }, + { + "epoch": 6.35, + "learning_rate": 2.138833923017931e-06, + "loss": 0.092, + "step": 15174 + }, + { + "epoch": 6.35, + "learning_rate": 2.1377861407112178e-06, + "loss": 0.09, + "step": 15175 + }, + { + "epoch": 6.35, + "learning_rate": 2.136738584396861e-06, + "loss": 0.0835, + "step": 15176 + }, + { + "epoch": 6.35, + "learning_rate": 2.135691254104977e-06, + "loss": 0.1081, + "step": 15177 + }, + { + "epoch": 6.35, + "learning_rate": 2.1346441498656688e-06, + "loss": 0.0438, + "step": 15178 + }, + { + "epoch": 6.36, + "learning_rate": 2.1335972717090358e-06, + "loss": 0.068, + "step": 15179 + }, + { + "epoch": 6.36, + "learning_rate": 2.132550619665168e-06, + "loss": 0.0942, + "step": 15180 + }, + { + "epoch": 6.36, + "learning_rate": 2.1315041937641534e-06, + "loss": 0.0728, + "step": 15181 + }, + { + "epoch": 6.36, + "learning_rate": 2.130457994036066e-06, + "loss": 0.1139, + "step": 15182 + }, + { + "epoch": 6.36, + "learning_rate": 2.12941202051098e-06, + "loss": 0.0853, + "step": 15183 + }, + { + "epoch": 6.36, + "learning_rate": 2.1283662732189624e-06, + "loss": 0.092, + "step": 15184 + }, + { + "epoch": 6.36, + "learning_rate": 2.12732075219007e-06, + "loss": 0.0503, + "step": 15185 + }, + { + "epoch": 6.36, + "learning_rate": 2.1262754574543566e-06, + "loss": 0.0582, + "step": 15186 + }, + { + "epoch": 6.36, + "learning_rate": 2.1252303890418714e-06, + "loss": 0.0838, + "step": 15187 + }, + { + "epoch": 6.36, + "learning_rate": 2.124185546982648e-06, + "loss": 0.0767, + "step": 15188 + }, + { + "epoch": 6.36, + "learning_rate": 2.123140931306722e-06, + "loss": 0.08, + "step": 15189 + }, + { + "epoch": 6.36, + "learning_rate": 2.1220965420441198e-06, + "loss": 0.0973, + "step": 15190 + }, + { + "epoch": 6.36, + "learning_rate": 2.1210523792248615e-06, + "loss": 0.0849, + "step": 15191 + }, + { + "epoch": 6.36, + "learning_rate": 2.1200084428789614e-06, + "loss": 0.0809, + "step": 15192 + }, + { + "epoch": 6.36, + "learning_rate": 2.118964733036426e-06, + "loss": 0.1049, + "step": 15193 + }, + { + "epoch": 6.36, + "learning_rate": 2.1179212497272582e-06, + "loss": 0.076, + "step": 15194 + }, + { + "epoch": 6.36, + "learning_rate": 2.1168779929814476e-06, + "loss": 0.0701, + "step": 15195 + }, + { + "epoch": 6.36, + "learning_rate": 2.1158349628289832e-06, + "loss": 0.0752, + "step": 15196 + }, + { + "epoch": 6.36, + "learning_rate": 2.1147921592998498e-06, + "loss": 0.0687, + "step": 15197 + }, + { + "epoch": 6.36, + "learning_rate": 2.1137495824240152e-06, + "loss": 0.0641, + "step": 15198 + }, + { + "epoch": 6.36, + "learning_rate": 2.1127072322314514e-06, + "loss": 0.0852, + "step": 15199 + }, + { + "epoch": 6.36, + "learning_rate": 2.1116651087521188e-06, + "loss": 0.0573, + "step": 15200 + }, + { + "epoch": 6.36, + "learning_rate": 2.110623212015972e-06, + "loss": 0.0811, + "step": 15201 + }, + { + "epoch": 6.36, + "learning_rate": 2.109581542052964e-06, + "loss": 0.0704, + "step": 15202 + }, + { + "epoch": 6.37, + "learning_rate": 2.108540098893029e-06, + "loss": 0.074, + "step": 15203 + }, + { + "epoch": 6.37, + "learning_rate": 2.107498882566107e-06, + "loss": 0.0836, + "step": 15204 + }, + { + "epoch": 6.37, + "learning_rate": 2.106457893102125e-06, + "loss": 0.0861, + "step": 15205 + }, + { + "epoch": 6.37, + "learning_rate": 2.1054171305310065e-06, + "loss": 0.0707, + "step": 15206 + }, + { + "epoch": 6.37, + "learning_rate": 2.1043765948826678e-06, + "loss": 0.0875, + "step": 15207 + }, + { + "epoch": 6.37, + "learning_rate": 2.1033362861870198e-06, + "loss": 0.083, + "step": 15208 + }, + { + "epoch": 6.37, + "learning_rate": 2.1022962044739602e-06, + "loss": 0.0617, + "step": 15209 + }, + { + "epoch": 6.37, + "learning_rate": 2.1012563497733873e-06, + "loss": 0.0751, + "step": 15210 + }, + { + "epoch": 6.37, + "learning_rate": 2.1002167221151915e-06, + "loss": 0.0753, + "step": 15211 + }, + { + "epoch": 6.37, + "learning_rate": 2.0991773215292566e-06, + "loss": 0.0736, + "step": 15212 + }, + { + "epoch": 6.37, + "learning_rate": 2.0981381480454575e-06, + "loss": 0.0571, + "step": 15213 + }, + { + "epoch": 6.37, + "learning_rate": 2.0970992016936685e-06, + "loss": 0.057, + "step": 15214 + }, + { + "epoch": 6.37, + "learning_rate": 2.096060482503748e-06, + "loss": 0.0662, + "step": 15215 + }, + { + "epoch": 6.37, + "learning_rate": 2.0950219905055548e-06, + "loss": 0.1205, + "step": 15216 + }, + { + "epoch": 6.37, + "learning_rate": 2.0939837257289395e-06, + "loss": 0.0759, + "step": 15217 + }, + { + "epoch": 6.37, + "learning_rate": 2.0929456882037458e-06, + "loss": 0.0748, + "step": 15218 + }, + { + "epoch": 6.37, + "learning_rate": 2.0919078779598123e-06, + "loss": 0.0819, + "step": 15219 + }, + { + "epoch": 6.37, + "learning_rate": 2.090870295026972e-06, + "loss": 0.0789, + "step": 15220 + }, + { + "epoch": 6.37, + "learning_rate": 2.0898329394350446e-06, + "loss": 0.0746, + "step": 15221 + }, + { + "epoch": 6.37, + "learning_rate": 2.0887958112138516e-06, + "loss": 0.0714, + "step": 15222 + }, + { + "epoch": 6.37, + "learning_rate": 2.0877589103932007e-06, + "loss": 0.0954, + "step": 15223 + }, + { + "epoch": 6.37, + "learning_rate": 2.0867222370028985e-06, + "loss": 0.1011, + "step": 15224 + }, + { + "epoch": 6.37, + "learning_rate": 2.0856857910727436e-06, + "loss": 0.0701, + "step": 15225 + }, + { + "epoch": 6.37, + "learning_rate": 2.0846495726325267e-06, + "loss": 0.1062, + "step": 15226 + }, + { + "epoch": 6.38, + "learning_rate": 2.0836135817120352e-06, + "loss": 0.0459, + "step": 15227 + }, + { + "epoch": 6.38, + "learning_rate": 2.0825778183410485e-06, + "loss": 0.0785, + "step": 15228 + }, + { + "epoch": 6.38, + "learning_rate": 2.0815422825493336e-06, + "loss": 0.1151, + "step": 15229 + }, + { + "epoch": 6.38, + "learning_rate": 2.0805069743666593e-06, + "loss": 0.0823, + "step": 15230 + }, + { + "epoch": 6.38, + "learning_rate": 2.0794718938227855e-06, + "loss": 0.0946, + "step": 15231 + }, + { + "epoch": 6.38, + "learning_rate": 2.078437040947462e-06, + "loss": 0.0926, + "step": 15232 + }, + { + "epoch": 6.38, + "learning_rate": 2.0774024157704364e-06, + "loss": 0.0794, + "step": 15233 + }, + { + "epoch": 6.38, + "learning_rate": 2.0763680183214485e-06, + "loss": 0.1051, + "step": 15234 + }, + { + "epoch": 6.38, + "learning_rate": 2.0753338486302334e-06, + "loss": 0.0725, + "step": 15235 + }, + { + "epoch": 6.38, + "learning_rate": 2.0742999067265125e-06, + "loss": 0.091, + "step": 15236 + }, + { + "epoch": 6.38, + "learning_rate": 2.0732661926400065e-06, + "loss": 0.0525, + "step": 15237 + }, + { + "epoch": 6.38, + "learning_rate": 2.07223270640043e-06, + "loss": 0.0917, + "step": 15238 + }, + { + "epoch": 6.38, + "learning_rate": 2.0711994480374895e-06, + "loss": 0.0621, + "step": 15239 + }, + { + "epoch": 6.38, + "learning_rate": 2.0701664175808843e-06, + "loss": 0.0761, + "step": 15240 + }, + { + "epoch": 6.38, + "learning_rate": 2.0691336150603124e-06, + "loss": 0.1114, + "step": 15241 + }, + { + "epoch": 6.38, + "learning_rate": 2.0681010405054535e-06, + "loss": 0.0802, + "step": 15242 + }, + { + "epoch": 6.38, + "learning_rate": 2.067068693945995e-06, + "loss": 0.1087, + "step": 15243 + }, + { + "epoch": 6.38, + "learning_rate": 2.0660365754116042e-06, + "loss": 0.0917, + "step": 15244 + }, + { + "epoch": 6.38, + "learning_rate": 2.0650046849319516e-06, + "loss": 0.0681, + "step": 15245 + }, + { + "epoch": 6.38, + "learning_rate": 2.063973022536698e-06, + "loss": 0.0731, + "step": 15246 + }, + { + "epoch": 6.38, + "learning_rate": 2.062941588255498e-06, + "loss": 0.081, + "step": 15247 + }, + { + "epoch": 6.38, + "learning_rate": 2.0619103821179975e-06, + "loss": 0.0566, + "step": 15248 + }, + { + "epoch": 6.38, + "learning_rate": 2.0608794041538426e-06, + "loss": 0.0903, + "step": 15249 + }, + { + "epoch": 6.38, + "learning_rate": 2.0598486543926623e-06, + "loss": 0.1197, + "step": 15250 + }, + { + "epoch": 6.39, + "learning_rate": 2.0588181328640856e-06, + "loss": 0.0987, + "step": 15251 + }, + { + "epoch": 6.39, + "learning_rate": 2.0577878395977347e-06, + "loss": 0.0553, + "step": 15252 + }, + { + "epoch": 6.39, + "learning_rate": 2.056757774623224e-06, + "loss": 0.077, + "step": 15253 + }, + { + "epoch": 6.39, + "learning_rate": 2.0557279379701632e-06, + "loss": 0.0711, + "step": 15254 + }, + { + "epoch": 6.39, + "learning_rate": 2.0546983296681554e-06, + "loss": 0.0822, + "step": 15255 + }, + { + "epoch": 6.39, + "learning_rate": 2.053668949746791e-06, + "loss": 0.1005, + "step": 15256 + }, + { + "epoch": 6.39, + "learning_rate": 2.0526397982356617e-06, + "loss": 0.0839, + "step": 15257 + }, + { + "epoch": 6.39, + "learning_rate": 2.0516108751643483e-06, + "loss": 0.1148, + "step": 15258 + }, + { + "epoch": 6.39, + "learning_rate": 2.050582180562427e-06, + "loss": 0.1279, + "step": 15259 + }, + { + "epoch": 6.39, + "learning_rate": 2.0495537144594678e-06, + "loss": 0.0718, + "step": 15260 + }, + { + "epoch": 6.39, + "learning_rate": 2.0485254768850317e-06, + "loss": 0.0816, + "step": 15261 + }, + { + "epoch": 6.39, + "learning_rate": 2.0474974678686764e-06, + "loss": 0.0784, + "step": 15262 + }, + { + "epoch": 6.39, + "learning_rate": 2.046469687439947e-06, + "loss": 0.075, + "step": 15263 + }, + { + "epoch": 6.39, + "learning_rate": 2.045442135628389e-06, + "loss": 0.0981, + "step": 15264 + }, + { + "epoch": 6.39, + "learning_rate": 2.044414812463541e-06, + "loss": 0.0727, + "step": 15265 + }, + { + "epoch": 6.39, + "learning_rate": 2.0433877179749273e-06, + "loss": 0.0525, + "step": 15266 + }, + { + "epoch": 6.39, + "learning_rate": 2.0423608521920723e-06, + "loss": 0.0993, + "step": 15267 + }, + { + "epoch": 6.39, + "learning_rate": 2.0413342151444928e-06, + "loss": 0.1002, + "step": 15268 + }, + { + "epoch": 6.39, + "learning_rate": 2.040307806861702e-06, + "loss": 0.065, + "step": 15269 + }, + { + "epoch": 6.39, + "learning_rate": 2.0392816273731973e-06, + "loss": 0.056, + "step": 15270 + }, + { + "epoch": 6.39, + "learning_rate": 2.0382556767084784e-06, + "loss": 0.0821, + "step": 15271 + }, + { + "epoch": 6.39, + "learning_rate": 2.0372299548970343e-06, + "loss": 0.0702, + "step": 15272 + }, + { + "epoch": 6.39, + "learning_rate": 2.0362044619683485e-06, + "loss": 0.1083, + "step": 15273 + }, + { + "epoch": 6.39, + "learning_rate": 2.0351791979518997e-06, + "loss": 0.0707, + "step": 15274 + }, + { + "epoch": 6.4, + "learning_rate": 2.0341541628771567e-06, + "loss": 0.0821, + "step": 15275 + }, + { + "epoch": 6.4, + "learning_rate": 2.0331293567735865e-06, + "loss": 0.055, + "step": 15276 + }, + { + "epoch": 6.4, + "learning_rate": 2.0321047796706393e-06, + "loss": 0.0862, + "step": 15277 + }, + { + "epoch": 6.4, + "learning_rate": 2.0310804315977717e-06, + "loss": 0.1026, + "step": 15278 + }, + { + "epoch": 6.4, + "learning_rate": 2.030056312584424e-06, + "loss": 0.0964, + "step": 15279 + }, + { + "epoch": 6.4, + "learning_rate": 2.0290324226600357e-06, + "loss": 0.104, + "step": 15280 + }, + { + "epoch": 6.4, + "learning_rate": 2.028008761854037e-06, + "loss": 0.101, + "step": 15281 + }, + { + "epoch": 6.4, + "learning_rate": 2.0269853301958554e-06, + "loss": 0.0903, + "step": 15282 + }, + { + "epoch": 6.4, + "learning_rate": 2.0259621277149023e-06, + "loss": 0.077, + "step": 15283 + }, + { + "epoch": 6.4, + "learning_rate": 2.0249391544405916e-06, + "loss": 0.1145, + "step": 15284 + }, + { + "epoch": 6.4, + "learning_rate": 2.0239164104023278e-06, + "loss": 0.0936, + "step": 15285 + }, + { + "epoch": 6.4, + "learning_rate": 2.0228938956295087e-06, + "loss": 0.0487, + "step": 15286 + }, + { + "epoch": 6.4, + "learning_rate": 2.0218716101515268e-06, + "loss": 0.1228, + "step": 15287 + }, + { + "epoch": 6.4, + "learning_rate": 2.020849553997768e-06, + "loss": 0.1073, + "step": 15288 + }, + { + "epoch": 6.4, + "learning_rate": 2.019827727197605e-06, + "loss": 0.0721, + "step": 15289 + }, + { + "epoch": 6.4, + "learning_rate": 2.018806129780416e-06, + "loss": 0.1021, + "step": 15290 + }, + { + "epoch": 6.4, + "learning_rate": 2.0177847617755595e-06, + "loss": 0.0533, + "step": 15291 + }, + { + "epoch": 6.4, + "learning_rate": 2.0167636232123967e-06, + "loss": 0.0427, + "step": 15292 + }, + { + "epoch": 6.4, + "learning_rate": 2.015742714120279e-06, + "loss": 0.082, + "step": 15293 + }, + { + "epoch": 6.4, + "learning_rate": 2.0147220345285524e-06, + "loss": 0.0755, + "step": 15294 + }, + { + "epoch": 6.4, + "learning_rate": 2.013701584466555e-06, + "loss": 0.1105, + "step": 15295 + }, + { + "epoch": 6.4, + "learning_rate": 2.012681363963621e-06, + "loss": 0.0404, + "step": 15296 + }, + { + "epoch": 6.4, + "learning_rate": 2.011661373049071e-06, + "loss": 0.1166, + "step": 15297 + }, + { + "epoch": 6.4, + "learning_rate": 2.0106416117522255e-06, + "loss": 0.0799, + "step": 15298 + }, + { + "epoch": 6.41, + "learning_rate": 2.009622080102398e-06, + "loss": 0.09, + "step": 15299 + }, + { + "epoch": 6.41, + "learning_rate": 2.008602778128892e-06, + "loss": 0.0517, + "step": 15300 + }, + { + "epoch": 6.41, + "learning_rate": 2.007583705861009e-06, + "loss": 0.1018, + "step": 15301 + }, + { + "epoch": 6.41, + "learning_rate": 2.00656486332804e-06, + "loss": 0.0705, + "step": 15302 + }, + { + "epoch": 6.41, + "learning_rate": 2.0055462505592725e-06, + "loss": 0.0839, + "step": 15303 + }, + { + "epoch": 6.41, + "learning_rate": 2.004527867583982e-06, + "loss": 0.0649, + "step": 15304 + }, + { + "epoch": 6.41, + "learning_rate": 2.003509714431442e-06, + "loss": 0.0991, + "step": 15305 + }, + { + "epoch": 6.41, + "learning_rate": 2.0024917911309207e-06, + "loss": 0.0896, + "step": 15306 + }, + { + "epoch": 6.41, + "learning_rate": 2.0014740977116743e-06, + "loss": 0.0785, + "step": 15307 + }, + { + "epoch": 6.41, + "learning_rate": 2.0004566342029586e-06, + "loss": 0.0783, + "step": 15308 + }, + { + "epoch": 6.41, + "learning_rate": 1.99943940063402e-06, + "loss": 0.0617, + "step": 15309 + }, + { + "epoch": 6.41, + "learning_rate": 1.9984223970340965e-06, + "loss": 0.1016, + "step": 15310 + }, + { + "epoch": 6.41, + "learning_rate": 1.997405623432419e-06, + "loss": 0.0764, + "step": 15311 + }, + { + "epoch": 6.41, + "learning_rate": 1.996389079858214e-06, + "loss": 0.0764, + "step": 15312 + }, + { + "epoch": 6.41, + "learning_rate": 1.9953727663407043e-06, + "loss": 0.0679, + "step": 15313 + }, + { + "epoch": 6.41, + "learning_rate": 1.9943566829091e-06, + "loss": 0.0698, + "step": 15314 + }, + { + "epoch": 6.41, + "learning_rate": 1.9933408295926095e-06, + "loss": 0.0677, + "step": 15315 + }, + { + "epoch": 6.41, + "learning_rate": 1.9923252064204313e-06, + "loss": 0.0525, + "step": 15316 + }, + { + "epoch": 6.41, + "learning_rate": 1.9913098134217612e-06, + "loss": 0.0881, + "step": 15317 + }, + { + "epoch": 6.41, + "learning_rate": 1.9902946506257825e-06, + "loss": 0.0696, + "step": 15318 + }, + { + "epoch": 6.41, + "learning_rate": 1.989279718061675e-06, + "loss": 0.095, + "step": 15319 + }, + { + "epoch": 6.41, + "learning_rate": 1.988265015758615e-06, + "loss": 0.0556, + "step": 15320 + }, + { + "epoch": 6.41, + "learning_rate": 1.9872505437457666e-06, + "loss": 0.1042, + "step": 15321 + }, + { + "epoch": 6.41, + "learning_rate": 1.986236302052291e-06, + "loss": 0.0944, + "step": 15322 + }, + { + "epoch": 6.42, + "learning_rate": 1.985222290707344e-06, + "loss": 0.1087, + "step": 15323 + }, + { + "epoch": 6.42, + "learning_rate": 1.9842085097400678e-06, + "loss": 0.0571, + "step": 15324 + }, + { + "epoch": 6.42, + "learning_rate": 1.9831949591796052e-06, + "loss": 0.1036, + "step": 15325 + }, + { + "epoch": 6.42, + "learning_rate": 1.9821816390550887e-06, + "loss": 0.0771, + "step": 15326 + }, + { + "epoch": 6.42, + "learning_rate": 1.981168549395647e-06, + "loss": 0.1008, + "step": 15327 + }, + { + "epoch": 6.42, + "learning_rate": 1.9801556902303986e-06, + "loss": 0.0746, + "step": 15328 + }, + { + "epoch": 6.42, + "learning_rate": 1.9791430615884622e-06, + "loss": 0.087, + "step": 15329 + }, + { + "epoch": 6.42, + "learning_rate": 1.9781306634989373e-06, + "loss": 0.0939, + "step": 15330 + }, + { + "epoch": 6.42, + "learning_rate": 1.9771184959909273e-06, + "loss": 0.083, + "step": 15331 + }, + { + "epoch": 6.42, + "learning_rate": 1.976106559093528e-06, + "loss": 0.0843, + "step": 15332 + }, + { + "epoch": 6.42, + "learning_rate": 1.9750948528358283e-06, + "loss": 0.0991, + "step": 15333 + }, + { + "epoch": 6.42, + "learning_rate": 1.9740833772469025e-06, + "loss": 0.0816, + "step": 15334 + }, + { + "epoch": 6.42, + "learning_rate": 1.973072132355828e-06, + "loss": 0.0678, + "step": 15335 + }, + { + "epoch": 6.42, + "learning_rate": 1.9720611181916716e-06, + "loss": 0.0686, + "step": 15336 + }, + { + "epoch": 6.42, + "learning_rate": 1.9710503347834975e-06, + "loss": 0.0766, + "step": 15337 + }, + { + "epoch": 6.42, + "learning_rate": 1.9700397821603546e-06, + "loss": 0.1118, + "step": 15338 + }, + { + "epoch": 6.42, + "learning_rate": 1.9690294603512915e-06, + "loss": 0.0656, + "step": 15339 + }, + { + "epoch": 6.42, + "learning_rate": 1.968019369385351e-06, + "loss": 0.0869, + "step": 15340 + }, + { + "epoch": 6.42, + "learning_rate": 1.9670095092915666e-06, + "loss": 0.0868, + "step": 15341 + }, + { + "epoch": 6.42, + "learning_rate": 1.9659998800989642e-06, + "loss": 0.0864, + "step": 15342 + }, + { + "epoch": 6.42, + "learning_rate": 1.9649904818365673e-06, + "loss": 0.096, + "step": 15343 + }, + { + "epoch": 6.42, + "learning_rate": 1.963981314533391e-06, + "loss": 0.0787, + "step": 15344 + }, + { + "epoch": 6.42, + "learning_rate": 1.962972378218437e-06, + "loss": 0.0652, + "step": 15345 + }, + { + "epoch": 6.42, + "learning_rate": 1.9619636729207116e-06, + "loss": 0.0911, + "step": 15346 + }, + { + "epoch": 6.43, + "learning_rate": 1.9609551986692067e-06, + "loss": 0.0688, + "step": 15347 + }, + { + "epoch": 6.43, + "learning_rate": 1.9599469554929116e-06, + "loss": 0.0916, + "step": 15348 + }, + { + "epoch": 6.43, + "learning_rate": 1.9589389434208063e-06, + "loss": 0.1026, + "step": 15349 + }, + { + "epoch": 6.43, + "learning_rate": 1.9579311624818674e-06, + "loss": 0.1018, + "step": 15350 + }, + { + "epoch": 6.43, + "learning_rate": 1.9569236127050596e-06, + "loss": 0.0661, + "step": 15351 + }, + { + "epoch": 6.43, + "learning_rate": 1.955916294119344e-06, + "loss": 0.0718, + "step": 15352 + }, + { + "epoch": 6.43, + "learning_rate": 1.954909206753678e-06, + "loss": 0.0781, + "step": 15353 + }, + { + "epoch": 6.43, + "learning_rate": 1.953902350637006e-06, + "loss": 0.0836, + "step": 15354 + }, + { + "epoch": 6.43, + "learning_rate": 1.9528957257982752e-06, + "loss": 0.0657, + "step": 15355 + }, + { + "epoch": 6.43, + "learning_rate": 1.9518893322664124e-06, + "loss": 0.0784, + "step": 15356 + }, + { + "epoch": 6.43, + "learning_rate": 1.9508831700703477e-06, + "loss": 0.0803, + "step": 15357 + }, + { + "epoch": 6.43, + "learning_rate": 1.9498772392390084e-06, + "loss": 0.0647, + "step": 15358 + }, + { + "epoch": 6.43, + "learning_rate": 1.9488715398013002e-06, + "loss": 0.0873, + "step": 15359 + }, + { + "epoch": 6.43, + "learning_rate": 1.9478660717861362e-06, + "loss": 0.0837, + "step": 15360 + }, + { + "epoch": 6.43, + "learning_rate": 1.946860835222416e-06, + "loss": 0.0908, + "step": 15361 + }, + { + "epoch": 6.43, + "learning_rate": 1.9458558301390353e-06, + "loss": 0.0487, + "step": 15362 + }, + { + "epoch": 6.43, + "learning_rate": 1.9448510565648813e-06, + "loss": 0.1196, + "step": 15363 + }, + { + "epoch": 6.43, + "learning_rate": 1.9438465145288377e-06, + "loss": 0.0769, + "step": 15364 + }, + { + "epoch": 6.43, + "learning_rate": 1.942842204059775e-06, + "loss": 0.0767, + "step": 15365 + }, + { + "epoch": 6.43, + "learning_rate": 1.9418381251865636e-06, + "loss": 0.0949, + "step": 15366 + }, + { + "epoch": 6.43, + "learning_rate": 1.940834277938065e-06, + "loss": 0.0808, + "step": 15367 + }, + { + "epoch": 6.43, + "learning_rate": 1.9398306623431327e-06, + "loss": 0.1119, + "step": 15368 + }, + { + "epoch": 6.43, + "learning_rate": 1.9388272784306163e-06, + "loss": 0.0912, + "step": 15369 + }, + { + "epoch": 6.44, + "learning_rate": 1.9378241262293583e-06, + "loss": 0.0776, + "step": 15370 + }, + { + "epoch": 6.44, + "learning_rate": 1.936821205768189e-06, + "loss": 0.062, + "step": 15371 + }, + { + "epoch": 6.44, + "learning_rate": 1.9358185170759394e-06, + "loss": 0.0768, + "step": 15372 + }, + { + "epoch": 6.44, + "learning_rate": 1.93481606018143e-06, + "loss": 0.095, + "step": 15373 + }, + { + "epoch": 6.44, + "learning_rate": 1.9338138351134773e-06, + "loss": 0.0929, + "step": 15374 + }, + { + "epoch": 6.44, + "learning_rate": 1.932811841900887e-06, + "loss": 0.0751, + "step": 15375 + }, + { + "epoch": 6.44, + "learning_rate": 1.9318100805724617e-06, + "loss": 0.0926, + "step": 15376 + }, + { + "epoch": 6.44, + "learning_rate": 1.9308085511569984e-06, + "loss": 0.073, + "step": 15377 + }, + { + "epoch": 6.44, + "learning_rate": 1.9298072536832834e-06, + "loss": 0.0421, + "step": 15378 + }, + { + "epoch": 6.44, + "learning_rate": 1.9288061881800945e-06, + "loss": 0.0576, + "step": 15379 + }, + { + "epoch": 6.44, + "learning_rate": 1.92780535467621e-06, + "loss": 0.106, + "step": 15380 + }, + { + "epoch": 6.44, + "learning_rate": 1.9268047532003962e-06, + "loss": 0.0936, + "step": 15381 + }, + { + "epoch": 6.44, + "learning_rate": 1.9258043837814164e-06, + "loss": 0.0728, + "step": 15382 + }, + { + "epoch": 6.44, + "learning_rate": 1.9248042464480255e-06, + "loss": 0.0798, + "step": 15383 + }, + { + "epoch": 6.44, + "learning_rate": 1.9238043412289696e-06, + "loss": 0.088, + "step": 15384 + }, + { + "epoch": 6.44, + "learning_rate": 1.9228046681529954e-06, + "loss": 0.1014, + "step": 15385 + }, + { + "epoch": 6.44, + "learning_rate": 1.92180522724883e-06, + "loss": 0.0671, + "step": 15386 + }, + { + "epoch": 6.44, + "learning_rate": 1.9208060185452037e-06, + "loss": 0.082, + "step": 15387 + }, + { + "epoch": 6.44, + "learning_rate": 1.9198070420708405e-06, + "loss": 0.0795, + "step": 15388 + }, + { + "epoch": 6.44, + "learning_rate": 1.918808297854453e-06, + "loss": 0.0988, + "step": 15389 + }, + { + "epoch": 6.44, + "learning_rate": 1.9178097859247502e-06, + "loss": 0.0596, + "step": 15390 + }, + { + "epoch": 6.44, + "learning_rate": 1.9168115063104355e-06, + "loss": 0.0862, + "step": 15391 + }, + { + "epoch": 6.44, + "learning_rate": 1.915813459040199e-06, + "loss": 0.089, + "step": 15392 + }, + { + "epoch": 6.44, + "learning_rate": 1.91481564414273e-06, + "loss": 0.113, + "step": 15393 + }, + { + "epoch": 6.45, + "learning_rate": 1.9138180616467105e-06, + "loss": 0.0811, + "step": 15394 + }, + { + "epoch": 6.45, + "learning_rate": 1.912820711580816e-06, + "loss": 0.0709, + "step": 15395 + }, + { + "epoch": 6.45, + "learning_rate": 1.911823593973714e-06, + "loss": 0.0679, + "step": 15396 + }, + { + "epoch": 6.45, + "learning_rate": 1.910826708854068e-06, + "loss": 0.0947, + "step": 15397 + }, + { + "epoch": 6.45, + "learning_rate": 1.9098300562505266e-06, + "loss": 0.1242, + "step": 15398 + }, + { + "epoch": 6.45, + "learning_rate": 1.9088336361917415e-06, + "loss": 0.0767, + "step": 15399 + }, + { + "epoch": 6.45, + "learning_rate": 1.9078374487063536e-06, + "loss": 0.0597, + "step": 15400 + }, + { + "epoch": 6.45, + "learning_rate": 1.9068414938229996e-06, + "loss": 0.0804, + "step": 15401 + }, + { + "epoch": 6.45, + "learning_rate": 1.9058457715703038e-06, + "loss": 0.0633, + "step": 15402 + }, + { + "epoch": 6.45, + "learning_rate": 1.9048502819768877e-06, + "loss": 0.0782, + "step": 15403 + }, + { + "epoch": 6.45, + "learning_rate": 1.903855025071366e-06, + "loss": 0.0845, + "step": 15404 + }, + { + "epoch": 6.45, + "learning_rate": 1.9028600008823517e-06, + "loss": 0.09, + "step": 15405 + }, + { + "epoch": 6.45, + "learning_rate": 1.901865209438437e-06, + "loss": 0.0917, + "step": 15406 + }, + { + "epoch": 6.45, + "learning_rate": 1.9008706507682217e-06, + "loss": 0.0871, + "step": 15407 + }, + { + "epoch": 6.45, + "learning_rate": 1.8998763249002928e-06, + "loss": 0.0657, + "step": 15408 + }, + { + "epoch": 6.45, + "learning_rate": 1.8988822318632305e-06, + "loss": 0.0579, + "step": 15409 + }, + { + "epoch": 6.45, + "learning_rate": 1.897888371685609e-06, + "loss": 0.0778, + "step": 15410 + }, + { + "epoch": 6.45, + "learning_rate": 1.8968947443959995e-06, + "loss": 0.0683, + "step": 15411 + }, + { + "epoch": 6.45, + "learning_rate": 1.8959013500229584e-06, + "loss": 0.0723, + "step": 15412 + }, + { + "epoch": 6.45, + "learning_rate": 1.8949081885950403e-06, + "loss": 0.0686, + "step": 15413 + }, + { + "epoch": 6.45, + "learning_rate": 1.8939152601407952e-06, + "loss": 0.0635, + "step": 15414 + }, + { + "epoch": 6.45, + "learning_rate": 1.8929225646887617e-06, + "loss": 0.0956, + "step": 15415 + }, + { + "epoch": 6.45, + "learning_rate": 1.8919301022674752e-06, + "loss": 0.1139, + "step": 15416 + }, + { + "epoch": 6.45, + "learning_rate": 1.890937872905463e-06, + "loss": 0.0914, + "step": 15417 + }, + { + "epoch": 6.46, + "learning_rate": 1.8899458766312484e-06, + "loss": 0.1168, + "step": 15418 + }, + { + "epoch": 6.46, + "learning_rate": 1.8889541134733401e-06, + "loss": 0.0744, + "step": 15419 + }, + { + "epoch": 6.46, + "learning_rate": 1.8879625834602488e-06, + "loss": 0.0642, + "step": 15420 + }, + { + "epoch": 6.46, + "learning_rate": 1.8869712866204748e-06, + "loss": 0.0822, + "step": 15421 + }, + { + "epoch": 6.46, + "learning_rate": 1.8859802229825109e-06, + "loss": 0.0684, + "step": 15422 + }, + { + "epoch": 6.46, + "learning_rate": 1.8849893925748474e-06, + "loss": 0.0957, + "step": 15423 + }, + { + "epoch": 6.46, + "learning_rate": 1.8839987954259609e-06, + "loss": 0.078, + "step": 15424 + }, + { + "epoch": 6.46, + "learning_rate": 1.883008431564327e-06, + "loss": 0.0764, + "step": 15425 + }, + { + "epoch": 6.46, + "learning_rate": 1.8820183010184157e-06, + "loss": 0.1057, + "step": 15426 + }, + { + "epoch": 6.46, + "learning_rate": 1.8810284038166815e-06, + "loss": 0.0593, + "step": 15427 + }, + { + "epoch": 6.46, + "learning_rate": 1.880038739987582e-06, + "loss": 0.0686, + "step": 15428 + }, + { + "epoch": 6.46, + "learning_rate": 1.8790493095595618e-06, + "loss": 0.0886, + "step": 15429 + }, + { + "epoch": 6.46, + "learning_rate": 1.8780601125610643e-06, + "loss": 0.0795, + "step": 15430 + }, + { + "epoch": 6.46, + "learning_rate": 1.8770711490205207e-06, + "loss": 0.0824, + "step": 15431 + }, + { + "epoch": 6.46, + "learning_rate": 1.8760824189663618e-06, + "loss": 0.0775, + "step": 15432 + }, + { + "epoch": 6.46, + "learning_rate": 1.8750939224270014e-06, + "loss": 0.0763, + "step": 15433 + }, + { + "epoch": 6.46, + "learning_rate": 1.8741056594308571e-06, + "loss": 0.0712, + "step": 15434 + }, + { + "epoch": 6.46, + "learning_rate": 1.8731176300063347e-06, + "loss": 0.0806, + "step": 15435 + }, + { + "epoch": 6.46, + "learning_rate": 1.8721298341818339e-06, + "loss": 0.0542, + "step": 15436 + }, + { + "epoch": 6.46, + "learning_rate": 1.8711422719857487e-06, + "loss": 0.07, + "step": 15437 + }, + { + "epoch": 6.46, + "learning_rate": 1.8701549434464684e-06, + "loss": 0.0852, + "step": 15438 + }, + { + "epoch": 6.46, + "learning_rate": 1.869167848592367e-06, + "loss": 0.1039, + "step": 15439 + }, + { + "epoch": 6.46, + "learning_rate": 1.8681809874518208e-06, + "loss": 0.0915, + "step": 15440 + }, + { + "epoch": 6.46, + "learning_rate": 1.8671943600531962e-06, + "loss": 0.0983, + "step": 15441 + }, + { + "epoch": 6.47, + "learning_rate": 1.8662079664248522e-06, + "loss": 0.047, + "step": 15442 + }, + { + "epoch": 6.47, + "learning_rate": 1.8652218065951432e-06, + "loss": 0.1026, + "step": 15443 + }, + { + "epoch": 6.47, + "learning_rate": 1.8642358805924144e-06, + "loss": 0.0711, + "step": 15444 + }, + { + "epoch": 6.47, + "learning_rate": 1.8632501884450082e-06, + "loss": 0.1029, + "step": 15445 + }, + { + "epoch": 6.47, + "learning_rate": 1.8622647301812546e-06, + "loss": 0.1079, + "step": 15446 + }, + { + "epoch": 6.47, + "learning_rate": 1.8612795058294775e-06, + "loss": 0.0836, + "step": 15447 + }, + { + "epoch": 6.47, + "learning_rate": 1.8602945154179997e-06, + "loss": 0.0754, + "step": 15448 + }, + { + "epoch": 6.47, + "learning_rate": 1.8593097589751318e-06, + "loss": 0.0884, + "step": 15449 + }, + { + "epoch": 6.47, + "learning_rate": 1.8583252365291814e-06, + "loss": 0.0766, + "step": 15450 + }, + { + "epoch": 6.47, + "learning_rate": 1.857340948108448e-06, + "loss": 0.0849, + "step": 15451 + }, + { + "epoch": 6.47, + "learning_rate": 1.8563568937412247e-06, + "loss": 0.0885, + "step": 15452 + }, + { + "epoch": 6.47, + "learning_rate": 1.855373073455794e-06, + "loss": 0.0845, + "step": 15453 + }, + { + "epoch": 6.47, + "learning_rate": 1.8543894872804368e-06, + "loss": 0.0834, + "step": 15454 + }, + { + "epoch": 6.47, + "learning_rate": 1.8534061352434263e-06, + "loss": 0.084, + "step": 15455 + }, + { + "epoch": 6.47, + "learning_rate": 1.8524230173730261e-06, + "loss": 0.0745, + "step": 15456 + }, + { + "epoch": 6.47, + "learning_rate": 1.8514401336974975e-06, + "loss": 0.1124, + "step": 15457 + }, + { + "epoch": 6.47, + "learning_rate": 1.8504574842450907e-06, + "loss": 0.0824, + "step": 15458 + }, + { + "epoch": 6.47, + "learning_rate": 1.8494750690440533e-06, + "loss": 0.0709, + "step": 15459 + }, + { + "epoch": 6.47, + "learning_rate": 1.8484928881226205e-06, + "loss": 0.0754, + "step": 15460 + }, + { + "epoch": 6.47, + "learning_rate": 1.8475109415090264e-06, + "loss": 0.0705, + "step": 15461 + }, + { + "epoch": 6.47, + "learning_rate": 1.8465292292314963e-06, + "loss": 0.0823, + "step": 15462 + }, + { + "epoch": 6.47, + "learning_rate": 1.8455477513182473e-06, + "loss": 0.089, + "step": 15463 + }, + { + "epoch": 6.47, + "learning_rate": 1.8445665077974928e-06, + "loss": 0.1006, + "step": 15464 + }, + { + "epoch": 6.47, + "learning_rate": 1.8435854986974399e-06, + "loss": 0.084, + "step": 15465 + }, + { + "epoch": 6.48, + "learning_rate": 1.8426047240462807e-06, + "loss": 0.1041, + "step": 15466 + }, + { + "epoch": 6.48, + "learning_rate": 1.8416241838722093e-06, + "loss": 0.0706, + "step": 15467 + }, + { + "epoch": 6.48, + "learning_rate": 1.840643878203412e-06, + "loss": 0.0696, + "step": 15468 + }, + { + "epoch": 6.48, + "learning_rate": 1.8396638070680685e-06, + "loss": 0.0779, + "step": 15469 + }, + { + "epoch": 6.48, + "learning_rate": 1.8386839704943449e-06, + "loss": 0.0984, + "step": 15470 + }, + { + "epoch": 6.48, + "learning_rate": 1.837704368510409e-06, + "loss": 0.0896, + "step": 15471 + }, + { + "epoch": 6.48, + "learning_rate": 1.836725001144417e-06, + "loss": 0.0775, + "step": 15472 + }, + { + "epoch": 6.48, + "learning_rate": 1.8357458684245234e-06, + "loss": 0.0882, + "step": 15473 + }, + { + "epoch": 6.48, + "learning_rate": 1.8347669703788683e-06, + "loss": 0.0753, + "step": 15474 + }, + { + "epoch": 6.48, + "learning_rate": 1.83378830703559e-06, + "loss": 0.0791, + "step": 15475 + }, + { + "epoch": 6.48, + "learning_rate": 1.832809878422821e-06, + "loss": 0.0847, + "step": 15476 + }, + { + "epoch": 6.48, + "learning_rate": 1.8318316845686834e-06, + "loss": 0.0809, + "step": 15477 + }, + { + "epoch": 6.48, + "learning_rate": 1.830853725501297e-06, + "loss": 0.0602, + "step": 15478 + }, + { + "epoch": 6.48, + "learning_rate": 1.8298760012487726e-06, + "loss": 0.0987, + "step": 15479 + }, + { + "epoch": 6.48, + "learning_rate": 1.828898511839211e-06, + "loss": 0.0938, + "step": 15480 + }, + { + "epoch": 6.48, + "learning_rate": 1.8279212573007099e-06, + "loss": 0.1222, + "step": 15481 + }, + { + "epoch": 6.48, + "learning_rate": 1.826944237661361e-06, + "loss": 0.0844, + "step": 15482 + }, + { + "epoch": 6.48, + "learning_rate": 1.8259674529492476e-06, + "loss": 0.0666, + "step": 15483 + }, + { + "epoch": 6.48, + "learning_rate": 1.824990903192445e-06, + "loss": 0.073, + "step": 15484 + }, + { + "epoch": 6.48, + "learning_rate": 1.8240145884190264e-06, + "loss": 0.0588, + "step": 15485 + }, + { + "epoch": 6.48, + "learning_rate": 1.8230385086570546e-06, + "loss": 0.0886, + "step": 15486 + }, + { + "epoch": 6.48, + "learning_rate": 1.8220626639345817e-06, + "loss": 0.0772, + "step": 15487 + }, + { + "epoch": 6.48, + "learning_rate": 1.8210870542796622e-06, + "loss": 0.1025, + "step": 15488 + }, + { + "epoch": 6.48, + "learning_rate": 1.8201116797203355e-06, + "loss": 0.0705, + "step": 15489 + }, + { + "epoch": 6.49, + "learning_rate": 1.8191365402846417e-06, + "loss": 0.0535, + "step": 15490 + }, + { + "epoch": 6.49, + "learning_rate": 1.8181616360006094e-06, + "loss": 0.0975, + "step": 15491 + }, + { + "epoch": 6.49, + "learning_rate": 1.8171869668962594e-06, + "loss": 0.0777, + "step": 15492 + }, + { + "epoch": 6.49, + "learning_rate": 1.8162125329996105e-06, + "loss": 0.0518, + "step": 15493 + }, + { + "epoch": 6.49, + "learning_rate": 1.815238334338667e-06, + "loss": 0.0665, + "step": 15494 + }, + { + "epoch": 6.49, + "learning_rate": 1.8142643709414355e-06, + "loss": 0.0883, + "step": 15495 + }, + { + "epoch": 6.49, + "learning_rate": 1.813290642835911e-06, + "loss": 0.0653, + "step": 15496 + }, + { + "epoch": 6.49, + "learning_rate": 1.8123171500500814e-06, + "loss": 0.0458, + "step": 15497 + }, + { + "epoch": 6.49, + "learning_rate": 1.8113438926119297e-06, + "loss": 0.0869, + "step": 15498 + }, + { + "epoch": 6.49, + "learning_rate": 1.8103708705494316e-06, + "loss": 0.092, + "step": 15499 + }, + { + "epoch": 6.49, + "learning_rate": 1.809398083890559e-06, + "loss": 0.0797, + "step": 15500 + }, + { + "epoch": 6.49, + "learning_rate": 1.808425532663266e-06, + "loss": 0.084, + "step": 15501 + }, + { + "epoch": 6.49, + "learning_rate": 1.8074532168955128e-06, + "loss": 0.0594, + "step": 15502 + }, + { + "epoch": 6.49, + "learning_rate": 1.8064811366152468e-06, + "loss": 0.0855, + "step": 15503 + }, + { + "epoch": 6.49, + "learning_rate": 1.8055092918504103e-06, + "loss": 0.0568, + "step": 15504 + }, + { + "epoch": 6.49, + "learning_rate": 1.8045376826289374e-06, + "loss": 0.0883, + "step": 15505 + }, + { + "epoch": 6.49, + "learning_rate": 1.8035663089787582e-06, + "loss": 0.1022, + "step": 15506 + }, + { + "epoch": 6.49, + "learning_rate": 1.80259517092779e-06, + "loss": 0.0616, + "step": 15507 + }, + { + "epoch": 6.49, + "learning_rate": 1.8016242685039486e-06, + "loss": 0.079, + "step": 15508 + }, + { + "epoch": 6.49, + "learning_rate": 1.8006536017351438e-06, + "loss": 0.084, + "step": 15509 + }, + { + "epoch": 6.49, + "learning_rate": 1.7996831706492745e-06, + "loss": 0.0961, + "step": 15510 + }, + { + "epoch": 6.49, + "learning_rate": 1.798712975274236e-06, + "loss": 0.0689, + "step": 15511 + }, + { + "epoch": 6.49, + "learning_rate": 1.7977430156379173e-06, + "loss": 0.0799, + "step": 15512 + }, + { + "epoch": 6.49, + "learning_rate": 1.7967732917681946e-06, + "loss": 0.0747, + "step": 15513 + }, + { + "epoch": 6.5, + "learning_rate": 1.795803803692947e-06, + "loss": 0.0868, + "step": 15514 + }, + { + "epoch": 6.5, + "learning_rate": 1.7948345514400356e-06, + "loss": 0.0968, + "step": 15515 + }, + { + "epoch": 6.5, + "learning_rate": 1.7938655350373246e-06, + "loss": 0.0793, + "step": 15516 + }, + { + "epoch": 6.5, + "learning_rate": 1.7928967545126674e-06, + "loss": 0.0978, + "step": 15517 + }, + { + "epoch": 6.5, + "learning_rate": 1.7919282098939095e-06, + "loss": 0.0584, + "step": 15518 + }, + { + "epoch": 6.5, + "learning_rate": 1.7909599012088908e-06, + "loss": 0.0498, + "step": 15519 + }, + { + "epoch": 6.5, + "learning_rate": 1.789991828485449e-06, + "loss": 0.0937, + "step": 15520 + }, + { + "epoch": 6.5, + "learning_rate": 1.789023991751403e-06, + "loss": 0.0835, + "step": 15521 + }, + { + "epoch": 6.5, + "learning_rate": 1.788056391034576e-06, + "loss": 0.0685, + "step": 15522 + }, + { + "epoch": 6.5, + "learning_rate": 1.7870890263627817e-06, + "loss": 0.0886, + "step": 15523 + }, + { + "epoch": 6.5, + "learning_rate": 1.7861218977638252e-06, + "loss": 0.0806, + "step": 15524 + }, + { + "epoch": 6.5, + "learning_rate": 1.7851550052655052e-06, + "loss": 0.0801, + "step": 15525 + }, + { + "epoch": 6.5, + "learning_rate": 1.7841883488956157e-06, + "loss": 0.0688, + "step": 15526 + }, + { + "epoch": 6.5, + "learning_rate": 1.7832219286819442e-06, + "loss": 0.1044, + "step": 15527 + }, + { + "epoch": 6.5, + "learning_rate": 1.7822557446522637e-06, + "loss": 0.0741, + "step": 15528 + }, + { + "epoch": 6.5, + "learning_rate": 1.781289796834349e-06, + "loss": 0.0708, + "step": 15529 + }, + { + "epoch": 6.5, + "learning_rate": 1.7803240852559667e-06, + "loss": 0.0867, + "step": 15530 + }, + { + "epoch": 6.5, + "learning_rate": 1.7793586099448745e-06, + "loss": 0.0651, + "step": 15531 + }, + { + "epoch": 6.5, + "learning_rate": 1.7783933709288249e-06, + "loss": 0.0988, + "step": 15532 + }, + { + "epoch": 6.5, + "learning_rate": 1.7774283682355642e-06, + "loss": 0.0839, + "step": 15533 + }, + { + "epoch": 6.5, + "learning_rate": 1.7764636018928249e-06, + "loss": 0.055, + "step": 15534 + }, + { + "epoch": 6.5, + "learning_rate": 1.7754990719283427e-06, + "loss": 0.083, + "step": 15535 + }, + { + "epoch": 6.5, + "learning_rate": 1.774534778369842e-06, + "loss": 0.0905, + "step": 15536 + }, + { + "epoch": 6.5, + "learning_rate": 1.7735707212450414e-06, + "loss": 0.0762, + "step": 15537 + }, + { + "epoch": 6.51, + "learning_rate": 1.7726069005816482e-06, + "loss": 0.0912, + "step": 15538 + }, + { + "epoch": 6.51, + "learning_rate": 1.7716433164073698e-06, + "loss": 0.0708, + "step": 15539 + }, + { + "epoch": 6.51, + "learning_rate": 1.770679968749901e-06, + "loss": 0.0814, + "step": 15540 + }, + { + "epoch": 6.51, + "learning_rate": 1.7697168576369373e-06, + "loss": 0.0873, + "step": 15541 + }, + { + "epoch": 6.51, + "learning_rate": 1.7687539830961564e-06, + "loss": 0.0748, + "step": 15542 + }, + { + "epoch": 6.51, + "learning_rate": 1.7677913451552386e-06, + "loss": 0.0954, + "step": 15543 + }, + { + "epoch": 6.51, + "learning_rate": 1.766828943841853e-06, + "loss": 0.0713, + "step": 15544 + }, + { + "epoch": 6.51, + "learning_rate": 1.7658667791836626e-06, + "loss": 0.0869, + "step": 15545 + }, + { + "epoch": 6.51, + "learning_rate": 1.7649048512083256e-06, + "loss": 0.0974, + "step": 15546 + }, + { + "epoch": 6.51, + "learning_rate": 1.763943159943493e-06, + "loss": 0.0788, + "step": 15547 + }, + { + "epoch": 6.51, + "learning_rate": 1.762981705416804e-06, + "loss": 0.0978, + "step": 15548 + }, + { + "epoch": 6.51, + "learning_rate": 1.7620204876558967e-06, + "loss": 0.0823, + "step": 15549 + }, + { + "epoch": 6.51, + "learning_rate": 1.7610595066883996e-06, + "loss": 0.0808, + "step": 15550 + }, + { + "epoch": 6.51, + "learning_rate": 1.7600987625419353e-06, + "loss": 0.0848, + "step": 15551 + }, + { + "epoch": 6.51, + "learning_rate": 1.7591382552441206e-06, + "loss": 0.1244, + "step": 15552 + }, + { + "epoch": 6.51, + "learning_rate": 1.7581779848225665e-06, + "loss": 0.0953, + "step": 15553 + }, + { + "epoch": 6.51, + "learning_rate": 1.7572179513048703e-06, + "loss": 0.0913, + "step": 15554 + }, + { + "epoch": 6.51, + "learning_rate": 1.7562581547186298e-06, + "loss": 0.1055, + "step": 15555 + }, + { + "epoch": 6.51, + "learning_rate": 1.7552985950914326e-06, + "loss": 0.0888, + "step": 15556 + }, + { + "epoch": 6.51, + "learning_rate": 1.7543392724508613e-06, + "loss": 0.0913, + "step": 15557 + }, + { + "epoch": 6.51, + "learning_rate": 1.7533801868244904e-06, + "loss": 0.0821, + "step": 15558 + }, + { + "epoch": 6.51, + "learning_rate": 1.7524213382398902e-06, + "loss": 0.1092, + "step": 15559 + }, + { + "epoch": 6.51, + "learning_rate": 1.7514627267246187e-06, + "loss": 0.0841, + "step": 15560 + }, + { + "epoch": 6.51, + "learning_rate": 1.7505043523062326e-06, + "loss": 0.0623, + "step": 15561 + }, + { + "epoch": 6.52, + "learning_rate": 1.7495462150122767e-06, + "loss": 0.0702, + "step": 15562 + }, + { + "epoch": 6.52, + "learning_rate": 1.7485883148702931e-06, + "loss": 0.0709, + "step": 15563 + }, + { + "epoch": 6.52, + "learning_rate": 1.7476306519078168e-06, + "loss": 0.1001, + "step": 15564 + }, + { + "epoch": 6.52, + "learning_rate": 1.7466732261523745e-06, + "loss": 0.079, + "step": 15565 + }, + { + "epoch": 6.52, + "learning_rate": 1.7457160376314852e-06, + "loss": 0.0705, + "step": 15566 + }, + { + "epoch": 6.52, + "learning_rate": 1.7447590863726649e-06, + "loss": 0.0983, + "step": 15567 + }, + { + "epoch": 6.52, + "learning_rate": 1.7438023724034215e-06, + "loss": 0.0806, + "step": 15568 + }, + { + "epoch": 6.52, + "learning_rate": 1.7428458957512496e-06, + "loss": 0.0615, + "step": 15569 + }, + { + "epoch": 6.52, + "learning_rate": 1.741889656443645e-06, + "loss": 0.0832, + "step": 15570 + }, + { + "epoch": 6.52, + "learning_rate": 1.7409336545080947e-06, + "loss": 0.0716, + "step": 15571 + }, + { + "epoch": 6.52, + "learning_rate": 1.7399778899720764e-06, + "loss": 0.1212, + "step": 15572 + }, + { + "epoch": 6.52, + "learning_rate": 1.7390223628630643e-06, + "loss": 0.1216, + "step": 15573 + }, + { + "epoch": 6.52, + "learning_rate": 1.738067073208527e-06, + "loss": 0.0805, + "step": 15574 + }, + { + "epoch": 6.52, + "learning_rate": 1.737112021035917e-06, + "loss": 0.0785, + "step": 15575 + }, + { + "epoch": 6.52, + "learning_rate": 1.7361572063726894e-06, + "loss": 0.1026, + "step": 15576 + }, + { + "epoch": 6.52, + "learning_rate": 1.735202629246291e-06, + "loss": 0.0766, + "step": 15577 + }, + { + "epoch": 6.52, + "learning_rate": 1.7342482896841572e-06, + "loss": 0.0695, + "step": 15578 + }, + { + "epoch": 6.52, + "learning_rate": 1.7332941877137233e-06, + "loss": 0.0673, + "step": 15579 + }, + { + "epoch": 6.52, + "learning_rate": 1.7323403233624136e-06, + "loss": 0.0663, + "step": 15580 + }, + { + "epoch": 6.52, + "learning_rate": 1.731386696657642e-06, + "loss": 0.0911, + "step": 15581 + }, + { + "epoch": 6.52, + "learning_rate": 1.7304333076268253e-06, + "loss": 0.0878, + "step": 15582 + }, + { + "epoch": 6.52, + "learning_rate": 1.7294801562973619e-06, + "loss": 0.0663, + "step": 15583 + }, + { + "epoch": 6.52, + "learning_rate": 1.7285272426966526e-06, + "loss": 0.0778, + "step": 15584 + }, + { + "epoch": 6.53, + "learning_rate": 1.7275745668520883e-06, + "loss": 0.0794, + "step": 15585 + }, + { + "epoch": 6.53, + "learning_rate": 1.7266221287910512e-06, + "loss": 0.0696, + "step": 15586 + }, + { + "epoch": 6.53, + "learning_rate": 1.7256699285409206e-06, + "loss": 0.0906, + "step": 15587 + }, + { + "epoch": 6.53, + "learning_rate": 1.724717966129067e-06, + "loss": 0.0507, + "step": 15588 + }, + { + "epoch": 6.53, + "learning_rate": 1.7237662415828493e-06, + "loss": 0.0617, + "step": 15589 + }, + { + "epoch": 6.53, + "learning_rate": 1.7228147549296271e-06, + "loss": 0.0886, + "step": 15590 + }, + { + "epoch": 6.53, + "learning_rate": 1.7218635061967505e-06, + "loss": 0.0585, + "step": 15591 + }, + { + "epoch": 6.53, + "learning_rate": 1.7209124954115608e-06, + "loss": 0.086, + "step": 15592 + }, + { + "epoch": 6.53, + "learning_rate": 1.7199617226013943e-06, + "loss": 0.0943, + "step": 15593 + }, + { + "epoch": 6.53, + "learning_rate": 1.7190111877935833e-06, + "loss": 0.0768, + "step": 15594 + }, + { + "epoch": 6.53, + "learning_rate": 1.718060891015445e-06, + "loss": 0.0918, + "step": 15595 + }, + { + "epoch": 6.53, + "learning_rate": 1.7171108322942975e-06, + "loss": 0.0945, + "step": 15596 + }, + { + "epoch": 6.53, + "learning_rate": 1.716161011657449e-06, + "loss": 0.0979, + "step": 15597 + }, + { + "epoch": 6.53, + "learning_rate": 1.7152114291322008e-06, + "loss": 0.1001, + "step": 15598 + }, + { + "epoch": 6.53, + "learning_rate": 1.714262084745848e-06, + "loss": 0.0842, + "step": 15599 + }, + { + "epoch": 6.53, + "learning_rate": 1.7133129785256797e-06, + "loss": 0.0716, + "step": 15600 + }, + { + "epoch": 6.53, + "learning_rate": 1.7123641104989786e-06, + "loss": 0.0983, + "step": 15601 + }, + { + "epoch": 6.53, + "learning_rate": 1.7114154806930138e-06, + "loss": 0.0482, + "step": 15602 + }, + { + "epoch": 6.53, + "learning_rate": 1.7104670891350572e-06, + "loss": 0.0885, + "step": 15603 + }, + { + "epoch": 6.53, + "learning_rate": 1.7095189358523668e-06, + "loss": 0.0932, + "step": 15604 + }, + { + "epoch": 6.53, + "learning_rate": 1.7085710208722006e-06, + "loss": 0.073, + "step": 15605 + }, + { + "epoch": 6.53, + "learning_rate": 1.7076233442218004e-06, + "loss": 0.0746, + "step": 15606 + }, + { + "epoch": 6.53, + "learning_rate": 1.7066759059284087e-06, + "loss": 0.0675, + "step": 15607 + }, + { + "epoch": 6.53, + "learning_rate": 1.7057287060192585e-06, + "loss": 0.0975, + "step": 15608 + }, + { + "epoch": 6.54, + "learning_rate": 1.7047817445215786e-06, + "loss": 0.0812, + "step": 15609 + }, + { + "epoch": 6.54, + "learning_rate": 1.7038350214625844e-06, + "loss": 0.1039, + "step": 15610 + }, + { + "epoch": 6.54, + "learning_rate": 1.7028885368694903e-06, + "loss": 0.0927, + "step": 15611 + }, + { + "epoch": 6.54, + "learning_rate": 1.7019422907695026e-06, + "loss": 0.0954, + "step": 15612 + }, + { + "epoch": 6.54, + "learning_rate": 1.7009962831898208e-06, + "loss": 0.0851, + "step": 15613 + }, + { + "epoch": 6.54, + "learning_rate": 1.700050514157635e-06, + "loss": 0.0976, + "step": 15614 + }, + { + "epoch": 6.54, + "learning_rate": 1.6991049837001362e-06, + "loss": 0.1027, + "step": 15615 + }, + { + "epoch": 6.54, + "learning_rate": 1.6981596918444953e-06, + "loss": 0.103, + "step": 15616 + }, + { + "epoch": 6.54, + "learning_rate": 1.6972146386178868e-06, + "loss": 0.0748, + "step": 15617 + }, + { + "epoch": 6.54, + "learning_rate": 1.6962698240474762e-06, + "loss": 0.092, + "step": 15618 + }, + { + "epoch": 6.54, + "learning_rate": 1.6953252481604198e-06, + "loss": 0.0901, + "step": 15619 + }, + { + "epoch": 6.54, + "learning_rate": 1.6943809109838716e-06, + "loss": 0.1037, + "step": 15620 + }, + { + "epoch": 6.54, + "learning_rate": 1.6934368125449752e-06, + "loss": 0.1248, + "step": 15621 + }, + { + "epoch": 6.54, + "learning_rate": 1.6924929528708633e-06, + "loss": 0.072, + "step": 15622 + }, + { + "epoch": 6.54, + "learning_rate": 1.6915493319886712e-06, + "loss": 0.0888, + "step": 15623 + }, + { + "epoch": 6.54, + "learning_rate": 1.690605949925519e-06, + "loss": 0.0678, + "step": 15624 + }, + { + "epoch": 6.54, + "learning_rate": 1.6896628067085264e-06, + "loss": 0.0959, + "step": 15625 + }, + { + "epoch": 6.54, + "learning_rate": 1.6887199023648016e-06, + "loss": 0.0664, + "step": 15626 + }, + { + "epoch": 6.54, + "learning_rate": 1.6877772369214507e-06, + "loss": 0.0693, + "step": 15627 + }, + { + "epoch": 6.54, + "learning_rate": 1.6868348104055642e-06, + "loss": 0.0777, + "step": 15628 + }, + { + "epoch": 6.54, + "learning_rate": 1.6858926228442362e-06, + "loss": 0.1035, + "step": 15629 + }, + { + "epoch": 6.54, + "learning_rate": 1.6849506742645449e-06, + "loss": 0.0704, + "step": 15630 + }, + { + "epoch": 6.54, + "learning_rate": 1.6840089646935686e-06, + "loss": 0.1098, + "step": 15631 + }, + { + "epoch": 6.54, + "learning_rate": 1.6830674941583736e-06, + "loss": 0.0978, + "step": 15632 + }, + { + "epoch": 6.55, + "learning_rate": 1.6821262626860246e-06, + "loss": 0.094, + "step": 15633 + }, + { + "epoch": 6.55, + "learning_rate": 1.6811852703035737e-06, + "loss": 0.0567, + "step": 15634 + }, + { + "epoch": 6.55, + "learning_rate": 1.6802445170380733e-06, + "loss": 0.098, + "step": 15635 + }, + { + "epoch": 6.55, + "learning_rate": 1.6793040029165596e-06, + "loss": 0.0864, + "step": 15636 + }, + { + "epoch": 6.55, + "learning_rate": 1.678363727966068e-06, + "loss": 0.0934, + "step": 15637 + }, + { + "epoch": 6.55, + "learning_rate": 1.6774236922136266e-06, + "loss": 0.0623, + "step": 15638 + }, + { + "epoch": 6.55, + "learning_rate": 1.676483895686255e-06, + "loss": 0.0902, + "step": 15639 + }, + { + "epoch": 6.55, + "learning_rate": 1.6755443384109694e-06, + "loss": 0.0686, + "step": 15640 + }, + { + "epoch": 6.55, + "learning_rate": 1.6746050204147734e-06, + "loss": 0.0762, + "step": 15641 + }, + { + "epoch": 6.55, + "learning_rate": 1.6736659417246714e-06, + "loss": 0.0639, + "step": 15642 + }, + { + "epoch": 6.55, + "learning_rate": 1.6727271023676506e-06, + "loss": 0.0844, + "step": 15643 + }, + { + "epoch": 6.55, + "learning_rate": 1.6717885023707004e-06, + "loss": 0.0582, + "step": 15644 + }, + { + "epoch": 6.55, + "learning_rate": 1.670850141760798e-06, + "loss": 0.1034, + "step": 15645 + }, + { + "epoch": 6.55, + "learning_rate": 1.6699120205649177e-06, + "loss": 0.0958, + "step": 15646 + }, + { + "epoch": 6.55, + "learning_rate": 1.6689741388100244e-06, + "loss": 0.086, + "step": 15647 + }, + { + "epoch": 6.55, + "learning_rate": 1.66803649652308e-06, + "loss": 0.08, + "step": 15648 + }, + { + "epoch": 6.55, + "learning_rate": 1.6670990937310293e-06, + "loss": 0.098, + "step": 15649 + }, + { + "epoch": 6.55, + "learning_rate": 1.6661619304608234e-06, + "loss": 0.0737, + "step": 15650 + }, + { + "epoch": 6.55, + "learning_rate": 1.665225006739396e-06, + "loss": 0.0956, + "step": 15651 + }, + { + "epoch": 6.55, + "learning_rate": 1.6642883225936779e-06, + "loss": 0.0917, + "step": 15652 + }, + { + "epoch": 6.55, + "learning_rate": 1.6633518780505964e-06, + "loss": 0.098, + "step": 15653 + }, + { + "epoch": 6.55, + "learning_rate": 1.6624156731370666e-06, + "loss": 0.0646, + "step": 15654 + }, + { + "epoch": 6.55, + "learning_rate": 1.6614797078799993e-06, + "loss": 0.0836, + "step": 15655 + }, + { + "epoch": 6.55, + "learning_rate": 1.6605439823063018e-06, + "loss": 0.0809, + "step": 15656 + }, + { + "epoch": 6.56, + "learning_rate": 1.659608496442865e-06, + "loss": 0.0779, + "step": 15657 + }, + { + "epoch": 6.56, + "learning_rate": 1.6586732503165793e-06, + "loss": 0.0667, + "step": 15658 + }, + { + "epoch": 6.56, + "learning_rate": 1.65773824395433e-06, + "loss": 0.1243, + "step": 15659 + }, + { + "epoch": 6.56, + "learning_rate": 1.6568034773829923e-06, + "loss": 0.0629, + "step": 15660 + }, + { + "epoch": 6.56, + "learning_rate": 1.6558689506294345e-06, + "loss": 0.1078, + "step": 15661 + }, + { + "epoch": 6.56, + "learning_rate": 1.654934663720522e-06, + "loss": 0.0776, + "step": 15662 + }, + { + "epoch": 6.56, + "learning_rate": 1.6540006166831046e-06, + "loss": 0.104, + "step": 15663 + }, + { + "epoch": 6.56, + "learning_rate": 1.6530668095440339e-06, + "loss": 0.0857, + "step": 15664 + }, + { + "epoch": 6.56, + "learning_rate": 1.6521332423301507e-06, + "loss": 0.091, + "step": 15665 + }, + { + "epoch": 6.56, + "learning_rate": 1.6511999150682888e-06, + "loss": 0.0714, + "step": 15666 + }, + { + "epoch": 6.56, + "learning_rate": 1.6502668277852774e-06, + "loss": 0.0698, + "step": 15667 + }, + { + "epoch": 6.56, + "learning_rate": 1.6493339805079367e-06, + "loss": 0.0865, + "step": 15668 + }, + { + "epoch": 6.56, + "learning_rate": 1.6484013732630833e-06, + "loss": 0.0685, + "step": 15669 + }, + { + "epoch": 6.56, + "learning_rate": 1.6474690060775178e-06, + "loss": 0.0712, + "step": 15670 + }, + { + "epoch": 6.56, + "learning_rate": 1.6465368789780446e-06, + "loss": 0.049, + "step": 15671 + }, + { + "epoch": 6.56, + "learning_rate": 1.6456049919914563e-06, + "loss": 0.0795, + "step": 15672 + }, + { + "epoch": 6.56, + "learning_rate": 1.6446733451445408e-06, + "loss": 0.0751, + "step": 15673 + }, + { + "epoch": 6.56, + "learning_rate": 1.643741938464073e-06, + "loss": 0.0589, + "step": 15674 + }, + { + "epoch": 6.56, + "learning_rate": 1.6428107719768293e-06, + "loss": 0.0733, + "step": 15675 + }, + { + "epoch": 6.56, + "learning_rate": 1.6418798457095753e-06, + "loss": 0.0658, + "step": 15676 + }, + { + "epoch": 6.56, + "learning_rate": 1.640949159689066e-06, + "loss": 0.0452, + "step": 15677 + }, + { + "epoch": 6.56, + "learning_rate": 1.6400187139420554e-06, + "loss": 0.0795, + "step": 15678 + }, + { + "epoch": 6.56, + "learning_rate": 1.6390885084952878e-06, + "loss": 0.1011, + "step": 15679 + }, + { + "epoch": 6.56, + "learning_rate": 1.638158543375502e-06, + "loss": 0.0588, + "step": 15680 + }, + { + "epoch": 6.57, + "learning_rate": 1.6372288186094287e-06, + "loss": 0.0611, + "step": 15681 + }, + { + "epoch": 6.57, + "learning_rate": 1.6362993342237931e-06, + "loss": 0.0871, + "step": 15682 + }, + { + "epoch": 6.57, + "learning_rate": 1.635370090245313e-06, + "loss": 0.1033, + "step": 15683 + }, + { + "epoch": 6.57, + "learning_rate": 1.6344410867006944e-06, + "loss": 0.0919, + "step": 15684 + }, + { + "epoch": 6.57, + "learning_rate": 1.6335123236166428e-06, + "loss": 0.0885, + "step": 15685 + }, + { + "epoch": 6.57, + "learning_rate": 1.6325838010198558e-06, + "loss": 0.0697, + "step": 15686 + }, + { + "epoch": 6.57, + "learning_rate": 1.631655518937022e-06, + "loss": 0.0664, + "step": 15687 + }, + { + "epoch": 6.57, + "learning_rate": 1.6307274773948244e-06, + "loss": 0.0457, + "step": 15688 + }, + { + "epoch": 6.57, + "learning_rate": 1.629799676419942e-06, + "loss": 0.0925, + "step": 15689 + }, + { + "epoch": 6.57, + "learning_rate": 1.6288721160390375e-06, + "loss": 0.0883, + "step": 15690 + }, + { + "epoch": 6.57, + "learning_rate": 1.6279447962787753e-06, + "loss": 0.0826, + "step": 15691 + }, + { + "epoch": 6.57, + "learning_rate": 1.6270177171658108e-06, + "loss": 0.0764, + "step": 15692 + }, + { + "epoch": 6.57, + "learning_rate": 1.6260908787267915e-06, + "loss": 0.0912, + "step": 15693 + }, + { + "epoch": 6.57, + "learning_rate": 1.6251642809883595e-06, + "loss": 0.0812, + "step": 15694 + }, + { + "epoch": 6.57, + "learning_rate": 1.624237923977151e-06, + "loss": 0.0999, + "step": 15695 + }, + { + "epoch": 6.57, + "learning_rate": 1.6233118077197884e-06, + "loss": 0.0605, + "step": 15696 + }, + { + "epoch": 6.57, + "learning_rate": 1.622385932242897e-06, + "loss": 0.0817, + "step": 15697 + }, + { + "epoch": 6.57, + "learning_rate": 1.6214602975730853e-06, + "loss": 0.0605, + "step": 15698 + }, + { + "epoch": 6.57, + "learning_rate": 1.6205349037369621e-06, + "loss": 0.101, + "step": 15699 + }, + { + "epoch": 6.57, + "learning_rate": 1.6196097507611274e-06, + "loss": 0.0774, + "step": 15700 + }, + { + "epoch": 6.57, + "learning_rate": 1.6186848386721743e-06, + "loss": 0.1025, + "step": 15701 + }, + { + "epoch": 6.57, + "learning_rate": 1.617760167496688e-06, + "loss": 0.0637, + "step": 15702 + }, + { + "epoch": 6.57, + "learning_rate": 1.6168357372612498e-06, + "loss": 0.046, + "step": 15703 + }, + { + "epoch": 6.57, + "learning_rate": 1.6159115479924259e-06, + "loss": 0.0925, + "step": 15704 + }, + { + "epoch": 6.58, + "learning_rate": 1.6149875997167851e-06, + "loss": 0.1007, + "step": 15705 + }, + { + "epoch": 6.58, + "learning_rate": 1.6140638924608853e-06, + "loss": 0.0874, + "step": 15706 + }, + { + "epoch": 6.58, + "learning_rate": 1.6131404262512774e-06, + "loss": 0.0554, + "step": 15707 + }, + { + "epoch": 6.58, + "learning_rate": 1.6122172011145053e-06, + "loss": 0.0789, + "step": 15708 + }, + { + "epoch": 6.58, + "learning_rate": 1.6112942170771073e-06, + "loss": 0.0864, + "step": 15709 + }, + { + "epoch": 6.58, + "learning_rate": 1.6103714741656152e-06, + "loss": 0.0571, + "step": 15710 + }, + { + "epoch": 6.58, + "learning_rate": 1.6094489724065476e-06, + "loss": 0.1119, + "step": 15711 + }, + { + "epoch": 6.58, + "learning_rate": 1.6085267118264235e-06, + "loss": 0.0649, + "step": 15712 + }, + { + "epoch": 6.58, + "learning_rate": 1.6076046924517518e-06, + "loss": 0.0511, + "step": 15713 + }, + { + "epoch": 6.58, + "learning_rate": 1.606682914309038e-06, + "loss": 0.0763, + "step": 15714 + }, + { + "epoch": 6.58, + "learning_rate": 1.6057613774247738e-06, + "loss": 0.0917, + "step": 15715 + }, + { + "epoch": 6.58, + "learning_rate": 1.6048400818254528e-06, + "loss": 0.0586, + "step": 15716 + }, + { + "epoch": 6.58, + "learning_rate": 1.6039190275375526e-06, + "loss": 0.0785, + "step": 15717 + }, + { + "epoch": 6.58, + "learning_rate": 1.60299821458755e-06, + "loss": 0.0894, + "step": 15718 + }, + { + "epoch": 6.58, + "learning_rate": 1.6020776430019113e-06, + "loss": 0.0793, + "step": 15719 + }, + { + "epoch": 6.58, + "learning_rate": 1.6011573128070978e-06, + "loss": 0.0858, + "step": 15720 + }, + { + "epoch": 6.58, + "learning_rate": 1.600237224029565e-06, + "loss": 0.0924, + "step": 15721 + }, + { + "epoch": 6.58, + "learning_rate": 1.5993173766957592e-06, + "loss": 0.0849, + "step": 15722 + }, + { + "epoch": 6.58, + "learning_rate": 1.5983977708321207e-06, + "loss": 0.0839, + "step": 15723 + }, + { + "epoch": 6.58, + "learning_rate": 1.597478406465085e-06, + "loss": 0.0739, + "step": 15724 + }, + { + "epoch": 6.58, + "learning_rate": 1.596559283621074e-06, + "loss": 0.0767, + "step": 15725 + }, + { + "epoch": 6.58, + "learning_rate": 1.5956404023265092e-06, + "loss": 0.0593, + "step": 15726 + }, + { + "epoch": 6.58, + "learning_rate": 1.5947217626078039e-06, + "loss": 0.0842, + "step": 15727 + }, + { + "epoch": 6.58, + "learning_rate": 1.593803364491362e-06, + "loss": 0.0605, + "step": 15728 + }, + { + "epoch": 6.59, + "learning_rate": 1.592885208003583e-06, + "loss": 0.0713, + "step": 15729 + }, + { + "epoch": 6.59, + "learning_rate": 1.5919672931708607e-06, + "loss": 0.1021, + "step": 15730 + }, + { + "epoch": 6.59, + "learning_rate": 1.5910496200195757e-06, + "loss": 0.0756, + "step": 15731 + }, + { + "epoch": 6.59, + "learning_rate": 1.5901321885761068e-06, + "loss": 0.0888, + "step": 15732 + }, + { + "epoch": 6.59, + "learning_rate": 1.5892149988668249e-06, + "loss": 0.0864, + "step": 15733 + }, + { + "epoch": 6.59, + "learning_rate": 1.5882980509180956e-06, + "loss": 0.083, + "step": 15734 + }, + { + "epoch": 6.59, + "learning_rate": 1.5873813447562735e-06, + "loss": 0.0715, + "step": 15735 + }, + { + "epoch": 6.59, + "learning_rate": 1.5864648804077122e-06, + "loss": 0.053, + "step": 15736 + }, + { + "epoch": 6.59, + "learning_rate": 1.5855486578987489e-06, + "loss": 0.0777, + "step": 15737 + }, + { + "epoch": 6.59, + "learning_rate": 1.584632677255723e-06, + "loss": 0.0909, + "step": 15738 + }, + { + "epoch": 6.59, + "learning_rate": 1.5837169385049623e-06, + "loss": 0.0655, + "step": 15739 + }, + { + "epoch": 6.59, + "learning_rate": 1.582801441672791e-06, + "loss": 0.0622, + "step": 15740 + }, + { + "epoch": 6.59, + "learning_rate": 1.5818861867855239e-06, + "loss": 0.0597, + "step": 15741 + }, + { + "epoch": 6.59, + "learning_rate": 1.5809711738694665e-06, + "loss": 0.0651, + "step": 15742 + }, + { + "epoch": 6.59, + "learning_rate": 1.5800564029509225e-06, + "loss": 0.0885, + "step": 15743 + }, + { + "epoch": 6.59, + "learning_rate": 1.579141874056187e-06, + "loss": 0.0718, + "step": 15744 + }, + { + "epoch": 6.59, + "learning_rate": 1.578227587211545e-06, + "loss": 0.0479, + "step": 15745 + }, + { + "epoch": 6.59, + "learning_rate": 1.5773135424432762e-06, + "loss": 0.0664, + "step": 15746 + }, + { + "epoch": 6.59, + "learning_rate": 1.5763997397776553e-06, + "loss": 0.0864, + "step": 15747 + }, + { + "epoch": 6.59, + "learning_rate": 1.5754861792409493e-06, + "loss": 0.081, + "step": 15748 + }, + { + "epoch": 6.59, + "learning_rate": 1.574572860859417e-06, + "loss": 0.0448, + "step": 15749 + }, + { + "epoch": 6.59, + "learning_rate": 1.5736597846593115e-06, + "loss": 0.0616, + "step": 15750 + }, + { + "epoch": 6.59, + "learning_rate": 1.5727469506668814e-06, + "loss": 0.0791, + "step": 15751 + }, + { + "epoch": 6.59, + "learning_rate": 1.5718343589083584e-06, + "loss": 0.0777, + "step": 15752 + }, + { + "epoch": 6.6, + "learning_rate": 1.5709220094099775e-06, + "loss": 0.087, + "step": 15753 + }, + { + "epoch": 6.6, + "learning_rate": 1.5700099021979642e-06, + "loss": 0.0565, + "step": 15754 + }, + { + "epoch": 6.6, + "learning_rate": 1.5690980372985354e-06, + "loss": 0.0675, + "step": 15755 + }, + { + "epoch": 6.6, + "learning_rate": 1.5681864147379022e-06, + "loss": 0.0759, + "step": 15756 + }, + { + "epoch": 6.6, + "learning_rate": 1.5672750345422716e-06, + "loss": 0.0632, + "step": 15757 + }, + { + "epoch": 6.6, + "learning_rate": 1.5663638967378337e-06, + "loss": 0.0824, + "step": 15758 + }, + { + "epoch": 6.6, + "learning_rate": 1.565453001350782e-06, + "loss": 0.0609, + "step": 15759 + }, + { + "epoch": 6.6, + "learning_rate": 1.5645423484072987e-06, + "loss": 0.0871, + "step": 15760 + }, + { + "epoch": 6.6, + "learning_rate": 1.5636319379335596e-06, + "loss": 0.103, + "step": 15761 + }, + { + "epoch": 6.6, + "learning_rate": 1.5627217699557352e-06, + "loss": 0.0919, + "step": 15762 + }, + { + "epoch": 6.6, + "learning_rate": 1.5618118444999885e-06, + "loss": 0.0721, + "step": 15763 + }, + { + "epoch": 6.6, + "learning_rate": 1.5609021615924702e-06, + "loss": 0.0606, + "step": 15764 + }, + { + "epoch": 6.6, + "learning_rate": 1.5599927212593325e-06, + "loss": 0.0896, + "step": 15765 + }, + { + "epoch": 6.6, + "learning_rate": 1.559083523526712e-06, + "loss": 0.1032, + "step": 15766 + }, + { + "epoch": 6.6, + "learning_rate": 1.558174568420746e-06, + "loss": 0.1055, + "step": 15767 + }, + { + "epoch": 6.6, + "learning_rate": 1.5572658559675602e-06, + "loss": 0.0868, + "step": 15768 + }, + { + "epoch": 6.6, + "learning_rate": 1.5563573861932756e-06, + "loss": 0.075, + "step": 15769 + }, + { + "epoch": 6.6, + "learning_rate": 1.555449159124005e-06, + "loss": 0.1048, + "step": 15770 + }, + { + "epoch": 6.6, + "learning_rate": 1.5545411747858574e-06, + "loss": 0.0805, + "step": 15771 + }, + { + "epoch": 6.6, + "learning_rate": 1.5536334332049274e-06, + "loss": 0.0666, + "step": 15772 + }, + { + "epoch": 6.6, + "learning_rate": 1.5527259344073087e-06, + "loss": 0.0646, + "step": 15773 + }, + { + "epoch": 6.6, + "learning_rate": 1.551818678419087e-06, + "loss": 0.1058, + "step": 15774 + }, + { + "epoch": 6.6, + "learning_rate": 1.5509116652663403e-06, + "loss": 0.0872, + "step": 15775 + }, + { + "epoch": 6.6, + "learning_rate": 1.5500048949751411e-06, + "loss": 0.0849, + "step": 15776 + }, + { + "epoch": 6.61, + "learning_rate": 1.5490983675715544e-06, + "loss": 0.0757, + "step": 15777 + }, + { + "epoch": 6.61, + "learning_rate": 1.5481920830816343e-06, + "loss": 0.107, + "step": 15778 + }, + { + "epoch": 6.61, + "learning_rate": 1.5472860415314317e-06, + "loss": 0.0731, + "step": 15779 + }, + { + "epoch": 6.61, + "learning_rate": 1.546380242946992e-06, + "loss": 0.1056, + "step": 15780 + }, + { + "epoch": 6.61, + "learning_rate": 1.5454746873543503e-06, + "loss": 0.078, + "step": 15781 + }, + { + "epoch": 6.61, + "learning_rate": 1.5445693747795366e-06, + "loss": 0.0975, + "step": 15782 + }, + { + "epoch": 6.61, + "learning_rate": 1.5436643052485723e-06, + "loss": 0.0566, + "step": 15783 + }, + { + "epoch": 6.61, + "learning_rate": 1.5427594787874766e-06, + "loss": 0.0795, + "step": 15784 + }, + { + "epoch": 6.61, + "learning_rate": 1.5418548954222512e-06, + "loss": 0.1206, + "step": 15785 + }, + { + "epoch": 6.61, + "learning_rate": 1.540950555178905e-06, + "loss": 0.0608, + "step": 15786 + }, + { + "epoch": 6.61, + "learning_rate": 1.5400464580834262e-06, + "loss": 0.0807, + "step": 15787 + }, + { + "epoch": 6.61, + "learning_rate": 1.539142604161804e-06, + "loss": 0.066, + "step": 15788 + }, + { + "epoch": 6.61, + "learning_rate": 1.5382389934400199e-06, + "loss": 0.0738, + "step": 15789 + }, + { + "epoch": 6.61, + "learning_rate": 1.5373356259440475e-06, + "loss": 0.0766, + "step": 15790 + }, + { + "epoch": 6.61, + "learning_rate": 1.5364325016998528e-06, + "loss": 0.0858, + "step": 15791 + }, + { + "epoch": 6.61, + "learning_rate": 1.535529620733397e-06, + "loss": 0.0799, + "step": 15792 + }, + { + "epoch": 6.61, + "learning_rate": 1.5346269830706295e-06, + "loss": 0.0814, + "step": 15793 + }, + { + "epoch": 6.61, + "learning_rate": 1.5337245887374962e-06, + "loss": 0.0911, + "step": 15794 + }, + { + "epoch": 6.61, + "learning_rate": 1.5328224377599376e-06, + "loss": 0.0756, + "step": 15795 + }, + { + "epoch": 6.61, + "learning_rate": 1.531920530163885e-06, + "loss": 0.1003, + "step": 15796 + }, + { + "epoch": 6.61, + "learning_rate": 1.5310188659752612e-06, + "loss": 0.0924, + "step": 15797 + }, + { + "epoch": 6.61, + "learning_rate": 1.5301174452199874e-06, + "loss": 0.0671, + "step": 15798 + }, + { + "epoch": 6.61, + "learning_rate": 1.5292162679239697e-06, + "loss": 0.0975, + "step": 15799 + }, + { + "epoch": 6.62, + "learning_rate": 1.528315334113114e-06, + "loss": 0.0744, + "step": 15800 + }, + { + "epoch": 6.62, + "learning_rate": 1.5274146438133154e-06, + "loss": 0.0805, + "step": 15801 + }, + { + "epoch": 6.62, + "learning_rate": 1.5265141970504648e-06, + "loss": 0.0738, + "step": 15802 + }, + { + "epoch": 6.62, + "learning_rate": 1.5256139938504445e-06, + "loss": 0.1022, + "step": 15803 + }, + { + "epoch": 6.62, + "learning_rate": 1.524714034239132e-06, + "loss": 0.0975, + "step": 15804 + }, + { + "epoch": 6.62, + "learning_rate": 1.5238143182423914e-06, + "loss": 0.0859, + "step": 15805 + }, + { + "epoch": 6.62, + "learning_rate": 1.5229148458860865e-06, + "loss": 0.0904, + "step": 15806 + }, + { + "epoch": 6.62, + "learning_rate": 1.522015617196072e-06, + "loss": 0.0979, + "step": 15807 + }, + { + "epoch": 6.62, + "learning_rate": 1.5211166321981963e-06, + "loss": 0.0959, + "step": 15808 + }, + { + "epoch": 6.62, + "learning_rate": 1.5202178909182997e-06, + "loss": 0.0619, + "step": 15809 + }, + { + "epoch": 6.62, + "learning_rate": 1.5193193933822136e-06, + "loss": 0.0899, + "step": 15810 + }, + { + "epoch": 6.62, + "learning_rate": 1.5184211396157655e-06, + "loss": 0.073, + "step": 15811 + }, + { + "epoch": 6.62, + "learning_rate": 1.5175231296447778e-06, + "loss": 0.0746, + "step": 15812 + }, + { + "epoch": 6.62, + "learning_rate": 1.516625363495059e-06, + "loss": 0.0975, + "step": 15813 + }, + { + "epoch": 6.62, + "learning_rate": 1.5157278411924148e-06, + "loss": 0.0644, + "step": 15814 + }, + { + "epoch": 6.62, + "learning_rate": 1.5148305627626458e-06, + "loss": 0.0867, + "step": 15815 + }, + { + "epoch": 6.62, + "learning_rate": 1.5139335282315427e-06, + "loss": 0.089, + "step": 15816 + }, + { + "epoch": 6.62, + "learning_rate": 1.51303673762489e-06, + "loss": 0.0824, + "step": 15817 + }, + { + "epoch": 6.62, + "learning_rate": 1.5121401909684663e-06, + "loss": 0.0783, + "step": 15818 + }, + { + "epoch": 6.62, + "learning_rate": 1.51124388828804e-06, + "loss": 0.0422, + "step": 15819 + }, + { + "epoch": 6.62, + "learning_rate": 1.5103478296093743e-06, + "loss": 0.0688, + "step": 15820 + }, + { + "epoch": 6.62, + "learning_rate": 1.509452014958227e-06, + "loss": 0.0618, + "step": 15821 + }, + { + "epoch": 6.62, + "learning_rate": 1.5085564443603473e-06, + "loss": 0.0895, + "step": 15822 + }, + { + "epoch": 6.62, + "learning_rate": 1.507661117841477e-06, + "loss": 0.0749, + "step": 15823 + }, + { + "epoch": 6.63, + "learning_rate": 1.5067660354273527e-06, + "loss": 0.0942, + "step": 15824 + }, + { + "epoch": 6.63, + "learning_rate": 1.5058711971437035e-06, + "loss": 0.0595, + "step": 15825 + }, + { + "epoch": 6.63, + "learning_rate": 1.504976603016247e-06, + "loss": 0.1109, + "step": 15826 + }, + { + "epoch": 6.63, + "learning_rate": 1.5040822530707e-06, + "loss": 0.0918, + "step": 15827 + }, + { + "epoch": 6.63, + "learning_rate": 1.5031881473327691e-06, + "loss": 0.0847, + "step": 15828 + }, + { + "epoch": 6.63, + "learning_rate": 1.5022942858281553e-06, + "loss": 0.0581, + "step": 15829 + }, + { + "epoch": 6.63, + "learning_rate": 1.5014006685825512e-06, + "loss": 0.0948, + "step": 15830 + }, + { + "epoch": 6.63, + "learning_rate": 1.500507295621646e-06, + "loss": 0.0823, + "step": 15831 + }, + { + "epoch": 6.63, + "learning_rate": 1.499614166971114e-06, + "loss": 0.0959, + "step": 15832 + }, + { + "epoch": 6.63, + "learning_rate": 1.4987212826566323e-06, + "loss": 0.0836, + "step": 15833 + }, + { + "epoch": 6.63, + "learning_rate": 1.4978286427038602e-06, + "loss": 0.0649, + "step": 15834 + }, + { + "epoch": 6.63, + "learning_rate": 1.4969362471384597e-06, + "loss": 0.1003, + "step": 15835 + }, + { + "epoch": 6.63, + "learning_rate": 1.4960440959860812e-06, + "loss": 0.0916, + "step": 15836 + }, + { + "epoch": 6.63, + "learning_rate": 1.4951521892723685e-06, + "loss": 0.0738, + "step": 15837 + }, + { + "epoch": 6.63, + "learning_rate": 1.4942605270229594e-06, + "loss": 0.0764, + "step": 15838 + }, + { + "epoch": 6.63, + "learning_rate": 1.4933691092634862e-06, + "loss": 0.0718, + "step": 15839 + }, + { + "epoch": 6.63, + "learning_rate": 1.4924779360195662e-06, + "loss": 0.0773, + "step": 15840 + }, + { + "epoch": 6.63, + "learning_rate": 1.4915870073168194e-06, + "loss": 0.0804, + "step": 15841 + }, + { + "epoch": 6.63, + "learning_rate": 1.4906963231808524e-06, + "loss": 0.0988, + "step": 15842 + }, + { + "epoch": 6.63, + "learning_rate": 1.4898058836372697e-06, + "loss": 0.0759, + "step": 15843 + }, + { + "epoch": 6.63, + "learning_rate": 1.4889156887116652e-06, + "loss": 0.0621, + "step": 15844 + }, + { + "epoch": 6.63, + "learning_rate": 1.4880257384296282e-06, + "loss": 0.0546, + "step": 15845 + }, + { + "epoch": 6.63, + "learning_rate": 1.487136032816736e-06, + "loss": 0.0796, + "step": 15846 + }, + { + "epoch": 6.63, + "learning_rate": 1.4862465718985653e-06, + "loss": 0.0723, + "step": 15847 + }, + { + "epoch": 6.64, + "learning_rate": 1.4853573557006807e-06, + "loss": 0.0789, + "step": 15848 + }, + { + "epoch": 6.64, + "learning_rate": 1.484468384248643e-06, + "loss": 0.0721, + "step": 15849 + }, + { + "epoch": 6.64, + "learning_rate": 1.4835796575680062e-06, + "loss": 0.0997, + "step": 15850 + }, + { + "epoch": 6.64, + "learning_rate": 1.4826911756843144e-06, + "loss": 0.095, + "step": 15851 + }, + { + "epoch": 6.64, + "learning_rate": 1.4818029386231082e-06, + "loss": 0.0576, + "step": 15852 + }, + { + "epoch": 6.64, + "learning_rate": 1.480914946409916e-06, + "loss": 0.0996, + "step": 15853 + }, + { + "epoch": 6.64, + "learning_rate": 1.4800271990702664e-06, + "loss": 0.0596, + "step": 15854 + }, + { + "epoch": 6.64, + "learning_rate": 1.4791396966296711e-06, + "loss": 0.0884, + "step": 15855 + }, + { + "epoch": 6.64, + "learning_rate": 1.4782524391136443e-06, + "loss": 0.0684, + "step": 15856 + }, + { + "epoch": 6.64, + "learning_rate": 1.47736542654769e-06, + "loss": 0.0636, + "step": 15857 + }, + { + "epoch": 6.64, + "learning_rate": 1.4764786589573033e-06, + "loss": 0.0788, + "step": 15858 + }, + { + "epoch": 6.64, + "learning_rate": 1.4755921363679758e-06, + "loss": 0.0973, + "step": 15859 + }, + { + "epoch": 6.64, + "learning_rate": 1.4747058588051854e-06, + "loss": 0.0763, + "step": 15860 + }, + { + "epoch": 6.64, + "learning_rate": 1.4738198262944103e-06, + "loss": 0.0712, + "step": 15861 + }, + { + "epoch": 6.64, + "learning_rate": 1.472934038861119e-06, + "loss": 0.0674, + "step": 15862 + }, + { + "epoch": 6.64, + "learning_rate": 1.4720484965307714e-06, + "loss": 0.1018, + "step": 15863 + }, + { + "epoch": 6.64, + "learning_rate": 1.4711631993288222e-06, + "loss": 0.1098, + "step": 15864 + }, + { + "epoch": 6.64, + "learning_rate": 1.4702781472807181e-06, + "loss": 0.0737, + "step": 15865 + }, + { + "epoch": 6.64, + "learning_rate": 1.469393340411902e-06, + "loss": 0.0964, + "step": 15866 + }, + { + "epoch": 6.64, + "learning_rate": 1.4685087787478025e-06, + "loss": 0.1032, + "step": 15867 + }, + { + "epoch": 6.64, + "learning_rate": 1.4676244623138468e-06, + "loss": 0.0856, + "step": 15868 + }, + { + "epoch": 6.64, + "learning_rate": 1.466740391135455e-06, + "loss": 0.1217, + "step": 15869 + }, + { + "epoch": 6.64, + "learning_rate": 1.4658565652380373e-06, + "loss": 0.0789, + "step": 15870 + }, + { + "epoch": 6.64, + "learning_rate": 1.4649729846470006e-06, + "loss": 0.073, + "step": 15871 + }, + { + "epoch": 6.65, + "learning_rate": 1.4640896493877443e-06, + "loss": 0.0972, + "step": 15872 + }, + { + "epoch": 6.65, + "learning_rate": 1.4632065594856527e-06, + "loss": 0.0651, + "step": 15873 + }, + { + "epoch": 6.65, + "learning_rate": 1.462323714966114e-06, + "loss": 0.1199, + "step": 15874 + }, + { + "epoch": 6.65, + "learning_rate": 1.4614411158545038e-06, + "loss": 0.1062, + "step": 15875 + }, + { + "epoch": 6.65, + "learning_rate": 1.460558762176194e-06, + "loss": 0.0922, + "step": 15876 + }, + { + "epoch": 6.65, + "learning_rate": 1.4596766539565432e-06, + "loss": 0.0987, + "step": 15877 + }, + { + "epoch": 6.65, + "learning_rate": 1.4587947912209076e-06, + "loss": 0.0748, + "step": 15878 + }, + { + "epoch": 6.65, + "learning_rate": 1.4579131739946361e-06, + "loss": 0.0895, + "step": 15879 + }, + { + "epoch": 6.65, + "learning_rate": 1.4570318023030738e-06, + "loss": 0.0804, + "step": 15880 + }, + { + "epoch": 6.65, + "learning_rate": 1.4561506761715482e-06, + "loss": 0.0968, + "step": 15881 + }, + { + "epoch": 6.65, + "learning_rate": 1.4552697956253903e-06, + "loss": 0.0778, + "step": 15882 + }, + { + "epoch": 6.65, + "learning_rate": 1.4543891606899196e-06, + "loss": 0.0658, + "step": 15883 + }, + { + "epoch": 6.65, + "learning_rate": 1.4535087713904484e-06, + "loss": 0.1075, + "step": 15884 + }, + { + "epoch": 6.65, + "learning_rate": 1.452628627752285e-06, + "loss": 0.0736, + "step": 15885 + }, + { + "epoch": 6.65, + "learning_rate": 1.4517487298007283e-06, + "loss": 0.086, + "step": 15886 + }, + { + "epoch": 6.65, + "learning_rate": 1.4508690775610656e-06, + "loss": 0.0852, + "step": 15887 + }, + { + "epoch": 6.65, + "learning_rate": 1.4499896710585859e-06, + "loss": 0.0803, + "step": 15888 + }, + { + "epoch": 6.65, + "learning_rate": 1.4491105103185666e-06, + "loss": 0.095, + "step": 15889 + }, + { + "epoch": 6.65, + "learning_rate": 1.4482315953662762e-06, + "loss": 0.1004, + "step": 15890 + }, + { + "epoch": 6.65, + "learning_rate": 1.4473529262269814e-06, + "loss": 0.0824, + "step": 15891 + }, + { + "epoch": 6.65, + "learning_rate": 1.4464745029259365e-06, + "loss": 0.0721, + "step": 15892 + }, + { + "epoch": 6.65, + "learning_rate": 1.4455963254883942e-06, + "loss": 0.0821, + "step": 15893 + }, + { + "epoch": 6.65, + "learning_rate": 1.4447183939395925e-06, + "loss": 0.0898, + "step": 15894 + }, + { + "epoch": 6.65, + "learning_rate": 1.4438407083047679e-06, + "loss": 0.092, + "step": 15895 + }, + { + "epoch": 6.66, + "learning_rate": 1.4429632686091498e-06, + "loss": 0.0992, + "step": 15896 + }, + { + "epoch": 6.66, + "learning_rate": 1.4420860748779596e-06, + "loss": 0.0756, + "step": 15897 + }, + { + "epoch": 6.66, + "learning_rate": 1.4412091271364105e-06, + "loss": 0.0674, + "step": 15898 + }, + { + "epoch": 6.66, + "learning_rate": 1.4403324254097128e-06, + "loss": 0.0817, + "step": 15899 + }, + { + "epoch": 6.66, + "learning_rate": 1.4394559697230637e-06, + "loss": 0.0912, + "step": 15900 + }, + { + "epoch": 6.66, + "learning_rate": 1.4385797601016538e-06, + "loss": 0.0836, + "step": 15901 + }, + { + "epoch": 6.66, + "learning_rate": 1.437703796570672e-06, + "loss": 0.0662, + "step": 15902 + }, + { + "epoch": 6.66, + "learning_rate": 1.4368280791552957e-06, + "loss": 0.1097, + "step": 15903 + }, + { + "epoch": 6.66, + "learning_rate": 1.4359526078806985e-06, + "loss": 0.0821, + "step": 15904 + }, + { + "epoch": 6.66, + "learning_rate": 1.4350773827720432e-06, + "loss": 0.0611, + "step": 15905 + }, + { + "epoch": 6.66, + "learning_rate": 1.4342024038544878e-06, + "loss": 0.1006, + "step": 15906 + }, + { + "epoch": 6.66, + "learning_rate": 1.4333276711531863e-06, + "loss": 0.0588, + "step": 15907 + }, + { + "epoch": 6.66, + "learning_rate": 1.4324531846932766e-06, + "loss": 0.0845, + "step": 15908 + }, + { + "epoch": 6.66, + "learning_rate": 1.4315789444998973e-06, + "loss": 0.0851, + "step": 15909 + }, + { + "epoch": 6.66, + "learning_rate": 1.4307049505981773e-06, + "loss": 0.0803, + "step": 15910 + }, + { + "epoch": 6.66, + "learning_rate": 1.4298312030132388e-06, + "loss": 0.0909, + "step": 15911 + }, + { + "epoch": 6.66, + "learning_rate": 1.4289577017701984e-06, + "loss": 0.1119, + "step": 15912 + }, + { + "epoch": 6.66, + "learning_rate": 1.4280844468941647e-06, + "loss": 0.0861, + "step": 15913 + }, + { + "epoch": 6.66, + "learning_rate": 1.4272114384102353e-06, + "loss": 0.1005, + "step": 15914 + }, + { + "epoch": 6.66, + "learning_rate": 1.426338676343505e-06, + "loss": 0.0662, + "step": 15915 + }, + { + "epoch": 6.66, + "learning_rate": 1.4254661607190624e-06, + "loss": 0.0832, + "step": 15916 + }, + { + "epoch": 6.66, + "learning_rate": 1.424593891561985e-06, + "loss": 0.0695, + "step": 15917 + }, + { + "epoch": 6.66, + "learning_rate": 1.4237218688973475e-06, + "loss": 0.0699, + "step": 15918 + }, + { + "epoch": 6.66, + "learning_rate": 1.4228500927502164e-06, + "loss": 0.0661, + "step": 15919 + }, + { + "epoch": 6.67, + "learning_rate": 1.4219785631456452e-06, + "loss": 0.0687, + "step": 15920 + }, + { + "epoch": 6.67, + "learning_rate": 1.421107280108689e-06, + "loss": 0.066, + "step": 15921 + }, + { + "epoch": 6.67, + "learning_rate": 1.4202362436643924e-06, + "loss": 0.1135, + "step": 15922 + }, + { + "epoch": 6.67, + "learning_rate": 1.41936545383779e-06, + "loss": 0.0978, + "step": 15923 + }, + { + "epoch": 6.67, + "learning_rate": 1.4184949106539137e-06, + "loss": 0.0523, + "step": 15924 + }, + { + "epoch": 6.67, + "learning_rate": 1.417624614137786e-06, + "loss": 0.105, + "step": 15925 + }, + { + "epoch": 6.67, + "learning_rate": 1.416754564314422e-06, + "loss": 0.085, + "step": 15926 + }, + { + "epoch": 6.67, + "learning_rate": 1.4158847612088344e-06, + "loss": 0.0774, + "step": 15927 + }, + { + "epoch": 6.67, + "learning_rate": 1.4150152048460198e-06, + "loss": 0.0513, + "step": 15928 + }, + { + "epoch": 6.67, + "learning_rate": 1.4141458952509745e-06, + "loss": 0.086, + "step": 15929 + }, + { + "epoch": 6.67, + "learning_rate": 1.4132768324486868e-06, + "loss": 0.0827, + "step": 15930 + }, + { + "epoch": 6.67, + "learning_rate": 1.4124080164641362e-06, + "loss": 0.0576, + "step": 15931 + }, + { + "epoch": 6.67, + "learning_rate": 1.4115394473222977e-06, + "loss": 0.052, + "step": 15932 + }, + { + "epoch": 6.67, + "learning_rate": 1.4106711250481353e-06, + "loss": 0.0713, + "step": 15933 + }, + { + "epoch": 6.67, + "learning_rate": 1.4098030496666126e-06, + "loss": 0.086, + "step": 15934 + }, + { + "epoch": 6.67, + "learning_rate": 1.4089352212026752e-06, + "loss": 0.0926, + "step": 15935 + }, + { + "epoch": 6.67, + "learning_rate": 1.408067639681272e-06, + "loss": 0.1034, + "step": 15936 + }, + { + "epoch": 6.67, + "learning_rate": 1.4072003051273408e-06, + "loss": 0.0879, + "step": 15937 + }, + { + "epoch": 6.67, + "learning_rate": 1.4063332175658107e-06, + "loss": 0.0597, + "step": 15938 + }, + { + "epoch": 6.67, + "learning_rate": 1.405466377021607e-06, + "loss": 0.0823, + "step": 15939 + }, + { + "epoch": 6.67, + "learning_rate": 1.404599783519648e-06, + "loss": 0.0752, + "step": 15940 + }, + { + "epoch": 6.67, + "learning_rate": 1.4037334370848376e-06, + "loss": 0.0663, + "step": 15941 + }, + { + "epoch": 6.67, + "learning_rate": 1.4028673377420821e-06, + "loss": 0.0863, + "step": 15942 + }, + { + "epoch": 6.67, + "learning_rate": 1.4020014855162755e-06, + "loss": 0.0547, + "step": 15943 + }, + { + "epoch": 6.68, + "learning_rate": 1.4011358804323095e-06, + "loss": 0.0713, + "step": 15944 + }, + { + "epoch": 6.68, + "learning_rate": 1.4002705225150593e-06, + "loss": 0.0967, + "step": 15945 + }, + { + "epoch": 6.68, + "learning_rate": 1.3994054117894018e-06, + "loss": 0.0731, + "step": 15946 + }, + { + "epoch": 6.68, + "learning_rate": 1.398540548280203e-06, + "loss": 0.0999, + "step": 15947 + }, + { + "epoch": 6.68, + "learning_rate": 1.3976759320123268e-06, + "loss": 0.0824, + "step": 15948 + }, + { + "epoch": 6.68, + "learning_rate": 1.3968115630106193e-06, + "loss": 0.0903, + "step": 15949 + }, + { + "epoch": 6.68, + "learning_rate": 1.3959474412999296e-06, + "loss": 0.0937, + "step": 15950 + }, + { + "epoch": 6.68, + "learning_rate": 1.3950835669050955e-06, + "loss": 0.0559, + "step": 15951 + }, + { + "epoch": 6.68, + "learning_rate": 1.394219939850948e-06, + "loss": 0.1039, + "step": 15952 + }, + { + "epoch": 6.68, + "learning_rate": 1.3933565601623123e-06, + "loss": 0.0779, + "step": 15953 + }, + { + "epoch": 6.68, + "learning_rate": 1.3924934278640068e-06, + "loss": 0.0767, + "step": 15954 + }, + { + "epoch": 6.68, + "learning_rate": 1.3916305429808375e-06, + "loss": 0.0572, + "step": 15955 + }, + { + "epoch": 6.68, + "learning_rate": 1.3907679055376089e-06, + "loss": 0.0967, + "step": 15956 + }, + { + "epoch": 6.68, + "learning_rate": 1.3899055155591167e-06, + "loss": 0.0821, + "step": 15957 + }, + { + "epoch": 6.68, + "learning_rate": 1.3890433730701513e-06, + "loss": 0.0777, + "step": 15958 + }, + { + "epoch": 6.68, + "learning_rate": 1.3881814780954916e-06, + "loss": 0.0568, + "step": 15959 + }, + { + "epoch": 6.68, + "learning_rate": 1.3873198306599168e-06, + "loss": 0.0594, + "step": 15960 + }, + { + "epoch": 6.68, + "learning_rate": 1.3864584307881878e-06, + "loss": 0.0879, + "step": 15961 + }, + { + "epoch": 6.68, + "learning_rate": 1.3855972785050675e-06, + "loss": 0.0866, + "step": 15962 + }, + { + "epoch": 6.68, + "learning_rate": 1.3847363738353091e-06, + "loss": 0.0754, + "step": 15963 + }, + { + "epoch": 6.68, + "learning_rate": 1.3838757168036598e-06, + "loss": 0.0802, + "step": 15964 + }, + { + "epoch": 6.68, + "learning_rate": 1.3830153074348562e-06, + "loss": 0.1218, + "step": 15965 + }, + { + "epoch": 6.68, + "learning_rate": 1.382155145753631e-06, + "loss": 0.0689, + "step": 15966 + }, + { + "epoch": 6.68, + "learning_rate": 1.3812952317847116e-06, + "loss": 0.0817, + "step": 15967 + }, + { + "epoch": 6.69, + "learning_rate": 1.3804355655528123e-06, + "loss": 0.0786, + "step": 15968 + }, + { + "epoch": 6.69, + "learning_rate": 1.3795761470826408e-06, + "loss": 0.0705, + "step": 15969 + }, + { + "epoch": 6.69, + "learning_rate": 1.3787169763989038e-06, + "loss": 0.0808, + "step": 15970 + }, + { + "epoch": 6.69, + "learning_rate": 1.3778580535262975e-06, + "loss": 0.0696, + "step": 15971 + }, + { + "epoch": 6.69, + "learning_rate": 1.3769993784895096e-06, + "loss": 0.0934, + "step": 15972 + }, + { + "epoch": 6.69, + "learning_rate": 1.376140951313224e-06, + "loss": 0.0787, + "step": 15973 + }, + { + "epoch": 6.69, + "learning_rate": 1.3752827720221128e-06, + "loss": 0.0771, + "step": 15974 + }, + { + "epoch": 6.69, + "learning_rate": 1.3744248406408467e-06, + "loss": 0.0788, + "step": 15975 + }, + { + "epoch": 6.69, + "learning_rate": 1.3735671571940835e-06, + "loss": 0.0711, + "step": 15976 + }, + { + "epoch": 6.69, + "learning_rate": 1.3727097217064767e-06, + "loss": 0.0902, + "step": 15977 + }, + { + "epoch": 6.69, + "learning_rate": 1.3718525342026733e-06, + "loss": 0.0761, + "step": 15978 + }, + { + "epoch": 6.69, + "learning_rate": 1.3709955947073117e-06, + "loss": 0.0823, + "step": 15979 + }, + { + "epoch": 6.69, + "learning_rate": 1.3701389032450252e-06, + "loss": 0.0721, + "step": 15980 + }, + { + "epoch": 6.69, + "learning_rate": 1.3692824598404398e-06, + "loss": 0.0955, + "step": 15981 + }, + { + "epoch": 6.69, + "learning_rate": 1.3684262645181702e-06, + "loss": 0.0907, + "step": 15982 + }, + { + "epoch": 6.69, + "learning_rate": 1.3675703173028266e-06, + "loss": 0.0557, + "step": 15983 + }, + { + "epoch": 6.69, + "learning_rate": 1.366714618219015e-06, + "loss": 0.0809, + "step": 15984 + }, + { + "epoch": 6.69, + "learning_rate": 1.3658591672913301e-06, + "loss": 0.0915, + "step": 15985 + }, + { + "epoch": 6.69, + "learning_rate": 1.3650039645443614e-06, + "loss": 0.0763, + "step": 15986 + }, + { + "epoch": 6.69, + "learning_rate": 1.3641490100026943e-06, + "loss": 0.0912, + "step": 15987 + }, + { + "epoch": 6.69, + "learning_rate": 1.3632943036908974e-06, + "loss": 0.0533, + "step": 15988 + }, + { + "epoch": 6.69, + "learning_rate": 1.362439845633542e-06, + "loss": 0.1142, + "step": 15989 + }, + { + "epoch": 6.69, + "learning_rate": 1.3615856358551904e-06, + "loss": 0.0713, + "step": 15990 + }, + { + "epoch": 6.69, + "learning_rate": 1.3607316743803922e-06, + "loss": 0.0435, + "step": 15991 + }, + { + "epoch": 6.7, + "learning_rate": 1.3598779612336965e-06, + "loss": 0.0534, + "step": 15992 + }, + { + "epoch": 6.7, + "learning_rate": 1.3590244964396403e-06, + "loss": 0.0946, + "step": 15993 + }, + { + "epoch": 6.7, + "learning_rate": 1.3581712800227576e-06, + "loss": 0.0693, + "step": 15994 + }, + { + "epoch": 6.7, + "learning_rate": 1.357318312007575e-06, + "loss": 0.0973, + "step": 15995 + }, + { + "epoch": 6.7, + "learning_rate": 1.3564655924186066e-06, + "loss": 0.0871, + "step": 15996 + }, + { + "epoch": 6.7, + "learning_rate": 1.3556131212803635e-06, + "loss": 0.0774, + "step": 15997 + }, + { + "epoch": 6.7, + "learning_rate": 1.354760898617351e-06, + "loss": 0.074, + "step": 15998 + }, + { + "epoch": 6.7, + "learning_rate": 1.3539089244540648e-06, + "loss": 0.0737, + "step": 15999 + }, + { + "epoch": 6.7, + "learning_rate": 1.3530571988149943e-06, + "loss": 0.0847, + "step": 16000 + }, + { + "epoch": 6.7, + "learning_rate": 1.3522057217246232e-06, + "loss": 0.0977, + "step": 16001 + }, + { + "epoch": 6.7, + "learning_rate": 1.351354493207423e-06, + "loss": 0.0814, + "step": 16002 + }, + { + "epoch": 6.7, + "learning_rate": 1.350503513287863e-06, + "loss": 0.0802, + "step": 16003 + }, + { + "epoch": 6.7, + "learning_rate": 1.3496527819904038e-06, + "loss": 0.0554, + "step": 16004 + }, + { + "epoch": 6.7, + "learning_rate": 1.348802299339499e-06, + "loss": 0.0923, + "step": 16005 + }, + { + "epoch": 6.7, + "learning_rate": 1.3479520653595957e-06, + "loss": 0.0737, + "step": 16006 + }, + { + "epoch": 6.7, + "learning_rate": 1.3471020800751322e-06, + "loss": 0.1104, + "step": 16007 + }, + { + "epoch": 6.7, + "learning_rate": 1.3462523435105436e-06, + "loss": 0.0921, + "step": 16008 + }, + { + "epoch": 6.7, + "learning_rate": 1.3454028556902498e-06, + "loss": 0.0635, + "step": 16009 + }, + { + "epoch": 6.7, + "learning_rate": 1.3445536166386708e-06, + "loss": 0.0707, + "step": 16010 + }, + { + "epoch": 6.7, + "learning_rate": 1.3437046263802178e-06, + "loss": 0.0919, + "step": 16011 + }, + { + "epoch": 6.7, + "learning_rate": 1.342855884939296e-06, + "loss": 0.0601, + "step": 16012 + }, + { + "epoch": 6.7, + "learning_rate": 1.3420073923402976e-06, + "loss": 0.0563, + "step": 16013 + }, + { + "epoch": 6.7, + "learning_rate": 1.3411591486076136e-06, + "loss": 0.0855, + "step": 16014 + }, + { + "epoch": 6.71, + "learning_rate": 1.3403111537656255e-06, + "loss": 0.0811, + "step": 16015 + }, + { + "epoch": 6.71, + "learning_rate": 1.3394634078387115e-06, + "loss": 0.0934, + "step": 16016 + }, + { + "epoch": 6.71, + "learning_rate": 1.3386159108512341e-06, + "loss": 0.0717, + "step": 16017 + }, + { + "epoch": 6.71, + "learning_rate": 1.3377686628275567e-06, + "loss": 0.0788, + "step": 16018 + }, + { + "epoch": 6.71, + "learning_rate": 1.3369216637920323e-06, + "loss": 0.0582, + "step": 16019 + }, + { + "epoch": 6.71, + "learning_rate": 1.3360749137690066e-06, + "loss": 0.0806, + "step": 16020 + }, + { + "epoch": 6.71, + "learning_rate": 1.3352284127828207e-06, + "loss": 0.0704, + "step": 16021 + }, + { + "epoch": 6.71, + "learning_rate": 1.3343821608578057e-06, + "loss": 0.0473, + "step": 16022 + }, + { + "epoch": 6.71, + "learning_rate": 1.3335361580182848e-06, + "loss": 0.0612, + "step": 16023 + }, + { + "epoch": 6.71, + "learning_rate": 1.3326904042885757e-06, + "loss": 0.0891, + "step": 16024 + }, + { + "epoch": 6.71, + "learning_rate": 1.33184489969299e-06, + "loss": 0.078, + "step": 16025 + }, + { + "epoch": 6.71, + "learning_rate": 1.3309996442558304e-06, + "loss": 0.0681, + "step": 16026 + }, + { + "epoch": 6.71, + "learning_rate": 1.3301546380013942e-06, + "loss": 0.1014, + "step": 16027 + }, + { + "epoch": 6.71, + "learning_rate": 1.3293098809539706e-06, + "loss": 0.0683, + "step": 16028 + }, + { + "epoch": 6.71, + "learning_rate": 1.3284653731378382e-06, + "loss": 0.0842, + "step": 16029 + }, + { + "epoch": 6.71, + "learning_rate": 1.3276211145772743e-06, + "loss": 0.0905, + "step": 16030 + }, + { + "epoch": 6.71, + "learning_rate": 1.326777105296546e-06, + "loss": 0.0827, + "step": 16031 + }, + { + "epoch": 6.71, + "learning_rate": 1.3259333453199119e-06, + "loss": 0.085, + "step": 16032 + }, + { + "epoch": 6.71, + "learning_rate": 1.3250898346716279e-06, + "loss": 0.1026, + "step": 16033 + }, + { + "epoch": 6.71, + "learning_rate": 1.324246573375938e-06, + "loss": 0.1129, + "step": 16034 + }, + { + "epoch": 6.71, + "learning_rate": 1.323403561457084e-06, + "loss": 0.092, + "step": 16035 + }, + { + "epoch": 6.71, + "learning_rate": 1.3225607989392953e-06, + "loss": 0.1044, + "step": 16036 + }, + { + "epoch": 6.71, + "learning_rate": 1.3217182858467936e-06, + "loss": 0.0621, + "step": 16037 + }, + { + "epoch": 6.71, + "learning_rate": 1.3208760222037986e-06, + "loss": 0.0796, + "step": 16038 + }, + { + "epoch": 6.72, + "learning_rate": 1.3200340080345208e-06, + "loss": 0.0831, + "step": 16039 + }, + { + "epoch": 6.72, + "learning_rate": 1.3191922433631632e-06, + "loss": 0.0672, + "step": 16040 + }, + { + "epoch": 6.72, + "learning_rate": 1.318350728213922e-06, + "loss": 0.0612, + "step": 16041 + }, + { + "epoch": 6.72, + "learning_rate": 1.3175094626109864e-06, + "loss": 0.0675, + "step": 16042 + }, + { + "epoch": 6.72, + "learning_rate": 1.316668446578535e-06, + "loss": 0.093, + "step": 16043 + }, + { + "epoch": 6.72, + "learning_rate": 1.3158276801407432e-06, + "loss": 0.0971, + "step": 16044 + }, + { + "epoch": 6.72, + "learning_rate": 1.314987163321778e-06, + "loss": 0.0573, + "step": 16045 + }, + { + "epoch": 6.72, + "learning_rate": 1.3141468961458003e-06, + "loss": 0.0859, + "step": 16046 + }, + { + "epoch": 6.72, + "learning_rate": 1.313306878636963e-06, + "loss": 0.0638, + "step": 16047 + }, + { + "epoch": 6.72, + "learning_rate": 1.3124671108194108e-06, + "loss": 0.0778, + "step": 16048 + }, + { + "epoch": 6.72, + "learning_rate": 1.3116275927172838e-06, + "loss": 0.0863, + "step": 16049 + }, + { + "epoch": 6.72, + "learning_rate": 1.3107883243547103e-06, + "loss": 0.0594, + "step": 16050 + }, + { + "epoch": 6.72, + "learning_rate": 1.3099493057558155e-06, + "loss": 0.0832, + "step": 16051 + }, + { + "epoch": 6.72, + "learning_rate": 1.3091105369447166e-06, + "loss": 0.0584, + "step": 16052 + }, + { + "epoch": 6.72, + "learning_rate": 1.3082720179455233e-06, + "loss": 0.1038, + "step": 16053 + }, + { + "epoch": 6.72, + "learning_rate": 1.3074337487823374e-06, + "loss": 0.0928, + "step": 16054 + }, + { + "epoch": 6.72, + "learning_rate": 1.3065957294792576e-06, + "loss": 0.0696, + "step": 16055 + }, + { + "epoch": 6.72, + "learning_rate": 1.3057579600603666e-06, + "loss": 0.0758, + "step": 16056 + }, + { + "epoch": 6.72, + "learning_rate": 1.3049204405497473e-06, + "loss": 0.0382, + "step": 16057 + }, + { + "epoch": 6.72, + "learning_rate": 1.3040831709714774e-06, + "loss": 0.0807, + "step": 16058 + }, + { + "epoch": 6.72, + "learning_rate": 1.3032461513496176e-06, + "loss": 0.0612, + "step": 16059 + }, + { + "epoch": 6.72, + "learning_rate": 1.3024093817082295e-06, + "loss": 0.0789, + "step": 16060 + }, + { + "epoch": 6.72, + "learning_rate": 1.3015728620713653e-06, + "loss": 0.0679, + "step": 16061 + }, + { + "epoch": 6.72, + "learning_rate": 1.300736592463071e-06, + "loss": 0.1048, + "step": 16062 + }, + { + "epoch": 6.73, + "learning_rate": 1.2999005729073855e-06, + "loss": 0.0882, + "step": 16063 + }, + { + "epoch": 6.73, + "learning_rate": 1.299064803428336e-06, + "loss": 0.0785, + "step": 16064 + }, + { + "epoch": 6.73, + "learning_rate": 1.2982292840499477e-06, + "loss": 0.0983, + "step": 16065 + }, + { + "epoch": 6.73, + "learning_rate": 1.2973940147962361e-06, + "loss": 0.0918, + "step": 16066 + }, + { + "epoch": 6.73, + "learning_rate": 1.2965589956912117e-06, + "loss": 0.0908, + "step": 16067 + }, + { + "epoch": 6.73, + "learning_rate": 1.2957242267588754e-06, + "loss": 0.069, + "step": 16068 + }, + { + "epoch": 6.73, + "learning_rate": 1.2948897080232248e-06, + "loss": 0.082, + "step": 16069 + }, + { + "epoch": 6.73, + "learning_rate": 1.2940554395082416e-06, + "loss": 0.0861, + "step": 16070 + }, + { + "epoch": 6.73, + "learning_rate": 1.2932214212379102e-06, + "loss": 0.094, + "step": 16071 + }, + { + "epoch": 6.73, + "learning_rate": 1.2923876532362022e-06, + "loss": 0.0741, + "step": 16072 + }, + { + "epoch": 6.73, + "learning_rate": 1.2915541355270845e-06, + "loss": 0.0595, + "step": 16073 + }, + { + "epoch": 6.73, + "learning_rate": 1.290720868134515e-06, + "loss": 0.0915, + "step": 16074 + }, + { + "epoch": 6.73, + "learning_rate": 1.2898878510824453e-06, + "loss": 0.0738, + "step": 16075 + }, + { + "epoch": 6.73, + "learning_rate": 1.2890550843948234e-06, + "loss": 0.0591, + "step": 16076 + }, + { + "epoch": 6.73, + "learning_rate": 1.2882225680955806e-06, + "loss": 0.0956, + "step": 16077 + }, + { + "epoch": 6.73, + "learning_rate": 1.2873903022086487e-06, + "loss": 0.0882, + "step": 16078 + }, + { + "epoch": 6.73, + "learning_rate": 1.2865582867579517e-06, + "loss": 0.0676, + "step": 16079 + }, + { + "epoch": 6.73, + "learning_rate": 1.2857265217674063e-06, + "loss": 0.0733, + "step": 16080 + }, + { + "epoch": 6.73, + "learning_rate": 1.2848950072609179e-06, + "loss": 0.0873, + "step": 16081 + }, + { + "epoch": 6.73, + "learning_rate": 1.2840637432623881e-06, + "loss": 0.0824, + "step": 16082 + }, + { + "epoch": 6.73, + "learning_rate": 1.2832327297957136e-06, + "loss": 0.057, + "step": 16083 + }, + { + "epoch": 6.73, + "learning_rate": 1.2824019668847775e-06, + "loss": 0.097, + "step": 16084 + }, + { + "epoch": 6.73, + "learning_rate": 1.2815714545534608e-06, + "loss": 0.0892, + "step": 16085 + }, + { + "epoch": 6.73, + "learning_rate": 1.2807411928256363e-06, + "loss": 0.0875, + "step": 16086 + }, + { + "epoch": 6.74, + "learning_rate": 1.2799111817251686e-06, + "loss": 0.0748, + "step": 16087 + }, + { + "epoch": 6.74, + "learning_rate": 1.2790814212759161e-06, + "loss": 0.0777, + "step": 16088 + }, + { + "epoch": 6.74, + "learning_rate": 1.27825191150173e-06, + "loss": 0.053, + "step": 16089 + }, + { + "epoch": 6.74, + "learning_rate": 1.2774226524264545e-06, + "loss": 0.0814, + "step": 16090 + }, + { + "epoch": 6.74, + "learning_rate": 1.2765936440739225e-06, + "loss": 0.0773, + "step": 16091 + }, + { + "epoch": 6.74, + "learning_rate": 1.275764886467965e-06, + "loss": 0.0506, + "step": 16092 + }, + { + "epoch": 6.74, + "learning_rate": 1.2749363796324043e-06, + "loss": 0.0743, + "step": 16093 + }, + { + "epoch": 6.74, + "learning_rate": 1.2741081235910547e-06, + "loss": 0.0992, + "step": 16094 + }, + { + "epoch": 6.74, + "learning_rate": 1.2732801183677246e-06, + "loss": 0.0694, + "step": 16095 + }, + { + "epoch": 6.74, + "learning_rate": 1.2724523639862153e-06, + "loss": 0.0733, + "step": 16096 + }, + { + "epoch": 6.74, + "learning_rate": 1.2716248604703153e-06, + "loss": 0.0818, + "step": 16097 + }, + { + "epoch": 6.74, + "learning_rate": 1.2707976078438132e-06, + "loss": 0.1016, + "step": 16098 + }, + { + "epoch": 6.74, + "learning_rate": 1.2699706061304884e-06, + "loss": 0.0834, + "step": 16099 + }, + { + "epoch": 6.74, + "learning_rate": 1.2691438553541112e-06, + "loss": 0.0431, + "step": 16100 + }, + { + "epoch": 6.74, + "learning_rate": 1.2683173555384476e-06, + "loss": 0.0985, + "step": 16101 + }, + { + "epoch": 6.74, + "learning_rate": 1.2674911067072514e-06, + "loss": 0.0744, + "step": 16102 + }, + { + "epoch": 6.74, + "learning_rate": 1.2666651088842773e-06, + "loss": 0.0862, + "step": 16103 + }, + { + "epoch": 6.74, + "learning_rate": 1.2658393620932652e-06, + "loss": 0.0848, + "step": 16104 + }, + { + "epoch": 6.74, + "learning_rate": 1.265013866357947e-06, + "loss": 0.0941, + "step": 16105 + }, + { + "epoch": 6.74, + "learning_rate": 1.264188621702055e-06, + "loss": 0.0776, + "step": 16106 + }, + { + "epoch": 6.74, + "learning_rate": 1.2633636281493078e-06, + "loss": 0.0564, + "step": 16107 + }, + { + "epoch": 6.74, + "learning_rate": 1.262538885723421e-06, + "loss": 0.0876, + "step": 16108 + }, + { + "epoch": 6.74, + "learning_rate": 1.2617143944481004e-06, + "loss": 0.093, + "step": 16109 + }, + { + "epoch": 6.74, + "learning_rate": 1.2608901543470475e-06, + "loss": 0.0993, + "step": 16110 + }, + { + "epoch": 6.75, + "learning_rate": 1.2600661654439505e-06, + "loss": 0.1058, + "step": 16111 + }, + { + "epoch": 6.75, + "learning_rate": 1.2592424277624948e-06, + "loss": 0.1112, + "step": 16112 + }, + { + "epoch": 6.75, + "learning_rate": 1.2584189413263604e-06, + "loss": 0.0813, + "step": 16113 + }, + { + "epoch": 6.75, + "learning_rate": 1.2575957061592159e-06, + "loss": 0.0847, + "step": 16114 + }, + { + "epoch": 6.75, + "learning_rate": 1.2567727222847237e-06, + "loss": 0.0952, + "step": 16115 + }, + { + "epoch": 6.75, + "learning_rate": 1.2559499897265426e-06, + "loss": 0.0755, + "step": 16116 + }, + { + "epoch": 6.75, + "learning_rate": 1.2551275085083215e-06, + "loss": 0.0503, + "step": 16117 + }, + { + "epoch": 6.75, + "learning_rate": 1.254305278653698e-06, + "loss": 0.0829, + "step": 16118 + }, + { + "epoch": 6.75, + "learning_rate": 1.2534833001863078e-06, + "loss": 0.0863, + "step": 16119 + }, + { + "epoch": 6.75, + "learning_rate": 1.2526615731297787e-06, + "loss": 0.0845, + "step": 16120 + }, + { + "epoch": 6.75, + "learning_rate": 1.2518400975077316e-06, + "loss": 0.0666, + "step": 16121 + }, + { + "epoch": 6.75, + "learning_rate": 1.2510188733437767e-06, + "loss": 0.0918, + "step": 16122 + }, + { + "epoch": 6.75, + "learning_rate": 1.250197900661523e-06, + "loss": 0.0965, + "step": 16123 + }, + { + "epoch": 6.75, + "learning_rate": 1.2493771794845644e-06, + "loss": 0.0821, + "step": 16124 + }, + { + "epoch": 6.75, + "learning_rate": 1.2485567098364937e-06, + "loss": 0.0706, + "step": 16125 + }, + { + "epoch": 6.75, + "learning_rate": 1.2477364917408964e-06, + "loss": 0.0577, + "step": 16126 + }, + { + "epoch": 6.75, + "learning_rate": 1.2469165252213445e-06, + "loss": 0.0698, + "step": 16127 + }, + { + "epoch": 6.75, + "learning_rate": 1.2460968103014104e-06, + "loss": 0.0853, + "step": 16128 + }, + { + "epoch": 6.75, + "learning_rate": 1.2452773470046543e-06, + "loss": 0.0955, + "step": 16129 + }, + { + "epoch": 6.75, + "learning_rate": 1.2444581353546326e-06, + "loss": 0.0921, + "step": 16130 + }, + { + "epoch": 6.75, + "learning_rate": 1.2436391753748955e-06, + "loss": 0.0521, + "step": 16131 + }, + { + "epoch": 6.75, + "learning_rate": 1.2428204670889766e-06, + "loss": 0.0614, + "step": 16132 + }, + { + "epoch": 6.75, + "learning_rate": 1.2420020105204122e-06, + "loss": 0.0705, + "step": 16133 + }, + { + "epoch": 6.75, + "learning_rate": 1.2411838056927295e-06, + "loss": 0.0681, + "step": 16134 + }, + { + "epoch": 6.76, + "learning_rate": 1.2403658526294448e-06, + "loss": 0.0632, + "step": 16135 + }, + { + "epoch": 6.76, + "learning_rate": 1.2395481513540719e-06, + "loss": 0.0803, + "step": 16136 + }, + { + "epoch": 6.76, + "learning_rate": 1.238730701890114e-06, + "loss": 0.0879, + "step": 16137 + }, + { + "epoch": 6.76, + "learning_rate": 1.2379135042610669e-06, + "loss": 0.0931, + "step": 16138 + }, + { + "epoch": 6.76, + "learning_rate": 1.2370965584904194e-06, + "loss": 0.078, + "step": 16139 + }, + { + "epoch": 6.76, + "learning_rate": 1.2362798646016561e-06, + "loss": 0.0907, + "step": 16140 + }, + { + "epoch": 6.76, + "learning_rate": 1.2354634226182506e-06, + "loss": 0.0921, + "step": 16141 + }, + { + "epoch": 6.76, + "learning_rate": 1.2346472325636717e-06, + "loss": 0.0967, + "step": 16142 + }, + { + "epoch": 6.76, + "learning_rate": 1.2338312944613828e-06, + "loss": 0.0937, + "step": 16143 + }, + { + "epoch": 6.76, + "learning_rate": 1.233015608334831e-06, + "loss": 0.0802, + "step": 16144 + }, + { + "epoch": 6.76, + "learning_rate": 1.2322001742074674e-06, + "loss": 0.0816, + "step": 16145 + }, + { + "epoch": 6.76, + "learning_rate": 1.2313849921027277e-06, + "loss": 0.0711, + "step": 16146 + }, + { + "epoch": 6.76, + "learning_rate": 1.2305700620440463e-06, + "loss": 0.1264, + "step": 16147 + }, + { + "epoch": 6.76, + "learning_rate": 1.2297553840548481e-06, + "loss": 0.0625, + "step": 16148 + }, + { + "epoch": 6.76, + "learning_rate": 1.2289409581585475e-06, + "loss": 0.1041, + "step": 16149 + }, + { + "epoch": 6.76, + "learning_rate": 1.228126784378555e-06, + "loss": 0.0655, + "step": 16150 + }, + { + "epoch": 6.76, + "learning_rate": 1.2273128627382768e-06, + "loss": 0.0905, + "step": 16151 + }, + { + "epoch": 6.76, + "learning_rate": 1.2264991932611036e-06, + "loss": 0.0756, + "step": 16152 + }, + { + "epoch": 6.76, + "learning_rate": 1.2256857759704255e-06, + "loss": 0.099, + "step": 16153 + }, + { + "epoch": 6.76, + "learning_rate": 1.2248726108896225e-06, + "loss": 0.075, + "step": 16154 + }, + { + "epoch": 6.76, + "learning_rate": 1.2240596980420704e-06, + "loss": 0.0853, + "step": 16155 + }, + { + "epoch": 6.76, + "learning_rate": 1.223247037451135e-06, + "loss": 0.0908, + "step": 16156 + }, + { + "epoch": 6.76, + "learning_rate": 1.2224346291401745e-06, + "loss": 0.0708, + "step": 16157 + }, + { + "epoch": 6.76, + "learning_rate": 1.2216224731325433e-06, + "loss": 0.0498, + "step": 16158 + }, + { + "epoch": 6.77, + "learning_rate": 1.2208105694515826e-06, + "loss": 0.0287, + "step": 16159 + }, + { + "epoch": 6.77, + "learning_rate": 1.2199989181206317e-06, + "loss": 0.1168, + "step": 16160 + }, + { + "epoch": 6.77, + "learning_rate": 1.2191875191630209e-06, + "loss": 0.098, + "step": 16161 + }, + { + "epoch": 6.77, + "learning_rate": 1.2183763726020726e-06, + "loss": 0.1242, + "step": 16162 + }, + { + "epoch": 6.77, + "learning_rate": 1.2175654784611024e-06, + "loss": 0.082, + "step": 16163 + }, + { + "epoch": 6.77, + "learning_rate": 1.216754836763422e-06, + "loss": 0.0883, + "step": 16164 + }, + { + "epoch": 6.77, + "learning_rate": 1.2159444475323268e-06, + "loss": 0.0557, + "step": 16165 + }, + { + "epoch": 6.77, + "learning_rate": 1.215134310791114e-06, + "loss": 0.0623, + "step": 16166 + }, + { + "epoch": 6.77, + "learning_rate": 1.2143244265630705e-06, + "loss": 0.0769, + "step": 16167 + }, + { + "epoch": 6.77, + "learning_rate": 1.2135147948714743e-06, + "loss": 0.0764, + "step": 16168 + }, + { + "epoch": 6.77, + "learning_rate": 1.212705415739599e-06, + "loss": 0.0581, + "step": 16169 + }, + { + "epoch": 6.77, + "learning_rate": 1.2118962891907105e-06, + "loss": 0.0961, + "step": 16170 + }, + { + "epoch": 6.77, + "learning_rate": 1.2110874152480622e-06, + "loss": 0.0848, + "step": 16171 + }, + { + "epoch": 6.77, + "learning_rate": 1.2102787939349091e-06, + "loss": 0.0636, + "step": 16172 + }, + { + "epoch": 6.77, + "learning_rate": 1.2094704252744916e-06, + "loss": 0.0711, + "step": 16173 + }, + { + "epoch": 6.77, + "learning_rate": 1.2086623092900441e-06, + "loss": 0.0806, + "step": 16174 + }, + { + "epoch": 6.77, + "learning_rate": 1.2078544460047991e-06, + "loss": 0.077, + "step": 16175 + }, + { + "epoch": 6.77, + "learning_rate": 1.207046835441975e-06, + "loss": 0.0656, + "step": 16176 + }, + { + "epoch": 6.77, + "learning_rate": 1.2062394776247865e-06, + "loss": 0.0556, + "step": 16177 + }, + { + "epoch": 6.77, + "learning_rate": 1.2054323725764439e-06, + "loss": 0.0943, + "step": 16178 + }, + { + "epoch": 6.77, + "learning_rate": 1.2046255203201407e-06, + "loss": 0.0666, + "step": 16179 + }, + { + "epoch": 6.77, + "learning_rate": 1.2038189208790718e-06, + "loss": 0.0834, + "step": 16180 + }, + { + "epoch": 6.77, + "learning_rate": 1.203012574276423e-06, + "loss": 0.0597, + "step": 16181 + }, + { + "epoch": 6.77, + "learning_rate": 1.2022064805353717e-06, + "loss": 0.0525, + "step": 16182 + }, + { + "epoch": 6.78, + "learning_rate": 1.2014006396790878e-06, + "loss": 0.0871, + "step": 16183 + }, + { + "epoch": 6.78, + "learning_rate": 1.2005950517307374e-06, + "loss": 0.077, + "step": 16184 + }, + { + "epoch": 6.78, + "learning_rate": 1.1997897167134708e-06, + "loss": 0.0911, + "step": 16185 + }, + { + "epoch": 6.78, + "learning_rate": 1.1989846346504396e-06, + "loss": 0.0771, + "step": 16186 + }, + { + "epoch": 6.78, + "learning_rate": 1.1981798055647865e-06, + "loss": 0.0922, + "step": 16187 + }, + { + "epoch": 6.78, + "learning_rate": 1.1973752294796426e-06, + "loss": 0.0934, + "step": 16188 + }, + { + "epoch": 6.78, + "learning_rate": 1.1965709064181374e-06, + "loss": 0.0966, + "step": 16189 + }, + { + "epoch": 6.78, + "learning_rate": 1.195766836403389e-06, + "loss": 0.0765, + "step": 16190 + }, + { + "epoch": 6.78, + "learning_rate": 1.1949630194585127e-06, + "loss": 0.0615, + "step": 16191 + }, + { + "epoch": 6.78, + "learning_rate": 1.1941594556066083e-06, + "loss": 0.0802, + "step": 16192 + }, + { + "epoch": 6.78, + "learning_rate": 1.1933561448707764e-06, + "loss": 0.0944, + "step": 16193 + }, + { + "epoch": 6.78, + "learning_rate": 1.1925530872741098e-06, + "loss": 0.0783, + "step": 16194 + }, + { + "epoch": 6.78, + "learning_rate": 1.1917502828396864e-06, + "loss": 0.0789, + "step": 16195 + }, + { + "epoch": 6.78, + "learning_rate": 1.190947731590586e-06, + "loss": 0.0661, + "step": 16196 + }, + { + "epoch": 6.78, + "learning_rate": 1.1901454335498752e-06, + "loss": 0.1021, + "step": 16197 + }, + { + "epoch": 6.78, + "learning_rate": 1.1893433887406158e-06, + "loss": 0.082, + "step": 16198 + }, + { + "epoch": 6.78, + "learning_rate": 1.188541597185865e-06, + "loss": 0.0746, + "step": 16199 + }, + { + "epoch": 6.78, + "learning_rate": 1.187740058908664e-06, + "loss": 0.1068, + "step": 16200 + }, + { + "epoch": 6.78, + "learning_rate": 1.186938773932056e-06, + "loss": 0.0991, + "step": 16201 + }, + { + "epoch": 6.78, + "learning_rate": 1.186137742279072e-06, + "loss": 0.09, + "step": 16202 + }, + { + "epoch": 6.78, + "learning_rate": 1.1853369639727376e-06, + "loss": 0.0617, + "step": 16203 + }, + { + "epoch": 6.78, + "learning_rate": 1.1845364390360692e-06, + "loss": 0.0599, + "step": 16204 + }, + { + "epoch": 6.78, + "learning_rate": 1.183736167492081e-06, + "loss": 0.0962, + "step": 16205 + }, + { + "epoch": 6.79, + "learning_rate": 1.1829361493637715e-06, + "loss": 0.0671, + "step": 16206 + }, + { + "epoch": 6.79, + "learning_rate": 1.1821363846741374e-06, + "loss": 0.0784, + "step": 16207 + }, + { + "epoch": 6.79, + "learning_rate": 1.1813368734461683e-06, + "loss": 0.0494, + "step": 16208 + }, + { + "epoch": 6.79, + "learning_rate": 1.1805376157028458e-06, + "loss": 0.0738, + "step": 16209 + }, + { + "epoch": 6.79, + "learning_rate": 1.1797386114671428e-06, + "loss": 0.0847, + "step": 16210 + }, + { + "epoch": 6.79, + "learning_rate": 1.1789398607620283e-06, + "loss": 0.0901, + "step": 16211 + }, + { + "epoch": 6.79, + "learning_rate": 1.1781413636104588e-06, + "loss": 0.0883, + "step": 16212 + }, + { + "epoch": 6.79, + "learning_rate": 1.1773431200353857e-06, + "loss": 0.0552, + "step": 16213 + }, + { + "epoch": 6.79, + "learning_rate": 1.1765451300597574e-06, + "loss": 0.0869, + "step": 16214 + }, + { + "epoch": 6.79, + "learning_rate": 1.175747393706509e-06, + "loss": 0.0581, + "step": 16215 + }, + { + "epoch": 6.79, + "learning_rate": 1.1749499109985728e-06, + "loss": 0.1278, + "step": 16216 + }, + { + "epoch": 6.79, + "learning_rate": 1.1741526819588679e-06, + "loss": 0.1016, + "step": 16217 + }, + { + "epoch": 6.79, + "learning_rate": 1.173355706610313e-06, + "loss": 0.0581, + "step": 16218 + }, + { + "epoch": 6.79, + "learning_rate": 1.1725589849758178e-06, + "loss": 0.0699, + "step": 16219 + }, + { + "epoch": 6.79, + "learning_rate": 1.1717625170782788e-06, + "loss": 0.0806, + "step": 16220 + }, + { + "epoch": 6.79, + "learning_rate": 1.1709663029405916e-06, + "loss": 0.0566, + "step": 16221 + }, + { + "epoch": 6.79, + "learning_rate": 1.1701703425856437e-06, + "loss": 0.0792, + "step": 16222 + }, + { + "epoch": 6.79, + "learning_rate": 1.1693746360363145e-06, + "loss": 0.0864, + "step": 16223 + }, + { + "epoch": 6.79, + "learning_rate": 1.1685791833154748e-06, + "loss": 0.0726, + "step": 16224 + }, + { + "epoch": 6.79, + "learning_rate": 1.1677839844459914e-06, + "loss": 0.0665, + "step": 16225 + }, + { + "epoch": 6.79, + "learning_rate": 1.1669890394507189e-06, + "loss": 0.08, + "step": 16226 + }, + { + "epoch": 6.79, + "learning_rate": 1.1661943483525074e-06, + "loss": 0.0594, + "step": 16227 + }, + { + "epoch": 6.79, + "learning_rate": 1.1653999111741998e-06, + "loss": 0.0887, + "step": 16228 + }, + { + "epoch": 6.79, + "learning_rate": 1.1646057279386325e-06, + "loss": 0.0723, + "step": 16229 + }, + { + "epoch": 6.8, + "learning_rate": 1.1638117986686336e-06, + "loss": 0.0681, + "step": 16230 + }, + { + "epoch": 6.8, + "learning_rate": 1.163018123387023e-06, + "loss": 0.0856, + "step": 16231 + }, + { + "epoch": 6.8, + "learning_rate": 1.1622247021166166e-06, + "loss": 0.0538, + "step": 16232 + }, + { + "epoch": 6.8, + "learning_rate": 1.1614315348802164e-06, + "loss": 0.0484, + "step": 16233 + }, + { + "epoch": 6.8, + "learning_rate": 1.160638621700624e-06, + "loss": 0.1014, + "step": 16234 + }, + { + "epoch": 6.8, + "learning_rate": 1.1598459626006308e-06, + "loss": 0.0939, + "step": 16235 + }, + { + "epoch": 6.8, + "learning_rate": 1.1590535576030204e-06, + "loss": 0.0629, + "step": 16236 + }, + { + "epoch": 6.8, + "learning_rate": 1.1582614067305697e-06, + "loss": 0.09, + "step": 16237 + }, + { + "epoch": 6.8, + "learning_rate": 1.1574695100060517e-06, + "loss": 0.0807, + "step": 16238 + }, + { + "epoch": 6.8, + "learning_rate": 1.1566778674522228e-06, + "loss": 0.0849, + "step": 16239 + }, + { + "epoch": 6.8, + "learning_rate": 1.1558864790918444e-06, + "loss": 0.0493, + "step": 16240 + }, + { + "epoch": 6.8, + "learning_rate": 1.1550953449476588e-06, + "loss": 0.1025, + "step": 16241 + }, + { + "epoch": 6.8, + "learning_rate": 1.1543044650424085e-06, + "loss": 0.0826, + "step": 16242 + }, + { + "epoch": 6.8, + "learning_rate": 1.1535138393988265e-06, + "loss": 0.0749, + "step": 16243 + }, + { + "epoch": 6.8, + "learning_rate": 1.1527234680396392e-06, + "loss": 0.0706, + "step": 16244 + }, + { + "epoch": 6.8, + "learning_rate": 1.1519333509875642e-06, + "loss": 0.079, + "step": 16245 + }, + { + "epoch": 6.8, + "learning_rate": 1.151143488265316e-06, + "loss": 0.0734, + "step": 16246 + }, + { + "epoch": 6.8, + "learning_rate": 1.1503538798955937e-06, + "loss": 0.1182, + "step": 16247 + }, + { + "epoch": 6.8, + "learning_rate": 1.1495645259010969e-06, + "loss": 0.083, + "step": 16248 + }, + { + "epoch": 6.8, + "learning_rate": 1.1487754263045138e-06, + "loss": 0.0921, + "step": 16249 + }, + { + "epoch": 6.8, + "learning_rate": 1.1479865811285262e-06, + "loss": 0.0853, + "step": 16250 + }, + { + "epoch": 6.8, + "learning_rate": 1.1471979903958087e-06, + "loss": 0.0741, + "step": 16251 + }, + { + "epoch": 6.8, + "learning_rate": 1.1464096541290326e-06, + "loss": 0.0754, + "step": 16252 + }, + { + "epoch": 6.8, + "learning_rate": 1.1456215723508512e-06, + "loss": 0.0952, + "step": 16253 + }, + { + "epoch": 6.81, + "learning_rate": 1.1448337450839208e-06, + "loss": 0.0692, + "step": 16254 + }, + { + "epoch": 6.81, + "learning_rate": 1.1440461723508867e-06, + "loss": 0.0656, + "step": 16255 + }, + { + "epoch": 6.81, + "learning_rate": 1.143258854174386e-06, + "loss": 0.0882, + "step": 16256 + }, + { + "epoch": 6.81, + "learning_rate": 1.142471790577051e-06, + "loss": 0.0911, + "step": 16257 + }, + { + "epoch": 6.81, + "learning_rate": 1.1416849815815045e-06, + "loss": 0.0977, + "step": 16258 + }, + { + "epoch": 6.81, + "learning_rate": 1.140898427210364e-06, + "loss": 0.0682, + "step": 16259 + }, + { + "epoch": 6.81, + "learning_rate": 1.1401121274862348e-06, + "loss": 0.0785, + "step": 16260 + }, + { + "epoch": 6.81, + "learning_rate": 1.1393260824317208e-06, + "loss": 0.0927, + "step": 16261 + }, + { + "epoch": 6.81, + "learning_rate": 1.1385402920694167e-06, + "loss": 0.0633, + "step": 16262 + }, + { + "epoch": 6.81, + "learning_rate": 1.137754756421907e-06, + "loss": 0.0938, + "step": 16263 + }, + { + "epoch": 6.81, + "learning_rate": 1.1369694755117733e-06, + "loss": 0.051, + "step": 16264 + }, + { + "epoch": 6.81, + "learning_rate": 1.136184449361586e-06, + "loss": 0.0968, + "step": 16265 + }, + { + "epoch": 6.81, + "learning_rate": 1.1353996779939135e-06, + "loss": 0.1161, + "step": 16266 + }, + { + "epoch": 6.81, + "learning_rate": 1.1346151614313095e-06, + "loss": 0.0944, + "step": 16267 + }, + { + "epoch": 6.81, + "learning_rate": 1.1338308996963255e-06, + "loss": 0.0622, + "step": 16268 + }, + { + "epoch": 6.81, + "learning_rate": 1.1330468928115036e-06, + "loss": 0.0881, + "step": 16269 + }, + { + "epoch": 6.81, + "learning_rate": 1.132263140799381e-06, + "loss": 0.1001, + "step": 16270 + }, + { + "epoch": 6.81, + "learning_rate": 1.131479643682486e-06, + "loss": 0.082, + "step": 16271 + }, + { + "epoch": 6.81, + "learning_rate": 1.1306964014833378e-06, + "loss": 0.0911, + "step": 16272 + }, + { + "epoch": 6.81, + "learning_rate": 1.129913414224454e-06, + "loss": 0.0798, + "step": 16273 + }, + { + "epoch": 6.81, + "learning_rate": 1.1291306819283353e-06, + "loss": 0.0896, + "step": 16274 + }, + { + "epoch": 6.81, + "learning_rate": 1.1283482046174832e-06, + "loss": 0.0488, + "step": 16275 + }, + { + "epoch": 6.81, + "learning_rate": 1.1275659823143903e-06, + "loss": 0.0652, + "step": 16276 + }, + { + "epoch": 6.81, + "learning_rate": 1.1267840150415388e-06, + "loss": 0.093, + "step": 16277 + }, + { + "epoch": 6.82, + "learning_rate": 1.1260023028214072e-06, + "loss": 0.0839, + "step": 16278 + }, + { + "epoch": 6.82, + "learning_rate": 1.125220845676467e-06, + "loss": 0.0774, + "step": 16279 + }, + { + "epoch": 6.82, + "learning_rate": 1.124439643629176e-06, + "loss": 0.0969, + "step": 16280 + }, + { + "epoch": 6.82, + "learning_rate": 1.1236586967019913e-06, + "loss": 0.0869, + "step": 16281 + }, + { + "epoch": 6.82, + "learning_rate": 1.1228780049173616e-06, + "loss": 0.0632, + "step": 16282 + }, + { + "epoch": 6.82, + "learning_rate": 1.1220975682977254e-06, + "loss": 0.0939, + "step": 16283 + }, + { + "epoch": 6.82, + "learning_rate": 1.121317386865518e-06, + "loss": 0.077, + "step": 16284 + }, + { + "epoch": 6.82, + "learning_rate": 1.120537460643163e-06, + "loss": 0.0857, + "step": 16285 + }, + { + "epoch": 6.82, + "learning_rate": 1.1197577896530776e-06, + "loss": 0.0703, + "step": 16286 + }, + { + "epoch": 6.82, + "learning_rate": 1.1189783739176774e-06, + "loss": 0.0719, + "step": 16287 + }, + { + "epoch": 6.82, + "learning_rate": 1.1181992134593612e-06, + "loss": 0.0721, + "step": 16288 + }, + { + "epoch": 6.82, + "learning_rate": 1.1174203083005263e-06, + "loss": 0.0799, + "step": 16289 + }, + { + "epoch": 6.82, + "learning_rate": 1.116641658463562e-06, + "loss": 0.0699, + "step": 16290 + }, + { + "epoch": 6.82, + "learning_rate": 1.1158632639708512e-06, + "loss": 0.0949, + "step": 16291 + }, + { + "epoch": 6.82, + "learning_rate": 1.1150851248447669e-06, + "loss": 0.0699, + "step": 16292 + }, + { + "epoch": 6.82, + "learning_rate": 1.1143072411076782e-06, + "loss": 0.0731, + "step": 16293 + }, + { + "epoch": 6.82, + "learning_rate": 1.1135296127819407e-06, + "loss": 0.0813, + "step": 16294 + }, + { + "epoch": 6.82, + "learning_rate": 1.1127522398899094e-06, + "loss": 0.081, + "step": 16295 + }, + { + "epoch": 6.82, + "learning_rate": 1.1119751224539278e-06, + "loss": 0.0699, + "step": 16296 + }, + { + "epoch": 6.82, + "learning_rate": 1.1111982604963345e-06, + "loss": 0.0821, + "step": 16297 + }, + { + "epoch": 6.82, + "learning_rate": 1.1104216540394596e-06, + "loss": 0.0959, + "step": 16298 + }, + { + "epoch": 6.82, + "learning_rate": 1.1096453031056265e-06, + "loss": 0.0907, + "step": 16299 + }, + { + "epoch": 6.82, + "learning_rate": 1.1088692077171503e-06, + "loss": 0.0513, + "step": 16300 + }, + { + "epoch": 6.82, + "learning_rate": 1.1080933678963379e-06, + "loss": 0.0845, + "step": 16301 + }, + { + "epoch": 6.83, + "learning_rate": 1.1073177836654913e-06, + "loss": 0.0945, + "step": 16302 + }, + { + "epoch": 6.83, + "learning_rate": 1.1065424550469028e-06, + "loss": 0.0662, + "step": 16303 + }, + { + "epoch": 6.83, + "learning_rate": 1.1057673820628612e-06, + "loss": 0.0858, + "step": 16304 + }, + { + "epoch": 6.83, + "learning_rate": 1.104992564735642e-06, + "loss": 0.0808, + "step": 16305 + }, + { + "epoch": 6.83, + "learning_rate": 1.1042180030875205e-06, + "loss": 0.0498, + "step": 16306 + }, + { + "epoch": 6.83, + "learning_rate": 1.1034436971407592e-06, + "loss": 0.0809, + "step": 16307 + }, + { + "epoch": 6.83, + "learning_rate": 1.1026696469176124e-06, + "loss": 0.094, + "step": 16308 + }, + { + "epoch": 6.83, + "learning_rate": 1.10189585244033e-06, + "loss": 0.1118, + "step": 16309 + }, + { + "epoch": 6.83, + "learning_rate": 1.1011223137311555e-06, + "loss": 0.0975, + "step": 16310 + }, + { + "epoch": 6.83, + "learning_rate": 1.1003490308123243e-06, + "loss": 0.0844, + "step": 16311 + }, + { + "epoch": 6.83, + "learning_rate": 1.099576003706062e-06, + "loss": 0.0935, + "step": 16312 + }, + { + "epoch": 6.83, + "learning_rate": 1.0988032324345898e-06, + "loss": 0.0727, + "step": 16313 + }, + { + "epoch": 6.83, + "learning_rate": 1.0980307170201222e-06, + "loss": 0.0765, + "step": 16314 + }, + { + "epoch": 6.83, + "learning_rate": 1.09725845748486e-06, + "loss": 0.074, + "step": 16315 + }, + { + "epoch": 6.83, + "learning_rate": 1.0964864538510022e-06, + "loss": 0.0684, + "step": 16316 + }, + { + "epoch": 6.83, + "learning_rate": 1.0957147061407414e-06, + "loss": 0.0592, + "step": 16317 + }, + { + "epoch": 6.83, + "learning_rate": 1.0949432143762596e-06, + "loss": 0.1047, + "step": 16318 + }, + { + "epoch": 6.83, + "learning_rate": 1.0941719785797333e-06, + "loss": 0.0862, + "step": 16319 + }, + { + "epoch": 6.83, + "learning_rate": 1.0934009987733318e-06, + "loss": 0.074, + "step": 16320 + }, + { + "epoch": 6.83, + "learning_rate": 1.092630274979214e-06, + "loss": 0.0656, + "step": 16321 + }, + { + "epoch": 6.83, + "learning_rate": 1.091859807219534e-06, + "loss": 0.1011, + "step": 16322 + }, + { + "epoch": 6.83, + "learning_rate": 1.091089595516439e-06, + "loss": 0.0674, + "step": 16323 + }, + { + "epoch": 6.83, + "learning_rate": 1.090319639892069e-06, + "loss": 0.0754, + "step": 16324 + }, + { + "epoch": 6.83, + "learning_rate": 1.0895499403685527e-06, + "loss": 0.0759, + "step": 16325 + }, + { + "epoch": 6.84, + "learning_rate": 1.0887804969680183e-06, + "loss": 0.0659, + "step": 16326 + }, + { + "epoch": 6.84, + "learning_rate": 1.0880113097125811e-06, + "loss": 0.0756, + "step": 16327 + }, + { + "epoch": 6.84, + "learning_rate": 1.0872423786243492e-06, + "loss": 0.0923, + "step": 16328 + }, + { + "epoch": 6.84, + "learning_rate": 1.0864737037254269e-06, + "loss": 0.0663, + "step": 16329 + }, + { + "epoch": 6.84, + "learning_rate": 1.085705285037909e-06, + "loss": 0.0682, + "step": 16330 + }, + { + "epoch": 6.84, + "learning_rate": 1.0849371225838812e-06, + "loss": 0.0596, + "step": 16331 + }, + { + "epoch": 6.84, + "learning_rate": 1.0841692163854244e-06, + "loss": 0.0642, + "step": 16332 + }, + { + "epoch": 6.84, + "learning_rate": 1.0834015664646114e-06, + "loss": 0.0716, + "step": 16333 + }, + { + "epoch": 6.84, + "learning_rate": 1.0826341728435096e-06, + "loss": 0.0658, + "step": 16334 + }, + { + "epoch": 6.84, + "learning_rate": 1.0818670355441741e-06, + "loss": 0.0611, + "step": 16335 + }, + { + "epoch": 6.84, + "learning_rate": 1.0811001545886568e-06, + "loss": 0.0533, + "step": 16336 + }, + { + "epoch": 6.84, + "learning_rate": 1.0803335299990015e-06, + "loss": 0.0934, + "step": 16337 + }, + { + "epoch": 6.84, + "learning_rate": 1.0795671617972424e-06, + "loss": 0.0777, + "step": 16338 + }, + { + "epoch": 6.84, + "learning_rate": 1.0788010500054102e-06, + "loss": 0.0946, + "step": 16339 + }, + { + "epoch": 6.84, + "learning_rate": 1.0780351946455258e-06, + "loss": 0.089, + "step": 16340 + }, + { + "epoch": 6.84, + "learning_rate": 1.077269595739604e-06, + "loss": 0.0611, + "step": 16341 + }, + { + "epoch": 6.84, + "learning_rate": 1.076504253309647e-06, + "loss": 0.096, + "step": 16342 + }, + { + "epoch": 6.84, + "learning_rate": 1.0757391673776573e-06, + "loss": 0.0851, + "step": 16343 + }, + { + "epoch": 6.84, + "learning_rate": 1.0749743379656263e-06, + "loss": 0.0889, + "step": 16344 + }, + { + "epoch": 6.84, + "learning_rate": 1.0742097650955375e-06, + "loss": 0.0625, + "step": 16345 + }, + { + "epoch": 6.84, + "learning_rate": 1.0734454487893687e-06, + "loss": 0.0934, + "step": 16346 + }, + { + "epoch": 6.84, + "learning_rate": 1.0726813890690913e-06, + "loss": 0.0325, + "step": 16347 + }, + { + "epoch": 6.84, + "learning_rate": 1.0719175859566634e-06, + "loss": 0.0627, + "step": 16348 + }, + { + "epoch": 6.84, + "learning_rate": 1.0711540394740416e-06, + "loss": 0.0725, + "step": 16349 + }, + { + "epoch": 6.85, + "learning_rate": 1.0703907496431743e-06, + "loss": 0.0745, + "step": 16350 + }, + { + "epoch": 6.85, + "learning_rate": 1.0696277164860002e-06, + "loss": 0.0632, + "step": 16351 + }, + { + "epoch": 6.85, + "learning_rate": 1.0688649400244544e-06, + "loss": 0.0666, + "step": 16352 + }, + { + "epoch": 6.85, + "learning_rate": 1.068102420280459e-06, + "loss": 0.1089, + "step": 16353 + }, + { + "epoch": 6.85, + "learning_rate": 1.0673401572759335e-06, + "loss": 0.0928, + "step": 16354 + }, + { + "epoch": 6.85, + "learning_rate": 1.0665781510327911e-06, + "loss": 0.0625, + "step": 16355 + }, + { + "epoch": 6.85, + "learning_rate": 1.06581640157293e-06, + "loss": 0.0707, + "step": 16356 + }, + { + "epoch": 6.85, + "learning_rate": 1.0650549089182482e-06, + "loss": 0.0622, + "step": 16357 + }, + { + "epoch": 6.85, + "learning_rate": 1.064293673090634e-06, + "loss": 0.0956, + "step": 16358 + }, + { + "epoch": 6.85, + "learning_rate": 1.0635326941119683e-06, + "loss": 0.0679, + "step": 16359 + }, + { + "epoch": 6.85, + "learning_rate": 1.0627719720041263e-06, + "loss": 0.073, + "step": 16360 + }, + { + "epoch": 6.85, + "learning_rate": 1.0620115067889747e-06, + "loss": 0.0908, + "step": 16361 + }, + { + "epoch": 6.85, + "learning_rate": 1.0612512984883694e-06, + "loss": 0.0869, + "step": 16362 + }, + { + "epoch": 6.85, + "learning_rate": 1.0604913471241629e-06, + "loss": 0.1014, + "step": 16363 + }, + { + "epoch": 6.85, + "learning_rate": 1.0597316527182e-06, + "loss": 0.0775, + "step": 16364 + }, + { + "epoch": 6.85, + "learning_rate": 1.0589722152923177e-06, + "loss": 0.0745, + "step": 16365 + }, + { + "epoch": 6.85, + "learning_rate": 1.058213034868345e-06, + "loss": 0.0947, + "step": 16366 + }, + { + "epoch": 6.85, + "learning_rate": 1.0574541114681046e-06, + "loss": 0.0809, + "step": 16367 + }, + { + "epoch": 6.85, + "learning_rate": 1.0566954451134093e-06, + "loss": 0.0749, + "step": 16368 + }, + { + "epoch": 6.85, + "learning_rate": 1.0559370358260667e-06, + "loss": 0.0803, + "step": 16369 + }, + { + "epoch": 6.85, + "learning_rate": 1.0551788836278776e-06, + "loss": 0.0862, + "step": 16370 + }, + { + "epoch": 6.85, + "learning_rate": 1.0544209885406342e-06, + "loss": 0.1013, + "step": 16371 + }, + { + "epoch": 6.85, + "learning_rate": 1.0536633505861216e-06, + "loss": 0.093, + "step": 16372 + }, + { + "epoch": 6.85, + "learning_rate": 1.0529059697861165e-06, + "loss": 0.0648, + "step": 16373 + }, + { + "epoch": 6.86, + "learning_rate": 1.0521488461623919e-06, + "loss": 0.0916, + "step": 16374 + }, + { + "epoch": 6.86, + "learning_rate": 1.0513919797367078e-06, + "loss": 0.0723, + "step": 16375 + }, + { + "epoch": 6.86, + "learning_rate": 1.0506353705308192e-06, + "loss": 0.0744, + "step": 16376 + }, + { + "epoch": 6.86, + "learning_rate": 1.0498790185664753e-06, + "loss": 0.0771, + "step": 16377 + }, + { + "epoch": 6.86, + "learning_rate": 1.0491229238654177e-06, + "loss": 0.0938, + "step": 16378 + }, + { + "epoch": 6.86, + "learning_rate": 1.0483670864493777e-06, + "loss": 0.0958, + "step": 16379 + }, + { + "epoch": 6.86, + "learning_rate": 1.0476115063400826e-06, + "loss": 0.0492, + "step": 16380 + }, + { + "epoch": 6.86, + "learning_rate": 1.0468561835592495e-06, + "loss": 0.0862, + "step": 16381 + }, + { + "epoch": 6.86, + "learning_rate": 1.046101118128594e-06, + "loss": 0.0814, + "step": 16382 + }, + { + "epoch": 6.86, + "learning_rate": 1.0453463100698135e-06, + "loss": 0.1073, + "step": 16383 + }, + { + "epoch": 6.86, + "learning_rate": 1.0445917594046073e-06, + "loss": 0.0732, + "step": 16384 + }, + { + "epoch": 6.86, + "learning_rate": 1.0438374661546646e-06, + "loss": 0.0696, + "step": 16385 + }, + { + "epoch": 6.86, + "learning_rate": 1.0430834303416658e-06, + "loss": 0.0517, + "step": 16386 + }, + { + "epoch": 6.86, + "learning_rate": 1.042329651987285e-06, + "loss": 0.0821, + "step": 16387 + }, + { + "epoch": 6.86, + "learning_rate": 1.0415761311131922e-06, + "loss": 0.0809, + "step": 16388 + }, + { + "epoch": 6.86, + "learning_rate": 1.0408228677410425e-06, + "loss": 0.086, + "step": 16389 + }, + { + "epoch": 6.86, + "learning_rate": 1.0400698618924888e-06, + "loss": 0.0722, + "step": 16390 + }, + { + "epoch": 6.86, + "learning_rate": 1.0393171135891754e-06, + "loss": 0.088, + "step": 16391 + }, + { + "epoch": 6.86, + "learning_rate": 1.0385646228527401e-06, + "loss": 0.0753, + "step": 16392 + }, + { + "epoch": 6.86, + "learning_rate": 1.0378123897048131e-06, + "loss": 0.0865, + "step": 16393 + }, + { + "epoch": 6.86, + "learning_rate": 1.037060414167017e-06, + "loss": 0.0897, + "step": 16394 + }, + { + "epoch": 6.86, + "learning_rate": 1.0363086962609648e-06, + "loss": 0.098, + "step": 16395 + }, + { + "epoch": 6.86, + "learning_rate": 1.0355572360082645e-06, + "loss": 0.0993, + "step": 16396 + }, + { + "epoch": 6.86, + "learning_rate": 1.0348060334305187e-06, + "loss": 0.0444, + "step": 16397 + }, + { + "epoch": 6.87, + "learning_rate": 1.0340550885493162e-06, + "loss": 0.099, + "step": 16398 + }, + { + "epoch": 6.87, + "learning_rate": 1.0333044013862437e-06, + "loss": 0.0992, + "step": 16399 + }, + { + "epoch": 6.87, + "learning_rate": 1.0325539719628785e-06, + "loss": 0.0673, + "step": 16400 + }, + { + "epoch": 6.87, + "learning_rate": 1.0318038003007924e-06, + "loss": 0.0799, + "step": 16401 + }, + { + "epoch": 6.87, + "learning_rate": 1.0310538864215503e-06, + "loss": 0.0739, + "step": 16402 + }, + { + "epoch": 6.87, + "learning_rate": 1.0303042303467027e-06, + "loss": 0.0847, + "step": 16403 + }, + { + "epoch": 6.87, + "learning_rate": 1.029554832097801e-06, + "loss": 0.0459, + "step": 16404 + }, + { + "epoch": 6.87, + "learning_rate": 1.0288056916963852e-06, + "loss": 0.0961, + "step": 16405 + }, + { + "epoch": 6.87, + "learning_rate": 1.0280568091639886e-06, + "loss": 0.0856, + "step": 16406 + }, + { + "epoch": 6.87, + "learning_rate": 1.0273081845221377e-06, + "loss": 0.0735, + "step": 16407 + }, + { + "epoch": 6.87, + "learning_rate": 1.026559817792352e-06, + "loss": 0.0828, + "step": 16408 + }, + { + "epoch": 6.87, + "learning_rate": 1.0258117089961405e-06, + "loss": 0.0677, + "step": 16409 + }, + { + "epoch": 6.87, + "learning_rate": 1.0250638581550076e-06, + "loss": 0.0815, + "step": 16410 + }, + { + "epoch": 6.87, + "learning_rate": 1.0243162652904492e-06, + "loss": 0.0893, + "step": 16411 + }, + { + "epoch": 6.87, + "learning_rate": 1.023568930423956e-06, + "loss": 0.0918, + "step": 16412 + }, + { + "epoch": 6.87, + "learning_rate": 1.0228218535770075e-06, + "loss": 0.0651, + "step": 16413 + }, + { + "epoch": 6.87, + "learning_rate": 1.0220750347710795e-06, + "loss": 0.0595, + "step": 16414 + }, + { + "epoch": 6.87, + "learning_rate": 1.0213284740276385e-06, + "loss": 0.089, + "step": 16415 + }, + { + "epoch": 6.87, + "learning_rate": 1.020582171368142e-06, + "loss": 0.0554, + "step": 16416 + }, + { + "epoch": 6.87, + "learning_rate": 1.0198361268140422e-06, + "loss": 0.0759, + "step": 16417 + }, + { + "epoch": 6.87, + "learning_rate": 1.0190903403867847e-06, + "loss": 0.0866, + "step": 16418 + }, + { + "epoch": 6.87, + "learning_rate": 1.0183448121078055e-06, + "loss": 0.0963, + "step": 16419 + }, + { + "epoch": 6.87, + "learning_rate": 1.0175995419985364e-06, + "loss": 0.0663, + "step": 16420 + }, + { + "epoch": 6.88, + "learning_rate": 1.0168545300803957e-06, + "loss": 0.0623, + "step": 16421 + }, + { + "epoch": 6.88, + "learning_rate": 1.0161097763748006e-06, + "loss": 0.0856, + "step": 16422 + }, + { + "epoch": 6.88, + "learning_rate": 1.015365280903159e-06, + "loss": 0.0765, + "step": 16423 + }, + { + "epoch": 6.88, + "learning_rate": 1.0146210436868675e-06, + "loss": 0.0778, + "step": 16424 + }, + { + "epoch": 6.88, + "learning_rate": 1.0138770647473206e-06, + "loss": 0.0706, + "step": 16425 + }, + { + "epoch": 6.88, + "learning_rate": 1.0131333441059045e-06, + "loss": 0.0771, + "step": 16426 + }, + { + "epoch": 6.88, + "learning_rate": 1.012389881783995e-06, + "loss": 0.0703, + "step": 16427 + }, + { + "epoch": 6.88, + "learning_rate": 1.011646677802962e-06, + "loss": 0.0617, + "step": 16428 + }, + { + "epoch": 6.88, + "learning_rate": 1.010903732184172e-06, + "loss": 0.097, + "step": 16429 + }, + { + "epoch": 6.88, + "learning_rate": 1.010161044948975e-06, + "loss": 0.0809, + "step": 16430 + }, + { + "epoch": 6.88, + "learning_rate": 1.0094186161187214e-06, + "loss": 0.0792, + "step": 16431 + }, + { + "epoch": 6.88, + "learning_rate": 1.0086764457147524e-06, + "loss": 0.0852, + "step": 16432 + }, + { + "epoch": 6.88, + "learning_rate": 1.007934533758399e-06, + "loss": 0.0726, + "step": 16433 + }, + { + "epoch": 6.88, + "learning_rate": 1.0071928802709886e-06, + "loss": 0.1071, + "step": 16434 + }, + { + "epoch": 6.88, + "learning_rate": 1.0064514852738406e-06, + "loss": 0.0572, + "step": 16435 + }, + { + "epoch": 6.88, + "learning_rate": 1.0057103487882624e-06, + "loss": 0.0718, + "step": 16436 + }, + { + "epoch": 6.88, + "learning_rate": 1.0049694708355583e-06, + "loss": 0.0643, + "step": 16437 + }, + { + "epoch": 6.88, + "learning_rate": 1.0042288514370258e-06, + "loss": 0.069, + "step": 16438 + }, + { + "epoch": 6.88, + "learning_rate": 1.0034884906139508e-06, + "loss": 0.079, + "step": 16439 + }, + { + "epoch": 6.88, + "learning_rate": 1.0027483883876166e-06, + "loss": 0.0699, + "step": 16440 + }, + { + "epoch": 6.88, + "learning_rate": 1.002008544779296e-06, + "loss": 0.1084, + "step": 16441 + }, + { + "epoch": 6.88, + "learning_rate": 1.0012689598102565e-06, + "loss": 0.096, + "step": 16442 + }, + { + "epoch": 6.88, + "learning_rate": 1.0005296335017557e-06, + "loss": 0.0603, + "step": 16443 + }, + { + "epoch": 6.88, + "learning_rate": 9.997905658750428e-07, + "loss": 0.0888, + "step": 16444 + }, + { + "epoch": 6.89, + "learning_rate": 9.990517569513624e-07, + "loss": 0.0721, + "step": 16445 + }, + { + "epoch": 6.89, + "learning_rate": 9.98313206751953e-07, + "loss": 0.0858, + "step": 16446 + }, + { + "epoch": 6.89, + "learning_rate": 9.975749152980429e-07, + "loss": 0.0755, + "step": 16447 + }, + { + "epoch": 6.89, + "learning_rate": 9.96836882610852e-07, + "loss": 0.0749, + "step": 16448 + }, + { + "epoch": 6.89, + "learning_rate": 9.96099108711599e-07, + "loss": 0.0614, + "step": 16449 + }, + { + "epoch": 6.89, + "learning_rate": 9.953615936214845e-07, + "loss": 0.1032, + "step": 16450 + }, + { + "epoch": 6.89, + "learning_rate": 9.946243373617104e-07, + "loss": 0.0466, + "step": 16451 + }, + { + "epoch": 6.89, + "learning_rate": 9.938873399534688e-07, + "loss": 0.0939, + "step": 16452 + }, + { + "epoch": 6.89, + "learning_rate": 9.931506014179427e-07, + "loss": 0.1175, + "step": 16453 + }, + { + "epoch": 6.89, + "learning_rate": 9.924141217763117e-07, + "loss": 0.0821, + "step": 16454 + }, + { + "epoch": 6.89, + "learning_rate": 9.916779010497424e-07, + "loss": 0.0936, + "step": 16455 + }, + { + "epoch": 6.89, + "learning_rate": 9.909419392594e-07, + "loss": 0.0785, + "step": 16456 + }, + { + "epoch": 6.89, + "learning_rate": 9.902062364264354e-07, + "loss": 0.0702, + "step": 16457 + }, + { + "epoch": 6.89, + "learning_rate": 9.89470792571997e-07, + "loss": 0.0996, + "step": 16458 + }, + { + "epoch": 6.89, + "learning_rate": 9.88735607717225e-07, + "loss": 0.0818, + "step": 16459 + }, + { + "epoch": 6.89, + "learning_rate": 9.880006818832522e-07, + "loss": 0.0734, + "step": 16460 + }, + { + "epoch": 6.89, + "learning_rate": 9.872660150912016e-07, + "loss": 0.1005, + "step": 16461 + }, + { + "epoch": 6.89, + "learning_rate": 9.865316073621945e-07, + "loss": 0.0835, + "step": 16462 + }, + { + "epoch": 6.89, + "learning_rate": 9.857974587173357e-07, + "loss": 0.1036, + "step": 16463 + }, + { + "epoch": 6.89, + "learning_rate": 9.8506356917773e-07, + "loss": 0.0803, + "step": 16464 + }, + { + "epoch": 6.89, + "learning_rate": 9.843299387644746e-07, + "loss": 0.068, + "step": 16465 + }, + { + "epoch": 6.89, + "learning_rate": 9.835965674986514e-07, + "loss": 0.0679, + "step": 16466 + }, + { + "epoch": 6.89, + "learning_rate": 9.82863455401345e-07, + "loss": 0.0814, + "step": 16467 + }, + { + "epoch": 6.89, + "learning_rate": 9.821306024936272e-07, + "loss": 0.0378, + "step": 16468 + }, + { + "epoch": 6.9, + "learning_rate": 9.813980087965625e-07, + "loss": 0.0549, + "step": 16469 + }, + { + "epoch": 6.9, + "learning_rate": 9.806656743312104e-07, + "loss": 0.0847, + "step": 16470 + }, + { + "epoch": 6.9, + "learning_rate": 9.799335991186186e-07, + "loss": 0.0792, + "step": 16471 + }, + { + "epoch": 6.9, + "learning_rate": 9.792017831798307e-07, + "loss": 0.0707, + "step": 16472 + }, + { + "epoch": 6.9, + "learning_rate": 9.784702265358826e-07, + "loss": 0.0938, + "step": 16473 + }, + { + "epoch": 6.9, + "learning_rate": 9.777389292078033e-07, + "loss": 0.0953, + "step": 16474 + }, + { + "epoch": 6.9, + "learning_rate": 9.770078912166114e-07, + "loss": 0.0846, + "step": 16475 + }, + { + "epoch": 6.9, + "learning_rate": 9.762771125833226e-07, + "loss": 0.0786, + "step": 16476 + }, + { + "epoch": 6.9, + "learning_rate": 9.7554659332894e-07, + "loss": 0.0914, + "step": 16477 + }, + { + "epoch": 6.9, + "learning_rate": 9.748163334744609e-07, + "loss": 0.0496, + "step": 16478 + }, + { + "epoch": 6.9, + "learning_rate": 9.740863330408778e-07, + "loss": 0.0769, + "step": 16479 + }, + { + "epoch": 6.9, + "learning_rate": 9.733565920491738e-07, + "loss": 0.1122, + "step": 16480 + }, + { + "epoch": 6.9, + "learning_rate": 9.726271105203255e-07, + "loss": 0.0569, + "step": 16481 + }, + { + "epoch": 6.9, + "learning_rate": 9.718978884752983e-07, + "loss": 0.0868, + "step": 16482 + }, + { + "epoch": 6.9, + "learning_rate": 9.711689259350587e-07, + "loss": 0.0809, + "step": 16483 + }, + { + "epoch": 6.9, + "learning_rate": 9.704402229205534e-07, + "loss": 0.0978, + "step": 16484 + }, + { + "epoch": 6.9, + "learning_rate": 9.697117794527311e-07, + "loss": 0.0694, + "step": 16485 + }, + { + "epoch": 6.9, + "learning_rate": 9.689835955525307e-07, + "loss": 0.0765, + "step": 16486 + }, + { + "epoch": 6.9, + "learning_rate": 9.68255671240882e-07, + "loss": 0.0612, + "step": 16487 + }, + { + "epoch": 6.9, + "learning_rate": 9.675280065387117e-07, + "loss": 0.071, + "step": 16488 + }, + { + "epoch": 6.9, + "learning_rate": 9.668006014669328e-07, + "loss": 0.0822, + "step": 16489 + }, + { + "epoch": 6.9, + "learning_rate": 9.660734560464557e-07, + "loss": 0.1086, + "step": 16490 + }, + { + "epoch": 6.9, + "learning_rate": 9.653465702981779e-07, + "loss": 0.0845, + "step": 16491 + }, + { + "epoch": 6.9, + "learning_rate": 9.64619944242997e-07, + "loss": 0.0677, + "step": 16492 + }, + { + "epoch": 6.91, + "learning_rate": 9.638935779017977e-07, + "loss": 0.0838, + "step": 16493 + }, + { + "epoch": 6.91, + "learning_rate": 9.631674712954586e-07, + "loss": 0.0735, + "step": 16494 + }, + { + "epoch": 6.91, + "learning_rate": 9.624416244448508e-07, + "loss": 0.0865, + "step": 16495 + }, + { + "epoch": 6.91, + "learning_rate": 9.6171603737084e-07, + "loss": 0.0667, + "step": 16496 + }, + { + "epoch": 6.91, + "learning_rate": 9.609907100942828e-07, + "loss": 0.0828, + "step": 16497 + }, + { + "epoch": 6.91, + "learning_rate": 9.602656426360246e-07, + "loss": 0.0794, + "step": 16498 + }, + { + "epoch": 6.91, + "learning_rate": 9.595408350169088e-07, + "loss": 0.0645, + "step": 16499 + }, + { + "epoch": 6.91, + "learning_rate": 9.588162872577688e-07, + "loss": 0.0823, + "step": 16500 + }, + { + "epoch": 6.91, + "learning_rate": 9.580919993794325e-07, + "loss": 0.1073, + "step": 16501 + }, + { + "epoch": 6.91, + "learning_rate": 9.573679714027173e-07, + "loss": 0.0848, + "step": 16502 + }, + { + "epoch": 6.91, + "learning_rate": 9.56644203348438e-07, + "loss": 0.0776, + "step": 16503 + }, + { + "epoch": 6.91, + "learning_rate": 9.559206952373933e-07, + "loss": 0.0622, + "step": 16504 + }, + { + "epoch": 6.91, + "learning_rate": 9.551974470903836e-07, + "loss": 0.098, + "step": 16505 + }, + { + "epoch": 6.91, + "learning_rate": 9.544744589281963e-07, + "loss": 0.0602, + "step": 16506 + }, + { + "epoch": 6.91, + "learning_rate": 9.53751730771615e-07, + "loss": 0.0916, + "step": 16507 + }, + { + "epoch": 6.91, + "learning_rate": 9.530292626414118e-07, + "loss": 0.0735, + "step": 16508 + }, + { + "epoch": 6.91, + "learning_rate": 9.523070545583535e-07, + "loss": 0.1054, + "step": 16509 + }, + { + "epoch": 6.91, + "learning_rate": 9.515851065432036e-07, + "loss": 0.0622, + "step": 16510 + }, + { + "epoch": 6.91, + "learning_rate": 9.508634186167098e-07, + "loss": 0.0743, + "step": 16511 + }, + { + "epoch": 6.91, + "learning_rate": 9.501419907996145e-07, + "loss": 0.0788, + "step": 16512 + }, + { + "epoch": 6.91, + "learning_rate": 9.494208231126578e-07, + "loss": 0.0596, + "step": 16513 + }, + { + "epoch": 6.91, + "learning_rate": 9.486999155765675e-07, + "loss": 0.0891, + "step": 16514 + }, + { + "epoch": 6.91, + "learning_rate": 9.47979268212067e-07, + "loss": 0.0679, + "step": 16515 + }, + { + "epoch": 6.91, + "learning_rate": 9.472588810398698e-07, + "loss": 0.0551, + "step": 16516 + }, + { + "epoch": 6.92, + "learning_rate": 9.465387540806836e-07, + "loss": 0.09, + "step": 16517 + }, + { + "epoch": 6.92, + "learning_rate": 9.458188873552054e-07, + "loss": 0.081, + "step": 16518 + }, + { + "epoch": 6.92, + "learning_rate": 9.450992808841297e-07, + "loss": 0.1218, + "step": 16519 + }, + { + "epoch": 6.92, + "learning_rate": 9.443799346881388e-07, + "loss": 0.0777, + "step": 16520 + }, + { + "epoch": 6.92, + "learning_rate": 9.436608487879118e-07, + "loss": 0.0809, + "step": 16521 + }, + { + "epoch": 6.92, + "learning_rate": 9.429420232041176e-07, + "loss": 0.089, + "step": 16522 + }, + { + "epoch": 6.92, + "learning_rate": 9.422234579574163e-07, + "loss": 0.0882, + "step": 16523 + }, + { + "epoch": 6.92, + "learning_rate": 9.415051530684671e-07, + "loss": 0.1084, + "step": 16524 + }, + { + "epoch": 6.92, + "learning_rate": 9.407871085579113e-07, + "loss": 0.0849, + "step": 16525 + }, + { + "epoch": 6.92, + "learning_rate": 9.400693244463921e-07, + "loss": 0.0915, + "step": 16526 + }, + { + "epoch": 6.92, + "learning_rate": 9.393518007545399e-07, + "loss": 0.0645, + "step": 16527 + }, + { + "epoch": 6.92, + "learning_rate": 9.386345375029793e-07, + "loss": 0.0911, + "step": 16528 + }, + { + "epoch": 6.92, + "learning_rate": 9.379175347123292e-07, + "loss": 0.0836, + "step": 16529 + }, + { + "epoch": 6.92, + "learning_rate": 9.372007924031989e-07, + "loss": 0.0984, + "step": 16530 + }, + { + "epoch": 6.92, + "learning_rate": 9.364843105961885e-07, + "loss": 0.0794, + "step": 16531 + }, + { + "epoch": 6.92, + "learning_rate": 9.357680893118937e-07, + "loss": 0.0728, + "step": 16532 + }, + { + "epoch": 6.92, + "learning_rate": 9.350521285709025e-07, + "loss": 0.0916, + "step": 16533 + }, + { + "epoch": 6.92, + "learning_rate": 9.343364283937928e-07, + "loss": 0.0867, + "step": 16534 + }, + { + "epoch": 6.92, + "learning_rate": 9.336209888011383e-07, + "loss": 0.0782, + "step": 16535 + }, + { + "epoch": 6.92, + "learning_rate": 9.329058098135024e-07, + "loss": 0.0911, + "step": 16536 + }, + { + "epoch": 6.92, + "learning_rate": 9.321908914514421e-07, + "loss": 0.0689, + "step": 16537 + }, + { + "epoch": 6.92, + "learning_rate": 9.31476233735511e-07, + "loss": 0.0549, + "step": 16538 + }, + { + "epoch": 6.92, + "learning_rate": 9.30761836686247e-07, + "loss": 0.0838, + "step": 16539 + }, + { + "epoch": 6.92, + "learning_rate": 9.300477003241848e-07, + "loss": 0.0678, + "step": 16540 + }, + { + "epoch": 6.93, + "learning_rate": 9.293338246698536e-07, + "loss": 0.0905, + "step": 16541 + }, + { + "epoch": 6.93, + "learning_rate": 9.286202097437725e-07, + "loss": 0.1143, + "step": 16542 + }, + { + "epoch": 6.93, + "learning_rate": 9.279068555664528e-07, + "loss": 0.0711, + "step": 16543 + }, + { + "epoch": 6.93, + "learning_rate": 9.271937621584026e-07, + "loss": 0.0915, + "step": 16544 + }, + { + "epoch": 6.93, + "learning_rate": 9.264809295401156e-07, + "loss": 0.0726, + "step": 16545 + }, + { + "epoch": 6.93, + "learning_rate": 9.25768357732082e-07, + "loss": 0.0738, + "step": 16546 + }, + { + "epoch": 6.93, + "learning_rate": 9.250560467547842e-07, + "loss": 0.101, + "step": 16547 + }, + { + "epoch": 6.93, + "learning_rate": 9.243439966286971e-07, + "loss": 0.0644, + "step": 16548 + }, + { + "epoch": 6.93, + "learning_rate": 9.236322073742887e-07, + "loss": 0.0633, + "step": 16549 + }, + { + "epoch": 6.93, + "learning_rate": 9.229206790120204e-07, + "loss": 0.0722, + "step": 16550 + }, + { + "epoch": 6.93, + "learning_rate": 9.222094115623404e-07, + "loss": 0.0666, + "step": 16551 + }, + { + "epoch": 6.93, + "learning_rate": 9.214984050456954e-07, + "loss": 0.0654, + "step": 16552 + }, + { + "epoch": 6.93, + "learning_rate": 9.207876594825216e-07, + "loss": 0.0772, + "step": 16553 + }, + { + "epoch": 6.93, + "learning_rate": 9.200771748932513e-07, + "loss": 0.0715, + "step": 16554 + }, + { + "epoch": 6.93, + "learning_rate": 9.193669512983038e-07, + "loss": 0.0896, + "step": 16555 + }, + { + "epoch": 6.93, + "learning_rate": 9.186569887180985e-07, + "loss": 0.0803, + "step": 16556 + }, + { + "epoch": 6.93, + "learning_rate": 9.179472871730366e-07, + "loss": 0.0934, + "step": 16557 + }, + { + "epoch": 6.93, + "learning_rate": 9.17237846683523e-07, + "loss": 0.1161, + "step": 16558 + }, + { + "epoch": 6.93, + "learning_rate": 9.165286672699459e-07, + "loss": 0.0756, + "step": 16559 + }, + { + "epoch": 6.93, + "learning_rate": 9.15819748952691e-07, + "loss": 0.0932, + "step": 16560 + }, + { + "epoch": 6.93, + "learning_rate": 9.151110917521366e-07, + "loss": 0.0653, + "step": 16561 + }, + { + "epoch": 6.93, + "learning_rate": 9.14402695688652e-07, + "loss": 0.08, + "step": 16562 + }, + { + "epoch": 6.93, + "learning_rate": 9.136945607825998e-07, + "loss": 0.076, + "step": 16563 + }, + { + "epoch": 6.93, + "learning_rate": 9.129866870543347e-07, + "loss": 0.0529, + "step": 16564 + }, + { + "epoch": 6.94, + "learning_rate": 9.122790745242049e-07, + "loss": 0.0748, + "step": 16565 + }, + { + "epoch": 6.94, + "learning_rate": 9.115717232125476e-07, + "loss": 0.065, + "step": 16566 + }, + { + "epoch": 6.94, + "learning_rate": 9.108646331396953e-07, + "loss": 0.0946, + "step": 16567 + }, + { + "epoch": 6.94, + "learning_rate": 9.10157804325974e-07, + "loss": 0.0623, + "step": 16568 + }, + { + "epoch": 6.94, + "learning_rate": 9.094512367916996e-07, + "loss": 0.0917, + "step": 16569 + }, + { + "epoch": 6.94, + "learning_rate": 9.087449305571838e-07, + "loss": 0.0906, + "step": 16570 + }, + { + "epoch": 6.94, + "learning_rate": 9.080388856427291e-07, + "loss": 0.0834, + "step": 16571 + }, + { + "epoch": 6.94, + "learning_rate": 9.07333102068626e-07, + "loss": 0.0968, + "step": 16572 + }, + { + "epoch": 6.94, + "learning_rate": 9.066275798551649e-07, + "loss": 0.0786, + "step": 16573 + }, + { + "epoch": 6.94, + "learning_rate": 9.059223190226251e-07, + "loss": 0.0709, + "step": 16574 + }, + { + "epoch": 6.94, + "learning_rate": 9.052173195912772e-07, + "loss": 0.0399, + "step": 16575 + }, + { + "epoch": 6.94, + "learning_rate": 9.045125815813883e-07, + "loss": 0.0838, + "step": 16576 + }, + { + "epoch": 6.94, + "learning_rate": 9.038081050132141e-07, + "loss": 0.0638, + "step": 16577 + }, + { + "epoch": 6.94, + "learning_rate": 9.031038899070033e-07, + "loss": 0.1018, + "step": 16578 + }, + { + "epoch": 6.94, + "learning_rate": 9.023999362830005e-07, + "loss": 0.0962, + "step": 16579 + }, + { + "epoch": 6.94, + "learning_rate": 9.016962441614363e-07, + "loss": 0.0945, + "step": 16580 + }, + { + "epoch": 6.94, + "learning_rate": 9.009928135625412e-07, + "loss": 0.0737, + "step": 16581 + }, + { + "epoch": 6.94, + "learning_rate": 9.002896445065312e-07, + "loss": 0.0624, + "step": 16582 + }, + { + "epoch": 6.94, + "learning_rate": 8.995867370136224e-07, + "loss": 0.0751, + "step": 16583 + }, + { + "epoch": 6.94, + "learning_rate": 8.988840911040165e-07, + "loss": 0.0876, + "step": 16584 + }, + { + "epoch": 6.94, + "learning_rate": 8.981817067979126e-07, + "loss": 0.087, + "step": 16585 + }, + { + "epoch": 6.94, + "learning_rate": 8.974795841154971e-07, + "loss": 0.1017, + "step": 16586 + }, + { + "epoch": 6.94, + "learning_rate": 8.967777230769537e-07, + "loss": 0.0746, + "step": 16587 + }, + { + "epoch": 6.94, + "learning_rate": 8.960761237024573e-07, + "loss": 0.0726, + "step": 16588 + }, + { + "epoch": 6.95, + "learning_rate": 8.953747860121731e-07, + "loss": 0.084, + "step": 16589 + }, + { + "epoch": 6.95, + "learning_rate": 8.946737100262614e-07, + "loss": 0.0726, + "step": 16590 + }, + { + "epoch": 6.95, + "learning_rate": 8.939728957648763e-07, + "loss": 0.0612, + "step": 16591 + }, + { + "epoch": 6.95, + "learning_rate": 8.932723432481572e-07, + "loss": 0.0952, + "step": 16592 + }, + { + "epoch": 6.95, + "learning_rate": 8.925720524962445e-07, + "loss": 0.0716, + "step": 16593 + }, + { + "epoch": 6.95, + "learning_rate": 8.918720235292655e-07, + "loss": 0.0679, + "step": 16594 + }, + { + "epoch": 6.95, + "learning_rate": 8.91172256367343e-07, + "loss": 0.0933, + "step": 16595 + }, + { + "epoch": 6.95, + "learning_rate": 8.90472751030591e-07, + "loss": 0.0544, + "step": 16596 + }, + { + "epoch": 6.95, + "learning_rate": 8.897735075391156e-07, + "loss": 0.0806, + "step": 16597 + }, + { + "epoch": 6.95, + "learning_rate": 8.890745259130196e-07, + "loss": 0.1142, + "step": 16598 + }, + { + "epoch": 6.95, + "learning_rate": 8.88375806172388e-07, + "loss": 0.0725, + "step": 16599 + }, + { + "epoch": 6.95, + "learning_rate": 8.876773483373091e-07, + "loss": 0.083, + "step": 16600 + }, + { + "epoch": 6.95, + "learning_rate": 8.869791524278604e-07, + "loss": 0.0751, + "step": 16601 + }, + { + "epoch": 6.95, + "learning_rate": 8.862812184641078e-07, + "loss": 0.0732, + "step": 16602 + }, + { + "epoch": 6.95, + "learning_rate": 8.855835464661145e-07, + "loss": 0.0782, + "step": 16603 + }, + { + "epoch": 6.95, + "learning_rate": 8.84886136453933e-07, + "loss": 0.0566, + "step": 16604 + }, + { + "epoch": 6.95, + "learning_rate": 8.841889884476118e-07, + "loss": 0.0722, + "step": 16605 + }, + { + "epoch": 6.95, + "learning_rate": 8.834921024671905e-07, + "loss": 0.0737, + "step": 16606 + }, + { + "epoch": 6.95, + "learning_rate": 8.827954785326975e-07, + "loss": 0.0725, + "step": 16607 + }, + { + "epoch": 6.95, + "learning_rate": 8.820991166641568e-07, + "loss": 0.0714, + "step": 16608 + }, + { + "epoch": 6.95, + "learning_rate": 8.814030168815868e-07, + "loss": 0.075, + "step": 16609 + }, + { + "epoch": 6.95, + "learning_rate": 8.807071792049948e-07, + "loss": 0.0669, + "step": 16610 + }, + { + "epoch": 6.95, + "learning_rate": 8.800116036543838e-07, + "loss": 0.1129, + "step": 16611 + }, + { + "epoch": 6.95, + "learning_rate": 8.793162902497465e-07, + "loss": 0.0735, + "step": 16612 + }, + { + "epoch": 6.96, + "learning_rate": 8.786212390110682e-07, + "loss": 0.0874, + "step": 16613 + }, + { + "epoch": 6.96, + "learning_rate": 8.779264499583284e-07, + "loss": 0.0807, + "step": 16614 + }, + { + "epoch": 6.96, + "learning_rate": 8.772319231114978e-07, + "loss": 0.0741, + "step": 16615 + }, + { + "epoch": 6.96, + "learning_rate": 8.765376584905395e-07, + "loss": 0.0974, + "step": 16616 + }, + { + "epoch": 6.96, + "learning_rate": 8.758436561154105e-07, + "loss": 0.0881, + "step": 16617 + }, + { + "epoch": 6.96, + "learning_rate": 8.751499160060605e-07, + "loss": 0.0807, + "step": 16618 + }, + { + "epoch": 6.96, + "learning_rate": 8.744564381824284e-07, + "loss": 0.1121, + "step": 16619 + }, + { + "epoch": 6.96, + "learning_rate": 8.737632226644465e-07, + "loss": 0.0894, + "step": 16620 + }, + { + "epoch": 6.96, + "learning_rate": 8.730702694720428e-07, + "loss": 0.0836, + "step": 16621 + }, + { + "epoch": 6.96, + "learning_rate": 8.723775786251354e-07, + "loss": 0.0837, + "step": 16622 + }, + { + "epoch": 6.96, + "learning_rate": 8.716851501436352e-07, + "loss": 0.0691, + "step": 16623 + }, + { + "epoch": 6.96, + "learning_rate": 8.709929840474463e-07, + "loss": 0.1011, + "step": 16624 + }, + { + "epoch": 6.96, + "learning_rate": 8.703010803564616e-07, + "loss": 0.0834, + "step": 16625 + }, + { + "epoch": 6.96, + "learning_rate": 8.696094390905718e-07, + "loss": 0.0848, + "step": 16626 + }, + { + "epoch": 6.96, + "learning_rate": 8.689180602696556e-07, + "loss": 0.1116, + "step": 16627 + }, + { + "epoch": 6.96, + "learning_rate": 8.682269439135871e-07, + "loss": 0.0851, + "step": 16628 + }, + { + "epoch": 6.96, + "learning_rate": 8.675360900422314e-07, + "loss": 0.081, + "step": 16629 + }, + { + "epoch": 6.96, + "learning_rate": 8.66845498675446e-07, + "loss": 0.0749, + "step": 16630 + }, + { + "epoch": 6.96, + "learning_rate": 8.661551698330828e-07, + "loss": 0.0705, + "step": 16631 + }, + { + "epoch": 6.96, + "learning_rate": 8.65465103534986e-07, + "loss": 0.0742, + "step": 16632 + }, + { + "epoch": 6.96, + "learning_rate": 8.647752998009873e-07, + "loss": 0.095, + "step": 16633 + }, + { + "epoch": 6.96, + "learning_rate": 8.640857586509155e-07, + "loss": 0.088, + "step": 16634 + }, + { + "epoch": 6.96, + "learning_rate": 8.633964801045913e-07, + "loss": 0.0972, + "step": 16635 + }, + { + "epoch": 6.97, + "learning_rate": 8.62707464181829e-07, + "loss": 0.1162, + "step": 16636 + }, + { + "epoch": 6.97, + "learning_rate": 8.620187109024314e-07, + "loss": 0.0883, + "step": 16637 + }, + { + "epoch": 6.97, + "learning_rate": 8.613302202861962e-07, + "loss": 0.092, + "step": 16638 + }, + { + "epoch": 6.97, + "learning_rate": 8.606419923529175e-07, + "loss": 0.0411, + "step": 16639 + }, + { + "epoch": 6.97, + "learning_rate": 8.599540271223728e-07, + "loss": 0.0708, + "step": 16640 + }, + { + "epoch": 6.97, + "learning_rate": 8.592663246143385e-07, + "loss": 0.0768, + "step": 16641 + }, + { + "epoch": 6.97, + "learning_rate": 8.58578884848582e-07, + "loss": 0.1036, + "step": 16642 + }, + { + "epoch": 6.97, + "learning_rate": 8.578917078448634e-07, + "loss": 0.09, + "step": 16643 + }, + { + "epoch": 6.97, + "learning_rate": 8.572047936229355e-07, + "loss": 0.1143, + "step": 16644 + }, + { + "epoch": 6.97, + "learning_rate": 8.565181422025448e-07, + "loss": 0.0634, + "step": 16645 + }, + { + "epoch": 6.97, + "learning_rate": 8.558317536034233e-07, + "loss": 0.0729, + "step": 16646 + }, + { + "epoch": 6.97, + "learning_rate": 8.551456278453063e-07, + "loss": 0.0873, + "step": 16647 + }, + { + "epoch": 6.97, + "learning_rate": 8.544597649479114e-07, + "loss": 0.0769, + "step": 16648 + }, + { + "epoch": 6.97, + "learning_rate": 8.53774164930955e-07, + "loss": 0.0891, + "step": 16649 + }, + { + "epoch": 6.97, + "learning_rate": 8.530888278141436e-07, + "loss": 0.0774, + "step": 16650 + }, + { + "epoch": 6.97, + "learning_rate": 8.524037536171781e-07, + "loss": 0.0801, + "step": 16651 + }, + { + "epoch": 6.97, + "learning_rate": 8.517189423597483e-07, + "loss": 0.0774, + "step": 16652 + }, + { + "epoch": 6.97, + "learning_rate": 8.510343940615417e-07, + "loss": 0.0767, + "step": 16653 + }, + { + "epoch": 6.97, + "learning_rate": 8.503501087422305e-07, + "loss": 0.0944, + "step": 16654 + }, + { + "epoch": 6.97, + "learning_rate": 8.496660864214868e-07, + "loss": 0.0833, + "step": 16655 + }, + { + "epoch": 6.97, + "learning_rate": 8.489823271189712e-07, + "loss": 0.0941, + "step": 16656 + }, + { + "epoch": 6.97, + "learning_rate": 8.482988308543383e-07, + "loss": 0.0652, + "step": 16657 + }, + { + "epoch": 6.97, + "learning_rate": 8.476155976472345e-07, + "loss": 0.0684, + "step": 16658 + }, + { + "epoch": 6.97, + "learning_rate": 8.469326275172995e-07, + "loss": 0.0905, + "step": 16659 + }, + { + "epoch": 6.98, + "learning_rate": 8.462499204841635e-07, + "loss": 0.0842, + "step": 16660 + }, + { + "epoch": 6.98, + "learning_rate": 8.455674765674493e-07, + "loss": 0.0685, + "step": 16661 + }, + { + "epoch": 6.98, + "learning_rate": 8.448852957867748e-07, + "loss": 0.0514, + "step": 16662 + }, + { + "epoch": 6.98, + "learning_rate": 8.442033781617486e-07, + "loss": 0.1013, + "step": 16663 + }, + { + "epoch": 6.98, + "learning_rate": 8.435217237119719e-07, + "loss": 0.0819, + "step": 16664 + }, + { + "epoch": 6.98, + "learning_rate": 8.428403324570378e-07, + "loss": 0.0854, + "step": 16665 + }, + { + "epoch": 6.98, + "learning_rate": 8.421592044165338e-07, + "loss": 0.0651, + "step": 16666 + }, + { + "epoch": 6.98, + "learning_rate": 8.414783396100357e-07, + "loss": 0.0824, + "step": 16667 + }, + { + "epoch": 6.98, + "learning_rate": 8.407977380571153e-07, + "loss": 0.0874, + "step": 16668 + }, + { + "epoch": 6.98, + "learning_rate": 8.401173997773381e-07, + "loss": 0.0745, + "step": 16669 + }, + { + "epoch": 6.98, + "learning_rate": 8.394373247902565e-07, + "loss": 0.0691, + "step": 16670 + }, + { + "epoch": 6.98, + "learning_rate": 8.387575131154202e-07, + "loss": 0.0817, + "step": 16671 + }, + { + "epoch": 6.98, + "learning_rate": 8.380779647723692e-07, + "loss": 0.0862, + "step": 16672 + }, + { + "epoch": 6.98, + "learning_rate": 8.373986797806377e-07, + "loss": 0.0664, + "step": 16673 + }, + { + "epoch": 6.98, + "learning_rate": 8.367196581597503e-07, + "loss": 0.0629, + "step": 16674 + }, + { + "epoch": 6.98, + "learning_rate": 8.360408999292236e-07, + "loss": 0.0735, + "step": 16675 + }, + { + "epoch": 6.98, + "learning_rate": 8.353624051085696e-07, + "loss": 0.0621, + "step": 16676 + }, + { + "epoch": 6.98, + "learning_rate": 8.346841737172917e-07, + "loss": 0.0812, + "step": 16677 + }, + { + "epoch": 6.98, + "learning_rate": 8.340062057748832e-07, + "loss": 0.0491, + "step": 16678 + }, + { + "epoch": 6.98, + "learning_rate": 8.333285013008329e-07, + "loss": 0.1196, + "step": 16679 + }, + { + "epoch": 6.98, + "learning_rate": 8.32651060314622e-07, + "loss": 0.0909, + "step": 16680 + }, + { + "epoch": 6.98, + "learning_rate": 8.319738828357193e-07, + "loss": 0.0886, + "step": 16681 + }, + { + "epoch": 6.98, + "learning_rate": 8.312969688835926e-07, + "loss": 0.0663, + "step": 16682 + }, + { + "epoch": 6.98, + "learning_rate": 8.306203184776985e-07, + "loss": 0.0835, + "step": 16683 + }, + { + "epoch": 6.99, + "learning_rate": 8.299439316374858e-07, + "loss": 0.1121, + "step": 16684 + }, + { + "epoch": 6.99, + "learning_rate": 8.29267808382398e-07, + "loss": 0.085, + "step": 16685 + }, + { + "epoch": 6.99, + "learning_rate": 8.285919487318694e-07, + "loss": 0.0473, + "step": 16686 + }, + { + "epoch": 6.99, + "learning_rate": 8.279163527053268e-07, + "loss": 0.0749, + "step": 16687 + }, + { + "epoch": 6.99, + "learning_rate": 8.272410203221881e-07, + "loss": 0.0831, + "step": 16688 + }, + { + "epoch": 6.99, + "learning_rate": 8.265659516018676e-07, + "loss": 0.0757, + "step": 16689 + }, + { + "epoch": 6.99, + "learning_rate": 8.258911465637675e-07, + "loss": 0.1019, + "step": 16690 + }, + { + "epoch": 6.99, + "learning_rate": 8.252166052272859e-07, + "loss": 0.0633, + "step": 16691 + }, + { + "epoch": 6.99, + "learning_rate": 8.245423276118136e-07, + "loss": 0.0781, + "step": 16692 + }, + { + "epoch": 6.99, + "learning_rate": 8.238683137367276e-07, + "loss": 0.0457, + "step": 16693 + }, + { + "epoch": 6.99, + "learning_rate": 8.231945636214068e-07, + "loss": 0.065, + "step": 16694 + }, + { + "epoch": 6.99, + "learning_rate": 8.225210772852122e-07, + "loss": 0.0584, + "step": 16695 + }, + { + "epoch": 6.99, + "learning_rate": 8.218478547475051e-07, + "loss": 0.0754, + "step": 16696 + }, + { + "epoch": 6.99, + "learning_rate": 8.211748960276378e-07, + "loss": 0.094, + "step": 16697 + }, + { + "epoch": 6.99, + "learning_rate": 8.205022011449526e-07, + "loss": 0.0867, + "step": 16698 + }, + { + "epoch": 6.99, + "learning_rate": 8.198297701187852e-07, + "loss": 0.0904, + "step": 16699 + }, + { + "epoch": 6.99, + "learning_rate": 8.191576029684667e-07, + "loss": 0.1052, + "step": 16700 + }, + { + "epoch": 6.99, + "learning_rate": 8.18485699713315e-07, + "loss": 0.0751, + "step": 16701 + }, + { + "epoch": 6.99, + "learning_rate": 8.178140603726437e-07, + "loss": 0.0655, + "step": 16702 + }, + { + "epoch": 6.99, + "learning_rate": 8.171426849657594e-07, + "loss": 0.0832, + "step": 16703 + }, + { + "epoch": 6.99, + "learning_rate": 8.164715735119599e-07, + "loss": 0.0788, + "step": 16704 + }, + { + "epoch": 6.99, + "learning_rate": 8.158007260305356e-07, + "loss": 0.0843, + "step": 16705 + }, + { + "epoch": 6.99, + "learning_rate": 8.151301425407699e-07, + "loss": 0.1059, + "step": 16706 + }, + { + "epoch": 6.99, + "learning_rate": 8.144598230619383e-07, + "loss": 0.0749, + "step": 16707 + }, + { + "epoch": 7.0, + "learning_rate": 8.137897676133077e-07, + "loss": 0.0613, + "step": 16708 + }, + { + "epoch": 7.0, + "learning_rate": 8.131199762141373e-07, + "loss": 0.0587, + "step": 16709 + }, + { + "epoch": 7.0, + "learning_rate": 8.124504488836815e-07, + "loss": 0.0732, + "step": 16710 + }, + { + "epoch": 7.0, + "learning_rate": 8.11781185641185e-07, + "loss": 0.0801, + "step": 16711 + }, + { + "epoch": 7.0, + "learning_rate": 8.111121865058846e-07, + "loss": 0.0597, + "step": 16712 + }, + { + "epoch": 7.0, + "learning_rate": 8.104434514970117e-07, + "loss": 0.0654, + "step": 16713 + }, + { + "epoch": 7.0, + "learning_rate": 8.097749806337851e-07, + "loss": 0.0664, + "step": 16714 + }, + { + "epoch": 7.0, + "learning_rate": 8.091067739354241e-07, + "loss": 0.0875, + "step": 16715 + }, + { + "epoch": 7.0, + "learning_rate": 8.084388314211311e-07, + "loss": 0.0907, + "step": 16716 + }, + { + "epoch": 7.0, + "learning_rate": 8.077711531101074e-07, + "loss": 0.0948, + "step": 16717 + }, + { + "epoch": 7.0, + "learning_rate": 8.071037390215442e-07, + "loss": 0.1112, + "step": 16718 + }, + { + "epoch": 7.0, + "learning_rate": 8.064365891746273e-07, + "loss": 0.0912, + "step": 16719 + }, + { + "epoch": 7.0, + "learning_rate": 8.057697035885315e-07, + "loss": 0.0744, + "step": 16720 + }, + { + "epoch": 7.0, + "learning_rate": 8.05103082282429e-07, + "loss": 0.0586, + "step": 16721 + }, + { + "epoch": 7.0, + "learning_rate": 8.04436725275477e-07, + "loss": 0.0533, + "step": 16722 + }, + { + "epoch": 7.0, + "learning_rate": 8.037706325868322e-07, + "loss": 0.0551, + "step": 16723 + }, + { + "epoch": 7.0, + "learning_rate": 8.031048042356393e-07, + "loss": 0.0567, + "step": 16724 + }, + { + "epoch": 7.0, + "learning_rate": 8.024392402410375e-07, + "loss": 0.059, + "step": 16725 + }, + { + "epoch": 7.0, + "learning_rate": 8.017739406221581e-07, + "loss": 0.039, + "step": 16726 + }, + { + "epoch": 7.0, + "learning_rate": 8.011089053981258e-07, + "loss": 0.0584, + "step": 16727 + }, + { + "epoch": 7.0, + "learning_rate": 8.004441345880542e-07, + "loss": 0.0674, + "step": 16728 + }, + { + "epoch": 7.0, + "learning_rate": 7.997796282110514e-07, + "loss": 0.0421, + "step": 16729 + }, + { + "epoch": 7.0, + "learning_rate": 7.9911538628622e-07, + "loss": 0.0638, + "step": 16730 + }, + { + "epoch": 7.0, + "learning_rate": 7.984514088326512e-07, + "loss": 0.0644, + "step": 16731 + }, + { + "epoch": 7.01, + "learning_rate": 7.977876958694319e-07, + "loss": 0.0453, + "step": 16732 + }, + { + "epoch": 7.01, + "learning_rate": 7.971242474156393e-07, + "loss": 0.0523, + "step": 16733 + }, + { + "epoch": 7.01, + "learning_rate": 7.964610634903458e-07, + "loss": 0.0488, + "step": 16734 + }, + { + "epoch": 7.01, + "learning_rate": 7.957981441126094e-07, + "loss": 0.0442, + "step": 16735 + }, + { + "epoch": 7.01, + "learning_rate": 7.951354893014885e-07, + "loss": 0.0343, + "step": 16736 + }, + { + "epoch": 7.01, + "learning_rate": 7.944730990760307e-07, + "loss": 0.0534, + "step": 16737 + }, + { + "epoch": 7.01, + "learning_rate": 7.938109734552735e-07, + "loss": 0.0261, + "step": 16738 + }, + { + "epoch": 7.01, + "learning_rate": 7.931491124582503e-07, + "loss": 0.0428, + "step": 16739 + }, + { + "epoch": 7.01, + "learning_rate": 7.924875161039869e-07, + "loss": 0.0633, + "step": 16740 + }, + { + "epoch": 7.01, + "learning_rate": 7.918261844114994e-07, + "loss": 0.0419, + "step": 16741 + }, + { + "epoch": 7.01, + "learning_rate": 7.911651173997959e-07, + "loss": 0.0469, + "step": 16742 + }, + { + "epoch": 7.01, + "learning_rate": 7.9050431508788e-07, + "loss": 0.0343, + "step": 16743 + }, + { + "epoch": 7.01, + "learning_rate": 7.898437774947442e-07, + "loss": 0.0416, + "step": 16744 + }, + { + "epoch": 7.01, + "learning_rate": 7.89183504639377e-07, + "loss": 0.0553, + "step": 16745 + }, + { + "epoch": 7.01, + "learning_rate": 7.885234965407551e-07, + "loss": 0.0456, + "step": 16746 + }, + { + "epoch": 7.01, + "learning_rate": 7.878637532178523e-07, + "loss": 0.0493, + "step": 16747 + }, + { + "epoch": 7.01, + "learning_rate": 7.872042746896324e-07, + "loss": 0.0584, + "step": 16748 + }, + { + "epoch": 7.01, + "learning_rate": 7.865450609750492e-07, + "loss": 0.0441, + "step": 16749 + }, + { + "epoch": 7.01, + "learning_rate": 7.858861120930516e-07, + "loss": 0.069, + "step": 16750 + }, + { + "epoch": 7.01, + "learning_rate": 7.852274280625815e-07, + "loss": 0.0656, + "step": 16751 + }, + { + "epoch": 7.01, + "learning_rate": 7.845690089025726e-07, + "loss": 0.0386, + "step": 16752 + }, + { + "epoch": 7.01, + "learning_rate": 7.839108546319485e-07, + "loss": 0.0722, + "step": 16753 + }, + { + "epoch": 7.01, + "learning_rate": 7.832529652696319e-07, + "loss": 0.0487, + "step": 16754 + }, + { + "epoch": 7.01, + "learning_rate": 7.825953408345277e-07, + "loss": 0.049, + "step": 16755 + }, + { + "epoch": 7.02, + "learning_rate": 7.819379813455407e-07, + "loss": 0.0345, + "step": 16756 + }, + { + "epoch": 7.02, + "learning_rate": 7.812808868215671e-07, + "loss": 0.0533, + "step": 16757 + }, + { + "epoch": 7.02, + "learning_rate": 7.806240572814927e-07, + "loss": 0.0435, + "step": 16758 + }, + { + "epoch": 7.02, + "learning_rate": 7.799674927442003e-07, + "loss": 0.0526, + "step": 16759 + }, + { + "epoch": 7.02, + "learning_rate": 7.793111932285613e-07, + "loss": 0.032, + "step": 16760 + }, + { + "epoch": 7.02, + "learning_rate": 7.786551587534386e-07, + "loss": 0.0518, + "step": 16761 + }, + { + "epoch": 7.02, + "learning_rate": 7.779993893376914e-07, + "loss": 0.0535, + "step": 16762 + }, + { + "epoch": 7.02, + "learning_rate": 7.773438850001669e-07, + "loss": 0.0357, + "step": 16763 + }, + { + "epoch": 7.02, + "learning_rate": 7.766886457597089e-07, + "loss": 0.0406, + "step": 16764 + }, + { + "epoch": 7.02, + "learning_rate": 7.760336716351502e-07, + "loss": 0.0399, + "step": 16765 + }, + { + "epoch": 7.02, + "learning_rate": 7.75378962645319e-07, + "loss": 0.0497, + "step": 16766 + }, + { + "epoch": 7.02, + "learning_rate": 7.747245188090335e-07, + "loss": 0.0407, + "step": 16767 + }, + { + "epoch": 7.02, + "learning_rate": 7.740703401451078e-07, + "loss": 0.0417, + "step": 16768 + }, + { + "epoch": 7.02, + "learning_rate": 7.734164266723409e-07, + "loss": 0.0706, + "step": 16769 + }, + { + "epoch": 7.02, + "learning_rate": 7.727627784095315e-07, + "loss": 0.0491, + "step": 16770 + }, + { + "epoch": 7.02, + "learning_rate": 7.721093953754677e-07, + "loss": 0.0575, + "step": 16771 + }, + { + "epoch": 7.02, + "learning_rate": 7.714562775889301e-07, + "loss": 0.0509, + "step": 16772 + }, + { + "epoch": 7.02, + "learning_rate": 7.708034250686936e-07, + "loss": 0.0662, + "step": 16773 + }, + { + "epoch": 7.02, + "learning_rate": 7.701508378335243e-07, + "loss": 0.0647, + "step": 16774 + }, + { + "epoch": 7.02, + "learning_rate": 7.694985159021762e-07, + "loss": 0.0538, + "step": 16775 + }, + { + "epoch": 7.02, + "learning_rate": 7.68846459293403e-07, + "loss": 0.0805, + "step": 16776 + }, + { + "epoch": 7.02, + "learning_rate": 7.681946680259467e-07, + "loss": 0.043, + "step": 16777 + }, + { + "epoch": 7.02, + "learning_rate": 7.675431421185431e-07, + "loss": 0.0544, + "step": 16778 + }, + { + "epoch": 7.02, + "learning_rate": 7.668918815899184e-07, + "loss": 0.0521, + "step": 16779 + }, + { + "epoch": 7.03, + "learning_rate": 7.662408864587933e-07, + "loss": 0.0247, + "step": 16780 + }, + { + "epoch": 7.03, + "learning_rate": 7.655901567438817e-07, + "loss": 0.0483, + "step": 16781 + }, + { + "epoch": 7.03, + "learning_rate": 7.649396924638841e-07, + "loss": 0.0487, + "step": 16782 + }, + { + "epoch": 7.03, + "learning_rate": 7.642894936375034e-07, + "loss": 0.0609, + "step": 16783 + }, + { + "epoch": 7.03, + "learning_rate": 7.636395602834223e-07, + "loss": 0.0555, + "step": 16784 + }, + { + "epoch": 7.03, + "learning_rate": 7.629898924203272e-07, + "loss": 0.04, + "step": 16785 + }, + { + "epoch": 7.03, + "learning_rate": 7.623404900668896e-07, + "loss": 0.0494, + "step": 16786 + }, + { + "epoch": 7.03, + "learning_rate": 7.616913532417781e-07, + "loss": 0.039, + "step": 16787 + }, + { + "epoch": 7.03, + "learning_rate": 7.610424819636508e-07, + "loss": 0.0592, + "step": 16788 + }, + { + "epoch": 7.03, + "learning_rate": 7.603938762511597e-07, + "loss": 0.0331, + "step": 16789 + }, + { + "epoch": 7.03, + "learning_rate": 7.597455361229466e-07, + "loss": 0.0446, + "step": 16790 + }, + { + "epoch": 7.03, + "learning_rate": 7.590974615976477e-07, + "loss": 0.0436, + "step": 16791 + }, + { + "epoch": 7.03, + "learning_rate": 7.584496526938933e-07, + "loss": 0.0654, + "step": 16792 + }, + { + "epoch": 7.03, + "learning_rate": 7.578021094303012e-07, + "loss": 0.0466, + "step": 16793 + }, + { + "epoch": 7.03, + "learning_rate": 7.571548318254874e-07, + "loss": 0.0442, + "step": 16794 + }, + { + "epoch": 7.03, + "learning_rate": 7.565078198980569e-07, + "loss": 0.0503, + "step": 16795 + }, + { + "epoch": 7.03, + "learning_rate": 7.558610736666061e-07, + "loss": 0.0542, + "step": 16796 + }, + { + "epoch": 7.03, + "learning_rate": 7.552145931497246e-07, + "loss": 0.0584, + "step": 16797 + }, + { + "epoch": 7.03, + "learning_rate": 7.545683783659963e-07, + "loss": 0.0533, + "step": 16798 + }, + { + "epoch": 7.03, + "learning_rate": 7.539224293339953e-07, + "loss": 0.0321, + "step": 16799 + }, + { + "epoch": 7.03, + "learning_rate": 7.53276746072289e-07, + "loss": 0.058, + "step": 16800 + }, + { + "epoch": 7.03, + "learning_rate": 7.526313285994391e-07, + "loss": 0.0487, + "step": 16801 + }, + { + "epoch": 7.03, + "learning_rate": 7.519861769339942e-07, + "loss": 0.0787, + "step": 16802 + }, + { + "epoch": 7.03, + "learning_rate": 7.513412910945006e-07, + "loss": 0.0295, + "step": 16803 + }, + { + "epoch": 7.04, + "learning_rate": 7.506966710994945e-07, + "loss": 0.0386, + "step": 16804 + }, + { + "epoch": 7.04, + "learning_rate": 7.500523169675056e-07, + "loss": 0.0329, + "step": 16805 + }, + { + "epoch": 7.04, + "learning_rate": 7.494082287170535e-07, + "loss": 0.044, + "step": 16806 + }, + { + "epoch": 7.04, + "learning_rate": 7.487644063666532e-07, + "loss": 0.0363, + "step": 16807 + }, + { + "epoch": 7.04, + "learning_rate": 7.481208499348102e-07, + "loss": 0.0662, + "step": 16808 + }, + { + "epoch": 7.04, + "learning_rate": 7.474775594400252e-07, + "loss": 0.0397, + "step": 16809 + }, + { + "epoch": 7.04, + "learning_rate": 7.468345349007855e-07, + "loss": 0.0364, + "step": 16810 + }, + { + "epoch": 7.04, + "learning_rate": 7.461917763355764e-07, + "loss": 0.0466, + "step": 16811 + }, + { + "epoch": 7.04, + "learning_rate": 7.455492837628731e-07, + "loss": 0.0587, + "step": 16812 + }, + { + "epoch": 7.04, + "learning_rate": 7.449070572011429e-07, + "loss": 0.0536, + "step": 16813 + }, + { + "epoch": 7.04, + "learning_rate": 7.442650966688469e-07, + "loss": 0.0586, + "step": 16814 + }, + { + "epoch": 7.04, + "learning_rate": 7.43623402184438e-07, + "loss": 0.0411, + "step": 16815 + }, + { + "epoch": 7.04, + "learning_rate": 7.429819737663602e-07, + "loss": 0.0405, + "step": 16816 + }, + { + "epoch": 7.04, + "learning_rate": 7.423408114330499e-07, + "loss": 0.0509, + "step": 16817 + }, + { + "epoch": 7.04, + "learning_rate": 7.41699915202938e-07, + "loss": 0.0519, + "step": 16818 + }, + { + "epoch": 7.04, + "learning_rate": 7.410592850944476e-07, + "loss": 0.0512, + "step": 16819 + }, + { + "epoch": 7.04, + "learning_rate": 7.404189211259905e-07, + "loss": 0.0485, + "step": 16820 + }, + { + "epoch": 7.04, + "learning_rate": 7.397788233159753e-07, + "loss": 0.0434, + "step": 16821 + }, + { + "epoch": 7.04, + "learning_rate": 7.391389916828018e-07, + "loss": 0.0478, + "step": 16822 + }, + { + "epoch": 7.04, + "learning_rate": 7.384994262448586e-07, + "loss": 0.0408, + "step": 16823 + }, + { + "epoch": 7.04, + "learning_rate": 7.378601270205299e-07, + "loss": 0.0521, + "step": 16824 + }, + { + "epoch": 7.04, + "learning_rate": 7.372210940281943e-07, + "loss": 0.0469, + "step": 16825 + }, + { + "epoch": 7.04, + "learning_rate": 7.365823272862183e-07, + "loss": 0.056, + "step": 16826 + }, + { + "epoch": 7.05, + "learning_rate": 7.359438268129626e-07, + "loss": 0.0539, + "step": 16827 + }, + { + "epoch": 7.05, + "learning_rate": 7.353055926267827e-07, + "loss": 0.0601, + "step": 16828 + }, + { + "epoch": 7.05, + "learning_rate": 7.346676247460205e-07, + "loss": 0.0615, + "step": 16829 + }, + { + "epoch": 7.05, + "learning_rate": 7.340299231890169e-07, + "loss": 0.0734, + "step": 16830 + }, + { + "epoch": 7.05, + "learning_rate": 7.333924879740994e-07, + "loss": 0.0506, + "step": 16831 + }, + { + "epoch": 7.05, + "learning_rate": 7.327553191195913e-07, + "loss": 0.0775, + "step": 16832 + }, + { + "epoch": 7.05, + "learning_rate": 7.321184166438089e-07, + "loss": 0.0427, + "step": 16833 + }, + { + "epoch": 7.05, + "learning_rate": 7.314817805650576e-07, + "loss": 0.0436, + "step": 16834 + }, + { + "epoch": 7.05, + "learning_rate": 7.308454109016383e-07, + "loss": 0.063, + "step": 16835 + }, + { + "epoch": 7.05, + "learning_rate": 7.302093076718442e-07, + "loss": 0.0398, + "step": 16836 + }, + { + "epoch": 7.05, + "learning_rate": 7.295734708939551e-07, + "loss": 0.062, + "step": 16837 + }, + { + "epoch": 7.05, + "learning_rate": 7.289379005862507e-07, + "loss": 0.0487, + "step": 16838 + }, + { + "epoch": 7.05, + "learning_rate": 7.283025967669998e-07, + "loss": 0.0399, + "step": 16839 + }, + { + "epoch": 7.05, + "learning_rate": 7.276675594544624e-07, + "loss": 0.0421, + "step": 16840 + }, + { + "epoch": 7.05, + "learning_rate": 7.270327886668937e-07, + "loss": 0.0248, + "step": 16841 + }, + { + "epoch": 7.05, + "learning_rate": 7.263982844225393e-07, + "loss": 0.0509, + "step": 16842 + }, + { + "epoch": 7.05, + "learning_rate": 7.257640467396365e-07, + "loss": 0.0543, + "step": 16843 + }, + { + "epoch": 7.05, + "learning_rate": 7.251300756364154e-07, + "loss": 0.0344, + "step": 16844 + }, + { + "epoch": 7.05, + "learning_rate": 7.244963711311003e-07, + "loss": 0.0437, + "step": 16845 + }, + { + "epoch": 7.05, + "learning_rate": 7.238629332419068e-07, + "loss": 0.0437, + "step": 16846 + }, + { + "epoch": 7.05, + "learning_rate": 7.23229761987041e-07, + "loss": 0.059, + "step": 16847 + }, + { + "epoch": 7.05, + "learning_rate": 7.225968573847031e-07, + "loss": 0.0455, + "step": 16848 + }, + { + "epoch": 7.05, + "learning_rate": 7.219642194530885e-07, + "loss": 0.0446, + "step": 16849 + }, + { + "epoch": 7.05, + "learning_rate": 7.21331848210377e-07, + "loss": 0.0713, + "step": 16850 + }, + { + "epoch": 7.06, + "learning_rate": 7.206997436747498e-07, + "loss": 0.0498, + "step": 16851 + }, + { + "epoch": 7.06, + "learning_rate": 7.200679058643733e-07, + "loss": 0.0443, + "step": 16852 + }, + { + "epoch": 7.06, + "learning_rate": 7.194363347974087e-07, + "loss": 0.0665, + "step": 16853 + }, + { + "epoch": 7.06, + "learning_rate": 7.188050304920113e-07, + "loss": 0.0543, + "step": 16854 + }, + { + "epoch": 7.06, + "learning_rate": 7.181739929663279e-07, + "loss": 0.0494, + "step": 16855 + }, + { + "epoch": 7.06, + "learning_rate": 7.175432222384981e-07, + "loss": 0.0478, + "step": 16856 + }, + { + "epoch": 7.06, + "learning_rate": 7.169127183266489e-07, + "loss": 0.0669, + "step": 16857 + }, + { + "epoch": 7.06, + "learning_rate": 7.162824812489067e-07, + "loss": 0.0413, + "step": 16858 + }, + { + "epoch": 7.06, + "learning_rate": 7.15652511023387e-07, + "loss": 0.0468, + "step": 16859 + }, + { + "epoch": 7.06, + "learning_rate": 7.150228076681954e-07, + "loss": 0.0416, + "step": 16860 + }, + { + "epoch": 7.06, + "learning_rate": 7.14393371201435e-07, + "loss": 0.0476, + "step": 16861 + }, + { + "epoch": 7.06, + "learning_rate": 7.137642016411972e-07, + "loss": 0.0451, + "step": 16862 + }, + { + "epoch": 7.06, + "learning_rate": 7.131352990055684e-07, + "loss": 0.0461, + "step": 16863 + }, + { + "epoch": 7.06, + "learning_rate": 7.125066633126221e-07, + "loss": 0.0596, + "step": 16864 + }, + { + "epoch": 7.06, + "learning_rate": 7.118782945804315e-07, + "loss": 0.0413, + "step": 16865 + }, + { + "epoch": 7.06, + "learning_rate": 7.112501928270565e-07, + "loss": 0.0538, + "step": 16866 + }, + { + "epoch": 7.06, + "learning_rate": 7.106223580705518e-07, + "loss": 0.0509, + "step": 16867 + }, + { + "epoch": 7.06, + "learning_rate": 7.099947903289639e-07, + "loss": 0.0664, + "step": 16868 + }, + { + "epoch": 7.06, + "learning_rate": 7.09367489620334e-07, + "loss": 0.0486, + "step": 16869 + }, + { + "epoch": 7.06, + "learning_rate": 7.087404559626898e-07, + "loss": 0.0577, + "step": 16870 + }, + { + "epoch": 7.06, + "learning_rate": 7.08113689374057e-07, + "loss": 0.0703, + "step": 16871 + }, + { + "epoch": 7.06, + "learning_rate": 7.074871898724489e-07, + "loss": 0.0454, + "step": 16872 + }, + { + "epoch": 7.06, + "learning_rate": 7.068609574758789e-07, + "loss": 0.045, + "step": 16873 + }, + { + "epoch": 7.06, + "learning_rate": 7.062349922023414e-07, + "loss": 0.0393, + "step": 16874 + }, + { + "epoch": 7.07, + "learning_rate": 7.056092940698323e-07, + "loss": 0.0451, + "step": 16875 + }, + { + "epoch": 7.07, + "learning_rate": 7.049838630963357e-07, + "loss": 0.0395, + "step": 16876 + }, + { + "epoch": 7.07, + "learning_rate": 7.04358699299832e-07, + "loss": 0.047, + "step": 16877 + }, + { + "epoch": 7.07, + "learning_rate": 7.037338026982877e-07, + "loss": 0.0573, + "step": 16878 + }, + { + "epoch": 7.07, + "learning_rate": 7.03109173309664e-07, + "loss": 0.0497, + "step": 16879 + }, + { + "epoch": 7.07, + "learning_rate": 7.024848111519189e-07, + "loss": 0.0619, + "step": 16880 + }, + { + "epoch": 7.07, + "learning_rate": 7.018607162429969e-07, + "loss": 0.0582, + "step": 16881 + }, + { + "epoch": 7.07, + "learning_rate": 7.012368886008369e-07, + "loss": 0.0479, + "step": 16882 + }, + { + "epoch": 7.07, + "learning_rate": 7.006133282433736e-07, + "loss": 0.0595, + "step": 16883 + }, + { + "epoch": 7.07, + "learning_rate": 6.999900351885258e-07, + "loss": 0.0317, + "step": 16884 + }, + { + "epoch": 7.07, + "learning_rate": 6.993670094542126e-07, + "loss": 0.0367, + "step": 16885 + }, + { + "epoch": 7.07, + "learning_rate": 6.987442510583409e-07, + "loss": 0.0383, + "step": 16886 + }, + { + "epoch": 7.07, + "learning_rate": 6.981217600188117e-07, + "loss": 0.0412, + "step": 16887 + }, + { + "epoch": 7.07, + "learning_rate": 6.974995363535186e-07, + "loss": 0.0623, + "step": 16888 + }, + { + "epoch": 7.07, + "learning_rate": 6.968775800803462e-07, + "loss": 0.0444, + "step": 16889 + }, + { + "epoch": 7.07, + "learning_rate": 6.962558912171735e-07, + "loss": 0.0342, + "step": 16890 + }, + { + "epoch": 7.07, + "learning_rate": 6.956344697818684e-07, + "loss": 0.0453, + "step": 16891 + }, + { + "epoch": 7.07, + "learning_rate": 6.950133157922934e-07, + "loss": 0.0417, + "step": 16892 + }, + { + "epoch": 7.07, + "learning_rate": 6.94392429266304e-07, + "loss": 0.0654, + "step": 16893 + }, + { + "epoch": 7.07, + "learning_rate": 6.937718102217461e-07, + "loss": 0.0427, + "step": 16894 + }, + { + "epoch": 7.07, + "learning_rate": 6.931514586764587e-07, + "loss": 0.0379, + "step": 16895 + }, + { + "epoch": 7.07, + "learning_rate": 6.925313746482765e-07, + "loss": 0.0404, + "step": 16896 + }, + { + "epoch": 7.07, + "learning_rate": 6.919115581550207e-07, + "loss": 0.0481, + "step": 16897 + }, + { + "epoch": 7.07, + "learning_rate": 6.912920092145048e-07, + "loss": 0.0624, + "step": 16898 + }, + { + "epoch": 7.08, + "learning_rate": 6.906727278445402e-07, + "loss": 0.0395, + "step": 16899 + }, + { + "epoch": 7.08, + "learning_rate": 6.900537140629282e-07, + "loss": 0.0761, + "step": 16900 + }, + { + "epoch": 7.08, + "learning_rate": 6.89434967887459e-07, + "loss": 0.0683, + "step": 16901 + }, + { + "epoch": 7.08, + "learning_rate": 6.888164893359206e-07, + "loss": 0.0487, + "step": 16902 + }, + { + "epoch": 7.08, + "learning_rate": 6.881982784260887e-07, + "loss": 0.0503, + "step": 16903 + }, + { + "epoch": 7.08, + "learning_rate": 6.875803351757371e-07, + "loss": 0.0449, + "step": 16904 + }, + { + "epoch": 7.08, + "learning_rate": 6.869626596026225e-07, + "loss": 0.052, + "step": 16905 + }, + { + "epoch": 7.08, + "learning_rate": 6.863452517245017e-07, + "loss": 0.0368, + "step": 16906 + }, + { + "epoch": 7.08, + "learning_rate": 6.85728111559123e-07, + "loss": 0.0517, + "step": 16907 + }, + { + "epoch": 7.08, + "learning_rate": 6.851112391242231e-07, + "loss": 0.049, + "step": 16908 + }, + { + "epoch": 7.08, + "learning_rate": 6.844946344375358e-07, + "loss": 0.0619, + "step": 16909 + }, + { + "epoch": 7.08, + "learning_rate": 6.838782975167846e-07, + "loss": 0.0458, + "step": 16910 + }, + { + "epoch": 7.08, + "learning_rate": 6.832622283796842e-07, + "loss": 0.0333, + "step": 16911 + }, + { + "epoch": 7.08, + "learning_rate": 6.826464270439426e-07, + "loss": 0.0416, + "step": 16912 + }, + { + "epoch": 7.08, + "learning_rate": 6.820308935272612e-07, + "loss": 0.046, + "step": 16913 + }, + { + "epoch": 7.08, + "learning_rate": 6.814156278473338e-07, + "loss": 0.0552, + "step": 16914 + }, + { + "epoch": 7.08, + "learning_rate": 6.80800630021845e-07, + "loss": 0.059, + "step": 16915 + }, + { + "epoch": 7.08, + "learning_rate": 6.801859000684719e-07, + "loss": 0.05, + "step": 16916 + }, + { + "epoch": 7.08, + "learning_rate": 6.795714380048868e-07, + "loss": 0.0466, + "step": 16917 + }, + { + "epoch": 7.08, + "learning_rate": 6.78957243848748e-07, + "loss": 0.0438, + "step": 16918 + }, + { + "epoch": 7.08, + "learning_rate": 6.783433176177134e-07, + "loss": 0.0654, + "step": 16919 + }, + { + "epoch": 7.08, + "learning_rate": 6.777296593294269e-07, + "loss": 0.0482, + "step": 16920 + }, + { + "epoch": 7.08, + "learning_rate": 6.771162690015298e-07, + "loss": 0.0327, + "step": 16921 + }, + { + "epoch": 7.08, + "learning_rate": 6.765031466516514e-07, + "loss": 0.0551, + "step": 16922 + }, + { + "epoch": 7.09, + "learning_rate": 6.758902922974165e-07, + "loss": 0.0455, + "step": 16923 + }, + { + "epoch": 7.09, + "learning_rate": 6.752777059564431e-07, + "loss": 0.0405, + "step": 16924 + }, + { + "epoch": 7.09, + "learning_rate": 6.74665387646336e-07, + "loss": 0.0494, + "step": 16925 + }, + { + "epoch": 7.09, + "learning_rate": 6.740533373846969e-07, + "loss": 0.0416, + "step": 16926 + }, + { + "epoch": 7.09, + "learning_rate": 6.734415551891193e-07, + "loss": 0.0524, + "step": 16927 + }, + { + "epoch": 7.09, + "learning_rate": 6.728300410771871e-07, + "loss": 0.0335, + "step": 16928 + }, + { + "epoch": 7.09, + "learning_rate": 6.722187950664794e-07, + "loss": 0.065, + "step": 16929 + }, + { + "epoch": 7.09, + "learning_rate": 6.716078171745643e-07, + "loss": 0.0502, + "step": 16930 + }, + { + "epoch": 7.09, + "learning_rate": 6.709971074190069e-07, + "loss": 0.0467, + "step": 16931 + }, + { + "epoch": 7.09, + "learning_rate": 6.703866658173574e-07, + "loss": 0.0301, + "step": 16932 + }, + { + "epoch": 7.09, + "learning_rate": 6.697764923871652e-07, + "loss": 0.0571, + "step": 16933 + }, + { + "epoch": 7.09, + "learning_rate": 6.691665871459674e-07, + "loss": 0.0473, + "step": 16934 + }, + { + "epoch": 7.09, + "learning_rate": 6.685569501112954e-07, + "loss": 0.0477, + "step": 16935 + }, + { + "epoch": 7.09, + "learning_rate": 6.679475813006742e-07, + "loss": 0.0532, + "step": 16936 + }, + { + "epoch": 7.09, + "learning_rate": 6.673384807316197e-07, + "loss": 0.0344, + "step": 16937 + }, + { + "epoch": 7.09, + "learning_rate": 6.66729648421638e-07, + "loss": 0.0632, + "step": 16938 + }, + { + "epoch": 7.09, + "learning_rate": 6.661210843882293e-07, + "loss": 0.0556, + "step": 16939 + }, + { + "epoch": 7.09, + "learning_rate": 6.655127886488877e-07, + "loss": 0.0644, + "step": 16940 + }, + { + "epoch": 7.09, + "learning_rate": 6.649047612211002e-07, + "loss": 0.0481, + "step": 16941 + }, + { + "epoch": 7.09, + "learning_rate": 6.642970021223383e-07, + "loss": 0.0634, + "step": 16942 + }, + { + "epoch": 7.09, + "learning_rate": 6.636895113700759e-07, + "loss": 0.0725, + "step": 16943 + }, + { + "epoch": 7.09, + "learning_rate": 6.630822889817734e-07, + "loss": 0.053, + "step": 16944 + }, + { + "epoch": 7.09, + "learning_rate": 6.62475334974887e-07, + "loss": 0.0539, + "step": 16945 + }, + { + "epoch": 7.09, + "learning_rate": 6.618686493668591e-07, + "loss": 0.0409, + "step": 16946 + }, + { + "epoch": 7.1, + "learning_rate": 6.612622321751306e-07, + "loss": 0.0357, + "step": 16947 + }, + { + "epoch": 7.1, + "learning_rate": 6.606560834171316e-07, + "loss": 0.0606, + "step": 16948 + }, + { + "epoch": 7.1, + "learning_rate": 6.600502031102863e-07, + "loss": 0.0367, + "step": 16949 + }, + { + "epoch": 7.1, + "learning_rate": 6.594445912720093e-07, + "loss": 0.0457, + "step": 16950 + }, + { + "epoch": 7.1, + "learning_rate": 6.588392479197115e-07, + "loss": 0.0349, + "step": 16951 + }, + { + "epoch": 7.1, + "learning_rate": 6.582341730707875e-07, + "loss": 0.0528, + "step": 16952 + }, + { + "epoch": 7.1, + "learning_rate": 6.576293667426337e-07, + "loss": 0.0508, + "step": 16953 + }, + { + "epoch": 7.1, + "learning_rate": 6.570248289526326e-07, + "loss": 0.0495, + "step": 16954 + }, + { + "epoch": 7.1, + "learning_rate": 6.564205597181627e-07, + "loss": 0.0423, + "step": 16955 + }, + { + "epoch": 7.1, + "learning_rate": 6.558165590565924e-07, + "loss": 0.059, + "step": 16956 + }, + { + "epoch": 7.1, + "learning_rate": 6.552128269852842e-07, + "loss": 0.0425, + "step": 16957 + }, + { + "epoch": 7.1, + "learning_rate": 6.546093635215888e-07, + "loss": 0.0472, + "step": 16958 + }, + { + "epoch": 7.1, + "learning_rate": 6.540061686828559e-07, + "loss": 0.0475, + "step": 16959 + }, + { + "epoch": 7.1, + "learning_rate": 6.534032424864211e-07, + "loss": 0.0521, + "step": 16960 + }, + { + "epoch": 7.1, + "learning_rate": 6.528005849496166e-07, + "loss": 0.0355, + "step": 16961 + }, + { + "epoch": 7.1, + "learning_rate": 6.521981960897639e-07, + "loss": 0.0614, + "step": 16962 + }, + { + "epoch": 7.1, + "learning_rate": 6.515960759241791e-07, + "loss": 0.0398, + "step": 16963 + }, + { + "epoch": 7.1, + "learning_rate": 6.509942244701717e-07, + "loss": 0.0477, + "step": 16964 + }, + { + "epoch": 7.1, + "learning_rate": 6.50392641745039e-07, + "loss": 0.0532, + "step": 16965 + }, + { + "epoch": 7.1, + "learning_rate": 6.497913277660716e-07, + "loss": 0.0637, + "step": 16966 + }, + { + "epoch": 7.1, + "learning_rate": 6.491902825505548e-07, + "loss": 0.0484, + "step": 16967 + }, + { + "epoch": 7.1, + "learning_rate": 6.485895061157665e-07, + "loss": 0.0621, + "step": 16968 + }, + { + "epoch": 7.1, + "learning_rate": 6.479889984789733e-07, + "loss": 0.0641, + "step": 16969 + }, + { + "epoch": 7.1, + "learning_rate": 6.47388759657438e-07, + "loss": 0.0537, + "step": 16970 + }, + { + "epoch": 7.11, + "learning_rate": 6.467887896684144e-07, + "loss": 0.0448, + "step": 16971 + }, + { + "epoch": 7.11, + "learning_rate": 6.461890885291477e-07, + "loss": 0.0457, + "step": 16972 + }, + { + "epoch": 7.11, + "learning_rate": 6.455896562568731e-07, + "loss": 0.0567, + "step": 16973 + }, + { + "epoch": 7.11, + "learning_rate": 6.449904928688244e-07, + "loss": 0.0442, + "step": 16974 + }, + { + "epoch": 7.11, + "learning_rate": 6.443915983822213e-07, + "loss": 0.056, + "step": 16975 + }, + { + "epoch": 7.11, + "learning_rate": 6.4379297281428e-07, + "loss": 0.0709, + "step": 16976 + }, + { + "epoch": 7.11, + "learning_rate": 6.431946161822078e-07, + "loss": 0.0519, + "step": 16977 + }, + { + "epoch": 7.11, + "learning_rate": 6.425965285032043e-07, + "loss": 0.0594, + "step": 16978 + }, + { + "epoch": 7.11, + "learning_rate": 6.41998709794458e-07, + "loss": 0.0358, + "step": 16979 + }, + { + "epoch": 7.11, + "learning_rate": 6.414011600731552e-07, + "loss": 0.0299, + "step": 16980 + }, + { + "epoch": 7.11, + "learning_rate": 6.408038793564708e-07, + "loss": 0.0414, + "step": 16981 + }, + { + "epoch": 7.11, + "learning_rate": 6.402068676615747e-07, + "loss": 0.0359, + "step": 16982 + }, + { + "epoch": 7.11, + "learning_rate": 6.396101250056252e-07, + "loss": 0.0433, + "step": 16983 + }, + { + "epoch": 7.11, + "learning_rate": 6.390136514057788e-07, + "loss": 0.0505, + "step": 16984 + }, + { + "epoch": 7.11, + "learning_rate": 6.384174468791771e-07, + "loss": 0.053, + "step": 16985 + }, + { + "epoch": 7.11, + "learning_rate": 6.378215114429587e-07, + "loss": 0.0392, + "step": 16986 + }, + { + "epoch": 7.11, + "learning_rate": 6.372258451142533e-07, + "loss": 0.0524, + "step": 16987 + }, + { + "epoch": 7.11, + "learning_rate": 6.366304479101815e-07, + "loss": 0.0684, + "step": 16988 + }, + { + "epoch": 7.11, + "learning_rate": 6.360353198478586e-07, + "loss": 0.0296, + "step": 16989 + }, + { + "epoch": 7.11, + "learning_rate": 6.354404609443909e-07, + "loss": 0.042, + "step": 16990 + }, + { + "epoch": 7.11, + "learning_rate": 6.348458712168771e-07, + "loss": 0.0531, + "step": 16991 + }, + { + "epoch": 7.11, + "learning_rate": 6.342515506824099e-07, + "loss": 0.0372, + "step": 16992 + }, + { + "epoch": 7.11, + "learning_rate": 6.336574993580702e-07, + "loss": 0.0533, + "step": 16993 + }, + { + "epoch": 7.11, + "learning_rate": 6.330637172609333e-07, + "loss": 0.0447, + "step": 16994 + }, + { + "epoch": 7.12, + "learning_rate": 6.324702044080678e-07, + "loss": 0.052, + "step": 16995 + }, + { + "epoch": 7.12, + "learning_rate": 6.318769608165332e-07, + "loss": 0.0692, + "step": 16996 + }, + { + "epoch": 7.12, + "learning_rate": 6.312839865033826e-07, + "loss": 0.0409, + "step": 16997 + }, + { + "epoch": 7.12, + "learning_rate": 6.306912814856625e-07, + "loss": 0.0613, + "step": 16998 + }, + { + "epoch": 7.12, + "learning_rate": 6.300988457804047e-07, + "loss": 0.0461, + "step": 16999 + }, + { + "epoch": 7.12, + "learning_rate": 6.295066794046423e-07, + "loss": 0.05, + "step": 17000 + }, + { + "epoch": 7.12, + "learning_rate": 6.289147823753939e-07, + "loss": 0.0657, + "step": 17001 + }, + { + "epoch": 7.12, + "learning_rate": 6.283231547096746e-07, + "loss": 0.0571, + "step": 17002 + }, + { + "epoch": 7.12, + "learning_rate": 6.2773179642449e-07, + "loss": 0.0354, + "step": 17003 + }, + { + "epoch": 7.12, + "learning_rate": 6.271407075368385e-07, + "loss": 0.0361, + "step": 17004 + }, + { + "epoch": 7.12, + "learning_rate": 6.26549888063711e-07, + "loss": 0.0458, + "step": 17005 + }, + { + "epoch": 7.12, + "learning_rate": 6.259593380220875e-07, + "loss": 0.0539, + "step": 17006 + }, + { + "epoch": 7.12, + "learning_rate": 6.253690574289451e-07, + "loss": 0.0529, + "step": 17007 + }, + { + "epoch": 7.12, + "learning_rate": 6.247790463012493e-07, + "loss": 0.0425, + "step": 17008 + }, + { + "epoch": 7.12, + "learning_rate": 6.241893046559633e-07, + "loss": 0.0618, + "step": 17009 + }, + { + "epoch": 7.12, + "learning_rate": 6.235998325100334e-07, + "loss": 0.0516, + "step": 17010 + }, + { + "epoch": 7.12, + "learning_rate": 6.23010629880405e-07, + "loss": 0.0726, + "step": 17011 + }, + { + "epoch": 7.12, + "learning_rate": 6.224216967840169e-07, + "loss": 0.0473, + "step": 17012 + }, + { + "epoch": 7.12, + "learning_rate": 6.218330332377953e-07, + "loss": 0.0543, + "step": 17013 + }, + { + "epoch": 7.12, + "learning_rate": 6.212446392586613e-07, + "loss": 0.0578, + "step": 17014 + }, + { + "epoch": 7.12, + "learning_rate": 6.206565148635269e-07, + "loss": 0.0423, + "step": 17015 + }, + { + "epoch": 7.12, + "learning_rate": 6.200686600692974e-07, + "loss": 0.0524, + "step": 17016 + }, + { + "epoch": 7.12, + "learning_rate": 6.194810748928714e-07, + "loss": 0.041, + "step": 17017 + }, + { + "epoch": 7.12, + "learning_rate": 6.188937593511368e-07, + "loss": 0.0781, + "step": 17018 + }, + { + "epoch": 7.13, + "learning_rate": 6.183067134609788e-07, + "loss": 0.0519, + "step": 17019 + }, + { + "epoch": 7.13, + "learning_rate": 6.177199372392673e-07, + "loss": 0.0491, + "step": 17020 + }, + { + "epoch": 7.13, + "learning_rate": 6.171334307028709e-07, + "loss": 0.0446, + "step": 17021 + }, + { + "epoch": 7.13, + "learning_rate": 6.165471938686474e-07, + "loss": 0.0611, + "step": 17022 + }, + { + "epoch": 7.13, + "learning_rate": 6.159612267534487e-07, + "loss": 0.0382, + "step": 17023 + }, + { + "epoch": 7.13, + "learning_rate": 6.15375529374116e-07, + "loss": 0.0452, + "step": 17024 + }, + { + "epoch": 7.13, + "learning_rate": 6.147901017474889e-07, + "loss": 0.0458, + "step": 17025 + }, + { + "epoch": 7.13, + "learning_rate": 6.142049438903896e-07, + "loss": 0.0655, + "step": 17026 + }, + { + "epoch": 7.13, + "learning_rate": 6.136200558196404e-07, + "loss": 0.0524, + "step": 17027 + }, + { + "epoch": 7.13, + "learning_rate": 6.130354375520542e-07, + "loss": 0.064, + "step": 17028 + }, + { + "epoch": 7.13, + "learning_rate": 6.124510891044333e-07, + "loss": 0.0449, + "step": 17029 + }, + { + "epoch": 7.13, + "learning_rate": 6.118670104935765e-07, + "loss": 0.0404, + "step": 17030 + }, + { + "epoch": 7.13, + "learning_rate": 6.112832017362713e-07, + "loss": 0.0543, + "step": 17031 + }, + { + "epoch": 7.13, + "learning_rate": 6.106996628493e-07, + "loss": 0.0434, + "step": 17032 + }, + { + "epoch": 7.13, + "learning_rate": 6.101163938494359e-07, + "loss": 0.0572, + "step": 17033 + }, + { + "epoch": 7.13, + "learning_rate": 6.095333947534421e-07, + "loss": 0.0562, + "step": 17034 + }, + { + "epoch": 7.13, + "learning_rate": 6.089506655780774e-07, + "loss": 0.0447, + "step": 17035 + }, + { + "epoch": 7.13, + "learning_rate": 6.083682063400919e-07, + "loss": 0.0567, + "step": 17036 + }, + { + "epoch": 7.13, + "learning_rate": 6.077860170562288e-07, + "loss": 0.0344, + "step": 17037 + }, + { + "epoch": 7.13, + "learning_rate": 6.072040977432225e-07, + "loss": 0.0537, + "step": 17038 + }, + { + "epoch": 7.13, + "learning_rate": 6.066224484178007e-07, + "loss": 0.0579, + "step": 17039 + }, + { + "epoch": 7.13, + "learning_rate": 6.0604106909668e-07, + "loss": 0.0518, + "step": 17040 + }, + { + "epoch": 7.13, + "learning_rate": 6.054599597965716e-07, + "loss": 0.0407, + "step": 17041 + }, + { + "epoch": 7.14, + "learning_rate": 6.048791205341809e-07, + "loss": 0.0791, + "step": 17042 + }, + { + "epoch": 7.14, + "learning_rate": 6.042985513262023e-07, + "loss": 0.0451, + "step": 17043 + }, + { + "epoch": 7.14, + "learning_rate": 6.037182521893247e-07, + "loss": 0.0292, + "step": 17044 + }, + { + "epoch": 7.14, + "learning_rate": 6.031382231402283e-07, + "loss": 0.0701, + "step": 17045 + }, + { + "epoch": 7.14, + "learning_rate": 6.025584641955862e-07, + "loss": 0.0321, + "step": 17046 + }, + { + "epoch": 7.14, + "learning_rate": 6.019789753720595e-07, + "loss": 0.0325, + "step": 17047 + }, + { + "epoch": 7.14, + "learning_rate": 6.013997566863084e-07, + "loss": 0.0744, + "step": 17048 + }, + { + "epoch": 7.14, + "learning_rate": 6.008208081549805e-07, + "loss": 0.0507, + "step": 17049 + }, + { + "epoch": 7.14, + "learning_rate": 6.002421297947181e-07, + "loss": 0.048, + "step": 17050 + }, + { + "epoch": 7.14, + "learning_rate": 5.996637216221545e-07, + "loss": 0.0519, + "step": 17051 + }, + { + "epoch": 7.14, + "learning_rate": 5.990855836539178e-07, + "loss": 0.0617, + "step": 17052 + }, + { + "epoch": 7.14, + "learning_rate": 5.985077159066211e-07, + "loss": 0.047, + "step": 17053 + }, + { + "epoch": 7.14, + "learning_rate": 5.979301183968789e-07, + "loss": 0.0532, + "step": 17054 + }, + { + "epoch": 7.14, + "learning_rate": 5.973527911412913e-07, + "loss": 0.0493, + "step": 17055 + }, + { + "epoch": 7.14, + "learning_rate": 5.96775734156454e-07, + "loss": 0.0379, + "step": 17056 + }, + { + "epoch": 7.14, + "learning_rate": 5.961989474589536e-07, + "loss": 0.0469, + "step": 17057 + }, + { + "epoch": 7.14, + "learning_rate": 5.956224310653702e-07, + "loss": 0.0569, + "step": 17058 + }, + { + "epoch": 7.14, + "learning_rate": 5.95046184992274e-07, + "loss": 0.0599, + "step": 17059 + }, + { + "epoch": 7.14, + "learning_rate": 5.944702092562315e-07, + "loss": 0.0446, + "step": 17060 + }, + { + "epoch": 7.14, + "learning_rate": 5.938945038737942e-07, + "loss": 0.043, + "step": 17061 + }, + { + "epoch": 7.14, + "learning_rate": 5.933190688615131e-07, + "loss": 0.035, + "step": 17062 + }, + { + "epoch": 7.14, + "learning_rate": 5.927439042359284e-07, + "loss": 0.0371, + "step": 17063 + }, + { + "epoch": 7.14, + "learning_rate": 5.921690100135713e-07, + "loss": 0.0493, + "step": 17064 + }, + { + "epoch": 7.14, + "learning_rate": 5.915943862109685e-07, + "loss": 0.0422, + "step": 17065 + }, + { + "epoch": 7.15, + "learning_rate": 5.91020032844638e-07, + "loss": 0.0691, + "step": 17066 + }, + { + "epoch": 7.15, + "learning_rate": 5.904459499310844e-07, + "loss": 0.0396, + "step": 17067 + }, + { + "epoch": 7.15, + "learning_rate": 5.898721374868132e-07, + "loss": 0.0283, + "step": 17068 + }, + { + "epoch": 7.15, + "learning_rate": 5.89298595528317e-07, + "loss": 0.0624, + "step": 17069 + }, + { + "epoch": 7.15, + "learning_rate": 5.887253240720814e-07, + "loss": 0.0461, + "step": 17070 + }, + { + "epoch": 7.15, + "learning_rate": 5.881523231345854e-07, + "loss": 0.0477, + "step": 17071 + }, + { + "epoch": 7.15, + "learning_rate": 5.875795927322981e-07, + "loss": 0.0541, + "step": 17072 + }, + { + "epoch": 7.15, + "learning_rate": 5.87007132881685e-07, + "loss": 0.052, + "step": 17073 + }, + { + "epoch": 7.15, + "learning_rate": 5.864349435991978e-07, + "loss": 0.0411, + "step": 17074 + }, + { + "epoch": 7.15, + "learning_rate": 5.858630249012842e-07, + "loss": 0.0537, + "step": 17075 + }, + { + "epoch": 7.15, + "learning_rate": 5.852913768043833e-07, + "loss": 0.057, + "step": 17076 + }, + { + "epoch": 7.15, + "learning_rate": 5.847199993249297e-07, + "loss": 0.0607, + "step": 17077 + }, + { + "epoch": 7.15, + "learning_rate": 5.841488924793437e-07, + "loss": 0.0495, + "step": 17078 + }, + { + "epoch": 7.15, + "learning_rate": 5.83578056284041e-07, + "loss": 0.0314, + "step": 17079 + }, + { + "epoch": 7.15, + "learning_rate": 5.83007490755434e-07, + "loss": 0.051, + "step": 17080 + }, + { + "epoch": 7.15, + "learning_rate": 5.824371959099174e-07, + "loss": 0.0399, + "step": 17081 + }, + { + "epoch": 7.15, + "learning_rate": 5.81867171763888e-07, + "loss": 0.0408, + "step": 17082 + }, + { + "epoch": 7.15, + "learning_rate": 5.812974183337283e-07, + "loss": 0.0347, + "step": 17083 + }, + { + "epoch": 7.15, + "learning_rate": 5.807279356358164e-07, + "loss": 0.0461, + "step": 17084 + }, + { + "epoch": 7.15, + "learning_rate": 5.801587236865214e-07, + "loss": 0.0652, + "step": 17085 + }, + { + "epoch": 7.15, + "learning_rate": 5.795897825022046e-07, + "loss": 0.0383, + "step": 17086 + }, + { + "epoch": 7.15, + "learning_rate": 5.790211120992218e-07, + "loss": 0.0328, + "step": 17087 + }, + { + "epoch": 7.15, + "learning_rate": 5.784527124939154e-07, + "loss": 0.0486, + "step": 17088 + }, + { + "epoch": 7.15, + "learning_rate": 5.778845837026259e-07, + "loss": 0.0435, + "step": 17089 + }, + { + "epoch": 7.16, + "learning_rate": 5.773167257416823e-07, + "loss": 0.0477, + "step": 17090 + }, + { + "epoch": 7.16, + "learning_rate": 5.767491386274082e-07, + "loss": 0.0498, + "step": 17091 + }, + { + "epoch": 7.16, + "learning_rate": 5.761818223761184e-07, + "loss": 0.0428, + "step": 17092 + }, + { + "epoch": 7.16, + "learning_rate": 5.756147770041209e-07, + "loss": 0.0436, + "step": 17093 + }, + { + "epoch": 7.16, + "learning_rate": 5.750480025277116e-07, + "loss": 0.0549, + "step": 17094 + }, + { + "epoch": 7.16, + "learning_rate": 5.744814989631842e-07, + "loss": 0.033, + "step": 17095 + }, + { + "epoch": 7.16, + "learning_rate": 5.739152663268222e-07, + "loss": 0.0581, + "step": 17096 + }, + { + "epoch": 7.16, + "learning_rate": 5.733493046349015e-07, + "loss": 0.0454, + "step": 17097 + }, + { + "epoch": 7.16, + "learning_rate": 5.727836139036902e-07, + "loss": 0.0521, + "step": 17098 + }, + { + "epoch": 7.16, + "learning_rate": 5.722181941494476e-07, + "loss": 0.0507, + "step": 17099 + }, + { + "epoch": 7.16, + "learning_rate": 5.716530453884283e-07, + "loss": 0.0441, + "step": 17100 + }, + { + "epoch": 7.16, + "learning_rate": 5.710881676368762e-07, + "loss": 0.0264, + "step": 17101 + }, + { + "epoch": 7.16, + "learning_rate": 5.70523560911026e-07, + "loss": 0.0546, + "step": 17102 + }, + { + "epoch": 7.16, + "learning_rate": 5.69959225227108e-07, + "loss": 0.0417, + "step": 17103 + }, + { + "epoch": 7.16, + "learning_rate": 5.693951606013448e-07, + "loss": 0.052, + "step": 17104 + }, + { + "epoch": 7.16, + "learning_rate": 5.68831367049949e-07, + "loss": 0.0548, + "step": 17105 + }, + { + "epoch": 7.16, + "learning_rate": 5.682678445891265e-07, + "loss": 0.058, + "step": 17106 + }, + { + "epoch": 7.16, + "learning_rate": 5.677045932350766e-07, + "loss": 0.064, + "step": 17107 + }, + { + "epoch": 7.16, + "learning_rate": 5.671416130039864e-07, + "loss": 0.0672, + "step": 17108 + }, + { + "epoch": 7.16, + "learning_rate": 5.665789039120395e-07, + "loss": 0.0542, + "step": 17109 + }, + { + "epoch": 7.16, + "learning_rate": 5.66016465975412e-07, + "loss": 0.0412, + "step": 17110 + }, + { + "epoch": 7.16, + "learning_rate": 5.654542992102685e-07, + "loss": 0.0584, + "step": 17111 + }, + { + "epoch": 7.16, + "learning_rate": 5.648924036327697e-07, + "loss": 0.0335, + "step": 17112 + }, + { + "epoch": 7.16, + "learning_rate": 5.643307792590669e-07, + "loss": 0.0471, + "step": 17113 + }, + { + "epoch": 7.17, + "learning_rate": 5.637694261053029e-07, + "loss": 0.0373, + "step": 17114 + }, + { + "epoch": 7.17, + "learning_rate": 5.632083441876123e-07, + "loss": 0.0523, + "step": 17115 + }, + { + "epoch": 7.17, + "learning_rate": 5.626475335221248e-07, + "loss": 0.0573, + "step": 17116 + }, + { + "epoch": 7.17, + "learning_rate": 5.620869941249585e-07, + "loss": 0.0413, + "step": 17117 + }, + { + "epoch": 7.17, + "learning_rate": 5.615267260122259e-07, + "loss": 0.0619, + "step": 17118 + }, + { + "epoch": 7.17, + "learning_rate": 5.609667292000331e-07, + "loss": 0.0465, + "step": 17119 + }, + { + "epoch": 7.17, + "learning_rate": 5.604070037044762e-07, + "loss": 0.0475, + "step": 17120 + }, + { + "epoch": 7.17, + "learning_rate": 5.598475495416422e-07, + "loss": 0.0465, + "step": 17121 + }, + { + "epoch": 7.17, + "learning_rate": 5.59288366727615e-07, + "loss": 0.0392, + "step": 17122 + }, + { + "epoch": 7.17, + "learning_rate": 5.58729455278465e-07, + "loss": 0.0409, + "step": 17123 + }, + { + "epoch": 7.17, + "learning_rate": 5.581708152102583e-07, + "loss": 0.0346, + "step": 17124 + }, + { + "epoch": 7.17, + "learning_rate": 5.576124465390532e-07, + "loss": 0.0457, + "step": 17125 + }, + { + "epoch": 7.17, + "learning_rate": 5.570543492808988e-07, + "loss": 0.0469, + "step": 17126 + }, + { + "epoch": 7.17, + "learning_rate": 5.564965234518383e-07, + "loss": 0.06, + "step": 17127 + }, + { + "epoch": 7.17, + "learning_rate": 5.559389690679063e-07, + "loss": 0.068, + "step": 17128 + }, + { + "epoch": 7.17, + "learning_rate": 5.553816861451267e-07, + "loss": 0.0508, + "step": 17129 + }, + { + "epoch": 7.17, + "learning_rate": 5.5482467469952e-07, + "loss": 0.0634, + "step": 17130 + }, + { + "epoch": 7.17, + "learning_rate": 5.542679347470958e-07, + "loss": 0.0611, + "step": 17131 + }, + { + "epoch": 7.17, + "learning_rate": 5.537114663038579e-07, + "loss": 0.0391, + "step": 17132 + }, + { + "epoch": 7.17, + "learning_rate": 5.531552693858022e-07, + "loss": 0.0572, + "step": 17133 + }, + { + "epoch": 7.17, + "learning_rate": 5.525993440089161e-07, + "loss": 0.0517, + "step": 17134 + }, + { + "epoch": 7.17, + "learning_rate": 5.520436901891768e-07, + "loss": 0.0466, + "step": 17135 + }, + { + "epoch": 7.17, + "learning_rate": 5.514883079425581e-07, + "loss": 0.0506, + "step": 17136 + }, + { + "epoch": 7.17, + "learning_rate": 5.509331972850229e-07, + "loss": 0.0642, + "step": 17137 + }, + { + "epoch": 7.18, + "learning_rate": 5.503783582325283e-07, + "loss": 0.0459, + "step": 17138 + }, + { + "epoch": 7.18, + "learning_rate": 5.498237908010229e-07, + "loss": 0.0536, + "step": 17139 + }, + { + "epoch": 7.18, + "learning_rate": 5.492694950064459e-07, + "loss": 0.0532, + "step": 17140 + }, + { + "epoch": 7.18, + "learning_rate": 5.487154708647324e-07, + "loss": 0.0545, + "step": 17141 + }, + { + "epoch": 7.18, + "learning_rate": 5.481617183918053e-07, + "loss": 0.049, + "step": 17142 + }, + { + "epoch": 7.18, + "learning_rate": 5.476082376035807e-07, + "loss": 0.0629, + "step": 17143 + }, + { + "epoch": 7.18, + "learning_rate": 5.470550285159704e-07, + "loss": 0.0462, + "step": 17144 + }, + { + "epoch": 7.18, + "learning_rate": 5.465020911448759e-07, + "loss": 0.047, + "step": 17145 + }, + { + "epoch": 7.18, + "learning_rate": 5.459494255061892e-07, + "loss": 0.0412, + "step": 17146 + }, + { + "epoch": 7.18, + "learning_rate": 5.453970316157964e-07, + "loss": 0.0722, + "step": 17147 + }, + { + "epoch": 7.18, + "learning_rate": 5.448449094895769e-07, + "loss": 0.0416, + "step": 17148 + }, + { + "epoch": 7.18, + "learning_rate": 5.442930591433992e-07, + "loss": 0.0392, + "step": 17149 + }, + { + "epoch": 7.18, + "learning_rate": 5.437414805931274e-07, + "loss": 0.0586, + "step": 17150 + }, + { + "epoch": 7.18, + "learning_rate": 5.431901738546142e-07, + "loss": 0.0482, + "step": 17151 + }, + { + "epoch": 7.18, + "learning_rate": 5.42639138943708e-07, + "loss": 0.0548, + "step": 17152 + }, + { + "epoch": 7.18, + "learning_rate": 5.420883758762474e-07, + "loss": 0.0484, + "step": 17153 + }, + { + "epoch": 7.18, + "learning_rate": 5.415378846680641e-07, + "loss": 0.0542, + "step": 17154 + }, + { + "epoch": 7.18, + "learning_rate": 5.409876653349822e-07, + "loss": 0.0526, + "step": 17155 + }, + { + "epoch": 7.18, + "learning_rate": 5.404377178928144e-07, + "loss": 0.047, + "step": 17156 + }, + { + "epoch": 7.18, + "learning_rate": 5.398880423573704e-07, + "loss": 0.0596, + "step": 17157 + }, + { + "epoch": 7.18, + "learning_rate": 5.393386387444499e-07, + "loss": 0.0485, + "step": 17158 + }, + { + "epoch": 7.18, + "learning_rate": 5.387895070698457e-07, + "loss": 0.0361, + "step": 17159 + }, + { + "epoch": 7.18, + "learning_rate": 5.382406473493395e-07, + "loss": 0.0459, + "step": 17160 + }, + { + "epoch": 7.18, + "learning_rate": 5.376920595987123e-07, + "loss": 0.0495, + "step": 17161 + }, + { + "epoch": 7.19, + "learning_rate": 5.37143743833729e-07, + "loss": 0.0529, + "step": 17162 + }, + { + "epoch": 7.19, + "learning_rate": 5.365957000701516e-07, + "loss": 0.0338, + "step": 17163 + }, + { + "epoch": 7.19, + "learning_rate": 5.360479283237319e-07, + "loss": 0.0757, + "step": 17164 + }, + { + "epoch": 7.19, + "learning_rate": 5.355004286102172e-07, + "loss": 0.0415, + "step": 17165 + }, + { + "epoch": 7.19, + "learning_rate": 5.34953200945344e-07, + "loss": 0.0604, + "step": 17166 + }, + { + "epoch": 7.19, + "learning_rate": 5.34406245344844e-07, + "loss": 0.0416, + "step": 17167 + }, + { + "epoch": 7.19, + "learning_rate": 5.338595618244347e-07, + "loss": 0.049, + "step": 17168 + }, + { + "epoch": 7.19, + "learning_rate": 5.333131503998334e-07, + "loss": 0.059, + "step": 17169 + }, + { + "epoch": 7.19, + "learning_rate": 5.327670110867444e-07, + "loss": 0.0566, + "step": 17170 + }, + { + "epoch": 7.19, + "learning_rate": 5.322211439008651e-07, + "loss": 0.0499, + "step": 17171 + }, + { + "epoch": 7.19, + "learning_rate": 5.316755488578884e-07, + "loss": 0.0497, + "step": 17172 + }, + { + "epoch": 7.19, + "learning_rate": 5.311302259734963e-07, + "loss": 0.0619, + "step": 17173 + }, + { + "epoch": 7.19, + "learning_rate": 5.305851752633628e-07, + "loss": 0.0443, + "step": 17174 + }, + { + "epoch": 7.19, + "learning_rate": 5.300403967431578e-07, + "loss": 0.0595, + "step": 17175 + }, + { + "epoch": 7.19, + "learning_rate": 5.294958904285363e-07, + "loss": 0.0332, + "step": 17176 + }, + { + "epoch": 7.19, + "learning_rate": 5.289516563351504e-07, + "loss": 0.0602, + "step": 17177 + }, + { + "epoch": 7.19, + "learning_rate": 5.284076944786454e-07, + "loss": 0.0471, + "step": 17178 + }, + { + "epoch": 7.19, + "learning_rate": 5.278640048746564e-07, + "loss": 0.0685, + "step": 17179 + }, + { + "epoch": 7.19, + "learning_rate": 5.273205875388109e-07, + "loss": 0.0361, + "step": 17180 + }, + { + "epoch": 7.19, + "learning_rate": 5.267774424867312e-07, + "loss": 0.0628, + "step": 17181 + }, + { + "epoch": 7.19, + "learning_rate": 5.262345697340254e-07, + "loss": 0.036, + "step": 17182 + }, + { + "epoch": 7.19, + "learning_rate": 5.256919692963003e-07, + "loss": 0.0546, + "step": 17183 + }, + { + "epoch": 7.19, + "learning_rate": 5.251496411891521e-07, + "loss": 0.0416, + "step": 17184 + }, + { + "epoch": 7.19, + "learning_rate": 5.246075854281696e-07, + "loss": 0.0642, + "step": 17185 + }, + { + "epoch": 7.2, + "learning_rate": 5.240658020289347e-07, + "loss": 0.0384, + "step": 17186 + }, + { + "epoch": 7.2, + "learning_rate": 5.235242910070182e-07, + "loss": 0.0674, + "step": 17187 + }, + { + "epoch": 7.2, + "learning_rate": 5.229830523779888e-07, + "loss": 0.0377, + "step": 17188 + }, + { + "epoch": 7.2, + "learning_rate": 5.224420861573997e-07, + "loss": 0.0552, + "step": 17189 + }, + { + "epoch": 7.2, + "learning_rate": 5.219013923608051e-07, + "loss": 0.0486, + "step": 17190 + }, + { + "epoch": 7.2, + "learning_rate": 5.213609710037415e-07, + "loss": 0.0612, + "step": 17191 + }, + { + "epoch": 7.2, + "learning_rate": 5.208208221017463e-07, + "loss": 0.0494, + "step": 17192 + }, + { + "epoch": 7.2, + "learning_rate": 5.202809456703451e-07, + "loss": 0.0492, + "step": 17193 + }, + { + "epoch": 7.2, + "learning_rate": 5.197413417250563e-07, + "loss": 0.0579, + "step": 17194 + }, + { + "epoch": 7.2, + "learning_rate": 5.192020102813888e-07, + "loss": 0.0604, + "step": 17195 + }, + { + "epoch": 7.2, + "learning_rate": 5.186629513548491e-07, + "loss": 0.0378, + "step": 17196 + }, + { + "epoch": 7.2, + "learning_rate": 5.18124164960927e-07, + "loss": 0.039, + "step": 17197 + }, + { + "epoch": 7.2, + "learning_rate": 5.175856511151123e-07, + "loss": 0.0578, + "step": 17198 + }, + { + "epoch": 7.2, + "learning_rate": 5.170474098328826e-07, + "loss": 0.0604, + "step": 17199 + }, + { + "epoch": 7.2, + "learning_rate": 5.165094411297111e-07, + "loss": 0.0532, + "step": 17200 + }, + { + "epoch": 7.2, + "learning_rate": 5.1597174502106e-07, + "loss": 0.0469, + "step": 17201 + }, + { + "epoch": 7.2, + "learning_rate": 5.154343215223856e-07, + "loss": 0.0451, + "step": 17202 + }, + { + "epoch": 7.2, + "learning_rate": 5.148971706491345e-07, + "loss": 0.0635, + "step": 17203 + }, + { + "epoch": 7.2, + "learning_rate": 5.143602924167479e-07, + "loss": 0.0566, + "step": 17204 + }, + { + "epoch": 7.2, + "learning_rate": 5.138236868406565e-07, + "loss": 0.0526, + "step": 17205 + }, + { + "epoch": 7.2, + "learning_rate": 5.13287353936286e-07, + "loss": 0.0509, + "step": 17206 + }, + { + "epoch": 7.2, + "learning_rate": 5.127512937190515e-07, + "loss": 0.0469, + "step": 17207 + }, + { + "epoch": 7.2, + "learning_rate": 5.122155062043632e-07, + "loss": 0.058, + "step": 17208 + }, + { + "epoch": 7.2, + "learning_rate": 5.116799914076198e-07, + "loss": 0.0455, + "step": 17209 + }, + { + "epoch": 7.21, + "learning_rate": 5.111447493442156e-07, + "loss": 0.072, + "step": 17210 + }, + { + "epoch": 7.21, + "learning_rate": 5.106097800295351e-07, + "loss": 0.0577, + "step": 17211 + }, + { + "epoch": 7.21, + "learning_rate": 5.100750834789559e-07, + "loss": 0.0573, + "step": 17212 + }, + { + "epoch": 7.21, + "learning_rate": 5.09540659707849e-07, + "loss": 0.047, + "step": 17213 + }, + { + "epoch": 7.21, + "learning_rate": 5.090065087315721e-07, + "loss": 0.0399, + "step": 17214 + }, + { + "epoch": 7.21, + "learning_rate": 5.084726305654819e-07, + "loss": 0.0516, + "step": 17215 + }, + { + "epoch": 7.21, + "learning_rate": 5.079390252249239e-07, + "loss": 0.0476, + "step": 17216 + }, + { + "epoch": 7.21, + "learning_rate": 5.074056927252335e-07, + "loss": 0.0331, + "step": 17217 + }, + { + "epoch": 7.21, + "learning_rate": 5.068726330817442e-07, + "loss": 0.0627, + "step": 17218 + }, + { + "epoch": 7.21, + "learning_rate": 5.063398463097758e-07, + "loss": 0.049, + "step": 17219 + }, + { + "epoch": 7.21, + "learning_rate": 5.058073324246448e-07, + "loss": 0.0441, + "step": 17220 + }, + { + "epoch": 7.21, + "learning_rate": 5.052750914416571e-07, + "loss": 0.0365, + "step": 17221 + }, + { + "epoch": 7.21, + "learning_rate": 5.047431233761124e-07, + "loss": 0.0519, + "step": 17222 + }, + { + "epoch": 7.21, + "learning_rate": 5.042114282432998e-07, + "loss": 0.0562, + "step": 17223 + }, + { + "epoch": 7.21, + "learning_rate": 5.036800060585023e-07, + "loss": 0.0381, + "step": 17224 + }, + { + "epoch": 7.21, + "learning_rate": 5.031488568369969e-07, + "loss": 0.0396, + "step": 17225 + }, + { + "epoch": 7.21, + "learning_rate": 5.026179805940501e-07, + "loss": 0.0411, + "step": 17226 + }, + { + "epoch": 7.21, + "learning_rate": 5.02087377344922e-07, + "loss": 0.0682, + "step": 17227 + }, + { + "epoch": 7.21, + "learning_rate": 5.015570471048637e-07, + "loss": 0.0456, + "step": 17228 + }, + { + "epoch": 7.21, + "learning_rate": 5.010269898891196e-07, + "loss": 0.0765, + "step": 17229 + }, + { + "epoch": 7.21, + "learning_rate": 5.004972057129253e-07, + "loss": 0.064, + "step": 17230 + }, + { + "epoch": 7.21, + "learning_rate": 4.999676945915088e-07, + "loss": 0.048, + "step": 17231 + }, + { + "epoch": 7.21, + "learning_rate": 4.994384565400912e-07, + "loss": 0.0618, + "step": 17232 + }, + { + "epoch": 7.21, + "learning_rate": 4.989094915738835e-07, + "loss": 0.0445, + "step": 17233 + }, + { + "epoch": 7.22, + "learning_rate": 4.983807997080925e-07, + "loss": 0.0666, + "step": 17234 + }, + { + "epoch": 7.22, + "learning_rate": 4.978523809579139e-07, + "loss": 0.06, + "step": 17235 + }, + { + "epoch": 7.22, + "learning_rate": 4.973242353385365e-07, + "loss": 0.052, + "step": 17236 + }, + { + "epoch": 7.22, + "learning_rate": 4.967963628651418e-07, + "loss": 0.0713, + "step": 17237 + }, + { + "epoch": 7.22, + "learning_rate": 4.962687635529018e-07, + "loss": 0.0548, + "step": 17238 + }, + { + "epoch": 7.22, + "learning_rate": 4.957414374169834e-07, + "loss": 0.0518, + "step": 17239 + }, + { + "epoch": 7.22, + "learning_rate": 4.952143844725421e-07, + "loss": 0.0585, + "step": 17240 + }, + { + "epoch": 7.22, + "learning_rate": 4.946876047347293e-07, + "loss": 0.0657, + "step": 17241 + }, + { + "epoch": 7.22, + "learning_rate": 4.941610982186873e-07, + "loss": 0.0384, + "step": 17242 + }, + { + "epoch": 7.22, + "learning_rate": 4.936348649395495e-07, + "loss": 0.0612, + "step": 17243 + }, + { + "epoch": 7.22, + "learning_rate": 4.931089049124405e-07, + "loss": 0.047, + "step": 17244 + }, + { + "epoch": 7.22, + "learning_rate": 4.925832181524803e-07, + "loss": 0.0511, + "step": 17245 + }, + { + "epoch": 7.22, + "learning_rate": 4.920578046747782e-07, + "loss": 0.0712, + "step": 17246 + }, + { + "epoch": 7.22, + "learning_rate": 4.915326644944374e-07, + "loss": 0.0619, + "step": 17247 + }, + { + "epoch": 7.22, + "learning_rate": 4.910077976265526e-07, + "loss": 0.0476, + "step": 17248 + }, + { + "epoch": 7.22, + "learning_rate": 4.904832040862129e-07, + "loss": 0.0398, + "step": 17249 + }, + { + "epoch": 7.22, + "learning_rate": 4.899588838884917e-07, + "loss": 0.0612, + "step": 17250 + }, + { + "epoch": 7.22, + "learning_rate": 4.894348370484648e-07, + "loss": 0.0294, + "step": 17251 + }, + { + "epoch": 7.22, + "learning_rate": 4.889110635811933e-07, + "loss": 0.0357, + "step": 17252 + }, + { + "epoch": 7.22, + "learning_rate": 4.883875635017332e-07, + "loss": 0.0531, + "step": 17253 + }, + { + "epoch": 7.22, + "learning_rate": 4.878643368251323e-07, + "loss": 0.0412, + "step": 17254 + }, + { + "epoch": 7.22, + "learning_rate": 4.873413835664298e-07, + "loss": 0.035, + "step": 17255 + }, + { + "epoch": 7.22, + "learning_rate": 4.868187037406602e-07, + "loss": 0.0401, + "step": 17256 + }, + { + "epoch": 7.23, + "learning_rate": 4.862962973628427e-07, + "loss": 0.0583, + "step": 17257 + }, + { + "epoch": 7.23, + "learning_rate": 4.857741644479974e-07, + "loss": 0.0446, + "step": 17258 + }, + { + "epoch": 7.23, + "learning_rate": 4.852523050111301e-07, + "loss": 0.0301, + "step": 17259 + }, + { + "epoch": 7.23, + "learning_rate": 4.847307190672424e-07, + "loss": 0.0308, + "step": 17260 + }, + { + "epoch": 7.23, + "learning_rate": 4.842094066313252e-07, + "loss": 0.0493, + "step": 17261 + }, + { + "epoch": 7.23, + "learning_rate": 4.836883677183656e-07, + "loss": 0.0656, + "step": 17262 + }, + { + "epoch": 7.23, + "learning_rate": 4.831676023433396e-07, + "loss": 0.0511, + "step": 17263 + }, + { + "epoch": 7.23, + "learning_rate": 4.826471105212149e-07, + "loss": 0.0567, + "step": 17264 + }, + { + "epoch": 7.23, + "learning_rate": 4.821268922669542e-07, + "loss": 0.047, + "step": 17265 + }, + { + "epoch": 7.23, + "learning_rate": 4.816069475955087e-07, + "loss": 0.0684, + "step": 17266 + }, + { + "epoch": 7.23, + "learning_rate": 4.810872765218255e-07, + "loss": 0.0433, + "step": 17267 + }, + { + "epoch": 7.23, + "learning_rate": 4.805678790608415e-07, + "loss": 0.0373, + "step": 17268 + }, + { + "epoch": 7.23, + "learning_rate": 4.800487552274868e-07, + "loss": 0.0461, + "step": 17269 + }, + { + "epoch": 7.23, + "learning_rate": 4.795299050366841e-07, + "loss": 0.056, + "step": 17270 + }, + { + "epoch": 7.23, + "learning_rate": 4.790113285033437e-07, + "loss": 0.0434, + "step": 17271 + }, + { + "epoch": 7.23, + "learning_rate": 4.784930256423748e-07, + "loss": 0.0303, + "step": 17272 + }, + { + "epoch": 7.23, + "learning_rate": 4.779749964686752e-07, + "loss": 0.0588, + "step": 17273 + }, + { + "epoch": 7.23, + "learning_rate": 4.774572409971334e-07, + "loss": 0.0565, + "step": 17274 + }, + { + "epoch": 7.23, + "learning_rate": 4.769397592426339e-07, + "loss": 0.0705, + "step": 17275 + }, + { + "epoch": 7.23, + "learning_rate": 4.7642255122005155e-07, + "loss": 0.0575, + "step": 17276 + }, + { + "epoch": 7.23, + "learning_rate": 4.7590561694424995e-07, + "loss": 0.058, + "step": 17277 + }, + { + "epoch": 7.23, + "learning_rate": 4.7538895643009066e-07, + "loss": 0.0532, + "step": 17278 + }, + { + "epoch": 7.23, + "learning_rate": 4.7487256969242277e-07, + "loss": 0.0531, + "step": 17279 + }, + { + "epoch": 7.23, + "learning_rate": 4.743564567460912e-07, + "loss": 0.0315, + "step": 17280 + }, + { + "epoch": 7.24, + "learning_rate": 4.738406176059318e-07, + "loss": 0.0385, + "step": 17281 + }, + { + "epoch": 7.24, + "learning_rate": 4.7332505228676937e-07, + "loss": 0.0513, + "step": 17282 + }, + { + "epoch": 7.24, + "learning_rate": 4.7280976080342434e-07, + "loss": 0.0437, + "step": 17283 + }, + { + "epoch": 7.24, + "learning_rate": 4.722947431707092e-07, + "loss": 0.0482, + "step": 17284 + }, + { + "epoch": 7.24, + "learning_rate": 4.7177999940342554e-07, + "loss": 0.0481, + "step": 17285 + }, + { + "epoch": 7.24, + "learning_rate": 4.712655295163715e-07, + "loss": 0.0416, + "step": 17286 + }, + { + "epoch": 7.24, + "learning_rate": 4.70751333524333e-07, + "loss": 0.0451, + "step": 17287 + }, + { + "epoch": 7.24, + "learning_rate": 4.702374114420927e-07, + "loss": 0.0611, + "step": 17288 + }, + { + "epoch": 7.24, + "learning_rate": 4.6972376328442095e-07, + "loss": 0.0486, + "step": 17289 + }, + { + "epoch": 7.24, + "learning_rate": 4.6921038906608375e-07, + "loss": 0.0418, + "step": 17290 + }, + { + "epoch": 7.24, + "learning_rate": 4.6869728880183486e-07, + "loss": 0.0622, + "step": 17291 + }, + { + "epoch": 7.24, + "learning_rate": 4.681844625064247e-07, + "loss": 0.0411, + "step": 17292 + }, + { + "epoch": 7.24, + "learning_rate": 4.676719101945937e-07, + "loss": 0.0574, + "step": 17293 + }, + { + "epoch": 7.24, + "learning_rate": 4.671596318810745e-07, + "loss": 0.0523, + "step": 17294 + }, + { + "epoch": 7.24, + "learning_rate": 4.666476275805931e-07, + "loss": 0.0486, + "step": 17295 + }, + { + "epoch": 7.24, + "learning_rate": 4.6613589730786557e-07, + "loss": 0.0604, + "step": 17296 + }, + { + "epoch": 7.24, + "learning_rate": 4.6562444107760227e-07, + "loss": 0.0412, + "step": 17297 + }, + { + "epoch": 7.24, + "learning_rate": 4.6511325890450374e-07, + "loss": 0.0475, + "step": 17298 + }, + { + "epoch": 7.24, + "learning_rate": 4.6460235080326264e-07, + "loss": 0.042, + "step": 17299 + }, + { + "epoch": 7.24, + "learning_rate": 4.64091716788565e-07, + "loss": 0.0475, + "step": 17300 + }, + { + "epoch": 7.24, + "learning_rate": 4.635813568750891e-07, + "loss": 0.0411, + "step": 17301 + }, + { + "epoch": 7.24, + "learning_rate": 4.630712710775054e-07, + "loss": 0.061, + "step": 17302 + }, + { + "epoch": 7.24, + "learning_rate": 4.625614594104755e-07, + "loss": 0.0551, + "step": 17303 + }, + { + "epoch": 7.24, + "learning_rate": 4.620519218886543e-07, + "loss": 0.0708, + "step": 17304 + }, + { + "epoch": 7.25, + "learning_rate": 4.6154265852668466e-07, + "loss": 0.0362, + "step": 17305 + }, + { + "epoch": 7.25, + "learning_rate": 4.61033669339207e-07, + "loss": 0.0756, + "step": 17306 + }, + { + "epoch": 7.25, + "learning_rate": 4.6052495434085185e-07, + "loss": 0.0703, + "step": 17307 + }, + { + "epoch": 7.25, + "learning_rate": 4.6001651354624087e-07, + "loss": 0.0462, + "step": 17308 + }, + { + "epoch": 7.25, + "learning_rate": 4.5950834696999127e-07, + "loss": 0.0519, + "step": 17309 + }, + { + "epoch": 7.25, + "learning_rate": 4.5900045462670684e-07, + "loss": 0.0645, + "step": 17310 + }, + { + "epoch": 7.25, + "learning_rate": 4.584928365309893e-07, + "loss": 0.05, + "step": 17311 + }, + { + "epoch": 7.25, + "learning_rate": 4.579854926974281e-07, + "loss": 0.061, + "step": 17312 + }, + { + "epoch": 7.25, + "learning_rate": 4.5747842314060486e-07, + "loss": 0.0557, + "step": 17313 + }, + { + "epoch": 7.25, + "learning_rate": 4.5697162787509686e-07, + "loss": 0.0476, + "step": 17314 + }, + { + "epoch": 7.25, + "learning_rate": 4.564651069154713e-07, + "loss": 0.053, + "step": 17315 + }, + { + "epoch": 7.25, + "learning_rate": 4.5595886027628765e-07, + "loss": 0.0552, + "step": 17316 + }, + { + "epoch": 7.25, + "learning_rate": 4.5545288797209875e-07, + "loss": 0.0505, + "step": 17317 + }, + { + "epoch": 7.25, + "learning_rate": 4.549471900174451e-07, + "loss": 0.0499, + "step": 17318 + }, + { + "epoch": 7.25, + "learning_rate": 4.544417664268652e-07, + "loss": 0.0369, + "step": 17319 + }, + { + "epoch": 7.25, + "learning_rate": 4.539366172148851e-07, + "loss": 0.0433, + "step": 17320 + }, + { + "epoch": 7.25, + "learning_rate": 4.534317423960266e-07, + "loss": 0.0449, + "step": 17321 + }, + { + "epoch": 7.25, + "learning_rate": 4.5292714198480135e-07, + "loss": 0.0608, + "step": 17322 + }, + { + "epoch": 7.25, + "learning_rate": 4.5242281599571223e-07, + "loss": 0.0682, + "step": 17323 + }, + { + "epoch": 7.25, + "learning_rate": 4.5191876444325986e-07, + "loss": 0.0345, + "step": 17324 + }, + { + "epoch": 7.25, + "learning_rate": 4.514149873419271e-07, + "loss": 0.0469, + "step": 17325 + }, + { + "epoch": 7.25, + "learning_rate": 4.5091148470619904e-07, + "loss": 0.0465, + "step": 17326 + }, + { + "epoch": 7.25, + "learning_rate": 4.504082565505452e-07, + "loss": 0.0575, + "step": 17327 + }, + { + "epoch": 7.25, + "learning_rate": 4.499053028894329e-07, + "loss": 0.055, + "step": 17328 + }, + { + "epoch": 7.26, + "learning_rate": 4.494026237373172e-07, + "loss": 0.0386, + "step": 17329 + }, + { + "epoch": 7.26, + "learning_rate": 4.489002191086478e-07, + "loss": 0.0713, + "step": 17330 + }, + { + "epoch": 7.26, + "learning_rate": 4.483980890178674e-07, + "loss": 0.0319, + "step": 17331 + }, + { + "epoch": 7.26, + "learning_rate": 4.478962334794068e-07, + "loss": 0.0569, + "step": 17332 + }, + { + "epoch": 7.26, + "learning_rate": 4.473946525076933e-07, + "loss": 0.0645, + "step": 17333 + }, + { + "epoch": 7.26, + "learning_rate": 4.468933461171432e-07, + "loss": 0.0429, + "step": 17334 + }, + { + "epoch": 7.26, + "learning_rate": 4.46392314322166e-07, + "loss": 0.0457, + "step": 17335 + }, + { + "epoch": 7.26, + "learning_rate": 4.458915571371647e-07, + "loss": 0.0455, + "step": 17336 + }, + { + "epoch": 7.26, + "learning_rate": 4.453910745765322e-07, + "loss": 0.0348, + "step": 17337 + }, + { + "epoch": 7.26, + "learning_rate": 4.4489086665465697e-07, + "loss": 0.0363, + "step": 17338 + }, + { + "epoch": 7.26, + "learning_rate": 4.44390933385912e-07, + "loss": 0.0432, + "step": 17339 + }, + { + "epoch": 7.26, + "learning_rate": 4.438912747846702e-07, + "loss": 0.0539, + "step": 17340 + }, + { + "epoch": 7.26, + "learning_rate": 4.4339189086529453e-07, + "loss": 0.058, + "step": 17341 + }, + { + "epoch": 7.26, + "learning_rate": 4.42892781642138e-07, + "loss": 0.0456, + "step": 17342 + }, + { + "epoch": 7.26, + "learning_rate": 4.42393947129548e-07, + "loss": 0.0764, + "step": 17343 + }, + { + "epoch": 7.26, + "learning_rate": 4.418953873418641e-07, + "loss": 0.0617, + "step": 17344 + }, + { + "epoch": 7.26, + "learning_rate": 4.413971022934138e-07, + "loss": 0.0425, + "step": 17345 + }, + { + "epoch": 7.26, + "learning_rate": 4.408990919985223e-07, + "loss": 0.0586, + "step": 17346 + }, + { + "epoch": 7.26, + "learning_rate": 4.4040135647150374e-07, + "loss": 0.0401, + "step": 17347 + }, + { + "epoch": 7.26, + "learning_rate": 4.399038957266644e-07, + "loss": 0.0491, + "step": 17348 + }, + { + "epoch": 7.26, + "learning_rate": 4.3940670977830614e-07, + "loss": 0.0355, + "step": 17349 + }, + { + "epoch": 7.26, + "learning_rate": 4.389097986407165e-07, + "loss": 0.069, + "step": 17350 + }, + { + "epoch": 7.26, + "learning_rate": 4.3841316232818064e-07, + "loss": 0.0486, + "step": 17351 + }, + { + "epoch": 7.26, + "learning_rate": 4.379168008549739e-07, + "loss": 0.0434, + "step": 17352 + }, + { + "epoch": 7.27, + "learning_rate": 4.3742071423536256e-07, + "loss": 0.045, + "step": 17353 + }, + { + "epoch": 7.27, + "learning_rate": 4.3692490248360755e-07, + "loss": 0.0497, + "step": 17354 + }, + { + "epoch": 7.27, + "learning_rate": 4.3642936561396063e-07, + "loss": 0.029, + "step": 17355 + }, + { + "epoch": 7.27, + "learning_rate": 4.359341036406639e-07, + "loss": 0.0545, + "step": 17356 + }, + { + "epoch": 7.27, + "learning_rate": 4.354391165779548e-07, + "loss": 0.0284, + "step": 17357 + }, + { + "epoch": 7.27, + "learning_rate": 4.3494440444006304e-07, + "loss": 0.0697, + "step": 17358 + }, + { + "epoch": 7.27, + "learning_rate": 4.344499672412039e-07, + "loss": 0.03, + "step": 17359 + }, + { + "epoch": 7.27, + "learning_rate": 4.3395580499559276e-07, + "loss": 0.06, + "step": 17360 + }, + { + "epoch": 7.27, + "learning_rate": 4.334619177174326e-07, + "loss": 0.0472, + "step": 17361 + }, + { + "epoch": 7.27, + "learning_rate": 4.3296830542092104e-07, + "loss": 0.0438, + "step": 17362 + }, + { + "epoch": 7.27, + "learning_rate": 4.3247496812024557e-07, + "loss": 0.0427, + "step": 17363 + }, + { + "epoch": 7.27, + "learning_rate": 4.319819058295893e-07, + "loss": 0.0448, + "step": 17364 + }, + { + "epoch": 7.27, + "learning_rate": 4.314891185631198e-07, + "loss": 0.05, + "step": 17365 + }, + { + "epoch": 7.27, + "learning_rate": 4.309966063350057e-07, + "loss": 0.0309, + "step": 17366 + }, + { + "epoch": 7.27, + "learning_rate": 4.305043691594024e-07, + "loss": 0.0388, + "step": 17367 + }, + { + "epoch": 7.27, + "learning_rate": 4.300124070504597e-07, + "loss": 0.0507, + "step": 17368 + }, + { + "epoch": 7.27, + "learning_rate": 4.2952072002231727e-07, + "loss": 0.045, + "step": 17369 + }, + { + "epoch": 7.27, + "learning_rate": 4.2902930808910946e-07, + "loss": 0.047, + "step": 17370 + }, + { + "epoch": 7.27, + "learning_rate": 4.285381712649628e-07, + "loss": 0.0591, + "step": 17371 + }, + { + "epoch": 7.27, + "learning_rate": 4.2804730956399255e-07, + "loss": 0.0318, + "step": 17372 + }, + { + "epoch": 7.27, + "learning_rate": 4.275567230003064e-07, + "loss": 0.0318, + "step": 17373 + }, + { + "epoch": 7.27, + "learning_rate": 4.2706641158800855e-07, + "loss": 0.0538, + "step": 17374 + }, + { + "epoch": 7.27, + "learning_rate": 4.265763753411922e-07, + "loss": 0.0387, + "step": 17375 + }, + { + "epoch": 7.27, + "learning_rate": 4.260866142739417e-07, + "loss": 0.0493, + "step": 17376 + }, + { + "epoch": 7.28, + "learning_rate": 4.2559712840033573e-07, + "loss": 0.0295, + "step": 17377 + }, + { + "epoch": 7.28, + "learning_rate": 4.2510791773444526e-07, + "loss": 0.0197, + "step": 17378 + }, + { + "epoch": 7.28, + "learning_rate": 4.2461898229033125e-07, + "loss": 0.033, + "step": 17379 + }, + { + "epoch": 7.28, + "learning_rate": 4.24130322082047e-07, + "loss": 0.0443, + "step": 17380 + }, + { + "epoch": 7.28, + "learning_rate": 4.236419371236389e-07, + "loss": 0.0478, + "step": 17381 + }, + { + "epoch": 7.28, + "learning_rate": 4.231538274291447e-07, + "loss": 0.0407, + "step": 17382 + }, + { + "epoch": 7.28, + "learning_rate": 4.226659930125965e-07, + "loss": 0.0503, + "step": 17383 + }, + { + "epoch": 7.28, + "learning_rate": 4.221784338880153e-07, + "loss": 0.0384, + "step": 17384 + }, + { + "epoch": 7.28, + "learning_rate": 4.216911500694165e-07, + "loss": 0.0419, + "step": 17385 + }, + { + "epoch": 7.28, + "learning_rate": 4.2120414157080457e-07, + "loss": 0.0671, + "step": 17386 + }, + { + "epoch": 7.28, + "learning_rate": 4.2071740840618045e-07, + "loss": 0.0418, + "step": 17387 + }, + { + "epoch": 7.28, + "learning_rate": 4.2023095058953303e-07, + "loss": 0.0592, + "step": 17388 + }, + { + "epoch": 7.28, + "learning_rate": 4.1974476813484654e-07, + "loss": 0.046, + "step": 17389 + }, + { + "epoch": 7.28, + "learning_rate": 4.1925886105609545e-07, + "loss": 0.0326, + "step": 17390 + }, + { + "epoch": 7.28, + "learning_rate": 4.187732293672475e-07, + "loss": 0.0493, + "step": 17391 + }, + { + "epoch": 7.28, + "learning_rate": 4.1828787308225927e-07, + "loss": 0.0503, + "step": 17392 + }, + { + "epoch": 7.28, + "learning_rate": 4.17802792215084e-07, + "loss": 0.0494, + "step": 17393 + }, + { + "epoch": 7.28, + "learning_rate": 4.173179867796662e-07, + "loss": 0.0586, + "step": 17394 + }, + { + "epoch": 7.28, + "learning_rate": 4.168334567899368e-07, + "loss": 0.051, + "step": 17395 + }, + { + "epoch": 7.28, + "learning_rate": 4.16349202259827e-07, + "loss": 0.0591, + "step": 17396 + }, + { + "epoch": 7.28, + "learning_rate": 4.158652232032545e-07, + "loss": 0.0404, + "step": 17397 + }, + { + "epoch": 7.28, + "learning_rate": 4.1538151963413043e-07, + "loss": 0.0403, + "step": 17398 + }, + { + "epoch": 7.28, + "learning_rate": 4.148980915663614e-07, + "loss": 0.0516, + "step": 17399 + }, + { + "epoch": 7.28, + "learning_rate": 4.1441493901383966e-07, + "loss": 0.063, + "step": 17400 + }, + { + "epoch": 7.29, + "learning_rate": 4.139320619904552e-07, + "loss": 0.0619, + "step": 17401 + }, + { + "epoch": 7.29, + "learning_rate": 4.134494605100858e-07, + "loss": 0.032, + "step": 17402 + }, + { + "epoch": 7.29, + "learning_rate": 4.12967134586606e-07, + "loss": 0.051, + "step": 17403 + }, + { + "epoch": 7.29, + "learning_rate": 4.124850842338779e-07, + "loss": 0.0308, + "step": 17404 + }, + { + "epoch": 7.29, + "learning_rate": 4.120033094657594e-07, + "loss": 0.0452, + "step": 17405 + }, + { + "epoch": 7.29, + "learning_rate": 4.1152181029609606e-07, + "loss": 0.0388, + "step": 17406 + }, + { + "epoch": 7.29, + "learning_rate": 4.1104058673873016e-07, + "loss": 0.0414, + "step": 17407 + }, + { + "epoch": 7.29, + "learning_rate": 4.105596388074928e-07, + "loss": 0.0416, + "step": 17408 + }, + { + "epoch": 7.29, + "learning_rate": 4.100789665162097e-07, + "loss": 0.0766, + "step": 17409 + }, + { + "epoch": 7.29, + "learning_rate": 4.095985698786964e-07, + "loss": 0.0513, + "step": 17410 + }, + { + "epoch": 7.29, + "learning_rate": 4.0911844890876183e-07, + "loss": 0.0473, + "step": 17411 + }, + { + "epoch": 7.29, + "learning_rate": 4.086386036202084e-07, + "loss": 0.0582, + "step": 17412 + }, + { + "epoch": 7.29, + "learning_rate": 4.0815903402682597e-07, + "loss": 0.0538, + "step": 17413 + }, + { + "epoch": 7.29, + "learning_rate": 4.076797401423993e-07, + "loss": 0.0553, + "step": 17414 + }, + { + "epoch": 7.29, + "learning_rate": 4.0720072198070725e-07, + "loss": 0.0634, + "step": 17415 + }, + { + "epoch": 7.29, + "learning_rate": 4.0672197955551773e-07, + "loss": 0.048, + "step": 17416 + }, + { + "epoch": 7.29, + "learning_rate": 4.0624351288059414e-07, + "loss": 0.0345, + "step": 17417 + }, + { + "epoch": 7.29, + "learning_rate": 4.057653219696844e-07, + "loss": 0.0466, + "step": 17418 + }, + { + "epoch": 7.29, + "learning_rate": 4.0528740683653864e-07, + "loss": 0.0434, + "step": 17419 + }, + { + "epoch": 7.29, + "learning_rate": 4.0480976749489254e-07, + "loss": 0.061, + "step": 17420 + }, + { + "epoch": 7.29, + "learning_rate": 4.0433240395847284e-07, + "loss": 0.0559, + "step": 17421 + }, + { + "epoch": 7.29, + "learning_rate": 4.0385531624100413e-07, + "loss": 0.0368, + "step": 17422 + }, + { + "epoch": 7.29, + "learning_rate": 4.033785043561989e-07, + "loss": 0.0419, + "step": 17423 + }, + { + "epoch": 7.29, + "learning_rate": 4.0290196831776264e-07, + "loss": 0.038, + "step": 17424 + }, + { + "epoch": 7.3, + "learning_rate": 4.024257081393934e-07, + "loss": 0.0545, + "step": 17425 + }, + { + "epoch": 7.3, + "learning_rate": 4.019497238347814e-07, + "loss": 0.0368, + "step": 17426 + }, + { + "epoch": 7.3, + "learning_rate": 4.014740154176056e-07, + "loss": 0.0479, + "step": 17427 + }, + { + "epoch": 7.3, + "learning_rate": 4.009985829015417e-07, + "loss": 0.0473, + "step": 17428 + }, + { + "epoch": 7.3, + "learning_rate": 4.0052342630025664e-07, + "loss": 0.0411, + "step": 17429 + }, + { + "epoch": 7.3, + "learning_rate": 4.0004854562740615e-07, + "loss": 0.0445, + "step": 17430 + }, + { + "epoch": 7.3, + "learning_rate": 3.9957394089664147e-07, + "loss": 0.0387, + "step": 17431 + }, + { + "epoch": 7.3, + "learning_rate": 3.990996121216062e-07, + "loss": 0.0619, + "step": 17432 + }, + { + "epoch": 7.3, + "learning_rate": 3.986255593159316e-07, + "loss": 0.0792, + "step": 17433 + }, + { + "epoch": 7.3, + "learning_rate": 3.981517824932446e-07, + "loss": 0.0331, + "step": 17434 + }, + { + "epoch": 7.3, + "learning_rate": 3.9767828166716535e-07, + "loss": 0.063, + "step": 17435 + }, + { + "epoch": 7.3, + "learning_rate": 3.9720505685130195e-07, + "loss": 0.0544, + "step": 17436 + }, + { + "epoch": 7.3, + "learning_rate": 3.967321080592579e-07, + "loss": 0.0414, + "step": 17437 + }, + { + "epoch": 7.3, + "learning_rate": 3.9625943530462787e-07, + "loss": 0.0573, + "step": 17438 + }, + { + "epoch": 7.3, + "learning_rate": 3.9578703860099987e-07, + "loss": 0.0431, + "step": 17439 + }, + { + "epoch": 7.3, + "learning_rate": 3.9531491796194975e-07, + "loss": 0.0329, + "step": 17440 + }, + { + "epoch": 7.3, + "learning_rate": 3.9484307340104886e-07, + "loss": 0.0559, + "step": 17441 + }, + { + "epoch": 7.3, + "learning_rate": 3.943715049318597e-07, + "loss": 0.0369, + "step": 17442 + }, + { + "epoch": 7.3, + "learning_rate": 3.939002125679381e-07, + "loss": 0.0628, + "step": 17443 + }, + { + "epoch": 7.3, + "learning_rate": 3.9342919632283093e-07, + "loss": 0.0454, + "step": 17444 + }, + { + "epoch": 7.3, + "learning_rate": 3.9295845621007635e-07, + "loss": 0.038, + "step": 17445 + }, + { + "epoch": 7.3, + "learning_rate": 3.924879922432068e-07, + "loss": 0.0705, + "step": 17446 + }, + { + "epoch": 7.3, + "learning_rate": 3.9201780443574367e-07, + "loss": 0.0651, + "step": 17447 + }, + { + "epoch": 7.31, + "learning_rate": 3.9154789280120287e-07, + "loss": 0.054, + "step": 17448 + }, + { + "epoch": 7.31, + "learning_rate": 3.9107825735309136e-07, + "loss": 0.0405, + "step": 17449 + }, + { + "epoch": 7.31, + "learning_rate": 3.906088981049083e-07, + "loss": 0.033, + "step": 17450 + }, + { + "epoch": 7.31, + "learning_rate": 3.901398150701452e-07, + "loss": 0.047, + "step": 17451 + }, + { + "epoch": 7.31, + "learning_rate": 3.8967100826228563e-07, + "loss": 0.0572, + "step": 17452 + }, + { + "epoch": 7.31, + "learning_rate": 3.8920247769480557e-07, + "loss": 0.0324, + "step": 17453 + }, + { + "epoch": 7.31, + "learning_rate": 3.8873422338117084e-07, + "loss": 0.0422, + "step": 17454 + }, + { + "epoch": 7.31, + "learning_rate": 3.8826624533484293e-07, + "loss": 0.0374, + "step": 17455 + }, + { + "epoch": 7.31, + "learning_rate": 3.8779854356927107e-07, + "loss": 0.0715, + "step": 17456 + }, + { + "epoch": 7.31, + "learning_rate": 3.873311180979011e-07, + "loss": 0.0487, + "step": 17457 + }, + { + "epoch": 7.31, + "learning_rate": 3.8686396893416797e-07, + "loss": 0.0439, + "step": 17458 + }, + { + "epoch": 7.31, + "learning_rate": 3.863970960915009e-07, + "loss": 0.0619, + "step": 17459 + }, + { + "epoch": 7.31, + "learning_rate": 3.8593049958331684e-07, + "loss": 0.0463, + "step": 17460 + }, + { + "epoch": 7.31, + "learning_rate": 3.854641794230296e-07, + "loss": 0.0467, + "step": 17461 + }, + { + "epoch": 7.31, + "learning_rate": 3.84998135624044e-07, + "loss": 0.0415, + "step": 17462 + }, + { + "epoch": 7.31, + "learning_rate": 3.845323681997526e-07, + "loss": 0.0458, + "step": 17463 + }, + { + "epoch": 7.31, + "learning_rate": 3.8406687716354695e-07, + "loss": 0.037, + "step": 17464 + }, + { + "epoch": 7.31, + "learning_rate": 3.836016625288053e-07, + "loss": 0.0711, + "step": 17465 + }, + { + "epoch": 7.31, + "learning_rate": 3.8313672430890016e-07, + "loss": 0.0652, + "step": 17466 + }, + { + "epoch": 7.31, + "learning_rate": 3.8267206251719756e-07, + "loss": 0.0621, + "step": 17467 + }, + { + "epoch": 7.31, + "learning_rate": 3.8220767716705133e-07, + "loss": 0.0326, + "step": 17468 + }, + { + "epoch": 7.31, + "learning_rate": 3.817435682718096e-07, + "loss": 0.0543, + "step": 17469 + }, + { + "epoch": 7.31, + "learning_rate": 3.81279735844815e-07, + "loss": 0.0264, + "step": 17470 + }, + { + "epoch": 7.31, + "learning_rate": 3.808161798993981e-07, + "loss": 0.0525, + "step": 17471 + }, + { + "epoch": 7.32, + "learning_rate": 3.803529004488848e-07, + "loss": 0.0604, + "step": 17472 + }, + { + "epoch": 7.32, + "learning_rate": 3.7988989750659235e-07, + "loss": 0.0645, + "step": 17473 + }, + { + "epoch": 7.32, + "learning_rate": 3.7942717108582663e-07, + "loss": 0.0479, + "step": 17474 + }, + { + "epoch": 7.32, + "learning_rate": 3.7896472119988927e-07, + "loss": 0.0339, + "step": 17475 + }, + { + "epoch": 7.32, + "learning_rate": 3.7850254786207406e-07, + "loss": 0.065, + "step": 17476 + }, + { + "epoch": 7.32, + "learning_rate": 3.780406510856649e-07, + "loss": 0.0603, + "step": 17477 + }, + { + "epoch": 7.32, + "learning_rate": 3.7757903088393886e-07, + "loss": 0.0492, + "step": 17478 + }, + { + "epoch": 7.32, + "learning_rate": 3.771176872701654e-07, + "loss": 0.0541, + "step": 17479 + }, + { + "epoch": 7.32, + "learning_rate": 3.7665662025760496e-07, + "loss": 0.0519, + "step": 17480 + }, + { + "epoch": 7.32, + "learning_rate": 3.7619582985951033e-07, + "loss": 0.0461, + "step": 17481 + }, + { + "epoch": 7.32, + "learning_rate": 3.757353160891253e-07, + "loss": 0.0587, + "step": 17482 + }, + { + "epoch": 7.32, + "learning_rate": 3.752750789596893e-07, + "loss": 0.0386, + "step": 17483 + }, + { + "epoch": 7.32, + "learning_rate": 3.748151184844295e-07, + "loss": 0.067, + "step": 17484 + }, + { + "epoch": 7.32, + "learning_rate": 3.743554346765699e-07, + "loss": 0.0416, + "step": 17485 + }, + { + "epoch": 7.32, + "learning_rate": 3.7389602754931973e-07, + "loss": 0.041, + "step": 17486 + }, + { + "epoch": 7.32, + "learning_rate": 3.734368971158875e-07, + "loss": 0.0373, + "step": 17487 + }, + { + "epoch": 7.32, + "learning_rate": 3.7297804338946807e-07, + "loss": 0.05, + "step": 17488 + }, + { + "epoch": 7.32, + "learning_rate": 3.725194663832521e-07, + "loss": 0.0446, + "step": 17489 + }, + { + "epoch": 7.32, + "learning_rate": 3.720611661104212e-07, + "loss": 0.0261, + "step": 17490 + }, + { + "epoch": 7.32, + "learning_rate": 3.716031425841482e-07, + "loss": 0.045, + "step": 17491 + }, + { + "epoch": 7.32, + "learning_rate": 3.7114539581759814e-07, + "loss": 0.0463, + "step": 17492 + }, + { + "epoch": 7.32, + "learning_rate": 3.706879258239293e-07, + "loss": 0.0356, + "step": 17493 + }, + { + "epoch": 7.32, + "learning_rate": 3.702307326162924e-07, + "loss": 0.0485, + "step": 17494 + }, + { + "epoch": 7.32, + "learning_rate": 3.697738162078268e-07, + "loss": 0.0504, + "step": 17495 + }, + { + "epoch": 7.33, + "learning_rate": 3.6931717661166764e-07, + "loss": 0.0378, + "step": 17496 + }, + { + "epoch": 7.33, + "learning_rate": 3.688608138409389e-07, + "loss": 0.0618, + "step": 17497 + }, + { + "epoch": 7.33, + "learning_rate": 3.6840472790876104e-07, + "loss": 0.0616, + "step": 17498 + }, + { + "epoch": 7.33, + "learning_rate": 3.6794891882824147e-07, + "loss": 0.0413, + "step": 17499 + }, + { + "epoch": 7.33, + "learning_rate": 3.6749338661248414e-07, + "loss": 0.0602, + "step": 17500 + }, + { + "epoch": 7.33, + "learning_rate": 3.6703813127458074e-07, + "loss": 0.064, + "step": 17501 + }, + { + "epoch": 7.33, + "learning_rate": 3.6658315282761757e-07, + "loss": 0.0716, + "step": 17502 + }, + { + "epoch": 7.33, + "learning_rate": 3.66128451284673e-07, + "loss": 0.0456, + "step": 17503 + }, + { + "epoch": 7.33, + "learning_rate": 3.656740266588177e-07, + "loss": 0.0486, + "step": 17504 + }, + { + "epoch": 7.33, + "learning_rate": 3.652198789631134e-07, + "loss": 0.0662, + "step": 17505 + }, + { + "epoch": 7.33, + "learning_rate": 3.64766008210613e-07, + "loss": 0.0621, + "step": 17506 + }, + { + "epoch": 7.33, + "learning_rate": 3.6431241441436617e-07, + "loss": 0.0472, + "step": 17507 + }, + { + "epoch": 7.33, + "learning_rate": 3.638590975874068e-07, + "loss": 0.0423, + "step": 17508 + }, + { + "epoch": 7.33, + "learning_rate": 3.634060577427667e-07, + "loss": 0.0613, + "step": 17509 + }, + { + "epoch": 7.33, + "learning_rate": 3.6295329489346776e-07, + "loss": 0.0548, + "step": 17510 + }, + { + "epoch": 7.33, + "learning_rate": 3.625008090525239e-07, + "loss": 0.0712, + "step": 17511 + }, + { + "epoch": 7.33, + "learning_rate": 3.6204860023294263e-07, + "loss": 0.0473, + "step": 17512 + }, + { + "epoch": 7.33, + "learning_rate": 3.6159666844772234e-07, + "loss": 0.0541, + "step": 17513 + }, + { + "epoch": 7.33, + "learning_rate": 3.611450137098538e-07, + "loss": 0.037, + "step": 17514 + }, + { + "epoch": 7.33, + "learning_rate": 3.606936360323166e-07, + "loss": 0.054, + "step": 17515 + }, + { + "epoch": 7.33, + "learning_rate": 3.6024253542808806e-07, + "loss": 0.0396, + "step": 17516 + }, + { + "epoch": 7.33, + "learning_rate": 3.597917119101335e-07, + "loss": 0.0425, + "step": 17517 + }, + { + "epoch": 7.33, + "learning_rate": 3.5934116549141027e-07, + "loss": 0.0495, + "step": 17518 + }, + { + "epoch": 7.33, + "learning_rate": 3.588908961848714e-07, + "loss": 0.0519, + "step": 17519 + }, + { + "epoch": 7.34, + "learning_rate": 3.5844090400345867e-07, + "loss": 0.0897, + "step": 17520 + }, + { + "epoch": 7.34, + "learning_rate": 3.579911889601073e-07, + "loss": 0.0397, + "step": 17521 + }, + { + "epoch": 7.34, + "learning_rate": 3.5754175106774147e-07, + "loss": 0.0531, + "step": 17522 + }, + { + "epoch": 7.34, + "learning_rate": 3.570925903392819e-07, + "loss": 0.0579, + "step": 17523 + }, + { + "epoch": 7.34, + "learning_rate": 3.5664370678763825e-07, + "loss": 0.0585, + "step": 17524 + }, + { + "epoch": 7.34, + "learning_rate": 3.561951004257147e-07, + "loss": 0.055, + "step": 17525 + }, + { + "epoch": 7.34, + "learning_rate": 3.5574677126640423e-07, + "loss": 0.0424, + "step": 17526 + }, + { + "epoch": 7.34, + "learning_rate": 3.5529871932259654e-07, + "loss": 0.0539, + "step": 17527 + }, + { + "epoch": 7.34, + "learning_rate": 3.5485094460716684e-07, + "loss": 0.0451, + "step": 17528 + }, + { + "epoch": 7.34, + "learning_rate": 3.5440344713298826e-07, + "loss": 0.0494, + "step": 17529 + }, + { + "epoch": 7.34, + "learning_rate": 3.5395622691292485e-07, + "loss": 0.0639, + "step": 17530 + }, + { + "epoch": 7.34, + "learning_rate": 3.5350928395982755e-07, + "loss": 0.0447, + "step": 17531 + }, + { + "epoch": 7.34, + "learning_rate": 3.530626182865471e-07, + "loss": 0.0295, + "step": 17532 + }, + { + "epoch": 7.34, + "learning_rate": 3.526162299059199e-07, + "loss": 0.0497, + "step": 17533 + }, + { + "epoch": 7.34, + "learning_rate": 3.5217011883077914e-07, + "loss": 0.0466, + "step": 17534 + }, + { + "epoch": 7.34, + "learning_rate": 3.5172428507394773e-07, + "loss": 0.0546, + "step": 17535 + }, + { + "epoch": 7.34, + "learning_rate": 3.5127872864823887e-07, + "loss": 0.0519, + "step": 17536 + }, + { + "epoch": 7.34, + "learning_rate": 3.5083344956646116e-07, + "loss": 0.0519, + "step": 17537 + }, + { + "epoch": 7.34, + "learning_rate": 3.5038844784141324e-07, + "loss": 0.0443, + "step": 17538 + }, + { + "epoch": 7.34, + "learning_rate": 3.4994372348588714e-07, + "loss": 0.0528, + "step": 17539 + }, + { + "epoch": 7.34, + "learning_rate": 3.494992765126637e-07, + "loss": 0.0274, + "step": 17540 + }, + { + "epoch": 7.34, + "learning_rate": 3.4905510693452273e-07, + "loss": 0.0516, + "step": 17541 + }, + { + "epoch": 7.34, + "learning_rate": 3.486112147642262e-07, + "loss": 0.0367, + "step": 17542 + }, + { + "epoch": 7.34, + "learning_rate": 3.481676000145362e-07, + "loss": 0.0601, + "step": 17543 + }, + { + "epoch": 7.35, + "learning_rate": 3.477242626982047e-07, + "loss": 0.0567, + "step": 17544 + }, + { + "epoch": 7.35, + "learning_rate": 3.4728120282797263e-07, + "loss": 0.0485, + "step": 17545 + }, + { + "epoch": 7.35, + "learning_rate": 3.4683842041657757e-07, + "loss": 0.0624, + "step": 17546 + }, + { + "epoch": 7.35, + "learning_rate": 3.463959154767449e-07, + "loss": 0.0641, + "step": 17547 + }, + { + "epoch": 7.35, + "learning_rate": 3.459536880211978e-07, + "loss": 0.035, + "step": 17548 + }, + { + "epoch": 7.35, + "learning_rate": 3.455117380626427e-07, + "loss": 0.0531, + "step": 17549 + }, + { + "epoch": 7.35, + "learning_rate": 3.450700656137862e-07, + "loss": 0.0611, + "step": 17550 + }, + { + "epoch": 7.35, + "learning_rate": 3.4462867068732253e-07, + "loss": 0.0546, + "step": 17551 + }, + { + "epoch": 7.35, + "learning_rate": 3.441875532959404e-07, + "loss": 0.0589, + "step": 17552 + }, + { + "epoch": 7.35, + "learning_rate": 3.437467134523198e-07, + "loss": 0.022, + "step": 17553 + }, + { + "epoch": 7.35, + "learning_rate": 3.4330615116912933e-07, + "loss": 0.0305, + "step": 17554 + }, + { + "epoch": 7.35, + "learning_rate": 3.428658664590356e-07, + "loss": 0.0283, + "step": 17555 + }, + { + "epoch": 7.35, + "learning_rate": 3.424258593346919e-07, + "loss": 0.0343, + "step": 17556 + }, + { + "epoch": 7.35, + "learning_rate": 3.419861298087468e-07, + "loss": 0.0567, + "step": 17557 + }, + { + "epoch": 7.35, + "learning_rate": 3.415466778938403e-07, + "loss": 0.052, + "step": 17558 + }, + { + "epoch": 7.35, + "learning_rate": 3.411075036026035e-07, + "loss": 0.0637, + "step": 17559 + }, + { + "epoch": 7.35, + "learning_rate": 3.406686069476606e-07, + "loss": 0.0489, + "step": 17560 + }, + { + "epoch": 7.35, + "learning_rate": 3.402299879416271e-07, + "loss": 0.0592, + "step": 17561 + }, + { + "epoch": 7.35, + "learning_rate": 3.3979164659711184e-07, + "loss": 0.0508, + "step": 17562 + }, + { + "epoch": 7.35, + "learning_rate": 3.393535829267114e-07, + "loss": 0.0409, + "step": 17563 + }, + { + "epoch": 7.35, + "learning_rate": 3.389157969430201e-07, + "loss": 0.0555, + "step": 17564 + }, + { + "epoch": 7.35, + "learning_rate": 3.384782886586213e-07, + "loss": 0.0502, + "step": 17565 + }, + { + "epoch": 7.35, + "learning_rate": 3.3804105808609047e-07, + "loss": 0.0659, + "step": 17566 + }, + { + "epoch": 7.35, + "learning_rate": 3.3760410523799526e-07, + "loss": 0.058, + "step": 17567 + }, + { + "epoch": 7.36, + "learning_rate": 3.371674301268979e-07, + "loss": 0.0324, + "step": 17568 + }, + { + "epoch": 7.36, + "learning_rate": 3.3673103276534616e-07, + "loss": 0.0626, + "step": 17569 + }, + { + "epoch": 7.36, + "learning_rate": 3.3629491316588546e-07, + "loss": 0.0406, + "step": 17570 + }, + { + "epoch": 7.36, + "learning_rate": 3.3585907134105253e-07, + "loss": 0.0371, + "step": 17571 + }, + { + "epoch": 7.36, + "learning_rate": 3.3542350730337516e-07, + "loss": 0.0589, + "step": 17572 + }, + { + "epoch": 7.36, + "learning_rate": 3.349882210653732e-07, + "loss": 0.0353, + "step": 17573 + }, + { + "epoch": 7.36, + "learning_rate": 3.345532126395579e-07, + "loss": 0.0669, + "step": 17574 + }, + { + "epoch": 7.36, + "learning_rate": 3.3411848203843357e-07, + "loss": 0.0595, + "step": 17575 + }, + { + "epoch": 7.36, + "learning_rate": 3.336840292744969e-07, + "loss": 0.0619, + "step": 17576 + }, + { + "epoch": 7.36, + "learning_rate": 3.332498543602347e-07, + "loss": 0.0613, + "step": 17577 + }, + { + "epoch": 7.36, + "learning_rate": 3.328159573081258e-07, + "loss": 0.0544, + "step": 17578 + }, + { + "epoch": 7.36, + "learning_rate": 3.323823381306446e-07, + "loss": 0.0532, + "step": 17579 + }, + { + "epoch": 7.36, + "learning_rate": 3.319489968402545e-07, + "loss": 0.0328, + "step": 17580 + }, + { + "epoch": 7.36, + "learning_rate": 3.3151593344941113e-07, + "loss": 0.0449, + "step": 17581 + }, + { + "epoch": 7.36, + "learning_rate": 3.310831479705634e-07, + "loss": 0.0474, + "step": 17582 + }, + { + "epoch": 7.36, + "learning_rate": 3.3065064041614914e-07, + "loss": 0.057, + "step": 17583 + }, + { + "epoch": 7.36, + "learning_rate": 3.302184107986017e-07, + "loss": 0.0557, + "step": 17584 + }, + { + "epoch": 7.36, + "learning_rate": 3.297864591303457e-07, + "loss": 0.0552, + "step": 17585 + }, + { + "epoch": 7.36, + "learning_rate": 3.2935478542379664e-07, + "loss": 0.0535, + "step": 17586 + }, + { + "epoch": 7.36, + "learning_rate": 3.289233896913624e-07, + "loss": 0.0573, + "step": 17587 + }, + { + "epoch": 7.36, + "learning_rate": 3.284922719454453e-07, + "loss": 0.0596, + "step": 17588 + }, + { + "epoch": 7.36, + "learning_rate": 3.2806143219843324e-07, + "loss": 0.0332, + "step": 17589 + }, + { + "epoch": 7.36, + "learning_rate": 3.2763087046271404e-07, + "loss": 0.0382, + "step": 17590 + }, + { + "epoch": 7.36, + "learning_rate": 3.272005867506611e-07, + "loss": 0.047, + "step": 17591 + }, + { + "epoch": 7.37, + "learning_rate": 3.2677058107464467e-07, + "loss": 0.0582, + "step": 17592 + }, + { + "epoch": 7.37, + "learning_rate": 3.263408534470236e-07, + "loss": 0.0572, + "step": 17593 + }, + { + "epoch": 7.37, + "learning_rate": 3.259114038801503e-07, + "loss": 0.0606, + "step": 17594 + }, + { + "epoch": 7.37, + "learning_rate": 3.2548223238637045e-07, + "loss": 0.0543, + "step": 17595 + }, + { + "epoch": 7.37, + "learning_rate": 3.250533389780175e-07, + "loss": 0.0382, + "step": 17596 + }, + { + "epoch": 7.37, + "learning_rate": 3.246247236674205e-07, + "loss": 0.0457, + "step": 17597 + }, + { + "epoch": 7.37, + "learning_rate": 3.241963864669018e-07, + "loss": 0.0439, + "step": 17598 + }, + { + "epoch": 7.37, + "learning_rate": 3.237683273887704e-07, + "loss": 0.0578, + "step": 17599 + }, + { + "epoch": 7.37, + "learning_rate": 3.233405464453321e-07, + "loss": 0.0534, + "step": 17600 + }, + { + "epoch": 7.37, + "learning_rate": 3.2291304364888256e-07, + "loss": 0.0599, + "step": 17601 + }, + { + "epoch": 7.37, + "learning_rate": 3.224858190117108e-07, + "loss": 0.0686, + "step": 17602 + }, + { + "epoch": 7.37, + "learning_rate": 3.220588725460971e-07, + "loss": 0.0503, + "step": 17603 + }, + { + "epoch": 7.37, + "learning_rate": 3.216322042643127e-07, + "loss": 0.0489, + "step": 17604 + }, + { + "epoch": 7.37, + "learning_rate": 3.2120581417862115e-07, + "loss": 0.0296, + "step": 17605 + }, + { + "epoch": 7.37, + "learning_rate": 3.2077970230128043e-07, + "loss": 0.0557, + "step": 17606 + }, + { + "epoch": 7.37, + "learning_rate": 3.203538686445373e-07, + "loss": 0.0553, + "step": 17607 + }, + { + "epoch": 7.37, + "learning_rate": 3.1992831322063324e-07, + "loss": 0.042, + "step": 17608 + }, + { + "epoch": 7.37, + "learning_rate": 3.1950303604180057e-07, + "loss": 0.046, + "step": 17609 + }, + { + "epoch": 7.37, + "learning_rate": 3.1907803712026173e-07, + "loss": 0.0723, + "step": 17610 + }, + { + "epoch": 7.37, + "learning_rate": 3.186533164682348e-07, + "loss": 0.0562, + "step": 17611 + }, + { + "epoch": 7.37, + "learning_rate": 3.182288740979267e-07, + "loss": 0.0347, + "step": 17612 + }, + { + "epoch": 7.37, + "learning_rate": 3.1780471002153755e-07, + "loss": 0.0586, + "step": 17613 + }, + { + "epoch": 7.37, + "learning_rate": 3.17380824251261e-07, + "loss": 0.0364, + "step": 17614 + }, + { + "epoch": 7.37, + "learning_rate": 3.169572167992818e-07, + "loss": 0.0607, + "step": 17615 + }, + { + "epoch": 7.38, + "learning_rate": 3.1653388767777237e-07, + "loss": 0.0533, + "step": 17616 + }, + { + "epoch": 7.38, + "learning_rate": 3.161108368989052e-07, + "loss": 0.0389, + "step": 17617 + }, + { + "epoch": 7.38, + "learning_rate": 3.1568806447483726e-07, + "loss": 0.0412, + "step": 17618 + }, + { + "epoch": 7.38, + "learning_rate": 3.152655704177221e-07, + "loss": 0.0473, + "step": 17619 + }, + { + "epoch": 7.38, + "learning_rate": 3.1484335473970564e-07, + "loss": 0.0459, + "step": 17620 + }, + { + "epoch": 7.38, + "learning_rate": 3.144214174529214e-07, + "loss": 0.0393, + "step": 17621 + }, + { + "epoch": 7.38, + "learning_rate": 3.139997585694998e-07, + "loss": 0.0354, + "step": 17622 + }, + { + "epoch": 7.38, + "learning_rate": 3.1357837810156e-07, + "loss": 0.0451, + "step": 17623 + }, + { + "epoch": 7.38, + "learning_rate": 3.1315727606121226e-07, + "loss": 0.0397, + "step": 17624 + }, + { + "epoch": 7.38, + "learning_rate": 3.127364524605636e-07, + "loss": 0.0487, + "step": 17625 + }, + { + "epoch": 7.38, + "learning_rate": 3.123159073117088e-07, + "loss": 0.0584, + "step": 17626 + }, + { + "epoch": 7.38, + "learning_rate": 3.1189564062673704e-07, + "loss": 0.0406, + "step": 17627 + }, + { + "epoch": 7.38, + "learning_rate": 3.1147565241772757e-07, + "loss": 0.0539, + "step": 17628 + }, + { + "epoch": 7.38, + "learning_rate": 3.110559426967552e-07, + "loss": 0.0452, + "step": 17629 + }, + { + "epoch": 7.38, + "learning_rate": 3.1063651147587913e-07, + "loss": 0.0421, + "step": 17630 + }, + { + "epoch": 7.38, + "learning_rate": 3.1021735876715973e-07, + "loss": 0.0597, + "step": 17631 + }, + { + "epoch": 7.38, + "learning_rate": 3.0979848458264293e-07, + "loss": 0.0551, + "step": 17632 + }, + { + "epoch": 7.38, + "learning_rate": 3.093798889343702e-07, + "loss": 0.0538, + "step": 17633 + }, + { + "epoch": 7.38, + "learning_rate": 3.0896157183437305e-07, + "loss": 0.0426, + "step": 17634 + }, + { + "epoch": 7.38, + "learning_rate": 3.0854353329467624e-07, + "loss": 0.0381, + "step": 17635 + }, + { + "epoch": 7.38, + "learning_rate": 3.081257733272969e-07, + "loss": 0.0315, + "step": 17636 + }, + { + "epoch": 7.38, + "learning_rate": 3.077082919442398e-07, + "loss": 0.0684, + "step": 17637 + }, + { + "epoch": 7.38, + "learning_rate": 3.072910891575076e-07, + "loss": 0.0511, + "step": 17638 + }, + { + "epoch": 7.38, + "learning_rate": 3.0687416497909185e-07, + "loss": 0.0481, + "step": 17639 + }, + { + "epoch": 7.39, + "learning_rate": 3.0645751942097623e-07, + "loss": 0.0658, + "step": 17640 + }, + { + "epoch": 7.39, + "learning_rate": 3.060411524951379e-07, + "loss": 0.0427, + "step": 17641 + }, + { + "epoch": 7.39, + "learning_rate": 3.05625064213545e-07, + "loss": 0.0536, + "step": 17642 + }, + { + "epoch": 7.39, + "learning_rate": 3.0520925458815576e-07, + "loss": 0.0519, + "step": 17643 + }, + { + "epoch": 7.39, + "learning_rate": 3.0479372363092506e-07, + "loss": 0.0414, + "step": 17644 + }, + { + "epoch": 7.39, + "learning_rate": 3.043784713537945e-07, + "loss": 0.0615, + "step": 17645 + }, + { + "epoch": 7.39, + "learning_rate": 3.0396349776870004e-07, + "loss": 0.0434, + "step": 17646 + }, + { + "epoch": 7.39, + "learning_rate": 3.0354880288757214e-07, + "loss": 0.0634, + "step": 17647 + }, + { + "epoch": 7.39, + "learning_rate": 3.0313438672232796e-07, + "loss": 0.0856, + "step": 17648 + }, + { + "epoch": 7.39, + "learning_rate": 3.0272024928488244e-07, + "loss": 0.0379, + "step": 17649 + }, + { + "epoch": 7.39, + "learning_rate": 3.023063905871382e-07, + "loss": 0.0397, + "step": 17650 + }, + { + "epoch": 7.39, + "learning_rate": 3.018928106409902e-07, + "loss": 0.0358, + "step": 17651 + }, + { + "epoch": 7.39, + "learning_rate": 3.014795094583278e-07, + "loss": 0.039, + "step": 17652 + }, + { + "epoch": 7.39, + "learning_rate": 3.0106648705103045e-07, + "loss": 0.0523, + "step": 17653 + }, + { + "epoch": 7.39, + "learning_rate": 3.0065374343096974e-07, + "loss": 0.0526, + "step": 17654 + }, + { + "epoch": 7.39, + "learning_rate": 3.002412786100106e-07, + "loss": 0.0791, + "step": 17655 + }, + { + "epoch": 7.39, + "learning_rate": 2.9982909260001023e-07, + "loss": 0.0536, + "step": 17656 + }, + { + "epoch": 7.39, + "learning_rate": 2.994171854128125e-07, + "loss": 0.0419, + "step": 17657 + }, + { + "epoch": 7.39, + "learning_rate": 2.9900555706026015e-07, + "loss": 0.036, + "step": 17658 + }, + { + "epoch": 7.39, + "learning_rate": 2.985942075541848e-07, + "loss": 0.0623, + "step": 17659 + }, + { + "epoch": 7.39, + "learning_rate": 2.981831369064103e-07, + "loss": 0.051, + "step": 17660 + }, + { + "epoch": 7.39, + "learning_rate": 2.9777234512875174e-07, + "loss": 0.0405, + "step": 17661 + }, + { + "epoch": 7.39, + "learning_rate": 2.973618322330174e-07, + "loss": 0.0401, + "step": 17662 + }, + { + "epoch": 7.4, + "learning_rate": 2.9695159823100896e-07, + "loss": 0.0441, + "step": 17663 + }, + { + "epoch": 7.4, + "learning_rate": 2.965416431345147e-07, + "loss": 0.0691, + "step": 17664 + }, + { + "epoch": 7.4, + "learning_rate": 2.9613196695532085e-07, + "loss": 0.0394, + "step": 17665 + }, + { + "epoch": 7.4, + "learning_rate": 2.9572256970520353e-07, + "loss": 0.0536, + "step": 17666 + }, + { + "epoch": 7.4, + "learning_rate": 2.953134513959277e-07, + "loss": 0.0639, + "step": 17667 + }, + { + "epoch": 7.4, + "learning_rate": 2.9490461203925513e-07, + "loss": 0.0285, + "step": 17668 + }, + { + "epoch": 7.4, + "learning_rate": 2.944960516469375e-07, + "loss": 0.0541, + "step": 17669 + }, + { + "epoch": 7.4, + "learning_rate": 2.940877702307199e-07, + "loss": 0.0451, + "step": 17670 + }, + { + "epoch": 7.4, + "learning_rate": 2.936797678023351e-07, + "loss": 0.0423, + "step": 17671 + }, + { + "epoch": 7.4, + "learning_rate": 2.9327204437351266e-07, + "loss": 0.0591, + "step": 17672 + }, + { + "epoch": 7.4, + "learning_rate": 2.9286459995597096e-07, + "loss": 0.0501, + "step": 17673 + }, + { + "epoch": 7.4, + "learning_rate": 2.9245743456142283e-07, + "loss": 0.0577, + "step": 17674 + }, + { + "epoch": 7.4, + "learning_rate": 2.9205054820157007e-07, + "loss": 0.0485, + "step": 17675 + }, + { + "epoch": 7.4, + "learning_rate": 2.916439408881111e-07, + "loss": 0.0518, + "step": 17676 + }, + { + "epoch": 7.4, + "learning_rate": 2.912376126327332e-07, + "loss": 0.0455, + "step": 17677 + }, + { + "epoch": 7.4, + "learning_rate": 2.908315634471126e-07, + "loss": 0.0631, + "step": 17678 + }, + { + "epoch": 7.4, + "learning_rate": 2.9042579334292333e-07, + "loss": 0.0396, + "step": 17679 + }, + { + "epoch": 7.4, + "learning_rate": 2.900203023318293e-07, + "loss": 0.0586, + "step": 17680 + }, + { + "epoch": 7.4, + "learning_rate": 2.8961509042548466e-07, + "loss": 0.0546, + "step": 17681 + }, + { + "epoch": 7.4, + "learning_rate": 2.892101576355366e-07, + "loss": 0.0392, + "step": 17682 + }, + { + "epoch": 7.4, + "learning_rate": 2.8880550397362705e-07, + "loss": 0.0524, + "step": 17683 + }, + { + "epoch": 7.4, + "learning_rate": 2.8840112945138553e-07, + "loss": 0.0387, + "step": 17684 + }, + { + "epoch": 7.4, + "learning_rate": 2.879970340804339e-07, + "loss": 0.0469, + "step": 17685 + }, + { + "epoch": 7.4, + "learning_rate": 2.875932178723906e-07, + "loss": 0.0548, + "step": 17686 + }, + { + "epoch": 7.41, + "learning_rate": 2.871896808388608e-07, + "loss": 0.0616, + "step": 17687 + }, + { + "epoch": 7.41, + "learning_rate": 2.867864229914452e-07, + "loss": 0.0609, + "step": 17688 + }, + { + "epoch": 7.41, + "learning_rate": 2.863834443417357e-07, + "loss": 0.0426, + "step": 17689 + }, + { + "epoch": 7.41, + "learning_rate": 2.85980744901313e-07, + "loss": 0.0461, + "step": 17690 + }, + { + "epoch": 7.41, + "learning_rate": 2.855783246817545e-07, + "loss": 0.0395, + "step": 17691 + }, + { + "epoch": 7.41, + "learning_rate": 2.851761836946265e-07, + "loss": 0.0499, + "step": 17692 + }, + { + "epoch": 7.41, + "learning_rate": 2.847743219514876e-07, + "loss": 0.0421, + "step": 17693 + }, + { + "epoch": 7.41, + "learning_rate": 2.8437273946388954e-07, + "loss": 0.0492, + "step": 17694 + }, + { + "epoch": 7.41, + "learning_rate": 2.839714362433754e-07, + "loss": 0.0464, + "step": 17695 + }, + { + "epoch": 7.41, + "learning_rate": 2.8357041230148153e-07, + "loss": 0.0623, + "step": 17696 + }, + { + "epoch": 7.41, + "learning_rate": 2.8316966764973417e-07, + "loss": 0.0539, + "step": 17697 + }, + { + "epoch": 7.41, + "learning_rate": 2.827692022996509e-07, + "loss": 0.0522, + "step": 17698 + }, + { + "epoch": 7.41, + "learning_rate": 2.823690162627446e-07, + "loss": 0.028, + "step": 17699 + }, + { + "epoch": 7.41, + "learning_rate": 2.819691095505173e-07, + "loss": 0.0448, + "step": 17700 + }, + { + "epoch": 7.41, + "learning_rate": 2.815694821744652e-07, + "loss": 0.0507, + "step": 17701 + }, + { + "epoch": 7.41, + "learning_rate": 2.811701341460737e-07, + "loss": 0.0512, + "step": 17702 + }, + { + "epoch": 7.41, + "learning_rate": 2.807710654768214e-07, + "loss": 0.0467, + "step": 17703 + }, + { + "epoch": 7.41, + "learning_rate": 2.803722761781824e-07, + "loss": 0.0349, + "step": 17704 + }, + { + "epoch": 7.41, + "learning_rate": 2.7997376626161644e-07, + "loss": 0.0453, + "step": 17705 + }, + { + "epoch": 7.41, + "learning_rate": 2.7957553573857876e-07, + "loss": 0.0674, + "step": 17706 + }, + { + "epoch": 7.41, + "learning_rate": 2.791775846205158e-07, + "loss": 0.0486, + "step": 17707 + }, + { + "epoch": 7.41, + "learning_rate": 2.787799129188673e-07, + "loss": 0.0415, + "step": 17708 + }, + { + "epoch": 7.41, + "learning_rate": 2.783825206450641e-07, + "loss": 0.0466, + "step": 17709 + }, + { + "epoch": 7.41, + "learning_rate": 2.779854078105304e-07, + "loss": 0.0504, + "step": 17710 + }, + { + "epoch": 7.42, + "learning_rate": 2.775885744266782e-07, + "loss": 0.0748, + "step": 17711 + }, + { + "epoch": 7.42, + "learning_rate": 2.771920205049139e-07, + "loss": 0.0579, + "step": 17712 + }, + { + "epoch": 7.42, + "learning_rate": 2.767957460566373e-07, + "loss": 0.05, + "step": 17713 + }, + { + "epoch": 7.42, + "learning_rate": 2.763997510932381e-07, + "loss": 0.0371, + "step": 17714 + }, + { + "epoch": 7.42, + "learning_rate": 2.7600403562609955e-07, + "loss": 0.0392, + "step": 17715 + }, + { + "epoch": 7.42, + "learning_rate": 2.756085996665969e-07, + "loss": 0.0545, + "step": 17716 + }, + { + "epoch": 7.42, + "learning_rate": 2.752134432260956e-07, + "loss": 0.0601, + "step": 17717 + }, + { + "epoch": 7.42, + "learning_rate": 2.7481856631595415e-07, + "loss": 0.0515, + "step": 17718 + }, + { + "epoch": 7.42, + "learning_rate": 2.7442396894752365e-07, + "loss": 0.0655, + "step": 17719 + }, + { + "epoch": 7.42, + "learning_rate": 2.7402965113214495e-07, + "loss": 0.0449, + "step": 17720 + }, + { + "epoch": 7.42, + "learning_rate": 2.7363561288115347e-07, + "loss": 0.039, + "step": 17721 + }, + { + "epoch": 7.42, + "learning_rate": 2.7324185420587457e-07, + "loss": 0.0347, + "step": 17722 + }, + { + "epoch": 7.42, + "learning_rate": 2.728483751176281e-07, + "loss": 0.0471, + "step": 17723 + }, + { + "epoch": 7.42, + "learning_rate": 2.7245517562772385e-07, + "loss": 0.0554, + "step": 17724 + }, + { + "epoch": 7.42, + "learning_rate": 2.720622557474628e-07, + "loss": 0.0563, + "step": 17725 + }, + { + "epoch": 7.42, + "learning_rate": 2.716696154881393e-07, + "loss": 0.0543, + "step": 17726 + }, + { + "epoch": 7.42, + "learning_rate": 2.7127725486104094e-07, + "loss": 0.067, + "step": 17727 + }, + { + "epoch": 7.42, + "learning_rate": 2.7088517387744426e-07, + "loss": 0.0382, + "step": 17728 + }, + { + "epoch": 7.42, + "learning_rate": 2.704933725486192e-07, + "loss": 0.0575, + "step": 17729 + }, + { + "epoch": 7.42, + "learning_rate": 2.7010185088582886e-07, + "loss": 0.0454, + "step": 17730 + }, + { + "epoch": 7.42, + "learning_rate": 2.697106089003276e-07, + "loss": 0.0555, + "step": 17731 + }, + { + "epoch": 7.42, + "learning_rate": 2.6931964660335985e-07, + "loss": 0.0613, + "step": 17732 + }, + { + "epoch": 7.42, + "learning_rate": 2.689289640061632e-07, + "loss": 0.0592, + "step": 17733 + }, + { + "epoch": 7.42, + "learning_rate": 2.685385611199698e-07, + "loss": 0.0351, + "step": 17734 + }, + { + "epoch": 7.43, + "learning_rate": 2.681484379559984e-07, + "loss": 0.0521, + "step": 17735 + }, + { + "epoch": 7.43, + "learning_rate": 2.677585945254646e-07, + "loss": 0.065, + "step": 17736 + }, + { + "epoch": 7.43, + "learning_rate": 2.673690308395738e-07, + "loss": 0.045, + "step": 17737 + }, + { + "epoch": 7.43, + "learning_rate": 2.6697974690952477e-07, + "loss": 0.03, + "step": 17738 + }, + { + "epoch": 7.43, + "learning_rate": 2.6659074274650533e-07, + "loss": 0.0582, + "step": 17739 + }, + { + "epoch": 7.43, + "learning_rate": 2.662020183616976e-07, + "loss": 0.0468, + "step": 17740 + }, + { + "epoch": 7.43, + "learning_rate": 2.6581357376627483e-07, + "loss": 0.0598, + "step": 17741 + }, + { + "epoch": 7.43, + "learning_rate": 2.6542540897140256e-07, + "loss": 0.0732, + "step": 17742 + }, + { + "epoch": 7.43, + "learning_rate": 2.650375239882386e-07, + "loss": 0.0396, + "step": 17743 + }, + { + "epoch": 7.43, + "learning_rate": 2.646499188279328e-07, + "loss": 0.0468, + "step": 17744 + }, + { + "epoch": 7.43, + "learning_rate": 2.642625935016274e-07, + "loss": 0.0367, + "step": 17745 + }, + { + "epoch": 7.43, + "learning_rate": 2.6387554802045246e-07, + "loss": 0.0477, + "step": 17746 + }, + { + "epoch": 7.43, + "learning_rate": 2.6348878239553677e-07, + "loss": 0.0628, + "step": 17747 + }, + { + "epoch": 7.43, + "learning_rate": 2.631022966379948e-07, + "loss": 0.0582, + "step": 17748 + }, + { + "epoch": 7.43, + "learning_rate": 2.6271609075893656e-07, + "loss": 0.0567, + "step": 17749 + }, + { + "epoch": 7.43, + "learning_rate": 2.6233016476946427e-07, + "loss": 0.0632, + "step": 17750 + }, + { + "epoch": 7.43, + "learning_rate": 2.619445186806713e-07, + "loss": 0.0567, + "step": 17751 + }, + { + "epoch": 7.43, + "learning_rate": 2.6155915250364094e-07, + "loss": 0.0392, + "step": 17752 + }, + { + "epoch": 7.43, + "learning_rate": 2.611740662494511e-07, + "loss": 0.0517, + "step": 17753 + }, + { + "epoch": 7.43, + "learning_rate": 2.607892599291695e-07, + "loss": 0.0436, + "step": 17754 + }, + { + "epoch": 7.43, + "learning_rate": 2.6040473355385954e-07, + "loss": 0.0579, + "step": 17755 + }, + { + "epoch": 7.43, + "learning_rate": 2.600204871345724e-07, + "loss": 0.0332, + "step": 17756 + }, + { + "epoch": 7.43, + "learning_rate": 2.596365206823537e-07, + "loss": 0.0377, + "step": 17757 + }, + { + "epoch": 7.43, + "learning_rate": 2.59252834208239e-07, + "loss": 0.0295, + "step": 17758 + }, + { + "epoch": 7.44, + "learning_rate": 2.588694277232595e-07, + "loss": 0.0364, + "step": 17759 + }, + { + "epoch": 7.44, + "learning_rate": 2.58486301238432e-07, + "loss": 0.0565, + "step": 17760 + }, + { + "epoch": 7.44, + "learning_rate": 2.5810345476477207e-07, + "loss": 0.0415, + "step": 17761 + }, + { + "epoch": 7.44, + "learning_rate": 2.5772088831328313e-07, + "loss": 0.0432, + "step": 17762 + }, + { + "epoch": 7.44, + "learning_rate": 2.5733860189496306e-07, + "loss": 0.0583, + "step": 17763 + }, + { + "epoch": 7.44, + "learning_rate": 2.5695659552079866e-07, + "loss": 0.0525, + "step": 17764 + }, + { + "epoch": 7.44, + "learning_rate": 2.565748692017722e-07, + "loss": 0.0465, + "step": 17765 + }, + { + "epoch": 7.44, + "learning_rate": 2.5619342294885387e-07, + "loss": 0.0576, + "step": 17766 + }, + { + "epoch": 7.44, + "learning_rate": 2.5581225677300925e-07, + "loss": 0.0588, + "step": 17767 + }, + { + "epoch": 7.44, + "learning_rate": 2.5543137068519407e-07, + "loss": 0.0535, + "step": 17768 + }, + { + "epoch": 7.44, + "learning_rate": 2.5505076469635734e-07, + "loss": 0.0495, + "step": 17769 + }, + { + "epoch": 7.44, + "learning_rate": 2.5467043881743814e-07, + "loss": 0.0792, + "step": 17770 + }, + { + "epoch": 7.44, + "learning_rate": 2.5429039305937095e-07, + "loss": 0.0626, + "step": 17771 + }, + { + "epoch": 7.44, + "learning_rate": 2.5391062743307717e-07, + "loss": 0.0569, + "step": 17772 + }, + { + "epoch": 7.44, + "learning_rate": 2.5353114194947457e-07, + "loss": 0.0197, + "step": 17773 + }, + { + "epoch": 7.44, + "learning_rate": 2.5315193661947013e-07, + "loss": 0.0386, + "step": 17774 + }, + { + "epoch": 7.44, + "learning_rate": 2.5277301145396283e-07, + "loss": 0.0483, + "step": 17775 + }, + { + "epoch": 7.44, + "learning_rate": 2.523943664638473e-07, + "loss": 0.0573, + "step": 17776 + }, + { + "epoch": 7.44, + "learning_rate": 2.5201600166000485e-07, + "loss": 0.0374, + "step": 17777 + }, + { + "epoch": 7.44, + "learning_rate": 2.5163791705331343e-07, + "loss": 0.0404, + "step": 17778 + }, + { + "epoch": 7.44, + "learning_rate": 2.512601126546399e-07, + "loss": 0.0402, + "step": 17779 + }, + { + "epoch": 7.44, + "learning_rate": 2.508825884748423e-07, + "loss": 0.0507, + "step": 17780 + }, + { + "epoch": 7.44, + "learning_rate": 2.5050534452477406e-07, + "loss": 0.0486, + "step": 17781 + }, + { + "epoch": 7.44, + "learning_rate": 2.5012838081527766e-07, + "loss": 0.0545, + "step": 17782 + }, + { + "epoch": 7.45, + "learning_rate": 2.4975169735718896e-07, + "loss": 0.0653, + "step": 17783 + }, + { + "epoch": 7.45, + "learning_rate": 2.493752941613348e-07, + "loss": 0.0716, + "step": 17784 + }, + { + "epoch": 7.45, + "learning_rate": 2.4899917123853647e-07, + "loss": 0.0518, + "step": 17785 + }, + { + "epoch": 7.45, + "learning_rate": 2.4862332859960426e-07, + "loss": 0.0181, + "step": 17786 + }, + { + "epoch": 7.45, + "learning_rate": 2.4824776625534064e-07, + "loss": 0.0417, + "step": 17787 + }, + { + "epoch": 7.45, + "learning_rate": 2.478724842165414e-07, + "loss": 0.0314, + "step": 17788 + }, + { + "epoch": 7.45, + "learning_rate": 2.4749748249399463e-07, + "loss": 0.034, + "step": 17789 + }, + { + "epoch": 7.45, + "learning_rate": 2.4712276109847723e-07, + "loss": 0.0728, + "step": 17790 + }, + { + "epoch": 7.45, + "learning_rate": 2.4674832004076166e-07, + "loss": 0.0472, + "step": 17791 + }, + { + "epoch": 7.45, + "learning_rate": 2.4637415933161155e-07, + "loss": 0.0512, + "step": 17792 + }, + { + "epoch": 7.45, + "learning_rate": 2.460002789817806e-07, + "loss": 0.0484, + "step": 17793 + }, + { + "epoch": 7.45, + "learning_rate": 2.4562667900201677e-07, + "loss": 0.0547, + "step": 17794 + }, + { + "epoch": 7.45, + "learning_rate": 2.4525335940305704e-07, + "loss": 0.0613, + "step": 17795 + }, + { + "epoch": 7.45, + "learning_rate": 2.44880320195634e-07, + "loss": 0.0636, + "step": 17796 + }, + { + "epoch": 7.45, + "learning_rate": 2.4450756139046906e-07, + "loss": 0.051, + "step": 17797 + }, + { + "epoch": 7.45, + "learning_rate": 2.4413508299827805e-07, + "loss": 0.0516, + "step": 17798 + }, + { + "epoch": 7.45, + "learning_rate": 2.4376288502976685e-07, + "loss": 0.0318, + "step": 17799 + }, + { + "epoch": 7.45, + "learning_rate": 2.433909674956336e-07, + "loss": 0.0468, + "step": 17800 + }, + { + "epoch": 7.45, + "learning_rate": 2.430193304065687e-07, + "loss": 0.0524, + "step": 17801 + }, + { + "epoch": 7.45, + "learning_rate": 2.4264797377325566e-07, + "loss": 0.0567, + "step": 17802 + }, + { + "epoch": 7.45, + "learning_rate": 2.422768976063683e-07, + "loss": 0.0401, + "step": 17803 + }, + { + "epoch": 7.45, + "learning_rate": 2.4190610191657246e-07, + "loss": 0.0437, + "step": 17804 + }, + { + "epoch": 7.45, + "learning_rate": 2.4153558671452524e-07, + "loss": 0.032, + "step": 17805 + }, + { + "epoch": 7.45, + "learning_rate": 2.411653520108803e-07, + "loss": 0.0429, + "step": 17806 + }, + { + "epoch": 7.46, + "learning_rate": 2.4079539781627693e-07, + "loss": 0.0478, + "step": 17807 + }, + { + "epoch": 7.46, + "learning_rate": 2.4042572414134877e-07, + "loss": 0.041, + "step": 17808 + }, + { + "epoch": 7.46, + "learning_rate": 2.4005633099672296e-07, + "loss": 0.045, + "step": 17809 + }, + { + "epoch": 7.46, + "learning_rate": 2.396872183930166e-07, + "loss": 0.0517, + "step": 17810 + }, + { + "epoch": 7.46, + "learning_rate": 2.3931838634083994e-07, + "loss": 0.0358, + "step": 17811 + }, + { + "epoch": 7.46, + "learning_rate": 2.3894983485079683e-07, + "loss": 0.0513, + "step": 17812 + }, + { + "epoch": 7.46, + "learning_rate": 2.3858156393347654e-07, + "loss": 0.0579, + "step": 17813 + }, + { + "epoch": 7.46, + "learning_rate": 2.3821357359946838e-07, + "loss": 0.0584, + "step": 17814 + }, + { + "epoch": 7.46, + "learning_rate": 2.3784586385934839e-07, + "loss": 0.0381, + "step": 17815 + }, + { + "epoch": 7.46, + "learning_rate": 2.374784347236858e-07, + "loss": 0.052, + "step": 17816 + }, + { + "epoch": 7.46, + "learning_rate": 2.3711128620304226e-07, + "loss": 0.046, + "step": 17817 + }, + { + "epoch": 7.46, + "learning_rate": 2.3674441830797147e-07, + "loss": 0.0637, + "step": 17818 + }, + { + "epoch": 7.46, + "learning_rate": 2.3637783104902057e-07, + "loss": 0.053, + "step": 17819 + }, + { + "epoch": 7.46, + "learning_rate": 2.3601152443672336e-07, + "loss": 0.0594, + "step": 17820 + }, + { + "epoch": 7.46, + "learning_rate": 2.3564549848161033e-07, + "loss": 0.0661, + "step": 17821 + }, + { + "epoch": 7.46, + "learning_rate": 2.3527975319420192e-07, + "loss": 0.0536, + "step": 17822 + }, + { + "epoch": 7.46, + "learning_rate": 2.349142885850131e-07, + "loss": 0.0418, + "step": 17823 + }, + { + "epoch": 7.46, + "learning_rate": 2.345491046645465e-07, + "loss": 0.05, + "step": 17824 + }, + { + "epoch": 7.46, + "learning_rate": 2.3418420144330268e-07, + "loss": 0.0583, + "step": 17825 + }, + { + "epoch": 7.46, + "learning_rate": 2.3381957893176543e-07, + "loss": 0.0828, + "step": 17826 + }, + { + "epoch": 7.46, + "learning_rate": 2.3345523714041973e-07, + "loss": 0.043, + "step": 17827 + }, + { + "epoch": 7.46, + "learning_rate": 2.330911760797361e-07, + "loss": 0.0615, + "step": 17828 + }, + { + "epoch": 7.46, + "learning_rate": 2.3272739576017945e-07, + "loss": 0.0439, + "step": 17829 + }, + { + "epoch": 7.46, + "learning_rate": 2.323638961922059e-07, + "loss": 0.0408, + "step": 17830 + }, + { + "epoch": 7.47, + "learning_rate": 2.3200067738626487e-07, + "loss": 0.0589, + "step": 17831 + }, + { + "epoch": 7.47, + "learning_rate": 2.3163773935279577e-07, + "loss": 0.0638, + "step": 17832 + }, + { + "epoch": 7.47, + "learning_rate": 2.3127508210223359e-07, + "loss": 0.0529, + "step": 17833 + }, + { + "epoch": 7.47, + "learning_rate": 2.309127056449989e-07, + "loss": 0.0459, + "step": 17834 + }, + { + "epoch": 7.47, + "learning_rate": 2.3055060999151003e-07, + "loss": 0.0486, + "step": 17835 + }, + { + "epoch": 7.47, + "learning_rate": 2.3018879515217418e-07, + "loss": 0.0327, + "step": 17836 + }, + { + "epoch": 7.47, + "learning_rate": 2.2982726113739195e-07, + "loss": 0.0491, + "step": 17837 + }, + { + "epoch": 7.47, + "learning_rate": 2.2946600795755502e-07, + "loss": 0.0493, + "step": 17838 + }, + { + "epoch": 7.47, + "learning_rate": 2.2910503562304843e-07, + "loss": 0.0534, + "step": 17839 + }, + { + "epoch": 7.47, + "learning_rate": 2.2874434414424606e-07, + "loss": 0.0509, + "step": 17840 + }, + { + "epoch": 7.47, + "learning_rate": 2.2838393353151745e-07, + "loss": 0.0737, + "step": 17841 + }, + { + "epoch": 7.47, + "learning_rate": 2.2802380379522093e-07, + "loss": 0.0368, + "step": 17842 + }, + { + "epoch": 7.47, + "learning_rate": 2.2766395494570825e-07, + "loss": 0.0512, + "step": 17843 + }, + { + "epoch": 7.47, + "learning_rate": 2.2730438699332446e-07, + "loss": 0.0422, + "step": 17844 + }, + { + "epoch": 7.47, + "learning_rate": 2.2694509994840353e-07, + "loss": 0.0436, + "step": 17845 + }, + { + "epoch": 7.47, + "learning_rate": 2.2658609382127384e-07, + "loss": 0.0599, + "step": 17846 + }, + { + "epoch": 7.47, + "learning_rate": 2.2622736862225382e-07, + "loss": 0.0471, + "step": 17847 + }, + { + "epoch": 7.47, + "learning_rate": 2.258689243616541e-07, + "loss": 0.0448, + "step": 17848 + }, + { + "epoch": 7.47, + "learning_rate": 2.2551076104977975e-07, + "loss": 0.0355, + "step": 17849 + }, + { + "epoch": 7.47, + "learning_rate": 2.251528786969237e-07, + "loss": 0.0359, + "step": 17850 + }, + { + "epoch": 7.47, + "learning_rate": 2.2479527731337437e-07, + "loss": 0.0344, + "step": 17851 + }, + { + "epoch": 7.47, + "learning_rate": 2.2443795690941127e-07, + "loss": 0.0577, + "step": 17852 + }, + { + "epoch": 7.47, + "learning_rate": 2.2408091749530402e-07, + "loss": 0.059, + "step": 17853 + }, + { + "epoch": 7.47, + "learning_rate": 2.237241590813155e-07, + "loss": 0.0708, + "step": 17854 + }, + { + "epoch": 7.48, + "learning_rate": 2.2336768167770084e-07, + "loss": 0.0472, + "step": 17855 + }, + { + "epoch": 7.48, + "learning_rate": 2.230114852947074e-07, + "loss": 0.0545, + "step": 17856 + }, + { + "epoch": 7.48, + "learning_rate": 2.2265556994257144e-07, + "loss": 0.0575, + "step": 17857 + }, + { + "epoch": 7.48, + "learning_rate": 2.2229993563152585e-07, + "loss": 0.0539, + "step": 17858 + }, + { + "epoch": 7.48, + "learning_rate": 2.2194458237179139e-07, + "loss": 0.0631, + "step": 17859 + }, + { + "epoch": 7.48, + "learning_rate": 2.2158951017358432e-07, + "loss": 0.0624, + "step": 17860 + }, + { + "epoch": 7.48, + "learning_rate": 2.2123471904710757e-07, + "loss": 0.0565, + "step": 17861 + }, + { + "epoch": 7.48, + "learning_rate": 2.20880209002563e-07, + "loss": 0.0399, + "step": 17862 + }, + { + "epoch": 7.48, + "learning_rate": 2.2052598005013803e-07, + "loss": 0.047, + "step": 17863 + }, + { + "epoch": 7.48, + "learning_rate": 2.2017203220001559e-07, + "loss": 0.0494, + "step": 17864 + }, + { + "epoch": 7.48, + "learning_rate": 2.1981836546236979e-07, + "loss": 0.0559, + "step": 17865 + }, + { + "epoch": 7.48, + "learning_rate": 2.1946497984736693e-07, + "loss": 0.0383, + "step": 17866 + }, + { + "epoch": 7.48, + "learning_rate": 2.1911187536516332e-07, + "loss": 0.0442, + "step": 17867 + }, + { + "epoch": 7.48, + "learning_rate": 2.1875905202590974e-07, + "loss": 0.0464, + "step": 17868 + }, + { + "epoch": 7.48, + "learning_rate": 2.1840650983974698e-07, + "loss": 0.0628, + "step": 17869 + }, + { + "epoch": 7.48, + "learning_rate": 2.1805424881681137e-07, + "loss": 0.036, + "step": 17870 + }, + { + "epoch": 7.48, + "learning_rate": 2.1770226896722369e-07, + "loss": 0.0673, + "step": 17871 + }, + { + "epoch": 7.48, + "learning_rate": 2.1735057030110362e-07, + "loss": 0.0502, + "step": 17872 + }, + { + "epoch": 7.48, + "learning_rate": 2.1699915282856087e-07, + "loss": 0.0514, + "step": 17873 + }, + { + "epoch": 7.48, + "learning_rate": 2.1664801655969735e-07, + "loss": 0.0612, + "step": 17874 + }, + { + "epoch": 7.48, + "learning_rate": 2.162971615046039e-07, + "loss": 0.048, + "step": 17875 + }, + { + "epoch": 7.48, + "learning_rate": 2.1594658767336685e-07, + "loss": 0.0522, + "step": 17876 + }, + { + "epoch": 7.48, + "learning_rate": 2.155962950760626e-07, + "loss": 0.0434, + "step": 17877 + }, + { + "epoch": 7.49, + "learning_rate": 2.152462837227598e-07, + "loss": 0.0527, + "step": 17878 + }, + { + "epoch": 7.49, + "learning_rate": 2.1489655362352034e-07, + "loss": 0.048, + "step": 17879 + }, + { + "epoch": 7.49, + "learning_rate": 2.1454710478839736e-07, + "loss": 0.0406, + "step": 17880 + }, + { + "epoch": 7.49, + "learning_rate": 2.1419793722743388e-07, + "loss": 0.044, + "step": 17881 + }, + { + "epoch": 7.49, + "learning_rate": 2.1384905095066634e-07, + "loss": 0.0478, + "step": 17882 + }, + { + "epoch": 7.49, + "learning_rate": 2.1350044596812335e-07, + "loss": 0.0491, + "step": 17883 + }, + { + "epoch": 7.49, + "learning_rate": 2.1315212228982584e-07, + "loss": 0.0352, + "step": 17884 + }, + { + "epoch": 7.49, + "learning_rate": 2.1280407992578578e-07, + "loss": 0.0437, + "step": 17885 + }, + { + "epoch": 7.49, + "learning_rate": 2.1245631888600848e-07, + "loss": 0.0512, + "step": 17886 + }, + { + "epoch": 7.49, + "learning_rate": 2.121088391804893e-07, + "loss": 0.0427, + "step": 17887 + }, + { + "epoch": 7.49, + "learning_rate": 2.117616408192147e-07, + "loss": 0.0554, + "step": 17888 + }, + { + "epoch": 7.49, + "learning_rate": 2.1141472381216554e-07, + "loss": 0.037, + "step": 17889 + }, + { + "epoch": 7.49, + "learning_rate": 2.11068088169315e-07, + "loss": 0.0567, + "step": 17890 + }, + { + "epoch": 7.49, + "learning_rate": 2.107217339006251e-07, + "loss": 0.0412, + "step": 17891 + }, + { + "epoch": 7.49, + "learning_rate": 2.103756610160512e-07, + "loss": 0.0677, + "step": 17892 + }, + { + "epoch": 7.49, + "learning_rate": 2.1002986952554428e-07, + "loss": 0.0494, + "step": 17893 + }, + { + "epoch": 7.49, + "learning_rate": 2.0968435943904075e-07, + "loss": 0.0517, + "step": 17894 + }, + { + "epoch": 7.49, + "learning_rate": 2.093391307664716e-07, + "loss": 0.0252, + "step": 17895 + }, + { + "epoch": 7.49, + "learning_rate": 2.0899418351776e-07, + "loss": 0.0519, + "step": 17896 + }, + { + "epoch": 7.49, + "learning_rate": 2.086495177028236e-07, + "loss": 0.0646, + "step": 17897 + }, + { + "epoch": 7.49, + "learning_rate": 2.0830513333156776e-07, + "loss": 0.0583, + "step": 17898 + }, + { + "epoch": 7.49, + "learning_rate": 2.0796103041389126e-07, + "loss": 0.0538, + "step": 17899 + }, + { + "epoch": 7.49, + "learning_rate": 2.0761720895968617e-07, + "loss": 0.0497, + "step": 17900 + }, + { + "epoch": 7.49, + "learning_rate": 2.0727366897883571e-07, + "loss": 0.0612, + "step": 17901 + }, + { + "epoch": 7.5, + "learning_rate": 2.069304104812131e-07, + "loss": 0.0552, + "step": 17902 + }, + { + "epoch": 7.5, + "learning_rate": 2.0658743347668487e-07, + "loss": 0.0488, + "step": 17903 + }, + { + "epoch": 7.5, + "learning_rate": 2.062447379751109e-07, + "loss": 0.0491, + "step": 17904 + }, + { + "epoch": 7.5, + "learning_rate": 2.0590232398634114e-07, + "loss": 0.0561, + "step": 17905 + }, + { + "epoch": 7.5, + "learning_rate": 2.0556019152021767e-07, + "loss": 0.0515, + "step": 17906 + }, + { + "epoch": 7.5, + "learning_rate": 2.0521834058657597e-07, + "loss": 0.0684, + "step": 17907 + }, + { + "epoch": 7.5, + "learning_rate": 2.0487677119524152e-07, + "loss": 0.043, + "step": 17908 + }, + { + "epoch": 7.5, + "learning_rate": 2.045354833560309e-07, + "loss": 0.0524, + "step": 17909 + }, + { + "epoch": 7.5, + "learning_rate": 2.0419447707875518e-07, + "loss": 0.053, + "step": 17910 + }, + { + "epoch": 7.5, + "learning_rate": 2.0385375237321759e-07, + "loss": 0.0464, + "step": 17911 + }, + { + "epoch": 7.5, + "learning_rate": 2.0351330924921031e-07, + "loss": 0.0389, + "step": 17912 + }, + { + "epoch": 7.5, + "learning_rate": 2.0317314771651996e-07, + "loss": 0.0343, + "step": 17913 + }, + { + "epoch": 7.5, + "learning_rate": 2.028332677849254e-07, + "loss": 0.0581, + "step": 17914 + }, + { + "epoch": 7.5, + "learning_rate": 2.0249366946419436e-07, + "loss": 0.0411, + "step": 17915 + }, + { + "epoch": 7.5, + "learning_rate": 2.021543527640879e-07, + "loss": 0.0523, + "step": 17916 + }, + { + "epoch": 7.5, + "learning_rate": 2.0181531769435936e-07, + "loss": 0.0686, + "step": 17917 + }, + { + "epoch": 7.5, + "learning_rate": 2.0147656426475537e-07, + "loss": 0.0555, + "step": 17918 + }, + { + "epoch": 7.5, + "learning_rate": 2.011380924850115e-07, + "loss": 0.0443, + "step": 17919 + }, + { + "epoch": 7.5, + "learning_rate": 2.007999023648588e-07, + "loss": 0.0613, + "step": 17920 + }, + { + "epoch": 7.5, + "learning_rate": 2.0046199391401844e-07, + "loss": 0.039, + "step": 17921 + }, + { + "epoch": 7.5, + "learning_rate": 2.001243671422004e-07, + "loss": 0.0493, + "step": 17922 + }, + { + "epoch": 7.5, + "learning_rate": 1.9978702205911137e-07, + "loss": 0.0779, + "step": 17923 + }, + { + "epoch": 7.5, + "learning_rate": 1.9944995867444806e-07, + "loss": 0.0601, + "step": 17924 + }, + { + "epoch": 7.5, + "learning_rate": 1.9911317699789822e-07, + "loss": 0.0622, + "step": 17925 + }, + { + "epoch": 7.51, + "learning_rate": 1.9877667703914417e-07, + "loss": 0.0615, + "step": 17926 + }, + { + "epoch": 7.51, + "learning_rate": 1.984404588078559e-07, + "loss": 0.0469, + "step": 17927 + }, + { + "epoch": 7.51, + "learning_rate": 1.9810452231370013e-07, + "loss": 0.034, + "step": 17928 + }, + { + "epoch": 7.51, + "learning_rate": 1.977688675663314e-07, + "loss": 0.0664, + "step": 17929 + }, + { + "epoch": 7.51, + "learning_rate": 1.974334945753975e-07, + "loss": 0.0598, + "step": 17930 + }, + { + "epoch": 7.51, + "learning_rate": 1.9709840335053965e-07, + "loss": 0.0547, + "step": 17931 + }, + { + "epoch": 7.51, + "learning_rate": 1.96763593901389e-07, + "loss": 0.0456, + "step": 17932 + }, + { + "epoch": 7.51, + "learning_rate": 1.9642906623757008e-07, + "loss": 0.0449, + "step": 17933 + }, + { + "epoch": 7.51, + "learning_rate": 1.9609482036869853e-07, + "loss": 0.0373, + "step": 17934 + }, + { + "epoch": 7.51, + "learning_rate": 1.9576085630438114e-07, + "loss": 0.0501, + "step": 17935 + }, + { + "epoch": 7.51, + "learning_rate": 1.9542717405421796e-07, + "loss": 0.0594, + "step": 17936 + }, + { + "epoch": 7.51, + "learning_rate": 1.9509377362779913e-07, + "loss": 0.0462, + "step": 17937 + }, + { + "epoch": 7.51, + "learning_rate": 1.9476065503471032e-07, + "loss": 0.0492, + "step": 17938 + }, + { + "epoch": 7.51, + "learning_rate": 1.9442781828452496e-07, + "loss": 0.0553, + "step": 17939 + }, + { + "epoch": 7.51, + "learning_rate": 1.9409526338681096e-07, + "loss": 0.0564, + "step": 17940 + }, + { + "epoch": 7.51, + "learning_rate": 1.9376299035112623e-07, + "loss": 0.069, + "step": 17941 + }, + { + "epoch": 7.51, + "learning_rate": 1.9343099918702424e-07, + "loss": 0.048, + "step": 17942 + }, + { + "epoch": 7.51, + "learning_rate": 1.9309928990404403e-07, + "loss": 0.0558, + "step": 17943 + }, + { + "epoch": 7.51, + "learning_rate": 1.927678625117224e-07, + "loss": 0.0586, + "step": 17944 + }, + { + "epoch": 7.51, + "learning_rate": 1.924367170195862e-07, + "loss": 0.0553, + "step": 17945 + }, + { + "epoch": 7.51, + "learning_rate": 1.9210585343715337e-07, + "loss": 0.0448, + "step": 17946 + }, + { + "epoch": 7.51, + "learning_rate": 1.9177527177393518e-07, + "loss": 0.0435, + "step": 17947 + }, + { + "epoch": 7.51, + "learning_rate": 1.914449720394329e-07, + "loss": 0.0372, + "step": 17948 + }, + { + "epoch": 7.51, + "learning_rate": 1.9111495424314009e-07, + "loss": 0.0541, + "step": 17949 + }, + { + "epoch": 7.52, + "learning_rate": 1.9078521839454466e-07, + "loss": 0.058, + "step": 17950 + }, + { + "epoch": 7.52, + "learning_rate": 1.9045576450312242e-07, + "loss": 0.039, + "step": 17951 + }, + { + "epoch": 7.52, + "learning_rate": 1.9012659257834464e-07, + "loss": 0.0564, + "step": 17952 + }, + { + "epoch": 7.52, + "learning_rate": 1.8979770262967378e-07, + "loss": 0.0444, + "step": 17953 + }, + { + "epoch": 7.52, + "learning_rate": 1.8946909466656115e-07, + "loss": 0.0515, + "step": 17954 + }, + { + "epoch": 7.52, + "learning_rate": 1.8914076869845587e-07, + "loss": 0.0674, + "step": 17955 + }, + { + "epoch": 7.52, + "learning_rate": 1.8881272473479152e-07, + "loss": 0.0523, + "step": 17956 + }, + { + "epoch": 7.52, + "learning_rate": 1.8848496278499938e-07, + "loss": 0.056, + "step": 17957 + }, + { + "epoch": 7.52, + "learning_rate": 1.8815748285849977e-07, + "loss": 0.0546, + "step": 17958 + }, + { + "epoch": 7.52, + "learning_rate": 1.8783028496470735e-07, + "loss": 0.0405, + "step": 17959 + }, + { + "epoch": 7.52, + "learning_rate": 1.875033691130257e-07, + "loss": 0.0504, + "step": 17960 + }, + { + "epoch": 7.52, + "learning_rate": 1.8717673531285398e-07, + "loss": 0.0564, + "step": 17961 + }, + { + "epoch": 7.52, + "learning_rate": 1.86850383573578e-07, + "loss": 0.0531, + "step": 17962 + }, + { + "epoch": 7.52, + "learning_rate": 1.8652431390457916e-07, + "loss": 0.0326, + "step": 17963 + }, + { + "epoch": 7.52, + "learning_rate": 1.8619852631523105e-07, + "loss": 0.0476, + "step": 17964 + }, + { + "epoch": 7.52, + "learning_rate": 1.8587302081489734e-07, + "loss": 0.0564, + "step": 17965 + }, + { + "epoch": 7.52, + "learning_rate": 1.8554779741293493e-07, + "loss": 0.0391, + "step": 17966 + }, + { + "epoch": 7.52, + "learning_rate": 1.8522285611869194e-07, + "loss": 0.0564, + "step": 17967 + }, + { + "epoch": 7.52, + "learning_rate": 1.848981969415087e-07, + "loss": 0.0552, + "step": 17968 + }, + { + "epoch": 7.52, + "learning_rate": 1.8457381989071767e-07, + "loss": 0.0359, + "step": 17969 + }, + { + "epoch": 7.52, + "learning_rate": 1.8424972497564143e-07, + "loss": 0.0379, + "step": 17970 + }, + { + "epoch": 7.52, + "learning_rate": 1.8392591220559586e-07, + "loss": 0.0391, + "step": 17971 + }, + { + "epoch": 7.52, + "learning_rate": 1.836023815898902e-07, + "loss": 0.0349, + "step": 17972 + }, + { + "epoch": 7.52, + "learning_rate": 1.8327913313782252e-07, + "loss": 0.0289, + "step": 17973 + }, + { + "epoch": 7.53, + "learning_rate": 1.8295616685868434e-07, + "loss": 0.0499, + "step": 17974 + }, + { + "epoch": 7.53, + "learning_rate": 1.8263348276176152e-07, + "loss": 0.063, + "step": 17975 + }, + { + "epoch": 7.53, + "learning_rate": 1.8231108085632664e-07, + "loss": 0.0428, + "step": 17976 + }, + { + "epoch": 7.53, + "learning_rate": 1.8198896115164676e-07, + "loss": 0.0729, + "step": 17977 + }, + { + "epoch": 7.53, + "learning_rate": 1.8166712365698335e-07, + "loss": 0.0325, + "step": 17978 + }, + { + "epoch": 7.53, + "learning_rate": 1.8134556838158457e-07, + "loss": 0.038, + "step": 17979 + }, + { + "epoch": 7.53, + "learning_rate": 1.8102429533469523e-07, + "loss": 0.0474, + "step": 17980 + }, + { + "epoch": 7.53, + "learning_rate": 1.8070330452554907e-07, + "loss": 0.0502, + "step": 17981 + }, + { + "epoch": 7.53, + "learning_rate": 1.8038259596337316e-07, + "loss": 0.0523, + "step": 17982 + }, + { + "epoch": 7.53, + "learning_rate": 1.8006216965738676e-07, + "loss": 0.0457, + "step": 17983 + }, + { + "epoch": 7.53, + "learning_rate": 1.797420256167981e-07, + "loss": 0.0496, + "step": 17984 + }, + { + "epoch": 7.53, + "learning_rate": 1.794221638508098e-07, + "loss": 0.051, + "step": 17985 + }, + { + "epoch": 7.53, + "learning_rate": 1.7910258436861782e-07, + "loss": 0.0541, + "step": 17986 + }, + { + "epoch": 7.53, + "learning_rate": 1.7878328717940597e-07, + "loss": 0.0444, + "step": 17987 + }, + { + "epoch": 7.53, + "learning_rate": 1.7846427229235462e-07, + "loss": 0.0563, + "step": 17988 + }, + { + "epoch": 7.53, + "learning_rate": 1.7814553971663207e-07, + "loss": 0.0434, + "step": 17989 + }, + { + "epoch": 7.53, + "learning_rate": 1.778270894613998e-07, + "loss": 0.0503, + "step": 17990 + }, + { + "epoch": 7.53, + "learning_rate": 1.7750892153581278e-07, + "loss": 0.0337, + "step": 17991 + }, + { + "epoch": 7.53, + "learning_rate": 1.7719103594901366e-07, + "loss": 0.0494, + "step": 17992 + }, + { + "epoch": 7.53, + "learning_rate": 1.768734327101429e-07, + "loss": 0.0563, + "step": 17993 + }, + { + "epoch": 7.53, + "learning_rate": 1.7655611182832767e-07, + "loss": 0.0568, + "step": 17994 + }, + { + "epoch": 7.53, + "learning_rate": 1.7623907331269174e-07, + "loss": 0.0511, + "step": 17995 + }, + { + "epoch": 7.53, + "learning_rate": 1.7592231717234454e-07, + "loss": 0.0627, + "step": 17996 + }, + { + "epoch": 7.53, + "learning_rate": 1.7560584341639208e-07, + "loss": 0.0462, + "step": 17997 + }, + { + "epoch": 7.54, + "learning_rate": 1.7528965205393266e-07, + "loss": 0.0687, + "step": 17998 + }, + { + "epoch": 7.54, + "learning_rate": 1.7497374309405346e-07, + "loss": 0.0531, + "step": 17999 + }, + { + "epoch": 7.54, + "learning_rate": 1.7465811654583608e-07, + "loss": 0.0545, + "step": 18000 + }, + { + "epoch": 7.54, + "learning_rate": 1.7434277241835106e-07, + "loss": 0.0449, + "step": 18001 + }, + { + "epoch": 7.54, + "learning_rate": 1.740277107206656e-07, + "loss": 0.0538, + "step": 18002 + }, + { + "epoch": 7.54, + "learning_rate": 1.7371293146183355e-07, + "loss": 0.041, + "step": 18003 + }, + { + "epoch": 7.54, + "learning_rate": 1.7339843465090322e-07, + "loss": 0.0338, + "step": 18004 + }, + { + "epoch": 7.54, + "learning_rate": 1.730842202969152e-07, + "loss": 0.0491, + "step": 18005 + }, + { + "epoch": 7.54, + "learning_rate": 1.727702884089011e-07, + "loss": 0.0327, + "step": 18006 + }, + { + "epoch": 7.54, + "learning_rate": 1.7245663899588372e-07, + "loss": 0.0536, + "step": 18007 + }, + { + "epoch": 7.54, + "learning_rate": 1.721432720668803e-07, + "loss": 0.0525, + "step": 18008 + }, + { + "epoch": 7.54, + "learning_rate": 1.71830187630897e-07, + "loss": 0.0436, + "step": 18009 + }, + { + "epoch": 7.54, + "learning_rate": 1.7151738569693432e-07, + "loss": 0.0405, + "step": 18010 + }, + { + "epoch": 7.54, + "learning_rate": 1.712048662739818e-07, + "loss": 0.0548, + "step": 18011 + }, + { + "epoch": 7.54, + "learning_rate": 1.7089262937102446e-07, + "loss": 0.0562, + "step": 18012 + }, + { + "epoch": 7.54, + "learning_rate": 1.7058067499703512e-07, + "loss": 0.0486, + "step": 18013 + }, + { + "epoch": 7.54, + "learning_rate": 1.7026900316098217e-07, + "loss": 0.0345, + "step": 18014 + }, + { + "epoch": 7.54, + "learning_rate": 1.6995761387182398e-07, + "loss": 0.0416, + "step": 18015 + }, + { + "epoch": 7.54, + "learning_rate": 1.6964650713851228e-07, + "loss": 0.0512, + "step": 18016 + }, + { + "epoch": 7.54, + "learning_rate": 1.693356829699877e-07, + "loss": 0.0593, + "step": 18017 + }, + { + "epoch": 7.54, + "learning_rate": 1.6902514137518422e-07, + "loss": 0.0458, + "step": 18018 + }, + { + "epoch": 7.54, + "learning_rate": 1.6871488236303023e-07, + "loss": 0.0503, + "step": 18019 + }, + { + "epoch": 7.54, + "learning_rate": 1.684049059424431e-07, + "loss": 0.0647, + "step": 18020 + }, + { + "epoch": 7.54, + "learning_rate": 1.6809521212233114e-07, + "loss": 0.0345, + "step": 18021 + }, + { + "epoch": 7.55, + "learning_rate": 1.6778580091159958e-07, + "loss": 0.044, + "step": 18022 + }, + { + "epoch": 7.55, + "learning_rate": 1.67476672319139e-07, + "loss": 0.055, + "step": 18023 + }, + { + "epoch": 7.55, + "learning_rate": 1.6716782635383677e-07, + "loss": 0.0437, + "step": 18024 + }, + { + "epoch": 7.55, + "learning_rate": 1.6685926302456912e-07, + "loss": 0.0442, + "step": 18025 + }, + { + "epoch": 7.55, + "learning_rate": 1.6655098234020671e-07, + "loss": 0.0404, + "step": 18026 + }, + { + "epoch": 7.55, + "learning_rate": 1.6624298430961027e-07, + "loss": 0.0417, + "step": 18027 + }, + { + "epoch": 7.55, + "learning_rate": 1.659352689416338e-07, + "loss": 0.0393, + "step": 18028 + }, + { + "epoch": 7.55, + "learning_rate": 1.6562783624512024e-07, + "loss": 0.0504, + "step": 18029 + }, + { + "epoch": 7.55, + "learning_rate": 1.6532068622890917e-07, + "loss": 0.0627, + "step": 18030 + }, + { + "epoch": 7.55, + "learning_rate": 1.6501381890182576e-07, + "loss": 0.0476, + "step": 18031 + }, + { + "epoch": 7.55, + "learning_rate": 1.6470723427269407e-07, + "loss": 0.0455, + "step": 18032 + }, + { + "epoch": 7.55, + "learning_rate": 1.6440093235032484e-07, + "loss": 0.0615, + "step": 18033 + }, + { + "epoch": 7.55, + "learning_rate": 1.640949131435221e-07, + "loss": 0.0529, + "step": 18034 + }, + { + "epoch": 7.55, + "learning_rate": 1.637891766610833e-07, + "loss": 0.0462, + "step": 18035 + }, + { + "epoch": 7.55, + "learning_rate": 1.6348372291179693e-07, + "loss": 0.0567, + "step": 18036 + }, + { + "epoch": 7.55, + "learning_rate": 1.6317855190444155e-07, + "loss": 0.0532, + "step": 18037 + }, + { + "epoch": 7.55, + "learning_rate": 1.6287366364779013e-07, + "loss": 0.0371, + "step": 18038 + }, + { + "epoch": 7.55, + "learning_rate": 1.6256905815060453e-07, + "loss": 0.0486, + "step": 18039 + }, + { + "epoch": 7.55, + "learning_rate": 1.6226473542164334e-07, + "loss": 0.0571, + "step": 18040 + }, + { + "epoch": 7.55, + "learning_rate": 1.6196069546965177e-07, + "loss": 0.0442, + "step": 18041 + }, + { + "epoch": 7.55, + "learning_rate": 1.6165693830336947e-07, + "loss": 0.0464, + "step": 18042 + }, + { + "epoch": 7.55, + "learning_rate": 1.613534639315295e-07, + "loss": 0.0358, + "step": 18043 + }, + { + "epoch": 7.55, + "learning_rate": 1.6105027236285265e-07, + "loss": 0.0372, + "step": 18044 + }, + { + "epoch": 7.55, + "learning_rate": 1.6074736360605414e-07, + "loss": 0.0567, + "step": 18045 + }, + { + "epoch": 7.56, + "learning_rate": 1.6044473766984258e-07, + "loss": 0.0631, + "step": 18046 + }, + { + "epoch": 7.56, + "learning_rate": 1.6014239456291436e-07, + "loss": 0.0556, + "step": 18047 + }, + { + "epoch": 7.56, + "learning_rate": 1.598403342939625e-07, + "loss": 0.0246, + "step": 18048 + }, + { + "epoch": 7.56, + "learning_rate": 1.5953855687166786e-07, + "loss": 0.0405, + "step": 18049 + }, + { + "epoch": 7.56, + "learning_rate": 1.592370623047046e-07, + "loss": 0.0566, + "step": 18050 + }, + { + "epoch": 7.56, + "learning_rate": 1.5893585060174134e-07, + "loss": 0.0605, + "step": 18051 + }, + { + "epoch": 7.56, + "learning_rate": 1.5863492177143226e-07, + "loss": 0.0511, + "step": 18052 + }, + { + "epoch": 7.56, + "learning_rate": 1.5833427582242933e-07, + "loss": 0.0528, + "step": 18053 + }, + { + "epoch": 7.56, + "learning_rate": 1.5803391276337565e-07, + "loss": 0.0491, + "step": 18054 + }, + { + "epoch": 7.56, + "learning_rate": 1.577338326029021e-07, + "loss": 0.0637, + "step": 18055 + }, + { + "epoch": 7.56, + "learning_rate": 1.574340353496362e-07, + "loss": 0.0585, + "step": 18056 + }, + { + "epoch": 7.56, + "learning_rate": 1.571345210121955e-07, + "loss": 0.0347, + "step": 18057 + }, + { + "epoch": 7.56, + "learning_rate": 1.568352895991876e-07, + "loss": 0.0553, + "step": 18058 + }, + { + "epoch": 7.56, + "learning_rate": 1.5653634111921445e-07, + "loss": 0.0647, + "step": 18059 + }, + { + "epoch": 7.56, + "learning_rate": 1.5623767558087033e-07, + "loss": 0.0487, + "step": 18060 + }, + { + "epoch": 7.56, + "learning_rate": 1.5593929299273837e-07, + "loss": 0.036, + "step": 18061 + }, + { + "epoch": 7.56, + "learning_rate": 1.556411933633961e-07, + "loss": 0.051, + "step": 18062 + }, + { + "epoch": 7.56, + "learning_rate": 1.5534337670141232e-07, + "loss": 0.0351, + "step": 18063 + }, + { + "epoch": 7.56, + "learning_rate": 1.5504584301534677e-07, + "loss": 0.062, + "step": 18064 + }, + { + "epoch": 7.56, + "learning_rate": 1.547485923137515e-07, + "loss": 0.0436, + "step": 18065 + }, + { + "epoch": 7.56, + "learning_rate": 1.5445162460517193e-07, + "loss": 0.0507, + "step": 18066 + }, + { + "epoch": 7.56, + "learning_rate": 1.5415493989814457e-07, + "loss": 0.042, + "step": 18067 + }, + { + "epoch": 7.56, + "learning_rate": 1.5385853820119478e-07, + "loss": 0.0687, + "step": 18068 + }, + { + "epoch": 7.56, + "learning_rate": 1.535624195228458e-07, + "loss": 0.0564, + "step": 18069 + }, + { + "epoch": 7.57, + "learning_rate": 1.5326658387160631e-07, + "loss": 0.0459, + "step": 18070 + }, + { + "epoch": 7.57, + "learning_rate": 1.5297103125598178e-07, + "loss": 0.0247, + "step": 18071 + }, + { + "epoch": 7.57, + "learning_rate": 1.526757616844665e-07, + "loss": 0.0634, + "step": 18072 + }, + { + "epoch": 7.57, + "learning_rate": 1.5238077516554817e-07, + "loss": 0.0413, + "step": 18073 + }, + { + "epoch": 7.57, + "learning_rate": 1.520860717077055e-07, + "loss": 0.0561, + "step": 18074 + }, + { + "epoch": 7.57, + "learning_rate": 1.5179165131941064e-07, + "loss": 0.0685, + "step": 18075 + }, + { + "epoch": 7.57, + "learning_rate": 1.514975140091246e-07, + "loss": 0.0342, + "step": 18076 + }, + { + "epoch": 7.57, + "learning_rate": 1.5120365978530393e-07, + "loss": 0.0438, + "step": 18077 + }, + { + "epoch": 7.57, + "learning_rate": 1.5091008865639412e-07, + "loss": 0.0521, + "step": 18078 + }, + { + "epoch": 7.57, + "learning_rate": 1.5061680063083284e-07, + "loss": 0.0508, + "step": 18079 + }, + { + "epoch": 7.57, + "learning_rate": 1.5032379571705224e-07, + "loss": 0.0525, + "step": 18080 + }, + { + "epoch": 7.57, + "learning_rate": 1.500310739234745e-07, + "loss": 0.0611, + "step": 18081 + }, + { + "epoch": 7.57, + "learning_rate": 1.4973863525851174e-07, + "loss": 0.0542, + "step": 18082 + }, + { + "epoch": 7.57, + "learning_rate": 1.494464797305706e-07, + "loss": 0.0486, + "step": 18083 + }, + { + "epoch": 7.57, + "learning_rate": 1.49154607348051e-07, + "loss": 0.0527, + "step": 18084 + }, + { + "epoch": 7.57, + "learning_rate": 1.4886301811933957e-07, + "loss": 0.0372, + "step": 18085 + }, + { + "epoch": 7.57, + "learning_rate": 1.4857171205281852e-07, + "loss": 0.0417, + "step": 18086 + }, + { + "epoch": 7.57, + "learning_rate": 1.4828068915686223e-07, + "loss": 0.0694, + "step": 18087 + }, + { + "epoch": 7.57, + "learning_rate": 1.4798994943983514e-07, + "loss": 0.0533, + "step": 18088 + }, + { + "epoch": 7.57, + "learning_rate": 1.4769949291009388e-07, + "loss": 0.0445, + "step": 18089 + }, + { + "epoch": 7.57, + "learning_rate": 1.474093195759896e-07, + "loss": 0.0391, + "step": 18090 + }, + { + "epoch": 7.57, + "learning_rate": 1.4711942944586e-07, + "loss": 0.0347, + "step": 18091 + }, + { + "epoch": 7.57, + "learning_rate": 1.4682982252803845e-07, + "loss": 0.0599, + "step": 18092 + }, + { + "epoch": 7.58, + "learning_rate": 1.4654049883085165e-07, + "loss": 0.0418, + "step": 18093 + }, + { + "epoch": 7.58, + "learning_rate": 1.462514583626129e-07, + "loss": 0.0516, + "step": 18094 + }, + { + "epoch": 7.58, + "learning_rate": 1.4596270113163336e-07, + "loss": 0.0362, + "step": 18095 + }, + { + "epoch": 7.58, + "learning_rate": 1.4567422714621193e-07, + "loss": 0.0384, + "step": 18096 + }, + { + "epoch": 7.58, + "learning_rate": 1.4538603641463867e-07, + "loss": 0.0628, + "step": 18097 + }, + { + "epoch": 7.58, + "learning_rate": 1.4509812894520026e-07, + "loss": 0.0494, + "step": 18098 + }, + { + "epoch": 7.58, + "learning_rate": 1.4481050474617008e-07, + "loss": 0.0564, + "step": 18099 + }, + { + "epoch": 7.58, + "learning_rate": 1.4452316382581711e-07, + "loss": 0.0468, + "step": 18100 + }, + { + "epoch": 7.58, + "learning_rate": 1.4423610619240026e-07, + "loss": 0.0259, + "step": 18101 + }, + { + "epoch": 7.58, + "learning_rate": 1.439493318541707e-07, + "loss": 0.066, + "step": 18102 + }, + { + "epoch": 7.58, + "learning_rate": 1.4366284081937077e-07, + "loss": 0.0486, + "step": 18103 + }, + { + "epoch": 7.58, + "learning_rate": 1.4337663309623718e-07, + "loss": 0.0437, + "step": 18104 + }, + { + "epoch": 7.58, + "learning_rate": 1.4309070869299558e-07, + "loss": 0.0525, + "step": 18105 + }, + { + "epoch": 7.58, + "learning_rate": 1.4280506761786383e-07, + "loss": 0.0565, + "step": 18106 + }, + { + "epoch": 7.58, + "learning_rate": 1.4251970987905316e-07, + "loss": 0.0485, + "step": 18107 + }, + { + "epoch": 7.58, + "learning_rate": 1.4223463548476702e-07, + "loss": 0.0419, + "step": 18108 + }, + { + "epoch": 7.58, + "learning_rate": 1.419498444431977e-07, + "loss": 0.04, + "step": 18109 + }, + { + "epoch": 7.58, + "learning_rate": 1.4166533676253315e-07, + "loss": 0.064, + "step": 18110 + }, + { + "epoch": 7.58, + "learning_rate": 1.413811124509501e-07, + "loss": 0.0479, + "step": 18111 + }, + { + "epoch": 7.58, + "learning_rate": 1.4109717151661873e-07, + "loss": 0.0364, + "step": 18112 + }, + { + "epoch": 7.58, + "learning_rate": 1.4081351396770027e-07, + "loss": 0.0612, + "step": 18113 + }, + { + "epoch": 7.58, + "learning_rate": 1.405301398123482e-07, + "loss": 0.0357, + "step": 18114 + }, + { + "epoch": 7.58, + "learning_rate": 1.4024704905870824e-07, + "loss": 0.0589, + "step": 18115 + }, + { + "epoch": 7.58, + "learning_rate": 1.399642417149172e-07, + "loss": 0.0462, + "step": 18116 + }, + { + "epoch": 7.59, + "learning_rate": 1.3968171778910632e-07, + "loss": 0.0495, + "step": 18117 + }, + { + "epoch": 7.59, + "learning_rate": 1.393994772893925e-07, + "loss": 0.0445, + "step": 18118 + }, + { + "epoch": 7.59, + "learning_rate": 1.3911752022389148e-07, + "loss": 0.0511, + "step": 18119 + }, + { + "epoch": 7.59, + "learning_rate": 1.388358466007067e-07, + "loss": 0.0437, + "step": 18120 + }, + { + "epoch": 7.59, + "learning_rate": 1.3855445642793507e-07, + "loss": 0.0508, + "step": 18121 + }, + { + "epoch": 7.59, + "learning_rate": 1.3827334971366346e-07, + "loss": 0.0594, + "step": 18122 + }, + { + "epoch": 7.59, + "learning_rate": 1.3799252646597428e-07, + "loss": 0.0374, + "step": 18123 + }, + { + "epoch": 7.59, + "learning_rate": 1.3771198669293885e-07, + "loss": 0.0643, + "step": 18124 + }, + { + "epoch": 7.59, + "learning_rate": 1.374317304026207e-07, + "loss": 0.0486, + "step": 18125 + }, + { + "epoch": 7.59, + "learning_rate": 1.3715175760307454e-07, + "loss": 0.0671, + "step": 18126 + }, + { + "epoch": 7.59, + "learning_rate": 1.3687206830235055e-07, + "loss": 0.0558, + "step": 18127 + }, + { + "epoch": 7.59, + "learning_rate": 1.3659266250848457e-07, + "loss": 0.0604, + "step": 18128 + }, + { + "epoch": 7.59, + "learning_rate": 1.3631354022951126e-07, + "loss": 0.0443, + "step": 18129 + }, + { + "epoch": 7.59, + "learning_rate": 1.36034701473452e-07, + "loss": 0.0468, + "step": 18130 + }, + { + "epoch": 7.59, + "learning_rate": 1.3575614624832257e-07, + "loss": 0.0434, + "step": 18131 + }, + { + "epoch": 7.59, + "learning_rate": 1.354778745621288e-07, + "loss": 0.0534, + "step": 18132 + }, + { + "epoch": 7.59, + "learning_rate": 1.3519988642286985e-07, + "loss": 0.0518, + "step": 18133 + }, + { + "epoch": 7.59, + "learning_rate": 1.3492218183853712e-07, + "loss": 0.0427, + "step": 18134 + }, + { + "epoch": 7.59, + "learning_rate": 1.3464476081711088e-07, + "loss": 0.0434, + "step": 18135 + }, + { + "epoch": 7.59, + "learning_rate": 1.3436762336656694e-07, + "loss": 0.05, + "step": 18136 + }, + { + "epoch": 7.59, + "learning_rate": 1.340907694948712e-07, + "loss": 0.0357, + "step": 18137 + }, + { + "epoch": 7.59, + "learning_rate": 1.3381419920998285e-07, + "loss": 0.0482, + "step": 18138 + }, + { + "epoch": 7.59, + "learning_rate": 1.3353791251984883e-07, + "loss": 0.0474, + "step": 18139 + }, + { + "epoch": 7.59, + "learning_rate": 1.332619094324128e-07, + "loss": 0.0785, + "step": 18140 + }, + { + "epoch": 7.6, + "learning_rate": 1.3298618995560618e-07, + "loss": 0.0443, + "step": 18141 + }, + { + "epoch": 7.6, + "learning_rate": 1.3271075409735824e-07, + "loss": 0.0615, + "step": 18142 + }, + { + "epoch": 7.6, + "learning_rate": 1.3243560186558148e-07, + "loss": 0.052, + "step": 18143 + }, + { + "epoch": 7.6, + "learning_rate": 1.3216073326818736e-07, + "loss": 0.0585, + "step": 18144 + }, + { + "epoch": 7.6, + "learning_rate": 1.3188614831307733e-07, + "loss": 0.0568, + "step": 18145 + }, + { + "epoch": 7.6, + "learning_rate": 1.316118470081418e-07, + "loss": 0.0485, + "step": 18146 + }, + { + "epoch": 7.6, + "learning_rate": 1.313378293612677e-07, + "loss": 0.0432, + "step": 18147 + }, + { + "epoch": 7.6, + "learning_rate": 1.3106409538032993e-07, + "loss": 0.0566, + "step": 18148 + }, + { + "epoch": 7.6, + "learning_rate": 1.307906450731977e-07, + "loss": 0.0438, + "step": 18149 + }, + { + "epoch": 7.6, + "learning_rate": 1.3051747844772922e-07, + "loss": 0.0453, + "step": 18150 + }, + { + "epoch": 7.6, + "learning_rate": 1.3024459551177926e-07, + "loss": 0.0348, + "step": 18151 + }, + { + "epoch": 7.6, + "learning_rate": 1.2997199627319047e-07, + "loss": 0.067, + "step": 18152 + }, + { + "epoch": 7.6, + "learning_rate": 1.2969968073979655e-07, + "loss": 0.0545, + "step": 18153 + }, + { + "epoch": 7.6, + "learning_rate": 1.2942764891942793e-07, + "loss": 0.0597, + "step": 18154 + }, + { + "epoch": 7.6, + "learning_rate": 1.291559008199017e-07, + "loss": 0.0481, + "step": 18155 + }, + { + "epoch": 7.6, + "learning_rate": 1.2888443644902937e-07, + "loss": 0.0404, + "step": 18156 + }, + { + "epoch": 7.6, + "learning_rate": 1.2861325581461582e-07, + "loss": 0.0408, + "step": 18157 + }, + { + "epoch": 7.6, + "learning_rate": 1.283423589244537e-07, + "loss": 0.054, + "step": 18158 + }, + { + "epoch": 7.6, + "learning_rate": 1.280717457863312e-07, + "loss": 0.0366, + "step": 18159 + }, + { + "epoch": 7.6, + "learning_rate": 1.2780141640802545e-07, + "loss": 0.0442, + "step": 18160 + }, + { + "epoch": 7.6, + "learning_rate": 1.2753137079730803e-07, + "loss": 0.0408, + "step": 18161 + }, + { + "epoch": 7.6, + "learning_rate": 1.272616089619405e-07, + "loss": 0.0519, + "step": 18162 + }, + { + "epoch": 7.6, + "learning_rate": 1.2699213090967667e-07, + "loss": 0.0511, + "step": 18163 + }, + { + "epoch": 7.6, + "learning_rate": 1.2672293664826364e-07, + "loss": 0.0352, + "step": 18164 + }, + { + "epoch": 7.61, + "learning_rate": 1.2645402618543745e-07, + "loss": 0.0438, + "step": 18165 + }, + { + "epoch": 7.61, + "learning_rate": 1.2618539952893084e-07, + "loss": 0.0405, + "step": 18166 + }, + { + "epoch": 7.61, + "learning_rate": 1.2591705668646093e-07, + "loss": 0.0393, + "step": 18167 + }, + { + "epoch": 7.61, + "learning_rate": 1.256489976657438e-07, + "loss": 0.0612, + "step": 18168 + }, + { + "epoch": 7.61, + "learning_rate": 1.2538122247448325e-07, + "loss": 0.0434, + "step": 18169 + }, + { + "epoch": 7.61, + "learning_rate": 1.2511373112037762e-07, + "loss": 0.0587, + "step": 18170 + }, + { + "epoch": 7.61, + "learning_rate": 1.2484652361111404e-07, + "loss": 0.0385, + "step": 18171 + }, + { + "epoch": 7.61, + "learning_rate": 1.2457959995437528e-07, + "loss": 0.0501, + "step": 18172 + }, + { + "epoch": 7.61, + "learning_rate": 1.243129601578319e-07, + "loss": 0.055, + "step": 18173 + }, + { + "epoch": 7.61, + "learning_rate": 1.2404660422914993e-07, + "loss": 0.0536, + "step": 18174 + }, + { + "epoch": 7.61, + "learning_rate": 1.2378053217598328e-07, + "loss": 0.0527, + "step": 18175 + }, + { + "epoch": 7.61, + "learning_rate": 1.235147440059814e-07, + "loss": 0.0435, + "step": 18176 + }, + { + "epoch": 7.61, + "learning_rate": 1.2324923972678483e-07, + "loss": 0.0398, + "step": 18177 + }, + { + "epoch": 7.61, + "learning_rate": 1.22984019346023e-07, + "loss": 0.0401, + "step": 18178 + }, + { + "epoch": 7.61, + "learning_rate": 1.2271908287132205e-07, + "loss": 0.0482, + "step": 18179 + }, + { + "epoch": 7.61, + "learning_rate": 1.2245443031029593e-07, + "loss": 0.0375, + "step": 18180 + }, + { + "epoch": 7.61, + "learning_rate": 1.221900616705518e-07, + "loss": 0.0529, + "step": 18181 + }, + { + "epoch": 7.61, + "learning_rate": 1.2192597695968922e-07, + "loss": 0.049, + "step": 18182 + }, + { + "epoch": 7.61, + "learning_rate": 1.2166217618529875e-07, + "loss": 0.0245, + "step": 18183 + }, + { + "epoch": 7.61, + "learning_rate": 1.213986593549621e-07, + "loss": 0.0407, + "step": 18184 + }, + { + "epoch": 7.61, + "learning_rate": 1.2113542647625655e-07, + "loss": 0.0319, + "step": 18185 + }, + { + "epoch": 7.61, + "learning_rate": 1.2087247755674603e-07, + "loss": 0.0481, + "step": 18186 + }, + { + "epoch": 7.61, + "learning_rate": 1.2060981260399008e-07, + "loss": 0.0426, + "step": 18187 + }, + { + "epoch": 7.61, + "learning_rate": 1.2034743162553707e-07, + "loss": 0.0647, + "step": 18188 + }, + { + "epoch": 7.62, + "learning_rate": 1.2008533462893102e-07, + "loss": 0.0501, + "step": 18189 + }, + { + "epoch": 7.62, + "learning_rate": 1.1982352162170475e-07, + "loss": 0.0459, + "step": 18190 + }, + { + "epoch": 7.62, + "learning_rate": 1.1956199261138335e-07, + "loss": 0.0471, + "step": 18191 + }, + { + "epoch": 7.62, + "learning_rate": 1.1930074760548416e-07, + "loss": 0.0546, + "step": 18192 + }, + { + "epoch": 7.62, + "learning_rate": 1.1903978661151893e-07, + "loss": 0.0464, + "step": 18193 + }, + { + "epoch": 7.62, + "learning_rate": 1.1877910963698502e-07, + "loss": 0.0338, + "step": 18194 + }, + { + "epoch": 7.62, + "learning_rate": 1.1851871668937753e-07, + "loss": 0.0367, + "step": 18195 + }, + { + "epoch": 7.62, + "learning_rate": 1.1825860777618158e-07, + "loss": 0.0676, + "step": 18196 + }, + { + "epoch": 7.62, + "learning_rate": 1.1799878290487233e-07, + "loss": 0.0543, + "step": 18197 + }, + { + "epoch": 7.62, + "learning_rate": 1.1773924208291931e-07, + "loss": 0.0646, + "step": 18198 + }, + { + "epoch": 7.62, + "learning_rate": 1.1747998531778215e-07, + "loss": 0.043, + "step": 18199 + }, + { + "epoch": 7.62, + "learning_rate": 1.1722101261691377e-07, + "loss": 0.0378, + "step": 18200 + }, + { + "epoch": 7.62, + "learning_rate": 1.1696232398775709e-07, + "loss": 0.0519, + "step": 18201 + }, + { + "epoch": 7.62, + "learning_rate": 1.167039194377484e-07, + "loss": 0.04, + "step": 18202 + }, + { + "epoch": 7.62, + "learning_rate": 1.1644579897431507e-07, + "loss": 0.0461, + "step": 18203 + }, + { + "epoch": 7.62, + "learning_rate": 1.1618796260487675e-07, + "loss": 0.0481, + "step": 18204 + }, + { + "epoch": 7.62, + "learning_rate": 1.1593041033684527e-07, + "loss": 0.0608, + "step": 18205 + }, + { + "epoch": 7.62, + "learning_rate": 1.1567314217762248e-07, + "loss": 0.0435, + "step": 18206 + }, + { + "epoch": 7.62, + "learning_rate": 1.1541615813460472e-07, + "loss": 0.0335, + "step": 18207 + }, + { + "epoch": 7.62, + "learning_rate": 1.1515945821517715e-07, + "loss": 0.0436, + "step": 18208 + }, + { + "epoch": 7.62, + "learning_rate": 1.1490304242671946e-07, + "loss": 0.0643, + "step": 18209 + }, + { + "epoch": 7.62, + "learning_rate": 1.146469107766035e-07, + "loss": 0.0506, + "step": 18210 + }, + { + "epoch": 7.62, + "learning_rate": 1.1439106327218897e-07, + "loss": 0.0478, + "step": 18211 + }, + { + "epoch": 7.62, + "learning_rate": 1.1413549992082994e-07, + "loss": 0.0453, + "step": 18212 + }, + { + "epoch": 7.63, + "learning_rate": 1.1388022072987503e-07, + "loss": 0.0422, + "step": 18213 + }, + { + "epoch": 7.63, + "learning_rate": 1.1362522570665945e-07, + "loss": 0.0464, + "step": 18214 + }, + { + "epoch": 7.63, + "learning_rate": 1.1337051485851403e-07, + "loss": 0.0552, + "step": 18215 + }, + { + "epoch": 7.63, + "learning_rate": 1.1311608819275955e-07, + "loss": 0.051, + "step": 18216 + }, + { + "epoch": 7.63, + "learning_rate": 1.1286194571671016e-07, + "loss": 0.049, + "step": 18217 + }, + { + "epoch": 7.63, + "learning_rate": 1.1260808743767005e-07, + "loss": 0.045, + "step": 18218 + }, + { + "epoch": 7.63, + "learning_rate": 1.1235451336293779e-07, + "loss": 0.0573, + "step": 18219 + }, + { + "epoch": 7.63, + "learning_rate": 1.1210122349979979e-07, + "loss": 0.0408, + "step": 18220 + }, + { + "epoch": 7.63, + "learning_rate": 1.118482178555369e-07, + "loss": 0.0228, + "step": 18221 + }, + { + "epoch": 7.63, + "learning_rate": 1.1159549643742329e-07, + "loss": 0.0352, + "step": 18222 + }, + { + "epoch": 7.63, + "learning_rate": 1.1134305925272204e-07, + "loss": 0.0594, + "step": 18223 + }, + { + "epoch": 7.63, + "learning_rate": 1.1109090630868958e-07, + "loss": 0.0703, + "step": 18224 + }, + { + "epoch": 7.63, + "learning_rate": 1.1083903761257452e-07, + "loss": 0.0261, + "step": 18225 + }, + { + "epoch": 7.63, + "learning_rate": 1.1058745317161557e-07, + "loss": 0.0367, + "step": 18226 + }, + { + "epoch": 7.63, + "learning_rate": 1.1033615299304467e-07, + "loss": 0.0434, + "step": 18227 + }, + { + "epoch": 7.63, + "learning_rate": 1.1008513708408387e-07, + "loss": 0.028, + "step": 18228 + }, + { + "epoch": 7.63, + "learning_rate": 1.0983440545195067e-07, + "loss": 0.0483, + "step": 18229 + }, + { + "epoch": 7.63, + "learning_rate": 1.0958395810385159e-07, + "loss": 0.0938, + "step": 18230 + }, + { + "epoch": 7.63, + "learning_rate": 1.0933379504698417e-07, + "loss": 0.0453, + "step": 18231 + }, + { + "epoch": 7.63, + "learning_rate": 1.0908391628854042e-07, + "loss": 0.0457, + "step": 18232 + }, + { + "epoch": 7.63, + "learning_rate": 1.0883432183570242e-07, + "loss": 0.0763, + "step": 18233 + }, + { + "epoch": 7.63, + "learning_rate": 1.0858501169564439e-07, + "loss": 0.0435, + "step": 18234 + }, + { + "epoch": 7.63, + "learning_rate": 1.0833598587553285e-07, + "loss": 0.0531, + "step": 18235 + }, + { + "epoch": 7.63, + "learning_rate": 1.0808724438252538e-07, + "loss": 0.0486, + "step": 18236 + }, + { + "epoch": 7.64, + "learning_rate": 1.0783878722377295e-07, + "loss": 0.074, + "step": 18237 + }, + { + "epoch": 7.64, + "learning_rate": 1.0759061440641538e-07, + "loss": 0.0639, + "step": 18238 + }, + { + "epoch": 7.64, + "learning_rate": 1.07342725937587e-07, + "loss": 0.05, + "step": 18239 + }, + { + "epoch": 7.64, + "learning_rate": 1.0709512182441539e-07, + "loss": 0.0577, + "step": 18240 + }, + { + "epoch": 7.64, + "learning_rate": 1.0684780207401379e-07, + "loss": 0.044, + "step": 18241 + }, + { + "epoch": 7.64, + "learning_rate": 1.0660076669349429e-07, + "loss": 0.0543, + "step": 18242 + }, + { + "epoch": 7.64, + "learning_rate": 1.0635401568995563e-07, + "loss": 0.0566, + "step": 18243 + }, + { + "epoch": 7.64, + "learning_rate": 1.0610754907049104e-07, + "loss": 0.0689, + "step": 18244 + }, + { + "epoch": 7.64, + "learning_rate": 1.0586136684218595e-07, + "loss": 0.0487, + "step": 18245 + }, + { + "epoch": 7.64, + "learning_rate": 1.0561546901211583e-07, + "loss": 0.0332, + "step": 18246 + }, + { + "epoch": 7.64, + "learning_rate": 1.0536985558734947e-07, + "loss": 0.0603, + "step": 18247 + }, + { + "epoch": 7.64, + "learning_rate": 1.0512452657494565e-07, + "loss": 0.0296, + "step": 18248 + }, + { + "epoch": 7.64, + "learning_rate": 1.0487948198195653e-07, + "loss": 0.0523, + "step": 18249 + }, + { + "epoch": 7.64, + "learning_rate": 1.0463472181542645e-07, + "loss": 0.0466, + "step": 18250 + }, + { + "epoch": 7.64, + "learning_rate": 1.0439024608238979e-07, + "loss": 0.044, + "step": 18251 + }, + { + "epoch": 7.64, + "learning_rate": 1.0414605478987427e-07, + "loss": 0.0507, + "step": 18252 + }, + { + "epoch": 7.64, + "learning_rate": 1.0390214794489984e-07, + "loss": 0.0752, + "step": 18253 + }, + { + "epoch": 7.64, + "learning_rate": 1.0365852555447642e-07, + "loss": 0.0402, + "step": 18254 + }, + { + "epoch": 7.64, + "learning_rate": 1.034151876256062e-07, + "loss": 0.0535, + "step": 18255 + }, + { + "epoch": 7.64, + "learning_rate": 1.031721341652847e-07, + "loss": 0.0374, + "step": 18256 + }, + { + "epoch": 7.64, + "learning_rate": 1.0292936518049745e-07, + "loss": 0.0459, + "step": 18257 + }, + { + "epoch": 7.64, + "learning_rate": 1.026868806782233e-07, + "loss": 0.0468, + "step": 18258 + }, + { + "epoch": 7.64, + "learning_rate": 1.0244468066543223e-07, + "loss": 0.0407, + "step": 18259 + }, + { + "epoch": 7.64, + "learning_rate": 1.0220276514908644e-07, + "loss": 0.0476, + "step": 18260 + }, + { + "epoch": 7.65, + "learning_rate": 1.0196113413613817e-07, + "loss": 0.0576, + "step": 18261 + }, + { + "epoch": 7.65, + "learning_rate": 1.0171978763353407e-07, + "loss": 0.0565, + "step": 18262 + }, + { + "epoch": 7.65, + "learning_rate": 1.0147872564821082e-07, + "loss": 0.0357, + "step": 18263 + }, + { + "epoch": 7.65, + "learning_rate": 1.0123794818709732e-07, + "loss": 0.0555, + "step": 18264 + }, + { + "epoch": 7.65, + "learning_rate": 1.0099745525711579e-07, + "loss": 0.0459, + "step": 18265 + }, + { + "epoch": 7.65, + "learning_rate": 1.0075724686517741e-07, + "loss": 0.0629, + "step": 18266 + }, + { + "epoch": 7.65, + "learning_rate": 1.0051732301818883e-07, + "loss": 0.0401, + "step": 18267 + }, + { + "epoch": 7.65, + "learning_rate": 1.0027768372304458e-07, + "loss": 0.055, + "step": 18268 + }, + { + "epoch": 7.65, + "learning_rate": 1.0003832898663357e-07, + "loss": 0.0564, + "step": 18269 + }, + { + "epoch": 7.65, + "learning_rate": 9.979925881583474e-08, + "loss": 0.0332, + "step": 18270 + }, + { + "epoch": 7.65, + "learning_rate": 9.95604732175215e-08, + "loss": 0.0459, + "step": 18271 + }, + { + "epoch": 7.65, + "learning_rate": 9.932197219855721e-08, + "loss": 0.0621, + "step": 18272 + }, + { + "epoch": 7.65, + "learning_rate": 9.908375576579754e-08, + "loss": 0.036, + "step": 18273 + }, + { + "epoch": 7.65, + "learning_rate": 9.884582392608921e-08, + "loss": 0.0349, + "step": 18274 + }, + { + "epoch": 7.65, + "learning_rate": 9.860817668627121e-08, + "loss": 0.0479, + "step": 18275 + }, + { + "epoch": 7.65, + "learning_rate": 9.837081405317472e-08, + "loss": 0.0517, + "step": 18276 + }, + { + "epoch": 7.65, + "learning_rate": 9.813373603362208e-08, + "loss": 0.041, + "step": 18277 + }, + { + "epoch": 7.65, + "learning_rate": 9.789694263443005e-08, + "loss": 0.045, + "step": 18278 + }, + { + "epoch": 7.65, + "learning_rate": 9.76604338624021e-08, + "loss": 0.0493, + "step": 18279 + }, + { + "epoch": 7.65, + "learning_rate": 9.742420972433831e-08, + "loss": 0.065, + "step": 18280 + }, + { + "epoch": 7.65, + "learning_rate": 9.718827022702881e-08, + "loss": 0.0346, + "step": 18281 + }, + { + "epoch": 7.65, + "learning_rate": 9.695261537725376e-08, + "loss": 0.0685, + "step": 18282 + }, + { + "epoch": 7.65, + "learning_rate": 9.671724518178882e-08, + "loss": 0.0492, + "step": 18283 + }, + { + "epoch": 7.66, + "learning_rate": 9.64821596473986e-08, + "loss": 0.0454, + "step": 18284 + }, + { + "epoch": 7.66, + "learning_rate": 9.62473587808399e-08, + "loss": 0.0623, + "step": 18285 + }, + { + "epoch": 7.66, + "learning_rate": 9.601284258886179e-08, + "loss": 0.0458, + "step": 18286 + }, + { + "epoch": 7.66, + "learning_rate": 9.577861107820774e-08, + "loss": 0.0386, + "step": 18287 + }, + { + "epoch": 7.66, + "learning_rate": 9.554466425560793e-08, + "loss": 0.0433, + "step": 18288 + }, + { + "epoch": 7.66, + "learning_rate": 9.531100212778698e-08, + "loss": 0.0519, + "step": 18289 + }, + { + "epoch": 7.66, + "learning_rate": 9.507762470146176e-08, + "loss": 0.0507, + "step": 18290 + }, + { + "epoch": 7.66, + "learning_rate": 9.484453198334242e-08, + "loss": 0.0406, + "step": 18291 + }, + { + "epoch": 7.66, + "learning_rate": 9.461172398012586e-08, + "loss": 0.08, + "step": 18292 + }, + { + "epoch": 7.66, + "learning_rate": 9.437920069850559e-08, + "loss": 0.0295, + "step": 18293 + }, + { + "epoch": 7.66, + "learning_rate": 9.414696214516628e-08, + "loss": 0.0366, + "step": 18294 + }, + { + "epoch": 7.66, + "learning_rate": 9.391500832678147e-08, + "loss": 0.0638, + "step": 18295 + }, + { + "epoch": 7.66, + "learning_rate": 9.368333925001916e-08, + "loss": 0.0536, + "step": 18296 + }, + { + "epoch": 7.66, + "learning_rate": 9.345195492153736e-08, + "loss": 0.0378, + "step": 18297 + }, + { + "epoch": 7.66, + "learning_rate": 9.322085534798963e-08, + "loss": 0.0539, + "step": 18298 + }, + { + "epoch": 7.66, + "learning_rate": 9.299004053601734e-08, + "loss": 0.0507, + "step": 18299 + }, + { + "epoch": 7.66, + "learning_rate": 9.275951049225518e-08, + "loss": 0.0471, + "step": 18300 + }, + { + "epoch": 7.66, + "learning_rate": 9.252926522332895e-08, + "loss": 0.0516, + "step": 18301 + }, + { + "epoch": 7.66, + "learning_rate": 9.22993047358578e-08, + "loss": 0.063, + "step": 18302 + }, + { + "epoch": 7.66, + "learning_rate": 9.206962903644978e-08, + "loss": 0.0518, + "step": 18303 + }, + { + "epoch": 7.66, + "learning_rate": 9.184023813170961e-08, + "loss": 0.0521, + "step": 18304 + }, + { + "epoch": 7.66, + "learning_rate": 9.161113202822868e-08, + "loss": 0.0747, + "step": 18305 + }, + { + "epoch": 7.66, + "learning_rate": 9.138231073259284e-08, + "loss": 0.0494, + "step": 18306 + }, + { + "epoch": 7.66, + "learning_rate": 9.115377425138017e-08, + "loss": 0.0513, + "step": 18307 + }, + { + "epoch": 7.67, + "learning_rate": 9.092552259115983e-08, + "loss": 0.0669, + "step": 18308 + }, + { + "epoch": 7.67, + "learning_rate": 9.069755575849215e-08, + "loss": 0.0526, + "step": 18309 + }, + { + "epoch": 7.67, + "learning_rate": 9.046987375992854e-08, + "loss": 0.0412, + "step": 18310 + }, + { + "epoch": 7.67, + "learning_rate": 9.024247660201602e-08, + "loss": 0.051, + "step": 18311 + }, + { + "epoch": 7.67, + "learning_rate": 9.001536429128932e-08, + "loss": 0.0542, + "step": 18312 + }, + { + "epoch": 7.67, + "learning_rate": 8.978853683427769e-08, + "loss": 0.0435, + "step": 18313 + }, + { + "epoch": 7.67, + "learning_rate": 8.956199423750033e-08, + "loss": 0.0373, + "step": 18314 + }, + { + "epoch": 7.67, + "learning_rate": 8.933573650746762e-08, + "loss": 0.052, + "step": 18315 + }, + { + "epoch": 7.67, + "learning_rate": 8.910976365068546e-08, + "loss": 0.0386, + "step": 18316 + }, + { + "epoch": 7.67, + "learning_rate": 8.888407567364865e-08, + "loss": 0.056, + "step": 18317 + }, + { + "epoch": 7.67, + "learning_rate": 8.865867258284422e-08, + "loss": 0.0648, + "step": 18318 + }, + { + "epoch": 7.67, + "learning_rate": 8.843355438475032e-08, + "loss": 0.0563, + "step": 18319 + }, + { + "epoch": 7.67, + "learning_rate": 8.820872108583955e-08, + "loss": 0.0504, + "step": 18320 + }, + { + "epoch": 7.67, + "learning_rate": 8.798417269257342e-08, + "loss": 0.0512, + "step": 18321 + }, + { + "epoch": 7.67, + "learning_rate": 8.775990921140565e-08, + "loss": 0.0432, + "step": 18322 + }, + { + "epoch": 7.67, + "learning_rate": 8.75359306487844e-08, + "loss": 0.055, + "step": 18323 + }, + { + "epoch": 7.67, + "learning_rate": 8.731223701114566e-08, + "loss": 0.0555, + "step": 18324 + }, + { + "epoch": 7.67, + "learning_rate": 8.708882830492093e-08, + "loss": 0.0449, + "step": 18325 + }, + { + "epoch": 7.67, + "learning_rate": 8.686570453653064e-08, + "loss": 0.0484, + "step": 18326 + }, + { + "epoch": 7.67, + "learning_rate": 8.664286571238967e-08, + "loss": 0.0633, + "step": 18327 + }, + { + "epoch": 7.67, + "learning_rate": 8.642031183890287e-08, + "loss": 0.0575, + "step": 18328 + }, + { + "epoch": 7.67, + "learning_rate": 8.619804292246626e-08, + "loss": 0.0379, + "step": 18329 + }, + { + "epoch": 7.67, + "learning_rate": 8.597605896946915e-08, + "loss": 0.0353, + "step": 18330 + }, + { + "epoch": 7.67, + "learning_rate": 8.575435998629311e-08, + "loss": 0.0421, + "step": 18331 + }, + { + "epoch": 7.68, + "learning_rate": 8.553294597930972e-08, + "loss": 0.0404, + "step": 18332 + }, + { + "epoch": 7.68, + "learning_rate": 8.5311816954885e-08, + "loss": 0.0475, + "step": 18333 + }, + { + "epoch": 7.68, + "learning_rate": 8.509097291937273e-08, + "loss": 0.0627, + "step": 18334 + }, + { + "epoch": 7.68, + "learning_rate": 8.487041387912342e-08, + "loss": 0.0532, + "step": 18335 + }, + { + "epoch": 7.68, + "learning_rate": 8.465013984047532e-08, + "loss": 0.048, + "step": 18336 + }, + { + "epoch": 7.68, + "learning_rate": 8.443015080976003e-08, + "loss": 0.0394, + "step": 18337 + }, + { + "epoch": 7.68, + "learning_rate": 8.421044679330026e-08, + "loss": 0.0458, + "step": 18338 + }, + { + "epoch": 7.68, + "learning_rate": 8.39910277974132e-08, + "loss": 0.0473, + "step": 18339 + }, + { + "epoch": 7.68, + "learning_rate": 8.377189382840378e-08, + "loss": 0.0588, + "step": 18340 + }, + { + "epoch": 7.68, + "learning_rate": 8.355304489257254e-08, + "loss": 0.0629, + "step": 18341 + }, + { + "epoch": 7.68, + "learning_rate": 8.333448099620889e-08, + "loss": 0.0505, + "step": 18342 + }, + { + "epoch": 7.68, + "learning_rate": 8.311620214559557e-08, + "loss": 0.0495, + "step": 18343 + }, + { + "epoch": 7.68, + "learning_rate": 8.289820834700757e-08, + "loss": 0.0716, + "step": 18344 + }, + { + "epoch": 7.68, + "learning_rate": 8.268049960670987e-08, + "loss": 0.04, + "step": 18345 + }, + { + "epoch": 7.68, + "learning_rate": 8.24630759309608e-08, + "loss": 0.051, + "step": 18346 + }, + { + "epoch": 7.68, + "learning_rate": 8.224593732600983e-08, + "loss": 0.067, + "step": 18347 + }, + { + "epoch": 7.68, + "learning_rate": 8.202908379809859e-08, + "loss": 0.0363, + "step": 18348 + }, + { + "epoch": 7.68, + "learning_rate": 8.181251535346102e-08, + "loss": 0.0484, + "step": 18349 + }, + { + "epoch": 7.68, + "learning_rate": 8.1596231998321e-08, + "loss": 0.0618, + "step": 18350 + }, + { + "epoch": 7.68, + "learning_rate": 8.138023373889581e-08, + "loss": 0.0392, + "step": 18351 + }, + { + "epoch": 7.68, + "learning_rate": 8.116452058139379e-08, + "loss": 0.0354, + "step": 18352 + }, + { + "epoch": 7.68, + "learning_rate": 8.094909253201555e-08, + "loss": 0.0366, + "step": 18353 + }, + { + "epoch": 7.68, + "learning_rate": 8.073394959695391e-08, + "loss": 0.059, + "step": 18354 + }, + { + "epoch": 7.68, + "learning_rate": 8.051909178239392e-08, + "loss": 0.034, + "step": 18355 + }, + { + "epoch": 7.69, + "learning_rate": 8.030451909450842e-08, + "loss": 0.0542, + "step": 18356 + }, + { + "epoch": 7.69, + "learning_rate": 8.009023153946805e-08, + "loss": 0.0481, + "step": 18357 + }, + { + "epoch": 7.69, + "learning_rate": 7.987622912343118e-08, + "loss": 0.0597, + "step": 18358 + }, + { + "epoch": 7.69, + "learning_rate": 7.966251185254847e-08, + "loss": 0.0338, + "step": 18359 + }, + { + "epoch": 7.69, + "learning_rate": 7.94490797329639e-08, + "loss": 0.053, + "step": 18360 + }, + { + "epoch": 7.69, + "learning_rate": 7.923593277081254e-08, + "loss": 0.0566, + "step": 18361 + }, + { + "epoch": 7.69, + "learning_rate": 7.902307097222061e-08, + "loss": 0.0374, + "step": 18362 + }, + { + "epoch": 7.69, + "learning_rate": 7.881049434330656e-08, + "loss": 0.0456, + "step": 18363 + }, + { + "epoch": 7.69, + "learning_rate": 7.859820289018105e-08, + "loss": 0.0602, + "step": 18364 + }, + { + "epoch": 7.69, + "learning_rate": 7.838619661894587e-08, + "loss": 0.0626, + "step": 18365 + }, + { + "epoch": 7.69, + "learning_rate": 7.817447553569613e-08, + "loss": 0.0377, + "step": 18366 + }, + { + "epoch": 7.69, + "learning_rate": 7.796303964651586e-08, + "loss": 0.0698, + "step": 18367 + }, + { + "epoch": 7.69, + "learning_rate": 7.775188895748354e-08, + "loss": 0.0536, + "step": 18368 + }, + { + "epoch": 7.69, + "learning_rate": 7.754102347466874e-08, + "loss": 0.0383, + "step": 18369 + }, + { + "epoch": 7.69, + "learning_rate": 7.733044320413218e-08, + "loss": 0.0372, + "step": 18370 + }, + { + "epoch": 7.69, + "learning_rate": 7.71201481519257e-08, + "loss": 0.055, + "step": 18371 + }, + { + "epoch": 7.69, + "learning_rate": 7.691013832409555e-08, + "loss": 0.0551, + "step": 18372 + }, + { + "epoch": 7.69, + "learning_rate": 7.670041372667803e-08, + "loss": 0.0466, + "step": 18373 + }, + { + "epoch": 7.69, + "learning_rate": 7.649097436570164e-08, + "loss": 0.0353, + "step": 18374 + }, + { + "epoch": 7.69, + "learning_rate": 7.6281820247186e-08, + "loss": 0.0641, + "step": 18375 + }, + { + "epoch": 7.69, + "learning_rate": 7.60729513771441e-08, + "loss": 0.0465, + "step": 18376 + }, + { + "epoch": 7.69, + "learning_rate": 7.58643677615778e-08, + "loss": 0.0322, + "step": 18377 + }, + { + "epoch": 7.69, + "learning_rate": 7.56560694064845e-08, + "loss": 0.0561, + "step": 18378 + }, + { + "epoch": 7.69, + "learning_rate": 7.544805631784946e-08, + "loss": 0.0574, + "step": 18379 + }, + { + "epoch": 7.7, + "learning_rate": 7.524032850165452e-08, + "loss": 0.0517, + "step": 18380 + }, + { + "epoch": 7.7, + "learning_rate": 7.503288596386827e-08, + "loss": 0.0456, + "step": 18381 + }, + { + "epoch": 7.7, + "learning_rate": 7.48257287104559e-08, + "loss": 0.0423, + "step": 18382 + }, + { + "epoch": 7.7, + "learning_rate": 7.461885674736935e-08, + "loss": 0.0628, + "step": 18383 + }, + { + "epoch": 7.7, + "learning_rate": 7.441227008055607e-08, + "loss": 0.0429, + "step": 18384 + }, + { + "epoch": 7.7, + "learning_rate": 7.420596871595465e-08, + "loss": 0.0385, + "step": 18385 + }, + { + "epoch": 7.7, + "learning_rate": 7.399995265949367e-08, + "loss": 0.0689, + "step": 18386 + }, + { + "epoch": 7.7, + "learning_rate": 7.379422191709729e-08, + "loss": 0.0384, + "step": 18387 + }, + { + "epoch": 7.7, + "learning_rate": 7.358877649467744e-08, + "loss": 0.0542, + "step": 18388 + }, + { + "epoch": 7.7, + "learning_rate": 7.338361639813941e-08, + "loss": 0.0519, + "step": 18389 + }, + { + "epoch": 7.7, + "learning_rate": 7.317874163338068e-08, + "loss": 0.0408, + "step": 18390 + }, + { + "epoch": 7.7, + "learning_rate": 7.29741522062899e-08, + "loss": 0.0381, + "step": 18391 + }, + { + "epoch": 7.7, + "learning_rate": 7.27698481227479e-08, + "loss": 0.0362, + "step": 18392 + }, + { + "epoch": 7.7, + "learning_rate": 7.256582938862777e-08, + "loss": 0.0502, + "step": 18393 + }, + { + "epoch": 7.7, + "learning_rate": 7.23620960097926e-08, + "loss": 0.0438, + "step": 18394 + }, + { + "epoch": 7.7, + "learning_rate": 7.215864799209993e-08, + "loss": 0.048, + "step": 18395 + }, + { + "epoch": 7.7, + "learning_rate": 7.195548534139729e-08, + "loss": 0.0833, + "step": 18396 + }, + { + "epoch": 7.7, + "learning_rate": 7.175260806352447e-08, + "loss": 0.0544, + "step": 18397 + }, + { + "epoch": 7.7, + "learning_rate": 7.155001616431234e-08, + "loss": 0.0559, + "step": 18398 + }, + { + "epoch": 7.7, + "learning_rate": 7.134770964958404e-08, + "loss": 0.0454, + "step": 18399 + }, + { + "epoch": 7.7, + "learning_rate": 7.114568852515602e-08, + "loss": 0.0659, + "step": 18400 + }, + { + "epoch": 7.7, + "learning_rate": 7.094395279683475e-08, + "loss": 0.0467, + "step": 18401 + }, + { + "epoch": 7.7, + "learning_rate": 7.074250247041892e-08, + "loss": 0.0348, + "step": 18402 + }, + { + "epoch": 7.7, + "learning_rate": 7.054133755169834e-08, + "loss": 0.0501, + "step": 18403 + }, + { + "epoch": 7.71, + "learning_rate": 7.034045804645506e-08, + "loss": 0.048, + "step": 18404 + }, + { + "epoch": 7.71, + "learning_rate": 7.013986396046446e-08, + "loss": 0.0406, + "step": 18405 + }, + { + "epoch": 7.71, + "learning_rate": 6.993955529949193e-08, + "loss": 0.0639, + "step": 18406 + }, + { + "epoch": 7.71, + "learning_rate": 6.973953206929507e-08, + "loss": 0.0416, + "step": 18407 + }, + { + "epoch": 7.71, + "learning_rate": 6.953979427562263e-08, + "loss": 0.0467, + "step": 18408 + }, + { + "epoch": 7.71, + "learning_rate": 6.93403419242178e-08, + "loss": 0.0445, + "step": 18409 + }, + { + "epoch": 7.71, + "learning_rate": 6.914117502081263e-08, + "loss": 0.0451, + "step": 18410 + }, + { + "epoch": 7.71, + "learning_rate": 6.894229357113036e-08, + "loss": 0.0568, + "step": 18411 + }, + { + "epoch": 7.71, + "learning_rate": 6.874369758089084e-08, + "loss": 0.0498, + "step": 18412 + }, + { + "epoch": 7.71, + "learning_rate": 6.85453870557995e-08, + "loss": 0.0486, + "step": 18413 + }, + { + "epoch": 7.71, + "learning_rate": 6.834736200155956e-08, + "loss": 0.0606, + "step": 18414 + }, + { + "epoch": 7.71, + "learning_rate": 6.814962242386092e-08, + "loss": 0.0529, + "step": 18415 + }, + { + "epoch": 7.71, + "learning_rate": 6.795216832838791e-08, + "loss": 0.0525, + "step": 18416 + }, + { + "epoch": 7.71, + "learning_rate": 6.775499972081601e-08, + "loss": 0.0445, + "step": 18417 + }, + { + "epoch": 7.71, + "learning_rate": 6.755811660681288e-08, + "loss": 0.0378, + "step": 18418 + }, + { + "epoch": 7.71, + "learning_rate": 6.736151899203847e-08, + "loss": 0.035, + "step": 18419 + }, + { + "epoch": 7.71, + "learning_rate": 6.716520688214267e-08, + "loss": 0.0318, + "step": 18420 + }, + { + "epoch": 7.71, + "learning_rate": 6.696918028276878e-08, + "loss": 0.0613, + "step": 18421 + }, + { + "epoch": 7.71, + "learning_rate": 6.677343919955114e-08, + "loss": 0.041, + "step": 18422 + }, + { + "epoch": 7.71, + "learning_rate": 6.657798363811641e-08, + "loss": 0.0591, + "step": 18423 + }, + { + "epoch": 7.71, + "learning_rate": 6.638281360408339e-08, + "loss": 0.034, + "step": 18424 + }, + { + "epoch": 7.71, + "learning_rate": 6.618792910306094e-08, + "loss": 0.038, + "step": 18425 + }, + { + "epoch": 7.71, + "learning_rate": 6.599333014065123e-08, + "loss": 0.0742, + "step": 18426 + }, + { + "epoch": 7.71, + "learning_rate": 6.57990167224487e-08, + "loss": 0.0403, + "step": 18427 + }, + { + "epoch": 7.72, + "learning_rate": 6.560498885403666e-08, + "loss": 0.0592, + "step": 18428 + }, + { + "epoch": 7.72, + "learning_rate": 6.541124654099507e-08, + "loss": 0.0598, + "step": 18429 + }, + { + "epoch": 7.72, + "learning_rate": 6.521778978889059e-08, + "loss": 0.0355, + "step": 18430 + }, + { + "epoch": 7.72, + "learning_rate": 6.502461860328546e-08, + "loss": 0.0573, + "step": 18431 + }, + { + "epoch": 7.72, + "learning_rate": 6.483173298973078e-08, + "loss": 0.0384, + "step": 18432 + }, + { + "epoch": 7.72, + "learning_rate": 6.4639132953771e-08, + "loss": 0.0596, + "step": 18433 + }, + { + "epoch": 7.72, + "learning_rate": 6.444681850094392e-08, + "loss": 0.05, + "step": 18434 + }, + { + "epoch": 7.72, + "learning_rate": 6.425478963677623e-08, + "loss": 0.0379, + "step": 18435 + }, + { + "epoch": 7.72, + "learning_rate": 6.406304636678795e-08, + "loss": 0.0476, + "step": 18436 + }, + { + "epoch": 7.72, + "learning_rate": 6.387158869649135e-08, + "loss": 0.0538, + "step": 18437 + }, + { + "epoch": 7.72, + "learning_rate": 6.368041663138758e-08, + "loss": 0.0675, + "step": 18438 + }, + { + "epoch": 7.72, + "learning_rate": 6.348953017697223e-08, + "loss": 0.0519, + "step": 18439 + }, + { + "epoch": 7.72, + "learning_rate": 6.329892933873316e-08, + "loss": 0.0381, + "step": 18440 + }, + { + "epoch": 7.72, + "learning_rate": 6.31086141221493e-08, + "loss": 0.0374, + "step": 18441 + }, + { + "epoch": 7.72, + "learning_rate": 6.291858453269073e-08, + "loss": 0.0483, + "step": 18442 + }, + { + "epoch": 7.72, + "learning_rate": 6.272884057581863e-08, + "loss": 0.0498, + "step": 18443 + }, + { + "epoch": 7.72, + "learning_rate": 6.253938225698863e-08, + "loss": 0.0433, + "step": 18444 + }, + { + "epoch": 7.72, + "learning_rate": 6.235020958164529e-08, + "loss": 0.037, + "step": 18445 + }, + { + "epoch": 7.72, + "learning_rate": 6.216132255522644e-08, + "loss": 0.0459, + "step": 18446 + }, + { + "epoch": 7.72, + "learning_rate": 6.197272118316111e-08, + "loss": 0.0669, + "step": 18447 + }, + { + "epoch": 7.72, + "learning_rate": 6.178440547087272e-08, + "loss": 0.044, + "step": 18448 + }, + { + "epoch": 7.72, + "learning_rate": 6.159637542377139e-08, + "loss": 0.0432, + "step": 18449 + }, + { + "epoch": 7.72, + "learning_rate": 6.140863104726391e-08, + "loss": 0.049, + "step": 18450 + }, + { + "epoch": 7.72, + "learning_rate": 6.122117234674485e-08, + "loss": 0.0487, + "step": 18451 + }, + { + "epoch": 7.73, + "learning_rate": 6.103399932760434e-08, + "loss": 0.0666, + "step": 18452 + }, + { + "epoch": 7.73, + "learning_rate": 6.084711199522142e-08, + "loss": 0.0438, + "step": 18453 + }, + { + "epoch": 7.73, + "learning_rate": 6.066051035496956e-08, + "loss": 0.0507, + "step": 18454 + }, + { + "epoch": 7.73, + "learning_rate": 6.047419441221003e-08, + "loss": 0.0572, + "step": 18455 + }, + { + "epoch": 7.73, + "learning_rate": 6.028816417230077e-08, + "loss": 0.054, + "step": 18456 + }, + { + "epoch": 7.73, + "learning_rate": 6.010241964058639e-08, + "loss": 0.0516, + "step": 18457 + }, + { + "epoch": 7.73, + "learning_rate": 5.991696082240817e-08, + "loss": 0.0338, + "step": 18458 + }, + { + "epoch": 7.73, + "learning_rate": 5.97317877230974e-08, + "loss": 0.0367, + "step": 18459 + }, + { + "epoch": 7.73, + "learning_rate": 5.954690034797428e-08, + "loss": 0.0392, + "step": 18460 + }, + { + "epoch": 7.73, + "learning_rate": 5.9362298702354546e-08, + "loss": 0.0513, + "step": 18461 + }, + { + "epoch": 7.73, + "learning_rate": 5.917798279154507e-08, + "loss": 0.0627, + "step": 18462 + }, + { + "epoch": 7.73, + "learning_rate": 5.899395262084162e-08, + "loss": 0.0666, + "step": 18463 + }, + { + "epoch": 7.73, + "learning_rate": 5.881020819553773e-08, + "loss": 0.0477, + "step": 18464 + }, + { + "epoch": 7.73, + "learning_rate": 5.862674952091141e-08, + "loss": 0.0544, + "step": 18465 + }, + { + "epoch": 7.73, + "learning_rate": 5.8443576602237316e-08, + "loss": 0.0605, + "step": 18466 + }, + { + "epoch": 7.73, + "learning_rate": 5.8260689444780135e-08, + "loss": 0.0631, + "step": 18467 + }, + { + "epoch": 7.73, + "learning_rate": 5.807808805379789e-08, + "loss": 0.0359, + "step": 18468 + }, + { + "epoch": 7.73, + "learning_rate": 5.78957724345397e-08, + "loss": 0.0467, + "step": 18469 + }, + { + "epoch": 7.73, + "learning_rate": 5.7713742592243604e-08, + "loss": 0.0416, + "step": 18470 + }, + { + "epoch": 7.73, + "learning_rate": 5.753199853214431e-08, + "loss": 0.0405, + "step": 18471 + }, + { + "epoch": 7.73, + "learning_rate": 5.735054025946429e-08, + "loss": 0.0299, + "step": 18472 + }, + { + "epoch": 7.73, + "learning_rate": 5.71693677794205e-08, + "loss": 0.037, + "step": 18473 + }, + { + "epoch": 7.73, + "learning_rate": 5.6988481097219886e-08, + "loss": 0.0462, + "step": 18474 + }, + { + "epoch": 7.73, + "learning_rate": 5.680788021806161e-08, + "loss": 0.0364, + "step": 18475 + }, + { + "epoch": 7.74, + "learning_rate": 5.6627565147138185e-08, + "loss": 0.0571, + "step": 18476 + }, + { + "epoch": 7.74, + "learning_rate": 5.6447535889632144e-08, + "loss": 0.047, + "step": 18477 + }, + { + "epoch": 7.74, + "learning_rate": 5.6267792450717115e-08, + "loss": 0.0395, + "step": 18478 + }, + { + "epoch": 7.74, + "learning_rate": 5.608833483556009e-08, + "loss": 0.0514, + "step": 18479 + }, + { + "epoch": 7.74, + "learning_rate": 5.590916304932026e-08, + "loss": 0.0673, + "step": 18480 + }, + { + "epoch": 7.74, + "learning_rate": 5.573027709714796e-08, + "loss": 0.0425, + "step": 18481 + }, + { + "epoch": 7.74, + "learning_rate": 5.555167698418351e-08, + "loss": 0.0336, + "step": 18482 + }, + { + "epoch": 7.74, + "learning_rate": 5.5373362715561706e-08, + "loss": 0.044, + "step": 18483 + }, + { + "epoch": 7.74, + "learning_rate": 5.519533429640955e-08, + "loss": 0.0461, + "step": 18484 + }, + { + "epoch": 7.74, + "learning_rate": 5.501759173184074e-08, + "loss": 0.0591, + "step": 18485 + }, + { + "epoch": 7.74, + "learning_rate": 5.4840135026966723e-08, + "loss": 0.069, + "step": 18486 + }, + { + "epoch": 7.74, + "learning_rate": 5.466296418688788e-08, + "loss": 0.0385, + "step": 18487 + }, + { + "epoch": 7.74, + "learning_rate": 5.4486079216696794e-08, + "loss": 0.053, + "step": 18488 + }, + { + "epoch": 7.74, + "learning_rate": 5.4309480121477185e-08, + "loss": 0.0384, + "step": 18489 + }, + { + "epoch": 7.74, + "learning_rate": 5.4133166906307213e-08, + "loss": 0.0696, + "step": 18490 + }, + { + "epoch": 7.74, + "learning_rate": 5.395713957625281e-08, + "loss": 0.0394, + "step": 18491 + }, + { + "epoch": 7.74, + "learning_rate": 5.3781398136374394e-08, + "loss": 0.0388, + "step": 18492 + }, + { + "epoch": 7.74, + "learning_rate": 5.360594259172347e-08, + "loss": 0.0303, + "step": 18493 + }, + { + "epoch": 7.74, + "learning_rate": 5.343077294734378e-08, + "loss": 0.0628, + "step": 18494 + }, + { + "epoch": 7.74, + "learning_rate": 5.3255889208270186e-08, + "loss": 0.0664, + "step": 18495 + }, + { + "epoch": 7.74, + "learning_rate": 5.3081291379529776e-08, + "loss": 0.0612, + "step": 18496 + }, + { + "epoch": 7.74, + "learning_rate": 5.290697946614076e-08, + "loss": 0.0577, + "step": 18497 + }, + { + "epoch": 7.74, + "learning_rate": 5.273295347311358e-08, + "loss": 0.0346, + "step": 18498 + }, + { + "epoch": 7.75, + "learning_rate": 5.255921340544978e-08, + "loss": 0.0435, + "step": 18499 + }, + { + "epoch": 7.75, + "learning_rate": 5.2385759268145375e-08, + "loss": 0.0461, + "step": 18500 + }, + { + "epoch": 7.75, + "learning_rate": 5.221259106618415e-08, + "loss": 0.0608, + "step": 18501 + }, + { + "epoch": 7.75, + "learning_rate": 5.203970880454434e-08, + "loss": 0.0501, + "step": 18502 + }, + { + "epoch": 7.75, + "learning_rate": 5.1867112488196425e-08, + "loss": 0.036, + "step": 18503 + }, + { + "epoch": 7.75, + "learning_rate": 5.1694802122099763e-08, + "loss": 0.047, + "step": 18504 + }, + { + "epoch": 7.75, + "learning_rate": 5.1522777711209284e-08, + "loss": 0.0485, + "step": 18505 + }, + { + "epoch": 7.75, + "learning_rate": 5.1351039260466585e-08, + "loss": 0.0572, + "step": 18506 + }, + { + "epoch": 7.75, + "learning_rate": 5.117958677481105e-08, + "loss": 0.0369, + "step": 18507 + }, + { + "epoch": 7.75, + "learning_rate": 5.100842025916986e-08, + "loss": 0.0381, + "step": 18508 + }, + { + "epoch": 7.75, + "learning_rate": 5.083753971846239e-08, + "loss": 0.0697, + "step": 18509 + }, + { + "epoch": 7.75, + "learning_rate": 5.06669451576014e-08, + "loss": 0.0833, + "step": 18510 + }, + { + "epoch": 7.75, + "learning_rate": 5.0496636581490734e-08, + "loss": 0.0553, + "step": 18511 + }, + { + "epoch": 7.75, + "learning_rate": 5.032661399502536e-08, + "loss": 0.0414, + "step": 18512 + }, + { + "epoch": 7.75, + "learning_rate": 5.015687740309139e-08, + "loss": 0.0601, + "step": 18513 + }, + { + "epoch": 7.75, + "learning_rate": 4.9987426810569337e-08, + "loss": 0.0443, + "step": 18514 + }, + { + "epoch": 7.75, + "learning_rate": 4.981826222232866e-08, + "loss": 0.0465, + "step": 18515 + }, + { + "epoch": 7.75, + "learning_rate": 4.964938364323213e-08, + "loss": 0.0329, + "step": 18516 + }, + { + "epoch": 7.75, + "learning_rate": 4.948079107813475e-08, + "loss": 0.0555, + "step": 18517 + }, + { + "epoch": 7.75, + "learning_rate": 4.931248453188264e-08, + "loss": 0.0561, + "step": 18518 + }, + { + "epoch": 7.75, + "learning_rate": 4.914446400931305e-08, + "loss": 0.0438, + "step": 18519 + }, + { + "epoch": 7.75, + "learning_rate": 4.897672951525545e-08, + "loss": 0.043, + "step": 18520 + }, + { + "epoch": 7.75, + "learning_rate": 4.8809281054530423e-08, + "loss": 0.043, + "step": 18521 + }, + { + "epoch": 7.75, + "learning_rate": 4.864211863195301e-08, + "loss": 0.0384, + "step": 18522 + }, + { + "epoch": 7.76, + "learning_rate": 4.847524225232714e-08, + "loss": 0.048, + "step": 18523 + }, + { + "epoch": 7.76, + "learning_rate": 4.8308651920448976e-08, + "loss": 0.0462, + "step": 18524 + }, + { + "epoch": 7.76, + "learning_rate": 4.814234764110803e-08, + "loss": 0.058, + "step": 18525 + }, + { + "epoch": 7.76, + "learning_rate": 4.7976329419084924e-08, + "loss": 0.0405, + "step": 18526 + }, + { + "epoch": 7.76, + "learning_rate": 4.781059725915027e-08, + "loss": 0.0474, + "step": 18527 + }, + { + "epoch": 7.76, + "learning_rate": 4.764515116606805e-08, + "loss": 0.0673, + "step": 18528 + }, + { + "epoch": 7.76, + "learning_rate": 4.747999114459556e-08, + "loss": 0.0538, + "step": 18529 + }, + { + "epoch": 7.76, + "learning_rate": 4.73151171994779e-08, + "loss": 0.0406, + "step": 18530 + }, + { + "epoch": 7.76, + "learning_rate": 4.715052933545572e-08, + "loss": 0.045, + "step": 18531 + }, + { + "epoch": 7.76, + "learning_rate": 4.698622755725968e-08, + "loss": 0.0565, + "step": 18532 + }, + { + "epoch": 7.76, + "learning_rate": 4.682221186961267e-08, + "loss": 0.0716, + "step": 18533 + }, + { + "epoch": 7.76, + "learning_rate": 4.665848227722869e-08, + "loss": 0.0302, + "step": 18534 + }, + { + "epoch": 7.76, + "learning_rate": 4.649503878481398e-08, + "loss": 0.0385, + "step": 18535 + }, + { + "epoch": 7.76, + "learning_rate": 4.6331881397067016e-08, + "loss": 0.0546, + "step": 18536 + }, + { + "epoch": 7.76, + "learning_rate": 4.616901011867736e-08, + "loss": 0.0644, + "step": 18537 + }, + { + "epoch": 7.76, + "learning_rate": 4.6006424954326834e-08, + "loss": 0.0714, + "step": 18538 + }, + { + "epoch": 7.76, + "learning_rate": 4.584412590868836e-08, + "loss": 0.0328, + "step": 18539 + }, + { + "epoch": 7.76, + "learning_rate": 4.568211298642711e-08, + "loss": 0.0542, + "step": 18540 + }, + { + "epoch": 7.76, + "learning_rate": 4.552038619220045e-08, + "loss": 0.0284, + "step": 18541 + }, + { + "epoch": 7.76, + "learning_rate": 4.535894553065689e-08, + "loss": 0.0417, + "step": 18542 + }, + { + "epoch": 7.76, + "learning_rate": 4.519779100643606e-08, + "loss": 0.0429, + "step": 18543 + }, + { + "epoch": 7.76, + "learning_rate": 4.503692262417203e-08, + "loss": 0.0593, + "step": 18544 + }, + { + "epoch": 7.76, + "learning_rate": 4.487634038848776e-08, + "loss": 0.0496, + "step": 18545 + }, + { + "epoch": 7.76, + "learning_rate": 4.471604430399845e-08, + "loss": 0.0484, + "step": 18546 + }, + { + "epoch": 7.77, + "learning_rate": 4.455603437531264e-08, + "loss": 0.0529, + "step": 18547 + }, + { + "epoch": 7.77, + "learning_rate": 4.439631060702887e-08, + "loss": 0.0473, + "step": 18548 + }, + { + "epoch": 7.77, + "learning_rate": 4.4236873003739025e-08, + "loss": 0.0572, + "step": 18549 + }, + { + "epoch": 7.77, + "learning_rate": 4.40777215700261e-08, + "loss": 0.0776, + "step": 18550 + }, + { + "epoch": 7.77, + "learning_rate": 4.391885631046422e-08, + "loss": 0.0429, + "step": 18551 + }, + { + "epoch": 7.77, + "learning_rate": 4.376027722961973e-08, + "loss": 0.0395, + "step": 18552 + }, + { + "epoch": 7.77, + "learning_rate": 4.3601984332051206e-08, + "loss": 0.0583, + "step": 18553 + }, + { + "epoch": 7.77, + "learning_rate": 4.344397762230834e-08, + "loss": 0.0364, + "step": 18554 + }, + { + "epoch": 7.77, + "learning_rate": 4.328625710493306e-08, + "loss": 0.047, + "step": 18555 + }, + { + "epoch": 7.77, + "learning_rate": 4.31288227844584e-08, + "loss": 0.0377, + "step": 18556 + }, + { + "epoch": 7.77, + "learning_rate": 4.2971674665410744e-08, + "loss": 0.0519, + "step": 18557 + }, + { + "epoch": 7.77, + "learning_rate": 4.2814812752306476e-08, + "loss": 0.038, + "step": 18558 + }, + { + "epoch": 7.77, + "learning_rate": 4.2658237049655325e-08, + "loss": 0.0486, + "step": 18559 + }, + { + "epoch": 7.77, + "learning_rate": 4.2501947561955914e-08, + "loss": 0.038, + "step": 18560 + }, + { + "epoch": 7.77, + "learning_rate": 4.2345944293702424e-08, + "loss": 0.057, + "step": 18561 + }, + { + "epoch": 7.77, + "learning_rate": 4.2190227249377933e-08, + "loss": 0.0457, + "step": 18562 + }, + { + "epoch": 7.77, + "learning_rate": 4.203479643345998e-08, + "loss": 0.0486, + "step": 18563 + }, + { + "epoch": 7.77, + "learning_rate": 4.1879651850414984e-08, + "loss": 0.031, + "step": 18564 + }, + { + "epoch": 7.77, + "learning_rate": 4.172479350470271e-08, + "loss": 0.0395, + "step": 18565 + }, + { + "epoch": 7.77, + "learning_rate": 4.1570221400774045e-08, + "loss": 0.0449, + "step": 18566 + }, + { + "epoch": 7.77, + "learning_rate": 4.1415935543072105e-08, + "loss": 0.066, + "step": 18567 + }, + { + "epoch": 7.77, + "learning_rate": 4.1261935936032226e-08, + "loss": 0.0485, + "step": 18568 + }, + { + "epoch": 7.77, + "learning_rate": 4.110822258408198e-08, + "loss": 0.0539, + "step": 18569 + }, + { + "epoch": 7.77, + "learning_rate": 4.0954795491636725e-08, + "loss": 0.0601, + "step": 18570 + }, + { + "epoch": 7.78, + "learning_rate": 4.080165466310959e-08, + "loss": 0.0432, + "step": 18571 + }, + { + "epoch": 7.78, + "learning_rate": 4.064880010290151e-08, + "loss": 0.023, + "step": 18572 + }, + { + "epoch": 7.78, + "learning_rate": 4.049623181540563e-08, + "loss": 0.0269, + "step": 18573 + }, + { + "epoch": 7.78, + "learning_rate": 4.0343949805006224e-08, + "loss": 0.0517, + "step": 18574 + }, + { + "epoch": 7.78, + "learning_rate": 4.019195407608312e-08, + "loss": 0.0415, + "step": 18575 + }, + { + "epoch": 7.78, + "learning_rate": 4.004024463300393e-08, + "loss": 0.0687, + "step": 18576 + }, + { + "epoch": 7.78, + "learning_rate": 3.988882148012963e-08, + "loss": 0.0628, + "step": 18577 + }, + { + "epoch": 7.78, + "learning_rate": 3.973768462181338e-08, + "loss": 0.073, + "step": 18578 + }, + { + "epoch": 7.78, + "learning_rate": 3.9586834062398385e-08, + "loss": 0.0492, + "step": 18579 + }, + { + "epoch": 7.78, + "learning_rate": 3.943626980622117e-08, + "loss": 0.0359, + "step": 18580 + }, + { + "epoch": 7.78, + "learning_rate": 3.9285991857608285e-08, + "loss": 0.0445, + "step": 18581 + }, + { + "epoch": 7.78, + "learning_rate": 3.913600022088182e-08, + "loss": 0.042, + "step": 18582 + }, + { + "epoch": 7.78, + "learning_rate": 3.898629490035166e-08, + "loss": 0.0442, + "step": 18583 + }, + { + "epoch": 7.78, + "learning_rate": 3.8836875900321036e-08, + "loss": 0.0384, + "step": 18584 + }, + { + "epoch": 7.78, + "learning_rate": 3.8687743225085397e-08, + "loss": 0.0476, + "step": 18585 + }, + { + "epoch": 7.78, + "learning_rate": 3.8538896878931307e-08, + "loss": 0.0588, + "step": 18586 + }, + { + "epoch": 7.78, + "learning_rate": 3.839033686613647e-08, + "loss": 0.0368, + "step": 18587 + }, + { + "epoch": 7.78, + "learning_rate": 3.82420631909719e-08, + "loss": 0.0332, + "step": 18588 + }, + { + "epoch": 7.78, + "learning_rate": 3.8094075857698645e-08, + "loss": 0.0394, + "step": 18589 + }, + { + "epoch": 7.78, + "learning_rate": 3.794637487057218e-08, + "loss": 0.0633, + "step": 18590 + }, + { + "epoch": 7.78, + "learning_rate": 3.779896023383578e-08, + "loss": 0.0437, + "step": 18591 + }, + { + "epoch": 7.78, + "learning_rate": 3.765183195172939e-08, + "loss": 0.0507, + "step": 18592 + }, + { + "epoch": 7.78, + "learning_rate": 3.7504990028479624e-08, + "loss": 0.0437, + "step": 18593 + }, + { + "epoch": 7.78, + "learning_rate": 3.735843446830867e-08, + "loss": 0.0341, + "step": 18594 + }, + { + "epoch": 7.79, + "learning_rate": 3.72121652754287e-08, + "loss": 0.0537, + "step": 18595 + }, + { + "epoch": 7.79, + "learning_rate": 3.706618245404525e-08, + "loss": 0.053, + "step": 18596 + }, + { + "epoch": 7.79, + "learning_rate": 3.692048600835163e-08, + "loss": 0.0479, + "step": 18597 + }, + { + "epoch": 7.79, + "learning_rate": 3.677507594253893e-08, + "loss": 0.0469, + "step": 18598 + }, + { + "epoch": 7.79, + "learning_rate": 3.6629952260784916e-08, + "loss": 0.0682, + "step": 18599 + }, + { + "epoch": 7.79, + "learning_rate": 3.648511496726181e-08, + "loss": 0.0481, + "step": 18600 + }, + { + "epoch": 7.79, + "learning_rate": 3.634056406613185e-08, + "loss": 0.026, + "step": 18601 + }, + { + "epoch": 7.79, + "learning_rate": 3.6196299561551684e-08, + "loss": 0.0359, + "step": 18602 + }, + { + "epoch": 7.79, + "learning_rate": 3.605232145766691e-08, + "loss": 0.0481, + "step": 18603 + }, + { + "epoch": 7.79, + "learning_rate": 3.590862975861642e-08, + "loss": 0.0629, + "step": 18604 + }, + { + "epoch": 7.79, + "learning_rate": 3.5765224468530254e-08, + "loss": 0.0513, + "step": 18605 + }, + { + "epoch": 7.79, + "learning_rate": 3.5622105591530677e-08, + "loss": 0.0601, + "step": 18606 + }, + { + "epoch": 7.79, + "learning_rate": 3.5479273131731054e-08, + "loss": 0.061, + "step": 18607 + }, + { + "epoch": 7.79, + "learning_rate": 3.53367270932381e-08, + "loss": 0.0681, + "step": 18608 + }, + { + "epoch": 7.79, + "learning_rate": 3.519446748014743e-08, + "loss": 0.05, + "step": 18609 + }, + { + "epoch": 7.79, + "learning_rate": 3.505249429655022e-08, + "loss": 0.0466, + "step": 18610 + }, + { + "epoch": 7.79, + "learning_rate": 3.4910807546525426e-08, + "loss": 0.0477, + "step": 18611 + }, + { + "epoch": 7.79, + "learning_rate": 3.476940723414757e-08, + "loss": 0.0587, + "step": 18612 + }, + { + "epoch": 7.79, + "learning_rate": 3.462829336347895e-08, + "loss": 0.0812, + "step": 18613 + }, + { + "epoch": 7.79, + "learning_rate": 3.448746593857744e-08, + "loss": 0.054, + "step": 18614 + }, + { + "epoch": 7.79, + "learning_rate": 3.4346924963490904e-08, + "loss": 0.0418, + "step": 18615 + }, + { + "epoch": 7.79, + "learning_rate": 3.420667044225834e-08, + "loss": 0.039, + "step": 18616 + }, + { + "epoch": 7.79, + "learning_rate": 3.406670237891208e-08, + "loss": 0.0389, + "step": 18617 + }, + { + "epoch": 7.79, + "learning_rate": 3.392702077747445e-08, + "loss": 0.0591, + "step": 18618 + }, + { + "epoch": 7.8, + "learning_rate": 3.378762564196003e-08, + "loss": 0.0511, + "step": 18619 + }, + { + "epoch": 7.8, + "learning_rate": 3.364851697637783e-08, + "loss": 0.0491, + "step": 18620 + }, + { + "epoch": 7.8, + "learning_rate": 3.3509694784724656e-08, + "loss": 0.0381, + "step": 18621 + }, + { + "epoch": 7.8, + "learning_rate": 3.337115907099064e-08, + "loss": 0.0555, + "step": 18622 + }, + { + "epoch": 7.8, + "learning_rate": 3.3232909839158166e-08, + "loss": 0.0475, + "step": 18623 + }, + { + "epoch": 7.8, + "learning_rate": 3.309494709320182e-08, + "loss": 0.0445, + "step": 18624 + }, + { + "epoch": 7.8, + "learning_rate": 3.2957270837087327e-08, + "loss": 0.0582, + "step": 18625 + }, + { + "epoch": 7.8, + "learning_rate": 3.281988107477041e-08, + "loss": 0.0415, + "step": 18626 + }, + { + "epoch": 7.8, + "learning_rate": 3.268277781020235e-08, + "loss": 0.0646, + "step": 18627 + }, + { + "epoch": 7.8, + "learning_rate": 3.254596104732222e-08, + "loss": 0.0427, + "step": 18628 + }, + { + "epoch": 7.8, + "learning_rate": 3.240943079006354e-08, + "loss": 0.0396, + "step": 18629 + }, + { + "epoch": 7.8, + "learning_rate": 3.2273187042350937e-08, + "loss": 0.0349, + "step": 18630 + }, + { + "epoch": 7.8, + "learning_rate": 3.213722980810019e-08, + "loss": 0.039, + "step": 18631 + }, + { + "epoch": 7.8, + "learning_rate": 3.200155909121927e-08, + "loss": 0.0676, + "step": 18632 + }, + { + "epoch": 7.8, + "learning_rate": 3.1866174895609504e-08, + "loss": 0.0424, + "step": 18633 + }, + { + "epoch": 7.8, + "learning_rate": 3.173107722515889e-08, + "loss": 0.0482, + "step": 18634 + }, + { + "epoch": 7.8, + "learning_rate": 3.1596266083754326e-08, + "loss": 0.0396, + "step": 18635 + }, + { + "epoch": 7.8, + "learning_rate": 3.146174147526826e-08, + "loss": 0.0362, + "step": 18636 + }, + { + "epoch": 7.8, + "learning_rate": 3.1327503403568713e-08, + "loss": 0.0699, + "step": 18637 + }, + { + "epoch": 7.8, + "learning_rate": 3.119355187251483e-08, + "loss": 0.0451, + "step": 18638 + }, + { + "epoch": 7.8, + "learning_rate": 3.105988688595574e-08, + "loss": 0.071, + "step": 18639 + }, + { + "epoch": 7.8, + "learning_rate": 3.0926508447733925e-08, + "loss": 0.0353, + "step": 18640 + }, + { + "epoch": 7.8, + "learning_rate": 3.0793416561682996e-08, + "loss": 0.0441, + "step": 18641 + }, + { + "epoch": 7.8, + "learning_rate": 3.066061123162878e-08, + "loss": 0.0409, + "step": 18642 + }, + { + "epoch": 7.81, + "learning_rate": 3.0528092461389325e-08, + "loss": 0.0437, + "step": 18643 + }, + { + "epoch": 7.81, + "learning_rate": 3.0395860254772705e-08, + "loss": 0.0463, + "step": 18644 + }, + { + "epoch": 7.81, + "learning_rate": 3.026391461557921e-08, + "loss": 0.0531, + "step": 18645 + }, + { + "epoch": 7.81, + "learning_rate": 3.013225554760357e-08, + "loss": 0.0588, + "step": 18646 + }, + { + "epoch": 7.81, + "learning_rate": 3.000088305462945e-08, + "loss": 0.0496, + "step": 18647 + }, + { + "epoch": 7.81, + "learning_rate": 2.986979714043159e-08, + "loss": 0.0386, + "step": 18648 + }, + { + "epoch": 7.81, + "learning_rate": 2.9738997808780312e-08, + "loss": 0.0568, + "step": 18649 + }, + { + "epoch": 7.81, + "learning_rate": 2.9608485063432612e-08, + "loss": 0.0283, + "step": 18650 + }, + { + "epoch": 7.81, + "learning_rate": 2.9478258908142153e-08, + "loss": 0.056, + "step": 18651 + }, + { + "epoch": 7.81, + "learning_rate": 2.9348319346651498e-08, + "loss": 0.042, + "step": 18652 + }, + { + "epoch": 7.81, + "learning_rate": 2.9218666382696548e-08, + "loss": 0.0487, + "step": 18653 + }, + { + "epoch": 7.81, + "learning_rate": 2.9089300020002097e-08, + "loss": 0.0411, + "step": 18654 + }, + { + "epoch": 7.81, + "learning_rate": 2.8960220262287398e-08, + "loss": 0.0536, + "step": 18655 + }, + { + "epoch": 7.81, + "learning_rate": 2.883142711326392e-08, + "loss": 0.0357, + "step": 18656 + }, + { + "epoch": 7.81, + "learning_rate": 2.8702920576632042e-08, + "loss": 0.0533, + "step": 18657 + }, + { + "epoch": 7.81, + "learning_rate": 2.8574700656086584e-08, + "loss": 0.0456, + "step": 18658 + }, + { + "epoch": 7.81, + "learning_rate": 2.8446767355313488e-08, + "loss": 0.0533, + "step": 18659 + }, + { + "epoch": 7.81, + "learning_rate": 2.8319120677988698e-08, + "loss": 0.0468, + "step": 18660 + }, + { + "epoch": 7.81, + "learning_rate": 2.8191760627782616e-08, + "loss": 0.057, + "step": 18661 + }, + { + "epoch": 7.81, + "learning_rate": 2.8064687208354534e-08, + "loss": 0.0376, + "step": 18662 + }, + { + "epoch": 7.81, + "learning_rate": 2.793790042335931e-08, + "loss": 0.0451, + "step": 18663 + }, + { + "epoch": 7.81, + "learning_rate": 2.781140027643958e-08, + "loss": 0.049, + "step": 18664 + }, + { + "epoch": 7.81, + "learning_rate": 2.7685186771231333e-08, + "loss": 0.0773, + "step": 18665 + }, + { + "epoch": 7.81, + "learning_rate": 2.7559259911363877e-08, + "loss": 0.0498, + "step": 18666 + }, + { + "epoch": 7.82, + "learning_rate": 2.743361970045655e-08, + "loss": 0.036, + "step": 18667 + }, + { + "epoch": 7.82, + "learning_rate": 2.7308266142119788e-08, + "loss": 0.0504, + "step": 18668 + }, + { + "epoch": 7.82, + "learning_rate": 2.718319923995738e-08, + "loss": 0.059, + "step": 18669 + }, + { + "epoch": 7.82, + "learning_rate": 2.705841899756423e-08, + "loss": 0.0408, + "step": 18670 + }, + { + "epoch": 7.82, + "learning_rate": 2.6933925418526352e-08, + "loss": 0.0384, + "step": 18671 + }, + { + "epoch": 7.82, + "learning_rate": 2.680971850642311e-08, + "loss": 0.0399, + "step": 18672 + }, + { + "epoch": 7.82, + "learning_rate": 2.668579826482498e-08, + "loss": 0.0501, + "step": 18673 + }, + { + "epoch": 7.82, + "learning_rate": 2.656216469729356e-08, + "loss": 0.0422, + "step": 18674 + }, + { + "epoch": 7.82, + "learning_rate": 2.6438817807381556e-08, + "loss": 0.0402, + "step": 18675 + }, + { + "epoch": 7.82, + "learning_rate": 2.631575759863614e-08, + "loss": 0.0441, + "step": 18676 + }, + { + "epoch": 7.82, + "learning_rate": 2.6192984074593364e-08, + "loss": 0.0355, + "step": 18677 + }, + { + "epoch": 7.82, + "learning_rate": 2.6070497238782632e-08, + "loss": 0.0491, + "step": 18678 + }, + { + "epoch": 7.82, + "learning_rate": 2.5948297094724463e-08, + "loss": 0.0732, + "step": 18679 + }, + { + "epoch": 7.82, + "learning_rate": 2.5826383645932706e-08, + "loss": 0.0557, + "step": 18680 + }, + { + "epoch": 7.82, + "learning_rate": 2.5704756895910122e-08, + "loss": 0.0381, + "step": 18681 + }, + { + "epoch": 7.82, + "learning_rate": 2.55834168481528e-08, + "loss": 0.0555, + "step": 18682 + }, + { + "epoch": 7.82, + "learning_rate": 2.5462363506149057e-08, + "loss": 0.0635, + "step": 18683 + }, + { + "epoch": 7.82, + "learning_rate": 2.5341596873379447e-08, + "loss": 0.0497, + "step": 18684 + }, + { + "epoch": 7.82, + "learning_rate": 2.522111695331342e-08, + "loss": 0.0408, + "step": 18685 + }, + { + "epoch": 7.82, + "learning_rate": 2.510092374941486e-08, + "loss": 0.0347, + "step": 18686 + }, + { + "epoch": 7.82, + "learning_rate": 2.4981017265138795e-08, + "loss": 0.0473, + "step": 18687 + }, + { + "epoch": 7.82, + "learning_rate": 2.486139750393246e-08, + "loss": 0.0364, + "step": 18688 + }, + { + "epoch": 7.82, + "learning_rate": 2.4742064469231998e-08, + "loss": 0.0576, + "step": 18689 + }, + { + "epoch": 7.82, + "learning_rate": 2.4623018164470214e-08, + "loss": 0.0521, + "step": 18690 + }, + { + "epoch": 7.83, + "learning_rate": 2.450425859306771e-08, + "loss": 0.0425, + "step": 18691 + }, + { + "epoch": 7.83, + "learning_rate": 2.438578575843731e-08, + "loss": 0.0526, + "step": 18692 + }, + { + "epoch": 7.83, + "learning_rate": 2.4267599663986286e-08, + "loss": 0.0629, + "step": 18693 + }, + { + "epoch": 7.83, + "learning_rate": 2.4149700313110813e-08, + "loss": 0.041, + "step": 18694 + }, + { + "epoch": 7.83, + "learning_rate": 2.4032087709199294e-08, + "loss": 0.0319, + "step": 18695 + }, + { + "epoch": 7.83, + "learning_rate": 2.3914761855632353e-08, + "loss": 0.0576, + "step": 18696 + }, + { + "epoch": 7.83, + "learning_rate": 2.3797722755783957e-08, + "loss": 0.0789, + "step": 18697 + }, + { + "epoch": 7.83, + "learning_rate": 2.368097041301698e-08, + "loss": 0.0592, + "step": 18698 + }, + { + "epoch": 7.83, + "learning_rate": 2.3564504830687617e-08, + "loss": 0.0625, + "step": 18699 + }, + { + "epoch": 7.83, + "learning_rate": 2.3448326012143197e-08, + "loss": 0.049, + "step": 18700 + }, + { + "epoch": 7.83, + "learning_rate": 2.3332433960724375e-08, + "loss": 0.0459, + "step": 18701 + }, + { + "epoch": 7.83, + "learning_rate": 2.3216828679760716e-08, + "loss": 0.03, + "step": 18702 + }, + { + "epoch": 7.83, + "learning_rate": 2.3101510172577335e-08, + "loss": 0.0712, + "step": 18703 + }, + { + "epoch": 7.83, + "learning_rate": 2.2986478442486026e-08, + "loss": 0.0626, + "step": 18704 + }, + { + "epoch": 7.83, + "learning_rate": 2.2871733492796366e-08, + "loss": 0.0527, + "step": 18705 + }, + { + "epoch": 7.83, + "learning_rate": 2.2757275326804606e-08, + "loss": 0.0462, + "step": 18706 + }, + { + "epoch": 7.83, + "learning_rate": 2.264310394780145e-08, + "loss": 0.0658, + "step": 18707 + }, + { + "epoch": 7.83, + "learning_rate": 2.252921935906871e-08, + "loss": 0.0548, + "step": 18708 + }, + { + "epoch": 7.83, + "learning_rate": 2.241562156388044e-08, + "loss": 0.0376, + "step": 18709 + }, + { + "epoch": 7.83, + "learning_rate": 2.2302310565499587e-08, + "loss": 0.0534, + "step": 18710 + }, + { + "epoch": 7.83, + "learning_rate": 2.2189286367185758e-08, + "loss": 0.0365, + "step": 18711 + }, + { + "epoch": 7.83, + "learning_rate": 2.207654897218636e-08, + "loss": 0.0515, + "step": 18712 + }, + { + "epoch": 7.83, + "learning_rate": 2.1964098383742138e-08, + "loss": 0.0346, + "step": 18713 + }, + { + "epoch": 7.84, + "learning_rate": 2.1851934605086057e-08, + "loss": 0.0352, + "step": 18714 + }, + { + "epoch": 7.84, + "learning_rate": 2.1740057639441093e-08, + "loss": 0.037, + "step": 18715 + }, + { + "epoch": 7.84, + "learning_rate": 2.1628467490023565e-08, + "loss": 0.0613, + "step": 18716 + }, + { + "epoch": 7.84, + "learning_rate": 2.1517164160042014e-08, + "loss": 0.0483, + "step": 18717 + }, + { + "epoch": 7.84, + "learning_rate": 2.1406147652693887e-08, + "loss": 0.061, + "step": 18718 + }, + { + "epoch": 7.84, + "learning_rate": 2.1295417971172184e-08, + "loss": 0.047, + "step": 18719 + }, + { + "epoch": 7.84, + "learning_rate": 2.1184975118657692e-08, + "loss": 0.0343, + "step": 18720 + }, + { + "epoch": 7.84, + "learning_rate": 2.107481909832676e-08, + "loss": 0.0623, + "step": 18721 + }, + { + "epoch": 7.84, + "learning_rate": 2.0964949913344635e-08, + "loss": 0.0503, + "step": 18722 + }, + { + "epoch": 7.84, + "learning_rate": 2.08553675668699e-08, + "loss": 0.045, + "step": 18723 + }, + { + "epoch": 7.84, + "learning_rate": 2.074607206205226e-08, + "loss": 0.0531, + "step": 18724 + }, + { + "epoch": 7.84, + "learning_rate": 2.0637063402033642e-08, + "loss": 0.0544, + "step": 18725 + }, + { + "epoch": 7.84, + "learning_rate": 2.0528341589945987e-08, + "loss": 0.0452, + "step": 18726 + }, + { + "epoch": 7.84, + "learning_rate": 2.04199066289168e-08, + "loss": 0.0463, + "step": 18727 + }, + { + "epoch": 7.84, + "learning_rate": 2.0311758522061355e-08, + "loss": 0.0462, + "step": 18728 + }, + { + "epoch": 7.84, + "learning_rate": 2.0203897272488283e-08, + "loss": 0.0365, + "step": 18729 + }, + { + "epoch": 7.84, + "learning_rate": 2.0096322883298435e-08, + "loss": 0.0499, + "step": 18730 + }, + { + "epoch": 7.84, + "learning_rate": 1.9989035357583785e-08, + "loss": 0.0336, + "step": 18731 + }, + { + "epoch": 7.84, + "learning_rate": 1.988203469842853e-08, + "loss": 0.057, + "step": 18732 + }, + { + "epoch": 7.84, + "learning_rate": 1.9775320908907992e-08, + "loss": 0.0614, + "step": 18733 + }, + { + "epoch": 7.84, + "learning_rate": 1.96688939920886e-08, + "loss": 0.0433, + "step": 18734 + }, + { + "epoch": 7.84, + "learning_rate": 1.9562753951031244e-08, + "loss": 0.0456, + "step": 18735 + }, + { + "epoch": 7.84, + "learning_rate": 1.9456900788786814e-08, + "loss": 0.0569, + "step": 18736 + }, + { + "epoch": 7.84, + "learning_rate": 1.9351334508396212e-08, + "loss": 0.048, + "step": 18737 + }, + { + "epoch": 7.85, + "learning_rate": 1.924605511289479e-08, + "loss": 0.0558, + "step": 18738 + }, + { + "epoch": 7.85, + "learning_rate": 1.9141062605309013e-08, + "loss": 0.0293, + "step": 18739 + }, + { + "epoch": 7.85, + "learning_rate": 1.9036356988655357e-08, + "loss": 0.0478, + "step": 18740 + }, + { + "epoch": 7.85, + "learning_rate": 1.893193826594586e-08, + "loss": 0.054, + "step": 18741 + }, + { + "epoch": 7.85, + "learning_rate": 1.8827806440181452e-08, + "loss": 0.0379, + "step": 18742 + }, + { + "epoch": 7.85, + "learning_rate": 1.872396151435307e-08, + "loss": 0.0671, + "step": 18743 + }, + { + "epoch": 7.85, + "learning_rate": 1.8620403491448334e-08, + "loss": 0.0569, + "step": 18744 + }, + { + "epoch": 7.85, + "learning_rate": 1.8517132374442636e-08, + "loss": 0.0433, + "step": 18745 + }, + { + "epoch": 7.85, + "learning_rate": 1.8414148166303603e-08, + "loss": 0.0435, + "step": 18746 + }, + { + "epoch": 7.85, + "learning_rate": 1.831145086999331e-08, + "loss": 0.0542, + "step": 18747 + }, + { + "epoch": 7.85, + "learning_rate": 1.8209040488462725e-08, + "loss": 0.0643, + "step": 18748 + }, + { + "epoch": 7.85, + "learning_rate": 1.8106917024655056e-08, + "loss": 0.0552, + "step": 18749 + }, + { + "epoch": 7.85, + "learning_rate": 1.800508048150684e-08, + "loss": 0.0314, + "step": 18750 + }, + { + "epoch": 7.85, + "learning_rate": 1.790353086194463e-08, + "loss": 0.0444, + "step": 18751 + }, + { + "epoch": 7.85, + "learning_rate": 1.7802268168887195e-08, + "loss": 0.0381, + "step": 18752 + }, + { + "epoch": 7.85, + "learning_rate": 1.770129240524554e-08, + "loss": 0.0374, + "step": 18753 + }, + { + "epoch": 7.85, + "learning_rate": 1.7600603573921794e-08, + "loss": 0.0617, + "step": 18754 + }, + { + "epoch": 7.85, + "learning_rate": 1.7500201677811414e-08, + "loss": 0.0455, + "step": 18755 + }, + { + "epoch": 7.85, + "learning_rate": 1.740008671979876e-08, + "loss": 0.0612, + "step": 18756 + }, + { + "epoch": 7.85, + "learning_rate": 1.730025870276264e-08, + "loss": 0.0338, + "step": 18757 + }, + { + "epoch": 7.85, + "learning_rate": 1.720071762957187e-08, + "loss": 0.0399, + "step": 18758 + }, + { + "epoch": 7.85, + "learning_rate": 1.71014635030875e-08, + "loss": 0.037, + "step": 18759 + }, + { + "epoch": 7.85, + "learning_rate": 1.70024963261628e-08, + "loss": 0.0506, + "step": 18760 + }, + { + "epoch": 7.85, + "learning_rate": 1.6903816101643267e-08, + "loss": 0.0563, + "step": 18761 + }, + { + "epoch": 7.86, + "learning_rate": 1.6805422832364414e-08, + "loss": 0.0376, + "step": 18762 + }, + { + "epoch": 7.86, + "learning_rate": 1.6707316521155093e-08, + "loss": 0.0475, + "step": 18763 + }, + { + "epoch": 7.86, + "learning_rate": 1.6609497170834154e-08, + "loss": 0.0729, + "step": 18764 + }, + { + "epoch": 7.86, + "learning_rate": 1.6511964784214907e-08, + "loss": 0.0427, + "step": 18765 + }, + { + "epoch": 7.86, + "learning_rate": 1.6414719364099552e-08, + "loss": 0.0478, + "step": 18766 + }, + { + "epoch": 7.86, + "learning_rate": 1.631776091328363e-08, + "loss": 0.0319, + "step": 18767 + }, + { + "epoch": 7.86, + "learning_rate": 1.6221089434554916e-08, + "loss": 0.0326, + "step": 18768 + }, + { + "epoch": 7.86, + "learning_rate": 1.612470493069118e-08, + "loss": 0.0592, + "step": 18769 + }, + { + "epoch": 7.86, + "learning_rate": 1.6028607404463547e-08, + "loss": 0.0405, + "step": 18770 + }, + { + "epoch": 7.86, + "learning_rate": 1.593279685863314e-08, + "loss": 0.0351, + "step": 18771 + }, + { + "epoch": 7.86, + "learning_rate": 1.5837273295955524e-08, + "loss": 0.0516, + "step": 18772 + }, + { + "epoch": 7.86, + "learning_rate": 1.574203671917629e-08, + "loss": 0.0365, + "step": 18773 + }, + { + "epoch": 7.86, + "learning_rate": 1.5647087131031023e-08, + "loss": 0.0584, + "step": 18774 + }, + { + "epoch": 7.86, + "learning_rate": 1.5552424534250876e-08, + "loss": 0.0532, + "step": 18775 + }, + { + "epoch": 7.86, + "learning_rate": 1.5458048931557002e-08, + "loss": 0.0438, + "step": 18776 + }, + { + "epoch": 7.86, + "learning_rate": 1.5363960325660565e-08, + "loss": 0.0395, + "step": 18777 + }, + { + "epoch": 7.86, + "learning_rate": 1.527015871926718e-08, + "loss": 0.0387, + "step": 18778 + }, + { + "epoch": 7.86, + "learning_rate": 1.517664411507247e-08, + "loss": 0.0553, + "step": 18779 + }, + { + "epoch": 7.86, + "learning_rate": 1.508341651576428e-08, + "loss": 0.0441, + "step": 18780 + }, + { + "epoch": 7.86, + "learning_rate": 1.4990475924023805e-08, + "loss": 0.0365, + "step": 18781 + }, + { + "epoch": 7.86, + "learning_rate": 1.4897822342521128e-08, + "loss": 0.0629, + "step": 18782 + }, + { + "epoch": 7.86, + "learning_rate": 1.4805455773919674e-08, + "loss": 0.048, + "step": 18783 + }, + { + "epoch": 7.86, + "learning_rate": 1.4713376220875097e-08, + "loss": 0.0478, + "step": 18784 + }, + { + "epoch": 7.86, + "learning_rate": 1.4621583686034169e-08, + "loss": 0.0553, + "step": 18785 + }, + { + "epoch": 7.87, + "learning_rate": 1.4530078172033669e-08, + "loss": 0.0357, + "step": 18786 + }, + { + "epoch": 7.87, + "learning_rate": 1.4438859681505935e-08, + "loss": 0.0346, + "step": 18787 + }, + { + "epoch": 7.87, + "learning_rate": 1.4347928217072205e-08, + "loss": 0.0639, + "step": 18788 + }, + { + "epoch": 7.87, + "learning_rate": 1.4257283781345943e-08, + "loss": 0.0427, + "step": 18789 + }, + { + "epoch": 7.87, + "learning_rate": 1.4166926376932843e-08, + "loss": 0.0477, + "step": 18790 + }, + { + "epoch": 7.87, + "learning_rate": 1.4076856006430828e-08, + "loss": 0.0519, + "step": 18791 + }, + { + "epoch": 7.87, + "learning_rate": 1.3987072672427827e-08, + "loss": 0.0408, + "step": 18792 + }, + { + "epoch": 7.87, + "learning_rate": 1.3897576377503996e-08, + "loss": 0.0455, + "step": 18793 + }, + { + "epoch": 7.87, + "learning_rate": 1.3808367124233945e-08, + "loss": 0.0512, + "step": 18794 + }, + { + "epoch": 7.87, + "learning_rate": 1.371944491518118e-08, + "loss": 0.0484, + "step": 18795 + }, + { + "epoch": 7.87, + "learning_rate": 1.363080975290032e-08, + "loss": 0.0457, + "step": 18796 + }, + { + "epoch": 7.87, + "learning_rate": 1.354246163994044e-08, + "loss": 0.044, + "step": 18797 + }, + { + "epoch": 7.87, + "learning_rate": 1.345440057884062e-08, + "loss": 0.0344, + "step": 18798 + }, + { + "epoch": 7.87, + "learning_rate": 1.336662657213217e-08, + "loss": 0.0477, + "step": 18799 + }, + { + "epoch": 7.87, + "learning_rate": 1.3279139622337512e-08, + "loss": 0.0592, + "step": 18800 + }, + { + "epoch": 7.87, + "learning_rate": 1.3191939731972414e-08, + "loss": 0.0622, + "step": 18801 + }, + { + "epoch": 7.87, + "learning_rate": 1.310502690354376e-08, + "loss": 0.0464, + "step": 18802 + }, + { + "epoch": 7.87, + "learning_rate": 1.3018401139548441e-08, + "loss": 0.0467, + "step": 18803 + }, + { + "epoch": 7.87, + "learning_rate": 1.2932062442475579e-08, + "loss": 0.0579, + "step": 18804 + }, + { + "epoch": 7.87, + "learning_rate": 1.284601081480985e-08, + "loss": 0.0445, + "step": 18805 + }, + { + "epoch": 7.87, + "learning_rate": 1.2760246259021503e-08, + "loss": 0.043, + "step": 18806 + }, + { + "epoch": 7.87, + "learning_rate": 1.2674768777578562e-08, + "loss": 0.0556, + "step": 18807 + }, + { + "epoch": 7.87, + "learning_rate": 1.258957837293684e-08, + "loss": 0.0445, + "step": 18808 + }, + { + "epoch": 7.87, + "learning_rate": 1.2504675047544379e-08, + "loss": 0.0539, + "step": 18809 + }, + { + "epoch": 7.88, + "learning_rate": 1.242005880384256e-08, + "loss": 0.038, + "step": 18810 + }, + { + "epoch": 7.88, + "learning_rate": 1.2335729644263882e-08, + "loss": 0.0474, + "step": 18811 + }, + { + "epoch": 7.88, + "learning_rate": 1.2251687571230853e-08, + "loss": 0.058, + "step": 18812 + }, + { + "epoch": 7.88, + "learning_rate": 1.2167932587160425e-08, + "loss": 0.0309, + "step": 18813 + }, + { + "epoch": 7.88, + "learning_rate": 1.2084464694459563e-08, + "loss": 0.0426, + "step": 18814 + }, + { + "epoch": 7.88, + "learning_rate": 1.200128389552857e-08, + "loss": 0.0314, + "step": 18815 + }, + { + "epoch": 7.88, + "learning_rate": 1.1918390192756646e-08, + "loss": 0.0452, + "step": 18816 + }, + { + "epoch": 7.88, + "learning_rate": 1.1835783588527438e-08, + "loss": 0.0374, + "step": 18817 + }, + { + "epoch": 7.88, + "learning_rate": 1.1753464085214605e-08, + "loss": 0.0611, + "step": 18818 + }, + { + "epoch": 7.88, + "learning_rate": 1.1671431685186252e-08, + "loss": 0.0487, + "step": 18819 + }, + { + "epoch": 7.88, + "learning_rate": 1.1589686390798272e-08, + "loss": 0.0504, + "step": 18820 + }, + { + "epoch": 7.88, + "learning_rate": 1.1508228204402116e-08, + "loss": 0.0609, + "step": 18821 + }, + { + "epoch": 7.88, + "learning_rate": 1.1427057128337026e-08, + "loss": 0.0435, + "step": 18822 + }, + { + "epoch": 7.88, + "learning_rate": 1.1346173164938913e-08, + "loss": 0.0394, + "step": 18823 + }, + { + "epoch": 7.88, + "learning_rate": 1.1265576316530358e-08, + "loss": 0.0647, + "step": 18824 + }, + { + "epoch": 7.88, + "learning_rate": 1.118526658542951e-08, + "loss": 0.048, + "step": 18825 + }, + { + "epoch": 7.88, + "learning_rate": 1.1105243973943413e-08, + "loss": 0.0576, + "step": 18826 + }, + { + "epoch": 7.88, + "learning_rate": 1.102550848437356e-08, + "loss": 0.0446, + "step": 18827 + }, + { + "epoch": 7.88, + "learning_rate": 1.0946060119011448e-08, + "loss": 0.0445, + "step": 18828 + }, + { + "epoch": 7.88, + "learning_rate": 1.0866898880140808e-08, + "loss": 0.032, + "step": 18829 + }, + { + "epoch": 7.88, + "learning_rate": 1.0788024770036487e-08, + "loss": 0.0517, + "step": 18830 + }, + { + "epoch": 7.88, + "learning_rate": 1.070943779096667e-08, + "loss": 0.0554, + "step": 18831 + }, + { + "epoch": 7.88, + "learning_rate": 1.063113794518955e-08, + "loss": 0.041, + "step": 18832 + }, + { + "epoch": 7.88, + "learning_rate": 1.0553125234955553e-08, + "loss": 0.0669, + "step": 18833 + }, + { + "epoch": 7.89, + "learning_rate": 1.0475399662508434e-08, + "loss": 0.0571, + "step": 18834 + }, + { + "epoch": 7.89, + "learning_rate": 1.0397961230080855e-08, + "loss": 0.0712, + "step": 18835 + }, + { + "epoch": 7.89, + "learning_rate": 1.0320809939898813e-08, + "loss": 0.0419, + "step": 18836 + }, + { + "epoch": 7.89, + "learning_rate": 1.0243945794181643e-08, + "loss": 0.0407, + "step": 18837 + }, + { + "epoch": 7.89, + "learning_rate": 1.0167368795136467e-08, + "loss": 0.0509, + "step": 18838 + }, + { + "epoch": 7.89, + "learning_rate": 1.0091078944964861e-08, + "loss": 0.0469, + "step": 18839 + }, + { + "epoch": 7.89, + "learning_rate": 1.0015076245861732e-08, + "loss": 0.0554, + "step": 18840 + }, + { + "epoch": 7.89, + "learning_rate": 9.939360700009782e-09, + "loss": 0.058, + "step": 18841 + }, + { + "epoch": 7.89, + "learning_rate": 9.863932309585044e-09, + "loss": 0.0546, + "step": 18842 + }, + { + "epoch": 7.89, + "learning_rate": 9.788791076758009e-09, + "loss": 0.0596, + "step": 18843 + }, + { + "epoch": 7.89, + "learning_rate": 9.713937003685836e-09, + "loss": 0.048, + "step": 18844 + }, + { + "epoch": 7.89, + "learning_rate": 9.63937009252236e-09, + "loss": 0.0426, + "step": 18845 + }, + { + "epoch": 7.89, + "learning_rate": 9.5650903454092e-09, + "loss": 0.0453, + "step": 18846 + }, + { + "epoch": 7.89, + "learning_rate": 9.491097764481316e-09, + "loss": 0.0373, + "step": 18847 + }, + { + "epoch": 7.89, + "learning_rate": 9.417392351868116e-09, + "loss": 0.038, + "step": 18848 + }, + { + "epoch": 7.89, + "learning_rate": 9.343974109685684e-09, + "loss": 0.0465, + "step": 18849 + }, + { + "epoch": 7.89, + "learning_rate": 9.270843040044553e-09, + "loss": 0.0528, + "step": 18850 + }, + { + "epoch": 7.89, + "learning_rate": 9.197999145048597e-09, + "loss": 0.0375, + "step": 18851 + }, + { + "epoch": 7.89, + "learning_rate": 9.125442426789478e-09, + "loss": 0.0471, + "step": 18852 + }, + { + "epoch": 7.89, + "learning_rate": 9.05317288735441e-09, + "loss": 0.0474, + "step": 18853 + }, + { + "epoch": 7.89, + "learning_rate": 8.981190528819517e-09, + "loss": 0.0502, + "step": 18854 + }, + { + "epoch": 7.89, + "learning_rate": 8.90949535325536e-09, + "loss": 0.0505, + "step": 18855 + }, + { + "epoch": 7.89, + "learning_rate": 8.838087362721404e-09, + "loss": 0.0682, + "step": 18856 + }, + { + "epoch": 7.89, + "learning_rate": 8.766966559270451e-09, + "loss": 0.0501, + "step": 18857 + }, + { + "epoch": 7.9, + "learning_rate": 8.696132944946422e-09, + "loss": 0.048, + "step": 18858 + }, + { + "epoch": 7.9, + "learning_rate": 8.625586521786577e-09, + "loss": 0.0481, + "step": 18859 + }, + { + "epoch": 7.9, + "learning_rate": 8.555327291818182e-09, + "loss": 0.0447, + "step": 18860 + }, + { + "epoch": 7.9, + "learning_rate": 8.485355257060734e-09, + "loss": 0.0438, + "step": 18861 + }, + { + "epoch": 7.9, + "learning_rate": 8.415670419524845e-09, + "loss": 0.0483, + "step": 18862 + }, + { + "epoch": 7.9, + "learning_rate": 8.34627278121447e-09, + "loss": 0.0544, + "step": 18863 + }, + { + "epoch": 7.9, + "learning_rate": 8.277162344123568e-09, + "loss": 0.0593, + "step": 18864 + }, + { + "epoch": 7.9, + "learning_rate": 8.20833911024055e-09, + "loss": 0.0641, + "step": 18865 + }, + { + "epoch": 7.9, + "learning_rate": 8.139803081540499e-09, + "loss": 0.0345, + "step": 18866 + }, + { + "epoch": 7.9, + "learning_rate": 8.071554259996283e-09, + "loss": 0.0442, + "step": 18867 + }, + { + "epoch": 7.9, + "learning_rate": 8.003592647568559e-09, + "loss": 0.0666, + "step": 18868 + }, + { + "epoch": 7.9, + "learning_rate": 7.935918246210206e-09, + "loss": 0.0498, + "step": 18869 + }, + { + "epoch": 7.9, + "learning_rate": 7.868531057867445e-09, + "loss": 0.0487, + "step": 18870 + }, + { + "epoch": 7.9, + "learning_rate": 7.801431084477618e-09, + "loss": 0.0575, + "step": 18871 + }, + { + "epoch": 7.9, + "learning_rate": 7.734618327968069e-09, + "loss": 0.03, + "step": 18872 + }, + { + "epoch": 7.9, + "learning_rate": 7.668092790259484e-09, + "loss": 0.0628, + "step": 18873 + }, + { + "epoch": 7.9, + "learning_rate": 7.60185447326478e-09, + "loss": 0.0436, + "step": 18874 + }, + { + "epoch": 7.9, + "learning_rate": 7.535903378887988e-09, + "loss": 0.0421, + "step": 18875 + }, + { + "epoch": 7.9, + "learning_rate": 7.470239509024258e-09, + "loss": 0.0411, + "step": 18876 + }, + { + "epoch": 7.9, + "learning_rate": 7.404862865560969e-09, + "loss": 0.0601, + "step": 18877 + }, + { + "epoch": 7.9, + "learning_rate": 7.33977345037773e-09, + "loss": 0.0532, + "step": 18878 + }, + { + "epoch": 7.9, + "learning_rate": 7.2749712653463756e-09, + "loss": 0.0517, + "step": 18879 + }, + { + "epoch": 7.9, + "learning_rate": 7.21045631232764e-09, + "loss": 0.0522, + "step": 18880 + }, + { + "epoch": 7.9, + "learning_rate": 7.146228593176707e-09, + "loss": 0.0609, + "step": 18881 + }, + { + "epoch": 7.91, + "learning_rate": 7.082288109740987e-09, + "loss": 0.0584, + "step": 18882 + }, + { + "epoch": 7.91, + "learning_rate": 7.018634863856788e-09, + "loss": 0.0631, + "step": 18883 + }, + { + "epoch": 7.91, + "learning_rate": 6.955268857353759e-09, + "loss": 0.0386, + "step": 18884 + }, + { + "epoch": 7.91, + "learning_rate": 6.892190092054885e-09, + "loss": 0.0589, + "step": 18885 + }, + { + "epoch": 7.91, + "learning_rate": 6.82939856977094e-09, + "loss": 0.0475, + "step": 18886 + }, + { + "epoch": 7.91, + "learning_rate": 6.766894292309367e-09, + "loss": 0.0332, + "step": 18887 + }, + { + "epoch": 7.91, + "learning_rate": 6.704677261464287e-09, + "loss": 0.0722, + "step": 18888 + }, + { + "epoch": 7.91, + "learning_rate": 6.64274747902649e-09, + "loss": 0.0487, + "step": 18889 + }, + { + "epoch": 7.91, + "learning_rate": 6.581104946774552e-09, + "loss": 0.062, + "step": 18890 + }, + { + "epoch": 7.91, + "learning_rate": 6.519749666481501e-09, + "loss": 0.0777, + "step": 18891 + }, + { + "epoch": 7.91, + "learning_rate": 6.458681639909259e-09, + "loss": 0.0363, + "step": 18892 + }, + { + "epoch": 7.91, + "learning_rate": 6.397900868814199e-09, + "loss": 0.0312, + "step": 18893 + }, + { + "epoch": 7.91, + "learning_rate": 6.3374073549438144e-09, + "loss": 0.0555, + "step": 18894 + }, + { + "epoch": 7.91, + "learning_rate": 6.277201100036712e-09, + "loss": 0.0323, + "step": 18895 + }, + { + "epoch": 7.91, + "learning_rate": 6.2172821058226196e-09, + "loss": 0.0629, + "step": 18896 + }, + { + "epoch": 7.91, + "learning_rate": 6.157650374023494e-09, + "loss": 0.0639, + "step": 18897 + }, + { + "epoch": 7.91, + "learning_rate": 6.098305906355739e-09, + "loss": 0.0521, + "step": 18898 + }, + { + "epoch": 7.91, + "learning_rate": 6.039248704523548e-09, + "loss": 0.0593, + "step": 18899 + }, + { + "epoch": 7.91, + "learning_rate": 5.980478770224452e-09, + "loss": 0.0407, + "step": 18900 + }, + { + "epoch": 7.91, + "learning_rate": 5.921996105147099e-09, + "loss": 0.0735, + "step": 18901 + }, + { + "epoch": 7.91, + "learning_rate": 5.863800710974588e-09, + "loss": 0.057, + "step": 18902 + }, + { + "epoch": 7.91, + "learning_rate": 5.805892589377804e-09, + "loss": 0.0584, + "step": 18903 + }, + { + "epoch": 7.91, + "learning_rate": 5.748271742023192e-09, + "loss": 0.0501, + "step": 18904 + }, + { + "epoch": 7.92, + "learning_rate": 5.690938170563876e-09, + "loss": 0.0345, + "step": 18905 + }, + { + "epoch": 7.92, + "learning_rate": 5.633891876650755e-09, + "loss": 0.0441, + "step": 18906 + }, + { + "epoch": 7.92, + "learning_rate": 5.577132861922518e-09, + "loss": 0.0659, + "step": 18907 + }, + { + "epoch": 7.92, + "learning_rate": 5.520661128011196e-09, + "loss": 0.0664, + "step": 18908 + }, + { + "epoch": 7.92, + "learning_rate": 5.464476676538821e-09, + "loss": 0.0475, + "step": 18909 + }, + { + "epoch": 7.92, + "learning_rate": 5.40857950912077e-09, + "loss": 0.0483, + "step": 18910 + }, + { + "epoch": 7.92, + "learning_rate": 5.352969627364646e-09, + "loss": 0.0289, + "step": 18911 + }, + { + "epoch": 7.92, + "learning_rate": 5.297647032869169e-09, + "loss": 0.0545, + "step": 18912 + }, + { + "epoch": 7.92, + "learning_rate": 5.242611727221958e-09, + "loss": 0.0543, + "step": 18913 + }, + { + "epoch": 7.92, + "learning_rate": 5.187863712007302e-09, + "loss": 0.0578, + "step": 18914 + }, + { + "epoch": 7.92, + "learning_rate": 5.133402988798386e-09, + "loss": 0.0533, + "step": 18915 + }, + { + "epoch": 7.92, + "learning_rate": 5.079229559160626e-09, + "loss": 0.0504, + "step": 18916 + }, + { + "epoch": 7.92, + "learning_rate": 5.025343424651663e-09, + "loss": 0.0543, + "step": 18917 + }, + { + "epoch": 7.92, + "learning_rate": 4.971744586819149e-09, + "loss": 0.0462, + "step": 18918 + }, + { + "epoch": 7.92, + "learning_rate": 4.918433047204075e-09, + "loss": 0.0544, + "step": 18919 + }, + { + "epoch": 7.92, + "learning_rate": 4.865408807339655e-09, + "loss": 0.0476, + "step": 18920 + }, + { + "epoch": 7.92, + "learning_rate": 4.812671868749119e-09, + "loss": 0.0469, + "step": 18921 + }, + { + "epoch": 7.92, + "learning_rate": 4.76022223294903e-09, + "loss": 0.0658, + "step": 18922 + }, + { + "epoch": 7.92, + "learning_rate": 4.7080599014470705e-09, + "loss": 0.05, + "step": 18923 + }, + { + "epoch": 7.92, + "learning_rate": 4.6561848757420425e-09, + "loss": 0.0531, + "step": 18924 + }, + { + "epoch": 7.92, + "learning_rate": 4.604597157324975e-09, + "loss": 0.0396, + "step": 18925 + }, + { + "epoch": 7.92, + "learning_rate": 4.553296747679125e-09, + "loss": 0.046, + "step": 18926 + }, + { + "epoch": 7.92, + "learning_rate": 4.502283648278871e-09, + "loss": 0.0542, + "step": 18927 + }, + { + "epoch": 7.92, + "learning_rate": 4.451557860590816e-09, + "loss": 0.0373, + "step": 18928 + }, + { + "epoch": 7.93, + "learning_rate": 4.401119386072683e-09, + "loss": 0.0404, + "step": 18929 + }, + { + "epoch": 7.93, + "learning_rate": 4.350968226174423e-09, + "loss": 0.0698, + "step": 18930 + }, + { + "epoch": 7.93, + "learning_rate": 4.301104382338217e-09, + "loss": 0.0423, + "step": 18931 + }, + { + "epoch": 7.93, + "learning_rate": 4.251527855996251e-09, + "loss": 0.0415, + "step": 18932 + }, + { + "epoch": 7.93, + "learning_rate": 4.202238648574053e-09, + "loss": 0.0588, + "step": 18933 + }, + { + "epoch": 7.93, + "learning_rate": 4.153236761488266e-09, + "loss": 0.0399, + "step": 18934 + }, + { + "epoch": 7.93, + "learning_rate": 4.1045221961477646e-09, + "loss": 0.0413, + "step": 18935 + }, + { + "epoch": 7.93, + "learning_rate": 4.056094953952538e-09, + "loss": 0.0467, + "step": 18936 + }, + { + "epoch": 7.93, + "learning_rate": 4.007955036294808e-09, + "loss": 0.0366, + "step": 18937 + }, + { + "epoch": 7.93, + "learning_rate": 3.960102444557912e-09, + "loss": 0.0394, + "step": 18938 + }, + { + "epoch": 7.93, + "learning_rate": 3.912537180117415e-09, + "loss": 0.046, + "step": 18939 + }, + { + "epoch": 7.93, + "learning_rate": 3.865259244340003e-09, + "loss": 0.0437, + "step": 18940 + }, + { + "epoch": 7.93, + "learning_rate": 3.818268638586809e-09, + "loss": 0.0555, + "step": 18941 + }, + { + "epoch": 7.93, + "learning_rate": 3.771565364205643e-09, + "loss": 0.0665, + "step": 18942 + }, + { + "epoch": 7.93, + "learning_rate": 3.7251494225409855e-09, + "loss": 0.052, + "step": 18943 + }, + { + "epoch": 7.93, + "learning_rate": 3.679020814926215e-09, + "loss": 0.0275, + "step": 18944 + }, + { + "epoch": 7.93, + "learning_rate": 3.633179542688048e-09, + "loss": 0.0419, + "step": 18945 + }, + { + "epoch": 7.93, + "learning_rate": 3.587625607143208e-09, + "loss": 0.0551, + "step": 18946 + }, + { + "epoch": 7.93, + "learning_rate": 3.542359009600649e-09, + "loss": 0.0553, + "step": 18947 + }, + { + "epoch": 7.93, + "learning_rate": 3.497379751362662e-09, + "loss": 0.0483, + "step": 18948 + }, + { + "epoch": 7.93, + "learning_rate": 3.4526878337226565e-09, + "loss": 0.0424, + "step": 18949 + }, + { + "epoch": 7.93, + "learning_rate": 3.408283257962941e-09, + "loss": 0.0603, + "step": 18950 + }, + { + "epoch": 7.93, + "learning_rate": 3.364166025362492e-09, + "loss": 0.0459, + "step": 18951 + }, + { + "epoch": 7.93, + "learning_rate": 3.320336137188074e-09, + "loss": 0.0425, + "step": 18952 + }, + { + "epoch": 7.94, + "learning_rate": 3.2767935946986794e-09, + "loss": 0.0311, + "step": 18953 + }, + { + "epoch": 7.94, + "learning_rate": 3.2335383991477507e-09, + "loss": 0.031, + "step": 18954 + }, + { + "epoch": 7.94, + "learning_rate": 3.1905705517776275e-09, + "loss": 0.0594, + "step": 18955 + }, + { + "epoch": 7.94, + "learning_rate": 3.1478900538239877e-09, + "loss": 0.048, + "step": 18956 + }, + { + "epoch": 7.94, + "learning_rate": 3.1054969065125174e-09, + "loss": 0.0366, + "step": 18957 + }, + { + "epoch": 7.94, + "learning_rate": 3.0633911110622417e-09, + "loss": 0.0571, + "step": 18958 + }, + { + "epoch": 7.94, + "learning_rate": 3.021572668684414e-09, + "loss": 0.0378, + "step": 18959 + }, + { + "epoch": 7.94, + "learning_rate": 2.9800415805802953e-09, + "loss": 0.05, + "step": 18960 + }, + { + "epoch": 7.94, + "learning_rate": 2.9387978479433754e-09, + "loss": 0.0345, + "step": 18961 + }, + { + "epoch": 7.94, + "learning_rate": 2.8978414719593726e-09, + "loss": 0.0558, + "step": 18962 + }, + { + "epoch": 7.94, + "learning_rate": 2.857172453806234e-09, + "loss": 0.0615, + "step": 18963 + }, + { + "epoch": 7.94, + "learning_rate": 2.8167907946519134e-09, + "loss": 0.0464, + "step": 18964 + }, + { + "epoch": 7.94, + "learning_rate": 2.776696495657705e-09, + "loss": 0.0455, + "step": 18965 + }, + { + "epoch": 7.94, + "learning_rate": 2.736889557976019e-09, + "loss": 0.0282, + "step": 18966 + }, + { + "epoch": 7.94, + "learning_rate": 2.6973699827503864e-09, + "loss": 0.0409, + "step": 18967 + }, + { + "epoch": 7.94, + "learning_rate": 2.6581377711176747e-09, + "loss": 0.04, + "step": 18968 + }, + { + "epoch": 7.94, + "learning_rate": 2.619192924205871e-09, + "loss": 0.0589, + "step": 18969 + }, + { + "epoch": 7.94, + "learning_rate": 2.58053544313297e-09, + "loss": 0.0428, + "step": 18970 + }, + { + "epoch": 7.94, + "learning_rate": 2.542165329012525e-09, + "loss": 0.0504, + "step": 18971 + }, + { + "epoch": 7.94, + "learning_rate": 2.504082582944767e-09, + "loss": 0.0526, + "step": 18972 + }, + { + "epoch": 7.94, + "learning_rate": 2.466287206025486e-09, + "loss": 0.0477, + "step": 18973 + }, + { + "epoch": 7.94, + "learning_rate": 2.428779199341591e-09, + "loss": 0.0429, + "step": 18974 + }, + { + "epoch": 7.94, + "learning_rate": 2.3915585639699977e-09, + "loss": 0.0603, + "step": 18975 + }, + { + "epoch": 7.94, + "learning_rate": 2.3546253009820717e-09, + "loss": 0.0509, + "step": 18976 + }, + { + "epoch": 7.95, + "learning_rate": 2.3179794114391863e-09, + "loss": 0.0395, + "step": 18977 + }, + { + "epoch": 7.95, + "learning_rate": 2.281620896392722e-09, + "loss": 0.0408, + "step": 18978 + }, + { + "epoch": 7.95, + "learning_rate": 2.2455497568907304e-09, + "loss": 0.0516, + "step": 18979 + }, + { + "epoch": 7.95, + "learning_rate": 2.2097659939679383e-09, + "loss": 0.0471, + "step": 18980 + }, + { + "epoch": 7.95, + "learning_rate": 2.174269608653523e-09, + "loss": 0.0664, + "step": 18981 + }, + { + "epoch": 7.95, + "learning_rate": 2.1390606019677796e-09, + "loss": 0.0344, + "step": 18982 + }, + { + "epoch": 7.95, + "learning_rate": 2.1041389749221208e-09, + "loss": 0.06, + "step": 18983 + }, + { + "epoch": 7.95, + "learning_rate": 2.069504728522409e-09, + "loss": 0.0398, + "step": 18984 + }, + { + "epoch": 7.95, + "learning_rate": 2.0351578637611834e-09, + "loss": 0.0421, + "step": 18985 + }, + { + "epoch": 7.95, + "learning_rate": 2.001098381628763e-09, + "loss": 0.0462, + "step": 18986 + }, + { + "epoch": 7.95, + "learning_rate": 1.9673262831021445e-09, + "loss": 0.0445, + "step": 18987 + }, + { + "epoch": 7.95, + "learning_rate": 1.9338415691527724e-09, + "loss": 0.0563, + "step": 18988 + }, + { + "epoch": 7.95, + "learning_rate": 1.9006442407432102e-09, + "loss": 0.0369, + "step": 18989 + }, + { + "epoch": 7.95, + "learning_rate": 1.8677342988260295e-09, + "loss": 0.0451, + "step": 18990 + }, + { + "epoch": 7.95, + "learning_rate": 1.8351117443504706e-09, + "loss": 0.0281, + "step": 18991 + }, + { + "epoch": 7.95, + "learning_rate": 1.8027765782513418e-09, + "loss": 0.0493, + "step": 18992 + }, + { + "epoch": 7.95, + "learning_rate": 1.7707288014601198e-09, + "loss": 0.0474, + "step": 18993 + }, + { + "epoch": 7.95, + "learning_rate": 1.7389684148960694e-09, + "loss": 0.0649, + "step": 18994 + }, + { + "epoch": 7.95, + "learning_rate": 1.7074954194729044e-09, + "loss": 0.0348, + "step": 18995 + }, + { + "epoch": 7.95, + "learning_rate": 1.6763098160965663e-09, + "loss": 0.0643, + "step": 18996 + }, + { + "epoch": 7.95, + "learning_rate": 1.645411605660785e-09, + "loss": 0.0491, + "step": 18997 + }, + { + "epoch": 7.95, + "learning_rate": 1.6148007890570694e-09, + "loss": 0.0546, + "step": 18998 + }, + { + "epoch": 7.95, + "learning_rate": 1.5844773671624958e-09, + "loss": 0.0262, + "step": 18999 + }, + { + "epoch": 7.95, + "learning_rate": 1.5544413408496994e-09, + "loss": 0.0554, + "step": 19000 + }, + { + "epoch": 7.96, + "learning_rate": 1.5246927109824338e-09, + "loss": 0.0643, + "step": 19001 + }, + { + "epoch": 7.96, + "learning_rate": 1.4952314784144606e-09, + "loss": 0.0504, + "step": 19002 + }, + { + "epoch": 7.96, + "learning_rate": 1.46605764399399e-09, + "loss": 0.0667, + "step": 19003 + }, + { + "epoch": 7.96, + "learning_rate": 1.4371712085603506e-09, + "loss": 0.0301, + "step": 19004 + }, + { + "epoch": 7.96, + "learning_rate": 1.4085721729406587e-09, + "loss": 0.0607, + "step": 19005 + }, + { + "epoch": 7.96, + "learning_rate": 1.3802605379609203e-09, + "loss": 0.0729, + "step": 19006 + }, + { + "epoch": 7.96, + "learning_rate": 1.3522363044315977e-09, + "loss": 0.041, + "step": 19007 + }, + { + "epoch": 7.96, + "learning_rate": 1.3244994731598238e-09, + "loss": 0.0508, + "step": 19008 + }, + { + "epoch": 7.96, + "learning_rate": 1.297050044942738e-09, + "loss": 0.0281, + "step": 19009 + }, + { + "epoch": 7.96, + "learning_rate": 1.2698880205697096e-09, + "loss": 0.0211, + "step": 19010 + }, + { + "epoch": 7.96, + "learning_rate": 1.2430134008201145e-09, + "loss": 0.0526, + "step": 19011 + }, + { + "epoch": 7.96, + "learning_rate": 1.2164261864677784e-09, + "loss": 0.0535, + "step": 19012 + }, + { + "epoch": 7.96, + "learning_rate": 1.190126378277645e-09, + "loss": 0.034, + "step": 19013 + }, + { + "epoch": 7.96, + "learning_rate": 1.164113977003556e-09, + "loss": 0.0369, + "step": 19014 + }, + { + "epoch": 7.96, + "learning_rate": 1.1383889833938011e-09, + "loss": 0.0365, + "step": 19015 + }, + { + "epoch": 7.96, + "learning_rate": 1.1129513981888994e-09, + "loss": 0.0465, + "step": 19016 + }, + { + "epoch": 7.96, + "learning_rate": 1.0878012221182676e-09, + "loss": 0.0481, + "step": 19017 + }, + { + "epoch": 7.96, + "learning_rate": 1.0629384559068812e-09, + "loss": 0.0401, + "step": 19018 + }, + { + "epoch": 7.96, + "learning_rate": 1.0383631002686133e-09, + "loss": 0.0436, + "step": 19019 + }, + { + "epoch": 7.96, + "learning_rate": 1.0140751559095663e-09, + "loss": 0.0551, + "step": 19020 + }, + { + "epoch": 7.96, + "learning_rate": 9.900746235269598e-10, + "loss": 0.034, + "step": 19021 + }, + { + "epoch": 7.96, + "learning_rate": 9.66361503812463e-10, + "loss": 0.0473, + "step": 19022 + }, + { + "epoch": 7.96, + "learning_rate": 9.429357974477526e-10, + "loss": 0.0478, + "step": 19023 + }, + { + "epoch": 7.96, + "learning_rate": 9.197975051045138e-10, + "loss": 0.0592, + "step": 19024 + }, + { + "epoch": 7.97, + "learning_rate": 8.969466274488802e-10, + "loss": 0.0388, + "step": 19025 + }, + { + "epoch": 7.97, + "learning_rate": 8.743831651369938e-10, + "loss": 0.0497, + "step": 19026 + }, + { + "epoch": 7.97, + "learning_rate": 8.521071188183349e-10, + "loss": 0.0483, + "step": 19027 + }, + { + "epoch": 7.97, + "learning_rate": 8.301184891323921e-10, + "loss": 0.0535, + "step": 19028 + }, + { + "epoch": 7.97, + "learning_rate": 8.084172767119924e-10, + "loss": 0.0634, + "step": 19029 + }, + { + "epoch": 7.97, + "learning_rate": 7.87003482179971e-10, + "loss": 0.064, + "step": 19030 + }, + { + "epoch": 7.97, + "learning_rate": 7.658771061525016e-10, + "loss": 0.0458, + "step": 19031 + }, + { + "epoch": 7.97, + "learning_rate": 7.450381492368763e-10, + "loss": 0.069, + "step": 19032 + }, + { + "epoch": 7.97, + "learning_rate": 7.244866120326155e-10, + "loss": 0.0442, + "step": 19033 + }, + { + "epoch": 7.97, + "learning_rate": 7.042224951292475e-10, + "loss": 0.0567, + "step": 19034 + }, + { + "epoch": 7.97, + "learning_rate": 6.842457991096396e-10, + "loss": 0.0524, + "step": 19035 + }, + { + "epoch": 7.97, + "learning_rate": 6.64556524547777e-10, + "loss": 0.0531, + "step": 19036 + }, + { + "epoch": 7.97, + "learning_rate": 6.451546720109836e-10, + "loss": 0.0392, + "step": 19037 + }, + { + "epoch": 7.97, + "learning_rate": 6.260402420554812e-10, + "loss": 0.0514, + "step": 19038 + }, + { + "epoch": 7.97, + "learning_rate": 6.072132352308302e-10, + "loss": 0.0479, + "step": 19039 + }, + { + "epoch": 7.97, + "learning_rate": 5.886736520788195e-10, + "loss": 0.0484, + "step": 19040 + }, + { + "epoch": 7.97, + "learning_rate": 5.70421493132356e-10, + "loss": 0.0726, + "step": 19041 + }, + { + "epoch": 7.97, + "learning_rate": 5.52456758915465e-10, + "loss": 0.0393, + "step": 19042 + }, + { + "epoch": 7.97, + "learning_rate": 5.347794499444004e-10, + "loss": 0.0583, + "step": 19043 + }, + { + "epoch": 7.97, + "learning_rate": 5.173895667287543e-10, + "loss": 0.0459, + "step": 19044 + }, + { + "epoch": 7.97, + "learning_rate": 5.00287109767017e-10, + "loss": 0.0605, + "step": 19045 + }, + { + "epoch": 7.97, + "learning_rate": 4.834720795510173e-10, + "loss": 0.0484, + "step": 19046 + }, + { + "epoch": 7.97, + "learning_rate": 4.669444765637021e-10, + "loss": 0.0448, + "step": 19047 + }, + { + "epoch": 7.97, + "learning_rate": 4.5070430128135724e-10, + "loss": 0.0276, + "step": 19048 + }, + { + "epoch": 7.98, + "learning_rate": 4.34751554169166e-10, + "loss": 0.0406, + "step": 19049 + }, + { + "epoch": 7.98, + "learning_rate": 4.190862356878711e-10, + "loss": 0.0535, + "step": 19050 + }, + { + "epoch": 7.98, + "learning_rate": 4.037083462860025e-10, + "loss": 0.0496, + "step": 19051 + }, + { + "epoch": 7.98, + "learning_rate": 3.8861788640542906e-10, + "loss": 0.0372, + "step": 19052 + }, + { + "epoch": 7.98, + "learning_rate": 3.738148564813582e-10, + "loss": 0.0401, + "step": 19053 + }, + { + "epoch": 7.98, + "learning_rate": 3.5929925693789504e-10, + "loss": 0.0407, + "step": 19054 + }, + { + "epoch": 7.98, + "learning_rate": 3.4507108819248347e-10, + "loss": 0.0543, + "step": 19055 + }, + { + "epoch": 7.98, + "learning_rate": 3.31130350655906e-10, + "loss": 0.0509, + "step": 19056 + }, + { + "epoch": 7.98, + "learning_rate": 3.174770447256226e-10, + "loss": 0.0465, + "step": 19057 + }, + { + "epoch": 7.98, + "learning_rate": 3.0411117079687244e-10, + "loss": 0.0361, + "step": 19058 + }, + { + "epoch": 7.98, + "learning_rate": 2.910327292526827e-10, + "loss": 0.0573, + "step": 19059 + }, + { + "epoch": 7.98, + "learning_rate": 2.7824172046941877e-10, + "loss": 0.0667, + "step": 19060 + }, + { + "epoch": 7.98, + "learning_rate": 2.6573814481456463e-10, + "loss": 0.0506, + "step": 19061 + }, + { + "epoch": 7.98, + "learning_rate": 2.535220026478324e-10, + "loss": 0.0436, + "step": 19062 + }, + { + "epoch": 7.98, + "learning_rate": 2.4159329431894253e-10, + "loss": 0.0639, + "step": 19063 + }, + { + "epoch": 7.98, + "learning_rate": 2.2995202017317418e-10, + "loss": 0.0805, + "step": 19064 + }, + { + "epoch": 7.98, + "learning_rate": 2.1859818054248416e-10, + "loss": 0.041, + "step": 19065 + }, + { + "epoch": 7.98, + "learning_rate": 2.075317757543882e-10, + "loss": 0.0455, + "step": 19066 + }, + { + "epoch": 7.98, + "learning_rate": 1.9675280612863056e-10, + "loss": 0.0531, + "step": 19067 + }, + { + "epoch": 7.98, + "learning_rate": 1.8626127197274302e-10, + "loss": 0.0634, + "step": 19068 + }, + { + "epoch": 7.98, + "learning_rate": 1.7605717358870623e-10, + "loss": 0.0579, + "step": 19069 + }, + { + "epoch": 7.98, + "learning_rate": 1.6614051127072929e-10, + "loss": 0.057, + "step": 19070 + }, + { + "epoch": 7.98, + "learning_rate": 1.565112853030293e-10, + "loss": 0.0649, + "step": 19071 + }, + { + "epoch": 7.98, + "learning_rate": 1.47169495963162e-10, + "loss": 0.0457, + "step": 19072 + }, + { + "epoch": 7.99, + "learning_rate": 1.3811514351869116e-10, + "loss": 0.0513, + "step": 19073 + }, + { + "epoch": 7.99, + "learning_rate": 1.2934822823051917e-10, + "loss": 0.0707, + "step": 19074 + }, + { + "epoch": 7.99, + "learning_rate": 1.2086875035066668e-10, + "loss": 0.0522, + "step": 19075 + }, + { + "epoch": 7.99, + "learning_rate": 1.1267671012227255e-10, + "loss": 0.0533, + "step": 19076 + }, + { + "epoch": 7.99, + "learning_rate": 1.0477210778181424e-10, + "loss": 0.0513, + "step": 19077 + }, + { + "epoch": 7.99, + "learning_rate": 9.715494355577726e-11, + "loss": 0.0557, + "step": 19078 + }, + { + "epoch": 7.99, + "learning_rate": 8.982521766287556e-11, + "loss": 0.0576, + "step": 19079 + }, + { + "epoch": 7.99, + "learning_rate": 8.278293031405148e-11, + "loss": 0.0567, + "step": 19080 + }, + { + "epoch": 7.99, + "learning_rate": 7.602808171247588e-11, + "loss": 0.0424, + "step": 19081 + }, + { + "epoch": 7.99, + "learning_rate": 6.956067205132754e-11, + "loss": 0.0524, + "step": 19082 + }, + { + "epoch": 7.99, + "learning_rate": 6.338070151712394e-11, + "loss": 0.043, + "step": 19083 + }, + { + "epoch": 7.99, + "learning_rate": 5.748817028750075e-11, + "loss": 0.056, + "step": 19084 + }, + { + "epoch": 7.99, + "learning_rate": 5.1883078531211886e-11, + "loss": 0.0527, + "step": 19085 + }, + { + "epoch": 7.99, + "learning_rate": 4.656542641034989e-11, + "loss": 0.0539, + "step": 19086 + }, + { + "epoch": 7.99, + "learning_rate": 4.153521407590511e-11, + "loss": 0.0525, + "step": 19087 + }, + { + "epoch": 7.99, + "learning_rate": 3.679244167442697e-11, + "loss": 0.0427, + "step": 19088 + }, + { + "epoch": 7.99, + "learning_rate": 3.2337109342472917e-11, + "loss": 0.0385, + "step": 19089 + }, + { + "epoch": 7.99, + "learning_rate": 2.8169217205498145e-11, + "loss": 0.0423, + "step": 19090 + }, + { + "epoch": 7.99, + "learning_rate": 2.428876538562719e-11, + "loss": 0.0612, + "step": 19091 + }, + { + "epoch": 7.99, + "learning_rate": 2.069575399388235e-11, + "loss": 0.064, + "step": 19092 + }, + { + "epoch": 7.99, + "learning_rate": 1.7390183133514373e-11, + "loss": 0.0468, + "step": 19093 + }, + { + "epoch": 7.99, + "learning_rate": 1.4372052900002432e-11, + "loss": 0.049, + "step": 19094 + }, + { + "epoch": 7.99, + "learning_rate": 1.1641363378833703e-11, + "loss": 0.0562, + "step": 19095 + }, + { + "epoch": 7.99, + "learning_rate": 9.198114648834023e-12, + "loss": 0.0558, + "step": 19096 + }, + { + "epoch": 8.0, + "learning_rate": 7.042306781057662e-12, + "loss": 0.0429, + "step": 19097 + }, + { + "epoch": 8.0, + "learning_rate": 5.173939836566888e-12, + "loss": 0.0506, + "step": 19098 + }, + { + "epoch": 8.0, + "learning_rate": 3.5930138708728523e-12, + "loss": 0.0493, + "step": 19099 + }, + { + "epoch": 8.0, + "learning_rate": 2.2995289261640295e-12, + "loss": 0.0358, + "step": 19100 + }, + { + "epoch": 8.0, + "learning_rate": 1.2934850424084489e-12, + "loss": 0.0476, + "step": 19101 + }, + { + "epoch": 8.0, + "learning_rate": 5.748822484719086e-13, + "loss": 0.0416, + "step": 19102 + }, + { + "epoch": 8.0, + "learning_rate": 1.4372056322820017e-13, + "loss": 0.0568, + "step": 19103 + }, + { + "epoch": 8.0, + "learning_rate": 0.0, + "loss": 0.0492, + "step": 19104 + }, + { + "epoch": 8.0, + "step": 19104, + "total_flos": 3.498653672693301e+18, + "train_loss": 0.223547331811361, + "train_runtime": 91750.5488, + "train_samples_per_second": 6.664, + "train_steps_per_second": 0.208 + } + ], + "max_steps": 19104, + "num_train_epochs": 8, + "total_flos": 3.498653672693301e+18, + "trial_name": null, + "trial_params": null +} diff --git a/training_args.bin b/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..8b1b50208b5713a1848bb03b9cf39bd33c748f72 --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:457d7e76c0fce8d03d8cfe3fdedf1b89cbf25e0f3a605b6976f387e928f69251 +size 3899